JOURNAL ONCOLOGY CLINICAL - PDF Free Download

Volume 34, Issue 15S, Part I of II

May 20, 2016

JOURNAL OF CLINICAL ONCOLOGY Official Journal of the American Society of Clinical Oncology

2016 ASCO Annual Meeting Proceedings

52nd Annual Meeting June 3-7, 2016 McCormick Place Chicago, IL

www.jco.org

52nd Annual Meeting of the American Society of Clinical Oncology June 3-7, 2016 Chicago, Illinois 2016 Annual Meeting Proceedings Part I (a supplement to the Journal of Clinical Oncology)

Copyright 2016 American Society of Clinical Oncology

Editor: Michael A. Carducci, MD Managing Editor: Amy Hindman Production Manager: Donna Dottellis

Requests for permission to reprint abstracts should be directed to Intellectual Property Rights Manager, American Society of Clinical Oncology, 2318 Mill Road, Suite 800, Alexandria, VA 22314. Tel: 571-483-1300; Fax: 571-366-9530; Email: [email protected]. Editorial correspondence and production questions should be addressed to Managing Editor, Annual Meeting Proceedings, American Society of Clinical Oncology, 2318 Mill Road, Suite 800, Alexandria, VA 22314. Email: [email protected]. Copyright © 2016 American Society of Clinical Oncology. All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without written permission by the Society. The American Society of Clinical Oncology assumes no responsibility for errors or omissions in this document. The reader is advised to check the appropriate medical literature and the product information currently provided by the manufacturer of each drug to be administered to verify the dosage, the method and duration of administration, or contraindications. It is the responsibility of the treating physician or other health care professional, relying on independent experience and knowledge of the patient, to determine drug, disease, and the best treatment for the patient. Abstract management and indexing provided by CONFEX, Cumberland, RI. Composition services and print production provided by Sheridan.

Volume 34, Issue 15S

Supplement to May 20, 2016

CONTENTS 2016 ASCO ANNUAL MEETING PROCEEDINGS Special Award Lecture Abstracts .............................................................................................................

1s

Plenary Session (Abstracts LBA1 - LBA4) .......................................................................................................................................

4s

Special Clinical Science Symposia (Abstracts 100 - 109) .............................................................................................................................................

6s

Breast Cancer—HER2/ER Scheduled presentations (Abstracts 500 - TPS636) .........................................................................................

9s

Breast Cancer—Triple-Negative/Cytotoxics/Local Therapy Scheduled presentations (Abstracts 1000 - TPS1105) .......................................................................................

43s

Cancer Prevention, Hereditary Genetics, and Epidemiology Scheduled presentations (Abstracts 1500 - TPS1594) ......................................................................................

69s

Central Nervous System Tumors Scheduled presentations (Abstracts LBA2000 - TPS2084) ............................................................................

93s

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics Scheduled presentations (Abstracts 2500 - TPS2609) ....................................................................................

114s

Developmental Therapeutics—Immunotherapy Scheduled presentations (Abstracts 3000 - TPS3114) ......................................................................................

142s

Gastrointestinal (Colorectal) Cancer Scheduled presentations (Abstracts 3500 - 4102) ............................................................................................

171s

continued on following page

Journal of Clinical Oncology (ISSN 0732-183X) is published 36 times a year, three times monthly, by the American Society of Clinical Oncology, 2318 Mill Road, Suite 800, Alexandria, VA 22314. Periodicals postage is paid at Alexandria, VA, and at additional mailing ofﬁces. Publication Mail Agreement Number 863289. Editorial correspondence should be addressed to Stephen A. Cannistra, MD, Journal of Clinical Oncology, 2318 Mill Road, Suite 800, Alexandria, VA 22314. Phone: 703-797-1900; Fax: 703-684-8720. E-mail: [email protected]. Internet: www.jco.org. Journal of Clinical Oncology® is a registered trademark of American Society of Clinical Oncology, Inc. POSTMASTER: Send address changes to American Society of Clinical Oncology, 2318 Mill Road, Suite 800, Alexandria, VA 22314. ASCO members should send changes of address to American Society of Clinical Oncology, 2318 Mill Road, Suite 800, Alexandria, VA 22314. Nonmembers should send changes of address to Journal of Clinical Oncology Customer Service, 2318 Mill Road, Suite 800, Alexandria, VA 22314. 2016 annual subscription rates, effective September 1, 2015: United States and possessions: individual, $625 one year, $1,188 two years; single issue, $40. International: individual, $867 one year, $1,647 two years; single issue, $50. Institutions: bundled (print 1 online): Tier 1: $1,016 US, $1,410 Int’l; Tier2: $1,172 US, $1,555 Int’l; Tier 3: $1,692 US, $2,060 Int’l; Tier 4: contact JCO for quote. Institutions: online only, worldwide: Tier 1: $856; Tier 2: $981; Tier 3: $1,415; Tier 4: contact JCO for quote. See http://jco.ascopubs.org/site/subscriptions/ ratecard.pdf for descriptions of each tier. Student and resident: United States and possessions: $303; all other countries, $421. To receive student/resident rate, orders must be accompanied by name of afﬁliated institution, date of term, and the signature of program/residency coordinator on institution letterhead. Orders will be billed at individual rate until proof of status is received. Current prices are in effect for back volumes and back issues. Back issues sold in conjunction with a subscription rate are on a prorated basis. Subscriptions are accepted on a 12-month basis. Prices are subject to change without notice. Single issues, both current and back, exist in limited quantities and are offered for sale subject to availability. JCO Legacy Archive (electronic back issues from January 1983 through December 1998) is also available; please inquire.

Volume 34, Issue 15S

Supplement to May 20, 2016

................................................................................................................................

Gastrointestinal (Noncolorectal) Cancer Scheduled presentations (Abstracts 4000 - TPS4154) .....................................................................................

206s

Genitourinary (Nonprostate) Cancer Scheduled presentations (Abstracts LBA4500 - TPS4585) ............................................................................

244s

Genitourinary (Prostate) Cancer Scheduled presentations (Abstracts 5000 - TPS5095) ....................................................................................

266s

Gynecologic Cancer Scheduled presentations (Abstracts 5501 - TPS5612) ......................................................................................

290s

Head and Neck Cancer Scheduled presentations (Abstracts 6000 - TPS6110) ......................................................................................

317s

Health Services Research and Quality of Care Scheduled presentations (Abstracts LBA6500 - TPS6632) .............................................................................

345s

Hematologic Malignancies—Leukemia, Myelodysplastic Syndromes, and Allotransplant Scheduled presentations (Abstracts 7000 - TPS7080) ....................................................................................

375s

Hematologic Malignancies—Lymphoma and Chronic Lymphocytic Leukemia Scheduled presentations (Abstracts 7500 - TPS7585) ....................................................................................

395s

Hematologic Malignancies—Plasma Cell Dyscrasia Scheduled presentations (Abstracts 8000 - TPS8073) ....................................................................................

416s

Lung Cancer—Non-Small Cell Local-Regional/Small Cell/Other Thoracic Cancers Scheduled presentations (Abstracts 8500 - TPS8581) .....................................................................................

435s

Lung Cancer—Non-Small Cell Metastatic Scheduled presentations (Abstracts 9000 - TPS9110) ......................................................................................

456s

Melanoma/Skin Cancers Scheduled presentations (Abstracts 9500 - TPS9600) ....................................................................................

484s

Patient and Survivor Care Scheduled presentations (Abstracts 10000 - TPS10145) ..................................................................................

510s

Pediatric Oncology Scheduled presentations (Abstracts 10500 - TPS10588) .................................................................................

547s

Sarcoma Scheduled presentations (Abstracts 11000 - TPS11073) ....................................................................................

569s

Tumor Biology Scheduled presentations (Abstracts 11500 - TPS11621) .....................................................................................

587s

Author Index ......................................................................................................................................................

618s

www.jco.org

American Society of Clinical Oncology 52nd Annual Meeting 2016 Abstracts Descriptions of Scientific Sessions Plenary Session The Plenary Session includes abstracts selected by the Scientific Program Committee as having practice-changing findings of the highest scientific merit. Highlights of the Day Sessions Highlights of the Day Sessions invite expert discussants to provide an overview of the previous day’s Oral Abstract presentations, focusing on key findings and putting abstracts into clinical context. Oral Abstract Sessions Oral Abstract Sessions include didactic presentations of abstracts of the highest scientific merit, as determined by the Scientific Program Committee. Experts in the field serve as discussants and provide comprehensive themed discussions of the findings from the abstracts. Clinical Science Symposia Clinical Science Symposia provide a forum for science in oncology, combining didactic lectures on a specific topic with abstract presentations. Experts in the field serve as discussants, placing studies in the appropriate context and critically discussing the applicability of the conclusions in clinical practice. Three special Clinical Science Symposia will be designated around specific pathways that cut across cancer types. Poster Discussion Sessions Select posters from the Poster Sessions will be discussed by expert discussants, with the abstract authors participating in a question and answer period as panel members. These sessions will be followed by networking with the discussants and authors. Poster Sessions Poster Sessions include selected abstracts of clinical research in poster format. Trials in Progress (TPS) abstracts are presented within a track’s Poster Session. Publication-Only Abstracts Publication-only abstracts were selected to be published online in conjunction with the Annual Meeting, but will not be presented at the Meeting. All presented and publication-only abstracts are citable to this Journal of Clinical Oncology supplement. For citation examples, please see the Letter from the Editor. This publication contains abstracts selected by the ASCO Scientific Program Committee for presentation at the 2016 Annual Meeting. Abstracts selected for electronic publication only are available in full-text versions online through ASCO.org and JCO.org. The type of session, the day, and the session start/end times are located to the right of the abstract number for scheduled presentations. To determine the location of the abstract session, refer to the Annual Meeting Program or the iPlanner, the online version of the Annual Meeting Program, available at am.asco.org. Dates and times are subject to change. All modifications will be posted on am.asco.org. The deadline for abstract submission for the 2017 Annual Meeting is Tuesday, February 7, 2017, at 11:59 PM (EST).

Letter from the Editor

T

he 2016 ASCO Annual Meeting Proceedings Part I (a supplement to the Journal of Clinical Oncology) is an enduring record of the more than 2,400 abstracts selected by the ASCO Scientific Program Committee for presentation at the 52nd ASCO Annual Meeting. Accepted abstracts not presented at the meeting are included in the online supplement to the May 20 issue of Journal of Clinical Oncology at JCO.org. The majority of abstracts selected for presentation are included here in full and are categorized by scientific track. After the Annual Meeting, abstracts can be accessed online through ASCO University’s Meeting Library (meetinglibrary.asco.org/abstracts). Online abstracts include the full list of abstract authors and their disclosure information. Late-Breaking Abstracts are represented here by abstract title and presenting author only. The full-text versions of these abstracts will be publicly released through ASCO.org during the Annual Meeting. Late-

Breaking Abstracts will also be included in the 2016 ASCO Annual Meeting Proceedings Part II, an online supplement to the June 20 issue of Journal of Clinical Oncology on JCO.org. Print versions of these abstracts will be available onsite at the Annual Meeting in the ASCO Daily News. All abstracts carry Journal of Clinical Oncology citations. The following are citation examples for print and electronic abstracts: J Clin Oncol 34:4s, 2016 (suppl; abstr LBA1) J Clin Oncol 34, 2016 (suppl; abstr e12000) Should you have any questions or comments about this publication, we encourage you to provide feedback by contacting us at [email protected]. Michael A. Carducci, MD Editor, 2016 ASCO Annual Meeting Proceedings

JOURNAL of CLINICAL ONCOLOGY ............................................................................... Ofﬁcial Journal of the American Society of Clinical Oncology Journal of Clinical Oncology (ISSN 0732-183X) is published 36

Orders and Payments

times a year, three times monthly, by the American Society of

P.O. Box 37211 Baltimore, MD 21279-3211

Clinical Oncology, 2318 Mill Road, Suite 800, Alexandria, VA 22314. Periodicals postage is paid at Alexandria, VA, and at additional mailing ofﬁces.

Important Tiers and Pricing Notes Additional rates along with tier descriptions are available online

Postmaster

at http://jco.ascopubs.org/site/subscriptions/ratecard.pdf

Send all changes of address for Journal of Clinical Oncology subscribers to: JCO Customer Service 2318 Mill Road, Suite 800 Alexandria, VA 22314 Editorial Correspondence (manuscript-related inquiries): Stephen A. Cannistra, MD, Editor-in-Chief Journal of Clinical Oncology 2318 Mill Road, Suite 800 Alexandria, VA 22314 Phone: 703-797-1900; Fax: 703-684-8720 E-mail: [email protected]; Internet: www.jco.org American Society of Clinical Oncology (membership-related inquiries):

1. Prices are in effect from September 1, 2015, through August 31, 2016. Prices are subject to change. 2. Print-only subscriptions or additional print subscriptions are available for $860 in the US and $1,244 outside the US. 3. For multisite licenses, please contact the appropriate agent for a quote. 4. Subscribers outside the US, add $100 per print subscription for expedited delivery. 5. Single-issue price: $40 US, $50 international. 6. Prices quoted are in US dollars, and payments must be made in US dollars. 7. Except on consortia orders, the publisher allows for a 5% discount on Tiers 1-3 to recognized subscription agents. Prices are subject to change without notice. Current prices are in effect for back volumes and back issues. Single issues, both

ASCO Member Services 2318 Mill Road, Suite 800 Alexandria, VA 22314

current and back, exist in limited quantities and are offered for sale

Phone: 703-299-0158; Toll-free: 888-282-2552 Fax: 703-299-0255 E-mail: [email protected]; Internet: www.asco.org Hours: Monday-Friday, 8:30 a.m.-5:00 p.m. Eastern Time Customer Service, Subscriptions, and Changes of Address: JCO Customer Service

subject to availability. Back issues sold in conjunction with a subscription are on a prorated basis. Advertising Sales The Walchli Tauber Group, Inc. 225 Old Emmorton Road, Suite 201 Bel Air, MD 21015 Phone: 443-512-8899; Fax: 443-512-8909 Internet: www.wt-group.com

2318 Mill Road, Suite 800 Alexandria, VA 22314 Phone: 703-519-1430; Toll-free: 888-273-3508; Fax: 703-518-8155 E-mail: [email protected] Internet orders/renewals: www.jco.org/subscriptions

Business-to-Business Sales Rick Werdann Springer Healthcare, LLC

2016 SUBSCRIPTION RATES

233 Spring Street New York, NY 10013

Individual Prices

Phone: 212-460-1523; Mobile: 646-209-1840

Domestic (US) Print 1 Online

E-mail: [email protected] Internet: www.SpringerHealthcare.com

Individuals in training $303 Individuals (1 year) $625

LICENSES AND CONSORTIA

International Print 1 Online

US, Canada, and Europe-Single Site License

Individuals in training $421

ASCO Customer Service 2318 Mill Road, Suite 800

Individuals (1 year) $867

Alexandria, VA 22314

Institutional Prices Domestic (US) Print 1 Online

Online Only

Tier Tier Tier Tier

$856 $981 $1,415 Call for quote

1 2 3 4

$1,016 $1,172 $1,692 Call for quote

International Print 1 Online

Online Only

Tier Tier Tier Tier

$856 $981 $1,415 Call for quote

1 2 3 4

$1,410 $1,555 $2,060 Call for quote

Phone: 1-888-282-2552 or 1-703-299-0158 Fax: 703-518-8155 E-mail: [email protected] US, Canada, and Europe-Multisite Licenses and Consortia David Charles eLicensing 92 Avenue du General de Gaulle 78600 Maisons-Lafﬁtte, France Phone/Fax: 133-1-39-12-29-29 E-mail: [email protected]

JOURNAL of CLINICAL ONCOLOGY .................................................................................... Ofﬁcial Journal of the American Society of Clinical Oncology Japan USACO Corporation 2-17-12 Higashi-Azuba Minato-ku Tokyo, Japan 106-0044 Phone: 181-3-3505-3529; Fax: 181-3-3505-6284 E-mail: [email protected]; Internet: www.usaco.co.jp

Free Public Access Journal of Clinical Oncology (JCO) provides free online access to original research articles older than one year at www.jco.org. In addition, all ASCO Special Articles, Rapid Communications, Editorials, Comments and Controversies articles, the Art of Oncology series, and Correspondence articles are free immediately upon publication.

China Charlesworth China Room 3305, Building 6, Jianwai SOHO No. 39 Dongsanhuan Zhonglu Chaoyang District, Beijing 100022 PR China Phone: 186-10-58696201; Fax: 186-10-58696201 E-mail: [email protected] Internet: www.charlesworth.com.cn (in Mandarin) and www.charlesworth.com

Disclaimer The ideas and opinions expressed in JCO do not necessarily reﬂect those of the American Society of Clinical Oncology (ASCO). The mention of any product, service, or therapy in this publication or in any advertisement in this publication should not be construed as an endorsement of the products mentioned. It is the responsibility of the treating physician or other health care provider, relying on independent experience and knowledge of the patient, to determine drug dosages and the best treatment for the patient. Readers are advised to check the appropriate medical literature and the product information currently provided by the manufacturer of each drug to be administered to verify approved uses, the dosage, method, and duration of administration, or contraindications. Readers are also encouraged to contact the manufacturer with questions about the features or limitations of any products. ASCO assumes no responsibility for any injury or damage to persons or property arising out of or related to any use of the material contained in this publication or to any errors or omissions.

India Publishers Communication Group Plot No. 692 B-3, Ashok Vatika Deoli Road, Khanpur New Delhi-110062 India Phone: 191-9891096706 E-mail: [email protected] South Korea EBSCO Korea Phone: 182-2-598-2571 E-mail: [email protected] Taiwan EBSCO Taiwan Phone: 1886 2-27751596 E-mail: [email protected] Latin America, Caribbean, Middle East, and North Africa Accucoms (US), Inc. West Point Commons 1816 West Point Pike, Suite 201 Lansdale, PA 19446 Phone: 215-393-5026 Fax: 215-660-5042 E-mail (Latin America): [email protected] E-mail (Middle East): [email protected] Internet: www.accucoms.com Permissions Requests Licensing, Rights, and Permissions Division American Society of Clinical Oncology 2318 Mill Road, Suite 800 Alexandria, VA 22314 Phone: 571-483-1722; Fax: 703-518-5094 E-mail: [email protected]

Copyright Copyright © 2016 by American Society of Clinical Oncology unless otherwise indicated. All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means now or hereafter known, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the Publisher. Printed in the United States of America. The appearance of the code at the bottom of the left column of the ﬁrst page of an article in this journal indicates the copyright owner’s consent that copies of the article may be made for personal or internal use, or for the personal or internal use of speciﬁc clients, for those registered with the Copyright Clearance Center, Inc. (222 Rosewood Drive, Danvers, MA 01923; 978-7508400; www.copyright.com). This consent is given on the condition that the copier pay the stated per-copy fee for that article through the Copyright Clearance Center, Inc., for copying beyond that permitted by Sections 107 or 108 of the US Copyright Law. This consent does not extend to other kinds of copying, such as copying for general distribution, for advertising or promotional purposes, for creating new collective works, or for resale. Absence of the code indicates that the material may not be processed through the Copyright Clearance Center, Inc. CPT © is a trademark of the American Medical Association. Journal of Clinical Oncology® is a registered trademark of American Society of Clinical Oncology, Inc.

ASCO Abstracts Policy Public Release of Abstracts The abstracts published in the 2016 ASCO Annual Meeting Proceedings Part I, including those abstracts published but not presented at the Meeting, were publicly released by ASCO at 5:00 PM (EDT) on Wednesday, May 18, 2016. These abstracts are publicly available online through ASCO.org, the official website of the Society. Late-Breaking Abstracts, which include all Plenary Abstracts, will be publicly released according to the following schedule: Abstracts presented in a press briefing or scientific presentation (whichever comes first) on · Late-Breaking Friday, June 3, will be publicly released Friday, June 3, through ASCO.org at 2:00 PM (EDT). Abstracts presented in a press briefing or scientific presentation (whichever comes first) on · Late-Breaking Saturday, June 4, will be publicly released Saturday, June 4, through ASCO.org at 7:30 AM (EDT). Abstracts presented in a press briefing or scientific presentation (whichever comes first) on · Late-Breaking Sunday, June 5, will be publicly released Sunday, June 5, through ASCO.org at 7:30 AM (EDT). Abstracts presented in a press briefing or scientific presentation (whichever comes first) on · Late-Breaking Monday, June 6, or Tuesday, June 7, will be publicly released Monday, June 6, through ASCO.org at 7:30 AM (EDT).

Abstracts will be available in Section D of ASCO Daily News on the day of their scientific · Late-Breaking presentation, with the exception of abstracts presented on Friday (these will appear in the Saturday issue) and those presented on Tuesday (these will appear in the Monday issue).

In the unlikely event that ASCO publicly releases an abstract in advance of the scheduled time, the release will be publicly announced on ASCO.org.

Abstract Notice All abstracts presented at and published in conjunction with the Annual Meeting are included in online supplements to the Journal of Clinical Oncology. The abstracts released on May 18, 2016, are included in the May 20 (Vol. 34, No. 15S) issue (2016 Annual Meeting Proceedings Part I), and the Late-Breaking Abstracts, released on a daily basis during the Meeting, are included in the June 20 (Vol. 34, No. 18S) issue (2016 Annual Meeting Proceedings Part II).

Copyright 2016 American Society of Clinical Oncology

ASCO’s Policy for Relationships with Companies In compliance with standards established by ASCO’s Policy for Relationships with Companies (Conflict of Interest; J Clin Oncol. 2013;31[16]:2043–2043) and the Accreditation Council for Continuing Medical Information (ACCME), ASCO’s intent is to promote balance, independence, objectivity, and scientific rigor through the disclosure of financial and other interests, and in the identification and management of potential conflicts. According to the ASCO’s Policy for Relationships with Companies, all authors are expected to provide general disclosure information for 11 disclosure categories of relationships with for-profit health care companies. Please visit asco.org/rwc for more information on the ASCO Conflict of Interest Policy and the Conflict of Interest Policy Implementation Plan for CME Activities.

2016 ASCO Annual Meeting Proceedings

Special Awards

ABSTRACTS American Society of Clinical Oncology 52nd Annual Meeting June 3-7, 2016 McCormick Place Chicago, IL SPECIAL AWARD LECTURE ABSTRACTS David A. Karnofsky Memorial Award and Lecture Saturday, June 4, 9:30 AM A lung cancer journey: 1975-2016. Paul A. Bunn Jr., MD, FASCO; University of Colorado School of Medicine, Denver, CO The 2016 Karnofsky award is an honor for me and is a reflection on the efforts of many dedicated individual scientists and physicians and patients, families and advocates whose efforts have led to major reductions in lung cancer mortality and considerable improvements in outcomes for those diagnosed with lung cancer. While lung cancer remains the leading cause of cancer mortality worldwide with more than 1.6 million deaths annually and the leading cause of cancer mortality in both men and women in developed countries, mortality rates have fallen dramatically in U.S. men since 1985 and in U.S. women over the past few years. Prevention efforts, especially in cigarette consumption, are responsible for the largest part of the mortality reduction and more needs to be done in this regard. Early detection with low dose spiral CT screening should play an increasing role in future mortality reductions. The majority of early-stage lung cancer patients are not cured with local therapy and the majority of all patients present with metastatic disease, highlighting the importance of systemic therapy. Systemic chemotherapy made small incremental gains in survival outcomes that came with considerable toxicity costs. More recently, scientific advances in the biology and molecular biology underlying lung cancer have borne fruit in lung cancer treatment outcomes with targeted and molecular therapies coupled with predictive biomarkers to aid in selection of the right drug for the right patient at the right time and with reduced toxicity. More recently, the introduction of immune therapies improved survival in second-line therapy and there is reason for optimism that such therapies applied to the right patients could improve first-line therapy outcomes in metastatic disease and potentially add to cure rates in earlier stages. What an exciting time for lung cancer investigation.

Science of Oncology Award and Lecture Sunday, June 5, 1:00 PM New cancer treatments emerging from studies of the VHL tumor suppressor protein and IDH oncoproteins. William G. Kaelin Jr., MD; Howard Hughes Medical Institute, Dana-Farber Cancer Institute, and Brigham and Women’s Hospital, Harvard Medical School, Boston, MA Loss of function germline VHL tumor suppressor gene mutations cause von Hippel-Lindau syndrome, characterized by an increased risk of clear cell renal carcinoma (ccRCC)—the most common form of kidney cancer—hemangioblastoma, and pheochromocytoma. Tumor development in this setting is caused by loss of the remaining normal VHL allele in a susceptible cell. Importantly, inactivation of the maternal and paternal VHL alleles, due to acquired mutations or hypermethylation, is also very common in non-hereditary ccRCC. The VHL protein, pVHL, is part of an ubiquitin ligase that targets the HIF transcription factor for degradation. Accordingly, pVHL loss upregulates HIF and HIF-responsive gene products (e.g., VEGF and Epo). This explains the hypervascularity of ccRCCs and their ability to cause erythrocytosis. We showed that inhibiting HIF, and particularly HIF2, is necessary and sufficient for renal cancer suppression by pVHL in preclinical models. This knowledge motivated the successful testing and eventual approval of drugs that inhibit VEGF (VEGF and VEGFR antagonists) or VEGF signaling (mTOR inhibitors) for ccRCC. Targeting HIF2 should, however, be more efficacious than targeting any one HIF2-responsive gene product. We found that a small molecule HIF2a inhibitor that has recently entered clinical trials is efficacious in preclinical pVHL-defective kidney cancer models. Our work on the antimyeloma drug lenalidomide suggests another way to target transcription factors such as HIF2, namely, small molecule mediated target destabilization. Some cancers, including some acute leukemias and gliomas, are driven by IDH1 or IDH2 mutations. These mutations cause the accumulation of mM amounts of R-2 hydroxyglutarate (R-2HG). R-2HG can modulate the activities of various 2-oxoglutarate-dependent enzymes, including the EglN prolyl hydroxylases that mark HIF for recognition by pVHL. We showed that the effects of R-2HG in leukemia are reversible and drugs that inhibit R-2HG production are promising in early clinical trials in leukemia patients.

1s

2s

Special Awards

ASCO–American Cancer Society Award and Lecture Monday, June 3, 3:00 PM Moving prevention mechanisms from the bench into the lung cancer clinic. Ethan Dmitrovsky, MD; The University of Texas MD Anderson Cancer Center, Houston, TX Lung cancer is the most common cause of cancer death for men and women. Even if all anti-smoking goals are met, lung cancer will remain a major cause of cancer death. This is because more lung cancers are now diagnosed collectively in never and former smokers than current ones. This is why cancer prevention is important. Behaviors that prevent lung cancers (smoking prevention and cessation) are essential, but pharmacologic interventions (chemoprevention) are also needed. We uncovered clinically relevant lung cancer chemoprevention mechanisms that can be targeted. These include retinoic acid receptor (RAR) and retinoid X receptor (RXR, rexinoid) pathways that trigger cyclin destabilization. This causes cell cycle arrest and repair of carcinogenic damage. Unexpectedly, the RAR pathway is silenced in lung cancer. This confers clinical retinoid resistance. Yet, rexinoids can overcome this resistance. Epidermal growth factor receptor-tyrosine kinase inhibitors (EGFR-TKIs) independently repress cyclin expression. This led us to combine an RXR with an EGFR-TKI agonist to combat lung cancers. To confirm translational relevance, mice were exposed to a potent carcinogen that caused lung cancers to readily form. Treatment with a rexinoid or an EGFR-TKI prevented these lung cancers. To learn if cyclin destabilization is a lung cancer prevention target, mice were also engineered to aberrantly express cyclins in the lung. This triggered lung cancers to develop. To establish clinical relevance, five clinical trials were conducted, spanning from phase 0 through phase II. These discovered that the same mechanisms engaged in the laboratory were active in clinical lung cancers. Clinical activity occurred even in lung cancer cases with KRAS mutations (an unmet medical need). Novel mechanisms were also found that destabilize cyclins and target aneuploidy, as will be discussed. Taken together, a mechanistic approach to cancer chemoprevention is necessary to reduce the heavy burden of lung cancer.

B. J. Kennedy Award and Lecture for Scientific Excellence in Geriatric Oncology Monday, June 6, 9:45 AM Improving older cancer patients’ treatment: Hope for the future. Pierre-Louis Soubeyran, MD, PhD; Institut Bergonie, ´ Bordeaux, France A major objective of geriatric oncology is to design treatment with an optimized efficacy toxicity ratio that will ensure both a good control of cancer while keeping the patient independent and at home. To achieve this goal, three major intermediate objectives should be completed: (a) identification of vulnerable patients at risk for unacceptable complications; (b) evaluation of the capacity of geriatric interventions to correct geriatric domains impairments and; (c) definition of the best strategy to tailor cancer treatment to patients’ status. To fulfill this goal, we may act through various approaches including implementation of geriatric assessment tools and interventions, biology of cancer, and biology of aging. Geriatric tools to identify vulnerable patients before treatment have been developed and are currently implemented in routine practice. Geriatric assessment questionnaires have demonstrated predictive value for unacceptable events occurring during treatment. Geriatric intervention and case management are currently evaluated in clinical trials. These initiatives should be pursued and extended. Some particular aspects should be reinforced including pharmaco-economic studies, pharmaco-epidemiology, and specific trials for vulnerable and frail patients. Biology of aging is improving and potential biomarkers already validated in the general population may be good candidates to include in our screening strategy. Better understanding of the biology of cancer will help design targeted therapies, which may be good candidates for the treatment of the elderly with cancer providing they have a good efficacy/toxicity ratio. As a perspective, this evolution towards a better understanding of both biology of cancer and biology of aging may help elucidate potential interactions between cancer and aging processes in older patients, which gives us hope for better-suited treatments in the near future.

Pediatric Oncology Award and Lecture Sunday, June 5, 11:30 AM Neuroblastoma: Advances through collaboration. Susan L. Cohn, MD; University of Chicago, Chicago, IL Pediatric oncologists have a long-standing track record for collaborative research though the cooperative groups and multi-institutional studies. Significant advances in the treatment of children with neuroblastoma and other cancers have been realized during the past 60 years as a result of this collaborative approach. In addition to conducting clinical trials, leaders in the pediatric cooperative groups had the foresight to collect and bank large numbers of clinically annotated tumor and germline samples. Studies conducted with these precious tissues have led to the discovery of powerful genomic biomarkers that have refined risk classification; provided insight into the

Special Awards

genetic basis of neuroblastoma; and directed us to oncogenic pathways that can be exploited therapeutically. The clinical and biological heterogeneity of neuroblastoma was recognized in the 1970’s, and tailored treatment approaches based on a combination of prognostic factors have been ongoing for decades. Successive risk-based cooperative group clinical trials have led to decreased toxicity and improved outcome for low- and intermediaterisk patients. In addition, high-risk patients have achieved higher survival rates with increasingly intensive, multimodality approaches. In 2010, the efficacy of immunotherapy plus isotretinoin following consolidation with myeloablation and stem cell transplant was demonstrated in a randomized Phase III Children’s Oncology Group (COG) study, leading to the first drug ever approved for neuroblastoma by the U.S. Food and Drug Administration (FDA). However, the progress observed in high-risk patients has come with a cost, and survivors remain at risk for a broad range of adverse outcomes including second malignant neoplasms (SMN) and early death. To ultimately cure patients with high-risk neuroblastoma and improve their quality of life, effective treatment strategies based on specific tumor targets and germline genomics with low toxicity profiles will be needed. The International Neuroblastoma Risk Group (INRG) Task Force has already collaborated to develop tools to facilitate these types of studies.

Gianni Bonadonna Breast Cancer Award and Lecture Saturday, June 4, at 4:45 PM Crosstalk between estrogen and growth factor receptor pathways: Implications for endocrine therapy resistance and new strategies to overcome it. C. Kent Osborne and Rachel Schiff; Dan L. Duncan Cancer Center and Lester and Sue Smith Breast Center, Baylor College of Medicine, Houston, TX Estrogen receptor alpha (ER)-positive breast cancer (BC), despite numerous genomic alterations, is usually dependent, at least initially, on the female hormone estrogen. ER regulates hundreds of genes, many of which are important for tumor growth and progression. Many patients with primary BC enjoy a normal life span because they received endocrine therapy, while others, including all patients with metastatic disease, demonstrate de novo or acquired resistance. Understanding the mechanisms for this resistance would have a major impact in overcoming treatment failure and improving survival of patients. Growth of BC cells is also regulated by various growth factors, oncogenic kinases, stress kinases, micro-environmental stimuli, and other cell survival factors. Polypeptide hormones such as insulin, insulin-like growth factors, epidermal growth factor and numerous other ligands for the HER family of receptors, integrins, and cytokines all regulate normal mammary development as well as BC progression. Interestingly many of these factors are themselves ER-regulated. In addition, ER does not act alone and it’s signaling is modified by a variety of co-regulatory proteins that form transcriptional complexes on DNA. Some of the co-activators are oncogenes themselves and can dramatically alter ER activity. Remarkably, many of these aforementioned pathways, which are up-regulated when ER is blocked, can work coordinately with ER to stimulate BC growth and survival. Numerous kinases phosphorylate and activate ER and/or its co-activators resulting in an altered ER cistrome and a new transcriptional program that often resembles that of the growth factors themselves. Recognition of this complex crosstalk has resulted in several new strategies to reverse endocrine therapy resistance using drug combinations to block compensatory pathways. Exciting new strategies include significant down-regulation of ER itself or its co-activator proteins or interfering with altered DNA binding and gene expression. Our understanding ER function is still incomplete despite more than 50 years of study.

3s

4s

Plenary Session

2016 ASCO Annual Meeting Proceedings

ABSTRACTS American Society of Clinical Oncology 52nd Annual Meeting June 3-7, 2016 McCormick Place Chicago, Illinois

LBA1

Plenary Session, Sun, 1:00 PM-4:00 PM

LBA2

Plenary Session, Sun, 1:00 PM-4:00 PM

A randomized trial (MA.17R) of extending adjuvant letrozole for 5 years after completing an initial 5 years of aromatase inhibitor therapy alone or preceded by tamoxifen in postmenopausal women with early-stage breast cancer. First Author: Paul E. Goss, Massachusetts General Hospital Cancer Center and Harvard Medical School, Boston, MA

A phase III randomized controlled trial of short-course radiotherapy with or without concomitant and adjuvant temozolomide in elderly patients with glioblastoma (CCTG CE.6, EORTC 26062-22061, TROG 08.02, NCT00482677). First Author: James R. Perry, Sunnybrook Health Sciences Centre, Toronto, ON, Canada

The full, final text of this abstract will be available at abstracts.asco.org at 7:30 AM (EDT) on Sunday, June 5, 2016, and in the Annual Meeting Proceedings online supplement to the June 20, 2016, issue of the Journal of Clinical Oncology. Onsite at the Meeting, this abstract will be printed in the Sunday edition of ASCO Daily News.

The full, final text of this abstract will be available at abstracts.asco.org at 7:30 AM (EDT) on Sunday, June 5, 2016, and in the Annual Meeting Proceedings online supplement to the June 20, 2016, issue of the Journal of Clinical Oncology. Onsite at the Meeting, this abstract will be printed in the Sunday edition of ASCO Daily News.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Plenary Session LBA3

Plenary Session, Sun, 1:00 PM-4:00 PM

LBA4

5s Plenary Session, Sun, 1:00 PM-4:00 PM

A phase III randomized clinical trial (RCT) of tandem myeloablative autologous stem cell transplant (ASCT) using peripheral blood stem cell (PBSC) as consolidation therapy for high-risk neuroblastoma (HR-NB): A Children’s Oncology Group (COG) study. First Author: Julie R. Park, Seattle Children’s Hospital and University of Washington School of Medicine, Seattle, WA

Phase III randomized controlled study of daratumumab, bortezomib, and dexamethasone (DVd) versus bortezomib and dexamethasone (Vd) in patients (pts) with relapsed or refractory multiple myeloma (RRMM): CASTOR study. First Author: Antonio Palumbo, University of Torino, Torino, Italy

The full, final text of this abstract will be available at abstracts.asco.org at 7:30 AM (EDT) on Sunday, June 5, 2016, and in the Annual Meeting Proceedings online supplement to the June 20, 2016, issue of the Journal of Clinical Oncology. Onsite at the Meeting, this abstract will be printed in the Sunday edition of ASCO Daily News.

The full, final text of this abstract will be available at abstracts.asco.org at 7:30 AM (EDT) on Sunday, June 5, 2016, and in the Annual Meeting Proceedings online supplement to the June 20, 2016, issue of the Journal of Clinical Oncology. Onsite at the Meeting, this abstract will be printed in the Sunday edition of ASCO Daily News.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

6s

Special Clinical Science Symposia

100

Clinical Science Symposium, Sat, 8:00 AM-9:30 AM

101

Clinical Science Symposium, Sat, 8:00 AM-9:30 AM

Checkmate 032: Nivolumab (N) alone or in combination with ipilimumab (I) for the treatment of recurrent small cell lung cancer (SCLC). First Author: Scott Joseph Antonia, Moffitt Cancer Center, Tampa, FL

A phase Ib dose escalation study of the OX40 agonist MOXR0916 and the PD-L1 inhibitor atezolizumab in patients with advanced solid tumors. First Author: Jeffrey R. Infante, Sarah Cannon Research Institute, Nashville, TN

Background: Patients (pts) with advanced (adv) SCLC after first-line platinumbased chemotherapy have limited options. Nivolumab, a programmed death-1 (PD1) immune checkpoint inhibitor, is approved for previously treated metastatic NSCLC in the US and for squamous NSCLC in the EU. Nivolumab + ipilimumab, a cytotoxic T-lymphocyte antigen-4 immune checkpoint inhibitor, has shown durable responses in multiple tumor types. CheckMate 032 was designed to evaluate nivolumab +/- ipilimumab in adv tumors including SCLC. Methods: AdvSCLC pts with progressive disease (PD) after $1 platinum-based chemotherapy, regardless of platinum sensitivity or tumor PD-1 ligand 1 (PD-L1) expression, were eligible. Pts received nivolumab ([mg/kg] N3 Q2W) or nivolumab + ipilimumab combination (N1 + I3 or N3 + I1 Q3W for 4 cycles then N3 Q2W). Primary endpoint was objective response rate (ORR). Additional endpoints were safety, overall survival (OS), progression-free survival (PFS), and biomarkers. Results: 180 pts were enrolled (n=80, N3; n=47, N1 + I3; n=53, N3 + I1). Among 127 pts in the N3 and N1 + I3 cohorts, 56% received $2 prior regimens and 30% were platinum resistant. Efficacy data are shown (Table). Responses were observed independent of platinum sensitivity and PD-L1 expression. Grade 3–4 treatment-related adverse events (TRAEs) occurred in 11% of pts in N3 and 32% of pts in N1 + I3; 5% and 13% discontinued due to TRAEs, respectively. One TR death due to myasthenia gravis occurred (N1 + I3 arm). Conclusions: Nivolumab and nivolumab + ipilimumab showed durable objective responses and manageable safety profiles, with possibly higher toxicities observed with the combination, in previously treated SCLC pts. Updated efficacy including OS by prior lines of therapy, safety, and biomarkers will be presented for the N3 and N1 + I3 cohorts. Efficacy and safety for the N3 + I1 cohort will be presented. Clinical trial information: NCT01928394.

Background: OX40 is a co-stimulatory receptor that is transiently expressed by T cells upon antigen recognition. In murine models, OX40 engagement by an agonist anti-OX40 antibody can promote durable tumor regression associated with co-stimulation of effector T cells and reduction of regulatory T cells. This dual mechanism of action is predicted to complement the activity of PD-L1 blockade. MOXR0916 is a humanized effector-competent agonist IgG1 monoclonal antibody (mAb) that targets OX40 and atezolizumab is an engineered humanized IgG1 mAb that targets PD-L1. Methods: A Phase I, open-label, multicenter study was conducted to evaluate the safety and pharmacokinetics (PK) of MOXR0916 and atezolizumab in patients (pts) with locally advanced or metastatic solid tumors. A 3+3 dose-escalation was conducted with a 21-day window to evaluate doselimiting toxicity (DLT). Escalating doses of MOXR0916 in combination with a fixed 1200 mg dose of atezolizumab were administered every 3 weeks (q3w). An expansion cohort to enable immune profiling of serial tumor biopsies was also enrolled. Prior immunotherapy with adequate washout was allowed if there was no history of Grade (G) $ 3 immune-mediated adverse events (AEs). Results: As of 7 Jan 2016, 25 pts were treated in 7 dose escalation cohorts (dose levels 0.8 to 600 mg) and 19 additional pts were treated in a serial biopsy cohort. The median number of prior therapies for metastatic disease was 2 (range 0-7), and 5 pts had received prior PD-1/PDL1 antibodies. No DLTs, G4/5 AEs attributed to study treatment, or related AEs leading to treatment discontinuation were reported. The majority of treatment-related AEs were G1 in severity; 1 related G3 event (pneumonitis responsive to corticosteroids) was reported. The PK of each mAb was consistent with its established single agent profile. Objective responses were observed; clinical and biomarker data will be updated. The regimen selected for dose expansion is MOXR0916 300 mg + atezolizumab 1200 mg q3w. Conclusions: The combination of MOXR0916 and atezolizumab was welltolerated. An expansion phase, with each agent administered at its recommended monotherapy dose, is ongoing in selected tumor types. Clinical trial information: NCT02410512.

c

ORR, % (n/N) Median DOR, mo Median OS, mo Median PFS, mo 1yr OS rate, %

N3a n=80

N1 + I3b n=47

13 (7/55) Not reached 3.55 1.38 27

31 (14/45) 6.90 7.75 3.35 48

15 pts had ,6 weeks minimum follow-up bMinimumfollow-up = 120 days c25 pts in N3 and 2 pts in N1 + I3 were non-evaluable for tumor response DOR = duration of response

a

102

Clinical Science Symposium, Sat, 8:00 AM-9:30 AM

Rate of durable complete response in ALL, NHL, and CLL after immunotherapy with optimized lymphodepletion and defined composition CD19 CAR-T cells. First Author: Cameron John Turtle, Fred Hutchinson Cancer Research Center, Seattle, WA Background: We are conducting a phase I/II trial in which patients (pts) with relapsed or refractory (R/R) CD19+ B cell malignancies receive CD19 CART cells comprised of a defined ratio of CD8+ and CD4+ CAR-T cells. Methods: Pts received lymphodepleting chemotherapy followed by infusion of a 1:1 ratio of CD8+:CD4+ CAR-T cells at one of 3 dose levels (2x105 to 2x107 CAR-T cells/kg). No pts were excluded due to lymphopenia (ALC , 0.5 in 22% of pts), circulating tumor or poor in vitro T cell proliferation. Results: Ninety pts with ALL (n = 36), NHL (n = 41) or CLL (n = 13) received CAR-T cells. Thirty-one of 33 ALL pts (94%) that have had restaging achieved CR by marrow flow cytometry. Two of 31 pts had molecular MRD. In vivo CAR-T cell expansion and persistence, and DFS were superior in pts who received cyclophosphamide (Cy) and fludarabine (Flu) lymphodepletion compared to those who received Cy without Flu. CAR-T cell persistence was limited in patients not receiving Flu by immune responses to the murine scFv of the CAR. In NHL, the ORR for pts treated with Cy/Flu and CAR-T cells at the maximally tolerated cell dose (n = 19; 2x106/kg) was 84% (47% CR). CAR-T cell expansion and persistence, and OS and PFS were better in Cy/Flu pts compared to those who received Cy alone (ORR 50%, CR 8%, n = 12). Remissions are ongoing in 9 of 10 Cy/Flu treated pts who achieved CR (range 1-9 months follow-up). All CLL pts had received prior ibrutinib. CAR-T cell therapy eliminated CLL from the marrow in 10 of 12 restaged pts (83%) and 6 (50%) achieved CR (Lugano criteria). None of the CLL pts who achieved CR has relapsed. In 87 pts evaluated for toxicity the rates of severe cytokine release syndrome (sCRS) and grade $ 3 neurotoxicity were 16% and 31%, respectively. Four pts (4%) died of complications of CRS and 1 pt experienced irreversible neurotoxicity. We identified serum biomarkers on day 1 after CAR-T cell infusion that correlated with subsequent sCRS or grade $ 3 neurotoxicity in ALL and NHL pts, allowing rational design of early intervention studies in high-risk pts. Conclusions: Adoptive therapy with CD19 CAR-T cells of defined subset composition results in durable CR in a high fraction of pts with R/R ALL, NHL and CLL. Clinical trial information: NCT01865617.

103

Clinical Science Symposium, Sun, 9:45 AM-11:15 AM

Programmed death-1 blockade in mismatch repair deficient colorectal cancer. First Author: Dung T. Le, Johns Hopkins University, Baltimore, MD Background: Mismatch repair deficient (dMMR) colon cancers are densely infiltrated with CD8+T cells and regress when treated with anti-programmed death-1 (PD-1) antibodies. This anti-tumor response is thought to be potentiated by the thousands of somatic mutations that when expressed as proteins result in hundreds of potentially immunogenic neo-antigens that can be recognized by the patient’s immune system. Methods: We previously reported a phase 2 study to evaluate the activity of pembrolizumab (pembro), a PD-1 antibody in treatment refractory dMMR colon cancers (NEJM 2015). We are reporting the expanded trial and updated data for the mismatch repair deficient CRC (dMMR, cohort A) and mismatch repair proficient CRC (pMMR, cohort B) cohorts. Pembro was administered at 10 mg/kg every 14 days in patients with 2 or more prior therapies. The coprimary endpoints were response and progression-free survival rate at 20 weeks. Secondary endpoints included disease control rate (DCR = CR + PR + SD), progression free survival (PFS), overall survival (OS) and safety. Results: A total of 53 patients (Cohort A: n = 28, Cohort B: n = 25) were treated. Median follow up time is 8.7 months. RR and DCR were 50% (14/ 28, 95% confidence interval (CI): 31-69%) and 89% (25/28) for dMMR CRC and 0% (0/25, 95% CI: 0-14%) and 16% (4/25) for pMMR CRC, respectively. Twenty-one of 28 dMMR CRC patients remain on study. Median PFS was not reached (NR) for dMMR CRC and 2.4 months for pMMR CRC (HR = 0.135; 95% CI 0.043 to 0.191; p=,0.0001). Median OS was NR for dMMR vs. 6 months for pMMR (HR = 0.247; 95% CI 0.117 to 0.589; p=0.001). For dMMR CRC, the PFS rates was 61% at 24 months and the OS rate was 66% at 24 months. Among patients with an objective response (n=14), only 1 has developed secondary resistance to pembro with a solitary brain metastasis after 4.6 months of therapy. Conclusions: Patients with dMMR CRC receive durable clinical benefit with pembrolizumab. Clinical trial information: NCT01876511.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Special Clinical Science Symposia 104

Clinical Science Symposium, Sun, 9:45 AM-11:15 AM

PD-L1 expression, Cancer Genome Atlas (TCGA) subtype, and mutational load as independent predictors of response to atezolizumab (atezo) in metastatic urothelial carcinoma (mUC; IMvigor210). First Author: Jonathan E. Rosenberg, Memorial Sloan Kettering Cancer Center, New York, NY Background: Biomarkers of clinical benefit of immunotherapies, such as non-synonymous mutations, mutational load, PD-L1 IHC and immune signatures, have not been systematically evaluated in mUC patients (pts) treated with anti–PD-L1 agents. Methods: Exploratory analyses of immune and genetic predictors of response to atezo were performed on archival tumor specimens from mUC pts (NCT02108652 cohort 2). PD-L1 expression in tumor samples was assessed by IHC (SP142 assay) and associated with response as determined by RECIST v1.1 (central review). Gene expression was quantified by RNA-seq (Illumina TruSeq RNA Access; n = 195); molecular subtypes were assigned as per TCGA. Mutations were detected by genomic profiling (FoundationOne 315-cancer gene panel; n = 150) to estimate overall load. Results: PD-L1 tumor-infiltrating immune cell (IC) status was associated with CD8+ T effector (Teff) gene expression levels (P , 0.0001). Notably, response to atezo was associated with high expression of CXCL9 (P = 0.0057) and CXCL10 (P = 0.0079). Evaluable samples were classified into luminal (n = 73) and basal (n = 122) TCGA subtypes. PD-L1 IC prevalence was enriched in basal (60%) vs luminal (23%) samples (P , 0.0001); elevated tumor cell PD-L1 was mostly restricted to the basal subtype (39% basal vs 4% luminal, P , 0.0001) and did not correlate with ORR. Teff expression was elevated in luminal cluster II and basal III/IV and not in luminal I. Responses occurred in all subtypes, but ORR was significantly higher in luminal II samples: 34% (P = 0.0017). While median mutational load was significantly increased in responders (12.4/Mb) vs nonresponders (6.4/Mb; P , 0.0001) and correlated with both PFS (P = 0.003) and OS (P = 0.014), these associations were independent of TCGA subtype. Conclusions: PD-L1 expression on IC was associated with response to atezo. Intrinsic TCGA subtype and mutational load each added predictive value. Simultaneous classification by these characteristics in mUC pts treated with atezo further defines the drivers of immune responsiveness and may suggest potential combination strategies. Clinical trial information: NCT02108652.

105

7s Clinical Science Symposium, Sun, 9:45 AM-11:15 AM

Hybrid capture-based next-generation sequencing (HC NGS) in melanoma to identify markers of response to anti-PD-1/PD-L1. First Author: Douglas Buckner Johnson, Vanderbilt University, Nashville, TN Background: Agents blocking programmed death-1/ligand (aPD1) induce durable responses in many melanoma patients (pts). Greater numbers of somatic mutations have been linked to high aPD1 response rates (RR) in other cancers. We assessed whether number and/or type of mutations (mut) using HC NGS correlated with outcomes to aPD1 in melanoma. Methods: HC NGS on 236-315 genes was performed on archival samples from pts who received aPD1 (comprising ~1.25 megabases [MB]). A novel algorithm extrapolated genomic mutational load (ML) from this panel. Intratumoral T cell clonal expansion was determined by T cell receptor (TCR) sequencing. ML and TCR clonality were correlated with with aPD1 outcomes. Results: Responders to aPD1 had higher ML compared to non-responders in discovery (median 45.6 vs. 3.9 mut/MB; p=0.003, n=32), and validation cohorts (37.1 vs. 12.8 mut/MB; p=0.002, n=33). RR was superior in high ML pts compared to intermediate/low (Table) as was progression-free survival (PFS) (median not reached [MNR] vs. 89 days vs. 86 days, p,0.001) and overall survival (OS) (MNR vs. 300 days vs. 375 days, p,0.001). Responders had more C.T transitions than non-responders (median 33.5 vs. 3.0, p,0.001). Melanomas with NF1 mutations had high ML (median 62.7 mut/MB) and high RR (74%); whereas BRAF mutant (12.0 mut/MB), NRAS mutant (17.6 mut/MB), and BRAF/NRAS/NF1wild-type (2.2 mut/MB) melanomas had lower median ML. TCR clonality did not correlate with response in these archival samples, nor did PD-L1 gene amplification (,1% incidence). Conclusions: ML as determined by HC NGS effectively stratified patients by likelihood of response. This approach may provide a widely available, clinically feasible predictor of response to aPD1. Cohort

Mutational load group

Response

No Response

P value

Discovery

High (n=11) Intermediate (n=11) Low (n=10) High (n=16) Intermediate (n=13) Low (n=4) High (n=27) Intermediate (n=24) Low (n=14)

9 (82%) 4 (36%) 1 (10%) 14 (88%) 3 (23%) 1 (25%) 23 (85%) 7 (29%) 2 (14%)

2 (18%) 7 (64%) 9 (90%) 2 (12%) 10 (77%) 3 (75%) 4 (15%) 17 (71%) 12 (86%)

0.003

Validation Combined

0.001 23.1 mut/MB; Intermediate: 3.2 – 23.1 mut/MB; Low: 3 P Mutation Wild type

2007

16.7 12.95

19.1 8.10

23.7 7.02

0.26 0.004

Oral Abstract Session, Sun, 8:00 AM-11:00 AM

An independently validated nomogram for individualized estimation of survival among patients with newly diagnosed glioblastoma: NRG oncology/RTOG 0525 and 0825. First Author: Haley R Gittleman, Case Western Reserve University, Cleveland, OH Background: Glioblastoma (GBM) is the most common primary malignant brain tumor. Nomograms are often used for individualized estimation of prognosis. This approach has been applied to GBM only once before. This study aimed to build and independently validate a nomogram to estimate individualized survival probabilities for patients with newly diagnosed GBM, using data from two independent NRG Oncology/Radiation Therapy Oncology Group (RTOG) clinical trials. Methods: This analysis included information on 799 (0525) and 555 (0825) eligible and randomized patients with newly diagnosed GBM and contained the following variables: age at diagnosis, race, gender, Karnofsky Performance Status (KPS), extent of resection, O6-methylguanine-DNA methyltransferase (MGMT) methylation status, and survival (months). Survival was assessed using Cox proportional hazards (CPH) regression, random survival forests (RSF), and recursive partitioning analysis (RPA), with adjustment for known prognostic factors. The models were developed and trained using the 0525 data and were independently validated using the 0825 data. Models were internally validated using 10-fold cross-validation, and individually predicted 6-, 12-, and 24-month survival probabilities were generated to measure the predictive accuracy and calibration against the actual survival status. Results: A final nomogram was built using the CPH model. Factors that increased the probability of shorter survival included greater age at diagnosis, male gender, lower KPS, not having total resection, and unmethylated MGMT status, all of which are consistent with observations in clinical practice. Conclusions: A nomogram that assesses individualized survival probabilities (6-, 12-, and 24-month) for patients with newly diagnosed GBM could be useful to healthcare providers for counseling patients regarding treatment decisions and optimizing therapeutic approaches. Free software for implementing this nomogram is provided via an internet interface: http://cancer4.case.edu/ rCalculator/rCalculator.html.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Central Nervous System Tumors 2008

Oral Abstract Session, Sun, 8:00 AM-11:00 AM

Phase 0 trial of AZD1775 in patients with first-recurrence glioblastoma. First Author: Nader Sanai, Barrow Neurological Institute, Phoenix, AZ Background: AZD1775 is a first-in-class Wee1 inhibitor with dual-function as a DNA damage sensitizer (via Cdk1-mediated G2-checkpoint arrest) and as a cytotoxic agent (via Cdk2-mediated DNA-destabilization). A Phase I study of AZD1775 for solid tumors suggested activity against brain tumors, but a preclinical study indicated minimal blood-brain-barrier penetration in mice. To resolve this controversy, we examined the pharmacokinetic (PK) and pharmacodynamics (PD) of AZD1775 in first-recurrence, de novo glioblastoma (GBM) patients. Methods: 20 adult patients received a single dose of AZD1775 prior to brain tumor resection and enrolled in either a doseescalation arm (100, 200, or 400 mg) or a time-escalation arm at 400 mg (4, 8-, or 24-hrs). Sparse PK blood sampling was performed. Tumor samples were collected intraoperatively and AZD1775 concentrations in plasma and tumor were determined by a validated LC-MS/MS method. PD endpoints were compared to matched archival tissue. Expression of efflux and uptake transporters in tumor samples were determined by qRT-PCR. Results: AZD1775 plasma exposure increased with increasing dose. AZD1775 tumor concentrations exhibited large inter-individual variability. At all doses and times, AZD1775 tumor concentrations (range, 98 – 4926 ng/g) were higher than the in vitro IC50 (42 ng/mL) for inhibition of cellular Wee1 activity. The tumor-to-plasma ratio was dose- and time-independent, ranging from 3.8 to 40.4 (mean, 11.7 6 8.5). Tumor-to-plasma ratios appeared to be associated with tumor mRNA levels of transporters ABCC1, ABCC3, and OATP1A2. Wee1 pathway suppression was inferred by abrogation of G2-arrest (phosphohistone-3, p= 0.0085) and intensified doublestrand DNA breakage (gamma-H2AX, p= 0.003) following drug exposure. No drug-related adverse events were associated with this study. Conclusions: Within 6 months of study activation, we obtained data to justify a Phase II trial for AZD1775 in GBM. Our findings show good CNS penetration for AZD1775 in humans (in contrast to results in mice), provide the first evidence of clinical biological activity in human GBM, and confirm the utility of Phase 0 trials as part of an accelerated paradigm for early drug development in brain tumor patients. Clinical trial information: NCT02207010.

2010

Clinical Science Symposium, Mon, 3:00 PM-4:30 PM

2009

95s Clinical Science Symposium, Mon, 3:00 PM-4:30 PM

Predicting isocitrate dehydrogenase genotype in malignant glioma with multimodality imaging markers. First Author: Biqi Zhang, Harvard Medical School, Boston, MA Background: Malignant gliomas with mutations in the isocitrate dehydrogenase (IDH) gene family confer longer overall survival relative to their IDH wild-type counterparts. Accurately determining IDH genotype preoperatively may have both prognostic and diagnostic value that contributes to treatment decision-making. The current study used a machinelearning algorithm to generate a model predictive of IDH genotype based on multi-modal imaging and clinical features. Methods: Preoperative magnetic resonance images were obtained for 120 patients with primary grade III (n = 35) and IV (n = 85) glioma in this retrospective study. IDH genotype was confirmed for grade III (32/35, 91%) and IV (22/85, 26%) tumors by immunohistochemistry, spectrometry-based mutation genotyping (OncoMap), or multiplex exome sequencing (OncoPanel). Patients were randomly assigned to either the training (n = 90) or validation cohort (n = 30). A total of 2970 imaging features were extracted from pre- and post-contrast T1-weighted, T2-weighted, and Apparent Diffusion Coefficient map. Using a random forest algorithm, non-redundant features were integrated with clinical data to generate a predictive model for IDH genotype. Results: IDH genotype prediction using our model achieved an accuracy of 86% (specificity = 90%, sensitivity = 83%) in the training cohort and 89% (sensitivity = 84%, specificity = 89%) in the validation cohort. The misclassification error using ten-fold cross-validation in the training set was 17%. Features with the highest predictive value included patient age as well as parametric intensity, texture, and shape features. Conclusions: Using a machine-learning algorithm, we achieved accurate prediction of IDH genotype in malignant gliomas with preoperative MRI and clinical features.

2011

Clinical Science Symposium, Mon, 3:00 PM-4:30 PM

Safety of pembrolizumab in combination with bevacizumab in recurrent glioblastoma (rGBM). First Author: David A. Reardon, Dana-Farber Cancer Institute/ Harvard Cancer Center, Boston, MA

Cost and resource utilization associated with glioblastoma among commercially insured adults in the United States (US). First Author: Shan Jiang, Pharmerit International, Bethesda, MD

Background: Inhibition of vascular endothelial growth factor (VEGF) may enhance immunotherapies and decrease cerebral edema among GBM patients (pts). This multicenter, randomized phase 2 study determined safety and anti-tumor activity of pembrolizumab (P), a PD-1 blocking monoclonal antibody (Mab) with and without bevacizumab (Bev), a humanized antiVEGF Mab among rGBM pts. Herein, we report the safety lead-in of this study. Methods: Pts in dose level 1 received P 200 mg Q3W and Bev 10 mg/ kg Q2W. Dose de-escalation levels -1 and -2 increased the P interval to 4 and 6 weeks, respectively if dose level 1 exceeded maximum tolerated dose (MTD) based on standard 3+3 criteria. Eligible pts were $ 18 years, had 1or 2 recurrences, no prior anti-VEGF treatment, and KPS $ 70. The primary endpoint was MTD/recommended phase 2 dose (RP2D) determination based on dose limiting toxicity (DLT) during cycle 1. MRI scans were obtained after each 6-week cycle. Results: 6 pts were treated (4 male, 2 female) at1st (n = 2) or 2nd (n = 4) recurrence. Median age was 48 years (range: 27-79), and KPS = 90 (n = 3), 80 (n = 2) and 70 (n = 1). Median time from initial GBM diagnosis was 11 mo (range 4-61). Median duration on study was 3 mo (range 1-9.3). The most common adverse events that were at least probably related to study therapy were fatigue (grade 2, n = 2; grade 1, n = 2) and hypertension (grade 3, n = 2; grade 1, n = 1). No pt experienced a DLT and none required dose modification, interruption or discontinuation due to toxicity although Bev was electively held in 1 pt for a dental procedure. All pts discontinued study therapy due to progressive disease (PD). Best radiographic response was partial response (n = 1), stable disease (n = 2) and PD (n = 3). Two pts remain alive (327 and 328 days) while 4 have died. Median overall survival is 6.8 months. Accrual to the randomized phase 2 part of the study recently completed including 30 pts treated with P and 39 treated with P+Bev. Conclusions: This is the first study to report safety/tolerability of combination PD-1 plus VEGF blockade in GBM. P and Bev were safely combined when administered according to approved dosing schedules. No DLT or unexpected toxicity was encountered. Clinical Trial NCT02337491; Support – The Ben and Catherine Ivy Foundation Clinical trial information: NCT02337491.

Background: Few studies have evaluated the economic burden of GBM—the most common type of high-grade brain tumors in adults. We assessed healthcare costs and resource utilization in newly-diagnosed GBM patients in a commercially-insured US population. Methods: The analysis included subjects identified in the IMS Pharmetrics Lifelink Plus claims database from 01/2009 to 03/2014 who were aged $18 years; had $1 malignant brain cancer diagnosis (ICD9-CM, 191.xx) (with first diagnosis date = “index” date); were treated with temozolomide (TMZ); had brain surgery #90 days prior to TMZ initiation; and were continuously enrolled for $12 months pre-index and $1 month post-index. Per-patient per-month (PPPM) costs (i.e., average allowed charges for all claims) and cumulative costs from 3 months pre- to up to 6, 12, or 60 months post-index were calculated. Results: 2,981 patients were included (median age: 56 years; 60% male). During the post-index period, the per-patient median (interquartile range, IQR) number of inpatient, emergency department (ED), and outpatient visits were 2 (1-5), 2 (1-3), and 22 (12-31), respectively; the median (IQR) length of stay was 5 (3-10) days. Table 1 presents cumulative per-patient costs by key categories and maximum follow-up time. Excluding TMZ, the 3 most costly medication classes prescribed (on PPPM-basis) were anticoagulants ($515), antidiabetic agents ($84), and anticonvulsants ($72). Conclusions: The economic burden of newly-diagnosed GBM is substantial, with costs exceeding $200,000 within a year post-diagnosis. Key cost drivers include inpatient care, chemotherapy, and radiation costs. Cumulative per-patient costs in US$ (% total) by maximum length of follow-up. 3-months pre-index to 3-months pre-index to 3-months pre-index to Cost category 6 months post-index 12 months post-index 60 months post-index Total costs $165,848 (100%) Inpatient $75,604 (46%) Chemotherapy $24,545 (15%) Radiation $32,463 (20%) ED $1,846 (1%) Others $31,390 (19%)

$212,331 $87,326 $43,761 $33,364 $2,383 $45,497

(100%) (41%) (21%) (16%) (1%) (21%)

$311,804 (100%) $114,563 (37%) $78,425 (25%) $37,775 (12%) $3,352 (1%) $77,689 (25%)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

96s 2012

Central Nervous System Tumors Poster Discussion Session; Displayed in Poster Session (Board #201), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

A first-in-human phase 1 study to evaluate the brain-penetrant PI3K/mTOR inhibitor GDC-0084 in patients with progressive or recurrent high-grade glioma. First Author: Patrick Y. Wen, Dana-Farber Cancer Institute, Boston, MA Background: GDC-0084 is a potent, oral, selective, brain-penetrant small molecule inhibitor of phosphoinositide 3-kinase (PI3K) and mammalian target of rapamycin (mTOR) kinase. A first-in-human, phase 1 dose escalation study was conducted in patients with high-grade glioma. Methods: GDC-0084 was administered orally, once daily on a continuous dosing schedule to evaluate the safety and pharmacokinetic (PK) characteristics. Plasma samples for PK analysis were collected on Day 1 and Day 8 or Day 15 of Cycle 1. Fluorodeoxyglucose positron emission tomography (FDG-PET) was performed at baseline and on-treatment. Results: 47 pts enrolled in eight successive dose escalation cohorts (2-65 mg). Doselimiting toxicities (DLTs) reported were one case of Gr 2 bradycardia and Gr 3 myocardial ischemia (15 mg), Gr 3 stomatitis (45 mg) and two cases of Gr 3 mucosal inflammation (65 mg); the MTD was 45 mg. The most frequent adverse events (AE) attributed to GDC-0084 were fatigue, hyperglycemia, nausea, hypertriglyceridemia, rash, hypophosphatemia, decreased appetite, diarrhea, and stomatitis. PK data showed linear and dose proportional increases in exposure, with a half-life supportive of once daily dosing (t1/2 ~19 hr). At a dose of 45 mg steady-state concentrations were consistent with antitumor activity observed in xenograft models. On FDG-PET, 5 of 27 (18.5%) evaluable patients show a metabolic PR. Additionally, at doses of $45 mg QD, a trend towards decreased median SUV in normal brain was observed, suggesting CNS penetration of the study drug. In two available exploratory specimens, GDC-0084 was detected at similar levels in brain tumor and brain tissue, with a brain tissue/tumor to plasma ratio of .1 and .0.5 for total and free drug, respectively. Of the evaluable patients, 26 patients (55.3%) had a best overall response of progressive disease, 19 patients (40.4%) had stable disease. Conclusions: The safety profile demonstrated classic Pi3K/mTOR-inhibitor related AEs. FDG-PET and concentration data from brain tumor tissue suggest that GDC-0084 crosses the blood brain barrier, with a uniform distribution throughout the brain. Clinical trial information: NCT01547546.

2014

Poster Discussion Session; Displayed in Poster Session (Board #203), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

2013

Poster Discussion Session; Displayed in Poster Session (Board #202), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

A phase I study of MLN0128 and bevacizumab in patients with recurrent glioblastoma and other solid tumors. First Author: Lakshmi Nayak, DanaFarber Cancer Institute, Boston, MA Background: The P13K/Akt/mTOR signaling axis plays a central role in cell growth and survival in a variety of cancers. MLN0128 is an oral TORC1/2 inhibitor. Bevacizumab is a monoclonal antibody targeting VEGF with activity in some cancers, including glioblastoma (GBM). Methods: This ongoing Cancer Therapeutic Evaluation Program (CTEP)-sponsored phase I trial uses a 3+3 dose escalation design to determine the maximum tolerated dose (MTD/recommended phase 2 dose (RP2D) of MLN0128 when administered daily with bevacizumab (10 mg/kg every 2 weeks) and includes a dose expansion at the MTD/RP2D for recurrent GBM (n=10) and recurrent endometrial/ovarian cancer (n=30) patients (pts). Eligibility includes histologically confirmed recurrent GBM or other advanced solid tumors, .18 years old, KPS. 60, adequate bone marrow and organ function. The starting dose of MLN0128 was 3 mg daily. Results: Thirty-six pts have enrolled (12 in dose escalation and 24 in dose expansion). The median age and KPS were 56 (range: 25-74) and 80 (range 70-100), and 25 were women. Histologic diagnosis included GBM (n=16), ovarian (n=14) and endometrial (n=6) cancer. Three MLN0128 dose levels were evaluated including 3,4 and 5 mg daily. Only 1 dose limiting toxicity (DLT) occurred (grade 3 fatigue at the 5 mg dose level). There were 31 additional related grade 3 events and 1 grade 4 event. Most common toxicities were rash, hypercholesterolemia, gastrointestinal symptoms, lymphopenia and fatigue. Median number of completed cycles was 2 (range, 1-17). Sixteen pts came off for disease progression, 8 withdrew, 2 patient/physician’s decision, 3 toxicity, 1 other; 6 remain on treatment. Best response to treatment among GBM pts was complete response (n=1), partial response (PR, n=3), stable disease (SD, n=6). Three ovarian cancer pts achieved PR and 8 SD, while 2 endometrial cancer pts achieved SD. Conclusions: The RP2D of daily MLN0128 is 5mg when combined with bevacizumab. The dose expansion phase continues to enroll. Clinical trial information: NCT02142803.

2015

Poster Discussion Session; Displayed in Poster Session (Board #204), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Safety and activity of nivolumab (nivo) monotherapy and nivo in combination with ipilimumab (ipi) in recurrent glioblastoma (GBM): Updated results from checkmate-143. First Author: David A. Reardon, Dana-Farber Cancer Institute and Harvard University School of Medicine, Boston, MA

Validation of the new glioma WHO classification in the french POLA network: Analysis of 1041 cases. First Author: Emeline Tabouret, Aix-Marseille University, AP-HM, Service de Neuro-Oncologie, CHU Timone, Marseille, France

Background: Patients (pts) with recurrent/progressive GBM have a poor prognosis and limited therapeutic options. The phase 1 CheckMate-143 study evaluates safety and tolerability of the anti-PD-1 monoclonal antibody nivo, alone or in combination with the anti-CTLA-4 antibody ipi, in pts with recurrent GBM. Preliminary results from a phase 1 safety cohort (Cohort 1, Methods) were presented previously. Herein, we present updated safety, overall survival (OS) data, and PD-L1 biomarker status for the completed phase 1 Cohorts 1 and 1b. Methods: Eligible pts had a first recurrence of GBM after radiation and temozolomide. In Cohort 1, 20 pts were randomized 1:1 to receive nivo 3 mg/kg (N3) every 2 weeks (Q2W) or nivo 1 mg/kg + ipi 3 mg/kg every 3 weeks (Q3W; N1+I3) for 4 doses followed by N3 Q2W. Pts in Cohort 1b (n = 20) received nivo 3 mg/kg + ipi 1 mg/kg Q3W (N3+I1) for 4 doses followed by N3 Q2W. Treatment continued until disease progression or unacceptable toxicity. PD-L1 status was determined by immunohistochemistry. Results: Accrual to Cohorts 1 and 1b was completed (N = 40). Median ages were 59 (range: 42–73), 55 (37–68), and 55 (27–73) years in the N3, N1+I3, and N3 +I1 arms, respectively. Median Karnofsky performance status (N = 40) was 90 (range: 70–90). Treatment-related adverse events (TRAEs) and treatment-related serious AEs (TRSAEs) are summarized (Table). Three pts on N1 + I3, 1 pt on N3 + I1, and 0 pts receiving N3 discontinued for TRAEs. Stable disease or better was achieved in 6/10, 4/ 10, and 9/20 pts receiving N3, N1+I3, and N3+I1, respectively. The 12-month OS was 40% (95% CI: 12–67) for N3, 30% (95% CI: 7–58) for N1+I3, and 25% (95% CI: 8–48) for N3+I1. PD-L1 expression was quantifiable in 34 pts (85%). Conclusions: Tolerability profiles in pts receiving nivo and nivo + ipi are consistent with observations in other tumor types, with no new safety signals. N3 was the best tolerated, and N3+I1 was better tolerated than N1+I3. Twelve-month OS rates are encouraging. Clinical trial information: NCT02017717.

Background: The upcoming WHO classification of gliomas will be refined taking into account the robust molecular alterations of gliomagenesis: the 1p19q codeletion and IDH1/2mutations. Methods: All cases of high-grade oligodendroglial tumors sent for central pathological review and included into the French nation-wide POLA network were reclassified according the upgrading of the 4th WHO classification that integrates pathological features and molecular data. Immunohistochemical expression of IDH1R132H, and ATRX, 1p19q codeletion, chromosome 7p gain and chromosome 10q loss by genomic-array and IDH1/2and histone K27M mutational statuses by sequencing were evaluated. Results: 1041 patients were included with a median age at diagnosis of 50.4 years (range, 17.1-84.4). Based on the new histo-molecular classification, diagnoses were: anaplastic oligodendroglioma with 1p19q codeletion (32.5%), anaplastic astrocytoma IDH mutated (IDHmt) (11.0%), anaplastic astrocytoma IDHwild-type (IDHwt) (5.3%), glioblastoma IDHmt (17.1%) and glioblastoma IDHwt (33.2%). Ten patients presented a midline tumor with the histone H3-K27M mutation. Both new and oldest WHO classifications were predictive for ProgressionFree Survival (PFS) and Overall Survival (OS) but the new histo-molecular classification was more discriminant with higher hazard ratio for PFS (1.603, 1.848, 5.025 and 6.135 versus 1.495, 2.705, 2.799 and 4.933) and for OS (3.588, 3.493, 11.020 and 14.708 versus 2.259, 4.340, 4.214 and 7.534). Grading (III versus IV) was not prognostic for IDHmt non 1p19q codeleted gliomas in univariate analyses (PFS, p=0.5; OS, p=0.8) and multivariate analysis (adjusted by age, type of surgery and first line treatment: PFS, p=0.13; OS, p=0.38). Among anaplastic astrocytoma IDHwt, cases presenting with 7p gain and 10q loss (55%) had a worse prognosis that the others for PFS (p=0.027), suggesting that all anaplastic astrocytoma IDHwt should not be considered as glioblastoma IDHwt. Conclusions: WHO histo-molecular classification of gliomas presented with high predictive and discriminative value, allowing the validation of three main molecular subgroups for the future neuro-oncological trials.

Cohort 1 TRAEs TRSAEs

Cohort 1b

AEs, n (%)

N3 (n = 10)

N1+I3 (n = 10)

N3+I1 (n = 20)

Any grade Grade 3–4 Grade 5 Any grade Grade 3–4 Grade 5

9 (90) 0 0 2 (20) 0 0

10 (100) 9 (90) 0 7 (70) 7 (70) 0

20 (100) 5 (25) 0 5 (25) 2 (10) 0

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Central Nervous System Tumors 2016

Poster Discussion Session; Displayed in Poster Session (Board #205), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Comprehensive mutation analysis in NRG Oncology/RTOG 9813: A phase III trial of RT + TMZ vs RT + nu for anaplastic astrocytoma and mixed anaplastic oligoastrocytoma (Astrocytoma Dominant). First Author: Erica Hlavin Bell, The Ohio State University, Columbus, OH

2017

97s

Poster Discussion Session; Displayed in Poster Session (Board #206), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Comprehensive mutation analysis in NRG Oncology/RTOG 9802: A phase III study of RT vs RT + PCV in high-risk low-grade gliomas (LGGs). First Author: Erica Hlavin Bell, The Ohio State University, Columbus, OH

Background: This study sought to examine the prognostic value of mutations within IDH1/2, ATRX, CIC, FUBP1, and the TERT promoter in a phase III study of grade III astrocytomas using multivariate analyses (MVAs). Methods: IDH mutations were determined by immunohistochemistry and/or next-generation sequencing. A custom Ion AmpliSeq DNA panel was used for mutation analysis; TERT promoter mutations were detected by Sanger sequencing. Progression-free survival (PFS) and overall survival (OS) were estimated using the Kaplan-Meier method. Hazard ratios (HRs) were calculated and tested using the Cox proportional hazard model and the log-rank test. MVAs were performed including age, treatment, surgery, and performance status as covariates. Results: Mutations within IDH1/2 were found in 55% (57/104), ATRX in 46% (28/61), the TERT promoter in 20% (11/55), CIC in 10% (6/61), and FUBP1 in 2% (1/61) of analyzed cases. In the univariate analyses on OS, IDH1/2 (HR=0.41; p,0.001) and ATRX mutations (HR=0.45; p=0.02) significantly correlated with improved OS, and TERT promoter mutations (HR=1.76; p=0.16) trended toward worse OS; on PFS, IDH1/2 (HR=0.50; p=0.004) and ATRX mutations (HR= 0.41; p=0.009) were significantly associated with improved PFS, and TERT promoter mutations (HR=2.0; p=0.08) trended toward worse PFS. Upon MVA, ATRX mutations were significantly associated with OS (HR=0.48; p=0.04) and PFS (HR=0.42; p=0.02), as were IDH1/2 mutations (OS: HR=0.49, p=0.01; PFS: HR=0.53, p=0.02). Conclusions: This study underscores the prognostic significance of IDH and ATRX. Importantly, this is the first study to examine the prognostic value of these mutations using rigorous MVA in a grade III astrocytoma cohort with prospectively-collected, well-annotated clinical data. Efforts to increase sample size and to perform predictive analyses are ongoing. Funding:U10CA21661, U10CA180868, U10CA180822, U10CA37422 (NCI), Bristol Myers Squibb and Merck &Co. Also, R01CA108633, R01CA169368, RC2CA148190, U10CA18085001 (NCI), Brain Tumor Funders Collaborative Grant, and The Ohio State University CCC (all to AC). Clinical trial information: NCT00004259.

Background: This study sought to examine the prognostic significance of mutations within IDH1/2, ATRX, CIC, FUBP1, and the TERT promoter in a prospective phase III study of high-risk LGG using multivariate analyses (MVAs). Methods: IDH mutation status was determined by immunohistochemistry and/or deep sequencing. A custom Ion AmpliSeq DNA panel was used for mutation analysis; TERT promoter mutations were detected by Sanger sequencing. Progression-free survival (PFS) and overall survival (OS) were estimated using the Kaplan-Meier method. Hazard ratios (HRs) were calculated and tested using the Cox proportional hazard model and the logrank test. MVAs were performed incorporating age, treatment, surgery, histology, and performance status as covariates. Results: Observed mutations, by frequencywere 74% (85/115) within IDH1/2, TERT promoter in 40% (32/80), CIC in 25% (17/69), ATRX in 24% (16/67), and FUBP1 in 10% (7/69). In the univariate analysis (UVA) on OS, IDH1/2 (HR = 0.40; p , 0.001), CIC (HR = 0.40; p = 0.04), and TERT promoter mutations (HR = 0.40; p = 0.007) were all statistically significant. For PFS, by UVA, IDH1/2 mutations (HR = 0.45; p = 0.003) were significantly associated with better outcome, and TERT promoter (HR = 0.62; p = 0.11) and CIC mutations (HR = 0.56; p = 0.12) trended toward better PFS. Upon MVA, TERT promoter (HR = 0.41; p = 0.04) and CIC mutations (HR = 0.23; p = 0.01) were significantly associated with OS, whereas IDH1/2 mutations were significantly associated with PFS (HR = 0.46; p = 0.005). Conclusions: This study underscores the prognostic value of TERT promoter, IDH, and CIC mutations and suggests that TERT and CIC may provide additional information beyond IDH. Most importantly, this is the first study to examine the prognostic effects of these mutations using rigorous MVA in a grade II cohort with prospectively-collected, well-annotated clinical data. Efforts to increase sample size and to perform predictive analyses are ongoing. Funding: U10CA21661, U10CA180868, U10CA180822, and U10CA37422 (NCI). Also, R01CA108633, R01CA169368, RC2CA148190, U10CA18085001 (NCI), Brain Tumor Funders Collaborative Grant, and The Ohio State University CCC (all to AC). Clinical trial information: NCT00003375.

2018

2019

Poster Discussion Session; Displayed in Poster Session (Board #207), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Poster Discussion Session; Displayed in Poster Session (Board #208), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Investigation of risk factors associated with fatigue in glioma patients. First Author: Elizabeth Vera, University of Texas Health Science Center-Houston, Houston, TX

Phase II part of EORTC study 26101: The sequence of bevacizumab and lomustine in patients with first recurrence of a glioblastoma. First Author: Wolfgang Wick, Neurology Clinic, Heidelberg, Germany

Background: Fatigue is a consistently reported severe symptom among patients with gliomas throughout the disease trajectory. Clinical variables and genomic pathways associated with fatigue in glioma patients have yet to be identified. Methods: Glioma patients undergoing neuropsychological evaluation at UT MD Anderson Cancer Center with available fatigue and genotyping data were included in this analysis. Clinical factors (performance status, tumor details, age, gender) were collected by chart review. Ratings on the FACT-BR item “I have a lack of energy” were taken within one year of diagnosis. A rating of 3 or 4 defined moderate/severe fatigue. Candidate genes in clock and inflammatory pathways associated with fatigue were identified from literature review. Seventy single nucleotide polymorphisms (SNPs) in 7 genes were available for study. Clinical factors and SNPs identified by univariate analyses were included in a multivariate logistic regression model with backward selection for moderate/severe fatigue. Results: The study included 136 patients (median age=48, 70% males). Moderate/ Severe fatigue was reported by 36%. A final model identified gender, performance status and SNPs rs2253820 and rs934945 as risk factors for moderate/severe fatigue. Female patients and patients with poor KPS were more likely to report fatigue. Each minor allele present for rs2253820 or rs934945 decreased odds of fatigue. These SNPs encode per1 and per2 genes, respectively, two important players of the circadian clock mechanism. No genes in the inflammatory pathway were associated with fatigue in the current study. Conclusions: Identifying patients at highest risk for fatigue during treatment allows for improved clinical monitoring and enrichment of patient selection for clinical trials.

Background: Both lomustine (LOM) and bevacizumab (BEV) are approved agents for recurrent or progressive glioblastoma. We aimed at evaluating the optimal treatment sequence (single agent vs sequential vscombination) of these agents in a randomized phase II design. Methods: Progressive patients after initial radiotherapy and temozolomide were randomized 2:2:2:1 between Arm 1: LOM 90 mg/m2 (max. 160 mg) every 6 weeks plus BEV (10 mg/ kg) every 2 weeks until progression, subsequent salvage treatment at the investigators best choice (MD choice); Arm 2: LOM 110 mg/m2 (max. 200 mg) every 6 weeks, at progression switch to BEV every two weeks; Arm 3: BEV every 2 weeks, at progression add LOM 90 mg/m2 (max. 160 mg) every 6 weeks and continue BEV; and Arm 4: LOM single agent 110 mg/m2 (max. 200 mg) every 6 weeks, at progression MD choice (arm 4). Eligibility criteria included performance status 0-2, adequate hematological, liver and renal function, and an interval since the end of radiotherapy of $ 3 months to rule out pseudoprogression. The primary endpoint was overall survival at 12 months (OS12), progression free survival (PFS) was measured from the start of study treatment until 1st PD. Response was locally assessed using RANO criteria. Results: Between Oct 2011 and July 2013 274 patients were randomized. Baseline characteristics were well balanced, and all treatment generally well tolerated. In arm 4, 43% received subsequent salvage BEV. Time to first progression was longer in initially BEV treated patients, but survival was similar in all arms. Conclusions: Initially BEV treated patients had longer PFS, but OS 12 or median OS were similar in the various schedules of BEV and LOM. Following the emergence of the data of the BELOB study, prior to any analysis arms 1 and 4 of this phase II trial were continued as a phase III trial. Clinical trial information: NCT01290939.

Clinical factors and SNPs on risk of moderate/severe fatigue in glioma patients (N=136)1 Factor B S.E.2 OR3 95% CI4 Gender Female Male Karnofsky Performance Status Poor (#80) Good ($90) SNPs5 rs2253820 rs934945

0.82

0.41

2.28 1

(1.02, 5.12)

1.37

0.54

3.95 1

(1.37, 11.42)

-0.79 -0.88

0.40 0.42

0.46 0.42

(0.21, 1.01) (0.18, 1.00)

1Logistic regression with backward selection at p,0.05 level; 2Standard Error; 3 Odds Ratio; 4Confidence Interval; 5Additive model.

Study treatment (per protocol population) n Arm Arm Arm Arm

1: 2: 3: 4:

LOM+BEV a` MD choice LOM, a` BEV BEV, a` LOM+BEV LOM, a` MD choice

OS 12 (95% CI)

Med OS Med PFS

70 32.9% (25.4, 41.1) 9.1 mo 73 27.4% (20.6, 35.2) 7.8 mo 70 32.9% (25.4, 41.1) 7.9 mo 36 41.7% (30.4, 53.7) 9.3 mo

4.2 1.5 2.8 1.6

mo mo mo mo

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

98s 2020

Central Nervous System Tumors Poster Discussion Session; Displayed in Poster Session (Board #209), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

2021

Poster Discussion Session; Displayed in Poster Session (Board #210), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Baseline plasma matrix metalloproteinase 9 (MMP9) to predict overall survival (OS) benefit from bevacizumab (BEV) in newly diagnosed glioblastoma (GBM): Retrospective analysis of AVAglio. First Author: Olivier L. Chinot, Aix-Marseille University, AP-HM, Service de Neuro-Oncologie, CHU Timone, Marseille, France

Residual tumor volume and change in tumor volume during adjuvant therapy to predict long-term survival in AVAglio: Phase 3 newly diagnosed glioblastoma patients treated with radiation, temozolomide, and bevacizumab or placebo. First Author: Benjamin M. Ellingson, University of California, Los Angeles, Los Angeles, CA

Background: BEV + radiotherapy/temozolomide (RT/TMZ) improves progression-free survival (PFS) (vs placebo [PBO] + RT/TMZ) in newly diagnosedGBM, but this does not translate into an OS benefit (AVAglio/NCT00943826). Data from recurrent GBM (Tabouret, Neuro Oncol 2014, 2015) suggest that MMP9 and MMP2 plasma levels may predict BEV-related tumor response and correlate with improved OS. Evaluation of MMP9 and MMP2 as potential biomarkers for patients likely to derive OS benefit from BEV is needed. We present data from a retrospective analysis of AVAglio, which assessed whether MMP9 baseline levels predicted OS benefit from BEV in newly diagnosed GBM. Methods: MMP9 and MMP2 levels were assessed (ELISA; R&D Systems) in baselineplasma samples (after surgery/before treatment) from 577/921 patients enrolled in AVAglio (PBO n=294; BEV n=283). Post-hoc analysis and PFS/OS multivariate models including MMP9–treatment interactions were carried out. Results: MMP9 distribution at baseline was comparable between arms (median [Quartile1–3]: PBO 82.4ng/mL [44.8–160.3]; BEV 83.6ng/mL [43.8–133.8]). Patient characteristics were generally balanced between MMP9 subgroups (per quartile), including for known prognostic factors. Patients with low MMP9 (,Quartile1) derived significant OS benefit from BEV, which translated into a median 5.2-month OS increase (HR 0.51, 95% CI 0.3420.76, p=0.0009; median 13.6 [PBO] vs 18.8 [BEV] months); a consistent PFS benefit was seen (HR 0.36, 95% CI 0.2420.54, p,0.0001). In patients with high MMP9, there was a trend for the HR to favor the PBO arm for OS (.Quartile 3 HR 1.21, 95% CI 0.8021.81). No clear prognostic value of MMP9 was shown, although an interaction was seen between treatment and MMP9 (p=0.03) for OS. MMP2 (by quartile) did not exhibit prognostic or predictive value. Conclusions: A post-hoc analysis of AVAglio suggested that baselineplasma MMP9 levels were predictive of OS benefit from the addition of BEV to RT/TMZ in newly diagnosed GBM. Clinical trial information: NCT00943826.

Background: Anti-angiogenic agents have raised doubts about the reliability of contrast enhancement as a surrogate of tumor burden in glioblastoma (GBM). Contrast-enhanced T1 subtraction maps increase lesion conspicuity in anti-angiogenic therapy. In the current study we used T1 subtraction maps to test whether post-surgical residual enhancing tumor volume and early changes in contrast enhancement during adjuvant therapy are prognostic of overall survival (OS) in patients with newly diagnosed GBM treated with radiation (RT) and temozolomide (TMZ) with or without bevacizumab (BV). Methods: A total of 798 patients (404 BV and 394 placebo) with newly diagnosed GBM enrolled in the AVAglio trial (Phase III; NCT00943826) had post-surgical, pre-RT MRI scans available. A total of 359 BV-treated and 333 placebo-treated patients had . 3 MRI scans for analyses. The volume of tumor with contrast enhancement plus central necrosis after T1 subtraction was quantified. Results: Patients with large (. 10cc) residual enhancing tumors after surgery have a significantly shorter OS regardless of chemotherapy (All patients, HR = 0.61, P , 0.0001; BV, HR = 0.62, P , 0.0001; Placebo, HR = 0.60, P , 0.0001). A decrease in enhancing tumor volume after RT and TMZ/TMZ+BV with respect to the pre-RT baseline predicted OS in the placebo arm (HR = 0.63, P , 0.0001) but not the BV arm (HR = 0.88, P = 0.49). If the post-RT/TMZ with or without BVscan is used as a new baseline for subsequent evaluation, change in enhancing tumor volume was a strong predictor of OS in both treatment arms (BV, HR = 0.74, P = 0.02; Placebo, HR = 0.66, P = 0.0003). Conclusions: Post-surgical residual enhancing tumor volume is prognostic for OS in newly diagnosed GBM patients. Change in enhancing tumor volume is useful as an early measure of response as indicated by correlation with OS in both treatment arms only after initial therapeutic changes have occurred. Thus, results suggest the first scan after radiation therapy should be used as the baseline for subsequent radiographic evaluation in both standard and anti-angiogenic treatment paradigms. Clinical trial information: NCT00943826.

2022

2023

Poster Discussion Session; Displayed in Poster Session (Board #211), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Poster Discussion Session; Displayed in Poster Session (Board #212), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Graded Prognostic Index for Gastroesophageal Cancer with Brain Metastases. First Author: Vidhya Karivedu, Fairview Hospital Cleveland Clinic, Cleveland, OH

Characterization of cell-free circulating tumor DNA in patients with brain metastases. First Author: Nadia Faiq, UC San Diego Moores Cancer Center, San Diego, CA

Background: Brain metastasis (BM) is a rare but serious neurologic complication of gastro esophageal cancers (GEC). The Disease Specific Graded Prognostic Assessment (DS-GPA) which is based solely on Karnofsky performance scale (KPS) is a commonly used prognostic index in patients with BM. We evaluated DS-GPA and other potential prognostic factors for overall survival (OS) in GECBM at our institution. Methods: With IRB approval, the Cleveland Clinic Brain Tumor and Neuro-Oncology Center’s database was used to identify GECBM patients treated between 2002 and 2014. Overall survival (OS) from the diagnosis of GECBM was the primary endpoint. Cox proportional hazards models with stepwise variable selection were used to identify independent prognostic factors. Results: Sixty three patients with median age of 62 years (range; 33-79), were included for analysis. KPS was 90-100 in 22 patients (38%), 70-80 in 30 (52%) and ,70 in 6 (10%) patients. Single BM was noted in 25 (40%), 2-3 BM in 19 (30%) and .3 BM in 19 (30%) patients. Forty nine patients (78%) were symptomatic at diagnosis. Twenty nine (46%) patients had supratentorial BM, 12 (19%) had infratentorial BM and 22 (35%) had both supra-and infratentorial BM at diagnosis. Initial therapy included whole brain radiation (WBRT) in 29 (48%), stereotactic radiosurgery (SRS) in 13 (21%), WBRT + SRS in 6 (10%), Surgery (S) + SRS in 2 (3%), S + WBRT in 9 (15%), S + SRS + WBRT in 1 (2%), while 1 patient received only chemotherapy. Median OS from diagnosis of BM was 4.5 months (95% C.I. 3.4-6.8). In multivariate analysis KPS, number of BM and symptoms were independent predictors of OS. Combining these factors a revised GPA with three groups was defined: unfavorable (total points ,10), intermediate (11-13), and favorable (.13 points) with median OS of 2.5, 5.0 and 8.8 months, respectively. Conclusions: A revised DS-GPA for GECBM based on KPS (performance status), number of BM and symptoms is proposed.

Background: Brain metastases are the most common malignant brain tumor and a significant cause of morbidity and mortality. As patients live longer from more effective systemic therapies, they are at increased risk of developing brain metastases. Genetic profiling of tumor tissue allows for targeted therapy, though biopsy of brain metastases is not always feasible. Cell-free circulating tumor DNA (ctDNA) testing carries diagnostic potential for targeted therapy without the need for repeat biopsies, and the ability to rapidly respond to changes in the molecular profile. Methods: 37 patients with brain metastases (15 lung, 12 breast, 5 ovarian, 3 renal, 2 melanoma) were tested prospectively with the Guardant360 ctDNA panel at a CLIAcertified, CAP-accredited clinical laboratory. Samples were analyzed for amplifications in 18 genes, single nucleotide variants in 68 genes, and select fusions, rearrangements and indels. Results: 29/37 (78.4%) patients tested harbored at least one detectable genomic alteration. The median number of alterations was 3 (range, 1-12). The most frequently detected alterations were in EGFR (19), TP53 (18), APC (6), NF1 (6), PIK3CA(6). The median ctDNA concentration was 0.50% (range, 0.10-43.80%). 11/37 patients had a subsequent test. 4/11 patients were tested within 50 days of initial testing; the median number of new alterations detected was 0.5 (range, 0-1). 7/11 patients were tested greater than 50 days of initial testing; the median number of new alterations detected was 2 (range, 0-4). Conclusions: CtDNA testing was able to detect alterations in a majority of patients with brain metastases. An increasing number of alterations were observed with time, suggesting the utility of ctDNA analysis in monitoring changes in tumor biology, response to therapy, and the ability to refine treatment plans with changes in the molecular profile.

Factors Number of BM 1 2-3 .3 Colon cancer specific GPA (KPS) 100 90 80 70 ,70 Symptoms Symptomatic Asymptomatic

No. of points

Hazard ratio

P valve

1.54 (1.08-2.18)

0.042

0.73 (0.54-0.99)

0.045

0.47 (0.23-0.97)

0.017

6 3 0 8 6 4 2 0 0 5

Revised GPA

No. Points

Median OS (months)

Unfavorable Intermediate Favorable

0-10 11-13 .13

2.5 5.0 8.8

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Central Nervous System Tumors 2024

Poster Session (Board #213), Sat, 1:00 PM-4:30 PM

2025

99s Poster Session (Board #214), Sat, 1:00 PM-4:30 PM

Blood baseline neutrophil count to predict bevacizumab efficacy in glioblastoma. First Author: Aurelie Bertaut, Georges-Francois Leclerc Cancer Center, Dijon, France

Translational impact of patient-derived glioblastoma tumorspheres. First Author: Maurizio Martini, Policlinico Universitario A. Gemelli, Istituto di Anatomia Patologica, Oncological Area, Rome, Italy

Background: Bevacizumab is frequently used at recurrence of glioblastoma. However, no biomarker could predict its efficacy. Methods: Exploratory cohort consisted in all patients treated with Stupp regimen for a glioblastoma in Dijon cancer center (N = 265). A validation cohort was TEMAVIR trial which compared 120 patients suffering from glioblastoma treated with Stupp regimen versus bevacizumab plus irinotecan followed by Stupp regimen plus bevacizumab. CSF3 expression was determined in a cohort of 202 patients from the cancer genome atlas. Prognosis role of CSF3expression was next determined in an independent cohort of 23 patients treated with bevacizumab for a recurrent glioblastoma. Results: In the exploratory cohort, bevacizumab increased median overall survival (OS) 18.7 vs 11.3 months, p = 0.0014. Upon multivariate analysis, the age, type of initial surgery, neutrophil count, Karnofsky status . 70%, bevacizumab administration were an independent prognosis factors of survival. We detected an interaction between bevacizumab use and neutrophil count at baseline. Cut off value of neutrophil count was set at 6000/mm3. Only patients with a high level of neutrophil benefited in term of overall survival from the bevacizumab administration (17.3 months (CI95% 14.6-20.4) for patients that received bevacizumab versus vs 8.8 months CI95% 6.4-10.3) for others (p , 0.0001)). Overall survival was not significantly improved for remaining patients treated with bevacizumab with low neutrophil count (21.6 months (CI95% 18.0-23.3) for patients that received bevacizumab versus 15.9 months (CI95%12.0-19.9) for the others (p = 0.7313)).Similar results were obtained in the TEMAVIR trial. Transcriptomic data underlined that expression of CSF3, the gene encoding G-CSF, was correlated with VEGF-A expression. In another independent cohort, we also observed this association between high CSF3expression in tumor and a better survival in patients treated with bevacizumab at recurrence. Conclusions: Only patients with a high level of neutrophil before initiation of bevacizumab treatment benefited from this therapy. Neutrophil count at baseline is a predictive biomarker of efficacy of bevacizumab for glioblastoma bearing patients

Background: Advances from glioma stem-like cell research, though increasing our knowledge on glioblastoma biology, do not influence clinical decisions yet. We conducted this study to explore the translational power of patient-derived glioblastoma tumorsphere cultures, which are known to be enriched for glioma stem-like cells. Methods: The relationship between growth of tumorspheres and outcome of donor patients was investigated in 52 glioblastoma tumors treated with surgical resection followed by radiotherapy and temozolomide. The effect of radiation (6 to 60 Gy) and of temozolomide (3.9 mM to 1 mM) on tumorsphere cell viability was related with the survival of donor patients. Results: Generation of tumorspheres was a highly significant independent factor for poor overall survival and progression-free survival in glioblastoma patients (p , 0.0001 and p = 0.0010, respectively). In general, tumorsphere cells were highly resistant to both radiation and temozolomide. Significantly shorter progression-free survival of the donor patients was associated with LD50 . 12 Gy of matched tumorsphere cultures (p = 0.0484). In addition, the activating phosphorylation level of ATM, Rad17 and to a lesser extent of Chk2 checkpoint proteins, induced by to DNA damage, was higher in the radioresistance tumorspheres in comparison to the sensitive ones. A direct relationship was found between sensitivity of tumorspheres to temozolomide and patient survival (p = 0.0167 and p = 0.0436 for overall survival and progression-free survival, respectively). Importantly, values for temozolomide IC50 , 50 mM, which are clinically meaningful since the plasma levels achieved in vivo are in this range, identified cases with significantly longer overall survival and progression-free survival (p = 0.0020 and p = 0.0016, respectively). Conclusions: Analysis of tumorsphere cultures does hold translational relevance by predicting the response of the parent tumor to radiation and, above all, to temozolomide. Dissecting the clonogenic core from the proliferating cell progenies in tumorspheres will orient future therapies for drug selection and treatment duration.

2026

2027

Poster Session (Board #215), Sat, 1:00 PM-4:30 PM

Therapy and outcomes of primary central nervous system lymphoma (PCSNL) in the United States: Analysis of the National Cancer Data Base (NCDB). First Author: Lindor Qunaj, Alpert Medical School of Brown University, Providence, RI Background: Although the role of radiation therapy (RT) and chemotherapy (Ch) in PCNSL has undergone a major evolution over the past decade, the application of these modalities in the community has not been studied. Our objective was to analyze the use of RT and Ch, and associated overall survival (OS), using the NCDB—an oncologic registry capturing ~70% of incident cancers in the US. Methods: We selected histologically confirmed PCNSL cases (diffuse large or unspecified B-cell lymphoma) reported to the NCDB between 2004 and 2012 excluding patients (pts) with HIV+ status. We identified the use of RT and Ch (single- or multiagent) as part of the initial therapy, and assessed factors associated with non-receipt of Ch in a multivariable modified Poisson model, adjusting for immortal-time bias by conditional landmark analysis at 3 months. OS was calculated from diagnosis and analyzed in a multivariable Cox model, reporting hazard ratio (HR) and 95% confidence intervals (CI). Results: Among 6,549 pts (median age, 67 years, 51% women), 71% received Ch (43% multiagent, 17% with RT), 11% RT alone, and 18% no active therapy for PCNSL. Between 2004 and 2012, the proportion receiving Ch increased from 65% to 73% (P for trend , .00001), while the proportion receiving any RT decreased from 36% to 22% (P, .00001). Non-receipt of Ct was significantly associated with age . 70 years, black race, lack of private insurance, higher comorbidity index, and unspecified histology (all P, .0001). Non-receipt of Ct was also significantly lower in academic/research cancer programs compared with community programs (adjusted relative risk, 0.61, CI, 0.49-0.76). OS at 3 years was 34.2% (CI, 32.9-35.5) for all pts, 49% (CI, 47-51) for pts receiving multi-agent Ct, 37% (CI, 34-39) for single-agent Ct, and 10% (CI, 712) for RT alone. Adjusting for confounding factors, OS was better after multi-agent Ct than after single-agent Ct (HR, 1.32, CI, 1.22-1.44) or RT alone (HR, 2.32, CI, 2.07-2.59). Conclusions: Disparities in the use of Ch for PCNSL are evident with regard to individual patient and facility characteristics. The proportion receiving Ch for PCNSL is increasing, and multiagent Ch is associated with better OS.

Poster Session (Board #216), Sat, 1:00 PM-4:30 PM

Phase 2 study of aldoxorubicin in relapsed glioblastoma. First Author: Morris D. Groves, Texas Oncology/The US Oncology Network, Austin, TX Background: Glioblastoma (GBM) is the most common primary brain tumor with a median survival of ~14 months after initial therapy. Relapsed patients usually receive bevacizumab with an ORR of~20%, and OS of ~31 weeks. Recently Prakash et al demonstrated in an orthotopic mouse model of GBM that IV aldoxorubicin (A) but not doxorubicin (D) significantly delayed tumor growth and prolonged survival by over 100%. D does not pnetrate the bloodbrain barrier. A is a prodrug of D that is attached to a pH sensitive linker that binds covalently to albumin and is released preferentially within tumors. We assessed the preliminary efficacy and safety of A administered to GBM patients who progressed after first line therapy. Methods: Patients . 18 yrs with 1st relapse GBM received either 250 mg/m2 (21) or 350 mg/m2(7) every 21 days until tumor progression, unacceptable toxicity or withdrawal. Tumors were assessed every 6 weeks by MRI and in some instances Dynamic Susceptibility Contrast (DSC) MRI. Safety monitoring included serum chemistries and CBC prior to each infusion and echocardiograms every 2 molnths. Results: Median age was 56 yrs (28-73); 19 females and 9 males. Subjects received 1-20 cycles of treatment. Median drug exposure for the 250 mg/m2 dose is 1042 mg/m2 D equivalents (Deq) (range 371-5768 mg/ m2 Deq) and for the 350 mg/m2 dose is 1912 mg/m2 Deq (range 4723647 mg/m2Deq). Best responses in 21 subjects according to MRI were 3 PR, 7 SD and 11 PD. Median overall survival has not been reached (range is 0.5-17.3+ months) with median f/u of 7 months (1-17+ months). Major grade 3 or 4 AEs include neutropenia (6 subjects) febrile neutropenia (2 subjects) mucositis (2 subjects), thrombocytopenia (2 subjects) and anemia (2 subjects). 2 subjects experienced treatment-related SAEs. No subjects had a decrease in their LVEF below 50% of expected values.Two subjects with apparent PD by scan underwent tumor debulking and had no microscopic evidence of disease. Another subject demonstrated no active malignancy upon DSC MRI scanning. Conclusions: Unlike doxorubicin, Aldoxorubicin appears to penetrate the blood-brain barrier in humans and is associated with objective tumor responses, stable disease and prolonged survival. Pseudoprogression occurs and may confuse responsiveness to A. Clinical trial information: NCT02014844.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

100s 2028

Central Nervous System Tumors Poster Session (Board #217), Sat, 1:00 PM-4:30 PM

Comparison of oncometabolite 2-hydroxyglutarate (2HG) levels in mutant isocitrate dehydrogenase (IDH) versus wild-type (WT) glioma tissues. First Author: Hao-Wen Sim, Princess Margaret Cancer Centre, University Health Network, Toronto, ON, Canada

2029

Poster Session (Board #218), Sat, 1:00 PM-4:30 PM

Validating RNAseq-signatures of vorinostat (VOR) sensitivity and resistance in patients with newly diagnosed glioblastoma (GBM) treated with VOR, temozolomide and radiation therapy in Alliance N0874/ABTC-0902. First Author: S. Keith Anderson, Alliance Statistics and Data Center, Mayo Clinic, Rochester, MN

Background: IDH mutations are common and prognostic in low-grade gliomas and secondary glioblastomas. IDH catalyzes the oxidative decarboxylation of isocitrate to a-ketoglutarate (aKG), and subsequently 2HG. Mutant IDH preferentially catalyzes the NADPH-dependent reduction of aKG to R-2-hydroxyglutarate (R-2HG), resulting in accumulation of R-2HG, relative to its enantiomer, S-2-hydroxyglutarate (S-2HG). Previous attempts to correlate plasma/urine 2HG levels with survival and responses to therapies were limited by inconsistent results. Thus, we analyzed R- and S-2HG levels, and ratio of R- to S-2HG (RRS), in glioma tissues from patients with and without IDH mutations. Methods: Fresh frozen glioma tissues were obtained from the University of Toronto Brain Tumor Bank. Diagnosis and IDH status were determined by microscopy and immunohistochemistry. Samples were analyzed using HPLC tandem mass spectrometry coupled with a CHIROBIOTIC R column. Results: Glioma tissues from 4 patients with IDH mutations and 4 WT patients were analyzed. Patient characteristics: M:F ratio: 4:0 for mutant, 3:1 for WT. Age range at surgery: 34-59 for mutant, 47-74 for WT. Tumor size (cm): 2.3-5.4 for mutant, 2.1-8.0 for WT. Location: frontal (2), temporal (1), occipital (1) for mutant, and parietal (1), temporal (2), occipital (1) for WT. Recurrence with treatment effect: 2 of 4 mutant, 3 of 4 WT. Median survival was longer in patients with mutant IDH gliomas (39.9 vs. 17.8 months, p = 0.018). Mutant IDH gliomas had higher total 2HG (mean 451548 vs 2728 ng/g, p = 0.055) and RRS (mean 537.61 vs 0.70, p = 0.022). RRS was 0.36, 0.57, 0.70 and 1.18 in 4 WT patients, and 4.15, 522.75, 543.22 and 1080.34 in 4 mutant patients. Disease-free survival was 6.3, 6.2, 28.3 and 72.0 months respectively for mutant patients with lowest to highest RRS. Conclusions: IDH mutations result in significantly higher RRS in glioma tissues. RRS in glioma tissue, compared with plasma/urine 2HG levels, appears to be a reliable correlate of IDH status. The potential utility of RRS as a prognostic marker warrants further evaluation.

Background: Five RNAseq-based gene set enrichment analysis (GSEA) signatures (sig) were previously developed in pre-clinical models to predict sensitivity or resistance to treatment with the HDAC inhibitor VOR. This study investigates if these signatures can identify groups of newly diagnosed GBM patients (pts) deriving benefit from being treated with VOR in addition to standard chemoradiation. Methods: Baseline tumor samples were obtained from 76 pts enrolled on the Phase II single arm trial N0874 (Galanis, et al., ASCO 2014). RNAseq-based GSEA sig scores were calculated for four VORsensitivity sig (9, 77, 80, and 139) and one VOR-resistance sig (79). Overall survival (OS) and progression free survival (PFS) were used to evaluate univariate cut-points for each signature. Recursive partitioning (rpart) was used to simultaneously evaluate age, MGMT status, and the five sig scores to identify patient groups with differences in OS and PFS. Kaplan-Meier analysis was used to evaluate OS and PFS using the best cut-points and partitions. Results: Higher scores for sig 77, 80, and 139 all had at least one univariate cut-point associated with a statistically significant improvement for PFS (HR 0.55, 0.54, and 0.5 respectively, all p , 0.05). Higher scores for sig 79 had a univariate cut-point related to lower PFS (HR = 1.7, p = 0.028). Only sig 139 had a univariate cut-point related to OS (HR = 0.45, p = 0.039). Rpart for OS and PFS resulted in 5 groups of pts for each. Pts with high 79-scores had shorter OS and PFS. Pts with low 139-scores had shorter PFS, even if sig79 scores were low. MGMT methylation status did not affect the negative association of sig79 with PFS and OS and positive association of sig139 with PFS. Conclusions: Pts with high scores for the VOR-resistant signature 79 have shorter PFS and OS and patients with high scores on the VOR-sensitive signature 139 have a longer PFS and OS. These results suggest that these signatures could be useful in identifying subgroups of GBM patients that might derive benefit from VOR treatment. Clinical trial information: NCT00731731.

2030

2031

Poster Session (Board #219), Sat, 1:00 PM-4:30 PM

Poster Session (Board #220), Sat, 1:00 PM-4:30 PM

Encouraging survival with Toca 511 and Toca FC compared to external lomustine control. First Author: Timothy Francis Cloughesy, UCLA, Los Angeles, CA

Precision medicine in recurrent glioblastoma: A feasibility trial conducted by the Ivy Foundation Early Phase Clinical Trials Consortium. First Author: Michael Prados, UC San Francisco, San Francisco, CA

Background: Recurrent glioblastoma (GBM) has an unmet need for effective therapies. Toca 511 (vocimagene amiretrorepvec), a retroviral replicating vector, encodes the transgene cytosine deaminase. Toca 511 selectively infects, persists and spreads in cancer cells. Subsequent oral administration of extended-release 5-fluorocytosine (Toca FC) produces 5-fluorouracil (5FU) within infected cells. 5-FU kills cancer cells and myeloid derived suppressor cells, leading to robust antitumor immune responses in animal models. In 2 Phase 1 studies, Toca 511 was administered by surgical resection and injection into the cavity wall (NCT01470794) or intratumoral injection by biopsy needle (NCT01156584). To provide context to the results observed, patients were compared to an external lomustine treated control (Courtesy Denovo Biopharma; Wick 2010). Methods: Demographics, overall survival (OS) and safety profile of Toca 511/Toca FC treated subjects with GBM in 1st or 2nd recurrence were compared to the lomustine control. Results: Toca 511/Toca FC treated patients were comparable to the lomustine control. Imbalances in KPS in favor of Toca 511/Toca FC treated patients were offset by younger age, higher number of subjects at 1strecurrence and lower steroid requirement in the lomustine control. Treatment with Toca 511/Toca FC, which was pooled from 2 Phase I studies, showed significant improvement in OS HR = 0.48, p , 0.001 with similar effect in the setting of surgical resection (OS HR 0.45, p = 0.003) and nonsurgical resection with a biopsy needle (OS HR 0.56, p = 0.060). Fewer related Grade $ 3 adverse events (AEs) were reported for Toca 511/Toca FC (2.5%) vs. lomustine (36.9%). There was a virtual absence of hematologic toxicity for Toca 511/Toca FC vs. lomustine (Grade $ 3 thrombocytopenia 23.8%). Discontinuations for AEs occurred in 0% for Toca 511/Toca FC vs. 4.8% for lomustine. Toca 511 is surgically delivered and treatment-emergent AEs regardless of attribution included incision site pain (20%), procedural pain (12.5%), and wound infection (5%) vs. 0%, 1.2%, 1.2% respectively for lomustine. Conclusions: Toca 511/Toca FC significantly improved survival and safety relative to lomustine. A Phase 2/3 trial has launched (NCT02414165)

Background: Pts with recurrent glioblastoma have median survival between 4 to 8 months. "Precision-based" therapies have not been used in this patient population. The goal of this trial was to determine the feasibility to make individualized treatment recommendations within at least 35 days of tumor tissue collection, using whole exome and RNA sequencing based upon those alterations. A sample size of at least 15 patients with sufficient tumor RNA and DNA was used to determine feasibility. Pts. had the option of accepting the recommendation or being treated using other treatment strategies. Adult patients with recurrent glioblastoma who were undergoing tumor resection at the time of relapse as part of their standard of care management were eligible. Methods: Tumor tissue was removed from both enhancing and nonenhancing regions of tumor and characterized by whole exome and RNA sequencing. Molecular profiles were done at Translational Genomics Research Institute (TGen), in a CLIA lab, matching genomic alterations with both cancer positioned and repositioned drugs. Up to 4 agents could be selected by a Molecular Tumor Board convened after each case to discuss prior patient treatment history, molecular alterations, and drug options. The treatment recommendation was based upon alterations from the enhancing tumor region. Non-enhancing tumor profiles, patient-derived xenografts, and circulating tumor DNA samples (obtained prior to and following surgery, and every 8 weeks) were used for correlative endpoints. Results: 20 patients were enrolled, and 16 had sufficient tumor content to profile. There were 3 feasibility failures (1 pt. took . 35 days to profile; 2 pts. did not have sufficient RNA content). Of the 16 eligible cases, the average time from surgery to Tumor Board was 27 days, with an average of 3.4 drugs recommended (range 1-4) including at least 1 repositioned agent in 13 cases. 7 pts accepted the Tumor Board recommendation. Molecular findings will be presented. Conclusions: It is possible to obtain useable genomic information at the time of relapse, and make treatment recommendations within weeks. Efficacy studies are being planned using this strategy. Clinical trial information: NCT02060890.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Central Nervous System Tumors 2032

Poster Session (Board #221), Sat, 1:00 PM-4:30 PM

The role of clinical characteristics and molecular biomarkers in low grade gliomas (LGG): A GICNO study. First Author: Alba Ariela Brandes, Azienda USL Bellaria-Maggiore Hospital, Bologna, Italy Background: Molecular characterization of low grade gliomas (LGG) has improved in recent years and is essential for diagnosis and treatment of these diseases. Moreover, clinical factors, such as age and the extent of surgery retain a prognostic role in LGG that were assessed for IDH1/2, 1p/19q codeletion, and MGMT methylation status. Methods: we retrospectively evaluated all adult LGG patients (pts) from our data warehouse who received surgery and had sufficient tissue to assess biomarkers characterization. IDH1/2 assessment was performed on formalin-fixed paraffin-embedded samples by PCR, MGMT by methylation specific PCR, 1p/19 codeletion by FISH. Results: 198 consecutive LGG were included. The median age was 38 (range:18–72). Median follow up was 74.0 months, 109 pts (55.1%) were , 40 years of age; 26 pts (13.1%) underwent biopsy, 119 pts (60.1%) partial resection, 53 pts (26.8%) complete resection. Twenty-eight pts (14.1%) were considered low risk (, 40 years with complete resection). IDH1/2 mutation was found in 79.8% of pts. 1p/19q codeletion was found in 41.4% of pts, MGMT methylation in 57.1% of pts. Median survival in low risk patients was 211.0 months (95%CI: 190.4–231.6) and was 145.3 months (95%CI: 108.5–182.2) in high risk patients (P = 0.006). Median survival for patients with IDH1/2 mutation was 159 months (95%CI: 103.3–214.7) and was 87.9 months in patients IDH1/2 wild type (95%CI: 61.1–114.6, P , 0.001). The impact of molecular biomarkers on survival is in the table. Multivariate analysis showed that clinical risk (P = 0.006), IDH1/2 mutation (P , 0.001) and 1p/19q codeletion (P = 0.03) were significantly correlated with overall survival. MGMT methylation was not significant. Conclusions: Molecular characteristic of LGG define the prognosis of these tumors. Moreover, clinical risk assessment retains a role even in the era of molecular characterization.

1p19q codel 1p19q non codel IDH1 mut, 1p19q IDH1 mut, 1p19q IDH1 mut, 1p19q IDH1 mut, 1p19q

2034

codel, MGMT met codel, MGMT unmet not codel, MGMT met not codel, MGMT unmet

Median OS (months)

95%CI

186.2 138.5 NR 189.4 202.7 109.1

128.2 244.2 91.0 186.0 / / / 105.3 112.8

5-year OS 10-year OS 94.7 84.9 100.0 92.9 95.8 90.7

73.8 51.5 95.7 61.9 72.9 44.1

Poster Session (Board #223), Sat, 1:00 PM-4:30 PM

Trends in utilization and impact of concurrent chemotherapy with radiation therapy for elderly patients with newly diagnosed glioblastoma: A review of the National Cancer Data Base. First Author: Jiayi Huang, Washington University School of Medicine in St Louis, St Louis, MO Background: To investigate the utilization and overall survival (OS) impact of concurrent chemotherapy in combination with radiation therapy (RT) for elderly glioblastoma (GBM) patients. Methods: Elderly patients (age . 70) with supratentorial and nonmetastatic GBM who received RT of 20-75 Gy with concurrent single-agent chemotherapy (chemoRT) or without (RT alone) during 2004-2012 were identified from the National Cancer Data Base (NCDB). RT fractionation was evaluated for those who received standardfractionated RT (SFRT) or abbreviated hypo-fractionated RT (HFRT). Hazard ratios (HR) and 95% confidence intervals (CIs) were determined using Cox regression analysis. OS rates were determined using the Kaplan-Meier method and compared using the log-rank test. Propensity score analysis was performed to reduce selection bias in treatment allocation. Results: A total of 5,252 patients were identified (RT alone: n = 1389; chemoRT: n = 3863). There was increasing utilization of chemotherapy since 2004, which reached 80% per year after 2010 (including . 66% for the subset of age . 80). ChemoRT was associated with significantly better OS than RT alone (HR 0.76; 95% CI 0.71-0.81; P, .001) on multivariate analysis. Similar OS benefit was demonstrated with 1202 pairs of propensity-matched patients (HR 0.79, 95% CI 0.73-0.86; P, .001). For the entire cohort, OS was 33% at 1 year and 12% at 2 years with chemoRT, versus 15% at 1 year and 3% at 2 years with RT alone (P, .001). For the matched pair, OS was 26% at 1 year and 9% at 2 years with chemoRT, versus 16% at 1 year and 4% at 2 years with RT alone (P, .001). Five thousand and one-hundred fifty-two patients (98%) were evaluable for RT fractionation (SFRT: n = 4621; HFRT: n = 531). HFRT was used selectively for a minority of patients and increased slightly after 2010 (approximately 10% per year for age # 80 and 20% for age . 80). Conclusions: Concurrent chemotherapy has been administered with RT for the majority of elderly GBM patients. Addition of chemotherapy to RT for elderly GBM patients significantly improved OS in routine clinical practice. HFRT utilization remains limited for elderly GBM patients.

2033

101s Poster Session (Board #222), Sat, 1:00 PM-4:30 PM

Phase I/II study of S49076, a multi-target inhibitor of c-MET, AXL, FGFR in combination with bevacizumab in patients with recurrent glioblastoma. First Author: Khe Hoang-Xuan, AP-HP, Hopital ˆ La Pitie´ Salpetri ˆ ere, ` IGCNO, Paris, France Background: There is no standard of care for the treatment of recurrent GBM (rGBM) and more effective treatments are clearly needed. S49076 is an orally bioavailable multikinase inhibitor of MET, AXL, FGFR1, 2, 3. The ability of S49076 to inhibit tumor growth in vivo was showed in a xenograft mouse model resistant to bevacizumab (BEV). Here, we report data from an ongoing Phase I study of S49076 in combination with BEV at first recurrence. Methods: Patients (pts) received escalating doses from 400 to 600 mg of S49076 orally, on a continuous once daily schedule of 28-day cycle, with BEV 10mg/kg on days 1 and 15. Primary objectives were to determine the Maximum Tolerated Dose (MTD), the Dose Limiting Toxicities (DLT) and to establish the Recommended Phase II Dose (RP2D) of S49076 +BEV. Secondary objectives included pharmacokinetics (PK), pharmacodynamics (PD) and antitumor activity (assessed every each cycle using RANO criteria). Results: Thirteen rGBM pts were included. Overall 75% (9/ 12) were male, median age was 56 years (range 34-70), and median KPS was 90 (range 70-100). Three pts were respectively treated at the dose levels 400 mg and 500 mg and 6 patients at 600 mg (due to 1 DLT observed; G3 asymptomatic ejection fraction decrease). This dose level was considered as the RP2D. The reported adverse events (AE) related to S49076 were mainly of G1 or G2 (2 AE $ G3). No grade 5 S49076 and/or BEV related AE has been observed. Best responses for the 12 evaluable pts were: 4 partial responses (PR) confirmed and 2 unconfirmed, 5 stable disease (SD) including 2 SD $ 3 months, and 1 progressive disease. Similar S49076 PK profile in combination to BEV has been observed in the first-in-human study where S49076 was used in monotherapy. PD results showed high and moderate MET amplification assessed by FISH in 2 pts which were not predictive of clinical response. No expression of MET and AXL were observed by immunohistochemistry (IHC) in any of the pts. Conclusions: The combination of BEV and S49076 has been well tolerated at the dose levels tested. These preliminary data does not suggest an increase of activity by adding S49076 to BEV. Clinical trial information: 2013-003079-37.

2035

Poster Session (Board #224), Sat, 1:00 PM-4:30 PM

A phase 2 trial of TRC105 with bevacizumab for bevacizumab refractory glioblastoma. First Author: Manmeet Singh Ahluwalia, Cleveland Clinic, Cleveland, OH Background: Endoglin is an endothelial cell membrane receptor highly expressed on tumor vessels that is essential for angiogenesis. It is upregulated by hypoxia and VEGF inhibition and may be an important mechanism of VEGF resistance. The combination of TRC105, an anti-endoglin monoclonal antibody, in combination with bevacizumab (Bev) has demonstrated activity in Bev refractory cancer patients with non-CNS tumors. We hypothesized that TRC105 administered with Bev would improve overall survival in glioblastoma patients who progressed on Bev. Methods: Patients with previously treated, recurrent glioblastoma (GBM) with radiographic progression following Bev therapy, KP status $ 70%, and adequate organ function were treated with TRC105 10 mg/kg weekly as a single agent (six patients) and then in combination with Bev 10 mg/kg every two weeks (16 patients). The primary endpoint for treatment with TRC105 and Bev was overall survival (OS), with the null hypothesis being OS of 4.0 months and alternative hypothesis being OS of 7.0 months. Results: Twenty-two patients (median age 54.5; M:F 16:6, median prior therapies 3) were enrolled; one patient did not receive TRC105 and was not included in the efficacy analysis. Initially, six patients were treated with TRC105 as a single agent. TRC105 was well tolerated in patients with GBM; each of six patients progressed by 1.40 months of treatment. Sixteen patients were treated with TRC105 and bevacizumab: adverse events (AEs) characteristic of each drug were not increased in frequency or severity during concurrent dosing. Fifteen patients were evaluable for efficacy of the combination and median overall survival (OS) was 5.75 months (95% CI: 4.21, 9.86), which exceeded the historic OS of 4.0 months seen in this patient population with Bev alone. No responses were seen by RANO criteria and median PFS was 1.81 months (95% CI: 1.25, 2.07) Conclusions: The combination of TRC105 and Bev was well tolerated in Bev refractory GBM patients. OS was 5.75 months, which exceeded the historic OS of 4.0 months. Treatment with the combination of TRC105 and Bev is currently being evaluated in Bev na¨ıve GBM patients in a randomized Phase 2 trial. Clinical trial information: NCT01564914.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

102s 2037

Central Nervous System Tumors Poster Session (Board #225), Sat, 1:00 PM-4:30 PM

Phase 1, multicenter, open-label, dose-escalation, study of marizomib (MRZ) and bevacizumab (BEV) in WHO grade IV malignant glioma (G4 MG). First Author: Daniela Annenelie Bota, UC Irvine Medical Center, Orange, CA

2038

Poster Session (Board #226), Sat, 1:00 PM-4:30 PM

Randomized phase II study of axitinib alone or combined with lomustine in patients with recurrent glioblastoma. First Author: Johnny Duerinck, UZ Brussel, Brussels, Belgium

Background: MRZ is an irreversible, brain-penetrant, pan-proteasome inhibitor (PI). It inhibits glioma cell proliferation and invasion in vitro, prolongs survival in in vivo mg tumor models, with little effect on normal neuronal stem cells, suggesting minimal neurotoxicity (Neuro Oncol 2015 Dec 17. pii: nov299). Intravenous (IV) MRZ has been administered to ~300 patients (pts) with solid tumors and hematologic cancers. This trial is evaluating the safety, pharmacodynamics, and efficacy of MRZ and BEV in BEV-na¨ıve pts with G4 mg who are in first or second relapse with no prior anti-angiogenic or PI therapy. Methods: Phase 1, dose-escalation (3+3 design) followed by doseexpansion at recommended Phase 2 Dose (RP2D); three dose cohorts - MRZ 0.55 (n = 6 pts), 0.7 (n = 3 pts), and 0.8 mg/m2 (n = 3 pts). MRZ infused IV (10 min) on Days 1, 8, & 15; BEV IV 10 mg/kg on Days 1 & 15 (over 28-Day Cycles). Tumor response is assessed every other Cycle by RANO criteria; blood proteasome inhibition assessed every Cycle. Results: In doseescalation, 12 pts enrolled: median age 53 yrs (44-61); 83% male; 83% Caucasian; Karnofsky Score . 80. As of 6 Jan 2016, duration of dosing is 533 weeks with treatment ongoing in 7 pts. MRZ/BEV was well tolerated; most common adverse events (AEs): headache (9), fatigue (7), hypertension (5), infusion site pain (5), and nausea (5). Grade 3 AEs related to MRZ or BEV: hypertension (2), confusional state (1), fatigue (1), hallucinations (1), and headache (1). One pt (cohort 1) had fatigue as DLT, but no DLTs at higher doses. MRZ caused . 70% inhibition of chymotrypsin-like activity on Cycle 1 Day 1 with 100% by Day 28. Transient hyperactivation of trypsin-like (T-L) and caspase-like (C-L) activities after the first 1-2 MRZ doses, followed by 40-60% inhibition of T-L and 10-30% inhibition of C-L activities in Cycle 3. RANO responses in the 3 cohorts: 4/12 partial response, 6/12 stable disease and 2/12 progressive disease. Conclusions: The MRZ/BEV combination was well tolerated with only one DLT at MRZ 0.5 mg/m2. To date, rapid and pronounced pan-proteasome inhibition has been observed with 4/12 pts achieving confirmed PRs. MRZ 0.8 mg/m2 is the RP2D in the ongoing doseexpansion stage. Clinical trial information: NCT02330562.

Background: vascular endothelial growth factor receptor (VEGFR) signal transduction mediates glioblastoma (GB) associated neo-angiogenesis. Axitinib, a small molecule TKI with high affinity and specificity for the VEGFRs, has demonstrated anti-tumor activity in patients with recurrent GB (J Clin Oncol 32:5s, 2014 [suppl; abstr 2018]). We investigated whether the combination of axitinib with lomustine (LOM) improves the outcome of pts with rGB as opposed to axitinib monotherapy. Methods: pts with rGB were randomized between single agent axitinib (AXI) vs. axitinib in combination with LOM (AXILOM) in an open label, randomized, phase II clinical trial. Pts in the AXI arm were allowed to cross over to AXILOM at progression. Sixmonth PFS served as the primary endpoint. Results: between February 2014 and July 2015, 56 pts were randomized 1:1 to AXI and AXILOM. Baseline characteristics were well balanced between both study arms (median age 56y [range 18-75]; 35M/21F; 15, 19, 11 and 11 pts had a WHO-PS of 0, 1, 2 and 3 respectively; MGMT promoter methylation: 11 pts methylated, 35unmethylated, 10- unknown). All pts had failed prior therapy with RT and temozolomide. Six-mth PFS% did not differ between both study arms (AXI: 21% (95% CI 7-37] vs. AXILOM: 17% [95% CI 2-32]; neither did OS differ between both study arms (6-mths OS AXI 57% [95% CI 38-76] vs. AXILOM 53% [95% CI 34-73]). Pts with a methylated MGMT-promoter had a superior PFS and OS on the AXIG-arm but not in the AXILOM-arm. The best response (by RANO criteria) was 0 CR/ 6 PR/ 8 SD in the AXI arm (BORR 21%) vs. 1 CR/ 9 PR/ 3 SD in the AXILOM arm (BORR 38%). Among the 12 pts in the AXI arm who crossed-over at the time of progression, BORR was 0%, following cross-over SD was documented in 8 pts, median PFS 14wks (range 11 to 34). AXILOM pts were at higher risk for grade 3/4 adverse events (most frequent gr3/4 AE on AXILOM vs AXI were: thrombocytopenia 3- vs 0 pts, hypertension 2 vs 2 pts, anorexia 3 vs 3 pts). Conclusions: combination of axitinib and lomustine does not improve the outcome of pts with rGB as compared to axitinib monotherapy, regardless of the MGMT promoter methylation status. The previously reported activity and safety profile of axitinib monotherapy in pts with rGB was confirmed. Clinical trial information: NCT01562197.

2039

2040

Poster Session (Board #227), Sat, 1:00 PM-4:30 PM

Poster Session (Board #228), Sat, 1:00 PM-4:30 PM

Association of increased survival in glioblastoma patients treated with dendritic cell vaccinations and temozolomide with increased activity of NK cells and ABCC3 expression. First Author: Gaetano Finocchiaro, Fondazione IRCCS Istituto Neurologico Carlo Besta, Milan, Italy

A clinical variable-based predictive model of delayed clearance of high dose methotrexate (HDMTX) in patients (pts) with primary CNS lymphoma (PCNSL). First Author: Yu Yuan, Wake Forest Baptist Medical Center, Winston-Salem, NC

Background: In the two stage phase I-II study DENDR-I, 24 patients with first diagnosis of glioblastoma (GBM) were treated with DC immunotherapy associated to standard radio-chemotherapy. To go to stage II progressionfree survival (PFS) 12 months after surgery (PFS-12) was expected in at least 8 cases. Evidence of immune response was investigated. Methods: Patients with post-surgery volume # 10 cc underwent leukapheresis before radiotherapy and chemotherapy with temozolomide (TMZ). Three intradermal injections of mature DC loaded with whole tumor lysate were done before and four after adjuvant TMZ. Results: The treatment was well tolerated. After a median follow up of 18.8 months 10 patients reached PFS12 (responders). Median overall survival (OS) was 19.4 mo comparing favorably with OS expected using the EORTC nomogram model 3 of 13.0 mo (p = 0.01). The immune activation of effector cells was assessed by intracellular staining of interferon-gamma (IFN-g) in PBLs of 5 responders and 6 non-responders. In responders DC vaccination induced a significant expansion of NK cells producing IFN- g after the second vaccine (p = 0.02 vs 1st vaccine). NK cells continued to produce high levels of IFN-g in response to and at the end of treatment (P = 0.004, responders vs non-responders). CD8+ T cells underwent a rapid expansion and produced higher levels of IFN-g compared to baseline (p , 0.02), but after the first administration of adjuvant TMZ primed CD8+ T cells failed to form an effector memory phenotype (CCR7negCD45RAnegCD62Llow/neg).We also studied in human PBL the expression of the multidrug-resistance protein Abcc3, conferring chemotherapy resistance to murine NK cells. Using Real Time-PCR, Abcc3 expression was found higher in responders vs non-responders before and after radiochemotherapy (p = 0.0005 and p = 0.02, respectively) and during vaccination and adjuvant TMZ (p , 0.01). Conclusions: These findings demonstrate that in DENDR-I patients with longer PFS DC vaccinations increase the number of active NK cells and the expression of Abcc3 and suggest that adjuvant chemotherapy interferes with effector memory formation.

Background: HDMTX remains the backbone of therapy for pts with PCNSL and delayed clearance may lead to higher rates of treatment-induced toxicities. It is unknown whether readily available pre-treatment clinical variables other than creatinine clearance (CrCl) are predictive of delayed clearance. Methods: A retrospective single-center study of pts (Age $ 18 years) receiving HDMTX ($ 1 g/m2, range 1-8g/m2) with biopsy confirmed PCNSL was performed. Variables studied included patient demographics (age, gender, ECOG); past medical history (DM, HTN, CHF, liver disease, MRI with contrast within 48 hours); MTX history (MTX dose [mg/m2], cycle number, number of prior HDMTX cycles, days between HDMTX cycles, history of prior delayed clearance [MTX level of $ 0.30 micromol/L at 48 hours]); lab abnormalities (CrCl, elevated liver enzymes, hypoalbuminemia [, 3.5g/dL], low total protein [ , 6 g/dL]); and medication use (steroid, rituximab, and concurrent use of renal tubular competitors). Results: We identified 34 pts with PCNSL who received 191 cycles of HDMTX from 2003-2014. 75/191 cycles had delayed clearance at 48 hours. Significant variables identified in univariate analysis (age, pretreatment creatinine, DM, HTN, hypoalbuminemia) were used to construct a multivariate model which demonstrated that delayed MTX clearance was significantly more common in pts with a history of DM (p = 0.0002, OR 1.32, 95% CI [1.14, 1.54]), HTN (p = 0.0013, OR 1.25, [1.09, 1.43]), and hypoalbuminemia (p = 0.025, OR 1.18, [1.02, 1.36]). Conclusions: Higher rates of delayed clearance following HDMTX administration were seen in pts with a history of DM, HTN, and hypoalbuminemia. A larger sample size is needed to generate a quantitative model from these preliminary observations that would allow individualized dosing modifications to avoid delayed MTX clearance.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Central Nervous System Tumors 2041

Poster Session (Board #229), Sat, 1:00 PM-4:30 PM

Safety and antitumor activity of hypofractionated stereotactic irradiation (HFSRT) with pembrolizumab (Pembro) and bevacizumab (Bev) in patients (pts) with recurrent high grade gliomas: Preliminary results from phase I study. First Author: Solmaz Sahebjam, Moffitt Cancer Center, Tampa, FL Background: High grade gliomas carry a grim prognosis despite current therapies. Several studies have shown that the combination of anti-PD-1 antibody with stereotactic radiosurgery generates robust and durable responses and improves survival in preclinical models of glioma. Pembro is a highly selective humanized monoclonal antibody against PD-1. This study evaluates the safety and anti-tumor activity of Pembro administered in combination with Bev and HFSRT in pts with recurrent high grade gliomas. Methods: This phase I study (3+3 design) explores the safety, tolerability, recommended phase II dose (RP2D), and antitumor activity of Pembro administered concurrently with HFSRT and Bev (NCT02313272). Adult pts with recurrent glioblastoma (GBM) or anaplastic astrocytoma (maximum diameter of target lesion # 3.5 cm) are eligible. Eligible patients receive HFSRT to the recurrent tumor (30 Gy delivered in 5 fractions) combined with Bev (10 mg/kg, intravenously Q2W) and Pembro (100 mg or 200 mg intravenously based on dose level, Q3W). Two dose levels of Pembro (100 mg and 200 mg Q3W) are explored. Once the RP2D is determined, an additional 20 patients will be enrolled in an expansion cohort. Response is assessed every 6W per RANO criteria. Effect of treatment on quality of life measures are evaluated. Results: As of Feb 2016, 6 pts with GBM were enrolled: 3 at Pembro 100 mg, 3 at Pembro 200 mg dose levels. No dose limiting toxicity or dose modifications were reported. Permanent discontinuation due to treatment-related AEs has not occurred. No treatment related neurologic AEs were observed. The most common drug-related AE was grade 1 fatigue (2 out of 6). At the time of abstract submission, all pts are continuing on study treatment and 3 pts (at Pembro 100 mg dose level) are evaluable for response. Durable disease control is seen in all 3 pts, including 1 CR (24+W) and 2 SDs (24+ W and 22+ W). Conclusions: Preliminary data demonstrate an acceptable toxicity profile for combination of HFSRT with Pembro and Bev in pts with recurrent high grade gliomas. Updated safety, efficacy data will be presented. Clinical trial information: NCT02313272.

2043

Poster Session (Board #231), Sat, 1:00 PM-4:30 PM

Safety, efficacy and survival of patients (pts) with primary CNS tumors in phase 1 (Ph1) trials: A 12-year single institution experience. First Author: Niamh Coleman, Royal Marsden Hospital, London, United Kingdom Background: Malignant primary CNS tumors constitute less than 2% of all malignancies and carry a dismal prognosis. Treatment (tx) options at relapse are limited. First-in-human solid tumor studies have historically excluded pts with CNS tumors due to the poor prognosis, concomitant drug interactions, concerns about excessive toxicities and limited efficacy. Methods: Retrospective data were collected on clinical tx and tumor characteristics of pts referred for consideration of Ph1 trials in the Royal Marsden Hospital between 06/2004 and 01/2016. Survival analyses were performed using the Kaplan-Meier method, Cox proportional hazards model, and associations with relevant clinical prognostic parameters were tested with Fischer’s Exact test. Statistical analysis was performed using Graphpad Prism 6.0 and SPSS 23. Results: 117pts with advanced, refractory primary CNS tumors were referred. At initial consultation, pts had a median ECOG PS of 1, median age of 49 years (range 18-70); 81pts (69%) were men, 90 (77%) had glioblastoma; 71pts (61%) were on anti-epileptic drugs, 76pts (65%) required steroid therapy. The median number of prior systemic therapies was 2. 75 pts (64%) were allocated to a Ph1 trial; 55 (73%) were treated with a study drug. 25pts (21%) were not allocated due to lack of slots on relevant trials and 14 (12%) were not allocated due to poor PS. The median PFS for pts treated on a Ph1 trial was 6.5 months (m) with a median OS of 11m (vs 4m in pts that did not proceed with a Ph1 trial) HR 0.49, 95% CI:0.27-.0.67, p = 0.0005. 5 pts (9%) had grade 3 drug-related toxicities. There were no study-related deaths. No correlation between RMH score, age, ECOG PS, previous lines of tx or neutrophil -lymphocyte ratio and outcome was found. Conclusions: We report a significant survival benefit for pts with primary CNS tumors treated on Ph1 trials. Toxicity and efficacy outcomes are comparable to the general Ph1 pt population. In the absence of an internationally recognized standard second line tx for pts with recurrent primary malignant CNS tumors, more Ph1 trials should allow enrolment of these pts and Ph1 trial participation should be considered at an earlier stage.

2042

103s Poster Session (Board #230), Sat, 1:00 PM-4:30 PM

EGFR gene amplification and variant III (EGFRvIII) mutation in primary and recurrent glioblastoma. First Author: Michael Weller, University Hospital Zurich, Zurich, Switzerland Background: Around 40% of glioblastomas exhibit amplification of the epidermal growth factor receptor (EGFR) gene; half of these tumors carry a specific genetic rearrangement characterized by deletion of exons 2-7 (EGFRvIII). The prognostic role of EGFRvIII in patients with EGFR-amplified glioblastoma and its expression in recurrent as opposed to newly diagnosed glioblastoma have not been defined. Such data are highly relevant for the clinical development of novel EGFRvIII-targeting agents. Methods: EGFRamplified glioblastomas from 109 patients treated according to standard of care were assessed for EGFRvIII expression and data were correlated with outcome. Pairs of primary and recurrentglioblastomas from 38 of these patients (including 25 patients with EGFRvIII-positive tumors) and from 33 glioblastoma patients without EGFR amplification were evaluated for changes in EGFR amplification and EGFRvIII expression at recurrence. Results: Among 109 patients with EGFR-amplified primary glioblastoma, 63 had EGFRvIII-positive tumors by immunohistochemistry or RT-PCR or both. The detection of EGFRvIII was not associated with inferior progressionfree survival (PFS) (HR = 0.85, 95%CI 0.58-1.25, p = 0.418) or overall survival (OS) (HR = 0.99, 95%CI 0.67-1.47, p = 0.961). EGFRvIII status did not change between primary and recurrent glioblastomas in 21 of 25 EGFRvIII-positive patients (84%). In 4 of 25 patients (16%), EGFRvIII positivity was lost at recurrence. One of 13 EGFR-amplified but primarily EGFRvIII-negative glioblastomas demonstrated EGFRvIII-positivity at recurrence. None of 33 initially EGFR non-amplified glioblastomas acquired EGFR amplification or EGFRvIII positivity at recurrence. Conclusions: EGFRvIII expression is not prognostic in patients with EGFRamplified glioblastomas. Most EGFRvIII-positive glioblastomas maintain positivity at recurrence. However, EGFRvIII expression may be lost or newly gained in small subsets of patients at recurrence. Thus, re-biopsy with reassessment of EGFRvIII status is recommended for recurrent glioblastoma patients enrolled in clinical trials using EGFRvIII-targeting agents.

2045

Poster Session (Board #232), Sat, 1:00 PM-4:30 PM

Feasibility and efficacy of concomitant chemoradiation (Ch-RT) in the management of newly diagnosed elderly glioblastoma (GB) patients: Results from the GLIOCAT study. First Author: Maria Martinez-Garcia, Medical Oncology, Hospital del Mar, Barcelona, Spain Background: Elderly GB patients have a dismal prognosis. The optimal treatment in this population remains to be established. Methods: We performed a multicenter retrospective and prospective study of newly diagnosed GB patients (pts) treated with standard Ch-RT (Stupp regimen). In the present substudy we analyzed clinical outcome and prognostic factors in elderly ( . 65 y) pts and compare these to younger pts from the global cohort. Results: Between 2005 to 2014, 432 pts were enrolled. There were 148 . 65y; 117 (79%) 65-75y and 31 (21%) . 75y with a mean age of 72y, mean KPS 80%, and 58.8% were males. In pts . 65y, gross total resection (GTR) was performed in 34 (24.5%), and biopsy in 29 (20.9%) with postoperative complications in 17 (11.5%). The time to initiate Ch-RT was 4.93 weeks (w); 127 (85.8%) finished Ch-RT and 109 (73.6%) completed adjuvant temozolomide (TMZ). Only 16 (10.8%) could discontinue steroids during the concomitant therapy. There were 28 (18.9%) patients with pseudoprogression. MGMT methylation status was studied in 103 pts of which 57 (55.3%) were methylated. The only statistical differences between . 65 and younger pts were: lower KPS (p = 0.01) and fewer pts initiating adjuvant TMZ (p = 0.015). Median follow-up is 16.76 months. In the global population progression-free survival (PFS) and overall survival (OS) were 8 months (95% CI, 7.49-8.5) and 15 months (95% CI, 7.49-8.5), respectively, compared to 7 (95% CI, 6.1-7.8), and 11 months (95% CI, 8.7-13.2) in . 65y (p , 0.001 and p = 0.033). For elderly pts, on multivariate analysis, GTR and methylated MGMTbut not KPS were independent predictors of OS and PFS, respectively. Conclusions: Standard Ch-RT is feasible in elderly pts even though this population had poor KPS and fewer pts that could start adjuvant TMZ. As expected, PFS and OS were poorer in . 65y. For elderly patients, type of resection and MGMT methylation remain the more relevant prognostic factors in newly diagnosed GB treated with Ch-RT and adjuvant TMZ. Randomized prospective trials are warranted.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

104s 2046

Central Nervous System Tumors Poster Session (Board #233), Sat, 1:00 PM-4:30 PM

Phase I study of single agent ibrutinib in recurrent/refractory primary (PCNSL) and secondary CNS lymphoma (SCNSL). First Author: Christian Grommes, Memorial Sloan Kettering Cancer Center, New York, NY Background: PCNSL is an aggressive primary brain tumor. Outcome and treatment options are poor for patients with recurrent/refractory disease. Response rates range between 30-60% with a progression free survival (PFS) of 2-6 months. Ibrutinib has shown promising clinical response in some Bcell malignancies. This phase I trial investigates ibrutinib in patients with recurrent/refractory (r/r) PCNSL and secondary CNS lymphoma (SCNSL). Methods: Eligible patients had r/r PCNSL or SCNSL, age $ 18, KPS $ 50, normal end-organ function, and unrestricted number and type of prior therapies. In patients with SCNSL disease, systemic disease needed to be absent. Results: Ten patients were enrolled (3 at 560 mg; 7 at 840 mg). Median age was 70 (range 21-80); 7 were women. Median ECOG was 1 (0: 1, 1: 5, 2: 4). 70% had PCNSL and 30% SCNSL. Six had parenchymal disease, 3 isolated cerebrospinal fluid (CSF) involvement and 1 both. Patients were highly pretreated: 2 (both SCNSL) had prior stem cell transplant; 3 (2 PCNSL, 1 SCNSL) prior radiation. The median prior CNS directed therapy was 2; all methotrexate regimens. Two patients withdrew despite clinical and radiographic response. Treatment has been well tolerated with 2 grade 4 toxicities (neutropenia and lymphopenia) that resolved after the drug was held. The most common toxicities observed were hyperglycemia, hypercholesteremia, and thrombocytopenia. After a median follow-up of 150 days, 9/10 patients were evaluated for response: 4 CR (3 in CSF; 1 in the parenchyma), 3 PR, and 2 PD; 78% overall response rate. In 2 patients response has not been confirmed in a 2ndtreatment assessment. Median time to first response was 28 days. The median PFS is 6 months. A higher CSF drug concentration was observed at higher doses of Ibrutinib and at steady state. Continued enrollment is ongoing at 840 mg in an expansion cohort. Molecular testing is ongoing to associate genomic alterations and outcome. Conclusions: Patients with CNS lymphoma tolerate Ibrutinib at 560 and 840mg well. Targeted agents might be an alternative therapeutic approach to be investigated for refractory/recurrent CNS lymphoma patients. Clinical trial information: NCT02315326.

2048

Poster Session (Board #235), Sat, 1:00 PM-4:30 PM

Radiologic progression types are treatment specific: An exploratory analysis of a phase 3 study of bevacizumab plus radiotherapy plus temozolomide for patients with newly diagnosed glioblastoma (AVAglio). First Author: Martha Nowosielski, Medical University Innsbruck, Innsbruck, Austria Background: Thetype of radiological progression following bevacizumab treatment failure is related to outcome in patients with progressive glioblastoma. In an exploratory analysis of the randomized, placebo-controlled phase III clinical trial in newly diagnosed glioblastoma of bevacizumab (bev) plus radiotherapy/temozolomide (AVAglio) we aimed to investigate whether these progression types (PT) are induced by therapy (anti-angiogenic versus sole cytotoxic chemotherapy), are related to outcome and may therefore reflect a particular resistance mechanism. Methods: MRI scans (baseline until IRF progress) of 621 patients (bev arm, n = 299; placebo arm, n = 322) were available for analysis in sufficient detail and quality. PT were categorized according to contrast enhancement behavior in T1 and T2 hyperintense signal change as published before (Nowosielski et al, Neurology 2014). Frequencies of the 5 PTs (cT1 flare-up, classic T1 progress, T2diffuse, T2-circumscribed and primary non-responder), time to development (PFS) and overall survival (OS) were assessed within each treatment arm. Results: Classic T1 progress and primary non responder were more common in the placebo than in the bev arm (56.8 % versus 25.1% and 21.7% versus 3.3%, respectively.) In the bev arm, cT1 flare up and the T2 circumscribed PT were more common (37.5% versus 10.9% and 21.7%. versus 3.4%, respectively). T2 diffuse was equally distributed between the treatment arms and showed longer PFS compared to the other PTs in both arms. In the bev arm, cT1 flare up as well as T2 diffuse showed longer OS than all the other PTs (17.8 months vs median in T2 diffuse not yet reached). Conclusions: Exploratory analysis of AVAglio imaging data showed that radiologic progression types seem to be specific for anti-angiogenic or temozolomide treatment failure. These progression types show differences in PFS and OS and may therefore reflect a particular resistance mechanism that may guide further post progression therapy. Clinical trial information: NCT00943826.

2047

Poster Session (Board #234), Sat, 1:00 PM-4:30 PM

Does the duration of adjuvant temozolomide impact survival in glioblastoma patients? A retrospective analysis from an Italian registry (GLIOSTRY). First Author: Roberta Ruda, Department of Neuro-Oncology, University of Turin and City of Health and Science, Turin, Italy Background: The Stupp regimen for newly diagnosed glioblastoma (GBM) consists of radiotherapy with concomitant and adjuvant temozolomide (TMZ) up to 6 cycles. A number of Institutions prolong the duration of the adjuvant chemotherapy up to 12 or more cycles, but it is unknown whether the prolongation of the adjuvant phase favorably impacts the length of survival. The aim of this retrospective study was to investigate this issue in an homogeneous population of GBM patients (age 18-70 years), receiving the Stupp regimen and the nitrosourea fotemustine at first relapse. Methods: Patients were collected through a national registry (Gliostry), that was established by the Italian Association for Neuro-Oncology (AINO). The survival of 556 patients receiving up to 6 cycles of TMZ was compared with that of 365 patients receiving more than 6 cycles of TMZ. Results: Median PFS following the Stupp regimen was 8.4 months for the group TMZ # 6 cycles and 14.3 months for the group TMZ ˃ 6 cycles (p , 0.001). The advantage in terms of PFS for patients receiving more than 6 cycles of TMZ was higher among patients with MGMT methylation (p , 0.001) and incomplete surgical resection (p , 0.001). Median OS following the Stupp regimen was 16.7 months for the group TMZ # 6 cycles and 26.3 months for the group ˃ 6 cycles (p , 0.001). The advantage in terms of OS for patients receiving more than 6 cycles of TMZ was higher among patients with MGMT methylation (p , 0.001) and incomplete resection (p , 0.001). The probability of response to salvage fotemustine was not influenced by the duration of adjuvant TMZ. Multivariate analysis of both PFS and OS following the Stupp regimen did not show an independent prognostic value for the duration of adjuvant TMZ. Conclusions: The prolongation of adjuvant TMZ with more than 6 cycles could be advantageous only in patients with MGMT methylation and in those with incomplete resection.

2049

Poster Session (Board #236), Sat, 1:00 PM-4:30 PM

Polymorphisms risk modeling for vascular toxicity in patients with glioblastoma treated on NRG Oncology/RTOG 0825. First Author: Renke Zhou, Baylor College of Medicine, Houston, TX Background: Vascular toxicities, including hypertension and thrombosis, are commonly associated with anti-angiogenic agents, such as Bevacizumab. Risk factors for toxicity have yet to be identified. RTOG 0825 evaluated the use of bev in patients with newly diagnosed GBM. This project evaluated the risk of and validated prediction models for vascular toxicity. Methods: Logistic regression with backward selection (p # 0.05) was used to build clinical models for thrombosis (Grade 3-4) and hypertension (Grade 24). A genome-wide analysis of genetic variation associated with the vascular toxicities was conducted. The top hits were incorporated into the risk modeling. Tagging variants were selected and examined individually and jointly and a final model developed. Results: Clinical and genotype data from 367 cases were included. The final clinical model for thrombosis included anticonvulsant use, body surface area, and blood urea nitrogen. Bev treatment was not associated with increased risk (12.1% vs 12.6%, bev vs placebo). The genome scan revealed five polymorphisms in protein coding genes of Tetraspanin 33, G protein-coupled receptor, and Cadherin 8 that significantly associated with thrombosis (p , 10-7). Significant genetic variants with adjustment for clinical variables resulted in the best model (AUC = 0.81), yielding a 1.7-fold increase in risk of thrombosis (p , 0.0001, 95% CI: 1.4-2.1). For hypertension only treatment arm (14.8% vs 7.9%, bev vs placebo) remained in the final clinical model. The two most significant variants (p , 10-7) generated the best model (AUC = 0.79), showing a 6.1fold increase in risk of hypertension per risk allele present (p , 0.0001, 95% CI: 3.3-11.1). Conclusions: Despite the absence of increased thrombosis with bev treatment, potential clinical and polymorphisms that predict increased risk were uncovered. In contrast, hypertension was highly associated with bev and risk assessment was enhanced by polymorphisms. These results support investigation of clinical and genomic factors to help optimize risk–benefit analyses of treatment options. Supported by NCI U10CA21661, U10CA180868, U10CA180822, U10CA37422, U24CA114734, NIH 1R01NR013707 and Genentech.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Central Nervous System Tumors 2050

Poster Session (Board #237), Sat, 1:00 PM-4:30 PM

Inter-interpreter reliability of neuropathological assessment of disease status after early repeat resection for suspected recurrent glioblastoma. First Author: Matthias Holdhoff, The Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins, Baltimore, MD

2051

105s Poster Session (Board #238), Sat, 1:00 PM-4:30 PM

Bendamustine for anaplastic gliomas. First Author: Marc C. Chamberlain, University of Washington, Seattle, WA

Background: Distinction of true progression from pseudo-progression in patients with suspected recurrent glioblastoma (rGBM) is an important and common diagnostic challenge in neuro-oncology. The “gold standard” for assessment of post-treatment disease status has remained surgery with histologic evaluation. However, after chemoradiation, tissue frequently contains a variable mixture of scar tissue and tumor cells. Neuropathologists are left to determine whether patients have active tumor or treatment effect. As validity and reliability are critical for any diagnostic test, our study assessed inter-interpreter reliability in the pathological assessment of disease status in this setting. Methods: Histopathology images from 13 patients with GBM who underwent surgery within one year of diagnosis for suspected rGBM were selected. 101 neuropathologists at 50 NCI-designated Cancer Centers received a case scenario of a patient with suspected rGBM, including histopathological information of the initial diagnosis. Participants were asked to grade each case for percentage of 1) active tumor, 2) treatment effect, 3) cellularity, 4) mitotic activity, and 5) to give an overall assessment of disease activity (‘active tumor’ vs. ‘pseudo-progression’ vs. ‘unable to classify’). Fleiss’ kappa statistics were calculated to measure the interinterpreter reliability among all readers. Kappa , 0.4 = marginal agreement, . 0.75 = excellent agreement. Results: We received responses from 48 pathologists from 30 centers, of whom 44 were neuropathologists by training (48% response rate). The kappa score for agreement in overall assessment of disease activity was 0.228. Individually, the Kappa scores were 0.288 for ‘active tumor’, 0.323 for ‘pseudo-progression’, and 0.036 for ‘unable to classify’. Conclusions: Our findings demonstrate marginal interinterpreter reliability in pathological assessment of disease status in rebiopsied GBM. More formal and consistent criteria are urgently needed as inconsistent assessment of disease status directly impacts on patient selection for clinical trials, as well as on decision making in clinical practice.

Background: There is no standard therapy for recurrent anaplastic glioma (AG). Salvage therapies include alkylator-based chemotherapy (temozolomide [TMZ] or lomustine [CCNU]), re-resection with or without carmustine implants, re-irradiation and bevacizumab. Bendamustine is a novel bifunctional alkylator with CNS penetration but never evaluated in AG. Objective: Assess response and toxicity of bendamustine in bevacizumab na¨ıve recurrent AG following prior radiotherapy (RT) and TMZ in a prospective Phase II trial. Methods: 26 adults (16 males; 10 females: median age 40 years {range 30-65}) with RT and TMZ refractory recurrent AG (14 anaplastic astrocytoma, 7 anaplastic oligodendroglioma, 5 anaplastic oligoastrocytoma) were treated with bendamustine. 12 patients were treated at 1st and 14 at 2nd recurrence. Prior salvage therapy included re-resection in 14 (12 with 2nd and 2 with 3rd resection), chemotherapy in 11 and re-RT in 2. A cycle of bendamustine was defined as 2 consecutive days of treatment (100mg/m2/day) administered once every 4 weeks (maximum number of cycles 6). Success of treatment was defined as progression free survival (PFS) at 6 months . 40%. Results: Grade 3 treatment-related toxicities included lymphopenia (11 patients), myalgia, pneumonia, diarrhea, leukopenia, allergic reaction and thrombocytopenia in 1 patient each. Three patients had grade 4 lymphopenia. No grade 5 toxicities were seen. One patient discontinued therapy due to toxicity (allergic reaction). Bendamustine dose was delayed in 3 patients (total of 5 events). There were no dose reductions. The median number of cycles of therapy was 3 (range 1-8). Best radiographic response was progressive disease in 12 (46%), stable disease in 13 (50%) and partial response in 1 (4%). Response did not differ by histology. Median PFS was 2.7 months (range 1-52 months), 6-month PFS was 27% and 12-month PFS was 8%. Conclusions: In this small prospective series of patients with recurrent AG refractory to TMZ and bevacizumab naive, bendamustine appears to have modest single agent activity though not meeting pre-specified criteria (40% 6-month PFS) and manageable toxicity. The study was supported in part through research funds provided by TEVA Pharmaceuticals and the National Comprehensive Cancer Network. Clinical trial information: NCT00823797.

2052

2054

Poster Session (Board #239), Sat, 1:00 PM-4:30 PM

Effect of controlled intratumoral viral delivery of Ad-RTS-hIL-12 + oral veledimex in subjects with recurrent or progressive glioma. First Author: Francois M. Lebel, ZIOPHARM Oncology, Boston, MA Background: The prognosis of subjects with glioblastoma multiforme (GBM) after first recurrence is extremely poor with a median overall survival (OS) of approximately 90 - 160 days. New therapies are needed, thus we developed an adenoviral vector, Ad-RTS-IL-12 (AD), administered intratumorally under the control of the RheoSwitch Therapeutic System (RTS) expression platform. Gene expression and IL-12 protein production is tightly controlled by activator ligand veledimex (V) and in our mouse model of glioma we have consistently shown a dose-response and survival benefit compared to temozolamide, bevacizumab, and a PD-1 inhibitor. Methods: In a multicenter Phase I dose escalation trial, subjects with recurrent or progressive Grade III or IV glioma undergoing resection were injected intratumorally with AD 2x1011viral particles and daily oral V for 15 doses, beginning on Day 0, prior to surgery. The study primary endpoint is safety and tolerability of AD+V with secondary endpoints including OS. Results: In the first cohort, consisting of 4 males and 3 females (ages 32-58), it was observed that V crossed the blood-brain barrier with 3066% of the V plasma level in the resected tumor sample (1766 ng/ml vs. 562 ng/ml). Peak serum IL-12 was observed at Day 3 (mean 23 pg/ml) and peak IFNg (mean 32 pg/ml) on Day 7. Peripheral blood samples showed an increase in CD8 T cells and an increase in the ratio of CD8/CD4 and CD8/FoxP3. AD+V was well tolerated with expected toxicities promptly reversing with discontinuance of V. Neurotoxicities were minimal and manageable. To date, 7/7 subjects are alive with 5 subjects having a follow-up of . 90 days and 3 . 160 days posttreatment. Conclusions: Intratumoral regulated IL-12 expression using AD+V in a salvage population with advanced and recurrent glioma has an acceptable and rapidly reversible safety profile. The intracranial administration of AD followed by oral V is clinically feasible and suggests possible benefit vs. historical control and provides a strong rational for dose-escalation as reviewed and approved by the independent Data Safety Monitoring Board. Clinical trial information: NCT02026271.

Poster Session (Board #241), Sat, 1:00 PM-4:30 PM

Genetic landscape of extreme responders with anaplastic oligodendroglioma: NRG Oncology/RTOG 9402. First Author: Matthias Holdhoff, The Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins, Baltimore, MD Background: The randomized multicenter RTOG 9402 trial showed a significant survival benefit in patients with 1p/19q co-deleted anaplastic oligodendrogliomas (AO) who received both radiation (RT) and chemotherapy (PCV regimen) versus RT alone. Interestingly, substantial separation of the survival curves was only seen after 7.3 years. We aimed to determine whether there are specific genetic alterations that can distinguish co-deleted AO patients who benefit from the addition of PCV from those who do not. Methods: Whole exome sequencing on matched tumor and normal DNA was performed on all available short-term (STS) and long-term survivors (LTS) who received RT+PCV. The hTERTand risk allele (G) of rs55705857 status of both cohorts of patients were also analyzed. Results: Seven STS (5 with survival of , 7.3y, 2 with later progression) and 8 LTS (7 with survival of $ 7.3y without progression) had sufficient material for whole exome sequencing. Several specimens could not be included due to limited quantities or degradation of the material. Among the specimens studied, there was no significant difference between the groups regarding age, performance status and extent of resection. The average number of somatic mutations for STS vs. LTS was 4 vs. 7. Most common mutations identified were IDH1 (in 57% STS and in 88% LTS), CIC (29% and 75%, respectively) and FUBP1 (71% and 50%, respectively). The hTERT promoter was mutated in 71% STS and in all LTS. Risk allele (G) of rs55705857 was identified in all patients (GA in 43% and 25%, respectively). Intriguingly, mutations of ESX1were found only in the LTS (38%) and not the STS group. Conclusions: Our results did not identify a distinct genetic signature that correlates with improved benefit from chemotherapy with PCV in patients with newly diagnosed 1p/19q codeleted AO; however, differences in alterations of ESX1 warrant further investigation.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

106s 2055

Central Nervous System Tumors Poster Session (Board #242), Sat, 1:00 PM-4:30 PM

Phase 1/2 trial of bevacizumab plus TPI 287, a brain penetrable antimicrotubule agent, in patients with recurrent glioblastoma. First Author: Samuel Aaron Goldlust, John Theurer Cancer Center, Hackensack, NJ Background: Microtubule inhibitors have demonstrated efficacy in preclinical models of glioblastoma (GBM), however, clinical benefit is hampered by subtherapeutic dose delivery. Designed to overcome this limitation, TPI 287 is a third-generation taxane that readily penetrates the blood brain barrier and evades mechanisms of P-glycoprotein mediated efflux. CB-017 is a multi-center phase 1/2 trial designed to determine the maximum tolerated dose and potential efficacy in patients treated with TPI 287 plus bevacizumab (BEV) for treatment of recurrent GBM. Results of the recently completed dose escalation stage of the trial are reported. Methods: GBM patients at first or second relapse after standard therapy and without prior exposure to anti-angiogenic agents were eligible for enrollment. BEV was administered at 10 mg/kg every 2 weeks and TPI 287 every 3 weeks via IV infusion. MRIs were obtained every six weeks with response assessed via RANO. TPI-287 dose escalation was based upon a standard 3+3 design. Results: As of January 2016, 24 patients were enrolled in seven TPI 287 dose-escalation cohorts (140-220 mg/m2). No DLTs were reported and myelosuppression (n = 3) was the only drug-related grade 3/4 adverse event. Increased frequency of myelosuppression and peripheral neuropathy in later cohorts, in conjunction with suggested activity, lead to halting dose escalation at 220 mg/m2. Among 20 patients evaluable for response, ORR was 60% (12/20; 3 complete, 9 partial). The median and 6-month PFS were 5.5 mo. [95% C.I. 4.1-6.9] and 37%, respectively. Median and 1-year OS were 12.9 mo. [95% C.I. 10.9-17.9] and 63%, respectively, after 20.1 mo. median follow-up. Conclusions: TPI 287 in combination with BEV is safe and well tolerated at doses up to 220 mg/m2, and survival compares favorably with historical controls. Advancement to the phase 2 stage of the trial is underway. Updated safety and efficacy results will be presented. Clinical trial information: NCT01933815.

2057

Poster Session (Board #244), Sat, 1:00 PM-4:30 PM

2056

Poster Session (Board #243), Sat, 1:00 PM-4:30 PM

Comparison of molecular alteration in glioblastoma tumors from old and young patients. First Author: Joanne Xiu, Caris Life Sciences, Phoenix, AZ Background: GBM patients (pts.) aged 70 yrs or older have a poorer prognosis than younger pts. IDH1/2 mutations are more prevalent in younger GBM pts. and confer a more favorable prognosis. We compared molecular alterations between younger pts. (YP; 18 , age , 45), and older pts. (OP; age . = 70), with additional stratification for wild type IDH1/2 status. Methods: GBM tumors submitted for tumor profiling between 2009 and Jan 2016 were tested with NextGen sequencing (SEQ), immunohistochemistry (IHC), fluorescent in-situ hybridization (FISH), fragment analysis (FA) and promoter methylation (Me). Retrospective analysis was performed on 375 GBM’s (YP = 197, OP = 178). Pediatric tumors were excluded. Chi-square was used for exploratory analyses and significance was defined as p , 0.05. Results: Alterations including overexpression of ALK (29% vs. 4.2%), RRM1 (47% vs. 32%) and mutations of ATRX (73% vs. 11%), BRAF (9.3% vs. 1.7%), IDH1 (24% vs. 3%), PDGFRA (7% vs. 0), PTPN11 (6.6% vs. 1%) and TP53 (58% vs. 26%) were significantly more prevalent in YP (N = 197). In contrast, PTEN mutation was significantly more frequent in OP (N = 178; 26% vs. 13%). PTEN loss by IHC was equal between YP and OP (22% vs. 21%). Pts. with known wild type IDH1/2 status were compared between YP and OP (N = 72 and 95). Significantly higher expression of TOPO1 was seen in YP (63% vs. 45%) while MGMT-Me was less common in YP (29% vs. 48%). The differences in ALK IHC (27% vs. 4.5%), mutations of BRAF (12% vs. 1%), PDGFRA (4%, vs. 0), PTPN11 (7% vs. 1%) and TP53% (42% vs. 25%) were significant. cKIT and PDGFRA co-amplifications were seen in 2 out of 3 tumors from OP with wild type IDH1/2 while no cKIT or PDGFRA amplification was seen in YP (0/6). No significant differences were seen in YP vs. OP for EGFRvIII (12% vs. 14%), PD-L1 expression on tumor cells (19% vs. 8.3%) or PD-1 expression on TIL (42% vs. 54%). Conclusions: Significant molecular differences were seen between older and younger GBM pts. Alterations seen in OP, including PTEN mutation and cKIT/PDGFRA coamplification were previously associated with recurrent tumors and poor survival, and may underlie the poor prognosis observed in OP. These results suggest and may guide which pts. will benefit from targeted therapies in future studies.

2058

Poster Session (Board #245), Sat, 1:00 PM-4:30 PM

Oncogenic correlates of epilepsy in glioblastoma. First Author: Sharon Berendsen, University Medical Center Utrecht/Brain Center Rudolf Magnus, Utrecht, Netherlands

Correlation of IDH mutation, 1p/19q co-deletion and 18FET-PET derived time-to-peak analysis as prognostic markers in glioma First Author: Bogdana Suchorska, Department of Neurosurgery LMU, Munich, Germany

Background: Epilepsy at presentation is a favorable prognostic factor in glioblastoma. In literature, deregulated expression of serum response factor (SRF) has been associated with epilepsy in animal models, and seems to enhance proliferation and invasion in multiple types of cancer. In this study, we analyze the biological mechanisms that associate with epilepsy in glioblastoma patients, and which might thus underlie this prognostic effect. Methods: Molecular classification, mRNA and miRNA expression-based (gene set) enrichment analyses were performed on fresh frozen tissue obtained from 76 tumors. Hospital records were screened for the presence of seizures at presentation of the disease. Protein expression patterns of gene set targets (ATRX, STAT5b) and markers of epithelial-mesenchymal transition (NF-kB, C/EBP-b and STAT3) were determined with immunohistochemistry on tissue microarrays containing 339 tumors. Results: Gene sets involved in hypoxia/HIF1-a, SRF, C/EBP-b and LPS-induced signaling pathways were significantly downregulated in glioblastoma patients who presented with epilepsy. On a protein level, epileptogenic tumors were characterized by a significant downregulation of phospho-STAT5b, a target of HIF1-a. No differential expression between patients with or without epilepsy was observed for ATRX, which is a target of SRF and has been associated with prognosis of glioma patients. Epilepsy status did not associate to EMT protein expression, molecular classification or miRNA expression. Conclusions: Epileptogenic GBMs show a downregulation of HIF1a/STAT5b and SRF signaling compared to glioblastomas that do not present with epilepsy. These genes are known to play oncogenic roles in glioblastoma and other types of cancer, and SRF, which is deregulated in non-tumoral epileptic brain, might provide insight in the tumor’s epileptogenicity.

Background: Stratification of glial tumors according to IDH mutation and 1p/ 19q co-deletion status will gain further importance for tumor classification and treatment in the near future. In analogy, factors derived from 18FET-PET uptake dynamics such as time-to-peak (TTP) analysis allow discrimination between different prognostic groups. The present study aims at correlating molecular tumor characteristics and TTP analysis and to analyze associations with clinical outcome. Methods: 300 patients with de novo WHO grade II-IV gliomas with 18FET-PET imaging at diagnosis were grouped according to IDH mutation and 1p/19q co-deletion status into 3 subgroups (IDH wildtype, IDH mut/non co-del 1p/19q and IDH mut/co-del 1p/19q). Clinical and imaging factors such as age, Karnofsky performance status, treatment and 18FET-PET derived TTP analysis were analyzed with regard to progression-free and overall survival (PFS and OS) via univariate and multivariate regression analysis. Results: PFS and OS was longest in the IDH mut/co-del 1p/19q subgroup followed by IDH mut/non co-del 1p/19q and IDH wt patients (p , 0.0001). Furthermore, outcome stratified by mean TTP of 17.5 minutes revealed significantly longer PFS and OS in patients with TTP . 17.5 minutes (p , 0.0001 for PFS and OS). Outcome analysis according to the mean TTP of 17.5 minutes within the three subgroups showed TTP to be correlated with PFS/OS within the IDH mut/ non co-del 1p/ 19q subgroup (p = 0.03 for OS/PFS), but not within the IDH mut/ co-del 1p/ 19q (p = 0.32 for PFS/p = 0.26 for OS) or the IDH wt (p = 0.63 for PFS and p = 0.05 for OS) subgroup. Conclusions: 18FET-PET derived dynamic analysis delivers additional prognostic information in the sub-group of IDH mut/ no co-del 1p/19q glioma concerning both progression free and overall survival.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Central Nervous System Tumors 2059

Poster Session (Board #246), Sat, 1:00 PM-4:30 PM

Profiling response to carmustine/bevacizumab in recurrent glioblastoma. First Author: Leonardo Rojas, Oncology Department, Centro Javeriano de Oncologı´a, Hospital Universitario San Ignacio, Bogota, Colombia Background: The prognosis for patients (pts) with recurrent glioblastoma is still poor with a median survival between 3 and 6 months. Angiogenesis is crucial for glioblastoma (Gb) growth, and anti-vascular endothelial growth factor agents are widely used in recurrent Gb pts. Following the results of BELoB trial we explored the efficacy of carmustine plus bevacizumab considering the prognostic value of several biomarkers. Methods: This retrospective study included 59 adult pts with histologically confirmed Gb who were treated with carmustine (BCNU 80 mg/m2 days 1-3 every 8 weeks) in combination with bevacizumab (10 mg/kg every 15 days) as a second line regimen. Response rate (RR), 6-month progression free survival (PFS6), progression-free survival (PFS) and overall survival (OS) were assessed according to CD133 mRNA expression, MGMT methylation (pMGMT) and IDH1/2, p53 and EGFRvIII mutation status. Results: The median age was 43-yo (23-70 years), 35 patients were male (59%), 78% of cases were treated with debulking surgery and 22% had undergone biopsy. The median follow-up was 18.6 months (95%CI 6.4-36), overall RR to BCNU/Bev combination was 54%, median PFS was 8,6 months (IC95% 2.4-14.7) and median OS was 27.1 months (IC95% 23.3-28.6). IDH1/2 mutations were found in 30%, pMGMT in 56%, p53 mutations in 29%, EGFRvIII in 24% and high CD133 mRNA expression in 57%. The variables that positively affected PFS were PS (p = 0.015), IDH+ (p = 0.05) and pMGMT+ (p = 0.009). Similarly, OS was significantly modified by the extent of surgical intervention (p = 0.005) and pMGMT+ (p = 0.05). Multivariate analysis showed that pMGMT and age affected OS in the study population (p = 0.018 and 0.04, respectively). The most frequent grade 3 or severe toxicities were hypertension (13 [22%]), fatigue (7 [12%]), thrombocytopenia (5 [8.4%]), deep vein thrombosis/pulmonary embolism (3 [5.0%]), and infections (2 [3.3%]). Conclusions: Combination of carmustine with Bev is well tolerated and may derive benefit in recurrent Gb. pMGMT and age were associated with good prognosis.

2061

Poster Session (Board #248), Sat, 1:00 PM-4:30 PM

Patient survival on the dose escalation phase of the Oncolytic Polio/ Rhinovirus Recombinant (PVSRIPO) against WHO grade IV malignant glioma (MG) clinical trial compared to historical controls. First Author: Annick Desjardins, Duke University Medical Center, Durham, NC Background: The live attenuated oral (SABIN) serotype 1 poliovirus vaccine was modified to contain a heterologous internal ribosomal entry site stemming from human rhinovirus type 2, creating PVSRIPO. PVSRIPO recognizes CD155, an oncofetal cell adhesion molecule and tumor antigen widely expressed ectopically in malignancy. We report the survival results of the dose-finding portion evaluating PVSRIPO delivered intratumorally by convection-enhanced delivery (CED) compared to a historical control group of patients (pts) treated at Duke. Methods: Eligible pts on trial were adults with recurrent supratentorial WHO grade IV MG; solitary tumor 1-5cm in diameter; $ 4 weeks after chemotherapy, bevacizumab or study drug; adequate organ function; KPS $ 70%; and positive anti-polio titer. A historical group of adult recurrent WHO grade IV mg pts treated at Duke between 1/01/ 07-12/15/14 was retrospectively identified based on presence of a solitary tumor 1-5cm in diameter, KPS $ 70%, and absence of clinical decline which would have prevented enrollment on the PVSRIPO trial. Results: Fifteen pts were treated in the dose escalation phase, dose levels 15 (1 each at dose levels 1 and 3, 7 at level 2, 2 at level 4, and 4 at level 5), while 124 pts were identified as historical controls. Patient characteristics were similar for age, gender, KPS, prior number of progressions, steroid dosage at enrollment, and prior bevacizumab failure; however, 80% of PVSRIPO pts versus 58.9% of control pts had a gross total resection at diagnosis. As of 1/15/16, a median OS of 12.6 months was observed for the PVSRIPO group versus 10.5 months for the historical controls, with 23.3% of PVSRIPO pts alive at 24 months versus 13.7% of historical controls. Conclusions: Infusion of PVSRIPO via CED seems to have a survival advantage when compared to historical control pts treated at the same institution, with a higher proportion of patients alive at 24 months. Reported pts were not treated at what was identified as the optimal dose of PVSRIPO (dose level -1). It is predicted that the pts treated at the optimal dose will have a greater increase in survival. Clinical trial information: NCT01491893.

2060

107s Poster Session (Board #247), Sat, 1:00 PM-4:30 PM

Phase IB trial of carboxyamidotriazole orotate (CTO) and radiotherapy (RT) with concurrent and adjuvant temozolomide (TMZ) in newly diagnosed glioblastoma (GBM). First Author: Antonio Marcilio Padula Omuro, Memorial Sloan Kettering Cancer Center, New York, NY Background: CTO is an oral inhibitor of non-voltage-dependent calcium signaling achieving simultaneous modulation of several receptor-mediated signaling pathways. A single-agent phase I trial determined the maximum tolerated dose (MTD) at 427 mg/m2, with a safe toxicity profile. A phase IB in combination with TMZ for TMZ-refractory glioma found therapeutic brain tissue concentrations and early evidence of activity, prompting this study in newly diagnosed GBM. Methods: Following a 3+3 design, pts received escalating doses of daily CTO (219-481mg/m2) added to the standard GBM RT regimen (60 Gy concurrent with TMZ 75 mg/m2 daily, followed by adjuvant TMZ 150-200 mg/m2 x 5/28 days). Results: Accrual is complete. All pts (N = 15) had GBM (methylated MGMT: 33%; unmethylated: 67%). ChemoRT was well tolerated at CTO doses of 219-481 mg/m2. CTO-related adverse events ( . 10%) included fatigue, nausea, constipation, ALT, rash, and headache; in addition to these, AE likely-related to TMZ included decreased platelets and neutrophils. No dose-limiting toxicities (DLT) were observed up to 481 mg/m2 (maximum administered dose- MAD), but cumulative toxicities emerging after the DLT evaluation period prompted de-escalation to 370 mg/m2, which was better tolerated. Exploratory efficacy analysis shows median PFS of 17m, and median OS not reached (median follow-up of survivors: 15m). The 1-y OS is 92% (95% CI 57-99%). Three pts have already completed 12+ cycles. One MGMT unmethylated completed cycle 12, discontinued CTO and has not recurred after 3+ months. One patient continuing CTO is on cycle 17 and has had a durable partial response for 12+ months. One patient continuing CTO has stable disease after 15+ cycles. PK data confirmed therapeutic levels starting at 219 mg/m2. One patient operated during study (481 mg/m2 cohort) showed tumor concentration of 5900 ng/g, with a plasma level of 3460 ng/mL. Conclusions: CTO combined with RT and TMZ is safe and well tolerated. The MAD was 481 mg/m2 and recommended phase II dose is 370 mg/m2. Given encouraging preliminary signals of activity including durable disease control, a randomized study is warranted. Clinical trial information: NCT01107522.

2062

Poster Session (Board #249), Sat, 1:00 PM-4:30 PM

Overall survival in patients with glioblastoma before and after bevacizumab approval. First Author: Derek Richard Johnson, Mayo Clinic, Rochester, MN Background: Glioblastoma (GBM) is an aggressive disease with limited therapeutic options. While bevacizumab was approved in 2009 for the treatment of patients with recurrent GBM, its impact on overall survival (OS) remains unclear. This study utilizes population-based cancer registry data to compare OS of patients diagnosed with GBM before versus after bevacizumab approval. Methods: Adult patients diagnosed with GBM were identified in the SEER database and divided into 2 cohorts: patients diagnosed in 2006-2008 (Pre-Bevacizumab Cohort, n=6,120) and patients diagnosed in 2010-2012 (Post-Bevacizumab Cohort, n=6,753). All patients were included irrespective of the treatments received. OS was assessed from diagnosis (dx) date up to 3-years post-dx. Results: Among 12,873 patients with GBM, the median age was 62 years, 41% were females, 31% underwent gross total resection, and 75% received radiation therapy. OS rates at 1 and 2 years post-dx were significantly higher for the Post-Bevacizumab Cohort versus the Pre-Bevacizumab Cohort (44 vs 40% and 21 vs 19%, p ,.01, Table). After adjusting for potential confounders, the hazard of death remained significantly lower in the Post-Bevacizumab Cohort (hazard ratio [HR] 0.91, p ,.01, Table). Survival was stable within the 2006-2008 period (median survival = 9 months for all years), but increased after year 2009 (median survival = 10 and 11 months for years 2010-2011 and 2012, respectively). Conclusions: Based on this large population-based study, the survival following glioblastoma diagnosis improved in 2010-2012 compared to 2006-2008. While the cause of this improvement cannot be proven in a retrospective analysis, the timing of the survival increase may indicate a potential benefit of bevacizumab on recurrent GBM survival. Pre-Bevacizumab Cohort OS rates (95% CI)* 1-year post-dx 2-years post-dx Median survival* (months) Adjusted HR Post vs Pre (95% CI)†

Post-Bevacizumab Cohort

40% (38 – 41) 44% (43 – 45) 19% (17 – 20) 21% (19 – 22) 9 11 0.91 (0.87 – 0.96)

p-value ,.01

* From Kaplan-Meier analyses. † HRs estimated from Cox regression models adjusted for age, gender, race, radiation therapy, and surgery type. A HR,1 indicates the Post-Bevacizumab Cohort had longer survival.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

108s 2063

Central Nervous System Tumors Poster Session (Board #250), Sat, 1:00 PM-4:30 PM

Phase I/II study of VAL-083 in patients with recurrent glioblastoma. First Author: Kent C. Shih, Tennessee Oncology, Nashville, TN Background: Glioblastoma (GBM) is the most common CNS tumor. Patients with recurrent GBM have few treatment options and dismal prognosis. O6methylguanine-DNA-methyltransferase (MGMT) is correlated with resistance to front-line systemic therapy with temozolomide (TMZ) and poor patient outcomes. Dianhydrogalactitol (VAL-083) is a bi-functional alkylating agent that readily crosses the blood-brain barrier and has demonstrated activity independent of MGMT in multiple GBM cell lines and cancer stem cells in vitro. VAL-083 showed promise against CNS tumors in prior NCI-sponsored clinical trials. The goal of this clinical trial is to determine the appropriate dose for VAL-083 for advancement to Phase III trials as a new treatment for recurrent GBM. Methods: Study Population: Patients must have recurrent GBM following surgery, radiation, TMZ and bevacizumab. Phase I: Open-label, single-arm dose-escalation study (3+3 design). Patients received VAL-083 on days 1, 2, 3 of a 21-day cycle, until MTD was reached. Phase II: 14 additional patients enrolled at MTD to further assess safety and outcomes. Results: Phase I: 29 patients were enrolled across 9 dose cohorts (1.5 - 50 mg/m2/d) and 40 mg/m2/d confirmed as the MTD. Myelosuppression was mild; no drug-related serious adverse events were reported at doses # 40 mg/m2/d. Dose limiting toxicity (G4 thrombocytopenia) was observed at higher doses. Platelet nadir occurred around day 20 and resolved rapidly and spontaneously. A dose-related and clinically meaningful survival improvement was observed. Pharmacokinetic analyses show dose-dependent linear systemic exposure with short 1-2h plasma terminal half-life; average Cmax 781 ng/mL at 40 mg/m2/d. Phase II: 14 patients have been enrolled at 40 mg/m2/d. To date, safety observations in the Phase II cohort are consistent with Phase I. Conclusions: VAL-083 at 40 mg/m2/d exhibits a favorable safety profile and a dose-related trend toward clinically meaningful improved survival in refractory GBM patients. We expect to present median overall survival in the Phase II expansion cohort and proposed Phase III study design. Clinical trial information: NCT01478178.

2065

Poster Session (Board #252), Sat, 1:00 PM-4:30 PM

PSMA ADC for progressive glioblastoma: Phase II Brown University Oncology Research Group Study. First Author: Heinrich Elinzano, Rhode Island Hospital, Providence, RI Background: Antiangiogenic agents targeting new blood vessel formation are approved to treat progressive glioblastoma (GBM), but limited data are available on the activity of vascular disrupting agents targeting established blood vessels in these tumors. Prostate specific membrane antigen (PSMA) is a transmembrane peptidase upregulated on endothelial cells of solid tumors including GBM, making it a potential therapeutic target. A PSMAtargeted monoclonal antibody conjugated with microtubule disrupting agent monomethyl auristatin E (MMAE) is undergoing evaluation in advanced prostate cancer (PSMA ADC, Progenics Pharmaceuticals). Unconjugated MMAE has demonstrated activity against GBM cell lines. Methods: We investigated the activity of PSMA ADC in a single arm Phase II study in progressive GBM patients following prior treatment with radiation, temozolomide and bevacizumab. 2.5 mg/kg PSMA ADC was administered intravenously every 3 weeks until disease progression, unacceptable toxicity or removal from study. Disease status was evaluated clinically and with contrast enhanced brain MRI every 3 cycles. Primary endpoint was objective response and secondary endpoints were progression-free survival (PFS), 6-month PFS and overall survival (OS). Results: 6 patients (4 males, 2 females, median age 63 years, KPS 70-90%) received PSMA ADC treatment. No objective responses were noted. 2/6 (33%) patients progressed after 1 cycle, 3/6 (50%) patients progressed after 2 cycles and 1/6 (17%) patient progressed after cycle 3. Median PFS was 6 weeks. Median OS was 2 months from study entry. Grade 3 and 4 adverse events included lymphopenia, dysphasia, syncope, neutropenia, leucopenia and fatigue. One patient had doselimiting toxicity (sigmoid perforation possibly related to study drug) within 30 days of study treatment and off-study carboplatin resulting in death. Immunohistochemistry staining showed no PSMA expression in viable tumor samples of 2 patients and low expression (relative to that seen in prostate cancer trials) in 1. Conclusions: PSMA ADC had no demonstrable activity in these patients with progressive GBM likely due to minimal expression of the PSMA target and was associated with dose-limiting toxicity. Clinical trial information: NCT01856933.

2064

Poster Session (Board #251), Sat, 1:00 PM-4:30 PM

Multicenter Phase IB trial of carboxyamidotriazole orotate (CTO) combined with temozolomide (TMZ) for recurrent glioblastoma (GBM) and other malignant gliomas (MG). First Author: Xuling Lin, Memorial Sloan Kettering Cancer Center, New York, NY Background: CTO is an oral inhibitor of non-voltage-dependent calcium signaling that results in simultaneous modulation of several receptormediated calcium-dependent signaling pathways. This study (NCT01107522) evaluated the combination of a standard TMZ schedule with escalating doses of CTO (219-812.5 mg/m2/day) in TMZ-refractory mg pts. Methods: Eligible pts ($ 18years) had recurrent mg and ECOG # 2. Using a 3+3 design, pts received escalating daily doses of CTO combined with TMZ 150 mg/m2 x 5/28 days. Primary objectives were safety, maximum tolerated dose (MTD), and recommended phase II dose (RP2D) of CTO; secondary objectives were antitumor activity and pharmacokinetic (PK) characterization. Results: Accrual has been completed, with N = 27 pts (8 women, median age 49, range 28-78); 44% had $ 2 recurrences and 15% previously failed bevacizumab. Histology was GBM in 16, anaplastic glioma in 11. Molecular studies revealed IDH-1/2mutation in 8/21 (38%), 1p19q codeletion in 1/17 (5%), and MGMT promoter methylation in 9/18 (50%). The combination was well tolerated with no dose-limiting toxicities (DLT) observed at doses 219-812mg/m2. Most frequent adverse events were fatigue, nausea, vomiting, constipation and headache, all grades 1 or 2. PK showed inter-patient variability, with no proportional drug exposure increase beyond 600mg. Given the lack of DLTs, and based on PK data, the 600mg flat dose was chosen as RP2D, and this level accrued 6 more pts. Analysis of drug concentrations in two pts undergoing surgery on study showed therapeutic concentrations (1705 ng/g and 6,200 ng/g). Exploratory efficacy analysis among evaluable pts (N = 23) showed a partial response in 5 and stable disease in 10, for a clinical benefit (PR+SD) rate of 65% (43-84%). Response duration ranged from 2-15 months. The 6m-PFS was 29% (95% CI 11-49) and 1y-OS was 50% (95% CI 27-69). Conclusions: In this heavily pre-treated population, CTO combined with TMZ was safe and well tolerated. The MAD was 812 mg/m2/day and RP2D was 600mg/day. CTO crosses the blood-brain barrier. Responses were seen, and given encouraging signals of activity, further development is warranted. Clinical trial information: NCT01107522.

2066

Poster Session (Board #253), Sat, 1:00 PM-4:30 PM

Next generation sequencing (NGS) of primary central nervous system tumors (CNST) as part of a personalized medicine cancer research program (PMRP) to identify alterations with sensitivity to targeted inhibitors. First Author: Tara L. Benkers, Swedish Neuroscience Institute, Seattle, WA Background: Primary CNST demonstrate remarkable intra-tumoral and intertumoral heterogeneity contributing to therapeutic resistance. We report implementation of a PMRP for patients with primary CNST. Methods: An Institutional Review Board approved prospective registration protocol was activated in September 2014 as a data repository for cancer patients undergoing genomic evaluation. NGS profiling of CNST was performed using a 68-gene alteration panel. Mutations (M) were classified as Actionable (approved clinical indication), Applicable (clinical trials or off-label indication) and Unknown significance. Results: As of December 2015, sequencing was completed on 98 primary CNST. Diagnoses included glioblastoma (51), anaplastic glioma (25), low-grade glioma (7), neuronalglial tumors (4), DLBCL (3), medulloblastoma (3), ependymoma (1), hemangiopericytoma (1), meningioma (2), pituitary adenoma (1). Mutations were identified in 97% of cases. 308 M were identified: 0 Actionable, 128 (42%) Applicable, and 180 (58%) of Unknown significance. The most common Applicable M included: TP53 (33), IDH1 (31), PTEN (14), TYMS (10), EGFR (8), TPMT (8), PIK3CA (6), BRAF (3), APC (2), CDKN2A (2), MYD88 (2), TET2 (2), KRAS (2) ATM (1), DPYD (1), IDH2 (1), JAK3 (1), and RET (1). The number of Applicable M in each case was: 0 (20%), 1 (38%), 2 (27%), 3 (10%), 4 (1%), 5 (1%). Impact of NGS testing was assessed: new therapy started (3%), declined off-label treatment option (1%), no actionable information (26%), stable on current therapy (58%), transitioned to hospice (7%). Clinical decisions were impacted by lack of evidence to change the treatment plan (36%) and off-label access and insurance coverage (68%). Conclusions: Primary CNST demonstrate a significant number of actionable mutations associated with sensitivity to targeted inhibitors along the PI3K/AKT/mTOR, EGFR, CDK, and BRAF pathways amongst others and matching to clinical trial therapies. CNST represent a unique opportunity for investigation of targeted therapies, however access to such therapies remains a barrier.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Central Nervous System Tumors 2067

Poster Session (Board #254), Sat, 1:00 PM-4:30 PM

2068

109s Poster Session (Board #255), Sat, 1:00 PM-4:30 PM

Pilot study of T cells redirected to EGFRvIII with a chimeric antigen receptor in patients with EGFRvIII+ glioblastoma. First Author: Donald M. O’Rourke, University of Pennsylvania, Philadelphia, PA

A phase I study of chemo-radiotherapy with plerixafor for newly diagnosed glioblastoma (GB). First Author: Reena Parada Thomas, Stanford University Hospital, Stanford, CA

Background: We have developed an autologous T cell product directed to the EGFR variant III mutation via lentiviral transduction of a chimeric antigen receptor (CAR), and opened an IND. We have initiated a pilot study enrolling patients with recurrent EGFRvIII-positive glioblastoma (GBM). Methods: Patients with recurrent GBM have their tumors screened for the presence of the EGFRvIII mutation by a next-generation sequencing based assay. Eligible patients undergo leukapheresis for collection of autologous T cells, which are genetically modified ex vivo to express the EGFRvIII CAR, expanded, and then cryopreserved for infusion. Results: We report results on the nine patients we have treated. To date, we have found that infusion of 15x108 CART-EGFRvIII cells is feasible to manufacture and infusion is safe, without evidence of off-tumor toxicity or cross-reactivity to wild type EGFR. No clinical or laboratory signs of systemic cytokine release syndrome have been observed, though elevations in serum IL-6 occur concurrently with CART-EGFRvIII expansion in the peripheral blood. One patient developed non-convulsive status epilepticus nine days after CART-EGFRvIII infusion, which resolved with standard treatment and anti-cytokine therapy. All patients have had significant expansion of CART-EGFRvIII cells between 710 days post-infusion, as measured by flow cytometry and quantitative PCR in peripheral blood samples. Five patients have undergone surgical resection of tumor between 6-120 days after infusion, and pathologic evaluation has demonstrated infiltration of activated CAR T cells, recruitment of new T cells (as assessed by next generation T cell receptor deep sequencing), and specific EGFRvIII target antigen loss in GBM cells in some cases. Conclusions: These findings provide evidence that CART-EGFRvIII cells are safe, without evidence of off-target toxicity or cytokine release syndrome, and are immunologically active. The data also suggest a mechanism of antigen editing by activated CART-EGFRvIII cells that track to specific EGFRvIII GBM cells. Clinical trial information: NCT02209376.

Background: Preclinical studies indicate that local recurrence after radiotherapy (RT) of Glioblastoma is dependent on recovery of tumor vasculature following RT via hypoxia-driven SDF-1 (CXCL12) secretion. We hypothesize that blocking the CXCL12/CXCR4 axis would improve local control. Methods: Newly diagnosed Glioblastoma patients were recruited to this Phase I study of a 4 week intravenous infusion of Plerixafor (Mozobil), a CXCR4 antagonist, with concurrent Temozolomide (TMZ) and radiation therapy. Three patients were treated with Plerixafor at 8.3 mg/kg/hr and six patients at 16.6 mg/kg/hr, while being monitored for dose limiting toxicities (DLTs) including grade $ 3 hematologic or non-hematologic adverse events. Patients underwent dynamic susceptibility contrast perfusion MRI (DSCMRI) for quantification of relative cerebral blood volume (rCBV) values by region-of-interest analysis. Pharmacokinetic (PK) analysis of Plerixafor plasma levels were collected. Results: Since August 2014, 9 patients completed therapy with upfront RT/TMZ and Plerixafor with no DLTs observed, leading to the recommended phase II dose of 16.6 mg/kg/hr. Currently, 7 of 9 patients are alive, with the longest survival since diagnosis being 18 months. Plasma Plerixafor levels reached our target of 100 ng/ml at the first time point, 7 days into therapy. DSC-MRI at 1 and 6 months post-RT revealed lower rCBV in the tumor bed compared to the postoperative pre-RT scan (P , 0.02, one way ANOVA). Conclusions: Plerixafor was safely escalated to the target dose of 16.6 mg/kg/hr with no DLTs observed. A marked decrease in rCBV in the radiation treatment field suggests enhanced local treatment effect in support of our hypothesis that inhibition of the SDF1/ CXCL4-mediated vasculogenesis pathway in the post-RT period enhances radiation. Recruitment into the phase II study is ongoing to evaluate these preliminary observations. Clinical trial information: NCT01977677.

2069

2070

Poster Session (Board #256), Sat, 1:00 PM-4:30 PM

Functional profiling of a glioblastoma (GBM) patient-derived cell line (PDCL) panel to identify cell-intrinsic differential radiation response. First Author: Mai Anh Huynh, Harvard Radiation Oncology Program, Boston, MA Background: It is not known whether GBM responds differently to radiation based on tumor cell-intrinsic factors. Newly developed PDCLs of GBM have potential to illuminate individual patient functional response to treatment at a scale not previously possible. We therefore assessed whether intrinsic differences in radiosensitivity are present in a large panel of clinically and genomically annotated GBM PDCLs. Methods: PDCLs were established from newly diagnosed (n = 25) and recurrent (n = 10) GBM patients. Whole exome sequencing and whole genome array comparative genomic hybridization was performed on the PDCLs, and patient outcomes data was collected. Highthroughput functional radiation screening was performed across all patients at varying doses. Response was calculated at 9 days after treatment based on ATP measurement using Cell Titer Glo to assess growth and recovery. Proliferation was integrated as a function of dose for each cell line (AUC). AUC was correlated with genomic candidate factors and clinical outcomes. Cell cycle phase and activation of DNA damage repair was assessed in a subset of PDCLs. Results: Mean AUC values were normally distributed across patient models (p = 0.67 Shapiro-Wilk). The mean AUC for p53 mutant lines was 4.3 vs. 2.9 for p53 intact lines (p = 0.03). Whereas p53 mutant tumors were generally relatively resistant, p53 wild-type tumors showed significant variability in radiosensitivity. Radioresistant lines displayed higher levels of 53BP1 and phospho-gH2ax foci. There was no difference in radiosensitivity between tumors derived from treatment-na¨ıve or recurrent tumors (p = NS). There was also no correlation of intrinsic radiosensitivity with patient outcome, though the analysis was limited by sample size and significant variability related to clinical prognostic factors. Conclusions: GBM PDCLs suggest the existence of tumor cell-intrinsic differences in radiosensitivity across patients that correlate with p53 mutation status and may relate to differences in the capacity for DNA repair. Further studies are needed to validate this in vivo and to determine whether PDCL may be useful functional indicators of clinical response.

Poster Session (Board #257), Sat, 1:00 PM-4:30 PM

Brain Metastases in Breast Cancer Network Germany (BMBC, GBG 79): Treatment patterns and clinical outcome of more than 1000 patients with brain metastases from breast cancer. First Author: Volkmar Mueller, University Hospital Eppendorf Frauenklinik, Hamburg, Germany Background: In order to improve knowledge about treatment and outcome of breast cancer patients with brain metastases (BM) we initiated the registry Brain Metastases in Breast Cancer Network Germany (BMBC, GBG79). Methods: Patients diagnosed with BM since 2000 were included retrospectively and prospectively. 1105 patients from 54 centers were analyzed. Results: Median age at diagnosis of BM was 51 years (20 – 87 years). 49.0 % of patients (472/1105) had HER2-pos., 20.5% (198/1005) triple-neg. and 30.5% (294/1005) luminal primary tumors. One year survival rate from diagnosis of BM was 35.7 % (CI95%: 32.6 – 38.8). Median time from first diagnosis of BM to death in the entire cohort was 7.1 month (CI95%: 5.8 – 7.9). HER2-positive patients had the longest median survival with 12.1 mo. (CI95%: 10.2 – 13.7) compared to 5.5 mo. (CI95%: 4.1 – 7.1) for luminal primary tumors and 4.1 mo. (CI95%: 3.1 – 4.8) for triple-negative patients (p , 0.001). Median OS of patients with BM diagnosed between 2000 and 2006 was 7.2 mo. (CI95%: 5.3 - 9.6) compared to 6.7 mo. (CI95%: 5.5 -8.1) of patients diagnosed between 2007 and 2012 (p = 0.848). No relevant OS improvements were observed in these time intervals for luminal (5.1 mo. [CI95%: 2.7-10.4] vs. 5.8 mo. [CI95%: 4.1 -8.0]; p = 0.541), triple negative (4.6 months [CI95%: 3.3 -7.4] vs. 4.0 mo [CI95%: 2.4 - 5.4]; p = 0.558) or HER2-pos. patients (10.7 mo. [CI95% 7.7- 13.5] vs. 12.6 [CI95% 9.4 – 15.5]; p = 0.320). 32.8% of HER2-pos. patients received HER2-directed therapies beyond the diagnosis of BM (55.5% trastuzumab, 53.5% lapatinib, 19.4% T-DM1 and 11.0% trastuzumab and pertuzumab). Those HER2-pos. patients receiving HER2-directed therapy after the diagnosis of BM lived longer than those without (median 17.7, CI95%: 15.0 – 22.0 vs. 12.9 mo., CI95%: 10.5 – 17.1, p = 0.015). Conclusions: Patients that received HER2-directed therapy after diagnosis of BM had an improved survival. However, we could not observe improvement of survival over the study period in any tumor subtype. These findings underline the high need for the development of new treatment strategies in patients with BM from breast cancer.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

110s 2071

Central Nervous System Tumors Poster Session (Board #258), Sat, 1:00 PM-4:30 PM

2072

Poster Session (Board #259), Sat, 1:00 PM-4:30 PM

Retrospective analysis of activity of pembrolizumab (pembro) in melanoma patients (pts) with brain metastasis (BM). First Author: Ibiayi Dagogo-Jack, Massachusetts General Hospital, Boston, MA

Descriptive analysis of 2419 patients with brain metastases of solid cancers: A real life cohort. First Author: Anna Sophie Berghoff, University of Vienna, Vienna, Austria

Background: Approximately 50% of pts with metastatic melanoma (MM) develop BM. Among MM pts, pembro, a humanized antibody against PD-1, is associated with objective response rates of 32-35% in the first line setting and 20-30% in ipilimumab-refractory pts. Median progression-free survival on pembro ranges from 4-6 months (mos). However, only 10% of pts included in these studies had BM. We, therefore, performed a retrospective analysis of MM pts treated at our institution to assess the activity of pembro in BM. Methods: We reviewed records of 113 consecutive pts with MM treated with pembro at Massachusetts General Hospital from May 1, 2014 to October 31, 2015. Twenty-four pts were excluded due to absence of available imaging for response evaluation. We evaluated time to progression (TTP) on pembro based on investigator assessment for the remaining 89 patients. Results: Thirty-six (40%) pts had BM prior to initiation of pembro, of whom 72% had received prior treatment for BM [SRS (50%), WBRT (3%), surgery (8%), or a combination of radiation and surgery (38%)]. With a median follow-up of 8.5 mos, 49 pts (55.1%) had disease progression (PD) on pembro. Intracranial (IC)-only PD was observed in 9 (10%) pts, one of whom developed BM on pembro. Median overall TTP at any site was 4 mos (90% CI: 2.7 to 7.8 mos). Median TTP at any site was 5.4 mos for those without BM (n = 53), 6.8 mos for those with treated BM (n = 26), and 1.2 mos for pts with untreated BM (n = 10). Based on Cox regression analysis adjusting for baseline factors, there was a significant (Wald chi-squared pvalue = 0.003) reduction in risk of PD in pts without BM (HR 0.18, 90%CI: 0.08 to 0.41) and in patients with treated BM (HR 0.25, 90% CI: 0.11 to 0.61) compared with those with untreated BM. Thirty-two (36%) pts were treated beyond PD (duration 21 days to 365+ days); half received local therapies. The 6-mos estimate of overall survival (OS) for the study population was 86% (90%CI: 77% to 91%). Median OS was not reached. Conclusions: Pembro has clinical activity in MM pts with BM, particularly those with previously treated BM. Prospective trials evaluating activity of pembro in MM patients with untreated BM are ongoing (NCT02085070).

Background: Profound knowledge of the baseline characteristics and the clinical course of brain metastases (BM) patients is the basis for reasonable clinical trial planning. Methods: A detailed chart review analysis of 2419 BM patients (50.5% male, 49.5% female, median age 59 years) patients treated for BM of solid cancers at the Medical University of Vienna between 1990 and 2011 was performed. Results: Primary tumor was lung cancer in 43.2%, breast cancer in 15.7%, melanoma in 16.4%, renal cell carcinoma in 9.1%, colorectal cancer in 9.3% and unknown in 1.4% patients. Rare primary tumor types included genito-urinary cancers (4.1%), sarcomas (0.7%), gastro-esophageal cancer (0.6%) and head and neck cancers (0.2%). A singular BM was present in 48.7% of patients, 2-3 BM in 27.7% and 23.5% presented with . 3 BM. Multiple BM were most frequently observed in breast cancer (30.6%) and least frequently in colorectal cancer (8.5%; p , 0.001). Time from primary tumor to BM diagnosis was shortest in lung cancer (median 11 months; range 1-162) and longest in breast cancer (median 44 months; 1 – 443; p , 0.001). Recursive partitioning analysis (RPA) and graded prognostic assessment (GPA) scores showed significant correlation with overall survival (both p , 0.001, log rank test). Evaluation of the disease status in the last 2 months prior to patient death showed intracranial progression in 35.9%, extracranial progression in 27.5% and combined extraand intracranial progression in 36.6% of patients. Conclusions: Our data highlights the heterogeneity in presentation and clinical course of BM patients in the everyday clinical setting and may be useful for rational planning of clinical studies.

2073

2074

Poster Session (Board #260), Sat, 1:00 PM-4:30 PM

The ARCAD METACER national cohort study of brain metastases in gastrointestinal cancers. First Author: Fabienne Portales, Institut regional ´ du Cancer de Montpellier (ICM), Montpellier, France Background: Brain metastases (BM) are quite rare and not well-documented in gastro-intestinal (GI) cancers. This prospective multicenter cohort study aimed to assess the overall survival (OS) in GI cancer patients with BM. Methods: From 2012 to 2014, 147 patients with synchronous or metachronous BM from histologically-proven GI primary tumors (PT) from 15 French centers were included. OS was defined as the time from BM diagnosis until death. Results: Patients were mainly women (64.6%); median age was 65.5 [28-92]. Histology was adenocarcinoma in 138 pts (95.8%). PT sites were mainly colorectal (71.9%), gastroesophageal (GE) (17.1%), biliary tract (7.6%). Initial staging was T3-T4 in 87 pts (65.5%), positive nodal status in 74 pts (55.2%). 61 pts (44.5) were M0, 75 pts (55.6%) were metastatic with 1, 2 or 3 sites in 42.2%, 36.3% and 21.5 %. Extra-cerebral metastases (ECM) were mostly lung (63.7%) and liver (59.3%). Median time between PT diagnosis and BM occurrence was 25.9 months [1.3-243.2]. 89 pts (60.5%) had PT or ECM progression. BM were unilateral in 62.5% patients, cerebral, cerebellar or both in 59.3%, 12.1% and 28.6%, respectively. BM management was anti-oedematic agents in 120 pts (84.5%), surgery in 25 pts (17.6%), whole-brain or stereotaxic radiotherapy in respectively 78 (54.1%) and 23 pts (16.0%). After a median follow-up of 28.7 months [0-32], median OS was 6.6 months [4.5-8.3]. Median OS was significantly higher in patients who underwent BM surgery, 12.1 (95%CI: 7.2-18.0) vs 4.6 (95%CI: 3.6-7.2) months (p= 0.010), in patients with WHO performance status 0-1 (8.4; 95%CI: 6.6-11.3) vs 2 (3.3; 95%CI: 2.4-5.9) (p= 0.001), and for cerebral (8.7; 95%CI: 5.9-12.3) vs cerebellar (4.6; 95%CI: 1.3-7.2) vs both (3.7; 95%CI: 2.4-7.5) localization (p= 0.013). In multivariate analysis, only BM localization was a significant prognostic factor for OS (HR = 1.7 and 1.8 for cerebellar and both localizations, p= 0.0168). Conclusions: Our study confirms that BM surgery improves poor survival of GI cancer patients with BM. Management and treatment of BM should be discussed in multidisciplinary meetings in regards of the prognostic factors identified, especially BM localization and PS. Authors thank the GERCOR for operational support.

Poster Session (Board #261), Sat, 1:00 PM-4:30 PM

Ofranogene obadenovec (VB-111), an anti-cancer gene therapy in combination with bevacizumab to improve overall survival compared to bevacizumab monotherapy in patients with rGBM: A phase 2 historically controlled trial. First Author: Andrew Jacob Brenner, The University of Texas Health Science Center at San Antonio, San Antonio, TX Background: Ofranergene obadenovec (VB-111) is a gene-therapy specifically targeting angiogenic endothelial cells thereby leading to tumor starvation. Safety and efficacy of VB-111 in combination with bevacizumab (BEV) were evaluated in recurrent Glioblastoma (rGBM) patients in this Phase 2 study and were compared to pooled BEV data generated using metaanalysis. Methods: VB-111 was administered at 1x1013 viral particles bimonthly until progression, followed by BEV standard of care (limited exposure cohort). The protocol was amended to add BEV 10mg/Kg biweekly combined with VB-111 bimonthly, until further progression (continuous exposure cohort). A systematic literature review identified reports of rGBM BEV monotherapy published between 1.1.2009 and 30.11.2015. Data were pooled in a meta-analysis and served as historical controls for comparison of VB-111 extended exposure cohort. Results: 46 patients at 4 sites received up to 13 doses of VB-111. Upon further progression, 24 received VB-111 with BEV, 22 were treated with BEV as standard of care. VB-111 was safe and well-tolerated alone and combined with BEV. Median OS was 59 weeks for continuous exposure versus 33 weeks for limited exposure cohort (p = 0.048). Eight reports of BEV monotherapy were identified, with a total of 694 rGBM patients. VB-111 continuous exposure had superior OS compared to the pooled meta-analysis of BEV monotherapy, 59 versus 39 weeks (p = 0.0295, log-rank test), with a Hazard Ratio of 0.62 (95% CI: 0.40-0.96). Of 46 patients who received VB-111, 25 patients spiked a fever post- dosing of VB-111 at least once, which was associated with increased median OS of 64 weeks versus 34 weeks among patients without fever (p = 0.03). Conclusions: VB-111 was safe and well-tolerated alone and combined with BEV in rGBM patients. VB-111 continuous exposure significantly increased OS compared to a historical BEV meta-analysis controls, and compared to VB-111 limited exposure. Toxicities were as expected in this population. The GLOBE phase 3 randomized controlled trial of VB-111 in rGBM is currently underway. Clinical trial information: NCT01260506.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Central Nervous System Tumors 2075

Poster Session (Board #262), Sat, 1:00 PM-4:30 PM

Prevalence, clinical risk factors and outcomes of patients with lung cancer presenting with brain metastases. First Author: Saiama Naheed Waqar, Washington University in St Louis, St. Louis, MO Background: There is limited data on the prevalence of brain metastases at presentation in patients with non-small cell lung cancer (NSCLC). We queried the National Cancer Database (NCDB) to determine the frequency of brain metastases in patients diagnosed with NSCLC and clinical risk factors associated with their presence. Methods: We identified patients with NSCLC diagnosed between 2010 and 2013 using the NCDB database, as during this time period, brain metastasis information was captured. Odds ratios (ORs) for presence of brain metastasis were calculated for individual pre-specified covariates of interest using logistic regression analysis. All statistical tests were 2-sided and p , 0.05 was considered significant. Results: Brain metastases were observed in 47,546 (9.4%) of the 504,151 patients with NSCLC during our study time period. On multivariate analysis the following variables were significantly associated with presence of brain metastasis: age (OR, 0.72 per 10 year increase in age; 95% confidence interval [CI]: 0.71, 0.73); adenocarcinoma histology (OR 2.75; 95% CI:2.67,2.83); tumor size . 3 cm (3.1 to 5 cm OR 1.55; 95% CI: 1.51, 1.60 and . 5 cm OR, 1.75; 95% CI: 1.70, 1.80); tumor grade $ II (grade II OR 2.11; 95% CI: 1.95, 2.29, grade III OR 3.87; 95% CI: 3.59, 4.18 and grade IV OR 3.60; 95% CI: 3.2, 4.05) and node positive disease N1 OR 1.86; 95% CI: 1.79, 1.94, N2 OR 2.08; 95% CI: 2.03, 2.14 and N3 OR 1.76; 95% CI :1.65, 1.77). We re-categorized age as , 70 or $ 70 years and the incidence of brain metastasis ranged from as low as 0.6% in patients with only 1 risk factor to as high as 19.5% in patients with all 5 risk factors. The median follow-up time for patients with brain metastases was 22 months. The median survival for patients with brain metastases was 6.08 months (95% CI 5.98-6.18), and their overall survival (OS) rates at 1, 2 and 3 years were 30.1%, 14.5% and 8.5% respectively. Conclusions: Brain metastases are seen in 9.4% of patients with NSCLC at presentation. If validated, our model may be used to guide the staging work-up for patients with newly diagnosed NSCLC without symptoms suggesting brain metastases.

2077

Poster Session (Board #264), Sat, 1:00 PM-4:30 PM

2076

111s Poster Session (Board #263), Sat, 1:00 PM-4:30 PM

Single fraction stereotactic radiosurgery (SRS) alone versus surgical resection and SRS boost for large brain metastases: A multi-institutional analysis. First Author: Roshan Sudhir Prabhu, Southeast Radiation Oncology Group, Levine Cancer Institute, Charlotte, NC Background: SRS dose is typically limited by brain metastasis (BM) size. Any benefit for the addition of surgery to SRS for large BM in patients (pts) with oligometastatic brain disease is not well characterized. Methods: Large BM was defined as $ 4 cc (2 cm diameter) prior to the study. The records of consecutive pts treated with single fraction SRS alone or surgery with either pre-operative (pre-SRS) or post-operative SRS (post-SRS [collectively SRS boost]) between 2005 - 2013 from 2 institutions were reviewed. Pre-SRS dose was 20% reduced compared to standard SRS dosing, with planned resection within 48 hours. Overall survival (OS) analysis used the KaplanMeier method. Otherwise, cumulative incidence with competing risk was used. Results: Overall, 250 pts with 260 treated large BM were included, of which 66 were treated with SRS alone and 194 with surgery and SRS boost (65 pre-SRS, 129 post-SRS). Median imaging follow-up period was 29 months for alive pts. Groups were well balanced except for lesion volume (GTV, median 5.9 cc vs. 9.3 cc, p , 0.001), median number of BM (1.5 vs. 1, p = 0.006), and median SRS dose (18 vs. 15 Gy, p , 0.001), respectively. Local recurrence (LR) was significantly lower with SRS boost (1-year LR: 36.7% vs. 18.8%, p = 0.004), which remained significant in multivariate analysis (MVA, hazard ratio [HR]: 0.33, 95% confidence interval (CI): 0.17–0.62, p , 0.001). There was no difference in radiation necrosis (RN) by resection status, but post-SRS had significantly increased RN rate compared with pre-SRS or SRS alone (reference) in MVA (HR: 2.9, 95% CI: 1.04-8.5, p = 0.05, 1-year RN: 22%, 4.8%, 12.3%, p , 0.001, respectively). OS was significantly higher with SRS boost (2-year OS: 35.5% vs. 19.5%, p = 0.013), which remained significant in MVA (HR 0.57, 95% CI 0.36-0.89, p , 0.01). Conclusions: Surgical resection with SRS boost (either pre-SRS or post-SRS) is associated with significantly reduced LR and improved OS compared with SRS alone for large ( $ 4 cc) BM, despite larger average lesion size in the SRS boost group. Post-SRS was associated with the highest rate of RN. Surgical resection with SRS boost may improve outcomes in pts with a limited number of large BM.

2078

Poster Session (Board #265a), Sat, 1:00 PM-4:30 PM

A phase 2 study on efficacy, safety and intratumoral pharmacokinetics of oral selinexor (KPT-330) in patients with recurrent glioblastoma (GBM). First Author: Morten Mau-Sørensen, Rigshospitalet, Copenhagen, Denmark

Complications of anticoagulant therapy in patients with brain metastases. First Author: Joshua David Gruhl, University of Nebraska Medical Center College of Medicine, Omaha, NE

Background: The nuclear export protein exportin 1, (XPO1) is overexpressed in a wide variety of cancers, including GBM. Selinexor (SEL) is an oral Selective Inhibitor of Nuclear Export XPO1 antagonist with brain penetration. SEL forces the nuclear retention & activation of tumor suppressor proteins (TSPs) such as p53, IkB, & p21. Methods: We conducted an open label, 4-Arm Phase II trial enrolling adults with KPS $ 60 and recurrent, bevacizumab na¨ıve, measurable GBM. Pts in Arm A received 50 mg/m2 of SEL BIW up to three times prior to resection of recurrent GBM to explore pharmacokinetic and pharmacodynamic effects of SEL in vivo; the other arms did not involve resection. Pts in Arm B received 50 mg/m2 SEL BIW, but the regimen was overly toxic. Therefore, arm B was closed and pts were randomized to either Arm C (60 mg, flat dose, BIW) or D (80 mg, flat dose, once weekly). Up to 30 pts will be accrued to arms C and D, each with a 30% target 6mPFS rate as the primary endpoint. Results: 7 pts (6/1 M/F, median age 57, 1–2 prior tx regimens (PTR)) were treated on Arm A, and 24 (19/5, M/ F, median age 51; 1–3 PTR) on Arm B. Accrual to Arms C & D continues, with 4 pts (2/2 M/F, median age 42; 2-3 PTR) as of 20-Jan-2016. Commonly reported G1/2 AE’s (Arms B, C, D) included thrombocytopenia (49%) anorexia (41%), nausea (37%), & fatigue (33%). G3 AEs reported in $ 2 pts included fatigue (19%) & neutropenia (11%). One related G4 AE (hypophosphatemia) was reported in 1 pt. Mean tumor SEL concentration in Arm A was 136 nM vs. 985.5 nM in simultaneously collected plasma. Analyses of molecular predictors of response and effects during treatment are ongoing. RANO defined best response (investigator determined) among 27 evaluable pts were: 3 partial responses (11%), 6 stable disease (22%) & 18 progressive disease (67%). The 6-months-PFS rate is immature, although 4 pts were on SEL for $ 6 cycles. Conclusions: SEL’s concentration achieved in GBM tissue exceeds the IC50 in vitro. The main toxicities were thrombocytopenia, fatigue, and anorexia, which are mostly Grade 1/2. Partial responses & stable disease were observed. Accrual continues. Clinical trial information: NCT01986348.

Background: Anticoagulation has long been viewed as carrying excessive risk of intracranial hemorrhage (ICH) in patients with brain metastases. However the data evaluating the exact incidence of ICH in patients on anticoagulant therapy (ACT) are sparse. This study was conducted to determine the incidence of ICH associated with ACT in adult patients with brain metastases Methods: Consecutive patients with brain metastases occurring from 20062014 were identified from a single institution database. Patients were categorized as having no outpatient anticoagulant therapy versus having outpatient anticoagulation therapy of greater than 1 month. Chi-square tests and Fisher’s exact test were used to compare rates of ICH by groups. Statistical analyses were carried out using SAS 9.3. Results: A total of 125 patients with brain metastases were analyzed. Of these, 64 had primary nonsmall cell lung cancer (51.2%), while the other primary sites included, breast (12%), melanoma and small cell lung cancer (7.2%) Overall, 12/125 (9.6%) patients had evidence of ICH. Neither the primary tumor site nor the number of brain metastases was associated with the development of ICH. Eight of 67 (11.94%) patients on outpatient ACT had evidence of ICH, compared with 4 of 58 (6.9%) patients not on ACT; however, this difference was not statistically significant (p = 0.33). In the subset of patients on enoxaparin, there was no difference in the incidence of ICH for daily versus twice daily dosing (p = 1.0). Those treated with stereotactic radiosurgery in addition to whole-brain radiotherapy (SRS+WBRT) were significantly more likely to experience ICH (p = 0.0014), however these were within the metastatic lesion and none of these events required any further intervention than ACT cessation. Conclusions: Treatment of metastases with SRS+WBRT was significantly correlated with ICH independent of ACT. Our finding that ACT does not significantly increase risk for ICH supports its use in these patients.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

112s TPS2079

Central Nervous System Tumors Poster Session (Board #265b), Sat, 1:00 PM-4:30 PM

A randomized, phase 3, open-label study of nivolumab versus temozolomide (TMZ) in combination with radiotherapy (RT) in adult patients (pts) with newly diagnosed, O-6-methylguanine DNA methyltransferase (MGMT)unmethylated glioblastoma (GBM): CheckMate-498. First Author: John H. Sampson, Duke University Medical Center, Durham, NC

TPS2080

Poster Session (Board #266a), Sat, 1:00 PM-4:30 PM

Phase 2 study to evaluate the clinical efficacy and safety of MEDI4736 (durvalumab) in patients with glioblastoma (GBM). First Author: David A. Reardon, Dana-Farber Cancer Institute/ Harvard Cancer Center, Boston, MA

Background: GBM, the most common primary brain tumor in adults, has an aggressive clinical course and a median survival of 15–18 months in clinical trial populations. Approximately 40% of pts with GBM have tumors with an MGMT-methylated promoter, which is associated with sensitivity to TMZ. New treatment options for pts with GBM are desirable, particularly for those with MGMT-unmethylated GBM, as TMZ efficacy is historically marginal in this population. Nivolumab, a fully human IgG4 monoclonal antibody to the programmed death-1 receptor, has shown clinical activity and increased survival in pts with metastatic melanoma, non–small-cell lung cancer, and renal cell carcinoma. CheckMate-498 (NCT02617589) is a phase 3 study designed to compare overall survival (OS) of nivolumab or TMZ, each in combination with RT, in pts with newly diagnosed MGMT-unmethylated GBM. In a companion phase 2 trial (CheckMate-548; NCT02667587), eligible pts with MGMT-methylated tumors (N = 320) will be randomized to RT + TMZ + nivolumab vs RT + TMZ + placebo, with OS as the primary objective. Methods: Eligibility criteria include newly diagnosed, histologically confirmed, MGMT-unmethylated (centrally confirmed) GBM in pts aged $ 18 years with Karnofsky performance status $ 70. Pts with prior treatment for GBM and recurrent or secondary GBM are ineligible. Approximately 550 pts will be randomized to receive nivolumab + RT followed by nivolumab, or TMZ + RT followed by TMZ. Treatment will continue until progression or unacceptable toxicity. The primary objective is to compare OS with nivolumab + RT vs TMZ + RT; secondary objectives include progressionfree survival and 2-year survival rate. Exploratory objectives include safety, biomarker analyses, and neurocognitive outcomes. Patients will be followed for safety and tolerability, tumor progression, and survival. At least 397 events, with an interim analysis after $ 298 events, will provide 90% power to detect a hazard ratio of , 0.72. Clinical trial information: NCT02617589.

Background: Outcome among newly diagnosed and recurrent GBM patients remains poor, and novel therapeutic approaches are needed. Blockade of Programmed death-1 (PD-1) and PD-ligand 1 (PD-L1) has shown promising clinical benefit among solid tumors, and data implicate PD-1/PD-L1 signaling as a significant contributor to immunosuppression in GBM. Specifically, PD-1 is expressed by many GBM infiltrating lymphocytes, PD-L1 is expressed by 61-100% of GBM tumors, and loss of the PTEN tumor suppressor gene, occurring in 40-50% of GBM tumors, leads to increased transcription and expression of PD-L1. Durvalumab (DUR) is a human IgG1 mAb against PD-L1. Bevacizumab (BEV) is a mAb against VEGF, and data suggest that VEGF inhibition may enhance the antitumor benefit of immunotherapies. Methods: This ongoing Phase 2, multicenter, open-label study (NCT02336165) is designed to evaluate the clinical efficacy and safety of DUR (10 mg/kg i.v. Q2W for 12 months) in 5 cohorts of newly diagnosed and recurrent GBM patients. Eligible patients include those who are newly diagnosed with GBM with unmethylated MGMT promoter and scheduled for radiotherapy (Cohort A); BEV-na¨ıve patients with recurrent GBM (Cohorts B, B2 and B3); and BEV-refractory patients with recurrent GBM (Cohort C). Cohort A patients receive DUR beginning with radiotherapy. Cohort B patients receive DUR as monotherapy, and Cohorts B2 and B3 receive DUR plus BEV at 10 mg/kg Q2W or 3 mg/kg Q2W, respectively. Cohort C patients receive DUR with continued BEV at 10 mg/kg Q2W. Primary endpoints are overall survival at 12 months (OS-12) (Cohort A), progression free survival at 6 months (PFS-6) (Cohorts B, B2, B3), and OS-6 (Cohort C). Secondary endpoints are safety/tolerability, median PFS/OS and overall response rate, DUR pharmacokinetics, and quality of life measures. Exploratory endpoints are neurologic function by NANO and immunocorrelative biomarkers. Enrollment as of 31Dec2015, with no DLTs reported during the 3+3 safety run-in, is: Cohort A, 5 of planned 37 patients; Cohort B, 31 of planned 30 patients (complete); Cohort C, 13 of planned 17 patients. Cohorts A and C are continuing enrollment; Cohorts B2 and B3 (n = 32 each) opened to enrollment in Dec 2015. Clinical trial information: NCT02336165.

TPS2081

TPS2082

Poster Session (Board #266b), Sat, 1:00 PM-4:30 PM

VERTU: Veliparib, radiotherapy (RT) and temozolomide (TMZ) trial in unmethylated MGMT glioblastoma (GBM). First Author: Mustafa Khasraw, NHMRC CTC, The University of Sydney, Camperdown NSW, Australia Background: Temozolomide (TMZ) offers minimal benefit in GBM patients (pts) with unmethylated O-6-methylguanine DNA methyltransferase (MGMT), that is the majority of GBMs. Pre-clinical data shows that the addition of the poly ADP ribose polymerase (PARP) inhibitor ABT-888 (veliparib,) to radiotherapy (RT) and TMZ is synergistic. Clinical studies demonstrated that veliparib is safe and tolerable when combined with either RT or TMZ but the triplet was poorly tolerated. VERTU examines using concurrent veliparib and RT (without TMZ) followed by concurrent veliparib and TMZ in newly diagnosed pts with GBM and unmethylated MGMT. Methods: A phase II trial of 120 pts randomized 2:1 to the experimental arm (n = 80) of veliparib 200mg bid concurrently with RT (60Gy in 30 fractions) followed by 6 months of veliparib (40mg bid, D 1-7) concurrently with TMZ (150-200mg/m2D 1-5) every 28 days. The standard treatment arm (n = 40) is concurrent RT and TMZ followed by 6 months of TMZ. MGMT methylation status is tested centrally. Primary endpoint is 6 months Progression Free Survival (6PFS) using RANO criteria with secondary endpoints including overall survival, toxicity and translational endpoints including expression levels of a DNA repair signature of XRCC1, ATM, RAD50, MSH2, PARP1, RAD51, and MRE11 correlated with clinical outcomes. VERTU has 80% power at alpha = 0.1 to detect a large increase in 6PFS from 53% to 65% in the experimental arm, using a Fleming’s one-stage design. An Independent Data Safety Committee is monitoring the study including evaluation of feasibility and safety after completion of RT in the first 60 patients (stage 1) before the remaining patients are accrued (Stage 2). Progress: VERTU commenced 28th Oct 2015. As of Feb 2016, 8 study sites are open in Australia. An additional 4-10 study centres will be activated. To date, 16 patients have been screened and 5 randomized. ANZCTR # ACTRN12615000407594. Clinical trial information: ACTRN12615000407594.

Poster Session (Board #267a), Sat, 1:00 PM-4:30 PM

A mutation-specific peptide vaccine targeting IDH1R132H in patients with newly diagnosed malignant astrocytomas: A first-in-man multicenter phase I clinical trial of the German Neurooncology Working Group (NOA-16). First Author: Michael Platten, Brain Tumor Immunology, German Cancer Research Center, Heidelberg, Germany Background: Mutations in the gene for isocitrate dehydrogenase type 1 frequently occur in diffuse gliomas resulting in a point mutation (IDH1R132H) in most of the cases. Preclinical studies have shown that mutation-specific T helper (Th) cell responses spontaneously occur in patients with IDH1-mutated gliomas and that a peptide vaccine encoding IDH1R132H is therapeutic in a humanized mouse tumor model. Methods: NOA-16 (ClinicalTrials.gov Identifier: NCT02454634) is a multicenter, first-in-man, phase I clinical trial, which is planned to enroll 39 patients with newly diagnosed malignant astrocytomas with IDH1R132H mutations at eight German sites. The target population is molecularly enriched for an unfavorable prognosis within the IDH1-mutated subgroup by mandating (i) absence of co-deletion of 1p/19q and (ii) loss of alphathalassemia/mental retardation syndrome X-linked (ATRX) expression. Patients aged $ 18 years with a Karnofsky performance score of $ 70 will receive radiotherapy only (cohort 1), combined radiochemotherapy with temozolomide (TMZ, cohort 2) or TMZ only (cohort 3) as the primary treatment based on the decision of the local investigator. Over a period of 32 weeks a total of eight vaccinations with an IDH1R132H peptide emulsified in incomplete Freund’s adjuvant produced at a central good manufacturing practice (GMP) site will be administered subcutaneously with topical imiquimod starting three months after the initiation of primary treatment. The primary end points are safety and immunogenicity as measured by IDH1R132H-specific antibody and T cell responses. Secondary outcome measures include progression-free survival (PFS) and overall response rate (ORR). Enrollment started in June 2015. The DMC reviewed the trial in December 2015 and suggested that the trial continues as planned after the first three patients had completed the first vaccinations without a regime-limiting toxicity (RLT). Eight patients have been enrolled at three sites (as of January 2016). Clinical trial information: NCT02454634.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Central Nervous System Tumors TPS2083

Poster Session (Board #267b), Sat, 1:00 PM-4:30 PM

TPS2084

113s Poster Session (Board #268a), Sat, 1:00 PM-4:30 PM

MEVITEM: A European, randomized, open-label Phase I/II of vismodegib in combination with temozolomide versus temozolomide alone in adult patients with recurrent/ refractory medulloblastoma presenting an activation of the Sonic Hedgehog pathway. First Author: Didier Frappaz, Centre Leon ´ Berard, ´ Lyon, France

Umbrella protocol for phase I/IIa trials of molecularly matched targeted therapies plus radiotherapy in patients with newly diagnosed glioblastoma without MGMT promoter methylation Neuro Master Match (N²M²). First Author: Anne Hertenstein, Heidelberg University Hospital, German Cancer Research Center (DKFZ), Heidelberg, DE, Germany

Background: Therapeutic options are limited for refractory/relapsed adult medulloblastoma with activation of sonic hedgehog pathway (SHH-MB). Inhibition of this pathway may offer an attractive therapy. Vismodegib suppresses SHH signaling by binding to and interfering with the SMO transmembrane receptor. We postulate that vismodegib together with chemotherapy may kill more tumor cells than chemotherapy alone by blocking cell proliferation at different molecular levels (vertical association strategy). Methods: MEVITEM is a multicenter, randomized (2:1 ratio), open-label, Phase I/II aiming to evaluate the safety and clinical activity of the association vismodegib (V: daily 150mg/d) + temozolomide (T: D-5: 150 mg/ m2 for Cycle 1 and 200 mg/m2 for subsequent cycles) versus T alone in adult patients with recurrent/refractory SHH-MB. Main eligibility criteria are: histologically confirmed recurrent/refractory SHH-MB for which no known curative therapy exists, not previously treated with T, with evidence of measurable disease and documented activation of SHH pathway. Pathological review and SHH pathway analysis are performed centrally by IHC. This Phase I/II includes i) a safety run aiming to evaluate the safety of V+T and ii) a Phase II part aiming to evaluate the clinical activity of V+T measured by the 6-month non-progression rate. Patients enrolled in the safety run are included in the evaluation of the Phase II part. Considering that V+T would be uninteresting if 6-month non-progression rate # 30% and promising if $ 55% and using Simon’s optimal two-stage design (type I error rate:5%, power:80%), a sample size of 25 evaluable patients is required for Arm V+T (including 9 in stage I). The sample size in Arm T alone is 13 patients. Considering the rarity of SHH-MB, the few therapeutic options and the promising results reported with V in SHH-MB: the Sponsor considered the enrolment of patients previously treated by T in a 3rdindependent and parallel arm with V as single agent (Arm V). To date, 18 patients were enrolled (Arm V+T: 7, Arm T: 4, Arm V: 7), the Phase II part is ongoing. Clinical trial information: NCT02029001.

Background: Patients with newly diagnosed glioblastoma without O6-methyl guanine O6-methylatransferase (MGMT) promoter methylation do not benefit from alkylating therapy. Clinical trials aiming at replacing temozolomide (TMZ) with targeted agents in unselected patient populations have failed to demonstrate any benefit. Advances in the understanding of glioblastoma at a molecular level along with technological progress have led to the identification of key genetic alterations in a timely manner allowing for treatment decisions in newly diagnosed patients. In this setting, these alterations not only refine the sub-classification of glioblastoma, but also allow for subset specific treatments, with contemporary analyses allowing for differentiating prognostic and predictive biomarkers. Methods: N2M2 is an open label phase I/II umbrella trial for patients with newly diagnosed glioblastoma without MGMT promoter methylation to show safety, feasibility and preliminary efficacy (decision for future randomized phase II/III) of treatment with targeted compounds in addition to radiotherapy based on thorough molecular characterization. N2M2 is formally divided into a DISCOVERY and a TREATMENT aspect. DISCOVERY includes the use of panel, whole exome and transcriptome sequencing, methylome analysis using Illumina EPIC arrays, and gene expression arrays to find new, unexpected targets and get a comprehensive view on affected pathways, with dedicated bioinformatics evaluation (GUIDE), a Molecular Tumor Board and a timely initiation (# 4 weeks) of the postoperative treatment. Stratification for TREATMENT takes place in seven subtrials according to the best matching molecular alteration. For the phase I parts, a maximum of 9 patients will be enrolled in each cohort. A Bayesian criterion is used for continuous monitoring of toxicity. In the phase II trials, progression-free survival at six months is used as endpoint for efficacy. Patients with no matching alteration receive the standard chemoradiotherapy with TMZ and serve as contemporary, non-randomized controls with an estimated PFS6 of 40%.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

114s

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics

2500

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

Phase Ib study of BGJ398 in combination with BYL719 in patients (pts) with select advanced solid tumors. First Author: David Michael Hyman, Memorial Sloan Kettering Cancer Center, New York, NY Background: Phosphatidylinositol-3-kinase (PI3K) and fibroblast growth factor receptor (FGFR) pathway dysregulation can co-occur in tumors. In this study (NCT01928459), safety, tolerability, and preliminary activity of the pan-FGFR (BGJ398) + a-specific PI3K (BYL719) inhibitors were evaluated and maximum tolerated dose (MTD) was established in pts w/ advanced solid tumors w/ PI3K catalytic subunit (PIK3CA) mutations 6 FGFR alterations. Methods: Pts w/ solid tumors received once-daily (QD) BGJ on D1-D21 of each 28-D cycle (C) and BYL continuously during escalation (PIK3CA-mutant 6 FGFRaltered) and expansion (arm 1, PIK3CA-mutant [n = 15]; arm 2, PIK3CA-mutant + FGFR-altered [n = 10]; arm 3, PIK3CA-mutant + FGFR-altered breast cancer [n = 10]). Results: Of 62 pts enrolled, 44% had $ 4 prior therapies. The MTD, 125 BGJ + 300 BYL, was determined based on 32 evaluable pts’ data during escalation. C1 doselimiting toxicities (DLTs) occurred in 4 pts (table). During expansion, 24 pts in arms 1 (n = 12), 2 (n = 6), and 3 (n = 6) received the MTD. At data cutoff (Nov 6, 2015), 4 pts were ongoing; median treatment duration was 64 D (range, 7-456 D). Common adverse events (AEs) included diarrhea (60%), fatigue (53%), nausea (48%), and on-target AEs (hyperphosphatemia [37%], hyperglycemia [36%]); most were G1/2. At the MTD, 61% of pts had $ 1 BGJ dose reduction; 61% of pts had $ 1 BYL dose reduction. Eight pts had partial responses (PRs), w/ 4 confirmed in urothelial, head and neck, melanoma, and anal cancer. One pt w/ FGFR3-TACC3 fusion urothelial carcinoma w/ PR had a complete shrinkage in target lesions lasting 4 mo. Conclusions: The MTD for BGJ + BYL was determined as QD 125 BGJ on D1-D21 of each C + 300 BYL continuously. Given the high number of reductions w/ chronic treatment, lower starting doses may be warranted in future studies. Responses were observed in tumors w/ specific genetic alterations; however, it is not known if BGJ + BYL in combination offers improved efficacy over single agents. Clinical trial information: NCT01928459.

Dose-escalation DLTs. Dose, mg BGJ

BYL

Pts Evaluable/Enrolled, n/N

DLTs

20 20 40 75 90 100

300 400 300 300 300 300

3/4 4/4 5/6 5/6 4/5 5/6

G4 hyperglycemia, n = 1 G3 stomatitis, n = 1

125

300

6/7

2502

G3 stomatitis, n = 1; pt received , 50% of doses due to non-DLT AEs, n = 1

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

Antitumor activity and safety of FPA144, an ADCC-enhanced, FGFR2b isoform-selective monoclonal antibody, in patients with FGFR2b+ gastric cancer and advanced solid tumors. First Author: Jeeyun Lee, Samsung Medical Center, Seoul, Korea, The Republic of Background: FGFR2b-overexpressing gastric cancer is a distinct setting with poor prognosis. FPA144, a humanized monoclonal IgG1 antibody directed against FGFR2b, has been engineered for enhanced ADCC and specific FGFR2 signaling blockade. FPA144-001 is a two-part first-in-human Phase 1 study of FPA144 monotherapy: Part 1 dose escalation in patients with advanced solid tumors (Part 1A) and advanced gastric cancer (Part 1B); and Part 2 expansion to evaluate the safety, PK and early efficacy in selected gastric cancer patients treated at the recommended dose (RD). Methods: Part 1A was a 3+3 design to assess safety and PK of FPA144 and to establish a RD. Patients with gastric cancer were enrolled in parallel Part 1B cohorts to provide gastric cancer PK. Part 2 eligibility requires a diagnosis of gastric cancer and tissue for FGFR2b expression analysis. Tumor testing is done centrally using a proprietary FGFR2b-selective IHC assay. Results: Part 1 of the study enrolled 27 patients – 19 in Part 1A and 8 in Part 1B, including 6 patients with FGFR2b-overexpressing gastric cancer tumors in Part 1B. No DLTs, treatment-related SAEs or treatment delays were observed during dose escalation; 15mg/kg q2w is the dose being explored in Part 2. Fatigue and decreased appetite were the most common treatment-related AEs. In Part 1A, a patient with recurrent bladder cancer treated at 3mg/kg had radiographic and metabolic disappearance of lesions by day 50. Of 6 FGFR2boverexpressing gastric cancer patients in Part 1B, 2 achieved PRs (1 confirmed, 1 unconfirmed at time of analysis) and 3 demonstrated SD as best response (2 confirmed, 1 unconfirmed). Initial PK analysis demonstrates that FPA144 exposure in subjects with gastric cancer is similar to that in other solid malignancies. Updated safety, PK, and efficacy data will be presented. Conclusions: FPA144, an afucosylated monoclonal antibody directed against FGFR2b, has a well-tolerated safety profile: no DLTs in dose escalation and no treatment-related SAEs to date. Early evidence of singleagent anti-tumor effect in patients with FGFR2b-overexpressing gastric cancer and bladder cancer has been observed. Clinical trial information: NCT02318329.

2501

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

Safety, pharmacokinetics (PK), pharmacodynamics (Pd), and antitumor activity in a phase 1b study evaluating anti-ErbB3 antibody KTN3379 in adults with advanced tumors alone and with targeted therapies. First Author: Gerald Steven Falchook, Sarah Cannon Research Institute, Denver, CO Background: KTN3379 is a fully human anti-ErbB3 antibody with YTE halflife extension engineering that uniquely binds domain 3 locking ErbB3 in an inactive configuration. ErbB3 is implicated in tumor growth and treatment resistance. A phase 1b study of KTN3379 alone and in combination was undertaken to evaluate safety, PK, Pd, and efficacy (NCT02014909). Methods: Patients (pts) received KTN3379 alone at 5-20 mg/kg (N=21) or in combination at 15 or 20 mg/kg IV q3w with cetuximab (N=16), erlotinib (N=8), vemurafenib (N=4) or trastuzumab (N=9). Safety monitoring and RECIST-based tumor assessments were conducted. PK was evaluated using a two-compartment pharmacokinetic model. Pd markers included serum ErbB3. Archival tumor was tested for neuregulin (NRG) and ErbB3. Results: 58 total pts were treated. No DLTs occurred in 21 pts receiving KTN3379 alone. DLTs of diarrhea, mucositis and rash were observed in 5 of 37 pts treated with combinations. The most common adverse events across all combinations were diarrhea, fatigue, nausea and rash. KTN3379 at or above 15 mg/kg was tolerated in all combination arms. Linear PK was observed at doses between 10 and 20 mg/kg and exceeded target concentration for antitumor activity. There was a trend toward slower clearance and longer half-life of KTN3379 compared to related compounds, consistent with the engineered YTE. Responses included 1 ongoing confirmed CR with cetuximab in a pt with HNSCC who had progressed on cetuximab alone, and 2 confirmed PRs with vemurafenib in pts with BRAF-mutant NSCLC, one of which previously progressed on dabrafenib (2 of 3 responders were NRG+). Analysis of Pd correlates is in progress and will be presented at ASCO. Conclusions: KTN3379 can be safely combined with standard doses of cetuximab, erlotinib, vemurafenib or trastuzumab at 15 to 20 mg/kg. Efficacy was observed including a CR in a pt with HNSCC with cetuximab and PRs in 2 pts with BRAF-mutant NSCLC in combination with vemurafenib. A phase 2 study of the combination of cetuximab and KTN3379 in HNSCC is planned, in addition to currently ongoing phase 1b studies in SCCHN and BRAF-mutant thyroid cancer. Clinical trial information: NCT02014909.

2503

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

Phase I study of sapacitabine and seliciclib in patients with advanced solid tumors. First Author: Sara M. Tolaney, Dana-Farber Cancer Institute, Boston, MA Background: Sapacitabine is an oral nucleoside analogue; the active metabolite CNDAC generates ssDNA breaks that are converted to dsDNA breaks (DSB) during subsequent replication, resulting in cell death. CNDACinduced DSB repair is dependent on homologous recombination (HR). Seliciclib is an oral CDK2, 7 and 9 inhibitor, and sensitizes cells to CNDAC by decreasing DSB repair via compromise of HR protein activation. This phase I study evaluates sequential and concomitant sapacitabine and seliciclib treatment. Methods: Dose escalation was conducted in patients with incurable solid tumors with sapacitabine b.i.d. x 7 consecutive days (d 1-7) followed by seliciclib b.i.d. x 3 consecutive days (d 8-10) or sapacitabine q.d. concomitantly with seliciclib q.d. x 5 days per week x 2 weeks (d 1-5, 8-12). MTD was the highest dose level at which less than one-third of at least 6 patients experienced cycle 1 DLT. Skin biopsies were obtained to assess DNA damage following sapacitabine and seliciclib treatment. Results: 67 patients were treated including 45 BRCA mutation carriers (BRCA +ves). MTDs are sapacitabine 50 mg b.i.d./seliciclib 1200 mg b.i.d. and sapacitabine 250 mg q.d./seliciclib 200 mg q.d. respectively. DLTs were reversible elevations in transaminase and bilirubin, neutropenia/febrile neutropenia and pneumonia. The most frequent grade 3/4 adverse events included elevations in ALT (10%), AST (13%), bilirubin (6%) and alkaline phosphatase (7%), neutropenia (21%), febrile neutropenia (6%), hyperglycemia (6%), hypokalemia (6%), and abdominal pain (7%). Skin biopsies showed a 2.3-fold increase in gH2AX staining post-sapacitabine (n = 16; p = 0.007) and a further 0.58-fold increase post-seliciclib (n = 12; p = 0.069). Six confirmed PRs occurred in BRCA +ves with pancreatic, ovarian and breast cancer. Response durations range from 49 to . 224 weeks in 4 ovarian and breast patients. SD was observed in 10 additional BRCA +ves with durations ranging from 26 to 81 weeks in 5 ovarian and breast cancer patients. Conclusions: Sequential and concomitant sapacitabine and seliciclib is safe with preliminary antitumor activity (35% PR + SD) in BRCA +ves, the status of which may be a potential biomarker for response across multiple tumor types. Clinical trial information: NCT00999401.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics 2504

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

Phase I trial of a first-in-class ATR inhibitor VX-970 as monotherapy (mono) or in combination (combo) with carboplatin (CP) incorporating pharmacodynamics (PD) studies. First Author: Brent O’Carrigan, The Royal Marsden/ Institute of Cancer Research, London, United Kingdom Background: ATR is a regulator of cellular responses to replication stress, where it signals DNA damage repair by homologous recombination. Many cells depend on ATR to survive DNA damage. VX-970 is a potent, selective ATR inhibitor with marked preclinical antitumor activity in combo with chemotherapy. Methods: Patients (pts) with advanced cancers were enrolled in 2 parts. Part A: N=1 pt cohorts received VX-970 QW; 3 + 3 cohorts were commenced if ³ G2 drug-related toxicities occurred. Part B: 3 + 3 pt cohorts received CP on d1 + VX-970 on d2 and 9 in 21-day cycles. Timed pre- and post-VX-970 tumor biopsies were assessed for pS345 CHK1 levels by IHC in part B. Hematologic toxicities were modeled for pts in part B. Results: 26 pts were treated; M/F 10/16, ECOG PS 0/1: 9/17. Median age: 66 yrs (range 49-76 years). VX-970 was generally well tolerated as mono or CP combo with mainly G1-2 toxicities. CP dose delays and reductions occurred in 3/3 pts (B1); 2/2 pts (B2); 0/3 pts (B3); 1/6 pts (B4) due to neutropenia and/or thrombocytopenia. Clinical data were consistent with toxicity modeling that predicted probabilities of #5% G4 neutropenia and ,1% thrombocytopenia at B4. Recommended phase 2 doses (RP2D) were DL A3 (mono) and B4 (CP combo). VX-970 PK was dose proportional with no CP interaction. VX-970 PK exposures at combo RP2D led to tumor regression in mouse models. Tumor biopsy studies showed decreased pCHK1 by 73-90% after VX-970. In part A, a pt with ATM loss colorectal cancer had RECIST complete response (19 mths+) and 4 pts had RECIST stable disease (SD). In part B: a pt with germline BRCA1 mutant and platinum-refractory, PARP inhibitor-resistant ovarian cancer with a somatic Y220C TP53 mutation had RECIST partial response for 6 mths and 8 pts had RECIST SD. Conclusions: VX-970 is generally well tolerated as mono and in CP combo with early evidence of PD and antitumor activity. VX-970 is now being tested in phase 2 combo trials and as mono in pts with DNA repair defects. Clinical trial information: 2013. Clinical trial information: #8209;005100. Clinical trial information: #8209;34. VX-970 dose (mg/m2)

CP dose (AUC)

# pts

Dose limiting toxicities

A1 A2 A3 A4

60 120 240 480

-

1 2 1 7

-

B1 B2 B3 B4

240 120 120 90

5 5 4 5

3 3 3 6

Febrile neutropenia Hypersensitivity Febrile neutropenia

Dose Level (DL)

2506

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

A phase I dose escalation trial of PT2385, a first-in-class oral HIF-2a inhibitor, in patients with advanced clear cell renal cell carcinoma. First Author: Kevin Dale Courtney, The University of Texas Southwestern Medical Center, Dallas, TX Background: Inactivation of the von Hippel Lindau (VHL) tumor suppressor occurs in the majority of clear cell renal cell carcinomas (ccRCC). VHL deficiency stabilizes the transcription factor hypoxia-inducible factor (HIF)2a, an oncogenic driver of ccRCC. PT2385 is a first-in-class, orally administered selective small molecule inhibitor of HIF-2a that disrupts the heterodimerization of HIF-2a with HIF-1b and blocks the transcription of several genes involved in oncogenesis. PT2385 demonstrated anti-tumor efficacy in mouse xenograft models of ccRCC. Methods: Patients (pts) with advanced ccRCC and at least one prior therapy with a VEGF inhibitor were treated with PT2385 twice daily (BID) in a 3+3 Phase I design to determine the recommended Phase II dose (RP2D) and evaluate safety, pharmacokinetics (PK) and pharmacodynamics (PD). Plasma PK were measured on days 1 and 15 and PD weekly. Results: 26 pts were enrolled in dose escalation receiving PT2385 from 100 to 1800 mg BID. Median number of prior therapies was 4. No dose limiting toxicities were observed. Exposure generally increased with dose up to the 800 mg cohort without a further increase from 800 to 1800 mg. At 800 mg, PT2385 was rapidly absorbed (Tmax = 2 h) with a Cmax of 3.1 mg/mL and a half-life of 17 h. Targeted Cmin of 280 ng/mL was exceeded in 15/16 pts receiving $ 800 mg. Circulating plasma levels of the HIF-2a transcriptional target erythropoietin (EPO) rapidly decreased within 24 hrs in 15/16 pts receiving $ 800 mg and remained suppressed during the 16 wk sampling period. Based on safety, PK, and EPO PD data, 800 mg BID was selected as the RP2D. An additional 17 of 25 planned patients have been enrolled in an expansion cohort. Among the 43 pts, the most common all-grade adverse events (AEs) were anemia (14), fatigue (13) and peripheral edema (10). Grade $ 3 treatment related AE’s were anemia (3), pulmonary embolism (1), and lymphopenia (1). To date, 1 CR, 1 PR, and 24 SD have been observed—of these pts, 7 are ongoing . 7 months. Conclusions: Continuous PT2385 administration was safe and tolerable and demonstrated early evidence of clinical activity in heavily pre-treated pts with advanced ccRCC. Clinical trial information: NCT02293980.

2505

115s

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

Phase I trial of a dual TOR kinase and DNA-PK inhibitor (CC-115) in advanced solid and hematologic cancers. First Author: Pamela N. Munster, University of California, San Francisco, San Francisco, CA Background: CC-115 is a potent inhibitor of DNA-PK and TOR kinase (DNAPKi/TORKi) with broad pre-clinical anti-tumor activity. Methods: Subjects with relapsed/refractory advanced solid and hematologic cancers enrolled in a first-in-human Phase 1a/b study of CC-115 given orally once or twice daily until disease progression in dose escalation and expansion cohorts. Results: 44 subjects evaluated across 10 dose escalation cohorts (0.5 – 40 mg) established 10 mg twice-daily (BID) as the recommended dose for cohort expansion in 62 additional subjects with glioblastoma multiforme (GBM: 14), head and neck squamous carcinoma (HNSCC: 18), chronic lymphocytic leukemia (CLL: 8), and ETS-overexpressing tumors (castrationresistant prostate cancer [CRPC: 12] and Ewing’s sarcoma [ES: 10]). The most common (. 20%) expansion phase related adverse events were fatigue (37%), nausea (31%), decreased appetite (29%), and hyperglycemia (24%). Linear, dose proportional exposure was observed with a terminal half life of 4 to 8 hrs (mean steady state Cmax 75 ng/mL, AUC0-24 776 ngxhr/mL at 10mg BID). Gastric pH-altering drugs and food did not impact PK. Plasma accumulation was minimal, renal clearance was negligible and brain penetration was confirmed in resected GBM (tumor/plasma ratio 0.7). Inhibition of TORC1 (pS6, p4EBP1) and TORC2 (pAKT) biomarkers in blood cells was exposure-dependent and consistently achieved with 10 mg BID dosing. DNA-PK inhibition was demonstrated ex vivo in circulating CLL cells from one subject dosed with CC-115. Tumor responses during dose escalation included 1 complete response (endometrial; .3yr), 1 partial response (melanoma), and stable disease (SD) in 18 (41%) subjects. In the efficacyevaluable expansion cohorts, SD was observed in CRPC (64%), HNSCC (53%), GBM (non-progression 21%), and ES (22%). One PR, and 3 PR with lymphocytosis (IWCLL), were seen in CLL. Conclusions: CC-115 was well tolerated with toxicities comparable with approved mTOR inhibitors. Evidence of TORC1/TORC2/DNA-PK pathway inhibition was observed as well as preliminary signals of broad anti-tumor activity. Phase 2 trials combining CC115 with androgen deprivation in CRPC and radiation in GBM are planned. Clinical trial information: NCT01353625.

2507

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

Phase 1, open-label, dose-escalation and expansion study of ABT-165, a dual variable domain immunoglobulin (DVD-Ig) targeting both DLL4 and VEGF, in patients (pts) with advanced solid tumors. First Author: Michael S. Gordon, Pinnacle Oncology Hematology/HonorHealth Research Institute, Scottsdale, AZ Background: Targeting angiogenesis through vascular endothelial growth factor (VEGF) pathway has been a successful clinical strategy in multiple malignancies. Delta like ligand-4 (DLL4) is a NOTCH pathway ligand involved in angiogenesis and maintenance of tumor initiating cells (TICs). ABT-165 is a DVD-Ig that inhibits both VEGF and DLL4 thereby potentially suppressing both tumor angiogenesis and TICs. Methods: In a 3+3 doseescalation (DE) design, ABT-165 was administered IV at doses ranging from 2.5 to 7.5 mg/kg once every 14 days (2 doses in 28-day cycle) (NCT01946074). After an amendment to address hypertension (HTN), at least 6 subjects per cohort were enrolled in subsequent DE cohorts starting at a dose of 1.25 mg/kg. Cohorts were preferentially enriched for ovarian cancer (OC) pts. Tumor and cardiac assessments were performed every 2 cycles. Results: As of January 6, 2016, 54 pts received at least one dose of ABT-165. The PK demonstrated target-mediated disposition with preliminary mean half-life ranging from 4.9 to 8.9 days. Biomarkers were consistent with engagement of both VEGF and DLL4 with NOTCH pathway modulation. No maximum tolerated dose was identified within the 28-day dose limiting toxicity (DLT) window; however, safety concerns beyond Cycle 1 precluded DE above 7.5 mg/kg. DLTs included grade 3 HTN and headache (HA). Treatment-related adverse events (AEs) occurring in $ 10% pts (including all dose levels and all grades) were: HTN (55.6%), HA (24.1%), fatigue (18.5%), diarrhea (14.8%) and nausea (11.1%). Other AEs of interest with respect to mechanism of action included proteinuria (9.3%), pulmonary HTN (7.4%), gastrointestinal perforation (GIP) (3.7%) and stroke (1.9%). 4/13 (30.8%) heavily pretreated OC pts had partial response, 1 of which (anti-VEGF na¨ıve) was on treatment for 14 Cycles. Conclusions: ABT165 displayed anti-VEGF-like toxicity including HTN and GIP. ABT-165 monotherapy has demonstrated anti-tumor activity in refractory OC. Longterm consequences of dual VEGF and DLL4 inhibition need to be evaluated. An ongoing phase Ib study will investigate combination of ABT-165 and chemotherapy. Clinical trial information: NCT01946074.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

116s 2508

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics Oral Abstract Session, Mon, 8:00 AM-11:00 AM

MRX34, a liposomal miR-34 mimic, in patients with advanced solid tumors: Final dose-escalation results from a first-in-human phase I trial of microRNA therapy. First Author: David S. Hong, Department of Investigational Cancer Therapeutics (Phase 1 Program), The University of Texas MD Anderson Cancer Center, Houston, TX Background: Each miRNA modulates expression of hundreds of genes across distinct cellular pathways, giving miRNA-based therapy potential to simultaneously repress multiple oncogenic processes in the tumor microenvironment, including growth and proliferation, resistance, cancer stem cells, metastasis, and immune evasion. The tumor suppressor miR-34a down-regulates expression of .30 oncogenes (eg, MET, MEK1, PDGFRa, CDK4/6, BCL2, WNT 1/3, NOTCH1, CD44), as well as genes involved in tumor immune evasion (eg, PD-L1, DGKz). MRX34, a liposomal miR-34a mimic, is a potential first-in-class miRNA therapy for cancer. Methods: Pts with advanced solid tumors (N=99) were enrolled in a standard 3+3 dose escalation trial of MRX34 infused IV on a biweekly (BIW) or daily x 5 (QDx5) schedule. The primary objective was MTD and recommended phase II dose (RP2D); secondary objectives were safety, PK, and biological and clinical activity. Results: Pts with HCC (42), RCC (2), acral melanoma (2), and other tumor types (53) were enrolled. Most were Caucasian (58%) and male (65%), with median age 60 y and 4 median prior therapies (range, 1–12). Most common AEs (% all/grade 3/4) were fever (64/2/0), chills (54/0/0), fatigue (49/4/0), back pain (44/8/0), nausea (40/2/0), vomiting (25/2/0), anorexia (25/2/0), and headache (25/0/0). Lab abnormalities (% grade 3/4) included lymphocytopenia (37/18), thrombocytopenia (22/2), elevated AST (21/3), neutropenia (19/9), and hyponatremia (18/3). Required steroid premedication reduced infusion-related AEs, and the QDx5 schedule improved safety and PK, with AUC increased ~10-fold on day 5 vs 1. The QDx5 MTD/RP2D was 70 mg/m2 for HCC and 93 mg/m2for non-HCC solid tumors. Analysis of RNA from WBCs showed dose-dependent repression of miR-34a target oncogenes, including FOXP1, BCL2, HDAC1, and CTNNB1. Three pts (HCC, RCC, acral melanoma) achieved prolonged confirmed PR per RECIST and 14 pts had SD (median duration, 136 days; range, 79–386). Conclusions: MRX34 has a manageable toxicity profile and evidence of activity in HCC, RCC and melanoma. Enrollment into phase Ia expansion cohorts continues, and phase II studies are being planned. Clinical trial information: NCT01829971.

2510

Poster Discussion Session; Displayed in Poster Session (Board #210), Sun, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sun, 11:30 AM-12:45 PM

2509

Poster Discussion Session; Displayed in Poster Session (Board #209), Sun, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sun, 11:30 AM-12:45 PM

Phase I study of anti-mesothelin antibody drug conjugate anetumab ravtansine (AR). First Author: George R. Blumenschein, Department of Thoracic/Head & Neck Medical Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX Background: AR (BAY 94-9343) comprises a novel fully human anti-mesothelin IgG1 antibody conjugated to the maytansinoid tubulin inhibitor DMR. A phase I study was conducted in patients (pts) with advanced solid tumors. Methods: Eligible pts with solid tumors refractory to standard treatment, adequate organ function, and ECOG performance status (PS) # 1 were enrolled. AR was administered IV on 2 schedules: every 21 days (q3w) in 45 pts (21 mesothelioma [Meso], 9 pancreatic, 5 breast, 4 ovarian cancer [OC], 6 other) at doses ranging from 0.15 to 7.5 mg/kg and in an expansion cohort at 6.5 mg/kg (12 Meso and 20 OC); and weekly (qw) in 71 pts (31 Meso and 40 OC) randomized to either 1.8 mg/ kg or 2.2 mg/kg. Pts were assessed for AEs weekly C1-C3 and on D1 in subsequent cycles. Tumor response was assessed q6w C1-C8 and q12w thereafter. Mesothelin expression in archival tumor samples was assessed by IHC (SP74, Ventana). Results: A total of 147 pts were evaluable: 64% female, mean age 61 yrs (range 18-88), PS 0/1 29%/71%. The 7.5 mg/kg q3w dose was not tolerated. The maximally tolerated dose (MTD) was 6.5 mg/kg q3w; 15 of 32 (47%) pts in MTD expansion had dose reduction to 5.5 mg/kg q3w. Adverse events (AEs) were similar to the mesothelin-ADC inhibitor class, including reversible keratopathy, asymptomatic liver function test increases, and gastrointestinal disorders. Drugrelated (DR) serious AEs were low (11 total; 5 at the q3w MTD) and highly similar to this class of inhibitors. No DR deaths were reported. One patient in each expansion cohort discontinued due to a reversible DRAE. Partial response (PR), stable disease (SD), and disease control rates (DCR) were 18%/47%/65% at the q3w MTD (n = 38), 6%/35%/41% at 1.8 mg/kg qw (n = 34), and 3%/50%/53% at 2.2 mg/kg qw (n = 36). At the q3w MTD, the PR/SD/DCR for OC pts (n = 22) was 9%/50%/59% and for Meso pts (n = 16) was 31%/44%/75%, respectively. Five of 16 (31%) Meso pts treated at the MTD had a durable PR ( . 600 days in 4); the response rate in 2ndline Meso pts was 50% (5/10). Durable responses were also observed in OC pts. Conclusions: Preliminary data with AR at 6.5 mg/kg q3w showed durable PRs in pts with advanced Meso. All DRAEs were reversible. A phase II trial in 2nd-line metastatic pleural Meso is ongoing (NCT02610140). Clinical trial information: NCT01439152.

2511

Poster Discussion Session; Displayed in Poster Session (Board #211), Sun, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sun, 11:30 AM-12:45 PM

Phase 1, open-label, dose-escalation and expansion study of ABBV-399, an antibody drug conjugate (ADC) targeting c-Met, in patients (pts) with advanced solid tumors. First Author: John H. Strickler, Duke University Medical Center, Durham, NC

A phase I study of SAR566658, an anti CA6-antibody drug conjugate (ADC), in patients (Pts) with CA6-positive advanced solid tumors (STs) (NCT01156870). First Author: Carlos Alberto Gomez-Roca, Institut Universitaire du Cancer de Toulouse, Toulouse, France

Background: The c-Met receptor is overexpressed in multiple tumors. ABBV399 is a first-in-class antibody-drug conjugate (ADC) comprised of the antic-Met antibody, ABT-700, and monomethyl auristatin E. Although prior efforts at targeting c-Met overexpression have met with limited clinical success, an ADC directed against c-Met represents a unique strategy to deliver a potent cytotoxin directly to c-Met+ tumor cells. Methods: In a 3+3 dose escalation design, ABBV-399 was administered at doses ranging from 0.15 to 3.3 mg/kg once every 21 days to pts with metastatic solid tumors (NCT02099058). The recommended single-agent phase 2 dose (RPTD) of ABBV-399 was then studied in a dose-expansion cohort in pts with c-Met+ (immunohistochemistry (IHC) $2+) non-small cell lung cancer (NSCLC). cMet overexpression was assessed by an IHC assay utilizing the SP44 antibody (Ventana). Results: As of January 4, 2016, 45 pts received at least 1 dose of ABBV-399. Dose-proportional increases of area under the curve (AUC) for ABBV-399 and total antibody were observed after single dose administration. Half-lives for ABBV-399 and total antibody were approximately 2-4 days. Dose-limiting toxicity (DLT) of febrile neutropenia occurred in 1 pt at 3 mg/kg and 1 pt (with septic shock) at 3.3 mg/kg. The RPTD of 2.7 mg/kg was chosen based primarily on safety and tolerability. There were no toxic deaths. Treatment-related adverse events occurring in $10% of pts (including all dose levels and all grades) were fatigue (20%), nausea (17.8%), neuropathy (13.3%), decreased appetite (13.3%), hypoalbuminemia (11.1%), and vomiting (11%). 3/10 evaluable c-Met+ NSCLC pts had a partial response; the duration of response ranged from 1+ to 5 months. There were no responses among 9 pts with c-Met negative tumors. Conclusions: ABBV-399 is well tolerated at the RPTD of 2.7 mg/kg and has demonstrated promising anti-tumor activity in pts with cMet+ NSCLC. The study will continue to assess the anti-tumor activity and safety of ABBV-399 in c-Met+ pts both as monotherapy and in combination with standard of care. Clinical trial information: NCT02099058.

Background: SAR566658 (SAR) is a maytansinoid-loaded ADC targeting CA6, a sialoglycotope of MUC-1 over-expressed in solid tumors and rarely in normal tissues. Methods: Phase I dose escalation and expansion study to evaluate SAR intravenous administration as single agent every 3 weeks (q3w) and 2 weeks (q2w). Primary objective included dose limiting toxicities (DLTs), maximum tolerated dose (MTD) and recommended dose (RD). Secondary objectives were PK, PD and preliminary efficacy assessment (RECIST1.1). Results: One hundred and fourteen patients (Pts) with heavily pretreated solid tumors expressing CA6 in $ 30% tumor cells with an intensity 2/3+ by immunohistochemistry were enrolled. SAR was administered across 11 dose levels (DLs) (10 to 240 mg/m² q3w and 120 mg/m² q2w). DLTs [1 grade (Gr) 3 diarrhea and 2 Gr3 keratitis] were observed at 240 mg/ m². Initial RD was 190 mg/m² however a high incidence of keratopathy was seen mainly at cycle 2. PK/safety simulation, preserving drug exposure and limiting keratopathy incidence, proposed two alternative schedules: 90 mg/ m² D1,D8 q3w and 120 mg/m² q2w. Most common adverse event (AE) was reversible Gr2/3 keratopathy in 41/114 (36%) (Gr3 in 9 Pts). Among these 41 Pts, 15/23 (65%) received 190 mg/m², 12/33 (36%) 150 mg/m², 6/8 (75%) 240 mg/m², 6/17 (35%) 90 mg/m² D1,D8 and 2/16 (13%) 120 mg/ m² q2w. Prophylaxis with vasoconstrictor and steroids eye drops implemented in 8 patients in alternative schedules prevented keratopathy. Other AEs were fatigue (32.6%), peripheral neuropathy (31.6%), GI disorders [(nausea (29%), abdominal pain (26%), diarrhea (25%)] and neutropenia (2.6%). Low grade liver and renal abnormalities were noted.Tumor regression was noted in about 60% Pts at 190 and 90 mg/m² D1,D8 and 35% Pts at 150 and 120 mg/m² q2w. One complete response (ovary), 8 partial responses (3 breast, 2 ovary, 1 NSCLC, 1 cervix and 1 bladder) and 39% stable disease were noted. Highest overall response rate (ORR) was observed in 2/15 at 90 mg/m² D1,D8 and 3/23 at 190 mg/m². Conclusions: SAR 90 mg/m² D1,D8 q3w provided a favorable safety profile and encouraging antitumor activity and is selected as the RD for further clinical development. Clinical trial information: NCT01156870.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics 2512

Poster Discussion Session; Displayed in Poster Session (Board #212), Sun, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sun, 11:30 AM-12:45 PM

2513

117s

Poster Discussion Session; Displayed in Poster Session (Board #213), Sun, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sun, 11:30 AM-12:45 PM

Preliminary activity in the first in human study of the first-in-class fatty acid synthase (FASN) inhibitor, TVB-2640. First Author: Emma Jane Dean, University of Manchester, The Christie Hospital NHS Foundation Trust, Manchester, United Kingdom

Phase I trial of first-in-class ATR inhibitor VX-970 in combination with gemcitabine (Gem) in advanced solid tumors (NCT02157792). First Author: Elizabeth R. Plummer, Newcastle University, Newcastle upon Tyne, United Kingdom

Background: FASN inhibition is a novel approach to cancer treatment involving selective disruption of palmitate biosynthesis that, in tumor cells, leads to apoptosis. TVB-2640 is an oral, first-in-class, small molecule reversible inhibitor of FASN that demonstrates in vivo antitumor effects. We previously reported the results of dose escalation and now present evidence of preliminary activity from the dose expansion. Methods: This ongoing multicenter Phase I trial enrolled patients (pts) with advanced solid tumors with adequate organ function. TVB-2640 was given orally once daily at the MTD (100 mg/m2) as monotherapy (mono) or in combination (combo) with weekly IV paclitaxel (80 mg/m2). Results: The most common AE’s seen in both groups [mono, N=53; combo, N=47] were: alopecia (57%), palmarplantar erythrodysesthesia (PPE) (36%), dry eye (13%), and increased lacrimation (11%). Gr 3 AE’s include corneal edema (3%) and PPE (10%). Other toxicities were # Gr 2 and only minor GI symptoms occurred. All toxicities were reversible on dose interruption and no enhancement of paclitaxel toxicity was observed when given with TVB-2640. 3 confirmed RECIST partial responses (cPR) all in combo, and multiple cases of prolonged stable disease (SD) (.16 wks) in both arms was seen. 16 NSCLC pts were accrued: 6 were KRAS Mut and, of those, 3 achieved SD with monotherapy for 17, 24 and 38 wks. 2 pts treated in combo had SD for 24 and 21 wks. One combo NSCLC pt (no prior taxane) was a cPR at week 8 and continues on study at wk 22. 3/12 breast cancer pts treated in combo had SD for 20 (TNBC), 24 and 25 wks. One combo breast cancer pt achieved a cPR at wk 12 and continues on study at wk 24. One cPR was seen in a combo treated pt with peritoneal carcinoma, along with a 58% reduction in CA-125 levels. Further reductions in CA-125 ranging from 58-98%, were seen in 5 of 12 ovarian pts. All ovarian pts had received prior taxanes and most, if not all, were taxane resistant upon study entry. Conclusions: TVB-2640 demonstrated prolonged SD when given in monotherapy and cPR when combined with weekly paclitaxel. Responses were seen across multiple tumor types, including KRASmut NSCLC, ovarian and breast cancer. Further exploration in specific tumor types is ongoing. Clinical trial information: NCT02223247.

Background: ATR is a regulator of the cellular response to replication stress, where it signals DNA damage repair through the homologous recombination pathway.Many cancer cells depend on ATR to survive DNA damage. VX-970 is a potent, selective inhibitor of ATR with marked preclinical anticancer activity in combination with DNAdamaging chemotherapy in preclinical models. A multicenter phase I study of VX-970 in combination with Gem was performed. Methods: Patients (pts) with advanced solid tumors measurable by RECIST 1.1 received IV VX-970 in combination with Gem in a 3+3 dose-escalation design. Gem was administered on days 1 and 8 and VX-970 on days 2, 9 and 16 of each 21-day cycle. Results: 50 pts were treated (28 M/22 F), median age 62 yrs (range 28-79 yrs), ECOG PS 0/1: 15/35. Primary tumors were colorectal (n=15), NSCLC (n=6), breast (n=4), pancreatic (n=2) and other (n=23). Gr 3/4 treatment-related AEs occurred in 25 pts. Maximum tolerated dose was not reached. VX-970 exposure was approximately linear based on Cmax and AUC0-‘ ; terminal T1/2 was »16 h. Best response was PR in 4 pts (breast, NPC, NSCLC, and CUP). Two pts with cancers not typically responsive to chemotherapy had prolonged SD: PFS was 415 days (papillary ca) and .191 days (GIST). Exposures achieved at »90mg/m2 correlated to those showing maximum preclinical activity. In a parallel phase I study this dose blocked ATR activity by »75% in patient tumor biopsies. Preliminary data from a non-randomized comparison of cohorts 1-4 (VX-970 , 90mg/m2 with Gem 875 mg/m2) and 5-7 (VX-970 $ 90mg/m2 with Gem 500mg/m2) showed median PFS of 8.3 and 29.3 weeks, respectively. Conclusions: The recommended phase II dose is VX-970 210 mg/m2 and Gem 1000 mg/ m2. Preliminary evidence of activity was observed. Studies in biomarker-defined NSCLC and SCLC are ongoing, as are additional studies with VX-970 and platinum in TNBC. Clinical trial information: NCT02157792.

2514

2515

Poster Discussion Session; Displayed in Poster Session (Board #214), Sun, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sun, 11:30 AM-12:45 PM

First-in-human trial of ONC201 in patients with refractory solid tumors. First Author: Mark N. Stein, Rutgers Cancer Institute of New Jersey, New Brunswick, NJ Background: ONC201 is an orally active, first-in-class small molecule activator of the integrated stress response that selectively upregulates ATF4 to trigger tumor cell death. Preclinical studies have established the compelling antitumor efficacy of the small molecule in a range of advanced solid tumors as well as its excellent safety profile in animals. A first-in-human phase I study of ONC201 was conducted to determine its recommended phase II dose (RP2D) in patients with advanced solid tumors. Methods: This openlabel study treated 10 patients during the single-patient cohort dose escalation portion of the trial. The human starting dose was 125mg, which represents 1/10th of the no-adverse-adverse-effect-level in animal toxicology studies using standard allometric conversion. Results: The RP2D was defined as 625 mg and no Grade . 1 drug-related adverse events occurred on this trial, which enrolled patients at increasing dose levels of 125mg, 250mg, 375mg, 500mg, and finally 625mg. Pharmacokinetic analysis revealed saturation of absorption at 375mg, suggesting dose escalation above 625mg was not warranted. The terminal half-life was 9.6 hours, Cmax values ranged from 1.5 - 7.5 ug/mL (~3.9-19.4 uM), and the AUC was 25 hr ug/L. Prolonged induction of caspase-cleaved keratin 18 and induction of TRAIL was observed in serum samples. Eight of 10 patients had stable disease and one patient with endometrial cancer underwent a mixed response. 17 of 25 additional patients have been accrued in an ongoing expansion phase. The available results from the expansion cohort confirm the excellent safety, and the pharmacokinetic and pharmacodynamic profiles of ONC201 at the recommended phase II dose of 625mg every three weeks. Conclusions: This study demonstrates that ONC201 is well tolerated and active at an oral dose of 625mg, warranting further evaluation in lead indications. Clinical trial information: NCT02250781.

Cohort

VX-970 dose (mg/m2)

Gem dose (mg/m2)

No. pts treated/ No. pts evaluable for DLTs

18 36 60 72 90 140 210 210 210 210

875 875 875 875 500 500 500 750 875 1000

3/3 3/3 4/3 7/6 6/6 8/6 3/3 3/3 7/6 6/6

1 2 3 4 5 6 7 8 9 10

No. DLTs (DLT)

2 (gr 3 thrombocytopenia; gr 3 elevated ALT and fatigue) 1 (gr 3 elevated AST) 1 (gr 3 elevated AST and ALT, gr 2 elevated ALP)

Poster Discussion Session; Displayed in Poster Session (Board #215), Sun, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sun, 11:30 AM-12:45 PM

Phase 1b of WNT inhibitor ipafricept (IPA, decoy receptor for WNT ligands) with carboplatin (C) and paclitaxel (P) in recurrent platinum-sensitive ovarian cancer (OC). First Author: Roisin Eilish O’Cearbhaill, Memorial Sloan Kettering Cancer Center and Weill Cornell Medical College, New York, NY Background: The WNT pathway is an important oncologic driver of OC. The first-inclass recombinant fusion protein IPA blocks WNT signaling through binding of WNT ligands. In patient-derived OC xenografts, IPA inhibits growth, reduces cancer stem cell frequency, promotes differentiation, synergizes with taxanes and shows greatest efficacy when given 2 to 3 days before the taxane. In Phase 1a, IPA was well tolerated with dysgeusia, fatigue, muscle spasms and anorexia as most common adverse events (AEs). Target modulation was observed in hair follicles at $ 2.5 mg/kg every 3 weeks (q3w). Consequent to the clinical experience with, mainly, vantictumab (VAN), a 2nd WNT inhibitor, serum biochemical markers of bone turnover were used to trigger bone-protective zoledronic acid (ZA)therapy, if appropriate. Methods: Dose escalation started with a standard 3+3 design for IPA/ C/P with q3w intravenous IPA on Day 1, in cycles 1 to 6 with C (AUC = 5 mg/mL $ min) and P (175 mg/m2). For enhanced bone safety the trial was revised to 6patient cohorts with a q3w regimen of IPA on Day 1 and C/P on Day 3 (IPA→C/P). Objectives were determination of maximum tolerated dose, safety, pharmacokinetics, pharmacodynamics and efficacy. Results: To date, 17 patients have been treated; the 1st 7 patients in 2 cohorts of q3w IPA/C/P (5 & 10 mg/kg) and the next 10 patients in 2 cohorts of q3w IPA→C/P (2 & 4 mg/kg). The revised safety schedule was implemented following fragility fractures in the VAN and IPA Phase 1 programs, including 1 G1 pelvic fracture in this study. All 10 IPA→C/P patients received ZA at baseline for postmenopausal status without further fragility fractures. IPA-related AEs $ 15% were nausea, fatigue, neutropenia, alopecia, anorexia and vomiting. Only related G $ 3 AEs were neutropenia (3 x G3, 1 x G4) and hypophosphatemia (1 x G3). Of the 17 evaluable patients, 6 (35%) had a best response of complete response (CR), 8 (47%) partial response (PR) and 3 (18%) stable disease. Conclusions: IPA is well tolerated in combination with sequential C/P. The revised safety plan resulted in baseline bone-protective ZA for 100% of patients. 82% of patients achieved CR or PR. Dose escalation continues. Clinical trial information: NCT02092363.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

118s 2516

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics Poster Discussion Session; Displayed in Poster Session (Board #216), Sun, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sun, 11:30 AM-12:45 PM

Phase 1b study of WNT inhibitor vantictumab (VAN, human monoclonal antibody) with paclitaxel (P) in patients (pts) with 1st- to 3rd-line metastatic HER2-negative breast cancer (BC). First Author: Monica M. Mita, CedarsSinai Samuel Oschin Comprehensive Cancer Institute, Los Angeles, CA

2517

Poster Discussion Session; Displayed in Poster Session (Board #217), Sun, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sun, 11:30 AM-12:45 PM

Co-targeting BRAF with mTOR inhibition in solid tumors harboring BRAF mutations: A phase I study. First Author: Shiraj Sen, The University of Texas MD Anderson Cancer Center, Houston, TX

Background: The WNT pathway is an important oncologic driver of BC. VAN is a first-in-class antibody that blocks canonical WNT signaling by interacting with 5 Frizzled receptors (1, 2, 5, 7, 8). In pt-derived BC xenografts, VAN inhibits growth, reduces cancer stem cell frequency, promotes differentiation and synergizes with taxanes. A 6-gene expression WNT biomarker, without prognostic value, predicts VAN efficacy in pt-derived BC xenografts. In Phase 1a, VAN was well tolerated except for bone toxicity mitigated with blood-based bone marker monitoring and zoledronic acid (ZA) administration. Target modulation was observed in tumor tissues at VAN $ 0.5 mg/kg every 2 weeks (q2w). Methods: Dose escalation started with intravenous VAN q2w using a standard 3+3 design, and was then pursued for q4w with 6-pt cohorts. P at 90 mg/m2 was given on Days 1, 8 and 15 of each 28-day cycle. Objectives were determination of maximum tolerated dose, safety, pharmacokinetics, pharmacodynamics, predictive biomarkers and efficacy. Results: 23 pts have been treated to date, the 1st 11 in 3 q2w cohorts (3.5, 7 & 14 mg/kg). After 2 Grade (G) 2 fragility fractures, 12 pts were treated with a revised safety plan in 2 q4w cohorts (3 & 5 mg/kg) without further fragility fractures. VAN-related adverse events (AEs) $ 10% were fatigue, constipation, neutropenia, diarrhea, nausea and abdominal pain. Only related G $ 3 AEs were G3 neutropenia (n = 3), leukopenia & pelvic pain. For q4w pts, baseline ZA, if indicated, increased monitoring and more sensitive bone marker triggers for ZA resulted in bone-protective ZA therapy by end of Cycle 1 in 100% of pts. Of 21 evaluable pts, 7 (33%) had a partial response and 6 (29%) stable disease. Conclusions: VAN is well tolerated in combination with paclitaxel. A revised safety plan appears to have addressed bone toxicity encountered early in the study. Updated results, including correlation of the 6-gene biomarker in pt tumors with progression-free and overall survival, will be presented Clinical trial information: NCT01973309.

Background: Vemurafenib, dabrafenib and trametinib are FDA approved for treating BRAFV600-mutant melanoma. Vemurafenib has activity in multiple BRAFV600-mutant non-melanoma, as well. Unfortunately, patients acquire resistance to BRAF targeted monotherapy. Because preclinical data suggest mTOR activation may mediate resistance, we conducted a phase I study adding mTOR inhibitor everolimus to vemurafenib and evaluated tolerability, maximum tolerated dose (MTD) and response in all patients including those who progressed on prior BRAF targeted monotherapy. Methods: We performed a dose escalation study of vemurafenib twice daily and everolimus daily in 28 day cycles with an expansion cohort treated at MTD. Results: Twenty patients (14 male) with BRAF-mutated tumors (17-V600E, 2-V600K, 1-G469A) were enrolled. 7 had melanoma, 5 had CNS tumors, 4 had thyroid cancer, and 1 each had appendiceal, colorectal, NSCLC, and unknown primary cancer. Median age of 18 adult patients was 64; 2 pediatric patients were 10 and 13. 10 patients had prior therapy with BRAF inhibitor (50%), 10 had prior phase I trial (50%), 18 had surgery (90%), 11 had radiation (55%), 14 had chemo (70%). No DLT was observed with vemurafenib 720 mg BID and everolimus 5 mg QD. 3 patients experienced DLTs (rash, fatigue) when everolimus 10 mg QD was added to vemurafenib. Overall, 13/20 patients (65%) responded: 4 patients (26%) had a partial response, 9 (47%) had stable disease as best response. Of the 10 patients who previously progressed on BRAF inhibitor without mTOR inhibition, 2 had partial response and 4 had stable disease with the addition of everolimus. Two patients with CNS tumors remain on protocol after 7 and 19 cycles. Conclusions: The combination of vemurafenib 720 mg BID and everolimus 5 mg QD is safe, well-tolerated, and has demonstrated activity in patients with BRAF-mutant advanced cancers, including those previously treated with a BRAF inhibitor. Studies to help identify patients warranting up-front dual BRAF/mTOR inhibition are ongoing. Clinical trial information: NCT01596140.

2518

2519

Poster Discussion Session; Displayed in Poster Session (Board #218), Sun, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sun, 11:30 AM-12:45 PM

Final results of a first-in-human study evaluating the safety, pharmacology and initial efficacy of MM-151, an oligoclonal anti-EGFR antibody in patients with refractory solid tumors. First Author: Wael A. Harb, Horizon Oncology Center, Lafayette, IN Background: MM-151 is a mixture of three IgG1 antibodies designed to bind simultaneously to non-overlapping EGFR epitopes and inhibit ligandmediated signal amplification, downregulate EGFR expression, and enhance immune-effector activities (ADCC, CDC). A Phase 1 study was completed to assess safety, tolerability, pharmacology and preliminary clinical activity of MM-151 alone and in combination with irinotecan. Methods: This study evaluated MM-151 when administered as a monotherapy and in combination with irinotecan. An expansion cohort was also enrolled to evaluate clinical activity in EGFR-refractory metastatic CRC patients (pts). Subset analyses and additional biomarker evaluations were performed in EGFR-driven indications. Results: A total of 111 patients were enrolled (87 pts on monotherapy). The most common tumor types were CRC (43 [39%]), NSCLC (11 [10%]) and SCCHN (14 [13%]). Weekly dose selection was previously reported. Reported here are final safety and biomarker data. Most adverse events were CTCAE Grades 1 and 2. The most common Grade 3 (G3) or higher non-infusion related reaction (IRR) AEs included EGFR-pathway toxicities such as maculopapular rash (11 [9.9%]), hypomagnesemia (10 [9%]), general rash (8 [7.2%]) and diarrhea (8 [7.2%]). G3 IRRs occurred in 8/57 (14%) of pts enrolled at the nonoptimized dosing guidelines vs. 1/57 (1.7%) of pts in the optimized dosing cohorts. Biomarker analyses revealed a complex set of resistance mechanisms. Notably, clinically meaningful durations were achieved in patients presenting with multiple resistance markers, including RAS/RAF mutations. Within a CRC subset, 13/29 (45%) achieved SD or PR at 3 cycles of treatment and 5/29 (17%) achieved a PR, with highly durable responses and disease control. Conclusions: Results to date demonstrate that MM-151 has an acceptable tolerability profile. Preliminary indications of objective clinical activity across both the EGFR-refractory and na¨ıve populations suggest potential for broad effect. Biomarker profiling also suggests that MM-151 may overcome mechanisms of resistance. Further clinical evaluation is underway. Clinical trial information: NCT01520389.

Poster Discussion Session; Displayed in Poster Session (Board #219), Sun, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sun, 11:30 AM-12:45 PM

Abexinostat (ABX) as a means to reverse pazopanib (PAZ) resistance: A phase 1 study in advanced solid tumor malignancies. First Author: Rahul Raj Aggarwal, University of California, San Francisco, San Francisco, CA Background: PAZ is a tyrosine kinase inhibitor of VEGFR approved for use in renal cell carcinoma (RCC). ABX is a potent pan-HDAC inhibitor (HDACi). Pre-clinical models suggest that HDACi-mediated epigenetic modulation of VEGF expression prevents PAZ resistance and potentiates efficacy. We therefore designed a Phase I clinical trial combining ABX with PAZ in pts with advanced solid tumors with an expansion cohort in RCC. Methods: The primary endpoint was the maximal tolerated dose (MTD) of PAZ plus ABX. Secondary endpoints included pharmacokinetics (PK) and efficacy. PAZ was dosed days 1-28 and ABX days 1-5, 8-12, and 15-19 of 28-day cycle (schedule A) with at a starting dose of 400 mg/day and 45 mg/m2orally twice daily respectively. An alternate ABX dosing schedule days 1-4, 8-11, and 1518 was investigated (schedule B) due to toxicity of Schedule A. Results: 52 patients (pts) (RCC; N = 23) with advanced solid tumors were enrolled (N = 22 schedule A; N = 30 schedule B). There were six dose-limiting toxicities including fatigue (N = 2), thrombocytopenia (N = 2), and elevated AST/ALT (N = 2). The most common grade $ 3 related adverse events observed were fatigue (13%), thrombocytopenia (12%), and diarrhea (10%). The MTD was PAZ 800 mg/day + ABX 45 mg/m2 BID on schedule B. 8 evaluable pts (19%) (N = 6 RCC; 2 thyroid) achieved partial tumor response (PR), with median duration of response of 9.2 months (1-33.2+). 7/9 (78%) of pts with prior disease progression on PAZ monotherapy had reduction in tumor burden on study. 15 out of 48 evaluable pts (31%) experienced stable disease or better for $ 6 months, and two previously PAZ-refractory pts with PRs remain on study for . 20 and 37 mos respectively. PK analyses did not reveal drug-drug interaction. Analyses of histone acetylation, circulating tumor DNA, and plasma metabolomic profile are ongoing. Conclusions: The combination of PAZ + ABX was well tolerated and durable tumor control ( . 3 yrs) was observed in RCC and thyroid cancer. Tumor regressions observed in majority of PAZ-refractory tumors preliminarily support the potential of ABX to reverse therapeutic resistance. A randomized phase 2 study with cross-over design is planned to further evaluate the combination of PAZ + HDACi. Clinical trial information: NCT01543763.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics 2520

Poster Discussion Session; Displayed in Poster Session (Board #220), Sun, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sun, 11:30 AM-12:45 PM

POSEIDON trial phase 1b results: Safety and preliminary efficacy of the isoform selective PI3K inhibitor taselisib (GDC-0032) combined with tamoxifen in hormone receptor (HR) positive, HER2-negative metastatic breast cancer (MBC) patients (pts) - including response monitoring by plasma circulating tumor (ct) DNA. First Author: Richard D. Baird, Cambridge Cancer Centre, Cambridge, United Kingdom Background: Endocrine therapy resistance remains a major challenge in the treatment of HR-positive breast cancer. The strategy of combining PI3KAKT-mTOR pathway inhibitors with endocrine therapy can improve clinical outcomes (eg. BOLERO-2, BELLE-2) but at the cost of increased toxicity. Taselisib is a potent, selective, beta-isoform sparing inhibitor of PI3K designed to have a better therapeutic index than first-generation pan-PI3K inhibitors. Methods: Pts with prior exposure to endocrine therapy were enrolled using a 3+3 design, testing taselisib in combination with tamoxifen 20mg qd. A (21 day on / 7 day off) intermittent schedule was also compared with daily continuous dosing (28/28). The primary objective was to determine the recommended phase II dose and schedule. Other objectives included: pharmacokinetics (PK), pharmacodynamics (PD) and serial monitoring of plasma ctDNA using digital PCR / tagged amplicon deepsequencing. Results: 30 pts were enrolled in 3 cohorts: taselisib 2mg (21/7), 4mg (21/7) and 4mg (28/28). The most common drug toxicities were diarrhea, nausea and anemia. No dose-limiting toxicities were seen. Three pts had reversible, late, grade $ 3 side effects (two colitis; one pneumonitis). Steady-state PK results: 4mg taselisib mean Cmax 66.6ng/mL, AUC 1016ng*hr/mL; mean endoxifen 7.63ng/mL; consistent with previous reports and low chance of drug-drug interaction. Three pts had a RECIST partial response; 15 pts stable disease; 4 pts not evaluated yet. The radiologic treatment responses seen in some pts were preceded by early changes in plasma PIK3CA ctDNA levels (eg. one pt with a response at 8 weeks had a .90% decrease in PIK3CAH1047R titres detected at 4 weeks). Conclusions: Taselisib in combination with tamoxifen is generally well tolerated. Preliminary evidence of anti-tumor activity was seen, in some patients preceded by a fall in plasma PIK3CA ctDNA levels. The recommended phase II dose of taselisib in combination with tamoxifen is 4mg on a daily continuous schedule. Clinical trial information: NCT02285179.

2522

Poster Session (Board #222), Sun, 8:00 AM-11:30 AM

Phase I dose-escalation trial of Sym004, a mixture of two anti-EGFR antibodies, in Japanese patients with advanced solid tumors. First Author: Ken Kato, Gastrointestinal Medical Oncology Division, National Cancer Center Hospital, Tokyo, Japan Background: Sym004 is a 1:1 mixture of two chimeric IgG1 antibodies targeting non-overlapping epitopes of the epidermal growth factor receptor (EGFR). Binding of both antibodies to EGFR leads to its internalization and degradation. The resulting down-regulation of EGFR results in antitumor activity that is superior to that of cetuximab and panitumumab in preclinical models; Sym004 can also overcome acquired resistance to cetuximab in vivo. We have evaluated the safety of Sym004 in Japanese patients (pts) with advanced solid tumors. Methods: Japanese adults aged $ 20 years with refractory or recurrent late-stage solid tumors, ECOG status of 0 or 1, and life expectancy of at least 3 months were eligible for the dose-escalation part of the trial (Part A). Sym004 doses tested were 6 mg/kg weekly (n = ’3+3’), 9 mg/kg loading dose followed by 6 mg/kg weekly (n = 6), 12 mg/kg weekly (n = 6), and 18 mg/kg q2w (n = 6). An expansion cohort of pts with esophageal squamous cell carcinoma (ESCC) (Part B) was enrolled at the Sym004 dose recommended based on part A. The primary objective was the safety and tolerability of Sym004 in Japanese pts. Secondary objectives included antitumor activity and pharmacokinetics. Results: 51 pts were treated (male/female: 39/12; median age 62 [40-73] years; ECOG performance status 0/1: 32/19). No dose-limiting toxicities were observed in Part A. A Sym004 dose of 12 mg/kg weekly was selected for use in Part B (n = 30). Pts in Part B received a median of 7 weeks of Sym004 therapy. All Part B pts had treatment-related adverse events (AEs); 18 had grade $ 3 AEs and 5 had serious AEs (cardiac arrest, lung infection, interstitial lung disease, toxic skin eruption, blood creatinine increase). Two pts had treatment-related AEs resulting in death (cardiac arrest and blood creatinine increase). 5 pts had a best overall response of partial response, 12 stable disease, 12 disease progression (1 pt not evaluable). The objective response rate was 16.7% (95% CI: 5.6–34.7%). Conclusions: Sym004 therapy was well tolerated with no DLTs at any dose studied. Evidence of antitumor activity, with objective responses, was observed in the expansion cohort of pts with advanced ESCC. Clinical trial information: NCT01955473.

2521

119s

Poster Session (Board #221), Sun, 8:00 AM-11:30 AM

A first-in-human phase 1 study of GC1118, a novel monoclonal antibody inhibiting epidermal growth factor receptor (EGFR), in patients with advanced solid tumors. First Author: Keun Wook Lee, Seoul National University Bundang Hosp, Seoul National University College of Medicine, SeongnamSi, Korea South Background: GC1118 is a novel anti-EGFR monoclonal antibody and has unique binding epitopes and affinity. We conduct a first-in-human phase 1 study to investigate the maximum tolerated dose (MTD), recommended phase 2 dose (RP2D), safety, pharmacokinetics, and preliminary efficacy of GC118 in patients with refractory solid tumors (NCT02352571). Methods: GC1118 was intravenously administrated over 2 hours on days 1, 8, 15 and 22 followed by additional 2 week rest in the first cycle, during which dose limiting toxicities (DLTs) were evaluated. If there was no tumor progression, GC1118 was administrated weekly thereafter. Dose escalation was guided by a Bayesian logistic regression model with overdose control. Results: Twenty-four patients, 17 with colorectal cancer (CRC), were treated with GC1118 at five dose levels: 0.3 mg/kg (n = 4), 1.0 mg/kg (n = 4), 3.0 mg/kg (n = 4), 4.0 mg/kg (n = 6) and 5.0 mg/kg (n = 6). In the 5.0 mg/kg cohort, two patients experienced DLTs, which were grade 3 skin toxicities. No DLTs were observed in the other cohorts. The MTD was determined as 5.0 mg/kg. GC1118-related adverse events (AEs; $ 10%) included skin toxicities [pruritus (58%), dermatitis acneiform (50%), dry skin (42%), paronychia (29%) and maculo-papular rash (25%)] and stomatitis (29%). Diarrhea developed in 2 patients ( # grade 2). AEs of grade $ 3 included skin (n = 2) and hepatic toxicities (n = 1). Pharmacokinetic study showed typical target-mediated drug disposition profiles via ligand-receptor coupling. Considering toxicity and pharmacokinetic data, 4.0 mg/kg was determined as the RP2D. Partial response (PR) was shown in 3 patients (all with CRC) and stable disease in 12. Median progression-free survival was 13.3 weeks (95% confidence interval: 5.1-21.1). Expansion cohorts were opened, and updated results will be presented. Conclusions: We determined the MTD as 5 mg/kg and the RP2D as 4 mg/kg. DLTs were grade 3 skin toxicities. GC1118 was generally well tolerated. Skin toxicities were the most common AEs and diarrhea was uncommon. GC1118 showed promising preliminary anti-tumor activity, especially in wild-type KRAS CRC with 3 PRs (18%) among 17 patients. Clinical trial information: NCT02352571.

2523

Poster Session (Board #223), Sun, 8:00 AM-11:30 AM

Phase 1b study of aldoxorubicin + gemcitabine in metastatic solid tumors. First Author: Kamalesh Kumar Sankhala, Sarcoma Oncology Center, Santa Monica, CA Background: Aldoxorubicin (A) is a novel prodrug of doxorubicin (D) that binds covalently to albumin, accumulates in tumors and releases D under acidic conditions. It has demonstrated significant efficacy against soft tissue sarcomas in a phase 2b study versus D. Gemcitabine (G), when combined with various chemotherapies (albumin/paclitaxel nanoparticles, liposomal D) has demonstrated efficacy in patients with pancreatic, ovarian and lung cancers. It was hypothesized that A+G might possess synergistic activity against solid tumors. This study was designed to identifiy the major toxicities of this combination and establish its MTD. Methods: Adults who had relapsed or not responded to prior chemotherapies were eligible. Initial cohorts were (i) A 170 mg/m2+ G 900 mg/m2 ); (ii) A 250 mg/m2 + G 900 mg/m2; and (iii) A 200 mg/m2 + G 750 mg/m2. A was administered on Day 1 and G on Days 1 and 8 of each 21 day cycle. Safety was monitored continuously, cardiac function was assessed every 2 months by echocardiogram, and tumor response every 6 weeks by CT scan. The MTD was defined as the dose level where , 3 of 6 subjects experienced a dose limiting toxicity. Results: 22 subjects have been entered with a median age of 60 years (2071) with 5 males and 17 females entered. Tumor types included 10 sarcomas (soft tissue and skeletal) and 4 ovarian cancers. 2 PRs (uterine leiomyosarcoma and ovarian cancer), 11 SDs and 2 PDs as best responses were documented. Major grades 3 and 4 adverse events included neutropenia in 7 subjects, thrombocytopenia in 13 subjects and anemia in 8 subjects. Febrile neutropenia occurred in 2 subjects and pneumoinia in 1 subject. 11 SAEs have been observed that include Klebsiella infection, sepsis, fever, increased anemia, small bowel obstruction, stomatitis, pneumonia, pleural effusion, thrombocytopenia, febrile neutropenia and UTI. No subjects exhibited an LVEF , 50% of normal at any evaluation; no treatment-related deaths occurred. The study is completing dosing of A 200 mg/m2 + G 500 mg/m2, considered the phase 2 doses. Conclusions: A + G can be administered safely at a recommended dose of 200 mg/m2 of A on Day 1 and 500 mg/m2 of G on Days 1 and 8 of a 21 day cycle. Thrombocytopenia is the major grade 3 or 4 AE that limits dosing. Clinical trial information: NCT02235688.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

120s 2524

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics Poster Session (Board #224), Sun, 8:00 AM-11:30 AM

Phase 1 multicenter, open-label study to establish the maximum tolerated dose (MTD) of two administration schedules of E7389 (eribulin) liposomal formulation in patients (pts) with solid tumors. First Author: Ishtiaq Husain Zubairi, Beatson West of Scotland Cancer Centre, Glasgow, United Kingdom Background: Eribulin, a microtubule dynamics inhibitor, is approved in the US for the treatment of pts with metastatic breast cancer who have previously received $ 2 chemotherapeutic regimens for metastatic disease including an anthracycline and a taxane in the adjuvant/metastatic setting. Preclinical evidence suggested improved efficacy and lower toxicity with a liposomal formulation of eribulin (eribulin-LF). The primary objective of this study was to determine the MTD and dosing schedule for eribulin-LF. Secondary objectives included determining safety, pharmacokinetics (PK), and efficacy. Methods: Using a conventional 3 + 3 study design, eligible pts received escalating doses (1.0 to 3.5 mg/m2) of eribulin-LF (60-min, i.v. infusion) on day (D) 1 of a 21-day cycle (schedule [S] 1) or on D1 and D15 of a 28-day cycle (S2). An expansion phase confirmed the safety of eribulin-LF in pts with endometrial, ovarian, or HER2-negative breast cancer. Results: Of 58 pts (S1: n = 20; S2: n = 38): 65% were female, median age was 62 yrs (range 33–75), 65% had an ECOG PS of 1, and 10 pts had metastatic breast cancer. Dose-limiting toxicities were: S1, grade (G) 4 hypophosphatemia (n = 1), G4 increased ALT/AST (n = 1), both at eribulin-LF 1.5 mg/m2; S2, G4 febrile neutropenia (n = 1; 1.5 mg/m2); G3 increased ALT/AST (n = 1), and G4 neutropenia (n = 1), both at 2 mg/m2. MTD was 1.4 mg/m2 in S1 and 1.5 mg/m2in S2; the latter dose and schedule were used in the expansion phase (n = 23). Treatment-emergent adverse events (TEAEs) included peripheral neuropathy (S1: 35%, S2: 8%) and neutropenia (S1: 20%, S2: 34%). TEAEs of G3 or higher occurred in 55% (S1) and 39% (S2). EribulinLF exhibited single-phase PK, and was quantifiable ~18 days postdose. Exposure was dose-dependent and t1/2 was ~35 hrs. Promising clinical activity was observed in pts with breast cancer (5/10 partial responses, 3/10 stable disease). Conclusions: In this phase 1 study, eribulin-LF was well tolerated, with promising activity in patients with breast cancer. Eisai Inc. Clinical trial information: NCT01945710.

2526

Poster Session (Board #226), Sun, 8:00 AM-11:30 AM

A phase 1 study of CRLX301, a novel nanoparticle-drug conjugate (NDC) containing docetaxel (DOC), in patients with refractory solid tumors. First Author: Ben Markman, Monash Cancer Center, Melbourne, Australia Background: CRLX301 is an investigational NDC with DOC covalently conjugated to a cyclodextrin-polyethylene glycol co-polymer designed to enhance delivery to tumor tissue. Preclinical studies of CRLX301 showed enhanced drug accumulation in tumor and greater tumor growth suppression in murine models compared to DOC. Methods: Patients (pt) with advanced solid tumors were treated with escalating doses of CRLX301 IV once every 21 days (Q3W). The first two cohorts were single pt cohorts and then adjusted to 3+3 design. Primary objective: MTD and/or RP2D. Secondary objectives: safety, anti-tumor activity and PK. Results: 20 pts were enrolled. All were evaluable for DLT. Prior lines of chemotherapy: median 3 (range 0 to 6); 8 pts had prior taxane therapy. Tumor types: cholangiocarcinoma (5), head and neck (3), lung (2) and others (10). All drug-related AEs were toxicities typically associated with DOC. The most common drug-related AEs (all grades) were fatigue (70%), infusion reaction (40%), nausea (35%), diarrhea (30%), neutropenia (30%), and peripheral neuropathy (20%). Most AEs were # grade 2. Two DLTs, reversible grade 3 transaminases with grade 2 bilirubin ( , 3 d) and grade 4 febrile neutropenia ( , 3 d), were observed at 90 mg/m2. One gr3 and one gr4 infusion reactions were observed at 90 mg/ m2; both resolved within 24 hours. Certain common AEs related to DOC, such as fluid retention, were not observed. MTD with Q3W dosing is 75 mg/ m2. A pt with B-RAF mutant unknown primary and resistant to a prior B-Raf inhibitor had tumor shrinkage per CT and partial metabolic response per FDG-PET. Stable disease lasting $ 3 cycles was observed in 5 pts (max 15 cycles). Plasma PK compared to commercial DOC at equivalent dose of 75 mg/m2: increased Cmax (5 – 8 fold) and increased AUC (60 fold) of total DOC and 25 fold less peak concentrations but more prolonged exposures of released DOC. Conclusions: The data suggest CRLX301 has a predictable and acceptable safety profile with preliminary signs of anti-tumor activity and distinct PK compared to DOC. MTD/RP2D were established as 75 mg/m2 for Q3W dosing. Weekly dosing and expansion cohorts in selected tumor types are planned. Clinical trial information: NCT02380677.

2525

Poster Session (Board #225), Sun, 8:00 AM-11:30 AM

Phase 1/2a trial of intravenous BAL101553, a novel tumor checkpoint controller (TCC), in advanced solid tumors. First Author: Juanita Lopez, The Royal Marsden/Institute of Cancer Research, Sutton, United Kingdom Background: BAL101553 (prodrug of BAL27862) is a small molecule TCC that binds microtubules and promotes tumor cell death through activation of the mitotic checkpoint. Antiproliferative effects appear driven by AUC; vascular disruption by Cmax . In a NSCLC xenograft mouse model, dosedependent vascular disruption with reduction of tumor perfusion was observed. Methods: Patients (pts) with advanced solid tumors who failed standard therapy, received 2-h intravenous (IV) BAL101553 on days 1, 8 and 15 of each 28-day cycle. The Phase 1 (P1) part of the study used an accelerated 3+3 dose-escalation design to determine the maximum tolerated dose (MTD). In Phase 2a (P2a), pts were randomized to assess whether MTD or sub-MTD doses provide better anti-tumor activity. Adverse events were assessed by CTCAEv4 grade (G); tumor response by RECIST 1.1 every 2 cycles. Assessments included PK, tumor biopsies, functional imaging, and analysis of circulating tumor, endothelial and endothelial progenitor cells (CTCs, CECs, CEPs). Results: 73 pts (39M/34F; median age 58 y; 29-80 y) received doses of 15-80 mg/m2 (N = 24, P1) and 30-60 mg/m2 (N = 49, P2a). DLTs in P1 at 60-80 mg/m2 were G2-3 gait disturbance (reversible) with G1-2 peripheral neuropathy. In P2a, myocardial ischemia (reversible ECG changes, troponin elevations) was observed in 3 pts (45-60 mg/m2), resulting in a recommended P2 dose (RP2D) of 30 mg/m2. Of 59 pts evaluable for efficacy to date, 39 received BAL101553 at 30 mg/m2 (as starting dose or adjustment). Of these, 1 pt (ampullary cancer) had a partial response for . 2 y; 10 pts had stable disease (3 pts . 4 cycles). Tumor biopsies showed effects on tumor cell proliferation and vascularization at $ 60 mg/m2, with time profiles of transient blood pressure elevations linking the vascular effect to Cmax . The CEC/CEP pattern supported vascular disruption at 60 mg/m2 and potentially increased vascular leakage at 30 mg/ m2. Conclusions: BAL101553 as 2-h IV infusion showed dose-dependent vascular effects that were dose-limiting and related to Cmax . The RP2D of 30 mg/m2is well tolerated, showed signals of antitumor activity and potentially increases vascular leakage, which may be useful for combination therapy. (NCT01397929). Clinical trial information: NCT01397929.

2527

Poster Session (Board #227), Sun, 8:00 AM-11:30 AM

Overall survival of participants compared to non-participants in a randomized-controlled trial (SELECT BC): A prospective cohort study. First Author: Kazutaka Narui, Breast and Thyroid Surgery, Yokohama City University Medical Center, Yokohama, Japan Background: Whether participation in a randomized-controlled trial provides benefits to patients is controversial. Some studies have showed that participants have longer prognosis than non-participants. However, in such studies, underlying bias always exists, such as patient condition, patient selection by physicians, and differences in the care provided. Therefore, to minimize such bias, we enrolled non-participants at the time of patient recruitment for a randomized-controlled trial, and prospectively compared the prognoses of participants and non-participants. Methods: This was a prospective cohort study accompanying the “SELection of Effective ChemoTherapy for Breast Cancer” (SELECT BC) study, which compared S-1 and taxane as first-line treatment for metastatic breast cancer and reported the non-inferiority of S-1 with respect to overall survival. Prognoses of trial participants and non-participants were compared. The endpoint was overall survival [Lancet Oncol 17:90-98, 2016]. All patients met the eligibility criteria set for SELECT BC. Results: From August 2009, when this study started, to July 2010, 228 patients were enrolled in SELECT BC, while 65 patients were registered as non-participants in this study. The median followup period was 32.5 months. The median overall survival of participants (36.8 months) was longer than that of non-participants (25.2 months) (hazard ratio [HR]: 1.48; 95% confidential interval [CI]: 1.06-2.07; p = 0.023). Standard chemotherapy was administered to 48 non-participants (74%); 34 (52%) of them received standard taxane regimen or S-1, which was to be administered if they participated in the trial. However, the median overall survival of these 34 patients (22.0 months) was shorter than that of the trial participants (36.8 months) (HR: 2.03; 95% CI: 1.35-3.06; p = 0.0007). Conclusions: Randomized-controlled trial participants had longer overall survival than non-participants. Several factors, such as mental status, relationship between physicians and patients, or difficulties in receiving the recommended treatment could have influenced the prognosis. Clinical trial information: 000020645.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics 2528

Poster Session (Board #228), Sun, 8:00 AM-11:30 AM

Genomic mutation profiling (GMP) and clinical outcome in patients (pts) treated with ribociclib (CDK4/6 inhibitor) in the Signature program. First Author: Julio Antonio Peguero, Oncology Consultants PA, Department of Research, Houston, TX Background: The Signature program comprises a series of 8 tissue-agnostic, mutation-specific, signal-seeking protocols without predetermined study sites, thereby bringing the protocol to the pt. Pts included in the ribociclib (LEE011) protocol had CDK4/6 pathway aberrations, identified via standard-of-care physician-directed profiling. Separate ribociclib phase 3 studies in breast cancer are ongoing. Methods: Pts with measurable advanced malignancies and no established standard therapy options were eligible. The primary objective was assessment of clinical benefit (CB; CR + PR + SD) at wk 16 by local investigator. Ribociclib was given orally (600 mg QD; 3 wk on, 1 wk off). Statistical design adaptively clustered pts into cohorts for independent analysis of early futility or efficacy. GMP of pt tumors was performed centrally for post hoc eligibility confirmation. Results: As of 28 Aug 2015, 106 pts were enrolled across 31 tumor types. Median age was 63 years (range, 19-86 years), 53% were women, and median number of prior antineoplastic therapies was 3 (range, 0-19). Baseline actionable alterations included 73 p16 (CDKN2A) mutations (mut; 68.9%); 10 cyclin D1 (CCND1) amplifications (amp; 9.4%); 9 CDK6 amp (8.5%); 7 CDK4 amp (6.6%); 5 each of CDK4 mut and cyclin D3 (CCND3) amp (4.7%); and 2 CDK6 mut (1.9%). 5 pts had . 1 eligible aberration. Major tumor types (in $ 4 pts) included sarcoma (13 [12%]) and bladder, triple-negative breast, and HNSCC (7 each [7%]). Rare tumors included mesothelioma (5 [5%]) and thyroid, penile, and adrenal (1 each [1%]). 12 analyzable cohorts were formed. Neutropenia (17% grade 3/4; 29% all grades) was the most frequent grade 3/4 drug-related AE seen in $ 10% of pts. At wk 16, CB was seen in 19 pts; 11 (58%) had p16 mut/loss. Preliminary antitumor activity was observed in 4 pts: 1 pt with urothelial cancer (CCND1 amp), 1 pt with sarcoma (CDK4 amp), 1 pt with unknown primary tumor (CDK6 amp), and 1 pt with ovarian cancer (CDK6 mut; reported after data cutoff). Conclusions: Ribociclib showed clinical activity in some pts. Correlation of CB and GMP is ongoing and will help to determine the significance of these mutations and potential for combination with other agents. Clinical trial information: NCT02187783.

2530

Poster Session (Board #230), Sun, 8:00 AM-11:30 AM

2529

121s

Poster Session (Board #229), Sun, 8:00 AM-11:30 AM

Too sick to enroll? Comorbidities limiting recruitment in early phase trials, review of over 1,100 clinical trials. First Author: Narjust Duma, Rutgers New Jersey Medical School, Newark, NJ Background: Early phase clinical trials (EPCT) evaluate the potential benefits of new regimens and help guide the care of oncology patients (pts).The inclusion/exclusion criteria in EPCT are usually rigorous and may exclude many pts commonly seen in clinical practice. Our objective was to study the most common comorbidities excluded in EPCT Methods: ClinicalTrials.gov was queried in December 1st of 2015. We reviewed the eligibility criteria of 1114 phase I/II trials including: age, organ function and comorbidities. Logistic regressions were completed and exclusion was studied as a binary variable based on the comorbidity Results: We reviewed 1114 trials, cancers included: 41% breast, 31% colorectal, 20% lung and 8% of trials included all solid tumors. 23% of the trials excluded pts . 70 years and 12% pts . 65 years. With regards to organ function, 423 (38%) trials required an albumin level . 3.0 g/dL, 355 (32%) creatinine levels , 1.5 mg/dL, and 333 (30%) liver function tests , 2.5 of the upper limit. Hematologic function was a significant exclusion factor: 434 (39%) trials required hemoglobin . 9 g/dL, 161 (14.5%) . 10 g/dL and 492 (44%) excluded pts with , 100,000 platelets. In terms of comorbidities, 570 (51%) excluded pts with myocardial infarction/angina within 12 months, 362 (33%) with atrial fibrillation, and 239 (22%) excluded pts on any anticoagulation therapy. Diabetes was not a significant factor of exclusion, only 10% of the trials excluded pts with “uncontrolled diabetes.” On univariable analysis, investigator initiated trials were less likely to exclude pts with cardiovascular disease (OR: 0.43, 95%CI: 0.25-0.74, p , 0.002) and mutation driven drugs were more likely to exclude pts with below normal hematologic values (OR: 2.39, 95%CI: 1.41-4.03, p , 0.001). No multivariable associations were significant. Conclusions: 23% of the trials excluded pts . 70 years or older, 33% pts with atrial fibrillation and 22% pts on any anticoagulation therapy, highlighting that a great percentage of EPCT are not designed to treat our aging oncology population. Investigators should review whether sufficient justification exists for every exclusion criterion before their incorporation in future trial protocols.

2531

Poster Session (Board #231), Sun, 8:00 AM-11:30 AM

Exposure-response methods and dose approval of new oncology drugs by FDA from 2005 to 2015. First Author: Kaitlyn Minchella, AstraZeneca, Waltham, MA

IND safety reporting: final results and best practices, from the Clinical Trials Transformation Initiative IND Safety Advancement Project. First Author: Raymond P. Perez, University of Kansas, Fairway, KS

Background: Exposure-response (E-R) analysis is critical for dose justification throughout drug development. The relationship between dose, exposure, and response is crucial for optimizing a drug’s safety and efficacy. We analyzed the E-R methods, the approved dose, and any dose or indication changes of all New Molecular Entities (NMEs) in oncology that were approved by FDA from 2005 to 2015. Methods: E-R analyses information was obtained from the “Clinical Pharmacology & Biopharmaceutics Reviews” of each NME (http://www.accessdata.fda.gov/scripts/cder/drugsatfda/index. cfm). E-R methods from both the FDA and sponsor were included in the analysis. If the same exposure-efficacy (E-E) or exposure-safety (E-S) method was used for multiple response endpoints, it was counted once. Different E-E or E-S methods used for the same NME were all counted. Dosage information was collected from the approved and revised labels. Results: From 2005 to 2015, FDA approved 59 NMEs in oncology. There were 44 small molecule drugs: 17 tyrosine kinase inhibitors (TKIs), 17 nonTKI molecular targeted agents, 9 chemotherapeutic agents, and 1 radioactive agent. Fifteen were large molecule drugs: 11 monoclonal antibodies, 2 anti-body drug conjugates, 1 enzyme, and 1 fusion protein. Methods used for E-E analysis and E-S analysis, respectively, were 25 and 32 logistic plot and regression analyses, 26 and 7 Kaplan Meier (KM) or Cox proportional hazard (Cox PH) analyses, 13 and 18 exploratory plots, 3 and 6 summary statistics, and 4 and 4 other analyses. Logistic regression and Cox PH analyses accounted for the majority of significant E-R relationships (13 and 2 for E-E, and 22 and 2 for E-S, respectively). Twenty-four drugs had a dosage and/or indication change on their label post-approval; in total, there were 13 new indications, 11 modified indications, 1 dose reduction, 1 additional dose option, and 1 dose schedule change. Conclusions: E-R analysis has played an important role in supporting dose labeling for oncology drugs. More opportunities still exist to advance the E-R methodology regarding safety and efficacy optimization and dose justification.

Background: As clarified in 2010, the FDA IND Safety Reporting Regulation requires sponsors to review safety data across all studies in an IND, analyze data in aggregate, assess causality, and issue reports only when it is reasonably likely that serious, unexpected event(s) were caused by the drug. A 2016 FDA analysis demonstrates that the number of expedited reports per IND per year is increasing and most reports (86%) are uninformative. We sought to understand 1) sponsor challenges to full implementation of the Regulation, 2) sponsor motivation to change safety reporting practice, 3) investigator site safety report management, and 4) best practices for communicating safety reports. Methods: Surveys of sponsors (n = 28) and investigator sites (n = 201) were conducted, followed by interviews. Best practices were shared at a July 2015 Expert Meeting. Results: Surveys of sponsors (n = 28) and investigator sites (n = 201) were conducted, followed by interviews. Best practices were shared at a July 2015 Expert Meeting. Conclusions: Sponsor best practices can improve IND safety reporting. Dissemination of these findings, sponsor/site education, and regulatory clarity offer the potential to further enhance safety in oncology trials.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

122s 2532

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics Poster Session (Board #232), Sun, 8:00 AM-11:30 AM

2533

Poster Session (Board #233), Sun, 8:00 AM-11:30 AM

EVESOR, the first model-based multi-parameter academic phase 1 trial meant to optimize the benefit/toxicity ratio of everolimus (EVE) and sorafenib (SOR) combination: Initial results. First Author: Benoit You, Institut de cancerologie ´ des Hospices Civils de Lyon (IC-HCL), CITOHL, Lyon, France

Prognostic factors for 90-day mortality in patients with solid malignancies treated on Phase 1 trials sponsored by the Cancer Therapy Evaluation Program (CTEP) of the National Cancer Institute. First Author: Kit Man Wong, University of Colorado, Denver, CO

Background: “Drug development of targeted agent associations is a challenge”. Previous traditional industry phase I trials failed to define the tolerable dose/dosing schedules of EVE & SOR combo. The optimal dosing schedule, maximizing the benefit/toxicity ratio, may be identified using modeling & simulation (M&S) of data derived from adequately designed multi-parameter trial. Methods: EVESOR was composed of 4 arms with different dosing schedules/doses of EVE & SOR (Continuous A & B arms EVE 5mg + SOR 200mg bid after reverse run-in periods; Intermittent arms: C, EVE qd and SOR bid for 7 days alternating every weeks; D, EVE continuous qd and SORA bid 3days-on 4 days-off). Dose-escalations planned in intermittent arms. Multi-parameter assessments: pharmacokinetics (PK); serial pharmacodynamics (PD) on signaling pathways (ERK, AKT, S6K) in PBMCs; kinetics of circulating tumor DNA & angiogenesis markers; repeated DCE-ultrasounds. The optimal doses/dosing schedules of EVE & SOR will be searched using M&S and then tested. Results: 26 solid tumor patients were enrolled: cholangiocarcinoma (n = 8); CRC (n = 5); breast (n = 3); pancreas (n = 3); endometrial (n = 2); others (n = 5). Grade 3-4 toxicity incidences were lower than expected: fatigue 23%; hypophosphatemia 11%; liver transaminase increase 11%. AEs were more frequent in continuous arms (A & B, 60%) than in intermittent arms (C & D, 40%). Two grade 3 DLTs occurred on arms C: post-biopsy liver abscess; toxidermia. Overall response rates (ORR): PR 11%; SD 77%. ORR was better in intermittent arms (C & D, SD 80%; PR 20%) than in continuous arms (A & B, PD 20%; SD 80%). Of note, cholangiocarcinoma: SD 62% with long responses . 10 months for 50% patients. No PK interaction between EVE & SOR was identified. PD interactions between SOR & EVE schedules were found on angiogenesis & RAS/PI3K signaling pathways. Conclusions: The novel design of multiparameter EVESOR trial aims at optimizing the doses/dosing schedules of the 2 drug combination, ultimately with M&S works. Initial outcomes suggest PD interactions, and better tolerance & efficacy of intermittent schedules warranting further investigations. Clinical trial information: NCT01932177.

Background: The appropriate selection of cancer patients with reasonable life expectancy is critical in Phase 1 trials. The ability to predict patient outcomes prior to enrollment, particularly their 90-day mortality (90DM), would improve selection, avoid harm in those unlikely to benefit, and maximize the meaningful data obtained. There is currently no standard prognostication strategy. We analyzed the prognostic factors of patients on Phase 1 trials sponsored by CTEP. Methods: 5,477 patients with solid tumors enrolled in CTEP Phase 1 trials from 2000 to 2015 were included. Patient, disease and lab characteristics within 30 days prior to first dose, study treatments and outcomes were extracted from a prospective database. Adjusted hazard ratios (aHR) for 90DM were analyzed by multivariate Cox regression. Results: The study population had a median age of 58 (18-90) years and 49% males. Most common tumor types were: 35% gastrointestinal, 11% breast, 10% genitourinary (GU); 94% of patients had ECOG 0-1. Most subjects received combination therapy (63%). The majority of trials tested a targeted agent (76%). 90DM was 16%; median duration on therapy (DoT) was 56 (0-1,358) days; response rate was 9%. 90DM was independently associated with worse ECOG (2 vs 0: aHR 3.8, 95% CI 2.5-5.7), high ALP ( . 2.5xULN vs , ULN: aHR 2.2, 95% CI 1.5-3.1), high LDH ( . 2.5xULN vs , ULN: aHR 2.3, 95% CI 1.7-3.1), low albumin ( , 3.0 vs . 3.5 g/dL: aHR 2.7, 95% CI 1.9-3.7), high bilirubin ( $ 1.3 vs , 1.3 mg/dL: aHR 1.6, 95% CI 1.2-2.2), and high white blood cells ( . 11 vs # 11 x109/L: aHR 1.7, 95% CI 1.3-2.2). Those on combination regimens were more likely to live in the first 90 days (vs monotherapy: aHR 0.75, 95% CI 0.59-0.95). Therapy type (cytotoxic, targeted, immunotherapy) and tumor site were not determinants of 90DM. Similar variables predicted DoT; additional factors associated with longer DoT were GU cancers and regimens containing noncytotoxic agents. Conclusions: This is the largest prognostic analysis of 90DM in Phase 1 trial patients. ECOG status, specific labs, and number of agents appear to be important factors. A prognostic tree algorithm will be presented.

2534

2535

Poster Session (Board #234), Sun, 8:00 AM-11:30 AM

The effect of comorbidity on participation in early phase cancer trials. First Author: Ibrahim Sheriff, King’s College London, London, United Kingdom Background: The elderly, ethnic minorities and the socioeconomically deprived are underrepresented in oncology clinical trials, including early phase studies. Comorbid burden may contribute to this, but the impact of additional diagnoses on participation and outcome of early phase cancer trials has not been previously analysed. Methods: We evaluated data collected over an 8 year period from an early phase clinic at a UK cancer centre. Comorbid burden was evaluated using the Charlson Comorbidity Index (CCI). Data on age, ethnicity, socioeconomic status and ECOG performance status were also collected. Univariate and multivariate logistic and Cox regression analyses were performed to quantify the effect of CCI on trial enrolment, and overall survival of patients on trial. We hypothesised that comorbid burden influences these outcomes. Results: 681 patients were referred and included in this analysis. Of these, 230 patients went on trial (33.8%). CCI was significantly associated with increasing age (Pearson’s x2 (10) = 19.02, P = 0.04) but not with socioeconomic or ethnic group. Trial enrolment and overall survival on trial was not affected by age, ethnicity or socioeconomic status. By contrast, patients with any significant comorbidity were less likely to be enrolled on a trial (CCI $ 2, adjusted odds ratio 0.27, p = 0.004, CI 0.11-0.66, compared to CCI = 0). CCI was not linked to overall survival on trial (CCI $ 2, adjusted hazard ratio 0.77, CI 0.32-1.84, compared to CCI = 0). Of the 451 patients who did not enrol on a trial, the main reasons for exclusion were patient choice in 98/451 (21.7%), poor ECOG performance status in 93/451 (20.6%) and ineligibility according to trial inclusion and exclusion criteria in 88/451 (19.5%). Comorbidity was directly cited as the reason for not enrolling in 21/451 patients (4.7%). Of these patients, 13/ 451 (2.9%) were excluded as their comorbidities rendered them ineligible for trial and 8/451 (1.8%) were excluded by clinician decision alone. Conclusions: Comorbidity affects the likelihood of enrolment on an early phase trial, but not overall survival once enrolled. These findings can further inform efforts to widen participation of underrepresented groups in early phase cancer trials.

Poster Session (Board #235), Sun, 8:00 AM-11:30 AM

SHIVA: Randomized phase II trial comparing molecularly targeted therapy based on tumor molecular profiling versus conventional therapy in patients with refractory cancer—PFS ratio from patients who crossed-over. First Author: Christophe Le Tourneau, Institut Curie, Paris, France Background: Several studies used the ratio of progression-free survival (PFS) on genotype-matched treatment to PFS on genotype-unmatched treatment to assess the efficacy of therapy guided by patients’ tumor molecular profiling [von Hoff et al., JCO 2010; Tsimberidou et al., CCR 2012]. We evaluated the PFS ratio from patients included in the SHIVA trial who crossed-over [Le Tourneau et al., Lancet Oncol 2015] (NCT01771458). Methods: The primary endpoint of the SHIVA trial was to compare PFS on molecularly targeted agents (MTAs) based on tumor molecular profiling and treatment at physician’s choice (TPC) in patients with any kind of cancer who had failed standard-of-care therapy. Experimental treatment included only marketed MTAs given outside their indications according to a pre-specified treatment algorithm. Patients were allowed to cross-over at disease progression. Response was evaluated according to RECIST 1.1 at randomization and at cross-over. We evaluated the proportion of patients with a PFS ratio (PFS2 on MTA/PFS1 on TPC) . 1.3 among the patients who crossed-over from the TPC arm to the MTA arm, using each patient as his/her own control. Patients with a censored PFS2 were considered as having progressive disease. Results: Among 741 patients enrolled in the SHIVA trial, 197 were randomized, and 95 crossed-over at disease progression, including 70 patients from TPC arm to MTA arm and 25 patients from MTA arm to TPC arm. Two patients crossed-over in the control arm without disease progression. Among the 68 patients who crossed-over from TPC arm to MTA arm with disease progression, median PFS1 was 2.0 months and median PFS2 was 2.1 months. The PFS ratio exceeded 1.3 in 35% of patients [95%CI: 24 - 46], including 20% of patients with PFS ratio . 2. The PFS ratio was 1-1.3, 0.7-1.0 and , 0.7 in 13%, 25%, and 26% of patients, respectively. Conclusions: We report a PFS ratio . 1.3 in 35% of patients who crossedover from the TPC arm to the MTA arm in the randomized SHIVA trial, suggesting a clinical benefit of the evaluated histology-agnostic approach in this subgroup of patients. Clinical trial information: NCT01771458.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics 2536

Poster Session (Board #236), Sun, 8:00 AM-11:30 AM

2537

123s

Poster Session (Board #237), Sun, 8:00 AM-11:30 AM

Comparison of genotype-specific progression free survival (PFS) outcomes in the control (ctrl) arms of randomized clinical trials (RCTs). First Author: Rossanna C. Pezo, Princess Margaret Cancer Centre, Toronto, ON, Canada

Geographical distribution of industry driven clinical trials performed for leukemia over the last ten years. First Author: Laura Vidal Boixader, INC Research, Barcelona, Spain

Background: PFS is often used as the primary outcome measure in RCTs involving patients (pts) with metastatic cancers. For RCTs evaluating new targeted agents, there is uncertainty about how somatic mutation status affects clinical outcomes under standard therapy. Our aim was to identify differences in genotype-specific PFS outcomes in the comparator arm of advanced solid tumor RCTs. Methods: A MEDLINE search was conducted to identify published RCT reports evaluating systemic therapies (tx) with PFS outcomes for wildtype (WT) versus (vs) mutant (MT) genotypes in the ctrl arm. Genotypespecific PFS data was available for cutaneous melanoma (BRAF), CRC (KRAS) and NSCLC (EGFR). Insufficient data was reported for other solid tumor genotypes, including HER2+ gastric, HPV+ SCC, PIK3CA breast, BRCA1/2 ovarian and KRAS NSCLC. We extracted data including sample size, type of tx, hazard ratios (HR), and genotypespecific median PFS (mPFS) for the ctrl arms. Trials were grouped according to type of tx, targeted (T) vs chemotherapy (C) and line of tx (first line (1L) vs non-1L). The t- test was used to compare mPFS difference and HRs for WT vs MT pts. Results: A total of 60 reports involving 29,631 pts were included in this analysis. For NSCLC, mPFS difference between WT and MT pts was 1.8 months (p = 0.01), in favor of EGFR MT pts receiving 1L C. For non-1L trials of T agents in NSCLC, mPFS difference was 6.7 months in favor of the EGFR MT pts (p = 0.05). In CRC and melanoma, there was no difference in mPFS for WT vs. MT genotypes. Mean difference in HRs between WT vs MT ctrl is listed in the Table below. Conclusions: EGFR MT NSCLC had a superior PFS under standard therapy with 1L C and non-1L T agents. For other groups analyzed, there was no difference in outcome under standard therapy according to genotype. Further reporting of RCTs that analyze outcome by genotype is required to establish standards for genotype-specific outcomes under standard therapy.

Background: The incidence of leukemia (L) is increasing worldwide, with an estimated 350,000 people diagnosed each year. One major motivation for looking abroad for drug clinical development is the lowered cost of running a clinical trial in developing countries, about one-half to one-third that in a developed country. We aim to describe variations on the geographical distribution of industry driven clinical trials performed for L over the last 10 years (2005-2015). Methods: INC clinical trial database was reviewed for L, all types and compared to industry sponsored clinical trials listed on Trialtrove, 2015, Citeline database. The geographic location of the trial was classified based on the following countries: North America (NA), Western Europe (WE), Eastern Europe (EE), Latin America (LA), Middle East and North Africa (MENA) and Asia Pacific (AP). A single site within a region was considered positive for regional trial involvement. Geographic distribution of INC L trials was compared to geographic distribution across all INC oncology clinical trials. Results: INC Research database included 61 trials: 72% included research sites in NA, 48% in WE, 25% in EE and 28% in AP. When compared with CiteLine´s database, including 1391 trials, similar percentages were observed except for a higher distribution in EE for INC (25% vs 10%) maybe reflecting the preferred choice of WE countries by pharmaceutical companies. When compared with INC geographical distribution for non-specific solid tumors, similar percentages to the L geographical distribution from CiteLines´s database were observed suggesting that industry is still going for same countries regardless cancer indication. A considerable gap between developing regions and developed countries is maintained through the last ten years with no significant signs of reduction. Conclusions: There are still important challenges associated with performing studies in cancer in less developed countries. For L, despite the impact of educational programs and networking on the management of this disease in developing countries, more effective measures need to be undertaken in order to offer a real chance for these patients to participate in clinical trials.

Tumor type (# trials) NSCLC (28) CRC (20) Melanoma (12)

2538

Tx line and type (C or T) 1LC Non-1 L C Non-1L T 1L C 1L C

Mean HR difference, WT vs MT (95% CI) 1.4 1.6 4.8 1.0

P value

(1.1-1.6) (0.1-3.0) (1.6-7.9) (0.9-1.1)

0.02 0.30 0.04 0.90

0.9 (0.0-6.7)

0.92

Poster Session (Board #238), Sun, 8:00 AM-11:30 AM

FDA analysis of the characteristics and relevance of serious adverse events that are Grade 1 or 2 in severity. First Author: M. Naomi Horiba, U.S. Food and Drug Administration, Silver Spring, MD Background: In New Drug Approval (NDA) applications, sponsors must provide an integrated summary of the safety of the drug product, including serious adverse events (SAEs), defined as adverse events (AE) resulting in death, life-threatening AEs, inpatient hospitalization or prolongation of hospitalization, persistent or significant incapacity or substantial disruption of the ability to conduct normal life functions, or a congenital anomaly/birth defect. The National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) grades the severity of AEs from 1 (mild) to 5 (death). This analysis attempts to quantify and characterize Grade 1-2 toxicities reported as SAEs. Methods: We reviewed SAEs from registrational trials submitted in nine NDAs for new molecular entities approved by FDA in 2014. Results: A total of 28,467 AEs were reported in 1,699 patients, of which 2,073 (7.3%; range 3.2- 10.9) were reported as SAEs. Of these 2,073 SAEs, 528 (25.5%; range 15.5-37.5) were Grade 1-2. Of 1,489 hospitalizations reported as SAEs, 93 (7.6%) reported no Grade 3-5 toxicities. Forty-four (8.3%) Grade 1 or 2 SAEs were not listed in the CTCAE dictionary, including dosing errors, catheter placement, and hospitalization for observation. A total of 65 (12.3%) Grade 1-2 SAEs were reported concomitantly with Grade 3-5 SAEs (e.g., Grade 2 fever with Grade 4 pneumonia). Of all Grade 1-2 SAEs, 24 (4.5%) were related to medication errors not resulting in toxicity; 57 (10.8%) were infections; 51 (9.6%) were neurologic complications; 44 (8.3%) were pyrexia; 28 (5.3%) were metabolic disturbances and 16 (3%) were abnormal laboratory findings; 12 (2.2%) were related to device/ catheter complications. Fourteen SAEs were erroneously reported as Grade 1-2; i.e., CTCAE only defined the events as Grade 3 to 5 in severity (e.g., septic shock). Conclusions: A significant amount of time is spent by applicants, investigators, cooperative groups, and FDA conducting analyses of both SAEs and AEs by CTCAE grade. In most cases, this is a duplication of efforts. Most reported Grade 1-2 events were, by definition, non-severe. Minor changes in the CTCAE definition and/or SAE definition could serve to streamline clinical trial processes.

2539

Poster Session (Board #239), Sun, 8:00 AM-11:30 AM

Feasibility of molecular profiling based assignment of cancer treatment (MPACT): A randomized NCI precision medicine study. First Author: Alice P. Chen, Early Clinical Trials Development Program, National Cancer Institute at the National Institutes of Health, Bethesda, MD Background: The NCI is conducting a double blind, randomized trial to evaluate whether refractory cancer patients (pts) are more likely to derive clinical benefit if treated with agents targeting a molecular abnormality in their tumors than if treated with non-matched therapeutic agents. The MPACT Trial (NCT01827384) was opened at the Developmental Therapeutics Clinic, NCI to test the feasibility of the design in 60 pts prior to opening as a multicenter study. Methods: Pts with progressive solid tumors after treatment (tx) with $ one line of standard therapy, with adequate organ function, KPS $ 70%, tumor amenable to percutaneous biopsy (bx) were eligible. Bx specimens were evaluated for mutations in one of 3 genetic pathways (DNA repair, PI3K, or RAS/RAF/MEK). If an actionable mutation (aMOI) was found in $ one pathway, pts were assigned tx using a rules-based algorithm (Genemed) that used a 2:1 randomization favoring agents potentially active in the mutational pathway versus the other drugs used for this trial that targeted other pathways. The tx arms are: AZD 1775 + carboplatin; or veliparib + temozolomide (DNA repair); everolimus (PI3K); and trametinib (RAS/RAF/MEK). Response rate (CR+PR) and/or 4-month PFS will be compared between the randomized groups. At first progression, pts molecular profiles are unblinded to clinicians. If a pt was treated with an agent from the unmatched drug set, pt is crossed over to targeted therapy. Results: Sixty pts have been enrolled and screened with NGS using a panel of 20 genes between February 2014 & July 2015. Pt characteristics: 33 pts male, median age 55 years (range 23-76), KPS 80-90%, 1-4 (range) of prior treatments. Most common bx site was liver. Eight pts had sarcomas, 52 had carcinomas. One pt had PE prior to bx. Only 4 bx failed to yield sufficient tumor and nucleic acid for the NGS assay. Two additional pts were successfully rebx. Of 55 pts, 29 (52%) had tx assignment. Average time from bx to assay report has been 6.6 days (range 5 – 12 days). 21 pts started assigned tx. Conclusions: These results demonstrate the feasibility of assigning tx based on real-time molecular profiling in the setting of a randomized study design. MPACT will open in other centers in 2016. Clinical trial information: NCT01827384.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

124s 2540

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics Poster Session (Board #240), Sun, 8:00 AM-11:30 AM

Symptom clusters in patients with advanced cancer in an early-phase clinical trials clinic. First Author: Goldy George, Department of Symptom Research, Department of Investigational Cancer Therapeutics (Phase 1 Program), The University of Texas MD Anderson Cancer Center, Houston, TX Background: Few data exist on patient-reported outcomes, specifically symptoms, in patients with advanced cancer who seek or are in treatment in phase I clinical trials of molecular targeted agents (phase I patients). We explored patient clusters based on symptom severity, and examined associated factors. Methods: Phase I patients completed the validated core MD Anderson Symptom Inventory (with 13 symptom and 6 symptom interference items, each rated on a 0-10 scale) and the Profile of Mood States. ECOG performance status was assessed. Cluster analysis was used. Results: In phase I patients (N = 248, 51% female, 90% ECOG 0-1, and 52% $ 60y), symptoms with highest reported mean severity were fatigue (4.6), drowsiness (3.7), pain (3.6), and lack of sleep (3.0). Highest symptom-related interference was with work (3.5), activity (3.4), and enjoyment of life (3.4). Two natural patient clusters based on reported symptom severity were seen: 38% of patients were in the high symptoms cluster (HSC), and 62% were in the low symptoms cluster (LSC). HSC patients reported greater tension, depression, anger, confusion and lower vigor (p , 0.001), were more likely to have poor PS (ECOG . = 2) (16.5% vs. 4.7%, p , 0.003) and reported higher symptom-related interference (5.1 vs. 1.8, p , 0.001) than LSC patients. Cluster membership did not vary by current enrollment on a phase I clinical trial (p . 0.05) or by the number of cycles completed (6.8 vs. 6.4 cycles, p . 0.05). Patients on a phase I trial reported less dyspnea (1.0 vs. 1.95, p , 0.001) and vomiting (0.53 vs. 1.07, p , 0.029) than patients not enrolled, but did not differ in other symptoms. Patients , 60y of age on a trial were more likely to be in the HSC than patients $ 60 y (60% vs. 41%, p , 0.038). Conclusions: Patientreported symptoms, particularly fatigue, drowsiness, pain and lack of sleep should be routinely assessed in phase I clinics, potentially improving attribution of adverse events. Also, current enrollment on a phase 1 trial does not appear to adversely impact symptoms. Treatment of symptoms may include counseling, especially for younger patients, given their higher reported symptom severity. Patient-reported outcomes can complement physician assessment to enhance treatment plans.

2542

Poster Session (Board #242), Sun, 8:00 AM-11:30 AM

2541

Poster Session (Board #241), Sun, 8:00 AM-11:30 AM

Clinical outcomes of patients with solid tumor enrolled in phase 1 clinical trials. First Author: Aaron Cleveland Denson, University of South Florida, Tampa, FL Background: Insufficient prognostic data can hamper patient enrollment into phase one clinical trials. Prior studies were limited by small numbers, selective patient population by tumor or therapy type. This study evaluated survival and prognostic factors in phase one clinical trial patients within all solid tumor types. Methods: All patients with solid tumors who participated in phase 1 therapeutic clinical trials from 1/2007 to 12/2013 at MCC were included. Data was extracted from the Oncore database and clinical charts. The primary endpoint was overall survival (OS) estimated using the KaplanMeir method. Results: We included 1408 patients enrolled in phase I clinical trials from 1/2007 to 12/2013 at Moffitt Cancer Center. The median age was 61 years (range 19-88) and 52% were male. Primary site of disease included Skin (26.6%), thoracic (23.5%), GI (17.5%), breast (8.7%), GU (7.2%), and others (16.3 %). Treatment types include targeted therapy (36.2%), immunotherapy (30.4%), a combination of chemo and targeted therapy (24.4%), or chemotherapy only (9%). Partial responses (PR) were seen in 6.5% of the patients, 37% had stable disease (SD) and 3.9% had a complete response (CR). Progressive disease (PD) was seen in 39% of patients. The median OS was 8.7 months (range: 0-85 months). Factors significantly associated with worse survival on multivariate analysis were history of DVT/PE , ECOG performance status $ 1, Liver metastasis, . 2 metastatic sites, . 2 prior systemic therapy, LDH . 618, wbc . 11. We validated the RMH prognostic score with a 95% CI of 0.361 (0.342, 0.381) as well as the MDACC prognostic score with 95% CI of 0.331 (0.313, 0.349). Serious AE (SAE) occurred in 56.6% of the patients with investigational drug related SAEs in 12% of the patients. The most common SAE included GI disorders, Surgical/Skin disorders, respiratory/thoracic/ mediastinal disorders, and investigations/blood disorders. Conclusions: This is one of the largest studies to assess clinical outcomes in a phase one clinical trial population which can help improve patient selection in order to design efficient and applicable Phase 1 clinical trials.

2543

Poster Session (Board #243), Sun, 8:00 AM-11:30 AM

Efficacy of a novel antibody-drug conjugate (ADC), ABT-414, as monotherapy in epidermal growth factor receptor (EGFR) amplified, recurrent glioblastoma (GBM). First Author: Martin J. Van Den Bent, Erasmus MC Cancer Center, Rotterdam, Netherlands

Outcomes in patients with relapsed/refractory Hodgkin lymphoma (RR-HL) with progression after autologous stem cell transplantation in the current era of novel therapeutics. First Author: Steven Michael Bair, Hospital of the University of Pennsylvania, Philadelphia, PA

Background: Recurrent GBM (rGBM) has dismal prognosis. Almost 50% GBM tumors harbor amplified (amp) EGFR. ABT-414 is a tumor specific ADC combining an antibody targeting a unique conformation of EGFR (ABT-806) to a microtubule cytotoxin, monomethyl auristatin F (MMAF). Here we report the safety and efficacy of ABT-414 monotherapy at recommended phase 2 dose (RPTD) in EGFR amp, rGBM. Methods: M12-356 (NCT01800695) is an open-label, phase 1, 3-arm study: Arm A (ABT-414+radiation/ temozolomide (TMZ) in newly diagnosed GBM (nGBM)), Arm B (ABT-414 +TMZ in nGBM as adjuvant therapy, or in rGBM) and Arm C (ABT-414 monotherapy in rGBM). Each arm had an escalation cohort to determine the RPTD and an expansion cohort to establish the safety and preliminary efficacy at RPTD. Results of Arm C expansion cohort at 1.25 mg/kg RPTD (IV infusion) are shown here. Eligible patients (pts) were adults with KPS score $ 70, EGFR amp (confirmed centrally), rGBM, normal end-organ function and no prior bevacizumab. Results: As of January 7, 2016, 48 EGFR amp, rGBM pts were treated in this cohort. The median age was 59 years (range, 35-80). Most pts had prior therapies: 40% had 1, 48% had 2, 10% had $ 3 prior therapies. Most common treatment emergent adverse events (TEAEs) ( $ 25% pts) were blurred vision (60%), headache, photophobia (29% each), dry eye, eye pain, fatigue (27% each). The most common serious AE ( . 1 pt) was seizure (8%). Grade 3/4 TEAEs ( . 1 pt) were keratitis (15%), corneal epithelial microcysts (8%), hemiparesis, hyperglycemia, muscular weakness, seizure (6% each), blurred vision, ulcerative keratitis (4% each). No dose-limiting toxicities were reported. Best RANO responses of 44 pts with complete data were: 2 partial responses, 18 stable disease, 24 progressive disease. The 6-month progression-free survival (PFS6) estimate was 30% [95% CI = 17, 44]. Conclusions: ABT-414 monotherapy, at 1.25 mg/kg RPTD, displayed frequent yet reversible ocular toxicities. An encouraging tumor stability/response and PFS6 were observed in this highly refractory EGFR amp, rGBM. A global randomized trial of ABT414, alone or with TMZ, vs. TMZ or lomustine, is underway in EGFR amp, rGBM (NCT02343406). Clinical trial information: NCT01800695.

Background: Patients (pts) with RR-HL who progress or relapse following autologous stem cell transplantation (ASCT) have a poor prognosis, with a historical median overall survival (OS) of 27 months (Crump, Blood 2008). More recently, these pts have been treated with novel therapies, particularly brentuximab vedotin (BV). We hypothesized that these agents have contributed to greater OS in this population. Methods: We conducted a retrospective study of all pts with RR-HL who underwent ASCT at the Univ. of Pennsylvania between 2004 and 2014. Therapy was given at discretion of the treating physician and reported data were collected through 10/2015. The primary endpoint was median OS from post-ASCT relapse. Results: Of 136 pts, 69 (50.7%) relapsed or progressed after ASCT. Median age was 37.5 years. Of pts who relapsed post-ASCT, 27 (40%) were exposed to at least one or more novel agents after relapse including 23 (34%) pts who received BV, 6 (9%) nivolumab, 5 (7%) lenalidomide, and 4 (6%) panobinostat. Median OS and follow-up were 34 and 72 months, respectively. Pts treated with novel agents had significantly prolonged OS compared to pts treated with conventional chemotherapy only (86 vs 22 months; p = 0.007). Other factors associated with improved OS in univariate analysis included age at relapse, response to ASCT, and radiation treatment (XRT) after postASCT relapse. Conclusions: Pts with RR-HL post-ASCT who are treated with novel agents may achieve prolonged OS when compared to chemotherapy alone. These results suggest improved outcomes when matched to historical controls and are likely a more accurate estimate of OS in the current era of expanded therapeutic options. Overall Age ,60 at relapse Age $60 at relapse Response to ASCT: CR/PR/SD Response to ASCT: PD Novel agent: Yes Novel agent: No Post-transplant XRT: Yes Post-transplant XRT: No

# of pts

Median OS, mos (95% CI)

69 60 9 33 31 27 41 46 21

34.1 (24.5 - 69.4) 46.1 (26.0 - 85.6) 19.0 (0.5 - n/a) 84.3 (33.8 - n/a) 22.4 (12.8 - 34.1) 85.6 (46.1 - n/a) 25.6 (12.8 - 33.8) 51.8 (26.0 - n/a) 24.5 (10.7 - 59.3)

P-value 0.0395 0.0024 0.0068 0.013

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics 2544

Poster Session (Board #244), Sun, 8:00 AM-11:30 AM

A phase I study of MEK162 and FOLFOX in chemotherapy-resistant metastatic colorectal cancer. First Author: May Thet Cho, City of Hope, Duarte, CA Background: Pre-clinical studies have shown that the combination of MEK inhibitors and 5-FU improves antitumor activity and that MEK inhibition overcomes both 5FU and platinum resistance. This phase I study was to determine the maximum tolerated dose (MTD) of the combination MEK162 and FOLFOX. Methods: Patients (pts) with metastatic colorectal cancer (mCRC) who progressed or failed prior 5FU, irinotecan, oxaliplatin and/or anti-EGFR therapy received twice daily MEK162 continuously or intermittently in combination with every-2-week FOLFOX without bolus 5-FU. Two dose levels of MEK162 (30mg and 45mg) were investigated in a standard 3 + 3 escalation design. Limited pharmacokinetic (PK) analysis of 5FU and oxaliplatin was performed at the MTD. An additional 6-patient expansion cohort of MEK162, a the defined MTD, was given BID on D1-D5 followed by FOLFOX on Day 6, repeated every 2 weeks. Dose limiting toxicity (DLT) was defined as any treatment-related grade (G) 3 or 4 nonhematological toxicity (with the exception of G3 diarrhea or vomiting , 48 hrs) or G 4 neutropenia or thrombocytopenia within the first 2 cycles (4 weeks) of the treatment. Results: 16 pts were enrolled in the continuous MEK162 arm (median age (range) 53 yrs (49-78); 11 men; ECOG 0/1 in 9/7 patients). No DLT was noted on the study. The MTD of MEK162 was 45 mg PO BID. An additional 6 pts (for a total of 12) were enrolled at the MTD for PK analysis and none of them developed DLT defining toxicities. A median of 8 cycles (range 1-19) was administered. Treatment-related $ grade 3 toxicities included anaphylaxis due to oxaliplatin (n = 1), CPK elevation (n = 2), neutropenia (n = 1), peripheral neuropathy (n = 3), thrombocytopenia (n = 1), retinal vascular disorder (n = 1), and acneiform rash (n = 1). 10 pts had SD at 2 months (m) by radiographic assessment, 5 of whom with stabilizations of . 4 m (4-10 m). There were no significant differences in either 5FU or oxaliplatin PK’s with or without MEK162. None of the intermittent MEK162 pts developed a DLT and all 6 progressed at 2 m. Conclusions: The combination of continuous MEK162 and FOLFOX has a manageable toxicity profile and promising antitumor activity in heavily pretreated mCRC patients. Clinical trial information: NCT02041481.

2546

Poster Session (Board #246), Sun, 8:00 AM-11:30 AM

Co-occurring genomic alterations and association with progression free survival in BRAFV600 mutated non-melanoma tumors treated with BRAF inhibitor. First Author: Shiraj Sen, The University of Texas MD Anderson Cancer Center, Houston, TX Background: Vemurafenib, dabrafenib and tremetinib are FDA approved for treating BRAFV600-mutated melanoma. Vemurafenib has activity in nonmelanoma BRAFV600-mutated tumors, as well. Unfortunately, most patients acquire resistance to BRAF monotherapy and mechanisms of resistance remain unknown. Methods: We analyzed CLIA certified clinical next generation sequencing data from BRAFV600E mutated non-melanoma tumors treated with BRAF inhibitor (BRAFi) from May 2012 to January 2016. We evaluated co-occurring genomic alterations and progression-free survival (PFS) of each patient. Results: Of the 30 patients treated with BRAFi monotherapy, 11 (37%) had NSCLC, 5 (17%) had colorectal cancer, 4 (13%) had cholangiocarcinoma, 3 (10%) had thyroid cancer, 3 (10%) had Erdheim Chester disease, and the remaining 4 (13%) had salivary gland carcinoma, glioblastoma, gliosarcoma, or unknown primary. Two of the 5 colorectal cancers were treated with BRAFi plus cetuximab. Three subsets of co-occurring genomic alterations were identified: 14 patients (47%) had no co-occuring alterations, 5 (17%) had PI3K/PTEN/mTOR pathway alterations, and 11 (37%) had other mutations: TP53 (n = 11), SMAD4 (n = 4), LKB1 (n = 2) and IDH (n = 2). Eight patients have ongoing response without progression on BRAFi monotherapy (PFS 222-805 days). Six of 8 responders (75%) have no co-occurring alterations, 2 of 8 (25%) have other mutations. All patients with co-alterations in the mTOR pathway progressed within 77 days. Tumors with mTOR pathway aberrations had a significantly lower median and mean PFS (mPFS, 53 days) compared to tumors with other cooccurring mutations (mPFS 206 days) and tumors without co-alterations (mPFS 276 days) (p = 1.03e-05). Conclusions: Co-occurring genomic alterations may help predict response to BRAFi therapy in BRAF-mutated tumors and PI3K/PTEN/mTOR pathway mutations may contribute to de novo resistance. Next generation sequencing is warranted on all BRAF mutated tumors as are further studies to address whether concurrently targeting cooccurring alterations will improve PFS. A trial combining BRAFi + mTOR inhibitor is underway (NCT01596140).

2545

125s

Poster Session (Board #245), Sun, 8:00 AM-11:30 AM

Phase I study of mTORC1/2 inhibitor BI 860585 as single agent or with exemestane or paclitaxel in patients with advanced solid tumors. First Author: Filippo G. De Braud, Istituto Naz. Tumori-Fond. IRCCS- Oncologia Medica, Milan, Italy Background: BI 860585 is a potent, selective ATP-competitive mTOR serine/threonine kinase inhibitor which has shown preclinical activity against various cancer types. This Phase I trial (NCT01938846) aims to determine the maximum tolerated dose (MTD) of BI 860585 as a single agent or combined with exemestane or paclitaxel in patients (pts) with advanced solid tumors. Methods: This study used a 3+3 design in 3 treatment arms (all in 28-day cycles): single-agent BI 860585 (5–300 mg/ day; Arm A), BI 860585 (40–220 mg/day) combined with fixed-dose exemestane (25 mg/day; Arm B), and BI 860585 (80–160 mg/day) combined with paclitaxel (60 or 80 mg/m2/week; Arm C). MTD is defined as the highest dose at which # 1 out of 6 pts experience a dose-limiting toxicity (DLT) during cycle 1. Results: To date, 41 pts have been treated in Arm A, 24 in Arm B and 17 in Arm C. DLTs have been observed in (n DLTs/n evaluable): 4/36 pts, and 4/20 pts in Arm A and B, respectively. No DLT has occurred in 12 pts in Arm C. DLTs (all grade 3) were diarrhea, rash and alanine/aspartate aminotransferase increased in Arm A; and stomatitis and rash in Arm B. The MTD was determined as 220 mg/day in Arm A. The most frequent treatmentrelated adverse events (AEs) have been: hyperglycemia (54%), diarrhea (39%) and nausea (37%) in Arm A; hyperglycemia (38%), diarrhea and rash (both 33%) in Arm B; and hyperglycemia (53%) and fatigue (47%) in Arm C; the majority of AEs were Grade # 2. No Grade 5 treatment-related AEs were observed. Preliminary best response data showed stable disease (SD) in Arm A (n = 18), SD (n = 6) and partial response (PR; n = 4) in Arm B, including a breast cancer pt pretreated with an mTORC1 inhibitor and exemestane, and SD (n = 8) and PR (n = 1) in Arm C. Disease control rates (PR+SD) were 44% in Arm A, 46% in Arm B and 59% in Arm C. Preliminary PK analysis showed linear PK with no food effect and no drug-drug interaction with exemestane or paclitaxel. Conclusions: MTD was determined as 220 mg/day for BI 860585 monotherapy. At MTD, AEs were mild-to-moderate and consistent with the mode of action of the compound. Early signs of efficacy were observed in varied tumor types with BI 860585 monotherapy and in combination. The study is ongoing. Clinical trial information: NCT01938846.

2547

Poster Session (Board #247), Sun, 8:00 AM-11:30 AM

A phase I/II study of TKM-080301, a PLK1-targeted RNAi in patients with adrenocortical cancer (ACC). First Author: Michael J Demeure, Ashion Analytics, Phoenix, AZ Background: Polo-like kinase 1 (PLK1) regulates critical aspects of tumor progression. High expression levels correlate with poor survival in ACC. TKM080301 is a lipid nanoparticle formulation of a siRNA against PLK1. Methods: TKM-080301 was evaluated in an open-label study with dose escalation and expansion phases in patients with refractory ACC. TKM080301 was infused IV over 30 minutes on days 1, 8, and 15 of a 28-day cycle. Primary study objectives included assessment of safety and anti-tumor activity by RECIST 1.1 after every 2 cycles. Results: Sixteen patients were treated at 0.6 or 0.75 mg/kg/week for up to 18 cycles. Eight received at least 2 cycles of study treatment and were evaluable for tumor response. Four had a best response of stable disease including one with a 13% reduction of target tumor diameter. One 51 yo man with metastatic intraperitoneal nonfunctional ACC had a partial response (target tumor reduction of 19% after cycle 2 and 49% after cycle 14). Residual tumor was resected and histopathology showed near-complete necrosis. Paired tumor/normal whole exome sequencing of this tumor revealed key somatic alterations including a missense mutation (Q331H) in TP53’s DNA binding domain, an NF1 nonsense mutation (S365*), and a missense ARID1A mutation (A438V). RNA sequencing showed elevated expression of PLK1 (FPKM = 13.5) in this tumor compared to a normal adrenal RNA control (FPKM = 0.95). Two subjects completed at least 6 cycles. Subjects were discontinued for progressive disease (7); infusion reactions (2); acute respiratory failure (1); elevated LFTs (1); or bowel obstruction (1). Most common AEs related to TKM-080301 were pyrexia (56%); chills (50%); back pain (31%); infusion reaction (31%); and nausea (25%). Serious AEs related to TKM-080301 were ECG T-wave inversion and musculoskeletal pain (1) and infusion reaction (1). Conclusions: TKM-080301 has been tolerated at a dose of 0.6 0.75 mg/kg for up to 18 cycles. Preliminary anti-tumor efficacy has been observed. A potential molecular therapeutic context of increased PLK1 expression with inactivation of p53 or NF1 was observed in a remarkable responder. Further evaluation of PLK1 as a therapeutic target of TKM080301 in ACC is warranted. Clinical trial information: NCT01262235.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

126s 2548

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics Poster Session (Board #248), Sun, 8:00 AM-11:30 AM

Dose escalation of POL6326 in combination with eribulin in HER2-negative relapsed metastatic breast cancer (mBCa) patients (pts). First Author: Sonia Pernas Simon, Catalan Institute of Oncology, Barcelona, Spain Background: POL6326 is a cyclic peptide and a potent, selective antagonist of the chemokine receptor CXCR4. Evidence suggests that CXCR4 inhibition interferes with the tumor-protective microenvironment and therefore sensitizes tumor cells to chemotherapy. Consequently, POL6326 may enhance the effects of cytoreductive therapy and is being developed as adjunctive cancer treatment. Methods: This open label dose escalation study was conducted in pts with relapsed mBC and with positive CXCR4 expression at tumor site to establish the Maximum Tolerated Dose (MTD) of the combination of POL6326 and eribulin. MTD was defined as the highest dose at which DLT occurs in no more than 1 of 6 pts. Cohorts received eribulin on days 2 and 9, flanked by POL6326 on days 1-3, and 8-10 of 21-day cycles. Pts were treated until disease progression or DLT. Results: 24 pts (median age 53 [38-74]; ECOG PS 1) wereenrolled in 8 cohorts. Eribulin mesilate was administered at 1.4 mg/m2 apart from two cohorts receiving 1.1 mg/m2. POL6326 was escalated from 0.5 to 3.5 mg/kg. MTD of the combination has not been achieved yet. The most commonly reported (Gr 3-4) adverse events were neutropenia (9/24 pts) and leucopenia (3/24 pts); 3/24 pts developed neutropenic fever and 1 patient on 1.4 mg/m2eribulin and 1.0 mg/kg POL6326 died from sepsis. 9/24 pts experienced (Gr 1-2) histaminergic reactions related to POL6326 that were manageable with anti-histamines. Increases of Cmax and AUC were linear among POL6326 doses 0.5-3.5 mg/ kg. Eribulin pharmacokinetics were in line with published values with no indication of PK interaction with POL6326. The median duration of treatment was 4 cycles [1-8] and 4 pts achieved a partial tumour response, all on the higher dose of eribulin. Conclusions: POL6326 in doses of 0.5-3.5 mg/ kg can be combined safely with eribulin 1.1-1.4 mg/m2 in pts with pretreated mBC. Incidence of neutropenia thus far has been similar to eribulin monotherapy. The combination with POL6326 may enhance cytoreductive potential of eribulin, and this treatment approach therefore warrants efficacy studies in mBC. Clinical trial information: NCT01837095.

2550

Poster Session (Board #250), Sun, 8:00 AM-11:30 AM

2549

Poster Session (Board #249), Sun, 8:00 AM-11:30 AM

Suppression of poly-ADP ribose (PAR) levels in PBMCs by veliparib (vel) as a pharmacodynamic (PD) marker associated with survival among women with BRCA1- or BRCA2- (BRCA)-associated metastatic breast cancer (MBC). First Author: Jeffrey N. Weitzel, City of Hope, Duarte, CA Background: PARP inhibitor therapy of BRCA-associated cancer exploits the concept of synthetic lethality, yet there have been limited PD markers of efficacy. We report here on baseline PAR and suppression of PAR in PBMCs as a PD marker of clinical outcome, as well as pharmacokinetics (PK). Methods: Pts received either carboplatin (carb) IV and oral vel (Phase I) or single agent vel (Phase II) followed by post-progression therapy with the combination (NCI protocol #8264; NO1-CM-2011-00038). Blood for PK analysis (phase I only) and PBMCs were collected at baseline, at 3hrs, and at additional later time points. Plasma vel and PAR were quantitated using LCMS assay and CTEP-validated reagents and methods, respectively. Results: 72 evaluable pts included 28 in Phase I, 44 in Phase II. PK samples were available for 25/28 pts in phase I; vel PK parameters are in line with values previously reported (Cl/F 20.9 L/h, V/F 173 L, t1/2 6.1 h). Response rate (RR) in phase I was 56% [CR rate 15%] and 3 remain in CR at 39, 42, and 46 months (2 on vel maintenance, 1 is off therapy). Paired baseline and 3 hr PBMC samples for PAR levels were available for 50 pts (18 phase I, 32 phase II). Higher baseline PAR levels (as a continuous variable) predict for better TTF (p , 0.01) and OS (p , 0.04), with marginal significance in PFS (p = 0.05) in Phase I patients. Using the median baseline PAR level from Phase I pts as a cut-point, there was increased hazard for the low PAR group (HR = 4.21 [1.39-12.8] in Phase I for TTF, and 2.1 [0.84-5.03] for Phase II). OS was worse for low PAR in both Phase I (HR = 2.75 [1.0-7.6]) and Phase II (HR = 3.7 [1.3-10.4]). Pts whose PAR level was reduced by more than 50% at 3 hrs also showed improved outcome. In multivariate analysis (including triple negativity, BRCA2 vs BRCA1, no. of sites, visceral disease, no. of prior Rx, ER/PR and age) only PAR (either the 50% reduction in PAR or baseline PAR , median, but not both as they were highly inversely correlated), was retained in the final prediction model for OS and TTF. Conclusions: Baseline level and reduction of PAR after treatment with vel associates with outcome among women with BRCA-associated MBC. Clinical trial information: NCT01149083.

2551

Poster Session (Board #251), Sun, 8:00 AM-11:30 AM

Phase I study of BPM 31510 in advanced solid tumors: Omics-based molecular correlation to outcome for patient stratification. First Author: Manish A. Shah, Weill Cornell Medicine, New York-Presbyterian Hospital, New York, NY

Phase I/II dose-escalation and expansion study of afuresertib + carboplatin and paclitaxel in recurrent ovarian cancer. First Author: Sarah Patricia Blagden, OCA Research Centre, Imperial College London, London, United Kingdom

Background: BPM 31510 targets the metabolic machinery of cancer cells to reverse the Warburg effect. The current study evaluates BPM 31510 as a 6day continuous infusion as a single agent and in combination with 3 standard weekly chemotherapy regimens. The study was designed to determine the Phase II dose and infer multi-omic patient-centric signatures identifying molecular predictive markers of clinical benefit and patient stratification. Methods: Eligible patients (pts) (aged $ 18 y) were relapsed/refractory to standard therapy. The monotherapy arm received IV BPM 31510 for 6 d in continuous infusion in 28-d cycles, and combination arms (gemcitabine, 5FU or docetaxel) were primed for 3 wks followed by weekly dosing in a 6 wk cycle. Doses were escalated in a 3+3 schema. Endpoints were safety, PK and multi-omics analysis. Tumor response is evaluated at wk 2 and then after every 2 cycles. Results: As of 10 Dec 2015, 85 pts have been enrolled. DLTs were reported at 171mg/kg in mono and at 137mg/kg in the gemcitabine arm (maximum administered dose) and were coagulopathy-related. Six of 66 pts (24%) maintained a minimum of Stable Disease for $ 4 cycles. Molecular signatures of response and safety were derived from the integrated omics and AI profiling of the Interrogative BiologyO` platform. Biomarker candidates correlating with favorable clinical response and safety were identified. pK levels of BPM 31510 was a driver of favorable response. These molecular correlates were independent of tumor type and prior therapy indicating a broad anti-tumor effect of BPM 31510. Novel multi-omic panels could stratify response before and 24h post treatment with AUC . 0.85. Conclusions: BPM 31510 is well tolerated as a monotherapy and in combination with chemotherapeutic agents and associated with early anti-tumor activity. The MTD of 110mg/kg in the gemcitabine arm is being established and used in a planned Phase II pancreatic cancer trial. Dose-escalation in the other 3 arms is ongoing to determine phase II doses. Molecular signatures will be utilized in phase II trials to guide the patient stratification and selection for a variety of cancer types that will guide the development plan for BPM 31510 as anticancer agent. Clinical trial information: NCT01957735.

Background: Patients (pts) with ovarian cancer (OC) often develop resistance to standard medical treatment (Rx) with platinum-taxane chemotherapy. Preclinical studies have shown that intra-nuclear activation of AKT by DNAdependent protein kinase inhibits cisplatin-mediated DNA damage in OC cell lines, which is reversible through AKT inhibition. The following study investigates whether the oral pan-AKT inhibitor afuresertib can restore sensitivity to platinum Rx and reverse acquired resistance to standard Rx in recurrent OC. Methods: In phase I, the maximum tolerated dose (MTD) was investigated; pts with recurrent OC were treated with afuresertib (orally qd) at doses escalating from 50 to 150 mg combined with paclitaxel (IV 175 mg/ m2) and carboplatin (AUC 5) q3w for # 6 cycles followed by afuresertib maintenance until progression/toxicity. Phase II investigated efficacy and safety of afuresertib at MTD plus paclitaxel and carboplatin followed by afuresertib maintenance. Pts were split into two cohorts: platinum resistant (PTR) or platinum refractory pts. Results: Phase I enrolled 29 pts; doselimiting toxicities (DLTs) were reported in 1 pt from the 125 mg cohort (G3 rash [n = 1]) and 2 pts from the 150 mg cohort (G3 rash in both pts with concurrent G2 febrile neutropenia and G2 lip swelling in 1 pt). The MTD of afuresertib was established as 125 mg. Phase II enrolled 30 heavily pretreated pts (median time since diagnosis = 2.1 y; median prior Rx lines = 3); 28 were PTR and 2 were platinum refractory. Tolerability of Rx was consistent with findings in phase I; G3/4 AEs included diarrhea (20%), fatigue (10%), rash (10%), vomiting (7%), and nausea (3%). Overall response rate (ORR) per RECIST v1.1 for pts in the platinum-resistant cohort was 32.1% (95% CI: 15.9–52.4) and median PFS was 7.1 months (95% CI: 6.3–9.0). ORR per GCIG CA125 in evaluable pts was 52.0% (n = 25; 95% CI: 31.3–72.2). Ad hoc analyses showed ORR to correlate with duration of sensitivity to prior platinum Rx and platinum-free intervals. Conclusions: An MTD of 125 mg afuresertib was established for the Rx of pts in combination with paclitaxel and carboplatin. Combination Rx was well tolerated and showed promising activity in pts with PTR OC. Clinical trial information: NCT01653912.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics 2552

Poster Session (Board #252), Sun, 8:00 AM-11:30 AM

Development of novel backbones for the treatment of peripheral T-cell lymphoma (PTCL): The pralatrexate/romidepsin doublet. First Author: Jennifer Effie Amengual, Columbia University Medical Center, New York, NY Background: Peripheral T-cell lymphoma (PTCL) is a heterogeneous group of lymphomas in which only 20-25% of patients experience long-term survival with CHOP chemotherapy. Recently several drugs have been approved for this entity including pralatrexate (P), romidepsin (R), and belinostat which have response rates ranging from 26%-29% as single agents. Based on our demonstration of synergy of P+R in preclinical models of TCL, we initiated a study on the safety and efficacy of P+R in a phase I-II study for relapsed or refractory lymphomas (NCT01947140). Methods: A 3+3 dose-escalation study started at P 10mg/m2 and R 12mg/m2 with escalation to P 25 mg/m2 and R 14 mg/m2. Patients were treated on 1 of 3 dosing schedules (weekly x 3 Q28D; weekly x 2 Q21D and QOW Q28D). The primary objective was to determine MTD and DLT; the secondary objective included describing ORR (CR+PR). Patients were required to have relapsed lymphoma of any subtype, ECOG PS , 2, and adequate organ and marrow function. There was no upper limit to the number of prior therapies or transplantation. Results: Twenty-five patients were enrolled and were evaluable for toxicity. Median age was 52 yrs (23-73) and 60% were male. The median number of prior therapies was 3 (range 1-16). Histologies included HL (N = 3), B-cell (N = 7 of which FL = 4) and T-cell (N = 15). The median number of cycles completed was 4 (range 18+ ongoing). There were 3 DLTs in cohort 4 (P 20mg/m2 & R 12mg/m2given weekly x 2 Q21D) consisting of 2 Grade 3 oral mucositis and 1 Grade 4 sepsis. The QOW Q28D schedule had no DLTs at equivalent and higher doses. The grade 3/4 toxicities reported in . 5% of patients were: neutropenia (32%), thrombocytopenia (32%), anemia (24%), oral mucositis (16%), hyponatremia (8%), pneumonia (8%) and sepsis (8%). Twenty-one patients were evaluable for response, 2 patients are currently on therapy. The ORR in the total, non-PTCL and PTCL populations was 52%; 33% (no CR) and 77% (38% CR) respectively. Presently 1 patient is awaiting safety and response analysis. Conclusions: These data support the lineage specific activity of the P+R combination, which is being expanded to a multicenter Phase II for PTCL. Clinical trial information: NCT01947140.

2553

127s

Poster Session (Board #253), Sun, 8:00 AM-11:30 AM

Pazopanib to suppress MET signaling in patients with refractory advanced solid tumors. First Author: Apurva K. Srivastava, Leidos Biomedical Research, Inc., Frederick National Laboratory for Cancer Research, Frederick, MD Background: Activation of MET signaling has been proposed as an adaptive response to pharmacological inhibition of VEGFR. We compared MET and phospho-MET levels in paired biopsies collected before and after pazopanib treatment of patients (pts) with advanced solid tumors enrolled in this single-agent arm of the Phase I trial of a pazopanib/tivantinib combination (NCT01468922). Methods: Eligible pts had histologically confirmed refractory cancers; $18 years old; and adequate bone marrow, hepatic, and renal function. Pts underwent pre-treatment tumor biopsy, received 7 days of pazopanib (800 mg/d, po) followed by post-treatment biopsy on day 8. The biopsies were flash frozen and one core processed into total cell lysates for MET analysis by validated ELISA (Srivastava et al, JCO 2011). Results: Paired biopsies were evaluable in 8 pts (age 19-59 yrs), and levels of full length MET (total) were measurable in 6 biopsy pairs, ranging from 3.1 - 626.3 fmol/mg protein (Table). Pazopanib treatment decreased total MET levels in 5 out of 6 pts; similarly, it decreased pMET levels in 3 of 3 pts. This decrease in MET levels did not appear to be related to sampling variability, based on H&E staining of the second pass biopsies. In vitro experiments using MET/VEGFR2 expressing cell lines confirmed decreased MET and pMET levels following pazopanib treatment. Conclusions: Pazopanib-associated decline in MET levels were contrary to the hypothesis of the trial that blockade of VEGFR would result in activation of MET. Pazopanib is known to inhibit multiple kinases including MET (MET IC50 6uM), so establishing a link between VEGFR blockade and MET response will need further study in clinical trials of VEGFR-targeted agents with greater target selectivity than pazopanib. Funded by NCI Contract No HHSN261200800001E. Clinical trial information: NCT01468922. Cancer Type

Biopsy Time

Total MET (fmol/mg Pr)

pY1234-35MET (fmol/mg Pr)

25

RCC

26

CUP

28

Ovarian

Pre Post Pre Post Pre Post Pre Post Pre Post Pre Post

73.5 54.0 5.2 12.1 626.3 120.7 3.1 ,LLQ 434.7 132.8 135.8 8.6

,LLQ ,LLQ ,LLQ ,LLQ 177.7 19.0 ,LLQ ,LLQ 168.1 66.2 1.7 ,LLQ

Patient #

29

NSLC

30

Esophageal

32

Colon

pY1356MET (fmol/mg Pr)

63.8 ,LLQ 72.4 19.7

LLQ, limit of quantitation

2554

Poster Session (Board #254), Sun, 8:00 AM-11:30 AM

Results of a phase I study of RX-5902: An orally bioavailable inhibitor of phosphorylated P68, targeting solid tumors. First Author: S. Gail Eckhardt, University of Colorado School of Medicine, Anschutz Medical Campus, Aurora, CO Background: RX-5902 is a novel compound that targets phosphorylated p68 RNA helicase (also known as DDX5), a member of the DEAD box family of RNA helicases. Phosphorylated p68 may play a vital role in cell proliferation and tumor/cancer progression. As a single agent, RX-5902 inhibits tumor growth, alters cell migration and enhances survival in a variety of in vivo animal xenograft tumor models (e.g., breast, ovarian, renal, pancreatic). We report the data from the first clinical study of RX-5902 as a single agent to treat solid tumors. Methods: This is a Phase 1 study (NCT02003092) designed to evaluate safety, tolerability and pharmacokinetics following increasing doses of RX-5902 at varying schedules. Primary objectives include safety, tolerability and dose limiting toxicities to identify the maximum tolerated dose and a recommended phase 2 dose and schedule (RP2D). Secondary objectives were pharmacokinetics (PK) and antitumor activity (RECIST v1.1). Eligible subjects (aged $ 18 years) with relapsed/refractory solid tumors received oral RX-5902 at 1, 3, 5 or 7 times per week for 3 weeks followed by 1 week of rest or for 4 weeks without a rest. Plasma concentrations were measured using a validated LC-MS/MS assay, and noncompartmental pharmacokinetic parameters were calculated using WinNonlin, Version 6.4. Results: As of January 2016, 18 subjects have been enrolled (8 Females, 10 males). No dose limiting toxicities or treatment related SAEs have been reported. Six subjects have experienced stable disease; three subjects are currently receiving treatment for . 1 year. The most common side effects were grade 1 related adverse events: nausea, vomiting and fatigue; no grade 2 related events have been reported. RX5902 was orally bioavailable with median Tmax of 2 hours and median elimination half-life of 12 hours. Conclusions: Data from this study support that RX-5902 is safe and well tolerated at the doses and schedules tested. Early anti-tumor activity has been observed. A recommended phase 2 dose for RX-5902 for the treatment of triple negative breast cancer and advanced ovarian cancer will be presented. Clinical trial information: NCT02003092.

2555

Poster Session (Board #255), Sun, 8:00 AM-11:30 AM

Results of a phase 1 study of single agent RX-3117: An oral antimetabolite nucleoside to treat solid tumors. First Author: Drew W. Rasco, South Texas Accelerated Research Therapeutics, San Antonio, TX Background: RX-3117 is an oral small-molecule antimetabolite, cyclopentyl pyrimidyl nucleoside that is activated by uridine cytidine kinase 2. RX-3117 has shown efficacy in xenograft models of gemcitabine resistant pancreatic, bladder and colorectal cancer. Emerging data from a Phase 1 clinical study of RX-3117 as a single agent in subjects with solid tumors is described below. Methods: The Phase 1 study (NCT02030067) is designed to evaluate safety, tolerability and pharmacokinetics (PK) following increasing doses and schedules of RX-3117 in eligible subjects (aged $ 18 years) with relapsed/refractory solid tumors. Primary objectives include safety and tolerability to determine the MTD and a recommended phase 2 dose and schedule (RP2D); secondary objectives were PK and antitumor activity (RECIST v1.1). Subjects received oral RX-3117 3, 5 or 7 times per week for 3 weeks followed by 1 week of rest or 4 weeks without a rest for up to 8 cycles. Plasma concentrations were measured using a validated LC-MS/MS assay, and noncompartmental pharmacokinetic parameters were calculated using WinNonlin, Version 6.4. Results: As of Jan 2016, 41 subjects have been enrolled (26 Females, 15 males). Eleven subjects experienced stable disease for 1 to 10 cycles; with 7 subjects receiving treatment from 104 to 276 days. A tumor burden reduction was seen in 3 subjects with pancreatic, breast and mesothelioma cancers. RX-3117 PK demonstrated a dose proportional exposure and was rapidly absorbed without a marked lag time, and with a median Tmax between 2 to 3 hours; accumulation was generally minimal. The most frequent related adverse events were moderate to severe anemia, mild to moderate fatigue and nausea, mild diarrhea, vomiting, and anorexia. Dose limiting toxicity of anemia was observed at 2000 mg administered 3 times per week. The recommended phase 2 dose and schedule will be 700 mg 5 times per week. Conclusions: Data from this study RX-3117 is safe and well tolerated at the doses and schedules tested. Early anti-tumor activity has been observed in pancreas, colorectal and mesothelioma cancers. A 2 stage Phase 2A design for pancreatic and bladder cancers is ongoing. Clinical trial information: NCT02030067.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

128s 2556

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics Poster Session (Board #256), Sun, 8:00 AM-11:30 AM

2557

Poster Session (Board #257), Sun, 8:00 AM-11:30 AM

A phase 1 trial of TRC102 (methoxyamine HCl) with temozolomide (TMZ) in patients with solid tumors and lymphomas. First Author: Robert S. Meehan, NCI NIH Division of Cancer Treatment and Diagnosis, Bethesda, MD

Efficacy and hematologic toxicity of carboplatin and talazoparib combination therapy in BRCA mutated patients. First Author: Mallika Sachdev Dhawan, University of California, San Francisco, San Francisco, CA

Background: The base excision repair (BER) pathway has been shown to promote resistance to both alkylating agents and antimetabolite chemotherapy. TRC102 inhibits BER by binding to chemotherapy-induced abasic sites; it potentiates the activity of alkylating agents including TMZ in murine models. In human xenograft studies, TRC102 efficiently enhanced the antitumor effect of TMZ regardless of cell line genetic characteristics, including O6-methylguanine DNA-methyltransferase (MGMT), mismatch repair (MMR), or p53 status. Methods: We conducted a phase 1 trial of TRC102 in combination with TMZ to determine the safety, tolerability, and maximum tolerated dose (MTD) of the combination in patients (pts) with refractory solid tumors. Eligibility criteria included adults with histologicallyconfirmed malignancies that had progressed on standard therapy, ECOG PS of 0-2, and adequate organ function. We followed a standard 3+3 design. Oral TRC102 and TMZ were administered daily, days 1-5, in 28-day cycles. We also evaluated the pharmacokinetic profile of TRC102 with TMZ. Antitumor responses were determined using RECIST criteria. Results: The escalation phase accrued between August 2013 and August 2015. MTD was reached at DL7 (TRC102 150mg PO D1-5, TMZ 150mg/m2 PO D1-5). The DLT at DL8 was hematologic. A total of 37 pts were enrolled, of whom 4 had a partial response (PR) (NSCLC, ovarian(2), and colon); one further PR occurred in a pt with colon cancer but was unconfirmed due to concurrent illness; 11 pts had stable disease (SD), 16 progressive disease (PD), and 5 were not evaluable. Three of the PR pts remain on study (20, 7, 5 months). Grade 3/4 adverse events included neutropenia (11%), anemia (8%), thrombocytopenia (5%), hemolysis (3%) or hypophosphatemia (3%). The half-life of TMZ was 2 hours, which was similar to reported monotherapy values; TRC102 half-life was 24 hours and its plasma concentrations increased with dose. Conclusions: The observation of 4 PRs and 11 SDs demonstrates that this combination is active and the side effect profile is manageable. The expansion phase of the trial is in progress. MGMT status, gH2Ax, caspase 3, and Ki67 will be evaluated in tumor biopsies from the ongoing expansion cohort. Clinical trial information: NCT01851369.

Background: Synergistic anti-tumor effects of PARPi and chemotherapy have been observed in preclinical models. This Phase 1 trial evaluates the tolerability, dose limiting toxicities (DLT) and efficacy of, the PARP inhibitor, talazoparib, in combination with carboplatin in patients with or without DNA repair mutations. We hypothesize that talazoparib may overcome carboplatin resistance but induce greater toxicity in patients with DNA repair defects. Methods: Talazoparib and carboplatin pharmacokinetics (PK), safety and anti-tumor activity were evaluated in a 3+3 dose escalation design. Genetic testing, PK, pharmacodynamic effects (PD), and alternate dose modeling were evaluated to better understand the interaction of carboplatin and talazoparib. Results: 24 patients were enrolled in 4 cohorts and treated with talazoparib 0.75 or 0.1 mg/day and carboplatin AUC 1 or 1.5 for 3/3 or 2/3 weeks. Tumor types included: breast, prostate, cholangiocarcinoma, ovarian, bladder and adenoid cystic carcinoma. DLTs were fatigue and thrombocytopenia; other grade 3/4 toxicities included fatigue (13%), neutropenia (33%), thrombocytopenia (33%), and anemia (58%). Post cycle 1 heme toxicities required dose delays/reductions in almost all patients. One complete response occurred in a germline BRCA1 (gBRCA1) breast cancer patient and two partial responses (PR) occurred in gBRCA2 patients (breast and bladder cancer); 11 patients (pts) had stable disease (SD) for $ 3 months, 5 pts are not yet evaluable. Thrombocytopenia (11.4% vs. 1.1% P , 0.001) and neutropenia (9.1% vs. 3.8% P , 0.001) were significantly more common in gBRCA carriers. PK modeling suggested that monthly carboplatin dosing in patients with DNA repair defects and every other week dosing in non-carriers may be more tolerable for PARPi and carboplatin combinations. Conclusions: Mutation carriers in BRCA or DNA repair genes responded better to the combination and responses were in prior progression on PARPi and carboplatin, but the combination caused increased heme toxicity in gBRCA or other DNA repair mutations. Less frequent carboplatin dosing may be required when given with PARPi in germ line mutation carriers. Clinical trial information: NCT02358200.

2558

2559

Poster Session (Board #258), Sun, 8:00 AM-11:30 AM

A phase 1 study of AR-42 in patients with advanced solid tumors, including nervous system tumors. First Author: Hugo Valencia, Ohio State University, Columbus, OH Background: We have shown that AR-42, an oral pan-histone deacetylase inhibitor, has clinical activity in some hematologic malignancies and preclinical activity in certain solid tumors, including neurofibromatosis type 2 (NF2)-associated meningiomas and schwannomas. Therefore, this phase 1 trial was initiated to investigate the dose limiting toxicities (DLTs) and the maximum tolerated dose (MTD) in patients with advanced solid tumors. Methods: This is a single-agent, standard 3+3, phase 1 dose-escalation study, with an additional expansion cohort at the MTD; mainly enrolling patients with malignancies of interest, based on our preclinical data. AR-42 was given in a 28-day cycle (3 times weekly, 3-week on, 1-week off). Plasma samples were collected following the first dose to assess pharmacokinetics in each patient. Results: Seventeen patients were enrolled in this study. Two patients were replaced and 15 patients were evaluable for safety and response. Patients had a median age of 49 (range: 20-80) and were heavily pretreated (41% had 3 prior regimens). Primary tumor sites included NF2associated tumors (n = 5), meningioma (n = 2), transitional cell carcinoma (n = 3), breast cancer (n = 2), and other (n = 5). The MTD was defined as 60 mg and the DLTs were two grade 3 thrombocytopenias; and one grade 4 psychiatric disorder, possibly related to AR-42 (at 80 mg). The most common toxicities were fatigue, cytopenias, electrolytes imbalance, nausea/vomiting and elevated creatinine, which were manageable. The best response was stable disease (53%, 8/15 patients) with a median progression free survival (mPFS) of 4.4 months (95% CI: 1.2-13.9). Interestingly, for NF2 and meningioma patients, the mPFS was 13.9 months. Individual plasma AR-42 Cmax and AUC0-24hr values ranged from 0.54 to 3.25 mM and 4.93 to 27.6 mM*Hr across the dose range, respectively. Conclusions: Single-agent oral AR-42 is well tolerated and associated with modest anti-tumor activity in patients with advanced solid tumors, especially those with NF2-asociated tumors and meningiomas. Pharmacokinetic data suggests active drug levels are achieved in plasma. In this study, the MTD is determined to be 60 mg, once daily, 3 times weekly, 3-week on and 1-week off. Clinical trial information: NCT01129193.

Poster Session (Board #259), Sun, 8:00 AM-11:30 AM

Results from a phase 1b/2 study of RX-0201, a novel AKT-1 antisense, combined with everolimus to treat metastatic clear cell renal carcinoma. First Author: Neeraj Agarwal, Huntsman Cancer Institute at the University of Utah, Salt Lake City, UT Background: RX-0201 is a novel, oligonucleotide that binds native AKT-1 mRNA, preventing downstream phosphorylation to p-AKT. RX-0201, in combination with everolimus, to treat metastatic clear cell renal carcinoma. In vitro RX-0201, in combination with everolimus, additively inhibited Caki1 cell growth. Methods: The Phase 1b/2 proof-of-concept, multicenter, open label study is conducted in 2 stages (NCT02089334). Stage 1 is a doseescalation Phase 1b/2 study of RX-0201 in combination with everolimus (10 mg daily). RX-0201 was given as a continuous intravenous infusion for 14 days followed by 7 days of rest. Subjects were enrolled at increasing doses of RX-0201 in a 3+3 design. The target dose of RX-0201 identified in Stage 1 is being studied further evaluated in Stage 2, which is a randomized, 2-arm study of RX-0201 in combination with everolimus versus everolimus alone. Plasma concentrations were measured in Stage 1 and noncompartmental pharmacokinetic parameters were calculated using WinNonlin, Version 6.4. Results: In Stage 1 of the Phase 1b study, subjects were treated with 125 mg/m2/day (n = 3), 200 mg/m2/day (n = 4), and 250 mg/m2/day (n = 3) RX-0201 in combination with 10 mg everolimus. Seven males and three females, median age 61 years, were treated. Subjects received 1-3 treatments prior to study entry (median = 1). The most common toxicities attributed to the combination were rash, mouth ulceration, decreased weight, thrombocytopenia, facial edema, fatigue, and pruritus. No significant events were attributed to RX-0201 alone. Most events (81%) were mild or moderate in severity. Based on the tolerability, 250 mg/m2/day RX-0201 dose was declared the recommended Stage 2 dose. Four subjects in the Stage 1 of Phase 1b study have experienced stable disease for 383, 191, 129 and 58 days; also a tumor burden reduction from 16% to 38.5% was seen in four subjects. RX-0201 PK demonstrated a dose proportional exposure. Conclusions: RX-0201, in combination with everolimus, appears to be safe and well tolerated in patients with metastatic renal cancer at doses up to 250 mg/m2/day. Stage 2 of the Phase 1b/2 clinical study is currently ongoing. Clinical trial information: NCT02089334.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics 2560

Poster Session (Board #260), Sun, 8:00 AM-11:30 AM

A phase 1 study of the PI3K/mTOR inhibitor PQR309 evaluating safety, pharmacokinetics (PK) and pharmacodynamics (PD) in patients (pts) with advanced solid tumors. First Author: Alex A. Adjei, Roswell Park Cancer Institute, Buffalo, NY Background: PQR is an oral pan-PI3K, mTORC1/mTORC2 inhibitor in clinical development. Methods: A Phase I trial of PQR to evaluate safety, PK and PD in pts with advanced solid tumors was performed, using the standard “3+3” design. Pts with ECOG PS of 0-1 were treated with escalating doses of PQR administered on a once daily continuous dosing schedule (qd). The dose limiting toxicity (DLT) period was the first 21 days of treatment. PK samples were obtained at predefined time points. Blood samples for the evaluation of glucose, insulin and c-peptide as PD markers were obtained at multiple time points. Results: 16 pts (11F:5M) were enrolled as of November 30 2015 and treated at 3 doses of PQR (80, 100 and 120 mg) qd. Median duration of therapy was 50 days (range 7 – 255). Adverse events (AEs) in $ 50% pts included fatigue, hyperglycemia, rash, nausea, vomiting, diarrhea and weight loss. Drug related grade 3 AEs seen in more than 1 pt were fatigue, hyperglycemia, rash and elevated transaminases. DLT’s were grade 3 fatique in 3 pts at 120mg and 1 pt at 100mg qd. However, dose interruptions at the 100mg dose indicated that this was not a tolerable dose. Thus the recommended phase II dose (RP2D) and MTD is 80 mg qd. PK analysis shows fast absorption (Tmax 1-2h), dose proportionality, an accumulation ratio of about 3.6 and an estimated T1/2 of 51 hours. Blood PD analysis demonstrated concentration dependent elevations in glucose, insulin and c-peptide within 2 hours after the administration of PQR. One pt with progressing mesothelioma had stable disease (SD) for 8.5 months. One pt each with leiomyosarcoma and ovarian cancer had SD for 3 months. Conclusions: Consistent with an earlier study, the MTD and RP2D of PQR is 80mg qd. Concentration dependent hyperglycemia has been confirmed as a mechanism-based toxicity and will be investigated as a PD marker. The concomittant increase in serum insulin suggests that insulin should not be the initial therapy to manage drug-induced hyperglycemia. Glucose transport inhibitors may be the treatment of choice in management of hyperglycemia. Recruitment in the 80mg qd expansion cohort with a focus on tumors with PI3 kinase pathway aberrations is ongoing. Clinical trial information: NCT02483858.

2562

Poster Session (Board #262), Sun, 8:00 AM-11:30 AM

Pharmacokinetic (PK) effects and safety of olaparib in combination with tamoxifen, anastrozole, or letrozole: Phase I study. First Author: Ruth Plummer, Northern Centre for Cancer Care, Newcastle upon Tyne, United Kingdom Background: The PARP inhibitorolaparib (Lynparza) is being evaluated in combination with other anticancer agents. We investigated the steady-state (ss) PK effects and safety of olaparib (tablet) co-administered with the antihormonal agents tamoxifen, anastrozole, or letrozole. Methods: An openlabel, nonrandomized study was conducted in 79 eligible patients with advanced solid tumors. During three consecutive treatment periods, patients received: (1) olaparib 300 mg twice daily (bd) for 5 days, followed by a 4-day washout period; (2) either tamoxifen 60 mg once daily (qd) for 4 days (loading dose) then 20 mg qd to day 26 (n=30), or anastrozole 1 mg qd to day 19 (n=23), or letrozole 2.5 mg qd to day 38 (n=26); (3) same as treatment period 2 concomitantly with olaparib 300 mg bd for 5 days. Blood samples for PK analysis were taken at ss. Safety was monitored throughout. Results: Olaparib AUC0-t and Cmax decreased by 27% and 20%, respectively, when co-administered with tamoxifen. Anastrozole and letrozole had no relevant impact on ss exposure to olaparib. Olaparib had minimal effects on ss exposure to tamoxifen, anastrozole or letrozole. Most AEs were of mild or moderate severity and as expected in this patient population. Conclusions: Exposure to olaparib decreased slightly when given with tamoxifen; tamoxifen exposure increased slightly when given with olaparib. These effects were considered unlikely to be clinically significant and no dose adjustments for either drugare required. There were no drug interactions between olaparib and letrozole or anastrozole. Safety data were consistent with the known safety profiles of study drugs. Clinical trial information: NCT02093351. Cmax GLS mean ratio

2561

Poster Session (Board #261), Sun, 8:00 AM-11:30 AM

A first-in-patient phase I study of BGB324, a selective Axl kinase inhibitor in patients with refractory/relapsed AML and high-risk MDS. First Author: Sonja Loges, Department of Oncology, Hematology, BMT with section Pneumology & Institute of Tumor Biology, University Medical Center HamburgEppendorf, Hamburg, Germany Background: The RTK Axl represents an independent prognostic factor and therapeutic target in AML. BGB324 is an orally available selective potent Axl inhibitor, which has antileukemic activity in preclinical models of AML. Methods: A standard 3 + 3 dose escalation study was performed to identify the maximum tolerated dose of BGB324 in patients with previously treated AML or high risk MDS. BGB324 was administered as an oral loading dose on days one and two followed by a reduced daily maintenance. Three dose levels were explored 400/100mg, 600/200mg and 900/300mg Results: To date, 13 patients with AML and 3 patients with MDS were treated in the first three dose levels. 6 patients remain on treatment and 11 are evaluable. Therapy has been well-tolerated and the MTD has not yet been reached. The majority of adverse events reported have been Grade 1 and 2. The most common related adverse events are diarrhea and fatigue. One patient with pre-existing cardiac conduction abnormalities discontinued study treatment because of prolonged QTc prolongation leading to an expansion of the first dose cohort to six patients. The use of a loading dose achieved steady state between three and six days which was maintained throughout treatment. One AML patient has an ongoing CRi . 5 months, one patient achieved clearance of circulating blasts accompanied by peripheral blood count recovery . 3 months and three AML patients experienced disease stabilisation for more than four months. The status of patients with MDS remained unchanged. Evidence of target inhibition in leukemic blasts was demonstrated by almost complete inhibition of Axl phosphorylation accompanied by reduction in phosphoErk and phosphoAkt signalling at day 21 of treatment. Conclusions: BGB324 can be safely administered for prolonged periods at doses that abrogate AXL signalling. Treatment with BGB324 exhibits anti-leukemic activity. Clinical trial information: NCT02488408.

2563

Poster Session (Board #263), Sun, 8:00 AM-11:30 AM

Dosing de novo two-drug combinations based on 32,894 patients in phase IIII clinical trials. First Author: Mina Nikanjam, UCLA, Los Angeles, CA Background: Metastatic cancers harbor complex genomic alterations. Combinations are needed to attenuate resistance. Knowledge of appropriate starting doses for novel drug combinations in clinical trials and practice is lacking. Literature review was used to help inform selection of safe starting doses for two-drug combinations involving a targeted agent. Methods: Phase I-III adult oncology clinical trial publications (Jan. 1, 2010 to Dec. 31, 2013; PubMed) were reviewed. The dose of drug used in each combination was compared to the single agent recommended dose (FDA-approved/ recommended phase 2 dose (RP2D)/maximum tolerated dose (MTD)). Dose percentages were calculated as: (safe dose of drug in combination/dose of drug as single agent at FDA/RP2D/MTD) X 100. Additive dose percentages were the sum of the dose percentage for each drug. Results: 144 studies (N = 8,568 patients; 95 combinations) of dual targeted agents and 372 studies (N = 24,326 patients; 248 drug combinations) with a targeted and cytotoxic agent were analyzed. Each agent could be administered at 100% of their standard dose in 51% (two targeted agents) and 38% of trials (cytotoxic and targeted agent), respectively. Median and lowest safe additive dose percentages are summarized in the Table. Dual targeted agent dosages were limited by overlapping targets, same drug class, and mTor inhibitors (lowest safe additive dose percentage of 60%), while targeted-cytotoxic combinations were limited by the presence of a PARP or HDAC inhibitor (lowest additive percentage of 41%). In the absence of these factors, the lowest safe additive dose percentage was 143% (dual targeted agents) and 82% (targeted-cytotoxic). Conclusions: The current findings can help inform safe starting doses for novel two-drug combinations in the context of clinical trials and practice.

AUC0-t GLS mean ratio

Comparison

Reference boundary*

Point estimate

90% CI

Point estimate

90% CI

T + O vs O A + O vs O L + O vs O T + O vs T A + O vs A L + O vs L

0.7–1.43 0.7–1.43 0.7–1.43 0.7–1.43 0.8–1.25 0.8–1.25

0.80 0.94 1.09 1.13 0.90 0.94

0.71–0.90 0.84–1.04 0.99–1.21 1.06–1.22 0.84–0.97 0.91–0.98

0.73 0.89 1.15 1.16 0.86 0.95

0.63–0.84 0.76–1.05 1.07–1.25 1.11–1.21 0.80–0.93 0.91–0.99

129s

Lowest Safe Additive Dose Percentage Lowest Safe Additive Dose Percentage with One Agent at 100% of Single Agent Recommended Dose Median Safe Additive Dose Percentage (Range)

Dual Targeted

Targeted-Cytotoxic

60% 143% (for non-overlapping targets/different classes) 125%

41% 82% (without HDAC/PARP inhibitors)

200% (60% – 200%)

117% (cytotoxic agent 100%) 136% (targeted agent 100%) 180% (41% - 200%)

*90% CIs falling within this boundary indicate lack of interaction. Abbreviations: O, olaparib; T, tamoxifen; A, anastrozole; L, letrozole; GLS, geometric least squares.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

130s 2564

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics Poster Session (Board #264), Sun, 8:00 AM-11:30 AM

Topical calcitriol (BPM31543) for the prevention of chemotherapy-induced alopecia (CIA): Efficacy findings from a phase I safety study. First Author: Mario E. Lacouture, Memorial Sloan Kettering Cancer Center, New York, NY Background: Cancer patients receiving taxane-based chemotherapy develop alopecia, which may lead to significant psychosocial, quality of life, and adherence issues. Besides a recently FDA-cleared scalp cooling device, there are no oral/ topical agents available prevent CIA. In murine studies, topical calcitriol reduced CIA, likely due to arrest of cell cycle in healthy hair follicles, and reduction in the sensitivity of follicular epithelium to chemotherapy. Methods: Up to 31 adult women with cancer (breast, gynecologic, soft-tissue/bone sarcoma) applied 1mL of BPM31543 to scalp bid, $ 5 days prior to initiation of taxane-based chemotherapy until 3 months or end of chemotherapy, in this dose-escalation study (5/10/20/40/60/80mg/mL cohorts). Periodic PK analysis, adverse event (AE) monitoring, patient self-assessment diaries (1-10 scale), and blinded photographic assessments were done. Results: Twenty-one subjects have been enrolled so far (evaluable, n = 13). PK data (n = 16; 5-40mg/mL) showed inter-individual variability, but no significant dose-dependent increase in systemic absorption (range, , 20-110 pg/mL). Treatment-related AEs (probably/possibly) were mild/ moderate, and included scalp pain (n = 1; 5 mg/mL), elevated vitamin D levels in 1 patient (20mg/mL) and with passage of renal calculus in another (n = 1; 40mg/ mL). All subjects reported changes in overall fullness, thickness, and volume of hair during chemotherapy, and experienced hair loss, with 85% showing a $ 75% increase in their rating (maximum, week 4) at the 5/10 mg/mL dose level. At the $ 20 mg/mL dose level, $ 75% hair loss was seen only in 43%. Hair loss/ thinning caused all subjects to change their hair style (onset, week 2; peak, weeks 5-6). Conclusions: Topical BPM31543 is safe and tolerable with potential activity at higher dosing levels. Clinical trial information: NCT01588522. Prevention of CIA with Topical Calcitriol: blinded photographic assessment of hair loss (7 sites). Maximal hair loss at any scalp site (Week 7 assessment vs. Baseline) -3

Dose level (mg/mL), n = 12 5

10

[Greatly decreased, . 75%]

-2

20

40

#009

#018, #019

#001

#007

#010

#002, #003

#004

#012, #014

2565

Poster Session (Board #265), Sun, 8:00 AM-11:30 AM

Population pharmacokinetic (PopPK) and exposure-response (ER) modeling of cabozantinib (C) in patients (pts) with renal cell carcinoma (RCC) in the phase 3 METEOR study. First Author: Steven Lacy, Exelixis, Inc, South San Francisco, CA Background: PopPK and ER models were developed to characterize the C concentration-time profile and the relationship between C exposure and efficacy endpoints in pts with RCC. Methods: The PopPK model was developed using nonlinear mixed effects modeling methodology (NONMEM v7.3) and incorporated 1650 measurable PK concentrations from 63 healthy volunteers (HV) and 282 pts with RCC. An extensive PK sampling strategy was employed for HV whereby C was administered as a single 20, 40 or 60 mg dose, while a single steady-state sample was collected at Day 29 and Day 57 for RCC pts initially receiving 60 mg C qd. Time-to-event Cox proportional hazard ER models were developed to characterize the relationship between various individual predicted C exposure measures and the relative hazard for specified efficacy endpoints including progressionfree survival (PFS). Results: A 2-compartment model with first-order elimination and two parallel (fast and slow), lagged first-order absorption processes adequately described the C concentration data. The oral clearance, terminal phase volume of distribution, and terminal half-life were estimated to be 2.23 L/hr, 319 L, and 99 hours, respectively, for a White male subject. Inter-individual variability of CL/F was estimated to be 46% CV. A statistically significant relationship was identified between the rate of progressive disease (PD) or death and the time-varying average daily C concentration. Increases in C concentration were predicted to decrease the rate of PD or death in a nonlinear manner. The best Emax model (p , 0.001) was achieved with EC50 = 100 ng/mL, a value below the typical individual predicted steady-state average concentration at 60 (1125 ng/mL) or 20 (375 ng/mL) mg/day. Conclusions: Based on the ER analysis, C was efficacious at all doses evaluated in METEOR, including those resulting from dose reductions, with higher exposures predicted to correlate with improved PFS. Clinical trial information: NCT02496208.

[Moderately decreased, 50-75%] -1

#016

[Slightly decreased, 25-49%]

2566

Poster Session (Board #266), Sun, 8:00 AM-11:30 AM

Phase I study of the PI3K/mTOR inhibitor gedatolisib (PF-05212384) in combination with docetaxel, cisplatin, and dacomitinib. First Author: Zev A. Wainberg, Department of Medicine, University of California, Los Angeles, CA Background: Inhibition of phosphatidylinositol-3-kinase (PI3K)-mediated signaling may overcome resistance to chemotherapy or human epidermal growth factor receptor (EGFR) inhibitors. Gedatolisib (G) is a dual inhibitor of PI3K/mammalian target of rapamycin (mTOR) in development for solid tumors. Methods: This is an ongoing study to determine the maximum tolerated dose (MTD) and safety of G in combination with: 1) docetaxel: Arm A, in castrate resistant prostate cancer, advanced breast cancer (BC), or nonsmall cell lung cancer (NSCLC) patients (pts); 2) cisplatin: Arm B, in urothelial transitional cell cancer (TCC), triple negative BC (TNBC), NSCLC or ovarian cancer pts; 3) the pan-EGFR TKI dacomitinib: Arm C, in refractory HER2+BC, HER2+esophagogastric cancer, head and neck cancer, or EGFRmutated NSCLC pts. Pts received a G lead-in dose 7 days (A, B) or 14 days (C) prior to day 1, followed by weekly intravenous (IV) G plus: docetaxel or cisplatin (75 mg IV every 3 weeks); or oral dacomitinib (30–45 mg/d). Secondary objectives included pharmacokinetics (PK), clinical efficacy (partial or complete response [PR or CR]), and associated biomarkers. Results: 74 pts (median prior therapies: 2; range: 0–6) received 90–260 mg/ week of G in: A, 21; B, 21; and C, 32. The top drug related adverse events: A: neutropenia, mucositis, alopecia; B: nausea, fatigue, vomiting; and C: mucositis, diarrhea, nausea, fatigue. Cycle 1 dose limiting toxicities (DLTs) were: A: grade 3 mucositis (n = 1); B: no DLT; C: grade 3 mucositis (n = 1), pneumonitis (n = 1), rash (n = 1), and grade 2 fatigue (n = 1). PK parameters for G were similar regardless of combination agent. Best overall response was A: 4/21 PRs; B: 8/21 PRs (5 TNBC, 2 NSCLC, 1 TCC), and C: 6/32, (1 CR, 5 PRs). In Arm B, 3 of the 8 PRs (2 NSCLC, 1 TCC) had prior platinum exposure. Next generation sequencing involving PI3K data from a subset of patients is under evaluation. Conclusions: G can be combined with docetaxel, cisplatin, or dacomitinib, with manageable toxicity profiles and preliminary antitumor activity, even when heavily pre-treated. Due to portfolio prioritization, dose escalation and expansion continues in Arm B, with closure of Arms A and C (MTD not reached). Clinical trial information: NCT01920061.

2567

Poster Session (Board #267), Sun, 8:00 AM-11:30 AM

A phase 1 study of eribulin mesylate (E7389), a novel microtubule targeting chemotherapeutic agent in children with refractory or recurrent solid tumors (excluding CNS), including lymphomas: a Children’s Oncology Group Phase 1 Consortium study (ADVL1314). First Author: Eric Schafer, Baylor College of Medicine/Texas Children’s Hospital, Houston, TX Background: Eribulin mesylate, a synthetic analog of Halichondrin B, is a novel microtubule dynamics inhibitor. It inhibits microtubule growth without effects on microtubule shortening and promotes nonproductive tubulin aggregate formation. We performed a phase 1 trial to determine the dose limiting toxicities (DLT), maximum tolerated dose (MTD) and pharmacokinetics (PK) of eribulin in children with refractory or recurrent solid tumors, including lymphomas. Methods: Eribulin was administered intravenously on days 1 and 8 in 21-day cycles. Three dose levels (1.1, 1.4 and 1.8 mg/m2/ dose) were evaluated using the rolling-6 design with additional patients enrolled into a PK expansion cohort at the MTD/recommended phase 2 dose (RP2D). PK samples were obtained following the Day 1 dose of cycle 1. Results: To date, 23 patients ages 3-17 (median 14) years have been enrolled, with 21 fully evaluable for toxicity. Subjects enrolled with 10 unique tumor types the most common of which were osteosarcoma (n = 9) and Ewing sarcoma (n = 4). During the dose escalation phase, DLTs occurred in 0/6 and 1/6 subjects at 1.1 and 1.4 mg/m2/dose, respectively. At 1.8 mg/m2/dose 2/ 5 subjects experienced dose-limiting (grade 4) neutropenia. The DLTs, which occurred in a single patient at 1.4 mg/m2/dose, included Grade 4 neutropenia and Grade 3 fatigue. No further DLTs have been observed in the expansion cohort at 1.4 mg/m2/dose. Grade 3/4 non-DLT non-hematologic toxicities across all dose levels and cycles included increases in ALT and AST, anorexia, nausea and hypokalemia. No episodes of $ Gr 2 QTc prolongation or peripheral neuropathy were reported. Eribulin exposure was dose-dependent and elimination half-life was ~36hr. Conclusions: Eribulin was well-tolerated in children with refractory/recurrent solid tumors with neutropenia identified as the primary DLT. The RP2D of eribulin mesylate is 1.4 mg/m2/dose on days 1 and 8 of a 21-day cycle. Clinical trial information: NCT02171260.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics 2568

Poster Session (Board #268), Sun, 8:00 AM-11:30 AM

Incidence, severity and factors associated with diarrhoea in phase I oncology studies: experience of 1002 consecutive cases. First Author: Nikolaos Diamantis, The Royal Marsden/Institute of Cancer Research, Sutton, United Kingdom Background: Diarrhea is a common toxicity related to cancer treatment. Diarrhea occurs at different degrees of severity both within and outside the dose limiting toxicity window, influencing the dose and schedule of novel anticancer drugs. We aimed to assess the incidence, severity and factors predisposing to diarrhea in patients (pts) enrolled in phase I studies. Methods: This is a retrospective study of 1002 separate phase I trial pt-cases treated at the Royal Marsden Hospital Drug Development Unit between January 2010 and December 2014. Data were collected from our electronic records for consecutive pts participating in 80 different phase I trials. Results: The incidence of diarrhea was 32.1%. Grade 1-5 diarrhea was reported in 21.9%, 6.3%, 3.1%, 0.1% and 0.1% pt-cases respectively. 69.5% episodes of diarrhea were reported as drug related. 5.6% of pts had a specific trial intervention due to diarrhea such as dose interruption or withdrawal from study. Time on study was not significantly different in pts who developed diarrhea (2.06 vs 1.9 months p = 0.14). The incidence of diarrhea was influenced by the primary site of the tumor (abdominal vs non abdominal, 35.3% vs 27.1%, p = 0.01) and the primary organ site specifically, p , 0.001. Pts who had previous abdominal surgery (n = 472) or colostomies (n = 71) had significantly higher incidence of diarrhea compared to those who did not, 38.4% vs 27.3% p = 0.0002 and 49% vs 31.3% p = 0.003 respectively. The administration route, (po vs iv, 38.4% vs 27.3% p = 0.008), trial design, (single agent vs combination, 29.2% vs 48%, p , 0.0001) and study drug mechanism of action (PI3K inh. vs other, 36.6% vs 28.2%, p = 0.002) also had a significant effect. Previous pelvic radiotherapy did not significantly influence the incidence of diarrhea. Conclusions: Diarrhea is a significant toxicity reported in approximately one third of pts enrolled in phase I clinical trials. Factors predisposing to diarrhea included colostomies, abdominal surgery, administration route and trial design. These data from across a wide variety of phase I studies should be used as benchmark of the incidence of diarrhea to inform trial design, inclusion and exclusion criteria of future early phase studies.

2570

Poster Session (Board #270), Sun, 8:00 AM-11:30 AM

First-in-human study of HM95573, a novel oral RAF inhibitor, in patients with solid tumors. First Author: Yong Sang Hong, Department of Oncology, Asan Medical Center, University of Ulsan College of Medicine, Seoul, South Korea Background: HM95573 is a novel and selective RAF kinase inhibitor which showed potent anti-tumor activities in BRAF, KRAS and NRAS mutant model in vitro and in vivo. This phase 1 trial evaluated the safety, tolerability, pharmacokinetics (PK), and anti-tumor activity of HM95573 in patients with solid tumors. Interim results from the phase 1 study are presented here. Methods: Patients who had solid tumor with BRAF, KRAS or NRAS mutation were enrolled. HM95573 was administered everyday with 21-day cycle and treatment cycle has been repeated until disease progression or dose-limiting toxicity (DLT). Dose escalation was performed with PK-guided dose escalation method until target AUC achieved or the first DLT observed. Safety assessments include adverse events (AE), laboratory test, ophthalmologic and dermatological assessments, physical examination, vital signs, and ECG. Serial blood samples were collected on Cycle1Day1, Cycle2Day1, and Cycle3Day1 for PK analysis. Tumor responses were assessed every 6 weeks. Results: Total 31 patients were enrolled in 5 cohorts and doses of each cohort were 50mg QD, 100mg QD, 200mg QD, 200mg BID and 300mg BID. Mean terminal half-life of HM95573 was within the range of 50 to 80 hours after single dosing. However, dose regimen was changed from QD to BID to maximize drug exposure with current formulation. Tumor types from 31 patients were colorectal cancer (68%), melanoma (23%), lung cancer (3%), GIST (3%) and bladder cancer (3%); the BRAF (45%), KRAS (45%) or NRAS (10%) mutation. One patient at 200mg BID experienced grade 3 skin rash as a DLT and up to 300 BID, the maximum tolerated dose was not reached. HM95573 was well tolerated up to 300mg BID; Most of AEs were generally Grade 1 or 2 and most frequent drug-related AE was rash (17%). One patient with NRAS mutant melanoma (200mg QD) experienced an unconfirmed partial response at 12 weeks, and 9 patients had stable disease. Conclusions: Overall, HM95573 was well tolerated up to 300 BID as exposure increased. Preliminary evidence of anti-tumor activity has been observed from 200mg QD, however, current phase 1 study will explore higher doses to evaluate the therapeutic potential of HM95573 further. Clinical trial information: NCT02405065.

2569

131s

Poster Session (Board #269), Sun, 8:00 AM-11:30 AM

An open label, randomised cross-over bioavailability study of oral paclitaxel (oraxol) compared to intravenous paclitaxel 80mg/m2. First Author: Christopher G. C. A. Jackson, University of Otago, Dunedin, New Zealand Background: Paclitaxel has poor oral bioavailability due to the active excretion by p-glycoprotein (Pgp) present on intestinal epithelial cells. Oraxol (Kinex) is a combination of oral paclitaxel and HM30181, a novel, orally administered, nonabsorbed specific inhibitor of intestinal Pgp. Oral paclitaxel would avoid the need for IV injections, avoid severe anaphylactic reactions and the side-effects of steroid pre-medication, cost savings and greater patient convenience. We report the absolute bioavailability of Oraxol compared to IV paclitaxel. Methods: We conducted an open label, randomized cross-over pharmacokinetic (PK) study to determine the bioavailability (PK, AUC) and safety of escalating doses of Oraxol over 2 days compared to IV paclitaxel 80mg/m2 over 1 hour in patients with incurable cancer receiving weekly paclitaxel. Following written informed consent, patients were randomized to receive either Oraxol followed one week later by IV paclitaxel or the reverse sequence. Six patients (4 F 2 M) received two daily doses of 270mg. Following review of safety and PK results, two patients were treated with Oraxol 274mg/m2 x 2 days, and then two patients received 313mg/m2 x 2 days. Results: Oraxol was well tolerated with no grade 3/4 drug toxicities at any of the dose levels. AUC0-‘ and Cmax increased with increasing Oraxol dose from 270mg to 274mg/m2 but not with dose escalation to 313mg/m2. Conclusions: Oraxol 615 mg/m2 over 2 days achieved paclitaxel AUC comparable to IV paclitaxel 80 mg/m2. Oraxol safety profile was acceptable without grade 3-4 toxicities. Oraxol represents an ideal drug candidate to replace IV paclitaxel. Clinical trial information: 12614000365662. Oraxol concentration normalised to 615 mg/m2 compared to IV paclitaxel 80 mg/m2 for all subjects. AUC0- ¥ (ng.hr/ml) AUC0-t (ng.hr/ml) Cmax0-24 (ng/ml) Cmax24-48 (ng/ml) Cmax (ng/ml) T1/2 (hr)

Oraxol

Paclitaxel IV

8638.8068155.07 (3074.24 – 29773.62) 6336.4263408.91 (2928.44 – 12566.67) 572.596301.88 (330.37 – 1122.34) 448.156229.01 (292.74 – 913.91)

6824.8563181.03 (4039.04 – 13862.91) 6289.2362397.61 (3911.52 – 10531.10)

2492.476362.46 (2025.67 – 3127.89) 21.8568.17 (11.38 – 42.04)

48.78630.76 (21.31 – 128.32)

2571

Poster Session (Board #271), Sun, 8:00 AM-11:30 AM

A phase I study of afatinib combined with carboplatin or carboplatin plus paclitaxel in patients (pts) with advanced solid tumors. First Author: Fangfei Gao, Kings College London, Guys Hospital, London, United Kingdom Background: The irreversible ErbB family blocker, afatinib, has demonstrated acceptable tolerability and promising efficacy when combined with cytotoxic agents. Methods: In an open-label, 3+3 dose escalation study, we evaluated the maximum tolerated dose (MTD), safety, pharmacokinetics (PK) and antitumor activity of afatinib in combination with carboplatin (regimen A) or carboplatin + paclitaxel (regimen B). The starting dose was afatinib 20 mg/day + carboplatin AUC6 (Day 1 of a 3-weekly cycle) in both regimens. Paclitaxel 175 mg/m2 was given on Day 1 of each cycle in regimen B. Carboplatin was reduced to AUC5 in case of toxicity. Afatinib was to be escalated from 20 to 40 mg then 50 mg; if $2/6 pts experienced doselimiting toxicity (DLT) at 40 mg, an intermediate dose of 30 mg was assessed. Results: 12 pts were treated in regimen A (8 male; median age 58 years) and 26 in regimen B (13 male; median age 61 years). The table shows dose cohorts and DLTs in both regimens. Although MTD was not reached for regimen A, it was decided not to escalate the dose further. Overall, the most frequent (any grade) drug-related adverse events (AEs) were rash/acne (75%), fatigue (67%) and diarrhea (58%) in regimen A and diarrhea (88%), rash/acne (73%) and fatigue (69%) in regimen B. Afatinib did not appear to alter carboplatin or paclitaxel PK. Partial response was reported in 3 pts (25%) in regimen A and 5 (19%) in regimen B. Conclusions: MTD was not reached for regimen A but the recommended phase 2 dose (RP2D) was afatinib 40 mg + carboplatin AUC6. MTD/RP2D for regimen B was afatinib 20 mg + carboplatin AUC5 + paclitaxel 175 mg/m2. At RP2D, AEs were manageable with antitumor activity observed. Clinical trial information: NCT00809133. Afatinib (mg/day) Regimen A 20 40 Regimen B 20

Paclitaxel (mg/m2)/ carboplatin (AUC) All pts/evaluable pts/pts with DLT -/6 -/6

3/3/0 9/6/1

175/6

7/6/2

20 40

175/5 175/5

6/6/0* 5/5/2

30

175/5

8/6/2

DLT (all G3)

Dermatitis acneiform Fatigue/C. difficile infection/diarrhea/ small intestinal hemorrhage/ dehydration/renal impairment/ neutropenic sepsis; Mucosal inflammation Febrile neutropenia; Fatigue/mucosal inflammation Stomatitis; Mucosal inflammation

* No DLT in first 3 pts; cohort expanded to 6 pts after 40 and 30 mg dose exceeded MTD.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

132s 2572

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics Poster Session (Board #272), Sun, 8:00 AM-11:30 AM

Matching degree between PI3K/AKT/mTOR (PAM) pathway mutations (mut) and therapy (ttx) as predictor of clinical benefit (ClinBen) in early trials. First Author: Juan Martin Liberal, Vall dHebron University Hospital, Barcelona, Spain Background: Drugs targeting aberrations in the PAM pathway are in clinical development but it is not completely clear if mut and ttx matching increases chances of ClinBen. Methods: Retrospective analysis of patients (pts) with molecular tumor profile treated with single agent PAM inhibitors (inh) in early trials at VHIO. Matching degrees were: supermatched (mut in PAM treated with a specific selective inh, such as PIK3CA mut - PI3Ka inh), matched (mut in PAM treated with a non-selective inh, such as PIK3CA mut AKT inh) and unmatched(PAM inh without PAM mut). ClinBen was defined as partial response or stable disease for . 4 months (m). Time to treatment failure (TTF) was defined as time between start of ttx and end for any cause. Results: From Jul 09 to Dec 15, 97 pts with known molecular tumor profile by Sequenom (82 pts) or Amplicon (15 pts) received 99 PAM inh. Median age 58 years (23-77); 60 F, 37 M. Tumor types were colorectal (CRC) 42%, breast (BC) 20%, gynecological (GYN) 19%, other 10%. Median prior ttx lines was 2 (0-8). Median time from diagnosis of metastatic disease until PAM inh was 28 m (CI95% 23-34). Most pts (77%) had $ 75% dose recommended. Ttx was matched to PAM mut in 70 pts. Of those, 59 were supermatched. PIK3CA mut were the most common (65%). ClinBen was seen in 24% pts (supermatched 30%, matched 18% and unmatched 13%). Supermatched had a better trend for ClinBen compared to matched (OR 2.0, p = 0.49) or unmatched (OR 2.7, p = 0.11). ClinBen was significantly lower (12%) if coexisting RAS mut (OR 0.3, p = 0.05) and higher in BC (45%, OR 5.8, p = 0.007) and GYN tumors (35%, OR 3.9, p = 0.04) compared to CRC (11%). In a multivariate regression, factors influencing ClinBen were supermatched ttx (OR 4.3, p = 0.10 vs matched; OR 4.87, p = 0.02 vs unmatched) and CRC primary (lower benefit, OR 0.16, p = 0.02). Median TTF was 2 m (CI95% 1.9-2.4). TTF was associated with matching degree (supermatched vs matched HR 0.3, p = 0.002; supermatched vs unmatchedHR 0.7, p = 0.2) and tumor type (BC vs CRC HR 0.42, p = 0.02; GYN vs CRC HR 0.49, p = 0.06) in a multivariate Cox model. Conclusions: Our results suggest that single agent PAM inh have the highest chance of ClinBen and longer TTF when given as supermatchedttx in nonCRC pts.

2574

Poster Session (Board #274), Sun, 8:00 AM-11:30 AM

2573

Poster Session (Board #273), Sun, 8:00 AM-11:30 AM

Phase I trial of paclitaxel, bevacizumab, and temsirolimus in advanced solid malignancies. First Author: Shannon Neville Westin, The University of Texas MD Anderson Cancer Center, Houston, TX Background: We sought to determine the MTD/recommended phase II (RP2D) dose of the combination of paclitaxel (P), bevacizumab (B), and temsirolimus (T) in patients (pts) with advanced solid tumors. Methods: Doses were escalated across 14 planned dose levels (DL) in a standard 3+3 design (Table). Responses were defined using RECIST 1.0. Results: To date, 61 pts have been enrolled. Four pts withdrew consent after one dose. Dose-limiting toxicities (DLTs) occurred at DL8 (G3 fatigue), DL10 (G3 hypertension, G4 pain), DL11 (G3 mucositis, G4 thrombocytopenia/neutropenia) and DL12 (G3 hyperglycemia). DL10 is being explored in expansion as the RP2D. Most common adverse events ( . 10%) were fatigue (74%; G3/4 5%), hypertriglyceridemia (49%, G3/4 5%), thrombocytopenia (38%, G3/4 7%), nausea (33%, G3/4 0%), mucositis (31%, G3/4 5%), constipation (30%, G3/4 0%), neutropenia (21%, G3/4 8%), diarrhea (20%, G3/4 0%), rash (16%, G3/4 0%), hypertension (13%, 02%), headache (10%, G3/4 2%), and vomiting (10%, G3/4 0%). Of 51pts evaluable for response, six had PR (12%) including NSCL (-64%, 13 months (m)), breast (-42%, 8m; -60% 3m), ovarian (-30%, 6m;), squamous cell cervical (-68%, +4m), and squamous cell tongue (-34%, 2m) cancer. Twenty pts (39%) had SD for 4 m or more including adenoid cystic (-3%, 10m), breast (-28%, 4m; +6%, 10m; +18%, 12m; -25%, 8m; -23%, 9m; -22%, 11m; +6%, 17m; +10%, 7m), chondrosarcoma (+8%, 19m; -19%, 6m), laryngeal/oropharyngeal (-19%, 5m; +12%, 7m), leiomyosarcoma (-10%, 38m; -10%, 8m), endometrial (-2%, 8m), gastric (-10%, 8m), ovarian (-20%, 18m) and renal (-10%, 7m) cancer. Conclusions: The combination of P, B, and T is well tolerated and demonstrates preliminary evidence of tumor activity in a variety of solid tumors, especially breast, gynecologic and squamous cancers. Studies to determine correlation of activity with tumor molecular profile are ongoing. Clinical trial information: NCT01187199.

Dose-escalation schedule. DL 1* 2* 3* 4* 5* 6* 7* 8* 9** 10** 11* 12** 13** 14**

B mg/kg, IV

T mg IV

P mg/m2 IV

5 Q3W 7.5 Q3W 7.5 Q3W 7.5 Q3W 10 Q3W 10 Q3W 10 Q3W 15 Q3W 10 Q2W 10 Q2W 15 Q3W 10 Q2W 10 Q2W 10 Q2W

12.5 12.5 20 20 20 25 25 25 25 25 25 25 25 25

30 Q3W 30 Q3W 30 Q3W 60 Q3W 60 Q3W 60 Q3W 80 Q3W 80 Q3W 30 QW x 3 40 QW x 3 120 Q3W 50 QW x 3 60 QW x 3 80 QW x 3

*21d cycle, ** 28d cycle

2575

Poster Session (Board #275), Sun, 8:00 AM-11:30 AM

A phase 1 dose escalation study of RXDX-105, an oral RET and BRAF inhibitor, in patients with advanced solid tumors. First Author: Manish R. Patel, Sarah Cannon Research Institute/Florida Cancer Specialists, Sarasota, FL

A first in human phase I study of receptor tyrosine kinase (RTK) inhibitor MGCD516 in patients with advanced solid tumors. First Author: Todd Michael Bauer, Sarah Cannon Research Institute, and Tennessee Oncology, PLLC., Nashville, TN

Background: RXDX-105 is a multikinase inhibitor that has demonstrated potent inhibition of RET. RXDX-105 is also active against BRAF. RET alterations are associated with the development of various types of cancer. Acquired BRAF mutations can result in constitutive activation of the MEK/ ERK signaling pathway, which fuels cancer growth. Methods: Pts with advanced solid tumors were enrolled in a Phase 1 dose escalation study with a standard 3 + 3 design to determine the recommended Phase 2 dose (RP2D) of RXDX-105, administered once daily on a continuous dosing schedule. Tumor response was assessed every 8 wks (RECIST v1.1). Treatmentemergent adverse events (AEs) were recorded according to NCI CTC v4.03. Pharmacokinetic (PK) analysis was performed. Results: To date, 45 pts (21 m and 24 F) received RXDX-105 across 8 dose levels (20 to 350 mg QD). Median age was 60 years (range 27–81). Median number of cycles was 2 (range 1 to 30). The PK data demonstrate an RXDX-105 half-life of 28 to 42 hrs. At 275 mg fed state, exposure reached the predicted efficacious concentration (Ceff ) based on preclinical data of RET and BRAF inhibition. The most common AEs were: fatigue (19 pts; 42%), vomiting (16 pts; 36%), nausea (16 pts; 36%), rash (15 pts; 33%), and constipation (12 pts; 27%). 4 DLTs occurred: G3 maculopapular rash (n = 1; 200 mg), G3 fatigue (n = 1; 275 mg), G3 diarrhea (n = 1; 275 mg fed state), and G3 hyperbilirubinemia (n = 1; 350 mg fed state). All DLTs resolved with dose hold. 2 SAEs were considered treatment-related: G2 headache, which occurred during hospitalization for disease progression, and G3 hyperbilirubinemia. No treatment-related deaths occurred. 1 durable PR was observed and disease shrinkage was noted in another 3 patients, 1 of whom has BRAF V600Emutant papillary thyroid cancer and continues on study after over 2 years with SD. Conclusions: In this ongoing study, the predicted efficacious exposure has been achieved and several patients have experienced clinical benefit. The safety profile to date indicates that RXDX-105 is tolerable at doses above the projected Ceff threshold. Dose escalation is ongoing. A Ph 1b basket study in pts with RET or BRAF alterations will commence after RP2D selection. Clinical trial information: NCT01877811.

Background: MGCD516 is an oral, potent small molecule inhibitor of a closely related spectrum of RTKs including RET, TRK family, DDR2, MET, AXL and split RTKs (VEGFR, PDGFR and KIT). MGCD516 has demonstrated antitumor activity in nonclinical cancer models harboring genetic alterations of MGCD516 targets including rearrangement of RET or NTRK or CHR4q12 amplification. Methods: Study objectives include evaluation for safety, pharmacokinetics (PK), pharmacodynamics (PD), the maximum tolerated dose (MTD) and clinical activity of MGCD516 in patients (pts) with advanced solid tumors. Eligible pts received a single dose for PK profiling followed by continuous daily dosing (QD) in 21 day cycles. Results: 32 unselected pts (14 men/18 women; median age 62 years; range 27-85) with advanced solid tumors were treated in escalating dose cohorts of 10, 20, 40, 80, 110, 150 or 200mg MGCD516. At 80mg, 1 of 6 evaluable pts experienced a DLT (Grade 3 palmar plantar erythrodysesthesia). At 200mg, 3 DLTs were observed among 3 evaluable pts (intolerable Grade 2 neuropathy, fatigue and stomatitis in 1 pt each), demonstrating 200mg exceeded the MTD. Treatment-related AEs ( . 15% of pts; Grade1-3) included hypertension, fatigue, diarrhea, nausea and decreased appetite. One treatment-related Grade 4 AE (febrile neutropenia) was reported. Prolonged stable disease (SD) has been observed in multiple pts including 3 pts with at least 17 weeks SD and 1 pt with 35 weeks SD. Preliminary PK data show that exposure increased dose proportionally with doses up to 200mg. At 150mg, the Cavg and AUC0-24 values at steady state (90.7 ng/mL and 2.18 mg$h/mL, resp.) exceed plasma exposure projections required for inhibition of key RTK targets and antitumor efficacy in nonclinical tumor models. Preliminary clinical PD data indicate dose dependent inhibition of the VEGF and MET pathways. Conclusions: The Phase 1b dose for MGCD516 was established at 150mg QD. MGCD516 shows favorable PK characteristics, on-target PD effects and is associated primarily with constitutional or GI-related AEs. Phase 1b enrollment began November 2015. Pts with NSCLC or other solid tumors with specific genetic alterations in MGCD516 target RTK genes are being enrolled. Clinical trial information: NCT02219711.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics 2576

Poster Session (Board #276), Sun, 8:00 AM-11:30 AM

Dose escalation study of ODM-203, a selective dual FGFR/VEGFR inhibitor, in patients with advanced solid tumours. First Author: Jordi Rodon Ahnert, Medical Oncology, Vall d’Hebron University Hospital and Universitat Autonoma de Barcelona, Barcelona, Spain Background: Inhibitors of fibroblast growth factor receptors (FGFR) are being developed for treatment of solid tumours with FGFR genetic alterations. ODM-203 is a small molecule with balanced inhibitory effects on both FGFR 1-4 and VEGFR 1-3 subtypes. We present the results of the firstin-man study. Methods: ODM-203 was evaluated in a 3+3 dose escalation design to identify the maximum tolerated dose (MTD), dosed daily in a 4 week cycle. PK profiles were taken on days 1 and 8, and safety labs, ECG, vital signs and adverse events were recorded on days 1, 4, 8, 15, 22, 29, week (wk) 6 and wk 8. Plasma markers of FGFR and VEGFR inhibition were measured at baseline, day 8 and day 29 and tumour response by RECIST criteria every 8 weeks. Patients continued ODM-203 treatment until disease progression or dose limiting toxicity (DLT). Results: 24 patients (median age 55, range 28-80 years) with advanced solid tumours received ODM-203 in doses between 50-800mg once/day. The longest treatment period so far is 40 wks (cholangiocarcinoma with FGFR fusion). Patients had various primary tumours, some having FGFR alterations. One DLT (corneal keratopathy) was recorded at 800mg/day but MTD was not identified. A single grade 4 AE (lipase increase) was reported. Most AE’s reported were G1-2, the most common being increased bilirubin (63%, G3 in 11 (46%) pts), diarrhoea (46%, G3 in 2 (8%) pts), alopecia (33%), jaundice (33%, G3 in 1 (4%) pt), stomatitis (29%) and increased phosphate (25%). Increased bilirubin was due to UGT1A1 inhibition by ODM-203 and resolved upon dose reduction/ discontinuation. After repeated dosing ODM-203 exposure increased more than dose-proportionally. Steady state was generally not reached by day 8. There were 2 partial responses (RCC and FGFR mutated salivary gland cancer) and 5 further patients achieved stable disease for at least 16 wks. Conclusions: ODM-203 elicited on-target adverse effects in addition to bilirubin changes. There was preliminary evidence of promising anti-tumour activity although FGFR-aberrant tumours were not specified. Further evaluation is ongoing at 600mg/day and will continue in an expansion study in patients with FGFR-aberrant tumours. Clinical trial information: NCT02264418.

2578

Poster Session (Board #278), Sun, 8:00 AM-11:30 AM

2577

133s

Poster Session (Board #277), Sun, 8:00 AM-11:30 AM

Pharmacokinetic (PK)-driven individualization of pazopanib therapy in patients with solid tumors: A phase 1 study (MC1112). First Author: Keith Christopher Bible, Mayo Clinic, Rochester, MN Background: Pazopanib (Pz) is a kinase inhibitor approved in renal cell carcinoma and soft tissue sarcoma (800 mg/day fasting) with activity also in other cancers. We found a strong correlation between attained Pz blood concentrations (CPz ) and RECIST PR in a prior study in thyroid cancer; no patients (pts) in the lowest CPz quintile attained PR (Lancet Oncol. 2010;11: 962-72). Methods: We posited that the proportion of pts attaining target steady state CPz (T-CPzS , . 30 mg/mL), and perhaps the proportion with potential to respond to therapy, might be increased by PK-guided Pz dosage modification. CPzS was assessed day 14 of each 28 day cycle. To increase CPzS , dosing was modified from fasting to fed day 28 in sub-T-CPzS pts, with escalation to 1000, then 1200, mg/day fed in later cycles in response to persistently sub-T-CPzS , if tolerated. Sample size of 54 gave 90% power (1sided, 5% significance level) to have at least 9 sub-T-CPzS pts; if . 3 of the 9 or 26.4% sub-T-CPzS pts convert to T-CPzS in later cycles, this strategy would be considered promising. Results: Accrual was completed, with 54 pts enrolled and 48 evaluable. Median age = 60 (range 28-84); median cycles = 2 (range 1-26). Of 48 evaluable pts, 15 (31.25%) attained sub-T-CPzS cycle 1; 4 (26.7%) of 15 attained T-CPzS upon dosing individualization in later cycles, meeting preset criteria to consider results promising. Eight pts with sub-T-CPzS altered dosing from fasting to fed, achieving 85.5% relative increase in CPzS . Failure to attain T-CPzS among the 15 initially sub-target pts was primarily due to cancer progression (7 pts, 46.7%). Of 48 evaluable pts, 6 attained PR (12.5%). CPzS was higher (mean 50.4 mg/mL [95% CI, 25.7 – 75.1] vs 41.7 mg/mL [95%CI, 11.1 – 72.2]) among those attaining PR. Of 54 enrolled, 3 (5.6%) had . 1 grade 4+ non-hematologic AE at least possibly related to study medication, including 1 death (pneumonitis, possibly related). Conclusions: Pz as conventionally administered results in sub-T-CPzS in . 30% of pts, with potential to be remediated by PK-guided dosage individualization. We next propose to assess the effects of this approach on response rates in kidney and thyroid cancers. Support: UM-1 CA186686-03 (NCT01552356). Clinical trial information: NCT01552356.

2579

Poster Session (Board #279), Sun, 8:00 AM-11:30 AM

Phase I pharmacokinetic study of single agent trametinib in advanced cancer patients with hepatic dysfunction: An NCI Organ Dysfunction Working Group (ODWG) study. First Author: Anna Spreafico, Princess Margaret Cancer Centre, Toronto, ON, Canada

Veliparib (ABT-888) extended-release formulations: A phase 1 study on safety, pharmacokinetics (PK), and bioavailability in patients with advanced solid tumors. First Author: Theresa Louise Werner, University of Utah Huntsman Cancer Institute, Salt Lake City, UT

Background: Trametinib (T), an orally available MEK 1/2 inhibitor, is FDA approved for BRAFV600E/K mutant advanced melanoma, at the single agent RP2D of 2 mg QD. Its clinical development has involved a variety of refractory cancers. A phase I study was conducted to establish the maximum tolerated dose (MTD), dose-limiting toxicity (DLT) and pharmacokinetic (PK) profile of T in advanced solid tumor patients (pts) with varying degrees of hepatic dysfunction (HD). Methods: Advanced cancer pts with ECOG #2, adequate renal and bone marrow functions, were stratified (NCI ODWG Criteria) into 4 HD groups: normal (NO), mild (ML), moderate (MD), severe (SV). T was given QD on a 28days cycle, with dose escalation occurring in sequential cohorts of pts within each HD group (starting dose: NO, ML: 2mg; MD: 1.5 mg; SV: 1mg). PK samples were collected at days 15-16 in cycle 1. Results: Between Mar ’14-Nov ’15, 30 (17 NO, 7 ML, 3 MD, 3SV) pts were treated. Median age 60, male 43%, ECOG#1: 97%. Common tumor types: GI non-CRC 17%, NSCLC 17%, uveal melanoma 13%. All cohorts drug-related AEs (All Grades/Grade 3-4): skin rash 73%/7%, increased transaminases 43%/7%, fatigue 40%/ 7%. No DLTs so far occurred in ML, MD and SV groups. Dose delay and dose reduction occurred in 47% and 20% of all pts respectively. PK data were available in 10 NO and 6 ML HD cohort pts (Table). Of 23 pts evaluable for response, 13 (56%) had stable disease and 2 (9%) pts partial response (NSCLC; ovarian). Median duration of response was 3.6 months (range 1-13.7+). Conclusions: The RP2D for trametinib in ML group, which has completed accrual, is 2 mg QD. The PK profile in NO and ML groups is comparable. Thus far, T appeared to be tolerated within all groups. Accrual in MD and SV groups is ongoing at 1.5 mg QD and 1 mg QD respectively. Clinical trial information: NCT02070549.

Background: Veliparib is a potent oral PARP1/2 inhibitor. Monotherapy trials with immediate release formulation veliparib (IR-v) showed antitumor activity in BRCA+ cancers. An extended release formulation (ER-v) may lower the peak-to-trough concentration ratio while maintaining exposure and improving tolerability, providing a rationale to conduct a phase 1 study (NCT01853306) of ER-v in pts with advanced tumors. Methods: Eligible pts had metastatic BRCA+ cancer, BRCA6 high-grade serous ovarian/primary peritoneal/fallopian tube cancer (OC), (part 1/2), or BRCA+ breast cancer (BC), or OC (part 3). Part 1 – PK assessment: pts received ER-v (3 different ER-vs tested) or IR-v 200 mg in fed/fasted state on days 1/3/5 (3-group, single-sequence crossover). Part 2 – 3+3 dose escalation: pts received ER-v daily (QD or BID) starting at 200 mg. Part 3 – safety expansion: ER-v was administered continuously at the RP2D. Primary objectives were to determine the bioavailability and food effect of ER-v vs IR-v, the MTD, and the RP2D/schedule of ER-v. ER-v and IR-v safety were also assessed. Results: As of 16 December 2014, 71 pts (96% female, 75% OC) received veliparib (part 1/2/3, n = 24/35/12). Pts were heavily pretreated and included platinum-refractory OC pts. ER-v had a longer Tmax (3.5 vs 1.7 h) and lower Cmax (629 vs 1471 ng/mL) than IR-v; AUC‘ was similar. The ER-v peak-totrough concentration ratio was low, and food intake moderately increased Cmax and Tmax. Three pts experienced DLTs: 2/5 pts treated at 600 mg BID ER-v (G2 asthenia and G3 N&V) and 1/3 pts treated at 800 mg QD ER-v (G3 seizure). The ER-v RP2D was determined to be 400 mg BID. G 3/4 TEAEs (part 1/2/3) occurred in 25/20/17% of pts; most common were thrombocytopenia (8/6/0%), small intestinal obstruction (4/0/8%), nausea (0/3/ 8%), and vomiting (0/3/8%). Gastrointestinal TEAEs were in line with historic IR-v data. Confirmed and unconfirmed responses were observed in pts with BC (6/16, 38%) and OC (8/44, 18%). Conclusions: ER-v was well tolerated in pts with advanced tumors and led to a reduced peak-to-trough ratio and more sustained exposure compared to IR-v. ER-v demonstrated single-agent antitumor activity. Clinical trial information: NCT01853306.

PK analysis for NO and ML groups.

AUC (ng*hr/ml) (SD) Cmax (ng/ml) (SD) Cmin (ng/ml) (SD) CL/F (L/hr) Tmax (hr) (range)

NO: 2mg QD (n=10) 90% CI

ML: 2 mg QD (n=6) 90% CI

507.5 6 189.0 (397.9 – 617.0) 30.37 6 12.15 (23.3 – 37.4) 16.30 6 6.47 (12.6 – 20.1) 4.51 6 1.85 (3.44 – 5.57) median: 2 (1-4)

387.7 6 201.3 (222.2 – 553.3) 29.53 6 15.88 (16.5 – 42.6) 12.14 6 6.95 (6.4 – 17.9) 6.20 6 2.82 (3.88 – 8.52) median: 1 (0-2)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

134s 2580

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics Poster Session (Board #280), Sun, 8:00 AM-11:30 AM

Investigational NEDD8-activating enzyme inhibitor pevonedistat (Pev) plus chemotherapy in patients (Pts) with solid tumors (Phase 1b study): Antitumor activity of pev plus carboplatin (Carbo)/paclitaxel (Pac). First Author: Todd Michael Bauer, Sarah Cannon Research Institute/Tennessee Oncology, PLLC, Nashville, TN Background: This is an open-label, multi-arm, dose-escalation study (NCT01862328) of pev (TAK-924/MLN4924) plus chemotherapy in pts with solid tumors. Safety and preliminary efficacy results have been reported (Lockhart et al, AACR-NCI-EORTC 2015); common adverse events in pts treated with pev + carbo/pac were fatigue (50%), nausea (38%), and G3 elevated AST/ALT (19%). We present previously unreported antitumor activity in pts treated with this combination. Methods: Twenty-six pts received pev IV 15, 20, or 25 mg/m2 on d 1, 3, and 5 with carbo AUC5 + pac 175 mg/ m2 on d 1 (lead-in cohort, n = 6: pev 15 mg/m2 + carbo AUC6) every 21 d. The combination appeared tolerable; MTD pev 20 mg/m2 + carbo AUC5 + pac 175 mg/m2. Response was assessed per RECIST v1.1; sparse PK sampling was carried out during C1. ERCC1 expression was assessed by IHC. Results: Nine responses were observed in 32 pts (28%) including those resistant to prior platinum/taxane therapy. Two pts (1 endometrial carcinoma, 1 bladder cancer) achieved a complete response (CR), and remain on study ( . 12 cycles). Seven pts (3 head and neck, 1 cholangiocarcinoma, 1 NSCLC, 1 sarcoma, 1 breast cancer) achieved a partial response (PR). The 9 responders received 4–18 cycles and had a median of 4 (range 1–8) prior therapies, including 4 with prior taxane (best response: 1 CR, 1 PR, 1 stable disease [SD], 1 unknown) and 8 with prior platinum (best response: 1 PR, 6 SD, 1 unknown). The majority had response within 2 cycles. Prolonged SD was seen in 3 pts (1 sarcoma, 1 endometrial cancer, 1 ovarian cancer), all 3 discontinued chemotherapy due to toxicity but continued to benefit from single-agent pev (SD duration 4.9–14.7 mos). Pev PK was unaffected in the presence of carbo/pac. Archival tumor sections from 23/30 pts were evaluable for ERCC1 staining. Conclusions: Pev+ carbo/pac showed broadbased antitumor activity in heavily pre-treated pts with solid tumors. Consistent with preclinical synergy between agents, lengthy treatment durations and improved responses in pts resistant to prior platinum/taxane therapy suggest the potential reversal of resistance by the addition of pev. Clinical trial information: NCT01862328.

2582

Poster Session (Board #282), Sun, 8:00 AM-11:30 AM

2581

Poster Session (Board #281), Sun, 8:00 AM-11:30 AM

A phase 1 study of the MDM2 inhibitor DS-3032b in patients (pts) with advanced solid tumors and lymphomas. First Author: Mrinal M. Gounder, Memorial Sloan Kettering Cancer Center, New York, NY Background: About 50% of tumors harbor inactivating mutations in TP53. Aberrant MDM2 signaling is an alternate mechanism to inactivate wild type (WT) p53. In preclinical studies, DS-3032b disrupted the MDM2-WTp53 interaction resulting in cell cycle arrest, senescence and/or apoptosis. We characterized the maximum tolerated dose (MTD), safety, tolerability, pharmacokinetic and pharmacodynamic (PD) profiles and efficacy of DS3032b in a phase 1 trial. Methods: Using accelerated and Bayesian design in the dose-escalation phase of this 2-part study (part 1), we escalated DS3032b from 15 to 240 mg orally once daily (QD) on a continuous or interrupted (QD 21/28 days) schedule. Results: Thirty-one of 34 pts enrolled in part 1 were evaluable for dose-limiting toxicities (DLTs). Median age was 59.5 years, 44% male, and 62% had $ 3 prior therapies. The majority of pts (79%) had wt TP53 and the most frequent tumor type was well/dedifferentiated liposarcoma (LPS; n = 15). MTD was 120 mg (interrupted) and 90 mg (continuous) schedule. The most common drug-related adverse events were hematologic and gastrointestinal. Six pts had DLTs, 5 were due to thrombocytopenia alone or with neutropenia. Of 26 pts evaluable for efficacy, none had an objective response but 77% had stable disease (SD); consistent with cell cycle arrest/senescence seen in preclinical studies. Tumor shrinkage was seen but did not meet RECIST criteria. Two pts with previously progressing LPS and carcinoid remain on study for 26+ mos and tumor sequencing showed MDM2 gene amplification. Exposures of DS3032b were significantly higher in pts with lower body weight. Induction of MIC- 1 (PD biomarker of WT p53 reactivation) correlated with drug exposure. Conclusions: DS-3032b has an acceptable safety profile at the MTD of 120 mg (QD 21/28). Clinical benefit with durable SD was seen in tumors with aberrant MDM2 signaling and WT p53. MDM2 is universally amplified in LPS and TP53 is the most commonly mutated gene in cancer; warranting further investigation of DS-3032b. Pts with melanoma or diffuse large B-cell lymphoma (DLBCL) are being enrolled in 2 dose-expansion cohorts (part 2) at the tentative recommended phase 2 dose 120 mg QD 21/28. Enrollment is ongoing (NCT01877382). Clinical trial information: NCT01877382.

2583

Poster Session (Board #283), Sun, 8:00 AM-11:30 AM

Updated efficacy and safety results from the phase I study of intermittent dosing of the dual MEK/RAF inhibitor, RO5126766 in patients (pts) with RAS/RAF mutated advanced solid tumours. First Author: Samuel John Harris, The Royal Marsden/Institute of Cancer Research, London, United Kingdom

Correlation of an Aurora A single nucleotide polymorphism (SNP) with clinical responses to alisertib in patients (pts) with advanced solid tumors. First Author: Huifeng Niu, Millennium Pharmaceuticals, Inc., a wholly owned subsidiary of Takeda Pharmaceutical Company Limited, Cambridge, MA

Background: RO5126766 is a novel MEK inhibitor with functional RAF inhibition. A dose escalation Phase 1 study, comparing 2 intermittent schedules, has previously been reported, with a recommended phase 2 dose (RP2D) of 4 mg twice weekly (Monday/Thursday (M/Th)). We now present results from the expansion cohort of this study, which aims to assess tolerability and preliminary activity in pts with KRAS, NRAS or BRAFmutated tumours. Methods: Pts with KRAS, NRAS or BRAF mutated tumours were treated with RO5126766 at the RP2D dose of 4 mg M/Th in 4 week cycles. Tumour assessment was performed with CT imaging every 2 cycles. Tumour pharmacodynamics was assessed in pre and post treatment biopsies where available. Results: To date, 18 evaluable pts (8 Non Small Cell Lung Carcinoma (NSCLC), 4 colorectal, 3 ovarian, 1 endometrial, 1 uterine carcinosarcoma, 1 melanoma) have been treated. Pts had received 1-7 (median 3) lines of treatment previously. Skin rashes were common G1-2 12pts, G3-4 4pts. Other adverse events were CK elevation, visual disturbances, peripheral oedema, mucositis, dry skin, diarrhoea and fatigue. Toxicities were manageable, however, dose reduction to 3.2 mg or 2.4 mg was required in 10 pts. Tumour size reduction was seen in 4/8 (50%) of pts with KRAS mutant NSCLC, of which 3/8 (37.5%) were RECIST partial responses (PR). The 8 NSCLC pts, had previously received 1-4 (median 2) lines of treatment, all having received platinum doublet chemotherapy. G12V, G12R and unknown codon 12/13 mutations were detected in responding NSCLC pts. RECIST PR was seen in 1 pt with KRAS mutated endometrial carcinoma and 1 pt with KRAS mutated ovarian cancer, whilst additionally, 1 pt with BRAF mutated ovarian cancer showed tumour size reduction. These two ovarian carcinoma pts had both received two separate MEK/BRAF inhibitors previously. Currently 5 pts overall remain on study and responses range from 4 -16.8 months. Recruitment is ongoing with plans to expand further in KRASmutated malignancies. Conclusions: These preliminary results, especially in KRAS mutant NSCLC pts who have been previously treated with standard of care chemotherapy, are encouraging. Clinical trial information: NCT02407509.

Background: Alisertib (MLN8237) is an investigational, oral, selective Aurora A kinase inhibitor. The Aurora A gene contains two functional SNPs (codon 31 and codon 57) that modify kinase activity. We assessed prognostic and predictive value of these SNPs. Methods: We did a correlative study of the Aurora A SNPs in The Cancer Genome Atlas (TCGA) database to evaluate their association with overall survival (OS), we then further assessed their potential association with clinical responses in 2 alisertib phase 2 trials. Whole blood samples were collected prior to alisertib study commencement and pt genotypes were analyzed by PCR. In study 1 (NCT01045421), single agent alisertib was investigated in pts with advanced solid tumors; Aurora A SNP profiles of 85 pts were obtained. In study 2 (NCT01091428), a randomized trial of alisertib plus weekly paclitaxel vs paclitaxel alone in pts with advanced recurrent ovarian cancers, Aurora A SNPs were genotyped in 122 pts (62 and 60 from each arm). The link between Aurora A SNPs and alisertib efficacy outcome was assessed. Results: The correlative TCGA study suggested prognostic significance for codon 57 SNP as all pts with solid tumors that carry VV and VI alleles had a significantly reduced OS compared with II homozygous pts [p , 0.0001, HR = 1.9 (VI) and 1.8 (VV)]. In study 1, pts with the VV allele at codon 57 (n = 53, 62%) had a significantly longer mPFS (p = 0.0213) in response to alisertib than pts with the IV and II alleles (n = 32, 38%). In study 2, pts with the VV allele at codon 57 receiving alisertib+paclitaxel (n = 47) had significantly improved mPFS (229 d) vs paclitaxel alone (n = 32; 117 d) (p = 0.0593, HR = 0.618). For the ITT population, improvement in mPFS was not statistically significant (230 vs 154 d) (p = 0.1534, HR = 0.74). In paclitaxel arm, pts with the VV variant had reduced mPFS vs ITT pts (117 vs 154 d), which was consistent with the VV variant being a poor prognostic biomarker as shown in TCGA. Higher kinase activity of Aurora A VV at codon 57 may explain the association with poor survival, chemo-resistance and better response to alisertib. Conclusions: Aurora A VV SNP at codon 57 may serve as a biomarker to predict disease outcome and response to alisertib. Clinical trial information: NCT01045421 and NCT01091428.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics 2584

Poster Session (Board #284), Sun, 8:00 AM-11:30 AM

Final report of a phase I study of veliparib (ABT-888) in combination with low-dose fractionated whole abdominal radiation therapy (LDFWAR) in patients with advanced solid malignancies and peritoneal carcinomatosis with a dose escalation in ovarian and fallopian tube cancers. First Author: Kim Anna Reiss, Johns Hopkins Univ, Batimore, MD Background: Combining low-dose radiation therapy with poly (ADP-ribose) polymerase (PARP) inhibition has been shown to enhance anti-tumor efficacy through potentiating DNA damage in tumor cells. We combined low dose fractionated whole abdominal radiation (LDFWAR) with escalating doses of veliparib (ABT888), an orally available, small molecule PARP inhibitor, in patients (pts) with peritoneal carcinomatosis from advanced solid tumor malignancies with a dose escalation in ovarian and fallopian cancer patients (OV). Methods: Pts were treated with escalating doses of veliparib, 40-400mg PO BID on days on days 1-21 of 3 28-day cycles on 6 dose levels. LDFWAR consisted of 21.6Gy in 36 fractions, 0.6 Gy twice daily on days 1 and 5 for weeks 1-3 of each cycle. Circulating tumor cells (CTCs) were serially collected. Quality of life (QoL) was assessed using the EORTC-QLQ-C30 questionnaire. Results: 32 pts were treated. Median follow-up was 45 months (10-50). The most common treatment-related gr3 and 4 toxicities were lymphopenia (59%), anemia (9%), thrombocytopenia (12%), neutropenia (6%), leukopenia (6%), nausea (6%), diarrhea (6%), anorexia (6%), vomiting (6%) and fatigue (6%). The MTD was determined to be 250mg PO BID (DL5). Median PFS was 3.6mo and median OS was 9.1mo. Survival on OV was not significantly longer than those with other cancers (9.4mo compared to 8.8mo, p = 0.885). QoL decreased for all groups during treatment, but was not seemingly worse at higher dose levels and the negative impact seemed to be less profound in OV patients. BRCA status was known for 14/18 patients with OV cancers. Conclusions: Veliparib plus LDFWAR is moderately well tolerated with gastrointestinal symptoms, fatigue and myelosuppression being the most common toxicities. In this very small study, BRCA mutation status did not appear to have an impact on PFS or OS of OV patients. DL5 (veliparib 250mg PO BID) is the RP2D for any future studies. Clinical trial information: NCT01264432. ALL PATIENTS (n = 32) OV (n = 18) BRCA+ (n = 5) BRCA- (n = 9)

2586

PFS median, mo

OS median, mo

3.63 2.73 1.77 3.66

9.05 9.35 8.96 8.03

Poster Session (Board #286), Sun, 8:00 AM-11:30 AM

2585

135s

Poster Session (Board #285), Sun, 8:00 AM-11:30 AM

A phase 1 dose-escalation study of the folic acid-tubulysin small molecule drug conjugate (SMDC) folate-tubulysin EC1456 in advanced cancer patients. First Author: Jasgit C. Sachdev, Scottsdale Healthcare, Paradise Valley, AZ Background: The folate receptor (FR) is highly expressed in several epithelial cancers (e.g. ovarian, NSCLC) but is expressed at low levels in most normal tissues. EC1456 is a novel, next generation SMDC of folic acid and the potent cytotoxic agent tubulysin B hydrazide (TubBH). EC1456 targets folate receptor positive (FR+) tumors to deliver TubBH intracellularly, leading to mitotic inhibition. Etarfolatide is a folate-containing tracer that chelates technetium-99m for SPECT imaging. 99mTc-etarfolatide noninvasively identifies FR+ malignant lesions to select pts for FR-targeted treatment. Phase 2 studies with the first generation therapeutic folate SMDC vintafolide have shown its utility in patient selection for FR-targeted therapy. Methods: The primary objective is to determine the MTD of EC1456 administered on 2 schedules BIW: days 1, 4, 8, 11 q 21 or 28 days (dosages: 0.5-4.5 mg/m2), or QW: days 1, 8 q 21 days (dosages: 1.5-10.0 mg/m2). Key inclusion criteria include: age $ 18 years, ECOG PS 0–1, and adequate endorgan function. 99mTc-etarfolatide scan to evaluate FR status is obtained on enrolled pts. Dose escalation follows the “3+3” protocol. Cycle 1 DLT evaluation must be completed for each dose level prior to dosing a new cohort. Results: 46 pts have been treated and are evaluable for cycle 1 toxicity. Median age is 66.5 years (range: 39-86); 27 pts are female. 23 pts have received 93 cycles of EC1456 BIW (median: 2; range: 1-22), and 23 pts have received 87 cycles of EC1456 QW (median: 2; range: 1-18). There have been no treatment-related deaths. One pt experienced uncomplicated G4 neutropenia. Two DLT’s have been observed, both on the QW schedule: G3 infusion reaction (4.5 mg/m2), and G3 headache (10.0 mg/m2). Stable disease (SD) . 4 months has been observed in 3 FR+ pts (mesothelioma, GEJ CA, SCLC) on the BIW schedule, and in 1 FR+ NSCLC pt and 1 FR nonevaluable pt with leiomyosarcoma on the QW schedule. Conclusions: Overall, both schedules of EC1456 appear to be well tolerated. Dose escalation is ongoing on both schedules. Anti-tumor activity of EC1456 is suggested by durable SD in pts with FR+ disease. Pharmacokinetic analyses are ongoing. Clinical trial information: NCT01999738.

2587

Poster Session (Board #287), Sun, 8:00 AM-11:30 AM

Phase 1 study of the PSMA-targeted tubulysin small-molecule drug conjugate EC1169 in patients with metastatic castrate-resistant prostate cancer (mCRPC). First Author: Michael J. Morris, Memorial Sloan Kettering Cancer Center and Weill Cornell Medical College, New York, NY

STM-01: Phase I EffTox study of aurora A kinase inhibitor alisertib (MLN8237) given in combination with selective VEGFR inhibitor pazopanib for therapy of solid tumors. First Author: Hiral A. Shah, University of Illinois at Chicago, Chicago, IL

Background: EC1169 is a small molecule drug conjugate designed to bind to prostate cancer (PCa) cells via prostate specific membrane antigen (PSMA) and, following endocytosis, to then release tubulysin B hydrazide (potent antimitotic). EC1169 administration to mice bearing PSMA-positive LNCaP and MDA PCa 2b xenografts has resulted in complete remissions and cures. EC0652, a novel tracer composed of a technetium-99m chelator and a PSMA-targeting moiety, is being developed to characterize whole body PSMA expression in real time. Rapid PSMA-specific uptake of 99mTcEC0652 into tumor tissue and rapid clearance from normal tissue results in enhanced tumor-to-background ratios to localize PSMA expressing tissue on SPECT imaging. Methods: Key inclusion criteria are ECOG PS 0-1, adequate organ function, and a diagnosis of mCRPC. Patients (pts) undergo 99m Tc-EC0652 SPECT scan prior to therapy. EC1169 is administered intravenously on 1 of 2 schedules every 21 days: 3 times weekly with 1 week off (TIW regimen), or once a week for 2 weeks with 1 week off (QW regimen). Dose escalation is based upon the “3+3” method. Study objectives include determination of EC1169’s maximal tolerated dose, safety, pharmacokinetics, and anti-tumor activity. 99mTc-EC0652 safety will be characterized. Results: Twenty-one pts are evaluable for cycle 1 toxicity (TIW: 5 pts; QW: 16 pts). All pts are Caucasian. Median age is 69.0 (range: 53-82). Median time on study is 6.7 weeks (range: 0.1-33.1). Median number (range) of administered EC1169 cycles is 4 (3-4) for TIW pts, and 2 (1-11) for QW pts. 9 of 21 pts have had drug-related (DR) adverse events (AE’s). No DRAE has occurred in . 1 TIW pt. Vomiting and fatigue were the only DRAE’s reported in $ 2 QW pts. There have been no on-study deaths, DR Grade 3-4 toxicity or serious adverse events, or dose limiting toxicity. 3 pts have demonstrated stable disease lasting more than 4 cycles. To date, no patient has demonstrated a 50% decrease in PSA. Dose proportionate increases in both Cmax and AUC for EC1169 have been observed. Conclusions: Both EC1169 and 99mTc-EC0652 appear to be well tolerated in mCRPC pts. Dose escalation and PK analyses are ongoing. Clinical trial information: NCT02202447.

Background: Pazopanib is a multi - kinase inhibitor that inhibits angiogenesis. Alisertib is a highly selective inhibitor of mitotic Aurora A Kinase that has been shown to disrupt chromosome segregation and in turn cell proliferation. At non-toxic concentrations, mitosis targeting agents exhibit antiangiogenic effects. Thus, the combination of these two agents may have a synergistic effect on tumor vasculature. The primary objective of this study is to determine the optimal tolerated dose (OTD) for alisertib and pazopanib, their safety and tolerability. Methods: This phase 1 study evaluated the OTD of alisertib taken orally, twice a day, on D1-7 in combination with pazopanib, taken orally, once a day, continuously in a 21 day cycle. There were a total of 4 dose levels tested, starting with 20 mg of alisertib and 400 mg of pazopanib. Disease response was assessed using the RECIST 1.1 every 2 cycles. Once the OTD was determined, a total of 9 patients were enrolled to assure safety and perform PK analysis. Results: A total of 27 patients received treatment and 18 of these patients had evaluable responses. The median age was 57 years (range 22-79), and approximately 74% of the patients had received at least three prior chemotherapy regimens. Dose limiting toxicities (DLT) occurred in dose level (DL) 2 (grade 3 colonic obstruction), DL 2+ (grade 4 thrombocytopenia and grade 3 mucositis), DL 3 (grade 3 AST/ALT elevation and grade 3 abdominal pain). The OTD was determined to be DL 2: alisertib 20 mg BID and pazopanib 600 mg daily. The most common grade 3 or 4 treatment related adverse events were neutropenia (22%), thrombocytopenia (12%), leukopenia (10%), anemia (10%), and increased AST (10%). Fourteen (78%) patients had stable disease and two (11%) patients had partial response. The median number of cycles was four, median progression free survival was 13.3 weeks (range (weeks); 1.9 -57.9) and median overall survival was not reached. Final PK analysis is ongoing. Conclusions: The combination of alisertib with pazopanib demonstrates manageable safety and early clinical evidence of antitumor activity in patients with advanced malignancies (NCT01639911). Clinical trial information: NCT01639911.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

136s

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics

2588

Poster Session (Board #288), Sun, 8:00 AM-11:30 AM

2589

Poster Session (Board #289), Sun, 8:00 AM-11:30 AM

Phase I study of X-82, an oral dual anti-VEGFR/PDGFR tyrosine kinase inhibitor, with everolimus in solid tumors. First Author: Benjamin R. Tan, Division of Oncology, Washington University School of Medicine, Saint Louis, MO

A phase I study of the CDK4/6 inhibitor, palbociclib plus 5-fluorouracil (5FU) in patients with advanced solid tumor malignancies (NCT01522989). First Author: Michael J. Pishvaian, Lombardi Comprehensive Cancer Center, Georgetown University, Washington, DC

Background: Anti-VEGFR tyrosine kinase inhibitors(TKI), combined with mTOR inhibitors, such as everolimus resulted in a significant response and progression-free survival benefit in pts with renal cell carcinoma. [Motzer]. In 2 Phase I studies of everolimus combined with sunitinib or lenvantinib, daily doses of everolimus above 2.5 mg and 5 mg, respectively, were not tolerated. [Molina]. X-82 is an oral multikinase inhibitor targeting VEGFR and PDGFR with a shorter half-life and limited tissue accumulation, thus, X-82 can potentially be combined with full therapeutic doses of everolimus. Methods: A 3+3 dose escalation design was utilized to determine the doselimiting toxicities (DLT) and the recommended Phase II dose (RP2D) of daily oral X82 plus everolimus at 10 mg PO daily for patients with solid malignancies. Eligibility include PS 0-1, measurable disease, adequate organ function and normal LVEF. Results: 21 pts (10 M/ 11F) with various malignancies were enrolled. 3 pts were not evaluable for DLT: 1 non-compliance, 1 consent withdrawal and 1 never initiated study therapy. 1 of 6 evaluable pts treated at dose level 4 experienced DLTs (gr 3 fatigue and hypophosphatemia). 1 of 3 pts in cohort 5 experienced a gr 3 mucositis. Due to inability to maintain starting doses beyond 1 cycle in 2 of 3 pts on cohort 5, our RP2D is X82 300 mg with everolimus 10 mg. RESPONSE: At cohort 4, 2 partial responses (PR) were noted in pts with renal cell carcinoma (RCC) who progressed through prior TKI and another RCC pt had marked necrosis of his recurrent tumor (SD by RECIST). At cohort 5, 1 pt with RCC had a 16% tumor shrinkage (SD). 10 pts with neuroendocrine tumors (NET) had SD with response duration ranging from 3-23 months. Conclusions: The RP2D is X82 at 300 mg PO daily plus everolimus 10 mg PO daily. Favorable responses seen in RCC and prolonged SDs for NETs warrant further investigations in these tumor types. Clinical trial information: NCT01784861.

Background: We demonstrated in mouse xenografts that CDK4/6 inhibition with palbociclib (P) is synergistic with 5FU in suppressing tumor growth. Based on these preclinical data, we initiated a Phase I trial to evaluate the safety and activity of P plus 5FU. Methods: Eligible patients (pts) had advanced, retinoblastoma (Rb)-positive (as demonstrated by IHC) solid tumors with an adequate performance status and intact organ function. The first stage of the study was a dose escalation to identify the recommended phase II dose (RP2D) of P in this combination (n = 21). The second stage is an ongoing schedule optimization, in which the timing of the P and 5FU are varied to identify the most effective sequence, as demonstrated by Ki67 suppression (n = 8 to date). All pts underwent tumor biopsies prior to treatment (tx), after 7 days of P, and again after the 5FU infusion. Results: Six pts were enrolled at a P dose of 50mg orally daily, Days 1-7, and IV 5FU as a 400mg/m2 bolus Day 8, and 2400mg/m2 continuous infusion Days 8-10, on an every 2 week cycle. However, 5/6 pts experienced significant delays in tx due to persistent myelosuppression, though only 1 met the definition of dose-limiting toxicity (DLT). The protocol was thus amended to remove the 5FU bolus. An additional 23 pts were enrolled. Most pts had colorectal (19) or breast (6) cancer; and pts had received a median 3 prior lines of Tx. 17 pts were DLT evaluable. The most common adverse events (AEs) were neutropenia (10), thrombocytopenia (4), nausea (N) (4), vomiting (V) (3), and mucositis (3). There were 4 DLTs (treatment delay (1), N/V (3)). The RP2D of P in this combination was 100mg orally daily. Efficacy was evaluable in 26/29 pts. One pt experienced a confirmed PR (Breast), and 6 pts had SD for a DCR of 27%. The median PFS in these 7 pts was 10 months, with 3 pts still on study. Tumor assessment of changes in Ki67 and Rb status are ongoing. Conclusions: The combination of P and 5FU is safe at a RP2D of P of 100mg. Accrual to schedule optimization is nearly complete, and the optimal schedule will be presented. This combination demonstrated promising disease control in this highly refractory population and should be tested in disease-specific Phase II trials. Clinical trial information: NCT01522989.

Dose Level

X82

Everolimus # pts

1

100 mg

10 mg

3

2

150 mg

10 mg

3

3

200 mg

10 mg

4

4

300 mg

10 mg

8

5

400 mg

10 mg

3

2590

pts with DLT

Responses

0

2 SD panc NET and carcinoid 2 SD Carcinoid and glioblastoma 0 3 SD Panc NET (2), carcinoid 1-Gr 3 fatigue, Gr 3 hypophosphatemia 2 PR- RCC (2) 3 SD- panc NET (2), thymic NET 1-Gr 3 mucositis 2 SD- RCC and pancreatic NET 0

Poster Session (Board #290), Sun, 8:00 AM-11:30 AM

TPS2591

Poster Session (Board #291a), Sun, 8:00 AM-11:30 AM

Exploratory responder analyses of greatest depth of response (DepOR) and survival in patients with metastatic non-small cell lung cancer (mNSCLC) treated with a targeted therapy or immunotherapy. First Author: Dickran Gano Kazandjian, Office of Hematology and Oncology Products/ Food and Drug Administration, Silver Spring, MD

First-in-human, open label, multicenter phase I of IPH4102, first-in-class humanized anti-KIR3DL2 monoclonal antibody, in relapsed/refractory cutaneous T-cell lymphomas. First Author: Martine Bagot, AP-HP, Hopital ˆ Saint-Louis, Unite´ d’Oncodermatologie, Departement ´ de Dermatologie, Paris, France

Background: RECIST criteria as a binary output may not fully capture the clinical benefit of targeted therapy and immunotherapy in mNSCLC. We sought to determine the association between the maximum reduction in tumor target lesions from baseline (DepOR) with overall survival (OS) for two drugs, an ALK inhibitor (ALKi) and a PD-1 inhibitor (PD-1i), which represent the targeted and immune therapy drug classes. Methods: Data from two randomized trials were pooled for each drug for exploratory analyses. Data from patients with any tumor shrinkage were grouped into 4 quartiles based on the greatest percent shrinkage in tumor target lesions from baseline and compared to data from patients with no tumor shrinkage. A retrospective exploratory responder analysis was performed to evaluate the association between the DepOR quartiles and OS for each drug using hazard ratios (HR, Cox proportional hazards model) Results: The exploratory responder analysis suggested that patients with the highest percentages of tumor shrinkage experienced the longest survival for both drugs. Specifically, pooled analysis of ALKi trials suggested incrementally larger HR for OS with increased tumor shrinkage by quartile. Pooled analysis of PD-1i trials suggested that greater than 50% tumor shrinkage was associated with more favorable OS HR compared to subgroups with 1-50% tumor shrinkage and with no tumor shrinkage (Table). Conclusions: Our analysis suggests that a positive association between the percentage of maximal tumor shrinkage with increasing effects on survival as measured by HR for both ALKi and PD-1i. These exploratory retrospective analyses results are hypothesis generating and further evaluation is necessary to understand these relationships.

Background: KIR3DL2 is consistently expressed in all subtypes of Cutaneous T-cell Lymphomas (CTCL), irrespectively of disease clinical stage, with the greatest expression in Sezary ´ Syndrome (SS) and transformed Mycosis Fungoides (MF), two subsets with high unmet need. KIR3DL2 belongs to the killer immunoglobulin (Ig)-like receptor (KIRs) family expressed on minor populations of NK, CD8 and CD4 T cells. IPH4102 is a first-in-class anti-KIR3DL2 monoclonal antibody (mAb). It depletes selectively KIR3DL2-expressing cells. Its modes of action include AntibodyDependent Cell-Cytoxicity (ADCC) and –Phagocytosis (ADCP). IPH4102 has potent efficacy in non-clinical models, in particular ex vivo autologous assays using primary CTCL cells. Methods: IPH4102-101 (NCT02593045) is a first-in-Human Phase I study of single-agent IPH4102 in relapsed/ refractory CTCL. The study has two sequential parts, a dose-escalation followed by cohort expansion. The primary objective is to assess the safety and tolerability of increasing doses by characterizing dose-limiting toxicity (DLT) and adverse events. A 3+3 design with accelerated titration is employed. IPH4102 first Human dose and subsequent dose-levels were selected based on a MABEL strategy and PK/PD modeling. The doseescalation portion aims to determine the maximal tolerated dose (MTD) or recommended Phase 2 dose (RP2D). Secondary objectives include PK, immunogenicity and signals of anti-neoplastic clinical activity. Exploratory objectives include identification of biomarkers of IPH4102 activity in skin lesions and on malignant blood cells when present. Eligible CTCL patients must have received at least 2 lines of systemic therapy. Centrally assessed KIR3DL2 expression in at least one involved organ is also required. In the expansion portion, two CTCL subtype-specific cohorts will include 20 additional patients to further explore the MTD or RP2D. CTCL subtypes will be selected according to dose-escalation findings. Enrollment into study IPH4102-101 started in November 2015. Cohorts 1-3 have been completed without DLT. Enrolment to cohort 4 started in January 2016. Clinical trial information: NCT02593045.

Crizotinib DepOR Subgroups N = 305 No tumor shrinkage (n = 12) 1-25% (n = 39) 26-50% (n = 70) 51-75% (n = 144) 76-100% (n = 40) Nivolumab DepOR Subgroups N = 355 No tumor shrinkage (n = 168) 1-25% (n = 70) 26-50% (n = 44) 51-75% (n = 45) 76-100% (n = 28)

OS HR (95% CI) 0.98 0.66 0.38 0.06

(0.36, (0.25, (0.15, (0.01,

2.69) 1.75) 0.96) 0.33)

0.54 0.50 0.09 0.14

(0.38, (0.32, (0.04, (0.06,

0.76) 0.78) 0.21) 0.31)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics TPS2592

Poster Session (Board #291b), Sun, 8:00 AM-11:30 AM

A phase I dose escalation and cohort expansion study of T-cell redirecting bispecific antibody against Glypican 3 in patients with advanced solid tumors. First Author: Kenji Hashimoto, Chugai Pharma Europe, London, United Kingdom

TPS2593

137s

Poster Session (Board #292a), Sun, 8:00 AM-11:30 AM

A phase I, open label, two part, safety and tolerability study of U3-1784 in patients with advanced solid tumours. First Author: Marika Ciprotti, Daiichi Sankyo Development Ltd., Gerrards Cross, United Kingdom

Background: Despite recent advance of immunotherapy, not all patient benefit. Reasons for the limited therapeutic effect include a low level of target protein expression or tumor immunogenicity. Bispecific antibodies to engage T-cells onto tumor are one approach to more active immunotherapy. However, management of on target/off tumor adverse effects including cytokine release can be a challenge. Glypican 3 (GPC3) is expressed across variety of solid tumors including hepatocellular carcinoma, while its expression in normal tissue is limited, making it an ideal target for a bispecific antibody. Our T-cell redirecting bispecific antibody against GPC3 is a fully humanised IgG4 bispecific antibody to recognise GPC3 on tumor with mutation to silence FcgR. Preclinical data show that this antibody is highly active against variety of GPC3-positive tumors, including poorly immunogenic models. Utilizing corticosteroids was effective in reducing cytokine release and widens the therapeutic window in preclinical study. Therefore, we designed a phase I dose escalation and cohort expansion study of this bispecific antibody in patients with GPC3-positive solid tumors. Methods: Approximately 50 patients with GPC3-positive solid tumor will be treated with this antibody weekly in dose escalation phase, until disease progression, occurrence of dose limiting toxicities (DLT) or discontinuation for other reasons. Subsequently, preliminary anti-tumor efficacy will be evaluated in patients with selected GPC3-positive tumor in a cohort expansion phase. Accelerated titration dose escalation design is utilised until grade 2 drug related event is observed, followed by 3+3 design. Modified continuous reassessment method will guide the dose escalation after the first event of DLT. Starting dose was determined based on the NOAEL and MABEL approach set as 0.003 mg/kg.

Background: Fibroblast Growth Factor Receptor 4 (FGFR4) is the fourth member of a family of FGF tyrosine kinase receptors that regulate cellular pathways involved in proliferation, differentiation and survival. Fibroblast Growth Factor 19 (FGF19), a natural ligand of FGFR4, has unique specificity for its receptor and regulates bile acid synthesis and hepatocyte proliferation in the normal liver. Although FGF signalling has been found to be of importance in a number of tumour types, including prostate, breast, pancreatic, pituitary and gynaecological tumours, growing evidence suggests that FGF19/FGFR4 pathway is a key driver in the development of hepatocellular carcinoma (HCC). U3-1784 is a fully human IgG1 monoclonal antibody and antagonist to FGF19 ligand. It selectively targets FGFR4 without binding to FGFR 1, 2 and 3. Methods: The study is divided into 2 parts: in part 1 (Dose Escalation Phase) multiple ascending doses (3 + 3 design) of U3-1784 are given to patients with histologically or cytologically confirmed advanced solid tumours; part 2 (Dose Expansion Phase) is carried out in up to 27 patients with histologically or cytologically confirmed HCC. Patients $ 18 years of age, performance status # 1 and whose malignancies are refractory to, intolerant of, or not eligible for standard treatment, or who decline standard therapy, or for whom no therapy with curative intent is available are eligible to enter this study. U3-1784 is administered intravenously every 2 weeks. Primary objectives are: safety and tolerability of U3-1784, maximum tolerated dose and/or safety of the maximum administered dose. Secondary objectives are: pharmacokinetic and pharmacodynamics (alphafetoprotein, bile acids, AST/ALT) of U3-1784, emergence of antidrug antibodies. Exploratory objectives are: tumour response by RECIST 1.1, overall response rate, disease control rate, time to progression, overall survival, duration of response/stable disease and additional biomarkers. Enrolment to cohort 1 began in January 2016. Clinical trial information: 2015-00267020.

TPS2594

TPS2595

Poster Session (Board #292b), Sun, 8:00 AM-11:30 AM

A Phase 1 study to assess the safety, pharmacokinetics (PK), pharmacodynamics (PD) and antitumor activities of daily oral BAL101553, a novel tumor checkpoint controller (TCC) in adult patients with advanced solid tumors. First Author: Rebecca Sophie Kristeleit, University College London, London, United Kingdom Background: BAL101553 is the pro-drug of BAL27862, a novel small molecule TCC that promotes tumor cell death by modulating the spindle assembly checkpoint (SAC). In a completed Phase 1 (P1) study, pre- and post-treatment tumor biopsies revealed antiproliferative and vascular disrupting PD effects with BAL101553 administered as a 2-hour intravenous infusion1. The time profile of transient blood pressure elevations linked the vascular effects of the drug to Cmax, while preclinical data suggest that effects on tumor cell proliferation are AUC-mediated; vascular toxicity was dose-limiting. Given this, dosing strategies that optimize the PK/PD profile, particularly those that minimize the Cmax, may allow higher doses and/or exposures of BAL27862 to be tolerated. Preclinical data with oral BAL101553 in various animal models, including colorectal, lung cancer and glioblastoma xenograft models, suggest excellent bioavailability and antitumor efficacy. Methods: This is an ongoing multicenter, open-label, Phase 1 dose-escalation study (NCT02490800) using a 3+3 design to determine the maximum tolerated dose (MTD), characterize dose-limiting toxicities (DLTs) and assess the PK, PD and antitumor activities of oral BAL101553. Eligible patients with advanced solid tumors, having failed standard therapy, receive once-daily oral capsules of BAL101553 in consecutive 28-day cycles. Adverse events are assessed using CTCAEv4; objective tumor response by RECIST 1.1 (every 2 cycles). The starting dose is 2 mg per day and dose escalation allows for dose doubling depending on observed toxicities. PD assessments include tumor biopsies and circulating tumor, endothelial and endothelial progenitor cells (CTCs, CECs, CEPs). Tumor biomarkers analyzed include BubR1, a checkpoint protein involved in the SAC and a potential biomarker for BAL101553 activity. Time-concentration PK profiles and repeated trough levels are assessed throughout the first two treatment cycles. 1Molife LR et al. J Clin Oncol 32:5s, 2014 (suppl; abstr 2562) Clinical trial information: NCT02490800.

Poster Session (Board #293a), Sun, 8:00 AM-11:30 AM

A phase I, open-label study of GSK2816126, an enhancer of zeste homolog 2 (EZH2) inhibitor, in patients with relapsed/refractory diffuse large B-cell lymphoma (DLBCL), transformed follicular lymphoma (tFL), other nonHodgkin’s lymphomas (NHL), multiple myeloma (MM) and solid tumor. First Author: Timothy Anthony Yap, Royal Marsden NHS Foundation Trust, London, United Kingdom Background: EZH2 is the methyltransferase component of the polycomb repressive complex 2 that represses transcription of target genes via trimethylation of histone H3 at lysine 27 (H3K27me3). These targeted genes are involved in fundamental cellular processes, e.g., development and cell differentiation. EZH2 is deregulated (via mutation or overexpression) in many cancers; increased expression correlates with more aggressive tumor progression and poor prognosis. GSK2816126, a selective and potent inhibitor of wild-type (WT) and mutant EZH2, can decrease H3K27me3 levels in vitro and has demonstrated antiproliferative activity in several cell lines derived from EZH2 WT/mutant cancers. Methods: This is a Phase I, openlabel, first-time-in-human, 2-part study. Part 1 is a dose-escalation phase to determine the safety, tolerability and recommended Phase 2 dose (RP2D) of GSK2816126 in patients ( $ 18 years) with relapsed/refractory DLBCL, tFL, other NHL, MM and solid tumors (all with ECOG PS of 0/1). An accelerated dose titration will be employed with one patient/dose level until the occurrence of a $ Grade 2 drug-related toxicity or dose-limiting toxicity; thereafter patients will be enrolled in a standard 3+3 design. A patient will receive a starting dose of 50mg GSK2816126 IV twice-weekly, and dose escalation will continue until an RP2D/MTD/MFD is determined. In Part 2, the clinical activity of GSK2816126 (objective response rate) will be evaluated in expansion cohorts of patients with MM, and with WT/mutant EZH2 germinal center B-cell–like DLBCL and tFL. Enrollment can be stopped early for futility. Additional study objectives include analysis of pharmacokinetics of GSK2816126 after single/repeat dosing, evaluation of the relationship between GSK2816126 exposure and safety/efficacy/ pharmacodynamic parameters, durability of response and progression-free survival. Recruitment is ongoing across four centers (USA and UK); 40 patients will be enrolled in Part 1 (currently 16 enrolled) and 129 in Part 2. Clinical trial information: NCT02082977.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

138s TPS2596

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics Poster Session (Board #293b), Sun, 8:00 AM-11:30 AM

TPS2597

Poster Session (Board #294a), Sun, 8:00 AM-11:30 AM

Phase II trial of the MEK1/2 inhibitor selumetinib (AZD6244) in adults with neurofibromatosis type 1 (NF1) and inoperable plexiform neurofibromas (PNs). First Author: Diana Bradford, Children’s National Medical Center, Washington, DC

First-in-human trial of microRNA cancer therapy with MRX34, a liposomal miR-34 mimic: Phase Ia expansion in patients with advanced solid tumors. First Author: Muhammad Shaalan Beg, Division of Hematology/Oncology, The University of Texas Southwestern Medical Center, Dallas, TX

Background: PNs in NF1 are a major source of morbidity, causing disfigurement, functional impairment, and pain. Selumetinib is an oral inhibitor of MEK 1/2, which may mediate anti-tumor effects in PNs by inhibiting Ras signaling. Selumetinib is currently undergoing evaluation in adult cancers, pediatric brain tumors and NF1 related PNs. Our phase I trial of selumetinib for children with NF1 and inoperable PNs demonstrated unprecedented activity with objective responses (PN volume decrease $ 20% using volumetric MRI analysis) in . 50% of patients (pts) enrolled. However, we have observed slow PN regrowth in several patients who required dose reductions for toxicity. This suggests that a certain selumetinib tissue concentration may be required for target inhibition and anti-tumor activity. Methods: This study will evaluate 1) the objective response rate to selumetinib in pts $ 18 years old with NF1 and inoperable, growing or symptomatic PNs, 2) PN and dermal neurofibroma (DN) biopsies prior to and on treatment with selumetinib for mechanisms of response and resistance, 3) the potential clinical benefit of selumetinib with evaluations of PN related pain, quality of life, and function, and 4) steady state pharmacokinetics and pharmacodynamics (cytokines, bone marrow derived precursor cells, and peripheral blood immune subsets). Willingness to undergo PN biopsies is required for eligibility. Pts will receive selumetinib at the adult recommended dose of 75 mg every 12 hours on a continuous dosing schedule (1 cycle = 28 days). PN biopsies will be analyzed for pERK, pAKT, pMEK, tumor kinome and transcriptome, and immune infiltrate. In addition, we will analyze and compare pERK, pAKT in DN. Response will be evaluated with periodic volumetric MRI analysis. Patient-reported outcome and adherence evaluations will be conducted. With a Simon two-stage design targeting a 45% response rate (90% power), if $ 5/20 pts in stage 1 respond, enrollment will be expanded to 35 evaluable patients and if $ 12/35 patients respond, selumetinib will be considered active. This CTEP-sponsored single site open label phase II study (NCT02644512) is currently enrolling patients. Clinical trial information: NCT02644512.

Background: MicroRNAs (miRNAs) are naturally occurring, short (~22 nucleotides), non-coding single-stranded RNAs that comprise a new class of post-transcriptional regulators of gene expression. Each miRNA modulates the expression of hundreds of genes across distinct cellular pathways. This broad activity gives miRNA-based therapy the potential to simultaneously repress multiple oncogenic processes in the tumor microenvironment, including growth and proliferation, resistance, cancer stem cells, metastasis, and immune evasion. The tumor suppressor miRNA miR-34a has been shown to down-regulate expression of . 30 oncogenes (eg, MET, MEK1, PDGFR-a, CDK4/6, BCL2, WNT 1/3, NOTCH1, CD44), as well as genes involved in tumor immune evasion (eg, PD-L1, DGKz). MRX34, a liposomeencapsulated miR-34a mimic, is a potential first-in-class miRNA therapy for cancer. Methods: The initial standard 3x3 dose-escalation portion of this trial (NCT01829971) is complete; here we describe the on-going phase Ia expansion in which 21 of ~80–100 planned patients have been enrolled. Eligible patients are $ 18 yrs, have refractory advanced solid tumors including HCC (Child-Pugh A only), RCC, melanoma, lung cancer (NSC or SC), or other solid tumors of potential interest, ECOG PS 0–2, acceptable hepatic, renal, and hematologic function, and anticipated life expectancy of . 3 m. Under required steroid premedication, MRX34 is given at the previously determined MTDs of 70 mg/m2 for HCC or 93 mg/m2 for non-HCC solid tumors in daily IV infusions for 5 consecutive days in week 1 with 2-weeks rest on a 21-day cycle (QDx5 schedule). The primary objective is safety of the recommended phase II dose (RP2D); secondary objectives are PK and biological and clinical activity. AEs are evaluated and recorded at each patient visit; antitumor activity is assessed by CT or MRI at screening, the end of cycle 2, and every even cycle thereafter. Down-regulation of cancer oncogenes, miRNA levels, and other potential pharmacodynamic markers are being evaluated, including with qRT-PCR and next-generation sequencing (NGS) analyses of white blood cells from blood samples and, as available, tumor and tissue biopsies. Clinical trial information: NCT01829971.

TPS2598

TPS2599

Poster Session (Board #294b), Sun, 8:00 AM-11:30 AM

Poster Session (Board #295a), Sun, 8:00 AM-11:30 AM

Open-label phase II clinical trial of orteronel (TAK-700) in metastatic or advanced nonresectable granulosa cell ovarian tumors: The GREKO II study. First Author: Laia Garrigos, Hospital del Mar, Barcelona, Spain

A Phase II Basket Study of the Oral TRK Inhibitor LOXO-101 in Adult Subjects with NTRK Fusion-Positive Tumors. First Author: Alexander E. Drilon, Memorial Sloan Kettering Cancer Center, New York, NY

Background: Granulosa-cell tumors (GCT) of the ovary are a rare entity characterized by a pathognomonic punctual mutation at the FOXL2 gene 402C→G (C134W). This alteration leads to an upregulaton of the enzyme CYP17 (key in androgens and 17-OH-progesterone synthesis). Interestingly, a previous study by our group has demonstrated clinical activity of ketoconazole, a well known CYP17 inhibitor (GREKO I trial). Additionally, cell cultures of both adult and juvenlie GCT showed an increased sensitivity to both ketoconazole and abiraterone compare to pancreatic tumors and lung adenocarcinoma. Orteronel (TAK700) is a selective inhibitor of 17, 20-lyase, that is being developed as an endocrine therapy for relevant hormonesensitive cancers such as prostate and breast cancer. Orteronel is expected to suppress sex hormone levels in both circulation and relevant hormone-dependent malignant tissue. Thus we aimed to test the clinical activity of such drug in GCT. Methods: An open-label phase II single arm clinical trial has been designed for women with metastatic or locally advanced nonresectable granulosa cell ovarian tumor that harbors the somatic mutation FOXL2 402C→G (C134W) and who have not received prior treatment with any CYP17 inhibitor. Treatment consists on Orteronel 300mg BID, given orally, continuously in a 28-day treatment cycle. The primary objective is to assess the clinical benefit rate. Since this is an extremely unfrequent disease, 10 Spanish institutions are involved. 8 for the 20 planned cases have already been recruited. An extensive translational research will be included in order to improve our scarce knowledge of GCT. Clinical trial information: NCT02101684.

Background: The TRK family of neurotrophin receptors, TRKA, TRKB, and TRKC (encoded by NTRK1, NTRK2, and NTRK3 genes, respectively) and their neurotrophin ligands regulate growth, differentiation and survival of neurons. Fusion of the kinase domain of the NTRK genes with various partners are oncogenic, and have been identified across many common tumor types, but generally at a frequency of , 5%. In rare tumor types, such as mammary analogue secrectory carcinoma (MASC) and secretory breast carcinoma, NTRK fusion genes have been identified in greater than 75% of cases. LOXO-101 is a potent, oral, ATP-competitive pan-TRK inhibitor with IC50 values in the low nanomolar range for inhibition of TRK family members in binding and cellular assays, with . 1000x selectivity over other kinases. LOXO-101 is the only selective pan-TRK inhibitor in clinical development and has demonstrated tumor inhibition in preclinical models and meaningful responses in an adult phase 1 trial. Methods: The phase II study of LOXO101 (NCT02576431) is an open-label, multi-center trial for adult patients with advanced solid tumors and primary CNS malignancies harboring NTRK fusions. Patients $ 18 years old with NTRK1, 2, or 3 fusion-positive cancers who have progressed following standard therapy are eligible and assigned to one of 8 cohorts in a basket trial design. Eight anatomically defined subgroups have been prespecified for patients with: non-small cell lung, thyroid, sarcoma, colorectal, salivary and biliary cancers, CNS malignancies, and all others. Patients are required to have RECIST or RANO measurable disease for eligibility for cohorts 1-7. Patients without measurable disease may enroll in the “all other anatomic sites” cohort, number 8. Patients will undergo radiographic evaluation for their disease at regular intervals. LOXO-101 is administered orally BID in capsule form at 100mg for continuous 28-day cycles. Archival tissue and fresh tissue are required for post hoc central review of NTRK fusion status as well as the characterization of other molecular pathologic features. Clinical trial information: NCT02576431.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics TPS2600

Poster Session (Board #295b), Sun, 8:00 AM-11:30 AM

A phase 1b/2 study of ibrutinib combination therapy in selected advanced genitourinary and gastrointestinal tumors. First Author: Jordan Berlin, Vanderbilt-Ingram Cancer Center, Nashville, TN Background: For patients with advanced genitourinary (GU) and gastrointestinal (GI) tumors, effective treatment options are limited. Ibrutinib (ibr) is a first-in-class, covalent inhibitor of Bruton’s tyrosine kinase (BTK), as well as a covalent inhibitor of other relevant kinases in solid tumor development pathways. In solid tumor models, ibr exhibited direct antitumor effects and enhanced the effects of other anticancer agents (Masso-Valles, Cancer Res 2015; Sagiv-Barfi, PNAS 2015; Chen, AACR-NCI-EORTC 2015, Gunderson Cancer Disc 2015). Preclinical models suggest that inhibition of BTK and interleukin-2–inducible kinase (ITK) may have efficacy by targeting the tumor microenvironment, which is composed of MAST cells, B cells, and other tumor-infiltrating leukocytes. Based on these preclinical results, we propose that ibr in combination with existing therapies may offer increased efficacy without overlapping toxicity in select solid tumors (Chanan-Khan, ASCO 2015). This study will evaluate ibr in combination with available effective therapies in patients with previously treated GU and GI tumors. Methods: This open-label, phase 1b/2 multicenter study (NCT02599324; PCYC-1128) will evaluate ibr in combination with existing therapies in up to 189 patients who have either metastatic renal cell carcinoma (RCC), advanced urothelial carcinoma (UC), advanced gastric adenocarcinoma (GA), or metastatic colorectal adenocarcinoma (CRC). All patients will receive oral ibr; patients will also receive oral everolimus for RCC; IV paclitaxel for UC; IV docetaxel for GA or IV cetuximab for CRC. The primary objective of phase 1b is to determine the safety of the combinations and establish the recommended phase 2 dose of ibr. The objectives of phase 2 are to determine overall response rate, progression-free and overall survival, and disease control rate, and to assess biomarkers. Eligible patients will have adequate hematologic, hepatic, and renal function, no prior treatment with study drugs or related agents, and no known CNS metastasis. Enrollment in the study has been initiated in the United States and European Union. Results: N/A Conclusions: N/A Clinical trial information: NCT02599324.

TPS2602

Poster Session (Board #296b), Sun, 8:00 AM-11:30 AM

A phase 1, multicenter, dose-escalation study of PRN1371, an irreversible covalent FGFR1-4 kinase inhibitor, in patients with advanced solid tumors, followed by expansion cohorts in patients with FGFR genetic alterations. First Author: Sarina Anne Piha-Paul, Department of Investigational Cancer Therapeutics (Phase 1 Program), The University of Texas MD Anderson Cancer Center, Houston, TX Background: FGFR inhibition is a promising therapeutic approach in a number of solid tumors where genetic alterations of FGFR drive tumor cell proliferation, and survival. PRN1371 is an oral, potent, irreversible inhibitor of FGFR1-4 that has shown high potency against many FGFR-aberrant cell lines. Methods: Part A of this phase 1 clinical trial explores ascending doses of PRN1371 in adult patients with advanced solid tumors in a "3 + 3" design, where cohorts of three patients are studied at each level until additional patients need to be added to better assess safety, establish the maximum tolerated dose and define the recommended phase 2 dose (RP2D). PRN1371 is dosed once daily in continuous, 28-day cycles. Part B studies two or three cohorts of 10 patients each with FGFR1-4 gene mutations, fusions, or truncations at the RP2D. The on-target effect of serum phosphorus and FGF23 increases are measured as potential pharmacodynamic biomarkers. Elevated serum phosphorus is managed with oral phosphate binding medications and a low phosphate diet, with dose interruptions and use of acetazolamide if certain thresholds are exceeded. Circulating tumor DNA from patients at baseline and during follow up is analyzed for FGFR genetic alterations. Pre and post-treatment tumor biopsies in Part B will be tested for a panel of pharmacodynamic and predictive biomarkers of FGFR inhibition. Clinical trial information: NCT02608125.

TPS2601

139s

Poster Session (Board #296a), Sun, 8:00 AM-11:30 AM

A randomized, double-blind, placebo-controlled study of ibrutinib, a Bruton tyrosine kinase inhibitor, with nab-paclitaxel and gemcitabine in the firstline treatment of patients with metastatic pancreatic adenocarcinoma (RESOLVE). First Author: Margaret A. Tempero, University of California, San Francisco, San Francisco, CA Background: Advanced pancreatic ductal adenocarcinoma (PDAC) is an aggressive malignancy with poor prognosis and short survival. Nabpaclitaxel/gemcitabine (Nab-P/GCB) is a preferred front-line treatment regimen. The Bruton tyrosine kinase (BTK) inhibitor ibrutinib (ibr) demonstrates antitumor activity in preclinical PDAC models, due in part to modulation of the tumor microenvironment. Ibr treatment inhibits mast-cell degranulation and decreases tumor-associated inflammation and desmoplasia; simultaneously, ibrutinib enhances the presence and activity of cytotoxic T cells in tumors (Affara et al, Cancer Cell, 2014; Masso-Valles et al, Cancer Res, 2015; Gunderson et al, Cancer Discov, 2015). This study will evaluate the efficacy of ibr in combination with Nab-P/GCB. Methods: The randomized, multicenter, double-blind, phase 2/3 RESOLVE trial (NCT02436668; PCYC-1137-CA) compares ibr vs. placebo, in combination with Nab-P/GCB, in approximately 320 patients with previously untreated metastatic pancreatic carcinoma. Eligibility criteria include confirmed metastatic pancreas carcinoma within 6 weeks of randomization; adequate organ function; and Karnofsky performance score $ 70. Key exclusion criteria include previous treatment (chemotherapy for primary PDAC, BTK inhibitor), and history of stroke or intracranial hemorrhage within 6 months. Patients will be randomized to oral ibr (560 mg) or matched placebo given daily in combination with Nab-P (125 mg/m2) and GCB (1,000 mg/m2) on days 1, 8, and 15 of a 28-day cycle until disease progression or treatment is no longer tolerated. A safety run-in phase evaluates the ibr combination prior to the randomization phase of the study. The primary endpoint is progression-free survival; secondary endpoints include overall survival and safety. Enrollment in the trial is ongoing. Results: N/A Conclusions: N/A Clinical trial information: NCT02436668.

TPS2603

Poster Session (Board #297a), Sun, 8:00 AM-11:30 AM

PATRIOT: A phase I study to assess the tolerability, safety and biological effects of a specific ataxia telangiectasia and Rad3-related (ATR) inhibitor (AZD6738) as a single agent and in combination with palliative radiation therapy in patients with solid tumours. First Author: Magnus Dillon, The Institute of Cancer Research, London, United Kingdom Background: Tumour control rates from radiation therapy (RT) are limited by normal tissue toxicities. Novel strategies are required to selectively sensitize tumour cells to radiation-induced DNA damage. The G2 cell cycle checkpoint is an attractive target for this, as normal cells will be protected by their intact G1 checkpoint, which is lost in the majority of cancer cells. ATR is an important mediator of the G2 checkpoint. Preclinical data suggest that ATR inhibition will sensitise to DNA damaging therapies, including RT. This multi-part phase 1 trial aims to assess safety and tolerability and preliminary anti-tumour activity of the ATR inhibitor AZD6738 as monotherapy and in combination with palliative RT, escalating both drug and radiation dose at a dose-fractionation relevant to radical treatment. The design aims to test a novel agent at the earliest stage of clinical development and assess safety in combination with RT, with the aim of moving to a radically-treated population if tolerated. Methods: Participants have advanced solid tumours without standard systemic therapy options. The trial comprises three parts: parts A and B will assess AZD6738 as a single agent in dose escalation to MTD (part A), followed by expansion cohorts enriched for defective DNA damage response (part B). Part C will assess AZD6738 in combination with palliative RT in which participants will receive 20 Gy in 10 fractions, with per cohort escalation of drug dose to monotherapy MTD if tolerated. At the highest tolerated combination dose, the RT dose will be escalated to 30 Gy in 15 fractions. Maintenance AZD6738 post-RT will be tested at the highest tolerated combination dose. The study opened in August 2014. The study is dose escalating in part A and part C has opened and treated its first patient. Part B will open when dose escalation has completed. PATRIOT is sponsored by The Royal Marsden, funded by the Cancer Research UK/AstraZeneca Combinations Alliance and supported by supply of free drug and distribution costs from Astra Zeneca. Clinical trial information: NCT02223923.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

140s TPS2604

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics Poster Session (Board #297b), Sun, 8:00 AM-11:30 AM

ROAR: a phase 2, open-label study in patients (pts) with BRAF V600E– mutated rare cancers to investigate the efficacy and safety of dabrafenib (D) and trametinib (T) combination therapy. First Author: Vivek Subbiah, The University of Texas MD Anderson Cancer Center, Houston, TX

TPS2605

Poster Session (Board #298a), Sun, 8:00 AM-11:30 AM

Phase I study of procaspase activating compound -1 (PAC-1) in the treatment of advanced malignancies. First Author: Oana Cristina Danciu, University of Illinois at Chicago, Chicago, IL

Background: Combination D + T is approved for the treatment of BRAF V600E/K–mutant unresectable or metastatic melanoma. BRAF V600E mutations have also been identified in some rare cancers, including anaplastic thyroid cancer, biliary tract cancer, gastrointestinal stromal tumor, nonseminomatous germ cell tumor, hairy cell leukemia (HCL), World Health Organization (WHO) grade I/II glioma, WHO grade III/IV glioma, multiple myeloma, and adenocarcinoma of the small intestine. US incidence rates of these rare tumors range from 0.07 to 6.3 per 100,000, with BRAF V600E mutation frequency ranging from 3% in high-grade glioma to 90% in HCL. Therefore, the ROAR study is underway in pts with these 9 rare BRAF-mutant cancers to assess the safety and efficacy of D + T combination therapy. Methods: This international, multicenter, open-label phase 2 study (NCT02034110) uses a Bayesian hierarchical statistical design that increases study power by borrowing information across histology cohorts based on the emerging response rate. Enrollment for each of the 9 cohorts (up to 25 pts each) may be stopped early for futility or efficacy if various criteria occur based on accrued data. Pts with BRAF V600E–mutant histologically or cytologically confirmed advanced disease with no available standard treatment options are eligible. Enrollment based on local BRAF V600E mutation results is permitted. Pts receive D (150 mg twice daily orally) and T (2 mg once daily orally) on a continuous dosing schedule until unacceptable toxicity, disease progression, or death occurs. Responses are assessed every 8 weeks per tumor-specific response criteria while pts are on study treatment. The primary study endpoint is overall response rate by investigator assessment. Secondary objectives include duration of response, progression-free survival, overall survival, and safety. Pharmacodynamic markers and quality of life will also be evaluated. If a histological cohort meets stopping criteria for early efficacy, an expansion cohort will be opened. This ongoing trial is currently recruiting pts at 48 centers in the United States, Canada, Europe, and South Korea. Clinical trial information: NCT02034110.

Background: Members of the caspase family of cysteine proteases are key players in both the initiation and execution of apoptosis; the activation of procaspase-3 to caspase-3 is a critical event in the apoptotic cascade. Procaspase-3 levels are elevated in: glioblastoma, breast cancer, colon cancer, lung cancer, lymphoma, neuroblastoma, melanoma, and liver cancer. As a consequence, caspase-3 levels are abnormally low in these tumors, allowing the tumors to avoid apoptosis. PAC-1 is a small molecule that directly activates procaspase-3 and induces apoptosis of cancer cells in culture. PAC-1 showed efficacy across a wide range of cancer cell lines, as well as in animal models of cancer, including brain cancer. This novel compound potently synergizes with chemotherapy agents (e.g.doxorubicin, temozolomide, etoposide, carboplatin). Methods: This is a Phase I dose escalation study with a modified- Fibonacci 3+3 design, consisting of two parts: to determine the maximum tolerated dose (MTD) of PAC-1 in advanced malignancies, and to determine the MTD of PAC-1 when combined with temozolomide in patients with primary brain tumors. For both parts the MTD dose level will expand to a total of 9 patients to ensure safety. Primary objectives: establish MTD, tolerability and toxicity. Secondary and correlative objectives: pharmacokinetics, pharmacodynamics, preliminary antitumor activity correlation with procaspase-3 expression in tumor, clinical response and adverse effects. Neurological symptoms of CNS toxicity will be assessed throughout the trial. Inclusion criteria: diagnoses of advanced malignancies (for part 1) and high grade glioma (for part 2), ECOG PS 0-2, adequate organ function. Exclusion criteria: received prior cytotoxic therapy in the last 3-6 weeks (duration based on prior therapy) or uncontrolled chronic illness. Administration and design: Part 1, PAC-1 (PO) is dosed at 75-450 mg daily (up to 5 dose levels) on days 1-21 on 28 days cycle. In Part 2, the first PAC-1 dose will be 1 dose lower than the PAC-1 MTD established in Part 1 (up to 3 dose levels).Temozolomide (PO) will be dosed at 150 mg/ m2 daily for 5 days starting on day 8 of each cycle. The study is currently enrolling patients in Part 1, NCT02355535. Clinical trial information: NCT02355535.

TPS2606

TPS2607

Poster Session (Board #298b), Sun, 8:00 AM-11:30 AM

NCI-MATCH (Molecular Analysis for Therapy Choice) a national signal finding trial. First Author: Barbara A. Conley, National Cancer Institute, Rockville, MD Background: Certain tumor types show dramatic responses to targeted treatments based on demonstration of the corresponding molecular abnormality. The NCI-MATCH study hypothesizes that patients (pts) across a range of histologies whose tumors share molecular abnormalities that have been shown to respond to a targeted treatment may respond to that same treatment, and that responsiveness or resistance may be predictable by concurrent molecular abnormalities. Methods: Pts with refractory solid tumors or lymphomas have molecular profiling on a proximate biopsy, using a validated, locked next generation sequencing (NGS) assay (143 genes; single nucleotide variants, copy number variants and fusions) and selected other molecular assays. The trial will have . 20 phase II targeted treatments (arms). Drugs are provided by various pharmaceutical companies and could be FDA approved or investigational single agents or combinations. The arms are placed under a “master protocol”, allowing new arms to be added to replace a filled cohort or to add a new molecularly based treatment. Specific rules are established to qualify the selection of the molecular subtype and to select the candidate treatment. A rules engine assigns pts to molecularly targeted treatments based on actionable mutation of interest (aMOI), as well as other factors that inform about likelihood of response. Each arm will enroll 35 pts (one stage design) with appropriate aMOI, regardless of histopathologic tumor type. The primary endpoint is response rate (5% vs 25%). Biopsy is requested upon progression. This trial opened in August 2015. Currently 798 of a planned 3000 patients have been screened. A protocoldirected pause in accrual occurred after 3 months, and interim analysis (feasibility) is ongoing. The current 10 arms and additional arms in the process of being added address molecular abnormalities in: EGFR, ERBB2, PIK3CA, PTEN, ALK, ROS, BRAF, NRAS, NF2, KIT, NF1, SMO, PTCH1, MET, FGFR, AKT, NTRK, cell cycle and PI3K pathway as well as other less common molecular abnormalities. Data from the NCI-MATCH trial is designed to identify signals from molecularly targeted treatment that will inform and accelerate further therapy development leading to clinical benefit. Clinical trial information: NCT02465060.

Poster Session (Board #299a), Sun, 8:00 AM-11:30 AM

SELECT-4: Phase I dose escalation trial of selumetinib (AZD6244, ARRY142886) in combination with durvalumab (MEDI4736) in patients with advanced solid tumors. First Author: Pasi A. Janne, Dana-Farber Cancer Institute, Boston, MA Background: Preclinical studies suggest that intermittent dosing with selumetinib, an oral, potent and highly selective, allosteric MEK1/2 inhibitor with a short half-life, should be evaluated in combination with durvalumab, a selective, high-affinity human IgG1 mAb, that blocks PD-L1 binding to PD-1 and CD80. Selumetinib may prime the immune response and also potentially inhibit T-cell function, thereby maximizing tumor cell damage and antigen presentation during the period of MEK inhibition; intermittent dosing with selumetinib will allow maximal relief of T-cell checkpoint blockade by durvalumab. The SELECT-4 study (NCT02586987) aims to evaluate this combination for the first time in pts. Methods: SELECT-4 is a Phase I, openlabel, multicenter, dose escalation study investigating the safety and tolerability of intermittent doses of selumetinib combined with durvalumab in pts with advanced solid tumors for which no standard therapy exists. Selumetinib (starting dose 50 mg po bid, increasing until the maximum tolerated dose is reached) will be given for a 7-day monotherapy run in, then on a 1 week on, 1 week off schedule, every 4 weeks in combination with durvalumab (1500 mg, iv) administered once every 4 weeks. Approximately 20–30 evaluable pts will be enrolled, with up to 6 pts in each dose escalation cohort and 6–10 evaluable pts in a paired biopsy expansion cohort. Additional cohorts may be explored as determined by the Safety Review Committee. Pts must have a WHO performance status of 0‒1 and measurable disease at baseline (RECIST 1.1). Tumor assessments will be performed at screening, then every 8 weeks until confirmed disease progression. The primary objective is to investigate the safety and tolerability of intermittent dosing of selumetinib combined with durvalumab. Secondary objectives include defining the recommended dose for selumetinib combined with durvalumab, preliminary assessment of anti-tumor activity, characterization of population pharmacokinetics for each drug when in combination, and assessment of the immunogenicity of durvalumab. Biomarkers will be assessed as an exploratory objective. Recruitment is ongoing. Clinical trial information: NCT02586987.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Clinical Pharmacology and Experimental Therapeutics TPS2608

Poster Session (Board #299b), Sun, 8:00 AM-11:30 AM

A Phase Ib, Open-Label, Multi-Center Study to Assess the Safety, Tolerability, Pharmacokinetics, and Anti-tumor Activity of AZD1775 Monotherapy in Patients with Advanced Solid Tumors: Expansion Cohorts. First Author: Todd Michael Bauer, Sarah Cannon Research Institute/Tennessee Oncology, PLLC, Nashville, TN Background: Many cancers are associated with DNA repair and cell cycle mutations that result in G1/S checkpoint deficiencies and high levels of endogenous damage and replication stress. This can lead to dependence on WEE1 kinase, which provides an important G2/M checkpoint, allowing repair of DNA damage prior to mitosis. AZD1775 is a highly selective, smallmolecule WEE1 inhibitor being developed for the treatment (tx) of advanced solid tumors. It has previously shown single-agent activity in patients (pts) with BRCA-1/2 mutations (Do, JCO 2015). We described initial safety and efficacy of the dose escalation portion of an open Phase Ib study (NCT02482311) of AZD1775 in pts with refractory solid tumors, and a dose of AZD1775 175 mg PO BID was identified (submitted to AACR 2016). Here we describe the expansion portion of this study in 6 matched cohorts: BRCA1/2-mutant and wildtype (WT) ovarian cancer (OvC); cyclin-E (CCNE1)amplified and non-amplified triple-negative breast cancer (TNBC); and any of MYC/MYCL/MYCN-amplified and all non-amplified small-cell lung cancer (SCLC). The objective of this portion of the study is to evaluate the safety, tolerability, and efficacy of AZD1775 monotherapy in these cohorts. Methods: 140 pts will be enrolled in the expansion portion: 20 pts with recurrent BRCA-1/2 WT OvC following . 3 prior therapies, $ 20 BRCA-1/2 mutant OvC pts following failure of PARP inhibitor therapy, 50 with recurrent TNBC following . 2 therapies ( $ 15 with cyclin E (CCNE1) amplified TNBC); and 50 with SCLC following . 1 therapy ( $ 15 with any of MYC/ MYCL/MYCN amplified SCLC). All pts must have measurable advanced solid tumors per RECIST, and adequate organ function. Pts will receive AZD1775 175 mg PO BID D1-3, 8-10 of a 21-day cycle. All pts will receive prophylactic antiemetics. Pts will be restaged every 2 cycles the first year, every 4 cycles thereafter, and continue until disease progression or intolerable toxicity. Tumor samples will be collected for correlative biomarker analysis. Clinical trial information: NCT02482311.

TPS2609

141s

Poster Session (Board #300a), Sun, 8:00 AM-11:30 AM

A Phase I/Ib study of MEK162 (binimetinib), a MEK inhibitor, in combination with carboplatin and pemetrexed in patients with non-squamous NSCLC. First Author: Safiya Karim, University of Saskatchewan, Saskatoon, SK, Canada Background: RAS is the most commonly mutated oncogene in lung cancer. While direct inhibitors of RAS have not yet been successfully developed, MEK inhibitors have shown activity in RAS mutated tumors including KRAS mutant NSCLC. Binimetinib (ARRAY, Boulder CO) is an oral highly selective inhibitor of MEK1/2 and has demonstrated preclinical and clinical activity. Distinct from other MEK inhibitors in clinical development, it has activity against NRAS mutant cancers, with improved PFS as a single agent in NRAS mutant melanoma compared to dacarbazine (NEMO). Methods: In this phase I/Ib study, binimetinib is combined with pemetrexed/carboplatin in advanced non-squamous NSCLC patients. Primary study objectives include determination of the RP2D of continuous binimetinib plus pemetrexed/ carboplatin, safety (NCI CTCAE v. 4) of the combination and exploration of efficacy (ORR per RECIST v.1.1). Population PK is being characterized and potential relationships between RAS mutation subtypes and response explored. A 3+3 dose-escalation design with 3 treatment groups will be used to determine the maximum administered dose and RP2D of the combination. The cohort will be expanded at the RP2D to a total of 35 patients with stratification by genotype (RAS/RAF wild type (10), KRAS G12C mutant (10), non-G12C mutant (10), BRAF/NRAS mutant (5)). Binimetinib 3045 mg BID daily (starting day -7 with a 48 hour break pre-chemotherapy) is combined with pemetrexed 500 mg/m2 IV and carboplatin AUC 5-6 mg*min/ mL. Patients who have completed 4-6 cycles of chemotherapy may continue to receive MEK162 until disease progression. Patient population: Patients must be chemo-na¨ıve with EGFR wild type, non-ALK rearranged nonsquamous NSCLC, have ECOG PS 0/1, adequate organ, marrow, and normal cardiac function. Those with prior retinal vein occlusion, coronary syndromes, active infection and significant comorbidities are excluded. Patients with stable treated CNS metastases are permitted. Accrual to cohort 1 is ongoing. Clinical trial information: NCT02185690.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

142s 3000

Developmental Therapeutics—Immunotherapy Oral Abstract Session, Sat, 1:15 PM-4:15 PM

Characterization of patients treated with a programmed cell death protein 1 inhibitor (anti-PD-1) past RECIST progression from a metastatic non-small cell lung cancer (mNSCLC) trial. First Author: Dickran Gano Kazandjian, Office of Hematology and Oncology Products/Food and Drug Administration, Silver Spring, MD Background: Anti-PD-1 monoclonal antibodies are immunotherapy agents that block T cell inhibitory signal pathways. Anecdotal cases have been reported of clinically important decrease in tumor size following initial evidence of RECIST-defined disease progression. We describe the findings in patients with treatment past RECIST-defined progression (TPP) observed in a trial of mNSCLC patients receiving an anti-PD-1. Methods: Data from a trial evaluating an anti-PD-1 for the treatment of patients with mNSCLC who progressed after platinum- doublet chemotherapy, demonstrating a clinical benefit for an anti-PD-1, was evaluated. Per protocol, patients were allowed to receive TPP if they were not experiencing rapid disease progression, had stable performance status, and written informed consent was obtained. We identified anti-PD-1-treated patients with TPP in this study and evaluated the changes in tumor burden following RECIST-defined progression in these patients. Results: In total, 71 patients received TPP with an anti-PD-1 and their best overall responses by RECIST v1.1 were progressive disease in 35 (49%), stable disease in 23 (32%), and partial response (PR) in 13 (18%); the PR rate was similar in patients with and without TPP. Among patients with TPP, RECIST-defined progression was based on an increase in the sum of longest diameter (SLD) of target lesions (TLs) in 37 (52%) patients, the appearance of new lesions (NLs) in 33 (46%), and/or unequivocal progression of non-target lesions (NTLs) in 28 (39%) patients. Among the TPP patients, 4 patients (5.6%) experienced additional tumor shrinkage. One patient with RECIST-defined progression based on $ 20% increase in SLD of TLs and three patients based on NLs or unequivocal progression of NTLs demonstrated a subsequent $ 30% decrease in SLD as compared to nadir. Conclusions: Our evaluation suggests that anti-PD-1-treated patients with mNSCLC generally do not benefit from TPP. The risks of continued treatment should be considered in light of the low likelihood of tumor shrinkage with TPP with an anti-PD-1 agent.

3001

Oral Abstract Session, Sat, 1:15 PM-4:15 PM

CheckMate 012: Safety and efficacy of first-line (1L) nivolumab (nivo; N) and ipilimumab (ipi; I) in advanced (adv) NSCLC. First Author: Matthew David Hellmann, Memorial Sloan Kettering Cancer Center, New York, NY Background: Nivo + ipi is approved in adv melanoma and has demonstrated clinical activity and manageable safety in various solid tumors. CheckMate 012 is a phase I study of nivo monotherapy or combined with other therapies in 1L adv NSCLC; we report updated results, including tumor growth dynamic (TGD) modeling, from N+I dosing schedules explored to optimize safety and permit synergistic activity. Methods: Patients ([pts] N=148, any NSCLC histology) received N+I (mg/kg) across 4 dose cohorts (Table). Primary objective was safety; secondary objectives were ORR (RECIST v1.1) and 24-wk PFS rate; exploratory endpoints were OS and efficacy by tumor programmed death ligand 1 (PD-L1) expression. The effect of N+I dosing was assessed by TGD modeling generated using individual tumor assessments; model-predicted tumor shrinkage at wk 12 was compared across cohorts. Results: Treatment-related (TR) adverse events (AEs) and select TRAEs were manageable (Table). TRAEs leading to discontinuation (DC) were comparable to nivo alone (10%), with no TR deaths. Across cohorts, ORRs ranged from 13%–39% (Table), and median duration of response was not reached. Responses were noted regardless of PD-L1 expression, with a higher magnitude of benefit in tumors that expressed PDL1. TGD modeling predicted enhanced tumor shrinkage for N3 + I1 schedules compared with nivo alone or any N1-containing schedule. Based on integrated efficacy/safety/TGD data, N3 Q2W + I1 Q6W was proposed for further evaluation. Conclusions: 1L therapy with N+I demonstrates clinical activity and a manageable safety profile. Updated safety and efficacy across cohorts (by histology, EGFR, smoking status, PD-L1 expression) and TGD modeling data will be presented. Clinical trial information: NCT01454102. N1 + I1 Q3W x 4 cycles, then N3 Q2W (n=31)

N1 Q2W + I1 Q6W (n=40)

N3 Q2W + I1 Q12W (n=38)

N3 Q2W + I1 Q6W (n=39)

77 29

73 35

74 29

69 28

13 0 6 6 3 0 0 13 13 10.6 55 16.6

5 8 8 10 0 0 0 8 25 4.9 NC 6.2

3 5 3 0 3 5 0 5 39 8.0 63 8.4

5 5 5 5 3 0 0 10 31 8.3 NC 7.7

TRAEs, % Any gr Gr 3–4 Select TRAEs, gr 3–4, % Skin Gastrointestinal Endocrine Hepatic Pulmonary Renal Hypersensitivity/Infusion TRAEs leading to DC, % ORR, % mPFS, mo 24-wk PFS rate, % Median follow-up, mo NC = not calculated; Gr = grade.

3002

Oral Abstract Session, Sat, 1:15 PM-4:15 PM

3003

Oral Abstract Session, Sat, 1:15 PM-4:15 PM

Phase Ib study of PF-05082566 in combination with pembrolizumab in patients with advanced solid tumors. First Author: Anthony W. Tolcher, START San Antonio, San Antonio, TX

Programmed death-1 blockade in mismatch repair deficient cancer independent of tumor histology. First Author: Luis A. Diaz, Johns Hopkins University, Baltimore, MD

Background: PF-05082566 (PF-2566) is a fully human IgG2 monoclonal antibody (mAb) that binds to human 4-1BB (CD137). Pembrolizumab is a humanized IgG4/kappa isotype mAb targeting PD-1. This Phase 1 study was designed to assess the overall safety, pharmacokinetics, pharmacodynamics (PD) and anti-tumor activity of this novel combination in patients with advanced solid tumors. Methods: Patients received PF-2566 (0.45 to 5.0 mg/kg) and pembrolizumab (2 mg/kg) intravenously on day 1 of 21-day cycles. Maximum tolerated dose was defined as the highest combination dose with a dose limiting toxicity (DLT) rate ,25% during the first 2 cycles per the TITE-CRM method. Results: Tumor types treated included non-small cell lung cancer (NSCLC) (n=6), renal cell carcinoma (RCC) (n=5), head & neck (H&N) (n=3), pancreatic (n=2), thyroid (n=2), and one each of small cell lung cancer (SCLC), colon, sarcoma, thymoma, and melanoma. All patients received between 0 and 9 lines of previous anticancer treatment. No DLTs were reported and 6 patients remain on therapy in this ongoing study. The number of cycles patients have received across all doses ranged from 2 to 19. Treatment emergent adverse events (AEs) were mostly Grade 1/2, with no apparent relationship between increasing doses of PF-2566 and frequency or severity of AEs. No patients discontinued study medication due to treatment related AEs. No substantial changes in inflammatory cytokines or its association with clinical symptoms were noted. The PF-2566 concentrations were above the preclinical predicted efficacious concentration and appeared to increase with increasing doses. Circulating PD responses (lymphocyte subsets) were observed across all doses of PF-2566. Six out of 23 treated patients (26%, 95% CI: 10.2%, 48.4%) had confirmed complete (CR) or partial response (PR) per RECIST 1.1: CR in SCLC (n=1), PRs in RCC (n=2), NSCLC (n=1), H&N (n=1) and anaplastic thyroid (n=1). Median duration of response has not been reached. Conclusions: The safety and efficacy profile of PF-2566 in doses up to 5.0 mg/kg in combination with pembrolizumab supports further investigation in patients with advanced solid tumors. Clinical trial information: NCT02179918.

Background: Mismatch repair deficiency (MRD) is feature of many cancers at a frequency of approximately 1 in 30 patients independent of tumor histology. Tumors with MRD are deficient in the repair of specific DNA replication errors and as a result accumulate hundreds to thousands of mutations per tumor genome. The high number of somatic mutations increase the chances for at least one of these mutations to result in a highly immunogenic neo-antigenic protein that can trigger a potent anti-tumor immune response in the presence of PD-1 blockade. Methods: To further test this hypothesis, we conducted a phase 2 study to evaluate the activity of pembrolizumab (pembro), a programmed death-1 (PD-1) antibody in MRD tumors independent of tumor histology using a basket design. Pembro was administered at 10 mg/kg every 14 days in patients with . 1 prior therapy. The co-primary endpoints were response and progression-free survival rate at 20 weeks. Secondary endpoints included disease control rate (DCR = CR +PR+SD), PFS, overall survival (OS) and safety. Results: A total of 29 patients were enrolled and treated on this study, including the following histologies: (endometrial: 9; pancreatic: 4; ampullary: 4; biliary: 3; small bowel: 3; gastric: 3; thyroid: 1; prostate: 1; sarcoma: 1). Median follow up time is 8.1 mos. Objective response and disease control rates were 48% (14/ 29, 95% confidence interval: 29-67%) and 72% (21/29), respectively. Twenty of 29 patients remain on treatment due to clinical benefit. Median overall Survival (OS) and progression-free survival (PFS) were 21 months and not reached (NR). The OS and PFS rates at 12 months were 79% and 54%, respectively, which support the durability of clinical benefit. Among the patients with an objective response, only 3 have developed secondary resistance to pembro with a median time to progression of 5 months. Conclusions: Independent of tumor histology, patients with advanced MRD cancers receive durable clinical benefit with Pembro. Clinical trial information: NCT01876511.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Immunotherapy 3004

Oral Abstract Session, Sat, 1:15 PM-4:15 PM

Correlation of WNT/b-catenin pathway activation with immune exclusion across most human cancers. First Author: Jason John Luke, University of Chicago Comprehensive Cancer Center, Chicago, IL Background: The presence of intra-tumoral CD8+T cells has been associated with clinical benefit to immunotherapy and patients can be categorized based on the presence or absence of a T cell-inflamed tumor microenvironment. Molecular mechanisms underpinning the absence of a T cell response are beginning to be understood with identification of the WNT/b-catenin pathway playing a major role in melanoma. Methods: Gene expression data from TCGA was segregated using a T cell-inflamed gene expression signature to identify patient samples by immune phenotype. Within the subset of each tumor histology lacking the T cell signature, exomic sequencing data was interrogated for mutations and pathway analysis was performed to identify gene expression changes predicted to activate the b-catenin signaling pathway. Immunohistochemistry was performed on human primary and metastatic specimens of multiple tumor histologies to correlate the presence of nuclear b-catenin with the lack of a T cell infiltrate. Results: Predicted activating mutations in CTNNB1 or inactivating mutations in Axin1, Axin2, APC1, APC2 were observed within 13 tumor histologies ranging in frequency of 9% (esophageal) to 67% (colon) of tumor specific samples. Pathway analysis identified 12 further histologies harboring increased expression of b-catenin pathway elements such as WNT ligands. Only 3 tumors (pancreatic, thyroid, paraganlioma) showed no activation correlating with absence of immune signatures. Immunohistochemistry demonstrated inverse correlations between b-catenin and CD8 +T cells. Conclusions: WNT/b-catenin pathway signaling correlates with T cell exclusion and is activated in most human solid tumors. Clinical development of b-catenin pathway inhibitors that combine with immunotherapies should be prioritized. Pathway activation by mutation Colon Melanoma (metastatic) Glioma – low grade Testicular Lung – Squamous Sarcoma Lung – Adenocarcinoma Uveal Melanoma Hepatocellular Kidney – Papillary Uterine Sarcoma Kidney – Chromophobe

3006

Pathway activation by increased expression Kidney – clear cell Mesothelioma Head and Neck Gastric Breast Ovarian Cervical Bladder Adrenocortical Uterine Prostate Glioblastoma

Oral Abstract Session, Sat, 1:15 PM-4:15 PM

A randomized, prospective evaluation comparing intensity of lymphodepletion prior to adoptive transfer of tumor infiltrating lymphocytes for patients with metastatic melanoma. First Author: Stephanie L. Goff, National Cancer Institute Surgery Branch, Bethesda, MD Background: Adoptive cell transfer (ACT), the infusion of large numbers of activated autologous lymphocytes, can mediate objective tumor regression in a majority of patients with metastatic melanoma (52/93, 56%). The addition and intensification of total body irradiation (TBI) to the preparative lymphodepleting chemotherapy regimen in sequential trials improved objective partial and complete response rates. Here, we evaluated the importance of adding TBI to the adoptive transfer of tumorinfiltrating lymphocytes (TIL) in a randomized fashion. Methods: 101 patients with metastatic melanoma, including 76 with M1c disease, were randomly assigned to receive nonmyeloablative chemotherapy with or without 1200 cGy TBI prior to transfer of TIL. The primary endpoints were complete response rate (as defined by Response Evaluation Criteria in Solid Tumors 1.0) and overall survival. Clinical and laboratory data were analyzed for correlates of response. Results: Complete response rates were 24% in both groups (12/50 v 12/51) and overall survival was also similar (median OS, 38.2 v 36.6 months; hazard ratio [HR], 1.11; 95% CI, 0.65 to 1.91, P= .71). Thrombotic microangiopathy was an adverse event unique to the TBI arm and occurred in 13/48 treated patients. With a median potential followup of 40.9 months, only one of the 24 complete responders recurred. Conclusions: Adoptive cell transfer with tumor-infiltrating lymphocytes can mediate durable complete regressions in 24% of patients with metastatic melanoma and a median survival greater than three years. Results were similar using chemotherapy preparative regimens with or without the addition of TBI . Clinical trial information: NCT01319565.

3005

143s Oral Abstract Session, Sat, 1:15 PM-4:15 PM

Correlation of peripheral and intratumoral T-cell receptor (TCR) clonality with clinical outcomes in patients with metastatic urothelial cancer (mUC) treated with atezolizumab. First Author: Samuel Funt, Memorial Sloan Kettering Cancer Center, New York, NY Background: The anti-programmed death-ligand 1 (PD-L1) antibody atezolizumab is active in patients (pts) with advanced mUC. We hypothesized that the TCR repertoire in tumors and blood may correlate with clinical benefit. Methods: 29 pts received atezolizumab 1200mg q3w on IMvigor 210 at MSKCC and had specimens available for analysis. High throughput DNA sequencing of the CDR3 region of the TCR beta chain was performed in baseline tumors (N=24) plus pre- and post-treatment peripheral blood mononuclear cells (PBMCs) (N=29 patients, 93 PBMC samples) using the immunoSEQ assay. Clonal dominance, clonal expansion and T-cell fraction were assessed for association with benefit (complete and partial response (CR, PR) vs. stable or progressive disease (SD, PD) and progression-free and overall survival (PFS, OS)). Results: The median number of T cells and unique TCRs sequenced in tumors was 1,402 and 1,085, respectively. The median number of T cells and unique TCRs sequenced in PBMCs was 147,493 and 81,705, respectively.Investigator reported best responses were 5 CR, 2 PR, 5 SD, and 17 PD.A combination of high T cell infiltration and clonality in tumors correlated with CR/PR compared to PD (p=0.02232, OR 11.5). The number of T cell clones in the tumor that expanded in the blood 3 weeks post-treatment was greater in CR/PR than in SD (p=0.01253) or PD pts (p=0.05565). Pts with lower baseline peripheral TCR clonality had improved PFS (p=0.0514) and OS (p=0.0116). Conclusions: In mUC patients treated with atezolizumab, high T cell infiltration and clonality in the tumor plus peripheral expansion of dominant tumor-resident TCR clones were associated with response. Low peripheral TCR clonality was associated with survival. These findings suggest that atezolizumab re-invigorates an oligoclonal population of tumor infiltrating lymphocytes and promotes their expansion in the blood. Low TCR repertoire clonality in the periphery at baseline may aid in preventing immune escape and may lead to improved OS. Similar observations have been seen in mice treated with radiation plus checkpoint blockade (Nature 520:373,2015). These findings warrant further study.

3007

Oral Abstract Session, Sat, 1:15 PM-4:15 PM

Efficacy of humanized CD19-targeted chimeric antigen receptor (CAR)modified T cells in children with relapsed ALL. First Author: Shannon L. Maude, Children’s Hospital of Philadelphia, Philadelphia, PA Background: Targeted immunotherapy with CTL019, CD19-specific CARmodified T cells, can produce potent, sustained responses in children with relapsed/refractory acute lymphoblastic leukemia (ALL). However, a subset of patients has limited persistence. We now report on retreatment with murine (CTL019) or humanized (CTL119) CD19-directed CAR T cells. Methods: Patient-derived T cells were transduced ex vivo with a lentiviral vector encoding a CAR with CD3z, 4-1BB, and murine or humanized antiCD19 scFv domains and activated/expanded with anti-CD3/CD28 beads. Results: Of 55 patients in CR at 1 month after CTL019 infusion (55/59 CR), 3 patients received a repeat infusion of murine CTL019 for CD19+ relapse and 17 for poor persistence at 3 and/or 6 months after initial infusion. Remission was achieved in 1/3 children treated for CD19+ relapse. Of 3 patients reinfused for CD19+ MRD, 1 progressed to CD19+ relapse, 1 became MRD- but had B cell recovery, and 1 had reduced MRD. Reinfusion induced B cell aplasia for a second time in 1/7 children treated for B cell recovery, while 6/7 children reinfused for CD19+ hematogones demonstrated continued B cell aplasia 6-21 months after repeat infusion. Of this group, 6 remained in remission 9-24 months after initial infusion, and 1 experienced a CD19- relapse. Eight children previously treated with CAR T cells (CTL019, n = 5; other, n = 3) were treated on a phase 1 study of humanized CTL119 for B cell recovery (n = 3), CD19+ relapse (n = 4), or no response to prior CAR T cells (n = 1). Cytokine release syndrome (CRS), seen in 4 patients, did not require vasopressor or respiratory support. Responses were seen in 4/8 patients with an ongoing CR of 7 months in 1 patient. 2/4 responding patients were previously resistant to reinfusion of murine CTL019. Conclusions: Reinfusion of murine CTL019 may prolong B cell aplasia in patients with early evidence of poor persistence but is less effective for B cell recovery, suggesting that examination of anti-CAR immune responses will be important to improve durable remission rates. Human CARs may overcome anti-murine immunity, and in the first study of a humanized anti-CD19 CAR, CTL119 induced remission in 50% of patients refractory to murine CAR T cells. Clinical trial information: NCT02374333.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

144s 3008

Developmental Therapeutics—Immunotherapy Oral Abstract Session, Sat, 1:15 PM-4:15 PM

Epigenetic control of CD4/CD8 lineage commitment and resistance to tumor infiltrating lymphocyte adoptive cell therapy for metastatic melanoma. First Author: David Michael Woods, NYU Langone Medical Center2, New York, NY Background: Adoptive cell therapy (ACT) with tumor infiltrating lymphocytes (TIL) is an effective approach for the treatment of metastatic melanoma, with objective response rates of approximately 50%. An increased understanding of the factors contributing to resistance to this promising therapy is needed. Methods: To determine the role of epigenetic mechanisms in TIL responses, cell infusion products from patients undergoing TIL ACT were separated by magnetic labeled antibodies into CD4 and CD8 populations, and assessed by an Illumina 450K DNA methylation array and chromatin immunoprecipitation of acetylated histone 3 with sequencing. Patient tumor samples were also assessed on a Nanostring platform for epigenetic gene and miR expression. The percentage of CD4 and CD8 T cells in TIL were assessed by flow cytometry. Results: Significant differences in DNA methylation of genes controlling CD4 lineage (CD4, CD8, RUNX3, XCL2, ZBTB7B) were found between responders (R) and non-responders (NR) (p , 0.05). Using a CD4/ CD8 differentiation signature, principal component analysis (PCA) revealed clustering based on response, with NR CD8 T cells more closely resembling CD4 T cells (p , 0.05). PCA of acetylated histone 3 also showed differences between R vsNR (p , 0.05). Nanostring analysis of tumor fragments showed decreased expression in NR in several epigenetic regulatory genes including DNA methyltransferases (DNMT1, DNMT3A) and histone deacetylases (HDAC8, SIRT4)(p , 0.05). MiR-148 and miR-185, known regulators of DMNT1, were increased in accordance with the patterns seen for DNMT1, a known regulator of CD4/CD8 lineage commitment. Higher levels of CD4/CD8 double positive TILs were found in NR infusion products. Conclusions: These results demonstrate that resistance to TIL therapy is characterized by altered DNA methylation and histone acetylation patterns accompanied by differences in genes regulating these marks. It is hypothesized that epigenetic dysregulation of CD4/CD8 lineage impairs the therapeutic efficacy of TIL therapy. These data highlight an unexplored mechanism of T-cell dysfunction characterizing resistance to TIL ACT.

LBA3010

Clinical Science Symposium, Mon, 3:00 PM-4:30 PM

Anti-CD19 chimeric antigen receptor T cells preceded by low-dose chemotherapy to induce remissions of advanced lymphoma. First Author: James Kochenderfer, Experimental Transplantation and Immunology Branch, National Cancer Institute, Bethesda, MD

The full, final text of this abstract will be available at abstracts.asco.org at 7:30 AM (EDT) on Sunday, June 5, 2016, and in the Annual Meeting Proceedings online supplement to the June 20, 2016, issue of the Journal of Clinical Oncology. Onsite at the Meeting, this abstract will be printed in the Monday edition of ASCO Daily News.

3009

Clinical Science Symposium, Mon, 3:00 PM-4:30 PM

Randomized, phase II dose optimization study of chimeric antigen receptor (CAR) modified T cells directed against CD19 in patients (pts) with relapsed, refractory (R/R) CLL. First Author: David L. Porter, Lymphoma Program, Abramson Cancer Center, University of Pennsylvania, Philadelphia, PA Background: Autologous T cells genetically modified to express a CD19targeting CAR (CTL019) can mediate potent anti-tumor effects in CLL and other CD19+ malignancies. Since CTL019 are self-replicating, and the optimal dose is unknown, we are performing a randomized phase II study of 2 CTL019 doses in R/R CLL. Methods: Eligible pts had . 2 prior therapies and progressed within 2 yrs of .2nd line therapy. Pts receive lymphodepleting chemotherapy before infusion. Stage 1 pts were randomized to 5x108 (high dose) or 5x107(low dose) CTL019; Stage 2 pts receive the chosen optimal dose. 35 pts median age 62 have been enrolled. 40% had deletion of p53, 23% failed ibrutinib, median prior therapies was 3 and all had active disease. Median follow-up is 9 mo (range 1-34). Results: 28 pts were treated on stage 1 with 24 evaluable for response (11 high dose, 13 low dose). 6/11 responded at the high dose (4 CRs) and 4/13 at the low dose (1 CR) (p=0.11) with similar toxicity. Therefore a dose of 5 x 108CTL019 was chosen as optimal for Stage 2; 21 pts have been treated at this dose with 17 evaluable for response (11 Stage 1, 6 Stage 2). 9 of the 17 responded (6 CR, 3PR). 5 remain in CR with median follow-up 26 mo (range 5-34) and1 progressed with CD19- disease. Response did not correlate with typical prognostic factors including age, prior therapies or mutated P53. In contrast, CAR T expansion correlated with response as robust CTL019 in-vivo expansion was noted in all responding pts with a median peak of 14.7% of CD3+ cells but 0.35% in non-responders. All 35 pts were evaluable for toxicity and 19 had delayed cytokine release syndrome (CRS) (7 gr 3-4) associated with high levels of IL6 and IFN-g. Tocilizumab successfully reversed CRS in 4 pts; 15 did not require intervention. Dose was not associated with CRS development or severity. Conclusions: This study identifies optimal dosing of CTL019 for R/R CLL and confirms that CTL019 undergo robust in vivo expansion with manageable CRS. CTL019 proliferation in vivo is a predictive biomarker of response. Additional biomarker studies to identify pts likely to respond will be presented. CTL019 cells can induce potent and durable responses in R/R CLL. Clinical trial information: NCT01747486.

3011

Clinical Science Symposium, Mon, 3:00 PM-4:30 PM

Sustained remissions with CD19-specific chimeric antigen receptor (CAR)modified T cells in children with relapsed/refractory ALL. First Author: Shannon L. Maude, Children’s Hospital of Philadelphia, Philadelphia, PA Background: Targeted immunotherapy with chimeric antigen receptor (CAR)-modified T cells can produce potent anti-tumor responses. We previously reported complete remissions (CR) and prolonged persistence in children and adults with relapsed/refractory (r/r) acute lymphoblastic leukemia (ALL) treated with CD19-specific CART cells (CTL019). We now report on outcomes and longer follow-up of 59 children with r/r ALL. Methods: T cells collected from the patient were transduced with a lentiviral vector encoding a CAR composed of anti-CD19 scFv, CD3z, and 4-1BB domains, activated/expanded ex vivo with anti-CD3/CD28 beads, and then infused at a dose of 107 to 108cells/kg with a transduction efficiency of 2.345%. 54/59 patients received lymphodepleting chemotherapy the week prior to cell infusion. Results: Of 59 patients aged 20mo-24y with CD19+ ALL, 44 had detectable disease prior to CTL019 cell infusion, while 15 were MRD-. 39 were treated for relapse after prior stem cell transplant (SCT). 15 patients had CNS disease within a year of infusion. At assessment 1 month after infusion, 55/59 (93%) were in CR. MRD , 0.01% by flow cytometry was achieved in 52 patients. CTL019 cells were detected in the CSF and no CNS relapses have been seen. With median follow-up 12 mo (1-43 mo), 34 patients had ongoing CR, with only 6 receiving subsequent therapy (5 SCT, 1 donor lymphocyte infusion), RFS was 76% (95% CI, 65-89%) at 6 mo and 55% at 12 mo (95% CI, 42-73%), and OS was 79% (95% CI, 69-91%) at 12 mo. 20 patients subsequently relapsed, 13 with CD19- disease. CTL019 persistence was accompanied by B cell aplasia, which continued up to last assessment (1-39 mo) in 24/34 patients with ongoing CR. Cytokine release syndrome (CRS) was seen in 88% of patients. Severe CRS requiring hemodynamic or respiratory support occurred in 27%, was associated with high disease burden, and was reversed with the anti-IL6R agent tocilizumab. We could predict development of severe CRS through regression modeling using IFNg, sgp130, and sIL1RA measured in the first 72h (sens 86%, spec 89%, AUC 0.93). Conclusions: Single-agent CTL019 immunotherapy can induce durable responses with control of CNS disease in patients with r/r ALL. Clinical trial information: NCT01626495.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Immunotherapy 3012

Clinical Science Symposium, Mon, 3:00 PM-4:30 PM

Treatment of EBV+ nasopharyngeal carcinoma with banked EBV-specific cytotoxic T cells. First Author: Susan Prockop, Memorial Sloan Kettering Cancer Center, New York, NY Background: The expression of EBV antigens in Nasopharyngeal Carcinoma (NPC) makes EBV+NPC a candidate for cellular therapy. Prior reports demonstrate the potential of autologous EBV directed T cell therapy for EBV +NPC. However, it is not possible to generate autologous EBV specific cytotoxic T cells (EBV CTLs) from all patients (pts). Methods: We treated 14 pts with metastatic EBV+NPC with adoptive transfer of EBV CTLs selected from a bank of . 300 GMP grade lines generated from EBV seropositive hematopoietic transplant donors by sensitization of T cells with irradiated Bcell lines transformed by EBV strain 95.8. EBV CTLs were selected based on matching for at least 2/8 HLA alleles and exhibiting EBV-specific cytotoxicity restricted by an HLA allele shared by the pt. Cells were administered in cycles of 3 weekly doses of 1-2 x 10e6/kg for a median of 2 cycles (range of 1-4 cycles). Results: Fourteen pts with a median age of 42 years (range 13-64) and median KPS of 90% (range 80-100%) with metastatic NPC were screened. Despite the fact that 9 pts were Asian, 3 Black and 1 Hispanic, an appropriate line was available for all 14 pts. Pts were treated for disease progression after first (N = 6) second (N = 7) or third (N = 1) line therapy. EBV CTLs infusions were well tolerated without toxicities. One pt achieved a complete response (documented by biopsy) lasting 21.2 months, two achieved partial responses lasting 3.5 and 6.7 months and one pt had stable disease lasting 7.4 months. All 4 pts who achieved CR, PR or SD are alive at 20.4 - 48.3 months post therapy. Time to initiation of subsequent therapy in the 10 pts who progressed ranged from 15 to 195 days. Of the 14 pts, two died of disease at 107 and 147 days after the start of therapy and one of toxicity due to subsequent therapy for progressive disease 43 months after the start of EBV CTL therapy. Eleven pts remain alive at a median of 18.1 months (3.0 -48.4 months) after start of therapy. Conclusions: Our findings show that the bank of EBV CTLs has sufficient diversity to treat nonCaucasian pts with NPC. We show that adoptive immunotherapy with banked EBV CTLs is well tolerated. EBV directed adoptive cellular therapy using banked 3rd party EBV CTLs may offer clinical benefit to some pts with metastatic NPC. Clinical trial information: NCT01498484.

3014

Poster Discussion Session; Displayed in Poster Session (Board #336), Sun, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sun, 4:45 PM-6:00 PM

3013

145s

Poster Discussion Session; Displayed in Poster Session (Board #335), Sun, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sun, 4:45 PM-6:00 PM

Pembrolizumab (pembro) plus ipilimumab (ipi) or pegylated interferon alfa2b (PEG-IFN) for advanced melanoma or renal cell carcinoma (RCC). First Author: Michael B. Atkins, Georgetown-Lombardi Comprehensive Cancer Center, Washington, DC Background: Pembro (anti–PD-1), ipi (anti–CTLA-4), and PEG-IFN (cytokine) are approved as monotherapy for treating melanoma. KEYNOTE-029 (NCT02089685) is a first-in-human, phase 1 study that includes independent pembro + ipi safety run-in and pembro + PEG-IFN dose-finding cohorts. Methods: Pts with advanced RCC with $ 1 prior therapy or melanoma with any number of prior therapies were eligible. Prior checkpoint therapy was not allowed. In the pembro + ipi safety run-in, pts received pembro 2 mg/kg Q3W + ipi 1 mg/kg Q3W for 4 doses, then pembro alone. Dose finding for pembro 2 mg/kg Q3W + PEG-IFN followed a toxicity probability interval-based design with 3-14 pts per dose level; the PEG-IFN dose was 1 mg/kg/wk at dose level (DL) 1 and 2 mg/kg/wk at DL2. Accrual continued until # 4/14 pts at a given DL experienced a dose-limiting toxicity (DLT) during the first 6 wk. Response was assessed by RECIST v1.1 at wk 12, every 6 wk until wk 30, and every 12 wk thereafter. The data analysis cutoff date was Oct 26, 2015. Results: The pembro + ipi run-in cohort enrolled 12 pts with melanoma and 10 with RCC. After 13.8-mo median follow-up, 5 pts remained on pembro and 59% had $ 1 grade 3-4 drug-related AE. The PR and SD rates were 25% (n = 3) and 42% (n = 5) for melanoma and 30% (n = 3) and 30% (n = 3) for RCC. The pembro + PEG-IFN cohort enrolled 4 pts with melanoma and 10 with RCC. After a median 13.2 mo of follow-up, 3 pts remained on treatment. DLTs occurred in 1/11 pts (9%) at DL1 (grade 3 depression). At DL2, 2/3 pts (67%) had a DLT (1 grade 3 elevated AST, 1 grade 4 suicide attempt). Grade 3-4 drugrelated AEs occurred in 57% of pts (55% at DL1, 67% at DL2), most commonly psychiatric disorders (21%). One pt (25%) with melanoma and 0 pts with RCC had PR; 2 pts (50%) with melanoma and 6 (60%) with RCC had SD. There were no drug-related deaths or unexpected AEs in either cohort. Conclusions: Pembro 2 mg/kg + ipi 1 mg/kg is tolerable and has promising antitumor activity in mostly heavily pretreated pts with advanced melanoma or RCC. This combination is being further evaluated in an ongoing melanoma expansion cohort. Pembro 2 mg/kg Q3W + PEG-IFN 1 mg/kg/wk is the maximum tolerated dose based on the DLT rate but appears to have limited antitumor activity. Clinical trial information: NCT02089685.

3015

Poster Discussion Session; Displayed in Poster Session (Board #337), Sun, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sun, 4:45 PM-6:00 PM

Pembrolizumab (pembro) in combination with dabrafenib (D) and trametinib (T) for BRAF-mutant advanced melanoma: Phase 1 KEYNOTE-022 study. First Author: Antoni Ribas, University of California, Los Angeles, Los Angeles, CA

Phase I study of the PD-L1 inhibitor, durvalumab (MEDI4736; D) in combination with a PARP inhibitor, olaparib (O) or a VEGFR inhibitor, cediranib (C) in women’s cancers (NCT02484404). First Author: Jung-min Lee, National Institutes of Health, Bethesda, MD

Background: The anti–PD-1 antibody pembro has demonstrated durable responses in ~30%-40% of patients (pts) with advanced melanoma. The combination of BRAF inhibitor D and MEK inhibitor T, the standard treatment for BRAF-mutant melanoma, leads to rapid responses and improved survival, but 50% of pts eventually experience disease progression around 12 mo. KEYNOTE-022 (NCT02130466) is a phase 1/2 study designed to assess safety and efficacy of the recommended doses of pembro in combination with D and T. Here, we report on the safety and efficacy of this triplet combination in BRAF-mutant melanoma in phase 1 testing. Methods: Key eligibility criteria included BRAFV600E/K-mutant stage III/IV advanced melanoma, $ 1 measurable lesion per RECIST v1.1, ECOG PS 0-1, and no prior treatment with BRAF/MEK inhibitors. A 3 + 3 design was used for dose escalation, with dose confirmation in an expansion cohort guided by the toxicity probability interval method. Pts received pembro 2 mg/kg Q3W + D 150 mg twice daily (BID) + T 2 mg daily (QD). AEs (NCI CTCAE v4.0) were collected throughout the study + 30 days; DLT observation window was the first 6 wk after dosing. Primary efficacy end point was ORR (RECIST v1.1, investigator). Results: As of the Jan 2016 data cutoff, 15 pts were enrolled across the dose determination and dose confirmation arms. The predefined criteria for progression to phase 2 were met, with DLTs reported in 3 pts: pt 1 had gr 4 neutropenia (treatment interrupted); pt 2 had gr 4 ALT increased (discontinued); pt 3 had gr 3 ALT, AST, and gamma-glutamyltransferase increased (discontinued). All events resolved. Overall, 10 (67%) pts experienced gr 3-4 treatment-related AEs, with 5 (33%) discontinuations; there were no treatment-related deaths. ORR (unconfirmed) in 15 pts was 60% (n = 9 PR, n = 2 SD, n = 3 PD). Based on these results, the recommended phase 2 regimen is pembro 2 mg/kg Q3W + D 150 mg BID + T 2 mg QD. Conclusions: The approved doses of pembro in combination with D + T administered concurrently provided a manageable toxicity profile in pts with BRAF-mutant melanoma. A phase 2 study will further evaluate safety and efficacy of this triplet combination as first-line therapy for BRAF-mutant melanoma. Clinical trial information: NCT02130466.

Background: Immune checkpoint blockade has shown clinical activity in subsets of advanced solid tumors. O and C have single agent activity in recurrent ovarian cancer (OvCa). We hypothesize O and C will complement the antitumor activity of D by increasing DNA damage through repair inhibition and hypoxia induction. This phase I study evaluated the safety and preliminary activity of the combinations of D+O and D+C. Methods: Eligible pts with PS 0-1 and good end organ function received D+O or D+C in parallel dose escalations in a 3+3 design. D+O included O tablets (200 or 300 mg bid) with D 10 mg/kg IV every 14d (dose level [DL] 1, 2) and O 300 mg bid with D 1500 mg IV every 28d (DL3). D+C tested C 20 or 30 mg daily with D 10 mg/kg IV every 14d (DL1, 2). The DLT period was one 28d cycle. Safety was assessed by CTCAEv4.0 and response by RECISTv1.1. Results: 19 pts (median age 66 [39-70], median 4 prior therapies [2-7]) were treated. D+O includes 10 OvCa and 2 triple negative breast Ca pts. D+C had 4 OvCa, 2 cervical Ca and 1 uterine leiomyosarcoma pts. The MTD for D+O is DL3. Gr3/ 4 D+O adverse events (AEs) include lymphopenia (2/12) and anemia (1/12). D+C gr3/4 AEs include hypertension (2/7), diarrhea (2/7), pulmonary embolism (2/7), pulmonary hypertension (1/7), and lymphopenia (1/7). 2 D +C DL1 pts required early discontinuation of C for pulmonary embolism and 1 pt on DL1 had C dose reduction due to recurrent gr2 fatigue and abdominal pain on cycle 2. 4 D+C DL2 pts had dose reduction on cycles 2 or 3 due to recurrent gr2 fatigue, abdominal pain and/or dyspnea. 1 PR (6+ mo) and 5 SD $ 4 mo (56%; [4-6+]) were seen in 9 evaluable D+O pts, yielding a 67% disease control rate (DCR). All were BRCA wild type. 2 PR (5, 4+ mo) and 2 SD $ 4 mo (4+ mo) were seen in 7 evaluable D+C pts, for a 57% DCR. Archival tissue samples for PD endpoints including PD-L1 expression and blood for PK are under analysis. Conclusions: The RP2D for D+O (O tablets 300 mg bid with M 1500 mg q28d) is tolerable and active in OvCa and TNBC pts without germline BRCA mutation. New D+C DLs will examine an intermittent C schedule with D 1500 mg every 28d. Phase II expansion studies of D+O in OvCa, TNBC, lung, and prostate cancers are now open to accrual. Clinical trial information: NCT02484404.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

146s 3016

Developmental Therapeutics—Immunotherapy Poster Discussion Session; Displayed in Poster Session (Board #338), Sun, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sun, 4:45 PM-6:00 PM

3017

Poster Discussion Session; Displayed in Poster Session (Board #339), Sun, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sun, 4:45 PM-6:00 PM

Safety, pharmacokinetics and efficacy of IMCgp100, a first-in-class soluble TCR-antiCD3 bispecific t cell redirector with solid tumour activity: Results from the FIH study in melanoma. First Author: Mark R. Middleton, University of Oxford, Oxford, United Kingdom

First-in-human phase 1 single-dose study of TRX-518, an anti-human glucocorticoid-induced tumor necrosis factor receptor (GITR) monoclonal antibody in adults with advanced solid tumors. First Author: Henry B. Koon, University Hospitals of Cleveland, Cleveland, OH

Background: T cell-based bispecific agents have shown activity in hematologic cancers, but solid tumor efficacy remains elusive. IMCgp100 is a bispecific biologic comprising an affinityenhanced TCR specific for gp100 and an anti-CD3 scFV. In vitro, IMCgp100 binds gp100+ melanoma cells causing redirection of cytotoxicity and induction of potent immune effects. Methods: The Phase I was conducted in HLA-A2+ pts with advanced melanoma, using a 3+3 design to define the MTD. Pts were treated with IMCgp100 (iv) weekly (QW, Arm 1) or daily (4QD3W, Arm 2) to evaluate safety, PK and efficacy. The recommended phase 2 regimen (RP2D-QW) was defined. Results: In the Ph I dose escalation,31 pts received doses from 5ng/ kg to 900ng/kg. In arm 1 dose-limiting toxicity of gr 3 or 4 hypotension was seen and associated with rapid trafficking of peripheral lymphocytes to skin and tumor. The MTD was determined to be 600ng/kg QW. IMCgp100 has an approximately dose-proportional profile with a plasma T1/2 of 5-6 hrs at the RP2D; pharmacodynamic effects of chemokine/cytokine release and lymphocyte infiltration into tumors was observed over 2 days post dose. Frequent related AEs (any grade) include rash (100%), pruritus (64%), pyrexia (50%), and periorbital edema (46%). Gr 3 or 4 related AE were observed in the first 3 weeks and include rash (23%), hypotension (6%) and lymphopenia (8%). Hypotension observed at the first doses was consistent with chemokine release (CXCL9, CXCL10) and movement of lymphocytes into tissues and, rarely associated with cytokine release syndrome (IL-6, IL10, IFNg). To mitigate risk of severe hypotension, flat dosing and an intrapatient escalation regimen of IMCgp100 were implemented. At the RP2D, 26 pts were evaluable for efficacy; confirmed PR were seen in 4 pts (2 uveal (UM) and 2 cutaneous (CM), including patients refractory to checkpoint agents) and 12 pts had SD. Of the 12 pts with SD, minor responses (.10% shrinkage) were seen in 3 pts. Conclusions: IMCgp100 is a first in class TCR anti-CD3 bispecific T cell redirector with a favorable safety profile and durable responses in melanoma (CM and UM). Ph II development in CM and UM is ongoing. Clinical trial information: NCT01211262.

Background: TRX518 is a novel, fully humanized agylcosyl IgG1 anti-GITR mAb. GITR is a member of the tumor necrosis factor receptor family expressed on T, B, & NK cells, and antigen presenting cells. GITR is expressed at high levels by Tregs and up-regulated following T cell activation; minimally expressed by nazve CD4+ and CD8+ T cells. Signaling through GITR abrogates Treg mediated suppression and enhances CD4+ and CD8+ T cell proliferation and TCR stimulation. GITR signaling may enhance host immune responses against tumor and aid in tumor rejection. This study evaluated the safety of a single dose of TRX518 in pts with advanced refractory solid tumors. Methods: Pts received a single dose of TRX518 on Day 1; doses ranged from 0.0001 to 8 mg/kg in 9 cohorts of up to 6 pts/cohort. The study was to determine safety and tolerability, maximum tolerated dose (MTD), pharmacokinetic (PK), and pharmacodynamic (PD) profiles of TRX518. Efficacy was evaluated with the immune related response criteria (irRC). Results: 40 pts enrolled; 10 melanoma; 9 NSCLC; 7 colorectal cancer; and 14 pts with 13 other solid tumors; Median age 57.5 yrs; Male 60%; Karnofsky PS 70-80% in 35% pts. Median prior systemic therapies 3 (range 1, 9); 30% with prior immunotherapy. TRX518 was well tolerated with no dose limiting toxicities, related serious adverse events or related treatment emergent adverse events (TEAE) . Grade 2. TEAE $ 15% included: cough, fatigue (28% each); vomiting, abdominal pain, nausea (18% each); dyspnea and anorexia (15% each). TRX518 exposure increased dose proportionally; mean half-life ranged from 179 to 364 hours. Saturation of T lymphocytes in peripheral blood was observed in all doses $ 0.5 mg/kg. Low levels of anti-drug antibodies was observed in 21 pts with a median titer of 1:184; little-to-no impact on TRX518 levels. Limited efficacy data was available (n = 28); 4 pts achieved a best response of stable disease (irSD). Conclusions: Single dose of TRX518 up to 8 mg/kg was safe and well tolerated. Further investigation is ongoing; a multi-dose study is planned to further evaluate anti-tumor activity. Clinical trial information: NCT01239134.

3018

3019

Poster Discussion Session; Displayed in Poster Session (Board #340), Sun, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sun, 4:45 PM-6:00 PM

Clinical activity and safety of pegylated human IL-10 (AM0010) in combination with anti-PD1. First Author: Aung Naing, Department of Investigational Cancer Therapeutics (Phase 1 Program), The University of Texas MD Anderson Cancer Center, Houston, TX Background: IL-10 has anti-inflammatory properties but stimulates the cytotoxicity and proliferation of CD8 T cells. The observation that T cell receptor mediated activation induces the expression of IL-10 receptors and PD1 on CD8 T cells, provides the mechanistic rationale for combining AM0010 and anti-PD1 in the clinic. Tolerability and anti-tumor activity of AM0010 alone was established in the ongoing phase 1 study. Objective responses were observed in pts with uveal melanoma, cutaneous T cell lymphoma and in 4 of 15 pts with RCC. Methods: Pts with advanced melanoma, RCC or NSCLC were treated with AM0010 (daily SC) and Pembrolizumab (q3wk IV). Tumor responses were monitored following irRC. Immune responses were measured by analysis of serum cytokines, activation of blood derived T cells, peripheral T cell clonality and immunohistochemistry of tumor infiltrating CD8 T cells. Results: In 19 pts, AM0010 10 mg/kg (n=13) or 20 mg/kg (n=6) in combination with anti-PD1 (2mg/kg) was well tolerated (observation period 10-15 months). All TrAEs were transient and DLTs or TrAEs leading to study discontinuation were not observed. There was no colitis, pneumonitis, or endocrine disruptions. G3/4 TrAEs were observed in 7 of 19 pts and included pruritus (1), anemia (3), thrombocytopenia (3) and malaise (1). Objective responses (PR/CR) were observed in 4 of 8 RCC pts, 2 of 5 NSCLC pts and 2 of 6 melanoma pts. 2 additional melanoma pts had tumor increase followed by decrease (pseudoprogression). AM0010 / anti-PD1 increased Th1 cytokines (IL-18, IFNg, IL-7) and the number and proliferation of PD1+ activated CD8 T cells while decreasing the proliferation of FoxP3+ Tregs and TGFb in the blood. AM0010 / anti-PD1 induced de-novo oligoclonal expansion of T cell clones in the blood without affecting total lymphocyte counts. AM0010 / anti-PD1 increased the number of tumor infiltrating Granzyme+ PD1+ CD8+ T cells in tumor biopsies of treated patients. Conclusions: AM0010 in combination with anti-PD1 is welltolerated. The clinical activity and the observed CD8 T cell activation encourages the continued exploration of AM0010 in combination with anti-PD1. Clinical trial information: NCT02009449. Best Response

SD

PR

CR

CBR (SD, PR, CR: > 24 weeks)

RCC (n=8) NSCLC (n=5) Melanoma (n=6)

4 3 2

3 2 2

1 -

5 4 2 (NR)

Poster Discussion Session; Displayed in Poster Session (Board #341), Sun, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sun, 4:45 PM-6:00 PM

A first-in-human, first-in-class phase I trial of the anti-CD47 antibody Hu5F9-G4 in patients with advanced cancers. First Author: Branimir I. Sikic, Stanford University School of Medicine, Stanford, CA Background: Hu5F9-G4 is a humanized monoclonal antibody that binds to CD47 and blocks its anti-phagocytic “don’t-eat-me” signal, leading to tumor phagocytosis and cross-presentation to T cells. CD47 is highly expressed on red blood cells (RBCs) and many human cancers, and less on normal tissues. Inhibition of CD47-SIRPa signaling by Hu5F9-G4 has broad antitumor activity in human tumor xenografts. In primate toxicology, Hu5F9-G4 caused a transient anemia that was mitigated by a single low Priming Dose enabling subsequent higher Maintenance Doses. Methods: Adults with solid tumors were enrolled in this two part dose-escalation study: Part A to determine the optimal Priming Dose and Part B to determine the optimal Maintenance Dose. Primary objectives were to assess safety and tolerability; secondary objectives were to assess PK and pharmacodynamics (PD). CD47 receptor occupancy was evaluated as a PD marker on peripheral blood RBCs. Results: To date, 16 patients have been enrolled, 11 in Part A and 5 in Part B. In Part A, 0.1, 0.3, and 1 mg/kg doses were well-tolerated. 2/2 pts at 3 mg/ kg had G3 DLTs: G3 abdominal pain; and G1 RBC hemagglutination [H] with G2 headache. The 1 mg/kg dose resulted in a mean RBC CD47 receptor occupancy of 91% 6 3% [95% CI 86 – 93%], indicating target saturation with no G3 anemia, and was selected as the Priming Dose. In Part B, the 1 mg/kg Priming Dose followed by 3 mg/kg Maintenance Dose was well tolerated. The study is ongoing with the current cohort at 1 mg/kg followed by 10 mg/kg weekly. Hu5F9-G4-related AEs include anemia (11 G1, 5 G2), hyperbilirubinemia (5 G1, 3 G2, 1 G3), headache (9 G1, 1 G2), H on peripheral blood smears (9 G1), nausea (3 G1), and retinal toxicity on serial retinal imaging (1 G2 cotton wool spot). Most adverse events were associated with the Priming Dose and were reversible. Mean peak Cmax levels were 78 6 32 mg/mL on the 3 mg/kg Maintenance Dose. Two patients (adenoid cystic ca) receiving weekly doses in Part A (0.1 mg/kg and 1 mg/kg) had stable disease for 16 and 8 months on continuous weekly drug dosing, respectively. Conclusions: Hu5F9-G4 was well tolerated at 3 mg/kg weekly, with a 1 mg/kg Priming Dose. No G3 anemia has occurred. Testing of higher Maintenance Doses is ongoing. Clinical trial information: NCT02216409.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Immunotherapy 3020

Poster Discussion Session; Displayed in Poster Session (Board #342), Sun, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sun, 4:45 PM-6:00 PM

Phase 2 trial of the indoleamine 2,3-dioxygenase pathway (IDO) inhibitor indoximod plus gemcitabine/nab-paclitaxel for the treatment of metastatic pancreas cancer: Interim analysis. First Author: Nathan Bahary, University of Pittsburgh Medical Center/ University Cancer Center, Pittsburgh, PA Background: Indoleamine 2,3 dioxygenase (IDO) is a tryptophancatabolizing enzyme that plays a key role in the normal regulation of peripheral immune tolerance. Tumors also employ this mechanism. In cancer, IDO can either be expressed by the tumor cells themselves, or induced indirectly in host antigen presenting cells. In these settings, IDO mediates an acquired immune tolerance, allowing evasion of immune mediated destruction. Indoximod is a broad IDO pathway inhibitor, shown to potentially interfere with multiple targets. The combination of gemcitabine and nabpaclitaxel is a current SOC for metastatic pancreas cancer. Pre-clinical models have demonstrated synergy between IDO inhibition with indoximod and chemotherapy. The Phase 1 results demonstrated safety of the combination and a promising preliminary objective response rate. Additionally, a delay in the occurrence of the observed responses and the durability of the observed responses suggested an immune mediated mechanism. Methods: Patients are treated with indoximod (1200mg BID continuous dosing) in combination with gemcitabine / nab-paclitaxel (1000mg/m2 / 125mg/m2 q week x3 per 4 week cycle) in continuous 4 weeks cycles. Eligible patients have treatment na¨ıve metastatic pancreatic cancer or 1st line salvage therapy after previous resection and adjuvant therapy. Treatment continues until disease progression or toxicity. Primary endpoint is overall survival. Secondary endpoints include response rate and progression free survival. Target enrollment is 80 patients in Phase 2. Results: At this interim analysis, 30 patients to date have been enrolled in Phase 1 / 2 and completed protocol treatment at least through imaging at the end of cycle 2. Of these, 11 (37%) have demonstrated an objective response by RECIST criteria including one patient with a confirmed complete response. One adverse event of immunological significance (colitis) was observed and required study withdrawl. Conclusions: The combination of indoximod and gemcitabine / nab-paclitaxel continues to demonstrate a promising objective response rate. Updated results will be presented. Clinical trial information: NCT02077881.

3022

Poster Discussion Session; Displayed in Poster Session (Board #344), Sun, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sun, 4:45 PM-6:00 PM

3021

147s

Poster Discussion Session; Displayed in Poster Session (Board #343), Sun, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sun, 4:45 PM-6:00 PM

Pilot trial of intratumoral (IT) G100, a toll-like receptor-4 (TLR4) agonist, in patients (pts) with Merkel cell carcinoma (MCC): Final clinical results and immunologic effects on the tumor microenvironment (TME). First Author: Shailender Bhatia, University of Washington - Fred Hutchinson Cancer Research Center, Seattle, WA Background: MCC is an aggressive skin cancer with suboptimal therapies. Despite persistent expression of Merkel cell polyomavirus in ~80% of pts, MCC tumors are able to evade host immunity. G100 consists of glucopyranosyl lipid-A (GLA), a TLR-4 agonist, administered IT. In preclinical models, G100 activates dendritic cells, T cells and other effector immune cells, and triggers local and systemic (abscopal) anti-tumor responses. G100 may overcome immune suppression in the MCC TME and lead to effective systemic anti-tumor responses. Methods: 10 MCC pts were enrolled to receive IT G100. Pts with loco-regional MCC (Cohort A) received G100 5 mg IT on days 1, 8 followed by surgery & radiation (RT) starting in week 4. Pts with metastatic disease (Cohort B) received G100 5 mg IT on days 1, 8 and 22 of a 6-week long cycle 1. For cycles 2-4, pts could receive 8 Gy RT to the injectable tumor on day 0 followed by G100 on days 1, 8, 15, 22, 29, and 35. Pre- and post-treatment tumor biopsies and blood were collected for immune monitoring. Results: All 10 pts (3 cohort A; 7 cohort B) completed 1 or more cycles of IT G100. Treatment-related AEs were mostly grade 1/2 (injection site reactions). 2/3 Cohort A pts with stage IIIB MCC are recurrence-free at 23+ and 19+ months (mo); one of these pts had pathologic CR after 2 G100 injections. Of 7 pts with stage IV MCC in Cohort B, 2 pts had PRs and are progression-free at 14+ and 13+ mo and 5 pts had PD. Multispectral immunohistochemistry of tumor biopsies from responders demonstrate increased inflammation with infiltration of CD8 and CD4 T cells following IT G100; T-cell receptor sequencing supports infiltration of new clonotypes rather than enrichment of existing ones. RNA expression analyses exhibit global activation of immune-related genes within the TME, including proinflammatory chemokines, cytokines, and PD-L1. Conclusions: In this pilot human trial in MCC pts, IT G100 had acceptable safety and encouraging clinical efficacy. Treatment induced inflammatory changes in the TME. The inflamed TME with PD-L1 expression after IT G100 suggest potential for synergy with inhibitors of the PD-1/PD-L1 axis. Clinical trial information: NCT02035657.

3023

Poster Discussion Session; Displayed in Poster Session (Board #345), Sun, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sun, 4:45 PM-6:00 PM

Phase I trial combining ipilimumab + high dose stereotactic radiation: Results and serum immune correlates. First Author: Chad Tang, The University of Texas MD Anderson Cancer Center, Houston, TX

Overall survival in patients with metastatic melanoma treated with concurrent ipilimumab and radiotherapy. First Author: Kristian Michael Koller, Penn State College of Medicine, Harrisburg, PA

Background: Preclinical models suggest systemic disease control with ipilimumab (IPI) and radiation. Here we conducted a phase I trial of IPI and high dose stereotactic body radiation therapy (SBRT) to determine the MTD and early efficacy signals. Methods: IPI 3mg/kg was administered q3 wks for 4 cycles. SBRT was sequenced concomitant (C1D2) or sequentially (C2D7) with IPI in 5 groups: concurrent liver 50 Gy in 4 fractions(Gp1), sequential liver 50 Gy(Gp2), concurrent lung 50 Gy(Gp3), sequential lung 50 Gy(Gp4), and sequential lung/liver 60 Gy in 10 fractions(Gp5). MTD for each group was determined using a 3+3 dose de-escalation design. Blood was collected for correlative analyses by the MDACC Immunotherapy Platform. Expression of multiple immune markers was assessed among 3 T cell populations via flow cytometry: total CD8+ [CD3+CD8+], CD4+ T effectors [CD4 Teffs : CD3 + CD4+FOXP3-], and CD4+ T regulatory cells [CD4 Treg : CD3+CD4+FOXP3 + CD127-/lo]). Significance was assessed via the Wilcoxon rank-sum test. Results: 35 pts initiated IPI, of which 31 completed SBRT. The most common histologies were: 8 NSCLC, 3 sarcoma, 3 colorectal, and 3 RCC. 2 DLTs were observed: a patient in Gp1 experienced gr3 pancreatitis and lipase elevation and a patient in Gp2 experienced gr3 bilirubin and AST increase. 13 pts (37%) experienced gr3+ AEs, with 2 pts taken off protocol due to toxicity. The most common gr3 AEs were colitis (n = 5, 14%) and rash (n = 3, 10%). The majority of AEs was self-limiting and required only outpatient supportive care. Response outside of the SBRT field was assessed in 31 pts with follow up imaging. 3 PRs were observed and 7/31 pts (23%) exhibited clinical benefit (CB = CR, PR, or SD $ 6 months). 4 of 4 pts who achieved CB exhibited an increased ratio of CD8:CD4 T cells after SBRT over pretreatment reference as compared to 1/7 in patients who did not achieve CB (P , 0.05). Patients treated with liver SBRT showed an increase in the proportion of CD8+ T cells expressing ICOS, GITR, and LAG3 compared to lung SBRT (all P , 0.05). Conclusions: The combination of SBRT and IPI (3mg/Kg) was safe with preliminary signs of efficacy. Correlative studies suggest differences in early systemic immune activation. Clinical trial information: NCT02239900.

Background: There is a growing body of evidence supporting the complementary roles of radiation therapy and immunotherapy in the treatment of malignancy. The mechanism of this phenomenon is unclear, but is thought to be mediated by synergistic mechanisms of immune activation. Ipilimumab (Ipi) is a monoclonal antibody used in the treatment of malignant melanoma that inhibits cytotoxic T-Lymphocyte antigen 4 (CTLA-4), an immunologic checkpoint on T cells. There are a few small studies suggesting an interaction between ipilimumab and radiation therapy, but larger systematic studies supporting these observations are limited. Methods: We conducted a retrospective analysis to evaluate all patients treated with ipilimumab for metastatic melanoma at a single institution from May 2011 to June 2015. Patients were grouped into those that had received concurrent radiation therapy while on ipilimumab (Ipi-RT) treatment, and those who did not. We then evaluated the treatment response following completion of ipilimumab. Primary outcome measured was rate of complete response (CR) in both radiated and distant disease sites, measured by RECIST criteria. Secondary outcome measures included ORR and OS. Results: A total of 101 patients received ipilimumab in the prespecified time frame. Seventy received Ipi-RT and 31 received ipilimumab without concurrent RT. Among the 70 who received Ipi-RT, 18 (25.7%) achieved a CR, compared to 2 out of 29 (6.45%) in the Ipi alone group [Odds ratio 5.02 (95%CI 1.09-23.2), p = 0.04]. Median OS was significantly increased in the concurrent Ipi-RT arm at 21 months vs 10 months for Ipi alone (p = 0.025). Conclusions: Patients treated with concurrent ipilimumab plus radiation therapy had a statistically significant increase in rates of complete response and median OS than those treated with ipilimumab alone. Prospective studies are needed to validate these findings.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

148s 3024

Developmental Therapeutics—Immunotherapy Poster Discussion Session; Displayed in Poster Session (Board #346), Sun, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sun, 4:45 PM-6:00 PM

A first-in-human study of REGN2810, a monoclonal, fully human antibody to programmed death-1 (PD-1), in combination with immunomodulators including hypofractionated radiotherapy (hfRT). First Author: Kyriakos P. Papadopoulos, START Center for Cancer Care, San Antonio, TX Background: Monotherapy with PD-1 pathway blockade is effective in a significant minority of patients with various malignancies. Combination therapy may further augment immune and clinical responses. This phase 1 multicenter study assessed the safety and preliminary anti-tumor activity of REGN2810, an anti-PD-1 antibody, as monotherapy and in combination with other anti-tumor treatments in patients with advanced solid tumors. Methods: REGN2810 was evaluated at 1, 3, or 10 mg/kg IV q 2 weeks (W) as monotherapy, and at 1 and 3mg/kg IV q 2W in combination with hfRT (either 9 Gy X 3 days or 6 Gy X 5 days in cycle 1 week 2), or with (CTX) or CTX + hfRT. A standard 3 + 3 dose escalation design was used, with additional patients (up to 6 total) to confirm safety. DLT monitoring period was 28 days. Tumor response (RECIST 1.1; in non-irradiated lesions) was evaluated every 8 weeks. Results: As of Feb 2016, 58 pts (median age 57.5 yrs, 57% male) have been treated in the dose escalation/safety confirmation portion of study: 18 at 1 mg/kg REGN2810 (mono: n= 6; with RT: n=12); 34 at 3 mg/kg (mono: n=6: with RT: n=12; with CTX: n=6; with RT+CTX: n=10); and 6 at 10 mg/kg alone. There were no DLTs. The most common drug-related adverse events were fatigue (n=14,24.1%), arthralgias (n=7,12.1%), nausea (n=6,10.3%), influenza like illness and rash (each n=5, 8.6%). Drug related adverse events $ grade 3 were transaminase elevation (n = 2), anemia (n=1), and anti Hu associated paraneoplastic encephalitis (n=1). Ongoing PK analyses indicate that steady-state is reached within 8 doses of q2W dosing with 2-3 fold accumulation. 43 pts were evaluable for response. Partial/ unconfirmed partial responses (uPR) were seen in 9/22 (40.9%) pts who received hfRT and in 2/21(9.5%) pts who received monotherapy. 27/43 (62.8%) pts had disease control (CR/PR/uPR/SD). Multiple expansion cohorts are evaluating REGN2810 at the recommended dose of 3 mg/kg q 2 weeks combined with hfRT and other anti-tumor regimens. Conclusions: REGN2810 has been well-tolerated to date. Responses appear augmented when hfRT is added to REGN2810, suggesting abscopal effects. Clinical trial information: NCT02383212.

3026

Poster Session (Board #348), Sun, 8:00 AM-11:30 AM

The peripheral blood TCR repertoire to facilitate patient stratification for immune checkpoint blockade inhibition in metastatic melanoma. First Author: Simone M. Goldinger, University Hospital Zurich, Zurich, Switzerland Background: Anti-PD-1 and/or anti-CTLA-4 antibodies show durable responses in metastatic melanoma. Measuring the diversity of T lymphocytes in pre-treatment liquid biopsies may help stratify patients for immunotherapy. Methods: In this retrospective blinded study, we used a multi-Nplex polymerase chain reaction (PCR) assay to measure TCR combinatorial diversity from genomic DNA in peripheral blood from melanoma patients treated with anti-CTLA-4 (n = 40) or anti-PD-1 (n = 36). A receiver operating characteristic (ROC) curve was used to determine the threshold for dichotomized analysis (i.e. low vs. high diversity) according to RECIST 1.1. Association with benefit in the low vs. high groups was assessed through a Fisher’s exact test. OS, PFS, and ORR were tested with ALC, ANC, LDH, S100 levels, and TCR diversity (DS) in a univariate model. Significant (p , 0.10) candidate prognostic factors from the univariate analysis were included in a multivariate Cox regression to define the added value of each variable on ORR, OS and PFS prediction. Results: A diversity score (DS) , 20% was associated with progressive or stable disease (p = 0.016) in antiCTLA-4 treated patients. In contrast, a DS , 20% was associated with partial or complete response in 5 of the 12 responders (p = 0.0021) receiving anti-PD-1, and all non-responders (24) had a high DS . 20%. All low DS patients were non-responders in the anti-CTLA-4 cohort (100% negative predictive value), and were all responders in the anti-PD1 cohort (100% positive predictive value). In multivariate analysis with ALC, ANC, LDH, and age, a DS , 20% was an independent predictive marker of response to antiPD-1 (p = 0.0001), with ALC . 1.4 (p = 0.027) and ANC . 4.8G/L (p = 0.016). A DS , 20% was found to be an independent prognostic marker of PFS (p = 0.040) with ANC . 4.8G/L (p = 0.0071). Conclusions: Patients with low DS did not benefit from anti-CTLA-4. In contrast, low DS patients receiving anti-PD-1 benefitted from the therapy. Thus, the TCR repertoire is a predictive biomarker of response that might help physicians choose best immune checkpoint modulators. Several studies are ongoing to validate these results.

3025

Poster Session (Board #347), Sun, 8:00 AM-11:30 AM

PD-L1 expression in advanced NSCLC: Primary lesions versus metastatic sites and impact of sample age. First Author: Anita Midha, AstraZeneca, Alderley Park, United Kingdom Background: Determination of PD-L1 levels in tumors can help establish patient suitability for PD-1/PD-L1 targeted immunotherapies in NSCLC. Biopsies may be taken from different sites (primary or metastatic) and may also vary in age, particularly if archived tissue is used. Understanding the impact of different sample types on PD-L1 expression will help determine suitability for testing. Methods: FFPE samples from 1546 patients screened for the ATLANTIC (NCT02087423) trial were successfully tested using the Ventana PD-L1 IHC (SP263) test positivity cut off $ 25% cells with tumor membrane expression. For patients screened for cohort 2, PD-L1 expression was reported in primary versus metastatic lesion, and by age, sex, ethnicity & smoking status. In addition, 88 commercial contemporaneous paired primary & metastatic samples were tested. Results: PD-L1 prevalence by sample age in ATLANTIC , 3 months, 31.2% (n = 1191); $ 3 months to 1 year 32.8% (n = 118), 1 - 3 years 29.1% (n = 173), . 3 years 13.3% (n = 83). For 112 patients, both archival & recently acquired (, 3 month) samples were available; concordance with recent samples was highest with archival samples less than 3 years old (76.2%). In patients screened for cohort 2, PD-L1 prevalence did not differ by demographic groups (Table 1). Prevalence in primary & metastatic samples was 35.0% & 33.2% retrospectively (p = 0.647). Concordance between paired contemporaneous commercial primary & metastatic samples was 39/44 (88.6%). Conclusions: PD-L1 expression was not enriched in any demographic group. Similarity of PD-L1 prevalence in metastatic and primary lesions and the concordance between paired samples supports the use of either for determination of PD-L1 status. Data also support use of samples up to 3 years old for determining PD-L1 status, which is relevant if a recent sample is unavailable. Clinical trial information: NCT02087423. Patients screened for ATLANTIC cohort 2 (n = 555)

Sex Ethnicity Smoking status Age

3027

Characteristic (n)

PD-L1 Prevalence

Male (348) Female (207) Caucasian (430) Asian (119) Other (6) Smoker (458) Non-smoker (97) . 65 (222) , 65 (333)

113 (32.5%) 69 (33.3%) 140 (32.6%) 41 (34.5%) 1 (16.7%) 151 (33.0%) 31 (32.0%) 76 (34.2%) 106 (31.8%)

P value 0.835 0.699 (Caucasian v Asian) 0.846 0.556

Poster Session (Board #349), Sun, 8:00 AM-11:30 AM

Tumor cell-free DNA copy number instability (CNI) to predict therapeutic response to immunotherapy prior to cycle 2. First Author: Glen J. Weiss, Cancer Treatment Centers of America, Goodyear, AZ Background: Tumorcell-free DNA (cfDNA) provides minimally invasive patient specific biomarkers to monitor tumor burden. Gains and losses of chromosomal regions have been detected in plasma as copy number aberrations (CNAs). Tregs are reported to be modulated by immunotherapy (immuno). We measured CNAs changes during treatment by computing a genomic copy number instability index (CNI) of cfDNA and Treg-specific demethylation region (TSDR) as measure for Tregs% of leukocytes compared to response. Methods: In this prospective study, prior to treatment and before each cycle, extracted plasma-DNA was subjected to shallow whole genome sequencing with a post mapping (HG19) read coverage of 24,000fold per 5.5Mbp bin. Read counts were transformed into (log2 ratio) Zvalues, and a CNI-score was calculated. % of TSDR+ leukocytes (TSDR%) were quantified with digital PCR from PBMC DNA. Primary endpoint was best overall response by imaging (irRECIST and RECIST 1.1). Hypotheses were: a) response to immuno is reflected by CNI change vs. baseline during therapy an b) alters TSDR%. Outcome was unblinded for analysis. Results: Of 27 enrolled patients (pts), 23 were assessable for response: 4 advanced melanoma (MEL), 2 renal cell carcinoma (RCC), 5 gastrointestinal, 4 pulmonary, 3 breast, 1 ovarian cancers, 3 pancreatic adenocarcinomas, and 1 sarcoma. MEL and RCC received interleukin-2, while the rest received anti-PD-1 with chemotherapy on trials. Median age was 58 years, 12 were women. CNI was measured in 69 samples. Mean baseline cfDNA was 8,076 (CI90th:2894-13,258cp/mL) and CNI was 2,222 (CI90th:1,1623,282). Pts with response or stable disease (n = 12) showed a significant decrease in CNI before cycle 2 (C2) (Mean: -1,386 vs.+271; p , 0.02), before cycle #3 (-2524 vs.+698, p , 0.005), and thereafter (p , 0.002) compared to pts with disease progression. cfDNA was not correlated to CNI nor response. TSDR% showed a decrease in all patients (p , 0.02) with immuno, independent of response. Conclusions: CNI change predicted response before C2 ~3-12 weeks prior to scan results, and therefore may serve as early predictor of therapeutic response to immuno. Immuno lowers Tregs % in blood, which might reflect T-cell activation.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Immunotherapy 3028

Poster Session (Board #350), Sun, 8:00 AM-11:30 AM

Diagnostic PD-L1 immunohistochemistry in NSCLC: Results of the first German harmonization study. First Author: Andreas H. Scheel, Institute of Pathology, University Hospital of Cologne, Cologne, Germany Background: Immunohistochemistry for PD-L1 (PD-L1 IHC) may be predictive for anti-PD-1 and anti-PD-L1 immunotherapy in non-small cell lung cancer (NSCLC). Several PD-L1 IHC assays with individual reagents and scoring-criteria are developed in parallel. Biomarker testing and clinical decision making would profit from one harmonized ’PD-L1 status’. Methods: Fifteen NSCLC resection specimens were centrally stained with the PD-L1 IHC assays 28-8, 22C3, SP142 and SP263. The slides were scored by nine German pathologists. Proportions of PD-L1 positive tumor cells (TC) and immune cells (IC) were scored according to a 6-step scoring system that integrates all cut-off criteria employed by the four assays. Interobserver and interassay concordance were analyzed with ’R’ Version 3.2.3. Results: Proportion-scoring of PD-L1 positive TC showed moderate interobserver concordance coefficients for the 6-step scoring system as well as a simplified 6-step score (Light’s kappa = 0.4 - 0.5). The integrated dichotomous proportion cut-offs ( $ 1%, $ 5%, $ 10%, $ 50%) showed good concordance coefficients (k = 0.6 - 0.8). The proportion-scoring of PDL1 positive IC yielded low interobserver concordance coefficients both for the 6-step-score (k , 0.2) and the dichotomous cut-offs (k = 0.12 - 0.25). The staining-intensity yielded low-to-moderate concordance for TC and IC. The assays 28-8 and 22C3 stained comparable TC-proportions. In some cases, SP142 stained fewer TC but more IC, SP263 stained more TC and IC. The differences in TC-proportions would translate into different classifications by any of the dichotomous cut-offs. Conclusions: The data indicate that unified PD-L1 IHC scoring-criteria for tumor cells are feasible while scoring of immune cells requires detailed training. The four tested PD-L1 assays do not show comparable staining-patterns in all cases of NSCLC. The results obtained by each assay are not interchangeable. Thus, more studies are required to archive a harmonized ’PD-L1 status’ in NSCLC. In particular, more data on the predictive value of one assay for multiple substances are needed.

3030

Poster Session (Board #352), Sun, 8:00 AM-11:30 AM

3029

149s

Poster Session (Board #351), Sun, 8:00 AM-11:30 AM

Association between T cell repertoire diversification and both clinical response as well as toxicity following immune checkpoint blockade in metastatic cancer patients. First Author: David Yoonsuk Oh, University of California San Francisco, Helen Diller Family Comprehensive Cancer Center, San Francisco, CA Background: Immune checkpoint inhibitors can elicit clinical responses to a range of tumor types but also significant immune-mediated toxicities (immune-related adverse events, or IRAEs). Both response and toxicity may be mediated by the generation or amplification of specific T cell clones, however it remains unclear whether changes in the T cell receptor (TCR) repertoire after checkpoint blockade are overlapping or distinct for these two outcomes. Methods: Next-generation TCR beta chain sequencing was performed on peripheral blood mononuclear cells from metastatic castrateresistant prostate cancer patients receiving ipilimumab and GM-CSF as part of a phase I/II trial to assess treatment-induced changes in the TCR repertoire and how they relate to clinical outcomes. Clinical response was defined by a . 50% PSA decline on treatment. Toxicity was defined as any IRAE developing on treatment. Results: We found that treating metastatic cancer patients with ipilimumab leads to T cell repertoire diversification as soon as 2 weeks after the first dose, as measured by a significant change in the clonality index which provides a metric of TCR repertoire diversity at each timepoint. Interestingly, clinical response is significantly correlated with increased repertoire diversity (p = 0.01 for change in clonality between weeks 0 and 2 for PSA responders, p = 0.055 for non-responders). At the same time, development of IRAEs during the study is also significantly correlated with increased diversity (p = 0.023 for change in clonality for IRAE patients, versus p = 0.057 for non_IRAE patients). Repertoire changes in IRAE patients also include increases in the frequency of TCR clones with treatment, including generation of novel clones. Conclusions: Early ipilimumab-induced changes in T cell clonality are associated with both clinical response and toxicity in prostate cancer. These observations are consistent with the overlap that is often seen clinically between clinical responders and IRAEs. Our results indicate that increased repertoire diversity may be a common initial event leading to both beneficial and deleterious outcomes with CTLA-4 blockade.

3031

Poster Session (Board #353), Sun, 8:00 AM-11:30 AM

Archival vs new tumor samples for assessing PD-L1 expression in the KEYNOTE-010 study of pembrolizumab (pembro) vs docetaxel (doce) for previously treated advanced NSCLC. First Author: Roy S. Herbst, Yale University School of Medicine, Yale Cancer Center, New Haven, CT

Occurrence of PDL1/2 copy number gains detected by FISH in adeno and squamous cell carcinomas of the lung and association with PDL1 overexpression in adenocarcinomas. First Author: Hans-Ulrich Schildhaus, InInstitute of Pathology, University Hospital Gottingen, ¨ Gottingen, ¨ Germany

Background: In KEYNOTE-010 (NCT01905657), pembro was superior to doce for OS in patients (pts) with PD-L1–positive advanced NSCLC that progressed after $2 platinumdoublet chemotherapy cycles (P = .0002 for 2 mg/kg, P , .0001 for 10 mg/kg for PD-L1 tumor proportion score [TPS] $50%; P= .0008 and , .0001 for PD-L1 TPS $1%). We compared outcomes in pts enrolled based on PD-L1 in archival vs new tumor samples. Methods: PD-L1 was assessed centrally by IHC (22C3 antibody) in archival or new (ie, no intervening therapy between collection and start of study drug) tumor samples. 1034 pts received pembro 2 or 10 mg/kg Q3W or doce 75 mg/m2Q3W for 24 mo or until progression, intolerable toxicity, or other reason. Response was assessed by RECIST v1.1 every 9 wk. Survival was assessed every 2 mo. Primary end points were OS and PFS in the TPS $50% and $1% populations. Pembro doses were pooled for this prespecified analysis. Results: Enrollment was based on archival samples in 456 pts (44%) and new samples in 578 (56%). Median (range) time between sample collection and PD-L1 assessment was 250 d (3-2510) and 11 d (1-371), respectively. PD-L1 TPS was $50% for 40% of archival and 45% of new samples. Archival samples were used in 48% of 222 pts with squamous and 43% of 724 pts with nonsquamous histology. Pembro significantly improved OS for TPS $50% and $1% in pts enrolled based on new and archival samples (Table). The PFS benefit pembro vs doce was similar in archival and new samples (Table). Conclusions: Pembro provided superior OS over doce regardless of whether new or archival samples were used to assess PD-L1. Incidence of PD-L1 TPS $50% was similar in archival and new samples. These data suggest a new biopsy may not be required for this predictive assay, sparing pts from risks associated with sample collection and avoiding resource utilization. Clinical trial information: NCT01905657.

Background: PDL1 expression measured by immunohistochemistry (IHC) is used as predictive biomarker for immune checkpoint inhibitors and is becoming clinical routine at least in pulmonary adenocarcinomas. We aimed to evaluate the presence of PDL1/2 gene cluster copy number gains in NSCLC by FISH and correlate these data to PDL1 IHC levels in adenocarcinomas (AC). Methods: We investigated 107 AC and 78 squamous cell carcinomas (SQCLC) for PDL1/2 FISH. Furthermore, AC were analyzed by IHC with the clone 28-8 stained on a DAKO platform. Cut-off level was 1% ( $ 5% highly positive). FISH was carried out with the ZytoLight SPEC CD274,PDCD1LG2/ CEN 9 Dual Color Probe. The FISH evaluation criteria included CD274, PDCD1LG2/CEN 9 ratio as well as average gene copy numbers, proportions of tumor cells with $ 4 PDL1/2 and $ 5 PDL1/2 signals as well as gene clusters. Results: 28/178 (16%) of NSCLC were FISH positive, 39% (11/ 28) thereof fulfilled the criteria for high level PDL1/2 amplification (PDL1,2/ CEN9 ratio $ 2.0 or average gene copy number per cell $ 6.0 or $ 10% of tumor cells containing $ 15 gene signals), 4% (1/28) for intermediate level copy number gain ( $ 50% of cells containing $ 5 gene signals) and 57% (16/28) for low level copy number gain ( $ 40% of tumor cells with $ 4 PDL1/2 signals). Overall PDL1/2 amplifications occur at almost equal frequencies in AC and SQCLC, 18% and 13% respectively, but high level amplifications are more often seen in SQCLC (7/78 vs. 4/107 in AC). FISH results were independent of KRAS, EGFR, ALK, ROS1 or MET status. 35/ 107 AC (33%) were PDL1 IHC positive (28 high, 7 low positivity). The correlation between IHC and FISH was statistically significant (Chi square: p , 0.001). 15/18 of FISH positive AC were IHC positive (vs. 3/66 IHC negatives). 75% (3/4) high level amplified AC were highly IHC positive. Conclusions: PDL1/2 gene copy number gains occur in AC and SQCLC, but high level amplifications are more frequent in SQCLC. PDL1/2 gains seem to be associated with PDL1 overexpression in a subset of AC. PDL1/2 FISH could contribute to our understanding of PDL1 expression and could therefore be a valuable adjunct biomarker in upcoming trials with PD1/PDL1 inhibitors.

TPS ‡50% Pembro/Doce OS Median, mo HR (95% CI) PFS Median, mo HR (95% CI)

TPS ‡1%

Archival n = 119/65

New n = 171/87

Archival n = 300/155

New n = 390/188

11.5/7.5 0.60 (0.40-0.90)

NR/8.3 0.44 (0.29-0.66)

10.5/8.3 0.70 (0.54-0.89)

12.6/8.6 0.64 (0.50-0.83)

3.9/4.0 0.64 (0.45-0.90)

6.3/4.3 0.54 (0.39-0.75)

2.9/3.8 0.81 (0.65-1.01)

4.1/4.2 0.86 (0.70-1.07)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

150s 3032

Developmental Therapeutics—Immunotherapy Poster Session (Board #354), Sun, 8:00 AM-11:30 AM

Multiplex bead-based measurement of humoral immune responses against tumor-associated antigens in stage II melanoma patients: Side study of the EORTC 18961 trial. First Author: Judith Michels, Gustave Roussy Comprehensive Cancer Center, Villejuif, France Background: Tumor associated antigen (TAA)-specific humoral immune responses have been previously shown to correlate with dismal prognosis in early stage melanoma patients. GM2 vaccination does not improve outcome for patients with stage II melanoma. Methods: The serum IgG antibody responses were assessed at baseline against a panel of 44 antigens by a bead-based multiplex assay in 970 stage II melanoma patients of the randomized EORTC 18961 trial, in which an adjuvant ganglioside GM2-KLH/QS21 vaccination (vacc) versus observation (obs) was compared. Cox multivariate models were used to assess the prognostic/predictive importance of antibody responses adjusted for gender, Breslow thickness (Breslow), ulceration status, confirmation of lymph nodenegative involvement by sentinel or elective node dissection (LN-negative) and treatment arm. Results: Significant prognostic clinical variables for relapse free survival (RFS) and overall survival (OS) were ulceration status, Breslow, LN-negative and gender. We found spontaneous antibody responses with a significant dismal or good prognostic impact on RFS and OS (table). Moreover predictive spontaneous antibody responses were depicted based on significant interaction with treatment for RFS (RhodN p=0.03; Rab38 p=0.04; RhodE2 p=0.08) and OS (RhodE2 p=0.004; Recoverin p=0.04; NA17.A p=0.05) respectively. Subgroups for RFS RhodN negative (n=849) HR(vacc vs. obs) 1.07, p=0.62; RhodN positive (n=121) HR(vacc vs. obs) 0.42, p=0.01; Rab38 negative (n=682) HR (vacc vs obs) 1.12, p=0.42; Rab38 positive (n=288) HR(vacc vs obs) 0.65, p=0.04 and OS RhodE2 negative (n=633) HR(vacc vs obs) 0.72, p=0.13; RhodE2 positive (n=337) HR(vacc vs obs) 1.85, p=0.01 were determined. Conclusions: This panel of TAA antibody responses defined a prognostic signature for patients with stage II melanoma. A set of antibody responses correlated with a beneficial or detrimental outcome for GM2 vaccination.

Multivariate cox regression for prognostic antibody responses. RFS

GM2 Rhod_E2 SSX2 CyclinB1 SCYE1v1

3034

OS

HR

p

HR

p

1.40 1.26 1.42 0.83 0.87

0.04 0.06 0.01 0.17 0.30

1.21 1.43 1.13 0.67 0.63

0.40 0.02 0.50 0.03 0.02

Poster Session (Board #356), Sun, 8:00 AM-11:30 AM

3033

Poster Session (Board #355), Sun, 8:00 AM-11:30 AM

Immune gene expression, survival outcome and response to PD-1/PD-L1 blockade: A TCGA pan-cancer analysis. First Author: Elia Seguı´, Hospital Clı´nic de Barcelona. Universitat de Barcelona, Barcelona, Spain Background: Immune infiltration predicts survival in particular cancertypes. However, differences in immune profiling and associations with patient outcomes among multiple cancer-types have not been fully explored. Methods: Clinical and RNA-seq data from 9,761 tumors representing 30 cancer-types were obtained from TCGA project. The expression of 35 immune-related gene signatures tracking various cell-types was evaluated, including PD1, CTLA4, CD8A and CD4 genes. Association of each signature with overall survival (OS) was evaluated using Cox models. Correlations between each signature and the overall response rates (ORR) following antiPD1/PDL1 monotherapy in each cancer-type (as reported in the literature) were evaluated across 16 cancer-types. Results: In the combined dataset, 15 out of 35 signatures (43%) mostly tracking cytotoxic T-cell response (e.g. CD8-T-cell, PD1 and CD8A) were highly correlated (mean coefficient [cc] = 0.69). The top 5 cancer-types with the highest percentages of samples expressing cytotoxic T-cell response signatures were lung adenocarcinoma (LUAD), thymoma, testicular cancer, melanoma and basal-like breast cancer. The top 5 cancer-types with the lowest percentages were chromophobe renal cell, pheochromocytoma/paraganglioma, prostate cancer, adrenocortical cancer and uveal melanoma. In terms of OS, expression of T-cell response signatures (as a continuous variable) was consistently associated with better outcomes in melanoma, endometrial adenocarcinoma, head and neck carcinoma, hepatocarcinoma, bladder cancer and LUAD, and with worse outcomes in low grade glioma, uveal melanoma, glioblastoma and clear cell renal carcinoma. Finally, high expression of CD8A gene, CD8-Tcell signature and PD1 gene was found highly correlated (cc = 0.68, 0.62 and 0.77, P, 0.001) with ORR following anti-PD1/PDL1 therapy. Conclusions: Significant differences in the expression of immune signatures exist within and across cancer-types. Whereas the association of cytotoxic Tcell response with OS might be dependent on cancer-type, response to antiPD1/PDL1 checkpoint inhibitors seems associated with the absolute levels of pre-existing cytotoxic T-cell response.

3035

Poster Session (Board #357), Sun, 8:00 AM-11:30 AM

Development and analytical performance of a molecular diagnostic for antiPD1 response on the nCounter Dx Analysis System. First Author: Brett Wallden, NanoString Technologies, Inc., Seattle, WA

Association of a specific innate immune response to DNA damage with DNA repair deficient colorectal cancers. First Author: Petros Tsantoulis, Hopitaux Universitaires de Geneve, Geneva, Switzerland

Background: Pembrolizumab is a humanized anti-PD1 antibody that is FDA approved for use in patients with advanced melanoma and in selected patients with metastatic non-small-cell lung cancer. It has also shown clinical activity in a number of other tumor types in clinical trials, but there is need for a precise and accurate test that can identify patients most likely to benefit from therapy. Several immune-related gene expression (Gx) signatures in formalin fixed, paraffin embedded (FFPE) tissue were previously reported to enrich for responders to pembrolizumab across different tumor types. We have developed a clinical trial assay, referred to here as the antiPD1 Gx test, based on genes repeatedly found to be associated with improved response to pembrolizumab in a number of cancers. Here we describe the development and analytical performance of the anti-PD1 Gx test in multiple tumor types. Methods: The anti-PD1 Predictor Score (PS) algorithm was trained using RNA from FFPE specimens from all cancer types in the KEYNOTE-012 trial and several cancer types in the KEYNOTE-028 trial (anal canal, biliary tract, colorectal, esophageal, and ovarian) to determine the final genes and weightings. Analytical precision from RNA, reproducibility from multiple tissue blocks, impact of intra-tumor heterogeneity, and sensitivity to RNA input amount were measured across operators using samples from multiple tumor types. The robustness of the assay was evaluated with the inclusion of adjacent non-tumor tissue. Results: The total standard deviation in anti-PD1 PS was , 5% of the score range with random error being the major source of variance. The assay was robust across the specified RNA input range and against the inclusion of non-tumor tissue. The major source of variability in Gx across multiple tumor types was associated with the tumors’ immune Gx signature rather than intra-tumor variability or even tumor type. Conclusions: The NanoString anti-PD1 Gx test is a robust assay, which profiles immune related Gx across multiple cancer types. The assay is well suited to clinical applications and its ability to identify responders to anti-PD1 therapy is being investigated in multiple indications in several studies.

Background: We previously described a molecular subtype of tumors that presents an innate immune response to intrinsic DNA damage. The associated gene expression pattern is mediated by the STING viral response pathway within the cancer cell and is defined by up-regulation of cytokines that drive lymphocytic infiltration and up-regulation of immune checkpoint such as PD-L1 and IDO1. We have developed a 44-gene signature, the DNA damage response deficient (DDRD) assay that can prospectively identify this molecular subtype (Mulligan et al 2014). We now apply this assay to colorectal cancer (CRC) to look for associations with DNA repair deficiency. Methods: We applied the DDRD assay to 72 stage I, 448 stage II, 836 stage III and 90 stage IV CRC patients with gene expression and exome sequencing data from the TCGA, Marisa and PETACC3 datasets (Kucherlapati et al., 2012, Marisa et al., 2012, Bosman et al., 2009). The association between DDRD positivity and MSI (microsatellite instability) status was tested using Chi-squared testing. A further analysis was performed within the TCGA cohort to determine if DDRD positivity is associated to general mutational burden in CRC, including evaluation of mutations in known DNA repair genes. Results: Approximately 35% of CRC were DDRD positive. Consistent with a loss of DNA repair capacity, the mean mutational burden was significantly higher in high DDRD score samples (Spearman’s correlation P = 0.0008). A strong association was found between MSI and DDRD positivity in all datasets (T-test P , 0.001) where around 80% of tumors known to be MSI positive were DDRD positive. Significantly 25% of MSS (microsatellite stable) tumors were DDRD-positive. Gene mutations associated with the MSS DDRD positive group included ATM, ATR, BRCA1/2 and other components of the Fanconi Anemia(FA)/BRCA pathway (P , 0.001 ). Conclusions: The DDRD assay is detected in patients with DNA repair deficiencies, particularly MSI or mutations in components of the FA/BRCA pathway. As this molecular subgroup is defined by an innate immune response and activation of immune checkpoint genes such as PD-L1, we hypothesize that DDRD positive CRC patients may benefit from therapies that specifically target immune biology.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Immunotherapy 3036

Poster Session (Board #358), Sun, 8:00 AM-11:30 AM

3037

151s

Poster Session (Board #359), Sun, 8:00 AM-11:30 AM

Relationship of baseline tumoral IFNg mRNA and PD-L1 protein expression to overall survival in durvalumab-treated NSCLC patients. First Author: Brandon W Higgs, MedImmune, Gaithersburg, MD

Association of plasma exosomal mRNA changes with long-term durable response to ipilimumab (IPI) in metastatic melanoma (MM) patients. First Author: Christine M. Coticchia, Exosome Diagnostics, Campridge, MA

Background: Biomarkers may help to identify non-small cell lung cancer (NSCLC) patients (pts) more likely to have improved outcomes to anti-PD-L1 therapy. From 112 durvalumab (D)-treated pts, we previously reported that high pretreatment levels of PD-L1 protein and IFNg mRNA expression in tumor biopsies were associated with higher objective response rates. We extend this analysis with mature survival outcome data that includes additional available pts; in addition, we examined changes in gene expression from on-treatment tumors in a subset of pts. Methods: CP1108/ NCT01693562 was a nonrandomized phase 1/2 trial evaluating D in pts with advanced NSCLC or other solid tumors. As of 14 DEC 2015, 298 previously treated NSCLC pts received 10 mg/kg Q2W of D #12 mos with a median 16.1 mos follow up. Expression profiling of 100 immune-associated genes using Fluidigm Biomark on frozen biopsies and immunostaining for tumoral PD-L1 on fresh or archival biopsies was conducted. 157 pts (55% squamous; 54% $2 prior therapy lines) had matched baseline mRNA and PD-L1 data of sufficient quality. Samples with detectable levels of IFNg mRNA were considered IFNg +. Pts with $25% tumor cells stained for PD-L1 at any intensity were scored PD-L1+. Of the 157 pts, 32 pts had pre- and post-treatment tumors evaluable for 100-gene mRNA analysis. Results: Pts who were IFNg+/PD-L1+ had a trend of longer overall survival compared to those who were IFNg-/PD-L1-, even after adjusting for gender, age, prior therapy lines, histology, ECOG, and smoking (Table 1). Following treatment with D, the transcript most highly induced was IFNg (~2-fold; p,0.016). Conclusions: NSCLC pts with elevated IFNg mRNA and PD-L1 protein expression in baseline tumor biopsies may have longer survival with D treatment. D treatment induces IFNg gene expression within the tumor microenvironment. Clinical trial information: NCT01693562.

Background: Identifying MM patients who will achieve long-term benefit from immunotherapy is an unmet need. We investigated using exosomal RNA (exoRNA) as a liquid biopsy approach to monitor response and aid in selecting therapy. Exosomes are released from cells, including tumor cells and can be extracted from biofliuds enabling non-invasive monitoring of a patient’s tumor status in real-time. However, the response signature must be identified against a background of exosomes from normal processes, such as exosomes released during maturation of reticulocytes. We have developed the ExosomeDx Depletion/Enrichment (EDDE) platform to remove nonrelevant exosomes from plasma. Using EDDE, we identified a plasma exoRNA signature associated with durable response to IPI. Methods: Plasma was obtained from 16 MM patients before IPI (baseline) and during the first 2-4 weeks of treatment (week 2/4). Response to therapy was determined by RECIST. Patients with . 12 months PFS from start of IPI were classified as durable responders (DR) and compared to all other patients (non-DR). Matched patient plasma samples at baseline and week 2/4 were processed both with a total exoRNA extraction method as well as with the EDDE platform to evaluate the strength of the signatures after enrichment. The exoRNA was analyzed by TaqMan qPCR on the OpenArray platform, examining the expression level of 607 mRNAs in each sample. Results: To date, we have evaluated a signature of IPI treated patients (n = 16). Of the 607 genes examined, the EDDE platform revealed 8 genes significantly increased at week 2/4 in most ( $ 3/5) of DR patients and decreased in the majority of non-DR patients. The signature includes immune-related and inflammatory signaling ligands and receptors such as BDKRB1, BMP8B, CNTFR and TNFSF11. In contrast, using the same gene selection criteria, total exoRNA revealed no significantly dysregulated genes. Conclusions: The EDDE platform depletes normal exosomes from plasma increasing the power to discriminate early exoRNA changes associated with DR to IPI. Our ongoing experiments are identifying exoRNA signatures associated with long-term PFS of MM patients treated with anti-PD-1 therapy.

OS by IFNg and PD-L1 status. IFNg +/PD-L1+ IFNg +/PD-L1‒ IFNg ‒/PD-L1+ IFNg ‒/PD-L1‒

# Pts (# events)

Median in months (95% CI)

Adjusted HRa; p

43 (11) 20 (8) 42 (16) 52 (34)

NR (8.8, NA) NR (6.5, NA) 9.7 (5.1, NA) 5.9 (4.1, 10.2)

0.40; 0.016 0.38; 0.021 0.64; 0.18

NR=Not Reached; NA=Not Applicable; a Univariate log-rank p,0.05 for all adjusted p ,0.05 (not shown)

3038

Poster Session (Board #360), Sun, 8:00 AM-11:30 AM

Association of response to programmed death 1 receptor or ligand (PD1/ PDL1) blockade with immune-related gene expression profiling across three cancer-types. First Author: Alejandro Navarro, Medical Oncology Department, Vall d’Hebron University Hospital, Barcelona, Spain Background: Anti-PD1/PDL1 drugs show consistent activity across various solid tumors. However, a substantial proportion of patients show primary resistance, and those who benefit do not benefit to the same extend. Here, we explored the predictive ability of immune-related gene expression in advanced lung cancer, head and neck and melanoma. Methods: Treatmentnaive (n = 45) or pre-treated (n = 31) tumor samples from 76 patients with lung non-squamous (n = 28), squamous cell lung cancer (n = 20), head and neck (n = 5) and melanoma (n = 23) were evaluated. RNA from formalinfixed paraffin-embedded tissue sections was analyzed using the NanoString PanCancer Immune Panel (730 genes). Patients were treated with pembrolizumab (n = 35), nivolumab (n = 28) or atezolizumab (n = 13). All genes, and 16 immune cell-type signatures, were linked to response and progression-free survival (PFS). Logistic regression models for overall response rate (ORR; defined as partial or complete response at 6-8 weeks) and non-progressive rate (NPR; non-progressive vs. progressive disease at 68 weeks), and Cox models for PFS, were used. Results: The ORR and NPR were 32.9% and 61.8%, respectively. Median PFS was 5.57 months (3.207.67). No gene was found significantly associated with ORR (False Discovery Rate [FDR] , 5%). Expression of 374 genes, including PD1 and PDL1, was found significantly associated with NPR (FDR = 0%). Six signatures (T-cells, CD8-T cells, NK-cells, CD4-activated, T-regulated and Th1-cells) showed significant associations with both NPR and PFS. The median PFS of the Th1cells high-tertile group was 12.3 months compared to 2.6 months of the Th1-cells low-tertile group (hazard ratio = 0.26, 0.10-0.65,P= 0.004). These associations were independent of sample timing, drug used or cancertype. Conclusions: Expression of immune-related genes in tumor samples identifies patients likely to have primary resistance to anti-PD1/PDL1 monotherapy. Results are consistent with the hypothesis that identification of a preexisting and stable adaptive immune response predicts clinical outcome regardless of cancer-type or anti-PD1/PDL1 inhibitor.

3039

Poster Session (Board #361), Sun, 8:00 AM-11:30 AM

LAG 3/MHC 2 and PD 1/PD L1 expression in non-small cell lung cancer patients. First Author: Yayi He, Division of Medical Oncology, Department of Medicine, University of Colorado Anschutz Medical Campus, Aurora, CO Background: Immunotherapy targeting the PD1/PDL1 checkpoint has shown promising efficacy in patients with non-small cell lung cancer (NSCLC). LAG-3/MHC II is another important checkpoint that may have a synergistic interaction with PD-L1/PD-1. In this study, we describe the protein expression of LAG 3, MHC 2, PD 1 and PD L1 and their associations with survival in patients with NSCLC. Methods: The expression of LAG3 (EPR4392, Abcam), MHC II (DP, DQ, DR) (CR3/43, Abcam), PD 1 (NAT 105, Cell marque) and PD L1 (22C3, Dako) was evaluated in 139 surgically resected specimens from patients with NSCLC by immunohistochemistry (IHC). Results: LAG 3 was expressed on the immune cells in 49 (35.3%) patients. MHC II was expressed in 37 (26.6%) patient tumor cells. 60 samples (43.2%) stained positive for PD 1 on the immune cells, and 25 (18.0%) had positive PD L1 expression on tumor cells. There was no LAG 3 or PD 1 expression on tumor cells. LAG 3 on immune cells was overexpressed in squamous carcinoma compared to adenocarcinoma (P = 0.044). MHC II on tumor cells was expressed less in squamous carcinoma (P = 0.043) and smokers (P = 0.004) compared to adenocarcinoma and non-smokers. LAG 3 expression on immune cells was correlated with the expression of PD 1 (P＜ 0.001) on immune cells and PD L1 (P = 0.006) on tumor cells. Expression of PD L1 on tumor cells and PD 1 on immune cells were positively correlated (P = 0.026). Expression of PD L1 on tumor cells and on immune cells was negatively correlated (P = 0.819). PD L1 (tumor cells) negative patients had a longer disease free survival (DFS) (1.85 years [95% CI 0.41–3.28] vs. 0.97 years [0.709–1.231]; P = 0.038). DFS was longest in both PD L1 and LAG 3 negative patients（PD L1 on tumor cells and LAG 3 on immune cells）, followed by patients for only PD L1 or LAG 3 positive, and DFS was shortest in patients positive for both PD L1 and LAG 3 (2.09 years [95% CI 0.25-3.93] vs. 1.76 years [0.67-2.85] vs. 0.67 years [0.17-1.18]; P = 0.018). Cox regression analysis showed that positive LAG 3 expression on immune cells was a poor prognostic factor for NSCLC patients (P = 0.040, EXP (B) 1.964, 95%CI 1.030-3.747). Conclusions: LAG 3 expression was correlated with PD 1/PD L1 expression and was associated with poor prognosis in patients with NSCLC.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

152s 3040

Developmental Therapeutics—Immunotherapy Poster Session (Board #362), Sun, 8:00 AM-11:30 AM

3041

Poster Session (Board #363), Sun, 8:00 AM-11:30 AM

Cytokine release syndrome (CRS) in patients treated with NY-ESO-1c259 TCR. First Author: Crystal Mackall, National Cancer Institute at the National Institutes of Health, Bethesda, MD

Identification of somatic mutations to predict development of autoimmune adverse events to immune therapy in melanoma. First Author: Keith Ryan Wells, University of Colorado Denver, Denver, CO

Background: Cytokine release syndrome (CRS) is observed with adoptive T-cell therapies (ACT) with varying frequency. CRS can be life threatening and may require steroids, which hinder the effectiveness of ACT. CRS prevalence and severity within the context NY-ESO-1C259T across 5 studies in 4 tumor types were examined: synovial sarcoma (SS), ovarian cancer (OC), myeloma and melanoma. Methods: A review of CRS including evaluation of concurrent AEs and reported symptoms, cytokine levels and CRP was performed in pts treated with NY-ESO-1C259T. AEs that are known manifestations of CRS were summarized for each pt. diagnosed with CRS. CRP was measured by the hospital laboratory. Serum cytokines were measured by Luminex. Results: Of 53 pts treated with NY-ESO-1C259T, 7 were diagnosed with CRS (Table). 2/7 pts received immunomodulatory treatment for mitigation of CRS: methylprednisolone 2 mg/kg x 2 doses with prednisone taper (n=1, G3) and Tocilizumab 4 mg/kg x 1 dose (n=1 G4). Diagnosis of CRS was made within the first 2 wks in all 7 pts and resolved within 32 days after the Tcell infusions. Onset of CRS coincided with T-cell expansion. Elevated IL-6 and IL-8 were observed during CRS. Timing of the increase of IL-6 and IL-8 suggests involvement in the effector phase of CRS instead of its induction. IL-5, IL-15, INF-g and GM-CSF showed no discernable pattern associated with CRS. Conclusions: CRS was observed in 13% of pts treated with NY-ESO-1C259T. Symptoms manifest in the first 2 wks and have been effectively managed with supportive care measures. Two pts required immunomodulatory therapy. While there are differences in the pt. populations, incidence of CRS with NY-ESO-1C259T appears to be of lower frequency and severity than reported with CD19 CAR-T therapy.

Background: Immunotherapy has quickly altered the treatment paradigm for advanced melanoma. Prior studies in melanoma using whole exome sequencing (WES) in patient tumor samples who have received immunotherapy have focused on neoantigen formation in correlation with clinical response to immunotherapy, but no studies to identify genes have been done. Understanding mechanisms of immune-related adverse events (irAE) at a genomic level may lead to improved rational treatment design and better patient selection. Methods: DNA was extracted from 87 tumor samples from 49 patients with metastatic melanoma treated with ipilimumab (anti-CTLA4) or with nivolumab or pembrolizumab (anti-PD-1) immunotherapy. WES was performed with exome capture on an Illumina HiSeq sequencer. Statistical analysis was performed with Chi-squared testing. Results: 27 patients with advanced or metastatic melanoma were treated with only ipilimumab, 11 with only anti-PD1, and 11 patients were treated with both anti-CLTA-4 and anti-PD1 either sequentially or in combination. irAE evaluated included rash, colitis, endocrinopathies, hepatitis, and pneumonitis. In 10 patients who developed colitis, 14 mutated genes were enriched in their tumor samples (P = 0.01-0.046) compared with patients without colitis. Analyzing all patients with any irAE versus no irAE treated with anti-CTLA-4, anti-PD1, or both (either sequential or combination) revealed 7 mutated genes enriched in patient samples who developed irAE (P = 0.02-0.04). Overall mutational burden of the tumor samples did not correlate with development of irAE. Conclusions: Autoimmune side effects are frequent and often serious side effects of immunotherapy for melanoma. Genetic analysis may provide insight in prediction of these side effects, understanding of tumor-immune response, and help direct therapeutic decisions.

Data summary. Trial/Patient Grade CRS/duration (days) Responder Fever/Rash/Diarrhea CNS symptoms Dyspnea/hypoxia Hypotension Fatigue CRP (mg/L) IL-6 peak/IL-8 (pg/mL)

SS 205

SS 207

SS 208

SS 261

SS 264

OC 101

OC 203

2/3 Y Y/Y/Y NR NR Y N 261 385/129

1/8 N Y/NR/Y Y NR NR N 157 114/671

3/24 Y Y/Y/Y Y NR NR Y 198 72/226

3/11 N Y/NR/Y NR Y NR Y 175 ND

4/11 Y Y/Y/NR NR Y Y NR 182 987/394

3/10 N Y/Y/Y Y Y Y Y ND 80/150

2/7 3 Y/Y/NR Y NR Y NR 143 ND

3 too early to assess response; NR – not reported; ND – not done

3042

Poster Session (Board #364), Sun, 8:00 AM-11:30 AM

3044

Poster Session (Board #366), Sun, 8:00 AM-11:30 AM

Mutational burden and activated T cell infiltration in lung squamous cell carcinomas with DNA repair mutations. First Author: Young Kwang Chae, Northwestern Medicine Developmental Therapeutics Institute, Chicago, IL

Inhibition of the novel immune checkpoint CEACAM1 to enhance anti-tumor immunological activity. First Author: Gal Markel, Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel

Background: Mutations in DNA repair genes are common in cancer, and may induce a state of hypermutability. Importantly, mutational burden is linked to enhanced response to immunotherapies, as potentially immunogenic neoantigens may drive a stronger anti-tumor immune response. Methods: We utilized mutation (n = 178) and mRNA expression (n = 501) data from the TCGA database for lung squamous cell carcinoma to analyze the link between DNA repair mutations, mutational burden, and predicted immune cell type tumor infiltration using a defined list of “immune metagenes” (Angelova et al,2015). Results: Tumors were infiltrated most by activated CD4 (36.1%) and CD8 (28.1%) T cells (top 9 of 28 total immune cell types listed in Table 1). This was further pronounced in tumors with higher mutational burden displaying an increase in activated CD4 and CD8 T cells (p , 0.05). Additionally, tumors with 0, 1, or 2-4 mutated homologous recombination (HR) DNA repair genes contained 221, 355, and 521 mutations, respectively. Those with 0, 1, or 2-3 mismatch repair (MMR) genes displayed mutation counts of 235, 341, and 786, respectively. Interestingly, tumors with HR mutations had increased infiltration by neutrophils, while those with MMR mutations were infiltrated by activated CD4 and activated CD8 T cells. Conclusions: Overall, mutational burden was linked to increased activated CD4 and CD8 T cells, and DNA repair mutations were linked to increased mutational burden. However, only MMR mutations were associated with activated CD4 and CD8 T cell infiltration, and may better serve as biomarkers to predict patient response to immunotherapy.

Background: Blockade of co-inhibitory immune receptors has become a promising approach for cancer immunotherapy. Carcinoembryonic antigenrelated cell adhesion molecule 1 (CEACAM1) is a recently characterized immune checkpoint. Expression of CEACAM1 on tumors induces a coinhibitory signal to CEACAM1-positive T and NK cells through homophilic interactions. Methods: CM-24 (MK-6018) is a humanized anti-CEACAM1 IgG4 antibody with high affinity and selectivity for CEACAM1. The effect of CM-24 alone or in combination was examined in vitro using tumorinfiltrating lymphocytes (TIL) and peripheral blood mononuclear cells. The anti-tumor efficacy of CM-24 was studied in vivo in subcutaneous xenograft and lung metastasis models of melanoma treated with IV injection of mlanoma-reactive TIL. Gene expression analysis of CEACAM1 and other genes of potential relevance to patient response to CEACAM1 were examined in a large cohort of human tumor samples. Results: CM-24 was demonstrated to be a potent blocker of intercellular CEACAM1-CEACAM1 interactions and reversed the inhibition of activated lymphocytes by restoring the phosphorylation of ZAP70. In addition, CM-24 enhanced the anti-tumor activity of TIL in the melanoma xenograft and lung metastasis models. CM24 also showed synergistic activity with immune checkpoint blockers targeting PD-1 or PD-L1. The gene expression analysis of CEACAM1 and other genes in 18,000+ solid tumor samples allowed us to identify tumor types which represent potential indications for CM-24 therapy and to identify gene expression profiles as candidate biomarkers of response to CM-24 as monotherapy or in combination with a PD-1 or PD-L1 antagonist. Conclusions: CM-24 increases the cytotoxic activity of lymphocytes against malignant cells and has shown effective anti-tumor activity in various in vitro and in vivo models. Treatment with CM-24 either as monotherapy or in combination with anti-PD-1 treatment such as pembrolizumab may be beneficial. First-in-man clinical trial with CM-24 is ongoing.

Immune cell type

Percent of patients with infiltrations (%)

Activated CD4 Activated CD8 Effector memory CD4 gd T cells Mature dendritic cells (DCs) Th17 T cells DCs Monocytes NK cells

36.1 28.1 10.6 9.6 7.0 5.8 5.0 4.8 4.8

*Anker and Chae contributed equally to this work

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Immunotherapy 3045

Poster Session (Board #367), Sun, 8:00 AM-11:30 AM

Phase Ib/II trial of expanded and activated autologous natural killer (NK) cells with trastuzumab (Tras) in refractory HER2+ metastatic breast cancer (MBC). First Author: Soo-Chin Lee, National University Cancer Institute, National University Health System, Singapore, Singapore, Singapore Background: A mechanism of Tras is antibody-dependent cell cytotoxicity (ADCC) exerted by NK cells. ADCC may be impaired in cancer patients. We hypothesize that this can be restored by NK activation, and developed a method to generate large numbers of activated NK cells from MBC patients (pts) and tested the effects of NK infusion with Tras. Methods: Peripheral blood harvested from 11 refractory HER2+ MBC pts aged 32-67 (median 59), was co-cultured for 10 days with irradiated K562-mb15-41BBL cells. Expanded NK cells were infused 24 hours (H) post-Tras, with interleukin-2, 1MU/m2 thrice weekly for 6 doses starting 24H pre-infusion. Tras was continued 3 weekly up to 8 cycles (C); pts with sustained stable disease (SD) at C4 or C6 received a 2nd and 3rdinfusion at C6 and C8 respectively. Blood was taken serially for NK cell counts and function. Results: 16 NK cell products were prepared for the 11 pts; 3 pts had a 2nd, and 2 a 3rd infusion. Median NK cell expansion was 337-fold (range 91-478). Products contained 81616% NK cells, virtually all CD16+ and highly cytotoxic: ADCC on HER2 + SKBR3 cells at 4H with Tras was 85.8610.1% at 2:1 effector:target ratio (E:T). Pts had 2-13 (median 6) prior lines of palliative therapy. 3 of 16 infusions were at 1st dose level (106 NK cells/kg), and 13 at target dose (107/ kg). NK infusion was well tolerated; worst toxicity was transient G2 thrombocytopenia (n = 1). There was no objective response but 7/10 evaluable pts, including 6 who had 107/kg at C1, had SD. Median progression free survival was 1.8 vs 8.5 months for pts who received 106/kg vs 107/kg cells at C1. NK cell counts increased from day (D) 1 (0.166 0.11x103/mL) to D8 (0.3260.19x103/mL) or D22 post-infusion (0.366 0.20x103/mL) (p , 0.01), although levels were only slightly higher than baseline (0.3160.22x103/mL). NK cells were 1368% of PBMCs on D1 and 1769% on D8; ADCC on SKBR3 cells with Tras without NK enrichment was 1164% on D1 and 1665% on D8 (2:1 E:T). Conclusions: This first-inhuman study shows that large numbers of highly potent NK cells can be reliably generated from MBC pts and safely infused with Tras. The disease control seen in heavily pre-treated pts is encouraging and warrants further studies. Clinical trial information: NCT02030561.

3047

Poster Session (Board #369), Sun, 8:00 AM-11:30 AM

Immune-related adverse events and survival in elderly patients with melanoma treated with ipilimumab. First Author: Idrees Mian, The University of Texas MD Anderson Cancer Center, Houston, TX Background: Immune-related adverse events (irAE) encompass wideranging toxicities in patients (pts) taking Ipilimumab. Experienced by many pts on Ipilimumab, they occasionally result in death. As irAE could be indicative of an enhanced immune response to therapy, pts developing these events potentially experience improved therapeutic response and survival. We evaluated the toxicity patterns of irAE in older pts with melanoma receiving Ipilimumab, and the association of irAE with their overall survival (OS). Methods: We identified melanoma pts 65 years or older, treated with Ipilimumab using the Surveillance Epidemiology and End Results (SEER) and Texas Cancer Registry (TCR) linked with Medicare data between 3/2011 and 12/2013. We defined 44 types of irAE based on 228 ICD-9 diagnosis codes; irAE had to occur within 3 months of last treatment and include one or more diagnosis codes, which could not be present within a 6-month window before treatment. We classified pts with irAE as severe if they had a related hospitalization or death within 2 weeks of the claim. We used log rank and multivariable Cox regression to analyze OS. Results: We identified 858 pts with mean age at diagnosis 69.1 years and mean age at treatment 74.8 years. More than half of pts (63.3%) had a Charlson Comorbidity Index of 0. Pts received a median of 3 treatment courses. 510 (60%) pts experienced irAE with 178 pts (20.7%) experiencing severe irAE. The most frequent irAE were colitis (17.5%), hypothyroidism (10.5%), dermatitis (5.4%) and hypophysitis (3.7%). Those with non-severe irAE had improved OS compared with no irAE or severe irAE with median survival years being 1.1, 0.9, and 0.6, respectively (p , 0.001). In the multivariable analysis, severe irAE pts were 28% more likely to die compared to pts with no irAE (p = 0.04). Conclusions: In this retrospective study our findings suggest that the incidence of irAE occurs in elderly melanoma pts at comparable rates to all melanoma pts treated with Ipilimumab. Hypothyroidism occurred more frequently than what has been reported in clinical trials. Pts with non-severe irAE had improved OS compared to pts with no irAE. Pts with severe irAE appeared to have the highest risk of death.

3046

153s

Poster Session (Board #368), Sun, 8:00 AM-11:30 AM

Chimeric antigen receptor engineered allogeneic CD27-negative T cells for the treatment of CD19+ leukemia. First Author: Aimee C. Talleur, St Jude Children’s Research Hospital, Memphis, TN Background: Relapsed/refractory pediatric acute lymphoblastic leukemia (ALL) has a poor prognosis. Adoptive immunotherapy with chimeric antigen receptor (CAR) engineered T cells targeted against a tumor associated antigen, such as CD19 on B-lineage leukemia, has shown immense promise in this patient population. An important aspect of CAR therapy is the source of effector cells, autologous or allogeneic. A barrier to using allogeneic T cells is that the na¨ıve subset may lead to graft-versus-host disease (GVHD). A potential strategy to circumvent this would be to use surface markers such as CD27 to distinguish between na¨ıve and memory T cells. CD27 is present on na¨ıve and central memory T cells, but not on effector memory and terminal effector memory T cells. Our objectives were to determine if allogeneic CD27- cells provide superior immunologic memory against pathogens commonly encountered in patients with refractory ALL and have decreased alloreactivity potential compared to CD27+ cells. We also sought to assess the cytotoxic potential of CD27- cells transduced with an anti-CD19 CAR vector against CD19+ leukemia cells. Methods: CD27- and CD27+ cell fractions were isolated from donor PBMC using magnetic microbead separation. Lymphocyte proliferation assays were performed to determine immunologic memory to antigen and mitogen stimuli. Alloreactivity potential was tested using one-way mixed lymphocyte reactions. CD27- cells were transduced with an anti-CD19 CAR and their cytotoxicity against a CD19+ leukemia cell line was determined in vitro via BATDA assay and in vivo in NSG mice. Results: CD27- cells had stronger immunologic memory against pathogens and were less alloreactive than CD27+ cells. CD27- cells expressing an anti-CD19 CAR demonstrated potent killing of a CD19+ leukemia cell line compared to untransduced CD27- cells, both in vitro and in vivo. Conclusions: CD27- allogeneic cells provide pathogen protection, have low GVHD potential and, when transduced with an anti-CD19 CAR, have cytotoxic capability against CD19+ leukemia cells. These findings make CD27- T cells a promising population to be used as effector T cells in CAR therapy, providing dual infection and leukemia control.

3048

Poster Session (Board #370), Sun, 8:00 AM-11:30 AM

Prolonged functional persistence of CD19CAR t cell products of defined CD4:CD8 composition and transgene expression determines durability of MRD-negative ALL remission. First Author: Rebecca Alice Gardner, Seattle Children’s Hospital, Seattle, WA Background: CD19 CAR-T cell has proven efficacy for treatment of aggressive ALL. However, limited CAR-T cell persistence and/orCD19 loss are barriers to therapeutic success. Methods: A Phase I trial (NCT02028455) of CAR-T cell products consisting of purified CD4 and CD8 T cell subsets expressing uniform levels of a 4-1BB second generation CD19-specific CAR were infused following either cyclophosphamide (Cy) or fludarabine (flu/cy). Leukemic response, B-cell aplasia (BCA) and T cell persistence were evaluated by flow cytometry. Results: Products were successfully manufactured in 39/40 consecutively enrolled patients (pts) (1-25y) despite having no screening assay of lymphocyte proliferation. The ALC at time of apheresis ranged from 140-4488/ml. Of evaluable pts, 91% (33/36) achieved MRD negative (neg) CR’s in a cohort of children and young adults. The DFS and OS at 12 months are 53% and 74%. Improved diseasefree survival (DFS) was observed in pts with ongoing BCA beyond day 63 (HR: 0.25 [95% CI: 0.08, 0.81], P = 0.01). For patients treated with Cy, durable engraftment of CAR T cells was positively influenced by CD19 antigen (Ag) burden (ALL + nonmalignant B cells) of . 15 % in the marrow at the time of CAR infusion (median of 4.2 v 1.4 months). This effect in 8 pts given Flu/Cy is currently under evaluation as the duration of BCA appears to also be influenced by lymphodepletion (median of 3.9 v 2.1 mo flu/cy v cy). Six patients had CD19-relapse. Conclusions: Our trial of CD19 CAR-T cell adoptive therapy has revealed a high rate of manufacturing success and MRD neg remissions. Despite high rates of MRD-neg remission, relapse is a significant etiology of treatment failure. The durability of remissions was influenced by the duration of CAR-T cell functional persistence of at least 63 days duration. Ag-driven proliferation affects the duration of B cell aplasia for pts given a Cy alone regimen. Lymphodepleting regimens also appear to influence the duration of B cell aplasia and potentially negate the effect of CD19 Ag burden. Accordingly, our program is continuing to investigate Flu/Cy conditioning regimens to improve CAR-T cell persistence. Clinical trial information: NCT02028455.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

154s 3049

Developmental Therapeutics—Immunotherapy Poster Session (Board #371), Sun, 8:00 AM-11:30 AM

3050

Poster Session (Board #372), Sun, 8:00 AM-11:30 AM

Towards a safer CD19 CART therapy with reduced risk of cytokine release syndrome (CRS). First Author: Lucas H. Horan, Eureka Therapeutics, Emeryville, CA

Efficacy and safety of Her2-targeted chimeric antigen receptor (CAR) T cells using MyD88/CD40 costimulation and iCaspase-9 suicide switch. First Author: Aaron E Foster, Bellicum Pharmaceuticals, Houston, TX

Background: Despite remarkable clinical success, adoptive T-cell therapy (ACT) with anti-CD19 chimeric antigen rector (CAR) T cells is hindered by both pro-inflammatory and immuno-suppressive adverse events. Hyperactivation of CD19 CARs can trigger life-threatening cytokine release syndrome (CRS), while xenogeneic sequences in existing constructs can lead to diminished CAR T cell efficacy due to immunogenecity. Methods: To ameliorate these problems we developed a novel chimeric T-cell therapy platform, ARTEMIS, that functionally matches the potency of CAR T cells, but dramatically reduces the release of cytokines upon killing of targetpositive tumors. To minimize the immunogenicity, we created a panel of fully-human CD19 antibodies using a human antibody library against cells expressing endogenous CD19. The CD19 antibodies were expressed as either ARTEMIS T cells or 2nd generation CD28z CAR T cells and tested for efficacy and safety both in vitro and in vivo. Results: In the Raji-lymphoma model with NSG mice, CD28z CART cells caused lethality within 24 hrs of Tcell injection into immunodeficient mice and led to pathologies resembling a pre-clinical CRS model. Strikingly, ARTEMIS T cells armed with the same anti-CD19 antibody showed no signs of toxicity and had potent anti-tumor effects. Both ARTEMIS and CAR T-cells engineered with our CD19 antibodies killed CD19+ cell lines with high specificity and potency. Compared to CAR, ARTEMIS T-cells have similar degranulation activity, intracellular cytokine production, and in vitro killing efficiencies. However, ARTEMIS Tcells release up to 1000-fold less cytokines and accumulate less exhaustion markers during in vitro killing assays. Conclusions: The therapeutic potential of CD19 CAR T cell therapy can be expanded with a better safety profile. The fully-human anti-CD19 antibody would reduce the risk of immunogenicity as compared with the mouse anti-CD19 antibody currently used in CAR T therapy. The ARTEMIS T-cell signaling platform was designed to be a safer ACT by relying on endogenous T-cell signaling factors. We show evidence that ARTEMIS begins to disentangle efficacy from CRS and holds the potential to be a clinically safer therapy by preventing ACT-triggered CRS.

Background: Efficacy of chimeric antigen receptor T cells (CAR-T) is dependent on their in vivo survival and expansion following adoptive transfer. Here, we show that constitutive expression of the MyD88/CD40 (MC) fusion protein dramatically enhances CAR-T proliferation resulting in improved antitumor efficacy against Her2+ solid tumors. Importantly, CAR-T cell levels and cytokine-related toxicity can be controlled by graduated administration of rimiducid to activate the inducible iCaspase-9 (iC9) suicide gene without detriment to tumor control. Methods: T cells were modified with a tricistronic retrovirus encoding iC9, a Her2-specific CAR (Her2.z) and constitutively active MC (iC9-Her2.z-MC), or control vectors (iC9-Her2.z and iC9-Her2. CD28.z) and compared for cytotoxicity, cytokine production and proliferation in coculture assays against Her2+ breast (SK-BR-3), ovarian (A2780, SKOV-3), gastric (N87, Kato III, SNU-1) and pancreatic (HPAC) cancer cell lines. CAR-T efficacy was assessed in NSG mice engrafted with tumor cell lines followed by i.t. or i.v. injection of T cells. Rimiducid was given i.p. at 0.5 mg/kg. Results: Her2-redirected T cells showed that in vitro, MC costimulation resulted in increased IL-2 production ( . 65-fold compared to iC9-Her2.CD28.z) which corresponded to enhanced proliferation and tumor elimination. In tumor xenograft models, iC9-Her2.z-MC CART cells were more effective at eliminating large Her2+ tumors compared to conventional CAR-T constructs (Her2.z and Her2.28.z). Bioluminescent imaging also showed robust engraftment and proliferation of CAR-T cells coexpressing MC. Importantly, cytokine-related toxicity could be completely resolved while maintaining anti-tumor efficacy by partial elimination of iC9Her2.z-MC CAR-T cells through systemic administration of rimiducid. Conclusions: MC represents a broadly applicable, potent T cell costimulatory signaling molecule in CAR-T cells, resulting in higher IL-2 production, improved survival and greater proliferative capacity. CARs containing MC demonstrate enhanced efficacy in vivo while retaining the safety profile through the iC9 suicide gene.

3051

3052

Poster Session (Board #373), Sun, 8:00 AM-11:30 AM

Poster Session (Board #374), Sun, 8:00 AM-11:30 AM

Targeting alpha fetoprotein with TCR engineered T cells in HCC. First Author: Andrew Gerry, Adaptimmune, Milton Park, Abingdon, United Kingdom

Role of genomic instability in immunotherapy with checkpoint inhibitors. First Author: George Yaghmour, West Clinic, Memphis, TN

Background: Alpha fetoprotein (AFP), an oncofetal protein, is transcriptionally repressed after birth. Reappearance of AFP in adult circulation indicates liver regeneration, hepatitis, chronic liver diseases, or malignant growth. When AFP positive, expression in HCC tends to be very high and homogeneous, making it an attractive immunotherapy target. Methods: A TCR specific for the HLA-A2-resricted peptide AFP158-165 (FMNKFIYEI) was identified and engineered to generate a panel of 12 affinity-enhanced TCRs (KD: 79.5-0.31mM). One clone (10mM KD), demonstrating enhanced potency against AFP-expressing liver tumor cell lines and no response to normal hepatocytes, was selected. Molecular mapping of each position of the target peptide was performed; the generated binding motif was searched against the human genome, identifying 165 potentially reactive peptides, which were synthesized, loaded onto T2 cells, and tested for reactivity by AFP TCR engineered T cells (AFP-T). Results: No safety concerns were identified. 126 normal cells and 42 tumor cell lines from various organ systems were screened for AFP-T reactivity; AFP-T showed cytokine secretion and cytotoxic activity against HCC cell lines in 2D culture and 3D microtissues, but no relevant response to any other HLA-A*0201-expressing cells. Alloreactivity was detected against a subset of HLA-A*0202 primary cells. A full alloreactivity screen performed using a panel of 51 EBVtransformed B cell lines covering 38 HLA-A, 63 HLA-B and 28 HLA-C alleles, demonstrated enhanced responses against HLA-C*0404 and HLAB*5103. AFP-T did not recognize the AFP peptide in the context of HLAA*0203. Patients with HLA-C*0404 and -B*5103 and -A*0202 will be excluded. Conclusions: A Phase I study will evaluate preconditioning- and cell- dose escalation to investigate the safety and anti-tumor activity of AFP-T in HLA-A*0201+ patients with AFP+ HCC and good residual liver function. Clinical safety measures for mitigating treatment-related hepatotoxicity include pre-treatment biopsy evidence of low AFP expression in noncancerous liver, and monitoring/management strategies within the protocol. Three cell-dose cohorts at 1x108, 1x109, 5x109 total transduced cells will be investigated.

Background: Checkpoint inhibitors have shown significant activity against a variety of cancers. However, biomarkers that predict responses to these agents have remained elusive. We hypothesized that tumors with greater genomic instability generate a more robust immune response that results in better survival. Methods: We retrospectively identified patients (pts) with solid tumors that were treated with checkpoint inhibitors and that had received commercially available next generation sequencing (NGS) between 2013-2015. Based on the skewed distribution of genomic alterations pts in the top quintile were compared to the other 4 quintiles combined. This procedure was done before examining mortality to avoid selecting cutoff points that could accentuate the mutation-outcome relationship. The functional significance of VUS’s in DNA damage repair pathways were predicted using PolyPhen. All p-values were two-sided and p-values less than 0.05 were considered statistically significant. Results: 50 pts were identified; most had either melanoma or NSCLC (top quintile 90% v others 86%, respectively). Median age (66 v 65) and median number of prior therapies (2 v 1) were similar. All pts had stage 4 disease. Top quintile pts had more genomic alterations (median 16.5) than the others (median 2) (p , 0.001); differences were both in median number of pathogenic mutations (3.5 vs. 1, p , 0.001) and VUS’s (12.5 vs. 1, p , 0.001). Pathogenic alterations in cell cycle regulatory genes (100% versus 48%, p = 0.017) were more common in the top quintile. Overall survival was superior for pts in the top quintile (722 days) versus the others (432 days) [HR 5.78; CI 1.40 – 15.12; p = 0.029]. Similar trends were seen regardless of histology. Progression free survival numerically favored the top quintile [not reached vs. 89 days, HR 2.11; CI 0.80-4.45; p = 0.154] as did objective response rate (50%) vs. others (20%), p = 0.101. PD1 and PDL1 status by immunohistochemistry were not associated with outcomes. Conclusions: The use of immune checkpoint blockade in tumors with higher mutational load was associated with improved overall survival. Our results suggest that the evaluation of tumor genomes may be predictive of immunotherapy benefit.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Immunotherapy 3053

Poster Session (Board #375), Sun, 8:00 AM-11:30 AM

PD-1 inhibitor-related pneumonitis in advanced cancer patients. First Author: Mizuki Nishino, DanaFarber Cancer Institute, Boston, MA Background: PD-1 inhibitor-related pneumonitis is an uncommon but serious immune-related adverse event (irAE). Details of radiographic patterns of this entity remain to be investigated. Methods: Patients (pts) who developed PD-1 inhibitor-related pneumonitis were identified among those treated in trials of nivolumab for advanced melanoma, lung cancer, and lymphoma. Chest CTs were reviewed by a consensus of 3 radiologists to assess CT findings and radiographic patterns of pneumonitis according to ATS/ERS classification of interstitial pneumonias, which were compared across tumor types and treatment regimens. Results: Among 170 pts treated in 10 different trials of nivolumab, either alone or in combination, 20 pts developed pneumonitis as an irAE (Grade: 1 in 5; 2 in 10; 3 in 5 pts). Of these, 13 (65%) were female, median age was 52 (range 28-71); 5 received nivolumab alone and 15 received combination therapy (with ipilimumab in 12, lirilumab in 3). Ten pts had melanoma, 6 had lymphoma, and 4 had lung cancer. Median time from therapy initiation to pneumonitis was 2.6 months (range 0.5-11.5), and was shorter in lung cancer than others (1.1 vs. 3.1 months; p = 0.008). Most pts (17/20; 85%) received corticosteroids; 3 (15%) also required infliximab. Among 19 pts with available CT, ground glass opacities were noted in all pts, with consolidation in 11. The most common distribution was multifocal (n = 9), followed by peripheral and lower (n = 3) and diffuse (n = 3). Radiographic pattern was cryptogenic organizing pneumonia (COP) in 12, non-specific interstitial pneumonia (NSIP) in 3, hypersensitivity pneumonitis (HP) in 2, and acute interstitial pneumonia (AIP)/acute respiratory distress syndrome (ARDS) in 2 pts. AIP/ARDS pattern had the highest pneumonitis grade, followed by COP, while NSIP and HP had lower grade (median Grade: 3, 2, 1, 1, respectively; p = 0.005). COP was the most common pattern in all tumors and treatment regimens. Conclusions: PD-1 pneumonitis showed a spectrum of radiographic patterns, which were associated with toxicity grades of pneumonitis. COP was the most common pattern across tumor types and therapeutic regimens. The radiographic pattern-based approach may contribute to better diagnose, categorize, and prognosticate in this entity.

3055

Poster Session (Board #377), Sun, 8:00 AM-11:30 AM

Avelumab (MSB0010718C; anti-PD-L1) in patients with advanced cancer: Safety data from 1300 patients enrolled in the phase 1b JAVELIN Solid Tumor trial. First Author: Karen Kelly, UC Davis Comprehensive Cancer Center, Sacramento, CA Background: Avelumab* is a fully human anti-PD-L1 IgG1 antibody showing preliminary efficacy in multiple tumor types. We report updated safety data of single-agent avelumab in patients (pts) with locally advanced or metastatic (LA/M) solid tumors from a phase 1b trial (NCT01772004). Methods: Pts from 16 different expansion cohorts (including NSCLC, gastric, ovarian, urothelial, mesothelioma, and breast), all unselected for PD-L1 expression, received avelumab 10 mg/kg IV Q2W until progression, unacceptable toxicity, or withdrawal. Treatment-emergent adverse events (AEs) were graded by NCI-CTCAE v4.0. Results: As of Nov 5, 2015, 1,300 pts received avelumab and were followed for $ 4 wks. Median age was 63 y (range 20-91), ECOG PS was 0 (37.7%), 1 (62.1%), or 2-3 (0.2%), and median number of prior lines of anticancer therapy was 2 (range 1-13). Median duration of treatment with avelumab and number of administrations were 11.5 wks (range 2-104) and 5 infusions (range 1-50), respectively. Treatment-related (TR) AEs occurred in 813 pts (62.5%), and the most common ( $ 5%) were fatigue (n = 212, 16.3%), infusion-related reaction (IRR; n = 209, 16.1%), nausea (n = 108, 8.3%), chills (n = 102, 7.8%), diarrhea (n = 79, 6.1%), and pyrexia (n = 72, 5.5%). Grade $ 3 TRAEs occurred in 124 pts (9.5%). The most frequent ( $ 0.5%) grade $ 3 TRAEs were GGT elevation (n = 9, 0.7%), IRR (n = 9, 0.7%), fatigue (n = 8, 0.6%), lipase elevation (n = 8, 0.6%), anemia (n = 7, 0.5%), and dyspnea (n = 6, 0.5%). Potential immune-related (ir) TRAEs were reported for 93 pts (7.2%); the most frequent ( $ 1.0%) were hypothyroidism (n = 45; 3.5%) and pneumonitis (n = 13; 1.0%). TRAEs resulted in permanent discontinuation for 79 pts (6.1%): 25 (1.9%) due to an IRR and 14 (1.1%) due to a potential irTRAE. TRAEs were considered the primary cause of death by the investigator for 5 pts (0.4%): radiation pneumonitis (1), pneumonitis (1), autoimmune hepatitis/liver failure (2), and respiratory distress/sepsis (1). Conclusions: Single-agent avelumab showed an acceptable safety profile in a heavily pretreated population and large dataset of pts with LA/M malignancies. Additional analyses and phase 3 trials are ongoing. *Proposed INN. Clinical trial information: NCT01772004.

3054

155s

Poster Session (Board #376), Sun, 8:00 AM-11:30 AM

Phase I clinical trial of combination imatinib and ipilimumab in patients with advanced malignancies. First Author: Matthew Reilley, The University of Texas MD Anderson Cancer Center, Houston, TX Background: Imatinib mesylate can induce rapid tumor regression, inhibit tumor immunosuppressive mechanisms, and release large amounts of antigenic debris from tumor cell death with subsequent increase in tumor antigen presentation. CTLA-4 blockade and imatinib combination therapy in mouse models of GIST synergistically reduces size of tumors mediated by imatinib-dependent intratumoral accumulation of CD8+ T-cells and suppression of Treg-cells. We hypothesized that combination therapy with imatinib and ipilimumab immunotherapy would be tolerable and may synergistically enhance anti-tumor T-cell activation in patients (pts). Methods: Primary objective of the dose-escalation study (3+3 design) was to establish the maximum tolerated dose (MTD) and recommended phase II dose of the combination. Eligible pts had metastatic or unresectable, solid tumors. In the expansion cohort, pts were required to have a known KIT mutation. Secondary objectives included evaluation of antitumor activity of the combination based on KIT mutation status and the predictive status of tumor-associated immune biomarkers. Results: 26 pts were dosed in the escalation portion of the study. No dose limiting toxicities were encountered. The most common grade 1 and 2 related adverse events (AEs) were nausea (57%), fatigue (50%), vomiting (46%), anorexia (35%), diarrhea, skin rash, and edema (each 27%). There were 4 (16%) grade 3 related AEs including fatigue, anemia, rash, and vomiting. There were no grade 4 AEs. The MTD was the highest planned dose level of Ipilmumab at 3mg/kg and Imatinib at 400mg twice daily. An additional 10 pts were treated at MTD. Among all pts, three responses were seen: one GIST (PR) and 2 melanoma pts (CR + PR). These responses were seen at higher dose levels. Stable disease was seen in 7 pts lasting an average of 6 months. Notably, both melanoma responders had KIT mutations; the GIST responder was wild type. Immune cell activity in tumor tissue biopsy specimens and serum is under analysis. Conclusions: In this phase I trial the combination was well tolerated. Responders in this trial suggest that this combination at MTD has antitumor activity in WT GIST and KIT-mutant melanoma and merits further investigation. Clinical trial information: NCT01738139.

3056

Poster Session (Board #378), Sun, 8:00 AM-11:30 AM

A phase 1 study of ramucirumab (R) plus pembrolizumab (P) in patients (pts) with advanced gastric or gastroesophageal junction (G/GEJ) adenocarcinoma, non-small cell lung cancer (NSCLC), or urothelial carcinoma (UC): Phase 1a results. First Author: Roy S. Herbst, Yale University School of Medicine, Yale Cancer Center, New Haven, CT Background: Hallmarks of tumor growth include angiogenesis and immunosuppression. This is the first study to combine R with P to target both processes simultaneously. Preliminary safety results from the dose-limiting toxicity (DLT) portion of the study (phase 1a) are reported. Methods: Eligible pts with advanced G/GEJ, NSCLC or UC and ECOG PS 0-1 who progressed on prior systemic therapy were enrolled. Two dosing regimens were evaluated in a phase 1a following a 3+3 design, schedule (S) 1 (R 8 mg/kg on Days 1&8 in G/GEJ only) and S2 (R 10 mg/kg on Day 1 in G/GEJ, NSCLC, and UC), given concomitantly with P 200 mg on Day 1 q3W. After phase 1A/DLT evaluation, each tumor type was expanded to 30 G/GEJ, 25 NSCLC, and 25 UC pts who will receive study treatment until confirmed disease progression or unacceptable toxicity (phase 1a). The primary objective was to assess safety and tolerability of two dosing regimens; preliminary efficacy will be examined in expansion cohorts. Results: As of data cutoff on 08-Jan-16, 10 pts were treated in phase 1a: 7 G/GEJ pts in S1 and 3 NSCLC pts in S2. S1 is currently ongoing, 1/5 evaluable pts had DLTs. In S2, no DLTs were observed and 3 expansion cohorts are fully enrolled. Treatment emergent adverse events (TEAEs) were reported for 9/10 pts, 5/10 pts had treatment-related AEs (TRAEs). The G/GEJ pt who experienced DLTs reported Grade (G) 3 TRAEs, including abdominal pain, hepatitis, pneumonia, colitis, peritonitis and G4 cholestasis; this pt discontinued study treatment. The other four pts reported G1-2 TRAEs including G1 fever, dysgeusia, dysphonia, asthenopia, and anorexia, and G2 hypothyroidism, dyspnea, pain, pruritus and asthenia; one of each, except pain (n = 2). Three deaths were reported (not related), 2 due to disease progression, and 1 sudden death. Conclusions: Preliminary data from the study did not reveal any unexpected safety signals. At the time of the phase 1a data cutoff, S2 was complete and the DLT observations for S1 are ongoing. An amendment is underway to explore additional tumor types. Additional safety and efficacy for S1 and S2 will be presented at the meeting. Clinical trial information: NCT02443324.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

156s 3057

Developmental Therapeutics—Immunotherapy Poster Session (Board #379), Sun, 8:00 AM-11:30 AM

3058

Poster Session (Board #380), Sun, 8:00 AM-11:30 AM

Biomarkers of immune checkpoint inhibitor response in metastatic breast cancer: PD-L1 protein expression, CD274 gene amplification, and total mutational burden. First Author: Jeffrey S. Ross, Albany Medical College, Albany, NY

Development of the Princess Margaret Immune Oncology Prognostic Index (PM-IPI): A novel prognostic score for patients (pts) treated in immune oncology (IO) phase I (P1) trials. First Author: Daphne Day, Princess Margaret Cancer Centre, Toronto, ON, Canada

Background: Predictive biomarkers for immune checkpoint inhibitors, such as PD-L1 (CD274) copy number, PD-L1 expression and tumor mutational burden (TMB), have not been widely studied in metastatic breast cancer (mBC). Methods: FFPE samples from 84 cases of BC, 71 ductal (IDC) and 13 lobular (ILC), were immunostained (IHC) using rabbit monoclonal PD-L1/ CD274 (SP142). Membranous PD-L1 staining was scored as negative, low positive (1 – 24%) and high positive ( = or . 25%) for both tumor cells and tumor infiltrating lymphocytes (TILs). Comprehensive genomic profiling (CGP) was used to measure PD-L1gene amplification and total TMB, calculated by counting mutations across a 1.25Mb region spanning 315 genes. TMB was also calculated for a separate cohort of 6,751 metastatic BC (mBC). Results: PD-L1 immunoreactivity was noted as low positive in 4/84 (5%) BC [2 tumor+/TIL+ and 2 tumor+/TIL-], and as low positive in the TIL component for 5 (6%) additional BC [tumor-/TIL+]; all 9 (100%) were IDCs. No PD-L1 immunoreactivity was noted in the TIL component of the remaining cases, despite the presence of prominent TILs in most cases. PDL1 protein expression correlated with decreased overall survival (OS) [100% PD-L1+ cases and 0% tumor-/TIL+ cases expired, (p = 0.002)], while showing a trend toward correlation with tumors that did not recur [100% PDL1+ cases and 100% tumor-/TIL+ did not recur, (p = 0.08)] and age at diagnosis in the ER+ subgroup [100% PD-L1+ cases diagnosed at a perimenopausal age and 100% tumor-/TIL+ cases diagnosed at a postmenopausal age, (p = 0.069)]. On multivariate analysis, tumor grade (p = 0.025) and disease recurrence independently predicted OS. PD-L1gene amplification was identified in only 57 (0.1%) of 6,751 mBC. High TMB, defined as greater than the top quartile across all indications, was found in 1,351 of 6,643 (20%) mBC cases. Conclusions: PD-L1 expression is found in both cancer cells and TILs and is a significant adverse prognostic factor in BC. PD-L1 amplification is extremely uncommon, whereas high TMB was found in 20% of mBC. The potential for these biomarkers to identify BC patients as candidates for immunotherapy warrants further study.

Background: Several previously published scores prognosticate survival for pts treated in cytotoxic and molecularly targeted P1 trials. IO is rapidly evolving in early drug development with unique response patterns and survival outcomes. Our aims were to develop a prognostic index for IO P1 trials, compare its performance with previously published P1 scores, and evaluate associations with IO treatment outcomes. Methods: We included 192 consecutive advanced solid tumor pts treated at Princess Margaret Cancer Centre from Aug 2012 to Aug 2015 in 13 IO P1 trials, targeting immune checkpoint and/or co-stimulatory molecules. Ten clinical factors were tested in univariate survival analysis. Multivariate analysis (MVA) was used to develop the PM-IPI. The ability of the PM-IPI to predict overall survival (OS), progression-free survival (PFS), 90-day mortality (90DM) and overall response rate (ORR) was compared with previously published P1 prognostic scores. Results: Median age was 58 y and 81% of pts had at least 1 prior systemic therapy (range 0-8). The most common tumor sites include melanoma (27%), lung (20%), urological (11%), head and neck (10%) and gastrointestinal (9%). Median PFS and OS were 13.4 and 73.6 wk respectively and 90DM was 16%. ORR was 20% by RECIST 1.1 or immunerelated response criteria (irRECIST). In MVA, ECOG PS . / = 1 (HR 3.2, p , 0.001), no. of metastatic sites . 2 (HR 2.0, p = 0.003) and albumin , lower limit of normal (HR 1.8, p = 0.007) were independent prognostic factors; comprising the PM-IPI. Pts with a score of 2-3 compared to pts with a score of 0-1 had significantly shorter OS (HR 3.4, p , 0.001), PFS (HR 2.3, p , 0.001), higher 90DM (OR 8.1, p , 0.001) and lower ORR (OR 0.4, p = 0.019). Additionally, the PM-IPI performed better than previously published P1 prognostic scores (Table). Conclusions: The PM-IPI prognosticates for survival and is associated with ORR and PFS in pts treated in IO P1 trials. Validation of this novel prognostic score in a large external cohort is ongoing.

3059

Poster Session (Board #381), Sun, 8:00 AM-11:30 AM

Safety profile of nivolumab administered as 30-minute (min) infusion: Analysis of data from CheckMate 153. First Author: David Michael Waterhouse, Oncology Hematology Care, Cincinnati, OH Background: Nivolumab,aprogrammed death-1 (PD-1) immune checkpoint inhibitor, is approved in the US for previously treated metastatic NSCLC and renal cell carcinoma, and for advanced melanoma, and in the EU for previously treated squamous NSCLC and advanced melanoma. Here, we assess the impact of infusion time on nivolumab safety from an ongoing, predominantly community-based trial. Methods: Patients who progressed during or after receiving $ 1 prior therapy were enrolled. The primary outcome of this study is incidence of grade (gr) 3–4 and 5 and select treatment-related adverse events (TRAEs). Patients received nivolumab 3 mg/kg IV Q2W; infusion time was shortened from 60 min to 30 min following an amendment to the study protocol. A comparison of toxicities by infusion time is reported in this analysis. Data for the overall population and elderly/ poor performance status patient subgroups are presented in separate abstracts. Results: A total of 322 and 355 patients, respectively, received nivolumab by 30-min or 60-min infusion; demographics for these patients are comparable to the overall population (median follow-up, 6 mo). Any gr (gr 3–4) TRAEs occurred in 53% (11%) and 51% (12%) of patients with 30-min or 60-min infusions, respectively. Gr 3-4 select AEs in $ 2% of patients given 30-min or 60-min infusions occurred in pulmonary (3% and 2%), hepatic (2% and 3%), and gastrointestinal (2% and 2%) categories. Any gr (gr 3–4) infusion reaction occurred in 3% and 2% ( , 1% and , 1%) of patients given 30-min or 60-min infusions, respectively (Table). Additional data by infusion time for this subgroup will be presented, including safety (use of steroids, incidence of discontinuation) and pharmacokinetics. Conclusions: Nivolumab3 mg/kg can be safely infused over 30 min, with a safety profile comparable to 60-min infusion and with a low incidence of infusion-related reactions. Clinical trial information: NCT02066636. Nivolumab 3 mg/kg Q2W 30-min infusion n = 322

Infusion reaction, n (%) Hypersensitivity Infusion-related reaction Anaphylaxis Bronchospasm

60-min infusion n = 355

Any gr

Gr 3–4

Any gr

Gr 3–4

9 (3) 5 (2) 2 ( , 1) 1 ( , 1) 1 ( , 1)

2 ( , 1) 0 0 1 ( , 1) 1 ( , 1)

6 (2) 1 ( , 1) 5 ( , 1) 0 0

3 ( , 1) 0 3 ( , 1) 0 0

OS PFS 90DM ORR

3060

C- index AUC

PM-IPI

Royal Marsden Index

Hammersmith Score

Nijmegen Score

0.71 0.66 0.75 0.64

0.65 0.63 0.69 0.59

0.59 0.58 0.70 0.56

0.59 0.58 0.68 0.58

Poster Session (Board #382), Sun, 8:00 AM-11:30 AM

A first-in-human phase I study of the anti-PD-1 antibody PDR001 in patients with advanced solid tumors. First Author: Aung Naing, Department of Investigational Cancer Therapeutics (Phase 1 Program), The University of Texas MD Anderson Cancer Center, Houston, TX Background: PDR001 is a humanized anti-PD-1 IgG4 antibody that blocks the binding of PD-L1 and PD-L2 to PD-1. PDR001 binds to PD-1 with high affinity and inhibits the biological activity of PD-1. Methods: Phase I dose escalation study of PDR001 given i.v. Q2W or Q4W to adult patients with advanced solid tumors. A Bayesian linear model estimating the dose– exposure relationship and a Bayesian logistic regression model with overdose control were used to guide dose escalation. Preliminary population PK analysis was conducted to estimate PK parameters and assess the impact of weight as a covariate on clearance and volume of distribution. Results: In the dose escalation part of the study, 58 patients have been treated with PDR001 at 5 dose levels and schedules: 1 mg/kg Q2W (16 pts); 3 mg/kg Q2W (15), 10 mg/kg Q2W (11), 3 mg/kg Q4W (6) and 5 mg/kg Q4W (10). 27/ 58 patients have been treated for . 3 cycles (cycle length: 28 days), and 27/ 58 remain on treatment at the time of abstract preparation. No dose limiting toxicity was reported. The most common AEs included: nausea, fatigue, anemia, diarrhea, dyspnea, vomiting, abdominal pain and headache. Three cases of hypothyroidism suspected to be related to PDR001 were reported. A single case of grade 3 autoimmune colitis was the only serious AE suspected to be related to PDR001. Approximately dose-proportional increases in exposure were observed from 1–10 mg/kg. PDR001 achieves an AUC0-336h of approximately 1000 mg*day/mL at cycle 3 with 3 mg/kg Q2W or 5 mg/kg Q4W. Based on the population PK analysis, a flat dose of 400 mg Q4W is predicted to achieve mean steady-state Ctrough concentrations of approximately 31 mg/mL (90% CI: 22–42 mg/mL), which exceeds the ex vivo EC50 for PD-1 blockade. The recommended phase 2 dose (RP2D) for PDR001 is therefore 400 mg Q4W. An alternative dosing regimen of 300 mg Q3W is expected to achieve similar exposure to 400 mg Q4W. Conclusions: Based on the preliminary data, PDR001 was well tolerated with a safety profile similar to those of other marketed anti-PD-1 antibodies. Based on the available PK and safety data described above, the RP2D of PDR001 has been declared as 400 mg i.v. Q4W, or 300 mg i.v. Q3W as an option for more convenient scheduling in combination treatment regimens. Clinical trial information: NCT02404441.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Immunotherapy 3061

Poster Session (Board #383), Sun, 8:00 AM-11:30 AM

Systemic inflammatory response syndrome (SIRS) with immune checkpoint inhibitors. First Author: Nitika Sharma, Vidant Health, Greenville, NC Background: Immune checkpoint blockade has emerged as an attractive treatment option in a wide spectrum of malignancies. Optimal use of these agents requires prompt recognition and management of immune-mediated toxicities. SIRS with immunotherapy is a result of exponential T-cell proliferation with marked elevation of inflammatory cytokines. C-reactive protein (CRP) is a reliable surrogate for IL-6 expression which mediates toxicity in cytokine release syndrome (CRS) (Lee CW et al. Blood 2014;124:188195.). CRS has been reported in patients receiving CAR-T therapy or BiTE antibodies but has not been described with agents that modulate CTLA-4 or PD-1. Methods: In this study we prospectively monitored 32 patients with progressive lung cancer who were put on 3 mg/kg nivolumab between June to December 2015 at a single institution. CRP was drawn before each treatment or in case of hospitalization. Results: Twelve patients (37.5%) were hospitalized with signs and symptoms of SIRS with one or more of the following - fever, tachypnea, tachycardia, hypotension, organ failure occurring 1-2 days to weeks after immunotherapy. The mean CRP pretreatment was found to be 38mg/L (0.1–192mg/L) (normal , 2.6mg/L). The mean CRP in patients who developed symptoms post treatment was 86mg/L (46-144mg/L) compared to 28mg/L (6-71mg/L) in asymptomatic patients. The symptomatic patients were treated with best supportive care, 1-2mg/kg equivalent dose of methylprednisolone. Eight patients were treated with anti- IL-6R monoclonal antibody, tocilizumab at a dose of 4 mg/ kg with improvement in inflammatory symptoms and CRP. Conclusions: SIRS is a serious and potentially life threatening complication of immunotherapy particularly noted in patients with lung cancer. Elevations in CRP seem to correlate with symptom severity. Prompt initiation of corticosteroids and IL-6 modulating therapy is critical to optimize the risk/ benefit profile of PD-1 targeted therapy. Symptoms (no. of patients) Generalized asthenia (1) Pneumonitis (3) Hypophysitis(2) Cerebritis, limbic encephalitis(3) Seizures (1) Hepatitis (2)

3063

Mean CRP before anti-PD-1

Mean CRP during SIRS

Mean CRP after steroids and anti-IL6

18 17 16 148 9

233 158 193 127 148 105

60 17 23 12 17 17

Poster Session (Board #385), Sun, 8:00 AM-11:30 AM

Immunogenicity of pembrolizumab (pembro) in patients (pts) with advanced melanoma (MEL) and non-small cell lung cancer (NSCLC): Pooled results from KEYNOTE-001, 002, 006, and 010. First Author: Marianne van Vugt, Merck & Co., Inc., Kenilworth, NJ Background: Monoclonal antibodies (mAb) can evoke immunogenicity, leading to production of antidrug antibodies (ADA). Pembro, a humanized mAb against PD-1, has demonstrated efficacy in NSCLC and MEL. Results from NSCLC cohorts of KEYNOTE-001 (NCT01295827) and KEYNOTE010 (NCT01905657) and from MEL cohorts of KEYNOTE-001, KEYNOTE002 (NCT01704287), and KEYNOTE-006 (NCT01866319) are presented. Methods: Samples were analyzed for antipembro antibodies using electrochemiluminescence immunoassay methodology following a standard 3tiered assay approach consisting of screening, confirmatory, and antibody titer assessment. Screening positives were tested in a confirmatory assay checking specificity of signal for pembro. In confirmed positive samples Ab titer, neutralizing capacity, drug exposure, efficacy, and AEs were assessed. The immunogenicity approach classified ADA samples as positive, negative, or inconclusive. Overall immunogenicity incidence was defined as proportion of treatment-emergent (TE) positive pts to the number of evaluable pts with a confirmed TE, non–TE-positive or negative immunogenicity status. Negative immunogenicity status was confirmed if ADA results of the pretreatment and postdose samples were negative in the confirmatory assay, and the pembro concentration was below the drug tolerance level in the last postdose sample. Results: Across cohorts, 2910 pts provided 11,886 pretreatment and posttreatment serum samples. Overall, 2632 pts (n = 1535 MEL, n = 1097 NSCLC) were included in the immunogenicity analysis. False-positive rates were 2.7% and 2.8% with Intertek and PPD assays, respectively. Of the 1087 (41%) evaluable pts, 19 (1.7%) had a TEpositive ADA, 10 (0.9%) were non–TE-positive, and 1058 (97.3%) had negative immunogenicity status. Initial results indicate a higher incidence rate in NSCLC vs MEL (2.5% vs 0.7%). In pts with a TE-positive ADA, no impact of antipembro antibodies on drug exposure, safety, or efficacy was observed. Conclusions: Pembro has a low potential to elicit the formation of ADA, and if ADA formation occurs, it has been shown to have no clinically relevant impact. Clinical trial information: NCT01295827, NCT01905657, NCT01704287, NCT01866319.

3062

157s

Poster Session (Board #384), Sun, 8:00 AM-11:30 AM

Treatment-related side effects as predictors of efficacy of check-point inhibitors (CPIs). First Author: Rebecca M. Prince, Princess Margaret Hospital, Toronto, ON, Canada Background: There is increasing interest in immunotherapy as a treatment for cancer with multiple agents studied in various solid tumors. CPIs activate the host immune system against tumor cells; as a consequence of this mechanism of action, treatment-related adverse events (AEs) can occur. We hypothesized that occurrence of AEs is associated with longer survival. Methods: A systematic review of MEDLINE (from inception until December 2015) identified randomized studies reporting the hazard ratio (HR) for overall survival (OS) as well as data on safety which allowed the calculation of the odds of AEs comparing experimental therapy to control. Safety and tolerability outcomes of interest were treatment discontinuation without progressive disease, grade 3/4 AEs as well as immune-related adverse events (irAEs). If reported, irAEs were extracted from manuscripts. Otherwise, irAEs were assumed to comprise colitis or diarrhea, increased transaminases or endocrinopathy (hypophysitis, thyroiditis, pancreatitis). The association between AEs and OS was explored using metaregression (simple linear regression weighted by sample size). Results: Of 130 articles identified initially, 18 met inclusion criteria. Included trials were all published after 2009 and explored CPIs in melanoma, renal cell carcinoma, prostate cancer, non-small cell and small cell lung cancers. CPIs studied were ipilimumab (n = 12), pembrolizumab (n = 2), and nivolumab (n = 7). There were nonsignificant positive associations between treatment discontinuation without progressive disease, grade 3/4 AEs, and irAEs, with worse survival (see Table). Conclusions: AEs with CPIs do not predict for improved OS. Positive associations between the odds of AEs and the HR for OS suggest that prolonged exposure to immunotherapeutics may be beneficial. Safety categories Treatment discontinuation without progression Any grade 3/4 AEs Reported irAE Presumed irAE

3064

R

p

0.44 0.40 0.19 0.40

0.18 0.22 0.81 0.22

Poster Session (Board #386), Sun, 8:00 AM-11:30 AM

Toxicity profile of approved anti-PD1 monoclonal antibodies in solid tumors: A systematic review and meta-analysis of randomized clinical trials. First Author: Ricardo Lima Barros Costa, Northwestern Medicine Developmental Therapeutics Institute, Robert H. Lurie Comprehensive Cancer Center of Northwestern University, Chicago, IL Background: Clinical trials supported the FDA approval of anti-PD1 antibodies nivolumab and pembrolizumab for treatment of advanced melanoma, renal cell carcinoma, and non-small cell lung cancer. This meta-analysis aimed to investigate the relative risk (RR) of adverse events (AEs) when compared to standard of care. Methods: PubMed search was conducted using the keywords or medical subject headings: “nivolumab” and “pembrolizumab”. Solid tumor randomized trials meeting the following inclusion criteria were included: at least one of the study arms consisting of nivolumab or pembrolizumab monotherapy; at least one of the control arms containing no anti-PD1 therapy. Treatment related (AEs) data were extracted according NCI CTCAE version 4.0 classification. The Q heterogeneity statistic was tested using a chi-square test. The percentage of between study variance was estimated using the I-squared statistic. Results: Our search yielded 442 published articles, after screening of the study titles 418 studies were excluded. After abstract and text review, eight phase III trials and one phase II randomized trial were included in our meta-analysis. A total of 5,666 patients were evaluable for toxicity. There was evidence of significant heterogeneity between studies. The random effect pooled RR for treatmentrelated all grade AEs and all grade 3 and 4 AEs was 0.878 (95% CI 0.8070.955; p= 0.002) and 0.292 (95% CI 0.192-0455; p , 0.001) favoring anti-PD1 therapy versus standard of care approach. The pooled RR for all serious and grade 3&4 serious AEs was 0.566(95% CI 0.307-1.041; p= 0.067) and 0.45 (95% CI 0.194-1.044; p = 0.063). The RR of treatmentrelated death was 0.448 (95% CI 0.185-1.085; p= 0.075). In this large cohort, all grade hypothyroidism, rash, pneumonitis and colitis were documented in 6.5%, 12.5%, 2.6%, and 1% of patients, respectively. Conclusions:PD1 inhibitors (nivolumab and pembrolizumab) are associated with lower risk of AEs when compared to standard of care therapy. Despite their clinical relevance, the risk of selected immune-related adverse events was lower in this large cohort when compared to previously published phase I and II trials.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

158s

Developmental Therapeutics—Immunotherapy

3065

Poster Session (Board #387), Sun, 8:00 AM-11:30 AM

3066

Poster Session (Board #388), Sun, 8:00 AM-11:30 AM

Tumor- and class-specific patterns of immune-related adverse events (irAEs) of immune checkpoint inhibitors (ICIs): A systematic review (SR). First Author: Daphne Day, Princess Margaret Cancer Centre, Toronto, ON, Canada

A phase I dose-escalation study of BGB-A317, an anti-programmed death-1 (PD-1) mAb in patients with advanced solid tumors. First Author: Jayesh Desai, Royal Melbourne Hospital, Melbourne, Australia

Background: ICIs targeting cytotoxic T-lymphocyte antigen 4 (CTLA-4), programmed cell death protein 1 (PD-1) and its ligand (PD-L1) produce a unique profile of toxicities driven by T cell tissue infiltration. The objective of this SR was to identify patterns and incidence of irAEs based on tumor type and ICI class. Methods: A SR was performed in MEDLINE, EMBASE and COCHRANE databases to identify prospective trials investigating single agent ICIs up to Nov 2015. Paired reviewers selected studies for inclusion and extracted data. Odds ratio (OR) with 95% confidence interval (CI) was used to quantify the effect of variables on the incidence of irAEs and contingency table p values were calculated using the chi-squared test. Results: We identified 48 clinical trials with a total of 6872 patients (pts), including 27 CTLA-4, 17 PD-1, 2 CTLA-4 vs PD-1 and 2 PD-L1 monoclonal antibody (mAb) trials. Nine were phase III and 39 were phase I/II trials. Grade 3/4 irAEs were more common with CTLA-4 mAbs compared with PD-1 mAbs (31% vs 10%). All grades colitis (OR 8.7, 95% CI 5.8-12.9), hypophysitis (6.5, 3.0-14.3) and rash (2.0, 1.8-2.3) were more common with CTLA-4 mAbs; whereas pneumonitis (6.4, 3.2-12.7), hypothyroidism (4.3, 2.9-6.3), arthralgia (3.5, 2.6-4.8) and vitiligo (3.5, 2.3-5.3) were more common with PD-1 mAbs. Comparison of irAEs from the 3 most studied tumor types in PD-1 mAbs trials [melanoma (n = 2048), non-small cell lung cancer (NSCLC, n = 1030) and renal cell carcinoma (RCC, n = 573)] showed that melanoma pts had a higher frequency of gastrointestinal and skin irAEs and lower frequency of pneumonitis (Table). Conclusions: We demonstrated that CTLA-4 and PD-1 mAbs have distinct irAE profiles presumably by checkpoint-specific immune modulation. Different immune microenvironments may drive histology-specific irAE patterns. Other tumor-dependent irAE profiles may be identified as data emerge from ICI trials.

Background: BGB-A317 is a humanized IgG4 anti-PD-1 mAb with high specificity and affinity (KD = 0.15 nM) for PD-1. It blocks PD-L1 and PD-L2 binding, and inhibits PD-1-mediated negative signaling in T-cell lines and tumor growth in a number of allogeneic xenograft models. Methods: A phase I, multicenter study was conducted to evaluate the safety, tolerability, pharmacokinetics (PK) and antitumor activity of BGB-A317 in patients (pts) with advanced solid tumors. A 3+3 dose escalation design was undertaken. Pts received escalating doses of BGB-A317 intravenously biweekly (Q2W). Additional pts were treated at 2 and 5 mg/kg either Q2W or Q3W to explore alternate schedules. Results: As of 15 Dec, 2015, 61 patients were treated across 4 dose-escalating cohorts of BGB-A317 Q2W (0.5 mg/kg, n = 3; 2 mg/ kg, n = 6; 5 mg/kg, n = 6 and 10 mg/kg, n = 6) and 4 dose-expansion cohorts (2 mg/kg, Q2W, n = 20; 2 mg/kg, Q3W, n = 2; 5 mg/kg Q2W, n = 17 and 5 mg/ kg, Q3W, n = 1). One DLT (1/6) of grade 3 colitis was observed in the 5 mg/kg Q2W cohort. Maximum tolerated dose was not reached. The most common treatment-emergent adverse events (AEs) were grade (G) 1-2 fatigue (25%), diarrhea (20%), nausea (16%), rash (13%), pruritus (11%) and abdominal pain (11%). Treatment–related G3 AEs included diabetic ketoacidosis (n = 1, 2 mg/kg Q2W), hypotension (n = 1, 2 mg/kg Q2W), colitis (n = 2, 5 mg/kg Q2W), elevated ALT (n = 1, 2 mg/kg Q2W), hyperglycaemia (n = 1, 2 mg/kg Q2W) and fatigue (n = 1, 5 mg/kg Q2W). PKs of BGB-A317 show doseproportional exposure increase from 0.5 to 10 mg/kg Q2W after the single dose administration with the elimination half-life of 11 to 17 days. Among 39 evaluable pts to date (10 ovarian, 6 colorectal, 6 RCC, 4 cutaneous SCC, 3 mesothelioma, 2 cervical, and 1 each of 8 others), 3 pts have partial response (2 to be confirmed) (1 gastric, 1 RCC and 1 cervical) and a further 13 pts exhibit stable disease. All 3 responding pts remain on treatment, ranging from 13 to 23 weeks. Conclusions: BGB-A317 demonstrates a favorable safety profile with AEs in keeping with the class effect. PKs are linear and early promising anti-tumor activity has been observed. The expansion cohorts are ongoing and an expanded phase IB study in selected cancer types is planned. Clinical trial information: NCT02407990.

Melanoma vs NSCLC

Melanoma vs RCC

irAE

OR

95% CI

pvalue

OR

95% CI

pvalue

Colitis Diarrhea Pruritus Rash Pneumonitis

4.2 1.9 2.4 1.8 0.4

1.3-14.0 1.5-2.5 1.9-3.1 1.4-2.3 0.3-0.7

0.01 , 0.001

No event in RCC 1.3 1.5 1.6 0.3

1.1-1.8 1.2-2.0 1.2-2.1 0.2-0.6

0.04 0.003 0.002 , 0.001

3067

Poster Session (Board #389), Sun, 8:00 AM-11:30 AM

Italian cohort of nivolumab Expanded Access Programme (EAP): Preliminary data from a real-world population. First Author: Lucio Crino, Clinical Oncology, S. Maria della Misericordia Hospital, Perugia, Italy Background: Nivolumab is the first checkpoint inhibitor approved for the treatment of squamous non small cell lung cancer (Sq-NSCLC) to show a survival benefit in a randomised phase III trial. The experience of patients and physicians in routine clinical practice is often different from those in a controlled clinical trial setting. The purpose of this analysis is to evaluate nivolumab use in real world setting. Methods: Nivolumab was available upon physician request for patients (pts) aged $ 18 years who had relapsed after a minimum of one prior systemic treatment for stage IIIB/stage IV Sq-NSCLC. Nivolumab 3 mg/kg is administered intravenously every 2 weeks to a maximum of 24 months. Pts included in the analysis had received $ 1 dose of nivolumab and were monitored for adverse events using Common Terminology Criteria for Adverse Events. Results: In total, 371 Italian pts participated in the EAP across 96 centres. Table 1 shows pts baseline characteristics. With a median follow-up of 2 months (range 0-7), 139 pts are not yet evaluable for response; the disease control rate (DCR) among others 232 pts evaluable for response is 35.8%: one patient (0,4%) with complete response, 32 pts (13.8%) with partial response and 50 pts (21.6%) with stable disease. DCR is comparable among pts regardless previous lines of therapy, brain metastasis, age and smoking habits. To date, median time of treatment is 2 months, with 42 patients treated beyond progression. Among 370 pts, 99 pts (27%) discontinued treatment for any reason except toxicity; 16 out of 370 discontinued due to AE (4%). Conclusions: To date, this is the largest clinical experience with nivolumab in current clinical practice: nivolumab seems to be an effective and safe therapy for pre-treated pts with Sq-NCSLC. These preliminary data provide insights into routine clinical use of nivolumab and seems to confirm data from pivotal trials.

3068

Poster Session (Board #390), Sun, 8:00 AM-11:30 AM

Comprehensive analysis of five key immune related adverse events (irAE) from immune checkpoint blockers (ICB) CTLA-4 and PD-1 inhibitors in cancer patients. First Author: Guillermo de Velasco, Dana-Farber Cancer Institute, Boston, MA Background: IrAEs have been described with ICB, though the overall risk of irAEs with these drugs remains unclear. Our aim was to determine the incidence and the relative risk (RR) of key irAEs from approved immune checkpoint blockers (ICB) to better characterize their safety profile. Methods: PubMed and Medline databases were searched from Jan1996 to Oct2015. Eligible studies included randomized phase II/III trials with nonICB control arms. Statistical analyses were conducted to calculate the summary incidence, RRs, and 95% CIs, using random-effects models. The 5 selected key irAEs included: colitis, transaminitis (increased AST), immunerelated rash, hypothyroidism, and pneumonitis. Results: A total of 6133 patients (pts) from 12 trials were included: 3499 pts assigned to singleagent ICB arms (nivolumab [n = 1298], pembrolizumab [n = 357], ipilimumab [n = 1844]) and 2634 to non-ICB arms. All-grade incidence of each selected key irAE with ICB was: colitis 3.1%, increased AST 7.8%, rash 16.7%, hypothyroidism 4.2 % and pneumonitis 2.4%. Compared to nonICB arms, the RR of all-grade irAE with ICB was: colitis 8.88 (95% CI, 5.09 to 15.5; P , 0.001), increased AST 1.89 (95% CI, 1.05 to 3.40; P = 0.034), rash 2.36 (95% CI, 1.39 to 3.99; P , 0.001), hypothyroidism 7.17 (95% CI, 3.57 to 14.4; P , 0.001) and pneumonitis 3.96 (95% CI, 0.68 to 23.2; P = 0.127). Grade 3/4 incidence of each selected key irAE with ICB was: colitis 2.0%, increased AST 2.0%, rash 1.0%, hypothyroidism 0.3% and pneumonitis 0.9%. Compared to non-ICB arms, the RR of grade 3/4 irAE with ICB was: colitis 6.83 (95% CI 2.68 to 17.4; P , 0.001) and increase AST 3.35 (95% CI 1.25 to 8.98; P = 0.016). Ipilimumab had higher RR of grade 3/4 colitis (p = 0.033) and all-grade rash (p = 0.047), than nivolumab/ pembrolizumab, but no other significant differences were found. Conclusions: This meta-analysis offers substantial evidence that ICB treatment is associated with a small but significant increase in the risk of several all-grade irAEs analyzed, as well as grade 3/4 colitis and transaminitis compared to non-ICB treated pts.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Immunotherapy 3069

Poster Session (Board #391), Sun, 8:00 AM-11:30 AM

3070

159s

Poster Session (Board #392), Sun, 8:00 AM-11:30 AM

Immune related adverse events as a biomarker in non-melanoma patients treated with programmed death 1 inhibitors (PD-1Is). First Author: Julia Judd, Temple University Health System, Philadelphia, PA

Is there a clinical benefit of anti-PD-1 in patients older than 75 years with previously treated solid tumour? First Author: Thierry Landre, Hopital ˆ Rene Muret, HUPSSD, UCOG 93 (APHP), Sevran, France

Background: The PD-1Is pembrolizumab (P) and nivolumab (N) have demonstrated clinical activity in multiple tumor types. They can lead to unique immune related adverse events (irAEs). Studies have shown conflicting results on whether the development of irAEs correlates with clinical response in melanoma patients (pts). Correlation of irAEs with better response in other cancer subtypes is unknown. Methods: We conducted a retrospective study of non-melanoma pts who received PD-1Is at our institution. The primary objectives were to correlate the development of any irAE, or any irAE requiring steroid use, with clinical response. Incidence of irAEs and steroid use was assessed via review of all relevant patient charts. Response was recorded at first restaging scan, and endpoints included overall response rate (ORR) per RECIST, overall survival (OS), and time to next therapy or death (TTNTD). Fisher’s exact tests were used to determine the association between irAE incidence and ORR and Kaplan-Meier curves with log-rank tests for the comparison of OS and TTNTD. Results: From November 2011-December 2015, 135 patients were treated with N or P; average age was 64 years, 88 (65%) were male. The majority received N (101, 75%). Cancer sub-type was as follows: non-small cell lung cancer (59, 44%), renal cell carcinoma (31, 23%), head and neck cancer (24, 18%), urothelial carcinoma (18, 13%), other (3, 2%). Overall, irAEs were noted in 60 (45.9%) with steroids required in 41 (30.4%). Endocrinopathies, dermatitis and colitis occurred in 25.9%, 7.4% and 6.7% of patients, respectively. Of the 112 pts for whom complete response data was available, we found no significant correlation between ORR and either incidence of any irAE (p = 0.21) or the use of steroids (p = 0.27). There was also no significant association of irAE incidence or steroid use with OS (p = 0.981, p = 0.954 respectively) or TTNTD (p = 0.277, p = 0.377 respectively). Conclusions: This retrospective review showed no association between the development of irAEs and outcomes in non-melanoma pts treated with PD1Is. This data may be limited by sample size and disease heterogeneity, or may reflect differences in biology as compared to patients with melanoma.

Background: The immune system undergoes profound transformations with age, and response patterns to immune challenges are therefore highly age dependent. Nivolumab, an anti-PD-1 immune checkpoint inhibitor, showed promising activity and tolerability in several solid tumours. However, in our knowledge, few data are available in patients older than 75 years. Methods: We performed an age subgroup analysis of published randomized control trials concerning nivolumab versus standard therapy in previously treated patients with advanced solid tumours. Overall Survival (OS) among the elderly ( $ 75 years) was compared with that of younger patients ( $ 65 to , 75 years). Hazard ratios (HRs) with their 95 % confidence interval (CI) were collected from the studies and pooled. A fixed-effect model was used. Results: Few studies have been published corresponding to our inclusion criteria. Two studies (CheckMate 017 for Squamous and CheckMate 057 for Non-Squamous) assessed nivolumab versus docetaxel in Non-Small Cell Lung Cancer (NSCLC) and one study (CheckMate 025) assessed nivolumab versus everolimus in Renal-Cell Carcinoma (RCC). Our pooled-analysis included 146 patients older than 75 years (72 NSCLC, 74 RCC) and 541 patients between 65 and 75 years (291 NSCLC and 250 RCC). All patients were mostly men (68%), with good Performance Status (0 or 1). For patients $75 years, OS of the nivolumab arm was inferior to that of the control arm (HR = 1.22; 95% CI 0.80-1.85; p = 0.36). Conversely, among younger patients ($65 to , 75 years), a statistically significant OS benefit was seen with nivolumab (HR = 0.62; 95 % CI 0.50-0.77; p , 0.0001). Similar trends were observed for PFS in patients with NSCLC (HR = 1.24; 95 % CI 0.73-2.09; p = 0.43 for patients $75 years and HR = 0.78; 95 % CI 0.601.01; p = 0.06 for patients $65 to , 75 years). Conclusions: Nivolumab survival benefit in patients older than 75 years appears uncertain. Several significant changes in the elderly individuals in regulatory elements of CD4+ and CD8+ T cells could explain the lack of effectiveness of nivolumab.These results should be interpreted with caution due to the small number of patients. Further studies are warranted to evaluate anti-PD-1 efficacy in elderly.

3071

3072

Poster Session (Board #393), Sun, 8:00 AM-11:30 AM

Poster Session (Board #394), Sun, 8:00 AM-11:30 AM

Frequency of PDL-1 expressed on circulating epithelial tumor cells (CETCs) compared to PDL-2 and role in immune evasion in cancer progression. First Author: Monika Pizon, Transfusion Center Bayreuth, Bayreuth, Germany

Identification of novel antagonistic anti-PD-1 antibodies that are nonblocking of the PD-1 / PD-L1 interaction. First Author: Craig Fenwick, MabQuest Sa, Pully, Switzerland

Background: The current cancer research strongly focuses on the immune therapies, where the programmed death-1 receptor (PD-1), with its ligands PD-L1 and PD-L2 play an important role. It is known that PDL-1 is frequently upregulated in a number of different cancers and the relevance of this pathway has been extensively studied and therapeutic approaches targeting PD-1 and PD-L1 have been developed. However, PD-L2 has not received as much attention and its role in modulating tumor immunity is less clear. To identify potential patients who may benefit from PD-1/PD-L1/PD-L2 targeted immunotherapeutics, we used a non-invasive, real-time liquid biopsy for determining PD-L1 and PDL-2 expression in CETCs of breast cancer patients. Methods: CETCs were determined from blood of 66 patients suffering from breast cancer. The number of vital CETCs and the expression of PDL-1 and PDL-2 were investigated using the maintrac method. Results: PDL-1 expressing CETCs were detected in 95 % of breast cancer patients whereas only 75% patients had PDL-2 positive CETCs. There was no association between the frequency of PDL-1 positive CETCs and progression of cancer disease. We found that the fraction of PDL-1 positive CETCs is significantly higher than the fraction of PDL-2 positive CETCs (63.3% vs. 32.9%; p ). Moreover, we observed a significant heterogeneity in PDL-1 and PDL-2 immunostaining intensity across CETCs from the same patients. Conclusions: Breast cancer patients have detectable CETCs with high frequency of PDL-1 regardless of stage of disease. PDL-1 seems to be a major factor in immune evasion and may be a promising target of anticancer therapies. Monitoring the frequency of PDL-1 positive CETCs could reflect individual patient’s response for an anti-PD-1/PDL-1 therapy.

Background: Monoclonal antibodies (mAbs) that act through PD-1 blockade have had significant clinical success as cancer immunotherapeutic agents particularly in melanoma and non-small cell lung cancers with considerably lower toxicities compared to similar therapies. Current anti-PD-1 Abs have relied on the blockade of the PD-1/PD-L1 interaction. Methods: A screening strategy was established to identify a panel of mAbs that bind with high affinity to diverse epitopes on PD-1. In vitro functional activity of the mAbs was evaluated by the capacity to restore proliferation to exhausted CD8 T cells from viremic HIV infected donors and using a mixed lymphocyte reaction assay (MLR). Results: Our studies show that antagonistic mAbs targeting PD-1 can be blocking or non-blocking of the PD-1/PD-L1 interaction. Compared to benchmark blocking mAbs (pembrolizumab and nivolumab), an epitope specific set of non-blocking anti-PD-1 mAbs have equivalent antagonistic activity in: 1) restoring proliferation to exhausted HIV specific CD8 T cells and 2) promoting proliferation and up-regulating IFNg production of CD4 T cells in a MLR assay. Importantly, combinations of blocking and non-blocking anti-PD-1 mAbs synergize in significantly improving functional activity as to that achieved with either therapeutic agent alone. Conclusions: A panel of non-blocking mAbs targeting PD-1 with strong antagonistic activity have been generated and extensively characterized in their ability to recover exhausted antigen-specific CD4 and CD8 T cells and to improve a variety of functions. Importantly, combination of non-blocking with blocking anti-PD-1 mAbs resulted in significantly improved functional activity. These results provide the scientific rationale to investigate whether combination therapy may enhance therapeutic benefit in cancer immunotherapy or whether non-blocking anti-PD-1 Abs may be effective in low PD-L1 positive tumors. Humanized non-blocker anti-PD-1 mAbs have been generated and are being advanced into clinical development.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

160s 3073

Developmental Therapeutics—Immunotherapy Poster Session (Board #395), Sun, 8:00 AM-11:30 AM

Peripheral blood T cell subset phenotype analysis in melanoma patients treated with combination nivolumab + ipilimumab compared to ipilimumab alone. First Author: Michael Andrew Postow, Memorial Sloan Kettering Cancer Center, New York, NY Background: The combination of nivolumab and ipilimumab (N+I) has superior clinical activity vs. ipilimumab (I) alone. Research is needed to understand the immunologic mechanisms unique to N+I that may be related to its increased efficacy. Using peripheral blood (PB) from a randomized study, we compared immunologic changes in patients (pts) treated with N+I to changes observed in pts treated with I alone. Methods: 22 patients treated at Memorial Sloan Kettering Cancer Center on clinical trial CA209069 (Checkmate 069) were consented to an IRB-approved correlative protocol for collection of PB samples. Patients were randomized either to receive the combination of N+I (n = 11) or I alone (n = 11). PB was collected at baseline and every 3 weeks. The frequency and phenotype of T cell populations was determined using flow cytometry. Pharmacodynamic (PD) changes from baseline to just before the 2nd dose (week 3) in PB T cell populations were compared between patients receiving combination N+I vs. I alone. Results: The combination of N+I resulted in distinct PD effects on PB T cell phenotype compared to I alone. Within the CD8+ T cell compartment, Ki67 +, CTLA-4+, and TIM3+ populations increased to a greater amount (p = 0.008, p = 0.002, and p = 0.004, respectively) in pts treated with N+I compared to I alone. Among CD4+ T cells, CTLA-4+ and FOXP3+ populations increased to a greater amount (p = 0.033 and p = 0.033, respectively) in pts treated with N+I compared to I alone. Among the CD4 +FOXP3+ (Treg) cells, there were significant differences in the change in frequency of TIM3+ and PD-1+ Tregs in pts treated with N+I compared to I alone (p = 0.008 and p , 0.001, respectively). There were no significant differences between pts treated with N+I and I alone in the change in ICOS +CD8+ or ICOS+CD4+ T cells. Conclusions: In this small, single-institution, retrospective analysis, unique immunologic changes were seen in the PB of pts treated with N+I compared to I alone in some, but not all, examined T cell populations. Whether the different PD immunologic effects are related to the increased clinical activity of N+I vs. I alone requires additional study.

3075

Poster Session (Board #397), Sun, 8:00 AM-11:30 AM

Updates on phase1b/2 trial of the indoleamine 2,3-dioxygenase pathway (IDO) inhibitor indoximod plus checkpoint inhibitors for the treatment of unresectable stage 3 or 4 melanoma. First Author: Yousef Zakharia, University of Iowa Hospitals and Clinics, Holden Comprehensive Cancer Center, Iowa City, IA Background: IDO is an enzyme that catalyzes tryptophan to kynurenine. Tryptophan depletion enhances the function of the suppressive Treg and inhibits the effector T cells. Kynurenine metabolites may augment the suppressive effect on the immune responses. The normal physiologic function of IDO is the regulation of acquired local and peripheral immune tolerance. In cancer, IDO can either be expressed directly by the tumor cells, or induced indirectly in host antigen presenting cells by the presence of tumor. In these settings, the IDO pathway mediates an acquired immune tolerance towards tumors, allowing tumors to thwart an immune response by the host. Therefore, IDO is an attractive target. Anti- CTLA-4 (ipilimumab) and anti-PD-1 (pembrolizumab and nivolumab) are monoclonal antibodies that block CTLA-4/ PD-1, enhancing immune responses against tumors. Tumor models have shown synergistic effect with anti-CTLA-4 and anti-PD-1 treatments in combination with indoximod providing a rationale for this trial. Methods: Metastatic melanoma patients were treated in a standard design (3 +3) dose escalation of oral indoximod twice daily (BID) with ipilimumab (3mg/kg q3 weeks x 4 doses) in Phase 1b. In Phase 2, indoximod is given with provider choice of Ipilimumab, pembrolizumab or nivolumab. In the event of progression, the provider can change therapy from one checkpoint inhibitor (anti-CTLA-4 or anti-PD-1) to another while continuing indoximod. Primary objective is overall response rate, secondary objectives median progression free and overall survival. Results: To date, 45 patients of a planned 105 are enrolled of which 9 patients comprised the Phase 1b cohort without DLT. The recommended Phase 2 dose was 1200 mg BID. 1 patient in Phase 1 demonstrated an ongoing CR currently at 11 months and 6 out of 9 patients are still alive (9-14 month from enrollment) and receiving additional treatment after coming off study. No SAEs or changes in ipilimumab dose due to indoximod addition were documented. Conclusions: The Phase 1b part has been successfully concluded without significant toxicities. Phase 2 is ongoing. Updated data to be presented. Clinical trial information: NCT02073123.

3074

Poster Session (Board #396), Sun, 8:00 AM-11:30 AM

Patient-derived tumor xenografts in humanized NSG-SGM3 mice: A new immuno-oncology platform. First Author: Li-Chin Yao, The Jackson Laboratory, Sacramento, CA Background: Humanized mice engrafted with tumors enable in vivo investigation of the interactions between the human immune system and human cancer. We have recently found that humanized NOD-scid IL2Rgnull (NSG) mice bearing patient-derived xenografts (PDX) allow efficacy studies of check-point inhibitors. Next generation NSG strains include triple transgenic NSG mice expressing human cytokines KITLG, CSF2, and IL-3 (NSG-SGM3). Methods: We provide a direct comparison of check-point inhibitors evaluation in NSG and NSG-SGM3 mice engrafted with CD34+ human hematopoietic progenitor cells (HPCs) from the same donor and implanted with PDX tumors. PDX tumors were engrafted into partially HLAmatched hu-NSG-SGM3 mice at 9 weeks post engraftment. Tumorinfiltrating myeloid cells were examined. Two PDX models and one cancer cell line with high PDL1 levels were used to test respond to anti-PD1 therapy in hu-NSG-SGM3 mice. Results: Reconstitution of human immune system in the blood was faster and more robust in NSG-SGM3 compared to NSG recipients throughout the course of the study (18 weeks). Human CD45 + cells reached 25% of total blood cells at week 4 in hu-NSG-SGM3 mice and at week 9 in hu-NSG mice. A majority of blood hCD45+ cells (55%) in huNSG-SGM3 at week 4 were CD33+ myeloid cells. By comparison, efficiency of human CD33+ cells (15%) in the circulation was poor in hu-NSG mice. Moreover, circulating hCD3+ T cells reached 10% at week 9 and included regulatory T cells (Tregs). Hu-NSG mice displayed comparable hCD3+ T cells in the blood only at 12-15 weeks and did not contain Tregs. Tumorinfiltrating myeloid cells will be discussed. High PD-1 expression by CD8 + T cells and Tregs were present within the breast cancer microenvironment suggesting anergy and immunosuppression in hu-NSG-SGM3 mice. Treatment with the anti-PD-1 receptor antibody pembrolizumab significantly reduced tumor growth and the PD-1 expression level on lymphocytes. Conclusions: PDX-bearing hu-NSG-SGM3 mice might serve as a new and improved platform for preclinical immuno-oncology efficacy studies.

3076

Poster Session (Board #398), Sun, 8:00 AM-11:30 AM

Discovery of novel, PD-L1-independent immune checkpoints using JEDI GFP-specific CD8 T cells. First Author: Ranjan Upadhyay, Icahn School of Medicine at Mount Sinai, New York, NY Background: Checkpoint blockade has emerged as an effective strategy for enhancing the endogenous T cell response against tumor-associated antigens. Its clinical benefit is maximized when combined with other immunomodulators or therapeutic maneuvers that induce immunogenic cell death. In order to find additional targetable suppressors of T cell activity, we have created an easily manipulable model of cancer immunoediting. Use of previous transgenic CD8 T cell models (e.g. OT-1) are confounded by difficulty in assessing tumor antigen (e.g. OVA) levels independent of MHC expression. Methods: Balb/c-derived A20 lymphoma cells expressing EGFP (A20-GFP-R0) were cultured with JEDI T cells that recognize the EGFP200d 208 peptide epitope presented on H-2K (Agudo et al., Nat Biotech 2015). This novel system permitted the flow sorting of rare ( , 0.1%) surviving cells based on antigen and MHC class I expression.Five iterations of this T-cell selective pressure allowed for the enrichment of JEDI-resistant clones (A20GFP-R5) that exclude antigen-loss and antigen presentation-deficient variants, two common mechanisms of immune escape. Results: A20-GFP-R5 exhibit marked (~100-fold) resistance to killing by JEDI T cells, despite atleast-equivalent levels of MHC-I and GFP expression compared to A20-GFPR0. T cell activation, as measured by surface expression of CD25 and CD69 as well as proliferation, is nearly identical in response to either R0 or R5, suggesting that there is no impairment of the TCR-peptide-MHC complex. However, we observe a severe deficiency in the production of effector cytokines TNFa and INFg in T cells cultured with R5. This effect seems to be independent of the PD-1 signaling pathway, since R5 has similar levels of PD-L1 expression, and treatment with PD-1 blocking antibodies cannot rescue the exhausted T cell phenotype. Conclusions: By naturally selecting for a rare subpopulation of cancer cells that are resistant to killing by CD8 T cells, we have found a PD-L1-independentcheckpoint pathway utilized by cancer to evade the immune response. Whole exome and RNA-seq studies are ongoing in order to identify the resistance mechanism(s) and to assess their use as potential immunotherapy targets.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Immunotherapy 3077

Poster Session (Board #399), Sun, 8:00 AM-11:30 AM

3078

161s

Poster Session (Board #400), Sun, 8:00 AM-11:30 AM

Immune cell activation among lung adenocarcinoma and squamous cell carcinoma intrinsic subtypes and CD274 (PD-L1) expression. First Author: Hawazin Faruki, GeneCentric, Durham, NC

Association between DNA level aberrations and immune cell infiltration in breast cancer. First Author: Anton Mikhailovich Safonov, Yale School of Medicine, New Haven, CT

Background: Gene expression based subtyping in lung Adenocarcinoma (AD) and lung Squamous Cell Carcinoma (SQ) classifies AD and SQ tumors into distinct subtypes with variable biologic and clinical features. Methods: Using previously published Bindea et al. immune cell gene signatures (24 in total) and the TCGA lung cancer gene expression datasets (AD n = 515, and SQ n = 501), we examined immune cell features in relation to AD and SQ subtypes. Signatures of both Innate Immune Cells (IIC) and Adaptive Immune Cells (AIC), as well as single immune-biomarkers (CTLA4, PDCD1, and CD274(PD-L1)) were examined in the 3 AD subtypes (Terminal Respiratory Unit (TRU), Proximal Proliferative (PP), and Proximal Inflammatory (PI)) and the 4 SQ subtypes (Primitive, Classical, Secretory, Basal). AIC signature associations with tumor subtype and with CD274 expression were evaluated using linear regression. Hierarchical clustering of immune signatures and individual gene correlations were also analyzed. Survival-signature associations were evaluated with cox proportional hazard models. Results: Hierarchical clustering of immune cell gene signatures provided separation of AD and SQ intrinsic subtypes. Immune activation was most prominent in the SQ Secretory subtype and in the AD PI subtype suggesting activation of both innate and adaptive immune cells. In contrast, the Classical subtype of SQ and the PP subtype of AD demonstrated lower immune activation. AD subtype and CD274 expression were similarly predictive of AIC features, however, among the SQ tumors, subtype associations were stronger (median F-test p-value and adjusted R-squared were 2.16e-24 and 0.20 for subtype versus 1.86e-09 and 0.07 for CD274). Survival analysis suggested Th17 and Tfh cells predicted improved survival in AD (p , 0.001). In SQ, only the Primitive subtype demonstrated significant immune cell expression associations with improved survival (p , 0.01). Conclusions: Intrinsic biologic subtypes of lung AD and SQ reveal key differences in immune cell activation, which were not always correlated with CD274 expression. Evaluation of subtypes as potential biomarkers for immunotherapy response should be investigated.

Background: What causes high versus low, or absent, immune cell infiltration in breast cancer is unknown. The goals of this analysis were to examine if total mutation load, neoantigen load, copy number variations (CNV), gene-level or pathway level somatic mutations or germline polymorphisms (SNP) are associated with the level of immune infiltration measured by immune metagene expression levels. Methods:: We used RNASeq, DNA copy number, mutation and germline SNP data from the TCGA representing 627 ER+, 207 HER2+ and 191 TNBC cancers. 13 published immune metagenes were used in correlation and multivariate linear regression analyses performed separately for the 3 major clinical subtypes. Pvalues were adjusted for multiple comparisons and permutation testing was used to assess false discovery rates. Results: Overall mutation, neoantigen and amplification, or deletion loads did not correlate strongly with any of the immune metagenes in any subtype (Spearman coefficient 0.2). In ER+ cancers, mutations in MAP2K4 and TP53 were associated with lower and higher levels of immune infiltration, respectively. In TNBC, mutations in MYH9 and HERC2 were associated with lower immune infiltration. None were found in HER2+ cancers. Three SNPs (rs425757, rs410232, rs470797) in the exonic regions of the FHPR1 and MLP genes were associated with low immune infiltration in ER+ cancers, none in the other subtypes. Two amplicons in TNBC and 3 amplicons in HER2+ cancers were associated with lower immune infiltration. We also identified alterations in several biological pathways that were associated with immune infiltration in different breast cancer subtypes. At the individual patient level, each pathway was affected at different genes and through distinct genomic mechanisms. Conclusions: Immune infiltration in breast cancer is not driven by a single global metric of genomic aberrations such as mutation, neoantigen or CNV loads, or by a few recurrent aberrations but by multiple different gene and pathway level associations that each affect small subsets of patients within each subtype.

3079

3080

Poster Session (Board #401), Sun, 8:00 AM-11:30 AM

First in human (FIH) study of an OX40 agonist monoclonal antibody (mAb) PF-04518600 (PF-8600) in adult patients (pts) with select advanced solid tumors: Preliminary safety and pharmacokinetic (PK)/pharmacodynamic results. First Author: Omid Hamid, The Angeles Clinic and Research Institute, Los Angeles, CA Background: Stimulating effector T cells is an attractive anti-cancer therapeutic strategy. PF-8600 is a novel fully human IgG2 agonistic mAb specific for human OX40, a tumor necrosis factor receptor expressed primarily on activated T cells. PF-8600 provides an immunotherapeutic approach distinct to PD-1 and CTLA-4. A FIH phase 1 study of PF-8600 to investigate the safety, maximum-tolerated dose, PK, immunomodulation and preliminary antitumor activity is on-going in pts with refractory melanoma, head and neck squamous cell, renal cell and hepatocellular carcinoma. Methods: PF-8600 was administrated intravenously every 14 days at doses 0.01 - 0.3 mg/kg. 28-day dose-limiting toxicity (DLT) was evaluated using a modified Toxicity Probability Interval design. PF-8600 PK profiling was completed during cycles 1 and 3. Starting at 0.1 mg/kg, each dose level is expanded to 10 patients for paired tumor biopsy assessments. Free and total OX40 receptor and markers of T cell proliferation (Ki67) were measured on CD4 and CD8 na¨ıve, central and effector memory cells. Results: As of 02 Nov 2015, 9 pts were enrolled in the dose-escalation phase of the study: 0.01 mg/kg (2pts), 0.1 mg/kg (3 pts) and 0.3 mg/kg (4pts). No DLT, drugrelated serious Adverse Events (AEs), immune related AEs or drug-related grade 3-5 AEs were observed. The most frequent drug-related AE was grade 2 fatigue (33.3%). 4 patients experienced best ORR of stable disease (RECIST), and 2 previously immunotherapy-treated patients have been on study for . 6 months. In the periphery, PF-8600 exhibited full target receptor occupancy and enhanced memory T cell proliferation at 0.3 mg/kg. Exposure (AUCt ) increased with increasing doses during cycle 1. Conclusions: These preliminary results demonstrate that PF-8600, an agonistic OX40 mAb, is safe up to 0.3 mg/kg, and biomarker endpoints are met. The highest planned dose cohorts (1.5 and 3 mg/kg) are enrolling and will be presented. Clinical trial information: NCT02315066.

Poster Session (Board #402), Sun, 8:00 AM-11:30 AM

A phase 2 study of NEO-102 (ensituximab), a novel chimeric monoclonal antibody, in adult patients (pts) with unresectable, metastatic colorectal cancer (mCRC). First Author: Richard D. Kim, H. Lee Moffit Cancer Center, Tampa, FL Background: NEO-102 is an investigational, novel, chimeric monoclonal IgG1 antibody raised against an immunogenic membrane fraction of an allogeneic CRC vaccine. It’s target has sequence homology to MUC5AC, but is preferentially expressed in pancreatic cancer and CRC. NEO-102 demonstrates exquisite specificity to CRC, with a mechanism of action through antibody-dependent cellular cytotoxicity (ADCC). A phase 1/2 study was conducted to evaluate NEO-102 monotherapy in pts with refractory mCRC. Here we describe results from the phase 2 part of the study. Methods: This is a single-arm, open-label, multi-center trial. Eligibility includes age $ 18 years, $ 2 lines of prior standard chemotherapy, measureable disease. An IHC-based companion diagnostic assay was used to select pts with NEO-102 expression in . 20% of tumor cells. NEO-102 at 3 mg/kg was administered IV every 2 weeks until unacceptable toxicity or disease progression. Primary endpoint was overall survival (OS). The minimum sample size was 43 pts, assuming that treatment with NEO-102 will improve OS by 40% (7.0 months [mo]) vs a historical control of 5 mo (one-sided significance level of 10% and 80% power). Additional endpoints were response rate (by RECIST 1.1) and analysis of pt PBMCs for ADCC and immune cytokine profiling. Results: A total of 53 pts were enrolled and were evaluable for safety, 48 pts were evaluable for OS, and 37 were evaluable for response. 26 pts were male, 32 were white, and mean number of prior therapies was 5 (range 2‒9). 15 pts (31%) remain alive as of 31 Dec 2015 with an ongoing median OS of 6.8 mo (range 1.8‒25+ mo). 18 pts (48%) had stable disease in target lesions at the end of the first course (day 57). 38 pts received $ 4 doses (max 15+) of NEO102. Grade 3 adverse events (AE) included anemia (1.3%), hyperbilirubinemia (0.9%), diarrhea (0.4%), fatigue (0.9%), headache (0.4%), nausea (0.4%), hypoxia (0.4%), and vomiting (0.4%). No grade 4 AEs were reported. Conclusions: In this phase 2 study of NEO-102 monotherapy in heavily pretreated pts with mCRC, OS was longer than the historical control, and NEO-102 was extremely well tolerated. Additional combination trials with NEO-102 are underway. Clinical trial information: NCT01040000.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

162s 3081

Developmental Therapeutics—Immunotherapy Poster Session (Board #403), Sun, 8:00 AM-11:30 AM

3082

Poster Session (Board #404), Sun, 8:00 AM-11:30 AM

Phase I study of gene mediated cytotoxic immunotherapy (GMCI) for patients with malignant pleural effusion (MPE). First Author: Charu Aggarwal, University of Pennsylvania, Philadelphia, PA

Anti-tumor activity of PEGylated human IL-10 (AM0010) in patients with pancreatic or colorectal cancer. First Author: Kyriakos P. Papadopoulos, START Center for Cancer Care, San Antonio, TX

Background: GMCI is a tumor-specific immune-stimulator through local delivery of aglatimagene besadenovec, an adenovirus-based vector expressing the HSV-1 thymidine kinase gene (AdV-tk) followed by administration of an anti-herpetic prodrug. Immunogenic tumor cell death and superantigen-mediated T cell stimulation leads to CD8 T cell dependent protection against tumor challenge in preclinical models. Methods: The primary end-point of this dose escalation trialwas safety of GMCI followed by standard chemotherapy. Eligibility included patients (pts) with MPE with a clinical indication for placement of pleural catheter, age . 18 yrs, ECOG PS 0-1, FEV1 . 40% predicted and adequate end organ function. Intra-pleural (IP) AdV-tk was administered at doses of 1x 1012 vector particles (vp) (Cohort 1); 1x1013 vp (Cohort 2); and 1x1013 vp plus celecoxib (Cohort 3). Three pts were treated per cohort with 10 pts in the expansion phase. Valacyclovir (2 gm PO TID x 14 days) started the day after AdV-tk followed by chemotherapy. Secondary end-points included response rate, progression free survival, overall survival and immune response. Results: From 2013 to 2015, 19 pts were enrolled and completed therapy: median age 68 years (range 42-89), 14 malignant mesothelioma (MM) (9 epithelioid, 3 sarcomatoid, 2 biphasic), 4 non-small cell lung cancer (NSCLC) and 1 breast cancer. All 3 pts in cohort 2 experienced transient cytokine release syndrome (CRS), with fever, nausea and chills (grade 2) plus hypotension in 1 pt. Addition of prophylactic celecoxib in cohort 3 (n = 13) reduced the incidence and severity of CRS (23% grade 1, 23% grade 2). Response according to RECIST was evaluable for 16 pts, one pt with MM had a PR, 12 SD, and 3 PD. Eight pts are alive and in active follow up (5-17 months), 3 pts with NSCLC are on active treatment with PFS 6.8-10.6 months and OS 9.4-11.8 months. Immune studies on pt samples are ongoing, and will be presented at the meeting. Conclusions: GMCI can be safely administered at high-doses IP followed by systemic chemotherapy. CRS symptoms due to immune stimulation after IP AdV-tk can be diminished with celecoxib prophylaxis. Phase II studies are warranted to further determine efficacy. Clinical trial information: NCT01997190.

Background: IL-10 induces activation of STAT3 in CD8 T cells leads to increased survival, proliferation and cytotoxicity of CD8 T cells. In preclinical studies, PEG-IL-10 induces CD8 mediated tumor rejection and synergizes with cytotoxic chemotherapies. Anti-tumor activity of AM0010 alone was established in the ongoing phase 1 basket trial. Objective responses were observed in pts with renal cell cancer (RCC; 4 of 15pts), uveal melanoma and cutaneous T cell lymphoma. Methods: Pts with advanced pancreatic cancer (PDAC) were treated daily with AM0010 alone (n = 24) or in combination with FOLFOX (5), capecitabine (5) or gemcitabine/nab-paclitaxel (6). Pts with colorectal cancer (n = 16) were treated with AM0010 alone. Tumor responses were monitored following irRC. Immune responses were monitored through analysis of serum cytokines, activation of blood derived T cells, immunosequencing for peripheral T cell clonality and immunohistochemistry for the infiltration of tumors by CD8 T cells. Results: In 24 PDAC pts and 16 CRC pts treated with AM0010 monotherapy (1-20 mg/kg), G3/4 TrAEs were observed in 16 pts, including anemia (5), thrombocytopenia (9), fatigue (2) rash (3) and transaminitis (1). Most TrAEs were transient, only one patient discontinued treatment due to anemia. Colitis, pneumonitis or endocrine abnormalities were not observed. In 16 pts with PDAC treated with AM0010 plus chemotherapy, TrAEs were observed in 11 pts, including thrombocytopenia (10). At 8 weeks, 8 of 17 pts with PDAC and in 4 of 15 pts with CRC on AM0010 alone had stable disease (SD). 11 of 15 PDAC pts had a reduction of CA 19-9. 4 patients (3 PDAC, 1 CRC) had prolonged PFS . 6 months. The median overall survival of PDAC was 5.1 months (n = 20), and CRC was 15.4 months (n = 13; observation . 15mo). 13 of 15 PDAC pts treated with AM0010 and chemotherapy had SD, 2 had a PR. AM0010 increased PD-1+ activated CD8 T cells and the de-novo oligoclonal expansion of T cell clones in the blood. AM0010 increased tumor infiltrating Phospho STAT3+, Granzyme+ and PD1+ CD8+ T cells in tumors. Conclusions: The preliminary clinical activity and the mechanistic data indicating enhanced immune stimulation encourages continued exploration of AM0010 in “immune resistant” cancers. Clinical trial information: NCT02009449.

3083

3084

Poster Session (Board #405), Sun, 8:00 AM-11:30 AM

AGI-134, a fully synthetic a-Gal-based cancer immunotherapy: Synergy with an anti-PD-1 antibody and pre-clinical pharmacokinetic and toxicity profiles. First Author: Sascha A. Kristian, Agalimmune Ltd., Sandwich, United Kingdom Background: Unlike virtually all non-primate mammals, humans lack a-Gal (Gal-a-1,3-Gal-b-1,4-GlcNAc) epitopes and produce antibodies against a-Gal-positive antigens. Anti-a-Gal (anti-Gal) antibodies are among the most abundant antibodies in humans and responsible for hyper-acute rejection of a-Gal-positive xenotransplants. Studies using rabbit-derived a-Gal glycolipids provided a preclinical proof of principle for a-Gal glycolipids as cancer immunotherapy, inducing a CD8+ T cell response to tumor associated antigens (TAAs) that correlated with protection from distal tumor growth in mice. In the current study we show that a fully synthetic, a-Gal glycolipid-like small molecule, AGI-134, displays potent anti-tumor activity by engaging the cellular and humoral immune system against TAAs. Methods: in vitro and in vivo studies. Results: AGI-134 inserts into the plasma membranes of mouse and human cancer cells in vitro. Anti-Gal IgM and IgG bind to the a-Gal labeled cancer cells and mediate an anti-cancer cell response via complement dependent cytotoxicity, antibody dependent cellular cytotoxicity, and cancer cell phagocytosis by antigen presenting cells in vitro. In anti-Gal producing a-1,3-galactosyltransferase knockout (GT KO) mice, intratumorally (IT) administered AGI-134 greatly inhibits the growth of untreated distal B16-F10 melanoma lesions in an a-Gal- and anti-Galdependent fashion. Moreover, we show evidence that AGI-134 acts synergistically with an anti-programmed cell death 1 receptor (PD-1) antibody. In a PK study in GT KO mice, AGI-134 shows low systemic exposure following IT and subcutaneous (SC) administration. In a non-GLP toxicity study in cynomolgus monkeys, 5 SC or intravenous AGI-134 doses ranging from 250 mg/kg/dose were well tolerated. The no-observed-adverse-effect level was considered to be 50 mg/kg SC. Conclusions: We propose that AGI-134 has potential as a new immunotherapy for the treatment of solid tumors. The data suggest that AGI-134 may be an excellent combination partner for checkpoint inhibitors in treating cancer. Currently, the drug is being prepared for a Phase I/II clinical trial.

Poster Session (Board #406), Sun, 8:00 AM-11:30 AM

Retrospective cohort analysis of adjuvant NY-ESO-1 vaccines in stage III melanoma. First Author: Michael William Lattanzi, New York University School of Medicine, New York, NY Background: Cancer-testis antigen NY-ESO-1 is a highly immunogenic melanoma antigen which has been adopted for adjuvant vaccine trials. Three such trials (NCT00124124, NCT00821652, NCT01079741) in high-risk melanoma patients were conducted at our center. Although exact treatment regimens differed between trials, all enrolled patients received an NY-ESO1-based peptide vaccine. Methods: All vaccine trial patients had histologically-proven melanoma and were without evidence of disease at trial entry; every stage III patient for whom records were available was included for analysis. For comparison, a control cohort of stage III patients who received no adjuvant therapy was identified via our clinicopathological database; every such patient in the database was included for analysis. Recurrence patterns were compared via a chi-squared test. Overall survival (OS) was estimated by Kaplan-Meier analysis, and a Cox proportional hazard model was employed to control for thickness and ulceration, and calculate a hazard ratio (HR). Results: Information was available for 62 vaccinated patients and 116 control patients (median follow-up: 62.8 and 30.5 months, respectively). The vaccine and control groups had the following characteristics, respectively: male (52% vs 62%), median thickness (2.0 vs 2.7), ulceration (41% vs 47%), positive lymph nodes (81% vs 93%), and nodular histotype (52% vs 56%); only lymph node status was significantly different (p = 0.02). At last follow-up, 35 patients (56%) in the treatment group had recurred, compared to 84 patients (64%) in the control group. At first recurrence, vaccinated patients tended to recur with resectable disease more often than control patients (79% vs 60%, p = 0.10), though this result was not significant. OS, however, was significantly longer in the vaccine group compared to the control group (median not reached vs 57.5 months, HR 0.58, 95%CI 0.34 – 0.99, p = 0.047). Conclusions: In this small retrospective cohort of stage III melanoma patients, adjuvant NY-ESO-1 vaccination was associated with improved OS. NY-ESO-1 vaccination may exert an immunologic advantage reflected in a pattern of locally resectable disease upon recurrence compared to diffusely metastatic recurrences.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Immunotherapy 3085

Poster Session (Board #407), Sun, 8:00 AM-11:30 AM

Randomized phase II study of WT1 peptide vaccine plus gemcitabine for advanced pancreatic ductal adenocarcinoma (PDAC): Clinical efficacy and immune response. First Author: Sumiyuki Nishida, Dept. of Respiratory Medicine, Allergy and Rheumatic Disease, Osaka University Graduate School of Medicine, Suita, Japan Background: Wilms’ tumor gene (WT1) is overexpressed in almost all malignancies including PDAC, and is supposed to be most promising tumor-associated antigen in cancer immunotherapies. To investigate superiority of WT1 vaccine in combination with gemcitabine (GEMWT1) compared to GEM alone (GEM) for patients (pts) with advanced PDAC, we designed the randomized phase II trial. Methods: HLA-A*02:01 and/or A*24: 02-positive pts with locally advanced or metastatic PDAC without prior treatment, or recurrence after surgery were enrolled. Pts were randomly assigned at a ratio of 1:1 to GEMWT1 or GEM. Pts in GEM were allowed to receive GEM + WT1 vaccine after disease progression. The primary end point was overall survival (OS) rate at 1 year. Secondary end points were progression free survival (PFS), WT1-specific immunogenic response, and safety. WT1 vaccine, which was composed of HLA-restricted 9-mer WT1 peptide (3mg/ body) and Montanide ISA51 adjuvant, was intradermally injected on days 1 and 15, and GEM was administrated at 1,000 mg/m2 on days 1, 8, and 15 in a 28-day cycle until disease progression. Results: A total of 91 pts were enrolled and 85 evaluable finally (GEMWT1, n = 42; GEM, n = 43). GEMWT1 improved 1-yr OS rate and prolonged PFS compared to GEM (Table). These were remarkable in pts with metastatic PDAC. WT1specific immunity was assessed with delayed-typed hypersensitivity (DTH) to WT1 peptide and tetramer assay of WT1-specific cytotoxic lymphocytes (WT1-CTLs). Median PFS was 195, 102, and 100 days for DTH(+), DTH(-), and GEM, respectively [HR (90% CI), 0.51 (0.31 to 0.85), 0.80 (0.50 to 1.28), and 1]. WT1-CTLs increased in almost all DTH(+) pts. Conclusions: GEM + WT1 vaccine prolonged PFS in pts with advanced PDAC. Clinical efficacy of this combination therapy was associated with the induction of WT1-CTL responses. Clinical trial information: UMIN000005248. Total

1-yr OS, % [90% CI] PFS, median (days) [90% CI] PFS, HR [90% CI]

3087

Metastatic

GEM+WT1 n = 42

GEM n = 43

GEM+WT1 n = 35

GEM n = 34

34.2 [22.4-46.3] 157 [102-193] 0.66 [0.44 – 0.99]

21.5 [12.2-32.6] 100 [57-141] p = 0.089

26.5 [15.1-39.4] 133 [87-166] 0.48 [0.30 – 0.77]

12.2 [4.8-23.3] 76 [54-106] p = 0.008

Poster Session (Board #409), Sun, 8:00 AM-11:30 AM

Adjuvant cancer peptide vaccine for pathological node-positive esophageal SCC patients who underwent preoperative therapy followed by R0 resection. First Author: Takushi Yasuda, Department of Surgery, Kinki University Faculty of Medicine, Osaka-Sayama, Osaka, Japan Background: Three HLA-A-24-restricted epitope peptides have been identified, which were derived from three cancer-testis antigens that are expressed only in esophageal squamous cell carcinoma (ESCC) cell lines: up-regulated lung cancer 10 (URLC10), cell division cycle associated 1 (CDCA1) and KH domain-containing protein overexpressed in cancer 1 (KOC1). This was an exploratory, prospective, phase II study in a single institute to elucidate the efficacy of an adjuvant vaccine using these three kinds of epitope peptides. Methods: Patients with ESCC who underwent R0 resection after preoperative chemotherapy or chemoradiotherapy (CRT) from December 2009 to September 2014 and had pathological positive nodes were enrolled. A peptide vaccine (each with 1 mg of epitope peptides and 1 ml of Montanide ISA 51 VG) were administered 10 times weekly and then 10 times biweekly only in patients with HLA-A2402. No other adjuvant therapy was given until recurrence occurred. We evaluated relapse-free survival (RFS) as the primary endpoint and overall survival (OS) as the secondary endpoint. Results: Thirty-four patients were enrolled in the vaccine group (VG) and 30 were in the control group (CG). There was no significant difference in sex, age, tumor location and pTNM stage between the two groups. The median duration of follow-up of censored patients was 42 months (range, 11–72). Recurrence rate was 17/34 (50.0%) in the VG, 4 of whom are now cancer-free after surgery or CRT, resulting in 13 (38.2%) with recurrence under treatment. On the other hand, recurrence was observed in 19/30 (63.3%) in the CG, all of whom already died or are under treatment. Five-year RFS rate of the VG vs. the CG was 44.7% vs. 33.9% (p = 0.212) and 5-year OS rates were 62.3% vs. 31.0% (p = 0.027), respectively. Patients with pN2-3 in the VG and CG had a 5-year RFS rate of 30.3% vs. 15.6% (p = 0.152) and OS rate of 48.8% vs. 19.9% (p = 0.027). Conclusions: Our study findings suggest that the cancer peptide vaccine suppresses postoperative recurrence and improves the survival of patients with ESCC. A randomized, multicenter, phase III study is underway to clarify the efficacy of these peptide vaccines as adjuvant therapy for ESCC. Clinical trial information: UMIN000003557.

3086

163s

Poster Session (Board #408), Sun, 8:00 AM-11:30 AM

In situ, therapeutic vaccination against refractory solid cancers with intratumoral Poly-ICLC: A phase I study. First Author: Chrisann Kyi, Tisch Cancer Center, Mount Sinai School of Medicine, New York, NY Background: Poly-ICLC, a double-stranded RNA complex, can directly activate dendritic cells and trigger NK cells to kill tumor cells. It can be given intramuscularly (IM) to induce systemic inflammation and intratumorally (IT) to induce immune infiltration of tumors. Methods: Patients (7 head and neck cancer (HNSCC), 1 melanoma) with recurrent metastatic disease who had failed prior systemic therapy were enrolled in the study. Patients received 2 cycles of treatment with each cycle consisting of 1mg Poly-ICLC 3x weekly IT for 2 weeks followed by IM boosters biweekly for 6 weeks with 2 week rest period. The 10-week cycle was repeated, followed by a 6-week no-treatment period. Evaluation of immune response was performed by immunohistochemistry (IHC) and RNA Sequencing (RNASeq) on tumor biopsies and blood collected throughout the study. Results: Two subjects completed 2 cycles of treatment; 6 subjects completed 1 cycle or less due to progressive disease. One patient achieved clinical benefit (stable disease, PFS 41 weeks) while the remaining patients had progressive disease. PolyICLC was well tolerated with principal side effects, fatigue and inflammation at injection site ( , grade 2). One case of grade 3 tumor necrosis was observed. In the patient with clinical benefit, IHC analysis of tumor showed increased CD4(60x), CD8(10x), PD1(20x) and PDL1(3x) compared to patients with progressive disease whose tumor biopsies showed unchanged/ decreased CD4, CD8, PD1, and PDL1 levels over treatment period. Furthermore, RNASeq analysis of the same patient’s tumor and PBMC showed dramatic changes such as upregulation of interferon-stimulated genes, chemokines, and genes associated with T cell activation and antigen presentation indicative of local and systemic immune activation in response to Poly-ICLC treatment. Conclusions: Poly-ICLC was well tolerated in advanced solid cancer patients, and generated local immune response in tumor microenvironment and systemic immune response as evident in the patient achieving clinical benefit. These results warrant further investigation, and are currently being explored in a larger multicenter adaptive phase II clinical trial. Clinical trial information: NCT01984892.

3089

Poster Session (Board #411), Sun, 8:00 AM-11:30 AM

Initial phase I/IIa trial results of an autologous tumor lysate, particle-loaded, dendritic cell (TLPLDC) vaccine in patients with solid tumors. First Author: Garth S Herbert, San Antonio Military Medical Center, San Antonio, TX Background: Tumor vaccines use various strategies to generate immune responses, commonly targeting specific and purported tumor-associated antigens instead of antigens unique to a patient’s tumor. We developed a process by which dendritic cells (DC) are exposed to a patient’s particular tumor antigens, creating a patient-specific DC vaccine in 48 hours. Here we describe initial results with the tumor lysate, particle-loaded, dendritic cell (TLPLDC) vaccine. Methods: This open label phase I/IIa trial includes patients with solid tumors of any stage, ECOG , 1, and . 4 months of life expectancy. Using as little as 1mg of tumor and 120ml blood from which DC are isolated, a personalized vaccine is created using a particle loading system. The primary vaccination series consists of four vaccines administered intradermally 1 month apart, and is followed by optional boosters every 3 months. Patients are followed per standard of care, and imaging results are centrally evaluated. Endpoints include vaccine safety and tumor response according to RECIST criteria. Results: To date 19 pts have been enrolled with 12 different tumor types. The median age is 57 (range 20 - 74). 18/19 had advanced-stage disease (stage III/IV). Median follow-up was 6.4 months (range 2.0 - 20.2). Most adverse events were mild, with no toxicities . grade 2. Of 17 patients with residual and measurable disease, 10 (58.8%) demonstrated clinical benefit consisting of 3 complete responses, 3 partial responses, and 4 with stable disease. Two patients treated in the adjuvant setting remain disease-free after 9 and 20 months. Conclusions: The TLPLDC vaccine is a scalable technology that generates a personalized DC vaccine in 48 hours, and requires less autologous tumor tissue and fewer DCs than previous DC vaccines. The vaccine is safe, with primarily grade 0/1 toxicities. Furthermore, it resulted in a 35% objective response rate in a variety of tumor histologies, warranting further study in larger trials.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

164s 3090

Developmental Therapeutics—Immunotherapy Poster Session (Board #412), Sun, 8:00 AM-11:30 AM

5-year follow up for 72 metastatic melanoma patients treated with eltrapuldencel-T (CLBS20) vaccine. First Author: Robert O. Dillman, CalCaladrius Biosciences, Inc., Irvine, CA Background: CLBS20 is a patient-specific therapeutic vaccine consisting of autologous dendritic cells loaded with antigens from irradiated cells from an autologous tumor cell line. It potentially presents the entire repertoire of unique antigens resulting from nonsynonymous mutations in a patient’s selfrenewing melanoma cells. In previously reported single arm and randomized phase 2 trials, CLBS was associated with a 72% 2-year overall survival (OS). In the randomized trial median OS for CLBS20-treated patients was more than twice that for a vaccine consisting of irradiated autologous tumor cells. Mild local injection site reactions was the most common toxicity. Methods: During 2001 to 2011 72 patients with stage 4 or recurrent stage 3 melanoma were treated with CLBS20 (Clinicaltrials.gov NCT00948480 & NCT00436930). Each had a metastatic melanoma lesion surgically resected, from which a tumor cell line was established, cells from which were incubated with autologous dendritic cells to produce the vaccine that was injected s.c. in 500 micrograms sargramostim (GM-CSF) weekly x 3 weeks and monthly x 5 months (mos). This report summarizes long-term survival data for all CLBS20-treated patients from date of first injection. Results: There were 45 men and 27 women. Median age is 52 years (17 to 83). Tumor sources were 37-lymph node, 15-soft tissue, 20-viscera. Prior therapies included 72-surgery, 41-chemotherapy, 33-Interleukin-2, 30interferon alpha, 24-GM-CSF, 19-radiation therapy (RT), and 15-RT for brain metastases. All 33 surviving patients were followed 5 years; none were lost to follow up. Median OS is 45.6 mos; 5-year OS is 46%. Median progression free survival (PFS) is 4.4 mos, 5-year PFS is 19%. The 20 patients with recurrent stage 3, have a 70% 5-year OS, median PFS 19.8 mos, and 35% 5-year PFS. For the 52 patients with stage 4, median OS is 41.0 mos, 5-year OS is 36%, and median PFS is 3.8 mos. One patient with refractory progressive measurable disease experienced a delayed complete response that was ongoing at 5 years. Conclusions: CLBS20 is associated with encouraging 5-year overall survival. Because of its unique mechanism of action and absence of toxicity, it should be evaluated further in similar patients. Clinical trial information: NCT00948480. Clinical trial information: NCT00436930.

3092

Poster Session (Board #414), Sun, 8:00 AM-11:30 AM

Systemic treatment with anti-PD-1 antibody nivolumab in combination with vaccine therapy in advanced pancreatic cancer. First Author: Jan Nesselhut, Praxisgemeinschaft fur ¨ Zelltherapie, Duderstadt, Germany Background: Systemic treatment with antibodies against PD-1 has shown promising results in many solid tumors with pancreatic cancer being one of the few exceptions. This may be due to failure of activation and/or recruitment of effector T cells into the tumor. An effective immune response against tumors requires the inhibition of inhibitory signals as well as the activation of antigen specific T-cell response. Thus, therapy efficacy may be improved by combination of anti-PD-1 therapy and dendritic cell vaccines. Methods: Seven patients with stage IV pancreatic cancer included in this pilot study received a treatment with nivolumab and dendritic cells (DC). After isolating monocytes from peripheral blood of the patients, antigen primed MoDC were generated using standard protocols. Nivolumab was given one day before the DC vaccine at a reduced dose of 1-2 mg/kg body weight. Cytokine release of DC and T-cell activity in relation to nivolumab therapy were measured using a standard mixed lymphocyte culture (MLC). Results: Using Recist criteria we observed 2 partial remissions with overall survival after onset of therapy of 13 months and 5 months respectively (at time of abstract submission). Both patients are still alive with an ongoing therapy response. Most patients tolerate the therapy well with only mild side effects upon nivolumab therapy. Cytokine measurement using MLC, showed a change in the cytokine release when nivolumab was added to the culture. Conclusions: At ASCO 2015 we demonstrated that the efficacy of DC based therapy can be improved by blockade of PD-L1 on dendritic cells improving the T-cell specific response. Here we show that systemic anti-PD1 therapy for patients with pancreatic cancer can be effective even at lower dose when combined with DC vaccine therapy.

3091

Poster Session (Board #413), Sun, 8:00 AM-11:30 AM

A phase 1 trial extension to assess immunologic efficacy and safety of primeboost vaccination with VXM01, an oral T cell vaccine against VEGF-receptor 2, in patients with advanced pancreatic cancer. First Author: Friedrich Hubertus Schmitz-Winnenthal, General, Visceral and Transplantation Surgery, University Clinics of Heidelberg, Germany, Heidelberg, Germany Background: VXM01 is an orally applied tumor vaccine based on live, attenuated salmonella bacteria carrying a eukaryotic expression plasmid, which encodes VEGFR-2. A recent randomized and placebo controlled phase I trial demonstrated in advanced pancreatic cancer patients safety, immunogenicity and T cell response related transient anti-angiogenic activity of 4 priming vaccinations applied within one week. As VEGFR2 specific T cell responses gradually declined after a peak response at day 21, an extension trial was conducted to explore whether monthly boost vaccinations can be safely administered and maintain increased vaccine specific T cell levels. Methods: 18 patients with advanced pancreatic cancer received a priming regimen with VXM01 followed by 6 monthly boost vaccinations starting on day 38 and 8 patients received placebo treatment in a randomized and blinded fashion. Immunomonitoring involved IFN-gamma Elispot analysis against VEGFR2 derived long peptides and was performed on blood samples harvested on days 0, 4, 14, 21 and 38 and 10 days after boost vaccination. Results: 12 out of 18 vaccinated patients showed considerable, in some cases high, increases in VEGFR2 specific T cell response over d0 levels. In the entire vaccinated group, VEGFR2 specific T cells peaked after 3 months with an average 4-fold increase over d0 levels. At study completion after 6 months VEGFR2 specific T cells were still increased over d0 values. Drug-related adverse events preferentially associated with boosting doses of VXM01 were decreases in lymphocyte count (22% vs 0) and increases in diarrhea (22% vs 0). Median overall survival was 9.3 months in the vaccinated group and 8.4 months in the placebo group. Within the vaccinated group, patients with a T-cell response to VXM01 showed a longer median overall survival (10.3 months) than patients without (5.4 months). Conclusions: Oral prime boost vaccination with VXM01 was well tolerated and is suitable to induce and maintain for several months effector T cell responses against VEGFR-2 in advanced stage pancreatic cancer patients. Clinical trial information: EudraCT 2011-000222-29.

3093

Poster Session (Board #415), Sun, 8:00 AM-11:30 AM

Single-agent LV305 to induce anti-tumor immune and clinical responses in patients with advanced or metastatic sarcoma and other cancers expressing NY-ESO-1. First Author: Neeta Somaiah, The University of Texas MD Anderson Cancer Center, Houston, TX Background: Immune approaches that stimulate specific anti-tumor cytotoxic T cells (CTLs) can have single agent activity, be synergistic with check point inhibitors (CPIs) and enhance adoptively transferred T cells. LV305 is a replication-incompetent, integration-deficient, hybrid viral vector based on the ZVex platform. It is designed to target dendritic cells in vivo and induce the expression of NY-ESO-1 in order to potently generate and expand anticancer CTLs. In this first-in-human study, the safety, immunogenicity and efficacy of LV305 were examined in patients (pts) with sarcoma or carcinoma. Methods: Pts with previously treated, advanced or metastatic malignancies expressing NY-ESO-1 (IHC) were eligible. In Part 1, Dose Escalation, 3 pts enrolled into each of 4 cohorts to receive 3 or 4 intradermal injections every 3 weeks of 108, 109 or 1010 vector genomes (vg) per dose. Cohort advancement followed a 3+3 design and safety was reviewed with a DSMB. In Part 2, additional pts were treated at the max dose/MTD. Results: In Part 1, no DLT or SAEs were observed in the 12 sarcoma patients; all related AEs were grade 1/2. In Part 2 (1010 vg dose), only 1 grade 3 AE was reported in 27 pts (12 sarcoma, 5 melanoma, 9 ovarian and 1 NSCLC). Of evaluable sarcoma pts, NY-ESO-1 specific CD4 12/23 (52%), CD8 12/23 (52%) and CD4 and/or CD8 16/23 (70%) were induced by LV305. Clinical responses included: sarcoma, 1 PR (recent SD to PR change {Study Day 579}, 606+ days), 13 SD (defined as . 84 days, range 136 – 585+ days); melanoma 4 SD; ovarian, 2 SD; and NSCLC 1 SD. In sarcoma pts, median PFS was 140 days and 7/10 pts with pretreatment PD had SD or PR following LV305. Conclusions: LV305 demonstrated acceptable safety up to 1010 vg and induced significant anti-NY-ESO-1 T cell responses in advanced malignancies. Clinical benefit (PR/SD) was seen in 14/24 pts with sarcoma, have been durable . 1 year in 5 pts and includes a late conversion of SD to PR. The median PFS for sarcomas was similar to that of trabectedin as reported in a recent US Ph3 study. The induction of immune responses, favorable safety profile and clinical activity indicate LV305 is an active agent that warrants further investigation. Clinical trial information: NCT02122861.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Immunotherapy 3094

Poster Session (Board #416), Sun, 8:00 AM-11:30 AM

Effect of cancer vaccine formulation on synergy with anti-CTLA-4 and antiPD-L1 checkpoint blockade therapy of cancer. First Author: Yared Hailemichael, The University of Texas MD Anderson Cancer Center, Houston, TX Background: A promising avenue to increase the efficacy of anti-CTLA-4 and anti-PD-(L)1 checkpoint blockade therapies is to enhance T cell induction through vaccination. Surprisingly, concurrent gp100 peptide vaccination decreased clinical efficacy of anti-CTLA-4 in patients with melanoma. As a result, it is currently unclear how to effectively combine checkpoint blockade therapy with vaccination. We used an animal model of gp100 peptide vaccination and CTLA-4/PD-L1 checkpoint blockade to understand this phenomenon and define parameters for successful combination of checkpoint blockade and vaccination. Methods: We treated established B16 melanoma tumors with anti-CTLA-4 therapy and concurrent vaccination with gp100/IFA and studied T cell expansion, trafficking and therapeutic effector function. Results: Anti-CTLA-4 monotherapy significantly increased intratumoral levels of TRP-2 melanoma antigen-specific Teffs, as well as total CD44hiCD11ahiCD8lo Teffs at the tumor site. Conversely, gp100/IFA vaccination-induced, gp100-specific CD8+ T cells accumulated at the inflamed vaccination site. Surprisingly, combination of gp100/IFA vaccination and anti-CTLA-4 therapy caused non-gp100-specific Teff, induced by anti-CTLA-4 therapy, to also accumulate at the vaccination site and not at the tumor site. This Teff sequestration at the vaccination site was dependent on IFN-g, CXCR3 and ICAM-1. Replacing the gp100/IFA formulation with a gp100-encoding viral vector or a non-persistent, water-based formulation potently synergized with anti-CTLA-4 or anti-PD-L1 resulting in a high proportion of complete cure. Conclusions: Persistent vaccine formulations can fail to synergize, or even diminish the efficacy of, CTLA-4/PD-(L)1 checkpoint–based cancer immune therapy through the induction of aberrant T cell trafficking to the vaccination site. A non-persistent vaccine formulation can reverse these undesirable effects and potently synergize with anti-CTLA-4 and/or anti-PD-(L)1 checkpoint blockade, resulting in significantly improved anti-tumor activity.

TPS3096

Poster Session (Board #417b), Sun, 8:00 AM-11:30 AM

A phase 1 study to evaluate the safety, tolerability, pharmacokinetics, immunogenicity, and antitumor activity of MEDI9447 alone and in combination with durvalumab (MEDI4736) in patients with advanced solid tumors. First Author: Johanna C. Bendell, Sarah Cannon Research Institute, Nashville, TN Background: The success of immune checkpoint inhibitors in various cancers has triggered the identification of novel approaches to modulate the immune system. CD73, an ectonucleotidase, catalyzes the rate-limiting step for the production of adenosine in the extracellular space. Adenosine may aid tumors in evading immune recognition and destruction. MEDI9447 is a human IgG1l monoclonal antibody (mAb) that selectively binds to and inhibits the ectonucleotidase activity of CD73. Durvalumab (MEDI4736) is a selective, high-affinity human IgG1 mAb that blocks PD-L1 binding to PD-1 (IC50 0.1 nM) and CD80 (B7.1; IC50 0.04 nM). Preclinical data suggests additive activity of MEDI9447 with checkpoint inhibitor antibodies.1 Methods: This is a first-time-in-human Phase 1, multicenter, open-label, dose-escalation and dose-expansion study (NCT02503774) of MEDI9447 administered as a single agent or in combination with durvalumab. Eligible patients include those with selected advanced solid tumors that have progressed, are refractory to, or are intolerant to standard therapy appropriate for the specific tumor type. The primary objectives are to assess safety and tolerability, describe dose-limiting toxicity (DLT), and determine the maximum tolerated dose. The secondary objectives are to evaluate antitumor efficacy, pharmacokinetics, immunogenicity, and biomarker activity in tumor biopsy specimens. Antitumor activity is assessed using immune-related Response Evaluation Criteria in Solid Tumors (irRECIST). Subjects in the monotherapy arm will receive one of four dose levels of MEDI9447; subjects in the combination arm will receive one of four dose levels of MEDI9447 and a fixed dose level of durvalumab. Subjects will receive MEDI9447 as monotherapy or in combination with durvalumab for up to 52 weeks. Dose escalation will follow a 3 + 3 design. Recruitment is ongoing. 1. Cancer Res 2015;75(15 Suppl):Abstract nr 285. doi:10.1158/1538-7445.AM2015285 Clinical trial information: NCT02503774.

TPS3095

165s

Poster Session (Board #417a), Sun, 8:00 AM-11:30 AM

A phase 1, first-time-in-human study of MEDI9197, a TLR7/8 agonist, administered intratumorally in subjects with a solid tumor cancer. First Author: Shilpa Gupta, Masonic Cancer Center, University of Minnesota, Minneapolis, MN Background: Toll-like receptors (TLRs) are proteins in the innate immune system that, once bound to their ligands, directly activate innate immune cells, including myeloid dendritic cells, plasmacytoid dendritic cells, monocytes/macrophages, and B cells. This activation may result in immune modulation such as activation of co-stimulatory molecules, production of antitumor and antiviral cytokines, and stimulation of cell-mediated NK and T cell immune responses involved in eradicating tumors. MEDI9197 is a small molecule imidazoquinoline agonist for human TLR7 and TLR8 that is designed to be water-insoluble, a property that confers a long-lasting depot effect. This agent is being developed as a potential anticancer therapy for intra-tumoral injection in patients with solid tumors. Methods: This is a phase 1, multicenter, open-label, dose-escalation, first-time-in-human study to evaluate the TLR7/8 agonist MEDI9197 delivered by intratumoral injection (to minimize systemic exposure and systemic immune activation but still stimulate targeted immune activation within the tumor). Subjects have metastatic/locally advanced solid tumors that have progressed on, are refractory to, or for which there is no standard of care therapy. Subjects must have a minimum of one lesion that is easily accessible (cutaneous or subcutaneous mass that is palpable and/or visualized by ultrasound for local injection). The study uses a 3 + 3 design to investigate a range of dose levels; subjects in each cohort will receive up to 4 doses of MEDI9197. The primary objective is to evaluate the safety and tolerability of MEDI9197 as determined by the primary endpoints of dose-limiting toxicities, adverse events, and laboratory parameters. Secondary objectives are to evaluate pharmacokinetics, changes in intratumoral pharmacodynamic (PD) biomarkers (lymphocyte populations and gene expression profiles), changes in systemic PD biomarkers, and antitumor activity (objective response, duration of response, disease control rate). Response will be determined using Response Evaluation Criteria in Solid Tumors (RECIST). Recruitment is ongoing, with a target enrollment of approximately 45 subjects. Clinical trial information: NCT02556463.

TPS3097

Poster Session (Board #418a), Sun, 8:00 AM-11:30 AM

Phase 1b/2 study of intratumoral Ad-RTS-hIL-12 + veledimex in patients with chemotherapy-responsive locally advanced or metastatic breast cancer. First Author: Heather L. McArthur, Memorial Sloan Kettering Cancer Center, New York, NY Background: Immune-based strategies involving T-cell activation have shown significant activity in multiple tumor types. The presence of immune elements in breast cancers has prognostic and predictive impact. Thus, strategies that optimize the interplay between breast cancer and the effected individual’s immune system may be therapeutic. Interleukin-12 (IL-12), a pro-inflammatory cytokine, reverses immune escape mechanisms induced by myeloid derived suppressor and dendritic cells which, in turn, improves the function of activated CD8+ T cells and promotes tumor stroma collapse. Because tumor neoantigens may be generated in response to chemotherapy, IL12-mediated immune modulation may be optimal in patients with chemotherapy-sensitive metastatic breast cancer. Ad-RTS-hIL-12 (Ad) is a novel gene therapy candidate expressing IL-12 under the control of an orallyadministered activator ligand, veledimex (V) through the proprietary RheoSwitch Therapeutic System (RTS). Methods: Open-label, phase 1b/2, single-arm, single-center study to evaluate the safety and tolerability of Ad+V in women with stable or responsive disease after $ 12-weeks of 1st or 2ndline chemotherapy. Up to 40 patients, including up to 8 HER2+patients will be placed on chemotherapy-holiday and enter the immunotherapy phase, consisting of a single cycle of Ad administered intratumorally (Day 1), along with V (80 mg QDx7). HER2-directed antibody therapy may be continued during the immunotherapy phase for women with HER2- disease. Women $ 18 years with histologically-confirmed locally advanced or metastatic breast cancer who have achieved a partial response (PR) or stable disease (SD) to 1st or 2nd-line chemotherapy are eligible. Secondary endpoints include 12 week overall response rate, disease control rate and impact on exploratory immune biomarkers. Summary: Ad+V is a novel gene therapy which controls local expression of IL-12 and may stimulate an anti-cancer T cell immune response. The ability to regulate the production of IL-12 by modulating V dosing may result in an improved therapeutic index in combination with standard of care. Clinical trial information: NCT02423902.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

166s TPS3098

Developmental Therapeutics—Immunotherapy Poster Session (Board #418b), Sun, 8:00 AM-11:30 AM

TPS3099

Poster Session (Board #419a), Sun, 8:00 AM-11:30 AM

Phase I/II study of mogamulizumab, an anti-CCR4 antibody targeting regulatory T cells in advanced cancer patients. First Author: George Plitas, Memorial Sloan Kettering Cancer Center, New York, NY

A Phase 1 study of MEDI1873 in adult patients with select advanced solid tumors. First Author: Ani Sarkis Balmanoukian, The Angeles Clinic and Research Institute, Los Angeles, CA

Background: Regulatory T cells (Treg) are a subset of CD4 T cells that preserve immune homeostasis by the establishment and maintenance of peripheral tolerance. This suppressive function however, limits anti-tumor immune responses and represents a critical obstacle to immunotherapy. Multiple preclinical models have demonstrated effective anti-tumor immunity by targeting Treg cells. Human tumor-infiltrating Treg cells express high levels of the chemokine receptor CCR4. Based on these findings we developed a Phase I/II clinical trial investigating the activity of Mogamulizumab, a fully humanized, IgG1, monoclonal, depleting, antiCCR4 antibody in patients with advanced solid organ malignancies. Methods: Eligible patients include those with histologically confirmed solid organ malignancies, adequate performance status, and organ function who are not eligible for, have declined, or have failed standard treatment. Up to 24 patients will be enrolled in the phase I study and the dose of Mogamulizumab will be escalated according to a standard 3+3 design. Mogamulizumab will be administered weekly for the first month and then every two weeks for up to a year or until progressive disease or unacceptable toxicity. The primary study end points are safety/tolerability and identification of the maximum tolerated dose (MTD), which will be taken forward in a phase II study. Secondary objectives include evaluating clinical activity as assessed by tumor response (RECIST v1.1 and Immune-related Response Criteria), progression-free survival, and overall survival. Exploratory objectives include determining the extent of tumor and peripheral blood Treg cell depletion as well as the correlative immunologic changes. Tumor biopsies will be available both pre-treatment and during treatment. The first two dose escalation cohorts have been accrued. The phase II study will continue to evaluate the safety and preliminary efficacy of mogamulizumab at the MTD in 48 patients with triple negative breast cancer, gastric adenocarcinoma, and non-small cell lung cancer. Clinical trial information: NCT02281409. Clinical trial information: NCT02281409.

Background: Directly activating T cells by delivering activating signals that augment those received via the T-cell receptor may enhance a patient’s response to tumor antigens. Glucocorticoid-induced tumor necrosis factor receptor family-related protein (GITR) is expressed on regulatory T cells (Tregs) and up-regulated on activated effector T cells where it is involved in regulating T-cell activation. MEDI1873, a GITR agonist, is under investigation in patients with advanced solid tumors. Based on preclinical studies, a GITR agonist is expected to induce an antitumor immune response by promoting effector T-cell activation and overcoming local immunosuppression by reducing the number of intratumoral Tregs1,2. Methods: This is a first-in-human, Phase 1, multicenter, open-label, dose escalation study in adult patients with recurrent or metastatic solid tumors (NCT02583165). The primary objectives are to assess safety, describe dose-limiting toxicities, and determine the maximum tolerated dose of single-agent MEDI1873. The secondary objectives are to evaluate antitumor activity based on RECIST Version 1.1 (includes objective response rate, disease control rate, duration of response, progression-free survival, and overall survival endpoints), pharmacokinetics, and immunogenicity of single-agent MEDI1873. In addition, pharmacodynamic biomarkers will be evaluated, including changes in the number and activation status of circulating and intratumoral lymphocyte populations compared with baseline. Eligible patients ( $ 18 years) will have histologically or cytologically confirmed advanced solid tumors and must have received prior treatment for recurrent or metastatic disease. In this dose escalation study, a modified 3+3 design will be followed. MEDI1873 will be administered until unacceptable toxicity or disease progression occurs. Recruitment for this study is ongoing. References 1. Cohen AD, et al. PLoS One 2010;5:e10436. 2. Coe D, et al. Cancer Immunol Immunother 2010;59:1367-77. Clinical trial information: NCT02583165.

TPS3100

TPS3101

Poster Session (Board #419b), Sun, 8:00 AM-11:30 AM

Phase II trial of umbilical cord blood-derived natural killer cells for multiple myeloma. First Author: Nina Shah, The University of Texas MD Anderson Cancer Center, Houston, TX Background: Multiple myeloma (MM) is an incurable disease characterized by immune dysregulation and exhaustion, whereby proliferation of malignant plasma cells is not checked by the native immune system. Allogeneic natural killer (NK) cells are active in various hematologic malignancies and may have a role against MM. We have recently completed a phase I, first-inhuman trial of NK cells derived from umbilical cord blood (CB-NK cells) in conjunction with high dose chemotherapy and autologous stem cell transplant (SCT). In this high-risk patient group, there was no GVHD or infusion reaction and 10/12 patients achieved VGPR or better. We are now conducting a phase II trial of this novel adoptive cell therapy at the MTD of 1 e8 CB-NK cells/kg. Methods: This is a single-arm, phase II trial of CB-NK cells for MM patients undergoing SCT. A total of 33 patients will be enrolled. Patients with symptomatic MM who are candidates for autologous SCT are eligible. CB units with at least 4/6 match at HLA-A, B and DR are chosen. When possible, CB units with potential NK alloreactivity (as determined by KIR typing) are prioritized. On day (-19) CB units are thawed and mononuclear cells (MNCs) are cultured in a gas permeable bioreactor with irradiated (100 Gy) K562-based aAPCs expressing membrane bound IL-21 “Clone 9.mbIL21” (2:1 feeder cell:MNC ratio) and IL-2 (100 IU/mL) . On day 7, cells are CD3-depleted and remaining cells are re-stimulated with aAPC feeder cells and cultured for an additional 7 days. Patients receive lenalidomide (10 mg orally daily) from days (-8) to day (-2). Melphalan 200 mg/ m2 is given IV on day (-7). CB-NK cells are infused on day (-5) with autologous SCT on day (0). Peripheral blood is drawn for correlative analyses (in vivo CB- NK persistence and phenotype) at day (-4), day (0) and weekly thereafter. Responses at day 90 are determined by international Myeloma Working Group criteria. The primary endpoint is rate of complete response (CR) at 90 days after SCT. Secondary endpoints include in vivo persistence of CB-NK cells, in vivo CB-NK phenotype, PFS and OS. Modified Bayesian analysis is being utilized for ongoing toxicity and efficacy monitoring. Study was initiated in December 2014. Clinical trial information: NCT01729091.

Poster Session (Board #420a), Sun, 8:00 AM-11:30 AM

Autologous genetically engineered NY-ESO-1c259T in HLA-A*02:01, HLA*02:05 and HLA*02:06 positive patients with NY-ESO-1 expressing tumors. First Author: Crystal Mackall, National Cancer Institute at the National Institutes of Health, Bethesda, MD Background: The immunogenic cancer testis antigen NY-ESO-1 is expressed in several cancers. While NY-ESO-1 is not expressed in vital tissues, it is expressed in approximately 60% in advance myeloma (MM), 15% of ovarian (OC), 32% of advanced melanoma (Mel), 32 % in NSCLC, and 70% of synovial sarcoma (SS) tumors (based on TCGA). Expression is correlated with tumor proliferation and high risk features. An autologous T cell product bearing human-derived affinity-optimized T cell receptor (TCR) which recognizes the SLLMWITQC-HLA-A*A2 peptide complex (NY-ESO-1c259-T) is investigated in several clinical trials Methods: Six clinical studies across 5 tumor types were initiated in HLA-A*02:01 positive patients (pts) with NYESO-1 expressing tumors. Fifty three pts have been treated with NY-ESO1c259-T as of Jan 2016; 27 in MM and 26 in solid tumors. NCT01352286: phase I/II study in post-ASCT setting in pts with relapsed or refractory MM: 25 pts enrolled. NCT01892293: phase I/IIa non-transplant study in pts with relapsed or progressive MM: 2 pts enrolled. NCT01343043: pilot study in unresectable, metastatic, or recurrent SS: 16 pts enrolled: cohort 1 completed (n = 12); 2 additional cohorts have been added to evaluate conditioning and antigen level (4 pts treated across these cohorts to date). NCT01567891: phase I study in pts with refractory or platinum resistant OC and/or have received more than 2 lines of prior chemotherapy: 6 patients enrolled. NCT01350401: phase I study in pts with metastatic melanoma: 4 pts enrolled. NCT02588612: phase I/II study in pts with relapsed metastatic NSCLC: no pts enrolled. The main objectives are safety, tolerability, and efficacy. Biomarker objectives include persistence in peripheral of NY-ESO1c259-T in blood, T-cell lineage and functionality of these cells over time; characterization of cytokine profiles, as well as NY-ESO-1 antigen levels post infusion. Clinical trial information: NCT01352286, NCT01892293, NCT01343043, NCT01567891, NCT01350401, NCT02588612.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Immunotherapy TPS3102

Poster Session (Board #420b), Sun, 8:00 AM-11:30 AM

A first-in-human phase I trial of NKG2D chimeric antigen receptor-T cells in AML/MDS and multiple myeloma. First Author: Sarah Nikiforow, DanaFarber Cancer Institute, Boston, MA Background: Canonical CAR-T cell constructs encode a single chain antibody variable fragment, costimulatory domain, and signaling domain of CD3 zeta. This restricts recognition to 1 tumor antigen and a limited set of cancers. This study employs a novel CAR fusing full-length human Natural Killer Group 2D (NKG2D) gene with the human CD3 zeta signaling domain. NKG2D CAR receives costimulation via naturally-expressed DAP10 and activates T cells directly to kill and secrete cytokines upon recognition of MIC-A, MIC-B, and UL-16 binding proteins. These NKG2D-ligands are upregulated in many solid tumors and hematologic malignancies but absent or poorly expressed on healthy tissues. In multiple murine cancer models, NKG2D CAR-T cells induced complete remissions, T-cell memory, and altered tumor microenvironment via cytokines. We demonstrated manufacture of autologous NKG2D CAR-T cells in a GMP-environment from healthy adults and patients with AML and myeloma was feasible. Following isolation of mononuclear cells, T cells are activated with anti-CD3 mAb and IL-2, undergo 2 transductions with SFG retroviral vector containing NKG2D CAR construct, and expand in media containing IL-2. Validation studies yielded consistent viability, robust cell expansion, high vector-driven NKG2D expression on T cells, potent IFN-g production during tumor cell co-culture, viral copy number/cell , 5, and no replication-competent retrovirus in NKG2D CAR-T (CM-CS1) cells. Methods: A phase 1 dose-escalation study to establish safety and feasibility of a single infusion of CM-CS1 T cells without lymphodepleting conditioning is currently enrolling subjects with AML/MDSRAEB or relapsed/refractory progressive multiple myeloma without standard therapy options (ClinicalTrials.gov NCT02203825). Dose-escalation will proceed in 4 cohorts [1x106 to 3x107 CM-CS1 T cells] according to a 3+3 design followed by expansion cohorts at the MTDs in AML/MDS and myeloma for an anticipated 24 subjects. As of February, 2016, 6 subjects have been treated. Cohort 1 was completed without DLTs. Additional studies may include lymphodepletion, multiple infusions, and cytoreductive chemotherapy in both hematologic and solid tumor malignancies. Clinical trial information: NCT02203825.

TPS3104

Poster Session (Board #421b), Sun, 8:00 AM-11:30 AM

A phase 1 study of enoblituzumab in combination with pembrolizumab in patients with advanced B7-H3-expressing cancers. First Author: Naiyer A. Rizvi, Columbia University Medical Center, New York, NY Background: Enoblituzumab, an Fc optimized humanized IgG1 monoclonal antibody that binds to B7-H3 (CD276), a member of the B7 family, is currently in Phase 1 testing. Enoblituzumab is Fc-engineered with 5 amino acid substitutions to enhance binding to the activating FcgR and decrease binding to the inhibitory FcgR. B7-H3 has limited expression in normal tissue but high expression in many cancers including melanoma (M), squamous cell cancer of the head and neck (SCCHN) and non-small cell lung cancer (NSCLC). B7-H3 overexpression correlates with poor prognosis in a broad range of cancers suggesting a potential role in enabling tumor immune escape. The hypotheses for combining enoblituzumab with pembrolizumab are: 1) immune-modulating agent combinations may mediate additive or synergistic antitumor activity and in tumors where neither single agent alone has substantial anti-tumor effect, 2) engagement of both innate and adaptive immunity, 3) targeting both B7-H3 and PD-1 may enhance the immune response against tumors via modulation of T-cell immunosuppression, 4) limited expression of B7-H3 on normal tissues may limit the risk of autoimmune related adverse events; thus enoblituzumab may be combined more readily with immune-modulating agents. Methods: This Phase 1 trial is a dose escalation/cohort expansion study (NCT02475213) enrolling patients with advanced B7-H3 expressinors. The escalation phase uses a 3+3+3 plan to enroll successive cohorts of patients with escalating doses of weekly intravenous (IV) enoblituzumab starting at 3 mg/kg, and 2 mg/kg IV pembrolizumab administered every 3 weeks. Response is first determined at 6 weeks and followed by irRECIST; based on response, both drugs will be given up to 1 year. A 3-cohort expansion phase will open at the established maximum tolerated dose (16 patients each) with M, SCCHN, and NSCLC. Progression on previous checkpoint inhibitor is allowed. Study objectives are the characterization of safety, tolerability, pharmacokinetics, pharmacodynamics, and preliminary antitumor activity of the combination. This is the first clinical trial evaluating combined targeting of B7-H3 and PD-1 in patients with advanced cancer. Enrollment is ongoing. Clinical trial information: NCT02475213.

TPS3103

167s

Poster Session (Board #421a), Sun, 8:00 AM-11:30 AM

Phase 1/2 study of durvalumab and tremelimumab as monotherapy and in combination in patients with unresectable hepatocellular carcinoma (HCC). First Author: Ghassan K. Abou-Alfa, Memorial Sloan Kettering Cancer Center and Weill Cornell Medical College, New York, NY Background: No approved standard-of-care treatment options exist for patients with unresectable HCC who progress on, are intolerant to, or have refused sorafenib therapy. The PD-1/PD-L1 pathway is an important immune checkpoint used by tumor cells to block antitumor responses. Durvalumab is a selective, high-affinity human IgG1 monoclonal antibody that blocks PDL1 binding to PD-1 (IC50 0.1 nM) and CD80 (B7.1; IC50 0.04 nM). Tremelimumab is an IgG2 monoclonal antibody that inhibits CTLA-4. Tremelimumab monotherapy can induce objective tumor remission with a favorable safety profile in patients with HCC.1Targeting both PD-L1 and CTLA-4 pathways provides nonredundant pathway blockade of critical immune checkpoints. Methods: This phase 1/2, multicenter, open-label, randomized study evaluates the safety, tolerability, and antitumor activity of durvalumab and tremelimumab each as monotherapy (20 and 10 mg/kg Q4W, respectively) and in combination (20 and 1 mg/kg Q4W, respectively) in patients with unresectable HCC with or without concomitant viral hepatitis B or hepatitis C infection who have progressed on, are intolerant to, or have refused treatment with sorafenib. Key exclusion criteria are prior exposure to immune-mediated therapy, hepatic encephalopathy, active or prior documented gastrointestinal variceal bleed, and any concurrent chemotherapy, immunotherapy, or biologic or hormonal therapy for cancer. The primary objective is to assess the safety and tolerability of durvalumab and tremelimumab each as monotherapy and in combination. Key secondary objectives are to evaluate clinical outcomes (including objective response rate, duration of response, and overall survival) and describe the pharmacokinetics, pharmacodynamics (including PD-L1 expression), and immunogenicity of durvalumab and tremelimumab as monotherapy and in combination. Recruitment is ongoing. 1. Sangro B, et al. J Hepatol. 2013; 59:81-8. Clinical trial information: NCT02519348.

TPS3105

Poster Session (Board #422a), Sun, 8:00 AM-11:30 AM

Phase 1, first-in-human, open label, dose escalation ctudy of MGD009, a humanized B7-H3 x CD3 dual-affinity re-targeting (DART) protein in patients with B7-H3-expressing neoplasms or B7-H3 expressing tumor vasculature. First Author: Anthony W. Tolcher, START San Antonio, San Antonio, TX Background: MGD009 is a B7-H3 x CD3 Dual Affinity Re-Targeting (DART) protein. DART proteins are bispecific, antibody-based molecules that can bind two distinct antigens simultaneously. MGD009 is designed to redirect T cells to eliminate B7-H3-expressing target cells through co-engagement of B7-H3 on the target cell and CD3 on the T cell, each in a monovalent binding manner. MGD009 contains a mutated human IgG1 Fc domain to reduce / eliminate effector function via binding to Fc gamma receptors (FcgRs) and complement, but retains binding to the neonatal Fc receptor (FcRn) enabling use of the IgG salvage pathway to prolong circulating its half-life. B7-H3 protein expression is extremely limited in normal human tissues but is overexpressed in a broad range of tumor types, including melanoma (M), NSCLC (LC), SCCHN (HN), mesothelioma (MESO) and urothelial cancer (UC). By binding to B7-H3 on tumor cells and CD3 on T lymphocytes, MGD009 can recruit cytotoxic T cells, irrespective of their ability to recognize cancer cells, and trigger T-cell activation, expansion, and T-cell mediated killing of B7-H3- expressing tumors. The limited expression of B7-H3 on normal cells should restrict the potential for cytolytic activity directed at normal tissues, and unintended toxicity in patients treated with MGD009. Methods: This multi-center, open-label trial is a Phase 1 dose escalation / cohort expansion study (NCT02628535). All patients must have advanced B7-H3+ tumors. Prior checkpoint inhibitor therapy will be allowed. Dose escalation uses a 3+3+3 design, with patients treated every 2 weeks with escalating doses of IV MGD009 (starting dose 0.3 ug/kg). Cohort expansions (N = 16/cohort) treated at the maximum tolerated dose will include patients with M, HN, MESO, UC and LC. Major objectives include characterization of the safety, pharmacokinetics, pharmacodynamics, and preliminary antitumor activity of MGD009 in this ongoing study. Clinical trial information: NCT02628535.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

168s TPS3106

Developmental Therapeutics—Immunotherapy Poster Session (Board #422b), Sun, 8:00 AM-11:30 AM

TPS3107

Poster Session (Board #423a), Sun, 8:00 AM-11:30 AM

Avelumab (MSB0010718C; anti-PD-L1) in combination with other cancer immunotherapies in patients with advanced malignancies: The phase 1b/2 JAVELIN Medley study. First Author: Antoni Ribas, Ronald Reagan UCLA Medical Center, Drug Information Center, Los Angeles, CA

ENGAGE-1: A first in human study of the OX40 agonist GSK3174998 alone and in combination with pembrolizumab in patients with advanced solid tumors. First Author: Jeffrey R. Infante, Sarah Cannon Research Institute, Nashville, TN

Background: Programmed death-1 receptor ligand (PD-L1) is a key therapeutic target in the reactivation of the immune response against multiple cancers. Additionally, 4-1BB agonists have been shown to enhance T-cell proliferation and anti-tumor immunity. Avelumab*, a fully human anti-PDL1 IgG1 antibody, and PF-05082566, a fully human anti-4-1BB (CD137) IgG2 antibody, are being investigated in separate trial programs. In preclinical models, anti-4-1BB/anti-PD-L1 combination therapy had significantly improved anti-tumor activity compared with monotherapy of either agent. This open-label, phase 1b/2 study (NCT02554812) evaluates the combination of avelumab + anti-4-1BB in patients (pts) with selected advanced solid tumors. Other avelumab + immune modulator combinations are planned. The goal of this study is to determine safety, tolerability, and clinical activity of these combination therapies. Methods: A total of approximately 147 pts with histologically confirmed advanced selected solid tumors, ECOG PS 0-1, and archival or fresh tumor tissue (not preselected for PD-L1 expression) will be enrolled in this combination study. During the phase 1b lead-in for the study arm with PF-05082566, pts with NSCLC that has progressed on standard therapy will be randomized (1:1:1) to receive avelumab (10 mg/kg; 1h intravenous [IV] infusion Q2W) + PF-05082566 (500 mg, 100 mg, or 20 mg; 1h IV infusion Q4W). Based on dose-limiting toxicity (DLT) analysis in these cohorts, phase 2 will continue enrolling pts with NSCLC with expanded enrollment criteria at selected dose levels. In addition, pts with melanoma or squamous cell carcinoma of the head and neck will be enrolled at a single dose level selected for phase 2. Treatment will continue until disease progression, unacceptable toxicity, or withdrawal. Primary endpoints are DLT (phase 1b) and confirmed objective response (RECIST 1.1). Secondary endpoints include progression-free survival, overall survival, duration of response, pharmacokinetics, immunogenicity, tumor biomarkers, and adverse events (NCI-CTCAE v4.03). Enrollment began in Nov 2015. *Proposed INN. Clinical trial information: NCT02554812.

Background: OX40 (CD134) is a potent costimulatory tumor necrosis factor receptor expressed on activated CD4+ and CD8+ T cells. OX40 agonism promotes T-cell division and survival, resulting in stimulation of both immune effector and memory functions, while also blocking the suppressive function of regulatory T cells. This holds potential to overcome immune resistance and enhance immune mediated anti-tumor activity with OX40 agonism, particularly in combination with checkpoint inhibition. GSK3174998 is a humanized IgG1 anti-OX40 agonistic monoclonal antibody identified through collaboration with MD Anderson Cancer Center and is currently in phase I development. Methods: This is an open-label, nonrandomized, multicenter study of GSK3174998 administered alone and in combination with pembrolizumab in patients (pts) with selected advanced or recurrent solid tumors: non-small cell lung cancer, squamous cell carcinoma of the head and neck, renal cell carcinoma, melanoma, bladder, soft tissue sarcoma, triple-negative breast cancer, and colorectal carcinoma displaying microsatellite instability. The study will be conducted in 2 parts in approximately 180 pts. Primary objectives are to determine safety, tolerability, and maximum tolerated or administered dose of GSK3174998 as a single agent (Part 1) and when administered in combination with 200 mg pembrolizumab (Part 2). Secondary objectives include evaluation of antitumor activity, response rate and duration, progression-free survival, and overall survival, pharmacokinetics, pharmacodynamic activity in the blood and tumor microenvironment, and immunogenicity. Adverse events will be monitored using NCI CTCAE v. 4.0. Radiographic imaging will be obtained every 12 weeks to assess clinical response defined by immune-related RECIST. Tumor biopsies and blood samples will be collected before and during treatment to allow investigation of candidate biomarkers which may predict clinical response. As of Feb 1, 2016, Cohorts 1 and 2 have completed without DLT and enrollment to Cohort 3 is ongoing. Clinical trial information: NCT02528357.

TPS3108

TPS3109

Poster Session (Board #423b), Sun, 8:00 AM-11:30 AM

Poster Session (Board #424a), Sun, 8:00 AM-11:30 AM

Phase I STORM study (KEYNOTE 200): Intravenous delivery of a novel oncolytic immunotherapy agent, Coxsackievirus A21 in combination with pembrolizumab in advanced cancer patients. First Author: Hardev S. Pandha, University of Surrey, Surrey, United Kingdom

Antibody-interferon-alpha fusion protein (IGN002) for the treatment of bcell non-Hodgkin lymphomas: A phase 1, first-in-human, dose-escalation trial. First Author: Patricia Ann Young, University of California, Los Angeles, Los Angeles, CA

Background: Coxsackievirus A21 (CVA21, CAVATAK) is a naturally occurring ICAM-1 targeted oncolytic immunotherapeutic virus. Tumor infection by CVA21 can increase levels immune-checkpoint molecules, immune-cell infiltration and enhancement of systemic antitumor immune response. Pembrolizumab is a human programmed death receptor-1 (PD-1) blocking antibody that has yielded significant solid tumor responses via reversal of tumor induced T-cell suppression. Preclinical studies in an immunecompetent mouse model of melanoma confirmed that combinations of IV CVA21 + anti-PD-1 mAbs mediated significantly greater antitumor activity compared to use of either agent alone. We postulate that the combination of CVA21+pembrolizumab may translate to a similar benefit in the clinic. We describe a Phase I study assessing safety and efficacy of IV CVA21 6 pembrolizumab in advanced cancer pts. Methods: The Phase I STORM: Systemic Treatment Of Resistant Malignancies: NCT02043665 (KEYNOTE 200): Primary objectives are to assess dose-limiting toxicities (DLT) of CVA21 6 pembrolizumab. Secondary objectives are to assess ORR as by irRECIST 1.1 criteria, PFS, and OS. Treatment: Part A: Pts are infused with CVA21 in 100 mL saline in Cohort 1 (n = 3), at a dose of 1 x 108 TCID50 , in Cohort 2 (n = 3) at a dose of 3 x 108 TCID50 and in Cohort 3 (n = 12-18) at a dose of 1 x 109 TCID50 on study days 1,3,5,22 and Q3W for 6 additional infusions. Part A enrollment is almost complete. Part B: Pts are infused with CVA21 in 100 mL saline + pembrolizumab. In Cohort 1 (n = 3), CVA21 is administered at a dose of 1 x 108 TCID50 , in Cohort 2 (n = 3) at a dose of 3 x 108 TCID50 and in Cohort 3 (n = ~80) at a dose of 1 x 109 TCID50 on study days 1,3,5,8,29,and Q3W for 6 additional infusions. Pembrolizumab is given in all cohorts at 200 mg IV Q2W from Day 8 for up to 2 years. Treatment with CVA21 6 pembrolizumab will continue until confirmed CR or PD (whichever comes first) per irRECIST 1.1 or DLT. Key eligibility: Pts with advanced disease considered appropriate treatment with CVA21 6 pembrolizumab, lesion(s) accessible for core biopsy, ECOG PS 0-1, no active cerebral metastases, no autoimmunity/immunosuppression. Clinical trial information: NCT02043665.

Background: Interferon alpha (IFNa) has anti-proliferative and pro-apoptotic effects against many cancers, including NHL, and promotes both innate and adaptive anti-tumor immunity, but the clinical utility of IFNa is hindered by dose-limiting toxicity due to systemic activation of IFN receptors. IGN002 is a novel recombinant protein comprising a chimeric IgG1 anti-CD20 antibody fused to human IFNa2b via a peptide linker. In preclinical studies against human B cell NHL, IGN002 exhibited stronger direct antiproliferative effects, complement-dependent cytotoxicity, antibodydependent cell-mediated cytotoxicity, and in vivo potency in xenograft models compared to rituximab (Timmerman et al, Blood 126:2762, 2015). Methods: This phase I, multicenter study (NCT02519270) was initiated in January 2016. Primary objectives are to evaluate the safety and tolerability of multiple weekly IV doses of IGN002 and determine the maximum tolerated dose (MTD) or recommended phase 2 dose (RP2D). Secondary objectives are to characterize the pharmacokinetics of IGN002, effects on circulating B cells, incidence of anti-IGN002 antibodies, and preliminary anti-tumor activity. Eligible patients have relapsed/refractory, measurable ( . 1.5 cm) CD20-positive NHL including diffuse large B cell, mantle cell, marginal zone, lymphoplasmacytic, follicular, transformed follicular, or primary mediastinal B-cell NHL, adequate organ functions, and ECOG PS 02. Exclusion criteria include small lymphocytic lymphoma/CLL; therapy with chemotherapy/small molecules, anti-CD20 antibodies or radiation within 28 days; allogeneic stem cell transplantation; CNS NHL; systemic autoimmune disease; or active infection. In the Dose-Escalation Stage, 8 cohorts of 3-6 patients will be treated at dose levels of 0.0001, 0.0003, 0.001, 0.003, 0.01, 0.03, 0.1, and 0.3 mg/kg. In Period 1, subjects will receive 2 weekly doses of IGN002, and then may receive up to 24 additional weekly doses in Period 2. In the Expansion Stage, 14 subjects will receive up to 24 weekly doses at the MTD or RP2D. These results will inform future studies of antibody-targeted IFNa therapy for NHL and other cancers. Clinical trial information: NCT02519270.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Developmental Therapeutics—Immunotherapy TPS3110

Poster Session (Board #424b), Sun, 8:00 AM-11:30 AM

TPS3111

169s

Poster Session (Board #425a), Sun, 8:00 AM-11:30 AM

Phase 1 study to evaluate the safety and tolerability of the CD40 agonistic monoclonal antibody APX005M in subjects with solid tumors. First Author: Johanna C. Bendell, Sarah Cannon Research Institute/Tennessee Oncology, Nashville, TN

IPI-549-01-A phase 1/1b first in human study of IPI-549, a PI3K-g inhibitor, as monotherapy and in combination with pembrolizumab in subjects with advanced solid tumors. First Author: Anthony W. Tolcher, START San Antonio, San Antonio, TX

Background: Immunotherapy is an effective approach for the treatment of solid tumors. In addition to reversing tumor-induced immune suppression, immune activating antibodies are being explored as the next generation of immuno-oncology therapeutics. CD40 is primarily expressed on antigen presenting cells such as dendritic and B cells, and initiates and regulates both innate and adaptive immunity and is essential for the activation of tumor-specific T cells. APX005M is a humanized IgG1 CD40 agonistic antibody developed using Apexigen’s APXiMAB discovery platform. APX005M binds with high affinity to human CD40 (Kd = 0.12nM), recognizes a unique epitope that overlaps with the CD40 ligand binding sites and induces its potent agonistic activity through crosslinking by Fc-gamma receptors. Methods: This is an ongoing, phase I, multicenter, open-label, first-in-human, dose-escalation study evaluating APX005M in patients with advanced solid tumors. Eligible subjects ( $ 18 years) will have an Eastern Cooperative Oncology Group performance status of 0 or 1, histologically documented solid tumor with no standard treatment options available, measurable disease by RECIST 1.1, adequate bone marrow, liver, and kidney function, and no toxicities related to prior treatments. The primary objectives of the study are to assess safety and tolerability and to define the maximum tolerated dose of APX005M. Secondary objectives include an evaluation of the pharmacokinetic, immune pharmacodynamic and antitumor effects of APX005M. Recruitment is ongoing, with a target enrollment of approximately 32 subjects across 3 centers in the United States. Clinical trial information: NCT02482168.

Background: IPI-549 is a potential first-in-class potent and selective PI3K-g inhibitor being developed as a novel orally administered immuno-oncology therapeutic in multiple cancer indications. Preclinical studies demonstrate a role for PI3K-g in polarization of immune suppressive myeloid cells in the tumor microenvironment. Inhibition of PI3K-g by IPI-549 enhanced antitumor immune responses and inhibited tumor growth in syngeneic solid tumor preclinical models. In addition, IPI-549 in combination with immune checkpoint inhibitors showed increased tumor growth inhibition compared to each single agent in multiple pre-clinical models. These data served as the scientific foundation for initiating a clinical trial testing IPI-549 as a potential immuno-oncology therapy. This first-in-human clinical study will evaluate the safety and tolerability, and determine the recommended Phase 2 dose (RP2D) of IPI-549 when administered as a monotherapy and in combination with pembrolizumab (NCT02637531) in solid tumors. Methods: This multi-part Phase 1/1b open-label trial will initiate with monotherapy dose escalation cohorts consisting of an accelerated dose escalation phase followed by a standard phase with a 3+3 design. Evaluation of the PK, PD, and safety data in these cohorts will lead to the determination of the maximum tolerated dose (MTD) and RP2D of IPI-549 monotherapy. Subsequently, combination dose escalation cohorts will be initiated in which the combination of IPI-549 and pembrolizumab will be evaluated. Expansion cohorts evaluating the safety, PK, PD, and preliminary clinical activity of IPI-549 as a monotherapy and in combination with pembrolizumab will occur following the dose escalation phase. All subjects in the trial will have advanced and/or metastatic carcinoma or melanoma, and will have failed to respond to standard therapies. Combination expansion cohorts will recruit subjects with non-small cell lung cancer or melanoma who must have received an anti-PD-1 or anti-PD-L1 antibody as their most recent treatment. This trial is currently enrolling patients in the US. Clinical trial information: NCT02637531.

TPS3112

TPS3113

Poster Session (Board #425b), Sun, 8:00 AM-11:30 AM

A phase I study of pembrolizumab in combination with enadenotucirev (EnAd) (SPICE) in subjects with metastatic or advanced carcinoma. First Author: Wael A. Harb, Horizon Oncology Center, Lafayette, IN Background: EnAd is a tumor-selective chimeric Ad11/Ad3 group B oncolytic adenovirus developed using a process of directed evolution. Phase I clinical studies have identified a well-tolerated dose and regimen for EnAd monotherapy. EnAd shows a high level of selective replication and cell killing for a broad range of carcinoma cell lines with little replication in normal and non-carcinoma cells. Woller et al. have shown that an oncolytic virus can stimulate immune cell activity and that this stimulation can synergize with checkpoint inhibitor (CPI) activity and can reverse CPI resistance. This emerging data supports the rationale for the combination study of EnAd with CPI, pembrolizumab (anti-PD-1 antibody). SPICE is a phase I multi-center, open label, nonrandomized clinical study in subjects with metastatic or advanced carcinoma. The study design allows for an initial dose escalation stage, followed by a dose expansion stage which will include treatment of tumors normally non-responsive to CPI. Methods: In the initial dose escalation phase, 3 cohorts of patients with metastatic or advanced colorectal, urothelial, squamous cell head and neck and salivary gland carcinomas are administered intravenously an increasing dose of enadenotucirev and pembrolizumab in a standard "3 + 3" design. Enrolment to cohort 1 began in January 2016. In both phases, the combination treatment will be administered in 21 day treatment cycles, each comprising three intravenous (IV) infusions of enadenotucirev separated by 48 hours (on Day 1, Day 3 and Day 5) and one IV infusion of pembrolizumab on Day 15. Eligible subjects will then enter a treatment period of up to six treatment cycles or until disease progression. The primary objectives of the study are to establish the MTD/ MFD of enadenotucirev and pembrolizumab treatment combination, to evaluate the safety and to recommend doses of enadenotucirev and pembrolizumab for future Phase II studies. Overall response, duration of response, clinical benefit rate and progression free survival are secondary endpoints which will be assessed according to RECIST Version 1.1 and irRECIST Version 1. Clinical trial information: NCT02636036.

Poster Session (Board #426a), Sun, 8:00 AM-11:30 AM

A phase I study in patients with a human papillomavirus type 16 positive oropharyngeal tumor treated with second generation synthetic long peptide vaccine conjugated to a defined adjuvant. First Author: Marije Slingerland, Leiden University Medical Center, Leiden, Netherlands Background: The synthetic long peptide (SLP) vaccine against the human papillomavirus type 16 (HPV16) oncoproteins E6 and E7 applied to highgrade premalignant HPV16-induced lesions of the vulva was safe and resulted in successful clearance of the lesion in half of the patients correlating with strong and broad T-cell responses against the vaccine peptides. However, in HPV16+ cervical cancer patients the vaccine responses were less vigorous and not sufficient for tumor regression. A second generation SLP vaccine was formulated, in which two E6 long peptides were conjugated to a synthetic Toll-like receptor 2 (TLR2) ligand called Amplivant. The TLR2 ligand-SLP conjugates were shown to enhance antigen presentation by dendritic cells, thereby improving induction of effective anti-tumor T cells in vitro and in preclinical studies. The two selected E6 peptides are immunodominant peptides containing both CD4+ and CD8+ T-cell epitopes in natural HPV16 immunity and upon vaccination with HPV16-SLP in previous clinical trials. Methods: In a recently initiated phase I clinical trial the two with Amplivant conjugated E6 peptides as a mixture (the vaccine is called Hespecta; HPV E Six Peptide Conjugated To Amplivant) were administered to HPV16+ oropharyngeal cancer patients, who first were treated with curative intent. Four dose escalation levels will be tested in six patients per peptide dose, namely 1, 5, 20 or 50 mg of conjugated peptides. The vaccine is administered intradermally three times with a three-week interval. The decision to start enrolment at the next dose level is made by assessing the safety after 4 out of 6 patients have completed the first follow-up visit after the third vaccination with the previous dose level. The dose escalation phase will be discontinued if in two or more patients a grade 3 or 4 vaccine-related toxicity occurs. The primary objective is to determine the ability of the vaccine to induce HPV16 E6-specific T-cell responses. The secondary objective is to study the safety of Hespecta. Currently, 2 patients have been enrolled. Clinical trial information: 2014-000658-12.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

170s TPS3114

Developmental Therapeutics—Immunotherapy Poster Session (Board #426b), Sun, 8:00 AM-11:30 AM

Randomized phase II trial of a DNA vaccine encoding prostatic acid phosphatase (PAP) versus GM-CSF adjuvant in patients with biochemically recurrent prostate cancer. First Author: Glenn Liu, University of Wisconsin Carbone Cancer Center, Madison, WI Background: We have previously reported the results from a phase I trial using a PAP-encoding DNA vaccine (pTVG-HP) in men with PSA-recurrent, non-metastatic, non-castrate, prostate cancer (clinical stage D0/M0). In that trial, patients received six intradermal immunizations, with GM-CSF coadministered as an adjuvant, over 12 weeks. No significant adverse events were detected, and CD4+ and CD8+ T cell immune responses to the PAP target antigen were elicited. The development of persistent Th1-type immunity was associated with favorable changes in PSA doubling time. A second trial, conducted in men with PSA-recurrent, non-metastatic, castrate-resistant prostate cancer (D0.5/M0) evaluated this vaccine administered over a 2-year period. In that trial there were no significant adverse events, the development of persistent Th1-biased PAP-specific immunity was detected, and many patients experienced a prolongation of PSA doubling time. Based on these results, we are currently conducting a multicenter trial using pTVG-HP plus GM-CSF adjuvant, versus GM-CSF adjuvant only, in patients with stage D0/M0 prostate cancer. Methods: Patients with non-castrate, non-metastatic prostate cancer (negative CT and bone scans), with rising PSA and a doubling time of , 12 months, are being randomly assigned to receive pTVG-HP + GM-CSF versus GM-CSF adjuvant. Treatments are being delivered intradermally every 2 weeks x 6, and then quarterly for up to 24 months. The primary endpoint is 2-year metastasis-free survival. Secondary endpoints include median progression-free survival, and determining whether immunization affects PSA doubling time. Laboratory endpoints include determining whether PAP-specific T-cell immunity is associated with progression-free survival. The trial was expanded in 2014 to include a total of 106 patients allowing for further exploratory biomarker analyses and to increase the statistical power, now enabling a . 90% power to detect an anticipated difference of 40% versus 70% in 2-year metastasisfree survival at the one-sided 5% significance level. Clinical trial information: NCT01341652.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Colorectal) Cancer 3500

Oral Abstract Session, Sun, 8:00 AM-11:00 AM

Validation of the Immunoscore (IM) as a prognostic marker in stage I/II/III colon cancer: Results of a worldwide consortium-based analysis of 1,336 patients. First Author: Jerome Galon, INSERM, Paris, France Background: Mounting evidence indicates an enhanced lymphocytic reaction as an informative prognostic indicator in colon cancer (CC). The IM methodology has been defined to quantify the in situ immune infiltrate. Methods: A Society for Immunotherapy of Cancer-led international consortium of 23 pathology expert centers from 17 countries was initiated to evaluate the standardized IM assay in routine clinical settings. Patient (pt) eligibility criteria were CC Stages I/II/III and no neo-adjuvant treatment. 1336 pts split into a training set (TS) and internal validation set (IVS) were quantified for IM using standardized procedure, immunohistochemistry using CD3 and CD8 antibodies, and quantification using digital pathology on whole slide section of primary CCs. Statistical analysis plan was pre-defined and all statistical analyses performed by blinded external statisticians. Primary study endpoint was time-to-recurrence (TTR); analyses were by Cox models stratified by enrolling center. Results: Median recurrent follow-up duration across centers was 5.9 years. Overall pts: 51% Male, median age 69 years, 19%/56%/25% stage I/II/III. Among pts with Stages I/II/III CC, in the TS TTR was shorter among 332 pts (48.1%) with Low-IM CC vs. 358 pts with High-IM CC (HR (95% CI), 0.35 (0.23-0.52); P , 0.0001). In the IVS (630 pts) TTR was also shorter among 303 pts with Low-IM CC vs. 327 pts with High-IM CC (HR (95% CI), 0.54 (0.34-0.84); P = 0.006). In both groups, results were independent of pt age, sex, tumor stage and sidedness. Among pts with Stage II CC, the difference in TTR between Low and High-IM was significant both in the TS (HR (95% CI), 0.27 (0.14-0.51); P , 0.0001) and in the IVS (HR (95% CI), 0.46 (0.24-0.87) P = 0.014); multivariate results similar (TS HR (95% CI), 0.28 (0.14-0.54); P , 0.0001) and IVS HR (95% CI), 0.46 (0.24-0.87); P = 0.0142)). Reproducibility of the standardized IM assay was validated across centers. Conclusions: The primary endpoint of the worldwide pre-specified IM study was reached. TTR was significantly longer in pts stages I/II/III with High-IM. Low-IM identified a subgroup of patients with high-risk stage II CC.

3501

171s Oral Abstract Session, Sun, 8:00 AM-11:00 AM

Nivolumab 6 ipilimumab in treatment (tx) of patients (pts) with metastatic colorectal cancer (mCRC) with and without high microsatellite instability (MSI-H): CheckMate-142 interim results. First Author: Michael J. Overman, The University of Texas MD Anderson Cancer Center, Houston, TX Background: Evidence supports use of nivolumab (N) in MSI-H mCRC. N, a fully human anti-PD-1 mAb and ipilimumab (I), a humanized anti-CTLA-4 mAb, have favorable safety & efficacy in other tumors. CheckMate-142, a phase 2 study, evaluates N 6 I in pts with mCRC, MSI-H and non-MSI-H. Methods: Pts had ECOG PS 0–1, and intolerance/ progression on $ 1 tx. MSI-H pts received N 3 mg/kg q2 wk (N3) or N 3 mg/kg + I 1 mg/kg q3 wk (N3+I1) x 4 doses followed by N3 until disease progression (PD) or other discontinuation. Initial evaluation of N+I at 3 doses was completed in non-MSI-H pts. Primary endpoint was investigator-reported ORR by RECIST 1.1; other endpoints were safety, OS, and PFS. Results: 33 (N3) and 26 (N3+I1) MSI-H pts, and 3 (N1+I1), 10 (N1 +I3), and 10 (N3+I1) non-MSI-H pts were enrolled. 82% (N3) and 92% (N3+I1) of MSIH and 100% of non-MSI-H pts had $ 2 prior regimens. 15% (N3) and 25% (N3+I1) of MSI-H pts had known BRAF V600E. 17 (52%; N3) and 19 (73%; N3+I1) MSI-H pts remain on tx. Efficacy results are shown in the Table. In MSI-H pts, tx-related adverse events (TRAEs) occurred in 26 (79%; N3) and 22 pts (85%; N3+I1); most common were diarrhea and fatigue (27% each; N3) and diarrhea (46%; N3+I1). Grade 3–4 TRAEs occurred in 7 (N3) and 8 pts (N3+I1). One pt on N3 had a Grade 5 TRAE (sudden death). In non-MSI-H pts median (95% CI) PFS was 1.4 mo (1.2–1.9; pooled N+I). Conclusions: N and N+I were well tolerated in most pts and demonstrated encouraging clinical activity and survival in MSI-H mCRC. This study is ongoing. Clinical trial information: NCT02060188.

MSI-Ha efficacy.

ORR, n (%) CR Confirmed PR SD PD Not determined/not reported Median duration of response (95% CI), mo Median PFS (95% CI), mo 4-mo PFS rate,b % Median OS (95% CI), mo 5-mo OS rate,c %

N3 (n = 33)

N3+I1 (n = 26)

9 (27) 0 9 (27) 8 (24) 11 (33) 5 (15) NR (4.2–NE) 5.3 (1.4–NE) 55 16.3 (8.3–NE) 75

4 (15) 0 4 (15) 17 (65) 3 (12) 2 (8) NR (NE–NE) NR (NE–NE) 80 NR (NE–NE) 100

NR, not reached; NE, not estimable aBy local screen bPFS Kaplan-Meier plot estimate, N3 = 17/33 events, N3+I1 = 4/26 events cPFS Kaplan-Meier plot estimate, N3 = 11/33 events, N3+I1 = 0/26 events

3502

Oral Abstract Session, Sun, 8:00 AM-11:00 AM

Clinical activity and safety of cobimetinib (cobi) and atezolizumab in colorectal cancer (CRC). First Author: Johanna C. Bendell, Sarah Cannon Research Institute, Nashville, TN Background: Atezolizumab (atezo; MPDL3280A) is an engineered antibody that inhibits binding of PD-L1 to its receptors, PD-1 and B7.1. Atezo has demonstrated monotherapy activity in a multitude of human tumor types. However, response rates in microsatellite stable (MSS) CRCs have been lower than in other indications. In preclinical models, targeted inhibition of MEK leads to upregulation of MHC I on tumor cells, induces intratumoral Tcell infiltration and enhances anti-PDL1 activity. We therefore conducted a Phase Ib study combining cobi (MEK inhibitor) and atezo in patients (pts) with advanced solid tumors. Methods: Cobi was escalated from 20 to 60 mg daily (21 days on/7 days off) and combined with atezo 800 mg IV q2w. Tumor-specific expansion cohorts, including KRAS-mutant CRC, and serial biopsy cohorts in solid tumors were opened upon determination of the MTD. Safety, tolerability and confirmed ORR by RECIST v1.1 were evaluated. Results: As of October 12, 2015, 23 CRC (22 KRAS mutant, 1 WT) pts were enrolled during escalation and expansion. No dose-limiting toxicities were observed, and expansion occurred at atezo 800 mg q2w and cobi 60 mg. Median follow-up for safety in CRC pts was 3.78 mo (range, 1.1-11.7). The most common treatment-related AEs included diarrhea (69.6%), fatigue (52.2%), dermatitis acneiform (43.5%), rash (34.8%), maculopapular rash (26.1%), pruritus (26.1%) and nausea (26.1%). Incidence of treatmentrelated G3-4 AEs was 34.8%. The only treatment-related G3-4 AE in $ 2 pts was diarrhea (8.7%). No G5 AEs were reported. The ORR was 17% (4 PR, 5 SD). Three responses were ongoing (range, 4.0 to 7.7 mo at time of data cutoff). Three responders were mismatch repair-proficient, and 1 was unknown. Response was not associated with baseline PD-L1 expression. Results from the serial biopsy cohort showed enhanced PD-L1 upregulation, CD8 T-cell infiltration and MHC I expression on treatment, providing mechanistic rationale for the combination. Conclusions: The combination of cobi and atezo in CRC is well tolerated at the maximum administered doses. These results show that pts with MSS CRC can respond to the combination of cobi and atezo, and provide support for continued evaluation of the combination. Clinical trial information: NCT01988896.

3503

Oral Abstract Session, Sun, 8:00 AM-11:00 AM

NCI9673: A multi-institutional eETCTN phase II study of nivolumab in refractory metastatic squamous cell carcinoma of the anal canal (SCCA). First Author: Van Karlyle Morris, The University of Texas MD Anderson Cancer Center, Houston, TX Background: The incidence of SCCA continues to rise annually in the US. 20% of patients (pts) will develop metastatic (met) disease which lacks a consensus approach to treatment. SCCA is largely driven by immune evasion of HPV-specific CD8 and CD4 T cells which promote oncogenesis for SCCA. Nivolumab (Nivo), a monoclonal antibody targeting PD-1 on T cells, promotes immune-mediated anti-tumor activity of T cells against HPV-positive cells in vitro. This is the first phase II trial of Nivo for pts with refractory met SCCA. Methods: Previously treated but immunotherapy na¨ıve met SCCA pts were eligible. PD-L1 expression was not required. HIV+ (CD4 count . 300/ uL) and hepatitis B/C pts were eligible. A Simon two-stage phase II trial (Ho: p , .05, Ha: p $ .20) was conducted. All pts were evaluated by RECIST Criteria 1.1. Pts received Nivo (3 mg/kg) IV every 2 weeks. Optional pretreatment and on-treatment tissue biopsies and plasma samples were collected for immune biomarkers and HPV/p16 status. All correlatives were evaluated at MDACC by the Immunotherapy Platform team and the Core Facility. Results: 39 pts were screened across the ETCTN network (May 2015 - October 2015); 37 pts were eligible. Median age was 56 years (interquartile range [IQR], 51.1-63.6); M:F was 12:25. Median number of prior therapies: 2 (range 1-8). All pts were evaluable for toxicity; 33 pts were evaluable for response. Median number of cycles: 6 (IQR, 3-10). Seven (21%) pts had a partial response and 19 (58%) pts had stable disease; disease control rate of 79%. Ten pts (one HIV+) remain on study [7 pts for . 6 months (M)]. Median progression-free survival was 4.1M. Common adverse events (AE): fatigue, nausea, and rash. Six pts had grade 3 AE’s: fatigue (N = 2) and one pt each of pneumonitis, rash, anemia, and hyperglycemia. Conclusions: Currently, there is no consensus approach for met SCCA. NCI9673 is the first prospective phase II trial of nivolumab in refractory metastatic SCCA. Single agent nivolumab demonstrated potentially meaningful activity and was well tolerated. Further evaluation of immune checkpoint therapy in met SCCA is justified. Updated clinical results will be presented. Exploratory correlative work is ongoing. Clinical trial information: NCT02314169.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

172s

Gastrointestinal (Colorectal) Cancer

3504

Oral Abstract Session, Sun, 8:00 AM-11:00 AM

Impact of primary (1o) tumor location on overall survival (OS) and progression-free survival (PFS) in patients (pts) with metastatic colorectal cancer (mCRC): Analysis of CALGB/SWOG 80405 (Alliance). First Author: Alan P. Venook, University of California, San Francisco, San Francisco, CA Background: 80405 found no OS or PFS difference when Bevacizumab (BV) or Cetuximab (Cet) was added to 1st-line FOLFOX or FOLFIRI in mCRC pts. As location of the 1° may affect mCRC outcome, we assessed the impact of 1° side (R v L) on OS and PFS in 80405 pts. Methods: 1° location was determined by chart review: 1137 pts w/ KRAS wt (codons 12 and 13) in main cohort; 252 pts w/ KRAS mut tumors treated w/ BV or Cet pre-amendment. R-sided = cecum to hepatic flexure; L-sided = splenic flexure to rectum. Transverse (T) = hepatic to splenic flexure. PFS per investigator. Kaplan Meier and Cox regression methods used. Results: KRAS wt pts: Median age = 59; synchronous = 78%. 1° site: R – 280 (25%); L – 689 (61%); T- 62 (5%); unsure – 106 (9%). OS & PFS (Table) difference by side statistically significant if adjusted for age, gender, BV / Cet, chemotherapy, prior therapy. There was a significant 1° side by biologic interaction (P int = 0.003, PFS and OS) but not by chemo, gender or RAS. OS, L-sided: Cet v BV, superiority (Log rank p = 0.04); R-sided: BV v Cet, superiority (p = 0.03). Results similar for PFS and if T colon allocated to R side. KRAS mut pts: 1°s: R - 35%; L- 50%. No statistically significant difference in any subset although OS favors L . R (only OS data shown). Conclusions: mCRC arising in the R v L colon are clinically different. In KRAS wt mCRC, pts w/ L-sided 1° tumor have superior OS and PFS v pts w/ R-sided 1°. Though not pre-planned analyses, OS and PFS were prolonged w/ Cet in L and w/BV in R but were poorer w/ Cet in R. Forthcoming molecular analysis of 1°s - e.g. BRAF, MSI, methylation may provide a biological explanation. For now, stratification in mCRC studies by R v L 1° sidedness is indicated. These data support BV in 1st line treatment for mCRC pts w/Rsided 1° tumor regardless of KRAS status. Support: U10CA180821, U10CA180882. Clinical trial information: NCT00265850. Right 1°

Left 1°

OS (mos) All KRAS wt Cet BV All KRAS mut

19.4 16.4 24.5 23.1

Hazard Ratio, 95% Confidence Interval (R v L, adjusted)

34.2 37.5 32.1 30.3

1.56 1.97 1.26 1.28

(1.32, (1.56, (1.00, (0.95,

1.84) 2.48) 1.58) 1.73)

11.5 12.0 11.1

1.25 (1.08, 1.46) 1.54 (1.25, 1.91) 1.03 (0.83, 1.28)

Log Rank p-value (adjusted) , 0.0001

PFS (mos) All KRAS wt Cet BV

3506

8.9 7.7 9.5

.002

Oral Abstract Session, Sun, 8:00 AM-11:00 AM

Association of primary (1°) site and molecular features with progression-free survival (PFS) and overall survival (OS) of metastatic colorectal cancer (mCRC) after anti-epidermal growth factor receptor (aEGFR) therapy. First Author: Michael Sangmin Lee, The University of North Carolina at Chapel Hill, Chapel Hill, NC Background: CRCs originating in the right colon (RC) have features distinct from those in the rectum/left colon, including more frequent CpG island methylator phenotype (CIMP-High) and BRAF mutation (MT). Prognosis and response to therapies, including aEGFR regimens, differ dramatically by 1° site in mCRC, but pathobiologic explanations remain unclear. Methods: 198 KRAS wild-type (WT) mCRC tumors were tested for CIMP status via bisulfite pyrosequencing and PCR amplification. BRAF and NRAS were sequenced. PFS on first aEGFR regimen was retrospectively determined in 167 patients. Univariate and multivariate Cox regression analyses with multiple imputations were performed. Tumors from 179 independent CRC patients were tested for promoter methylation (Illumina HumanMethylation450) and gene expression. Results: Inferior PFS with aEGFR therapy was associated with RC CRC (HR = 1.56, CI 1.01-2.41), CIMP-High (HR = 2.38, 1.47-3.85), BRAF MT (HR = 2.15, 1.26-3.65), and NRAS MT (HR = 2.11, 1.23-3.65). On multivariate analysis, CIMP status (p = 0.041), BRAF MT (p = 0.038), and NRAS MT (p = 0.012) remained significant, but 1° site did not (p = 0.27). Inferior OS was associated with RC CRC (HR = 1.45, 1.04-2.01), CIMP-High (HR = 1.53, 1.08-2.16), BRAF MT (HR = 2.46, 1.61-3.75), and NRAS MT (HR = 1.70, 1.03-2.81). RC 1° site was also associated with CIMP-High (OR = 2.35, 1.22-4.54) and BRAF MT (OR = 5.45, 2.47-12.03). Independent data from MDACC and TCGA showed RC 1° site and CIMP-High were both strongly associated with hypermethylation of promoters and suppression of expression of the EGFR ligands EREG and AREG. 1° site was also strongly associated with consensus molecular subtypes (CMS) 1 and 3 (p , 0.001, p = 0.03), which were previously shown in mCRC patients to have worse outcomes than the ‘classic’ CMS2 subtype. Conclusions: CRC 1° site is associated with OS and PFS after aEGFR therapy. Molecular analyses suggest that BRAF MT, NRAS MT, molecular subtypes, and tumor methylation account for the effect and may provide a biologic explanation for the association with anatomic location.

3505

Oral Abstract Session, Sun, 8:00 AM-11:00 AM

The relationship between primary tumor sidedness and prognosis in colorectal cancer. First Author: Deborah Schrag, Dana-Farber Cancer Institute, Boston, MA Background: Clinical trials including CALGB/SWOG 80405 reveal differences in overall survival for patients with colorectal cancer based on the location of the 1° tumor yet this information is not routinely included in study design, reporting or patient counseling. Objective: To evaluate the impact of 1° tumor location on prognosis among stage-specific cohorts of patients with colorectal cancer. Methods: The Surveillance Epidemiology and End Results Program (SEER) tracks sidedness of primary colorectal cancers. We characterized the 1° site using SEER data as R-sided 1° = cecum to transverse colon; L-sided 1° = splenic flexure to sigmoid descending colon; 1° rectum = rectosigmoid and rectal. We identified patients with diagnoses of primary colorectal cancer in a SEER region between 2007 and 2011 and followed through 2013 by stage at diagnosis. We measured median and3-year survival and used Kaplan-Meier plots and Cox regression used to compare OS across groups. Age, gender, race, ethnicity, and year of diagnosis were included in adjusted Cox models to estimate the hazard ratio (HR) for death of rectal and right-sided relative to left-sided tumors. Results: Right sided 1° CRC had inferior prognosis to both left-sided and rectal cancers and this difference persisted after adjusting for differences in other clinical and demographic characteristics. The association was less consistent for stage I and II cancers. Conclusions: In a population-based series of patients with stage III and IV CRC, patients with R-sided tumors had inferior survival. This heterogeneity argues for consistent reporting of 1° site and further research including molecular sub-typing to better understand the mechanisms underpinning this phenotypic difference in prognosis. Stage/location of primary tumor Stage IV Left Right Rectal Stage III Left Right Rectal

3507

N

Median survival

3-year survival probability %

Unadjusted HR (95% CI)

Adjusted HR (95% CI)

4784 7579 4392

17.0 8.7 17.4

27 16 26

1.0 1.40 (1.35-1.46) 0.99 (0.94-1.04)

1.0 1.20 (1.15-1.25) 1.02 (0.97-1.07)

6394 13748 5745

n/a 59 n/a

71 62 70

1.0 1.39 (1.32-1.46) 1.02 (0.96-1.08)

1.0 1.17 (1.11-1.23) 1.10 (1.04-1.17)

Oral Abstract Session, Sun, 8:00 AM-11:00 AM

CREST: Randomised phase III study of stenting as a bridge to surgery in obstructing colorectal cancer—Results of the UK ColoRectal Endoscopic Stenting Trial (CREST). First Author: James Hill, Manchester Royal Infirmary, Manchester, United Kingdom Background: Uncertainty remains about the efficacy and safety of endoluminal stenting as an alternative to emergency surgery in patients with potentially curable obstructing left sided colorectal cancer. Emergency presentations still account for 20% of all colorectal cancer cases; obstruction is present in 80% of these. Methods: Patients presenting in the emergency setting with left-sided colonic obstruction needing urgent decompression and radiological features of a carcinoma were randomized to either: endoluminal stenting as a bridge to surgery or surgical decompression. Patients were stratified according to curative intent based on pre-operative staging investigations. A combined endoscopic/fluoroscopic technique was standardized in stent workshops with elective surgery performed 1-4 weeks later. Results: 246 patients from 39 units were randomized between 2009 and 2014 with 98% complying with allocated treatment. 92% were treated with curative intent. 30-day post-operative mortality (5.3% vs 4.4%) and length of hospital stay [15.5 days (IQR 10-26) vs 16 days (10-27)] were similar with stenting and surgery. Stenting achieved relief of obstruction in 82% of patients and reduced stoma formation; 69% emergency surgery v 45% with stenting as a bridge to surgery (p , 0.001). There were no significant differences in QoL at 3 and 12 months or critical care utilization. 1 year mortality for potentially curative patients did not differ by treatment group. Conclusions: CReST is the largest trial of endoluminal stenting in obstructing colorectal cancer. In patients fit enough to undergo surgery, stenting as a bridge to surgery reduced stoma formation without a detrimental effect on one-year survival. Post-operative mortality, length of hospital stay, critical care usage and Quality of Life were not different between the two treatment groups. Clinical trial information: 13846816.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Colorectal) Cancer 3508

Oral Abstract Session, Sun, 8:00 AM-11:00 AM

A randomized trial comparing mesorectal excision with or without lateral lymph node dissection for clinical stage II, III lower rectal cancer: Primary endpoint analysis of Japan Clinical Oncology Group study JCOG0212. First Author: Shin Fujita, Department of Surgery, Tochigi Cancer Center, Utsunomiya, Japan

3509

173s Clinical Science Symposium, Mon, 11:30 AM-1:00 PM

Clonality patterns of driver mutations (mut) to reveal spatial-temporal genomic heterogeneity in colorectal cancer (CRC). First Author: Rodrigo Dienstmann, Oncology Data Science (ODysSey) Group, Vall d’Hebron Institute of Oncology, Barcelona, Spain

5-year RFS 73.4% (68.5-77.7%) 73.3% (68.3-77.6%) 1.07 (0.81-1.41) 5-year OS 92.6% (89.3-94.9%) 90.2% (86.5-92.9%) 1.25 (0.85-1.84) 5-year LRFS 87.7% (83.8-90.7%) 82.4% (78.0-86.1%) 1.37 (0.98-1.93)

Background: The mutant allele fractions (MAFs) of a driver gene may influence response to matched targeted therapies, such as KRAS and antiEGFR antibodies in CRC. We performed a comprehensive analysis of clonalsubclonal frequencies of driver alterations in unpaired prim (prim) vs metastatic (met) CRC and studied their potential clinical implications. Methods: 642 CRC samples with targeted mut analysis from 2010-2015 (Sequenom in 469, MiSeq in 173; prim in 496, met in 110). MAFs of each variant were normalized for tumor purity in the sample. Non-parametric statistics and Cox models are described. Results: APC mut were found in 50% (median MAF 0.5; reference group for comparisons), TP53 mut in 65% (0.67; p = .035), KRAS mut in 46% (0.56; p = .2), NRAS mut in 4% (0.51; p = .9), BRAF mut in 8% (0.28; p , .001) and PIK3CA mut in 16% (0.39; p , .001). Half of samples with KRAS/PIK3CA coexisting mut had significantly lower PIK3CA than KRAS MAFs (0.27 [n = 34] vs 0.46 [n = 39]; p , .001). We only found higher MAFs in met vs prim after chemo +/- anti-EGFR for TP53 (0.77 [n = 70] vs 0.56 [n = 32]; p = .004) and BRAF (0.44 [n = 8] vs 0.26 [n = 42]; p = .04). BRAF MAFs were higher in non-V600 than V600 (0.45 [n = 5] vs 0.27 [n = 48]; p = .02). In 31 patients (pts) treated with 41 anti-BRAF regimens, no correlation between BRAF V600 MAFs and time to treatment failure was found (p = .36). Pts with any RAS mut tumors, irrespective of MAFs, had worse time from recurrence to death (TRD) as compared to RAS/BRAF/PIK3CA wild-type (43 vs 34 months [m], HR 1.39; p = .004), but coexisting PIK3CA mut, either clonal or subclonal, did not further impact survival (31 m and 35 m, HR 1.07; p = .7). Pts with BRAF V600 mut had worse TRD (21 m, HR 2.2; p , .001), but not those with BRAF non-V600 mut (46 m, HR 1.35; p = .55). Conclusions: Our results indicate clonality of RAS mut and subclonality of BRAF V600 mut and a subset of PIK3CA mut. Differences in prim vs met sites for TP53 and BRAF V600 MAFs suggest acquired copy number events and clonal selection after therapy. BRAF V600 MAFs in prim tissue did not predict benefit with targeted drugs in met setting. Despite significant differences in prognosis according to mut in driver oncogenes, their clonality patterns did not influence survival.

3510

3511

Background: Mesorectal excision (ME) is the standard surgery for lower rectal cancer. However, lateral pelvic lymph node metastasis is occasionally found, and ME with lateral lymph node dissection (LLND) is the standard procedure in Japan. We conducted a randomized trial to confirm the noninferiority of ME alone to ME with LLND in terms of efficacy and reported the ME group tended to show lower postoperative morbidity than the ME+LLND group. This is the primary analysis of the relapse-free survival (RFS) as the primary endpoint. Methods: Eligibility criteria included histologically proven rectal cancer; clinical stage II/III; main lesion located in the rectum and the lower margin below the peritoneal reflection; no lateral pelvic lymph node swelling; PS of 0 or 1; patient age 20-75 years. After surgeons had confirmed R0 resection by the ME procedure, patients were randomized intraoperatively to ME alone or ME with LLND. The primary endpoint was RFS and the non-inferiority margin of the hazard ratio (HR) was 1.34. The planned sample size was 700 with a power of 75% and a one-sided alpha of 5%. Results: A total 701 patients enrolled from 33 institutions were randomized to ME+LLND (n=351) or ME (n=350) between June 2003 and August 2010. The 5-year RFS was 73.4% and 73.3% in the ME+LLND group and the ME group, respectively. The HR was 1.07 [90.9% CI 0.84-1.36 (.1.34)]; thus the non-inferiority of ME was not confirmed (p=0.055). The 5-year overall survival (OS) and the 5-year local-recurrence-free survival (LRFS) were shown in the Table. The numbers of patients with local recurrence were 25 (7.1%) and 44 (12.6%) in the ME+LLND group and the ME group, respectively (p=0.02). Conclusions: The non-inferiority of ME to ME with LLND was not confirmed. ME with LLND significantly reduced local recurrence after surgery compared with ME alone. Clinical trial information: C000000034. ME+LLND (95% CI)

ME (95% CI)

HR (95% CI)

Clinical Science Symposium, Mon, 11:30 AM-1:00 PM

Clinical outcome and benefit of oxaliplatin in colon cancer according to intrinsic subtypes: Results from NRG Oncology/NSABP C-07. First Author: Kay L. Pogue-Geile, NSABP/NRG Oncology, Pittsburgh, PA Background: In the C-07 trial for stage II/III colon cancer, oxaliplatin added to 5-fluorouracil + leucovorin (FULV) provided significant but modest absolute benefit for DFS. We hypothesized that molecular subtypes would be associated with differential prognosis and benefit from oxaliplatin. Methods: C-07 cases were divided into discovery (n = 848) and validation (n = 881) cohorts based on the chronological order of tissue block submission. Patients (pts) from a discovery cohort were subtyped with the CRCA classifier described by Sadanandam et al (enterocyte, goblet-like, transit amplifying, stem-like, and inflammatory). A re-estimated centroid using 72 genes included in a custom nCounter gene code set was used to determine subtypes and their association with oxaliplatin benefit in the discovery cohort. The validation cohort was then prospectively examined with a prespecified statistical analysis plan. Post-hoc analysis included examination of the entire cohort with CRCA, CCS and CMS subtyping methods. Results: In both discovery and validation cohorts, stem-like subtype was associated with poor prognosis and lack of benefit from oxaliplatin. While stage III C-07 pts with an enterocyte subtype had a statistically significant benefit from oxaliplatin in the discovery cohort (Hazard Ratio [HR] = 0.223), only a trend for benefit was observed in the validation cohort (HR = 0.525, P= 0.141). Failure to validate may have been due to limited power or a limited number of genes defining the enterocytye subtype because the code set was designed before subtypes were described. Examination of the different subtyping methods shows that all 3 methods robustly identified the same poor prognostic pts (stem-like, CCS-3 and CMS-4). Conclusions: The apparent clinical utility of subtyping includes the identification of the stem-like subtype because it can be robustly identified, and have a very poor prognosis. Patients with stem-like subtype in our retrospective study received no benefit from oxaliplatin. Future clinical trials are required to determine the extent of the clinical utility of subtyping for oxaliplatin benefit. Support: U10-CA-180868, -189867, -180822; Sanofi Clinical trial information: NCT00004931.

Clinical Science Symposium, Mon, 11:30 AM-1:00 PM

The potential of circulating tumor DNA (ctDNA) to reshape the design of clinical trials testing adjuvant therapy in patients with early stage cancers. First Author: Jeanne Tie, Walter and Eliza Hall Institute of Medical Research, Melbourne, Australia Background: The conventional approach to testing the benefit of adjuvant therapies in patients (pts) with relatively favorable prognoses is to follow a large number of pts for long periods of time, hoping that mature outcome data will document an improved outcome compared to control pts. We reasoned that the design of such trials could be improved if pts with minimal residual disease could be identified a priori through the presence of ctDNA, and the effects of adjuvant therapy then assessed through serial ctDNA assays. Methods: We carried out a prospective trial in 231 pts with Stage II colon cancer. Serial plasma samples were collected every 3 months starting 4-10 weeks after surgery. Somatic mutations in pts’ tumors were identified via sequencing of 15 genes commonly mutated in colon cancer. We then designed personalized assays to quantify ctDNA in plasma samples. Adjuvant chemotherapy was administered at clinician discretion, blinded to ctDNA analysis. Results: Somatic mutations were identified in 230 (99.6%) of tumors. Matching ctDNA was detected in the immediate post-operative period in 14 of 178 (8%) pts not treated with chemotherapy, 11 of whom had recurred (79%) at a median follow-up of 27 months. In contrast, recurrence occurred in only 16 (10%) of the 164 pts with negative ctDNA not treated with chemotherapy (HR 15.66, log-rank P,0.0001). ctDNA was detected in the immediate post-operative period in 6 of 52 pts who went on to receive chemotherapy. The ctDNA status turned from positive to negative during adjuvant treatment phase in all 6 pts (100%) but became positive again following completion of chemotherapy in 2 pts, both of whom have recurred. In patients with serial samples available, the median lead-time between ctDNA detection and radiologic-recurrence was 167 days. Conclusions: Detection of ctDNA in pts with resected stage II colon cancer provides direct evidence of residual disease. As well as defining pts at very high risk of later radiologic-recurrence, serial ctDNA analysis may provide an early readout of adjuvant treatment effect. Including ctDNA analyses would increase the efficiency of clinical trials testing the benefit of adjuvant treatment. Clinical trial information: ACTRN12612000326897.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

174s 3512

Gastrointestinal (Colorectal) Cancer Poster Discussion Session; Displayed in Poster Session (Board #209), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 1:15 PM-2:30 PM

FOLFIRINOX combined to targeted therapy according RAS status for colorectal cancer patients with liver metastases initially non-resectable: A phase II randomized Study—Prodige 14 – ACCORD 21 (METHEP-2), a unicancer GI trial. First Author: Marc Ychou, Montpellier Cancer Institute, Montpellier, France

3513

Poster Discussion Session; Displayed in Poster Session (Board #210), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 1:15 PM-2:30 PM

FOLFIRINOX as induction treatment in rectal cancer patients with synchronous metastases (RCSM): Results of the FFCD 1102 phase II trial. First Author: Jean Baptiste Bachet, Pitie´ Salpetri ´ ere ` Hospital, Paris, France

Background: Liver metastases (LM) from colorectal cancer (CRC) are initially resectable in only 10-15% of patients (pts). The conversion to resectability following induction chemotherapy is an important strategy to increase survival. Our study was designed to determine the most appropriate chemotherapy (associated with a targeted therapy) for CRC pts with LM considered as initially unresectable. Methods: This French phase II, multicenter, prospective trial, randomized pts between bi-chemotherapy (BiCT) versus tri-chemotherapy (TriCT). The population was initially stratified by targeted therapy depending on KRAS status and then by RAS status (from 02 Dec 2013 due to the change in cetuximab’s [Cet] marketing authorization): Cet for wt(K)RAS pts and bevacizumab (Bev) for mtRAS pts. The hypothesis was to increase the rate of LM resection (R0-R1) from 50% with BiCT to 70% with TriCT (bilateral a-test 5%; power 90%). Results: 256 patients were randomized in 33 sites from February 2011 till April 2015: 126 BiCT (FOLFIRI [56 pts]; FOLFOX4 [70 pts]) and 130 TriCT (FOLFIRINOX). The resection rate (R0 or R1; CI95%) of the LM was 45.2% [36; 54] for pts treated with BiCT vs 56.9% [48; 66] for TriCT (p = 0.062).The LM resection rate (R0 or R1; CI95%) was 44.7% [35; 55] for pts treated with Bev (mtRAS) vs 55.6% [47; 64] for Cet (wtRAS) (p = 0.087). At the time of data analysis, the median follow-up (CI95%) was 22.5 months [19.6;29.5] for the BiCT pts and 23.5 months [19.8; 28.8] for the TriCT pts and at analysis 78 patients had died. The median overall survival (OS) is significantly different (p = 0.048): in the TriCT Arm the median OS was not reached and is 36 months [23.5;40.6] in the BiCT Arm. The severe toxicity rate was 37.6% for BiCT vs 41.7% for TriCT (p = 0.503). 38 BiCT pts and 34 TriCT pts had surgical complications, with two deaths in each arm. Conclusions: First line FOLFIRINOX chemotherapy, in association with a targeted therapy, showed a higher rate of LM R0/R1 resections than standard BiCT (FOLFIRI or FOLFOX4) combined with the same targeted therapy, with a statistically significant difference in terms of OS. Clinical trial information: 2009-012813-22.

Background: Optimal therapeutic strategy in patients (pts) with RCSM remains discussed and many front line options can be discussed to best treat primary tumor and metastatic disease: surgery (S), radiotherapy (RT), chemoradiotherapy (CRT) or chemotherapy (CT). Upfront aggressive chemotherapy may be particularly interesting in this setting. Methods: Pts with histologically proven RCSM received FOLFIRINOX: oxaliplatin (O) 85 mg/m2 d1 + irinotecan (I) 180 mg/m2 d1 + leucovorin 400 mg/m2 d1 followed by 5FU 400 mg/m2 bolus d1 and 2,400 mg/m246h continuous infusion biweekly ; 8 cycles were mandatory then therapeutic strategy was investigators’ choice. Eligibility included adequate organ function, age $ 18 years, PS # 2, no prior RT or CT. CT-scan and MRI were centrally reviewed. The primary endpoint was disease control rate at 4 m (4m DC). With a Simon 2-stage design, a targeted (H1) 4m DC . 75% was defined (unilateral alpha of 5% and power of 90%). Results: 65 pts were enrolled (07/2012 to 02/2015): male 78%; median age 61 years; PS 0-1 99%; liver metastases 92%; $ 2 metastatic sites 63%. All pts received at least 1 cycle of CT; 85% of pts received the 8 planned cycles. Median relative dose-intensities of O, I, 5FU bolus and 5FU continuous were 93.5%, 96.5%, 86% and 96%, respectively. The main grade 3-4 toxicities were neutropenia (29%), diarrhea (12%), abdominal pain (9%), fatigue (8%) and nausea (5%). Local symptoms (rectal bleeding, rectal syndrome and/or sub-occlusion), present in 72% of pts at baseline, were, 16% and 10% at 2 and 4 m, respectively. Primary objective was met with a 4m DC of 94% (95% CI, 86.3-97.8). Objective response rate (RECIST) was 86%. After FOLFIRINOX, 36 pts (55%) had a RT or CRT, 28 (43%) had a resection of the rectal tumor and 18 (28%) had a surgery and/or ablation of their metastases. With a median follow-up of 21.5 m (95% CI, 15.4-24.7), median PFS was 10.9 m (95% CI, 8.8-12.3) with rates of PFS at 6 and 12 m of 82% and 41%. Conclusions: Front line aggressive systemic CT with FOLFIRINOX allows a good local and distant control of RCSM, with acceptable toxicities and leaves the opportunity to decide best locoregional treatment and surgery of metastatic lesions on a controlled disease at 4m. Clinical trial information: NCT01674309.

3514

3515

Poster Discussion Session; Displayed in Poster Session (Board #211), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 1:15 PM-2:30 PM

Bevacizumab or cetuximab plus chemotherapy after progression with bevacizumab plus chemotherapy in patients with wtKRAS metastatic colorectal cancer: A randomized phase II study (Prodige 18 –UNICANCER GI). First Author: Sandrine Hiret, ICO Rene´ Gauducheau, Nantes, France Background: Second-line treatment with chemotherapy plus bevacizumab (Bev) or cetuximab (Cet) is now recognized as a standard treatment in mCRC. The main objective of this French multicenter, prospective open randomized trial was to evaluate the Progression Free Survival (PFS) rate at 4 months in patients receiving chemotherapy plus Bev or Cet after failure with Bev plus chemotherapy. Methods: The main eligibility criteria were disease progression after first-line chemotherapy: 5-FU with irinotecan or oxaliplatin associated with bevacizumab in wtKRAS mCRC. Patients were treated with cross over chemotherapy (FOLFIRI or mFOLFOX6) with bevacizumab (Arm A) or cetuximab (Arm B) until progression or limiting toxicity. The tumor response was evaluated every 6 weeks until progression using RECIST 1.1. Results: 133 patients, 86 male(64.7%), PS 0 (n = 74, 57.8%), 1 (n = 54, 42.2%), in 25 sites in France, were included between October 2010 and May 2015. Most frequent chemotherapy regimens delivered were mFOLFOX6 + Bev (n = 41) or Cet (n = 42); FOLFIRI + Bev (n = 25) or Cet (n = 25). The PFS rate at 4 months was 79% in Arm A and 66.7% in Arm B (p = 0.09). Secondary objectives included median PFS: 7.1 months (Arm A) vs 5.6 months (Arm B) (HR = 1.43; 95%CI [0.99-2.06] p = 0.06), and median OS: 15.9 months (Arm A) vs 10.6 months (Arm B) (HR = 1.44; 95%CI [0.952.18] p = 0.08). Conclusions: In wtKRAS mCRC patients progressing after bevacizumab plus chemotherapy, continuation beyond progression with bevacizumab and crossover chemotherapy is associated with a numerically higher but not statistically significant median PFS and OS compared to cetuximab plus chemotherapy. Final data with KRAS and NRAS analysis will be presented at the meeting. Clinical trial information: NCT01442649.

Poster Discussion Session; Displayed in Poster Session (Board #212), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 1:15 PM-2:30 PM

MAVERICC, a phase II study of mFOLFOX6-bevacizumab (BV) vs FOLFIRIBV as first-line (1L) chemotherapy (CT) in patients (pts) with metastatic colorectal cancer (mCRC): Outcomes by tumor location and KRAS status. First Author: Heinz-Josef Lenz, University of Southern California, Norris Comprehensive Cancer Center, Keck School of Medicine, Los Angeles, CA Background: Standard 1L mCRC treatment (tx) includes a CT backbone (e.g. modified leucovorin [LV]/5fluorouracil [5-FU]/oxaliplatin [mFOLFOX6] or LV/ 5-FU/ irinotecan [FOLFIRI]) and a biologic (e.g. BV). The preferred CT backbone for anti-VEGF tx is unknown. MAVERICC (NCT01374425), a global, randomized, open-label, phase 2 trial, assessed efficacy and safety of mFOLFOX6-BV vs FOLFIRI-BV in pts with mCRC. PFS and OS were comparable between FOLFIRI-BV and mFOLFOX-BV (Lenz et al, ASCO GI 2016). Exploratory analyses compared clinical outcomes of the BV-containing tx groups by primary tumor location and baseline (BL) KRAS status. Methods: Pts with mCRC ($1 measurable metastatic lesion, ECOG performance status #1) were randomized 1:1 to BV (5 mg/kg) + mFOLFOX6 or FOLFIRI every 2 wks. Time to PFS and OS for each tx group were estimated by Kaplan–Meier methods, hazard ratios (HR) were estimated by Cox regression, p-values were based on log-rank tests stratified by BL ERCC-1 status and region. Results: 376 pts were randomized: median age, 61 yr; white race, 83%; US region, 85%. BL: KRAS mutant (MUT), 34%; KRAS wild type (WT), 55%; KRAS unknown, 11%; left-sided primary tumor, 59%. Efficacy results are shown. Conclusions: There was a trend toward improved PFS and OS in the FOLFIRI-BV group, regardless of primary tumor location or BL KRAS status. PFS benefit from FOLFIRIBV was statistically significant in left-sided tumors only. PFS and OS were numerically greater in pts with WT vs MUT KRAS status. Clinical trial information: NCT01374425.

ITT mFOLFOX6-BV (n=188) FOLFIRI-BV (n=188) HRa Primary tumor location Right mFOLFOX6-BV (n=75) FOLFIRI-BV (n=79) HRa Left mFOLFOX6-BV (n=113) FOLFIRI-BV (n=109) HRa KRAS Exon 2c (by tx) mFOLFOX6-BV (n=107) FOLFIRI-BV (n=101) HRa MUT mFOLFOX6-BV (n=65) FOLFIRI-BV (n=63) HRa KRAS Exon 2c (all tx) WT (n=208) MUT (n=128) HRa a

Median PFS (mo)

Median OS (mo)

10.1 12.6 0.79

23.9 27.5 0.76

10.0 11.2 0.88

22.7 27.4 0.76

10.2 13.8 0.71b

24.1 27.5 0.86

11.2 14.0 0.83

26.1 36.7 0.77

9.5 12.3 0.72

22.5 26.9 0.83

12.5 10.9 1.29

28.8 24.6 1.34

FOLFIRI vs mFOLFOX6 bp=0.040, others p.0.05 cKRAS status of 40 pts unknown

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Colorectal) Cancer 3516

Poster Discussion Session; Displayed in Poster Session (Board #213), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 1:15 PM-2:30 PM

3517

175s

Poster Discussion Session; Displayed in Poster Session (Board #214), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 1:15 PM-2:30 PM

MiR 31 3p as a predictive biomarker of cetuximab efficacy effect in metastatic colorectal cancer (mCRC) patients enrolled in FIRE-3 study. First Author: Pierre Laurent-Puig, Hopital ˆ Europeen ´ Georges Pompidou, Assistance Publique Hopitaux ˆ de Paris, Paris, France

HER2 amplification as a negative predictive biomarker for anti-epidermal growth factor receptor antibody therapy in metastatic colorectal cancer. First Author: Kanwal Pratap Singh Raghav, The University of Texas MD Anderson Cancer Center, Houston, TX

Background: MiR 31 3p expression has been shown to be predictive of survival in RASwild-type (WT) mCRC patients treated with anti-EGFR therapies. We aimed to validate the predictive value of miR 31 3p in FIRE-3, a randomized clinical trial, based on a cut-off threshold of miR 31 3p expression identified in previous studies. Methods: MiR 31 3p expression was measured by qRTPCR after extraction from 370 RAS WT paraffin-embedded tumor samples. 191 patients were treated with FOLFIRI plus bevacizumab and 179 FOLFIRI plus cetuximab in first-line therapy. Patients were defined as low or high miR 31 3p expressors based on a pre-specified cut-off threshold. Primary objective was to assess cetuximab benefit on overall (OS) and progression free survival (PFS) in the low expressor group (a = 0.025). Interaction between miR 31 3p expression and treatment was tested (a = 0.10) and assessed using a propensity score. Secondary objectives were to evaluate treatment effect on survival in high expressors and objective response rate (ORR) in both high and low expressors. Cox and logistic multivariate regressions were used to estimate hazard ratios (HR) and odds ratios (OR) respectively. Results: In low expressors (67% of studied patients), the risk of death was significantly decreased in patients treated with cetuximab as compared to bevacizumab (HR = 0.60 [0.42–0.85], p = 0.005; median OS: 39 vs. 28 months respectively). No significant difference was seen in high expressors (HR = 1.10 [0.70–1.75] p = 0.67; median OS: 20 months for both treatment arms). MiR 31 3p was found to be predictive of cetuximab effect on OS and PFS (interaction test: p = 0.07 and p = 0.08 respectively). For patients whom ORR data were available, we showed a benefit for cetuximab in low expressors (OR = 3.37 [1.70–6.67] p = 0.0005). Furthermore, miR 31 3p was found to be predictive of cetuximab effect on ORR (significant interaction: p = 0.06). Conclusions: Low miR 31 3p expressors have a significantly better OS and ORR when treated with FOLFIRI plus cetuximab as compared to FOLFIRI plus bevacizumab. MiR 31 3p is predictive of cetuximab effect on OS, PFS and ORR and should be used to select patients who will have greater benefit from cetuximab therapy.

Background: HER2 amplification (HERamp), seen in 5% of KRAS wildtype (WT) metastatic colorectal cancers (mCRC), is associated with resistance to anti-epidermal growth factor receptor antibodies (antiEGFRabs). The purpose of this study was to validate the predictive impact of HERamp in mCRC. Methods: We performed systematic analyses of RAS and BRAF WT mCRC patients (pts) across 2 distinct cohorts. We tested HERamp in cohort 1 (N = 97) using immunohistochemistry and dual in-situ hybridization (HERamp: HER2/CEP17 $ 2.2). We validated these findings in cohort 2 (N = 99), which comprised of 37 cases of HERamp mCRC pts identified by next-generation sequencing (HERamp: $ 4 copies) and 62 HER2 non-amplified (HER2NA) pts treated previously with antiEGFRabs who served as controls. The primary objective was to compare progression-free survival (PFS) in pts treated with antiEGFRabs. PFS and overall survival (OS) were estimated using Kaplan Meier method and compared using log rank test. Results: HERamp was seen in 14 (14 %) of RAS/BRAF WT pts in cohort 1. In this cohort, median OS (29.1 v 45.1 months (m), P = 0.78) and PFS on first line therapy without an antiEGFRab (PFS1) (9.7 v 8.4 m, P = 0.70) was similar between HERamp and HER2NA pts. A total of 66 pts in cohort 1 received antiEGFRab after first line therapy. Median PFS on antiEGFRab therapy (PFS2) was significantly shorter in pts with HERamp compared to HER2NA tumors (2.9 v 8.1 m, hazard ratio (HR) 5.0, P , 0.0001). These findings were confirmed in cohort 2, in which 69 pts received antiEGFRab after first line therapy and median PFS2 was significantly shorter for HERamp pts compared to HER2NA pts (2.8 v 9.3 m, HR 6.6, P , 0.0001) with a similar OS (P = 0.86) and PFS1 (P = 0.62). Conclusions: HER2 amplification in mCRC is a predictive biomarker for lack of efficacy of antiEGFRab therapy. This magnitude of effect is comparable to RAS mutations; the only other validated predictive biomarker for antiEGFRabs, and affects 1 in 8 patients currently receiving these agents. Patients with RAS/RAF WT mCRC should be screened for HER2 amplification prior to treatment with antiEGFRabs and should be considered for early referral to clinical trials.

3518

3519

Poster Discussion Session; Displayed in Poster Session (Board #215), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 1:15 PM-2:30 PM

Association of tumor infiltrating lymphocytes (TILs) with molecular subtype and prognosis in stage III colon cancers (CC) from a FOLFOX-based adjuvant chemotherapy trial. First Author: Frank A. Sinicrope, Mayo Clinic, Rochester, MN Background: High levels of TILs are associated with favorable prognosis in some solid tumors. We examined TIL levels in CC subtyped using mutations in KRAS or BRAF oncogenes as well as DNA mismatch repair (MMR) status, and examined association with patient (pt) prognosis. Methods: Stage III CC (N=2293) from a randomized trial of adjuvant FOLFOX + cetuximab (NCCTG N0147) were analyzed for TILs; dichotomized as high (. 4 TILs per HPF) vs low at light microscopy. We determined the association of TILs with clinicopathological features, mutations in KRAS exon 2 or BRAF(V600E), MMR status, and DFS by multivariate Cox regression analyses. Results: Higher TIL densities were found in tumors with wild-type (wt) vs mutated (mut) KRAS (p=.038) or BRAF (p,.0001). Among tumors with mut BRAFV600E or mut KRAS, high vs low TILs were found in 19.2% and 9.5%, respectively. In tumors with mut BRAF, high TILs were detected in 31.2% of dMMR vs 9.5% of pMMR cancers (p,.0001). TILs were increased 3.6-fold in dMMR vs proficient (p) MMR tumors (p,.0001). Also associated with high vs low TILs were female vs male (p=.011), white vs non-white (p=.008), proximal vs distal (p,.0001), lower T stage (p=.004), fewer positive/examined nodes (p=.015), and high vs low grade (p=.0001). High vs low TILs were associated with better DFS overall [HR=0.76 (0.57-1.01), p=0.047], and among tumors with wt KRAS [HR= 0.61 (0.41-.89), p=.006] or wt KRAS and BRAF [HR= 0.56 (0.34-.92), p=.013]. TILs were not prognostic in tumors with mut KRAS (p=.71; TIL*KRAS interaction p= 0.097) or mut BRAF (p=.18). High vs low TILs were associated with favorable DFS in pt tumors with dMMR [HR= 0.55 (.30-1.01), p=.045], but not pMMR (p=.301; interaction p=0.17]. No significant differences were found by treatment arm. Conclusions: Significantly lower TILs are found in tumors with mut vs wt KRAS or BRAF suggesting reduced immunogenicity. Among tumors with wt, but not mut, KRAS and BRAF, high vs low TILs were associated with better survival. TILs were increased 3.6-fold in dMMR vs pMMR, and favorably prognostic within dMMR pts. These results for wt tumors and those with dMMR suggest that immune activation improves survival outcomes.

Poster Discussion Session; Displayed in Poster Session (Board #216), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 1:15 PM-2:30 PM

Prognostication using molecular (mol) markers and clinicopathological (clpath) features in high-risk stage II/III colon cancer (CC). First Author: Rodrigo Dienstmann, Sage Bionetworks, Seattle, WA Background: AJCC v.7 staging and clpath covariates do not accurately predict outcome of CC patients (pts) eligible for adjuvant chemotherapy (adjCh). Mol markers, microsatellite instability (MSI) and mutations (mut) in BRAF and KRAS, may further improve prognostication in high-risk CC. Methods: Data from 4,796 stage II/III pts accrued to phase 3 trials PETACC3 and N0147 (train set) were used to construct Cox models for overall survival (OS). Variables included stage (pT, pN); clpath (# lymph nodes [LNs] positive, # LNs assessed, tumor grade, primary site, age, gender, adjCh); and mol (MSI, BRAF and KRAS mut). Final models (1. TNM; 2. TNM + mol; 3. TNM + clpath; 4. TNM + clpath + mol) were internally (train set) and externally (test set) validated on 597 stage II/III treated pts from non-clinical trial cohorts. Results: All examined traits were statistically significant for OS prediction (Table 1). We found significant interaction between site and MSI (p=.03) in multivariate models, with worse OS for MSI vs MSS left CC (HR 1.22) and better OS for right CC (HR 0.70). Models 1 to 4 associated with C-indices of 0.65, 0.68, 0.70, 0.72, respectively, in train set; and 0.68, 0.71, 0.73, 0.74 in test set. Corresponding time-dependent AUCs (5 years summary) were 0.54, 0.66, 0.73, 0.74 in train set; and 0.55, 0.68, 0.72, 0.73 in test set. Conclusions: Incorporation of mol markers (MSI, BRAF and KRAS mut) improves prognostic estimation in stage II/III CC pts treated with adjCh. We propose their prospective assessment in larger clinical cohorts, including untreated pts, and development of online calculators that could aid in prognostication (model-based staging) and patient/physician communication.

Multivariable OS Cox model in train set. Variable pT2 vs pT1 pT3 vs pT1 pT4 vs pT1 pN1 vs pN0 pN2 vs pN0 # LNs positive .=12 vs ,12 LNs assessed High vs Low/Moderate grade Right vs left Age (years) Male vs female FOLFIRI vs 5FU FOLFOX vs 5FU MSI vs MSS BRAF mut vs wt KRAS mut vs wt

HR 1.06 2.22 4.23 1.76 2.88 1.07 0.70 1.29 1.53 1.01 1.36 1.01 0.70 0.72 1.80 1.45

CI95% 0.52 1.24 2.32 1.17 1.74 1.05 0.59 1.09 1.33 1.00 1.16 0.80 0.59 0.57 1.48 1.26

2.19 4.36 8.33 2.39 3.72 1.08 0.82 1.51 1.81 1.02 1.54 1.30 0.92 0.90 2.28 1.72

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

176s 3520

Gastrointestinal (Colorectal) Cancer Poster Discussion Session; Displayed in Poster Session (Board #217), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 1:15 PM-2:30 PM

Immunologic profiling of consensus molecular subtype (CMS) stratified colorectal cancer (CRC) primary and liver metastectomy specimens: Implications for immune targeting of proficient mismatch repair CRC. First Author: Matthew Reilley, The University of Texas MD Anderson Cancer Center, Houston, TX

3521

Poster Discussion Session; Displayed in Poster Session (Board #218), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 1:15 PM-2:30 PM

Final results of STAR-01: A randomized phase III trial comparing preoperative chemoradiation with or without oxaliplatin in locally advanced rectal cancer. First Author: Carlo Aschele, Medical Oncology Unit, General Hospital, La Spezia, Italy

Background: While checkpoint blockade therapy has shown promise in mismatch repairdeficient CRC, these represent a small fraction of overall cases. Limited data exists regarding the immune microenvironment of mismatch repair-proficient CRC, hindering efforts to develop novel immunotherapy treatments for the majority of CRC patients. Methods: Quantification of immune subsets was evaluated by immunohistochemical (IHC) analyses in 23 primary MSI tumors, 45 primary MSS tumors and 34 untreated liver metastatectomy (LM) specimens for both center of tumor (CT) and invasive margin (IM) regions. Immunologic data, tumor microsatellite status, and consensus molecular subtype (CMS) were evaluated for possible correlations. Results: In CT, we found average CD3, CD4 and CD8 T cell infiltrates to be 6%, 3%, and 2% respectively for LM vs 9%, 4% and 5% for MSI vs 7%, 4% and 4% for MSS. In IM, we found average CD3, CD4, and CD8 T cell infiltrates to be 24%, 11% and 12% for LM vs 29%, 16% and 18% for MSI vs 25%, 14% and 12% for MSS. Higher T cell infiltrates were found in IM as compared to CT (p , 0.01) and there were fewer CD8 T cell infiltrates found in metastases as compared to primary tumors (p , 0.01). We observed higher average frequency of CD68 macrophages in LM in both CT (26%) and IM (55%) as compared to MSI (CT = 21%; IM = 32%; p , 0.01) and MSS (CT = 18%; IM = 32%; p , 0.01). Average CT PD-L1 expression on tumor cells was higher in MSI (27%) vs MSS (9%) and LM (3%). We did not observe significant differences between LM, MSI and MSS for CD45RO and granzyme-B expression. Stratified by CMS subtypes, CD8 T cell infiltration (p , 0.01) and PD-L1 expression on tumor cells (p , 0.01) were significantly higher in CMS1 (immune subtype). CMS3 (metabolic subtype) demonstrated higher expression of ICOS in CT (p , 0.05) and OX40 (CT p = 0.05; IM p , 0.01) than other subtypes. Conclusions: Our findings support evidence for efficacy of PD-1/PD-L1 blockade in CMS1 and suggest that CMS3 CRC subtype may benefit from unique immunotherapy combinations. These findings have important implications for future immunotherapy approaches in CRC and merit further investigation.

Background: STAR-01 is a randomized phase III trial investigating the effect of adding oxaliplatin (OXA) to preoperative (preop) 5-fluorouracil (FU)-based pelvic chemoradiation (CRT) in patients (pts) with resectable locally advanced rectal cancer (LARC). No benefit from adding OXA could be demonstrated on primary tumor response to preoperative chemoradiation (Aschele, J Clin Oncol, 2011). Methods: Eligibility required a resectable, biopsy-proven rectal adenocarcinoma within 12 cm from the anal verge with radiological evidence of perirectal fat or nodal involvement. Randomization was between infused FU (225 mg/msq/day) concomitant to external-beam pelvic radiation (50.4 Gy in 28 daily fractions) (arm A) or the same regimen + weekly OXA (60 mg/msq x 6) (Arm B). Overall survival (OS) was the primary endpoint (252 deaths required to have a log-rank test with 80% power to detect a 30% reduction in mortality rates at the 5% two-sided significance level). Results: From November 2003 through August 2008, 739 eligible patients were enrolled at 41 Italian institutions. With a median follow up of 8.8 years (IQR 8.1-9.9), 132 and 109 deaths were recorded in Arm A and B, respectively (HR 0.82, CI 0.64-1.06, p=0.126). The corresponding 5 and 10 years OS rates were 77.6 vs 80.4 % and 62.3 vs 67.4 % (Arm A vs B). Preoperative/intraoperative progressions were thrice higher in Arm A (28, vs 10 in Arm B) while recurrences and deaths without recurrence were evenly distributed between the 2 arms (83/81 and 14/19 in Arm A/B, respectively). The corresponding event-free survival rates were 70.6 vs 74.2 % and 66.3 vs 69.2 % in Arm A vs B, at 3 and 5 years, respectively (HR 0.89, CI 0.69-1.15, p=0.374). Conclusions: This study did not meet its primary endpoint of a 30% reduction in mortality rates. The observed differences are consistent with a more limited impact on OS and warrants further investigation with pooled analyses of similar studies investigating OXA added to FP-based preop CRT. The large reduction in the number of early progressions is coherent with an effect of OXA on micrometastatic disease but was not paralleled by an improvement in recurrences after surgery.

3522

3523

Poster Discussion Session; Displayed in Poster Session (Board #219), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 1:15 PM-2:30 PM

Phase II trials of cetuximab plus combined modality therapy (CMT) in squamous cell carcinoma of the anal canal (SCCAC) with and without human immunodeficiency virus (HIV) infection. First Author: Madhur Garg, Montefiore Medical Center, Bronx, NY Background: SCCAC is a human papilloma virus (HPV) associated cancer that is potentially curable with CMT, commonly associated with HIV infection, and characterized by high local regional failure (LRF) rates. Since cetuximab enhances the effect of radiation in HPV-associated oropharyngeal SCC, we evaluated the efficacy and toxicity of cetuximab plus CMT in SCCAC. Methods: We performed 2 concurrently planned and conducted phase II trials of cetuximab plus CMT in stage I-III SCCAC without (E3205) and with (AMC045) HIV infection. CMT included cisplatin (75 mg/m2) and 5-FU (1000 mg/m2/day x 4 days) x 2 cycles plus RT (45-54 Gy), plus 2 cycles of neoadjuvant cisplatin/5-FU in the first 28 patients in E3205 prior to a study amendment. Cetuximab (400 mg/m2 IV, then 250 mg/ m2 IV weekly x 8 weeks) began 1 week prior to CMT. The studies were designed to detect at least a 50% reduction in 3-year LRF rate (35% to 17.5%, 1-sided alpha 0.10, power 90%). Secondary objectives included progression free survival (PFS) and overall survival (OS). Results: The characteristics of the study population and efficacy results are summarized below, indicating higher rates of node positive disease in AMC 045 (47%) and E3205 (54%) compared with R9811 (26%) and ACTII (32%). The majority of patients completed protocol therapy in AMC045 (82%) and E3205 (79%), rates of grade 4 toxicity were similar (32% in E3205, 26% AMC45), and 4% had treatment-associated deaths. Conclusions: Patients with SCCAC with and without HIV infection exhibited similar rates of completing therapy (80%) and clinical outcomes with cetuximab plus CMT. LRF rates were lower compared with historical data using similar definitions. The findings suggest that patients with stage I-III HIV-associated SCCAC should be treated with curative intent similar to immunocompetent patients, and that addition of cetuximab to CMT may reduce LRF. Clinical trial information: 00324415, 00316888. No. Node positive 3 year LRF Rate 3 year PFS Rate 3 year OS Rate

E3205

AMC 045

61 54% 23% (95% CI 13%, 36%) P under H0 = 0.03 68% (95% CI 55%, 79%) 83% (95% CI 71%, 91%)

45 47% 16% (95% CI 7%, 30%) P under H0 = 0.003 82% (95% CI 66%, 91%) 79% (95% CI 63%, 89%)

Poster Discussion Session; Displayed in Poster Session (Board #220), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 1:15 PM-2:30 PM

Salvage surgery with abdominoperineal excision of the rectum (APER) following loco-regional failure after chemoradiation (CRT) using mitomycin (MMC) or cisplatin (CisP), with or without maintenance 5FU/CisP chemotherapy (CT) in squamous cell carcinoma of the anus (SCCA) and the impact on long-term outcomes: Results of ACT II. First Author: Robert Glynne-Jones, Mount Vernon Cancer Centre, Middlesex, United Kingdom Background: Concurrent CRT is standard treatment for patients with SCCA. We explored the results of surgical salvage with APER in ACT II, which compared 5FU/CisP with 5FU/ MMC concurrent with a uniform RT dose (50.4Gy, 28 daily fractions of 1.8Gy), and a second randomization to maintenance chemotherapy. Methods: The ACT II trial recruited 940 patients. We examined the overall survival of patients from the time of salvage surgery post CRT according to 4 time periods: Group A - operated # 6 months of CRT (n = 19); Group B - between . 6 and # 12 months (n = 36); Group C - between . 12 and # 24 months (n = 28); Group D - after . 24 months (n = 18) respectively. Results: There were 291 disease failures (31%) overall - males, total radiation dose, and tumor stage were independent predictors of local failure. Of 291 relapses, surgical salvage with APER was attempted in 107 (37%) loco-regional failures of whom 101 were evaluable and 53 subsequently died. The 2-year OS rates after local disease failure from time of salvage APER to death from any cause was 54% (43%-63%). Rates of OS beyond 2 years are unreliable given the small number of patients at risk after 2 years. Death rate after salvage APER was 15/19(79%), 21/36(58%), 12/28(43%) and 5/18(28%) in Groups A, B, C and D. Conclusions: In the management of SCCA, local failure may benefit from early detection and salvage by radical surgery. Results of radical surgery performed in ACT II in 37% compare with 56% in ACT I. These observations suggest close imaging with MRI and clinical surveillance may be helpful in the first 2 years, and imply the need for a trial to test further systemic interventions such as chemotherapy peri-operatively. Clinical trial information: 26715889. Time from salvage surgery until death Median OS (months), IQR Hazard Ratio (HR), (95% CI) p-value

Group A N = 19

Group B N = 36

Group C N = 28

Group D N = 18

Overall N = 101

9.6 (5.8 to 26.3) 21.1 (11.7 to 118.1) 47.7 (15.7 to NR) NR 30.3 (11.7 to 118.1) 1.00 (baseline) 0.53 (0.27 to 1.03) 0.33 (0.15 to 0.70) 0.31 (0.11 to 0.85) HR, (95% CI) 0.062 0.004 0.024

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Colorectal) Cancer 3524

Poster Session (Board #221), Sat, 8:00 AM-11:30 AM

Regorafenib in previously treated metastatic colorectal cancer (mCRC): Analysis of age subgroups in the open-label phase IIIb CONSIGN trial. First Author: Eric Van Cutsem, University Hospitals Leuven, Leuven, Belgium Background: In the phase 3 CORRECT trial, regorafenib improved survival vs placebo in treatment-refractory mCRC. The large, single-arm, phase 3b CONSIGN trial (NCT01538680) was designed to provide continued access to regorafenib for patients with mCRC and to further characterize regorafenib safety. In CONSIGN, adverse events (AEs) and progression-free survival (PFS) were consistent with results reported in phase 3 trials. We performed a subgroup analysis of CONSIGN to assess outcomes by age. Methods: Patients with mCRC who progressed after standard therapies and had ECOG PS 0-1 received regorafenib 160 mg daily for the first 3 weeks of each 4-week cycle until disease progression, death, or unacceptable toxicity. The primary endpoint was safety. PFS, assessed per investigator, was the only efficacy measure collected. Results: A total of 2872 patients were assigned to treatment (,65: n=1720; $65: n=1152); n=1713 and n=1151, respectively, were evaluable for safety. ECOG PS 0/1 were 49%/51% (,65) and 44%/55% ($65). The median (range) treatment duration was 2.5 months (0.03–30.4) and 2.3 months (0.03–28.5) in patients ,65 and $65, respectively; mean (SD) percent of planned dose was 76% (20) and 74% (20), respectively. Most patients experienced a regorafenib-related AE (,65: 91%; $65: 92%; Table). AEs led to treatment discontinuation in 25% (,65) and 26% ($65) of patients. Treatmentemergent NCI-CTCAE v4.0 hepatic grade $3 laboratory toxicities (,65; $65) included increased bilirubin (14%; 12%), increased AST (8%; 5%), and increased ALT (7%; 4%). Estimated median PFS (95% CI) in the ,65 group was 2.7 months (2.6, 2.8) and in the $65 group was 2.6 months (2.5, 2.7). Conclusions: The safety profile of regorafenib was generally comparable in patients ,65 and $65 years of age. PFS was similar across age subgroups. Clinical trial information: NCT01538680. Drug-related AEs, n (%)

6-12

> 12-18

> 18-24

> 24-30

> 30-36

> 36

No at risk (OS) in Arm (A/B) time-wise hazard ratio of OS (95% CI)

297/295

272/273

229/230

177/176

125/127

95/77

69/55

1.23 0.89 0.77 1.01 0.48 0.51 0.78 (0.65-2.32) (0.54-1.49) (0.48-1.21) (0.64-1.61) (0.27-0.85) (0.22-1.17) (0.47-1.28)

Poster Session (Board #227), Sat, 8:00 AM-11:30 AM

Effect of pharmacogenetic-based selection of first-line chemotherapy on response rate and R0 surgery in metastatic CRC patients. First Author: Albert Abad, Hosp Univ Germans Trias I Pujol, Barcelona, Spain Background: In a previous work we demonstrated that pts harboringTYMS3’UTR+6 bp/+6 bp and ERCC1-118 C/T or C/C genotypes may better receive capecitabine instead of 5FU in an OXA-based first-line treatment. The objective of the present work was to demonstrate that prospective selection of first-line treatment according to these genetic variants could improve the outcome of metastatic CRC pts. Methods: We designed a phase II comparative clinical trial (SETICC, SElected Treatment In Colorectal Cancer; NCT01071655) in which 195 CRC pts with initially unresectable metastasis were randomized 2:1 to receive XELOX plus bevacizumab in a control arm or, based on our previous results, chemotherapy according to their TYMS-3’UTR and ERCC1-118 genotypes plus bevacizumab in in the experimental arm (table 1). Results: In 161 evaluable pts the overall best objective response (Complete +Partial response) was 47% (95 CI: 33.9-61.1) in the control arm and 65% (95 CI: 55.2-73.9) in the experimental arm (p = 0.04). In the Per Protocol (PP) population (N = 180), 20% (N = 37) were converted to resectable, 16 pts in the control arm and 21 in the experimental arm. The R0 surgery was 43% (N = 7) in the control arm and 86% (N = 18) in the experimental arm (p = 0.018). Median PFS (10 months; 95 CI: 8.5-11.6 vs 9.4 months; 95% CI: 6.7-12.1 months) and overall survival (19.3 months; 95% CI: 15.5-23.1 vs. 16.5 months; 95% CI: 13.5-19.5.) was better for pts in the experimental arm (p:NS). Conclusions: This study shows for the first time that prospective selection of first-line treatment according to TYMS3’UTR and ERCC1-118 polymorphisms significantly improves the ORR and R0 liver metastasis resection of CRC pts with non-initially resectable liver metastases. This approach should be taken into account in this kind of pts in order to increase the probability of R0 surgery success rate. Clinical trial information: NCT01071655. ITT Treatments’ assignment according to genotypes (Martinez-Balibrea et al, Eur J Cancer 2008). Genotypes Treatment XELOX XELIRI FUOX FUIRI

TYMS-3’UTR +6bp/+6bp +6bp/+6bp +6bp/-6bp or -6bp/-6bp +6bp/-6bp or -6bp/-6bp

ERCC1-118 T/T C/T or C/C T/T C/T or C/C

3531

Poster Session (Board #228), Sat, 8:00 AM-11:30 AM

Final results of PRODIGE 9, a randomized phase III comparing no treatment to bevacizumab maintenance during chemotherapy-free intervals in metastatic colorectal cancer. First Author: Thomas Aparicio, Hopital ˆ Avicenne, AP-HP, Bobigny, France Background: Conflicting results are reported for maintenance treatment with bevacizumab (bev) during chemotherapy free intervals (CFI) in metastatic colorectal cancer (mCRC). Methods: The objective was to compare the tumor control duration (TCD) by either bev maintenance (Arm A) or no treatment during CFI (Arm B) after induction chemotherapy (CT) (12 cycles of FOLFIRI + bev). CT was reintroduced at progression (8 cycles) and then a new CFI. The randomization was performed before induction CT. TCD was defined by the time between randomization and tumor progression during a CT sequence (Aparicio T et al, Dig Liver Dis, 2015). Per Protocol (PP) population was defined as patients (pts) with at least one CT reintroduction after first progression during CFI. We present the final analysis of the trial. Results: From March 2010 to July 2013, 491 pts were randomized. The median age was 64.6 years [range: 27 - 89], 64% of the pts were men, 93% of the pts were OMS 0-1 and 218 (44%) pts had a non-resected primary tumor. A progression during induction CT happen in 85 (17%) pts, 261 (53%) pts had at least one reintroduction, 107 (22%) pts had two and 51 (10%) pts had three or more re-introductions. Multivariate analysis in all pts revealed that WHO performance status $ 2, unresected primary tumor and BRAF mutation were associated with a shorter TCD. Unresected primary tumor and BRAF mutation were associated with a shorter overall survival (OS). Grade 3-4 toxicities were observed in 80% of the pts in arm A and 79% in arm B. Conclusions: The alternate irinotecan-based CT sequences with or without bev maintenance during CFI, revealed an impressive TCD. Bev maintenance monotherapy did not improved TCD, progression free survival (PFS) or OS. Clinical trial information: NCT00952029. Efficacy criteria.

All patients TCD PFS OS PP Population TCD PFS OS

Arm A Maintenance Median in months

Arm B No Maintenance Median in months

N = 246 15.08 9.20 21.65 N = 124 (50%) 17.77 9.86 27.47

N = 245 14.98 8.90 21.98 N = 137 (56%) 23.26 9.49 28.58

HR [95% CI]; p-value 1.09 [0.87 - 1.37]; p = 0.43 0.92 [0.76,1.10]; p = 0.34 1.05 [0.86,1.28]; p = 0.65 1.18 [0.87 ; 1.59]; p = 0.29 0.89 [0.69 ; 1.13]; p = 0.33 1.09 [0.82 ; 1.45]; p = 0.56

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Colorectal) Cancer 3532

Poster Session (Board #229), Sat, 8:00 AM-11:30 AM

Obesity in metastatic colorectal cancer: Pooled analysis of FFCD trials. First Author: Thomas Aparicio, Hopital ˆ Avicenne, AP-HP, Bobigny, France Background: Previous studies showed that high and low body mass index (BMI) are associated with worst prognosis in early-stage colorectal cancer (CRC) and low BMI is associated with a worst prognostic in metastatic CRC (mCRC). We aimed to assess obesity according to time, efficacy outcomes according to BMI, dose reduction in obese patients (pts). Methods: A pooled analysis of individual data from 2085pts enrolled into 8 FFCD first line mCRC trials from 1991 to 2013 was performed. Obese pts (BMI $ 30) were compared to normal or overweight pts (BMI: 18.5-29). Skinny pts (BMI , 18.5) were compared to pts with BMI . 18.5. Analyses were stratified by trials, overall pooled hazard ratios (HR and 95% CI) were calculated using a fixed effect model. Cochrane Q tests were used to study heterogeneity between trials: for all analyses trials were homogenous. Medians of survivals were calculated with the Kaplan-Meier method. Interaction tests were performed between obese effect and sex, age (# 70 vs . 70 years) and chemotherapy with antiangiogenics. Results: Twelve percent of the enrolled pts were obese (7.7% for 1991-1999, 14.5% for 2000-2006, 13.6% for 2007-2013). Comparison of obese pts vs pts with BMI 18.5-29 revealed a median overall survival (OS) of 19.5 vs 16.9 months(m), HR = 1.15 [1.001.33] p = 0.06, a median progression free survival (PFS) of 7.9 vs 7.3m, HR = 1.06 [0.93-1.22] p = 0.36, and an objective response rate (RR) of 42% vs 38.5%, OR = 1.02 [0.77-1.37] p = 0.87. A significant OS and RR improvement were observed in pts with BMI . 25 compared to pts with BMI 18.5-24: 18.5 vs 16.3m, HR = 1.13 [1.02-1.25] p = 0.02, and 42% vs 36% OR = 1.23 [1.01-1.50] p = 0.04 respectively. Subgroup analysis revealed that obesity was associated with better OS and PFS in men (p = 0.03, 0.05 respectively). OS and PFS were significantly lower in skinny pts HR = 0.75 [0.60-0.94] p = 0.01, and HR = 0.75 [0.60-0.93] p = 0.01 respectively. Dose reduction at the first cycle was more frequent in pts with BMI $ 30 vs BMI 18.5-29: 34% vs 17% p , 0.001 (especially in obese pts with bsa . 2m2). Conclusions: Overweight and obesity are significantly associated with better OS compared to normal weight in mCRC, mainly observed in men. The pejorative prognostic of a BMI , 18.5 is confirmed. A dose reduction is frequent in obese pts.

3533

179s Poster Session (Board #230), Sat, 8:00 AM-11:30 AM

Predictive validity of NeoAdjuvant Rectal (NAR) Score and pathologic complete response (ypCR) for overall survival (OS) as surrogate endpoints in rectal cancer clinical trial. First Author: Greg Yothers, NRG Oncology, and The University of Pittsburgh, Pittsburgh, PA Background: ypCR (absence of primary tumor and nodal metastases) and more recently NAR score (weighted combination of ypN and downstaging of T) have been used as primary endpoints in neoadjuvant clinical trials for rectal cancer. Unlike OS, both surrogate endpoints are available shortly after surgery. While both have demonstrated patient-level association with OS, neither has been tested for trial-level association. Methods: Mean difference in NAR score and log odds ratios (ORs) for ypCR were tested for linear association with log hazard ratios (HRs) for OS. Treatment (trt) comparisons were weighted by the inverse of the SE of the log HR for OS, giving precise estimates more weight. Steiger’s Z test was used to compare predictive validity. NAR = ((5 ypN – 3 (cT – ypT) + 12)^2) / 9.61. Results: Data from 6 randomized trt comparisons from 5 international clinical trials of contemporary neoadjuvant therapy for rectal cancer were analyzed in this metaanalysis including 4705 patients and 931 deaths overall (Table). For all 6 comparisons, point estimates for both NAR and ypCR were concordant with OS (same direction of trt effect). NAR (R-squared = 0.1292) had numerically greater predictive validity for OS than did ypCR (R-squared = 0.0169), (p = 0.38). Conclusions: NAR has nominally greater predictive validity for OS than does ypCR. Efforts to identify or improve short term surrogate endpoints for OS should continue. Support: CA-180868, -189867, -180822; Sanofi; Roche. Trial

Control

Experimental

NSABP R-04

5-FU

Capecitabine

NSABP R-04

No Oxaliplatin

Oxaliplatin

RT

ChemoRT

GCR3

FFCD 9203

ChemoRT

Chemo a` ChemoRT

Polish

Std RT

Short Course RT

RT

ChemoRT

EORTC 22921

Patients (Deaths)

OS HR (95%CI)

NAR Mean Dif (95%CI)

ypCR OR (95%CI)

1481 (248) 1215 (193) 687 (125) 95 (24) 286 (85) 941 (256)

0.83 (0.65, 1.07) 0.92 (0.69, 1.22) 0.98 (0.76, 1.28) 1.46 (0.64, 3.33) 1.02 (0.67, 1.57) 0.98 (0.82, 1.18)

-1.13 (-2.59, 0.34) -0.33 (-1.94, 1.29) -1.57 (-3.64, 0.49) 4.16 (-0.79, 9.12) 7.39 (3.86, 10.92) -4.80 (-6.53, -3.07)

1.20 (0.93, 1.54) 1.15 (0.87, 1.53) 3.77 (1.84, 7.73) 0.94 (0.31, 2.86) 0.04 (0.01, 0.31) 3.00 (1.79, 5.03)

RT = Radiotherapy; Chemo = Chemotherapy; Std = Standard

3534

Poster Session (Board #231), Sat, 8:00 AM-11:30 AM

3536

Poster Session (Board #233), Sat, 8:00 AM-11:30 AM

Impact of genetic polymorphisms of VEGF pathway on the response to bevacizumab in metastatic colorectal cancer (mCRC): Ancillary study of PRODIGE 9 trial. First Author: Camille Sibertin-Blanc, Department of Digestive Oncology. Aix-Marseille University - Assistance Publique Hopitaux ˆ de Marseille ; (2)UMR S-910 INSERM, Marseille, France

Final results from a phase III trial evaluating panitumumab (pmab) + best supportive care (BSC) vs BSC in chemorefractory wild-type (WT) KRAS exon 2 and WT RAS metastatic colorectal cancer (mCRC). First Author: Tae Won Kim, Department of Oncology, Asan Medical Center, University of Ulsan College of Medicine, Seoul, South Korea

Background: Bevacizumab (Bev) is widely used in mCRC but there is no validated predictive factor. Several studies have reported the impact of SNPs in the Vascular Endothelial Growth Factor-A (VEGF-A) pathway genes on the Bev’s efficacy but results in the mCRC are yet inconstant. PRODIGE 9 is a phase III trial that compare the tumor control duration (TCD) under a 1st-line chemotherapy followed by either Bev maintenance (Arm A) or no treatment during chemotherapy-free intervals (CFI) (Arm B).The aim of our study was to evaluate the impact of genetic polymorphisms of VEGF A, VEGFR1, VEGFR2 and HIF1a on TCD, duration of 1stCFI, progression-free survival (PFS) and overall survival (OS). Methods: We included patients (pts) of the PRODIGE 9 study for whom biological consent was signed and a DNA sample was available. Ten polymorphisms were genotyped on germ line DNA using real-time Polymerase Chain Reaction. DNA was extracted from blood samples. Survival analyses were done using Kaplan-Meier method and Cox model. For tRR, a logistic regression was used. Results: Among the 491 pts randomized in PRODIGE 9, DNA was available for 314 pts: 160 in Arm A and 154 in Arm B . The median TCD, OS and PFS, were respectively 15.7, 22.6 and 9.2 months (m). For VEGFR-1 rs9582036, pts with CC genotype had significantly better TCD (median [95%CI] 22.4 m [14.7- NR]) compared to pts with AA and CA genotypes (14.4 m [11.7-17.1] (p = 0.04)) in arm A whereas there was no difference in the other arm. This tendency was observed also for the 1st CFI (median 4.6 m for pts CC vs 4.1 m for AC and AA), for the median OS (28.2 vs 22.5 m), the RR, but it was not statistically significant (table). Conclusions: In this prospective study, VEGFR-1 rs9582036 variants were associated with clinical outcome in terms of TCD for patients with mCRC treated by Bev-based therapy. Clinical trial information: NCT00952029.

Background: In the primary analysis of study 20100007, pmab + BSC significantly improved OS and PFS vs BSC in WT KRAS exon 2 mCRC and, in the first prospective phase 3 study, WT RAS (exons 2, 3, 4 of KRAS and NRAS) mCRC. We report the final analysis and evaluate BRAF status from study 20100007. Methods: Anti-EGFR na¨ıve patients (pts) with WT KRAS exon 2 mCRC were randomized 1:1 to pmab + BSC or BSC. On-study crossover was prohibited. KRAS exon 2 and RAS status were assessed centrally. The primary endpoint was OS in WT KRAS exon 2 mCRC; secondary endpoints were OS in WT RAS mCRC and PFS and safety in both WT populations. Pts were followed for survival for $2 years after the last pt was randomized and a final analysis conducted. Results: 377 pts with WT KRAS exon 2 mCRC were enrolled; 84% had died at the final analysis (72% in the primary analysis). RAS ascertainment was 86%, BRAF was 84%. Pmab + BSC significantly improved OS and PFS vs BSC in WT KRAS exon 2 and WT RAS mCRC (See Table). No OS benefit was seen in mutant (MT) RAS mCRC with pmab (HR=1.09; 95% CI=0.56–2.16; P=0.80). BRAF mutations (descriptive analysis) were associated with poor prognosis (BSC; WT BRAF [n=142] vs MT BRAF [n=11], HR=0.33; 95% CI=0.1620.65). No new toxicities were seen. Conclusions: Pmab significantly improved OS and PFS in chemorefractory WT KRAS exon 2 and WT RAS mCRC vs BSC, validating the importance of RAS testing at diagnosis. BRAF mutations appeared to be prognostic. Clinical trial information: NCT01412957.

median (m) [95%CI] TCD TCD OS 1st CFI *Not reached

Arm A

Arm B

15.4 [12.5 ;17.4] 17.3 [13.4 ; 23.3]; p = 0.11 CC CA+AA p CC CA+AA p N = 14 N = 146 N = 16 N = 138 22.4 [14.75- ] 14.4 [11.7-17.1] 0.04 NR [9.5- ] 15.7 [12.7-21.9] 0.08 28.2 [18.1-42.8] 22.5 [18.6-24.6] 0.5 22.66 [17.2- ] 22.1 [19.5-25.2] 0.28 4.6 [1.6-13.3] 4.14 [0.5-29.0] 0.30 3.15 [1.6-18.2] 4 [0.5-22.2] 0.49

WT KRAS exon 2, n Median OS, mo (95% CI) Median PFS, mo (95% WT RAS, n Median OS, mo (95% CI) Median PFS, mo (95% WT RAS/WT BRAF, n Median OS, mo (95% CI) Median PFS, mo (95% WT RAS/MT BRAF, n Median OS, mo (95% CI) Median PFS, mo (95%

Pmab + BSC

BSC

HR (95% CI)

189 10.0 (8.7–11.3)

188 7.4 (5.8–9.3)

P Value

0.74 (0.59–0.93)

0.009

CI)

3.6 (3.4–5.3) 142 10.0 (8.7–11.6)

1.7 (1.6–1.9) 128 6.9 (5.2–7.9)

0.54 (0.43–0.67)

,0.001

0.72 (0.55–0.94)

0.015

CI)

5.2 (3.5–5.3) 128 10.2 (8.7-11.7)

1.7 (1.6–2.2) 114 7.4 (5.7-10.0)

0.45 (0.35–0.59)

,0.001

0.75 (0.57-0.99)

0.044

CI)

5.3 (3.6-5.4) 9 4.1 (3.8-13.9)

1.8 (1.6-2.6) 11 3.0 (1.3-4.1)

0.45 (0.34-0.60)

,0.001

0.39 (0.10-1.51)

0.160

CI)

1.5 (0.8-3.7)

1.3 (0.9-1.8)

0.28 (0.07-1.08)

0.050

HR = hazard ratio

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

180s 3537

Gastrointestinal (Colorectal) Cancer Poster Session (Board #234), Sat, 8:00 AM-11:30 AM

A multi-centric randomized phase II trial evaluating dual targeting of the EGFR with cetuximab and afatinib versus cetuximab alone in patients with chemotherapy refractory wtKRAS metastatic colorectal cancer (UCGI 25: A UNICANCER trial). First Author: Helene Senellart, Institut de Cancerologie de l’Ouest, Nantes, France Background: Promising results using cetuximab and erlotinib to treat wtKRAS metastatic colorectal cancer (mCRC) were reported by Weickhardt (J Clin Oncol 2012; 30:1505-12). Combining a monoclonal antibody and a tyrosine kinase inhibitor may be a treatment option for mCRC patients (pts). We investigated dual targeting with cetuximab and afatinib in chemotherapy refractory wtKRAS mCRC. Methods: wtKRAS mCRC pts (ECOG # 1) who failed chemotherapy with irinotecan and oxaliplatin were eligible. Due to low accrual and the frequent use of an anti-EGFR antibody in the first- or secondline, the study was amended, after 19 inclusions, to allow pre-treatment with anti-EGFRs. This multicenter, prospective, open phase II trial randomized pts (2:1) to cetuximab (500mg/m2/two weeks) and oral afatinib (40mg/day) (Arm A) or cetuximab alone (500mg/m2/two weeks) (Arm B). Treatment continued until progression (evaluated every six weeks using RECIST 1.1) or limiting toxicity. At progression (Arm B) afatinib could be added to cetuximab. The primary objective was the non-progression rate at six months (Fleming’s one-step design, one-sided a = 5%, b = 10%, H0: 30%; H1: 50%; 75 pts needed, 49 in Arm A). Results: 75 pts were included between November 2012 and May 2015 in 14 French sites. We randomized 51 pts in Arm A and 24 in Arm B. The non-progression rate (CI95%) at six months was 17.0% in Arm A [7.7-30.8] and 20.8% in Arm B [7.1-42.2]. Median PFS (CI95%) and OS (CI95%) were 4.1 months [2.7; 5.4] and 13.6 months [7.7; 19.9] in Arm A; and 2.7 months [1.4; 4.7] and 13.4 months [4.7; 20.4] in Arm B. The objective response was 26.0% in Arm A and 8.3% in Arm B. 38 pts (76%) in Arm A vs 14 (58%) in Arm B had AEs of grade 3 or 4. Conclusions: ORR results are encouraging with cetuximab and afatinib. The median PFS has to be interpreted considering the criterion allowing pretreatment with anti-EGFRs. Cross-over in Arm A could explain the similar median OS in both arms. Additional data will be presented (NRAS, subgroup of pts not pre-treated with anti-EGFRs, pts treated with cross-over). Clinical trial information: NCT01919879.

3538

Poster Session (Board #235), Sat, 8:00 AM-11:30 AM

Efficacy of panitumumab vs cetuximab in patients with wild-type KRAS exon 2 metastatic colorectal cancer treated with prior bevacizumab: Results from ASPECCT. First Author: Marc Peeters, Antwerp University Hospital, Edegem, Belgium Background: The phase 3 ASPECCT trial in patients (pts) with chemorefractory wild type (WT) KRAS exon 2 metastatic colorectal cancer (mCRC) demonstrated that panitumumab (pmab) was noninferior to cetuximab (cmab) for overall survival (OS). A previous subgroup analysis of hazard ratios (HRs) suggested that pts who had received prior bevacizumab (bev; any line, at any point before study start) in the pmab arm may have had better outcomes vs pts in the cmab arm (Price et al. Lancet Oncol 2014;15:569). Methods: Pts were randomized 1:1 to receive pmab or cmab. The subset of pts who had received prior bev were analyzed based on the final analysis of ASPECCT. Results: 999 pts were randomized and treated: 499 pmab and 500 cmab. The prior bev subset included 126 pts in the pmab arm (25%) and 132 pts in the cmab arm (26%). Pts in the pmab arm had longer median OS and progression-free survival (PFS) and higher objective response rates (ORR) compared with pts in the cmab arm. Results are shown (table). After adjustment for baseline covariates including ECOGperformance status, number of metastatic sites, and baseline lactate dehydrogenase (LDH), OS hazard ratio (HR) was 0.65 (95%CI=0.49-0.85) with pmab vs cmab in pts who had received prior bev. Pts in the pmab and cmab arms who did not receive prior bev had similar OS, PFS, and ORR. Post-progression antitumor therapy was similar between the pmab (47%) and cmab arms (52%) in pts who received prior bev. Conclusions: In ASPECCT, pts with WT KRAS exon 2 mCRC who received prior bev-containing regimens may have derived greater benefit with pmab versus cmab monotherapy. Clinical trial information: NCT01001377. Prior Bev Treatment Pmab (N = 126) OS, events (%) Median (95% CI), mos Hazard ratio (95% CI) PFS, events (%) Median (95% CI), mos Hazard ratio (95% CI) ORR, % (95% CI)*

Cmab (N = 132)

108 (85.7) 122 (92.4) 11.3 (9.3–14.2) 9.8 (8.3–11.7) 0.73 (0.56–0.96) 124 (98.4) 130 (98.5) 4.7 (3.2–4.9) 3.2 (3.0-4.8) 0.84 (0.66–1.08) 22.3 (15.3–30.8) 15.0 (9.3–22.4)

No Prior Bev Treatment Pmab (N = 373)

Cmab (N = 368)

338 (90.6) 334 (90.8) 10.0 (9.1–11.2) 9.9 (9.0–11.1) 1.02 (0.87–1.19) 362 (97.1) 360 (97.8) 3.8 (3.2–4.8) 4.7 (3.3–4.8) 1.03 (0.89–1.19) 21.9 (17.8–26.5) 21.5 (17.4–26.1)

*Evaluable pts.

3539

Poster Session (Board #236), Sat, 8:00 AM-11:30 AM

3540

Poster Session (Board #237), Sat, 8:00 AM-11:30 AM

Non-randomized phase II study to assess the efficacy of pembrolizumab (Pem) plus radiotherapy (RT) or ablation in mismatch repair proficient (pMMR) metastatic colorectal cancer (mCRC) patients. First Author: Neil Howard Segal, Memorial Sloan Kettering Cancer Center, New York, NY

Females versus males: Clinical features and outcome differences in large molecularly selected cohort of mCRC patients. First Author: Marta Schirripa, Division of Medical Oncology, USC Norris Comprehensive Cancer Center, Los Angeles, CA

Background: Pem alone is not active in pMMR mCRC. RT or tumor ablation alone does not result in objective systemic benefit. However, these local modalities can induce systemic immunity (abscopal effect) that can be augmented by checkpoint blockade in clinical and preclinical models. The PD-1 pathway is an important immune checkpoint to inhibit antitumor responses. A Phase II study to evaluate the safety and abscopal effect of Pem after RT or ablation in pts with pMMR mCRC is ongoing (NCT02437071). Methods: Pts with histologically/cytologically confirmed, unresectable/ recurrent mCRC, who have received $ 2 standard therapies, underwent palliative RT (cohort 1) or ablation (cohort 2) and received 200 mg pembrolizumab Q3W for up to 24 months. The primary objective is Overall Response Rate (ORR) according to RECIST v1.1 in a non-targeted lesion. Nine pts are enrolled in stage 1 in each cohort; 15 additional pts are enrolled in stage 2 following 1 observed response in a Simon two-stage design. Correlative studies include PDL-1 status and immune correlates.Results: As of 14 Jan 2016, 26 pts [mean age 55 (range 18-80); 62% male; ECOG 0/1: (12%/88%)] received a median of 4 doses of Pem (range 1-10). While drugrelated AEs were reported in 73% of pts, all were either grade 1 or 2; the most frequent were fatigue (23%), rash (15%), and nausea (15%). 19 pts were evaluable for response. Interim ORR is 9% (1 of 11) in the RT cohort. No responses were observed in the ablation cohort. Updated data will be presented.Conclusions: Pem after RT or ablation is feasible with a tolerable safety profile, with one patient achieving a PR in non-irradiated lesions after RT (abscopal effect). Accrual to the RT + Pem arm continues. Clinical trial information: NCT02437071.

Background: The impact of gender differences in terms of OS and of association with clinical and molecular characteristics remains unclear. Previous analyses were limited by small sample size, significant heterogeneity in stage and poor data. We planned the present analysis to determine the prognostic impact of gender and possible associated clinical features in a molecularly selected mCRC pts population. Methods: Metastatic CRC pts referred to UO Oncologia Medica AOUP from 2000 to 2014 were selected for the inclusion in this analysis based on BRAF and/or RAS mut status availability . The following characteristics were collected: age, ECOG-PS, grade, time to metastases, primary tumour site, resection of primary tumour, pT, pN status, mucinous histology, number and sites of mts (peritoneum, lung, or distant lymph nodes), at the time of diagnosis. Chi-square tests were used to compare clinical characteristics between males (M) and females (F). Stepwise Cox regression analysis was performed to identify the prognostic factors affecting OS in m and F separately. The impact of each significant prognostic factor in affecting outcome according to gender was estimated by interaction test. Results: 1045 pts were included. Main pts’ characteristics were the following: F/M = 41/59%; median age = 63 (range 24-92); ECOGPS 0/1-2 = 76/24%; synchronous/metachronous = 71/29%; RAS-BRAF mutational status (wt-wt/mut-wt/wt-mut) = 37/54/9%. F presented more frequently with younger age ( # 65) (p , 0.001), peritoneal metastasis (p = 0.009), BRAF mutant tumor (p = 0.033) and locoregional node involvement (p = 0.040) compared to M. Among all RAS wild-type, F had worse OS compared to m (median OS 31.7 vs 39.5 months, HR = 1.33 [95%CI: 1.04 1.71] p for interaction = 0.011). This outcome difference was confirmed after adjustment for BRAF status (HR = 1.22 [95%CI:0.95-1.57], p for interaction = 0.044), but not in the multivariate model (HR = 0.99 [95%: 0.76-1.28], p for interaction = 0.90). Conclusions: In the present analyses we confirmed an association of female gender with negative prognostic factors such as BRAFmutation, peritoneal metastases, N status. No significant differences were observed in terms of outcome.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Colorectal) Cancer 3541

Poster Session (Board #238), Sat, 8:00 AM-11:30 AM

Association of the number of natural killer cells and B-cell area ratios in lymph nodes with survival in patients with stage II colon cancer. First Author: Kazutake Okada, Tokai University, Kanagawa, Japan Background: Patients with many retrieved lymph nodes (LNs) have good outcomes in Stage II colon cancer. We previously reported that the longest diameter of retrieved LNs correlates with the number of LNs and can be used as a prognostic factor (Int J Colorectal Dis 2015). LNs consist mainly of T cells, B cells, NK cells, and histiocytes. We studied the relations of the number of NK cells and the T-cell and B-cell area ratios in longest-diameter LNs to the number of retrieved LNs and outcomes. Methods: The subjects were 320 patients with stage II colon cancer. One LN with the maximum long-axis diameter was selected, and was immunostained with CD56 as an NK-cell marker, CD3 as a T-cell marker, and CD20 as a B-cell marker. The number of CD56-positive cells were examined and the mean number per 0.093 mm2 was calculated by 6 power fields. CD3 and CD20 positive area ratio was determined by image analysis. The relations were evaluated with the use of Spearman’s rank correlation coefficient. The cut-off values were calculated by ROC curve analysis. Results: The number of LNs was 14.8 6 10.1 (mean 6 SD). The number of CD56-positive cells was 10.7 6 9.6. The CD3-positive area ratio was 0.39 6 0.08. The CD20-positive area ratio was 0.42 6 0.10. The correlation coefficients for the number of LNs with the number of CD56-positive cells, CD3-positive area ratio and CD20-positive area ratio were 0.30 (p , 0.001), -0.03 (p = 0.569) and 0.10 (p = 0.072), respectively. The number of LNs positively correlated with the number of CD56-positive cells. The median follow-up period was 118 months. Patients with 12 or more CD56-positive cells (n = 97) had a significantly better survival (HR, 0.64; 95%CI, 0.41-0.97; p = 0.040). Patients with a CD20positive area ratio of 0.43 or greater (n = 160) had a significantly better survival (HR, 0.64; 95%CI, 0.44-0.92; p = 0.015). Multivariate analysis revealed the number of NK cells is an independent prognostic factor for relapse-free survival. Conclusions: A high number of NK cells in longestdiameter LN was associated with a higher number of retrieved LNs and better survival. NK cells in the longest LNs may reflect the host antitumor immune response in patients with Stage II colon cancer.

3543

Poster Session (Board #240), Sat, 8:00 AM-11:30 AM

Modified FOLFOXIRI (mFOLFOXIRI) plus cetuximab (cet), followed by cet or bevacizumab (bev) maintenance, in RAS/BRAF wild-type (wt) metastatic colorectal cancer (mCRC): Results of the phase II randomized MACBETH trial by GONO. First Author: Carlotta Antoniotti, U.O. Oncologia Medica 2, Azienda Ospedaliero-Universitaria Pisana, Pisa, Italy Background: The combination of triple chemotherapy regimens with an antiEGFR reported promising activity with some safety concerns in phase II trials. MACBETH trial aimed at evaluating the activity and safety of 1st-line mFOLFOXIRI plus cet and at exploring the role of maintenance with cet or bev in unresectable mCRC patients (pts). The study initially enrolled KRAS wt pts and, after an amendment in Oct 2013, only RAS/BRAF wt pts. Methods: Eligibility criteria included measurable, unresectable, RAS/BRAF wt disease (centrally screened), age 18-75 yrs, no prior therapy for advanced disease. Pts were randomized to receive up to 8 cycles of mFOLFOXIRI + cet (cet 500 mg/sqm, iri 130 mg/sqm, oxa 85 mg/sqm, I-LV 200 mg/sqm, 5FU 2400 mg/sqm over 48h q2w), followed by cet (arm A) or bev (arm B) until progression. The trial was a phase II randomized non-comparative study, with primary endpoint 10 months-Progression Free Rate (10m-PFR). Results: Between Nov 2011 and Feb 2015, 323 pts from 21 Italian centers were screened. Out of 143 randomized pts, 116 were RAS/BRAF wt and were therefore included in the modified ITT (mITT) population (arm A/B N = 59/57). Pts’ characteristics were: median age 60 yrs, ECOG PS 0 89%, synchronous metastases 84%, liver-only disease 45%. At a median follow up of 25.5 months, 10m-PFR was 52% in arm A and 41% in arm B (primary endpoint not met), with median PFS of 11.2 and 9.3 months. Response rate was 76% in 106 evaluable pts and 70% (arm A/B 68%/72%) in the mITT population; only 2 pts per arm experienced PD as best response. Resection rate was 38% (arm A/B 46%/30%), 65% (arm A/B 71%/58%) in the liveronly subgroup. Main grade 3-4 toxicities during induction were neutropenia (33%), febrile neutropenia (3%), diarrhoea (19%), skin rash (18%), stomatitis (6%). Conclusions: Neither of the two arms met the primary endpoint. However, 4 months-induction with mFOLFOXIRI + cet was feasible and had relevant activity, leading to high conversion rate. This may positively affect OS results, still immature (an update will be presented). Clinical trial information: NCT02295930.

3542

181s Poster Session (Board #239), Sat, 8:00 AM-11:30 AM

SIRFLOX study: Novel approach to define depth of response (DpR) within a volumetric model in patients with metastatic colorectal cancer (mCRC). First Author: Volker Heinemann, Comprehensive Cancer Center, LudwigMaximilian-University of Munich, Munich, Germany Background: DpR has recently been proposed as a measure of efficacy that predicts the long-term treatment outcome for mCRC. DpR is a continuous measure that defines the nadir of tumor response, typically by the sum of the longest diameters of all target lesions compared to baseline. A novel volumetric model was applied to the independent blinded reader RECIST data from SIRFLOX, a multicenter RCT of 530 patients with non-resectable, liveronly or liver-dominant mCRC that compared first-line FOLFOX (6 bev) plus Y-90 resin microspheres [SIRT] to FOLFOX (6 bev) alone [Control]. Methods: Spherical tumor volume was estimated from the longest unidimensional length for # 5 target hepatic lesions (RECIST 1.0) in the SIRFLOX ITT population. Finite mixture modelling was used to assess baseline tumor distribution and identify the optimal cut-point for subpopulations within which any potential predictors of DpR could be described. Results: There was a 7% greater decrease in median liver tumor volume in SIRT vs. Control (-75% vs. -68%; P = 0.036). Deepest response (median time to nadir) occurred 60 days later in SIRT than Control (266 vs. 206 days; P , 0.001). A larger treatment effect and longer time to nadir were observed in patients with . 12% hepatic tumor burden (median DpR -90% in SIRT vs. -77% in Control, P = 0.003; median nadir 298 vs. 196 days, P , 0.001), compared to those with # 12% tumor burden (median DpR -93% vs. -94%, P = 0.763; median nadir 243.5 vs. 220 days, P = 0.152). PFS in the liver by competing risk analysis was significantly longer in patients with . 12% hepatic tumor burden receiving SIRT vs. Control (median 27.2 vs. 13.1 months, P = 0.003), whereas complete response [CR] for SIRT vs. Control was more common in patients with # 12% liver tumor burden (12.9% vs.3.5%; P = 0.001). Conclusions: The present study integrates baseline tumor volume into the analysis of volumetric DpR modelled from existing unidimensional target lesion assessment. In SIRFLOX, the addition of SIRT to standard chemotherapy significantly increased hepatic DpR. The impact of SIRT on PFS was greatest in patients with a baseline tumor burden . 12%, whereas the impact on CR rate was greater where tumor burden was , 12%. Clinical trial information: NCT00724503.

3544

Poster Session (Board #241), Sat, 8:00 AM-11:30 AM

Phase 2 results: Encorafenib (ENCO) and cetuximab (CETUX) with or without alpelisib (ALP) in patients with advanced BRAF-mutant colorectal cancer (BRAFm CRC). First Author: Josep Tabernero, Vall d’Hebron Institute of Oncology, Barcelona, Spain Background: BRAFm CRC carries a poor prognosis, with survival of 5– 6 months after failure of first-line therapy. Reactivation of epidermal growth factor receptor (EGFR) signaling with BRAF inhibition and uninhibited PI3K signaling may explain the limited efficacy of BRAF inhibitor monotherapy in patients (pts) with BRAFm CRC. This study investigated the BRAF inhibitor ENCO and the anti-EGFR antibody CETUX with or without the PI3Ka inhibitor ALP (BYL719) in pts with advanced BRAFm CRC. Methods: In this ongoing phase 1b/randomized phase 2 study (ClinicalTrials.gov: NCT01719380), phase 1b did not identify a maximum tolerated dose for either combination. Based on general tolerability of the triplet, the ENCO dose for phase 2 was chosen to be the same in both arms. In phase 2, pts with advanced BRAFm CRC failing $ 1 prior therapy were randomized 1:1 to a doublet (ENCO 200 mg PO QD and CETUX per label) or triplet (ENCO, CETUX, and ALP 300 mg PO QD). Progression-free survival (PFS) was the primary endpoint. Secondary endpoints included overall response rate (ORR) and overall survival (OS). Results: A total of 102 pts were randomized (triplet, n = 52; doublet, n = 50). Pts in each group had a median of 2 prior therapies. A planned PFS analysis comparing the triplet to the doublet after 73 events showed a HR (95% CI) of 0.69 (0.43–1.11; P= 0.064) with median PFS (95% CI) of 5.4 (4.1–7.2) mo and 4.2 (3.4–5.4) mo and confirmed ORR (95% CI) of 27% (16%–41%) and 22% (12%–36%), respectively. With 35 events, interim OS analysis (triplet vs doublet) showed a HR (95% CI) of 1.21 (0.61–2.39); median OS was 15.2 mo on the triplet and not reached on the doublet. Grade 3/4 adverse events (AEs), regardless of causality, were reported in 79% (triplet) and 58% (doublet) of pts. Grade 3/4 AEs in . 10% of pts in either arm were anemia (17% vs 6%), hyperglycemia (13% vs 2%), and increased lipase (8% vs 18%) for the triplet and doublet, respectively. Conclusions: Combinations of ENCO and CETUX with or without ALP showed promising clinical activity, including improved PFS and OS relative to historical data. Adding ALP to the doublet may add a PFS benefit as well as additional toxicity. Clinical trial information: NCT01719380.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

182s

Gastrointestinal (Colorectal) Cancer

3545

Poster Session (Board #242), Sat, 8:00 AM-11:30 AM

Sorafenib (Soraf) and irinotecan (Iri) combination for pretreated RASmutated metastatic colorectal cancer (mCRC) patients: A multicentre randomized phase II trial (NEXIRI 2-PRODIGE 27). First Author: Emmanuelle Samalin, Institut regional ´ du Cancer de Montpellier (ICM), Montpellier, France Background: Sorafenib and irinotecan combination (NEXIRI) showed promising efficacy with a 65% disease control rate (DCR) in pretreated mutated (mt) KRAS mCRC. In this single arm study, CCND1 rs9344 A/A polymorphism had a predictive value (Samalin et al. 2014).This multicentre randomized phase II trial aimed to determine the 2-month progression-free survival rate (2-PFS) of NEXIRI vsIri or Soraf monotherapies in these patients (pts) after failure of all approved active drugs at the time of the study. Methods: Pts PS#1 with progressive nonresectable mtKRAS (then RAS) mCRC pretreated with irinotecan, oxaliplatin, fluoropyrimidines and bevacizumab (none with regorafenib), were randomized in 3 arms: NEXIRI (biweekly Iri IV 120, 150, 180mg/m² at C3 combined with a fixed dose of Soraf 400mg twice daily) vs Iri (180mg/m²) alone vs Soraf alone, until progression or toxicity, with cross-over to NEXIRI at progression. Primary endpoint was the 2-PFS (RECIST v1.1). Pharmacokinetic, pharmacogenetics and pathologic translational studies were undertaken. Results: We included 173 pts (age 62 [31-82]; PS 0/1: 38/61%) between 2012 and 2014 in 17 French centres. Main results were (median follow-up 17.5 months): See table. Conclusions: We confirmed the NEXIRI regimen efficacy in a randomized study for refractory mtRAS mCRC pts. CCND1 rs9344 may identify patients who benefit from this combination. These results justify comparing NEXIRI to Regorafenib or TAS 102 monotherapies in CCND1 rs9344 A/A pts subgroup. Other results from ancillary studies will be presented at the meeting. Clinical trial information: NCT01715441.

Median treatment duration (months)* Grade 3/4 toxicities (%)* Neutropenia (febrile) Diarrhea Hand-foot syndrome Hypertension 2-PFS (%)$ DCR/PR (%)$ Cross-over to NEXIRI n(%) Median PFS (months) Median OS (months) CCND1 rs9344 genotype response A/A A/G or G/G

NEXIRI (n=57)

Iri (n=56)

3

2

Soraf (n=57) 2.5

16/2 (5) 26/0 17/0 10/0 59 [39-66] 59/4 3.7 [2.2-4.9] 7.0 [5.8-9.4] PD SD 1 11 14 15

6/0 (0) 7/0 0/0 2/0 23 [10-33] 25/0 42(75) 1.9 [1.7-2.1] 6.3 [4.8-8] PD SD 8 0 23 9

0/0 (0) 7/0 16/0 10/0 22 [8-30] 22/0 27(47) 2.1 [1.9-2.5] 5.1 [3.7-7.7] PD SD 9 0 22 6

3546

Poster Session (Board #243), Sat, 8:00 AM-11:30 AM

IND.210: A randomized phase 2 trial of FOLFOX6/bevacizumab with or without reovirus (Reo) in patients (pts) with metastatic colorectal cancer (mCRC). First Author: Patricia A. Tang, Tom Baker Cancer Centre, University of Calgary, Calgary, AB, Canada Background: Reo, a serotype 3 reovirus, may preferentially infect and destroy cells with RAS activation. This trial evaluated the progression-free survival (PFS) of untreated mCRC pts treated with FOLFOX6/bevacizumab (q2 weekly) alone or with Reo (3x1010TCID50 on day 1-5 [cycles 1, 2, 4, 6, 8 then alternate cycles]). Methods: The trial had 80% power to detect improvement in PFS from 8.5 to 15.5 mo [Hazard ratio (HR) 0.55, a = 0.1]. Primary analysis adjusted for stratification factors (prior adjuvant chemotherapy, KRAS status) with sensitivity analyses (SA) adjusted for additional prespecified factors (gender, age, baseline LDH and Alk Phos). Secondary endpoints included overall response rate (RR), quality of life (EORTC QLQ C30), overall survival (OS). Results: Following safety run-in of 6 pts, 103 pts were randomized to FOLFOX6/bevacizumab with (n = 51) or without (n = 52) Reo. Pt characteristics: median age 60 (range 31-79), male (61.2%), ECOG 0: 45.6%, 1: 50.5% and 2: 3.9%. At median follow-up 13 months, PFS was significantly shorter in the Reo arm [median 7.3 vs 9.1 mo, HR 1.59 (95% CI 1.00 - 2.53) p = 0.046; SA HR 1.65, p = 0.04]. RR was higher in Reo treated pts [52.9 vs 34.6%, OR 2.09 (95% CI 0.96-4.55), p = 0.06; SA p = 0.03], but with a shorter median duration of response (5.1 versus 8.5 mo, p = 0.028). In subgroup analysis female pts had higher RR and longer OS than male pts but test for interaction was significant only for RR. There was no significant difference in hepatectomy rate (Reo 7.8% vs 3.8%), median OS (p = 0.48), or relative dose intensity of FOLFOX6 or bevacizumab. Reo pts had worse QLQ C30 physical functioning (mean change score from baseline -3.9 vs. 1.94; p = 0.008). The most common adverse events (AEs) of any grade attributed to Reo were fatigue (51%), fever (49%), chills (44%), and anorexia (33%). With respect to grade 3+ AEs, pts on Reo experienced more hypertension (26.3 vs 3.8%, p = 0.001), more proteinuria (22.8 vs 1.9%, p = 0.001); and less abdominal pain (3.5 vs 17.3, p = 0.02) than the control arm. Conclusions: While the addition of Reo to FOLFOX6/bevacizumab increased RR; PFS, toxicity and QoL were significantly worse. Further follow-up of maturing OS data is planned. Clinical trial information: NCT01622543.

*170 pts $18 (6/4/8) non-evaluable pts

3547

Poster Session (Board #244), Sat, 8:00 AM-11:30 AM

3548

Poster Session (Board #245), Sat, 8:00 AM-11:30 AM

Efficacy and safety results in patients with impaired renal and hepatic function in the RECOURSE trial. First Author: Robert J. Mayer, DanaFarber Cancer Institute, Boston, MA

Phase Ib study of cabozantinib plus panitumumab in KRAS wild-type (WT) metastatic colorectal cancer (mCRC). First Author: John H. Strickler, Duke University Medical Center, Durham, NC

Background: TAS-102 contains an antineoplastic nucleoside, trifluridine, and a thymidine phosphorylase inhibitor, tipiracil, at a molar ratio of 1:0.5. The efficacy and safety of TAS-102 in patients (pts) with metastatic colorectal cancer refractory/intolerant to standard therapies were confirmed in the RECOURSE trial; enrollment criteria included $2 prior lines of standard chemotherapy. Results of RECOURSE showed significant improvement in overall survival (OS) in pts treated with TAS-102 compared with the placebo (pbo) group. Here we report results of RECOURSE in pts classified according to renal and hepatic function. Methods: RECOURSE patients’ kidney and liver functions were classified according to calculated creatinine clearance and National Cancer Institute criteria, respectively. Results: Significant OS benefits for TAS-102 were observed in normal and mild renal and hepatic subgroups compared with pbo (Table). In TAS-102 groups, there were no marked safety differences between pts with normal function and mild renal or hepatic dysfunction. TAS-102 pts with moderate renal impairment had favorable survival benefit, but higher incidence of treatment-related Grade 3 adverse events (AEs) and serious AEs (SAEs), as well as dose reductions compared with pts in other renal subgroups. Conclusions: TAS-102 demonstrated a survival benefit for normal/mild pt subgroups. TAS-102 pts with moderate renal dysfunction had a modest OS benefit with higher frequency of treatment-related AEs. Clinical trial information: NCT01607957.

Background: Anti-EGFR therapy improves survival for patients (pts) with RAS WT mCRC, but nearly all pts develop therapeutic resistance. MET amplification (amp) is a well-described driver of acquired resistance to antiEGFR therapy. Cabozantinib (C) is an oral inhibitor of multiple tyrosine kinases, including VEGFR1-3 and MET. Panitumumab (P) is an anti-EGFR monoclonal antibody. We designed a phase Ib study of C+P to target MET amp and enhance the activity of anti-EGFR therapy. Methods: Eligible pts with chemotherapy refractory KRAS WT mCRC were enrolled in a 3+3 dose finding cohort (Dose Find) to identify the recommended phase II dose (RPTD). Prior anti-EGFR therapy was permitted. Cycle length was 28 days. Pts were then enrolled in a single-arm expansion cohort (EXP) and treated at the RPTD. The EXP cohort included a 2-week C monotherapy lead-in. The objectives of the EXP cohort were to better describe the safety, tolerability, and efficacy of C+P. Response assessment occurred every 2 months (mos) using RECIST version 1.1. Peripheral blood was sequenced for . 54 gene mutations and focal amps, including MET (Guardant Health, Inc.). Results: As of 11/24/2015, 16 pts were evaluable for toxicity (6 Dose Find; 10 EXP), and 14 pts were evaluable for response (6 Dose Find; 8 EXP). 13 pts had received prior anti-EGFR therapy. There were no dose limiting toxicity events in the Dose Find cohort (0/6). The RPTD was C 60mg PO daily and P 6 mg/kg IV Q2 weeks. In the Dose Find and EXP cohorts, there were no grade 4 or 5 treatment related adverse events (TRAEs). The most frequently reported (all grades . 30%) TRAEs were (n, %) acneiform rash (10, 63%), oral mucositis (9, 56%), diarrhea (8, 50%), increased AST (7, 44%), paronychia (6, 38%), increased ALT (5, 31%), and fatigue (5, 31%). Median PFS was 3.7 mos (95% C.I., 2.3–7.4 mos). Median OS was 7.5 mos (95% C. I., 6.4–12.1 mos). 2 pts (14%) had a confirmed partial response, and 5 pts had PFS . 6 months. Conclusions: In this heavily pre-treated patient population, the combination of C+P demonstrated an acceptable safety profile and encouraging clinical activity at the RPTD. Recruitment for the EXP cohort is now complete. Blood biomarker analysis will be presented. Clinical trial information: NCT02008383.

TAS-102 vs Pbo

Any TAS-102–Related

Any TAS-102–Related

All TAS-102

Overall Survival

Grade ‡3 AE

SAE

Dose Reductions

HR (95% CI)*

n (%)

n (%)

n (%)

Normal

0.63 (0.50-0.80)

171/325 (52.6)

25/325 (7.7)

55/325 (16.9)

Mild G1

0.71 (0.53-0.95)

71/169 (42)

20/169 (11.8)

16/169 (9.5)

Mild G2

0.44 (0.21-0.92)

18/34 (52.9)

5/34 (14.7)

2/34 (5.9)

Renal Normal

0.64 (0.51-0.81)

138/306 (45.1)

28/306 (9.2)

33/306 (10.8)

Mild

0.71 (0.53-0.96)

94/178 (52.8)

16/178 (9.0)

29/178 (16.3)

Moderate

0.85 (0.47-1.56)

29/47 (61.7)

6/47 (12.8)

11/47 (23.4)

Hepatic

*Intention-to-treat population; all other data from as-treated population.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Colorectal) Cancer 3549

Poster Session (Board #246), Sat, 8:00 AM-11:30 AM

Two-stage hepatectomy for colorectal metastases: Association of pathologic response to intensified preoperative chemotherapy with second stage completion and longer survival. First Author: Frederic Bibeau, Institut Val d’Aurelle, Montpellier, France Background: The two-stage surgical resection (TSR) of bilobar colorectal liver metastases (CRLM) is widely used and has shown encouraging survival results. The risk of drop-out after the first-stage hepatectomy remains high and associated with poor survival rates. Our objective was to evaluate the predictive factors of long-term survival, based on the pathologic response to an intensified systemic chemotherapy administered preoperatively. Methods: Data from 899 patients treated for CRLM in our institution were collected prospectively between January 2003 and August 2013. We evaluated the pathologic response to preoperative chemotherapy, and its impact on the second-stage completion and on survival. The Tumour Regression Grade (TRG), the Blazer grade, and the modified-TRG were used to classify patients as responders (TRG and mTRG 1-3, Blazer 0-1) or nonresponders (TRG and mTRG 4-5, Blazer 2) after the first stage Results: Sixtyseven patients were eligible for the TSR first stage. All patients underwent an intensified chemotherapy in combination with a biotherapy (Bevacizumab or Cetuximab) in 38cases. Responders in the three classifications (TRG: p = 0.033; mTRG: p = 0.03, Blazer: p = 0.005), and initial metastases number (p = 0.001) were independent predictive factors for the second-stage completion. Triple chemotherapy was significantly associated with responders in the three classifications (TRG and mTRG: 73.7% versus 26.3% p , 0.0001 ; Blazer : 84.2% versus 15.8% p = 0.001). Median overall survival (OS) of patients who completed TSR was significantly different (44, 84 versus 18,39 months; p , 0.0001). Conclusions: A good pathologic response to intensified preoperative chemotherapy is associated with completion of the second stage of TSR, and thus with a longer survival. Knowing this response before the first-stage resection may allow avoiding useless resections for patients who will not benefit from this strategy. Updated relapse-free survival and OS and particular patterns of pathologic response will be presented for this meeting.

3551

Poster Session (Board #248), Sat, 8:00 AM-11:30 AM

Influence of KRAS exon 2 mutation variants as well as NRAS- and BRAFmutations on outcome of patients with metastatic colorectal cancer (mCRC) receiving combination chemotherapy with or without bevacizumab. First Author: Dominik Paul Modest, Department of Hematology and Oncology, Klinikum Grosshadern, University of Munich, Munich, Germany Background: To explore the impact of KRAS, NRAS and BRAF mutations as well as KRAS mutation variants in patients (pts.) with mCRC receiving first-line combination chemotherapy with or without bevacizumab. Methods: 1239 pts. from 5 randomized trials (FIRE-1, FIRE-3, AIOKRK0207, AIOKRK0604, RO91) were included into the analysis. Outcome was retrospectively evaluated by KaplanMeier method, log rank tests and cox regression. Results: In 664 tumors no mutation was detected, 462 tumors were diagnosed with KRAS mutation, 39 tumors with NRAS mutation and 74 tumors with BRAF mutation. This categorization was associated with significant differences in progression-free (PFS) (P,0.01) and overall survival (OS) (P,0.01). Furthermore, KRAS exon 2 mutation variants were associated with heterogeneous OS (p=0.01). In particular G12D and G12V subtypes that occur with high frequency among KRAS exon 2 variants were associated with rather favourable OS, whereas other less frequent subtypes (G12C, G13D, G12S) correlated with less favourable outcome. Outcome data are summarized in table 1. Multivariate analyses will be presented at the meeting. Conclusions: In this retrospective evaluation, KRAS exon 2 mutation variants G12D and G12V were associated with rather favourable OS while KRAS G12C, G12S and G13D mutant tumors were associated with less favourable OS. Expectedly, any mutations in KRAS and NRAS were associated with inferior OS of mCRC patients compared to patients with tumors not being mutated in any of these genes. Subgroups No mutation (n=664) KRAS mutation (n=462) G12D (n=152) G12V (n=92) G13D (n=71) G12S (n=34) G12C (n=28) G12A (n=22) NRAS mutation (n=39) BRAF mutation (n=74)

Progression-free survival months (95% CI)

Overall survival months (95% CI)

10.3 (9.7-10.8)

26.9 (25.2-28.5)

9.5 (8.9-10.1)

21.0 (18.5-23.5)

10.5 (9.0-11.9) 9.5 (8.2-10.8) 8.8 (7.6-10.0) 9.3 (7.1-11.4) 10.1 (6.4-13.8) 9.8 (7.7-11.8) 9.5 (8.3-10.7)

25.2 23.5 17.6 19.1 16.8 24.0 21.3

7.4 (6.2-8.5)

(22.2-28.2) (16.7-30.3) (13.3-21.9) (15.2-23.0) (15.6-18.0) (16.4-31.5) (14.9-27.8)

3550

183s Poster Session (Board #247), Sat, 8:00 AM-11:30 AM

An international phase III randomized, non-inferiority trial comparing 3 vs 6 months of oxaliplatin-based adjuvant chemotherapy for colon cancer: Compliance and safety of the phase III Japanese ACHIEVE trial. First Author: Tetsuya Eto, Department of Gastroenterology, Tsuchiura Kyodo General Hospital, Tsuchiura, Japan Background: Oxaliplatin/Fluoropyrimidine chemotherapy is an established adjuvant treatment for colon cancer (CC). Neurotoxicity from oxaliplatin is cumulative, dose limiting, and potentially irreversible. The International Duration Evaluation of Adjuvant Chemotherapy (IDEA) project was established to prospectively analyze data from several randomized trials to test whether 3-month (arm 3) of oxaliplatin-based adjuvant (FOLFOX4/ mFOLFOX6 or XELOX) is non-inferior in terms of disease-free survival to 6-month (arm 6) treatment in patients (pts) with stage III CC, with less toxicity. Methods: We present the compliance and safety of the phase III ACHIEVE trial (JFMC47) as one of the IDEA project. Results: From 2012 to 2014, 1,313 patients were randomized from 244 centers and 1,301 (651 in arm 3 and 650 in arm 6) included in the intention-to-treat (ITT) population. Baseline characteristics and proportion of mFOLFOX6 vs XELOX were well balanced (25.0% vs 75.0%). Among the safety population (N = 1,277, 98% of ITT population), there was a treatment-related death. Compliance was higher in arm 3 than in arm 6: treatment completion rate in pts receiving all the planned cycles were 86.3% in arm 3 and 61.4% in arm 6, and consistent higher rate either mFOLFOX6 or XELOX was observed in arm 3 than arm 6 (87.6% vs 69.6% in mFOLFOX6; 85.9% vs 58.7% in XELOX). Therapy was permanently interrupted due to toxicity in 11.5% in arm 3 and 28.7% in arm 6. The overall incidence of grade3-4 toxicity events was 28.7% in arm 3 and 42.8% in arm 6. Grade 2 or higher neuropathy was more frequent in arm 6 than in arm 3: 36.5 % vs 13.6%. However, toxicity events depended on the FU backbone with more grade 3-4 neutropenia on mFOLFOX6 (30.4% vs 12.4%), and more grade 3-4 anorexia (1.9% vs 5.1%) and grade 3-4 diarrhea (1.3% vs 5.5%) on XELOX. Conclusions: Both mFOLFOX6 and XELOX were safe and well tolerated. The grade3-4 main toxicities did vary according to FU backbone. Observed toxicity was significantly higher in the 6-month arm compared to that of the 3-month arm; compliance was lower in the standard 6-month arm. Clinical trial information: UMIN 000008543.

3552

Poster Session (Board #249), Sat, 8:00 AM-11:30 AM

Comparative molecular analyses of colon versus rectal tumors. First Author: John Marshall, Lombardi Comprehensive Cancer Center, Georgetown University, Washington, DC Background: Colorectal cancer (CRC) is a heterogeneous disease. There is limited data on colon and rectal tumor molecular differences. Methods: A total of 8174 CRC tumors submitted to Caris Life Sciences for IHC (protein expression), ISH (gene amplification), and NGS sequencing between 2009 and 2015 were studied. Only tumors with origins that were clearly defined as ascending, transverse, or descending colon or rectum were included in this study. We excluded any cases without clear designation. Chi-square tests determined molecular differences between colon and rectal tumors. Results: A total of 2,010 tumors met our inclusion criteria and were examined. We compared primary (1o) colon (n = 502) with 1o rectal (n = 872) tumors; the colon 1os had higher rates of ATM (7.5% vs. 3.2%, p = 0.01), BRAF (24% vs. 3.4%, p , 0.01), CTNNB1 (4% vs. 0.3%, p = 0.003), IDH1 (2% vs. 0%, p = 0.01), GNAS (3.3% vs. 1%, p = 0.04), HNF1A (15% vs. 9%, p = 0.01), JAK3 (8% vs. 2.2%, p = 0.002), PIK3CA (24% vs. 12%, p = , 0.01), RB1 (1.7% vs. 0%, p = 0.02), and SMAD4 (17% vs. 12%, p = 0.05) mutations. However, mutations in APC and TP53, and amplification of HER2, were higher in rectaltumors (67% vs. 54% and 72% vs. 60% [both p , 0.01] and 5.4 % vs. 1.5% [p , 0.01]). Colon 1o tumors had higher MSI (17% vs. 0.7%, p , 0.01), EGFR (64% vs. 42%, p , 0.01), PD-1 (55% vs. 44%, p = 0.01), and TS (36% vs. 28%, p , 0.01) expression. In contrast, rectal 1o tumors had higher expression of MGMT (64% vs. 52%, p = 0.002), TLE3 (33% vs. 26%, p = 0.03), TOPO1 (52% vs. 36%, p , 0.01), and TUBB3 (41% vs. 30%, p , 0.01). When we excluded MSI+ tumors and analyzed only the MSI- tumor differences, colon tumors still had a greater frequency of BRAF (13% vs. 4.5%, p = 0.003) and PIK3CA (22% vs. 12%, p = 0.009) mutations, and higher EGFR expression (63% vs. 51%, p = 0.047) than rectal tumors, whereas, PTEN (56% vs. 41%, p = 0.004), TLE3 (43% vs. 29%, p = 0.004) and TOPO1 (50% vs. 38%, p = 0.03) were overexpressed in rectal tumors. Conclusions: Molecular profile differences between colon and rectal tumors suggest different carcinogenic pathways and biology. Low frequency mutations in several druggable genes, or protein overexpression (e.g., HER-2, PIK3CA, TOPO1) provide therapeutic targets. Our findings are consistent with the consensus molecular subtyping of CRC.

11.7 (9.7-13.6)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

184s 3553

Gastrointestinal (Colorectal) Cancer Poster Session (Board #250), Sat, 8:00 AM-11:30 AM

Erlotinib added to bevacizumab as maintenance therapy and health-related quality of life in patients with metastatic colorectal cancer: Results of the GERCOR DREAM phase III trial. First Author: Amelie Anota, Methodology and Quality of Life Unit, Department of Oncology, EA 3181, University Hospital;French National Platform Quality of Life and Cancer, Besancon, France Background: The DREAM phase III clinical trial aimed to assess the efficacy and tolerability of maintenance bevacizumab (B) plus erlotinib (E) after a bevacizumab-based induction therapy in patients with unresectable metastatic colorectal cancer (Tournigand C. et al. Lancet Oncol. 2015). The addition of E to B improved maintenance progression-free survival (PFS) (median B: 4.9months; median B+E: 5.4 months; Hazard Ratio (HR) = 0.78 [0.68-0.96], P = 0.036) and overall survival (OS). This study presents the results of health-related quality of life (HRQoL) data (secondary endpoint). Methods: HRQoL was assed using EuroQoL EQ-5D generic questionnaire before maintenance randomization (T0), at 2 (T1) and 4 months (T2). Analysis of baseline missing data profile was done comparing responders to non responders according to baseline patients’ characteristics and OS. The longitudinal analysis of the visual analogue scale (VAS) of the EQ-5D was done with a linear mixed model (LMM) by including the arm, the time and their interaction. Time to HRQoL deterioration (TTD) was defined as the time from randomization to the first deterioration of 5-point at least of the VAS score as compared to T0. Univariate cox regression analyses were performed to estimate HR of the treatment effect with its 95% confidence interval. Results: Between 2007 and 2011, 700 patients were included. After induction therapy, 452 patients without disease progression were randomized in B arm (N = 228) or B+E arm (N = 224). Among them, 111 patients (24.5%) completed the EQ-5D at T0, 78 (17.3%) at T1 and 54 (11.9%) at T2. Effects introduced in the LMM were not significant: arm effect (coeff = 0.93 (ref = Arm A), P = 0.83), time effect (coeff = 0.25, P = 0.88) and treatment by time interaction (coeff = 1.80, P = 0.42) reflecting a similar HRQoL among arms throughout its measurement. Median of TTD was 4.5 months (2.7-NA) for B vs. 4.1 months (2.3-NA) for B+E (HR = 1.46 [0.67-3.19], P = 0.34). Conclusions: These HRQoL results suggest that the added value of erlotinib to bevacizumab is not associated with a HRQoL deterioration, while it allows to improve maintenance PFS and OS. Clinical trial information: NCT00265824.

3555

Poster Session (Board #252), Sat, 8:00 AM-11:30 AM

3554

Poster Session (Board #251), Sat, 8:00 AM-11:30 AM

Bevacizumab plus chemotherapy (bev/CT) as first-line therapy for patients with potentially resectable metastatic colorectal cancer (mCRC): Final results of the French noninterventional PICASSO study. First Author: David Malka, Gustave Roussy Cancer Campus, Villejuif, France Background: Few prospective studies specifically focused on the management of potentially resectable patients with mCRC. This prospective noninterventional cohort evaluated patients (pts) with mCRC and no evidence of new metastatic disease (NED) after 1st line bev/CT followed or not by secondary metastases resection. Methods: Previously untreated mCRC pts with potentially resectable liver and/or lung metastases treated with 1st line bev/CT were included and followed for 36 months. Decisions for patient enrollment, treatment initiation, and surgery metastases, were taken by a multidisciplinary team. The primary endpoint was frequency of pts with NED with or without secondary resection after a 1st line treatment with bev/CT. Secondary endpoints included progression-free survival (PFS), overall survival (OS), relapse-free survival, safety and criteria to define pts unresectability. Results: Between September 2010 and June 2012, 218 pts were included. Out of the 205 evaluable pts (efficacy population), 64% were men, mean age was 66 years, 96% were ECOG 0-1, 86% had liver only metastases. After the 1st line bev/CT, 104 pts (51%) had liver and/or lung metastases resection. The primary endpoint, NED, was reached by 92 pts (48%) [95% CI, 40;55%] including 88 pts after surgery (85%[76;91%] of resected patients). NED also occurred in 4 pts who did not underwent metastasis resection (4%[1;11%] of unresected pts) due to a complete response before surgery. Median PFS was 11.5 months [10.4;12.4] in the efficacy population and 15.7 months [12.9;18.9] in NED pts. The OS rate at 36 months was 52% [44;56%] and 77% [66;85%] for NED pts. In the safety population (n = 210), grade $ 3 adverse events (AEs) occurred in 92 pts (44%), treatment related AEs occurred in 17 pts (8%). Overall, 15 pts (7%) died of an AE, of which only 1 patient (0.5%) had a treatment related AE leading to death. Conclusions: In this non-interventional study conducted in real-life setting in potentially resectable mCRC pts treated with bev/CT in 1st line, half of pts were resected and showed no detectable metastatic disease and prolonged OS. Clinical trial information: NCT01343901.

3556

Poster Session (Board #253), Sat, 8:00 AM-11:30 AM

Bevacizumab plus chemotherapy as first-line therapy for elderly patients with metastatic colorectal cancer: Interim results of the noninterventional CASSIOPEE study. First Author: Eric Francois, Department of Medical Oncology, Centre Antoine-Lacassagne, Nice, France

Onset of neutropenia as an indicator of treatment response in the phase III RECOURSE trial of TAS-102 vs placebo in patients with metastatic colorectal cancer. First Author: Atsushi Ohtsu, National Cancer Center Hospital East, Kashiwa, Japan

Background: Approximately half of the patients (pts) with metastatic colorectal cancer (mCRC) are elderly. A real-life setting study was conducted in elderly pts ($ 75 years-old) with mCRC treated in daily practice with bevacizumab (bev) plus chemotherapy, in order to improve the knowledge on this population and to contribute in optimizing treatment strategy. Methods: CASSIOPEE is a prospective, multicenter, French, noninterventional study conducted over 24 months. Pts aged 75 years or older with mCRC and treated in 1stline with bev and chemotherapy were eligible. The primary objective was to describe progression-free survival (PFS). Secondary objectives included the description of pts characteristics, overall survival, bev regimen, safety and autonomy criteria such as Lawton Instrumental Activities of Daily Living Scale (IADL 4 items) and Balducci score. Results of a planned interim analysis after 12 months of follow-up are presented (except for PFS on a longer period). Results: Between March 2012 and June 2014, 403 pts were included. Out of the 351 pts (efficacy population), 52% were men, mean age 81 (64), 80% had primary tumor located in the colon, the main metastatic sites were liver (65%), including liver only (38%). Median PFS was 9.2 months [8.4;10.4] in the efficacy population and 9.8 months for pts aged , 80, 9.2 months for pts aged $ 80 or # 85 and 8.3 months for pts aged . 85. The PFS rate at 12 months was 35.3%. On the IADL scale, change from baseline to 6 months was stable (61%), improvement (28%) and deterioration (10%). Balducci score was stable from baseline to 6 months: harmonious (38% vs. 36%), intermediate (21% vs. 23%) and frail (3% vs. 4%), respectively. In the safety population (n = 382), adverse events grade $ 3 occurred in 34% pts, of which 28% bev related. Overall, 3% pts died of an AE, of which 25% bev related. Conclusions: CASSIOPEE, a non-interventional real-life setting study, showed treatment benefit andacceptable safety of 1st line bev combined with chemotherapy in elderly pts. After 12 months of follow-up, PFS was 9.2 months, that is comparable with PFS of randomized studies. Clinical trial information: NCT01555762.

Background: TAS-102 is comprised of an antineoplastic thymidine-based nucleoside analog, trifluridine (FTD), and a thymidine phosphorylase inhibitor, tipiracil. Primary results of the RECOURSE trial demonstrated a significant improvement in overall survival (OS) and progression-free survival (PFS) with TAS102 vs placebo (pbo) in patients (pts) with metastatic colorectal cancer refractory/ intolerant to standard therapies. Neutropenia is a common TAS-102–associated adverse event and it has been hypothesized to be associated with a relatively high FTD concentration in pts. Methods: RECOURSE data were analyzed post hoc for correlations between onset of neutropenia (Grade 3/4) and survival benefit. Results: Of 533 pts given TAS-102, 75 (14%) developed Grade 3/4 neutropenia in treatment cycle 1, 86 (16%) in cycle 2, and 39 (7%) in cycle $3. Onset of neutropenia at any cycle was associated with longer median OS and PFS compared with no neutropenia. A consistent survival benefit was observed regardless of the cycle of initial onset of neutropenia, as demonstrated by the hazard ratio (against cycle-matched pbo control groups) and corresponding median OS differences (Table). Conclusions: An association between occurrence of Grade 3/4 neutropenia and survival benefit was observed. The data indicate that such survival benefit occurred regardless of whether the initial onset of neutropenia occurred after cycle 1, cycle 2, or later. Further analyses are required to fully determine whether FTD pharmacokinetics correlate with TAS-102 efficacy and onset of neutropenia, and whether cycle initiation delays affect response. Clinical trial information: NCT01607957.

All RECOURSE (as treated) Earliest cycle of Grade 3/4 neutropenia onset in TAS-102 arm 1 2 $3 None

TAS-102/Pbo* n (%)

TAS-102/Pbo* Median OS (months)

Hazard Ratio (95% CI)

533 (100)/265 (100)

7.1/5.3

0.68 (0.58-0.81)

75 (14)/265 (100) 86 (16)/215 (81) 39 (7)/48 (18) 333 (62)/265 (100)

9.7/5.3 8.7/6.3 13.8/10.2 5.5/5.3

0.45 0.56 0.36 0.97

(0.32-0.64) (0.41-0.78) (0.17-0.75) (0.81-1.16)

*Number of pbo pts who entered cycle; no pbo pts exhibited neutropenia.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Colorectal) Cancer 3557

Poster Session (Board #254), Sat, 8:00 AM-11:30 AM

Onset of neutropenia as an indicator of treatment response in the randomized phase II of TAS-102 vs placebo in Japanese patients with metastatic colorectal cancer (Study J003-10040030). First Author: Tomohiro Nishina, National Hospital Organization Shikoku Cancer Center, Matsuyama, Japan Background: TAS-102 is comprised of an antineoplastic thymidine-based nucleoside analog, trifluridine (FTD), and a thymidine phosphorylase inhibitor, tipiracil hydrochloride. Primary results of the J003-10040030 randomized phase 2 trial demonstrated a significant improvement in overall survival (OS) and progression-free survival (PFS) favoring TAS-102 vs placebo (pbo) in patients (pts) with metastatic colorectal cancer refractory/intolerant to standard therapies (Yoshino T, et al. Lancet Oncology 2012). Neutropenia is the most common TAS102-associated adverse event and it has been hypothesized to reflect a relatively high FTD concentration in pts. Methods: Study J003 data were analyzed post hoc for correlations between initial onset of Grade 3/4 neutropenia and OS benefit. Results: Of 112 pts given TAS-102, 31 (28%) developed Grade 3/4 neutropenia in treatment cycle 1, 19 (17%) in cycle 2, and 8 (7%) in $ cycle 3. Onset of Grade 3/4 neutropenia at cycle 1 and 2 was associated with longer median OS compared to pts without neutropenia as summarized by the hazard ratio (against cycle-matched placebo control groups) and the corresponding median OS differences (see Table). Conclusions: Pts experienced Grade 3/4 neutropenia in cycle 1 and 2 showed a greater OS benefit over pbo. In contrast, pts who did not experience Grade 3/4 neutropenia appear to show no OS benefit. The occurrence of Grade 3/4 neutropenia might become an on-treatment predictive biomarker of TAS-102 and TAS-102 treatment should not be discontinued permanently due to the onset of Grade 3/4 neutropenia. The correlation between efficacy of TAS-102 and onset of neutropenia warrants further analyses. Clinical trial information: 090880.

All J003 (Full Analysis Set) Earliest cycle of Grade 3/4 neutropenia onset in TAS-102 arm 1 2 $3 None 1

TAS-102/PBO1 n (%)

TAS-102/PBO1 median OS (months)

Hazard ratio (95% CI)

112(100)/57(100)

9.0/6.6

0.56 (0.39-0.81)

31 (28)/57 (100) 19 (17)/18 (32) 8 (7)/9 (16) 54 (48)/57 (100)

13.0 /6.6 Not reached /9.8 10.3 /10.8 6.3 /6.6

0.31 0.28 1.07 1.05

(0.17, (0.11, (0.34, (0.70,

0.56) 0.75) 3.36) 1.58)

3558

185s Poster Session (Board #255), Sat, 8:00 AM-11:30 AM

Curative treatment for patients with synchronous liver metastases and peritoneal carcinomatosis of advanced colorectal cancer (aCRC): A multicenter study of the French Association of Surgery. First Author: Rea lo Dico, Hopital Lariboisiere AP-HP, service de chirurgie digestie et cancerologie, ´ 75010 PARIS, Paris, France Background: Patients with synchronous peritoneal carcinomatosis (PC) and liver metastasis (LM) of colorectal cancer (CRC) are generally considered for exclusive systemic palliative chemotherapy. Aggressive surgical approaches combining hepatectomy and peritoneal resection with curative intent remain controversial in such a setting and based on no solid data. The aim of this prospective cohort was to assess morbidity, mortality, disease-free survival (DFS) and overall survival (OS) of patients with PC and LM of aCRC origin treated with an aggressive therapeutic approach combining surgical treatment of liver and peritoneal lesions followed by hyperthermic intraperitoneal chemotherapy (HIPEC). Methods: All patients with PC and synchronous LM who undergone cytoreductive surgery and LM resection in the same time followed by HIPEC were registered in the prospective database of the French Association of Surgery and analysed. The primary endpoint was survival from the time of surgery. Results: From 1993 to 2015, 146 patients were analyzed. After a mean follow-up of 36 months, the median OS and DFS, were respectively 27,2 and 9.5 months. The median number of LM was 4 (range, 1–8).Postoperative morbidity Grade III - IV (Dindo Classification) rate was 14.8% with no postoperative death reported. Patients with complete cytoreductive surgery (n = 132) had higher OS than those with incomplete cytoreduction (n = 14) with a median of 29, versus 4 months respectively (p = 0.0001). Rectal primary tumor, PCI of 13 or more, positive nodal status of the primitive tumor, and more than 3 LM were not identified as independent factors for poor OS. Conclusions: This multicenter study is the largest series currently reported confirming that prolonged survival can be achieved in selected patients suitable for PC and LM surgery. Multimodality treatment, including surgical treatment of PC and LM with curative intent, and heated intraperitoneal chemotherapy had acceptable in terms of morbidity and allows prolonged survival.

Number of PBO pts who entered cycle– no pbo patient exhibited neutropenia.

3559

Poster Session (Board #256), Sat, 8:00 AM-11:30 AM

Comparison of three molecular methods to detect mutations in KRAS, NRAS, BRAF and PIK3CA in metastatic colorectal cancer (mCRC). First Author: Cristina Santos, Translational Research Laboratory and Department of Medical Oncology, Institut Catala` d’Oncologia-IDIBELL, L’Hospitalet de Llobregat, Spain Background: Mutational analysis of RAS in mCRC has showed an improvement in patient (pts) selection for anti-EGFR drugs. The standard techniques (Techs) have an analytical sensitivity (Sns) of 1-5%. Digital PCR (dPCR) can detect mutated alleles at .05-.1% Sns, but it is unclear if this offers a clinical advantage. Objective:To compare the Sns and feasibility to detect point mutations in KRAS, NRAS, BRAF and PIK3CAby Cobas, Therascreen assays and quantitative dPCR (Fluidigm), and correlate them with progression-free survival (PFS) when treated with anti-EGFR- or bevacizumab (bev)-based therapies (Tx). Methods: 585 pts datasets treated with anti-EGFR- or bev-based Tx were pooled from several mCRC trials and retrospective series from the TTD/RTICC Spanish network. Concordance and predictability were evaluated using kappa and Harrel’s C index. Results: Out of the 585 pts, 385 were male (65.8%), median age 65, 192 pts (32.8%) included in trials (122 anti-EGFR (20.8%) and 70 bev (12%)), 393 pts (67.2%) treated as per label indication [143 anti-EGFR (24.4%) and 250 bev (42.7%)]. There were 4 invalid results for Therascreen and 6 for Cobas and dPCR, respectively (Table). No differences in PFS or OS in bev group were observed by molecular profiling with any platform. Globally, in pts who received anti-EGFR Tx, PFS was significantly better in RAS/BRAF wt groups (HR = 1.45, p = .016 for Cobas; HR = 1.78, p , .001 for Therascreen; and HR = 1.45, p = .005 for dPCR. Differences in OS in anti-EGFR group were detected for sensitive (HR = 2.041, p = .011) and chemorefractary (HR = 2.72, p = .001) pts by dPCR and considering all mutations. dPCR only outperformed the other Techs when predicting OS. Fine tunning of the optimal Sns threshold of quantitative Techs will be presented. Conclusions: Although the rate of mutations between Techs is different, the predictability of PFS to anti-EGFR Tx is similar in the first-line setting, but higher sensitivity methods tend to perform better in chemorefractory patients Mutations KRAS (exon2/3) RAS extended (exon2/3/4) Any mutation

Cobas N%

Kappa

Therascreen N%

Kappa

dPCR N%

Kappa dPCR/Cobas

209 36 273 47 333 57

.87 .73 .67

208 36 276 47 326 56

.71 .63 .58

262 45 333 57 361 62

.77 .64 .60

3560

Poster Session (Board #257), Sat, 8:00 AM-11:30 AM

Prospective analysis of CEA, circulating DNA (cDNA) and circulating tumor cells (CTC) kinetics in patients (pts) treated for a metastatic colorectal cancer (mCRC). First Author: David Sefrioui, Digestive Oncology Unit, IRON group, Rouen Hospital, University of Normandy, Rouen, France Background: CEA, cDNA and CTC have been reported as useful markers in mCRC pts treated with chemotherapy (CT)-based regimen. We previously reported that CEA kinetic (threshold 0.05) was associated with outcome [J Clin Oncol 2008;26:3681-6]. We designed a prospective muticentric trial for CEA kinetic validation and to evaluate cDNA and CTC kinetics. Methods: mCRCpts, PS 0-2, with CEA $ 5mg/L, who started a CT were included. Plasma were taken from baseline (T0) to the fourth cycle for CEA and at T0 and 6 weeks (W6) for cDNA and CTC kinetics. Cell-free DNA (cfDNA) was quantified by fluorimetric method, circulating tumour DNA (ctDNA) by Digital PCR, and CTC (CTC+ vs CTC-) by ScreenCellaˆ method. Primary endpoint was control disease (CD) (response/stable vs progressive disease (PD)) (RECIST 1.1) according to CEA kinetic. Secondary endpoints were progression-free survival (PFS) and overall survival (OS) according to T0 median CEA, cfDNA, ctDNA, CTC +/- and kinetics. Results: A total of 200 mCRC pts were included with a median follow-up of 12 months (m) (range 141). CtDNA was detected in 90% from the 95 RAS/BRAF mutated tumours. Mean CEA, cfDNA and ctDNA significantly decreased from T0 to W6 (respectively 436 vs 244, p , 0.0001; 98 vs 35, p = 0.01; 20.3 vs 5.2, p , 0.0001). A total of 54.7% were CTC+ at T0 versus 60% at W6 (NS). CEA kinetic (# 0.05 threshold) was associated with CD (90 vs 52%, p , 0.0001), PFS (8 vs 3 m; Hazard ratio (HR) 0.5, p , 0.0001) and OS (16 vs 12.5m, HR 0.62, p = 0.021). Median (low vs high) cfDNA and ctDNA at T0 were associated with PFS (7.8 vs 5.6m, HR 1.7, p = 0.0009 and 7.5 vs 5m, HR 2.1, p = 0.001, respectively) and OS (18.7 vs 13m, HR 2, p , 0.0001 and 19 vs 12m, HR 2.3, p = 0.0007, respectively). Among tested threshold, a decrease of ctDNA from T0 to w6 (factor . 2.6) was associated with CD (80 vs 20%, p , 0.0001) with a trend for PFS (p = 0.11). CEA kinetic . 0.05 (adjusted Odds Ratio (ORa) = 22.2), ctDNA kinetic # 2.6 (ORa = 15.1) and median cfDNA (ORa = 0.10) were independent factor of PD. Conclusions: This prospective study confirmed that CEA kinetic is clinically relevant for CT monitoring in mCRC. Our results also highlight that cfDNA and ctDNA are both associated with outcome.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

186s 3561

Gastrointestinal (Colorectal) Cancer Poster Session (Board #258), Sat, 8:00 AM-11:30 AM

3563

Poster Session (Board #260), Sat, 8:00 AM-11:30 AM

Exploring outcomes of RAS-mutant (RAS mut) advanced colorectal cancer (aCRC) treated with chemotherapy: Analysis from 2254 patients (pts) in randomised clinical trials (RCTs). First Author: Jenny F. Seligmann, University of Leeds, Leeds, United Kingdom

A phase II and co-clinical study of an AKT inhibitor in patients (pts) with biomarker-enriched, previously treated metastatic colorectal cancer (mCRC). First Author: A. Dasari, The University of Texas MD Anderson Cancer Center, Houston, TX

Background: RAS mut status predicts lack of benefit from anti-EGFR agents in aCRC, but its impact on prognosis and chemotherapy outcomes is less clear. Previously we have reported that the poor outcomes of BRAF mut cancer aCRC are due to accelerated decline following 1st-line progression, not chemoresistance. In the same dataset we perform a detailed analysis of outcomes of RAS mut pts throughout the aCRC treatment pathway to help guide clinical decision-making in this underserved population. Methods: RAS was assessed in tumors of 2254 BRAF wild-type (wt) pts in 3 RCTs: COIN (n = 1158), FOCUS (n = 706, KRAS only) and PICCOLO (n = 390). End-points were progression free survival (PFS), response rate (RR), and OS. Treatments received were 1st line OxFU (COIN), 1st line OxFU or IrFU (FOCUS) or 2nd line Irinotecan (Ir) (PICCOLO). Analyses were adjusted for known negative prognostic markers: poor performance status, primary tumour in-situ, right tumour location, peritoneal mets and high platelet count, and were performed using Cox proportional hazards models and logistic regression. Results: 1101 pts (48.8%) were RAS mut. RAS mut status conferred worse OS in both 1st-line studies (COIN adj HR = 1.38, p , 0.001; FOCUS adj HR = 1.33, p , 0.001), and at the point of starting 2nd-line treatment (adj HR = 1.33, p = 0.014). Compared with wt, RAS mut pts treated with 1st-line combination chemotherapy had inferior treatment outcomes (RR 42.2% vs 51.7%; adj OR = 0.69,p , 0.001) and PFS (6.4 vs 8.0 mths, adj HR = 1.24, p , 0.001). RAS mut pts were just as likely as wt to receive 2nd-line treatment after 1st-line progression in COIN (51.4% vs 51.7%) and were well represented in PICCOLO (53.6%). RAS-mut pts treated with 2nd-line Ir had inferior outcomes than wt (RR 7.7% vs 12.7%; PFS 3.1 vs 4.9 months) but neither were statistically significant (p = 0.10 and p = 0.42, respectively). Conclusions: RAS-mut status is an independent poor prognostic marker in aCRC. This is partly driven by relative chemoresistance in both the 1st and 2nd-line settings, observed following adjustment for known poor prognostic factors. New approaches in treating RAS mut aCRC are urgently required.

Background: The PI3K / AKT pathway has several roles in CRC growth and progression through downstream targets including the MAPK pathway. Based on strong preclinical mCRC and non-mCRC clinical data, we hypothesized that PI3K pathway inhibition may have clinical activity in mCRC pts with biomarkers of PI3K pathway activation. Methods: Previously treated mCRC pts were enrolled into 2 cohorts (A: PTEN loss by IHC; B: PIK3CA mutation, mt; table). Since MAPK pathway mts have been shown to reduce sensitivity to PI3K pathway inhibition, KRAS and BRAF mts were used as negative selectors. Pts were administered the oral, allosteric AKT inhibitor, MK2206 in this single institution, CTEP-sponsored trial. Primary end point was response rate and secondary end points were progression free survival (PFS), overall survival (OS) and safety. Paired biopsies were used to establish patient derived xenografts (PDX) for a co-clinical trial. Other planned translational studies included correlation of biomarkers in primary & metastatic sites and pharmacodynamic assessment of PI3K pathway inhibition. Results: In 18 pts (median age 56; median prior systemic therapies 3), common treatment related adverse events (AEs) were hyperglycemia, nausea / vomiting (67% each) diarrhea (61%) and rash (50%) Grade $ 3 AEs were uncommon and included hyperglycemia, elevated liver tests (17% each) and fatigue (11%). Of these, 2 were grade 4 AEs of diarrhea and elevated bilirubin. There were no radiographic responses; all pts in cohort A had progressive disease (PD) while 2 in cohort B had stable disease (SD) as best response (table). Median PFS was 1.8 months (m) and median OS was 6.8 m. Translational work in PDX, paired tumor biopsies and peripheral blood samples is ongoing and will be presented at the conference. Conclusions: In contrast to robust preclinical data, single agent MK-2206 does not have activity in mCRC pts even when carefully enriched. However, extensive, preplanned translational studies provide us with a unique opportunity to evaluate the reasons for this discrepancy. Clinical trial information: NCT01802320. Cohort A B

3564

Poster Session (Board #261), Sat, 8:00 AM-11:30 AM

Phase 1b extension study of cancer stemness inhibitor BB608 (napabucasin) administered in combination with FOLFIRI +/- bevacizumab (Bev) in patients (pts) with advanced colorectal cancer (CRC). First Author: Bert H. O’Neil, Indiana University, Simon Cancer Center, Indianapolis, IN Background: BB608 (aka Napabucasin, BBI-608) is an oral first-in-class cancer stemness inhibitor targeting Stat3-driven gene transcription. Antitumor activity was observed in vitro and in vivo. BB608 showed safety and encouraging signs of anti-cancer activity in phase I/II studies. Methods: A phase Ib extension multi-center study in pts with advanced CRC was undertaken to confirm the RP2D and signs of anti-cancer activity of BB608 in combination with FOLFIRI +/- Bev. BB608 was administered continuously at 240 mg BID in combination with FOLFIRI (5-FU 400 mg/m2 bolus with 2400 mg/m2, irinotecan 180 mg/m2, and leucovorin 400 mg/m2 infusion) +/- Bev 5 mg/kg, administered bi-weekly until disease progression, unacceptable toxicity, or other discontinuation criterion. Results: 46 pretreated CRC pts who had failed an average of . 2 prior lines of therapy were enrolled; including 20 pts (43.5%) previously progressed on FOLFIRI +/Bev. Of the 46 pts, 14 received FOLFIRI and 32 FOLFIRI with Bev in combination with BB608. There was no dose-limiting or unexpected toxicity or significant pharmacokinetic interactions. Most common adverse events (AEs) included grade 1/2 diarrhea, nausea, vomiting and fatigue. Grade 3 AEs observed in 15 pts included diarrhea (9), fatigue (3), dehydration/ hyponatremia (1), hypokalemia (1) and burning in rectum (1) resolved with dose reduction and/or supportive care. Disease control (PR+SD) was observed in 37 of 40 evaluable pts (93%), with partial response (PR) in 11 pts (28%) (RECIST 1.1 30-65% regression), and stable disease with tumor regression in 21 pts (53%). Among 19 pts who had progressed on FOLFIRI +/- Bev previously and were evaluable for tumor assessment, disease control (PR+SD) was observed in 17 pts (90%), tumor regression was observed in 15 pts (80%) of which 6 pts achieved PR (32%). Conclusions: This phase Ib extension study confirmed that BB608 can be safely combined with FOLFIRI +/- Bev, and shows encouraging signs of anti-tumor activity in CRC pts, including pts who had previously progressed on FOLFIRI +/- Bev. Clinical trial information: NCT02024607.

3565

PIK3CA

AKT

KRAS

BRAF

PTEN IHC

Enrolment

Best Response

wt mt

wt wt

wt wt

wt wt

Loss Any

9 9

PD x 9 SD x 2

Poster Session (Board #262), Sat, 8:00 AM-11:30 AM

KRAS and BRAF as prognostic biomarkers in patients undergoing surgical resection of colorectal cancer liver metastasis: A systematic review and meta-analysis. First Author: Francesco Passiglia, Department of Surgical, Oncological and Oral Sciences, Section of Medical Oncology, University of Palermo, Palermo, Italy Background: Clinical trials investigated the potential role of both KRAS and BRAF mutations, as prognostic biomarkers, in colorectal cancer (CRC) patients who underwent surgical treatment of liver metastasis (CLM), showing conflicting results. This meta-analysis aims to review all the studies reporting survival outcomes (recurrence free survival (RFS), and/or overall survival (OS)) of patients undergoing resection of CLM, stratified according to KRAS and/or BRAF mutation status. Methods: Data from all published studies reporting survival outcomes (RFS and/or OS) of CRC patients who received resection of CLM, stratified by KRAS and/or BRAF mutation status were collected by searching in PubMed, Cochrane Library, American Society of Clinical Oncology and European Society of Medical Oncology meeting proceedings. Pooled hazard ratios (HRs) and 95% confidence intervals (95% CIs) were calculated for both the OS and/or RFS. Results: Seven eligible trials (1403 patients) were included. Pooled analysis showed that KRAS mutations predicted a significant worse both RFS (HR: 1.65; 95% CI: 1.23 – 2.21) and OS (HR: 1.86; 95% CI: 1.51 – 2.30) in patients who underwent surgical resection of CLM. BRAF mutations were also associated with a significant worse OS (HR: 3.90; 95% CI: 1.96 – 7.73) in this subgroup of patients. Conclusions: This meta-analysis suggests both KRAS and BRAF mutations as negative prognostic biomarkers associated with worse survival outcomes in patients undergoing hepatic resection of CLM. Such evidences support the introduction of new treatment decision models, taking into account the tumor molecular profile in order to individualize both systemic and loco-regional treatment strategies.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Colorectal) Cancer 3566

Poster Session (Board #263), Sat, 8:00 AM-11:30 AM

The association between serum folate level and toxicity of capecitabine. First Author: Stephen Lam Chan, The Chinese University of Hong Kong, Hong Kong, Hong Kong Background: During capecitabine treatment, it has been suggested that the rate of toxicity differs between Western and Eastern patients (pt), and serum folate is associated with more toxicity in the Caucasian population. We conducted a prospective study to determine the association between serum folate level and toxicity during capecitabine treatment in Chinese pt. Methods: Key inclusion criteria: colon cancer pt with a plan of capecitabine or capecitabine-based treatment. Exclusion criteria included ECOG PS . 2; concomitant administration of radiotherapy or irinotecan; creatinine clearance , 30ml/min. After consent, relevant data of each pt were documented at baseline and serum was taken for assay of folate level. Pt were started 3-weekly capecitabine 2500mg/m2/day D1-14 alone or capecitabine 2000mg/m2/day D1-14 when combined with oxaliplatin. The toxicity of the first 4 cycles was determined by investigators according to CTCAE 4.0. The primary objective was to determine the association between serum folate level and the rate of $ Grade 2 toxicity. Results: Total 193 patients were recruited from Oct 2013 to Sept 2015, of which 136 of them were eligible. The median age was 60 years (Range 26-82). Tumor stage: Stage II (13; 9.6%); Stage III (78; 57.3%) and Stage IV (45; 33.1%). 36 pt had capecitabine alone and 100 pt underwent capecitabine combinational treatment. The rate of G2 andG3 toxicity was 30.9% and 12.5%, respectively. The G2 toxicity rate (type) was 49.3% (Nausea); 11.8% (Vomiting); 25.0% (diarrhea); 13.2% (mucositis); 5.9% (pigmentation); 25.7% (Hand-foot reaction); 19.1% (neutropenia). The mean serum folate was 27.4nmol/L (range: 12.8-45.4; SD: 27.4). Serum folate level was associated with more $ G2 toxicity (p = 0.0043; OR = 1.086) and remained an independent predictor of $ G2 toxicity (p = 0.0186; OR = 1.072) after adjustment with demographics, renal and liver function parameters. Using cutoff at 75thpercentile, more pts had $ G2 toxicity in the high folate group (91.4% vs. 69.3%; p = 0.0183) than the low folate group. Conclusions: Serum folate level is associated with higher rate of $ G2 toxicity during capecitabine treatment in Chinese population. Further dietary studies are under planning. (Supported by HMRF 01120226).

3568

Poster Session (Board #265), Sat, 8:00 AM-11:30 AM

187s

3567

Poster Session (Board #264), Sat, 8:00 AM-11:30 AM

A randomized multicenter phase II study of FOLFIRI plus either panitumumab (Pmab) or bevacizumab (Bmab) as second-line treatment for wild-type KRAS exon 2 metastatic colorectal cancer (mCRC) with exploratory biomarker analysis by liquid biopsy: WJOG6210G. First Author: Kohei Shitara, National Cancer Center Hospital East, Chiba, Japan Background: The objective of this phase II trial wasto compare efficacy of Pmab plus FOLFIRI and Bmab plus FOLFIRI as second-line chemotherapy for patients (pts) with wild-type (WT) KRAS exon 2 mCRC, associated with comprehensive biomarker analysis. Methods: Pts with WT KRAS exon 2 mCRC refractory to first-line chemotherapy containing oxaliplatin and Bmab were randomly assigned to Pmab plus FOLFIRI or Bmab plus FOLFIRI. The primary endpoint was OS; secondary end points included PFS, ORR, safety and translational research. Results: Of 121 randomly assigned pts, 117 pts were eligible for analysis. OS was almost similar between two arms (HR, 1.16; 95% CI, 0.76 to 1.77) with median OS of 16.2 months in Pmab and 13.4 months in Bmab (Stratified Log-rank P=0.499). Median PFS was 6.0 months in Pmab and 5.9 months in Bmab (HR, 1.14; 95% CI, 0.78 to 1.66). ORR was 46.2% in Pmab and 5.7% in Bmab. KRAS, NRAS or BRAF mutations in circulating tumor DNA were identified in 19 (17.4%) of 109 pts whose baseline plasma samples were available for biomarker analysis. For pts with any of these mutations, Pmab showed worse OS than Bmab (HR, 0.42) meanwhile better trend was observed in Pmab for all WT (HR, 1.21; P for interaction =0.026). Pmab showed better OS than Bmab in patients with low serum VEGF-A (# median) level, (HR 1.92) while Bmab showed better OS among those with high level (HR 0.67; P for interaction=0.016, table). High serum levels of HGF, amphiregulin, EGF and Tenascin C were associated with worse OS among all pts. Conclusions: OS was almost similar with second-line FOLFIRI plus either Pmab or Bmab in pts with WT KRAS exon 2 mCRC. Oncogenic mutation in circulating tumor DNA and serum VEGF-A may be candidate biomarkers to stratify pts, which warrants further evaluation. Clinical trial information: UMIN000005216. Mutation in circulating tumor DNA All Wild

N of patients Median OS (month) HR 95% CI P for interaction

3569

Serum VEGF-A

RAS or BRAF mutant

Low (£ median)

High (>median)

Pmab

Bmab

Pmab

Bmab

Pmab

Bmab

Pmab

46 18.9

44 16.1

8 5.4

11 8.2

30 22.4

24 13.4

24 10.7

1.21 0.74 - 1.99

0.42 0.15 - 1.12

31 13.7 0.67 0.37 - 1.21

1.92 0.98 - 3.76

0.026

Bmab

0.016

Poster Session (Board #266), Sat, 8:00 AM-11:30 AM

Prognostic value of programmed death-ligand 1 (PD-L1) expression in patients with stage III colorectal cancer. First Author: Shigehiro Koganemaru, Department of Medical Oncology, Toranomon Hospital, Tokyo, Japan

Association of poor compliance with adjuvant chemotherapy with poorer survival in patients with rectal cancer: A NCDB analysis (N=14,742). First Author: Zhaomin Xu, University of Rochester Medical Center, Rochester, NY

Background: The programmed death-1 (PD-1)/PD-L1 pathway is a negative feedback mechanism that suppresses the activity of T cells. PD-L1 was expressed on tumor cells (TCs) or tumor-infiltrating immune cells (ICs) in several malignancies. We previously reported that high PD-L1 expression on TCs and ICs were associated with poor and better prognosis in patients with stage IIIb colorectal cancer, respectively. We reported whether these findings were validated in patients with stage III colorectal cancer. Methods: Formalin-fixed paraffin-embedded (FFPE) blocks specimens were obtained from the patients with stage III colorectal cancer who underwent surgical resection and adjuvant chemotherapy at our institution between January 2009 and July 2012. We excluded patients who received radiotherapy or chemoradiotherapy before surgery. PD-L1 expression was evaluated by immunohistochemistry (IHC) on TCs and ICs. Specimens were scored as IHC low (L) or high (H), when , 5% or $ 5% of cells were PD-L1 positive, respectively. Results: Two hundred and thirty-five patients were included in this analysis. Median age was 63 years (range 32-84). One hundred and seventy-four and 61 patients received fluoropyrimidine and fluoropyrimidine combined with oxaliplatin as adjuvant chemotherapy, respectively. The number of patients with PD-L1 expression on TCs/ICs (H/H, H/L, L/H, and L/L) was 0, 19, 36, and 180, respectively. The median followup time of this study was 52.9 months. Patients with high PD-L1 expression on TCs had significantly shorter disease-free survival (DFS) than patients with low expression (hazard ratio [HR] 2.36; 95% confidence interval [CI], 1.21-4.62; p = 0.012). Meanwhile, patients with high PD-L1 expression on ICs were associated with longer DFS than patients with low expression (HR 0.40; 95% CI, 0.16-0.98; p = 0.046). Conclusions: High PD-L1 expression on TCs negatively affect the survival, whereas high PD-L1 expression on ICs is associated with favorable prognosis of patients with stage III colorectal cancer.

Background: Current NCCN treatment guidelines for locally advanced rectal cancer include neoadjuvant chemoradiation, followed by total mesorectal excision and adjuvant chemotherapy. The aim of this study was to identify factors associated with hospital-specific rates of adjuvant chemotherapy and examine the subsequent impact on survival among stage II/III rectal cancer patients. Methods: National Cancer Data Base was queried for clinical stage II/III rectal cancer patients (2006-2011) who underwent neoadjuvant chemoradiation followed by surgical resection. A multivariable logistic regression was used to identify patient and hospital factors associated with receipt of adjuvant chemotherapy. A Cox proportional hazards model was used to estimate the adjusted effect of receiving adjuvant therapy on 5-year overall survival (OS). Results: 14,742 patients were included. Approximately 68% of the cohort did not receive adjuvant chemotherapy. Older age, higher comorbidity score, and complete pathologic response were associated with lower odds of receiving adjuvant therapy. After risk adjustment, there was a 22-fold difference in the rate of adjuvant therapy use between the worst performing hospital and best performing hospital (3.1 to 67.7%). Use of adjuvant therapy was associated with better 5-year OS after controlling for patient confounders (HR = 0.65, 95% CI = 0.59-0.71). Conclusions: Adjuvant chemotherapy was associated with a 35% increased odds of overall survival. Despite this, only a third of eligible patients received adjuvant therapy. There is poor compliance and wide variation across hospitals in adherence to NCCN guidelines. In light of these concerning findings, new interventions such as comprehensive geriatric assessment and physical rehabilitation should be explored to see if they help address this disparity. Cox proportional hazard model: 5-year overall survival stratified by pathologic response. pCR n = 1,727

ypT0-2 n = 5,130

ypT3-4 n = 7,885

ypN0 n = 9,865

ypN+ n = 4,877

% Who received adjuvant chemotherapy 28.0 31.8 33.1 28.6 39.1 Adjuvant chemotherapy (HR) Yes 0.40 (0.23- 0.67) 0.59 (0.48-0.73) 0.73 (0.66- 0.81) 0.60 (0.52-0.69) 0.77 (0.68-0.87)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

188s 3570

Gastrointestinal (Colorectal) Cancer Poster Session (Board #267), Sat, 8:00 AM-11:30 AM

Influence of mRNA expression of fibroblast growth factor 2 (FGF2) in colorectal cancer (CRC) cell lines and in patients with metastatic colorectal cancer (mCRC) treated with FUFIRI or mIrOx (FIRE1). First Author: Arndt Stahler, Klinikum Grosshadern, Munich, Germany

3571

Poster Session (Board #268), Sat, 8:00 AM-11:30 AM

Impact of UGT1A1 genotype on overall survival in Japanese advanced colorectal cancer patients treated by irinotecan-based regimens. First Author: Wataru Ichikawa, Showa University, Tokyo, Japan

Background: Our aim was to evaluate fibroblast growth factor 2 (FGF2) expression in CRC cell lines exposed to 5-Fluorouracil (5-FU) and validate results with clinical data from mCRC patients receiving first-line treatment. Methods: 5-FU IC50 was determined in the CRC cell lines DLD1, LoVo, and HCT-15 applying MTT assays. mRNA expression of FGF2 was measured by RT-qPCR using HPRT as reference gene. Threshold values were determined by minimum p-value method. FGF2 expression was knocked down using sh (short hairpin) RNA in vitro with eGFP as control. In vitroresults were validated by data of the randomized FIRE1 trial (5-FU/LV/irinotecan [FUFIRI] vs. irinotecan/oxaliplatin [mIrOx] using Nanostring. 187 patients were included in this analysis. Results: Decreasing sensitivity of DLD1 [IC50 : 8.8 mM, 95% CI: 6.0 – 13.1 mM], LoVo [IC50 : 10.0 mM, 95% CI: 8.6 – 11.5 mM] and HCT15 [IC50 : 15.4 mM, 95% CI: 13.3 – 17.9 mM] for 5-FU was associated with relative FGF2 expression levels [DLD1: 1.0; LoVo: 154.0; HCT-15: 92.4]. Knocking down FGF2 expression [rel. expression DLD1: sheGFP : 1.00, shFGF2 : 0.02, p = 0.009; LoVo: sheGFP : 1.00, shFGF2 : 0.19, p = 0.032] correlated with a significant decrease of IC50 for DLD1 [100 to 32% (95% CI: 29.9 – 34.8 %), p , 0.001] and LoVo [100 to 64% (95% CI: 55.2% 74.1%), p = 0.001]. In FIRE1, high (n = 89) vs. low (n = 98) FGF2 expression was not correlated significantly with PFS [8.0 vs. 8.2 months, HR: 0.87, 95% CI: 0.65 – 1.17, p = 0.370], but with OS [17.9 vs. 23.5 months, HR: 0.70, 95% CI: 0.51 – 0.96, p = 0.025]. Higher FGF2 expression was significantly associated with worse objective response rate [PR and CR (n = 84) vs. SD and PD (n = 21); 24.85 vs. 31.89, p = 0.049]. Effects in FIRE1 were noticed in patients treated with FUFIRI as well as mIrOx. Conclusions: FGF2 expression might reflect chemosensitivity in CRC cell lines as a knockdown of FGF2 led to a decrease of IC50 for 5-FU in vitro. In FIRE1, high FGF2 expression was associated with lower response rate and overall survival significantly. Interfering the FGF2 system might be a modulator for acquired chemoresistance.

Background: The association between UGT1A1 genotypesand clinical outcomes of irinotecan (IRI)-based regimens have been controversial. We conducted an open-label, multicenter, prospective observational study to evaluate the association between UGT1A1 genotypes and safety/efficacy of IRI-based regimens in Japanese patients (pts) with advanced colorectal cancer (aCRC) (NCT 01039506). We reported previously that the progression-free survival (PFS), the primary endpoint, in the UGT1A1 hetero group were non-inferior to those in the wild group, although the homo group had worse PFS as compared with the wild group (ASCO 2015, Abst No. 3525 and ECC 2015, Abst No. 2100). We present the updated overall survival (OS), the secondary endpoint, in terms of UGT1A1 genotypes. Methods: This study included 1,376 pts who had histologically confirmed aCRC and received IRI-based regimens. UGT1A1 genotypes were categorized into three groups: wild (*1/*1), hetero (*1/*6, *1/*28), and homo (*6/*6, *6/*28, *28/ *28). The non-inferiority of hetero or homo to wild in OS was investigated under the non-inferiority margin of hazard ratio (HR) of 1.25, using propensity score weighting to adjust for baseline characteristics. The association of OS with the degree of IRI dose reduction was also examined. Results: The efficacy analysis was performed using 1,345 pts (data cutoff date of October 10, 2015). The median follow-up time was 15.0 months (m) (IQR 6.9–27.7). The median OS according to UGT1A1 genotypes were 21.3 (95% CI, 18.6–22.6) m in wild, 18.8 m (16.6–21.4) in hetero, 15.4 m (12.4–18.6) in homo. The HRs of hetero to wild was 0.991 (0.858–1.144; non-inferiority p = 0.0016) and that of homo to wild was 1.279 (1.026–1.595, non-inferiority p = 0.8365). The HRs to wild were 1.119 (0.845–1.483, p = 0.4324) and 1.438 (1.056–1.958, p = 0.0421) in the homo group treated with the initial dose 80% or more and with the initial dose less than 80% of IRI, respectively. In patients treated with the initial dose less than 80% of IRI, the HR of homo to wild was 1.419 (0.996–2.023, p = 0.0529). Conclusions: OS in the UGT1A1 hetero group was non-inferior to the wild group. However, the homo group had worse OS as compared with the wild group.

3573

3574

Poster Session (Board #270), Sat, 8:00 AM-11:30 AM

The Nationwide Cancer Genome Screening Project for Gastrointestinal Cancer in Japan (GI-SCREEN): MSI-status and cancer-related genome alterations in advanced colorectal cancer (CRC)—GI-SCREEN 2013-01CRC sub-study. First Author: Takeshi Kajiwara, Department of Gastrointestinal Medical Oncology, Shikoku Cancer Center, Ehime, Japan Background: We initiated a nationwide screening project in Japan to detect rare cancer-related genome alterations in CRC (GI-SCREEN 2013-01-CRC) or non-CRC GI cancer and to facilitate the enrollment of patients (pts) in clinical trials for targeted therapies (Shitara Ket, al ASCO 2015 TPS4134). In this sub-study, we evaluated relationship between MSI status and oncogenic genome alterations as well as clinicopathological features. Methods: Pts with advanced CRC who are or will be treated with systemic chemotherapy were eligible. A total of 36 mutations of KRAS, NRAS, BRAF, PIK3CA in genomic DNA from FFPE tumor specimens were simultaneously analyzed by Luminex (xMAP) technology. Using the MSI Analysis System (Promega) composed of 5 mononucleotide markers, MSI status was analyzed with or without paired DNA from FFPE normal tissue. Results: Among 853 pts enrolled from Feb. 2014 to Jan. 2015, 839 tumor samples have been analyzed for both oncogenic genome alterations and MSI. Mutations in KRAS exon 2, other KRAS or NRAS, BRAF and PIK3CA were detected in 34.0%, 8.0%, 4.6%, and 8.6%, respectively. MSI status was evaluable in 805 pts, of which MSI-H was detected in 15 pts (1.9%). These MSI-H were also detected without normal tissue in this study. Among MSI-H pts, 6 were associated with BRAF V600E mutation, 3 with KRAS exon 2 mutation, and 2 with PIK3CA mutation (overlapping with KRAS exon2). BRAF+MSI-H was associated with older age at presentation and worse OS than non-BRAF +MSI-H (median age, 71.5 vs. 47yrs; median OS, 6.9 vs. 18.1 months).We also found rare germline variants in microsatellite of NR21 and BAT25, and these allele frequencies were 0.33% and 0.25%, respectively. Conclusions: The frequency of MSI-H advanced CRC was 1.9% in Japan. BRAF+MSI-H was associated with worse prognosis. MSI test and multioncogenes mutation analysis were feasible without normal tissue, although rare variant of microsatellite markers may affect the decision of MSI status.

Poster Session (Board #271), Sat, 8:00 AM-11:30 AM

Genomic distinctions between colon and rectal cancer in young patients. First Author: Joshua E. Meyer, Fox Chase Cancer Center, Philadelphia, PA Background: Rectal cancer is increasing in incidence in young patients, with emerging data demonstrating differences when compared to colon cancer in tumor biology. Additional data demonstrate differences in the genetic basis of colorectal cancer (CRC) based on patient age at diagnosis. This study investigated genomic differences between rectal and colon cancers in the context of patient age. Methods: To determine genomic differences in tumors from younger and older patients with CRC, DNA was extracted from formalin-fixed, paraffin-embedded sections from 4699 cases of CRC, representing 698 cases of rectal cancer and 4001 of colon cancer. Comprehensive genomic profiling was performed on hybridization-captured, adaptor ligation-based libraries to a mean specimen median sequencing depth of 688X for all coding exons and selected introns of up to 403 cancer-related genes. Patients were binned by age. P values were calculated using Fisher’s exact test. Odds ratios were calculated and 95% confidence intervals (95% CI) were calculated using standard methods. Results: FAM123B showed more frequent truncating mutations and deletions in patients under age 40 with rectal cancer as compared to colon cancer (3.7:1 in patients age , 40; p = 0.022). In older age groups this ratio decreased and then reversed (age 40-55: 1.3:1; p = 0.425 age 56-75: 1:1.5; p = 0.136 age . 75: 1:4.3; p = 0.227). Other genes with differential truncation and deletion mutation rates between rectal and colon cancer within age-matched cohorts included SMAD2 (p = 0.031), MLL2 (p = 0.027), ETV6 (p = 0.017), ASXL1 (p = 0.012), and ARID1A (p = 0.057). Gene amplification events were compared between rectal and colon cancers and demonstrated differences in FLT3 (OR = 0.59, 95%CI: 0.39-0.91), JAK2 (OR = 0.12, 95%CI: 0.01-1.06), CD274 (PD-L1) (OR 0.07, 95%CI: 0-0.84), PDCD1LG2 (OR = 0.07, 95%CI: 00.84), and PRKC1 (OR = 0.07, 95%CI: 0-0.84). Conclusions: In this analysis, multiple genomic differences were demonstrated between colon and rectal cancers. These lend support to the subdivision of colorectal cancer by anatomic site. These findings also raise possible therapeutic implications of colorectal subsite origin, and warrant further study.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Colorectal) Cancer 3575

Poster Session (Board #272), Sat, 8:00 AM-11:30 AM

Prognostic utility of the presence of extramural vascular invasion in stage II and III colorectal cancer: Data from the FACS randomised controlled trial. First Author: Sian Alexandra Pugh, Southampton University Hospitals Trust, Southampton, United Kingdom Background: Estimates of recurrence after resection of colorectal cancer (CRC) with curative intent form the basis of cancer staging and treatment planning. Adjuvant treatment is recommended in stage III and “high-risk” stage II patients. A key feature commonly used to define patients as “highrisk” is the presence of extramural vascular invasion (EMVI). There is a need to define the prognostic utility of EMVI in an accurately staged prospectively followed-up population managed in the modern era. Methods: Observational analysis of data from the FACS (follow-up after colorectal cancer surgery) trial after 5 years of follow-up. All participants had undergone treatment with curative intent for stage I-III primary CRC, with microscopically clear margins, no evidence of metastases on axial imaging and CEA , 10mg/l following completion of treatment. Data on EMVI were extracted from pathology reports and Cox proportional-hazards regression and Kaplan-Meier curves used to compare survival. Results: EMVI was identified in 21% (150/ 703) of cases and was associated with a higher risk of recurrence (EMVI +ve 38/150, 25%; EMVI -ve 80/553, 15% p = 0.002) which equated to a hazard ratio of 1.92 (95% CI 1.30-2.82). This effect was seen predominantly in stage III (EMVI +ve 31/88, 35%; EMVI -ve 23/140, 16% p = 0.001; HR 2.43 95%, CI 1.42-4.12) but not in stage II disease (EMVI +ve 7/58, 12%; EMVI -ve 44/275, 16% p = 0.45; HR 0.76 95%CI, 0.34-1.69). Overall survival followed the same pattern as DFS: stage II (HR 0.98, 95%, CI 0.551.74 p = 0.94) and stage III (HR 1.98, 95%CI, 1.25-3.13 p = 0.004.) Adjusting for adjuvant chemotherapy did not significantly alter the hazard ratios. Recurrences from EMVI positive CRC were less likely to be surgically resectable (EMVI +ve 8/38, 21%; EMVI -ve 40/80, 50% p = 0.003). Conclusions: Analysis of a cohort that underwent rigorous staging to exclude metastatic disease before trial entry indicates that whilst the presence of EMVI corresponds to a higher risk of recurrence in stage III CRC, this is not seen in stage II disease. This may have implications for decision-making regarding adjuvant therapy and highlights the need for a randomised trial in this setting. Clinical trial information: 41458548.

3577

Poster Session (Board #274), Sat, 8:00 AM-11:30 AM

3576

189s Poster Session (Board #273), Sat, 8:00 AM-11:30 AM

Immune profile and survival outcomes in stage 2 colon cancer. First Author: Peter Gibbs, Royal Melbourne Hospital, Melbourne, Australia Background: Tumor associated immune response impacts outcomes in cancer. In colon cancer (CC), a good immune response, as represented by a dense lymphocytic infiltrate, is known to be associated with improved overall survival (OS). To date studies have used a subjective scoring system and improved OS has been presumed to solely be the consequence of reduced cancer recurrence. The relationship with deficient mismatch repair (dMMR) status, a good prognostic marker that is typically associated with an immune infiltrate, remains unexplored. We examined an objectively determined Immune Profile (IP) and survival outcomes in stage 2 CC. Methods: Stage 2 CC cases were identified from a hospital registry that prospectively records comprehensive point of care data, including recurrence free survival (RFS) and OS. MMR status was determined by immunohistochemistry. The density of CD3 and CD8 T-cells within each tumor was assessed by immunostaining and automated image analysis. A pattern recognition algorithm scored CD3 and CD8 density at the tumor core (TC) and invasive margin (IM). Raw scores for each region (CD3TC , CD3IM , CD8TC and CD8IM ) were added and categorised as IP High or IP Low. Survival analyses used the Kaplan–Meier method and log-rank test. Results: We included 463 subjects with stage II CC, median age 70.4 years, with median follow-up 57.7 months. 93 (16.5%) tumors were dMMR. 220 (47.5%) tumors were categorised as IP High. IP High was associated with improved survival outcomes compared to IP Low, including RFS (HR 0.25, p , 0.001), post-recurrence survival (HR 0.20, p , 0.001), cancer-specific survival (HR 0.08, p , 0.001) and OS (HR 0.10, p , 0.001). The improved RFS for IP high cases was independent of MMR status (dMMR: HR 0.10, p = 0.03; pMMR: HR 0.27, p, 0.001). In patients without recurrence IP High was associated with reduced non-cancer mortality (HR 0.10, p , 0.001). Conclusions: Using an automated and objective measure, we have confirmed that immune infiltration is strongly associated with improved RFS and OS outcomes in stage II CC, independent of MMR status. We have also shown that a good immune response (IP High) is associated with post recurrence survival (where cancer recurs), and with reduced non-cancer mortality in patients that remain recurrence free.

3578

Poster Session (Board #275), Sat, 8:00 AM-11:30 AM

High intra- and inter-tumoral heterogeneity of RAS mutations in colorectal cancer. First Author: David Tougeron, Gastroenterology Department, Poitiers University Hospital, Poitiers, France

Differences in relapse-free survival (RFS) and survival after relapse (SAR) in right (R) versus left (L) stage I-III colon cancer (CCa). First Author: Hagen F. Kennecke, BC Cancer Agency, Vancouver, BC, Canada

Background: Anti-epidermal growth factor receptor monoclonal antibodies (anti-EGFR mAbs) are a standard treatment in metastatic colorectal cancer (mCRC) with wild-type (WT) RAS status. However, approximately 30% of patients with RAS WT mCRC are non-responders to anti-EGFR mAbs, possibly due to undetected tumoral subclones harboring RAS mutations. The aim of this study was to analyze the distribution of RAS mutations in different areas of the primary tumor, metastatic lymph nodes and distant metastasis. Methods: A retrospective cohort of 18 patients with a CRC was included in the study. All tumors were tested for KRAS, NRAS (codon 12, 13, 59, 61, 117 and 146) and BRAFV600E mutations using transparietal sections. Six were KRAS mutant, one NRAS mutant and one BRAF mutant. Multiregion analysis was performed in 60 spatially separated tumor areas according to the pTNM staging, including 49 in primary tumors (submucosa (pT1), musculosa (pT2) and/or subserosa (pT3)), 10 in metastatic lymph nodes and 1 in metastasis. KRAS, NRAS and BRAF mutations were tested using pyrosequencing. Intra-tumoral heterogeneity is defined by the presence of different tumoral subclones within the same tumor mass. Inter-tumoral heterogeneity consists in the presence of different tumor subclones in various tumor sites (i.e. primary tumor, metastatic lymph nodes or metastases). Results: Among the 60 tumor areas, 35 were KRAS mutated (in 11 tumors), 8 NRAS mutated (in 3 tumors) and one BRAF mutated(in 1 tumor). In primary tumors (n = 18), intra-tumoral heterogeneity for RAS mutation was found in 8 patients (44%). Inter-tumoral heterogeneity for RAS mutation between primary tumors and metastatic lymph nodes or distant metastasis (n = 11) was found in 6 patients (54%). Moreover, 5 tumors had multiple RAS mutated subclones in the same tumor (28%). Indeed, we observed the coexistence of different KRAS mutations (n = 3) as well as KRAS and NRAS mutations (n = 4). Conclusions: A high proportion of CRC presented intraand/or inter-tumoral heterogeneity, which has relevant clinical implications for anti-EGFR mAbs prescription. These results suggest the need for multiple RAS testing in different part of the same tumor and/or more sensitive techniques.

Background: Genomic analyses describe significant differences between R and L sided CCa. Four distinct Consensus Molecular Subtypes (CMS1-4) have recently been described: CMS1 is predominantly R, stage II and appears to do well in early stages, but has inferior SAR. CMS2 is predominantly L and has a more favorable SAR. We hypothesize that RFS and SAR outcomes are different in R versus L CCa. Methods: Patients diagnosed with stage I-III CCa were identified from the BC Cancer Agency GI Cancers Outcomes Unit. Disease location, demographic, pathologic, treatment, recurrence and mortality information was prospectively collected. R included cecal to transverse tumors; L included splenic flexure to sigmoid tumors. Kaplan Meier survival analysis was performed by stage and stratified by use of adjuvant chemotherapy. RFS was measured from date of surgery until recurrence and censored at death or last follow-up. SAR was measured from relapse, until death or last follow-up. Cox-proportional hazard models were used to control for baseline clinicopathologic characteristics. Results: Of 5378 patients with stage I-III CCa, 46% were R and 54% L. The median age was 67. Compared to L, R sided was significantly associated with females (52 vs 43%), grade 3 (23 vs 10%), stage II (39 vs 35%), . 12 lymph nodes sampled (60 vs 45%) and no adjuvant chemotherapy (54 vs 47%). 5Y survival outcomes and hazard ratios (HR) are listed. In multivariate analysis, R was associated with superior RFS for only stage II disease, while L experienced better SAR overall and in stages II and III. Conclusions: R vs L sided CCa has a prognostic impact in both early stage and relapsed disease, independent of adjuvant therapy. Until better molecular classification is available, R versus L CCa may be a meaningful reflection of molecular subtypes and represent relevant stratification and prognostic factors. 5Y RFS (%) N Any Stage II III

5378 1946 2840

Any Relapse Relapsed II Relapsed III

1452 631 821

R

L

75 73 85 78 64 66 5Y SAR (%) 7 15 11 21 5 12

HR (95% CI) p-value 0.093 , 0.001 0.245 , 0.001 , 0.001 , 0.001

R vs L

p-value

1.1 (1.0-1.2) 0.322 1.3 (1.1-1.7) 0.007 1.0 (0.8-1.1) 0.598 HR (95% CI) 0.8 (0.7-0.9) , 0.001 0.7 (0.6-0.9) 0.009 0.8 (0.7-0.9) , 0.001

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

190s 3579

Gastrointestinal (Colorectal) Cancer Poster Session (Board #276), Sat, 8:00 AM-11:30 AM

3580

Poster Session (Board #277), Sat, 8:00 AM-11:30 AM

Genetic variants of ATM and XRCC3 to predict efficacy of TAS-102 in patients with refractory metastatic colorectal cancer. First Author: Mitsukuni Suenaga, Division of Medical Oncology, USC Norris Comprehensive Cancer Center, Los Angeles, CA

Genetic variants of hENT-1 to predict efficacy of TAS-102 in patients with refractory metastatic colorectal cancer. First Author: Wu Zhang, Division of Medical Oncology, USC Norris Comprehensive Cancer Center, Los Angeles, CA

Background: TAS-102 is an oral combination of trifluridine (FTD) and tipiracil hydrochloride. Incorporation of tri-phosphorylated FTD into DNA is the main anti-tumor mechanism of action. We hypothesized that homologous recombination (HR) of DNA double-strand breaks (DSB) will modulate TAS102 efficacy. We therefore tested whether genetic polymorphisms in the HR pathway are associated with clinical outcomes in patients with refractory metastatic colorectal cancer (mCRC) treated with TAS-102. Methods: We analyzed genomic DNA extracted from 104 blood samples of two different cohorts: an evaluation set of 52 patients receiving TAS-102 (median age 61 years, male 44%, median follow-up time 6.4 months); and a control set of 52 patients receiving regorafenib without history of TAS-102 treatment (median age 64 years, male 44%, all patients deceased). Ten single nucleotide polymorphisms (SNPs) of seven genes in HR pathway (ATM, BRCA1, BRCA2, XRCC3, FANCD2, H2AX, RAD51) were analyzed by PCR-based direct sequencing for association with progression-free survival (PFS) and overall survival (OS). Candidate SNPs were selected by their frequency and potential function. Results: In univariate analysis, patients carrying any A allele in XRCC3 rs861539 had significant longer PFS (3.8 vs. 2.3 months, HR 0.44, 95%CI: 0.21-0.92, p = 0.024) and OS (15.6 vs. 6.3 months, HR 0.25, 95%CI: 0.08-0.79, p = 0.012) than those with the G/G variant. Patients with any G allele in ATM rs609429 had a longer OS compared to those with the C/C variant (8.7 vs. 4.4 months, HR 0.37, 95%CI: 0.14-0.99, p = 0.022). In multivariate analysis, XRCC3 rs861539 showed marginal significance in PFS (p = 0.091) and OS (p = 0.056). ATM rs609429 remained significant for OS (p = 0.020). No significant differences were observed in the control group. Conclusions: Genetic variants in the HR pathway, ATM rs609429 and XRCC3 rs861539, may serve as predictive and prognostic markers in mCRC patients receiving TAS-102.

Background: Trifluridine (FTD) is an active cytotoxic component of TAS102, and incorporation of tri-phosphorylated FTD into DNA confers its antitumor effect. Recent reports suggest that decreased expression of human equilibrative nucleoside transporter (hENT) and thymidine kinase 1 (TK-1) results in decreased nuclear intake of FTD. We tested whether gene polymorphisms involved in FTD metabolism are associated with outcomes in patients with refractory metastatic colorectal cancer (mCRC) treated with TAS-102. Methods: We analyzed genomic DNA extracted from 104 blood samples of two different cohorts: an evaluation set of 52 patients receiving TAS-102 (median age 61 years, male 44%, median follow-up time, 6.4 months), and a control group of 52 patients receiving regorafenib without history of TAS-102 treatment (median age 64 years, male 44%, all patients deceased). Single nucleotide polymorphisms (SNPs) of genes in FTD metabolism (TK-1, hENT-1) were analyzed by PCR-based direct sequencing. The candidate SNPs were selected by their frequency and potential function. Associations between selectedSNPs and PFS and OS were evaluated by Kaplan-Meier and log-rank tests in the overall population. Cox proportional hazard regression model was used in multivariate analyses. Results: In univariate analysis, median PFS for patients with any hENT-1 rs760370 G allele was 3.5 vs. 2.1 months for those carrying the A/A genotype (HR 0.44, 95%CI: 0.23-0.83, p = 0.004). Median OS was 8.7 and 5.3 months in the two groups, respectively (HR 0.27, 95%CI: 0.10-0.70, p = 0.003). Among patients with any hENT-1 rs9394992 T allele, median PFS was 3.4 vs. 1.9 months for those with the C/C variant (HR 0.48, 95%CI: 0.25-0.91, p = 0.011). Median OS was 8.7 vs. 4.4 months (HR 0.21, 95% CI: 0.08-0.51, p , 0.001) in the two groups, respectively. In multivariate analysis, hENT-1 rs760370 and hENT-1 rs9394992 genotypes remained significantly associated with PFS and OS. No significant differences were observed in the control group. Conclusions: hENT-1 germline polymorphisms may serve as predictive and prognostic markers in mCRC patients treated with TAS-102.

3581

3582

Poster Session (Board #278), Sat, 8:00 AM-11:30 AM

Polymorphisms of genes encoding for vitamin D binding protein and Wnt5a to predict outcome for mCRC patients treated with first-line FOLFIRI and bevacizumab: Data from FIRE-3 trial. First Author: Martin D. Berger, Division of Medical Oncology, USC Norris Comprehensive Cancer Center, Los Angeles, CA Background: Vitamin D mainly exerts its inhibitory influence on the development of colon cancer by inhibiting the canonical Wnt signaling pathway, whereas its role in modulating the non-canonical Wnt pathway is less clear. Additionally, there are preliminary data suggesting that vitamin D may inhibit tumor angiogenesis. We hypothesize that SNPs in genes involved in Vitamin D and non-canonical Wnt5a signaling are associated with outcome in metastatic colorectal cancer (mCRC) patients (pts) treated with first-line FOLFIRI and bevacizumab (bev). Methods: 295 pts with mCRC enrolled in the phase III FIRE-3 trial and treated with first-line FOLFIRI and bev were included in this study. Genomic DNA was extracted from formalin fixed paraffin embedded tissue. 10 functional SNPs in 10 genes (GC, CAMK2B, CYP24A1, CYP27B1, VDR, DKK1, CST5, ROR2, Wnt5a, CASR) were analyzed by PCR-based direct sequencing. Results: Baseline characteristics were as follows: female/male = 99/196; median age = 65y; primary tumor site right/left = 74/215;median PFS/OS = 10.2/24.2 months. G alleles carriers of the GC rs4588 SNP encoding for the vitamin D binding protein showed a longer median overall survival (OS) than those with a TT genotype (25.0 vs 15.9 months) in both univariate (HR 2.18, p = 0.002) and multivariate analysis (HR 1.73, p = 0.043). Pts with a CC genotype of the Wnt5a rs1829556 SNP yielded a higher response rate compared to those harboring any T allele (75% vs 56%, p = 0.006) which also translated into better progression-free (PFS) (11.5 vs 10.1 months, HR 1.45, p = 0.016) and OS (29.0 vs 22.1 months, HR 1.56, p = 0.011). These associations remained significant in multivariate analysis. Conclusions: This is the first report demonstrating that the GC rs4588 SNP encoding for the vitamin D binding protein might serve as a prognostic marker in pts with mCRC. Additionally we provide the first evidence that the Wnt5a rs1829556 SNP might be both a prognostic and predictive biomarker for mCRC pts treated with first-line FOLFIRI and bev. Targeting the Wnt5a ligand and the vitamin D binding protein might be a promising approach to improve our treatment options against mCRC.

Poster Session (Board #279), Sat, 8:00 AM-11:30 AM

Contribution of tumor stroma ratio to a better selection of stage II colon cancer patients into high and low risk groups: A population based study. First Author: Sanne Kjaer-Frifeldt, Department of Pathology, Odense University Hospital, and Danish Colorectal Cancer Group South, Odense, Denmark Background: Despite the well-known uncertainty on efficiency, adjuvant chemotherapy is today offered to a subgroup of stage II colon cancer patients with high risk of recurrence, currently based on T-stage, lymphovascular invasion, malignancy grade, perforation and inadequate lymph node sampling. Recent research has focused on the stroma in epithelial tumors as a prognostic biomarker in various malignant epithelial neoplasms, including colon cancer. Patients with a high percentage of stroma within the tumor, determined by the Tumor-Stroma-Ratio (TSR), show a poor prognosis. Methods: The study included all patients diagnosed with stage II colon cancer in Denmark in 2003. Patients receiving adjuvant chemotherapy were excluded. The TSR was scored on routine H&E tissue sections. Stroma-high (. 50% stroma) and stroma-low (# 50% stroma) groups were evaluated regarding Overall Survival (OS) and Recurrence-Free Cancer Specific Survival (RF-CSS). The currently defined risk groups were stratified according to TSR, and a prognostic index was created using the coefficients obtained from a collective multiple Cox regression, based on RF-CSS data. Results: Of 580 patients, 107 (18.4%) were scored as stroma-high and 473 (81.6%) as stroma-low. TSR showed prognostic impact in both univariate (OS: p= 0.0075, HR = 1.58 (1.14-2.20), RF-CSS: p, 10-5, HR = 2.47 (1.53-3.98)) and multiple COX analyses (OS: p= 0.0007, HR = 1.66 (1.24-2.22), RFCSS: p= 0.0001, HR = 2.22 (1.58-3.27)). When comparing the generated RF-CSS risk index based on traditional high-risk, and including the TSR, the TSR contributed significantly to the index, p= 0.0001. When stratifying according to TSR, stroma high patients showed significantly poorer 5 years RF-CSS in both low (stroma high: 76%, stroma low; 89%; p , 10-5) and high (stroma high: 49%, stroma low: 71%; p = 0.01) risk groups. Conclusions: This population based study confirms the prognostic importance of TSR in both uni- and multivariate analyses. Furthermore, including TSR in the current risk assessment of patients with stage II colon cancer, may contribute to a more appropriate definition of high and low risk groups than todays practice.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Colorectal) Cancer 3583

Poster Session (Board #280), Sat, 8:00 AM-11:30 AM

Association of polymorphisms with tumor response in rectal cancer patients treated with capecitabine +/- oxaliplatine and radiation: Pharmacogenetic analysis of ACCORD-12/PRODIGE-2 trial. First Author: Valerie Boige, INSERM U1147, Paris, France Background: We examined whether 66 germline polymorphisms (SNPs) in 10 candidate genes (ERCC1, ERCC2, ERCC4, GSTP1, MTHFR, SOD2, TYMS, XPA, XRRC1, and XRCC3) would predict clinical outcome in the ACCORD-12 phase III trial which randomly compared neoadjuvant radiotherapy (RT) plus capecitabine (CAP45) with dose-intensified radiotherapy plus capecitabine and oxaliplatin (CAPOX50) in T3-4 Nx M0 resectable rectal cancer. Methods: A candidate-gene association study was conducted in 316 patients (n = 161 in the CAPOX50 and n = 155 in the CAP45 arm). The primary end-point was response according to the Dworak score in each arm. Logistic regressions were used to assess univariate/multivariate associations. The Storey and Tibshirani method based on the control of false discovery rate was used (q-value , 0.10 (adjusted p-value) considered as true discovery). Multivariate models adjusted on treatment arm and T stage were performed to determine prognostic and predictive values of SNPs for tumor response. Results: In univariate analysis, five SNPs in ERCC2 (rs1799787: OR = 3.46 [1.76; 6.79] and rs13181: OR = 0.28 [0.14; 0.58]), XPA (rs3176683: OR = 0.18 [0.05; 0.66]), MTHFR (rs7553194: OR = 4.60 [1.48; 14.3]) and ERCC1 (rs10412761: OR = 0.34 [0.17; 0.68]) were significantly associated with the Dworak score in the CAPOX50 arm, and none in the CAP45 arm. All but one (XPA) were confirmed in the multivariate analysis. Interaction was significant for ERCC2 (rs1799787, p= 0.05) and XPA (rs3176683, p= 0.008), and marginally significant for MTHFR (rs7553194, p= 0.053), suggesting a predictive effect of these SNPs for response to oxaliplatin-based chemoRT, and a prognostic effect of ERCC1 rs10412761 for response to chemoRT (+/- oxaliplatin). MTHFRrs7553194 was associated with overall survival in the multivariate analysis. Conclusions: This pharmacogenetic analysis shows that SNPs in four DNA repair/folate metabolic pathways genes have a significant association with response to chemoRT in patients with locally advanced rectal cancer. The predictive effect of ERCC2 and XPA may identify patients who benefit from oxaliplatin-based chemoRT.

3585

Poster Session (Board #282), Sat, 8:00 AM-11:30 AM

Incidence of colorectal cancer among young adults in the state of Colorado. First Author: Christopher Hanyoung Lieu, University of Colorado Cancer Center, Aurora, CO Background: Colorectal cancer (CRC) incidence has decreased overall, both nationally and across Colorado, since 1990. Since National Comprehensive Cancer Network (NCCN) guidelines recommend screening beginning at age 50 for individuals with average risk, little is known about the CRC incidence in individuals under age 50. However, from 1992 to 2011 there has been a continuous increase in the rate of CRC in patients under 50, which has not been evaluated by geographic region. We examined CRC incidence, focusing on age, in order to better understand CRC in the under-50 population in Colorado. Methods: We analyzed 39,575 CRC cases from the Colorado Central Cancer Registry, examining effects of age, gender, stage at diagnosis, disease location, first treatment received, and length of disease. We calculated the annual percentage change over the 23-year study period (1990 – 2013) to analyze changes in age-adjusted incident rates. Results: The age-adjusted incidence of CRC was lower in individuals under 50, ranging from 4.3 to 6.8 per 100,000 person-years compared to individuals over 50, where incidence ranged from 104.7 to 178.2 per 100,000. However, increasing age-adjusted incidence rates averaging 0.8% per year (p , 0.05) were observed in the under-50 population, while ageadjusted rates fell on average 2.5% per year (p , 0.05) in the over-50 population. Age-adjusted incidence rose in males under 50 by 1.0% per year (p , 0.05) and increased 0.8% per year in females. The increase in incidence appeared to be driven mostly by an increase in stage IV CRC in the under-50 population with an increase of 2.4% per year from 2003 through 2013 (p , 0.05) compared to an increase in early-stage CRC of 0.8% per year. Conclusions: The absolute incidence of CRC continues to fall in the state of Colorado in individuals over age 50. However, incidence is rising in individuals under age 50, particularly among males, and the increase in incidence appears to be driven by an increase in late-stage diagnosis. Additional research into CRC in individuals under age 50 is necessary to better characterize this rise in incidence and possible causative mechanisms.

3584

191s Poster Session (Board #281), Sat, 8:00 AM-11:30 AM

A systems model of BCL-2 dependent apoptosis to predict stage II CRC patients benefiting from adjuvant chemotherapy and as a prognostic tool for stage III CRC patients with increased risk of recurrence. First Author: Jochen Prehn, Royal College of Surgeons in Ireland, Centre for Systems Medicine, Department of Physiology & Medical Physics, Dublin, Ireland Background: Over the past decades the TNM staging system has remained the best indicator for the clinical outcome of CRC patients, and it is used to stratify patients for adjuvant chemotherapy after surgical resection of the tumor. Currently, adjuvant chemotherapy is not recommended for stage II patients. Previously, we developed a systems model of the BCL2 family of proteins (DR_MOMP) to assess cancer cells’ resistance to induce mitochondrial outer membrane permeabilization (MOMP), based on absolute protein levels and the interaction of pro- and anti-apoptotic BCL2 family proteins. Methods: Reverse Phase Protein Arrays (RPPA) were used to determine BAK, BAX, BCL2, BCL(X)L and MCL1 protein levels in primary tumours collected from two distinct cohorts: 1) 138 stage II CRC patients from a completed clinical trial, randomised to adjuvant 5-FU-based chemotherapy or observation only, 2) 128 stage III CRC patients treated with adjuvant 5-FU-based chemotherapy. DR_MOMP was employed to calculate the amount of cellular stress required to induce MOMP based on tumour protein levels. Protein level (RPPA) and mRNA expression (SeqV2 RSEM) data from the TCGA COAD cohort was used to validate the findings. Results: 1) Stage II patients randomised to observation only and classified as high-risk by DR_MOMP had an approximately 2-fold increased risk of death compared to patients receiving chemotherapy or classified as low-risk (HR 2.4; 95% CI 1.2-4.8; p-value = 0.02). 2) Stage III patients treated with chemotherapy and classified as high-risk had a more than 10-fold increased risk of death compared to low-risk patients (HR 10.6; 95% CI 2.4-46.3; pvalue , 0.0001). The hazard ratio (HR) was similar in 261 stage II-IV patients of the TCGA COAD cohort (HR 10.6; 95% CI 1.2-12.5; p-value = 0.01). Conclusions: Our system holds promise as a novel predictive biomarker to stratify stage II CRC patients for adjuvant chemotherapy and as a prognostic biomarker for stage III patients. It could be combined with traditional staging to assess risk and in subsequent clinical management of patients.

3586

Poster Session (Board #283), Sat, 8:00 AM-11:30 AM

Genetic variants of R-spondin genes to predict clinical outcome in mCRC patients (pts) treated with first line FOLFIRI and bevacizumab (FOLFIRI/ BEV) in FIRE-3 cohort. First Author: Yan Ning, University of Southern California Norris Comprehensive Cancer Center, Los Angeles, CA Background: The secreted Wnt agonists of the R-spondin (RSPO) family, RSPO1-4, play important roles as Wnt coactivators by interacting with LGR receptors, leading to stabilization of Frizzles at the cell surface and resulting in various Wnt-related cellular and biological processes, including tumorigenesis. The novel Rspo-Wnt-VegfC-Vegfr3 pathway has been reported as bypass angiogenic pathway during tumor angiogenesis. Also, genomic alterations of RSPO genes were found in CRC. Therefore, we tested whether polymorphisms (SNPs) in RSPO1 (rs4074961, rs1134025), RSPO2 (rs601558, rs555008) and RSPO3 (rs1892172, rs10457487) could predict clinical outcome in mCRC pts treated with first line FOLFIRI/BEV enrolled in FIRE3 trial. Methods: Genomic DNA was isolated from tissue samples of 295 pts treated with first-line FOLFIRI/BEV in FIRE-3 trial. Median follow up: 39.6 mos; Median PFS: 11.3 mos. and Median OS: 25 mos. PCR-based DNA sequencing was used to determine SNPs. Results: Our data showed that RSPOs snps predict clinical outcome by pts tumor location and KRAS status. In left-side tumor pts, the RSPO2 rs555008 T/T genotype was significantly associated with shorter OS [23.7 (18.4-27.5) mos] compared to any G allele [31.9 (25-40) mos] in both univariate (P = 0.011) and multivariate analyses (P = 0.05). The RSPO3 rs10457487 any C [11.7 (10.1-13) mos] was associated with longer PFS compared to A/A [9.1 (7.510.1) mos] in both univariate (P = 0.009) and multivariate analyses (P = 0.024). For KRAS wild type pts, the RSPO2 rs555008 T/T was significantly associated with shorter OS [22.3 (17.5-26.4) mos] compared to any G allele [28.4 (22.7-35) mos] in univariate (P = 0.011). The RSPO3rs10457487 any C allele was significantly associated with longer PFS [10.8 (9.7-12) mos] compared to A/A [9.1 (7.6-10.1) mos] in univariate (P = 0.025) and had a similar trend in multivariate analyse (P = 0.08). Conclusions: This study provides the first evidence that RSPO2 rs555008 and RSPO3 rs10457487 could be prognostic markers for OS and PFS in mCRC pts treated with first line FOLFIRI/BEV. The significant correlation with outcomes was dependent on tumor location and KRAS status.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

192s 3587

Gastrointestinal (Colorectal) Cancer Poster Session (Board #284), Sat, 8:00 AM-11:30 AM

Tumor mutational burden as a potential biomarker for PD1/PD-L1 therapy in colorectal cancer. First Author: Thomas J. George, University of Florida, Gainesville, FL Background: Biomarkers capable of predicting response to checkpoint inhibitor therapies represent a significant clinical need. Increased tumor neoantigenic burden has been linked to PD1/PD-L1 therapeutic response in several conditions including metastatic melanoma, non-small cell lung carcinoma and microsatellite instable (MSI-H) colorectal cancer (CRC). However, the challenges and high cost associated with neo-antigen discovery has shifted focus towards more efficient methods of response stratification. As such, mutational burden determination from whole exome sequencing and comprehensive genomic profiling (CGP) has emerged as a potential solution. In CRC, MSI-H may serve as a predictive biomarker for activity of PD1/PD-L1 therapy, reflecting an unquantified immunogenic mutational burden. Herein, we explore the ability to quantify tumor mutational burden (TMB) as a potential predictive biomarker of PD1/PD-L1 therapy in CRC. Methods: Formalin-fixed, paraffin embedded tissue sections from 2013 cases of CRC were sequenced with CGP (FoundationOne assay). MSI status and PD-L1 gene amplification were determined as previously described. TMB was calculated by counting mutations across a 1.25Mb region spanning 315 genes. Patients were classified as TMB high or low using the top quartile threshold and microsatellite instable (MSI-H) or stable (MSS) using a computational algorithm developed by Foundation Medicine. Results: MSS tumors were observed in 1934 of 2013 (95.2%) cases and MSI-H tumors were observed in 79 of 2013 (3.9%) cases. 79 of 79 (100%) MSI-H cases were TMB high (range 16.8-72.7 mut./Mb) and 1510 of 1934 (78.1%) MSS cases were TMB low (range 0.0-8.0 mut./Mb). Consequently, 424 of 1934 (21.9%) MSS cases were confirmed as TMB high (range 8.843.1 mut./Mb). Of note, less than 0.1% of cases featured PD-L1 gene amplifications. Conclusions: Tumor mutational burden as estimated by CGP, as opposed to MSI assessment alone, significantly increases the number of patients with metastatic CRC who may benefit from checkpoint inhibitor based therapeutic approaches. Investigations to validate this biomarker are ongoing.

3589

Poster Session (Board #286), Sat, 8:00 AM-11:30 AM

Emerging genomic landscape and therapeutic targets in young patients with colorectal cancer (CRC). First Author: Christopher Hanyoung Lieu, University of Colorado, Denver, CO Background: Patients younger than 50 years comprise roughly 10% of the total incidence of CRC, but the incidence in young patients is increasing. This study investigated genomic differences between younger and older CRC tumors to identify potential therapeutic targets for this patient population. Methods: DNA was extracted from formalin-fixed, paraffin-embedded sections from 4699 cases of colorectal cancer (4001 colon and 698 rectal). Comprehensive genomic profiling was performed on hybridization-captured, adaptor ligation-based libraries to a mean specimen median sequencing depth of 688X for all coding exons and selected introns of up to 403 cancerrelated genes. Younger patients with CRC were defined as , 40 years and older patients with CRC were defined as $ 55 years. P values were calculated using Fisher’s exact test; false discovery rate (FDR) and 95% confidence intervals (95%CI) were also calculated. Results: Gene alteration rates in the younger and older cohorts were similar in a majority of the genes analyzed. In the colon cancer analysis, alterations in CTNNB1 (p , 0.001, FDR = 0.08) and MLH1 (p , 0.001, FDR = 0.03) were found to be more common in younger patients compared to older patients. Older patients were more likely to have alterations in APC (p , 0.01, FDR , 0.01) and FAM123B (p , 0.01, FDR = 0.03). In the combined colon and rectal analysis, CTNNB1 (p , 0.001, FDR = 0.07) and MLH1 (p , 0.001, FDR = 0.02) alterations were found to be more common in younger patients with CRC, and APC (p , 0.01, FDR , 0.01) was more commonly altered in older patients with CRC. When focusing on truncating mutations and deletions only, young rectal cancer patients were found to have a log ratio of 3.7 for lesions in FAM123B (p = 0.02, 95%CI: 0.08-0.96) compared to colon cancer. CTNNB1 was found to have a 1.5X rate of mutation in colon compared to rectal cancer, not reaching statistical significance. Conclusions: In this analysis, younger patients and older patients demonstrated similar overall rates of genomic alterations. However, younger patients with CRC exhibited an increased rate of alteration in CTNNB1 and MLH1. These data may provide support to explore the WNT signaling pathway as a potential therapeutic target for younger patients with CRC.

3588

Poster Session (Board #285), Sat, 8:00 AM-11:30 AM

Polymorphisms in toll-like receptor (TLR) genes in the prediction of outcome for cetuximab-based treatment in patients with metastatic colorectal cancer (mCRC). First Author: Satoshi Okazaki, USC Norris Comprehensive Cancer Center, Los Angeles, CA Background: TLR7 and TLR9 signaling pathways are implicated in the regulation of immune system through type-I interferon induction. Immune responses within the tumor microenvironment may influence the efficacy of chemotherapy. TLR7 and TLR9 agonists showed promising results in preclinical and/or clinical trials for cancer patients (pts), in particular in association with cetuximab (cet). We tested the hypothesis that genetic variations in TLR7 and TLR9, and their downstream molecule IRF5 and IRF7, will be associated with outcome in mCRC pts receiving cet-based chemotherapy. Methods: This study included 3 independent cohorts: pts treated with FOLFIRI + cet in the FIRE-3 trial as a discovery set (FIRE3-Cet cohort, n = 299); pts treated with FOLFIRI + bevacizumab in FIRE-3 trial as a control set (FIRE3-Bev cohort, n = 293), pts treated with FOLFOX (or SOX) + cet in JACCRO-CC05/06 trial as a validation set (Japanese cohort, n = 76). Genomic DNA was isolated from tissue samples. 6 single nucleotide polymorphisms (SNPs) in TLR7, TLR9, IRF5, and IRF7 were analyzed by PCR and direct sequencing. These SNPs were tested for the association with PFS and OS. Results: Main patients characteristics’ were the following: FIRE3-Cet cohort M/F 68/32%, median age 64, KRAS-status wild/mutant 82/18%; FIRE3-Bev cohort M/F 66/34%, median age 64, KRAS-status wild/mutant 84/16%; Japanese cohort M/F 58/42%, median age 63, all pts were KRAS-wild type. Median follow-up times were 41.8, 40.9, and 24.7 months, respectively. Among KRAS-wild type pts in the discovery set (FIRE3-Cet), pts with TLR7 rs3853839 G/G variant showed a trend toward longer PFS compared to those with any C (median 10.0 vs. 11.8 months, HR 1.39, P= 0.092). This preliminary association was confirmed in the Japanese cohort, and pts with G/G genotype showed a PFS benefit compared to those carrying any C (univariate: 9.1 vs. 11.6 months, HR 2.04, P= 0.005, multivariate: HR 2.02, 95%CI: 1.14-3.55, P= 0.015). In the control set (FIRE3-Bev), this correlation of TLR7rs3853839 with PFS was not observed. Conclusions: Our findings suggest that TLR7 rs3853839 may predict outcome of cet-based chemotherapy in KRAS-wild type mCRC pts.

3590

Poster Session (Board #287), Sat, 8:00 AM-11:30 AM

Genetic variations associated with cancer cachexia pathways to predict survival in metastatic colorectal cancer (mCRC): Results from FIRE-3 and TRIBE. First Author: Yuji Miyamoto, Division of Medical Oncology, USC Norris Comprehensive Cancer Center, Los Angeles, CA Background: Cancer cachexia is a multifactorial syndrome that commonly affects mCRC patients (pts) leading to progressive functional impairment. Cachexia occurs in 70% of pts with advanced cancer. The Activin/TGF-beta/ SMAD pathway, known as cachexia pathway, plays a critical role in development and progression of CRC. We aimed to evaluate whether single nucleotide polymorphisms (SNPs) of genes involved in the cachexia pathway predict clinical outcomes in bevacizumab (BV) treated mCRC pts. Methods: Genomic DNA was obtained from mCRC pts receiving bevacizumab plus FOLFIRI as first-line treatment and analyzed by using PCRbased direct sequencing. Nine functional SNPs in 5 genes (INHBA, MSTN, ACVR2B, SMAD2 and FOXO3) were tested in a discovery cohort of 294 pts in FIRE3 trial (NCT00433927), then validated in 230 pts in TRIBE trial (NCT00719797). Results: In FIRE3 BV arm, main characteristics were the following: male/female = 194/100; median age = 65; RAS-wildtype/mutant = 198/83; median PFS = 10.2 months; median OS = 24.2 months, median follow-up time = 40.8 months. Among RAS-wildtype pts, SMAD2 rs1792689 A/- (N = 34) achieved a significantly worse PFS compared to G/G variant carriers (N = 157) in the univariate (median PFS 9.2 vs 10.8 months respectively, HR = 1.50 [95%CI 1.02-2.22], p = 0.037) and in the multivariate analysis (HR = 1.58 [95%CI 1.06-2.34], p = 0.023). Among RASmutant pts, ACVR2B rs2268753 T/T (N = 30) showed a significant worse OS compared to C/- variant carriers (N = 46) in the univariate (median OS 8.8 vs 11.2 months respectively, HR = 1.81 [95%CI 1.06-3.09], p = 0.029) and in the multivariate analysis (HR = 2.13 [95%CI 1.19-3.81], p = 0.011). Among female pts, INHBA rs17776182 A/- (N = 31) showed a significant worse PFS compared to G/G variant carriers (N = 59) in the univariate (median PFS 8.8 vs 10.1 months respectively, HR = 1.61 [95%CI 1.012.57], p = 0.045) and in the multivariate analysis (HR = 1.75 [95%CI 1.082.85], p = 0.024). The validation analyses using TRIBE cohort are ongoing. Conclusions: Our study showed for the first time that variations in genes regulating cancer cachexia may be affect prognosis of mCRC pts treated with BV based chemotherapy.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Colorectal) Cancer 3591

Poster Session (Board #288), Sat, 8:00 AM-11:30 AM

3592

193s Poster Session (Board #289), Sat, 8:00 AM-11:30 AM

The Nationwide Cancer Genome Screening Project in Japan, SCRUM-Japan GI-SCREEN: Efficient identification of cancer genome alterations in advanced colorectal cancer. First Author: Takeshi Kato, Kansai Rosai Hospital, Amagasaki, Japan

Association of gene signature to identify molecular subtypes with clinical outcomes of 1st-line cetuximab (cet) treatment for metastatic colorectal cancer (mCRC). First Author: Yu Sunakawa, Division of Medical Oncology, Showa University Northern Yokohama Hospital, Yokohama, Japan

Background: We conduct the nationwide cancer genome screening project in Japan from February 2014. Since February 2015, we introduced Next Generation Sequencing method to detect cancer genome alterations in advanced colorectal cancer (aCRC), called as SCRUM-Japan GI-SCREEN 2013-01-CRC. The objective is to evaluate the frequency of cancer genome alterations in aCRC and to identify patients who are candidate for clinical trial for corresponding targeting agents. Methods: This study is ongoing with the participation of 20 major cancer centers. Patients with aCRC who plan to or receive systemic chemotherapy were eligible. Twenty ng of DNA and 10 ng of RNA were extracted from formalin-fixed paraffin embedded (FFPE) tumor samples and were analyzed by the Oncomine Cancer Research Panel (OCP) which allows to detect gene mutation, copy number variant (CNV) and fusions across 143 genes in a CLIA certified CAP accredited laboratory. The detected genomic variant data were classified according to whether genetic driver of cancer, including gain- and loss-of-function or single nucleotide variant based on the Oncomine Knowledgebase. Results: As of October 31st in 2015, a total of 393 aCRC patients were enrolled and 360 samples were available. Out of 360 samples, 345 samples were analyzed and 15 samples are currently under analysis. The sequence with the OCP was successfully performed in 257 tumors (74.5%). The most frequently detected mutations were TP53 (66.9%), APC (56.4%), KRAS (38.9%), PIK3CA (12.1%), and BRAF (10.9%), and CNVs were ERBB2 (3.5%), FLT3 (2.3%), MYC (1.9%), and FGFR1 (1.6%). No gene fusion was detected so far. Patients with BRAF, PIK3CA or FGFR1 mutation were enrolled in early clinical trials. We plan to conduct the investigator initiated studies targeting aCRC patients with ERBB2 amplification and those with BRAF non-V600E mutations. Conclusions: This nationwide screening system is efficient to detect rare mutations in aCRC. This novel knowledge provides an intriguing background to investigate new target approaches in these patients and represents a progress toward more precision medicine. Clinical trial information: UMIN000016343.

Background: Recently, several groups have reported on classifications of CRC using gene expression data to identify distinct subtypes (Guinney J, et al. Nat med 2015). Some studies found one subtype which has poor prognosis and does not respond to cet (Sadanandam A, et al. Nat Med 2013; De Sousa E Melo F, et al. Nat Med 2013). However, few subtyping studies focused on efficacy of cet as an initial therapy for mCRC. We therefore investigated whether the gene signature will predict outcomes in mCRC patients (pts) received 1st-line cet treatment. Methods: We analyzed total RNA isolated from tissue samples of 77 KRAS wild-type pts enrolled in 2 phase II trials (JACCRO CC-05:UMIN000004197 or CC-06: UMIN000007022) evaluating cet plus oxaliplatin-based chemotherapy as 1st-line treatment. Gene expression levels were measured by HTG EdgeSeq Oncology Biomarker Panel, which is comprised of probes targeting 2560 genes implicated in numbers of pathways, using next generation sequencing for quantitative analysis of targeted RNAs. Univariate Cox regression analysis was conducted for all genes that passed QC filtering. Hierarchical clustering was performed using genes under 0.005 of the Cox pvalue. Results: The patient cohort comprised 57% males, median age of 63 years, 90% of performance status 0 and 15% of right colon cancer. The Cox regression analysis identified 24 genes to be with less than 0.005 of p-value for overall survival (OS). CDX2, which recently identified to be a novel prognostic marker in colon cancer (N Engl J Med 2016), ranked as first gene associated with OS as well as progression-free survival (PFS) (p= 5.56310-5, FDR.p= 0.058 and p= 5.09310-5, FDR.p= 0.053, respectively). The hierarchical clustering using the 24 genes could classify all participants into 3 groups and the Kaplan-Meier curve of the groups had a significant difference in OS (mean 13.6 vs. 19.8 vs. 28.1 months, p= 2.84e-06) but not in PFS. Conclusions: Multigene-expression signature may predict outcomes of 1stline cet treatment in mCRC pts. These preliminary data warrants further confirmation clinical trials: validation of our findings in FIRE3 trial is ongoing.

3593

3594

Poster Session (Board #290), Sat, 8:00 AM-11:30 AM

Primary tumor location and bevacizumab effectiveness in metastatic colorectal cancer patients. First Author: Wen-zhuo He, Sun Yat-sen University Cancer Center; State Key Laboratory of Oncology in South China; Collaborative Innovation Center for Cancer Medicine, Guangzhou, China Background: We previously reported that bevacizumab effectiveness was associated with the primary tumor location of metastatic colorectal cancer (mCRC). However, several important clinical and pathological features were not balanced when patients were grouped by the primary tumor location. Methods: From 2004 to 2013, we identified 740 patients with mCRC treated with oxaliplatin or irinotecan based chemotherapy (CT group) and 244 patients treated with bevacizumab plus oxaliplatin or irinotican based chemotherapy as first-line setting (CT + B group) from Sun Yat-sen university cancer center. We selected patients using propensity score analyses based on unbalanced characteristics (gender, mucinous histology, stage at diagnosis, and lactate dehydrogenase). The primary outcome was overall survival (OS). Kaplan-Meier curves with log-rank tests were used to detect difference. The last time follow-up was conducted at 30th December 2015. Results: 58 right-side colon, 86 left-side colon and 99 rectal cancer patients were included in CT group (total: 243) while 78 right-side colon, 86 left-side colon and 80 rectal cancer patients were included in CT + B group (total: 244). Patients in CT + B group had a similar OS in comparison with CT group only when the primary tumor located at right-side colon (median OS were 20.2 months for CT + B group versus 20.5 for CT group, P = 0.851). For leftside colon cancer patients, those in CT + B group had longer OS than CT group (26.3 versus 23.1 months, P = 0.021). For rectal cancer patients, significantly longer OS were also observed in CT + B than CT group (26.3 versus 21.1 months, P = 0.014). Conclusions: Our data suggested that patients with right-side colon cancer could not get survival benefit from the addition of bevacizumab, and the different response to bevacizumab between right-side and left-side colon cancer were not caused by routinely clinical and pathological features.

Poster Session (Board #291), Sat, 8:00 AM-11:30 AM

Association of family history and survival in patients with colorectal cancer. First Author: Dawn Q. Chong, Massachusetts General Hospital and Harvard Medical School, Boston, MA Background: A family history of colorectal cancer (CRC) in first-degree relatives (FDRs) increases the risk of CRC. However, beyond rare hereditary CRC syndromes, the influence of family history on CRC survival remains uncertain. Methods: We conducted a pooled analysis of 5,010 incident CRC cases from 6 prospective cohort studies and 2 population-based case-control studies within the International Survival Analysis in CRC Consortium (ISACC). Cox proportional hazards models were used to estimate overall survival (OS) and CRC-specific survival (CSS) in relation to family history of CRC in FDRs, and number of affected FDRs, adjusting for age, gender, body mass index, smoking status, use of aspirin/non-steroidal anti-inflammatory drugs, history of screening endoscopy, stage at diagnosis, tumor location and study site. We conducted subgroup analyses by age, gender, tumor location and stage. Results: 819 (16.3%) patients reported a family history of CRC. There were 1,580 (31.5%) total deaths over a median follow-up time of 4.6 years, of which 1,046 (66.2%) were due to CRC. Individuals with a family history of CRC were more likely to have undergone screening endoscopy (p , 0.001), have proximal colon tumors (p = 0.04) and non-metastatic CRC (p = 0.003). Having a family history of CRC was not associated with OS [hazard ratio (HR) 1.03; 95% confidence interval (CI) 0.89-1.19] or CSS (HR 1.13; 95% CI 0.95-1.36) after multivariate adjustment. Compared to patients without a family history, those with $ 2 affected FDRs had a HR of 0.92 (95% CI 0.59-1.43) for OS, and HR of 0.98 (95% CI 0.55-1.76) for CSS. There was no association between age at diagnosis of the affected FDR with OS (ptrend = 0.47) or CSS (ptrend = 0.16). In subgroup analyses, family history was associated with worse CSS in individuals diagnosed at age # 70 years (HR 1.45; 95% CI 1.11-1.89), and in patients with distal colon cancer (HR 1.45; 95% CI 1.03-2.04). Conclusions: Among CRC patients,family history is generally not associated with CRC survival. However, family history may be associated with worse prognosis in individuals diagnosed at age # 70 years, and in patients with distal colon cancer, suggesting a possible distinct pathogenic mechanism underlying a common genetic predisposition.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

194s 3595

Gastrointestinal (Colorectal) Cancer Poster Session (Board #292), Sat, 8:00 AM-11:30 AM

Genetic variants in immune response genes to predict clinical outcome in mCRC patients treated with FOLFIRI/cetuximab (FIRE-3) or with first line cetuximab-based chemotherapy (JACCRO CC-05/06 AR). First Author: Jordan David West, Division of Medical Oncology, USC Norris Comprehensive Cancer Center, Los Angeles, CA Background: Cetuximab can exert its anti-tumor mechanisms by stimulating of the immune system via ADCC (antibody-dependent cell-mediated cytotoxicity). Immune response through T-cell activation may also play a role in anti-tumor efficacy of chemotherapy. Our group previously showed immune response gene polymorphisms (IDO1, CD24, PDL1) were associated with clinical outcome in mCRC treated with cetuximab-based therapy (J Clin Oncol 31, 2013 (suppl; abstr 3567). Here we validated these findings in two independent cohort mCRC treated with cetuximab-based therapy (FIRE3 cetuximab arm and JACCRO CC-05/06 AR). Methods: Genomic DNA was isolated from tissue samples of 2 cohorts of mCRC patients. Cohort 1: 225 patients with all RAS wildtype treated in first-line with FOLFIRI/CET(arm A) from the FIRE-3 trial. Median age 64 yrs, male 71%, median OS 28.7 mos. Cohort 2: 74 Japanese KRASwt pts treated either with 1st-line modified FOLFOX6 (n = 27/57) or SOX (n = 47/67) in combination with CET. 57% males of pts, median age of 63 years and median OS were 33.9 mos, respectively. PCR-based direct Sanger sequencing was used to determine polymorphisms. Results: Our data showed that CD24 rs52812045 AA/GA genotype was significantly associated with shorter PFS compared to GG genotype in multivariate analysis {9.2(6.7,13.8) vs 11.8 (9.4, 18.9) mos, HR: 2.12; p = 0.018} in Japanese cohort, which is consistent with our previous findings. In FIRE3 cohort, IDO1rs3739319 GG genotype was significantly associated with longer PFS compared to AA/GA genotype in multivariate analysis {10.5(8.8, 12.6) vs 9.2(7.9, 10.6) mos, HR: 0.67; p = 0.011}. Also, IDO1rs3739319 showed a trend significance in OS {AA/GA: 33.1(27.6, 41.2) vs GG: 23.920.6, 36.6) mos; HR: 0.70; p = 0.063}. Recursive petitioning analysis of both cohorts showed IDO1 rs 3739319 is the most important snp to distinguish OS. Conclusions: Our data validated previously findings that polymorphisms in immune response gens IDO1 and CD24 may predict clinical outcomes in cetuximab-based therapy. Our findings indicate to combine anti-EGFR and immune checkpoint inhibitors in future clinical trials.

3597

Poster Session (Board #294), Sat, 8:00 AM-11:30 AM

Blood-based biomarkers in patients (pts) with metastatic colorectal cancer (mCRC) treated with FOLFOX or FOLFIRI plus bevacizumab (Bev), cetuximab (Cetux), or Bev plus Cetux: Results from CALGB 80405 (Alliance). First Author: Andrew B. Nixon, Duke University Medical Center, Durham, NC Background: CALGB 80405 was a randomized phase III trial comparing FOLFOX or FOLFIRI + Bev, Cetux or both. Results showed no significant differences between Bev and Cetux in progression-free survival (PFS) or overall survival (OS). The primary objective of our study was to validate 7 biomarkers previously identified to be predictive for outcome for either Bev or Cetux. Methods: Baselineplasma samples were analyzed via multiplex ELISA technology for Ang-2, sCD73, sHER3, HGF, IL-6, SDF-1, and VEGFD. The markers were correlated with PFS and OS using univariate Cox proportional hazards models stratified by chemotherapy (chemo); predictive markers were tested by adding a treatment by analyte interaction term to the model. Exploratory analyses included a chemo by biologic by marker term. Results: Baseline samples were available from 715 KRAS wild type pts treated with Bev or Cetux. No prespecified biomarkers were predictive for OS or PFS after multiple testing correction. However, lower levels of VEGF-D were associated with greater benefit from Bev for PFS (uncorrected interaction p = 0.049; HR 0.69 [95% CI 0.48-1.00]). Based on exploratory analyses, a potential 3-way interaction between chemo, biologic, and VEGFD levels with respect to PFS was observed (p = 0.028). In subset analyses, VEGF-D was found to predict for benefit from Bev in pts treated with FOLFOX (p = 0.0028; HR 0.50 [95% CI 0.31-0.79]), but not in pts treated with FOLFIRI (p = 0.53; HR 1.27 [95% CI 0.59-2.73]). IL-6 was strongly prognostic for PFS and OS (p , 0.0001). Conclusions: While the prespecified criteria for validation of these markers was not met, these data suggest a role for VEGF-D in mediating resistance to Bev, particularly for pts receiving FOLFOX. These findings are consistent with prior observations and highlight the potential importance of alternate VEGF ligands in response to anti-VEGF therapeutics. In addition, IL-6 was found to be strongly prognostic, highlighting the importance of inflammation in mCRC. Additional analyses are ongoing. Support: U10CA180821, U10CA180882. ClinicalTrials.gov Id: NCT00265850

3596

Poster Session (Board #293), Sat, 8:00 AM-11:30 AM

Clinical correlation with codon-specific mutations in metastatic colorectal cancer. First Author: Sophia C. Kamran, Harvard Radiation Oncology Program, Boston, MA Background: Mutational status in colorectal cancer (CRC) impacts response to EGFR inhibition, prognosis, and may affect response to chemoradiation. It is unclear if obesity or other clinical characteristics, including smoking and diabetes, increases the predilection for certain genotypes. The purpose of the following study was to explore the relationship between clinical factors, codon-specific mutations and overall survival (OS). Methods: Metastatic colorectal cancer patients were retrospectively analyzed; clinical characteristics including smoking, diabetes, and body mass index (BMI) were extracted from medical records. Genotyping was performed for . 150 mutations across 15 commonly mutated cancer genes including NRAS, KRAS, PIK3CA, BRAF, and PTEN as part of their clinical management and recorded. Logistic regression was used to establish associations between clinical characteristics and genotypes. Results: 394 metastatic CRC patients were identified. Median age was 58 (range 25-89). 219 (56%) were male. 202 (51%) were non-smokers. 61 had diabetes. 272 patients had a BMI $ 25. 136 patients had KRAS mutations, the most common of which was G12D (37%). Diabetes was associated with a higher incidence of KRAS (odds ratio [OR] 1.9 95% CI 1.1-1.3; p = 0.02) and PIK3CA mutations (OR 2.4 95% CI 1.1-5.0; p = 0.02), but not OS. KRAS mutations were not significantly associated with BMI $ 25 nor OS. BRAF mutations were significantly associated with poorer OS (hazard ratio [HR] 1.9 95% CI 1.32.7; p , 0.001). Current smokers were more likely to have CTNNB1 mutations (OR 11.3 95% CI 1.5-83.2; p = 0.02) and G12D KRAS mutations (OR 5.6 95% CI 1.0-30.3). Current smokers were less likely to have BMI $ 25 (OR 0.44; p = 0.02). Patients with BMI # 18.5 have an association with poor OS (HR 2.6 95% CI 1.1-6.4; p = 0.03), although this needs to be corroborated with a larger sample size. Conclusions: A statistically significant association was found between diabetes and KRAS and PIK3CA mutations. The association of BRAF mutations and decreased overall survival was significant, concordant with previous reports. Smokers were more likely to harbor G12D KRAS mutations, and low BMI (# 18.5) trended toward poorer overall survival.

3598

Poster Session (Board #295), Sat, 8:00 AM-11:30 AM

Comparative molecular analyses of BRAF-V600E mutant tumors: Colorectal cancers (CRC) vs. melanomas. First Author: Shelly Ann Christiansen, Georgetown University Hospital, Washington, DC Background: Little is known about the molecular characteristics of BRAF mutant (mt) CRC. It is unknown whether BRAF mt CRC have molecular and biological profiles similar to BRAF mutant melanomas (Mel). Methods: A total of 5139 tumor samples (CRC, 4007 and Mel, 1132) submitted to Caris Life Sciences for IHC (protein expression), ISH (gene amplification), and NGS sequencing between 2009 and 2015 were retrospectively studied. Chisquare tests determined differences. Results: The rate of BRAF-V600E mutation was 7% in CRC (n = 270), and 30% in Mel (n = 334). Most frequently co-mutated genes in BRAF mt CRC were TP53 (56%), APC (26%), and PIK3CA (19%), and most frequently overexpressed proteins were TOP2A (90%), EGFR (77%), and cMET (57%). Most frequently comutated genes in BRAF mt Mel were CDKN2A (28%), ROS1 (19%), and TP53 (13%), and most frequently overexpressed proteins were PD1+ TILs (75%), and TS (71%). When compared with BRAF mt Mel, BRAF mt CRCs had a greater frequency of TP53 (56% vs. 13%), APC (26% vs. 3%), PIK3CA (19% vs. 1%), and SMAD4 (18% vs. 0%) mutations (all p-values , 0.01); however, mutations in CDKN2A (28% vs. 19%) and ROS1 (19% vs. 12%) appeared at higher rates in mt Mel (statistical significance not reached). BRAF mt CRC had a higher frequency of P-glycoprotein (PGP) (52% vs. 9%), cMET (57% vs. 13%), EGFR (77% vs. 6%), and HER2 (4% vs. 0%) overexpression (all p-values , 0.001), but lower PD1+ TILs (61% vs. 75%; p = 0.012) and ERCC1 (17% vs. 41%; p = 0.004) when compared with mt Mel. Co-occurring RAS mutations were rare, seen in 3 CRC and 2 Mel pts. MEK1 (1/31) and MEK2 (1/31) mutations were detected only in Mel. Mismatch repair deficiency and MSI were seen in 34% of BRAF mt CRC. On examining PD1+ TILs and PDL1 tumor expression in MSI-high (H) and MSI–stable (S) CRC, and comparing with Mel, PD1+ TILs were found in 75% of Mel, 80% of MSI-H CRC, and 56% of MSI-S CRC (Mel or MSI-H CRC vs. MSI-S CRC; p , 0.01). PDL1 was positive in 10% of MSI-H and 15.8% of MSI-S CRC and 15% of Mel. Conclusions: BRAF mt CRCmay carry molecular alterations that are distinct from BRAF mt Mel, suggesting different carcinogenic pathways, and potential resistance mechanisms. Drug combinations that target these alterations (e.g., cMET, EGFR) may warrant further investigation.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Colorectal) Cancer 3599

Poster Session (Board #296), Sat, 8:00 AM-11:30 AM

RAS family mutation patterns in a large cohort of CRCs. First Author: Joanne Xiu, Caris Life Sciences, Phoenix, AZ

3600

195s Poster Session (Board #297), Sat, 8:00 AM-11:30 AM

Molecular markers and survival after recurrence in stage III colon cancers from NCCTG N0147 and NSABP C-08 adjuvant chemotherapy trials. First Author: Frank A. Sinicrope, Mayo Clinic, Rochester, MN

Background: Mutated RAS family members confer resistance to anti-EGFR therapy. We hypothesized RAS family mutations have variable molecular and clinical features including in metastasis of CRC. Methods: We analyzed 3678 CRCs (1992 local tumors and 1685 metastases (mets) including 620 liver, 211 lung, 181 peritoneum, 137 ovary, 19 brain, 15 bone and 4 adrenal) profiled between 2014 and 2016. NextGen sequencing was performed with Illumina MiSeq (TruSeq panel) and NextSeq (Agilent SureSlect XT assay). Microsatellite instability (MSI) was assessed. Results: KRAS mutation occurred in 49% (1791/3640) of CRCs including exon 2 (43.1%), exon 4 (3.3%) and exon 3 (2.7%). Mutation rates were highest in bone (67%), then lung (56%), brain (53%), and no KRAS mutation was found in 4 adrenal mets. Codon 61/146 mutations occurred in brain mets (40%) . local disease (11.6%, p = 0.0234), lung (8.4%), liver (11.5%), ovarian (8%) and peritoneal mets (9.3%). KRAS mutations were more prevalent in MSIlow CRCs (846/1604, 53%) vs MSI-high CRCs (36/101, 36%, p = 0.001). KRAS G12V was the most mutually exclusive of MSI-high (1/199) while 11/ 138 G13D tumors (8%), 14/274 G12D (5%) tumors and 2/8 K117N CRCs were MSI-positive. NRAS mutation occurred in 3.7% (133/3620) of CRCs, frequently in exon 3 (2.0%), then exon 2 (1.7%), but not in exon 4 or 5. Mutation rate was noted in adrenal (1/4, G12D), brain (1/19, G12D) and ovary (5% or 7/137), but not in bone mets (0/15). NRAS Q61 mutations were more prevalent vs G12/13 in CRCs of local disease (55% vs. 37%), liver (59% vs. 41%) and lung (50% vs. 33%) but were lower than codon 12/13 mutations in ovary (43% vs 57%) and peritoneal (20% vs. 80% ) mets. NRAS Q61K mutations were frequent in local disease (27% of NRAS mutations) and liver mets (30%), but were absent in lung mets (0 of 6 NRAS mutations) and rare in MSI-high CRCs (1/99, A18T). HRAS mutation occurred in 0.3% (11/3223) CRCs, 9 in local disease, including 1 G12D mutation, and 8 variants of unknown significance in exon 2/3. Additional 2 A59T mutations occurred in a liver met and a mesenteric LN. Conclusions: Different codon preferences are mutated within the RAS family in CRC. RAS family mutation codon preferences in metastatic CRC warrant further investigation to elucidate key drivers and personalized therapeutics.

Background: The identification of biomarkers that can predict patient survival after recurrence (SAR) of cancer is important to guide clinical decisionmaking. We determined the association of molecular markers in the primary tumor with SAR in patients with stage III colon carcinomas treated with adjuvant FOLFOX-based chemotherapy in two clinical trials. Methods: In randomized trials of adjuvant FOLFOX 6 cetuximab (NCCTG N0147) or FOLFOX + bevacizumab (NSABP C-08), we determined the association of somatic BRAFV600E and KRAS exon 2 mutations and DNA mismatch repair (MMR) status in colon adenocarcinomas with patient SAR. Cox models were utilized and adjusted for clinicopathological features and time-to-recurrence data. The interaction effect of the anatomic site of the primary tumor on the impact of biomarkers on SAR was determined. Results: Among patients with colon cancer recurrence [N0147 (N = 871); C-08 (N = 524)], those with tumors harboring mutant BRAFV600E had significantly worse SAR compared to those with non-mutated BRAF and KRAS genes [HR 2.45 (95% CI, 1.853.25, Padjusted , .0001]. Mutant BRAFV600E was also associated with significantly poorer outcome for tumors with proficient (p) DNA mismatch repair (pMMR) [HR 2.64 (95% CI, 1.96-3.57), Padjusted , .0001] and with deficient (d) MMR [HR 2.70 (95% CI, 1.23-5.93), Padjusted = .014]. Significant interactions were found for MMR (P = .027) and KRAS (P = .014) status by primary tumor site for SAR, with significantly better SAR for dMMR tumors of the proximal vs distal colon [HR 0.57 (95% CI, 0.39-0.81), Padjusted = .023], and significantly worse SAR for mutant KRAS tumors of the distal colon [codon Conclusions: Patients whose tumors harbored BRAFV600E mutations and were dMMR or pMMR had significantly shorter SAR. The significant adverse impact of KRAS mutations on SAR was limited to patients with distal cancers, indicating that KRAS should be interpreted by primary tumor site. These molecular markers may, therefore, inform patient management and treatment decisions at the time of tumor recurrence. Clinical trial number: NCCTG NO147, NCT00079274; NSABP C08, NCT00096278

3601

3602

Poster Session (Board #298), Sat, 8:00 AM-11:30 AM

Poster Session (Board #299), Sat, 8:00 AM-11:30 AM

Association of right-sided colon cancer with poor efficacy of cetuximab in patients with RAS wild-type metastatic colorectal cancer. First Author: Dalyong Kim, Department of Oncology, Asan Medical Center, University of Ulsan College of Medicine, Seoul, South Korea

Population-based assessment of disease-free survival (DFS) as a surrogate endpoint (SE) of 5-year overall survival (5Y OS) in stage I-III rectal (RC) and colon cancer (CC). First Author: Richard M. Lee-Ying, University of Calgary Tom Baker Cancer Centre, Calgary, AB, Canada

Background: Right-sided colon cancer (RC) is distinct from left-sided colorectal cancer (LC) in terms of embryology, pathology and genetics. Previous reports have shown that RC have more KRAS and BRAF mutations and poorer clinical outcomes than LC. We investigated whether primary tumor location affects cetuximab efficacy in patients with RAS wild-type (WT) metastatic colon cancer determined by extended RAS genotyping. Methods: Extended RAS analysis by SequenomMassARRAY technology platform (OncoMap) targeting KRAS (exon 2, 3, 4), NRAS (exon 2, 3, 4), PIK3CA and BRAF (exon 15) was performed in tumors from 307 patients treated with cetuximab as salvage treatment. Tumors with mutated RAS (n = 127) and synchronous primary tumors (n = 10) were excluded. The border of right-sided colon and left-sided colon was defined as splenic flexure. Results: A total of 170 patients were included (RC versus LC, 23 and 147). More mutant BRAF (39.1% vs. 5.4%), poorly differentiated (17.4% vs. 3.4%) and mutant PIK3CA (13% vs. 1.4%) tumor and peritoneal involvement (26.1% vs. 8.8%) were observed in RC. Progression-free survival (PFS) was significantly shorter in RC (2.0 vs 4.9 months, P = 0.004). Overall survival (OS) was 4.1 months in RC and 13.0 months in LC (P = 0.002). In multivariate analysis, BRAF mutations, RC, poorly differentiated histology and peritoneal involvement were associated with shorter overall survival (adjusted hazard ratios, 2.96, 1.63, 2.97 and 2.12, respectively). Conclusions: In RAS WT colon cancer treated with cetuximab as salvage treatment, right-sided colon cancer had a poor survival. Further investigation is required to verify the difference in cetuximab efficacy according to primary site of metastatic colorectal cancer.

Background: 3Y DFS is a validated SE for 5Y OS in adjuvant chemotherapy (AC) CC trials, while no SE have been established for radiation (RT) in RC trials. The objective of this study was to validate 3Y DFS as a SE for 5Y OS in RC and CC. Methods: Patients diagnosed with stage I-III RC and CC referred to any 1 of 5 regional cancer centers in British Columbia between 1990 and 2009 were followed by the GI Cancer Outcomes Unit. Clinical, pathological, and treatment characteristics were prospectively collected, with DFS and OS measured from the date of definitive surgery until relapse or death. RC patients who received RT and CC patients who received AC were propensityscore matched to those who did not, within each cancer center, using an absolute standard difference of #0.2. Individual level associations between DFS and 5Y OS were calculated using Spearman’s rho . Propensity-score adjusted treatment effects, by center were calculated using R2. Results: Of 3153 patients with RC, 2564 received RT and 589 did not. Of 4638 patients with CC 2563 received AC and 2075 did not. Patients with AC and RT were younger and more likely to have grade 3, lymphovascular and perineural invasion, and stage III disease. RC patients also had higher CEA levels and received more CT. After matching, most patients at the extremes of age and with stage I-II disease were removed, resulting in 1544 RC and 2226 CC patients with no significant differences between treatment groups. There was a strong correlation between 2, 3 and 4Y DFS and 5Y OS by Spearman’s rho and R2. Crude recurrence rates for both RC and CC were similar, peaking in the first year and declining to zero after eight years. Conclusions: These data support 3 year DFS as a SE for 5Y OS, for RT in RC and AC in CC. Correlations remained significant at both individual and across-centers for both RC and CC. The recurrence rates for both RC and CC are similar. Association with 5Y OS. RC

2Y DFS 3Y DFS 4Y DFS

CC

Rho (95% CI)

R2 (95% CI)

Rho (95% CI)

R2 (95% CI)

0.91 (0.89-0.92) 0.92 (0.90-0.93) 0.93 (0.92-0.94)

0.88 (0.67-1.10) 0.98 (0.93-1.02) 0.93 (0.80-1.06)

0.88 (0.86-0.90) 0.90 (0.88-0.91) 0.91 (0.90-0.93)

1.00 (0.99-1.01) 0.96 (0.89-1.04) 0.98 (0.95-1.02)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

196s 3603

Gastrointestinal (Colorectal) Cancer Poster Session (Board #300), Sat, 8:00 AM-11:30 AM

Hospitalizations and deaths within 60 days of fluoropyrimidine chemotherapy initiation among Medicare patients with colorectal cancer. First Author: Gabriel Brooks, Dana-Farber Cancer Institute, Boston, MA Background: Fluoropyrimidine (FP) drugs are the basis of chemotherapy for colorectal cancer (CRC). FP chemotherapy carries risk for severe toxicity and death, however the incidence of these complications among elderly patients is poorly described. Methods: We used the SEER-Medicare data to identify adults age $ 65 who received FP chemotherapy (5-FU or capecitabine) for CRC. All subjects were diagnosed with CRC in 2003-11. Intent of chemotherapy was classified as adjuvant for patients diagnosed with locoregional colon cancer who had colectomy and initiated chemotherapy within 180 days of diagnosis. Therapy intent was classified as palliative for patients with metastatic CRC and for patients who initiated chemotherapy for CRC .180 days after diagnosis. Patients with rectal cancer were excluded from the adjuvant therapy group due to overlapping toxicities of FP drugs and radiation. We then identified first hospitalizations and deaths occurring within 60 days of FP initiation. Hospitalizations were considered FPassociated when a diagnosis code for neutropenia, thrombocytopenia, diarrhea, gastroenteritis, stomatitis, or mucositis was present on the hospital claim. Deaths were considered FP-associated when death followed FPassociated hospitalization and occurred within 60 days of FP initiation. Results:See table. Conclusions: 8% of Medicare patients receiving FP chemotherapy for CRC were hospitalized with symptoms of FP toxicity within 60 days of chemotherapy initiation. Further research is warranted to investigate the contribution of evaluable risk factors (including polymorphisms of the DPYD gene) to FP-associated hospitalizations and deaths. Events within 60 days of FP initiation Subjects (n) Hospitalization, any (%) - Hospitalization with FP toxicity (%) Death, any cause (%) - Death after FP-associated hospitalization (%)

3605

Colon cancer, adjuvant Rx (95% CI)

Colorectal cancer, palliative Rx (95% CI)

13,268 18.9 (18.3-19.6) 8.7 (8.2-9.2) 1.9 (1.7-2.2) 0.7 (0.6-0.9)

11,134 24.9 (24.1-25.7) 8.0 (7.5-8.50) 7.1 (6.6-7.6) 1.4 (1.2-1.7)

Poster Session (Board #302), Sat, 8:00 AM-11:30 AM

3604

Poster Session (Board #301), Sat, 8:00 AM-11:30 AM

Identifying SNPs associated with progression-free survival (PFS) and overall survival (OS) in patients with KRAS wildtype and mutant metastatic colorectal cancer (mCRC) using Random Survival Forests (RSF). First Author: Dongyun Yang, USC Norris Comprehensive Cancer Center, Los Angeles, CA Background: Patients (pts) with newly diagnosed mCRC treated with 1st-line bevacizumab (Bev) based therapy have median OS over 2 years. All pts eventually progress: response rates are »60% and PFS »11 months. RAS mutation is a predictive and prognostic biomarker. RAS mt patients don’t benefit from anti-EGFR antibodies; however RAS signaling may play a role in pathways associated with bevacizumab efficacy. We examined whether genes involved in angiogenesis, Wnt, immune, and other related pathways predict efficacy of bevacizumab based chemotherapy (CX) in pts with KRAS wt and mt mCRC. Methods: Among pts with mCRC who enrolled in 2 phase III clinical trials (TRIBE and Fire-3), those treated in FOLFIRI+bevacizumab arm were included. Genomic DNA was extracted from blood samples. 27 single nucleotide polymorphisms (SNPs) were genotyped by PCR-based direct sequencing. PFS and OS were endpoints. RSF was used to identify markers associated with endpoints - stratifying by KRAS. RSF is a robust non-parametric method which does not require the proportionality assumption between covariates and survival, and has built-in algorithm to impute missing data. RSF grows an ensemble of trees to provide stable estimates of prediction accuracy versus a single tree with recursive partitioning. Results: Patient characteristics, outcomes, and associated SNPs by KRAS status are shown in table. T/T of TBK1 was associated with lower PFS and OS rates in both KRAS wt and mt compared to A/A or A/T. Associations between 4 SNPs and PFS and OS varied by KRAS (see Table). Conclusions: TBK1 gene may play a role in disease progression regardless of KRAS, but CCL2, CBP, b-catenin and FOXL2 may be involved in disease progression differently by KRAS. M/F Median age Median follow-up, months Median PFS Median OS TBK1 rs7486100 (immune) CCL2 rs4586 (immune) CBP rs129963 (Wnt) b-catenin rs3864004 (Wnt) FOXL2 rs1192493 (Wnt)

3606

KRAS wt (n = 345)

KRAS mt (n = 141)

220/125 63 45.0 10.5 25.6 PFS, OS PFS, OS PFS, OS

84/57 62 49.9 9.3 23.1 PFS OS PFS, OS

Poster Session (Board #303), Sat, 8:00 AM-11:30 AM

Is the neutrophil-lymphocyte ratio (NLR) a predictive and prognostic factor in rectal cancer patients treated with neoadjuvant chemoradiation (nCRT)? First Author: Ui Sup Shin, Korea Institute of Radiological and Medical Sciences, Seoul, South Korea

Identifying predictive SNPs in patients with metastatic colorectal cancer (mCRC) using Random Survival Forests. First Author: Shu Cao, Department of Preventive Medicine, USC Norris Comprehensive Cancer Center, Los Angeles, CA

Background: Response to nCRT is a key prognostic indicator for rectal cancer and may be influenced by the systemic inflammatory response. The purpose of this study was to determine the relationship between pre- and posttreatment NLR and nCRT treatment response and prognosis among patients with locally advanced rectal cancer. Methods: The primary medical records of a consecutive cohort of localized rectal cancer patients treated with nCRT and surgical resection at a single tertiary institution between 1/1998 to 6/ 2015 were reviewed and the pre- and post-nCRT NLR was recorded. The association of NLR with tumor response and recurrence free survival (RFS) were assessed using logistic (tumor response) and Cox (survival) multivariable regression. Results: A total of 927 patients were analyzed. Median follow-up was 49 months (IQR 23-74). There were 162 (17.2%) patients with complete response (ypT0N0, CR) and 505 (54.5%) with poor response (ypT3/4 or N+, PR). nCRT was associated with an increase in the NLR: mean pre-nCRT NLR = 3.0 (62.4); post-nCRT NLR = 5.6 (65.7) (p , 0.0001). Mean pre-nCRT NLR was lower in CR (2.7 61.6) than non-CR (3.1 62.6) patients, p = .01; however post-nCRT NLR was not different between the response groups. After adjusting for covariates, pre-nCRT NLR $ 5 was significantly associated with lower rate of complete treatment response (OR 0.35, p = 0.027). However, only the post-nCRT NLR, and not the pre-, was associated with RFS (HR; 1.74, 95% confidence interval; 1.26-2.42, p = 0.001). Conclusions: The NLR in rectal cancer patients is both predictive of tumor response to nCRT and prognostic for recurrence free survival. It may therefore be considered for identifying patients eligible for novel organ preserving treatment strategies reliant upon good pathologic response to nCRT and for the delivery of personalized rectal cancer care.

Background: We have investigated many candidate genes in a pathway approach and showed significant association with response, PFS and OS, without understanding their clinical importance when comparing the different pathways. In this study, we used random survival forests (RSF), an ensemble tree method for analysis of survival data, to identify the most predictive genes. Methods: In this study, we analyzed 27 single nucleotide polymorphisms (SNPs) of 24 genes in 745 patients with mCRC receiving first-line FOLFIRI plus bevacizumab in three randomized phase III trials: PROVETTA (n = 220, male/female 129/91, median age 63 years), TRIBE arm A (n = 228, male/female 138/90, median age 60 years) and FIRE3 arm B (n = 297, male/female 195/102, median age 65 years). The SNPs were involved in the following pathways: tumor budding, angiogenesis, HIPPO, macrophage and autophagy. In RSF procedure, 1000 trees were constructed, with each tree constructed on a bootstrap sample from the original cohort. The variables were selected by Variable Importance and Minimal Depth. Results: Patient characteristics and outcomes are shown in the table below. In overall patients, CBP (rs129963) in Wnt pathway was identified to be the important predictor for PFS, ß-catenin (rs3864004) in tumor budding/Wnt pathway and DSCR1 (rs6517239) in HIPPO pathway were identified to be predictive for OS. In male, DSCR1 (rs6517239) was important for both PFS and OS. In female, PDGFRB (rs2302273) and HRG (rs2228243) from macrophage pathway were found predictive for PFS. HRG (rs2228243) was also predictive for OS. In patients with tumor on rightsided colon, EPS15 (rs17567) was predictive in PFS. In patients with tumor on left-sided colon, CBP (rs9392) and CBP (rs129963) were identified for PFS and OS, respectively. Conclusions: Our study identified ß-catenin, DSCR1, CBP, PDGFRB, HRG and EPS15 may serve as prognostic markers in mCRC patients treated with bev based chemotherapy.

Median Median Median Median

age, years follow-up, months PFS OS

Total patients (n = 745)

Male (n = 462)

Female (n = 283)

Right-sided (n = 178)

Left-sided (n = 492)

62 49.6 10.3 26.1

64 50.4 10.3 25.4

61 49.6 10.3 27.3

62 44.9 9.4 21.5

62 49.3 10.9 30.3

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Colorectal) Cancer 3607

Poster Session (Board #304), Sat, 8:00 AM-11:30 AM

3608

197s Poster Session (Board #305), Sat, 8:00 AM-11:30 AM

Association of FBXW7 missense mutations (mt) with unfavorable prognosis in metastatic colorectal cancer (mCRC) patients (pts). First Author: Krittiya Korphaisarn, The University of Texas MD Anderson Cancer Center, Houston, TX

Assessment of tumor-infiltrating lymphocytes and immune-checkpoints expression in metastatic colorectal cancer patients. First Author: Aurelien Gobert, Pitie-Salp ´ etri ˆ ere ` University Hospital, Medical Oncology, Paris, France

Background: FBXW7 functions as a ubiquitin ligase tagging dominant oncogenic proteins such as cyclin E, MYC, Jun and Notch for degradation by the proteasome system. FBXW7mt are noted throughout the gene and have been shown to have differential effects depending on location and type. Recent data suggests that missense mt lead to greater loss of FBXW7 function than nonsense or truncation mt but the clinicopathological factors associated with FBXW7 missense mt and clinical outcomes in mCRC have not been described. Methods: Data from pts with mCRC whose tumors were evaluated by next generation sequencing for hotspot mt through the Assessment of Targeted Therapies Against Colorectal Cancer (ATTACC) program at MD Anderson Cancer Center were obtained. Alterations in FBXW7 were identified and their implications on clinical, pathological features and overall survival (OS) were evaluated. Results: Of 854 mCRC pts in the ATTACC dataset, 574 had data on FBXW7 of which 45 pts (7.8%) had mt. Of these, 38 were missense mt while others were nonsense (5), insertions or deletions (1 each). R465H mt in exon9 and R497Q mt in exon10 were the most common missense mt (15.8% each). None of the clinicopathological features including age, gender, tumor site & grade, KRAS, BRAF, PIK3CA mt or microsatellite status were associated with FBXW7 missense mt. Pts with FBXW7 missense mt tumors had significantly inferior OS compared with pts with FBXW7 wild type tumors in both univariate (median OS 36.2 months, mos 95% confidence interval, CI 18.7-38.5 in mt vs median OS 47.5 mos 95% CI 44.2-50.9 mos; p = 0.003) and multivariate analyses after correcting for other known prognostic factors including KRAS, BRAF status and tumor grade (HR = 1.8, 95%CI: 1.14-2.86; p = 0.01). In contrast, other alterations in FBXW7 did not appear to impact prognosis (median OS 42.9 mos, 95% CI 21.1 – 64.8 mos). Conclusions: To date, this is the largest clinical dataset of FBXW7 missense mt which showed negative prognostic impact on survival in mCRC without any association with other known prognostic factors. Further studies on identification of downstream pathways underlying this worse prognosis and potential therapeutic targets are required.

Background: The immune system is known to be involved in oncogenesis and has become a target as confirmed with the immune-checkpoints (ICP) inhibitors in several cancer types. Here we evaluated the expression of ICP and lymphocyte infiltration in a metastatic colorectal cancers with MicroSatellite-Stability phenotype population, to better understand the potential mechanisms underlying a lack of response with ICP inhibitors in this population. Methods: We analyzed the correlation between the quantity and location of lymphocytic infiltrate (primary and metastasis) and ICP expression (PD-1, PD-L1, PD-L2, TIM-1) on lymphocytes, tumoral and endothelial cells with known prognostic factors in a large cohort of colorectal cancers treated in our institution between January 2009 and January 2015. For this purpose, a tissue-microarray composed of 130 samples related to 93 patients has been constructed. The results of TIL and ICP were scored as negative/low or high. Results: This immune assessment was usable for 76 patients, with 61 primary tumors and 35 metastastic sites including liver (n = 18), peritoneum (n = 11) and others (n = 5). There was no difference of TIL expression between primary tumors and metastatic sites (p = 0.79). Surprisingly, higher TIL and CD4/CD8 ratio were observed in liver metastases as compared to other sites (p = 0.0097). PD-L1 expression on lymphocytes was found to be correlated with a T4 tumor infiltration in comparison with less infiltrated tumor (p , 10^3), but there was no difference according lymph nodes involvement. The TIM-1 expression was found on tumor cells and was associated with a younger age than 60y.o. (p = 0,0278) and with a primitive located in the right colon (p = 0,045). Of note, 4 out of 17 cases with available matched primary-liver metastasis gained PD-1 expression associated with an increase of lymphocytic infiltration in metastastic sites. Conclusions: To our knowledge, this is the first study reporting the expression of these immune checkpoints and TIL in a metastatic colorectal cancer population. The TIM-1 and PD-L1 expressions were respectively associated with a younger age at diagnosis, a right colon involvement and also a locally more infiltrating tumor.

3609

3610

Poster Session (Board #306), Sat, 8:00 AM-11:30 AM

Disentangling the association between statins, cholesterol, and colorectal cancer: A nested case-control study. First Author: Ronac Mamtani, Abramson Cancer Center of the University of Pennsylvania, Philadelphia, PA Background: Prior studies have found an association between statin use and reduced risk of colorectal cancer (CRC). We hypothesized that these findings may be due to systematic bias, and examined the independent association of CRC risk with statin use, serum cholesterol, and change in cholesterol concentration. Methods: 22,163 colorectal cancer cases and 86,538 matched controls between 1995 and 2013 were identified within a UK primary care database. Conditional logistic regression estimated CRC risk with statin use, serum total cholesterol (mmol/L), and change in total cholesterol level. Results: We confirmed a decreased risk of CRC with statin use (long-term: OR, 0.95; 95% CI, 0.91-0.99; short-term: OR, 0.92; 95% CI, 0.85-0.99). However, to assess whether the observed association may result from indication bias, a subgroup analysis was conducted among patients prescribed a statin. In this subgroup (n = 5,102 cases, n = 19,032 controls), the risk of CRC was not significantly different among those who continued statin therapy and those who discontinued (OR, 0.98; 95% CI, 0.79-1.22). Increased serum cholesterol was independently associated with decreased risk of colorectal cancer (OR, 0.89 per mmol/L increase; 95% CI, 0.87-0.91). Decreases in serum total cholesterol . 1 mmol/L $ 1 year prior to cancer diagnosis were associated with subsequent CRC (statin users: OR, 1.25; 95 CI%, 1.03-1.53; nonusers: OR, 2.36; 95 CI%, 1.78-3.12) Conclusions: Although the risk of CRC was lower in statin users versus nonusers, no difference was observed among those who continued versus discontinued statin therapy, suggesting the potential for indication bias. The association between decreased serum cholesterol and CRC risk suggests a cholesterol-lowering effect of undiagnosed malignancy. Further studies are needed to assess the clinical utility of serum cholesterol as a biomarker for risk or early detection of CRC.

Poster Session (Board #307), Sat, 8:00 AM-11:30 AM

Re-evaluating the optimal threshold for lymph node (LN) harvest in colon cancer: Insights from a population-based study. First Author: Joseph Del Paggio, Division of Cancer Care & Epidemiology, Cancer Research Institute, Queen’s University, Kingston, ON, Canada Background: Colorectal cancer guidelines recommend $ 12 LNs be harvested during resection; however, this threshold is based on limited evidence. While numerous studies show that LN yield is associated with survival, the optimal threshold is inconsistent in the literature. Here, we evaluate thresholds associated with both LN positivity and survival using a population-based dataset. Methods: Treatment records were linked to the Ontario Cancer Registry to identify all colon cancer patients during 20022008. Surgical pathology reports were reviewed for a random 25% sample. Modified Poisson regression was used to identify factors associated with LN positivity, and Cox proportional Hazards model was used for survival analysis. Sequential regression analysis using multiple thresholds and Pearson/ martingale residuals were used to evaluate the optimal threshold for LN positivity and survival. Results: 25613 patients underwent colon cancer resection. 7519 (~25%) randomly selected cases were reviewed; 5508 met eligibility criteria. On adjusted analysis, younger age (p , 0.001), left sided tumours (p = 0.003), higher T stage (p , 0.001), and greater LN harvest (p = 0.007) were associated with greater likelihood of LN positivity. Regression analyses with multiple cut-points suggested that beyond 12-14 LN there was little incremental increase in LN positivity. Results were consistent on Pearson residual plot. Cox model analysis showed that increased LN harvest was associated with improved cancer-specific survival for stage 2 and 3 disease. Sequential Cox analyses with multiple cut-points in stage 2 disease suggested no gain in survival beyond 20 LNs; for stage 3 disease, no threshold effect was observed. Martingale residual analysis confirmed the thresholds. Conclusions: While greater LN harvest is associated with improved outcome in colon cancer, these data show a disconnect between the number of LNs needed to accurately determine LN positivity, and the optimal LNs needed for improved survival. Although the historically stated threshold of requiring a LN harvest of $ 12 may insure accurate LN staging, the threshold for optimal survival requires a significantly greater harvest.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

198s 3611

Gastrointestinal (Colorectal) Cancer Poster Session (Board #308), Sat, 8:00 AM-11:30 AM

Lymph-node yield in 23,773 rectal cancer patients undergoing surgery between 2000 and 2011 and its impact on staging accuracy - Results from a nationwide observational study. First Author: Benjamin Garlipp, Otto von Guericke University Hospital, Magdeburg, Germany Background: Lymph node (LN) sampling in rectal cancer surgery has a prognostic impact, possibly because more pN+ cancers are correctly diagnosed if more LN are examined. However, despite increasing numbers of LN per specimen, the proportion of pN+ cases has remained largely constant in recent years. We examined the development of LN sampling in rectal cancer in Germany within a 12-year period. Methods: Data from the prospective German Quality Assurance in Rectal Cancer observational study were used. All patients undergoing radical surgery without neoadjuvant treatment from 01/01/2000 to 12/31/2011 were included. The number of LN (total and tumor-involved) per specimen, the proportions of specimens with $ 12 LN examined and of pN+ cases were analyzed for each year. To compensate for factors that may have impacted on LN diagnostics over the study period, results were compared between hospitals that did and did not report an increase in LN number per specimen between 2000/2001 and 2010/2011. Results: A total of 23.773 patients were analyzed. The mean number of LN per specimen increased from 16.2 to 20.8 during the study period, as did the proportion of specimens with $ 12 LN examined (from 73.6% to 93.2%; p , 0.001). pN+ was diagnosed in 44.1%, with nonsignificant variability during the study period (p = 0.130). The mean number of tumor-involved LN varied between 2.2 and 2.6 (p = 0.331). Examination of . 12 LN did not lead to a further increase in tumor-involved LN found in pT2 and pT4 specimens (p = 0.256 and p = 0.655). Only in pT3 specimens with more than 20 LN examined, a slight increase in positive LN was observed compared to specimens with only 12 examined LN. Neither in centers that did nor in those that did not report an increase in LN examined per specimen was there a significant change in pN+ cases observed during the study period. Conclusions: LN sampling became more extensive during the study period but this did not lead to significant tumor upstaging. Evaluation of . 12 LN only marginally reduces the number of missed pN+ cases. Mechanisms other than staging accuracy may be responsible for the beneficial effect of extensive LN sampling reported in studies.

3613

Poster Session (Board #310), Sat, 8:00 AM-11:30 AM

3612

Poster Session (Board #309), Sat, 8:00 AM-11:30 AM

Open versus laparoscopic surgery for low rectal cancer: a large multicenter cohort study in Japan. First Author: Koya Hida, Kyoto University, Kyoto, Japan Background: Laparoscopic surgery for rectal cancer is widely performed all over the world and several randomized controlled trials have been reported. However, the usefulness of laparoscopic surgery compared with open surgery has not been demonstrated sufficiently, especially for the low rectal area. Therefore, a large multicenter cohort study with more than 1000 cases of low rectal cancer was planned to settle this issue. Methods: The data of patients with clinical stage II–III low rectal cancer below the peritoneal reflection were collected and analyzed retrospectively. The operations were performed from 2010 to 2011 and the cases were followed up until 2015. Short-term outcomes and long-term prognosis were analyzed. Results: A total of 1608 cases were collected from 69 institutes, and 1500 cases were eligible for analysis. The cases were matched using propensity scores (482 open cases and 482 laparoscopic cases). The median follow-up periods of each group were 3.6 years and 3.4 years, respectively. The median age was 64 years and the proportion of men was 69%. The mean height of the tumor from the anal verge was 4.6 cm. Preoperative treatment was performed in 35% of the patients. The conversion rate from laparoscopic to open surgery was 5.2%. Estimated blood loss during laparoscopic surgery was significantly less than that during open surgery (90 ml vs 625 ml, p , 0.001). Overall, the occurrence of complications after laparoscopic surgeries was less than that after open surgeries (30.3% vs 39.2%, p = 0.005), and the proportion of anal preservation was higher in the laparoscopic group than in the open group (60.0% vs 53.3%, p = 0.037). Three-year overall survival rates of R0 resection cases were 91.7% (95% CI 88.6—94.0) and 92.0% (89.0—94.2) in the laparoscopic and open groups, respectively, and no significant difference was shown between the two groups. No significant difference was observed in relapse-free survival between the two groups (72.1%, 67.6—76.1 vs 75.1%, 70.8—79.0). Conclusions: Even for advanced, very low rectal cancer below the peritoneal reflection, laparoscopic surgery could be considered a useful option based on the short- and longterm results of our large cohort study.

3614

Poster Session (Board #311), Sat, 8:00 AM-11:30 AM

A phase II study of oxaliplatin/5FU/bevacizumab and concurrent pelvic radiation in patients with simultaneous primary and metastatic rectal cancer: The Chrome-B trial. First Author: Michael Michael, University of Melbourne, Melbourne, Australia

Genotype-driven study of weekly irinotecan and capecitabine in combination with neoadjuvant radiotherapy of locally advanced rectal cancer. First Author: Zhen Zhang, Fudan University Shanghai Cancer Center, Shanghai, China

Background: Current chemotherapy regimens used during chemoradiation (CRT) are adequate for radiosensitization but suboptimal for systemic control. The aim of this study was to assess tolerability, and local/systemic benefits of a new regimen delivering intensive chemotherapy plus bevacizumab (Bev) and radical radiotherapy in an interdigitating manner. Methods: Phase II study for patients (pts) with untreated simultaneous symptomatic primary & metastatic rectal cancer. Pts received, in 12 weeks, 3 courses of FOLFOX-Bev and pelvic radiation 50.4Gy with concurrent Oxaliplatin/5-FU/Bev. A) FOLFOX chemotherapy: Oxaliplatin (Ox) 100mg/ m2, Leucovorin 200mg/m2, 5FU 400mg/m2 bolus, all day 1, and 5FU CI 2.4g/m2/46 hours given in weeks 1, 6, and 11. B) Bev 5mg/kg in weeks 1, 3, 5, 9, 11. C) Pelvic CRT: 25.2Gy in 3 weeks, 1.8Gy/fr, with concurrent Ox 85mg/m2 day 1 and 5FU CI 200mg/m2/day given in weeks 3-5, and 8-10. All pts staged with CT, MRI and FDG-PET before and post treatment. The primary endpoint was tolerability. Results: 30 pts treated. The mean age: 57 years. Rectal primary MRI stages: T2 3%, T3 73% and T4 24%. All pts had metastasis: liver 80%, lung 10%. 42% of pts had $ 3 sites of metastatic disease. 24 pts (80%) completed the 12 week treatment regimen and the planned radiation dose. Overall 80% of pts received the planned number of Ox courses. 77% and 80% of pts received $ 75% of protocol Ox and Bev dose. Treatment related grade (Gr) 3 toxicities were: neutropenia 40%, diarrhea 7%, radiation perineal skin reaction 13%, 1 patient each with hypertension and enterocolits. Gr 4 neutropenia in 23%. Febrile neutropenia in 3 (10%) pts, no septic deaths. The overall CT/MRI response rates for rectal primary and metastatic disease were 62% and 48% respectively. The PET metabolic response rate for rectal primary (CR+PR) was 100% (CR 39%) and for metastatic disease 90% (CR 21%). Conclusions: It is feasible to deliver intensive chemotherapy plus bevacizumab and radical radiotherapy in an interdigitating manner treating both primary and metastatic rectal cancer simultaneously. High response rates are encouraging. This treatment design warrants further investigations. Clinical trial information: ACTRN12611000033943.

Background: In our previous study, The MTD of weekly irinotecan was escalated to 80 mg/m2 in patients with the UGT1A1 6/6 genotype and to 65 mg/m2 in those with the 6/7 genotype. This is a sequent platform study for patients with 6/6 genotype. Methods: Patients with rectal cancer clinical stage T3-4, N0-2 eligible for preoperative chemoradiotherapy were screened for the UGT1A1 genotype. The dose of weekly irinotecan was 80 mg/m2 for the 6/6 genotype patients, whereas the dose of capecitabine was fixed to 625 mg/m2 bid d1-5 weekly. Concurrent intensity-modulated radiation therapy was given to the pelvis (50 Gy/ 25Fxs). Results: A total of 52 patients (median age, 54 years) were enrolled. Fifty-one patients completed radiotherapy schedule as planned and 42 (81%) patients completed more than three cycles of weekly irinotecan. Forty-three patients (83%) underwent radical surgery and R0-resection was accomplished in all patients. Five of the 9 patients refused to have surgery and accepted watch and wait strategy after achieving clinical complete response. Of the 43 resected patients, pathological complete response was observed in 13 (30%), and another 15 showed microfoci of residual tumour (35%). Grade 3/4 leucopenia/ neutropenia occurred in 11 patients (21%), and the main grade 3 nonhematologic toxicity was diarrhea (n = 9, 17%). Conclusions: Our data exhibits manageable toxicities and an encouraging high pCR rate in the UGT1A1 6/6 genotype cases. A phase III trial is ongoing to evaluate the value of irinotecan in neoadjuvant therapy (NCT02605265). Clinical trial information: NCT01474187.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Colorectal) Cancer 3615

Poster Session (Board #312), Sat, 8:00 AM-11:30 AM

3616

199s Poster Session (Board #313), Sat, 8:00 AM-11:30 AM

Results of the phase II TRUST trial of induction treatment with FOLFOXIRI + bevacizumab (BV) followed by chemo-radiotherapy (CRT) plus BV and surgery in locally advanced rectal carcinoma (LARC). First Author: Gianluca Masi, Azienda Ospedaliero-Universitaria Pisana, Istituto Toscano Tumori, Pisa, Italy

Intergroup randomized phase III study of postoperative oxaliplatin, 5-fluorouracil and leucovorin (mFOLFOX6) vs mFOLFOX6 and bevacizumab (Bev) for patients (pts) with stage II/ III rectal cancer receiving pre-operative chemoradiation. First Author: Al Bowen Benson, Robert H. Lurie Comprehensive Cancer Center of Northwestern University, Chicago, IL

Background: Induction chemotherapy (CT) is a promising option in LARC. FOLFOXIRI + BV is an effective treatment in metastatic colorectal cancer. Methods: This is a phase II multicentric single-arm single-stage trial. Primary endpoint was 2-year disease-free survival (DFS). Patients (pts) with LARC at , 12 cm from the anal verge, cN+ or cT4 or high risk cT3 (MRI criteria) underwent 6 cycles of FOLFOXIRI (irinotecan 165 mg/m2, oxaliplatin 85 mg/m2, folinate 200 mg/m2 and 5FU 3200 mg/m2 in 48 h) + BV (5 mg/kg) every 2 weeks followed by CRT (50.4 Gy + 5FU 225 mg/m2/day or capecitabine 825 mg/m2/bid continuously + BV 5 mg/kg on days 1, 15, 28). Surgery was planned 8 weeks after CRT. Results: We enrolled 48 pts. Main characteristics were: median age, 53 years (30-74); cT2/T3/T4, 4%/60%/ 36%; cN0/N+, 4%/96%. 46 pts completed induction CT: 1 patient (pt) died due to bowel perforation and sepsis and 1 discontinued CT after acute kidney injury. Main grade (G) 3/4 toxicities during induction were neutropenia (42%), febrile neutropenia (4.2%), diarrhea (12.5%). Clinical Response Rate (CRR) after induction was 77%. One pt underwent surgery after induction CT and 45 started CRT. After the first 13 pts, the protocol was amended and the schedule of capecitabine modified (800 mg/m2/bid 5 days/week) due to an excessive rate of G3 hand-foot syndrome (23%) and proctitis (23%). After amendment all pts completed CRT with acceptable toxicity: in particular G3-4 toxicities included only 6.2% proctitis. CRR after CRT was 78%. One pt had early progressive disease (PD) after CRT and died. 44 pts underwent surgery: low anterior resection 90%, abdomino-perineal resection 7%. R0 resection was achieved in 98% of pts. Early (90 days) postsurgical complication rate was 31% with 17% of anastomotic dehiscences, all solved. Pathologic complete response rate was 36%. At a median follow up of 21 months, 8 pts had PD (2 local recurrences) and the estimated 2yDFS is 78%. Conclusions: Induction CT with FOLFOXIRI + BV followed by CRT is feasible and highly active in LARC. The rate of early post-surgical complications is not negligible. Preliminary results for DFS are promising. Clinical trial information: 2011-003340-45.

Background: Metastatic colorectal cancer trials have shown improved outcomes when Bev was added to chemotherapy. E5204 was designed to test whether the addition of Bev to postoperative mFOLFOX6 could improve OS in pts with stage II/ III rectal cancer after preoperative chemoradiation. Methods: Eligible pts had rectal adenocarcinoma , clinical Stage T3N0M0, T4N0M0, TanyN1-2M0 before neoadjuvant therapy; had received preoperative chemoradiation (may have included oxaliplatin); had a completely resected tumor and able to begin adjuvant treatment by day 56. Pts were randomized to mFOLFOX6 (Arm A) or mFOLFOX6 and Bev (5mg/kg) (Arm B) administered every 2 wks for 12 cycles. Results: E5204 was terminated on 4/29/09 due to poor accrual. 355 pts were registered (17% of accrual goal). Median follow-up was 72 mos. There was no significant difference in OS between arm A and arm B (one-sided stratified log rank p = 0.876, HR = 0.72, 95% CI: 0.41, 1.26). 5-yr OS was 88.3% (95% CI: 52.3, 92.4) in arm A and 83.7% (95% CI: 77.1, 88.6) in arm B. No significant difference was observed in DFS (two-sided stratified log-rank p = 0.299, HR = 1.25, 95% CI: 0.82,1.90). 5-yr DFS rate was 71.2% (95% CI: 63.4, 77.6) in arm A and 76.5% (95% CI: 69.2, 82.3) in arm B. The incidence rate of treatmentrelated grade $ 3 AEs was 68.8% (119/173, 95% CI: 61.3, 75.6) on arm A, and 70.7% (123/174, 95% CI: 63.3, 77.3) on arm B. Arm B pts had earlier discontinuation of therapy due to AEs and pt withdrawal than arm A (32.4% vs 21.5%, p = 0.029).The most common treatment-related AEs were neutropenia, leukopenia, neuropathy, diarrhea w/o prior colostomy and fatigue. Conclusions: At 17% of planned accrual, the trial did not meet its primary end point. The addition of Bev to postoperative mFOLFOX6 did not significantly improve OS in pts with stage II/ III rectal cancer after preoperative chemoradiation. These results are consistent with adjuvant colon cancer trials with Bev. About two-thirds of pts developed grade 3 or higher toxicity, and in a subset, pt-reported outcome endpoints were not different between the 2 arms. Clinical trial information: E5204.

3617

3618

Poster Session (Board #314), Sat, 8:00 AM-11:30 AM

A randomized phase III trial of 1-year adjuvant chemotherapy with oral tegafur-uracil (UFT) vs. surgery alone in stage II colon cancer: SACURA trial. First Author: Yoshiki Kajiwara, Department of Surgery, National Defense Medical College, Saitama, Japan Background: Efficacy of adjuvant chemotherapy in stage II colon cancer patients is still controversial. The SACURA trial is a phase III study to evaluate the superiority of 1-year adjuvant treatment with oral tegafur-uracil (UFT) to surgery alone for stage II colon cancer in a large population. Methods: 20-80 aged patients with curatively resected stage II colon cancer were randomly assigned to the surgery alone group or UFT group (UFT at 500-600 mg/day as tegafur in 2 divided doses for 5 days, followed by 2-day rest, then repeated for 1 year). Primary endpoint was disease-free survival (DFS), and the secondary endpoints were overall survival (OS), recurrencefree survival (RFS), and safety. Sample size was 2000 determined with an assumed hazard ratio (HR) 0.729, respectively, a two-sided significance level of 5%, and a power of 90%. Results: A total of 1982 patients (997 in the surgery alone group and 985 in the UFT group) were included in the efficacy analysis. Median follow-up was 69.5 months, the median age at enrollment was 66 years, and stage IIA/IIB/IIC were 84%/13%/3%, respectively. The 5-year DFS rate was 78.4% in the surgery alone group and 80.2% in the UFT group. The HR for DFS was 0.91 (95%CI, 0.75-1.10; p = 0.307), and superiority of the UFT group was not demonstrated. Approximately 9% of patients in the both groups experienced second cancers, which consist 41.7% of the DFS events. The 5-year RFS rate was 84.6% and 87.2% (HR, 0.82; 95%CI, 0.65-1.04), and the 5-year OS rate was 94.3% and 94.5% (HR, 0.93; 95%CI, 0.66-1.31) in the surgery alone group and the UFT group, respectively. The completion rate of the 1-year UFT was 60.8%. The Incidence of grade 3/4 adverse events in the UFT group was 3.7% for ALT elevation, 3.3% for AST elevation, 1.4% for hyperbilirubinemia, 2.6% for anorexia, 1.9% for nausea, 1.4% for diarrhea, and , 1% for hematologic toxicities. Conclusions: Superiority of 1-year adjuvant treatment with UFT to surgery alone was not demonstrated in patients with stage II colon cancer. Clinical trial information: NCT00392899.

Poster Session (Board #315), Sat, 8:00 AM-11:30 AM

Preoperative staging with magnetic resonance imaging (MRI) and endorectal ultra-sonography (ERUS) for locally advanced rectal cancer (LARC) after chemoradiotherapy (CRT): Accuracy with histopathologic findings. First Author: Marc Van Den Eynde, Institut Roi Albert II, Cliniques universitaires St-Luc, UCL, Brussels, Belgium Background: Pre-operative CRT followed by total mesorectal excision (TME) is nowadays the standard for LARC (cT3-T4N0 or cTxN+). TME could be avoided for patients with pathologic complete response. Our aim was to determine whether preoperative staging (MRI/ERUS) after CRT could accurately predict a pathological response. Methods: Between 1998 and 2014, 331 patients with LARC who underwent TME after CRT were identified. Only patients with available and complete results for both preoperative MRI and ERUS (T and N staging) and pathological staging were analyzed. Relevant categories of tumor downstaging (yT0N0: complete; yT1-2N0: present; yT3-T4N0 or yTxN+: absence) were compared. Accuracy of MRI and EUS was assessed using the ROC method and agreement between them by the kappa test. Mc-Nemar’s test evaluated accuracy difference. Results: 154 patients (sex ratio M/F 1.96; mean age 62.0 years-old; upper (15.5%), mid (26.6%) and low rectum (57.7%)) were analyzed. T-stage was accurately estimated by MRI in 53.2% (overstating (O): 35.1%, understaging (U): 11.7%) and by ERUS in 55.8% (O: 33.8%, U: 10.4%); Kappa coefficient: 0.489 (moderate agreement). N-stage was accurately estimated by MRI in 64.3% (O: 23.4%, U: 12.3%) and by ERUS in 63.6% (O: 14.9%, U: 21.4%); Kappa coefficient: 0.326 (fair agreement). Accuracies in the prediction of absence of tumor downstaging were similar with MRI (66.9%) and ERUS (74.0%; p = 0.25) both with an ability to predict non-responders (sensitivity) of respectively 85.6% and 87.6% (specificity of 35.1% and 50.9% respectively). Accuracies in the prediction of complete tumor downstaging were similar for MRI (85.1%) and ERUS (84.4%; p = 0.491) with a low ability to predict complete responders (sensitivity MRI: 25.0%, ERUS: 30.0%; specificity: MRI: 94.0%, ERUS: 92.5%). Conclusions: After CRT, preoperative MRI and ERUS accurately predict an absence of tumor downstaging but seem unhelpful for adequately estimate a complete pathological response and a possible non operative patient management.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

200s 3619

Gastrointestinal (Colorectal) Cancer Poster Session (Board #316), Sat, 8:00 AM-11:30 AM

3620

Poster Session (Board #317), Sat, 8:00 AM-11:30 AM

PRODIGE 2 phase III trial neoadjuvant in rectal cancer: Quality of life and results at 5 years. First Author: Pierre-Luc Etienne, Clinique de l’Armoricaine, Saint-Brieuc, France

Validity of Adjuvant! Online in elderly patients with stage III colon cancer based on 2,794 patients from the ACCENT database. First Author: Christiana Matthaiou, B.O. Cyprus Oncology Centre, Nicosia, Cyprus

Background: The ACCORD 12 -PRODIGE 2 trial randomly assigned patients to receive radiotherapy (RTH) 45 Gy (CAP-45) with capecitabine or RTH 50 Gy with capecitabine and weekly oxaliplatin(CAPOX-50). The primary objective (complete sterilization of the operative specimen) was not met. This is an update of results at 5 years. Methods: Between November 2005 and July 2008, 598 patients were randomized. Inclusion criteria were: adenocarcinoma of the rectum accessible to digital examination, stage T3 or resectable T4 (or T2 distal anterior rectum) NX M0, , 80 years, PS 0-1. Updated carcinologic results were analyzed using the Kaplan-Meier method and Cox proportional-hazards models. Quality of life was assessed with the anal sphincter conservative treatment questionnaire (ASCT) and by reporting of erectile dysfunction (ED) [satisfaction degree varies from 1(ED) to 7 (normal)]. The evolution of the ED score and selected items from ASCT questionnaire was analyzed with a linear mixed model (LMM). Results: In the ITT population (median follow-up of 5 years), results were in CAP-45 arm and in CAPOX-50 arm, respectively : local recurrence rate 8.8% vs 7.8% (HR for 5y-local RFS 0.92[0.51 – 1.66], p = 0.78), disease free survival (DFS) 60.4% vs 64.7% (p = 0.25, HR = 0.86 [0.66-1,11]), overall survival (OS) 76.4% vs 81.9% (p = 0.056, HR = 0.71[0.50-1 .01]) and specific OS (p = 0.19, HR = 0.75[0.50 – 1.13]). No significant difference was observed in the evolution of the ED score between the two arms (p = 0.29). From ASCT questionnaire, social life disturbance rated from 1 to 4 (greater to no disturbance), with a 5y-median score of 3 (1-4) was not different between the two arms during follow-up (p = 0.15). Moreover, for the global satisfaction score rated from 1 to 7 (unsatisfied to very satisfied), and for the total score, no statistically significant difference was noted (p = 0.43 and p = 0.38 respectively). Conclusions: The quality of life and ED were not different between the two arms after five years of follow-up like the oncologic results as specific OS, DFS and local recurrence. The addition of oxaliplatin to chemoradiation with capecitabin has a minor impact on natural history of the locally advanced rectal cancer. Clinical trial information: NCT00227747.

Background: Adjuvant! Online is a tool used for clinical decision making in patients (pts) with early stage colon cancer. As the tool’s construction and validation has never been published, it remains whether Adjuvant! Online can accurately predict outcomes for elderly pts (age 70+) with node positive colon cancer who received adjuvant chemotherapy. Methods: Individual data from elderly pts with stage III colon cancer who enrolled into ACCENT trials were entered into the Adjuvant! Online program, and predicted probabilities of 5-year overall survival (OS) and recurrence-free survival (RFS) were obtained. Median predictions were then compared with known rates and their 95% CIs, overall and within pt subgroups. As co-morbidities were not known for ACCENT pts but required for calculator entry, pts were assumed to have either “minor” or “average for age” co-morbidities in two separate analyses. Results: A total of 2,794 elderly pts from 14 randomized studies were included. When “minor” comorbidities were assumed, the median predicted 5-year OS rate of 63% matched the actual rate of 63% (95% CI: 62% to 65%); when “average for age” comorbidities were assumed, the median prediction dropped to 58%, outside the CI for the actual rate. On the other hand, assuming “minor” comorbidities gave a median 5year RFS prediction of 61%, outside the 95% CI for the actual rate of 56% (95% CI: 55% to 58%); assuming “average for age” comorbidities yielded an improved median prediction of 57%. Conclusions: Adjuvant! Online is reasonably accurate overall for predicting outcomes in elderly patients with stage III colon cancer who participated in clinical trials, though accuracy may differ between 5-year RFS and 5-year OS predictions when a fixed degree of comorbidities is assumed.

3621

3622

Poster Session (Board #318), Sat, 8:00 AM-11:30 AM

Effects of a structured exercise program on physical activity and healthrelated fitness in colon cancer survivors: One year feasibility results from the NCIC CTG CO.21 (CHALLENGE) trial. First Author: Kerry S. Courneya, University of Alberta, Edmonton, AB, Canada Background: There is strong interest in testing lifestyle interventions to improve cancer outcomes, however, the optimal methods for achieving behaviour change in large scale pragmatic trials is unknown. In this context, we report the 1 year feasibility results for behaviour change in NCIC CTG CO.21 Colon Health and Life-Long Exercise Change (CHALLENGE) Trial. Methods: Between 2009 and 2014, the CHALLENGE Trial randomized 273 high risk stage II and III colon cancer survivors who had recently completed chemotherapy at 50 centers in Canada and Australia to a structured exercise program (SEP; n = 136) or health education materials (HEM; n = 137). The primary feasibility outcome in an a prior defined interim analysis was a difference between randomized groups of $ 5 metabolic equivalent task (MET) hours/week in self-reported recreational physical activity (PA) once a minimum of 250 participants had reached 1 year. Secondary feasibility outcomes included cardiorespiratory fitness, body weight and circumferences, and objective physical functioning. Results: The SEP group reported an increase from baseline to 1 year in self-reported recreational PA of 15.6 MET hours/week compared to an increase of 5.1 MET hours/week in the HEM group (between group difference = +10.5 [95% CI = +3.1 to +17.9]; p = 0.002). The SEP group also improved relative to the HEM group in predicted VO2max (p = 0.068), 6 minute walk (p , 0.001), 30 second chair stand (p , 0.001), 8 foot up and go (p = 0.004), and sit and reach (p = 0.08). Conclusions: The behaviour change intervention in the CHALLENGE Trial produced a substantial increase in self-reported recreational PA that met the feasibility criterion for trial continuation, resulted in objective fitness improvements, and is consistent with the amount of PA associated with improved colorectal cancer outcomes in observational studies. Clinical trial information: NCT00819208.

Poster Session (Board #319), Sat, 8:00 AM-11:30 AM

Planned safety analysis of the ACTS-CC 02 trial: A randomized phase III trial of S-1/oxaliplatin (SOX) versus UFT/LV as adjuvant chemotherapy for highrisk stage III colon cancer. First Author: Takuya Miura, Department of Gastroenterological Surgery, Hirosaki University Graduate School of Medicine, Hirosaki, Japan Background: The ACTS-CC 02 trial was designed to verify the superiority of postoperative adjuvant chemotherapy with S-1/oxaliplatin (SOX) over Japanese standard treatment with UFT/LV in terms of disease-free survival in patients (pts) with high-risk stage III colon cancer. To date, no phase III trials evaluating SOX as postoperative adjuvant chemotherapy have been reported. We report the results of a planned safety analysis. Methods: Pts who underwent curative resection for high-risk stage III colon cancer were randomly assigned to either UFT/LV (300–600 mg/day of UFT according to body surface area [BSA] and 75 mg/day of LV on days 1-28, every 35 days, 5 courses) or SOX (100 mg/m2 of oxaliplatin on day 1 and 80–120 mg/day of S1 according to BSA on days 1-14, every 21 days, 8 courses). The dose and adverse events (AEs) in each course were recorded. Results: From April 2010 through October 2014, a total of 966 pts were enrolled at 260 institutions. The safety analysis set (derived by excluding pts who received no protocol treatment from the full analysis set which excluded pts who withdrew informed consent) comprised 472 pts in UFT/LV and 460 in SOX. The median age was 65.0 years, with men accounting for 54.2%, an ECOG PS of 0 in 94.2%, and colon + rectosigmoid cancer in 86.4%.Demographic factors were well balanced. The population of pts who received the planned 6-month protocol treatment was 76.9% in UFT/LV and 65.8% in SOX. The overall incidences of any grade AEs in UFT/LV and SOX were 91.3% and 98.7%, respectively. The incidences of grade $ 3 AEs were 15.9% and 35.7% respectively, including neutropenia (1.5% v 17.2%), thrombocytopenia (0.6% v 2.8%), increased AST (2.1% v 0.7%), increased ALT (3.0% v 0.9%), anorexia (2.3% v 3.5%), nausea (0.8% v 2.0%), diarrhea (8.1% v 5.4%), fatigue (1.1% v 1.5%) and sensory neuropathy (0.2% v 4.6%). There was 1 treatment-related death (fulminant hepatitis) in UFT/LV. Conclusions: The incidence of AEs associated with SOX as postoperative adjuvant chemotherapy for colon cancer was considered acceptable, although slightly higher than that of UFT/LV. It was similar to that reported for adjuvant chemotherapy with FOLFOX or CapeOX. Clinical trial information: JapicCTI-101073.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Colorectal) Cancer 3623

Poster Session (Board #320), Sat, 8:00 AM-11:30 AM

Predicted benefit of alternative post-treatment surveillance strategies in stage II and III colon cancer survivors. First Author: Johnie Rose, Case Comprehensive Cancer Center, Case Western Reserve University, Cleveland, OH Background: Early diagnosis of recurrence may provide the opportunity for cure among colon cancer survivors. Randomized studies comparing posttreatment surveillance strategies suggest that patients receiving more frequent testing are more likely to undergo curative treatment upon recurrence. However, uncertainty about the cancer-specific mortality benefit of intensive surveillance has led to variation in surveillance recommendations across professional societies. We developed a simulation model to predict outcomes for different surveillance strategies within a hypothetical population of survivors. Methods: The Colorectal Cancer Surveillance and Recurrence (CRCSuRe) model generates natural history events in a simulated population of colon cancer survivors based on patient-level data from the Clinical Outcomes of Surgical Therapy (COST) Trial [NEJM.2004;350(20):205059]. Individuals are at time-varying risks for progression to recurrence, recurrence symptoms, metastatic recurrence, and death. Any schedule of followup testing using carcinoembryonic antigen (CEA), computed tomography (CT), and colonoscopy can be simulated in a population with defined age and stage distribution. We used CRCSuRe to compare three distinct surveillance regimens—all of which complied with one or more published guidelines—and no surveillance in a population with mean age of 65 years and equal proportions of stage II and III disease. Results: Predicted clinical outcomes are shown in the table. Conclusions: Increased intensity of surveillance is associated with more frequent detection of resectable recurrences and improved survival. However, modest predicted survival differences suggest that riskadjusted surveillance strategies are needed to improve patient outcomes.

TPS3624

201s Poster Session (Board #321a), Sat, 8:00 AM-11:30 AM

An open-label, multi-center, phase 2 study of switch maintenance with TAS102 plus bevacizumab following oxaliplatin or irinotecan-based fluoropyrimidine-containing induction chemotherapy in patients with metastatic colorectal cancer: ALEXANDRIA study. First Author: Mohamed E. Salem, Lombardi Comprehensive Cancer Center, Georgetown University, Washington, DC

* In months if not stated; yrs = years.

Background: Maintenance therapy with a fluoropyrimidine plus bevacizumab is a widely accepted strategy in the treatment of mCRC, having been shown significant improvement in progression-free survival (PFS), predominantly in the CAIRO 3 study (Simkins LH, Lancet 2015). TAS-102 is a novel nucleoside antitumor agent consisting of trifluridine and tipiracil hydrochloride. In the randomized phase 3 RECOURSE trial, TAS-102 demonstrated a significant improvement in overall survival (OS) and PFS, and resulted in a 44% disease control rate, compared with placebo in patients with refractory CRC. (Mayer RJ, NEJM 2015). Methods: ALEXANDRIA (NCT02654639) is an open-label, multicenter, single arm phase 2 study designed to evaluate efficacy and safety of switch maintenance with TAS-102 plus bevacizumab following induction chemotherapy in mCRC patients. Eligibility criteria include histologically confirmed mCRC, ECOG # 1, acceptable organ and bone marrow function, and patients must have had 16 to 20 weeks of standard first line therapy with bevacizumab plus oxaliplatin- or irinotecan-containing regimens and have stable disease or better. Forty-five patients will be enrolled on the study to receive 35 mg/m2 TAS-102 po bid on Days 1-5 and Days 8-12 of each 28day cycle. Bevacizumab, 5 mg/kg IV, will be administered every 14 days. This treatment cycle will be repeated every 28 days until patients demonstrate disease progression, experience unacceptable toxicity or withdraw consent. CT scans will be performed every 8 weeks. The primary endpoint is PFS, and secondary endpoints include OS, safety, ORR, DCR, TTP, and DDC. This study is designed to have 80% power (2-sided alpha level of 0.05) to test the median PFS of 8 mo vs. 5 mo (historical control) using a one-sample log-rank test. The time-to-event endpoints (PFS, OS, TTP, and DDC) will be estimated using the Kaplan-Meier method. Estimates of the rates (ORR and DCR) will be presented with their 95% confidence intervals. Clinical trial information: NCT02654639.

TPS3625

TPS3626

High Intensity Surveillance* Outcomes

CEA q3 x 2 yrs; then q6 x 3 yrs CT chest/abdomen/pelvis q12 Colonoscopy At 12 and 48 Resectable recurrences 42.1% Detected by symptoms 13.1% 5-year relative survival of patients who recur 19.1%

Medium Intensity

Low Intensity

None

q6 x 5 yrs q12 At 12 38.0% 20.6% 18.0%

– At 24 At 60 25.8% 65.2% 14.7 %

– – – 23.7% 100.0% 13.8%

Poster Session (Board #321b), Sat, 8:00 AM-11:30 AM

PARADIGM study: A multicenter, randomized, phase III study of mFOLFOX6 plus panitumumab or bevacizumab as first-line treatment in patients with RAS (KRAS/NRAS) wild-type metastatic colorectal cancer. First Author: Kei Muro, Aichi Cancer Center Hospital, Nagoya, Japan Background: Optimal combination of monoclonal antibody (anti-VEGF vs. anti-EGFR antibody) with standard chemotherapy as first-line treatment in patients (pts) with RAS (KRAS/NRAS) wild-type metastatic colorectal cancer (mCRC) remains controversial. The FIRE-3 study demonstrated a significant improvement in overall survival (OS) with anti-EGFR over bevacizumab in pts with KRAS exon 2 wild-type mCRC, while the CALGB 80405 study did not. This study is designed to compare panitumumab with bevacizumab plus mFOLFOX6 in pts with RAS wild-type chemotherapy-naive mCRC. Methods: Eligible pts are aged 20-79 years and ECOG performance status 01 with histologically/cytologically confirmed RAS wild-type mCRC. A total of 800 pts will be randomly assigned in a 1:1 ratio to mFOLFOX6 plus panitumumab or bevacizumab and stratified by institution, age (20-64 vs. 6579 years), and liver metastases (present vs. absent). Treatment regimens include oxaliplatin 85 mg/m2, concurrent with l-leucovorin 200 mg/m2, followed by 5-fluorouracil (5-FU) iv bolus 400 mg/m2 at day 1, followed by 46-hour 5-FU infusion 2400 mg/m2 and either panitumumab 6 mg/kg or bevacizumab 5 mg/kg at day 1 every two weeks. The primary endpoint is OS; the study was designed to detect the OS hazard ratio of 0.76, with a onesided type I error of 0.025 and 80% power. An interim analysis is planned for the OS when approximately 70% of the targeted 570 events have occurred. A large-scale exploratory biomarker substudy to identify potential biomarker candidates using tumor tissue and circulating tumor DNA (ctDNA) is also underway (Study ID: NCT02394834). Tumor tissue samples will be collected before chemotherapy. Plasma samples will be obtained before and after chemotherapy. Paired pretreatment DNA from primary tissue and ctDNA will be analyzed using an NGS-based platform to evaluate potential biomarkers for each treatment arm. Pretreatment and post-treatment ctDNA will also be analyzed to investigate resistance mechanisms for each treatment. As of January 2016, 98 pts have been randomized. Results are expected in 2020. Clinical trial information: NCT02394795.

Poster Session (Board #322a), Sat, 8:00 AM-11:30 AM

Epigenetic priming prior to pembrolizumab in microsatellite-stable (MSS) advanced colorectal cancer. First Author: Adrian Gerard Murphy, Johns Hopkins Sidney Kimmel Comprehensive Cancer Center, Baltimore, MD Background: Pembrolizumab has been shown to result in compelling responses and progression free survival in advanced MMR-d CRC. However, the vast majority (~95%) of patients with advanced CRC have MMR proficient (MMR-p) tumors which do not respond to PD1 blockade. Hypermethylation of DNA promoter regions by DNA methyltransferases (DNMT) and histone deacetylation by histone deacetylases (HDAC) result in the silencing of tumor suppressor genes. Preclinical studies have shown that combined HDAC/DNMT inhibition can induce susceptibility to immune checkpoint therapy and inhibit tumor growth by inducing re-expression of tumor suppressors silence by DNA methylation. Methods: This is a single institution, open-label, randomized pilot study (NCT02512172) enrolling patients with advanced MMR-p CRC who have progressed on $ 1 line of chemotherapy (n = 30). Patients are randomized to one of 3 arms: A: oral 5azacitidine (5AZA, CC486), B: romidepsin or C: combined 5AZA and romidepsin. Patients will have a pre-treatment biopsy followed by 14 days of epigenetic therapy followed by a second biopsy. All groups then begin pembrolizumab 200 mg combined with epigenetic therapy until disease progression. Eligibility: Histologically confirmed MMR-p CRC and have received $ 1 line of chemotherapy but # 3 lines, ECOG 0-1, measurable/ biopsiable disease. Those with primary refractory disease are ineligible for this study There will be a safety run-in for all arms (n = 3 patients), if 0-1 patients have a DLT (dose limiting toxicity), accrual will continue, if $ 2 experience DLT, we will halt accrual to that arm. Toxicity is assessed using standard NCI CTCAE v4.0. Statistics: Primary endpoints include evaluation of changes in CD45R0+ve tumor infiltrating lymphocytes (memory T cells) in tumor biopsies and to determine safety/feasibility of these combined therapies. Secondary endpoints include estimating progression-free and overall survival. This study was activated in January 2016 and is expected to complete accrual within 12 months. Clinical trial information: NCT02512172.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

202s TPS3627

Gastrointestinal (Colorectal) Cancer Poster Session (Board #322b), Sat, 8:00 AM-11:30 AM

Randomized phase II study of first-line FOLFOX plus panitumumab (pan) versus 5FU plus pan in elderly RAS and BRAF wild-type (wt) metastatic colorectal cancer (mCRC) patients (pts): The PANDA study. First Author: Roberto Moretto, Azienda Ospedaliero-Universitaria Pisana, Istituto Toscano Tumori, Pisa, Italy Background: Based on available data, a reasonable upfront treatment for elderly mCRC pts is a fluoropyrimidine-based monotherapy plus bevacizumab, irrespectively of RAS status. Up today, data about the treatment of elderly mCRC pts with chemotherapy plus anti-EGFRs are scarce. FOLFOX plus pan is a standard first-line option for RAS wt mCRC. Adjustments of chemo-dosage are commonly applied in routinary practice to elderly pts, while evidences about the combination of an anti-EGFR with a fluoropyrimidine-based monotherapy are lacking. On the basis of these considerations, we designed the present randomized phase II trial of upfront pan in combination with FOLFOX or 5FU (both at adjusted dosage) in elderly RAS and BRAFwt unresectable mCRC pts. Methods: This is a prospective, open-label, multicenter phase II trial in which initially unresectable and previously untreated elderly ($ 70 years) RAS and BRAF wt mCRC pts are randomized to pan (6 mg/kg) plus simplified FOLFOX (oxaliplatin 85 mg/ sqm, L-leucovorin 200 mg/sqm, 5-fluoruracil 2400 mg/sqm 48 hcontinuous infusion) or 5FU (L-leucovorin 200 mg/sqm, 5-fluoruracil 2400 mg/sqm 48 h-continuous infusion) every 14 days up to 12 cycles, followed by pan alone until disease progression. RAS and BRAF analyses are centralized. Geriatric assessment by means of G8 screening tool and CRASH score is scheduled at baseline. Primary endpoint is progression-free survival (PFS). Secondary endpoints are response rate, overall survival, safety profile, early response, association of G8 score with clinical outcome and treatmentrelated toxicity, association of CRASH score risk categories with treatmentrelated toxicity. Assuming an expected median PFS time $ 9.65 months with one or both experimental regimens, a sample size of 90 patients in each arm will guarantee a power of 90% for a one-sided Brookmeyer-Crowley test, with a type I error rate of 5%, against the null hypothesis of a median PFS time # 6 months. A pick-the-winner strategy will drive further experimental developments for one or both the two combinations. The study is sponsored by GONO Cooperative Group. EUDRACT 2015-003888-10. Clinical trial information: 2015-003888-10.

TPS3629

Poster Session (Board #323b), Sat, 8:00 AM-11:30 AM

TRIBE-2 by GONO group: A phase III strategy study in the first- and secondline treatment of unresectable metastatic colorectal cancer (mCRC) patients. First Author: Chiara Cremolini, Azienda OspedalieroUniversitaria Pisana, Istituto Toscano Tumori, Pisa, Italy Background: Chemotherapy plus bevacizumab (bev) is a standard first-line treatment for unresectable mCRC patients. The intensity of the chemotherapy backbone may be modulated from one to three drugs regimens according to patients’ general conditions and comorbidities, treatments’ objectives, and disease characteristics. TRIBE study demonstrated a significant advantage in progression-free (PFS) and overall survival (OS) for FOLFOXIRI plus bev as compared with FOLFIRI plus bev. Based on recent evidences supporting the de-intensification of the upfront regimen after 46 months, alternating induction and maintenance phases is a valuable option. Moreover, the prolonged inhibition of angiogenesis, and in particular the continuation of bev beyond disease progression, is an efficacious strategy in the treatment of mCRC patients. Based on these considerations, we designed the present phase III study aiming at comparing two different strategies in the first- and second-line treatment of mCRC. Methods: TRIBE2 is a prospective, open-label, multicentric phase III randomized trial in which unresectable mCRC patients, previously untreated for the metastatic disease, are randomized to receive first-line mFOLFOX-6 plus bevacizumab followed by FOLFIRI plus bevacizumab after progression (arm A) or FOLFOXIRI plus bevacizumab followed by re-introduction of the same regimen after progression (arm B). All combination treatments are repeated up to 8 cycles and followed by maintenance with a fluoropyrimidine plus bev, in order to shorten the duration of more intensive treatments while prolonging as more as possible oxaliplatin- and irinotecan-free intervals. The primary endpoint is Progression Free Survival 2 (PFS2), defined as the time from randomization to disease progression on any treatment given after 1st progression or death. Assuming a median PFS2 of 15 months in arm A, 654 subjects will be needed to detect a HR for PFS2 of 0.77 in favour of arm B with two-sided alpha and beta errors of 0.05 and 0.20, respectively. The trial is now ongoing at 58 Italian centers and from Feb 2015 to Jan 2016 212 patients have been randomized. Clinical trial information: NCT02339116.

TPS3628

Poster Session (Board #323a), Sat, 8:00 AM-11:30 AM

A phase 1, first-in-human, open label, dose escalation study of MGD007, a humanized gpA33 x CD3 DART molecule, in patients with relapsed/ refractory metastatic colorectal carcinoma. First Author: John D. Powderly, Carolina BioOncology Institute, Huntersville, NC Background: Colorectal (CRC) cancer is the fourth most commonly diagnosed cancer in the U.S. and the second most common cause of cancer related deaths. Recent advances in immunotherapy for multiple malignancies benefit only a specific minority of patients with colorectal cancer but serve as proof of principle. The Dual-Affinity Re-Targeting, or DART, technology, is a proprietary antibody-based platform that has been designed to engage multiple targets with a single molecule. MGD007 is a novel gpA33 x CD3 DART protein that targets gpA33-positive cells for recognition and elimination by co-engagement of CD3-expressing T lymphocytes. Glycoprotein A33 (gpA33) is a 43 kDa membrane-bound glycoprotein normally restricted to the surface of normal human colon and small bowel epithelial cells. In addition, the gpA33 antigen is homogeneously expressed at high levels in . 95% of primary and metastatic human CRC. MGD007 mediates redirected T cell killing of gpA33-expressing CRC cell lines in vitro and blocks transplantable CRC cell line xenografts in preclinical tumor models. These observations suggest that gpA33 that could be an attractive target for the immunotherapy of CRC. This study represents the first clinical evaluation of DARTmolecules in patients with solid tumors. Methods: The Phase I doseescalation two-arm study of MGD007 will enroll patients with metastatic relapsed/refractory CRC. The study will employ a 3+3+3 design to explore the safety, PK, immuno-regulatory activity and preliminary anti-tumor activity of MGD007. Patients will receive either weekly infusions of MGD007 (Arm A) or every three-week infusions (Arm B) in 6-week cycling until confirmed disease progression or intolerable toxicity. Two dose expansion cohorts will be initiated once dose and schedule are established from dose escalation. Response will be assessed every 6 weeks per RECIST v1.1 and immune RECIST to account for response patterns observed with immunotherapies. The study is ongoing in the U.S. Clinical trial information: NCT02248805.

TPS3630

Poster Session (Board #324a), Sat, 8:00 AM-11:30 AM

Regorafenib dose optimization study (ReDOS): A phase II randomized study of lower starting dose regorafenib compared to standard dose regorafenib in patients with refractory metastatic colorectal cancer (mCRC). First Author: Tanios S. Bekaii-Saab, The Ohio State University Comprehensive Cancer Center, Columbus, OH Background: Regorafenib is an oral multikinase inhibitor that blocks several protein kinases involved in angiogenesis and oncogenesis. It was recently shown to provide a survival benefit in refractory metastatic colorectal cancer. Despite the observed benefits, toxicities such as palmar-plantar erythrodysesthesia syndrome (PPES- which occurs early in the first 1-2 weeks) and fatigue have limited its use. Multiple practices across the US adopted various dosing or interval scheduling despite the absence of supportive data. Methods: ReDOS (NCT02368886) is an Academic and Community Cancer Research United (ACCRU) network-led randomized phase II study of lower dose regorafenib compared to standard dose regorafenib in patients with refractory mCRC. Patients are randomized 1:1 to receive an escalating dose (Arm A: 80 mg daily Week 1, 120 mg daily Week 2,160 mg daily week 3 then 1 week off followed by Cycle 2) vs. standard dose (Arm B: 160 mg daily for 21 days/then 1 week off). Within each treatment arm, patients are further randomized to receive either pre-emptive or reactive Clobetasol for prevention/treatment of PPES (sub-arms A1, A2, B1 and B2). Eligibility criteria include histologically confirmed mCRC, ECOG # 1, acceptable organ function, and patients must have failed all standard regimens including appropriate biologics. The primary endpoint is the 8 week planned continuation rate defined as the proportion of patients in each arm who complete 2 cycles of treatment and who intend to initiate cycle 3 if no progression is noted. Assuming an 8-week planned continuation rate of 75% in the control group (B), and desiring an improvement to 90% in the experimental group (A), a one-sided test with alpha = 0.20 and power of 80% will require a sample size of 110 patients enrolled to both regorafenib arms (A and B) with 28 patients randomized to treatment sub-arm (A1, A2, B1, B2). An additional 10 patients will be enrolled to account for dropout, cancellations, ineligibles, etc, to give a total sample size of 120 patients. The total study duration is expected to be approximately 2 years. Clinical trial information: NCT02368886.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Colorectal) Cancer TPS3631

Poster Session (Board #324b), Sat, 8:00 AM-11:30 AM

KEYNOTE-164: Phase 2 study of pembrolizumab for patients with previously treated, microsatellite instability-high advanced colorectal carcinoma. First Author: Dung T. Le, Sidney Kimmel Cancer Center at Johns Hopkins University, Baltimore, MD Background: A subset of advanced colorectal cancers (CRC) is characterized by mismatch repair (MMR) deficiency, leading to a microsatellite instabilityhigh (MSI-H) phenotype, features of which are a high mutational burden and lymphocytic infiltrates. Pembrolizumab is a monoclonal anti–PD-1 antibody designed to block the interaction between PD-1 and its ligands PD-L1 and PD-L2. In the phase 2 KEYNOTE-016 study, pembrolizumab provided an ORR of 62% in patients (pts) with progressive MMR-deficient metastatic CRC vs 0% in pts with MMR-proficient CRC. The multicenter, phase 2 KEYNOTE-164 trial (NCT02460198) will evaluate the efficacy and safety of pembrolizumab in pts with previously treated, advanced MSI-H CRC. Methods: Key eligibility criteria include age $ 18 y; advanced CRC; MSI-H phenotype evidenced by $ 2 allelic shifts using a PCR-based assay, or lack of expression of $ 1 MMR protein (MLH1, MSH2, MSH6, PMS2) by IHC; prior treatment with fluoropyrimidine, oxaliplatin, and irinotecan; ECOG PS 0-1; no active autoimmune disease or brain metastases; and no prior anticancer therapy within 2 wk of study treatment. Pembrolizumab 200 mg Q3W is to be administered for 35 cycles or until disease progression, unacceptable toxicity, pt withdrawal, or investigator decision. Clinically stable pts with RECIST-defined progression may continue pembrolizumab until a scan performed $ 4 wk later confirms progression. Pts who complete all 35 cycles or discontinue pembrolizumab following a complete response and experience progression may be eligible for 1 y of pembrolizumab retreatment. Response is to be evaluated every 9 wk per RECIST v1.1 by central imaging vendor review and per RECIST adapted for immunotherapy response patterns. Adverse events (AEs) are to be assessed throughout treatment and for 30 d thereafter (90 d for serious AEs) and graded per NCI CTCAE v4.0. Pts are to be followed for survival every 9 wk. ORR per RECIST v1.1 by central imaging vendor review is the primary end point; secondary end points include PFS, OS, disease control rate, and DOR; ORR will also be assessed by the investigator. Enrollment will continue until ~60 pts have enrolled. Clinical trial information: NCT02460198.

TPS3633

Poster Session (Board #325b), Sat, 8:00 AM-11:30 AM

A phase 1b/2 study combining MM-151 + nal-IRI + 5-FU + leucovorin in RAS-wildtype metastatic colorectal cancer (mCRC). First Author: Emily Chan, Vanderbilt University Medical Center, Nashville, TN Background: Colorectal cancer is a leading cause of cancer death worldwide and the third most common cancer in men and women. Despite recent approvals of new agents in refractory disease, there has been little change in the composition of first-line therapies in recent years, which generally include chemotherapy in combination with a biologic, either an EGFR-targeted antibody or a VEGF inhibitor. Investigational agents currently in development represent attempts to improve anti-neoplastic efficacy within these established drug classes. MM-151 is a oligoclonal mixture of three IgG1, antiEGFR antibodies designed to bind distinct non-overlapping EGFR epitopes and inhibit ligand-mediated signal amplification. MM-151’s molecular composition enables antagonism of clinically relevant EGFR ligand mixtures, EGFR down-regulation and immune effector function (ADCC, CDC). Nal-IRI is a nanoliposomal formulation of irinotecan. In a randomized phase-III study (NAPOLI-1) of patients with metastatic pancreatic cancer previously treated with gemcitabine-based therapy, nal-IRI + 5-FU/LV demonstrated its safety and significant clinical activity, increasing overall survival (OS) and progression-free survival (PFS) relative to 5-FU/LV. The current study will characterize preliminary safety, initial efficacy and biomarker / pharmacodynamic profile of MM-151 + nal-IRI + 5-FU/LV in mCRC patients who are treatment na¨ıve to an EGFR inhibitor or irinotecan-based chemotherapy. Methods: In the dose finding cohorts of this Phase 1b/2 study, a modified toxicity probability interval approach (mTPI) will be utilized to determine the maximum tolerated dose (MTD) of MM-151 in combination with nal-IRI, 5FU and leucovorin. Enrollment will be determined by the safety profile observed, with approximately 8-12 pts anticipated. In the Phase 2 portion of the study, approximately 20-30 pts will receive the combination at the defined MTD. The primary objectives of this phase of the study is to characterize initial efficacy, obtain additional safety data, and complete pharmacodynamic and biomarker evaluation through the collection of pretreatment and on-study samples.

TPS3632

203s Poster Session (Board #325a), Sat, 8:00 AM-11:30 AM

A multicenter phase I/II study of TAS-102 with nintedanib in patients (pts) with metastatic colorectal cancer (mCRC) refractory to standard therapies (N-TASK FORCE): EPOC1410. First Author: Yasutoshi Kuboki, Department of Gastrointestinal Oncology, National Cancer Center Hospital East, Kashiwa, Japan Background: In a RECOURSE phase III study, TAS-102 significantly improved progression-free survival (PFS) and overall survival (OS) over placebo in patients (pts) with metastatic colorectal cancer (mCRC) refractory to standard therapies. We reported the combination of TAS-102 with bevacizumab (C-TASK FORCE trial) showed promising anti-tumor activities with mild toxicities (Kuboki Y, et al. ASCO 2015). Nintedanib is a triple angiokinase inhibitor of vascular endothelial growth factor receptors (1,2,3), platelet-derived growth factor receptors (a, b), and fibroblast growth factor receptors (1,2,3). A global phase III study, called as the LUME-Colon 1, comparing nintedanib monotherapy versus placebo in pts with mCRC refractory to standard therapies is ongoing, on the basis of promising antitumor activities with mild toxicities in a phase I clinical study of nintedanib monotherapy (Mross et al.). Therefore, we initiated phase I/II study to assess efficacy and safety for the combination of TAS-102 with nintedanib. Methods: The key eligibility criteria were pts with mCRC refractory or intolerant to fluoropyrimidine, irinotecan, oxaliplatin, anti-angiogenesis inhibitor and anti-EGFR antibody (if wild-type RAS) and without prior regorafenib and TAS-102; at least one measurable lesion; and ECOG performance status of 0 or 1. Phase I part was designed to determine the recommended phase II dose (RP2D) in a “3+3” cohort-based dose escalation design of nintedanib (150mg BID every day on level 1 and 200mg BID every day on level 2) with a fixed dose of TAS-102 (35 mg/m2 BID on days 1–5 and 8–12 q4w). Primary endpoint of the phase II part was an investigator-assessed PFS rate at 16 weeks in pts receiving the combination with RP2D. Using a single stage binomial design, this study required 52 pts, with an investigator-assessed PFS rate at 16 weeks of 40% deemed promising and 25% unacceptable (one-sided alpha = 0.1; beta = 0.2). Secondary endpoints included OS, PFS, objective response rate, disease control rate, safety, and pharmacokinetic parameters. The enrollment to phase II part began in January 2016. Clinical trial information: UMIN000017114. Clinical trial information: 000017114.

TPS3634

Poster Session (Board #326a), Sat, 8:00 AM-11:30 AM

First-line FOLFOX-4 plus panitumumab followed by 5-FU/LV plus panitumumab or single-agent panitumumab as maintenance therapy in patients (pts) with RAS wild-type, metastatic colorectal cancer (mCRC): The VALENTINO study. First Author: Filippo Pietrantonio, Fondazione IRCCS Istituto Nazionale Tumori, Milano, Italy Background: First-line FOLFOX-4 plus panitumumab is a standard treatment of RAS wild-type mCRC. Trials with first-line oxaliplatin-based chemotherapy showed that a deintensification-reintroduction strategy has no inferior outcomes when compared to treatment until progression. In the biologics era, the continuation of fluoropyrimidines and bevacizumab in the maintenance setting showed an increase of progression-free survival (PFS) as compared to treatment holiday. Few data are available at present for the optimal maintenance strategy with anti-EGFR monoclonal antibodies when used in association with first-line chemotherapy. This study aims to evaluate the efficacy of maintenance with panitumumab alone as compared to panitumumab with 5-FU/LV following FOLFOX-4 and panitumumab in pts with RAS wild-type mCRC. Methods: This is an open label, multicenter, randomized phase II study. Up to 224 pts will be enrolled and treated with induction FOLFOX-4 and panitumumab for 8 cycles. Then, in presence of disease control, they will receive at 1:1 ratio maintenance treatment panitumumab alone or 5-FU/LV plus panitumumab until progression or unacceptable toxicity. Tumor assessment is planned every 8 weeks. Primary objective is PFS: the sample size is calculated on the basis of a noninferiority hypothesis of median PFS with panitumumab alone as compared to 5-FU/LV and panitumumab, taking into account a median PFS of 10 months observed in the PRIME trial. An overall sample size of 224 pts achieves 90% power to detect a probability of 50% in the control group and a maximum difference of 15% in the study group, with a significant level of 0.1. The drop-out rate is 15%. Secondary objectives are safety, quality of life, response rate, duration of response, time to progression, time to treatment failure, overall survival. Exploratory objectives: identification of potential biomarkers of primary resistance by Next-generation sequencing (NGS) and in-situ hybridization techniques on archival tissues; evaluation of changes in plasma biomarkers by digital PCR or NGS; pharmacogenetic studies. Clinical trial information: NCT02476045.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

204s TPS3635

Gastrointestinal (Colorectal) Cancer Poster Session (Board #326b), Sat, 8:00 AM-11:30 AM

Open-label, randomized, multicenter, phase II trial designed to compare the efficacy of CAPTEM combination versus FOLFIRI as second line treatment in patients (pts) who have progressed on or after first-line oxaliplatin-containing chemotherapy for advanced, MGMT methylated, RAS mutated colorectal cancer (CRC). First Author: Rosa Berenato, Fondazione IRCCS Istituto Nazionale Tumori, Milan, Italy Background: Epigenetic silencing of O6-methylguanine-DNA methyltransferase (MGMT) during colorectal tumorigenesis is responsible for diminished DNA-repair of O6-alkylguanine adducts, leading to an enhancing chemosensitivity to alkylating agents such as temozolomide (TMZ). In two phase II studies, we showed that TMZ is effective in heavily pre-treated pts with advanced CRC and MGMT promoter methylation. Other studies in carcinoids cell lines demonstrated synergistic cell kill if 5-fluoracile and TMZ were delivered in a schedule-dependent manner. The CAPTEM regimen consists of TMZ for 5 days at 75 mg/m2/day bid on days 10–14 and capecitabine 750 mg/m2/day bid on days 1–14 of a 28-day cycle. Given the potential synergy of capecitabine and TMZ in MGMT methylated CRC, we planned a randomized study to evaluate the efficacy of CAPTEM versus FOLFIRI after failure of prior first-line oxaliplatin-based treatment in patients with advanced, MGMT methylated, RAS mutated CRC. Methods: This is an open-label, multicenter, randomized phase II trial. Up to 82 pts will be enrolled and randomized in a 1:1 ratio to receive FOLFIRI or CAPTEM as second-line treatment for up to 6 months or up to disease progression or unacceptable toxicity. Efficacy assessments will be performed every 8 weeks. Primary objective is progression-free survival (PFS). According to the results of the GERCOR study, the median PFS during second-line treatment with crossover from FOLFOX to FOLFIRI was 2.3 months. A one-sided log rank test with an overall sample size of 82 subjects achieves 90% power at a 5% significance level to detect an increase in median PFS from 2 months of the control group to 4 months. Secondary objectives: safety, quality of life, response rate and overall survival. Exploratory objectives: identification of potential biomarkers associated with activity of TMZ in tumor tissue (MGMT immunohistochemistry, methylBEAMing and next-generation sequencing) and in plasma (methylBEAMing and miRNA). Clinical trial information: NCT02414009.

TPS3637

Poster Session (Board #327b), Sat, 8:00 AM-11:30 AM

FOLFOX or CAPOX perioperative chemotherapy versus postoperative chemotherapy for locally advanced colon cancer: OPTICAL study. First Author: Yanhong Deng, Sixth Affiliated Hospital of Sun Yat-sen University, Guangzhou, China Background: In patients with high risk stage II and stage III colon cancer (CC), curative surgery followed by adjuvant chemotherapy with FOLFOX or CAPOX regimens has become a standard treatment. However, 20 to 30 % of these patients will develop distant metastasis, which ultimately result in death. Perioperative chemotherapy is a promising strategy with potential benefits that could be more effective at eradicating micrometastases. Moreover, shrinking tumor before surgery not only facilitate removal of all the tumor by the surgeon but also reduce tumor cell spreading during the procedure. With recent advances in radiology, preoperative computed tomography is a robust method for measuring the depth of tumor invasion and identifying the CC patients with poor prognosis, who may benefit from perioperative chemotherapy. We conducted the present randomized study to explore whether perioperative chemotherapy with FOLFOX or CAPOX regimens compared with postoperative chemotherapy could improve diseasefree survival in patients with radiologically staged, locally advanced, but resectable colon cancer. Methods: This trial is a two-arm, multicenter, open labelled, prospective, randomized phase III studies. 738 eligible patients with locally advanced (T4 or T3 with extramural depthS5 mm) CC patients will be randomly assigned, in a 1:1 ratio, to receive either perioperative (6 cycles of mFOLFOX6 or 4 cycles of CAPOX followed by surgery and a further 6 cycles of mFOLFOX6 or 4 cycles of CAPOX) or postoperative chemotherapy (surgery followed by 12 cycles of mFOLFOX6 or 8 cycles of CAPOX). The primary objective of this study is 3-year disease-free survival. Secondary objectives are efficacy in terms of R0 resection rate, overall survival (OS), relapse-free survival (RFS), down-staging of primary tumors, and tolerability of perioperative therapy and postoperative complications. Clinical trial information: NCT02572141.

TPS3636

Poster Session (Board #327a), Sat, 8:00 AM-11:30 AM

The FUNNEL: A molecular multiplex triage for precision medicine in metastatic colorectal cancer. First Author: Francesco Leone, Istituto di Candiolo, Fondazione del Piemonte per l’Oncologia, IRCCS, Candiolo, Italy Background: The goal of precision medicine is to match a selective drug with a genetic lesion that predicts for drug sensitivity. Over 50% of metastatic colorectal cancer (mCRC) patients (PTS) harbor KRAS, NRAS and BRAF mutations not responsive to anti-EGFR therapies. Recent data from our patient-derived mCRC xenografts platform support the presence of further negative predictors of response to anti-EGFRs in a further 10-15% of cases (Bertotti A. et al, Nature 2015). Therapeutic resistance to EGFR blockade was overcome in tumour graft models through therapies targeting actionable genes, suggesting the need for an expanded molecular classification of mCRC requiring clinical validation. Methods: We translated this list of potential mCRC drivers in the customized molecular diagnostic panel FUNNEL. FUNNEL primary goal is to identify PTS harboring low prevalence (~3%) markers potentially predictive for response to available targeted therapies other than anti-EGFR. To this end, the paraffin sample of 1000 newly diagnosed mCRC PTS from 8 Italian cancer centers are centrally genotyped with Sequenom and NanoString technologies. Centralized DNA extraction and analysis is performed after histology revision in the FUNNEL pathology hub. Whenever possible, patients with ensuing specific molecular features will be recruited, after standard primary treatment, in proof-ofconcept trials with a targeted therapy already proven positive in the matching preclinical model. Beyond ‘molecular triage’ purposes, FUNNEL will build an integrated repository of life-time clinical and molecular information, including proteomics from initial tissue micro arrays and dynamic data of centralized disease assessments imaging coupled with longitudinal liquid biopsies collected, until death, whenever a significant clinical event occurs (baseline, response to therapy and progression of sequential treatment’s lines). Altogether these data will give a comprehensive insight of mCRC clonal evolution laying the groundwork to design further innovative trials, and, eventually, inform patient care. EudraCT 2016-000448-33, Funded by AIRC Grant 9970; Italian Ministry of Health grant NET-2011-02352137. Clinical trial information: 2016-000448-33.

TPS3638

Poster Session (Board #328a), Sat, 8:00 AM-11:30 AM

A phase II clinical trial platform utilizing total neoadjuvant therapy (TNT) in rectal cancer: Nrg-GI002. First Author: Thomas J. George, NSABP/NRG Oncology, and The University of Florida, Pittsburgh, PA Background: Improvements in outcomes for locally advanced rectal cancer (LARC) have plateaued due to an inability to consistently deliver adjuvant therapy and thus far ineffective novel therapies. Systematic testing of new chemotherapy and radiation sensitizers are needed to advance treatment outcomes. This randomized phase II modular clinical trial platform utilizes Total Neoadjuvant Therapy (TNT) with parallel experimental arms in LARC. The experimental arms are not intended for direct comparison, but to test a variety of sensitizers or hypotheses in a consistent and relatively homogenous high-risk patient (pt) population with correlative biomarkers. Success of any given experimental arm will be determined by achievement of pathologic endpoints compared to a control arm. Methods: This NCTN multi-arm randomized phase II trial serves as a modular platform to assess novel sensitizers to neoadjuvant chemotherapy and/or chemoradiotherapy (chemoRT) in LARC. Eligibility includes LARC as defined by any ONE of the following criteria: distal location (cT3-4 # 5cm from the anal verge, any N); bulky (any cT4 or tumor within 3mm of the mesorectal fascia); high risk for metastatic disease (cN2); or not a candidate for sphincter-sparing surgical resection. After randomization, all pts receive neoadjuvant FOLFOX x 4mo → chemoRT (capecitabine with 50.4Gy) → surgical resection 8-12 wks later. Based on promising phase I results, the first experimental arm will assess the activity of veliparib added to standard chemoRT (capecitabine + RT). Primary endpoint is to demonstrate an improvement in Neoadjuvant Rectal Cancer (NAR) score for the experimental arm vs control representing a 20% relative risk reduction in DFS HR and 3-4% absolute OS improvement. Secondary endpoints include comparisons of OS, DFS, toxicity, pCR, cCR, therapy completion, negative surgical margins, sphincter preservation, sphincter function including quality of life, and exploratory assessments of molecular predictors of response and distant failure. Target accrual is 79 evaluable pts per arm with additional arms added through protocol amendments. NCT # pending. Support: U10 CA-180868, -180822, UG189867; AbbVie.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Colorectal) Cancer TPS3639

Poster Session (Board #328b), Sat, 8:00 AM-11:30 AM

3640

205s Poster Session (Board #269), Sat, 8:00 AM-11:30 AM

KEYNOTE-177: First-line, open-label, randomized, phase 3 study of pembrolizumab versus investigator-choice chemotherapy for mismatch repairdeficient or microsatellite instability-high metastatic colorectal carcinoma. First Author: Luis A. Diaz, Sidney Kimmel Cancer Center at Johns Hopkins University, Baltimore, MD

Randomized phase 3 study of S-1 versus capecitabine in the first-line treatment of metastatic colorectal cancer (mCRC): The SALTO study of the Dutch Colorectal Cancer Group. First Author: Cornelis J. A. Punt, Department of Medical Oncology, Academic Medical Center, University of Amsterdam, Amsterdam, Netherlands

Background: Mismatch repair–deficient (dMMR) tumors are characterized by high mutational load and lymphocyte infiltration and may be good candidates for immune checkpoint blockade. Pembrolizumab is a monoclonal antibody against PD-1 that is designed to block its interaction with PD-L1 and PD-L2 and thus allow an antitumor immune response. In the KEYNOTE-016 proof-of-concept study, pembrolizumab showed promising antitumor activity against dMMR tumors in patients (pts) with treatmentrefractory metastatic colorectal carcinoma (mCRC). KEYNOTE-177 is an international, randomized trial designed to evaluate the efficacy and safety of pembrolizumab compared with standard-of-care (SOC) chemotherapy in the first-line setting for dMMR or microsatellite instability–high (MSI-H) mCRC. Methods: Key eligibility criteria include age $ 18 y, confirmed MSIH or dMMR mCRC, ECOG PS 0-1, no active autoimmune disease or brain metastases, and no prior therapy for metastatic disease. Pts are to be randomized 1:1 to receive either pembrolizumab 200 mg Q3W or investigator’s choice of SOC chemotherapy. Chemotherapy must be chosen prior to randomization; options include mFOLFOX6 or FOLFIRI alone or in combination with bevacizumab or cetuximab. Treatment is to continue until PD, unacceptable toxicity, pt/investigator decision, or completion of 35 cycles (pembrolizumab only). Response is to be evaluated every 9 wk per RECIST v1.1 by central imaging vendor review and per RECIST adapted for immunotherapy response patterns. Eligible pts may continue pembrolizumab beyond initial RECIST-defined progression. Pts in the SOC arm who have PD and meet crossover criteria may be eligible to receive pembrolizumab for up to 17 treatment cycles. AEs are to be assessed throughout treatment and for 30 d thereafter (90 d for serious AEs) and graded per NCI CTCAE v4.0. Pts are to be followed for survival every 9 wk. PFS per RECIST v1.1 is the primary end point; OS and ORR are key secondary end points; others include duration of response and health-related quality of life. Planned enrollment in KEYNOTE-177 is 270 pts. Clinical trial information: NCT02563002.

Background: Hand-foot syndrome (HFS) is a common side effect of capecitabine. S-1 is an oral fluoropyrimidine with comparable efficacy to capecitabine in gastrointestinal cancers but associated with a lower incidence of HFS in Asian patients (pts). The Dutch SALTO study (NCT01918852) was designed to compare the incidence of HFS between S1 and capecitabine as 1stline treatment in mCRC pts. Methods: Previously untreated mCRC pts, WHO performance status (PS) 0-2, and planned treatment with fluoropyrimidine monochemotherapy were randomized between capecitabine 1250 mg/m2 for pts , 70 years of age, and 1000 mg/m2 for pts $ 70 years of age, b.i.d. day 1-14 (arm A) and S-1 30 mg/m2 b.i.d.day 1-14 (arm B). Cycles were administered q 3 wks. Cotreatment with bevacizumab 7.5 mg/kg/3wks iv was optional. Pts were stratified for use of bevacizumab, PS, serum LDH, and institution.Toxicity was assessed prior to each cycle by the investigator using NCI CTC 4.0 and by patients using a diary. Tumor response was assessed every 9 wks. Primary endpoint was the incidence of HFS. A total of 150 pts was required to demonstrate a difference in HFS of $ 20% with 90% power (a = 0.05, 2-tailed test). Secondary endpoints were grade 3 HFS, other toxicities, progression-free survival (PFS), response rate (RR), overall survival (OS). Results: A total of 161 pts were randomized between January 2014 - July 2015, 81 in arm A and 80 in arm B. Median age is 73 yrs (50-86), 10% of pts have PS2, 58% were planned to receive bevacizumab. Median follow-up is 9.4 months, 39 pts (24%) are still on study. Pts received a median of 6 cycles. In arm A and B the investigatorassessed all grade HFS is 70% and 41% (p .00025), and grade 3 HFS 20% and 4% (p .0025), resp. Other grade $ 3 toxicity only differed for anorexia (2.5% and 12.5%, p .018). All grade HFS assessed by 91 pts (preliminary results) is 82% and 57% (p .014), and grade 3 16% and 4% (p .08). Median PFS is 8.3 months and not significantly different between treatment arms (p .88, HR 0.97 [0.67-1.41]). Data on OS are not mature. Conclusions: Treatment with S-1 results in a significantly lower incidence of HFS compared to capecitabine in pts with mCRC, without compromising efficacy. Clinical trial information: NCT01918852.

4102

Poster Session (Board #94), Sat, 8:00 AM-11:30 AM

Genomic profiling and efficacy of anti-EGFR therapy in appendiceal adenocarcinoma. First Author: Maria Ignez Freitas Melro Braghiroli, Memorial Sloan Kettering Cancer Center, New York, NY Background: Metastatic appendiceal adenocarcinoma (AAC) is a heterogenous disease and a large proportion of patients present with diffuse metastases in the peritoneal cavity. Cytotoxic and targeted therapies are typically extrapolated from colorectal adenocarcinoma (CRC), however, the efficacy is not well studied. Herein we investigated the genetic profiles of these tumors in an effort to identify molecular characteristics and potentially actionable mutations, as well as the response to anti-EGFR therapy in RAS/ BRAF wild type (wt) AAC. Methods: We identified patients (pts) with ACC treated at MSKCC from 2002 to 2016 who had undergone molecular profiling, either by next generation sequencing using our MSK-IMPACT platform, or by MALDI-TOF mass spectroscopy genotyping (Sequenom). MSK-IMPACT tumors and matched normal samples were analyzed on a 410 gene panel. Sequenom provided an 8 gene panel including KRAS, NRAS, BRAF, and PIK3CA. Via an IRB approved waiver, we collected tumor histology and evaluated those who were RAS/RAF wt and were treated with antiEGFR therapy. Results: To date, we identified a total 109 AAC patients, of whom 60 had Sequenom testing and 49 had MSK-IMPACT. Among pts analyzed with MSK-IMPACT, 34 had mucinous adenocarcinoma, 6 adenocarcinoma, 9 adenocarcinoma ex goblet cell carcinoid. In total 194 alterations were identified with a median 3.9 alterations/patient (range 0-10). Alterations were seen most commonly in KRAS (57%), GNAS (33%), TP53 (29%), and SMAD4 (6%). Potentially treatable alterations were present in 12% of patients and included BRAF V600E (1), MTOR (2), ERBB2 (1) and NTRK(2). Of the total 109 pts, 51 (47%) were RAS/BRAF wt. Of those, 12 evaluable patients received anti-EGFR therapy with either panitumumab or cetuximab. There were no responders. Conclusions: Mutational sequencing in AAC indicates that 12% have mutations in genes such as BRAF V600E, MTOR, ERBB2 and NTRK with the potential to expand investigational options. Additionally, in RAS/BRAF wt pts, panitumumab/cetuximab does not appear to have therapeutic efficacy comparable to historic controls in RAS/RAF wt CRC.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

206s 4000

Gastrointestinal (Noncolorectal) Cancer Oral Abstract Session, Mon, 8:00 AM-11:00 AM

A multicenter randomized phase III trial of neo-adjuvant chemotherapy followed by surgery and chemotherapy or by surgery and chemoradiotherapy in resectable gastric cancer: First results from the CRITICS study. First Author: Marcel Verheij, Department of Radiation Oncology, The Netherlands Cancer Institute/Antoni van Leeuwenhoek Ziekenhuis, Amsterdam, Netherlands Background: The mainstay of potentially curative treatment of gastric cancer is radical surgical resection. Because most patients in the Western world present with advanced stages long-term survival remains poor at about 25%, with local recurrences as part of treatment failure in up to 80% of cases. Postoperative chemoradiotherapy (CRT) and perioperative chemotherapy (CT) have demonstrated a survival benefit over surgery alone. The current randomized phase III CRITICS-study (NCT00407186) investigated whether chemoradiotherapy after neo-adjuvant chemotherapy and adequate (D2) surgery leads to improved overall survival (OS) in comparison with postoperative chemotherapy. Furthermore, toxicity of both treatment regimens was explored. Methods: Patients with stage Ib-IVa resectable gastric cancer were randomized after diagnosis. Neo-adjuvant CT was prescribed in both arms and consisted of 3 courses of epirubicin, cisplatin/oxaliplatin and capecitabine (ECC/EOC). After gastric cancer resection, patients received another 3 courses of ECC/EOC or CRT (45 Gy in 25 fractions combined with weekly cisplatin and daily capecitabine). Primary endpoint is OS; secondary endpoints are: disease free survival, toxicity profile and quality of life. Results: Between January 2007 and April 2015, 788 patients from The Netherlands, Sweden and Denmark were randomized (393 CT; 395 CRT). Baseline characteristics were well balanced with 70% males and a median age of 61 years. 84% completed 3 cycles before surgery. In the CT arm 46% and in the CRT arm 55% completed treatment according to protocol. After a median follow-up of 50 months, 405 patients have died. The 5-year survival is 41.3% for CT and 40.9% for CRT (p=0.99). Toxicity was mainly hematological (grade III or higher: 44% vs 34%; p=0.01) and gastrointestinal (grade III or higher: 37% vs 42%; p=0.14) for CT and CRT, respectively. Conclusion: No significant difference in overall survival was found between postoperative chemotherapy and chemoradiotherapy. Clinical trial information: NCT00407186.

4002

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

LBA4001

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

FAST: An international, multicenter, randomized, phase II trial of epirubicin, oxaliplatin, and capecitabine (EOX) with or without IMAB362, a first-in-class anti-CLDN18.2 antibody, as first-line therapy in patients with advanced CLDN18.2+ gastric and gastroesophageal junction (GEJ) adenocarcinoma. First Author: Salah-Eddin Al-Batran, Institute of Clinical Cancer Research, Nordwest Hospital, Frankfurt Am Main, Germany

The full, final text of this abstract will be available at abstracts.asco.org at 7:30 AM (EDT) on Sunday, June 5, 2016, and in the Annual Meeting Proceedings online supplement to the June 20, 2016, issue of the Journal of Clinical Oncology. Onsite at the Meeting, this abstract will be printed in the Monday edition of ASCO Daily News.

4003

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

Discontinuation of first-line chemotherapy (CT) after 6 weeks of CT in patients (pts) with metastatic squamous-cell esophageal cancer (MSEC): A randomized phase II trial. First Author: Antoine Adenis, Centre Oscar Lambret, Lille, France

Phase III randomized study of sorafenib plus doxorubicin versus sorafenib in patients with advanced hepatocellular carcinoma (HCC): CALGB 80802 (Alliance). First Author: Ghassan K. Abou-Alfa, Memorial Sloan Kettering Cancer Center and Weill Cornell Medical College, New York, NY

Background: Even though there is no evidence to support the use of CT in MSEC, many physicians treat pts with good ECOG performance status (PS) with fluorouracil (5FU)/platinum-based CT. Therefore, in order to estimate overall survival (OS) in MSEC, we designed a discontinuation phase 2 trial in pts free from progression (PD) after 6 weeks (wks) of CT. Methods: PS,3 MESC pts were treated with 1st-line 5FU/platinum-based CT, and underwent tumor assessment at 6 wks. Pts free from PD after 6 wks of CT were randomized (1:1) to receive CT continuation (arm A) or CT discontinuation plus BSC (arm B). In arm B, pts were allowed to restart CT in case of PD. Primary endpoint was ITT 9-month OS rate (calculated from randomization date). The sample size has been calculated to provide an estimate of ITT 9-month OS rate 6 12.5% in arm A. Arm B served as an internal control, without formal comparison intent.With an anticipated 58% dropout rate at 1st tumor assessment, 106 pts were needed to randomize 62 pts. Secondary objectives were tolerance, PFS, quality of life, and medical costs. Results: 105 pts were included, and 101 received CT (FOLFOX: 76/101; LV5FU2-CDDPq2w: 18/101; TPF: 4/101, FU-CDDPq3w: 3/101). 67/101 pts free from PD at 6 wks were randomized. Baseline pts characteristics were as follows: median age: 64; male gender: 54/67; PS #1/2: 61/6; number of metastatic sites 1/ $2: 31/36; BMI,18.5kg/m²: 6/67; prior exposure to CT (combined to radiation therapy and/or surgery): 37/67. 64/67 pts were eligible and treated (arm A 31, arm B 33). CTs were LV5FU2-CDDPq2w 7/31, FOLFOX 24/31. ITT 9-months OS rate was 50% (85%-CI: 37-62) for arm A and 44% (85%CI: 31-56) for arm B. PFS after randomization was 2.8 mo (95% CI: 1.7-5) for arm A, and 1.4 mo (95% CI: 1.4-2.7) for arm B. Tolerance was good, as expected with the CT we used. Conclusions: Inpts with MESC free from PD after 6 wks of 5FU/platinum-based CT who were randomized to the CTcontinuation arm, 9-month OS rate in arm A was 50%. Despite a trend in OS and PFS favoring arm A, it does not appear that CT continuation provides much clinical benefit over CT discontinuation plus BSC, in such pts. Results of quality of life and medical costs are awaited. Clinical trial information: NCT01248299.

Background: An exploratory analysis of a randomized phase II study in HCC comparing doxorubicin (D) alone to doxorubicin plus sorafenib (D+S) showed a significant improvement in overall survival favoring D+S (JAMA, 2011). The results appeared promising compared to the historic outcomes seen in the pivotal sorafenib (S) trials. CALGB 80802 was designed to determine if D+S improved survival compared to S alone. Methods: Patients with histologically proven advanced HCC, no prior systemic therapy and Child-Pugh A were randomized to receive D 60 mg/m2 every 21 days plus S 400 mg PO twice daily (D+S) or S alone. For bilirubin $ 1.3x normal, D and S doses were halved. D was maxed out at 360 mg/m2. The study was stratified by extent of disease (locally advanced; metastatic), the primary endpoint was overall survival (OS); and secondary endpoint progression-free survival (PFS). The final analysis was to occur when 364 events were observed among 480 total patients, with 90% power to detect a 37% increase in median OS (10.7 to 14.7 months; 1-sided a = 0.05). Results: The Alliance DSMB halted the study after accrual of 346 patients (173 on each of D+S and S) when a futility boundary was crossed at a planned interim analysis. With 107 events in each arm, median OS was 9.3 months (95%CI 7.1-12.9) for D+S, and 10.5 months (95% CI 7.4-14.3) for S with a hazard ratio (HR) 1.06 (95% CI 0.8- 1.4) for D+S vs. S. Median PFS was 3.6 (95% CI 2.8-4.6) and 3.2 months (95% CI 2.3-4.1), respectively (HR = 0.90, 95% CI 0.72-1.2). There were 38 deaths on treatment: 18 on D+S and 20 on S. Among these 8 [sepsis (1), dysphagia (1), pneumonia (1), cardiac (2), hepatic failure (2), and not otherwise specified (1)] on D+S, and 3 [fatigue (1), hepatic failure (1), and a secondary malignancy (1)] on S, were at least possibly related to treatment. A maximum grade 3 or 4 only hematologic adverse events (AE) occurred in 37.8% of patients on D+S and 8.1% of patients on S. Nonhematologic AEs were comparable, in 63.6% and 61.5% of patients, respectively. Conclusions: The addition of D to S resulted in higher toxicity and did not improve OS or PFS. The S median OS of about 10 months is consistent with previous reports. NCI Grant U10CA180821 Clinical trial information: NCT01015833.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Noncolorectal) Cancer 4005

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

NETTER-1 phase III: Efficacy and safety results in patients with midgut neuroendocrine tumors treated with 177Lu-DOTATATE. First Author: Jonathan R. Strosberg, Moffitt Cancer Center, Tampa, FL Background: There are limited therapeutic options for patients with advanced midgut NETs progressing on first-line somatostatin analog therapy. The purpose of this phase III trial was to evaluate the efficacy and safety of 177 Lu-DOTA0-Tyr3-Octreotate (Lutathera) in patients with advanced, progressive sstr positive midgut NETs. Methods: 230 patients with grade 1-2 metastatic midgut NETs were randomized to Lutathera, 7.4 GBq every 8 weeks (x 4 administrations) vs Octreotide LAR 60 mg every 4 weeks. Primary endpoint was PFS (RECIST 1.1) with tumor assessment every 12 weeks. Secondary objectives included ORR, OS, toxicity and QoL. Results: In the intent-to-treat population (ITT), the median PFS was not reached for Lutathera and was 8.4 months with control (p , 0.0001, HR 0.21). There were 23 centrally confirmed disease progressions or deaths in the Lutathera arm and 67 in the Octreotide LAR 60 mg arm. The objective radiographic response rate (ORR) was 18% with Lutathera and 3% with control (p = 0.0008). Besides the scintigraphic 111In-pentetreotide tumor uptake score (Krenning scale . = 2), tumor burden and Ki67 grade had no significant effect on clinical efficacy outcomes (PFS, OS, TTP) in the Cox regression models. Interim OS analysis (13 deaths in Lutathera group and 22 in control group; p = 0.019) strongly suggests an improvement in OS. Only 5% (6 patients) experienced dose modifying toxicity with Lutathera. Grade 3 or 4 adverse events of neutropenia, thrombocytopenia and lymphopenia occurred in 1%, 2% and 9% of patients in Lutathera arm vs. none in controls. Conclusions: The NETTER-1 trial provides evidence for a clinically meaningful and statistically significant increase in PFS and ORR, and suggests a potential survival benefit in patients with advanced midgut NETs treated with Lutathera in both ITT and PP analyses. Lutathera safety profile was found to be very favorable. Clinical trial information: NCT01578239.

4007

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

MAESTRO: A randomized, double-blind phase III study of evofosfamide (Evo) in combination with gemcitabine (Gem) in previously untreated patients (pts) with metastatic or locally advanced unresectable pancreatic ductal adenocarcinoma (PDAC). First Author: Eric Van Cutsem, University Hospitals Leuven, Leuven, Belgium Background: Hypoxia in PDAC is associated with disease progression and poor prognosis. Evo is a hypoxia-activated prodrug of Br-IPM that is preferentially activated under hypoxic conditions. The addition of Evo to Gem significantly improved PFS in a randomized phase II trial in advanced PDAC (NCT01144455). Methods: MAESTRO was an international, randomized, double-blind, placebo-controlled phase III trial of Evo/Gem vs Placebo (Pbo)/ Gem in pts with locally advanced unresectable or metastatic PDAC (NCT01746979). Evo 340 mg/m2 or matched Pbo and Gem 1,000 mg/m2 were administered IV on days 1, 8, and 15 of a 28-day cycle. Key eligibility criteria included ECOG PS 0/1. The primary endpoint was overall survival (OS). 660 patients were to be randomized 1:1 to obtain 508 events (deaths) to assure 90% power to detect a hazard ratio (HR) of 0.75 for OS with a twosided a of 0.05. Secondary endpoints included PFS and objective response rate (ORR). Results: From Jan 2013 to Nov 2014, 693 pts were randomized (346 Evo/Gem, 347 Pbo/Gem). Baseline characteristics were balanced. Median OS was 8.7 mo with Evo/Gem vs 7.6 mo with Pbo/Gem; HR = 0.84 (95% CI: 0.71–1.01, p = 0.059). Median PFS was 5.5 mo with Evo/Gem vs 3.7 mo with Pbo/Gem; HR = 0.77 (95% CI: 0.65–0.92, p = 0.004). Best ORR was 20% with Evo/Gem vs 16% with Pbo/Gem; odds ratio (OR) = 1.32 (95% CI: 0.88–1.97, p = 0.17). Confirmed ORR was 15% with Evo/Gem vs 9% with Pbo/Gem; OR = 1.90 (95% CI: 1.16 – 3.12, p = 0.009). Most common non-hematologic AEs of nausea (47%), decreased appetite (35%) and vomiting (33%) were similar across arms. Hematologic AEs of neutropenia, thrombocytopenia and anemia were more frequent with Evo/Gem. AEs leading to death were 9% with Evo/Gem vs 11% with Pbo/Gem. AEs leading to treatment discontinuation were 17.9% on Evo/Gem and 15.6% on Pbo/Gem. Conclusions: The primary endpoint was not met as difference in OS time was not statistically significant. Evo/Gem showed signs of antitumor activity with longer PFS and higher ORR. The safety profile was similar to that previously reported. Clinical trial information: NCT01746979.

LBA4006

207s Oral Abstract Session, Mon, 8:00 AM-11:00 AM

ESPAC-4: A multicenter, international, open-label randomized controlled phase III trial of adjuvant combination chemotherapy of gemcitabine (GEM) and capecitabine (CAP) versus monotherapy gemcitabine in patients with resected pancreatic ductal adenocarcinoma. First Author: John P. Neoptolemos, University of Liverpool, Liverpool, United Kingdom

The full, final text of this abstract will be available at abstracts.asco.org at 2:00 PM (EDT) on Friday, June 3, 2016, and in the Annual Meeting Proceedings online supplement to the June 20, 2016, issue of the Journal of Clinical Oncology. Onsite at the Meeting, this abstract will be printed in the Monday edition of ASCO Daily News.

4008

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

PET-PANC: Multi-centre prospective diagnostic accuracy and clinical value trial of FDG PET/CT in the diagnosis and management of suspected pancreatic cancer. First Author: Paula Ghaneh, University of Liverpool, Liverpool, United Kingdom Background: Pancreatic cancer diagnosis and staging is challenging. FDG PET/CT adds anatomic localization to functional data. The aim of this study was to determine the impact of FDG PET/CT in addition to standard diagnostic workup in patients with suspected pancreatic cancer. Methods: Patients with suspected pancreatic cancer underwent FDG PET/ CT following multi-detector CT (MDCT). FDG PET/CT scans were reviewed and quality assured centrally. Diagnosis, staging and planned management were recorded before and after FDG PET/CT. Reference standard was histology or clinical outcome. Primary outcome measure was incremental diagnostic value of FDG PET/CT in addition to MDCT. Sample size was 500 patients, following interim analysis; 80% power to detect increase in sensitivity from 81% to 90% and specificity from 66% to 80%. Secondary outcome measures were changes in diagnosis, staging, and management; cost effectiveness was estimated. Results: Between January 2011 and April 2013 589 patients with suspected pancreatic cancer underwent MDCT and FDG PET/CT in 18 UK centres. 550 patients had complete data and in range FDG PET/CT. 261 patients (47%) had pancreatic ductal adenocarcinoma (PDAC). For the diagnosis of PDAC, both sensitivity (92.7% [95% CI 89.6%, 95.9%] compared to 88.5% [95% CI 84.6%, 92.4%], p=0.010) and specificity (75.8% [95% CI 70.8%, 80.7%] compared to 70.6% [95% CI 65.3%, 75.8%] p=0.023) were significantly higher for FDG PET/CT than MDCT. FDG PET/CT correctly changed the staging of PDAC in 56 patients (14%) (p=0.001). FDG PET/CT influenced management in 250 (45%) of patients. FDG PET/CT stopped futile resection in 58 patients (20%) due to have surgery. FDG PET/CT was associated with a QALY gain of 0.0157 (95% CI -0.0101, 0.0430) and cost saving of £645 (95% CI -£1314, £2743). In the base case model FDG PET/CT dominated MDCT alone and is likely to be cost effective for the UK NHS. Conclusions: FDG PET/CT provided significant incremental diagnostic benefit in the diagnosis of pancreatic cancer and had a significant influence on the staging and management of patients. FDG PET/CT was cost effective at current reimbursement rates for FDG PET/ CT to the UK NHS. Clinical trial information: 73852054.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

208s 4009

Gastrointestinal (Noncolorectal) Cancer Poster Discussion Session; Displayed in Poster Session (Board #1), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 3:00 PM-4:15 PM

Avelumab (MSB0010718C; anti-PD-L1) in patients with advanced gastric or gastroesophageal junction cancer from JAVELIN solid tumor phase Ib trial: Analysis of safety and clinical activity. First Author: Hyun Cheol Chung, Yonsei Cancer Center, Seoul, Korea, The Republic of Background: Avelumab (proposed INN) is a fully human anti-PD-L1 IgG1 antibody under clinical investigation in multiple cancers. We report safety and clinical activity of avelumab as a first-line maintenance (Mn) or second-line (2L) therapy in patients (pts) with advanced gastric or gastroesophageal junction cancer (GC/GEJ; NCT01772004), including activity associated with PD-L1 expression. Methods: Pts with GC/GEJ received avelumab at 10 mg/kg IV Q2W until progression, unacceptable toxicity, or withdrawal. Response was assessed every 6 wks (RECIST 1.1). Best overall response and progression-free survival (PFS) were evaluated. Adverse events (AEs) were graded by NCI CTCAE v4.0. PD-L1 expression was assessed by IHC. Results: As of Oct 23 2015, 151 pts were treated with avelumab (62 pts, 2L; 89 pts, Mn) and followed for median of 49 wks (range 9-84). Treatment-related adverse events (TRAEs) of any grade occurred in 89 pts (58.9%); most common ( . 10%) were infusion-related reaction (19 [12.6%]) and fatigue (16 [10.6%]). Grade $ 3 TRAEs were reported in 15 pts (9.9%); fatigue, asthenia, increased GGT, thrombocytopenia, and anemia occurred in . 1 pt (2 each; 1.3%). There was 1 treatment-related death (hepatic failure/autoimmune hepatitis). Fourteen pts had an unconfirmed response: 2L 6/62 (9.7%), all PRs; Mn 8/89 (9.0%), 2 CRs, 6 PRs. In 2L and Mn pts, disease control rate was 29.0% and 57.3%, and median PFS was 6.0 wks (95% CI: 5.7, 6.4) and 12.0 wks (95% CI: 9.9, 17.6), respectively. PDL1 expression was evaluable in 74 pts (22/62 2L, 52/89 Mn). Activity based on a $ 1% cutoff for tumor cell staining is shown in the table. Conclusions: Single-agent avelumab had an acceptable safety profile and promising clinical activity in unselected pts with GC/ GEJ treated in Mn and 2L settings. These data represent the largest study of anti-PD-(L)1 agents in GC/GEJ. Two randomized phase III trials of avelumab in GC are open. Clinical trial information: NCT01772004. 2L (n = 22)

ORR, % (95% CI) Median PFS, wks (95% CI)

Mn (n = 52)

PD-L1+ (n = 11)

PD-L12 (n = 11)

PD-L1+ (n = 20)

PD-L12 (n = 32)

18.2 (2.3, 51.8) 6.3 (5.4, 18.0)

9.1 (0.2, 41.3) 10.4 (4.1, 21.9)

10.0 (1.2, 31.7) 17.6 (6.0, 24.1)

3.1 (0.1, 16.2) 11.6 (5.7, 14.1)

4010

Poster Discussion Session; Displayed in Poster Session (Board #2), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 3:00 PM-4:15 PM

CheckMate-032: Phase I/II, open-label study of safety and activity of nivolumab (nivo) alone or with ipilimumab (ipi) in advanced and metastatic (A/M) gastric cancer (GC). First Author: Yelena Yuriy Janjigian, Gastrointestinal Oncology Service, Department of Medicine, Memorial Sloan Kettering Cancer Center, New York, NY Background: Chemotherapy-refractory GC is a uniformly fatal illness and an unmet therapeutic need. Nivo is a fully human anti-PD-1 monoclonal antibody with a manageable safety profile and efficacy in solid tumors. We report results of a completed A/M GC cohort of CheckMate-032 study evaluating nivo monotherapy and nivo with ipi in pts with solid tumors. Methods: Pts with A/M, histologically confirmed GC, esophageal (EC), or gastroesophageal junction cancer (GEC), irrespective of PD-L1 status, who had progressed on chemotherapy, were treated with nivo 3 mg/kg Q2W (N3), nivo 1 mg/kg + ipi 3 mg/kg (N1+I3), or nivo 3 mg/kg + ipi 1 mg/kg (N3+I1) Q3W x 4 cycles, followed by nivo 3 mg/kg Q2W until confirmed disease progression or intolerable toxicity. Primary endpoint was ORR; other endpoints included safety, OS, and biomarker status. Results: 160 pts with A/M GC were treated with N3 (n = 59), N1+I3 (n = 49), or N3+I1 (n = 52). Baseline characteristics were comparable across cohorts; the majority of pts had $ 2 prior regimens. Treatment-related AEs (TRAE) of any grade occurred in 70% (N3), 84% (N1+I3), and 75% (N3+I1) of pts; Grade 3-4 TRAEs occurred in 17%, 45%, and 27% of pts. 12% of pts stopped therapy due to treatment toxicity: 5% (N3), 22% (N1+I3), and 12% (N3+I1). Treatment-related serious AEs (TRSAEs) of any grade and Grade 3-4 TRSAE occurred in 10% and 5% (N3), 43% and 35% (N1+I3), and 23% and 15% (N3+I1) of pts. There was one Grade 5 TRSAE of tumor lysis syndrome (N3+I1). 154 (96%) pts were evaluable for efficacy with the confirmed ORR 16%: 14% (N3), 26% (N1+I3), and 10% (N3+I1), including 2 pts with CR (1 in N3; 1 in N1+I3) and a disease control rate (ORR+SD) of 38%. OS data are in the Table below with 15 pts (9%) remaining on therapy. Updated biomarker data will be presented. Conclusions: TRAEs for nivo and nivo + ipi were consistent with those previously reported. Clinical activity and OS in pts with chemotherapy refractory disease are encouraging. These data support ongoing evaluation of nivo and nivo + ipi in A/M GC/EC/GEC. Clinical trial information: NCT01928394.

OS Rate, % (95% CI) 6 mo 12 mo Median OS, mo (95% CI)

4011

Poster Discussion Session; Displayed in Poster Session (Board #3), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 3:00 PM-4:15 PM

A randomized, open-label, two-arm phase II trial comparing the efficacy of sequential ipilimumab (ipi) versus best supportive care (BSC) following firstline (1L) chemotherapy in patients with unresectable, locally advanced/ metastatic (A/M) gastric or gastro-esophageal junction (G/GEJ) cancer. First Author: Markus H. Moehler, Johannes-Gutenberg Universitat ¨ Mainz, Mainz, Germany Background: Pts with advanced gastric cancer have a poor prognosis with median overall survival (OS) of ~1 yr. Ipi is a monoclonal antibody that enhances T-cell activation and T-effector cell tumor infiltration and activity by inhibiting CTLA-4 receptor interaction with its ligands. This study (NCT01585987) evaluated efficacy of ipi vs BSC as sequential/maintenance therapy in pts with unresectable locally A/M G/GEJ cancer following 1L chemotherapy with platinum plus fluoropyrimidine. Methods: Eligible pts $18 years without progressive disease (PD) following 1L chemotherapy were randomized to ipi 10 mg/kg Q3W for 4 doses then ipi 10 mg/kg Q12W for up to 3 yrs, or BSC. Primary endpoint was immunerelated progression-free survival (irPFS); secondary endpoints were PFS by modified WHO criteria (mWHO) and OS. Post-interim analysis (IA), the study was stopped. Updated OS are presented. Results: Of 143 pts screened, 57 were randomized to each arm. At the IA, improvement in irPFS with ipi vs BSC was not observed (HR=1.44 [80% CI: 1.09, 1.91], p=0.097). Treatment-related adverse events (AEs) occurred in 41/57 (72%) of ipi pts and 25/45 (56%) pts on active BSC. The most frequent ipirelated AEs of any grade were pruritus (32%), diarrhea (25%), fatigue (23%), and rash (18%), with no gastrointestinal perforations in the ipi arm. Conclusions: The irPFS was similar and median OS for both arms was approximately 1 yr. AEs were consistent with other ipi studies. These results suggest ipi in combination with chemotherapy warrants further study in pts with A/M G/GEJ cancer. Clinical trial information: NCT01585987.

Survival. Ipi (n=57)

BSC (n=57)

51 (90) 2.9 1.6–5.2

p-value

HR (80% CI)

40 (70) 4.9 3.5–6.5

0.097

1.44 (1.09–1.91)

52 (91) 2.7 1.4–3.0

40 (70) 4.9 3.5–6.1

0.034

1.59 (1.20–2.10)

25 (44) 16.8 11.8–23.1

25 (44) 12.1 9.3–NE

0.643

0.87 (0.60–1.27)

36 (63) 12.7 10.5–18.9

30 (53) 12.1 9.3–NE

4012

N3

N1+I3

N3+I1

49 (35–62) 36 (21–51) 5.0 (3.4–12.4)

54 (39–67) 34 (19–50) 6.9 (3.6–NA)

43 (29–57) NA 4.8 (3.0–9.1)

Poster Discussion Session; Displayed in Poster Session (Board #4), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 3:00 PM-4:15 PM

Phase I/II safety and antitumor activity of nivolumab (nivo) in patients (pts) with advanced hepatocellular carcinoma (HCC): Interim analysis of the CheckMate-040 dose escalation study. First Author: Anthony B. ElKhoueiry, University of Southern California Norris Comprehensive Cancer Center, Los Angeles, CA Background: For pts with advanced HCC on sorafenib (sor), overall survival (OS) is 11 mo; median OS with best supportive care (BSC) post-sor failure is 7–8 mo.Safety and preliminary antitumor efficacy of nivo, a fully human IgG4 mAb PD-1 inhibitor, was evaluated in a multiple ascending-dose, phase 1/2 study in pts with advanced HCC with clinical follow-up to 3 yrs. Methods: Pts had histologically-confirmed, advanced HCC, Child-Pugh (CP) score # 7, and previously failed, refused, or were intolerant of sor. Dose escalation occurred in 3 parallel cohorts by etiology: no active hepatitis virus infection, HBV-infection, or HCV-infection. Pts received nivo 0.1–10 mg/kg for up to 2 yrs. The primary endpoint was safety. Secondary endpoints included antitumor activity by RECIST 1.1 and DOR; exploratory endpoints included biomarker assessment. Results: 51 pts were enrolled and treated with nivo: baseline CP scores were 5 (n=44) or 6 (n=7), 76% had extrahepatic metastasis, 12% had vascular invasion, and 73% had prior sor. 10 pts remain on study; 41 discontinued, most (n=35) due to PD and 1 due to a treatment-related adverse event (TRAE) of ALT and AST increase. TRAEs occurred in 39 pts (77%); most common were rash and AST increase (20% each). Grade 3-4 TRAEs were seen in 10 pts (20%); most common were AST increase (10%), lipase and ALT increase (6% each). A maximum tolerated dose was not reached. Efficacy data are reported in the Table. 48 pts were evaluable for response. Responses were observed in pts with and without quantifiable PD-L1 as assessed by IHC. Antiviral responses in HCV-infected pts have been observed. Conclusions: Nivo was well tolerated with a manageable safety profile. Treatment produced durable responses and disease stabilization across all dose levels and cohorts. Reported OS rates were favorable relative to historical data for BSC. Clinical trial information: NCT01658878.

a

irPFS Events, n (%) Median (mos) 95% CI PFS by mWHOa Events, n (%) Median (mos) 95% CI OSa Events, n (%) Median (mos) 95% CI OSb Events, n (%) Median (mos) 95% CI a

at IA bat study end; NE, not estimable.

Efficacy Total (N=48) ORR CR, n (%) PR, n (%) SD, n (%) PD, n (%) Not evaluable Median DOR, mo Pts with decline in tumor burden from baseline, n (%) Median OS,a (95% CI), mo OS Rate,a % 6 mo 9 mo 12 mo 18 mo a

7 (15) 3 (6) 4 (8) 24 (50) 15 (31) 2 (4) 23.7 17 (37) 15.1 (9.6, 28.6) 67 67 59 48

Based on entire pt population.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Noncolorectal) Cancer 4013

Poster Discussion Session; Displayed in Poster Session (Board #5), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 3:00 PM-4:15 PM

4014

209s

Poster Discussion Session; Displayed in Poster Session (Board #6), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 3:00 PM-4:15 PM

Efficacy of combined endoscopic resection and chemoradiotherapy for clinical stage I esophageal squamous cell carcinoma (ESCC): A singlearm confirmatory study (JCOG0508). First Author: Manabu Muto, Department of Clinical Oncology, Kyoto University Hospital, Kyoto, Japan

Phase III study of intraperitoneal paclitaxel plus s-1/paclitaxel compared with s-1/cisplatin in gastric cancer patients with peritoneal metastasis: PHOENIX-GC trial. First Author: Hironori Ishigami, The University of Tokyo, Tokyo, Japan

Background: For clinical(c) stage I submucosal (T1b) thoracic ESCC, surgery with lymph node dissection is the standard treatment. However, pathologic diagnosis after resection sometimes reveals mucosal cancer (T1a), which is treatable by endoscopic resection (ER) alone. T1b ESCC is also treatable with chemoradiotherapy (CRT) but has risk of loco-regional failure and cardiopulmonary adverse events (AEs). We investigated the efficacy and safety of combined ER+CRT for cT1b ESCC. Methods: The eligibility criteria included: histologically proven thoracic ESCC, T1b diagnosed by endoscopic ultrasound, cN0M0, primary tumor size # 5 cm and esophageal lumen circularity # 3/4, and age 20-75 years. After ER, additional treatment was indicated by histologic diagnosis: Group A, pT1a with negative resection margin and no vascular invasion -no additional treatment; Group B, pT1b with negative resection margin and pT1a with vascular invasion -prophylactic CRT; Group C, pT1b with positive resection margin -definitive CRT. Chemotherapy comprised of 5-fluorouracil (700 mg/ m2/day, days 1-4 and 29-32, civ), and cisplatin (70 mg/ m2/day, days 1 and 29). Radiotherapy dose was 41.4 Gy/23 fr, delivered to the loco-regional lymph-node area (Group B) or 50.4 Gy/28 frs with a boost to the primary site (Group C). Primary endpoint was 3-year overall survival (OS) of Group B. Key secondary endpoint was 3-year OS of all enrolled patients (pts). The required sample size was 82 for primary analysis, with one-sided alpha of 0.05 and power of 90%, based on the expected and threshold 3-year OS as 90% and 80%. Results: Of 177 pts registered in 2006-2012 [M/F, 147/30; Age, 63 years (42-75); lesion diameter, 2.5cm (0.5-5.0)], ER was performed in 176 pts (Group A/B/C, 74/87/15). There were no grade 3/4 AEs due to ER; 8 of 96 pts who received CRT suffered from late cardio-pulmonary AEs. The 3-year OS was 90.7% (90% CI; 84.0-94.7) of 87 pts in Group B and 92.6% (90% CI; 88.5-95.2) overall. Conclusions: Combined ER+CRT for T1b ESCC is considered to be comparable to surgery in efficacy and can be new minimally invasive treatment option. Clinical trial information: 000000553.

Background: Intraperitoneal (IP) paclitaxel (PTX) provides sustained high local concentrations, and its efficacy has been shown in ovarian cancer. We developed a regimen combining IP PTX with S-1/PTX for the treatment of gastric cancer patients, and obtained promising results with a one-year overall survival (OS) rate of 78% in a phase II study. This multicenter phase III study evaluated the efficacy of IP PTX plus S-1/PTX compared to standard systemic chemotherapy. Methods: Eligibility criteria included pathologically confirmed gastric adenocarcinoma, peritoneal metastasis, and no or shortterm (, 2 months) prior chemotherapy. Patients were randomized 2:1 to an IP (IP PTX plus S-1/PTX; IP PTX 20 mg/m2, intravenous [IV] PTX 50 mg/m2 on days 1 and 8, and S-1 80 mg/m2/day on days 1-14, q3 weeks) arm or a SP (S-1/cisplatin; IV cisplatin 60 mg/m2 on day 8, and S-1 80 mg/m2/day on days 1-21, q5 weeks) arm. Randomization was stratified by center, having received prior chemotherapy, and the extent of peritoneal disease. The primary endpoint was OS. Secondary endpoints were response rate and safety. Results: Between October 2011 and November 2013, 183 patients were enrolled, and 164 patients were included in the efficacy analysis. Baseline patient characteristics were balanced between the two arms except for ascites. Of 45 patients with massive ascites (beyond the pelvic cavity), 38 (84%) were randomized to the IP arm and 7 (16%) to the SP arm. The median OS for IP and SP were 17.7 and 15.2 months, respectively (stratified log-rank test, p = 0.080; hazard ratio [HR] 0.72, 95% confidence interval [CI] 0.49-1.04, p = 0.081). For a sensitivity analysis using a stratified Cox regression model, adjusting for the baseline ascites, the HR was 0.59 (95% CI 0.39-0.87, p = 0.0079). The overall response rate was 53% in the IP arm, and 60% in the SP arm (p = 1.0). Both regimens were tolerable, and there were no treatment-related deaths. Conclusions: The primary analysis did not show the statistical superiority of the IP regimen. The sensitivity analysis, considering the imbalance of ascites, suggested clinical efficacy of IP PTX in gastric cancer with peritoneal metastasis. Clinical trial information: UMIN000005930.

4015

4016

Poster Discussion Session; Displayed in Poster Session (Board #7), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 3:00 PM-4:15 PM

Phase III trial of s-1 plus oxaliplatin (SOX) vs s-1 plus cisplatin (SP) combination chemotherapy for first-line treatment of advanced gastric cancer (AGC): SOPP study. First Author: Min-Hee Ryu, Department of Oncology, Asan Medical Center, University of Ulsan College of Medicine, Seoul, South Korea Background: S-1 plus cisplatin (SP) is one of the standard first-line chemotherapies for AGC in the East Asia. Oxaliplatin is generally less toxic and more convenient than cisplatin. This study compared the safety and efficacy of S-1 plus oxaliplatin (SOX) with those of SP. Methods: This SOPP study was a multi-center, randomized, open-label, phase III study to evaluate whether SOX was non-inferior / superior to SP in terms of progression-free survival (PFS) according to RECIST v1.1. Patients (pts) with metastatic or recurrent gastric or gastroesophageal junction adenocarcinoma and no prior chemotherapy were randomized 1:1 to receive either SOX (S-1 80 mg/m2/ day on D1-14, oxaliplatin 130 mg/m2 on D1, every 3 weeks) or SP (S-1 80 mg/m2/day on D1-14, cisplatin 60mg/m2 on D1, every 3 weeks) until disease progression or unacceptable toxicities. Results: Between October 2012 and October 2014, a total of 338 pts were randomized from 10 sites in Korea. The median age was 56 years (range, 26-80). 98% of pts had ECOG performance status 0-1. 51% of pts had measurable lesions. With a median follow-up of 15.6 months (range, 0.1-33.3) in surviving pts, SOX was significantly non-inferior but not superior to SP in PFS (median 5.6 months vs. 5.7 months; HR 0.85, 95% CI 0.67-1.07, p = 0.169). In pts with measurable disease, overall response rates (ORR) were similar between SOX and SP (58% vs. 60%, p = 0.7). Overall survival (OS) of both groups was not different, either (median 12.9 vs. 11.4 months; HR 0.86, 95% CI 0.661.11, p = 0.242). Treatment was generally well tolerated in both arms. While grade 3/4 neutropenia and febrile neutropenia were more common in SP, grade 3/4 peripheral neuropathy and thrombocytopenia were more common in SOX. Conclusions: SOX was non-inferior to SP in terms of PFS, ORR, and OS. The two regimens were well tolerated with different toxicity profiles. SOX regimen can be also recommended as a first-line treatment of AGC. Clinical trial information: NCT01671449.

Poster Discussion Session; Displayed in Poster Session (Board #8), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 3:00 PM-4:15 PM

Efficacy and safety findings from DREAM: A phase III study of DHP107 (oral paclitaxel) vs IV paclitaxel in patients with gastric cancer after failure of firstline chemotherapy. First Author: Yoon-Koo Kang, Asan Medical Center, University of Ulsan, Seoul, South Korea Background: Paclitaxel is currently only available as an IV formulation. DHP107 is a novel oral formulation composed of lipid ingredients and paclitaxel. DHP107 has demonstrated comparable efficacy, safety and pharmacokinetics to IV paclitaxel as second-line therapy in patients (pts) with advanced gastric cancer (AGC). DREAM (#NCT01839773) is a multicenter, open-label, prospective, randomized phase III study of pts with histologically or cytologically confirmed, unresectable recurrent/AGC after failure of first-line therapy. Methods: Pts were randomized 1:1 to either DHP107 (200 mg/m2 po bid d1, 8 & 15 q4 w) or IV paclitaxel (175 mg/m2 infused d1 q3 w). Pts were stratified by ECOG PS, disease status and prior treatment with response evaluation (RECIST v1.1) every 6 w (61 w). Primary endpoint: non-inferiority of progression-free survival (PFS); secondary endpoints: overall response rate (ORR), overall survival (OS), safety. Results: 236 pts (n = 118 in each arm) were randomized; baseline characteristics were balanced. Prior first-line therapy was fluoropyrimidine + platinum in 93% of pts. Median PFS (per protocol) was 3.3 m (95% CI 1.9–4.0) for DHP107 and 2.7 m (1.9–2.9) for paclitaxel (HR = 0.86; 95% CI 0.65–1.14; p = 0.294). Median OS (ITT) was 9.7 m (95% CI 7.1–11.5) for DHP107 vs 8.9 m (95% CI 7.1–12.2) for paclitaxel (HR = 1.04; 95% CI 0.76–1.41; p = 0.824). ORR was 17.8% for DHP107 (CR 4.2%; PR 13.6%) vs 25.4% for paclitaxel (CR 3.4%; PR 22.0%; p = 0.155); DCR was similar between arms. Nausea, vomiting, diarrhea and mucositis were more common with DHP107; peripheral neuropathy was more frequent with paclitaxel. Grade $ 3 adverse events (AEs) were infrequent; the most common was neutropenia (42% vs 55%); febrile neutropenia was rare (5.9% vs 2.5%). Serious AEs and treatment discontinuations were comparable; there were no hypersensitivity reactions with DHP (vs 2.5% with paclitaxel). Conclusions: DHP107 was non-inferior to paclitaxel in terms of PFS when given as second-line treatment for AGC. Other efficacy parameters and safety signals were comparable. DHP107 is the first oral paclitaxel with proven efficacy and safety for the treatment of AGC. Clinical trial information: NCT01839773.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

210s

Gastrointestinal (Noncolorectal) Cancer

4017

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

Phase III randomized study of second line ADI-peg 20 (A) plus best supportive care versus placebo (P) plus best supportive care in patients (pts) with advanced hepatocellular carcinoma (HCC). First Author: Ghassan K. Abou-Alfa, Memorial Sloan Kettering Cancer Center, New York, NY Background: Arginine (Ar) depletion is a putative target in HCC, which lacks the citrulline (Ci) to Ar repleting enzyme argininosuccinate synthetase. A is an Ar degrading enzyme cloned from M. hominis and produced in E. coli and conjugated with polyethylene glycol. Methods: Pts with histologically proven advanced HCC and Child-Pugh (CP) up to B7 who failed or were intolerant to prior systemic therapy, were 2:1 randomized to A 18 mg/m2vs P IM injection weekly. The study was stratified by the combination of geographic region (Asia vs Other) and prior treatment status (sorafenib (S) vs other (O) failure). The primary endpoint was overall survival (OS) with secondary endpoints progression-free survival (PFS), safety and Ar and Ci levels correlatives. The study was planned for 422 A and 211 P, with 93% power to detect a 1.6 months (m) increase in median OS (4 to 5.6 m; 1-sided a = 0.025). Results: 635 pts were enrolled: median age 61, 82% male, 60% Asian, 52% hepatitis B, 26% hepatitis C, 13% alcohol and 7% NASH, 76% stage IV and 30% vascular invasion, CP A 90.6%, 70% progressed on S, 11.2% on O and 18.8% did not tolerate either. The mean number of doses was 14 for A and 17 for P. Median OS was 7.8 m for A vs 7.4 for P (p = 0.884, HR=1.022 (95% CI: 0.847, 1.233)) and median PFS 2.6 vs 2.6 (p = 0.075, HR=1.175 (95% CI: 0.964, 1.432)). The most common grade 3 and above AEs in both groups were fatigue (A: 1.9% and P: 3.3%), and decreased appetite (A: 1.9%, and P: 1.4%). Hypersensitivity reactions including anaphylaxis occurred in 2.1% pts on A. There were 15% deaths on A arm within 30 days of last dose vs 10.4% on P (progressive disease 83%, GI hemorrhage 8%, unknown/other 9%) Pts with Ar depletion for .8 weeks had a median OS of 12.3 m compared to 7.3 m (P = 0.0032) for # 4 weeks. Similarly, pts with Ci increase for .8 weeks had a median OS of 11.6 m, compared to 3.5 m (P,0.0001) for #4 weeks. Conclusions: A did not demonstrate an OS benefit in the population of advanced HCC with failed prior systemic therapy. A was well tolerated. An OS benefit associated with prolonged Ar depletion was noted. Strategies to enhance prolonged arginine depletion and/or synergize the effect of A are underway Clinical trial information: NCT01287585.

4019

Poster Discussion Session; Displayed in Poster Session (Board #11), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 3:00 PM-4:15 PM

4018

Poster Discussion Session; Displayed in Poster Session (Board #10), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 3:00 PM-4:15 PM

TACE 2: A randomized placebo-controlled, double-blinded, phase III trial evaluating sorafenib in combination with transarterial chemoembolisation (TACE) in patients with unresectable hepatocellular carcinoma (HCC)— Background. First Author: Tim Meyer, UCL Cancer Institute, University College London, London, United Kingdom Background: TACE is regarded as the standard of care for patients (pts) with intermediate stage HCC while sorafenib (S) is the current standard for advanced disease. TACE 2 was designed to determine whether TACE + S improves progression free survival (PFS) compared to TACE alone. Methods: Pts with HCC were randomised 1:1 to continuous S (400mg BD) or placebo (P). Inclusion criteria included unresectable, liver confined HCC, patent main portal vein, ECOG PS # 1 and Child Pugh A liver score. Study drug was commenced at randomisation and TACE performed at 2-5 weeks using drug eluting beads (DEB) loaded with 150mg doxorubicin. Further TACE was performed according to radiological response and patient tolerance. Primary outcome measure was PFS. Secondary outcome measures included overall survival (OS) and toxicity. Target recruitment was 412 to detect a (Hazard Ratio) HR of 0.72 with 2-sided significance a = 0.05 and 85% power. Treatment HR in the intention to treat population was estimated by Cox survival model. A planned interim futility analysis was performed at 45% of trial events. Results: The interim futility analysis, performed July 2015, led to trial termination in December. At the interim analysis, 294 were randomised to S (n = 147) or P (n = 147) from 20 UK sites. Median age was 67 yrs and 169 (58%) were PS 0. In 229 (79%) cirrhotic pts, liver disease was: 54 (24%) HCV; 30 (13%) HBV; 99 (43%) alcohol; 86 (38%) other. Median follow-up was 15.2 months. Median PFS for the S and P group was 7.8 (95% CI 5.9, 10.0) and 7.7 (95% CI 5.9, 10.5) months; (HR) of 1.03 (95% CI 0.75, 1.42 p = 0.85). For the S and P groups: median OS was 18.8 (95% CI 12.3, 24.0) and 19.6 (95% CI 14.8, 24.0) months; there were 77 and 78 SAEs; 195 and 256 TACE procedures. Median duration of S and P was 5.9 and 7.7 months; median of patient average daily dose was 649mg and 800mg. Conclusions: The TACE 2 trial provides no evidence that addition of S to DEB TACE improves PFS or OS in European pts with intermediate HCC. Alternative systemic therapies need to be evaluated in combination with TACE to improve outcome for this patient population. Clinical trial information: ISRCTN93375053.

4020

Poster Discussion Session; Displayed in Poster Session (Board #12), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 3:00 PM-4:15 PM

A phase II, double-blind study of galunisertib+gemcitabine (GG) vs gemcitabine+placebo (GP) in patients (pts) with unresectable pancreatic cancer (PC). First Author: Davide Melisi, Digestive Molecular Clinical Oncology, University of Verona, Verona, Italy

Genomic profiling to distinguish poorly differentiated neuroendocrine carcinomas arising in different sites. First Author: Emily K. Bergsland, UC San Francisco, Helen Diller Family Comprehensive Cancer Center, San Francisco, CA

Background: In vivo preclinical models, transforming growth factor-beta (TGFb) inhibitors enhanced gemcitabine activity. In this study, patients (pts) with Stage II to IV unresectable PC were treated with GG or GP to evaluate their efficacy and safety. Methods: Pts were randomized 2:1 (stratified for ECOG PS, disease stage and previous gemcitabine use) to GG or GP. Oral galunisertib 150 mg BID was given as intermittent dosing (14 days on/14 days off per cycle); gemcitabine was administered per label. The primary endpoint was overall survival (OS). Using a Bayesian augmented control (BAC) design assuming a hazard ratio (HR) of 0.7 for OS, the study had 90% power to identify if the probability of HR , 1 was . 0.85. Secondary endpoints included progression-free survival (PFS), overall response rate (ORR), biomarkers and safety (CTCAE v4). Results: 156 pts were randomized to GG (N = 104) or GP (N = 52). Baseline characteristics were balanced between groups. HR based on BAC was 0.8 with 95% credible interval (0.60, 1.09) and PR (HR , 1) = 0.92, indicating a positive study. Efficacy data are summarized in the table. For the subgroup of pts with baseline TGFb1 levels # 4224 pg/mL (n = 117), median OS was 10.9 vs 7.2 months (GG vs GP) with unadjusted HR (95% CI) = 0.68 (0.44, 1.04), logrank p = 0.076). The most frequent CTC grade 3/4 adverse events possibly related to study treatment (GG vs GP) were anemia (7.8% vs 13.5%), neutropenia (32.0% vs 26.9%) and thrombocytopenia (7.8% vs 9.6%). 46% vs 43% of pts had . 50% decrease in CA19-9 and 43% vs 41% had . 50% decrease in TGFb1 (GG vs GP). For both biomarkers a reduction in concentration correlates with improved OS. Conclusions: GG resulted in improvement of OS and PFS in pts with PC, with a manageable toxicity profile as compared to GP. TGFb1 and CA19-9 reduction correlated with OS. Pts with lower levels of TGFb1 may have greater benefit from treatment with galunisertib. Clinical trial information: NCT01373164.

Background: Extrapulmonary poorly differentiated neuroendocrine carcinomas (EPNEC) occur in nearly any site. Traditionally treated like small cell lung carcinoma (SCLC), optimal therapy has not yet been defined. We examined genomic alterations in EP-NEC of different sites in comparison to SCLC and to each other. Methods: Genomic profiling was performed on 867 NEC, including 593 SCLC and 274 EP-NEC (pathologist-confirmed poorly differentiated morphology) arising in the gastrointestinal (GI) tract and pancreas: 123 pancreas, 92 colon, and 59 “other GI” (esophageal, stomach, small intestine). Hybridization-captured, adaptor ligation based libraries were used to a mean coverage depth of 600X for 192 cancer-related genes. All classes of genomic alterations were identified (copy number losses, amplifications, rearrangements, and short variant mutations). Results: There were 9 genes with alterations in . 15% of tumors in any group (Table 1). Only TP53 crossed the 15% threshold in every group; MEN1 and DAXX . 15% specific to pancreas, APC and KRAS . 15% specific to colon, and CCNE1 . 15% specific to “other GI”. Every EP-NEC group had a lower rate of alteration for TP53 and RB1than SCLC. Conclusions: In this dataset, molecular alterations in GI tract and pancreas EP-NEC differ from SCLC and from each other. TP53 and RB1alterations are less common in EP-NEC and other genes more frequently altered than in SCLC. These findings suggest that optimal therapy for EP-NEC may be site-specific and different from SCLC. An in-depth analysis of site-specific alterations and targetable mutations will be presented. Correlation with histopathologic subtype is planned.

GG Median (95% CI) OS (months) HR (95% CI) OS Median (95% CI) PFS (months) HR (95% CI) PFS ORR (95% CI) (%)

GP

9.10 (7.43, 12.2) 7.59 (4.04, 9.92) 0.89 (0.61, 1.30) 3.65 (2.86, 5.39) 2.79 (1.81, 3.68) 0.80 (0.55, 1.15) 8.7 (4.0, 15.8) 1.9 (0.1, 10.3)

Unadjusted logrank p = 0.215 Fisher’s exact P = 0.116

Genes with alterations in . 15% of tumors in any group (% with alterations). Gene TP53 RB1 CDKN2A MEN1 CDKN2B DAXX APC KRAS CCNE1 S, P, C or O

SCLC (N = 593)

Pancreas (N = 123)

Colon (N = 92)

Other GI (N = 59)

90%P,C,O 67% P,C,O 3%P,O 1%P 1%P,O 0%P 2%C 4%C 4%O

18%S,C,O 10% S,C,O 21%S,C 33%S,C,O 16%S,C 20%S,C,O 3%C 7%C 2%O

59%S,P 34%S,P 5%P,O 3%P 1%P,O 0%P 47%S,P,O 37%S,P,O 1%O

49%S,P 29% S,P 25%S,C 2%P 19%S,C 0%P 8%C 3%C 17%S,P,C

: Statistically significant difference compared to S= SCLC, P= Pancreas, C= colon, O=

Other GI

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Noncolorectal) Cancer 4021

Poster Session (Board #13), Sat, 8:00 AM-11:30 AM

Phase II study of chemoselection with docetaxel plus 5-fluorouracil and cisplatin induction chemotherapy and subsequent conversion surgery for locally advanced unresectable esophageal cancer. First Author: Tomoya Yokota, Shizuoka Cancer Center, Sunto-Gun, Japan

4023

211s Poster Session (Board #15), Sat, 8:00 AM-11:30 AM

Metabolic characteristics of esophageal carcinoma in a veteran population. First Author: Theresa Ratajczak, Wright State University, Dayton, OH

Background: This multicenter phase II trial assessed the safety and efficacy of docetaxel plus 5-fluorouracil and cisplatin (DCF) induction chemotherapy (IC) and subsequent conversion surgery (CS) for locally advanced unresectable esophageal cancer (LAUEC). Methods: Esophageal cancer patients (pts) with clinical T4 disease and/or unresectable supraclavicular lymph node metastasis were eligible. The treatment starts with 3 cycles of DCF-IC, followed by CS if resectable, or by concurrent radiation plus chemotherapy with 5-fluorouracil and cisplatin (CF-RT) if not resectable. In case of CF-RT, the resectability is re-evaluated at 30 Gy-40Gy, followed by CS if resectable, or by additional CF-RT with total irradiation dose of 60Gy. If resectable after completion of CF-RT, CS is performed. The primary endpoint is 1-year overall survival (OS). We assumed a null hypothesis with 50% and expected 65% of 1-year OS. Results: Characteristics of 48 pts enrolled were as follows: median age, 66 years; PS0/1, 28/20; primary tumor sites, Ut/Mt/Lt, 13/33/ 2; T3/T4a/T4b, 5/1/42; N0/N1/N2/N3, 4/12/22/10; M0/M1, 39/9. Eleven pts (22.9%) developed febrile neutropenia related to DCF-IC. CS was performed in 20 pts (41.7%), including DCF-CS (n = 18), DCF-CF-RT40GyCS (n = 1), and DCF-CF-RT60Gy-CS (n = 1). R0 resection was confirmed in 19 pts (39.6%). Grade $ 3 postoperative complications includes 1 recurrent laryngeal nerve palsy, 1 lung infection, 1 wound infection, 1 pulmonary fistula, and 1 dysphagia; but no remarkable postoperative complications were observed in pts who underwent CS. The overall cure rate including R0 resection or the clinical complete response after CF-RT was 47.9%. The estimated 1-year OS was 67.7% and lower limit of 80% confidence interval was 59.5%. The OS for pts who underwent R0 resection was significantly longer than those who did not undergo R0 resection (median survival time: not reached vs. 326 days, p , 0.01). There was 1 treatment-related death in patient who received DCF-CF-RT60Gy. Conclusions: DCF-IC followed by CS showed promising signs of tolerability and efficacy in pts with LAUEC. Clinical trial information: 000011089.

Background: Esophageal cancer is projected to increase by 35% through 2025 in the U.S. We investigated whether the presence of metabolic syndrome components differed between these two cancers in a veteran population. History of gastroesophageal reflux disease (GERD) was also evaluated. Methods: Records of patients diagnosed with esophageal carcinoma between 1996-2014 were reviewed. Patients with incomplete documentation were excluded. Information on demographics, BMI before diagnosis, hypertension, hyperlipidemia, hyperglycemia, GERD, and diabetes mellitus (DM) type 2 was collected. Statistical analysis was performed using IBM SPSS Statistics 22.0 software and proper data-dependent statistical testing. Results: Of the 118 eligible patients, 81 were classified as having adenocarcinoma (EAC) and 37 as having squamous cell carcinoma (SCC). Age at diagnosis for EAC and SCC was 67.169.7 vs. 67.06 10.8 years, (p = 0.83). Compared to SCC, patients with EAC were more likely to be white than African American (98.8% vs. 59.5%, p , 0.001). Veterans with EAC had a higher mean pre-diagnosis BMI than those with SCC (29.766.5 vs. 25.068.0 kg/m2, p , 0.001). Patients with EAC were more likely to have hyperlipidemia (59% vs. 35%, p = 0.015), GERD (51% vs. 22%, p = 0.003), hyperglycemia (44.4% vs. 27.0%, p = 0.07) and type 2 DM (37% vs. 19%, p = 0.049) than those with SCC. The two groups did not differ on HTN (59.3% vs. 56.8%, p = 0.80). Pre-diagnosis BMI (OR = 1.11; 95% CI = 1.01 to 1.21) and GERD (OR = 3.60; 95% CI = 1.27 to 10.19) were independent predictors of type of cancer – i.e., significant after controlling for the other variables. Hyperlipidemia (OR = 1.94 [0.63 to 5.92]) and Type 2 DM (OR = 1.08 [0.32 to 3.67]) were not independent predictors of type of esophageal cancer. Conclusions: We found that type of cancer (EAC or SCC) was associated with presence of certain components of metabolic syndrome (i.e., pre-diagnosis BMI, GERD, hyperlipidemia, and diabetes mellitus type 2) with the former two being independent predictors of EAC when compared to SCC. With the increasing incidence of obesity, GERD, and metabolic syndrome in the Western world, it is important to evaluate the relationship of common morbidities to esophageal carcinoma, which for EAC has also risen.

4024

4025

Poster Session (Board #16), Sat, 8:00 AM-11:30 AM

Poster Session (Board #17), Sat, 8:00 AM-11:30 AM

Prediction of malignant progression of Barrett’s oesophagus: A complete systematic review and meta-analysis. First Author: Kiran Altaf, Department of Surgery, St. Helen’s and Knowsley Teaching Hospitals, NHS Foundation Hospitals, Prescot, United Kingdom

Robotic gastrectomy for gastric cancer: Subgroup analysis of a multicenter prospective comparative study of robotic versus laparoscopic gastrectomy. First Author: Joong-Min Park, Department of Surgery, Chung-Ang University College of Medicine, Seoul, Korea, The Republic of

Background: Barrett’s Oesophagus (BO) is a precursor for Oesophageal Adenocarcinoma (OAC). Currently, there are no biomarkers in clinical practice that predict the malignant progression in BO. We sought to clarify the effectiveness of common genetic aberrations as potential biomarkers through a systematic review and meta-analyses. Methods: MEDLINE, EMBASE and the Cochrane Library were searched by two independent reviewers to identify all clinical studies that assessed efficacy of p53, p16, Ki67 and DNA content abnormalities as prognostic markers in BO. The main outcome measure was development of OAC. Results: 104 clinical studies, with 12,353 samples were identified. Mutation [DOR 10.91 (2.46-48.42), sensitivity 47% (41-53%), specificity 92% (90-94%), PLR 4.71 (1.7312.78), NLR 0.65 (0.50-0.85), AUC 0.7921] and loss of p53 [DOR 16.16 (6.09-42.88), sensitivity 31% (25-38%), specificity 98% (97-98%), PLR 6.66 (4.51-9.84), NLR 0.41 (0.11-1.47), AUC 0.9228] were found to be superior to the other p53 abnormalities (Loss of heterozygosity and Overexpression). Ki-67 was also noted to have high sensitivity in identifying high risk patients [DOR 5.54 (3.40-9.05), sensitivity 82% (78-85%), specificity 48% (44-52%), PLR 1.59 (1.20-2.10), NLR 0.42 (0.34-0.51), AUC 0.7607]. Aneuploidy [DOR 12.08 (8.09-18.03), sensitivity 53% (50-57%), specificity 87% (85-88%), PLR 4.26 (2.92-6.20), NLR 0.42 (0.32-0.55), AUC 0.8461], tetraploidy [DOR 5.87(2.56-13.4), sensitivity 46% (3953%), specificity 85% (82-87%), PLR 3.47 (1.98-6.05), NLR 0.65 (0.450.94), AUC 0.7926] and loss of Y chromosome [DOR 9.23 (4.34-19.63), sensitivity 68% (56-79%), specificity 80% (73-86%), PLR 2.67 (1.295.50), NLR 0.49 (0.35-0.68), AUC 0.8073] also predicted the malignant development with respectable accuracy but p16 aberrations (Hypermethylation, Mutation and Loss) failed to demonstrate any advantage over the other biomarkers studied. Conclusions: Loss and mutation of p53 and Ki-67 effectively predict malignant progression in BO. A panel of biomarkers would be more suited to be included in surveillance programme. This will need to be confirmed in large, prospective clinical trials with costefficiency analyses.

Background: Robotic gastrectomy for gastric cancer has been proven to be a feasible and safe minimally invasive procedure. However, our previous multicenter prospective study indicated that robotic gastrectomy is not superior to laparoscopic gastrectomy. This study aimed to identify which subgroups of patients would benefit from robotic gastrectomy over conventional laparoscopic gastrectomy. Methods: A prospective multicenter comparative study was conducted. Surgical outcomes were compared between the robotic and laparoscopic groups. The analysis was conducted in each subgroup according to the patients’ obesity status, the type of gastrectomy, and the extent of lymph node dissection. Results: A total of 434 patients were enrolled for the robotic (n = 223) and laparoscopic (n = 211) groups. According to obesity and gastrectomy type, there was no difference in the estimated blood loss (EBL), number of retrieved lymph nodes (RLN), complication rate, open conversion rate, and the length of hospital stay between the robotic and laparoscopic groups. According to the extent of lymph node dissection, the robotic group showed a significantly lower EBL than the laparoscopic group after D2 dissection (98.9 ml in the robotic group and 140.5 ml in the laparoscopic group, p = 0.021), while there was no difference in EBL in patients that did not undergo D2 dissection (96.5 ml in the robotic group and 82.6 ml in the laparoscopic group, p = 0.365). Further subgroup analysis showed that non-obese patients who underwent D2 lymph node dissection benefited from less blood loss with robotic gastrectomy (83.5 ml in the robotic group and 146.4 ml in the laparoscopic group, p = 0.002). Conclusions: Non-obese patients undergoing D2 lymph node dissection can benefit from less blood loss with a robotic surgery system. Clinical trial information: NCT01309256.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

212s 4026

Gastrointestinal (Noncolorectal) Cancer Poster Session (Board #18), Sat, 8:00 AM-11:30 AM

4027

Poster Session (Board #19), Sat, 8:00 AM-11:30 AM

Validation of the JCOG prognostic index in advanced gastric cancer of the SPIRITS and G-SOX trials, using individual patient data. First Author: Daisuke Takahari, Department of Gastroenterology, Cancer Institute Hospital of Japanese Foundation for Cancer Research, Tokyo, Japan

Influence of induction chemotherapy (IC) in trimodality-eligible esophageal cancer patients: Secondary analysis of a randomized trial. First Author: Yusuke Shimodaira, The University of Texas MD Anderson Cancer Center, Houston, TX

Background: While the Royal Marsden Hospital (RMH) prognostic model had been established in Western patients with advanced gastric cancer(AGC) (Chau, JCO, 2004), we identified four risk factors; performance status (PS) $ 1, number of metastatic sites $ 2, no prior gastrectomy, and serum alkaline phosphatase (ALP) . normal range, and proposed Japan Clinical Oncology Group (JCOG) prognostic index based on the number of these factors (good risk 0,1; moderate risk 2,3; poor risk 4) in the JCOG9912 trial (Takahari, Oncologist, 2014). Methods: To validate our JCOG prognostic index, it was applied to the combined cohort of other two Japanese phase III trials for AGC; SPIRITS trial comparing between S-1 and S-1 + Cisplatin (SP) (Koizumi, Lancet Oncol., 2008) and G-SOX trial comparing between SP and S-1 + Oxaliplatin (SOX) (Yamada, Ann. Oncol., 2015). Results: 936 (94.5%) out of the totally 990 patients randomized in these trials, whose complete data were available for multivariate analyses, were included in the present study (S-1 n=150, SP n=470, SOX n=316). The median survival time (MST) for all patients was 13.2 months, and each risk factor of the JCOG index remained a significant prognostic factor (Table). Three risk groups categorized by the JCOG prognostic index identified highly significant survival differences (compared with good risk group [n=338]; HR, [95%CI]; 1.71[1.46-2.01], p,0.001 in moderate risk group [n=537] and 3.32[2.474.46], p,0.001 in poor risk group [n=61]), associated with MST of 17.2, 12.0 and 7.8 months, respectively. Good and moderate risk groups categorized by RMH index showed no significant survival differences (compared with good [n=314]; HR, [95%CI];1.10[0.94-1.28], P =0.23 in moderate [n=598] and 1.98[1.27-3.10], P =0.003 in poor [n=24]), associated with MST of 14.9, 13.0 and 6.9 months, respectively. Conclusions: JCOG prognostic index was validated and can be used for patient stratification in the future clinical trials.

Background: The primary analysis of a phase II randomized trial did not show benefit from IC for overall survival (OS) or pathologic complete response (pathCR). A secondary analysis was done to identify if any subgroups benefited from IC. Methods: 126 patients were randomly assigned to IC plus trimodality or trimodality. Multivariate recursive partitioning analysis (RPA) was applied to identify subgroups with differing OS experience. Additionally, proportional hazards regression with interactions was carried out to identify differential effect of IC. Results: The median follow-up was 6.7 years (range, 3.3 to 9.8 years), and the median OS was 3.8 (95% CI: 2.6, 5.8) years for 126 patients. OS showed univariate association with tumor length (p = 0.03), T stage (p = 0.02), N stage (p = 0.04), clinical stage (p = 0.01), and tumor grade (p , 0.001). IC did not improve OS after additional follow-up (p = 0.13) or the pathCR rate (p = 0.09). However, when tumor grade and IC were modeled together, the effect of IC differed between the two grade groups. Among patients with well/moderate tumor grade (n = 59) IC impressively prolonged OS (74% vs. 50% alive at 5 years, respectively, but not for poor grade (31% vs 28% alive at 5 years, respectively; interaction p = 0.04; IC p = 0.03). RPA confirmed this subgroup; finding 4 EC phenotypes by clinical variables for OS: (1) Very-high risk (poor grade and cN2/N3), (2) High risk (poor grade, cN0/N1, and cIII stage), (3) Moderate risk (poor grade, cN0/N1, and cI/II stage or well/moderate grade without IC), and (4) Low risk (Well/moderate grade and had IC). Patients’ 5-year OS in these subgroups were 11%, 27%, 48%, and 74%, respectively (p , 0.001). Conclusions: The secondary analysis of a randomized trial provides a compelling suggestion that IC may be highly beneficial for low-risk subgroup of EC and a prospective evaluation is warranted.

Factor PS $ 1 (vs. 0) No. of metastatic sites .2 (vs. ,2) Prior gastrectomy(-) (vs. (+)) ALP .(vs. , normal range)

4028

Hazard Ratio

95% CI

P value

1.44 1.45 1.55 1.19

1.23-1.67 1.24-1.71 1.31-1.84 1.02-1.38

,0.0001 ,0.0001 ,0.0001 0.03

Poster Session (Board #20), Sat, 8:00 AM-11:30 AM

Laparoscopy-assisted versus open D2 distal gastrectomy for advanced gastric cancer: Results from a randomized phase II multicenter clinical trial (COACT 1001). First Author: Young Woo Kim, Graduate School of Cancer Science and Policy & Research Institute & Hospital, National Cancer Center, Goyang, South Korea Background: For advanced gastric cancer (AGC), D2 gastrectomy is the standard treatment worldwide, and this procedure shows improved survival. One systematic review and several retrospective studies showed that overall survival rate and disease-free survival following laparoscopic D2 gastrectomy for AGC was not significantly different from ODG. In addition, we showed that the compliance rate of D2 lymph node dissection in LADG was not different from that in ODG in gastric cancer patients. Laparoscopic D2 gastrectomies are technically challenging, and their oncologic safety has not been proven by a prospective randomized controlled trial. This study was a multicenter, prospective, randomized phase II study to evaluate the feasibility of LADG with D2 lymph node dissection compared with ODG for AGC treatment. Methods: Patients with cT2-T4a and cN0-2 (AJCC 7thstaging system) distal gastric cancer were randomly but not blindingly assigned to LADG or ODG groups using fixed block sizes with a 1:1 allocation ratio. The primary endpoint was the noncompliance rate of the lymph node dissection, which was used to evaluate feasibility. Secondary endpoints included 3-year disease-free survival, 5-year overall survival, complications, and surgical stress response. Results: Between Jun 2010 and Oct 2011,204 patients were enrolled and underwent either LADG (n = 105) or ODG (n = 99). Of those, 196 patients (100 in LADG and 96 in ODG) were included in the intention-to-treat analysis. There were no significant differences in the overall noncompliance rate of lymph node dissection between LADG and ODG groups (47.0% and 43.2%, respectively; p = 0.648). In the subgroup analysis, the noncompliance rate in the LADG group was significantly higher than in the ODG group for clinical stage III disease (52.0% vs. 25.0%, p = 0.043). Three-year disease-free survival was not different in between the groups (LADG, 80.1%; ODG, 81.9%; p = 0.448). Postoperative complication rate and surgical stress response were not different between the groups. Conclusions: LADG was feasible for AGC treatment based on the noncompliance rate of D2 lymph node dissection. Clinical trial information: NCT01088204.

4029

Poster Session (Board #21), Sat, 8:00 AM-11:30 AM

Outcomes in older patients with localized gastric adenocarcinoma treated preoperatively. First Author: Nikolaos Charalampakis, The University of Texas MD Anderson Cancer Center, Houston, TX Background: Older patients with localized gastric adenocarcinoma (LGAC) have substantial postoperative morbidity and mortality; however, postoperative outcomes of these patients who receive preoperative chemotherapy and/or chemoradiation have not been reported. We examined the impact of age at baseline on potential predictors of postoperative outcomes. Methods: Patients with LGAC who were first treated with chemotherapy and/ or chemoradiation followed by surgery (n = 203) formed two groups: (1) $ 65 years old (n = 70) and (2) , 65 years old (n = 133). We assessed postoperative morbidity and mortality as well as overall survival (OS) and progression-free survival (PFS). Potential predictors of 90-day postoperative outcomes were identified i) by age groups and ii) other clinical covariates. Descriptive statistics and survival analyses were utilized. Results: 90-daypostoperative morbidity was similar in older and younger patients (61% vs 58%; p = 0.6549). 90-day mortality was similar (3% vs 0%; p = 0.1178). Major Clavien grade III/IV complications were similar in older patients (17% vs 12%; p = 0.3919). OS and PFS were also similar for both groups (p = 0.8629 and p = 0.558, respectively). Other factors, such as Charlson comorbidity index (p = 0.0114), length of hospital stay (p = 0.0009), median operative time (p = 0.006) and presence of diabetes (p = 0.0282) were strong predictors of postoperative complications. Conclusions: Our data suggest that there is no significant difference in postoperative morbidity and mortality between younger and older patients who have localized gastric adenocarcinoma and received preoperative treatment. Rather than age, comorbidities have more impact on postoperative complications and outcomes.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Noncolorectal) Cancer 4030

Poster Session (Board #22), Sat, 8:00 AM-11:30 AM

4031

213s Poster Session (Board #23), Sat, 8:00 AM-11:30 AM

Prognostic value of circulating tumor cells in advanced gastroesophageal adenocarcinomas in the randomized trial PRODIGE 17- MEGA (Unicancer GI-AGEO). First Author: Simon Pernot, Hopital ˆ Europeen ´ GeorgesPompidou, Paris, France

Preoperative chemoradiotherapy and the long-term run in curative treatment of locally advanced oesophagogastric junction adenocarcinoma: Update of the POET phase III study. First Author: Michael Stahl, Department of Medical Oncology, Kliniken Essen-Mitte, Essen, Germany

Background: Circulating tumor cells (CTC) with CellSearch Veridex, is a prognostic marker in breast, colon and prostate cancer approved by the Food and Drug Administration but is not validated in gastroesophageal adenocarcinomas (GEA). Objective: To evaluate the prognostic value of CTC in advanced GEA before and during treatment (Tx). Methods: 134 patients (pts) were enrolled in the biological substudy of the randomized trial PRODIGE 17 ACCORD 20 MEGA testing FOLFOX +/- panitumumab or rilotumumab in 1st-line Tx of locally advanced or metastatic GEA. Blood samples were collected on CellSave tubes on day (D)0 and D28, after 2 cycles. CTC were detected with CellSearch Veridex. Prognostic value of CTC was analyzed using the Cox model. Results: Quantification of CTC at D0 and D28 was achieved for respectively 106 and 65 pts. The median of CTC at D0 was 1 [range: 0-415], with 40 (38%), 17 (16%), 10 (9%), and 39 (37%) pts with 0, 1, 2, or . 2 CTC, respectively. A CTC level at D0 $ 2 was significantly associated with worse PFS and OS. The number of CTC decreased under Tx regardless of Tx arm (p , 0.001). The persistence of CTC $ 2 at D28 was associated with even poorer PFS and OS as compared to D0. After adjustment on tumor location, histological type (diffuse vs intestinal), age, ECOG PS and the number of metastases (1 vs . 1), CTC $ 2 at D0 remained an independent prognostic factor for PFS (HR: 1.71; 95%CI, 1.06-2.77; p = 0.03) and OS (HR: 2.03; 95%CI, 1.19-3.48; p = 0.01). No interaction was found between CTC and Tx received. Conclusions: Quantification of CTC before Tx could be a useful prognostic tool in advanced GEA. A CTC threshold $ 2 seems to have better discriminant value than the presence or absence of CTC. CTC quantitation at D28 may help to adapt Tx early in non-responding pts. Clinical trial information: NCT01443065.

Background: In 2009 we published the results of a German phase III trial comparing preoperative chemotherapy (CT) with preoperative chemoradiotherapy (CRT) in locally advanced esophago-gastric junction adenocarcinoma (EGJ AC) (JCO 27:851). Three-year survival was increased from 28% to 48% by adding radiotherapy, but results did not reach statistical significance (p = 0.07). Methods: Fifty-nine pts vs 60 pts were randomized into preoperative CT vs CRT, respectively. Median observation time reached 10.5 years. Results: The rate of R0-resection was comparable between treatment arms (68% vs 72%). With regard to OS there was a strong trend in favour of CRT: survival rate at 5 years reached 24.4% (group A) and 39.5% (group B) (HR 0.65, 95%CI 0.42-1.01; two-sided log-rank p = 0.055). Moreover, we observed a significant improvement in local PFS after complete resection (R0/R1) by CRT compared to CT (HR 0.42, 0.19-0.93; p = 0.03). Disease free survival after complete resection was also improved by adding radiotherapy to CT (HR 0.61, 0.36-1.01, p = 0.053). Overall, there was no difference in survival between pts with Siewert type I and those with type II cancer, but pts with type II cancer appear to benefit even more from radiotherapy (type I HR 0.71; type II HR 0.60). Conclusions: Our results suggest that long-term survival of pts with locally advanced EGJ AC is best achieved with preoperative CRT. POET remains the single phase III trial comparing chemo- with chemoradiotherapy followed by surgery in this patient population. We urgently need a confirmatory trial to define preoperative CRT as standard treatment.

CTC

pts (%)

PFS (months)

40 (38%) 66 (62%)

7,8 5,9

CTC 0-1 57 (54%) CTC $ 2 49 (46%)

7,6 5,6

D28 CTC CTC +

45 69%) 20 (31%)

8,3 5,1

CTC 0-1 53 (82%) CTC $ 2 12 (18%)

8,3 2,9

D0 CTC CTC +

4032

HR [IC95]

p-value

OS (months)

0,02

17,1 11,5

1,59 [1,04-2,43]

0,03

16,9 11,1

1,90 [1,08-3,36]

0,03

17,4 11,1

1,67 [1,06-2,61]

4,79 0,0001 [2,38-9,67]

17,9 5,4

HR [IC95]

p-value

1,80 [1,09-2,95]

0,02

1,85 [1,17-2,92]

0,01

1,24 [0,66-2,35]

0,51

3,13 [1,55-6,33]

0,004

Poster Session (Board #24), Sat, 8:00 AM-11:30 AM

A multicentre, open-label phase II study of irinotecan, capecitabine, and oxaliplatin (IXO) as first-line treatment in patients with metastatic gastric or gastroesophageal junction (GEJ) adenocarcinoma. First Author: Arthur Gregory Ang Lui, University of Alberta, Cross Cancer Institute, Edmonton, AB, Canada Background: Metastatic gastric and GEJ cancer is a leading cause of cancerrelated mortality. Systemic chemotherapy remains the primary treatment but optimal agents and schedule remain controversial. This study determined the safety and efficacy of first-line IXO in metastatic gastric/GEJ adenocarcinoma. Methods: The Bryant&Day two-stage design was used with dual primary endpoints of response rate (RR) and toxicity. As specified by design, after accrual of 15 patients (pts), if , 7 pts responded or $ 6 pts experienced dose limiting toxicity (DLT), accrual would be terminated concluding IXO was not active or too toxic; otherwise, second stage accrual would occur to 50 pts and if . 19 pts responded and DLT occurred in # 17 pts, IXO would be declared active and nontoxic. Eligible pts with HER2 unamplified/unknown metastatic gastric/GEJ adenocarcinoma were treated with IXO in a 21-day (D) cycle at either dose level 1 (DL1) (D1 O-100mg/m2 IV & I-160mg/m2 IV, D2-15 X-1900mg/m2/day PO) or dose level -1 (DL-1) (D1 O-85mg/m2 IV & I120mg/m2 IV, D2-15 X-1425mg/m2/day PO). Secondary endpoints were overall survival (OS), progression free survival (PFS), and health-related quality of life (HRQL). Results: After accrual of 9 pts at DL1, RR was 78% with 56% DLTs thus doses were adjusted to DL-1 without modifying sample size. Stage I accrual at DL-1 yielded a RR of 66.7% while 6.3% of pts had DLT allowing for accrual to stage 2. In total, 50 pts (48 and 49 pts evaluable for efficacy and safety, respectively) were enrolled in an intention to treat analysis and received a median of 7 cycles (range 1-21). Overall RR was 54% (26/48) with a disease control rate of 85.4%. DLTs were experienced in 22% (11/49) (DL1 = 5/9 (56%); DL-1 = 6/40 (15%)). The most frequent grade 3/ 4 AEs were diarrhea, neutropenia, fatigue, hypokalemia, and nausea. Median PFS and OS were 11.0 and 12.9 months, respectively. HRQL data is yet to be analyzed. Conclusions: IXO demonstrates impressive ORR, PFS, OS, and acceptable toxicity compared to standard triplet chemotherapy regimens. IXO should be evaluated in phase III trials in this patient population. Clinical trial information: NCT01129310.

4033

Poster Session (Board #25), Sat, 8:00 AM-11:30 AM

Results of a phase I study of GS-5745 in combination with mFOLFOX in patients with advanced unresectable gastric / GE junction tumors. First Author: Manish A. Shah, Weill Cornell Medicine, New York-Presbyterian Hospital, New York, NY Background: GS-5745 is a monoclonal antibody that inhibits matrix metalloproteinase 9 (MMP9), an extracellular enzyme involved in matrix remodeling, tumor growth, and metastasis. Preclinical studies demonstrate that MMP9 inhibition alters the tumor microenvironment, associated with greater chemotherapy penetration and improved anti-tumor immunity. Methods: We performed a phase I study of GS-5745 in combination with various chemotherapy regimens (Bendell, ASCO 2015). The expanded cohort of GC evaluated mFOLFOX with GS-5745 800 mg IV every 2 weeks. Serial blood samples were evaluated for correlative evidence of MMP-9 inhibition. Clinical Trial Information: NCT01803282. Results: As of January 12, 2016, 40 patients (pts) with metastatic GC were treated (14 remain on study). Grade 3 or higher treatment-emergent AEs (TEAEs) included neutropenia (18%), nausea (10%), and neutrophil count decreased (10%). The most common serious TEAEs were abdominal pain, GI hemorrhage, hyponatremia, nausea, pyrexia and septic shock each occurring in 2 pts (5%). Median progression free survival (PFS) for all 40 pts is 7.4 months (90% confidence interval (CI) = [5, 12]), with a median duration of response (DOR) of 9.4 months and an objective response rate (ORR) of 50%. Of those pts, 30 pts were chemotherapy na¨ıve (eg. 1stline), and demonstrated a median PFS of 12 (90% CI = [5.5, 18]) months, with a median DOR of 11 months and an ORR of 57%. Collagen neoepitope decreased with continued treatment, demonstrating on-target effects. Conclusions: Adding GS-5745 to mFOLFOX is feasible and well tolerated, and associated with encouraging activity. Biomarker studies demonstrate MMP-9 inhibition. For treatment na¨ıve pts, the median PFS was considerably longer than expected (12 months). GS5745 is now being examined in a registrational phase III study in 1st line GC with mFOLFOX chemotherapy. Clinical trial information: NCT01803282.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

214s 4034

Gastrointestinal (Noncolorectal) Cancer Poster Session (Board #26), Sat, 8:00 AM-11:30 AM

4035

Poster Session (Board #27), Sat, 8:00 AM-11:30 AM

Oncological outcomes of function preserving gastrectomy for early gastric cancer: A multicenter case-controlled analysis comparing pylorus-preserving gastrectomy versus conventional distal gastrectomy. First Author: Masaki Aizawa, Niigata Cancer Center Hospital, Niigata, Japan

Comparative molecular analyses of esophageal cancer: Adenocarcinoma vs. squamous cell carcinomas and impact on outcome. First Author: Mohamed E. Salem, Lombardi Comprehensive Cancer Center, Georgetown University, Washington, DC

Background: The aim of this study was to clarify the oncological safety of pylorus-preserving gastrectomy (PPG) compared with conventional distal gastrectomy (DG). Methods: From three institutions specializing in cancer, the medical records for a cohort of 2,898 consecutive patients who had undergone DG (n = 2,208) or PPG (n = 690) for clinical Stage I gastric cancer between January 2006 and December 2012 were analyzed. A propensity score for each patient was estimated based on 38 preoperative clinical and tumor related factors. After propensity score matching, 1,004 patients (DG = 502, PG = 502) were included in the analysis. The overall survival, relapse free survival and occurrence of secondary gastric cancer were then compared. The median (range) observation period was 48.6 (1-109.8) months. Results: The patients who received a PPG were relatively young age (58.7 6 10.4 years) and had a high performance status, middle gastric tumor, solitary tumor, clinical T1, and clinical N0. The 5-year overall survival was 98.4 % for the PPG group and 96.6% for the DG group (Hazard ratio: 0.48, 95%CI: 0.21 - 1.09, P= 0.07). The 3-year relapse-free survival was 99.5% for the PPG group and 98.0% for the DG group (HR: 0.39, 95%CI: 0.12 – 1.33, P= 0.12). Postoperative secondary gastric cancer was encountered in 8 (1.6 %) in the PPG group and 4 patients (0.8 %) in the DG group. No significant differences in either overall survival, relapse free survival or the occurrence of secondary gastric cancer were observed between the two groups. Conclusions: Given the adequate estimation of the clinical tumor stage, the oncological safety of PPG for clinical T1N0 gastric cancer in the middle stomach was comparable to that of DG.

Background: Patients (pts) with esophageal cancer (EsophCa) have a poor prognosis and limited treatment options. The effect of histological subtype on tumor molecular profile remains unknown. Here we aim to compare the molecular aberrations in esophageal adenocarcinoma (EAC) and squamous cell carcinoma (ESCC). Methods: EsophCa tumors submitted to Caris Life Sciences for IHC (protein expression), ISH (gene amplification), and NGS sequencing between 2009 and 2015 were studied and correlated with pt outcomes. Chi-square tests determined differences between histological subtypes. Kaplan-Meier methodology estimated survival. Results: A total of 966 tumors (883 EACs and 113 ESCCs) were examined. Most frequently mutated genes were TP53 (71%), BRCA2 (10%), HNF1A (9%), APC (8%), SMAD4 (6.3%), PIK3CA (4.4%), ATM (5.3%), cMET (3%), ERBB2 (2.2%), PTEN (2%), and NOTCH1 (1%). When we compared EACs with ESCCs, KRAS (6.5%), NRAS (1.4%), GNAS (1%), and BRAF (0.7%) mutations were seen only in EACs, whereas, NFE2L2 mutations (R34P and E79G), and HER-2/neu overexpression (12%) and amplification (20%), were seen only in ESCCs (p , 0.001). EACs had higher overexpression of P-glycoprotein (PGP) (51% vs. 8%, p , 0.01) compared with ESCCs. ESCCs showed higher PD-L1 expression (19% vs. 6%, p , 0.01), but there was no difference in the frequency of PD-1 expression on tumor-infiltrating lymphocytes. ESCCs also had higher overexpression of ERCC1 (55% vs. 36%, p = 0.01), MGMT (62% vs. 49%, p = 0.01), EGFR (93% vs. 75%, p = 0.003), TLE3 (70% vs. 34%, p , 0.001), RRM1 (52% vs. 36%, p = 0.006), PTEN (67% vs. 50%, p = 0.001), and TOPO1 (76% vs. 61%, p = 0.002) compared with EACs. In a small subset of pts, survival data were available, and low PGP expression was associated with prolonged survival (HR = 8.7, p = 0.004). Conclusions: Molecular profile differences between EACs and ESCCs indicate different carcinogenic pathways, and biology, that may influence response to therapy. Low frequency mutations in several druggable genes may provide therapeutic opportunities. Correlation of low PGP with prolonged survival implicates PGP as a prognostic biomarker, and highlights the importance of targeting multi-drug resistance.

4036

4037

Poster Session (Board #28), Sat, 8:00 AM-11:30 AM

Efficacy and safety of rhLTa-Da with cisplatin and fluorouracil in patients with metastatic esophageal squamous cell carcinoma: A randomized, multicenter, open-labelled, controlled, phase IIb trial. First Author: Yu-hong Li, Department of Medical Oncology, Sun Yat-sen University Cancer Center, Guangzhou, China Background: Recombinant human lymphotoxin-a derivative (rhLTa-Da) is the LTa derivative without N-terminal 27 amino acid residues. Our prior studies showed that the addition of rhLTa-Da to cisplatin-based treatment was tolerable and had an encouraging response rate in metastatic esophageal squamous cell carcinoma (mESCC). This randomized, multicenter, controlled, phase II b trial was conducted to evaluate the efficacy and safety of rhLTa-Da with cisplatin and fluorouracil (PF) as the first-line therapy for mESCC (ChiCTR-TRC-11001197). Methods: Patients from 15 centers in China were randomly assigned (1:1:1) into three arms (arm A, PF + 10 mg/ m2/d rhLTa-Da; arm B, PF + 20 mg/m2/d rhLTa-Da, arm C, PF alone). The primary endpoints included progression-free survival (PFS) and confirmed overall response rate (ORR). An explanatory analysis was performed to evaluate the role of serum tumor necrosis factor receptor-II (TNFR-II) in predicting the efficacy of rhLTa-Da. Results: Between September 2010 and May 2013, 150 patients were enrolled. Baseline patients characteristics were similar between arms. No significant difference in either PFS or ORR was observed between these arms. The median PFS was 3.9 months (95% confidence interval [CI]: 2.9-5.3 months) in arm A, 5.7 months (95% CI: 3.2-6.8 months) in arm B, and 4.9 months (95% CI: 3.1-6.1 months) in arm C (P = 0.248); the ORR was 27.1% in arm A, 46.8% in arm B, and 41.7% in arm C (P = 0.073). However, among patients with low levels of serum TNFRII, the median PFS was significantly longer in arm B than in other two arms (7.2 months [95% CI: 5.1-8.6 months] in arm B vs. 3.5 months [95% CI: 1.7-5.5 months] in arm A [P= 0.022] and 4.0 months [95% CI: 3.26.3 months] in arm C [P = 0.027]). The addition of rhLTa-Da to PFdidn’t add AEs statistically. Conclusions: Although the primary endpoint was not met, the addition of high-dose rhLTa-Da to PF regimen may prolong PFS in mESCC patients with low levels of serum TNFR-II. Further prospective study in selected patients is warranted. Clinical trial information: ChiCTR-TRC11001197.

Poster Session (Board #29), Sat, 8:00 AM-11:30 AM

Preliminary safety data from KEYNOTE-059: Pembrolizumab plus 5-fluorouracil (5-FU) and cisplatin for first-line treatment of advanced gastric cancer. First Author: Charles S. Fuchs, Dana-Farber Cancer Institute, Boston, MA Background: Standard first-line treatment for advanced gastric cancer includes combination chemotherapy with a platinum agent and a fluoropyrimidine. Pembrolizumab (pembro) has shown promising antitumor activity as monotherapy in patients with advanced gastric cancer. We report preliminary safety data for patients with advanced gastric cancer treated with pembro + cisplatin + 5-FU in the multicohort, phase 2 KEYNOTE-059 study (NCT02335411). Methods: Eligible patients were aged $ 18 y and had HER2– advanced gastric or gastroesophageal junction adenocarcinoma, ECOG PS 0-1, and no prior systemic therapy for metastatic disease. Patients received pembro 200 mg + 5-FU 800 mg/m2 (or capecitabine 1000 mg/m2 in Japan) + cisplatin 80 mg/m2 Q3W for 6 cycles followed by pembro + 5-FU (or capecitabine) for up to 2 y or until confirmed progression, intolerable toxicity, or investigator decision. Results: Of the 18 patients treated, 67% were men, and median age was 58 y. As of the Oct 9, 2015 data cutoff date, median follow-up duration was 5.5 mo (range, 4.0-7.3). There were no treatment-related deaths and only 1 patient (6%) discontinued treatment because of an AE (stomatitis), which was considered by the investigator to be unrelated to pembro or chemotherapy. 17 patients (94%) experienced treatment-related AEs of any grade, most commonly stomatitis (n = 7, 39%), decreased appetite (n = 6, 33%), nausea (n = 5, 28%), and neutropenia/ decreased neutrophils (n = 11, 61%) without neutropenic fever and unrelated to pembro. 12 patients (67%) experienced grade 3-4 treatmentrelated AEs; none were attributed to pembro. AEs attributed to pembro occurred in 7 patients (39%); the most common were diarrhea, dysgeusia, hyperthyroidism, and nausea (n = 2 each); all were grade 1/2. 8 patients (44%) experienced AEs of special interest, regardless of attribution by investigator, including hyperthyroidism, hypothyroidism, infusion-related reaction, pruritus, and vasculitis; all were grade 1/2. Conclusions: These data suggest the combination of pembro, cisplatin, and 5-FU has a manageable safety profile as first-line therapy in patients with advanced gastric cancer. Clinical trial information: NCT02335411.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Noncolorectal) Cancer 4038

Poster Session (Board #30), Sat, 8:00 AM-11:30 AM

Prognostic value of the metastatic lymph node (N) ratio in the adjuvant chemoradiotherapy in stomach tumors (ARTIST) phase III trial First Author: Soonil Lee, Department of Internal Medicine, Dankook University Hospital, Dankook University College of Medicine, Cheonan, South Korea Background: The proportion between metastatic and examined lymph nodes (N ratio) has been proposed as an independent prognostic factor in gastric cancer (GC) patients. The aim of this study is to investigate the relationship between the metastatic N ratio and the prognosis of GC after curative D2 surgery Methods: We retrospectively reviewed the data of 458 ARTIST patients who underwent D2 gastrectomy followed by adjuvant chemotherapy with capecitabine plus cisplatin (XP, n = 228) or chemoradiotherapy (XPRT, n = 230). Disease-free survival (DFS) of the patients was analyzed in order to evaluate the influence of N ratio on treatment outcome. Four N ratio categories (0% v 1-9% v 10-25% v . 25%) were employed, and statistical analysis was performed using adjusted Cox regression and stratified survival analysis. Results: Among 458 GC patients enrolled in ARTIST, the proportion of patients with at least 15 lymph nodes examined was 99%. As expected, there was a significant interaction between the N staging and N ratio. At multivariate analysis, N ratio was retained as an independent prognostic factor for DFS: HR for N ratio 0%, 1; N ratio 1-9%, 1.061; N ratio 10-25%, 1.202; and N ratio . 25%, 3.571. Furthermore, there also was a significant difference in DFS between XPRT and XP arms for patients with higher N ratio. In patients with N ratio . 25%, the 5-year DFS was 55% v 28% for XPRT and XP arms, respectively (HR, 0.527; 95% CI, 0.307 to 0.904; P = 0.020). Conclusions: In patients with curatively resected GC, N ratio is independently associated with DFS. Although this finding warrants further investigation in prospective studies, benefit with chemoradiotherapy in D2 resected GC seems to be limited to those with N ratio . 25% Clinical trial information: NCT00323830.

4040

Poster Session (Board #32), Sat, 8:00 AM-11:30 AM

Phase II study of intraperitoneal paclitaxel plus S-1/oxaliplatin for gastric cancer with peritoneal metastasis: SOX+IP PTX trial. First Author: Yoshiyuki Fujiwara, Osaka Medical Center for Cancer and Cardiovascular Disease, Osaka, Japan Background: Intraperitoneal (IP) chemotherapy is a promising treatment option for gastric cancer with peritoneal metastasis. We previously reported the safety and efficacy of IP paclitaxel (PTX) plus S-1/PTX in phase I and II studies. S-1/oxaliplatin (SOX) showed non-inferiority to S-1/cisplatin in PFS, and is considered as a new standard regimen for the first-line treatment of advanced gastric cancer on an outpatient basis in Japan. We designed a new regimen combining IP PTX with SOX in order to maximize systemic effects as well as local effects in the peritoneal cavity. The recommended dose of IP PTX was determined to be 40 mg/m2 in a phase I study. A phase II study of SOX plus IP PTX regimen was performed in gastric cancer patients with peritoneal metastasis. Methods: Gastric cancer patients with peritoneal metastasis confirmed by diagnostic imaging, laparoscopy or laparotomy were enrolled. PTX was administered intraperitoneally at 40 mg/m2 on days 1 and 8. Oxaliplatin was administered intravenously at 100 mg/m2 on day 1, and S1 was administered at 80 mg/m2/day for 14 consecutive days, repeated every 21 days. The primary endpoint was the 1-year overall survival (OS) rate. Secondary endpoints were response rate, negative conversion rate on peritoneal cytology and safety. Results: Sixty patients were enrolled and fully evaluated for OS and toxicity. The median number of courses was 10 (range 1-19). The 1-year OS rate was 71% (95% CI, 58-81%). The overall response rate was 67% (95% CI, 22-96%) in 6 patients with target lesions. Cancer cells ceased to be detected by peritoneal cytology in 35 (75%) of 47 patients. The incidences of grade 3/4 hematological and non-hematological toxicities were 55% and 44%, respectively. The frequent grade 3/4 toxicities included neutropenia (50%), leukopenia (27%), anemia (18%) febrile neutropenia (12%) and anorexia (12%). Catheter obstruction and infection of the peritoneal implantable port were observed in one patient each. There were no treatment-related deaths. Conclusions: Combination chemotherapy of SOX plus IP PTX regimen is well tolerated and active in gastric cancer patients with peritoneal metastasis. Clinical trial information: UMIN000012834.

4039

215s Poster Session (Board #31), Sat, 8:00 AM-11:30 AM

A 30 gene panel as prognostic for survival outcomes in clinically resectable gastric cancer. First Author: Matthew Porembka, UT Southwestern Medical Center, Dallas, TX Background: Current pre-treatment risk stratification modalities for gastric cancer (GC) are inadequate. Methods: Somatic mutation data from 19 cancer types sequenced by The Cancer Genome Atlas with published genegene interaction networks was integrated using NTriPath in order to identify altered pathways unique to GC. Gene expression profile of 565 treatmentnaive GC patients who underwent resection with curative intent at Yonsei University (Seoul, South Korea) from 1999-2006 were used generate molecular subtypes based on consensus clustering. A predictive risk score (RS) was developed based on high and low risk molecular subtypes. The prognostic utility of the RS was validated using the profiling data from the 300 GC patient Asian Gastric Cancer Cohort. Patients with the lowest and highest quintile risk score were defined as low and high risk, respectively. The remaining patients were grouped as an intermediate risk group. Results: Four molecular subtypes were identified. Five-year survival for the low, intermediate, and high risk groups were 72% (95% confidence interval [CI], 0.58 to 0.81), 53% (0.46 to 0.60), and 30% (0.19 to 0.42), respectively (P , 0.001). On multivariate analysis, the RS was significantly associated with survival independent of AJCC staging (HR 1.13 [1.00 to 1.27] per unit increase in RS). Conclusions: The 30 gene risk score is prognostic for GC patients and can be use at the pre-treatment setting to help guide management. Multivariate Analysis Stage I II III IV Lauren type (Diffuse VS Intestinal) Perineural invasion Lymphovascular invasion RS (per 1 unit increase)

4041

P-value

HR (95% CI)

Reference 0.16 0.04 , 0.001 0.82 0.02 0.03 0.046

– 2.01 (0.76, 5.34) 2.70 (1.03, 7.11) 7.17 (2.74, 18.75) 0.95 (0.63, 1.44) 1.57 (1.07, 2.30) 1.69 (1.04, 2.74) 1.13 (1.00, 1.27)

Poster Session (Board #33), Sat, 8:00 AM-11:30 AM

Candidate predictive biomarkers for response to olaparib in advanced gastric cancer (AGC). First Author: Darren R Hodgson, AstraZeneca, Macclesfield, United Kingdom Background: A Phase II trial in 124 patients (pts) with AGC (NCT01063517; Study 39) showed that, as 2nd-line therapy, the PARP inhibitor olaparib (Lynparza; tablets) plus paclitaxel, followed by maintenance olaparib (O/P arm), significantly improved overall survival (OS) vs paclitaxel alone (P arm) both in the overall population and in pts with low ATM protein levels (ATMlow), for whom the trial was enriched (Bang et al JCO 2015). The current study explored the relationship of other candidate biomarkers with ATM levels and clinical outcomes. Methods: Next-generation DNA sequencing (Foundation Medicine) of tumor samples from Study 39 pts was performed, followed by comprehensive genomic analysis. Candidate biomarkers included microsatellite instability (MSI)-like status (overall mutation load), ATM mutations, mutations in other genes involved in homologous recombination repair (HRRm), and TP53 and ARID1A mutations. ATM levels were determined by IHC at study entry. Associations between genetic markers were assessed by contingency table analysis, while associations between individual genetic markers and clinical outcome were assessed by Cox proportional hazards modeling. Results: 55/124 pts (ATMlow , n = 28; ATM positive [+ve], n = 27) had $ 1 sample evaluable for tumor genetics (EFTG): 6 pts (11%) had loss-of-function ATM mutations (all in ATMlow pts) and 35 pts (64%) had a TP53 mutation (TP53m). HRRm (including in ATM), ARID1A mutations and MSI-like profiles were found in 11, 19 and 7 pts, respectively, most of whom were ATMlow . OS HRs (95% CI) for O/P vs P in the EFTG population, ATMlow pts, ATM +ve pts and TP53m pts were 0.34 (0.16–0.74), 0.17 (0.05–0.65), 0.65 (0.24–1.73) and 0.515 (0.21–1.29), respectively. Pt numbers were small, but these exploratory analyses suggest that 1) ATMlow pts who receive O/P have the longest OS, regardless of TP53m status; 2) the OS HR in ATMlow pts is lower than for subsets of pts with other candidate biomarkers; and 3) an OS benefit is seen with O/P for pts without ATM mutations. Conclusions: For pts with AGC, ATM level remains the leading predictive biomarker for determining response to olaparib treatment. Clinical trial information: NCT01063517.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

216s 4042

Gastrointestinal (Noncolorectal) Cancer Poster Session (Board #34), Sat, 8:00 AM-11:30 AM

An update on the randomized phase III POST trial: S-1 based doublet as an adjuvant chemotherapy for curatively resected stage III gastric cancer. First Author: Choong-kun Lee, Yonsei Cancer Center, Division of Medical Oncology, Department of Internal Medicine, Yonsei University College of Medicine, Seoul, South Korea

4043

Poster Session (Board #35), Sat, 8:00 AM-11:30 AM

Randomized phase 2 study of paclitaxel (PTX), trastuzumab (T) with or without MM-111 in HER2 expressing gastroesophageal cancers (GEC). First Author: Crystal Shereen Denlinger, Fox Chase Cancer Center, Philadelphia, PA

Background: We report an updated results from a randomized, multicenter, phase III trial to compare S-1 plus docetaxel (DS) with S-1 plus cisplatin (SP) as an adjuvant chemotherapy for stage III gastric cancer (GC) patients. (ClinicalTrials.gov: NCT01283217). Methods: Stage III GC patients received curative gastrectomy with D2 lymphadenectomy were randomized into equal groups to receive adjuvant chemotherapy of 8 cycles of DS (S-1 70mg/m2/ day on days 1-14, docetaxel 35mg/m2 on days 1 and 8) or SP (S-1 70mg/m2/ day on days 1-14, cisplatin 60mg/m2 on day 1) every 3 weeks. The primary endpoint was 3-year disease-free survival (DFS) rate. Results: Between November 2010 and July 2013, 153 patients (75 patients to DS and 78 patients to SP) were enrolled from 8 institutions in Korea. After the capecitabine plus oxaliplatin was approved based on the CLASSIC study, it was decided to close the study early. With a median follow up duration of 37 (range, 1–57) months, the 3-year DFS rate was 49.3% in DS group (median DFS: 27.7 months) and 56.4% in SP group (median DFS: 44.0 months). Both group did not reach median OS with 29 (38.7%) and 27 (34.6%) deaths in DS and SP groups, respectively. There was no statistically significant difference in DFS (Hazard Ratio [HR], 1.251; 95% confidence interval [CI], 0.787–1.989; P = 0.343) or OS (HR, 1.200; 95% CI, 0.709–2.030; P = 0.497) between two groups. The most common grade 3 or 4 adverse event was neutropenia (42.7% in DS and 38.5% in SP, P= 0.351). SP group had more grade 3 or 4 anemia (1.3% vs. 11.5%, P= 0.037), whereas grade 3 or 4 hand-foot syndrome (4.1% vs. 0%, P= 0.025) and mucositis (10.7% vs. 2.6%, P= 0.001) were more common in DS group. Fifty-one patients (68%) in DS group and 52 (66.7%) in SP groups finished planned treatment. The median relative dose intensity (RDI) was 0.86 for both docetaxel and S-1 in DS group, while 0.83 for cisplatin and 0.86 for S-1 in SP group. Conclusions: Our results suggest that S-1 plus cisplatin or docetaxel is an effective and tolerable option for patients with curatively resected stage III gastric cancer. Clinical trial information: NCT01283217.

Background: HER2 overexpression occurs in , 20% of GEC, and T-based front-line therapy improves survival. HER3 is overexpressed in , 87% of GEC and is associated with poor prognosis. HER3 is activated by its ligand heregulin (HRG) to form a potent signaling heterodimer with HER2 and is emerging as a key tumorigenic node. MM-111, a bispecific antibody designed to inhibit HRG-activated HER3 signaling in HER2+ tumors, was evaluated in second-line treatment of advanced HER2+ GEC in combination with weekly PTX and TRAS. Methods: Patients (pts) who progressed following fluoropyrimidine/platinum-based therapy +/- T for advanced disease or within 6 months of neoadjuvant/adjuvant therapy were randomized 1:1 to 80 mg/m2 PTX on days 1, 8 and 15 of a 28 day cycle; T at 4 mg/kg load/2 mg/ kg/weekly; with (Arm A) or without (Arm B) MM-111 at 20 mg/kg/weekly. Primary endpoints were progression-free survival (PFS) in the intent-to-treat (ITT) population and HRG-high population (pts with HRG $ median). Enrollment of 120 pts was planned with 70 or 38 PFS events required to detect a 67% or 50% PFS improvement in the ITT- or HRG-high population, respectively (80% power, alpha = 0.1). Results: The study was closed early by the Data Monitoring Committee based on lack of efficacy within the ITT. 88 pts were randomized; 71 pts received prior T; 54 pts had HER2+ status centrally confirmed; 57 pts had assessable HRG mRNA. Median PFS in the ITT was 9.6 weeks (95% CI 8.0-15.7) for Arm A and 23.3 weeks (95% CI 12.1-27.0) for Arm B, HR 2.08 (p = 0.01; 95% CI 1.17-3.68). Median OS in ITT was 32.1 weeks (95% CI 20.4-48.0) and 56.1 weeks (95% CI 40.7-NR) for Arm A and Arm B respectively, HR 2.12 (p = 0.045, 95% CI 1.0-4.5). Any grade treatment emergent adverse events with . 15% incidence across both arms were diarrhea, anemia, decreased appetite, alopecia, fatigue, nausea, vomiting, asthenia, neutropenia, constipation, and cough. Conclusions: MM-111 did not improve PFS or OS in HER2+ GEC when added to PTX/T. HRG expression was lower than anticipated in the study population, potentially contributing to the effect of MM-111 on PFS and OS in the ITT. PTX/T alone appears promising in second-line therapy of GEC and further investigation is warranted. Clinical trial information: NCT01774851.

4044

4045

Poster Session (Board #36), Sat, 8:00 AM-11:30 AM

Poster Session (Board #37), Sat, 8:00 AM-11:30 AM

Outcomes with adjuvant chemo-radiation (ACRT) in patients (pts) with localized gastric cancer (GC): Analysis of National Cancer Data Base (NCDB). First Author: Afsaneh Barzi, University of Southern California Norris Comprehensive Cancer Center, Los Angeles, CA

Effect of definitive concurrent therapy with or without surgery on resectable esophageal cancer survival: An analysis of the National Cancer Data Base. First Author: Kushal Naik, Rollins School of Public Health, Emory University, Atlanta, GA

Background: The choice of adjunctive therapy (AT) in localized GC varies by tumor location (TL): cardia (cGC) and non-cardia (nGC). We assessed the effectiveness of 3 commonly used ATs: ACRT, Neoadjuvant Chemo-radiation (NACRT), and Perioperative Chemotherapy (PCh). Methods: 3 cohorts were defined corresponding to AT modality. Pts with GC and known TL who underwent definitive surgery and AT were eligible. 21477 pts were included in the AT cohorts. Overall survival (OS) was used as primary outcome. TL, TNM Stage, demographics, and facility type (FT), were included in the Cox model. Results: Median OS for ACRT was 42.1 months (95% CI 40.2-44.1) which was significantly higher vs. NACRT 35.9 (34.2-37.9: HR 1.17) and PCh 34.2 (32.7-35.8: HR 1.12) p , .0001. Stratifying cohorts for cGC, ACRT remained superior with median OS of 36.4 (34.5-39.0) vs. NACRT 34.0 (32.5-35.6: HR 1.13) and PCh 33.3 (31.2-35.6: HR 1.11) p , .0001. These number for nGC were 48.3 (45.6-52.0), 35.1 (31.0-40.9: HR 1.18), and 35.9 (33.5-38.3: HR 1.20) p , .0001. PCh downstaged average tumor size of 4.5 to 4cm in cGC (p , 0.001) but this effect was not present in nGC. In 10796 pts with cGC treatment distribution was 24%, 55%, and 21% for ACRT, NACRT, and PCh. These numbers for nGC were 10681, 55%, 7%, and 38%. Conclusions: Sequence and type of AT makes a significant difference in the outcome of GC. Within the limitations of a registry analysis, ACRT appears to be the best AT for both cGC and nGC. The distribution of patients among AT modalities reflects a difference based on cGC and nGC. This choice is not supported by these results and should be re-evaluated.

Background: Locally advanced resectable esophageal cancers (rEC) are managed either with concurrent chemo-radiation followed by surgery (CRSx) or concurrent chemo-radiation alone (cCR). There is insufficient evidence comparing the overall survival (OS) of these two groups in a large population. Methods: The National Cancer Data Base (NCDB) was queried for rEC cases diagnosed from 2003 to 2011. Patients with previous cancers, cervical rEC, clinical stage T1N0 or metastatic disease were excluded. Only adeno-, squamous or mixed adeno-squamous carcinomas treated with cCR or CRSx were included. cCR was defined as chemotherapy and radiotherapy given within 30 days of each other. CRSx was defined as cCR followed by surgical resection within 90 days of initiation of cCR. The OS in the two groups was compared using Kaplan-Meier methods and extended Cox-proportional hazard models. Results: A total of 14,495 eligible patients were identified; of those, 10,645 (73.4%) received cCR and the rest were treated with CRSx. Univariate survival analysis showed that survival following CRSx was significantly better than that after cCR with a mortality hazard ratio (HR) of 0.48 and a 95% confidence interval (CI) from 0.45 to 0.50 (p , 0.001). The odds of receiving CRSx was higher among patients with having stage III disease, private insurance, living more than 10 miles away or in areas with higher median income or greater proportion of high school-educated residents. Patients over 70 years of age, males, African-Americans, with comorbidities, squamous cell EC, lesions of upper third of esophagus, or those treated at community programs were most likely to receive cCR alone. After adjusting for age, sex, race, insurance, area-based measures of income and education, treatment facility type, co-morbidities, primary site, grade, histology and stage, the HR for CRSx compared to cCR alone was 0.66 (95% CI: 0.45-0.96, p = 0.03) and the median survival with CRSx was longer (14.3 vs. 34.7 months). Conclusions: Data from the NCDB strongly support the inclusion of surgery after concurrent chemoradiotherapy for patients with locally advanced, resectable esophageal cancer.

Summary of results. N (%) (cGC, nGC) Age Male/Female (%) Race: White/Black/East-Asia/Hispanic/others (%) Differentiation: well-moderate/poor/unknown (%) FT: Academic/non-academic Insurance: Private/Medicare/Medicaid/Others (%) Pathological Stage 0, I, II, III, IV (%)

ACRT

NACRT

PCh

8517 (40) (2618, 5899) 63 66/34 58/18/7/3/12/1/1 70/26/4 35/64 43/41/9/7 0/13/32/48/7

6630 (31) (5892, 738) 62 82/18 87/5/1/1/4/1/1 51/36/13 50/50 51/38/5/6 3/23/34/37/2

6330 (29) (2286, 4044) 64 66/34 63/15/5/3/12/1/1 67/27/6 48/52 43/43/7/7 1/17/30/43/8

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Noncolorectal) Cancer 4046

Poster Session (Board #38), Sat, 8:00 AM-11:30 AM

Updated results for the advanced esophageal carcinoma cohort of the phase 1b KEYNOTE-028 study of pembrolizumab. First Author: Toshihiko Doi, National Cancer Center Hospital East, Chiba, Japan Background: PD-L1 expression is associated with poor prognosis in esophageal cancer. Pembrolizumab (pembro) is a monoclonal antibody against PD-1 that helps to restore antitumor immune surveillance. The multicohort, phase 1b KEYNOTE-028 (NCT02054806) trial was designed to evaluate the safety and efficacy of pembro in patients with PD-L1+ advanced solid tumors. Updated results from the esophageal carcinoma cohort of KEYNOTE-028 are reported. Methods: Key eligibility criteria included advanced PD-L1+ squamous cell carcinoma or adenocarcinoma of the esophagus or gastroesophageal junction, failure of standard therapy, and ECOG PS 0-1. Patients received pembro 10 mg/kg every 2 wk for up to 2 y or until confirmed progression, intolerable toxicity, or investigator decision. Response was assessed every 8 wk for the first 6 mo and every 12 wk thereafter. Primary end point was ORR per RECIST v1.1 by investigator review. Results: In the 23 patients enrolled, median age was 65 y, 83% were men, 74% had squamous histology, and 87% received $ 2 prior therapies for metastatic disease. As of the Nov 4, 2015, data cutoff date, median follow-up duration was 7.1 mo (range, 1.3-19.4 mo). 9 patients (39%) had treatment-related adverse events (AEs), most commonly decreased appetite (n = 3, 13%). 4 (17%) patients experienced grade 3 treatment-related AEs (decreased lymphocytes in 2 patients and decreased appetite, liver disorder, and pruritic rash in 1 patient each). There were no treatment-related deaths or discontinuations. Immune-mediated AEs included grade 2 hypothyroidism in 2 patients (9%) and adrenal insufficiency and pruritic rash in 1 patient (4%) each. ORR was 30% (95% CI, 13%-53%), and the stable disease rate was 9% (95% CI, 1%-28%). 4 of 7 responses were ongoing at data cutoff, with a median duration of response not reached (range, 5.5-11.8 + mo). Conclusions: After additional follow-up, pembro continued to show manageable toxicity and durable antitumor activity in patients with heavily pretreated, PD-L1+ advanced esophageal carcinoma. Phase 2 and 3 trials of pembro in patients with esophageal carcinoma are ongoing. Clinical trial information: NCT02054806.

4048

Poster Session (Board #40), Sat, 8:00 AM-11:30 AM

4047

217s Poster Session (Board #39), Sat, 8:00 AM-11:30 AM

Efficacy of preoperative chemotherapy regimens in patients with initially unresectable locally advanced gastric adenocarcinoma: Capecitabine and oxaliplatin (XELOX) or with epirubicin (EOX). First Author: Yan Wang, Fudan University ZhongShan Hospital, 180 Fenglin Road, Shanghai, Shanghai, China Background: Preoperative chemotherapy for initially unresectable locally advanced gastric cancer is thought to downsize tumors for curative resection and there upon prolong survival, but there is no best recommended regimen. We assessed the effectiveness of EOX (capecitabine, oxaliplatin and epirubicin) compared with XELOX (capecitabine and oxaliplatin) as preoperative chemotherapy for initially unresectable locally advanced gastric cancer. Methods: This is a prospective observational study. The trial was registered on ClinicalTrials.gov (NCT02192983). Patients with unresectable locally advanced gastric cancer were performed EOX regimen or XELOX regimen at the discretion of the investigators. They were assessed for response every 2 cycles by CT (computed tomography) scan. A multidisciplinary team reassessed resectability after 4 cycles. If disease became resectable, radical surgery would be performed within 4-6 weeks after the last treatment cycle. The primary endpoint was the response rate. Secondary end points included the R0 resection rate, survival and adverse events. Results: From November 2008 to May 2015, 242 patients were enrolled; 112 of them were assigned to EOX regimen and 130 to XELOX regimen. The response rates were 33.0% and 33.8% respectively in EOX group and XELOX group (P= 0.997). After 4 cycles of chemotherapy, 63 patients (56.3%) in EOX group and 81 patients (62.3%) in XELOX group received radical operation (P= 0.408). There was no significant difference in progress-free survival (PFS, 12.0m vs. 15.4m, P= 0.925) and overall survival (OS, 25.7m vs. 29.0m, P= 0.783) in two groups. In addition, more adverse effects occurred in EOX group, such as more leucopenia (22.3% vs. 10.0%, P= 0.014), neutropenia (23.2% vs. 11.5%, P= 0.025), fatigue (11.6% vs. 3.8%, P= 0.041) and vomiting (10.7% vs. 2.3%, P= 0.015). Conclusions: For unresectable locally advanced gastric cancer patients, XELOX regimen showed similar effects in response rate, radical resection rate and survival benefits, but with less toxicity effects.

4049

Poster Session (Board #41), Sat, 8:00 AM-11:30 AM

Phase III randomized study of elective nodal irradiation plus erlotinib combined with chemotherapy for esophageal squamous cell carcinoma. First Author: Shi Xiu Wu, Hangzhou Cancer Hospital, Hangzhou Zhejiang, China

Phase II clinical trial of S-1 plus nanoparticle albumin-bound paclitaxel as first-line chemotherapy for patients with metastatic gastric cancer. First Author: Rui-hua Xu, Sun Yat-sen University Cancer Center, Guangzhou, China

Background: The addition of anti-EGFR therapy to chemoradiotherpay in the treatment of unresectable esophageal squamous cell carcinoma (ESCC) is not well defined, as well as the role of elective nodal irradiation (ENI). This phase III trial was designed to determine whether the addition of erlotinib and/or ENI to concurrent conventional chemoradiotherapy containing cisplatin/paclitaxel (CP) improved survival. Methods: This study was a 2x2 factorial design, with erlotinib and ENI as factors. We randomized Chinese patients with ESCC to: (A) ENI/CP/erlotinib; (B) ENI/CP; (C) conventionalfield irradiation (CFI)/CP/erlotinib; (D) CFI/CP. The primary endpoint was overall survival (OS); secondary endpoints included progression-free survival, local-regional failure rate, and toxicity. Three hundred and forty-four patients and 242 events were needed for 85% power to detect a hazard ratio (HR) of 0.76 (2-year OS rate from 35% to 45% for each factor) if no interaction between the two factors was observed. Results: From December 2007 to December 2014, 352 patients were recruited. Median follow-up was 16.5 months for all patients at this analysis. There were no significant interactions between the two factors of ENI and erlotinib (P = 0.999). Twoyear OS rate and median OS was 54% and 35.4 months for ENI (A+B) versus 42% and 20.5 months for CFI (C+D; HR = 0.69; 95% CI, 0.52 to 0.91; P = 0.008), respectively. Two-year OS rate and median OS was 52% and 24.9 months for erlotinib (A+C) versus 44% and 20.9 months for control (B +D; HR = 0.77; 95% CI, 0.57 to 1.00; P = 0.049), respectively. Patients treated with ENI plus erlotinib (A) achieved a 2-year OS rate of 58% and a median OS of 47.2 months. Hematologic toxicities grade 3 or above were the most common adverse events. Erlotinib was associated with a higher rate of all-grade rash events. Conclusions: Compared with conventional chemoradiotherapy, the addition of either ENI or erlotinib significantly improved survival among Chinese patients with ESCC, with the best of benefits in those treated with ENI plus erlotinib. The incidence rates of adverse-event were acceptable in this study. Clinical trial information: NCT00686114.

Background: Little was known about nanoparticle albumin-bound paclitaxel (Nab-PTX) in gastric cancer. As the first phase II trial, it aimed to evaluate the efficacy and safety of S-1 plus Nab-PTX as first-line treatment for patients with metastatic gastric cancer. Methods: A total of 73 gastric cancer patients with metastatic and measurable lesions (M/F 45/28, median age 53, ECOG 0-2) were enrolled in first-line setting from Sun Yat-sen University Cancer Center between 2011 and 2015. They were orally treated with S-1 in doses of 40 mg (BSA , 1.25 m2), 50 mg (1.25 # BSA , 1.50 m2) and 60 mg (BSA $ 1.50 m2) b.i.d. on days 1-14 in combination with Nab-PTX (200240 mg/m2, divided on days 1 and 8, intravenously during 30 minutes) of each 21-day cycle. Treatment was planned for 4-6 cycles, or until progression, unacceptable toxicity, and discontinuation. Primary endpoint was progression-free survival (PFS) and secondary endpoints were overall response rate (ORR), overall survival (OS), and adverse events (AEs). Results: Of the 73 patients, 4 (5.5%) had complete responses, 35 (47.9%) had partial responses (PRs), 25 (34.2%) had stable diseases, 4 (5.5%) progressed and 5 (6.8%) were not evaluable, with median number of cycles as 4 (range 1-6). The ORR and DCR were 53.4% and 87.7%, accordingly. The median PFS was 8.77 months (95% confidence interval (CI) 6.88–10.66). The median OS was 14.70 months (95% CI 9.57–19.83). Most AEs were mild without treatment-related death. Grade 3 to 4 AEs occurred in 22 patients (30.1%) and grade 4 AE (neutropenia) occurred in only one of them. Grade 3 AEs included leukopenia (13.7%), neutropenia (12.3%), anemia (5.5%), thrombocytopenia (1.4%), diarrhea (6.8%), vomiting (2.7%), stomatitis (1.4%), peripheral neuropathy (1.4%), and hand-foot syndrome (1.4%). After 2 to 6 cycles, 7 patients (6 lymph node, 1 liver) achieved good PRs and received gastrectomy plus metastasectomy. 30 (41.1%) patients had S-1 monotherapy as maintenance with median number of cycles as 4 (range 1–20). Conclusions: S-1 plus Nab-PTX is an encouraging option in first-line treatment for patients with metastatic gastric cancer, with promising efficacy, safety, and convenience. Clinical trial information: NCT02229058.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

218s 4050

Gastrointestinal (Noncolorectal) Cancer Poster Session (Board #42), Sat, 8:00 AM-11:30 AM

Prospective randomized phase II study of concurrent chemoradiotherapy versus chemotherapy alone in stage IV esophageal squamous cell carcinoma. First Author: Tao Li, Department of Radiation Oncology, Sichuan Cancer Hospital & Institute, Chengdu, China Background: Stage IV ESCC carries a poor prognosis with a median survivals of 6-9 months. The standard treatment has traditionally been chemotherapy. Palliative radiotherapy was used for symptom relief. The optimal treatment for stage IV ESCC has not yet been established. The aim of this study was to compare the efficacy and safety of CCRT versus chemotherapy alone in patients with stage IV ESCC. Methods: Stage IV ESCC with a performance status of 0-1 were randomly assigned to the CCRT group and the chemotherapy group. Both groups of patients received at least 2 cycles of chemotherapy with cisplatin and docetaxel every 3 weeks. Patients in CCRT group received 50-60 Gy/25-30 fractions of concurrent radiotherapy to the esophageal primary tumor. The primary end point was overall survival (OS). The secondary end points were progression-free survival (PFS), object response rate (ORR) of primary tumor and toxicity. Results: Between Aug. 2013 and Oct. 2015, 60 patients were enrolled and divided into the CCRT group (n = 30) and the chemotherapy group (n = 30). The baseline clinical characteristics of the 2 groups were similar. Patients in the CCRT group received a mean 54.7 Gy of radiotherapy and a mean 3.6 cycles of chemotherapy, whereas patients in the chemotherapy group received a mean 3.8 cycles. The ORR of the primary tumor was higher in the CCRT group than in the chemotherapy group (83.3% vs. 46.7%, p = 0.001). At a median follow-up of 16 months, median PFS (9.3 vs. 4.7 months, p = 0.021) and median OS (18.3 vs. 10.2 months, p = 0.001) were significantly longer in the CCRT than that in the chemotherapy group. Overall survival rates at 1, 2 years were 73.3%, 43.3% in the CCRT group and 46.6%, 26.7% in chemotherapy group (p = 0.030) Although $ grade 3 neutropenia was significantly more frequent in the CCRT group than that in the chemotherapy group (33.3% vs. 20.0 %, p , 0.05), the rates of other toxicities did not differ. Conclusions: CCRT was well tolerated and was associated with longer PFS and OS than chemotherapy alone in patients with stage IV ESCC. Controlled randomized, multi-center trials are required to determine whether CCRT is a primary treatment option for patients with stage IV ESCC.

4052

Poster Session (Board #44), Sat, 8:00 AM-11:30 AM

PD-L1 expression in Epstein-Barr virus-infected gastric cancers. First Author: Sarah Derks, dana-farber cancer institute, Boston, MA Background: Gastric cancer (GC) is a deadly disease with limited treatment options. Recent studies with PD-1 inhibitor Pembrolizumab have shown promising results in GC. Key questions remain which GC subclass may respond best. In a comprehensive molecular analysis of gastric adenocarcinomas as part of The Cancer Genome Atlas (TCGA) we identified that 15% of Epstein Barr Virus (EBV)+ GCs have amplification of chromosomal region 9p24.1, the locus of PD-1 ligands PD-L1 and PD-L2. As EBV+ cancers have multiple mechanisms to induce PD-L1 expression, we hypothesize that PDL1 status may differ specifically in EBV+ GCs, regardless of 9p24.1 status, indicating a broader potential role for PD-1 blockade in this GC subtype. Methods: PD-L1 expression was analyzed with immunohistochemistry (IHC) using whole tissues slides of EBV+ (n = 32) and EBV negative (n = 49) GCs. The transcriptional landscape of EBV+ and EBV negative GCs was determined using mRNA sequencing results from the TCGA study. Results: With IHC we showed that EBV+ GCs have PD-L1 expression in tumor and immune cells in 50% (16/32) and 94% (30/32) of GCs respectively. This in contrast to EBV negative GCs that have PD-L1+ tumor and immune cells in only 10% (5/49) and 39% (19/49) of cases respectively (P, 0.001). Furthermore, PD-L1+ immune cells infiltrated the center of the majority of EBV+ GCs, while in EBV negative GCs PD-L1+ immune cells remained at the invasive margin (P = 0.007). mRNA sequencing analyses of 12 EBV+ GCs and 10 EBV- GCs showed that PD-L1+/EBV+ GCs have enrichment of an interferon-g driven gene signature. Single-sample gene set enrichment analyses on a bigger series of 214 individual GCs demonstrated that besides EBV+ GCs (n = 19), microsatellite instable (MSI) GCs (n = 47) have strong enrichment of IFN-g response gene expression, in contrast to genomic stable (GS) (n = 42) and chromosomal instable (CIN) (n = 106) gastric cancer subtypes (P, 0.001). Conclusions: As the presence of PD-L1+ tumor and/or tumor infiltrating immune cells as well as an interferon-mediated adaptive immune response are associated with PD-1 blockade vulnerability in other cancers, we here provide a strong rationale for testing of PD-1 blockade in EBV+ GC and for evaluating EBV and MSI status as variables for response.

4051

Poster Session (Board #43), Sat, 8:00 AM-11:30 AM

How many lymph nodes should be removed to define an optimal D2 lymphadenectomy for gastric cancer in the modern era? An analysis of 2,947 patients from a two-institution database in China. First Author: Changming Huang, Fujian Medical University Union Hospital, Fuzhou, China Background: Although current guidelines suggest that 16 or more lymph nodes (LNs) are required for the appropriate staging of gastric cancer, the effect on survival of the minimum number of examined LNs (eLNs) in the different types of gastrectomy remains unclear. Methods: We retrospectively analyzed 2662 patients who underwent curative gastrectomy with D2 lymphadenectomy for gastric cancer at Fujian Medical University Union Hospital from Jan 2000 to Dec 2010 and randomly divided them into development (70%, n＝1863) and validation (30%, n＝799) datasets. An additional external validation was performed using the data set (n＝285) that was collected during the same period from the Sun Yat-sen University Cancer Center in Guangzhou, China. A hypothetical TNM classification (hTNM) was proposed based on eLNs and prognosis. Results: The mean numbers of nodes removed during radical distal and total gastrectomy were 2669.6 and 29610.7, respectively (p＜0.01). The optimal LNs count thresholds were determined to be 16 and 21 for patients who underwent curative distal and total gastrectomy, respectively. The hTNM staging system had higher linear trend and likelihood ratioX2 scores and lower AIC values than the seventh AJCC TNM classification; thus, the hTNM staging system exhibited superior prognostic stratification. Similar results were found in the two validation datasets. Conclusions: To maximize the survival benefit after radical total gastrectomy (RTG) for gastric cancer, a minimum of 21 LNs should be removed. For patients undergoing RTG, the hTNM staging system may predict survival more accurately and discriminatively, and this proposed system could have good clinical utility.

4053

Poster Session (Board #45), Sat, 8:00 AM-11:30 AM

Clinicopathological features of program death ligand-1 (PD-L1) expression with tumor-infiltrating lymphocytes (TILs), mismatch repair (MMR) and Epstein-Barr virus (EBV) status in a large cohort of gastric cancer. First Author: Akihito Kawazoe, Department of Gastroenterology and Gastrointestinal Oncology, National Cancer Center Hospital East, Chiba, Japan Background: Antibody-based therapies for targeting programmed death 1 (PD-1) or its ligand PD-L1 (anti-PD1/PD-L1 therapies) have shown promising results in gastric cancer (GC). PD-L1 expression on tumor cell membrane and tumor-infiltrating lymphocytes (TILs) has been proposed as predictive biomarkers for anti-PD-1/PD-L1 therapies in several types of malignancies. TCGA has recently reported that PD-L1 gene was frequently amplified in Epstein-Barr virus (EBV)-positive GC. In addition, mismatch-repair deficiency (D-MMR) was associated with better clinical outcomes after antiPD-1/PD-L1 therapies in solid tumors. However, little is known about clinicopathological features of PD-L1 expression with TILs, MMR and EBV status in a large cohort of GC cases. Methods: We performed a tissue microarray analysis in 487 advanced GC patients who underwent a gastrectomy without preoperative chemotherapy. PD-L1 expression on tumor cell membrane, densities as well as expressions of lymphocyte-associated markers (CD3, CD4, CD8, and FOXP3) of TILs and MMR status were evaluated by immunohistochemistry. EBV status was evaluated by in situ hybridization. Results: PD-L1 expression, D-MMR, and EBV were identified in 22.8, 5.1, and 5.1% of cases, respectively. PD-L1 expression was more frequently observed in elderly (P = 0.002), male (P = 0.029), poorly differentiated adenocarcinoma with solid-type histology (P , 0.001), D-MMR (P , 0.001), and EBV-positive status (P = 0.001). Regarding PD-L1 expression and TILs status, a strong association was observed between PD-L1 expression and high densities of either CD3 (+), CD8 (+), or FOXP3 (+) TILs (P , 0.001). In multivariate analysis, high density of CD8 (+) TILs was significantly associated with better survival (hazard ratio, 0.62; 95% CI, 0.39 to 0.99; P = 0.047), while PD-L1 expression and densities of other subtypes of TILs were not independent prognostic factors. Conclusions: In GC, PD-L1 expression was associated with distinct clinicopathological features including high density of TILs, D-MMR and EBV-positive status, but not a prognostic factor.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Noncolorectal) Cancer 4054

Poster Session (Board #46), Sat, 8:00 AM-11:30 AM

Neoadjuvant chemotherapy versus chemoradiation prior to esophagectomy: Impact on rate of complete pathologic response and survival in esophageal cancer patients. First Author: Albert C. Lockhart, Washington University School of Medicine in St. Louis, St. Louis, MO Background: At this time, short-and long-term outcomes among locally advanced esophageal cancer patients receiving neoadjuvant chemotherapy versus chemoradiation therapy prior to esophagectomy remain poorly characterized with conflicting findings among various institutions. Methods: Esophageal cancer patients receiving either neoadjuvant chemotherapy or chemoradiation prior to esophagectomy were identified using the National Cancer Data Base (NCDB). Univariate analysis compared patient, tumor, and postoperative outcome characteristics. Logistic regression was performed to identify variables associated with achieving pCR. KaplanMeier analysis was performed to compare overall median survival by neoadjuvant therapy type and pCR status. Finally, a Cox proportional hazards model was fitted to identify variables associated with increased mortality hazard. Results: From 2006 – 2012, 916/7,338 (12.5%) of patients received neoadjuvant chemotherapy while 6,422 (87.5%) received neoadjuvant chemoradiation. Neoadjuvant chemoradiation patients were more likely to achieve pCR (17.2% versus 6.4%, p , 0.001) and less likely to have positive margins (5.6% versus 11.5%, p , 0.001) than neoadjuvant chemotherapy patients, with no difference in 30- or 90-day mortality. Achieving pCR was associated with improved overall median survival (59.5 months 6 4.0 versus 30.1 months 6 0.76 for those with persistent disease, p , 0.001). On logistic regression, neoadjuvant chemoradiation therapy was independently associated with achieving pCR (Odds Ratio 2.75, 2.01 – 3.77, p , 0.001). Despite improvement in pCR rate with neoadjuvant chemoradiation, neoadjuvant therapy type was not independently associated with long-term survival (HR 1.12, 95% CI 0.97 – 1.30, p = 0.12). Conclusions: While neoadjuvant chemoradiation is more successful in downstaging esophageal cancer prior to esophagectomy, this therapy was not independently prognostic for improved long-term survival. Other factors affecting long-term survival among pathologic complete responders and among patients with persistent disease should be investigated to clarify this association.

4056

Poster Session (Board #48), Sat, 8:00 AM-11:30 AM

4055

219s Poster Session (Board #47), Sat, 8:00 AM-11:30 AM

miR-125a-induced cellular switch to elicit a response to anti-HER2 targeted therapy in gastric cancer cells. First Author: Elad Hikri, Sackler Faculty Of Medicine, Tel-Aviv University, Tel-Aviv, Israel Background: HER2 (ERBB2) amplification in gastric cancer ranges from 623%, accordingly Trastuzumab has been incorporated into the treatment arsenal of HER2-enriched gastric cancer. We had previously demonstrated the miR-125a-3p induces overexpression of HER2 in basal-like breast cancer cells and sensitizes them to anti-HER2 therapy (AACR-NCI-EORTC 2015). We aimed to study the effect of miR-125a-3p as a potential modulator of the ERBB2/HER2 pathway in HER2-negative gastric adenocarcinoma. Methods: We generated stable KATO-III cells that overexpress miR125a-3p and control cells that overexpress scrambled miRNA. Relative mRNA level of ERBB2 was measured by qPCR and its protein expression and localisation were examined by western blot and immunofluorescence staining. Moreover, the effect of miR-125a-3p alone or combined with antiHER2 therapies on cellular proliferation was evaluated using EdU incorporation and XTT assays. Results: miR-125a-3p-overexpressing KATO cells showed a significant increase in the expression level of ERBB2 mRNA and protein as well as a stronger immunofluorescence staining of ERBB2 on cell membrane compared with control cells. Trastuzumab reduced cell growth and proliferation of miR-125a-3p-overexpressing KATO cells. Furthermore, this antiproliferative effect was enhanced following pre-treatment of the miR-125a-3p-overexpressing KATO cells with lapatinib, a dual ErbB1 and ErbB2 receptor tyrosine kinase inhibitor, prior tp trastuzumab administration. Conclusions: Our results indicate that miR-125a-3p is capable of inducing a shift in the expression and function of ERBB2 pathway that may convert the fate of gastric cancer cells to effectively dispose them to antiHER2 therapies. In an era of personalized medicine, our study proposes a means to enlarge the patient population that may benefit from anti-HER2 therapies.

4057

Poster Session (Board #49), Sat, 8:00 AM-11:30 AM

A propensity-matched analysis comparing hepatic recurrence after curative gastrectomy followed by adjuvant HIPEC to surgery alone for advanced gastric cancer. First Author: Satoshi Murata, Department of Surgery, Shiga University of Medical Science, Otsu, Japan

Impact of earlier adjuvant chemotherapy after surgery on prognosis in patients with operable gastric cancer: A propensity score matched analysis. First Author: Mi-Jung Kim, Center for Gastric Cancer, Research Institute and Hospital, National Cancer Center, Goyang, Korea South

Background: Prevention of recurrence after curative (R0) surgery is one of the most important subjects to be resolved for gastrointestinal (GI) cancers. We have previously shown that hyperthermic intraperitoneal chemotherapy (HIPEC) following R0 gastrectomy is effective for preventing peritoneal recurrence. However, little is known about the preventive effect of HIPEC for GI cancers on hepatic recurrence. The aim of this study was to compare hepatic or peritoneal recurrence-free survival between patients with or without adjuvant HIPEC after R0 gastrectomy for an advanced gastric cancer (GC). Methods: Patients with advanced GC who underwent gastrectomy with a curative intent in the Shiga University of Medical Science hospital between 1998 and 2012 were included in this single center, retrospective, propensity score-matched cohort study. Cox proportional hazards regression models were used to evaluate the association between adjuvant HIPEC and hepatic or peritoneal recurrence-free survival. The Kaplan-Meier method was used to calculate the survival rate. P values , 0.05 were considered statistically significant. HIPEC was performed using CDDP and MMC, or combined with 5-FU in 5 L of perfusate of saline maintained at 42–43°C for 30 min. Results: A total of 186 patients with an advanced GC, in which the pathological depth of invasion was beyond the muscularis propria, were included in the study. There was no significant difference in clinicopathological factors between the matched cohorts (with or without HIPEC). There were significant differences in the hepatic recurrence-free survival (hazard ratio [HR] for with HIPEC versus without HIPEC: 0.10, 95%CI: 0.012 to 0.83, P: 0.033) and peritoneal recurrence-free survival (HR: 0.20, 95%CI: 0.068 to 0.61, P: 0.005). The 3- and 5-year overall survival rate was 94.0% and 86.8% in patients with HIPEC and 59.1% and 53.4% in patients without HIPEC (log-rank: P , 0.0001). Conclusions: Adjuvant HIPEC performed with R0 gastric surgery showed a preventive effect on hepatic recurrence, as well as peritoneal recurrence and survival benefits for patients with advanced GC compared with R0 surgery alone.

Background: Adjuvant chemotherapy improves survival after curative D2 gastrectomy in patients with operable gastric cancer. However, the optimal timing of initiation of adjuvant chemotherapy after surgery is not clearly defined. Methods: Patients with operable gastric cancer who had undergone adjuvant chemotherapy after D2 gastrectomy for stage II and III gastric cancer from January 2009 to December 2012 at National Cancer Center, South Korea were included. Patients were classified into the two groups ( # 6 weeks vs. . 6 weeks) according to the time to initiation of adjuvant chemotherapy after surgery. Propensity score matching (PSM) was performed to adjust the difference of baseline clinicopathologic characteristics between the two groups. Survival outcomes were compared between groups after PSM. Results: Of 484 enrolled patients, 367 patients started adjuvant chemotherapy within 6 weeks after surgery (early group), and 117 patients started adjuvant chemotherapy more than 6 weeks after surgery (late group). After PSM, 232 patients were included (116 in the early group vs. 116 in the late group). In pre-PSM analysis, baseline characteristics were significantly different for age, surgical complication, the type of chemotherapy regimen and the degree of chemotherapy completion between the early and late groups. With median follow up of 60 months (range, 37.3 – 85.5), the 5-year overall survival (OS) rate was 71.4% in the early group and 74.5% in the late group (P = 0.864). The 5-year disease-free survival (DFS) was 66.6% vs. 69.8%, respectively (P = 0.801). After PSM, baseline characteristics were well balanced between groups, but there was no significant difference for survival outcomes between groups: for 5-year OS rate, 76.4% vs. 75.1% (P = 0.741), for 5-year DFS rate, 70.2% vs. 70.4% (P = 0.550). Conclusions: This study suggests that adjuvant chemotherapy is equally effective regardless of the initiation of adjuvant chemotherapy within 6 weeks after surgery. Clinicians should consider allowing adequate time for patient recovery after D2 gastrectomy.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

220s 4058

Gastrointestinal (Noncolorectal) Cancer Poster Session (Board #50), Sat, 8:00 AM-11:30 AM

The Nationwide Cancer Genome Screening Project in Japan SCRUM-Japan GI-SCREEN: Efficient identification of cancer genome alterations in advanced non-colorectal (Non-CRC) gastrointestinal cancer. First Author: Atsuo Takashima, Gastrointestinal Medical Oncology Division, National Cancer Center Hospital, Tokyo, Japan Background: We conduct the nationwide cancer genome screening project in Japan since 2015 using Next Generation Sequencing in advanced noncolorectal gastrointestinal (GI) cancer (aNon-CRC), called as SCRUM-Japan GI-SCREEN 2015-01-Non CRC. The objective is to evaluate the frequency of cancer genome alterations in aNon-CRC and to identify patients (pts) who are candidate for clinical trial for corresponding targeting agents. Methods: Pts with aNon-CRC including esophageal cancer (EC), gastric cancer (GC), and other types of aNon-CRC. Twenty ng of DNA and 10 ng of RNA were extracted from FFPE tumor samples and were analyzed by the Oncomine Cancer Research Panel (OCP) which allows to detect gene mutation, copy number variant (CNV) and fusions across 143 genes in a CLIA certified CAP accredited laboratory. The detected genomic variant data were classified according to whether genetic driver of cancer including gain- and loss-offunction or single nucleotide variant based on the Oncomine Knowledgebase. Results: As of October 31st in 2015, a total of 243 GC and 85 EC pts were enrolled and 235 GC and 83 EC samples were available. A total of 231 GC and 79 EC samples were analyzed and four GC and EC samples each are currently under analysis. The sequence with the OCP was successfully performed in 174 GC (75.3%) and 52 EC (65.8%). The most frequently detected mutations were TP53 (48.3%), PIK3CA (6.9%), KRAS, and SMAD4 (5.7%) in GC and were TP53 (76.9%), NFE2L2 (25.0%), CDKN2A, and PIK3CA (9.6%) in EC. Those of CNVs were ERBB2 (8.6%), CCNE1 (5.2%), FGFR2, and KRAS (4.0%) in GC and were CCND1 (44.2%), EGFR (13.5%), and SOX2 (9.6%) in EC. WIPF2-ERBB2 fusion and EGFR vIII were detected in GC and FGFR3-TACC3 fusion was detected in EC. Pts with PIK3CA or AKT1 mutation, MET or FGFR2 amplification and FGFR3-TACC3 fusion were enrolled in early clinical trials. Conclusions: This nationwide screening system is efficient to detect rare mutations in GI cancer. This novel knowledge provides an intriguing background to investigate new target approaches and represents a progress toward more precision medicine. Clinical trial information: UMIN000016344.

4060

Poster Session (Board #52), Sat, 8:00 AM-11:30 AM

Long-term outcomes of laparoscopic distal gastrectomy compared with open distal gastrectomy for clinical stage I gastric adenocarcinoma (KLASS-01): A multi-center prospective randomized controlled trial. First Author: Hyung-Ho Kim, Department of Surgery, Seoul National University Bundang Hospital, Seoul National University College of Medicine, Seongnam, South Korea Background: Laparoscopic distal gastrectomy (LDG) has numerous advantages over open distal gastrectomy (ODG) for defeating gastric cancer. However, there is no direct evidence regarding oncological safety of LDG due to lack of long-term results from phase III randomized controlled trials. Methods: A phase III, multi-center, open-label, non-inferiority, prospective randomized controlled trial comparing LDG with ODG was conducted. The primary endpoint was 5-year overall survival. Inclusion criteria of patients were pathologically-proven gastric adenocarcinoma and preoperative clinical stage I (AJCC/UICC 6th edition). Results: A total of 1416 patients were randomly assigned to laparoscopic group (n = 705) or open group (n = 711) between Jan. 2006 and Aug. 2010. Among them, 1377 patients including 681 laparoscopic group and 696 open group were eligible for intention-totreat (ITT) analysis. For per-protocol (PP) analysis, 1203 patients, 618 laparoscopic group and 585 open group, were included. By ITT analysis, the 5-year overall survival (OS) rates were very similar in the two groups (95.8% in the laparoscopic group and 95.9% in the open group; log-rank P = 0.774). The lower limit of the 95% confidence interval for the 5-year overall survival rate difference was greater than the pre-specified non-inferiority margin, 5%. The 5-year relapse-free survival (RFS) rates were also similar in the two groups (92.5% in the laparoscopic group and 94.3% in the open group; logrank P = 0.317). By PP analysis, the 5-year OS rates (96.4% in the laparoscopic group and 96.3% in the open group; log-rank P = 0.929) and the 5year RFS rates (92.9% in the LDG and 94.7% in the ODG group; log-rank P = 0.368) were also similar in the two groups. Conclusions: The long–term survival after laparoscopic distal gastrectomy is not inferior to that of open distal gastrectomy in clinical stage I gastric cancer. Our study suggests that the laparoscopic distal gastrectomy is an oncologically acceptable alternative to open distal gastrectomy for clinical stage I gastric cancer. Clinical trial information: NCT00452751.

4059

Poster Session (Board #51), Sat, 8:00 AM-11:30 AM

Esophagogastric junction cancer and classification as esophagus cancer according to molecular fingerprint. First Author: Williams Fernandes Barra, Universidade Federal do Para - Nucleo de Pesquisas em Oncologia, Belem, Brazil Background: The 7th edition of UICC staging system moved GEJ cancers from gastric staging group to esophagus cancers. Since clinical management is strongly influenced by this staging system, we looked at molecular fingerprints of GEJ tumors and compared to gastric and esophagus profiles, to elucidate whether GEJ cluster in the gastric or esophagus cancer groups, according to mRNA expression pattern, since this might represent tumor identity. Methods: The clinical and expression data were downloaded from TCGA. The total number of samples was 1167 ( 450 stomach, 196 esophagus and 521 colon samples). Both Principal Component Analysis (PCA) and Heat Map (HM) plots were performed in R platform, using Log2 transformation of RPKM normalized data. Differential Expression Analysis was also performed in R, using RAW data and the DESeq2 package. Benjamini and Hochberg False Discovery Rate corrections were used. The mRNAs were tagged as differentially expressed if they met the following criteria: i) FDR adjusted p-value , 0.05; and ii) |Log2(fold-change)| . 2 (expressions greater or lower than 4 fold change). Results: Esophagus squamous cell carcinoma (ESCC) clustered apart of the others tumors, while adenocarcinomas (AC) clustered all together. The HM also demonstrates that ESCC belongs to a different group, while AC of esophagus looks like AC of the cardia and non cardia regions. Even distal gastric cancers are quite similar to AC of the lower esophagus, demonstrating that esophagus AC relies much closer to gastric cancers than to esophagus cancers. We also analyzed 521 samples of colon AC. Both the PCA and HM images clearly demonstrated that AC of the colon are much different than AC of stomach and esophagus, strengthening the results. Conclusions: By using robust molecular fingerprints, in a large series of tumors from an import cancer data bank, it was strongly demonstrated that GEJ tumors look much more like as gastric cancers than esophagus cancers, despite of tumor heterogeneity.

4061

Poster Session (Board #53), Sat, 8:00 AM-11:30 AM

Peripheral natural killer cells are a prognostic factor in advanced oesogastric adenocarcinoma and are associated with intestinal types in the randomized trial PRODIGE17-ACCORD20 (UNICANCER GI). First Author: Magali Terme, INSERM U970 - Paris-Descartes University, Paris, France Background: Natural Killer cells (NK) are associated with a better prognosis in many types of tumors. Their value is not well known in oesogastric adenocarcinoma (OGA), but tumor infiltrating NK seem to be associated with a good prognosis. The quantification of NK in peripheral blood of patients (pts) could be a simple and reproducible prognostic factor. Objective: To evaluate the prognostic impact of circulating NK among pts with advanced OGA. Methods: 134 pts included in the randomized trial PRODIGE 17-ACCORD 20 (MEGA) (FOLFOX alone or with either panitumumab or rilotumumab in 1st-line treatment of advanced OGA) and having consented to the biological study were collected. Rate of NK in total circulating leukocytes was measured by flow cytometry after immunophenotyping (CD56+ CD3-) at diagnosis. Prognostic effect of NK was analysed through the Cox model by testing several thresholds determined by the distribution of this parameter. Results: NK were measured in 85 pts (63%) included in the study. The median rate was 8.5% with a range of [0.6-34.8]. Patients with NK . 17% (top quintile) had significantly higher PFS (9.2 months vs. 5.8 months; HR = 0.56 [0.31-1.00], p = 0.04), and OS (25.8 months vs. 11.5 months; HR 0.47, [0.23-0.96], p = 0.03). This effect was maintained by adjusting for age, ECOG, location of primitive tumor, histological type (intestinal vs diffuse) and number of metastases (PFS: HR 0.44; [0.23-0.83]; p = 0.01, OS: HR 0.44; [0.21-0.92], p = 0.03). There was no interaction between NK and treatment received. In gastric adenocarcinomas, the rate of NK was significantly lower in diffuse forms than in intestinal forms (p = 0.02). The prognostic value of NK was maintained in the intestinal sub-group, but not in the diffuse subgroup. Conclusions: Measurement of circulating NK in advanced AOG is a simple and reproducible prognostic tool. Pts with NK . 17% have a significantly better OS and PFS. Finally, the low rate of NK in diffuse forms suggests low immunogenicity of this type of tumor, which could be related to a lower antigenic load and potentially with poorer efficacy of immunotherapy. Clinical trial information: NCT01443065.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Noncolorectal) Cancer 4062

Poster Session (Board #54), Sat, 8:00 AM-11:30 AM

Morbidity of laparoscopic distal gastrectomy with D2 lymphadenectomy compared with open distal gastrectomy for locally advanced gastric cancer: Short term outcomes from multicenter randomized controlled trial (KLASS02). First Author: Hyuk-Joon Lee, Department of Surgery and Cancer Research Institute, Seoul National University College of Medicine, Seoul, South Korea Background: Although lots of benefits of laparoscopic gastric cancer surgery have been reported, strong evidences of laparoscopic surgery are still limited, especially in locally advanced gastric cancer which requires extensive lymph node dissection. Herein, we report the short term clinical outcomes of KLASS-02, a multicenter randomized controlled trial comparing laparoscopic distal gastrectomy (LDG) with D2 lymphadenectomy compared with open distal gastrectomy (ODG). Methods: The enrollment criteria include histologically confirmed gastric adenocarcinoma with cT2-4a and N0-1. Primary end-point was relapse free 3-year survival. Secondary end-points included 30-day morbidity, 90-day mortality, postoperative pain, and recovery. Results: A total of 1,050 patients were randomly assigned to LDG (n = 526) or ODG group (n = 524) between November 2011 and April 2015. After excluding patients who received bypass or no surgery, 1,011 patients were analyzed for modified intention-to-treat analysis. Estimated blood loss was significantly less in LDG (155.3 ml) than in ODG (231.2 ml, p , 0.001). Total number of retrieved lymph nodes was comparable between groups (LDG = 46.6 vs. ODG = 47.4, p = 0.46). Lengths of both resection margins were comparable between groups. The 30-day overall complication rate was 16.4% in LDG and 24.3% in ODG with statistical significance (p = 0.002). Postoperative analgesics use and patients’ reported pain visual analog scale was less in LDG than in ODG. First day of flatus was shorter in LDG (3.53 d vs. 3.71 d, p = 0.027) and post-operative hospital stay was shorter in LDG (8.1 d vs. 9.3 d, p = 0.003). Mortality rate within 90 postoperative days was comparable between groups (0.38% in LDG vs. 0.57% in ODG). Conclusions: LDG with D2 lymphadenectomy for locally advanced gastric cancer has benefits of less complication rate, faster recovery, and less pain without compromising oncologic safety, compared with open surgery. Clinical trial information: NCT01456598.

4064

Poster Session (Board #56), Sat, 8:00 AM-11:30 AM

Correlation between mismatch repair deficiency (MMRd), microsatellite instability (MSI) and survival in MAGIC. First Author: Elizabeth Catherine Smyth, Royal Marsden NHS Foundation Trust, London and Sutton, United Kingdom Background: MMRd and MSI are prognostic for survival in many cancers and predict resistance to fluoropyrimidines in colon cancer. We examined the relationship between MMRd, MSI and survival in patients (pts) with resectable gastroesophageal cancer randomised to surgery alone or perioperative ECF chemotherapy in the MRC MAGIC trial. Methods: For MMR, tumour sections were assessed for expression of MMR proteins MLH1, MSH2, MSH6 and PSM2. MSI status was determined using the Promega MSI Analysis System (NR-21, BAT-26, BAT-25, NR-24, MONO-27). MMR status was correlated with MSI and the relationship between MMR status and survival was assessed. Results: MMRd results were available for 302 pts (66% resected pts). Sixteen (5%), 3 (1%), 2 (1%) and 18 (6%) pts were deficient in MLH1, MSH2, MSH6 and PMS2 respectively. Fourteen of 16 MLH1-ve tumours (88%) with available MSI results had MSI compared to 2/269 (1%) MLH1+ve tumours. Pts with MLH1-ve tumours treated with chemotherapy plus surgery had inferior median survival compared to MLH1+ve pts (7.2m vs 22.3m) whereas for pts treated with surgery alone the effect was in the opposite direction; median survival for MLH-ve pts was not reached, that for MLH1+ve pts was 20.3m. The interaction test between MHL1 and treatment was significant p=0.01. MSH2, MSH6 and PMS2 did not demonstrate similar relationships. Conclusions: In MAGIC, 88% of MLH1-ve pts demonstrated MSI and MLH1 deficiency was associated with a positive prognostic effect in pts treated with surgery alone and a differentially negative prognostic effect in pts treated with chemotherapy. These results are consistent with those reported for MSI status in this dataset. If independently validated, either MSI or MMRd (specifically MLH1) status on pre-operative biopsies could be used to select patients for peri-operative chemotherapy.

MLH1 status and overall survival in MAGIC. Chemotherapy patients

Patients Events Median OS (months) 95% CI HR 95% CI p-value

Surgery patients

MLH1 +ve

MLH1 -ve

MLH1 +ve

MLH1 -ve

122 77 22.3 (16.1, 35.7)

8 7 7.2 (0.1, NR) 2.73 (1.25, 5.96) 0.012

148 101 20.3m (16.7, 27.8)

6 1 NR (4.9, NR) 0.17 (0.02, 1.27) 0.085

*only patients with date of surgery included in survival analysis

4063

221s Poster Session (Board #55), Sat, 8:00 AM-11:30 AM

Randomized phase II trial comparing 4 doublets as a front-line treatment in Asian patients with recurrent/metastatic gastric cancer. First Author: Hongjae Chon, Medical Oncology, CHA Bundang Medical Center, CHA University, Seongnam, Korea., Seongnam, Korea, The Republic of Background: Despite a large number of randomized trials, there is no global consensus as to the standard regimen for front-line chemotherapy of recurrent/metastatic gastric cancer. We conducted a randomized phase II study to compare 4 doublet regimens, S-1 and cisplatin (SP); FOLFOX; docetaxel and 5-FU (DF); and paclitaxel and 5-FU (TF), for their therapeutic efficacy and safety as a first line treatment. Methods: Patients with no prior chemotherapy for recurrent/metastatic gastric adenocarcinoma were enrolled and randomized evenly to each regimen. The primary endpoint was progression free survival (PFS), and secondary endpoints were overall survival (OS), response rate (RR), and safety profile. Analysis was performed by intention to treat. Results: A total of 177 patients were enrolled from MAR 2010 to MAY 2015. Four treatment arms were well-balanced: median age 57, male 69.5%, and ECOG performance 0/1/2 = 61/37/2%. At data cutoff (DEC 31, 2015), median PFS were 8.6 (3.5-13.7) months for SP, 5.8 (2.98.7) months for FOLFOX, 6.8 (2.4-11.2) months for DF, and 4.6 months for TF (3.6-5.6) (p = 0.054). Median OS were 14.7 months for SP, 11.3 months for FOLFOX, 11.7 months for DF, and 10.7 months for TF (p = 0.208), and RR were 17.8% for SP, 21% for FOLFOX, 16% for DF, and 33% for TF). In terms of platinum- and taxane-based regimens, the PFS and OS did not differ significantly (PFS: 7.2 vs. 5.7 months; p = 0.207, OS: 13.3 vs. 11.7 months; p = 0.701). Toxicity was generally well tolerated, with grade 34 neutropenia observed in 33.3% of SP, 9.3% of TF, 55.6% of DF, and 45.5% of FOLFOX arm, while febrile neutropenia was observed in only 1 patient in SP arm and 2 patients in DF arm. There was no treatment-related mortality. Conclusions: Although none of the 4 regimens revealed significant superiority over others in the treatment of recurrent/metastatic gastric cancer, SP regimen showed a tendency of increased PFS and OS compared with other regimens. Clinical trial information: NCT01283204.

4065

Poster Session (Board #57), Sat, 8:00 AM-11:30 AM

KRAS gene amplification to define a distinct molecular subgroup of gastroesophageal adenocarcinoma. First Author: David Xu, University of Chicago, Chicago, IL Background: KRAS mutation is rare (, 5%) in gastroesophageal cancer (GEC). However, the incidence of KRAS gene amplification (amp+), consequent protein levels, and prognostic and/or therapeutic implications are unknown. Methods: 410 GEC samples and 30 cell lines were assessed for KRAS gene copy number (GCN) by fluorescence in situ hybridization (FISH) (n = 90), Kras expression by selected reaction monitoring mass spectrometry (Kras-SRM-MS) (n = 393), and Kras-SRM level evaluated for correlation with KRAS amp+ status (n = 73). Survival analysis was performed comparing KRAS amp+ versus non-amp+ patients. When possible, concurrent 315 gene next-generation sequencing was also performed. Four KRAS-amplified xenograft lines (CAT-2,12,14,15) were established from malignant effusions. Tumorigenic activity of KRAS amp+ lines (CAT lines, MKN-1) were assessed using MTT and soft agar assays in vitro and subcutaneous xenograft models, compared to non-amp+ lines. Inhibitory assays were performed using KRAS siRNA and CRIPSR, and commercial inhibitors targeting downstream effectors MEK and/or PIK3CA. Results: KRAS FISH revealed clustered gene amp+ in 28.9% (26/90); these patients had worse prognosis than non-amp+ patients. GCN significantly correlated with Kras expression. All KRAS amp+ cell lines significantly overexpressed Kras protein and were tumorigenic in xenograft subcutaneous models. KRAS siRNA and KRAS CRISPR of KRAS amp+ cell lines demonstrated inhibition in MTT viability and soft agar assays, compared to appropriate controls, and demonstrated significant and durable xenograft growth reduction. Conversely, inhibition using MEK and/or PI3K inhibitors demonstrated only transient growth reduction in vivo. Conclusions: KRAS gene amp+ was observed in a large subset (26%) of GEC patients, which correlated with extreme expression by mass spectrometry. Established xenograft lines serve as models to investigate therapeutic strategies for KRAS amp+ patients. Inhibition using MEK/PIK3CA inhibitors provided transient benefit for KRAS amp+ tumors while durable inhibition was observed with Kras protein knockdown, suggesting potential benefit from novel siRNA therapeutics currently in development.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

222s 4066

Gastrointestinal (Noncolorectal) Cancer Poster Session (Board #58), Sat, 8:00 AM-11:30 AM

4067

Poster Session (Board #59), Sat, 8:00 AM-11:30 AM

Assessment of preoperative liver function in patients with hepatocellular carcinoma: The albumin-indocyanine green evaluation (ALICE) grade. First Author: Takashi Kokudo, Graduate School of Medicine, The University of Tokyo, Tokyo, Japan

Survival benefit of liver resection for hepatocellular carcinoma associated with portal vein invasion: A Japanese nationwide survey. First Author: Takashi Kokudo, Graduate School of Medicine, The University of Tokyo, Tokyo, Japan

Background: Most patients with hepatocellular carcinoma (HCC) have underlying liver disease, therefore, precise preoperative evaluation of the patient’s liver function is essential. Methods: We developed a simple grading system incorporating only two variables, namely, the serum albumin level and the indocyanine green retention rate at 15 minutes, to assess the preoperative liver function, based on the overall survival of 1868 patients with HCC who underwent liver resection. We then tested the model in a European cohort (n = 70) and analyzed the predictive power of the grading system for the postoperative short-term outcome in a Japanese cohort. Results: The Albumin-Indocyanine Green Evaluation (ALICE) grading system was developed in a randomly assigned training cohort: linear predictor = 0.663 3 log10 ICG R15 (%) - 0.0718 3 albumin (g/dL) (cut-off value: -2.20 and -1.39). This new grading system showed a predictive power for the overall survival similar to the Child-Pugh grading system in the validation cohort. Determination of the ALICE grade in Child-Pugh A patients allowed further stratification of the postoperative prognosis (median survival time: grade 1, 12.5 y; grade 2, 6.45 y; grade 3, 3.37 y). This result was reproducible in the European cohort. Determination of the ALICE grade allowed better prediction of the risk of postoperative liver failure and mortality (ascites: grade 1, 2.1%; grade 2, 6.5%; grade 3, 16.0%; mortality: grade 1, 0%; grade 2, 1.3%; grade 3, 5.3%) than the previously reported model based on the presence/absence of portal hypertension. The ALICE grade 2 was further divided into 2 groups (2a and2b) using the median linear predictor (-1.88) as a cut-off value. Although major resection did not affect the incidence of ascites or the mortality rate in the ALICE grade 2a group, major resection in the ALICE grade 2b group had a significantly higher incidence of ascites and mortality rate. Conclusions: This new grading system is a simple and objective method for prediction of the postoperative long-term and short-term outcomes. This new system could be potentially used worldwide for surgical decision making in patients with HCC.

Background: The presence of portal vein tumor thrombosis (PVTT) in patients with hepatocellular carcinoma (HCC) is regarded as indicating an advanced stage, and liver resection (LR) is not recommended. The aim of this study was to evaluate the survival benefit of LR for HCC patients with PVTT through the analysis of the data from a Japanese nationwide survey. Methods: We analyzed data for 6,474 HCC patients with PVTT registered between 2000 and 2007. Of these patients, 2,093 patients who underwent LR and 4,381 patients who received other treatments were compared. The propensity scores were calculated for 1,786 patients in the LR group and 3,758 patients in the non-LR group and we successfully matched 1,229 patients (68.8% of the LR group). Results: The median survival time (MST) in the LR group was 1.93 years longer than that in the non-LR group (2.74 years vs 0.81 years; P , 0.001) and 1.03 years longer than that in the nonLR group (2.41 years vs 1.38 years; P , 0.001) in a propensity scorematched cohort. A subgroup analysis revealed that LR provides a survival benefit regardless of the Child-Pugh grade, etiology of HCC, and tumor number. The survival benefit was not statistically significant only in patients with PVTT invading the main trunk or contralateral branch. In the LR group, the postoperative mortality rate was 1.4% (29 patients) and the multivariate analysis identified liver cirrhosis (hazard ratio [HR] 1.31), Child-Pugh class B (HR 1.69), and R2 resection (HR 1.60) as significant risk factors for the overall survival other than tumor related factors. Conclusions: As long as the PVTT is limited to the first-order branch, LR leads to a longer survival outcome than non-surgical treatment, especially in the presence of good liver function.

4068

4069

Poster Session (Board #60), Sat, 8:00 AM-11:30 AM

Poster Session (Board #61), Sat, 8:00 AM-11:30 AM

Comparison of cooled-probe microwave and radiofrequency ablation treatment in incipient hepatocellular carcinoma: A phase III randomized controlled trial with 6-year follow-up. First Author: Jie Yu, Chinese PLA General Hospital, Beijing, China

Squamous cell carcinoma of gallbladder: A Surveillance Epidemiology and End Results (SEER) database review of demographics and survival. First Author: Yi-Tsung Lu, Department of Medicine, John H. Stroger Jr, Hospital of Cook County, Chicago, IL

Background: According to randomized controlled trial (RCT) requirement, the therapeutic effectiveness of cooled-probe microwave ablation (MWA) versus radiofrequency ablation (RFA) on early-stage , 5cm hepatocellular carcinoma (HCC) was investigated to find a better approach for minimally invasive thermal ablation. Methods: From October 2008 to June 2015, 403 cases of biopsy-proved HCC patients were involved in a RCT study, 203 (265 nodules) cases were treated with ultrasound guided percutaneously MWA and 200 (251 nodules) cases were treated with RFA. The patients were then followed up with contrast enhanced imaging including contrast enhanced ultrasound, magnetic resonance imaging or computed tomography to evaluate treatment response prospectively, including technique effectiveness (TE), local tumor progression (LTP), intrahepatic metastases, extrahepatic metastases, complications, disease free survival(DFS) and overall survival(OS). Survival was analysed using the Kaplan–Meier method. The study has been registered in Clinical-Trials.gov and the identifier number is NCT02539212. Results: The follow-up period was 35.2 (range 2.0-81.9) months. MWA needed less applicator number, ablation session, puncture number, ablation duration and expense (P, 0.05). The TE rates were 99.6% (264/265) in MWA and 98.8% (248/251) in RFA. The 1-, 3-, 5-year LTP were 1.1%, 4.3%, 11.4% for MWA versus 2.1%, 5.8%, 19.7% for RFA, respectively(P= 0.11). The 1-, 3-, 5-year OS were 96.4%, 81.9% and 67.3% for MWA and 95.9%, 81.4% and 72.7% for RFA, respectively (P = 0.91). There was no difference in 5-year intrahepatic metastatic (P= 0.30) and extrahepatic metastatic rates(P= 0.12) between two groups. There was also no difference in major complications between MWA(3.4%) and RFA (2.5%) (P= 0.59). Conclusions: Both MWA and RFA are suitable options for early-stage HCC. MWA shared the advantages of less ablation duration, puncture, session and expense. Therefore, MWA seems a safe and promising treatment of HCC. Clinical trial information: NCT02539212.

Background: Squamous cell carcinoma of the gallbladder (SCCGB) is a rare subtype of gallbladder cancer (GBC) with its distinct biological behavior, different from gall bladder adenocarcinoma (ACGB). We aimed to study the demographics and survival of patients with SCCGB by reviewing the SEER database. Methods: From 1973-2012, we identified 17328 patients with primary GBC from the SEER data and extracted 384 SCCGB cases to examine patient characteristics and clinical outcomes by histology and stage. Associated factors to SCCGB were estimated with a multivariable logistic regression model. Adjusted hazards ratios (AHR) were calculated using Cox models. Results: 65% of SCCGB patients were women. The ethnic makeup was 55% Caucasian, 25% Hispanic, 12% Afro-American, and 8% other. Compared with ACGB, SCCGB was associated with non-Caucasian ethnicity, less primary tumor resection, more carcinoma in situ, and lower N stage at diagnosis. Survivals varied among different ethnicities in SCCGB patients, and the Hispanic patients had the best outcome (HR 0.83, p = 0.037 vs nonHispanic Caucasian). Further analysis revealed SCCGB in Hispanic patients was associated with younger age, and when factoring this, Hispanic ethnicity was not an independent predictor of survival. Survival analysis revealed that better outcome was correlated with the performance of primary surgery (AHR 0.33, p = 0.001), lower grade (AHR 0.49, p = 0.03), and lower stage (AHR 0.35, p = 0.003). The year of diagnosis was not correlated with outcome as the survival of SCCGB patients before and after 1998 was not significantly different. Conclusions: SCCGB has its distinct demographics different from ACGB. Primary surgery remained the only curative measure, and patients undergoing surgery had longer survival, regardless of grade or stage. In a disease so heavily associated with cholelithiasis, the younger age of diagnosis in Hispanics may be due to the higher prevalence of cholelithiasis in younger Hispanics. Similar survival across 30 years of SEER registry indicates the need for continued investigation into the treatment strategies in patients with SCCGB.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Noncolorectal) Cancer 4070

Poster Session (Board #62), Sat, 8:00 AM-11:30 AM

A phase 2 study of galunisertib, a novel transforming growth factor-beta (TGF-b) receptor I kinase inhibitor, in patients with advanced hepatocellular carcinoma (HCC) and low serum alpha fetoprotein (AFP). First Author: Sandrine J. Faivre, Service d’Oncologie Medicale, ´ Clichy, France Background: TGFb signaling is associated with HCC progression. Here we report the results of a cohort of patients with HCC and normal serum AFP treated with galunisertib as part of a larger phase 2 study. Methods: Eligible patients with HCC included: those who had progressed on or were ineligible to receive sorafenib, advanced Child-Pugh A/B7 HCC, AFP , 1.5x ULN, ECOG PS # 1, measurable disease per RECIST 1.1, and # 1 prior systemic regimen. Galunisertib was administered as intermittent dosing of 14 days on/ off (300 mg/day, 28 days = 1 cycle). Primary objectives were time-toprogression (TTP) and evaluation of biomarker changes (TGFb1 and Ecadherin). Secondary objectives included overall survival (OS), toxicity (CTCAE, v 4.0) and pharmacokinetics. Results: 40 patients were enrolled. 90% of patients were male; median age: 68 years; PS = 0/1, 60%/40%; Child-Pugh A/B 75%/23%; etiology: hepatitis C 18%, hepatitis B 15%, alcohol 23%; multiple 10%. Overall, 85% of patients had received prior sorafenib. TTP was 4.16 months 90%CI: (2.76, 5.89). Median OS was 16.8 months (90%CI: 10.4,24.1). Four patients discontinued treatment due to adverse events (AEs). Grade 3/4 possibly treatment related AEs were observed in 6 patients: 1 occurrence of each of the following: hypoalbuminemia, neutropenia, thrombocytopenia, abdominal pain, constipation, drug eruption, pruritus and gall bladder perforation. Median baseline serum TGFb1 was 3.39 ng/mL (range: 1.16-37.5) and E-cadherin was 6.33 mg/mL (range: 1.45-18.9). TGFb1 decline of . 20% occurred in 74% of patients. In patients with TGFb1 reduction . 20%, median OS was 21.8 months vs 7.91 months for patients with reduction of # 20% (Log Rank p = 0.002). E-cadherin reduction of . 20% occurred in 46% of patients. Conclusions: HCC patients with normal AFP and with TGFb1 reduction showed improvement in OS compared to patients with non-TGFb1 reduction. Clinical trial information: NCT01246986.

4072

Poster Session (Board #64), Sat, 8:00 AM-11:30 AM

Tolerability and activity of tepotinib in Asian patients with advanced hepatocellular carcinoma (HCC). First Author: Shukui Qin, Medical Oncology Department, Nanjing Bayi Hospital, Nanjing, China Background: Patients (pts) with advanced HCC have a poor prognosis, particularly if tumor c-Met activity is aberrant. Current first-line therapy is sorafenib, which has limited efficacy and is not widely used in Asia. Tepotinib (MSC2156119J) is a highly selective c-Met inhibitor with favorable safety and promising antitumor activity, particularly in c-Met+ tumors. Translational modeling established a recommended phase II dose (RP2D) of 500 mg/day. All patients enrolled into a phase Ib trial of tepotinib in Asian pts with HCC have completed the dose-limiting toxicity (DLT) evaluation period and data are reported. Methods: Asian adults with confirmed HCC of BCLC Stage C, Child-Pugh Class A liver function without encephalopathy, and ECOG PS 0–2 were recruited to this 3+3 doseescalation study (tepotinib 300 or 500 mg/day, 21-day cycle; protocol amendment added a 1,000 mg/day cohort, n = 6). In total, 12 patients were to be enrolled at the RP2D. c-Met expression was retrospectively determined by IHC. PK was analyzed by non-compartmental methods. Results: 27 pts were enrolled (median age 57 [38-69]; male 23; ECOG PS 0/1 11/15). 7 received tepotinib 300 mg/day, 14 500 mg/day, and 6 1,000 mg/day. No DLTs were observed. 22 pts experienced treatment-related treatmentemergent adverse events (TRTEAEs); most were grade # 2. The most common TRTEAEs were diarrhea (n = 10), elevated AST (7), nausea (7), and elevated ALT (6). 15 grade $ 3 TRTEAEs occurred, including grade 3 increased lipase levels (n = 3) and grade 3 diarrhea (n = 2). Best overall response (BOR) in pts with c-Met+ (n = 7) tumors was partial response (n = 2), stable disease (SD; n = 2), and progressive disease (PD; n = 3). The 2 pts with PRs were treated with tepotinib 500 and 1,000 mg/day, respectively; AFP levels in these pts were at least stabilized during response to tepotinib. Of 18 pts with c-Met– tumors, 6 had a BOR of SD, 10 PD, and 2 were not evaluable. Conclusions: Data indicate that tepotinib at doses of up to 1,000 mg/day is well tolerated by Asian pts with HCC. Efficacy appeared to be greatest in the c-Met+ population. The ongoing phase II part of this study is comparing the efficacy and safety of first-line tepotinib and sorafenib in pts with c-Met+ HCC. Clinical trial information: NCT01988493.

4071

223s Poster Session (Board #63), Sat, 8:00 AM-11:30 AM

Dexamethasone for the prevention of transcatheter arterial chemoembolization-induced fever, nausea, vomiting, and anorexia in patients with hepatocellular carcinoma: A randomized, double-blind, placebocontrolled trial. First Author: Sadahisa Ogasawara, Department of Gastroenterology and Nephrology, Graduate School of Medicine, Chiba University, Chiba, Japan Background: Fever, nausea, vomiting, and anorexia are the most frequent adverse events of transcatheter arterial chemoembolization (TACE) in patients with hepatocellular carcinoma (HCC). In this study the efficacy of dexamethasone in preventing these events was evaluated. Methods: Patients with HCC who had a Child–Pugh score of A or B and who lacked macrovascular invasion/extrahepatic metastasis were included in this study. Patients were randomly assigned to either a dexamethasone regimen [day 1, intravenous dexamethasone (20 mg) and granisetron (3 mg) before TACE; day 2 and 3, intravenous dexamethasone (8 mg)] or a control regimen [day 1, intravenous placebo (saline) and granisetron (3 mg); day 2 and 3, intravenous placebo]. The primary endpoint was a complete response (CR), which was defined as no grade 1 or higher fever, nausea, vomiting, or anorexia, according to CTCAE (version 4.0), and no use of rescue therapy for 120 h after TACE. Results: From October 2010 to June 2013, 120 patients were randomly assigned. One patient randomly assigned to the dexamethasone regimen did not receive TACE and was therefore excluded from analysis. Thus, 59 patients were randomly assigned to the dexamethasone regimen and 60 to the control regimen and underwent intention-to-treat analysis. The overall CR rate was greater with the dexamethasone regimen than with the control regimen (47.5% vs. 10.2%; P , 0.001). The overall incidence of fever, nausea and vomiting, and anorexia within 120 h following TACE were 28.8%, 27.1%, and 33.9% in the dexamethasone regimen and 72.9%, 47.5%, and 78.0% in the control regimen, respectively. The rate of fever, nausea and vomiting, and anorexia were significantly higher in patients receiving the control regimen than those receiving the dexamethasone regimen (fever: P , 0.001, nausea and vomiting: P = 0.022, and anorexia: P , 0.001). Both treatment regimens were generally well tolerated. Conclusions: The dexamethasone regimen was more effective than the control regimen for the prevention of TACE-induced fever, nausea, vomiting, and anorexia in patients with HCC. Clinical trial information: UMIN000004317.

4073

Poster Session (Board #65), Sat, 8:00 AM-11:30 AM

Tremelimumab: A monoclonal antibody against CTLA-4—In combination with subtotal ablation (trans catheter arterial chemoembolization (TACE), radiofrequency ablation (RFA) or cryoablation) in patients with hepatocellular carcinoma (HCC) and biliary tract carcinoma (BTC). First Author: Austin G. Duffy, National Cancer Institute at the National Institutes of Health, Bethesda, MD Background: Tremelimumab is a fully human monoclonal antibody that binds to CTLA-4 expressed on the surface of activated T lymphocytes and results in inhibition of B7-CTLA-4-mediated downregulation of T-cell activation. Transcatheter arterial chemoembolization (TACE), radiofrequency (RFA) and cryoablation (CA) have been shown to induce a peripheral immune response which may enhance the effect of anti-CTLA4 treatment. Methods: Patients with HCC [Childs Pugh A/B7; BCLC B/C; ECOG 0/1; postsorafenib (BCLC stage C only)] or refractory BTC were enrolled in a study of Tremelimumab combined with subtotal TACE, RFA or CA performed on week 6. All BTC patients received RFA in combination with tremelimumab. Tumor biopsies were performed at baseline and at time of RF/TACE. Results: 41 pts enrolled (32 HCC, 9 BTC). Characteristics: M:F 31:10; Median age 54(range 42-76); In HCC pts cirrhosis present in 22pts, BCLC Stage B/C: 9/23; Hepatitis B/C/neg: 5/18/9. 14 pts received TACE, 19 underwent RFA (inc all 9 BTC pts), 5 CA during week 6 of tremelimumab therapy. 3 pts did not receive an ablative procedure (due to PD). No DLT encountered. Most common toxicity was pruritus. Of N = 17 pts evaluable for response outside of TACE/RFA-treated lesion 4 (23.5%) achieved confirmed partial responses. 10 of 12 pts with quantifiable HCV experienced a marked reduction in viral load. 6-week tumor biopsies showed clear increase in CD8+ T cells only in pts showing a clinical response. Flow cytometry of PBMC revealed statistically significant changes in CD4/Treg and CD8/Treg ratio only in patients showing clinical response. Median PFS for evaluable HCC population was 5.7m. Conclusions: Tremelimumab in combination with subtotal TACE, RFA or CA in patients with advanced HCC and BTC is safe and leads to the accumulation of intratumoral CD8+ T cells, activation of CD4+ and CD8+ T cells in peripheral blood in responding patients. Encouraging clinical activity seen with objective confirmed responses, PFS 5.7m and possibly surrogate reductions in HCV viral load. Clinical trial information: NCT01853618.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

224s 4074

Gastrointestinal (Noncolorectal) Cancer Poster Session (Board #66), Sat, 8:00 AM-11:30 AM

A phase (Ph) II study of the efficacy and safety of the cMET inhibitor capmatinib (INC280) in patients (pts) with advanced hepatocellular carcinoma (HCC). First Author: Tawesak Tanwandee, Division of Gastroenterology, Department of Medicine, Faculty of Medicine Siriraj Hospital, Mahidol University, Bangkok, Thailand Background: Overexpression/activation of cMET occurs in 20‒48% of HCC, and predicts worse survival. Capmatinib (INC280) is a potent, selective cMET inhibitor that causes regression of cMET dysregulated (cMET+) animal solid tumor models at well-tolerated doses. This open-label, single-arm study is evaluating the safety and efficacy of INC280 in pts with cMET+ advanced HCC, who have received no prior systemic therapy (NCT01737827). Methods: The primary endpoint is time to progression (TTP); secondary endpoints are ORR, DCR, PFS, OS, safety and tolerability, and PK. Eligible pts were aged $ 18 years, ECOG PS # 2, Child-Pugh A, and had no prior cMET/HGF-targeted therapy. Based on responding pts in preliminary studies, an eligibility criterion was amended to specify that tumors must have high cMET status (IHC 3+, IHC 2+ and cMET gene copy number [GCN] $ 5 by FISH, or cMET GCN $ 5 and unknown IHC); low cMET status is defined as IHC 2+ and GCN , 5, IHC 1+ and any GCN, or cMET GCN $ 5 and IHC 0+. Results: As of September 28, 2015, 36 pts were enrolled: 8 pts in a dosedetermining part (INC280 300 mg BID) and 28 pts in Ph II dose-expansion groups (INC280 600 mg BID [capsules], except for one pt who received the wrong dose of 300 mg BID). In expansion pts (8 with high cMET and 20 with low cMET status; 89% male, 57% ECOG PS 0, 57% with distant metastases) median age was 55.5 years (range 34–74). Treatment was discontinued in 24/28 (86%) pts, mainly due to disease progression (54%); treatment is ongoing in 4 (14%) pts. The most common AEs, regardless of causality, were increased AST (39%), nausea (36%), diarrhea (32%), and vomiting (32%). The most common Grade 3/4 AEs, regardless of causality, were increased AST (29%), and hyperbilirubinemia (11%). INC280 PK was similar to that seen in non-HCC pts. Partial responses (PRs) were seen in 3/28 pts (ORR 11%). All responders had high cMET status; there were 3/8 PRs and 1/8 pts had stable disease in the high cMET group (ORR 38%; DCR 50%). Conclusions: Oral INC280 600 mg BID was well tolerated with a manageable safety profile, and showed activity in pts with high cMET status HCC. Confirmation of these initial positive results is warranted to establish the relevance of INC280 in HCC. Clinical trial information: NCT01737827.

4076

Poster Session (Board #68), Sat, 8:00 AM-11:30 AM

4075

Poster Session (Board #67), Sat, 8:00 AM-11:30 AM

TTP or PFS as a surrogate endpoint in advanced hepatocellular carcinoma treated with systemic therapy. First Author: Dae-Won Lee, Seoul National University Hospital, Seoul, Korea South Background: While overall survival (OS) is the most important survival endpoint in oncology clinical trials, it can be affected by sequential treatment after progression or by crossover from control to experimental arms and requires longer follow up duration compared to surrogate endpoints. Because death resulting from liver cirrhosis might confound the treatment outcome, time to progression (TTP) is suggested as a reliable surrogate endpoint compared to progression free survival (PFS) in the clinical trials of hepatocellular carcinoma (HCC). However, the correlation between TTP/PFS and OS has never been studied. Methods: We searched Pubmed and Embase data to obtain data source. Eligible studies were randomized controlled phase III trials which evaluated the efficacy of systemic chemotherapy or molecular targeted therapy in advanced HCC. The association between TTP/ PFS and OS was assessed by Spearman rank correlation coefficient (rs) and linear regression analysis. The association of treatment effects as shown by the hazard ratio (HR) of TTP/PFS and OS in each trial was also investigated. Results: Ten studies with a total of 20 treatment arms and 6689 patients were included. Median TTP/PFS was significantly correlated with median OS. The rs value between TTP/PFS and OS was 0.74 (95% confidence index (CI) 0.45 – 0.89, p , 0.001) and the corresponding R2 was 0.47. There was no difference in rs value (0.72 vs. 0.69) and R2 value (0.38 vs. 0.35) between TTP and OS, and PFS and OS. Incremental benefit from the study treatment in TTP/PFS from each trial was correlated with incremental benefit in OS. The rs value and R2 value between log (HRTTP/PFS ) and log (HROS ) was 0.76 (95% CI 0.25 – 0.94, p = 0.009) and 0.61. Conclusions: Our study results suggest that either TTP or PFS could serve as a surrogate marker for OS in the clinical trials of advanced HCC. TTP was not a superior surrogate marker compared to PFS. In addition, this trial level analysis shows that the gain of TTP/PFS leads to the gain of OS.

4077

Poster Session (Board #69), Sat, 8:00 AM-11:30 AM

Platinum rechallenge in patients with advanced biliary tract carcinoma (ABTC) after failure of gemcitabine (GEM)-platinum combination: A national AGEO retrospective study. First Author: Cindy Neuzillet, Henri Mondor University Hospital, Creteil, ´ France

Final results of a phase III randomized controlled trial comparing modified gemcitabine + oxaliplatin (mGEMOX) to gemcitabine+ cisplatin in management of unresectable gall bladder cancer (GBC). First Author: Atul Sharma, All India Institute of Medical Sciences, New Delhi, India

Background: There is no validated option beyond GEM-platinum standard 1st-line (L1) chemotherapy (CT) for ABTC. In ovarian and colon cancers, platinum rechallenge demonstrated benefit in patients who responded to a prior platinum-based CT and with a significant platinum-free interval. We aimed to assess the usefulness of platinum-based CT as 2nd (L2) or 3rd-line (L3) treatment in ABTC. Methods: Patients with ABTC who received platinum-based CT as L2 or L3 were identified from a national retrospective AGEO cohort of ATBC patients receiving L2 CT after failure of GEM-platinum L1 CT at 18 French institutions. Progression-free survival (PFS) and overall survival (OS) were estimated by the Kaplan-Meier method. Results: Among 686 patients who received GEM-platinum L1 CT, 231 were given L2 CT; among them, 92 (39.8%) received a platinum-based CT as L2 (n = 70) or L3 (n = 25). Median age was 65 years (range: 25-81); 50 (54.3%) patients were men and 42 were women. The primary tumor was intrahepatic, extrahepatic and gallbladder carcinoma in 44/28/19 patients. L1 CT was GEMOX in 83 (90.2%) patients and GEMCIS in 9. At the time of platinum rechallenge, 82 (92.1%) patients had metastatic disease; ECOG PS was 0-1 in 73.6%; CA19.9 was . 400 IU/ml in 36.2%. Rechallenge regimens were 5FUcisplatin (n = 53), FOLFOX/XELOX (n = 28), GEMOX/GEMCIS (n = 9), and carboplatin-based CT (n = 5). Overall, median L1 PFS (mPFS) was 7.2 months and median OS (mOS) from L1 was 17.4 months; mPFS and mOS from platinum rechallenge were 4.4 months (L2: 4.8; L3: 3.4) and 7.7 months (L2: 7.7; L3: 7.7), respectively. 82 patients were evaluable according to RECIST criteria; 12 (14.6%) had partial response and 40 stable disease (48.8%) (disease control rate [DCR] 63.4%). PS (0-1 vs 2-3) and CA19.9 (# vs . 400 IU/ml) were significantly associated with longer PFS (HR 2.41, p = 0.0016 and HR 1.88, p = 0.0350); PS, CA19.9, and reason for L1 discontinuation (toxicity/other vs progression) were significantly associated with longer OS (HR 2.17, p = 0.0019; HR 1.69, p = 0.0491; HR 1.77, p = 0.0122, respectively). Conclusions: Platinum rechallenge is an interesting option yielding a DCR over 60% in patients with ABTC.

Background: Combination of gemcitabine and platinum (cisplatin/ oxalipatin) though standard for advanced/unresectable GBC has not been directly compared. This study was designed to see if OS with mGEMOX is equivalent to GemCis. Methods: Single center phase III randomized trial. Primary end point was OS in 2 groups. Main Secondary end points were: 1. PFS in 2 groups; 2. RR in two groups. Considering median OS of 9.5 month in mGEMOX (as per our previous study) and 11.7month in GemCis 108 patients were required in each arm to have g 62 month with 80% power. To account for protocol violation and lost to follow up additional 22 patients in each arm were enrolled. Treatment protocol: Arm A- mGEMOX Oxaliplatin 80 mg/m2 and Gemcitabine 900 mg/m2 day1 and 8 maximums of 6 cycles, Arm B- GemCis Cisplatin 25 mg/m2 Gemcitabine 1000 mg/m2 day1 and 8 maximums of 8 cycles, repeated every 3 weeks. Results: Between February 2010 and July 2015, 260 patients with unresectable/metastatic GBC subjects were randomly assigned in a 1:1 ratio to mGEMOX or GemCis. 243 patients (119 in mGEMOX and 124 in GemCis) received at least 1 dose of chemotherapy and were considered analyzable. Subjects in 2 groups were well matched for baseline characteristics. Median number of cycles was 4 in both the groups. 33.4% in mGEMOX and 25.8% in GemCis completed the assigned treatment. Median follow up of surviving patients is 10.5 months (range 4.5-47). Median OS in mGEMOX is 9 months (95% CI = 7.77-10.22) and 8 months in GemCis arm (95% CI = 7.40-8.59); p = 0.152. Median PFS in mGEMOX is 6 months (95% CI = 4.72-7.27) and 4.5 months in GemCis arm (95% CI = 3.44-5.55); p = 0.123. 29.4% (35) and 18.5% (23) patients survived 12 month or longer in mGEMOX and GemCis arms respectively (p = 0.442). 23.5% and 22.6% subjects achieved objective CR/PR in mGEMOX and GemCis arms respectively. Toxicity was comparable except more grade 3/4 diarrhea, and grade 1/2 peripheral neuropathy in mGEMOX arm. Conclusions: This study confirms that 6 cycles of mGEMOX is equivalent to 8 cycles of GemCis and one of these may be chosen for initial treatment. Clinical trial information: CTRI/2010/091/001406.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Noncolorectal) Cancer 4078

Poster Session (Board #70), Sat, 8:00 AM-11:30 AM

4079

225s Poster Session (Board #71), Sat, 8:00 AM-11:30 AM

Safety and antitumor activity of nivolumab (nivo) in patients (pts) with advanced hepatocellular carcinoma (HCC): Interim analysis of doseexpansion cohorts from the phase 1/2 CheckMate-040 study. First Author: Bruno Sangro, Clinica Universidad de Navarra and CIBERehd, Pamplona, Spain

Next-generation sequencing (NGS) survey of biliary tract cancer (BTC) to reveal the association between tumor somatic variants and chemotherapy resistance. First Author: Daniel H. Ahn, The Ohio State University Comprehensive Cancer Center, Arthur G. James Cancer Hospital, Richard J. Solove Research Institute, Columbus, OH

Background: HCC tumors are associated with chronic inflammation that can promote an immunosuppressive environment; anti-PD-1 therapy may counter this inhibition. Nivo, a fully human IgG4 monoclonal antibody PD-1 inhibitor, was initially evaluated in a multiple ascending-dose, phase 1/2 study in pts with advanced HCC; it was well tolerated with antitumor activity in different etiologies, across lines of therapy, justifying an expansion phase. Interim results are presented. Methods: Pts had histologically confirmed, advanced HCC and Child-Pugh class A. Dose expansion at nivo 3 mg/kg occurred in 4 cohorts: uninfected sorafenib (sor) na¨ıve/intolerant, uninfected sor progressors, HCV-, and HBV-infected. Primary endpoint was confirmed overall response rate (ORR) by RECIST 1.1. Secondary endpoints included OS, PFS, time to progression, and biomarker assessment. Results: Dose expansion enrolled 206 pts; 75% had extrahepatic metastasis, 7% vascular invasion, and 64% prior sor. Across cohorts, pts received a median of 5–6 doses (range: 1–19). Treatment-related AEs (TRAEs) occurred in 104 pts (50%); the most frequent were fatigue (17%) and pruritus (12%). Grade 3/4 TRAEs were seen in 28 pts (14%); most common were ALT and AST increases (3% each). 68 of 174 evaluable pts (39%) had a decline in tumor burden. Preliminarily, 91 pts (55%) had $ 18 wks follow-up and/or PD. ORR for these pts was 9% (8/91) [14% (3/22) uninfected sor na¨ıve/intolerant; 7% (2/27) uninfected sor progressors; 14% (3/21) HCV-infected; (0/21) HBVinfected]. 6 mos OS rate was 69% (95% CI, 0.43–0.85). Responses were observed in pts with and without quantifiable PD-L1 measured by IHC. Antiviral responses in HCV- and HBV-infected pts have been observed as measured by declines in HCV RNA and quantitative HBV surface antigen. Conclusions: AEs were consistent across nivo cohorts and similar to profiles in other tumor types. Results are preliminary and may underestimate response; data indicate activity across all etiologic subtypes and lines of therapy, supporting ongoing study of nivo in HCC. Clinical trial information: NCT01658878.

Background: BTC are uncommon and associated with a dismal prognosis. Gemcitabine and platinum-combinations (GP) form the standard approach for treating advanced BTC with a modest improvement in survival. To identify potential biomarkers for response to GP in BTC, we used NGS to evaluate the genomic BTC landscape and identify whether mutations affecting DNA repair were associated with GP resistance. Methods: Pretreatment formalinfixed, paraffin-embedded (FFPE) samples from 183 patients (pts) with BTC treated with GP were analyzed with a commercial targeted NGS platform. Genes with incidence . 10% were included in our analysis. Cox regression models were used to determine the association between mutations, progression free survival (PFS) and overall survival (OS). Survival was estimated using the Kaplan Meier method and compared by the log-rank test. Results: Of the 183 pts analyzed, 181 (99%) had at least one actionable mutation. The most common mutations included CDKN2A (29%), TP53 (27.9%), KRAS (22.4%), and ARID1A (13.7%). Considering genes with an incidence . 10%, no individual gene was independently predictive of GP response. In pts with unresectable BTC who received GP as first-line therapy, the joint status of CDKN2A, TP53 and ARID1A were associated with PFS (P = 0.0004) and OS (P = , 0.0001). Pts with mutations in CDKN2A and TP53 with wild-type ARID1A were identified as a poor prognostic cohort with a median PFS of 2.63 months (mos) and OS 5.22 mos. Pts with mutant ARID1A regardless of the single mutational status of TP53 or CDKN2A had statistically similar PFS and OS. In a single pt who exhibited mutations in all three genes, the median PFS was 20.37 mos and median OS not reached. Conclusions: In the largest exploratory analysis of this nature to date in BTC, we found that the presence of three prevalent and mutually exclusive mutations (TP53, CKDN2A and ARID1A) represent distinct pt cohorts. These mutations are prognostic and may represent a predictive biomarker to GP response. Prospective studies are needed to validate these findings, including the incorporation of novel agents that exploit the genomic instability observed with these mutations with GP in BTC.

4080

4081

Poster Session (Board #72), Sat, 8:00 AM-11:30 AM

Poster Session (Board #73), Sat, 8:00 AM-11:30 AM

Impact of integrating insulin like growth factor-1 (IGF-1) in model for end stage liver disease (MELD) scores on predicting hepatocellular carcinoma patients’ survival. First Author: Reham Abdel-Wahab, Clinical Oncology Department, Assiut University, Assiut, Egypt

Integrating insulin-like growth factor-1 (IGF-1) score into Barcelona Clinic Liver Cancer (BCLC) and Cancer of the Liver Italian Program (CLIP) scores. First Author: Reham Abdel-Wahab, Clinical Oncology Department, Assiut University, Assiut, Egypt

Background: The status of the liver reserve in hepatocellular carcinoma (HCC) affects patient’s survival. In clinical practice, models for end-stage liver disease (MELD) score is used to predict overall survival (OS). Our aim is to evaluate whether integrating IGF-1 can provide additional value to MELD and sodium (Na)-MELD scoring systems in predicting patients’ OS. Methods: Plasma IGF-1 was measured in both training (n = 310) and validation (n = 155) cohorts from MD Anderson Cancer Center. We created MELD-IGF and Na-MELD-IGF scores from training cohort then we validate our results. Cox models were used to determine the association of IGF-1 with OS after adjustment for MELD and Na-MELD. Harrell’s C-index and Ustatistics were used to compare the predictive capacity of MELD, Na-MELD, MELD-IGF, and Na-MELD-IGF. Results: IGF-1 was significantly associated with OS after adjusting for MELD or Na-MELD in both cohorts. Patients with IGF # 26ng/ml had significantly higher hazard ratio (HR) compared to patients with the same MELD or Na-MELD but with IGF-1 . 26ng/ml among the training cohort (Table 1). MELD-IGF score had a higher C-index (0.59 and 0.72) compared to MELD score (0.56 and 0.66) (P = 0.017 and 0.011), among training and validation cohort respectively. While, the C-index for MELDNa-IGF was 0.633 compared to 0.629 for MELDNa score in the training cohort, the C-index for MELDNa-IGF and MELDNa scores were 0.72 and 0.66 (P = 0.31 and 0.003) respectively in the validation cohort. Conclusions: Our study indicate that MELD-IGF score is a significant predictor for patients’ OS after being validated into two independent cohorts. Future studies are warranted for international validation of this score.

Background: Barcelona Clinic Liver Cancer (BCLC) and Cancer of the Liver Italian Program (CLIP) Staging are the only systems that include Child Tarcott Pugh score (CTP) to assess hepatic function. Recently, we developed and validated Insulin like growth factor-1 CTP (IGF-CTP) score which possessed a higher ability in predicting patients’ overall survival (OS) than CTP. Our current study aimed at evaluating whether integrating IGF-CTP into BCLC and CLIP can provide additional value in predicting OS. Methods: In both training (n = 310) and validation (n= 155) from MD Anderson Cancer Center, we used cox models to evaluate OS after adjustment for BCLC and CLIP parameters. Harrell’s C-index and U-statistics were used to assess the predictive capacity of the IGF-CTP-BCLC and IGF-CTP-CLIP. Results: In the training cohort, patients with IGF #26ng/ml had a shorter OS after adjusting for the effect of BCLC and CLIP with hazard ratio (HR) and 95% confidence interval (95%CI); [2.19 (1.6 – 2.98) and 1.52 (91.1 – 2.1)], respectively. A similar trend was found in the validation cohort and after adjusting for the effects of BCLC and CLIP parameters. C-index showed that integrating IGFCTP score was significant in CLIP but not in BCLC, while, adding IGF-1 (cutoff point; 26ng/ml) to both BCLC and CLIP showed a higher predictive ability. (Table1) Conclusions: Integrating plasma IGF-1 into BCLC and CLIP led to a significantly better prediction for patients’ OS in two independent cohorts. Integrating IGF-CTP was significant only in CLIP score. However, BCLC classes are more heterogeneous because it is a treatment allocation system not only a prognostic system, therefore, future studies of BCLC in this setting are warranted for international validation of these scores.

MELD and IGF

MELDNa and IGF

Models

Training cohort HR (95% CI)

Validation cohort HR (95% CI)

MELD IGF-1 ( # 26 vs . 26) MELD*IGF MELD-IGF MELDNa IGF-1 ( # 26 vs . 26) MELDNa*IGF MELDNa-IGF

1.08 (1.04 – 1.12) 1.96 (1.45 – 2.66) 1.197 (1.09 – 1.32) 1.07 (1.05 – 1.09) 1.08 (1.06 – 1.11) 1.93 (1.42 – 2.62) 1.13 (1.06 – 1.2) 1.07 (1.05 – 1.08)

1.12 (1.05 - 1.19) 2.31 (1.39 – 3.86) 0.92 (0.78 – 1.1) 1.05 (1.03 – 1.07) 1.09 (1.04 – 1.14) 2.35 (1.39 – 3.98) 0.97 (0.86 – 1.09) 1.04 (1.02 – 1.06)

CLIP IGF-CTP-CLIP IGF-CLIP BCLC IGF-CTP-BCLC IGF-BCLC

Training cohort C index (95% CI) [P value]

Validation cohort C index (95% CI) [P value]

0.705 (0.66 – 0.75) 0.715 (0.67 – 0.76) [0.006] 0.711 (0.67 – 0.75) [,0.0001] 0.64 (0.6 – 0.68) 0.64 (0.61 – 0.67) [0.2] 0.67 (0.63 – 0.71) [,0.0001]

0.7 (0.63 – 0.77) 0.72 (0.65 – 0.8) [0.015] 0.73 (0.65 – 0.81) [,0.0001] 0.54 (0.5 – 0.58) 0.57 (0.54 – 0.6) [0.3] 0.65 (0.58 – 0.73) [,0.0001]

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

226s 4082

Gastrointestinal (Noncolorectal) Cancer Poster Session (Board #74), Sat, 8:00 AM-11:30 AM

Adjuvant therapy for gallbladder carcinoma (GBCA): Analysis of the National Cancer Data Base (NCDB). First Author: Kalyan C. Mantripragada, Rhode Island Hospital, Providence, RI Background: GBCA has a poor prognosis even when it is resectable. Guidelines recommend consideration of adjuvant chemoradiation (ACRT) or chemotherapy (ACH), but without high level evidence. Our objective was to evaluate ACH and ACRT in a nationwide GBCA dataset, while addressing potential sources of bias. Methods: We selected patients (pts) with T2-3N0 or T1-3N1-2 (N+) GBCA reported in 2004-2011 to the NCDB—an oncology registry capturing ~70% of cancers in the United States (US). We excluded pts managed without surgery, with grossly positive margins or medical contraindications to chemotherapy. We defined ACH/ACRT as treatment initiated within 3 months of surgery. We addressed missing data by multiple imputation, immortal time bias by landmark analysis (excluding pts with , 4 months of follow-up), and indication bias by propensity score analysis (inverse probability of treatment weighting, IPTW, and 1:1 matching). We verified balance of all available confounders, and assessed sensitivity to unobserved confounding. Overall survival (OS) was compared in Cox models reporting hazard ratio (HR) with 95% confidence intervals (CI). Results: Among 4,802 eligible pts, 1,384 (29%) received ACH (57% single-, 33% multi-agent, 10% unspecified), and 651 (14%) ACRT. ACH was less frequently applied in pts . 70 years old, Hispanic women, and pts with comorbidities. It was more common in T3 or N+ tumors, tumors . 5 cm in size, or with microscopically positive margins. Unadjusted 3-year (y) OS was 51% for T2, 26% for T3 and 31% for N+ GBCA. There was no evidence of improved OS after application of ACH (absolute OS difference at 3 y: 0.3%; IPTW estimate of HR = 1.00, CI, 0.92-1.10, matching estimate: HR = 0.98, CI, 0.88-1.08), and no evidence of heterogeneity by stage (P= .24). The effect of ACRT was heterogeneous, with significantly improved OS for T3 or N+ GBCA (OS difference at 3 y: 3.2%; IPTW HR = 0.83, CI, 0.73-0.95; matching HR = 0.84, CI, 0.72-0.99). Conclusions: Up to 2011, ACH was applied to a minority of GBCA pts in the US community, and without an evident OS advantage. ACRT modestly improved OS in T3 or N+ disease. Regimens with known superior activity in metastatic GBCA (cisplatin/ gemcitabine) or novel approaches should be investigated.

4084

Poster Session (Board #76), Sat, 8:00 AM-11:30 AM

4083

Poster Session (Board #75), Sat, 8:00 AM-11:30 AM

Circulating biomarkers in a phase II study of hypofractionated proton beam therapy (H-PBT) for hepatocellular carcinoma (HCC) and intrahepatic cholangiocarcinoma (ICC). First Author: Theodore S. Hong, NSABP/NRG Oncology, and The Massachusetts General Hospital, Boston, MA Background: To explore the association of circulating biomarkers with outcome of patients treated with H-PBT for HCC and ICC. Methods: Pts enrolled at MGH in a phase II study (NCT00976898) of H-PBT for localized HCC (n = 21) or ICC (n = 20) (Hong, JCO 2016) had plasma biomarkers measured at baseline and weekly during PBT, and included growth factors, inflammatory cytokines, soluble receptors, and markers of hypoxia and fibrosis. Associations of PFS and OS were analyzed by dichotomizing at the observed median of baseline levels and reported as a hazard ratio (HR) estimated by proportional hazards regression, with exact p-values based on a 2-sided logrank test. Results: 2 yr OS and PFS was 49% and 30% for HCC pts and 32% and 17% for ICC pts. Baseline levels and changes of biomarkers of tumor microenvironment (hypoxia, fibrosis, inflammation) were significantly correlated with PFS and OS for HCC (Table, select OS sig values), but not ICC. However, baseline levels of plasma HGF were significantly correlated with OS of both ICC and HCC pts. Conclusions: The outcome of PBT inboth HCC and ICC was associated with baseline values and on-treatment changes in circulating biomarkers. While HGF data suggest a common mechanism of resistance for HCC/ICC, markers of microenvironment suggest differential mechanism, consistent with the different biology and outcomes. Clinical trial information: NCT00976898. HCC: OS Baseline HR

P

HCC: PFS

Post-PBT HR

Growth Factors HGF 8.7 0.001 0.2 PlGF 6.0 0.011 0.7 Ang-2 4.3 0.027 1.6 Soluble Receptors sMet 5.8 0.009 1.6 Inflammatory Cytokines IL-8 17.4 .0002 0.7 SDF1a 9.3 .0008 1.2 Hypoxia CAIX 7.0 0.009 NA Fibrosis TIMP1 22.4 .0002 0.9

4085

P

Baseline HR

ICC: OS

Post-PBT

P

HR

Baseline

P

HR

P

ICC: PFS

Post-PBT HR

P

Baseline HR

P

Post-PBT HR

P

0.236 2.4 0.115 0.05 0.044 3.3 0.043 0.5 0.351 1.5 0.419 0.7 0.476 0.619 3.0 0.048 0.9 0.832 1.7 0.303 0.5 0.231 2.1 0.159 0.5 0.173 0.745 2.5 0.087 1.1 0.859 1.0 0.968 3.4 0.068 0.9 0.899 0.8 0.667 0.503 3.0 0.036

1.2

0.800 1.7 0.329 0.8 0.781 0.8 0.750 1.3 0.693

0.536 4.4 0.007 0.785 2.7 0.062

0.5 1.2

0.274 1.7 0.321 0.5 0.285 0.7 0.511 1.2 0.751 0.823 2.2 0.136 0.6 0.525 1.1 0.922 0.2 0.074

0.061 4.9 0.006

6.2

0.042 1.5 0.507 0.8 0.735 1.5 0.444 0.7 0.639

0.857 6.0 0.003

1.3

0.681

Poster Session (Board #77), Sat, 8:00 AM-11:30 AM

Segmental yttrium-90 radioembolization versus chemoembolization for localized hepatocellular carcinoma. First Author: Siddharth A Padia, University of Washington, Seattle, WA

Comprehensive profiling of biliary tract cancers (BTC) to reveal molecular heterogeneity with implications for matched targeted therapies (MTT). First Author: Tamara Sauri, Vall d Hebron University Hospital, Barcelona, Spain

Background: Transarterial chemoembolization (TACE) remains the standard of care for localized, surgically unresectable hepatocellular carcinoma (HCC). However, yttrium-90 radioembolization (Y90) may yield high response rates when delivered to a single segment in ablative-level radiation doses. This study compared Y90 to TACE for localized HCC, with assessments of toxicity, tumor response, and survival. Methods: Over a 6-year period, 178 patients underwent treatment for 235 tumors with either Y90 or TACE performed to a single hepatic Couinaud segment. Under IRB approval, patient and tumor characteristics were tabulated. TACE was performed utilizing doxorubicin; Y90 was delivered using an ablative-level dosing strategy. Toxicity, index tumor and overall response based on modified RECIST criteria, and overall survival were compared. Results: 101 patients underwent segmental Y90 for 132 tumors. 77 patients underwent segmental TACE for 103 tumors. Age, gender, etiology of liver disease, degree of cirrhosis (Child-Pugh A 60%, B 34%, C 6%), and tumor stage (BCLC A 34%, B 17%, C 42%, D 7%) were similar between the groups. Y90 patients had higher rates of portal venous invasion or infiltrative tumor compared to TACE patients (18% v 1%, p,0.001; 23% v 9%, p=0.004, respectively). TACE trended towards higher rates of bilirubin toxicity (p=0.06). Toxicities ($ grade 3) were otherwise similar. Y90 demonstrated higher complete response rates for both the index tumor and overall compared to TACE, with longer times to local tumor progression. Survival was similar between the two groups. Conclusions: When performed in a segmental, ablative-level dose strategy, Y90 demonstrates higher complete tumor response rates and local tumor control when compared to TACE, with similar toxicity profiles. Despite a greater proportion of patients with highrisk tumor characteristics in the Y90 group, similar overall survival was achieved.

Background: BTC is an heterogeneous disease that includes intrahepatic cholangiocarcinoma (IHC), extrahepatic cholangiocarcinoma (EHC), gallbladder carcinomas (GBC) and ampulloma (AM), typically present an advanced stage at diagnosis and chemotherapy (chemo) has limited efficacy. We describe potentially targetable genomic alterations (ptGA) of BTC and impact of MTT in these patients (pts). Methods: From 2011 to 2015, 104 chemorefractory patients (pts) with BTC (62 IHC, 23 EHC, 15 GBC, 4 AM) were referred to molecular profiling at our institution. Archived tumor samples were analyzed using different panels (FoundationOne in 27, Amplicon-MiSeq in 25, MassARRAY-Sequenom in 23, Fusion-Nanostring in 31 and protein expression in 13) and pts were offered MTT in Phase 1 trials according to specific eligibility criteria and logistics. Results: Median age was 57 y (27-82); metastatic sites were liver (81%), nodes (43%) and lung (21%); median time to progression on first-line chemo was 5.1 months (m) (CI95% 4.1-6.3); 27% received 2 or more chemo lines. At least one ptGA was detected in 49% of the samples, with proportional increase according to the size and spectrum of gene alterations tested (p,0.001). Type of ptGA differed across tumor types (see Table). Additional ptGA in IHC (BRAF D594, ERBB2 S310, RNF43 and BRCA2 mut) and EHC (BRAF G469 and ERBB2 S310 mut) were found in single cases. A total of 22 pts were enrolled in Phase 1 trials, 13 of them with a direct MTT (5 PI3K, 2 MEK, 2 MET, 2 IDH1, 1 ERBB2 and 1 FGFR inhibitor [inh]). Partial response was observed in 1 case (MET ampl IHC with MET inh) and prolonged stable disease in 3 pts (1 FGFR2 fusion IHC with FGFR inh; 1 IDH1 mut IHC with IDH1 inh; 1 NRAS mut GBC with MEK inh). Conclusions: The diverse landscape of ptGA in BTC is a rich source for MTT in early clinical trials, and has the potential to improve outcomes of individual pts. Expanding gene panels to identify these driver alterations (mut, ampl and fusions) is key for precision medicine in BTC.

Category Index tumor response (%)

Overall response (%)

Local tumor progression (%) Overall survival (%)

CR PR SD PD CR PR SD PD 6 months 1 year 6 months 1 year 2 years

Y90

TACE

P-value

90 7 2 1 83 12 2 3 2 4 98 92 84

73 19 5 3 58 26 8 8 3 15 99 92 77

,0.001

,0.001

0.992 0.040 0.680 0.585 0.454

IDH1 mutation [mut] IDH2 mut FGFR2 fusion ERBB2 amplification [ampl] PIK3CA mut RAS mut MET ampl

IHC (n=62)

EHC (n=23)

GBC (n=15)

AM (n=4)

18% 9% 7% 4% 13% 11%

6% 12% 6% 24% -

10% 80% 30% 10% -

67% -

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Noncolorectal) Cancer 4086

Poster Session (Board #78), Sat, 8:00 AM-11:30 AM

mRECIST to predict survival in advanced hepatocellular carcinoma: Analysis of two randomized phase II trials comparing nintedanib versus sorafenib. First Author: Tim Meyer, UCL Cancer Institute, University College London, London, United Kingdom Background: Response Evaluation Criteria in Solid Tumors (RECIST) has been shown to be a poor surrogate for survival benefit with targeted therapy in advanced hepatocellular cancer (HCC). We investigated whether response evaluated using modified RECIST (mRECIST) predicted overall survival (OS) using comparable data from two Phase II trials conducted to evaluate the efficacy/safety of nintedanib versus sorafenib as first-line treatment of mainly Caucasian patients (study 1199.37; NCT01004003) or Asian patients (study 1199.39; NCT00987935) with advanced HCC. Methods: Data from patients (n=188) treated with either nintedanib or sorafenib were pooled for analyses. Cox regression and Kaplan–Meier survival analyses were used to explore differences in OS between the responders and non-responders according to RECIST 1.0 and mRECIST criteria. Two multivariate Cox proportional hazards selection models were conducted to identify baseline variables that predict survival and if RECIST response (analysis 1) and mRECIST response (analysis 2) predict survival at the 0.2 level of significance. Results: Response rates were 4.3% (n=8) by RECIST and 14.9% (n=28) by mRECIST. Discordance between RECIST/mRECIST evaluation was most common for assessment of partial response (n=22; 11.7%) and stable disease (n=24; 12.8%). OS was significantly longer in patients with response compared to patients without response; RECIST: hazard ratio (HR) 0.32 (95% CI 0.13–0.81), p=0.0113 and mRECIST: 0.53 (95% CI 0.33–0.85), p=0.0079. In both multivariate models, presence of macrovascular invasion (MVI) and presence of extrahepatic spread (EHS) at baseline were associated with worse OS. RECIST response and mRECIST response were both significant predictors of survival (see table). Conclusions: Objective mRECIST response is a surrogate for OS and as such should be considered a valid endpoint for use in HCC trials. Clinical trial information: NCT01004003 and NCT00987935.

Analysis 1 MVI absent EHS absent RECIST response Analysis 2 MVI absent EHS absent mRECIST response

4088

OS hazard ratio (95% CI)

p-value

0.72 (0.51‒1.01) 0.71 (0.49‒1.01) 0.40 (0.16‒1.01)

0.0571 0.0591 0.0514

0.76 (0.54‒1.07) 0.72 (0.50‒1.03) 0.61 (0.37‒0.99)

0.1094 0.0721 0.0431

Poster Session (Board #80), Sat, 8:00 AM-11:30 AM

Comprehensive molecular profiling and analysis of mutual exclusivity of genetic aberrations (MEGA) of intra- and extrahepatic cholangiocarcinomas (IHC and EHC) evaluation of prognostic features and potential targets for intervention. First Author: Maeve Aine Lowery, Memorial Sloan Kettering Cancer Center, Weill Cornell Medical College, New York, NY Background: Varying driver genetic aberrations have been identified among different anatomic and clinical subtypes of IHC and EHC, with suggestive prognostic and/or predictive benefit and potential therapeutic implications. Methods: Archival FFPE samples from patients who consented prospectively to study were analyzed using the MSK-IMPACT platform, a hybridization capture based next generation sequencing assay for targeted deep sequencing of all exons and selected introns of 410 key cancer genes. Tumor and matched normal libraries were sequenced on an Illumina HiSeq 2500. Sequencing output was processed using a custom analysis pipeline. Demographic and treatment data was prospectively collected. Fisher’s exact tests were performed to look for MEGA and associations between clinical characteristics and genetic alterations. Results: 137 pts were studied: 78% EHC and 22% IHC; 50% male, from metastatic sample (70%) and primary (30%). Median mutations per sample: 3, median sample coverage was 680X. Most commonly altered genes in IHC were IDH1 (30%), ARID1A (23%) BAP1 (20%), p53 (20%) and FGFR2 gene fusions (14%), Tendency towards MEGA was seen between multiple genes in IHC including TP53: IDH1 (p = 0.01), IDH1:KRAS (p = .03), TP53:BAP1 (p = 0.03), IDH1: FGFR2 (p = 0.04). Genetic alterations more common to EHC versus IHC included KRAS, SMAD4, STK11 and p53 (p , 0.05). FGFR2 fusions and BAP1 alterations were not identified in EHC. 120 pts received 1st line chemotherapy for advanced disease (77% gemcitabine/platinum), median TTP was 9.1 months and did not differ for pts with IDH1 mutations or FGFR2 gene fusions. Conclusions: Distinct patterns of genetic alterations and MEGA both within IHC and between IHC and EHC were identified. Response to 1st line chemotherapy did not differ between different molecular subgroups; these prospective data provide a contemporary benchmark for better understanding and development of targeted therapies in molecularly profiled IHC and EHC.

4087

227s

Poster Discussion Session; Displayed in Poster Session (Board #79), Sat, 8:00 AM-11:30 AM, Discussed in Poster Discussion Session, Sat, 3:00 PM-4:15 PM

Stereotactic body radiotherapy (SBRT) as an alternative to transarterial chemoembolization (TACE) for hepatocellular carcinoma (HCC). First Author: Eli Sapir, Department of Radiation Oncology, Hadassah Hebrew University Medical Center, Jerusalem, Israel Background: TACE is a standard treatment for patients (pts) with HCC. SBRT is a newer, noninvasive therapy. There have been no comparative studies to date. Thus, we examined TACE and SBRT outcomes. Methods: After IRB approval, institutional HCC and radiotherapy databases were queried for pts receiving TACE or SBRT for 1-2 tumors and combined with a diagnosis and billing code query for HCC, TACE, and SBRT. Pts with main branch portal venous (PV) involvement, extrahepatic spread, and Child Pugh C cirrhosis were excluded, since they are not candidates for TACE. Inverse probability weighting, via a propensity score, was used to adjust for imbalances between treatment groups. Local control (LC), overall survival (OS), and toxicity were compared. LC for was defined as no tumor growth within or immediately adjacent to the TACE cavity or original tumor. Results: From 2006-2014, 84 pts with 114 tumors were treated with TACE and 125 pts with 173 tumors with SBRT. Median follow-up was 28 months (1.8-103) and 16 months (0.2-98) for pts treated with TACE or SBRT, respectively (p,0.001). Pts treated with TACE were younger (61 vs 65 yrs, p=0.01) and had slightly larger tumors (2.9 vs 2.3 cm, p,0.001). In propensity adjusted analysis, 1and 2-yr LC favored SBRT: 97% and 91% for SBRT and 41% and 18%, for TACE (HR 18.8, 95% CI 6.7-52.7, p,0.001). Increasing tumor size (HR 1.2 per cm, 95% CI 1.05 - 1.23, p,0.001) and partial PV tumor thrombus (HR 7, 95% CI 2.73 - 15.2, p,0.001) predicted for worse LC in pts treated with TACE, but not with SBRT. Grade 3+ toxicity occurred after 14% and 7% of TACE and SBRT treatments, respectively (p=0.05). From HCC diagnosis, SBRT was initiated a mean of 9 months later than TACE (p,0.001), and more patients underwent liver transplantation after TACE (18% vs. 8%, p=0.01). After adjustment for baseline liver function and transplantation, overall survival was not significantly different (HR = 0.73, 95% CI 0.48 – 1.12, p =0.15). Conclusions: SBRT is a safe alternative to TACE for 1-2 tumors, and provides better LC, with no difference in OS. Prospective comparative trials of TACE, SBRT, and other ablative therapies are warranted.

4089

Poster Session (Board #81), Sat, 8:00 AM-11:30 AM

External validation of a prognostic score in patients (pts) with high-grade gastrointestinal neuroendocrine carcinomas (GI-NECs). First Author: Angela Lamarca, Department of Medical Oncology, The Christie NHS Foundation Trust, Manchester, United Kingdom Background: Prognostic markers for risk-stratification of pts with GI-NECs are lacking. We aimed to externally validate our previously-designed prognostic score (Training Cohort [TC]) derived from 109 pts (Lamarca et al, ASCO 2015). The score included five variables (presence of liver metastases, alkaline phosphatase (ALK), lactate dehydrogenase (LDH), ECOG performance status (PS) and Ki67) and was able to prognosticate overall survival (OS). Methods: Pts diagnosed with GI-NECs were identified retrospectively from 5 European centres (April’00 – Dec’15); pts with data for all 5 variables identified from the TC were eligible for the Validation Cohort (VC). To externally validate the score (power: 80% and two-tailed a-error: 0.05), a minimum of 82 pts was required. The primary end-point was OS; Cox regression and Kaplan-Meier estimates were applied; and univariate/ multivariable analyses performed. Results: Of 209 pts identified; 178 were eligible. The median follow-up time was 11.8 months (range 0.2167.4); 121 pts (68%) had died at time of the analysis. Median age: 58.7 yrs (range 22.2-85.5); 62% were male, 81% metastatic, 18% foregut, 3% midgut, 41% pancreas, 17% hindgut and 21% unknown primary. The median Ki67 was 55% (21-100), ECOG PS 0: 35%, 1: 46%; 66% received chemotherapy. The baseline median ALK and LDH were 133 IU/l (30-2363) and 347 IU/l (107-3786), respectively; there were significant differences (p , 0.05) in Ki67 and LDH between TC and VC. The estimated median OS was 16.1 months (95%CI 14.6-19.4). Both score (p , 0.001) and stage (p = 0.002) were significant in the univariate analysis. On multivariable analysis (adjusted for stage), the score was prognostic for OS (HR 1.4, 95%CI 1.11.6; p = 0.001) validating previous TC results. The score classified pts into different groups with incremental risk of death: groups A/B (0-2 points; 117 pts (66%); median OS 20.7 months [95%CI 17.2-28.1]) and C/D (3-6 points; 61 pts (34%); median OS 8.1 months [95%CI 4.8-9.4]). Conclusions: This international external series validates the capability of our prognostic score to predict OS in pts with GI-NECs; these results may inform clinical decision-making and trial design.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

228s 4090

Gastrointestinal (Noncolorectal) Cancer Poster Session (Board #82), Sat, 8:00 AM-11:30 AM

Everolimus (EVE) in advanced, nonfunctional, well-differentiated neuroendocrine tumors (NET) of gastrointestinal (GI) or lung origin: Second interim overall survival (OS) results from the RADIANT-4 study. First Author: James C. Yao, The University of Texas MD Anderson Cancer Center, Houston, TX Background: TheRADIANT-4 study met its primary end point with a median progressionfree survival (PFS) of 11.0 months (mo) for EVE vs 3.9 mo for placebo (PBO; HR = 0.48 [95% CI 0.35-0.67]; P , .00001) in patients (pts) with advanced, progressive, welldifferentiated NET of lung or GI origin. A planned first interim OS analysis favored EVE but did not cross the boundary for significance (HR = 0.64; 95% CI 0.40-1.05; P= .037). We report survival results from preplanned second interim OS analysis. Methods: 302 pts with advanced, progressive, well-differentiated, nonfunctional lung/GI NET were randomized (2:1) to EVE (10 mg/d) or PBO, each with best supportive care. Pts were stratified by tumor origin, WHO performance status, and prior somatostatin analogue use. Primary end point was PFS; OS was the key secondary end point. OS was estimated using the Kaplan–Meier method and the treatment groups were compared using a one-sided log-rank test. The Lan-DeMets method with O’Brien-Fleming type stopping boundary was used to control the cumulative type I error rate. HR was estimated by a stratified Cox proportional hazards model. Results: By data cutoff (Nov 30, 2015), 101 pts died; 66 of 205 (32%) in the EVE group and 35 of 97 (36%) in the PBO group. Median duration of study follow-up was 33.4 mo. EVE was associated with a 27% reduction in the estimated risk of death compared to PBO (HR = 0.73; 95% CI 0.481.11; P = .071 [the P-value threshold to claim significance for OS was 0.001982]). Estimated survival rate at 2 years was 77% with EVE and 62% with PBO (Table). Conclusions: The findings from this second interim OS analysis also suggested a trend for survival benefit with EVE, although statistical significance was not achieved. The final OS analysis will be performed when approximately 191 deaths have occurred. Clinical trial information: NCT01524783.

Estimated OS rates. Kaplan-Meier Estimates (95% CI) at 6 mo 12 mo 18 mo 24 mo 30 mo

4092

EVE 10 mg/d (n = 207) 94.9 88.9 81.5 76.9 67.4

(90.7-97.2) (83.5-92.6) (75.1-86.4) (70.0-82.4) (59.8-73.8)

PBO (n = 97) 90.3 82.2 73.5 61.5 58.8

(82.2-94.8) (72.6-88.7) (62.7-81.6) (50.0-71.1) (47.2-68.7)

Poster Session (Board #84), Sat, 8:00 AM-11:30 AM

Everolimus in combination with Octreotide LAR in first line setting for patients with neuroendocrine tumors (I.T.M.O. study): A 5-years update. First Author: Emilio Bajetta, Istituto di Oncologia, Policlinico di Monza, Monza, Italy Background: We previously presented data of this phase II study showing that the combination of Everolimus and Octreotide LAR for advanced neuroendocrine tumors (NETs), in the first line setting, is an active and safe treatment. At a median follow-up period of 277 days, the median time to progression (TTP) and the overall survival (OS) have not been reached (ASCO 2013 Abs. 4136). Methods: We performed a phase II, multicenter trial using a Simon two-stage minmax design. Patients with advanced well differentiated, previously untreated neuroendocrine tumors of the gastroenteropancreatic (GEP) tract and of the lung, received Octreotide LAR 30 mg every 28 days in combination with Everolimus 10 mg per day, continuously. The primary endpoint was objective response rate (ORR). Currently we performed an analysis of “long responder” patients and of OS and TTP after 5 years. Results: A total of 50 patients (58% males) were enrolled; 17 (34%) of these patients have received treatment for more than 2 year. The median exposure to study drugs is 519.5 weeks (range 48-2024). Currently 3 patients are still taking advantage from the above mentioned treatment. The primary tumor site was: liver 2 pts, pancreas 11 pts, small intestine 8 pts, lung 7 pts and unknown in 8 pts About 70% of these patients had no carcinoid Syndrome and 50% received surgery at the primary tumor site. The ORR (RP+RC) was 18%: CR 1 pt, PR 8 (22%) pts, SD 27 (75%) pts. The median TTP is 33.6 months (95% CI 18,7 – 41,2). The median OS is 61,9 months (95% CI 49,8 – n.d.). Conclusions: Everolimus in combination with octreotide LAR has shown to be active in advanced NETs. The current analysis showed a further prolongation of TTP, and a long exposure to the study drug without major side effects in the long term. The treatment with two drugs has induced objective responses, but also long duration SD. Because of scarce therapeutic options, phase III studies for this combination are desirable to improve the therapeutic possibilities for this disease. Acknowledgements: Italian Trials in Medical Oncology (I.T.M.O.) group, Giacinto Facchetti Foundation and Novartis.

4091

Poster Session (Board #83), Sat, 8:00 AM-11:30 AM

The clinicopathologic features and treatment of 607 hindgut neuroendocrine tumor (NET) patients at a single institution. First Author: Seung Tae Kim, Samsung Medical Center, Seoul, South Korea Background: The clinicopathologic features of hindgut neuroendocrine tumor (NET) as well as the treatment outcomes are not well known. There are currently no published data on treatment outcomes for patients with metastatic hindgut NET. The aim of this study was to conduct a comprehensive analysis of clinicopathologic features, treatments and survival in hindgut NET patients. Methods: Among patients who were pathologically diagnosed with hindgut NET at Samsung Medical Center between March 2001 and February 2015, 607 were analyzed in this study. Hindgut NETs were defined as NETs that originated from the transverse and distal colon, rectum, and anus. Results: Primary sites included 81 colon (13.3%) and 526 rectum (86.7%). According to the WHO classification, 578 patients (95.2%) had grade 1 NETs, 17 (2.8%) grade 2 NETs, and 12 (2.0%) had neuroendocrine carcinoma (NEC). Forty-two patients (6.9%) had extensive disease, while the majority (93.1%, 565 patients) only exhibited localized disease. The fiveand ten-year survival rates of 565 localized NET patients were 98.1% and 95.3%, respectively. The median OS in 42 patients with extensive disease was 24.8 months (95% CI, 10.7-38.8). Among 565 patients with localized disease, the majority (484 patients, 85.7%) were treated with endoscopic procedure by gastroenterologists. For 42 patients with extensive disease, 17 patients were managed by supportive care, 3 by concurrent chemoradiotherapy (CCRT), and 22 by systemic therapy. Among these 22 patients, 12 patients received only first-line therapy, 8 had second-line, and only 2 patients had third-line therapy. As first-line chemotherapy, the most commonly used regimens were etoposide plus cisplatin (N = 7) and long acting octreotide (N = 7). During treatment courses, the most commonly used regimen was long-acting octreotide. The median OS in 22 metastatic NET patients receiving systemic therapy was 19.3 months (95% CI, 3.2-35.3). Conclusions: This analysis provides useful information about the clinicopathologic features, treatments and survival outcomes for hindgut NET patients.

4093

Poster Session (Board #85), Sat, 8:00 AM-11:30 AM

Association of disease progression, health-related quality of life (HRQoL), and utility in patients (pts) with advanced, nonfunctional, well-differentiated gastrointestinal (GI) or lung neuroendocrine tumors (NET) in the phase 3 RADIANT-4 trial. First Author: Simron Singh, Odette Cancer Center, Sunnybrook Health Sciences Centre, Toronto, ON, Canada Background: Post hoc analyses were performed to determine if disease progression is associated with decline in HRQoL and utility scores using data from RADIANT-4, a phase 3 trial that showed significantly prolonged progression-free survival (PFS) with everolimus + best supportive care (BSC) vs placebo + BSC in pts with advanced, progressive, nonfunctional GI or lung NET. Methods: Pooling data from both arms, 284 pts were analyzed from baseline to study end. HRQoL was measured with FACT-G, a validated questionnaire with 4 domains: physical (PWB), social/family (SWB), emotional (EWB), and functional wellbeing (FWB). FACT-G was completed at baseline, every 8 wks until mo 12 after randomization, and every 12 wks thereafter. Association between disease progression and HRQoL outcomes was assessed by fitting linear mixed models. Based on a review of existing mapping functions and relevance for the RADIANT-4 population, 2 mapping algorithms were selected to translate FACT-G scores into EQ-5D utility scores: Young, Med Decis Making 2015 (UK value set); Teckle, Health Qual Life Outcomes 2013 (US value set). Results: The difference in FACT-G total score pre- vs postprogression was significant: 79.7 vs 74.8 (difference: 4.91; 95% CI: 3.71, 6.11). This difference may also be clinically relevant based on published ranges for minimal important difference (Yost & Eton, Eval Health Prof 2005). Differences in subscale scores were: PWB 22.4 vs 20.9 (1.5; 95% CI: 1.05, 1.95); EWB 17.6 vs 16.4 (1.14; 95% CI: 0.78, 1.49); SWB 21.6 vs 20.9 (0.69; 95% CI: 0.24, 1.14); and FWB 18.2 vs 16.9 (1.34; 95% CI: 0.86, 1.82). Mean “Teckle” utility was 0.826 (95% CI: 0.815, 0.836) preprogression and 0.795 (95% CI: 0.783, 0.807) postprogression; mean “Young” utility was 0.779 (95% CI: 0.763, 0.796) preprogression and 0.725 (95% CI: 0.706, 0.744) postprogression. Conclusions: Disease progression in pts with advanced, nonfunctional, welldifferentiated GI or lung NET is associated with a significant decline in HRQoL and utility scores. Effective therapy to prolong PFS may delay a decline in HRQoL and utility.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Noncolorectal) Cancer 4094

Poster Session (Board #86), Sat, 8:00 AM-11:30 AM

4095

229s Poster Session (Board #87), Sat, 8:00 AM-11:30 AM

Curative resection in digestive neuroendocrine neoplasms: Recurrence-free survival rate and definition of a risk score for recurrence. First Author: Elettra Merola, Digestive and Liver Diseases Unit, Sant’Andrea Hospital, Rome, Italy

Phase I, multi-center, open-label, dose-escalation study of pasireotide LAR (PAS) in patients with advanced neuroendocrine tumors (NET). First Author: Jonathan R. Strosberg, H. Lee Moffitt Cancer Center & Research Institute, Tampa, FL

Background: Surgery with radical intent is the only curative option for digestive neuroendocrine neoplasms (DNENs), but clinical practice shows disease-free patients recurring even after years. Our aims were to determine, in a consecutive series of DNENs treated by radical surgery, the recurrencefree survival (RFS) rate (primary end point), and to define a risk score for recurrence (secondary end point). Methods: Retrospective analysis of sporadic pancreatic (PNEN) or small intestine NEN (SINENs) who were treated by R0/R1 surgery between 1993 and 2014, and had a minimum 12-month follow-up after resection. Patients were classified according to the ENETS TNM/grading system. Survival analysis was performed by Kaplan-Meier and comparisons with log-rank, while risk factor analysis was carried out by Coxregression model; cut-off values were identified by ROC curves. Results: 215 patients (91 PNENs, 124 SINENs) were analyzed, with a median ki67 of 3% (range 1%-20%); 17.2% of them had a stage IV disease at diagnosis. Median RFS of the overall population was 61 months (5-yr rate: 49.5%). Four "risk factors" for recurrence were significant at multivariate analysis: PNEN vs SINEN (HR 1.44; P = 0.04), ki67 . 4% (HR 1.71; P , 0.01), stage IV since diagnosis (HR 2.09; P , 0.01), T parameter . 2 (HR 2.54; P , 0.01). Three "risk categories" for recurrence were identified: low risk (0 risk factors; median RFS 118 months), moderate risk (1-2 risk factors; median RFS 61 months), high risk (3-4 risk factors; median RFS 19 months; P , 0.01). Comparing the categories by Cox-regression method, results showed: HR "moderate vs low risk" 2.99, P = 0.01; HR "high vs moderate risk" 2.80; P , 0.01; HR "high vs low risk" 9.39; P , 0.01. Conclusions: Almost 50% of DNENs treated by curative surgery recur in 5 years. In our population, major risk factors for recurrence were tumor primary site, ki67, presence of metastases at diagnosis and T parameter. "Risk categories" might help in selecting patients who may benefit from adjuvant treatments.

Background: Pasireotide, a novel somatostatin analog, has been previously investigated but the maximum tolerated dose (MTD) has not been determined in pts with advanced NET. We report results of a planned interim analysis of a phase I dose-escalation (DE) study to determine the MTD, characterize safety, tolerability, PK, and efficacy trends in pts with advanced NET. Methods: Pts were enrolled in 2 phases: DE phase (to determine the MTD) at a starting dose of 80mg PAS i.m. followed by a dose expansion (DX) phase (evaluate safety and preliminary efficacy). Associations between PK parameters and clinical outcomes (probability of bradycardia, % tumor shrinkage, and glucagon levels) were evaluated using regression analysis. Bradycardia is defined as heart rate , 40 bpm. Results: As of July-2015, 29 pts (15, DE; 14, DX) were treated with 80 mg (13pts) and 120 mg (16pts) doses. No protocol defined dose-limiting toxicities were observed in the study; however in a post hoc analysis a higher incidence of bradycardia was seen with 120 mg (31.3%) vs 80 mg (0%). At data cut-off, 92.3% in 80 mg and 75% pts in 120 mg had discontinued treatment; primarily due to disease progression (44.8%) or adverse events (AEs; 27.6%). In the PK analysis, probability of bradycardia showed a moderate increase with increasing PAS concentration that was not statistically significant [Odds ratio for every 1.5x increase (95% CI): daytime, 1.672 (0.381-7.338); nighttime, 2.024 (0.494-8.292)]. No significant association was observed between PAS concentrations and glucagon levels. Two partial radiographic responses (PRs) were observed, both in the 120mg dose. PAS concentrations correlated with % tumor shrinkage although the association was not statistically significant (P= 0.08). The most common AEs in the 80 mg vs 120 mg were hyperglycemia (76.9% vs 81.3%), fatigue (53.8% vs 50%) and nausea (53.8% vs 31.3%). Conclusions: MTD was defined at 120 mg for PAS in pts with advanced NET. Although objective radiographic responses are rarely observed with SSA, 2 PRs were observed among 16 pts in the 120mg cohort. Bradycardia appears to be a dose-limiting effect, however the mechanism and clinical significance are uncertain. Clinical trial information: NCT01364415.

4096

4097

Poster Session (Board #88), Sat, 8:00 AM-11:30 AM

Exploratory analysis of tumor growth rate (TGR) with lanreotide depot/autogel (LAN) in patients (pts) with neuroendocrine tumors (NETs) from the CLARINET study. First Author: Martyn E. Caplin, Royal Free Hospital, London, United Kingdom Background: Antitumor activity of LAN in intestinal and pancreatic NETs was demonstrated in the phase 3 CLARINET study (NCT00353496), based on significantly prolonged progression-free survival (PFS) vs placebo (PBO). TGR has been proposed as a novel measure with additional insights into tumor response in several other tumor types. Here, we undertook an exploratory analysis of TGR in NET pts in CLARINET. Methods: Pts with metastatic intestinal/pancreatic NETs received LAN 120 mg or PBO every 4 wks for 96 wks. Target lesions were assessed by central radiologic review based on RECIST v1.0. TGR (% variation of tumor volume per month) was calculated from sum-of-longest-diameters (SLD) of original target lesions (excluding new ones) on 2 CT scans during defined periods: 12–24 wks prior to randomization vs baseline (pretreatment); and baseline vs each visit or between consecutive visits. TGR pre- and post-treatment were compared between treatment groups. Results: Almost all pts were classified as stable on RECIST after 12 wks treatment (LAN, 90/96; PBO, 94/98), while a large proportion exhibited TGR changes. An immediate reduction in mean TGR was observed after only 12 wks treatment with LAN but not with PBO, resulting in a significant difference in TGR between treatments which was maintained at subsequent study visits (Table). Conclusions: TGR seems to provide an early and more precise characterization of treatment activity than RECIST criteria in NET pts. The potential of TGR requires validation as a prespecified outcome measure in a prospective study. Clinical trial information: NCT00353496. TGR values at each study visit with LAN vs PBO. Treatment LAN, PBO, period (wks) N mean (95% CI) mean (95% CI) Pretreatment 0–12 12–24 24–36 36–48 48–72 72–96

200 4.1 (2.6, 5.6) 196 1.2 (-0.4, 2.7) 174 2.0 (0.5, 3.5) 154 1.8 (0.4, 3.2) 132 –0.3 (–1.7, 1.3) 108 1.3 (0.4, 2.1) 90 0.6 (–0.5, 1.7)

3.3 4.1 3.2 4.0 3.1 3.0 3.1

(1.7, (2.6, (1.7, (2.6, (1.5, (2.0, (1.7,

4.8) 5.6) 4.6) 5.4) 4.7) 4.0) 4.6)

Difference, mean (95% CI)

P-value

0.8 (–1.4, 3.0) –2.9 (–5.1, –0.8) –1.2 (–3.3, 0.9) –2.2 (–4.2, –0.3) –3.4 (–5.6, –1.1) –1.7 (–3.0, –0.4) –2.6 (–4.4, –0.7)

0.46 0.008 0.28 0.03 0.003 0.009 0.007

Poster Session (Board #89), Sat, 8:00 AM-11:30 AM

Functional status of neuroendocrine tumors among elderly patients: A large population-based study using SEER-Medicare data. First Author: Chan Shen, The University of Texas MD Anderson Cancer Center, Houston, TX Background: Functional neuroendocrine tumors (NETs) secrete bioactive amines and cause carcinoid syndrome, with its incumbent symptom burden of flushing and diarrhea. However, the frequency of carcinoid syndrome in the NET population has never been rigorously evaluated, nor has the relationship between patient characteristics and tumor function. The objective of our study was to assess the proportion of NETs patients with carcinoid syndrome, and identify clinical factors associated with tumor function. Methods: We identified NET patients diagnosed between 1/2000 and 12/ 2011 from the SEER-Medicare database, excluding those under age 65, enrolled in HMOs or without continuous enrollment in Medicare Parts A and B 3 months before and 3 months after the NET diagnosis. We identified patients with functional NETs by requiring that they had at least two claims indicative of carcinoid syndrome during the 6-month time window described above. We compared the demographic and clinical characteristics between patients with and without functional NETs using Chi-squared tests. We used the Cochran-Armitage trend test to test the trend in the proportion of patients with functional NETs. Results: Among the study cohort of 10,226 NET patients aged 65 and over, 1922 (19%) had carcinoid syndrome. The proportion of functional NETs reported increased significantly during the study period, growing from 10.7% in 2000 to 18% in 2011 (p , 0.001). Compared to patients without syndrome, patients with functional NETs were more likely to have regional/distant stage than localized stage (p , 0.001), grade 1 or 2 than grade 3 or 4 histology (p , 0.001), and have primary tumors in the small bowel or cecum as opposed to the lung or colorectum (p , 0.001). We also found that female patients and non-Hispanic white patients were more likely to report carcinoid syndromes (p , 0.001). Conclusions: This first population-based study identified a potential upward trend in the proportion of US Medicare patients over 65 years old reported to have functional NET. The study also demonstrated tumor functionality to be significantly associated with histology grade, stage, and primary site.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

230s 4098

Gastrointestinal (Noncolorectal) Cancer Poster Session (Board #90), Sat, 8:00 AM-11:30 AM

Circulating neuroendocrine transcripts and gene cluster analysis to predict the efficacy of peptide radioreceptor therapy. First Author: Lisa Bodei, European Institute of Oncology, Milan, Italy

4099

Poster Session (Board #91), Sat, 8:00 AM-11:30 AM

Clinicopathological and demographic factors associated with development of distant metastasis among patients with locoregional neuroendocrine tumor (NET). First Author: A. Dasari, The University of Texas MD Anderson Cancer Center, Houston, TX

Background: Somatostatin receptors (SSR) are overexpressed by neuroendocrine tumors (NETs). Peptide receptor radionuclide therapy (PRRT) targets the tumor cells through SSRs. Prediction of efficacy is critical since SSR expression or chromogranin A (CgA) levels have limited (20-60%) predictive ability. We assessed the utility of a blood-based multigene NET transcript analysis (NETest) to predict PRRT efficacy in 2 PRRT-treated NET cohorts. The algorithm comprises . 50 genes + clusters defining growth factor signaling (GFS) and metabolism (MTB). Methods: We investigated 2 independent 177Lu-PRRT-treated groups: #1 (n= 72), median total activity 18.5GBq (6.5-27.8) and #2 (n= 24), 24.5GBq (11.4-29.7). Baseline evaluations: grade (Ki67), SSR imaging (SRI), CgA (ELISA), and NETest (qRT-PCR - multianalyte algorithmic analyses). Using #1 we mathematically devised a predictive response index (PRI) of . 50 NET genes, MTB and GFS clusters & Ki67). We then assessed this PRI for PRRT efficacy in #2. RECIST criteria were used to assess disease control (responder vs non-responder). Statistical analyses: multiple regression, Kaplan-Meier survival, Chi2. Results: Cohorts were comparable by grade, staging (IV) and SRI uptake. #2 had more advanced disease (progressive 100% vs. 74%, p, 0.01). At restaging, disease control rates were similar (68% and 83%, p= 0.23) and median PFS was not reached in either cohort (follow-up 15 and 9.5 months, respectively). Baseline CgA was not different (6446133 vs. 6846295ng/ ml). NETest levels were significantly elevated in #2 (6766% vs. 5163%, p= 0.03), consistent with the progressive disease profile. PRRT response was not predicted by grading (p= 0.15), SRI (p= 0.41) or CgA (p= 0.12). In #1 the PRI accurately (. 90%) predicted responders (95%) and nonresponders (90%), significantly better than SRI (Level 4: 43% accuracy: Chi2= 41, p, 0.0001). In #2 the PRI was also predictive (83%) and significantly more accurate than SRI (46%, Chi2= 21, p, 0.01). Conclusions: NET gene analysis in blood significantly outperformed (p, 0.01) CgA, Ki67 and SRI for prediction of PRRT efficacy. PRRT efficacy can be accurately predicted (83-94%) by pre therapy blood transcript analysis.

Background: Incidence oflocoregional NETs is increasing. However, the association between the risk of development of metastasis and clinicopathological/demographic factors remains poorly understood. We assessed the association between baseline characteristics at diagnosis and risk of metastatic spread. Methods: From the Surveillance, Epidemiology and End Results (SEER)-Medicare database, we identified NET patients diagnosed between 1/2002 and 12/2011 by ICD-O-3 codes. We further restricted the study cohort to patients with locoregional disease at diagnosis. We excluded patients under age 65, without continuous Medicare Parts A and B enrollment or enrolled in health maintenance organizations. Development of metastatic disease was identified by capturing at least two claims indicative of secondary malignant neoplasms until 12/31/2013. Demographic and clinical characteristics between patients with and without development of metastasis were compared using Chi-squared tests. Time to development of metastasis was assessed using the Kaplan Meier method and compared using log rank test. Results: At the time of diagnosis, of the 5472 eligible patients, 67% had localized disease, 70% were grade 1 or 2 and were either gastroenteropancreatic (53%), pulmonary (31%) or of unknown origin (15%). During the study period, 1618 (30%) patients developed metastatic disease. We found that the risk of developing distant metastasis varied significantly by grade (22% in grade 1 to 57% in grade 4), stage at diagnosis (15% in local vs 70% in regional), and primary site (13% in colon and rectum to 44% in pancreas). The log rank tests confirmed the significant differences in disease progression by the above tumor characteristics with pvalues less than 0.01. Conclusions: This is the largest population-based study to date evaluating disease progression patterns among Medicare patients over 65 years old with locoregional NETs and showed that approximately 28% of patients developed distant metastases within two years from diagnosis. Histology grade, stage at diagnosis and primary site were significantly associated with risk of development of metastatic disease.

4100

4101

Poster Session (Board #92), Sat, 8:00 AM-11:30 AM

Therapeutic impact and timing of gastrointestinal malignancy genomic profiling: A single-institution experience. First Author: Kristin Lynn Koenig, The Ohio State University Wexner Medical Center, Columbus, OH Background: Tumor genomic profiling has become critical in the identification of targeted therapeutic options for patients (pts) with advanced malignancies. Mutational frequencies and their therapeutic importance vary among tumor types. This analysis was undertaken to characterize the landscape of genomic alterations in gastrointestinal (GI) malignancies found in a large academic institutional practice, and to determine the frequency of alteration-specific targeted therapy selection based on genomic profiling. Methods: Adult pts with GI malignancies presenting to the Ohio State University Comprehensive Cancer Center oncology clinics were offered next generation sequencing through FoundationOne testing as part of routine clinical care. Institutional review board approval was obtained to retrospectively analyze results from FoundationOne testing performed between 2012 and 2015. Results: 271 pts with GI malignancies underwent successful genomic profiling. A total of 1203 genomic alterations were found, with an average of 4.5 per tumor (range 0-20); 365 (30%) of these were potentially actionable and most often found in colorectal or gastroesophageal tumors. 14 pts (5.3%) had actionable alterations in MET, CDKN2A/B, FGFR2, KRAS, BRAF, or NF2 that led to enrollment in genotype-directed clinical trials or off label use of targeted therapies beyond standard of care. Pt performance status at the time of genomic alteration identification was a significant factor in precluding genotype-directed therapy. One variant of unknown significance involving FGFR2 identified at initial testing subsequently became actionable and led to pt enrollment on a clinical trial. One pt with rectal cancer was found to have a KRAS wild-type and BRAF mutant primary but KRAS mutant and BRAF wild-type liver metastasis. Conclusions: Genomic profiling of GI malignancies through FoundationOne testing is feasible and can lead to genotype-directed therapy selection; however, it should be considered early in the pt’s course to optimize use of targeted therapies through clinical trials. Consideration should be given to serial tumor testing to identify emerging genomic alterations for optimal therapy selection.

Poster Session (Board #93), Sat, 8:00 AM-11:30 AM

Peritoneal mesothelioma: Evaluation of chemotherapeutic agents used for HIPEC through the RENAPE database. First Author: Brice Malgras, Department of Surgical Oncology, Hopital ˆ Lariboisiere, ` AP-HP, Paris, France Background: peritoneal mesothelioma (PM) is a rare and rapidly fatal neoplasm with an estimated survival of one year. The introduction of cytoreductive surgery with hyperthermic intraperitonal chemotherapy (CRS + HIPEC) has greatly improved overall survival rates. Many chemotherapeutic agents have been used in HIPEC therapy but nonetheless without difference in terms of patient survival. The aim of the study was to evaluate the prognostic impact of the different cytotoxic agents used in HIPEC to treat PM. Methods: from 1989 to 2014, 260 patients with PM were treated with CRS + HIPEC in 20 tertiary French centers and prospectively included in the RENAPE database. Inclusion criteria were age # 80, performance status # 2, no significant comorbidities and no extraperitoneal metastases. Results: 5 cytotoxic agents (CA) were used for HIPEC (Cisplatin (C), Doxorubicin (D), Mitomycin (M), oxaliplatin (O), irinotecan(I)) including 6 combination regimens (C, C+D, C+M, M, O, O+I). Regimen did not influence postoperative survival. But, regimen that use 2 CA (C+D, C+M, O+I) improve overall survival compared to regimen using 1 CA (C, M, O), P = 0.03 HR[95% CI] = 0.54 [0.31-0.95]. No difference was found according to major complications (Grade $ 3) between the 2 groups. The 2 groups were comparable according to median age, gender, neoadjuvant chemotherapy, asbestos exposure, CC0 resection, follow up, histologic subtypes, except for median PCI which was higher in 1 CA group (15) than in 2 CA group (19) (P = 0.006). In CC0 patients overall survival was still improved in 2 CA group versus 1 CA group (P = 0.01, HR[95%CI] = 0.34[0.14-0.83]). Conclusions: CRS + HIPEC using 2 CA drugs seems to be associated with an improvement in overall survival in patients with PM, in particular when given to patients with high PCI. No difference was found in term of postoperative complication between HIPEC using 2 CA versus 1 CA.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Noncolorectal) Cancer 4103

Poster Session (Board #95), Sat, 8:00 AM-11:30 AM

Peri-operative modified FOLFIRINOX in resectable pancreatic cancer: A pilot study. First Author: Robert de Wilton Marsh, Kellogg Cancer Center, NorthShore University Health System, Evanston, IL Background: Surgical resection followed by adjuvant gemcitabine or 5-FU is standard for resectable pancreatic cancer (PC). FOLFIRINOX has substantial activity in metastatic PC. The efficacy and tolerability of peri-operative modified FOLFIRINOX (mFFx) in resectable PC is unknown. Methods: Patients (pts) with ECOG PS 0/1 and resectable PC confirmed by multidisciplinary review received 4 cycles of mFFX (oxaliplatin 85 mg/m2, irinotecan 180 mg/m2, leucovorin 400 mg/m2 day 1, 5-FU 2400 mg/m2 x 48 hours, peg-filgrastim 6 mg SQ day 4) both pre- and post-surgery. Results: 21 enrolled pts started therapy (median age: 63 (range 47-78), 81% ECOG 0). 20/21 pts completed 4 cycles of preoperative (pre-op) mFFX (1 withdrew for toxicity). 17/21 pts underwent resection (3 had metastatic disease intra-operatively). 14/21 completed 4 cycles of postoperative (postop) mFFX (2 pts expired within 90 days of surgery: 1 had an MI and 1 had GI bleed). Radiographic response by RECIST after pre-op mFFx: 1 CR, 3 PR, 16 SD, 0 PD, (ORR 4/20, 20%). Of the 17 (81%) patients who had curativeintent surgery, 16/17 (94%) had R0 resections. Treatment effect was seen in 10/17 (59%) specimens with 1 pathologic CR. Dose modifications were required in 30% pre-op (neuropathy 1; thrombocytopenia 2; neutropenia 3; diarrhea 2) and in 26% post-op (neuropathy 3; diarrhea 3; allergic reaction 1). Grade 3/4 adverse events occurred in 23% and 14% pts pre-op and in 26% and 0.06% pts post-op, respectively. In the 20 pts who completed preop mFFX, median OS was 33.4 months (mo) (95% CI: 13-not estimable); PFS was 11.4 mo (95% CI: 3.2 - 17.8). In the 17 pts who underwent surgical resection, median DFS was 9.7 mo (95% CI: 7.1-16.0). Conclusions: Eight cycles of mFFX is a feasible and tolerable peri-operative regimen in good PS pts with resectable PC. No pts progressed on pre-op mFFX. A high (94%) R0 resection rate in those pts undergoing surgery was observed. The OS of 33.4 mo compares favorably with historical data. Further study of mFFX may be warranted in the peri-op setting. Clinical trial information: NCT01660711.

4105

Poster Session (Board #97), Sat, 8:00 AM-11:30 AM

4104

231s Poster Session (Board #96), Sat, 8:00 AM-11:30 AM

Final analysis of stage 1 data from a randomized phase II study of PEGPH20 plus nab-Paclitaxel/gemcitabine in stage IV previously untreated pancreatic cancer patients (pts), utilizing Ventana companion diagnostic assay. First Author: Andrea J. Bullock, Beth Israel Deaconess Medical Center, Boston, MA Background: Poor outcome in pancreatic ductal adenocarcinoma (PDA) is associated partly with stromal hyaluronan (HA) accumulation, which compromises chemotherapy perfusion. PEGPH20 (PEGylated recombinant human hyaluronidase) potentiates chemotherapy by degrading HA. An ongoing, phase II, randomized study is comparing PEGPH20 + nab-paclitaxel (Nab) + Gemcitabine (Gem) (PAG) vs Nab+Gem (AG) in untreated Stage IV PDA. Interim data were presented at ASCO 2015 using a prototype assay to determine the HA-high subpopulation. Final efficacy data are now available with a newly developed Ventana HA companion diagnostic assay and scoring algorithm, which identifies HA-high (staining in the extracellular matrix [ECM] $ 50% of tumor surface) pts. Methods: Pts received PEGPH20 3 mg/kg twice weekly (C1), then weekly (C2+) with standard AG dosing. The protocol was amended after a temporary clinical hold (Apr-Jul 2014) for an imbalance in thromboembolic (TE) events to exclude high-TE-risk pts and add enoxaparin (LMWH) prophylaxis to both treatment arms. Primary endpoints are PFS and TE events. PFS and ORR data through Dec 2014 are for pts enrolled up to clinical hold (Stage 1); TE data through Dec 2015 (Stage 2) are outlined below. Results: 135 pts were treated (74 PAG, 61 AG). PFS results are shown below (median follow-up 7 mo). In HA-high pts receiving PAG vs AG, ORR was 55% (1 CR) vs 33%. TE events (PAG vs AG) were: Stage 1 (no LMWH) 42% vs 25%; Stage 2 (+LMWH 1 mg/kg/d) 7% vs 4%; overall (+LMWH 40 mg/d or 1 mg/kg/d) 12% vs 9%. No Grade 5 events have occurred with 1 mg/kg/day LMWH. Conclusions: Pts with HA-high tumors receiving PAG vs AG showed clinically meaningful improvements in PFS and ORR. TE events reduced with LMWH. A global phase III trial of PAG is scheduled to initiate in Q1 2016. Clinical Trial Information: NCT01839487. mPFS, mo Population Treated HA-High (n = 43 w/ evaluable HA data) HA-Low (n = 75 w/ evaluable HA data)

4106

PAG

AG

Hazard Ratio [95% CI]

5.5 (n = 74) 9.2 (n = 22)

5.2 (n = 61) 6.3 (n = 21)

0.73 [0.46-1.15] 0.48 [0.16-1.48]

5.3 (n = 44)

4.3 (n = 31)

0.69 [0.38-1.25]

Poster Session (Board #98), Sat, 8:00 AM-11:30 AM

Initial survival outcomes for the AGITG GAP study – a phase II study of perioperative nab-paclitaxel and gemcitabine for resectable pancreatic ductal adenocarcinoma (PDAC). First Author: Andrew Barbour, University of Queensland, Brisbane, Australia

Plasma biomarker for detection of early-stage pancreatic cancer and risk factors for pancreatic malignancy using antibodies for apolipoprotein-A2 isoforms. First Author: Kazufumi Honda, National Cancer Center Research Institute, Tokyo, Japan

Background: Nab-paclitaxel (nab-PC) and gemcitabine (GEM) are effective for metastatic PDAC. Efficacy of nab-PC and GEM as perioperative chemotherapy (CX) for resectable PDAC is unknown. The GAP trial assessed feasibility and outcomes of preoperative CX in operable PDAC patients (pts). Methods: A multi-centre single arm, phase 2 trial. ECOG 0/1 pts with operable PDAC received 2 cycles of preoperative nab-PC 125mg/m2 and GEM 1000mg/m2on D1, D8, D15 (28 day cycle) and 4 cycles postoperatively if PDAC resected. Response to CX was measured via CT (baseline (BL) and prior surgery) and FDG-PET scans (BL and D15 of cycle 1). Progression–free survival (PFS) defined as period from registration to progression and recurrence-free survival (RFS) is from surgery to relapse. Median follow up for PFS was 27.5 months (mths). Results: 42 pts were enrolled with 39 analysed for response (3 pts excluded – 1 had no PDAC and 2 unresectable at BL), 41 analysed for safety (1 unresectable at BL with no treatment given). The median tumour size was 26mm (IQR 25-35). 93% of pts received nabPC and GEM preoperatively but only 60% postoperatively. 78% of those pts tolerated adjuvant CX. 5/41 pts did not have surgery (2 progressed, 2 refused, 1 had sepsis). 36 had surgery, of which 6 were unresectable, 1 had no PDAC, 15 had R0 ( . 1mm) resection and 14 had R1 ( , 1mm). PFS (n = 39) was 12.3mths and RFS (n = 28) was 11.7mths (R0 pts 17.7 mths vs R1 pts 10 mths), with 7 pts progression free. 10/39 (26%) pts had a partial response (PR) to neoadjuvant CX while 26 (66%) had stable disease (SD). 15 (48%) had partial metabolic response on PET. Pts with PR on CT were 2.45 times (p = 0.01) more likely to have R0 resection but no correlation was seen with D15 PET response. As per NCCN guidelines at central review 29/39 pts were resectable (7 borderline and 3 unresectable). Median PFS for these pts classed as resectable was 12.3 mths, 21/29 were resected with median RFS 17.2 mths. Conclusions: GAP pts with PR from CX were more likely to have R0 resections and longer RFS. The RFS was similar to previous trials and support use of perioperative CX. Nab-PC and GEM is a well tolerated platform for future studies of preoperative therapy. Clinical trial information: ACTRN12611000848909.

Background: We have previously revealed that circulating levels of apolipoprotein A2 (apoA2) isoforms apoA2-ATQ/AT (C-terminal truncations of apoA2 homo-dimer) are significantly decreased in early-stage pancreatic cancer in comparison with healthy controls on mass spectrometry (MS)based assays. We report here the development of novel enzyme-linked immunosorbent assays (ELISAs) for measuring apoA2-ATQ/AT and their clinical applicability for the early detection of pancreatic cancer. Methods: Plasma and serum concentrations of apoA2-isoforms were measured in three independent cohorts (two Japanese cohorts and one pancreatic reference set collected by the US National Cancer Institute’s Early Detection Research Network (NCI EDRN), each comprising healthy control subjects and patients with pancreatic cancer or gastroenterological disease (n = 1156). These cohorts included 151 cases of stage-I/II pancreatic cancer. Results: Significant reductions in apoA2-ATQ/AT levels were recognized in patients with pancreatic cancer in comparison with healthy controls in both independent Japanese cohorts (p = 1.34*10-18 and 5.09*10-39). Areas under the receiver operating characteristic curve (AUCs) were . 0.92 for distinguishing patients with stage-I/II pancreatic cancer from healthy controls in the Japanese cohorts. The AUCs of apoA2-ATQ/AT to distinguish patients with pancreatic cancer from healthy controls were higher than those of CA19-9 in both Japanese cohorts. Better discrimination of pancreatic cancer was also observed with apoA2-ATQ/AT than with CA199 in the blind test using NCI EDRN data; combining apoA2-ATQ/AT with CA19-9 led to significantly improved AUC compared to CA19-9 alone Conclusions: ELISA for apoA2-isoforms offers a good biomarker for early detection of pancreatic cancer. We have started experimental screening for pancreatic cancer in the Kobe Cancer Screening Network.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

232s 4107

Gastrointestinal (Noncolorectal) Cancer Poster Session (Board #99), Sat, 8:00 AM-11:30 AM

Multicenter randomized phase II study comparing alternate-day oral therapy using S-1 with the standard regimen as a first-line treatment for patients with locally advanced and metastatic pancreatic cancer: PAN-01 study. First Author: Masayuki Sho, Department of Surgery, Nara Medical University, Kashihara, Japan Background: S-1 is an active agent for the treatment of pancreatic cancer. GEST study has previously shown the noninferiority of S-1 to gemcitabine in advanced pancreatic cancer. However, the standard regimen of 4 weeks of administration followed by 2 weeks of rest frequently causes adverse effects. To induce the effect of S-1 while reducing toxicity, alternate-day administration may be a treatment option. The aim of this study was to clarify the efficacy and toxicity of alternate-day administration of S-1 compared to the standard regimen for advanced pancreatic cancer. To this end, we conducted multicenter randomized phase II study, PAN-01. Methods: Patients who met the study criteria, including histological confirmation, age . 20, Performance Status of 0 or 1, no prior chemotherapy/radiotherapy, and adequate organ functions, were randomly assigned at a ratio of 1:2 to standard daily administration (Arm A) or alternate-day administration (Arm B). The primary end point was overall survival (OS). Secondary end points were safety, response rate, progression-free survival (PFS), and time to treatment failure. Results: A total of 190 patients were enrolled and 185 were subject to final analysis (Arm A: 64, Arm B: 121). The median OS for Arms A and B were 10.4 and 9.4 months, respectively (hazard ratio, 1.19; 95% CI, 0.86-1.64). Data did not indicate the noninferiority of alternate-day administration to standard regimen in overall survival. Furthermore, the median PFS were 4.2 and 3.0 months, respectively (hazard ratio, 1.65; 95% CI, 1.20-2.29), indicating that alternate-day administration was significantly worse than standard regimen in tumor progression. Anorexia (all grade: Arm A vs Arm B (%): 58.5 vs 50.0, P = 0.04), fatigue (60.0 vs 42.6, P = 0.02), pigmentation (24.6 vs 7.4, P , 0.001) and pneumonitis (7.7 vs 1.6, P = 0.03) were more common in daily administration than alternate-day administration. Conclusions: This study failed to demonstrate the noninferiority of alternate-day administration of S-1 to standard regimen as a first-line chemotherapy in unresectable pancreatic cancer. Clinical trial information: UMIN000008604.

4109

Poster Session (Board #101), Sat, 8:00 AM-11:30 AM

4108

Poster Session (Board #100), Sat, 8:00 AM-11:30 AM

Genetic heterogeneity and survival among pancreatic adenocarcinoma (PDAC) patients with positive family history. First Author: Gloria M. Petersen, Mayo Clinic, Rochester, MN Background: Availability of cancer gene mutation panel-based screening will identify increasing numbers of patients who carry diverse germline mutations. Prevalence of mutations and survival by carrier status among PDAC patients has not been well described. Methods: Using a hereditary cancer panel, we sequenced 25 cancer susceptibility genes in lymphocyte DNA from 303 well-characterized PDAC patients (47% female, mean age (+SD) 64.9+10.3 years, 98% White, 3% Ashkenazi Jewish heritage) in the Mayo Clinic Familial Pancreatic Cancer Registry. Kindreds that contained at least two first degree relatives (FDR) with PDAC met criteria for Familial Pancreatic Cancer (FPC), while the remaining were familial, but not FPC. Results: Overall, 35 of 303 (12%) patients carried a deleterious mutation in one of 10 genes in the 25-gene panel. Of the 186 FPC patients, 24 (13%) were carriers, and 11/117 (9%) patients with family history not meeting FPC criteria were carriers. Prevalence of (n) in PDAC patients were BRCA2 (11), ATM (8), CDKN2A (4), CHEK2 (4), MUTYH/MYH (3 heterozygotes, not biallelic), BRCA1 (2), and 1 each in BARD1, NBN, PALB2, and PMS2. Novel mutations were found in ATM, BARD1, and PMS2. Among individuals with deleterious mutations in genes associated with breast cancer, 13/27 (48%) had a FDR with breast cancer vs non-carriers (76/276 (28%); p = .02); among only women with these deleterious mutations, the proportions were 9/12 (75%) and 39/129 (30%), respectively; p = .003). There was no significant difference in median survival overall or by stage among mutation carriers versus non-carriers: Overall sample: 14.2 months (mos) vs 12.7 mos, respectively; p = 0.78; Resectable: 32.4 mos, n = 10 vs 27.3 mos, n = 78; Locally Advanced: 14.5 mos, n = 6 vs 10.6 mos, n = 72; Metastatic: 8.8 mos, n = 15 vs 7.8 mos, n = 98). Conclusions: With 12% prevalence of deleterious mutations, susceptibility gene testing in PDAC patients with a positive family history is warranted regardless of meeting FPC criteria. These findings will inform genetic risk counseling for family members. While survival did not differ, further analysis is warranted. The diversity of genetic susceptibility offers future opportunities for targeted therapies.

4110

Poster Session (Board #102), Sat, 8:00 AM-11:30 AM

Pathological response to neoadjuvant gemcitabine plus nabpaclitaxel in pancreatic adenocarcinoma to improve survival. First Author: Rafael AlvarezGallego, Centro Integral Oncologico ´ Clara Campal- Hospital Madrid Norte Sanchinarro, Madrid, Spain

RUCAPANC: An open-label, phase 2 trial of the PARP inhibitor rucaparib in patients (pts) with pancreatic cancer (PC) and a known deleterious germline or somatic BRCA mutation. First Author: Susan M. Domchek, University of Pennsylvania, Philadelphia, PA

Background: nab-paclitaxel + gemcitabine is approved for the treatment of metastastic pancreatic cancer. So far, few studies have investigate this regimen in patients with nonmetastatic disease. Here we describe our experience of using gemcitabine and nab-paclitaxel as neoadjuvant. Methods: Patients with resectable, borderline resectable or locally advanced pancreatic cancer were treated with gemcitabine (1000mg/m2 days 1, 8 and 15) and nab-paclitaxel (125mg/m2 days 1, 8 and 15) as neoadjuvant treatment. Some patients were treated also with chemoradiotherapy after completion of chemotherapy treatment. Response were assessed by FDGPET, CA199 levels and EUS-elastography. Results: 58 patients were included into the study and treated wit gemcitabine plus nab-paclitaxel. The median number of cycles were 3 (range 2-14). At diagnosis 15 patients presented with resectable disease, 16 with borderline tumors and 27 patients had locally advanced cancer. Twenty-four patients received treatment with chemo-radiotherapy after chemotherapy and 35patients (60%) underwent surgery (12 resectable, 8 borderline and 15 with locally advanced cancer). Poor pathological tumor response, defined by Ryan Grade Regression Tumor (GRT) grade 2 or 3, was found in 18 patients and 17 had good pathological response (GRT grade 0 or 1). Seventeen patients presented progression disease during neoadjuvant treatment and 6 patients were inoperable. Median overall survival (mOS) and progression free survival (PFS) was 16.6 months (m) and 8,8 m respectively for the entire population. In resected patients mOS and PFS were 17,4 and 13,8 m respectively. mOS was 30,6 m in patients with good pathological and 16,5 m in poor responding patients (p = 0.008). Likewise, PFS was 16,6 m in patients with good pathological response versus 8,9 m in poor responding patients (p = 0,019). In a multivariate analysis no treatment with chemoradiotherapy (p = 0.006) and good GRT (p = 0.006) impacts in overall survival. Conclusions: Neoadjuvant treatment with gemcitabine plus nab-paclitaxel improve PFS and OS in patients with non metastatic pancreas adenocarcinoma who undergo surgery and show a good pathological responding tumor.

Background: Pts with PC have a poor prognosis and limited treatment options. Around 9% of pts with PC have a germline or somatic BRCA mutation. Recent studies showed that rucaparib effectively treats pts with platinumsensitive, relapsed, high-grade ovarian carcinoma and a BRCA mutation. RUCAPANC investigated the efficacy and safety of rucaparib in pts with PC and a known deleterious BRCA mutation. Methods: RUCAPANC (NCT02042378) enrolled pts with PC with measurable, relapsed disease who received 1–2 prior chemotherapy regimens in the locally advanced/ metastatic setting. Pts received oral rucaparib (600 mg twice daily) until disease progression. The primary endpoint was objective response rate (ORR) by RECIST v1.1. Safety was also evaluated. Results: Nineteen pts were treated; 3 pts are ongoing. Median age was 57 years (range 41–75), 58% were male, 68% received $ 2 prior chemotherapy regimens (including adjuvant treatment), 79% were ECOG Performance Status of 1, and 79% had a BRCA2mutation by local testing. The confirmed ORR was 11% (1 partial response [PR] and 1 complete response [CR]); 1 additional patient had an unconfirmed PR. All pts with a response received only 1 prior line of therapy. The duration of confirmed responses was 36 and 49 weeks (both ongoing). The disease control rate (PR or stable disease [SD] for $ 12 weeks) was 32% (6/19) in all pts and 50% (3/6) in pts with only 1 prior regimen. Four pts had SD, 9 had PD, and 3 were not evaluable for response. One pt with SD is ongoing at 60 weeks of treatment. Enrollment was stopped due to lack of responses in the first 15 pts evaluated; the 3 PRs occurred in the last 4 pts enrolled. Common treatment-related AEs included nausea and anemia (37% each). Common treatment-related grade $ 3 AEs included anemia (26%), thrombocytopenia (16%), and fatigue (11%). Conclusions: Rucaparib provided clinical benefit to pts with advanced PC and a BRCA mutation and demonstrated an acceptable safety profile. Less heavily pretreated pts derived durable benefit, suggesting that rucaparib may be an option earlier in the treatment course. These findings may inform future study designs in pts with advanced PC. Clinical trial information: NCT02042378.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Noncolorectal) Cancer 4111

Poster Session (Board #103), Sat, 8:00 AM-11:30 AM

Phase II study of modified FOLFIRINOX for chemotherapy-na¨ıve patients with metastatic pancreatic cancer. First Author: Makoto Ueno, Division of Hepatobiliary and Pancreatic Oncology, Kanagawa Cancer Center, Yokohama, Japan Background: FOLFIRINOX is considered one of the standard chemotherapy regimens for chemotherapy-na¨ıve pts. with MPC, but carries an unfavorable adverse event (AE) profile, especially in Japanese Phase II study of full dose FOLFIRINOX, the incidence of grade 3 or more neutropenia and febrile neutropenia was 77.8% and 22.2% respectively. Objective: The purpose of this study was to evaluate the efficacy and safety of modified FOLFIRINOX regimen: intravenous oxaliplatin 85 mg/m2, irinotecan 150 mg/m2, 5-FU infusion 2,400 mg/m2over 46 h, no bolus 5-FU. Methods: This study was an open-label, multicenter, single-arm phase II study. The primary endpoint was overall survival and the incidence of grade 3 or more neutropenia. All patients did not receive prophylactic pegfilgrastim. Results: Sixty-nine pts. were enrolled from 39 institutions in Japan. Median age was 62.6 years (range 42.4 - 74.2). All pts. had no prior treatment and had distant metastatic lesions. 68.1% of the pts. had a PS 0, the primary site of the tumor was the head of the pancreas in 44.9% of pts., 20.3% of pts. had a biliary stent, and 5.8% of pts. experienced recurrence after resection. The UGT1A1 genotype was wild type in 79.7%, heterozygous (*1/*6, *1/*28) in 20.3%. Median overall survival was 11.2 months (95% confidence interval [CI], 9.0-) and the 1-year survival proportion was 0.481. Median progression-free survival was 5.5 months (95% CI, 4.1-6.7). The response rate was 37.7% (95% CI, 26.3-50.2) and disease control rate was 78.3% (95% CI, 66.7–87.3). The incidence of grade 3 or more neutropenia was 47.8%. Other major grade 3 or more toxicities were febrile neutropenia (8.7%), thrombocytopenia (2.9%), anemia (4.3%), anorexia (15.9%), diarrhea (10.1%), nausea (8.7%), cholangitis (5.8%) and peripheral sensory neuropathy (5.8%). Serious adverse events occurred in 6 pts. (8.7%). All AE proportions were less than those in the previous Japanese full dose Phase II study. One patient died due to interstitial pneumonia related to a treatment. Conclusions: This is the first prospective study of modified FOLFIRINOX in Asia. Modified FOLFIRINOX of this study has an improved safety profile with maintained efficacy in MPC without prophylactic pegfilgrastim. Clinical trial information: UMIN000013301.

4113

Poster Session (Board #105), Sat, 8:00 AM-11:30 AM

Optimized economic evaluation for the United States (US) of nab-paclitaxel plus gemcitabine (NAB-P+GEM), FOLFIRINOX (FFX), and gemcitabine (GEM) as first-line treatment for metastatic pancreatic cancer (mPDA). First Author: Mahdi Gharaibeh, Center for Health Outcomes & PharmacoEconomic Research, Tucson, AZ Background: Per conference feedback we optimized our 2015 ASCO reported independent economic evaluation of GEM, NAB-P+GEM, and FFX in mPDA by improving survival curve fitting (exponential v. Weibull distributions) and modifying disease monitoring. Methods: We used published direct efficacy estimates of GEM v. FFX or NAB-P+GEM, complemented by FFX v. NAB-P+GEM efficacy estimated with Bucher’s indirect comparison method. Kaplan Meier curve fit was tested for exponential and Weibull parametric distributions. A state transition model with full time horizon was used to estimate the life years (LY) and quality-adjusted life years (QALY) gained and incremental cost-effectiveness (ICER) and cost-utility ratios (ICUR) (payer perspective; 3%/year discount rate; 2015 US$). We did not benchmark to categorical thresholds as pharmacoeconomic analysis is to inform, not determine, policy or clinical practice. Results: Per Bucher method, HR for NAB-P+GEM vs. FFX was 1.47 (95%CI = 1.10-1.96) for PFS and 1.26 (95% CI = 0.95-1.67, p = ns) for OS indicating superiority in PFS for FFX but no superiority in OS for either regimen.In direct comparison, differentials for FFX over GEM were +0.372 LY and +0.257 QALY gained for ICER of $252,474/LY and ICUR of $365,530/QALY at incremental cost of $93,941; for NAB-P+GEM over GEM +0.182 LY and +0.130 QALY for ICER of $136,202/LY and ICUR of $190,349/QALY at incremental cost of $24,751. In indirect comparison, differentials for FFX over NAB-P+GEM were +0.19 LY and +0.127 QALY gained for ICER of $363,470/LY and ICUR of $544,803/QALY at an incremental total cost of $69,190. Conclusions: In the absence of a statistically significant (indirectly estimated) OS benefit of either NAB-P+GEM or FFX, the economic benefits in terms of cost-savings and incremental cost-effectiveness and cost-utility favor NAB-P+GEM over FOLFIRINOX therapy in mPDA.

4112

233s Poster Session (Board #104), Sat, 8:00 AM-11:30 AM

Do pancreatic cancer (PDA) stem cell markers predict biologic behavior? First Author: Ibrahim Halil Sahin, Icahn School of Medicine at Mount Sinai St Luke’s Roosevelt Hospital, New York, NY Background: Studies indicate heterogeneous clinical behavior arises in part from variant clones and cancer stem cells (CSC) which are CD44 (+) and Epithelial Specific Antigen (ESA) (+) (Lee, 2008). We investigated the impact of CSC markers on survival in PDA patients who had recurrence following surgical resection (SR). Methods: N=69 patients (pts) diagnosed between 01/04 and 12/14 at Memorial Sloan Kettering who had SR followed by lung or liver recurrence, were identified. Pts eligible if they had consented to IRB #00-032/#06-107, permitting research tissue analysis, and if they had primary tumor tissue available. We assessed CD44 and ESA expression by immunohistochemistry (IHC) staining. Staining intensity was scored as weak (1), moderate (2) and strong (3); The number of (+) cells was scored as: few (1), patchy (2) and diffuse (3). IHC staining was considered (+) if total score $4. Overall survival (OS) and time from relapse to death (TRD) were evaluated using Wilcoxon signed-rank test Results: Of 69 pts, N=25 primary tissue samples were available to assess CD44 and N=23 samples for ESA. All pts were deceased at the time of analysis. No difference in OS was observed between pts with CD44(+) and CD44(-) (Table1). Pts with ESA (-) had worse OS compared to pts with ESA (+) (P=0.008). TRD was 3.4 and 8.8 mo in CD44(+) and CD4(-) pts respectively (P=0.2). Improved TRD was observed in ESA (+) pts compared to ESA (-) pts(P=0.02). Sub-analysis of CD44 (+) and ESA (-) pts (group1) and CD44 (-) and ESA (+) pts (group2) showed a trend to worse OS and TRD in group1. Conclusions: Observed survival outcomes were improved for pts with ESA (+). Loss of ESA expression and related worse outcomes could be due to epithelial-mesenchymal transition and acquisition of CSC characteristics. The observed trend to worse TRD in CD44 (+) may indicate a potential biomarker role for rapid disease progression following recurrence. Respecting small sample size, these data suggest CSC biomarkers may have prognostic value in PDA and further investigation is warranted.

OS (mo) TRD (mo)

4114

CD44 (+)/CD44 (-) (13/12)

ESA (+)/ESA (-) (16/7)

CD44 (+) and ESA (-)/CD44 (-) and ESA (+)(4/9)

16.1/17.2 (P=0.5) 3.4/8.8 (P=0.2)

34.3/8.74 (P=0.008) 10.7/3.2 (P=0.02)

8.7/36.5 (P=0.08) 3.2/19.6 (P=0.13)

Poster Session (Board #106), Sat, 8:00 AM-11:30 AM

Evofosfamide combined with gemcitabine/nab-paclitaxel in patients with previously untreated locally advanced or metastatic pancreatic adenocarcinoma (PAC): Results of a phase I trial. First Author: Mitesh J. Borad, Mayo Clinic, Scottsdale, AZ Background: Gemcitabine + nab-paclitaxel (G/nab-P) is a standard first-line option for advanced PAC. Evofosfamide (Evo, previously known as TH-302) is a hypoxia-activated prodrug of bromo-isophosphoramide mustard that is preferentially activated under hypoxic conditions. The recent phase III MAESTRO study suggested that Evo + G has clinical activity (overall survival: HR = 0.84, P = 0.0589). Encouraging preclinical activity for Evo + G/nab-P and the clinical importance of G/nab-P provided a rationale for study of this triplet. Methods: This phase I, dose-escalation trial (NCT02047500) recruited pts with previously untreated locally advanced or metastatic PAC. A classical 3 + 3 design used 3 planned Evo dose levels: 170 mg/m2, 240 mg/ m2, and 340 mg/m2 on days 1, 8, and 15 of a 28-day cycle, combined with nab-P 100 mg/m2 followed by G 800 mg/m2. Primary objectives: safety and tolerability, maximum tolerated dose (MTD), and the recommended phase II dose (RP2D) of Evo + G/nab-P. Secondary objectives: RECIST response rate and safety. Results: At the cut-off date of 7 Dec 2015, 17 pts were enrolled (median age 62.4 [50.0–78.2]; male 7). 7 received Evo at 170 mg/m2, 3 at 240 mg/m2, and 7 at 340 mg/m2. A pt treated with Evo 170 mg/m2 had 2 dose-limiting toxicities (DLTs; both grade 3/4 hematologic events); no other DLTs were observed and Evo 340 mg/m2 was established as the MTD. The most common treatment-related adverse events (TRAEs) were neutropenia (n = 13), thrombocytopenia (12), anemia (10), alopecia (10), and fatigue (10). Serious TRAEs reported in 4 pts were pancytopenia, febrile neutropenia, pneumonia, sepsis, lung infiltration, acute kidney injury, erythematous rash, and hypotension. One TRAE (sepsis) led to death in the Evo 170 mg/m2 cohort. Best overall response: partial response (9/17, 52.9%), stable disease (5/17, 29.4%), not evaluable (3/17, 17.6%). Conclusions: Evo plus attenuated G/nab-P showed a manageable safety profile with evidence of activity. Clinical trial information: NCT02047500.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

234s 4115

Gastrointestinal (Noncolorectal) Cancer Poster Session (Board #107), Sat, 8:00 AM-11:30 AM

Systematic review of resection rates and clinical outcomes after FOLFIRINOX-based treatment in patients with locally advanced pancreatic cancer. First Author: Steffi Rombouts, University Medical Center Utrecht, Utrecht, Netherlands Background: FOLFIRINOX prolongs survival in patients with metastatic pancreatic cancer and may also benefit patients with LAPC. Furthermore, it may convert a substantial number into resectable tumors. Previous studies combined patients with LAPC and borderline resectable pancreatic cancer, which hampers the interpretation of outcomes with FOLFIRINOX in LAPC. The aim of this review was to provide an overview of the (R0) resection rate and clinical outcomes after FOLFIRINOX-based therapy for locally advanced pancreatic cancer (LAPC). Methods: PubMed, Embase and the Cochrane library were systematically searched for studies published up to August 31st, 2015. Primary outcome was the (R0) resection rate. Results: Fourteen studies involving 365 patients with LAPC were included. A modified chemotherapy regimen was described in 3 studies and FOLFIRINOX dose reductions in up to 65% of patients. Radiotherapy was given in 57% of all patients. Total resection rate was 28% (77% R0) with a peri-operative mortality of 3%. Median overall survival ranged from 8.9 to 25.0 months. Median survival after resection was 24.9 months, based on one study. In total, 6 out of 85 (7%) resection specimens with available data showed a complete pathologic response. Grade 3 to 4 toxicity occurred in 23% of patients. Data of patients treated with FOLFIRINOX without radiotherapy were available from 292 patient: resection rate was 12% (70% R0) with 15.7 months median overall survival and 19% grade 3 to 4 toxicity. Conclusions: FOLFIRINOX-based treatment for patients with LAPC seems safe and achieves high (R0) resection rates and overall survival, despite the frequent administered modified regimes and dose reductions during treatment.

4117

Poster Session (Board #109), Sat, 8:00 AM-11:30 AM

4116

Poster Session (Board #108), Sat, 8:00 AM-11:30 AM

Preoperative gemcitabine-nab-paclitaxel (G-NP) for (borderline) resectable (BLR) or locally advanced (LA) pancreatic ductal adenocarcinoma (PDAC): Feasibility results and early response monitoring by Diffusion-Weighted (DW) MR. First Author: Jean-Luc Van Laethem, Erasme University Hospital, Brussels, Belgium Background: G-NP has recently emerged as an effective combination treatment for metastatic PDAC. We aimed to evaluate tolerability and activity of induction/neoadjuvant G-NP before surgery. In addition, we evaluated DW-MR imaging to monitor tumoral changes and to predict early response to chemotherapy compared with RECIST evaluation. Methods: Patients (pts) with ECOG PS 0/1, cytologically proven PDAC, normal bilirubin level and BLR or LA disease were included and received at least one cycle of G 1000 mg/m2 and NP 125 mg/m2, weekly 3week/4. Additional cycles were indicated according to sequential tumor restaging for resectability evaluation; complementary chemoradiation (CRT) (50.4 Gy/28 fr) with capecitabine (825 mg/m2 BID) was administered in case of persisting non resectability. Sequential DW-MR was performed at baseline and weekly during cycle 1 and monthly for subsequent cycles. DW-derived quantitative parameters (DW tumour volume and histogram based apparent diffusion coefficient (H-ADC) and pure diffusivity (H-D)) were compared between responders and non-responders, based on clinical/biochemical (CA 19.9) morphological and pathological response (pR) Results: 23 pts received therapy (M/F: 13/10; median age: 63, ECOG 0/1: 12/23), 13 BLR and 10 LA PDAC; median number of cycles: 3. Eight pts received CRT. Related grade III and IV toxicity occurred in 15 and 1 pts. No death occurred during the whole sequence. Best RECIST responses were 8 PR, 12 SD and 3 PD. Ten/16 pts (62%) with an elevated CA19.9 level at baseline decreased CA 19.9 levels by at least 50 %. R0 (n = 7)/R1(n = 6) resection was performed in 13/23 pts (56%).There were 0 pCR, 3 major pR and 10 minor/no pR. In 16 analyzed pts for DW-MR images, DW tumor volume, H-ADC and H-D variations were significantly different between responders and non-responders at the end of cycle 1. Conclusions: Induction chemotherapy with G-NP is feasible and active before surgery; our data suggest a potential interest of monitoring response to preop G-NP chemotherapy by using DW-MR for the early discrimination of responders/non-responders. Clinical trial information: NCT01715142.

4118

Poster Session (Board #110), Sat, 8:00 AM-11:30 AM

Safety, pharmacokinetics, pharmacodynamics, and antitumor activity of necuparanib combined with nab-paclitaxel and gemcitabine in patients with metastatic pancreatic cancer: Updated phase 1 results. First Author: Eileen Mary O’Reilly, Memorial Sloan Kettering Cancer Center, New York, NY

Does clinical stage predict pathologic stage in pancreatic adenocarcinoma? First Author: Gyulnara G. Kasumova, Surgical Outcomes Analysis and Research, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA

Background: Necuparanib (“necu”) inhibits multiple heparin-binding growth factors, chemokines, and adhesion molecules in nonclinical studies. It is being evaluated in a Phase 1/2 trial with nab-paclitaxel (nabP) + gemcitabine (gem) in patients (pts) with metastatic pancreatic cancer. Updated Phase 1 safety/pharmacokinetic (PK)/pharmacodynamic (PD) data are provided. Methods: Necu was given as daily escalating s.c. doses in combination with i.v. 125 mg/m2 nabP and 1000 mg/m2 gem (Days 1, 8, 15 of each 28-day cycle). The necu start dose was 0.5 mg/kg and was escalated until the maximum tolerated dose was determined (which was 5 mg/kg). The protocol was amended to include nabP after 2 necu + gem cohorts were completed. Results: 39 pts in 7 cohorts (0.5 and 1 mg/kg + gem; 1, 2, 4, 6, and 5 mg/kg + nabP + gem) received necu. No new safety issues have been observed in this Phase 1 trial over the past year and no increase in known AEs of chemotherapy across doses was observed. Exposure to necu was observed starting at 2 mg/kg. Release of heparin-binding proteins (e.g. HGF, a PD marker) from heparan sulfate stores increased with dose and plateaued at 45 mg/kg. 3 (8%) pts remain alive; 19 (49%) pts have lived . 1 year, 8 (21%) . 1.5 years, and 4 (10%) . 2 years; 33.7 months is the longest survival to date. 16 pts treated with necu + nabP + gem completed Cycle 1 and had $ 1 scan on treatment; 8 (50%) achieved RECIST partial response and 6 (38%) achieved stable disease, for a disease control rate of 14/16 (88%); median overall survival (OS) in this subset was 16.0 months. Median OS of patients treated with $ 1 dose of necu + nabP + gem (n = 24) was 12.4 months. Of 15 CA19.9 evaluable pts, 15 (100%) had . 20%, 14 (93%) had . 50%, and 7 (47%) had . 90% decreases from baseline. Conclusions: Acceptable tolerability and encouraging signals of activity continue to be observed with the combination, which is being further evaluated in an ongoing randomized, double-blind, Phase 2 trial. Clinical trial information: NCT01621243.

Background: When faced with a diagnosis of pancreatic cancer, patients and providers rely upon stage of disease to formulate a treatment plan. The possibility of undertreatment of early-stage patients has been raised. We performed a large national retrospective analysis to evaluate the consistency between clinical and pathologic staging. Methods: National Cancer Data Base was queried for pancreatic adenocarcinoma 2004-2013. Those who underwent resection with or without chemotherapy and had defined AJCC clinical and pathologic stages were included for analysis. Clinical and pathologic stages were compared; if pathologic.clinical stage defined as up-staged; if clinical.pathologic defined as downstaged. Characteristics were compared by x2. Cochran-Armitage trend test used to evaluate trends in consistency of staging over time. Results: In 22,076 patients, clinical and pathologic stages were consistent 64.4% of the time. 68.8% of those down-staged received neoadjuvant therapy. Patients younger than 65 were more likely to be down-staged (56.0 vs 44.0%) while those older were up-staged (57.6 vs 42.4%) (p,0.0001). Private insurance holders were more likely to be downstaged (45.1%) vs those with Medicare were up-staged (56.4%) or have consistent staging (50.9%) (p,0.0001). Majority were treated at academic vs community centers (58.9 vs 41.1%) with down-staging more frequent at academic centers (67.8 vs 32.3%, p,0.0001). Over time, consistency between clinical and pathologic staging has declined (p,0.0001), with more patients being up-staged (p,0.0001). Conclusions: Our ability to accurately predict clinical stage of pancreatic cancer has not improved over time. An increased proportion of patients are upstaged based on resection pathology. The so-called “national failure to operate” may in part be a reflection of this poor correlation and underlying advanced disease. Better prognostic tools and novel treatment strategies including neoadjuvant treatment may be warranted. STAGING TREATMENT

Consistent

UP-stage

DOWN-stage

N (%) Total

p-value

Surgery First 12267 (86.3) 6051 (91.3) 388 (31.2) 18706 (84.7) ,0.0001 Neoadjuvant 1939 (13.7) 576 (8.7) 855 (68.8) 3370 (15.3) Total 14206 (64.4) 6627 (30.0) 1243 (5.6) 22076 (100.0)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Noncolorectal) Cancer 4119

Poster Session (Board #111), Sat, 8:00 AM-11:30 AM

A randomized, double-blinded, placebo-controlled phase II trial of gemcitabine (gem) plus nab-paclitaxel (nab-P) plus apatorsen (A) or placebo (Pl) in patients (pts) with metastatic pancreatic cancer (mPC): The RAINIER trial. First Author: Andrew H. Ko, University of California, San Francisco, San Francisco, CA

4120

235s Poster Session (Board #112), Sat, 8:00 AM-11:30 AM

Nab-paclitaxel plus gemcitabine or plus simplified LV5FU2 as first-line therapy in patients with metastatic pancreatic adenocarcinoma: A GERCOR randomized phase II study (AFUGEM). First Author: Pascal Hammel, Beaujon Hospital, Clichy, France

Background: Over expression of heat shock protein 27 (Hsp27) in PC promotes tumor growth and metastasis, and is associated with poor prognosis. A is an antisense oligonucleotide that binds to Hsp27 mRNA and inhibits production of Hsp27 protein. This randomized, double-blinded, phase II trial evaluates the efficacy of gem/nab-P plus A or Pl in pts with mPC. Methods: Pts with untreated mPC were randomized 1:1 to Arm A (gem, nabP, A) or Arm B (gem, nab-P, Pl). 3 loading doses of 600mg A IV or Pl IV were given, then 600 mg A or Pl weekly with chemotherapy in 28 day cycles. Both arms received gem 1000mg/m2 IV +nab-P 125mg/m2 IV on days 1, 8, and 15. Pts were re-staged every 2 cycles. Serum Hsp27 levels were drawn at baseline and on treatment. The primary endpoint was overall survival (OS); secondary endpoints were progression free survival (PFS), response rate (RR), CA 19-9 response, and toxicity. Results: 132 pts were randomized: median age 66 yrs, 57% male, 47% ECOG 0. Toxicities were similar between arms. The most frequent G 3/4 treatment-related AEs reported in . 10% of subjects included anemia (20%), neutropenia (17%), fatigue (16%), thrombocytopenia (14%), and nausea (13%) on Arm A and anemia (27%), neutropenia (19%), thrombocytopenia (13%), and fatigue (11%) on Arm B. Serious AEs were numerically higher in arm A (16 vs 9 pts). Overall RR on both arms was 18%. With a median follow-up of 9.2 mo, median PFS was 2.7 vs 3.8 months for arms A and B, respectively (hazard ratio 1.0), while median OS was 5.3 vs. 6.9 months (HR 1.2). Of the 14% of pts with high baseline serum level of Hsp27 ( . 9.3 ng/mL), median PFS and OS was longer in arm A than in arm B (PFS 3.3 vs 0.9 mo, HR 0.4; OS 3.3 vs 1.0 mo, HR 0.6). Conclusions: The addition of A to chemotherapy for pts with first-line mPC did not improve outcomes in the ITT population, although OS for the entire cohort did worse than expected. There was a trend toward improvement in PFS and OS in the A arm for pts with high baseline serum Hsp 27, a poorerprognosis subgroup. Further study of A plus chemotherapy may be warranted in mPC pts preselected for this biomarker. Clinical trial information: NCT01844817.

Background: nab-paclitaxel plus gemcitabine regimen (NABGEM) became a standard in patients (pts) with metastatic pancreatic adenocarcinoma (mPAC). The AFUGEM trial evaluated the efficacy and safety of nabpaclitaxel plus simplified LV5FU2 (NABFU). Methods: Main eligibility criteria were: untreated mPAC, ECOG PS # 2. Pts received nab-paclitaxel (125 mg/m² on d1, 8 and 15) plus gemcitabine (1000 mg/m² on d1, 8 and 15) or plus simplified LV5FU2 (leucovorin 400 mg/m², 5FU bolus 400 mg/ m² followed by 2400 mg/m² given as a 46-hour continuous infusion on d1 and 15), one cycle every 4 weeks in both arms, ratio 1:2, respectively. The primary end point was observed progression-free survival (PFS) rate at 4 months (m). A Fleming 2-stage design was used with a targeted (H1) 4mPFS rate of 50% in NABFU arm and an uninteresting 35% rate (H0, unilateral alpha of 5% and power of 80%). Results: 114 pts were included: NABGEM n = 39, NABFU n = 75. Baseline characteristics were well balanced: median age 66 years (45-86), male gender 61.4%, head tumor 43.8%, $ 2 metastatic sites 36.8%, ECOG PS 0-1 84.2% and 2 15.8%; at the exception of pain more frequent in the NABFU arm (38.5% vs 56.0%). In mITT population, the observed 4m-PFS rates were of 53.9% [95% CI 38.2%-69.5%] in NABGEM and 55.4% [95% CI 44.1%-66.7%] in NABFU; objective response rates were 39.5% [95% CI 23.9-55.0] and 35.1% [95% CI 24.3-46.0], respectively. After a median follow-up of 16.9 m, median PFS were 4.9 m [95% CI 2.1-7.7] in NABGEM and 6.0 m [95% CI 4.1-7.6] in NABFU; median overall survivals were 9.2 m [95% CI 6.013.6] and 11.4 m [95% CI 8.8-16.6], respectively [HR = 0.64; 95% CI 0.41-1.01]. Grade 3-4 adverse events were more frequent in NABGEM (78.9%) than in NABFU (71.6%) with more anemia (10.5% vs 5.4%), thrombocytopenia (10.5% vs 0%), ALAT (13.2% vs 4.1%), ASAT (13.2% vs 5.4%) and bilirubin (5.3% vs 1.4%) increase but less mucositis (2.6% vs 8.1%) and sensitive neuropathy (7.9% vs 18.9%). Conclusions: Nabpaclitaxel plus simplified LV5FU2 appears at least as active as gemcitabine plus nab-paclitaxel in mPAC. The toxicity profile is safe and tolerable. This regimen deserves evaluation in a phase III trial. Clinical trial information: NCT01964534.

4121

4122

Poster Session (Board #113), Sat, 8:00 AM-11:30 AM

Comparative molecular analyses of pancreatic cancer (PC): KRAS wild type vs. KRAS mutant tumors and primary tumors vs. distant metastases. First Author: Mohamed E. Salem, Lombardi Comprehensive Cancer Center, Georgetown University, Washington, DC Background: Targeted therapy has a minimal role in PC, partly because PC has not yet been well molecularly characterized. Little is known about the molecular characteristics of the subset of PC that doesn’t carry KRAS mutations. Better knowledge would enhance our ability to develop targeted therapies. Methods: PC tumors submitted to Caris Life Sciences for IHC (protein expression), ISH (gene amplification), and NGS sequencing between 2009 and 2015 were studied. Chi-square tests determined differences. Results: A total of 2426 PC tumors were examined. KRAS mutations (85%) were the most frequent genetic alteration. Other commonly mutated genes were TP53 (63%), SMAD4 (13%), BRCA2 (12%), ATM/APC/NTRK1 (5% each), BRCA1 (4%) and cMET/PIK3CA (3% each). BRAF mutations were seen in 6% of the RAS WT tumors. When compared with KRAS MT, KRAS WT tumors had a greater frequency of BRCA1 (9% vs. 3%, p = 0.05), CTNNB1 (5% vs. 0.2%; p , 0.01), GNAS (4% vs. 1.5%, p = 0.02), and FGFR2 (1.2% vs. 0.1%, p , 0.01) mutations, whereas SMAD4 and TP53 mutations were higher in KRAS MT tumors (15% vs. 5%, p = 0.02; 68% vs. 28%, p , 0.01, respectively). KRAS MT tumors had higher expression (66% vs. 49%, p , 0.01) and amplification (2.5% vs. 0%, p = 0.04) of cMET, and higher expression of EGFR (90% vs. 82%, p = 0.04), whereas KRAS WT tumors had higher HER2 expression (2% vs. 0.4%, p , 0.01) and amplification (6% vs. 0.7%, p , 0.01). Comparing 1o (n = 1099) with Met (n = 1327) PC, 1o tumors had a higher frequency of “low” ERCC1 (81% vs. 63%, p , 0.01) and “low” RRM1 (87% vs. 77%, p , 0.01), and higher PD1+ TILs (45% vs. 34%, p , 0.01). Conversely, cMET (62% vs. 51%, p = 0.006), PDL1 (10% vs. 5%, p = 0.01), and TOPO1 (65 vs. 33, p , 0.01) were overexpressed at higher rates in Met PC. Lung Mets, had a higher expression of HER2 (2.3% vs. 0.5%; , 0.01), and PD1 TILs (50% vs. 33%; p = 0.03) than liver Mets, whereas, the PIK3CA mutation rate was higher in liver Mets (4% vs. 0%; p = 0.03). Conclusions: Genomic differences between KRAS WT vs. MT tumors suggest different carcinogenic pathways and tumor biology. 1o tumors may carry genetic alterations that are distinct from distant Mets. Mutations in druggable genes (e.g., HER-2, PIK3CA, BRCA2) may provide therapeutic opportunities.

Poster Session (Board #114), Sat, 8:00 AM-11:30 AM

Redefining the positive margin in pancreatic cancer: Impact of distance on survival. First Author: Arsen Osipov, Cedars-Sinai Medical Center, Los Angeles, CA Background: Significant controversy exists regarding prognostic significance of margin clearance and what constitutes a positive margin (R1) in pancreatic ductal adenocarcinoma (PDA). We performed a comprehensive archival analysis of all surgical resection margins (SRM) to determine the effect on locoregional recurrence and survival, and impact of adjuvant therapy in PDA. Methods: We identified 105 patients with resected stage IIII PDA from 2007-2014. SRMs for each patient were identified for pancreatic, anterior, bile duct, and posterior margins (posterior surface, uncinate and vascular groove). Three pathologists reviewed all archival surgical specimens and recategorized each margin as: tumor at ink/transected, , 0.5 mm, 0.5-1 mm, . 1-2 mm or . 2mm from the inked surface. The significance of margins and clinical variables was assessed on disease-free survival (DFS) and overall survival (OS) using multivariate cox proportional hazards modeling. Estimates for DFS and OS were calculated using KaplanMeier methods. Results: The pancreatic, anterior and bile duct SRMs were not significant (p . 0.05) predictors of DFS and OS. However, increasing posterior margin clearance up to 2 mm was a significant predictor of DFS (p = 0.001) and OS (p = 0.001). Margin clearance in excess of 2 mm did not impact clinical outcomes DFS (p = 0.08) and OS (p = 0.08). Dichotomizing the posterior SRM at 2 mm revealed it to be an independent predictor of OS (HR 0.31 p = 0.008) and DFS (HR 0.46 p = 0.05) on multivariate analysis (median OS 23.2 [ # 2mm] vs 60 [ . 2mm] months; median DFS 13.9 [ # 2mm] vs 27.3 [ . 2mm] months). Adjusting for multiple covariates, a margin status of . 2mm, was a significant predictor of OS in patients who received adjuvant chemotherapy (ACT) (HR 0.31 p = 0.03), yet this difference was mitigated in patients receiving adjuvant chemoradiotherapy (HR 0.40 p = 0.19). Conclusions: Among all SRMs in PDA, the posterior margin is the most clinically significant. Achieving SRMs of at least 2 mm has a significant impact on long term clinical outcomes. The addition of radiotherapy to ACT mitigates the negative prognostic significance of a posterior margin less than 2 mm. Comprehensive archival validation in a larger data set may help redefine R1 in PDA.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

236s 4123

Gastrointestinal (Noncolorectal) Cancer Poster Session (Board #115), Sat, 8:00 AM-11:30 AM

Overall survival of patients with pancreatic adenocarcinoma and BRCA1 or BRCA2 germline mutation. First Author: Kyaw Lwin Aung, Princess Margaret Cancer Center, Toronto, ON, Canada Background: Pathogenic BRCA1/2 germline mutations are found in 4.6% of patients with pancreatic adenocarcinoma (PDAC). Long-term overall survival (OS) of these patients is largely unknown. BRCA1/2 mutations are associated with DNA double strand break repair deficiency and more data are needed to establish whether PDAC patients with a BRCA1/2 germline mutation achieve better OS with DNA cross linking agents such as platinum. Methods: The objective was to evaluate OS including survival outcomes achieved with platinum based chemotherapy in a cohort of BRCA1/2germline mutant PDAC patients. Primary endpoint was 5 year OS for patients with stage I & II operable disease and 2 year OS for patients with inoperable stage III & IV advanced disease. Evaluable patients, diagnosed between January 1997 and December 2015, were identified from the Ontario Pancreas Cancer Study prospective database. Results: Fifty seven patients, 25 female and 32 male, were identified: 14 BRCA1 mutant, 42 BRCA2 mutant, and 1 double mutant. Median age was 63 years (range 29-84). Twenty six patients (45%) had prior cancer diagnoses (13 breast, 3 prostate, 3 bladder, 2 ovarian, 2 thyroid, 1 colon, 1 melanoma and 1 lymphoma). At a median follow up of 18 months (range 2-87), 40 (70%) have died. Twenty four (42%) had a primary tumor resection (4 stage I + 20 stage II). Their median OS was 19.5 months and 5 year OS 17% (95%CI 6-35%). Thirty three patients (58%) had advanced disease (8 inoperable stage III + 25 stage IV). Their median OS was 9.6 months and 2 year OS 24% (95%CI 11.5-43%). Twenty advanced patients received platinum (oxaliplatin, cisplatin or carboplatin) either as part of first line or subsequent lines of palliative chemotherapy. Their median OS was 15.3 months and 2 year OS 35% (95%CI 18-56%). Of remaining 13 advanced patients, treatment details were not available for one. For 12 advanced patients who did not receive platinum chemotherapy, median OS was 8.3 months and 2 year OS 0%. Conclusions: In this large cohort of germline BRCA1/2 mutant PDAC patients, OS for early stage disease was similar to historical controls, whereas OS for advanced disease was superior to historical controls, particularly for patients who received platinum based chemotherapy.

4125

Poster Session (Board #117), Sat, 8:00 AM-11:30 AM

PANOVA: A pilot study of TTFields concomitant with gemcitabine for frontline therapy of advanced pancreatic adenocarcinoma. First Author: Fernando Rivera, Hospital Universitario Marques ´ de Valdecilla, Santander, Spain Background: TTFields are alternating electric fields delivered to the region of the tumor by means of non-invasive transducer arrays surrounding the abdominal region. TTFields have an anti-mitotic effect by interfering with mitotic spindle formation, leading to apoptosis. In pancreatic cancer, TTFields decreased proliferation and clonogenic potential in vitro, and reduced tumor volume in vivo. The PANOVA trial is the first study testing TTFields in pancreatic cancer patients. Methods: Twenty advanced pancreatic cancer patients were enrolled in a prospective study of TTFields concomitant with weekly gemcitabine. All patients had unresectable tumors, an ECOG performance score of 0-1 and no prior therapy. The primary endpoint was the incidence and severity of treatment emergent adverse events. Results: The median age was 73 (range – 49-81) and most patients (80%) had an ECOG score of 1. Thirteen patients (65%) had distant metastases. Median compliance with TTFields was 78% (14 hours/day), with median duration of 5 months. Fourteen patients (70%) had serious AEs during the study period. Ten patients (50%) had treatment-related skin toxicity, of which only 2 were grade 3, both resolved with appropriate treatment. No TTFields-related serious AEs were reported. The median PFS was 8.3 months (95% CI 4.3, 10.3): 10.3 months (95% CI 2.8, NA) in locally advanced patients and 5.7 months (95% CI 3.8 - 10.3) in patients with metastatic disease. PFS rate at 6 months was 56%. Of the evaluable tumors, 30% had partial response and another 30% stable disease. The median OS for all patients was 14.9 months: It was not reached after 15 months in locally advanced patients, and was 8.3 months (95% CI 4.314.9) in patients with metastatic disease. The 1-year survival rate was 55% (86% in locally advanced and 40% in metastatic disease). Conclusions: TTFields concomitant to gemcitabine are tolerable and safe for advanced pancreatic cancer patients, with promising clinical outcome compared to published data for gemcitabine alone. An extension of the current protocol is testing patients treated with gemcitabine, nab-paclitaxel and TTFields. Clinical trial information: NCT01971281.

4124

Poster Session (Board #116), Sat, 8:00 AM-11:30 AM

Impact of second-line treatment (2L T) in advanced pancreatic cancer (APDAC) patients (pts) receiving first line Nab-Paclitaxel (nab-P) + Gemcitabine (G): An Italian multicentre real life experience. First Author: Guido Giordano, Oncologia Medica, Ospedale Sacro Cuore di Gesu’ Fatebenefratelli, Benevento, Italy Background: There is no standard of care for 2L T in APDAC. Nab-P + G combination has showed superior efficacy compared to G alone in MPACT phase III study. Here we report data of a retrospective analysis evaluating outcomes of 2L T following first line Nab-P + G. Methods: Clinical records of 250 APDAC were retrospectively reviewed, evaluating survival outcomes in pts receiving 2L T according each treatment schedule. OS and PFS were estimated by Kaplan-Meier method with 95% CI. Pearson correlation analysis was performed to investigate correlation between first line outcomes and 2L T response or survival. Results: At time of data analysis among 221 pts who experienced disease progression 122 (55%) received 2L T and 99 (45%) best supportive care (BSC). Baseline characteristics of pts receiving 2L T were: median age 66 (range 39-79), M/F:66/56, ECOG PS 0/1/2: 41/ 55/26. 35.5% had multiple organ involvement and 28.7% had received adjuvant treatment. In 2L T pts FOLFOX/XELOX, FOLFIRI, FOLFIRINOX (classic or modified) and single agent therapies were used in45%, 22%, 18% and 15% respectively. Median OS in pts receiving 2L T was 13.5 months (95% CI 12.659-14.341) vs 6.8 months in pts receiving BSC (95% CI 5.567-8.033), p , 0.0001. According to 2L T schedule median OS was 12.9 (95% CI 11.772-14.028), 13.2 (95% CI 10.909-15.491), 13.8 (95% CI 11.069-16.531), 12.3 (95% CI 10.620-13.980) in pts FOLFOX/ XELOX, FOLFIRI, FOLFIRINOX (classic or modified) and other single agent therapies respectively, with no significant differences between each group. A trend to significant correlation was observed between response to first line treatment and response to 2L T (Pearson correlation 0.298, Sig 2 tails p = 0.06) as well as between longer first line PFS ( $ 6 months) and 2L PFS $ 4 months (Pearson correlation 0.321, Sig 2 tails p = 0.002). Conclusions: Our data show that 2L T is an option to improve outcome for more than a half of pts following first line Nab-P + G. Both response and PFS during first line Nab-P + G treatment may influence 2L T outcomes.

4126

Poster Session (Board #118), Sat, 8:00 AM-11:30 AM

Updated overall survival (OS) analysis of NAPOLI-1: Phase 3 study of nanoliposomal irinotecan (nal-IRI, MM-398), with or without 5-fluorouracil and leucovorin (5-FU/LV), vs 5-FU/LV in metastatic pancreatic cancer (mPAC) previously treated with gemcitabine (gem)-based therapy. First Author: Andrea Wang-Gillam, Siteman Cancer Center, Washington University, Saint Louis, MO Background: NAPOLI-1 is a global, randomized Phase 3 study evaluating nal-IRI—a nanoliposomal irinotecan—with or without 5-FU/LV vs 5-FU/LV in 417 mPAC patients previously treated with gem-based therapy. Primary survival analysis was based on 313 events. nal-IRI+5-FU/LV significantly improved OS (primary endpoint), 6.1 months (mo) vs 4.2 mo; with 5-FU/LV (unstratified hazard ratio [HR] = 0.67; P = 0.012). The primary endpoint was supported by improved progression-free survival, time to treatment failure, objective response and CA19-9 response rates, and manageable toxicities. An updated analysis of OS, 6- and 12-month-survival estimates, and safety is presented. Methods: The updated descriptive analysis of OS, based on 378 events (25 May 2015), includes data from all randomized patients across the 3 arms. The final OS analysis will also be presented. Results: After 378 OS events, nal-IRI+5-FU/LV (n = 117) retained an OS advantage relative to 5FU/LV (n = 119): 6.2 mo (95% confidence interval [CI], 4.8–8.4) vs 4.2 mo (95% CI, 3.3–5.3) with an unstratified HR of 0.75 (P = 0.0417). In contrast, there was no OS advantage with nal-IRI monotherapy (n = 151) vs 5-FU/LV (n = 149): 4.9 mo [95% CI, 4.2–5.6] vs 4.2 mo [95% CI, 3.6–4.9], HR = 1.08; P = 0.5. Six-month survival estimates were 53% (95% CI, 44–62%) for nalIRI+5-FU/LV vs 38% (95% CI, 29–47%) for 5-FU/LV; 12-month survival estimates were 26% (95% CI, 18–35%) for nal-IRI+5-FU/LV vs 16% (95% CI, 10–24%) for 5-FU/LV. With events in nearly all patients, the OS curves converge at ~20 mo with 19 patients (16.2%) surviving beyond 20 mo. This is a reason for attenuation of the HR estimate and unstratified log rank pvalue. The most common $ grade 3 adverse events occurring at a $ 2% incidence in nal-IRI-containing arms were neutropenia, diarrhea, vomiting, and fatigue. Conclusions: In an updated analysis, the median OS benefit for nal-IRI+5-FU/LV over 5-FU/LV was maintained with a similar safety profile. nal-IRI+5-FU/LV may be a new standard of care for mPAC patients previously treated with gem-based therapy. Clinical trial information: NCT01494506.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Noncolorectal) Cancer 4127

Poster Session (Board #119), Sat, 8:00 AM-11:30 AM

Genomic profiling of pancreas ductal adenocarcinoma (PDA), actionability, and correlation with clinical phenotype. First Author: Emmet Jordan, Memorial Sloan Kettering Cancer Center, New York, NY Background: Molecular profiling (MP) in cancer has enabled identification of potential actionable drug targets that has prompted attempts to discover clinically validated biomarkers and driver alterations for therapeutic clinical trials. This prospective study assessed if comprehensive genetic analysis of archival paraffin embedded PDA tissue is feasible within a clinically relevant timeframe and can provide predictive and/or prognostic information along with potential targets for therapy. Methods: Archival FFPE samples from patients (pts) who consented prospectively to study were analyzed using the MSK-IMPACT platform, a hybridization capture based next generation sequencing assay for targeted deep sequencing of all exons and selected introns of 410 key cancer genes. Tumor and matched normal libraries were sequenced on an Illumina HiSeq 2500. Sequencing output was processed using a custom analysis pipeline. Demographic and treatment data were prospectively collected. Fisher’s exact tests were performed to look for associations between clinical characteristics and genetic alterations. Results: N = 211 PDA pts were enrolled from Mar 2014-Jan 2016. We report results from the first 201 pts (table 1). N = 121 had curative surgery, 51% remain disease free at time of analysis. CDKN2B (p = 0.019), RB1 (p = 0.019), MDC1 (p = 0.042) and TGFBR2 (p = 0.042) were associated with disease recurrence (DR) post resection. 55% pts were never smokers, SMAD3 mutations (MUT) were seen only in smokers (p = 0.04). RECQL4 MUT were more common in pts , 50 yrs (p = 0.0002). TGFBR2 MUT were more common in metastatic disease (p = 0.009). For the 73 pts who received 1st line platinum, CDKN2A/ 2B deletions (DEL) were only observed with stable disease/partial response as compared to progressive disease (p = 0.10, 0.18). Conclusions: MP is feasible in PDA pts in a clinically relevant timeframe using FFPE tissue from both primary and metastatic sites. We identified MUT associated with DR, age at diagnosis (dx), platinum response and smoking. N/% Stage at dx: I/II III IV Mean MUT per sample Most frequent MUT KRAS TP53 CDKN2A SMAD4 ARID1A Most frequent CNA CDKN2A/B DEL SMAD4 DEL ERBB2 AMP

4129

101 (50%) 31 (15%) 69 (34%) 4 91% 68% 21% 16% 12%

4128

237s Poster Session (Board #120), Sat, 8:00 AM-11:30 AM

A phase Ib extension study of cancer stemness inhibitor BB608 (napabucasin) in combination with gemcitabine and nab-paclitaxel (nab-PTX) in patients (pts) with metastatic pancreatic cancer. First Author: Bassel F. ElRayes, Winship Cancer Institute of Emory University, Atlanta, GA Background: BB608 (aka napabucasin, BBI-608) is an oral first-in-class cancer stemness inhibitor that works by targeting Stat3-driven gene transcription. Preclinical anti-tumor activity was observed in vitro and in vivo. BB608 has shown clinical safety and encouraging signs of anti-cancer activity in multiple cancer types in phase I/II studies. Methods: A phase Ib extension, open label, multi-center study in pts with metastatic pancreatic ductal adenocarcinoma (mPDAC) was undertaken to confirm the RP2D, PK profile and signs of anticancer activity of BB608 in combination with gemcitabine and nab-PTX. BB608 was administered at 240 mg BID in combination with gemcitabine 1000 mg/m2 and nab-PTX 125 mg/m2 administered weekly for 3 out every 4 weeks until progression of disease, unacceptable toxicity, or other discontinuation criterion. Results: Of 37 pts enrolled, 31 (84%) were treatment-na¨ıve and 6 (16%) received neoadjuvant systemic therapy. There were no dose-limiting or unexpected toxicities. Most common adverse events (AE’s) related to BB608 included grade 1 diarrhea, abdominal pain, nausea, and fatigue with grade 3 adverse events observed in 6 pts: 1 pt (diarrhea), 1 pt (dehydration), 3 pts (fatigue, hypokalemia) and 1 pt (hyponatremia). No significant pharmacokinetic interactions were observed. Disease control (PR+SD) was observed in 27 of 29 evaluable pts (93%), with 23 pts (79.3%) with tumor regression of which 10 pts (34.5%) achieved PR (RECIST 1.1: 33 - 78% regression). Of the 8 pts who were nonevaluable for treatment response, 3 stopped treatment due to clinical progression, 1 due to non-compliance, and 4 withdrew consent. Among 37 pts enrolled (intent-to-treat), disease control (PR+SD) was observed in 27 pts (73%), with tumor regression observed in 23 pts (62.2%) of which 10 pts achieved PR (27%). Conclusions: This phase Ib extension study confirmed that BB608 can be combined with gemcitabine and nab-PTX and shows encouraging signs of anti-tumor activity in pts with pancreatic cancer. Clinical trial information: NCT02231723.

14% 3% 2%

Poster Session (Board #121), Sat, 8:00 AM-11:30 AM

4130

Poster Session (Board #122), Sat, 8:00 AM-11:30 AM

Breast cancer resistance protein (ABCG2) as a potential biomarker for gemcitabine and nab-paclitaxel sensitivity. First Author: Kenneth H. Yu, Memorial Sloan Kettering Cancer Center, Weill Cornell Medical College, New York, NY

A randomized phase 2 study of the Bruton tyrosine kinase (Btk) inhibitor acalabrutinib alone or with pembrolizumab for metastatic pancreatic cancer (mPC). First Author: Michael J. Overman, The University of Texas MD Anderson Cancer Center, Houston, TX

Background: PDAC is a leading and rising cause of cancer mortality. Despite this, modern cytotoxics (C) can induce tumor responses and extend life. Breast cancer resistance protein (ABCG2) is an ATP-binding cassette (ABC) transporter identified as a molecular cause of multidrug resistance in diverse cancer cells. We have previously shown that expression profiling of circulating tumor and invasive cells (CTICs) can predict response and resistance to C, predominantly 5-FU based. The current study explores the ability to predict treatment response based on the expression of ABCG2 in a uniform population of patients receiving gemcitabine (G) and nab-paclitaxel (Nab). Methods: 33 patients receiving frontline treatment with G and Nab for treatment of metastatic PDAC were enrolled into a prospective trial from Jan 2014 to the present. Six mL of peripheral blood was collected prior to treatment, after 2 months of treatment and at disease progression. CTICs were isolated using a cell-invasion assay, total RNA extracted and expression of 83 genes used as candidate biomarkers of drug response were measured by qPCR. Patients were followed for PFS and OS. Results: CTICs were isolated, and gene expression profiles were obtained in all 33 patients prior to initiating C and in 21 patients at 1st line disease progression. Expression of ABCG2 predicts treatment response and progression free survival. 12/33 (36%) patients were ABCG2 high and 7/33 (21%) were ABCG2 low, the remainder were intermediate. Clinical benefit was seen for patients with low v high ABCG2 expression (PFS 5 mo v 3 mo, p = 0.028, HR 0.15; OS 8.6 mo v 3.4 mo, p = 0.020). Conclusions: Gene expression profiling of CTICs from patients with metastatic PDAC treated with G and Nab identifies ABCG2 expression as a candidate biomarker of response and prognosis. A prospective study to validate these findings is warranted. Clinical trial information: NCT01474564.

Background: The importance of the tumor microenvironment in the biology of mPC has been increasingly recognized. However, immune checkpoint blockade has previously shown limited clinical responses in mPC. Preclinical studies in PC indicate promising anti-tumor effect from acalabrutinib alone and in combination with pembrolizumab. Methods: This was a Phase 2, multicenter, open-label, randomized (1:1; n = 38/arm) study (NCT02362048) evaluating the Btk inhibitor, acalabrutinib (ACP-196) alone (mono) or in combination (combo) with pembrolizumab. This study included a safety run-in that assessed the first 6 patients (pts) enrolled in the combo arm for dose-limiting toxicity (DLT). Adults with histologically confirmed pancreatic ductal adenocarcinoma who had received $ 1 prior systemic therapy for mPC were eligible. Acalabrutinib 100 mg PO BID was administered days 1-21 and pembrolizumab 200 mg IV day 1 of each 3-week cycle. Response was assessed every 2 cycles using RECIST 1.1. Blood was collected for exploratory correlative analyses. Results: Preliminary results are provided for the first 58 treated pts (26 mono arm, 32 combo arm). The median number of prior treatments was 2. No DLTs occurred. Grade 3-4 events that occurred in $ 2 pts were dehydration (12%), anemia (12%) and hypotension (8%) on the mono arm and were anemia (9%) and abdominal pain (9%) on the combo arm. No Grade 5 treatment-related events were reported. 44 pts (21 mono arm, 23 combo arm) were evaluable for response. In the mono arm, 4 stable disease (SD) were reported. In the combo arm, 3 partial responses (2 PR, 1 PRu), including 1 pt with near resolution of bulky peritoneal disease, and 5 SD were reported. All 3 PRs occurred in pts with a strong family history of pancreatic or breast cancer. Germline testing identified BRCA2VUS in one responder. At the time of the data cut, the median time on study for pts with responses was 3.7 months (range 1.4-4.1). Conclusions: The combination of acalabrutinib and pembrolizumab has a favorable benefit/risk profile and encouraging preliminary antitumor activity in pretreated mPC, particularly in a subpopulation of pts with familial mPC. Clinical trial information: NCT02362048.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

238s TPS4131

Gastrointestinal (Noncolorectal) Cancer Poster Session (Board #123a), Sat, 8:00 AM-11:30 AM

TPS4132

Poster Session (Board #123b), Sat, 8:00 AM-11:30 AM

ESOPEC: A prospective randomized controlled multicenter phase III trial comparing perioperative chemotherapy to neoadjuvant chemoradiation in patients with adenocarcinoma of the esophagus (NCT02509286). First Author: Jens Hoeppner, University of Freiburg - Medical Center, Freiburg, Germany

A phase 3 randomized, double-blind, placebo-controlled study to evaluate the efficacy and safety of GS-5745 combined with mFOLFOX6 as first-line treatment in patients with advanced gastric or gastroesophageal junction adenocarcinoma. First Author: Johanna C. Bendell, Sarah Cannon Research Institute, Nashville, TN

Background: Large randomized controlled trials (RCT) comparing neoadjuvant chemoradiation (neoCRT) (CROSS trial) or perioperative chemotherapy (periCTX) (MAGIC/FFCD 9703 trials) plus surgery with surgery alone have been carried out in non-metastasized esophageal adenocarcinoma. These RCT showed significant survival benefits for combined modality treatment. As prospective head-to-head comparisons of neoCRT versus periCTX applying contemporary treatment regimens are lacking, the present trial was started to define a future standard of care. Methods: This investigator-initiated multicenter prospective two-arm RCT is comparing the efficacy of neoCRT (CROSS regimen) followed by surgery versus periCTX and surgery (FLOT regimen) for the treatment of localized esophageal and junctional adenocarcinoma. Eligibility for the trial is given in cT1 cN+ cM0 and cT2-4a cNx cM0 tumors. The trial will include 438 participants who are 1:1 randomized to one of the two study arms and stratified by N-stage and study site. Patients enrolled in the neoCRT study arm will receive preoperative radiation therapy (41.4Gy in 23 fractions of 1,8Gy: days 1-5, days 8-12, days 15-19, days 22-26 and days 29-31) and concurrent carboplatin (2mg/ml/min) and paclitaxel (50mg/m2) on day 1, 8, 15, 22, 29. Patients in the periCTX study arm will receive 4 preoperative cycles and 4 postoperative cycles of 5-FU 2600 mg/m² (24h), d1 leucovorin 200 mg/m², d1 oxaliplatin 85 mg/m², d1 docetaxel 50mg/m2, d1 every two weeks (q2w). The primary endpoint of the trial is overall survival. Secondary endpoints are progressionfree survival, recurrence-free survival, site of failure, postoperative morbidity and mortality, duration of hospitalization and quality of life. Biological substudies on biomarkers of esophageal adenocarcinoma are included. Current status: The trial was approved by the competent ethics committee and the German Federal Institute for Drugs and Medicine Devices. Patient recruitment has begun in Feburary 2016. Overall 18 German sites will commence recruitment in 2016. Clinical trial information: NCT02509286.

Background: GS-5745 is a monoclonal antibody that inhibits matrix metalloproteinase 9 (MMP9), an extracellular enzyme involved in matrix remodeling, tumor growth, and metastasis. Inhibiting MMP9 is expected to block paracrine signaling and metastasis and to alter the immune microenvironment within the tumor. GS-5745 (800 mg every 2 weeks) with mFOLFOX6 was examined in a Phase 1b study in 40 patients with gastric and GEJ adenocarcinoma (GS-US-296-0101), and demonstrated encouraging activity without added toxicity. Additionally, decreased serum biomarkers suggested inhibition of MMP9 enzymatic activity by these agents. These data support the hypothesis that GS-5745 treatment inhibits MMP9 activity and that the inhibition may lead to improved clinical outcomes. Methods: This phase 3, randomized, double-blind, multicenter study investigates the efficacy and safety of GS-5745 combined with mFOLFOX6 in subjects with untreated gastric and GEJ adenocarcinoma. Total of 430 eligible subjects with advanced gastric and GEJ cancer will be randomized in a 1:1 manner to mFOLFOX6 plus GS-5745 or mFOLFOX6 plus placebo. Stratification factors include ECOG status (0 v 1), geographic region (Latin America v All other countries), and primary tumor site (gastric v GEJ). CT or MRI scans will be performed every 8 weeks to evaluate response to treatment. mFOLFOX6 will be administered on Days 1 and 15 of each 28-day treatment cycle for a total of 6 cycles followed thereafter by leucovorin (LV) and 5-fluorouracil (5-FU) dosing on Days 1 and 15 of each 28-day treatment cycle until disease progression. GS-5745/placebo 800 mg will be infused on Days 1 and Day 15 of each 28 day cycle until disease progression. Primary endpoint is OS, and secondary endpoints include PFS, ORR (RECIST 1.1), and safety. The study is designed to have an 85% power to detect clinically meaningful improvement in overall survival at the one-sided significance level of 0.025. The association of exploratory biomarkers with study drug response will also be evaluated. Enrollment opened Oct. 2015. Clinical trial information: NCT02545504.

TPS4133

TPS4134

Poster Session (Board #124a), Sat, 8:00 AM-11:30 AM

Poster Session (Board #124b), Sat, 8:00 AM-11:30 AM

EORTC-1203: Integration of trastuzumab (T), with or without pertuzumab (P), into perioperative chemotherapy (CT) of HER-2 positive stomach cancer—INNOVATION trial. First Author: Anna Dorothea Wagner, Department of Oncology, Lausanne University Hospital, Lausanne, Switzerland

Maintenance therapy with avelumab (MSB0010718C; anti-PD-L1) vs continuation of first-line chemotherapy in patients with unresectable, locally advanced or metastatic gastric cancer: The phase 3 JAVELIN Gastric 100 trial. First Author: Markus H. Moehler, Johannes-Gutenberg University Mainz, Mainz, Germany

Background: Approximately 10-20% of patients with gastric cancer (GCa) have HER-2 positive tumors. The addition of T to cisplatin/fluoropyrimidinebased CT improved survival in metastatic HER-2 positive GCa. When P was added to T and CT, a significant increase in the histopathological complete response rate was observed in HER-2 positive breast cancer. This study aims to investigate the added value of combining both HER-2 targeting drugs with perioperative CT for GCa and gastroesophageal-junction cancer (GEJCa). Methods: This is a randomized, open-label phase II trial. Biopsies from patients with resectable GCa or GEJCa (UICC tumor stage Ib-III) will be centrally screened for HER-2. Over a 4-year accrual period, 225 patients with HER-2 positive cancers will be centrally randomized (1:2:2 ratio) across 52 centers in 10 European and 3 non-European countries in one of the three arms: - Control arm: Cisplatin (80 mg/m2 every 3 weeks) and capecitabine (1000 mg/m2 twice daily every 2 out of 3 weeks), or 5-FU (800 mg/m2/day for 5 days every 3 weeks) for 3 cycles before and after surgery - Experimental arm 1: CT plus T (8 mg/kg loading dose, followed by 6mg/kg every 3 weeks) Experimental arm 2: CT plus T plus P (840mg every 3 weeks) T and/or P will be administered on day 1 of every CT cycle and continued afterwards for a total of 17 cycles. Stratification will be done by histological subtype (intestinal/non-intestinal); region (Korea vs Europe); location (GEJCa vs nonGEJCa) and HER-2 (immunohistochemistry 3+ vs IHC 2+/FISH+). The primary objective is to detect an increase in the pathological response rate (RR, , 10% residual tumor cells) as determined by central review, from 15% to 30% with addition of T or T and P. The RR in each experimental arm will be tested versus 15% with a one-sided type I error of 5%. If both tests are positive, the experimental arm with the higher RR will be recommended if the difference exceeds 5%. The recruitment has started in September 2015. Clinical trial information: NCT02205047.

Background: Programmed death-1 receptor ligand (PD-L1) is a key therapeutic target in the reactivation of the immune response against multiple cancers. Avelumab* is a fully human anti-PD-L1 IgG1 antibody that has shown promising efficacy and an acceptable safety profile in patients (pts) with adenocarcinoma of the stomach or gastroesophageal junction (AS/GEJ) treated in first-line (1L) and second-line settings. This open-label phase 3 study (NCT02625610) compares maintenance (Mn) treatment with singleagent avelumab vs continuation of 1L chemotherapy in pts with AS/ GEJ. Methods: The primary objective of this global, multicenter trial is to demonstrate superiority of Mn therapy with avelumab vs continuation of 1L chemotherapy. Primary endpoint is overall survival or progression-free survival. Approximately 666 eligible pts will receive induction chemotherapy and upon completion, approximately 466 pts without disease progression will be randomized to receive treatment in the Mn phase. Main eligibility criteria include: histologically confirmed unresectable locally advanced or metastatic (LA/M) AS/GEJ, ECOG PS 0-1, no prior chemotherapy for LA/M disease, no prior therapy with any drug targeting T cell coregulatory proteins, and no concurrent anticancer treatment or immunosuppressive agents. Pts are not preselected for PD-L1 expression; HER2+ pts are excluded. During the induction phase, pts receive chemotherapy (oxaliplatin + 5-fluorouracil + leucovorin or oxaliplatin + capecitabine) for 12 wks. Pts entering the Mn phase are randomized to receive either avelumab 10 mg/kg as a 1h intravenous infusion Q2W or continuation of 1L chemotherapy. Treatment is given until disease progression, unacceptable toxicity, or consent withdrawal. Secondary endpoints include best overall response, quality of life (assessed via EQ-5D-5L, EORTC QLQ-C30, and EORTC QLQ-STO22), safety as per NCI-CTCAE v4.03, and tumor biomarkers. Responses are evaluated according to RECIST 1.1 and adjudicated by a blinded independent review committee. Trial enrollment began in Dec 2015. *Proposed INN. Clinical trial information: NCT02625610.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Noncolorectal) Cancer TPS4135

Poster Session (Board #125a), Sat, 8:00 AM-11:30 AM

TPS4136

239s Poster Session (Board #125b), Sat, 8:00 AM-11:30 AM

Avelumab (MSB0010718C; anti-PD-L1) + best supportive care (BSC) vs BSC 6 chemotherapy as third-line treatment for patients with unresectable, recurrent, or metastatic gastric cancer: The phase 3 JAVELIN Gastric 300 trial. First Author: Yung-Jue Bang, Seoul National University Hospital, Seoul, Korea, The Republic of

A randomized, multicenter, controlled study to compare perioperative chemotherapy of oxaliplatin combined with TS-1 (SOX) versus SOX or oxaliplatin with capecitabine (XELOX) as post-operative chemotherapy in locally advanced gastric adenocarcinoma with D2 dissection (RESOLVE Trial). First Author: Jiafu Ji, Beijing Cancer Hospital, Beijing, China

Background: Programmed death-1 receptor ligand (PD-L1) is a key therapeutic target in the reactivation of the immune response against multiple cancers. Avelumab* is a fully human anti-PD-L1 IgG1 antibody that has shown promising efficacy and an acceptable safety profile in multiple tumor types, including adenocarcinoma of the stomach or gastroesophageal junction (AS/GEJ). This open-label phase 3 trial (NCT02625623) compares avelumab + best supportive care (BSC) vs BSC 6 chemotherapy as third-line treatment for patients (pts) with AS/GEJ. Methods: The primary objective of this global, multicenter trial is to demonstrate superiority, defined by overall survival, of avelumab + BSC vs BSC 6 chemotherapy. Approximately 330 pts stratified by region (Asia vs non-Asia) will be randomized. Main eligibility criteria include: histologically confirmed unresectable locally advanced or metastatic AS/GEJ, fresh or archival tumor tissue for PD-L1 expression assessment, ECOG PS 0-1, 2 prior lines of systemic treatment, no prior therapy with any drug targeting T cell coregulatory proteins, and no concurrent anticancer treatment or immunosuppressive agents. Pts are not preselected for PD-L1 expression. Pts receive either BSC with avelumab 10 mg/kg as a 1h intravenous infusion Q2W or BSC 6 chemotherapy (physician’s choice of irinotecan 150 mg/m2 Q2W or paclitaxel 80 mg/m2 weekly for 3 out of 4 weeks, in a 4-week treatment cycle for pts eligible to receive chemotherapy). Pts not eligible for chemotherapy will receive BSC only. Treatment is given until disease progression, unacceptable toxicity, or consent withdrawal. Secondary endpoints include progression-free survival, objective response rate, quality of life (assessed via EQ-5D-5L, EORTC QLQC30, and EORTC QLQ-STO22), safety as per NCI-CTCAE v4.03, and tumor biomarkers. Responses are evaluated according to RECIST 1.1 and adjudicated by a blinded independent review committee. Trial enrollment began in Dec 2015. *Proposed INN. Clinical trial information: NCT02625623.

Background: Perioperative treatment of locally advanced gastric cancer (LAGC) has always been argued by eastern and western scholars. MAGIC study has showed that peri-operative chemotherapy with 3 cycles of ECF has increased 13% on 5yOS compared with surgery alone; however, ACTS-GC or CLASSIC showed adjuvant chemotherapy after D2 dissection could achieve the similar survival benefit. So in this prospective randomized phase III study, we aims to compare the survival benefit as well as the safety for SOX (oxaliplatin/TS-1) as periopertative therapy versus SOX or XELOX as postoperative therapy after D2 dissection. Methods: RESOLVE Trial (ClinicalTrials.gov, NCT01534546) is a 3-arm, randomized, multi-center, openlabel phase III trial. Primary endpoint is 3yDFS, and secondary endpoints include 5yOS, R0 resection rate and safety. Main eligibility criteria include gastric or gastro-esophageal junction adenocarcinoma, clinical stages cT4b/ N+M0 or cT4aN+M0.Patients are randomized 1:1:1 with 2 strata: site and Lauren classification (diffuse or intestinal type).We anticipate that 3yDFS of perioperative chemotherapy of SOX is superior than postoperative XELOX, and 3y DFS of SOX is non-inferior to XELOX as adjuvant chemotherapy, a = 0.05, with power of 80%; number of subjects per treatment arm is 353, totally 1059. Arm A and Arm B will receive standard gastrectomy with D2 Lymphadenectomy first and 8 cycles of adjuvant XELOX or SOX later. [XELOX: capecitabine 1000 mg/m2, bid, d1~14 q3W; oxaliplatin 130mg/ m2,d1, q3W; SOX: TS-1: 40~60mg bid,d1~14 q3W oxaliplatin F130mg/m2 d1,q3W]. Arm C will receive 3 cycles of neoadjuvant SOX first, and then standard gastrectomy with D2 lymphadenectomy, and 5 cycles of adjuvant SOX followed by 3 cycles of S-1 monotherapy. First patient entered into the study in Aug 2012. As of Dec 31, 2015, a total of 823 subjects were screened and 769 patients were recruited by 27 Chinese centers. Up to now, there were 251 patients are on treatment and 86 patients have died. Clinical trial information: NCT01534546.

TPS4137

TPS4138

Poster Session (Board #126a), Sat, 8:00 AM-11:30 AM

Pembrolizumab versus paclitaxel as second-line therapy for advanced gastric or gastroesophageal junction (GEJ) adenocarcinoma: Phase 3 KEYNOTE-061 study. First Author: Atsushi Ohtsu, National Cancer Center Hospital East, Kashiwa, Japan Background: Pembrolizumab (pembro) is a humanized monoclonal antibody against PD-1 that prevents binding of PD-1 to PD-L1 and PD-L2 and permits activation of an antitumor immune response. In KEYNOTE-012, pembro showed a manageable safety profile and a 22% ORR in pts with advanced gastric cancer. The randomized, open-label, phase 3 KEYNOTE-061 study (NCT02370498) is designed to compare the efficacy and safety of pembro with those of standard-of-care paclitaxel in the second-line treatment of advanced gastric cancer. Methods: KEYNOTE-061 is designed for patients with metastatic or unresectable gastric or GEJ adenocarcinoma that progressed after first-line treatment with platinum + fluoropyrimidine doublet chemotherapy. Patients with HER2/neu-positive tumors are eligible if they have documented progression on a regimen that also included trastuzumab. Other key eligibility criteria include measurable disease per RECIST v1.1, ECOG PS 0-1, no chemotherapy within 2 wk of first dose of study drug, and provision of a newly obtained or archival tumor sample for assessing PD-L1 status. Eligible pts are randomized 1:1 to receive pembro 200 mg Q3W or paclitaxel 80 mg/m2 IV on days 1, 8, and 15 of each 28-d cycle. Treatment will continue until disease progression, intolerable toxicity, refusal by pt or investigator, or completion of 24 mo of therapy (pembro arm only). Pts in the pembro arm who have a CR after $ 24 wk may discontinue after $ 2 doses following initial CR. Clinically stable pts who progress per RECIST v1.1 may continue pembro at the discretion of the investigator until a confirmatory CT scan performed $ 4 wk later. Response will be assessed every 6 wk per RECIST v1.1 by central imaging vendor review and per RECIST adapted for immunotherapy response patterns. AEs will be assessed throughout treatment and for 30 d thereafter (90 d for serious AEs). Primary efficacy end points are PFS per RECIST v1.1 and OS in pts with PD-L1+ tumors. Secondary end points include PFS and OS in all pts, time to progression, ORR, and duration of response. Enrollment in KEYNOTE-061 is ongoing and will continue until up to 720 pts are enrolled. Clinical trial information: NCT02370498.

Poster Session (Board #126b), Sat, 8:00 AM-11:30 AM

KEYNOTE-062: Phase 3 study of pembrolizumab alone or in combination with chemotherapy versus chemotherapy alone as first-line therapy for advanced gastric or gastroesophageal junction (GEJ) adenocarcinoma. First Author: Josep Tabernero, Vall d’Hebron University Hospital, Barcelona, Spain Background: Pembrolizumab (pembro) is a monoclonal antibody against PD1 designed to block its interaction with PD-L1 and PD-L2 and permit an antitumor immune response. In KEYNOTE-012, pembro showed a 22% ORR (RECIST v1.1, central review) and manageable safety profile in patients (pts) with advanced gastric cancer. The randomized, phase 3 KEYNOTE-062 study (NCT02494583) is designed to compare the efficacy and safety of pembro alone or in combination with cisplatin + fluoropyrimidine with those of cisplatin + fluoropyrimidine as first-line therapy for PD-L1+/HER2– advanced gastric or GEJ adenocarcinoma. Methods: Key eligibility criteria include age $ 18 y, locally advanced/metastatic PD-L1+/HER2– gastric or GEJ adenocarcinoma, ECOG PS 0-1, no active autoimmune disease, brain metastases, or prior therapy for advanced disease. Pts are randomized 1:1:1 to pembro 200 mg Q3W (arm 1), pembro + cisplatin 80 mg/m2 Q3W + 5fluorouracil (5-FU) 800 mg/m2 on days 1-5 of each Q3W cycle (arm 2), or placebo Q3W + cisplatin + 5-FU (arm 3); 5-FU may be replaced with capecitabine 1000 mg/m2 twice daily on days 1-14 of each cycle. Randomization is stratified by region (Europe/North America/Australia vs Asia vs rest of world), disease status (locally advanced vs metastatic), and chosen fluoropyrimidine (5-FU vs capecitabine). Arm 1 is open label; in arms 2 and 3, assignment to pembro vs placebo is double blind. All treatments will continue for 35 cycles or until progressive disease, unacceptable toxicity, or pt/investigator decision. Response will be evaluated every 6 wk per RECIST v1.1 by central imaging vendor review and per RECIST adapted for immunotherapy response patterns; eligible pts may continue treatment beyond initial RECIST-defined progression. AEs will be assessed throughout treatment and for 30 d thereafter (90 d for serious AEs) and graded per NCI CTCAE v4.0. Pts will be followed for survival every 3 mo. OS and PFS per RECIST v1.1 are the primary study end points; secondary end points include ORR and duration of response. Enrollment in KEYNOTE-062 is ongoing and will continue until ~750 pts have enrolled. Clinical trial information: NCT02494583.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

240s TPS4139

Gastrointestinal (Noncolorectal) Cancer Poster Session (Board #127a), Sat, 8:00 AM-11:30 AM

TPS4140

Poster Session (Board #127b), Sat, 8:00 AM-11:30 AM

Pembrolizumab for previously treated metastatic adenocarcinoma or squamous cell carcinoma of the esophagus: Phase 2 KEYNOTE-180 study. First Author: Manish A. Shah, Weill Cornell Medical College and New York Presbyterian Hospital, New York, NY

KEYNOTE-181: Phase 3, open-label study of second-line pembrolizumab vs single-agent chemotherapy in patients with advanced/metastatic esophageal adenocarcinoma. First Author: Toshihiko Doi, National Cancer Center Hospital East, Chiba, Japan

Background: PD-L1 is frequently overexpressed in esophageal cancer. Pembrolizumab is a humanized monoclonal antibody that targets the PD-1 receptor and blocks interaction with PD-L1 and PD-L2. In the multicohort, phase 1b KEYNOTE-028 trial, pembrolizumab showed manageable toxicity, a 30.4% ORR, and median duration of response not reached in 23 patients (pts) with PD-L1+ advanced esophageal cancer. The multicenter, phase 2 KEYNOTE-180 trial (NCT02559687) is designed to further evaluate pembrolizumab as a monotherapy in pts with previously treated advanced/metastatic esophageal cancer. Methods: Key eligibility criteria include age $ 18 y, advanced/metastatic adenocarcinoma or squamous cell carcinoma of the esophagus or advanced/metastatic Siewert type I adenocarcinoma of the esophagogastric junction (EGJ), measurable disease, documented progression during or after 2 prior standard lines of therapy, ECOG PS 0-1, no active autoimmune disease or brain metastases, and provision of a tumor sample for retrospective biomarker analysis. Pts with metastatic Siewert type I EGJ adenocarcinoma must have known HER2 status and, if HER2+, must have documented progression on treatment containing trastuzumab. Eligible pts are to receive pembrolizumab 200 mg Q3W for 35 cycles (~2 y) or until progression, unacceptable toxicity, or investigator or pt decision. Response is to be assessed every 9 wk per RECIST v1.1 and RECIST adapted for immunotherapy response patterns. Treatment may be discontinued for pts who have a CR, and eligible pts may continue treatment beyond initial RECIST-defined progression. AEs will be assessed throughout treatment and for 30 d thereafter (up to 90 d for serious AEs) and graded per NCI CTCAE v4.0. Pts are to be followed for survival every 9 wk. The primary efficacy end point is ORR per RECIST v1.1 by central imaging vendor review. Secondary end points include PFS, OS, and duration of response. Exploratory analyses include evaluation of immune-related gene expression profiles and PD-L1 expression status as predictors of pembrolizumab efficacy. Enrollment in KEYNOTE-180 began in December 2015 and will continue until approximately 100 pts are enrolled. Clinical trial information: NCT02559687.

Background: Tumors can hijack the PD-1 pathway to suppress immune control. Overexpression of the PD-1 ligand PD-L1 in esophageal cancer may be associated with a poor prognosis. Pembrolizumab (pembro) is a monoclonal antibody that can reinstate the antitumor autoimmune response by blocking the PD-1–ligand interaction. It demonstrated antitumor activity (ORR, 30.4%; DOR, 40 weeks) and was well tolerated in patients with PD-L1 + advanced esophageal cancer (phase 1b KEYNOTE-028). The open-label, randomized, phase 3 KEYNOTE-181 trial (NCT02564263) will compare the tumor response to pembro relative to single-agent chemotherapy in previously treated advanced/metastatic (a/m) esophageal cancer. Methods: Key eligibility criteria include age $ 18 years, a/m adenocarcinoma or squamous cell carcinoma of the esophagus or Siewert type I adenocarcinoma of the esophagogastric junction (EGJ), measurable disease, documented PD during/after first-line therapy, ECOG PS 0-1, no active autoimmune disease or CNS metastases, and provision of a tissue sample for biomarker analysis. HER2 status must be determined for patients with EGJ adenocarcinoma; if HER2+, there must be documentation of PD on trastuzumab-containing therapy. Eligible patients are to be randomized (unblinded) 1:1 to IV pembro 200 mg Q3W or investigator’s choice of paclitaxel 80-100 mg/m2 on days 1, 8, and 15 Q4W, docetaxel 75 mg/m2 Q3W, or irinotecan 180 mg/m2 Q2W. Treatment is to continue until documented PD, intolerable toxicity, patient/ investigator decision, or (for pembro) until 35 cycles (~2 years). AEs are to be monitored throughout treatment and for 30 days thereafter (90 days for serious AEs). Tumor response will be assessed Q9W per RECIST v1.1 and adapted immune-related RECIST (irRECIST; central imaging vendor review). All patients will be followed Q9W for OS until death, withdrawal of consent, or end of study. Primary efficacy end points are PFS (per RECIST v1.1, blinded central imaging vendor review) and OS. Secondary end points include ORR (per RECIST v1.1, blinded central imaging vendor review). Approximately 600 patients will be enrolled. Clinical trial information: NCT02564263.

TPS4141

TPS4142

Poster Session (Board #128a), Sat, 8:00 AM-11:30 AM

Poster Session (Board #128b), Sat, 8:00 AM-11:30 AM

TAGS, a randomized, double-blind, phase 3 study evaluating TAS-102 plus best supportive care vs placebo plus best supportive care in patients with metastatic gastric cancer refractory to standard treatments. First Author: David H. Ilson, Memorial Sloan Kettering Cancer Center, New York, NY

Multicenter feasibility study of chemoradiation, trastuzumab and pertuzumab in resectable HER2+ esophageal carcinoma: The TRAP study. First Author: Stephanie van der Woude, Academisch Medisch Centrum, Amsterdam, Netherlands

Background: TAS-102 is comprised of an antineoplastic thymidine-based nucleoside analog, trifluridine, and a thymidine phosphorylase inhibitor, tipiracil, at a molar ratio of 1:0.5. The global phase 3 RECOURSE trial demonstrated a significant prolongation of overall survival (OS) and progression-free survival (PFS) with TAS-102 vs placebo (pbo) in patients (pts) with metastatic colorectal cancer who had been treated with $ 2 lines of prior standard chemotherapy. A phase 2 study conducted in Japan that evaluated efficacy and safety of TAS-102 in pts with metastatic gastric cancer who had failed 1 or 2 lines of standard chemotherapy demonstrated that TAS-102 was well tolerated, and TAS-102 treatment led to an OS of 8.7 months and disease control rate of 65.5%. The present study will investigate efficacy and safety of TAS-102 plus best supportive care (BSC) vs pbo plus BSC in pts with metastatic gastric cancer refractory to standard treatments. Methods: This phase 3, double-blind, placebo-controlled multinational study (NCT02500043) will randomize ~500 pts with metastatic gastric cancer who previously received $ 2 prior regimens for advanced disease. Eligible pts will be randomized 2:1 to TAS-102 plus BSC or pbo plus BSC, and stratified by region, ECOG performance status, and prior treatment with ramucirumab. TAS-102 or pbo 35 mg/m2/dose will be administered orally twice daily on days 1-5 and 8-12 of each 28-day cycle until discontinuation criteria are met. Primary endpoint is OS; secondary endpoints are PFS, quality of life (QoL), and safety. Computed tomography scans are performed every 8 weeks and QoL assessments every 4 weeks. One interim analysis for efficacy and futility is planned after approximately half of the total target events (384 deaths) are observed. If the primary endpoint is met and a favorable benefit/risk ratio is demonstrated for TAS-102, pts in the pbo arm who meet study eligibility criteria will be offered the option to cross over to open-label TAS-102. As of January 2016, no pts were randomized and recruitment is ongoing. Clinical trial information: NCT02500043.

Background: Despite multimodality treatment the outcome of esophageal adenocarcinoma (EAC) is poor. In EAC, human epidermal growth factor receptor 2 (HER2) is overexpressed in 19% to 43% of cases. Based on the positive results of the ToGA trial in metastatic esophagogastric cancer, HER2 targeting is currently investigated in patients treated with curative intent. Most patients, however, develop resistance to trastuzumab. Recent studies in breast cancer evaluating the effect of combined HER2/HER3 targeting with trastuzumab and pertuzumab showed improved efficacy. To explore if combined HER2 targeting added to multimodality treatment is feasible and has potential benefit in HER2 overexpressing resectable EAC, we are currently conducting the TRAP study. The primary objective is the feasibility of preoperative treatment with chemoradiation combined with pertuzumab and trastuzumab. Secondary objectives include: toxicity, surgery withdrawal rate, R0 resection rate, pathological reponse and pharmacokinetics. Methods: Patients with resectable EAC who are in a good clinical condition and considered operable are treated in one of the 9 participating centers with paclitaxel 50 mg/m2 and carboplatin AUC = 2 on days 1, 8, 15, 22 and 29. Trastuzumab is administered at a dose of 4 mg/kg on day 1, followed by 2 mg/ kg at week 2-6. From week 7 onwards trastuzumab is dosed at 6 mg/kg every 3 weeks. Pertuzumab is administered on day 1, 22, 43, 64, and 85 using a fixed dose of 840 mg. Radiotherapy is delivered to a total dose of 41.4 Gy, starting the first day of chemotherapy. Surgical resection follows in week 14. Treatment with trastuzumab and pertuzumab during chemoradiation is regarded feasible if $ 80% of patients complete treatment. Primary analysis will be performed in all patients who received at least one dose of pertuzumab and/or trastuzumab. Currently, 21 of planned 40 patients have been enrolled. The data safety monitoring board did not find any safety concerns after the first 10 operated patients. Clinical trial information: NCT02120911.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Noncolorectal) Cancer TPS4143

Poster Session (Board #129a), Sat, 8:00 AM-11:30 AM

A randomized phase II trial of systemic chemotherapy with or without trastuzumab followed by surgery in HER2 positive advanced gastric or esophagogastric junction adenocarcinoma with extensive lymph node metastasis: Japan Clinical Oncology Group study JCOG1301C (Trigger study). First Author: Masanori Tokunaga, Division of Gastric Surgery, Shizuoka Cancer Center, Nagaizumi, Japan

TPS4144

241s Poster Session (Board #129b), Sat, 8:00 AM-11:30 AM

The BRIGHTER trial: A phase III randomized double-blind study of BBI-608 + weekly paclitaxel versus placebo (PBO) + weekly paclitaxel in patients (pts) with pretreated advanced gastric and gastro-esophageal junction (GEJ) adenocarcinoma. First Author: Manish A. Shah, Weill Cornell Medicine, New York-Presbyterian Hospital, New York, NY

Background: The efficacy of trastuzumab for HER2-positive advanced gastric cancer was shown in ToGA trial, associated with a higher response rate. However, it is unclear if addition of trastuzumab to preoperative chemotherapy is effective for HER2-positive resectable gastric cancer. In Japan, preoperative chemotherapy with S-1 plus cisplatin (SP) is considered to be a promising treatment option for gastric cancer patients with extensive lymph node metastases because its efficacy was shown in single arm phase II trial (JCOG0405). Addition of trastuzumab to preoperative SP may enhance the anti-tumor activity and improve overall survival of patients with apparent para-aortic or regional lymph node metastasis. Methods: This study is designed as a randomized screening phase II study to compare preoperative SP plus trastuzumab and SP for HER2-positive gastric cancer with apparent para-aortic lymph nodes ( $ 1.0 cm in long axis) and/or regional lymph nodes ( $ 1.5 cm in short axis) metastasis (UMIN000016920). In both arms, an infusion of cisplatin (60 mg/m2/day, day 1) and oral S-1 (80 mg/m2/day, days 1–14) repeated every 3 weeks are given for 3 courses, and an additional infusion of trastuzumab (first course, 8 mg/kg; second course onward, 6 mg/ kg, day 1, every 3 weeks) are also given in SP plus trastuzumab arm. After R0 resection, adjuvant chemotherapy with S-1 is performed for 1 year after surgery. The planned sample size was 130 patients in total with a 1-sided alpha of 0.2, a power of 0.75, and a 10% increase in the 3-year OS (70% vs 80%), and the patients will be accrued from 58 Japanese institutions over 3 years. The primary endpoint is overall survival. Because Japanese medical insurance does not reimburse the perioperative use of trastuzumab, this study is conducted under Advanced Medical Care B, a governmental program to allow an off-label use. As of January 2016, no patients have been enrolled yet. Clinical trial information: UMIN000016920.

Background: BBI-608 is an orally-administered first-in-class cancer stemness inhibitor. By targeting Stat3, BBI-608 blocks cancer stem cell (CSC) self-renewal and survival through suppressing stemness pathways, including Stat3 and b-catenin as well as immune checkpoint gene expression. Potent anti-tumor and anti-metastatic activity was observed in preclinical models, with marked synergy between BBI-608 and paclitaxel. Moreover, cancer stemness genes, such as stat3 and b-catenin, two poor prognostic biomarkers in many cancer types, predict sensitivity to BBI-608. Encouraging anticancer activity in refractory gastric and GEJ adenocarcinoma was observed in a phase Ib (Stephenson et al, ASCO 2014 abstr) and a subsequent phase II study including 46 gastric or GEJ adenocarcinoma pts (Becerra et al, ASCO 2015 abstr). On the basis of these data, a phase III trial is being conducted in North America, South America, Europe, Australia, and Asia. Methods: This study (ClinicalTrials.gov NCT02178956) will assess the efficacy of BBI-608+paclitaxel vs PBO+paclitaxel in pts with pre-treated, advanced gastric and GEJ adenocarcinoma (target n = 680). Pts must have failed one prior line of therapy containing a fluoropyrimidine/platinum doublet for unresectable or metastatic disease. Pts are randomized in a 1:1 ratio to receive BBI608 480 mg or PBO twice daily continuously plus paclitaxel 80 mg/m2 IV, weekly, for 3 of every 4 weeks. Treatment will continue until disease progression, death, intolerable toxicity, or patient/ investigator decision to stop. Primary endpoint is overall survival (OS) in the general study population; secondary endpoints include progression free survival (PFS), OS and PFS in a predefined biomarker-positive subpopulation, objective response rate, disease control rate, and safety. In addition, blood, plasma, and archival tissue will be assessed for pharmacokinetic and biomarker analyses and quality of life will be measured. As of February 2016, 364 pts were randomized and recruitment is ongoing. Clinical trial information: NCT02178956.

TPS4145

TPS4146

Poster Session (Board #130a), Sat, 8:00 AM-11:30 AM

Poster Session (Board #130b), Sat, 8:00 AM-11:30 AM

A randomized, double-blind, placebo-controlled phase III study of ramucirumab versus placebo as second-line treatment in patients with hepatocellular carcinoma and elevated baseline alpha-fetoprotein following first-line sorafenib (REACH-2). First Author: Andrew X. Zhu, Massachusetts General Hospital Cancer Center, Boston, MA

PHOCUS: A phase 3 randomized, open-label study comparing the oncolytic immunotherapy Pexa-Vec followed by sorafenib (SOR) vs SOR in patients with advanced hepatocellular carcinoma (HCC) without prior systemic therapy. First Author: Ghassan K. Abou-Alfa, Memorial Sloan Kettering Cancer Center, New York, NY

Background: Ramucirumab (Ram) is a human IgG1 monoclonal antibody that inhibits ligand activation of the vascular endothelial growth factor receptor 2. The Phase III REACH study assessed Ram in the treatment of patients with advanced hepatocellular carcinoma (HCC) after prior sorafenib. Ram was well-tolerated. A significant improvement in overall survival (OS) in the overall population (N = 565) was not demonstrated (HR = 0.87; p = 0.14). However, a meaningful improvement in OS was observed in a prespecified subgroup of patients with baseline alpha-fetoprotein (AFP) $ 400 ng/mL (N = 250) (HR = 0.67, p = 0.006; median OS Ram 7.8 months vs. placebo 4.2 months). Methods: Designed to confirm the treatment effects of Ram in patients with elevated AFP, REACH-2 is a randomized, double-blind, placebo-controlled, global Phase III study of Ram and best supportive care (BSC) versus placebo and BSC in patients with HCC and elevated baseline AFP following prior therapy with sorafenib; Child-Pugh score , 7; Barcelona Clinic Liver Cancer Stage C or Stage B disease not amenable/refractory to locoregional therapy; AFP $ 400 ng/mL; and ECOG performance status 0‒1. Patients with a history of hepatic encephalopathy, clinically meaningful ascites, liver transplant, or hepatic locoregional therapy after sorafenib are not eligible. Eligible patients will be randomized 2:1 to receive Ram (8 mg/kg) or placebo as an intravenous infusion on day 1 of each 14-day cycle until radiographic or clinical disease progression or criterion for discontinuation are met. The primary objective is to assess the OS for patients treated with Ram versus placebo, and assumes an OS HR of 0.7 (85% power; 2-sided type I error 0.05). Target enrollment is 399 patients with a final analysis at 318 events (20% censoring). Secondary objectives include progression-free survival, objective response rate, safety, and patient-focused outcomes. Other objectives are assessments of biomarkers relevant to angiogenesis and HCC. Clinical trial information: NCT02435433.

Background: Pexa-Vec (pexastimogene devacirepvec; JX-594) is an oncolytic and immunotherapeutic vaccinia virus designed to selectively replicate in and destroy cancer cells. It causes direct oncolysis accompanied by tumor vascular disruption and anti-tumor immunity mediated by expression of the transgene GM-CSF. SOR, a multi-targeted tyrosine kinase inhibitor, is the standard of care for first-line systemic treatment of advanced HCC. Both preliminary preclinical and clinical data suggest complementary anti-tumor effects of a sequential combination of Pexa-Vec followed by SOR possibly by targeting the tumor vasculature via different mechanisms (Heo et al., Mol Ther 2011). A randomized phase II dose-finding study with 3 Pexa-Vec intratumoral (IT) liver injections in first line advanced HCC patients showed an acceptable safety profile and a significant increase in overall survival (OS) in the highest dose group (109pfu) (Heo et al., Nat Med 2013). Methods: This global, randomized, open-label, phase III study will compare the efficacy and tolerability of Pexa-Vec followed by SOR vs SOR in advanced HCC patients. Eligible patients are Child-Pugh A, ECOG PS 0-1, BCLC B/C, have at least 1 measurable and viable liver tumor (based on radiographic assessment) injectable under imaging guidance, have liver tumor mass , 50% of the total liver volume, no invasion of the inferior vena cava, and had no prior systemic therapy for HCC. 600 patients will be randomized 1:1 to either 3 Pexa-Vec IT injections administered every 2 weeks at a dose of 109pfu (D1, W2, W4) followed by 400 mg BID SOR starting at W6 or to SOR 400 mg BID (from D1). Patients are allocated to treatment arm using minimization by center, HCC etiology, presence of extra-hepatic disease, vascular invasion, PS 0-1, AFP levels. The primary endpoint is OS (1-sided stratified log-rank test; a = 0.025, 86% power, HR 0.83). Secondary endpoints include time to progression, progression-free survival, overall response rate and disease control rate (radiographic evaluation every 6 weeks). In addition, safety, biomarkers and quality of life will be evaluated. Clinical trial information: NCT02562755.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

242s TPS4147

Gastrointestinal (Noncolorectal) Cancer Poster Session (Board #131a), Sat, 8:00 AM-11:30 AM

A randomized, multicenter, phase 3 study of nivolumab vs sorafenib as firstline treatment in patients (pts) with advanced hepatocellular carcinoma (HCC): CheckMate-459. First Author: Bruno Sangro, Liver Unit, Clinica Universidad de Navarra and CIBERHD, Pamplona, Spain Background: HCC is the fifth most prevalent cancer globally and the secondleading cause of cancer-related deaths. The high mortality rate is typically due to the late stage of disease at diagnosis and a lack of effective treatment options. Sorafenib, the only systemic agent approved to treat advanced HCC, has shown only a modest survival benefit. More effective treatment options are needed. Nivolumab, a fully human IgG4 monoclonal antibody programmed death receptor-1 (PD-1) inhibitor, has demonstrated overall survival (OS) benefit in pts with metastatic melanoma, non–small-cell lung cancer, and advanced renal cell carcinoma. Preliminary data from the CheckMate-040 trial suggest that nivolumab has clinical activity and is tolerable in pts with HCC, including those with hepatitis B or hepatitis C virus (HCV) infection.1 CheckMate-459 is a phase 3, randomized, open-label study (NCT02576509) designed to compare the efficacy of nivolumab and sorafenib in pts with advanced HCC. Methods: Eligibility criteria include age $ 18 years, histologically confirmed advanced HCC, $ 1 measurable untreated lesion, ECOG performance status of 0 or 1, and no prior systemic therapy. Additional criteria include completion of locoregional therapy for HCC $ 4 weeks prior to baseline scan and Child-Pugh class A. Pts with fibrolamellar, sarcomatoid HCC; mixed cholangiocarcinoma and HCC; or prior liver transplant will be excluded. An estimated 726 pts will be randomized 1:1 to receive nivolumab or sorafenib until disease progression or unacceptable toxicity. Pts will be stratified by etiology, vascular invasion and/or extrahepatic spread, and geography. CheckMate-459 began in Nov 2015; estimated primary completion date is May 2017. Primary objectives are OS and time to progression. Secondary objectives include overall response rate, progression-free survival, and evaluation of the relationship between PD-L1 expression and efficacy. Exploratory patient-reported measures, including effects of treatment on health status and quality of life, will also be evaluated. Reference:1El-Khoueiry, et al. J Clin Oncol. 2015;33(suppl):LBA 101. Clinical trial information: NCT02576509.

TPS4149

Poster Session (Board #132a), Sat, 8:00 AM-11:30 AM

Randomized phase II study of cisplatin and etoposide versus temozolomide and capecitabine in patients (pts) with advanced G3 non-small cell gastroenteropancreatic neuroendocrine carcinomas (GEPNEC): A trial of the ECOG-ACRIN Cancer Research Group (EA2142). First Author: Jennifer Rachel Eads, University Hospitals Seidman Cancer Center, Case Comprehensive Cancer Center, Case Western Reserve University, Cleveland, OH Background: Poorly differentiated (G3) GEPNECs are rare tumors for which little prospective data are available. Historically these tumors have been treated akin to small cell lung cancer with platinum and etoposide based on histologic similarities. Emerging pathologic and clinical data suggest heterogeneity amongst G3 GEPNECs based on histology (small cell vs. nonsmall cell), Ki-67, response to therapy (platinum based vs. temozolomide based) and survival. Temozolomide based therapy appears the most promising alternative option. The proposed study will provide the first prospective data regarding the role of cisplatin and etoposide vs. temozolomide and capecitabine for G3 GEPNECs and their correlation with clinical parameters. Methods: This is a multi-center, randomized phase II trial. Pts with locally advanced/unresectable or metastatic G3 GEPNEC are randomized to receive capecitabine 750 mg/m2 PO every 12 hours days 1-14 and temozolomide 200 mg/m2 PO daily days 10-14 (Arm A) or cisplatin 25 mg/m2 IV daily days 1-3 and etoposide 100 mg/m2 IV daily days 1-3 (Arm B). Cycle length: 28 days (Arm A), 21 days (Arm B). Eligibility criteria include G3 non-small cell histology, Ki-67 proliferative index 20-100%, $ 10 mitotic figures per 10 high powered fields, measureable disease, no prior chemotherapy. Primary endpoint is progression free survival (PFS); secondary endpoints include response rate (RR), overall survival (OS) and toxicity. Tissue specimens will be collected for central pathology review with assessment of Ki-67 and its correlation with PFS, OS and response. CT, PET and octreoscan images will be banked. A total of 126 pts will be needed to accrue 120 eligible cases to detect an improvement in PFS from 6 months in the control arm (Arm B) to 10 months in the experimental arm (Arm A) with 90% power and a one-sided significance level of 0.10 using a log-rank test. Planned accrual is 4 pts per month over 30 months with an additional 12 months of follow-up. The trial was activated in November 2015. Clinical trial information: NCT02595424.

TPS4148

Poster Session (Board #131b), Sat, 8:00 AM-11:30 AM

REMINET: A European, multicentre, PHASE II/III randomized double-blind, placebo-controlled study evaluating lanreotide as maintenance therapy after first-line treatment in patients with non-resectable duodeno-pancreatic neuroendocrine tumours. First Author: Come Lepage, CHU Le Bocage HGE, INSERM U866, Dijon, France Background: Patients (pts) with metastatic or locally advanced, nonresectable, Well-Differentiated Duodeno-Pancreatic (WDDP) NETs are treated following European guidelines. Pts with more aggressive disease, i.e. progressive and/or symptomatic metastases, significant hepatic invasion ( . 30-50%), and/or bone metastases, combination chemotherapy is considered; otherwise biotherapy (everolimus or sunitinib) is an option. The recommendations are to stop chemotherapy once disease control is obtained. The concept of maintenance therapy is well established in several cancer types. In this study we will evaluate the role of a somatostatin analogue (LAN) as maintenance in patients with stable/responding disease after 1L therapy (which may have been interrupted for tolerability issues) until progression. Somatostatin analogues are well tolerated and have demonstrated antiproliferative effect making them ideal candidates for maintenance therapy. Methods: REMINET is an academic randomized, double-blind, placebo-controlled, phase II/III study. A total of 222 adults patients pts with a metastatic (synchronous or metachronous) or locally advanced, nonresectable, grade 1 or 2 WDDP NETs (WHO 2010 classification; Ki-67 # 20%) and documented stable disease or objective response after 1L therapy at least 4 weeks prior to randomization, will be enrolled and randomly assigned in a 1:1 ratio to receive 120 mg LAN or placebo, every 28 days, until disease progression or unacceptable toxicity. Stratification factors: Centre, Grade 1 vs. 2, 1L treatment (chemotherapy vs. biotherapy). The primary endpoint of the phase III part is Progression-Free Survival (PFS) assessed by the investigators (RECIST v1.1). The aim of the phase II part is to demonstrate a 6-months PFS . 45% in LAN arm. Main secondary endpoints are PFS according to central review, overall survival, safety and quality of life. Frozen blood samples (-80°C) will be BioBanked for ancillary studies. The study is currently open in Europe. Status: A total of 13 patients are already randomized. Clinical trial information: NCT02288377.

TPS4150

Poster Session (Board #132b), Sat, 8:00 AM-11:30 AM

A phase 2, open-label, multicenter study of durvalumab (MEDI4736) 6 tremelimumab in patients (pts) with metastatic pancreatic ductal adenocarcinoma (mPDAC): ALPS. First Author: Eileen Mary O’Reilly, Memorial Sloan Kettering Cancer Center, New York, NY Background: PDAC is a lethal malignancy; most patients present with metastatic disease and have a median survival of , 1 year with current standard cytotoxic therapies. Thus there is a major unmet need for new treatment approaches. Combined blockade of immune checkpoints is being explored across multiple malignancies. Dual blockade of programmed cell death-1 (PD-1) and cytotoxic T-lymphocyte-associated antigen-4 (CTLA-4) pathways is the most promising immuno-oncology combination approach to date. Durvalumab (D) is a selective, high affinity, engineered human IgG1 mAb that blocks programmed cell death ligand-1 (PD-L1) binding to PD-1 and CD80. Tremelimumab (T) is a selective human IgG2 mAb inhibitor of CTLA-4. In a Phase 1/2 study (NCT01693562), D monotherapy showed preliminary evidence of antitumour activity in several tumor types, including PDAC1. Encouraging activity and manageable tolerability of the combination of D + T were reported in a Phase 1b NSCLC study (NCT02000947)2. Here we describe ALPS (NCT02558894), a Phase 2 study of D 6 T in mPDAC. Methods: This is a Phase 2, open-label, multicenter, 2-part study to determine the efficacy and safety of D as monotherapy or in combination with T in pts with mPDAC whose disease has progressed on 5-fluorouracil- or gemcitabine-containing chemotherapy (CT). Eligible pts must not have had . 1 prior CT regimen or other systemic therapy for recurrent/mPDAC. Proximate archival or new tumor biopsy procurement is required for PD-L1 and biomarker analyses. In Part A (lead-in), pts will be randomized (1:1) to receive D 1500 mg i.v. q4w; or D 1500 mg i.v. q4w + T 75 mg i.v. q4w for 4 doses, followed by D 1500 mg i.v. q4w for a total of up to 12 months of treatment. The Part A primary endpoint is objective response rate using investigator assessments (RECIST v1.1). Secondary endpoints include DoR; disease control rate; PFS; OS; safety and tolerability; health-related QoL; PK; and immunogenicity. The design of Part B will be based on findings from Part A. Recruitment is ongoing in Part A. 1. Segal NH, et al. Ann Oncol 2014;25 (Suppl. 4):1058PD; 2. Antonia SJ, et al. J Clin Oncol 2015;33(Suppl. 15): 3014. Clinical trial information: NCT02558894.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Gastrointestinal (Noncolorectal) Cancer TPS4151

Poster Session (Board #133a), Sat, 8:00 AM-11:30 AM

TPS4152

243s Poster Session (Board #133b), Sat, 8:00 AM-11:30 AM

SWOG S1505: A randomized phase II study of perioperative mFOLFIRINOX vs. gemcitabine/nab-paclitaxel as therapy for resectable pancreatic adenocarcinoma. First Author: Davendra Sohal, Cleveland Clinic, Cleveland, OH

POLO: A randomized phase III trial of olaparib maintenance monotherapy in patients (pts) with metastatic pancreatic cancer (mPC) who have a germline BRCA1/2 mutation (gBRCAm). First Author: Talia Golan, The Oncology Institute Sheba Medical Center, Tel-Hashomer, Israel

Background: Clinical outcomes after curative therapy for resectable pancreatic ductal adenocarcinoma (PDA) remain suboptimal. Series show that 70-85% of patients die of systemic recurrence. Improved overall survival (OS) in the metastatic setting with the use of multi-agent chemotherapy regimens (FOLFIRINOX, gemcitabine/nab-paclitaxel) holds the promise of progress in the curative setting as well. However, aggressive systemic therapy is usually not feasible after major pancreatic surgery. Therefore, early control of systemic disease by increased preoperative chemotherapy delivery may improve outcomes. Furthermore, the perioperative platform facilitates early identification of patients with chemotherapy-resistant tumors and allows prospective biomarker studies, which can be used to test targeted therapies in the future. Methods: This is a randomized phase II study intended to choose the most promising perioperative regimen to test in a larger trial. Eligibility requirements include adult patients with an ECOG PS of 0 or 1, a confirmed histopathologic diagnosis of PDA, and resectable disease as confirmed by central radiology review: no involvement of the celiac, common hepatic, or superior mesenteric arteries (and, if present, variants); no involvement, or , 180° interface between tumor and vessel wall, of the portal or superior mesenteric veins; patent portal vein/splenic vein confluence; no metastases. Treatment includes 12 weeks [either 6 doses of mFOLFIRINOX (5-fluorouracil, irinotecan, oxaliplatin – without bolus 5-FU and leucovorin), or 9 doses of gemcitabine/nab-paclitaxel, on standard schedules] of preoperative chemotherapy, followed by surgical resection and 12 weeks of identical postoperative chemotherapy. Primary outcome is 2-year OS, using a “pick the winner” design with minimum two-year OS of 40% assuming a 58% alternative hypothesis, 88% power, and a 1-sided a of 0.05, providing 90% probability of selecting the better regimen with a total sample size of 112 patients. Correlative studies are planned. The study opened through the National Clinical Trials Network (ClinicalTrials.gov Registry Number: NCT02562716). Clinical trial information: NCT02562716.

Background: Defective double-strand DNA break repair caused by a gBRCAm is a risk factor for mPC. In the general population of mPC pts, the prevalence of a gBRCAm may be around 4.5%; but, in some populations of mPC pts, such as Ashkenazi Jews, the prevalence could be up to 15%. Monotherapy with the PARP inhibitor olaparib (Lynparza) has led to tumor responses in a Phase II trial that enrolled gBRCAm pts with a variety of solid tumor types, including in pts with mPC (Kaufman et al JCO 2014: Study 42; NCT01078662). This trial supported preclinical data showing that gBRCAmdefective tumors are intrinsically sensitive to PARP inhibitors, and led to a doubleblind, placebo-controlled Phase III trial (NCT02184195; POLO) of olaparib ‘switch maintenance’ monotherapy in pts with mPC and a gBRCAm who have not progressed on first-line platinum-based chemotherapy. Methods: Eligible pts with mPC must have documented disease control after completing $16 weeks of a first-line, platinum-based regimen, with a deleterious gBRCAm, which will be assayed and confirmed by Integrated BRACAnalysis (Myriad Genetic Laboratories) during the trial. Pts are randomized (3:2) to olaparib tablets (300 mg orally bd) or placebo. The primary endpoint is PFS, determined by blinded independent central review using RECIST 1.1. The primary PFS analysis will be performed after »87 PFS events (»60% maturity) using a log-rank test. Enrollment began in Q4 2014. As of 16 January 2016, 635 pts have been screened. Of 590 pts for whom BRCA1/2 mutation testing results are currently available, 46 pts (7.8%) with a gBRCAm have been identified: 10 pts were known to have a gBRCAm at screening, while 36/580 (6.2%) had a newly identified gBRCAm. Outside the USA or Israel (where screened populations may be enriched for Ashkenazi Jews), a gBRCAm was newly identified in 23/568 pts (4%). The target number for randomization is »145 pts across »90 centers worldwide. Clinical trial information: NCT02184195. Country

Screened (n)

Spain France Korea USA Israel UK Italy Germany Australia Belgium Canada Netherlands Total

95 85 81 75 73 61 61 44 29 7 6 4 621*

*The breakdown by country is based on an initial 621 pts screened.

TPS4153

Poster Session (Board #134a), Sat, 8:00 AM-11:30 AM

Randomized phase 2 study of the safety, efficacy, and immune response of GVAX pancreas (with cyclophosphamide) and CRS-207 with or without nivolumab in patients with previously treated metastatic pancreatic adenocarcinoma (STELLAR). First Author: Dung T. Le, Johns Hopkins University, Baltimore, MD Background: A heterologous prime-boost vaccination strategy using GVAX pancreas and CRS-207 is showing promise in patients with pancreatic adenocarcinoma (PDA) (Le, JCO 2015). Furthermore, blockade of the immune checkpoint programmed death-1 (PD-1) is active in some cancers. Combinatorial strategies aimed at priming tumor antigen-specific T cells while simultaneously blocking negative checkpoints may be necessary to improve outcomes in PDA. GVAX is composed of allogeneic pancreatic cancer cells modified to express GM-CSF and induces a broad response against multiple tumor antigens. GVAX is given with low-dose cyclophosphamide (CY) to inhibit regulatory T cells. CRS-207 is live-attenuated Listeria monocytogenes engineered to express the tumor-associated antigen mesothelin. CRS-207 boosts responses against mesothelin and is unique in its capacity to stimulate both innate and adaptive immunity by activating T cells and NK cells. Nivolumab is an antibody against PD-1. Methods: This is a phase 2 study comparing CY/GVAX and CRS-207 with or without nivolumab in subjects with PDA who failed only one chemotherapy regimen for metastatic disease. Subjects are randomized in a 1:1 ratio to receive either 2 doses of CY/nivolumab/GVAX and 4 doses of nivolumab/CRS207 (Arm A) or 2 doses of CY/GVAX and 4 doses of CRS-207 (Arm B). The primary objective is to compare OS between Arms A and B. Secondary/ exploratory objectives include: assessment of safety and clinical responses (tumor assessments and CA19-9 levels) and correlation of Lm- and mesothelin-specific T cell and other immunological responses with OS, progression-free survival and best overall response. Clinical trial information: NCT02243371.

TPS4154

Poster Session (Board #134b), Sat, 8:00 AM-11:30 AM

GLOBAL BALLAD: An International Rare Cancers Initiative trial to evaluate the potential benefit of adjuvant chemotherapy for small bowel adenocarcinoma (IRCI 002). First Author: T.R. Jeffry Evans, Beatson West of Scotland Cancer Centre and University of Glasgow, Glasgow, United Kingdom Background: Small bowel adenocarcinoma (SBA) is a rare cancer (, 5% of gastrointestinal tumours) with poor prognosis at all stages. There are no RCTs of adjuvant treatment following definitive surgery in stage I–III SBA, but the proven benefits of fluoropyrimidine-based adjuvant therapy (FP), with/ without oxaliplatin (Ox), in colorectal cancer suggest this may be worthwhile. GLOBAL BALLAD aims to answer 2 questions in resected stage I–III SBA. (A)Where the clinician is uncertain, what is the value of adjuvant postoperative chemotherapy over observation? (B)Where the clinician is convinced of the value of adjuvant treatment, what is the value of adding Ox to FP chemotherapy? Methods: This is an open-label, randomised, controlled multi-centre trial with disease free survival as the primary endpoint. A standard frequentist phase III trial design results in an infeasible sample size. A hybrid approach has thus been adopted with a standard randomised phase II screening design augmented, using Bayesian techniques, by subjective clinician estimates of treatment effect based on external evidence. The study is designed to have 80–90% power to detect HR = 0.775 for each question at the 20% 1-sided significance level. If results are statistically significant, the study data will be combined with clinician estimates of treatment benefit to give an overall estimate. The numbers required are 580–880 patients. For question A, patients are randomised 1:1 between FP+/-Ox and observation. For question B, randomisation is 1:1 FP v FP +Ox. To ease recruitment choice of FP is allowed. Patients allocated to chemotherapy in A may also be randomised to have Ox or not, boosting recruitment to randomisation B and trial efficiency. All patients will have pretreatment tumour samples and venous blood collected for genomic studies for exploratory prognostic and predictive markers. All patients complete a questionnaire for risk factors for SBA. The study is currently open in UK and France with funding from Cancer Research UK (grant CRUK/12/041) and Programme Hospitalier De Recherche Clinique respectively. It is under internal review in Japan. EudraCT No: 2013-003047-29 Clinical trial information: 15070952.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

244s LBA4500

Genitourinary (Nonprostate) Cancer Oral Abstract Session, Sun, 8:00 AM-11:00 AM

Atezolizumab (atezo) as first-line (1L) therapy in cisplatin-ineligible locally advanced/metastatic urothelial carcinoma (mUC): Primary analysis of IMvigor210 cohort 1. First Author: Arjun Vasant Balar, Perlmutter Cancer Center, NYU Langone Medical Center, New York, NY

The full, final text of this abstract will be available at abstracts.asco.org at 7:30 AM (EDT) on Sunday, June 5, 2016, and in the Annual Meeting Proceedings online supplement to the June 20, 2016, issue of the Journal of Clinical Oncology. Onsite at the Meeting, this abstract will be printed in the Sunday edition of ASCO Daily News.

4501

Background: Minimal antitumor activity of existing therapies and the observation of immune dysfunction in bladder cancer have prompted evaluation of immunotherapy in this malignancy. Nivolumab (fully human IgG4 programmed death-1 immune checkpoint inhibitor antibody) monotherapy has shown survival benefit in patients (pts) with melanoma, lung cancer, and renal cell carcinoma. Here, we report efficacy and safety of nivolumab monotherapy in pts with mUC after $1 prior line of platinum-based therapy in an open-label, multicenter phase I/II study (NCT01928394). Methods: Pts (unselected by PD-L1 expression status) with mUC received nivolumab 3 mg/kg intravenously every 2 weeks until progression or discontinuation. Primary endpoint was objective response rate (ORR; RECIST 1.1). Other endpoints included safety, duration of response (DOR), progression-free survival (PFS), and overall survival (OS). Results: Of 78 treated pts (median age 65.5 years; range, 31–85), 65.4% had received $2 prior therapies. At a median follow-up of 213 days (range, 22–499), 33.3% of pts remain on therapy; primary reason for treatment discontinuation was disease progression. Median number of doses was 8.5 (range, 1–34); 70.5% received .4 doses. Efficacy findings are shown in the table. Outcomes by PD-L1 expression will be included in the presentation. Grade 3 or 4 treatment-related adverse events (TRAEs) occurred in 20.5% of pts; most frequent were increased lipase and increased amylase (3.8% each) and fatigue, decreased neutrophils, and dyspnea (2.6% each). Grade 5 TRAEs occurred in 2.6% of pts (pneumonitis [n=1] and thrombocytopenia [TTP; n=1]). No grade 3 or 4 pneumonitis or TTP was reported. Conclusions: Nivolumab monotherapy demonstrated promising efficacy and acceptable safety in previously treated, unselected pts with mUC. OS data are encouraging. Clinical trial information: NCT01928394. Parameter ORR (confirmed), % (95% CI) Median PFS, months (95% CI) Median OS, months (95% CI) 12-month OS rate, % (95% CI) Median time to response, months (SD) Median DOR, months (95% CI)

4502

Oral Abstract Session, Sun, 8:00 AM-11:00 AM

Safety and efficacy of durvalumab (MEDI4736), a PD-L1 antibody, in urothelial bladder cancer. First Author: Christophe Massard, Drug Development Unit, Institut Gustave Roussy Cancer Centre, Villejuif, France Background: A Phase 1/2 dose escalation and dose expansion study is evaluating the safety and efficacy of durvalumab, a modified human IgG1 mAb that blocks PD-L1 binding to PD-1 and CD80, in solid tumors including urothelial bladder cancer (UBC). Methods: Eligible patients (pts) have inoperable or metastatic UBC, ECOG PS 0-1, with no prior anti-PD-1/PD-L1 exposure. Durvalumab 10 mg/kg is administered IV q2w for up to 12 mos. PD-L1 expression on tumor cells (TC) and tumor-associated immune cells (IC) is assessed with the Ventana SP263 assay (PD-L1+ defined as $ 25% of TC or IC staining for PD-L1). Response is investigator assessed per RECIST v1.1. Results: As of 20 Nov 2015, 61 pts (40 PD-L1+, 21 PD-L1‒; mean age 65 yrs [range 34-81]; 69% male; ECOG 0/1: 28%/72%; 51% with $ 2 prior systemic treatments) received a median of 4 doses (range 1-26). Drug-related AEs occurred in 64%; most frequent were fatigue (13%) and diarrhea (10%). Three pts had drug-related Grade $ 3 AEs: infusion-related reaction, tumor flare, and acute kidney injury (biopsy-proven nephritis). No drug-related AEs led to death. Forty-two pts were response evaluable with treatment initiated $ 12 wks before data cutoff (median follow-up 6.5 mos [range 0.8-14.8]; Table). PD-L1 expression in TC or IC both appeared to independently enrich for treatment response; 15/16 responders were PD-L1+ in TC or IC. ORR was 25% (3/12) in pts with liver metastases and 50% (4/8) in pts with lymph node‒only disease. Responses are ongoing in 15/16 pts; longest duration of response is ongoing at 11.3+ mos (median not reached). Conclusions: Durvalumab has an acceptable safety profile with evidence of clinical activity in heavily pretreated UBC pts. Defining PD-L1 status based on TC and IC expression effectively separates the pt population into 2 subgroups with distinct response rates. Nearly all responses occurred in the PD-L1+group. Studies of durvalumab in UBC are ongoing. Clinical trial information: NCT01693562.

ORR (confirmed/unconfirmed) by PD-L1 expression and localization. ORR PD-L1 Expression by Location All evaluable TC IC TC or IC

PD-L1 Status

n/N (%)

95% CI

Any + ‒ + ‒ + ‒

16/42 (38.1) 8/15 (53.3) 8/27 (29.6) 11/18 (61.1) 5/24 (20.8) 15/28 (53.6) 1/14 (7.1)

23.6‒54.4 26.6‒78.7 13.8‒50.2 35.7‒82.7 7.1‒42.2 33.9‒72.5 0.2‒33.9

Oral Abstract Session, Sun, 8:00 AM-11:00 AM

Efficacy and safety of nivolumab monotherapy in metastatic urothelial cancer (mUC): Results from the phase I/II CheckMate 032 study. First Author: Padmanee Sharma, The University of Texas MD Anderson Cancer Center, Houston, TX

4503

Nivolumab All treated pts (N=78) 24.4 (15.3235.4) 2.8 (1.525.5) Not estimable (NE) (7.02NE) 51.6 (37.0264.5) 1.5 (2.1) NE (5.52NE)

Oral Abstract Session, Sun, 8:00 AM-11:00 AM

Limited versus extended pelvic lymphadenectomy in patients with bladder cancer undergoing radical cystectomy: Survival results from a prospective, randomized trial (LEA AUO AB 25/02). First Author: Juergen E. Gschwend, Department of Urology, Klinikum rechts der Isar, Technical University of Munich, Munich, Germany Background: The extent of pelvic lymph node dissection (PLND) in bladder cancer patients (pts.) undergoing radical cystectomy may affect survival according to retrospective studies. The German Urologic Oncology Group (AUO) reports mature data of the first prospective, randomized clinical trial to evaluate the impact of a limited versus an extended PLND. Methods: Pts. with high-grade T1 or invasive urothelial bladder cancer (cT2-T4a) from 16 German centers were randomized 1:1 to receive a limited versus an extended PLND at the time of radical cystectomy. Limited PLND included 6 fields (bilateral obturator, internal and external iliac nodes) and extended PLND defined 14 fields (in addition bilateral deep obturator fossa, presacral, paracaval, interaortocaval and paraaortal nodes up to the inferior mesenteric artery). Pts. with neoadjuvant chemo- or radiotherapy were excluded, adjuvant chemotherapy was allowed. The primary endpoint was recurrence-free survival (RFS). Cancer-specific survival (CSS) was a secondary endpoint. The planned total sample size was 400 to detect an improvement in 5-year RFS from 50% to 65% (90% power, 2-sided test, p , 0.05) in the extended arm. Results: In total, 375 of 437 pts. were randomized from 02/2006 to 08/ 2010 and eligible for intention-to-treat analysis (191 limited and 184 extended PLND). The tumor was locally confined (# pT2 pN0) in 49.6% of patients and 24.0% were node positive (pN+). The median number of dissected nodes was 19 in the limited and 32 in the extended arm. The 5year RFS rate was 62.0% in the limited compared to 69.3% in the extended arm which was statistically not significant (Hazard ratio (HR) = 0.80, 95% Confidence Interval (CI) (0.54-1.19); log-rank p = 0.28). The 5-year CSS rate was improved from 66.2% in the limited to 77.5% in the extended arm which was statistically not significant (HR = 0.70, 95%CI 0.45-1.10; logrank p = 0.13). Conclusions: We observed a trend but no significant difference toward improved RFS and CSS with an extended PLND. The rate of recurrence was lower than expected in the limited arm, which might be due to the high number of resected nodes in this arm. Clinical trial information: NCT01215071.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Genitourinary (Nonprostate) Cancer 4504

Oral Abstract Session, Sun, 8:00 AM-11:00 AM

Mature results of the GETUG 13 phase III trial in poor-prognosis germ-cell tumors (GCT). First Author: Karim Fizazi, Institut Gustave Roussy, University of Paris Sud, Villejuif, France Background: Until 2014, standard treatment for poor-prognosis GCT was 4 BEP plus surgery and cure was achieved in only 50% of patients (pts) (IGCCCG 1997). The tumor marker decline rate identified pts with a better outcome (J Clin Oncol 2004, 22: 3868-76). The GETUG 13 phase III trial established that switching pts with an unfavorable decline to intensified chemotherapy results in improved progression-free survival (PFS) (Lancet Oncol 2014; 15: 1442-50). We assessed the long-term efficacy and toxicity in pts treated in GETUG 13. Methods: 263 pts with IGCCCG poor-prognosis GCT were treated with 1 BEP. hCG and AFP were assessed at day 21: 1) 51pts with a favorable decline continued BEP (Fav-BEP); 2) 203 pts with an unfavorable decline were randomized to receive either BEP (Unfav-BEP) or a dose-dense regimen (Unfav-dose-dense), consisting of paclitaxel-BEP plus oxaliplatin x 2 cycles, followed by 2 cycles of cisplatin, ifosfamide, and bleomycin + G-CSF. PFS and overall survival (OS) (logrank) and long-term toxicity (NCI-CTC criteria) were assessed. Results: The median follow-up is 5.6 years (range 0.3; 11.9). The 5-year PFS rate is 60% in the Unfav-dosedense arm vs 47% in the Unfav-BEP arm (HR: 0.65 [0.43-0.97]; p=0.037). The 5-year OS rate is 70.4% and 60.8%, respectively (HR: 0.69 [0.431.11]; p=0.12). Side effects evolved favorably, with 3 pts in the Unfav-dosedense arm reporting grade 3 motor neurotoxicity at 1 year but no reported toxicity $ grade 2 after year 2. The prognostic value of the tumor marker decline was confirmed: 70% vs 47% for 5-year PFS (p=0.006), and 78% vs 61% for OS (p=0.02). Salvage high-dose chemotherapy plus a stem cell transplant was implemented in 8% in the Unfav-dose-dense arm and 17% in the Unfav-BEP arm (p=0.035). Conclusions: With a mature follow-up, GETUG 13 shows that pts with poor-prognosis GCT and an unfavorable tumor marker decline after 1 BEP who are treated with intensified chemotherapy achieve significantly improved PFS, numerically better OS, minimal long-term toxicity, and a reduced need for high-dose salvage chemotherapy plus a stem cell transplant. These data support integrating this strategy as a standard of care for these rare pts. Clinical trial information: NCT00104676.

4506

Oral Abstract Session, Sun, 8:00 AM-11:00 AM

4505

245s Oral Abstract Session, Sun, 8:00 AM-11:00 AM

High-dose chemotherapy (HDCT) and autologous peripheral-blood stem cell transplant (PBSCT) for relapsed metastatic germ-cell tumors (mGCT): The Indiana University (IU) experience. First Author: Nabil Adra, Indiana University Melvin and Bren Simon Cancer Center, Indianapolis, IN Background: Patients (pts) with mGCT who relapse after initial chemotherapy (CT) can still be cured with second-line and even third-line regimens. We previously reported series of 184 pts treated with HDCT between 1996-2004 (N Engl J Med 2007;357:340-8). We now report survival outcomes of 364 pts with relapsed mGCT treated with HDCT and PBSCT at IU between 2004-2014. Methods: 364 consecutive pts with mGCT who progressed after cisplatin-based combination CT. 341 pts received 2 consecutive courses of HDCT consisting of 700mg/m2 carboplatin and 750mg/m2etoposide, each for 3 consecutive days, and each followed by PBSCT. 23 pts received only a single course of HDCT due to progressive disease or toxicity. Kaplan-Meier methods were used for progression free survival (PFS) analysis. Results: Median age was 32. At time of starting HDCT, median AFP was 7.5 (range, 1-21,347) and median hCG was 37.1 (range, 0.5-178,140). Of the 364 pts, 230 had complete remission (CR) without relapse during a median follow-up of 29 months. Of the 302 pts who received treatment as second-line therapy, 195 pts were disease-free. 35 of 62 pts who received treatment as third-line or later therapy were disease-free. There were 6 treatment-related deaths. Secondary leukemia developed in 5 pts. Conclusions: This large single-institution study further demonstrates that relapsed mGCT are curable by HDCT plus PBSCT even when used in third-line or later therapy. Variable (N)

2-year PFS

All pts (364)

57%

HDCT

-

-Second-line (302)

59%

-Third-line or subsequent (62)

43%

Platinum Sensitivity -Sensitive (242) -Refractory (122) Salvage AFP

30% 58%

-.1000 ng/mL (28)

36% -

-#1000 mIu/mL (274)

65%

-.1000 mIu/mL (90)

34%

Histology

-

-Seminoma (79)

90%

-Nonseminoma (285)

48%

Response to initial CT -CR or partial remission (PR) with normal serum markers (289) -Less than CR or PR (75) IGCCCG risk

-

-

-Intermediate (38)

55% 59%

-Mediastinal nonseminoma (20)

20% -

-yes (64)

35%

-no (300)

61%

4507

,0.001

,0.001

,0.001

38%

-Testicular/retroperitoneal (344) Brain metastasis

,0.001

29% 80%

-Poor (175)

0.04

64%

-Good (151)

Primary site

0.02

,0.001

71%

-#1000 ng/mL (336) Salvage hCG

p value

,0.001

,0.001

Oral Abstract Session, Sun, 8:00 AM-11:00 AM

Overall survival (OS) in METEOR, a randomized phase 3 trial of cabozantinib (Cabo) versus everolimus (Eve) in patients (pts) with advanced renal cell carcinoma (RCC). First Author: Toni K. Choueiri, Dana-Farber/Brigham and Women’s Cancer Center, Boston, MA

Long-term overall survival (OS) with nivolumab in previously treated patients with advanced renal cell carcinoma (aRCC) from phase I and II studies. First Author: David F. McDermott, Beth Israel Deaconess Medical Center, Boston, MA

Background: Cabo is an inhibitor of tyrosine kinases including MET, VEGF receptors, and AXL, which are oncogenic drivers in RCC. In the METEOR trial (NCT01865747), Cabo showed a statistically significant improvement in progression-free survival (PFS, the primary endpoint) compared with Eve in pts with previously treated RCC (median of 7.4 vs 3.8 mo; HR=0.58, 95% CI 0.45–0.75; P ,0.001) and improved the objective response rate (Choueiri NEJM 2015). The safety profile of Cabo was acceptable and similar to other VEGFR TKIs in this population. Interim analysis of the secondary endpoint of OS with a minimum follow-up of 6 mo revealed a favorable trend for Cabo vs Eve (HR=0.67, 95% CI 0.51–0.89; P=0.005). Here we present the final OS results, arising from a second interim analysis. Methods: Pts with measurable clear cell RCC, KPS $70, and $1 prior VEGFR TKI were randomized 1:1 to Cabo (60 mg qd) or Eve (10 mg qd) stratified by MSKCC risk group and number of prior VEGFR TKIs (1 or $2). The study was designed to detect a HR for OS of 0.75 (80% power, 2-sided a=0.04). Results: From Aug 2013–Nov 2014, 658 pts were randomized. As of 31 Dec 2015, with a minimum follow-up of 13 mo, 320 deaths were recorded (140 for Cabo and 180 for Eve). 74 (22%) pts remained on therapy in the Cabo arm vs 25 (8%) pts in the Eve arm. The secondary endpoint of improved OS for Cabo-treated pts was met. The median OS was 21.4 mo for Cabo vs 16.5 mo for Eve, with a 33% reduction in the rate of death (HR 0.67, 95% CI 0.53 to 0.83, P=0.0003). Landmark estimates of survival at 18 mo were 58% in the Cabo arm vs 47% of the Eve arm. OS benefit with Cabo was consistently observed across all prespecified subgroups including MSKCC risk group, number and type of prior VEGFR TKIs, prior anti-PD-1/PD-L1 treatment, location and extent of tumor metastases, and tumor MET expression level. SAEs were consistent with the safety profile previously reported. Conclusions: Cabo is the only agent to demonstrate a significant benefit in OS, PFS, and ORR in a Phase 3 trial in previously treated pts with advanced RCC. Cabo is an important new treatment option for these patients. Clinical trial information: NCT01865747.

Background: Nivolumab, a programmed death-1 (PD-1) immune checkpoint inhibitor, was recently approved by the FDA for patients with clear-cell aRCC previously treated with anti-angiogenic therapy based on results from the phase III CheckMate 025 study, which showed significantly longer median OS for nivolumab vs everolimus (25.0 vs 19.6 months; P = 0.002) with a minimum follow-up of 14 months (Motzer et al, N Engl J Med 2015;373: 1803–13). Because of the early survival benefit seen in the phase III study, there is increased interest in the potential for nivolumab to provide long-term OS. Here, we report long-term OS results from phase I and II nivolumab studies. Methods: Patients with Eastern Cooperative Oncology Group performance status #2 and prior systemic treatment with 1–5 regimens received nivolumab (1 or 10 mg/kg) every 2 weeks in a phase I open-label study (NCT00730639; McDermott et al, J Clin Oncol 2015;33:2013–20). Patients with Karnofsky performance status $70% and prior treatment with 1–3 regimens in the metastatic setting received nivolumab (0.3, 2, or 10 mg/ kg) every 3 weeks in a phase II randomized study (NCT01354431; Motzer et al, J Clin Oncol 2015;33:1430–37). In both studies, OS was estimated by the Kaplan–Meier method. Results: 34 patients with aRCC were treated in the phase I study. At a minimum follow-up of 50.5 months, objective response rate (ORR) was 29% and the median duration of response was 12.9 months. The 3- and 5-year OS rates were 41% and 34%. 167 patients with aRCC were treated in the phase II study. At a minimum follow-up of 38 months, ORR was 21% and the median duration of response was 22 months. The 3-year OS rate was 35%. Additional data will be presented including 4-year OS rate from the phase II study and long-term OS by subgroups. Conclusions: With about one-third of patients treated with nivolumab alive at 5 years in the phase I study and 3 years in the phase II study, this is the longest follow-up reported to date with any anti-PD-1/PD-L1 agent in aRCC. Potential predictors of long-term survival with nivolumab in this previously treated population are being explored. Clinical trial information: NCT00730639 and NCT01354431.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

246s 4508

Genitourinary (Nonprostate) Cancer Poster Discussion Session; Displayed in Poster Session (Board #131), Mon, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Mon, 4:45 PM-6:00 PM

4509

Poster Discussion Session; Displayed in Poster Session (Board #132), Mon, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Mon, 4:45 PM-6:00 PM

FDA analysis of treatment beyond disease progression disease (PD) in patients with metastatic renal cell carcinoma (mRCC) treated with nivolumab vs. everolimus. First Author: Chana Weinstock, Food and Drug Administration, Silver Spring, MD

Treatment beyond progression with nivolumab (nivo) in patients (pts) with advanced renal cell carcinoma (aRCC) in the phase III CheckMate 025 study. First Author: Bernard J. Escudier, Department of Cancer Medicine, Institut Gustave Roussy, Villejuif, France

Background: Traditional measurement of disease progression by RECIST v1.1 criteria may not fully capture patient benefit from PD-1 and PD-L1 inhibitors. Contemporary immunotherapy protocols generally allow patients to continue treatment beyond investigator-assessed radiographic PD as long as there is ongoing clinical benefit. However, the frequency of responses beyond PD is unknown, and whether this strategy is beneficial to the patient remains unclear. We present a review of the data from patients with mRCC treated beyond PD with nivolumab on CheckMate 025 (a randomized, phase 3 study of nivolumab vs. everolimus in locally advanced or mRCC). Methods: Subjects on CheckMate 025 who received any treatment beyond radiographic PD were identified. Dates of progression, treatment discontinuation, and corresponding tumor measurements were analyzed. Results: Treatment beyond radiographic PD occurred in 171 of patients treated with nivolumab. Median treatment duration beyond radiographic progression was 3.2 months (range 0.1-23.4). There were 5 patients (2.9% of those treated beyond progression) with a partial response after initially meeting criteria for RECIST 1.1-defined PD. In these 5 patients, small changes in tumor burden led to assessments of progression and response. Median survival in the 171 patients on the nivolumab arm treated beyond PD was 28.1 months. Median survival was 25.0 months among the 410 patients randomized to nivolumab and 16.6 months among the 134 patients who did not receive nivolumab after PD. Conclusions: Responses beyond investigator-assessed PD in mRCC patients treated with nivolumab appear to be rare, similar to data from treatment beyond PD with nivolumab in melanoma (FDA review). However, it is possible that some patients may derive clinical benefit that is not captured by radiographic assessments, and further investigation is needed. This data may help inform future treatment decisions and/or trial design in patients treated with immune checkpoint inhibitors. Clinical trial information: NCT01668784.

Background: Immunotherapy response patterns differ from traditional therapies, and pts may benefit from treatment after initial RECIST progression (CCR 2009;15:7412–20). We investigated pts treated beyond progression (TBP) with nivo in study CheckMate 025—nivo vs everolimus (eve) in previously treated pts with aRCC (NCT01668784). Methods: Treatment beyond progression was allowed in pts who had investigatorassessed clinical benefit and tolerated study drug. Pts TBP continued to receive nivo $4 wk after first RECIST version 1.1–defined progression to account for any delayed scan results; pts not TBP (NTBP) discontinued treatment after first progression. Pts without progression were excluded from this analysis. Results: Of 406 nivo pts treated, 38% were TBP; 36% were NTBP (of 397 eve pts treated, 17% were TBP; 36% were NTBP—current analysis for nivo only). Baseline characteristics were generally similar except for higher Karnofsky performance status (KPS) $90 with TBP vs NTBP (72% vs 62%) and less bulky tumor burden (18% vs 26%). Median overall duration of treatment (DOT) was 8.8 (TBP) and 2.3 mo (NTBP). From randomization to progression, objective response rate was 20% and 14%; median time to response was 1.9 and 3.7 mo; duration of response was 5.6 and 7.0 mo for TBP and NTBP pts, respectively. Treatment-related adverse events occurred in 71% of pts TBP and 70% of pts NTBP before first progression. Characteristics at first progression are shown (Table). Median DOT after first progression was 3.4 mo. Of 140 pts TBP with tumor measurements before and after progression, 14% had $30% tumor burden reduction from first progression. Median overall survival was 28.1 (TBP) vs 15.0 mo (NTBP); P,0.001.Conclusions: Treatment beyond progression with nivo can be associated with tumor shrinkage after progression. Evaluating disease characteristics at first progression may facilitate decision making to continue nivo treatment beyond progression. Clinical trial information: NCT01668784.

4510

Poster Discussion Session; Displayed in Poster Session (Board #133), Mon, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Mon, 4:45 PM-6:00 PM

Characteristics at first progression, %

TBP

NTBP

New lesions Increase in target lesions Site of new lesions Bone Liver KPS $90 Small (,13 cm) to bulky ($13 cm) tumor burden change Quality of life FKSI-DRS, median

41 55 5 5

44 43 14 8

73 7 31

48 13 27

4511

Poster Discussion Session; Displayed in Poster Session (Board #134), Mon, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Mon, 4:45 PM-6:00 PM

First-line sunitinib versus pazopanib in metastatic renal cell carcinoma (mRCC): Results from the international metastatic renal cell carcinoma database consortium (IMDC). First Author: Jose Manuel Ruiz Morales, Tom Baker Cancer Centre - University of Calgary, Calgary, AB, Canada

ECOG 1808: Randomized phase II trial of sunitinib with or without gemcitabine in advanced kidney cancer with sarcomatoid features. First Author: Naomi B. Haas, Abramson Cancer Center of the University of Pennsylvania, Philadelphia, PA

Background: Sunitinib (SU) and pazopanib (PZ) are standards of care for first-line treatment of metastatic renal cell carcinoma (mRCC). However, how the efficacy of these drugs translates into effectiveness on a population-based level is unknown. Methods: We used the IMDC to assess overall survival (OS), progression-free survival (PFS), response rate (RR) and performed proportional hazard regression adjusting for IMDC prognostic groups. Second-line OS2 and PFS2 were also evaluated. Results: We obtained data from 7438 patients with mRCC treated with either first line SU (n=6519) or PZ (n=919) with an overall median follow-up of 40.4 months (m) (CI95% 39.2 – 42.1). Baseline characteristics are in Table 1. There were no significant differences in IMDC prognostic groups with favorable, intermediate and poor risk patient distributions being 23%, 57%, 20% respectively for sunitinib and 24%, 58%, 18% respectively for pazopanib (p=0.36). There was no OS difference between sunitinib and pazopanib (22.3m vs 22.6m respectively, p=0.65). When adjusted for IMDC criteria, the HR of death for PZ vs SU is 1.03 (95%CI 0.92-1.17, p=0.58). There was no PFS difference between sunitinib and pazopanib (8.4m vs 8.3m respectively, p=0.17). When adjusted for IMDC criteria, the HR for PFS for PZ vs SU 1.08 (95%CI 0.981-1.19, p=0.12). There was no difference in RR between sunitinib and pazopanib (30% vs 28% respectively, p=0.15). We also found no difference in any second-line treatment between either postSU vs. post-PZ groups for OS2 (12.8m vs. 12.7 m, p=0.91) and PFS2 (3.7 vs. 4.5 m, p=0.50). Conclusions: We confirmed in real-world practice, that SU and PZ have similar efficacy in the first-line setting for mRCC and do not affect outcomes with subsequent second-line treatment.

Background: Patients (pts) with RCC with sarcomatoid (sarc) features have poorer outcomes than those with non-sarc RCC. Sarc RCC demonstrated a 15.4% response rate (RR) to doxorubicin + gemcitabine (G) in E8802. A phase I trial of combined sunitinib/ gemcitabine (SG) suggests activity in RCC with sarc. We sought to evaluate the effectiveness of SG and sunitinib alone (S) in sarc RCC. Methods: Pts were randomized 1:1 to each arm using a two-stage design. RCC pts with any sarc change and # 1 prior nonvascular endothelial growth factor tyrosine kinase inhibitor were stratified into prognostic groups: good (clear cell, , 20% sarc, PS 0), intermediate (20-50% sarc, PS 0), and poor (non-clear cell or . 50% sarc or PS 1). The primary endpoint was RR. Secondary endpoints were progression-free survival (PFS), overall survival (OS) and safety. For SG, the null RR was 15% and a 30% RR was of interest. For S alone, a 20% RR was of interest vs. a 5% null rate. Success is defined as 10 of 43 SG pts or 4 of 37 S pts. Accrual goals were 45 pts in the SG arm and 40 in the S arm. Results: Both arms met protocol criteria for continuing to stage 2 of accrual. 47 pts were randomized to SG and 40 to S alone. 3 pts withdrew before treatment, 1 of whom was ineligible. The analyzable set includes eligible treated pts registered before June 2015. RR, PFS and OS are shown in the table. The most common adverse events were anemia, nausea, fatigue, hypertension, and decreased neutrophil count. Grade $ 3 events were experienced by 29 in the SG arm and 17 in the S arm. Conclusions: This is the largest and first randomized trial for sarc RCC. The RR on S alone meets protocol criteria for success. For SG, we await results from recently enrolled pts (expected by the ASCO meeting). Though preliminary, the results suggest a role for the addition of chemotherapy in this difficult to treat subset of RCC. Clinical trial information: NCT01164228.

Baseline characteristics at first-line treatment.

Randomized Assessable as of 1/23/16 Response Rate 2-stage Exact 90% CI Median PFS 90% CI Median OS 90% CI Median cycles received Good Prognosis Intermediate Prognosis Poor Prognosis

Characteristic Gender Median Age KPS ,80% Dx to Tx interval,1yr Hypercalcemia Anemia Neutrophilia Thrombocytosis Prior Nephrectomy

Sunitinib

Pazopanib

P value

M71%/F29% 62.5 13% 55% 13% 45% 13% 15% 86%

M70%/F30% 65.3 14% 51% 13% 44% 12% 15% 88%

0.64 ,0.01 0.70 0.04 0.61 0.57 0.17 0.77 0.09

SG

S

47 35 7/35 = 20.0% 10.1 – 34.9% 23 weeks 13-25 weeks 41 weeks 32-90 weeks 3 5 (14.3%) 6 (17.1%) 24 (68.6%)

40 36 4/36 = 11.1% 5.1 – 34.0% 13 weeks 12-25 weeks 33 weeks 29-59 weeks 2 7 (19.4%) 7 (19.4%) 22 (61.1%)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Genitourinary (Nonprostate) Cancer 4512

Poster Discussion Session; Displayed in Poster Session (Board #135), Mon, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Mon, 4:45 PM-6:00 PM

4513

247s

Poster Discussion Session; Displayed in Poster Session (Board #136), Mon, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Mon, 4:45 PM-6:00 PM

DNA methylation profiling of renal cell carcinomas subtypes to identify epiclusters linked to cell ontogeny. First Author: Gabriel G. Malouf, PitieSalpetriere Hospital, Paris, France

Phase 2 trial results of DN24-02, a HER2-targeted autologous cellular immunotherapy in HER2+ urothelial cancer patients (pts). First Author: Dean F. Bajorin, Memorial Sloan Kettering Cancer Center, New York, NY

Background: Although genotype-phenotype correlations have been described for different subtypes of renal cell carcinoma (RCC), it is unknown if DNA methylation profiles correlate with morphological or ontology based phenotypes. Here we test the hypothesis that DNA methylation signatures of different RCC can discriminate between putative precursor cells in the nephron. Methods: We performed deep profiling of DNA methylation using Digital Restriction Analysis of DNA Methylation (DREAM) based on next generation sequencing analysis of methylation-specific signatures in diverse histopathological RCC subtypes (n = 22). We furthermore validated our findings in an independent dataset of diverse histopathological RCC subtypes (n = 44) and in The Cancer Genome Atlas Clear Cell (n = 424) and Chromophobe Renal Cell Carcinoma (n = 66) Datasets. Results: Our data provide the first mapping of methylome epi-signature and indicates that RCC subtypes can be grouped into two major epi-clusters: C1 which encompasses clear-cell RCC (n = 4), papillary RCC (n = 5), mucinous and spindle cell carcinomas (n = 2) and translocation RCC (n = 5); C2 which comprises oncocytoma (n = 2) and chromophobe RCC (n = 4). Interestingly, C1 epicluster displayed three fold more hypermethylation as compared to C2 epicluster. Of note, differentially methylated regions between C1 and C2 epiclusters occur in gene bodies and intergenic regions, instead of gene promoters. These data was validated in an independent dataset using supervised clustering. Transcriptome analysis of C1 epi-cluster suggests a functional convergence on Polycomb targets, whereas C2 epi-cluster displays DNA methylation defects. Finally, our epigenetic ontogeny signature was associated with worse outcomes of patients with clear-cell RCC, suggesting that cell of origins matters for RCC prediction purposes. Conclusions: Our data defines two epi-clusters that can discriminate between distinct RCC subtypes and for the first time, to our knowledge, define the epigenetic basis for proximal versus distal tubule derived kidney tumors. Genes defining ontogeny epi-signature might be used in the future to predict patients outcome.

Background: DN24-02 is an autologous cellular immunotherapy consisting of antigen-presenting cells (APC) cultured with BA7072, a recombinant HER2-derived antigen (HER500) linked to granulocyte macrophage colonystimulating factor. NeuACT (NCT01353222) is an open-label, randomized, phase 2 trial of adjuvant DN24-02 or standard of care (SOC) surveillance in HER2+ urothelial cancer pts with high relapse risk after surgery. Study objectives were overall survival (OS), disease recurrence-free survival (DRFS), product potency, and immunologic response. Methods: Pts expressing HER2 were randomized 1:1 to DN24-02 or SOC (target = 180 pts). DN24-02 pts received infusions every 2 wk (for a total of 3). Product potency was assessed using CD54 upregulation as a measure of APC activation. Antigen-specific cellular response was evaluated by T cell proliferation and IFN-g ELISPOT; humoral response was measured by ELISA. OS and DRFS were compared using Cox regression analysis. Results: In 142 enrolled pts with 13.2 mo of follow-up, OS was 37.0 (DN24-02) vs 22.2 mo (SOC) (hazard ratio [HR] 0.96; p = 0.87), and DRFS was 11.9 (DN24-02) vs 10.1 mo (SOC) (HR 0.99; p = 0.97). In subgroup OS analyses, more favorable HRs were observed in pts with lower disease burden (i.e., lack of lymph node involvement), no prior neoadjuvant chemotherapy, and lower neutrophil/lymphocyte ratios. DN24-02 increased APC activation, in vitro IL2 and IFN-g accumulation, serum cytokines (IL-2, IFN-g, and TNF-a), and antigen-specific T cell responses. BA7072 and HER500 antibody titers were significantly elevated from baseline through month 16. Conclusions: No statistical differences were observed in OS or DRFS; however, interpretation of these results is restricted by limited enrollment and short follow-up. Pts with certain characteristics of lower tumor burden had the most notable HRs for OS. APC activation and cytokine patterns were consistent with an immune prime boost. DN24-02–induced immune response persisting . 1 y suggests T cell activation and immune memory generation, consistent with mobilized antitumor immune responses. Together, these results suggest that additional investigation may be warranted. Clinical trial information: NCT01353222.

4514

4515

Poster Discussion Session; Displayed in Poster Session (Board #137), Mon, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Mon, 4:45 PM-6:00 PM

Avelumab (MSB0010718C; anti-PD-L1) in patients with metastatic urothelial carcinoma from the JAVELIN solid tumor phase 1b trial: Analysis of safety, clinical activity, and PD-L1 expression. First Author: Andrea B. Apolo, Genitourinary Malignancies Branch, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD Background: Avelumab* is a fully human anti-PD-L1 IgG1 antibody under clinical investigation in multiple cancers. We report updated safety and efficacy associated with avelumab as a 2nd-line therapy in patients (pts) with metastatic urothelial carcinoma (mUC; NCT01772004). Methods: Pts with mUC unselected for PD-L1 expression received avelumab 10 mg/kg IV Q2W until progression, unacceptable toxicity, or withdrawal. Tumors were assessed every 6 wks (RECIST 1.1). Unconfirmed objective response rate (ORR), progression-free survival (PFS), and overall survival (OS) were evaluated. Adverse events (AEs) were graded by NCI-CTCAE v4.0. PD-L1 expression was assessed by IHC. Results: As of Oct 7, 2015, 44 pts (27 [61.4%] with visceral metastasis) were treated with avelumab (median 14 wks [range 2-56]) and followed for a median of 11 mos (range 10-13). Median age was 68 y (range 30-84), ECOG PS was 0 (43.2%) or 1 (56.8%), and median number of prior therapies was 2 (range 1-6). Treatment-related (TR) AEs occurred in 30 pts (68.2%); the most common ($10%) were infusion-related reaction (20.5%), fatigue (20.5%), asthenia (11.4%), and nausea (11.4%). Grade $3 TRAEs were asthenia, myositis, decreased appetite, and elevated CPK or AST (each 1 event) and no treatment-related deaths occurred. ORR was 18.2% (8 pts; 95% CI: 8.2, 32.7) with 2 CRs and 6 PRs; 4 were ongoing. SD was observed in 17 pts (38.6%); disease-control rate was 56.8%. PD-L1 expression was evaluable in 35 pts. Using a $5% cutoff for tumor cell staining, 12/35 [34.3%] were PD-L1+; ORR was 50.0% in PD-L1+ pts (6/12; 95% CI: 21.1, 78.9) vs 4.3% in PD-L1- pts (1/23; 95% CI: 0.1, 21.9). PFS rate at 24 wks was 58.3% (95% CI: 27.0, 80.1) in PD-L1 + pts vs 16.6% (95% CI: 4.2, 36.0) in PD-L1-. ORR in pts +/- baseline visceral metastasis was 18.5% (5/27) and 17.6% (3/17), respectively. OS at 12 mos was 50.9% (95% CI: 32.6, 66.6) for the overall population. Conclusions: Avelumab showed an acceptable safety profile and promising clinical activity in pts with mUC. Greater activity in pts with PD-L1+ tumors was observed. A randomized phase 3 trial of avelumab in pts with mUC is underway. *Proposed INN. Clinical trial information: NCT01772004.

Oral Abstract Session, Sun, 8:00 AM-11:00 AM

Updated efficacy and . 1-y follow up from IMvigor210: Atezolizumab (atezo) in platinum (plat) treated locally advanced/metastatic urothelial carcinoma (mUC). First Author: Robert Dreicer, University of Virginia School of Medicine, Charlottesville, VA Background: A majority of mUC patients (pts) treated with chemo ultimately progress, and subsequent therapy is often accompanied by low ORRs and short OS. Atezo (MPDL3280A) has been found to be safe and effective in plat treated pts. Here we report updated efficacy in pts with . 1-y median follow up. Methods: IMvigor210 (NCT02108652) cohort 2 mUC pts who progressed during/ following plat received atezo 1200 mg IV q3w until loss of clinical benefit. PD-L1 status was centrally scored on tumor-infiltrating immune cells (IC; SP142 IHC assay): IC2/3, 1, 0. Confirmed RECIST v1.1 ORR (RECIST, central review) and modified (m)RECIST ORR (investigator) were co-primary endpoints. DOR, OS and safety were secondary (data cut Nov 27, 2015). Results: 310 efficacy/safety evaluable pts had median age 66 y; 37% had prior cystectomy; 41% had $ 2 prior regimens for mUC; 39% had chemo within past 3 mo. Responses were consistent in IC2/3 pts (26% RECIST ORR; 29% mRECIST) as well as across all IC subgroups, and enriched in pts with higher IC status (Table). RECIST responses were seen in pts with ECOG PS1 (8% ORR; 16/193), CrCl , 60 mL/min (13%; 14/ 110), Hb , 10 g/dL (7%; 5/69), visceral mets (10%; 24/243) and $ 4 prior mUC regimens (8%; 2/24). Responses were ongoing in 37/46 pts (80%) per RECIST and 46/60 (77%) per mRECIST. mDOR was not reached in IC2/3, IC1/2/3 and all comers. With longer follow up (median 14.4 mo [range 0.2-17.1]) an evolution of responses including additional CRs was seen. mOS and 1-y OS were notable in all IC subgroups (Table). Atezo remains well tolerated. Conclusions: Atezois an efficacious monotherapy in heavily pretreated mUC pts. Durable responses occurred across predefined IC subgroups, including pts with poor prognostic factors. Clinically meaningful OS was seen. With sustained efficacy coupled with favorable safety, atezo shows promise as a new standard of care in plat treated mUC (expanded access study: NCT02589717). Clinical trial information: NCT02108652. RECIST n IC2/3

100

IC1/2/3

207

Allb a

Not estimable;

310 b

ORR, % 95% CI 26 18, 36 18 13, 24 15 11, 19

mRECIST CR, n 12 14 17

ORR, % 95% CI 29 20, 39 23 17, 29 19 15, 24

OS CR, n 8 16 18

mOS, mo 95% CI

1-y OS, % 95% CI

11.9 9.0, NEa 9.0 7.1, 10.9 7.9 6.7, 9.3

50 40, 60 40 33, 47 37 31, 42

IC0, 1, 2/3

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

248s 4516

Genitourinary (Nonprostate) Cancer Poster Discussion Session; Displayed in Poster Session (Board #138), Mon, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Mon, 4:45 PM-6:00 PM

4517

Poster Discussion Session; Displayed in Poster Session (Board #140), Mon, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Mon, 4:45 PM-6:00 PM

Avelumab (MSB0010718C; anti-PD-L1) in patients with advanced adrenocortical carcinoma from the JAVELIN solid tumor phase Ib trial: Safety and clinical activity. First Author: Christophe Le Tourneau, Institut Curie, Paris, France

Efficacy of BGJ398, a fibroblast growth factor receptor (FGFR) 1-3 inhibitor, in patients (pts) with previously treated advanced/metastatic urothelial carcinoma (mUC) with FGFR3 alterations. First Author: Sumanta K. Pal, City of Hope, Duarte, CA

Background: Avelumab* is a fully human anti-PD-L1 IgG1 antibody under clinical investigation in multiple cancers. We report safety and clinical activity of single-agent avelumab in patients (pts) with metastatic adrenocortical carcinoma (mACC), a rare malignancy for which there are limited therapeutic options (NCT01772004). Methods: Pts with mACC who had progressed after platinum-based therapy and were unselected for PD-L1 expression were treated with avelumab at 10 mg/kg IV Q2W until progression, unacceptable toxicity, or withdrawal. Prior and ongoing treatment with mitotane was permitted. Tumors were assessed every 6 wks (RECIST 1.1). Objective response rate (ORR) and progression-free survival (PFS) were evaluated. Adverse events (AEs) were graded by NCI-CTCAE v4.0. Results: As of Oct 23, 2015, 37 pts were treated with avelumab (median 8 wks [range 2-48]) and followed for a median of 13 wks (range 0-58). Median age was 50 y (range 23-71), ECOG PS was 0 (37.8%) or 1 (62.2%), and median number of prior treatments for metastatic disease was 1 (range 0-5). Treatment-related (TR) AEs occurred in 23 pts (62.2%); the most common (. 10%) were nausea (6 [16.2%]), fatigue (5 [13.5%]), pyrexia (5 [13.5%]), and infusion-related reaction (5 [13.5%]), all of grade 1/2. Grade $3 TRAEs occurred in 5 pts (13.5%; each 1 event): hyperkalemia, increased ALT, GGT, or transaminase, sepsis, spinal cord infection, and pneumonitis. Potential immune-related TRAEs occurred in 4 pts (10.8%): 3 pts with grade 1/2 events and 1 pt with grade 3 pneumonitis that resolved. There were no treatment-related deaths. Among 19 pts with $13 wks f/u, unconfirmed ORR was 10.5% (2 partial responses; 95% CI: 1.3, 33.1). Stable disease was observed in 5 pts (26.3%); disease control rate was 36.8% (7/19). Median PFS was 7.6 wks (95% CI: 5.9, 23.9), and PFS rate at 12 wks was 30.3% (95% CI: 12.3, 50.7). Conclusions: Avelumabshowed an acceptable safety profile and clinical activity in pts with mACC, a dataset representing the first study to date of an anti-PD-(L)1 agent in this rare tumor type. Analyses of activity, including response correlated with PD-L1 expression, are ongoing. *Proposed INN. Clinical trial information: NCT01772004.

Background: FGFR3 alterations are frequent in UC (» 21%) and have a putative role in pathogenesis and progression. Antitumor activity observed in pts with UC in a phase 1 trial of the FGFR inhibitor BGJ398 (NCT01004224) led to initiation of an extended cohort of genetically selected pts to further characterize BGJ398 activity in UC. Methods: Eligible pts had mUC with activating FGFR3 mutations/fusions and prior platinum-based chemotherapy unless contraindicated. Pts received BGJ398 at the recommended phase 2 dose (RP2D)/schedule (125 mg once daily, 3 weeks on/1 week off). Overall response rate (ORR) and safety were evaluated. Results: At data cutoff (Nov 2, 2015), 33 pts with FGFR3-altered mUC (mutation [n = 31], fusion [n = 2]; based on mUC tissue) were enrolled. Bone or liver metastases were present in 42% and 48% of pts, respectively. Most pts (76%) had $ 2 prior therapies. The Kaplan-Meier estimate of median duration of exposure was 13.3 weeks (95% CI, 7.0-22.3 weeks); 15 pts (45%) remained on treatment at data cutoff (max, 42+ weeks). The ORR in 25 evaluable pts was 36% and included 1 unconfirmed complete response (CR; pathologically confirmed in a pt with persistent bone metastases after 2 cytotoxic regimens) and 8 partial responses (4 confirmed). Most responses were observed after 2 treatment cycles. Of 7 pts with stable disease, 6 had tumor reductions ranging from 10%-26%. Safety was similar to prior reports of the RP2D/ schedule, with most adverse events (AEs) being G1/2 and manageable. Frequent AEs (all G; G3/4) were hyperphosphatemia (42%; 6%), constipation (36%; 0%), fatigue (36%; 6%), and elevated serum creatinine (36%; 3%). Conclusions: BGJ398 monotherapy is well tolerated and has encouraging activity in heavily pretreated pts with mUC. Responses were observed in pts with multiple prior therapies and in pts with bone metastases. This trial validates the biological role of FGFR3 and clinical benefit conferred by its blockade in UC. The 36% ORR observed in these patients is notable given their poor prognosis and limited therapeutic options. The CR in the pt with osseous metastases is unique and warrants further exploration. Clinical trial information: NCT01004224.

4518

4519

Poster Discussion Session; Displayed in Poster Session (Board #141), Mon, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Mon, 4:45 PM-6:00 PM

Poster Discussion Session; Displayed in Poster Session (Board #142), Mon, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Mon, 4:45 PM-6:00 PM

Genomic evolution in muscle-invasive bladder cancer resistant to neoadjuvant chemotherapy. First Author: David Liu, Dana-Farber Cancer Institute, Boston, MA

Differences in somatic DNA damage repair (DDR) gene alterations between primary and secondary muscle invasive bladder cancer (MIBC). First Author: Eugene J Pietzak, Memorial Sloan Kettering Cancer Center, New York, NY

Background: The majority of patients with muscle-invasive bladder cancer (MIBC) treated with neoadjuvant cisplatin-based chemotherapy (NAC) have residual disease at time of cystectomy. Changes in the tumor genomic landscape after NAC may inform genetic resistance patterns and treatment strategies. Methods: Matched pre- and post-NAC tumor samples were obtained from 31 patients with MIBC who had gross residual disease ($ pT2) at cystectomy, followed by whole exome sequencing of these “trios” (preand post-NAC tumor with matched germline samples). Linear regression was used to adjust for differences in sample purity and predict changes in mutation and neoantigen burden from pre- to post-NAC. Phylogenetic analysis of serial tumor biopsies was performed to identify alterations gained or lost after treatment. Results: A wide proportion of mutations (5-81%, median = 31%) and neoantigens (4-77%, median = 32%) in the post-NAC tumor were not seen in the matched pre-NAC tumor. After adjusting for differences in sample purity, there was no statistically significant overall change in the mutational or neoantigen burden post-NAC (mean = -0.67 mut/Mb [p = 0.06] and -0.88 neoantigens/Mb [p = 0.10] respectively), but the change was strongly associated with pre-NAC mutation and neoantigen levels (R2= 0.44, p , 0.0001; R2= 0.40, p = 0.0002 respectively): tumors with higher baseline levels experienced greater decreases. While no single gene was significantly enriched for nonsynonymous mutations exclusively in pre- or post-NAC samples, 32% (10/31) of patients had either a gain or loss of alterations in potentially actionable genes, including PIK3CA and MTOR. Four post-NAC tumors harbored focal E2F3 amplification, including one acquired after NAC. Conclusions: These results suggest that NAC decreases pre-existing mutation and neoantigen burden while simultaneously revealing new mutations and neoantigens. This may occur by elimination of less fit subclones and induction of new mutations from DNA damage. Further, potentially targetable mutations are lost and acquired after NAC; this heterogeneity suggests that rebiopsy in metastatic/recurrent settings after NAC may be necessary to accurately inform clinical management.

Background: Alterations in DDR genes, including the nucleotide excision repair gene ERCC2, have been shown to be associated with response to systemic chemotherapy in MIBC (Van Allen, 2014). We hypothesized that patients with initial non-invasive bladder cancer treated with intravesical BCG who progress to “Secondary” MIBC (S-MIBC) would have fewer DDR gene alterations, than patients initially presenting with “Primary” MIBC (P-MIBC). Methods: Using data from TCGA and an institutional genomic research database of 545 patients, we identified 33 patients with S-MIBC and matched them by stage, gender, and age to 76 P-MIBC patients (Discovery set). We then validated our findings in a prospective cohort of 89 consecutive patients sequenced on an institutional clinical protocol with a CLIA-certified targeted exon-capture assay; 23 with S-MIBC and 66 with P-MIBC. Somatic alterations in DDR genes were compared between S-MIBC and P-MIBC tumors. Results: Patient demographics were evenly distributed in both the Discovery and Validation sets (Table). Within the Discovery set, somatic mutations in ERCC2 occurred in 20% of P-MIBC tumors (15/76), but were absent in S-MIBC tumors (0/33) (p = 0.01). No differences were seen in other DDR genes. These results were confirmed in our validation set, as 18% of P-MIBC tumors (12/66) had ERCC2 mutations, but none of the S-MIBC tumors (0/23, p = 0.03). Conclusions: Somatic ERCC2 mutations occurred frequently in P-MIBC tumors (19%, 27/142), but were absent in S-MIBC tumors (0/56), suggesting that tumors progressing to muscle invasive disease after intravesical BCG may be less likely to respond to systemic chemotherapy. Discovery Cohort

N= Median Age (IQR) Male Stage T2 T3 T4 Node Positive Metastasis ERCC2 Any DDR Gene ERCC3 ERCC4 ATM BRCA1 BRCA2 FANCA FANCC ATR CHEK2 MSH2 MSH3 MSH6 POLE

S-MIBC

P-MIBC

33 68 (63, 76) 82%

76 73 (66, 78) 79%

45% 36% 18% 39% 0 0 9 (27%) 1 0 2 1 3 1 2 1 2 1 0 1 1

34% 51% 14% 42% 0 15 (20%) 31 (41%) 0 0 7 2 4 1 2 2 3 0 1 0 2

Validation Cohort p= 0.2 0.8 0.4

0.5 0.01 0.4

S-MIBC

P-MIBC

23 67 (60, 74) 91%

66 64 (56, 70) 88%

52% 39% 9% 44% 48% 0 5 (22%) 0 1 2 0 1 0 0 1 0 0 0 0 0

63% 29% 9% 36% 39% 12(18%) 20 (30%) 0 1 5 1 2 1 0 1 1 0 0 1 1

p= 0.3 0.9 0.1

0.6 0.6 0.03 0.6

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Genitourinary (Nonprostate) Cancer 4520

Poster Session (Board #143), Mon, 1:00 PM-4:30 PM

Circulating tumor cells (CTC) and HER-2 status on CTC and primary tumor in urothelial cancer (UC) patients refractory to platinum based chemotherapy. First Author: Maria De Santis, University of Warwick, Cancer Research Centre, Coventry, United Kingdom

4521

249s Poster Session (Board #144), Mon, 1:00 PM-4:30 PM

FIESTA: A phase Ib and pharmacokinetic trial of AZD4547 in combination with gemcitabine and cisplatin. First Author: Robert Hugh Jones, Cardiff University and Velindre Cancer Centre, Cardiff, United Kingdom

Background: There are no established serum markers to monitor UC progression, response to therapy and to evaluate the risk for metastases. This study was set up to define the percentage of CTC positive UC pts at progression after platinum based combination therapy, potential CTC cut-off levels for drug screening and the HER2 status on CTC in comparison to the primary tumor. Methods: Blood samples from pts who fulfilled the inclusion criteria were analysed with CellSearch- CTC Kit and -Tumor Phenotyping Reagent HER2/neu. Results were reported as number of CTC/7.5 mL whole blood. Paraffin embedded tumor tissue was evaluated for HER2 positivity (FISH) and compared to that on CTC. The correlation between CTC parameters, the HER2 status of the primary tumor, and the Bellmunt prognostic groups (0-3) were investigated using descriptive statistics. The planned sample size of 70 pts should allow for a two-sided 95% confidence interval with a width of +/- 12% for the expected rate of 50% CTC positive pts. Results: 70 pts with platinum refractory UC were enrolled in 8 institutions and 69 pts fulfilled all inclusion criteria and CTC of 63 pts were evaluable. 66/69 pts had measurable metastases (38 visceral metastases of which 23 were in the liver). 21, 29, 14 and 5 pts belonged to Bellmunt prognostic groups 0,1,2,3, respectively. In 42 of 63 evaluable pts (66%) CTC ($ 1 CTC) were detectable, and 29 (46.0%), 25 (39.7%) and 23 (36.5%) pts had more than 2, 3 and 5 CTC/7.5ml, respectively. In 3 pts HER2-positive CTC were detected; HER2 gene amplification on the primary tumor was found in 6/46 cases (13%). All patients with HER2-positive primary tumors also harboured CTC. HER2-positive CTC were found in 2 pts with HER2 amplified primary tumors and in one patient with a HER2negative primary tumor. In 3/5 pts with HER2 amplified tumors CTC were HER2 negative. HER2 positive CTC were significantly correlated with the Bellmunt risk group 3 (60%, CI 23.1- 88.2) but not with the HER2 status of the primary tumor. Conclusions: CTC are present in a considerable number of platinum refractory UC patients and might serve as a drug screening biomarker as well as for the detection of targetable molecular alterations. Clinical trial information: EudraCT 2013-000124-34.

Background: gemcitabine (G) plus cisplatin (C) is a standard-of-care, combination chemotherapy regimen for patients (pts) with bladder cancer (BC), in both neoadjuvant and palliative settings. AZD4547 is a selective fibroblast growth factor receptor (FGFR) tyrosine kinase inhibitor with potential to improve outcomes for tumors with FGFR mutation and/or overexpression, as commonly seen in BC. We have conducted a phase Ib dose escalation trial, combining GC with AZD4547. Methods: eligible pts were treated with escalating doses of oral AZD4547 (40mg, 80mg,100mg BD, 2 weeks on/1 week off) and GC (G 1000mg/m2 IV, days 1&8; C 70mg/m2 IV, day 1), every 21 days. Pts were prospectively screened, then monitored, for retinal pigmented epithelial detachment (RPED). Results: 19 pts were recruited to 3 dose levels at 3 UK centers. There were no treatment-related deaths or serious treatment-emergent adverse events. One dose-limiting toxicity was seen at 80mg BD of AZD4547 (G4 thrombocytopenia) and 2 at 100mg BD (G4 thrombocytopenia; G3 hyponatremia). The maximum tolerated dose was therefore defined as 80mg BD. 3 pts completed 3 or more cycles, without significant toxicity, at this dose level. The recommended dose for subsequent studies is therefore 80mg BD. RPED was seen in 4 of 6 pts at 80mg BD and 4 of 9 pts at 100mg BD. Only one reported symptoms prior to RPED screening. In all cases, RPED resolved spontaneously, on withdrawal of AZD4547. A protocol amendment permitted re-starting of AZD4547 in 2 pts, after resolution of RPED. In both cases, RPED recurred. Tmax pharmacokinetic data for the 3-drug combination were in line with those previously reported for AZD4547 alone; Cmax values, in the presence vs absence of GC, gave an 8-day accumulation factor of approximately 2. Conclusions: a recommended dose of 80mg BD, 2 weeks on/1 week off, has been determined for AZD4547 in combination with GC, with manageable toxicity. A comparison of toxicities of GC with and without AZD4547 is under way in a randomised expansion cohort of advanced bladder cancer patients. FIESTA was developed by the UK NCRI Bladder Cancer Clinical Studies Group, sponsored by the University of Leeds, funded by the Cancer Research UK/Astra Zeneca Combinations Alliance. Clinical trial information: 44149443.

4522

4523

Poster Session (Board #145), Mon, 1:00 PM-4:30 PM

Poster Session (Board #146), Mon, 1:00 PM-4:30 PM

Comparison of upper tract urothelial carcinoma and urothelial carcinoma of the bladder to reveal key differences in mutational profile and load. First Author: Sumanta K. Pal, City of Hope, Duarte, CA

mRNA expression levels and prognostic value of PD1/PDL1 and CTLA4 pathways genes in a large series of 155 bladder tumors. First Author: Geraldine Pignot, Institut Paoli-Calmettes, Marseille, France

Background: Upper tract urothelial carcinoma (UTUC) is generally thought to have a more aggressive clinical phenotype as compared to urothelial carcinoma of the bladder (UCB). We interrogated genomic differences between these entities. Methods: DNA was extracted from 40 microns of FFPE sections from 295 and 195 consecutive patients with UCB and UTUC, respectively. Comprehensive genomic profiling (CGP) was performed on hybridization-captured, adaptor ligation based libraries to a mean coverage depth of 627X for up to 315 cancer-related genes plus 37 introns from 14 genes frequently rearranged in cancer. The CGP assay included base substitutions (SUB), INDELs, copy number alterations (CNA) and fusions/ rearrangements. Genomic alterations (GAs) linked to drugs on the market or under evaluation in mechanism driven clinical trials were considered clinically relevant GAs (CRGAs). A 2-sided Z-test for comparison of population proportions was used to juxtapose the frequency of GAs and CRGAs in UTUC and UCB. Results: There were 70% male and 30% female patients with a mean age of 66 years. Although both UCB and UTUC harbored a similar number of GAs/sample (6.4 for both), CRGAs were slightly more common with UCB (2.6 vs 2.0). The most frequent GA was in TP53, and this was more common in UCB (56% vs 46%, P = 0.03). In contrast, the next most frequent GA (CDKN2A) was more common in UTUC (43% vs 34%, P = 0.047). Amongst CRGAs, differences in FGFR3 alteration approached statistical significance, with a higher frequency in UTUC (28% vs 21%, P = 0.08). A borderline increase in frequency of ERBB2 alterations in UCB was also seen (17% vs 11%, P = 0.07). Mutational load was higher in UCB as compared to UTUC (6.6 GAs/MB vs 5.5 GAs/MB). Conclusions: These results highlight key differences in the frequency of GAs present in UTUC and UCB. In addition to explaining the clinical phenotype of these disease, they may inform ongoing and future trials of targeted therapy in this population.

Background: Immunotherapy in bladder cancer seems to give promising results. The purpose of this study is to evaluate the involvement of immune response during bladder carcinogenesis and to compare expression levels in non-muscle invasive bladder cancer (NMIBC) versus muscle-invasive bladder cancer (MIBC). Methods: Expression levels of 3 genes involved in PD1 pathway (PD1, PDL1 and PDL2), 4 genes involved in CTLA4 pathway (CTLA4, CD28, CD80 and CD86) and 27 other immune genes (including LAG3 and TIM3) were assessed by quantitative real time RT-PCR and by immunochemistry in normal and tumoral human bladder samples. Tumors were obtained from 155 patients (25 women and 130 men, mean age 68 years). All patients have signed an informed consent. Pathological tumor staging showed: 71 NMIBC (29 low grade pTa, 14 high grade pTa and 28 high grade pT1), and 84 MIBC ($ pT2) all of high grade. Results were coupled to a survival analysis. Results: PD1 and PDL1 were significantly overexpressed in MIBC compared to normal bladder tissue (59.5 vs 6.7% and 60.7 vs 0% respectively, p , 0.01), whereas the proportion of overexpression was low in NMIBC (22.5% and 4.2% respectively). Similarly, CTLA4 and one of its ligands CD80 were significantly overexpressed in MIBC as compared to normal bladder tissue (84.5 vs 20.0% and 92.9 vs 6.7% respectively, p , 0.01), whereas CD80 was individually overexpressed in 46.5% of NMIBC without overexpression of CTLA4. There was a high concordance rate between expression levels of the two pathways, according to Spearman test (0.88, p , 0.0001). The results of the molecular analysis were confirmed by immunohistochemistry with a good correlation. There was no correlation between mRNA expression levels of the studied genes and prognosis in terms of relapse / progression for NMIBC and in terms of recurrence-free and overall survival for MIBC. Conclusions: Our results confirm the role of immune checkpoints in bladder carcinogenesis, with a good correlation between RT-PCR and immunochemistry analysis, and opening new therapeutic perspectives, especially for invasive tumors. However, these deregulations do not concern NMIBC and seem not to be associated with survival outcomes.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

250s 4524

Genitourinary (Nonprostate) Cancer Poster Session (Board #147), Mon, 1:00 PM-4:30 PM

Adjuvant versus neoadjuvant chemotherapy for muscle-invasive bladder cancer (MIBC): Analysis of the National Cancer Database (NCDB). First Author: Guru Sonpavde, University of Alabama at Birmingham Comprehensive Cancer Center, Birmingham, AL Background: Neoadjuvant chemotherapy (NC) preceding radical cystectomy (RC) is the standard for muscle invasive bladder cancer (MIBC). The efficacy of post-RC adjuvant chemotherapy (AC) is unproven, although meta-analyses and retrospective analyses support its efficacy. We conducted a retrospective analysis of the NCDB to compare the overall survival (OS) with NC vs. AC. Methods: The NCDB was employed including patients (pts) with new diagnoses of urothelial carcinoma of bladder in the US from 2004-2013. Pts with clinical stage T2-T4aN0 who underwent RC were included. Those with prior malignancy and prior radiotherapy were excluded. Multivariate analyses were conducted to determine whether NC conferred differential impact on OS compared with AC after controlling for baseline clinical stage, age, race, year of diagnosis, Charlson Comorbidity Index (CCI), number of lymph nodes examined at RC and gender. Owing to suboptimal durations between RC and chemotherapy and significant discord between clinical and pathologic staging in cT2N0 pts, sub-analyses were conducted in such pts with time from first chemotherapy to RC , 6 months (mo) in the NC group, and time from RC to first chemotherapy , 4 mo in the AC group. Results: A total of 7704 pts were eligible for analysis, of whom 1624 received NC, 839 received AC and 5241 underwent RC alone. On multivariate analysis, NC was associated with significantly longer OS compared to AC (HR = 1.19 [95% CI: 1.05-1.36], p = 0.008). Sub-analyses comparing the cT2N0 group that received NC followed by RC within 6 mo vs. the modeled and congruent AC group that received RC followed by AC within 4 mo showed no significant OS difference (HR = 0.93 [95% CI: 0.64-1.36]). Limitations of a retrospective analysis apply. Conclusions: Overall, NC conferred better OS than AC in MIBC patients with cT2-T4aN0, although OS appeared similar in modeled pts receiving more optimal NC and AC. These data suggest caution in interpreting results from meta-analyses and retrospective analyses supporting AC as an alternative to NC, and suggest that NC should remain the preferred standard for MIBC. A trial comparing NC vs. AC should be considered.

4526

Poster Session (Board #149), Mon, 1:00 PM-4:30 PM

4525

Poster Session (Board #148), Mon, 1:00 PM-4:30 PM

Subclassification and outcome prediction of patients with muscle invasive urothelial carcinoma (MIUC) treated by radical cystectomy (RC) with a NanoString based molecular screening. First Author: Sebastien Ly Rinaldetti, Department of Hematology and Oncology, University Medical Centre Mannheim, Medical Faculty Mannheim, University of Heidelberg, Mannheim, Germany Background: Transcriptome expression studies identified distinct bladder cancer subtypes of prognostic relevance and closely related with breast cancer molecular subclasses. Here we developed a molecular subtypescreening in a chemotherapy naive patient cohort with MIUC. Methods: FFPE samples of MIUC (pT2-T4, G2-3, exclusion of squamous carcinoma) were collected from a cohort of 47 patients (collection period from 1998-2012) treated exclusively with RC and bilateral lymphadenectomy. 80 genes were selected using two approaches. First, genes were selected from the prognostic gene signatures published previously (basal, luminal & p53-like). Second, we reanalyzed published microarray data with gene set enrichment analysis. Gene expression was measured with the NanoString nCounter technology after RNA extraction with a bead based system (Stratifyer). Results: Unsupervised hierarchical clustering separated the tumors into 3 clusters (Pearson coefficient . 0,75): basal, luminal and infiltrated. These subtypes present a distinct OS and DSS, over a follow-up period up to 12 years. The luminal subtype revealed an aggressive phenotype such as an OS of 24 months (p = 0,002) and a DSS of 38 months (p = 0,017). In contrast the infiltrated subtype showed best outcome (OS: 93 mo, DSS: 128 mo) whereas basal patients formed an intermediate risk group (OS: 87 mo, DSS: 108 mo). In silico validation of our gene panel with two recent microarray studies (GSE48276, GSE13507) confirmed the prognostic outcomes for MIUC. Our study confirms a highly subtype specific expression of drug targets. Genes like ERBB2 (HER2neu), ERBB3 and the FGFR gene family are exclusively expressed in the luminal subtype (p , 0,003) whereas EGFR is higher in the basal subtype (p = 0,031). Conclusions: This nCounter based molecular screening allows a risk stratification and survival prediction based on three main subtypes: basal, luminal and infiltrated. As the expression of drug targets is highly subtype specific, this molecular subclassification is needed in future clinical studies in order to underline its translational benefit by means of personalized therapy.

4527

Poster Session (Board #150), Mon, 1:00 PM-4:30 PM

Phase 2 trial of dovitinib in Bacillus Calmette-Guerin (BCG) refractory urothelial carcinoma (UC) with tumor FGFR3 mutations or overexpression: Hoosier Cancer Research Network GU12-157. First Author: Noah M. Hahn, Johns Hopkins Kimmel Cancer Center, Baltimore, MD

Programmed death-ligand 1 (PD-L1) characterization of circulating tumor cells (CTCs) in muscle invasive and metastatic bladder cancer patients. First Author: Archana Anantharaman, University of California, San Francisco, San Francisco, CA

Background: FGFR3 aberrations are common in low grade non-muscle invasive bladder cancer (NMIBC) tumors but decrease in frequency in metastatic UC. This multisite single-arm trial assessed the clinical and pharmacodynamic activity of dovitinib (an oral FGFR3/VEGFR2 inhibitor) in a treatment resistant, molecularly enriched NMIBC population. Methods: Patients (pts) with BCG refractory NMIBC despite at least 2 prior intravesical therapy regimens with increased tumor FGFR3 expression by centrally analyzed immunohistochemistry (IHC+) or FGFR3 mutations (Mut+) assessed in a CLIA-certified lab were eligible. Pts received dovitinib 500 mg once daily (5 days on / 2 days off) on q28d cycles. The primary endpoint was 6-month TURBT-confirmed complete response (CR) rate. Results: Between 11/2013 and 10/2014, 13 pts enrolled (10 IHC+ Mut-, 3 IHC+ Mut+). Demographics included: median age 70 years; 85% male; CIS (3 pts), Ta/T1 (8 pts), and Ta/T1 + CIS (2 pts); median prior regimens 3; median time from last treatment 6 months. Pts received a median of 4 dovitinib cycles. Dose reductions were required in 10 pts (77%). Treatment related grade 3/4 events included: fatigue, hypertension, hypertriglyceridemia (2 pts, 15% each); hepatotoxicity, stomatitis, rash (1 pt, 8% each). 6-month complete response rate was 8% (0% in IHC+ Mut-; 33% in IHC+ Mut+). The primary endpoint was not met. One pt remains a CR at 15+ months. One pt (8%) developed muscle-invasion on study. Pharmacodynamically active (94-5812 nM) dovitinib concentrations in urothelial tissue were observed in all 9 PK-evaluable pts. Reductions in pFGFR3 IHC staining were observed post-dovitinib treatment. A concordant urine FGFR3 mutation was noted in 1 of the 3 Mut+ pts. Conclusions: Oral dovitinib consistently achieved biologically active concentrations within the urothelium and demonstrated pharmacodynamic FGFR3 inhibition. Long-term administration of dovitinib was not feasible due to frequent treatment related toxicity. Absent clinical activity in FGFR3 IHC+ Mut- pts suggests that this should not be an entry criterion for future trials targeting FGFR3 in UC. Clinical trial information: NCT01732107.

Background: Recent studies indicate that PD-1 and PD-L1 checkpoint inhibitors have activity in chemotherapy refractory patients with muscle invasive (MIBC) and metastatic (mBCa) bladder cancer. PD-L1 expression on tumor cells or lymphocytes may correlate with response to therapy. To identify potential patients who may benefit from PD-1/PD-L1 targeted immunotherapeutics, we utilized a non-invasive blood test to evaluate PD-L1 protein expression in CTCs and WBCs of bladder cancer patients. Methods: Blood samples from 22 patients with MIBC or mBCa were collected at UCSF and shipped to Epic Sciences. All nucleated cells were subjected to immunofluorescent (IF) staining and CTC identification by fluorescent scanners and algorithmic analysis. Cytokeratin (CK+) expressing CTCs (CK+, CD45-, intact nuclei morphologically distinct from WBCs) as well as CK- CTCs were enumerated and correlated with available clinical data. A subset of patient samples underwent further genetic characterization by FISH. Results: CTCs were detected in 19/22 (86.4%) patients, inclusive of CK(+) CTCs (12/22, 55%), CK(-) CTCs (14/22, 70%), CK(+) CTC Clusters (6/22, 27%), and apoptotic CTCs (12/22, 54%). Seven of 22 (32%) patients had PD-L1(+) CTCs; 4 of these patients had exclusively CK(-)/CD45(-)/PD-L1(+) CTCs. A subset of CTCs were secondarily confirmed as bladder cancer via FISH. High PD-L1(+)/CD45(-) CTC burden was associated with poorer overall survival (HR: 3.29, log-rank p = 0.041) Conclusions: Patients with MIBC and mBCa who have detectable PD-L1(+) CTCs, and patients with high PD-L1(+) CTCs have worse survival compared to patients with low or absent PD-L1 expression.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Genitourinary (Nonprostate) Cancer 4528

Poster Session (Board #151), Mon, 1:00 PM-4:30 PM

Circulating cell-free DNA profiling of patients with advanced urothelial carcinoma of the bladder. First Author: Rebecca J Nagy, Guardant Health, Inc., Redwood City, CA Background: Urothelial carcinoma of the bladder (UCB) exhibits one of the highest somatic mutation burdens. Circulating cell-free DNA (cfDNA) is obtained non-invasively from peripheral blood and appears detectable in most patients (pts) with advanced UCB. We report cfDNA profiling of patients with advanced UCB using biopsy-free cfDNA sequencing. Methods: Pts with advanced UCB that underwent cfDNA analysis using Guardant360 were identified. A 70-gene cfDNA next generation sequencing (NGS) panel from a CLIA-licensed, CAP-accredited laboratory (Guardant Health, Inc.) offers complete exon sequencing for 29 cancer genes, critical exons in 39 genes and amplifications (16 genes), fusions (6 genes) and indels (3 genes) harvested from 10 mL of peripheral blood. Results: Of 71 patients with advanced UCB, cfDNA was detectable in 61 patients (85.9%). The median age was 70 years (range 45-89). The most common recurrent somatic mutations were in TP53 (n = 35), BRCA1/2 (n = 20), ARID1A (n = 17), FGFR2/3 (n = 17), KRAS/RAF1/BRAF/MAPK (n = 15), and NF1 (n = 12), MET (n = 10) and ERBB2 (n = 10). The most common genes with increased copy numbers were kinase genes (ERBB2, PIK3CA, EGFR, RAF1, BRAF, MET and FGFR1) and cell-cycle controlling genes (MYC, CCNE1, CDK4/6). The cfDNA somatic alteration burden appeared similar to those of other major malignancies. Serial cfDNA profiling of pts receiving chemotherapy revealed the clonal evolution of mutations in genes associated with resistance including BRCA2, NF1 and GATA3. Analysis of additional patients is ongoing in this expanding dataset. Conclusions: cfDNA was frequently detected in patients with advanced UCB, and alterations were most frequently seen in TP53, kinase genes, epigenetic modifiers and cell-cycle controlling genes, which appears similar to those previously reported from muscle-invasive UCB tumor tissue NGS. Given that cfDNA offers a noninvasive means of profiling tumor DNA, identification of the mechanisms of resistance targeted with this test should be evaluated for impact on response.

4530

Poster Session (Board #153), Mon, 1:00 PM-4:30 PM

Comparative effectiveness of treatment strategies for urothelial cancer of the bladder (UCB) with clinical lymph node involvement (cN+). First Author: Matt D. Galsky, Icahn School of Medicine at Mount Sinai, New York, NY Background: Patients (pts) with cN+ UCB are high risk, but potentially curable. Such pts are excluded from neoadjuvant chemotherapy (chemo) trials and pooled with pts with distant metastases in 1st line chemo trials not suited to define the role of combined modality therapy. To address this evidence void, we performed a comparative effectiveness analysis. Methods: We included cTanyN1-3M0 UC pts from the National Cancer Database (2003-2012) treated with multi-agent chemo and/or cystectomy (cyst). We used multi-state survival analysis allowing for delayed entry to assess overall survival (OS) according to various treatment strategies. Effectiveness was estimated with multivariable adjustment for tumor-, patient-, and facility-level characteristics. To disentangle the effectiveness of specific approaches, we additionally censored for adjuvant chemo and adjusted for pathological stage. Results: Among 1,756 pts (cN1=48%, cN2=45%, cN3=7%), 1,104 underwent cyst and 652 were treated with chemo alone. Of the cyst pts, 363 received pre-op chemo and 328 received adjuvant chemo. Crude 5-year OS for cyst alone, chemo alone, pre-op chemo, and adjuvant chemo was 19%, 13%, 31%, and 26%, respectively. Multivariable multi-state models censored for start of adjuvant chemo demonstrated that pre-op chemo improved OS, an effect associated with improved pathological stage (Table). Survival of pts treated with chemo alone was similar to pts undergoing cyst alone, while adjuvant chemo independently improved OS (Table). Conclusions: In this large observational cohort, a sizeable subset of pts with cN+ UCB achieved durable disease control. Treatment with combined modality therapy, with chemo administered pre- or post-cyst, was associated with significantly improved OS. Multivariable adjusted hazard ratios (HR) for OS (reference: cyst +/- adjuvant). Analysis

Treatment

Overall

Chemo alone Pre-op chemo Chemo alone Pre-op chemo Pre-op chemo Adjuvant chemo

Censored for adjuvant chemo Censored for adjuvant chemo, adjusted for path stage Adjusted for path stage

HR (95% CI) 1.34 0.93 1.15 0.80 1.14 0.69

(1.18-1.53) (0.78-1.11) (0.99-1.35) (0.65-0.97) (0.98-1.33) (0.57-0.83)

4529

251s Poster Session (Board #152), Mon, 1:00 PM-4:30 PM

Randomized, placebo-controlled phase II trial (MAJA): Efficacy results of maintenance vinflunine after cisplatin chemotherapy (CT) in patients with advanced urothelial carcinoma (UC)—SOGUG 2011-02. First Author: Albert Font, Institut Catala` d’Oncologia, Hospital Universitari Germans Trias i Pujol, Badalona, Spain Background: Vinflunine (VFL) is a microtubule inhibitor approved by EMA as treatment after platinum progression in metastatic UC. We evaluated whether maintenance VFL delays progression after response to CT. Methods: Patients (pts) with measurable disease, locally recurrent/metastatic UC and adequate organ function with radiological response or stabilization after 4-6 cycles (cy) of a cisplatin/gemcitabine chemotherapy (carboplatin allowed after cy 4) were randomized (R) 1:1 to receive VFL 320 or 280mg/m2 (in case of PS1, age $ 75 years, prior pelvic radiotherapy (RT) or CrCl , 60ml/min) every 21 days vs best supportive care (BSC), until disease progression. Primary endpoint was progression free survival (PFS). With a median PFS considered unacceptable for the experimental arm of 4 months (m) (p0) and acceptable 6.5m (p1). 78 eligible were required for an a-error of 0.05 (one-tailed test) and a 0.1 b-error. Results: 88 pts from 21 institutions of SOGUG were R between 04/2012-01/2015. Forty five in the VFL arm and 43 in the BSC arm. 1 pt was not treated. 1 of 87 treated pts was not eligible due to an excess of time between the last dose of cisplatin and the start of VFL. At the beginning of treatment: median age 64 years [42-83]; PS1 50%; Hb , 10g/dl 8%; liver metastasis 21%. Pts in the VFL arm received a median of 6 cy per patient [1–43]. 10 pts (22.2%) continue in treatment in VFL arm. Most common G3/4AEs (%pts) in VFL arm were constipation (13.6), neutropenia (15.9), fatigue (15.9), and 1 febrile neutropenia (2.3). After a median of 12.2 m of follow-up [0.5-41.4], 59% of pts have progressed and 43% of pts have died in the VFL arm; 81% and 62% pts respectively in the BSC arm. The median PFS was of 6.5 m (1.3-11.7) in the VFL arm and 4.6 m (3.1-6) the BSC arm; HR 0.56 (IC95%; 0.34-0.93, p = 0.024). After progression, 34% of pts received treatment at VFL arm and 60% of pts at BSC arm. Conclusions: Maintenance VFL in patients with disease control after first line cisplatin-based chemotherapy significantly reduces 44% the risk of progression with an acceptable tolerability profile. Trial is maturing to assess the impact of maintenance VFL in survival. Clinical trial information: NCT01529411.

4531

Poster Session (Board #154), Mon, 1:00 PM-4:30 PM

Function of microRNA activity by ActMiR in bladder cancer. First Author: Ana Collazo Lorduy, Icahn School of Medicine at Mount Sinai, New York, NY Background: MicroRNAs (miRs) play a key role in cancer, in tumorigenesis and tumor progression. In the past years, miR expression signatures have been reported as prognostic biomarkers in bladder cancer (BC). However, miR’s expression does not always correlate with activity. We recently developed a novel computational method, named ActMiR, for explicitly inferring the activity of miRs based on the changes in expression levels of target genes. The objective was to validate the inferred miRs’ activity using BC as a model. Methods: We applied ActMiR in 405 BC cases from the TCGA database for which information from both mRNA and miR expression was available. Different BC cell lines (5637 and HT1376 for basal BC, SW780 and HT1197 for p53-like BC, and RT4 and RT112 for luminal BC) as well as the immortalized urothelial cell line (HUC) were used to perform the in vitro functional validation. RNA was extracted basally and after inhibition of miR106-5p with specific anti-miR inhibitor; miR and target genes’ expression was assessed by qRT-PCR. Results: ActMir analyses revealed that a subset of 306 out of 1044 miRNAs were differentially expressed between tumor and normal samples at p-value , 10-4. More importantly, at p-value , 10-4, 155 out of 556 miRNAs were functionally active. From these, only four (miR106b, miR532, mir556, and mir134) were significantly differentially expressed, functionally active and showed prognostic significance in the TCGA BC cohort. We chose to further analyze mir106b because it showed the biggest and most significant difference in activity between normal and cancer tissues (p-value , 2*10-11). As inferred by ActMir, mir106b-5p was significantly overexpressed in all cancer cells when compared to HUC. We used p53 cell lines to confirm that inhibition of mir106b-5p showed upregulation of the predicted target genes, because ActMir had predicted that miR106 expression and activity were both associated with overall survival in this BC subtype. Conclusions: Our results underscore the value of ActMir for inferring miR activity in BC from a systems biology perspective. It endorses ActMir as a promising tool for studying casual effects of miR activity on target genes, and ultimately for determining survival outcomes of BC patients.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

252s

Genitourinary (Nonprostate) Cancer

4532

Poster Session (Board #155), Mon, 1:00 PM-4:30 PM

4533

Poster Session (Board #156), Mon, 1:00 PM-4:30 PM

Anti-tumor activity, safety and pharmacokinetics (PK) of AGS15E (ASG15ME) in a phase I dose escalation trial in patients (Pts) with metastatic urothelial cancer (mUC). First Author: Daniel Peter Petrylak, Yale University, New Haven, CT

Anti-tumor activity, safety and pharmacokinetics (PK) of ASG-22CE (ASG22ME; enfortumab vedotin) in a phase I dose escalation trial in patients (Pts) with metastatic urothelial cancer (mUC). First Author: Jonathan E. Rosenberg, Memorial Sloan Kettering Cancer Center, New York, NY

Background: SLITRK6 is a cell adhesion molecule highly expressed on several tumors, including mUC. ASG-15ME is an antibody drug conjugate (ADC) that delivers a small molecule microtubule-disrupting agent, monomethyl auristatin E (MMAE), to tumor cells expressing SLITRK6. Methods: mUC pts unselected for SLITRK6 expression (determined by an IHC assay) and previously treated with $ 1 prior chemo regimen or unfit for Cisplatin were enrolled using a modified continual reassessment method design. Disease assessments (DA) were performed every 8 weeks (wks) using RECIST v 1.1. ASG15ME was administered IV weekly for 3 out of every 4 wks until no further benefit. 6 dose levels were studied: 0.1, .25, .5, 0.75, 1, or 1.25 mg/kg. Results: Of analyzed tumor tissues, 85% (n=46) were SLITRK6 positive (56% had H-score $150). As of 1/21/16, 43 pts were enrolled. Anti-tumor activity was seen at dose levels $ 0.5 mg/kg (n=37 reported here). Median age= 64 y; 100% ECOG PS # 1; 22 pts (56%) had $ 2 lines of prior therapy. 1 pt had a confirmed complete response (CR) and 10 pts had partial response (PR) (ORR (PR and CR) =30%) including 3/10 pts (30%) with liver metastases. Median time on study was 14 wks. 37 pts (95%) had adverse events (AEs). The most common AE was fatigue (56%). 20 pts (51%) had Grade 3/4 AEs with 35% considered related. 9 pts (23%) had eye disorders (1 Gr 3). 4 pts (11%) had protocol defined dose limiting toxicities. There were 2 deaths unrelated to study drug. Serum concentrations of ASG-15ME decreased multi-exponentially and exposure was dose-proportional. ASG15ME half-life is 2.3 days. The median progression free survival is 16 wks (unplanned exploratory analysis). Enrollment is open at active dose levels; updated results will be presented. Conclusions: ASG-15ME targeting SLITRK6 is an ADC in mUC that is well tolerated and active. These results warrant ongoing study of ASG-15ME in mUC pts. Clinical trial information: NCT01963052.

Background: Nectin-4 is a cell adhesion molecule expressed on several tumor types, including mUC. ASG-22ME is an antibody drug conjugate (ADC) that delivers a small molecule microtubule-disrupting agent, monomethyl auristatin E (MMAE), to tumor cells expressing Nectin-4. Methods: Pts with solid tumors including mUC treated with $ 1 prior chemo regimen or mUC pts unfit for Cisplatin were enrolled using a modified continual reassessment method design. Pts were prescreened for Nectin-4 expression (IHC assay) and enrolled if H-score $150. Disease assessments (DA) were performed every 8 weeks (wks) using RECIST v 1.1. ASG-22ME was administered IV wkly for 3 out of every 4 wks until no further clinical benefit. 4 dose levels were studied: .5, .75, 1, or 1.25 mg/kg. Results: Of analyzed tumor tissues, 93% (n=243) were Nectin-4 positive (83% had H-score $ 150). As of 1/21/16, 39 solid tumor pts were enrolled; 31 with mUC reported here. Median age= 65 y; 100% ECOG PS # 1; 21 pts (68%) had $ 2 lines of prior therapy for mUC. 7 pts (25%) had a PR (ORR (PR and CR) =25%), including 3/10 pts (30%) with liver metastasis. Median time on study was 15 wks. 30 pts (97%) had adverse events (AEs). The most common AE was fatigue (42%). 18 pts (58%) had Grade 3/4 AEs, with 44% considered related. 7 pts (23%) had eye disorders (Grade 1/2). 2 pts (6%) had protocol defined dose limiting toxicities. There were 2 deaths unrelated to study drug. Serum concentration of ASG-22ME decreased multi-exponentially and exposure was dose-proportional. ASG-22ME half-life is 1.5 days. Anti-tumor activity was seen at all dose levels. The median progression free survival is 17 wks (unplanned exploratory analysis). Enrollment is open at active dose levels: updated results will be presented. Conclusions: ASG-22ME targeting Nectin-4 is an ADC in mUC that is well tolerated and active. These results warrant ongoing study of ASG-22ME in mUC pts. Clinical trial information: NCT02091999.

Dose (mg/kg)

mUC Pts only.

Evaluable pts (n=37) ORR, n (%) Responders (PR/CR) Median Duration on Rx (Wks) DCR (SD/PR/CR) n (%)

0.5

0.75

1

1.25

8 1 (13)

14 4 (29)

10 4 (40)

5 2 (40)

17

24

20

10

3 (38)

10 (71)

8 (80)

4 (80)

Evaluable pts = $ 1 dose of drug and had $ 1 post-baseline DA.

Dose (mg/kg) Evaluable pts (n=28) ORR, n (%) Responders (PR/CR) Median Duration on Rx (Wks) DCR (SD/PR/CR) n (%)

0.5

0.75

1

1.25

2 1 (50)

11 3 (27)

11 1 (9)

4 2 (50)

65

24

31

12

2 (100)

9 (82)

7 (64)

3 (75)

Evaluable pts = $ 1 dose of drug and $ 1 post-baseline DA.

4534

Poster Discussion Session; Displayed in Poster Session (Board #139), Mon, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Mon, 4:45 PM-6:00 PM

A phase II study of cabozantinib in patients (pts) with relapsed or refractory metastatic urothelial carcinoma (mUC). First Author: Andrea B. Apolo, Genitourinary Malignancies Branch, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD Background: Hepatocyte growth factor (HGF) and its receptor (MET) are activated in urothelial carcinoma (UC). In translational studies, cabozantinib, a multiple receptor tyrosine kinase inhibitor primarily targeting MET and VEGFR2, reversed HGF-driven UC cell growth and invasion. Methods: In this phase II study, pts received cabozantinib 60 mg/day in 28day cycles in 3 cohorts: 1) mUC, 2) bone-only mUC, and 3) metastatic rare bladder histology. The primary objective was response rate (RR) by RECIST v1.1. Results: Of67 pts enrolled, 70% were male, median age: 63 (22–82), KPS 80% in 73% of pts. Primary sites: bladder (69%), upper tract (25%), both (6%). No. of prior therapies: 1/2/3/$4 (range 1–6) in 34/39/16/9% of pts, respectively; 85% had visceral metastases (lung, liver, bone); sites of metastases included LN/lung/bone/liver/soft tissue/carcinomatosis in 69/ 36/22/17/21/2% of pts, respectively. Grade 3 adverse events (AEs) in $ 2 pts: fatigue 9%, HTN 7%, hypophosphatemia 6%, diarrhea 8%, AST/ALT increase 7%, thrombocytopenia 4%, lymphopenia 3%, hyponatremia 3%, and proteinuria 3%. Grade 4 AEs: hypomagnesemia 3% and lipase increase 1%. Cohorts 1/2/3 enrolled 48/6/13 pts, respectively. Of 41 evaluable pts in cohort 1, there was 1 complete response (CR), 7 partial responses (PR) (RR = 19.5%; 95% CI: 8.8–34.9%), 18 stable disease (SD), and 15 progressive disease (PD). Median progression-free survival (mPFS): 3.7 mos (95% CI: 2.3–6.5); 6-month PFS: 38.0% (95% CI: 23.3–52.6%). Median overall survival (mOS): 8.2 mos (95% CI: 5.2–10.3); 6-month OS: 65.1% (95% CI: 48.3–77.6%); 12-month OS: 25.7% (95% CI: 13.0–40.3%). Median follow-up was 23.7 mos. Of 5 evaluable pts in cohort 2, 3 (60%) had improvement by NaF FDG/PET. mPFS: 5.3 mos (95% CI: 1.8–8.3); mOS: 9.3 mos (95% CI: 3.6–12.5). Of 10 evaluable pts in cohort 3, there were 0 CR/ PR, 5 SD, and 5 PD. mPFS: 2.9 mos (95% CI: 1.8–3.7); mOS: 4.6 mos (95% CI: 2.6–8.0). Conclusions: Cabozantinib has clinical activity in pts with relapsed or refractory mUC with manageable toxicities. Further studies are underway to correlate responses with MET expression, immune subsets, CTCs, and cytokine and mutational profiles. Clinical trial information: NCT01688999.

4535

Poster Session (Board #157), Mon, 1:00 PM-4:30 PM

Differential expression of PD-L1 expression in high grade T1 (HGT1) v. muscle invasive urothelial carcinoma (MIUC) and its prognostic implications. First Author: Stephanie A. Mullane, Bladder Cancer Center, Dana-Farber Cancer Institute, Brigham and Women’s Cancer Center, Boston, MA Background: Programmed death-1 (PD-1) receptor negatively regulates T cell-mediated responses.PD-1 ligand (PD-L1) can be expressed on both tumor cells and mononuclear (MNC) infiltrates in MIUC. Increased MNC PDL1 positive (PD-L1+) staining is correlated with better overall survival in metastatic UC. Several trials are now exploring the role of checkpoint inhibitors in non-MIUC (NMIUC). However, differences in PD-L1 expression between NMIUC (such as HGT1 tumors) and MIUC remain unknown. Methods: Formalin-fixed paraffin embedded (FFPE) tumor samples from 109 clinically annotated HGT1 UC patients (pts) were retrieved. All pts were initially diagnosed with HGT1 UC by transurethral resection (TUR), subsequently received BCG, and were followed for a median of 7.4 years. PD-L1 expression was evaluated by immunohistochemistry using a mouse monoclonal anti-PD-L1 antibody (405.9A11). The assay was validated using FFPE cell line controls known to be positive or negative for PD-L1 expression by flow cytometry. PD-L1+ was defined as $ 5% tumor cell membrane staining. MNC PD-L1 expression was scored based on the extent of infiltrate and percentage of positive cells. Fisher’s exact tests assessed the associations of disease outcome (no recurrence, recurrence, or progression) with MNC presence, PD-L1+ MNC, and PD-L1+ tumor cells. Comparison analysis with PD-L1 expression results in MIUC is presented. Results: Among 109 HGT1 NMIUC pts, PD-L1+ expression in tumor cells was seen in 6 (6%) pts and in MNC in 40 (38%) pts. PD-L1+ expression was not correlated with disease outcomes in tumor cells (p = 0.3) nor in MNC (p = 0.47). The pattern of PD-L1 staining in tumor cells (but not MNC) was significantly different when comparing MIUC and HGT1, with less positivity seen in HGT1 tumors (p = 0.001). Among three patients who had paired TUR and cystectomy samples, one patient changed from negative to positive PD-L1 MNC. Conclusions: PD-L1 is widely expressed in MNC, but not in tumor cells in HGT1 NMIUC with no correlation to recurrence or progression to muscle invasion. PD-L1 expression in tumor cells is significantly higher in MIUC compared to HGT1 NMIUC.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Genitourinary (Nonprostate) Cancer 4536

Poster Session (Board #158), Mon, 1:00 PM-4:30 PM

Patterns of chemotherapy administration in bladder preservation therapy (BPT) for muscle-invasive bladder cancer (MIBC). First Author: Tracy Lynn Rose, UNC Hospital, Durham, NC Background: Trimodality BPT in MIBC includes a maximal transurethral resection followed by concurrent chemoradiotherapy as an alternative to cystectomy in appropriately selected patients, or as a potential treatment in non-cystectomy candidates. Several chemotherapy regimens are used in BPT, but little is known about current practice patterns. This report describes chemotherapy utilization patterns and associated outcomes with BPT in MIBC. Methods: Data were collected from the Retrospective International Study of Cancers of the Urothelial Tract (RISC), a database of 3,024 consecutive patients from 29 international academic centers from 2005 to 2013. Patients with clinical T2-T4aN0M0 urothelial cancer of the bladder were included. Results: 269 patients received BPT. Compared with the 1,448 patients who received radical cystectomy, BPT patients were older (p = 0.01), less likely to have ECOG performance status of 0 or 1 (p , 0.01), and more likely to have clinical T4a disease (p , 0.01). 41% of BPT patients received concurrent chemotherapy with radiation. Patients receiving chemotherapy were younger (p = 0.01) and less likely to have T3 or T4 disease (p , 0.01) compared to patients who received radiotherapy alone. 25% of patients treated with chemotherapy received cisplatin only and 21% received carboplatin only. Additional regimens included gemcitabine alone (8%), paclitaxel (8%), 5-FU+mitomycin (5%), as well as others. Chemotherapy-treated patients had an adjusted HR for death of 0.89 (95% CI 0.61-1.29) compared to those who received radiotherapy alone after controlling for age, clinical T stage, and Charlson comorbidity index (median overall survival 26.5 vs 26.2 months, p = 0.37). There were no significant differences in survival among chemotherapy regimens. Only 10 patients (4%) eventually underwent cystectomy after BPT. Conclusions: A minority of patients undergoing BPT receive concurrent chemotherapy. The choice of chemotherapy used varies widely in clinical practice, with no clear standard. Salvage cystectomy is rarely performed. Continued research is needed on the comparative effectiveness among chemotherapy regimens in BPT.

4538

Poster Session (Board #160), Mon, 1:00 PM-4:30 PM

Prognostic significance of DNA damage repair (DDR) mutations in patients with urothelial carcinoma (UC) and associations with tumor infiltrating lymphocytes (TILs). First Author: Russell Bailey McBride, Department of Pathology & The Institute for Translational Epidemiology, Icahn School of Medicine at Mount Sinai, New York, NY Background: Somatic mutations in DDR genes have been associated with improved outcomes in patients with muscle-invasive bladder cancer postcystectomy, even in the absence of perioperative chemotherapy. This counterintuitive finding has been hypothesized to result from the accumulation of missense mutations encoding antigens that trigger anti-tumor immunity. We sought to validate whether DDR mutations in UC were associated with improved outcomes, and to test whether this association is mediated by TILs. Methods: We performed immunohistochemical staining for CD8 (Dako) and hybridization-based next-generation sequencing (MSKIMPACT) of all exons and selected introns of 26, previously identified DDR genes in 77 cystectomy specimens (MSH cohort). Mutation rates per case were dichotomized (any vs. none). TILs were measured using an intensity weighted H-Score of CD8+ expression. Clinical information was complete on 63 cases and included age, sex, TNM stage, and neo-adjuvant therapy. Progression-free survival (PFS), disease-specific survival (DSS), and overall survival (OS) were estimated using proportional hazard models. Statistical analyses were performed using SAS 9.3. Results: At least one DDR mutation was found in 32/63 (50.8%) of the cases. The most common mutations were BRCA1 (7.9%), BRCA2 (6.3%), and ATRX (6.3%). Patients with vs. without DDR mutations did not differ by age, sex, TNM stage, or neoadjuvant therapy. The mean CD8+ H-score was numerically higher in those with DDR mutations (20.2) than in non-mutants (15.9) (Wilcoxon 2-sided p-value = 0.10). Patients with DDR mutations had roughly half the risk of recurrence (HR = 0.53 95% CI [0.26-1.11], p = 0.09) or disease-specific death (HR = 0.51 95% CI [0.24-1.09], p = 0.08). Conclusions: DDR mutations were associated with improved outcomes in patients with UC post-cystectomy, though our sample size was inadequate to reach statistical significance. We also identified a non-significant association between CD8+ tumor expression and DDR mutations. Prospective validation of the association of DDR mutations with clinical outcomes and proposed mechanisms are indicated.

4537

253s Poster Session (Board #159), Mon, 1:00 PM-4:30 PM

Natural history of clinical complete response to neoadjuvant chemotherapy for urothelial carcinoma of the bladder: Updated single-institution experience. First Author: Justin T. Matulay, Department of Urology, Columbia University Medical Center, New York, NY Background: Approximately 30-40% of patients with muscle-invasive bladder cancer (MIBC) who undergo neoadjuvant chemotherapy (NAC) have a complete pathologic response at radical cystectomy. It is unknown if patients identified via endoscopic evaluation as complete responders can safely undergo surveillance. We describe our updated experience with a cohort of such patients. Methods: Retrospective review of 44 patients with proven MIBC ($cT2) and clinical complete response (cCR) to NAC. Each patient underwent platinum based NAC and cCR was defined as negative urine cytology, normal TURBT, and normal cross-sectional imaging. Patients were followed post-NAC with cytology, cystoscopy with biopsy, and crosssectional imaging. Results: Mean age of the cohort was 67.2 6 9.6 years with a median follow-up of 37.4 6 38.3 months. The NAC regimens were: MVAC 47.7% (21), gemcitabine/cisplatin 31.8% (14), other 18.2% (7), and unknown 2.3% (1). Table 1 summarizes the clinical stage and outcomes of the cohort. Kaplan-Meier analysis shows a 5-yr cancer-specific survival (CSS) for the immediate and non-immediate RC groups of 87.5% vs. 100%, p=0.182, respectively. The overall mean CSS is 134.1 6 10.2 months. The mean CSS for immediate RC versus non-immediate RC is 134.1 months and 99.4 months, respectively, but the difference is not statistically significant. Conclusions: In a highly select subgroup of patients who have no evidence of measurable disease following NAC for MIBC, vigilant surveillance results in a 57% chance of avoiding radical surgery and an overall survival of 72.7% after a median follow-up of 38.5 months. Future studies should focus on identifying the clinical, pathologic, and molecular factors associated with a durable complete response. Clinical stage and post-cT0 outcomes.

No. pts Pre-NAC clinical stage (No. pts) Median days from MIBC diagnosis to NAC (range) No. death/total No. No. cancer related death/total No.

4539

Immediate Cystectomy

Delayed Cystectomy

Recurrence with Bladder in situ

Bladder preserved, NED

8 cT2 (8) 55 (5-103)

11 cT3 (1), cT2 (10) 20 (1-78)

11 cT2 (11) 27 (3-121)

14 cT3 (1). cT2 (13) 47 (11-118)

4/8 2/8

4/11 3/11

3/11 0/11

1/14 0/14

Poster Session (Board #161), Mon, 1:00 PM-4:30 PM

Genomic predictors of recurrence (R) or progression (P) in high grade T1 (HGT1) non-muscle invasive (NMI) bladder cancer. First Author: Joaquim Bellmunt, Dana-Farber Cancer Institute, Boston, MA Background: HGT1 NMI tumors have a 40% risk of R and 21% risk of P. Distinguishing aggressive versus indolent HGT1 disease is critical to decide which patients will benefit from early cystectomy. We examined genetic features associated with recurrence and progression. Methods: FFPE tumor/ normal (T/N) samples were available from a clinically annotated cohort with HGT1 (7.4 y f/u), who were managed in a uniform manner using TURB and BCG. Clinical outcome was classified as: non-recurrent disease , 4 years (good outcome, GO), recurrent (R), or progression to MI or M1 (PD). 37 primary T/N matched sample pairs were subject to WES followed by Broad standard analysis pipelines, yielding 27 QC-passed T/N pairs (exome coverage . 23 Mb). Results: TP53 mutations (8/27) were most common in PD samples (2-R/ 6-PD, P = 0.009 by Fisher’s test, FDR = 0.2) while RB1 mutations (6/27) were enriched in GO samples (5-GO/ 1-PD, P = 0.024, FDR = 0.24). Other recurrent mutations not enriched by clinical subtype ( . 4) included ERBB3, KDM6A, STAG2, and KMT2C. Three mutational signatures were identified from single nucleotide variants: APOBEC (67%), C . T at CpG (4%), COSMIC-5 (unknown origin) (29%). The COSMIC-5 signature was enriched in GO samples (P = 0.03 by Mann-Whitney between GO vs PD +R (median 32% vs 4%). The sample with highest mutation-rate had more than 80% COSMIC-5 signature mutations, and an ERCC2 S246Y mutation. Two most common focal copy number events were CDKN2A deletions (2-GO/ 3-R/4-PD) and 1q23 amplification (AMP) spanning MCL1 (2-GO/4-R/2-PD) and PVRL4 (1-GO/3-R/3-PD). Recurrent focal AMP ( . 4 samples) also included MDM2(3-GO/2-R/1-PD), CCNE1 (3-R/2-PD), YWHAZ (1-GO/1-R/ 2-PD), and ERBB2 (2-GO/1-R/1-PD). Combined AMPs in E2F3/CCND1/ CCNE1 were enriched in PD or R samples (5-R/4-PD, P = 0.01 Fisher’s test, FDR = 0.14) and co-occurred with TP53 mutation (6 of 8 TP53 mutations, P = 0.006, FDR = 0.12). Conclusions: Analysis of mutations and copy number alterations in HGT1 bladder tumors may improve prediction of good outcome vs. recurrence or progression, enabling a management strategy of early cystectomy for those with poor prognosis, and continued observation for those with predicted good outcome.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

254s 4540

Genitourinary (Nonprostate) Cancer Poster Session (Board #162), Mon, 1:00 PM-4:30 PM

Trends in neoadjuvant chemotherapy (NAC) use for muscle-invasive bladder cancer (MIBC): An updated report using the National Cancer Database (NCDB). First Author: Paulo Gustavo Bergerot, Federal University of Sao Paulo (UNIFESP), Sao Paulo, Brazil

4541

Poster Session (Board #163), Mon, 1:00 PM-4:30 PM

Fixed tissue ChIP-seq (FiT-Seq) of archived FFPE clinical bladder cancer (BC) samples to reveal tumor-specific enhancer and super-enhancer profiles. First Author: Paloma Cejas, Dana-Farber Cancer Institute, Boston, MA

Background: NAC is a standard of care for MIBC. A report from the NCDB cites increasing use during a 5-yr timeframe (2006-2010; Reardon et al Eur Urol 2015). We examine trends over a 10-yr span and include unique populations (cT1 or N+) who receive this modality. Methods: Patients (pts) with MIBC (ICD-O-3 8120/8130) diagnosed from 2004 to 2013 were identified who had (1) received cystectomy, (2) no prior radiotherapy, and (3) complete clinicopathologic data. Univariate and multivariate analyses were conducted to determine whether demographic or clinicopathologic criteria were associated with use or non-use of NAC. Trend analyses were conducted to assess cumulative use of NAC, segregating pts with cT2-4/N0 disease. Results: 17,180 pts were included in the analysis, 2,622 of whom received NAC. From 2004 to 2013, an increase in NAC use from 14.8% to 40.0% was seen overall, with an increase from 8.4% to 35.5% in pts with cT2-4/N0 disease. Rates of NAC use for cT2-4/N0 disease rose in parallel to use for cT1 and N+ disease. While rates of NAC rose over this 10-year span, use of adjuvant chemotherapy was stable, ranging from 6.9% to 12.1%. On multivariate analysis, increasing age, non-White race, lower income, treatment in a community setting, high Charlson Index Group risk, and a lack of private insurance were associated with decreasing utilization of NAC (P , 0.01 for each). Conclusions: A broader longitudinal analysis of the NCDB highlights a more profound increase in NAC use over the past decade and also implies use in patients with cT1 and N+ disease. Use of adjuvant chemotherapy showed no substantial change over this 10-year span. Our study also highlights socioeconomic barriers to NAC use, such as income level, race and treatment setting. Interventions to improve NAC use in these settings should be considered.

Background: Application of ChIP-seq to the vast majority of clinical tissue samples has not been possible due to the extensive tissue fixation. We developed Fixed-Tissue chromatin immuno-precipitation sequencing (FiTseq), which reveals the epigenomes of FFPE tissue samples and enables many previously impossible translational studies of chromatin states to be done in archived clinical specimens. No chromatin state maps from primary BC have been described to date. Methods: Using multiple histone marks measured with FiT-seq we generate chromatin state maps for a 10-year old BC sample demonstrating a robust capability to study a wide variety of genomic regions. Six BC tumors were further analyzed with H3K4me2 to identify enhancers. The study was then expanded to a cohort of 24 high grade T1 clinically annotated bladder cancer samples with extreme phenotypes. Results: In the first 6 samples, we observed the existence of a cis-regulatory pattern characteristic for BC that distinguishes it from other tumor types including seminoma, breast, and colorectal. The BC enhancer pattern shows H3K4me2-defined super-enhancers near key bladder-cancer oncogenes, such as GATA3 and CCND1 in the majority of samples. The top-ranked super-enhancer in one case mapped to FGFR3, a driver oncogene in ~12% of bladder cancers. In the expansion cohort of 24 samples, we observed two main clusters in the unsupervised clustering based on the pattern of enhancers. Interestingly, one of the clusters corresponded to a subset of bladder tumors that had specific chromatin features with H3K4me2 marked enhancers at the vicinity of genes that where enriched in the p53 pathway. Conclusions: We show for the first time chromatin state maps from primary BC and describe the enhancer profiles of BC that shows distinct lineage characteristics. Our analysis suggests that histologically similar BC have different chromatin states, which may drive differential biology.

4542

4543

Poster Session (Board #164), Mon, 1:00 PM-4:30 PM

Poster Session (Board #165), Mon, 1:00 PM-4:30 PM

Sperm count in Swedish clinical stage I testicular cancer patients following modern adjuvant treatment. First Author: Kristina Weibring, Karolinska University Hospital, Stockholm, Sweden

Genome-wide association study of cisplatin-induced peripheral neuropathy (CIPN) in testicular cancer survivors. First Author: Omar El Charif, University of Chicago, Chicago, IL

Background: Little is known regarding sperm production following modern adjuvant treatment in clinical stage I (CSI) testicular cancer (TC). Methods: 182 TC patients aged 18-50 were prospectively included during 2001-2006 at any given time within 5 years of orchiectomy. Semen samples were delivered post op and 6, 12, 24, 36 and 60 months (T0-T60) after completion of therapy. Sperm concentration (SC) and total sperm number (TSN) were used as measurements of testicular function. Results: Treatment groups were: radiotherapy 1.8 Gy x 14 to a total dose of 25.2 Gy to the paraaortic and ipsilateral iliac lymph nodes (RT n = 70), one cycle of adjuvant BEP (Bleomycin, Etoposide, Cisplatin, 5 day regimen) (BEP n = 62), one cycle of adjuvant Carboplatin AUC 7 (Carb n = 22), and patients managed by surveillance (SURV n = 28). The mean SC and mean TSN increased slightly over time, T0-T60, except for the RT group where an obvious decrease was seen at T6 followed by recovery. At T12 compared to baseline, there was a significant increase in mean SC in the surveillance, BEP and Carb groups. A numerical but not significant increase in mean TSN at T12 compared to baseline was seen in all treatment groups except the RT group. At T12, no significant differences between the treatment groups BEP, Carb or RT overall, in mean SC or mean TSN compared to the surveillance group (p 0.11 and p 0.28) were found. Longitudinal studies showed that the majority of the patients in all four groups showed no deterioration in SC and TSN one year post treatment and most seemed to be rather stable thereafter. Except for in the BEP group, azospermia was observed in all treatment groups in a total of 11 patients. Only one azospermic patient became normospermic (Carb group). Conclusions: No clinically significant long-term effect on sperm concentration or total sperm number associated to modern adjuvant treatment in CSI TC was found. However, a number of patients have low sperm counts after orchiectomy and remain subfertile also after adjuvant treatment. Spermbanking should be offered preferably before orchiectomy as assisted reproductive measures may be necessary regardless of any treatment given.

Background: CIPN is a potentially permanent side effect of cisplatin chemotherapy. CIPN remains a major clinical challenge due to lack of effective treatment, impact on quality of life and unexplained inter-individual variability. Methods: Testicular cancer patients (n = 847) given cisplatin-based therapy (median dose: 400 mg/m2) were assessed for responses to the validated EORTC QLQ-CIPN20 questionnaire. Associations were evaluated between frequency of sensory neuropathy and cumulative cisplatin dose, smoking history and age. Using the Illumina HumanOmniExpressExome chip and with imputation, 5.1 million SNPs passed quality control (QC) for GWAS inclusion. A gene-based method, PrediXcan, was used to consider associations between the genetically determined component of gene expression and CIPN. Results: Sensory neuropathy was common (57% having any symptom). CIPN sensory items (n = 8) indicated excellent internal consistency (alpha coefficient = 0.88). Using each patient’s mean sensory neuropathy score, we sorted them into 3 ordinal groups according to severity: 43% none, 41% a little, 16% quite a bit/very much. Only age and smoking status, not dose, were significantly related to sensory neuropathy (OR = 1.04 [95% CI 1.03-1.06] and 1.5 [95% CI 1.2-2.0], respectively). The top 2 SNPs (P = 5 x 10-7) from a GWAS of 677 patients that passed QC, were each associated with greater risk of sensory neuropathy (OR = 1.9 [95% CI 1.52.4] and 1.9 [95% CI 1.5-2.5]). These SNPs are expression QTLs for LYPD3 (rs12797447), SNX8 and GSTT1 (rs4757366) in lymphoblastoid cell lines. PrediXcan identified lower predicted expression of RPRD1B in whole blood (5453 genes tested) met genome-wide significance for association with increased sensory neuropathy risk (P = 3 x 10-6). Conclusions: Our traditional GWAS identified several SNPs associated with CIPN that were functionally important due to their association with gene expression. Our gene-based test implicated RPRD1B, known to play a role in cell cycle regulation, as associated with CIPN. Future studies will focus on functional validation, analysis of a replication set and meta-analysis with other CIPN studies to identify drug-independent variants.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Genitourinary (Nonprostate) Cancer 4544

Poster Session (Board #166), Mon, 1:00 PM-4:30 PM

4545

255s Poster Session (Board #167), Mon, 1:00 PM-4:30 PM

Prognostic role of PD-L1 expressing tumor infiltrating lymphocytes in testicular germ cell tumors. First Author: Michal Chovanec, 2nd Department of Oncology, Faculty of Medicine, Comenius University; National Cancer Institute, Bratislava, Slovakia

Carboplatin AUC 10 in metastatic seminoma: Role of early PET scanning to reduce the amount of treatment while maintaining outcome—A phase 2 study (CarPET). First Author: Jonathan Shamash, St Bartholemew’s Hospital, London, United Kingdom

Background: Testicular germ cell tumors (TGCTs) are nearly universally curable malignancies. Nevertheless, standard chemotherapy is not curative in a small subgroup of patients. Previously we showed that PD-L1 overexpression is associated with worse prognosis in TGCTs, while tumor infiltrating lymphocytes (TILs) are prognostic in different types of cancer. The aim of this study was to evaluate the prognostic value of PD-L1 expressing TILs in TGCTs. Methods: Surgical specimens from 223 patients with TGCTs (215 with primary testicular tumor, 8 with extragonadal GCT) were included into this translational study. TILs expressing PD-L1 were detected by immunohistochemistry using anti-PD-L1 monoclonal antibody and PD-L1 expression was scored semiquantitatively by multiplicative quickscore (QS) method. Assessments were performed in all histopathologic subtypes of TGCTs, germ cell neoplasia in situ (GCNIS) and results were correlated with clinical outcome. Results: PD-L1 positive TILs were found significantly more often in seminomas (95.9% of patients) and embryonal carcinomas (91.0%) compared to yolk sac tumors (60.0%), choriocarcinomas (54.6%), teratomas (35.7%) or GCNIS (79.7%) (all p , 0.05). Similarly, percentage of PD-L1 expressing TILs was highest in seminoma (62.0% of all TILs), followed by embryonal carcinoma (43.2%), yolk sac tumor (32.3%), teratoma (24.2%) and choriocarcinoma (21.4%). PD-L1 positive lymphocytes between tubules were present in 79.7% of GCNIS cases. Patients with seminoma with high infiltration of PD-L1 positive TILs (QS $ 10) had non-significantly better progression-free survival (HR = 2.07, 95% CI 0.6 – 7.16, p = 0.28) and significantly better overall survival (HR = 6.98, 95% CI (1.75 – 27.91, p = 0.047) opposite to patients with lower expresion of PD-L1 (QS , 10). In other histopathological subtypes, PD-L1 expressing TILs were not prognostic. Conclusions: In this translational study, we showed for the first time prognostic value of PD-L1 expressing TILs in seminoma. Our data imply that PD-L1 positive TILs could play a protective role in seminoma patients and we suppose, that presence of these TILs could be responsible for better prognosis of seminoma compared to nonseminoma.

Background: Carboplatin montherapy for metastatic seminoma has not been pursued following two studies showing reduced progression-free survival compared to combination therapy. The doses were in-retrospect suboptimal and increased doses based on glomerular filtration rate have been associated with improved outcome. We postulated that an early response to carboplatin might allow de-escaltion of therapy without compromising efficacy Methods: Patients with metastatic untreated seminoma with IGCCCG good prognosis were recruited. PET scanning was performed prior to and post 1st cycle of carboplatin AUC 10 – based on an EDTA ( uncorrected) clearance. Those with a Deauville score of , 3 ( metabolic response) were given a total of 3 cycles, the rest received 4. Treatment was given every 21 days. Results: 48 patients were recruited ( 46 testicular, 2 extragonadal). 14 stage 2a, 23 stage 2b, 10 stage 2c The median age was 36 (24-66). 13 had an EDTA . 120 ml/min. One patient withdrew after 1 cycle due to severe Raynauds. He received radiotherapy and is progression-free. 20 had a metabolic response after 1 cycle and received 3 cycles. Overall radiologic response was 32 CR, 12 PR, 3 SD. With a median follow up of 22 months PFS was 97%, overall survival 100% at 2 years. Toxicity (Grade 3/4) neutropenia 19%, thrombocytopenia 13%, fatigue 2%, vomiting 2%, febrile neutropenia1%. Tinnitus (grade 1 in 12.4%, grade 2 in 4%), neuropathy - grade 1 in 8%. 15% of patients required blood and 9% required platelets overall. Conclusions: Carboplatin monotherapy is effective with low toxicity. PET scanning allows de escalation of therapy – comparison against combination therapy is warranted. Clinical trial information: EudracT 2009009882-33.

4546

4547

Poster Session (Board #168), Mon, 1:00 PM-4:30 PM

Suggested reclassification strategy applied to intermediate and poor risk nonseminomatous germ cell tumors (NSGCT): A two-institution combined analysis. First Author: Andrea Necchi, Fondazione IRCCS Istituto Nazionale Tumori, Milan, Italy Background: IGCCCG risk classification is used since 1997 to allocate metastatic GCT patients (pts). Recently, its applicability and survival estimates have been called into question. Guidelines recommend 4 cycles of BEP or VIP for both intermediate (int) and poor risk disease. We sought to revisit the outcomes and treatments in a population of int and poor risk NSGCT from Indiana and INT Milano. Methods: Data on consecutive pts receiving first-line chemotherapy (CT) from 1990 to 2014 were collected. Kaplan-Meier method was used to estimate relapse-free survival (RFS) and overall survival (OS). Cox regression analyses were used to evaluate potential prognostic factors of RFS and OS in an univariable model. Forward stepwise selection was used to construct a multivariable (MV) model. A risk factor (RF) model was then constructed and compared with IGCCCG classification using the concordance statistics (CS). Results: 647 pts were identified; median age was 27 years (range 13-60), 115 had a mediastinal primary (PMNSGCT), 60 (9.3%) brain, 131 (20.3%) liver, and 37 (5.7%) bone metastases (mets). 437 (67.5%) had received BEP CT. RF in the MV model were PMNSGCT (HR = 3.23, 95% CI = 2.28-4.59), brain mets (HR = 2.29, 95% CI = 1.48-3.55), lung mets (HR = 1.75, 95% CI = 1.22-2.50) and age at diagnosis ( $ 30 vs , 30, HR = 1.67, 95% CI = 1.22-2.29). CS were improved based on the RF model compared to IGCCCG classification (RFS: 0.63 vs 0.58; OS: 0.65 vs 0.59). For int risk, there were no differences between 3 (n = 26) and 4 cycles BEP (n = 160) or BEPx3+EPx1 (n = 31) for both RFS (p = 0.31) and OS (p = 0.055). Conclusions: In this contemporary cohort, the outcomes of int risk NSGCT have improved compared to historical estimates. Many int risk pts would not require BEPx4 to attain a cure. Using new RF, improved risk stratification was observed, which gives further evidence that a reclassification of GCT is needed. N IGCCCG Int IGCCCG Poor 0 RF 1 RF 2 RF $ 3 RF

232 415 126 294 169 58

2-yr RFS (95% CI)

2-yr OS (95% CI)

80.6 56.0 79.8 72.5 52.8 27.2

91.3 74.5 89.9 86.2 75.7 42.4

(74.8-85.3) (51.0-60.7) (71.5-85.8) (66.8-77.3) (44.8-60.1) (16.4-39.2)

(86.5-94.4) (69.9-78.5) (82.8-94.1) (81.5-89.8) (68.2-81.6) (29.0-55.3)

Poster Session (Board #169), Mon, 1:00 PM-4:30 PM

Chemotherapy and cognitive function in testicular cancer: A prospective, longitudinal cohort study. First Author: Ian N. Olver, University of South Australia, Adelaide, Australia Background: Neuropsychological assessments suggest chemotherapy may lead to cognitive impairment but causal links remain unclear. Methodological limitations can be addressed by prospective longitudinal studies using different chemotherapy regimens, appropriate controls and adjusting for confounders. Methods: This 16 centre, prospective longitudinal observational study accrued 145 patients with testicular cancer and analysing those with sufficient data, compared 2 groups, surgery only (n = 41) and surgery + chemotherapy (n = 61). Cognition was assessed by CogHealth, a 10-minute online playing-card tool that overcomes language limitations, assessment burden, and practice/ceiling effects. Quality of life, fatigue, anxiety, depression and self-perceived cognitive function were also assessed. Linear mixed models were used to compare changes in scores from baseline ( # 6 months after orchidectomy and before chemotherapy) and follow-up (12-18 months later). Results: There were no significant differences between the 2 groups from baseline to follow-up across the cognitive tasks; complex decision-making, visual learning, working memory or visual attention. Self-perceived cognitive function matched measured performance. However, there were significant interactions between groups (i.e. the surgery + chemotherapy group had a different trajectory over time compared to the surgery group) for psychomotor function (covarying age, p , 0.001) and for physical wellbeing (covarying radiotherapy, p = 0.008). The surgery + chemotherapy group scored significantly worse than the surgery group at baseline on psychomotor function (p = 0.03, d = -0.85), physical wellbeing (p , 0.0001, d = 0.80), fatigue (p = 0.01, d = 0.53) and anxiety (p = 0.005, d = -0.80), but there were no significant differences at follow-up. For both groups, anxiety (p = 0.03), functional (p = 0.001) and emotional wellbeing (p = 0.002) improved over time. Conclusions: For patients with testicular cancer there were no significant differences between groups for objective or self-reported cognitive function after 12 months, with only psychomotor function and physical wellbeing worse in the chemotherapy group at baseline. Clinical trial information: 12609000545268.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

256s 4548

Genitourinary (Nonprostate) Cancer Poster Session (Board #170), Mon, 1:00 PM-4:30 PM

4549

Poster Session (Board #171), Mon, 1:00 PM-4:30 PM

Extension of overall survival in patients with metastatic renal cell carcinoma who received HD IL-2 followed by targeted therapy and/or immune checkpoint blockade from the PROCLAIM registry. First Author: Joseph Clark, Loyola University Medical Center, Maywood, IL

Quality of life (QoL) and overall survival (OS) in patients (pts) with advanced clear-cell renal cell carcinoma (aRCC) treated with nivolumab (NIVI) vs. everolimus (EVE) in the phase III CheckMate 025 study. First Author: David Cella, Northwestern University, Feinberg School of Medicine, Chicago, IL

Background: High dose IL-2 (HD IL-2) can provide durable responses in patients with metastatic renal cell carcinoma (mRCC). PROCLAIM (www. proclaimregistry.com, NCT01415167) is an IL-2 registry with over 44 participating sites that captures real-world patient population survival and outcomes. Herein, we report on contemporary patient experience sequencing HD IL-2 with targeted therapy (TT) and immune checkpoint blockade (ICB) in mRCC. Methods: Patients were prospectively enrolled into the registry as of 2011 and must have received at least one dose of HD IL-2 for this analysis. Statistics and survival analysis were performed on datasets as of December 2, 2015. Results: The mOS for all mRCC patients (n=411) was not reached (NR), with a median follow-up of 21 months. The overall response rate (ORR) for the 382 patients with available data was 17.8%. The mOS for those who experienced complete response (CR, n=15), partial response (PR, n=53), or stable disease (SD, n=145) was not reached while in patients with progressive disease (PD, n=169), mOS was 17 months. Survival for patients treated with TT or ICB following HD IL-2 were further analyzed and described in Table 1. Treatments prior to HD IL-2 include chemotherapy (n=12), TT (n=74), immunotherapy (n=8), radiation (n=64), and surgery (n=389). Conclusions: This analysis of the national IL-2 registry reveals that HD IL-2 therapy followed by TT and/or ICB is potentially associated with survival benefit. HD IL-2 continues to be a valuable treatment option for eligible patients with mRCC. Clinical trial information: NCT01415167.

Background: In the CheckMate 025 study (NCT01668784), pts treated with nivo had an OS benefit vs eve (hazard ratio=0.73; 98.5% CI, 0.57 to 0.93), and median scores on the Functional Assessment of Cancer Therapy–Kidney Symptom Index–Disease Related Symptoms (FKSI-DRS) scale increased from baseline (BL) and were significantly (P,0.05) improved with nivo vs eve through week 104 (NEJM 2015;373:1803–13). Here we present a full analysis of QoL in relation to OS. Methods: QoL was measured by the kidneyspecific FKSI-DRS questionnaire at BL and every 4 weeks. Chi-square analyses were used to assess differences in the proportion of pts who experienced meaningful improvement (2-point change) with nivo vs eve. OS was assessed by subgroups of pts with FKSI-DRS improvement ($2-point increase), no change (,2-point increase or decrease), and deterioration ($2-point decrease) at week 8 using the Kaplan–Meier method. Results: BL QoL was collected for 86% of all randomized pts; 361/410 (nivo) and 343/ 411 (eve). Over the course of the study, 55% of pts experienced meaningful DRS improvement with nivo vs 37% with eve (P,0.001). Median (95% CI) time to improvement in DRS occurred at 4.7 (3.7–7.5) months with nivo and was not estimable (NE) with eve due to the limited number of pts who experienced improvement. Association of change in QoL at week 8 and OS is shown in the table. Results using the EQ-5D questionnaire of general health status will be presented. Conclusions: In this global study of pts with aRCC, treatment with nivo vs eve resulted in a significantly greater proportion of pts experiencing meaningful QoL improvement, and improvement or no change in QoL at 8 weeks was associated with an OS benefit with nivo vs eve. These results suggest that the assessment of QoL may be helpful for evaluating the extent of OS benefit in clinical practice. Clinical trial information: NCT01668784.

Survival for patients with and without therapy after HD IL-2. Groups (treatment after HD IL-2) Targeted Therapy Targeted Therapy & Immune Checkpoint Blockade Immune Checkpoint Blockade No Targeted Therapy or Immune Checkpoint Blockade

N

mOS (months)

1, 2, 3 Year Survival (%)

190 13 (12 aPD-1, 1 aPD-1+aCTLA-4)

35.5 NR

81,63,50 100,80,80

12 (7 aPD-1, 2 aPDL-1, 3 aPD-1+aCTLA-4) 196

NR NR

90,79,79 76,65,65

FKSI-DRS

N

TT may include inhibitors of VEGF, mTOR, or, EGFR pathways.

Improvement 91 No change 122 Deterioration 89

4550

4551

Poster Session (Board #172), Mon, 1:00 PM-4:30 PM

A randomized phase II study to compare the efficacy of upfront bi-monthly rotations between pazopanib (PAZ) and everolimus (EVE) versus sequential treatment of first-line PAZ and second-line EVE until progression in patients with metastatic clear cell renal cell cancer (ccRCC) (ROPETAR trial). First Author: Geert A. Cirkel, University Medical Center Utrecht, Utrecht, Netherlands Background: Targeted agents have improved survival in ccRCC patients but resistance always develops. Since accumulating evidence suggests that resistance is partially reversible after drug withdrawal we tested the hypothesis that bi-monthly rotations of treatment with PAZ and EVE delays resistance compared with the standard of PAZ followed by EVE at progression. Methods: Phase 2, open-label, randomized (1:1), multicenter study. Eligibility: $ 18 years, treatment naive progressive metastatic ccRCC, ECOG PS # 2. First-line treatment consisted of either a bi-monthly alternating treatment schedule with PAZ 800mg/d and EVE 10mg/d (rotating arm, ROT) or PAZ 800mg/day (control arm, CON). After progression according RECIST 1.1 (PD) patients made a final rotation to either PAZ or EVE monotherapy (ROT) or initiated EVE (CON) as second-line treatment. Primary endpoint was survival until first PD or death (PFS1). Secondary endpoints included time to second PD or death (PFS2) and toxicity.Accrual of 100 patients was planned to detect an increase in 1-year PFS of 50 to 80% (power 90%, a = 0.05, 2-tailed test). Results: Between 2012 and 2014, 101 patients were randomized. Overall Memorial Sloan Kettering Cancer Center risk categories (MSKCCrc): favorable 26%, intermediate 58% and high-risk 15%. Baseline characteristics and risk categories were wellbalanced. One year PFS1 for ROT was 46% (95% CI:34-62) and 31% (95% CI:20-48) for CON (p = 0.13; 2-sided). Median PFS1 for ROT was 7.4mo (95% CI:5.1-19.0) and 8.9mo (95% CI:6.6-11.9) for CON (p = 0.32). Hazard ratio stratified by MSKCCrc (sHR) for PFS1 was 0.81 (95% CI:0.491.32;p = 0.39) in favor of ROT. Median PFS2 for ROT was 23.0mo (95% CI: . 10.9) and 14.5mo (95% CI: . 11.9) for CON. sHR for PFS2 was 1.05 (95% CI:0.59-1.84;p = 0.88). Mucositis (35% vs 8%), anorexia (39% vs 20%) and dizziness (14% vs 0%) were more prevalent in ROT during firstline treatment. Conclusions: Bi-monthly rotating treatment with PAZ and EVE did not lead to a delay in drug resistance, prolonged PFS or less toxicity. Clinical trial information: NCT01408004.

Nivo median OS (95% CI)

N

28.1 (23.6–NE) NE (27.4–NE) 26.7 (17.3–NE)

47 95 128

Eve Hazard ratio (95% CI) median OS (95% CI) nivo vs eve 19.3 (14.5–NE) NE (19.3–NE) 24.3 (18.7–NE)

0.62 (0.37–1.06) 0.68 (0.44–1.05) 0.99 (0.66–1.48)

Poster Session (Board #173), Mon, 1:00 PM-4:30 PM

Prospective evaluation of circulating cell-free DNA sequencing in patients with metastatic renal cell carcinoma treated with pazopanib plus abexinostat. First Author: Jim Leng, University of California, San Francisco, San Francisco, CA Background: Renal cell carcinoma (RCC) exhibits somatic alterations in genes controlling hypoxia and chromatin states including VHL. Analysis of circulating cell-free DNA (cfDNA) as a means to detect these and other potentially actionable mutations in RCC across histologic subtypes has not been previously fully characterized. Methods: cfDNA was isolated from pretreatment and post-progression plasma samples of sixteen patients with metastatic RCC (N = 12 clear cell; N = 4 papillary) enrolled on a clinical trial of pazopanib and abexinostat (NCT01543763) and analyzed using a cfDNA next-generation sequencing (NGS) panel of 68 genes (Guardant Health). The lower limit of detection was 0.1% mutant allele frequency (MAF) at each nucleotide position. Results: 16 pretreatment and 5 post-progression samples were evaluable. 49 somatic mutations were reported across the entire cohort (mean = 3.06 mutations per patient; range: 0-7). Somatic alterations were detected in 94% (n = 15) of pretreatment samples and 4 of 5 post-progression samples with mean cfDNA MAFs of 0.70% (range = 0.1% - 5.6%) and 1.65% (range 0.10-5.5%) respectively. The most frequently mutated genes were VHL (44% of patients), TP53 (38%), FGFR2 (38%), and KRAS (19%). Evidence of PI3K pathway activation was detected in 25% of pts (N = 2 with activating PIK3CA mutations; N = 2 with PTEN loss-offunction). Activating mutations in MET and IDH1 were identified in two of the four pts with papillary RCC. 2 of 5 pts analyzed to date had novel somatic mutations detected at the time of progression. Conclusions: Analysis of cfDNA in metastatic renal cell carcinoma yields similar frequency of VHL mutations compared to TCGA, as well as potentially actionable mutations across both clear cell and papillary RCC subtypes. PI3K pathway alterations are common and suggest the potential to select patients for combination PI3K/VEGF targeting approach. Detection of novel, actionable mutations using cfDNA at the time of progression is feasible and may provide insights into mechanisms of resistance by non-invasive methods. Longitudinal analysis of the entire study cohort and correlations with radiographic response are ongoing. Clinical trial information: NCT01543763.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Genitourinary (Nonprostate) Cancer 4552

Poster Session (Board #174), Mon, 1:00 PM-4:30 PM

4553

257s Poster Session (Board #175), Mon, 1:00 PM-4:30 PM

Correlation of response with overall survival (OS) for nivolumab vs everolimus in advanced renal cell carcinoma (aRCC): Results from the phase III CheckMate 025 study. First Author: Robert J. Motzer, Memorial Sloan Kettering Cancer Center, New York, NY

Subgroup analyses and updated overall survival from the phase II trial of lenvatinib (LEN), everolimus (EVE), and LEN+EVE in metastatic renal cell carcinoma (mRCC). First Author: Thomas E. Hutson, Texas Oncology, Dallas, TX

Background: In CheckMate 025 (NCT01668784), nivolumab demonstrated superior OS (25.0 mo; 95% CI, 21.8–not estimable [NE]) vs everolimus (19.6 mo; 95% CI, 17.6–23.1) and a higher objective response rate (25% vs 5%, P, 0.001) in previously treated patients with aRCC (N Engl J Med 2015;373:1803–13). Here, we investigated OS benefit with nivolumab vs everolimus by tumor response. Methods: Patients were randomized 1:1 to nivolumab 3 mg/kg intravenously every 2 wk or everolimus 10 mg orally once daily. We conducted a landmark analysis of OS by best response (complete/ partial response; responders), stable disease [SD], and progressive disease [PD]) within 4 mo of randomization. OS was assessed using the Kaplan– Meier method. Results: For the 410 and 411 patients randomized to nivolumab and everolimus, median time to response was 3.5 mo (range: 1.4–24.8) and 3.7 mo (1.5–11.2), respectively. By mo 4 with nivolumab, 19% of patients were responders, 30% had SD, and 40% had PD; with everolimus, 3% of patients were responders, 45% had SD, and 31% had PD. The remainder of patients had discontinued, died, or had an unevaluable response by 4 mo. Median (95% CI) OS is shown (table). For nivolumab, OS rates at 12 and 18 mo were 96% and 89% in responders, 91% and 81% in patients with SD, and 70% and 50% in patients with PD by 4 mo. For everolimus, OS rates at 12 and 18 mo were 93% and 77% in responders, 90% and 79% in patients with SD, and 54% and 35% in patients with PD by 4 mo. Analyses of each subgroup by disease features will be presented. Conclusions: Patients who had a best response of SD within 4 mo of starting treatment had a similar survival trajectory compared with responders. Patients with a best response of progression had an improved OS with nivolumab vs everolimus. Clinical trial information: NCT01668784.

Background: In the randomized phase 2 study of LEN, EVE, and LEN+EVE in mRCC, LEN+EVE significantly improved the primary endpoint median progression-free survival (PFS) vs EVE (14.6 vs 5.5 months [mos]; hazard ratio [HR] 0.40; 95% confidence interval [CI] 0.24–0.68; P= 0.0005). LEN alone (7.4 mos) also improved median PFS vs EVE (HR 0.61; 95% CI 0.38–0.98; P= 0.048). We report PFS outcomes by patient subgroups. Methods: Patients with progressive clear cell mRCC following 1 VEGFtargeted therapy received LEN (24 mg/d), EVE (10 mg/d), or LEN+EVE (18+5 mg/d). Exploratory subgroups included baseline tumor size (sum of tumor diameters), Memorial Sloan-Kettering Cancer Center (MSKCC) risk group, and metastasis site. The data cutoff for the primary analysis was June 13, 2014; updated cutoff (for overall survival [OS] only) was July 31, 2015. Results: Median PFS was improved in LEN+EVE vs EVE in all examined subgroups (Table). In an updated analysis of OS, the median was 25.5 mos (95% CI 16.4–32.1), 19.1 mos (95% CI 13.6–26.2), and 15.4 mos (95% CI 11.8–20.6) for LEN+EVE, LEN, and EVE, respectively. In comparison with EVE, overall survival was longer for LEN+EVE, a trend that was maintained through the cutoff for the updated OS analysis (HR 0.59; 95% CI 0.36–0.96; P= 0.065). Notably, the HR remains , 0.6, and the upper limit of the CI no longer crosses 1. Conclusions: The PFS advantage with LEN +EVE vs EVE was maintained in all examined subgroups. The updated OS analysis demonstrated a trend towards survival benefit with LEN+EVE vs EVE in patients with mRCC following 1 VEGF-targeted therapy. Clinical trial information: NCT 01136733.

Nivolumab, n (%) Median OS (95% CI) 12-mo rate, % (95% CI) 18-mo rate, % (95% CI) Everolimus, n (%) Median OS (95% CI) 12-mo rate, % (95% CI) 18-mo rate, % (95% CI)

4554

Responders

SD

PD

77 (19) NE (24.1–NE) 96 (88–99) 89 (80–95) 14 (3) NE (12.4–NE) 93 (59–99) 77 (45–92)

121 (30) NE (22.7–NE) 91 (84–95) 81 (73–87) 186 (45) 25.0 (22.9–NE) 90 (85–93) 79 (72–84)

164 (40) 14.0 (11.3–16.9) 70 (63–77) 50 (42–57) 127 (31) 9.8 (6.1–12.2) 54 (45–63) 35 (27–43)

Poster Session (Board #176), Mon, 1:00 PM-4:30 PM

Median PFS, mos n

LEN n = 52

n

LEN+EVE n = 51

n

EVE n = 50

MSKCC risk Favorable

11

18.4

12

20.1

12

9.8

Intermediate

18

7.2

19

14.6

19

5.5

Poor

23

5.6

20

5.6

19

3.5

Baseline tumor size , Median

27

7.4

24

14.7

25

5.5

$ Median

25

7.2

27

11.2

25

5.3

Metastasis site Bone Visceral organs Lymph nodes

23 49 32

7.2 7.4 8.0

16 41 29

5.4 9.5 14.7

17 44 31

3.6 5.5 5.5

4555

HR (95% CI); P-value (LEN+EVE vs EVE) 0.25 (0.08–0.76); P= 0.009 0.35 (0.15–0.80); P= 0.0243 0.44 (0.20–0.98); P= 0.0936 0.34 (0.16–0.71); P= 0.0035 0.39 (0.19–0.80); P= 0.0134 0.41 (0.18–0.94) 0.44 (0.26–0.77) 0.28 (0.14–0.58)

Poster Session (Board #177), Mon, 1:00 PM-4:30 PM

Characterizing the outcomes of metastatic papillary renal cell carcinoma (papRCC). First Author: J. Connor Wells, Tom Baker Cancer Centre, Calgary, AB, Canada

High dose Interleukin 2 (HD IL2) as 1st line treatment in metastatic renal cell carcinoma (mRCC): A 10-year single centre experience. First Author: Shien Chow, The Christie NHS Foundation Trust, Manchester, United Kingdom

Background: Metastatic papRCC is the second most prevalent type of renal cell carcinoma, next to clear cell RCC (ccRCC). Outcome on therapy in metastatic papRCC remains poorly characterized in the targeted therapy (TT) era. Methods: Data from 5637 patients in the International Metastatic Renal Cell Carcinoma Database Consortium (IMDC) were retrospectively analyzed. All patients had metastatic disease and received at least one line of TT. Clinical outcomes were compared between papRCC and ccRCC. Results: 466/5637 (8%) of mRCC patients had papillary histology. Overall survival (OS) in papRCC patients was significantly lower than ccRCC patients, 13.8 months vs 21.9 months (p , 0.0001). Progression free survival (PFS) was also lower in papRCC than ccRCC, 4.7 months vs 7.3 months (p , 0.0001). IMDC prognostic groups were able to stratify patients with metastatic papRCC into favorable, intermediate and poor risk groups (34.1 months, 17.0 months, 6.0 months, respectively (p , 0.0001)). Sunitinib was the most common first line agent (62.9%); no differences in OS were observed between various TTs including mTOR inhibitors. Response rate (RR) to TT was 10.2%, lower than ccRCC (30.5%). 49.5% received second-line therapy. Conclusions: To our knowledge, this is the largest analysis of metastatic papRCC patients. Patients with mRCC with papillary histology have poorer OS, PFS, and RR compared to those with clear cell RCC. PapRCC patients can be stratified using IMDC prognostic factors and this data can be used as benchmarks for planning future clinical trials.

Background: Targeted therapy has become the treatment of choice for majority of mRCC cases in recently. For most, treatment benefit is short-lived and given with palliative intent. HD IL2 is an effective treatment capable of durable remission but its unselected use was difficult due to modest response rate and considerable toxicity. Using a set pathology criteria as selection tool, we have been able to show improved outcomes in our previous report (Shablak et al 2011) . We present a 10-yr experience in our centre. Methods: Prospective pathology selection began in 2006. Tumour specimens were reviewed and stratified into Favourable or Other categories based on constitution of histology growth patterns. HD IL2 was preferentially offered to Favourable category. Outcome evaluation included response rates, survival, and treatment tolerance. Multivariate analysis were performed to explore potential prognostic and predictive factors. Results: Total patients reviewed were 145 (2003 – 2013). Among prospective cohort with Favourable pathology n = 106, overall response rate was 48.1% with CR rate of 21.6%. Median OS was 49.4 months. Factors linked to significantly better response and/or survival include . 50% alveolar growth pattern, higher cycle 1 tolerance ( . 18 doses), , 3 metastatic organ sites , and . 6/12 duration between diagnosis and start of treatment. CAIX (Carbonic anhydrase 9) is prognostic but not predictive of response. Toxicities were those expected of IL2 but was manageable on general medical wards, with no treatmentrelated death. Importantly most CRs were durable with 76% (23/30) remained relapse-free (median 39 months follow up). Conclusions: Our experience shows that HD IL2 remains an effective and safe treatment in well-selected cases of mRCC. Results from this large prospective series confirms similar outcomes to our previously reported retrospective series. Given the prospect of long-term remission, fit patients with Favourable histology and low disease burden should be considered for HD IL2 in experienced centre. Better understanding has been gained from this in-depth analysis especially that of possible predictors and strategies likely improve treatment outcome.

Outcome response to TT for ccRCC and papRCC. Overall Survival (months; 95% CI) PFS (months; 95% CI) IMDC Prognostic Score (months; 95% CI) Favourable Intermediate Poor RR to 1st Line TT Complete Response Partial Response Stable Disease Progressive Disease

ccRCC (n = 5016)

papRCC (n = 466)

21.85 (20.86-22.93) 7.33 (6.92-7.72) 41.9 (38.0-44.8) (n = 780) 24.0 (22.8-25.1) (n = 2173) 7.1 (6.5-8.0) (n = 984)

13.76 (12.45-16.1) 4.73 (4.14-5.22) 34.1 (18.6-49.1) (n = 43) 17.0 (13.4-18.7)(n = 200) 6.0 (4.1-7.9) (n = 116)

110/4389 (2.5%) 1234/4389 (28%) 1950/4389 (44%) 1095/4389 (25%)

n (%) 3/391 (0.77%) 37/391 (9.46%) 217/391 (55.5%) 134/391 (34.3%)

p , 0.0001 p , 0.0001 p , 0.0001

p , 0.0001

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

258s

Genitourinary (Nonprostate) Cancer

4556

Poster Session (Board #178), Mon, 1:00 PM-4:30 PM

Medullary renal cell carcinoma (RCC): Genomics and treatment outcomes. First Author: Maria Isabel Carlo, MSKCC, New York, NY Background: Medullary RCC is a rare and aggressive type of kidney cancer that primarily affects young adults with sickle cell trait. Recent reports have identified loss of expression of SMARCB1, a chromatin remodeling gene, as a molecular hallmark of this cancer. There is no defined standard treatment for this entity, although most patients receive platinum-based therapy based on small retrospective series (JCO 29; 2011: Abstract 386). Methods: We performed a retrospective analysis on 36 patients given a pathologic diagnosis of medullary RCC at MSKCC from 1995 to 2015. The goal was to characterize the genomics and report on clinical outcomes with systemic therapy. Results: Median age was 28 years (range 12-72), most patients were male (75%), African American (70%; n = 20/29 with race reported) and presented with metastatic disease (83%); 100% (n = 27/27 tested) had sickle cell trait. All patients with available tissue (n = 23) lacked SMARCB1 expression by IHC. Analysis with in-situ hybridization for SMARCB1 (n = 8) revealed loss of heterozygosity with concurrent translocation in 6, and biallelic loss in 2 (Table). Next-generation targeted sequencing (NGS) showed no VHL mutations. Overall survival (OS) in 32 evaluable patients was 5.8 months (95% CI 4.1-10.9). 11 patients received platinum-based therapy; 4 had a partial response, 2 had stable disease, 4 had progressive disease, 1 was non-evaluable; their median PFS was 2.9 months (95% CI 2.0-Not Reached). 3 patients received VEGF or mTOR inhibitors; none had a response. Conclusions: Outcome was mostly poor in this cohort of patients with medullary RCC. We confirmed uniform loss of SMARCB1 as the key molecular feature in this tumor; novel agents targeting chromatin remodeling, such as EZH2 inhibitors, should be explored in this disease. Patient

SMARCB1 Expression (IHC)

1

No

2

No

3 4

No No

5

No

6

No

7

No

8

No

SMARCB1 FISH Analysis

Targeted NGS

Hemizygous loss / translocation Hemizygous loss / translocation Homozygous loss Hemizygous loss / translocation Homozygous loss

N/A

Hemizygous loss / translocation Hemizygous loss / translocation Hemizygous loss / translocation

4558

N/A N/A AKT3; CDKN2B; ERBB2; GPS2; MLL; MTOR; POLE; SPEN; TET1 MTOR; FH; STK40; MLL; POLE; IDH; SETD2; RASA1; AR; ATRX NF1; NSD1 PDGFRB ERG

Poster Session (Board #180), Mon, 1:00 PM-4:30 PM

4557

Poster Session (Board #179), Mon, 1:00 PM-4:30 PM

Outcomes based on prior VEGFR TKI and PD-1 checkpoint inhibitor therapy in METEOR, a randomized phase 3 trial of cabozantinib (C) vs everolimus (E) in advanced renal cell carcinoma (RCC). First Author: Thomas Powles, Barts Cancer Institute, Queen Mary University of London, London, United Kingdom Background: Optimizing the sequence of systemic therapy to improve outcomes in patients (pts) with advanced RCC remains a clinical challenge. C, a tyrosine kinase inhibitor (TKI), was recently evaluated in pts with RCC and $ 1 prior VEGFR TKI in the Phase 3 METEOR trial (NCT01865747). The trial met its primary endpoint, significantly improving progression-free survival (PFS) compared with E (median [m] 7.4 vs 3.8 mo, HR = 0.58, 95% CI 0.45–0.75; P , 0.001; Choueiri NEJM 2015). Here clinical outcomes by prior sunitinib (S) and pazopanib (P) VEGFR TKI therapy and by prior antiPD-1/PD-L1 therapy are evaluated. Methods: 658 pts were randomized (stratified by MSKCC risk groups and number of prior VEGFR TKIs) 1:1 to receive C (60 mg qd) or E (10 mg qd). Clinical outcome measures included PFS (primary endpoint), ORR, overall survival (OS), and safety. Results: Demographics and baseline characteristics of pts who received only S (n = 267) or only P (n = 171) as prior VEGFR TKI therapy were similar and were also balanced for C vs E. Efficacy outcomes by prior therapy were consistent with the primary PFS analysis. The number of prior VEGFR TKI therapies did not affect efficacy outcomes for C vs E. In pts with S as the only prior TKI, mPFS was 9.1 mo for C vs 3.7 mo for E (HR 0.43, 95% CI 0.32–0.59), while for pts with only prior P, mPFS was 7.4 mo for C vs 5.1 mo for E (HR 0.67, 94% CI 0.45–0.99). The ORR by independent radiology review with C was 16% for pts with only prior S and 19% for only prior P compared with 3% and 3.6% with E, respectively. In pts with prior anti-PD1/PD-L1 therapy (n = 32; primarily nivolumab), the HR was 0.22 (95% CI 0.07–0.65) in favor of C. The mPFS for C could not be estimated, and was 4.1 mo for E; ORR was 22% for C and 0 % for E. The most common AEs reported for pts with only prior S, only prior P, and prior anti-PD-1/PD-L1 therapy were consistent with the overall safety profile in this population. Conclusions: Efficacy outcomes by prior VEGFR TKI therapy and prior antiPD-1/PD-L1 therapy consistently favor C, as was seen in the primary analysis. C should be considered a new treatment option in previously treated pts with RCC regardless of prior systemic therapy. Clinical trial information: NCT01865747.

4559

Poster Session (Board #181), Mon, 1:00 PM-4:30 PM

Efficacy of cabozantinib (C) vs everolimus (E) in patients (pts) with advanced renal cell carcinoma (RCC) and bone metastases (mets) from the phase III METEOR study. First Author: Bernard J. Escudier, Institut Gustave-Roussy, Villejuif, France

A prospective multicenter phase II study of sorafenib combined with cisplatin plus gemcitabine for the treatment of patients with advanced renal collecting duct carcinoma (NCT01762150). First Author: Xinan Sheng, Peking University Cancer Hospital & Institute, Beijing, China

Background: In pts with metastatic RCC, bone mets are prognostic for poor outcomes and represent an unmet medical need. The Phase 3 METEOR trial (NCT01865747) assessed the efficacy and safety of C vs E in pts with RCC and $1 prior VEGFR TKI. The trial met its primary endpoint of improving progression-free survival (PFS; HR 0.58, 95% CI 0.45–0.75; P,0.001) (Choueiri NEJM 2015). As C has shown clinical activity in bone mets in prostate cancer pts (Smith JCO 33 2015 abs 139), we evaluated clinical outcomes in RCC pts with bone mets in METEOR. Methods: 658 pts were randomized (stratified by MSKCC risk group and number of prior VEGFR TKIs) 1:1 to receive C (60 mg qd) or E (10 mg qd). Clinical outcome measures included PFS, ORR, overall survival (OS), and safety. Exploratory endpoints included bone scan response (BSR) in pts with bone scan lesions at baseline (Brown Nucl Med Commun 2012 33:384), incidence of skeletal-related events (SRE), and changes in bone turnover markers. Results: 142 pts had bone mets at baseline; of these, 112 had visceral mets also. MSKCC risk groups in pts with bone mets were consistent with the overall study population.PFS HRs of C vs E were 0.33 (bone; 95% CI 0.21—0.51) and 0.26 (bone+visceral; 95% CI 0.16—0.43). Median PFS with C was 7.4 mo in pts with bone mets and 5.6 mo in pts with bone and visceral mets compared to 2.7 mo and 1.9 mo with E respectively. The ORR with C was 17% and 20%, respectively. BSR by IRC was 18% (C) vs 10% (E). 12% (C) and 14% (E) of randomized pts had at least one SRE, including 4 (C) and 8 (E) cases of spinal cord compression. For pts with an SRE pre-randomization, the incidence of post-randomization SREs was 16% (C) and 34% (E) and included 0 (C) and 5 (E) cases of spinal cord compression. Reductions in bone markers were greater with C vs E (Table). The most common AEs in pts with bone mets were consistent with the safety profile in the overall study population. Conclusions: Treatment with C was associated with improved PFS and ORR in pts with or without bone mets as compared to E. This clinical benefit was supported by the outcomes of bone metastasis-related endpoints. Clinical trial information: NCT01865747.

Background: Collecting duct carcinoma (CDC) of the kidney is a rare and aggressive neoplasm without any approved treatment. The anti-VEGF drugs can produce definite efficacy on renal clear cell carcinoma, but limited for advanced CDC. Gemcitabine plus cisplatin (GC) has showed some activity in patients with advanced CDC. This study was designed to evaluate the efficacy and safety of sorafenib combined with GC as first line treatment in patients with advanced CDC. Methods: This study is a prospective multicenter, single-arm phase II trial. Patients with metastatic, or unresectable CDC by central pathological confirmation and $ 1 measurable lesion per RECIST criteria, ECOG PS # 2, and adequate organ function are enrolled. All patients received sorafenib (400mg Bid) combined with GC (cisplatin: 25mg/m2, d1-3; gemcitabine: 1000 mg/m2, d1,d8, for 4 cycles) until disease progression, unacceptable toxicity, or study discontinuation for any other reason. The primary endpoint is PFS. Overall survival, disease control rate, and safety will also be assessed. Results: Between Jun 2012 and Dec 2015, 26 patients with metastatic CDC were enrolled at 5 Chinese centers after central pathological confirmation. In these patients, 25 patients received prior nephrectomy. The sites of metastases included lung, mediastinal lymph node, local recurrence, retroperitoneal lymph node, skeleton and peritoneum. There were 0 complete and 8 partial responses for an objective response rate of 30.8% (95% CI: 13.1 to 48.5%). Fourteen patients (53.8%) were noted to have stable disease. The disease control rate was 83.8%. Median PFS and OS was 8.3 (95% CI: 2.9 to 13.7) and 11.5 months (95% CI 10.5 to 12.5), respectively. The treatment was well tolerated and toxicities were most grade 1/2. Grade 3/4 toxicities included neutropenia (5/26), thrombocytopenia (3/26), hand-foot syndrome (9/26). Conclusions: This is the largest study about advanced CDC and is the first prospective, multicenter, phase II study showing that sorafenib combined with GC regimen is active and safe as first line therapy in patients with advanced CDC. Clinical trial information: NCT01762150.

C

P1NP CTx

E

Baseline*

Week 5*

Baseline*

Week 5*

48.5 0.40

38.9 0.19

49.9 0.43

44.1 0.43

*Median, ug/L.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Genitourinary (Nonprostate) Cancer 4560

Poster Session (Board #182), Mon, 1:00 PM-4:30 PM

Immunomodulation by HDAC inhibition: Results from a phase II study with entinostat and high-dose interleukin 2 in metastatic renal cell carcinoma patients (CTEP#7870). First Author: Roberto Pili, Roswell Park Cancer Institute, Buffalo, NY Background: Immunosuppressive factors such as regulatory T cells (Tregs) and myeloid-derived suppressive cells (MDSCs) limit the efficacy of immunotherapies. Histone deacetylase (HDAC) inhibitors have been shown to have immunomodulatory effects. We have previously reported that the class I selective HDAC inhibitor entinostat has synergistic antitumor effects in combination with high dose interleukin-2 (IL-2) in a renal cell carcinoma model by down-regulating Foxp3 expression and function of Tregs. Thus, we have conducted a Phase I/II clinical study with entinostat and high dose IL-2 in patients (pts) with metastatic clear cell renal cell carcinoma. Methods: The primary objectives were to evaluate the safety, tolerability and efficacy of this combination strategy. The main eligibility criteria were clear cell histology, no prior treatments, and being fit to receive high dose IL-2. The phase I portion consisted of two dose levels of entinostat (3 and 5 mg, PO every 14 days) and a fixed standard dose of IL-2 (600,000 units/kg every 8 hrs). To test our hypothesis, the fixed sample size at the Phase II dose level is 36 with a type I/II error of 10%. If 11 or more of the pts have a response, the hypothesis that the response rate is # 20% is rejected. Results: Dose levels 1 and 2 were completed without DLTs and 5 mg was the Phase II recommended dose for entinostat. The most common transient grade 3/4 toxicities were hypophosphatemia (16%). lymphopenia (15%), and hypocalcemia (7%) We have enrolled 47 pts (44 at dose level 2), and 38 have completed one cycle (84 days) treatment. Four pts were not evaluable. 13 pts have achieved objective response (34%; 10 PR, 3 CR). The overall median PFS is 16.1 months (6.2, 27.8) and the median PFS of responders is 28.5 months (12,NR). To date the median OS is 65.3 months (52.6, 65.3). Decreased Tregs have been observed following treatment and lower numbers have been associated with response. Conclusions: The results from this clinical trial suggest that entinostat may increase the therapeutic effect of high dose IL-2 by modulating immunosuppressive cells. Clinical trial information: NCT01038778.

4562

Poster Session (Board #184), Mon, 1:00 PM-4:30 PM

Association of high cost sharing and target therapy initiation among Medicare patients with metastatic renal cell carcinoma. First Author: Pengxiang Li, University of Pennsylvania, Philadelphia, PA Background: Targeted therapies (TTs) extend survival in mRCC, but high outof-pocket (OOP) costs may limit access. This is particularly true for oral TTs covered under Medicare Part D wherein patients without low-income subsidies (non-LIS) typically face 25%-33% coinsurance (initial coverage phase), followed by ~50% coinsurance (coverage gap phase), and then 5% coinsurance (catastrophic phase). On the other hand, LIS patients face nominal cost sharing of # $5 under Part D. In contrast, infusible TTs are covered by Medicare Part B and available at low OOP costs to both groups given supplemental insurance. Methods: We used 2011-2013 100% Medicare claims to examine oral TT (sorafenib, sunitinib, everolimus, pazopanib, or axitinib) initiation rates among fee-for-service non-LIS RCC patients newly diagnosed with metastases in the liver, lung, or bone, as compared to their LIS counterparts. We also compared initiation rates for infusible TTs (temsirolimus or bevacizumab) across the two groups. Logistic regressions controlling for patient (e.g. age, gender, race, metastatic site, Charlson score, nephrectomy) and plan characteristics (e.g. Part D plan type and formulary coverage, prior authorization (PA) for oral TTs) were used to model initiation rates. Results: The final sample included 1554 mRCC patients (mean age 72 years, 60% male, 85% White, mean Charlson score 1.1). On average, 86% of oral TTs covered by the patient’s plan formulary were subject to PA. Non-LIS patients (n = 1080) were less likely than LIS patients (n = 474) to have an oral TT claim within 6 months of first metastases claim (25% vs. 37%; adjusted odds ratio [adj-OR] = 0.52, p , 0.001). Initiation rates of infusible TTs were similar across both groups (11% vs. 12%; adj-OR = 1.11, p = 0.687). Non-LIS patients were less likely to initiate any (oral or infusible) TT within 6 months of metastases diagnosis (33% vs. 45%; adj-OR = 0.45, p , 0.001). Extensive subgroup and sensitivity analyses showed consistent findings. Conclusions: High cost sharing was associated with reduced and/or delayed access to oral TTs under Medicare Part D. Lower initiation of oral TTs was not offset by an increase in use of infusible TTs.

4561

259s Poster Session (Board #183), Mon, 1:00 PM-4:30 PM

Prognosis of brain metastasis (BM) in metastatic renal cell carcinoma (mRCC): Experience from Gustave Roussy (IGR). First Author: Annalisa Guida, Univeristy of Modena and Reggio Emilia, Modena, Italy Background: BM occur in 2-16% of patients (pts) with mRCC and are associated with a poor outcome. However, BM appear at different stage of the disease, and various therapies can be proposed. The aim of this study was to find out potential prognostic factors in this heterogeneous group of pts. Methods: Metastatic RCC pts with radiological evidence of BM, followed at IGR from 1997 to 2015 were analyzed. Pts characteristics, treatments and outcome data were collected. Tumor burden (TB) at the time of BM was defined according to RECIST 1.1. OS was calculated from date of BM diagnosis to death for any cause or last follow-up using the Kaplan-Maier method and modelled with the Cox-regression model. Results: 162 pts were included in this retrospective study. Most pts (95%) had clear-cell histology. Half pts (54%) had synchronous metastatic disease. In 36% of cases BM occurred before the start of 1-line therapy, in 37% before the start of 2-line. The most common metastatic sites at the time of BM diagnosis were: lung 79%, lymph nodes 54% and bones 34%. Using International Metastatic Renal Cancer Database Consortium (IMDC) prognostic score, pts were classified as good 24%, intermediate 54% and poor 22% risk when BM was diagnosed. Two third of pts (66%) were diagnosed of a single BM, while 14% had $ 3. Neurological symptoms (NS) were present in 61% of cases. Local treatments (LT) were: surgery 28%, stereotaxic radiotherapy (SRT) 30%, whole-brain radiotherapy (WBRT) 24%, while 25% did not receive local treatment. With a median follow up of 44 months (mo), median OS for pts with BM was 12 mo. In multivariate analysis (table), NS, TB, time to BM (TTBM), number of line of therapy, and the use of a local therapy appear significantly correlated with OS. Conclusions: This is the largest series of BM from mRCC. In this series, characteristics of the disease appear to be more important than the type of local treatment of BM in terms of prognosis. Covariate TTBM yr ,1 1-5 .5 TB mm . 100 Line 0 1 $2 NS Yes LT No Surgery SRT WBRT

4563

HR (95CI)

p

1 0.45 (0.26-0.78) 0.36 (0.19-0.71)

0.004 0.003

2.32 (1.44-3.76)

, 0.001

1 2.05 (1.18-3.56) 2.64 (1.34-5.20)

, 0.01 , 0.005

1.8 (1.10-2.96)

0.018

1 0.09 (0.04-0.20) 0.17 (0.09-0.32) 0.36 (0.20-0.65)

, 0.001 , 0.001 0.001

Poster Session (Board #185), Mon, 1:00 PM-4:30 PM

A randomized phase II trial of interleukin-2/interferon-a plus bevacizumab versus interleukin-2/interferon-a in metastatic renal cell carcinoma (mRCC): Results from the Danish Renal Cancer Group (DARENCA) study 1. First Author: Frede Donskov, Department of Oncology, Aarhus University Hospital, Aarhus, Denmark Background: The potential benefit of combining the antiangiogenic agent bevacizumab (BEV) with Interleukin-2/Interferon-alpha (IL2/IFN) immunotherapy is unknown. This randomized, open-label, phase 2 study compared IL2/IFN/BEV with IL2/IFN in patients with clear-cell metastatic renalcell carcinoma (mRCC). Methods: We randomly assigned 118 patients (1:1) after stratification by MSKCC risk group to receive IFN 3 MIU sc once daily and IL2 2.4 MIU/m2 sc twice daily, 5 days per week, weeks 1 and 2 every 28day-cycle, for a maximum of 9 months; or supplemented with BEV 10 mg/kg, every two weeks iv until progression, unacceptable toxicity, withdrawal of consent or a maximum of 1 year following no evidence of disease (NED). The primary end point was progression-free survival. Secondary end points included objective response rate and overall survival. Results: Baseline characteristics were well balanced between the IL2/IFN/BEV (N = 59) and IL2/IFN (N = 59) arms; metastasis-free interval , 1 year, (75% vs 76%); prior nephrectomy, (85% vs 86%); MSKCC favourable/intermediate risk group, (51%/49% vs 52%/48%); three or more disease sites, (41% vs 44%), respectively. The median progression-free survival was 8.0 mo (95% CI, 4.2 to11.9) with IL2/IFN/BEV and 8.1 mo (95% CI, 5.1 to 11.0) with IL2/IFN, p = 0.47. There was no difference in secondary endpoints between IL2/IFN/ BEV and IL2/IFN; median time to treatment failure, (8.3 mo vs 5.6 mo, p = 0.18), response rate (44.1% vs 28.8%, p = 0.13), surgery of residual disease (17.0% vs 17.0%, p = 1.0), patients achieving NED (8.5% vs 8.5%, p = 1.0), and median overall survival (27.8 mo vs 35.1 mo, p = 0.19), respectively. TKI following progression was well balanced between arms (80% vs 73%). No new or unexpected toxicity was observed. Most common grade 3/4 adverse events for IL2/IFN/BEV and IL2/IFN were fatigue (64.4% vs 61.1%), flu-like symptoms (37.3 vs 41%) and thrombosis (6.8% vs 18.6%, p = 0.01), respectively. Conclusions: The addition of bevacizumab to Interleukin-2/Interferon did not add efficacy in mRCC. No new or unexpected safety findings were identified. Clinical trial information: ClinicalTrials.gov, NCT01274273.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

260s 4564

Genitourinary (Nonprostate) Cancer Poster Session (Board #186), Mon, 1:00 PM-4:30 PM

Differentiation of low grade from high grade clear cell renal cell carcinoma neoplasms using a CAD algorithm on four-phase CT. First Author: Heidi Coy, UCLA Department of Radiological Sciences, Los Angeles, CA Background: Up to 70% of renal masses are found incidentally on imaging. Pretreatment percutaneous biopsy is commonly used to diagnose tumor grade. A non-invasive method to assess grade on imaging would be of clinical value, especially in characterization of clear cell renal cell carcinoma (ccRCC), the most common subtype with the highest metastatic potential. In addition, patients with different Fuhrman tumor grades have a different prognosis and therapeutic pathway. The purpose of our study was to differentiate low grade from high grade ccRCC tumors using a CAD algorithm on four-phase CT. Methods: With IRB approval for this HIPAA-compliant retrospective study, our pathology and imaging databases were queried to obtain a cohort of histologically proven ccRCC with preoperative CT imaged with a four-phase renal mass protocol. A whole lesion 3D contour was obtained in all phases with proprietary software. The CAD algorithm determined a circular ROI with a 0.5cm diameter region of peak lesion enhancement # 300HU within the 3D lesion contour. All contours were confirmed by a genitourinary radiologist. A model was derived using logistical regression with CAD derived peak lesion enhancement from all four phases, %wash-in and %wash-out of tumor enhancement, and tumor size as input. Discrimination was evaluated using receiver operator characteristic (ROC) curves. Results: 112 patients (64% men, 36% women) with 118 unique ccRCCs were analyzed. In the low grade cohort, there were 83 (70%) lesions (Fuhrman grade 1 = 17, Fuhrman grade II = 66), and 35 (30%) lesions in the high grade cohort (Fuhrman grade III = 27, Fuhrman grade IV = 8). Mean low grade lesion size was 2.85 cm (range = 0.8-11.1) and mean high grade lesion size was 6.35 cm (range 1.0-14.7). The model demonstrated discriminatory power in predicting low grade from high grade ccRCC lesions with an area under the curve of 0.765 (0.667-0.893 95% CI). Conclusions: Differentiation of Fuhrman grades on imaging can aid the clinician in stratifying patients to the appropriate therapeutic pathway, and assist in determining which patients can undergo watchful waiting, surgical intervention, or which may be candidates for clinical trials.

4566

Poster Session (Board #188), Mon, 1:00 PM-4:30 PM

Molecular profiling of renal medullary carcinoma to reveal frequent alterations in chromatin remodeling genes and to identify EZH2 as a relevant therapeutic target. First Author: Jianjun Gao, The University of Texas MD Anderson Cancer Center, Houston, TX Background: Renal medullary carcinoma (RMC) is a rare and highly aggressive malignancy that affects primarily young adults with sickle cell trait. The prognosis of RMC is poor with a median survival of about 12 months from diagnosis. It has been previously shown that loss of heterozygosity of SMARCB1 occurs in the majority of cases leading to protein loss. However, a comprehensive molecular profiling of this disease remains unknown. Methods: Whole-exome sequencing (WES) of 9 RMC matched tumor-normal cases and 1 unmatched case were performed. In addition, single nucleotide polymorphism (SNP) analysis was done using the Affymetrix SNP 6.0 array. To test differentially expressed genes between RMC cases and normal kidney, RNA sequencing was carried out. The effect of EZH2 inhibition on in vitro RMC tumor cell proliferation was assessed using the MTT assay. Results: WES analysis of 9 paired RMC cases identified 233 nonsynonymous somatic mutations (median: 21; range: 12-44). We observed frequent mutations in chromatin remodeling genes in 7 of 10 RMC cases, including mutations in SMARCB1 (n = 2) and SMARCA5 gene (n = 1). SNP analysis demonstrated that only the case with SMARCA5 somatic mutation harbored SMARCB1 monoallelic loss. However, all RMC cases had loss of SMARCB1 protein by immunohistochemistry. Ingenuity pathway analysis using RNA sequencing data identified EZH2 as one of the top over-expressed upstream regulators (Z-score = 3.7; p-value = 1.6x10-6). RT-PCR and western blot analyses confirmed that EZH2 expression is elevated at both mRNA and protein levels corresponding to SMARCB1 loss in RMC tumors. In vitro assessment of a novel RMC tumor cell line derived from the case harboring SMARCB1 deletion showed that EZH2 inhibitor GSK343 inhibited proliferation of this cell line. Conclusions: Our data reveal that 30% of RMC cases showed inactivation of SMARCB1 through somatic mutations or monoallelic loss, suggesting that other additional epigenetic mechanisms may lead to gene inactivation. In addition, our data suggest that EZH2 may be a relevant therapeutic target in RMC.

4565

Poster Session (Board #187), Mon, 1:00 PM-4:30 PM

Anlotinib versus sunitinib as first line treatment for metastatic renal cell carcinoma (mRCC): Preliminary results from a randomized phase II clinical trial. First Author: Ai-Ping Zhou, Cancer Hospital, Chinese Academy of Medical Sciences, Beijing, China Background: Anlotinib is a potent oral, small molecular TKI with a favorable safety profile in phase I clinical trial which mainly targets VEGFR1/2/3, FGFR2. This multicenter randomized phase II trial (NCT02072031) compared efficacy and safety of anlotinib and sunitinib as first line treatment in mRCC patients with clear cell being the predominant component. Methods: Patients were randomized in 2: 1 to receive anlotinib or sunitinib. Anlotinib was administered at dose of 12 mg once daily for 2 weeks followed by 1 week rest, sunitinib at dose of 50mg once daily for 4 weeks followed by 2 weeks off. The primary endpoint was progression free survival (PFS). Secondary end points included overall survival (OS), overall response rate (ORR), safety. Results: 133（93 with anlotinib, 40 with sunitinib） patients were enrolled. Data cutoff was May 15, 2015. Both group had similar PFS (11.3 vs. 11.0 months，p = 0.30), ORR (24.4% vs. 23.3%), 6-week disease controll rate (97.8% vs. 93.0%, p = 0.33). OS for both group has not yet reached. Patients treated with anlotinib, as compared with those treated with sunitinib, had a significant less incidence of grade 3 or 4 side effects (28.9% vs. 55.8%, p = 0.0039), especially less grade 3 or 4 thrombocytopenia (0 vs. 11.6%, p = 0.003), neutropenia (0.0 vs. 9.3%, p = 0.009). As for all grades of side effects, anlotinib had less hand-foot syndrome (41.1% vs. 65.1%, p , 0.015), eyelid edema (2.2% vs. 25.4%, p , 0.0001), hair depigmentation (0.0% vs. 14.0%, p = 0.0009), skin yellowing (0.0 vs. 37.2%, p , 0.0001), neutropenia (3.3% vs. 44.2%, p , 0.0001), thrombocytopenia (11.1% vs. 58.1%, p , 0.0001), anemia (4.4% vs. 34.9%, p , 0.001). Also, anlotinib had a trend of less hypertension (50% vs. 67.4%, p = 0.0647), increase of creatinine (14.4% vs. 27.9%, p = 0.095), hypophosphatemia (13.3％ vs. 27.9%, p = 0.085). Whereas, anlotinib had a trend of higher incidence of hypercholesterolemia (25.6% vs.11.6%, p = 0.073) than sunitinib. Conclusions: Anlotinib was shown to have similar efficacy to sunitinib with favorable safety profile in patients with metastatic RCC. Clinical trial information: NCT02072031.

4567

Poster Session (Board #189), Mon, 1:00 PM-4:30 PM

Third- and fourth-line treatment in patients with RCC: Results of a prospective multicentric non-randomised study (IVOIRE study). First Author: Eric Voog, Clinique Victor Hugo, Le Mans, France Background: Sequential targeted therapy is standard in patients (pts) with clearcell renal carcinoma (RCC). Clinical Practice Guidelines exists only for first and second lines (ESMO recommendations). Exploratory recommendations for next treatments (ttt) have been proposed according to their risk (MSKCC). In line 3, pts without prior everolimus were proposed to receive it. Pts who have failed 2nd line ttt with everolimus were proposed to rechallenge TKI only if they experimented a disease control . 6 months in 1st line. For line 4, TKI or everolimus were rechallenged according to the same previously quoted criteria. Methods: A prospective multicentric study has included pts treated according to these guidelines. The objectives were overall survival and for each line best response (RECIST), disease control time (time from day 1 of treatment to the day before the next ttt) and grade 3/4 toxicity. Results: 270 advanced or metastatic renal cancer pts have been included in 17 French centers between Sept 1st, 2011 and Sept 1st, 2014: 73% men; median age 65.5 years [35-90]; Clear RCC: 86%; grade III-IV Fuhrmann: ¨ 54%. Results are summarized in table. Poor risk pts are not presented here. 39% and 21% of pts receiving a line 1 were able to receive a line 3 and 4 and clinical benefit remained high : 54% and 36% respectively for TKI and 51% for everolimus (L3). OS was 59 months [50-73]. More data will be shown at the meeting. Conclusions: For pts who responded . 6 months to TKI, rechallenging the same ttt in a subsequent line could provide high clinical benefit with acceptable toxicity. TTT

Criteria

TKI

n Best response

Everolimus

Others ttt

Median DCT Months [Q1-Q3] n Best response

Median DCT Months [Q1-Q3] n

L1 n = 270

L2 n = 195

L3 n = 106

L4 n = 58

248 CR+RP : 46% SD : 30 % PD : 16 % NA : 8 % 12.9 [6.4-23.1]

87 CR+RP : 28 % SD : 38 % PD : 20 % NA : 14 % 7.1 [3.4-13.4]

66 CR+RP : 15 % SD: 39 % PD: 35 % NA : 11 % 6.2 [3.2-12.1]

33 RP : 12 % SD: 24 % PD: 30 % NA : 34 % 3.7 [1.4-9.4]

5 SD : 4 pts PD : 1 pt

28 RP : 4 % SD: 46 % PD: 36 % NA : 14 % 5.3 [2.0-8.1]

8 SD: 3 pts PD: 2 pts NA : 3 pts

Not done

101 RP : 9 % SD: 44 % PD: 26 % NA : 21 % 6.9 [2.8-12.4]

1.6 [0.8 - 4.2]

17

7

12

17

BR: Best Response; Not Assessable (NA) = ongoing treatment/imaging not done; DCT: Disease Control Time ; Q1/3 : 1st /3rd Quartile.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Genitourinary (Nonprostate) Cancer 4568

Poster Session (Board #190), Mon, 1:00 PM-4:30 PM

Phase 1 study of CB-839, a small molecule inhibitor of glutaminase (GLS), alone and in combination with everolimus (E) in patients (pts) with renal cell cancer (RCC). First Author: Funda Meric-Bernstam, Department of Investigational Cancer Therapeutics (Phase 1 Program), Sheikh Khalifa Bin Zayed Al Nahyan Institute for Personalized Cancer Therapy, Department of Surgical Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX Background: CB-839 is a selective inhibitor of GLS, a key enzyme in the utilization of glutamine by many cancer cells. CB-839 has broad activity in preclinical models, including RCC where GLS is highly expressed. In RCC cell lines, CB-839 combines synergistically with E in vitro and in vivo, strongly inhibiting cell growth as well as glucose and glutamine metabolism. Previously reported data from this study have demonstrated that CB-839 achieved robust GLS inhibition in blood and tumors at well tolerated doses and that BID dosing with meals provided optimal PK and tolerability. Methods: CX-839-001 is an ongoing Phase 1 study evaluating the safety, PK, PD and efficacy of CB-839 in advanced/metastatic cancer patients. After completing an all-comers dose escalation of monotherapy CB-839, an RCC expansion cohort (n = 11) was opened to evaluate the efficacy of monotherapy CB-839 and was further expanded after achieving a preplanned overall response rate. In addition, the combination of CB-839 with standard E (10 mg po QD; CBE) is also being studied in RCC patients, with dose escalation ongoing (3+3 design; 400 mg BID starting dose) and two expansion cohorts for clear cell and papillary RCC planned (n = 9 each). Results: Monotherapy: Safety data from 75 pts who received 600-1000 mg BID revealed a low rate of CB-839related Gr3/4 AEs (4/75 pts). Nineteen heavily pretreated (median of 3 prior therapies) RCC pts (10 clear cell, 6 papillary, 3 other) have been enrolled on the BID dosing regimen. One pt with RCC has achieved a confirmed Partial Response (PR) that remains ongoing after 8.3 months on therapy. Radiographic Stable Disease (SD) or PR was observed in 9 of 15 (60%) efficacy-evaluable RCC pts. Duration of clinical benefit in pts with SD or better ranged between 2.1 and 8.3+ mo (median of 4.6 mo). CBE combination: Eight pts have initiated treatment with escalating doses of CB-839 (400 – 600 mg) in combination with E to date. Conclusions: Monotherapy CB-839 has been well tolerated and demonstrated evidence of efficacy in a subset of patients, including an ongoing confirmed PR, which has resulted in further expansion of this cohort. The combination of CB-839 with E is being studied. Clinical trial information: NCT02071862.

4570

Poster Session (Board #192), Mon, 1:00 PM-4:30 PM

4569

261s Poster Session (Board #191), Mon, 1:00 PM-4:30 PM

Contribution of interleukin-22, a T cell secreted cytokine, to renal cell carcinoma progression and association with poor outcome in RCC patients. First Author: Marc Johannes Weidenbusch, Medizinische Klinik und Poliklinik IV, Klinikum der Universitat ¨ Munchen, ¨ Munich, Germany Background: Immunotherapy holds promise in RCC therapy. However, tumor-induced immune mechanisms add to resistance in RCC and little is known about how immunotherapy affects this. Interleukin-22 (IL-22) directly stimulates tumor cells, hence we examined the effect of IL-22 and its receptor (IL22R) on both RCC overall survival (OS) and progression free survival (PFS) and characterized the biological effect of IL-22 on RCC cells. Methods: RNA-seq data on 413 RCC cases from TCGA were analyzed by cbioportal. Intratumoral IL-22 and IL22R expression-stratified OS and PFS were calculated with Kaplan-Meier. Results were validated in an independent cohort of 40 RCC patients. FFPE-tissues were stained for IL-22 and IL22R. An array of eleven RCC cell lines was screened for IL22R expression and their biological behavior was characterized. Results: IL22R expression in RCC patients inversely correlates with median OS (46.12 vs. 85.45 mo; p , 0,01) and PFS (72.9 vs.91.33 mo; p , 0.05), and higher IL22 expression is also associated with reduced median OS (47.04 vs. 85.45 mo; p = 0.0107mo). In the validation cohort, the association was confirmed for IL22R and IL-22 on the protein level. RCC cell lines have heterogeneous IL22R expression and IL22R+ cells increase their invasiveness, proliferation, hypoxic survival and resistance to RCC therapy in the presence of IL-22 in a dose-dependent manner. Conclusions: We show an effect of IL-22 on RCC outcomes in two independent patient cohorts. Both intratumoral IL22 and IL22R are associated with worse prognosis and these findings are linked to RCC biology in vitro. We suggest IL22R as a prognostic marker and for the first time provide evidence for potential negative effects of immunotherapy by inducing T cell-derived IL-22 in RCC.

4571

Poster Session (Board #193), Mon, 1:00 PM-4:30 PM

Outcomes of metastatic chromophobe renal cell carcinoma (chrRCC) in the targeted therapy era: Results from the International Metastatic Renal Cell Cancer Database Consortium. First Author: Steven Yip, University of Calgary, Calgary, AB, Canada

Phase II study of pazopanib plus interferon alfa as first-line therapy of advanced renal cell carcinoma: A Spanish Oncology Genitourinary Group (SOGUG) study. First Author: Xavier Garcia del Muro, Institut Catala` d’Oncologia, l Hospitalet de Llobregat, Barcelona, Spain

Background: Treatment outcomes are poorly characterized in patients with metastatic chromophobe renal cell cancer (chrRCC), a subtype of renal cell carcinoma. Methods: A retrospective data analysis was performed using the IMDC dataset to determine metastatic chrRCC treatment outcomes in the targeted therapy era. All 5621 patients received one or more lines of targeted therapy for their metastatic renal cell cancer (mRCC). Kaplan-Meier and summary statistics were used. Overall survival rates and overall response rates were calculated. Results: Out of all 5621 IMDC database patients, 114 (2%) patients were identified to have metastatic chrRCC. Baseline characteristics and risk factors are outlined in table 1. Metastatic chrRCC patients with IMDC favorable (20%), intermediate (57%) and poor risk (23%) had median overall survivals of 31.4, 27.3 and 6.5 months, respectively. Patients with metastatic chrRCC had a similar overall survival compared to patients with clear cell mRCC (20.8 months (95% CI 19.9 - 21.7) vs 21.8 months (95% CI 20.9 - 22.9), respectively). The overall response rate of targeted therapy was 20% (4% CR, 16% PR, 49% SD, 31% PD). 52/114 (46%) metastatic chrRCC received second line systemic therapy. Conclusions: Outcomes in metastatic chrRCC and clear cell mRCC are similar when treated with conventional targeted therapies. To the authors’ knowledge this is the largest series of metastatic chrRCC patients and these results set new benchmarks for clinical trial design and patient counseling.

Background: Combination of a VEGF receptor inhibitor with immunotherapy could be a strategy to enhance antitumor efficacy in advanced RCC. A phase I trial by our group showed that pazopanib plus interferon alfa (IFN) is feasible and safe (J Clin Oncol 32, 2014; suppl 4 abstr 450). The purpose of this study is to assess the activity and toxicity of this combination in patients (pts) with advanced renal cell carcinoma (RCC). Methods: Untreatedpts with clear cell RCC, ECOG PS 0-1 and adequate hematological, cardiac, renal and hepatic function, were included in this phase II study, receiving pazopanib at 800mg daily plus IFN 3 million U subcutaneously 3 times a week. Primary endpoint was response rate (RECIST v.1.1). A Simon’s two-stage sequential design was used (P0 = 20%, P1 = 40%, a = 0.05, b = 0.20), requiring at least 10 responses for the study to be considered positive. Results: Thirty three pts were enrolled at 10 centers. Median age was 59 (37-75). PS: 021,1-12 pts. 25 were males. Heng prognostic group: favorable-12, intermediate- 21 pts. 90% had undergone prior nephrectomy. The median number of cycles administered per patient was 7.4, although at the time of the analysis 18 pts remained on therapy. With a median follow-up of 9.3 months, 9 (27%) pts achieved partial response and 20 (61%) had stable disease. Median progression-free survival (PFS) was 14.8 months (95% CI, 4.35-25.2) and median overall survival was not reached. Most common grade 3-4 toxicities were hypertension (18%), asthenia (12%), mucositis (12%), hepatotoxicity (9%) and diarrhea (6%). Other toxicities of varying grades included fever, arthralgia, headache, hand-foot syndrome, emesis and dysgeusia. Conclusions: The combination ofpazopanib and IFN is safe and provides promising PFS. These results support the further exploration of alternative combinations of pazopanib with new immunotherapeutic agents, which are more active and better tolerated than IFN. Clinical trial information: NCT01513187.

Baseline characteristics & first-line therapy in metastatic chrRCC. Baseline Characteristics & First-Line Therapy Gender Anemia Thrombocytosis Neutrophilia Karnofsky Performance Status , 80 Less than 1 year from time of diagnosis to first-line therapy First-line therapy

Metastatic chrRCC Patient Group (N= 114) 56% Male, 44% Female 51/104 15/105 14/104 19/106 48/114 Sunitinib 68/113 Sorafenib 16/113 Bevacizumab 1/113 Temsirolimus 11/113 Pazopanib 12/113 Everolimus 5/113

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

262s 4572

Genitourinary (Nonprostate) Cancer Poster Session (Board #194), Mon, 1:00 PM-4:30 PM

4573

Poster Session (Board #195), Mon, 1:00 PM-4:30 PM

Transcriptomic profiling of collecting duct carcinoma to reveal metabolic and immune aberrations. First Author: Gabriel G. Malouf, Pitie-Salpetriere Hospital, Paris, France

Comprehensive genomic profiling to identify clinically relevant genomic alterations in patients with advanced penile cancers. First Author: Andre Poisl Fay, PUCRS School of Medicine, Porto Alegre, Brazil

Background: Collecting duct carcinoma (CDC) is a kidney cancer subtype that is thought to arise from principal cells in distal parts of the collecting ducts. Some studies suggest an overlap of CDC with upper tract urothelial carcinoma (UTUC), making the pathological diagnosis challenging. Furthermore, the transcriptomic profile of CDC is not fully known. Methods: CDC samples (n = 15) collected from several French institutions were centrally reviewed by a national expert pathologist. As a control, muscle-invasive UTUC samples (n = 10) were also analyzed. RNA sequencing of CDC was performed and compared to that of urothelial (UTUC and bladder) and renal cell carcinoma (RCC) subtypes (clear-cell, papillary, translocation and chromophobe). Tumor infiltrating lymphocytes (TIL) were also analyzed using CD4, CD8 and CD3 staining. Results: CDC displays a unique transcriptomic signature among kidney cancer subtypes, with a putative cell of origin in the distal convoluted tubules. Hierarchical unsupervised clustering reveals that the CDC signature is closer to that of other RCC subtypes than to UTUC, which is similar to that of bladder carcinoma. CDC is characterized by a metabolic shift, with impairment of oxidoreductase activity, pyruvate metabolism and the tricarboxlyic acid cycle, as well as an immunogenic response. We found that the overall median TIL percentage in CDC assessed using CD3 staining was 22% (range: 0%-50%), with a higher statistically significant percentage in metastatic versus non-metastatic tumors (p = 0.04). Similarly, the median CD8 TIL percentage was 11% (range: 0%25%), with a trend toward a higher percentage in metastatic versus nonmetastatic tumors (p = 0.08). Genes and pathways differentially altered between CDC and UTUC point to a basal-like phenotype of CDC in contrast to the luminal-like signature of UTUC. Finally, we identified a set of genes wich might be used in clinic to differentiate between CDC and UTUC with good accuracy. Conclusions: CDC is unique among RCC subtypes that display a pathognomonic transcriptomic signature. Furthermore, CDC is both immunogenic and a metabolic disease, indicating that targeting these processes might provide therapeutic options for patients with CDC.

Background: Penile cancer (PC) is a rare disease with incidence highest in the developing world. Advanced disease has poor clinical outcome as effective systemic therapies are lacking. Comprehensive genomic profiling (CGP) was performed to identify clinically relevant genomic alterations (CRGAs) that may help to understand the biology underlying PC and to explore targeted therapy opportunities. Methods: DNA was extracted from 40 microns of FFPE sections from 37 patients with penile cancers. CGP was performed on hybridization-captured, adaptor ligation based libraries to a mean coverage depth of 627X for up to 236 cancer-related genes plus 47 introns from 19 genes frequently rearranged in cancer. CRGAs were defined as genomic alterations (GAs) linked to targeted therapies approved/experimental in mechanism-driven clinical trials. Results: All but a single case were squamous cell carcinoma, and were predominantly from western countries. The median age of the PC patients was 62 years (range 25 - 87 years). HPV-16, HPV-6, and HPV-11 were found in 8 (21.6%), 1 (2.7%), and 1 (2.7%) patients, respectively, with the rest being HPV negative. There were 223 total GA (mean 6 GA/specimen) involving 73 individual genes and 91 CRGA were found (mean 2.5 CRGA/ specimen) involving 29 genes (Table 1). EGFR amplifications occurred in 22% of cases with a median copy number of 19, and were mutually exclusive from PIK3CA alterations (8/9 mutations) in 24% of cases. The CCND1/FGF3/4/19 amplicon was found in another 8.1% of cases and FGFR3 and FBXW7 mutations in another 5.4% of cases each, with these cases lacking the other oncogenic drivers listed above. No significant relationship of GA to HPV status was observed. Conclusions: CGP identified CRGAs in 97.3% of PC cases, with EGFR amplifications and PIK3CAalterations occurring in a mutually exclusive fashion, with both potentially able to benefit from targeted therapy. Studies are in progress to assay cases from endemic countries to assess whether the biology is distinct from cases of western origin, particularly in light of HPV status.

TPS4574

TPS4575

Poster Session (Board #196a), Mon, 1:00 PM-4:30 PM

A phase 3 study of first-line durvalumab (MEDI4736) 6 tremelimumab versus standard of care (SoC) chemotherapy (CT) in patients (pts) with unresectable Stage IV urothelial bladder cancer (UBC): DANUBE. First Author: Thomas Powles, Barts Cancer Institute, Queen Mary University of London, London, United Kingdom Background: Cisplatin-based CT is the standard first-line treatment for metastatic UBC and carboplatin-based regimens are utilized for cisplatinineligible pts. Despite a relatively high response rate, responses are generally short-lived and almost all pts experience disease progression highlighting a major unmet medical need in UBC. Immune checkpoint blockade has shown activity in pts with CT-resistant UBC, and evidence suggests targeting both programmed cell death-1 (PD-1) and cytotoxic T-lymphocyte-associated antigen-4 (CTLA-4) checkpoints provides for non-redundant pathway blockade and synergy. Durvalumab is a selective, high affinity, engineered human IgG1 mAb that blocks programmed cell death ligand-1 (PD-L1) binding to PD-1 and CD80. Tremelimumab is an anti-CTLA-4 mAb of the lgG2 kappa isotype. In a Phase 1/2 study (NCT01693562), single-agent durvalumab showed preliminary evidence of antitumor activity across several tumor types, including UBC. Encouraging clinical activity and manageable tolerability were reported in a Phase 1b study of durvalumab + tremelimumab in NSCLC (NCT02000947). Methods: DANUBE is a randomized, open-label, multicenter, global Phase 3 study (NCT02516241) of durvalumab 6 tremelimumab versus SoC CT in treatment-na¨ıve pts with unresectable and/or metastatic UBC. Pts will be randomized 1:1:1 to receive durvalumab 1500 mg i.v. every 4 wks (q4w) for up to 12 mos; durvalumab 1500 mg i.v. q4w + tremelimumab 75 mg i.v. q4w for 4 doses, followed by durvalumab 1500 mg i.v. q4w for up to 12 mos; or SoC (cisplatin + gemcitabine or carboplatin + gemcitabine) for up to 6 cycles. Pts will be stratified according to cisplatin eligibility, PD-L1 status (PD-L1+: $ 25% tumor membrane or tumor-associated immune cells stained), and visceral metastasis. The primary endpoint is PFS using investigator assessment (RECIST v1.1). Secondary endpoints include OS; proportion of pts alive and progression free at 12 mos, ORR, DOR, and DCR using investigator assessment; time to second progression; HRQOL; PK; immunogenicity; and safety and tolerability. Recruitment is ongoing. Clinical trial information: NCT02516241.

Gene CDKN2A NOTCH1 PIK3CA EGFR CCND1 FBXW7 BRCA2 FGFR3

Short Variants

CNA

Rearrangement

16 9 8 0 0 4 3 3

4 0 1 8 5 0 0 0

1 3 0 0 0 0 0 0

Poster Session (Board #196b), Mon, 1:00 PM-4:30 PM

A phase 2 study of JNJ-42756493, a pan-FGFR tyrosine kinase inhibitor, in patients (pts) with metastatic or unresectable urothelial cancer (UC) harboring FGFR gene alterations. First Author: Arlene O. Siefker-Radtke, The University of Texas MD Anderson Cancer Center, Houston, TX Background: Current treatment options for metastatic UC remain limited, and tolerability is a concern in many pts. Alterations in the fibroblast growth factor receptor (FGFR) pathway are implicated in development of nonmuscle–invasive bladder cancer; FGFR3 mutations or FGFR translocations are present in ~10-20% of pts with metastatic UC. Treatment with panFGFR (FGFR1-4) inhibitor JNJ-42756493 resulted in promising antitumor activity in a phase 1 trial in pts with advanced solid tumors, including 3 partial responses among 8 pts with UC (Tabernero J, et al. J Clin Oncol. 2015). Safety was manageable and as anticipated with a potent FGFR inhibitor. This phase 2 open-label study of JNJ-42756493 (NCT02365597) is being conducted to determine efficacy and safety of 2 different dose regimens in pts with metastatic or unresectable UC with specific FGFR translocations or mutations. Methods: Eligible pts must have disease progression following chemotherapy for metastatic or surgically unresectable UC or relapse within 12 mos of the last dose of neoadjuvant/adjuvant chemotherapy. There is no limit on number of prior lines of treatment. Prior immunotherapy (eg, programmed-death 1/death-ligand 1 inhibitor) is allowed. Enrollment of cisplatin-ineligible (chemotherapy na¨ıve) pts is allowed. Pts must have measurable disease (RECIST v1.1); Eastern Cooperative Oncology Group performance status # 2; and adequate bone marrow, liver, and kidney function (creatinine clearance $ 40 mL/min). Pts with uncontrolled cardiovascular disease or baseline phosphate persistently above the upper limit of normal are not eligible. Eligible pts are randomized to 1 of 2 oral dose regimens of JNJ-42756493 in a 28-d cycle: an intermittent (10 mg/d for 7 d followed by 7 d off drug) or a continuous (6 mg/d) regimen until a dose regimen is selected. Approximately 110 pts will be treated at the selected dose regimen. Primary end point is objective response rate; secondary end points include progression-free survival, duration of response, overall survival, safety, biomarker, and pharmacokinetic assessments. This trial is currently enrolling at 93 sites in 14 countries. Clinical trial information: NCT02365597.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Genitourinary (Nonprostate) Cancer TPS4576

Poster Session (Board #197a), Mon, 1:00 PM-4:30 PM

TPS4577

263s Poster Session (Board #197b), Mon, 1:00 PM-4:30 PM

KEYNOTE-057: Phase 2 study of pembrolizumab for patients (pts) with Bacillus Calmette Guerin (BCG)-unresponsive, high-risk non-muscleinvasive bladder cancer (NMIBC). First Author: Ashish M. Kamat, The University of Texas MD Anderson Cancer Center, Houston, TX

BISCAY, a phase Ib, biomarker-directed multidrug umbrella study in patients with metastatic bladder cancer. First Author: Thomas Powles, Barts Cancer Institute, Barts Health, and The Royal Free London NHS Foundation Trust, London, United Kingdom

Background: The majority of pts presenting with bladder cancer ( . 70,000 cases/y in the United States) have NMIBC. Despite standard of care therapy with transurethral resection of bladder tumor [TURBT] and BCG administration, a large percentage of pts will have disease recurrence/ progression. PD-L1 is widely expressed in urothelial tumors. Pembrolizumab, an anti–PD-1 antibody, blocks the interaction of PD-1 and its ligands, PD-L1 and PD-L2. KEYNOTE-057 (NCT02625961) is a single-arm, open-label, phase 2 study to evaluate pembrolizumab in pts with high-risk NMIBC unresponsive to BCG. Methods: Key inclusion criteria include age $ 18 y; BCG-unresponsive NMIBC (high-grade Ta, T1, and/or carcinoma in situ [CIS] despite adequate BCG treatment); ineligible for/or declines radical cystectomy; ECOG PS 0-2. Pts must have undergone $ 2 cystoscopic procedures with the most recent # 8 wk of study start, including complete TURBT (tissue sample must be available). Pts will receive pembrolizumab 200 mg Q3W for 24 mo or until disease recurrence or progression, unacceptable toxicity, pt withdrawal, or investigator decision. Pts will be placed into cohorts by presence (cohort A) or absence (cohort B) of CIS based on tissue pathology at screening. Tumors will be evaluated using cystoscopy and urine cytology every 12 wk for the first 2 y, every 24 wks for the next 2 y, and every 52 wk thereafter. CT imaging will be used to assess for upper tract and metastatic or nodal disease. At 18 mo, pts with no evidence of disease may discontinue treatment. Pts with low-grade Ta recurrence will be allowed to undergo repeat TURBT and remain on treatment; low-grade recurrence will not be considered treatment failure. AEs will be monitored throughout the study and for 30 days (90 days for SAEs and ECIs) after study end and graded per CTCAE v4.0. The primary end points are complete response for pts with CIS (cohort A) and disease-free survival for those without CIS (cohort B); secondary end points include progression-free survival, overall survival, duration of response, and PD-L1 expression in relation to clinical outcome. Enrollment is ongoing and will continue until ~260 pts are enrolled. Clinical trial information: NCT02625961.

Background: Despite progress with immunotherapy (IO), there remains a significant unmet medical need for patients with metastatic urothelial bladder cancer (UBC). A patient’s molecular tumour profile enables a targeted small molecule (SM) + IO approach to treatment, particularly in PD-L1 -ve patients. Tumour biomarker profiling to enrol for trials assessing a single targeted agent leads to high screen failure rates and disappointment for patients who are biomarker-negative. BISCAYavoids these issues through centralised screening aligned to a portfolio of targeted SM+IO treatments, such that all patients are allocated to a treatment group determined by their tumour biomarker profile. Methods: Patients with metastatic UBC who have failed at least one prior platinum regimen are eligible. The primary objective is safety and tolerability of new combinations of SM and IO durvalumab (D) PD-L1 antibody treatment. Secondary objectives include overall response rate (ORR), and evaluating markers of response to combination treatment. Tumour samples are evaluated centrally utilising next generation sequencing and biomarker results are used to allocate patients to treatment (Table 1). The study will explore whether adding a SM to IO therapy in patients with specific biomarker expression may trigger increased neoantigen release and improve immunosensitisation. Biomarker prevalence and overlap, mechanism of action and preclinical data enabled a treatment allocation algorithm to be developed with the following adjusted prevalences: AZD4547 alone or in combination with D (~11%), D+O (~19%), D+AZD1775 (~46%) and D alone (~22%). 20 patients in each cohort will detect a 20% improvement over D ORR (80% probability). Additional combinations are being considered for inclusion as faster recruiting cohorts complete. Clinical trial information: NCT02546661. Biomarkers and treatment allocation. Treatment Group

Target

Tumour Biomarkers

Prevalence (%)

AZD4547 monotherapy D + AZD4547 D + olaparib (O) D + AZD1775

FGFRi FGFRi PARPi Wee1i

FGFR3 mut/fusions FGFR3 mut/fusions HRR gene panel CDKN2A homozygous loss/inactivation mut RB1 homozygous loss/inactivatuion mut CCNE1 amp MYC, MYCL, MYCN amp

11 11 28 35 24 11 17 22

D monotherapy

TPS4578

Poster Session (Board #198a), Mon, 1:00 PM-4:30 PM

TPS4579

unselected

Poster Session (Board #198b), Mon, 1:00 PM-4:30 PM

HCRN GU14-188: Neoadjuvant pembrolizumab (P) and gemcitabine (G) with or without cisplatin (C) in muscle invasive urothelial cancer (MIUC). First Author: Christopher J. Hoimes, Case Comprehensive Cancer Center at Seidman Cancer Center, Cleveland, OH

A phase II single arm multi-center trial evaluating the efficacy of pembrolizumab in the treatment of patients (pts) with incurable platinum refractory germ cell tumors (GCT). First Author: Nabil Adra, Indiana University Melvin and Bren Simon Cancer Center, Indianapolis, IN

Background: Neoadjuvant chemotherapy (NAC) with C prior to cystectomy prolongs survival and increases the cure rate in MIUC. Most MIUC patients, however, are not cured with NAC or are cisplatin ineligible. Both G and C have favorable immunomodulatory activity with ability to increase antigen presentation, inhibit and reduce myeloid derived suppressor cells, and prevent PD-1 dependent CD4+ T cell tolerization. Single agent PD-1 antagonism with P has been shown to have activity in metastatic urothelial cancer. Hoosier Cancer Research Network (HCRN) GU14-188 is a phase Ib/II trial (NCT02365766) evaluating tolerability and efficacy of the combination of G and P with or without C in cisplatin -eligible and –ineligible MIUC patients. Methods: Up to 12 subjects will be enrolled on the Phase Ib portion of the trial performed in a standard 3+3 design where a recommended phase II dose of P will be found or stopped due to pre-specified dose limiting toxicities (DLT). An additional 70 subjects that are either cisplatin eligible or ineligible will be treated on a phase II portion and treated with GC+P (cisplatin eligible) or GP (cisplatin ineligible). Eligible subjects include those with clinical stage II or III bladder or upper tract disease with urothelial or mixed variant histology. GC is given every 21 days for 4 cycles; G is given on a 28 day cycle for 3 cycles. All patients receive overlapping therapy with P every 3 weeks for a total of five doses, followed by consolidation extirpative surgery. Primary endpoint is the pathologic muscle invasive response rate (PaIR, p # T1) and will be reported for the cisplatin –eligible and –ineligible cohorts. Secondary endpoints include relapse free survival and overall survival. Exploratory immune and histopathologic aims are also being performed. Clinical trial information: NCT02365766.

Background: Initial cisplatin-based combination chemotherapy will cure 80% of pts with metastatic GCT. Pts who relapse after initial chemotherapy can still be cured with second and even third-line regimens. These regimens include high-dose chemotherapy (HDCT) or salvage standard-dose chemotherapy (SDCT). Despite remarkable results with salvage chemotherapy, about 15% of pts with metastatic GCT are currently incurable and have limited further treatment options. Tumors utilize the programmed death receptor 1 (PD-1) pathway to suppress immune control. Pembrolizumab is a potent highly selective IgG4 humanized monoclonal antibody designed to block the interaction between PD-1 and its ligands, PD-L1 and PD-L2. Pembrolizumab has demonstrated activity in various malignancies. Investigating immune therapy with pembrolizumab is a novel approach for salvage therapy in pts with metastatic GCT. Methods: This is a multi-center, single arm, open label phase 2 trial of pembrolizumab 200mg IV Q3weeks as salvage therapy for pts with incurable GCT. Eligible pts are adult males or females with metastatic GCT (seminoma or non-seminoma) who progressed after first line cisplatin-based chemotherapy and after at least 1 salvage regimen (HDCT or SDCT). Pts with treated stable brain metastasis are eligible. Treatment with systemic corticosteroids/immunosuppressants within 7 days of first trial treatment is not permitted. All pts must provide a biopsy sample for PD-L1 expression evaluation and other exploratory analyses including tumor infiltrating lymphocyte and gene expression analysis. Pts are eligible irrespective of PD-L1 expression status. Pembrolizumab will be given for up to 52 weeks or until disease progression or unacceptable toxicity. Primary objective is overall response rate (ORR) using immune related response criteria (irRC) and secondary objectives are ORR using RECIST criteria, assess toxicity and tolerability of pembrolizumab, and duration of disease response. Using a Simon two-stage minimax design, we plan to enroll a total of 20 pts. This study is currently enrolling pts. Clinical trial information: NCT02499952.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

264s TPS4580

Genitourinary (Nonprostate) Cancer Poster Session (Board #199a), Mon, 1:00 PM-4:30 PM

Avelumab (MSB0010718C; anti-PD-L1) in combination with axitinib as first-line treatment for patients with advanced renal cell carcinoma. First Author: James M. G. Larkin, The Royal Marsden NHS Foundation Trust, London, United Kingdom Background: Targeted agents have improved outcomes in advanced renal cell carcinoma (aRCC), but patients (pts) inevitably develop resistance. Programmed death-1 receptor ligand (PD-L1) is a key therapeutic target in the reactivation of the immune response against multiple cancers, including RCC. Avelumab* is a fully human IgG1 anti-PD-L1 antibody showing clinical activity in several tumor types. Axitinib is a receptor tyrosine kinase inhibitor (TKI) indicated for treatment of aRCC after failure of one prior systemic therapy; axitinib also shows clinical activity as a first-line (1L) therapy. In recent trials, anti-PD-1/TKI combinations have shown promising clinical efficacy in aRCC, and preliminary data suggest better tolerability for the combination with axitinib. This open-label, multicenter, dose-finding study (NCT02493751) evaluates the novel combination of avelumab + axitinib as 1L therapy in pts with aRCC. Methods: The primary objective for this global phase Ib study is to assess safety and tolerability of avelumab + axitinib, defined by dose-limiting toxicities, and determine the maximum tolerated dose (MTD). Eligibility criteria include: histologically confirmed aRCC with a clear cell component, archival or fresh tumor biopsy, ECOG PS 0-1, no prior immunotherapy, and no prior systemic therapy for aRCC. Up to 55 pts will receive avelumab (10 mg/kg or 5 mg/kg as a 1h IV infusion Q2W) + axitinib (5 mg or 3 mg orally BID) in dose-finding and dose-expansion parts and treatment will be given until confirmed disease progression, unacceptable toxicity, withdrawal, or study termination. Other objectives include assessment of objective tumor response (RECIST 1.1), disease control, progression-free survival, OS, duration of response, time to response, pharmacokinetics, pharmacodynamics, immunogenicity, tumor tissue biomarkers, and safety (NCI-CTCAE v4.03). Enrollment began in Oct 2015. A phase 3 trial of avelumab + axitinib vs sunitinib monotherapy as 1L treatment of aRCC is planned. *Proposed INN. Clinical trial information: NCT02493751.

TPS4582

Poster Session (Board #200a), Mon, 1:00 PM-4:30 PM

A randomized, open label, multicenter phase 2 study to evaluate the efficacy of sorafenib in patients (pts) with advanced renal cell carcinoma (RCC) after radical resection of metastases: RESORT trial. First Author: Giuseppe Procopio, Fondazione IRCCS Istituto Nazionale dei Tumori, Milano, Italy Background: Complete response with targeted agents in metastatic renal cell carcinoma (mRCC) is rarely achieved. Retrospective data seem to indicate that metastasectomy (Mtx) may improve survival in selected pts. The RESORT study was designed to evaluate whether Mtx followed by sorafenib may provide additional clinical benefit. Methods: Overall 132 mRCC pts undergoing complete Mtx will be randomized 1:1 to receive either sorafenib or best supportive care (BSC) for 52 weeks or until disease recurrence. Patients will be stratified by prognostic factors including: time from nephrectomy to Mtx, site of disease and number of lesions. Key inclusion criteria: maximum 3 metastatic lesions; complete removal of all metastatic lesions; no tumor cells seen microscopically (R0 resection); no more than three months from Mtx. The primary endpoint is recurrence-free survival (RFS), defined as the time from randomization to diagnosis of disease relapse or death. Secondary endpoints are: overall survival (OS) and safety. Exploratory endpoints: evaluation of vascular endothelial growth factors (VEGF) and fibroblast growth factor (FGF) levels. Evaluation of the association between baseline CTCs count and RFS. Sorafenib will be administered at the dose of 400 mg orally once daily for 21 days and then increased to 400 mg twice daily if the patient has not experienced grade 2 skin toxicity or any other grade 3 toxicities. Imaging assessments will be performed every 12 weeks. Statistical plan:Kaplan-Meier methods will be used for survival analyses and the log-rank test to compare sorafenib arm and BSC arm according to the stratification factors. The hazard ratio (HR) will be determined at the 95% confidence interval (CI). The hypothesis is to increase the RFS time from 12 months (BSC arm) to 18 months (sorafenib arm), corresponding to a 50% improvement. With a study power of 80% and a 1-sided error of 0.15, the estimated number of events required is 86. Forthy-four out of 132 pts have been enrolled. Clinical trial information: NCT01444807.

TPS4581

Poster Session (Board #199b), Mon, 1:00 PM-4:30 PM

A phase I/Ib, open-label, dose-finding study to evaluate safety, pharmacodynamics, and efficacy of pembrolizumab (MK-3475) in combination with vorinostat in patients with advanced renal or urothelial cell carcinoma. First Author: Roberto Pili, Indiana University Mel and Brian Simon Cancer Center, Indianapolis, IN Background: Novel strategies to enhance the activity of immune-checkpoint inhibitors are under development. Our group has reported that the HDAC inhibitor entinostat suppresses regulatory T (Treg) cell function, enhances the anti-tumor effects of immunotherapies. A Phase I/II clinical study with entinostat and high dose interleukin 2 in renal cell carcinoma patients is showing some promising results. Preliminary data from our group also suggest that entinostat enhances the anti-tumor activity of PD-1 blockade in a renal cell carcinoma model. The current study will test the following hypotheses: 1) HDAC inhibition by vorinostat will decrease Tregs and enhance the anti-tumor activity of immune-check point inhibition; 2) Vorinostat will re-sensitize refractory patients previously treated with immunecheck point inhibitors. Methods: This is a Phase I study to assess the immunomodulatory activity of the HDAC inhibitor vorinostat in patients with urothelial or renal cell carcinoma receiving the PD-1 inhibitor pembrolizumab. Key eligibility criteria include previously treated urothelial or renal cell carcinoma. Treatment cycles consist of daily oral dosing of vorinostat (Days 1 through 14) and intravenous dosing of pembrolizumab (Day 1, Cycle 1). The two-drug regimen will be given every 3 weeks. The starting dose of vorinostat will be 100 mg orally once per day for Days 1 through 14. The dose of pembrolizumab will be 200 mg IV on Day 1, every 21 days. Patients will be enrolled according to a standard 3 + 3 design and two dose levels of vorinostat will be tested (100 and 200 mg). The primary endpoint is to determine the recommended Phase 2 dose (RP2D) based on the safety and tolerability of this combination. Once the RP2D is identified, two cohorts of patients with anti-PD-1/PD-L1 sensitive disease or anti-PD-1/ PD-L1 resistant disease will be treated. Secondary endpoints include assessing the objective response rate and progression-free survival. Correlative studies include assessment of PD-L1 (2) expression, immune cell subsets, and microRNAs in tumor and blood samples. Clinical trial information: NCT02619253.

TPS4583

Poster Session (Board #200b), Mon, 1:00 PM-4:30 PM

ADAPTeR: A phase II study of anti-PD1 (nivolumab) therapy as pre- and postoperative therapy in metastatic renal cell carcinoma. First Author: Mark Stares, The Royal Marsden NHS Foundation Trust, London, United Kingdom Background: Cytoreductive nephrectomy may be beneficial in patients with metastatic clear cell renal cell carcinoma (mccRCC). However, there are currently no standard pre-operative systemic therapies. The anti-PD1 agent, Nivolumab, has demonstrated survival improvements in patients with mccRCC, but responses are limited to only 25% of those treated. This highlights the need to identify predictive biomarkers of response and resistance to guide treatment decisions. Methods: ADAPTeR is a single center exploratory phase II study to assess the safety of nivolumab in patients with mccRCC undergoing nephrectomy. Exploratory endpoints include correlating changes in biomarkers with clinical outcomes. Key inclusion criteria are: histologically confirmed mccRCC with no prior systemic therapy. 19 patients will receive Nivolumab, 3mg/kg every two weeks, for eight weeks prior to nephrectomy. Nivolumab will then be restarted and continued for as long there is clinical benefit. Peripheral blood and multi-region samples of tumour tissue will be taken at baseline, the time of nephrectomy and disease progression. Next generation sequencing of tumour tissue DNA and RNA will be used to map somatic mutations and predict candidate neo-antigens. Tumour samples will be evaluated by multi-parametric flow cytometry allowing phenotypic characterization of tumour infiltrating lymphocytes (TILs), focusing on the expression of key co-inhibitory and co-stimulatory molecules, functional markers and correlation with clinical outcomes. The relative abundance of immune cell subsets including CD4+FoxP3+/CD4 + FoxP3-/CD8+ T lymphocyte and myeloid compartments will be quantified and correlated with clinical outcomes Multiplex immunohistochemistry and quantitative automated immunofluorescence will be used to map the tumour immune landscape, assessing both distribution and spatial relationships between TILs and myeloid compartments. Quantitative high-throughput sequencing will be used to characterise the intratumoral T cell repertoire and T cell clonality longitudinally, through therapy, with tracking of these clones in the peripheral blood. ADAPTeR is open to recruitment. Clinical trial information: NCT02446860.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Genitourinary (Nonprostate) Cancer TPS4584

Poster Session (Board #201a), Mon, 1:00 PM-4:30 PM

TARIBO trial: Targeted therapy with or without nephrectomy in metastatic renal cell carcinoma (mRCC)—Liquid biopsy for biomarkers discovery. First Author: Elena Verzoni, Fondazione IRCCS Istituto Nazionale Tumori, Milan, Italy Background: Two randomized trials in the cytokine era clearly showed that cytoreductive nephrectomy (CN) increased life expectancy in metastatic renal cell carcinoma (mRCC) patients (pts). The survival benefit of first-line sunitinib and pazopanib has been demonstrated, but the majority of pts enrolled in the pivotal phase III studies had undergone CN. Therefore it is unclear if similar survival benefit could be achieved without CN with these new targeted agents. Methods: A total of 270 mRCC pts will be randomized to receive CN followed by TKIs vs upfront TKIs. Patients will receive pazopanib 800 mg orally daily or sunitinib 50 mg daily, 4 weeks on/ 2 weeks off. The choice of TKI will be done according to investigator’s clinical practice. Primary objective: To compare clinical benefit, as measured by overall survival (OS), provided by CN followed by TKIs vs upfront TKIs in pts with mRCC. Secondary objectives: To compare clinical benefit, as measured by progression-free survival (PFS) and response rate (RR) provided by CN followed by TKIs vs upfront TKIs, Safety. Exploratory study: Evaluation of predictive role of CTCs count and ctDNA at baseline, before and after surgery (in pts undergoing CN), 24 weeks after randomization and at the time of disease progression Key inclusion criteria: Favorable or intermediate MSKCC or Heng prognostic risk group; diagnosis of RCC with a clear-cell component; resectable asymptomatic mRCC with primary in tumor in place; up to 3 different metastatic sites; $ 3 metastatic lesions. Key exclusion criteria: Widespread disease ( . or = 4 metastatic organ sites) Oligometastatic disease suitable of metastasectomy ( , 3 lesions confined at one organ site) Statistical plan: The sample size was calculated in order to compare 5-year OS between subjects randomized to receive CN followed by TKIs and those randomized to receive upfront TKIs. A total of 191 deaths will yield 80% power to detect a hazard ratio of 1.5 of upfront TKIs vs CN followed by TKIs with an overall type 1 error of 0.05 (two-sided log-rank test). Such a HR corresponds to an increase in the 5-year OS, from an anticipated value of 10% for TKIs to 21.5% for CN followed by TKIs. Clinical trial information: NCT02535351.

TPS4585

265s Poster Session (Board #201b), Mon, 1:00 PM-4:30 PM

ECOG 8141: A prospective phase II trial of neoadjuvant systemic chemotherapy followed by extirpative surgery for patients with high grade upper tract urothelial carcinoma. First Author: Jean H. Hoffman-Censits, The Sidney Kimmel Cancer Center at Thomas Jefferson University, Philadelphia, PA Background: Upper tract urothelial cancer (UTUC) accounts for approximately 5% of all urothelial cancers. At presentation, 30% of patients demonstrate invasive and/or locally advanced disease, 30-40% have regional lymph node involvement, and 20% harbor metastatic disease. Fiveyear cancer specific survival (CSS) rates for pT2 and pT3 tumors are 73% and 40%, respectively. For the more common urothelial bladder cancer, cisplatin based neoadjuvant chemotherapy (NAC) is supported by Level 1 evidence with a 5% absolute improvement in overall survival compared to cystectomy alone, and a pathologic complete response rate (pCR) of 38%. There are no large scale prospective data on the use of NAC in UTUC. NCCN guidelines for UTUC recommend consideration of adjuvant chemotherapy in selected patients. Poor renal function after radical nephroureterectomy (RNU) limits cisplatin candidacy, highlighting the potential greater urgency for NAC in this population compared to bladder cancer. A large retrospective study of NAC prior to RNU for UTUC demonstrated a 14% pCR rate. Favorable oncologic outcomes with NAC for UTUC have been reported compared to historic controls. Methods: Sixty patients with high grade UTUC without metastases who are planned for RNU will be enrolled. The presence of high grade disease on biopsy histology or a positive urine cytology with a radiographically visible upper tract tumor is required. Patients with CrCl . 50 will receive accelerated MVAC and those with CrCl . 30 and , 50 will receive carboplatin and gemcitabine, both for 4 cycles. We hypothesize that NAC for high-grade UTUC prior to RNU will be well tolerated, and result in an 18% pCR rate. Distant recurrence-free survival, event-free survival, bladder cancer-free survival, cancer specific survival, and renal functional outcomes will be measured. Blood and pathologic samples will be collected for evaluation of markers of chemotherapy response/resistance. Clinical trial information: NCT02412670.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

266s 5000

Genitourinary (Prostate) Cancer Oral Abstract Session, Mon, 8:00 AM-11:00 AM

The PROMIS study: A paired-cohort, blinded confirmatory study evaluating the accuracy of multi-parametric MRI and TRUS biopsy in men with an elevated PSA. First Author: Hashim Uddin Ahmed, Division of Surgery and Interventional Science, University College London, London, United Kingdom

5001

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

A randomized phase III trial between adjuvant docetaxel and surveillance after radical prostatectomy for high risk prostate cancer: Results of SPCG12. First Author: Goran Ahlgren, Department of Urology, Skane University Hospital, Lund University, Malmo, ¨ Sweden

Background: Multi-parametric magnetic resonance imaging (MP-MRI) as a triage test might allow men to avoid unnecessary transrectal ultrasound-guided (TRUS) prostate biopsy since many men with an elevated PSA do not have clinically significant prostate cancer. Methods: PROMIS (ClinicalTrials.gov/ NCT01292291) is a prospective, paired-cohort confirmatory study representing level Ib evidence. PROMIS tested MP-MRI and TRUS-biopsy with respect to an accurate reference test, 5mm sampling of the prostate with template prostate mapping biopsy (TPM-biopsy). In 11 UK centres, men with an elevated PSA up to 15ng/ml with no prior biopsy underwent MP-MRI followed by both TRUS-biopsy and TPM-biopsy. The conduct and reporting of each test was performed blind to the other test results. Clinically significant cancer was defined as Gleason $4+3 and/or maximum cancer core length $6mm. Results: Between May 2012 and December 2015, 576 men underwent all 3 tests. The reference TPM-biopsy detected clinically significant cancer in 230 (40%). For MP-MRI, sensitivity was 93% [95%CI 88-96], specificity 41% [36-46], positive predictive value 51% [46-56] and negative predictive value 89% [83-94]. TRUS-biopsy was significantly less sensitive than MP-MRI; 48% [95%CI 42-55] (p=,0.0001). TRUS-biopsy also had significant worse negative predictive value than MP-MRI: 74% [69-78] vs. 89% [83-94] (p,0.0001). We considered two scenarios compared to the current standard of TRUS-biopsy for all. First, using MP-MRI as a triage test would, at a minimum, avoid a primary biopsy in 158 (27%) men with 14 (2%) fewer cases of clinically significant cancer detected. Second, if biopsy targeted to MP-MRI findings achieves the sensitivity of TPM-biopsy, triage with MP-MRI would again avoid primary biopsy in 158 (27%) but detect 17 (3%) more cases of clinically significant cancer than the standard TRUS-biopsy for all pathway. Conclusions: TRUS-biopsy performs poorly both in detecting and ruling-out clinically significant prostate cancer. MP-MRI as a triage test can identify one quarter of men who might safely avoid unnecessary biopsy, without impairing the detection of clinically significant cancer. Clinical trial information: NCT01292291.

Background: Docetaxel has proved to prolong survival in advanced castrate resistant prostate cancer (PCa) and we therefore started this trial to evaluate if six courses of docetaxel improves biochemical disease free survival (BDFS) after radical prostatectomy for high risk PCa. Methods: A total of 459 patients were randomised in 2005-2010 in this multinational openlabeled phase III study, to receive either 6 cycles of adjuvant docetaxel 75mg/m2 q 3 weeks (Arm A) or Surveillance (Arm B). Primary end-point was a rising PSA .0.5ng/ml. High risk prostate cancer was defined as pT2 with a positive margin if Gleasonscore (GS) 4+3 or higher, pT3b .GS 3+4, any lymph node positive disease with .GS 3+4, patients were followed for 5 years with PSA every 3 months. The study was powered to show a 15% difference at 5 years follow up. Results: All six cycles were completed by 79.1% of the patients. Salvage radiotherapy before the primary end-point was reached was given to 8% in Arm A and 10% in Arm B. Mean age was 62.2 years, mean baseline PSA 0.156, 83.7% had pT3 disease and 37.5% had GS 8-10 at randomisation. Of the 308 patients that had a lymph node dissection, 55 (17.5%) had metastasis. Median follow up was 56.8 months. The endpoint was reached in 43.2% of patients; 47.9% in Arm A and 38.9% in Arm B. In a Kaplan-Meier analysis there was no significant difference between the BDFS curves (p=0.078), but the curve of Arm A crossed the curve of Arm B at 15 months and it was parallel at a 10% lower level beyond 24 months. There were 6 deaths from prostate cancer in Arm A and only 3 in Arm B. Febrile neutropenia occurred in 18.7% of the patients in Arm A. No deaths were related to treatment. In a Cox multivariate analysis excluding lymph node status, GS (p,0.001), pT-stage (p=0.002) and positive surgical margin (p=0.009) were significant predictors of progression while randomisation arm (p=0.09) did not reach significance. Conclusions: Adjuvant docetaxel without hormonal therapy did not improve BDFS after radical prostatectomy for high risk prostate cancer. Instead, docetaxel as monotherapy seems to generate a more rapid biochemical progression in a subgroup of patients. Further analysis of this subgroup is warranted. Clinical trial information: NCT00376792.

5002

5003

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

Neoadjuvant enzalutamide (ENZA) and abiraterone acetate (AA) plus leuprolide acetate (LHRHa) versus AA+ LHRHa in localized high-risk prostate cancer (LHRPC). First Author: Eleni Efstathiou, Alexandra General Hospital of Athens, University of Athens, Athens, Greece

A randomized trial of a shorter radiation fractionation schedule for the treatment of localized prostate cancer. First Author: Charles N Catton, Department of Radiation Oncology, Princess Margaret Cancer Centre, Toronto, ON, Canada

Background: AA added to medical castration augments cytoreduction in LHRPC though wide range of persistent cancer is observed. To build on these findings we conducted a study examining AA+ LHRHa with or without ENZA. Methods: We performeda neoadjuvant study of 24 wks of AA+ENZA +LHRHa vs AA+LHRHa (randomized 2:1) in pts with LHRPC (clinical stage T1c/T2 with biopsy Gleason $ 8, or $ T2b with Gleason $ 7 and PSA . 10 ng/mL). We primarily studied treatment effect on pathologic stage. Secondary endpoints include: residual tumor measures; tumor volume, cellularity, tumor epithelial volume (TEV: epithelial component [cellularity %] tumor volume), steroid metabolites (LCMS) and key protein/RNA markers(AR signaling [AR-N ARC19, ARV7, CYP17, PSA], glucocorticoid receptor [GR], neuroendocrine [CD56, CgA], Ki67,p53, RB). Sample size provided 80% power for difference in # pT2N0 per chi-square test at significance of 0.1. Results: Sixty-five treated pts underwent prostatectomy No Grade $ 4 adverse event (AE) reported. Main Grade 3 AEs: LFT increase 19 pts (13 AA +ENZA), hypertension 13 AA+ENZA. PSA nadir # 0.1 ng/mL: 40/44 (91%) in ENZA+ AA+LHRHa vs 17/21 (81%). Pathologic downstaging ( # pT2N0) occurred in 13/44 (30%) AA+ENZA+ LHRHa pts vs 11/21 (52 %) AA +LHRHa (p = 0.07) including 2 pCR. TEV and stage were aligned. Despite high rates of PSA # 0.1, wide range of viable tumor was observed: volume (08.64cc), cellularity (0-90%), TEV (0-5.58cc) Metabolomic data in line with previous report. Stage # pT2N0 was associated with lower Ki67 (p 0.005) and increased expression of canonical AR signaling components AR-N, ARC19, CYP17 (p 0.005) but not PSA (p 0.17). ARV7 presence was more frequent in ENZA+AA+LHRH treated tumors (p 0.05). Four CD56/ CgA, 1 Rb- and 1 p53 altered, tumors identified. Conclusions: Findings do not favor adding ENZA to augment AA+LHRHa efficacy in LHRPC. Expression of neuroendocrine markers, p53, loss of RB, are infrequent in persistent cancers. Canonical AR signaling and low Ki67 correlate with lower stage. ARV7 was detected more frequently in the combination arm. Study results emphasize need to consider biologic heterogeneity when applying AR targeting therapies. Clinical trial information: NCT01946165.

Background: Men with localized prostate cancer (PC) are often treated with high dose radiotherapy (RT) over 8-9 weeks. The a-b ratio which describes the dose-response of tumors and normal tissues to fractionated RT is low for PC. Hence, hypofractionation RT may be more efficacious in PC. Objective: To determine whether an 8-week course of escalated dose conformal RT can be compressed safely, and with similar efficacy into a 4-week course in intermediate risk PC. Methods: Men with intermediate risk PC (T1-2 Gleason 6 and PSA 10-20 ng/ml or T2b-c Gleason 6 and PSA , 20 ng/ml or T1-2 Gleason 7 and PSA , 20 ng/ml) were randomized to conventional (CON) RT, 78Gy in 39 fractions over 8 weeks or hypofractionated (HYP) RT, 60Gy in 20 fractions over 4 weeks, without hormone therapy. RT was planned to respect predefined dose constraints to a risk-adapted volume that included prostate +/- base of seminal vesicles. Daily image guidance was mandated and RT plans underwent real-time central review. The primary outcome is biochemical-clinical failure (BCF) defined by any of: PSA failure (nadir+2), hormonal intervention, clinical local or distant failure, or death. The trial was designed to show that the 5-year BCF of the HYP RT regimen is no higher than CON RT by up to 7.5% (hazard ratio [HR] up to 1.32) with 85% power and one-sided a = 5%. Acute and late GU/GI toxicity were assessed using RTOG criteria. Results:Between 2006 and 2011, 1,206 men from 27 sites in Canada, Australia and France were allocated to HYP RT (608) or CON RT (598). Mean age was 71 (48-88) years. Baseline characteristics were similar between arms. Median follow-up is 6.0 years. To date, 164 patients receiving HYP RT experienced a BCF event compared to 173 in the CON RT group. The BCF event rate at 5 years in both arms was 21%. The observed HR is 0.96 with 90% CI, 0.80 to 1.15. Overall 75 patients have died in each group. GU/GI toxicity grade . 3 was comparable in the acute period; however, late toxicity favored the HYPO RT arm: 3.5% vs. 5.4%, diff = -1.9%, 95% CI, - 4.3 to 0.43%. Conclusion: The HYP RT regimenwas not inferior to conventional RT with no increase in acute or late toxicity. Thus, it is a consideration for men with intermediate risk PC. Clinical trial information: NCT00304759.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Genitourinary (Prostate) Cancer 5004

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

Quality of life (QOL) analysis from CHAARTED: Chemohormonal androgen ablation randomized trial in prostate cancer (E3805). First Author: Linda J. Patrick-Miller, The University of Chicago Medical Center, Chicago, IL Background: Androgen deprivation therapy plus docetaxel (ADT+D) for metastatic hormone sensitive prostate cancer (mHSPC) improves overall survival over ADT. However, docetaxel’s adverse event profile that can diminish QOL. Methods: Patients were randomized to ADT+D (6 cycles) or ADT. QOL instruments including Functional Assessment of Cancer TherapyProstate (FACT-P), FACT-Taxane, FACIT-Fatigue, and Brief Pain Inventory (BPI) were collected at baseline, 3, 6, 9 & 12 months (mo) post randomization. Wilcoxon signed rank tests examined change over time. Mixed effect models compared QOL between arms at each time point. Results: 790 patients were randomized to ADT+D (N = 397) or ADT alone (N = 393) and completed QOL assessments (90% baseline; 84% 3 mo; 74% 9 mo; 69% 12 mo). ADT+D patients reported significant decline in FACT-P at 3 mo (p , .001), but did not differ significantly from baseline to 12 mo (p = .38). ADT +D was associated with significantly worse FACT-P scores at 3 mo (p = 0.02), yet significantly better scores at 12 mo (p = .04). ADT+D patients reported significantly worse FACIT-F scores at 3 mo (p , .001). ADT+D and ADT patients both reported significantly worse follow-up FACT-Taxane scores at all time points (P , .001) and significantly worse BPI scores at 9 and 12 mo (p , .05). BPI scores did not differ significantly between arms over time. Conclusions: Although ADT+D is associated with decreased QOL at 3 mo, 12 mo QOL was better for ADT+D than ADT patients. While both ADT+D and ADT patients report some increased symptoms over time, this study suggests that ADT+D not only provides a survival benefit, but also preserves a better QOL for mHSPC longer than ADT alone. Clinical trial information: NCT00309985. Mixed effects model FACT-P total score1. Difference between ADT+D & ADT2 Baseline 3 months 6 months 9 months 12 months

Estimate

SE

p-value

-1.00 -3.09 0.90 0.29 2.85

1.28 1.32 1.34 1.37 1.39

0.43 0.02 0.50 0.84 0.04

5005

267s Oral Abstract Session, Mon, 8:00 AM-11:00 AM

PRINCE: A phase III study comparing intermittent docetaxel therapy versus continuous docetaxel therapy in patients with castration-resistant prostate cancer. First Author: Hannes Cash, Department of Urology, Charite´ Universitatsmedizin ¨ Berlin, Berlin, Germany, Berlin, Germany Background: A randomized phase III study investigating the non-inferiority of an intermittent docetaxel compared to a continuous docetaxel treatment for patients with castration-resistant prostate cancer (CRPC). Methods: 187 patients were randomized to either an intermittent docetaxel treatment or to a continuous docetaxel treatment until discontinuation (docetaxel application in both arms was weekly 35mg/m2 or three-weekly 75mg/m2). The intermittent arm received docetaxel for one study sequence of 12 weeks and then paused until clinical disease progression, defined by one of the following criteria: increased serum PSA . 4ng/ml with an 50% increase over baseline level, radiological or symptomatic progression. The primary endpoint was one-year survival, which was tested for non-inferiority (efficacy margin 12.5 %). Secondary endpoints were overall survival (OS), progression free survival (PFS), median time to treatment failure (TTF) and toxicity. Final recruitment was not reached. Results: Of 156 eligible patients, 78 were allocated to each arm. One-year survival was 72.6% in the continuous arm vs. 75.8% in the intermittent arm and median OS 18.3 months (95% CI: 1521.5) vs. 19.3 months (95% CI: 4.7-6.1) (P= .535). The intermittent treatment met the non-inferiority criteria in one year survival (difference to lower 95% CI: -0.1201), but not for OS, according to the result of a post-hoc analysis. The differences between the study arms in PFS and TTF were not significant. The median treatment holiday in the intermittent arm was 15 weeks (range 1-69), or 38% of the overall treatment duration. Safety profiles of both study arms were comparable. Conclusions: The intermittent docetaxel chemotherapy was non-inferior to a continuous therapy in one-year survival. It was well tolerated and may present a treatment option for patients with CRPC.

1 Adjusted for age (# 59 v 60-69 v $ 70; disease extent (high v low), local therapy (Y/N), ECOG PS (0 v 1/2 ), baseline physical well-being (# 20 v 20 - # 25 v . 25), baseline pain (0/1 v 2/3 v $ 4) 2ADT+D score - ADT score *Sanofi provided study drug and ECOG-ACRIN grant.

5006

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

5007

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

Cabazitaxel vs docetaxel in chemotherapy-naive (CN) patients with metastatic castration-resistant prostate cancer (mCRPC): A three-arm phase III study (FIRSTANA). First Author: A. Oliver Sartor, Tulane University, New Orleans, LA

TAXYNERGY: Randomized trial of early switch from first-line docetaxel (D) to cabazitaxel (C) or vice versa with circulating tumor cell (CTC) biomarkers in patients (pts) with metastatic castration-resistant prostate cancer (mCRPC). First Author: Scott T. Tagawa, Weill Cornell Medicine, New York, NY

Background: The Phase III TROPIC study (NCT00417079) reported significant improvement in overall survival (OS) for cabazitaxel 25 mg/m2IV Q3W plus prednisone 10 mg PO QD (P) vs. mitoxantrone plus P in mCRPC pts previously treated with a docetaxel (D)-containing regimen. The FIRSTANA study examined if cabazitaxel 20 mg/m² (C20) or 25 mg/m² (C25) IV Q3W plus P is superior to docetaxel 75 mg/m² (D75) IV Q3W plus P in terms of OS in CN mCRPC pts. Methods: In this multinational, open label phase III study, mCRPC pts, ECOG PS 0-2, who had progressed after castration were randomized 1:1:1 to C20, C25 or D75 IV Q3W plus P. The primary endpoint was OS. Key secondary endpoints were safety, progression free survival (PFS), tumor PFS, tumor response (RECIST 1.1), PSA response, PSA PFS, pain response, pain PFS, time to skeletal-related events (SRE) and health-related quality of life (HRQOL). Results: Between May 2011 and April 2013, 1168 pts were randomized (C20=391, C25=389, D75=388). Baseline demographics and disease characteristics were similar across cohorts. The median number of treatment cycles was 9 for all dose groups. In the ITT analysis, median OS was 24.5 months for C20, 25.2 months for C25 and 24.3 months for D75. HR for C20 vs. D75 was 1.009 (0.85 to 1.197, p=0.9967) and for C25 vs. D75 was 0.97 (0.819 to 1.16, p=0.7574), indicating that C20 and C25 were not superior to D75 in terms of OS. PFS was 4.4 months for C20, 5.1 months for C25 and 5.3 months for D75 (NS). Tumor responses were superior in C25 (41.6%) compared to D75 (30.9%), p=0.0370. Other secondary endpoints did not significantly differ across dose groups. Adverse events (AEs) grade 3-4 were 41.2% in C20, 60.1% in C25 and 46.0% in D75; pts discontinuing treatment due to an AE were 25.2% in C20, 31.7% in C25 and 33.9% in D75. Febrile neutropenia, diarrhea and hematuria were more frequent in C25; peripheral neuropathy, peripheral edema, alopecia and nail disorders were more frequent in D75. Conclusions: C20 and C25 did not demonstrate superiority for OS compared to D75 in CN mCRPC pts. Among secondary endpoints only tumor responses were significantly superior for C25. AEs were less frequent in C20 for most categories. Clinical trial information: NCT01308567.

Background: TAXYNERGY (NCT01718353), a randomized phase II trial, evaluated early taxane switch with CTC biomarkers to study mechanisms of sensitivity/resistance in men with chemo-na¨ıve mCRPC. Methods: Pts were randomized 2:1 to first-line D or C. Pts not achieving $30% PSA decline by Cycle 4 (C4) switched to the other taxane. Pts remained on treatment until progression, unacceptable toxicity, investigator decision or study cut-off. CTCs were collected at multiple time points. Co-primary endpoints: achievement of confirmed $50% PSA response vs historical control (TAX327) and analysis of drug target engagement in post-treatment CTCs by androgen-receptor (AR) nuclear localization (ARNL). Results: 63 pts were randomized to D (n = 41) or C (n = 22). Median age was 71 yr; ECOG PS was #1 in 93.7%. Baseline characteristics in both arms were similar; 44.4% received prior potent AR-targeted therapy. 33 pts (52.4%; D 48.8%, C 59.1%) had $30% PSA decline by C4 and did not switch taxane. Of 30 pts (47.6%) not achieving $30% PSA decline by C4, 15 (23.8%; D 29.3%, C 13.6%) switched taxane and 15 discontinued. 8/15 (53.3%) pts who switched taxane achieved $50% PSA response. Overall, 35 pts (55.6%) had confirmed $50% PSA response; the pre-defined endpoint of lower limit of the 10% 1-sided CI of 47.5% exceeded the historical control rate of 45.4%. In pts with $50% PSA decline, mean %ARNL in CTCs decreased by 6.5%, while in pts without 50% PSA decline %ARNL increased by 6.1% (P = 0.029). Composite PFS was 9.1 mo (95% CI 4.9, 11.7); median OS was not reached after median follow up of 14 mo. Key toxicities in pts on D or C throughout or who switched, respectively, included diarrhea (48%; 63%; 60%), nausea (30%; 53%; 47%), neutropenia (19%; 32%; 13%) and febrile neutropenia (19%; 16%; 0%). Conclusions: TAXYNERGY, the first prospective trial to report ARNL in CTCs as a correlative taxane biomarker, met its co-primary clinical and biomarker endpoints, with an early switch in taxane resulting in a higher PSA response rate vs TAX327. Consistent with the proposed MOA, ARNL may indicate taxane sensitivity/resistance. Funding: Sanofi Clinical trial information: NCT01718353.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

268s

Genitourinary (Prostate) Cancer

5008

Oral Abstract Session, Mon, 8:00 AM-11:00 AM

Phase III non-inferiority study of cabazitaxel (C) 20 mg/m2 (C20) versus 25 mg/m2 (C25) in patients (pts) with metastatic castration-resistant prostate cancer (mCRPC) previously treated with docetaxel (D). First Author: Johann S. De Bono, The Royal Marsden NHS Foundation Trust and The Institute of Cancer Research, London, United Kingdom Background: The Phase III TROPIC study (NCT00417079) reported a significant improvement in overall survival (OS) for C plus prednisone (P) (25 mg/m2once every 3 weeks plus 10 mg orally once daily) versus mitoxantrone plus P (Hazard Ratio [HR] 0.70; P , 0.0001) in pts with mCRPC previously treated with D. This PROSELICA study (NCT01308580) was designed to determine the relative efficacy and safety profile of C20 plus P compared with C25 plus P. Methods: In this randomized, open-label, multinational phase III study, pts with mCRPC and ECOG performance status 0–2, who progressed after treatment with D, were stratified (ECOG, RECIST, region) and randomized 1:1 to C20 or C25. To show that C20 could preserve $ 50% of the efficacy benefit showed by C25 in TROPIC, the HR of C20 vs C25 for the primary endpoint OS could not exceed 1.214 under 1-sided 98.89% confidence level adjusted after interim analyses. Secondary endpoints included progression free survival (PFS), safety, PSA, pain and tumor responses and quality of life. Results: From April 2011 to December 2013, 1200 pts were randomized (C20 n = 598; C25 n = 602). Patient characteristics were similar for C20 and C25. Median number of C cycles was 6 for C20 and 7 for C25. The median survival of C20 and C25 did not differ significantly and the HR boundaries (99% confidence level) were within the non-inferiority margins assumptions, therefore meeting the study’s non-inferiority endpoint. PSA and RECIST response rates were higher in C25 (see Table). Grade 3–4 adverse events: 39.7% C20; 54.5% C25. Grade 4 laboratory neutropenia: 21.3% C20; 48.6% C25. Neutropenic sepsis/infection: 2.2% C20; 6.1% C25. Conclusions: In pts with mCRPC progressing after treatment with D, C20 demonstrates non-inferiority for OS compared with C25 and an improved overall safety profile. Clinical trial information: NCT01308580. Outcome (HR) Median OS, mos Median PFS, mos PSA response, % ORR, n/N

5010

C20 N = 598

C25 N = 602

C20 vs C25

13.4 (12.19 to 14.88) 2.9 (2.79 to 3.45) 29.5 (25.6 to 33.3) 50/271 18.5%

14.5 (13.47 to 15.28) 3.5 (3.12 to 3.94) 42.9 (38.8 to 47.1) 60/256 23.4%

Upper Limit of 98.89% Confidence Interval (CI): 1.184 HR (95% CI): 1.099 (0.974 to 1.24) P , 0.0001

Clinical Science Symposium, Tue, 9:45 AM-11:15 AM

Co-targeting androgen receptor (AR) and DNA repair: A randomized ETS gene fusion-stratified trial of abiraterone + prednisone (Abi) +/- the PARP1 inhibitor veliparib for metastatic castration-resistant prostate cancer (mCRPC) patients (pts) (NCI9012)—A University of Chicago phase II consortium trial. First Author: Maha Hussain, University of Michigan, Ann Arbor, MI Background: In preclinical CRPC models, PARP1 inhibition synergizes with AR targeted therapy and induces DNA damage, especially in ETS fusionpositive tumors. We hypothesized: co-targeting PARP-1 + AR is superior to AR inhibition and ETS is a response predictor. Methods: Eligible mCRPC pts underwent metastatic (mets) disease biopsy (bx), were stratified by ETS status and randomized to Abi (Arm A) or Abi + Veliparib (Arm B). The primary endpoint was confirmed PSA response rate (RR, $ 50% decline). Secondary endpoints included safety, objective RR (ORR), progression free survival (PFS), and if DNA repair gene deficiency (DRD; homozygous deletions or deleterious mutations: BRCA 1, BRCA 2, ATM, FANCA, PALB2, RAD51B, RAD51C) predicts response. 148 pts stratified by ETS status to be randomized to detect a 20% PSA RR improvement assuming a 5% 1-sided type I error and 80% power. Primary analysis: PSA RR logistic model to test the interaction of ETS status and treatment with a significance threshold of 0.15. Results: 185 eligible pts underwent mets bx (89 soft tissue, 96 bone); 159 (86 %) had adequate tissue, 35% were ETS +ve (IHC). 153 pts (88% white, 8% black, median age 68 years, median PSA 35.4 ng/ml) were randomized: Arms A/B: 74/79. Hyperglycemia was the most common G3/4 AE ($ 5% in both arms). PSA RR: Arm A 63%, Arm B 67%. ORR: Arm A 36%, Arm B 49%. Median PFS: Arm A 8.5 months (m) (95% CI: 8.1 – 13.5), Arm B 11m (95% CI: 8.1 – 13.8). More Arm B pts were on therapy for 12 (34% vs 30%) and 18 + cycles (20% vs 16%). There were no differences by ETS status. Exploratory analysis from sequencing mets tissue (N=66) was performed. 18 pts (27%) had DRD; .80% in both arms had PSA response. Median PFS (95% Cl) by DRD status: no DRD: 5.8 ms (4.2 - 8.2) vs DRD 13.5 ms (8.2 - NR). Conclusions: This first mets biomarker-based trial was feasible. The trend is in favor of Abi + Veliparib with regard to PSA RR, ORR, therapy duration and mPFS. Exploratory analysis suggest that DRD is associated with better response. Updated clinical and sequencing data will be reported. Clinical trial information: NCT01576172.

5009

Clinical Science Symposium, Tue, 9:45 AM-11:15 AM

Inherited mutations in DNA repair genes in men with metastatic castrationresistant prostate cancer. First Author: Peter Nelson, Fred Hutchinson Cancer Research Center, Seattle, WA Background: Inherited mutations in genes that regulate DNA repair processes such as BRCA1 and BRCA2 associate with increased risks of developing lethal prostate cancer (PC), and with response to poly-ADP ribose polymerase (PARP) inhibition and platinum chemotherapy. While the prevalence of such mutations in men with localized PC unselected for family predisposition is insufficient to warrant routine testing, the frequency of germline DNA repair gene (DRG) mutations in men with metastatic prostate cancer (mPC) is not established. Methods: We recruited men with mPC (n = 569) through five IRB-approved studies enrolling across seven institutions designed to ascertain molecular aberrations in mPC. Men were unselected for family history of cancer and were documented to have mPC based on a biopsy of a metastatic site. We isolated germ-line DNA from blood lymphocytes, saliva, or tissue uninvolved with cancer and used multiplex sequencing assays to assess germline mutations in 20 DRGs associated with autosomal dominant cancer predisposition syndromes. Results: Sixty-six deleterious DRG germline mutations were identified in 65 men (11.4%), including BRCA2 (n = 29, 5.1%), ATM (n = 10, 1.8%), CHEK2 (n = 7, 1.2%), BRCA1 (n = 5, 0.9%), PALB2 (n = 3, 0.5%), RAD51D (n = 3, 0.5%), ATR (n = 2, 0.3%) and 1 man each with mutation in FAM175A, GEN1, MRE11A, MSH2, MSH6, RAD51C, and NBN. One patient had mutation in ATM and CHEK2. Overall, the frequency of germline DRG mutations in mPC significantly exceeded the 5% frequency identified in localized PC (n = 499; p , 0.001) which included men with high-risk PC, and 3.1% in the Exome Aggregation Consortium (ExAC) database, which includes individuals with cancer (n = 60,706; p, 0.005). Conclusions: The identification of a germline aberration in a DNA repair pathway gene has implications for treatment selection in view of the potential for therapeutic responses to PARP inhibitors and platinum chemotherapy, and also for identifying families at risk for cancer predisposition. These results suggest that men with mPC, regardless of family history of cancer, should be considered for germline genetic testing of DNA repair genes.

5011

Clinical Science Symposium, Tue, 9:45 AM-11:15 AM

Correlation of DNA damage response (DDR) gene alterations with response to neoadjuvant (neo) dose-dense gemcitabine and cisplatin (ddGC) in urothelial carcinoma (UC). First Author: Gopa Iyer, Memorial Sloan Kettering Cancer Center, New York, NY Background: Individual DDR genetic alterations have been reported to be predictive for neo chemotherapy response in UC. We sought to identify whether alterations in a DDR gene set correlate with response in UC patients (pts) enrolled onto a phase II study of neo ddGC. Methods: This trial evaluated 46 pts treated with 6 cycles of ddGC (Balar et al, GU ASCO 2016). 63% of pts undergoing cystectomy achieved pathological response defined as # pT1. Exon capture sequencing (MSK-IMPACT) of 341 to 410 genes was performed on pretreatment transurethral resection specimens and matched germline DNA on 24/ 46 pts. Fourteen of these 24 pts were responders to ddGC. Only deleterious alterations (truncating or functionally validated) in a pre-defined DDR gene set (Table 1) were considered when comparing responders vs non-responders. Results: The median number of somatic mutations (muts) per sample was 10 (range, 0-36) with an average coverage of 550X. The most frequently altered DDR genes were BRCA2 and ATR (12% each). Eleven samples (46%) harbored at least one DDR gene alteration (4 had 3 alterations, 1 had 2, and 6 had 1). The median number of DDR alterations in responders vs non-responders was similar (1 vs 2); however, 5 of 14 responders had deleterious DDR alterations vs 0 of 10 non-responders (p = 0.053). These included 4 nonsense muts, one each in BRCA2, ATR, ERCC5, and RECQL4, and one ERCC2 N238S point mut. This preponderance of deleterious DDR alterations in responders was similar to that observed in a previously published cohort (13 alterations in responders vs 4 in non-responders, p = 0.016) (Van Allen et al, Cancer Discovery 2014). Conclusions: Lack of deleterious DDR alterations in non-responders in this study is biologically plausible and worthy of further study. Sequencing of the remaining pts on this trial is ongoing to characterize the positive and negative values for this DDR gene set in this patient cohort. Clinical trial information: NCT01589094. DDR gene set. ATM

CHEK2

MLH1

PMS1

RAD51D

ATR BARD1 BLM BRCA1 BRCA2 BRIP1 CHEK1

ERCC2 ERCC3 ERCC4 ERCC5 FANCA FANCC MDC1

MRE11A MSH2 MSH6 MUTYH NBN PALB2 PARP1

PMS2 POLD1 POLE RAD50 RAD51 RAD51B RAD51C

RAD52 RAD54L RECQL4 XRCC2

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Genitourinary (Prostate) Cancer 5012

Poster Discussion Session; Displayed in Poster Session (Board #269), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

5013

269s

Poster Discussion Session; Displayed in Poster Session (Board #270), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

AR-V7 and efficacy of abiraterone (Abi) and enzalutamide (Enza) in castration-resistant prostate cancer (CRPC): Expanded analysis of the Johns Hopkins cohort. First Author: Emmanuel S. Antonarakis, The Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins, Baltimore, MD

AR-V7 and CTC heterogeneity biomarkers additively to predict patient (pt) outcomes with taxanes relative to approved AR targeted therapy. First Author: Howard I. Scher, Sidney Kimmel Center for Prostate and and Urologic Cancers and Memorial Sloan-Kettering Cancer Center, New York, NY

Background: We previously reported an association between ARV7 detection and poor outcomes with Abi/Enza in 62 CRPC pts (NEJM 2014). Here, we expand our analysis to 202 pts, and also report the prognostic significance of CTC detection in addition to ARV7 detection. Methods: We prospectively enrolled CRPC pts starting Abi or Enza, and examined the prognostic value of CTC detection (+ vs -) and ARV7 detection (+ vs -) using a CTC-based ARV7 mRNA assay. We examined $50% PSA responses, PSA progression free survival (PSA-PFS), clinical/radiologic PFS (PFS), and overall survival (OS). We constructed multivariable (MVA) models adjusting for PSA, Gleason sum, number of prior hormone therapies, prior Abi/Enza use, prior taxane use, presence of visceral mets, and ECOG score. We also separately examined the 1st-line (Abi/Enza-na¨ıve) and 2nd-line (post-Abi or Enza) novel hormonal therapy (NHT) settings. Results: 202 pts were enrolled (median f/u 12.9 mo, range 1.3–35.8 mo). CTC+/ARV7+ pts were more likely to have Gleason $8 (p= .05), M1 disease at diagnosis (p= .01), higher PSA (p, .01), prior Abi/ Enza use (p= .03), prior taxane use (p= .02), and ECOG $1 (p= .01). Outcomes for the overall cohort, and separately for the 1st-line and 2nd-line NHT cohorts, are shown in the Table. All correlations remained significant in MVA models. Conclusions: The negative prognostic impact of ARV7 in pts receiving Abi/Enza is confirmed in this expanded analysis, and applies to both the 1st-line and 2nd-line NHT settings. This study is the first to suggest that our CTC-based ARV7 assay should be interpreted using 3 separate prognostic categories (CTC, CTC+/ARV7-, CTC+/ARV7+).

Background: An unmet need in the management of mCRPC is how to use approved AR signaling inhibitors (AR tx) and taxanes (T) to maximize survival for the individual patient. Pts with AR-V7+ CTCs (V7+) and, separately, high phenotypic CTC heterogeneity, demonstrate improved OS with T over AR Tx relative to pts with no V7+ CTCs and low CTC heterogeneity. We studied both markers in concert in relation to clinical outcomes. Methods: 191 blood samples from 161 unique pts were collected prior to initiation of AR Tx (n=128) or T (n=63). CTCs were detected and analyzed on the Epic Sciences platform, identifying 20 discrete phenotypic digital pathology cell features inclusive of AR & CK protein expression to define 15 unique CTC phenotypes present at variable frequencies in the samples studied. Heterogeneity, the diversity in the number of types of cells present, was quantified with a Shannon Index. Using the Epic nuclear V7 immunofluorecent assay, heterogeneity scores and V7 status were associated with outcomes alone and in combination. Results: 63 (33%) pts were identified by either the V7 or High Heterogeneity (HH) biomarkers (20 V7+/HH+; 14 V7+/HH-; & 29 V7-/HH+). The addition of HH increased clinical sensitivity by 85% over V7 alone. V7+/HH+ pts had the worst outcomes with AR Tx and favored taxane tx (80% reduction of risk, p=0.011) after adjusting for line of tx and other pre-tx markers. Compared to biomarker(-), V7+/HH- and V7-/HH+ patient classes also had worse OS with AR Tx than with T. Conclusions: Characterization of AR-V7 status and CTC heterogeneity alone identifies pts with worse outcomes on AR Tx relative to T. Used together, the results may provide significant clinical utility to individual pts by informing the decision to select a tx from which he is most likely to benefit. Prospective validation of these biomarkers is ongoing.

Overall (N = 202)

N (%) PSA response PSA-PFS (mo) PFS (mo) OS (mo)

5014

CTC n = 53 (26.2%)

CTC+/ ARV7 n = 113 (56.0%)

CTC+/ ARV7+ n = 36 (17.8%)

75% 11.3 13.9 28.7

52% 6.2 7.7 29.5

14% 2.1 3.1 11.2

1st-Line NHT (N = 124)

P , , , ,

.001 .001 .001 .001

2nd-Line NHT (N = 78)

CTC n = 36 (29.0%)

CTC+/ ARV7 n = 73 (58.9%)

CTC+/ ARV7+ n = 15 (12.1%)

P

CTC n = 17 (21.8%)

CTC+/ ARV7 n = 40 (51.3%)

CTC+/ ARV7+ n = 21 (26.9%)

P

86% 12.7 21.6 29.7

66% 7.3 10.1 30.7

27% 2.9 4.1 21.5

, .001 , .001 , .001 .003

53% 6.4 6.2 18.8

28% 4.4 5.3 13.0

5% 1.1 2.8 8.5

.003 , .001 , .001 , .001

Poster Discussion Session; Displayed in Poster Session (Board #271), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Association of plasma cell-free DNA concentration [cfDNA] with outcome from taxane therapy (TT) for castration resistant prostate cancer (CRPC). First Author: Niven Mehra, Royal Marsden NHS Foundation Trust and The Institute of Cancer Research, Sutton, United Kingdom Background: Biomarkers are needed to guide CRPC treatment; we evaluated [cfDNA] pre- and post-TT, associating this with outcome. Methods: cfDNA was isolated and quantified from 1 mL of plasma, prospectively collected from CRPC patients (PTS) recruited to two phase III trials, utilizing QIAamp Circulating Nucleic Acid and Quant-IT Picogreen HS DNA kits. FIRSTANA (NCT01308567) compared docetaxel (D; 75mg/m2) with cabazitaxel (C; 20 or 25mg/m2) as 1st line TT; PROSELICA (NCT01308580) compared C at 20 or 25 mg/m2 as 2nd line TT. Results: 1400 samples from 315 FIRSTANA PTS and 1102 samples from 256 PROSELICA PTS were available. PTS with two baseline (BL) samples (taken 1-7 days apart prior to course 1 pre-TT), had a median coefficient of variation (CV) of 23.6%, with higher median [cfDNA] in 2nd line PTS (28.4 vs 16.2 ng/mL, p,0.001) and in PTS with visceral disease (27.5 vs 19.6 ng/mL, P=0.003). Prognostic factors associated with BL [cfDNA] included Hb (Spearman r=0.45, n=570), LDH (r=0.40, n=566), ALP (r=0.37, n=570) and PSA (r=0.34, n=566), all p,0.001. In both studies the association of BL cfDNA with OS and rPFS was stronger than that of C2 and C4 changes. Median OS by BL [cfDNA] quartiles in FIRSTANA was 38.8, 29.9, 21.8 and 15.1 months (m) and in PROSELICA 18.1, 17.8, 12.1 and 9.3m. In FIRSTANA median rPFS by BL [cfDNA] quartiles was 16.8, 11.4, 9.7, and 10.8m, and in PROSELICA 16.4, 10.3, median not reached and 10.1m. The Table reports univariate (UVA) and multivariate (MVA) Cox regression analyses for BL [cfDNA] and outcome. 95% of PTS have sufficient cfDNA (.5ng/mL) for targeted next generation sequencing (NGS). Conclusions: High BL [cfDNA] associates with advancing disease, poor prognosis and shorter rPFS on TT. NGS of these samples is ongoing to further elucidate the clinical utility of cfDNA biomarker studies. Clinical trial information: NCT01308580 and NCT01308567. MVA†

UVA BL cfDNA (log10) FIRSTANA OS rPFS PROSELICA OS rPFS Combined OS rPFS

HR

95%CI

P

HR

95%CI

P

3.3 2.2

2.4-4.4 1.6-3.0

,0.001 ,0.001

2.0 2.0

1.4-3.0 1.3-2.9

0.001 0.001

1.8 1.6

1.5-2.2 1.2-2.1

,0.001 ,0.001

1.5 1.3

1.1-2.2 0.9-2.0

0.023 0.208

3.1 2.4

2.6-3.8 1.9-3.0

,0.001 ,0.001

1.6 1.7

1.2-2.1 1.2-2.3

0.002 0.001

Univariate OS Comparison vs V7-/HHV7+/HH+ (n=20) V7+/HH- (n=14) V7-/HH+ (n=29) Either Biomarker+ (n=63)

5015

HR of AR

HR of T

HR: 22.7, p , 0.0001 (n=8) HR: 12.7, p , 0.0001 (n= 8) HR: 4.7, p = 0.0156 (n=17) HR: 7.7, p , 0.0001 (n = 33)

HR: 3.3, p = 0.0007 (n=12) HR: 4.0, p = 0.005 (n= 6) HR: 1.4, p = 0.54 (n=12) HR: 2.6, p = 0.0142 (n = 30)

Poster Discussion Session; Displayed in Poster Session (Board #272), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

HSD3B1 and resistance to androgen deprivation therapy in prostate cancer. First Author: Jason W.D. Hearn, University of Michigan, Ann Arbor, MI Background: The somatic mutation HSD3B1(1245A . C) has been mechanistically linked to castration-resistant prostate cancer by encoding a mutant enzyme that augments intratumoral dihydrotestosterone (DHT) synthesis. Given the HSD3B1(1245C) allele is also frequently found in the germline, we hypothesized men inheriting this variant allele would exhibit resistance to androgen deprivation therapy (ADT). Methods: We determined HSD3B1 genotype retrospectively in men treated with ADT for postprostatectomy biochemical failure. We analyzed progression-free survival (PFS), distant metastasis-free survival (DMFS), and overall survival (OS) according to HSD3B1 genotype using Kaplan-Meier methods. Demographic and treatment characteristics were compared to assess for potential confounders using Fisher’s exact test and Kruskal-Wallis analysis of variance. Multivariable analyses were performed to assess the independent predictive value of HSD3B1 genotype on outcomes. Results were independently validated in two external cohorts, including a second post-prostatectomy cohort and a metastatic cohort. Results: 443 patients were included: 118 in the primary cohort, 137 in the post-prostatectomy validation cohort, and 188 in the metastatic validation cohort. In the primary cohort, median PFS diminished as a function of the number of variant alleles inherited (6.6 years in homozygous wild-type men, 4.1 years in heterozygotes, and 2.5 years in homozygous variants; P = 0.01). Median DMFS likewise diminished according to the number of variant alleles inherited (9.1 vs. 6.8 vs. 3.6 years, respectively; P = 0.01). Finally, OS decreased with the number of variant alleles inherited (5-year and 10-year OS: 82% and 55% vs. 74% and 35% vs. 58% and 0%; P = 0.006). On multivariable analysis, the impact of HSD3B1 genotype on metastasis (hazard ratio (HR) 2.8; P = 0.023) and death (HR 3.3; P = 0.016) was maintained. Findings in the external cohorts independently validated the profound impact of HSD3B1(1245C) on outcomes, including overall survival. Conclusions: Inheritance of the HSD3B1 (1245C) allele that enhances dihydrotestosterone synthesis predicts innate resistance to ADT in prostate cancer.

† Included ECOG, M1-site, Hb, Neu, Lym, Plt, ALP, Alb, LDH, PSA, BL pain.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

270s 5016

Genitourinary (Prostate) Cancer Poster Discussion Session; Displayed in Poster Session (Board #273), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Fine tuning metabolism of biochemically active abiraterone metabolites to optimize anti-androgen therapy in prostate cancer. First Author: Nima Sharifi, Cleveland Clinic, Cleveland, OH Background: Abiraterone inhibits CYP17A1 and blocks androgen synthesis, prolonging survival in castration-resistant prostate cancer (CRPC). Abiraterone is metabolized in patients to D4A, which has even greater anti-tumor activity and structural similarities to endogenous steroidal 5a-reductase substrates, such as testosterone. Other steroidal abiraterone metabolites might also harbor biochemical activity that could be relevant to clinical response or resistance to abiraterone. Furthermore, clinical manipulation of abiraterone metabolites might be an approach to optimizing anti-androgen therapy. Methods: We synthesized 3 5a-reduced and 3 5b-reduced abiraterone metabolites as standards for mass spectrometry analysis and to interrogate biochemical function. Serum was analyzed by mass spectrometry in 12 patients treated with abiraterone acetate plus prednisone for CRPC and in 16 patients treated on a clinical trial of abiraterone acetate (1000 mg) plus prednisone for 2 cycles, followed by the addition of dutasteride (3.5 mg) for 4 cycles. The activity of selected novel abiraterone metabolites was tested in assays for steroidogenic enzymes, androgen receptor binding studies, gene expression studies and xenograft models. Results: D4A is converted to at least 3 5a-reduced and 3 5b-reduced metabolites. The initial 5a-reduced metabolite, 5a-abi, is more abundant than D4A in patients with prostate cancer taking abiraterone, and is an androgen receptor (AR) agonist, which promotes prostate cancer progression. In a clinical trial of abiraterone alone, followed by abiraterone plus dutasteride, 5a-abi and downstream metabolites are depleted, while D4A concentrations rise, effectively blocking production of a tumor promoting metabolite and permitting D4A accumulation. Furthermore, dutasteride does not deplete three 5b-reduced metabolites, which were also clinically detectable, demonstrating the specific biochemical effects of pharmacologic 5a-reductase inhibition on abiraterone metabolism. Conclusions: Our findings suggest a previously unappreciated and biochemically specific method of clinically fine tuning abiraterone metabolism to optimize therapy.

5017

Poster Discussion Session; Displayed in Poster Session (Board #274), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Randomized phase II study of AKT blockade with ipatasertib (GDC-0068) and abiraterone (Abi) vs. Abi alone in patients with metastatic castrationresistant prostate cancer (mCRPC) after docetaxel chemotherapy (A. MARTIN Study). First Author: Johann S. De Bono, The Royal Marsden NHS Foundation Trust and The Institute of Cancer Research, London, United Kingdom Background: mCRPC commonly has hyperactivated PI3K/Akt signaling, usually via PTEN loss; androgen receptor (AR) blockade activates Akt, supporting mCRPC cell survival. We hypothesized that co-inhibition of Akt with Ipatasertib (Ipat), a potent novel oral ATP-competitive inhibitor of Akt, and of AR by Abi would improve outcome in mCRPC. Methods: Patients (pts) with mCRPC, previously treated with docetaxel, were double-blinded randomized 1:1:1 to three arms: Ipat-400, Ipat-200, or placebo (pbo), all in combination with Abi 1000 mg and prednisone 10 mg, daily (NCT01485861). Pts were stratified by prior enzalutamide (yes or no), number of chemotherapies (one vs more), and progression by PSA only vs other. Primary endpoint was radiographic progression-free survival (rPFS), with key secondary endpoint of overall survival (OS). Exploratory endpoints included assessment of the effect of PI3K/Akt pathway alterations in archival tumor tissue on rPFS. Results: 253 patients (pts) were randomized with 173 rPFS events (68%) at the primary analysis. Compared to pbo, Ipat-400 showed increased rPFS (median 8.2 vs 6.4 m; HR = 0.75; p= 0.17) with consistent trends in subgroup analyses. At 41% OS events, analysis showed an increased OS for Ipat-400 vs pbo (median 18.9m vs 15.6m; HR = 0.72; p= 0.22). Results for the Ipat-200 vs pbo showed less robust HR (rPFS HR = 0.94, p= 0.75; OS HR = 0.96, p= 0.87) - a potential dose effect for this combination. mCRPC with PTEN loss had superior rPFS benefit from Ipat400 over pbo. AEs more common with Ipat were generally consistent with PI3K/Akt pathway inhibitor class and included diarrhea, nausea, vomiting, asthenia, rash, decreased appetite, and hyperglycemia. These AEs were dose-dependent, manageable, and reversible with Abi dose intensity maintained. Conclusions: This study provides evidence that Ipat in combination with Abi may improve rPFS and OS in pts with mCRPC post docetaxel, with potentially increased benefit in patients with decreased PTEN expression. Clinical trial information: NCT01485861.

5018 Poster Discussion Session; Displayed in Poster Session (Board #275), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

5019

Development and validation of genomic signature to predict ADT treatment failure. First Author: Jeffrey Karnes, Mayo Clinic, Rochester, MN

Clinical and genomic characterization of metastatic small cell/ neuroendocrine prostate cancer (SCNC) and intermediate atypical prostate cancer (IAC): Results from the SU2C/PCF/AACRWest Coast Prostate Cancer Dream Team (WCDT). First Author: Eric Jay Small, University of California, San Francisco, San Francisco, CA

Background: Androgen deprivation therapy (ADT) is one of the main treatment options for locally advanced, recurrent and metastatic prostate cancer. Neuroendocrine prostate cancer (NEPC) is inherently less sensitive to ADT. NEPC can be observed de novo (e.g., small cell prostate cancer) but more commonly arises after exposure to ADT. In some patients, even short exposure to ADT may ‘unmask’ a NEPC phenotype. We hypothesized that a gene expression signature of NEPC when measured in primary tumor specimens (RP) may be useful for predicting patients with innate resistance to ADT. Methods: Expression profiles of 786 PCa patients treated with RP were obtained from the Decipher GRID database. These were split into training (n=360) and validation (n=426) sets and stratified by the receipt of adjuvant ADT (n=222) or no adjuvant ADT (n=564). A literature review of ADT resistance and neuroendocrine genes identified 1,632 genes as candidates. This set was further filtered, using logistic regression and forward feature selection to select a genomic ADT resistance signature (ARS). ARS was trained using a generalized linear model with lasso regularization. Survival c-index, Kaplan Meier and Cox regression analysis was used to compare survival differences between treated and untreated patients with high and low ARS scores (defined by median split). Results: In validation cohorts, the ARS was predictive of metastasis in cohorts receiving adjuvant ADT (10-year metastasis free survival c-index of 0.64 (95% CI 0.55-0.71) as compared to 0.50 (95% CI 0.36-0.63) in patients not treated with ADT). Among ADT treated patients, those with low ARS scores had a 10 year MFS of 86%, versus 68% in those with high ARS scores (p=0.03). In multivariable analysis adjusting for confounding variableswith metastasis as an endpoint, ARS validated with a significant interaction with ADT treatment (p=0.02) which demonstrates ARS’s adjuvant ADT predictive capability. Conclusions: An ADT resistance signature validated as a predictive biomarker with a significant interaction term for predicting metastasis after hormonal treatment. ARS may allow for identification of patients that may be optimal candidates for trials of novel systemic agents

Poster Discussion Session; Displayed in Poster Session (Board #276), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Background: SCNC and a novel pathologic subtype, IAC, comprise a growing proportion of mCRPC patients resistant to androgen signaling inhibitors such as Abiraterone (Abi) or Enzalutamide (Enz). We sought to characterize these non-adenocarcinoma (adeno) subtypes in a prospective biopsy study. Methods: Eligible mCRPC pts underwent a metastasis biopsy (bx), and were followed for clinical outcomes. Tissue was both frozen and formalin fixed/ paraffin embedded (FFPE). Independent review of FFPE specimens was undertaken by 3 pathologists. Frozen specimens underwent laser capture micro-dissection prior to RNA sequencing (seq). Machine learning was used to derive histology-specific expression signatures. Signature accuracy was evaluated with leave-pair-out cross-validation and application to an independent data set. Results: 226 of 300 planned mCRPC pts (74% resistant to Abi and/or Enz) have undergone bx (including 123 bone, 61 node, and 23 liver bx) with a 78% evaluable biopsy rate. Adeno was identified in 39%, non-adeno in 41% (SCNC in 12%, IAC in 29%), with other mixtures in the remaining 20%. Median overall survival (OS) from time of biopsy for pts with non-adeno histologies was 12.8 months (mos) [ IAC OS = 19.1 mos; SCNC OS = 12.8 mos] versus 25.8 mos in adeno pts (p=0.023). RNAseq data are available from 94 bx. Transcriptional signatures were developed which accurately distinguish adeno, SCNC and IAC. IAC appears to have a signature intermediate between adeno and SCNC. When applied to an independent data set (Beltran et al, Ca Disc 2011), these transcriptional signatures identified SCNC with 100% accuracy, and predict for IAC differentiation in a subset of non-SCNC tumors. Conclusions: A majority of CRPC metastases exhibit non-adenocarcinoma features, which are associated with a shortened survival. IAC and SCNC are genomically distinct, and newly derived transcriptional signatures based on these bx can be used to identify SCNC in an independent data set. Integration of whole exome with RNA-seq data are ongoing to identify pathways up-regulated in IAC and SCNC.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Genitourinary (Prostate) Cancer 5020

Poster Discussion Session; Displayed in Poster Session (Board #277), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Confirmatory analysis to determine associations between platinumsensitivity, molecular signature of combined tumor suppressor defects and aggressive variant prostate carcinomas (AVPC). First Author: Paul Gettys Corn, The University of Texas MD Anderson Cancer Center, Houston, TX Background: Clinically defined AVPC is associated with platinum-sensitivity and a molecular profile of combined ($2) alterations in PTEN, Tp53 and/or RB1 (Aparicio et al. Clin Cancer Res 2013;19 and Clin Cancer Res 2015;1). Based on these findings we conducted a validation study of cabazitaxel (CAB) plus or minus carboplatin (CARB) in men with CRPC stratified for the presence or absence of AVPC criteria (Corn et al. ASCO 2015; Funding: Sanofi). Methods: Clinical data were updated and available specimens were assembled for analysis. Immunohistochemistry was performed for AR, Ki67, RB1 and PTEN. DNA was extracted from remaining tumor cells and subject to next-generation sequencing to include Tp53. Results: At a median follow up of 21.6 months (mo), median progression free survival (mPFS) is 4.6 mo (95%CI 3.5-5.8) with CAB vs 7.4 mo (95% CI 5.6-8.3) with CAB/CARB (p = 0.004). In men meeting AVPC criteria, mPFS is 3.8 mo (95% CI 2.8-5.7) with CAB vs 5.6 mo (95% CI 4.4-8.0) with CAB/CARB (p = 0.012). After 86 deaths, median overall survival (mOS) is 17.4 mo (95% CI 12.6-25.9) with CAB vs 19.2 (95% CI 17.0-29.3) with CAB/CARB (p=0.489). 53% (42/79) of CAB patients and 23% (19/81) of CAB/CARB patients received additional salvage platinum-based therapy after progression. We identified 124 samples from 92 patients, of which 95 contained sufficient tumor for analysis. Of 54 analyzed to date, 38 belonged to men receiving CAB/CARB (30 AVPC) and 16 to men receiving CAB (10 AVPC). #10% of cells stained for AR, RB1 and PTEN in 6 (15%), 17 (42.5%) and 5 (12.5%) of 40 AVPC tumors and 0 (0%), 5 (35.7%) and 2 (14.3%) of non-AVPC tumors respectively. Tp53 sequencing is ongoing. Conclusions: Adding CARB to CAB is safe and significantly improves mPFS and response rates in men with mCRPC with a trend towards improved mOS. Men meeting AVPC clinical criteria benefit most from the CAB/CARB combination. We observed a high frequency of tumor suppressor defects in the characterized samples. Further analysis of the rate of their combination, association with AVPC criteria and benefit from CARB will be informative. Clinical trial information: NCT01505868.

5022

Poster Discussion Session; Displayed in Poster Session (Board #279), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

5021

271s

Poster Discussion Session; Displayed in Poster Session (Board #278), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Does short-term androgen depletion add to high dose radiotherapy (80 Gy) in localized intermediate risk prostate cancer? Final analysis of GETUG 14 randomized trial (EU-20503/NCT00104741). First Author: Bernard M. Dubray, CRLCC Henri Becquerel, Rouen, France Background: Multi-center randomized trial to evaluate the addition of 4month androgen deprivation to high dose radiotherapy in intermediate risk localized prostate adenocarcinoma patients (pts). Methods: eligible pts were randomly assigned to high dose conformal radiotherapy (prostate 80 Gy / 40 fractions; seminal vesicles 46 Gy / 23 fractions) either alone (group RT) or in combination with 4-month androgen deprivation (flutamide + triptorelin starting 2 months before radiotherapy, group AD-RT). Lymphadenectomy was mandatory when the risk of node involvement was . 10% (Partin). The primary endpoint was survival without clinical / biochemical relapse at 5 years. Secondary endpoints included overall survival, toxicity (CTCAE v3) and quality of life (QLQ-C30, PR-25). The a-priori sample size was 450 patients, 225 per arm (0.90 power to detect an increase from 75 to 85%, bilateral a = 0.05). Results: 377 pts were included between September 2003 and June 2010. The inclusions were prematurely closed, due to slow accrual. Intent-to-treat analysis was made for 370 pts (191 RT, 179 AD-RT). Prognostic factors were well balanced between the two arms. The median follow-up duration was 84 months (range: 3 to 132). At 5 years, the probabilities of survival without clinical / biochemical relapse were 76% [95% CI: 69% – 81%] and 84% [78% – 89%] in RT and AD-RT groups, respectively (p = 0.02). Overall survival probabilities were 94% [90% - 97%] and 93% [88% - 96%] respectively (p = 0.54).Cumulative incidence of biochemical failure were 21% [15% – 26%] and 10% [6% – 15%], respectively (p , 0.01). The probabilities of being free of grade 3-4 toxicities were 96% and 95% (p = 0.69) for digestive tract, 93% and 95% (p = 0.44) for urinary tract. Conclusions: 4 months of androgen blockade improves event-free survival at 5 years in pts with intermediate risk prostate adenocarcinoma when treated with high dose radiotherapy. Longer follow-up is required to demonstrate an impact on overall survival. Clinical trial information: EU-20503 / NCT00104741.

5023

Poster Discussion Session; Displayed in Poster Session (Board #280), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Final results of 2-dose fractionation of 177Lu-J591 for progressive metastatic castration-resistant prostate cancer (mCRPC). First Author: Scott T. Tagawa, Weill Cornell Medicine, New York, NY

Disease-free survival (DFS) as a surrogate for overall survival (OS) in localized prostate cancer (CaP). First Author: Christopher Sweeney, Dana-Farber Cancer Institute, Harvard Cancer Center, Boston, MA

Background: Phase 1 and 2 studies of single dose 177Lu-J591 are published. We hypothesized dose fractionation may improve safety and efficacy. Methods: Patients with progressive mCRPC; no selection for PSMA expression. Initially, 6 cohorts of 3-6 pts got 2 doses of 177Lu-J591 2 wks apart (20 mCi/m2, escalating to 45 mCi/m2 x2). Subsequently, pts enrolled in 2 expansion cohorts at the recommended phase 2 doses (RP2D). Planar 177Lu-J591 imaging was semi-quantitatively scored. Endpoints: PSA changes and survival (OS); CTC count (CellSearch) changes in expansion cohorts. Results: 49 pts, median age 74.1 yrs (range 55-95), PSA 44.9 (1.9-766.5); 83.7% with bone, 61.2% LN, 40.8% visceral mets. 8.2% CALGB (Halabi) low, 34.7% intermediate, 57.1% high risk group. RP2D’s of fractionated 177Lu-J591 were 40 mCi/m2 x2 or 45 mCi/m2 x2 with option for GCSF. PSA changes and OS in Table. Accurate targeting of 177 Lu-J591 was seen in 79.6%. Pts with lower PSMA expression by imaging were less likely to respond (p=0.07). Of 25 with CTC counts, 14 declined, 8 stably favorable, 3 increased. RP2D associated with more PSA declines (p=0.036) and longer OS (p=0.004), even after controlling for CALGB prognostic grouping (adjusted HR 0.42 [95% CI 0.21, 0.84] p=0.01). Predictable, reversible myelosuppression was seen. 36 (73.5%) pts had grade 3/4 heme toxicities; 19 (57.6%) had Gr 4 heme tox in RP2D cohorts with 45.4% receiving prophylactic platelet transfusions (median 1, range 1-4) and 6 GCSF (without fever). 14 (28.6%) had infusion reactions (without pre-meds), with 1 pt having Gr 2 and withdrawing prior to 2nddose. 5 (10.2%) had transient Gr 1/2 AST/ ALT. Conclusions: Fractionated 177Lu2J591 is well tolerated with predictable, reversible myelosuppression and PSA and CTC declines. With dose-fractionation, the cumulative dose MTD is 14228% higher than single dose MTD with similar toxicity. Prior dose-response is confirmed. Clinical trial information: NCT00538668.

Background: Advances in the treatment of localized CaP have led to decreased recurrences and improved OS. The Intermediate Clinical Endpoints in CaP (ICECaP) Working Group is conducting an individual patient data (IPD) meta-analysis of possible surrogate endpoints for localized CaP trials. We hypothesized that DFS is a surrogate for OS. Methods: By June 2013, we systematically identified 102 eligible randomized trials (completed or ongoing) comparing treatments in localized CaP and collected IPD from trialists. DFS was defined from randomization (R) to the first evidence of clinical recurrence (loco-regional or distant) or death from any cause; or was censored at the date of last follow-up. OS was defined from R to death from any cause. We evaluated the surrogacy of DFS with OS using a meta-analytic 2-stage validation model where 2 conditions must hold to claim DFS is a surrogate for OS (Buyse et al, 2000 & 2011 table). The secondary objective evaluated surrogacy of time to disease recurrence (TDR) with disease specific survival (DSS), defined analogously to DFS and OS but with nonCaP deaths censored. Results: By January 2016, IPD from 16,999 men randomized in 20 mature trials were available for analysis. 32% of the men were from prostatectomy trials, about 30% of the men had intermediate and 50% high-risk disease and 86% were , 74 yo. With median follow-up of 10.2 years, 32% (N = 5,370) men had died and 30% of these deaths (N = 1,592) were due to CaP. Conclusions: DFS can be used as a surrogate of OS and TDR as a surrogate of DSS.

Group Total Low dose RP2D 40 x2 cohort 45 x2 cohort

N

Any PSA decline

> 30% PSA decline

> 50% PSA decline

Median OS mo [95% CI]

49 16 33 16 17

55.1% 37.5% 66.7% 50.0% 87.5%

32.7% 12.5% 42.4% 25.0% 58.8%

16.3% 6.3% 21.2% 12.5% 29.4%

22.9 [16.2, 29.7] 14.6 [9.9, 19.4] 27.7 [15.8, 39.6] 19.6 [9.1, 302] 48.3 [16.0, 80.6]

Condition 1: Correlation between the ICE and true endpoint

Condition 2: Correlation between study specific treatment effect on endpoints

Meta-analytic 2-stage validation Model

R2 from weighted linear regression of t-year Kaplan Meier estimates (95% CI)

Plackett’s copula model estimate of Kendall Tau correlation (95% CI)

R2 from weighted linear regression of treatment effects (log HR) (95% CI)

DFS as a surrogate for OS

10y OS vs 5y DFS 0.80 (0.67,0.86) 8y OS vs 5y DFS 0.85 (0.76,0.90) 10y DSS vs 5y TDR 0.73 (0.56,0.81) 8y DSS vs 5y TDR 0.79 (0.67,0.85)

0.85(0.85,0.86)

0.80(0.60,0.87)

0.71(0.70,0.72)

0.61(0.29,0.75)

TDR as surrogate for DSS

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

272s

Genitourinary (Prostate) Cancer

5024

Poster Session (Board #281), Sat, 1:00 PM-4:30 PM

The Prostate Cancer Registry: Do patients with metastatic castrationresistant prostate cancer (mCRPC) differ according to metastatic status at diagnosis? First Author: Simon Chowdhury, Guy’s Hospital, London, United Kingdom Background: The mCRPC patient population is heterogeneous, and clinical outcomes are influenced by clinical presentation and treatment history. Using the Prostate Cancer Registry (NCT02236637) database, we assessed whether the subset of mCRPC patients with metastatic (M) status at initial diagnosis represents patients with different clinical and disease characteristics in a real-life context. Methods: We report findings from the first 1323 consecutive patients in the Prostate Cancer Registry, a 3-year, prospective, observational study of the routine management of . 3000 patients with mCRPC across Europe. Results: Of the 1323 patients analyzed, 549 had distant metastases at initial diagnosis (M1, 41.5%), 526 did not (M0, 39.8%), and 248 were not evaluable (Mx, 18.7%). Conclusions: In this sample of 1323 men with mCRPC in routine practice, a higher percentage with M1 at diagnosis presented with aggressive and extended disease at mCRPC. M status at initial diagnosis may be a critical indicator of future treatment and prognosis in mCRPC, and follow-up of these patients over the course of the study will clarify how these differences and subsequent treatment may affect clinical outcomes.

Characteristics. M stage At initial diagnosis Patients, n Mean (SD) patient age, years $ 75 years, n (%) Gleason score 8-10, n (%) Prior to study entry Years from diagnosis to CRPC, median (range) Years from first M diagnosis to study entry, median (range) Radical prostatectomy, n (%) Prostate radiotherapy, n (%) At study entry Prostate-specific antigen, ng/mL, median (range) Bone lesions, n (%)a 0 1-20 . 20 ECOG PS, n (%)a 0-1 2 Any analgesic therapy, n (%) SD, standard deviation.

a

M0

M1

Mx

526 73.6 (7.2) 241 (45.8) 218 (44.6)

549 70.6 (8.6) 187 (34.1) 331 (66.5)

248 74.2 (7.7) 128 (51.6) 118 (52.7)

6.2 (0-20) 2.0 (0-15)

2.4 (0-16) 2.5 (0-13)

5.5 (0-29) 1.9 (0-11)

204 (38.8) 300 (57.0)

31 (5.6) 49 (8.9)

69 (27.8) 84 (33.9)

45.9 (0.0-5910.0)

61.7 (0.0-10710.0)

46.1 (0.0-2530.5)

49 (13.5) 208 (57.5) 49 (13.5)

8 (1.9) 228 (55.2) 75 (18.2)

14 (7.9) 110 (61.8) 21 (11.8)

436 (87.2) 51 (10.2) 192 (36.5)

426 (83.4) 71 (13.9) 274 (49.9)

201 (87.4) 21 (9.1) 111 (44.8)

Some patients not evaluable.

5025

Background: AA men have higher prostate cancer (PC) incidence and mortality than CAU men, potentially related to biologic or socioeconomic factors. A retrospective subgroup analysis suggested greater overall survival (OS) benefit in AA metastatic castrationresistant PC (mCRPC) pts treated with sip-T compared with the overall population (McLeod et al, AUA 2012, abstract 953). Data on AA and CAU pts from PROCEED (NCT01306890), a phase 4 registry of mCRPC pts receiving sip-T, may further delineate treatment benefit for AA men. Methods: Men age $18 y with mCRPC treated with sip-T were enrolled in PROCEED. Baseline pt characteristics, treatment history, safety, OS, ACIs, and time to ACIs (tACI) were recorded. Results: 902 pts received $1 sip-T infusion, including 231 (11.7%) AA men. Primary Gleason score 5 was more common in CAU pts (p,0.05). In AAs, median baseline prostate-specific antigen (PSA) was higher (p,0.001), and PSA doubling time (PSADT; p,0.01) was shorter (Table). Other baseline pt characteristics were similar between groups. A trend toward lower percentage of AAs receiving radical prostatectomy before sip-T (p=0.06) was observed. Post–sip-T, both groups had similar time to first (CAU 7.3 vs AA 6.9 mo) and second (CAU 28.7 vs AA 27.1 mo) ACI. Over 30% of pts had not received an ACI 12 mo after sip-T (CAU 32.3% vs AA 31.9%). Conclusions: Despite higher baseline PSA in AA pts, which may predict less clinical benefit post–sip-T, CAU and AA pts had similar tACI after sip-T. Additional follow-up will assess potential OS differences in AA and CAU pts. Clinical trial information: NCT01306890. Characteristic, n (%) Gleason sum 7 $8 Primary Gleason score 5 Positive lymph nodes Bone metastasis 1-10 .10 ECOG PS 0 1 Alkaline phosphatasea, U/L Lactate dehydrogenasea, U/L Time from diagnosisa, mo PSAa, ng/mL PSADTa, mo a

5026

Poster Session (Board #283), Sat, 1:00 PM-4:30 PM

Blood-based predictive biomarkers for overall survival (OS) in patients (pts) receiving the immunotherapy sipuleucel-T (sip-T). First Author: William K. Oh, Division of Hematology and Medical Oncology, Mount Sinai School of Medicine, New York, NY Background: Sip-T is an FDA-approved immunotherapy for the treatment of asymptomatic or minimally symptomatic metastatic castration-resistant prostate cancer (mCRPC). While baseline prostate-specific antigen correlates with prolonged OS in pts treated with sip-T, biomarkers predictive of OS remain elusive. Gene expression signatures in peripheral blood (PB) correlate with prognosis in mCRPC (Ross et al, Lancet Oncol. 2012). This study sought to identify a candidate PB gene signature predictive of OS benefit with sip-T. Methods: PB mononuclear cells from sip-T or control (CN) arm pts in the IMPACT trial (NCT00065442) were collected prior to leukapheresis. Affymetrix gene chip screening evaluated gene expression; association of gene expression with OS was performed using a multivariate Cox proportional hazards regression model, where OS was fit to gene expression and the 2003 baseline Halabi prognostic variables. The top 50 gene candidates were selected for qPCR confirmation. Results: 242 pt samples were analyzed (163 sip-T, 79 CN). 5 genes were associated with OS; high expression of SNTB1 and CHI3L2 and low expression of SYNGR3, AURKC, and ZNF268 were associated with improved OS in sip-T–treated pts but not in CN arm pts. A composite gene score (CGS) was calculated incorporating expression of these genes. In a CGS tertile analysis, OS in the top and middle sip-T tertiles was significantly better compared with the corresponding CN groups (Table). The lowest sip-T tertile had a median OS similar to that of the CN arm. Conclusions: CGS based on baseline gene expression may predict OS outcomes for pts receiving sip-T. Compared with the CN arm, OS was significantly improved in the top two tertiles of sip-T–treated pts; OS in the bottom tertile was similar to the CN arm. External validation is currently underway. Clinical trial information: NCT00065442. Tertilea

N (CN)

N (Sip-T)

Median OS, mo (95% CI) CN

Median OS, mo (95% CI) Sip-T

Hazard ratiob

P-valueb

Top Middle Bottom

26 25 25

53 52 52

21.1 (15.5, NE) 21.6 (14.2, NE) 23.8 (12.8, NE)

27.1 (23.0, NE) 24.6 (18.4, NE) 18.6 (13.6, 26.3)

0.41 0.45 1.16

0.007 0.02 0.65

Poster Session (Board #282), Sat, 1:00 PM-4:30 PM

Characteristics and anticancer interventions (ACIs) in African American (AA) and Caucasian (CAU) patients (pts) treated with sipuleucel-T (sip-T): Realworld experience from the PROCEED registry. First Author: Andrew J. Armstrong, Duke University Medical Center, Duke Cancer Institute Divisions of Medical Oncology and Urology, Duke University, Durham, NC

CAU (n=1,711 or as noted)

AA (n=231 or as noted)

516 (30.2) 875 (51.1) n=1478 208 (14.1)* 538 (31.4) n=1430 997 (69.7) 248 (17.3) 1130 (66.0) 501 (29.3) 81 185 59.4 13.9† 8.36**

67 (29.0) 106 (45.9) n=184 17 (9.2)* 78 (33.8) n=191 136 (71.2) 29(15.2) 141 (61.0) 80 (34.6) 87 192 72.0 33.2† 8.01**

Median. *p=0.03; †p,0.001; **p=0.007 for CAU vs AA.

5027

Poster Session (Board #284), Sat, 1:00 PM-4:30 PM

Comprehensive genomic profiling of neuroendocrine carcinoma of the prostate. First Author: Philip J. Stephens, Foundation Medicine, Inc., Cambridge, MA Background: Neuroendocrine carcinoma of the prostate (NCAP) is an aggressive high grade malignancy that often presents as metastatic disease. Current treatments of this tumor have only modest benefit leading investigators to query whether comprehensive genomic profiling (CGP) would have potential to discover novel targets of therapy. Methods: DNA was extracted from 40 microns of FFPE sections from 37 consecutive cases of relapsed/metastatic NCAP. CGP was performed on hybridization-captured, adaptor ligation based libraries to a mean coverage depth of 583X for up to 315 cancer-related genes plus 37 introns from 14 genes frequently rearranged in cancer. Genomic alterations (GA) included base substitutions (SUB), INDELs, copy number alterations (CNA) and fusions/rearrangements. Clinically relevant GA (CRGA) were defined as GA linked to drugs on the market or under evaluation in mechanism driven clinical trials. Results: The mean age of the men in this study was 65.1 years (range 43 to 83 years). All (100%) cases were positive for neuroendocrine markers on immunohistochemical staining and were Stage IV at the time of CGP. Samples used for sequencing were obtained from the primary tumor in 9 (24%) of NCAP and from metastatic sites in 28 (76%) of NCAP (12 liver, 6 LN, 2 each from bladder, pelvis and soft tissue, and 1 each from rectum, bone, urethra and ureter. There were 213 total GA (5.8 GA/sample) and 47 CRGA (1.3 CRGA/ sample). The most frequent GA were non-CRGA mutations in TP53 (68%) and RB1 (51%). TMPRSS:ERG fusions were identified in 32% of cases whereas AR was altered in 8% (1 mutation and 2 amplifications). The most frequent CRGA involved PTEN (32%), BRCA2 (14%), FGFR1 (5%), PIK3CA (5%) and AKT2 (3%). No alterations in BRAF were identified. Clinical responses to MTOR inhibitors in patients with MTOR pathway alterations will be presented. Conclusions: NCAP has distinctive genomic alterations from classic acinar CAP including reduced frequencies of alterations in TMPRSS: ERG and AR and frequent RB1 mutations. Multiple alterations in the MTOR pathway identified in this infrequent tumor type suggest that these patients may be candidates for MTOR inhibitors and other targeted therapies.

a Tertiles by CGS. bCompared with corresponding CN arm tertile. CI = confidence interval; NE = not estimable.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Genitourinary (Prostate) Cancer 5028

Poster Session (Board #285), Sat, 1:00 PM-4:30 PM

Clinical characteristics of metastatic castration-resistant prostate cancer (mCRPC) patients with DNA repair (DNAr) defects. First Author: Michael Kolinsky, The Royal Marsden, The Institute of Cancer Research, Sutton, United Kingdom Background: Recent data show that 20-30% of mCRPC patients have defects in DNAr genes. We have evaluated the clinical phenotype of these patients. Methods: Medical records of mCRPC patients treated at the Royal Marsden Hospital from 2005-2015 with deleterious DNAr defects identified by NGS were reviewed. Kaplan-Meier survival analyses and Log Rank testing were pursued. Results: Overall, 36 patients with DNAr defects were identified: 21 BRCA2, 8 ATM, 2 BRCA1, 1 HDAC2, 1 FANCA, 1 PALB2, 1 CDK12, 1 BLM. Median age at diagnosis was 61 years, with 24/34 (71%) metastatic at diagnosis, 22/29 (76%) with Gleason score $ 8, and median PSA at diagnosis of 86. Median follow-up was 62 (21-179) months (m), with 19 deaths. Median time to CRPC, overall survival (OS) from CRPC, and OS were 11, 67, and 111 m, respectively. Patients with BRCA2 defects were less likely to present with metastatic disease (55% vs 93%, p = 0.024), but had worse OS (69m vs 167m, p = 0.012) and OS from CRPC (50m vs 108m, p = 0.007), than patients with defects in other DNAr genes. Response to systemic therapies are presented in the table below. Conclusions: Highgrade cancers and metastatic disease at diagnosis were frequent in this cohort. PSA response rates to therapy were similar to those reported in unselected mCRPC. BRCA2 aberrant cancers may have a worse prognosis and be less responsive to AR targeted therapies than patients with other DNAr defects. AR targeting pre-chemo

AR targeting post-chemo

docetaxel

cabazitaxel

carboplatin

PARPi

$50% PSA response [BRCA2 vs other] RECIST response Median PFS in m

71.4% (20/28) 75% (12/16) 25% (1/4) 46.2% (12/26) 71.4% (5/7) 36.4% (8/22) [75% vs 50%, p = 0.355] [23% vs 56%, p = 0.021†] [75% vs 67%, p = 0.610] [71% vs 89%, p = 0.050] [33% vs 0%, p = 0.655] [53% vs 33%, p = 0.205] 16.7% (1/6) 42.9% (3/7) 33.3% (3/9) 55.6% (5/9) 0% (0/3) 42.9% (6/14) 2.8 4.3 7.0 7.4 0.7 6.2 [2.8 vs 2.3, p = 1.000] [4.1 vs 4.5, p = 0.821] [6.0 vs 7.3, p = 0.966] [4.5 vs 8.4, p = 0.242] [2.1 vs 0.7, p = 0.480] [8.9 vs 3.6, p = 0.126] Median OS from start in m 24.1 33.9 43.6 22.2 4.6 15.9 [BRCA2 vs other] [24.1 vs 8.9, p = 0.678] [24.1 vs 41.4, p = 0.060] [35.5 vs 43.6, p = 0.053] [21.0 vs 38.1, p = 0.009†] [4.6 vs 4.9, p = 0.918] [15.4 vs 16.9, p = 0.628] †

statistically significant

TPS5030

Poster Session (Board #287), Sat, 1:00 PM-4:30 PM

Analysis of AR-driven resistance mechanisms in a phase 1/2 study of EPI506, a novel AR N-terminal domain inhibitor, in men with metastatic castration-resistant prostate cancer (mCRPC) with progression after enzalutamide or abiraterone. First Author: Kim N. Chi, British Columbia Cancer Agency, Vancouver, BC, Canada Background: Androgen receptor (AR) signaling remains essential for CRPC progression and current inhibitors, such as abiraterone and enzalutamide, focus on blocking androgen synthesis and/or preventing androgen binding to the receptor. Eventually, tumor escape mechanisms enable AR signaling to become reactivated despite these interventions and represent a challenge to treatment of these tumors. Known AR-driven resistance mechanisms include AR copy number increase, gain-of-function mutations, and constitutively active AR splice variants. EPI-506 is a first-in-class, highly specific small molecule AR N-terminal domain (NTD) inhibitor that has potential to overcome shortcomings of current hormone therapies by targeting AR transcriptional activity. By targeting the NTD, EPI-506 can affect a broader AR population, including variants implicated in resistant mCRPC tumors. Methods: This is an ongoing Phase I study (3+3 design) which will be followed by a 3-arm Phase II study in CRPC patients that are progressing after abiraterone and/or enzalutamide. Biomarkers of AR driven resistance will be explored using circulating tumor cell (CTC) and plasma derived cell free DNA (cfDNA). CTCs and cfDNA samples will be collected at serial timepoints (baseline, 12-weeks, and progression). Presence of AR splice variants, including AR-V7, will be analyzed using the Epic Sciences CTC platform to detect AR C-terminal loss followed by immunofluorescence or RT-PCR detection of AR-V7. cfDNA will be subjected to targeted sequencing of AR and 74 other prostate cancer related genes using an Ion Ampliseq Custom DNA panel for mutations and copy number changes. Information gathered will be used to evaluate the clinical activity of EPI-506 in the context of known AR resistance mechanisms. The clinical study will be the first to evaluate the novel NTD inhibitor EPI-506 in men with enzalutamideand/or abiraterone-failure mCRPC, and is the first agent with the potential to inhibit both canonical and variant-mediated AR signaling. Clinical trial information: NCT02606123.

273s

5029

Poster Session (Board #286), Sat, 1:00 PM-4:30 PM

Prognostic significance of free testosterone levels during chemotherapy with carboplatin plus docetaxel (CD) in metastatic castration- and docetaxelresistant prostate cancer (mDRPC). First Author: Christoph W. Reuter, Dept. Hematology, Hemostasis, Oncology and Stem Cell Transplantation, Hannover Medical School, Hannover, Germany Background: Carboplatin plus docetaxel may be effective in mDRPC. Platinum(II)-complexes have been shown to interfere with steroid biosynthesis lowering testosterone levels by inhibiting the cholesterol side chain cleavage enzyme (CYP11A1), 3b-hydroxysteroid dehydrogenase (HSD3B1,2) and 17a hydroxylase/C17,20-lyase (CYP17A1). Methods: Docetaxel failure/resistance was defined according to the Prostate Cancer Working Group (PCWG2 2007) criteria. Treatment consisted of at least 2 cycles of carboplatin AUC5 iv for 30 min on day 1 every 4 weeks (q4w), docetaxel (35 mg/m2) iv for one hour on days 1, 8, (15) plus prednisone 2x5mg/day orally after receiving informed consent until disease progression or occurrence of intolerable adverse effects. Efficacy measures were done following PCWG2 recommendations. Free testosterone (FT) levels were measured before (n = 70) and during CD chemotherapy (n = 62). Results: Of the 93 pts. treated since February 2005, 95.7% had bone, 41.9% lymph node, 26.9% liver and 19.4% lung metastases. At the time of the current analysis, the median follow-up time was 15.4 months. The objective response rate was 38.6% and the disease control rate 71.4% in the 57 pts. with measureable disease. Response of prostate-specific antigen ( $ 50%) was observed in 46/93 (49.5%) patients. Median PFS was 7.2 months (CI 95% 6.2, 8.1) and median OS was 15.6 months (CI 95% 11.5, 19.8). The most common reversible grade 3/4 toxicity was leukopenia/ neutropenia (40.0/38.1%). Median FT levels were 0.65 pg/ml before and , 0.18 pg/ml during carboplatin/docetaxel treatment (nadir levels, p , 0.001; detection limit , 0.18 pg/ml). Median serum androgene levels (T+DHT) were 0.1 ng/ ml before and below the detection limit of , 0.05 ng/ml during DC treatment. In multivariate analyses, LDH, PSA response, FT nadir levels below the detection limit ( , 0.18 pg/mL) during CD treatment were associated with longer OS (p , 0.05). Conclusions: These data suggest that CD may be an important second-line treatment option for mDRPC patients by inhibiting the testosterone biosynthesis.

5031

Poster Session (Board #288), Sat, 1:00 PM-4:30 PM

Age, race, and survival of men with de novo distant metastatic prostate cancer (M1PC). First Author: Brandon David Bernard, Dana-Farber Cancer Institute, Boston, MA Background: Older age at diagnosis is associated with a higher frequency of de novo M1PC and worse cancer-specific survival (CSS). In the PSA era, outcomes with androgen deprivation therapy in clinical trials for M1PC are similar for whites (W) and blacks (B) however it is unknown if this is true at the population level; the relationship between age, race and CSS is also unknown. The aim of this study was to describe the frequency and rate of de novo M1PC and CSS for a contemporary cohort of men with de novoM1PC and assess for associations with age and race. Methods: The Surveillance, Epidemiology, and End Results (SEER) registry was used to identify W and B men diagnosed with de novoM1PC (pelvic node only disease excluded) from 2004-2012. Age groups were defined as # 74 or $ 75 years. Frequency, rate and CSS were determined for W and B in the two age groups. Descriptive endpoints were used to define associations of age and race with frequency, rate and CSS. Frequencies, age-adjusted rates and survival were calculated using SEER*Stat. Chi-square tests were used to evaluate the associations between age group, race and frequency; a 2-sided p , 0.05 was considered statistically significant. Results: From 2004-2012, prostate cancer was diagnosed in 452,698 men. The table shows frequency (% M1PC of all PC within race/age group), rate per 100,000 and CSS of men with de novo M1PC (*p , 0.05 between and within age groups). Compared to men # 74, there was a significantly higher frequency of de novoM1PC in men $ 75 regardless of race (p , 0.0001). CSS was also significantly worse in men $ 75 regardless of race. Within both age groups, the proportion of de novoM1PC was greater for B than for W (p , 0.0001). Conclusions: The rate of de novoM1PC is significantly higher and CSS significantly lower in men $ 75, irrespective of race. Given the poor CSS in men $ 75, due consideration for chemohormonal therapy in men fit for docetaxel should be undertaken despite older age. £ 74 De novo M1PC Frequency (%)* Rate per 100,000* CSS Median (months) 1 yr (95% CI) 5 yr (95% CI)

‡ 75

W

B

W

B

8,732 (2.9) 3.1

2,793 (4.6) 8.3

8,091 (9.9) 58.4

1,165 (12.6) 108.9

36

34

22

20

82.2 (81.3-83.1) 34.6 (33.2-36.0)

80.1 (78.5-81.7) 33.4 (30.9-35.9)

66.8 (65.6-68.0) 22.7 (21.3-24.2)

64.4 (61.2-67.4) 21.1 (17.7-24.8)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

274s 5032

Genitourinary (Prostate) Cancer Poster Session (Board #289), Sat, 1:00 PM-4:30 PM

Expression and genetic variants in stress-related signaling pathways in lethal and nonlethal prostate cancer. First Author: Donghao Lu, Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden Background: Recent data suggest that dysregulated stress-related signaling pathways may play a role in prostate cancer progression. We aim to compare expression of selected genes in the adrenergic, glucocorticoid, dopaminergic, and serotoninergic signaling pathways in tumor tissue from men with lethal and nonlethal prostate cancer, and to explore whether genetic variants may explain potential variation. Methods: We included 504 Swedish men diagnosed with incidental prostate cancer through TURP during 1977-1998 with up to 28 years of follow up. For those with tumor tissue available (N = 396), we measured mRNA expression of 223 selected genes included in predetermined stress-related signaling pathways. Using DNA from normal prostate tissue (N = 262), we genotyped 14 stress-related genes (29 SNPs). Associations of pathway gene expression and genetic variants with lethal prostate cancer were assessed as primary outcomes, and with tumor biomarkers as secondary outcomes. Results: Differential mRNA expression of genes in the serotoninergic pathway was associated with risk of lethal prostate cancer (p = 0.007); similar but somewhat weaker associations were noted for the adrenergic (p = 0.014) and glucocorticoid (p = 0.020) pathways. None of the studied genetic variants were associated with risk of lethal cancer after correcting for multiple testing, but a link was suggested for variants of HTR2A (rs2296972; OR 0.5, 95% CI 0.3-0.8, p = 0.002) and NR3CI (rs33388; OR 1.6, 95% CI 1.0-2.4, p = 0.035) genes (within the serotoninergic and glucocorticoid signaling pathways) in over-dominant models. No significant association between genotypes and mRNA expression of any individual gene in the corresponding pathways were observed. Genetic variants of HTR2A and NR3CI were associated with increased prostatic intraepithelial neoplasia and cell proliferation (all p , 0.05), respectively. Conclusions: Our findings lend some support to the hypothesis that dysregulated signaling in stress-related molecular pathways, particularly the serotoninergic, is involved in prostate cancer progression. If confirmed, this finding could open up new paths for research targeted towards reducing prostate cancer progression.

5034

Poster Session (Board #291), Sat, 1:00 PM-4:30 PM

Circulating tumor DNA (ctDNA) burden and actionable mutations in treatment-na¨ıve metastatic castration-resistant prostate cancer (mCRPC). First Author: Alexander William Wyatt, Vancouver Prostate Centre, University of British Columbia, Vancouver, BC, Canada Background: The molecular landscape underpinning response to therapy in mCRPC patients is undefined. The analysis of plasma cell-free DNA (cfDNA) has recently emerged as a minimally-invasive method to explore tumor characteristics and therefore has potential utility as a predictive biomarker. However, the broad applicability of cfDNA remains unclear, especially in treatment-na¨ıve mCRPC where tumor burden is typically lower. Methods: To dissect the landscape of tumor-derived cfDNA we collected baseline cfDNA samples from 36 chemotherapy-na¨ıve mCRPC patients enrolled in an ongoing randomized phase II cross-over trial of abiraterone vs enzalutamide (NCT02125357) and performed deep targeted sequencing using a custom NimbleGen SeqCap EZ Choice panel of 72 mCRPC-related genes. 10ng of cfDNA was used for library construction and minimum read depth was ~500X. Results: Using a conservative detection threshold of . 2% allelic frequency, we identified a total of 53 somatic coding mutations in 36 baseline cfDNA samples, with 22/36 samples harboring at least one mutation. Mutations were highly consistent with advanced prostate cancer, falling for example in the AR (n = 5), SPOP (n = 3), PTEN (n = 3) and TP53 (n = 9). The median allelic frequency of coding mutation was 15.8% (range 2.0-74.9), with 10 patients harboring at least one mutation at greater than 25% frequency. Assuming heterozygosity, this suggests that circulating tumor DNA (ctDNA) comprises a majority of the total cfDNA fraction in almost a third of the cohort. Known activating mutations in PIK3CA (E545K, H1074R) were detected in in two patients, and several DNA damage repair genes, including BRCA2, ATM and FANCD2, harbored deleterious mutations and evidence of bi-allelic loss. Conclusions: Our data suggests that a high ctDNA fraction amenable to next-generation sequencing is present in the majority of chemotherapy-na¨ıve mCRPC patients starting 1st line abiraterone or enzalutamide. The readiness of detection of clinically-actionable and informative mutations suggests that targeted cfDNA sequencing can form the basis of a practical biomarker for clinical trial enrichment.

5033

Poster Session (Board #290), Sat, 1:00 PM-4:30 PM

PSA response in black and white patients treated with abiraterone acetate (AA) for metastatic castrate-resistant prostate cancer (mCRPC). First Author: Sundhar Ramalingam, Duke University Medical Center, Durham, NC Background: Pivotal trials of AA for patients with mCRPC enrolled few black patients (pts), a population with a higher mortality from prostate cancer. Evidence suggests differences in androgen receptor signaling between black (B) and white (W) PC pts; therefore, we retrospectively investigated PSA response in B vs W pts treated with AA. Methods: We performed a 2:1, W vs B case-control study. Pts were identified through pharmacy records for AA use in mCRPC between May 1, 2008 and June 16, 2015 at Duke University, with pts matched for prior docetaxel use. Pts with prior enzalutamide were excluded. Baseline prognostic factors of performance status, PSA, Gleason score, alkaline phosphatase, albumin and hemoglobin (hgb) were identified. The primary objective was to compare the rate of a $ 90% PSA decline from baseline PSA with AA therapy between groups. Results: Baseline characteristics among pts (n = 45 B, n = 90 W) were similar with the exception of median hgb (B = 11.4 g/dl, W = 12.3 g/dl). The proportion of black pts achieving a $ 90% PSA decline was 37.8%, vs. 28.9% for white pts (p = 0.296; Table). Statistically significant differences were found in the proportion of pts achieving a $ 50% PSA decline [B = 68.9%, W = 48.9% (p = 0.028)] and $ 30% PSA decline [B = 77.8%, W = 54.4% (p = 0.008)]. Primary AA-refractory disease (PSA increase as best response) trended to be more in W = 31.1% than in B = 15.6% pts (p = 0.052). Median treatment duration (B = 9.36 months, W = 8.25 months) did not differ (Wilcoxon p = 0.44). Median overall survival [B = 27.3 months (95% CI, 13.9, not estimable), W = 24.8 months (95% CI 19, 31.6) (p = 0.669)] and median time to PSA progression [B = 11.0 months (95% CI 4.3, 18.0), W = 9.4 months (95% CI: 6.2, 13.0) (p = 0.917)] did not differ. Conclusions: Black pts may have a higher PSA response to AA than white pts. An ongoing prospective clinical study (NCT01940276) is evaluating outcomes between black and white pts treated with AA. PSA response outcome $ 30% PSA decline $ 50% PSA decline $ 90% PSA decline PSA increase as best response

5035

# Black pts (n = 45)

% Black pts

# White pts (n = 90)

% White pts

P-value

35 31 17 7

77.8% 68.9% 37.8% 15.6%

49 44 26 28

54.4% 48.9% 28.9% 31.1%

0.008 0.028 0.296 0.052

Poster Session (Board #292), Sat, 1:00 PM-4:30 PM

Profiling of circulating tumor (ct)-DNA for potentially actionable targets in prostate cancer (PCa). First Author: Guru Sonpavde, University of Alabama at Birmingham Comprehensive Cancer Center, Birmingham, AL Background: Genomic profiling of metastatic PCa tissue is challenging and circulating tumor cells may not recapitulate entire tumor biology. Cell-free ctDNA is detectable in most patients (pts) with advanced cancer and may capture relevant alterations in cancer cells from different tumor sites. We studied the frequency of ctDNA alterations in potentially actionable genes in pts with PCa. Methods: Patients with PCa that underwent ctDNA analysis using Guardant360 were identified. A 70-gene ctDNA next generation sequencing panel from a CLIA-licensed, CAP-accredited laboratory (Guardant Health, Inc.) was used to sequence all exons in 29 cancer genes, critical exons in 39 genes, and amplifications (16 genes), fusions (6 genes), indels (3 genes), harvested from 10 mL of peripheral blood. Results: We studied 449 pts with PCa (median age 71, range 42-92) with at least one ctDNA test. ctDNA alterations were detectable in 386 patients (84%). Genes with the highest frequency of somatic alteration including amplification (*) in the overall group were: TP53 (54%), AR* (47%), FGFR1/2/3* (20%), MYC* (19%), PIK3CA* (19%), CDK4/6* (16%), EGFR (15%), BRAF *(14%), APC (13%), NF1 and MET (each 12%). A subset of 128 pts with confirmed metastatic castration-resistant prostate cancer (mCRPC) had a similar distribution and frequency of genomic alterations compared to the overall group. Recurrent somatic mutations in AR, PIK3CA, APC and CTNNB1 were identified in the mCRPC cohort at similar frequencies compared to prior studies of tumor tissue analyses. 38 mCPRC pts were confirmed to receive $ 4 cycles of 1st or 2nd line taxane chemotherapy and had f/u data; TP53 mutations were associated with longer PFS (PCWG2 criteria; p = 0.032) and PIK3CA mutations were associated with shorter PFS (p = 0.024) following taxane therapy, using a log rank test. Serial ctDNA profiling available in 10 pts revealed clonal evolution of alterations in AR, PIK3CA, BRCA2 and NF1. Conclusions: In this large dataset, ctDNA was frequently detected in pts with PCa with an alteration profile similar to that seen in tumor tissue. ctDNA analysis offers a non-invasive means of serially profiling tumor DNA to identify treatment targets and emerging markers of resistance.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Genitourinary (Prostate) Cancer 5036

Poster Session (Board #293), Sat, 1:00 PM-4:30 PM

Association of androgen receptor variant 9 (AR-V9) mRNA expression levels in metastatic tissue with resistance to abiraterone acetate/prednisone (AA/ P). First Author: Manish Kohli, Mayo Clinic, Rochester, MN Background: AA/P is an approved treatment for metastatic castrate resistant prostate cancer (mCRPC), for which there are no validated resistance/ response markers. We conducted a prospective trial to evaluate if Androgen Receptor (AR) or AR variants (AR-Vs) mRNA expression in tissue metastases can predict early resistance to AA/P. Methods: Metastatic sites of CRPC stage patients (pts) initiating pre-chemo AA/P were serially biopsied twice, prior to (pre-AA/P) and after 12 weeks of treatment (post-AA/P). Composite response at 12 weeks (primary endpoint) was evaluated with PSA, RECIST, imaging and symptoms (per PCWG2). Pre-AA/P metastases mRNA expression levels {for AR, AR-Vs (AR-V3, AR-V7, AR-V9, AR23, AR45), Chromogranin-A (CHGA); and four cell cycle division genes}; Serum PSA, testosterone, CHGA levels; Gleason Score (GS) at initial diagnosis; high V/s low volume disease; time on hormone therapy to mCRPC stage were all included in a logistic regression model for predicting AA/P resistance at 12 weeks. A final multivariate analysis (MVA) model fitted only those factors thought to be clinically relevant or with an entry threshold of p # 0.2 in univariate analysis. Results: Between 6/2013 & 8/2015 of 92 pts enrolled, 78 had complete gene expression & outcome data at the time of analysis (12/ 2015). Gene expression at pre and post AA/P was available for 65/78 pts. Median pre-AA/P PSA was 14.6 ng/ml (IQR: 8.0 - 45.6). Progression at 12 weeks was observed in 32/78 pts. At the univariate level elevated pre-AA/ P expression of AR-V3 (p = 0.03), AR-V7 (p = 0.17), AR-V9 (p = 0.02), ARV7/FL ratio (p = 0.05), AR-V9/FL ratio (p = 0.03) and cell division cycle gene CDC45 (p = 0.11) met the threshold for inclusion into MVA. Elevated AR-V9 expression alone was associated with progression at 12 weeks in the MVA (OR: 4.0; CI 1.31 – 12.2; C-Index: 0.62). Additionally pts with elevated ARV9 expressions at either biopsy time point (pre or post-AA/P 12-week) were more likely to progress on AA/P (OR: 3.3; 95% CI: 1.11 – 9.92; p = 0.03; Cindex: 0.63). Conclusions: Elevated AR-V9 mRNA expression in metastases is associated with early progression on AA/P and could serve as a biomarker of resistance. Clinical trial information: NCT# 0195364.

5037

275s Poster Session (Board #294), Sat, 1:00 PM-4:30 PM

TMPRSS-ERG fusion in men with prostate cancer (PCa) and non-prostate malignancies: Defining a role for comprehensive genomic profiling (CGP) to guide clinical care. First Author: Primo Lara, University of California, Davis, Sacramento, CA Background: Genomic rearrangements involving androgen-regulated TMPRSS & the ETS transcription factor ERG are found in ~50% of PCa tumors. Thus identification of TMPRSS-ERG is taken as a marker of prostatic origin. We assessed TMPRSS-ERG in a large series of genomic profiles of patients with various advanced cancers, including PCa and other malignancies. Methods: CGP of 315 cancer genes was applied to 50ng of DNA from submitted cases to assess all classes of genomic alterations. TMPRSS-ERG frequency across histologic subtypes was determined. CGP data from an index case of PCa with histologic evolution from adeno- to pure squamous cell cancer (SCC) is described. Results: TMPRSS-ERG was identified in 288 of 5,149 tumors in men; it was not detected in tumors from females. PCa accounted for 250 cases. Frequency by tumor type is shown. The index case is a patient with Gleason 9 PCa s/p surgery and radiation who developed a large left retrovesical mass pathologically proven to be pure SCC. TMPRSSERG was detected by CGP in the SCC tumor, primary PCa, and metachronous PCa metastasis. The patient is benefiting from androgen deprivation. Phylogenetic trees showing clonal and histopathologic evolution of PCa will be presented. Conclusions: In this large CGP dataset, TMPRSS-ERG fusion was seen in ~30% of PCa regardless of histologic type but was rarely detected in non-PCa tumors occurring in men. CGP can identify TMPRSS-ERG in PCa presenting as variant histology. As TMPRSS-ERG is almost exclusively associated with PCa, CGP should be considered in appropriate cases of metastatic cancer in men where the anatomic site of origin is uncertain as there are potential therapeutic implications. Subtype Prostate acinar ADC Prostate NE Ca Prostate Ca, undiff’d Prostate ductal ADC Undiff NE Ca of UP ADC, UP SCC, UP Ca of UP/UP (NOS) Lung Small Cell Bladder Ca, NOS Urothelial Ca Pancreas Ca, NE Pancreas Ca, NOS Lung SCC

# of male patients

# harboring TMPRSS-ERG

%

715 51 40 19 236 939 267 697 320 39 663 89 244 830

214 18 11 7 9 8 5 6 3 2 2 1 1 1

30 35 28 37 3.8 0.9 1.9 0.9 0.9 5.1 0.3 1.1 0.4 0.1

Ca-carcinoma; ADC-adenoCa; NE-neuroendocrine; UP-unknown primary; SCC-squamous cell cancer; NOS-not otherwise specified.

5038

Poster Session (Board #295), Sat, 1:00 PM-4:30 PM

Prostate cancer enhanced mRNA detection assay in whole blood as predictive biomarker of tumor sensitivity to targeted androgen inhibition for men with metastatic castration-resistant prostate cancer (mCRPC). First Author: Daniel Costin Danila, Memorial Sloan Kettering Cancer Center and Weill Cornell Medical College, New York, NY Background: Enzalutamide (ENZ) and abiraterone + prednisone (AA) have transformed the care of pts with mCRPC. Qualified predictive biomarkers are needed to guide treatment selections. Studies of androgen receptor (AR) splice variants, especially ARv7, may predict resistance to ENZ and AA. We report on analytical validation and testing of whole blood ARv7 assay for pts with mCRPC starting AR-targeted therapies. Methods: Men with progressive mCRPC were enrolled prospectively on a biomarker protocol to collect blood in PAXgene tubes before starting treatment with ENZ or AA. An RTPCR based assay to detect KLK2, KLK3, HOXB13, GRHL2, FOXA1 and ARv7 in whole blood was analytically validated in 2 independent laboratories. VCaP spikes established the limit of detection, and healthy volunteers were used as controls. Samples were independently tested with the validated signature. Results: A standard operating protocol was generated and applied to 45pts starting ENZ (AA- and chemo- na¨ıve) and 33pts starting AA (prior ENZ .6 months; 22 pts were chemo-na¨ıve; 8 pts progressed on ENZ cohort). ARv7 detection trended at higher levels in pts with prior ENZ exposure (Ct=26) compared to patients starting ENZ (Ct=30.5). Conclusions: In our study, ARv7+ve pts starting AA after ENZ had ,50% PSA declines, but did not predict de novo resistance to ENZ. Further analysis of ARv7 status as a predictive biomarker, in parallel with other biomarkers, is ongoing. PSA response from baseline

ARv7 status

N (%, 95%CI)

‡50%PSA decline N (%, 95%CI)

True responder (‡50% max decline)

Acquired resistance (rise after ‡50% max decline)

De novo resistance (50 to £70 yrs n=49

>70 yrs n=25

66.7 20 13.3 26.7 0 33.3 0 6.7 20

67.3 34.7 32.7 32.7 20.4 28.6 14.3 2 16.3

60 44 44 36 32 16 16 8 4

Poster Discussion Session; Displayed in Poster Session (Board #338), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Salivary duct carcinoma: Clinical outcomes and prognostic factors in 157 patients and results of androgen deprivation therapy in recurrent disease (n=31)—Study of the Dutch head and neck society (DHNS). First Author: Eline Boon, Radboud University Medical Center Nijmegen, Nijmegen, Netherlands Background: Salivary Duct Carcinoma (SDC) is a rare, highly aggressive subtype of salivary gland cancer, which is androgen receptor (AR) positive in most pts. In recurrent disease androgen deprivation therapy (ADT) can be used, however only limited data are available. The aim of this study was to evaluate clinical outcome, to identify prognostic factors for overall survival (OS) and to retrospectively analyze efficacy of ADT. Methods: The nationwide network and registry of histo- and cytopathology (PALGA) identified all adult pts with SDC between 1990 - 2014 in the Netherlands. Lymph node ratio (LNR) was defined as the number of positive LN divided by the total number of LN after neck dissection. Survival analyses were conducted using Kaplan-Meier method. The Cox proportional hazard model was used to identify prognostic factors on uni- and multivariable analyses. Results: A total of 157 pts with SDC were identified. Median age was 66 years, 77% were male. AR and HER2neu were positive in 100% and 44%, respectively. Median number of positive LN was 4 (range 0–97) and median LNR was 0.20. Median OS was 44 months (95% CI 30-48). Multivariable analyses identified male gender, high N-stage and high LNR as significant poor prognostic factors for OS. After median 14 months (range 1-87) 84 pts (54%) had recurrent disease: local recurrence in n = 19 (23%), regional recurrence in n = 31 (37%) and distant metastases in n = 72 (86%). Only 12 (14%) of recurrent pts had local and/or regional recurrence without distant metastases. 31 pts with recurrent SDC were treated with ADT: 4 pts had PR, 10 pts had SD and 17 pts had PD. Median PFS of all ADT treated pts was 3.8 months (range 1-25) and median OS 4 months (range 1–64). In pts with PR or SD (45%) median PFS was 11 months (range 3-25) and median OS was 29 months (range 3-64). Conclusions: More than half of the SDC pts developed recurrent disease. Male gender, high N-stage and high LNR were correlated with worse OS. In this largest series of ADT treated SDC pts, we found clinical benefit (PR + SD) in 45% of pts for a median duration of 11 months and median OS of 29 months. We recommend considering ADT treatment in recurrent SDC.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Head and Neck Cancer 6017

Poster Discussion Session; Displayed in Poster Session (Board #339), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

6018

321s

Poster Discussion Session; Displayed in Poster Session (Board #340), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Preliminary results for the advanced salivary gland carcinoma cohort of the phase 1b KEYNOTE-028 study of pembrolizumab. First Author: Roger B. Cohen, University of Pennsylvania, Philadelphia, PA

Phase I extension clinical study of BB503, a first-in-class cancer stemness kinase inhibitor, in adult patients with advanced head and neck cancer. First Author: Gregory Michael Cote, Massachusetts General Hospital, Boston, MA

Background: Patients (pts) with treatment-refractory salivary gland cancer have limited treatment options. The PD-1 pathway is upregulated in many tumors leading to immune response suppression. Pembrolizumab (pembro), an anti–PD-1 antibody, blocks interaction between PD-1 and PD-L1 and PDL2. The nonrandomized KEYNOTE-028 (NCT02054806) trial evaluated the safety and efficacy of pembro in 20 advanced solid tumor cohorts. Results from the salivary gland carcinoma cohort are presented. Methods: Key inclusion criteria included advanced (unresectable and/or metastatic) salivary gland carcinoma excluding sarcomas and mesenchymal tumors, failure of systemic therapy, ECOG PS 0-1, and PD-L1 expression in $ 1% of tumor or stroma cells by IHC. Pts received pembro 10 mg/kg every 2 wk for up to 24 mo or until confirmed progression, intolerable toxicity, death or withdrawal of consent. Response assessed every 8 wk for the first 6 mo and every 12 wk thereafter. The primary end point was ORR per RECIST v1.1 by investigator review. Results: As of Dec 10, 2015, 26 pts were enrolled in this cohort. Median age was 57.0 y; 88.5% were male; 73.1% had an ECOG PS of 1; 73.1% had received prior therapy for metastatic disease. At data cutoff, median follow-up duration was 61.9 wk (range, 8.7-88.4). Treatmentrelated adverse events (TRAEs) occurred in 22 (84.6%) pts; TRAEs in $ 15% of pts were diarrhea (n = 4), decreased appetite (n = 4), pruritus (n = 4), and fatigue (n = 8). 3 (11.5%) pts had grade 3-5 TRAEs which resulted in 1 death (interstitial lung disease); 2 (7.7%) pts discontinued pembro because of a TRAE (grade 2 arthritis; grade 3 hepatitis). Confirmed ORR was 11.5% (PR, n = 3; 95% CI, 2.4-30.2); median duration of response was 17.0 wk (range, 15.1-53.4+). Pts achieving PR had adenocarcinoma (n = 2) and high-grade serous carcinoma (n = 1). Stable disease rate was 46.2% (n = 12; 95% CI, 26.6-66.6). The 6-mo OS rate was 70.4%; the 6-mo PFS rate was 20.7%. At data cutoff, 2 pts were on treatment. Conclusions: In this heavily pretreated population, pembro showed promising antitumor activity. Clinical benefit of pembrolizumab in advanced salivary gland carcinoma will be further investigated in the phase 2 KEYNOTE-158 trial (NCT02628067). Clinical trial information: NCT02054806.

Background: BB503 (BBI-503) is an oral first-in-class cancer stemness kinase inhibitor. By targeting multiple serine-threonine stemness kinases, BB503 inhibits Nanog and other cancer stemness pathways. A Phase I clinical trial of BB503 demonstrated safety and signs of anti-cancer activity in patients (pts) with advanced solid tumors. Dose escalation and RP2D expansion accrual included a cohort for pts with refractory head and neck cancer (HNCA). Methods: Pts with metastatic, unresectable HNCA who had exhausted standard therapies were enrolled. BB503 was administered orally, daily, in continuous 28-day cycles. Pts were assessed for safety and preliminary activity including Disease Control Rate (DCR), defined as stable disease of at least 8 weeks or objective partial (PR) or complete response (CR) per RECIST 1.1. Results: 33 pts were enrolled: 13 with adenoid cystic carcinoma (ACC), 15 with head and neck squamous cell carcinoma (HNSCC), and 5 with parotid/salivary tumors. All ACC pts received prior radiation (XRT), 92% prior surgery, and 54% prior systemic therapy. For HNSCC and parotid/salivary pts, 90% received prior XRT, 80% prior surgery, and 90% prior systemic therapy. Pts were treated with BB503 10 mg to 300 mg daily, which was well tolerated. Grade 3 AE included rapidly reversible diarrhea (n = 5) and nausea/vomiting (n = 1). For evaluable pts with HNSCC and parotid/salivary tumors (n = 16), 19% achieved PR and 25% achieved disease control for $ 24 weeks. The 4 non-evaluable pts had clinical deterioration prior to receiving $80% of 1 cycle of study drug per protocol. For ITT HNSCC or parotid/salivary pts (n = 20), median overall survival (mOS) is 35 weeks. For all ACC pts (ITT, n = 13), 38% had disease control for $ 24 weeks, 69% have survived more than 1 year, and mOS is not reached. Conclusions: BB503 demonstrated tolerability and encouraging signs of anti-cancer activity in pts with refractory HNCA, including partial response and prolonged disease control. Further clinical investigation of BBI503 for HNCA is warranted. Clinical trial information: NCT01781455.

6019

6020

Poster Discussion Session; Displayed in Poster Session (Board #341), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Clinical relevance of lymph node ratio in resected oral cavity squamous cell carcinoma in patients with N2 disease. First Author: Ali Hosni, Princess Margaret Cancer Centre, Toronto, ON, Canada Background: Lymph node ratio (LNR, number of positive lymph nodes/total number of excised lymph nodes) has been shown to be associated with outcomes in multiple malignancies. In this study, the impact of LNR on distant metastasis (DM) and overall survival (OS) in oral cavity squamous cell carcinoma (OSCC) was investigated. Methods: Retrospective review of pN02 OSCC patients (pts) treated between 1994-2012 with curative surgery with neck dissection (ND) +/- postoperative radiotherapy (PORT) with or without concurrent chemotherapy (CCT). LNR was subjected to multivariable analysis (MVA) of DM and OS, adjusted for pT3-4, extracapsular extension (ECE), high grade (G3), lymphovascular invasion (LVI), perineural invasion (PNI), and tumor subsite. Results: Overall 914 pts were identified; median age: 61 yr (18-92); median follow-up: 51 months (1–189); pT3-4: 283 (31%); pN-classification: N0: 482 (53%), N1: 128 (14%), N2a: 6 (0.5%); N2b: 225 (24.5%); N2c: 73 (8%); median number of dissected nodes: 36 (6-125); median number of pN+: 2 (1-49); median LNR for pN+ pts: 0.06; ECE: 187 (20%); G3: 147 (16%); LVI: 115 (13%); PNI: 416 (46%). Bilateral ND was performed in 367 (40%); PORT was used in 452 (49%); and CCT in 80 (9%). The 5-yr distant control (DC) and OS were 89% and 70%; respectively. pT3-4 (p,0.001), G3 (p,0.001), LVI (p=0.0013), and PNI (p,0.001) were all associated with high LNR. On MVA, LNR .0.06 was associated with more DM (HR=1.8; 95%CI=1.1-3.1; p=0.017) and lower OS (HR=2; 95%CI=1.4–2.8; p,0.001). In subgroup analysis of pN2 pts (n=304): higher LNR (.0.14) was associated with lower 5yr- DC (67%, p=0.014) and OS (25%, p,0.001), and on MVA within the pN2 subgroup, both higher LNR (p,0.001) and ECE (p=0.006) were associated with lower OS. Conclusions: High LNR is associated with higher rate of DM and lower OS in OSCC. LNR should be assessed in pN2 pts in future prospective trials to select patients for adjuvant therapies.

Poster Discussion Session; Displayed in Poster Session (Board #342), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Does addition of neck ultrasonography to physical examination, in follow-up of patients with early stage, clinically node negative oral cancers, influence outcome? A randomized control trial (RCT). First Author: Anil D’Cruz, Tata Memorial Hospital, Mumbai, India Background: We previously reported results of a RCT demonstrating superiority of elective neck dissection (END) over therapeutic neck dissection (TND) in early oral cancer patients with clinically node negative neck1. We now report the results of the second question in this trial –elucidating the role of ultrasonography (USG) during follow-up, based on the hypothesis that addition of imaging would detect neck nodes earlier and improve salvage and survival (NCT00193765). Methods: Patients with lateralized T1 or T2 squamous carcinoma of oral cavity after initial surgery (excision of primary with/without neck dissection) were randomized to follow up with physical examination plus neck ultrasound (PE+USG) performed in a standardized manner versus physical examination (PE). The primary end point was overall survival (OS)1. Results: This report includes results of 500 patients in the same cohort reported earlier1with a median follow up of 39 months; 252 patients were randomized to PE+USG and 244 to PE. Both arms were balanced for stratification factors and surgical procedure (END vs. TND). There were 118 recurrences with 67 deaths in PE+USG and 109 recurrences with 62 deaths in PE, respectively. There was no OS difference between PE +USG and PE in unadjusted analysis (3-year OS 73.3% and 73.8%, respectively, HR = 1.02, 95%CI 0.73-1.45, p = 0.89) and after adjustment (HR = 0.81, 95%CI 0.51-1.29, p = 0.37) for stratification factors, prognostic factors, surgical treatment (END vs. TND) and an interaction term between two study questions, in a Cox model. END vs. TND continued to be highly significant for OS in this model (HR = 0.54, 95%CI 0.32 - 0.92, p = 0.02). Within TND (wait and watch) arm there was no significant difference between PE+USG and PE (3-year OS 67.3% and 67.6% respectively, HR = 0.96, 95%CI 0.62–1.5, p = 0.86). Conclusions: Neck ultrasound confers no survival advantage over physical examination in postoperative follow-up of clinically node negative early stage oral cancer patients. 1. N Engl J Med 2015;373:521-9 Clinical trial information: NCT00193765.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

322s 6021

Head and Neck Cancer Poster Discussion Session; Displayed in Poster Session (Board #343), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

6022

Poster Discussion Session; Displayed in Poster Session (Board #344), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

ECLYPS: Multicenter trial of FDG-PET/CT to detect residual nodal disease in locally advanced head-and-neck squamous cell carcinoma (LAHNSCC) after chemoradiotherapy (CRT). First Author: Tim Van Den Wyngaert, Antwerp University Hospital, Edegem, Belgium

Individual patient data meta-analysis of the IMPACT of treatment site on the outcome of head and neck cancer patients treated within 6 randomized trials. First Author: Paolo Bossi, Head and Neck Cancer Medical Oncology Unit, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy

Background: FDG-PET/CT surveillance after CRT in LAHNSCC results in non-inferior overall survival (OS) compared to routine neck dissection (ND) (Mehanna H, ASCO 2015). However, optimal imaging frequency, interpretation in p16+ patients (pts), and value of image quantification remain uncertain. Methods: Prospective multicenter trial of FDG-PET/CT 12wks after end of CRT in pts with newly diagnosed LAHNSCC (stage IVa/b), with scanner (cross-)calibration, standardized image reconstruction, and blinded central review (Hopkins criteria). Reference standard was histology or . 1y negative clinical follow-up after imaging. The primary end-point was the negative predictive value (NPV) for nodal recurrence (target . 90% with 95% CI excluding 85%). Secondary analyses assessed time-dependency of test characteristics, value of quantitative SUV assessment, and impact of p16 status. Results: Of 152 pts recruited in 7 centers, 125 had adequate primary tumor control after CRT and entered follow-up (median 20.4mos), during which 23 (18.4%) neck recurrences occurred (ND rate 16%; 2y OS 85.9%). Overall NPV of FDG-PET/CT was 92.1% (95% CI 86.9 – 95.3; H0 : NPV = 85% p= 0.012), with sensitivity 65.2% (95% CI 44.9 – 81.2) and specificity 91.2% (95% CI 84.1 – 95.3). Sensitivity was time-dependent, decreasing from 83.3% at 3mos to 59.7% at 12mos after imaging (p= 0.04). Specificity was stable over time. Visual FDG-PET/CT analysis was prognostic for OS (HR 11.27; p, 0.0001) and time to nodal recurrence (SHR 12.36; p, 0.0001), as was p16+ status (HR 0.16 and SHR 0.34; both p= 0.02). Quantitative assessment (SUV70 ) was superior to local read (p= 0.01), but not Hopkins criteria. For neck recurrences , 6mos after imaging, SUV70 in p16+ pts was higher compared to p16- pts, affecting optimal SUV70 cut-off values. Conclusions: FDG-PET/CT can rule out neck disease with high probability 12 wks after CRT in LAHNSCC, but sensitivity was lower for late recurrences, which may warrant an additional surveillance scan. SUV analysis outperformed local visual read, but not Hopkins criteria. Optimal absolute SUV thresholds may depend on p16 status. Clinical trial information: NCT01179360.

Background: The impact of centre expertise on outcome of head and neck cancer (HNC) patients (pts) treated with chemotherapy (CT) +/- radiotherapy (RT) has been sparsely evaluated. Pts selection, RT quality, compliance with full dose administration, kind and extent of supportive care implementation may all explain possible benefits of being treated in highly experienced centers. Methods: We performed an individual patient data meta-analysis of 6 randomized trials in HNC conducted in Italy between 1983 and 2001. Different RT approaches (curative, postoperative), fractionations (conventional, altered) and combinations with CT (sequential, alternated, concomitant) were administered. The main aim was the association between center type (coordinating site- CS - vs non-coordinating site - NCS) and overall survival (OS), progression-free survival (PFS) and death within 3 months since treatment start (3MD). In univariable analyses center type was the main investigated variable and the trial was modeled as a random factor. Multivariable covariates analysed were: site of disease, performance status, T and N stage, treatment type; interaction between CS, NCS and treatment was also considered. Results: Globally, the outcome of 1022 patients was analyzed. Median follow up was 76 months (48-224). CS enrolled a higher median number of pts in respect to NCS (178 vs 115). In univariable analysis, NCS showed a higher risk of 3MD (OR 1.82; 1.093.04). In multivariable analysis, no difference was observed in OS, PFS and 3MD between CS and NCS; models with interaction centre-treatment showed that pts treated with RT and CT in NCS had a significantly worse OS (HR 1.47; 1.08-2.02) and PFS (HR 1.54; 1.09-2.18). On the contrary no difference on outcome was identified with conventional/altered RT as single treatment. Conclusions: In this pooled analyses of several trials on advanced HNC, we saw that OS was significantly better at CS as compared to NCS when treatment consisted of combination of CT and RT. Our study adds to available evidence suggesting that high case volumes correlate with clinical outcome in HNC. In clinical trials on RT and CT, case volume may need to be a stratifying factor.

6024

Poster Discussion Session; Displayed in Poster Session (Board #346), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

Predicting overall survival (OS) and progression-free (PFS) for oropharynx cancers (OPC) in NRG Oncology RTOG0129/0522 with nomograms. First Author: Carole Fakhry, Johns Hopkins University, Baltimore, MD

ABSTRACT WITHDRAWN

Background: Therapeutic de-intensification for OPC underscores the need for accurate patient-specific assessment of risk. Current risk stratification is based on HPV tumor status, tobacco use, nodal and tumor stage. These do not include other important predictors of prognosis. Therefore, a patientspecific approach to estimate survival was explored. Methods: Patients with T2-T4, N0-N3 OPC enrolled in RTOG0129 or 0522, available p16 immunohistochemistry (surrogate for HPV) and tobacco history were eligible for retrospective analysis of factors associated with OS and PFS. RTOG0129 compared standard versus accelerated fractionation (AFX) cisplatin based chemoradiation, and RTOG0522 compared cisplatin-AFX 6 cetuximab. Nomograms were created based on Cox proportional hazards regression models. Concordance index was used for model discrimination and validation. Results: Study population included 493 patients. Factors independently associated with worse OS included age . 50, poor performance status (Zubrod 1), p16 negativity, # high school education, . 10 pack-years tobacco, anemia, advanced tumor (T4) and nodal stage (N2c-3). Significant interactions were observed including p16 and performance status. All these factors, along with marital status and gender were included into a nomogram to estimate 2- and 5-year OS based upon the absence or presence of each of these patient characteristics. Factors significantly associated with PFS were similar, with the addition of weight loss $ 5% and marital status. A nomogram that accounts for the relative contribution of each of these prognostic factors provides 2- and 5-year estimates of PFS. Both nomograms had high concordance indices (0.76 for OS, 0.70 for PFS) and were well calibrated. Conclusions: The developed nomograms incorporate an expanded list of prognostic factors and allow for individual patient-specific estimates of OS and PFS. Use of these nomograms will facilitate decision-making for patients and providers and will refine patient selection criteria for deintensification trials.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Head and Neck Cancer 6025

Poster Session (Board #347), Sat, 1:00 PM-4:30 PM

6026

323s Poster Session (Board #348), Sat, 1:00 PM-4:30 PM

Final results of the randomized phase II DeLOS-II trial: Induction chemotherapy (IC) followed by radiotherapy (R) vs. cetuximab (E) plus IC and R for functional larynx preservation in resectable laryngeal and hypopharyngeal cancer (LHSCC). First Author: Andreas Dietz, University of Leipzig, Leipzig, Germany

Phase II trial of combination treatment with paclitaxel, carboplatin and cetuximab (PCE) as first-line treatment in patients with recurrent and/or metastatic squamous cell carcinoma of the head and neck (CSPOR-HN02). First Author: Makoto Tahara, National Cancer Center Hospital East, Chiba, Japan

Background: To evaluate the impact of E added throughout IC and R on the rate of twoyear functional laryngectomy free survival (fLFS). Methods: Untreated patients (pts) with stage III/IV LHSCC, amenable to resection by total laryngectomy, were randomized to three cycles of IC with TPF (docetaxel and cisplatin 75 mg/m² day 1 and F 750 mg/m²/ day on days 1-5) followed by R (69,6Gy) without (arm A) or with (arm B) standard dose of cetuximab for 16 weeks throughout IC and R. In case of non-response after the first IC cycle, salvage laryngectomy was performed. The primary objective was a 2-year fLFS rate above 35% (lower bound of 80%-CI) in the experimental arm. Results: Of180 pts randomized (7.2007-9.2012), 173 fulfilled ITT criteria. Due to 4 therapy related deaths among the first 64 randomized pts, F was omitted from IC in 2.2009 with no further treatment related deaths in the remaining 112 pts. Table 1 summarizes rates of response, laryngectomy, fLFS and OS. The primary objective of an fLFS above 35% was met, however, the control arm also met objectives. 31.8% (arm A) and 22.7% (arm B) of patients did not respond to first cycle IC. Conclusions: IC with TPFE/TPE was feasible and showed a trend towards being more effective compared to TPF/TP in terms of rates of fLFS and OS. The early response rate to IC of 77.3% and the 1 and 2 year fLFS rates of 63.6% and 46.6%, respectively compare very well to previous larynx preservation trials and suggest effective treatment selection and stratification by early response evaluation to IC. Clinical trial information: NCT00508664.

Background: The standard of care for first-line treatment of recurrent and/or metastatic squamous cell carcinoma of the head and neck (R/M SCCHN) is combination treatment with cisplatin, 5-FU and cetuximab (PFE). However, this regimen requires hospitalization to ensure proper hydration and continuous infusion of 5-FU, and causes severe nausea and anorexia. We evaluated the efficacy and safety of paclitaxel, carboplatin and cetuximab (PCE) as first-line treatment in patients with R/M SCCHN. Methods: Eligibility included recurrent and/or metastatic, histologically proven SCC of the oropharynx, oral cavity, hypopharynx or larynx; PS 0-1; adequate organ function; no suitable local therapy for R/M SCCHN; and no prior systemic chemotherapy for R/M SCCHN. Chemotherapy consisted of paclitaxel 100mg/m2 on day 1, 8; carboplatin AUC 2.5 on day 1, 8, repeated every 3 weeks for up to 6 cycles; and cetuximab at an initial dose of 400mg/ m2, followed by 250mg/m2weekly until disease progression or unacceptable toxicities. Primary endpoint was overall response rate (ORR). Secondary endpoints were safety, treatment completion rate, progression-free survival, overall survival, and clinical benefit rate (CBR). Planned sample size was 45 patients. Results: Forty-seven subjects were accrued from July 2013 and Oct 2014. Of 45 evaluable, 40 were male; median age was 63 years; ECOG PS was 0/1 in 23/22 cases; site was the hypopharynx/oropharynx/oral cavity/ larynx in 17/11/10/7 cases; and 36/9 cases were smokers/non-smokers, respectively. ORR was 37.8% (95% CI 30.9–48.6). Median overall survival was 14.7 months (95% CI, 12.1–Not Reached) and progression-free survival was 5.2 months (95% CI 3.9–5.6). Grade 3 or 4 adverse events included neutropenia (68%), skin reaction (15%), fatigue (9%) and febrile neutropenia (9%). A potentially treatment-related death occurred in one patient with intestinal pneumonia. Conclusions: The PCE regimen shows promising activity with acceptable toxicity and can be provided in the outpatient clinic. Further studies are needed to compare PCE with PFE in this population. Clinical trial information: UMIN000010507.

Efficacy and outcome data.

Localisation: Larynx Hypopharynx TPF TP ORR* (n = 126) TPF TP Laryngectomy fLFS at 6 months fLFS at 1 year fLFS at 2 years OS at 2 years

Arm A TP(F) – > R

Arm B TP(F) – > R plus cetuximab

: 44 41 30 55 : 94.7% 79.1% 25.9% 65.9% 55.3% 44.7% 68.2%

: 42 46 31 57 : 80.0% 94.9% 33.0% 77.3% 63.6% 46.6% 69.3%

OR (95%CI)

0.5680 0.7068 0.9268 0.9508

(0.2905-1.1105) (0.3842-1.3001) (0.5094-1.6863) (0.4997-1.8091)

OR: Odds Ratio; *at final examination

6027

Poster Session (Board #349), Sat, 1:00 PM-4:30 PM

6028

Poster Session (Board #350), Sat, 1:00 PM-4:30 PM

Phase II trial of carboplatin/paclitaxel and cetuximab, followed by carboplatin/paclitaxel/cetuximab and erlotinib, in metastatic or recurrent squamous cell carcinoma of the head and neck. First Author: Aarti K. Bhatia, Yale School of Medicine & Yale Cancer Center, New Haven, CT

Clinical outcome after multicenter, open-label phase II trial on post-surgery chemoradiation in combination with cetuximab in squamous cell carcinoma of the head and neck with high risk of locoregional recurrence. First Author: Christiane Matuschek, University Hospital, Duesseldorf, Germany

Background: Cetuximab (C) improves outcomes when added to chemotherapy in patients with squamous cell carcinoma of the head and neck (SCCHN). C resistance is associated with nuclear localization of the epidermal growth factor receptor (EGFR). We hypothesized that an EGFR tyrosine kinase inhibitor (TKI) would improve response to C through inhibition of nuclear translocation of EGFR and undertook a Phase II study of total EGFR blockade, adding Erlotinib (E) to chemotherapy and C in patients with recurrent/metastatic SCCHN. Methods: 24 patients were enrolled. Following baseline tumor assessment and biopsy, they were treated with carboplatin (AUC 2 weekly)/paclitaxel (80mg/m2 weekly) and C (400mg/m2 loading, then 250mg/m2weekly thereafter). After the first 21-day cycle, biopsy was repeated, then therapy resumed with the addition of E (150mg daily x 2 days followed by 100mg daily) then a repeat biopsy after cycle 2. Treatment continued until disease progression or unacceptable toxicity. Restaging was done prior to cycles 2 and 4 and every 9 weeks thereafter. Primary end point was objective response rate (ORR) by RECIST v 1.1 – complete (CR) and partial (PR) responses. Secondary end points: toxicity, overall survival (OS) and laboratory correlates. Results: Median age for the group was 65.5 years. Patients received a median of 6 (range 1-17) cycles. 12/24 (50%) had objective responses (12 PR, 0 CR). 1 patient had a pathological complete response and 2 were consolidated with definitive radiation with no recurrence at 9 and 41 months. 7/24 (29%) had stable disease. For intent to treat population, median progression-free survival (PFS) and OS were 5.8 and 8.1 months, while for per protocol population, median OS was 10.6 months. 5 patients are alive at the time of analysis. Dose modification for toxicity was required in 50% of patients. Main toxicities were cytopenia, diarrhea, neuropathy, hypomagnesemia and rash. Conclusions: Total EGFR blockade demonstrated excellent efficacy and tolerability in this small sample. ORR, PFS, OS are comparable to historical controls treated with the EXTREME regimen. Correlative analyses are ongoing. Clinical trial information: NCT01316757.

Background: We performed an open label phase 2 trial and investigated the efficacy and toxicity of additional cetuximab during and after adjuvant radiochemotherapy (RCT) in head and neck cancer. Methods: 83 Patients with SCCHN were eligible for the study. 61.6 Gy (1.8/2.0/2.2Gy, d1-36) were administered using an integrated boost IMRT-technique. Cis-DDP (20 mg/m2, d1-5 and d29-33) and 5-FU (continuous infusion: 600 mg/m2, d1d5 + d29-33) were given concurrently. Cetuximab started 7 days prior to RCT at 400 mg/m2 followed by weekly doses of 250 mg/m2. Maintenance cetuximab began after RCT at 500 mg/m² every 2 weeks for six months. The study was conducted at 10 investigational sites recruiting in Germany from May 2008- December 2010. Results: The number of patients enrolled was 83. Five patients did not meet inclusion criteria leaving 78 for analysis. Median follow up was 1.5 years. 18 patients had events (death or progression) of which 10 were locoregional relapses. 2 year overall survival, disease free survival, and locoregional tumor control were 86% (95%CL 7995%), 77% (95% CL 66-86%), and 82% (95%CL 93-78%). A total number of 1542 AEs including 196 SAEs were documented. No formerly unknown toxicities were observed. One patient died during RCT. The death was considered to be treatment related, but not related to cetuximab. Acute grade 3+4 toxicities not taking into account skin toxicity outside the radiation portals were observed in 58% of patients, which was below the predefined rate of 67% that was considered as maximally tolerable in these patients. With the expected exception of skin toxicity outside the radiation portals, most grade 3+4 toxcities regress within 6 weeks after RCT. Grade 3 +4 late toxicities ( . day 90) were observed in 34% of patients. 20% of patients refuse further treatment with cetuximab after completion of RCT. Conclusions: Adjuvant RCT with concomitant and maintenance cetuximab is feasible and results in a favorable clinical outcome in high risk SCCHNC. Clinical trial information: EudraCT No. 2007-002659-17.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

324s

Head and Neck Cancer

6029

Poster Session (Board #351), Sat, 1:00 PM-4:30 PM

Interim results from a phase II study of CC-486 in previously treated patients (pts) with locally advanced/metastatic nasopharyngeal cancer (NPC). First Author: Ricard Mesia, Medical Oncology Department, Institut Catala` d’Oncologia (ICO) L’Hospitalet, University of Barcelona, Barcelona, Spain

6030

Poster Session (Board #352), Sat, 1:00 PM-4:30 PM

Impact of molecular prescreening for genomically-guided trials in head and neck cancer (HNC). First Author: Neus Baste, Medical Oncology Department, Vall d’Hebron University Hospital, Barcelona, Spain

Background: For pts with advanced NPC, no approved chemotherapy is available in the second-line setting or beyond, and treatment options are limited. Hypermethylation contributes to the initiation and progression of NPC; thus, epigenetic modifying agents may be a useful therapeutic strategy. Single-agent CC-486, an oral formulation of azacitidine, has shown promising results in pts with relapsed/refractory solid tumors, including NPC. Here we report interim results from the phase II study of CC-486 in previously treated pts with recurrent locally advanced/metastatic NPC. Methods: Pts with locally advanced/metastatic, progressive NPC and 1 to 2 previous treatments (including a platinum-containing regimen) received oral CC-486 300 mg/day (d) d 1-14 of a 21-d cycle. In addition, the first 6 Asian-Pacific (AP) pts received CC-486 200 mg/d d 1-14 of cycle 1; if well tolerated, subsequent AP pts received 300 mg/d as above. Pts were treated until disease progression or unacceptable toxicity. The primary endpoints were independently assessed overall response rate and progression-free survival. Secondary endpoints included overall survival, disease control rate, safety, and pharmacokinetic analyses in AP pts. Results: As of Nov 12, 2015, 36 pts were enrolled and received CC-486 (6 pts at 200 mg and 30 pts at 300 mg). The median age was 54 y, and most pts were men (78%) with an ECOG PS of 1 (58%); 36% were Asian. Of 10 pts evaluable for efficacy, 3 (30%) had an unconfirmed partial response, and 2 (20%) had stable disease by independent review ( $ 16 weeks). In 34 pts evaluable for safety, grade 3/ 4 adverse events (AEs) were hematologic and included mainly neutropenia (3 pts at 200 and 4 pts at 300 mg). Grade 2 nonhematologic AEs included vomiting (1 pt at 200 mg and 8 pts at 300 mg) and nausea (3 pts each at 200 and 300 mg). Dose reductions due to AEs occurred in 3 and 5 pts at the 200 and 300 mg dose levels, respectively. 26 pts remain on treatment. Conclusions: Single-agent CC-486 was well tolerated in pts with recurrent locally advanced/metastatic NPC; grade 3/4 AEs were mainly hematologic. Initial efficacy results are encouraging. More follow-up data will be presented at the meeting. Clinical trial information: NCT02269943.

Background: HNC encompass a group of highly heterogeneous diseases, including oral, oropharyngeal and laryngeal squamous cell carcinomas (SCC), nasopharyngeal (NP), nasosinusal (NS) and salivary gland (SG) tumors. Chemotherapy has shown limited efficacy in recurrent (rec)/metastatic (met) disease. Genomic studies have increased our understanding of HNC biology, identifying potential targetable mutations (mut). We evaluated the impact of precision medicine efforts in HNC at Vall d’Hebron Institute of Oncology (VHIO). Methods: From 10/2010 to 10/2015, 49 patients (pts) with metastatic HNC underwent tumor molecular profiling at VHIO. Mut detection was performed using multiplex mass-spectrometry (Sequenom, n = 21) or NGS (AmpliconSeq, n = 28).Profiling was performed on metastatic (n = 21) or primary sites (n = 19), with missing information in remaining cases (n = 9). Results: Most frequent tumor type was SCC (n = 20), followed by SG (n = 18), NP (n = 9) and NS (n = 2). Median age at diagnosis was 51 years, 55% stage IV. Rec/met disease to nodes or lungs was seen in 95% of SCC, while liver met was detected in 33% NP and 39% SG. PIK3CA (5/49) and CDKN2A mut (5/28) were found across all subtypes. TP53 mut were enriched in SCC (8/9, p , 0.001). SG had PTEN (1/10), NF1 (1/3), HRAS (3/18), ERBB3 (1/10) and NRAS (1/18) mut. One NP harbored NRAS mut (1/9). The prevalence of targetable mut varied according to tumor type (44% SG, 15% SCC, 36% N/NS; p = 0.03). In total, 18/49 pts were enrolled into 26 phase 1 trials (ph1), 37% accrual rate. Median prior treatment lines were 2 (0-5). 9/26 ph1 were matched therapies (mTher): 6 in SG (PIK3CA, PTEN or ERBB3 mut), 2 in SCC (CDKN2A or PTEN mut) and 1 in NP (NRAS mut). Stable disease was observed in 5 SG treated with PI3K pathway inhibitors (inh), and 1 partial response in the NP treated with PI3K plus MEK inh. Considering pts as their own controls, median time to treatment failure (TTF) during mTher (9 months, range 3-23) was similar to TTF while on prior first / second treatment line (6.9 months, range 4-110; p = 0.68). Conclusions: Genomic characterization of HNC identifies pts likely to benefit from a mTher in ph1, with promising targets being PI3K pathway mut in SG and NRAS mut in NP.

6031

6032

Poster Session (Board #353), Sat, 1:00 PM-4:30 PM

Systemic therapies for recurrent/metastatic nasopharyngeal carcinoma (RM NPC). First Author: Amy Prawira, Princess Margaret Cancer Centre, Toronto, ON, Canada Background: No randomized trials exist to guide clinical trial designs in RM NPC. As all published studies are single-arm trials, reliable historical progression-free survival (PFS) and overall survival (OS) data are not available. We aim to analyze existent literature to estimate the relative efficacy of available systemic regimens in RM NPC, as well as provide estimates of historical PFS and OS. Methods: We conducted a systematic search of MEDLINE, EMBASE and the Cochrane Library to March 2015. Clinical trials (in English only) investigating systemic therapies in adult patients (pts) with RM NPC were included. All relevant studies were assessed for quality using Downs and Blacks (DB) checklist (scored out of 27). Aggregate data analysis and student t-test were performed for all identified studies (model A). For studies that published analyzable Kaplan-Meier (KM) curves, survival data were extracted and marginal proportional hazards models were constructed (model B). Results: A total of 56 studies were identified and included in model A, 26 of which had analyzable KM curves and were included in model B. The 26 studies had significantly higher mean DB scores than the remaining 30 (17.3 [95% CI 15.5-19.0] vs 13.7 [95% CI 12.2-15.1], p = 0.002). For 2nd line+ pts, the estimated median OS [mOS] by model A is 11.5 mo (95% CI 10.1-12.9), while model B estimated this at 12.5 mo (95% CI 11.9-13.4). Table 1 presents the relative efficacy for first-line [FL] treatment regimens: combination [C] vs single-agent [S], and platinum [P] vs nonplatinum [NP] based therapies. Conclusions: Our analysis supports the use of platinumbased combination chemotherapy in the first-line setting. We present the first aggregate estimates of OS for patients in 2nd line+ setting, which could inform the design of future clinical trials for RM NPC. Model A/mo (95% CI) FL C mPFS FL S FL P mPFS FL NP FL C mOS FL S FL P mOS FL NP

8.4 (6.9-9.8) 3.5 (1.1-5.9) 8.3 (7.0-9.6) 3.5 (1.1-5.9) 17.8 (14.2-21.4) 8.2 (0.0-16.7) 17.4 (13.9-20.8) 8.2 (0.0-16.7)

P-value

Model B/HR (95% CI)

P-value

0.007

0.48 (0.41-0.56)

, 0.0001

0.007

0.48 (0.41-0.56)

, 0.0001

0.020

1.16 (0.98-1.38)

0.084

0.023

1.16 (0.98-1.38)

0.080

Poster Session (Board #354), Sat, 1:00 PM-4:30 PM

Organ preservation with daily concurrent chemoradiotherapy using retrograde superselective intra-arterial infusion for locally advanced tongue cancer: analysis of therapeutic results in 101 cases. First Author: Kenji Mitsudo, Department of Oral and Maxillofacial Surgery, Yokohama City University Graduate School of Medicine, Yokohama, Japan Background: Extended surgery for patients with locally advanced oral tongue squamous cell carcinoma markedly reduces the patient’s quality of life (QOL). To preserve function while improving locoregional control and survival rates, concurrent chemoradiotherapy represents one of the standard treatment modalities for definitive treatment of locoregionally advanced oral cancer. Retrograde superselective intra-arterial chemotherapy for oral cancer has the advantage of delivering a high concentration of the chemotherapeutic agents to the tumor bed, it can be used to provide daily concurrent chemoradiotherapy for patients with advanced oral cancer. The purpose of this study was to evaluate the therapeutic results and rate of organ preservation for patients with locally advanced tongue cancer treated with retrograde superselective intra-arterial chemotherapy and daily concurrent radiotherapy. Methods: Between August 2006 and June 2015, 101 patients with tongue squamous cell carcinoma (20 with stage II, 25 with stage III and 56 with stage IV disease, M0) underwent intra-arterial chemoradiotherapy. Catheterization from the superficial temporal and occipital arteries was performed. And treatment consisted of superselective intra-arterial chemotherapy (docetaxel 50-70 mg/m2, cisplatin 125-175 mg/m2) and daily concurrent radiotherapy (50-70 Gy, median 60 Gy) for 5-7 weeks. Results: The median follow-up for all patients was 36 months (range, 3113 months). After intra-arterial chemoradiotherapy, primary site complete response was achieved in 96 (95.0%) of 101 cases. Twenty three patients (22.8%) died. Using the Kaplan-Meier method, three-year overall and disease free survival rates were 78.7% and 67.3%, respectively. Three-year local control rate was 86.1% (stage II: 100%, stage III: 88.9%, stage IV: 80.3%). Conclusions: Daily concurrent chemoradiotherapy using retrograde superselective intra-arterial infusion for locally advanced tongue cancer can result in organ preservation and provide good overall survival, thus preserving organs and contributing to patients’ QOL.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Head and Neck Cancer 6033

Poster Session (Board #355), Sat, 1:00 PM-4:30 PM

Response to induction chemotherapy for locally advanced nasopharyngeal cancer to predict prognostic impact of chemosensitivity. First Author: Jiade J. Lu, Shanghai Proton and Heavy Ion Center, Shanghai, China Background: Nasopharyngeal cancer (NPC) is highly sensitive to chemotherapy and radiotherapy, and concurrent chemoradiotherapy (CRT) is the standard treatment for locally advanced NPC (LA-NPC).The effectiveness of platinum-based induction chemotherapy (ICT) with novel agent(s)such as taxane on overall or disease-free survival has been suggested in prospective trials. The purpose of this study is to evaluate the clinical value of observed response to ICT in predicting treatment outcome in patients with LA-NPC. Methods: Between 4/2007 and 8/2014, 687 patients with LA-NPC(i.e., T3, T4, or N+) were treated with platinum-based ICT(100 patients with taxane +/-5FU or gemcitabine [TP/TPF/GP], and 17 with 5FU [PF]) and evaluated for response with MRI were identified. Among the 120 patients who had stable disease (SD) or progression (PD) after ICT (ICT-resistant), 117 were successfully matched for diagnosis/follow-up time, gender, TNM stage, ICT regimen, CRT regimen (none vs. low vs. high dose cisplatin vs. targeted agents), and adjuvant chemotherapy (none vs. TP/TPF/GP vs. PF) to 117of the remaining 567 patients who achieved complete or partial response (CR or PR) after ICT (ICT-sensitive) by propensity score analysis. Results: The median follow-up time was 34 months (range 3-97 months). The overall survival (OS), progression-free survival (PFS), and local/regional/distantmetastatic relaps free survival (LRFS/RRFS/DMFS) rates were significantly better in the ICT-sensitive group (Table). Conclusions: Response to ICT is a significant predictive factor not only for OS, PFS, and DMFS but also for locoregional disease control in patients with LA-NPC. Prospective study is needed to confirm such findings then evaluate whether more aggressive CRT and/or adjuvant chemotherapy regimens may improve the prognosis of the ICT-resistant LA-NPC patients. Treatment outcomes in ICT-sensitive vs ICT-resistant LA-NPC patients. 3-year (%) 5-year (%) OS PFS LRFS RRFS DMFS

6035

ICT-resistant

ICT-sensitive

ICT-resistant

ICT-sensitive

P value

79.3 47.1 73.0 80.8 77.2

94.9 86.4 57.6 80.8 66.9

66.1 31.1 97.9 97.5 90.8

84.2 72.8 94.4 97.5 76.2

0.006 ,0.001 ,0.001 ,0.001 0.005

Poster Session (Board #357), Sat, 1:00 PM-4:30 PM

6034

325s Poster Session (Board #356), Sat, 1:00 PM-4:30 PM

Randomised study of 2 low doses of sunitinib to modulate tumor microvasculature prior to chemotherapy in nasopharyngeal carcinoma (NPC). First Author: Chee Seng Tan, National University Cancer Institute, National University Health System, Singapore Background: NPC is chemo- and radiosensitive but relapse rate is high and progression free survival (PFS) short for stage 4 disease. Almost all NPCs overexpress VEGF that promotes immature vasculature and poor tumor perfusion. We conducted a randomised phase 2 study of 2 low doses of sunitinib to promote maturity of tumor microvasculature prior to standard cisplatin-gemcitabine therapy in stage 4 NPC patients with the primary objectives to determine safety, tolerability and response rate (RR), and secondary objectives to determine PFS and effects on microvasculature. Methods: Eligible patients were randomised to receive sunitinib 12.5mg (ARM A) or 25mg (ARM B) daily for 7 days prior to each 3 weekly cycle of D1 cisplatin 75mg/m2 + D1, D8 gemcitabine 1000mg/m2, for 3 cycles followed by concurrent chemoradiotherapy in stage 4A/B (locally advanced) patients, and for 6 cycles in stage 4C (metastatic) patients. Enumeration of endothelial cell (CD34 antibody) and pericyte (SMA antibody) counts at 200x magnification by immunohistochemistry was performed by a blinded pathologist on tumor biopsies at baseline and cycle 1 day 7 sunitinib treatment to evaluate vascular normalization. Results: 33 patients were recruited (15 Arm A; 18 Arm B). The overall RR to induction chemotherapy was 82% (27/ 33) with 3 complete responses; 13/15 Arm A and 14/18 Arm B; 2 were non evaluable. At a median follow up of 23 months in stage 4A/B patients, 7/26 (27%) patients had progressed; median PFS was 9 months for stage 4C. G3/ 4 neutropenia without fever and thrombocytopenia without bleeding occurred in 67% and 40% in both Arms, respectively; resulting in dose delay of a week in 47% patients in Arm A and 50% in Arm B. Sunitinib alone induced early rise in plasma EBV DNA titre in 14/17 patients with detectable titres. 3/4 patients with detectable EBV DNA after induction chemotherapy relapsed within 12 months. Increase in vascular maturation reflected by increased SMA/CD34 score was seen in 75% of patients after 1 week of sunitinib in both arms. Conclusions: Sunitinib prior to chemotherapy is tolerable and active in stage 4 NPC, and induces early evidence of microvascular maturation, warranting further clinical studies. Clinical trial information: NCT01309633.

6036

Poster Session (Board #358), Sat, 1:00 PM-4:30 PM

Standard or split TPF induction chemotherapy followed by bioradiation: ICRAT randomized phase II study. First Author: Carmen Stromberger, Dpts. for Radiation Oncology, Comprehensive Cancer Center, Charite´ University Medicine Berlin, Berlin, Germany

A phase 1 trial of vorinostat in combination with concurrent chemoradiation therapy in the treatment of advanced stage head and neck squamous cell carcinoma. First Author: Theodoros Nicholas Teknos, The James Cancer Hospital Ohio State University, Columbus, OH

Background: TPF induction chemotherapy (IC) is a reasonable treatment for locally advanced (LA) squamous cell carcinoma (SCC) of the head and neck. Substantial toxicity may compromise compliance and thereby efficacy. In this 3-armed trial, we compared two TPF-IC regimens followed by cetuximab and radiation (bioradiation) vs. chemoradiation (CRT) in terms of efficacy, toxicity & survival. Methods: Eligible patients with LASCC of the oro- (ORO), hypopharynx (HYP) and oral cavity (OC) wererandomly assigned 1:1:1 to receive either 3 cycles of standard TPF (arm A) with TP (75 mg/m2, day 1) and F (750 mg/m2/24h days 1-4) or split-course TPF (arm B) with TP (40 mg/ m2, days 1 + 8) and F (1500 mg/m2/24h days 1 + 8) followed by cetuximab (250mg/m2) bioradiation vs. PF-CRT (arm C) to 72 Gy. Exploratory endpoints: best overall response (BOR; RECIST1.1), hematologic acute toxicity (CTCAE v.4.02) of TPF-IC, on time application of bioradiation week 10 6 7 days & survival (progression-free: PFS; metastasis-free: MFS; recurrencefree: RFS; overall: OS). Results: 94 patients were randomized across 8 centres (A: 34, B: 30, C: 30). Median age: 56 years. Tumorsite: 49% ORO, 34% HYP, 17% OC. 89% had IMRT. BOR (complete, partial remission) was similar for both TPF-arms, A 88% , B 83%. Hematologic toxicity did not improve with split-course TPF-IC (neutropenia G4/5: B 13/3%, A 29/0%; thrombocytopenia G4: B 0%, A 3%). Two patients died from neutropenic sepsis in split-course arm B. Compliance to TPF-IC was higher in the standard arm (A 77%, B 57%). Bioradiation started on time (range: 213 weeks) in 85% and 73% and was completed in 94% and 90% of patients in A and B, respectively. At 1-year PFS was 74%, 73% and 43%, MFS was 94%, 78% and 87%, RFS was 76%, 82% and 69%, and OS was 85%, 80% and 59% for A, B and C, respectively. According to univariate Cox regression analysis, TPF-IC was not associated with improved OS (hazard ratio: 1.16 [95% CI: 0.8-1.6; P = n.s]) Conclusions: Split-course TPF-IC lowered compliance and increased hematologic toxicity over standard TPF-IC with similar efficacy. Standard TPF-IC should not be modified outside of clinical trials. This trial was not statistically powered for survival differences. Some secondary end-points are pending. Clinical trial information: NCT01181401.

Background: Vorinostat (Zolinza) is a potent HDAC inhibitor that sensitizes head and neck squamous cell carcinoma (HNSCC) to cytotoxic therapy while sparing normal epithelium The primary objective of this study was to determine the maximally tolerated dose (MTD) of Vorinostat in combination with concurrent chemoradiation therapy. Secondary and tertiary endpoints included therapeutic response and delineation of acute and late toxicities. Methods: Twenty seven of 29 consented patients with Stage III/IV unresectable HNSCC completed the protocol. Patients received Vorinostat in combination with standard cisplatin/carboplatin chemotherapy (every 21 days) and intensity modulated radiation therapy (IMRT, 70Gy). Doses of Vorinostat were escalated in sequential cohorts of three patients. Toxicities were determined using standard NCI-CTCAEv4.03. Dose limiting toxicities (DLT) were defined as any adverse event of Grade 3 or higher. Results: Of the 27 patients on this study, 17/27 were p16/HPV+. The MTD of Vorinostat was 200mg delivered on M,W,F in combination with concurrent chemoradiation therapy. Twenty five of 27 patients (92.59%) had a complete response (CR) to therapy and 24/27 (88.89%) are alive and disease-free at a mean followup of 32.1 months (range 2-66 months). Two patients (7.41%) had residual disease at the completion of therapy and died of disease. One patient had a CR but developed lung metastases at 12 months. He is currently alive with disease at 38 months. The regimen was well tolerated and resulted in no treatment breaks during radiotherapy. Patients experienced a mean weight loss of 11.69% (range 0.79-20.61%) along with low rates of radiation related dermatitis (22/27 Grade 1) and oral mucositis (0/27 Grade 4, 15/27 Grade 3, 6/27 Grade 2, 6/27 Grade 1). No patient developed osteoradionecrosis, 4/27 developed pharyngeal stenosis and 1/27 developed Grade 3 xerostomia. Conclusions: Vorinostat in combination with concurrent chemoradiation therapy is a safe and highly effective treatment regimen. There was a high rate of complete response to therapy with toxicity rates comparable, if not more favorable, to existing treatment standards. Clinical trial information: NCT01064921.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

326s 6037

Head and Neck Cancer Poster Session (Board #359), Sat, 1:00 PM-4:30 PM

6038

Poster Session (Board #360), Sat, 1:00 PM-4:30 PM

Clinical predictive factors of overall survival and locoregional failure in advanced laryngeal cancer treated with definitive chemoradiation. First Author: Adeel M Khan, Cleveland Clinic, Cleveland, OH

Comorbidity and survival in locally advanced laryngeal cancer. First Author: Collin Francis Mulcahy, The University of Texas MD Anderson Cancer Center, Houston, TX

Background: Definitive chemoradiation (CRT) has become an established organ preservation treatment for patients (pts) with locoregionally advanced laryngeal cancer. Although most pts will experience long-term disease control, locoregional failure (LRF) remains a concern for a subset of pts. This study is aimed to identify predictors of LRF and overall survival (OS) after larynx-preservation therapy. Methods: Patients with AJCC stage III-IVB squamous cell carcinoma of the larynx treated with definitive CRT between 1993 and 2014 were identified at a single large tertiary institution in an IRB-approved registry. Patient, tumor, treatment, and outcome data were collected. The Cox proportional hazards method was used to identify predictors of LRF and OS. Results: There were 117 pts included in the study with median follow-up of 31 months (range: 1.8–185). Median age at diagnosis was 59 years, 74% were male, 57% were current smokers, and 21% had a history of alcohol abuse. Fourteen percent presented with T2 tumors, 70% T3 tumors, 16% had T4 disease and 62% were node-positive. Radiation therapy was administered with either a 3-field approach in 61% or with intensity modulated radiation therapy in 37%. Chemotherapy consisted of either cisplatin and 5-fluorouracil (60%) or single agent cisplatin (33%). Twenty (17%) LRFs were observed with 17 occurring within 2 years of treatment. Thirty-seven deaths (32%) were observed with median OS time of 116 months. Univariate analysis showed a higher LRF rate in pts with heavy alcohol and $ 20 pack-year smoking history who were currently smoking at time of diagnosis (HR 4.2, 95% CI 1.7-10.6, p = 0.002). These pts also had a lower OS rate (HR 2.2, 95% CI 1.0-4.9, p = 0.05). Their median OS time was notably reduced: 52 vs 119 months. T-stage, N-stage, or CRT regimen were not associated with LRF or OS. Conclusions: These data suggest that pts treated with definitive CRT for stage III-IVB laryngeal cancer who fail locoregionally do so within 24 months. Heavy alcohol history combined with significant smoking history and continued smoking was a significant predictor of LRF and decreased OS. Emphasis on smoking cessation and alcohol reform is warranted.

Background: We aim to quantify the relationship between pretreatment comorbidity and survival outcomes for patients with locally-advanced larynx cancer. Methods: We retrospectively reviewed records of patients treated for T3 and T4 laryngeal squamous cell cancer at MD Anderson Cancer Center between 1985 and 2011 after institutional review board approval. Baseline pretreatment comorbidity data were collected and age-adjusted Charlson comorbidity index (CCI) was calculated for each case. Kaplan-Meier and Cox proportional hazards modeling were used to determine associations with survival. Results: Among 548 patients with median age of 59 years (range 31-91), 58% were treated with larynx preservation and the rest with total laryngectomy and adjuvant radiotherapy. 19% received induction chemotherapy (54% platinum + taxane, 40% platinum-based). 34% received concurrent therapy (84% platinum-based). 237 patients (43%) suffered at least one comorbid condition prior to therapy. Cardiovascular diseases were the most common comorbidity (43%). Recursive partitioning showed CCI . 3 associated with poorer outcomes (p , 0.001). For all patients, 5-yr and 10yr overall survival (OS) for patients with CCI # 3 (n = 443, 81%) were superior to CCI . 3 (n = 105, 19%), (P, 0.0001), though 5 and 10-yr disease-specific survival (DSS) were non-different (70% vs. 66% at 5-yr and 60% vs. 66% at 10-yr, P= 0.17). 5-yr and 10-yr non-cancer cause specific survival (NCCSS) improved for age adjusted CCI # 3 (88% vs. 67% at 5-yr and 68% vs. 28% at 10-yr, P, 0.0001). Multivariate analysis revealed that age adjusted CCI . 3 and nodal staging as independent predictors for worse OS after Bonferroni correction (P = 0.0002 and P = 0.0001, respectively). Competing risk analysis showed risk of non-cancer death exceeded death from disease after 3 years for CCI . 3; while non-cancer risk of demise never exceeded cancer-specific risk during follow up for CCI # 3. Conclusions: The age-adjusted Charlson comorbidity index is a significant predictor of NCCSS and OS for locally-advanced larynx cancer patients, but is not associated with DSS. After 3 years, patients with CCI . 3 have non-cancer mortality risk greater than that of cancer-related death.

6039

6040

Poster Session (Board #361), Sat, 1:00 PM-4:30 PM

Prognostic value of plasma fibrinogen level and cervical nodal necrosis in stage IVA/B nasopharyngeal carcinoma patients who had positive cervical nodal metastasis. First Author: Mei Lan, Department of Radiation Oncology, Sun Yat-sen University Cancer Center, State Key Laboratory of Oncology in South China, Collaborative Innovation Center of Cancer Medicine, Guangzhou, China Background: Effective prognostic factors for stage IVA/B nasopharyngeal carcinoma (NPC) patients are limited. Fibrinolysis activation and cervical nodal necrosis (CNN) have been observed with a higher risk of progression and metastasis in head and neck cancers. The objective of this study was to investigate the prognostic value of plasma fibrinogen (FIB) level and CNN in stage IVA/B NPC patients who had positive cervical nodal metastasis. Methods: This was an institutional review board-approved retrospective study of 702 patients with newly diagnosed T4N1-2 or T1-4N3 NPC who were treated with definitive radiation therapy6chemotherapy, between January 2008 and December 2011. Coagulation tests including FIB, activated partial thromboplastin time were measured in all patients before treatment. Patients were also categorized into CNN group and non-CNN group. Disease-specific survival (DSS), disease-free survival (DFS) and distant metastasis-free survival (DMFS) were calculated using the KaplanMeier method, and differences were compared with log-rank test. Results: The death risk of patients with FIB . 4g/L was over 3-fold compared to patients with FIB # 4 g/L. The 5-year DSS, DFS and DMFS rates for patients with FIB . 4 g/L and FIB # 4 g/L were 60.3% and 75.4%, 56.6% and 69.0%, 55.3% and 68.2%, respectively (P , 0.001 for all). Further subgroup analysis demonstrated that survival decreased according to a gradual increase of FIB. The 5-year DSS, DFS and DMFS rates of the CNN and non-CNN groups were 62.1% and 82.5%, 44.7% and 81.0%, 48.9% and 81.2%, respectively (P , 0.004 for all). The distant metastasis risk of patients with both FIB . 4 g/L and CNN was over 7-fold compared with those without FIB increase or CNN. Multiple analysis showed that only FIB . 4 g/L, CNN were independent negative prognostic factors for DSS, DFS and DMFS. EBV-DNA was not an independent prognostic factor in multivariate analysis. Conclusions: Plasma FIB level and CNN presented to be stronger independent prognostic factors than EBV-DNA for stage IVA/B NPC patients who had positive cervical nodal metastasis. Studies about the mechanisms of how FIB and CNN affect survival are warranted.

Poster Session (Board #362), Sat, 1:00 PM-4:30 PM

Diagnosis of HPV driven head and neck cancer: Comparing p16 based algorithms with the RNAscope HPV-test. First Author: Haitham Mirghani, Gustave Roussy Cancer Campus, Villejuif, France Background: Accurate identification of HPV-driven oropharyngeal cancer (OPC) is a major issue and none of the current diagnostic approaches is ideal. An in situ hybridization (ISH) assay that detects high-risk HPV E6/E7 mRNA, called the RNAscope HPV-test, has been recently developed. Studies have suggested that this assay may become a standard to define HPV-status. Methods: To further assess this test, we compared its performance against the strategies that are used in routine clinical practice: p16 immunohistochemistry (IHC) as a single test and algorithms combining p16-IHC with HPV-DNA identification by PCR (algorithm-1) or ISH (algorithm-2). Results: Results: 105 OPC specimens were analyzed. The prevalence of HPV-positive samples varied considerably: 67% for p16-IHC, 54% for algorithm-1, 61% for algorithm-2 and 59% for the RNAscope HPV-test. Discrepancies between the RNAscope HPV-test and p16-IHC, algorithm-1 and 2 were noted in respectively 13.3%, 13.1%, and 8.6%. The 4 diagnostic strategies were able to identify 2 groups with different prognosis according to HPV-status, as expected. However, the greater survival differential was observed with the RNAscope HPV-test [HR: 0.19, 95% confidence interval (CI), 0.07–0.51, p = 0.001] closely followed by algorithm-1 (HR:0.23, 95% CI, 0.08–0.66, p = 0.006) and algorithm-2 (HR:0.26, 95% CI, 0.1–0.65, p = 0.004). In contrast, a weaker association was found when p16-IHC was used as a single test (HR: 0.33, 95% CI, 0.13–0.81, p = 0.02). Conclusions: Our findings suggest that the RNAscope HPV-test and p16based algorithms perform better that p16 alone to identify OPC that are truly driven by HPV-infection. The RNAscope HPV-test has the advantage of being a single test

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Head and Neck Cancer 6041

Poster Session (Board #363), Sat, 1:00 PM-4:30 PM

6042

327s Poster Session (Board #364), Sat, 1:00 PM-4:30 PM

Correlation of hypomagnesemia from cetuximab with survival in metastatic squamous cell cancer of the head and neck (SCCHN). First Author: Sabarish Ram Ayyappan, University Hospitals Seidman Cancer Center,Case Comprehensive Cancer Center, Cleveland, OH

Proteomic and mRNA expression analysis identifies PDL-1 expression as a marker of treatment failure in head and neck cancer. First Author: Heath Devin Skinner, The University of Texas MD Anderson Cancer Center, Houston, TX

Background: Cetuximab (C) improves survival and response rate in metastatic SCCHN. It also causes hypomagnesemia by renal electrolyte wasting. We evaluated incidence of hypomagnesemia and correlated this with clinical outcomes. Methods: This is a retrospective study of 46 patients with metastatic SCCHN treated with C alone or in combination with chemotherapy from 2003 – 2014. Patients received IV magnesium (Mg) supplementation on the day of infusion as per protocol. Clinical experimental variables: mg at baseline, on day 1 of each cycle and post therapy. Co-variates included age, sex, stage, skin toxicity, number of C infusions and concurrent chemotherapy. The predictive value of serum mg change on progression free survival (PFS) and overall survival (OS) was evaluated using the KaplanMeier method and Cox models. Results: Twenty six (57 %) patients developed hypomagnesemia that required replacement therapy. Fifty-four % had a 20% or greater change from baseline levels and nine patients (20%) developed grade 2-4 hypomagnesemia. Median time to first low mg was 39 days (3, 595). By univariate analysis, . 20% drop of mg was associated with better PFS (p = 0.02). By univariate Cox regression, % change of Mg, number of C infusions and time-to-lowest mg were significantly associated with PFS. There was an association between skin rash and drop of mg [ . 20% drop in 63.6 % with skin rash vs 30.8% without (p = 0.044)]. Median PFS was 4.1 months (95% CI- 2.7-6.4) and median OS was 9.4 months (95% CI- 6.6- 11.2). In the multivariate Cox proportional hazards analysis, age, percentage change of mg and skin rash were independently associated with OS and PFS (table 1). Conclusions: The incidence of C induced hypomagnesemia was 57% . Change in mg correlated with PFS and OS. Akin to skin rash, hypomagnesemia as a marker of outcome in SCCHN warrants further study.

Background: With the exception of human papilloma virus (HPV), there are no effective biomarkers in Head and Neck Cancer (HNSCC). To identify novel biomarkers in HPV-negative HNSCC we utilized a screening approach including reverse phase protein array (RPPA) and mRNA expression array on HNSCC tumor specimens from multiple cohorts of patients. Methods: All cohorts were locally advanced HPV-negative HNSCC treated with surgery and post-operative radiotherapy. RPPA examining 144 proteins and phosphoproteins was performed on tumor samples from 70 patients. Illumina mRNA gene array was performed on a separate cohort of 102 patients. Immunohistochemical validation studies for PDL-1 expression were performed on HNSCC tumors from 115 patients. Results: Cox regression analysis was performed on the RPPA and several proteins were associated with locoregional recurrence (LLR), including PDL-1 (p = 0.0034). PDL-1 was associated with LRR on multivariate analysis accounting for significant clinical variables (p = 3.2x10-4). On mRNA expression array, both PDL-1 and PDL-2 were in the top ten genes associated with LRR (8th and 3rd most significant; p = 6.1x10-05 and p = 1.1x10-05 respectively). On multivariate analysis taking into account clinical variables PDL-1 mRNA remained significantly associated with LRR (p = 0.021). To validate this observation, we performed immunohistochemical analysis on tumors to examine tumor and stromal PDL-1 and PD-1 as well as the degree of CD8 and CD4 positive immune infiltrate. Only tumor PDL-1 was significantly associated with LRR (p = 0.003) on univariate analysis. On multivariate analysis taking into account clinical variables, tumor PDL-1 expression remained significantly associated with LRR (p = 0.021). In all cohorts, patients with high PDL-1 expression had between a 35-40% absolute increase in 2 year LRR rate compared to tumors with low expression. Further, PDL-1 expression was associated with LRR regardless of p53 mutation status. Conclusions: The current study demonstrates that tumor PDL-1 expression is highly associated with LRR in HNSCC and argues that targeting this immune checkpoint in combination with radiation could provide significant benefit.

Multivariate analysis of covariates with PFS and OS. OS Factor Age (per year increase) % Change of magnesium Skin toxicity (Yes vs. no)

6043

PFS

HR

95% C.I

p-value

HR

95% C.I

p-value

1.07 0.12 0.18

1.01-1.14 0.02-0.89 0.08-0.44

0.026 0.038 0.0001

1.07 0.03 0.19

1.02-1.14 0.003-0.21 0.08-0.44

0.011 0.0007 , 0.0001

Poster Session (Board #365), Sat, 1:00 PM-4:30 PM

6044

Poster Session (Board #366), Sat, 1:00 PM-4:30 PM

A correlative analysis of PDL-1, PD-1, and EGFR, HER2, HER3 expression in oropharyngeal squamous cell carcinoma (OPSCC). First Author: Conor Ernst Steuer, Emory Univ, Atlanta, GA

Readmission and emergency department visits in the peri-treatment period for head and neck cancers patients in Ontario. First Author: Antoine Eskander, University of Toronto, Toronto, ON, Canada

Background: Agents targeting the PD1/PDL1 pathway are currently under investigation for squamous cell cancer of the head and neck (SCCHN). We sought to explore the potential association of the PD1/PDL1 pathway with clinical characteristics, outcome and other biologically relevant alterations (EGFR, HER3, HER3) in SCCHN using an institutional patient and tissue database. Methods: Protein expression was assessed by IHC on tissue microarray sections (EGFR, HER2, HER3) and or whole tissue sections (PD1/PDL1) using monoclonal antibodies. Expression of EGFR, HER2, HER3 was quantified in tumor cells while PD1/PDL1 was quantified on tumor cells, peritumoral stroma and stroma lymphocytes using intensity, percentage of positively staining cells and H-Score (product of intensity and percentage). Maximum PD1 lymphocyte density was measured on a scale of 0 to 4 (none, rare, moderate, high, very high). Significant associations (p , 0.05) between biomarkers and patient outcome were tested using descriptive and inferential statistics, logistic regression and Cox proportional hazards models in SAS 9.3. Results: We employed 100 OPSCC tissue samples for the analysis. Median age was 59 years, p16 positive (71%), male (81%), never smokers (18%), and 77% with stage 3 or 4. 25% of all cases were PDL1 positive. PD1 lymphocyte score of 2-4 was recorded in 65% of cases. There was a higher proportion of PDL1 (+) tumors in p16(+) OPSCC (87%) than PDL1 negative tumors (65%, p = 0.047). PDL1 status highly correlated with advanced nodal disease on multivariate analysis (5.53 (CI 1.06-28.77), p = 0.042). There was no correlation between PDL1, PD1 and EGFR, HER2, or HER3 expression. There was no association of PDL1 status with disease free or overall survival. Lower density of PD1+ lymphocytes in peritumoral stroma was associated with increased risk of death on multivariate analysis (HR = 3.17(1.03-9.78), p = 0.045). Conclusions: There was no association between PD1/PDL1 and expression of EGFR, HER2, or HER3 in OPSCC. PDL1 expression on tumor cells correlates with p16 positivity in OPSCC and advanced nodal status while low density of PD1 on peritumoral stromal lymphocyte density correlates with increased mortality.

Background: Mucosal head and neck squamous cell cancers are often managed with multimodality treatment which can be associated with significant toxicity. The objective of this study was to assess emergency department visits and non-elective hospitalizations for head and neck cancer patients during and immediately after their treatment. Methods: A cohort of patients treated for head and neck squamous cell carcinoma with curative intent was developed using administrative data. Emergency department visits and hospitalizations in the peri-treatment 90 day period following surgery, radiation, or chemoradiathion was determined. If a second treatment was initiated prior to the completion of 90 days, the attributable risk period was changed to the second treatment. Results: Cohort of 3,920 patients (1,317 larynx/hypopharynx; 2,603 oral cavity/oropharynx) from 2008-2012. The number of non-elective hospitalizations (per 100 patient days) attributable to surgery, radiation and chemoradiation was 0.35, 0.13, and 0.3 for larynx/hypopharynx and 0.11, 0.17, and 0.23 for oral cavity/ oropharynx respectively. The number of emergency department visits (per 100 patient days) attributable to surgery, radiation and chemoradiation was 0.89, 0.34, and 0.7 for larynx/hypopharynx and 0.44, 0.36, and 0.55 for oral cavity/oropharynx respectively. Multimodality treatment rendered higher readmission and emergency department visits. Conclusions: Patients undergoing treatment for head and neck cancer have significant non-elective hospitalizations and visits to the emergency department in the peritreatment period. Rates are higher in patients receiving multi-modality treatment, specifically, patients receiving chemoradiation have much higher non-elective hospitalizations and visits to the emergency department compared to those receiving radiotherapy alone or surgery alone. Further research is required to determine predictors of these outcomes and to identify high risk groups for which quality improvement interventions can be focused.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

328s 6045

Head and Neck Cancer Poster Session (Board #367), Sat, 1:00 PM-4:30 PM

BERIL-1: Biomarker results from targeted sequencing of circulating tumor DNA (ctDNA) and archival tissue in a randomized phase II study of buparlisib (BKM120) or placebo plus paclitaxel in patients with head and neck squamous cell carcinoma (HNSCC). First Author: Lisa F. Licitra, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy Background: A phase II, placebo-controlled study of buparlisib (BUP) and paclitaxel (PAC) in patients (pts) with platinum-pretreated recurrent/ metastatic HNSCC (NCT01852292) met its primary endpoint, demonstrating improved clinical benefit with BUP + PAC vs PAC alone: median progression-free survival (mPFS) was 4.6 vs 3.5 months, overall response rate (ORR) was 39% vs 14%, and median overall survival (mOS) at data cutoff was 10.0 vs 6.5 months. Exploratory analyses were conducted to identify biomarkers of improved clinical benefit. Methods: A total of 158 pts received BUP + PAC or placebo + PAC (n = 79 each). ctDNA from 109 plasma samples and genomic DNA from 84 archival tissue samples collected at screening were analyzed by next-generation sequencing with panels of 542 and 44 genes, respectively. Human papillomavirus (HPV) status was assessed by immunohistochemistry in 143 archival tissue samples. Results: Based on ctDNA, in 43 (39%) pts with TP53 alteration, mOS (8.4 vs 5.7 months; hazard ratio [HR] 0.49, 95% CI: 0.22–1.12) and ORR (33% vs 8%) were improved with BUP + PAC vs PAC alone. In 83 (76%) pts with low mutational load ( , 13 variants), mOS was improved (12.6 vs 6.4 months; HR 0.57, 95% CI: 0.31–1.07). TP53 alteration and mutational load in ctDNA were not associated with improved mPFS. Based on archival tissue, in 52 (62%) pts with TP53 alteration, mPFS (5.0 vs 2.2 months; HR 0.45, 95% CI: 0.23–0.89), mOS (11.0 vs 5.8 months; HR 0.45, 95% CI: 0.21–0.98), and ORR (38% vs 8%) were improved for BUP + PAC vs PAC alone. In 115 (80%) pts with HPV-negative status, mPFS (3.7 vs 2.3 months), mOS (9.1 vs 5.8 months), and ORR (40% vs 11%) were improved. PIK3CA alteration (9 [8%] pts in ctDNA; 16 [13%] pts in archival tissue) and phosphatidylinositol 3-kinase pathway alteration (27 [25%] pts in ctDNA; Lui et al. Cancer Discov 2013) were not associated with improved tumor response. Conclusions: Exploratory analyses suggest that BUP + PAC may be effective in pts with TP53-altered, HPV-negative HNSCC, or with low mutational load. Further analyses of OS and prognostic/predictive markers will be presented. Clinical trial information: NCT01852292.

6047

Poster Session (Board #369), Sat, 1:00 PM-4:30 PM

Pilot study of a modified yoga program for head and neck cancer (HNC). First Author: Barbara A. Murphy, Vanderbilt University Medical Center, Nashville, TN Background: HNC causes late musculoskeletal impairment (MSI), physical symptoms & psychologic distress. A modified yoga intervention using poses, breath work, meditation & relaxation may improve outcomes. We conducted a wait list control study testing a modified Yoga program for HNC survivors. Primary aims: feasibility (recruitment, compliance), satisfaction, safety, MSI and need for pose modifications. Secondary aim: preliminary efficacy for range of motion (ROM), posture, pain, and psychological distress. Methods: Eligible subjects: completed therapy . 3m prior to study, cancer free, and medically cleared. Study arms: individualized yoga instruction tiw x 4 weeks then biw x 4 weeks or wait list control. Yoga was taught by a certified instructor. Fidelity Measures: home practice guide with16 poses for neck-shoulders-back, 8 week instruction plan detailing progression of poses and practices, assessment tools for the instructor to capture MSI and document pose modifications, videotaped yoga sessions and home practice diary. Demographics & treatment data captured at baseline. MSI and symptoms assessed at baseline, 4, & 8 weeks using questionnaires, posture analysis software, & measures of jaw-neck-shoulder ROM. Results: 73 patients recruited: 40 consented, 11 ineligible. Pt characteristics: median age 63.1 yrs, 62.5% male, 90% white, 32% single, 33% income , $30,000. Of 20 intervention subjects, 5 discontinued (1 recurrence, 2 medical conditions, 2 work). Median # sessions completed 19/ 20; median # of home practices reported: 35/36. Median program satisfaction 10/10 for 10 assessed components. No adverse events were reported. Seventeen impairments were noted: neck (80%), shoulder (67%), posture (53%), frailty/deconditioning (53%), and jaw (53%). Mean # impairments 5.27 (SD 1.97): Exploratory analysis of efficacy measures indicates potential benefit for the following: pain, anxiety and depression, and ROM. Conclusions: A modified yoga program in HNC pts is safe. Recruitment rates and satisfaction were high. Demographics of HNC pts did not limit recruitment. Significant limitations in movement requiring pose modifications were found. Preliminary efficacy data supports further yoga studies. 1R21CA173202-01A1 NIH/NCI Clinical trial information: NCT01951664.

6046

Poster Session (Board #368), Sat, 1:00 PM-4:30 PM

Molecular decision tree algorithms to predict individual recurrence pattern for locally advanced nasopharyngeal carcinoma. First Author: XiangBo Wan, The Sixth Affiliated Hospital of Sun Yat-sen University, Guangzhou, China Background: This study is to design multiple molecular and clinicopathological variable integrated decision tree algorithms to predict individual recurrence pattern patterns (with VS without recurrence) for locally advanced nasopharyngeal carcinoma (NPC). Methods: A total of 136 locally advanced NPC patients, originated from a randomized controlled phase III trial, was included. For each patient, the expression levels of 33 tumor biomarkers in tissue, 3 Epstein-Barr virus related serological antibody titer and 5 clinical clinicopathological variables, were collected to construct the decision tree algorithm. Three algorithm classifiers, augmented by the adaptive boosting algorithm for variable selection and classification, were designed to predict individual recurrence pattern. The classifiers were trained in the training subset and further tested using a 10-fold cross-validation scheme in the validation subset. Results: In total, the expression level of 13 molecules expression in tissue, including AKT1, Aurora-A, Bax, Bcl-2, N-Cadherin, CENP-H, HIF-1a, LMP-1, C-Met, MMP-2, MMP-9, Pontin and Stathmin, and N stage were selected to construct three 10-fold cross-validation decision tree classifiers. These classifiers displayed high predictive sensitivity (87.2-93.3%), specificity (69.0-100.0%), and overall accuracy (84.595.2%) to predict recurrence pattern individually. Multivariate analyses confirmed the decision tree classifier was an independent prognostic factor to predict individual recurrence (algorithm 1: HR 0.07, 95% CI 0.03-0.16, P , 0.01; algorithm 2: HR 0.13, 95% CI 0.04-0.44, P , 0.01; algorithm 3: HR 0.13, 95% CI 0.03-0.68, P = 0.02). Conclusions: Multiple molecular and clinicopathological variable integrated decision tree algorithms may individually predict the recurrence pattern for NPC. This decision tree algorism provides a potential tool to select patients with high recurrence risk for intensive follow-up, and to diagnose recurrence at an earlier stage for salvage treatment in the NPC endemic region.

6048

Poster Session (Board #370), Sat, 1:00 PM-4:30 PM

Racial disparity in oncologic and patient-reported quality of life (PROs) outcomes in patients with locally advanced head and neck squamous cell carcinomas (HNSCC) enrolled in a randomized phase II trial. First Author: Margaret Winn Kendrick, Seattle Cancer Care Alliance/Univ of Washington, Seattle, WA Background: Using PROs from our previously completed clinical trial (NCT00410826), we conducted an exploratory analysis to identify PRO associations with clinical factors and oncologic outcomes. Methods: From 12/2006 to 10/2011, 204 patients(pts) were randomized to cisplatin 100mg/m2 on radiation (XRT) days(d)1, 22, and 43 concurrent with 70Gy XRT (Arm A) or the same treatment with erlotinib 150mg daily starting XRT d7 until XRT completion (Arm B). Using the validated University of Washington QOL form, PROs were measured on XRT d-7 (Arm B only), d0, d30, and d180. PRO associations with clinical factors and oncologic outcomes were explored using linear mixed, logistic and cox models, while adjusting for potential confounders. Results: One hundred eighty nine (93%) [Arm A (N = 97); Arm B (N = 92)] pts consented to PRO collection. PRO scores were available in Arm A: 55 (56%) pts at d0, 65 (67%) at d30, and 45 (46%) at d180 and Arm B: 82 (89%) pts at d-7, 58 (63%) at d0, 69 (75%) at d30, and 41 (44%) at d180. Demographic and tumor characteristics were wellbalanced apart from more females in Arm A [20 (19%) vs 8 (8%); p = 0.03]. There were 7 (16%) black pts in Arm A and 13 (13%) in Arm B (p = 0.28). There was no change in mean PRO scores from d-7 on d0 (p = 0.83). Scores were significantly lower in both arms at d30 (estimated difference from baseline -13.21 (CI -18.32, -8.1), with no significant difference by Arm (p = 0.71). D180 remained lower than d-7 for Arm A (estimated difference from baseline -6.56 (CI -12.30, -0.82) p = 0.03), however were not significantly different from d-7 (baseline) in Arm B (p = 0.68). After adjusting for potential confounders on multivariable analysis, black race was an independent predictor for inferior PRO scores [OR -15.87; 95% CI (-21.11, -10.63); p , 0.0001] at all timepoints, complete response rates [OR 0.23; 95% CI (0.1, 0.53); p , 0.001] and overall survival [OR 3.94; 95% CI (1.89, 8.24); p , 0.001]. Conclusions: PRO worsened during chemoradiation, but returned to baseline in pts on Arm B. Black pts had inferior PRO and overall survival, meriting further investigation.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Head and Neck Cancer 6049

Poster Session (Board #371), Sat, 1:00 PM-4:30 PM

6050

329s Poster Session (Board #372), Sat, 1:00 PM-4:30 PM

Activity of afatinib administered in a window pre-operative study in squamous cell carcinoma of the head and neck (SCCHN) : EORTC-90111. First Author: Jean-Pascal H. Machiels, Universite Catholique de Louvain Cliniques Universitaires St-Luc, Brussels, Belgium

Prognostic factors for overall survival of patients with head and neck soft tissue sarcoma based on 25 years data from the Netherlands Cancer registry. First Author: Lotte Ten Thije, Radboud University Medical Center, Nijmegen, Netherlands

Background: Targeted agents (TA) are often investigated in unselected endstage cancer patients (pts), making difficult to fully evaluate their activity and impairing translational research. One way to resolve this issue is to perform "window" studies where a TA is given between the diagnosis and surgery. Afatinib (A) is an oral ErbB family blocker that improves PFS of pts with recurrent SCCHN after platinum failure. Methods: This study was a randomized, multicenter, phase II window of opportunity trial. Treatmentna¨ıve SCCHN pts selected for primary curative surgery were randomized (5:1 ratio) to receive A during 14 days (day -15 until day -1) before surgery (day 0) or no treatment (B). Tumour biopsies, FDG/PET, and MRI were performed at diagnosis and at surgery. The primary end point was metabolic FDG-PET/CT response (centrally reviewed; H0 = 10%, H1 = 30%, Alpha = 10% 1-sided, Power = 90%); according to EORTC guidelines). Other endpoints included response by RECIST, DCE MRI, DWI MRI, and translational research. Results: 30 pts were randomized in 4 centers: 25 to A and 5 to B. 23 pts randomized to A were eligible (median age: 59, male/female: 16/7, oral cavity/oropharynx: 19/4; ECOG 0/1: 15/8, stage II/III/IV: 4/4/15). One pt discontinued A after 11 days for gr3 renal failure and diarrhea with delayed surgery by 24 days. Another pt had gr3 acneiform rash. Of the 24 pts who completed the planned A treatment, 2 had delayed surgery by 4 and 2 days (gr2 colitis and organizational issues). No surgical complications were related to A. Of the 23 eligible pts, 16 had a partial metabolic FDG/PET response (PMR): 70% (95% CI: 47-87%). There were no complete responses and 7 pts had stable disease (SMD). The proportion of responders to A was significantly higher than 10% (p , 0.001). There was no CMR/PMR in the no-treatment arm (n = 3). Conclusions: This study demonstrates that multinational challenging window studies are feasible. Preliminary results show that A given for 2 weeks to newly diagnosed SCCHN pts induces a high rate of FDG-PET response and can be safely administered before surgery. Definitive results will be presented at ASCO 2016. Translational research to evaluate predictive biomarkers is ongoing Clinical trial information: NCT01538381.

Background: Head and neck soft tissue sarcomas (HNSTS) account for 2% of head and neck cancers. The aim of this research is to establish prognostic factors for overall survival (OS) in patients (pts) with HNSTS in general and dermatofibrosarcoma protuberans (DFSP) in particular. Methods: We retrieved data from the Netherlands Cancer Registry (NCR) on pts diagnosed with HNSTS (WHO classification) from 1989 to 2013, including pts’ vital status. OS was analyzed using univariable and multivariable-adjusted Cox proportional hazards models, for the total group as well as for the subset of surgically treated patients. In the Cox analysis, we used multiple imputations for missing data on tumor grade, size, depth, and resection status. Results: The most common subtypes in the total of 1573 pts were high-grade sarcoma not otherwise specified, leiomyosarcoma and DFSP. Median age of pts was 66 years, the majority was male (64%). The 5-year and 10-year OS were 57% and 43%, respectively. OS did not change over time. The 5- and 10-year OS were 73 and 69% in pts , 18 years (n = 178), 78% and 71 % in pts 18-34 years (n = 152), 72% and 64% in pts 35-64 years (n = 439), and 40% and 20% in elderly ( $ 65 years, n = 804), respectively. OS was better for surgically treated pts (5-year: 62% versus 33%, 10-year: 47% versus 24%). Among pts who had surgery (n = 1285, 82%), multivariable analysis on the imputed dataset identified old age ( $ 65 years), high tumour grade, presence of metastases at time of diagnosis, and residual disease (R1-2) as independent unfavorable prognostic factors affecting OS. Pts with DFSP of the head and neck (n = 212) had significantly worse OS compared with DFSP in the rest of the body (n = 1458) (p , 0.001) with 10 year OS of 70% versus 92%, respectively. Conclusions: We analyzed the largest population of pts suffering from HNSTS so far. Survival did not improve in 25 years despite improvements in diagnostics and treatments. While prognosis was better for pts who were surgically treated, old age, tumors exhibiting aggressive behavior and incomplete surgery negatively impacted survival in this group. OS of head and neck DFSP was significantly worse compared with the rest of the body.

6051

6052

Poster Session (Board #373), Sat, 1:00 PM-4:30 PM

Poster Session (Board #374), Sat, 1:00 PM-4:30 PM

A prognostic model combining CD4/CD8 ratio and N stage predicts the risk of distant metastasis for patients with nasopharyngeal carcinoma treated by intensity modulated radiotherapy. First Author: Chang-Juan Tao, Zhejiang Cancer Hospital, Zhejiang, China

Analysis of immune infiltrates in a genomically characterized clinical cohort of head and neck squamous cell carcinoma (HNSCC) patients (pts). First Author: Nicole Grace Chau, Medical Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA

Background: Distant metastasis is a poor prognostic factor in nasopharyngeal carcinoma (NPC). We aimed to evaluate the correlation between circulating lymphocyte subsets and clinical variables, and design an effective prognostic model for distant metastasis-free survival (DMFS) in NPC. Methods: Subsets of circulating lymphocytes were determined in 719 nonmetastatic NPC patients before treatment, and overall survival and DMFS was monitored. Significant prognostic factors were identified using univariate and multivariate analyses. Results: The percentage of CD19+ lymphocytes correlated negatively with TNM stage (r = –0.082, P = 0.028). Patients with higher CD4/CD8 ratios ( $ 1.77) showed better 5-year DMFS than patients with lower ratios (91.9% vs. 85.4%, P , 0.001). Multivariate analysis revealed that CD4/CD8 ratio (HR, 0.450; 95% confidence interval [CI], 0.266–0.760; P = 0.003), T stage (HR, 2.3549; 95% CI, 1.409 – 3.930; P = 0.001) and clinical N stage (HR, 2.294; 95% CI, 1.370 – 3.839; P = 0.002) were independent factors affecting DMFS. The prognostic N-R model we developed divides patients into three groups: (1) low-risk (early N stage and CD4/CD8 ratio $ 1.77); (2) intermediate-risk (advanced N stage or CD4/CD8 ratio , 1.77) and (3) high-risk (advanced N stage and CD4/CD8 ratio , 1.77) of distant metastasis. Conclusions: Our prognostic model, based on clinical N stage and CD4/CD8 ratio, may predict the risk of distant metastasis, allowing individualized treatment for NPC.

Background: Characterization of immune infiltrates and correlation with clinical and genomic alterations in HNSCC may facilitate the discovery of prognostic or predictive biomarkers to better select pts for immune checkpoint therapy. Methods: We previously reported the clinical utility of targeted next-generation sequencing of tumor DNA derived from 213 HNSCC pts treated at our center (Chau/Li CCR 2016). In this same cohort, we performed immunohistochemical analysis on formalin-fixed paraffin embedded tumor samples to determine PD-L1, PD-L2, PD-1, FOXP3, LAG3, GITR, TIM3, CD3, CD4, and CD8 expression. Survival estimates were obtained by Kaplan-Meier method. Differences in progression free and overall survival (PFS, OS) between pts with high or low PD-L1, PD-L2, PD-1, GITR, TIM3 expression, which was scored semiquantitatively by staining intensity (0, negative; 1+, weak; 2+, moderate; 3+, intense) and extent ( , 5%, 525%, 25-50%, 50-75%, . 75%), and absolute average number per 4 HPF for GITR, FOXP3, LAG3, CD3, CD4, CD8 tumor infiltrating lymphocytes, were assessed by univariate and multivariate survival analyses using Cox proportional hazards model. Immune marker expression was correlated with clinical and genomic factors. Results: Initial analysis (n = 36 pts) revealed expression of PD-L1 (50%, 18/36), PD-L2 (31%, 11/35), PD-1 (37%, 13/ 35), GITR (30%, 10/33), TIM3 (76%, 26/34), and lymphocyte expression (absolute number . 40) was identified for FOXP3 (26%), LAG3 (20%), CD3 (68%), CD4 (21%), CD8 (86%). Poorer OS appeared to be associated with PD-1+ (p = 0.06), TIM3+ (p = 0.26) and GITR+ (p = 0.17). CD8+ alone was not predictive of survival, but if CD8+ T-cells displayed exhaustion markers (TIM3, PD-1) and the tumor was PD-L1+, the outcome was worse. No significant correlations were observed between presence of immune infiltrates and tumor site, stage, age, sex, smoking, p16 status, somatic alterations, copy number variations, PFS or OS. Conclusions: Initial analysis reveals co-expression of CD8+/TIM3+/PD-1+ T-cells and PD-L1+ tumors may be associated with poorer survival outcomes, suggesting that targeting of multiple checkpoints in these tumors may be warranted.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

330s 6053

Head and Neck Cancer Poster Session (Board #375), Sat, 1:00 PM-4:30 PM

Association of FASL and FAS polymorphisms, enrolled in extrinsic apoptosis pathway, with head and neck squamous cell carcinoma risk and outcomes. First Author: Vitor Teixeira Liutti, Department of Internal Medicine, Faculty of Medical Sciences, University of Campinas, Campinas, Brazil Background: Apoptosis plays an important role in origin of head and neck squamous cell carcinoma (HNSCC). Inherited genetic alterations, such as single nucleotide polymorphisms (SNPs), may influence the individual apoptotic capacity, having development of tumors as consequence. To the best of our knowledge, the roles of FASL c.-844C . T and FAS c.-671A . G SNPs in HNSCC risk, clinic pathological aspects, overall survival and progression free survival are unclear, and therefore these were the aims of our study. Methods: DNA of 463 patients and 470 controls were analyzed by PCR-RFLP. Patients were treated according to the Institutional protocol. The statistical analyses were realized using chi-square, logistic regression model, multifactor dimensionality reduction (MDR), Kaplan-Meier, and univariate and multivariate Cox analyses. Results: FASL TT genotype was more frequent in overall HNSCC patients (27.9% vs 16.2%, P = 0.001) and in those with SCC of oral cavity (30.0% vs 16.2%, P= 0.006), pharynx (29.9% vs 16.2%, P= 0.007), and larynx (25.4% vs 16.2%, P= 0.03) than in controls. Carriers of the genotypes were under a 3.24, 5.86, 2.93 and 2.54-fold increased risks of overall HNSCC and SCC of the mentioned subsites than others, respectively. An excess of FASL CT or TT plus FAS AA or AG combined genotype was seen in overall HNSCC patients compared to controls (91.1% vs 84.1%, P= 0.04); carriers of the genotype were under a 2.31-fold increased risk overall HNSCC than others. Among smokers, FASL TT and FAS AA genotypes were associated with 165.89 and 81.05-fold increased risks of HNSCC (P, 0.001), respectively. FASL c.-844C . T, FAS c.-671A . G SNPs and tobacco was the best interaction MDR model for risk of overall HNSCC and SCC of oral cavity, pharynx and larynx (P, 0.001). The median follow-up time of HNSCC patients was 46.0 months (1.6-166.0); no association of SNPs and patients´ survival was seen in study. Conclusions: Our data present preliminary evidence that inherited abnormalities in FASL c.844C . T and FAS c.-671A . G SNPs are determinants of overall HNSCC and SCC of oral cavity, pharynx and larynx, particularly among smokers, possibly due to their action in tumor carcinogenesis.

6055

Poster Session (Board #377), Sat, 1:00 PM-4:30 PM

Defining the immunologic phenotypes and their prognostic impacts on head and neck squamous cell cancer (HNSCC). First Author: HyeRyun Kim, Division of Medical Oncology, Department of Internal Medicine, Yonsei University College of Medicine, Seoul, South Korea Background: To investigate the frequency and distributing patterns of programmed death-ligand 1 (PD-L1), other immune checkpoints (ICP) and their prognostic impacts in resected HNSCC. Methods: PD-L1 expression and its pattern on tumor cells (TC) and tumor-infiltrating immune cells (IC) were investigated, using SP142 assay, in micro-arrayed tumors from 402 HNSCC. PD-L1 on TC and IC can be categorized into 4 groups according to their percentage; TC0 or IC0, , 1%; TC1 or IC1, $ 1% and , 5%; TC2 or IC2, $ 5% and , 50%; TC3 or IC3, $ 50%. PD-L1 positivity was defined as $ 5% of TC or IC by immunohistochemistry. Tumor infiltrating lymphocytes (TIL) and their various ICP, i.e. programmed death-1 (PD-1), lymphocyteactivation gene-3 (LAG-3), and inducible T-cell co-stimulator (ICOS), were evaluated. The densities of TIL and various ICP were semiquantitatively scored, and positivity was defined as higher than the median value of each marker. The association between various immune markers, recurrence-free survival (RFS) and overall survival (OS) were analyzed. Results: The predominant primary sites were oral cavity (50.7%) and oropharynx (30.3%). The proportions of stage were 28.9% stage I, 10.7% stage II, 18.4% stage III, and 42.0% stage IVA/B. In a Cox proportional hazard model adjusted for age, gender, smoking, stage and HPV status, high PD-L1 expression on IC was an independent prognostic factor for better RFS and OS (RFS: adjusted hazard ratio [AHR], 0.49; P = .0007; OS: AHR, 0.37; P = .04). High CD3+ or CD8+ TIL density, expression of FOXP3, and PD-1 in TIL were independently associated with better prognosis. HNSCC is further characterized by tumors that express PD-L1 exclusively on TC or IC. Notably, TC3 and IC3 represent distinct populations with , 1% overlap, suggesting that PD-L1 might be regulated by the distinct mechanisms in TC and IC subsets. IC3 exhibit higher CD8+TIL density compared to TC3, suggesting that PD-L1 on IC is regulated by adaptive mechanisms and reflects preexisting immunity. Conclusions: PD-L1 on IC is independent predictors of OS in resected HNSCC. This highlights the importance of comprehensive assessment on both TC and IC. Exploratory analysis of tumor gene expression will be presented.

6054

Poster Session (Board #376), Sat, 1:00 PM-4:30 PM

E3311 trial of transoral surgery for oropharynx cancer: Implementation of a novel surgeon credentialing and quality assurance process. First Author: Shuli Li, Dana-Farber Cancer Institute, Boston, MA Background: ECOG-ACRIN (E3311) examines the role of transoral head and neck surgery in treatment deintensification for stage III/IV p16+ oropharyngeal cancer (OPC). Patients with lateralized, resectable T1-T2 OPC and N1-N2b with no matted nodes undergo transoral resection and neck dissection, followed by risk-based adjuvant therapy. Intermediate risk patients [negative margins, N2 6 extranodal extension (ENE) # 1mm] are randomized to 50Gy vs. 60Gy radiation. We established credentialing criteria and ongoing quality assurance (QA) for transoral surgeons wishing to accrue to E3311. Methods: Each surgeon attests to experience of . 20 cases of transoral resection of OPC, using either transoral robotic or laser microsurgery. Surgeons then submit paired surgical pathology and operative reports for 10 transoral resections within the past 24 months. Nine experienced head and neck surgeons established criteria for approval, review case submissions, on-hold status and ongoing QA. Positive margin(s) were permitted on only 1 of 10 cases during initial credentialing or while accruing to E3311. Results: Since 2013, 107 surgeons applied for credentialing. After peer-review, 71 surgeons were credentialed for E3311 at 51 different centers, with 49 surgeons having accrued . 1 and 31 having accrued $ 3 patients (accrual per surgeon 1-24, median = 4). Twelve surgeons were asked to provide additional, replacement cases, where histology or margin status was unclear or deemed insufficient during review. Of these, 10 were approved upon re-review. Two surgeons withdrew their applications, and 22 have not submitted cases for credentialing review. Ongoing QA has led to review of two surgeons with positive margin(s) in their first 5 cases. The composite QA stopping rule of 22% combining grade III/IV bleeding and positive margins was not met during planned DSMC review (August 2015), with 8% grade III/IV bleed rate and 3% positive margins. Conclusions: Initial surgeon credentialing and ongoing QA results in low rates of positive margins and grade III/IV oral bleeding rates in a multicenter trial of transoral surgery. Clinical trial information: NCT01898494.

6056

Poster Session (Board #378), Sat, 1:00 PM-4:30 PM

Immune-related gene expression signatures as predictive biomarkers for outcome after concurrent chemoradiation in patients with locally advanced oropharyngeal carcinomas. First Author: Inge Tinhofer, Dept. of Radiooncology and Radiotherapy, Charite´ Unversity Hospital and German Cancer Research Center Heidelberg (DKFZ)/German Cancer Consortium (DKTK) Partner Site Berlin, Berlin, Germany Background: Immune-related gene expression signatures as candidate predictive biomarkers for cisplatin (CDDP)- and mitomycin C (MMC)-based chemoradiation (CRTX) were established in oropharyngeal carcinoma (OPC) patients from the ARO04-01 phase III trial. Potential overlap between the immune signatures of CRTX efficacy and the previously reported ‘inflamed’ signature of pembrolizumab activity was determined. Methods: Archival tumor specimens from OPC patients treated within the ARO04-01 study with concurrent MMC-CTRX (N = 24) or CDDP-CTRX (N = 23) were included. FFPE-extracted RNA was analyzed using the PanCancer Immune Profiling Panel on the NanoString nCounter system. Gene expression signatures predictive for 3yr-OS were established by Spearman Rank correlation and hierarchical cluster analysis. Survival estimates according to the immune signatures were determined by Kaplan-Meier analysis. The potential interference with the HPV status was evaluated in multivariate Cox regression models. Results: Signature 1 was associated with OS (P, 0.001) and PFS (P= 0.006) after CDDP-CRTX but not MMC-CRTX. Its positive predictive value for 3yr-OS was 87.5%, and the negative predictive value was 100%; AUC = 0.97 [95% CI, 0.89-1.0]. Signature 1 was enriched for genes of the adaptive immune system, mainly affecting T-cell receptor and PD-1 signaling. A partial overlap with genes from the ‘inflamed’ pembrolizumab signature was observed. In contrast, signature 2 which showed a significant association with outcome for both CRTX regimens was enriched for genes of the innate immune system, regulating Toll-like receptor, TNF receptor and NFkB signaling. Despite a higher prevalence of signature 1 in HPV+ tumors, its predictive value for OS and PFS after CDDP-CRTX was independent of the HPV status. Conclusions: We demonstrate for the first time an impact of adaptive immunity and PD-1 signaling on the efficacy of CDDP-CRTX which may be important for inclusion of PD-1/PD-L1 inhibitors into curative treatment settings. The NanoString technology for immune gene expression analysis should provide a useful tool for patient stratification.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Head and Neck Cancer 6057

Poster Session (Board #379), Sat, 1:00 PM-4:30 PM

FGFR1 as a prognostic biomarker and potential therapeutic target in head and neck squamous cell carcinoma. First Author: Stefan M. Willems, Department of Pathology, University Medical Center Utrecht, Utrecht University, Utrecht, Netherlands Background: FGFR1 (fibroblast growth factor receptor 1) is a promising therapeutic target in multiple types of solid tumors, including head and neck squamous cell carcinoma (HNSCC). FGFR inhibitors have shown great therapeutic value in preclinical models. However, resistance remains a major setback. In this study we have investigated the prognostic value of FGFR1 expression in HNSCC, the therapeutic relevance of targeting FGFR with AZD4547, and potential resistant mechanisms. Methods: Immunohistochemistry and fluorescence in situ hybridization were applied on tissue microarrays to investigate FGFR1 protein expression and FGFR1 gene copy-numbers in 452 HNSCC. The sensitivity of HNSCC cell lines to AZD4547, either as single or combination treatment with gefitinib, was assessed using long-term colony formation assays and biochemical analysis. Results: FGFR1 protein overexpression occurred in 82% (36/44) of HPV-positive HNSCC and 75% (294/392) of HPV-negative HNSCC and relates with poor overall survival and disease-free survival in HPV-negative HNSCC (HR: 2.97; 95% CI: 1.73-6.49; P = 0.001, HR: 1.51; 95% CI: 1.03-2.31; P = 0.036). Moreover, the FGFR1 gene was amplified in 3% (3/110) of HPV-negative oropharyngeal HNSCC. Treatment of the high FGFR1 expressing cell line CCL30 with AZD4547 reduced cell proliferation and FGFR signaling. Two FGFR-amplified cell lines, SCC147 and BICR16, were resistant to AZD4547 treatment due to EGFR signaling. Combined AZD4547 and gefitinib treatment synergistically inhibited proliferation of resistant cell lines. Conclusions: We identify high FGFR1 expression as a candidate prognostic biomarker in HPV-negative HNSCC. Furthermore, we provide a rationale for treating FGFR1-expressing HNSCC with the FGFR inhibitor AZD4547 and FGFR inhibitor-resistant HNSCC with the combination of AZD4547 and gefitinib.

6059

Poster Session (Board #381), Sat, 1:00 PM-4:30 PM

Biomarkers of prognosis in patients with differentiated thyroid cancer: Results from the DECISION trial. First Author: Carol Elaine Pena, Bayer HealthCare Pharmaceuticals, Whippany, NJ Background: Sorafenib (SOR) is a multikinase inhibitor approved to treat locally recurrent or metastatic, progressive, differentiated thyroid cancer (DTC) refractory to radioactive iodine treatment. We examined serum thyroglobulin (Tg) and candidate plasma biomarkers to identify prognostic and/ or predictive biomarkers in patients (pts) from the Phase 3 DECISION trial of SOR in DTC. Methods: Serum for Tg was collected at baseline (BL) and during treatment. 15 plasma proteins were tested in BL samples. Biomarkers were analyzed by Cox regression for correlation with PFS, OS and disease control rate (DCR). Optimal biomarker cutoffs (OBC) were determined using a maximum chi-square method testing between the 25th and 75th percentiles. Results: BL Tg and plasma biomarker data were available for 403 (96.6%) and 395 (94.7%) of the 417 DECISION pts, respectively. Elevated BL Tg was robustly associated with poor prognosis in terms of PFS (HR = 2.03 for high vs low Tg, p , 0.0001; using OBC of 1547 pmol/L, equivalent to the 58th %ile) and DCR (odds ratio [OR] = 0.32, p = 0.01; OBC, 3047 pmol/L = 68th%ile) in multivariate analyses including both treatment arms. Pts experiencing a $ 30% reduction in Tg during treatment had longer PFS than those not reaching a 30% decrease in both the placebo (HR = 0.49, p = 0.009) and SOR (0.61, p = 0.022) cohorts. 76% of SOR pts achieved a Tg decrease $ 30% vs 14% of placebo pts (p , 0.001). Elevated BL VEGF-A (higher than the OBC) was also strongly associated with poor PFS (HR = 1.82, p = 0.0003), OS (HR = 2.13, p = 0.0067), and DCR (OR = 0.30, p = 0.0046). Baseline Tg and VEGF-A together were prognostic for PFS (HR = 2.12, p , 0.00001). Elevated BL VEGF-C (above OBC) and TGF-b1 (continuous variable) and low BL E-cadherin (continuous variable) were associated with poor DCR, while elevated BL vimentin was associated with poor OS (continuous variable). None of the biomarkers examined were able to predict benefit from SOR in DTC. Conclusions: Elevated BL Tg and VEGF-A were associated with poor prognosis in DTC, as were elevated VEGF-C, TGFb1, and low E-cadherin. None of the biomarkers were predictive of response to sorafenib. Serum Tg changes over time may be a pharmacodynamic biomarker of tumor response and/or progression. Clinical trial information: NCT00984282.

6058

331s Poster Session (Board #380), Sat, 1:00 PM-4:30 PM

Predisposing germline mutations in young patients with oral cavity tumors identified by next generation sequencing. First Author: Florentia Fostira, NCSR Demokritos, Athens, Greece Background: Oral cavity squamous cell cancers (SCCOC) and in particular oral squamous cell tongue cancers (SCCOT) occur with a rising incidence in younger patients often lacking the typical risk factors of tobacco use, alcohol use and human papilloma virus infection. While SCCOT tumors in young and older patients appear genetically similar, the cause for the increasing incidence in young patients remains unknown. We therefore sought to identify germline mutations that can predispose to SCCOT in young patients, unselected for family history. Methods: We prospectively collected 27 young HPV negative patients with SCCOC. Informed consent was provided from all the individuals that underwent genetic testing. Patients were treated with surgery followed by radiation or chemoradiation based on prognostic features. DNA extracted from peripheral blood lymphocytes was analyzed by a customized 94-gene panel through massive parallel sequencing. The minimum base coverage was 50x, while the variant calling was performed with GATK. Tumor samples from all the patients with definitely pathogenic mutations were subsequently tested for loss of heterozygosity (LOH) of the wild-type allele. Results: The median age of diagnosis was 33.2 (range 1552 years). Germline loss-of-function mutations in three genes were identified in 18.5% (5/27) of the individuals tested, while 10 patients carried variants of unknown clinical significance (VUS) in six additional cancer genes. CDKN2Awas the most frequently mutated gene seen in 11.1% of the individuals and recurrence was not observed in any of these patients. Moreover, LOH in these tumors, represented by either reduction or loss of the wild type allele was confirmed. None of these patients had family history of cancer suggestive of an underlying genetic defect. Conclusions: This is the first study that correlates germline mutations in three cancer predisposing genes with SCCOC in young patients, suggesting that early age of diagnosis alone indicates a risk factor. These data may have significant clinical implications for individuals and their families, while potentially aberrant pathways amenable to targeted therapeutic strategies can be highlighted.

6060

Poster Session (Board #382), Sat, 1:00 PM-4:30 PM

Correlation of anti-tumor adaptive immunity with clinical response in a phase II “Window” trial of neoadjuvant cetuximab in patients with resectable stage III-IV head and neck squamous carcinoma (HNSCC). First Author: Robert L. Ferris, University of Pittsburgh Cancer Institute, Pittsburgh, PA Background: Despite near universal EGFR expression in HNSCC, the antiEGFR antibody, cetuximab, improves survival in only 15% of patients. Activation of immune cell subsets might contribute to anti-tumor responses. We investigated the association of cetuximab response with FcgR polymorphism, hypothesized to influence the potency of cetuximab-FcgR binding and NK cell-mediated cytotoxicity, and peripheral EGFR-specific T cells, markers of tumor-antigen specific cellular immunity. Methods: We conducted a prospective phase II “window” trial in patients with previously untreated, resectable stage III-IVA HNSCC (n = 40) using cetuximab (400 mg/m2 then 250mg/m2/week) for 3-4 weeks preoperatively, followed by adjuvant chemoradiation with or without cetuximab. Biomarkers in peripheral blood (PBL) and tumor infiltrating lymphocytes (TIL) were correlated with clinical response and acneiform rash. In the brief window setting,, response was measured using changes in pre/post tumor diameter on CT scans as a continuous variable ( $ 10% decrease). Results: Median follow up for progression free survival (PFS) was 24 mo (range 1–54 mo). 3-year PFS was 84% (95% CI = 71%-98%). Objective response was observed in 11 of 33 (33%) evaluable patients. Adjusted for baseline size, response was associated with improved PFS (p = 0.058). Response was also associated with higher circulating EGFR-specific T cells after neoadjuvant cetuximab (p , 0.04). FcgRIIIa VV/VF genotypes on natural killer cells were favorable; tumors shrank by median 10% vs +3.8% growth in FF expressors. Rash during window exposure was not associated with response, however was associated with FcgR genotype (FF . VF . VV; p = 0.077). Conclusions: In previously untreated HNSCC patients, response to neoadjuvant cetuximab was associated with enhanced cellular immunity against EGFR and improved PFS. Response modestly correlated with FcgRIIIa genotype but not rash. EGFR-specific T cells frequency is a novel immune biomarker of cetuximab response, warranting validation in larger cohorts. Clinical trial information: NCT01218048.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

332s 6061

Head and Neck Cancer Poster Session (Board #383), Sat, 1:00 PM-4:30 PM

6062

Poster Session (Board #384), Sat, 1:00 PM-4:30 PM

The prognostic significance of Notch 1 intracellular domain nuclear levels in head and neck squamous cell carcinoma. First Author: Theodoros Rampias, Biomedical Research Foundation of the Academy of Athens (BRFAA), Athens, Greece

Natural history following recurrence after definitive locoregional treatment in 1000+ cases of locally advanced head and neck squamous cell carcinoma. First Author: Jonathan Eric Leeman, Department of Radiation Oncology, Memorial Sloan Kettering Cancer Center, New York, NY

Background: An impressive finding of whole-exome sequencing studies in head and neck cancer is a high incidence of loss-of-function mutations within the NOTCH1gene. The vast majority of these mutations occurred at the ligand binding domains and were predicted to inhibit the proteolytic cleavage of Notch1 receptor and the translocation of its intracellular domain (Notch1ic) to nucleus. Our aim was to investigate the prognostic significance of nuclear Notch1ic for HNSCC. Methods: Nuclear Notch1ic protein levels were evaluated in tumors in a cohort of 101 HNSCC patients using automated quantitative protein analysis (AQUA). Our immunofluorescence protocol was optimized to detect only the Notch1ic and also to create a cytoplasmic tumor mask for the strict quantification of Notch1ic levels that have translocated to nucleus. The objectives were to determine the association of nuclear Notch1ic with efficacy outcomes, in terms of patients’ overall survival (OS) and disease-free survival (DFS), by Kaplan-Meier and Cox proportional regression analysis. Results: Median follow-up time was 90 months. Survival analysis highlighted the favorable prognostic significance of nuclear Notch1ic levels for patients’ outcome. Superior DFS and significantly lower risk for disease relapse was revealed for the patients with higher nuclear Notch1ic levels by Kaplan-Meier (p = 0.006) and univariate Cox regression analysis (HR: 0.398; 95% CI: 0.201-0.790; p = 0.008). Moreover, multivariate Cox regression models, adjusted for age, tumor stage, grade, nodal status, alcohol consumption, smoking behavior, age and gender highlighted the independent prognostic value of nuclear Notch1ic (HR: 0.304; 95% CI: 0.131-0.706; p = 0.006) for the favorable DFS of the patients. Finally, a trend between nuclear Notch1ic and OS of HNSCC patients was also observed. Conclusions: In the nucleus, Notch1ic binds the DNA binding protein RBPJ and recruits MAML resulting in formation of the NOTCH coactivator complex that activates transcription of genes involved in cell-cycle control and differentiation. Therefore, HNSCCs that retain nuclear Notch1ic levels may represent a district subgroup of tumors with less aggressive behavior.

Background: Squamous cell carcinomas of the head and neck (HNSCC) arising from different subsites have unique biologic features and differing clinical outcomes. Herein, we report on the natural history following recurrence in a large cohort of locally advanced HNSCC patients treated with definitive loco-regional treatment with modern radiation techniques (IMRT). Methods: We reviewed outcomes from 1062 consecutive locally advanced HNSCC patients treated with definite chemo-RT or surgery with postoperative RT +/- chemotherapy from 2001-2013 at a large cancer center, including 734 cases of oropharyngeal carcinoma (OPC), 155 oral cavity carcinoma (OCC), 126 laryngeal carcinoma (LRX) and 47 hypopharyngeal carcinoma (HPX). We analyzed outcomes across sub-sites with a focus on survival following locoregional failure (LRF) or distant metastasis (DM). Median follow-up was 57.7 months in surviving patients. Results: Crude rates of LRF and DM, respectively, by sub-site were: OPC 9.4% and 12%, OCC 25.8% and 12.9%, LRX 19.8% and 18.2%, HPX 31.9%, 29.8%. Time from LRF to death was not significantly different between sub-sites. Across all sub-sites, patients who underwent salvage surgery or salvage reirradiation following LRF experienced longer survival (surgery vs no surgery, 21.7 vs 6.0 mo, p , 0.01, re-RT vs no re-RT, 21.9 vs. 10.3 mo, p = 0.04). Time from DM to death was significantly shorter in OCC compared to other sub-sites (OCC median 3.9 months, OPC 13.4 mo, LRX 17.5 mo, HPX 7.5 mo, p , 0.01). Metastatic patients treated after 2006 experienced longer survival across all sub-sites (13.2 vs 6 mo, interquartile ranges [6.228.7] vs [3.2-13.4], p , 0.01). Conclusions: Salvage surgery and re-RT following LRF are associated with improved survival. DM from OCC portends inferior outcomes compared to other sub-sites. Metastatic patients treated after 2006, when cetuximab was FDA approved for HNSCC, experienced longer survival. Understanding the natural history of recurrent/metastatic HNSCC after definitive loco-regional treatment suggests several unique features which influence outcomes and may affect interpretation of clinical trials in these patients.

6063

6064

Poster Session (Board #385), Sat, 1:00 PM-4:30 PM

Comparison of outcomes following recurrence or metastasis in HPV+ and HPV- oropharyngeal carcinoma treated with definitive chemoradiation using IMRT. First Author: Jonathan Eric Leeman, Department of Radiation Oncology, Memorial Sloan Kettering Cancer Center, New York, NY Background: Outcomes following recurrence or metastasis of Human Papillomavirus (HPV)-related oropharyngeal carcinoma (OPC) have not been studied extensively. Herein, we report patterns of failure and the natural history following recurrence in a large cohort of locally advanced OPC patients treated with definitive loco-regional treatment using modern radiation techniques (IMRT). Methods: We reviewed outcomes from 734 consecutive locally advanced OPC patients treated with definitive chemoradiation using IMRT between July 2001 and December 2013 at a large cancer center. HPV/ p16 status was available for 417 OPC patients (56.8%) with 302 HPV+ patients (72.4%) and 115 HPV- patients (27.6%). We compared patterns of failure between HPV+ and HPV- OPC cases as well as survival following locoregional recurrence (LRR) recurrence or distant metastasis (DM). Median follow-up was 63.8 months in surviving patients. Survival rates were estimated using the Kaplan-Meier method and compared with log-rank testing. Results: Five year rates of local control, regional control, distant control and overall survival in HPV+ vs HPV- OPC were 94.6% vs. 84.2% (p = 0.001), 93.2% vs 85.9% (p = 0.011), 92.2% vs 79.1% (p , 0.001) and 86% vs 69.2% (p , 0.001). Time to progression was not significantly different between HPV+ and HPV- OPC (10.3 vs 10.6 months, respectively). The proportion of patients with isolated LRR did not differ between the two groups: 78% of LRRs were isolated in HPV+ OPC vs. 77% in HPV- OPC. Time from DM to death was not significantly different between HPV+ and HPVOPC (16.1 vs. 20.2 months, p = NS) but time from isolated LRF to death was significantly longer in HPV+ OPC compared to HPV- (median 36.5 vs 18.8 months, p = 0.04). Conclusions: HPV+ and HPV- OPC display similar proportions of local, regional and distant failures. Survival following DM is independent of HPV/p16 status but survival following LRR in HPV+ OPC is favorable compared to HPV- disease. HPV status is an important factor to consider in the design of clinical trials for recurrent OPC, particularly in the setting of isolated locoregional recurrence.

Poster Session (Board #386), Sat, 1:00 PM-4:30 PM

Prevalence and sociodemographic predictors of depression in patients with head and neck cancer - results from a national study. First Author: Nosayaba Osazuwa-Peters, Saint Louis University Cancer Center, Saint Louis, MO Background: There is an increased risk of depression after a diagnosis of head and neck cancer (HNC) due to associated morbidity and quality of life issues. However, no nationally representative study has quantified prevalence and predictors of depression across head and neck subsites. We describe age-adjusted prevalence, and sociodemographic predictors of depression among patients with HNC cancer using a national dataset. Methods: We performed a cross-sectional analysis of 76,331 cases of HNC (international classification of disease, 9th edition), using the Nationwide Inpatient Sample 2008 – 2013 data. Weighted, multivariate logistic regression analysis was used to examine association between patients’ sociodemographic and tumor anatomical site with and diagnosis of a major depressive disorder. Results: Overall prevalence of major depressive disorder in HNC was 9.2% with the highest prevalence being among patients with larynx cancer (31.3%). In the multivariate analysis, there were lower odds of depression among patients with cancer of salivary glands (odds ratio [OR] = 0.75, 95% = 0.67 – 0.85), gum (OR = 0.80, 95% = 0.67 – 0.95), and nasopharynx (OR = 0.82, 95% = 0.73 – 0.93), compared to larynx cancer. Sociodemographic predictors of depression included female gender (OR = 1.80, 95% = 1.70 – 1.90), and having Medicaid (OR = 1.09, 95% = 1.01 – 1.24) vs. private payer. Additionally, there were lower odds of depression among Blacks compared to Whites (OR = 0.57, 95% = 0.52 – 0.63), among patients in the west (OR = 0.85, 95% = 0.77 – 0.92) and south (OR = 0.82, 95% = 0.76 – 0.88, while those in the Midwest had higher odds of depression (OR = 1.12, 95% = 1.03 – 1.21). Finally, having advanced comorbidities (OR = 1.21 per one unit increase in Elixhauser comorbidity index, 95% = 1.19 – 1.23) was associated with higher odds of depression in these patients. Conclusions: Depression rates in HNC vary depending on anatomic subsite, and almost 1 out of 3 patients with cancer of the larynx may be depressed. With such a burden, it may be necessary for more psychological interventions to be integrated into clinical guidelines for the treatment HNC, particularly in high depression risk subsites such as the larynx.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Head and Neck Cancer 6065

Poster Session (Board #387), Sat, 1:00 PM-4:30 PM

Does HPV ISH add to the diagnostic performance of P16 IHC in oropharyngeal cancer? Is it worth the extra cost? First Author: Nagashree Seetharamu, Hofstra Northwell School of Medicine, Lake Success, NY Background: Human Papilloma Virus (HPV) related oropharyngeal squamous cell cancer (OPSCC) is distinct from that attributable to tobacco use, with better prognosis and treatment response. While HPV status doesn’t currently influence treatment decisions, clinical trials are exploring deintensification strategies for HPV+ OPSCC. While no gold standard tests exist, immunohistochemistry (IHC) for P16, which is a surrogate marker for HPV and HPV DNA in-situ hybridization (ISH) are commonly used for determining HPV status. Discordance between these tests is frequent, and no guidelines exist with regards to their use. As we advance towards HPVspecific protocols, accurate assessment of HPV status is essential. We evaluated the concordance of HPV P16 and HPV ISH for OPSCC, with special focus on never-smokers, in who tumors are likely HPV-related. Methods: Charts of patients with OPSCC diagnosed between 2012 and 2014 were reviewed. Data included age, gender, stage, smoking, primary site, and HPV status. Descriptive statistics were used to summarize the data. Results: 50 patients were analyzed. 39 were male (78%) and median age was 61.5 years. Most common primary site was base of tongue (n = 28, 56%) followed by tonsil (n = 19, 38%). Most patients had advanced disease: 16 stage III (32%), 31 stage IVA (62 %), 1 stage IVB (2%). P16 IHC was performed in 48 (96%) and ISH in 43 (86%). Wide spectrum (WS) ISH was utilized in 23 (46%) and HPV 16/18 ISH in 23 (46%). P16 stained strong and diffuse in 35 (70%) and weak in 5 (10%). HPV ISH was positive in 20 (40%); 9 WS and 11 HPV 16/18. All HPV ISH positive tumors stained for P16 (100%). Amongst never-smokers (n = 20), P16 was diffusely positive in 19/19 tested (100%), whereas HPV ISH was positive in 8/17 (47%). Conclusions: Discordance between P16 IHC and HPV ISH is high. P16 is more sensitive in never-smokers. We conclude that HPV ISH has low utility and recommend the utilization of P16 IHC only for diagnostic and prognostic purposes in OPSCC pending availability of improved testing techniques.

6067

Poster Session (Board #389), Sat, 1:00 PM-4:30 PM

Risk of distant metastases in head and neck carcinoma patients and myeloperoxidase (MPO) expression. First Author: Antonio Lopez-Pousa, Hospital Sant Pau, Barcelona, Spain Background: Tumor-associate neutrophils (TAN) represent a significant portion of the inflammatory cell infiltrate in head and neck carcinoma (HNSCC). Several studies have shown a significant association between TAN and poor clinical outcome in HNSCC patients. Myeloperoxidase (MPO) is a peroxidase enzyme expressed in TAN. MPO expression could be a biomarker for the identification and quantification of TAN. The objective of our study was to evaluate the transcriptional expression of MPO in HNSCC tumors, and to evaluate the relationship between MPO expression and the appearance of distant metastases Methods: RT-PCR was used to determine mRNA expression levels of MPO in 160 consecutive patients with HNSCC treated between 2004-2011. Continuous values of mRNA expression were analyzed using a classification and regression tree (CART) method considering the appearance of distant metastases as the dependent variable. During the followup period, 23 patients (14.4%) presented with distant metastases. Results: mRNA expression level of MPO for patients with distant metastases was significantly higher than for patients without (P = 0.007). The CART method classified patients in two categories: high (n = 60, 37.5%) or low (n = 100, 62.5%) expression level of MPO. The 5-year distant metastasis free survival was 66.7% for MPO high-expression level and 93.3% for MPO lowexpression level (P , 0.001) patients. Considering patients with a low MPO expression as the reference group, patients with a high MPO expression had a 4.5 fold higher risk of appearance of distant metastases (CI 95%: 1.7-11.6). Patients with high risk of distant based on clinical and pathological variables, including patients with locally advanced tumors (cT3-T4), with advanced regional spread (cN2-N3), or with neck node metastases with extracapsular spread, who had a high MPO expression (n = 42) had a 5-year distant metastasis free survival of 59.2%. Conclusions: We found a significant relationship between mRNA MPO expression level and the appearance of distant metastases in HNSCC patients. The combination of clinicalpathological data with the MPO expression can be used to identify patients with a high risk of distant metastases

6066

333s Poster Session (Board #388), Sat, 1:00 PM-4:30 PM

Analysis of chemotherapy selection for locally advanced squamous cell carcinoma of head and neck (SCCHN) in a commercially insured population in the United States. First Author: Hyunseok Kang, Johns Hopkins University School of Medicine, Baltimore, MD Background: Multiple standard of care options exist for locally advanced SCCHN. We analyzed treatment plans submitted for commercially insured patients by practicing medical oncologists for pre-authorization to look into patterns of practice. Methods: Consecutive chemotherapy treatment plans for a population of 2 million covered lives across the US for newly diagnosed oral cavity, oropharynx, larynx and hypopharynx cancers submitted to insurers using the eviti Connect portal for chemotherapy pre-authorization between June 1, 2011 and June 30, 2015 were analyzed. Results: A total of 393 treatment plans were submitted for 340 newly diagnosed patients; 68 (20%) patients were from academic centers (AC) and 272 (80%) patients were from community practices (CP). Plans for chemotherapy before and during radiotherapy were identified in 61 (18%) patients, plans for definitive concurrent chemoradiation were identified in 217 (64%) patients, and plans for adjuvant chemoradiation after surgery were identified in 62 (18%) patients; no significant difference in treatment approach was seen between AC and CP. Only 9 (15%) patients at AC submitted plans for clinical trials. CP proposed upfront concurrent chemoradiation in larynx SCC more frequently than AC (79% vs 56%, p = 0.06). Single agent cisplatin, single agent cetuximab and single agent carboplatin were the 3 most commonly proposed regimens concurrent with radiotherapy, and there was no significant difference between AC and CP, although CP tended to use cetuximab more often in non-oropharynx SCCs (24% vs 10%, p = 0.16). For oropharynx SCC, unknown HPV status was reported more frequently in CP compared to AC (51% vs 31%, p = 0.04). In logistic regression, there were tendencies for more cetuximab use in older patients (OR 1.05, p = 0.09) and following induction chemotherapy (OR 2.03, p = 0.11). Conclusions: Among commercially insured patients with locally advanced SCCHN, clinical trials are not commonly proposed even at AC. HPV status was not routinely reported in oropharynx SCC, especially in CP. Single agent cisplatin continues to be the most commonly used regimen for concurrent chemoradiation in both AC and CP.

6068

Poster Session (Board #390), Sat, 1:00 PM-4:30 PM

The prognostic role of gender, race and human papillomavirus (HPV) in oropharyngeal squamous cell cancer (OPC) and non-oropharyngeal head and neck squamous cell cancer (non-OP HNC). First Author: Carole Fakhry, Johns Hopkins University, Baltimore, MD Background: HPV tumor status is a well-established prognostic marker in OPC. Gender and racial/ethnic differences in prognosis have not been well explored in OPC due to limited numbers of women and non-whites in studies. Additionally, the prognostic role for HPV in non-OP HNC remains controversial. Therefore, survival differences were explored among OPCs by gender and race in the context of HPV tumor status and non-OP HNCs by HPV tumor status. Methods: Retrospective multi-institution study including OPCs and non-OP HNCs of oral cavity, larynx and nasopharynx diagnosed 1995-2012. Women and non-whites were oversampled. Race/ethnicity categories used included White non-Hispanic, Black non-Hispanic and Asian/Hispanic of any race. HPV tumor detection consisted of p16 immunohistochemistry and HPV in situ hybridization (ISH; for HPV16 and any high-risk HPV). Tumors that were HPV16 and/or high-risk positive were considered HPV ISHpositive. Kaplan Meier and Cox proportional hazards models were used to evaluate overall survival (OS). Results: Study population included 239 OPCs and 625 non-OP HNCs with median follow-up time of 42.5 and 42.3 months, respectively. Univariate factors associated with better OS for OPC included younger age, lower stage, female gender, Asian/Hispanic race/ ethnicity, not currently smoking, and HPV-positive tumor status (ISH: HR 0.34, p , 0.001; p16: HR 0.46, p = 0.01). After controlling for HPV ISH, age, current smoking, and stage, risk of death was significantly lower for women relative to men (HR 0.54, p = 0.04) and Asian/ Hispanics compared to Whites (HR 0.35, p = 0.06). Among non-OP HNCs, univariate factors associated with OS included age, stage, race, current smoking and alcohol use, and anatomic site. Tumor HPV ISH and p16 were not associated with OS. In multivariate analysis, significant risk factors for better OS included younger age, lower stage and not currently smoking (each p , 0.01). Conclusions: In OPC, there are marked differences in survival by gender and race, even after accounting for HPV tumor status. In non-OP HNCs, HPV tumor status whether measured by ISH or p16 has no prognostic significance.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

334s 6069

Head and Neck Cancer Poster Session (Board #391), Sat, 1:00 PM-4:30 PM

6070

Poster Session (Board #392), Sat, 1:00 PM-4:30 PM

The evidence on effectiveness of upper neck irradiation vs whole neck irradiation as elective neck irradiation among node negative nasopharyngeal cancer: A meta analysis. First Author: Jayson L. Co, University of Santo Tomas Hospital, Manila, Philippines

Sequential boost versus integrated boost intensity-modulated radiation therapy with concurrent chemotherapy for locally-advanced head and neck cancer (LAHNC). First Author: Anthony Cmelak, Vanderbilt-Ingram Cancer Center, Nashville, TN

Background: Nasopharyngeal carcinoma (NPC) is a central tumor with rich lymphatic network and a propensity for bilateral cervical lymph node metastasis. There is an orderly pattern of lymph node involvement in NPC as opposed to other head and neck squamous cell carcinoma where skip metastasis is present. These lymph nodes are prophylactically irradiated even in node negative disease. However, there is no current standard for prophylactic neck irradiation in node negative or limited retropharyngeal node (RP) positive NPC. Limiting radiation to the upper neck spares lower neck soft tissues and thyroid gland. The aim of this paper is to synthesize the current evidence regarding effectiveness of upper neck irradiation (UNI) versus whole neck irradiation (WNI) as prophylactic neck irradiation in node negative or limited RP node positive NPC. Methods: Search of relevant articles were done from 2000 to October 2015. Critical appraisal and metaanalysis of the eligible studies were undertaken to assess the effectiveness of UNI versus WNI as prophylactic neck irradiation in node negative or limited involved retropharyngeal node NPC. Results: One randomized controlled trial and 4 retrospective cohorts were included in this meta-analysis. Only one randomized controlled trial investigated on the use of prophylactic UNI versus WNI and showed no confirmed nodal relapse in both arms. Pooled analysis of four retrospective studies showed an odds ratio of 1.68 (CI- 0.644.40) in nodal recurrence, whether in-field or out-of-field recurrence. There was no significant difference on the rate of nodal recurrence when UNI versus WNI as prophylactic node irradiation was used for node negative or limited RP positive NPC. Pooled data from three studies showed no significant difference in 5-year distant metastasis and overall survival between UNI and WNI with an odds ratio of 1.09 (0.58-2.03) and 0.99 (O.60-1.63) respectively. Conclusions: In node negative or limited RP positive NPC, it may be a reasonable option to treat only the upper neck (levels II, III and VA) without compromising nodal control, distant metastasis and overall survival.

Background: Despite IMRT technological advances, toxicity in LAHNC pts remains extreme. Different IMRT techniques have been developed, including sequential boost (SeqB) and simultaneous integrated boost (SIB). However, the relative efficacy and toxicities of each technique remain unexplored. Methods: 209 pts with LAHNC were treated with SeqB (n = 68) or SIB (n = 141). A matched cohort multi-institution analysis was used. All 209 pts received 69.3Gy in 33 fxs of 2.1Gy to gross disease. SiB pts received 1.6Gy daily in 33 fxs (52.8Gy) to prophylactic nodes; SeqB pts received 2.1Gy for 24 fxs (50.4Gy) to gross disease and prophylactic nodes, followed by a boost of 18.9Gy in 9 fxs to gross disease. All pts received concurrent chemotherapy, 95% with weekly paclitaxel (40mg/m2) and carboplatin (AUC 1). Targets and organs at risk (OARs) contours, dose constraints, dosimetric plan evaluation, and toxicity assessment were alike and performed by a single experienced physician. Weekly toxicities were graded using the CTCAE. Tumor control and survival were estimated via KaplanMeier. Results: Median FU was 30.6 mo. At 3 years, OS was 69% for SeqB and 79% for SIB cohort (p = 0.13). 3-yr DFS was not different (63% vs. 73%; p = 0.27). There were no significant differences in local, regional, or distant RFS. There were no differences in weight loss (p = 0.291), incidence of Gtube (p = 0.494), or in prolonged G-tube dependence (p = 0.465). Rates of grade 3 or 4 dysphagia were higher for SiB than for SeqB (81% vs. 55%, , 0.001) as well as dermatitis (78% vs. 58%, p = 0.012). Fewer SIB pts received the total 69.3Gy IMRT dose due to acute toxicity (93% vs 100%; p = 0.03). OARs mean dose, Dmax, and percent volume receiving 30Gy and 70Gy were not statistically different between the two cohorts. Conclusions: There were no significant differences in tumor control or survival outcomes between the two IMRT delivery approaches. The SeqB IMRT technique resulted in lower grade 3 and 4 acute radiation dermatitis and dysphagia compared to SiB, and more pts completed treatment with SeqB than did pts treated with SiB IMRT. SeqB should be considered for LAHNC to lower treatment toxicity and improve delivery.

6071

6072

Poster Session (Board #393), Sat, 1:00 PM-4:30 PM

Phase II prospective trial of cetuximab and radiotherapy for locally advanced, squamous cell carcinomas of the head and neck in patients .70 years old or with comorbidities not-eligible for platinum-based chemotherapy. First Author: Paul Swiecicki, University of Michigan Health System Comprehensive Cancer Center, Ann Arbor, MI Background: We designed a single arm phase II trial to determine the efficacy and toxicity of concurrent cetuximab (C) radiotherapy (RT) in patients non-eligible for standard chemotherapy. Non-eligibility was defined as age . 70 years (yrs), co-morbidities that preclude platinum-based chemotherapy, or creatinine clearance (CrCl) , 30 mL/min. Primary endpoint was progression free survival. Methods: 22 stage III/IVA patients (pts) were enrolled. The median age was 70 years (range: 40-85), 50% had eGFR with , 90 mL/ min, 64% were males, 36% were females; 64% Caucasian, 36 % had an ECOG performance status (PS) of 0, 64% ECOG PS of 1. All pts received C 400 mg/m2 day (d) 0, then 250 mg/m2 d 7 and weekly during RT. Pts were treated with a planned 70 Gray (Gy) in 35 fractions over 7 weeks to the gross tumor and 50-60 Gy to subclinical target volumes.Results: As of January 2016 with a median follow up of 22.8 months and 22 evaluable pts, the median PFS was 30.07 months (95% CI: 2.93, . 50). The overall survival was 33.8 months (95% CI: 9.2, . 50). 6 pts (27%) have recurred, all of whom have died. Treatment related grade 3 toxicities were seen in 7 pts (32%), no grade 4 toxicities were seen. The most common grade 3 toxicities were mucositis (5 pts), dysphagia (4 pts), and dermatitis (3 pts). Conclusions: The results of this phase II prospective study suggest that in patients . 70 yrs old or with comorbidities, precluding platinum-based chemotherapy, tolerate treatment with C+RT well, though survival rates are lesser when compared with historical controls. Clinical trial information: NCT00904345.

Poster Session (Board #394), Sat, 1:00 PM-4:30 PM

Tumor’s flare-up and patterns of recurrence in patients (pts) with recurrent and/or metastatic (R/M) head and neck squamous cell carcinoma (HNSCC) treated with anti-PD1/PD-L1 inhibitors. First Author: Esma Saada-Bouzid, Centre Antoine Lacassagne, Nice, France Background: About 20% of R/M HNSCC pts have an objective response with anti-PD1/PD-L1 inhibitors after platinum failure. Some pts, however, experience a rapid flare-up during treatment. We evaluated the relationship between a flare-up occurence and the patterns of recurrence in R/M HNSCC pts treated with anti-PD1/PD-L1. Methods: Data from R/M HNSCC pts treated with anti-PD1/PD-L1 in 4 French Cancer Centers between 09/2012 and 09/2015 were retrospectively collected. Pts had an imaging performed at least 2 weeks prior to baseline imaging to estimate tumor growth kinetics (TGK) before and on immunotherapy. TGK before starting immunotherapy (TGKPRE ) was defined as the difference of the sum of the largest diameters of the target lesions (according to RECIST) per unit of time between prebaseline and baseline imaging, while TGK on immunotherapy (TGKPOST ) was defined as the same difference between baseline and last imaging performed on treatment. The TGK ratio (TGKR ) was defined as TGKPOST / TGKPRE . A flare-up was defined as a TGKR . 1, while 0 , TGKR # 1 and TGKR , 1 indicated tumor deceleration and tumor shrinkage, respectively. An exact Fisher test was used for statistical comparison. Results: Thirty-five pts with available pre-baseline imaging and evaluable disease were identified. Median age was 63 years [35-84], 28 pts were male (80%). Primary tumor location was oral cavity, oropharynx, larynx, and hypopharynx in 12 (34%), 12 (34%), 5 (14%), and 6 (17%) pts, respectively. Patterns of recurrence included exclusive distant metastases in 11 pts (31%), exclusive locoregional recurrence in 14 pts (40%), and both in 10 pts (29%). Ten pts (29%) experienced a flare-up, while 8 pts (23%) had tumor shrinkage and 17 pts (49%) a deceleration of tumor growth. Interestingly, a flare-up was observed in none of the 11 pts (0%) with exclusive distant metastases, in 5/ 14 pts (36%) with exclusive loco-regional recurrence, and in 5/10 pts (50%) with loco-regional recurrence and distant metastases (p = 0.06). Conclusions: Our preliminary results suggest that anti-PD1/PDL1 should be used cautiously in R/M HNSCC pts with loco-regional recurrence.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Head and Neck Cancer 6073

Poster Session (Board #395), Sat, 1:00 PM-4:30 PM

Treatment of laryngeal cancer: The University of Michigan experience. First Author: Francis P. Worden, University of Michigan Comprehensive Cancer Center, Ann Arbor, MI Background: Previous success of Phase II trials from 1995- 2005, led us to develop personalized treatment selection based on tumor & patient (pt) characteristics using neoadjuvant bioselection—1 cycle of chemotherapy (NB) to select pts with stage III/IV cancer for laryngectomy (L) or chemoradiation (CRT). Methods: From 2002-2012, 247 previously untreated laryngeal cancer pts were treated. Treatment of 95 early disease (Stage I/II) pts included endoscopic resection (35%), RT alone (53%) or CRT for deeply invasive T2 lesions (12%). Of 152 pts with stage III/IV disease, NB was utilized in 71(47%) & CRT in 49 (32%) pts refusing NB. Primary L was recommended for severely compromised larynx in 26 (21%) or candidates for supraglottic L (n = 6). Propensity of treatment approach for advanced disease pts was modeled using logistic regression modeling & overall (OS) & disease specific survival (DSS) were analyzed with Cox proportional hazards models stratified by propensity score. Median follow up: 48 months.Results: Estimated 5 yr OS & DSS was 75% (95% C.I. 68-81%) & 83% (70-85%), respectively for the entire cohort. DSS was 91% (81-96%) for pts with Stage I/II & 78% (70-85%) for pts with stage III/IV disease. OS did not differ for stage I/II vs III/IV pts (p = 0.24). For 152 advanced disease pts, 5 year OS (& DSS) ranged from 78% (78%) for primary L to 76% (79%) for NB and 63% (68%) for primary CRT. Larynx preservation was 77% for primary CRT & 66% for the NB group. A propensity-stratified Cox proportional hazards model controlling for known prognostic factors showed NB was associated with a 0.33 hazard reduction [95% C.I., 0.13,0.89], (p = .03); current smoking & stage IV disease were associated with increased hazard in the multivariable model. Among non-NB pts, the multivariable propensity-stratified model showed surgical treatment was associated with significantly reduced hazard (0.09 [0.02, 0.51] p = 0.007)—trends for worse prognosis in older & stage IV pts. Conclusions: Our results confirm high long-term survival rates for an individualized treatment approach, particularly for Stage III/IV pts with NB. Future challenges include improving outcomes for pts with T2 cancers, decreasing rates of salvage L in pts with advanced disease & identifying predictive biomarkers.

6075

Poster Session (Board #397), Sat, 1:00 PM-4:30 PM

Neoadjuvant carboplatin, nab-paclitaxel and cetuximab prior to standard of care chemoradiotherapy for N2b or unresectable squamous head and neck cancer. First Author: Jared Weiss, Lineberger Comprehensive Cancer Center at the University of North Carolina, Chapel Hill, NC Background: Although induction studies of TPF have been negative in locally advanced SCCHN, phase II studies of weekly carboplatin (CbP), paclitaxel and cetuximab (C225) have shown promising results. Nab-paclitaxel-based chemotherapy has demonstrated a higher response rate (RR) than solventbased paclitaxel in SCC of the lung with favorable toxicity. We evaluated an induction regimen using nab-paclitaxel, CbP and C225 followed by concurrent chemoradiation. Methods: Patients with treatment na¨ıve SCCHN of any site with $ N2b disease or that was unresectable by strict criteria were eligible. Subjects were screened for risk of anaphylaxis to C225 with ELISA for IgE antibodies against galactose-a-1,3-galactose (alphagal); negativity was required for subjects residing in the SE of the US. Patients were treated with nab-paclitaxel 100mg/m2, CbP AUC 2 and C225 400mg/m2 week 1 then 250mg/m2for six weeks, followed by standard of care chemoradiotherapy (CRT). Results: 39 subjects were treated and 30 had a response (76.9%); there was no progressive disease. Primary sites were: oropharynx (OPX) (26), larynx (3) oral cavity (OC) (9), hypopharynx (1). Grade 3/4 toxicity included rash (11), decreased neutrophil count (4), decreased white blood cells (2), fatigue (1), palmar-plantar erythrodysesthesia syndrome (1), febrile neutropenia (1), hypocalcemia (1), and hypokalemia (1); there was no anaphylaxis to C225.1 year survival is 89% (95% CI 73-96%) and two-year survival is 80% (62-90%). Median followup for survival for living patients is 1.82 years. 1 year PFS was 81% (64-90%). Median followup for PFS for patients without progression or death is 1.49 years. Amongst p16+ oropharynx patients, 1 year PFS and OS were both 89% (64-97%). Amongst all other patients, 1 year PFS and OS were 74% and 73%, respectively. Conclusions: The combination of nab-paclitaxel, CbP and C225 is tolerable and active against locally advanced SCCHN. Phase III study of this induction regimen followed by standard of care chemoradiotherapy versus standard of care chemoradiotherapy alone is indicated. Clinical trial information: NCT01412229.

335s

6074

Poster Session (Board #396), Sat, 1:00 PM-4:30 PM

High-dose versus weekly cisplatin definitive chemoradiotherapy for HPVrelated oropharyngeal squamous cell carcinoma of the head and neck. First Author: Cesar Augusto Perez, James Graham Brown Cancer Ctr, Louisville, KY Background: Definitive concurrent chemoradiation (CRT) is a standard treatment for patients (pts) with squamous cell carcinoma of the oropharynx (SCCOPx). We compared the outcomes and toxicity of high-dose cisplatin (HDC) versus weekly (WC) definitive CRT for patients with HPV-related SCCOPx. Methods: All pts with p16 positive SCCOPx treated with definitive CRT with cisplatin between 2010 and 2014 at a single institution were reviewed. CTCAE v 4.03 toxicity criteria was used. The Kaplan-Meier method was used to estimate event-free survival (EFS) and the overall survival (OS). Results: Of the55 pts included, 85% were males and 57% had a history of smoking . 10 pack-years. Median age at diagnosis was 55.4 years (40.3-80.0 years). 22 pts were treated with HDC at doses of 100 mg/ m2 on days 1 and 22; 33 were treated with WC at 40 mg/m2. Groups were well balanced in respect to sex (p = 0.454) and smoking history (p = 0.799). The median total dose of cisplatin was 200 mg/m2 for both cohorts, with a mean of 195 mg/m2 for the HDC group and 189 mg/m2 for the WC group. At median follow-up of 27 months, there was one local failure and no distant failures in the HDC cohort. In the WC group, there were 6 total failures (2 local and 4 distant). Estimated 2-year EFS was better in HDC cohort as compared to WC (96% vs. 75%; p = 0.05), despite a longer follow-up in pts with HDC (32 months vs. 19 months). There was no significant difference in 2-year OS (95% vs. 94%; p = 0.4). In terms of toxicity, weight loss was marginally higher in the HDC arm compared to WC arm; all weight loss grades (100% vs 87%, p = 0.12) and weight loss . grade 1 (77% vs 60%; p = 0.19). However, gastric tube dependence at six months following therapy was similar (13% vs. 12%, p = 1.0). Acute renal injury of any grade was higher in HDC group (55% vs. 36%, p = 0.18) but grade 3 or 4 hematological toxicity was similar (20% vs 23%, p = 0.76). Conclusions: HPV-positive SCCOPx treated with definitive CRT with either HDC or WC had similar toxicity profile for several parameters. HDC has better EFS when compared with WC and this seems to be driven by increased distant failure rates, although the OS was similar. A longer follow-up will be needed to better assess if the OS remains similar.

6076

Poster Session (Board #398), Sat, 1:00 PM-4:30 PM

TPF plus cetuximab (E) induction chemotherapy (ICT) followed by biochemoradiation (BCRT) with weekly E plus weekly cisplatin (P) or carboplatin (C) in stage III/IV squamous cell carcinoma of the head and neck (HNSCC): A randomized phase II EORTC trial. First Author: Pol M. Specenier, Antwerp University Hospital, Edegem, Belgium Background: To test the safety of BCRT with P and C after TPF-E. Methods: Stage III/IV unresectable, HNSCC pts received up to 4 cycles (cy) of TPF-E (P and docetaxel [D] 75 mg/m² d 1, 5-FU 750 mg/m²/d d1-5, cetuximab [E, 400/250 mg/m²/ wk]), with prophylactic antibiotics, no G-CSF. Pts with no PD after 4 cy were randomly assigned to BCRT (70 Gy/7 weeks/2 Gy fractions), weekly E, with either P 40 mg/m²/wk or C AUC 1.5/wk. Primary endpoint: feasibility of BCRT ( . 80% dose intensity (DI) of RT, P or C, and E). Results: 46 patients started TPF-E, 30 started BCRT. Median age: 57 (48-72), 41 male. WHO 0/1: 29/17. Stage III/IV: 4/42. 4 oral cavity, 24 oropharynx, 12 hypopharynx, 4 larynx, 2 unknown. 34, 4, 6, 2 pts received 4,3,2, and 1 TPF-E cy. Median DI (mg/m²/wk): D: 24.4, P: 24.4, 5-FU: 1220.9, E: 255.4. ICT was discontinued in 12 pts (6 toxicity [1 grade 5], 1 protocol violation, 2 medical decision, 1 pt refusal, 1 death [cause unknown], 1 cerebrovascular accident). 3 were not randomized: 1 PD under ICT, 1 RT start prior to randomization, 1 comorbidity/TPF-E toxicity. Median DI on BCRT (mg/m²/wk): P arm: P: 34.4, E: 238.1; C arm: C: AUC 1.5/wk, E: 243.2. P and C were stopped early in 7 and 4 pts, respectively. RT was stopped in 2 pts with P and interrupted in 1 and 4 pts with P and C, respectively. Conclusions: Only 34/46 pts completed 4 TPF-E. 30/46 started BCRT. 53 % and 47 % of the randomized pts failed to meet the feasibility criteria with P and C, respectively. Study closed prematurely. Clinical trial information: NCT00646659.

Selected grade 3-5 toxicity. ICT

BCRT

N = 46 grade 3 Skin Diarrhea Constipation Mucositis Nausea Vomiting Gastrointestinal other Febrile neutropenia Infection Neurologic Pneumonitis Anemia Leukopenia Neutropenia Thrombocytopenia Renal failure Hyponatremia Hypokalemia Hypocalcemia Hypercalcemia Hypomagnesemia Hearing loss Fatigue Weight loss Sinustachycardia Hypertension Hypotension Left ventricular dysfunction Cardiac other

4 1 2 3 3 2 4 3 6 3 4 17 11 2 9 6 7 6

grade 4

grade 5

Grade 3

P

C

N = 15

N = 15 Grade 4

3 1 6 1

1 3* 1 2 1 2

Grade 3

Grade 4

4

1

3 2 6 1 3 1

9 7 1

1

1

7 14 1

6 2 1

1 4 3 3 5

4 1

1 1

2 1 1

1 2 1 3 2

6 1

2

2 1 1

1 1 1 1 1

*: 2 colon perforations, 1 ileus

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

336s 6077

Head and Neck Cancer Poster Session (Board #399), Sat, 1:00 PM-4:30 PM

Changes in body composition and prognostic importance of sarcopenia in patients receiving CTRT for oropharyngeal cancer. First Author: Shrujal S. Baxi, Memorial Sloan Kettering Cancer Center, New York, NY Background: Concurrent chemoradiation (CTRT) improves outcomes in oropharyngeal cancer (OPC) but causes toxicities that alter body composition. In other cancers, loss of lean body mass (sarcopenia) is associated with poor prognosis. Changes in body composition in OPC, and the clinical importance of these alterations, have not been studied. Methods: Using a retrospective design, 234 OPC patients (mean age, 58 years (range 27 to 83); mean BMI, 29 kg/m2) receiving definitive CTRT were studied. Skeletal muscle cross-sectional area (CSA) at the third lumbar vertebra (L3) and thigh were assessed by CT-scan prior to and after completion of CTRT (median: 13.5 wks; range 2-27) and normalized to stature to obtain skeletal muscle index (cm2/m2). Sarcopenia was defined using previously published cutoffs. Cox models were used to assess the relationship between sarcopenia and recurrence-free survival (RFS) and overall survival (OS). Wilcoxon rank-sum tests were used to assess changes in percent loss in weight and L3 and thigh muscle CSA in patients hospitalized post-CTRT. Results: Prior to CTRT, 68% of patients were classified as sarcopenic. HPV-testing was completed in 168 (72%); 149 were positive. Post-CTRT, body mass (kg) decreased a median of 13.8% (IQR: 10.1% to 18.7%, p , .001) while muscle CSA at L3 and thigh decreased by 11.2% (IQR: 6.9% to 16.8%, p , .001) and 13.6% (IQR: 5.5% to 21.2%, p , .001), respectively; 89% were classified as sarcopenic. Older age and male were significant baseline predictors of sarcopenia postCTRT (p’s , 0.05). Fifty-eight (25%) patients were hospitalized within 3 months of CTRT. Median follow-up was 34 months; in this period, 26 recurrences and 19 deaths were observed. Greater percent weight loss and thigh muscle CSA loss were positively associated with risk of hospitalization (p’s , 0.01); baseline or post-treatment sarcopenia were not associated with RFS or OS (p’s . 0.05). Conclusions: A significant proportion of OPC patients present with clinical sarcopenia, while CTRT increases the proportion of subjects with this condition.Decreases in body and muscle mass are associated with hospitalization. The long-term clinical impact of these losses remains to be determined.

6079

Poster Session (Board #401), Sat, 1:00 PM-4:30 PM

6078

Poster Session (Board #400), Sat, 1:00 PM-4:30 PM

Poor prognostic factors in human papilloma virus-positive head and neck cancer: Who should not be candidate of de-escalated treatment? First Author: Shin Hye Yoo, Department of Internal Medicine, Seoul National University Hospital, Seoul, Korea, The Republic of Background: Human papilloma virus (HPV)-associated locally advanced head and neck squamous carcinoma (LA HNSCC) shows favorable treatment outcome, and treatment de-escalation is actively under-investigated. However, not all the HPV+ LA HNSCCs are curable, and some of them shows poor prognosis. The purpose of this study is to identify poor prognostic factors in HPV+ LA HNSCC patients. Methods: Patients who were diagnosed as LA HNSCC and tested for HPV status from 2000 to 2015 were included in this retrospective analysis. Demographic data (age, gender), smoking status, comorbidity, and clinical information about stage, pathology, type of definite treatment were obtained. HPV typing was conducted using HPV DNA genotyping chip or liquid bead microarray. Expression of p16 was assessed by immunohistochemistry. Univariate and multivariate analyses were done with factors associated with overall survival (OS) by Cox regression analysis. Results: One hundred fifty two HPV+ LA HNSCC patients were included in this study. Male patients were 82.2% and 43.4% were current or ex-smoker. Oropharyngeal cancer was 84.2%. In univariate analysis, old age, poor performance, non-oropharynx, advanced T stage (T3, 4), N0 stage, HPV genotype 18 were significantly associated with poor OS. Smoking was not associated with OS. In multivariate analysis, poor performance (ECOG PS$1), non-oropharyngeal location were independently associated with shorter overall survival (HR 4.82, p = 0.003; HR 7.36, p = 0.001, respectively). Furthermore, HPV genotype 18 positivity was also independent poor prognostic factor for OS in HPV+ LA HNSCC (HR 11.41, p , 0.001). Conclusions: Non-oropharyngeal cancer, poor performance status, HPV genotype 18 were independent poor prognostic factors in HPV+ LA HNSCC. These patients should not be candidate of de-escalated treatment.

6081

Poster Session (Board #403), Sat, 1:00 PM-4:30 PM

Prophylactic antibiotics to prevent pneumonia in patients treated with chemoradiotherapy (CRT) for locally advanced head and neck carcinoma (LAHNC). First Author: Janneke Ham, Radboudumc, Nijmegen, Netherlands

Impact of age on treatment outcomes of head and neck squamous cell carcinoma (HNSCC) patients. First Author: Jitlada Juengsamarn, Ramathibodi Hospital, Bangkok, Thailand

Background: Platinum-based CRT is the treatment of choice for LAHNC, but induces a high rate of acute toxicity, including dysphagia and aspiration pneumonia. Aspiration pneumonia often requires hospitalization and negatively influences quality of life (QoL). We hypothesized that prophylactic antibiotics during CRT in LAHNC patients (pts) can prevent pneumonia and hospital admissions. Methods: A multicenter randomized trial was performed. The standard treatment group (STG) received standard care; the intervention group (IG) received prophylactic amoxicillin/clavulanic acid from day 29 until 14 days after completion of CRT. LAHNC pts treated with CRT were registered before start and randomized between day 21-28, unless they received antibiotics 14 days before planned randomization. The primary objective was to measure a reduction in the rate of pneumonias. A confirmed pneumonia was defined as a previously non-existing infiltrate on chest X-ray or $ 3 of the following features: sustained fever, rales or rhonchi on chest auscultation, sputum with substantial leukocytes or sputum culture showing a pathogen. A suspected pneumonia was defined as 2 of these features. Results: A total of 106 pts were registered, of which 94 pts were randomized: 48 pts were allocated to the STG and 46 pts to the IG. Pts were equally randomized by minimization for smoking, COPD status, HPV status in oropharyngeal cancer, degree of weight loss, primary tumor site, participating centre and tube feeding at time of randomization. A confirmed or suspected pneumonia during CRT and follow up till 3.5 months was observed in 20 (41.7%) of 48 pts in the STG, and 20 (43.5%) of 46 pts in the IG, respectively (P = 0.56). Hospitalization for any cause (i.e., fever, dehydration, mucositis or pneumonia), was necessary in 17 of 48 pts (35.4%) in the STG versus 9 of 46 pts (19.6%) in the IG (P = .08). Conclusions: Prophylactic antibiotics during CRT for LAHNC did not reduce the rate of pneumonias. However, there is a trend towards a reduced rate of hospitalizations, which potentially may lead to a better QoL and reduction in costs. Currently, further microbiological and cost-effectiveness analyses are being performed. Clinical trial information: NCT01598402.

Background: Impact of age on treatment outcomes of head and neck squamous cell carcinoma (HNSSC) patients. Methods: We retrospectively HNSCC patients treat at Ramathibodi hospital between 1/2008 and 12/ 2013. Nasopharyngeal carcinoma was excluded. We evaluated the effect of age including other risk factors on treatment outcomes. Results: Total of 601HNSCC patients with median age of 66 years (range 25-95) was identified. At the time of diagnosis, 43% of patients were $ 65 years, while 26% and 16% were . 70 and . 75 years respectively. In elderly patients ( $ 65years), 58% presented with locally advanced stage (stage III-IV). Increase of age was a worse prognostic factor for OS with HR 2.93 (95%CI 2.3-3.7, P , 0.01). In multivariate analysis, hypoalbuminemia was a poor prognostic factor for overall survival (OS). In locally advanced HNSCC elderly patient, chemoradiotherapy (CRT) did not improve OS when comparing with radiotherapy alone, HR 0.87 (95% CI 0.58-1.30, P 0.37). Conclusions: Age was associated with poor OS. Hypoalbuminemia was a poor prognosis factor for OS. CRT in locally advanced HNSCC elderly patients did not improved OS when compared with radiotherapy alone. Median OS Overall (n = 601) Primary site Oral cavity (39%) Oropharynx (19%) Hypopharynx (14%) Larynx (22%)

·· ·· · · · · · ·

$ 65 years vs , 65 years (n = 339) BMI in elderly Charlson score in elderly

Hazard ratio

95% Confidence interval

28.6 (0.03-93)

67.3 vs 20.2

2.93 0.66

(2.3-3.7) P , 0.01 (0.28-1.57) P 0.35

0.84

(0.73-0.97) P 0.02

Albumin in elderly

0.95

(0.92-0.97) P 0.01

Surgery in elderly

0.72

(0.55-0.93) P 0.01

CRT vs RTalone in elderly CRT Cisplatin vs non-cisplatin regimen

0.87 0.87

(0.58-1.30) P 0.37 (0.53-1.43) P 0.80

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Head and Neck Cancer 6082

Poster Session (Board #404), Sat, 1:00 PM-4:30 PM

A network meta-analysis of the sequencing and types of systemic therapies with definitive radiotherapy in locally advanced squamous cell carcinoma of head and neck (LASCCHN). First Author: Keemo Delos Santos, Sunnybrook Research Institute, Toronto, ON, Canada Background: Many randomized controlled trials (RCTs) have evaluated definitive radiotherapy (RT) with different sequencing (induction (I+RT), concurrent (CRT), adjuvant, induction + concurrent (I+CRT)) and types of systemic therapies (chemotherapy (chemo), EGFR inhibitors) in the treatment of LASCCHN. Some therapies have not been compared directly. We performed a network meta-analysis to enable direct and indirect comparisons of all existing treatment modalities for LASCCHN simultaneously. Methods: A systematic review was conducted using MEDLINE, EMBASE, ASCO abstracts, ASTRO abstracts and Cochrane Central of Registered Trials using Cochrane methodology to identify RCTs up to February 2015 investigating sequencing or types of systemic treatments that involved the same definitive radiotherapy in the arms for the RCT. Two reviewers independently reviewed the RCTs and discrepancies were resolved either by discussion or by a third reviewer. The Parmar method was used to extract OS hazard ratios (HR). A Bayesian network meta-analysis (NMA) with random effects was constructed using WinBUGS to compare the relative efficacy of different sequencing and types of systemic therapies simultaneously. Results: Fifty-nine RCTs involving 11,991 patients and 13 different treatment strategies were identified. Selected comparisons are shown in the table. Conclusions: Our findings support that CRT remains the standard treatment for LASCCHN. Concurrent radiotherapy with an EGFR inhibitor does not confer an OS benefit for patients who are eligible for CRT (chemo). Induction taxane-based chemotherapy followed by CRT showed a strong trend for OS benefit when compared with CRT; this warrants further evaluation in RCTs. Comparisons CRT (chemo) vs. RT CRT (EGFR) vs. RT CRT (EGFR) vs. CRT (chemo) I (no taxane) + RT vs. RT I (taxane-based) + RT vs. RT I (taxane-based) + RT vs. CRT (chemo) I (taxane-based) + CRT (chemo) vs. RT I (no taxane) + CRT (chemo) vs. CRT (chemo) I (taxane-based) + CRT (chemo) vs. CRT (chemo)

6084

HR

95% CrI

0.72 0.77 1.06 0.88 0.64 0.89 0.60 1.02 0.83

0.66-0.78 0.55-1.08 0.75-1.52 0.66-1.00 0.43-0.95 0.59-1.33 0.48-0.74 0.82-1.24 0.67-1.01

Poster Session (Board #406), Sat, 1:00 PM-4:30 PM

337s

6083

Poster Session (Board #405), Sat, 1:00 PM-4:30 PM

HPV cell free DNA as a marker of response to chemoradiation for HPVpositive oropharngeal cancers. First Author: Daniel S. Higginson, Memorial Sloan Kettering Cancer Center, New York, NY Background: Circulating cell free plasma DNA(cfDNA) can be a sensitive biomarker of tumor burden that is accessible by simple phlebotomy. Virally induced cancers present unique advantages for cfDNA technologies in that viral genomes present a cancer-specific, broadly applicable target for assessment by quantitative PCR(qPCR) without any additional DNA processing steps that reduce DNA yield. In this pilot study, we hypothesize that qPCR of HPV16 cfDNA can serve as a marker of disease burden after a course of chemoradiation for HPV-associated oropharyngeal cancer(OPC). Methods: Patients were consented and enrolled on MSKCC specimen protocol #06-107. 8-10 mL of blood was collected in EDTA-lined vacutainer tubes before the first fraction of radiation and after the completion of radiation. Within 30 minutes of phlebotomy, samples were centrifuged at 820g for 10 minutes. The upper plasma layer was further centrifuged at 16,000g for 10 minutes to remove any remaining cell debris. DNA was extracted from the upper plasma layer (1-3mL) using the Qiagen QIAamp Circulating Nucleic Acid kit. Primers for quantitative PCR were selected via alignment of all known sequences of the HPV16 and HPV33 strains. Small amplicons ( , 100bp) were selected to maximize assay sensitivity. Results: 25 head and neck cancer patients were accrued. 22 had HPVassociated OPC defined by ISH and/or p16 immunohistochemistry. Of the 14 HPV+ OPC plasma samples analyzed, 12 exhibited detectable HPV16 cfDNA with copy numbers ranging from 1 to ~10,000 copies per mL of plasma and 1 exhibited HPV33 cfDNA for an overall sensitivity of 13/14 (93%). Of the 7 patients who have also completed chemoradiation therapy, all 7 no longer exhibited any detectable HPV16 cfDNA. None of 5 samples with non-head and neck malignancies had detectable HPV cfDNA. Conclusions: HPV16 or HPV33 cfDNA can be readily identified in almost all HPV-associated OPC patients analyzed to date. The copy numbers of HPV16 cfDNA per mL plasma vary widely, likely due to a differing number of integration events. Additional work is needed to determine if residual HPV16 or HPV33 cfDNA at completion of treatment is prognostic of recurrence.

6085

Poster Session (Board #407), Sat, 1:00 PM-4:30 PM

Phase II trial of cetuximab and radiation in low risk, HPV positive patients with locally advanced squamous cell carcinoma of the oropharynx (SCCOP). First Author: Paul Swiecicki, University of Michigan Health System Comprehensive Cancer Center, Ann Arbor, MI

Clinical genomic testing of anaplastic thyroid carcinoma (ATC) and need for integration into future prospective clinical trials. First Author: Vlad Sandulache, Department of Head and Neck Surgery, University of Texas MD Anderson Cancer Center, Houston, TX

Background: We designed a single arm phase II trial to determine the efficacy and toxicity of concurrent cetuximab (C) radiotherapy (RT) in patients with low risk SCCOP. Low risk was defined by tumor stage ( , T4, N3), Human Papilloma Virus-16 positive, and , 10 pack year smoking history. Primary endpoint was rate of recurrence.Methods: 38 patients (pts) with stage III/IVA were enrolled. The median age was 58.5 years (range: 44-76), 87 % males, 13% females; 97 % Caucasian; 88% ECOG 0, 12% ECOG 1. All patients received C 400 mg/m2 day (d) 0 then 250 mg/m2 d 7 and weekly during RT. Pts were treated with a planned 70 Gray (Gy) in 35 fractions over 7 weeks to the gross tumor and 50-60 Gy to subclinical target volumes. Results: As of January 2016 with a median follow up of 20.2 months, 38 pts were evaluable. One pt recurred with distant metastases and died 11 months post C-RT. Median PFS and OS have not been reached. Grade 3 treatment related toxicities were seen in 23 pts (60%), no grade 4 toxicities were seen. The most common grade 3 toxicities included mucositis (12 pts), dehydration (5 pts), dysphagia (4 pts), and leukopenia (4 pts). Twelve (32%) patients had at least one dose of C omitted, primarily due to toxicity. Conclusions: Patients with low risk, HPV positive SCCOP, treated with C + RT, achieve excellent control of disease, though the acute toxicity profile appears to be more significant when compared to historical controls with platinum and radiation. Clinical trial information: NCT01663259.

Background: ATC is an aggressive disease requiring rapid diagnosis. Targeted therapeutics provide ATC patients (pts) with previously unavailable treatment options. Development of high throughput next generation sequencing (NGS) platforms provides clinicians with the ability to characterize the genomic background of ATC. In ATC, a rapid, blood based test is ideal for several reasons: Lack of available tumor; Expedited results allowing for expanded treatment options; Enhanced enrollment in clinical trials. Methods: We evaluated consecutive ATC pts treated at MD Anderson between 8/2015 and 12/2015. All pts underwent genomic profiling of tumor and cell free circulating DNA (cfDNA, blood based) using NGS. Results: 14 patients were included. The most commonly mutated genes noted on both platforms were BRAF (7/14) and TP53 (11/14). Concordance between tumor and cfDNA data was high for BRAF, PIK3CA, NRAS and moderate for TP53 (Table). Twelve of 14 pts were slated for treatment, one pt elected hospice and another was observed following outside treatment. Seven pts were initiated on targeted therapy or immunotherapy based in part on tumor genomic data. Mean time to results was 13.6 days with blood-based method vs. 16.8 days with tumor-based method. Conclusions: This is one of the largest cohorts of ATC pts who have undergone blood and tumor based genomic profiling. Based on these data, we recommend utilization of both tumor based and cfDNA analysis in order to maximize sensitivity for putative targets. Integration of cfDNA in clinical trials is recommended, as turn-around time is somewhat shorter than that for tissue based testing.

Genetic alterations identified in ATC patients. Genetic alteration Tissue based testing Blood based testing

TP53

BRAF

Other

N=7 N=10 N=9 PIK3CA(H1047R, E524K, (R196, R282W, R175H, E258*, (BRAFV600E) H1047R, E542K) E258K, T170M, M246I, R248Q, NRAS, HRAS, CDKN2A, SMARCB1, Y220C) PTEN, TERT N=5 N=22 N=11 (BRAFV600E) PIK3CA(H1047R, E524K, (R282W, H193R, R267W, H1047R, E542K) R175H, R273S, A189D, M246I, NRAS,BRCA2, AR, PDGFRA, JAK2, Q192*, P27T, G266R, R248Q) CDK4, MAP2K2, PTEN, EGFR, CCNE1, APC, SMO, CCND2, BRCA1, TERT, MET (2)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

338s 6086

Head and Neck Cancer Poster Session (Board #408), Sat, 1:00 PM-4:30 PM

6087

Poster Session (Board #409), Sat, 1:00 PM-4:30 PM

PACSA: Phase II study of pazopanib in patients with progressive recurrent or metastatic (R/M) salivary gland carcinoma (SGC). First Author: Joel Guigay, Centre Antoine-Lacassagne, Nice, France

Low-dose versus high dose radioactive iodine ablation differentiated thyroid carcinoma: a prospective randomized study. First Author: Dina R. Ibrahim, Ain Shams Unversity, Cairo, Egypt

Background: SGC of head and neck (SGCHN) are rare tumors including adenoid cystic carcinoma (ACC) and non-ACC, with no standard systemic treatment for R/M patients (pts). Pazopanib (Pb) is an oral small molecule inhibitor of VEGFR, PDGFR and KIT. We conducted a multicenter single arm phase II study to assess the antitumor activity of Pb in ACC and non-ACC SGCHN. Methods: Pts with confirmed progressive R/M SGCHN received Pb 800 mg daily until progression (PD). Response was assessed every 12 weeks (RECIST). Primary endpoint was the 6-mo PFS rate. Secondary endpoints included response rate (CR/PR), overall survival, toxicity. For ACC pts, the trial was designed with inacceptable and promising 6-mo PFS rates of 20% and 40%, requiring 43 evaluable pts (alpha error = 0.07, power = 0.93). For non-ACC pts, trial was exploratory on 20 pts. A study of tumor growth rates is ongoing. Results: From 2013 to 2015, 72 pts were enrolled: 49 ACC and 20 non-ACC (3 ineligible excluded) including 11 adenocarcinoma (ADK), M:F = 32:37, median age 59 yrs (range 27-84), PS 0-1 = 42:27, prior radiation: 64 (93%), prior systemic therapy: 40 (58%). The most frequent adverse events (AE) (% of pts) were fatigue (81%), nausea (59%), diarrhea (54%), high blood pressure (BP) (43%), and anorexia (33%). 31/69 pts experienced grade 3 AE, primarily fatigue and high BP. Two grade 4 AE (cerebrovascular stroke, hepatitis) and 1 death (infection) were possibly related to treatment. Median FU was 13 mo. Among 46 ACC pts evaluable for efficacy (3 non progressive excluded), best response was 1 (2%) PR, 35 (76%) SD, 10 (22%) PD. 4 pts withdrew, 7 discontinued therapy due to AE. 6-mo PFS rate was 43% (95%CI = 29;57) with 20 pts without PD at 6 mo, median PFS was 5.9 mo. Median OS was 16.6 mo with 1 year OS = 64.8%. For 18 non-ACC pts evaluable for efficacy (2 non progressive excluded), best response was 1 PR (6%), 13 SD (72%), 4 PD (22%). 4 pts discontinued therapy due to AE. 6-mo PFS rate was 50% (95%CI = 29;71), median PFS was 6.7 mo (8 mo for ADK), median OS was not reached with 1 year OS = 75%. Conclusions: There were no new safety signals, tolerance was as expected. PACSA is positive with a 6-mo PFS rate . 40% observed in ACC and nonACC pts. Pb is promising and deserves further studies. Clinical trial information: NCT02393820.

Background: Following total thyroidectomy of differentiated thyroid carcinoma (DTC), some patients are treated with adjuvant radioactive iodine I131 (RAI). The present study compares the success of ablation with 30 millicurie (mci) versus 80 mci. Methods: From June 2013 to October 2015, we randomized the DTC patients to either 30 mci or 80 mci RAI after surgery. The study included patients with T1-T3, N0-N1, M0 tumors (The American Joint Cancer Committee AJCC 7th edition). For each patient baseline serum thyroglobulin (sTg), thyroglobulin antibody (Tg Ab) were performed. Six months post-ablation successful thyroid ablation was defined as a negative whole body scan and undetectable serum thyroglobulin. Results: Out of 50 patients with DTC, 45 patients fulfilled the eligibility criteria. Total thyroidectomy was performed in 27/45 (60%). 26/45 (57.8%) of these patients received 30 mci while19/45 (42.2%) patients received 80 mci. The median age was 37 and 36.5 in the arms 80 and 30 mci respectively. Females represented 35/45 (77.8%) of patients. Histopathology showed papillary carcinoma in 42/45 (93.3%) and follicular carcinoma in 3/45 (6.7%) of patients. T2 tumors predominated in 10/19 (52.6%), and 15/26 (57.7%) of the 80 and 30 mci arms respectively. According to the American Thyroid Association risk classification, all of the patients in both arms had low risk disease. Success of ablation was achieved in 15/19 (78.9%) of the am 80 mci, and in 15/26 (60%) of the arm 30 mci (p 0.13). No patients developed distant metastases in both arms. The 80 mci arm patients had longer hospital admission compared to the 30 mci arm (p 0.008). Only 6/26 (23.1%) patients in the 30 mci arm were isolated for 2-4 days, whereas all the 80 mci arm patients were isolated for 3-5 days. Conclusions: Both 80 mci and 30 mci of radioactive Iodine (I-131) have similar success rate in the ablation of thyroid remnant of low risk differentiated thyroid cancer patients. The 30 mci I-131 dose is associated with fewer side effects, shorter hospital admission duration, and less expensive in low risk DTC patients. Clinical trial information: 1578/2013.

6088

6089

Poster Session (Board #410), Sat, 1:00 PM-4:30 PM

Poster Session (Board #411), Sat, 1:00 PM-4:30 PM

Phase II study of lenvatinib in patients with differentiated, medullary, and anaplastic thyroid cancer: Final analysis results. First Author: Shunji Takahashi, Department of Medical Oncology, Cancer Institute Hospital of Japanese Foundation for Cancer Research, Tokyo, Japan

Response to lenvatinib treatment in patients with radioiodine-refractory differentiated thyroid cancer (RR-DTC): Updated results from SELECT. First Author: Andrew G. Gianoukakis, David Geffen School of Medicine at UCLA, Torrence, CA

Background: Thyroid cancer is classified into differentiated (DTC), medullary (MTC), and anaplastic (ATC) types. Lenvatinib (LEN) significantly prolonged progression-free survival (PFS) vs placebo in the phase III SELECT trial of patients (pts) with 131I-refractory DTC (RR-DTC). Initial results of this phase 2 trial (Study 208) of LEN in RR-DTC, MTC, and ATC were reported with a cutoff date of 15 Jun 2014. Here we report the final results of this study. Methods: This was a singlearm, open-label, phase 2 study conducted in Japan with an updated data cutoff of 9 Jul 2015. Pts received LEN 24 mg/d in 28-d cycles until progressive disease or development of unacceptable toxicity. Primary endpoint was safety; secondary endpoint was efficacy as assessed by PFS, overall survival (OS), overall response rate (ORR), and disease control rate (DCR). Results: 51 Pts, including 25 pts with RR-DTC (previously 23), 9 pts with MTC, and 17 pts with ATC (previously 11) were enrolled. All pts had at least one treatment-emergent adverse event (TEAE). The most common any-grade TEAEs included hypertension (90%), decreased appetite (78%), palmar-plantar erythrodysaesthesia syndrome (77%), fatigue (73%), proteinuria (61%), stomatitis (57%), and diarrhea (55%). Incidences of Grade 3 and 4 TEAEs were similar among subgroups (RR-DTC, 72%; MTC, 100%; ATC, 88%). Of note, only 1 pt discontinued treatment due to a TEAE. There were 4 fatal serious AEs, all considered unrelated to LEN. Tumor shrinkage occurred in most pts, including pts with ATC. Median duration of treatment was 5.5 mos (range, 0.7–33.1) for pts with ATC; 8 received LEN for more than 6 mos. Efficacy parameters are shown in the table below. Conclusions: In this study, LEN showed tumor shrinkage in almost all pts with advanced thyroid cancer, including pts with ATC. Toxicities were manageable with dose modifications. LEN efficacy especially in ATC warrants further investigation. Clinical trial information: NCT01728623.

Background: Lenvatinib (LEN) significantly prolonged median progressionfree survival (PFS) vs placebo (PBO) in the primary analysis of the phase 3 SELECT study of RR-DTC (18.3 vs 3.6 mos; HR 0.21; 99% CI 0.14–0.31; P, 0.001). Median duration of overall response (DOR; per RECIST 1.1) to LEN was not reached at the time of the primary analysis. Here we present new efficacy data with a focus on DOR. Methods: The data cutoff was 31 Aug 2015 (LEN: n = 261, PBO: n = 131); primary endpoint was PFS. DOR was examined for patients (pts) who had partial (PR) or complete responses (CR) and by subgroup (age, sex, tumor subtype, baseline disease burden, baseline Eastern Cooperative Oncology Group score, metastasis site, prior vascular endothelial growth factor [VEGF] therapy). For this analysis, tumor assessments were per investigator. We also report clinical features of 2 pts with RR-DTC with prolonged responses to LEN. Results: At cutoff, median PFS was 19.4 mos (95% CI 14.8–29.3) for LEN and 3.7 mos (95% CI 3.5–5.4) for PBO (HR 0.24; 99% CI 0.17–0.35; P, 0.0001). 157 Pts (60.2%) responded to LEN and median DOR was 30 mos (95% CI 18.4–35.2). 3 Pts (2.3%) responded to PBO and the median DOR was 14.7 mos (95% CI 7.5–not evaluable [NE; median not yet reached]). Median DOR to LEN was similar among subgroups except in pts with greater disease burden (tumor size # 35mm: NE; 35–60mm: 27.5 mos; 60–92mm: 18.0 mos; . 92mm: 15.7 mos), and pts with liver metastasis (yes: 15.7 mos; no: 31.3 mos). 2 Pts with prolonged DOR are still receiving LEN: 1 woman (Hurthle ¨ cell follicular thyroid cancer; no prior VEGF therapy) received LEN since Nov 2011 and achieved CR with a DOR of 46 mos (ongoing) and a PFS of 48 mos; 1 man (Hurthle ¨ cell variant of papillary thyroid carcinoma; no prior VEGF therapy) received LEN since Jul 2012 and achieved PR with a DOR of 35 mos (ongoing) and a PFS of 41 mos (of note, this pt has multiple metastases in the lung, abdomen, and bone). Conclusions: In this analysis, LEN PFS benefit was maintained along with a notable DOR. The case studies profiled demonstrate that an extended DOR to LEN can occur in pts with RR-DTC who show significant variation in the severity of disease. Clinical trial information: NCT01321554.

Median PFS (95% CI)―mos. Median OS (95% CI)―mos. ORR―n (%)* DCR―n (%)

RR-DTC (n = 25)

MTC (n = 9)

ATC (n = 17)

25.8 (18.4–NE) 31.8 (31.8–NE) 17 (68) 25 (100)

9.2 (1.8–NE) 12.1 (3.8–NE) 2 (22) 9 (100)

7.4 (1.7–12.9) 10.6 (3.8–19.8) 4 (24) 12 (71)

NE, not evaluable. *All partial responses.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Head and Neck Cancer 6090

Poster Session (Board #412), Sat, 1:00 PM-4:30 PM

6091

339s Poster Session (Board #413), Sat, 1:00 PM-4:30 PM

Phase II trial of nintedanib in patients with recurrent or metastatic salivary gland cancer: A multicenter phase II study. First Author: Youjin Kim, Samsung Medical Center, Seoul, Korea, The Republic of

Pembrolizumab for advanced papillary or follicular thyroid cancer: preliminary results from the phase 1b KEYNOTE-028 study. First Author: Janice M. Mehnert, Rutgers Cancer Institute of New Jersey, New Brunswick, NJ

Background: Salivary gland cancers (SGC) are histologically uncommon but heterogeneous, accounting for less than 5% of all head and neck cancers. No specific therapy including targeted agents has been associated with improved clinical outcome in recurrent/metastatic SGC. Recent studies suggest VEGF and PDGFR might play an important role in SGC. Nintedanib is a potent small molecule triple receptor tyrosine kinase inhibitor (VEGFR1-3, FGFR 1-2, and PDGFR a/ß). We sought to determine nintedanib’s antitumor activity in patients with recurrent or metastatic SGC. Methods: This openlabel, multicenter, phase II, single-arm study was conducted at 11 hospitals in Korea. Patients with pathologically confirmed recurrent and/or metastatic SGC who had undergone cytotoxic chemotherapy were enrolled. Nintedanib was given orally at 200mg twice daily until disease progression or unacceptable toxicity. The primary endpoint was response rate. Secondary endpoints were progression-free survival, overall survival, toxicity, and disease control rates. Simon two-stage optimal design was used. Results: Median age was 54 years, 62% of patients were female, and 95% had an ECOG performance status of 0 or 1. The majority of patients had adenoid cystic carcinoma (65%), and 40% of all of patients had received at least two prior chemotherapies. After 20 patients were enrolled, the study was stopped because no responder was observed at Stage I. There was no partial response, but the disease control rate was 75% (15/20). The median duration of stable disease was 8.3 months (range 3.5-12.4). At the time of data cutoff with median follow-up of 9.5 months, median overall survival had not been reached, and the progression-free survival rate at 6 months was 60% (CI.0.34-0.79). Grade 3 adverse events included liver enzyme elevation (25%) and nausea/vomiting (5%). Five patients required dose reduction. Conclusions: Single agent nintedanib did not show partial response, but did achieve a 75% control rate with long stabilization in these SGC patients. Nintedanib was tolerable and toxicities were manageable. Given the high rate and long duration of disease control, further investigation is needed. Clinical trial information: NCT02558387.

Background: Treatment options are limited for patients (pts) with advanced thyroid cancer. PD-L1 expression in thyroid cancer tumors has been shown to correlate with poor prognosis. Pembrolizumab (pembro), an anti–PD-1 antibody, blocks the interaction between PD-1 and PD-L1 and PD-L2. We assessed the safety and efficacy of pembro in PD-L1+ advanced (unresectable and/or metastatic) thyroid cancer. Methods: KEYNOTE-028 (NCT02054806) is a nonrandomized trial of pembro in 20 types of advanced solid tumors. Key eligibility criteria for this cohort included papillary or follicular subtypes of advanced thyroid cancer, failure of standard therapy, ECOG PS 0-1, and PD-L1 expression in $ 1% of tumor or stroma cells by IHC. Pembro 10 mg/kg was given every 2 wk for up to 24 mo or until confirmed progression, intolerable toxicity, death, or withdrawal of consent. Response was assessed every 8 wk for the first 6 mo and every 12 wk thereafter. The primary end point was ORR per RECIST v1.1 by investigator review. Results: Of the 22 pts enrolled, median age was 60.5 y; 40.9% were male; 68.2% vs 31.8% had papillary vs follicular carcinoma; 36.4% had an ECOG PS of 1; 50.0% received $ 2 prior therapies for metastatic disease (18 pts were radioactive iodine refractory; 7 pts received prior sorafenib; 1 pt received prior lenvatinib). Median follow-up duration as of Dec 10, 2015, was 73.5 wk (range, 29.4-87.0). 18 (81.8%) pts had treatment-related adverse events (TRAEs); those occurring in $ 15% of pts were diarrhea (n = 7) and fatigue (n = 4). 1 TRAE of grade $ 3 (grade 3 colitis) occurred; no pts died or discontinued pembro because of a TRAE. 2 pts had a PR for an ORR (confirmed) of 9.1% (95% CI, 1.1-29.2); median duration of response was not yet reached (range, 35.3-44.1+ wk) by the data cutoff. The stable disease rate was 54.5% (n = 12, 95% CI, 32.2-75.6). The 6-mo OS rate was 100%; the 6-mo PFS rate was 58.7%. At the data cutoff, 6 pts remained on treatment. Conclusions: Pembro shows promising antitumor activity in advanced papillary or follicular thyroid cancer which progressed on standard therapies. The clinical benefit of pembrolizumab in advanced thyroid cancer will be further investigated in the phase 2 KEYNOTE-158 trial (NCT02628067). Clinical trial information: NCT02054806.

6092

6093

Poster Session (Board #414), Sat, 1:00 PM-4:30 PM

Chest X-ray prior to thyroidectomy: Is it really needed? First Author: Alexandra Mikhael, Cleveland Clinic, Cleveland, OH Background: Chest X-ray (CXR) prior to thyroid surgery is routinely ordered despite the lack of evidence to establish its utility. Unnecessary testing would be adding medical, financial and personal expenses without benefit. This study aimed to determine the utility of preoperative CXR in patients undergoing thyroidectomy. Methods: 594 consecutive patients who underwent thyroidectomy at the Cleveland Clinic were included in the study. Patients without preoperative CXR were excluded Retrospective review was performed to collect data including demographics, CXR findings, anesthesia records and pathological data including cancer stage and presence of metastasis. Patients with findings on CXR were compared to those without, to see if there was any correlation with difficult intubation or stage of cancer. Results: 494 patients (83%) of patients reviewed had a preoperative CXR. 67% had cancer on surgical pathology. 56 % had at least one abnormality reported on the reading. Skeletal abnormalities (25%) were the most frequent abnormality reported followed by tracheal deviation (16%), cardiomegaly (12%), lung nodules (6%; 3% subcentimetric and 0.5% diffuse nodularity), emphysema (2%) and pleural effusion (1%). Of 78 patients with tracheal deviation on CXR only 5% had a difficult intubation during surgery. CXR was shown to impact management in 4%. Tracheal deviation was more commonly seen in non-cancer cases compared to cancer patients (27% vs. 12%, p , 0.001). Amongst patient with cancer a higher T stage ( $ 2) had higher proportions of tracheal deviation compared to T = 1 (17% vs. 8 %, p , 0.001). Although patients with metastatic cancer (n = 81) compared to other cancer patients had a higher proportion of any abnormality on CXR (57% vs. 44%, p = 0.045) there was no significant difference for tracheal deviation or lung nodules on CXR. Of patients with nodules on CXR (n = 29) only 14% patients were shown to have metastatic disease. Conclusions: The utility of preoperative CXR in patients undergoing thyroidectomy including for thyroid cancer is very limited. In the climate of value-based care, routine use of this modality may be considered redundant, and should be ordered only if otherwise clinically indicated.

Poster Session (Board #415), Sat, 1:00 PM-4:30 PM

Predictors of neck reoperation (NRO) in 35,538 Californians with differentiated thyroid cancer. First Author: Theresa H.M. Keegan, Center for Oncology Hematology Outcomes Research and Training (COHORT), UC Davis Comprehensive Cancer Center, Sacramento, CA Background: Our prior work documented variation in initial management patterns and survival for subgroups of patients (pts) with differentiated thyroid cancer (DTC) in California. Therefore, we hypothesized that patient, tumor, and hospital facility characteristics at initial thyroid surgery would predict NRO, an event associated with increased mortality. Methods: Using the California Cancer Registry linked to hospital records from the California Office of Statewide Health Planning and Development (OSHPD), we identified pts with initial thyroid surgery for papillary (PTC) or follicular thyroid cancer during 1991-2012. We defined NRO from OSHPD as a neck node surgery 91 days to 5 years from the initial thyroid surgery. In pts at least 5years from initial total/near total thyroidectomy (1991-2008; N = 24,473), we performed multivariable logistic regression to identify demographic, tumor, and hospital characteristics associated with NRO. Results are presented as adjusted odds ratios (OR) and 95% confidence intervals (CI). Results: NRO was identified in 1,741 (4.9%) pts, in decreasing rates with time from diagnosis. Post-thyroidectomy radioiodine (RAI) was administered to 75.8% of those with NRO; whereas 56.6% of those without NRO received RAI. In multivariable models, NRO was associated with younger age (15-34: OR 1.41, CI 1.17-1.69; versus $ 55 years) male sex (OR 1.29, CI 1.131.47), and Asian/Pacific Islander (API: OR 1.19, CI 1.02-1.39) or Hispanic (OR 1.26, CI 1.07-1.49) race/ethnicity. Tumor characteristics were strongly associated with increased odds of NRO, including PTC (OR 1.59, 1.192.12), increasing T stage, extrathyroidal extension (OR 2.21, CI 1.94-2.52), and N stage (OR 3.42, CI 3.00-3.93). The odds of NRO increased from 1991-2008 and were higher in those initially treated with RAI. Hospital thyroid cancer surgery volume was not associated with NRO. Conclusions: NRO is becoming increasingly frequent, with tumor characteristics, younger age, API or Hispanic race/ethnicity, and male sex but not hospital volume associated with NRO. This study identifies subgroups of pts at higher risk for NRO and subsequent mortality due to persistent or more biologically aggressive disease.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

340s 6094

Head and Neck Cancer Poster Session (Board #416), Sat, 1:00 PM-4:30 PM

Late systemic symptom (SS) frequency in head and neck cancer (HNC) survivors as measured by the Vanderbilt Head and Neck Symptom Survey – General Symptoms Subscale (GSS). First Author: Elizabeth Marie WulffBurchfield, Vanderbilt University Hospital, Nashville, TN

6095

Poster Session (Board #417), Sat, 1:00 PM-4:30 PM

Phase II trial of eribulin for recurrent or metastatic salivary gland cancers. First Author: Keith D. Eaton, University of Washington, Seattle, WA

Background: Neuroinflammation and central sensitization from cancer and its therapy may result in late SS. Studies of SS in HNC survivors are lacking, due in part to the absence of comprehensive, targeted tools. The GSS was developed to meet this need and promote research in this arena. We are conducting a mixed-methods, cross-sectional, descriptive study in HNC survivors to validate the GSS. Methods: Inclusion: Cancer-free HNC patients age . 20, greater than 1 year post treatment. Procedures: Patients completed self-report measures, including the GSS, Profile of Mood States – Short Form, Neurotoxicity Rating Scale, Body Image Quality of Life Inventory, and a Quality of Life scale. Patients with moderate-to-severe SS, as determined by a score of $ 4 on at least one GSS item, were offered participation in a structured interview exploring impact. Statistics: Frequency distributions were used to summarize patient characteristics and GSS item responses. Results: 66 patients out of a planned 100 have been recruited to date. Patient characteristics: male 77%, Caucasian 95%, median age 63.8 years, and median time since last treatment 38 months. 52% of patients experienced one or more moderate-to-severe SS. SS frequencies (percent of patients experiencing any grade / percent of patients experiencing moderateto-severe) were as follows: fatigue 41%/20%, with 21% reporting moderateto-severe effects on daily life; trouble falling asleep 33%/8%; trouble staying asleep 48%/22%; unexplained sweating 20%/8%; feeling hot 30%/5%; feeling cold 49%/22%; weight loss 15%/2%; anorexia 22%/11%; memory or thought processing problems 46%/20%; sadness or depression 30%/8%; anxiety 36%/11%; widespread pain 50%/23%, with 12% reporting average pain $ 4/10 within 1 week of enrollment. Qualitative data regarding the impact of SS will be presented. Conclusions: The frequency, severity and significance of late SS in HNC survivors is substantial and mitigation is essential. Areas of needed research include prevention of neuroinflammatory toxicities, treatments minimizing symptom burden, and support programs for those destined to survive with complex late effects.

Background: Salivary gland cancers (SGC) are rare, biologically and histologically diverse malignancies with no standard of care therapy in the recurrent/metastatic setting. Eribulin is a microtubule inhibitor with established efficacy in previously treated breast cancer. The purpose of this study was to determine the anticancer activity of eribulin in advanced SGC. Methods: In a single center two-stage single arm phase II design, patients with progressive, recurrent and/or metastatic SCG of the head and neck or other sites, with documented disease progression within the past 6 months and not amenable to curative intent therapy were eligible. Any number of prior cytotoxic chemotherapies was allowed. Patients received eribulin 1.4 mg/m2 IV on Days 1 and 8 of a 21-day cycle until disease progression or unacceptable toxicities. Response was assessed by CT imaging every 6 weeks. The primary endpoint was overall response rate by RECIST 1.1. Secondary endpoints were duration of response (DOR), progression free survival (PFS), disease control rate (DCR), and grade . = 3 toxicities. Results: Between May 2012 and August 2015, 29 patients were enrolled. Median age was 63 (range 34-75); 20 (69%) were male. The most common histologies were adenoid cystic (n = 10) and adenocarcinoma (n = 5). Prior radiation was administered in 25 (86%). Number of prior cytotoxic therapies were: 0 (12, 41%), 1 (13, 45%), 3(3, 10%), 5 (1, 3%). Neutropenia was the most common toxicity (Grade 3 (9, 31%), Grade 4 (4, 14%)). Other grade 3 toxicities were: febrile neutropenia, increased QTc, fatigue and peripheral neuropathy in a single patient each. Median follow-up is 2 years, with 5 patients continuing on treatment. The median number of cycles administered was 4 (range 2-28). Response by RECIST criteria was observed in 3/29 (10%) with 2 PR and 1 CR , 20/29 (69%) patients demonstrated a decrement in tumor size at first assessment. Median PFS was 3 months, DCR was 26/29 (90%) and median DOR was 17.2 months. Conclusions: In this phase II trial of eribulin for advanced SGC, we observed response rates similar to those reported to other cytotoxic therapies and superior to those observed in trials of TKIs. Toxicities were similar to those observed in prior trials with eribulin. Clinical trial information: NCT01613768.

6096

6097

Poster Session (Board #418), Sat, 1:00 PM-4:30 PM

Poster Session (Board #419), Sat, 1:00 PM-4:30 PM

Phase II study of regorafenib in progressive, recurrent/metastatic adenoid cystic carcinoma. First Author: Alan Loh Ho, Memorial Sloan Kettering Cancer Center, New York, NY

Does margin status affect oncologic outcomes and survival in patients with papillary thyroid cancer? First Author: Gauri Bhuchar, Cleveland Clinic, Cleveland, OH

Background: Recurrent/metastatic adenoid cystic carcinoma (R/M ACC) is incurable. Regorafenib (BAY 73-4506; Bayer) is a tyrosine kinase inhibitor (TKI) that targets receptor tyrosine kinases (RTKs), including vascular endothelial growth factor receptor (VEGFR), fibroblast growth factor receptor (FGFR), and platelet-derived growth factor receptor (PDGFR), each thought to be activated in ACC. In this phase II trial, we tested if regorafenib is an effective therapy for R/M ACC. Methods: Patients had incurable R/M ACC of any primary site. Evidence of disease progression on imaging performed within 6 months of enrollment ornew/worsening disease-related symptoms was required. Patients were started at 160 mg daily for 3 weeks in a 4-week cycle. To limit drug toxicities, the protocol was later amended to start patients at 120 mg daily; if dose reduction was not performed/required, escalation to 160 mg was allowed. There are two RECIST v1.1 evaluable primary endpoints: 1) the proportion of patients alive at 6 months without disease progression, and 2) best overall response rate. Among 38 patients, if at least 21 are alive and free of progression at 6 months, or at least 5 have a response, the study would be deemed positive. Results: 38 patients were registered. 15 were started at 160 mg. Only two of the 23 started at 120 mg were increased to 160 mg. No CRs or PRs have been observed. 17 have had SD for 6 months or more; only 1 of 8 patients currently on therapy has been treated , 6 months. Whole exome (16 tumors) and RNAseq (9 tumors) analyses were performed, results are pending. Conclusions: The study failed to meet the primary endpoints, but regorafenib may elicit disease control for a subset of ACC patients. Final clinical and exploratory analyses evaluating the genomic and transcriptomic landscape of ACCs are pending. Clinical trial information: NCT02098538.

Background: It is widely known that gross extrathyroidal extension (GEE) portends poor prognosis in patients with papillary thyroid cancer (PTC). There is controversy regarding the prognostic value of microscopic extrathyroidal tumor extension (MEE). American Joint Committee on Cancer 7th edition upstages patients with MEE. However, recent studies have shown no difference in cause-specific mortality and tumor recurrence in patients with MEE and those with clear margins. This is a retrospective, single center study that further elucidates the role of MEE in cause-specific mortality and tumor recurrence in patients with PTC. Methods: We conducted a cohort study of consecutive patients with PTC and without distant metastasis that underwent thyroidectomy at the Cleveland Clinic between 2003 and 2008. Pathology slides were reviewed to determine margin status. Recurrence between groups was compared using Pearson chi-square test. Kaplan-Meier estimates of survival were calculated and compared using Cox proportional hazards models. Results: 287 patients met the aforementioned study criteria and were included in the survival analysis. Of these, 21 patients were excluded from the recurrence analysis as they continued to have persistent disease after surgical resection. Median age was 49 years; 22% were males. 202 (70%) patients had clear margins, 69 (24%) had MEE and 16 (6%) had GEE. 5.2% of the patients with clear margins (N = 194) had recurrence compared to 16.4% of the patients with MEE (N = 61) (OR = 3.6, p = 0.004) and 25% of the patients with GEE (N = 8) (OR = 6.1, p = 0.020). 10 year survival of patients with clear margins was 99.5%, with MEE was 95.6% (HR = 4.6, CI = 0.77-27.60, p = 0.095) and with GEE was 75% (HR = 30.2, CI = 5.50-165.99, p , 0.001). Survival in patients with GEE was worse than those with MEE (HR = 6.3, p = 0.016). Conclusions: Contrary to recent literature that challenges the current staging guidelines, our study shows that rate of disease recurrence is significantly higher in patients with microscopic extrathyroidal extension when compared to those with clear margins. However, we agree that there is no significant difference in survival between the two groups.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Head and Neck Cancer 6098

Poster Session (Board #420), Sat, 1:00 PM-4:30 PM

Utility or futility of preoperative thyroglobulin in patients undergoing thyroidectomy for differentiated thyroid cancer. First Author: Rushad Patell, Cleveland Clinic, Cleveland, OH Background: Thyroglobulin (TG) has an established role to detect recurrence of differentiated thyroid cancer (DTC). The role of preoperative TG measurement is more ambiguous, yet commonly performed in practice. This study aimed to determine if preoperative TG correlates with stage or metastatic burden. Methods: A retrospective review of consecutive patients undergoing thyroidectomy with preoperative TG, at Cleveland Clinic was conducted. Data collected included demographics, preoperative TG levels, pathology of resected specimen, stage of cancer and presence of metastasis. Associations between presurgical TG levels with size of largest malignant nodule, total thyroid size, stage of cancer and metastasis using Wilcoxon rank sum tests and Spearman correlations were made. Receiver Operator Curve (ROC) was generated to assess utility of presurgical TG to detect metastasis. Results: Study population consisted of 598; patients with positive TG antibodies were excluded (n = 176). 422 patients were included in the final analysis. Median age was 55 years and 76 % were female. 78% patients had malignancy on pathology of which papillary thyroid cancer was most frequent (92%). 18% had TG levels . 500 ng/mL and 29% , 35 ng/ mL. 4% had distant metastasis and 18 % lymph node metastasis. Median length of follow up was 52 months. Preoperative TG led to additional imaging in 4%. A significant positive correlation between presurgical TG and thyroid size (rho, 0.49, p , 0.001) and size of malignant nodule (rho 0.27, p , 0.001) was observed. Using a cut-off of 500 for presurgical TG to detect metastasis sensitivity was 10.3% (95% CI 2.5-18.2), specificity 90.0% (95% CI 86.0-93.9), positive predictive value 21.4%, (95% CI 6.2-36.6) and negative predictive value 79.1% (95% CI 74.1-84.2). Area under curve for ROC generated was not significant (0.46 95% CI 0.38-0.54), indicating poor diagnostic ability of TG to predict metastasis. Conclusions: Although preoperative TG correlated significantly with the size of gland and T stage there was no correlation with metastasis. Presurgical TG is of limited utility in today’s value based healthcare environment.

6100

Poster Session (Board #422), Sat, 1:00 PM-4:30 PM

Radiation therapy and low dose doxorubicin in thyroid cancer: A phase II study. First Author: Eric Jeffrey Sherman, Memorial Sloan Kettering Cancer Center, New York, NY Background: The role of external beam radiation therapy has remained controversial in the management of differentiated thyroid cancer (DTC). There is an absence of prospective studies in this area despite the importance of locoregional control (LRC) in the neck. We performed a prospective phase II study to better evaluated both LRC and toxicity with intensity modulated radiation therapy (IMRT). Methods: We enrolled 26 patients in a prospective single institution phase II study. The first 7 did not receive chemotherapy; the next 19 received weekly doxorubicin (10 mg/m2) concurrently with IMRT (70Gy). Inclusion criteria included non-anaplastic, nonmedullary DTC that is either grossly recurrent after surgery or unresectable with a Karnofsky Performance status . 60%. The primary objective was 2year LRC rate of 90% in the group receiving concurrent chemotherapy compared to a historical LRC rate of 40%. Swallow evaluation was collected every 6 months for 2 years (not reported). Results: The median age was 63 years (29-83); 65% were male and 50% had distant metastatic disease at time of study entry; histology was papillary (65%), poorly differentiated (27%) and Hurthle Cell (8%). The BRAF V600E mutation status was known in 9 patients, 6 of which were positive. All patients completed IMRT without delay. Patients received a median of 7 (5-9) doses of doxorubicin. Of the 26 patients, 5 have died (none related to treatment) of which 4 had metastases at time of IMRT initiation. Median followup is 420 days (59-864 days). Only one patient has had proven failure in the neck (RT alone group). No proven LR failure has been noted in the chemotherapy group. No patients required permanent PEG use. Conclusions: Concurrent doxorubicin and IMRT appears to be effective in controlling LR disease in patients with recurrent or resectable DTC. Longer term follow up and toxicity data (swallowing evaluations) does need to be collected and analyzed. Clinical trial information: NCT01882816.

6099

341s Poster Session (Board #421), Sat, 1:00 PM-4:30 PM

Enhancing radioiodine (RAI) incorporation into BRAFV600E-mutant, RAIrefractory thyroid cancers with the BRAF inhibitor vemurafenib: A pilot study. First Author: Lara Dunn, Memorial Sloan Kettering Cancer Center, New York, NY Background: For patients with radioactive iodine (RAI)-refractory, recurrent/ metastatic thyroid cancer, treatment entails indefinite exposure to tyrosine kinase inhibitors and their associated toxicities. The mitogen activated protein kinase (MAPK) signaling pathway is commonly activated in thyroid cancer and suppresses genes required for iodine uptake and retention, leading to RAI refractoriness. In a previous pilot study, the MEK inhibitor selumetinib restored RAI uptake; however, the response was attenuated in patients with BRAF-mutant tumors. Methods: We are conducting a study to determine whether direct targeting of BRAF activity with vemurafenib in patients with BRAF-mutant thyroid cancer reverses RAI-refractoriness. The primary objective is to determine the proportion of patients with restoration of RAI uptake sufficient to enable therapeutic RAI. Patients were stimulated with thyrotropin alfa and underwent 124I PET scans prior to and 4 weeks after treatment with vemurafenib (960mg oral twice daily). Patients whose second 124I PET scan demonstrated that . 2000 cGy could be delivered to a lesion using , 300 mCi of 131I, received therapeutic 131I while continuing vemurafenib. Results: 12 patients with BRAF-mutant thyroid cancer were enrolled; 2 were removed from study prior to the second 124I PET scan and were replaced. The median age was 68 (range 43-72) with 7 (58%) men. Vemurafenib increased 124I PET lesional dosimetry sufficient to warrant therapeutic 131I in 4 of 9 evaluable patients. Of these, 3 exhibited tumor regression (1 PR, 2 SD) at 6 months following RAI and 1 has yet to undergo follow-up imaging. Decrease in thyroglobulin levels and tumor regressions on vemurafenib prior to RAI were observed. Analysis of transcriptional signatures to assess MAPK pathway inhibition and thyroidspecific gene expression is ongoing. Conclusions: These preliminary results suggest that vemurafenib can enhance RAI incorporation and efficacy in a subset of patients with BRAF-mutant thyroid cancer. We are currently investigating whether MAPK or thyroid-specific transcriptional output markers in tumor specimens identify patients most likely to respond. Clinical trial information: NCT02145143.

TPS6101

Poster Session (Board #423a), Sat, 1:00 PM-4:30 PM

A phase 3, randomized, open-label study of first-line durvalumab (MEDI4736) 6 tremelimumab versus standard of care (SoC; EXTREME regimen) in recurrent/metastatic (R/M) SCCHN: KESTREL. First Author: Tanguy Y. Seiwert, University of Chicago, Chicago, IL Background: Current therapies for pts with R/M SCCHN achieve poor survival, so more effective and better tolerated treatments are needed. Tumors can evade immune detection by exploiting inhibitory immune checkpoints such as programmed cell death-1 (PD-1) and cytotoxic T-lymphocyteassociated antigen-4 (CTLA-4), commonly expressed in SCCHN. Durvalumab (D) is a selective, high affinity, engineered human IgG1 mAb that blocks binding of programmed cell death ligand-1 (PD-L1) to PD-1 and CD80. Tremelimumab (T) is a selective human IgG2 mAb inhibitor of CTLA-4. Targeting both PD-L1 and CTLA-4 pathways has potential for synergistic antitumor effects. In a Phase 1b study in pts with advanced NSCLC (NCT2000947), D+T combination regimens demonstrated clinical activity and manageable tolerability profiles. The Phase 3 KESTREL study (NCT02551159) compares D 6 T with SoC EXTREME regimen for first-line treatment of R/M SCCHN. Methods: KESTREL is an open-label, multicenter, global study of pts with R/M SCCHN (oral cavity, oropharynx, hypopharynx or larynx) who have received no prior systemic chemotherapy (unless part of multimodality treatment for locally advanced disease). Pts (n = 628) will be stratified by PD-L1 status, smoking history, tumor location, and then HPV status (oropharyngeal cancer) and randomized (2:1:1) to receive flat-doses of T 75 mg q4w (max 4 doses) + D 1500 mg q4w; D 1500 mg q4w; or SoC EXTREME regimen (carboplatin or cisplatin + 5FU + cetuximab), all until PD. The D+T combination will be assessed vs SoC in terms of co-primary endpoints PFS and OS. D+T vs SoC will be further assessed in terms of ORR; DoR; proportion of pts alive and progression free at 12 months; OS at 24 months; secondary progression; safety and tolerability; pharmacokinetics; immunogenicity; and HRQoL. The efficacy of D monotherapy vs both D+T and SoC will also be tested. Exploratory endpoints include blinded independent central review of antitumor activity (immune-related RECIST v1.1) and potential biomarkers of progression/response. Pts are enrolled from USA, Canada, Taiwan, South Korea, Greece and Spain (total of: 23 countries, 191 sites planned). Clinical trial information: NCT02551159.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

342s TPS6102

Head and Neck Cancer Poster Session (Board #423b), Sat, 1:00 PM-4:30 PM

An international, multicenter, randomized, double-blind, placebocontrolled, parallel-group phase 2 study of palbociclib (an oral CDK4/6 inhibitor) plus cetuximab in patients with recurrent/metastatic (R/M) squamous cell carcinoma of the head and neck (SCCHN). First Author: Bhumsuk Keam, Department of Internal Medicine, Seoul National University Hospital, Seoul, South Korea Background: SCCHN is the sixth most common cancer worldwide. Current second-line treatments for human papillomavirus (HPV)–negative, R/M SCCHN have marginal effectiveness and significant toxicity. Therefore, improved treatment options are needed. Inhibition of the cyclin D1/cyclindependent kinase (CDK)4/Rb axis using the selective CDK4/6 inhibitor palbociclib (PAL) represents a rational treatment approach for patients (pts) with incurable, cetuximab-naive, HPV-negative SCCHN because loss of RB1 is infrequent but loss of CDKN2A expression or amplification of CCND1 occurs in most cases; the majority of a large cell line panel responded at low nanomolar concentrations of PAL; and PAL exhibited significant singleagent activity in a series of HNSCC patient-derived xenograft (PDX) models. A combinatorial approach with the epidermal growth factor receptor (EGFR) inhibitor cetuximab may further enhance antitumor effects because EGFR overexpression is considered an important driver in SCCHN pathogenesis; EGFR inhibition demonstrated strong synergy in combination with PAL in vitro; and PAL plus cetuximab showed significant antitumor activity in an HNSCC PDX model. A phase 1 trial established the safety of combining PAL with cetuximab in pts with incurable SCCHN, with a decrease in tumor target lesions observed in approximately 50% of pts (Michel, ASCO 2015, Abs: 6043). Methods: Patients with HPV-negative, cetuximab-naive R/M SCCHN after failure of 1 platinum-containing regimen will be randomized 1:1 to receive either 125 mg PAL orally daily for 21 days followed by a 7-day break in combination with weekly cetuximab (initial dose, 400 mg/m2 followed by 250 mg/m2) or placebo plus the same regimen of cetuximab. The primary objective is to compare overall survival for both treatment arms. Secondary and exploratory objectives include evaluation of progression-free survival, PK/PD relationships, patient-reported outcome measures, and biomarker analyses. Clinical trial information: NCT02499120.

TPS6104

Poster Session (Board #424b), Sat, 1:00 PM-4:30 PM

TPS6103

Poster Session (Board #424a), Sat, 1:00 PM-4:30 PM

De-ESCALaTE: Comparison of early and late toxic events in low-risk HPV positive oropharyngeal carcinoma patients treated with chemoradiotherapy, cetuximab versus cisplatin. First Author: Hisham Mohamed Mehanna, University of Birmingham, Birmingham, United Kingdom Background: The rapid increase in oropharyngeal squamous cell carcinoma (OPSCC) incidence in the developed world has been attributed to a rise in Human Papillomavirus (HPV) infection. HPV related OPSCC is considered a distinct disease entity affecting a younger and healthier patient population. As such, acute and late toxicities, and the subsequent treatment strategy of those toxicities, have a significant impact on healthcare systems and societies, as well as patient lives. Standard platin-based chemoradiotherapy is associated with considerable acute toxicity and long term sequelae. DeESCALaTE is a pragmatic, multi-centre, open label randomised clinical trial addressing the need to establish less toxic therapy in low-risk HPV positive head and neck cancer. Methods: Patients with low-risk HPV+ OPSCC are randomised to receive radiotherapy (70G in 35F) and either cisplatin (100 mg/m2 x 3) or cetuximab (400 mg/m2 loading dose followed by weekly 250 mg/m2) and will be followed up for two years. Patients who have a significant smoking history (i.e. more than 10 pack years) and advanced nodal status (i.e. N2b, N2c or N3) are excluded from this trial. The primary outcome measure is the total number of acute (occurring during treatment or up to 90 days after treatment) and late (occurring between 90 days and two years after treatment) severe or life threatening (Grades 3-5) events, which has been chosen because these events result in significant morbidity and a reduction would be considered beneficial by both patients and clinicians. The sample size calculation is based on the total number of toxicity events rather than the number of patients affected. This ensures that the risk of underestimation of the toxicity burden is minimised. Secondary outcomes are overall survival, quality of life and feeding tube use. Current enrolment into the De-ESCALaTE trial [ISRCTN number: ISRCTN33522080] is 239 randomised patients with a target total of 304 patients. The IDSMC last reviewed the trial in May 2015 and suggested that the trial continue as planned. Clinical trial information: ISRCTN33522080.

TPS6105

Poster Session (Board #425a), Sat, 1:00 PM-4:30 PM

Patritumab (P) or placebo (PBO) plus cetuximab (C) and platinum-based therapy in squamous cell carcinoma of the head and neck (SCCHN): a phase 2 study. First Author: Kevin J. Harrington, Royal Marsden Hospital/Institute of Cancer Research, London, United Kingdom

A phase 2 study of tarloxotinib bromide (TRLX) in patients (Pts) with recurrent or metastatic (R/M) squamous cell carcinoma of the head and neck (SCCHN) or skin (SCCS). First Author: Danny Rischin, Peter MacCallum Cancer Centre, East Melbourne, Australia

Background: P is a fully human monoclonal antibody against human epidermal growth factor receptor 3. P blocks activation by the ligand, heregulin (HRG), inducing receptor internalization. Evidence is growing that HRG presence determines disease progression and survival; in a phase 2 study in non–small-cell lung cancer, P + erlotinib increased progression-free survival (PFS) in high HRG mRNA expression (HRG-high) patients. A phase 1b study in SCCHN demonstrated safety, tolerability and tumor response of P + C + cisplatin or carboplatin and informed the P phase 2 dose. January 2016 (N = 15) response rate = 47%; best responses = 3 complete response (CR), 4 partial response (PR) and 8 stable disease (SD). This phase 2 study (NCT02633800) evaluates first-line P + C + platinum (P arm) vs. PBO + C + platinum (PBO arm) in recurrent and/or metastatic (R/M) SCCHN. The primary objective is to evaluate PFS in the HRG-high population (P vs. PBO arms). Methods: This is arandomized, controlled, double-blind first-line study in Europe. Patients aged $ 18 years with confirmed R/M SCCHN, ECOG performance status # 1 and documented HRG expression (per archived or fresh biopsies) are eligible. The primary efficacy endpoint is PFS. Secondary endpoints include overall survival, overall response rate, pharmacokinetics and safety (including human antihuman antibody incidence). Approximately 105 patients will be stratified 2:1 by HRG status (70 high; 35 low), and then randomized 1:1 to the P or PBO arm. All patients will receive either intravenous P (18mg/kg loading dose [LD]; 9mg/kg maintenance dose [MD] every 3 weeks [q3w]) or PBO, and C (400mg/m2 LD; 250mg/m2 MD weekly) + # 6 cycles of cisplatin (100mg/m2 q3w) or carboplatin (area under the curve of 5). Patients demonstrating CR, PR or SD will be treated with P/ PBO + C + # 6 cycles platinum for the study duration (until all patients have died or $ 13 months postrandomization of last patient); those benefiting may continue treatment uninterrupted in an open-label extension phase until progressive disease, toxicity or withdrawal. Survival status will be obtained $ 13 months after discontinuation. The first patient was dosed December 2015 and enrollment is open. Clinical trial information: NCT02633800.

Background: R/M SCCHN/SCCS have a poor prognosis. Molecularly targeted EGFR therapies provide some clinical benefit but are limited by systemic ontarget side effects such as rash and diarrhea. SCCHN is a hypoxic tumor type. The presence of hypoxia has been shown to be associated with poorer outcomes and shorter survival. Additionally, tumor hypoxia upregulates WT EGFR signalling through several HIF-dependent mechanisms. TRLX is a hypoxia-activated prodrug that releases an irreversible pan-ErbB TKI targeting WT EGFR, mutant EGFR and HER2. Hypoxic tumor targeting using TRLX may allow a greater therapeutic and less dose-limiting systemic toxicity than seen with current EGFR TKIs. A Phase 1 study in pts with advanced solid tumors established the MTD of TRLX administered as a 1-hour weekly IV infusion at 150 mg/m2. The most common treatment-related adverse events were dose-dependent and included rash, QT prolongation, nausea, infusion reaction, vomiting, diarrhea and fatigue. Methods: A multicenter Phase 2 trial was initiated to evaluate the safety and activity of single-agent TRLX in pts with R/M SCCHN or SCCS (NCT02449681). The primary endpoint is RECIST 1.1 response rate. Secondary endpoints include PFS, duration of response, OS, PK and safety, as well as hypoxia PET imaging with [18F]HX-4 and molecular analyses of tumor tissue and plasma for biomarkers that may be associated with tumor response (including exome and mRNA analysis, EGFR-GRB2 proximity-ligand assay, and Veristrat plasma assay). Up to 68 pts will be enrolled with at least 10 pts in each of three disease specific groups (P16-positive oropharyngeal carcinoma, P16negative SCCHN, SCCS). Key eligibility criteria include measureable disease according to RECIST 1.1 and ECOG performance status 0-2. Prior systemic therapy is not required for SCCS. Prior EGFR TKI therapy is not allowed; prior EGFR antibody therapy is allowed. TRLX (150 mg/m2) is administered by IV infusion over 60 minutes on Days 1, 8, 15 and 22 of a 28day cycle. The study design incorporates a Simon two-stage design (alpha = 0.05; beta = 0.20) based on response rate. Recruitment is ongoing. Clinical trial information: NCT02449681.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Head and Neck Cancer TPS6106

Poster Session (Board #425b), Sat, 1:00 PM-4:30 PM

TPS6107

343s Poster Session (Board #426a), Sat, 1:00 PM-4:30 PM

Phase ib trial of dose-escalating AZD1775 in combination with concurrent radiation and cisplatin for intermediate and high risk head and neck squamous cell carcinoma. First Author: Bhishamjit S. Chera, The University of North Carolina at Chapel Hill, Chapel Hill, NC

Phase IB study of pembrolizumab in combination with chemoradiotherapy (CRT) for locally-advanced squamous cell carcinoma of the head and neck (LA-SCCHN). First Author: Steven Francis Powell, Sanford Health, Sioux Falls, SD

Background: Patients with significant tobacco histories (intermediate risk) and those who are HPV negative (high risk) have modest to poor outcomes and have been shown to have more disruptive p53 mutations. AZD1775 is a highly selective oral tyrosine kinase inhibitor of Wee1. Wee 1 inhibits cyclindependent kinases (CDK) 1 and 2. When DNA is damaged (e.g. radiation, chemotherapy), Wee1 inhibition is expected to abrogate the G2-M checkpoint leading to premature mitotic entry and consequently cell death. P53deficient tumors rely more heavily on the G2-M checkpoint and thus may be particularly sensitive to Wee1 inhibition. We hypothesize that AZD1775 in combination with chemoradiotherapy will improve clinical outcomes for patients with intermediate/high risk HNSCC. We are conducting an open label, single arm Phase 1b study (using a traditional 3+3 design) to identify the maximum tolerated dose (MTD) and recommended phase 2 dose of AZD1775 when added to standard of care chemoradiotherapy for the treatment of intermediate/high risk squamous cell cancer of the head and neck. Methods: Eligible patients will have T1-T4, N0-N3, M0 intermediate and high risk HNSCC defined as: primary tumors located in the oral cavity, larynx, or hypopharynx; HPV/p16 negative oropharynx; and HPV/p16 positive with . 10 pack-years smoking history. All patients will receive Intensity Modulated Radiotherapy Treatments as per standard of care. The total dose will be 70 Gy at 2Gy/fx, 35 fractions, Mon to Fri, for 7 weeks. Cisplatin at a dose of 40 mg/m2 IV will be infused day 1 of each week of radiation. AZD1775 will be taken orally twice daily Mon to Wed during the 7 week course of chemoradiotherapy. The starting dose level for AZD1775 is 50mg and escalates in increments of 50 mg for three additional levels (100 mg, 150 mg, and 200 mg). With a maximum of 6 patients enrolled at each dose level, a maximum of 24 patients will be enrolled. The MTD will be determined by the development of dose limiting toxicities 7 weeks of therapy. We will perform correlative studies on archival tissue (p53 mutational status) and mid-treatment biopsies (phosphor-CDK1(Tyr-15) and gH2AX levels). Clinical trial information: NCT02585973.

Background: The host immune response is critical for the clearance of SCCHN during CRT. Recent translational research has demonstrated that the programmed death receptor 1 (PD-1) and its ligand (PD-L1, PD-L2) interaction may play a role in immune escape of SCCHN during CRT. Pembrolizumab is a humanized IgG4 monoclonal antibody that blocks this interaction and may restore T-cell activity in this setting. PD-1 inhibition in combination with cisplatin-based CRT suggests synergy in a pre-clinical SCCHN mouse model. Clinical data on the safety and efficacy of this approach is needed to determine the potential of this therapy for the definitive management of LA-SCCHN. Methods: This is a single-arm, multi-site, openlabel trial of pembrolizumab used in combination with cisplatin-based, definitive CRT in patients with stage III-IVB SCCHN. 39 patients will be enrolled with safety and efficacy as the co-primary endpoints. Treatment consists of a loading dose of pembrolizumab 200 mg IV given 7 days prior to initiation of CRT and continued every 3 weeks during CRT and following completion of CRT for a total of 8 doses. CRT will consist of cisplatin 40 mg/ m2 IV weekly x 6 doses given concurrently with radiation fractionated at 2 Gy once daily for 35 fractions (total 70 Gy). Efficacy will be measured by the complete response (CR) rate on treatment end imaging (100 days postcompletion of CRT) or pathologic CR for those who undergo salvage surgery. Safety will be assessed by NCI-CTCAE v4.0 criteria. Secondary endpoints include best overall response rate (ORR), progression-free survival (PFS), overall survival (OS), locoregional control (LRC), and distant metastasis (DM) rate at 1- and 2- years. Patient reported outcomes (PRO) will be measured using the FACT H&N v4.0 survey. Correlative research will include analysis of circulating immunocyte populations during therapy and evaluation of tumor PD-L1 expression at baseline and in participants who have salvage surgery. The study is currently open with 6 of 39 participants enrolled. Clinical trial information: NCT02586207.

TPS6108

TPS6109

Poster Session (Board #426b), Sat, 1:00 PM-4:30 PM

Poster Session (Board #427a), Sat, 1:00 PM-4:30 PM

ORCA-2: A phase I study of olaparib in addition to cisplatin-based concurrent chemoradiotherapy for patients with high risk locally advanced squamous cell carcinoma of the head and neck. First Author: Martin David Forster, University College London, London, United Kingdom

A phase I study of metformin in combination with cisplatin and radiation in locally advanced head and neck squamous cell carcinoma. First Author: Trisha Michel Wise-Draper, University of Cincinnati Cancer Institute, Cincinnati, OH

Background: Standard care for locally advanced squamous cell carcinoma of the head and neck (HNSCC) is cisplatin-based chemoradiotherapy (C-CRT). Use of intensity modulated radiotherapy (IMRT) reduces toxicity to normal tissue but despite aggressive primary management many patients with high risk disease develop recurrence. The PARP-1 inhibitor olaparib inhibits DNA damage repair and may potentiate the anti-tumour activity of C-CRT. Preclinical data demonstrate potent radio-sensitisation but the optimal dosing schedule to enhance anti-tumour activity, whilst allowing normal tissue recovery, has not been investigated clinically. ORCA-2 aims to determine the recommended dose and duration of olaparib to give in combination with weekly cisplatin and IMRT. Methods: This is a phase I trial in which olaparib will be escalated by increasing both the dose (50, 100, 150 & 200 mg bd) and duration (3, 4 & 5 days). Olaparib tablets will be taken orally twice daily (bd) on a weekly dosing schedule (initially 50 mg bd for 3 sequential days), cisplatin 35 mg/m2weekly and IMRT 70 Gy in 35 fractions. Primary endpoint is occurrence of dose limiting toxicities over 13 weeks (7 weeks C-CRT and 6 weeks follow up). A novel dual escalation design called Product of Independent Beta Probabilities Escalation (PIPE) is being used. Escalation decisions are based on a pre-specified maximum allowable target toxicity level of 33%, predicted prior probabilities of toxicity in each dose-duration combination and cumulative toxicity data from a minimum of 2 patients in each cohort. This design allows flexibility and adaptability to provide better estimates of the recommended dose-duration combinations for further investigation than a standard 3+3 escalation. Eligible patients have newly diagnosed, histologically confirmed, high risk, locally advanced HNSCC and WHO performance status 0-1. Blood and fresh tissue will be collected for exploratory biological studies including evaluation of pharmacodynamic markers of PARP inhibition. The trial opened on 28 September 2015 and recruitment is ongoing. Clinical trial information: NCT02308072.

Background: Diabetic patients with head and neck squamous cell carcinoma (HNSCC) receiving the oral hypoglycemic, metformin, demonstrated improved survival compared to those not on metformin when treated with curative intent cisplatin and radiation (CRT) in retrospective analyses. Preclinically, metformin has been shown to inhibit mTOR, a signaling pathway that regulates cell growth and survival, commonly activated in HNSCC. Pretreatment with metformin resulted in a decrease in oral cavity tumors in mice. Here we are prospectively investigating escalating doses of metformin in combination with CRT in locally advanced HNSCC in non-diabetic patients, and the mechanism by which metformin may exert its effect. Methods: Patients with locally advanced HNSCC receive oral metformin administered in escalating doses (1000mg BID, 850mg TID or 1000mg TID) for 7-14 days prior to initiation of cisplatin (100 mg/m2 administered on days 1, 22 and 43) along with concurrent radiation therapy (70 Gy). Initially, metformin was escalated to assigned cohort dose over 7 days prior to CRT. However, quick escalation was poorly tolerated and therefore, the protocol was modified to allow a slower dose escalation over 14 days. Metformin is continued throughout treatment. Blood samples are collected before and after metformin as well as during chemotherapy. Flow cytometry to detect percent circulating immune activated cells (CD8+ and PD-1+) and clinical laboratory tests including serum glucose, vitamin B12 and C-peptide levels are being performed. Clinical trial information: NCT02325401. Dose-escalation schedule. Dose Dose Level

Metformin Daily (in divided doses)

Cisplatin Days 1, 22, 43

Level -1 Level 1 Level 2 Level 3 Expansion

1500mg 2000mg 2550mg 3000mg MTD

100mg/m2 100mg/m2 100mg/m2 100mg/m2 100mg/m2

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

344s TPS6110

Head and Neck Cancer Poster Session (Board #427b), Sat, 1:00 PM-4:30 PM

Immunotherapy with pembrolizumab in surgically resectable head and neck squamous cell carcinoma. First Author: Ravindra Uppaluri, Washington University School of Medicine, St. Louis, MO Background: Patients with locally advanced, surgically resectable Stage III/IV human papillomavirus (HPV)-negative head and neck squamous cell carcinomas (HNSCC) suffer from locoregional and distant metastatic (LRR/DM) failures. This population is thus an ideal target for checkpoint blockade therapy as the mutational landscape of these carcinogen-induced malignancies bears a rich array of antigenic targets for immune based therapeutics. Methods: Trial design and statistical methods This is a Phase II neoadjuvant and post-operative pembrolizumab trial. After informed consent and registration, tumor biopsy and peripheral blood will be obtained and patients will receive a single pre-operative dose of pembrolizumab (10 mg/kg). After two to three weeks, patients undergo definitive surgery where tissue is again harvested. Patients then proceed with standard post-operative adjuvant therapy dictated by the pathologic report. Highrisk patients with extracapsular extension and/or positive margins receive additional maintenance pembrolizumab for the first year after treatment. Sample size justification: Historical data has shown a 35% rate of LRR and DM at 1 year in HNSCC patients undergoing multimodality treatment and our primary endpoint is to reduce LRR/DM to 15%. We will recruit 46 patients with 31 of these expected to be high risk and qualify for post-operative pembrolizumab. This cohort will allow us to detect with 80% power (alpha = 0.05) a difference of at least 20% in the 1year LRR/DM rate. Correlative studies: Clinical outcomes will be correlated with tumor PD-L1 expression, neoantigen burden as defined by tumor exome sequencing and RNA-Seq in matched pre-and post-pembrolizumab treated samples. Finally, patient specific tumor infiltrating lymphocyte cultures will be used to define antigen-specific T cells targeted by pembrolizumab. Major eligibility criteria Patients must have stage III/IV HPV-negative HNSCC with planned primary surgical management. Current enrollment 10 of 46 patients enrolled Trial registry number NCT02296684 Clinical trial information: NCT02296684.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Health Services Research and Quality of Care LBA6500

Oral Abstract Session, Mon, 9:45 AM-12:45 PM

Global differences in cancer drug prices: A comparative analysis. First Author: Daniel A. Goldstein, Rabin Medical Center, Petach Tkvah, Israel

The full, final text of this abstract will be available at abstracts.asco.org at 7:30 AM (EDT) on Monday, June 6, 2016, and in the Annual Meeting Proceedings online supplement to the June 20, 2016, issue of the Journal of Clinical Oncology. Onsite at the Meeting, this abstract will be printed in the Monday edition of ASCO Daily News.

6501

345s Oral Abstract Session, Mon, 9:45 AM-12:45 PM

Clinical use of the 21-gene assay and patient experiences in early-stage breast cancer. First Author: Steven J. Katz, University of Michigan, Ann Arbor, MI Background: The 21-gene assay enables oncologists to tailor chemotherapy in earlystage, estrogen-receptor positive breast cancer. Few studies have studied the impact of test results on chemotherapy recommendations and receipt, and on patient experiences. Methods: We surveyed 3781 women with breast cancer treated in 2013-14 selected by the Georgia and Los Angeles SEER Registries (response n= 2593, 69%). Women reported their oncologist’s recommendations, chemotherapy receipt, and treatment decision satisfaction. SEER registries linked Genomic Health, Inc test results to patients. We selected 1287 patients with Stage I/II, ER+, HER2- disease and categorized them into 3 groups 1) node-negative favorable (no high-risk features); 2) node-negative, less favorable (age at diagnosis , 50 yrs or Grade 3 tumor); and 3) node-positive. Regression models - adjusted for comorbidity, education, income, race, location, and sampling design - examined test scores and clinical factors on chemotherapy recommendation and receipt. Those tested rated their satisfaction on a 5-point Likert scale (1=not satisfied, 5=totally satisfied). Results: Overall 53% of selected women were tested (57%, 65%, and 31%, for groups 1, 2, & 3). Recommendations for chemotherapy and chemotherapy receipt were consistent with risk scores (Table), and racial differences were not significant in regression models. Most (82%) tested patients reported that the test helped decision making yet 14% did not recall their test status. The x̅(SD) testing satisfaction score was 4.6(0.7). Conclusions: Oncologists personalize treatment based on assay results, even in node-positive disease. High satisfaction and absence of disparities suggests that precision medicine advances have improved the quality of systemic breast cancer treatment. Node-neg, favorable (n=718)

Node-neg, unfavorable (n=283)

Node pos (n=286)

Recommended Against/ Got Recommended Against/ Got Recommended Got Neutral/For Chemo Neutral/For Chemo Against/Neutral/For Chemo No test (n=608) RS 0-17 (n=427) 18-30 (n=204) .30 (n=48)

54 / 21 / 25

14

25 / 9 / 66

59

9 / 9 / 82

83

78 / 10 / 12

3

64 / 11 / 25

6

49 / 12 / 39

22

38 / 19 / 43 0 / 0 / 100

38 100

20 / 25 / 55 3 / 5 / 92

52 94

9 / 22 / 69 0 / 0 / 100

64 100

%s reported. RS: Recurrence Score

6502

Oral Abstract Session, Mon, 9:45 AM-12:45 PM

Price trajectory of individual cancer drugs following launch. First Author: Noa Gordon, Davidoff Centre, Rabin Medical Centre, Petach Tikva, Israel Background: Cancer drug launch prices have increased in recent years. However, it is not fully understood how individual drug prices change over time following launch. For example, following the approval of imatinib, the price had increased threefold over a decade. The objective of this study was to measure the price trajectory of 10 cancer drugs following their launch into the US marketplace. Methods: We studied the quarterly changes in prices of average monthly doses for a cohort of 10 patented IV cancer drugs that were approved by the FDA between 1997 and 2012. In order to account for discounts and rebates, we used the Average Sales Price (ASP) published by the Center for Medicare and Medicaid Services (CMS). US inflation rates were obtained from the United States Department of Labor, and prices were adjusted for inflation. For each drug we calculated the cumulative drug price change. Data was analyzed using IBM SPSS Statistics software. Results: With an average follow-up period of 8 years, the median percent of price change for all drugs was 22% (range 3-74%). After adjusting for inflation, the median percent of price change was 6% (range -10-44%). Rituximab and trastuzumab follow a similar pattern in price increase over time and inflation-adjusted prices rose since approval by 44% and 40% respectively. The inflation-adjusted price of pemetrexed rose by 26%. We clustered drugs for indication, year of approval and company but did not find any significant trends. Conclusions: Cancer drug prices may change substantially following launch. Prices may increase by as much as 44% even after adjusting for inflation. When discussing value, we must take into account that prices are not always static. Generic Name

Follow-up time (years)

ASP change (US$)

ASP change (%)

Inflation-adjusted ASP change (%)

rituximab trastuzumab pemetrexed panitumumab bevacizumab nab-paclitaxel ipilimumab cetuximab denosumab pertuzumab

11 11 11 8 11 10 4 11 4 2

3041 2396 2632 2020 2258 1578 4660 78 127 143

74 69 52 25 24 20 11 8 7 3

44 40 26 12 3 2 8 -10 4 3

6503

Oral Abstract Session, Mon, 9:45 AM-12:45 PM

Access to accredited cancer hospitals within federal exchange plans under the Affordable Care Act. First Author: Kenneth L. Kehl, The University of Texas MD Anderson Cancer Center, Houston, TX Background: The Affordable Care Act (ACA) of 2010 expanded access to health insurance in the U.S. However, concern is increasing about the breadth of provider networks within ACA exchange plans. For patients with cancer, narrow networks may restrict access to specialized care and clinical trials. To assess the scope of this problem, we analyzed how often exchange networks in the four most populous states on the federal marketplace include hospitals accredited by the American College of Surgeons Commission on Cancer (CoC) and NCI-designated cancer centers. Methods: We downloaded publicly available machine-readable 2016 provider network data for individual marketplace plans in Florida, Texas, Illinois, and Pennsylvania. We extracted a list of in-network hospitals for each plan. Since many plans share the same networks, our unit of analysis was each unique network. We matched CoC-accredited hospitals and NCI-designated centers (cancer centers and comprehensive cancer centers) in these states to their National Provider Identification (NPI) numbers. We linked the NPI numbers to the provider networks to assess how often CoC-accredited hospitals and NCI centers were included in each network. Inclusion of NCI centers was compared across states with the chi-square test. Results: 1181 plans were identified and linked to in-network facility NPIs; these plans contained 87 unique facility networks. 86 networks (99%) contained at least one of the 306 CoC-accredited hospitals located in these four states. 51 networks (59%) included an NCI-designated cancer center. The proportion of networks including an NCI center varied by state, ranging from 27% in Florida to 82% in Pennsylvania (P=0.01). Conclusions: Almost all federal exchange networks in our sample contained at least one CoC-accredited hospital. Most networks included at least one hospital affiliated with an NCI-designated cancer center, but there was substantial variation in access to NCI centers across states. Work is ongoing to compare the remaining federal exchange networks to state-based exchanges and other public and private insurance networks, and further research is necessary to assess the impact of narrow networks on patients and clinical research.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

346s

Health Services Research and Quality of Care

6504

Oral Abstract Session, Mon, 9:45 AM-12:45 PM

6505

Oral Abstract Session, Mon, 9:45 AM-12:45 PM

Health plan selection and out-of-pocket costs for cancer patients in the health insurance exchange. First Author: Callie Lynne Gable, Duke University, Durham, NC

A disease management pilot program in a Medicare-age population with cancer. First Author: Marcus A. Neubauer, The US Oncology Network, McKesson Specialty Health, Seattle, WA

Background: Little is known about variation in out-of-pocket costs for cancer patients in exchange-based health insurance plans. Despite annual maximums, it is unclear to what extent the Affordable Care Act limits out-ofpocket costs, and what role indicators like metal tier and annual maximums should play in consumer plan selection. Methods: This was a "secret shopper" study to describe out-of-pocket costs incurred by health insurance exchange plans available to residents of 6 North Carolina counties. In MarchApril 2015, HealthCare.gov was searched for individual coverage for a hypothetical, 55-year-old, non-smoking man with chronic myelogenous leukemia (CML) taking imatinib. CML was chosen since it typically requires years of potentially expensive, prescription, anticancer therapy. Metal tier was varied, and patient income was varied from 100%-300% of the federal poverty level (FPL). Using HealthCare.gov data for each plan, total annual out-of-pocket costs were calculated as annual premium plus maximum annual out-of-pocket cost. Results: The hypothetical patient was eligible for 141 bronze and 210 silver plans with variation in income. High-deductible plans (deductible . $1300) accounted for all bronze and 54% of silver plans.At 100% of FPL, total silver out-of-pocket costs after subsidies were considerably lower than bronze costs. At 200% of FPL, total out-of-pocket costs were similar between silver and bronze plans. At 300% of FPL, the hypothetical patient would pay, on average, $1470 per year less out-ofpocket by selecting a bronze rather than silver plan. Conclusions: For a cancer patient on expensive therapy, metal tier or stated annual out-ofpocket maximums were unreliable indicators of financial protection. Patients face numerous plan choices, with costs varying considerably within and across metal tiers. For a cancer patient expecting high out-of-pocket costs, the conventional wisdom of choosing based on metal tier alone may result in higher costs. Research should focus on optimizing plan selection for patients with cancer.

Background: Cancer care delivery and reimbursement reform are focused on controlling costs and improving quality, thereby increasing value. Medicareage patients have more cancer and more co-morbidities, and are at higher risk of complications from treatment. A program (Innovent Oncology, IO) conducted at Texas Oncology (TxO) and sponsored by Aetna employs evidence-based treatment Pathways (PW), a disease management call center and an advance care planning (ACP) program with the goal of improving patient care and reducing total costs. Methods: Over two years (6/1/ 13 to 5/31/15), IO enrolled 390 Aetna Medicare Advantage pts with PW diagnoses, treated by TxO. PW compliance was tracked. Innovent Patient Support Services (PSS) calls occurred at regular intervals during chemotherapy. Symptoms and quality of life were assessed. PSS nurses also introduced ACP principles. Hospice enrollment (% patients admitted to hospice who expired during the program) and PSS patient satisfaction (% of patients very satisfied with PSS services) were also tracked. For cost evaluations, benchmark costs are derived from Aetna’s administrative claims data for the matched cohort. For inpatient and ER events, we use an average cost per day or per ER visit to approximate cost. The matched cohort’s data are adjusted for diagnosis and age to match the Innovent group’s profile. Results: For first year and second year, respectively, PW adherence was 81% and 84%; Hospice enrollment was 55% and to be determined; PSS patient satisfaction was 93% and 94%. The percent savings for the program are indicated in the Table. Conclusions: Oncology practices are capable of improving the quality of cancer care and reducing costs. We show that a practice-based program that applies rigor to clinical pathways; uses an affiliated call center to help manage older patients receiving chemotherapy; and is properly funded by a payer sponsor can be a successful model to achieve these important goals.

% FPL

Average annual silver out-of-pocket maximum + premium

Average annual bronze out-of-pocket maximum + premium

100 200 300

$1792 $7279 $10,266

$6188 $6895 $8796

6506

Oral Abstract Session, Mon, 9:45 AM-12:45 PM

Human capital costs of obtaining oral anticancer medications. First Author: Yu-Ning Wong, Fox Chase Cancer Center, Philadelphia, PA Background: Oral anticancer medications (OAMs) are commonly prescribed for various cancers. Initiating OAMs can involve extensive effort from clinic staff, including obtaining prior authorization and copayment assistance and coordinating with specialty pharmacies. We conducted a retrospective review to characterize potential delays in treatment (Tx). Methods: We included prostate and kidney cancer patients (pts) prescribed on-label OAMs at an academic center between 8/1/2014 and 8/31/2015. The following was extracted from nursing notes and pt charts: demographics, date of OAM prescription (Rx), date of first dose, number and reason for staff phone calls required to obtain OAM, copay amount and whether financial assistance was requested. We used Fisher’s exact and Kruskal-Wallis tests to compare differences between OAMs. Results: In 116 pts (55% kidney cancer, 45% prostate cancer) with 149 unique Rx, median age 65 (27-88), 85% male, 89% white, 78% had Rx drug coverage. See Table for results by drug. In pts with information available (56 unique Rx), quoted average/median copay/ month was $1,224/$175 ($0-9453), but dropped to $42/$3 per month ($0$1056) after copayment assistance program utilization. Conclusions: The initiation of on-label OAMs was a labor intensive process and on average it took 2 weeks to obtain a drug. The significant OAM dependent differences found may be related to specific drug plans, assistance programs or unmeasured variables and needs to be further validated in a larger sample. Outof-pocket costs were variable and required considerable staff effort to mitigate via financial assistance. The process of initiating OAMs can be a barrier to care and should be a focus of quality improvement initiatives. # Rx Rx transferred to a different pharmacy (%) Avg days from Rx – Tx Start Rx requiring 2-4 calls by staff to obtain drug (%) Rx requiring 5-8 calls by staff to obtain drug (%) Reason for calls: Prior Authorization (%) Reason for calls: Instructions (%) Copay assistance utilized (%)

Abiraterone

Enzalutamide

Sunitinib

Pazopanib

Everolimus

Axitinib

Total

P-Value

26 35 20 62 12 43 45 38

28 32 15 57 18 33 51 46

26 50 19 42 15 40 38 35

22 50 14 73 14 26 46 27

10 10 7 20 20 38 54 40

37 11 10 32 14 41 50 27

149 32 14 49 15 36 47 35

0.003 0.001 0.009 0.974 0.884 0.96 0.632

Percent $ savings (control vs investigational use). Chemotherapy and Supportive Care Benchmark Cost Actual Cost Savings Percent

6507

$11,420,791 $9,081,351 20.48%

Inpatient

ER

Total

$3,925,662 $305,247 $15,651,701 $3,323,072 $281,140 $12,685,563 15.35% 7.90% 18.95%

Oral Abstract Session, Mon, 9:45 AM-12:45 PM

Can lay health workers achieve the triple aim? Results from the Engagement of Patients with Advanced Cancer (EPAC) trial. First Author: Manali I. Patel, Stanford Hospital, Menlo Park, CA Background: Rising cancer costs demand innovative approaches to improve patient satisfaction and outcomes while reducing expenditures. We developed a novel model that integrates lay health workers (LHWs) into care to assist with documentation of patients’ goals and symptoms. We partnered with the VA to implement the approach and to study its efficacy on satisfaction, utilization, and costs. Methods: Newly diagnosed patients with stage 3 and 4 and recurrent cancer were randomly assigned to receive either the LHW intervention or usual care alone. Analyses were performed using intentto-treat. x², t-tests, or Wilcoxon Rank Sum were used to compare baseline characteristics and outcomes (satisfaction, utilization, and costs) between the intervention and usual care groups. Mean differences in total costs of care were compared using Wilcoxon Rank Sum. We used regression analysis to estimate differences in satisfaction between baseline and six months between the study arms. Results: Between August 15, 2013 and February 2, 2015, 213 patients were randomized, with 108 in the intervention and 105 in usual care. Median age was 69; 99% were male. The majority (78%) were non-Hispanic white with 6% non-Hispanic black, 2% Hispanic, and 2% Asian Pacific Islander. The majority (54%) were unmarried. The average travel distance was 92 miles. Lung cancer was the most common diagnosis (36%). The majority had stage 4 disease (54%). Six months after enrollment, 45% of patients had died in both study arms. More patients in the intervention had a documented goals of care (93% versus 29%; p,0.001) and advance directive (78% versus 38%; p,0.001) and received hospice services (40% versus 23%; p=0.008). Satisfaction was also higher among patients in the intervention group (p,0.001) compared to patients in usual care alone. Patients in the intervention had lower mean total healthcare costs as compared to patients in usual care alone, although not statistically significant ($134,901 versus $116,123; p=0.2). Conclusions: Incorporating timely documentation of patients’ goals of care and provision of ongoing patient support via LHWs improved patient satisfaction and reduced total healthcare expenditures.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

Health Services Research and Quality of Care 6508

Oral Abstract Session, Mon, 9:45 AM-12:45 PM

Associations between job supports and job retention among working patients with colorectal cancer. First Author: Christine Marie Veenstra, University of Michigan, Ann Arbor, MI Background: Serious illness may cause substantial economic hardship, exacerbated by illness- or treatment-related job loss. Colorectal cancer (CRC) is a common, serious illness with a rising incidence among younger, workingage patients. Job supports may help working patients with CRC retain their jobs and alleviate financial burden. Methods: In 2011-14, we surveyed Stage III CRC patients from Detroit and Georgia SEER registries. Job retention was the outcome measure. Availability of job support benefits (employer-based health insurance, flexible schedule, paid sick leave, extended sick leave, disability, unpaid leave) was the primary independent variable. We assessed associations between social determinants, environment, job supports and job retention, using a multivariate logistic regression model first controlling only for patient-level social and environmental covariates and then adding in job supports. Results: 1301/1901 eligible patients returned completed surveys (68% response rate). Only the 546 patients who were working at the time of diagnosis were included in this study. 45% of working patients reported that they stopped working altogether as a result of cancer diagnosis and treatment. In the individual- and area-level adjusted model, only education and household income were significantly associated with job retention. In the fully adjusted model including job support benefits, education and income were no longer statistically significant and only employer-based health insurance (OR = 2.63; 95% CI = 1.34-5.14; p , 0.01), flexible work schedule (OR = 6.19; 95% CI = 3.56-10.78; p , 0.01) and paid sick leave (OR = 2.26; 95% CI = 1.114.61; p = 0.03) were significantly associated with job retention. Conclusions: Nearly half of working patients were unable to retain their jobs due to CRC diagnosis and treatment. Employer-based health insurance, flexible work schedule, and paid sick leave were associated with significantly greater likelihood of job retention. Based on our findings, an investment in job support benefits by employers and government could provide long-term societal dividends by enabling job retention and reducing financial burden, especially among young working patients.

6509

347s

Poster Discussion Session; Displayed in Poster Session (Board #1), Sat, 1:00 PM-4:30 PM, Discussed in Poster Discussion Session, Sat, 4:45 PM-6:00 PM

The opportunity cost (OC) of performing radical prostatectomy (RP) by low volume (LV) providers (Prov). First Author: Sarmad Sadeghi, USC Norris Comprehensive Cancer Center, Los Angeles, CA Background: Evidence suggests a link between experience and improved oncological and non-oncological outcomes for RP. We explored the frequencies of RPs by LV Prov that occur within 100 miles of highest volume Prov and the potential savings associated with referring them to highest volume Prov. Methods: We analyzed Medicare Provider Utilization and Payment Data for 2012 and 2013, SEER Data 2003-2012. A Markov model was used to compare outcomes of high and low volume Prov in terms of effectiveness (oncological and non-oncological) and cost of oncological treatment (Tx) failure. Medicare Prov were grouped according to the number of RPs. Prov with below median volume were designated as LV Prov. Using the Prov zip codes we calculated the travel distance required for referral. Results were extrapolated using SEER data to non Medicare patients. Opportunity costs of Tx failure were calculated for RPs within 100 miles of highest volume Prov. Results: 15% of Prov performed more than 30% of the RP with at least 30 cases per year. There were 241,740, and 238,590 prostate cancer cases, and 32 and 33% of these underwent RP, in 2012 and 2013. The Markov model projected that referral to high volume Prov was associated with 50 fewer biochemical failures resulting in $1.8 mil savings (over 20 years) as well as 233 fewer post-op hospitalization days per 1000 referrals. Conclusions: While reduced LOS does not impact reimbursement, it does free up societal resources and therefore results in savings. Costs of avoidable Tx failures also represent significant savings. The potential for improved outcomes and reduced costs of care represent an OC associated with performing RP by LV Prov. Referrals within 100 miles are feasible and can improve outcomes and save societal costs. National Extrapolations Opportunity for

Medicare Data Prov Grp N Prov Prov Vol 2012

All .=30 50th pctl 75th pctl 2013 All .=30 50th pctl 75th pctl

499 72 247 376 470 66 234 352

N RP (%)

11-255 10168 (100%) .=30 3374 (33%) ,16 3124 (31%) ,23 5480 (54%) 11-236 9456 (100%) .=30 3021 (32%) ,16 2945 (31%) ,23 5087 (54%)

% RP 200 V/ capita Bulgaria / 66 Croatia / 81 Czech Republic / 91 Estonia / 69 Latvia / 62 Lithuania / 79 Poland / 81 Portugal / 81 Romania / 55

Cyprus / 105 Denmark / 163 Finland / 125 Greece / 127 Hungary / 105 Ireland / 164 Italy / 161 Malta / 134 Slovakia / 107 Slovenia / 139 Spain / 129 United Kingdom / 136

Austria / 266 Belgium / 227 France / 212 Germany / 265 Luxembourg / 323 Netherlands / 264 Sweden / 223

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.

372s 6620

Health Services Research and Quality of Care Poster Session (Board #103), Sat, 1:00 PM-4:30 PM

6621

Poster Session (Board #104), Sat, 1:00 PM-4:30 PM

Is there a link between value as defined by ESMO-MCBS and uptake of new drugs? First Author: Peter Lindgren, Institute of Health Economics, Lund, Sweden

Health care resource utilization (HCRU) in relapsed/refractory multiple myeloma (RRMM): Results from PREAMBLE. First Author: Hartmut Goldschmidt, Heidelberg University Hospital, Heidelberg, Germany

Background: There has been an increased interest in ways of measuring the value of new therapeutic options in oncology. Recently the European Society for Medical Oncology Magnitude of Clinical Benefit Scale (ESMO-MCBS) have been introduced as one such measure. The objective of the scale is to assist decisions about value and cost-effectiveness of new therapies by classifying therapeutic advances according to their relative clinical benefit based on pre-defined criteria. We hypothesize that the perceived value of a new drug would be a major driver of its widespread use, and therefore that one way of establishing the extent of which the scale captures value would be how well it correlates with actual use. Methods: This was a retrospective study analyzing data on sales from the member states of the European Union plus Norway and Switzerland. The study included drugs launched between 2004 and 2013 also having a published ESMO-MCBS score. In cases where more than one score had be assigned, we used the one related to the first published trial. We analyzed uptake of the drugs for a time period up to three years after launch. Uptake was defined as sales (V) per case. Death caused by the type of cancer the drug being indicat