IASLC ATLAS OF EGFR TESTING IN LUNG CANCER

Transcription

1 INTERNATIONAL ASSOCIATION FOR THE STUDY OF LUNG CANCER IASLC ATLAS OF EGFR TESTING IN LUNG CANCER Conquering Thoracic Cancers Worldwide EDITED BY TONY S. MOK, MD DAVID P. CARBONE, MD, PHD FRED R. HIRSCH, MD, PHD Del18 E709X T790M G719X Ins20 S7681 Ins19 Del19 L858R L861Q

2

3 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER

4 International Association for the Study of Lung Cancer, Aurora, CO, U.S.A. Editors: Tony S. Mok, MD David P. Carbone, MD, PhD Fred R. Hirsch, MD, PhD An IASLC publication published by Editorial Rx Press, North Fort Myers, FL, U.S.A. Cover and interior design by Amy Boches, Biographics IASLC Office: IASLC, East Colfax Ave., Unit 10, Aurora, Colorado 80011, U.S.A. First Editorial Rx Press Printing October ISBN: Copyright 2017 International Association for the Study of Lung Cancer All rights reserved Cover images: Top image: Courtesy of Dara Aisner, MD. Bottom left image: Reprinted by permission from Macmillan Publishers Ltd: Modern Pathology; 2012; 25: , Bottom right image: Reprinted from Cancer Sci. 2016; 107: courtesy of Yoshihisa Kobayashi, MD, Tetsuya Mitsudomi, MD, PhD, under a Creative Commons License CC-BY-NC- ND Without limiting the rights under copyright reserved above, no part of this publication may be reproduced, stored in or introduced into a retrieval system, or transmitted in any form, or by any means without prior written permission. While the information in this book is believed to be true and accurate as of the publication date, neither the IASLC nor the editors nor the publisher can accept any legal responsibility for any errors or omissions that may be made. The publisher makes no warranty, express or implied, with response to the material contained therein.

5 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER EDITED BY TONY S. MOK, MD DAVID P. CARBONE, MD, PHD FRED R. HIRSCH, MD, PHD AN INTERNATIONAL ASSOCIATION FOR THE STUDY OF LUNG CANCER PUBLICATION Editorial Rx Press North Fort Myers, FL

6 Acknowledgments IASLC acknowledges the generous funding and support provided by AstraZeneca for the IASLC Atlas of EGFR Testing in Lung Cancer. The coeditors and contributors also acknowledge the assistance of Jacinta Wiens, MS, PhD, Scientific Affairs Project Manager, IASLC, for coordinating the project; the editorial assistance of Joy Curzio and Lori Alexander, MTPW, ELS, MWC; the graphic arts talents of Amy Boches, Biographics; and, the publishing support of Deb Whippen, Editor and Publisher, Editorial Rx Press, for the publication of this text.

7 Contents Contributors Abbreviations...8 Manufacturers...9 Introduction Therapeutic Perspectives EGFR Testing Sample Acquisition, Processing, and General Diagnostic Procedures EGFR Gene Mutations Types of Assays for EGFR Testing Reporting, Interpretations, and Quality Assurance Access to Testing Guidelines and Algorithms Summary and Future Perspectives...65 References...69

8 6 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER Contributors Editors Tony S. Mok, MD Professor Department of Clinical Oncology Chinese University of Hong Kong Prince of Wales Hospital Hong Kong, China David P. Carbone, MD, PhD Professor and Director James Thoracic Center The Ohio State University Comprehensive Cancer Center Ohio, U.S.A. Fred R. Hirsch, MD, PhD Professor of Medicine and Pathology Pia and Fred R. Hirsch Endowed Chair University of Colorado Cancer Center CEO, International Association for the Study of Lung Cancer (IASLC) Colorado, U.S.A. Contributing Authors Myung-Ju Ahn, MD, PhD Professor Division of Hematology-Oncology Department of Medicine Samsung Medical Center Sungkyunkwan University School of Medicine Seoul, South Korea Dara L. Aisner, MD, PhD Associate Professor Department of Pathology School of Medicine University of Colorado Anschutz Medical Campus Colorado, U.S.A. Sanja Dacic, MD, PhD Professor Department of Pathology University of Pittsburgh Medical Center Pennsylvania, U.S.A. Leora Horn, MD, MSc Associate Professor of Medicine Director, Thoracic Oncology Program Assistant Vice Chairman for Faculty Development Department of Medicine Vanderbilt University Medical Center Tennessee, U.S.A. Peter B. Illei, MD Assistant Professor of Pathology Assistant Professor of Oncology Department of Pathology Johns Hopkins University School of Medicine Maryland, U.S.A. Yuichi Ishikawa, MD, PhD Chief, Division of Pathology Director, Clinicopathology Center Vice Director, Cancer Institute Keio University The Cancer Institute Hospital of JFCR Tokyo, Japan Pasi A. Jänne MD, PhD Director, Lowe Center for Thoracic Oncology Director, Belfer Center for Applied Cancer Science Senior Physician Professor of Medicine Dana-Farber Cancer Institute Harvard Medical School Massachusetts, U.S.A. Keith M. Kerr, FRCPath Professor Department of Pathology Aberdeen University Medical School, Aberdeen Royal Infirmary Aberdeen, Scotland, United Kingdom Philip Mack, PhD Professor Director of Molecular Pharmacology Department of Internal Medicine Division of Hematology and Oncology UC Davis Comprehensive Cancer Center California, U.S.A. Alberto M. Marchevsky, MD Director Pulmonary and Mediastinal Pathology Cedars-Sinai Medical Center Clinical Professor of Pathology David Geffen UCLA School of Medicine California, U.S.A.

9 CONTRIBUTORS 7 Geoffrey R. Oxnard, MD Assistant Professor of Medicine Dana-Farber Cancer Institute Harvard Medical School Massachusetts, U.S.A. Walter Weder, MD Professor and Chair Department of Thoracic Surgery University Hospital Zurich Zurich, Switzerland Keunchil Park, MD, PhD Professor Division of Hematology-Oncology Department of Medicine Samsung Medical Center Sungkyunkwan University School of Medicine Seoul, South Korea Luis Paz-Ares, MD Associate Professor Chair, Medical Oncology Department University Complutense Hospital Doce De Octubre Madrid, Spain Solange Peters, MD, PhD CHUV Medical Oncology Service Chief Chair of Thoracic Oncology Associate Professor Oncology Department Centre Hospilalier Universitaire Vaudois Lausanne, Switzerland Daniel SW Tan, MBBS, MRCP, PhD Senior Consultant, Division of Medical Oncology Principal Investigator, Cancer Therapeutics Research Laboratory National Cancer Centre Singapore Singapore Erik Thunnissen, MD, PhD Consultant Pathologist VU University Medical center Amsterdam, The Netherlands Ming Sound Tsao, MD, FRCPC Pathologist, Senior Scientist, and Professor M. Qasim Choksi Chair in Lung Cancer Translational Research Princess Margaret Cancer Centre, University Health Network Department of Laboratory Medicine and Pathobiology University of Toronto Toronto, Canada Jacinta Wiens, PhD Scientific Affairs Project Manager International Association for the Study of Lung Cancer Colorado, U.S.A. Ignacio I. Wistuba, MD Professor and Chair Anderson Clinical Faculty Chair for Cancer Treatment and Research Department of Translational Molecular Pathology The University of Texas MD Anderson Cancer Center Texas, U.S.A. James Chih-Hsin Yang, MD, PhD Distinguished Professor, Director Graduate Institute of Oncology College of Medicine, National Taiwan University Director, Department of Oncology National Taiwan University Hospital Taipei City, Taiwan Yasushi Yatabe, MD, PhD Chief Department of Pathology and Molecular Diagnostics Aichi Cancer Center Nagoya, Japan Rex Yung, MD Pulmonologist Greater Baltimore Medical Center Maryland, U.S.A.

10 8 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER Abbreviations The following abbreviations are used in the text without being expanded. ALK: Anaplastic lymphoma kinase BRAF: B-raf proto-oncogene cfdna: Cell-free DNA ctdna: Circulating-tumor DNA EDTA: Ethylenediaminetetraacetic acid EGFR: Epidermal growth factor receptor FISH: Fluorescent in situ hybridization FDA: Food and Drug Administration H&E: Hematoxylin and eosin HER2: Human epidermal growth factor receptor 2 HGF: Hepatocyte growth factor IHC: Immunohistochemistry KRAS: Kirsten rate sarcoma viral oncogene homolog MAPK: Mitogen-activated protein kinase MET: MET proto-oncogene NSCLC: Non-small cell lung cancer PCR: Polymerase chain reaction PD-L1: Programmed cell death ligand-1 P13KCA: Phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit α PTEN: Phosphatase and tensin homolog RET: RET proto-oncogene ROS1: c-ros oncogene 1 RTK: Receptor tyrosine kinase SCLC: Small cell lung cancer STAT: Signal transducer and activator of transcription TTF1: Thyroid transcription factor-1

11 MANUFACTURERS 9 Manufacturers The following manufacturers and their EGFR testing-related products are noted in this Atlas. The location given for each manufacturer is not the only location; most manufacturers have offices worldwide. Agena Bioscience San Diego, California, USA MassARRAY systems, OncoCarta Panel v1.0, Lung Carta Panel Applied Biosystems Foster City, California, USA SNapShot Multiplex Kit Hologic Marlborough, Massachusetts, USA CyoLyt Solution Illumina, Inc. San Diego, California, USA MiSeq Sequencing System, HiSeq Sequencing System Qiagen N.V. Venlo, the Netherlands therascreen EGFR RCQ PCR Kit Roche Molecular Systems, Inc. Pleasanton, California, USA cobas EGFR Mutation Test v2 Streck Omaha, Nebraska, USA Cell-Free DNA BCT Thermo Fisher Scientific Waltham, Masschusetts, USA Oncomine Dx Target Test, Ion Personal Genome Machine (PGM) (Ion Torrent)

12

13 Introduction By Tony S. Mok, David P. Carbone, and Fred R. Hirsch The discovery of somatic driver mutations in EGFR over a decade ago set the stage for science-based precision medicine in the management of advanced NSCLC. Since then, research has achieved many milestones that have transformed the clinical management of this disease. Starting with the IPASS study, we demonstrated that selection of patients by the presence of EGFR mutations rather than clinical/pathologic characteristics for treatment with first-line EGFR TKIs resulted in improved outcomes with less toxicity compared to chemotherapy. We now have more potent second-generation EGFR TKIs that may be superior to the first-generation ones, and have also found that EGFR TKIs may potentially be beneficially combined with other agents, such as bevacizumab. Identification of EGFR exon 20 T790M mutations as a common resistance mechanism, and selectively targeted by the third-generation EGFR TKI osimertinib, defined standard therapy for patients with acquired resistance to EGFR TKIs. All of these developments have helped to establish this new molecular treatment paradigm, making precision medicine a reality for many thousands of patients with lung cancer. Testing for EGFR mutations was the key to this success. From the time when EGFR testing became available at academic centers to general acceptance of EGFR testing as an essential aspect of lung cancer management globally, intense research efforts have been dedicated to the advancement in technology, standardization, validation, and interpretation of this analysis. Now that we are blessed with multiple efficient platforms, we need consensus on the establishment and optimization of EGFR testing. The IASLC Atlas of EGFR Testing in Lung Cancer is a useful guidebook for this purpose. We aim to provide the pathologists with know-how information on EGFR testing, and to provide the clinicians with the know-why information on how to interpret the results. From the retrieval and handling of tumor sample to the different available assays, and from interpretation of results to reporting and quality assurance, the Atlas is a comprehensive yet user friendly compendium for the general oncology readership. We have also summarized the relevant clinical data supporting the application of EGFR testing in patients with either treatment-naïve or EGFR TKI-resistant disease. It is the vision of IASLC to disseminate these

14 12 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER important educational messages around the world, and I hope you will share our vision to improve the quality of lung cancer care, and thus improve and prolong the lives of our lung cancer patients.

15 Therapeutic Perspectives 1 By Tony S. Mok, Solange Peters, Daniel SW Tan, Luiz Paz-Ares, and Keunchil Park Non-small cell lung cancer (NSCLC) is the leading cause of death around the globe, resulting in more than 1 million deaths worldwide annually (GLOBOCAN 2015). The 5-year survival rate for patients with stage IV disease is less than 5% (Siegel 2017). The identification of specific somatic aberrations has led to a personalized therapeutic approach, resulting in substantially improved outcomes in NSCLC. In Europe and North America, 10% to 20% of patients with NSCLC possess an activating EGFR gene mutation (Moran 2009). Incidence of this driver alteration is more prevalent in East Asia, affecting 50% to 60% of patients with NSCLC. Regardless of geographic region, prevalence is higher among never-smokers, women, and the adenocarcinoma subtype (Shi 2014). EGFR Mutation Incidence, Associated Treatment Outcomes The EGFR gene is located on the short arm of chromosome 7 (7p11.2) and encodes a 170-kDa type I transmembrane growth factor receptor with tyrosine kinase activity. EGFR belongs to the HER/ErbB family of receptor tyrosine kinases. Intracellular signaling is mediated mainly through the RAS/RAF/MEK/MAPK pathway, the PI3K/PTEN/AKT pathway, and the STAT pathway. Downstream EGFR signaling ultimately leads to increased proliferation, angiogenesis, metastasis, and decreased apoptosis. Gain-of-function or activating mutations of EGFR were first identified in 2004, leading to constitutive tyrosine kinase activity (Lynch 2004). Given the magnitude of the benefit of targeted therapy associated with this disease, management of EGFR-mutated NSCLC is a major issue worldwide, particularly in regions with high prevalence. An EGFR mutation status should be systematically analyzed with sequencing as the criterion standard for patients who have advanced NSCLC with nonsquamous histology (Novello 2016). Activating (and sensitizing) EGFR mutations are predictive for response to EGFR tyrosine kinase inhibitors (TKIs), including first-generation gefitinib and erlotinib, second-generation afatinib and dacomitinib, and the third-generation drug osimertinib. In patients with advanced NSCLC, both first- and second-generation EGFR TKIs are standard first-line treatment options. Such treatments result in improved response and

16 14 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER progression-free survival rates, better tolerability, and superior quality of life compared with platinum-based chemotherapy in the first-line setting, as demonstrated in several randomized trials (Mok 2009; Maemondo 2010; Mitsudomi 2010; Rosell 2012; Sequist 2013; Lee 2014; Wu 2014; Wu 2015; Zhou 2015). Exon 19 LREA frame-deletion mutations and the single-point substitution mutation L858R in exon 21 are the most frequent in NSCLC and are considered to be so-called classic mutations, accounting for 85% to 90% of all EGFR mutations (Mok 2009; da Cunha Santos 2011). Most of the large randomized trials comparing EGFR TKIs with chemotherapy enrolled patients with classic mutations; only four trials included patients with uncommon EGFR mutations. Uncommon mutations comprise approximately 10% of EGFR mutations and are defined as all mutations excluding exon 19 deletion and L858R. The objective response rate to TKIs for patients with classic mutations is approximately 60%. Between 50% and 60% of classic EGFR mutations are exon 19 LREA frame deletions (Lee 2015). Patients with this specific activating mutation have consistently had better outcomes with EGFR TKIs than patients with L858R single-point substitutions (Jackman 2006; Jackman 2009; Riely 2006; Lee 2015). A recent meta-analysis evaluating seven trials (1,649 patients) found that EGFR TKIs significantly prolonged progression-free survival (HR: 0.37; 95% CI: ) in all subgroups compared with chemotherapy. TKI treatment benefit was 50% greater (HR: 0.24; 95% CI: ) for patients with exon 19 deletions than for those with exon 21 L858R substitution (HR: 0.48; 95% CI: ; P = 0.001). Patients with exon 19 deletions had a greater overall survival than those with exon 21 L858R substitution after gefitinib or elotinib therapy. Interestingly, exon 21 L858R substitutions, rather than exon 19 deletions, have been associated with longer overall survival for TKI treatment-naive patients (Shigematsu 2005). In the meta-analysis by Lee et al., chemotherapy resulted in significantly greater progression-free survival for patients with exon 21 L858R substitutions than for patients harboring exon 19 deletions (median progression-free survival, 6.1 vs. 5.1 months; P = 0.003). Within this context, the findings of a pooled analysis focusing on the two randomized trials comparing afatinib and chemotherapy LUX-Lung 3 and LUX Lung-6 trials suggested this TKI could improve overall survival compared with chemotherapy among patients with exon 19 deletions but not for patients with L858R substituted disease (Yang 2015). The cause of this difference in response to EGFR TKIs, or possibly chemotherapy, by EGFR mutation subtype is not known. However, in the randomized trial in which afatinib was compared with gefitinib (LUX-Lung 7), the overall survival did not differ significantly in the two treatment arms among patients with EGFR exon 19 deletion (Paz-Ares 2017). EGFR TKIs Despite the first encouraging report in 1948 by Karnofsky et al. that advanced lung cancer could respond to cytotoxic chemotherapy, progress in controlling advanced or metastatic lung cancer has been slower than expected (Kennedy 1998). Until the early 2000s, the standard of care for patients with advanced NSCLC was a third-generation platinum-based chemotherapy doublet, irrespective of histopathology, for patients with good performance statuses and best supportive care for patients with poor performance statuses. At the dawn of the new millennium, the thoracic oncology community experienced a temporary

17 THERAPEUTIC PERSPECTIVES 15 disappointment following an observation that survival outcomes of modern chemotherapy regimens had reached a therapeutic plateau (Schiller 2002). Fortunately, that was merely the beginning of a completely new treatment paradigm. The initial studies (IDEAL-1 and -2) of gefitinib, the first EGFR TKI, evaluated the agent as second- and third-line treatment for pretreated patients with advanced NSCLC. A oncedaily dose of 250 or 500 mg showed meaningful antitumor activity (overall response rate, 9% to 19%; overall survival, 6 to 8 months) and provided symptom relief with mild toxicities (Fukuoka 2003; Kris 2003). In two large randomized phase III trials (INTACT-1 and -2), chemotherapy-naive patients with unresectable stage III and IV NSCLC were randomly assigned to receive gefitinib (250 or 500 mg daily) or placebo in combination with platinumdoublet chemotherapy. There was no added benefit for survival, time to progression, or response rate compared with standard chemotherapy alone (Giaccone 2004; Herbst 2004). Based on the early results of these trials, gefitinib received accelerated approval in 2003 by the US FDA as monotherapy treatment for patients with locally advanced or metastatic NSCLC after failure of both platinum- and docetaxel-based chemotherapies. Erlotinib, another reversible EGFR TKI, was approved by the FDA in 2004 as a singleagent treatment for patients with locally advanced or metastatic NSCLC who had disease progression after other treatments, including at least one prior chemotherapy regimen. Approval was based on the early results of a randomized placebo-controlled double-blind trial that demonstrated improved overall survival for patients treated with erlotinib (6.7 months) compared with placebo (4.7 months; HR: 0.70; p < 0.001) (Shepherd 2005). Response Prediction of EGFR TKIs In the IDEAL-1 study, the response rate was greater for Japanese patients than for non- Japanese patients (27.5% vs. 10.4%; odds ratio = 3.27; P = ), which suggested that potential ethnic and/or genetic factors may be involved in gefitinib response (Fukuoka 2003). Investigators soon discovered that a mutation in the EGFR kinase domain is strongly associated with response to gefitinib (Lynch 2004; Paez 2004), as discussed previously. The introduction of EGFR TKIs in the management of advanced EGFR-mutated NSCLC has opened the door to precision medicine and paved the way for studies of other solid tumors, such as ALK-rearranged and ROS-1-rearranged NSCLC. Several randomized phase III trials comparing EGFR TKIs with the standard platinumbased chemotherapy doublet in chemotherapy-naive patients with EGFR-mutated NSCLC have been conducted following encouraging results from trials on EGFR TKIs for patients with advanced NSCLC whose disease did not respond cytotoxic chemotherapies. EGFR TKIs demonstrated superior outcomes compared with standard chemotherapy in terms of objective response rate (60% to 80% vs. 25% to 35%, respectively) and progression-free survival (10 to 14 months vs. 5 to 6 months, respectively) with better tolerability (Haaland 2014). These results formed the basis of approval of first-generation EGFR TKIs, such as gefitinib and erlotinib, and second-generation agents, such as afatinib and this drug class is the current standard of care for the first-line management of EGFR-mutated NSCLC. One strategy to further improve first-line management of EGFR-mutated NSCLC includes the addition of an antiangiogenic agent to an EGFR TKI. A phase II Japanese trial demonstrated superior progression-free survival with erlotinib and bevacizumab (16.0 months)

18 16 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER compared with erlotinib alone (9.7 months) (Seto 2014). Confirmatory trials are ongoing in Japan. As was previously described, afatinib was approved for first-line treatment of advanced EGFR-mutated NSCLC, and a prospective head-to-head randomized phase IIb comparative study in the first-line setting demonstrated better progression-free survival, time to treatment failure, and overall response rate compared with gefitinib, which suggests that a second-generation EGFR TKI is not interchangeable with a first-generation agent (Park 2016). A prospective randomized phase III trial of dacomitinib also demonstrated superior progression-free survival compared with the first-generation EGFR TKIs (median 14.7 vs 9.2 months) (Mok 2017). EGFR TKI Resistance Almost all patients who had a response to first- or second-generation EGFR TKIs eventually had resistance to these drugs. The most common mechanism of resistance, accounting for approximately 50% of all cases, is the acquired T790M mutation in exon 20 (Kobayashi 2005; Yun 2008; Yu 2013). Upon occurrence of T790M mutation at exon 20 of the EGFR gene, the quinazoline-based first-generation EGFR TKI can no longer bind to the ATP binding pocket at the receptor, thus losing efficacy on inhibition of downstream signalling. The biology and epidemiology of the T790M resistance mutation has been well studied. The incidence of this mutation is similar among ethnicities and among different EGFR TKIs. Preclinical studies have shown that the more potent EGFR inhibition with second-generation EGFR TKIs, such as afatinib, may overcome the T790M mutation-associated acquired resistance, but severe toxicities associated with inhibition of normal EGFR receptor at this dose level would be prohibitive for clinical use. Afatinib was given to patients with EGFR-mutated NSCLC that was nonresponsive to prior treatment with gefitinib and/or erlotinib in the LUX-Lung 1 and LUX-Lung 4 trials (Miller 2012, Katakami 2013). The overall response rate was 7% to 8%, with a median progression-free survival of 3 to 4 months, suggesting that second-generation EGFR TKIs cannot overcome T790M-associated acquired resistance. Other mechanisms of resistance are relatively heterogeneous, and these may include HER2 and/or MET amplification, PIK3CA and/or BRAF mutation, and small cell lung cancer transformation (Stewart 2015). The incidence of each of these mechanisms of resistance is much lower than that of the T790M mutation, and only rarely do these mutations or amplifications occur concurrently with T790M mutation. Third-generation EGFR TKIs were designed to target the T790M mutation while sparing wild-type EGFR, as first reported by Zhou et al (Zhou 2009). These authors studied three closely related pyrimidines namely WZ3146, WZ4002, and WZ8040 and found WZ4002 to be most potent against EGFR T790M. In an in vivo study of WZ4002, tumor regressions were significant compared with vehicle alone in both EGFR L858R/T790M and deletion E746_A750/T790M-containing murine models. However, the compound was never developed into a commercial drug. The first and only commercially available third-generation EGFR TKI is osimertinib. The FDA granted regular approval on March 30, 2017, to osimertinib for the treatment of patients with metastatic EGFR T790M mutation-positive NSCLC, as detected by an FDA-approved test, and disease progression during or after EGFR TKI therapy. Molecular structure is

19 THERAPEUTIC PERSPECTIVES 17 distinct from WZ4002 but the background concept of a pyrimidine-based structure and the idea of the electrophilic functionality residing in different regions of the chemical structure were similar (Cross 2014). The authors of a preclinical study reported the IC 50 of osimertinib against cell line with L858R/T790M at 1 nm against L858R/T790M, which is approximately 200 times greater potency than wild-type EGFR. Jänne et al. reported results from the first phase I/II multicenter study (253 patients). Five dose cohorts (20, 40, 80, 160, and 240 mg daily), were studied in patients with EGFR mutation-positive disease who had prior benefit from first- or second-generation EGFR TKIs but subsequently experienced disease progression. Both patients with or without T790M resistance were included. Focusing on the T790M-positive subgroup, the overall response rate was 61% and progression-free survival was 9.6 months (Jänne 2015). The finding was supported by a single-arm phase II study in which a daily dose of 80 mg of osimertinib was evaluated in 210 patients with T790M-positive disease (Yang 2017). The response rate was 64% (95% CI: 57%-71%) and the median progression-free survival was 8.6 months. Toxicities were well tolerated and included diarrhea (47%), rash (40%), nausea (22%), and decreased appetite (21%). Grade 3 or higher toxicities were uncommon. AURA 3 is the first randomized phase III study to compare osimertinib with standard platinum-based doublet chemotherapy in patients with T790M-positive lung cancer (Mok 2017). To be eligible, patients must have had treatment with an EGFR TKI that was unsuccessful according to RECIST criteria and proven with the presence of T790M mutation confirmed by repeat biopsy. Tumor response rates were 71% and 31% for the osimertinib and chemotherapy groups, respectively. Progression-free survival was significantly longer for the osimertinib group (median, 10.1 vs. 4.4 months; HR: 0.30; 95% CI: , P = ) The overall toxicity profile of osimertinib is also more favorable than chemotherapy. The study has defined osimertinib to be the standard of care for T790M-positive acquired resistance; therefore, patients who experience treatment failure of a first-line EGFR TKI should be tested for the presence of a T790M mutation. However, management of T790Mnegative resistance continues to be controversial. Chemotherapy is the current standard, and ongoing research is investigating the role of immunotherapy in this population. Although third-generation EGFR TKIs have demonstrated impressive efficacy, development of acquired resistance is inevitable. Little is known about the mechanisms of acquired resistance to these agents, and active investigation is ongoing.

20

21 EGFR Testing 2 By Ming Sound Tsao, Ignacio I. Wistuba, and Yasushi Yatabe In 2013, the College of American Pathologists (CAP), International Association for the Study of Lung Cancer (IASLC), and the Association for Molecular Pathology (AMP) published the molecular testing guideline in lung cancer, which included information about testing of EGFR gene mutations (Lindeman 2013). The guideline, which was subsequently endorsed by ASCO (Leighl 2014), has served as an international standard for EGFR mutation testing. The guideline update, authored by experts from each of the three organizations, is expected to be published by the end of For this update, a systematic review was conducted of all applicable data since the initial publication. The update includes recommendations on a range of additional molecular tests in lung cancer that are used either routinely or to identify patients who might benefit from novel therapies in clinical trials. Several updated recommendations are related to EGFR testing (Box 1). The recommendations offered in the 2013 guideline largely have been reaffirmed, except for the following points (Lindeman 2017). Any cytology sample with adequate cellularity and preservation may be tested. The development of sensitive molecular testing techniques allows cytology specimens to be used in clinical practice, and the findings of the systematic review demonstrated excellent performance of smear preparations, if the tumor cells were cytologically confirmed in samples (Roy-Chowdhuri 2016, Rekhtman 2016, Malapelle 2017a). In patients with advanced lung cancer, which represents two-thirds of the lung cancer population, cytology specimens often are the only samples available for molecular testing to guide treatment decisions. Analytic methods must be able to detect mutations in a sample with 20% or more malignant cell content. In the 2013 guideline, Sanger sequencing was still one of the recommended testing methods, but evidence for the clinical utility of more sensitive PCR-based techniques has been established in subsequent studies. Along with the widespread availability of these sensitive techniques, Sanger sequencing is not considered suitable for clinical use, particularly for biopsy specimen testing.

22 20 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER IHC is not appropriate for EGFR mutation testing. The 2013 guideline indicated that EGFR IHC testing does not have a role in clinical practice, but IHC tests with mutation-specific EGFR antibodies are allowed when the sample is extremely limited. The findings of subsequent studies have shown that these antibodies have poor sensitivity for many exon 19 deletions, insensitivity for less-common mutations (eg, codon 719 mutations), and false-positive results with exon 20 insertions (Kitamura 2010). With the availability of very sensitive detection techniques and circulating cfdna, EGFR mutation-specific IHC was considered to have no role in routine clinical practice. In its endorsement of the CAP/IASCL/AMP guideline, ASCO suggested routine molecular testing of early-stage disease, similar to the testing paradigm for patients with breast cancer (Leighl 2014). The updated guideline also addresses this point, encouraging such testing but at the discretion of the individual institution and local oncology team. The updated guideline is more focused than the original version on testing of other genes, including ROS1, BRAF, RET, HER2, RAS, and MET. Furthermore, testing for secondary mutations that cause therapeutic resistance, multigene panel testing, and liquid biopsy application also are described. The updated guideline also includes the following recommendations related to EGFR testing. T790M. The results of the AURA trials have demonstrated improved outcome using osimertinib after the development of resistance to first- and second-generation EGFR tyrosine kinase inhibitors (TKIs), and the findings of a systematic review support the increased therapeutic benefit (Janne 2015, Goss 2016, Mok 2017, Yang 2017, Lindemann 2017). (See Chapter 1.) The updated guideline recommends EGFR T790M testing to select patient candidates for treatment with a third-generation EGFR TKI. The results of some studies have shown that a small proportion of T790-mutated tumor cells that existed in a bulk tumor-cell population caused the resistance (Inukai 2006, Engleman 2006, Chmielecki 2011, Su 2012, Chen 2016). In clinical samples, the number of T790Mmutated tumor cells is almost always less than the number of cells that bear only the sensitizing EGFR mutations (Engelman 2006, Chmielecki 2011). Therefore, the sensitivity to detect the EGFR T790M resistance mutation was set to detect as low as 5% of mutant allele. A commercial real-time PCR allele-specific assay, droplet digital PCR, and nextgeneration sequencing are currently the preferred methods to meet the requirement. Liquid biopsy. There is much enthusiasm regarding the development of noninvasive methods for EGFR mutation testing. For example, so-called liquid biopsy, in which blood is used as the test sample, has been evaluated in several studies (Oxnard 2014, Qiu 2015, Mok 2015). (See Chapter 3.) A systematic review of this new diagnostic modality demonstrated insufficient evidence for use as a molecular test that involves circulating tumor cells. However, testing of cfdna provided useful information as an alternative molecular test when tissue is limited and/or insufficient for regular tissue-based molecular testing. Because of the high specificity (95.6%; 95% CI: 83.3% 99.0%) but only intermediate sensitivity (66.4%; 95% CI: %) of cfdna testing, a negative result does not represent tumor negativity for EGFR mutation (Lindemann 2017).

23 EGFR TESTING 21 Box 1. Recommendations for EGFR Testing in the Updated College of American Pathologists (CAP), International Association for the Study of Lung Cancer (IASLC), and the Association for Molecular Pathology (AMP) Molecular Testing Guideline Patient Selection for EGFR Mutation Testing 1.1a: Recommendation: EGFR molecular testing should be used to select patients for EGFR- targeted tyrosine kinase inhibitor (TKI) therapy; patients with lung adenocarcinoma should not be excluded from testing based on clinical characteristics. 1.2: Recommendation: In the setting of lung cancer resection specimens, EGFR testing is recommended for adenocarcinomas and mixed lung cancers with an adenocarcinoma component, regardless of histologic grade. In the setting of full excised lung cancer specimens, EGFR testing is not recommended in lung cancers that lack any adenocarcinoma component, such as pure squamous cell carcinomas and pure small cell carcinomas. 1.3: Recommendation: In the setting of more limited lung cancer specimens (i.e., biopsies or cytology) where an adenocarcinoma component cannot be completely excluded, EGFR testing may be performed in cases showing squamous or small cell histology but clinical criteria (e.g., young age and lack of smoking history) may be useful in selecting a subset of these samples for testing. 1.4: Recommendation: To determine EGFR status for initial treatment selection, primary tumors or metastatic lesions are equally suitable for testing. 1.5: Expert consensus opinion: In patients with multiple, apparently separate, primary lung adenocarcinomas, each tumor may be tested, but testing of multiple different areas within a single tumor is not necessary. Specimen Testing for EGFR Mutation 2.1a: Recommendation: EGFR mutation testing should be ordered at the time of diagnosis for patients presenting with advanced-stage disease stage IV according to the 7th edition Tumor Node Metastasis (TNM) staging system that meets the therapeutic requirements, or at time of recurrence or progression in patients who originally presented with lower-stage disease but were not previously tested. 2.2a: Expert consensus opinion: EGFR testing of tumors at diagnosis from patients presenting with stage I, II, or III disease is encouraged, but the decision to do so should be made locally by each laboratory in collaboration with the oncology team. 2.3: Recommendation: Tissue should be prioritized for EGFR, ALK, and ROS1 testing. Turnaround Times for Test Results 3.1: Expert consensus opinion: EGFR results should be available within 2 weeks (10 working days) of receiving the specimen in the testing laboratory. 3.2: Expert consensus opinion: Laboratories with average turnaround times beyond 2 weeks must make a more rapid test available either in-house or through a reference laboratory in instances of clinical urgency. 3.3: Expert consensus opinion: Laboratory departments should establish processes to ensure that specimens that have final histopathologic diagnosis are sent to outside molecular pathology laboratories within 3 working days of receiving requests and to intramural molecular pathology laboratories within 24 hours. Specimen Processing 4.1: Expert consensus opinion: Pathologists should use formalin-fixed, paraffin-embedded specimens or fresh, frozen, or alcohol-fixed specimens for polymerase chain reaction (PCR)-based EGFR mutation tests. Other tissue treatments (e.g., acidic or heavy-metal fixatives, or decalcifying solutions) should be avoided in specimens destined for EGFR testing. 4.2: Expert consensus opinion: Cytologic samples are also suitable for EGFR testing, with cell block, cell pellets and smears. continued on next page

24 22 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER Specimen Requirements 5.1: Expert consensus opinion: Pathologists should determine the adequacy of specimens for EGFR testing by assessing cancer cell content and DNA quantity and quality. 5.2: Expert consensus opinion: Each laboratory should establish the minimum proportion and number of cancer cells needed for mutation detection during validation. 5.3: Expert consensus opinion: A pathologist should assess the tumor content of each specimen and either perform or guide a trained technologist to perform microdissection for tumor cell enrichment, when needed. Performing EGFR Testing 6.1: Recommendation: Laboratories may use any validated EGFR testing method with sufficient performance characteristics. 6.2: Expert consensus opinion: Laboratories should use EGFR test methods that are able to detect mutations in specimens with at least 20 % cancer cell content. 6.3: Expert consensus opinion: Clinical EGFR mutation testing should be able to detect all individual mutations that have been reported with a frequency of at least 1% of EGFR-mutated lung adenocarcinomas. 6.4: Recommendation: Immunohistochemistry (IHC) for total EGFR IHC and EGFR-mutation specific IHC is not recommended for selection of EGFR TKI therapy. 6.5: Recommendation: EGFR copy number analysis (i.e., fluorescence in situ hybridization [FISH] or chromogenic in situ hybridization [CISH]) is not recommended for selection of EGFR TKI therapy. New Recommendations Key Question IV: What testing is indicated for patients with targetable mutations who have relapsed on targeted therapy? IVA. EGFR T790M Strong Recommendation: In patients with lung adenocarcinoma who harbor sensitizing EGFR mutations and whose disease has progressed after treatment with an EGFR-targeted TKI, physicians must use EGFR T790M mutational testing when selecting patients for third-generation EGFR-targeted therapy. Recommendation: Laboratories testing for EGFR T790M mutation in patients with secondary clinical resistance to EGFR-targeted TKIs should deploy assays capable of detecting EGFR T790M mutations in as few as 5% of mutant alleles. Key Question V: What is the role of testing for circulating, cell-free DNA (cfdna), for patients with lung cancer Va. No Recommendation: There is currently insufficient evidence to support the use of cfdna molecular methods for the diagnosis of primary lung adenocarcinoma. Vb. Recommendation: In some clinical settings in which tissue is limited and/or insufficient for molecular testing, physicians may use a cfdna assay to identify EGFR mutations. Vc. Expert Consensus Opinion: Physicians may use cfdna methods to identify EGFR T790M mutations in patients with lung adenocarcinoma with disease progression or secondary clinical resistance to EGFR-targeted tyrosine kinase inhibitors; testing of the tumor sample is recommended if the plasma result is negative. Vd. No Recommendation: There is currently insufficient evidence to support the use of circulating tumor cell (CTC) molecular analysis for the diagnosis of primary lung adenocarcinoma, the identification of EGFR or other mutations, or the identification of EGFR T790M mutations at the time of EGFR TKI resistance. Patient Candidates for Testing In early studies, the incidence of EGFR kinase domain mutations was more prevalent in the East Asian population, never-smokers, women, and the adenocarcinoma subtype, as summarized in Tables 6 and 9 in the 2013 guideline (Lindeman 2013). The guideline includes recommendations for patient selection for EGFR testing (Box 1).

25 EGFR TESTING 23 Although the number of studies on squamous cell carcinoma testing is limited, the updated CAP/IASLC/AMP guideline addresses the molecular tests for lung cancers that do not have an adenocarcinoma component. The guideline states that molecular testing is recommended for tumors with a histology other than adenocarcinoma when clinical features indicate a higher probability of the presence of an oncogenic driver. Major clinical features include young age and the absence of tobacco exposure. Ample data demonstrate association of EGFR, ALK, and ROS1 alterations with no or minimal tobacco exposure. Although it is generally recognized that EGFR mutations rarely occur in patients with squamous cell carcinoma, the authors of some studies (especially from East Asian countries) have reported the presence of EGFR mutations in 3% to 13% of patients with this type of lung cancer. However, the prevalence of nonsmokers is higher among these patients than among patients without EGFR mutations (Hata 2013, Dearden 2013). The findings suggest that light or absent tobacco exposure is a sufficient rationale to prompt testing, regardless of sample types or nonadenocarcinoma morphology. However, the response rates and survival benefit associated with EGFR TKI therapies among these patients appear to be lower than among patients with adenocarcinoma (Hata 2013, Xu J 2016, Liu 2017, Joshi 2017). In terms of patient age, the findings of the systematic review for the 2013 CAP/IASLC/ AMP guideline showed that the mean age of patients with EGFR-mutated tumors was significantly less than that for patients without EGFR-mutated tumors (Lindeman 2013). However, prevalence of younger age is more frequently reported for patients with ALK and ROS1 alterations. The updated guideline recommends testing for all patients younger than 50 years who have nonadenocarcinoma histology. It is important to emphasize that in the current practice of lung cancer pathology, routine performance of histologic markers (TTF-1/mucin and P40/P63) on poorly differentiated NSCLC specimens is mandatory, both in the context of more accurate histologic classification of the tumors and especially in the context of the CAP/IASLC/AMP guideline (Loo 2010, Mukhopadhyay 2011, Travis 2011, Warth 2012, Rekhtman 2012, Rekhtman 2013). Timing of EGFR Mutation Testing Primary Diagnosis Knowledge of EGFR mutation status is crucial to therapeutic decision-making for patients with advanced nonsquamous cell NSCLC given that EGFR TKIs are used as first-line treatment of patients with EGFR-sensitizing mutations. Such knowledge is also essential when treating patients with recurrence after surgical resection of earlier-stage disease. This principle was articulated in the 2013 CAP/IASLC/AMP guideline and is reaffirmed in the update. Because lung cancer is at an advanced stage at the time of diagnosis in two-thirds of patients, only small biopsy or cytology specimens are available for testing in most patients. Except for bone specimens that have been fixed in acidic solutions (eg, Bouin or bone-decalcifying solutions) or heavy-metal fixatives (eg, Zenker s, B5, or B-Plus fixative, or acid zinc formalin), all types of small biopsy or cytology specimens are suitable for mutation testing; this has been confirmed in reports from various laboratories around the world (Table 1) (Shiau 2014, Shi 2014, Vigliar 2015, Yatabe 2015). The main issue in the timing of EGFR testing is whether testing should be ordered by the treating physician (known as bespoke testing) or by the pathologists responsible

26 24 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER for the primary diagnosis (reflex testing). Bespoke testing remains the customary practice in many parts of the world because it may avoid the cost of testing in patients with early-stage disease. In contrast, bespoke testing may result in a substantial delay in the initiation of first-line treatment for patients with advanced disease, potentially negatively affecting the outcome. Lim et al compared bespoke and reflex EGFR testing and found the median time from consultation to treatment decision was significantly shorter (0 vs. 22 days, P = ) for reflex testing, as was the time to treatment start (16 vs. 29 days, P = 0.004) (Lim 2015). Similarly, Cheema Table 1. Published Data on Sample Types Used in EGFR Testing Around the World Sample Available for Testing (%) Study Source of Data Surgical Biopsy Cytology Yatabe 2015 China Hong Kong Indonesia Japan South Korea Philippines Singapore Taiwan Thailand Vietnam (annotated as biopsy) Shiau 2014 Ontario, Canada Sholl 2015 United States, multiinstitutional trial a Shi 2014 Multiple countries in Asia b (PIONEER trial) 88.6 (annotated as biopsy) 11.4 Hlinkova 2013 Eastern Europe, single institution a Lung Cancer Mutation Consortium. b PIONEER trial et al. compared 74 patients who had reflex EGFR testing with 232 patients who had routine bespoke testing (Cheema 2016). Reflex testing was associated with a shorter median time from first medical oncology visit to treatment (26 vs. 36 days, P = 0.071) and time to optimal first-line systemic therapy (24 vs. 36 days, P = 0.036), as well as greater quality of biomarker testing, as indicated by the percentage of unsuccessful tests (4% vs. 14%, P = 0.039). A substantial pitfall in reflex testing is the common lack of knowledge of disease stage at the time of initial biopsy examination by the pathologist. It is expected that 30% to 40% of patients have early-stage disease and may be treated by surgical resection. Awareness of EGFR mutation status is not crucial for primary treatment because there is no adjuvant role for EGFR TKIs; therefore, EGFR testing is not required. The 2013 CAP/IASLC/AMP guideline stated that EGFR testing of tumors at diagnosis from patients presenting with stage I, II, or III disease is encouraged but the decision to do so should be made locally by each laboratory, in collaboration with its oncology team. (Lindeman 2013). Analogous to the use of adenocarcinoma histology to limit EGFR testing, whether testing should be done in patients with early-stage disease is largely an economic issue because these patients with EGFR mutations may be cured by surgery alone and do not require EGFR TKI therapy. However, metastatic recurrence is expected to develop in 40% to 60% of patients with early-stage disease, which requires treatment similar to that for advanced disease. Overall,

27 EGFR TESTING 25 15% of patients with nonsquamous NSCLC may not need testing at all. Reflex testing on tumors of all stages may streamline pathology workload, mitigate turnaround time for testing, minimize testing failure rates when testing small samples, and fast-track repeat biopsy when it is required. Disease Progression and Monitoring EGFR TKIs such as gefitinib, erlotinib, and afatinib are the standard of treatment for patients with EGFR-mutated tumors because of the high response rates (55% to 78%) and substantially longer progression-free survival rates for these patients (Mok 2011). However, resistance and relapse develop within a short time in most patients, caused by the T790M mutation in exon 20 of the EGFR kinase domain (approximately 50% of cases), amplification of the MET oncogene, or mutations of the PI3KCA gene. Other changes, including SCLC transformation, also have been also described in approximately 4% of cases. (PMID: ) Osimertinib is the first FDA-approved TKI for patients with EGFR T790M-mutated NSCLC who had disease progression during treatment with an EGFR TKI (Goss 2016, Mok 2017). These data highlight the importance of repeat biopsy to identify the underlying mechanisms of resistance to EGFR TKIs, particularly the actionable EGFR T790M mutation, MET amplification, and SCLC transformation. A new EGFR acquired-resistance mutation (C797S) has been noted during treatment with osimertinib (Yu HA 2015); no specific effective treatment, however, has been developed for this new mutation. The isolation of cfdna in blood samples from patients with lung cancer has provided an alternative source of tumor material that can be used as a biopsy surrogate for EGFR T790M mutation testing (Bordi 2015, Oxnard 2016, Sacher 2016). cfdna genotyping testing approaches allow serial assessments of repeat blood samples from patients with lung cancer, which may be especially important when monitoring for acquisition of second-site EGFR mutation in patients with NSCLC who have disease progression during treatment with an EGFR TKI (Oxnard 2016). Eligibility for osimertinib therapy is determined by the presence of an EGFR T790M mutation. Currently, the companion diagnostic for this drug is approved by the FDA for both tissue biopsy and plasma cfdna testing. Therefore, liquid biopsy provides an attractive alternative to tissue biopsy and could help widen the selection of eligible patients with the acquired T790M mutation. The results of a retrospective analysis showed that, for patients with tumors that were positive for T790M mutations on plasma testing, outcomes with osimertinib were equivalent to those for patients who had positive results on a tissue-based assay (Oxnard 2016). These findings suggest that, with the availability of validated plasma T790M assays, a tumor biopsy for T790M genotyping may be avoided in some patients. However, because of the 30% rate of false-negative results for plasma genotyping reported in the study, a tumor biopsy is required to determine the presence of a T790M mutation when the results of a plasma assay are negative. Conclusion The 2013 CAP/IASLC/AMP guideline has served as an international guideline for molecular testing in lung cancer. The anticipated update of this guideline generally reaffirms the recommendations made in the original guideline, with a few additional recommendations

28 26 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER based on the findings of systemic reviews conducted since the initial publication. These new recommendations address such topics as testing of cytology samples, testing methods, testing by IHC, testing for T790M acquired-resistance mutations, and testing of cfdna in blood. Although EGFR testing is recommended mainly for patients with adenocarcinoma or for patients in whom adenocarcinoma cannot be ruled out, testing is also recommended for tumors with nonadenocarcinoma histologies when clinical features indicate a higher probability for the presence of an oncogenic driver.

29 Sample Acquisition, Processing, and 3 General Diagnostic Procedures By Philip Mack, Rex Yung, Walter Weder, and Dara L. Aisner The standard principles for lung cancer diagnosis apply when acquiring sufficient tissue for definitive diagnosis and characterization of lung cancer at the highest stage and with minimal risk to the patient. Adequate characterization is defined as the ability to distinguish cell type by IHC because the diagnosis of NSCLC-not otherwise specified is unsatisfactory in the era of cell type-specific cytotoxic therapy, limiting direction to perform follow-up testing for targeted therapy-associated actionable biomarkers. Surgical biopsies, including excision biopsy of cutaneously accessible metastatic lymph nodes, mediastinoscopy, or video-assisted thoracoscopic surgery (VATS) will provide the greatest amount of tissue for biomarker analysis. The majority of lung cancer cases may continue to be diagnosed at later stages (ie, multistation IIB, IIIA/B, and IV) until advances in early detection occur and screening is routinely established; less-invasive procedures that combine diagnosis and staging are preferable. The only noninvasive approach to tissue acquisition and examination involves spontaneously expectorated or induced sputum. The general diagnostic yield from sputum is in the range of 5% to 20%, depending on the location of tumor, technician familiarity with processing techniques, and cytologic diagnosis of lung cancer from sputum. Studies on EGFR and other biomarker testing in sputum are limited. The only controlled paired study comparing sputum testing with EGFR-proved primary tumor demonstrated sensitivity of 30% to 50% (Hubers 2013). As with the other minimally invasive techniques discussed further in this chapter, harvesting of sufficient tissue amounts for multiple biomarker analyses is an even greater challenge with sputum. Pleural effusion with positive results of cytologic testing will provide a stage IVa diagnosis; however, sensitivity of pleural cytology for pleural metastases is between 40% and 70%. In addition, previous reports of EGFR testing on pleural samples showed that results were dependent on the number of viable cells recovered in a cell block (Kimura 2006). Testing by next-generation sequencing (Buttitta 2013) or cfdna (Park 2017), which is explained in greater detail in this chapter, may be more promising. Pleural extension and metastases is most effectively diagnosed by VATS or medical thoracoscopy, the latter of which allows

30 28 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER for simultaneous treatment of surgical and chemical pleurodesis by insertion of a long-term drainage catheter, which also improves lung re-expansion. Bronchoalveolar lavage is the least invasive of the bronchoscopy approaches and requires the least specialized training to perform. This technique, however, also provides the lowest diagnostic yield for lung cancer (Xin 2016), although incorporation of next-generation sequencing and cfdna testing can also improve detection rates for EGFR mutations (Buttitta 2012; Park 2017). Nevertheless, a bronchoscopy performed specifically for cancer diagnosis should be planned with procedures that maximize tissue acquisition for predictive biomarker testing. Comprehensive sampling of multiple accessible targets should be considered (provided it does not compromise patient safety) because it is often unclear whether enlarged lymph nodes are malignant or only reactive (Zhang 2012; Hyo 2016). Aside from transbronchial brushings, endobronchial and transbronchial forceps biopsies, and cryobiopsies, the most versatile biopsy technique may be transbronchial needle aspiration (TBNA) because it can be used to sample lymph nodes, as well as endobronchial lesions and parenchymal masses. Although conventional TBNA-acquired tissue can be tested for EGFR mutations (Horiike 2007), the advent of the endobronchial ultrasound (EBUS)-adapted bronchoscope (Nakajima 2007) is a definite improvement because this tool provides real-time image guidance, allowing for accurate sampling of lymph nodes (Navani 2012). However, mediastinoscopy remains the criterion standard when these less-invasive methods fail to provide an accurate diagnosis. There are different methods of processing tissue collected using EBUS-TBNA (outside of typical sampling of lymph nodes and other masses) to enhance diagnostic yield for cancercell typing and for molecular testing (Yung 2012; van der Heijden 2014). These methods should be standardized for a given institution. Given the improvements in biomarker testing regarding use of smaller quantities of tissue, the choice for sample acquisition should be customized for the local institution according to the available experience and expertise of the radiologists, interventional pulmonologists, gastroenterologists, and surgeons. Ideally, cases are discussed at a multidisciplinary tumor board, and decisions regarding biomarker testing should be predetermined for the patient if biomarker reflex testing is not a part of the pathologist s standard operating procedure. Quality, Amount, and Location of a Tissue Biopsy Numerous factors regarding the quantity and quality of tumor tissue for molecular analysis should be considered. Key among these factors is the quantity of tumor available, either as formalin-fixed paraffin embedded (FFPE) tissue or, in some cases, preserved in the format of cytopathology materials (eg, smears and touch preps). As most assay approaches to evaluating EGFR mutations involve PCR-based strategies, testing can be performed on relatively small quantities of tissue, even on exceptionally scant samples. Standard minimally invasive approaches that are often performed in an interventional radiology setting, such as transbronchial biopsy, EBUS-fine-needle aspiration, and core biopsies, may yield sufficient tissue for molecular analysis. However, balancing the needs for adequate primary diagnosis and for subsequent molecular and ancillary studies requires careful consideration of tissue management. Specialized strategies can be used to optimize tissue handling and management; however, these strategies are often time- and labor-intensive (Aisner 2016). Measures

31 SAMPLE ACQUISITION, PROCESSING, AND GENERAL DIAGNOSTIC PROCEDURES 29 to ensure preservation of tissue from initial diagnostic procedures should be maximized to reduce the need for repeat biopsy. When considering the quantity of tissue collected, practicing pathologists should know the minimum quantity needed by the laboratory performing the molecular biomarker testing, including for EGFR mutations, to best identify and preserve samples that are adequate for subsequent analysis. However, when it is uncertain whether the size of a sample is sufficient, the sample should be referred to the molecular laboratory for evaluation because testing may sometimes be possible on samples that are initially perceived to be insufficient. The possibility of testing also largely depends on the capacity of the testing laboratory to individually tailor the tumor enrichment and extraction in difficult cases. Body fluids are often used for molecular testing. For example, pleural fluid is often a sample type used for EGFR mutation analysis. The ability to use a pleural fluid sample is often highly dependent on a combination of three factors: the tumor cellularity of the specimen (pleural fluids often contain a high proportion of nontumor cells); the ability of the laboratory to enrich for tumor, which can be exceptionally challenging or impossible with pleural fluid samples; and the analytic sensitivity of the platform used for molecular testing. For example, a pleural fluid with a high degree of nontumor cells in which enrichment is not feasible would not be appropriately tested with a low analytic-sensitivity methodology, such as Sanger sequencing. A common question posed to a molecular laboratory is What is the absolute minimum number of tumor cells needed for testing? Few data are available regarding a minimum number of cells, and the answer is likely to depend on variables at each testing laboratory. In addition, the number of cells visible on a single section stained with H&E may not be representative of what remains in a FFPE block. Furthermore, the ability to test exceptionally scant specimens may be largely dependent on the quality of the nucleic acids extracted from the specimen, a feature that cannot be determined by histopathologic examination. The quality of nucleic acids derived from solid tumor samples depends on postprocedural specimen handling. Key factors relate to duration of formalin exposure, histopathologic processing parameters, exposure to other agents (specifically decalcifying agents or acids), and elapsed time in paraffin. Although optimal formalin exposure time has not been established for lung cancer samples, this time has been determined in other clinical settings, most notably breast cancer. Exposure of fewer than 6 hours and more than 72 hours has been associated with increased error rates in ancillary testing in this setting. On the basis of this information, it is suggested that lung cancer samples not be subjected to more than 72 hours of formalin exposure, including histopathologic processing time. When possible, formalin exposure should be limited to fewer than 48 hours. Other factors that can be mitigated include use of only neutral buffered formalin (because unbuffered formalin will render samples unsuitable for molecular testing), avoidance of decalcification treatments when bone metastases are sampled (Aisner 2016), and avoidance of other acid-based solutions, such as Bouin s fixative. Slides that are cut from a paraffin block and left for extended periods of time may undergo oxidation, resulting in reduced-quality nucleic acids. To the extent possible, tissue should be preserved in paraffin blocks rather than on archived unstained slides.

32 30 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER The quality of nucleic acids derived from non-ffpe cytopathology preparations are often superior to FFPE-derived specimens, owing to lack of exposure to formalin. Ethanol- and methanol-based fixatives used in most cytopathology specimens are generally less damaging to nucleic acids; however, there are insufficient data regarding additional variables that may affect the nucleic acid quality, such as coverslip mounting media or method of coverslip removal. Best Practices for Sample Collection Limited data exist at this time to suggest evidence-based best practices for sample collection. For tissue samples intended for FFPE, immediate deposition into formalin can reduce postprocedural degradation that may occur if a sample is kept in saline or other media or remains dry until receipt in the pathology laboratory. Numerous studies have demonstrated that rapid onsite evaluation can improve diagnostic yield, as well as procurement of additional tissue for molecular and ancillary studies when fine-needle aspiration is performed. Cytopathologists who are trained in the use of additional tissue for molecular and ancillary studies can also effectively triage samples into the appropriate preparations (eg, smears, spins, and cell blocks) to maximize the likelihood of successful testing. In 2016, the College of American Pathologists convened a multidisciplinary guideline group to address the outstanding questions related to specimen acquisition and handling. These highly anticipated guidelines are expected to be of significant assistance in identifying best practices in this rapidly evolving field. Circulating Tumor DNA from Plasma In the absence of sufficient tumor tissue for molecular analysis, examination of ctdna isolated from patient plasma has been shown to be a viable alternative for most patients with stage IV disease. Tumors often shed DNA into peripheral circulation, and the extent to which this occurs appears to be based on a variety of factors, including stage, disease burden, metastatic dissemination, tumor aggressiveness, proliferation, and apoptotic rate. DNA released into circulation in this manner is referred to as cfdna or ctdna. Although these terms are often used interchangeably, they have distinct meanings: cfdna refers to the totality of mutant and wild-type DNA shed into body fluids from both normal and tumor cells, and ctdna refers to the specific subset that is tumor in origin. Identification of tumor-specific DNA abnormalities in cfdna, such as EGFR-activating mutations, requires assays that are both highly sensitive (able to identify very small proportions of mutant alleles in a large pool of wild-type alleles) and highly specific (not prone to false-positive findings). The need for an assay with a specificity approaching 100% stems from the observation that, in most cases, ctdna mutations are detected at extremely low mutant allele frequencies (MAFs). The MAF is defined as the percentage of mutant alleles at a specific locus; thus, a MAF of 1% indicates a ratio of one mutant sequence to 99 wild-type sequences. Large-scale analysis of patients with NSCLC adenocarcinoma has shown that the vast majority with mutations that were detectable in cfdna also had MAFs of less than 1% (Mack 2016). When using a technique that requires such high levels of sensitivity, any assay that risks generation of false-positive results, even at very low frequencies, may result in misdiagnosis. (See Chapter 5 for a discussion of assays appropriate for detection of EGFR

33 SAMPLE ACQUISITION, PROCESSING, AND GENERAL DIAGNOSTIC PROCEDURES 31 mutations in cfdna.) Regardless of the assay used, great care must be taken to maximize detection of EGFR mutations during the sample-acquisition stage. Plasma Sample Preparation and Storage Due to the highly sensitive assays required for mutation detection in cfdna, all precautions must be taken to avoid contamination. Processing of plasma for ctdna analysis should be conducted by trained personnel in a dedicated blood laboratory. It is essential to avoid any risk of cross-contamination with mutation-positive samples (such as assay controls), and processing should take place in facilities that are isolated from PCR- amplification products. However, perhaps the biggest threat to ctdna analysis is the release of genomic DNA from white blood cells during sample preparation. Inadvertent lysis of white blood cells can significantly increase the total amount of cfdna in the tube, diluting the MAF. Thus, plasma is preferable to serum. When standard blood tubes, such as EDTA tubes are used, best practice includes rapid processing and a double-spin procedure. It is recommended that samples be processed within 4 hours of being collected (El Messaoudi 2013); if immediate processing is not possible, tubes should be kept at 4 o C. Centrifugation of blood at 1,000 to 2,000 g for 10 to 20 minutes separates the plasma from the cell fractions. However, a second centrifugation step is recommended because this initial process fails to remove all cells and cellular debris (Normanno 2017). This step is typically accomplished by transferring the plasma supernatant from the first spin into a 15-mL conical tube and centrifuging a second time at a minimum of 3,000 g for another 10 minutes. If possible, a refrigerated centrifuge should be used. The plasma supernatant from the second spin should be carefully pipetted into prelabelled freezer vials and stored at -80 o C in aliquots to avoid freeze thaw cycles. Alternatively, the use of a blood-collection tube that contains stabilizers that prevent both nuclease-mediated DNA degradation and lysis of nucleated cells can be used. A commonly used example is the Cell-Free DNA BCT (Streck). Although stabilizer tubes are expensive, their advantage is the ability to be stored at room temperature for days before processing without compromising the recovery of cfdna. If an outside service for ctdna analysis is being used, it is essential to strictly follow the service s plasma collection procedures, as they are developed and optimized for its analytical approaches. Other Considerations The timing of the blood collection relative to patient treatment is a crucial but often unaddressed factor for maximizing ctdna yield. Patients who are undergoing treatment will often have limited or nondetectable levels of ctdna. For example, Mok et al. demonstrated that after two cycles of chemotherapy (with or without sequential erlotinib), detection of EGFR mutations in cfdna was significantly ablated, particularly for patients who received an effective EGFR tyrosine kinase inhibitor (Mok 2015). The dynamic drop in the presence of mutant-positive alleles is likely a product of reduced tumor burden; however, even cytostatic responses to therapy may have a dramatic effect in the shedding of DNA because apoptosis typically occurs in proliferating cells. Thus, blood for cfdna analysis should be collected before treatment is initiated or at the time of disease progression. In patients with disease progression during treatment with an EGFR tyrosine kinase inhibitor, ctdna analysis can be instrumental in identifying emergent resistance mechanisms, particularly

34 32 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER in cases where secondary tissue biopsies are unfeasible or uninformative. For instance, detection of the EGFR T790M gatekeeper mutation in plasma can guide use of osimertinib, as discussed in Chapter 1. However, the absence of mutation positivity in plasma should not be construed as a definitive negative result. Small, indolent tumors may not shed sufficient DNA into peripheral circulation for detection. In treatment-naïve tumors, detection rates for a known EGFR-activating mutation in tissue range from 60% to 85%. Detection rates are lower still for patients with disease in the initial stages of progression (Qiu 2015). If a well-validated approach is used, positive detection of an actionable mutation in plasma ctdna can be used to guide clinical decision making; however, a lack of detection should be taken as inconclusive it cannot be determined whether the tumor is truly negative or is simply not shedding sufficient DNA for detection in plasma.

35 EGFR Gene Mutations 4 By James Chih-Hsin Yang, Pasi A. Jänne, Myung-Ju Ahn, and Leora Horn Sensitizing EGFR gene mutations are the most common actionable driver mutations found in patients with NSCLC, occurring in approximately 10% to 15% of white patients and as many as 50% of Asian patients (Hirsch 2009, Shi 2015, Skov 2015). These mutations occur within EGFR exons 18 21, which encode a portion of the EGFR kinase domain, and are a prime example of the complexity of the disease at the molecular level. Mutations involving exons 18, 19, and 21 are considered predictive of sensitivity to EGFR tyrosine kinase inhibitors (TKIs), such as gefitinib, erlotinib, afatinib, and icotinib; mutations in exon 20, however, are typically resistant to therapy, with the exception of S768I mutations and A763_Y764insFQEA variants (Figure 1) (Asahina 2006, Sasaki 2007, Ladanyi 2008, Wu 2008, Arcila 2013, Beau- Faller 2014, Baek 2015, Chiu 2015, Cheng 2015, Yang 2015a, Naidoo 2015). In addition, some uncommon mutations may be associated with primary drug sensitivity and resistance, whereas other rare EGFR mutations are of less clear clinical significance (Dahabreh 2010). Figure 1. Location of hotspots for EGFR gene mutations in chromosome 7p11.2, with a focus on exons and corresponding percentages of responsiveness to EGFR tyrosine kinase inhibitors. Reprinted by permission from Macmillan Publishers Ltd: Modern Pathology; 2012; 25: , 2012.

36 34 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER Common Mutations The two most common EGFR gene mutations are L858R and exon 19 deletion. The L858R mutation within exon 21 results in an amino acid substitution from a leucine (L) to an arginine (R) at position 858 in EGFR. L858R mutations are seen in approximately 43% of EGFR mutation-positive NSCLC (Mitsudomi 2010). L858R occurring as a single mutation is associated with sensitivity to treatment with EGFR TKIs. In contrast, L858R mutations occurring in combination with either A871E, L747S, de novo G719S, de novo T790M, L861P, or R776G mutations are associated with disease progression (Table 1) (Yeh 2013). EGFR exon 19 deletions are in-frame deletions occurring within exon 19 and are found in approximately 48% of EGFR-mutated lung tumors (Mitsudomi 2010). As many as 53 different exon 19 deletions have been described E746_A750 being most common, accounting for approximately 60% of exon 19 deletions. The majority of these deletions are associated with response to EGFR TKIs; however, deletion 19 unspecified in combination with either exon 20 insertion unspecified or V769M were noted to be associated with disease progression (Yeh 2013). Biochemical analyses of the deletion 19 and L858R mutations showed that exon 19 deletions appeared to be more sensitive than the L858R substitution to erlotinib inhibition (Carey 2006). In a meta-analysis of seven randomized trials (1,649 patients) comparing first-line treatment with first- and second-generation EGFR TKIs (950 patients) and chemotherapy (699 patients), the authors examined the effects of mutation subtype and clinical characteristics on progression-free survival rates (Lee 2015). Across all EGFR mutation subtypes, treatment with an EGFR TKI was associated with a 63% reduction in the risk of disease progression or death compared with chemotherapy (HR: 0.37; 95% CI: ; P < 0.001). The vast majority of patients harbored common EGFR mutations: 872 patients had exon 19 deletions, and 686 patients had exon 21 L858R substitutions. Subgroup analyses demonstrated that patients with exon 19 deletions had a 50% greater progression-free survival benefit when treated with an EGFR TKI than patients with exon 21 L858R substitutions (P < for interaction). Another mutation, EGFR exon 20 insertion, is seen in 4% to 9% of EGFR mutation-positive NSCLC (Naidoo 2015). This mutation is most often between amino acids 767 and 774 of exon 20 and confers primary resistance to EGFR TKIs (Sasaki 2007, Wu 2008, Arcila 2013, Yasuda 2013, Oxnard 2013, Beau-Faller 2014, Yang 2015a). Structural and molecular analyses have shown that, although this mutation may activate EGFR, it does so without increasing receptor affinity for EGFR TKIs (Yasuda 2013). One exception may be the A763_Y764insFQEA variant, which appears to be sensitive to EGFR TKIs (Yasuda 2013, Naidoo 2015). Secondary Mutations In most patients treated with an EGFR TKI, resistance develops within approximately 10 to 12 months. The T790M mutation is a second-site mutation occurring within exon 20 of EGFR that has been detected in approximately 50% to 60% of patients in whom acquired resistance to EGFR TKIs develops (Table 2) (Kobayashi 2005, Pao 2005). This so-called gatekeeper mutation, which involves a threonine-to-methionine substitution in exon 20, increases the affinity of mutant EGFR for ATP, thereby competitively inhibiting the binding ability of reversible EGFR TKIs (Yun 2008). Osimertinib was recently approved for patients with T790M mutation resistance to a first- or second-generation EGFR TKI, such

37 EGFR GENE MUTATIONS 35 Table 1. EGFR Gene Mutations Associated with Disease Progression + Mutation Number of Instances Responses to Therapy Number of Unique Progressive Disease a Patients b Studies c Study A763V Chou 2005 A859T Cappuzzo 2006, Taron 2005 Del19 A767-V Chou 2005 Del19 unspecified + ins Wu JY 2011 unspecified Del19 unspecified + V769M Wu JY 2011 E709G + G719C Wu 2011 E711K Pallis 2007 G719A + S768I Wu SG 2008, Wu JY 2008 G721D Ichihara 2007 G729R Pallis 2007 I744M Hsieh 2006 I759T Takahashi 2010 Ins insGVV Ichihara 2007 Ins20 A767-V769dupASV Wu JY 2008 Ins20 D770-N771insD Wu JY 2008 Ins20 P772-H773insYNP Wu JY 2008 H773Y Ins20 S768-D770dupSVD Wu JY 2008 Ins20 SVD Yang CH 2008 Ins20 unspecified Cappuzzo 2007 K806E Wu JY 2008 K860E Pallis 2007 L703F Pallis 2007 L747P Kimura 2007 L777G Ichihara 2007 L838P + E868G Hsieh 2006 L858R + A871E Wu JY 2011 L858R + G719S Hata 2010 L858R + L747S (de novo) Wu JY 2011 L858R + L861P Pallis 2007 L858R + T790M (de novo) Wu JY 2008, Ichihara 2007, Yang CH 2008, Jackman 2009, Tokumo 2006 L858R + R776G Wu JY 2008 L861Q + G719S Chou 2005 continued on next page

38 36 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER N826Y Wu JY 2011 N842S Wu JY 2011 S768I + V769L Asahina 2006 S784F Wu JY 2011 T847A + G863S Pallis 2007 T847I Wu JY 2011 T790M (de novo) Jackman 2009, Donovan 2009 V774M Wu JY 2011 V802I Jackman 2009 V852I Cappuzzo 2005 a Disease progression is defined by the Response Evaluation Criteria in Solid Tumors (RECIST) criteria (7), following treatment with an EGFR tyrosine kinase inhibitor (TKI, gefitinib or erlotinib therapy) as best response. b The number of unique patients does not match the number of responses to therapy because one patient may have been treated with EGFR TKIs in different lines of treatment and, therefore, may have multiple associated instances of response to therapy. c The number of unique studies refers to the number of different studies that encompass all of the patients with a particular mutation in the DNA Inventory to Refine and Enhance Cancer Treatment (DIRECT) database. Table 2. EGFR Gene Mutations in Patients with Acquired Resistance to EGFR Tyrosine Kinase Inhibitors (TKIs) Acquired Resistance Number of Unique Initial EGFR Status Mutation a Patients Studies b Study Del19 E746-A750 T790M 14 5 Costa 2008, Chen 2009, Balak 2006, Bean 2007, Engelman 2007 Del19 E746-T751insA T790M 1 1 Balak 2006 Del19 E746-T751insV T790M 1 1 Bean 2008 Del19 L747-A750insP T790M 1 1 Chen 2009 Del19 L747-E749;A750P T790M 3 2 Engelman 2007, Bean 2008 Del19 L747-P753insQ T790M 1 1 Ruppert 2009 Del19 L747-P753insS T790M 2 2 Balak 2008, Engelman 2007 Del19 L747-S752 T790M 2 2 Chen 2009, Balak 2006 Del19 L747-T751 T790M 1 1 Bean 2008 Del19 L747-T751;K754E T790M 3 2 Engelman 2007, Bean 2008 Del19 unspecified T790M 3 1 Onitsuka 2010 L858R D761Y 1 1 Balak 2006 L858R L747S 1 1 Costa 2008 L858R T854A 1 1 Bean 2008 L858R T790M 21 6 Chen 2009, Balak 2006, Engelman 2007, Bean 2008, Onitsuka 2010, Inukai 2006 Wild type T790M 4 1 Chen 2009 a Resistance mutations were acquired after treatment with an EGFR TKI. b The number of unique studies refers to the number of different studies that encompass all of the patients with a particular mutation in the DNA Inventory to Refine and Enhance Cancer Treatment (DIRECT) database.

39 EGFR GENE MUTATIONS 37 as gefitinb, erlotinb, or afatinib (Janne 2015). Germline EGFR T790M mutations occur in approximately 1% of white patients with NSCLC; cancers in such patients often also contain a second activating EGFR mutation. In addition, patients with germline T790M mutations may have familial cancer syndromes (Bell 2005). Studies have found that patients carrying a germline T790M mutation have a high lifetime risk of the development of lung cancer; this risk is as high as 31% among never-smoking genetic carriers (Gazdar 2014, Bell 2005). In addition, T790M mutations may be detected de novo in conjunction with an EGFR sensitizing mutation and are associated with decreased sensitivity to EGFR TKIs (Wu 2011). In a metaanalysis of three randomized controlled trials and 15 observational studies, pretreatment T790M mutations were more likely to be present with L858R mutations than with exon 19 deletions (Chen 2016). This association may provide an explanation for better progressionfree survival for patients with exon 19 deletions than for patients with L858R mutations (Jackman 2006, Riely 2006, Yang 2015b). T790M is most often seen in the cis position with an L858R mutation or exon 19 deletion; however, it can occur in the trans position as well (Kobayashi 2005). Among patients with resistance to osimertinib, a third mutation, C797S, was identified along with the activating mutation and T790M (Thress 2015). Preclinical studies have shown that, when C797S and T790M mutations were present in the trans position, cells were resistant to third-generation TKIs but sensitive to combined treatment with first- and third-generation inhibitors. When these mutations were in cis conformation, no TKIs were able to suppress EGFR activity (Niederst 2015). Uncommon EGFR Mutations Studies in which patients with uncommon EGFR mutations have been analyzed as a single group have often shown that the response to EGFR TKIs is lower among this group than among patients with either of the common mutations alone (L858R and exon 19 deletions) (Watanabe 2014, Bek 2015, Chiu 2015, Arrieta 2015,). However, when analyses are performed on individual mutations or on smaller selective subsets, substantial clinical heterogeneity clearly exists. The most frequently detected exon 18 mutation is G719X, followed by the E709X mutation (Beau-Faller 2014, Cheng 2015). Mutations in G719A, G719C, and G719S, among others, are seen in approximately 2% to 3% of patients with EGFR-positive NSCLC. These mutations are substitution at position 719 in EGFR; substitution from a glycine (G) to an alanine (A), cysteine (C), or serine (S) are associated with responses to an EGFR TKI but are less sensitive than the common mutations in vitro (Yeh 2013, Yun 2007, Kancha 2009). A retrospective analysis of patients pretreated with an EGFR TKI who received afatinib as part of the Afatinib Compassionate-Use Program following disease progression during one prior regimen of an EGFR TKI and chemotherapy included 10 patients with G719X mutations, and the median time to treatment failure was 2.6 months (Heigener 2015). In contrast, among patients with E709X mutations, the median time to treatment failure was 12.2 months with afatinib therapy. In a prospective pool analysis of 18 patients who had not been treated with an EGFR TKI and had G719X mutation and received afatinib, the response rate was 78%, with a progression-free survival of 13.8 months and an overall survival of 26.9 months (Yang 2015b). The L861Q mutation is seen in 2% of patients with EGFR mutations. This mutation occurs in exon 21 because of a substitution at position 861 in EGFR, from a leucine (L) to a glutamine

40 38 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER (Q), and is associated with disease control. However, patients with these mutations are less likely to have a response to therapy than patients with the common mutations deletion 19 and L858R (Yeh 2013). A post-hoc analysis of 838 patients from LUX-Lung 2, LUX-Lung 3, and LUX-Lung 6 showed uncommon EGFR mutations in 100 (12%) of the patients; 16 of these mutations were L861Q substitutions (Yang 2015a). The overall response rate associated with afatinib for patients with L861Q mutations was 56.3%, with a median progression-free and overall survival rate of 8.2 months and 17.1 months, respectively. In an analysis of next-generation sequencing data, rare EGFR mutations were identified in approximately 20% of samples (Costa 2016), a higher proportion than has been estimated (Mitsudomi 2010, Krawczyk 2015). Retrospective data from the DNA Inventory to Refine and Enhance Cancer Treatment (DIRECT) database provides a comprehensive review of known and reported EGFR mutations, including 188 unique EGFR mutations occurring in 207 different combinations, as well as of responses to targeted therapy (Yeh 2013). The information in this electronic catalog of EGFR mutations in NSCLC paired with clinical outcome was collected with use of a retrospective PubMed medical-subject heading search to identify patient-level, mutation-specific, drug-response data from different studies of EGFR-mutant NSCLCs (Yeh 2013). Electronic queries of DIRECT will result in a customized report for clinical use at the bedside. These data are continuously updated for public use. A growing number of genomic profiling services that include pertinent clinical relevance data are becoming available for community use. Molecular Mechanisms EGFR proteins belong to the ErbB family, which consists of ErbB1 (EGFR), ErbB2 (HER2), ErbB3 (HER3), and ErbB4 (HER4). ErbB family proteins are membrane proteins that bind to extracellular ligands and form homodimer or heterodimer with each other (Figure 2) (Blume- Jensen 2001). Dimerization of ErbB proteins plus ligand binding results in phosphorylation of tyrosine kinase in the intracellular domain, leading to activation of tyrosine kinase and subsequently activating the downstream pathways (Ferguson 2008). Dimerization of ErbB proteins are usually pre-formed before ligand binding. However, ErbB dimerization is not sufficient for activation of ErbB downstream pathways. Activation occurs only after ErbB dimerization and ligand binding (Tao 2008). The results of studies suggest that tetramers and clusters of ErbB proteins are frequently seen with ligand binding and ErbB activation (Clayton 2005). Of note, there is no known ligand for HER2 and no intracellular active kinase domain for HER3. Phosphorylation of different EGFR tyrosine kinases through dimerization of EGFR and ATP binding to the kinase pocket leads to diverse downstream pathway activation. There are two important activated pathways: one leads to activation of RAS, RAF, MEK, and ERK and eventually results in cellular proliferation, and the second pathway activates PI3K and AKT and prevents apoptosis of cells (Figure 2). ErbB family downstream signals are controlled through binding to different adaptor proteins and different binding sites of ErbB family proteins, and protein is recruited to phosphorylation sites of intracellular domain (Figure 3). EGFR is the most important member of the ErbB family. EGFR is present in most epithelial, mucosal, and glandular cells, and its primary function is to maintain the structure

41 EGFR GENE MUTATIONS 39 Figure 2. ERBB signaling pathways. Binding of a family of specific ligands to extracellular domain of ERBB leads to formation of homo- and heterodimers. In this case, HER2 is a preferred dimerization partner and heterodimers containing HER2 mediate a stronger signal than homodimers. Dimerization consequently stimulates intrinsic tyrosine kinase activity of the receptors and triggers autophosphorylation of specific tyrosine residues within the cytoplasmic domain. These phosphorylated tyrosines serve as specific binding sites for several signal transducers that initiate multiple signaling pathways including mitogen-activated protein kinase (MAPK), phosphatidyl inositol 3 kinase (PI3K)-AKT and signal transducer and activator of transcription protein (STAT) 3 and 5 pathways. These eventually result in cell proliferation, migration and metastasis, evasion from apoptosis, or angiogenesis, all of which are associated with cancer. P85 and p110 is a regulatory and catalytic subunit of phosphatidyl inositol 3 kinase (PI3K), respectively. STAT, SRC and mtor are also activated by ERBB sinaling. AR, amphiregulin; BTC, betacellulin; EPR, epirefulin; ERK, extracellular signal-regulated kinase; HB-EGFR, heparin binding EGF; MEK, MAP an ERK kinase; mtor, mammmalian target of rapamycin; NRG, neuregulin; TGF, transforming growth factor. Reprinted with permission from Cancer Sci. 2007;98: EGFR OLCg Gab1 PTP2c/Shp1 Src Cb1 Chk1 HER2 Stat5 Shc/Grb2 PI3K Grb7 HER4 HER3 Figure 3. Protein recruited to phosphorylation sites of ErbB intracellular sites. (Adapted from Bradshaw R, Dennis EA (eds). Functioning of Transmembrane Receptors in Signaling Mechanisms. Academic Press, p Used with permission.)

42 40 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER of these cells by replacing apoptotic cells in the surface with newly divided and differentiated cells from the stem cells. The number of EGFR molecules per cell range from 20,000 in physiologic conditions to more than several million per cell in pathologic conditions. Overexpression of EGFR has been reported in more than 65% of patients with NSCLC (Franklin 2002). Gene amplification shown by fluorescence in situ hybridization is the main mechanism of EGFR overexpression. EGFR mutation in the kinase domain (exons 18-21) was discovered after the successful administration of EGFR TKIs to patients with lung cancer (Paez 2004, Lynch 2004). Most of the patients with NSCLC who had a response to EGFR TKIs had one of these mutations in their tumors. The mutations found in patients with NSCLC are always in frame, which suggests that these mutations are functional. The fact that inhibition of the activating EGFR mutation by an EGFR TKI resulted in dramatic tumor response in most patients suggests that these activating EGFR mutations play a dominant role in maintaining the survival of these cancer cells. Among patients with EGFR-positive NSCLC, 80% also had EGFR gene amplification. Patients who had only EGFR gene amplification but no EGFR mutations did not have a response to gefitinib and had short progression-free survival times, suggesting that the presence of EGFR mutation but not EGFR gene amplification is the mechanism for maintaining the addicted EGFR pathways in patients with NSCLC who had a response to EGFR TKIs (Fukuoka 2011). Mutation of EGFR resulted in alteration of affinity of EGFR tyrosine kinase to ATP and EGFR TKIs (Yun 2008). This is due to the three-dimensional structural alterations in the EGFR tyrosine kinase domain (Figure 4). Resistance mutations, such as T790M, may accumulate in cancer cells through survival advantage by selective pressure after exposure of EGFR mutant cells to gefitinib, erlotinib, or afatinib (Ercan 2010). Mutant-specific EGFR TKIs, such as osimertinib, were designed to inhibit both activating and T790M EGFR mutations (Finlay 2014). However, a novel mutation, C797S, can emerge in response to prolonged exposure to osimertinib (Figure 5) (Thress 2015). Several resistance mechanisms to EGFR TKIs have been suggested (Table 3). L792 M790 P794 M793 C797 Figure 4. Three-dimensional structural alterations in the EGFR tyrosine kinase domain. Structural modeling of CO-1686 binding to EGFR T790M. The EGFR T790M kinase is shown in a ribbon representation (green) with the bound CO-1686 in orange. The aminopyrimidine binds to the hinge residue Met793 through hydrogen bonding (yellow dashed lines). The C5-CF3 substitution points to the gatekeeper residue Met790. Both C2 and C4 substitutions adapt a U-shaped binding mode. The piperazine ring is facing an open space in the solvent exposure area. The meta-acrylamide points to Cys797 and forms the covalent bond. Reprinted with permission from Cancer Discov. 2013; 3:

43 EGFR GENE MUTATIONS 41 Figure 5. A 3-dimensional structure of the C797S allele in the EGFR gene. The structure of a typical tyrosine-specific kinase domain of EGFR is shown in the middle panel, along with an ATP analog, 5 -adenylyl-imidodiphosphate (AMP-PNP). The P-loop and the activation loop are in light blue and light green, respectively. Leucine 858 (L858), which is frequently mutated in lung cancer, is highlighted in pink. Similarly labeled is threonine 790 (purple), which is replaced by a methionine in some advanced lung tumors and confers resistance to first-generation tyrosine kinase inhibitors (TKIs). The squared area is magnified in the side panels and shows the structure of the T790M mutant in a complex with either a first-generation TKI, erlotinib (right), or with a second-generation TKI, afatinib (left). The latter structure shows a replacement of cysteine 797 to second- and thirdgeneration EGFR TKIs. The chemical structures of both erlotinib and afatinib are shown. Reprinted with permission from Semin Cell Dev Biol. 2016;50: Table 3. Possible Resistance Mechanisms to EGFR Tyrosine Kinase Inhibitors (TKIs) Primary resistance EGFR modification Alternative pathway activation Acquired resistance EGFR modification Alternative pathway activation Histologic Other First- and Second- Generation Exon 20 insertion BIM deletion T790M Amplification Bypass MET amplification HGF overexpression AXL overexpression HER2 amplification Downstream PTEN loss PI3K mutation BRAF V600E SCLC transformation (RB loss) Epithelial-mesenchymal transition Drug efflux (ABCG2) RTK internalization Third-Generation: Osimertinib EGFR TKI C797S G769S/R L792F/H T790M loss Bypass MET amplification HER2 amplification Downstream BRAF V600E NRAS mutation SCLC transformation Third-Generation: Rociletinib C797S L718Q L844V Amplification SCLC transformation Epithelial-mesenchymal transition Third-Generation: WZ4002 C797S L718Q L844V Bypass IGF1R activation Downstream ERK1 and ERK2 activation (Cortot 2014, Piotrowska 2015, Ercan 2015, Tricker 2015, Eberlein 2015, Wlater 2013, Oxnard 2015, Park 2016, Ou 2016, Ham 2016, Ou 2017, Chen 2017).