Biomarker of Immunotherapy for Future Lung Cancer Treatment

Biomarker of Immunotherapy for Future Lung Cancer Treatment May 6, 2017 TJCC Teh-Ying Chou, MD, Ph. D, MBA

Evolution of Lung Cancer Classification – Molecular Pathology 2000 2008 Non-small cell lung cancer Adenocarcinoma TARGETS TODAY (2015) EGFR ALK ROS-1 FUTURE TARGETS KRAS and others BRAF HER 2 RET Molecular Pathology-based Personalized Medicine of. DDR 2 Lung. METCancer FGFR 1 PI 3 K Large-cell carcinoma & others Small-cell lung cancer Squamous cell carcinoma The colors denote different histological subtypes Small cell lung cancer Lancet 2013; 382: 709 -19 Adenocarcinoma with actionable mutations Large-cell carcinoma Squamous cell carcinoma without oncogenic alteration Small-cell carcinoma Squamous cell carcinoma with oncogenic alteration

lung cancer pathology 1999 2004 2015 Molecular Pathology-based Precision Medicine of Lung Cancer Histology Targeted Therapy Immunotherapy Precision Medicine

Biomarker of Immunotherapy for Future Lung Cancer Treatment • Immunotherapy for Lung Cancer • Biomarkers of Immunotherapy for Lung Cancer • PD-L 1 IHC: Prediction Biomarker for Anti-PD 1/PD-L 1 Immune Checkpoint Inhibitor Therapy • PD-L 1 IHC: The Blueprint Project • PD-L 1 IHC: The IASLC Atlas • PD-L 1 IHC: The Future Perspectives

Immunotherapy

Immune Checkpoint Inhibition

PD-L 1 immunohistochemistry as a biomarker Presented By Keith Kerr at 2015 ASCO Annual Meeting

Slide 11 Presented By Justin Gainor at 2015 ASCO Annual Meeting

Slide 14 Presented By Justin Gainor at 2015 ASCO Annual Meeting

Predictive value? Atezolizumab Detection antibody* SP 142 Nivolumab 28 -8 Pembrolizumab 22 C 3 IHC platform Ventana Dako Tested cells NSCLC (IC and TC) UC (IC) Lung (TC) NSCLC (TC) UC (TC and stroma) PD-L 1+ as ≥ 5% of TCs PD-L 1+ as ≥ 50% of TCs Estimated PD-L 1 prevalence in NSCLC TC 1/2/3 & IC 0* TC 1/2/3 & IC 1/2/3‡ 11% 26% IC 1/2/3 & TC 0* 30% ~46%1 PD-L 1 expression correlates with efficacy *TC 3 or IC 3 = TC ≥ 50% or IC ≥ 10% PD-L 1+; TC 2/3 or IC 2/3 = TC or IC ≥ 5% PD-L 1+; TC 1/2/3 or IC 1/2/3 = TC or IC ≥ 1% PD-L 1+; TC 0 and IC 0 = TC and IC < 1% PD-L 1+, respectively. ‡POPLAR (Spira, et al. ASCO 2015); Study-003 (Brahmer, et al. ASCO 2014; KEYNOTE-001 (Garon, et al. AACR 2015) 1. Kerr KM, et al. J Thorac Oncol 2015 ~25%1

Mutational Load and Immunotherapy Presented By Justin Gainor at 2015 ASCO Annual Meeting

Mutational Load and Immunotherapy Mutation Load and Immunotherapy Presented By Christine Walko at 2016 ASCO Annual Meeting

Slide 17 Presented By Padmanee Sharma at 2016 ASCO Annual Meeting

Overlap between responders and non-responders make it difficult to use mutational load as a predictive biomarker Presented By Padmanee Sharma at 2016 ASCO Annual Meeting

Highest Mutation Burden Subset Displays Marked Improvement in OS Presented By Alexandra Snyder Charen at 2016 ASCO Annual Meeting

Why Does Mutational Load Matter? Presented By Alexandra Snyder Charen at 2016 ASCO Annual Meeting

Issues with the biomarkers • Heterogeneity- multiple tumors and multiple passes within a tumor • Interval between biopsy and treatment • Primary versus metastatic disease • Antibody and staining conditions • Defining a positive results (cut-offs): – – Cell type expressing PD-L 1 (immune cell vs. tumor or both) Location of expression- cell surface vs. intracellular vs. stromal Intensity, percent of cells “positive” Distribution- patchy vs. diffuse, intratumoral vs. peripheral

“Companion/Complementary Testing” is the gold standard basis of therapy selection, while mutation load and neoantigen measurement etc. are “surrogate testing” as predictive biomarkers

The Blueprint Project: Comparing PD-L 1 IHC Diagnostics for Immune Checkpoint Inhibitors

Different Classes of In Vitro Diagnostics 39

Blueprint Project: Study Design • Two-phase study to gain sufficient data and rigor: Feasibility on small cohort evaluated at Dako and Ventana To assess the four PD-L 1 diagnostics on the same NSCLC cases and gather initial data in two stages: Phase 1 • Step 1: Evaluate analytical comparability by quantifying and comparing PD-L 1 expression on tumour and immune cells • Step 2: Clinical agreement was assessed through comparisons of patient classification and agreement using various combinations of assays and PD-L 1 staining thresholds Phase 2 TBD: Proposed larger, statistically powered study that will be designed from the Phase 1 “information gathering” 40

Blueprint Project: Analytical Comparison of PD-L 1 Staining 22 C 3 28 -8 SP 263 SP 142 • Three assays (22 C 3, 28 -8, Immune Cells Tumour Cells SP 263) demonstrate similar analytical performance with respect to percentages of tumour cells positive and dynamic range • • • SP 142 consistently labels fewer tumour cells 22 C 3 28 -8 SP 263 SP 142 All assays labeled immune cells, but there was increased analytical variability across assays and readers with immune-cell staining vs tumour-cell staining There is generally higher agreement between observers when assessing TPS than when assessing ICPS 41

Blueprint Project: Step 1 Study Results Analytical comparison of tumor and immune-cell–membrane staining for each of the 39 NSCLC cases, using all four PD-L 1 assays Tumour Cells 90 22 C 3 28 -8 80 SP 142 SP 263 100 % Immune Cell Staining % Tumour Cell Staining 100 70 60 50 40 30 20 10 90 22 C 3 28 -8 80 SP 142 SP 263 70 60 50 40 30 20 10 0 0 • • • Immune Cells 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 Cases Data points represent the mean score from three pathologists, for each assay, with each NSCLC case Identical PD-L 1 expression reported is represented by superimposed points No clinically relevant cutoff is applied 42

Blueprint Project: Clinical Diagnostic Comparison • • 36. 9% of the cases studied showed discrepant results for PD-L 1 expression between the assays There is the potential for different diagnostic results according to the key clinical cutoffs if assays and algorithms are mismatched – • • The assays, the cutoffs, and therapeutics may not be interchangeable The results of this preliminary study should not alter current guidelines as indicated for each therapeuticdiagnostic validated combination pair Further studies are required SP 142 TC 1 IC 1 (30) 22 C 3 1% (26) 28 -8 1% (26) 1 0 6 5 SP 263 25% (20) 1 0 0 19 0 0 0 Overall agreement with each selected cutoff Assay 1% TPS TC 1/IC 1 25% TPS 22 C 3 38/38 (100%) 31/38 (81. 6%) 30/38 (78. 9%) 28 -8 38/38 (100%) 29/38 (76. 3%) SP 142 20/38 (52. 6%) 38/38 (100%) 15/38 (39. 5%) SP 263 32/38 (84. 2%) 23/38 (60. 5%) 38/38 (100%) Agreement rates between assays and cutoffs are shown All four assays are in agreement for 19 cases 43

The Blue Print Project Courtesy of Professor Fred Hirsch 44

Practical Considerations for PD-L 1 Testing IVDs vs LDTs • Formal HTA Assessment of PD-L 1 Dx Test Is Not Required in Most Markets Class III In Vitro Diagnostics 1, 2 Laboratory-Developed Tests 1, 2 Developed in the form of a kit for general use Created by local laboratories, at which use laboratories is restricted FDA approval of kits required CE marking required No requirement for premarket review, plans in place for future No requirement for premarket review, manufacturer selfdeclares Highly regulated PMA process before being marketed Clinical safety or performance evaluation required Clinical validation is not required for their use Clinical safety or performance evaluation required CE = Conformité Européene (European Conformity); Dx = diagnostic; HER 2 = human epidermal growth factor receptor 2; HTA = health technology assessment; IVDs = in vitro diagnostics; LDTs = laboratory-developed tests; PMA = premarket approval. 1. In Vitro Companion Diagnostic Devices. FDA Guidance. August 6, 2014. http: //www. fda. gov/downloads/medicaldevices/deviceregulationandguidance/guidancedocuments/ucm 262327. pdf. Accessed September 2016. 2. Understanding Europe’s New Medical Devices Regulation (MDR). Key changes contained in the proposed MDR and their impact on manufacturers. EMERGO white paper: July 2016, 45 anticipated publishing date early 2017.

Assays and LDTs: Mix and Match? Drug Anti–PD-L 1 Assay Definition of Positivity Outcome By choice Assay validated for drug in Definition validated in trial for drug Predictable based upon Known trial data By choice Any trial-validated assay Definition validated in trial for drug of choice Uncertain By choice Any trial-validated assay Definition validated with Very uncertain the assay Not known By choice LDT using any clone Unknown Not known Very, very uncertain Risk to Patient? Not known Courtesy of Professor Keith Kerr 46

The “Consensus” we may end up with… Cancer Patients (to be treated by Anti-PD 1 / Anti-PDL 1) Reflex Testing? PDL 1 IHC accompanied with / complimentary to Drug A PDL 1 IHC for Drug B PDL 1 IHC for Drug C + + + Drug B Drug A Drug C 47

1. 2. Goals: Compare analytical performance of 4 assays (22 C 3, 28 -8, SP 142, SP 263) using the staining protocols used in respective clinical trials Compare the treatment-determining scoring algorithm developed for each assay and used in clinical trials Journal of Thoracic Oncology Vol. 12 No. 2: 208 -222

Blueprint phase 2 - Goals • Validation of Blueprint phase 1 results using different types of clinical samples (resection, small biopsy, cytology cell block) • Comparability and heterogeneity of PD-L 1 assay results in surgical tumor resection, core needle and FNA samples prepared from same tumor. • Inter-observers concordance among larger panel of pulmonary pathologists • Concordance of PD-L 1 scoring using standard light microscopy vs. digital images accessed by web-based system.

Blueprint Phase 2 (2017) Steering Committee Funders PFE/EMD Serono Astra. Zeneca Technical facilitator BMS Dako Ventana IASLC Pathology expertise Execution Team Core Team Regulatory/public advocates Neutral observers Genentech FDA Histo. Gene X EMA Merck Technical expertise

There is literature comparing different PD-L 1 IHC assays but…… • Mostly of limited value in the testing debate • Based upon LDTs • Tell us nothing about the comparability of trialvalidated assays • May not be clinically practical BUT (and inevitably) Many demonstrate differences in outcome on the same cases

IASLC Atlas of PD-L 1 IHC Testing

The Need of Liquid Biopsies in Oncology • Diagnosis of Cancer • Prognosis Prediction • Therapy Response Prediction • Risk Assessment for Relapse • Monitoring Tumor Burden • Treatment Selection

Liquid Biopsies Capture Molecular Heterogeneity of Tumor

PDAC CTC Cluster Count Criteria Sub-classification of PDAC CTC Cluster Sub-classification #1 CTC cluster contain = 1 CTC Sub-classification #2 CTC cluster contain = 2 CTC Sub-classification #3 CTC cluster contain = 3 CTC Sub-classification #3 CTC cluster contain >= 5 cells CTC Cluster Criteria Contain one CTC with WBC Size > 15 um Contact nuclear stain Contain 2 CTC in any shape with x WBCs Size >20 um Contact nuclear stain Contain 3 CTC in any shape with x WBCs Size >25 um Contact nuclear stain CTC cluster more than 5 cells contain at least one CTC Represented Figures Red: Pan. CK Green: CD 45 Blue: DAPI