Clinical Need and diagnostic challenge-the upcoming landscape of checkpointinhibitors ESP/ESMO Marina Chiara GARASSINO Amsterdam 2017 September, 4th 29 European Congress of Pathology
Disclosures BMS ROCHE MSD ASTRA ZENECA
DC = dendritic cell; NK = natural killer. 1. Dranoff G. Nat Rev Cancer. 2004;4:11 22; 2. Janeway CA, et al. Immunobiology: The Immune System in Health and Disease. 6th ed. New York, NY: Garland Science; 2004. Cells of the Immune System Innate immune system: involving proteins (chemokines and cytokines) and cells, is considered to be the first line of immune defense and does not generate an antigenspecific response 1,2 Mast cell B cell Adaptive immune system: mediated by B and T cells is highly specific and capable of generating an antigenspecific response 1,2 Induction requires presentation of antigens by cells of the innate immune system Macrophage NK cell Complement protein Basophil Eosinophil Neutrophil DC Granulocytes γδ T cell NKT cell CD4 + cell T cell CD8 + cell Antibodies Innate Immunity (rapid response) Adaptive Immunity (slow response, memory) Adapted from Dranoff G. 1
Explanation of the Molecular Mechanisms of Checkpoint Inhibitors and Other Key Emerging Immunologic Strategies Chen L, et al. Nat Rev Immunol. 2013;13(4):227-242.
Regulation of T-Cell Activation: Balancing Activating and Inhibitory Signals Immune checkpoints limit, or check, an ongoing immune response Prevents damage to the body s healthy tissues Negative co-stimulation, also called co-inhibition, helps shut down immune responses PD-1, CTLA-4, and LAG-3 are examples of co-inhibitory checkpoint molecules Amplitude and quality of a T-cell response is regulated by a balance of activating and inhibitory signals APC/ Tumor B7-2 (CD86) PD-L1 PD-L2 CD40 CD137L OX40L MHC PD-1 B7-1 (CD80) LAG-3 CD28 CD40L CD137 OX40 Activation B7-1 (CD80) CTLA-4 Inhibition TCR Inhibition Inhibition T cell Inhibition Activation Activation Activation CTLA-4 = cytotoxic T-lymphocyte antigen-4; LAG-3 = lymphocyte activation gene-3; PD-1 = programmed death-1; PD-L1 = programmed death-ligand 1. Pardoll DM. Nat Rev Cancer. 2012;12:252 264.
L.Chen et al, Nature Reviews Immunology, 2013 Apr;13(4):227-42 Co-signalling in T cells
Mutational Load Creates Neoantigens Stephens PJ, et al. Nature. 2012;486(7403):400-404. Fritsch EF, et al. Oncoimmunology. 2014;3:e29311
Mutational Heterogeneity in Cancer: Altered Proteins Contain Neo-Epitopes for Immune Recognition Fig. 2 Estimate of the neoantigen repertoire in human cancer. Does mutational load correlate with response to immune checkpoint blockade? Ton N. Schumacher, and Robert D. Schreiber Science 2015;348:69-74 Published by AAAS
Colorectal Cancers Are Generally Unresponsive to PD-1 Blockade, but the MSI-High Subset Has a High Mutational Load Lawrence MS, et al. Nature. 2013;499(7457):214-218. Microsatellite instability (MSI): Genetic hypermutability resulting from deficient mismatch repair (dmmr), present in ~15% colon cancers and in some other tumor types Topalian SL, et al. J Clin Oncol. 2013;31(suppl): Abstract 3002.
Immunoscore Meta-analysis of 124 published articles studying the impact of cytotoxic T cells, memory T cells, and T-helper subpopulations with regards to prognosis of patients with cancer (20 cancer types analyzed) Galon J, et al. Science. 2006;313(5795):1960-1964. Fridman et al, Nature Review Cancer, 2012
BIOMARKERS PD-L1 MUTATIONAL BURDEN IFN-gamma Others
Positive Negative Expression of PD-L1 may be heterogeneous and varies with antibody used H&E E1L3N SP142 1 mm Immunofluorescence shows stroma and epithelial staining are often concordant and adjacent Green = Cytokeratin Blue = Nuclei Red = PD-L1 (SP142) Unpublished Data: J McLaughlin, K Schalper and D Rimm (Yale Pathology) Unpublished Data: J McLaughlin,K Schalper and D Rimm (Yale Pathology)
PD-L1 Immunohistochemistry: Expression Heterogeneity and Potential for Sampling Error Biopsy Core 1 18g needle = 800 µm Biopsy Core 2 13
% Immune Cell Staining The Blueprint Project: Comparing PD-L1 IHC Diagnostics for Immune Checkpoint Inhibitors 100 90 Dako: 22-C3 and 28-8 Ventana: SP 263 Three assays (22C3, 28-8, SP263) demonstrate similar analytical performance with respect to percentages of tumor cells positive and dynamic range -SP142 consistently labels fewer tumor cells All assays label immune cells but there is less precision in analytical performance than with tumor cell labeling There is generally higher agreement between observers when assessing TPS than when assessing ICPS % Tumor Staining % Tumor Staining 80 70 60 50 40 30 20 10 0 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 Cases 22C3 100 90 80 70 60 50 40 30 20 28-8 SP142 SP263 Ventana: SP 142 Hirsch et al, JTO 2016 14 10 0 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 Cases 22C3 28-8 SP142 SP263
Comparison of Two Studies 100 90 80 % Tumor Staining 70 60 50 40 30 20 10 0 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 Cases 22C3 28-8 SP142 SP263 The Blueprint Project: Phase II How far is it safe to deviate from trial-validated practice? How should LDTs be validated? Rimm D et al. CMSTO 2016; Hirsch FR et al AACR 2016
What is a PDL1 positive tumor? Atezolizumab Durvalumab 25% / 90%
First line
KEYNOTE-024 study design: Pembrolizumab vs chemotherapy in first-line NSCLC Reck M, et al. New Engl J Med 2016;
KEYNOTE-024 study: PFS and OS Assessed per RECIST v1.1 by blinded, independent central review Data cut-off: May 9, 2016 Data cut-off: May 9, 2016 Reck M, et al. New Engl J Med 2016;
CheckMate-026 study: PFS and OS Treatment-related adverse events of Grade 3 or 4 occurred in 17.6% and 50.6% of patients treated with nivolumab or chemotherapy, respectively Carbone DP, N Engl J Med 2017 22, 376(25):2415-2426
MYSTIC TRIAL press release
Second and further lines PD-L1 is not the only criteria of choice
CHECKMATE-017 study (SQUAMOUS): OS Nivolumab vs docetaxel OS by PD-L1 Expression Presented By David Spigel at 2015 ASCO Annual Meeting
Overall survival (%) Overall survival (%) Overall survival (%) CHECKMATE-057 study (NON SQUAMOUS): OS Nivolumab vs docetaxel 100 1% PD-L1 expression 100 5% PD-L1 expression 100 10% PD-L1 expression 90 mos (mo) 90 mos (mo) 90 mos (mo) 80 70 Nivolumab Docetaxel 17.7 9.0 80 70 Nivolumab Docetaxel 19.4 8.1 80 70 Nivolumab 19.9 Docetaxel 8.0 60 60 60 50 50 50 40 40 40 30 20 Nivolumab Docetaxel 30 20 Nivolumab Docetaxel 30 20 Nivolumab Docetaxel 10 0 HR (95% CI) - 0.58 (0.43, 0.79) 0 3 6 9 12 15 18 21 24 27 30 Time (months) 10 0 HR (95% CI) - 0.43 (0.30, 0.62) 0 3 6 9 12 15 18 21 24 27 30 Time (months) 10 0 HR (95% CI) - 0.40 (0.27, 0.58) 0 3 6 9 12 15 18 21 24 27 30 Time (months) Borghaei H, et al. N Engl J Med 2015;373:1627 39 (suppl appendix)
CHECKMATE-057: sub-group analysis Treatment effect on OS in predefined sub-groups
KEYNOTE-010: Pembrolizumab vs docetaxel Herbst RS, Presented December 20, 2015
KEYNOTE-010: Pembrolizumab vs docetaxel: OS Time (months) Grade 3 5 treatment-related adverse events were less common with pembrolizumab than with docetaxel: 13% of patients given 2 mg/kg, 16% given10 mg/kg, and 35% of patients given docetaxel). Herbst RS, Lancet 2016; 387: 1450 50
OAK Phase III Study Design: atezolizumab vs docetaxel a A prespecified analysis of the first 850 patients provided sufficient power to test the co-primary endpoints of OS in the ITT and TC 1/2/3 or IC 1/2/3 subgroup ( 1% PD-L1 expression TC: Tumor cells; IC: tumor-infiltrating immune cells Rittmeyer A, et al. Lancet. 2017 Jan 21;389(10066):255-265
OAK: atezolizumab vs docetaxel: OS a Stratified HR Treatment-related adverse events of Grade 3 or 4 occurred in 15% and 43% of patients treated with atezolizumab or docetaxel, respectively Rittmeyer A, et al. Lancet. 2017 Jan 21;389(10066):255-265
OS (%) OS (%) OS by PD-L1 expression Chechmate 057 (Nivolumab) 10 0 90 80 70 60 50 40 30 20 Niv o Doc 1% PD-L1 expression level Median OS (mo) Nivo 17.2 Doc 9.0 10 HR (95% CI)=0.59 (0.43, 0.82) 0 0 3 6 9 1 1 1 2 2 2 2 5 8 1 4 7 10 0 90 80 70 60 50 <1% PD-L1 expression level Median OS(mo) Nivo 10.4 Doc 10.1 40 30 Niv o 20 Doc HR 10(95% CI)=0.90 (0.66, 1.24) 0 0 3 6 9 1 1 1 2 2 2 Time (months) 2 5 8 1 4 7 > % 1 PD-L1 HR 0.59 < % 1 PD-L1 HR 0.90 Keynote 10 (Pembrolizumab).Rizvi NA, et al. Lancet Oncol 2015 R Herbst et al. Lancet 2016 Barlesi, et al. ESMO 2016 OAK Trial (Atezolizumab) TC3 or IC3 HR: 0.41 TC0 and IC0% HR: 0.75 PD-L1 > 50% HR: 0.53 PD-L1 < 1% HR: 0.76 In the II line setting, do we really select patients according a more favorable HR?
ATLANTIC study: durvalumab Overall Survival in Cohorts 2 and 3 (full analysis set) Treatment-related adverse events of Grade 3 occurred in 8.3% of patients in cohort 2 and 18% of patients in cohort 3 Garassino MC, WCLC 2016
TUMOUR MUTATIONAL BURDEN & NEOANTIGENS
Total exome mutations (mutations/mb) Total Exome Mutations vs Genes in FoundationOne Panel CheckMate 026 TMB Analysis: Nivolumab in First-line NSCLC 100 50 10 1 1 10 50 100 FoundationOne panel a (mutations/mb) a Based on in silico analysis filtering on 315 genes in FoundationOne comprehensive genomic profile (Foundation Medicine, Inc, Cambridge, MA, USA) 1 1. Frampton GM, et al. Nat Biotechnol 2013;31:1023 1031 3 5
Data Interoperability of Whole Exome Sequencing (WES) Based Mutational Burden Estimates from Different Laboratories Concordance varied from 3-89% High-quality WES fro archival FFPE material is challenging Clinical interpretation is not standardized across labs Qiu et al, International Journal of Molecular Sciences, 2016
PFS (%) PFS by Tumor Mutation Burden Subgroup CheckMate 026 TMB Analysis: Nivolumab in First-line NSCLC High TMB Low/medium TMB 100 90 Median PFS, months (95% CI) Nivolumab Chemotherapy n = 47 n = 60 9.7 (5.1, NR) 5.8 (4.2, 8.5) 100 90 Median PFS, months (95% CI) Nivolumab Chemotherapy n = 111 n = 94 4.1 (2.8, 5.4) 6.9 (5.5, 8.6) 80 70 HR = 0.62 (95% CI: 0.38, 1.00) 80 70 HR = 1.82 (95% CI: 1.30, 2.55) 60 60 50 Nivolumab 50 40 40 30 20 Chemotherapy 30 20 Chemotherapy 10 10 Nivolumab No. at Risk Nivolumab Chemotherapy 0 0 3 6 9 12 15 18 21 Months 47 30 26 21 16 12 4 1 60 42 22 15 9 7 4 1 0 0 3 6 9 12 15 18 21 24 Months 111 54 30 15 9 7 2 1 1 94 65 37 23 15 12 5 0 0 > 240 somatic mut < 100/100-240 somatic mut 3 7
OS (%) OS by Tumor Mutation Burden Subgroup CheckMate 026 TMB Analysis: Nivolumab in First-line NSCLC High TMB Low/medium TMB 100 90 80 Median OS, months (95% CI) Nivolumab Chemotherapy n = 47 n = 60 18.3 (11.4, NR) 18.8 (11.3, NR) HR = 1.10 (95% CI: 0.64, 1.88) 100 90 80 Median OS, months (95% CI) Nivolumab Chemotherapy n = 111 n = 94 12.7 (9.9, 16.1) 13.2 (9.5, 15.2) HR = 0.99 (95% CI: 0.71, 1.40) 70 1-y OS rate = 64% vs 60% 70 60 60 1-y OS rate = 54% vs 53% 50 Chemotherapy 50 No. at Risk Nivolumab Chemotherapy 40 30 20 10 0 68% received nivolumab as crossover and/or post-study treatment Nivolumab 0 3 6 9 12 15 18 21 24 Months 47 41 35 33 30 24 13 4 0 60 56 48 45 36 34 19 9 1 40 30 20 10 0 55% received nivolumab as crossover and/or post-study treatment 0 3 6 9 12 15 18 21 24 Months 111 99 82 69 59 47 28 11 3 94 87 71 60 50 38 21 6 2 Chemotherapy Nivolumab 27 0 1
TMB (no. of missense mutations) Analysis of the Association Between TMB and PD-L1 Expression CheckMate 026 TMB Analysis: Nivolumab in First-line NSCLC 1000 High TMB 316 243 100 Low/medium TMB 32 10 0 25 50 75 100 PD-L1 (% tumor expression) a a All patients had 1% PD-L1 tumor expression There was no association between TMB and PD-L1 expression in patients with 1% PD-L1 tumor expression 3 9
PFS (%) PFS by TMB Subgroup and PD-L1 Expression CheckMate 026 TMB Analysis: Nivolumab in First-line NSCLC 100 Nivolumab Arm 100 Chemotherapy Arm 75 High TMB, PD-L1 50% 75 50 50 No. at Risk High TMB, PD-L1 50% High TMB, PD-L1 1 49% Low/medium TMB, PD-L1 50% Low/medium TMB, PD-L1 1 49% 25 0 Low/medium TMB, PD-L1 50% High TMB, PD-L1 1 49% Low/medium TMB, PD-L1 1 49% 0 3 6 9 12 15 18 21 24 Months 16 13 10 8 8 6 2 0 0 31 17 16 13 8 6 2 1 0 41 21 12 6 2 2 1 0 0 70 33 18 9 7 5 1 1 1 25 0 High TMB, PD-L1 1 49% Low/medium TMB, PD-L1 50% 0 3 6 9 12 15 18 21 Months 32 24 13 12 7 5 2 1 28 18 9 3 2 2 2 0 41 30 14 10 5 4 2 0 53 35 23 13 10 8 3 0 Low/medium TMB, PD-L1 1 49% High TMB, PD-L1 50% 4 0
Neoantigens In Cancer Schumaker and Schreiber, Science 2015
Is Neoantigen burden meaningful? Or just a different way of looking at mutation burden?
Is Neoantigen burden meaningful? Examination of clonality may help Schumaker and Schreiber, Science 2015
High Mutation Burden, Low heterogeneity McGraham; Science2017
Fehrenbacher L, Lancet 2016; 387: 1837 46
Immunosuppressive tumour phenotypes Teng et al, Cancer 2015
Tumors Use Complex, Overlapping Mechanisms to Evade and Suppress the Immune System 3 2 Priming and Activation CD28/B7.1, CD137/CD137L OX40/OX40L, CD27/CD70 HVEM, GITR, IL-2, IL-12 CTLA4/B7.1 PD-L1/PD-1 PD-L1/B7.1 Prostaglandins Cancer antigen presentation TNF-, IL-1 IFN-, CD40L/CD40 CDN, ATP HMGB1, TLR IL-10, IL-4, IL-13 Lymph node Tumour Stimulatory factors Blood vessel Trafficking of T cells to tumours CX3CL1, CXCL9, CXCL1 CCL5 Infiltration of T cells into tumours LFA1/ICAM1 Selectins VEGF Endothelin-B receptor Recognition of cancer cells by T cells T cell receptor Reduced pmhc on cancer cells 4 5 6 1 Release of cancer cell antigens Immunogenic cell death Tolergenic cell death Inhibitors Killing of cancer cells IFN-, T-cell granule content TIM-3/phospholipids PD-L1/PD-1 PD-L1/B7.1 IDO TGF- BTLA VISTA LAG-3 Arginase MICA/MICB B7-H4 7 Chen DS and Mellman I. Immunity. 2013 Jul 25;39(1):1-10.
Cancer immune contexture scenarios and personalized medicine for immunotherapy PD-L1 Adapted from Giraldo NA, et al. Curr Opin Immunol 2014;27:8 15
We are far from this simplification
Conclusions PD-L1 is the factor for which we have more robust data to select not the only Mutational Burden plays an important role, but it is not easily reproducible and transferable in clinical practice Neoantigens can be a similar way to look at the TMB Clinical characteristics are important (smoking habits) It is possible that the selection will be through multiple tests A new world of drugs is coming Pathologists are key for the future personalization of cancer treatment
Thank YOU Very Grazie