IMMUNOTHERAPY FOR GASTROINTESTINAL CANCERS Dr Elizabeth Smyth Cambridge University Hospitals NHS Foundation Trust ESMO Gastric Cancer Preceptorship Valencia 2018
DISCLOSURES Honoraria for advisory role Servier, Celgene, BMS, Five Prime Therapeutics, Gritstone Oncology
PRINCIPLES OF IMMUNOTHERAPY Antigen presenting Can you generate cytotoxic T-cells? T-cell trafficking Can the T-cells get to the tumour? Peptide-MHC recognition Can the T-cells see the tumour? PD-L1 on tumour/inhibitory cytokines Can the T-cells be deactivated? Chen et al. Clin Cancer Res 2012;18:6580-6587
BLOCKADE OF PD-1 OR CTLA-4 SIGNALING IN IMMUNOTHERAPY Anti-CTLA4 antibodies (ipilimumab, tremilimumab) block a negative regulatory signal during T-cell priming. Anti-PD-1 antibodies (pembrolizumab, nivolumab) block the negative regulatory signal of PD-1 which is expressed on T-cells during long term antigen exposure. Ribas A. N engl J med 2012;366:2517-2519.
IMMUNE CHECKPOINT BLOCKADE IN GASTROESOPHAGEAL CANCER A mixed bag of results
IMMUNE CHECKPOINT BLOCKADE IN GASTROESOPHAGEAL CANCER Take home message 1 Anti-PD-1 therarapy is superior to best supportive care in chemorefractory gastroesphageal cancer patients Supporting data: ATTRACTION 2
NIVOLUMAB IN CHEMOREFRACTORY GASTRIC CANCER ATTRACTION-02 ATTRACTION-02 Patient Characteristics ECOG 0 vs. 1 29% vs. 71% Site of disease Gastric vs. other 82% vs. 18% Prior regimens 2 vs. 3 vs. 4 20% vs. 40% vs. 40% Bang et al, Lancet 2017
ATTRACTION-02 Response rates and duration ORR 12% RECIST response rates are modest Responses seen in PD-L1 positive and negative patients Median time to response was 1.6m (1.4-7.0m) Responses also seen as late at 7 months Boku et al, ESMO 2017
ATTRACTION-02 Updated overall survival results Boku et al, ESMO 2017
ATTRACTION-02 Overall survival by PD-L1 status 14% tested population were PD-L1 positive PD-L1 antibody 28-8 pharmdx assay PD-L1 positivity was defined as staining in 1% or more of tumour cells. Ka ng et al, Lancet 207,
PEMBROLIZUMAB IN CHEMOREFRACTORY GC KEYNOTE-059 Pts with recurrent or metastatic gastric or GEJ adenocarcinoma; ECOG PS 0/1; HER2/neu negative*; no prior PD-1/PD-L1 tx,(n = 259) Cohort 1 2 prior lines chemotherapy Cohort 2 Treatment naive Cohort 3 Treatment naïve PD-L1+ Pembrolizumab 200 mg Q3W Pembrolizumab 200 mg Q3W + Cisplatin 80 mg/m 2 Q3W + 5-FU 800 mg/m 2 Q3W or Capecitabine 1000 mg/m 2 BID Q3W Pembrolizumab 200 mg Q3W Treatment until PD, 24m or, intolerable toxicity, or withdrawal of consent. *HER2/neu positive allowed in cohort 1 if prior trastuzumab administered. Primary endpoints: ORR, safety; secondary endpoints: DoR, PFS, OS Fuchs CS et al, JAMA Oncology 2018
KEYNOTE-059 Study design Pts with recurrent or metastatic gastric or GEJ; ECOG PS 0/1; no prior PD-1/PD-L1 Cohort 1 2 prior lines chemotherapy Pembrolizumab 200 mg Q3W Treatment until PD, 24m, intolerable toxicity, or withdrawal of consent. Keynote -059 patient characteristics (N = 259) Median age, yrs (range) 62 (24-89) Geographic region, n (%) US vs. East Asia vs. Other 48% vs. 13% vs. 39% ECOG PS 0 vs. 1 41% vs. 58% Tumour site (%) Gastric vs. GOJ 48% vs. 51% Prior therapies 2 vs. 3 vs. 4 52% vs. 29% vs. 19% Fuchs CS et al, JAMA Oncology 2018
PEMBROLIZUMAB IN CHEMOREFRACTORY GC KEYNOTE-059 Change From BL (%) 120 100 80 60 40 20 0-20 -40-60 -80-100 PD-L1 positive PD-L1 negative PD-L1 expression unknown ORR 11.6% 9% in MSS *Included pts with measurable disease at BL and 1 post-bl assessment (n = 223). RECIST response rates are modest (identical to nivolumab in ATTRACTION-02) Responses in PDL1 positive and negative patients Treatment Exposure and Duration of Response Majority of responses are early Median duration of response: All patients 8.4m PD-L1 positive 16.3m PD-L1 negative 6.9m Confirmed Responders (n = 30) CR PR PD Death Ongoing pembrolizumab 0 2 4 6 8 10 12 14 16 18 20 22 24 Fuchs CS et al, JAMA Oncology 2018
KEYNOTE-059 ORR according to PD-L1 status and line of Tx PD-L1 status Line of Treatment PD-L1 and 3 rd Line Positive (n = 148) Negative (n = 109) 3rd (n = 134) 4th (n = 125) Positive (n = 75) Negative (n = 58) ORR (%) 15.5 (10.1-22.4) 6.4 (2.6-12.8) 16.4 (10.6-23.8) 6.4 (2.8-12.2) 22.7 (13.8-33.8) 8.6 (2.9-19.0) Response rates PD-L1 positive vs. PD-L1 negative (15.5% vs 6.4% ) Combination PD-L1 positive and earlier line of treatment (3 rd ) ORR 23% PD-L1 assay is 22C3 antibody using CPS score. CPS score = (number positive cells (IC, tumour)/tumour cells) x 100 PD-L1 positive if 1%
CONTROVERSY Does ATTRACTION 2 only support use of nivolumab in Asian patients?
NIVOLUMAB AND PEMBROLIZUMAB IN CHEMOREFRACTORY GASTROESOPHAGEAL CANCER Median OS (m) mos PD-L1 positive mos PD-L1 negative ORR 12% 5.26 5.22 6.05 Change From BL (%) 120 100 80 60 40 20 0-20 -40-60 -80-100 PD-L1 positive PD-L1 negative PD-L1 expression unknown ORR 11.6% 15.5% vs 6.4% PD-L1+ vs - CHECKMATE 032 ATTRACTION-02 KEYNOTE 059 cohort 1 nivolumab arm ORR 12% ORR 19% vs 12% PD-L1+ vs. - Median OS (m) mos PD-L1 positive mos PD-L1 negative 5.6 5.8 4.6 Median OS (m) 6.2
IMMUNE CHECKPOINT BLOCKADE IN GASTROESOPHAGEAL CANCER Take home messages Anti-CTLA4 therapy is not better than BSC as maintenance in 1L GC Supporting data: Bang et al, CCR 2017 Anti-PD-L1 therapy is not better than chemotherapy in chemorefractory GC patients Supporting data: JAVELIN 300 Anti-PD-1 therapy is not better than paclitaxel in PD-L1 CPS1 2L GC patients Supporting data: KEYNOTE 061
NEGATIVE TRIALS OF IMMUNE CHECKPOINT BLOCKADE IN GC Ipilimumab All best supportive care Pembrolizumab Paclitaxel Avelumab Chemotherapy Bang YJ, et al. Clin Cancer Res 2017;23:5671 5678; Shitara K, et al. Lancet 2018;392:123 133; Bang YJ, et al. Ann Oncol 2018 (in press)
KEYNOTE 061 Pembrolizumab vs paclitaxel in 2L GC patients Shitara et al, Lancet 2018
KEYNOTE 061 Pembrolizumab vs paclitaxel in 2L GC patients all patients Shitara et al, Lancet 2018
KEYNOTE 061 Pembrolizumab vs paclitaxel in 2L GC patients CPS 1 Shitara et al, Lancet 2018
KEYNOTE 061 Progression free survival in CPS 1 Shitara et al, Lancet 2018
KEYNOTE 061 OS in CPS 1 population Shitara et al, Lancet 2018
KEYNOTE 061 OS in different populations Shitara et al, Lancet 2018
KEYNOTE 061 ORR and duration in CPS 1 Shitara et al, Lancet 2018
KEYNOTE 061 TOXICITY
PEMBROLIZUMAB IN TREATMENT NAÏVE GC Pts with recurrent or metastatic gastric or GEJ adenocarcinoma; ECOG PS 0/1; Cohort 3 Treatment naïve PD-L1+ Pembrolizumab 200 mg Q3W Treatment until PD, 24m, intolerable toxicity, or withdrawal of consent. ORR 26%
IMMUNE CHECKPOINT BLOCKADE IN MGC The future? Zamarin et al, Pharmacology & Therapeutics 2015
COMBINATIONS: ANTI-CTLA4 + ANTI-PD-1 CHECKMATE-032 CHECKMATE-032 Patient Characteristics Site of disease Gastric vs. GOJ 37% vs. 63% Prior regimens 1 vs. 2 vs. 3 vs. 4 20% vs. 34% vs. 27% vs. 18% Janjigian et al, J Clin Onc 2018.
CHECKMATE-032 Radiological responses by treatment arm ORR 12% ORR 24% ORR 8% PD-L1 positive ORR 19% PD-L1 negative ORR 12% PD-L1 positive ORR 40% PD-L1 negative ORR 22% PD-L1 positive ORR 23% PD-L1 negative ORR 0% Janjigian et al, J Clin Onc 2018.
CHECKMATE-032 Radiological responses by treatment arm Progression free survival Overall survival NIVO1 + IPI3 NIVO1 + IPI3 Janjigian et al, J Clin Onc 2018.
COMBINATION: CHEMOTHERAPY PLUS ANTI-PD-1 KEYNOTE-059 Pts with recurrent or metastatic gastric or GEJ adenocarcinoma; ECOG PS 0/1; HER2/neu negative*; no prior PD-1/PD-L1 tx) Cohort 2 Treatment naive Pembrolizumab 200 mg Q3W + Cisplatin 80 mg/m 2 Q3W + 5-FU 800 mg/m 2 Q3W or Capecitabine 1000 mg/m 2 BID Q3W Treatment until PD, 24m, intolerable toxicity, or withdrawal of consent. All patients PD-L1 positive PD-L1 negative ORR 60% 69% 38% Wainberg et al, ESMO 2017.
COMBINATION: ANTI-ANGIOGENESIS + ANTI-PD-1 JVDF trial Chau et al, ASCO GI 2018
BIOMARKERS FOR ICB IN GC MSI KEYNOTE 061 PD-L1 CPS > 10 KEYNOTE 061 T-cell inflamed signature KEYNOTE059 Shitara et al, Lancet, June 4 2018 Fuchs et al, JAMA Oncol May 10 2018
MSI-H PD-L1 1>20 TCI GEP high Non-responder GC BIOMARKERS FOR IMMUNE CHECKPOINT BLOCKADE MSI, PD-L1 AND GENE EXPRESSION Fuchs et al, JAMA Oncol May 10 2018
IMMUNE ENVIRONMENT IN A HETEROGENEOUS DISEASE Subtype characteristics Immune characteristics Copy number changes: Low immune score Low IFNγ signature Rare in metastatic patients PD-L1: tumour ++ TILs +++ High IFNγ signature Rare in metastatic patients PD-L1: tumour ++ TILs ++ High IFNγ signature Genomically bland Diffuse type rarely PD-L1+ Subsets may have TIL Oesophageal TCGA Nature 541, 169 175 (12 January 2017) Derks et al, Oncotarget. 2016 May 31;7(22):32925-32 Böger et al, Oncotarget. 2016 Apr 26; 7(17): Ock et al, Nat Commun. 2017; 8: 1050. Sohn et al, Clin Cancer Res. 2017 Jul 26.
CIN GASTROESOPHAGEAL CANCER IMMUNE ENVIRONMENT CIN-GC characterised by high somatic copy number alterations (SCNA) Copy number alterations associated with low immune gene expression CNA associated with benefit from IO in melanoma Gastric Cancer TCGA, Nature 2014. Davoli, Science. 2017 Jan 20;355(6322).
CIN GASTROESOPHAGEAL CANCER IMMUNE ENVIRONMENT In TGCA dataset CIN tumours have lowest INFγ signature GC-CIN tumours have the lowest proportion of T-cell inflamed tumours Specific mutations associated with inflamed subtype: PIK3CA, ATM, RHOA and CDH1 Amplifications associated with immune ignorance: ERBB2, MYC, VEGFA Maron et al, ASCO SITC 2017
CIN GASTROESOPHAGEAL CANCER IMMUNE ENVIRONMENT 3 oesophageal adenocarcinoma genomic signatures identified using whole exome NGS Mutagenic subtype associated with neoantigen and immune infiltrates Subtype characteristics Oesophageal TCGA Nature 541, 169 175 (12 January 2017) Secrier et al, Nature Genetics 48, 1131 1141 (2016)
IMMUNOTHERAPY FOR GASTROESOPHAGEAL CANCER CONCLUSIONS Anti-PD-1 therapy is a new standard for patients with chemorefractory gastroesophageal cancer Single agent activity is modest MSI-H, rare but sensitive to IO. All patients should be tested. PD-L1 enriches for response Combinations (chemotherapy, immunotherapy) increase efficacy Subtype biology and interaction with immune system key to designing effective personalised immunotherapy combinations.