Present state of the art of immune checkpoint inhibitors in oncology. John Haanen, MD PhD

Transcription

1 Present state of the art of immune checkpoint inhibitors in oncology John Haanen, MD PhD ESMO Advanced course in Immuno-oncology, Amsterdam Febr 2018

2 Disclosures I have provided consultation, attended advisory boards, and/or provided lectures for: Pfizer, MSD, BMS, IPSEN, Novartis, Roche/Genentech, NEON Therapeutics and Gadeta, for which NKI received honoraria Through my work NKI received grant support from BMS, MSD, Novartis, and NEON Therapeutics I declare no conflict of interest

3 Immune Checkpoint inhibitors Immune checkpoints play an important role in immune tolerance Cancer hijacks many of these peripheral tolerance mechanism to escape the immune system Inhibition of a single immune checkpoint can be enough to break this cancer induced tolerance (anti-ctla4, anti-pd-1/pd-l1) Combination of some of these inhibitors appear more powerful

4 Value of enhancing endogenous tumor-specific T cell responses: Checkpoint blockade Adjusted from Drake et al., Nat Rev Clin Oncol 2014

5 Ipilimumab for treatment of metastatic melanoma Pre-treated-pts +/- gp100 HLA-A2 3mg/kg Re-induction possible naive-pts + DTIC 10 mg/kg Maintenance possible 1 Year 2 Year Ipi + gp100 N=403 44% 22% Ipi + pbo N=137 46% 24% gp100 + pbo N=136 25% 14% Hodi et al 2010 NEJM Ipilimumab+ DTIC N=250 Placebo+ DTIC N=252 1 Year 2 Year 3 Year Robert et al NEJM 2011

6 Updated OS results from CheckMate 066 trial in BRAF wt advanced melanoma Decrease of the risk of death 58% vs chemotherapy Atkinson et al. abstract 3774 SMR 2015

7 Pembrolizumab vs ipilimumab as 1 st line treatment of metastatic melanoma Treatment Arm Median (95% CI), mo Rate at 12 mo HR (95% CI) P Pembrolizumab Q2W NR (NR-NR) 74.1% 0.63 ( ) < Pembrolizumab Q3W NR (NR-NR) 68.4% 0.69 ( ) < Ipilimumab NR (12.7-NR) 58.2% Decrease of risk of death of pembrolizumab 31 to 37% vs ipilimumab Robert et al NEJM 2015

8 Change From Baseline in Tumor Size, % Anti-PD1 Demonstrates Broad Antitumor Activity Melanoma 1 (N=655) KEYNOTE NSCLC 2 (N=262) KEYNOTE H&N 3 (N=132) KEYNOTE Urothelial 4 (N=33) KEYNOTE Gastric 5 (N=39) KEYNOTE TNBC 6 (N=32) KEYNOTE chl 7 (N=29) KEYNOTE Mesothelioma 8 (N=25) KEYNOTE Ovarian 9 (N=26) KEYNOTE SCLC 10 (N=20) KEYNOTE Esophageal 11 (N=23) KEYNOTE Courtesy of G Long 1. Daud A et al ASCO; 2. Garon EB et al. ESMO 2014; 3. Seiwert T et al ASCO; 4. Plimack E et al ASCO; 5. Bang YJ et al ASCO; 6. Nanda R et al. SABCS 2014; 7. Moskowitz C et al ASH Annual Meeting; 8. Alley EA et al AACR; 9. Varga A et al ASCO; 10. Ott PA et al ASCO; 11. Doi T et al ASCO.

9 FDA approvals for immune checkpoint blockers, diagnostic tests, and treatment combinations Melanoma Ipilimumab monotherapy Pembrolizumab 2 nd -line Nivolumab 2 nd -line Ipi adjuvant Nivo/ipi combo Nivo 1st-line Pembro 1st-line Nivo mel biomarker Pembro MSI-hi tumors 2 nd line with biomarker Nivo MSI-hi CRC 2 nd line with biomarker Pembro bladder 2 nd line Pembro/chemo NSCLC combo 2 nd line Avelumab bladder 2 nd line Non-Melanoma Nivo lung cancer squamous 2 nd line Pembro lung 2 nd line with biomarker Nivo lung cancer nonsquamous 2 nd line with biomarker Nivo kidney 2 nd line Nivo Hodgkin 4 th line Pembro head/neck 2 nd line Atezo lung cancer 2 nd line Nivo head/neck 2 nd line Durvalumab bladder 2 nd line with biomarker Avelumab Merkel cell Ca 1 st line Courtesy of S. Topalian (JAMA 2017) Atezo bladder cancer 2 nd line with biomarker Pembro lung cancer 1 st line with biomarker Nivo bladder 2 nd line Pembro Hodgkin 4 th line

10 Melanoma has become from a disease that gave cancer a bad name to a model disease for I-O Current I-O treatment options for melanoma Stage III disease Neo-adjuvant/adjuvant trials Stage IV disease combination therapy

11

12 EORTC 18071/CA : Study Design Randomized, double-blind, phase 3 study evaluating the efficacy and safety of ipilimumab in the adjuvant setting for high-risk melanoma High-risk, stage III, completely resected melanoma R N = 475 INDUCTION Ipilimumab 10 mg/kg Q3W 4 MAINTENANCE Ipilimumab 10 mg/kg Q12W up to 3 years N = 951 N = 476 INDUCTION Placebo Q3W 4 MAINTENANCE Placebo Q12W up to 3 years Week 1 Week 12 Week 24 Stratification factors Stage (IIIA vs IIIB vs IIIC 1-3 positive lymph nodes vs IIIC 4 positive lymph nodes) Regions (North America, European countries, and Australia) Enrollment Period: June 2008 to July 2011 Treatment up to a maximum of 3 years, or until disease progression, intolerable toxicity, or withdrawal Q3W = every 3 weeks; Q12W = every 12 weeks; R = randomization.

13 Patients Alive and Without Recurrence (%) RFS (per IRC) Ipilimumab Placebo Events/patients 264/ /476 HR (95% CI) a 0.76 (0.64, 0.89) Log-rank P value a Median RFS, months (95% CI) 41% 30% 27.6 (19.3, 37.2) 17.1 (13.6, 21.6) a Stratified by stage provided at randomization. CI = confidence interval O N Number of patients at risk Years Ipilimumab Placebo

14 Patients Alive (%) OS Ipilimumab Placebo Deaths/patients 162/ /476 HR (95.1% CI) a 0.72 (0.58, 0.88) Log-rank P value a % a Stratified by stage provided at randomization % O N Number of patients at risk Years Ipilimumab Placebo

15 Patient Disposition and Treatment Ipilimumab (n = 471) Placebo (n = 474) Discontinuation, % Reasons for discontinuation, % Normal completion (received study drug for entire 3 years) Disease recurrence AE related to study drug Other reasons a Median doses, per patient, n Receiving 1 maintenance dose, % Receiving 7 doses (1 year of therapy), % a Includes AE unrelated to study drug, both related and unrelated to study drug, patient request, poor/noncompliance, death, pregnancy, patient no longer eligible, other.

16 Phase III trials in the adjuvant setting for stage III and IV disease A Phase 3, Randomized, Double-blind Study of Adjuvant Immunotherapy With Nivolumab Versus Ipilimumab After Complete Resection of Stage IIIb/c or Stage IV Melanoma in Subjects Who Are at High Risk for Recurrence (CheckMate-238) Adjuvant Immunotherapy With Anti-PD-1 Monoclonal Antibody Pembrolizumab Versus Placebo After Complete Resection of High-risk Stage III Melanoma: A Randomized, Double- Blind Phase 3 Trial of the EORTC Melanoma Group (KEYNOTE-054) A Phase III Randomized Trial Comparing Physician/Patient Choice of Either High Dose Interferon or Ipilimumab to Pembrolizumab in Patients With High Risk Resected Melanoma

17

18 CA : CA : Study Design Patients with high-risk, completely resected stage IIIB/IIIC or stage IV melanoma Stratified by: 1:1 n = 453 n = 453 1) Disease stage: IIIB/C vs IV M1a-M1b vs IV M1c 2) PD-L1 status at a 5% cutoff in tumor cells NIVO 3 mg/kg IV Q2W and IPI placebo IV Q3W for 4 doses then Q12W from week 24 IPI 10 mg/kg IV Q3W for 4 doses then Q12W from week 24 and NIVO placebo IV Q2W Follow-up Maximum treatment duration of 1 year Enrollment period: March 30, 2015 to November 30, 2015

19 Study Overview Primary endpoint RFS: time from randomization until first recurrence (local, regional, or distant metastasis), new primary melanoma, or death Secondary endpoints OS Safety and tolerability RFS by PD-L1 tumor expression HRQoL Current interim analysis Primary endpoint (RFS), safety, and HRQoL DMFS (exploratory) Duration of follow-up: minimum 18 months; 360 events DMFS = distant metastasis-free survival; HRQoL = health-related quality of life

20 Baseline Patient Characteristics NIVO (n = 453) IPI (n = 453) Median age, years Male, % Stage, IIIB+IIIC, % Macroscopic lymph node involvement (% of stage IIIB+IIIC) Ulceration (% of stage IIIB+IIIC) Stage IV, % M1c without brain metastases (% stage IV) PD-L1 expression 5%, % BRAF mutation, % LDH ULN, % Most of the patients had cutaneous melanoma (85%), and 4% had acral and 3% had mucosal melanoma All 905 patients are off treatment; median doses were 24 (1-26) in the NIVO group and 4 (1-7) in the IPI group 397 patients completed 1 year of treatment (61% of the NIVO group and 27% of the IPI group)

21 RFS (%) Primary Endpoint: RFS NIVO IPI 100 Events/patients 154/ / Median (95% CI) NR NR (16.6, NR) HR (97.56% CI) 0.65 (0.51, 0.83) Log-rank P value < % 61% 66% 53% NIVO IPI Months Number of patients at risk NIVO IPI

22 RFS (%) RFS (%) Subgroup Analysis of RFS: PD-L1 Expression Level PD-L1 Expression Level <5% PD-L1 Expression Level 5% NIVO IPI NIVO IPI Events/patients 114/ /286 Events/patients 31/152 57/154 Median (95% CI) NR 15.9 (10.4, NR) Median (95% CI) NR NR 100 HR (95% CI) 0.71 (0.56, 0.91) 100 HR (95% CI) 0.50 (0.32, 0.78) % % % % NIVO IPI NIVO IPI Number of patients at risk Months Number of patients at risk Months NIVO NIVO IPI IPI

23 RFS: Prespecified Subgroups Subgroup No. of events/no. of patients NIVO 3 mg/kg IPI 10 mg/kg Unstratified HR (95% CI) Overall Overall 154/ / (0.53, 0.81) Age <65 years 106/ / (0.51, 0.84) 65 years 48/120 59/ (0.45, 0.97) Sex Male 99/ / (0.53, 0.88) Female 55/195 73/ (0.44, 0.89) Stage (CRF) Stage IIIb 41/163 54/ (0.44, 1.00) Stage IIIc 79/ / (0.49, 0.87) Stage IV M1a-M1b 25/62 35/ (0.38, 1.05) Stage IV M1c 8/20 8/ (0.37, 2.66) Not reported 1/2 0/0 Stage III: Ulceration Absent 58/201 94/ (0.42, 0.82) Stage III: Lymph node involvement Present 60/153 64/ (0.51, 1.04) Not reported 2/15 5/ (0.07, 2.00) Microscopic 41/125 55/ (0.47, 1.07) Macroscopic 72/ / (0.46, 0.84) Not reported 7/25 7/ (0.21, 1.72) PD-L1 status <5%/indeterminate 123/ / (0.56, 0.90) 5% 31/152 57/ (0.32, 0.78) BRAF mutation status Mutant 63/187 84/ (0.52, 1.00) Wild-type 67/ / (0.43, 0.79) Not reported 24/69 17/ (0.45, 1.54) Unstratified HR (95% CI) NIVO IPI

24 DMFS (%) Exploratory Endpoint: DMFS for Stage III Patients NIVO IPI % Events/patients 93/ /366 Median (95% CI) NR NR HR (95% CI) 0.73 (0.55, 0.95) Log-rank P value % Number of patients at risk NIVO IPI NIVO IPI Months

25 Treatment-Related Select Adverse Events NIVO (n = 452) IPI (n = 453) AE, n (%) Any grade Grade 3/4 Any grade Grade 3/4 Skin 201 (44.5) 5 (1.1) 271 (59.8) 27 (6.0) Gastrointestinal 114 (25.2) 9 (2.0) 219 (48.3) 76 (16.8) Hepatic 41 (9.1) 8 (1.8) 96 (21.2) 49 (10.8) Pulmonary 6 (1.3) 0 11 (2.4) 4 (0.9) Renal 6 (1.3) 0 7 (1.5) 0 Hypersensitivity/infusion reaction 11 (2.4) 1 (0.2) 9 (2.0) 0 Endocrine Adrenal disorder 6 (1.3) 2 (0.4) 13 (2.9) 4 (0.9) Diabetes 2 (0.4) 1 (0.2) 1 (0.2) 0 Pituitary disorder 8 (1.8) 2 (0.4) 56 (12.4) 13 (2.9) Thyroid disorder 92 (20.4) 3 (0.7) 57 (12.6) 4 (0.9) Median time to onset of treatment-related select AEs was generally shorter for patients receiving IPI (range weeks) than for those receiving NIVO (range weeks)

26 Conclusions Nivolumab showed a clinically and statistically significant improvement in RFS vs the active control of high-dose ipilimumab for patients with resected stages IIIB/IIIC and stage IV melanoma at high risk of recurrence (HR = 0.65, P < ) 18-month RFS rates were 66% for nivolumab and 53% for ipilimumab Benefit for nivolumab was observed across the majority of prespecified subgroups tested, including PD-L1 and BRAF mutation status Nivolumab has a superior safety profile in comparison with ipilimumab, with fewer grade 3/4 AEs and fewer AEs leading to treatment discontinuation Nivolumab has the potential to be a new standard treatment option for patients with resected stage IIIB, IIIC, and IV melanoma regardless of BRAF mutation

27 New developments in adjuvant and neoadjuvant trials An Open-label, Phase IB Study of NEO-PV-01 + Adjuvant With Nivolumab in Patients With Melanoma, Non-Small Cell Lung Carcinoma or Transitional Cell Carcinoma of the Bladder Phase II Study to Identify the Optimal neoadjuvant Combination Scheme of Ipilimumab and Nivolumab in Stage III Melanoma Patients (OPACIN-neo) A Phase II, Randomised, Open Label Study of Neoadjuvant Dabrafenib, Trametinib and / or Pembrolizumab in BRAF V600 Mutant Resectable Stage IIIB/C Melanoma A Phase 1b Trial of Neoadjuvant CXCR4 antagonist (X4P- 001) Alone and With Pembrolizumab in Patients With Resectable Melanoma

28 Antonia et al., NEJM 2017

29 PACIFIC: Study Design Phase III, Randomized, Double-blind, Placebocontrolled, Multicenter, International Study Patients with stage III, locally advanced, unresectable NSCLC who have not progressed following definitive platinum-based ccrt ( 2 cycles) Durvalumab 10 mg/kg q2w for up to 12 months N=476 Co-primary endpoints PFS by BICR using RECIST v1.1* OS 18 years or older WHO PS score 0 or 1 Estimated life expectancy of 12 weeks Archived tissue was collected All-comers population 1 42 days post-ccrt R 2:1 randomization, stratified by age, sex, and smoking history N=713 Placebo 10 mg/kg q2w for up to 12 months N=237 Key secondary endpoints ORR (per BICR) DoR (per BICR) Safety and tolerability PROs *Defined as the time from randomization (which occurred up to 6 weeks post-ccrt) to the first documented event of tumor progression or death in the absence of progression. ClinicalTrials.gov number: NCT BICR, blinded independent central review; ccrt, concurrent chemoradiation therapy; DoR, duration of response; NSCLC, non-small cell lung cancer; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; PROs, patient-reported outcomes; PS, performance status; q2w, every 2 weeks; RECIST, Response Evaluation Criteria in Solid Tumors; WHO, World Health Organization Antonia et al., NEJM 2017

30 Baseline Characteristics (ITT) Durvalumab (N=476) Placebo (N=237) Age Median (range), years 65 years, % 64 (31 84) (23 90) 45.1 Male, % WHO performance status score, %* 0 / / / 51.5 Smoking status, % Current / Former / Never 16.6 / 74.4 / / 75.1 / 8.9 Disease stage, % IIIA / IIIB 52.9 / / 45.1 Histology, % Squamous / Non-squamous 47.1 / / 57.0 PD-L1 status, % Known: TC <25% / TC 25% Unknown 39.3 / / Prior chemotherapy, % Induction / Definitive ccrt 25.8 / / 99.6 Prior radiotherapy, %* <54 Gy 54 to 66 Gy >66 to 74 Gy Best response to prior ccrt, % CR / PR / SD / PD 1.9 / 48.7 / 46.6 / / 46.8 / 48.1 / 0 *Not reported or missing (durvalumab, placebo, total): WHO performance status (0.4% each), prior radiotherapy (0.2%, 0.4%, 0.3%). Other: durvalumab, 2.5%; placebo, 2.1%; total, 2.4%. No sample collected or no valid test result. Not evaluable/not applicable: durvalumab, 2.3%; placebo, 2.1%; total, 2.2%. ccrt, concurrent chemoradiation therapy; CR, complete response; ITT, intention-to-treat; PD, progressive disease; PD-L1, programmed cell death ligand-1; PR, partial response; SD, stable disease; TC, tumor cell; TC 25%, 25% PD-L1 expression on tumor cells; TC <25%, <25% PD-L1 expression on tumor cells; WHO, World Health Organization Antonia et al., NEJM 2017

31 Patient Disposition Durvalumab (N=476) Placebo (N=237) Received treatment, n (%) 473 (99.4) 236 (99.6) Completed 12 months of treatment, n (%) 202 (42.7) 71 (30.1) Discontinued study treatment, n (%) Patient decision Adverse event Severe non-compliance to protocol Disease worsening Development of study specific discontinuation criterion Other 241 (51.0) 14 (3.0) 73 (15.4) 1 (0.2) 148 (31.3) 1 (0.2) 4 (0.8) 153 (64.8) 12 (5.1) 23 (9.7) 1 (0.4) 116 (49.2) 1 (0.4) 0 Received subsequent therapy after discontinuation, n (%) 145 (30.5) 102 (43.0) Progressed by BICR as of data cutoff for interim PFS analysis, n Median follow-up was 14.5 months (range ) Antonia et al., NEJM 2017

32 PFS probability PFS by BICR (Primary Endpoint; ITT) Stratified hazard ratio, 0.52 (95% CI, ) Two-sided P< Durvalumab (N=476) Placebo (N=237) Median PFS (95% CI), months 16.8 ( ) 5.6 ( ) 12-month PFS rate (95% CI) 55.9% ( ) 35.3% ( ) 18-month PFS rate (95% CI) 44.2% ( ) 27.0% ( ) Placebo Durvalumab No. at risk Durvalumab Placebo Time from randomization (months) Antonia et al., NEJM 2017

33 PFS Subgroup Analysis by BICR (ITT) Sex Age at randomization Smoking status Disease stage Histology Best response to ccrt PD-L1 status EGFR status Durvalumab Placebo Unstratified HR* No. of patients (95% CI) All patients ( ) Male ( ) Female ( ) <65 years ( ) 65 years ( ) Smoker ( ) Non-smoker ( ) Stage IIIA ( ) Stage IIIB ( ) Squamous ( ) Non-squamous ( ) CR 9 7 PR ( ) SD ( ) 25% ( ) <25% ( ) Unknown ( ) Mutant ( ) Wild-type ( ) Unknown ( ) Favors durvalumab Favors placebo *Hazard ratio and 95% CI not calculated if the subgroup has less than 20 events. BICR, blinded independent central review; CI, confidence interval; CR, complete response; HR, hazard ratio; ITT, intention-to-treat; EGFR, epidermal growth factor receptor Antonia et al., NEJM 2017

34 Probability of death or distant metastasis Time to Distant Metastasis or Death by BICR (ITT) Stratified hazard ratio, 0.52 (95% CI, ) Two-sided P< No. at risk Durvalumab Placebo Median time (95% CI), months Durvalumab Placebo Time from randomization (months) Durvalumab 23.2 (23.2 NR) Placebo 14.6 ( ) Antonia et al., NEJM 2017

35 Pneumonitis or Radiation Pneumonitis Pneumonitis (grouped terms) or radiation pneumonitis, n (%)* Durvalumab (N=475) Placebo (N=234) Any grade 161 (33.9) 58 (24.8) Grade 3/4 16 (3.4) 6 (2.6) Grade 5 5 (1.1) 4 (1.7) Leading to discontinuation 30 (6.3) 10 (4.3) Safety analysis set (all-causality). *Pneumonitis/radiation pneumonitis was assessed by investigators with subsequent review and adjudication by the study sponsor. In addition, pneumonitis, as reported in the table, is a grouped term, which includes acute interstitial pneumonitis, interstitial lung disease, pneumonitis, and pulmonary fibrosis. Two patients randomized to placebo received at least one dose of durvalumab and were considered part of the durvalumab arm for safety reporting. Antonia et al., NEJM 2017

36 Summary Durvalumab demonstrated a statistically significant and robust improvement in PFS versus placebo (HR 0.52; P<0.0001; median improvement of >11 months) at a planned interim analysis PFS improvement with durvalumab was observed across all prespecified subgroups Durvalumab demonstrated a clinically meaningful benefit in ORR (28.4% vs 16.0%; P<0.001), with durable responses versus placebo (median DoR not reached vs 13.8 months) Patients receiving durvalumab had a lower incidence of new lesions, including new brain metastases, compared with patients receiving placebo The safety profile of durvalumab was consistent with that of other immunotherapies and with its known safety profile as monotherapy in patients with more advanced disease; 1 no new safety signals were identified The study remains blinded to OS Antonia et al., NEJM 2017

37 Melanoma has become from a disease that gave cancer a bad name to a model disease for I-O Current I-O treatment options for melanoma Stage III disease Neo-adjuvant/adjuvant trials Stage IV disease combination therapy

38 Why combining (immuno)therapies?

39 Cancer Immunogram Tumor foreignness Mutational load Tumor sensitivity to immune effectors MHC expression IFN-g sensitivity General immune status Lymphocyte count Absence of inhibitory tumor metabolism LDH, glucose utilization Immune cell infiltration Intratumoral T cells Absence of soluble inhibitors IL6->CRP/ESR Absence of Checkpoints PD-L1 Blank, Haanen et al. Science 2016

40 Chen & Mellman Nature 2017 Cancer Immunophenotypes

41 Combinations with immunotherapy Melero, Haanen Nat Rev Canc 2015

42 Most combinations have anti-pd1/pdl1 as backbone Melero, Haanen Nat Rev Canc 2015

43 Combinations Anti-CTLA4 + anti-pd1/pdl1 IDO1/2 inhibitor + anti-pd1 Oncolytic virus + anti-pd1 Anti-LAG3 + anti-pd1 Combination with angiogenesis inhibitors IMpower 150 (NSCLC) IMmotion 151(RCC)

44 PD1 blockade synergizes with actla-4 Curran et al., PNAS 2010

45 Immunotherapy melanoma CTLA4 + PD-1 blockade (ipilimumab + nivolumab) Wolchok et al. ASCO 2013 # 9012

46 Overall Survival Results From a Phase III Trial of Nivolumab Combined With Ipilimumab in Treatment-naïve Patients With Advanced Melanoma (CheckMate 067) James Larkin, 1 Vanna Chiarion-Sileni, 2 Rene Gonzalez, 3 Piotr Rutkowski, 4 Jean-Jacques Grob, 5 C. Lance Cowey, 6 Christopher D. Lao, 7 Dirk Schadendorf, 8 Pier Francesco Ferrucci, 9 Michael Smylie, 10 Reinhard Dummer, 11 Andrew Hill, 12 John Haanen, 13 Michele Maio, 14 Grant McArthur, 15 Dana Walker, 16 Linda Rollin, 16 Christine Horak, 16 F. Stephen Hodi, 17,* Jedd D. Wolchok 18,* 1 Royal Marsden Hospital, London, UK; 2 Oncology Institute of Veneto IRCCS, Padua, Italy; 3 University of Colorado Cancer Center, Denver, CO, USA; 4 Maria Sklodowska-Curie Memorial Cancer Center & Institute of Oncology, Warsaw, Poland; 5 Hospital de la Timone, Marseille, France; 6 Texas Oncology-Baylor Charles A. Sammons Cancer Center, Dallas, TX, USA; 7 University of Michigan, Ann Arbor, MI, USA; 8 Department of Dermatology, University of Essen, Essen, Germany; 9 European Institute of Oncology, Milan, Italy; 10 Cross Cancer Institute, Alberta, Canada; 11 Universitäts Spital, Zurich, Switzerland; 12 Tasman Oncology Research, QLD, Australia; 13 Netherlands Cancer Institute, Amsterdam, The Netherlands; 14 University Hospital of Siena, Siena, Italy; 15 Peter MacCallum Cancer Centre, Victoria, Australia; 16 Bristol-Myers Squibb, Princeton, NJ, USA; 17 Dana-Farber Cancer Institute, Boston, MA, USA; 18 Memorial Sloan-Kettering Cancer Center and Weill Cornell Medical College, New York, NY, USA; *Contributed equally to this study. Presented at AACR 2017 by Larkin

47 CheckMate CA : 067: Study Study Design Design Randomized, double-blind, phase III study to compare NIVO+IPI or NIVO alone to IPI alone* N=314 NIVO 1 mg/kg + IPI 3 mg/kg Q3W for 4 doses then NIVO 3 mg/kg Q2W Unresectable or Metatastic Melanoma Previously untreated 945 patients Randomize 1:1:1 Stratify by: BRAF status AJCC M stage Tumor PD-L1 expression <5% vs 5%* N=316 NIVO 3 mg/kg Q2W + IPI-matched placebo Treat until progression or unacceptable toxicity N=315 IPI 3 mg/kg Q3W for 4 doses + NIVO-matched placebo Database lock: Sept 13, 2016 (median follow-up ~30 months in both NIVO-containing arms) *The study was not powered for a comparison between NIVO and NIVO+IPI Presented at AACR 2017 by Larkin

48 Response To Treatment NIVO+IPI (N=314) NIVO (N=316) IPI (N=315) ORR, % (95% CI)* 58.9 ( ) 44.6 ( ) 19.0 ( ) Best overall response % Complete response Partial response Stable disease Progressive disease Unknown Median duration of response, months (95% CI) NR (NR NR) 31.1 (31.1 NR) 18.2 (8.3 NR) *By RECIST v1.1; NR = not reached. At the 18-month DBL, the CR rate for NIVO+IPI, NIVO and IPI was 12.1%, 9.8% and 2.2%, respectively Database lock: Sept 13, 2016, minimum f/u of 28 months

49 Progression-Free Survival Median PFS, mo (95% CI) NIVO+IPI (N=314) NIVO (N=316) IPI (N=315) 11.7 ( ) 6.9 ( ) 2.9 ( ) HR (95% CI) vs. IPI 0.42 ( ) 0.54 ( ) -- HR (95% CI) vs. NIVO 0.76 ( ) Wolchok et al., NEJM 2017

50 Overall Survival Median OS, mo (95% CI) NIVO+IPI (N=314) NIVO (N=316) IPI (N=315) NR NR (29.1 NR) 20.0 ( ) HR (98% CI) vs. IPI 0.55 ( )* 0.63 ( )* -- HR (95% CI) vs. NIVO 0.88 ( ) *P< Wolchok et al., NEJM 2017

51 Subsequent Therapies: All Randomized Patients NIVO+IPI (N=314) NIVO (N=316) IPI (N=315) Any subsequent therapy, n (%)* 129 (41) 169 (54) 225 (71) Systemic therapy 100 (32) 140 (44) 196 (62) Anti-PD-1 agents 30 (10) 32 (10) 132 (42) Anti-CTLA-4 19 (6) 83 (26) 12 (4) BRAF inhibitors 40 (13) 57 (18) 68 (22) MEK inhibitors 30 (10) 38 (12) 39 (12) Investigational agents** 8 (3) 6 (2) 15 (5) Median time to subsequent systemic therapy, mo (95% CI) 2 year % of pts free of subsequent therapies NR (NR NR) 26.8 (18.0 NR) 8.5 ( ) *Patients may have received more than 1 subsequent therapy (e.g. radiation, surgery and systemic therapies) **Other than investigational immunotherapy, BRAF inhibitors, and MEK inhibitors Presented at AACR 2017 by Larkin

52 OS (%) OS (%) Outcomes Observed at PDL1 1% Cutoff PD-L1 Expression Level <1% PD-L1 Expression Level 1% <1% PD-L1 NIVO+IPI NIVO IPI 1% PD-L1 NIVO+IPI NIVO IPI Median OS, mo (95% CI) NR 23.5 (26.5 NR) (13.0 NR) 18.6 ( ) Median OS, mo (95% CI) NR NR 22.1 ( ) HR (95% CI) vs NIVO 0.74 ( ) HR (95% CI) vs NIVO 1.03 ( ) % % 67% % 41% % ORR of 54.5% for NIVO+IPI and 35.0% for NIVO 10 ORR of 65.2% for NIVO+IPI and 55.0% for NIVO Months Patients at risk: NIVO+IPI NIVO IPI Months Patients at risk: NIVO+IPI NIVO IPI Presented at AACR 2017 by Larkin

53 Safety Summary With an additional 19 months of follow-up, safety was consistent with the initial report 1 NIVO+IPI (N=313) NIVO (N=313) IPI (N=311) Patients reporting event, % Any Grade Grade 3-4 Any Grade Grade 3-4 Any Grade Grade 3-4 Treatment-related adverse event (AE) Treatment-related AE leading to discontinuation Treatment-related death, n (%) 2 (0.6) a 1 (0.3) b 1 (0.3) b Most select AEs were managed and resolved within 3-4 weeks (85 100% across organ categories) ORR was 70.7% for pts who discontinued NIVO+IPI due to AEs, with median OS not reached a Cardiomyopathy (NIVO+IPI, n=1); Liver necrosis (NIVO+IPI, n=1). Both deaths occurred >100 days after the last treatment. b Neutropenia (NIVO, n=1); colon perforation (IPI, n=1) Larkin J, et al. NEJM 2015;373:23 34.

54 Rapid complete remission after combination immunotherapy with anti-ctla4 and anti-pd1 Chapman et al., NEJM 2015

55 Conclusions (1) Combination of ipilimumab + nivolumab is highly effective in patients with metastatic melanoma with ORR of 58.9%, PFS of 11.7 months and 2 year OS rate of 58%

56 Conclusions (2) NIVO+IPI and NIVO significantly improved OS and PFS vs. IPI alone in patients with untreated advanced melanoma In descriptive analyses, NIVO+IPI resulted in numerically higher OS, PFS and ORR vs. NIVO alone Results consistently favored NIVO+IPI across clinically relevant subgroups, including PD-L1 expression <5% or <1%, mutant BRAF, and elevated LDH Although similar prolongation of OS was observed with NIVO and NIVO+IPI for PD-L1 expression 5% or 1%, NIVO+IPI resulted in higher ORR regardless of PD-L1 expression For NIVO+IPI, median DOR and time to subsequent therapy are still not reached The safety profile of the combination showed high rate of grade 3-4 IR toxicity, but early discontinuation due to AEs did not preclude benefit

57 Epacadostat Plus Pembrolizumab in Patients With Advanced Melanoma: Phase 1 and 2 Efficacy and Safety Results From ECHO- 202/KEYNOTE-037 O. Hamid, 1 T. F. Gajewski, 2 A. E. Frankel, 3 T. M. Bauer, 4 A. J. Olszanski, 5 J. J. Luke, 2 A. S. Balmanoukian, 1 E. V. Schmidt, 6 B. Sharkey, 7 J. Maleski, 7 M. M. Jones, 7 T. C. Gangadhar 8 1 The Angeles Clinic and Research Institute, Los Angeles, CA, USA; 2 University of Chicago, Chicago, IL, USA; 3 University of Texas Southwestern Medical Center, Dallas, TX, USA*; 4 Sarah Cannon Research Institute/Tennessee Oncology, Nashville, TN, USA; 5 Fox Chase Cancer Center, Philadelphia, PA, USA; 6 Merck & Co., Inc., Kenilworth, NJ, USA; 7 Incyte Corporation, Wilmington, DE, USA; 8 Abramson Cancer Center of the University of Pennsylvania, Philadelphia, PA, USA Presentation #1214O Session: Melanoma and Other Skin Tumours Presented at the ESMO Annual Meeting 2017 Madrid, Spain September 9, 2017

58 IDO1 Enzyme and Epacadostat Tumors may evade immunosurveillance through numerous mechanisms (eg, immune checkpoint inhibition of T-cell activation and upregulation of the IDO1 enzyme) IDO1 is an IFNγ-induced intracellular enzyme that catalyzes the first and ratelimiting step of tryptophan degradation in the kynurenine pathway 1 In tumors, depletion of tryptophan and production of kynurenine and other metabolites shift the local TME to an immunosuppressive state that helps tumor cells evade immunosurveillance 1 Epacadostat is a potent and specific oral inhibitor of the IDO1 enzyme 2 Combining epacadostat with an immune checkpoint inhibitor may improve patient outcomes IDO1, indoleamine 2,3 dioxygenase 1; IFNγ, interferon gamma; PD-1, programmed death 1; PD-L1, programmed death ligand Moon YW, et al. J Immunother Cancer. 2015;3: Liu X, et al. Blood. 2010;115(17):

59 Objective and Study Design Objective: To report the efficacy, safety, and tolerability data for epacadostat plus pembrolizumab in patients with advanced melanoma in the phase 1/2 ECHO-202/KEYNOTE-037 study (NCT ) Dose Escalation Epacadostat 25 mg BID + Pembrolizumab 2 mg/kg Q3W Epacadostat 50 mg BID + Pembrolizumab 2 mg/kg Q3W Epacadostat 100 mg BID + Pembrolizumab 2 mg/kg Q3W Epacadostat 300 mg BID + Pembrolizumab 200 mg Q3W Phase 1b Phase 2 Safety Expansion Epacadostat 50 mg BID + Pembrolizumab 200 mg Q3W Epacadostat 100 mg BID + Pembrolizumab 200 mg Q3W Epacadostat 300 mg BID + Pembrolizumab 200 mg Q3W Open-Label Cohort Expansion Epacadostat 100 mg BID + Pembrolizumab 200 mg Q3W Tumor cohorts Melanoma, SCCHN, UC, NSCLC (TPS 50%,* TPS <50%), RCC, TNBC, OC, MSI-High CRC,* DLBCL,* GC,* HCC* Phase 1/2 Melanoma Patients (N=65) 18 years of age Histologically or cytologically confirmed melanoma Life expectancy >12 weeks ECOG PS 0 or 1 ALT, AST, ALP <2.5x ULN; conjugated bilirubin <2.0x ULN No previous IDO inhibitor or immune checkpoint inhibitor treatment Phase 2 requirements Documented BRAF V600Emutation status or consent for BRAF mutation testing during screening Ocular melanoma excluded ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BID, twice daily; CRC, colorectal cancer; DLBCL, diffuse large B-cell lymphoma; ECOG PS, Eastern Cooperative Oncology Group performance status; GC, gastric cancer; HCC, hepatocellular carcinoma; IDO, indoleamine 2,3 dioxygenase; MSI, microsatellite instability; NSCLC, non-small cell lung cancer; OC, ovarian cancer; Q3W, every 3 weeks; RCC, renal cell carcinoma; SCCHN, squamous cell carcinoma of the head and neck; TNBC, triple-negative breast cancer; TPS, tumor proportion score; UC, urothelial cancer; ULN, upper limit of normal. * Ongoing patient enrollment at data cutoff (June 9, 2017). Ongoing patient enrollment at time of European Society for Medical Oncology presentation (September 9, 2017).

60 Patient Disposition Epacadostat Plus Pembrolizumab, P1/2 Advanced Melanoma Phase 1/2 Melanoma (N=65; Enrollment Complete) Treatment Ongoing (n=32) Treatment Completed* (n=6) Treatment Discontinued (n=27) Disease progression (n=15) Adverse event (n=6) Treatment related (n=4) Death (n=3) Patient decision (n=1) Other (n=2) Median (range) follow-up: 45+ (2 to 144+) weeks Median (range) epacadostat exposure: 34+ (<1 to 145+) weeks * Patients received 24 months of study treatment or achieved complete response and elected to discontinue treatment following the protocoldefined minimum amount of treatment ( 24 weeks before discontinuation and 2 cycles of combination treatment beyond the date of initial complete response).

61 Best Objective Response by RECIST v1.1 Epacadostat Plus Pembrolizumab, P1/2 Advanced Melanoma Per-protocol evaluable,* n (%) All Patients (N=65) Treatment-Naive for Advanced Disease, All E Doses (n=54) Treatment-Naive for Advanced Disease, E 100 mg (n=39) n=63 n=53 n=38 ORR (CR+PR) 35 (56) 29 (55) 22 (58) CR 9 (14) 7 (13) 3 (8) PR 26 (41) 22 (42) 19 (50) SD 10 (16) 9 (17) 6 (16) DCR (CR+PR+SD) 45 (71) 38 (72) 28 (74) PD or death 18 (29) 15 (28) 10 (26) Not evaluable n=2 n=1 n=1 Response Observed Across Patient Subgroups (N=63*) For all patients, based on irrecist (n=63*): ORR=59% (9 CR, 28 PR); DCR=75% (10 SD) Liver metastases yes (n=24) vs no (n=39): 46% vs 62% ORR M1c (n=35) vs non-m1c (n=28): 49% vs 64% ORR BRAF-mutation positive (n=18) vs negative (n=43): 50% vs 56% ORR LDH normal (n=39) vs elevated (n=23): 62% vs 48% ORR

62 Best Percentage Change From Baseline in Target Lesions Epacadostat Plus Pembrolizumab, P1/2 Advanced Melanoma All Patients Treatment-Naive Patients PD-L1 Positive PD-L1 Negative PD-L1 Unknown E 100 mg BID Other E Doses CR, complete response; DCR, disease control rate; E, epacadostat, irrecist, immune-related RECIST; MEL, melanoma; ORR, objective response rate; PD, progressive disease; PD-L1, programmed death ligand 1; PR, partial response; RECIST, Response Evaluation Criteria in Solid Tumors; SD, stable disease. Note: PD-L1 positive defined as MEL Score 2 (membranous PD-L1 expression in 1% of tumor cells or inflammatory cells in nests of tumor cells) as assessed by immunohistochemistry using the 22C3 antibody. * SD per target lesion but PD per RECIST v1.1. PR per target lesion but PD per RECIST v1.1. PR per target lesion but SD per RECIST v1.1. PD per target lesion per RECIST v1.1 (PR per irrecist). ǁ CR per target lesion but PD per new lesion per RECIST v1.1 (PR per irrecist); CR per target lesion but PR per RECIST v1.1 (nontarget lesions still present).

63 Percentage Change From Baseline in Target Lesions Over Time All Patients Epacadostat Plus Pembrolizumab, P1/2 Advanced Melanoma

64 Progression-Free Survival Epacadostat Plus Pembrolizumab, P1/2 Advanced Melanoma All Patients All Treatment-naive, All E Doses Treatment-naive E 100 mg BID BID, twice daily; E, epacadostat.

65 Immune related adverse events Epacadostat Plus Pembrolizumab, P1/2 Advanced Melanoma Adverse Event,* n (%) All Grade (N=65) Grade 3/4 (N=65) Total 12 (19) 6 (9) Hypothyroidism 4 (6) 0 Severe skin reactions 4 (6) 3 (5) Colitis 2 (3) 2 (3) Uveitis 2 (3) 0 Autoimmune hepatitis 1 (2) 1 (2) * Adverse events of special interest include adverse events with an immune-related cause, regardless of attribution to study treatment by the investigator. Includes grade 3 rash generalized, rash maculopapular, rash pruritic, and erythema multiforme minor. Includes iritis and uveitis. Includes grade 3 elevations of alanine aminotransferase and aspartate aminotransferase, and biopsy confirmed.

66 Conclusions Epacadostat Plus Pembrolizumab, P1/2 Advanced Melanoma Epacadostat plus pembrolizumab demonstrated promising antitumor activity in patients with advanced melanoma All Patients (n=63) 56% ORR (14% CR) 71% DCR 12.4 month median PFS* 49% landmark 18-month PFS* Treatment-Naive for Advanced Disease, All E Doses (n=53) 55% ORR (13% CR) 72% DCR 22.8 month median PFS* 52% landmark 18-month PFS* Treatment-Naive for Advanced Disease, E 100 mg (n=38) 58% ORR (8% CR) 74% DCR Median PFS not reached* 55% landmark 18-month PFS* Epacadostat plus pembrolizumab demonstrated a favorable safety profile in these phase 1/2 melanoma patients that is consistent with previous reports; 1,2 the incidence of related grade 3/4 toxicity was 20% These results support the ongoing phase 3 investigation of epacadostat plus pembrolizumab in patients with advanced melanoma (ECHO-301/KEYNOTE- 252; N=706, fully accrued)

67 Oncolytic Virotherapy Promotes Intratumoral T Cell Infiltration and Improves Anti-PD-1 Immunotherapy Ribas et al., Cell 2017

68 Ribas et al., Cell 2017 Study design: Phase 1b study 21 patients with stage IIIB/C or IV melanoma

69 Ribas et al., Cell 2017 Best overall response

70 Ribas et al., Cell 2017 Objective responses observed in all stages

71 Ribas et al., Cell 2017 PFS and OS

72 Converting Cold into Hot Tumors by Combining Immunotherapies Haanen Cell 2017

73 Efficacy of BMS (relatlimab), a Monoclonal Antibody That Targets Lymphocyte Activation Gene-3 (LAG-3), in Combination With Nivolumab in Patients With Melanoma Who Progressed During Prior Anti PD-1/PD-L1 Therapy in All-Comer and Biomarker-Enriched Populations Paolo Antonio Ascierto, 1 Petri Bono, 2 Shailender Bhatia, 3 Ignacio Melero, 4 Marta Nyakas, 5 Inge Marie Svane, 6 James Larkin, 7 Carlos A. Gomez-Roca, 8 Dirk Schadendorf, 9 Reinhard Dummer, 10 Aurélien Marabelle, 11 Christoph Hoeller, 12 Matthew Maurer, 13 Christopher Harbison, 13 Priyam Mitra, 13 Satyendra Suryawanshi, 13 Kent Thudium, 13 Eva Muñoz-Couselo 14 1 Istituto Nazionale Tumori Fondazione "G. Pascale," Napoli, Italy; 2 Comprehensive Cancer Center, Helsinki University Hospital, Helsinki, Finland; 3 University of Washington, Seattle Cancer Care Alliance, Fred Hutchinson Cancer Research Center, Seattle, WA; 4 Clinica Universidad de Navarra, Pamplona, Spain; 5 Oslo University Hospital, Oslo, Norway; 6 Copenhagen University Hospital, Herlev, Denmark; 7 Royal Marsden Hospital, NHS Foundation Trust, London, United Kingdom; 8 Institut Universitaire du Cancer, Oncopole, Toulouse, France; 9 Westdeutsches Tumorzentrum, University Hospital Essen & German Cancer Consortium, Essen, Germany; 10 UniversitätsSpital Zürich, Skin Cancer Center University Hospital, Zürich, Switzerland; 11 Gustave Roussy, Paris, France; 12 Medical University of Vienna, Vienna, Austria; 13 Bristol-Myers Squibb, Princeton, NJ; 14 Vall d Hebron Institute of Oncology, Barcelona, Spain Presented at ESMO 2017 by Ascierto

74 Potential Role of LAG-3 in T-Cell Exhaustion and Anti PD-1 Resistance LAG-3 regulates a checkpoint pathway that limits the activity of T cells 1 LAG-3 and PD-1 receptors are overexpressed and/or co-expressed on tumor-infiltrating lymphocytes in melanoma 2,3 I-O therapy naive: LAG-3 may limit I-O response + Nivolumab + Relatlimab PD-1 PD-L1 Tumor or other infiltrating cell Effector CD4 + /CD8 + T cell + Antigen LAG-3 MHC II PD-L1 PD-1 MHC II LAG-3 PD-1 I-O therapy experienced: LAG-3 may contribute to resistance + Nivolumab Acquired resistance + Nivolumab + Relatlimab Nivolumab Relatlimab (BMS /anti LAG-3) I-O, immuno-oncology; MHC II, major histocompatibility complex class II; PD-1, programmed death-1; PD-L1, programmed death ligand Grosso JF et al. J Clin Invest. 2007;117: Goding SR et al. J Immunol. 2013;190: Taube JM et al. Clin Cancer Res. 2015;21:

75 Study Rationale and Design Relatlimab + nivolumab demonstrated tolerability, peripheral T-cell activation, and preliminary clinical activity in advanced solid tumors (dose-escalation cohort; NCT ) 1 Antitumor activity was previously reported in 55 patients with melanoma who progressed during prior anti PD-1/PD-L1 therapy (ORR, 12.5%; DCR, 54%) LAG-3 expression correlated with higher rates of response and disease control a,2 Updated results in this ongoing study are presented (median follow-up, not reached [range, 0.1+ to 53+ weeks]) Dose Escalation N = 8 (advanced solid tumors) Relatlimab (80 mg) + Nivolumab (240 mg) IV Q2W Dose Expansion N = 262 Study Endpoints (dose expansion) Co-Primary: Preliminary efficacy and safety/tolerability Other: Immunogenicity, QTc, PK, PD, biomarkers Efficacy: Melanoma (progressed during prior I-O) n = 68 b Safety: All patients DCR, disease control rate; IV, intravenous; ORR, objective response rate; PD, pharmacodynamics; PK, pharmacokinetics; Q2W, every 2 weeks; QTc, corrected QT interval. a LAG-3 expression was evaluated by IHC. b Sixty-one patients were response evaluable. 1. Lipson E et al. J Immunother Cancer. 2016;4(suppl 1): Ascierto et al. J Clin Oncol. 2017;35(suppl) [abstract 9520].

76 Prior Therapies Mel Prior PD-(L)1 n = 68 Prior radiotherapy, n (%) 19 (28) Prior systemic therapy, n (%) 68 (100) Immunotherapy 68 (100) Anti CTLA-4 a 39 (57) Anti PD-1/PD-L1 b 68 (100) Best response to prior anti PD-1/PD-L1 c CR 1 (1.5) PR 12 (18) SD 20 (29) PD 31 (46) BRAF inhibitors 19 (28) MEK inhibitors 13 (19) Number of systemic regimens 1 16 (24) 2 21 (31) 3 31 (46) Median (range) 2 (1 5) Patients in the melanoma prior PD-(L)1 cohort were heavily pretreated, with approximately 77% having 2 prior therapies Most patients (57%) also received prior anti CTLA-4 therapy 46% of patients had a best response of PD to prior anti PD-1/PD-L1 therapy a Four patients received anti PD-1/PD-L1 + anti CTLA-4, 8 patients received anti CTLA-4 after anti PD-1/PD-L1, and 33 patients received anti CTLA-4 before anti PD-1/PD-L1; prior anti CTLA-4 therapy was not reported in 2 patients b Thirty-three patients received nivolumab, 33 patients received pembrolizumab, and 2 patients received other therapies. c Response in 1 patient was reported as not applicable.

77 BOR, best overall response. a Response-evaluable patients; all progressed during prior anti PD-1/PD-L1 therapy. b Immune-cell LAG-3 expression (percent of positive cells within invasive margin, tumor, and stroma) evaluated by IHC in tumor sections with antibody clone 17B4. c Tumor response evaluated by investigator per Response Evaluation Criteria in Solid Tumors v1.1. d One response was unconfirmed. e Occurred prior to first radiographic scan. Antitumor Activity of Relatlimab + Nivolumab Mel Prior PD-(L)1 a ORR, n (%) c 95% CI BOR, n (%) c All n = 61 7 (11.5) d 4.7, 22 LAG-3 1% b n = 33 6 (18) d 7, 35.5 ORR was 11.5% and DCR was 49% LAG-3 expression ( 1%) enriched for response Median duration of response was not reached (range, 0.1+ to 39+) CR 1 (1.6) 1 (3.0) PR 6 (9.8) d 5 (15) d SD 23 (38) 15 (45) PD 25 (41) 8 (24) Clinical progression e 6 (9.8) 4 (12) DCR (CR + PR + SD), n (%) c 95% CI 30 (49) 36, (64) 45, 80

78 Best percent change in sum of target lesion Best Change in Target Lesion Size by LAG-3 and PD-L1 Expression 100 LAG-3 1% n = LAG-3 < 1% n = LAG-3 Unknown n = diameters from baseline a,b % with tumor reduction % with tumor reduction % with tumor reduction Pink: PD-L1 1% Blue: PD-L1 < 1% Gray: PD-L1 unknown a Six patients with clinical progression prior to their first scan and 1 with PD due to a new symptomatic brain metastasis prior to getting full scans were not included. b One patient with best change from baseline > 30% had a best response of SD.

79 Summary With additional follow-up, clinically meaningful antitumor activity with relatlimab + nivolumab was observed in a heavily pretreated population of patients with melanoma who had progressed during prior anti PD-1/PD-L1 therapy Responses were more likely in patients with LAG-3 expression 1%, while PD-L1 expression did not appear to enrich for response The combination is well tolerated, with a safety profile similar to that of nivolumab monotherapy Further investigations of the combination of relatlimab + nivolumab are ongoing, in melanoma and other tumor types, in both I-O naive and I-O experienced patients

80 ESMO IMMUNO-ONCOLOGY CONGRESS 2017 Primary PFS and safety analyses of a randomised Phase III study of carboplatin + paclitaxel +/ bevacizumab, with or without atezolizumab in 1L non-squamous metastatic NSCLC (IMpower150) Martin Reck, 1 Mark A. Socinski, 2 Federico Cappuzzo, 3 Francisco Orlandi, 4 Daniil Stroyakovskii, 5 Naoyuki Nogami, 6 Delvys Rodríguez-Abreu, 7 Denis Moro-Sibilot, 8 Christian A. Thomas, 9 Fabrice Barlesi, 10 Gene Finley, 11 Claudia Kelsch, 12 Anthony Lee, 12 Shelley Coleman, 12 Yijing Shen, 12 Marcin Kowanetz, 12 Ariel Lopez-Chavez, 12 Alan Sandler, 12 Robert Jotte 13 esmo.org

81 Rationale for combining atezolizumab + bevacizumab In addition to its known anti-angiogenic effects, bevacizumab s inhibition of VEGF has immune modulatory effects 3. Priming and activation (APCs and T cells) 4. Trafficking of T cells to tumours (CTLs) Normalisation of the tumour vasculature for increased T-cell tumour infiltraton 4-7 Promotion of T-cell priming and activation via dendritic cell maturation Infiltration of T cells into tumours (CTLs, endothelial cells) 2. Cancer antigen presentation (dendritic cells/ APCs) 1. Release of cancer cell antigens (cancer cell death) 7. Killing of cancer cells (immune and cancer cells) 6. Recognition of cancer cells by T cells (CTLs, cancer cells) Establishing an immune-permissive tumour microenvironment by decreasing MDSC and Treg populations 7-11 Atezolizumab s T-cell mediated cancer cell killing may be enhanced through bevacizumab s reversal of VEGF-mediated immunosuppression 1. Ferrara N, et al. Nat Rev Drug Discov, Gabrilovich DI, et al. Nat Med, Oyama T, et al. J Immunol, Goel S, et al. Physiol Rev, Motz GT, et al. Nat Med, Hodi FS, et al. Cancer Immunol Res, Wallin JJ, et al. Nat Commun, Gabrilovich DI, Nagaraj S. Nat Rev Immunol, Roland CL, et al. PLoS One, Facciabene A, et al. Nature, Voron T, et al. J Exp Med, Figure adapted from Chen DS, Mellman I. Immunity, Reck M, et al. IMpower150 PFS analysis.

82 Survival follow-up IMpower150 study design Maintenance therapy (no crossover permitted) Stage IV or recurrent metastatic non-squamous NSCLC Chemotherapy-naive a Tumour tissue available for biomarker testing Any PD-L1 IHC status Stratification factors: Sex PD-L1 IHC expression Liver metastases N = 1202 R 1:1:1 Arm A Atezolizumab b + Carboplatin c + Paclitaxel d 4 or 6 cycles Arm B Atezolizumab b + Carboplatin c + Paclitaxel d + Bevacizumab e 4 or 6 cycles Arm C (control) Carboplatin c + Paclitaxel d + Bevacizumab e 4 or 6 cycles Atezolizumab b Atezolizumab b + Bevacizumab e Bevacizumab e Treated with atezolizumab until PD by RECIST v1.1 or loss of clinical benefit AND/OR Treated with bevacizumab until PD by RECIST v1.1 The principal question is to assess whether the addition of atezolizumab to Arm C provides clinical benefit a Patients with a sensitising EGFR mutation or ALK translocation must have disease progression or intolerance of treatment with one or more approved targeted therapies. b Atezolizumab: 1200 mg IV q3w. c Carboplatin: AUC 6 IV q3w. d Paclitaxel: 200 mg/m 2 IV q3w. e Bevacizumab: 15 mg/kg IV q3w. Reck M, et al. IMpower150 PFS analysis.

83 IMpower150 study populations and objectives ITT-WT a (87% of patients) Teff-high WT a High T-effector gene signature expression ITT All randomised patients a WT refers to patients without EGFR or ALK genetic alterations. Teff-low WT a Low T-effector gene signature expression EGFR/ALK + (13% of patients) 1 Co-primary objectives Investigator-assessed PFS in ITT-WT Investigator-assessed PFS in Teff-high WT OS in ITT-WT The T-effector (Teff) gene signature is defined by expression of PD-L1, CXCL9 and IFNγ and is a surrogate of both PD-L1 IHC expression and pre-existing immunity (Kowanetz M, et al. WCLC, 2017). Reck M, et al. IMpower150 PFS analysis.

84 Biomarkers in IMpower150 IMpower150 provided the opportunity to evaluate multiple strategies to enrich for PFS, including T-effector (Teff) gene signature expression and PD-L1 IHC The Teff gene signature is defined by mrna expression of 3 genes (PD- L1, CXCL9 and IFNγ) and is a surrogate for both PD-L1 expression and pre-existing immunity In the OAK study, the Teff gene signature appeared to be a more sensitive biomarker of PFS benefit for monotherapy atezolizumab vs docetaxel than PD-L1 IHC expression 1 PD-L1 expression was evaluated using the SP142 IHC assay, as defined in the Phase III OAK study of atezolizumab vs docetaxel 2 1. Kowanetz M, et al. WCLC 2017 [abstract MA 05.09]. 2. Rittmeyer A, et al. Lancet, Reck M, et al. IMpower150 PFS analysis.

85 Baseline characteristics Baseline characteristics in ITT Arm A: atezo + CP (N = 402) Arm B: atezo + bev + CP (N = 400) Arm C (control): bev + CP (N = 400) Median age (range), years 63 (32-85) 63 (31-89) 63 (31-90) Sex, male, n (%) 241 (60%) 240 (60%) 239 (60%) ECOG PS, 0, n (%) 180 (45%) 159 (40%) 179 (45%) Tobacco use history, n (%) Current smoker Previous smoker Never smoker 98 (24%) 227 (57%) 77 (19%) 90 (23%) 228 (57%) 82 (21%) 92 (23%) 231 (58%) 77 (19%) Liver metastases, yes, n (%) 53 (13%) 53 (13%) 57 (14%) EGFR mutation, positive, n (%) 46 (11%) 35 (9%) 45 (11%) ALK rearrangement, positive, n (%) 9 (2%) 13 (3%) 21 (5%) Teff gene signature expression, high, n (%) a 177 (44%) 166 (42%) 148 (37%) Of those tested KRAS mutation, positive, n (%) 36 (29%) 47 (44%) 38 (33%) PD-L1 expression, n (%) b TC2/3 or IC2/3 TC1/2/3 or IC1/2/3 TC0 and IC0 137 (34%) 213 (53%) 188 (47%) 140 (35%) 209 (52%) 191 (48%) 133 (33%) 195 (49%) 205 (51%) IC, tumour-infiltrating immune cells; TC, tumour cells. a The Teff gene signature high cut-off 1.91 was used. b 1 patient in Arm A had unknown PD-L1 IHC expression. TC2/3 or IC2/3 = TC or IC 5% PD-L1+; TC1/2/3 or IC1/2/3 = TC or IC 1% PD-L1+; TC0 and IC0 = TC and IC < 1% PD-L1+. Data cutoff: September 15, 2017 Reck M, et al. IMpower150 PFS analysis.