1. Targeting KRAS and down streaming pathways Federico Cappuzzo Istituto Toscano Tumori Ospedale Civile Livorno-Italy

2. Istituto Toscano Tumori – Livorno, Italy Molecular events in lung cancer Adenocarcinoma Unknown 53.8% KRAS 27% EGFR 9.5% EGFR resistance mutations 0.8% HER2 0.9% BRAF 1.7% PI3K 2.6% ALK 3.7% Squamous-cell carcinoma

3. KRAS mutations in NSCLC KRAS is the most frequently mutated oncogene in NSCLC outside Asia. Mutations in ~20% of tumors KRAS mutations occur in lung adenocarcinoma and less frequently in the squamous cell carcinoma subtype KRAS mutations are usually associated with a history of tobacco use (transversion mutations), even if have been detected in 10% of never smoker (transition mutations) The vast majority of KRAS mutations involves codons 12 or 13 Istituto Toscano Tumori – Livorno, Italy

4. Questions on KRAS mutations in lung cancer Are KRAS mutations prognostic? Are KRAS mutations predictive for chemotherapy sensitivity? Are KRAS mutations predictive for targeted therapy sensitivity? Has KRAS any influence on sensitivity to checkpoint inhibitors? Any drug really working? Istituto Toscano Tumori – Livorno, Italy

5. KRAS mutations and NSCLC prognosis Istituto Toscano Tumori – Livorno, Italy

6. KRAS mutations and prognosis Over 50 studies published Different methods for detection (IHC versus PCR) Conflicting results ReferenceNMutated (%)p value for OS Cappelletti258270.7 Grossi249190.07 Kern44360.16 Ma718140.31 Tsao450260.4 Mascaux*3779180.01 *2005 meta-analysis Istituto Toscano Tumori – Livorno, Italy

7. No prognostic effect of KRAS mutations in the LACE-Bio pooled analysis Istituto Toscano Tumori – Livorno, Italy Shepherd et al, JCO 2013 All PatientsAdenocarcinoma Patients

8. Pooled analysis of prognostic and predictive effect of KRAS mut in patients treated with EGFR-TKIs Istituto Toscano Tumori – Livorno, Italy Zer et al, ESMO 2014

9. Prognostic effect of KRAS mutation in placebo arm only 9 All patients Median OS 4.7 v 5.0 m HR 1.095, p=0.48 Resected Adenocarcinoma Advanced adenocarcinoma Median OS 4.7 v 4.4 m HR 0.91, p=0.63 Median OS 5.2 v 5.1 m HR 1.0, p=0.98 Time in months Istituto Toscano Tumori – Livorno, Italy Zer et al, ESMO 2014

10. Prognostic Effect of KRAS Codon 12 Mutation Subtype 10 mOS 6.3 mo (n=75) mOS 3.9 mo (n=12) mOS 1.8 mo (n=21) p=0.01 Istituto Toscano Tumori – Livorno, Italy Zer et al, ESMO 2014

11. KRAS mutations as predictive factors for survival with adjuvant chemotherapy Istituto Toscano Tumori – Livorno, Italy

12. KRAS mutations are not predictive for adjuvant CT in the LACE-Bio pooled analysis Istituto Toscano Tumori – Livorno, Italy Shepherd et al, JCO 2013 Patients with KRAS mutPatients with KRAS wild-type Interaction p value=0.37

13. Adjuvant CT potentially detrimental in codon 13 KRAS mutations: LACE-Bio pooled analysis Istituto Toscano Tumori – Livorno, Italy Shepherd et al, JCO 2013 Patients withcodon 12 KRAS mutPatients with codon 13 KRAS mut

14. KRAS mutations as predictive factors for survival to EGFR-TKIs Istituto Toscano Tumori – Livorno, Italy

15. Predictive Effect of KRAS Mutation in patients treated with EGFR-TKIs 15 Istituto Toscano Tumori – Livorno, Italy Zer et al, ESMO 2014

16. KRAS Mutant KRAS WT Predictive effect of KRAS mutations All Patients HR=1.13, p=0.395 HR=0.91, p=0.18 Interaction p=0.17 16 Istituto Toscano Tumori – Livorno, Italy Zer et al, ESMO 2014

17. Predictive effect of KRAS mutation Adenocarcinoma EGFR WT KRAS Mutant KRAS WT HR=1.04, p=0.84 HR=0.83, p=0.19 Interaction p=0.31 17 Istituto Toscano Tumori – Livorno, Italy Zer et al, ESMO 2014

18. Predictive effect of KRAS mutation, Codon 12 subtype in advanced adenocarcinoma G12D, G12S, G12A, G12R Median OS 3.3 v 6.4m HR 0.43, p=0.08 Interaction p=0.003 G12C or G12V Median OS 8.2 v 4.3 m HR 1.64, p=0.04 18 Istituto Toscano Tumori – Livorno, Italy Zer et al, ESMO 2014

19. (n=350) Placebo (N=973) Randomization stratified by: histology, stage, prior adjuvant chemo, EGFR FISH status, smoking status, country (n=623) Erlotinib 150mg/day Up to 4 cycles of platinum-based doublet Tumor samples EGFR IHC+ and/or EGFR FISH+  90 d  180 d RADIANT trial design Radiology assessment: every 3 months on treatment and yearly during long-term follow up Primary endpoint: DFS Secondary endpoints: Overall survival (OS); DFS and OS in patients with del19/L858R (EGFR M+) No adjuvant chemotherapy 2:1 Stage IB–IIIA NSCLC Complete surgical resection 2-yr treatment period Istituto Toscano Tumori – Livorno, Italy Kelly K, ASCO 2014

20. RADIANT: DFS and OS in the whole study population Istituto Toscano Tumori – Livorno, Italy DFSOS Erlotinib Placebo Kelly K, ASCO 2014

21. RADIANT Trial: Prognostic Effect of KRAS: Placebo Arm, All patients DFS OS HR=1.14 (95% CI: 0.72, 1.81) HR=1.19 (95% CI: 0.67, 2.12) 21 Istituto Toscano Tumori – Livorno, Italy Shepherd et al, ESMO 2014

22. Predictive Value of KRAS on DFS: All Randomized 22 Interaction HR: 0.87; P=0.64 KRAS M+ KRAS WT HR: 0.81 (95% CI: 0.48, 1.38) HR: 0.93 (95% CI: 0.73, 1.19) 22 Istituto Toscano Tumori – Livorno, Italy Shepherd et al, ESMO 2014

23. 23 Predictive Value of KRAS on OS: All Randomized Interaction HR: 0.90; P=0.78 KRAS M+ KRAS WT HR: 1.05 (95% CI: 0.56, 1.99) HR: 1.15 (95% CI: 0.85, 1.55) 23 Istituto Toscano Tumori – Livorno, Italy Shepherd et al, ESMO 2014

24. Can we use KRAS testing for precluding an EGFR-TKI to pretreated NSCLC? Istituto Toscano Tumori – Livorno, Italy

25. KRAS mutations are predictive for lack of response to EGFR-TKIs ReferenceN% MutatedRR (%) Massarelli7022.80 Miller8022.50 BR2111816.95.0 Eberhadt26421.08.0 Zer Codon 12 Codon 13 79 68 5 19.31.3 1.5 0 25 Istituto Toscano Tumori – Livorno, Italy

26. Gefitinib versus docetaxel Overall survival by biomarkers 0.51.01.52.0 HR (gefitinib vs docetaxel) and 95% CI Overall EGFR FISH + EGFR FISH - EGFR expression + EGFR expression - EGFR mutation + EGFR mutation - K-RAS mutation + K-RAS mutation - 0 Favors gefitinibFavors docetaxel Istituto Toscano Tumori – Livorno, Italy

27. TAILOR Study Design ERLOTINIB 150 mg po, daily ERLOTINIB 150 mg po, daily DOCETAXEL 75 mg/m2 iv day 1,22 OR 35 mg/m2 iv day 1,8,15,29 DOCETAXEL 75 mg/m2 iv day 1,22 OR 35 mg/m2 iv day 1,8,15,29 R 1:1 CROSS OVER NOT ALLOWED CROSS OVER NOT ALLOWED Garassino MC, et al Lancet Oncol 2013 N 222 EGFR wild-type Istituto Toscano Tumori – Livorno, Italy

28. Prognostic impact of KRAS Univariate analysis PFSOS Garassino MC, et al ESMO 2013 Istituto Toscano Tumori – Livorno, Italy HR 1.02, p=0.93HR 1.28, p=0.14

29. Predictive impact of KRAS on OS KRAS mutantsKRAS wild-type Test for interaction Univariate p=0.82; Multivariate =0.97 HR 0.81 (95%CI 0.45-1.47 ) p=0.492HR 0.79 (95%CI 0.57-1.10) p=0.167 Garassino MC, et al ESMO 2013 Istituto Toscano Tumori – Livorno, Italy HR 0.81, p=0.492HR 0.79, p=0.167

30. Can we use KRAS testing for selecting patients candidate for checkpoint inhibitors? Istituto Toscano Tumori – Livorno, Italy

31. Clinical Development of Inhibitors of PD-1 Immune Checkpoint TargetAgentMoleculeCompanyDevelopment PD-1Nivolumab- BMS-936558 Fully human IgG4 mAbBristol-Myers SquibbPhase II, III multiple tumors Pidilizumab CT-011 Humanized IgG1 mAbCureTechPhase II multiple tumors Lambrolizumab MK-3475 Humanized IgG4 mAbMerckPhase I-II AMP-224Recombinant PD-L2-Fc fusion protein GlaxoSmithKlinePhase I PD-L1BMS-936559Fully human IgG4 mAbBristol-Myers SquibbPhase I MedI-4736Engineered human IgG1 mAb MedImmunePhase I MPDL-3280AEngineered human IgG1 mAb GenentechPhase I-II Istituto Toscano Tumori – Livorno, Italy

32. a ORR includes investigator-assessed unconfirmed and confirmed PR. Patients first dosed at 1-20 mg/kg by Oct 1, 2012. Data cutoff: Apr 30, 2013. Patients With PR, % Response by Smoking Status (ORR a ) n = 43 n = 10 MPDL3280A Phase Ia: Response by Smoking Status - NSCLC Former/current smokers Never smokers Patient Smoking Status (n = 53) Soria et al et al. ESMO 2013 Istituto Toscano Tumori – Livorno, Italy

33. Activity of pembrolizumab according to clinical characteristics NORR (%) Total23621 Previous treatment236 Treatment naive4226 Previuos treated19420 Histology230 Non-squamous19123 Squamous3918 Smoking history230 Current/Former16527 Never659 Garon et al et al. ESMO 2014 Istituto Toscano Tumori – Livorno, Italy

34. Nivolumab more effective in smokers Variable ORR, % (n/N) [95% CI] P-value Smoking exposure0.018 ≤5 pack-yrs0 (0/14) [0, 23] >5 pack-yrs30 (20/66) [20, 43] Time since quitting0.22 Current smoker27 (6/22) [11, 50] 1–5 yrs prior46 (6/13) [19, 75] 6–15 yrs prior17 (2/12) [2, 48] >15 yrs prior18 (6/33) [7, 36] 0 20 80 60 40 Months Since Treatment Initiation 10 0 PFS (%) PFS by smoking exposure Never/minimal smokers (mPFS 1.7 months) Former/current smokers (mPFS 2.2 months) HR (95% CI) = 0.41 (0.22, 0.74), P = 0.003 mPFS = median Progression-free survival Istituto Toscano Tumori – Livorno, Italy Hellmann et al et al. ASCO 2014

35. Istituto Toscano Tumori – Livorno, Italy Altered protein contain new epitopes for immune recognition, providing a common denominator for immunotherapy High mutational rates may contribute to increased immunogenicity

36. Response to anti-PD-1 is independent of EGFR or KRAS mutation status: nivolumab as an example SubgroupORR, % (n/N) [95% CI] EGFR status Mutant17 (2/12) [2.1, 48.4] Wild-type20 (11/56) [10.2, 32.4] Unknown15 (9/61) [7.0, 26.2] KRAS status Mutant14 (3/21) [3.0, 36.3] Wild-type25 (9/36) [12.1, 42.2] Unknown14 (10/72) [6.9, 24.1] Change in tumour size, % -100 -80 -60 -40 120 -20 0 20 40 60 80 100 Patients EGFR mutation status Mutant Unknown Wild-type -100 -80 -60 -40 120 -20 0 20 40 60 80 100 Patients Change in tumour size, % KRAS mutation status Mutant Unknown Wild-type CA209-003: phase 1 follow-up study, up to 5 prior lines of therapy, NSCLC cohort Brahmer J, et al. ASCO 2014 (Abstract 8112). Istituto Toscano Tumori – Livorno, Italy

37. Response to anti-PD-1 is independent of EGFR or KRAS mutations: pembrolizumab and MPDL 3280A SubgroupNORR (%) Total5323 EGFR status46 Mutant617 Wild type4023 KRAS status37 Mutant10 Wild type2730 MPDL 3280APembro SubgroupNORR (%) Total23621 EGFR status Mutant3614 Wild type214- KRAS status Mutant3928 Wild type117- Istituto Toscano Tumori – Livorno, Italy Garon et al et al. ESMO 2014Soria et al et al. ESMO 2013

38. There is any drug really working in KRAS mutated NSCLC? Istituto Toscano Tumori – Livorno, Italy

39. Selumetinib Selumetinib (AZD6244, ARRY-142886) is a potent and selective allosteric inhibitor of MEK 1/2 1 Tendency for greater sensitivity to selumetinib in BRAF/RAS-mutant cell lines 2 1. Yeh et al. Clin Cancer Res 2007;13:1576–83; 2. Davies et al. Mol Cancer Ther 2007;6:2209–19 RasRafMEK 1/2ERK 1/2 Selumetinib Cell viability inhibition IC 50 (µM) Cell line RAS/BRAF mutation 0 5 10 15 20 25 30 None KRAS BRAF NRAS

40. Selumetinib: preclinical/early phase clinical studies Preclinically, the combination of selumetinib and docetaxel demonstrated tumor regression in a KRAS-mutant cancer 1 Phase I study showed that selumetinib plus docetaxel had a manageable tolerability profile 2 1. Holt et al. Br J Cancer 2012; 2 Kim et al. Mol Can Ther 2011; HCT-116 human tumor xenografts (KRAS-mutant) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0 2 4 6 8 10 12 14 Mean tumour volume (cm3) +/-s.e.m Control Selumetinib-25mg kg-1 per qd Docetaxel-15mg kg-1 Combo Istituto Toscano Tumori – Livorno, Italy

41. Phase II, double-blind, randomized, placebo- controlled, multi-center trial; NCT00890825 Docetaxel was administered every 21 days; selumetinib/placebo administered daily Following completion of patient enrollment, the primary endpoint was changed from PFS to OS, without changing the sample size ‡ –OS analysis was planned for after approximately 58 events; HR 0.57, 80% power assuming a 1-sided 10% significance level Selumetinib 75 mg BID + docetaxel 75 mg/m 2 Placebo BID + docetaxel 75 mg/m 2 Endpoints Primary OS Secondary PFS ORR Duration of response Change in tumor size Alive and progression- free at 6 months Safety and tolerability Randomization (1:1 ratio) Patients Locally advanced or metastatic NSCLC (stage IIIB-IV) Failed first-line therapy Confirmed KRAS mutant tumor* WHO PS 0-1 Excluding symptomatic brain metastases Janne PA et al, Lancet Oncol 2014

42. Overall survival There was a numerical increase in OS (median follow-up 7.2 mo); hazards non-proportional –56/83 deaths (67% maturity): selumetinib + docetaxel 29/43, placebo + docetaxel 27/40 Median OS Selumetinib + docetaxel, n=439.4 mo Placebo + docetaxel, n=405.2 mo HR 0.80; 80% CI 0.56, 1.14; 1-sided p=0.2069* Proportion of patients alive Days 0 50 100 150 200 250 300 350 400 450 500 550 600 650 Number at risk 1.0 0.8 0.6 0.4 0.2 0.0 43 40 41 35 36 27 35 20 28 18 25 15 14 13 6666 6565 2424 131321 Istituto Toscano Tumori – Livorno, Italy Janne PA et al, Lancet Oncol 2014

43. Trametinib and pemetrexed for KRAS mutated NSCLC Istituto Toscano Tumori – Livorno, Italy Mazieres J et al. WCLC 2013

44. Tivantinib: Study Design Randomized, placebo-controlled, double-blind clinical trial RANDOMIZERANDOMIZE Erlotinib 150 mg PO QD + Placebo 28-day cycle Erlotinib 150 mg PO QD + Placebo 28-day cycle Erlotinib 150 mg PO QD + ARQ 197 360 mg PO BID 28-day cycle Erlotinib 150 mg PO QD + ARQ 197 360 mg PO BID 28-day cycle Endpoints 1° PFS 2° ORR, OS Subset analyses Crossover: ORR NSCLC Inoperable locally adv/ metastatic dz. ≥1 prior chemo (no prior EGFR TKI) 33 sites in 6 countries Study accrual over 11 months (10/08-9/09) Randomization stratified by prognostic factors incl. sex, age, smoking, histology, performance status, prior therapy and best response, and geography (U.S. vs. ex- U.S.) PD Istituto Toscano Tumori – Livorno, Italy

45. 5.02.0 Tivantinib: PFS in Histologic and Molecular Subgroups ARQ197/erlotinibPlacebo/erlotinib Unadjusted HR (95% CI) NMedian PFS (95% CI, weeks) Squamous Cell26 / 2413.7 (8.0‒18.1)8.4 (7.9‒21.0) Non-Squamous Cell58 / 5918.9 (15.0‒31.1)9.7 (8.0‒16.0) c-MET FISH >419 / 1815.4 (8.1‒24.4)15.3 (7.1‒16.3) c-MET FISH >58 / 1124.1 (16.3‒NE)15.6 (7.9‒31.4) EGFR mutant6 / 1124.1 (8.0‒32.1)21.0 (8.1‒36.0) EGFR wt51 / 4813.7 (8.1‒18.1)8.1 (7.9‒9.9) KRAS mutant10 / 59.7 (7.9‒NE)4.3 (1.1‒8.0) KRAS wt49 / 4515.4 (8.1‒18.1)9.9 (8.0‒16.0) 1.00 Favors ARQ 197/Erlotinib Favors Erlotinib/placebo HR=0.70 HR=0.18 0.51.5 HR=1.01 HR=1.23 HR=0.45 HR=0.71 HR=1.05 HR=0.71 Istituto Toscano Tumori – Livorno, Italy

46. MARQUEE phase III study design Primary end-point: OS Istituto Toscano Tumori – Livorno, Italy

47. MARQUEE: Forest plot for OS in key subgroups Istituto Toscano Tumori – Livorno, Italy

48. Conclusions KRAS mutations are not prognostic in early nor advanced NSCLC KRAS mutations are not predictive for OS in: –Early NSCLC exposed to adjuvant chemotherapy –Early NSCLC exposed to EGFR-TKIs –Advanced NSCLC exposed to EGFR-TKIs KRAS mutations predict no response to EGFR-TKIs KRAS mutations cannot be used for precluding a treatment with –Adjuvant Chemotherapy –EGFR-TKIs in second or subsequent lines –In patients candidate for checkpoint inhibitors No drugs are currently available in clinical practice Several compounds are under investigation Istituto Toscano Tumori – Livorno, Italy