1. VEGF-Targeted Therapies: Resistance and Revisiting Our Assumptions Lee M. Ellis, MD Depts. of Surgical Oncology and Cancer Biology U.T. M.D. Anderson Cancer Center Houston, Texas, USA A reality check now that we have clinical data.

2. Disclosures Ad hoc consulting –Genentech/Roche, sanofi-aventis, Bayer Most of the data involves bevacizumab as this drug is approved in more disease types than others in the class Thank you for sharing slides –Lillian Siu –Axel Grothey/Eric Van Cutsem –Josep Tabernero/Carmen Allegra Data collected from searches on Pubmed, Clinicaltrials.gov, ASCO, and AACR websites

3. Anti-angiogenic Assumptions: Then and Now Assumptions 20-40 yrs agoAssumptions 2002-2010What we know from clinical trial results (in 2012 Angio inhibition would induce dormancy in all tumor types Angio inhibition would provide benefit across tumor types Benefit is tumor dependent and context dependent (+/- chemo) Little discussion of multiplicity of angiogenic factors Other angiogenic factors are important and may contribute to resistance Dual targeting of bypass pathways have not led to major advances Resistance would not occurResistance is inevitableContinuation of therapy may be of some benefit Did not consider consequences of induction of hypoxia VEGF inhibitors may increase tumor aggressiveness In GBM, VEGF inhibitors may increase invasion and metastasis, but patients still benefit from therapy Did not consider consequences of withdraw Preclinical and anecdotes- Withdraw may lead to “flare” No hard data to support that withdraw lead to “flare” Did not think about biomarkersBiomarkers are elusiveMaybe? Need validation

4. Resistance to Anti-VEGF Therapy Introduction/Background Resistance in preclinical studies –And were these studies confirmed in the clinic Resistance/Sensitivity Beyond Progression

5. Almost All Malignancies Are Inherently Resistant To VEGF Targeted Therapies (Exception-rcc) To Understand Resistance We Need to Better Understand Mechanisms of Action

6. MOA: Do We (I) Have Any Clue? Ellis, Hicklin. Nat Rev Ca 2008 Proposed MOA199520002005201020122015? Anti-angiogenic (EC proliferation) +++ ++? Anti-angiogenic (Bone Marrow Derived Progenitor Cells) ++++? “Normalization” with improved delivery of chemo +++? Direct effect on tumor+++? Vascular “constriction”+++++ ? Offset effects of stress++ ? Disrupt the cancer stem cell niche ++? Immune function+++ ? The benefits of a long career is that you can change you mind many, many times

7. Anti-VEGF Therapy: My Theory on Different Mechanisms of Action in Different Tumor Systems Renal Cell Carcinoma/NETs Colon Carcinoma With diverse MOAs, understanding resistance is likely to be tumor dependent, and/or site dependent.

8. Resistance to Anti-VEGF Therapy Introduction/Background Resistance in preclinical studies –And were these studies confirmed in the clinic Resistance/Sensitivity Beyond Progression

9. Most Resistance Manuscripts and Reviews Focus on Redundant Angiogenic Pathways Ellis, Hicklin, CCR 2008 c-Met

10. FGF

11. FGFs

13. Hypothesis Generating Clinical Study in CRC Blood drawn at each cycle and at progression

14. Cytokines Increased Prior to Progression On FOLFIRI + Bev Kopetz JCO 2009

15. Results of a Recently Reported Clinical Trial Targeting FGF-R in CRC Clinicaltrial.gov search May, 2012 The only randomized Phase III trial with FGF inhibitor Brivanib And others at higher IC50s…. Diaz-Padilla, Siu Exp Opin Invest Drugs 2011

16. CO.20 Trial: Cetuximab +/- Brivanib K-Ras WT Chemo-refractory CRC Cetuximab + Placebo Cetuximab + Brivanib P Value PFS3.45.0<0.0001 (HR=0.72) RR7.213.60.044 Siu et al

17. Commentary/ Discussion This is but a single study-some hints of activity –Biomarker Driven Study: »Phase II Study of Second-line Irinotecan Plus Brivanib, a Dual Tyrosine Inhibitor of VEGFR and FGFR, in Metastatic Colorectal Cancer Patients Enriched for Elevated Levels of Plasma FGF Following Progression on Bevacizumab-based Treatment: Mike Overman, MDACC There are numerous bypass pathways Kopetz/Heymach JCO: B-FGF, HGF, PlGF, Eotaxin, MMP-9, and more Not all tumors are p-NETs

18. VEGFR-1 (Flt-1) NRP-1/ NRP-2 VEGF-DVEGF-C VEGF-B VEGF-A PlGF VEGFR-2 (Flk-1/KDR) VEGFR-3 (Flt-4) Vasculogenesis Angiogenesis Lymphangiogenesis PlGF

19. Induction of PlGF by Anti-VEGF Therapy

20. The PlGF Debate!

21. VEGFR-1 (Flt-1) NRP-1/ NRP-2 VEGF-DVEGF-C VEGF-B VEGF-A PlGF VEGFR-2 (Flk-1/KDR) VEGFR-3 (Flt-4) Vasculogenesis Angiogenesis Lymphangiogenesis BEVACIZUMAB* VEGF-TRAP 18F1 1121B TG-403 Tyrosine Kinase Inhibitors Sunitinib* Sorafenib* Pazopanib* Axitinib* Motesanib Cedirinib Brivanib Many, many others VEGF Targeted Agents in the Clinic or In Clinical Trials Ellis, Hicklin Nat Rev Ca. 2008 * FDA approved agents Do we see improved outcomes in patients treated with agents that target PlGF/VEGFR-1?

22. VELOUR Study Design Metastatic Colorectal Cancer RANDOMIZERANDOMIZE Aflibercept 4 mg/kg IV, day 1 + FOLFIRI q2 weeks Aflibercept 4 mg/kg IV, day 1 + FOLFIRI q2 weeks Placebo IV, day 1 + FOLFIRI q2 weeks Placebo IV, day 1 + FOLFIRI q2 weeks 1:1 Disease Progression Death 600 600 600 Stratification factors: - ECOG PS (0 vs 1 vs 2) - Prior bevacizumab (Y/N) Primary endpoint: Overall survival Sample size: HR 0.8, 90% power and a 2-sided type I error 0.05 Final analysis of OS: Analyzed at 863 th death event using a 2-sided nominal significance level of 0.0466 (α spending function)

23. Overall Survival, ITT Population Cut-off date = February 7, 2011; Median follow-up = 22.28 months Presented at ESMO/WCGC meeting 2011, Barcelona: Abstract O-0024 1.4 mo increase in median OS

24. FDA Approvals Of VEGF-Targeted AgentsThat Have Been in Clinical Development for “Awhile” BevacizumabAfliberceptTKIs CRCPending - Neg front line Ph II trial Pending (Regorafenib) NSCLCNeg BreastwithdrawnNeg GBMNeg RCC P-NETs HCC If PlGF/VEGFR-1 is important in VEGF/VEGFR-2 resistance, Aflibercept and TKIs should be more effective PlGF/VEGFR-1 Targeting

25. Commentary/ Discussion Agents that target VEGF + PlGF are no more efficacious than those that target VEGF –In the only head-head Phase III study in CRC, Chemo + Bev vs Chemo + TKI (Cediranab), the primary endpoint for Cediranib was not met

26. C-Met/HGF DrugClass of Drug ARQ197TKI GSK1363089 (XL880)TKI AMG 102MoAB HGF AMG 208TKI AV-299MoAB HGF PF-02341066TKI PF-04217903TKI MP470TKI MGCD265TKI MetMabMoAB Met Modified Yap et al. Cancer Drug Targets 2011

27. Hypothesis: Combination Therapy with a VEGF-R TKI and c- Met TKI Is Better Than Either Therapy Alone “In conclusion, our study indicates a role for HGF/c-Met pathway in development of resistance to antiangiogenic therapy and suggests a potential strategy to circumvent resistance to vascular endothelial the clinic.”

28. Combination VEGF/Met Targeted Trials

29. Other Resistance Pathways Not Discussed Mostly refuted Untested in the clinic Under investigation: Are we targeting ECs or Cancer Stem Cells or Both? (Notch inhibitors slow in development)

30. Resistance to Anti-VEGF Therapy Introduction/Background Resistance in preclinical studies –And were these studies confirmed in the clinic Resistance/Sensitivity Beyond Progression

31. PFS in Patients Receiving TKIs After Prior VEGFR TKIs 2008

33. Continuation of VEGF-Targeted Therapy After First Line Progression in mCRC TrialSTUDY DESIGNOUTCOMECOMMENTSASCO 2012 Abstract VELOURFOLFIRI +/- Aflibercept after progression OS HR = 0.817* PFS HR = 0.76 OS HR Prior Bev 0.86 ~30% Received Prior Bev Abstract #3505 CORRECTRegorafenib vs Placebo in refractory patients OS HR= 0.77* PFS HR = 0.49 All Received Prior Bev Abstract #3502 TML18147Bev + alternate chemo after progression on Bev + first line chemo Primary endpoint of improved OS met* All Received Prior Bev Abstract #CRA3503 * Met primary endpoint

34. Primary Endpoint Allegra Abstract #3505

35. TML18147-OS*: ITT population (“Bev Beyond Progression”) *From randomisation OS estimate Time (months) 1.0 0.8 0.6 0.4 0.2 0 0612182430384248 Number at risk Chemo41029316251247320 Bev + chemo409328189642913410 Chemo Bev + chemo 9.8 11.2 Arnold, Abstract #CRA3503 Chemo (n=411) Bev + chemo (n=409) Median time to event (mos) 9.811.2 HR (95% CI)0.81 (0.69–0.94) p-value0.0062

36. 1.00 0.50 0.25 0 0.75 200100500150300250350 Days from randomization Survival distribution function Placebo N=255 Regorafenib N=505 Regorafenib Placebo Median 1.9 mos 1.7 mos (95% CI) (1.9–2.1) (1.7–1.7)) Hazard ratio: 0.49 (95% CI: 0.42–0.58) 1-sided p-value: <0.000001 CORRECT: OS and PFS Regorafenib vs BSC Grothey et al. ASCO GI 2012 (LBA 385). 1.00 0.50 0.25 0 0.75 200100500150300250400350450 Days from randomization Survival distribution function Placebo N=255 Regorafenib N=505 Median 6.4 mos 5.0 mos 95% CI (5.9–7.3) (4.4–5.8) Hazard ratio: 0.77 (95% CI: 0.64–0.94) 1-sided p-value: 0.0052 Primary endpoint met prespecified stopping criteria at interim analysis (1-sided p<0.009279 at approximately 74% of events required for final analysis) Regorafenib Placebo Overall Survival Progression-Free Survival

37. Conclusions The more we learn about the tumor vasculature, the less sure we about the MOA of VEGF-targeted agents There are numerous redundant pathways that mediate resistance to VEGF-targeted agents –Will we need to inhibit multiple pathways (>2) to obtain meaningful clinical benefit? Is this feasible????? (tox) VEGF inhibition beyond progression provides some degree of patient benefit…i.e. complete resistance may not occur Biomarkers of sensitivity and resistance will improve patient outcomes--- in my opinion, the only way to make a major advance at this stage –Michael Maitland to discuss

38. Thank You for Your Attention

39. Bev + standard 1L chemo (either oxaliplatin or irinotecan-based) (n=820) Standard 2L chemo (oxaliplatin or irinotecan-based)* until PD Bev (2.5mg/kg/wk) + standard 2L chemo (oxaliplatin or irinotecan- based)* until PD PD Key eligibility criteria Histologically confirmed diagnosis of metastatic colorectal cancer ≥3 months of standard 1L bev plus chemo PD <3 months after last bev administration Primary endpointOS from randomisation Secondary endpointsPFS, overall response rate (ORR), OS from start of 1L therapy TML18147 Study Design (Phase III) * Cross-over: Oxaliplatin → Irinotecan Irinotecan → Oxaliplatin * Cross-over: Oxaliplatin → Irinotecan Irinotecan → Oxaliplatin PD = disease progression Randomize 1:1 Arnold, Abstract #CRA3503

40. CORRECT Study Design Multicenter, randomized, double-blind, placebo-controlled, phase III –2:1 randomization –Strat. factors: prior anti-VEGF therapy, time from diagnosis of mCRC, geographical region Global trial: 16 countries, 114 active centers –1,052 patients screened, 760 patients randomized within 10 months Secondary endpoints: PFS, ORR, DCR Primary Endpoint: OS 90% power to detect 33.3% increase (HR=0.75), with 1-sided overall a=0.025 Primary Endpoint: OS 90% power to detect 33.3% increase (HR=0.75), with 1-sided overall a=0.025 Regorafenib + BSC 160 mg orally once daily 3 weeks on, 1 week off Regorafenib + BSC 160 mg orally once daily 3 weeks on, 1 week off Placebo + BSC 3 weeks on, 1 week off R 2:1 2:1 mCRC after standard therapy Grothey et al. ASCO GI 2012 (LBA 385).