1. Colorectal Discussion Beyond Progression: Is this progress? Alan P. Venook, MD University of California, SF

2. Disclosures Roche / Genentech, Regeneron and Sanofi have provided me with background information I have served on Data Safety Monitoring Boards related to studies with Bevacizumab and Brivanib I helped design the original CRC Bevacizumab trial and prescribe Bevacizumab I had no involvement in the Aflibercept or Brivanib trials nor have I ever prescribed the drugs Thanks to Drs. Arnold, Allegra and Siu being timely Thanks to Lee Ellis and Napo Ferrara for trying to educate me about VEGF biology PRESENTED BY:

3. Colorectal abstracts Review data (briefly) Mechanism of action (minimal) Clinical impact Research implications Next steps PRESENTED BY:

4. Theme: Beyond Progression TML – BEV* beyond progression –Initiated by AIO; completed by Roche / Genentech VELOUR - 2 nd line Aflibercept in pts PD on Bev –Regeneron / Sanofi CO.20 – Cetuximab +/- Brivanib –NCIC + AGITG * BEV = Bevacizumab PRESENTED BY:

5. Themes: Beyond Progression TML – Bev beyond progression OS IMPROVED –Initiated by AIO; completed by Roche / Genentech VELOUR - 2 nd line Aflibercept in pts PD on Bev –Regeneron / Sanofi SAME BENEFIT +/- PRIOR BEV CO.20 – Cetuximab +/- Brivanib CLOSE BUT NOT QUITE –NCIC + AGITG PRESENTED BY:

6. SWOG 0600/iBET : A Phase III of Irinotecan and Cetuximab With or Without Bevacizumab in Patients With mCRC That Progressed During First-Line Therapy Gold, Grothey et al.. Primary endpoint: OS Secondary endpoints: PFS, objective tumor response, tolerability, and safety June 2007 Second line PD RANDOMIZERANDOMIZE RANDOMIZERANDOMIZE CT + dual biologic arm removed mCRC KRAS wild-type previously treated with Bevacizumab and oxaliplatin- based CT (n=1260) mCRC KRAS wild-type previously treated with Bevacizumab and oxaliplatin- based CT (n=1260) Bevacizumab 5 mg/kg + (FOLFIRI or Irinotecan + Capecitabine) Bevacizumab 5 mg/kg + (FOLFIRI or Irinotecan + Capecitabine) Cetuximab + (FOLFIRI or Irinotecan + Capecitabine) Cetuximab + (FOLFIRI or Irinotecan + Capecitabine) 6

7. 7 BRiTE Registry: Bevacizumab Regimens: Investigation Treatment Effects Grothey et al, ASCO, 2007 A longer median OS in the BBP subgroup was observed compared with the No BBP subgroup (Table 4). A longer survival beyond 1 st PD was observed in the BBP subgroup versus the No BBP subgroup. Median OS: 19.9 v. 31.8 mos OS Beyond PD: 9.5 v. 19.2 mos

8. SWOG 0600/iBET : A Phase III of Irinotecan and Cetuximab With or Without Bevacizumab in Patients With mCRC That Progressed During First-Line Therapy Gold, Grothey et al.. Primary endpoint: OS Secondary endpoints: PFS, objective tumor response, tolerability, and safety June 2007 – October 2010 Second line PD RANDOMIZERANDOMIZE RANDOMIZERANDOMIZE CT + dual biologic arm removed mCRC KRAS wild-type previously treated with Bevacizumab and oxaliplatin- based CT (n=1260) mCRC KRAS wild-type previously treated with Bevacizumab and oxaliplatin- based CT (n=1260) Bevacizumab 5 mg/kg + (FOLFIRI or Irinotecan + Capecitabine) Bevacizumab 5 mg/kg + (FOLFIRI or Irinotecan + Capecitabine) Cetuximab + (FOLFIRI or Irinotecan + Capecitabine) Cetuximab + (FOLFIRI or Irinotecan + Capecitabine) 8 N = 68

9. Bev + standard first-line CT (either oxaliplatin or irinotecan-based) (n=820) Randomize 1:1 Standard second-line CT (oxaliplatin or irinotecan- based) until PD Bev (2.5 mg/kg/wk) + standard second-line CT (oxaliplatin or irinotecan- based) until PD PD TML – Treatment Multiple Lines Cross-over: Oxaliplatin → Irinotecan Irinotecan → Oxaliplatin Cross-over: Oxaliplatin → Irinotecan Irinotecan → Oxaliplatin Study conducted in 220 centres in Europe and Saudi Arabia Primary endpointOverall survival from randomization Statistics: Designed to detect 30% (HR 0.77) improvement in median OS (90% power, 2-sided 5%)

10. TML features and findings Very select patient population –Tolerant of long-term Bev, not curable –PFS 1 st line > 3 months Responses infrequent (4-5%) Median 4 months additional Bev Chemotherapy backbone did not matter BEV beyond PD improves OS 1.4 months (HR = 0.81) PRESENTED BY:

11. TML: missing in abstract presentation Impact of subsequent treatment KRAS status –Only 40% received anti-EGFR BRAF status Primaries in place ? OS from diagnosis, not randomization The word: BIOMARKER PRESENTED BY:

12. What else does TML teach us? Affirms the limited utility of Registry studies regarding interventions and outcomes: –BRITE: 9.5 v. 19.2 OS beyond PD –TML: 9.8 v. 11.2 BRiTE findings not replicated; the publication* could be cited as an example of the pitfalls of Registry data * Grothey et al, JCO, 2008 PRESENTED BY:

13. Aflibercept Fusion protein of key domains from human VEGF receptors 1 and 2 with human IgG Fc¹ Blocks all human VEGF-A isoforms, VEGF-B, and placental growth factor (PlGF)² High affinity – binds VEGF-A and PlGF more tightly than native receptors 1.Holash J et al. Proc Natl Acad Sci USA. 2002;99:11393-11398. 2.Tew WP et al. Clin Cancer Res. 2010;16:358-366.

14. VELOUR – VEGF-trap with irinotecan in coLOrectal cancer after failURe of oxaliplatin 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) OS: 12.06 v 13.5 HR = 0.817 p =0.0032

15. ASCO 2012 15 Consistency of OS and PFS With and Without Prior Bevacizumab Prior BevacizumabNo Prior Bevacizumab Placebo/ FOLFIRI (n=187) Aflibercept/ FOLFIRI (n=186) Δ Placebo/ FOLFIRI (n=427) Aflibercept/ FOLFIRI (n=426) Δ OS (months) (95.34% CI) 11.7 (9.8-13.8) 12.5 (10.8-15.5) 0.8 12.4 (11.2-13.5) 13.9 (12.7-15.6) 1.5 PFS (months) (99.99% CI) 3.9 (2.9-5.4) 6.7 (4.8-8.7) 2.8 5.4 (4.2-6.7) 6.9 (5.8-8.2) 1.5 Interaction between “treatment arm” and “prior bevacizumab” factor was not significant at the 2-sided 10% level (P=0.57 for OS; P=0.2 for PFS) Response rate 8.4 11.7 12.4 23.3

16. NCIC CTG CO.20: Schema RANDOMIZERANDOMIZERANDOMIZERANDOMIZE Brivanib + Cetuximab n = 376 Placebo + Cetuximab n = 374 1° endpoint: Overall Survival Total sample size: 750 Trial: 2008 - 2011 Median follow-up: updated analysis: 34 months (694 deaths) 1:1 Stratify by Center and ECOG 0/1 versus 2

17. What do we learn from NCI CTG CO.20? PRESENTED BY: PFS improvement does not mean OS as well The study was powered such that 0.7 mos would not be statistically significant TML & VELOUR were powered differently –1.4 mos OS benefit Ask yourselves: is 1.4 mos clinically significant ?

18. Colorectal abstracts Review data (briefly) Mechanism of action (minimal) Clinical Impact Research Implications Next steps PRESENTED BY:

19. Anti-VEGF mechanism of action: evolving Ellis, Hicklin. Nat Rev Ca 2008 Proposed MOA199520002005201020122015? Anti-angiogenic (EC proliferation) +++ ++? Anti-angiogenic (Bone Marrow Derived Progenitor Cells) ++++? “Normalization” improve delivery chemo +++? Direct effect on tumor+++? Vascular “constriction”+++++ ? Offset effects of stress++ ? Disrupt the cancer stem cell niche ++? Immune function+++ ? Courtesy Lee Ellis, MD Difficult to explain mechanism of VEGF inhibition beyond progression when we are not sure of the mechanism(s) of action in the first place

20. Colorectal abstracts Review data (briefly) Mechanism of action (minimal) Clinical Impact Research Implications Next steps PRESENTED BY:

21. Clinical Impact -- Regulatory Drugs likely to be approved by FDA –Aflibercept in 2 nd line metastatic CRC w/ FOLFIRI –Regorafenib in “refractory” metastatic colorectal cancer Label addition for old drug –Bevacizumab beyond progression Drug with uncertain future –Brivanib awaits other maturing trials (will not discuss further since may never be available) PRESENTED BY:

22. Clinical Impact – Precision Medicine ? Current Practice Pattern in US While we say CONTINUUM OF CARE, in practice: FOLFOX / BEV 1 st line until neurotoxicity +/- PD, Then KRAS status: –(FOLF) Irinotecan +/- EGFR Ab –FOLFOX + EGFR Not established –Regorafenib –Aflibercept –Clinical Trial PRESENTED BY:

23. Clinical Impact: Whither FOLFOX / BEV? Hurwitz, et al. NEJM, 2004 = 4.7 mos

24. Clinical Impact: Whither FOLFOX / BEV? History Lesson Hurwitz data: BEV a difference-maker with IFL IN RETROSPECT: BEV approved “with FU regimen”; FOLFOX > IFL FOLFOX / BEV: 2 nd -line in BEV naïve patients Subsequent studies: no increase RR, marginal OS Although not limited by neurotoxicity, FOLFIRI less popular as backbone BY DEFAULT: FOLFOX / BEV 1 st -line metastatic CRC PRESENTED BY:

25. Clinical Impact – Paradigm shift? Current Practice Pattern in US While we say CONTINUUM OF CARE, in practice: FOLFOX / BEV 1 st line until neurotoxicity +/- PD, Then KRAS status: –(FOLF) Irinotecan +/- EGFR Ab –FOLFOX + EGFR Not established –Regorafenib –Aflibercept –Clinical Trial PRESENTED BY: BEV beyond PD

26. Clinical Impact – Paradigm shift Anti-VEGF therapy: Interpreting TML If TML correctly predicts that continuing 1 st -line BEV beyond PD improves outcomes in 2 nd -line: Why would that not apply for 3 rd line ? For subsequent line(s) ? For EGFR Ab’s in KRAS wt patients? Should we be stopping BEV before elective surgery? What toxicities justify stopping BEV? Treatment Holidays? Are Aflibercept and Regorafenib extensions of this effect? PRESENTED BY:

27. Clinical Impact: Anti-VEGF therapy What do we tell our patients? WHY: Once initiated, discontinuing anti-VEGF therapy may lead to a small survival disadvantage Anti-VEGF therapy works in subsequent lines as well FOREVER: In fact, it may be a mistake to discontinue anti-VEGF therapy at any time and that effects all the decisions we will make over the course of your disease WHEN ? So while anti-VEGF therapy is a standard 1 st line treatment, it may make more sense to use it in 2 nd and subsequent lines rather than up-front PRESENTED BY:

28. Colorectal abstracts Review data (briefly) Mechanism of action(minimal) Clinical Impact Research Implications Next steps PRESENTED BY:

29. Implications: from Clinical Care to Research Is FOLFOX / BEV the optimal 1 st line choice? Can we distinguish BEV v. Aflibercept ? Is Regorafenib different ? Have we reached the ceiling with VEGF as a target? What about EGFR antibodies? When? Should we revisit dual antibodies? WHO BENEFITS and WHO DOES NOT? PRESENTED BY:

30. Put some “precision” into Precision Medicine Treating many patients to help few Missed opportunity and added toxicities Paucity of biomarkers for colorectal cancer: –KRAS: who does NOT benefit from EGFR inhibition –MSI: who may not need chemotherapy –BRAF: who does very poorly overall Need to inform choice of treatment by biomarkers if at all possible PRESENTED BY:

31. Colorectal abstracts Review data (briefly) Mechanism of action (minimal) Clinical Impact Research Implications Next steps PRESENTED BY:

32. CALGB/SWOG # 80405 Untreated advanced or mCRC N =1142 Bevacizumab + FOLFOX or FOLFIRI q 2 wks Cetuximab + FOLFOX or FOLFIRI q 2 wks Screen for eligibility Send tumor tissue block to SWOG PCO Randomize Patients w/ Wild type K-ras tumor Shameless advertising: clinical and correlative results expected 2013 +/- Re-open: 6/09 Closed to accrual: 2/12 Patients enrolled: N = 2334 (total) 1177 (final endpt)

33. Search for anti-VEGF biomarkers Bevacizumab –P–Plasma; blood mRNA & DNA; tumor IHC, mRNA, DNA –p–pVEGF-A putative biomarker (JCO, 2012) Aflibercept –P–Plasma for pK and pD; VEGF Regorafenib –A–Anticipate extensive biomarker analysis –Y–Yes/no nature of outcome may yield biomarker Brivanib –P–Plasma 83-analyte (data analysis) PRESENTED BY:

34. COST of CARE The Elephant in the Room

35. Meropol and Schulman, JCO, 2007 Cost of Cancer Care: Issues and Implications

36. Impact on Cost of Care: back of the envelope Bevacizumab –$2864 per 400 mg vial* –Average weekly dose = 175 mg Regorafenib –Sorafenib $8377 / month Aflibercept PRESENTED BY: $$ per UCSF pharmacy $$$ unknown

37. Colorectal cancer in 2012: my reality check These are modest advances A minority of patients appear to benefit And the costs are unsustainable THE CHALLENGE Actually deliver on promise of personalized medicine To do so, we need better tools to predict outcomes And it must be affordable PRESENTED BY:

38. CRC 2012: making it affordable Multiple Choice (may select more than one) 1.Implement an agency like NICE PRESENTED BY:

39. NICE: National Institute for Health and Clinical Excellence Critical analysis of cost:benefit Determines what will be covered by National Health Insurance in UK Approval does not mean coverage Non-approval does not mean non-coverage http://guidance.nice.org.uk/CG131 PRESENTED BY:

40. CRC 2012: how to live within our means Multiple Choice (may select more than one) 1.Implement an agency like NICE 2.Let the market take care of things (as with generic drugs) 3.Await a bold cost-slashing move by Pharma (e.g. Cipla) 4.Give the FDA (or other panel) authority over drug pricing 5.Moratorium on studies powered to find clinically insignificant impact (to be defined) 6.Make the discovery of predictive biomarkers as high a priority as finding new drugs or indications PRESENTED BY: