1. Predictive Biomarkers of EGFR Targeted Therapy: Beyond KRAS Josep Tabernero, MD PhD Vall d’Hebron University Hospital and Vall d’Hebron Institute of Oncology (VHIO) Barcelona, Spain

2. NCIC CTG CO.17: mCRC Cetuximab vs BSC HR OS: ITT 0.7  KRAS wt 0.55 1 Jonker, DJ et al. NEJM; 357:2040-8,2007 2 Karapetis, CS. et al. NEJM;359:1757-65,2008 EGFR inhibitors: beyond KRAS

3. Baseline Predictive Biomarkers MUTATIONS KRAS mut NRAS mut, BRAF mut PIK3CA mut TP53 mut, PTEN mut RECEPTOR EXPRESSION EGFR (HER1), HER- 2/neu, HER-3, HER-4, IGF1R LIGAND EXPRESSION EGF, TGF-α, HB-EGF, amphiregulin (AREG), betacellulin, epiregulin (EREG) and epigen DOWNSTREAM EFECTORS EXPRESSION MPK-1 (DUSP-1), DUSP-4, DUSP-6

4. Mutant BRAF 11/79 (14%) Wild-type BRAF 68/79 (86%) Responders0/11 (0%) ‡ 22/68 (32%) ‡ Non-responders11/11 (100%) ‡ 46/68 (68%) ‡ Mutant KRAS 34/113 (30%) Wild-type KRAS 79/113 (70%) Responders2/34 (6%)*22/79 (28%)* Non-responders32/34 (94%)*57/79 (72%)* BRAF mutational status on wild-type KRAS tumours (n=79) KRAS mutational status (evaluable patients n=113) mCRC patients treated with panitumumab or cetuximab, n=114 *p<0.05 (p=0.011) ‡ p<0.05 (p=0.029) Di Nicolantonio, et al. J Clin Oncol 2008;26:5705-12 BRAF mutations

5. a Cochran–Mantel–Haenszel testVan Cutsem E, et al. J Clin Oncol 2011; 29:2011-9 BRAF mutations

6. European Consortium: –Refractory CRC patients treated with irinotecan + cetuximab –1022 tumour samples –773 samples of quality De Roock, W et al. Lancet Oncol 2010;11:753-62 BRAF, NRAS, PIK3CA mutations

7. Response Rate Overall Survival De Roock, W et al. Lancet Oncol 2010;11:753-62 BRAF, NRAS, PIK3CA mutations

8. Ir vs IrCs non-inferior efficacy primary endpoint PFS at 12 wks IrPan irinotecan + pan’mab IrPan irinotecan + pan’mab IrCs irinotecan + c’sporin IrCs irinotecan + c’sporin Ir irinotecan alone Ir irinotecan alone Ir vs IrPan superior efficacy primary endpoint OS Ir irinotecan alone Ir irinotecan alone KRAS mutated or unknown KRAS-wt (c.12/13 & 61) PICCOLO study target total n = 1200 (including 494 accrued under previous design) eligibility as before Seymour et al. Proc ASCO 2011:A3523 BRAF, NRAS mutations

9. IrPan better Ir better *Adjusted HRs, 95% CIs KRAS wt: 460 pts, 312 events HR=0.91 (0.73, 1.14), p=0.44 Double wt : 348 pts, 230 events; HR=0.87 (0.67, 1.13), p=0.30 BRAF mut: 63 pts, 53 events; HR=2.03 (1.13, 3.64) NRAS mut: 21 pts, 15 events HR=4.59 (1.19, 17.67) KRAS 146 mut: 17 pts, 14 events HR=1.32 (0.30, 5.81) Any mut: 99 pts, 80 events HR=2.03 (1.26, 3.28) All wt: 264 pts, 171 events; HR=0.86 (0.63, 1.16), p=0.32 Seymour et al. Proc ASCO 2011:A3523 BRAF, NRAS mutations OS

10. Presented by: BRAF, NRAS mutations Oliner et al. Proc ASCO 2013:A3511 PRIME study: FOLFOX +/- Panitumumab Overall Survival

11. Bertotti A et al. Cancer Discov 2011;1:508-23 Yonesaka K et al. Sci Transl Med 2011;3:99ra86 Quadruple* wt non-responders (n=11 xenopatients) 36.4% (P<0.001) NormalHER2 amplification HER2 FISH - HER2 *KRAS/NRAS/ BRAF/PIK3CA HER-2 amplification HER-2 amplHER-2 non-ampl N13220 mPFS (d)89149 p=NS mOS (d)307515 p=0.0013 mCRC treated with cetuximab (DFCI)

12. Scartozzi M et al. Oncologist 2011;16:53-60 Scartozzi M et al. Ann Oncol 2012;23:1706-1712 HER-3 overexpression HER-3 +HER-3 - N4440 mPFS (m)2.86.3 p<0.0001 mOS (m)10.513.6 p=0.01 mCRC treated with irinotecan + cetuximab PFS OS

13. 1 Singh, AB et al. Cell Signal; 17:1183-1193,2005 2 Shelly, M et al. J Biol Chem; 273:10496-10505,1998 3 Khambata-Ford, S. et al. J Clin Oncol; 25:3230-3237, 2007 4 Tabernero J et al, J Clin Oncol 2010;28:1181-1189 5 Jacobs B et al, J Clin Oncol 2009;27:5068-5074 EGFR ligands: –1 in C. Elegans –4 in Drosophila –7 in mammals: EGF, TGF-α, HB-EGF, amphiregulin (AREG), betacellulin, epiregulin (EREG) and epigen 1 –EREG and AREG bind more weakly to EGFR than EGF but much more potently and more prolonged –EREG preferentially activates heterodimers 2 High gene expression levels of EREG and AREG predict response to cetuximab 3-5 EGFR Ligands

14. Ligands: AREG, EREG, TGFalpha Presented by: Tabernero J et al, J Clin Oncol 2010;28:1181-1189 mCRC 1 st -line treatment Cetuximab (window) + FOLFIRI

15. Combimarker: K-Ras wt and high EREG –Pre-especified threshold 1 –Minimum threshold: 169/384 (44%) –All comers  394 (100%)HR: 0.7 –K-Ras wt  230 (58%) HR: 0.55 –Combimarker  169 (44%)HR: 0.46 1 Khambata-Ford, S. et al. J Clin Oncol; 25:3230-3237, 2007 2 Jonker, D et al. Proc ASCO 2009 Ligands: AREG, EREG, TGFalpha

16. On-treatment Predictive Biomarkers (“under pressure”) MUTATIONS KRAS mut EGFR mut RECEPTOR EXPRESSION HER-2/neu, HER-3, IGF1R RECEPTOR AMPLIFICATION C-Met

17. Misale S, et al. Nature 2012 Diaz E, et al. Nature 2012 KRAS mut/ampl under pressure

18. Initial response to cetuximab followed by PD Quantitative analysis of KRAS(Q61H) mutant DNA in plasma, as assessed by BEAMing Misale S, et al. Nature 2012 Diaz E, et al. Nature 2012 KRAS mut/ampl under pressure

19. Bardelli A, et al. Cancer Discov 2013;e C-Met amplification

20. Montagut C, et al. Nature Med 2012;18:221-223 Morelli MP et al. Proc ASCO 2013:A3512 EGFR mutations EGFR ectodomain mutation (S492R) that prevents Cetuximab binding, but retain binding to panitumumab S492R EGFR mutation was detected in 4 (7%) out of 55 pts (MDACC) KRAS wt after treatment with cetuximab

21. CD73 – 5’nucleotidase (5’-NT) Mizumoto N et al, Nat Med 2002;8:358-365 Stagg J et al, PNAS 2010;107:1547-1552 The release of extracellular ATP in response to cell death or stress activates immune responses The hydrolysis of extracellular ATP into adenosine acts as a negative feedback mechanism to prevent excessive inflammation and tissue damage CD73 converts AMP to adenosine and is considered as a an immunosuppressor inducer Anti-CD73 mAb therapy delays tumor growth in immune-competent mice and inhibits the development of metastasis, through the induction of adaptive immune response

22. CALGB80203 Study Design – Biomarker program A: FOLFIRI B: FOLFIRI + Cetuximab C: FOLFOX D: FOLFOX + Cetuximab 1 st line Metastatic Colorectal Cancer n = 238 Key inclusion criteria: EGFR positive Ras status not required Accrual: 01/21/2004-12/03/2004 Total patients for biomarker analysis (n=103) Hurwitz H, et al. Proc ASCO 2013

23. Prognostic Analyses: Chemotherapy only population 52 patients Hurwitz H, et al. Proc ASCO 2013

24. Prognostic Analyses: Chemotherapy only, WT KRAS population 29 patients Hurwitz H, et al. Proc ASCO 2013

25. Predictive Analyses: CD73 and Her3 Overall Population Population: 52 w/o + 51 w Cetuximab Hurwitz H, et al. Proc ASCO 2013

26. Predictive Analyses: CD73 and Her3 Wild type KRAS Population Population: 29 w/o + 26 w Cetuximab Hurwitz H, et al. Proc ASCO 2013

27. Author’s Conclusions & Personal Comments Multiple Her axis members and CD73 were identified as candidate prognostic markers for patients treated with FOLFOX/FOLFIRI chemotherapy CD73 and Her3 were identified as candidate predictive markers for benefit/lack of benefit from cetuximab ➔ Candidate markers were selected from the literature and confirmed –only as candidates- in the study

28. + Predictive Genes & outcome KRAS wt population 144 KRAS wt mCRC treated with cetuximab 4-gene predictive classifier (AREG, EREG, DUSP6, SLC26A3) constructed using multivariate analysis with two-layer five-fold cross-validation Patients below the classifier cutpoint had PFS and disease control rates similar to those of patients with KRAS mutant mCRC. Khambata-Ford, S. et al. J Clin Oncol 2007;25:3230-3237. Baker JB, et al. Br J Cancer 2011;104:488-495

29. These results –Merit confirmation in other randomized studies –Suggest potentially intersecting roles for tumor inflammation, Her family cross talk, and Ras signaling –Suggest potential novel patient selection and combination treatment approaches ➔ Important correlative translational study: ➔ Negative, prematurely closed study, poor recruitment ➔ In this circumstances, high tumor acquisition rate (43%) ➔ Hypothesis generating study ➔ No power to confirm any previous finding… ➔ …but raises new potential prognostic/predictive biomarkers to be explored in the future Author’s Conclusions & Personal Comments

30. What’s next? Reproduce these results in large cohorts datasets, especially for the new candidates, i.e. CD73 Evaluate these proposed candidates in trials with Panitumumab, in order to address the class-effect or the Ig subtype-effect (IgG1 vs IgG2) Integrate different correlative studies into large datasets Make the results open access

31. Acknowledgements ASCO Scientific & Educational Committee The authors, especially Herb Hurwitz, for the very constructive interaction Special claim for implementing more correlative studies and for data sharing for the best interest of our patients