Final results of the phase III, randomised, double-blind AVOREN trial of first-line bevacizumab + interferon-a2a in metastatic renal cell carcinoma Escudier B, Bellmunt J, Negrier S, Bajetta E, Melichar B, Bracarda S, Ravaud A, Golding S, Jethwa S on behalf of the AVOREN investigators

Disclosures Honoraria from Antigenics, Bayer Healthcare, Genentech, GSK, Novartis, Pfizer, F. Hoffmann-La Roche Ltd, Wyeth Consultant for Antigenics, Bayer Healthcare, Inate Pharma, F. Hoffmann-La Roche Ltd, Novartis, Wyeth

AVOREN study design Endpoints Treatment primary:* OS IFNa + Bevacizumab (n=327) Nephrectomised patients with advanced RCC (n=649) Stratification: Country MSKCC risk group PD 1:1 IFNa + placebo (n=322) PD Endpoints primary:* OS secondary: PFS, TTP, TTF, RR, safety Treatment bevacizumab/placebo 10mg/kg i.v. q2w IFN 9MIU s.c. t.i.w. (maximum 52 weeks) *PFS is the primary endpoint for regulatory approval in the USA Escudier, et al. Lancet 2007

Summary of published AVOREN data Final analysis of PFS performed at the time of the interim analysis significant increase from 5.4 to 10.2 months when bevacizumab is combined with IFN (HR=0.63; p=0.0001)1 good safety profile By reducing the IFN dose for safety issues PFS benefit is maintained2 decreased incidence of grade 3/4 events2 1. Escudier, et al. Lancet 2007; 2. Melichar, et al. Ann Oncol 2008

Independent review of PFS and ORR Investigator1 IRC2 IFN + Bevacizumab (n=327) IFN + placebo (n=322) IFN + Bevacizumab (n=288) IFN + placebo (n=281) ORR (%) 31 13 12 p value <0.0001 Median PFS (months) 10.2 5.4 10.4 5.5 HR (95% CI) 0.63 (0.52–0.75) 0.57 (0.45–0.72) 1. Escudier, et al. Lancet 2007; 2. Roche, data on file

Objectives Final analysis of OS Clinical cut-off September 2008 Median follow-up: 22 months Statistical considerations required 445 events from 649 randomised patients 80% power to detect an improvement in OS from 13 to 17 months corresponding to an HR of 0.76 at a two-sided overall significance level of 0.05

Final OS: unstratified and stratified analyses Cox regression p value HR 95% CI Log-rank Wilcoxon Unstratified 0.91 0.76–1.10 0.3360 0.2046 Stratified* 0.86 0.72–1.04 0.1291 0.0969 *Stratified by Motzer score and region

Probability of survival Final OS 1.0 0.8 0.6 0.4 0.2 Probability of survival IFN + Bevacizumab (n=327) IFN + placebo (n=322) HR=0.86 (95% CI: 0.72–1.04) p=0.1291 (stratified*) 21.3 23.3 0 6 12 18 24 30 36 42 Time (months) Patients at risk (n) IFN + Bevacizumab 327 278 237 194 157 124 84 27 IFN + placebo 322 262 216 177 141 113 78 22 *Stratified by Motzer score and region

Multiple Cox regression analysis for OS A multiple Cox regression model controls for several predetermined baseline prognostic factors that influence survival independent of treatment Variables included in the analysis gender, age, Motzer score, location of metastases (lung, bone, liver), body weight loss, number of sites, baseline SLD, region, baseline VEGF, and some lab values (albumin, creatinine, alkaline phosphatase, WBC count, platelets) Adjustment for these factors resulted in an improved treatment effect HR=0.78 (95% CI: 0.63–0.96); p=0.0219

Final OS in reduced-dose IFN population 1.0 0.8 0.6 0.4 0.2 Probability of survival Bevacizumab + reduced-dose IFN (n=131) Bevacizumab + IFN (n=327) 23.3 26.0 0 6 12 18 24 30 36 42 Time (months) Patients at risk (n) Bevacizumab + reduced-dose IFN 131 116 102 85 72 57 38 16 Bevacizumab + IFN 327 278 237 194 157 124 84 27

Censoring patients at time of crossover (n=13) IFN + Bevacizumab (n=327) IFN + placebo (n=322) Patients with event, n (%) 220 (67.3) 224 (69.6) Patients without events, n (%) 107 (32.7) 98 (30.4) Median OS, months (95% CI) 23.3 (20–27) 20.8 (18–24) HR (95% CI) 0.84 (0.70–1.02) p value (Log-rank test)* 0.0766 Derived measure *Stratified by Motzer score and region

Censoring crossover patients 1.0 0.8 0.6 0.4 0.2 Probability of survival IFN + Bevacizumab (n=327) IFN + placebo (n=322) HR=0.84 (95% CI: 0.70–1.02) p=0.0766* 20.8 23.3 0 6 12 18 24 30 36 42 Time (months) Patients at risk (n) Bevacizumab + IFN 327 278 237 194 157 124 84 27 IFN + placebo 322 262 216 173 131 101 69 19 *Stratified by Motzer score and region

Summary of subsequent medical therapies Treatment, n (%) IFN + Bevacizumab (n=327) IFN + placebo (n=322) Total patients with ≥1 treatment 180 (55) 202 (63) VEGF inhibitors Sunitinib 83 (25) 92 (29) Sorafenib 60 (18) 50 (16) Bevacizumab 10 (3) 12 (4) Other* 7 (2) 6 (2) mTOR inhibitors‡ 14 (4) Cytokines 32 (10) 52 (16) Chemotherapy§ 28 (9) 47 (15) *Protein TKI, pazopanib, erlotinib, blinded sorafenib, blinded sunitinib, angiogenesis inhibitors NOS, VEGF inhibitor NOS ‡Temsirolimus, everolimus (RAD001) §Antimetabolites, vinca alkaloids and antineoplastic agents

Regional variation in subsequent treatment with TKIs IFN + Bevacizumab IFN + placebo Therapy Western Europe (n=180) Eastern Europe and other (n=147) Western Europe (n=177) Eastern Europe and other (n=145) TKIs, n (%) Sunitinib 52 (29) 31 (21) 64 (36) 28 (19) Sorafenib 51 (28) 9 (6) 48 (27) 2 (1)

Regional variation in OS IFN + Bevacizumab IFN + placebo Western Europe (n=180) Eastern Europe and other (n=147) Western Europe (n=177) Eastern Europe and other (n=145) Events, n (%) 120 (67) 100 (68) 125 (71) 99 (68) Median OS, months (95% CI) 24.5 (21–29) 23.1 (17–27) 23.7 (21–28) 17.1 (13–22) HR (95% CI)* 0.93 (0.71–1.21) 0.92 (0.71–1.20) p value (log-rank)* 0.5922 0.5336 *Stratified by Motzer score and region

OS by post-protocol therapies IFN + Bevacizumab vs IFN + placebo (n) Median OS IFN + Bevacizumab (months) IFN + placebo (months) HR (95% CI) Subsequent TKI*‡ 113 vs 120 38.6 33.6 0.80 (0.56–1.13) Subsequent sunitinib 83 vs 92 43.6 39.7 0.88 (0.58–1.35) Subsequent sorafenib 60 vs 50 30.7 0.73 (0.44–1.20) *Subsequent therapy defined as any post-protocol therapy, any line (before or after PD) ‡TKIs include sunitinib, sorafenib, pazopanib, erlotinib, blinded sorafenib, blinded sunitinib and unspecified protein TKI

Subgroup analysis of OS in AVOREN Baseline risk factor Total (n) HR HR 95% CI All patients 649 0.86 0.72–1.04 Sex Female Male 192 457 0.90 0.84 0.64–1.27 0.66–1.05 Age (years) <65 ≥65 410 239 0.78 0.99 0.61–0.99 0.73–1.35 Motzer score Favourable Intermediate Poor 180 366 55 0.86 0.83 0.57–1.30 0.65–1.06 0.47–1.59 Baseline VEGF > median No Yes 192 0.74 0.88 0.50–1.10 0.62–1.24 Per cent body weight loss 10 >10 501 81 0.88 0.60 0.70–1.09 0.35–1.04 0.2 0.5 1 2 5

Subgroup analysis of OS in AVOREN (cont’d) Baseline risk factor Total (n) HR HR 95% CI No. of metastatic sites 1 2 >2 152 242 252 0.81 1.02 0.74 0.52–1.27 0.73–1.41 0.55–0.99 Lung metastases No Yes 173 473 1.10 0.77 0.73–1.66 0.61–0.96 Tumour in lung only No Yes 565 81 0.88 0.68 0.72–1.08 0.36–1.28 Bone metastases No Yes 521 125 0.87 0.88 0.70–1.09 0.58–1.32 Liver metastases No Yes 508 138 0.76 1.30 0.62–0.95 0.85–1.98 0.2 0.5 1 2 5

OS in patients without liver metastases at baseline Probability of survival 1.0 0.8 0.6 0.4 0.2 Median OS IFN + Bevacizumab (n=258) IFN + placebo (n=250) HR=0.76 (95% CI: 0.62–0.95) p=0.0155* 21.4 27.5 0 6 12 18 24 30 36 42 Time (months) Patients at risk (n) Bevacizumab + IFN 258 228 204 169 140 112 77 24 IFN + placebo 250 211 173 140 111 91 61 17 *Stratified by Motzer score and region

Conclusions Although not statistically significant, a trend towards improved OS was observed with bevacizumab + IFN combination. The results have been confounded by post-protocol bevacizumab subsequent therapies Patients with bevacizumab + reduced doses of IFN had similar OS to the whole bevacizumab + IFN population Treatment effect is significant when adjusted for prognostic factors This analysis confirms the activity of this treatment in first line metastatic RCC with good or intermediate risk

Acknowledgements All patients who participated All investigators, nurses and study coordinators F. Hoffmann-La Roche Ltd