1. A Meta-Analysis of Overall Survival Data from Three Trials of Bevacizumab and First-Line Chemotherapy as Treatment for Patients with Metastatic Breast Cancer Baylor-Sammons Cancer Center, Texas Oncology, US Oncology, Dallas, TX; Mount Vernon Cancer Centre, London, England; Dana-Farber Cancer Institute, Boston, MA; Institut Curie, Paris, France; Mayo Clinic, Jacksonville, Florida; Michiana Hematology Oncology, South Bend, IN; Vall d'Hebron University Hospital, Barcelona, Spain; BioOncology, Genentech, S San Francisco, CA; Indiana University Melvin and Bren Simon Cancer Center, Indianapolis, IN Joyce O’Shaughnessy, David Miles, Robert Gray, Véronique Diéras, Edith A. Perez, Robin Zon, Javier Cortés, Xian Zhou, See-Chun Phan, Kathy Miller ASCO, 2010

2. 2 Introduction Bevacizumab (BV), a monoclonal antibody, inhibits vascular endothelial growth factor (VEGF), a key mediator of angiogenesis. 3 randomized trials (E2100, AVADO, RIBBON-1) have demonstrated significantly improved progression-free survivial (PFS) for BV combined with different chemotherapies as first-line metastatic breast cancer (MBC) treatment. PFS improved when BV combined with chemotherapy regardless of hormone receptor status, sites of metastases, disease-free interval (DFI), or prior adjuvant taxane use.

3. 3 General Study Designs Optional Second- line Chemo + BV ( AVADO and RIBBON-1 only ) Chemo + No BV Chemo + BV Treat until PD RANDOMIZE Previously Untreated MBC RIBBON-1 Capecitabine, Taxane, or Anthracycline AVADO Docetaxel E2100 Paclitaxel

4. 4 BV=bevacizumab, PL=placebo, PFS=progression-free survival, ORR=objective response rate, OS=overall survival. * Permitted continuing on BV or crossing over to BV. † Analyses based on IRF assessments. ‡ 15 mg/kg cohort. Comparison of the Studies E2100AVADO*RIBBON-1* No. of patients722488 ‡ 1237 GeographyUS (90%)Ex-USUS (50%) Randomization ratio (BV:PL) 1:1 2:1 Chemotherapy Paclitaxel weekly Docetaxel Capecitabine, Docetaxel/nab-Paclitaxel, Doxorubicin/Epirubicin Primary EndpointPFS † PFS Key Secondary Endpoints OS, ORR OS, ORR, 1-yr survival

5. 5 Overview of Efficacy Results from the Individual Studies in the Pooled Analysis E2100AVADO RIBBON-1 (Cape) RIBBON-1 (Tax/Anthra) Non- BV BV Non- BV BV* Non- BV BV Non- BV BV Median PFS, mo 5.811.37.98.85.78.68.09.2 Stratified HR (95% CI) 0.48 (0.39–0.61) 0.62 (0.48–0.79) 0.69 (0.56–0.84) 0.64 (0.52–0.80) p-valuesp<0.0001p=0.0003p=0.0002p<0.0001 BV=bevacizumab, Cape=capecitabine, Tax/Anthra=taxane/anthracycline. * 15 mg/kg cohort.

6. 6 Objective To quantify treatment benefit of BV in combination with first-line chemotherapy for MBC patients by analyzing patient data from 3 randomized controlled trials in a pooled fashion

7. 7 Statistical Methods Efficacy and safety data were pooled from the studies E2100, AVADO, RIBBON-1: Cape, and RIBBON-1: Tax/Anthra. PFS was defined in the primary analysis of the individual study statistical analysis plans. Stratified HR estimate from Cox model used the individual studies (E2100, AVADO, RIBBON-1: Cape, RIBBON-1: T/Anth) as the only stratification factor. Kaplan–Meier curves were used to determine duration of PFS and OS. Median OS followup: StudyMedian OS followup (mo) E210035 AVADO29 RIBBON-1: Cape23 RIBBON-1: Tax/Anthra26

8. 8 Patient Characteristics, Pooled Population Non-BV (n  1008) BV (n  1439) Age, median 55 yr56 yr Triple-negative disease, %2625 Disease-free interval (≤24 mo), %3937 Prior adjuvant chemo, %6462 Taxane2224 Anthracycline5248 Visceral disease, %7169 ≥3 metastatic sites, %3841

9. 9 Progression-Free Survival, Pooled Population Non-BV (n=1008) BV (n=1439) Median, mo6.79.2 HR (95% CI)0.64 (0.57–0.71)

10. 10 Analysis of PFS by Subgroups

11. 11 Objective Response Rate* *Includes only patients with measurable disease at baseline. Non-BV (n=788) BV (n=1105) 50 0 45 40 35 30 25 20 15 10 5 32 49 Objective response rate (%)

12. 12 Overall Survival, Pooled Population Non-BV (n=1008) BV (n=1439) Median, mo26.426.7 HR (95% CI)0.97 (0.86–1.08) 1-yr survival rate (%) 7782

13. 13 Analysis of OS by Subgroups

14. 14 Use of Subsequent Systemic Therapies in AVADO and RIBBON-1 Studies* % Non-BV (n  654) BV (n  1071) Any chemotherapy7165 Bevacizumab5140 Any hormonal therapy2523 # of subsequent anti-cancer agents ≥42723 31512 22726 11015 *Data not available from E2100.

15. 15 Safety, Causes of Death* % Non-BV (n  982) BV (n  1679) Total deaths55.851.3 MBC51.547.4 Treatment-related1.82.1 Other1.41.5 Missing1.00.3 *Safety evaluable patient population.

16. 16 Grade ≥3 Selected Adverse Events (AEs), Pooled Population % Non-BV (n  982) BV (n  1679) Neutropenia7.110.0 Sensory neuropathy8.59.5 Hypertension1.29.0 Febrile neutropenia3.56.5 Venous thromboembolic event3.82.8 Proteinuria02.3 Arterial thromboembolic event0.31.6 Bleeding0.41.5 Left ventricular systolic function0.21.5 Wound dehiscence0.30.8 Fistula0.30.5 GI perforation0.30.5 RPLS0<0.1 RPLS=Reversible posterior leukoencephalopathy syndrome.

17. 17 PFS -HR=0.64, 36% reduction in risk of PD or death -2.5 month improvement in median PFS -Improvements across key clinical subpopulations ORR -17% increase vs controls OS -No statistically significant difference Summary of Pooled Efficacy Analysis

18. 18 Conclusions Significant PFS advantage but no OS difference with BV across 3 first-line studies and in pooled analysis -In MBC, the ability of Phase III trials to demonstrate treatment effect upon OS depends on the duration of survival post-progression (SPP) -Higher chance of affecting OS in populations with short SPP (20 month SPP in these 3 first-line trials) Patients with adverse prognostic features benefit from BV as do patients with more indolent disease Low incidence of treatment-related deaths with BV Safety profile consistent with previous BV experience Broglio and Berry. 2009. J Natl Cancer Inst. 101:1642-49. Saad, Katz, and Buyse. 2010. J Clin Oncol. 28:1958-62. Burzykowski, et al. 2008. J Clin Oncol. 26:1987-92.

19. 19 Acknowledgments We would like to thank all of the patients and all of the investigators who participated in the E2100, AVADO, and RIBBON-1 studies from: AustraliaItalyRomaniaUkraine AustriaLithuaniaRussiaUnited Kingdom BelgiumMexicoSingaporeUSA BrazilNetherlandsSouth AfricaUruguay CanadaNorwaySouth Korea ChinaPanamaSpain FrancePeruSweden GermanyPhilippinesSwitzerland GreecePolandTaiwan GuatemalaPortugalThailand

20. 20 Back-up

21. 21 Database Cutoff Dates from First-line MBC Studies Used in Pooled Analysis E2100AVADORIBBON-1 PFS Feb, 2005 (Interim analysis) Gray, et al. JCO, 2009 (IRF results) Oct, 2007 Miles, et al. ASCO 2008 Jul, 2008 Robert, et al. ASCO 2009 OS Oct, 2006 US PI Apr 30, 2009 (Pre-specified OS update, includes exploratory PFS update) Miles, SABCS 2009 Feb 23, 2009 (Pre-specified OS update) Unpublished MBC=metastatic breast cancer, OS=overall survival, PFS=progression-free survival.