1. A PHASE 2 STUDY OF BEVACIZUMAB PLUS ERLOTINIB IN PATIENTS WITH GEMCITABINE- REFRACTORY METASTATIC PANCREATIC CANCER Poster 4 AH Ko, E Dito, B Schillinger, AP Venook, EK Bergsland, WM Korn, MA Tempero University of California, San Francisco Comprehensive Cancer Center, San Francisco, CA

2. Background Fewer than half of patients with advanced pancreatic cancer receive additional systemic therapy following progression on a gemcitabine- based regimen 1 There is no standard of care second-line therapy for advanced pancreatic cancer Erlotinib and/or bevacizumab have been tested in combination with gemcitabine in the first-line setting 2-5  Erlotinib plus gemcitabine was approved for use in this setting in 2005 based on a small, but significant, survival advantage when erlotinib was added to gemcitabine 4 Since dual inhibition of EGFR and VEGF may confer additive or even synergistic benefit, 6-8 we tested the combination of erlotinib and bevacizumab in patients with previously treated advanced pancreatic cancer 1 Schrag D, et al. J Clin Oncol. 2007;25:203s. Abstract 4524; 2 Kindler HL, et al. J Clin Oncol. 2005;23:8033–8040; 3 Kindler HL, et al. J Clin Oncol. 2007;25:199s. Abstract 4508; 4 Moore MJ, et al. J Clin Oncol. 2007;25:1960-1966; 5 Kindler HL, et al. J Clin Oncol. 2006;24. Abstract 4040; 6 Petit AM, et al. Am J Pathol. 1997;151:1523–1530; 7 Herbst RS, et al. J Clin Oncol. 2005;23:2544–2555; 8 Dancey JE, Chen HX. Nat Rev Drug Discov. 2006;5:649–659.

3. Study Endpoints Primary endpoints  Overall survival at 6 months  Safety Secondary endpoints  Response rate  TTP  Proportion of patients with at least a 50% decline in serum CA19-9 levels

4. Study Design Treatment continued until Disease progression Intercurrent illness Unacceptable toxicity Patient withdrawal Day 1 21 Bevacizumab (15 mg/kg) Cycle 1Additional cycles Erlotinib 150 mg daily CA 19-9 measurements every cycle 42 CT scans every 2 cycles

5. Eligibility Criteria Histologically confirmed diagnosis of pancreatic cancer Extrapancreatic metastases Previous treatment with up to 3 systemic therapies for advanced disease, including at least 1 gemcitabine-based regimen Previous therapy with either a VEGF/VEGFR inhibitor or an EGFR inhibitor, but not both, is permitted

6. Patient Characteristics CharacteristicPatients (N = 32) Median age, years (range) 60 (36 – 82) Female:male, % 56:44 Ethnicity, no. (%) Caucasian (non-Hispanic) Hispanic African American Asian American/Pacific Islander 24 (75) 3 (9) 2 (6) ECOG performance status, no. (%) 0 1 14 (44) 18 (56) Number of prior lines of systemic therapy, no. (%) 1 2 3 21 (66) 9 (28) 2 (6)

7. Patient Characteristics, cont’d CharacteristicPatients (N = 32) Prior VEGF/VEGFR therapy, no. (%) 8 (25) Prior EGFR therapy, no. (%) 7 (22) Prior Whipple resection, no. (%) 7 (22) Prior radiation therapy, no. (%) 9 (28) Elevated baseline CA19-9 levels (> 2 x ULN), no. (%) 21 (66)

8. Study Treatment Treatment cycles, median no. (range) 2 (1 – 7) Patients (n = 31) receiving 2 or fewer therapy cycles, no. (%) 23 (74) Patients requiring dose reduction of erlotinib, no. 4 Reasons for study discontinuation, no. (%) Progressive disease Self-withdrawal due to poor tolerance Adverse event Prolonged treatment interruption for palliative radiation therapy 25 (81) 2 (6) a 3 (10) b 1 (3) a One patient secondary to asthenia and grade 3 rash; 1 patient secondary to asthenia. b One patient with grade 3 upper gastrointestinal bleed; 1 patient with unexplained severe abdominal pain not related to disease progression; 1 patient with pulmonary embolism and increasingly symptomatic pleural effusion.

9. Treatment-Associated Toxicity Toxicity, no. (%) Evaluable patients (n = 30) Rash Grade 1 Grade 2 Grade 3 24 (80) 5 (17) 18 (60) 1 (3) Diarrhea Grades 1-2 Grade 3 8 (27) 7 (23) 1 (3) Hypertension (grade 3) 4 (13) Gastrointestinal bleeding Grade 1 Grade 2 Grade 3 2 (7) 1 (3) a 0 (0) 1 (3) b a Rectal bleeding likely secondary to hemorrhoids. b Upper gastrointestinal bleed secondary to duodenal ulcers. Patient had been on low-molecular weight heparin at the time.

10. Treatment-Associated Toxicity, cont’d Toxicity, no. (%) Evaluable patients (n = 30) Venous thromboembolic events Grade 2 Grade 3 Grade 4 (pulmonary embolism) 5 (17) 2 (7) a 0 (0) 3 (10) Dyspnea 2 (7) b Pancreatogastric fistula 1 (3) Suspected hemorrhage into intrapulmonic metastases 1 (3) a One patient had a Mediport-associated deep vein thrombosis and 1 had an asymptomatic superior mesenteric vein thrombosis. b Etiology was uncertain in these 2 cases. There was no evidence of interstitial pneumonitis.

11. Efficacy Measure (median follow-up 418 days)Patients (n = 28) Best response, no. (%) Partial response Stable disease ≥2 cycles Progressive disease 1 (3.6) 7 (25.0) a 20 (71.4) CA19-9 biomarker decline of ≥25% (patients with >2 x ULN as baseline, n = 21), no. (%) 4 (19.0) Estimated median TTP, days (95% CI) 40 (35 – 59) Estimated median overall survival, days (95% CI) 104 (87 – 133) Estimated 6-month survival rate, % 27 a Includes 1 patient with a decrease in size of peritoneal metastases, and 1 with decreased ascites.

12. Time to Tumor Progression 0 20 40 60 80 100 Patients without progression ( %) 20406080100120140 Days

13. Overall Survival 0 20 40 60 80 100 Patients surviving ( %) 0100200300400500600 Days

14. Outcomes Following Progression on Gemcitabine RegimennRR (%)Median survival Reference Paclitaxel185.617.5 weeksOettle, 2000 Raltitrexed1904.3 monthsUlrich-Pur, 2003 Raltitrexed/irinotecan19166.5 monthsUlrich-Pur, 2003 5-FU/LV/oxaliplatin3023.325 weeksTsavaris, 2005 23N/A21 weeksOettle, 2005 9-nitrocamptothecin58792 daysBurris, 2005 Pemetrexed523.820 weeksBoeck, 2007 Capecitabine/erlotinib30116.7 monthsKulke, 2007 Erlotinib/bevacizumab313.6104 daysKo, 2008

15. Conclusions and Future Directions The combination of bevacizumab and erlotinib has modest efficacy in patients with gemcitabine-refractory metastatic pancreatic cancer The combination of bevacizumab and erlotinib can be safety administered to patients with metastatic pancreatic cancer  Treatment-related toxicities, including rash, hypertension, and thromboembolic events, were observed in ranges that were expected when combining these 2 classes of agents We are currently measuring baseline and post-treatment levels of circulating tumor and endothelial cells to assess whether these cells have prognostic or predictive utility in this disease setting Effective treatment in this patient population remains a significant unmet need. Given the fragile nature and poor prognosis of this patient population, future studies should focus on refining the subset of patients most likely to benefit from salvage therapy