Abstract
BACKGROUND: Phase 3 studies of bevacizumab in patients with advanced pancreatic cancer (APCA) demonstrated no improvement in outcome. To the authors' knowledge, no validated predictive biomarkers for bevacizumab exist, although emerging data suggest that subsets of patients with APCA may benefit from treatment with bevacizumab. The authors evaluated baseline serum albumin (b-alb) as a predictive biomarker in a pooled analysis from 7 prospective clinical trials of gemcitabine-based therapy with or without bevacizumab. METHODS: Data were collected from individual databases from 7 prospective clinical trials. Patients were grouped by exposure to bevacizumab and by b-alb level (≥ 3.4 g/L or < 3.4 g/dL). Overall survival (OS), time to disease progression (TTP), overall response rate, and disease control rate (overall response rate plus stable disease lasting ≥ 16 weeks) were compared between groups. Univariate and multivariable analyses of prognostic factors were performed. RESULTS: A total of 264 patients were included. The median age was 59 years (range, 31 years-85 years) and all patients had stage IV disease per TNM staging. Normal b-alb was associated with significantly improved median OS (10.2 months vs 4.1 months; P = .0001), median TTP (6.2 months vs 3.7 months; P = 0.0488), and disease control rate (71% vs 46%; P = .007) for patients receiving bevacizumab, but not for those treated without bevacizumab. Multivariable analysis revealed a significant influence of normal b-alb on OS (P = .0008) and TTP (P = .033). CONCLUSIONS: Patients with APCA with normal b-alb derive benefit from treatment with bevacizumab. Future prospective investigations of bevacizumab in patients with APCA should consider selecting patients with normal b-alb to maximize potential benefit.