A Bayesian network meta-analysis of systemic regimens for advanced pancreatic cancer.

4015 Background: For advanced pancreatic cancer, many regimens have been compared with gemcitabine (G) in randomized control trials (RCTs). Few have been compared with each other directly in RCTs and the relative efficacy and safety among them are unclear. Methods: A systematic review was performed through MEDLINE, EMBASE, Cochrane Central Register of Contorlled Trials and ASCO meeting abstracts up to Jan 2013 to identify RCTs that included metastatic pancreatic cancer comparing the following regimens: G, G+5-flourouracil (GF), G+capecitabine (GCap), G+S1 (GS), G+cisplatin (GCis), G+oxaliplatin (GOx), G+erlotinib (GE), G+Abraxane (GA) and FOLFIRINOX. Studies were reviewed by two authors and discrepancies were resolved by consensus or by a third author. Data including overall survival (OS), progression-free survival (PFS), response rate (RR), and side-effects were extracted. A Bayesian network meta-analysis with random effects was performed using WinBUGS to compare all regimens simultaneously. Results: Twenty-two studies involving 6,252 patients were identified, with 21 RCTs involving G, 4 with GF, 3 with GCap, 2 with GS, 6 with GCis, 3 with GOx, 1 with GE, 1 with GA and 1 with FOLFIRINOX. Median OS, PFS and RR for G arms from all trials were similar, suggesting the absence of significant clinical heterogeneity among RCTs. For OS, the results of the Bayesian network meta-analysis found that the probability that FOLFIRINOX was the best regimen was 71%, while it was 19% for GS, 7% for GA and 2% for GE respectively. The OS hazard ratio (HR) for FOLFIRINOX vs. GS was 0.82 (95% credible region (CR): 0.53-1.35), the OS HR for FOLFIRINOX vs. GA was 0.77 (95% CR: 0.51-1.23), and the OS HR for FOLFIRINOX vs. GE was 0.67 (95% CR: 0.45-1.08). Similar ranking and probabilities were observed for the best regimen for PFS. Conclusions: FOLFIRINOX appeared to be the best regimen for advanced pancreatic cancer probabilistically, with a trend towards improvement in OS and PFS when compared with GS, GA, or GE by indirect comparisons. In the absence of direct pairwise comparisons of these regimens from RCTs, network meta-analysis helps synthesize evidence and inform decision making.