Background: There are no head-to-head comparisons evaluating the efficacy of the main polychemotherapy regimens used for patients with pancreatic cancer in the adjuvant setting. We aimed to describe the relative efficacy of modified FOLFIRINOX (mFOLFIRINOX), gemcitabine plus capecitabine (GEM-CAP) and gemcitabine plus nab-paclitaxel (GEM-NAB) in this setting using a Bayesian network approach.

Methods: We collected data from the ESPAC-4, PRODIGE 24 and APACT trials. Disease-free survival (DFS), according to the investigators, and overall survival (OS) for the three polychemotherapy regimens were compared using gemcitabine as the reference arm. We ran Markov chain Monte Carlo simulations with a fixed-effect model to generate the posterior distribution of the hazard ratios (HRs) using non-informative priors. Relative efficacy was measured by HRs, surface under cumulative ranking and rankograms.

Results: mFOLFIRINOX was the chemotherapy regimen most likely to be the most effective in the adjuvant setting (98.9% and 89.6% probability for DFS and OS, respectively). GEM-NAB marginally improved DFS (HR = 0.97, 95% credible interval (95% CrI) = 0.77–1.21) and OS (HR = 0.98, 95% CrI = 0.76–1.25) when compared to GEM-CAP. However, GEM-NAB had the highest chances of being the second most active chemotherapy regimen (61.4% and 52.5% probability for DFS and OS, respectively), whereas GEM-CAP was less likely to represent the second most active regimen (37.7% and 40.1% probability for DFS and OS, respectively).

Conclusion: For patients eligible and fit enough to undergo adjuvant treatment with mFOLFIRINOX, this constitutes the treatment of choice. For those with contraindications to mFOLFIRINOX, while both GEM-NAB and GEM-CAP can be considered appropriate alternatives, GEM-NAB is likely the most effective regimen.