Proof of concept and design of an externally controlled trial for patients with gastro-enteropancreatic neuroendocrine carcinomas based on the randomized phase II BEVANEC trial

Aim: Randomized trials are very challenging to perform in rare cancers such as gastro-enteropancreatic neuroendocrine carcinomas. We hypothesized that the randomized BEVANEC trial of FOLFIRI + /- bevacizumab could have been designed with an external control arm (ECA) and led to similar results. Methods: For this proof of concept, BEVANEC-EMU was emulated as a prospective single experimental arm (EA) trial of FOLFIRI + bevacizumab (BEVANEC data) vs an ECA of FOLFIRI from real world data (RWD) with similar inclusion criteria and primary endpoints. Results: RWD characteristics were similar to those of BEVANEC. Missing data were observed for some patients in RWD (LDH, PAL and performance status). After weighting by the inverse of propensity-score, patient's characteristics were as well balanced between the ECA –n = 66) and the EA of BEVANEC-EMU (n = 59) as between the 2 randomized arm of BEVANEC trial. The 6-months OS rates were 61.1 % [95 %CI 48.9;76.5] in the EA and 53.4 % [95 %CI 42.2;67.6] in the ECA. In BEVANEC-EMU, the primary endpoint was met in both arms, as observed in the BEVANEC trial. The same conclusions were drawn when a hybrid ECA (generated by a mixture of RWD and prospective data from BEVANEC). A custom randomization algorithm which incrementally incorporate RWD into a prospective comparative externally controlled trial is described. Conclusions: The BEVANEC trial could have been led as an externally-controlled trial. This proof of concept led to the design of REWENEC-01, a funded prospective comparative externally-controlled trial for the evaluation of an immunotherapy combination in association with FOLFIRI in GEP NEC patients.