Randomized controlled trials (RCTs) in oncology, especially in rare diseases, are often unfeasible for ethical, long recruitment periods and high cost reasons. In this setting, single-arm trials (SATs) are used, but inadequate for demonstrating the benefit of a new treatment. To address these challenges, innovative methodologies, such as trials incorporating external (or synthetic) control arms (ECAs) have emerged to complement the data from a SAT to move closer to the standards of RCTS for making comparisons. In the setting of hybrid ECA (Hyb_ECA, a control arm obtained by a mixture of prospective data and external or historical data), we proposed a dynamic algorithm of randomization to overcome some limitation of clinical trials incorporating an ECA (ECA-CT).

We have simulated a motivation ECA-CT in patients with neuroendocrine carcinomas (NECs) treated in the second line of chemotherapy. We used data from the BEVANEC RCT, with FOLFIRI+Bevacizumab as a treatment arm (RTA) and FOLFIRI as a control arm (RCA). The RCA was replaced with a Hyb_ECA composed of both a proportion p of RCA and a proportion 1 – p of data from two other studies (CEPD and RBNEC cohorts). A propensity score (PS) approach was used to balance measured confounders between the RTA and the Hyb_ECA. Overall survival as well as its bootstrap 95%CI was estimated in the Hyb_ECA and compared with that of the RCA. However, to maintain the benefit of randomization for future ECA-CT, we developed a dynamic minimization algorithm. Our approach relies on the method proposed by Han et al. (2009) and allows to gradually integrate during the inclusion phase, patients from external sources.

In the motivation ECA clinical trial, we obtained a Hyb_ECA similar to the RCA in term of patient's baseline characteristics and outcome define in BEVANEC trial (12-month overall survival was 32.4% in RCA and ranged from 32.23% to 32.88% in Hyb_ECA with 50 to 75% of external data). When applying the proposed randomization algorithm on a 1:1 ratio design with 75% external data in the hyb_ECA, RTA and Hyb_ECA were well balanced as well as the stratification factors (Fig. 1A/1B).

Externally controlled trials might be an appropriate compromise between RCTs and SATs in rare diseases. The proposed randomization algorithm could be an appropriate and useful tool when designing and performing future clinical trials involving Hyb_ECA.