(C-158) Feasibility of Supplementing Single-Arm Trials with External Control Arms – Evaluation of German Real-World Data

Background: While single-arm trials (SATs) are gaining in popularity, they are often seen critically by regulators. Real-world data (RWD) may provide an opportunity to improve the interpretability of SATs by supplying external control arms.

Objectives: To determine the number of breast cancer and amyotrophic lateral sclerosis (ALS) SATs registered in the European Union (EU) for which an external control arm based on German RWD can be constructed.

Methods: We extracted trial information on all breast cancer or ALS SATs registered in the EU’s clinical trial registers between 2004 and 2023. For each trial, we determined whether inclusion/exclusion criteria and the primary endpoint could be identified in one of five German RWD sources. The main outcome measure was the number of trials that could be feasibly supplemented with an external control arm based on one of the RWD sources investigated. This was defined by any trial for which all important inclusion/exclusion criteria and at least one primary endpoint could be identified in a RWD source.

Results: We identified 379 breast cancer SATs and 11 ALS SATs. Taking into account all important eligibility criteria and the primary endpoints, 2 breast cancer SATs and 11 ALS SATs could feasibly be supplemented with a RWD-derived external control arm. In 93 breast cancer trials at least one outcome was ascertainable in a RWD source, and all important eligibility criteria were recorded in a RWD source for 35 trials. Nine ALS trials had at least one primary endpoint ascertainable in RWD sources, and 2 trials had all important eligibility criteria recorded in a RWD source.

Conclusions: For the investigated phenotypes and data sources, SATs with RWD-derived external control arms are rarely suitable to establish treatment effects of medicines when requiring all important eligibility criteria and a primary endpoint to be available. Externally controlled SATs should be designed with limitations of the source of external controls in mind.