A Framework for Assessing the Viability of an Externally Controlled Arm for a Single-Arm Trial

OBJECTIVES: A single-arm trial with an external control arm (ECA) is one in which the patients in the control group do not participate in the trial. The number of approved ECA trials is on the increase, due largely to their practical advantages, including sample size and ethical considerations. Given this, we developed a framework to make initial assessments on the viability of an ECA in meeting regulatory requirements (ie, ECA checklist). The purpose of the ECA checklist is to inform exploratory discussions with regulatory agencies, who would expect study sponsors to establish the appropriateness of the proposed design, data sources, and statistical analyses.

METHODS: The ECA checklist is based on relevant United States Food and Drug Administration guidance documents and the literature on over 40 accepted ECA trials across different therapeutic areas. It is structured on the critical aspects of design and analytical considerations on the use of real-word data and past trials, presented as 10 thematic questions with recommendations on possible solutions to issues arising from each theme. We evaluated the checklist against 5 ECA trials.

RESULTS: Ideally, all thematic checklist responses would be affirmative to be most confident of regulatory acceptance of the ECA; non-affirmative responses would require adequate rationale, guided by the associated recommendations for progression. In our evaluation against 5 approved ECA trials, we obtained affirmative responses for at least 8 themes.

CONCLUSIONS: Use of the ECA checklist will enable decision-makers to make an initial viability assessment of the ECA and make an informed decision on required next steps to help ensure acceptance of the ECA design. Such steps may include an assessment of the feasibility of initiating a disease natural history study, an appointment with the relevant regulatory agency for explorative discussions, or an exploration of alternative study designs, such as a randomized controlled trial.