How to improve the quality of comparisons using external control cohorts in single-arm clinical trials?

TL;DR

This paper proposes a structured 3-step methodology to improve the validity of indirect comparisons using external control cohorts in single-arm clinical trials, especially relevant for accelerated drug approvals in oncology.

Contribution

It introduces a detailed 3-step process for incorporating external controls into single-arm trials to enhance evidence quality, addressing a critical gap in non-randomized study comparisons.

Findings

A 3-step process for external control integration is proposed.

Careful methodological considerations are crucial for valid indirect comparisons.

Post-market evaluation remains vital due to inherent limitations.

Abstract

PURPOSE Providing rapid answers and early acces to patients to innovative treatments without randomized clinical trial (RCT) is growing, with benefit estimated from single-arm trials. This has become common in oncology, impacting the approval pathway of health technology assessment agencies. We aimed to provide some guidance for indirect comparison to external controls to improve the level of evidence following such uncontrolled designs. METHODS We used the illustrative example of blinatumomab, a bispecific antibody for the treatment of B-cell ALL in complete remission (CR) with persistent minimal residual disease (MRD). Its approval relied on a single-arm trial conducted in 86 adults with B-cell ALL in CR, with undetectable MRD after one cycle as the main endpoint. To maximize the validity of indirect comparisons, a 3-step process for incorporating external control data to such single-arm trial data is proposed and detailed, with emphasis on the example. RESULTS The first step includes the definition of estimand, i.e. the treatment effect reflecting the clinical question. The second step relies on the adequate selection of external controls, from previous RCT or real-world data (RWD) obtained from patient cohort, registries, or electronic patient files. The third step consists in chosing the statistical approach targeting the treatment effect of interest, either in the whole population or restricted to the single-arm trial or the external controls, and depending on the available individual-level or aggregrated external data. CONCLUSION Validity of treatment effect derived from indirect comparisons heavily depends on carefull methodological considerations that are included in the proposed 3-step procedure. Because the level of evidence of a well conducted RCT cannot be guaranteed, post-market authorization evaluation is even more important than in standard settings.