Feasibility of Randomized Controlled Trials for Cancer Drugs Approved by the Food and Drug Administration Based on Single-Arm Studies.

Background: The US Food and Drug Administration (FDA) introduced an Accelerated Approval (AA) pathway to expedite patient access to new drugs. AA accepts less rigorous trial designs, including single-arm studies (SAS), owing to perceived lack of feasibility of timely randomized controlled trials (RCTs). Methods: We designed hypothetical RCTs with endpoints of overall response rate (ORR), progression-free survival (PFS), and overall survival (OS) for FDA approvals based on SAS for solid tumors during 2010-2019. Existing standards of care served as controls. RCTs were designed to detect a difference with power of 0.80, α-error of 5% (2-sided), and 1:1 randomization. Accrual duration was estimated based on participation by less than 5% of eligible patients derived from cancer-specific incidence and mortality rates in the United States. Results: Of 172 (18.0%) approvals during the study period, 31 (18.0%) were based on SAS. Median sample size was 104 (range = 23-411), and 77.4% were AA. All studies reported ORR, 55% reported duration of response, 19.4% reported PFS, and 22.5% reported OS. Median sample sizes needed to conduct RCTs with endpoints of ORR, PFS, and OS were 206, 130, and 396, respectively. It would have been theoretically possible to conduct RCTs within duration comparable with that required by SAS for 84.6%, 94.1%, and 80.0% of approvals with endpoints of ORR, PFS, and OS, respectively. Conclusion: An overwhelming majority of FDA approvals based on SAS should be feasible as RCTs within a reasonable time frame. Given the collateral harms to patients and to scientific rigor, drug approval based on SAS should only be permitted under exceptional circumstances.