Progression-free survival in control arms of clinical trials: Analysis of FDA cancer drug approvals between 2014 and 2023.

11031 Background: FDA regularly approves new cancer drugs based on the results of surrogate endpoints, often using expedited programs. When assessing the efficacy of new anticancer treatments, the choice of the control arm plays a crucial role. Herein, we evaluated the progression-free survival (PFS) performance of control arms in trials that led to FDA approvals for oncology treatments. Methods: We investigated FDA drug approvals issued between January 2014 and December 2023 for the treatment of advanced cancers, available on the FDA.gov website. We analysed publications of phase III or double-arm phase II trials that prompted approvals. For each clinical trial, we collected: i) The assumed median PFS (amPFS) value of the control arm, generally indicated in the study protocol for the sample size calculation; ii) The control arm’s recorded median PFS (rmPFS) and its confidence interval (CI). Control arms of each trial were considered under- or out-performing if amPFS exceeded the highest CI of rmPFS or was inferior than the lowest CI of rmPFS, respectively. All other cases were defined as “within the range”. Results: We found 72 trials leading to new cancer drug approvals. Immune checkpoint inhibitors (ICIs) and small molecules were investigated in the majority of cases, in 27 (37.5%) and 19 (26.4%) trials, respectively. Control arms under- or out-performed in 24/72 (33.3%) and 8/72 (11.1%) cases, respectively. In the remaining 40/72 (55.5%) trials, control arms performances were “within the range”. Interestingly, in 12/27 (44.4%) trials leading to ICIs approval, control arms underperformed. We found that in 6/14 (42.9%), 5/16 (31.3%) and 5/16 (31.3%) trials leading to new approvals in lung, breast and genitourinary cancers, respectively, the control arms underperformed. In 18/24 (75.0%) underperforming control arms, the discrepancy between amPFS and rmPFS was larger than 20%. Among the 19 trials with underperforming control arm and leading to expedited approval, in 7 (36.8%) cases a final overall survival (OS) advantage was not observed. Conclusions: The PFS benefit observed in several trials leading to FDA approval of new anticancer drugs may have been influenced by shorter-than-expected PFS in the control arms. Accurate analysis of control arms outcomes in clinical trials is essential for a comprehensive assessment of the efficacy of experimental drugs.