Evolution of clinical benefit in cancer trials: comparing pivotal approval data with updated efficacy

Many oncology drugs are approved by the Food and Drug Administration (FDA) based on interim analyses from randomized controlled trials (RCTs) using immature efficacy data. Although expedited approvals expedite access, they create uncertainty about the durability of benefit and long-term efficacy. We aimed to systematically compare initial and updated results from RCTs supporting FDA oncology approvals, evaluating longitudinal changes in relative efficacy.

All phase II-III RCTs supporting FDA oncology drug approvals from 2006-2025 that reported both initial and updated results were included. Relative efficacy (overall survival [OS] and progression-/event-free survival [PFS/EFS]) between groups was quantified using restricted mean survival time (RMST) ratio, with changes over time expressed as Ratio of updated-to-initial RMST ratio (RR). Meta-analyses and meta-regressions were used to estimate pooled effects and identify trial-level factors. Multiple sensitivity analyses, including those based on RMST difference (RMSD), were conducted to assess robustness.

A total of 165 RCTs were included. For OS, mean RR was 1.052 (95% CI, 1.029-1.063) and meta-analytic RR 1.023 (1.015-1.031, I ² = 15.5%); for PFS/EFS, mean RR was 1.131 (1.102-1.153) and meta-analytic RR 1.099 (1.078-1.121, I ² = 69.2%), indicating significant relative survival improvement for experimental arms as follow-up matured. 84.7% of OS and 92.5% of PFS/EFS comparisons showed increased RRs. Sensitivity analyses showed normalized RMSDs of 0.072 months/year (OS) and 0.156 months/year (PFS/EFS) in experimental arms. Shorter initial follow-up and less mature control-arm data were associated with greater relative efficacy gains of experimental treatments.

With extended follow-up, most FDA-approved oncology drugs with available updated data showed more durable survival benefits. Given the inherent selection of trials with publicly available updates and reconstructable survival curves, these findings should not be interpreted as a global validation of FDA approval decisions, but they underscore the need for dynamic, evidence-maturity-based oversight with ongoing reassessment and strengthened post-marketing surveillance.