Characterization of inactive control arms and post-protocol therapy in advanced and metastatic oncology trials.

96 Background: Inactive control arms using placebo or best supportive care (BSC) may be warranted to measure the risks and benefits of novel therapies in oncology randomized controlled trials (RCTs). However, in the advanced or metastatic setting, an inactive control arm is only ethical when no therapeutic options are available. This study assesses the frequency and potential impacts on endpoints of post-progression therapy in oncology RCTs conducted in these settings that led to Food and Drug Administration (FDA) approval using inactive control arms. Methods: We indexed all registered oncology RCTs leading to FDA approvals from 2009 to 2024 found on ClinicalTrials.gov, excluding RCTs conducted in maintenance, consolidation, or perioperative settings. We included RCTs comparing anticancer agents against placebo or BSC in the advanced or metastatic settings, where post-protocol data for the control arm were reported (n = 18). We descriptively analyzed trial characteristics, post-protocol therapies, crossover, and clinical outcomes. We used two-sample z-tests and Welch’s t-test for comparison. To investigate the association between post-protocol therapy rates and endpoint outcomes, we conducted a simple linear regression analysis. Results: Among trials comparing anticancer agents to placebo or BSC in the advanced or metastatic settings, median rates of post-progression treatment were higher in the control groups than in experimental groups (46.9% vs 29.4%; p < 0.001 (z-test)). Trials with crossover had a higher median rate of post-protocol therapy in the control arm than those without (66.0% vs 40.7%; p < 0.001 (z-test)). An inverse correlation was observed between the proportion of controls receiving subsequent treatment and the hazard ratio, with a 0.07 decrease in the hazard ratio for each 10% increase in the proportion of control patients receiving post-protocol care (R² = 0.56; p < 0.001). The rates of patients with post-protocol anticancer therapy were different between trials using overall survival (OS) or progression-free survival (PFS) as the regulatory endpoint (32.1% vs 67.2%; p = 0.0001 (Welch’s t-test)). Conclusions: Significantly more and nearly half of inactive control arm patients received subsequent anticancer therapies, suggesting an active control arm may have been indicated. Moreover, these differences are associated with a larger effect size. When comparing trials using PFS versus OS as primary endpoints for approval in our study, control-arm patients in PFS-endpoint trials are significantly more likely to receive post-protocol therapy. Consequently, the interpretation of PFS in these trials is more susceptible to differences in post-protocol therapies. Our findings advocate for greater scrutiny by regulators and ethics committees to ensure that vulnerable patients have access to appropriate, high-quality care.