Clinical Equipoise for Trials of Novel Biologic Therapies, Therapeutic Success Rates, and Predictors of Success: A Meta-Analysis.

PURPOSE: The demand for more rapid access to novel biologic therapies than randomized controlled trials can deliver is a topic of ongoing study and debate. We aimed to inform this debate by estimating therapeutic success from phase III trials comparing novel biologic therapies with standard of care and identifying predictors of success. METHODS: This was a meta-analysis of phase III trials evaluating novel biologic therapies in advanced breast, colorectal, lung, and prostate cancers. Therapeutic success was defined as statistically significant results for the primary end point favoring novel biologic therapies. RESULTS: Of 119 included phase III trials (76,726 patients), therapeutic success was 41%, with a statistically significant relative reduction in disease progression and death for novel biologic therapies over standard of care of 20% and 8%. Therapeutic success did not improve over time (pre-2010, 33%; 2010 to 2014, 44%; P = .2). Predictors of success were a biomarker-selected population (odds ratio, 4.74; 95% CI, 2.05 to 10.95) and progression-free survival end point compared with overall survival (odds ratio, 5.22; 95% CI, 2.41 to 11.39). Phase III trials with a biomarker-selected population showed a larger 28% progression-free survival benefit than phase III trials overall (hazard ratio, 0.72; 95% CI, 0.70 to 0.75) but similar 8% overall survival benefit (hazard ratio, 0.92; 95% CI, 0.90 to 0.94). Therapeutic success of phase III trials with and without a preceding phase II trial were 43% and 30%, respectively. CONCLUSION: Therapeutic success of novel biologic therapies in phase III trials, including therapies with a matching predictive biomarker, was modest and has not significantly improved over time. Equipoise remains and supports the ongoing ethical and scientific requirement for phase III randomized controlled trials to estimate treatment efficacy and assess the value of potential biomarkers.