A systematic review and pooled analysis of hypothesized versus observed effect sizes in early phase cancer prevention clinical trials.

Early phase cancer prevention trials are designed to demonstrate safety, tolerability, feasibility, and signals of efficacy of preventive agents. Yet it is often observed that many trials fail to detect intervention effects. We conducted a systematic review and pooled analyses of recently completed early phase chemoprevention trials to gain in depth insight on the failure of detecting efficacy signals by comparing hypothesized effect sizes to the corresponding observed effect sizes. Single or multi-arm efficacy chemoprevention trials conducted under the Phase 0/I/II Cancer Prevention Clinical Trials Program of the Division of Cancer Prevention, National Cancer Institute between 2003 and 2019 were evaluated. A total of 59 chemoprevention trials were reviewed. Twenty-four studies were efficacy or biomarker trials with complete information on hypothesized and observed effect sizes and included in this analysis. The majority of the trials (n=18) were multi-arm randomized studies of which 15 trials were blinded. The pooled estimate of the observed to hypothesized effect size ratio was 0.57 (95% CI: 0.42-0.73, p<0.001) based on a random effects model. There were no significant differences detected in the ratio of observed to hypothesized effect sizes when conducting various subgroup analyses. The results demonstrate that the majority of early phase cancer chemoprevention trials have substantially smaller observed effect sizes than hypothesized effect sizes. Sample size calculations for early phase chemoprevention trials need to balance the potential detectable effect sizes with realistic and cost-effective accrual of study populations, thereby, detecting only intervention effects large enough to justify subsequent large scale confirmatory trials.