ABSTRACT
BACKGROUND: In a competitive landscape with many ongoing adjuvant randomised controlled trials (RCTs), the prevalence of trials that failed to recruit their targeted sample size and were inadequately powered is unclear. The aims of the study are (i) to determine the percentage of trials with accrual and statistical power failure and (ii) to evaluate their potential impact on the drug development process. MATERIALS AND METHODS: A systematic review was carried out to identify adjuvant phase III oncology RCTs reported between 2013 and 2023 across all solid tumours. No restrictions were applied regarding the type of intervention or journal of publication. The percentage of trials with accrual failure and power failure was estimated as well as their association with the efficacy endpoints. Logistic regression models were used to estimate the odds ratio (OR) and its 95% confidence interval (CI). RESULTS: A total of 282 RCTs met the inclusion criteria with a median sample size of 661 patients and a median accrual period of 4.3 years. Most of these studies were superiority trials (83.0%). Accrual failure was observed in 22.0% of the studies, finishing recruitment without achieving the targeted sample size. Overall, 39.7% of the studies experienced power failure, having less power than specified in the protocol at the date of the read-out. Among superiority RCTs evaluating intermediate survival endpoints, only 31.1% presented statistically significant results. Trials with power failure were less likely to present statistically significant results (37.9% versus 21.9%, P = 0.04). The association was consistent across all cancer types. In the subset of non-inferiority trials, 35.0% formally demonstrated non-inferiority of the experimental arm. CONCLUSIONS: Nearly 40% of adjuvant phase III RCTs experienced power failure, and the reduction in power significantly impacted the final study results. There is a need for procedural refinements in the design and implementation of future adjuvant RCTs to mitigate these fallacies.
