Association of enrollment duration on trial outcome in phase 3 randomized controlled trials: A pancancer analysis spanning 67 years.

Abstract

e13597 Background: Randomized controlled trials (RCTs) inform the standard of care in clinical oncology, yet a substantial proportion yield negative results. Difficulty with patient enrollment is a problem commonly encountered in the conduct of large multicenter RCTs, leading to slow enrollment. Slow enrollment can lead to increased financial costs, delayed reporting, and decreased relevance of trial results if standards of care shift. The effects of slow enrollment on trial success are understudied. We hypothesized that longer enrollment periods would be inversely correlated with trial success, after adjustment for year of study. Methods: Phase 3 RCTs studying malignant hematologic and solid tumor conditions available on the HemOnc.org website through December 31, 2021, were included in the analysis. HemOnc.org contains nearly all phase 3 RCTs of systemic anticancer therapy registered on ClinicalTrials.gov (with results) or with publications indexed in PubMed. Start year and end year of enrollment and comparative efficacy of control and experimental arms of RCTs were extracted. RCTs were deemed positive if the experimental arms were superior in superiority trials or non-inferior in non-inferiority trials, or if control arms were inferior in superiority trials. The remaining RCTs were deemed negative. Logistic regression was used to analyze the relationship between length of enrollment period and trial outcome, adjusted for year of study enrollment start. Results: The analysis included 3072 phase 3 RCTs spanning 99 different cancer types, enrolling from 1955-2019. Breast cancer was the most studied condition (n=554 trials) followed by non-small cell lung cancer (n=314 trials). 1513 (49.2%) of RCTs reported positive trial outcomes. The overall median enrollment period was 4 years (range 1-16 years). Negative trials took 1 year longer, on average, to complete enrollment (Table). The logistic regression model showed that shorter enrollment periods were associated with trial success. Later enrollment start year was also independently associated with trial success (Table). Conclusions: Our study showed that shorter enrollment period was significantly associated with positive trial outcomes. The success rate of trials has also increased over time; this could reflect both more efficacious drugs as well as more conservative enrollment criteria. Our main limitation is that we do not have data on early termination or closure for futility; these two issues might arise in both fast and slow accruing studies, which would bias results towards the null. The results of this analysis emphasize the importance of rapid phase 3 trial enrollment.[Table: see text]