The frequency of use and enrollment impact of patient-centered outcomes in prostate cancer clinical trials.

Abstract

115 Background: Cancer treatments should help patients live either longer or better by improving patient quantity or quality of life. However, some trial endpoints do not correlate with overall survival, and are not noticeable to patients. Selecting endpoints for clinical trials that reflect these goals could improve the discovery process by ensuring trial results are of direct interest to patients, and perhaps by improving enrollment rates to trials. However, how frequently prostate cancer clinical trials use patient-centered outcomes, and how outcome type impacts trial enrollment, is unknown. Methods: On October 23, 2022, we queried ClinicalTrials.gov for full text records of phase 2-3 prostate cancer trials started after January 1, 2007. We extracted primary outcomes and other trial characteristics using a custom Python script. Two reviewers categorized each outcome into categories (e.g., overall survival, response rate, patient reported measure). As previously studied, the only valid surrogate for overall survival is metastasis-free survival. We considered these two outcomes or any outcome directly noticeable to a patient to be a patient-centered outcome. For completed or terminated trials, we defined ‘sufficient accrual’ as attaining 85% of a trial’s goal enrollment. We identified associations between trial outcome types and sufficient trial enrollment with chi-square tests and logistic regression. Results: Of 1,717 prostate cancer trials, only 37% used a patient-centered outcome, with 6% using overall or metastasis-free survival. Among 318 Phase 3 trials, 49% used a patient-centered outcome and 26% used overall or metastasis-free survival. Of 731 completed or terminated prostate cancer trials, 55% of trials and 68% of phase 3 trials reached sufficient enrollment (85% of enrollment goal). On multivariable analysis, trials with an overall survival endpoint had higher odds of sufficient enrollment (OR 8.0 [95% CI 2.2-33.5], p < 0.01), but trials with any patient-centered outcome had lower odds of sufficient enrollment (OR 0.25 [95% CI 0.11-0.54], p < 0.01). Conclusions: Less than half of prostate cancer trials use an outcome that is clearly patient-centered (i.e., affecting overall survival or an outcome noticeable to a patient). Further work is needed to clarify the use, understanding, and effect of outcome selection in cancer trials. Realigning trial efforts with patient-centered goals could improve the efficiency, productivity, and applicability of clinical trials.