Bayesian Multi‐Criteria Decision Analysis of Net Clinical Benefit in ARASENS : Overall and Japanese Subgroup Results

OBJECTIVE: To assess the benefit-risk of adding darolutamide to androgen-deprivation therapy (ADT) plus docetaxel in ARASENS, in the overall population and the Japanese subgroup, using Bayesian multi-criteria decision analysis (MCDA). METHODS: Bayesian MCDA was conducted using published aggregate data from ARASENS. A three-criterion model was evaluated based on overall survival (OS), grade 3/4 adverse events, and treatment discontinuation. Additional models included five criteria in the overall cohort (adding time to castration-resistant prostate cancer [TTCRPC] and undetectable prostate-specific antigen) and four criteria in the Japanese subgroup (adding TTCRPC). Prespecified weights summing to one were applied, and 100 000 posterior simulations estimated the probability that the triplet provided greater utility (P[ΔU > 0]). Weight-scenario sensitivity analyses were conducted. RESULTS: In the overall cohort, the darolutamide triplet achieved net clinical benefit across weight scenarios except the toxicity-only, driven by OS/TTCRPC benefits with minimal safety penalties. In the Japanese subgroup, the three-criterion model yielded P(ΔU > 0) of 0.601, reflecting wide OS uncertainty and a higher discontinuation rate. When TTCRPC was incorporated, P(ΔU > 0) increased to 0.975. However, this increase was driven primarily by the TTCRPC component, while the OS contribution remained imprecise due to the small subgroup size. CONCLUSIONS: Bayesian MCDA integrated multiple clinical endpoints to examine how differing value preferences influence decision-making. Scenario-based analysis demonstrated a consistent benefit of adding darolutamide to ADT and docetaxel, while revealing that varying efficacy-safety weightings affected confidence in overall utility among Japanese.