Background: Randomized nmCRPC trials comparing darolutamide (D), apalutamide (A) and enzalutamide (E) have not been reported. Safety of these therapeutics has important implications in assessing patient risk-benefit concerns. Matching-adjusted indirect comparison (MAIC) is a method to perform indirect treatment comparisons adjusting for cross-trial heterogeneity. Objective: To compare the safety outcomes of D vs. A or E using MAIC. Methods: Data from the ARAMIS (D vs. placebo [PBO]), SPARTAN (A vs. PBO) and PROSPER (E vs. PBO) trials were used. Key safety outcomes including adverse events (AEs) that have central nervous system relevance were compared using anchored MAIC. Individual patient level data (IPD) from ARAMIS were selected and re-weighted to match the inclusion criteria and baseline characteristics published in SPARTAN and PROSPER (no access to their IPD). The Benjamini-Hochberg approach was applied to adjust for multiplicity. The D vs A MAIC matched on 7 covariates: age, prostate-specific antigen (PSA) level and doubling time, Eastern Cooperative Oncology Group (ECOG), Gleason score, bone-sparing agent use and prior surgery. Sensitivity analyses were conducted matching on different sets of covariates. D vs. E were matched on age, region, PSA level and doubling time, ECOG, Gleason score and bone-sparing agent use. Risk difference (RD) ([DARO – PBO_ARAMIS] – [ENZA – PBO_PROSPER]) and odds ratio (OR) (OR_ARAMIS/OR_PROSPER) were calculated. RD<0 or OR<1 indicate lower AE risk for D. Results: For D vs. A, the effective sample sizes (ESS) of D and its placebo (PBO) arm were 604 and 391 after matching. Fall, fracture, and rash were statistically significantly lower for D vs. A (Table). For D vs. E, the ESS of D and PBO arm were 580 and 395, respectively. Fall, dizziness, mental impairment, hypertension, fatigue and severe fatigue were statistically significantly lower for D vs. E. Conclusions: After adjusting for trial differences, D showed favorable safety profile in fall, dizziness, mental-impairment, hypertension, rash, fatigue, and fracture.

^a All grades AEs with the exception of severe fatigue (grade 3+) * Raw and multiplicity adjusted p-value <0.05 ** Raw p-value <0.05 NR=not reported in PROSPER