Number needed to harm (NNH) in patients with non-metastatic, castration-resistant prostate cancer (nmCRPC): A final analysis from the darolutamide (D), enzalutamide (E), and apalutamide (A) clinical trials.

326 Background: Second generation androgen receptor inhibitors (ARI) D, A and E, approved for the treatment of nmCRPC, offer benefit in overall survival (OS) and metastasis-free survival (MFS), but do so with different risk-benefit profiles. Contextualizing risks of adverse events (AEs), or harm to patients, can be challenging. NNH is a trial-based measure that evaluates the number of patients who need to receive treatment before a harmful outcome occurs with the intervention, compared to the control. Thus, a higher NNH reflects a lower incremental likelihood of harm. This study is an update to a previous NNH analysis, utilizing final data from the ARI randomized controlled trials (RCTs). Methods: MEDLINE and EMBASE were searched for relevant RCTs. AEs occurring in ≥5% of either RCT arm were extracted from the final ARAMIS (D), SPARTAN (A) and PROSPER (E) publications. NNH by AE (all grade, grade 3 & 4) were calculated over the duration of each RCT based on reported rates, using the inverse of the absolute risk increase [1 / (experimental AE rate - control AE rate)]. An exploratory analysis considered the application of results to a US population, assuming trial-based AE rates to be representative of real-world rates. Results: Results show a trend of higher (i.e., favorable) NNH for AEs for D in ARAMIS across all grade and grade 3-4 AEs (Table), indicating a lower likelihood of incremental harm across AEs including fatigue, falls, hypertension and fracture for D versus A and E. Focusing on fatigue, fall and fractures, and selecting treatment with the highest (D) and lowest (A or E) NNH for an estimated 24,000 US high risk nmCRPC patients annually, treatment with D would translate to 4,073 fewer all-grade fatigue events, 2,928 fewer falls, and 2,383 fewer fractures compared to treatment with A (falls) or E (fatigue or fracture). Conclusions: The findings show a consistent trend of higher NNH (favorability) for D compared to A and E. The differences in their respective AE profiles are noteworthy not only for patients, but also for healthcare systems considering the overall risk-benefit amongst these three ARIs. Comparing D, A, and E directly in a single study may further inform their comparative profiles.[Table: see text]