OBJECTIVES: Treatment decision-making for metastatic castration-sensitive prostate cancer (mCSPC) is challenged by unclear comparative effectiveness and widely varied costs of multiple competing strategies. The objective of this study is to compare the effectiveness and safety of systematic therapies for mCSPC.

METHODS: Bibliographic databases (Medline, Embase, and Central), regulatory documents (FDA and European Medicines Agency), and trial registries (ClinicalTrials.gov and EU trial register) were searched from inception to November 2019 for randomized controlled trials testing active drugs that added to androgen deprivation therapy (ADT) for mCSPC. Cochrane risk-of-bias tool version 2 were used to assess trial quality. Bayesian network meta-analysis (NMA) was used to estimate relative effects of competing strategies. In addition to combing published constant hazard ratios (HR), we reconstructed survival data from Kaplan Meier curves to enable parametric survival NMA and that allows time-varying HR.

RESULTS: Seven trials with 7,236 patients were included. Risk of bias is a concern for trials with open label, missing data, or unprespecified analysis. Ordered from the most to the least effective, treatments that improved the overall survival are abiraterone, apalutamide, and docetaxel, HR (95% credible interval [CI]) 0.64 (0.56-0.73), 0.67(0.51-0.88), and 0.80 (0.72-0.89); treatments that improved radiographic progression-free survival (rPFS) are: enzalumide, abiraterone, apalutamide, and docetaxel, HR (95% C) 0.39 (0.30-0.51), 0.45 (0.40-0.51), 0.48 (0.39-0.59), and 0.67 (0.61-0.74). Allowing time-varying HR produced similar treatment rankings. Serious adverse events (SAE) were substantially increased for docetaxel and slightly increased for abiraterone, odds ratio (95%CI) 104.17 (24.85-1012.32) and 1.42(1.11-1.83).

CONCLUSIONS: Abiraterone provided the largest OS benefit with slightly increased risk of SAE. Apalutamide offered comparable OS benefit with abiraterone without increasing SAE risk. Enzalutamide, although delayed rPFS to the greatest extent, did not show OS benefit based on the available evidence.