Heterogeneity in Treatment Effects of Reduced Versus Standard Dose of Cabazitaxel in Metastatic Castration‐Resistant Prostate Cancer

BACKGROUND: (C25) in metastatic castration-resistant prostate cancer (mCRPC), one-variable-at-a-time subgroup analysis suggested possible heterogeneity in treatment effect (HTE) of C25 versus C20 among study participants. Novel predictive HTE analysis approaches may provide an in-depth understanding of such results. METHODS: We analyzed patient-level data from 1200 patients with mCRPC who were randomized in the PROSELICA trial. Outcomes included overall survival (OS) and progression-free survival (PFS). Using baseline characteristics, patients were stratified into quartiles based on either quantitative baseline risk of poor outcome (risk modeling) or predicted individualized treatment effect (ITE) using a causal survival forest algorithm (effect modeling). Treatment effects were measured as differences in restricted mean survival time (RMST). RESULTS: For risk modeling, the OS effect of C25 increased with risk quartiles: -0.07 months (95% CI, -1.60 to 1.46) in the lowest risk quartile and 1.67 months (95% CI, 0.25 to 3.10) in the highest risk quartile. For effect modeling, the OS effect ranged from -0.17 months (95% CI, -3.01 to 2.68) in the lowest ITE quartile to 0.57 months (95% CI, -2.27 to 3.41) in the highest ITE quartile. Both approaches demonstrated greater C25 benefit in patients with extensive previous treatment and baseline disease burden. PFS effects remained consistent across all quartiles. CONCLUSIONS: The OS effect of C25 versus C20 may vary based on baseline characteristics in post-docetaxel mCRPC. Patients with extensive treatment history and disease burden may benefit more from C25.