Cabazitaxel Versus Abiraterone or Enzalutamide for Poor-prognosis Metastatic Castration-resistant Prostate Cancer After Docetaxel: A Phase 2 Trial with a Circulating Tumor DNA Analysis.

[BACKGROUND AND OBJECTIVE] Whether cabazitaxel or an androgen receptor pathway inhibitor (ARPI) is the optimal treatment option for poor-prognosis metastatic castration-resistant prostate cancer (mCRPC), progressing on docetaxel, remains unclear. There are limited prospective data supporting a preference for one of these treatments and few candidate biomarkers to inform individual patient management. This study aims to compare the clinical efficacy of cabazitaxel versus ARPIs in patients with poor-prognosis mCRPC who have progressed on docetaxel, and to evaluate the prognostic and predictive utility of circulating tumor DNA (ctDNA) in this treatment-refractory population.

[METHODS] A multicenter, open-label, phase 2b trial randomized poor-prognosis mCRPC patients to an ARPI (1000 mg abiraterone plus prednisone or 160 mg enzalutamide daily) or cabazitaxel (25 mg/m every 3 wk plus prednisone daily). The primary endpoint was the clinical benefit rate (CBR) at 12 wk. The secondary endpoints included radiographic progression-free survival (rPFS), overall survival (OS), and PSA50 response. Genomic analyses on plasma utilized targeted cell-free DNA sequencing at baseline, 12 wk, and progression.

[KEY FINDINGS AND LIMITATIONS] In total, 106 patients were randomized. The CBR at 12 wk was 62.3% (66/106), with no difference between treatments (p = 0.54). Between groups, rPFS and OS (median follow-up of 30.9 mo) were not different. PSA50 was higher in the ARPI arm (47.2%) than in the cabazitaxel arm (26.9%; p = 0.04). Prior ARPI exposure (in 37.7%) predicted inferior outcomes on ARPIs but not on cabazitaxel. Adverse events of grade ≥3 were more frequent with cabazitaxel (65.4% vs 30.2%). A high baseline ctDNA fraction correlated with reduced rPFS and OS; plasma AR copy number status was not associated with outcomes, but PTEN alterations were linked with shorter OS (hazard ratio: 1.9, multivariable p = 0.02).

[CONCLUSIONS AND CLINICAL IMPLICATIONS] No significant differences in CBR or time-to-event endpoints were observed between cabazitaxel and ARPIs. However, prior ARPI exposure, a higher baseline ctDNA fraction, and PTEN alterations were strongly prognostic.