308 Background: Both enzalutamide and abiraterone are approved for the treatment of mCRPC, each demonstrating improved overall survival (OS) versus placebo. However, it is unclear what the optimal sequencing of these two therapies should be to maximize clinical outcomes and survival. Here, we compare clinical outcomes among patients with mCRPC who received enzalutamide first followed immediately by abiraterone (enza-to-abi) or the opposite sequence (abi-to-enza). Methods: A retrospective review of consecutive mCRPC patients treated at Johns Hopkins with enza-to-abi or abi-to-enza was conducted. The combined PFS (PFS = PFS1 + PFS2) was the primary endpoint, measured from the start of the first therapy (i.e. enza or abi) until disease progression on the subsequent therapy (i.e.abi or enza). The OS from the start of the first therapy until death was the secondary endpoint. Cox proportional hazards multivariable regression analysis was performed to determine whether one sequence was better than the other after adjusting for baseline characteristics. Results: 71 patients were identified: 58 received abi-to-enza and 13 received enza-to-abi. Comparisons of baseline characteristics between groups identified differences in PSA levels (P = 0.007), hemoglobin (P < 0.001), and presence of visceral disease (P = 0.035). The abi-to-enza group had a longer combined PFS than the enza-to-abi group: median 16.3 vs 12.5 mo (HR 0.53, P = 0.04). There was also a numeric improvement in OS in the abi-to-enza group compared to the enza-to-abi group: median 29.0 vs 21.0 mo (HR 0.51, P < 0.10). In multivariable analyses incorporating PSA level, hemoglobin, visceral disease and prior docetaxel use, both combined PFS (HR 2.59, P = 0.03) and OS (HR 4.59, P < 0.01) demonstrated improved outcomes with the abi-to-enza sequence. Conclusions: This hypothesis-generating study potentially suggests superior PFS and OS in men with mCRPC receiving abiraterone then enzalutamide (compared to enzalutamide then abiraterone), although this could be due to baseline imbalances or the small sample size of this study. Prospective validation of this concept is ongoing (NCT02125357).
Comparison of performance of exponential, Cox proportional hazards, weibull and frailty survival models for analysis of small sample size data
