Early assessment of PSA response in CRPC patients treated with enzalutamide (Enza) or abiraterone (Abi).

Abstract

249 Background: Abi and Enza are used in treating metastatic CRPC. Primary resistance is well described but little data exists for early treatment responders. Further characterization of patients with early PSA declines was the goal of this study. Methods: Single institution, retrospective reviews were performed on 84 CRPC patients (pts) treated with Abi or Enza. PSAs were recorded for the duration of treatment. The primary end point was to describe receiver operating characteristics (ROC) of early PSA changes, including sensitivities and specificities, as a predictor of later treatment response (defined as ≥ 50% decrease in PSA from baseline). 12 additional clinical covariates were also evaluated as treatment response predictors. Results: 30/63 pts in the Abi treatment group and 12/44 pts in the Enza treated group had a PSA treatment response at some point. Among the “eventual” Abi responders, some achieved a response early; 16/30 at 4 weeks and 22/30 at 8 weeks. Of Enza “eventual” responders, 7/12 achieved response at 4 weeks. For early responders (either 4 or 8 weeks), Abi pts median treatment time was 336 days. Enza early responders median treatment was time 303 days. For ROC analysis of PSA changes at 4 & 8 weeks, in terms of predicting ‘eventual’ responders, the Abi group had an area under the curve (AUC) of 0.86 ± 0.07 and 0.93 ± 0.08, respectively. Enza ROC analysis of PSA change at 4 weeks had an AUC of 0.97 ± 0.03. Sensitivities and specificities were also described for various PSA change thresholds in each group. Ideal cut-off points for Abi at 4 weeks was a 37% decrease in PSA (sensitivity: 0.74, specificity: 0.97) and at 8 weeks was a 33% decrease (0.82, 0.97). For Enza ideal cut-off at 4 weeks was a 31% PSA decrease (0.82, 0.93). Covariate analysis of the Abi pts indicated that metastasis at diagnosis (P = 0.048) and prior taxanes predicted resistance to treatment (P = 0.019), as measured by < 50% decline in PSA. All covariates in the Enza pts failed to reach significance. Conclusions: Results suggest that early PSA changes for both Enza and Abi may be a reliable way for clinicians to predict long-term PSA response. This may allow for earlier modifications to be made in patient management for those not achieving an early PSA response.