Purpose: We constructed a biomarker-survival modeling framework to explore the relationship between prostate-specific antigen (PSA) kinetics and overall survival (OS) in metastatic castration-resistant prostate cancer (mCRPC) patients following oral administration of 1,000 mg/day of abiraterone acetate (AA).Experimental Design: The PSA-survival modeling framework was based on data from two phase III studies, COU-AA-301 (chemotherapypretreated,n¼ 1,184)andCOU-AA-302(chemotherapy na€ ve, n ¼ 1,081), and included a mixed-effects tumor growth inhibition model and a Cox proportional hazards survival model.Results: The effect of AA on PSA kinetics was significant (P < 0.0001) and comparable between the chemotherapy-na€ ve and -pretreated patients. PSA kinetics [e.g., PSA nadir, PSA response rate (!30%, 50%, and 90%), time to PSA progression, PSA doubling time (PSADT)] were highly associated with OS in both populations. The model-based posttreatment PSADT had the strongest association with OS (HR $0.9 in both populations). The models could accurately predict survival outcomes. After adjusting for PSA kinetic endpoints, the treatment effect of AA on survival was no longer statistically significant in both studies, and the Prentice criteria of surrogacy were met for the PSA kinetic endpoints. A strong correlation was also observed between PSA and radiographic progression-free survival.Conclusions: The analysis revealed a consistent treatment effect of AA on PSA kinetics and strong associations between PSA kinetics and OS in chemotherapy-pretreated and -na€ ve patients, thereby providing a rationale to consider PSA kinetics as surrogacy endpoints to indicate clinical benefit in AA-treated patients with mCRPC regardless of chemotherapy treatment.

<p>Supplemental Methods, Table 1 and Figures 1-2. Appendix I: TGI PSA kinetic model, Appendix II: Survival analysis. Supplementary Table 1. Parameter estimates of the final PSA kinetic model. Supplementary Figure 1. Diagnostic plot of the observed and predicted PSA concentrations for the PSA kinetic models. Supplementary Figure 2. Comparison of predicted survival probability based on the final multivariate PSA-survival model to the observed survival probability over time.</p>

<div>Abstract<p><b>Purpose:</b> We constructed a biomarker-survival modeling framework to explore the relationship between prostate-specific antigen (PSA) kinetics and overall survival (OS) in metastatic castration-resistant prostate cancer (mCRPC) patients following oral administration of 1,000 mg/day of abiraterone acetate (AA).</p><p><b>Experimental Design:</b> The PSA-survival modeling framework was based on data from two phase III studies, COU-AA-301 (chemotherapy pretreated, <i>n</i> = 1,184) and COU-AA-302 (chemotherapy naïve, <i>n</i> = 1,081), and included a mixed-effects tumor growth inhibition model and a Cox proportional hazards survival model.</p><p><b>Results:</b> The effect of AA on PSA kinetics was significant (<i>P</i> < 0.0001) and comparable between the chemotherapy-naïve and -pretreated patients. PSA kinetics [e.g., PSA nadir, PSA response rate (≥30%, 50%, and 90%), time to PSA progression, PSA doubling time (PSADT)] were highly associated with OS in both populations. The model-based posttreatment PSADT had the strongest association with OS (HR ∼0.9 in both populations). The models could accurately predict survival outcomes. After adjusting for PSA kinetic endpoints, the treatment effect of AA on survival was no longer statistically significant in both studies, and the Prentice criteria of surrogacy were met for the PSA kinetic endpoints. A strong correlation was also observed between PSA and radiographic progression-free survival.</p><p><b>Conclusions:</b> The analysis revealed a consistent treatment effect of AA on PSA kinetics and strong associations between PSA kinetics and OS in chemotherapy-pretreated and -naïve patients, thereby providing a rationale to consider PSA kinetics as surrogacy endpoints to indicate clinical benefit in AA-treated patients with mCRPC regardless of chemotherapy treatment. <i>Clin Cancer Res; 21(14); 3170–7. ©2015 AACR</i>.</p></div>

<div>Abstract<p><b>Purpose:</b> We constructed a biomarker-survival modeling framework to explore the relationship between prostate-specific antigen (PSA) kinetics and overall survival (OS) in metastatic castration-resistant prostate cancer (mCRPC) patients following oral administration of 1,000 mg/day of abiraterone acetate (AA).</p><p><b>Experimental Design:</b> The PSA-survival modeling framework was based on data from two phase III studies, COU-AA-301 (chemotherapy pretreated, <i>n</i> = 1,184) and COU-AA-302 (chemotherapy naïve, <i>n</i> = 1,081), and included a mixed-effects tumor growth inhibition model and a Cox proportional hazards survival model.</p><p><b>Results:</b> The effect of AA on PSA kinetics was significant (<i>P</i> < 0.0001) and comparable between the chemotherapy-naïve and -pretreated patients. PSA kinetics [e.g., PSA nadir, PSA response rate (≥30%, 50%, and 90%), time to PSA progression, PSA doubling time (PSADT)] were highly associated with OS in both populations. The model-based posttreatment PSADT had the strongest association with OS (HR ∼0.9 in both populations). The models could accurately predict survival outcomes. After adjusting for PSA kinetic endpoints, the treatment effect of AA on survival was no longer statistically significant in both studies, and the Prentice criteria of surrogacy were met for the PSA kinetic endpoints. A strong correlation was also observed between PSA and radiographic progression-free survival.</p><p><b>Conclusions:</b> The analysis revealed a consistent treatment effect of AA on PSA kinetics and strong associations between PSA kinetics and OS in chemotherapy-pretreated and -naïve patients, thereby providing a rationale to consider PSA kinetics as surrogacy endpoints to indicate clinical benefit in AA-treated patients with mCRPC regardless of chemotherapy treatment. <i>Clin Cancer Res; 21(14); 3170–7. ©2015 AACR</i>.</p></div>

<p>Supplemental Methods, Table 1 and Figures 1-2. Appendix I: TGI PSA kinetic model, Appendix II: Survival analysis. Supplementary Table 1. Parameter estimates of the final PSA kinetic model. Supplementary Figure 1. Diagnostic plot of the observed and predicted PSA concentrations for the PSA kinetic models. Supplementary Figure 2. Comparison of predicted survival probability based on the final multivariate PSA-survival model to the observed survival probability over time.</p>