Development and Validation of a Novel Prognostic Model for Predicting Overall Survival in Treatment-naïve Castration-sensitive Metastatic Prostate Cancer

Abstract

BackgroundThere has been growth in the treatment options for castration-sensitive metastatic prostate cancer (mPCa), but without clear guidance for risk stratification. ObjectiveTo identify clinical parameters associated with overall survival (OS) and establish a prognostic model for use with treatment-naïve castration-sensitive mPCa. Design, setting, and participantsA retrospective review of 304 patients treated at Kyoto University Hospital was performed. A prognostic model was created using clinical parameters associated with OS. The model was externally validated in an independent cohort of 520 patients. Outcome measurements and statistical analysisMultivariable analysis was performed to identify the clinical parameters associated with OS. Risk scores were calculated using Cox proportional hazards analysis for each combination of risk factors, and patients were grouped into categories based on those scores. Results and limitationsOver 80% of the cohort had a Gleason sum score ≥8. The median OS was 53mo among patients with CHAARTED high-volume PCa (n=172) and 131mo among those with low-volume PCa (n=100). Independent factors associated with OS were extent of disease score ≥2 or the presence of liver metastasis; lactate dehydrogenase >250U/L; and a primary Gleason score of 5. The median OS for the high-, intermediate-, and low-risk groups according to the new model were 28mo, 59mo, and not reached, respectively; the corresponding values in the validation cohort were 41mo, 63mo, and not reached. Harrell's C-index was 0.649. ConclusionsOur simple and reproducible prognostic model for treatment-naïve castration-sensitive mPCa could aid in risk stratification and treatment selection. Patient summaryWe identified clinical parameters associated with prognosis in castration-sensitive metastatic prostate cancer and established a reproducible prognostic model that could be used to guide treatment decisions.