Performing and optimizing individualized actuarial prediction of post-prostatectomy PSA control with a Clinical Outcome Prediction Expert (COPE)

Abstract

Rational treatment decision requires accurate projection of the clinical course of a patient. Current methods in clinical outcome analysis mostly focus on population data. We investigated the applicability and optimization of the widely used actuarial method to project an individual's clinical outcome. We designed and implemented COPE (Clinical Outcome Prediction Expert), which performs, assesses and optimizes individualized actuarial prediction. The program was applied to analyze a post-prostatectomy database. Stratified actuarial curves are used to project individual outcomes. The area under the receiver operator characteristic (ROC) curve was used to measure predictive performance. We applied COPE to search for the optimal cut-off time and cut-off probability for prediction. The pre-treatment PSA (prostate-specific antigen), the Gleason score and the AJCC (American Joint Commission on Cancer) clinical T-stage were used as predictors. We found that the optimal range of the cut-off probability was 65% to 75% and the cut-off time was 44 to 52 months for all predictors. Optimization greatly simplifies the risk stratification and improves the predictive power of the Gleason score. The optimized multivariate risk score has the highest ROC area of 0.77 among all predictors. This study shows that ROC-optimized risk stratification improves the accuracy of clinical outcome prediction.