Validity of the CAPRA Score to Predict Biochemical Recurrence-Free Survival After Radical Prostatectomy. Results From a European Multicenter Survey of 1,296 Patients

Abstract

The CAPRA (Cancer of the Prostate Risk Assessment) score from the University of California, San Francisco provides a new statistical model to predict recurrence-free survival and pathological tumor stage after radical prostatectomy. It was originally developed using data from the CaPSURE (Cancer of the Prostate Strategic Urologic Research Endeavor) registry. To calculate the score, which ranges from 0 to 10, 5 clinical variables are needed, ie prostate specific antigen, Gleason sum, clinical tumor grade, percentage of positive biopsies and patient age. To date, the only external validation of the CAPRA score has been conducted using the SEARCH (Shared Equal Access Regional Cancer Hospital) database. The present study uses a German database to contribute to existing validation work and to test transferability of the CAPRA score to a sample that differs fundamentally from the SEARCH sample in terms of clinical features. Data from 1,296 German patients after radical prostatectomy were used for validation. Mean followup was 56.5 (+/-35.4) months. Accuracy of prediction of recurrence-free survival and pathological tumor stage with the CAPRA score was analyzed using Kaplan-Meier analysis, proportional hazards regression, logistic regression and graphic representation. For the external validation of the CAPRA score, the underlying clinical variables of our study group were unfavorable compared to the original cohort from the CaPSURE data set. The recurrence-free survival rate decreased after 3 and 5 years from 100% to 97%, respectively, in the CAPRA score 0 to 1 group, and from 44% to 31%, respectively, in the CAPRA score of 7 or higher group. The hazard ratios of a biochemical recurrence per 1-group increase were 1.50 (95% CI 1.43-1.56) for the CAPRA sum score, 1.62 (95% CI 1.53-1.71) for the 7-group CAPRA score and 3.52 (95% CI 3.00-4.12) for the 3-group CAPRA score. Concordance indices between 0.78 and 0.81 suggested good predictive accuracy. Of the 5 CAPRA constituents 4 independently predicted recurrence-free survival, ie prostate specific antigen, Gleason sum, cT stage and percent of positive biopsies. Positive margins occurred in 13.1% of patients with a CAPRA score of 0 to 1 vs 62% of patients with a score of 7 to 10 (p <0.001). Organ confined tumors were present in 97.7% of patients with a CAPRA score of 0 to 1 vs 19.3% of those with a score of 7 to 10 (p <0.001). Despite different clinical features in the present patient cohort and the CaPSURE data set, the accuracy of the CAPRA nomogram in predicting recurrence-free survival was high. These results underscore the effectiveness and the clinical applicability of the CAPRA score which, in addition to patient counseling, may also be used for risk stratification in clinical studies.