Evaluation of four pre-operative models for prediction of biochemical recurrence after radical prostatectomy in localised prostate cancer.

Background: Biochemical recurrence (BCR) can be seen in the early or late period after radical prostatectomy (RP). Various models have been developed to predict BCR.

Objective: In our study, we evaluated the accuracy of four pre-operative models (GP score, PRIX, D'Amico risk classification, CAPRA) in predicting BCR after RP in Turkish patients.

Methods: Age, preoperative total prostate-specific antigen (PSA) values, clinical stages, total number of cores taken in biopsy, number of positive cores, preoperative biopsy Gleason score (GS), follow-up time and presence of BCR after RP were recorded. BCR was defined as a total PSA value >0.2 ng/dL twice consecutively after RP. Classifications or scoring was performed according to pre-operative models. The 1-, 3- and 5-year BCR-free rates of the patients were determined for each model. Also, the accuracy of four predictive models for predicting 1-, 3- and 5-year BCR was evaluated.

Results: For all pre-operative models there was a statistically significant difference between risk groups in BCR-free rates at 1-, 3- and 5-year after RP (P < .001). The Harrell's concordance index for 1-year BCR predictions was 0.802, 0.831, 0.773 and 0.745 for the GP score, PRIX, CAPRA and D'Amico, respectively. For 3-year BCR predictions, it was 0.798, 0.791, 0.723 and 0.714 for the GP score, PRIX, CAPRA and D'Amico and respectively. Finally, The Harrell's concordance index for 5-year BCR predictions was 0.778, 0.771, 0.702 and 0.693 for the GP score, PRIX, CAPRA and D'Amico, respectively.

Conclusion: In the prediction of BCR, the accuracy of GP scoring and PRIX seems slightly higher than CAPRA and D'Amico risk classification. Surely our results should be supported by head to head comparisons within other larger cohorts.