Early prostate-specific antigen relapse after radical retropubic prostatectomy: prediction on the basis of preoperative and postoperative tumor characteristics.

OBJECTIVES: This study was undertaken to distinguish between patients who will and will not benefit from a retropubic radical prostatectomy (RRP) for clinically localized prostatic carcinoma (PCa) on the basis of preoperative and postoperative tumor characteristics.
METHODS: Data of 318 consecutive patients who underwent RRP for clinically localized PCa were reviewed. Preoperative characteristics used included clinical stage, findings on transrectal ultrasonography, prostate-specific antigen (PSA) values, Gleason grade, number of positive biopsies, number of biopsies containing any Gleason grade 4 and/or 5 cancer, and number of biopsies with predominant (>50% of cancerous tissue) Gleason grade 4 and/or 5 cancer. Postoperative characteristics included pathologic stage, Gleason grade, margin status, cancer volume, and volume of Gleason grade 4 and/or 5 cancer. The impact on biochemical relapse after RRP were calculated by Cox regression and CART (classification and regression tree) analysis to establish low, intermediate, and high risk of recurrence.
RESULTS: Of patients who underwent RRP, 66% showed no evidence of relapse after a follow-up of 42 months. All preoperative and postoperative characteristics showed a significant association with biochemical relapse. Cox regression of preoperative characteristics showed the number of positive biopsies with predominant Gleason grade 4 and/or 5 cancer to be the most accurate predictor of failure (p < 0.0001), followed by the number of positive biopsies and PSA. CART analysis distinguished between four risk groups on the basis of the same characteristics as in the Cox regression. The low-risk group consisted of 232 patients (75.1%) and the high-risk group of 17 patients (5.5%); corresponding Kaplan-Meier curves showed a 2-year PSA-free survival rate of 97% for the low-risk group and 20% for the high-risk group. Cox regression of postoperative characteristics recognized the volume of Gleason grade 4 and/or 5 as the characteristic with the strongest association with biochemical failure. CART analysis distinguished between four risk groups, using the volume of high-grade cancer as the most influential characteristic. The corresponding Kaplan-Meier curves showed for the low-risk group (n = 79; 29.6%) a PSA-free survival rate of 96% after 42 months and for the high-risk group (n = 47; 17.6%) a 21% PSA-free survival rate after 42 months.
CONCLUSION: For preoperative and postoperative estimation of biochemical recurrence after RRP, a quantitative analysis of high-grade cancer, expressed by the number of preoperative biopsy cores containing high-grade cancer and the volume of cancer, proved to be the best predictor of relapse. CART analysis might be useful in advising patients for their best therapy options. However, defined characteristics of risk groups should be evaluated with new prospective data before they are used routinely.