Radical prostatectomy and positive surgical margins: relationship with prostate cancer outcome

ABSTRACT

Introduction Positive surgical margins (PSMs) are an adverse factor that may predict a worse outcome in patients submitted to radical prostatectomy (RP). However, not all of these cases will evolve to biochemical (BCR) or clinical (CR) recurrence, therefore relationship between PSMs and these recurrent events has to be correlated with other clinical and pathologic findings to indicate complementary treatment for selected patients. Materials and Methods Of 1250 patients submitted to open retropubic radical prostatectomy (RRP), between March 1991 and June 2008, the outcome of 161 patients with PSMs and of 67 without PSMs as a control group, comprising a total of 228 cases were retrospectively reviewed. A minimum follow-up time of 2 years after surgery was considered. BCR was determined when PSA ≥ 0.2ng/mL. CR was determined whenever there was clinical evidence of tumor. Chi-square test was used to correlate clinical and pathologic variables with PSMs. Time interval to biochemical recurrence was analyzed by the Kaplan-Meier product limit analysis using the log-rank test for comparison between groups. Univariate and multivariate Cox stepwise logistic regression models were used to identify significant predictors of risk of shorter intervals to BCR. Results Prostate circumference margin was the most common site with 78 cases (48.44%). Regarding the outcome of 228 cases from both groups, BCR occurred in 68 patients (29.82%), and CR in 10 (4.38%). Univariate analysis showed statistically significant associations (p < 0.001) between presence of PSMs with BCR, but not with CR (p = 0.05). At follow-up of the 161 patients with PSMs, only 61(37.8%) presented BCR, while 100 (62.8%) did not. BCR correlated with pathologic stage; Gleason score; preoperative PSA; tumor volume in the specimen; capsular and perineural invasion; presence and number of PSMs. CR correlated only with angiolymphatic invasion and Gleason score. Considering univariate ...