Accuracy of preoperative endorectal MRI in predicting extracapsular extension and influence on neurovascular bundle sparing in radical prostatectomy

To evaluate the accuracy of presurgical endorectal MRI (eMRI) for local staging before radical prostatectomy (RP) and its influence on neurovascular bundle (NVB) resection during radical prostatectomy. A total of 385 patients with histologically proven prostate cancer (PCa) have been included in this retrospective study between 2004 and 2008. All patients underwent preoperative eMRI at 1.5 T before open RP. Staging results by eMRI were compared with the histopathological findings. The presence of positive surgical margins and extent of nerve-sparing procedure were evaluated. Subgroup analysis of low-risk group and intermediate to high-risk group based on D'Amico criteria was conducted. In 294 (76.4%) patients, pathological stage was correctly predicted, 69 patients (17.9%) were understaged and 22 (5.7%) overstaged. Overall sensitivity, specificity, negative and positive predictive value for predicting extracapsular extension (ECE) were 41.5, 91.8, 78.0 and 69.0%, respectively. One hundred and fifty-two (48.4%) of the patients classified as stage cT2 by eMRI underwent bilateral NVB sparing, whereas 14 (19.7%) patients with reported ECE underwent bilateral NVB sparing (P < 0.01). Overall positive surgical margin rate was 14.8%. Sensitivity of predicting ECE and positive predictive value were lower in the low-risk group than in the intermediate and high-risk group. eMRI is effective in predicting extracapsular extension in an intermediate to high-risk group. Preoperative eMRI in patients with low-risk criteria is not recommended as a routine assessment modality. eMRI findings did appear to influence surgical strategy as patients with imaging findings suggesting > cT2 disease were less likely to undergo NVB sparing.