Prediction of focal extracapsular extension at radical prostatectomy: Relative merit of transrectal ultrasound, endorectal magnetic resonance imaging, prostate specific antigen, prostate specific antigen density, and systematic biopsy.

We conducted a study to compare the relative merits of prostate specific antigen (PSA), PSA density (PSAD), transrectal ultrasound (TRUS), endorectal magnetic resonance imaging (MRI), and systematic biopsy in the prediction of focal extracapsular extension (ECE) at radical prostatectomy. A retrospective review of patients who underwent TRUS, endorectal MRI, and radical prostatectomy at our institution was performed. Patients with a diagnosis of prostate cancer who were thought to be surgical candidates by digital rectal examination and TRUS underwent endorectal MRI prior to radical prostatectomy. Imaging, PSA, PSAD, and systematic biopsy results (tumor grade and fraction of positive systematic biopsies) were correlated with step-sectioned, radical prostatectomy pathologic data. Data was analyzed for the entire prostate and on each individual side. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and likelihood ratios were calculated for each modality, and receiver operating characteristic (ROC) curves were generated. Stepwise logistic regression analysis was used to weigh the relative contributions of preoperative parameters in predicting ECE. Data was collected from 54 patients who had sextant systematic biopsy, imaging, and radical prostatectomy. A total of 24 sides demonstrated ECE (19 patients, 5 with bilateral ECE). When assessed for the dominant prostate side and on a side-for-side basis, MRI had the highest sensitivity and NPV for detecting focal ECE. MRI also had the highest PPV, and TRUS had the highest specificity for side-for-side analysis. For the dominant prostate side, PSA had the highest specificity and PPV for detecting focal ECE. Of note, significant overlap was demonstrated in the 95% confidence intervals of all modalities with each other for all analyses. ROC analyses found MRI and Gleason sum to be superior for the dominant prostate side assessment and MRI and the fraction of positive systematic biopsies to be superior for a side-for-side analysis. Optimal likelihood ratios for positive test results were seen for PSA (dominant prostate side) and MRI (side-for-side), and for negative test results for MRI. Logistic regression demonstrated MRI and Gleason sum to be powerful predictors of ECE. Thus, we would conclude that endorectal MRI and tumor grade provide unique information in the prediction of focal ECE in select patients.