Multiparametric MRI Improves Accuracy of Clinical Nomograms for Predicting Extracapsular Extension of Prostate Cancer

Abstract

To compare the accuracy of multiparametric magnetic resonance imaging (MP-MRI) with the Partin tables and Memorial Sloan-Kettering (MSK) nomogram for predicting extracapsular extension (ECE) in prostate cancer and to create a tool for clinicians to estimate pathologic ECErisk. A retrospective review of 112 patients who underwent 3T MP-MRI of the prostate and radical prostatectomy was performed. Regression analyses were carried out to identify predictors of ECE. Predictive accuracy of models based on nomogram and MP-MRI were compared. A total of 33 of patients (29%) had ECE on MP-MRI whereas 26 patients (23%) had ECE on final pathology. Mean age was 62.8years and mean prostate-specific antigen was 8.2ng/dL. MRI was a significant predictor of ECE that was independent of age, prostate-specific antigen, Gleason score, clinical stage, and percent positive cores on biopsy. Sensitivity, specificity, positive predictive value, and negative predictive value of MP-MRI for ECE were 84.6%, 87.2%, 66.7%, and 94.9%, respectively. Areas under the curve for Partin and MSK nomograms for predicting ECE were 0.85 and 0.86, respectively. Area under the curve increased to 0.92 and 0.94, respectively, when MP-MRI was added to each nomogram. We provide an online tool that integrates Partin or MSK nomogram results with ECE status determined from MRI to predict pathologic ECE. Within the typical range of risks for ECE provided by the clinical nomograms (ie, 15%-40%), MRI was useful for predicting pathologic ECE. MP-MRI may be a useful adjunct for clinically staging prostate cancer. MP-MRI improved accuracy of existing clinical nomograms for prediction of pathologic ECE.