The Accuracy of Prostate Magnetic Resonance Imaging Interpretation: Impact of the Individual Radiologist and Clinical Factors

Abstract

ObjectiveTo compare test performance of multiparametric magnetic resonance imaging (mpMRI) for detection of prostate cancer between individual radiologists using the Prostate Imaging Reporting and Data System (PI-RADS) and to identify clinical factors that may predict test performance. Materials and MethodsWe examined our database of consecutive men who received prostate mpMRI prior to biopsy between September 2014 and December 2016 (n = 459). Test performance (eg, sensitivity, specificity, positive predictive value [PPV] and negative predictive value) were defined with PI-RADS classification 4 or 5 considered test positive and Gleason score ≥7 on biopsy from any targeted core considered outcome positive. Multivariate logistic regression was performed to identify clinical variables that affect test performance. ResultsNo significant differences in test performance were found among individual radiologists. Prior biopsy (odds ratio [OR] 0.10, P = .01), radiologist experience >500 prostate mpMRI (OR 0.18, P = .04), transition zone location (OR 0.10, P = .04), and posterior location (OR 0.04, P = .03) were predictors of diminished sensitivity. Location of the mpMRI lesion in the TZ was a predictor of improved specificity (OR 2.53, P = .04). Increasing age (OR 1.07, P <.01) and prostate-specific antigen (OR 1.10, P <.01) predicted increased PPV, while prior biopsy predicted decreased PPV (OR 0.50, P <.01). ConclusionAlthough variation exists in test performance among individual radiologists using PI-RADS, significant differences were not observed. Additional prostate mpMRI experience was not beneficial in improving accuracy of interpretation. Nonmodifiable patient variables—including prostate lesion location, prior biopsy history, prostate-specific antigen, and age—are predictive of prostate mpMRI test performance.