Evaluation of the PI-RADS Scoring System for Classifying mpMRI Findings in Men with Suspicion of Prostate Cancer

PURPOSE To evaluate the ESUR scoring system (PI-RADS) for multiparametric MRI of the prostate in clinical routine and to define a reliable way to generate an overall PI-RADS score. METHODS Retrospective analysis of all patients with a history of negative prebiopsies, who underwent 3 Tesla multiparametric MRI from October 2011 to April 2013 (n = 143): PI-RADS scores for each single modality were defined. To generate the overall PI-RADS score, an algorithm based approach summing up each single-modality score to a sum-score was compared to a more subjective approach, weighting the single modalities dependent on the radiologist's impression. Because of ongoing cancer suspicion 73 patients underwent targeted mpMRI-ultrasound image fusion rebiopsy. For this group thresholds for tumor incidences and malignancy were calculated. RESULTS 39 (53%) out of 73 targeted rebiopsies were cancer positive. The PI-RADS score correlated well with tumor incidence (AUC of 0.86, 95% CI 0.78 to 0.94) and malignancy (AUC 0.84, 95% CI 0.68 to 0.99). Regarding the sum-score a threshold of ≥ 10 turned out to be reliable for cancer detection (sensitivity 90%, specificity 62%) and for ≥ 13 for indicating higher malignancy (Gleason ≥ 4 + 3) (sensitivity 80%, specificity 86%). To generate the overall PI-RADS score, the use of an algorithm based approach was more reliable than that of the approach based on the radiologist's impression. CONCLUSION The presented scoring system correlates well with tumor incidence and malignancy. To generate the overall PI-RADS score, it seems to be advisable to use an algorithm based instead of a subjective approach.