Comparison of the Prostate Imaging Reporting and Data System (PI-RADS) Version 1 and 2 in a Cohort of 245 Patients with Histopathological Reference and Long-Term Follow-Up

OBJECTIVE: To compare the performance of PI-RADSv2 with PI-RADSv1 in patients with elevated PSA before biopsy. METHODS: 245 patients with elevated PSA underwent mpMRI before biopsy between May 2011 and December 2014 at 3.0 Tesla without endorectal coil. Patients underwent transrectal ultrasound-guided systematic 12-core biopsy followed by radical prostatectomy (N = 68), radiation therapy (N = 91) or clinical follow-up for at least two years (N = 86). All exams were scored on a per-patient basis according to PI-RADSv1 and PI-RADSv2. ClinsigPC was defined as Gleason score ≥7 (including 3+4 with prominent but not predominant Gleason 4 component), and/or tumour volume of ≥0.5cc, and/or tumour stage ≥T3a. RESULTS: In 144 patients (58.8%) a ClinsigPC was found within two years after mpMRI. The PI-RADSv1 and PI-RADSv2 overall assessment scores were significantly higher (P < 0.001) in patients with ClinsigPC as compared to patients without ClinsigPC. ROC analysis showed an area under the curve of 0.82 (CI 0.76–0.87) for PI-RADSv1 and 0.79 (CI 0.73–0.85) for PI-RADSv2 (P: NS). A threshold score of 3 exhibited sensitivities of 88.2% and 79.2% (P = 0.001) and specificities of 64.4% and 67.3% (P: NS) with PI-RADSv1 and PI-RADSv2, respectively. CONCLUSIONS: The mpMRI scoring systems PI-RADSv1 and PI-RADSv2 yield similar accuracy to detect ClinsigPC in patients with elevated PSA, although clinicians should be aware that when an overall assessment score of 3 is used as a threshold for a positive mpMRI, PI-RADSv2 has lower sensitivity than PI-RADSv1. Nevertheless, PI-RADSv2 is preferable over PI-RADSv1 because it has the advantage of providing well-defined instructions on how to determine the overall assessment category.