Assessment of PI-RADS v2 Categories ≥ 3 for Diagnosis of Clinically Significant Prostate Cancer - Beyond the Abstract

mpMRI and its reporting by PI-RADs v2 scoring system has become standard clinical practice. This IRB approved retrospective cohort study assessed the diagnostic accuracy of PI-RADS v2 categories ≥ 3 to detect clinically significant prostate cancer (csPCa).  In this study, 3T (Tesla) mpMRI lesions reported as PI-RADS v2 categories ≥ 3 were correlated to the 3D model of the prostate showing cancer foci from Transperineal Mapping Biopsy (TPMB).

Clinically significant prostate cancer was defined as GS (Gleason score) of ≥ 7 and/or tumor volume ≥ 0.5 cc per PI-RADs v2 document guidelines.  We evaluated two outcomes for csPCa:  group 1 for PCa with GS ≥7 and group 2 for PCa with tumor volume ≥ 0.5 cc. Any sector in the prostate that was not observed on MRI and did not have a positive biopsy result on 3DTPMB was recorded as negative MRI and having a cancer negative result. We evaluated sensitivity, specificity, PPV and NPV at the lesion and person level for PI-RADS v2 categories ≥ 3 to diagnose csPCa. PI-RADS v2 score of 5 had high PPV and specificity (100%) at the lesion level and person level in detection of csPCa, however, PS (PI-RADS score) of 3 and 4 had high false positives in our study.

56 clinically significant cancer lesions were found by TPMB; 10 cancer lesions of GS ≥ 7 and 19 cancer lesions of GS 6 with tumor volume > 0.5 cc were missed by mpMRI. Radiologist agreement on the presence or absence of certain MR features was varied. Health care providers should carefully consider these limitations of PI-RADS v2 categories as the approaches to the prostate cancer screening and diagnosis continue to evolve. Future revisions to PI-RADS should take into account the limitations in terms of accuracy and variable inter-rater reliability identified in this study.

Written by: Nayana Patel, M.D., Associate Professor of Radiology, Abdominal Imaging Fellowship Program Director, University of Colorado Denver SOM, Aurora, Colorado

Read the Abstract