Comparing the diagnostic accuracy of PI-RADS Version 1 and PI-RADS Version 2 using a radiologicpathologic correlation
Sara Fardin, Michael Nguyentat, Shane Knipping, Alexander Ushinsky, Layla Shirkhoda, Alessandra Miranda Aguirre, Christopher Green, Edward Uchio, Chandana Lall and Roozbeh Houshyar
University of California, Irvine, USA
: J Clin Exp Oncol
Introduction: Multiparametric MRI (mpMRI) has been developed as an adjuvant tool for prostate cancer screening. MpMRI according to PI-RADSv1 criteria improved detection of prostate cancer, but studies evaluating PI-RADSv2 are lacking. Utilizing radiologic pathologic correlation through a combination of TRUS guided 12-core and MRI/TRUS fusion biopsies; we compared the diagnostic accuracy of mpMRI according to PI-RADSv1 versus PI-RADSv2. Methods: We retrospectively reviewed charts of patients with mpMRI utilizing PI-RADSv1 and subsequent combination biopsy (TRUS-guided 12-core biopsy plus MRI/TRUS fusion biopsy). The mpMRI studies were then re-graded to PI-RADSv2, by a radiologist blinded to biopsy results. An mpMRI with PI-RADS score of 3-5 was considered positive. Clinically significant cancer was defined as Gleason Score â¥7 or 6 with â¥50% tumor on core. Chi-square and ROC analyses were performed. Results: Our analysis included 76 men with a total of 124 identified prostate lesions on mpMRI. For clinically significant cancer, PIRADSv1 had a NPV, PPV, sensitivity, and specificity of 100%, 33.0%, 100%, and 20.2%, respectively. PI-RADSv2 had values of 100%, 41.2%, 100%, and 43.8%, respectively. Chi-square analysis showed a significant association between clinically significant cancer and both PI-RADSv1 score (p=0.004) and PIRADSv2 score (p<0.001). On ROC analysis, the AUC was 0.819 and 0.856 for PI-RADSv1 and PI-RADSv2, respectively. Discussion: MpMRI utilizing PI-RADSv2 criteria achieves a higher specificity and PPV. While both PI-RADS scores show statistically significant correlation with clinically significant cancer, the correlation was stronger using PI-RADSv2. Furthermore, there was an AUC increase of 0.037 when using PI-RADSv2 criteria. Compared to PI-RADSv1, PI-RADSv2 appears to have superior diagnostic accuracy for detecting clinically significant prostate cancer.
Sara Fardin has completed her Medical School in Tehran University of Medical Sciences and Radiology-Molecular Imaging Research Fellowship in the University of Pennsylvania. She is currently working as a Research Fellow in University of California, Irvine.