PURPOSE: The main aim of the study was to compare the diagnostic performance of Prostate Imaging Reporting and Data System (PI-RADS) versions 1 and 2 for detection of prostate carcinoma (PCa) and clinically significant prostate carcinoma (CSPCa). The second aim was to evaluate the potential benefit of adding the apparent diffusion coefficient (ADC) and prostate specific antigen (PSA) density to the standard evaluation protocol. METHODS: A total of 167 consecutive patients with elevated PSA underwent magnetic resonance imaging. The images were evaluated prospectively using both versions of the PI-RADS and the results compared with 12-core template biopsy and magnetic resonance/transrectal ultrasound fusion biopsy. Receiver-operating characteristic (ROC) curves were compared for each scoring system using DeLong\'s test. The area under the curve (AUC) was calculated for ADC and PSA density for lesions scored 4. RESULTS: PI-RADS V2 had high discriminative ability for PCa prediction with an AUC of 0.824 (95% CI 0.763 to 0.885), compared to an AUC of 0.724 (95% CI 0.654 to 0.794) for PI-RADS V1 (p = 0.0335). ADC demonstrated a higher discriminative ability with an AUC of 0.702 (95% CI 0.548 to 0.856) in CSPCa prediction. Using the obtained ADC threshold of 828x10^-6 mm^2/s improved specificity to 86.73% with a sensitivity of 60.38%. CONCLUSION: PI-RADS version 2 exhibited significantly higher discriminative ability for PCa and CSPCa detection compared to PI-RADS version 1. Using the ADC can improve the tumor predictability of PI-RADS version 2 in lesions scored 4.

Department of Radiology Na Homolce Hospital Prague Czech Republic

Department of Urology Thomayer Hospital Prague Czech Republic

National Institute of Public Health Prague Czech Republic

Radiology Department 3rd Faculty of Medicine Charles University Faculty Hospital Kralovske Vinohrady Prague Czech Republic