Assessment of PI-RADS v2 categories ≥ 3 for diagnosis of clinically significant prostate cancer

Nayana U. Patel*, Kimberly E. Lind, Kavita Garg, David Crawford, Priya N. Werahera, Sajal S. Pokharel

*Corresponding author for this work

Research output: Contribution to journalArticle


Purpose: To assess the diagnostic accuracy of PI-RADS v2 categories ≥ 3 to detect clinically significant prostate cancer (csPCa) against histopathology of Transperineal Mapping Biopsy (TPMB). Materials and methods: IRB-approved retrospective cohort study included 47 men who had 3.0 T multi-parametric MRI (mpMRI) and TPMB of prostate. Two radiologists independently evaluated T2, DWI, ADC map, and DCE images using PI-RADS v2 categories. A third radiologist served as tie-breaker. PI-RADS v2 score (PS) ≥ 3 lesions were correlated with 3D model of TPMB (3DTPMB) results based on prostate sectors. Two groups of csPCa status were separately analyzed for accuracy measures at lesion and person levels: Group 1 with GS (Gleason Score) ≥ 7 and group 2 with tumor volume ≥ 0.5 cc. Inter-rater reliability for PS and MR lexicon was calculated. Results: Forty-seven patients with 3DTPMB had at least one lesion with PS ≥ 3 on mpMRI. PS of 5 had high PPV and high specificity of 100% at the lesion and person levels. Sensitivity of a PS ≥ 3 was 68.27% for group 1 and was 48.39% for group 2. Specificity was 93.56% for group 1 and was 95.53% for group 2. At the person level, sensitivity of PS ≥ 3 was 81.25% for group 1 and was 82.35% for group 2. Specificity was 32.26% for group 1 and was 53.85% for group 2. Conclusion: PI-RADS v2 category of 5 had high PPV and specificity; however, combined PS ≥ 3 had mixed performance in detection of csPCa.

Original languageEnglish
Pages (from-to)705-712
Number of pages8
JournalAbdominal Radiology
Issue number2
Early online date31 Aug 2018
Publication statusPublished - 15 Feb 2019
Externally publishedYes



  • Clinically significant prostate cancer
  • Multi-parametric MRI
  • PI-RADS v2
  • Prostate cancer
  • Prostate MRI

Cite this