Multiparametric MR Imaging for Detection of Clinically Significant Prostate Cancer: A Validation Cohort Study with Transperineal Template Prostate Mapping as the Reference Standard

Purpose: To evaluate the diagnostic performance of multiparametric (MP) magnetic resonance (MR) imaging for prostate cancer detection by using transperineal template prostate mapping (TTPM) biopsies as the reference standard and to determine the potential ability of MP MR imaging to identify clinically significant prostate cancer. Materials and Methods: Institutional review board exemption was granted by the local research ethics committee for this retrospective study. Included were 64 men (mean age, 62 years [range, 40-76]; mean prostate-specific antigen, 8.2 ng/mL [8.2 mu g/L] [range, 2.1-43 ng/mL]), 51 with biopsy-proved cancer and 13 suspected of having clinically significant cancer that was biopsy negative or without prior biopsy. MP MR imaging included T2-weighted, dynamic contrast-enhanced and diffusion-weighted imaging (1.5 T, pelvic phased-array coil). Three radiologists independently reviewed images and were blinded to results of biopsy. Two-by-two tables were derived by using sectors of analysis of four quadrants, two lobes, and one whole prostate. Primary target definition for clinically significant disease necessary to be present within a sector of analysis on TTPM for that sector to be deemed positive was set at Gleason score of 3+4 or more and/or cancer core length involvement of 4 mm or more. Sensitivity, negative predictive value, and negative likelihood ratio were calculated to determine ability of MP MR imaging to rule out cancer. Specificity, positive predictive value, positive likelihood ratio, accuracy (overall fraction correct), and area under receiver operating characteristic curves were also calculated. Results: Twenty-eight percent (71 of 256) of sectors had clinically significant cancer by primary endpoint definition. For primary endpoint definition (>= 4 mm and/or Gleason score >= 3+4), sensitivity, negative predictive value, and negative likelihood ratios were 58%-73%, 84%-89%, and 0.3-0.5, respectively. Specificity, positive predictive value, and positive likelihood ratios were 71%-84%, 49%-63%, and 2.-3.44, respectively. Area under the curve values were 0.73-0.84. Conclusion: Results of this study indicate that MP MR imaging has a high negative predictive value to rule out clinically significant prostate cancer and may potentially have clinical use in diagnostic pathways of men at risk. (C) RSNA, 2013