The diagnostic accuracy of multiparametric magnetic resonance imaging before biopsy in the detection of prostate cancer

Objectives: To determine the extent to which clinically significant prostate cancer (csPCa) can be detected in a routine National Health Service setting in men with no previous biopsy, when multiparametric magnetic resonance imaging (mpMRI) is introduced into the diagnostic pathway.

Patients and methods: In all, 1 090 mpMRIs were performed between July 2013 and April 2016 in biopsy-naïve men with an abnormal prostate-specific antigen level and/or digital rectal examination. Data were collected from patient records at the Royal Devon and Exeter NHS Foundation Trust. mpMRI Prostate Imaging Reporting and Data System (PI-RADS) scores were compared to transperineal or transrectal ultrasonography (TRUS)-guided biopsy findings as the reference standard. csPCa was defined as Gleason score of ≥3+4. The diagnostic accuracy of mpMRI was also assessed.

Results: The mpMRI was interpretable in 1 023 men and 792 underwent biopsy, of which 106 were transperineal. The median number of cores taken in transperineal and TRUS-guided biopsy were 10 and 6, respectively. The detection rate of csPCa was 37%; csPCa rose from 15% of PI-RADS 1 and 2 to 86% of PI-RADS 5. The sensitivity, negative predictive value, specificity, and positive predictive value were 82%, 85%, 59% and 54%, respectively. The study is limited by its retrospective nature and lack of reporting of follow-up for 'missed cancers'. Men with low mpMRI PI-RADS were also less likely to undergo biopsy. Whilst this selection bias may overestimate the detection rate of csPCa, this reflects the shared decisions patients and clinicians make in day-to-day practice outside of research centres.

Conclusion: In a routine clinical setting, the higher the mpMRI PI-RADS, the greater the detection rate of csPCa in biopsy-naïve men. A normal mpMRI does not exclude csPCa; however, mpMRI may have utility in informing shared-decision making on whether to proceed to biopsy and subsequent treatment.