AUA 2017: Radiologist Experience Level Does Not Predict the Accuracy of Prostate MRI Interpretation for Clinically Significant Prostate Cancer: Are Consensus Reads the Answer?

Boston, MA ( At this morning’s Uroradiology poster session at the 2017 AUA Annual meeting, Dr. Kim and colleagues from Washington University presented their study assessing radiologist experience with prostate MRI imaging for predicting clinically significant prostate cancer. Certainly in this era of multi-parametric MRI imaging with ever increasingly advanced technology, accurate PIRADS scoring of prostate lesions is crucial.

In this study, the authors used their prospectively maintained prostate MRI database to assess 32 patients that underwent a prostate MRI and subsequent prostate biopsy. The primary outcome was to examined the predictive accuracy of a PIRADS 4 or 5 read for clinically significant (Gleason 7+) prostate cancer in a blinded fashion. Four radiologists of varying levels of experience independently interpreted all prostate MRIs and were blinded to the clinical outcomes. The authors defined an ‘over-read’ as a PIRADS 4 or 5 lesion with resulting biopsy as either benign or Gleason 6 disease, and an ‘under-read’ was defined as a PIRADS 1-3 lesion with resulting biopsy of Gleason 7 or greater prostate cancer. Between radiologists there was no difference in accuracy of prostate MRI interpretation (ranging from 56-75%). If two or more radiologists were in agreement (ie. a ‘majority-read’) this improved interpretation over the individual assessments, although this was not statistically significant (p=0.16). The strength of this study is an appropriate study design strategy to evaluate the question of whether consensus interpretation among radiologists should be further evaluated, however the main limitation is a small sample size inadequate for drawing clinically meaningful conclusions.

It is important for expert prostate MRI interpretation, since these opinions drive further targeted biopsies in appropriately selected individuals. This study is hypothesis generating, particularly the utilization of a consensus interpretation for improving diagnostic accuracy, and should be evaluated in larger, multi-institutional studies.

Presented By: Eric Kim, Washington University, St. Louis, MO, USA

Co-Authors: Joel Vetter, Anup Shetty, Kathryn Fowler, Aaron Mintz, Cary Siegel, Gerald Andriole, Robert Grubb III

Written By: Zachary Klaassen, MD, Urologic Oncology Fellow, University of Toronto, Princess Margaret Cancer Centre
Twitter: @zklaassen_md

at the 2017 AUA Annual Meeting - May 12 - 16, 2017 – Boston, Massachusetts, USA