SUO 2017: Prostate MRI Interpretation Varies Substantially Across Radiologists
Methods: Participants were a consecutive cohort of men at Stanford who underwent mpMRI and MR-US fusion biopsy between 2014 and 2016. Subjects were either biopsy-naïve or had undergone prior biopsy. All MRIs were read by 1 of 9 different radiologists using PIRADS scoring system. MRIs were not re-read prior to biopsy by another radiologist. Biopsy histopathology served as the standard for evaluating presence of cancer. Primary outcomes were the PIRADS score distribution and diagnostic accuracy across 9 radiologists. Clinically significant cancer was defined as Gleason ≥7. Multivariable logistic regression analysis was used to evaluate the association between age, PSA, PIRADS score, and radiologist in predicting significant cancer. Sensitivity analysis included radiologist prostate mpMRI case volume and changes in accuracy over time.
Results: A total of 409 subjects with 509 MRI lesions were analyzed. While the mean number of lesions assigned per patient (1.2) did not differ across radiologists, substantial variation existed in PIRADS distribution and cancer yield. Significant cancer detection ranged from 0-23% for PIRADS 3, 23-65% for PIRADS 4 and 40-80% for PIRADS 5 across radiologists. The proportion of men with PIRADS < 3 found to have clinically significant cancer varied from 7-50%. AUC was 0.70 (range of 0.63-0.80) for all prostate cancers and 0.72 (range of 0.62-0.82) for clinically significant cancer. Multivariable analysis showed that PIRADS score (p<0.001) and radiologist (p=0.03) were independent predictors of cancer. Neither individual radiologist case volume, nor study period impacted results.
Conclusion: The authors concluded that considerable variability in PIRADS score assignment and diagnostic accuracy exists across radiologists in prostate MRI. Therefore, the authors suggest that internal validation of MRI scan interpretations with biopsy findings be performed before using MRI to make important clinical decisions.
Presented by: Nancy Wang, Stanford, CA, USA
Written by: Hanan Goldberg, MD, Urologic Oncology Fellow (SUO), University of Toronto, Princess Margaret Cancer Centre, @GoldbergHanan, at the 18th Annual Meeting of the Society of Urologic Oncology, November 29-December 1, 2017 – Washington, DC