Abstract | November 8, 2021

Accuracy of Prostate Imaging Reporting and Data System Scores Among Varying Radiologists: A Retrospective Study

Presenting Author: Robert H. Drury, B.S., Medical Student, 4th Year, Tulane University School of Medicine, New Orleans, Louisiana, New Orleans, LA

Coauthors: Scott Brimley, MD, Department of Urology, PGY1, University of Kentucky, Lexington, Kentucky; Jacob Greenberg, BS, Medical Student, Tulane University, New Orleans, Louisiana; Robert Drury, BS, Medical Student, Tulane University, New Orleans, Louisiana; Quinn Smith, MD, Department of Radiology, PGY1, Creighton University, Omaha, Nebraska; Brian Dick, MD, Department of Urology, PGY1, University of California (San Francisco), San Francisco, California; Michael Polchert, BS, Medical Student, Tulane University, New Orleans, Louisiana; Jonathan Silberstein, MD, Department of Urology, Tulane University, New Orleans, Louisiana; Raju Thomas, MD, FACS, FRCS, MHA, Department of Urology, Tulane University, New Orleans, Louisiana; Louis Krane, MD, Department of Urology, Tulane University, New Orleans, Louisiana.

Learning Objectives

  1. Describe the diagnostic algorithm for detecting prostate cancer, including digital rectal exams, prostate-specific antigen levels, prostate biopsies, and multiparametric magnetic resonance imaging/PI-RADS scores. 
  2. Discuss the accuracy and prognostic benefits of PI-RADS scores. 

Background/Knowledge Gap: Multiparametric magnetic resonance imaging (mpMRI) helps detect prostate cancer. Radiologists reading prostate mpMRI use the Prostate Imaging Reporting And Data System (PI-RADS) to estimate risk of malignancy. Though used by large expert centers and small individual medical centers alike, PI-RADS accuracy varies between radiologists, necessitating internal validation. We evaluated PI-RADS score accuracy between our large institution and external, private radiologists. 

Methods/Design: Data was analyzed from an Institutional Review Board approved, prospectively maintained database of 96 patients who had undergone prostate biopsies using fusion technology from December 2018 to September 2020. Demographic, radiologic, and biopsy data were collected and reviewed. Statistics were performed using commercial software. 

Results/Findings: Results from our institutional radiologists and private radiologists were compared. Across 96 patients total, our institution found 89 lesions and external radiologists found 81 lesions. No statistically significant differences in patient demographics were seen between groups. Patients had an average prostate-specific antigen (PSA) of 8.23 ng/ml, PSA density of 0.187 ng/ml2, and MRI volume of 57.6 ml. The clinically significant cancer (grade group ≥2) detection rate was 15.79% for PI-RADS 3, 20.0% for PI-RADS 4, and 44.44% for PI-RADS 5 at our institution. The rate was 10.0% for PI-RADS 3, 16.67% for PI-RADS 4, and 33.33% for PI-RADS 5 for private radiologists. When comparing cancer detection rates per PI-RADS score between groups, no statistically significant differences were seen. 

Conclusions/Implications: The PI-RADS scoring system is often subjective to individual radiologists. Even among patients with high PI-RADS scores, our large expert institution found relatively low clinically significant cancer rates. Private radiologists showed comparable PI-RADS readings. Though the potential benefit of PI-RADS scoring appears great, further teamwork between urologists and radiologists is needed to increase PI-RADS score accuracy.

Download Abstract
Back to Abstracts