PI-RADS Not As Effective with Prostate MRI As Hoped

April 22, 2020

Use of the categorization tool produces results that vary widely across institutions.

Using the Prostate Imaging Reporting and Data System (PI-RADS) with MRI produces varying results when trying to pinpoint whether a patient has prostate cancer, new research has revealed.

In an international study, conducted in multiple centers and published on April 21 in Radiology, researchers, led by Antonio C. Westphalen, M.D., Ph.D., a radiologist with the University of California at San Francisco, determined the positive predictive value (PPV) of PI-RADS was low.

“PI-RADS must be reliable to guide the diagnostic process, assist with management decisions, and improve patient outcomes,” the authors wrote. “Therefore, the wide variability in PPV could hinder managing physician confidence in the system, affecting the broad acceptance and use of PI-RADS.”

Based on an evaluation of results from more than 3,400 men, the investigators determined that PI-RADS scores of 3 or 4 had a PPV of 35 percent and 49 percent, respectively, and they ranged between 27 percent and 65 percent.

Initially, the PI-RADS system was created to help standardize the analysis of the MRI images used to detect the presence and severity of prostate cancer. But, previous studies conducted in individual centers have cast doubt on whether smaller institutions that might not have staff as familiar with PI-RADS can use the system as well as large facilities. Consequently, a more thorough evaluation of how PI-RADS performs with MRI in various practice settings would help in determining performance benchmarks, as well as quality improvement initiatives.

In an effort to better understand the effectiveness of the PI-RADS system, Westphalen’s team conducted a retrospective study that measured the PPV of radiologists who used the categorization tool. The investigation, conducted in 26 centers in the United States, Canada, Brazil, the Netherlands, and South Korea, included 3,449 men with an average age of 65 years who had suspected or biopsy-confirmed, untreated prostate cancer. Most patients received prostate MRI imaging at either 1.5T (451 patients) or 3T (2,981 patients).

According to study results, the MRI imaging pinpointed 5,082 lesions. Most patients were categorized as PI-RADS 4, with PI-RADS 3 being the next closest rating. Overall, investigators found, PPV results were low with scores of 2, 3, and 4, ranging from 27 percent to 65 percent. Such a wide variation could create problems, Westphalen’s team said. Not only could it impede the ability to assess radiologist or institutional PI-RADS performance based on benchmarks, but it could also reduce payer and policymaker support for prostate MRI.

Consequently, they wrote, the hope behind this study’s findings is that it will prompt additional research, as well as education and quality efforts, to reduce variability in PI-RADS use.