Perspectives and Concerns about PI-RADS and Variability

Editor:

We applaud Dr Westphalen and colleagues for their efforts to evaluate the daily performance of Prostate Imaging Reporting and Data System (PI-RADS), published in the July 2020 issue of Radiology (1). They concluded that the wide variation and overall low positive predictive value (PPV) is due to PI-RADS, and that this “could hinder managing physician confidence in the system, affecting the broader acceptance and use of PI-RADS.” We agree PI-RADS updates must focus on quality, but do not agree with the conclusion that PI-RADS is the cause of the unreliability, hindering physician confidence and decreasing acceptance.

Is it fair to ascribe the PPV entirely to PI-RADS? Let us remember what PI-RADS is and what it was designed to do: It is a reporting and data system designed to promote global standardization and diminish variation in the acquisition, interpretation, and reporting of prostate multiparametric MRI (2). The assessment score depends on the image data and expertise of the radiologist and is but a single data point in the prostate cancer diagnostic pathway from presentation to diagnostic biopsy pathology. PI-RADS must always be combined with clinical factors like serum prostate-specific antigen kinetics, family history, digital rectal examination findings, and previous biopsy results.

PI-RADS is very helpful for decreasing the diagnosis of indolent disease through biopsy avoidance (3). This has been achieved without impairing detection of clinically significant prostate cancer. It is helpful in identification of targets, and MRI-guided biopsy leads to increased detection of cancers grade group 2 or greater compared with systematic-transrectal US biopsy, especially after prior biopsy that was negative for cancer (4).

Whereas the PI-RADS assessment may be a factor in variation, it is likely not as impactful as the heterogeneity in patients, magnet field strength and vendors, endorectal coil use, image quality, reader experience, threshold of biopsy positivity at MRI, biopsy targeting methods, systematic core samples, pathologist expertise (there is a known wide variability in Gleason score assignments), and importantly the experience of biopsy operators. These variables were not adequately controlled for in the study report. In fact, the PI-RADS PPV results are in line with a recent meta-analysis (5).

We therefore conclude that the variability in PPV described by Dr Westphalen and colleagues is most certainly related to all the factors in the prostate MRI pathway, and that PI-RADS behaves well.