LAUSR

PurposeTo determine whether peripheral zone PI-RADS 4 observations can be further risk-stratified.MethodsThis was an IRB-approved HIPAA-compliant retrospective diagnostic accuracy study. Peripheral zone PI-RADS 4 observations prospectively identified at the study institution from 8/1/2015 to 12/31/2016 (n = 170 in 149 mpMRIs) were reviewed independently by two blinded genitourinary radiologists on the basis of (a) PI-RADS v2 shape, (b) pattern of peripheral zone sparing, and (c) rationale for PI-RADS 4 designation. Reference standard was targeted MR–ultrasound fusion biopsy and detection of Gleason 7+ prostate cancer. Positive predictive values (PPVs) were calculated. Predictors were assessed with binary logistic regression.ResultsPI-RADS 4 lesions with a DWI score of 4 were more likely to represent Gleason 7+ prostate cancer (p = 0.008–0.01; Reader 1 PPV: 53%; Reader 2 PPV: 48%). Pattern of peripheral zone sparing and most lesion shapes were not predictive (p > 0.05); however, oval lesions were predictive for Reader 1 (PPV = 59%, p = 0.03) and lentiform lesions were predictive for Reader 2 (PPV = 74%, p = 0.01). Lesions scored as “not meeting PI-RADS 4 criteria” had significantly lower PPV (p = 0.016–0.003; Reader 1 PPV: 14%, Reader 2 PPV: 16%).DiscussionPeripheral zone PI-RADS 4 lesions with a DWI score of 4 are more likely Gleason 7+ cancer than those with a DWI score of 3. Lesions overcalled as PI-RADS 4 have PPV similar to published PI-RADS 3 data. Lesion shape and peripheral zone sparing in general do not predict Gleason 7+ cancer within PI-RADS 4 observations.