Our results demonstrate that detection of overall and CS-PCa on FB may be higher in lesions identified on bpMRI compared to those on mpMRI. In various group sub-analyses, bpMRI performed either similarly or somewhat superior to mpMRI. This study adds to the literature supporting bpMRI as an appropriate alternative to mpMRI for PCa evaluation and validates it among a racially diverse patient population [5, 9, 12, 15, 18,19,20].

We found the odds of detecting overall PCa on FB were higher among PI-RADS 3 lesions using bpMRI than mpMRI, with trends similar but non-significant for CS-PCa. This suggests that PI-RADS 3 lesions detected on bpMRI may require a more aggressive workup, particularly when deciding whether a targeted biopsy is indicated for these typically lower-risk lesions. Similar findings were reported in a study of 123 men that found bpMRI had higher specificity than mpMRI for the detection of CS-PCa [21]. However, in a retrospective review of 235 patients, mpMRI had a higher detection of CS-PCa upon biopsy than bpMRI among patients with PI-RADS 4 lesions [20]. Therefore, larger datasets are needed to better resolve the differences per PI-RADS lesion.

Our analysis of men enrolled in AS protocol had similar outcomes when comparing bpMRI to mpMRI. Similarly, a review of 101 AS patients, who underwent diagnostic FB with bpMRI and follow-up confirmatory MRI with mpMRI, found that confirmatory mpMRI failed to detect higher rates of missed lesions, miscategorized PI-RADS risk assessments, or variation in tumor size, compared to bpMRI alone [22]. Our study supports prior evidence that bpMRI is a sufficient alternative to mpMRI in AS patients, including in high-risk populations.

Our study population was highly diverse, comprising nearly 50% NHB men and nearly 30% Hispanic men. When stratified by race/ethnicity, we found no differences in the odds of yielding PCa or CS-PCa on biopsied lesions identified by either MRI protocol. That said, our results showed a trend towards an increased odds of detecting CS-PCa in NHB men among lesions identified on bpMRI, which approached, but did not reach, significance. A larger study population may be needed to determine if there is a significant difference. Overall, prior work on this topic has been hindered by poor enrollment of racial/ethnic minorities and a lack of reporting of racial/ethnic demographic data. Studies comparing mpMRI to bpMRI in somewhat diverse cohorts have found similar performance for detecting CS-PCa [23,24,25]. However, NHW men were still vastly overrepresented in these studies, and they did not report sub-analyses stratified by race/ethnicity. Furthermore, a prior study at our institution showed NHB men have higher odds of detecting overall and CS-PCa on FB (both bpMRI and mpMRI) compared to NHW men [26]. Our study adds to the body of evidence that bpMRI is an appropriate diagnostic modality alternative to mpMRI for NHB and Hispanic men.

Our study has several strengths and limitations. Strengths include sample size and enrollment of a diverse patient population. Additionally, we utilized mixed-effects logistic regression, which considers patient-specific and lesion-specific factors when analyzing the odds of detecting PCa on biopsy. Routine multidisciplinary conferences were utilized to review challenging cases, and the urologists and radiologists involved were consistent throughout the study. Limitations include its retrospective nature. Additionally, we did not perform a head-to-head comparison of mpMRI and bpMRI protocols; our patients received either mpMRI or bpMRI based on date ordered (per our institution’s decision to change the imaging protocol). We also did not collect data on which patients received MRI using 1.5T vs 3T instruments. However, studies comparing 1.5T vs 3T instruments have found similar PI-RADS scoring when read by fellowship-trained body radiologists [27, 28].

All exams in this study meet the updated technical parameters recommended in PI-RADS V2.1. Due to logistical challenges in 2020 and the early half of 2021, there was a delay in implementing the imaging interpretation criteria of PI-RADS V2.1 at our institution. All the studies performed during this time were reported using the PI-RADS V2. Recent literature has shown that upgrade and downgrade rates are not significantly different between PI-RADS V2 and V2.1 [29].

Furthermore, our study did not compare the PI-RADS 4 lesions between the 2 groups to differentiate between the dominant PI-RADS 4 lesions on bpMRI and the upgraded PI-RADS 4 lesions on mpMRI that were upgraded by DCE (DWI score 3 with positive DCE). There is varied literature on the benefits of DCE with some studies showing no added benefit to including DCE [9, 13], while other studies have demonstrated benefit only in inexperienced readers [30].

Additionally, our study may be subject to selection bias given that we did not match patients based on patient characteristics between mpMRI and bpMRI protocols. Of note, our patient population had imbalances in the proportion of PI-RADS 3 and 4 lesions between mpMRI and bpMRI protocols. We performed stratified analyses by PI-RADS score to control for this variation, but caution must be applied given the imbalances. Finally, six fellowship-trained body radiologists reviewed our MRI scans, which may contribute to greater inter-reader variability. However, this reflects the reality of clinical practice, and therefore, this study provides results that can be expected in a true clinical setting, as opposed to a strictly controlled research study.