RISK STRATIFYING PIRADS 4 LESIONS AND IDENTIFYING THOSE WITH VERY HIGH CANCER DETECTION RATES ON PROSTATE MRI.

Although MRI has aided in prostate cancer diagnosis, including with risk stratification and targeted-biopsies, and have high negative predictive values, false positives have continued to be a limitation, and is thought to be approximately 40-60% for PIRADS 4 or "high suspicion" lesions, with positives defined pathologically as at least Gleason 3+4. We hypothesized that this low positive predictive value for PIRADS 4 suggests heterogeneity within this category, that PIRADS 4 lesions can be further risk-stratified, and that there is a subset of PIRADS 4 lesions that are higher suspicion, and would therefore have higher cancer detection rates when biopsied. The aim of this retrospective cohort study is to determine whether this subset of PIRADS 4 lesions can be identified on biparametric prostate MRI (bpMRI). All bpMRIs performed;at a single healthcare system;from July 2021 to December 2022;and interpreted by a single board-certified radiologist with over ten years experience;were identified as part of our;HIPAA-compliant, IRB-approved investigation. Studies with PIRADS 4 lesions were retrospectively reviewed. A;strategy of inclusion and exclusion was used to identify lesions that can be categorized into a higher suspicion group, "4c (concern)". Inclusion criteria included focal, homogeneous, circumscribed, T2 moderate to marked hypodensity and marked restricted diffusion for peripheral zone lesions,;while exclusion criteria included any possible mimicker of prostate cancer, or;with any other uncertainty that the lesion could definitively represent a cancer. Then, radiology-pathology correlation was performed for all the patients with "4", "4c" and PIRADS 5 lesions who underwent prostate biopsies. Statistical analysis was performed using Fisher's exact test. 413 bpMRIs were identified (3 PIRADS 1, 136 PIRADS 2, 100 PIRADS 3, 120 PIRADS 4, 42 PIRADS 5, and 12 nondiagnostic). 45;of 120 PIRADS 4 were categorized as "4c", with 75;remaining as "4". 53 out of;75 "4", 38 out of 45 "4c", and 30 out of 42;PIRADS 5 underwent biopsies. 35, 5, and 13 of "4"; 36, 2, and 0 of "4c"; 26, 1, and 3 of PIRADS 5 had ≥Gleason 3+4 (clinically significant cancer or "csPCa"), Gleason 3+3 (clinically insignificant cancer;or "ciPCa"), and negative biopsy, respectively. The cancer detection rates for csPCa (defined as csPCa/(csPCa+ciPCa+negative biopsy));for patients who underwent biopsy for "4" was 66% and "4c" was 95%, which was statistically significant (p=0.0016), while that for all PIRADS 4 was 78% and PIRADS 5 was 87%. (Please see Table 1 and Figure 1). There is a distinct subset of PIRADS 4 lesions that can be identified on biparametric prostate MRI. When biopsied, these lesions, termed "4c", have cancer detection rates higher than other PIRADS 4 lesions and similar to those of PIRADS 5 or "very high suspicion" lesions. These findings indicate that PIRADS 4 lesions can be further risk stratified on MRI. Risk stratifying and, if possible, identifying;PIRADS 4 lesions that are higher or "very high suspicion";will mitigate the limitation of false positives on prostate MRIs, and aid;patient management and prostate cancer diagnosis. For example, in the context of a negative prostate biopsy of a patient with a "very high suspicion" PIRADS 4c lesion, it is not unreasonable to consider a repeat biopsy or close clinical followup. [ABSTRACT FROM AUTHOR]