Evaluation of ancillary multiparametric MRI features in PI-RADS 3 prostate lesions and their association with prostate cancer.

Abstract

26 Background: Prostate cancer (PCa) remains the leading malignancy among men in Colombia and a major cause of cancer-related mortality worldwide. Multiparametric MRI (mpMRI) has become a cornerstone in PCa detection and biopsy targeting. However, PI-RADS 3 lesions continue to represent a diagnostic gray zone, accounting for 20–30% of mpMRI findings, with reported clinically significant PCa (csPCa) detection rates ranging from 11–25%. Management of these indeterminate lesions remains a relevant clinical challenge. Our institution has observed higher-than-expected csPCa prevalence in PI-RADS 3 lesions, underscoring the need to evaluate mpMRI features and clinical parameters that may refine risk stratification. Methods: Retrospective, single-center study of men >18 years with PI-RADS 3 lesions on mpMRI who underwent targeted prostate biopsy at a low and middle-income country (LMICs) between January 2019 and December 2024. mpMRI were acquired on 1.5T and 3T scanners following PI-RADS v2.1 standards. Imaging was reviewed independently by two abdominal radiologists with ≥10 years of experience. Clinical (age, PSA, PSA density), imaging (lesion size, location, morphology, diffusion/ADC restriction, contrast enhancement), and histopathology (Gleason score, ISUP grade group) data were collected. Clinically significant PCa was defined as Gleason ≥3+4. Univariate, bivariate, and multivariate logistic regression analyses were performed to identify independent predictors of csPCa. Results: A total of 82 patients met inclusion criteria. Overall PCa prevalence was 58%, with csPCa in 42% notably higher than previously reported rates for PI-RADS 3 lesions. On multivariate analysis, low apparent diffusion coefficient (ADC) values, linear or irregular lesion morphology, and elevated PSA density were independently associated with csPCa (p < 0.05). Traditional mpMRI features such as T2 signal intensity, qualitative diffusion restriction, and early dynamic contrast enhancement did not show statistically significant associations. The predictive model demonstrated acceptable discrimination (AUC = 0.726). These findings could suggest that integration of quantitative ADC metrics, PSA density, and lesion morphology may enhance risk stratification and therefore clinical management. Conclusions: Ancillary imaging and clinical features, specifically quantitative ADC values, PSA density, and lesion morphology emerged as robust predictors of csPCa. Incorporating these parameters into diagnostic pathways could ameliorate the proper selection of patients requiring prostate biopsies while improving early detection of aggressive disease. Our results highlight the need to refine PI-RADS guidelines particularly in the PI-RADS 3 category, and adapt biopsy strategies to diverse healthcare settings.