Abstract
Objective
PI-RADS 3 lesions are equivocal on prostate MRI, complicating biopsy decisions. We aimed to develop and externally validate a simple pre-biopsy risk stratification model using routinely available clinical and MRI variables to guide selective biopsy.
Methods
In this multicenter retrospective study, men with PI-RADS 3 lesions undergoing transperineal biopsy at two tertiary hospitals (Jan 2023–Dec 2025) were included. One center served as the development cohort and the other as external validation. Clinically significant prostate cancer (csPCa) was defined as ISUP Grade Group ≥2. Candidate predictors included age, abnormal digital rectal examination (DRE), PSA density (PSAD), lesion location, and maximum lesion diameter. A multivariable logistic regression model was developed and validated externally.
Results
Among 275 men (153 development, 122 validation), csPCa was detected in 37 (24.2%) and 31 (25.4%), respectively. PSAD and age were the main predictors, with other variables contributing to overall risk estimation. The model showed good discrimination (AUC 0.818 development, 0.846 validation), acceptable calibration, and decision curve analysis indicated potential clinical utility.
Conclusion
This preliminary model using routinely available clinical and MRI variables may support pre-biopsy risk stratification in men with PI-RADS 3 lesions. However, further validation in larger and more diverse cohorts is required before routine clinical implementation.