Development and validation of a pre-stratified strategy for predicting clinically significant prostate cancer in PI-RADS 3 lesions

The optimal management of patients with Prostate Imaging Reporting and Data System (PI-RADS) 3 lesions remains a clinical dilemma, primarily due to the lesions’ intermediate risk profile. This study aimed to develop a risk-adapted decision-making strategy by pre-stratifying patients according to biopsy history and lesion location. This retrospective study included 919 men with PI-RADS 3 lesions on mpMRI who underwent systematic prostate biopsy between January 2022 and June 2025. Patients were stratified according to biopsy history and lesion location. Multivariable logistic regression was employed to identify independent predictors of clinically significant prostate cancer (csPCa). The model’s discriminative performance was assessed using the area under the receiver operating characteristic curve (AUC). Clinical utility and net benefit were evaluated via decision curve analysis (DCA). Among the overall cohort of 919 patients—684 in the training set and 235 in the validation set—152 (16.5%) were diagnosed with csPCa. csPCa was more frequently detected in biopsy-naive patients and in those with peripheral zone (PZ) lesions. In the subgroup of biopsy-naive patients without PZ lesions, multivariable logistic regression identified age (OR = 1.07, 95% CI 1.01–1.13; P < 0.05), f/tPSA (OR = 0.0001352, 95% CI 2.823e − 7 - 0.02962; P < 0.01), and prostate‑specific antigen density (PSAD) (OR = 5.34, 95% CI 1.69–24.84; P < 0.05) as independent predictors of csPCa. The final model combining these three variables showed excellent discrimination, with an AUC of 0.839 in the training cohort. Using a 10% probability threshold, the optimized biopsy strategy would avert 44.6% of immediate biopsies while missing only 7.6% of csPCa cases with a sensitivity of 86% and a specificity of 72.5%. The application of the model to the validation cohort achieved an AUC of 0.746 and the optimized biopsy strategy would have deferred biopsy in 43.8% of patients while missing 14.7% of csPCa. This study proposed a novel, risk-adapted management strategy for patients with PI-RADS 3 lesions. This optimized approach would have deferred nearly half of unnecessary biopsies while maintaining a low rate of missed csPCa diagnoses.