Prostate cancer (PCa) is a leading malignancy among men, yet identifying clinically significant PCa (csPCa) remains challenging. Despite advances with mpMRI and PI-RADS scoring, PI-RADS 3 and 4 lesions still result in potentially unnecessary biopsies. This study explores whether integrating PSA-based biomarkers and clinical parameters can improve risk assessment and minimize avoidable interventions in a Taiwanese cohort.

Detailed medical records of 165 patients who underwent mpMRI and MRI-ultrasound fusion-guided biopsy between April 2021 and August 2024 were retrospectively reviewed. Patients were stratified into PI-RADS 3-4 and PI-RADS 5 groups. csPCa was defined as an ISUP grade ≥2. Logistic regression models were used to identify predictors of csPCa. The diagnostic performance of PI-RADS alone and in combination with PSA density (PSAD) and PSA velocity (PSAV) was assessed using receiver operating characteristic (ROC) analysis and decision curve analysis (DCA).

csPCa was detected in 30.8% of PI-RADS 3-4 lesions. PSAD ≥0.20 ng/mL/mL was an independent predictor of csPCa (OR: 12.167, p = 0.042). In the overall cohort, the combination of PI-RADS 3-4 with PSAD ≥0.20 ng/mL/mL and PSAV ≥0.75 ng/mL/year improved diagnostic accuracy (AUC: 0.624) compared to PI-RADS alone. In the PI-RADS 3-4 subgroup, the combined model achieved an AUC of 0.673. DCA demonstrated that this combined model provided the highest net benefit within a threshold probability range of 25%-50%, reducing unnecessary biopsies by approximately 40%. In the PI-RADS 3-4 subgroup, the combination model also yielded the greatest net benefit within the threshold range of 18%-50%.

Integrating PSAD and PSAV improves risk stratification in PI-RADS 3-4 lesions, offering a practical approach to optimize biopsy decision-making and minimize unnecessary procedures. DCA findings support the clinical utility of this model in identifying patients who may safely defer biopsy while maintaining high detection rates of csPCa.