Incremental value of prostate-specific antigen density beyond PI-RADS and conventional MRI features for predicting the highest ISUP grade group in patients with prostate cancer diagnosed at initial systematic prostate biopsy.

OBJECTIVE: To evaluate the incremental value of prostate-specific antigen density (PSAD) beyond the Prostate Imaging Reporting and Data System (PI-RADS) score and conventional magnetic resonance imaging (MRI) features for predicting the highest International Society of Urological Pathology (ISUP) grade group at initial systematic biopsy. METHODS: This retrospective study included 171 patients with prostate cancer diagnosed by initial systematic biopsy between February 2022 and February 2026. All patients underwent 3.0-T multiparametric MRI within 15 days before biopsy. The highest ISUP grade group across all biopsy cores was the primary outcome. A baseline model, an MRI-extended model, and a combined model incorporating PSAD were sequentially developed using proportional odds ordinal logistic regression. Internal validation was performed with 2,000 bootstrap resamples. Firth logistic regression and sensitivity analyses were used to assess robustness. RESULTS: The numbers of patients with ISUP grade groups 1-5 were 26, 10, 9, 40, and 86, respectively; 135 patients (78.9%) had grade group ≥3. Median PSAD increased from 0.20 ng/mL2 in grade group 1 to 1.38 ng/mL2 in grade group 5 (P<0.001). After adjustment for age, PI-RADS score, MRI-suspected local progression, and multizonal involvement on MRI, each doubling of PSAD was associated with higher ISUP grade group (odds ratio [OR], 2.04; 95% confidence interval [CI], 1.59-2.62; P<0.001). Adding PSAD significantly improved model fit (likelihood ratio χ2=37.83; P<0.001) and increased the bootstrap-corrected ordinal C-statistic from 0.660 to 0.744. Paired out-of-bag analysis showed a median increase of 0.094 in the ordinal C-statistic (95% CI, 0.012-0.171). The combined model achieved an out-of-bag area under the receiver operating characteristic curve of 0.835 for grade group ≥3. The association remained robust after excluding patients with total prostate-specific antigen >100 ng/mL (OR, 2.36; 95% CI, 1.68-3.29) and among patients with PI-RADS scores ≤3 (OR, 2.60; 95% CI, 1.46-5.58; P<0.001). CONCLUSIONS: PSAD was independently associated with the highest biopsy ISUP grade group and provided incremental predictive information beyond PI-RADS and conventional MRI features. PSAD may also aid in identifying patients with low PI-RADS scores but high pathological grade groups, although external validation is required.