Prostate-specific antigen density (PSAD) is commonly used in the diagnostic evaluation and risk stratification of prostate cancer, particularly in guiding decisions for biopsy and active surveillance. However, its prognostic role in predicting postoperative outcomes such as biochemical recurrence (BCR) or persistent PSA after radical prostatectomy remains uncertain.

This retrospective study included patients diagnosed with localized prostate cancer who underwent radical prostatectomy. Patients with locally advanced or metastatic disease, or those who received neoadjuvant therapy, were excluded. PSAD was calculated preoperatively using initial PSA and prostate volume. The primary outcomes were BCR, defined as a post-nadir PSA rise >0.2 ng/mL, and a composite endpoint of BCR or persistent PSA, defined as postoperative PSA persistently >0.2 ng/mL without reaching nadir.

Of 82 patients included in the discovery cohort. The mean age was 68.1 ± 8.7 years, with a mean initial PSA of 15.6 ± 18.3 ng/mL and a mean prostate volume of 81.8 ± 41.9 cc. The average PSA density was 0.42 ± 0.38. Most patients had clinical stage T1–T2a (54.9%), and the most common post-operative Gleason score was 3+4 (48.8%). During follow-up, BCR occurred in 13 patients (15.9%), persistent PSA in 2 (2.4%), and 15 (18.3%) met the composite endpoint. Salvage therapy was performed in 16 patients (19.5%). ROC analysis demonstrated that PSAD had an AUC of 0.635 (95% CI: 0.522–0.739, p = 0.181) for predicting BCR, and an AUC of 0.594 (95% CI: 0.480–0.701, p = 0.323) for the combined outcome. At a cutoff of >0.60, specificity was high (~90%), but sensitivity was low (40–46%). Univariate logistic regression revealed that PSA density was significantly associated with BCR (OR: 4.51; 95% CI: 1.17–17.41; p = 0.029), along with advanced clinical stage T3 (OR: 6.00; 95% CI: 1.03–34.95; p = 0.046). However, in multivariate analysis, neither PSAD (OR: 3.46; 95% CI: 0.77–15.56; p = 0.106) nor clinical stage remained statistically significant. For the combined outcome of BCR + persistent PSA, PSAD showed a borderline association in univariate analysis (OR: 3.39; 95% CI: 0.91–12.58; p = 0.068), but was not significant in multivariate analysis (OR: 2.52; 95% CI: 0.59–10.72; p = 0.212).

Although PSA density >0.60 demonstrates high specificity for predicting postoperative oncologic failure. However, it had limited sensitivity and lack of independent predictive value in multivariate analysis. These findings suggest that PSAD should be interpreted in the context of additional clinical conditions for optimal risk stratification.