Whole-gland (WG) prostate-specific antigen (PSA) density (PSAD) has proven useful in diagnosing to be beneficial in localized prostate cancer (PCa). This study aimed to evaluate the predictive performance of WG and zonal (transition zone [TZ] and peripheral zone [PZ]) PSAD in predicting PCa and clinically significant PCa (csPCa) in prostate MRI.
A retrospective analysis was conducted on consecutive patients who underwent multiparametric MRI and MRI/US fusion-guided biopsy between March 2019 and July 2024. TZ-PSAD, PZ-PSAD, and WG-PSAD were calculated using in-house AI models. Optimal thresholds for TZ-PSAD and PZ-PSAD were determined using the Youden index from receiver operating characteristic (ROC) curve analyses with five-fold cross-validation, whereas 0.15 ng/mL2 was applied as the threshold for WG-PSAD. Statistical comparisons were performed using Wilcoxon rank-sum, χ2, and Fisher's exact tests. Logistic regression (LR) and area under the ROC curve (AUC) analyses with DeLong's test were conducted to evaluate diagnostic performance.
The study cohort included 774 consecutive patients (median age = 67 years [interquartile range {IQR}: 61-71], median WG-PSAD = 0.11 ng/mL2 [IQR: 0.07-0.17], median TZ-PSAD = 0.22 ng/mL2 [IQR: 0.12-0.41], median PZ-PSAD = 0.13 ng/mL2 [IQR: 0.16-0.34]). Among these patients, 475 had PCa and 341 had csPCa. The mean optimal thresholds for TZ-PSAD and PZ-PSAD were 0.20 ng/mL2 and 0.21 ng/mL2, respectively, for PCa, whereas they were 0.26 and 0.23, respectively, for csPCa. Multivariable LR identified TZ-PSAD (OR = 2.00, p = 0.03) and WG-PSAD (OR = 2.40, p = 0.02) as significant predictors of PCa. For csPCa, TZ-PSAD was the only independent predictor (OR = 2.13, p = 0.02) among PSAD measurements. TZ-PSAD showed a superior AUC for both PCa (0.79 ± 0.05) and csPCa (0.77 ± 0.02) compared to WG-PSAD (0.77 ± 0.06 for PCa, 0.76 ± 0.03 for csPCa) and PZ-PSAD (0.69 ± 0.06 for PCa, 0.70 ± 0.04 for csPCa; p < 0.001).
Both TZ-PSAD and WG-PSAD are strong predictors of PCa, but TZ-PSAD is a superior predictor of csPCa compared to WG-PSAD and PZ-PSAD. Further prospective studies are warranted to validate these findings.
NCT03354416.
The Prostate. 2025 Feb 25 [Epub ahead of print]
Omer Tarik Esengur, Emma Stevenson, Hunter Stecko, Nathan S Lay, Dong Yang, Jesse Tetreault, Ziyue Xu, Daguang Xu, Enis C Yilmaz, David G Gelikman, Stephanie A Harmon, Maria J Merino, Sandeep Gurram, Bradford J Wood, Peter L Choyke, Peter A Pinto, Baris Turkbey
Molecular Imaging Branch, National Cancer Institute, National Institutes of Health, Bethesda, Maryland, USA., NVIDIA Corporation, Santa Clara, California, USA., Laboratory of Pathology, National Cancer Institute, National Institutes of Health, Bethesda, Maryland, USA., Urologic Oncology Branch, National Cancer Institute, National Institutes of Health, Bethesda, Maryland, USA., Center for Interventional Oncology, National Cancer Institute, National Institutes of Health, Bethesda, Maryland, USA.