J Urol. 2025 May;213(5):590-599. doi: 10.1097/JU.0000000000004437. Epub 2025 Apr 9.
ABSTRACT
PURPOSE: The purpose of this study was to compare performance of Stockholm3 in an external validation with commonly used prostate cancer biomarkers and risk calculators.
MATERIALS AND METHODS: SEPTA was a multicenter trial validating Stockholm3 in a racially/ethnically diverse population of men meeting local care guidelines for prostate biopsy (2019-2023). In total, 2115 (98%) men with complete data for risk calculators and biomarkers were included. The primary outcome was detection of Grade Group ≥ 2 (GG ≥ 2) cancer. Predictors included Stockholm3, free/total PSA ratio, PSA density, European Randomized Study of Screening for Prostate Cancer-4, Prostate Biopsy Collaborative Group, and Prostate Cancer Prevention Trial version 2 risk calculators. Performance characteristics were computed at clinically used thresholds for each risk score. ROC analysis, graphical calibration assessment, and decision curve analysis were performed.
RESULTS: Among 2115 men, median age was 63 years (IQR: 58-68), median PSA was 6.1 ng/mL (IQR: 4.5-9.0), 415 (20%) had a prior negative prostate biopsy, and 356 (17%) had an MRI performed before biopsy. There were 1200 (56.7%) benign biopsies performed, 307 (14.5%) GG1 cancers detected, and 608 (28.8%) GG ≥ 2 cancers detected. The Stockholm3 test had superior discrimination (all P < .001) compared with all evaluated biomarkers and risk calculators with an AUC of 0.82 vs 0.72 for free/total PSA, 0.76 for PSA density, 0.77 for European Randomized Study of Screening for Prostate Cancer-4, 0.74 for Prostate Biopsy Collaborative Group, and 0.78 for Prostate Cancer Prevention Trial version 2. Decision curve analysis demonstrated superior performance of Stockholm3, showing the highest positive net benefit. Compared with free/total PSA, Stockholm3 could reduce unnecessary biopsies by 44% while maintaining a 0.95 sensitivity.
CONCLUSIONS: Stockholm3 outperforms other commonly used biomarkers and risk calculators for detecting GG ≥ 2 cancer in a diverse population.
PMID:40202133 | DOI:10.1097/JU.0000000000004437
