Introduction:

Our study aimed to compare the performance of Stockholm3 in an external validation with other commonly used prostate cancer biomarkers and risk calculators, including free/total PSA ratio, PSA density, ERSPC-4/SWOP, PBCG, and PCPT-v2, using the ethnically and racially diverse SEPTA cohort.

Methods:

The SEPTA study was a multi-center trial that validated Stockholm3 in an ethnically and racially diverse population of men undergoing prostate biopsy in North America between 2019 and 2023. Among the SEPTA cohort, 2,115 (98%) with complete data for risk calculation were included in the analysis. The primary outcome measure was detection of Grade Group ≥2 (GG≥2) prostate cancer. The analysis included Stockholm3, age, free and total PSA, prostate volume, and prostate biopsy outcome. Risk was calculated for all biomarkers and risk calculators, and operating and performance characteristics were computed at clinically used thresholds. Receiver operating characteristic (ROC) analysis, graphical calibration estimation, and decision curve analysis were performed for all biomarkers and risk calculators.

Results:

2,115 men were included, 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 prior to biopsy.  There were 1,200 (57%) benign biopsies performed, 307 (15%) GG1 cancers detected, and 608 (29%) GG≥2 cancers detected.  The Stockholm3 test outperformed all evaluated biomarkers and risk calculators in the overall cohort, with an area under the curve (AUC) of 0.82 (95% CI: 0.80-0.84, p<0.001). The AUCs for the prediction tools were: 0.72 (95% CI: 0.70-0.74) for free/total PSA, 0.76 (95% CI: 0.74-0.78) for PSA density, 0.77 (95% CI: 0.75-0.79) for ERSPC-4/SWOP, 0.74 (95% CI: 0.72-0.76) for PBCG, and 0.78 (95% CI: 0.76-0.80) for PCPT-v2. PBCG was best calibrated, followed by Stockholm3, which showed an underprediction of GG≥2 prostate cancer. Decision curve analysis demonstrated superior performance of Stockholm3, showing the highest positive net benefit.

Conclusion:

Stockholm3 outperforms other commonly used biomarkers and risk calculators for detecting clinically significant prostate cancer in an ethnically and racially diverse population.

Funding: Vetenskapsrådet (Swedish Research Council, 2020-01609; Eklund, Grönberg), Cancerfonden (Swedish Cancer Society, CAN 180649; Eklund, Grönberg), and A3P Biomedical (H.T.V.)