Inflammatory indices beyond PSA density for prostate cancer detection in men with PSA 4–10 ng/mL

Abstract BACKGROUND Whether routinely available systemic inflammatory indices provide incremental diagnostic information beyond prostate-specific antigen density (PSAD) in men with prostate-specific antigen (PSA) concentrations of 4–10 ng/mL remains uncertain. METHODS We retrospectively pooled 237 men undergoing initial prostate biopsy at two hospitals, including 110 with biopsy-detected prostate cancer and 127 with benign prostatic hyperplasia. Logistic regression models included PSAD alone (Model 0), PSAD plus ln(PIV) (Model 1), or PSAD plus ln(SIRI) (Model 2). We compared discrimination, model fit, prediction error, calibration, exploratory operating points, and internal validation using 1000 bootstrap resamples. RESULTS AUROCs were 0.835, 0.848, and 0.846 for Models 0, 1, and 2, respectively. The increases of 0.013 and 0.011 were not statistically significant in paired DeLong tests (P = 0.244 and P = 0.357). Both extended models improved fit and reduced apparent Brier scores. At 90.0% sensitivity, specificity was 43.3%, 54.3%, and 52.8%; the corresponding specificity increases were 11.0 percentage points (95% bootstrap interval − 10.9 to 27.7) and 9.4 percentage points (− 13.7 to 25.6). Optimism-corrected AUROCs were 0.834, 0.843, and 0.842. CONCLUSIONS PIV and SIRI improved model fit in this selected cohort but did not establish reliable incremental discrimination or clinical utility beyond PSAD. External evaluation in representative biopsy populations is required before either index is used to inform biopsy decisions.