Development and validation of a clinical nomogram based on lasso-logistic regression for predicting prostate cancer with PSA 4-20.0 ng/mL: a retrospective study

Background Given the established diagnostic utility of the prostate health index (PHI) in prostate cancer (PCa), this study sought to incorporate PHI into a clinically applicable prediction model alongside conventional parameters, with the goal of refining biopsy selection in men presenting with PSA values of 4–20 ng/mL. Methods We retrospectively collected clinical data from patients undergoing prostate biopsy at tertiary medical centers in China. Candidate variables were screened using least absolute shrinkage and selection operator (LASSO) regression, and the selected predictors were incorporated into a multivariable logistic regression model, which was subsequently presented as a nomogram. Model performance was evaluated in both the training and validation cohorts in terms of discrimination, calibration, and clinical utility. Results A total of 314 patients were included, with 219 assigned to the training cohort and 95 to the validation cohort. LASSO regression identified prostate volume, blood glucose, low-density lipoprotein, triglycerides, urinary leukocyte count, hypertension, Prostate Imaging–Reporting and Data System score, platelet-to-lymphocyte ratio, albumin, the fPSA/tPSA ratio, and PHI as candidate variables. Multivariate analysis demonstrated that triglycerides, PI-RADS score, ALB, and PHI were independent predictors of PCa. The nomogram achieved good discriminatory performance, with an area under the receiver operating characteristic curve of 0.75 in the training cohort. Calibration curves and the Hosmer–Lemeshow test indicated good agreement between predicted and observed outcomes. Consistent performance was observed in the validation cohort. Conclusions Our findings suggest that integrating clinical parameters into a PHI-based model can enhance the stratification of prostate cancer risk, potentially reducing unnecessary biopsies and improving patient outcomes.