Risk Stratification of Suspected Clinically Localized Prostate Cancer Using Multi-Indicator Functional Differentiation: A Predictive Model With fPSA/tPSA, PHI, PCA3, and MRI.

BACKGROUND

Accurate risk stratification is crucial for managing men with suspected clinically localized prostate cancer, particularly those with total prostate-specific antigen (tPSA) in the diagnostic grey zone (4-10 ng/mL). This study aimed to develop and validate a predictive model integrating multi-dimensional indicators to distinguish clinically insignificant prostate cancer from significant disease.

METHODS

This retrospective cohort study analysed 242 patients with suspected clinically localized prostate cancer who underwent biopsy from January 2020-December 2021. Patients were stratified into low-risk (n = 118) and high-risk (n = 124) groups based on biopsy pathology. Key biomarkers including free prostate-specific antigen (fPSA)/tPSA ratio, Prostate Health Index (PHI), and Prostate Cancer Antigen 3 (PCA3) score were measured before biopsy. Multiparametric magnetic resonance imaging (mp-MRI) parameters were also assessed.

RESULTS

The high-risk group had significantly lower percentage of free to total prostate‑specific antigen (%fPSA) and prostate volume, but higher PHI, PCA3 scores, and positive Prostate Imaging-Reporting and Data System (PI-RADS) findings (all p < 0.05). Multivariate analysis identified %fPSA, PHI, PCA3 score, prostate volume, PI-RADS score ≥ 4, and index lesion diameter as independent predictors. A nomogram incorporating these factors demonstrated excellent discrimination, with an area under the curve (AUC) of 0.885 in the development cohort, remaining robust upon 10-fold cross-validation (AUC = 0.871). Temporal validation in an independent cohort (n = 80) yielded an AUC of 0.863.

CONCLUSIONS

The developed nomogram, integrating initial fPSA/tPSA screening, PHI risk quantification, PCA3 molecular confirmation, and magnetic resonance imaging (MRI) features, provides an effective tool for personalized risk stratification, with direct comparative analyses confirming its advantage over single-modality approaches.