Abstract
BACKGROUND: This study aimed to develop and externally validate nomograms for predicting prostate cancer (PCa) and clinically significant prostate cancer (csPCa) in patients with prostate-specific antigen (PSA) between 4 and 20 ng/mL. METHODS: Nomograms were developed using data from patients with PSA of 4–20 ng/mL who underwent prostate MRI and biopsy at our institution (n = 440). The outcomes were the presence of csPCa and PCa. Significant variables identified through univariate logistic analysis and LASSO regression analysis were used to construct four nomograms separately for lesions located in the peripheral and transitional zones. These nomograms were subsequently validated and evaluated using an external independent cohort of patients obtained from the Prostate Imaging: Cancer AI (PI-CAI) database (n = 313). RESULTS: Age, Prostate Imaging Reporting and Data System (PI-RADS) score, apparent diffusion coefficient (ADC) value, and PSA density (PSAD) were independent predictors in the prediction model for csPCa in the peripheral zone (PZ), showing an area under the curve (AUC) of 0.934 in the external validation cohort. For csPCa in the transitional zone (TZ), PI-RADS score, ADC value, and PSAD were independent predictors, with an AUC of 0.903. Additionally, PI-RADS score and ADC value were independent predictors for PCa in PZ, with an AUC of 0.882, while PI-RADS score and PSAD were independent predictors in TZ, with an AUC of 0.764. Calibration curves indicated good agreement, and decision curve analyses (DCAs) confirmed the clinical benefits of the nomograms. CONCLUSION: Our diagnostic nomograms are simple, feasible, and demonstrate strong performance in predicting csPCa and PCa. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s12894-026-02055-y.