Abstract
This study aimed to develop and validate a diagnostic model for prostate cancer (PCa) by integrating magnetic resonance imaging (MRI) parameters with the immunohistochemical expression of p504s, CK5/6, and Ki-67. A total of 448 patients undergoing prostate needle biopsy were included and randomly allocated into training (70 %) and validation (30 %) cohorts. Clinical data, MRI findings, and biomarker expression levels were analyzed. Multivariate logistic regression identified independent predictors, which were used to construct a diagnostic nomogram. Compared to controls, PCa patients had significantly higher PSA levels, lower f-PSA/t-PSA ratios, a greater frequency of palpable nodules, higher CC/C ratios, lower ADC values, increased p504s and Ki-67 positivity, and reduced CK5/6 expression. Seven variables were ultimately identified as independent predictors for the model. The resulting nomogram demonstrated excellent discrimination, with an area under the curve (AUC) of 0.971 in the training set and 0.977 in the validation set. It significantly outperformed a model using clinical indicators alone. This combined MRI-biomarker model shows high diagnostic accuracy for PCa and could potentially aid clinical decision-making and reduce unnecessary biopsies. External validation is required prior to clinical application.