Construction and verification of a prostate cancer risk prediction model based on traditional screening methods

Abstract

Abstract Background Timely and accurate diagnosis of prostate cancer (PCa) is of paramount importance in guiding treatment and reducing the suffering and death of patients. This study aimed to construct a risk prediction model for PCa based on prostate-specific antigen (PSA) levels, digital rectal examination (DRE), and transrectal ultrasonography (TRUS) to develop a screening tool with better clinical performance. Methods We retrospectively analysed 1593 patients who underwent transrectal ultrasound-guided biopsy (TRUSB) between June 2000 and February 2023. Patients were randomly divided into a training set of 1115 cases (70%) and a validation set of 478 patients (30%). A PCa risk prediction model was established using the R software. The performance of the model was examined based on calibration curves, receiver operating characteristic (ROC) curves, decision curve analysis (DCA), and clinical impact curves (CIC). Results Serum PSA levels, DRE results, prostatic border, shape, hypoechoic area, and seminal vesicle condition were associated with pathological outcomes. The areas under the (ROC) curves of the training and verification sets were 0.885 and 0.879, respectively. The optimal cut-off value was 0.477. The calibration curves indicated good calibration, and the DCA and CIC results demonstrated good clinical practicality. Subsequently, we developed an online calculator (https://jiwentong0.shinyapps.io/dynnomapp/) with six variables to screen high-risk patients. Conclusions This study incorporated the results of three traditional screening methods to establish a highly accurate model for predicting PCa before biopsy. With this model, we aim to provide a non-invasive and cost-effective tool for PCa screening.