The following is a summary of “Prediction of false-positive PI-RADS 5 lesions on prostate multiparametric MRI: development and internal validation of clinical-radiological characteristics based nomogram,” published in the April 2024 issue of Urology by Cheng et al.

This study endeavors to construct a comprehensive risk model incorporating clinical and radiological parameters to accurately predict false-positive lesions classified as The Prostate Imaging Reporting and Data System (PI-RADS) 5. Leveraging data from 612 biopsy-naïve patients who underwent multiparametric magnetic resonance imaging (mpMRI) before prostate biopsy, a meticulous amalgamation of clinical variables and radiological features extracted from mpMRI scans was meticulously curated. Subsequently, lesions were stratified into distinct training and validation cohorts through random allocation. Utilizing stepwise multivariate logistic regression analysis with backward elimination, variables demonstrating significant disparities were identified. 

A diagnostic nomogram was then meticulously formulated within the training cohort and subjected to rigorous validation within the independent validation cohort. Calibration curve assessment and receiver operating characteristic (ROC) analysis were further employed to evaluate the model’s performance. Notably, among the 296 PI-RADS 5 lesions delineated in 294 patients, 132 and 56 lesions were corroborated as clinically significant prostate cancer within the training and validation cohorts, respectively. The resultant diagnostic nomogram, predicated on prostate-specific antigen density, maximum lesion diameter, zonality of the lesion, apparent diffusion coefficient minimum value, and apparent diffusion coefficient minimum value ratio, exhibited commendable discrimination with a C-index of 0.821 in the training cohort and 0.871 in the validation cohort. 

Furthermore, the calibration curve underscored robust agreement between estimated and observed outcomes across both cohorts. Upon determining the optimal cutoff values via ROC analysis, the nomogram showcased promising diagnostic performance metrics, with a sensitivity of 90.6%, specificity of 67.9%, positive predictive value (PPV) of 61.7%, and negative predictive value (NPV) of 92.7% within the validation cohort. Ultimately, the developed nomogram holds substantial promise in effectively discerning false-positive PI-RADS 5 lesions from clinically significant ones, thereby potentially obviating the need for unnecessary prostate biopsies.

Source: bmcurol.biomedcentral.com/articles/10.1186/s12894-024-01465-0