Diagnostic accuracy of multiparametric magnetic resonance imaging combined with clinical parameters in the detection of clinically significant prostate cancer: A novel diagnostic model.

Abstract

To evaluate the diagnostic accuracy of multiparametric magnetic resonance imaging in the detection of prostate cancer, according to Prostate Imaging Reporting and Data System, and the usefulness of combining clinical parameters to improve patients' risk assessment. Overall, 201 patients underwent multiparametric magnetic resonance imaging investigation with a 3-T magnet and a 32-channel body coil based on triplanar high-resolution T2-weighted, diffusion-weighted and T1-weighted dynamic contrast-enhanced imaging before, during and after intravenous administration of paramagnetic contrast agent. Random transrectal ultrasound-guided biopsy was carried out for all eligible patients. If a Prostate Imaging Reporting and Data System ≥3 lesion was present, a targeted biopsy with magnetic resonance imaging-transrectal ultrasound fusion-guided system was carried out. Sensitivity, specificity, positive predictive value and negative predictive value of Prostate Imaging Reporting and Data System ≥3 lesions for the detection of prostate cancer were 65.1%, 54.9%, 43.1% and 75.0% respectively, with an accuracy of 64.2% (55.1-72.7%). At uni- and multivariate analysis, age ≥70years and prostate-specific antigen density ≥0.15ng/mL/mL were significantly associated with prostate cancer. A new risk model named "modified Prostate Imaging Reporting and Data System" was created considering age and prostate-specific antigen density in addition to the Prostate Imaging Reporting and Data System score showing an improved correlation with prostate cancer compared with the Prostate Imaging Reporting and Data System alone (area under curve 71.4%, 95% confidence interval 62.2-80.5 vs area under curve 62.6%, 95% confidence interval 52.1-73; P≤0.0001). The accuracy of Prostate Imaging Reporting and Data System alone in the diagnosis of prostate cancer might be suboptimal, whereas a novel risk model based on the combination of multiparametric magnetic resonance imaging data with clinical parameters could offer higher discrimination and improve the ability of diagnosing clinically significant disease.