Combining Prostate Imaging Reporting and Data System, Histogram Analysis, and Prostate-Specific Antigen Density to Determine the Risk of Prostate Cancer in Patients With Prostate-Specific Antigen of 4 to 20 ng/mL.

Abstract

To develop regression models using Prostate Imaging Reporting and Data System (PI-RADS), histogram analysis, and prostate-specific antigen density (PSAD) to predict prostate cancer (PCa) and clinically significant PCa (CSPCa) in patients with prostate-specific antigen of 4 to 20 ng/mL. In total, 195 PCa and 386 noncancer patients with prostate-specific antigen of 4 to 20 ng/mL were divided into development and validation cohorts. Magnetic resonance imaging results of them were assessed by PI-RADS scores and histogram analysis-corrected PI-RADS (PI-RADSh) scores. Diagnostic efficiencies for PCa and CSPCa of these scores plus PSAD were evaluated with logistic regression and receiver operating characteristic curve analysis. Prostate-specific antigen density + PI-RADSh score showed significantly higher area under the receiver operating characteristic curve for PCa (0.956) and CSPCa (0.960), which were higher than PI-RADS (0.909 and 0.926), PI-RADSh (0.921 and 0.940), and PSAD + PI-RADS (0.943 and 0.949) (all P < 0.05). Incorporation of PSAD and histogram analysis raised the diagnosis efficiencies of PI-RADS for PCa and CSPCa.