Value of clinical parameters and MRI with PI-RADSV2 in predicting seminal vesicle invasion of prostate cancer

Abstract

Objective To investigate the usefulness of magnetic resonance imaging (MRI) with Prostate Imaging Reporting and Data System version 2 (PI-RADSV2) and clinical parameters in predicting seminal vesicle invasion (SVI). Material and methods In this retrospective study, we identified 569 prostate cancer patients who underwent radical prostatectomy with MRI before surgery. SVI was interpreted with PI-RADSV2. Clinical parameters such as the prostate-specific antigen (PSA) and Gleason score (GS) were analyzed for the prediction of SVI. Logistic regression models and receiver operating characteristic (ROC) curves were used to evaluate SVI based on clinical parameters and MRI with PI-RADSV2. Results The median age at presentation was 67 years (43–85 years). The median PSA level was 6.1 ng/mL (2.2–72.8 ng/mL). There were 113 patients with a biopsy GS of ≥ 8. A total of 34 patients (6.0%) were interpreted to have SVI by MRI of which 20 were true positive, and 52 patients (9.1%) had true SVI in the final pathologic analysis. In multivariable analysis, PSA (HR: 1.03, 95% CI: 1.00–1.07), biopsy GS ≥ 8 (HR: 4.14, 95% CI: 2.12–8.09), and MRI with PI-RADSV2 (HR: 14.67, 95% CI: 6.34–33.93) were significantly associated with pathologic SVI. The area under the curve of the model based on the clinical parameters PSA and GS plus MRI (0.862) was significantly larger than that of the model based on clinical parameters alone (0.777, p < 0.001). Conclusions MRI with PI-RADSV2 using the clinical parameters PSA and GS was effective in predicting SVI.