Staging accuracy of MRI of the prostate with special reference to the influence of the time of last ejaculation on the detection of seminal vesicle invasion

Abstract

To evaluate the Prostate Imaging Reporting and Data System, version 2.1 (PIRADS V2.1) criteria for seminal vesicle invasion (SVI) and examine whether the timing of last ejaculation influences the detection of SVI. The study population consisted of 68 patients (34 with SVI, 34 without SVI, matching groups by age and prostate volume) who underwent PIRADS V2.1-compliant multiparametric magnetic resonance imaging (MRI; 34 at 1.5 T, 34 at 3 T). Before the examination, the time of last ejaculation (38/68≤5 days, 30/68>5 days) was collected via a questionnaire. The five PIRADS V2.1 criteria for SVI with subsequent overall assessment were evaluated retrospectively by two independent examiners (examiner 1 with >10 years of experience, examiner 2 with 6 months of experience) in a single-blinded fashion for all patients using a questionnaire and a six-point scale (0=no, 1=very likely not, 2=probably not, 3=possible, 4=probable, 5=certain). E1 achieved high specificity (100%) and positive predictive value (PPV; 100%) in the overall assessment, independent of the time of last ejaculation (sensitivity=76.5%, negative predictive value [NPV]=81%). The area under the curve (AUC) value was 0.882; for E2, it was 0.765. At ≤5 days, the AUC values of E1 and E2 differed significantly (0.867 versus 0.681, p=0.016), as did the diffusion restriction criterion (0.833 versus 0.681, p=0.028). E1 showed high AUC values independent of time. E2 had better values for all criteria at >5 days than at ≤5 days. There were no significant differences between the examiners in all observations at >5 days. The PIRADS V2.1 criteria are well suited for an experienced examiner to detect SVI independent of time point. An inexperienced examiner will benefit from patients being abstinent >5 days prior to MRI.