Review of the accuracy of multi-parametric MRI prostate in detecting prostate cancer within a local reporting service.

Abstract

Multi-parametric magnetic resonance imaging of the prostate is crucial in detecting prostate cancer (CaP) and staging local disease. The Prostate Imaging Reporting and Data System (PIRADS) scoring system is used to assess and classify lesions and enables communication between clinicians and radiologists. This study aimed to assess the accuracy of PIRADSv2 in detecting CaP using histopathology specimens within our local service. This retrospective study included 192 patients between September 2016 and May 2019. All had mpMRI prostate examinations prior to biopsy or prostatectomy. Lesions on MRI were assigned a PIRADS score and comparison made with histopathology results. Gleason score ≥7 was considered as clinically significant prostate cancer (csCaP). We calculated accuracy, sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for detecting all CaP and csCaP. In the PIRADS 3 group, 32% were Gleason 6 and 32% were Gleason 7 lesions. In the PIRADS 4 group, 37% were Gleason 6 and 41% were Gleason ≥7. For PIRADS 5 lesions, 32% were Gleason 6 and 68% were Gleason ≥7. For all CaP, sensitivity was 84.7%, specificity 54.6%, PPV 82.3% and NPV 58.8%. For csCaP Gleason ≥7, PIRADS cut-off ≥3 had sensitivity, specificity, PPV and NPV of 95.7%, 39.3%, 47.5% and 94.1%, respectively, and cut-off ≥4 had sensitivity, specificity, PPV and NPV of 84.3%, 53.3%, 50.9% and 85.5%. This study confirms PIRADS has high accuracy, sensitivity and NPV for detecting all CaP and csCaP. A high NPV may obviate need for biopsy in low-risk patients.