Abstract
This study was aimed to evaluate the frequency of false-positive and false-negative results in multiparametric prostate magnetic resonance imaging (mpMRI), to investigate the potential contributing factors, and to analyze the clinical variables that may affect the diagnostic accuracy of mpMRI. This retrospective study included 1269 patients who underwent mpMRI between 2018 and 2024, followed by prostate biopsy within 3 months. Patients with Prostate Imaging-Reporting and Data System (PI-RADS) 1 to 3 lesions but positive biopsy results were classified as "false-negative," whereas those with PI-RADS 4 to 5 lesions and negative biopsy results were considered "false-positive." mpMRI findings were compared with clinical parameters such as prostate-specific antigen (PSA), PSA density, and age. Intra-observer consistency was assessed using a 2nd mpMRI reading performed by the same experienced radiologist. mpMRI may yield false results, particularly in cases involving anterior/apical lesions or suboptimal image quality. Although the number of false-negative cases was low (n = 21), the possibility of missing clinically significant prostate cancers (ISUP Grade Group ≥ 2, Gleason ≥ 3 + 4) remains an important concern due to the potential for delayed diagnosis and management. reevaluation by experienced radiologists significantly may enhance diagnostic precision. Incorporating PSA density and standardized imaging quality criteria into routine practice may improve the overall diagnostic performance of mpMRI. A total of 203 patients (15.99%) were identified as having discordant results: 1.65% (n = 21) were false-negatives, and 14.34% (n = 182) were false-positives. Among the 182 index false-positive cases (defined as PI-RADS 4-5 with a negative biopsy within 3 months), 78 patients had visible lesions on mpMRI. During limited follow-up, 7 of these were later diagnosed with prostate cancer on repeat biopsy, including 4 with clinically significant disease. As repeat biopsies were not routinely performed, the number of false-positive cases at index evaluation remained 182. PSA density, advanced age, and lesion location (particularly the anterior and apical zones) were identified as potential contributors to false results. Intra-observer agreement for mpMRI interpretation was low, with a Cohen kappa coefficient of 0.099, indicating significant discordance.