MRI Quality and Reader Experience in Organized Prostate Cancer Screening: Insights from the PROBASE Trial

BACKGROUND: The diagnostic performance of prostate magnetic resonance imaging (MRI) critically depends on image quality and reader expertise. The PROBASE trial is a prospective population- and prostate-specific antigen (PSA)-based prostate cancer (PCa) screening study enrolling men aged 45 yr. OBJECTIVE: In this predefined substudy, we evaluated the impact of MRI quality and expert reference reading on clinically significant PCa (csPCa; International Society of Urological Pathology grade ≥2) detection. DESIGN, SETTING, AND PARTICIPANTS: This analysis included 516 participants who underwent multiparametric MRI and combined MRI-targeted biopsy and systematic biopsy after screening with PSA ≥3 ng/ml. Local Prostate Imaging Reporting and Data System (PI-RADS) scores were compared with reference readings by experienced uroradiologists. MRI quality was assessed using Prostate Imaging Quality (version 1; PI-QUAL v1). OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: Detection rates, positive predictive values (PPV), and negative predictive values (NPV) were calculated at an underlying biopsy threshold of PI-RADS score of ≥4. Image quality was compared between the study sites. RESULTS AND LIMITATIONS: Reference reading yielded higher NPV (92% vs 86%), PPV (57% vs 50%), and sensitivity (83% vs 69%) than local reading. Local reading missed substantially more csPCa in PI-RADS 1-2 (17.1% vs 2.9%) and resulted in more false-positive MRI findings in PI-RADS 4-5 (n = 76 vs n = 59) than reference reading. Lower MRI quality (PI-QUAL 1-3/5) was associated with reduced csPCa detection (46% vs 62%) and true negative rate (84% vs 95%). PI-QUAL differed between the study sites. PI-RADS classification of reference reading missed small or diffusely infiltrative csPCa, often in lower quality scans. Limitations include using local interpretations to define biopsy targets, limiting retrospective comparison, and possibly underestimating reference-reader performance. CONCLUSIONS: Expert reference MRI reading substantially improves csPCa detection and can reduce unnecessary biopsies in a screening setting. High-quality imaging, standardized protocols, and experienced reading should be considered essential components of future MRI-based screening strategies. TRIAL REGISTRATION: The trial is registered with the ISRCTN (International Randomized Controlled Trial Number) registry, registration number ISRCTN37591328, and can be accessed at https://www.isrctn.com/ISRCTN37591328. The study protocol can be accessed at doi: 10.1016/j.eururo.2013.05.022.