P0364 Establishment and Validation of a Central Reader Pool for Magnetic Resonance Enterography (MRE) for Use in Crohn’s Disease Clinical Trials

Abstract Background Clinical trials in Crohn’s disease (CD) currently rely on ileocolonoscopy to assess disease activity and monitor treatment response. There is growing interest in utilising non-invasive tests for this purpose. Magnetic resonance enterography (MRE) is non-invasive, more tolerable for patients, and allows assessment of all bowel layers and surrounding tissues. Validated systems for scoring CD severity on MRE enable standardised reporting and include Magnetic Resonance Index of Activity (MaRIA), a simplified version (MaRIAs) and the Lemann Index (LI). However, to be utilized in clinical trials, inter-reader reliability needs to be established. Here, we report the use of a standardised protocol to create a central reader pool for MRE scoring in CD and report its validation by application to the PROFILE1 clinical trial cohort. Methods Five accredited gastrointestinal radiologists blindly reported a total of 18 scans across 2 rounds of reading, with a consensus meeting between the rounds (Table 1). Each scan was scored using MaRIAs, MaRIAs-E (modified MaRIAs which includes small bowel segments), and LI. Intraclass correlation coefficient (ICC) was calculated to assess inter-reader reliability. To validate this reliability assessment in a clinical trial cohort, 71 scans from the PROFILE trial were double-read and ICC calculated. Results The central reader pool showed good inter-reader reliability after two rounds of training (ICC>0.70, with 95%CI lower bound>0.5). This high level of reliability was maintained in the trial validation (Figure 1). Conclusion Using this protocol we have successfully validated a central reader pool dedicated to severity scoring of CD that can be utilised in clinical trial setting. The ICC metrics in training rounds reflect good agreement between readers, suggesting that any single reader chosen from the pool would return a similar result to any other reader. The similarly high ICC calculated from the trial validation set suggests that the protocol used for establishing the central reader pool was robust, and high reliability of readers persisted during the PROFILE trial. Importantly, this methodology is transferable to other modalities where expert readers are asked to score images against validated scoring systems. References Nurulamin M. Noor et al., “A Biomarker-Stratified Comparison of Top-down versus Accelerated Step-up Treatment Strategies for Patients with Newly Diagnosed Crohn’s Disease (PROFILE): A Multicentre, Open-Label Randomised Controlled Trial,” The Lancet Gastroenterology & Hepatology 9, no. 5 (May 1, 2024): 415–27