P131 The simplified MARIA score is strongly correlated with the MARIA and Clermont scores to analyse the luminal activity of Crohn’s disease (CD) in magnetic resonance and is easier to calculate

Abstract

Abstract Background Two scores have shown their interest in evaluating the luminal activity of CD in magnetic resonance (MR), the MARIA and Clermont scores, but their calculation can be complicated in clinical practice. Recently, the Spanish authors validated a new simplified MARIA score (MARIAs) and retrospectively showed a very good correlation with the initial MARIA score (1). The purpose of this study was to validate this correlation in an independent cohort and to compare its feasibility with other scores. Methods This was a retrospective analysis of all MR performed in our department to evaluate the luminal activity of CD. Two independent radiologists each calculated the MARIA, Clermont and MARIAs scores (wall thickness > 3mm, oedema, comb sign, ulcerations), each blind to the other and blind to clinical activity. In case of disagreement between the two radiologists, a third reader analyzed MR activity. Correlations were calculated between these three tests for the entire cohort. In addition, concordance tests were performed for predetermined thresholds of 7 to 11 for the MARIA score, 8.5 to 12.5 for the Clermont score and 1 to 2 for the MARIAs score. The duration of the analysis was reported for each reading. Isolated colonic CD were excluded from the study. Results One hundred and twenty-one CD were included (65 in clinical and biological activity, 33% ileal phenotype, mean age: 38.5 years, sex ratio M/F = 54.5%). The agreement between the MARIA score between 7 and 11, the Clermont score between 8.5 and 12.5 and the MARIAs score between 1 and 2 was 0.92 (Fleiss kappa test). The correlation between the MARIA score and its simplified score was very strong (r = 0.93; 95% CI: 0.9–0.95) as well as between the Clermont score and the MARIAs score (r = 0.92 95% CI: 0.89–0.95). Inter-observer agreement was significantly higher in the calculation of the MARIAs score (96%) compared with 80% for the MARIA and Clermont scores (p = 0.04). Reading time was significantly faster for the MARIAs score (3.45 ± 0.4 min) compared with the MARIA (10.05 ± 1.2 min) and Clermont (13.5 ± −3.5 min) scores (p < 0.01). Positive predictive values for deep remission (clinical remission and calprotectin < 250 µg/g stool) were 95% at thresholds <1, <7 and <8.5 for the MARIAs, MARIA and Clermont scores, respectively. Conclusion In this independent cohort, we confirm the very strong correlation between the simplified MARIA score and the two reference scores (MARIA and Clermont) in the analysis of MC luminal activity as well as its rapid calculation and strong inter-observer agreement.