Magnetic resonance enterography to predict subsequent disabling Crohn's disease in newly diagnosed patients (METRIC-EF)-multivariable prediction model, multicentre diagnostic inception cohort.

objectivesMagnetic resonance enterography (MRE) is a first-line investigation to diagnose Crohn's disease (CD), but its role for prognostication is unknown. Accordingly, we assessed the predictive ability of prognostic models including MRE scores (MRE Global Score (MEGS), simplified MR Index of Activity (sMARIA), and Lémann index (LI)) against models using clinical predictors alone for the development of modified Beaugerie disabling CD (MBDD) within 5 years of diagnosis.

methodsThis was a multicentre, diagnostic inception cohort of patients with newly diagnosed CD across 9 UK hospitals, followed for 4 years or more. We censored development of MBDD ≤ 90 days from diagnosis, and used time-to-event models using Royston-Parmer flexible parametric models.

resultsWe included 194 patients, median age 29, IQR 22-44 years, 52% female. Within 5 years of diagnosis, 42% (81/194) developed MBDD. In univariable analysis, initial steroid requirement was associated with increased risk of developing MBDD (HR 2.11 (95% CI 1.36, 3.26). The baseline clinical model had 49% (39, 60) sensitivity and 66% (57, 74) specificity for predicting the top 40% of patients with the greatest risk of developing MBDD, and 86% (77, 92) sensitivity and 35% (27, 45) specificity for predicting the development of MBDD in patients with an absolute risk of ≥ 10%. There was no significant difference in sensitivity when the MEGS, sMARIA, or LI were added to the baseline clinical model.

conclusionsAddition of MRE scores at diagnosis to a multivariable model comprising clinical predictors did not improve prediction of MBDD within 5 years of diagnosis. KEY POINTS: Question Magnetic resonance enterography (MRE) is important for diagnosing and monitoring Crohn's disease (CD), but primary research evaluating its prognostic role is lacking. Findings Adding MRE findings at diagnosis to a multivariable model comprising clinical predictors did not improve the prediction of disabling CD within 5 years of diagnosis. Clinical relevance When tested in a prospective, multicentre trial, current MRE activity and damage scores at diagnosis did not reliably predict whether patients would subsequently develop disabling CD. Notwithstanding this finding, MRE remains an essential tool for diagnosis and monitoring.