Diagnostic Performance of Magnetic Resonance Enterography Disease Activity Indices Compared with a Histological Reference Standard for Adult Terminal Ileal Crohn's Disease: Experience from the METRIC Trial

Abstract

Background and aims: The simplified magnetic resonance enterography [MRE] index of activity [sMARIA], London, and 'extended' London, scoring systems are widely used in Crohn's disease [CD] to assess disease activity, although validation studies have usually been single-centre, retrospective, and/or used few readers. Here, we evaluated these MRE indices within a prospective, multicentre, multireader, diagnostic accuracy trial.

Methods: A subset of participants [newly diagnosed or suspected of relapse] recruited to the METRIC trial with available terminal ileal [TI] biopsies was included. Using pre-specified thresholds, the sensitivity and specificity of sMARIA, London, and 'extended' London scores for active and severe [sMARIA] TI CD were calculated using different thresholds for the histological activity index [HAI].

Results: We studied 111 patients [median age 29 years, interquartile range 21-41, 75 newly diagnosed, 36 suspected relapse] from seven centres, of whom 22 had no active TI CD [HAI = 0], 39 mild [HAI = 1], 13 moderate [HAI = 2], and 37 severe CD activity [HAI = 3]. In total, 26 radiologists prospectively scored MRE datasets as per their usual clinical practice. Sensitivity and specificity for active disease [HAI >0] were 83% [95% confidence interval 74% to 90%] and 41% [23% to 61%] for sMARIA, 76% [67% to 84%] and 64% [43% to 80%] for the London score, and 81% [72% to 88%] and 41% [23% to 61%] for the 'extended' London score, respectively. The sMARIA had 84% [69-92%] sensitivity and 53% [41-64%] specificity for severe CD.

Conclusions: When tested at their proposed cut-offs in a real-world setting, sMARIA, London, and 'extended' London indices achieve high sensitivity for active TI disease against a histological reference standard, but specificity is low.

Keywords: Crohn’s disease; imaging; magnetic resonance enterography.

Figure 1.

Flow chart of the study population.

Figure 2.

Distribution of the activity scores against HAI values stratified by newly diagnosed and suspected relapse patients. [a] sMARIA, [b] London, and [c] ‘extended’ London score. ND, newly diagnosed; SR, suspected relapse; HAI, histological activity index.

Figure 3.

ROC curves for detecting active disease [HAI >0] for [a] sMARIA, [b] London, and [c] ‘extended’ London scores. Grey bars report 95% confidence intervals at the pre-specified thresholds. HAI, histological activity index; ROC, receiver operating characteristic.

Figure 4.

ROC curves for detecting active disease [HAI >0] stratified by newly diagnosed [grey] and suspected relapse [orange] patients for [a] sMARIA, [b] London, and [c] ‘extended’ London scores. Grey bars report 95% confidence intervals at the pre-specified thresholds. HAI, histological activity index; ROC, receiver operating characteristic.

Figure 5.

ROC curves for sMARIA for detecting severe disease stratified by newly diagnosed [grey] and suspected relapse [orange] patients. Grey bars report 95% confidence intervals at the pre-specified thresholds. ROC, receiver operating characteristic.