Variation in Endoscopic Activity Assessment and Endoscopy Score Validation in Adults with Eosinophilic Esophagitis

Abstract

Background & Aims Eosinophilic esophagitis (EoE) is assessed endoscopically (endoscopic activity), based on grades of edema, rings, exudates, furrows, and strictures (EREFS). We examined variations in endoscopic assessments of severity, developed and validated 3 EREFS-based scoring systems, and assessed responsiveness of these systems using data from a randomized placebo-controlled trial of patients with EoE. Methods For the development set, 5 gastroenterologists reviewed EREFS findings from 266 adults with EoE and provided endoscopist global assessment scores (EndoGA, scale of 0 to 10); variation (ΔEndoGA) was assessed using linear regression. We evaluated simple scores (features given arbitrary values from 0 to 3) and developed 2 scoring systems (adjusted score range, 0–100). We then fitted our linear regression model with mean EndoGA to data from 146 adults recruited in centers in Switzerland and the United States between April 2011 and December 2012. For the validation set, we collected data from 120 separate adults (recruited in centers in Switzerland and the United States between May 2013 and July 2014), assessing regression coefficient-based scores using Bland-Altman method. We assessed the responsiveness of our scoring systems using data from a randomized trial of patients with EoE given fluticasone (n=16) or placebo (n=8). Results The distribution of EndoGA values differed among endoscopists (mean ΔEndoGA, 2.6±1.8; range 0–6.6). We developed 2 regression-based scoring systems to assess overall and proximal and distal esophageal findings; variation in endoscopic features accounted for more than 90% of the mean EndoGA variation. In the validation group, differences between mean EndoGA and regression-based scores were small (ranging from –4.70 to 2.03), indicating good agreement. In analyses of data from the randomized trial, the baseline to end of study change in patients given fluticasone was a reduction of 24.3 in simple score (reduction of 4.6 in patients given placebo, P=.052); a reduction of 23.5 in regression-based overall score (reduction of 6.56 in patients given placebo, P=.12), and a reduction of 23.8 (reduction of 8.44 in patients given placebo, P=.11). Conclusion Assessments of endoscopic activity in patients with EoE vary among endoscopists. In an analysis of data from a randomized controlled trial, we found that newly developed scoring systems are no better than simple scoring system in detecting changes in endoscopic activity. These results support the use of a simple scoring system in evaluation of endoscopic activity in patients with EoE.