Development and validation of multiparametric models based on computed tomography enterography to determine endoscopic activity and surgical risk in patients with Crohn's disease: A multi-center study

Abstract

ObjectiveTo develop novel multiparametric models based on computed tomography enterography (CTE) scores to identify endoscopic activity and surgical risk in patients with Crohn's disease (CD). MethodsWe analyzed 171 patients from 3 hospitals. Correlations between CTE outcomes and endoscopic scores were assessed using Spearman's rank correlation analysis. Predictive models for moderate to severe CD were developed, and receiver operating characteristic (ROC) curves were constructed to determine the area under the ROC curve (AUC). A combined nomogram based on CTE scores and clinical variables was also developed for predicting moderate to severe CD and surgery. ResultsCTE scores were significantly correlated with endoscopy scores at the segment level. The global CTE score was an independent predictor of severe (HR = 1.231, 95% CI: 1.048–1.446, p = 0.012) and moderate-to-severe Simplified Endoscopic Scores for Crohn's Disease (SES-CD) (HR = 1.202, 95% CI: 1.090–1.325, p < 0.001). The nomogram integrating CTE and clinical data predicted moderate to severe SES-CD and severe SES-CD scores in the validation cohort with AUCs of 0.837 and 0.807, respectively. The CTE score (HR = 1.18; 95% CI: 1.103–1.262; p = 0.001) and SES-CD score (HR = 3.125, 95% CI: 1.542–6.33; p = 0.001) were independent prognostic factors for surgery-free survival. A prognostic nomogram incorporating CTE scores, SES-CD and C-reactive protein (CRP) accurately predicted the risk of surgery in patients with CD. ConclusionThe newly developed CTE score and multiparametric models displayed high accuracy in predicting moderate to severe CD and surgical risk for CD patients.