Prediction of celiac disease at endoscopy

Area of Science:

• Gastroenterology
• Immunology
• Epidemiology

Background:

• Celiac disease diagnosis and detection guidelines are often unclear.
• The prevalence and predictors of celiac disease require further investigation, particularly in diverse populations.

Purpose of the Study:

• To prospectively determine the prevalence and predictors of celiac disease in Lebanese patients undergoing esophagogastroduodenoscopy (EGD).
• To evaluate diagnostic strategies for celiac disease to improve detection rates and minimize unnecessary procedures.

Main Methods:

• A cross-sectional study of 999 patients undergoing EGD, involving questionnaires, serologic testing (tissue transglutaminase [tTG], deamidated gliadin peptide [DGP]), and duodenal biopsies.
• Multivariate logistic regression was used to identify independent predictors of celiac disease.
• Diagnostic strategies involving clinical suspicion, endoscopic findings, serology, and biopsy based on prediction models were compared.

Main Results:

• Celiac disease prevalence was 1.5%, with 1.8% showing villous atrophy (Marsh 3).
• Positive tTG was the strongest predictor (OR 131.7), followed by endoscopic villous atrophy (OR 64.8), eczema history, anemia, and Shiite ethnicity.
• A biopsy strategy based on prediction models achieved 93%-100% sensitivity for celiac disease with 22%-26% unnecessary biopsies, significantly outperforming strategies relying solely on clinical suspicion or excluding pre-EGD serology.

Conclusions:

• Standard clinical suspicion and endoscopic findings alone lead to a high miss rate for celiac disease.
• A prediction model incorporating easily obtainable patient information prior to or during EGD maximizes celiac disease diagnosis while minimizing unnecessary duodenal biopsies.