Defining Thresholds of Antibody Levels Improves Diagnosis of Celiac Disease

BACKGROUND & AIMS: The European Society for Pediatric Gastroenterology and Nutrition proposed guidelines for the diagnosis of celiac disease, stating that duodenal biopsy is no longer needed if patients have symptoms and levels of immunoglobulin A anti-tissue transglutaminase (IgA anti-tTG) more than 10-fold the cut-off value. We evaluated the accuracy of this guideline in a well-characterized population using different commercial assays. METHODS: We analyzed levels of IgA anti-tTG in serum samples from 104 consecutive pediatric and adult patients who were not deficient in IgA and were diagnosed with celiac disease from August 1, 2000, to December 31, 2009. We also analyzed serum samples from 537 consecutive patients without celiac disease (controls), collected from May 1, 2004, to October 12, 2006, who underwent intestinal biopsy analysis. Serum levels of antibodies were quantified using assays from Bio-Rad, INOVA, Genesis, and Thermo Fisher. RESULTS: The likelihood ratio (probability of a specific result in patients divided by the probability of the same result in controls) for celiac disease increased with levels of IgA anti-tTG in all assays. Depending on the assay, the likelihood ratio for levels greater than 10-fold the cut-off value ranged from 111 to 294. The percentage of patients with celiac disease with levels of IgA anti-tTG greater than 10-fold the cut-off value ranged from 41% to 61%, depending on the assay. For levels of anti-tTG greater than 10-fold the cut-off value, the post-test probabilities for celiac disease (probability of disease, based on pretest probability and test result) were, depending on the assay, 89%-96% and 53%-75% for pretest probabilities (probability of disease depending on symptoms) of 7% and 1%, respectively. CONCLUSIONS: To diagnose celiac disease based on serologic factors, it might be best to define thresholds for levels of IgA anti-tTG based on a predefined likelihood ratio or post-test probability, instead of a multiple of a cut-off value. Patients with a high pretest probability and levels of anti-tTG greater than 10-fold the cut-off value have a high probability for having celiac disease, aiding clinical decision making.