Utility of Testing Patients, on Presentation, for Serologic Features of Celiac Disease

Abstract

Celiac disease shares features of other disorders. It can be diagnosed conclusively only based on duodenal histology analysis, which is not practical for screening purposes. Serologic analysis might be used to identify candidates for biopsy analysis. We aimed to develop a simple diagnostic approach that all clinicians could follow to increase the percentage of patients accurately diagnosed with celiac disease at initial presentation. We performed a retrospective analysis of data from 752 patients (88 with celiac disease, none were IgA deficient) who attended a UK district general hospital from January 2007 through December 2008 and underwent biopsy analysis and serologic tests to measure endomyseal antibodies and IgA antibodies against tissue transglutaminase (tTG). Patients avoiding gluten in their diet were excluded. Patients were assigned to 1 of 4 groups: high-risk (based on presence of anemia, chronic diarrhea, unintentional weight loss, or dermatitis herpetiformis), low-risk (based on such factors as dyspepsia, abnormal liver function, ataxia, or chronic cough), nutrient deficiency (based on levels of iron, vitamins B12 and D, or folate), or screening (because they had type 1 diabetes or a family history of celiac disease). Patients with celiac disease were identified using the modified Marsh criteria (grades 1-3) for interpreting duodenal histology. We compared clinical category, serology profiles, and biopsy results between patients with and without celiac disease. Celiac disease was diagnosed in 64 of 565 patients in the high-risk group (11%), 14 of 156 patients in the low-risk group (9%; P = .47 compared with high-risk group), 7 of 28 patients in the nutrient-deficiency group, and 3 of 3 patients in the screening group. Among 71 patients who tested positive for both antibodies (tTG and endomyseal antibodies), the positive predictive value for celiac disease was 97%; a negative test result for tTG had a negative predictive value of 98%. Among 708 patients with normal-looking biopsy samples, only 62 had celiac disease (9%). Among 44 patients with abnormal biopsy samples, 26 had celiac disease (59%). Based on a retrospective analysis, patients with and without celiac disease cannot be distinguished based on clinical features. Patients who present with symptoms of celiac disease should be tested for tTG, to identify candidates for duodenal biopsy analysis.