Intestinal Biopsy is not Always Required to Diagnose Celiac Disease

A Retrospective Analysis of Combined Antibody Tests

Annemarie Bürgin-Wolff; Buser Mauro; Hadziselimovic Faruk


BMC Gastroenterol. 2013;13(19) 

In This Article



Included in this retrospective study were serum samples from 268 patients on a gluten-containing diet. The samples were collected in hospitals or medical services throughout Switzerland, Germany and Austria. The sera were then sent to the Institute for Coeliac Disease in Liestal, Switzerland, where the antibody determinations were performed without any knowledge of each patient's clinical condition. All patients from whom we received a jejunal histology report and clinical data were included in the study. At the beginning of our studies, all patients with symptoms suggestive of CD underwent small intestinal biopsy; antibody determinations were performed at the same time. This diagnostic procedure gradually changed with time as serological tests gained increasing importance in diagnosis, and for an undefined period patients were sometimes selected for biopsy when IgA anti tTG or EMA were positive. The patients suffered from gastrointestinal symptoms such as diarrhea, constipation, poor weight gain, chronic vomiting, abdominal pain, flatulence, and failure to thrive; or disorders such as unexplained weight loss in adults, iron-deficiency anemia, lassitude, psychiatric disorders, short stature, and diabetes type 1. IgA-deficient CD patients were excluded. Serum for antibody determinations was obtained within 2 months before or one month after endoscopic intervention.

Sample Analysis

All samples were analyzed for antibodies against tTG, ngli, and dpgli by fully automated fluoroenzyme immunoassay tests (Elia Celikey IgA, Elia Gliadin IgA, Elia Gliadin IgG, Elia Gliadin DP IgA and Elia Gliadin DP IgG; Phadia [now Thermo Fisher Scientific], Freiburg, Germany) performed on the Phadia 100 instrument in accordance with the manufacturer's instructions. With the help of ROC curves, we calculated the optimal cut-off values for our sample. For all analyzes except IgA anti-dgpli, we found the best sensitivity and specificity to be consistent with the recommendations of the respective manufacturer. For IgA anti- dpgli. we found that using a cut off value of 7 instead of 10 resulted in somewhat higher sensitivity (78% instead of 71%) while maintaining the same specificity.

For each antigen, we determined the best cut-off value within our sample with respect to the sum of false-positive and false-negative results: IgA anti tTG = 7, IgA anti-ngli= 7, IgG anti-ngli = 7, IgA anti-dpgli = 7, and IgG anti-dpgli = 10. EMA was analyzed by indirect immunofluorescence on monkey esophagus sections: cut-off = serum dilution: 1:5.

Statistical Analysis

In addition to the usual descriptors for diagnostic tests, such as sensitivity, specificity, and positive and negative predictive values, the quantity "efficiency" was used: the efficiency of a diagnostic test is its percentage of correct outcomes. Because predictive values are dependent on the prevalence of the disease in the study population, we also calculated the positive and negative likelihood ratios for CD, which are independent of the pretest probability of disease. The likelihood ratio is the ratio of the probability that a patient with the disease has a particular test result divided by the probability that a patient without the disease has the same test result. Contingency tables were evaluated using "Fisher's Exact Test". To test whether the replacement or addition of a diagnostic test improves the outcome to a statistically significant degree, McNemar's test for significant changes was applied. Both tests were used in their precise form (not only asymptotic) as available in the software package StatXact version 6.3.0. (Cytel Software Corporation Cambridge, MA, USA).