Article Text

Download PDFPDF
Review: IgA endomyseal and transglutaminase antibodies had high specificity for diagnosis of coeliac disease
  1. Neil E Gibson, MD
  1. University of Alberta, St Albert, Alberta, Canada

    Statistics from

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

 Q What are the test properties of antigliadin antibody (AGA), endomyseal antibody (EMA), and transglutaminase antibody (tTG) and their subtypes for diagnosis of coeliac disease in children and adults?

    Clinical impact ratings Gastroenterology ★★★★★★☆


    Embedded ImageData sources:

    Medline (1966 to October 2003) and EMBASE/Excerpta Medica (1974–2003).

    Embedded ImageStudy selection and assessment:

    studies that allowed extraction of sensitivity and specificity data and had no fatal methodological flaws. Exclusion criteria: control group did not receive the gold standard (biopsy), biopsy criteria not described, patient disease status for the test under consideration was known, control groups included patients with Marsh I or II biopsy lesions, or AGA tests done without a commercial ELISA kit or before 1990.

    Embedded ImageOutcomes:

    sensitivity and specificity.


    55 studies met the selection criteria. The table summarises the sensitivity and specificity of each test by patient population.

    Weighted pooled sensitivity (sens) and specificity (spec) of tests for diagnosing celiac disease*


    Endomyseal and transglutaminase antibodies (IgA class) had pooled sensitivities ⩾90% and specificities ⩾97% for diagnosis of coeliac disease in children or adults. Antigliadin antibody (IgA class) had sensitivities of 75–95% and specificities of 80–95%.

    Author response

    Although we agree with Dr. Gibson’s calculation of likelihood ratios, we disagree with the statement that a confirmatory biopsy for positive serology is not required. The recommendation of a confirmatory biopsy is based on systematic review1 and is supported by major gastrointestinal societies.

    1 Rostom A, Dubé C, Cranney A, et al. Celiac disease. Evid Rep Technol Assess (Summ) 2004;(104):1–6.


    The results of the meta-analysis by Rostom et al confirm that antibody tests can be useful for diagnosis of coeliac disease. It appears that tTG and EMA are superior to AGA and that there is congruity between adults and children. This review raises questions about their utility for screening and as alternatives to endoscopy with biopsy.

    The positive and negative likelihood ratios are 231 and 0.08, respectively, for IgA EMA-HU, and 20 and 0.1 for IgA tTG-GP. If the latter is restricted to adults, the likelihood ratios are 18 and 0.15, respectively. These statistics can be helpful when using clinical suspicion to establish a diagnosis. If a patient presenting with a typical history has a 30% probability of coeliac disease, then a positive EMA gives a post-test probability of 98%, and a negative test yields a post-test probability of 3%. Both results are useful and may preclude confirmation with endoscopy. Exceptions might be patients who do not respond to dietary therapy or have IgA deficiency, thus rendering the serology doubtful.

    Endoscopy still has a role in patients with symptoms but borderline serology, as well as in patients in whom positive proof is needed to motivate dietary compliance. Some clinicians may be uncomfortable making a diagnosis based on serology rather than histology. This study, however, reassures clinicians that a serological diagnosis is robust.

    Unfortunately, because the individual studies were of known, finite populations, the results of the meta-analysis may not be generalisable to asymptomatic screening.

    In summary, IgA EMA-HU and IgA tTG-HP appear to be sensitive and specific tests that can be used to make a diagnosis of coeliac disease without resorting to endoscopy in the context of moderate clinical suspicion. The authors recommend confirmatory biopsy for positive tests based on consensus within the discipline; however, based on this study, this step may not be absolutely required for all patients. Further studies are also needed to assess their use to screen asymptomatic populations.

    View Abstract


    • For correspondence: Dr A Rostom, University of Ottawa, Ottawa, Ontario, Canada. arostom{at}

    • Source of funding: Agency for Health Research and Quality.