Discrepancies between histology and serology for the diagnosis of coeliac disease (1)

Editor – Sweis and colleagues showed discrepancies between histology and serology in the diagnosis of coeliac disease (CD) (Clin Med August 2009 pp 346–8), and suggest we reduce our reliance on serology testing in diagnosing and excluding CD. However, we feel there are major reasons to reconsider this.

The numbers reported here must be interpreted carefully: 10 out of 26 CD patients who received serologic testing were seronegative. This 38.5% occurrence of seronegative CD is misleading. In the spirit of Bayes theorem, the more common the condition we are testing, the greater the percentage of false negative results.¹ In this case, all 26 patients were selected due to the diagnosis of CD, meaning the prevalence in this group was already 100%. Therefore, this group is bound to have a high number of false negative tests. The authors correctly state that a small number of cases of CD will be missed by relying on serology alone, but the true prevalence is unknown, and this number is likely to be much lower than 38.5%.

In addition, the predictive value of using an ELISA-based method to detect tissue transglutaminase autoantibody (tTG) remains open to discussion. There are currently numerous tTG assays available, all with varying performances. The International tTG Workshop for CD performed head-to-head comparisons of various commercial and laboratory-based tTG assays. For this workshop, assays reported sensitivities ranging from 82% to 93%, underscoring the marked variability in assay performance.² Given these findings, the lack of positive serology in a proportion of their biopsy-proven coeliacs could be assay dependent.

Finally, even though intestinal biopsy is the gold standard method to diagnose CD, it is not without its short comings. The sensitivity of histology is largely dependent on the site and number of biopsy samples taken.^3,4 Negative histology often excludes a diagnosis of CD. However, a proportion of these patients have CD-like gastrointestinal symptoms, which might be attributed to the subtle changes seen in microscopic enteritis that could go undetected.⁵

In all, we agree that it is important not to rely on serology alone for the diagnosis of CD, but to allow serology to increase or decrease your estimation of risk of disease. However, considering the lifelong implications of a diagnosis of CD, one should still maintain a degree of suspicion and also take great care in interpreting villous atrophy in the absence of autoantibodies in any patient.

• © 2010 Royal College of Physicians