Reply

Abstract

We appreciate the comments of Dr Bürgin–Wolff and Prof Hadziselimovic. They raise 2 issues: firstly they demonstrate that the inclusion of immunoglobulin A (IgA) anti-gliadin antibodies (IgA-AGA) and immunoglobulin G (IgG) anti-gliadin antibodies (IgG-AGA) could be helpful in diagnosing celiac disease (CD) if combined with IgA antiendomysial antibodies (IgA-EMA), and IgA anti-tissue transglutaminase (IgA-tTG) antibody results in a high-risk group. Secondly, they suggest the possibility of obviating the need for a duodenal biopsy in the high-risk group due to the high positive predictive value of their proposed decision tool. We have examined our data and similarly used all 4 antibodies to test in our low-risk group. We found all 4 antibodies were positive in 16/77 with CD and in 2/1923 with non CD. This gave the test a specificity of 99.9%, a positive predictive value (PPV) of 88.8%, and sensitivity of 21%. If we required all 4 antibodies to be negative, as an exclusion test for CD, then we found this to have a negative predictive value (NPV) of 99.6%. Current immunological testing using IgA-EMA and tTG antibodies for CD are regularly reported as achieving a high sensitivity and specificity. However, the imperfections of the antibody tests are often overlooked in real clinical practice and in primary care, patients may be started on a gluten-free diet without a duodenal biopsy or referral to a gastroenterologist1Sinclair D. Duncan H. What happens to patients with positive tissue transglutaminase and endomysium antibody results in general practice?.J Clin Pathol. 2004; 57: 943-945Crossref PubMed Scopus (23) Google Scholar. We feel that it is important to test the antibodies in a low-risk group and even with a PPV of only 88%, if all 4 antibodies are used, we would still recommend a duodenal biopsy. We have also previously reported the use of a decision tool for CD in adults using pre-test symptoms of iron deficiency anemia, diarrhea, and weight loss2Hopper A.D. Cross S.S. Hurlstone D.P. et al.Pre-endoscopy serological testing for coeliac disease: evaluation of a clinical decision tool.BMJ. 2007; 334: 729-732Crossref PubMed Scopus (138) Google Scholar. In this study, the high-risk group prevalence of CD was 9.6%, which was somewhat less than the 59% reported by the authors in the decision tool above (we are unsure of their inclusion criteria in this group). Evaluating our clinical decision tool by combining the biopsy of the high-risk group and those patients with a positive tTG result gave a sensitivity of 100% and a NPV of 100% but only a specificity of 60.8% and PPV of 9.3%. Therefore we feel we could exclude a number of patients who did not require a duodenal biopsy due to a high NPV but still recommend performing a duodenal biopsy in all patients suspected of CD due to the low PPV and specificity. We currently still use duodenal biopsy in our center in all patients suspected of CD, regardless of antibody profile as it ensures validity of the diagnosis, allows an assessment of the degree of histologic severity as a baseline (should symptoms not improve), and influences any advice about family screening. Confirmation of the diagnosis based on resolution of symptoms with gluten exclusion alone (or by using a virtual histology technique as suggested above) may ultimately result in confusion for the patient. This point can be illustrated in patients with irritable bowel syndrome (who have had a normal duodenal biopsy) but may benefit from a gluten-free diet3Parker T.J. Naylor S.J. Riordan A.M. et al.Management of patients with food intolerance in irritable bowel syndrome: the development and use of an exclusion diet.J Hum Nutr Diet. 1995; 8: 159-166Crossref Scopus (65) Google Scholar. Two-Step Approach for Diagnosing Celiac DiseaseClinical Gastroenterology and HepatologyVol. 6Issue 10PreviewIn their recent article (Clin Gastroenterol Hepatol 2008;6:314–320) Hopper et al found that adopting a 2-step approach by using IgA tissue transglutaminase (tTG) first and then IgA endomysial antibody (EMA) increases the positive predictive value for diagnosing celiac disease from 28.6% to 71.7%. Furthermore, use of nondeamidated IgA/IgG antigliadin antibody (AGA) conferred no additional benefit. Our results (Figure 1) illustrate that there is a 99.8% positive predictive value if not only IgA tTG/EMA but also IgA and IgG AGA are positive (calculated according to Bayes formula). Full-Text PDF