Tuesday 11 March 2014
Diagnostics in Celiac Disease
Promotor: Prof. dr E.E.S. Nieuwenhuis
Defence: 11 March 2014
The prevalence of celiac disease (CD) is approximately 1% worldwide. The disease is characterized by an immunological response to gluten, the storage protein in wheat, barley and rye. This response causes intestinal inflammation but can also be detected serologically by measuring disease specific (auto-)antibodies. Detecting this immunological response against gluten through histological and/or serological methods is used in the diagnosis of CD. However, none of these methods is perfect yet, so in this thesis we aimed at improving diagnostic strategies in CD.
Histological lesions in CD are graded using the Marsh classification. Finding an increased number of intra-epithelial lymphocytes (IELs) and crypt hyperplasia (Marsh II), generally with villous atrophy (Marsh III), was considered to be the gold standard for the diagnosis of CD. However, several factors might make evaluation of small intestinal histology suboptimal. In Chapter 2 we therefore studied the inter-observer variability in the histological diagnosis of CD and found that the agreement between 2 pathologists was only moderate, with a Kappa value of 0.486. More importantly in 7.4% of the cases a discrepancy in the diagnosis of CD between both pathologists was found, which occurred most commonly when the quality of the biopsy specimen was suboptimal. So, we cannot totally rely only on histology when diagnosing CD. In order to support the pediatric gastroenterologist in making the correct diagnosis, a pathology report should therefore include a detailed statement about the quality of the biopsy samples and the extent to which the pathologist is confident with the diagnosis. In addition, when a discrepancy between serology and histology is found, the first step should be to revise the biopsies.
In Chapter 2 we also showed that performing CD3 staining in order to detect IELs could lead to a different diagnosis. Because CD3 staining was not done systematically, we studied the additional value of this immunohistochemical staining in Chapter 3, where we concluded that CD3 staining should be performed whenever there is a discrepancy between serology and the conclusion of the pathologist based on the routine sections. In our study this strategy resulted in an additional diagnosis of Marsh I in 5.0% of the studied patients, while in 0.6% of the cases a Marsh I lesion could be withdrawn after assessment of CD3 stains. More importantly, in 5.7% of the patients the diagnosis of CD was missed on routine stains. Finally, in 0.6% of the cases the diagnosis of CD could be rejected after evaluation of the CD3 sections.
Secondly, we studied methods to improve non-invasive tests in the diagnosis of CD. HLA typing is the best available test to exclude CD, because virtually all patients with CD are either HLA-DQ2.5 or HLA-DQ8 positive. However, the heterodimer HLA-DQ2.2 has also been detected in patients with CD. In Chapter 4 we therefore studied the frequency of HLA-DQ2.2 in CD. We found that the 5.8% of CD patients, who lacked both HLA-DQ2.5 and HLA-DQ8, were all HLA-DQ2.2 positive. This heterodimer should therefore also be considered as positive when screening for CD.
Subsequently, we focused on serological tests as markers for CD. Immunoglobulin A (IgA) antibodies against tissue-transglutaminase (tTGA) or endomysium (EMA) are considered to be specific and sensitive screening tools for the disease, although they do not reach 100% accuracy. In Chapter 5 we found that the newly developed antibodies against deamidated gliadin peptides (a-DGP) in general do not outperform EMA and tTGA. Because both EMA and tTGA have been reported to be less sensitive in children <2 years of age, we also studied their performance in this specific subgroup. Surprisingly both EMA and tTGA performed better than was expected from previous literature, but still the Immunoglobulin G class a-DGP were superior, showing 100% accuracy. Adding Immunoglobulin G a-DGP to the diagnostic work-up of CD might therefore be beneficial in young children.
In Chapter 6 we studied in a retrospective design whether a tTGA of at least 10 times the upper limit of normal (100 U/ml) can be used to make the diagnosis of CD without needing a biopsy. We showed that all symptomatic patients with a tTGA of at least 100 U/ml, who all also had positive EMA and responded well to a gluten free diet, indeed had histological lesions compatible with CD. In Chapter 7, we subsequently confirmed these findings in a prospective design, adding more strength to this approach. By applying these criteria, which are now implemented in the new ESPGHAN guidelines for the diagnosis of CD, a small intestinal biopsy will not be needed in almost 40% of children suspected to have CD. Finally, in Chapter 8 we showed that patients with very high tTGA also have a more severe disease: they have more extra-intestinal symptoms and more histological involvement. In addition they more often carry multiple disease associated HLA-types.