This article is mentioned in:

Introduction

Non-invasive examination methods are increasingly important in diagnosing celiac disease (CD). New options for diagnosing CD have been discovered as a result of progress in immunology (1). The detection of circulating autoantibodies is becoming more frequent clinically. As a result, many new, highly heterogeneous cases of CD have been diagnosed, and consequently, recognition of the incidence of CD in the pediatric and adult populations has increased. Detection of anti-endomysial antibodies (AEA) and anti-tissue transglutaminase antibodies (t-TG), characteristic for their high sensitivity and specificity, play important roles in diagnosing and clinically monitoring CD (2,3). Nevertheless, histopathology remains the major tool for clinicians, and particularly gastroenterologists, for definitively diagnosing CD (4,5). A variety of studies have documented that AEA, and t-TG, are highly specific in predicting CD (6,7). There are also seronegative data, which is particularly true for the adult population (8). The diagnosis in children is more successful (9), as pediatric CD has clear features in comparison to adult CD (10). It is often difficult to diagnose CD quickly, due to the fact that CD has a wide range of symptoms. However, we do not know if the same problems exist in the diagnosis of pediatric patients. We decided to test this hypothesis on a population of children in Eastern Slovakia. The objective of this study was to compare the levels of AEA and t-TG (IgA, IgG) in the plasma with that of histological abnormalities in pediatric patients with CD.

Materials and methods

Patients

Between 2005 and 2010, a program which searched for CD in the pediatric population was performed. All samples were obtained from the Teaching Faculty Hospital (Košice, Slovakia). In total, 4,247 biopsy samples were used. Informed consent was provided by the parents of the children. This study was approved by the Ethics Committee of Pavol Jozef Safarik University in Kosice, Kosice, Slovak Republic.

Serum analysis

Sera from all children with digestive disorders and chronic non-specific gastroenteritis were sampled during hospitalization, in addition to routine biochemical investigations. Blood sera were separated using routine biochemical methodology (11). Serological testing kits were provided by INOVA Diagnostics (San Diego, CA, USA). AEA (IgA) and t-TG (IgA and IgG) were examined using immunofluorescence and other immunological methods. For AEA examination, binding to human umbilical cord tissue was used and for t-TG we used the classical immunological ELISA test. AEA was investigated by indirect immunofluorescence neutrophil inhibitory factor (NIF), and the limit value for t-TG was 10 U/ml. Only samples where AEA were negative were selected. Regardless of the outcome of the serology, biopsies from the duodenum were subsequently performed.

Biopsies

Each patient underwent between three to five endoscopic biopsies from the bulb and third duodenal portions and the samples were fixed in a 10% formalin solution. Duodenal biopsy specimens were obtained by standard procedures, including orientation, processing and interpretation in order to avoid, or minimize, their misinterpretation (12). The samples were hematoxylin & eosin (H&E) stained and evaluated in accordance with the Marsh-Oberhuber system (13,14).

Results

Samples

From the 4,247 samples, we found 44 new cases of CD in children with AEA negativity, but these instances always coexisted with histological abnormalities of the small intestine (Table I). A t-TG (IgA) value >10 U/ml was detected in 4 patients, and a t-TG (IgG) value >10 U/ml was detected in another 4 patients. All others t-TG (IgA) and t-TG (IgG) values were <10 U/ml. There was elevated t-TG (IgA) or t-TG (IgG), but none of the patients exhibited increases in both values. All values are in the context of the AEA negativity. As shown in the Table I, AEA negativity is always visible in Marsh I histological grade, as well as in the advanced grade of histological abnormalities, MIII. Values of t-TG (IgA) and t-TG (IgG) >10 U/ml were visible only in the advanced Marsh III grade of histological abnormalities, remaining <10 U/ml in histological grade Marsh I. Of the 44 samples, nine were diagnosed as Marsh I, two as Marsh IIIA, 14 as Marsh IIIB and 19 as Marsh IIIC. Variable damage, including inflammatory lesions with fine, or total, villous atrophy of the small intestine was visible in our investigated materials. In the cases with partial villous atrophy, the differentiated intraepithelial inflammatory florid process (Fig. 1) was visible as well. In cases where it was difficult to obtain a morphological image, which prevented the exact evaluation of mucous membrane characteristics, we were able to find only residual structures of crypts with compensatory cryptic hyperplasia and hyperplasia of Brunner’s glands. Regarding the material characteristics, the main feature was severe structural disorder, resulting in completely smoothed mucous membrane relief, which consisted of mixed dense inflammation infiltrate with predominant intraepithelial lymphocytes and plasma cells (Fig. 2).

Figure 1 Section of the duodenum with total villous atrophy and well-activated inflammatory process throughout the mucosa. 1, epithelium; 2, inflammatory process and plasmatic cells. Magnification, ×200. H&E, hematoxylin & eosin.

Figure 2 Sample with total villous atrophy (blue arrows) and epithelial erosion (red arrows), including inflammatory process of the small bowel. Magnification, ×100. H&E, hematoxylin & eosin.

Table I Overview of immunological evaluation values and associated histopathological description in pediatric CD.

Table I

Overview of immunological evaluation values and associated histopathological description in pediatric CD.

Patient	t-TG IgA	t-TG IgG	AEA	Marsh
1	2.6	3.8	Negative	I
2	2.7	2.4	Negative	I
3	1.0	1.6	Negative	I
4	4.4	2.8	Negative	I
5	2.7	2.5	Negative	I
6	3.2	4.4	Negative	I
7	1.0	5.4	Negative	I
8	2.6	3.8	Negative	I
9	2.7	2.4	Negative	I
10	7.2	6.8	Negative	M3A
11	7.1	4.2	Negative	M3A
12	3.0	3.5	Negative	M3B
13	2.6	4.9	Negative	M3B
14	3.4	9.1	Negative	M3B
15	2.5	1.4	Negative	M3B
16	11.2	2.7	Negative	M3B
17	7.8	7.3	Negative	M3B
18	4.1	4.5	Negative	M3B
19	1.0	1.8	Negative	M3B
20	1.0	4.0	Negative	M3B
21	3.0	3.5	Negative	M3B
22	2.6	4.9	Negative	M3B
23	3.4	9.1	Negative	M3B
24	2.5	1.4	Negative	M3B
25	12.2	2.7	Negative	M3B
26	1.7	89.3	Negative	M3C
27	2.5	4.8	Negative	M3C
28	5.1	10.1	Negative	M3C
29	6.1	3.5	Negative	M3C
30	2.5	2.6	Negative	M3C
31	2.4	6.9	Negative	M3C
32	5.1	4.0	Negative	M3C
33	17.2	4.8	Negative	M3C
34	3.8	5.9	Negative	M3C
35	8.3	8.0	Negative	M3C
36	3.3	1.7	Negative	M3C
37	2.7	3.9	Negative	M3C
38	1.7	89.3	Negative	M3C
39	2.5	4.8	Negative	M3C
40	5.1	10.1	Negative	M3C
41	15.6	8.2	Negative	M3C
42	8.3	8.0	Negative	M3C
43	3.3	1.7	Negative	M3C
44	2.7	3.9	Negative	M3C

[i] CD, celiac disease; AEA, anti-endomysial antibody; t-TG, anti-tissue transglutaminase antibody.

Discussion

In this study, we wanted to demonstrate the lack of association between autoantibodies and lesions in certain cases of pediatric CD. This may justify the necessity of biopsies for the accurate diagnosis of CD. We show lesions, found in our samples, that are diagnostically important. The findings of Rostami et al(15) demonstrated that AEA have limited value in CD screening programs. Their sensitivity in patients with total villous atrophy (Marsh IIIC) was 100%, but with partial atrophy (Marsh IIIA) the sensitivity of AEA decreased to 3l%. Furthermore, a high IgA t-TG titer correlates with the histopathological finding of Marsh IIIC (16,17). Another study revealed that positive AEA are observed mostly in CD patients with severe tissue damage due to partial or total villous atrophy, but they have a low sensitivity in CD patients with partial villous atrophy in spite of the use of different substrates (18,19). Additionally, a study by Tărmure et al(20) demonstrated that in patients with Marsh I and Marsh II, the sensitivity of AEA was lower. These results demonstrated that the autoantibody titers may correlate with the damage to the small intestine. In our study, there have been negative antibodies with Marsh IIIC, demonstrating that serological screening is not 100% accurate for the diagnosis of CD. That serological tests would be negative was known for adult CD patients but was not previously known for pediatric CD patients. It is possible that clinical symptomatology is poor in certain children, while the disease remains active, and these diagnostic problems require a solution. Therefore, a biopsy must always be performed, so that an expert pathologist may accurately distinguish CD from other diseases (21).

The literature has provided evidence that AEA titers are the first to disappear following the introduction of a gluten-free diet, as CD is induced by gluten and related cereal proteins. Midhagen et al(22) demonstrated that antibody titers decreased sharply within the first month following the introduction of a gluten-free diet. It is possible that weak reactions of AEA or t-TG are caused by the ingestion of gluten-free food, but the degree of histological damage remains high. Fernández-Salazar et al(23) documented that AEA negativity is not rare in patients with CD. These results demonstrate that sensitivity to serology varies according to the intestinal damage and the introduction of a gluten-free diet. The results of other studies also support this hypothesis (24,25). Therefore, we recommend multiple endoscopic biopsies from the proximal and distal duodenum (26,27). Furthermore, in cases of low IgA t-TG titers or in patients with IgA deficiency, intestinal biopsies should remain mandatory (28). The results of certain studies have revealed that only patients with a IgA t-TG level ≥100 U/ml and whose symptoms improve upon the introduction of a gluten-free diet may not require a biopsy of the small intestine to confirm CD (29,30). Although these results are promising, in our opinion, biopsies should not be rejected. However, there are children who are AEA positive and exhibit a normal mucosal villous morphology of the small intestine, which indicates that the diagnostic criteria for CD should be re-evaluated (31). These disparities were described for other antibodies [e.g., AGA (IgA), ARA (IgA) and JAB (IgA)], which is also important for the pathologist to avoid rejection of biopsies based on insufficient evidence (32). Certain studies report that an accurate diagnosis is supported by the genetic analysis of HLADQ2 or HLADQ8 (33,34), but these tests are not yet standard practice in all workplaces, and HLA association alone is inadequate for providing an explanation of the inheritance pattern. For children with a biopsy consistent with CD, in addition to a positive response to a gluten-free diet, HLA testing is not required (35). Di Osdado et al(36) found variations in the expression levels of 109 genes by comparing Marsh III and Marsh 0 lesions, which have roles in the proliferation and differentiation pathways. These results are crucial and encouraging for medical practice, as they may lead to alternative future genetic analyses and the discovery of combinations with new candidate genes for CD. Recent studies have focused on ways to speed up the diagnosis of CD (37,38), but according to our study, biopsies remain necessary. Until we find a combination of diagnostic methods that are 100% reliable, we should continue to perform biopsies. We believe that currently biopsies are the optimum way for the gastroenterologist to accurately diagnose CD in at risk pediatric groups, as well as in children with unspecified long-term related digestive problems. Histological results are important for clinicians, but particularly for the gastroenterologist, in order to establish tests and interpret all available data (39,40).

It has been many years since CD was considered to be a rare disease and experts have agreed that the incidence of CD will increase, since more programs have started to investigate the ‘celiac iceburg’ (41). Autoantibodies are considered to only be helpful in clinical practice, with regards to CD screening. In accordance with this, a biopsy of the small intestine is always required for the accurate diagnosis of CD. We recommend a minimum of four endoscopic biopsies to be obtained from the bulb and third duodenal portions, as a section of the biopsy may be damaged, and the absence of certain material that should be investigated may result in incorrect histopathological interpretation. Slice descriptions should include all relevant information, and the histopathological report should be clear and conform to the recommended standards for intestinal biopsies. We consider this information to be particularly important for pathologists, which is in accordance with the views expressed in other studies (42). We recommend the use of autoantibodies only as a supportive method, as a control in gluten-free diets or as an early predictor of CD with necessary further follow-up. Recent European Society of Pediatric Gastroenterology, Hepatology and Nutrition (ESPGHAN) guidelines state that in cases where a child manifests clear symptoms of CD and demonstrates particularly high levels of t-TG, with positive HLA testing, the physician may consider omitting a biopsy. However, there are some instances where pediatric CD has remained undiagnosed, providing support for a revision to the diagnostic criteria in a clinical, endoscopic and serological context. However, since there are no reliable data on the incidence of CD in children in the East region of the Slovak Republic, our results may be considered to be valid for other countries as well, given that many countries have no screening, only local screening or use different diagnosis methods. The incidence of CD in pediatric patients is higher than previously assumed.

Recent and current literature indicates that celiac disease is underdiagnosed. From our results we conclude that the screening program should be considerably more intense, particluarly in pediatrics. In numerous cases it is difficult to diagnose only from the serum. Biopsies should be part of the investigation, regardless of the outcome of serology.

Acknowledgements

This study was supported by the IGA MZCR: NT 13424-4/2012 grant.

References