Celiac disease (CD) is a gluten-induced multiorgan inflammatory disorder, with the small intestine being the primary target of inflammation. The CD diagnosis is based on a small intestinal biopsy demonstrating increased number of intraepithelial lymphocytes, crypt hyperplasia, and villous flattening, which resolves on a gluten-free diet (1). The aetiology of CD is multifactorial, with both genetic and environmental factors involved in disease development (2). Human leukocyte antigen (HLA) genes encoding class II molecules DQ2 and DQ8 are key genetic risk factors. More than 90% of the patients express the DQ2 molecule and most of the remaining patients carry the DQ8 molecule (3).
For a long time, CD was considered a rare disease only affecting children in Europe with an estimated prevalence of 1/1000. However, recent population-based screening studies from different parts of the world have revealed much higher prevalences (eg, ∼1% in many European countries and in the United States) (4–8).
The onset of CD may occur at any age once gluten has been introduced into the diet (9). The symptoms vary considerably between individuals and also with age. In children younger than 2 years of age gastrointestinal symptoms and failure to thrive are common. In older children and adults, the symptoms are often nonspecific, such as abdominal pain, anaemia, osteoporosis, fatigue, and even depression. Consequently, the diagnosis is easily delayed or even missed (4).
The often vague symptoms call for reliable markers to identify subjects from whom a biopsy may be warranted to verify the diagnosis. The serological markers most commonly used in clinical routine are immunoglobulin (Ig) A antibodies against gliadin (AGA-IgA), tissue transglutaminase (tTG-IgA), and endomysium (EMA-IgA) (10). However, some studies suggest that AGA-IgA should be discarded in view of its poor specificity and replaced with tTG-IgA (11,12). Hence, there is a need for more detailed knowledge concerning the age-dependent production of current serological markers in children, as well as to further develop simple, reliable, and inexpensive diagnostic tests. The aim of this study was to investigate the age-dependent serum levels and occurrence of elevated CD–related antibodies in young children, to define the optimal serological procedure when selecting for small intestinal biopsy.
PATIENTS AND METHODS
Celiac Disease Cases and Control Subjects
A prospective register of all new CD cases in children younger than 15 years of age was established in 1991, surveying 40% of the Swedish population in this age group (13). All of the CD cases reported from 1992 to 1995 were included in a case referent study (14), which was the basis for the present study. Besides a biopsy verified diagnosis, a blood sample collected at the biopsy occasion was the inclusion criterion. Thus included were 430 CD children. Excluded were 2 children with IgA deficiency and the analysis was finally based on 428 CD children (281 girls and 147 boys, median age at diagnosis 16 months; range 7.5 months–14 years) (Fig. 1, Table 1).
CD was diagnosed according to the revised European Society for Pediatric Gastroenterology, Hepatology, and Nutrition (15) criteria, including histological examination of a biopsy specimen taken while on a gluten-containing diet, followed by clinical remission on gluten free diet. The biopsy was considered diagnostic for CD when assessment of the small-intestine mucosa demonstrated at least a partial villous atrophy (16,17).
Two different control groups were included in the study, referred to as disease controls and healthy controls. The disease control group encompassed 133 children admitted to the Department of Pediatrics, Umeå University Hospital, during 1992 to 2002 on the suspicion of having CD. All of these had a small intestinal mucosal evaluation without revealing any degree of villous atrophy. Children with IgA deficiency (n = 4) were excluded and the remaining 129 children were included in the disease control group (58 girls and 71 boys; median age 37 months; range 8.5 months–14.6 years) (Fig. 1, Table 1). The healthy control group consisted of children recruited as population-based referents in the case referent study (14), and for ethical reasons no biopsies were performed. The latter group consisted of 87 children (54 girls and 33 boys; median age 29 months; range 14 months–11 years) (Fig. 1, Table 1).
All children included in the present study were younger than 15 years of age and had normal levels of s-IgA (s-IgA >0.06 g/L). For the CD cases and disease controls, blood samples were collected within 1 week from the biopsy occasion. Genotyping for HLA-DQ2 was performed in all children from whom a whole blood sample was available: 176 of 428 children with CD and 72 of 87 healthy controls. Genotyping for HLA-DQ8 was performed in all HLA-DQ2–negative children.
AGA-IgA was determined by enzyme-linked immunoassay (ELISA) (Gliadin IgA VARELISA, Phadia, Freiburg, Germany) in accordance with the manufacturer's instructions. Sera from patients and controls were tested diluted 1:101 and antibody levels were expressed as arbitrary units (U/mL) calculated from a 6-point calibrator curve. Samples with antibody levels above the highest calibrator point were further diluted and retested. The optimal cutoff was 15 U/mL, as estimated from receiver operator characteristics analysis (18).
The tTG-IgA was measured by ELISA (Celikey, Phadia) in accordance with the manufacturer's instructions. Sera from patients and controls were tested diluted 1:101, and antibody levels were expressed as arbitrary units (U/mL) calculated from a 6-point calibrator curve. Samples with antibody levels above the highest calibrator point were further diluted and retested. Optimal cutoff was 4 U/mL, as estimated from receiver operator characteristics analysis, in which EMA-positive CD children comprised the disease group.
Analysis of EMA-IgA was performed with indirect immunofluorescence using tissue sections from marmoset monkey oesophagus mounted on glass slides. Undiluted sera and sera diluted 1:10 with phosphate buffered saline containing 1% bovine serum albumin were applied to the slides, which were incubated for 30 minutes at room temperature. After washing with phosphate buffered saline, the sections were covered with fluorescein conjugated rabbit Fab′2 antihuman IgA (Dako, Copenhagen, Denmark) for 30 minutes, washed with phosphate buffered saline, and examined by fluorescence microscopy. Sera yielding fluorescent binding to the endomysial structure in dilution 1:10 were regarded as positive and were further diluted to determine the final titre for which fluorescence was detected.
Total serum IgA levels were analysed by nephelometry (BN Pro Spec System, Dade Behring, Marburg, Germany). Children with values below 0.06 g/L were considered IgA deficient and excluded from the study.
The class II alleles DQB1*0201/0202 and DQA1*0501 (HLA-DQ2) together with DQB1*0302 (HLA-DQ8) were determined with polymerase chain reaction with sequence-specific primers, using high resolution primers (Dynal Biotech, Oslo, Norway) (19). DNA was extracted with DTAB/CTAB (Sigma Aldrich, Stockholm, Sweden) (20), and amplifications were performed in a Perkin Elmer 9600 thermal cycler. After amplification, the products were visualized by electrophoresis using a 2% agarose gel stained with ethidium bromide.
Small Intestinal Biopsy
A small intestinal biopsy specimen was taken at the level of, or distally to, the ligament of Treitz with a Watson or Storz pediatric capsule under fluoroscopic control. In a few cases, biopsies were taken during upper endoscopy. All of the specimens were first examined by the local pathologist. Thereafter, all of the biopsies were classified according to Alexander (21) by a single pathologist experienced in CD morphology and classification.
Informed consent was obtained from parents, or when applicable participating children. The study was approved by the research ethical committees of all Swedish medical faculties.
The optimal cutoff values for the different antibody markers were estimated from receiver operator characteristics analyses. AGA-IgA and tTG-IgA levels were expressed as median with 10th–90th percentiles U/mL and EMA-IgA titres as median with 10th to 90th percentiles. The Kruskal-Wallis test was used to estimate differences in antibody levels between groups. The sensitivity and specificity of the different tests were calculated with 95% confidence intervals (CIs) using the exact binominal method, and also were compared using McNemar chi-square test. SPSS for Windows, version 12.0.1, (SPSS, Chicago, IL) and GraphPad Prism, version 4, (GraphPad Software, San Diego, CA) were used for statistical analyses.
Among CD children, elevated levels of AGA-IgA were found in 411 of 428 (96%) CD cases (median 258; 10th–90th percentiles 28 to >1000 U/mL), tTG-IgA in 385 of 428 (90%) cases (236; 4 to >1000 U/mL), and EMA-IgA in 383 of 428 (89%) cases (1:400; <1:10–1:1600) (Fig. 2A and B). Elevated levels of all 3 antibodies were found in 364 of 428 (85%) cases, whereas 4 of 428 (1%) had normal levels of all antibodies tested. In the disease control group, 8 of 129 (6%) children had elevated AGA-IgA, 3 of 129 (2%) were tTG-IgA positive, and 4 of 129 (3%) were EMA-IgA positive, whereas in the healthy control group 3 of 87 (3%) children were AGA-IgA positive, 2 of 87 (2%) were tTG-IgA positive, and 4 of 87 (5%) were EMA-IgA positive. There was no significant difference in the levels of AGA-IgA and tTG-IgA between disease control and healthy control subjects. All 5 control children with elevated levels of tTG-IgA and 3 of 11 control children with positive AGA-IgA also had increased levels of EMA-IgA (Fig. 2A and B).
Age-related Antibody Variation
The majority of the CD children (327/428; 76%) in the study were younger than 2 years of age at diagnosis. To explore how the prevalence and magnitude of raised antibody levels varied with age, all children were divided into 4 age groups (Table 1). Concordance with respect to positive and negative antibody results for each marker and age group is depicted in Table 2.
Among CD children younger than 12 months, elevated AGA-IgA was found in 78 of 82 (95%) cases (479; 36 to >1000 U/mL), whereas elevated tTG-IgA was found in 54 of 82 (66%) cases (69; 0.6 to 790 U/mL), and EMA-IgA was increased in 51 of 82 (62%) cases (Fig. 3A and B). In CD children ages 12 to 17.9 months, AGA-IgA was elevated in 166 of 169 (98%) cases (417; 53 to >1000 U/mL), tTG-IgA in 155 of 169 (92%) cases (307; 10 to >1000 U/mL), and EMA-IgA in 157 of 169 (93%) cases. All CD children ages 18 to 23.9 months had elevated AGA-IgA, and all but 1 also had elevated tTG-IgA as well as EMA-IgA. In CD children older than 24 months, elevated AGA-IgA was found in 91 of 101 (90%) cases (86; 15 to 524 U/mL), and all of them were tTG-IgA positive (175; 28 to 629 U/mL), and 100 of 101 (99%) also EMA-IgA positive (Fig. 3A and B). Thus, the AGA-IgA levels of CD children older than 24 months were lower than in the other 3 age groups, whereas the lowest levels of tTG-IgA were found among children younger than 12 months (Fig. 3A and B).
One disease control and 2 healthy control children older than 24 months had elevated levels of all 3 antibodies (Table 2). One of these children had normal mucosa. The other 2 had not undergone a biopsy, but they were both HLA-DQ2 positive.
Including all children, the sensitivity of AGA-IgA was 96% compared with 90% for tTG-IgA and 89% for EMA-IgA (P < 0.0001) (Table 3). The specificity of AGA-IgA was 95% compared with 98% for tTG-IgA, and 96% for EMA-IgA (not significant). Because the prevalence of antibodies differed between younger and older CD children, we also present the sensitivities and specificities of the different antibody tests for children younger and older than 18 months (Table 3). In children younger than 18 months, AGA-IgA had a sensitivity of 97% compared with 83% for both tTG-IgA and EMA-IgA (P < 0.0001). In contrast, in children older than 18 months the sensitivity of AGA-IgA was 94% compared with 99% for both tTG-IgA and EMA-IgA (P = 0.01 and 0.04, respectively) (Table 3). Combining AGA-IgA and tTG-IgA in the youngest age group resulted in a sensitivity of 98% when at least 1 of the tests was positive.
Of 176 genotyped CD cases, 146 (83%) carried both the DQA1*0501 and DQB1*0201 alleles of the HLA-DQ2 heterodimer, and 20 CD children (11%) only had the DQB1*0201 allele. The DQB1*0302 allele of the HLA-DQ8 heterodimer was carried by 15 (8.5%). Two children (1%) had none of the alleles typed for. These 2 children were both younger than 18 months and had increased AGA-IgA but normal tTG-IgA levels. One child was also EMA-IgA positive. Both children were followed up with biopsy after a period of gluten-free diet, and at that occasion both of them showed improvement of the mucosal structure of the small intestine.
In the healthy control group, 19 of 72 genotyped subjects (26%) carried the DQB1*0201 or 0202 allele together with the DQA1*0501 allele, and 8 of 72 (11%) subjects carried genes encoding only 1 chain of the DQ2 heterodimer. The number of control subjects without any of the DQ2-encoding alleles was 45 of 72 (63%), and of those 15 carried the DQB1*0302 allele.
There was no significant difference in antibody levels for CD children or controls carrying the DQ2 or the DQ8 alleles, respectively. However, control children with at least 1 of the DQ2 alleles or the DQ8 heterodimer had higher AGA-IgA levels (P < 0.01) than children without any of these alleles (data not shown).
The main finding in the present study was that raised levels of AGA-IgA were more prevalent than tTG-IgA and EMA-IgA in CD children younger than 18 months of age. In contrast, in older CD children both tTG-IgA and EMA-IgA were more prevalent than AGA-IgA.
The importance of antibody assessment in predicting childhood CD has increased along with the increased proportion of children with minor or atypical symptoms (6). Since the discovery of tTG as the main autoantigen in CD, anti-tTG antibodies have become an important tool in identifying CD and in screening of risk groups for the disease (22). Comparisons of the different CD-related antibodies in young children are sparse, but there are reports suggesting that a considerable proportion of the youngest CD children lack EMA despite positive serum levels of AGA (23). Therefore, it is important to elucidate the relation between CD-related antibodies during active disease in young children.
Our present study encompasses a large group of children with biopsy-verified CD diagnosed during a period when Sweden experienced an epidemic of the disease in young children (13). As many as 76% of the included CD children were younger than 2 years of age, which gave us a unique opportunity to study how the occurrence of serum antibodies in childhood correlates with age. In the present study, as many as 34% of the included CD children younger than 12 months of age had normal serum levels of tTG-IgA. However, the presence of tTG-IgA increased with age and already at 18 months most children had an autoantibody response against both tTG and EMA. The presence of AGA-IgA, however, decreased with age and was least prevalent in children older than 2 years of age.
Our results show that autoantibodies can develop rapidly after an intestinal exposure to an antigen. A majority of children diagnosed with CD during their first year of life actually had elevated tTG-IgA and EMA-IgA, and the youngest child with highly elevated levels was only 7.5 months of age. The regulation of the humoral immune response resulting in development of AGA and autoantibody production in CD is not well understood. The tTG-IgA response of CD subjects originates mainly from the intestinal lesion (24), whereas AGA-IgA–producing cells also reach the peripheral circulation (25), which may explain why elevated AGA-IgA can precede detectable tTG-IgA levels in serum. The time at which gluten is introduced into the diet, as well as the amount and duration of gluten exposure, will also most likely influence the serum levels of CD-related antibodies.
According to the North American Society for Pediatric Gastroenterology, Hepatology, and Nutrition, AGA analysis can be abandoned for routine childhood CD testing (11). Our results support their conclusions for symptomatic children older than 18 months of age. However, AGA-IgA should still be used in combination with tTG-IgA in children younger than 18 months of age to increase the likelihood that all children with CD are identified. In our study as many as 17% of the CD children in the youngest age group would have remained undiagnosed if tTG-IgA had been used alone. However, the use of AGA in older children and adults seems doubtful, as there are several reports of increased AGA levels in patients with various enteral diseases other than CD (26,27). Furthermore, in the present study 6% of all disease control subjects had elevated AGA-IgA despite normal intestinal mucosal morphology. However, because of the comparatively low number of control children younger than 2 years of age, the estimate is uncertain and the percentage of children with increased AGA-IgA may be even higher.
Recently, it was reported that an ELISA based on synthetic deamidated gliadin peptides had a higher diagnostic accuracy than conventional assays based on native gliadin (28). Although that study was carried out with an adult population, it is possible that AGA-IgA assessment with a more homogenous antigen than native gliadin also may improve the test performance in children, which would benefit the diagnosis of CD in children without detectable serum levels of tTG-IgA.
In line with previous reports (23,29), we found only a few control subjects with increased serum levels of tTG-IgA or EMA-IgA. Three control children had increased levels of both autoantibodies and AGA, and we cannot exclude that these children will develop CD later in life. However, 2 of the healthy control subjects with elevated EMA-IgA were found to have normal levels of these antibodies on a later occasion. It was recently reported that EMA-IgA levels can fluctuate over time and even disappear spontaneously in children exposed to dietary gluten without the development of CD (30). Thus, only a prolonged follow-up will reveal whether the control children with elevated EMA-IgA observed in our study will develop CD later in life. Whether the elevated levels of AGA in the remaining control subjects were caused by inflammatory diseases other than CD, or whether they predict forthcoming CD, could not be determined by the present study.
Notably, during the Swedish CD epidemic high amounts of gluten were introduced abruptly into the infant diet, and often after breast-feeding had been discontinued (13,14). This infant feeding pattern possibly also enhanced the production of AGA and anti-tTG antibodies. After the epidemic period, the incidence of symptomatic CD decreased, most likely owing to changed infant feeding with a more gradual gluten introduction during ongoing breast-feeding and other, not yet identified factors. To what extent the findings in our study mirror the present situation remains to be investigated.
In summary, our results suggest that in children ages 18 months or older both tTG-IgA and EMA-IgA are sufficiently accurate to be used as a single antibody marker, whereas a large proportion of younger CD children lack these antibodies. Therefore, when selecting children for small intestinal biopsy, the detection of a combination of AGA-IgA and tTG-IgA is optimal for identifying untreated CD in children younger than 18 months.
We thank all of the participating children and their families; Susanne Walther, administrative assistant, Department of Public Health and Clinical Medicine, Umeå University; Anna-Karin Åberg, medical laboratory technologist, Clinical Microbiology and Immunology, Örebro University Hospital; Roger Stenling, MD, PhD, Department of Pathology, Umeå University, for diagnostic support; and collaborators at all of the participating pediatric departments, in particular Göte Forsberg and Mats Lundström (Umeå), Lars Danielsson (Danderyd), Henry Ascher (Göteborg), Jan Ejderhamn (Huddinge), Ulf Jansson (Jönköping), Karin Fälth-Magnusson (Linköping), Bertil Cavell (Lund), Lars Stenhammar (Norrköping), Kristjan Arinbjarnarson (Skellefteå), Mats Eriksson (Uddevalla), Urban Myrdal (Västerås), and Bo Lindquist (Örebro).
1. Meeuwisse G. Diagnostic criteria in coeliac disease. Acta Paediatr Scand 1970; 59:461–463.
2. Kagnoff MF. Overview and pathogenesis of celiac disease. Gastroenterology 2005; 128:10–18.
3. Sollid LM, Thorsby E. HLA susceptibility genes in celiac disease: genetic mapping and role in pathogenesis. Gastroenterology 1993; 105:910–922.
4. Green PH, Jabri B. Coeliac disease. Lancet 2003; 362:383–391.
5. Rewers M. Epidemiology of celiac disease: what are the prevalence, incidence, and progression of celiac disease? Gastroenterology 2005; 128:47–51.
6. Maki M, Mustalahti K, Kokkonen J, et al
. Prevalence of celiac disease among children in Finland. N Engl J Med 2003; 348:2517–2524.
7. Fasano A, Berti I, Gerarduzzi T, et al
. Prevalence of celiac disease in at-risk and not-at-risk groups in the United States: a large multicenter study. Arch Intern Med 2003; 163:286–292.
8. Carlsson AK, Axelsson IE, Borulf SK, et al
. Serological screening for celiac disease in healthy 2.5-year-old children in Sweden. Pediatrics 2001; 107:42–45.
9. Abdulkarim AS, Murray JA. Review article: the diagnosis of coeliac disease. Aliment Pharmacol Ther 2003; 17:987–995.
10. Farrell RJ, Kelly CP. Diagnosis of celiac sprue. Am J Gastroenterol 2001; 96:3237–3246.
11. Hill ID, Dirks MH, Liptak GS, et al
. Guideline for the diagnosis and treatment of celiac disease in children: recommendations of the North American Society for Pediatric Gastroenterology, Hepatology, and Nutrition. J Pediatr Gastroenterol Nutr 2005; 40:1–19.
12. Reeves GE, Squance ML, Duggan AE, et al
. Diagnostic accuracy of coeliac serological tests: a prospective study. Eur J Gastroenterol Hepatol 2006; 18:493–501.
13. Ivarsson A, Persson LA, Nystrom L, et al
. Epidemic of coeliac disease in Swedish children. Acta Paediatr 2000; 89:165–171.
14. Ivarsson A, Hernell O, Stenlund H, et al
. Breast-feeding protects against celiac disease. Am J Clin Nutr 2002; 75:914–921.
15. Walker-Smith JA, Guandalini S, Schmitz J, et al
. Revised criteria for diagnosis of coeliac disease. Report of Working Group of European Society of Paediatric Gastroenterology and Nutrition. Arch Dis Child 1990; 65:909–911.
16. Oberhuber G. Histopathology of celiac disease. Biomed Pharmacother 2000; 54:368–372.
17. Marsh MN. The natural history of gluten sensitivity: defining, refining, and re-defining. Q J Med 1995; 88:9–13.
18. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988; 44:837–845.
19. Olerup O, Aldener A, Fogdell A. HLA-DQB1 and -DQA1 typing by PCR amplification with sequence-specific primers (PCR-SSP) in 2 hours. Tissue Antigens 1993; 41:119–134.
20. Gustincich S, Manfioletti G, Del Sal G, et al
. A fast method for high-quality genomic DNA extraction from whole human blood. Biotechniques 1991; 11:298–300, 302.
21. Alexander JO. The small intestine in dermatitis herpetiformis. In: Alexander JO, editor. Dermatitis Herpetiformis. Major Problems in Dermatology, Volume 4. London: Saunders; 1975. pp. 236–280.
22. Dieterich W, Ehnis T, Bauer M, et al
. Identification of tissue transglutaminase as the autoantigen of celiac disease. Nat Med 1997; 3:797–801.
23. Burgin-Wolff A, Gaze H, Hadziselimovic F, et al
. Antigliadin and antiendomysium antibody determination for coeliac disease. Arch Dis Child 1991; 66:941–947.
24. Marzari R, Sblattero D, Florian F, et al
. Molecular dissection of the tissue transglutaminase autoantibody response in celiac disease. J Immunol 2001; 166:4170–4176.
25. Hansson T, Dannaeus A, Kraaz W, et al
. Production of antibodies to gliadin by peripheral blood lymphocytes in children with celiac disease: the use of an enzyme-linked immunospot technique for screening and follow-up. Pediatr Res 1997; 41:554–559.
26. Kull K, Uibo O, Salupere R, et al
. High frequency of antigliadin antibodies and absence of antireticulin and antiendomysium antibodies in patients with ulcerative colitis. J Gastroenterol 1999; 34:61–65.
27. Paimela L, Kurki P, Leirisalo-Repo M, et al
. Gliadin immune reactivity in patients with rheumatoid arthritis. Clin Exp Rheumatol 1995; 13:603–607.
28. Sugai E, Vazquez H, Nachman F, et al
. Accuracy of testing for antibodies to synthetic gliadin-related peptides in celiac disease. Clin Gastroenterol Hepatol 2006; 4:1112–1117.
29. Hansson T, Dahlbom I, Rogberg S, et al
. Recombinant human tissue transglutaminase for diagnosis and follow-up of childhood coeliac disease. Pediatr Res 2002; 51:700–705.
30. Simell S, Hoppu S, Hekkala A, et al
. Fate of five celiac disease–associated antibodies during normal diet in genetically at-risk children observed from birth in a natural history study. Am J Gastroenterol 2007; 102:2026–2035.