Assays for immunoglobulin (Ig) A antibodies against tissue transglutaminase (anti-tTG), endomysium (EmA) and for IgG antibodies against deamidated gliadin peptides (anti-dGli) in serum have a high sensitivity and specificity for coeliac disease (CD) in children (1,2); however, in young children (up to 2 years old), antibodies against native gliadin (anti-nGli) are still often assumed to have the best performance. At young age, IgA-EmA have been reported to have a low sensitivity (3–8), with maximum values of 89%. The sensitivity of IgA-anti-tTG ranged between 83% and 90% (1,6,8,9). Specificity of IgG-anti-nGli was only 77% at high sensitivity (5). IgA-anti-nGli were claimed to be best in finding CD in young children (6) with sensitivity between 82% and 97% and specificity between 88% and 94% (1,5–8). The diagnostic performance of anti-dGli in young children still needs to be established. Recently, it was shown that in children younger than 2 years with signs of chronic enteropathy and high serum levels of anti-nGli but normal values of anti-tTG and of EmA, CD can be predicted by high serum levels of anti-dGli (10).
The aim of our study was to compare the diagnostic performance of assays for IgA and IgG-anti-dGli, IgA and IgG-anti-nGli, IgA and IgG-anti-tTG, and IgA-EmA in the diagnosis of CD at young age.
Sera of 184 children younger than 2 years old with and without CD were retrospectively investigated. The patients were recruited from the Children's Hospital of the Clinical Centre “Sankt Georg” Leipzig, (Germany), University Hospital Leuven (Belgium), University Children's Hospitals of Leipzig, Tübingen, München, and Giessen (Germany), Children's University Hospital Graz (Austria), and from the Department of Paediatrics of the University Medical Centre Leiden (the Netherlands). The patients comprised 42 children with CD and 142 controls (100 girls and 84 boys, mean age 1.34 years, range 0.4–2.0 years). Sera were withdrawn at the time of the diagnostic duodenal biopsies. All of the patients were biopsied under normal diet due to suspicion of CD or other gastrointestinal disorders. Intestinal pathology of all of the patients with CD was in accordance with Marsh 2 or Marsh 3 criteria (11,12). All of the patients with CD improved under gluten-free diet (GFD). The study was approved by the ethical committee of the University of Leipzig and of the local ethical committees of the participating centres.
IgA and IgG antibodies against deamidated gliadin analogous fusion peptides (anti-dGli), anti-nGli, anti-tTG, and IgA-EmA were measured (blinded to the histological diagnosis) with test kits from EUROIMMUN Medizinische Labordiagnostika, Lübeck, Germany. EmA were estimated by indirect immunofluorescence analysis using a combination of primate oesophagus, primate small intestine, and primate liver. The analyses were performed by EUROIMMUN. Cutoffs were as suggested by the manufacturer (Table 1).
The data were evaluated by receiver operating characteristic (ROC) analysis. The area under the ROC curves (AUC) was calculated. Differences between ROC curves were evaluated by pairwise comparison according to χ2 analysis. An error probability P < 0.05 was considered statistically significant. Noninferiority testing was performed if there was no statistically significant difference. For noninferiority testing, the lower end of 90% confidence intervals (CI) of differences between AUCs was considered. Noninferiority was assumed if the lower end of this CI was not below a zone of diagnostic indifference of 0.01.
For the cutoffs suggested by the manufacturer, diagnostic accuracies, sensitivities, and specificities were calculated. Significance of differences (P < 0.05) was evaluated applying McNemar test. For noninferiority testing, the proportions of children being false-negative or false-positive were compared between 2 tests. The z test was applied for calculation of the 90% CI of differences in proportions and for comparison of sensitivities and specificities in the different age groups.
From all of the tests investigated, only assays for IgG-anti-dGli, IgA-anti-tTG, and IgA-EmA had high specificity (≥0.96) connected with high sensitivity (≥0.86), high positive predictive values (≥0.52 and ≥0.69 at pretest probabilities of 0.05 and 0.1, respectively), and high negative predictive values (≥0.99 and ≥0.98 at pretest probabilities of 0.05 and 0.1, respectively; Table 1).
The IgG-anti-tTG assay showed the highest specificity (0.993). The specificity of the 2 anti-dGli tests (0.972 and 0.965) was comparable with that of the assays for IgA-anti-tTG (0.972) and IgA-EmA (0.958). The specificity of the IgG-anti-nGli test was significantly lower than that of all of the other tests. The IgG-anti-tTG-test had a significantly higher specificity than the IgA-anti-nGli assay. The 2 anti-dGli assays were not inferior to the IgG-anti-nGli test.
The IgG-anti-nGli test exhibited the highest sensitivity (0.929). The sensitivity of the IgG-anti-dGli test (0.857) was as high as that of the IgA-anti-tTG and the IgA-EmA assay, and higher than that of IgA-anti-dGli (0.810), IgA-anti-nGli (0.785), and IgG-anti-tTG (0.476) assays. The low sensitivity of the latter test differed significantly from that of all of the other tests.
Positive likelihood ratios (PLR) were lowest for IgG-anti-nGli. These antibodies, however, showed the also lowest negative likelihood ratio (NLR).
The diagnostic odds ratios (DOR) as an integrated measure of sensitivity and specificity were lowest for IgA and IgG-anti-nGli (48.4 and 26.3, respectively) and highest for IgA-anti-tTG and IgG-anti-dGli (207 and 164, respectively). Similarly, ROC analysis revealed that from all of the antibody tests examined, the AUCs of the IgA-anti-tTG and IgG-anti-dGli assay were highest (0.951 and 0.950, respectively). Statistical evaluation of the ROC analysis showed that the IgG-anti-dGli assay was not inferior to that of IgA-anti-nGli, IgG-anti-nGli, IgG-anti-tTG, and IgA-EmA. The IgA-tTG assay was not inferior to the IgA-anti-dGli, IgA-anti-nGli, and the IgA-EmA assay.
Mean concentrations of IgA- and IgG-anti-dGli, IgA-anti-tTG, and IgA-EmA were 50-fold higher in patients with CD than in controls. Median concentrations were even increased at least 200-fold. For IgA- and IgG-anti-nGli and IgG-anti-tTG, the factor of increase was lower (Table 1).
Positive and negative predictive values of the tests in the children up to 2 years are shown in Figure 1 as a function of pretest probability. Positive predictive values of the anti-dGli, anti-tTG, and EmA tests were higher (≥0.52 at a pretest probability of 0.05 and ≥0.69 at a pretest probability of 0.1) than that of anti-nGli assays (≤0.37 at a pretest probability of 0.05 and ≤0.55 at a pretest probability of 0.1). The negative predictive value of the IgG-anti-tTG test was lowest (0.97 at a pretest probability of 0.05 and 0.94 at a pretest probability of 0.1). Only IgA- and IgG-anti-dGli, IgA-anti-tTG, and IgA-EmA showed a high positive predictive value in combination with a high negative predictive value.
In 19 (45%) of our patients with CD up to 2 years old, all 7 antibody tests were positive (Table 2). In 12 additional patients with CD (29%), all but the IgG-anti-tTG antibodies were above the cutoffs. In 3 of the patients with CD, antibody positivity was restricted to IgG-anti-nGli antibodies. In an additional 2 patients, none of the tests was positive. There were 6 children (14%) with negative IgA-anti-tTG among the patients with CD. Details of these patients are shown in Table 3.
In 89 (63%) of the controls up to 2 years old, none of the tests was positive (Table 2). In another 35 controls (25%), only the concentration of IgG-anti-nGli was elevated. In 3 controls (2%), only IgA- and IgG-anti-nGli were positive. In 2 controls (1%), IgA-anti-tTG together with IgG-anti-nGli were elevated. In 2 additional controls (1%), all but IgA-anti-tTG and IgG-anti-tTG were positive. In the remaining 11 controls, different combinations of antibody positivity were observed, which occurred only once. There was a total of 4 controls with increased IgA-anti-tTG. In all of them, the concentration of IgA-anti-tTG was higher than 2-fold the cutoff. In 2 of them, an elevated concentration of IgG-anti-dGli was also found. In 1 of the 2 controls, the antibody concentration could be controlled after 19 months (both antibodies positive) and after 23 months (both antibodies negative).
Except for IgG-anti-nGli, the specificity of all of the antibody tests was high for CD for patients up to 2 years old; however, the assays with high specificity (IgA- and IgG-anti-dGli and anti-tTG, IgA-anti-nGli and EmA) exhibited a low sensitivity, whereas the low specificity of the IgG-anti-nGli test was associated with a high sensitivity. Using paired indicators of diagnostic performance such as sensitivity and specificity or PLR and NLR can be a disadvantage in comparing the performance of competing tests, especially if 1 test does not outperform the other on both indicators. The DOR can be used as a single indicator of diagnostic performance and combines the strengths of sensitivity and specificity and of PLR and NLR and is independent on prevalence (13). Regarding the DOR, the performances of the IgA- and IgG-anti-nGli were worst.
For further evaluation of the tests, calculation of predictive values at a defined pretest probability is useful. In contrast to screening conditions with a prevalence of approximately 1%, in paediatric gastroenterology centres, a higher pretest probability must be assumed. At a pretest probability of CD of approximately 0.05, all of the tests except the assays for IgA- and IgG-anti-nGli result in posttest probabilities of at least 0.52 (Fig. 1). The posttest probability of IgA- and IgG-anti-nGli tests was <0.4. A high positive predictive value reduces the number of patients who need to undergo endoscopy. The negative predictive value is high for all of the tests (≥0.99), except for IgG-anti-tTG (0.97). A high negative predictive value ensures that the number of patients with CD among the children with negative test results is at minimum (at most 1%).
Thus, from all of the tests investigated, only assays for IgG-anti-dGli, IgA-anti-tTG, and IgA-EmA had high specificity (≥0.96) connected with high sensitivity (≥0.86), with high positive predictive values (≥0.52 and ≥0.69 at pretest probabilities of 0.05 and 0.1, respectively) and negative predictive values (≥0.99 and ≥0.98 at pretest probabilities of 0.05 and 0.1, respectively). Therefore, these assays should be preferred over tests measuring anti-nGli.
A selection bias in our study cannot be excluded. First, biopsies are more likely to be performed in symptomatic patients with positive results for CD-specific antibodies such as IgA-EmA and IgA-anti-tTG, which favour these tests. Furthermore, most children in the present study had not been rechallenged with gluten to confirm the diagnosis of CD, which was required according to the previous ESPGHAN criteria on the diagnosis of CD (14). Recently it was suggested that routine gluten challenge in patients younger than 2 years is not necessary when patients have villous atrophy in combination with positive EmA (7). According to ESPGHAN guidelines, a later gluten challenge should be performed only if villous atrophy was found in children younger than 2 years who are negative for CD-specific antibodies to confirm CD as a cause of the enteropathy (15). Six of our patients were classified as coeliacs in the absence of increased IgA-anti-tTG (however, one of them was positive for IgA-EmA and this patient and another positive for routine assay of IgA-anti-tTG outside the present study). The latter finding shows that occasionally diverse results may be obtained when tests of different suppliers are compared (16). In fact, none of the remaining 4 patients was rechallenged with gluten until now, mainly due to their young age. According to the ESPGHAN suggestions (15), such a provocation should not be performed before the age of 5 years. Among the 4 IgA-anti-tTG–negative patients with CD, there was 1 child with selective IgA deficiency. Interestingly, in this child, not only IgA-anti-tTG but also IgG-anti-dGli were negative. Another of the 4 patients was positive 64 days before but not at the day of endoscopy. The latter patient is suspected to have started the GFD too early. Due to the histological findings and recovery of the children under GFD, we must assume but cannot guarantee that the IgA-anti-tTG–negative patients with CD are correctly classified. Mucosal lesions due to infections or allergy may have been considered CD. Confirmation or exclusion of the diagnosis of CD is therefore needed and should be performed using clinical follow-up, exclusion of IgA deficiency, HLA analysis, and/or gluten provocation. Our finding of a high percentage (14%) of IgA-anti-tTG–negative children with CD (Table 1) is in line with the observation of a lower sensitivity of antibody assays in young children (1,6,8,9).
Although we included one of the largest number of young children published so far, the study is not powered to detect potential differences between the tests with both high sensitivity and specificity (IgA-anti-tTG, IgA-EmA, IgG-anti-dGli).
In conclusion, our results do not support the use of anti-nGli assays in the diagnosis of CD in children up to 2 years old. IgA-anti-tTG, IgA-EmA, and IgG-anti-dGli perform better than anti-nGli. Despite the good performance of these antibody tests, duodenal biopsy still represents the criterion standard to which these tests have to be compared, especially in young children, with the prospect of a lifelong and life-altering diagnosis.
1. Prause C, Richter T, Koletzko S, et al. New developments in serodiagnosis of childhood celiac disease: assay of antibodies against deamidated gliadin. Annals N Y Acad Sci
2. Prause C, Ritter M, Probst C, et al. Antibodies against deamidated gliadin as new and accurate biomarkers of childhood coeliac disease. J Pediatr Gastroenterol Nutr
3. Bürgin-Wolff A, Gaze H, Hadziselimovic F, et al. Antigliadin and antiendomysium antibody determination for coeliac disease. Arch Dis Child
4. Ghedira I, Sghiri R, Ayadi A, et al. Anticorps anti-endomysium, anti-réticuline et anti-gliadine, intérêt dans le diagnostic de la maladie coeliaque chez l’enfant. Pathol Biol
5. Tonutti E, Visentini D, Bizzaro N, et al. The role of antitissue transglutaminase assay for the diagnosis and monitoring of coeliac disease: a French–Italian multicentre study. J Clin Pathol
6. Lagerqvist C, Dahlbom I, Hansson T, et al. Antigliadin immunoglobulin A best in finding celiac disease in children younger than 18 months of age. J Pediatr Gastroenterol Nutr
7. Wolters VM, van de Nadort C, Gerritsen SAM, et al. Is gluten challenge really necessary for the diagnosis of celiac disease in children younger than age 2 years? J Pediatr Gastroenterol Nutr
8. Maglio M, Tosco A, Paparo F, et al. Serum and intestinal celiac disease-associated antibodies in children with celiac disease younger than 2 years of age. J Pediatr Gastroenterol Nutr
9. Basso D, Guariso G, Fogar P, et al. Antibodies against synthetic deamidated gliadin peptides for celiac disease diagnosis and follow-up in children. Clin Chem
10. Barbato M, Maiella G, Di Camillo C, et al. The anti-deamidated gliadin peptide antibodies unmask celiac disease in small children with chronic diarrhoea. Dig Liver Dis
11. Marsh MN. Gluten, major histocompatibility complex, and the small intestine. A molecular and immunobiologic approach to the spectrum of gluten sensitivity (“celiac sprue”). Gastroenterology
12. Oberhuber G, Granditsch G, Vogelsang H. The histopathology of coeliac disease: time for a standardized report scheme for pathologists. Eur J Gastroenterol Hepatol
13. Glas AS, Lijmer JG, Prins MH, et al. The diagnostic odds ratio: a single indicator of test performance. J Clin Epidemiol
14. 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
15. Husby S, Koletzko S, Korponay-Szabó IR, et al. European Society of Paediatric Gastroenterology, Hepatology, and Nutrition guidelines for the diagnosis of coeliac disease. J Pediatr Gastroenterol Nutr
16. van Meensel B, Hiele M, Hoffman I, et al. Diagnostic accuracy of ten second-generation (human) tissue transglutaminase antibody assays in celiac disease. Clin Chem
Keywords:Copyright 2012 by ESPGHAN and NASPGHAN
antibodies; coeliac disease; deamidated gliadin; diagnosis; tissue transglutaminase