See “Screening for Celiac Disease” by Hill on page 414; “Case Finding for Celiac Disease Is Okay, But Is It Enough?” by Catassi and Leonetti on page 415; and “Epidemiology of Coeliac Disease and Comorbidity in Norwegian Children” by Størdal et al on page 467.
Celiac disease (CD) is an immune-mediated enteropathy in which gluten-related peptides trigger an immune response in some individuals carrying human leukocyte antigen (HLA) risk alleles (1,2). The disease may present with a wide range of symptoms resulting in often delayed or absent diagnosis (3–5) with negative effect on quality of life (6,7). This has resulted in a debate concerning general CD screening (8–10). Serological markers indicative of CD have been available since the early 1980s. Enzyme-linked immunosorbent assay (ELISA) tests of IgA antibodies directed against the CD autoantigen tissue transglutaminase-2 (tTG) are most commonly used today. In pediatric populations, high tTG-IgA levels are highly suggestive of CD (11,12). The reliability of tTG-IgA in a mass-screening situation, that is, when targeting the general population, has not been studied.
The present study is part of the Exploring the Iceberg of Celiacs in Sweden (ETICS) study, a population-based CD mass screening among 12-year-old Swedish children (4,13). We evaluated a serological screening strategy based on the determination of serum IgA (s-IgA), tTG-IgA, tTG-IgG, endomysial antibodies (EMA) IgA, and EMA-IgG and also the possible role of sequential analysis of HLA risk alleles and/or EMA-IgA in a mass screening.
During 2005 and 2006, all 12-year-olds in 5 regions in Sweden were invited to the population-based CD mass screening study (the study was performed in cooperation with the county councils of Västerbotten, Stockholm, Östergötland, Kronoberg, and Skåne, and the study was undertaken within the Centre for Global Health Research at Umeå University). In total, 10,041 children were invited, and informed consent was obtained from their parents before blood samples were collected. Blood samples from 7208 children without previously known CD were collected (48% boys and 52% girls). Details of the study are presented elsewhere (4,13).
All of the samples were analyzed for serum tTG of isotype IgA (tTG-IgA) and total s-IgA. If there was a low level of s-IgA (<0.5 g/L), tTG of isotype IgG (tTG-IgG) was analyzed. In all children with tTG-IgA >4 U/mL or tTG-IgG >6 U/mL and those with borderline levels of tTGIgA (2–4 U/mL) or tTG-IgG (3–6 U/mL) in combination with EMA positivity of the respective isotypes, a small intestinal biopsy was recommended. Criteria and the outcomes are depicted in Figure 1.
In addition, blood samples from all children with tTG-IgA levels between 4 and 20 U/mL were analyzed for EMA-IgA after the screening to evaluate whether the sequential analysis of EMA would make CD screening more efficient.
Analysis of HLA alleles was performed in all 192 children in whom an intestinal biopsy was recommended and in a random selection of 1167 children without a CD diagnosis. The referents were frequency matched for sex and study site at the group level.
Serum levels of tTG-IgA and IgG were measured by ELISA (Celikey Phadia, Freiburg, Germany). All serum samples were diluted 1:101, and the antibody levels, expressed as arbitrary units (units per milliliter), were calculated from a 6-point calibrator curve. Intraassay variability for tTG-IgA was 4.9% to 8.7% and for tTG-IgG it was 3.6% to 7.2%. Interassay variability was 3.2% to 4.5% and 1.5% to 4.9%, respectively.
EMA-IgA and EMA-IgG were measured with an indirect immunofluorescence technique using tissue sections from marmoset monkey esophagus mounted on glass slides (Euroimmun, Lübeck, Germany). Sera yielding fluorescent binding to the endomysial structure in a 1:5 dilution were regarded as positive and were further diluted to determine the final titer at which fluorescence was detected.
Total s-IgA was analyzed using a routine nephelometric method according to the manufacturer's instructions for use (BN Pro Spec System, Dade Behring, Marburg GmbH, Germany). Children with serum levels <0.06 g/L were considered IgA deficient, whereas values <0.5 g/L were considered as low s-IgA.
DNA extracted from whole blood was analyzed for detection of alleles encoding for the HLA class II DQ subunits. The alleles were detected using an EU-DQkit (Eurospital SpA, Trieste, Italy) according to the manufacturer's instructions.
Case Ascertainment and Diagnostic Criteria
All children fulfilling any of the serological criteria (Fig. 1) were invited to clinical follow-up by a pediatrician and were recommended to undergo further investigation with a small intestinal biopsy, either by oral capsule or upper endoscopy. Children with a normal biopsy taken by oral capsule but with persistent elevated levels of serological markers were re-investigated by upper endoscopy as described earlier (13). Biopsy specimens were initially diagnosed by a pathologist at the local hospital and later in a blinded fashion by 1 pathologist, in common for all cases, with expertise in gastrointestinal pathology. If there was disagreement, a third pathologist with expertise in gastrointestinal pathology reviewed the biopsies, and a consensus diagnosis was achieved. Criteria for CD diagnosis were a small intestinal mucosa with villous atrophy (Marsh 3a–c) or a borderline mucosa (Marsh 1–2), that is, >30 intraepithelial lymphocytes per 100 enterocytes, and the latter had to be in combination with symptoms and/or other signs compatible with CD and with clinical and serological response to a gluten-free diet (13). In total, 192 children fulfilled serological criteria for biopsy and a small intestinal biopsy was performed in 184. Among these, 153 new cases were identified, as reported previously (4,13).
Data handling and statistical analyses were performed with SPSS version 19 (SPSS Inc, Chicago, IL). Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated for different cutoffs of tTG-IgA and for different hypothetical combinations of serological markers, followed by analysis of HLA risk alleles in individuals with positive serology and EMA-IgA, if tTG is <20 U/mL. Receiver operating characteristic (ROC) curves for tTG-IgA were plotted. This study was approved by the ethical review board of Umeå University, Umeå, Sweden.
Characteristics of Serological Tests and Genetic Markers
s-IgA was analyzed in 7161 children and 170 (2.4%) had total s-IgA <0.5 g/L. Selective IgA deficiency (defined as total s-IgA <0.06 g/L) was found in 27 children (0.4%). Of those, 2 (7.4%) were later found to have CD. The prevalence of IgA deficiency among screening-detected patients with CD in our population was 1.3%. Median s-IgA was 1.38 g/L (interquartile range 1.03–1.82).
The median tTG-IgA level in cases with CD (n = 153) was 25.2 U/mL (interquartile range 9.13–62.9) and it was 0.10 U/mL (interquartile range 0.01–0.25) in the remaining children without CD (n = 7055). The manufacturer's recommended cutoff for tTG-IgA was 5 U/mL, and by lowering the cutoff to 4 U/mL (see also Fig. 1), we identified 32 additional children who were recommended to undergo a small intestinal biopsy, 17 of whom received a CD diagnosis. Eight children with tTG-IgA ranging from 2.45 to 101 U/mL declined biopsy.
The area under the ROC curve calculated for the study population of 7208 children was 0.988 for the tTG-IgA test. Optimal cutoff calculated from the ROC curve was 3.0 U/mL. Figure 2 depicts the outcome of investigation for CD with small intestinal biopsy in relation to the levels of tTG-IgA and tTG-IgG. In the group with an elevation of tTG-IgA >10 times the upper limit of normal (>50 U/mL), all individuals received a CD diagnosis. There were 3 cases with tTG-IgA approximately 30 U/mL with normal mucosa (range 29–32), whereas all children with tTGIgA >32 U/mL were diagnosed as having CD. Negative results of small intestinal biopsy are common at lower levels of tTG-IgA. Only 2 cases with CD (40%) of 5 biopsied were diagnosed in the group as having low s-IgA; both were diagnosed as having selective IgA deficiency (s-IgA <0.06 g/L). Ten children had their diagnosis based on Marsh 1 enteropathy. Median tTG-IgA in this group was 31 U/mL (range 5–101). Their characteristics have previously been reported in the Journal(13).
All screening-detected cases with CD carried HLA-DQ2 or HLA-DQ8 or both. DQ2 was most commonly related to CD in both boys and girls. Among the controls, 55.7% of the boys and 50.4% of the girls possessed the CD-associated HLA-DQ haplotype. The prevalences of the different haplotypes in cases with CD and in controls are summarized in Table 1.
Accuracy of Different Hypothetical Screening Strategies
To evaluate the hypothetical diagnostic accuracy of different levels of tTG cutoffs, both isolated and in combination with sequential testing for HLA-DQ alleles, or EMA-IgA, we calculated the sensitivity, specificity, PPVs, and NPVs (Table 2). All combinations had robust accuracy, with sensitivity ranging from 87.6% to 100%, specificity from 99.5% to 99.8%, PPV from 79.7% to 89.7%, and NPV from 99.7% to 100% (Table 2). Efforts to increase sensitivity by applying a low tTG-IgA cutoff resulted in an increased number of negative small intestinal investigations and lower PPV. With a tTG-IgA cutoff of 4 U/mL, adding sequential testing of HLA risk alleles would have reduced the number of negative and unnecessary small intestinal investigations by 2 and also reduced the number of families declining biopsy by 2. Requiring EMA positivity when tTG-IgA is <20 U/mL with the same tTG-IgA cutoff would have reduced the number of negative investigations by 8 and also reduced the number of families declining biopsy by 2, but 2 cases with CD would have been missed.
Our results show that tTG-IgA is a robust and reliable tool for identifying CD in a mass screening situation. By adding supportive sequential testing for EMA-IgA or alleles encoding for HLA-DQ, performance is improved. All identified cases with CD carried at least 1 CD-associated HLA-DQ haplotype, and the prevalence of CD risk alleles among controls was also high (>50%).
In this study, approximately 10% of Swedish children born in 1993 were screened for CD. Given the distribution of study sites involved, we consider this study representative for Swedish sixth graders at large. It is possible that we did not identify all of the cases with CD in the screened population because we cannot rule out that there could be cases with CD with tTG-IgA <2 U/mL. It is known that there are a few seronegative cases with CD that will not be identified by screening with conventional serological markers (14); however, given the large sample size and the prioritized sensitivity in our screening strategy, false-negative cases would have had a low effect on the estimated values for specificity, sensitivity, and the predictive values. It would have been interesting also to test all children for EMA-IgA to see whether there were tTG-IgA–negative but EMA-IgA–positive cases with CD. Furthermore, 8 children with tTG-IgA ranging from 2.45 to 101 U/mL declined biopsy. Several studies have pointed out the difficulty in identifying cases with CD (3–7), which has resulted in a discussion about the relevance of CD mass screening. Nevertheless, many questions remain; the most important concerns what benefits individuals with “silent CD” would gain from starting and maintaining treatment with a gluten-free diet. It was beyond the scope of this article to consider all possible benefits and drawbacks of CD mass screening. Instead, our aim was to evaluate the accuracy of different screening tools and screening strategies based on the outcome of the Swedish ETICS study. Two main screening strategies have been proposed (8). In one approach, the starting point is identification of individuals at genetic risk for developing CD, thus those carrying the necessary HLA-DQ2 and/or HLA-DQ8 alleles. According to this strategy, those with a genetic susceptibility are further tested with serological markers. Theoretically this approach is suitable if a repeated screening is planned. With the other approach, the whole population is screened with serological markers. Both strategies include a mandatory verifying small intestinal biopsy, which also is in concordance with the screening of “asymptomatic” CD risk groups in the updated European Society for Pediatric Gastroenterology, Hepatology, and Nutrition CD guidelines (15).
In the present study, we started with the analysis of s-IgA and tTG-IgA antibodies as recommended in the present guidelines of the European Society for Pediatric Gastroenterology, Hepatology, and Nutrition and the North American Society for Pediatric Gastroenterology, Hepatology, and Nutrition. Previous economic evaluations have found that analysis of total s-IgA is not cost-effective in mass screening for CD (16), and this view is supported by our findings because only 2 cases with CD were identified at the cost of analyzing s-IgA in 7208 children. We found that tTG-IgA levels >20 U/mL, and especially >50 U/mL (10 times the upper limit of normal recommended by the manufacturer), were highly predictive of CD diagnosis after small intestinal biopsy. At lower levels, some supportive sequential analysis should be used to reduce the number of unnecessary biopsies to a minimum, both to save resources and to minimize invasive investigations. In the present study, we analyzed EMA-IgA titers in children with tTG-IgA levels between 2 and 20 U/mL. Requiring EMA positivity would effectively reduce the number of negative small intestinal investigations, but cases would have been missed if this strategy had been adopted. An alternative would be to sequentially analyze alleles encoding for HLA-DQ. In our study, all cases with CD carried alleles encoding for either HLA-DQ2 or -DQ8 or both, and analysis did increase predictability, but to a lesser extent than analysis of EMA-IgA (notably without missing any case with CD). A further benefit with HLA analysis is that a negative result can be used to rule out the risk of developing CD in the future. Disadvantages of EMA analyses are that the method is time-consuming, analyzer dependent, and expensive in comparison with analysis of tTG-IgA (16). We used a lower cutoff and indeed identified several extra cases with CD, although the specificity in this group was lower. Whether this would be acceptable in mass screening is a matter for discussion because we have shown that this would result in many unnecessary small intestinal investigations. It is worth noting, however, that this finding implies that discrete elevations of tTG-IgA levels also need to be considered and that modestly elevated serum concentrations of tTG-IgA should not be used alone to exclude CD. In a clinical setting, it is important for the pediatrician to evaluate all available information, and in many cases, these children should be retested after a period of time. This also highlights the risk of missed diagnoses when commercially available CD tests can be bought and used freely without any evaluation by a physician.
It is generally assumed that the prevalence of CD risk alleles in Europe is between 30% and 40% (17,18). We found a significantly higher prevalence of >50% in our screened control population, which may contribute to the high CD prevalence in Sweden. Hence, starting with analysis of HLA risk alleles would consequently be less effective in a mass screening because the risk alleles are prevalent in the majority of the population. When choosing between the different screening strategies presented, both pose different practical and ethical dilemmas. Exposing the population to testing for genetic risk may result in difficulties for the individual to understand the implications of the test results, and because 30% to 55% of the population carry the HLA-DQ2 or -DQ8 risk alleles, and relatively few will develop CD, such a strategy may create a great deal of unnecessary anxiety among those carrying the risk alleles. Fewer individuals, however, would need repeated testing. Based on the study, it seems that a future screening program should be tailored to the characteristics of the specific target population.
In summary, elevated levels of tTG-IgA have good predictability for biopsy-verified CD diagnoses, and we believe that in a mass screening for CD, tTG-IgA would be a robust marker. We also conclude that some supportive sequential test should be used when tTG-IgA levels are low. Analyses of HLA risk alleles and of EMA-IgA titer have both their advantages and disadvantages.
We thank all of the children and their families, and the personnel working in the study: research nurses, laboratory personnel and administrative staff, and collaborators within the school health services and pediatric departments.
1. Sollid LM. Coeliac disease: dissecting a complex inflammatory disorder. Nat Rev Immunol
2. Ludvigsson JF, Leffler DA, Bai JC, et al. The Oslo definitions for coeliac disease and related terms. Gut
3. Bingley PJ, Williams AJ, Norcross AJ, et al. Undiagnosed coeliac disease at age seven: population based prospective birth cohort study. BMJ
4. Myleus A, Ivarsson A, Webb C, et al. Celiac disease
revealed in 3% of Swedish 12-year-olds born during an epidemic. J Pediatr Gastroenterol Nutr
5. Katz KD, Rashtak S, Lahr BD, et al. Screening for celiac disease
in a North American population: sequential serology and gastrointestinal symptoms. Am J Gastroenterol
6. Norstrom F, Lindholm L, Sandstrom O, et al. Delay to celiac disease
diagnosis and its implications for health-related quality of life. BMC Gastroenterol
7. Roma E, Panayiotou J, Karantana H, et al. Changing pattern in the clinical presentation of pediatric celiac disease
: a 30-year study. Digestion
8. Mearin ML, Ivarsson A, Dickey W. Coeliac disease: is it time for mass screening
? Best practice & research. Clin Gastroenterol
9. Fasano A. Should we screen for coeliac disease? Yes. BMJ
10. Evans KE, McAllister R, Sanders DS. Should we screen for coeliac disease? No. BMJ
11. Alessio MG, Tonutti E, Brusca I, et al. Correlation between IgA tissue transglutaminase antibody ratio and histological finding in celiac disease
. J Pediatr Gastroenterol Nutr
12. Dahlbom I, Korponay-Szabo IR, Kovacs JB, et al. Prediction of clinical and mucosal severity of coeliac disease and dermatitis herpetiformis by quantification of IgA/IgG serum antibodies to tissue transglutaminase. J Pediatr Gastroenterol Nutr
13. Webb C, Halvarsson B, Norstrom F, et al. Accuracy in celiac disease
diagnostics by controlling the small-bowel biopsy process. J Pediatr Gastroenterol Nutr
14. Abrams JA, Diamond B, Rotterdam H, et al. Seronegative celiac disease
: increased prevalence with lesser degrees of villous atrophy. Dig Dis Sci
15. Husby S, Koletzko S, Korponay-Szabo IR, et al. European Society for Pediatric Gastroenterology, Hepatology, and Nutrition guidelines for the diagnosis of coeliac disease. J Pediatr Gastroenterol Nutr
16. Shamir R, Hernell O, Leshno M. Cost-effectiveness analysis of screening for celiac disease
in the adult population. Med Decis Making
17. Tjon JM, van Bergen J, Koning F. Celiac disease
: how complicated can it get? Immunogenetics
18. Hadithi M, von Blomberg BM, Crusius JB, et al. Accuracy of serologic tests and HLADQ typing for diagnosing celiac disease
. Ann Intern Med