Celiac disease (CD) is a common chronic autoimmune disorder with a prevalence around 1% in Europe (1) and as much as 3% in a recent Swedish screening study (2). The diagnosis is based on a combination of serology tests and pathological mucosa from intestinal biopsies (3,4). Immunoglobulin A (IgA) tissue transglutaminase (tTG) is considered the best test with sensitivity and specificity of >95% but performs less well in children younger than 2 years and IgA-deficient individuals (5). The combined IgG/IgA deamidated gliadin antibody test (IgG/IgA-deamidated gliadin peptide antibodies [DGP]) or the IgG-DGP also performs well with both high sensitivity and specificity (5–7). DGP has replaced antibodies against native gliadin (8) because of much better test characteristics. DGP has been proposed to be a good complement to tTG (9–11), especially in case of IgA deficiency (12).
Despite excellent test characteristics for DGP in research laboratories, evaluations in everyday clinical practice of the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV), as well as the cost of introducing the test, are lacking. In 2008, analyses against DGP were introduced into routine medical care at the immunological laboratories in Stockholm (IgG-DGP) and Gothenburg (IgG/IgA-DGP), the 2 largest cities in Sweden, as a complement to IgA-tTG.
We conducted a population-based chart review of test characteristics, as used in clinical practice, to establish whether DGP test kits had improved the workup for CD in children and to what cost.
All of the consecutive patients younger than 18 years who were tested for tTG and/or DGP and underwent duodenal biopsy because of suspected CD in Stockholm and Gothenburg (roughly one-third of Sweden's population according to Statistics Sweden; www.scb.se) between September 1, 2008, and March 31, 2010, were included (n = 630, Fig. 1). Medical records were reviewed for clinical presentation and duodenal biopsy pathology report. Serology data were obtained from the laboratory report databases at the departments of clinical immunology of the participating hospitals.
Children were excluded if the CD serology preceding the duodenal biopsy had not been performed in one of the departments of clinical immunology of the participating hospitals (n = 93 [15%], Fig. 1). If CD serology had been performed in other laboratories, it would have been difficult to confirm test values and type of test kit used.
Definition of CD
All of the participants had been referred to a pediatric gastroenterology unit because of suspected CD and underwent duodenal biopsy under general anesthesia. It was mandatory in all of the participating centers to take 4 biopsies from the distal duodenum and 2 biopsies from the duodenal bulb. CD was defined as Marsh 3 (total villous atrophy, subtotal villous atrophy, or partial villous atrophy) (13). Two individuals had crypt hyperplasia (Marsh 2) and were also regarded as having CD. They both had gastrointestinal symptoms and positive serology titers. Pathology reports of Marsh 1 (intraepithelial lymphocytosis) were not regarded specific enough to represent CD, irrespective of serology results (13). Individuals with Marsh 0 (normal mucosa) were regarded as not having CD (13).
Serology Tests Used
The serology test kits and cutoff values used (cutoff values as suggested by the manufacturers were used) are summarized in the online-only Supplemental Digital Content 1 (http://links.lww.com/MPG/A149). In Stockholm, the tests GAFx3 enzyme linked immunosorbent assay (ELISA) (Euroimmun Medizinische Labordiagnostika AG, Lübeck, Germany) and Celikey, ELISA (Phadia AB, Uppsala, Sweden) were used. In Gothenburg QUANTA Lite Celiac DGP Screen ELISA (INOVA Diagnostics Inc, San Diego, CA), and Celikey, ImmunoCAP (Phadia AB, Uppsala, Sweden) were used.
Direct medical costs for serology analyses were based on figures provided by the economy units of the participating immunology departments. The mean cost (or charge) that health care providers had actually paid for the serology analyses during the study period and the total number of analyses performed in children during the study period were provided.
Direct medical costs of duodenal biopsies in general anesthesia were based on figures provided by the economy units of the participating gastroenterology and anesthesiology departments. Direct costs included physician costs, pathologist costs, facilities, and/or hospitalization costs for anesthetic and endoscopic procedures. We did not include indirect patient costs (time lost from parents’ work or transportation costs) in the model.
Sensitivity, specificity, PPV, and NPV for IgA-tTG and IgG-DGP used alone or in combination were calculated for children undergoing duodenal biopsy. The number of gastroscopies owing to false-positive serologies was also calculated. Receiver operating characteristics (ROC) curves were plotted. The number of CD cases that would likely have been missed if DGP had not been included in the CD workup was calculated assuming that a negative tTG would not have led to a duodenal biopsy in the absolute majority of tested children.
Cost per diagnosed case of CD was calculated in 2 scenarios:
1. Costs for all of the pediatric analyses of tTG and DGP performed in the participating hospitals during the study period and costs for all of the gastroscopies performed in the study population were divided by all of the new cases of CD.
2. In an alternative scenario, the number of cases that would have been missed without the use of DGP was subtracted from the total number of CD cases in the study. We excluded all of the costs for DGP in the population (approximation based on costs for DGP in Stockholm) and all of the costs of gastroscopies of children with a positive DGP and a negative tTG. Total cost for serologies and endoscopies in this scenario was divided with cases that would have been detected even without use of DGP.
Five hundred thirty-seven children underwent duodenal biopsy and fulfilled criteria for participation, of which 278 (52%) had CD (Fig. 1). Age ranged from 12 months to 18 years (mean 7.6 years); 71 (13%) were younger than 2 years and 308 (57%) were girls (Table 1). Specific IgA values were available in 366 children, of whom 16 (4%) had IgA deficiency (Table 1). Symptoms preceding workup for suspected CD were gastrointestinal complaints (n = 358 [66%]), nongastrointestinal symptoms (n = 117 [22%]), and screening of risk groups (eg, heredity for CD, diabetes mellitus, or Down syndrome, and so on, n = 62 [12%]) (Table 1).
Diagnostic Characteristics of tTG and DGP in the Total Sample
tTG had been analyzed in 530 (99%) of children and DGP had been analyzed in 408 (76%) (Fig. 1). The ROC curve was significantly better for tTG compared with DGP (Fig. 2). The ROC curve restricted to children younger than 2 years was based on few individuals (n = 71) and should be interpreted with caution. DGP was not superior to tTG among children younger than 2 years (online-only Supplemental Digital Content 2, http://links.lww.com/MPG/A150). In post-hoc analyses, we also made ROC curves for children younger than 3, 4, 5, 6, and 7 years, all of which showed a tendency to tTG being better than DGP; however, ROC curves in all of the age groups had narrowly overlapping confidence intervals, and results should still be interpreted with caution (online-only Supplemental Digital Content 2, http://links.lww.com/MPG/A150). Among children undergoing duodenal biopsy for suspected CD, sensitivity, specificity, PPV, and NPV for tTG were 94%, 86%, 88%, and 93%, respectively (Table 2). Corresponding values for DGP were 91%, 26%, 51%, and 78% (Table 2). If intermediate values had been regarded as negative, the sensitivity, specificity, PPV, and NPV would have been 91%, 91%, 92%, and 90% for tTG (Table 2) and 88%, 44%, 57%, and 81% for DGP (Table 2). If intraepithelial lymphocytosis had been regarded as CD, the sensitivity, specificity, PPV, and NPV would have been 86%, 92%, 94%, and 88% for tTG (Table 2) and 90%, 26%, 57%, and 74% for DGP (Table 2). tTG performed better than DGP, regardless of test kits used (online-only Supplemental Digital Content 3, http://links.lww.com/MPG/A151; and online-only Supplemental Digital Content 4, http://links.lww.com/MPG/A152) or cutoff levels chosen (Fig. 2 and online-only Supplemental Digital Content 2–4, http://links.lww.com/MPG/A150, http://links.lww.com/MPG/A151, http://links.lww.com/MPG/A152).
There were 149 children who were tTG negative and DGP positive (ie, children who would likely not have undergone duodenal biopsy if DGP had not been used; Fig. 1). In this subgroup, only 8 of 149 had villous atrophy, corresponding to a PPV of 5%, or 20 negative duodenal biopsies per new case of CD. In case intermediate values would have been regarded as negative, PPV for a positive DGP together with a negative tTG would have been 6% (7/115). All of the children undergoing duodenal biopsy in spite of negative serologies (tTG or DGP alone or both combined) had gastrointestinal symptoms suggestive of CD. The 3 children in this group who had villous atrophy improved on a gluten-free diet and had a normal mucosa on control biopsy. Patient characteristics of children who were only tested for either tTG (n = 129) or DGP (n = 7) did not differ markedly from the total sample, except for the fact that only 6 of 136 children were younger than 2 years in this group (data not shown).
No false-positives were found among the 143 (53%) CD cases who had a tTG value >80 U/mL (ie, 10 times the cutoff for a clearly positive value stated by the manufacturer). False-positive values of DGP ranged from intermediate values to >200 U/mL (data not shown).
Diagnostic Characteristics of Children Younger Than 2 Years and of IgA-deficient Children
Among the 71 children younger than 2 years with normal IgA, the proportion of positive serologies representing CD (ie, PPV) were 96% (25/26) for tTG and 48% (24/50) for DGP (Table 3). No patient with CD younger than 2 years would have been missed if DGP had not been used in this sample (1 child younger than 2 years had a negative tTG, but also a negative DGP and had been biopsied because of extensive symptoms). In 16 IgA-deficient children, 11 were DGP-positive, of which 5 had CD, corresponding to a PPV of 45%.
During the study period, 26,180 unique children underwent serologic testing (tTG and/or DGP) for suspected CD in the participating study centers. Cost of all serologic testings for CD in children of all ages during the study period, when tests for both tTG and DGP were available, was €51 per test kit. If DGP would not have been available, the price would have been €42 per test kit.
The mean cost of endoscopy in general anesthesia during the study period was €1100. The total cost for CD work-up in the study population was €1,925,840, or €6878 per new case of CD (total cost of all serologies and all duodenal biopsies divided by all new cases of CD). If DGP would not have been available, the total workup cost would have been lower, but resulted in more missed cases of CD. In a scenario without DGP, 8 cases of CD could have been missed (for patient characteristics, see Fig. 1). In this setting, the cost of using DGP to find these 8 cases of CD, that would possibly otherwise have been missed, amounts to €399,520 or €49,940 per extra CD case.
We also compared different test strategies among the 401 children who had been tested for both tTG and DGP. The consequences of different testing strategies with regard to number of CD cases missed, number of unnecessary duodenal biopsies, and cost per detected CD case are summarized in Table 4. A test strategy requiring both a positive tTG and a positive DGP rendered the best specificity and the lowest cost (Table 4). A strategy requiring either a positive tTG or a positive DGP rendered the highest sensitivity and the highest cost (Table 4). The use of tTG alone was the best tradeoff between sensitivity, specificity, and cost (Table 4).
If duodenal biopsies would not have been performed in the 143 children with tTG >80 U/mL, total cost of workup for CD during the study period would have decreased by €157,300 (8%).
We reviewed performance of tTG and DGP as used when diagnosing consecutive cases of CD in clinical practice. Of 278 cases, 8 could possibly have been missed if only tTG but not DGP had been available. The cost of detecting the extra 8 cases by introducing DGP was considerable in terms of number of unnecessary duodenal biopsies and money. DGP was not superior to tTG in children younger than 2 years, and was of questionable value in IgA-deficient children.
To our knowledge, this is the first evaluation of the performance and cost of DGP as prospectively tested in clinical practice in an unselected population seeking medical care. We were able to include all of the consecutive patients who underwent duodenal biopsy during the CD workup in the 2 largest Swedish cities, resulting in a large cohort for study. We had access to all medical chart data including original laboratory results and biopsy reports for all of the study participants as well as cost data for serology analyses and gastroscopies. Our findings of good test performance for tTG and poor test performance for DGP were consistent, regardless of analysis method used and regardless of which of the 2 independent and officially certified clinical laboratories that had carried out the analyses (online-only Supplemental Digital Content 3, http://links.lww.com/MPG/A151; and online-only Supplemental Digital Content 4, http://links.lww.com/MPG/A152).
Our study also has limitations. The absolute majority of children with negative serologies in primary care was naturally not referred for further workup and hence did not undergo duodenal biopsies. Therefore, we were able to evaluate the diagnostic performance in clinical practice post primary care screening, whereas diagnostic performance at the primary care level could not be calculated. Despite the relatively large size of the present study, the subanalyses of DGP performance in children younger than 2 years and in IgA-deficient children are based on small numbers, and those results should be interpreted with caution. Misclassification of CD status based on histological samples is possible (14), but because duodenal biopsies were reviewed by experienced pathologists in routine medical care, with no knowledge of the present study or its hypothesis, any misclassification of CD status is most likely nondifferential.
In a few earlier studies of tTG and DGP, consecutive patients with villous atrophy have been compared with patients with normal duodenal biopsies (6–10) and/or to healthy, unbiopsied controls (6,15). In all of the listed studies, serology analyses have been performed retrospectively in research laboratories, and performance of IgA-tTG and IgG-DGP has been excellent with mostly both high sensitivity (60%–99%) and specificity (80%–100%) regardless of test kit, but dependent on cutoff limits used. Most studies, including ours, used cutoff values as suggested by the manufacturers.
Our findings that performance of tTG is extremely good, but that sensitivity can be increased if tTG and DGP are combined (positive tTG or DGP required) are in concordance with earlier studies (7,9,11,15); however, the relatively poor test performance of DGP found in our study, with a large proportion of false-positives has only been described once before in a study of 116 children with high risk for CD (16). In that study, as well as in ours, serology samples were analyzed prospectively in standard medical care as opposed to all of the earlier studies, which may be part of the explanation for the dramatic difference in results regarding DGP specificity.
Earlier studies have concluded that DGP does not outperform tTG (not even in children younger than 2 years) (8,17), but that the combination of tTG and DGP has higher sensitivity than tTG alone (15), which is also in concordance with our findings. Some studies have suggested that DGP should be included as a complement to tTG in future diagnostic algorithms for CD (9–11), but monetary costs and unnecessary duodenal biopsies have hitherto not been investigated. Given the considerable costs and the high proportion (42%) of unnecessary biopsies when including DGP in the diagnostic algorithm, as we have shown in the present study, we support the recommendation that IgG-DGP is used only in children with IgA deficiency (5).
Our findings that there were no false-positive tTG values >80 U/mL and that it is potentially possible to safely exclude the biopsy in a subset of patients with suspected CD are in line with revised guidelines regarding CD (4).
In conclusion, tTG was superior to DGP, even in children younger than 2 years. IgG-DGP was of questionable value even in IgA-deficient children. The introduction of DGP as a complement to tTG for the workup of suspected CD in children rendered few extra cases of CD and generated considerable costs and many unnecessary gastroscopies performed under general anesthesia.
1. Dubé C, Rostom A, Sy R, et al. The prevalence of celiac disease in average-risk and at-risk Western European populations: a systematic review. Gastroenterology 2005; 128:S57–S67.
2. Myléus 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 2009; 49:170–176.
3. Ribes-Koninckx C, Mearin M, Korponay-Szabó I, et al. Coeliac disease diagnosis: ESPGHAN 1990 criteria or need for a change? Results of a questionnaire. J Pediatr Gastroenterol Nutr 2012; 54:136–160.
4. 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 2012; 54:136–160.
5. Leffler DA, Schuppan D. Update on serologic testing in celiac disease. Am J Gastroenterol 2010; 105:2520–2524.
6. Agardh D. Antibodies against synthetic deamidated gliadin peptides and tissue transglutaminase for the identification of childhood celiac disease. Gastroenterology 2007; 5:1276–1281.
7. Monzani A, Rapa A, Fonio P, et al. Use of deamidated gliadin peptide antibodies to monitor diet compliance in childhood celiac disease. J Pediatr Gastroenterol Nutr 2011; 53:55–60.
8. 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 2009; 49:52–58.
9. Basso D, Guariso G, Bozzato D, et al. New screening tests enrich anti-transglutaminase results and support a highly sensitive two-test based strategy for celiac disease diagnosis. Clin Chim Acta 2011; 412:1662–1667.
10. Rozenberg O, Lerner A, Pacht A, et al. A new algorithm for the diagnosis of celiac disease. Cell Mol Immunol 2011; 8:146–149.
11. Brusca I, Carroccio A, Tonutti E, et al. The old and new tests for celiac disease: which is the best test combination to diagnose celiac disease in pediatric patients? Clin Chem Lab Med 2011;50:111–7.
12. Villalta D, Tonutti E, Prause C, et al. IgG antibodies against deamidated gliadin peptides for diagnosis of celiac disease in patients with IgA deficiency. Clin Chem 2010; 56:464–468.
13. Ludvigsson JF, Green PH. Clinical management of coeliac disease. J Intern Med 2011; 269:560–571.
14. Corazza GR, Villanacci V, Zambelli C, et al. Comparison of the interobserver reproducibility with different histologic criteria used in celiac disease. Gastroenterology 2007; 5:838–843.
15. Jaskowski TD, Donaldson MR, Hull CM, et al. Novel screening assay performance in pediatric celiac disease and adult dermatitis herpetiformis. J Pediatr Gastroenterol Nutr 2010; 51:19–23.
16. Parizade M, Bujanover Y, Weiss B, et al. Performance of serology assays for diagnosing celiac disease in a clinical setting. Clin Vaccine Immunol 2009; 16:1576–1582.
17. 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 2009; 55:150–157.