Celiac disease (CD) is caused by a permanent intolerance to ingested gluten in genetically susceptible individuals, leading to immunologically mediated inflammatory damage to the upper small intestine (1). The diagnosis of CD is based on the histological identification of gluten-sensitive enteropathy (2) together with the detection of anti-tissue transglutaminase antibodies (tTGA) and/or endomysial antibodies (EmA) (3–5). The only treatment for CD is lifelong adherence to a gluten-free diet (GFD) aiming at improving gastrointestinal and malabsorption symptoms as well as preventing complications. Under a GFD, decrease in EmA and tTGA IgA titers are generally observed in adherent patients (6–10). Periodical clinical evaluations together with serial tTGA IgA measurements are now recommended as a follow-up strategy to monitor compliance with a GFD (11,12).
Since the identification of tissue transglutaminase (tTG) as the major antigen of EmA (13), quantitative solid-phase enzyme-linked immunosorbent assay (ELISA) assays for the detection of tTGA have been developed. Guinea pig tTGs, the initial antigenic substrate used in these assays, have been replaced by recombinant human tTGs, providing second-generation assay sensitivities ranging from 70% to 100% and specificities ranging from 85% to 100% (4,5). Measurement of IgA antibody to human recombinant tTG is now considered the most efficient single serologic test for the detection of CD by the American Gastroenterological Association Institute (14). In addition, quantitative radiobinding assays (RBA) have been developed and have been reported to be more sensitive and specific than ELISAs (15,16). Such an RBA assay has been used in our center since the late 1990s, and we sought to retrospectively evaluate its performance in pediatric CD at diagnosis and also during monitoring of GFD. The aims of the present study were to compare our RBA-tTGA assay to one of the second-generation commercial ELISA kits, analyze the potential relation between tTGA levels as measured by RBA and histological severity of the disease at diagnosis, and confirm the time course of tTGA clearance during GFD.
A cohort of 80 pediatric patients with confirmed CD studied in the pediatric gastroenterology unit of our hospital since 1998 (median age at diagnosis 6.6 years, range 0.8–16.6 years; male/female sex ratio 28/52) were retrospectively studied. CD diagnosis was confirmed by CD serology and duodenal histology or, in patients with subnormal or noncontributive biopsy, on the basis of a clinical response to a GFD. A single child with a subnormal biopsy was diagnosed with latent CD based on strongly positive CD serology and a family history of CD. None of the patients showed immunoglobulin A (IgA) deficiency. Depending on the availability of biopsy sections and serum samples at diagnosis or during GFD, 3 partially overlapping subgroups of children with CD were studied. Comparison of RBA and ELISA tTGA IgA detection assays was performed in 28 children with CD for whom a diagnosis serum sample was available and in 150 serum samples from healthy controls.
A review of histological data at diagnosis was performed in 34 children with CD for whom RBA-tTGA IgA testing at diagnosis had been performed in our center. Among these patients, 24 presented with clinical symptoms, whereas 10 were asymptomatic.
Retrospective analysis of CD serology during GFD, initiated after diagnosis (n = 22) and/or an ill-tolerated gluten challenge (n = 17), was carried out in 34 patients who were all tTGA positive at diagnosis. The median follow-up time after initiation or reinitiation of GFD was 60 months (range 12–84 months). All of the patients were judged to be adherent to GFD by an experienced physician (J.S.) based on careful interrogation of the children and their parents, disappearance of gastroenterological symptomatology, and normalization of biological markers of intestinal absorption. They were considered to be in clinical remission from CD within 12 months following initiation of GFD.
Duodenal biopsies were reviewed by 2 pathologists blinded to the patients’ clinical history and serology results. Reporting was based on the modified Marsh criteria. A modified Marsh score ≥2 was reported as positive.
Detection of EmA IgA was performed using an indirect immunofluorescence assay including monkey esophageal tissue according to the manufacturer's instructions (Eurospital, Trieste, Italy). The presence of EmA was evaluated by 2 observers. The intensity of fluorescence observed in the muscularis mucosa was reported in a semiquantitative way (negative, 0.5+, 1+, 2+, 3+).
Detection of tTGA IgA was performed according to a previously described quantitative RBA assay procedure (15,17) that was based on the use of radiolabeled tTG produced in vitro. One microgram of full-length tTG cDNA subcloned into the pGEM3z vector (Promega, Charbonnieres, France) was used to generate 35S-tTG using a TNT SP6-coupled transcription/translation kit (Promega) according to the manufacturer's instructions and in the presence of 1 mCi/mL 35S-methionine (GE Healthcare, Orsay, France). Translated and labeled antigen was purified by chromatography on a NAP-5 column (GE Healthcare). For each analysis, 17,500 cpm of labeled antigen was incubated with 2 μL of each patient's serum at 4°C overnight in 96-well plates. IgA immune complexes were captured with 20 μL of goat anti-human IgA-agarose (Sigma, Lyon, France) for 1 hour at 4°C. After transfer of the reactions into 96-well MultiScreenHV filtration plates (Millipore, Saint Quentin Fallavier, France), captured complexes were washed 10 times using a MultiScreenHTS Vacuum Manifold. The amount of labeled antigen bound by specific autoantibodies was quantified using a microbeta trilux counter (EGC Instruments, Evry, France) after addition into each well of 10 μL of Wallac Optiphase Supermix scintillation cocktail. Each experiment was technically validated using a positive control (pooled positive sera 4500 cpm; optimal range 3500–4500 cpm; mean ± 2 standard deviations [SD] of 10 different runs) and 2 negative controls (pooled negative sera 75 cpm, optimal range 55–95 cpm; mean ± 2 SD of 10 different runs). Each sample was tested in duplicates. Results were expressed as the mean counts per minute of the duplicates. Positivity cutoff (mean counts per minute ± 3 SD found in the serum of healthy individuals) for the detection of tTG autoantibodies was 150 cpm. This cutoff was also validated in adult disease controls (lupus and autoimmune hepatitis). The RBA assay is a quantitative assay in which immune complexes are formed in excess of labeled antigen. Titration curves (serial dilution of test samples from 1:1–1:1024) are logarithmic (R2 > 0.98). At all of the dilutions, counts per minute ratios between test samples are conserved. Intra- and interassay variabilities of our RBA-tTGA IgA assay were 5.3% and 11%, respectively.
Diagnosis serum samples were also tested for tTGA IgA using a commercial ELISA assay according to the manufacturer's instructions (Eurospital). The results were defined as negative when <7 U/mL and positive if ≥7 U/mL.
Results were analyzed using Prism software version 5 (GraphPad Software Inc, La Jolla, CA).
Comparison of 2 Techniques for Detection of tTGA
We retrospectively compared our RBA-tTGA IgA assay with one of the second-generation ELISA kits using recombinant human tTG in 28 diagnosis serum samples from patients with confirmed CD and 150 samples from healthy controls. Both techniques showed extremely high specificity (98.7% for RBA and 100% for ELISA). All of the CD samples were found to be RBA-tTGA positive. A single child with CD showed negative detection of tTGA by ELISA but was EmA positive. These results correspond to a high sensitivity of both assays (100% for RBA and 96.4% for ELISA). Concordance with EmA was 89.2% for ELISA and 92.8% for RBA (2 samples were tTGA positive with RBA and ELISA but EmA negative). In the entire cohort of 80 children, the overall sensitivity of RBA-tTGA and EmA assays at diagnosis was 96.2% and 97.4%, respectively, with a concordance rate between the 2 assays of 93.6%.
CD Serology and Histology at Diagnosis
In 34 patients, we retrospectively analyzed the potential relation between tTGA IgA levels at diagnosis, before initiation of GFD, and the severity of histological lesions as assessed by blinded reevaluation of duodenal biopsies according to the modified Marsh score (2). All 34 patients displayed EmA antibodies. Thirty-one patients were tTGA positive, with median levels of 1827 cpm (range 290–5659). Three patients (8.8%) were negative for tTGA IgA (<150 cpm) and IgG (data not shown). In all 3, diagnosis biopsies showed partial (2 cases) or total (1 case) villous atrophy of the mucosa. Among tTGA-positive (and EmA-positive) patients, isolated increased numbers of intraepithelial lymphocytes in the absence of villous atrophy were observed in a single child diagnosed as having latent CD. Overall, as shown in Figure 1, there was no relation between tTGA IgA levels and the severity of histological lesions at diagnosis, nor between the histological score and the presentation at diagnosis (symptomatic vs asymptomatic patients, data not shown).
tTGA and EMA During Gluten-free Diet
Thirty-four tTGA-positive patients in clinical remission of CD were retrospectively analyzed for CD serology at diagnosis and/or after gluten challenge, and during primary or secondary GFD. In children whose presentation at diagnosis or relapse was typical (gastrointestinal symptoms), baseline tTGA IgA levels were found to be higher than in those with atypical presentation or minimal symptoms (3371±1729 cpm [n = 10] vs 1727±1294 cpm [n = 12], respectively; P = 0.0229). Higher levels were also observed in children showing only malabsorption symptoms than in those with atypical presentation (3480±1747 cpm [n = 9] vs 1727±1294 cpm [n = 12], respectively; P = 0.0428). Upon initiation of GFD, the rate of positivity for EmA IgA declined rapidly (Fig. 2A). At year 3, 86.7% of the children showed no detectable EmA. By contrast, the rate of positivity of tTGA IgA decreased more slowly, with 60% of patients still positive for tTGA at year 3. At year 5, all of the patients were EmA negative, whereas only 47.8% were tTGA negative. Of note, 5 tTGA-negative/EmA-positive follow-up samples (Fig. 2B) were in fact scored weakly positive (fluorescence intensity 0.5) for EmAs.
The levels of tTGA generally declined during GFD (Fig. 2B). A single child showed no significant modification of tTGA levels during 6 years of follow-up (ranging from 1958 at initiation to 1442 cpm), despite normalization of EmA at month 12, apparent good compliance with GFD, and clinical remission from CD. In the others, tTGA levels decreased rapidly during the first year of GFD (66% decrease at months 6 and 12) and more slowly thereafter (reaching 91% decrease at month 60). In a single patient, after an initial decrease between day 0 and month 6, tTGA levels reached a plateau (around 1100 cpm) that persisted throughout the 48-month follow-up period, whereas EmAs remained undetectable from month 18.
At initiation of GFD, tTGA levels were lower in children who became tTGA negative during GFD than in those who remained tTGA positive during follow-up (mean±SD 2080±1554 cpm vs 3688±1435 cpm, respectively, P = 0.0119). In the same vein, the median time for tTGA normalization was shorter in patients with initial tTGA levels <3000 cpm than in children with baseline tTGA >3000 cpm (19 and 58 months, respectively; P = 0.0004). Interestingly, the median time for EMA negativation in these 2 subgroups of patients did not significantly differ (12 months, if baseline tTGA <3000 cpm vs 18 months if tTGA >3000 cpm; P = 0.2, data not shown).
We sought to compare our RBA-tTGA assay to a commercial second-generation ELISA kit using recombinant human tTG. In a small series of 28 children with CD at diagnosis, we observed a high sensitivity of both ELISA and RBA assays. These results are in agreement with a previous evaluation of 6 different human tTG-based ELISA kits reporting sensitivities at diagnosis ranging from 71% to 100%, with an 88% sensitivity for the particular kit used in the present study (5). In addition, they corroborate the recent results of an international tTGA workshop showing that the highest sensitivity (93%) and specificity (100%) was achieved by RBA and that several ELISAs had the potential to reach RBA performances (16).
High rate of concordance between RBA-, ELISA-tTGA IgA and EmA assays was observed at CD diagnosis in our study; however, during GFD, using RBA, we found higher proportions of tTGA-positive and EmA-negative patients than previously reported with ELISA both in pediatric and adult CD. In fact, a dramatic decrease in ELISA-tTGA levels, and EMA seroreversion are regularly observed in the first months following initiation of GFD, leading to negative CD serology in compliant patients (8,17–20). In a 2007 study, it was shown that after 1 year on GFD, ELISA-tTGA IgAs were undetectable in 92% of 74 children (20). This is in striking contrast to the high rate of RBA-tTGA seropositivity found after 2 years of GFD in our study (80.6%) among children who were in clinical remission from CD. Our results are, however, in agreement with those reported by another group also using RBA for tTGA detection, and showing that by year 2 after starting GFD, 64.1% of adult patients with CD were still tTGA positive (21). A 2009 international tTGA workshop showed that RBA identified more frequently than ELISA the sera from patients with CD on GFD as positive (16). Although we did not directly compare both techniques during GFD, higher sensitivity of RBA, an assay based on the formation of antigen/antibody complexes in liquid phase and excess of antigen, may explain this discrepancy. In fact, RBA is best suited for the detection of low levels of high-affinity antigens, whereas ELISA is a capacity assay (22). RBA remains the criterion standard for sensitive detection of type 1 diabetes mellitus autoantibodies such as anti-glutamic acid decarboxylase and anti-insulin antibodies (23).
We did observe, like others, a continuous decrease in tTGA levels during GFD, but the time required for seronegativation (median 46 months) was longer than previously reported (7–9). Patients with extremely high tTGA levels (>3000 cpm) at diagnosis or relapse took longer to achieve normal levels. Overall, our observations made in children who were allegedly compliant with GFD, EmA negative within 18 months (median), and in clinical remission from CD within 12 months of GFD suggest that a decreasing trend in tTGA levels rather than absolute levels may be used as a surrogate marker of adherence to GFD. It is generally recognized that negative CD serology (based on EmA and ELISA-tTGA) may document absence of major dietary indiscretions but miss minor indiscretions both in children (24) and adults (25–27). The slow kinetics of RBA-tTGA clearance, presumably related to superior sensitivity of RBA at low concentrations of antibodies, may thus allow the detection of more discrete dietary lapses and help to promote long-term adherence to GFD through adapted follow-up strategy (9).
Two patients (5.8%), although in clinical remission from CD, displayed persisting moderate tTGA levels with undetectable EmAs during GFD follow-up. This may indicate discrete intakes of gluten (9). Alternatively, such persistence in tTGA IgA positivity may reflect residual reactivity toward gliadin-independent tTG domains (21). In addition, the presence of anti-Ig antibodies, which has been described in the serum of patients with CD (28), may have interfered with our RBA assay and led to false-positive tTGA detection.
RBA assays are presently “in house” techniques and recent evaluations have showed variable assay performances (16) in terms of interassay coefficients of variability and result reporting (units) among different laboratories. This may hamper the interpretation of RBA-tTGA monitoring during GFD if performed in different centers. Efforts in standardizing the various assays and international quality control surveys are thus needed for optimal monitoring of RBA-tTGA during GFD in large cohort of patients.
We did not find any correlation between tTGA levels as measured by RBA and the degree of histological damage at diagnosis. This is in contrast to previous studies showing that high levels of tTGA correlated with the presence of villous atrophy with Marsh scores of ≥3 in adult and pediatric CD (29,30). The small size of our series, the low number of patients with partial degrees of mucosal damage (only 2.9% of biopsies with a score of 2), and the sampling of tissue during diagnostic biopsy may explain this discrepancy. Of note, there was no relation between the clinical presentation at diagnosis and the histological score, suggesting that histology is independent both of clinical symptoms and RBA-tTGA levels, which were found to be correlated at diagnosis.
It is not clear whether EmAs and tTGAs play a role in the pathogenesis of CD and tissue injury (31). Autoantibodies could merely be an epiphenomenon not related to CD pathogenic mechanisms but rather to a state of heightened immune responsiveness in genetically predisposed individuals (32). The lack of correlation of tTGA levels with histological features of CD at diagnosis (the present study) or during GFD in most studies in adults (19,27,33) supports this possibility. In this regard, the fact that a subset of patients showed at diagnosis or after gluten challenge extremely high tTGA levels (>3000 cpm) requiring longer times to normalize upon GFD is reminiscent of high responders to alloantigens in the transfusion setting (34).
In conclusion, we show in the present study that both ELISA and RBA provide high sensitivity for the detection of tTGA IgA antibodies at diagnosis. During GFD, the inherent characteristics of RBA, best suited for the detection of extremely low concentrations of antibodies, are presumably responsible for higher tTGA positivity rates than previously reported with ELISA. A decreasing trend for RBA-tTGA levels most likely represents a better surrogate marker of compliance with GFD than absolute normal levels. In light of recent evaluations showing variable performances of ELISA-tTGA or RBA-tTGA tests at diagnosis or during GFD, present ongoing efforts in standardizing the various assays will permit definition of the optimal strategy for monitoring of GFD in larger studies (16).
1. Branski D, Troncone R. Celiac disease: a reappraisal. J Pediatr 1998; 133:181–187.
2. Revised criteria for diagnosis of coeliac disease. Report of Working Group of European Society of Paediatric Gastroenterology and Nutrition. Arch Dis Child 1990;65:909-11.
3. Bardella MT, Trovato C, Cesana BM, et al. Serological markers for coeliac disease: is it time to change? Dig Liver Dis 2001; 33:426–431.
4. 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.
5. Wong RC, Wilson RJ, Steele RH, et al. A comparison of 13 guinea pig and human anti-tissue transglutaminase antibody ELISA kits. J Clin Pathol 2002; 55:488–494.
6. Dickey W, Hughes DF, McMillan SA. Disappearance of endomysial antibodies in treated celiac disease does not indicate histological recovery. Am J Gastroenterol 2000; 95:712–714.
7. Hill PG, McMillan SA. Anti-tissue transglutaminase antibodies and their role in the investigation of coeliac disease. Ann Clin Biochem 2006; 43:105–117.
8. Dipper CR, Maitra S, Thomas R, et al. Anti-tissue transglutaminase antibodies in the follow-up of adult coeliac disease. Aliment Pharmacol Ther 2009; 30:236–244.
9. Zanini B, Lanzarotto F, Mora A, et al. Five year time course of celiac disease serology during gluten free diet: results of a community based “CD-Watch” program. Dig Liver Dis 2010; 42:865–870.
10. Liu E, Li M, Emery L, et al. Natural history of antibodies to deamidated gliadin peptides and transglutaminase in early childhood celiac disease. J Pediatr Gastroenterol Nutr 2007; 45:293–300.
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. Pietzak MM. Follow-up of patients with celiac disease: achieving compliance with treatment. Gastroenterology 2005; 128:S135–S141.
13. Dieterich W, Ehnis T, Bauer M, et al. Identification of tissue transglutaminase as the autoantigen of celiac disease. Nat Med 1997; 3:797–801.
14. Rostom A, Murray JA, Kagnoff MF. American Gastroenterological Association (AGA) Institute technical review on the diagnosis and management of celiac disease. Gastroenterology 2006; 131:1981–2002.
15. Bonamico M, Tiberti C, Picarelli A, et al. Radioimmunoassay to detect antitransglutaminase autoantibodies is the most sensitive and specific screening method for celiac disease. Am J Gastroenterol 2001; 96:1536–1540.
16. Li M, Yu L, Tiberti C, et al. A report on the International Transglutaminase Autoantibody Workshop for Celiac Disease. Am J Gastroenterol 2009; 104:154–163.
17. Bazzigaluppi E, Lampasona V, Barera G, et al. Comparison of tissue transglutaminase-specific antibody assays with established antibody measurements for coeliac disease. J Autoimmun 1999; 12:51–56.
18. Burgin-Wolff A, Dahlbom I, Hadziselimovic F, et al. Antibodies against human tissue transglutaminase and endomysium in diagnosing and monitoring coeliac disease. Scand J Gastroenterol 2002; 37:685–691.
19. Midhagen G, Aberg AK, Olcen P, et al. Antibody levels in adult patients with coeliac disease during gluten-free diet: a rapid initial decrease of clinical importance. J Intern Med 2004; 256:519–524.
20. Killander A, Arnell H, Hagenas L, et al. Omitting control biopsy in paediatric coeliac disease: a follow-up study. Acta Paediatr 2007; 96:1190–1194.
21. Tiberti C, Bonamico M, Dotta F, et al. Evidence of a selective epitope loss of anti-transglutaminase immunoreactivity in gluten-free diet celiac sera: a new tool to distinguish disease-specific immunoreactivities. Clin Immunol 2006; 121:40–46.
22. Sodoyez JC, Koch M, Lemaire I, et al. Influence of affinity of antibodies upon their detection by liquid phase radiobinding assay and solid phase enzyme linked immunosorbent assay. Demonstration using monoclonal antibodies raised against rDNA human proinsulin. Diabetologia 1991; 34:463–468.
23. Bingley PJ, Bonifacio E, Mueller PW. Diabetes Antibody Standardization Program: first assay proficiency evaluation. Diabetes 2003; 52:1128–1136.
24. Troncone R, Mayer M, Spagnuolo F, et al. Endomysial antibodies as unreliable markers for slight dietary transgressions in adolescents with celiac disease. J Pediatr Gastroenterol Nutr 1995; 21:69–72.
25. Kaukinen K, Sulkanen S, Maki M, et al. IgA-class transglutaminase antibodies in evaluating the efficacy of gluten-free diet in coeliac disease. Eur J Gastroenterol Hepatol 2002; 14:311–315.
26. Leffler DA, Edwards George JB, Dennis M, et al. A prospective comparative study of five measures of gluten-free diet adherence in adults with coeliac disease. Aliment Pharmacol Ther 2007; 26:1227–1235.
27. Vahedi K, Mascart F, Mary JY, et al. Reliability of antitransglutaminase antibodies as predictors of gluten-free diet compliance in adult celiac disease. Am J Gastroenterol 2003; 98:1079–1087.
28. Hallgren J, Knutson F, Lavo B, et al. Increased mucosal synthesis of rheumatoid factor (RF) in coeliac disease. Clin Exp Immunol 1996; 103:94–98.
29. Donaldson MR, Firth SD, Wimpee H, et al. Correlation of duodenal histology with tissue transglutaminase and endomysial antibody levels in pediatric celiac disease. Clin Gastroenterol Hepatol 2007; 5:567–573.
30. Donaldson MR, Book LS, Leiferman KM, et al. Strongly positive tissue transglutaminase antibodies are associated with Marsh 3 histopathology in adult and pediatric celiac disease. J Clin Gastroenterol 2008; 42:256–260.
31. Kagnoff MF. Celiac disease: pathogenesis of a model immunogenetic disease. J Clin Invest 2007; 117:41–49.
32. Cianci R, Cammarota G, Lolli S, et al. Abnormal synthesis of IgA in coeliac disease and related disorders. J Biol Regul Homeost Agents 2008; 22:99–104.
33. Tursi A, Brandimarte G, Giorgetti GM. Lack of usefulness of anti-transglutaminase antibodies in assessing histologic recovery after gluten-free diet in celiac disease. J Clin Gastroenterol 2003; 37:387–391.
34. Schonewille H, Brand A. Does an alloimmune response to strong immunogenic red blood cell antigens enhance a response to weaker antigens? Transfusion 2008; 48:958–963.