Main

Celiac disease is characterized by malabsorption, an abnormal immune response to dietary gluten, and signs of a local inflammation in the gut accompanied by development of a flattened intestinal mucosa(1–3). The diagnostic signs, formalized in the original ESPGAN (European Society of Paediatric Gastroenterology and Nutrition) Interlaken criteria for celiac disease in childhood, require three intestinal biopsies to verify the diagnosis(4, 5)(before dietary treatment, after dietary treatment, and after 3-6 mo of gluten challenge).

In practial terms also levels of serum IgA AGA are used diagnostically, and in many cases constitute the major screening method(6–8). IgA EMA, directed against antigens associated with the reticulin sheathe surrounding the smooth muscle fibers in monkey esophagus, are also frequently used in the screening for untreated celiac disease(9–12). Although both these methods, especially when used in combination, are very valuable tools in celiac disease, they do not necessarily mirror the production of IgA in the gut, because the major part of the IgA produced locally against gut immunogens is secreted into the lumen(13, 14). Furthermore, changes in antibody levels are always somewhat delayed in relation to the actual production of the antibodies in question.

As there is an obvious need, particularly in young children, to use cells or serum available in the peripheral blood for diagnosis and to confirm the diagnosis as quickly as possible after a gluten challenge, we wanted to use a novel method for determination of circulating IgA-producing cells toward gut immunogens to study the gut-associated immune response to gliadin in these patients. Our strategy took as one of its starting points the recent demonstrations first from Czerkinsky et al.(15) and later from own work(16) that a peroral immunization with an antigen, which gives rise to a mucosal immune response, is mirrored by increased levels of circulating IgA-producing cells against the immunogen as early as 6 to 12 d after immunization and that the ELISPOT method for measuring these circulating IgA-producing cells appears both more sensitive and more rapid for the detection of such gut-associated responses than is serum IgA levels for the same immunogen.

Therefore, we have in this study used the ELISPOT method(17) to enumerate circulating IgG, IgA, and IgM AGA SFC(18) at different phases of diagnosis in a wide spectrum of children with suspected celiac disease. One particular aim of the study was to investigate whether the ELISPOT method can be used as an improved technique for rapid follow-up of anti-gliadin reactivity after gluten challenge.

METHODS

Patients. Eighty-two children (46 girls, 36 boys, median age 4 y, range 1-18 y) with celiac disease or gastrointestinal symptoms and suspected malabsorption were studied. All patients were diagnosed at the Department of Paediatrics, Uppsala University Hospital. Celiac disease was found in 57 of the investigated children and so far disease has been verified in 42 according to the original ESPGAN criteria for celiac disease in childhood(4) and in 15 according to the revised criteria for diagnosis of celiac disease(19). Twenty-five of the children did not have celiac disease (disease controls), 10 had other food intolerances, 1 had inflammatory bowel disease, and 14 had miscellaneous gastrointestinal disorders.

Small intestinal biopsy was performed, and blood samples were collected before dietary treatment and before and after gluten challenge. Blood samples were also collected in 23 patients after 2 wk of gluten challenge. All the challenged children had a normal gluten-containing diet.

Seven normal children (5 girls, 2 boys, median age 4 y, range 1-9 y) without any gastrointestinal symptoms and without relatives with celiac disease served as controls. All studies on patients and healthy controls were performed in accordance with the ethical rules of the hospital, and for ethical reasons intestinal biopsies were not performed in the healthy controls.

Preparation of lymphocytes. Peripheral blood lymphocytes were isolated from heparinized venous blood samples (3-7 mL) by Ficoll-Paque(Pharmacia, Uppsala, Sweden) separation.

Antibody-producing cells. The ELISPOT method developed by Czerkinsky et al.(17) was used with minor modifications(20) to analyze Ig-producing cells (IgG, IgA and IgM) and gliadin specific Ig-producing cells (IgG AGA, IgA AGA, and IgM AGA) in peripheral blood. 96-well flat-bottomed plastic plates (Dynatech, Chantilly, USA) were coated at 4 °C overnight with anti-IgG (Tago, Burlingame, CA), anti-IgA (Dako, Glostrup, Denmark) and anti-IgM (Cappel, Durham, NC). For detection of gliadin-specific SFC, the plates were coated with crude gliadin (Sigma Chemical Co., St. Louis, MO) dissolved in ethanol, and diluted to 800 μg/mL in PBS (pH 7.2).

The plates were washed three times in distilled water, and cells were added at a concentration of 104 cells/well (IgG, IgA, and IgM) or 106 cells/well (IgG AGA, IgA AGA, and IgM AGA). The plates were then incubated for 3 h at 37 °C in 5% CO2. After washing, the plates were incubated with biotinylated anti-human antibodies diluted in PBS to 1 μg/mL specific for IgG (Tago), IgA (Tago), and IgM (Tago), respectively.

After incubation at 4 °C overnight, the cells were washed, avidin-alkaline-phosphatase (Dako) was added, and the plates were incubated another two hours at room temperature. Finally, the bound antibodies were visualized by adding the chromogen 5-bromo-4-chloro-3-indolyl-phosphate (BCIP, 2.3 mmol/L, Sigma Chemical Co.).

The number of spots in each well were counted with a reversed microscope and the antibody response was expressed as the number of SFC/106 MNC(SFC/106 MNC). Antigen-specificity was demonstrated by the absence of SFC on plates coated with human serum albumin, BSA, or PBS.

Anti-gliadin antibodies. IgG and IgA antibodies to gliadin in serum were determined with an ELISA method(6, 21). EIA microplates (Costar, Cambridge, MA) were coated with crude gliadin(Sigma Chemical Co.) dissolved in 70% ethanol and diluted in PBS to a concentration of 100 μg/mL. The plates were incubated at 37 °C for 2 h and then washed with 0.05% Tween 20 in PBS (PBS-T).

Patient sera were diluted in PBS-T with 1% BSA to suitable concentrations. Patient serum (100 μL) was added, and the plates were incubated for 1 h at room temperature and washed with PBS-T. Peroxidase-conjugated anti-IgG (Dako) or anti-IgA (Dako) diluted in PBS (1/1000 for anti-IgG and 1/300 for anti-IgA) was added, and after 1 h of incubation at room temperature the plates were washed. A substrate solution, containing 6 mg of 1,2-fenylenediamine dihydrochloride (Dako) and 5 μL of 30% H2O2 dissolved in 50 mL of 0.1 M citrate buffer (pH 5.0), was added. The reaction was stopped after 15 min with 1 M H2SO4 and read in a micro ELISA reader (Organon Teknika, Turnhout, Belgium) at 492 nm.

Positive sera were used as standards and the antibody concentration was expressed as arbitrary units/mL (AU/mL). Normal range for IgA AGA: <200 AU/mL (younger than 5 y), <72 AU/mL (older than 5 y). Normal range for IgG AGA: <38 AU/mL (younger than 5 y), <21 AU/mL (older than 5 y).

Anti-endomysium antibodies. IgA EMA were analyzed by indirect immunofluorescence microscopy using fixed cryostat sections of monkey esophagus (Scimedx, Densville, NJ) as an antigen substrate(22). Patient sera were diluted 1/10 in PBS and incubated 30 min at room temperature.

After washing, the reaction was detected with FITC rabbit anti-human IgA conjugate (Dako), diluted 1/20 in PBS and applied to the sections for 30 min. Positive sera were titrated further (1/50, 1/100, 1/500, and 1/1000), and the antibody titer was defined as the highest serum dilution yielding fluorescence.

Intestinal biopsy. Small intestinal biopsies were performed with a Watson capsule of pediatric size from the distal duodenum or proximal jejunum (the ligament of Treitz). Sections from the paraffin-embedded biopsies were stained with hematoxylin and eosin, van Gieson stain, and Alcian blue-PAS as used routinely.

All specimens were studied histologically by the same pathologist (W.K.), and normality was defined as a villous/crypt ratio of 2:1 or more(23) and an IEL count within normal limits (1 or less IEL/5 epithelial cells)(24, 25). The histopathologic changes were classified as: increased numbers of IEL without signs of atrophy, partial/subtotal villous atrophy (a villous/crypt ratio distinctly less than 2:1), and total villous atrophy (flat mucosa).

Statistics. The numbers of SFC/106 MNC are reported as median (with 10th and 90th centiles, in brackets). The Mann-Whitney U test (two-tailed) was used for calculation of differences in SFC values between different patient groups, and the Wilcoxon signed rank test was used to compare SFC values before and after gluten challenge.

Fischer's exact test was used for comparison of the proportions of patients with increased numbers of SFC and antibody titers after gluten challenge. In the evaluation of the different tests, the following statistical definitions have been used: sensitivity (the ability of a test to give a positive finding when the person tested has the disease), specificity (the ability of the test to give a negative finding when the person tested is free of disease), positive predictive value (the probability of the presence of disease when the test is positive), and negative predictive value (the probability of the absence of disease when the test is negative)(26).

RESULTS

Antibody production by peripheral blood MNC. The antigen specificity of the ELISPOT assay was shown by the absence of spots in uncoated wells and in wells coated with human serum albumin or BSA. The numbers of total IgA, IgM, and IgG SFC were relatively constant and independent of the diagnosis, age, and dietary treatment (Fig. 1).

Figure 1
figure 1

Numbers of IgA, IgM, and IgG SFC expressed as SFC/106 MNC in healthy controls, disease controls (children with gastrointestinal symptoms without celiac disease), treated celiac disease, untreated celiac disease, after 2 wk and after 12 wk of gluten challenge. Boxes mark 25th, 50th, and 75th, and end of bars 10th and 90th centiles.

Full size image

Gliadin SFC in children with untreated or treated celiac disease and controls. Figure 2 shows the numbers of IgA AGA, IgM AGA and IgG AGA SFC in peripheral blood from healthy control subjects, disease control subjects (see “Methods”), and treated or untreated patients with celiac disease. The numbers of IgA AGA SFC in untreated patients with celiac disease (median 25.0 (4.6-44.0) IgA AGA SFC/106 MNC) were significantly higher than in both control groups and in the group with treated celiac disease (p < 0.0001). The numbers of IgM AGA SFC were also increased in untreated patients (median 4.0 (0.2-14.9) IgM AGA SFC/106 MNC) compared with the other groups (p < 0.05), whereas no such difference was observed in the level of IgG AGA SFC. None of the 30 treated celiac disease patients on a gluten-free diet had more than 9 IgA AGA SFC, and only 1/25 of the patients with other gastrointestinal disorders than celiac disease had more than 6 IgA AGA SFC.

Figure 2
figure 2

Numbers of IgA, IgM, and IgG anti-gliadin SFC (IgA AGA, IgM AGA, and IgG AGA SFC) expressed as SFC/106 MNC in healthy controls, disease controls (children with gastrointestinal symptoms without celiac disease), treated celiac disease, and untreated celiac disease. Boxes mark 25th, 50th, and 75th, and end of bars 10th and 90th centiles (p values vs control groups).

Full size image

Comparison between IgA gliadin SFC and the mucosal condition. We next wanted to relate IgA AGA SFC levels with the degree of mucosal atrophy (Fig. 3). Seventy small intestinal biopsies were taken from 64 children (four underwent biopsy twice and one also had a third biopsy). There were 17 new cases of celiac disease, and 15 of the samples were taken from patients on a gluten-free diet. The numbers of IgA AGA SFC/106 MNC were significantly increased (p < 0.0001) for 22 children (14 with untreated celiac disease and 8 patients after 12 wk of gluten challenge) with total villous atrophy (median 16.0 (4.4-48.7) IgA AGA SFC/106 MNC) compared with 14 nonceliac disease children with normal mucosa [2.0 (0-6.3) IgA AGA SFC/106 MNC]. The numbers of SFC were also increased (p < 0.0001) for 13 patients (3 untreated patients and 10 children after 12 wk of gluten challenge) with partial/subtotal villous atrophy [20.0 (3.6-35.6) IgA AGA SFC/106 MNC] compared with the children with normal biopsies. There were no significant difference between the two atrophy groups (total and partial/subtotal villous atrophy), and 9 out of 15 children on a gluten-free diet had a normal mucosa [median for the 15 children on a gluten-free diet was 1.0 (0-2.0) IgA AGA SFC/106 MNC].

Figure 3
figure 3

Numbers of IgA anti-gliadin SFC (IgA AGA SFC) expressed as SFC/106 MNC in untreated and challenged celiac disease patients and controls compared with the histopathology of the small intestinal mucosa. Lines represent median values (VA = villous atrophy;part-subtotal VA = a villous/crypt ratio distinctly less than 2:1;p values vs normal).

Full size image

Almost all patients with untreated celiac disease had abnormally high counts of IEL (e.g. more than 20 IEL/100 epithelial cells) within the flat surface epithelium. A mucosa with normal villous/crypt ratio but increased numbers of IEL was present in five of the celiac disease patients on a gluten-free diet [median 0 (range 0-3) IgA AGA SFC/106 MNC] and in three of the disease controls, two with cow's milk protein intolerance, and one with unspecific gastrointestinal symptoms (1, 2, and 5 IgA AGA SFC/106 MNC). Three of the children with suspected celiac disease had increased numbers of IEL only after 12 wk of gluten challenge (2, 10, and 14 IgA AGA SFC/106 MNC). These three children were after a prolonged gluten challenge diagnosed as having celiac disease, in accord with their clinical symptoms.

Gliadin SFC during gluten challenge. In 20 children with the initial diagnosis celiac disease, based on biopsy findings and clinical remission on a gluten-free diet, the numbers of IgA AGA SFC/106 MNC had increased significantly (p < 0.0001) to median 14.0 (2.5-50.5) as early as after 2 wk of gluten challenge, and after 12 wk of challenge the levels of IgA AGA SFC [median 13 (5.5-32.0)] were still elevated (p< 0.0001) (Fig. 4). No increase was seen for either IgM AGA SFC or IgG AGA SFC during challenge (data not shown).

Figure 4
figure 4

Numbers and median values of IgA anti-gliadin SFC (IgA AGA SFC) expressed as SFC/106 MNC before a gluten challenge, after 2 wk and after 12 wk of gluten challenge in 20 children with and 3 without celiac disease (p values vs before).

Full size image

Figure 4 also shows three children with nonceliac disease during gluten challenge. All these children had initially a suspected celiac disease. However, none of them had a mucosal relapse or clinical symptoms after 12 wk of gluten challenge, and no increase in the numbers of IgA, IgM, or IgG AGA SFC were seen after 2 wk or after 12 wk of gluten challenge. The serum IgA AGA and IgG AGA titers followed a different pattern and increased from 2 wk to 12 wk of gluten challenge (data not shown).

Sensitivity and specificity of IgA gliadin SFC in detection of celiac disease. Table 1 summarizes the sensitivity and specificity of gliadin SFC, serum AGA, and EMA during gluten challenge. The mean numbers ± 2 SD of IgA AGA SFC in the 7 healthy and 25 disease controls were 7 IgA AGA SFC/106 MNC, and this value was chosen as the cutoff level for the ELISPOT test. Fifteen of the 17 untreated children with celiac disease had a positive ELISPOT test, whereas 31/32 children with nonceliac disease fed a normal diet had a negative test. One of the two false negatives was 10 y of age. Of untreated patients, 16/17 had IgA EMA (titer 1/10-1/1000).

Table 1 Comparison of the validity of the IgA gliadin spot-forming cell test (IgA AGA SFC), the serum IgA and IgG gliadin assay (IgA AGA and IgG AGA), and the serum IgA endomysium test (IgA EMA) for detection of coeliac disease: samples were taken before dietary treatment, after 2 wk, and after 12 wk of gluten challenge
Full size table

When comparing the numbers of SFC to the serum AGA and serum EMA concentrations during a gluten challenge, the IgA AGA SFC values increased to above the cutoff level in 15 of 20 children with celiac disease after 2 wk of challenge, whereas the serum IgA AGA titers increased to above the cutoff levels in 6 of the 20 children (p < 0.01) and the serum IgG AGA titers in 3/20 (p < 0.001) after 2 wk of gluten challenge. Seven of the 20 children (p < 0.05) had IgA EMA after 2 wk of gluten challenge (titer 1/10-1/100). After 12 wk of challenge 24/28 children had a positive ELISPOT test, 21/28 had increased titers of serum IgA AGA, and 21/28 were IgA EMA positive (titer 1/10-1/500). When the serum IgA AGA and serum IgA EMA assays were used in combination, 17/17 untreated patients, 9/20 patients after 2 wk of gluten challenge, and 23/28 patients after 12 wk of gluten challenge had increased serum levels of either IgA AGA or IgA EMA or both.

DISCUSSION

This study was concerned with the possibility that antibody-producing cells in the circulation, measured by the ELISPOT technique, may mirror a disease-associated gut immune response toward gluten in such a way that detection of this antibody production may provide a useful complementary diagnostic method for celiac disease in children. The ELISPOT technique may also be useful in studies of antibodies produced during a pathogenic antigen challenge in the gut, because antibodies present in serum represent only the resulting fraction from production and elimination by absorption or consumption.

The results show that the occurrence of circulating IgA AGA-producing cells is a remarkably reproducible finding in children with celiac disease, defined either by the classical ESPGAN “Interlaken criteria”(4) or by the revised criteria for diagnosis of celiac disease(19). There was similarly a very reproducible and rapid increase in the numbers of IgA antigliadin-producing cells after 2 wk of gluten challenge, which indicates that the appearance of antibody-producing cells in blood may follow the same dynamics in relation to gluten antigen exposure as has been demonstrated previously for experimental antigens, such as peroral influenza vaccine, where a full response to antibody production in peripheral blood is seen at 6-8 d after the challenge(15, 16). These results in mainly young children are interestingly somewhat different from those obtained by Lycke et al.(18) with a similar methodology in a limited number of adult patients with celiac disease. In these adults only 2/6 had an IgA anti-gliadin response measured by the ELISPOT method after gluten challenge, despite the fact that such antibody-producing cells could in most of these patients be identified in the gastrointestinal mucosa. We do not know the reason for this discrepancy, but if it mirrors a reproducible difference between young children and adults, it would be tempting to speculate about a pathogenic relevance of the finding which would deserve further investigation. It is believed that the presence of specific IgA-producing cells in the circulation is part of a mechanism whereby an immune response to an antigen present in one part of the mucosal system is generalized to the whole mucosal system. Differences in this “spreading” of the immune response might then also hypothetically contribute in determining whether a particular local antigenic challenge may or may not lead to pathologic consequences. In more general terms, the possibility to measure both antibodies in serum and the appearance of migrating antibody-producing cells in individuals challenged with an antigen, such as gluten in the gut, may help to elucidate when an immune response to a gut-associated antigen will have pathologic consequences. It is known that this is not always the case, because children with a previous history of gluten enteropathy may have increased serum levels of AGA without morphologic signs of gut damage after gluten challenge(11).

Concerning the diagnostic potentials of measuring antibody-producing cells as a complement to determination of serum titers of AGA, the present data indicate that such a potential may indeed exist. Thus, first, there were a few children who showed clinical and histopathologic signs of gluten enteropathy after 12 wk of gluten challenge who had a positive ELISPOT test while being negative in the ELISA test. Second, and possibly more important, the ELISPOT test gave a more reliable response than both the gliadin ELISA test and a standard test for EMA at the early time point after challenge that was studied. In 15/20 children with celiac disease this increase in AGA-producing cells was apparent as early as after 2 wk of gluten challenge, whereas the increase in serum EMA was slower, and only 7/20 patients were positive after 2 wk of challenge. In combination, the high sensitivity as well as specificity and the possibility for a diagnosis from a blood sample early after challenge represent attractive features for a diagnostic procedure for celiac disease in children.

A major drawback for extension studies on the clinical feasibility of such an assay is of course the amount of work needed to perform an ELISPOT assay instead of an ELISA. It remains to be seen, therefore, whether enough simplified methods for measuring antibody production from peripheral blood lymphocytes can be developed to make it possible to introduce such methodology in routine clinical work. In the meantime, however, the ELISPOT method may provide an additional research tool both for diagnosis and for evaluation of therapies in celiac disease.