Main

Elimination diet followed by double-blind placebo-controlled food challenge is considered the gold standard for the diagnosis of food allergy in infants and children (1). Immediate reactions after food challenge can be easily identified, whereas delayed reactions can be difficult to recognize (2). Therefore, nonclinical criteria for diagnosing food allergy are required. Among proposed diagnostic tests, SPTs and measurement of serum-specific IgE antibodies to foods are commonly performed in children with suspected food allergy. However, these tests should be regarded as only suggestive of food allergy (3,4).

IgE-positive cells have been demonstrated in the intestinal mucosa by immunoelectron microscopy (5). However, conflicting results have been reported on the presence of IgE-containing cells in the mucosa of the small intestine and on their relation to serum-specific IgE antibodies in uncontrolled studies in patients with food allergy (6–11). So far, no comparative data are available on the diagnostic accuracy of SPT to foods versus positive IgE cells in the intestinal mucosa, and on the correlation between duodenal IgE-containing cells and the total serum IgE level.

We were therefore interested first, in clearly ascertaining the value of duodenal IgE-positive cells in the diagnosis of food allergy in children, and second, in comparing the finding of duodenal mucosal IgE-containing cells with SPT results and serum IgE levels.

METHODS

Patients. Thirty-one children with a history of symptoms which could be caused by food allergy were consecutively enrolled in the study. Complaints were vomiting in 24 cases, eczema in 14, diarrhea in 5, abdominal pain in 3, asthma in 2, urticaria in 2, angioedema in 1, failure to thrive in 1.

All children underwent endoscopy of the intestinal upper tract (Olympus GIF-XP 20) before beginning an elimination diet. Duodenal biopsies were taken in all children. The children were investigated while having symptoms or within 24 h from the last appearance of clinical signs of allergy. Immunologic studies were carried out in all patients.

To ascertain the presence of clinical food hypersensitivity, children were put on a 14-d restricted diet. During this period, patients were allowed to eat only soy milk, tea, water, sucrose, rice, carrots, lettuce, potatoes, rabbit's meat, pears, and olive oil, if none of these foods was suspected of eliciting clinical reactions and SPT results were negative (12). None of these infants was breast-fed. Calcium supplements were given. If symptoms improved on this diet, oral challenges to common foods were performed. Each food was given for 4 d at home. Children were visited before and after the challenge and if reactions occurred. Challenges for foods suspected of eliciting adverse reactions, on the basis of positive immunologic tests or if a suggestive history was present, were performed at the hospital (12). Open challenge tests were performed in infants under 12 mo of age, and double-blind placebo-controlled food challenges in children over 1 y of age (1). Briefly, children were administered increasing doses of food or placebo every 20 min. The total amount of food was equal to the average daily intake, e.g. one egg. Patients were observed for 5 h. If no symptom occurred, parents gave the food at home for 4 d. As a placebo, a soy formula (Humana SL, Humana Italia, Milan, Italy) was used in the children who were tolerant. In the children with suspected soy allergy and in soy challenge, an amino acid-derived formula (Nutri-junior, Nutricia, Milan, Italy) was used. Positive responses to placebo were never found. In one case provocation was not performed because the child had a convincing history of a severe anaphylactic reaction and a positive SPT result. All children who improved after the elimination diet had at least one positive food challenge. No children were on antihistamines, oral steroids, or β-agonists at the time of the study.

According to the presence or absence of clinical reactions to oral food challenge, children were included in the food-allergic patient or the control group, respectively. The food-allergic patient group consisted of 13 children (8 boys and 5 girls; age range, 2 mo to 4 y, mean 15 ± 13 mo) with positive food challenge (see Table 1). Thirteen of 20 positive provocation tests elicited reactions within 12 h from the end of the challenge. These early reactors presented the following symptoms: vomiting in 8 cases, urticaria in 5, rash in 5, angioedema in 4, diarrhea in 4, eczema in 1, asthma in 1, rhinitis in 1. Seven single delayed reactions which started more than 12 h after challenge occurred: eczema in 6 cases and diarrhea in 1. Two children had early reactions and then developed delayed symptoms: eczema in 2 cases and diarrhea in 1. Eleven children were found to be allergic to cow's milk, 5 to egg, 3 to wheat, and 1 to lamb's meat.

Table 1 Clinical data, results of immunologic studies, early (<12 h) and delayed (>12 h) symptoms after oral provocation in patients with food allergy
Full size table

The control group consisted of 18 patients (12 boys and 6 girls; age range, 5 mo to 10 y, mean, 44 ± 41 mo) who did not have food allergy (negative oral food challenge results). The age of the control group was significantly higher than that of the patient group (p < 0.001). Control patients were referred with the following symptoms: vomiting in 17 cases, eczema in 3, diarrhea in 2, abdominal pain in 2, asthma in 1, failure to thrive in 1. Parents of all patients gave written consent.

Immunologic studies. All children were given SPTs with food extracts (1:20, w:v) (Bayropharm, Milan, Italy) of hen's egg, cow's milk, goat's milk, pea, wheat, oat, corn, barley, rice, bean, peanut, pear, peach, apple, banana, tomato, potato, carrot, soy, cod, tuna, salmon, beef, pork, lamb's meat, chicken, and rabbit's meat. The size of the skin test wheal was compared with histamine chloride, 1 mg/mL, and the diluent. SPTs were considered positive if the diameter of the wheal was greater than 3 mm, after subtracting the diameter of the wheal of the diluent (12) and was graded in comparison with the histamine wheal: + + + + if greater than twice the area of the histamine standard, + + + if twice the area of the histamine standard, + + if equal to the area of the histamine standard, and + if half the area of the histamine standard (13). Serum circulating IgE antibodies to foods were measured by Phadebas RAST (radioallergosorbant test) (Pharmacia, Uppsala, Sweden). Total serum IgE was measured by Phadebas PRIST (Pharmacia, Uppsala, Sweden).

Immunofluorescent study. Duodenal biopsies were immediately fixed in periodate-lysine-paraformaldehyde, which can preserve antigenicity better than conventional fixative (14). The periodate-lysine-paraformaldehyde solution was prepared as follows: 0.1 M dibasic sodium phosphate was added to 0.2 M lysine-HC1 in distilled water until a pH of 7.4 was achieved. The solution was then diluted to 0.1 M lysine with 0.1 M sodium phosphate buffer, pH 7.4 (resulting phosphate concentration was 0.05 M). After the fixation with the paraformaldehyde solution, solid sodium m-periodate was added. Then, the tissues were washed with three changes of cacodylate 0.1 M buffer solution for 30 min each, and then overnight at 4°C with cacodylate 0.1 M buffer solution to which 7% sucrose had been added. Excess washing solution was removed, and the specimens were rinsed with four changes of acetone for 10 min each. Dehydrated specimens were treated with xylol for 30 min and embedded in paraffin. Serial sections were cut at about 2-3 µm on a sliding microtome (LKBF, Sweden). The specimens were dried on a plate at 37°C. For histology, adjacent tissue sections were subjected to conventional mast cell staining procedures with toluidine blue or were stained with hematoxylin and eosin. For immunohistochemistry, adjacent sections were treated with xylol for 15 min, then with ethanol, and washed with water. After treatment with PBS buffer, the specimens were incubated overnight with commercially prepared fluorescent antisera specific for human IgE (Dako, Copenhagen, Denmark) at 4°C. The sections were washed with three changes of PBS buffer. The specimens were examined under a Leitz Ortholux incident-light fluorescent microscope. The lamina propria cells with IgE fluorescence were counted by a observer unaware of the clinical data. Counting was performed with a ×40 objective. In each case the number of cells in 10 different adjacent areas of the mucosa above the muscularis mucosae was counted, giving the number of cells/10 visual fields. Nasal polyps were used as a positive control tissue (15), whereas negative controls were carried out by the absence of staining in positive sections after blocking antigenic binding sites with nonconjugated antisera. The specificity of the anti-IgE antiserum was confirmed by lack of cytoplasmic immunofluorescence on monoclonal bone marrow preparations using antisera detected against IgA, IgG, IgM, and IgD.

Adjacent tissue sections stained with toluidine blue or with hematoxylin and eosin were examined by conventional transmission microscopy to identify IgE-positive cells (4). Furthermore, the number of mast cells/10 visual fields was counted.

Data analysis. Data are presented as mean ± SD. ANOVA was used to compare continuous variables. Logarithmically transformed serum IgE levels showed a normal distribution. We considered 95% of the number of duodenal IgE-positive cells in control subjects as the normal value. Sensitivity, specificity, and predictive accuracy of diagnostic tests were calculated using standard formulas (16). Differences in the ability of diagnostic tests to discriminate children with food allergy were compared by McNemar's test with Yates continuity correction. A χ2 test with Yates continuity correction was used to compare differences in percentages of food-allergic and control children who had elevated total serum IgE. Correlations were examined by linear regression analysis and by Spearman rank correlation coefficient (rS).

RESULTS

Immunologic studies. The mean logarithmic total serum IgE levels in the food-allergic children (2.12 ± 0.62) were significantly elevated compared with that of the control group (1.44 ± 0.23) (p < 0.01). Total serum IgE were elevated for age (17) in 10 out of 13 (76%) children with food allergy and in 2 out of 18 control subjects (11%) (p < 0.001).

Positive SPT reactions to at least one food were found in 10 of 13 (76%) children in the food-allergic group and in 4 of 18 (22%) children in the control group (p < 0.01). In the food-allergic group, 8 children had positive SPT reactions to cow's milk, 6 to egg, 4 to wheat, 3 to pea, 3 to peanut, 2 to oat, 2 to barley, 2 to potato, 2 to pork, 2 to beef, 1 to soy, 1 to pear, 1 to carrot, 1 to bean, 1 to lamb's meat, and 1 to tomato. SPT reactions were positive to 16 of the 20 (80%) offending foods. Eight children were positive to cow's milk, 4 to egg, 3 to wheat, and 1 to lamb's meat (Table 1). In the control group, SPT reactions were positive to cow's milk in 2 cases, to egg in 2, and to pear in 1.

Specific serum IgE antibodies to one or several foods were found in 11 (84%) children in the food-allergic group and in 3 (16%) children in the control group (p < 0.001). In the food-allergic group, 7 children had specific serum IgE antibodies to cow's milk, 7 to egg, 3 to wheat, 2 to rice, 2 to potato, 2 to soy, 2 to pea, 2 to peanut, 2 to lamb's meat, 1 to beef, 1 to pork, 1 to pear, 1 to barley, and 1 to oat. Specific serum IgE antibodies to 14 of the 20 (70%) offending foods were found. Seven children had IgE antibodies to cow's milk, 4 to egg, and 3 to wheat (Table 1). In the control group, specific serum IgE antibodies to egg were found in 2 cases, to cow's milk in 1, and to wheat in 1. Among the early reactors group, one child was SPT-negative and 3 food-specific IgE antibody-negative. Two children who had isolated delayed reactions to cow's milk had both negative SPT and food-specific IgE antibodies. Results of both SPTs and specific serum IgE antibodies for the antigens most frequently associated with clinical reactions were found to have poor positive predictive accuracy and sensitivity, but high specificity and negative predictive accuracy. Total serum IgE level gave similar values of sensitivity, specificity, and predictive accuracy (Table 2).

Table 2 Sensitivity, specificity, and positive and negative predictive accuracy of number of duodenal IgE-containing cells, total serum IgE, SPT, and specific serum IgE antibodies for the two food antigens most frequently associated with food hypersensitivity
Full size table

There was no significant difference between results of SPT and allergen-specific serum IgE antibodies in identifying children with egg or cow's milk allergy. Furthermore, no significant differences was found between total serum IgE level and results of SPT, or food-specific IgE antibodies, or SPT-positive food-specific IgE antibodies in identifying allergic subjects. Positive SPT and RAST reactions were not associated with the time of onset of symptoms.

Patients 3 and 11 (Table 1) who had only atopic dermatitis after challenge were followed up in the outpatient clinic to have confirmation of the positive cow's milk challenge. No eczematous lesion appeared on a cow's milk-free diet during a 1-y follow-up.

IgE-containing cells in the duodenal mucosa. The distribution of the number of IgE-positive cells in the biopsy specimens from the food-allergic children (153.24 ± 83.13) was significantly elevated compared with that of the control group (18.4 ± 18.90) (p < 0.01) (Fig. 1). The number of IgE-positive cells was significantly more elevated in children who developed isolated delayed reactions (82.66 ± 23.69) than in control subjects (p < 0.01). This number was not significantly different between children who developed isolated delayed reactions and the remaining food-allergic children. The prevalence of elevated numbers of IgE-containing cells in food-allergic children was 92%. The number of duodenal IgE-positive cells correlated with the total serum IgE level (r = 0.62; p < 0.001) (Fig. 2) (rS = 0.64; p < 0.001). Age (17 children under 18 mo versus 14 children over 18 mo), sex (11 female versus 20 male), or symptoms did not significantly influence the number of IgE-positive cells or the total IgE concentrations in serum.

Figure 1
figure 1

Number of IgE-positive cells in 13 children with food allergy and in 18 control children. Mean values are indicated by horizontal lines. The means were 153.24 and 18.4 IgE-positive cells/10 fields for food-allergic and control children (p < 0.01), respectively.

Full size image
Figure 2
figure 2

Plot of log (total serum IgE) against numbers of duodenal mucosal IgE-positive cells (r = 0.62; p < 0.001).

Full size image

Values of sensitivity, specificity, and positive and negative predictive accuracy calculated for the number of duodenal IgE-containing cells and total serum IgE level for clinical hypersensitivity to at least one food are listed in Table 3. In general, the diagnostic accuracy of IgE-containing cells was high. However, there was no significant difference between the number of IgE- containing cells and total serum IgE level in identifying children with food allergy.

Table 3 Sensitivity, specificity, and positive and negative predictive accuracy of number of duodenal IgE-containing cells and total serum IgE value for the presence of clinical food allergy to at least one food in 31 children
Full size table

Among 12 food-allergic children with an elevated number of duodenal IgE-positive cells, two had negative SPTs and specific IgE antibodies to the offending foods, and one of them had a normal serum IgE level. Both children had isolated delayed clinical reactions after food challenges (Table 1).

The number of duodenal IgE-containing cells in children with clinical reactions to cow's milk was found to have a high but not absolute diagnostic accuracy. Sensivity and negative predictive accuracy of duodenal IgE-positive cells in children with clinical hypersensitivity to egg were high, but positive predictive accuracy and specificity were poor (Table 2). There was no significant difference between the number of IgE-containing cells and the results of SPT, or food-specific serum IgE antibodies, or SPT-positive allergen-specific IgE antibodies or serum IgE levels in identifying children with clinical hypersensitivity to cow's milk and egg.

In children with food allergy, a marked inflammatory cell infiltration which includes lymphocytes, mostly plasma cells, polymorphs, and some eosinophils was observed in the duodenal mucosa. The number of mast cells was 3.6 per 10 visual fields. In the control group, mast cells were virtually absent and the infiltrate poor. IgE-positive cells were judged to be principally mononuclear cells. Most of them were recognized as plasma cells; mast cells represented 2.3% of IgE-positive cells.

DISCUSSION

The results of this study showed that IgE-positive cells in the duodenal mucosa are significantly increased in children with food allergy. These findings were consistent with previous reports describing an association of duodenal IgE-containing cells with cow's milk allergy (7–9). Uncontrolled Finnish studies (6,10,11) found different results in terms of IgE-positive cells at the site of the intestinal mucosa. However, it is not easy to make a comparison between our results and theirs because of the different experimental design and immunohistochemical techniques used and because of the lack of antisera specificity, especially before 1980 (18). These differences may also explain our observation of IgE-positive cells being mainly plasma cells. This finding is in accordance with some previous studies (5,8), but at variance with the results of others (19) which showed predominantly mast cells.

Further, our data show that sensitivity, specificity, and predictive indices of duodenal IgE-positive cells are similar to those of SPTs and specific serum IgE antibodies to foods. The diagnostic accuracy of SPTs and specific serum IgE antibodies to cow's milk and egg was comparable to that reported in previous studies (3,4). The measurement of mucosal duodenal IgE-positive cells provided no diagnostic advantage over total serum IgE in predicting clinical food allergy. As the prevalence of food allergy in our population was higher than in the general population (20), normal numbers of IgE-containing cells should be considered useful to rule out clinical food hypersensitivity. Increased numbers of IgE immunoreactive cells should be regarded only as suggestive of a reaction to foods (21).

Schrander et al. (8) determined in 16 children with cow's milk intolerance the diagnostic value of duodenal mucosal IgE-positive cells, stained by the immunoperoxidase technique, and of specific serum IgE antibodies in comparison to cow's milk challenge. The specificity and positive predictive accuracy of the number of IgE-positive cells were comparable to ours, but sensitivity was lower. However, these authors did not describe how the food challenge was performed and which symptoms were induced. They carried out a qualitative rather than a quantitative evaluation of the immunohistochemical stainings. SPTs were not studied.

In children who have delayed clinical reactions to foods, immunologic tests are considered to have no diagnostic value (22,23). Interestingly, we found that delayed reactors had a significantly elevated number of mucosal duodenal IgE-positive cells, even in the absence of elevated total serum IgE levels, specific serum IgE antibodies, or positive SPT to the offending food. Thus, our findings suggest that not only acute onset reactions but also delayed reactions, occurring at least 12 h after food exposure, are IgE-mediated (24). However, the diagnostic value of this observation needs to be further clarified.

In some delayed reactors, symptoms cannot be easily related to the reintroduction of a food, because of a long time lag or the variable course of the disease. Therefore, we followed up patients 3 and 11 (Table 1) to confirm the positive cow's milk challenge. During a 1-y follow-up, no eczematous lesion appeared in children given a cow's milk-free diet.

In our study population, children with food allergy were significantly younger than were control children. This is in agreement with studies (25,26) which reported a higher prevalence of adverse reactions to food in infancy and early childhood. Total serum IgE levels increase during childhood (17), but were found to be lower in the control group. Therefore, we think that the different age of the two groups did not influence our results.

Previous studies have offered evidence of intestinal mucosal damage in some patients with food allergy, by documenting 1) minor morphologic changes in the mucosa of the small intestine (27–30), 2) an increased gastrointestinal permeability (31–34), and 3) reactions on the gastric mucosa after intragastric food challenge under endoscopy (35). However, the mechanisms responsible for intestinal alterations in food-allergic patients are poorly understood (36). Several studies indicate that immediate hypersensitivity reactions are involved. In children with food allergy, elevated IgE levels have been found both in feces (37) and intestinal washings (38), and the fecal eosinophil cationic protein concentration is enhanced (39). Our results extend these previous observations by showing a significant correlation between clinical food hypersensitivity and an increased number of duodenal mucosal IgE-positive mast cells as well as IgE-producing plasma cells.

The detection of allergen-specific IgE antibodies in intestinal washings in food-intolerant patients with both negative SPTs and food-specific serum IgE antibodies has been interpreted as the result of a strictly local IgE synthesis (38). In contrast, there is some support for the view that the whole mucosal immune system is involved in patients with atopic diseases (40). Moreover, local intestinal IgE production was found to be unlikely involved to any large extent, because IgE-positive B cells represented less than 2% of intestinal Ig-producing cells (41). Further, serum IgE concentrations have been found to be associated with the number of basophil IgE-receptors (42). In agreement with these observations, we found that the number of duodenal mucosal IgE-positive cells was correlated with IgE levels in blood, suggesting a systemic IgE production. This is true also in the patients with gastrointestinal complaints. Our data of IgE immunoreactive cells in the duodenal mucosa in the absence of positive SPT or allergen-specific IgE antibodies do not exclude a strictly local production of IgE. Alternatively, these findings may be explained by the poor accuracy of immunologic tests.

Our study indicates that occurrence of cutaneous, gastrointestinal, and respiratory symptoms after food intake is associated with IgE-mediated reactions in the duodenal mucosa. We found that duodenal IgE-positive cells had a high but imperfect diagnostic accuracy which was comparable to that of total serum IgE, SPT, and food-specific IgE antibodies. Thus, our results confirm that an allergen elimination diet followed by food challenge is the most accurate means of diagnosing food allergy. However, we found increased IgE-containing cells in all delayed reactors despite the fact that one-third had negative immunologic tests. Baehler et al. (22) found that no patients with delayed reactions to cow's milk challenge had positive SPTs or specific serum IgE antibodies. A larger study is required to establish whether our findings in the delayed reactor group may be of diagnostic value.