Main

Immune-mediated hypersensitivity to CMP affects between 2 and 7.5% of infants(1, 2). It is considered that immunologically mediated hypersensitivity to CMP occurs predominantly via type I and type IV hypersensitivity reactions, and these are associated with different clinical presentations and pathology(3); however, not only is there heterogeneity in the immunologic response to CMP, but there is heterogeneity in the gastrointestinal response. Children with immediate, classical CMA type I-like reactions usually have IgE antibodies to CMP. They have a rapid onset of symptoms with diarrhea and vomiting, and sometimes anaphylaxis, after milk ingestion(4, 5). There is no information on the upper small intestinal mucosa in these children because it is unethical to take biopsies, but animal studies suggest that intestinal anaphylaxis is not associated with any gross change in mucosal architecture(6, 7). Other children have a syndrome of chronic diarrhea and failure to thrive on a milk-containing diet(8). These children often do not have IgE antibodies(9), and the intestinal mucosa at biopsy shows a thin mucosa with patchy, mild to moderate partial villous atrophy. This syndrome is termed CMSE. The need to differentiate this condition from celiac disease makes it ethical to carry out small bowel biopsies in CMSE. Definitive diagnosis of CMA and CMSE depends on double-blind challenge, but because there are ethical grounds as to whether these children should be challenged when thriving on a milk-free diet, it is accepted that diagnosis can be made on the basis of clinical and laboratory findings(11). Virtually all bottle-fed children have precipitating antibodies to CMP in their serum(12), and although type III hypersensitivity reactions to foods are not commonly considered, there is evidence that these may occur in the intestine(13).

There is now good evidence that cell-mediated hypersensitivity mediated by lamina propria T cells can cause villous atrophy and crypt cell hyperplasia(14, 15). Activated T cells are present in the lesions of the best studied food sensitive enteropathy, celiac disease(16–18), and gluten-reactive T cell clones can be isolated from celiac mucosa(19, 20). Although T cell hypersensitivity has been proposed to be of pathogenetic importance in CMSE, no study has addressed this in the target organ, the gut.

The capacity of T cells to produce tissue damage is largely controlled by the repertoire of proinflammatory cytokines they produce(21–24). Th1 cells produce mainly IL-2 and IFN-γ upon activation and are responsible for cell-mediated immunity and delayed-type hypersensitivity tissue damage. In contrast, Th2 cells produce IL-4, IL-5, and IL-10, with IL-4 and IL-5 mainly promoting IgE responses and eosinophilia. In the case of hypersensitivity to cow's milk it might be expected that children with an immediate IgE-type response may have T cells that secrete Th2 cytokines on milk stimulation, whereas those with late-onset CMSE may have Th1-type cytokine profile. So far there are only four studies in which cytokine production in cow's milk hypersensitivity has been assessed, all using PBL. In the first study, it was shown that lymphocytes from children with immediate-onset symptoms to CMP produced less IFN-γ when stimulated in vitro with CMP than control children or those with delayed-onset symptoms(25). In the second study, lymphocytes from a heterogeneous group of cow's milk-sensitive children, some with villous atropy, were cultured in vitro with CMP and released the same amounts of IFN-γ and IL-4 as controls(26). In the third study, blood lymphocytes from patients with CMA had reduced IFN-γ production after polyclonal activation, a defect which was not seen when the children became milk-tolerant(27). Finally, Nakajima et al.(28) produced T cell lines specific for alpha s1-casein from patients with CMA. CD8+ cell lines secreting IFN-γ and IL-4 were isolated in a surprisingly high frequency from these patients(28). Earlier studies using in vitro assays for the production of l eukocyte inhibition factor (which is likely to be a mixture of cytokines) suggested that children with slow onset symptoms to cow's milk had greater responses than immediate responders(29), suggestive of heightened T cell immunity.

In this study we have assessed spontaneous cytokine production by ELISPOT in the blood and lamina propria of children with untreated CMSE to determine whether there is preferential secretion of Th1-type cytokines. We have also compared the cytokine profiles in blood of children with CMA and children with CMSE.

METHODS

Patients and samples. This study received ethical approval from the Hackney and District Health Authority. Fourteen children with a clinical history suggestive of CMSE (12 boys, two girls, aged 3-21 mo, mean age 12.7 mo) were studied (see Table 1). Ten children who had been assessed for failure to thrive (seven boys, three girls, aged 3-24 mo, mean age 17.3 mo), but with normal mucosal morphology, served as control subjects. Diagnosis of CMSE was established in concordance with the current European Society for Pediatric Gastroenterology and Nutrition criteria(10) and was based on history, clinical symptoms, mucosal morphology and histology, and immediate clinical improvement on a cow's milk-free diet. All children had a history of chronic diarrhea (diarrhea for more than 14 d with negative stool virology and bacteriology) and failure to thrive subsequent to the introduction of CMP containing formula/cow's milk into their diet. The history of symptoms averaged 6.4 mo (range, 2-10 mo), with patients 2, 3, 4, and 8, experiencing a weight loss of more than 2 SD in age-related centiles. Additional skin problems (anal and/or circumoral dermatitis) were encountered in patients 2, 9, 11, and 12, and patient 10 had recurrent rectal bleeding. All children underwent a Crosby capsule biopsy in the duodenum, and specimens were immediately examined under a dissecting microscope (×25 magnification) for their gross appearance. One specimen was processed for routine histology, and another for the isolation of LPL. Blood for parallel investigation was obtained from an i.v. canula already in place. In addition, blood mononuclear cells from five children (three boys, two girls, aged 5-12 mo) with rapid onset of symptoms at the introduction of cow's milk were also studied. Within 1 h of exposure to cow's milk, all developed diarrhea and vomiting, and three of the five developed wheezing. One developed a facial rash. In two of the five, IgE responses were confirmed by radioimmunosorbent and radioallergosorbent tests, and in three by skin-prick test.

Table 1 Patients' details (CMSE): clinical response to milk-free diet (weight gain within 1 mo)
Full size table

Histologic analysis. One duodenal specimen from each patient and control subject was fixed in formalin for routine hematoxylin and eosin staining. The histologic analysis was carried out without knowing the patients' symptoms, the initial mucosal appearance, or results of the cytokine assays.

Separation of peripheral blood mononuclear cells. Peripheral blood mononuclear cells were obtained from 1-2 mL of venous heparinized blood layered on a Ficoll-Hypaque density gradient using standard procedures. The cells in the buffy coat were collected and washed three times in RPMI 1640 medium (Flow, Irvine, Scotland) containing sodium bicarbonate, mercaptoethanol, penicillin, streptomycin, gentamicin, and supplemented with 10% heat-inactivated FCS. After resuspension in 1 mL of medium, cell numbers were determined using a hemocytometer. Viable cells were identified by trypan blue (0.1%) exclusion and comprised more than 95% before use. Two to four cytocentrifuge preparations per sample were performed. Between 15 000 and 20 000 cells were placed on microscope slides using a cytospin centrifuge at 700 rpm (200 × g). Slides were allowed to dry and were stored at -70 °C in airtight containers with desiccant (silica gel, Merck-BDH) until immunostaining.

Isolation of lamina propria mononuclear cells. Mononuclear cells were isolated from the biopsies by collagenase digestion as previously described(30). Briefly, specimens were freed of epithelial cells by initial incubation in 20 mL of Hanks' balanced salt solution free of calcium and magnesium (Flow Laboratories, McLean, VA) for 45 min at room temperature. They were then placed into RPMI medium supplemented with 10% FCS containing 1.5 mg/mL type 1 collagenase (Sigma Chemical Co., St. Louis, MO). Tubes were incubated at 37 °C in a 5% CO2-95% air water-saturated atmosphere with vigorous pipetting every 10-15 min. The enzyme treatment was concluded when the biopsies were no longer macroscopically visible, which in virtually all cases took 2-3 h. Cells were washed in RPMI medium, resuspended in 5 mL of medium, and filtered through a glass wool column. After a final wash, cells were resuspended in 500 μL of medium, counted, and checked for viability and purity. In all cases viability was greater than 95%, and because of prior removal of epithelial cells, enterocyte contamination was minimal (<5%). Two cytocentrifuge preparations per sample were performed (5 000-10 000 cells per slide) and stored until immunostaining as described above.

Enumeration of IFN- γ -, IL-4-, IL-5-, and IL-10-secreting cells. Nitrocellulose-bottomed microtiter wells (Millipore Co., Bedford, MA) were coated for 3 h with cytokine-specific monoclonal capture antibody (0.7-1 μg/well, in 100 μL of PBS). Unadsorbed antibody was removed by three washes with PBS, and the blood and mucosal cells, respectively, were dispensed into the wells and incubated for 20 h in 100-μL volumes at 37 °C in a humidified atmosphere of 5% CO2 in air. Assays involving blood lymphocytes were performed in triplicate for each cytokine and with mucosal cells were done in duplicate. After 20 h of incubation the wells were washed eight times with PBS containing 0.05% Tween 20 (200 μL/well). A biotin-labeled cytokine-specific MAb was then added in 100-μL volumes/well (0.15-0.3 ng/well, diluted in PBS), and incubated for 3 h at room temperature. Capture and detecting antibodies to IFN-γ (capture antibody D1K, detection antibody 7-B6-1) were purchased from Chromogenix AB, Molndal, Sweden, IL-4 antibodies (capture antibody IL-4-I, detection antibody IL-4-II) from Mabtech AB, Stockholm, Sweden, and IL-5 (capture antibody TRFK5, detection antibody TRFK4) and IL-10 (capture antibody JES3-9D7, detection antibody JES3-12G81) antibodies from Pharmingen, San Diego, CA. After eight washes with PBS + Tween, streptavidin alkaline phophatase (1:1000, Mabtech AB, Stockholm, Sweden) was added and incubated for 1-2 h at 37 °C. The wells were washed six times with buffer containing Tris (pH 7.6), and 100-μL aliquots of 5-bromo-4-chloro-3-indoyl phosphate/nitro blue tetrazolium substrate solution (Bio-Rad Laboratories, Hercules, CA) were added. The plates were left at room temperature for 30 min. The enzyme reaction was stopped by washing three times with distilled water. The wells were allowed to dry, and the number of spots per well was enumerated using a dissecting microscope (×25 magnification). The frequency of SPC/105 input cells was then calculated. Most of the counts were made by one of the authors (A.C.H.). The reproducibility and accuracy of the counts were confirmed in selected samples by E.J.B. In some experiments cells were incubated with cycloheximide (50 ng/mL) for 3 or 5 h before the ELISPOT assay was performed as usual. In preliminary experiments the optimal concentrations of all reagents was tested.

Positive control experiments for ELISPOTs. Positive controls for IFN-γ and IL-4 SFC were 48 h phytohemagglutinin-stimulated peripheral blood blasts. LPS-stimulated (1 μg/mL for 24 h) peripheral blood mononuclear cells were used as positive controls for IL-10 SFC. The positive control of IL-5 SFC was a CD4+ T cell line from an individual atopic to Dermatophagoides pteronyssinus, which secretes large amounts of IL-5(31) and was the kind gift of Dr. C. J. Corrigan.

Evaluation of spot size. The sizes of the spots were determined by computer-assisted image analysis (SeeScan, Cambridge, England). Briefly, spots were visualized on a monitor, and their diameters were measured using a mouse-controlled cursor. In wells with many homogenously distributed spots, a total of 20 random spots were measured, whereas in wells with fewer spots each single spot was analyzed.

T cell depletion. T cells were removed from cell suspensions using a magnetic bead separation technique. Briefly, single cell suspensions were prepared, and the percentage of T cells was determined by differential counts of CD3 immunostained cytopreparations. After incubation in an excess of anti-CD3 (UCHT-1, Dako Ltd., High Wycombe, Bucks, UK) for 30 min on ice, cells were washed twice and resuspended in vials containing 1 mL of RPMI/10% FCS. Meanwhile dynabeads in a volume to match the optimal dynabeads to target cell ratio (5-10:1 for mononuclear cell suspensions, dynabead concentration 4 × 108 dynabeads/mL) were washed three times in 1 mL of PBS/2% FCS (pH 7.4) using a magnet and kept on ice. Beads were then added to the cell suspensions and put on a rotator for 60 min at 2-4 °C. The vials with the cell/bead mixture were then placed in the magnetic particle concentrator for 2-3 min to collect the rosetted cells, and the T cell-depleted supernatant (negative fraction) was taken off. This step was repeated twice, and the finally obtained negative fraction was washed, and cells were counted and then used for cytocentrifuge preparations and the ELISPOT assay.

Enumeration of T cells among blood and mucosal mononuclear cells. T lymphocyte percentages were estimated on cytospin preparations stained with monoclonal anti-CD3 antibody (UCHT-1) using an indirect immunoperoxidase method. Briefly, cells were fixed with acetone, after which anti-CD3 antibody was added for 60 min. Peroxidase-conjugated rabbit anti-mouse IgG (dilution 1:50, Dako Ltd.) supplemented by 4% human serum was used as the secondary antibody and was followed by a 20-min incubation period with diaminobenzidine. After a final wash with water, slides were counterstained with Mayer's hemalum (Merck-BDH) for 1-2 min. In control slides where the primary antibody was replaced by a Tris-containing buffer, less than 1% of cells stained positive.

Statistical analysis. Data were first tested to determine whether they conformed to a sample from a normally distributed population. If they did, comparison between the control groups was done by a t test; if the normality test failed, then a two-tailed Mann-Whitney test was used.

RESULTS

Histologic and clinical findings. Histologic assessment revealed patchy villous atrophy and slight crypt hyperplasia in all patients with CMSE, with an increase of intraepithelial lymphocytes in patients 3, 4, 6, 8, and 9, and more pronounced mucosal inflammation (increase of lymphocytes and plasma cells) in patients 8 and 9. All patients showed a quick clinical response to milk exclusion with a normalization of stools and skin within days as well as a continuous weight gain, particularly within the first month (Table 1). All children remained well and continued to thrive on cow's milk-free diet (follow-up 10-16 mo).

Cell yields of mucosal mononuclear cells. A median of 0.27 × 106 mononuclear cells (0.05-0.55 × 106) was obtained per biopsy. T cells made up 41.5% (37.5-53.5%) of the cell preparation. Neither cell yields nor T cell percentages differed significantly between the patient and control groups.

Frequencies of cytokine-secreting cells in peripheral blood mononuclear cells. ELISPOTs for IFN-γ- and IL-4-secreting cells were performed on 29 blood samples (14 CMSE, 10 control, five CMA) and for IL-5 and IL-10 on 27 samples (12 CMSE, 10 control, five CMA) because insufficient blood was available from two of the children with CMSE to perform all the assays. In all groups the same hierarchy of cytokine secretion was found, with the highest frequency for IFN-γ-secreting cells, followed by IL-4, IL-5, and IL-10 (Fig. 1,A-D). However, frequencies were significantly higher in patients with CMSE and CMA when compared with control subjects (IFN-γ, medians of 48 versus 60 versus 13 spots/105 mononuclear cells, p < 0.0001 for control versus CMSE and control versus CMA; IL-4, medians 32 versus 65 versus 6 spots, p < 0.0001 for control versus CMSE and control versus CMA; IL-5, medians of 31 versus 35 versus 2 spots, p < 0.0001 for control versus CMSE and control versus CMA; and IL-10, medians of 4 versus 9 versus 1 spot, p < 0.01 for control versus CMSE and control versus CMA; Fig. 1). There was no difference between children with CMSE and those with CMA for IFN-γ, IL-5, and IL-10, but IL-4 ELISPOTs were significantly raised in CMA compared with CMSE (p < 0.01).

Figure 1
figure 1

Peripheral blood ELISPOTs. The frequency of cells secreting IFN-γ (A), IL-4 (B), IL-5 (C), and IL-10 (D) in peripheral blood mononuclear cells isolated from children with untreated CMSE and CMA, and control subjects. Each point represents a single patient.

Full size image

Frequencies of cytokine-secreting cells in lamina propria mononuclear cells. Twenty-four samples (14 CMSE, 10 control) from duodenal mucosa were assessed for IFN-γ and for IL-4 secretion, and 22 samples (12 CMSE, 10 control) for IL-5 and IL-10. Insufficient cells were recovered from two of the children with CMSE to enable all the assays to be performed. The same hierarchy of cytokine secretion was found as in the blood, with IFN-γ-secreting cells by far outnumbering the other cytokines investigated (Fig. 2,A-D). Note also the very high levels of cytokine-secreting cells in LPL compared with PBL. Patients with CMSE had significantly higher mean frequencies of IFN-γ- and IL-4-secreting cells than controls (IFN-γ, 2785 versus 1182 spots/105 mononuclear cells, p < 0.0001; IL-4, 230 versus 48 spots, p < 0.0001). Examples of this large increase are shown in Figure 3. Frequencies of IL-5-secreting cells were not increased in CMSE compared with controls. Numbers of IL-10-secreting cells were higher in controls than in children with CMSE (13 versus 3 spots, p < 0.001).

Figure 2
figure 2

Lamina propria ELISPOTs. The frequency of cells secreting IFN-γ (A), IL-4 (B), IL-5 (C), and IL-10 (D) in lamina propria mononuclear cells isolated from children with untreated CMSE and CMA and control subjects. Each point represents a single patient.

Full size image
Figure 3
figure 3

IFN-γ and IL-4 ELISPOTs by lamina propria mononuclear cells isolated from a control child (top) and a child with CMSE (bottom). Input cell number was 5000 cells/well for the IFN-γ wells and 40 000 cells for the IL-4 wells (original magnification, ×25).

Full size image

Inhibition of spot formation by cycloheximide. In two children with CMSE, sufficient cells were isolated from LPL to carry out cycloheximide inhibition studies to ensure that we were not measuring cytokines released from the cytoplasm of dead cells or membrane-bound cytokines being released into the medium. Incubation with cycloheximide resulted in a dramatic inhibition of spot formation in a time-dependent fashion. Thus an incubation period of 3 h reduced numbers of IFN-γ SFC by 84.8-92.2%, and those of IL-4 SFC by 85.7-92.5%. An incubation period of 5 h led to a 100% reduction of spot formation (Table 2).

Table 2 Inhibition of spot formation by cycloheximide
Full size table

ELISPOT size. The diameter of the ELISPOTs was used as a semiquantitative way of analyzing the amount of cytokine protein being secreted by individual cells from various compartments (Fig. 4). In the blood of the CMSE patients, the diameters of IFN-γ spots were significantly greater than in CMA and control subjects (1.35 versus 0.60 versus 0.81, p < 0.001). IL-4 ELISPOTs were larger in children with CMA than those with CMSE (1.45 versus 1.05, p < 0.0001). The increase in IL-4 ELISPOT size in the blood of children with CMA was particularly striking and is shown in Figure 5. In the lamina propria, IFN-γ, IL-4, and IL-5 spots were significantly larger in CMSE patients than in control subjects (IFN-γ, 0.78 versus 0.55, p < 0.001; IL-4, 1.08 versus 0.34, p < 0.001; and IL-5, 1.22 versus 0.30, p < 0.001).

Figure 4
figure 4

Size of the ELISPOTs in peripheral blood mononuclear cells of children with CMSE and CMA, and control subjects, and the lamina propria mononuclear cells of children with CMSE and control subjects.

Full size image
Figure 5
figure 5

IL-4 ELISPOTs by peripheral blood mononuclear cells from a control child (top) and a child with CMA (bottom). Note that, as well as increased numbers of ELISPOTs in the child with CMA, the spots are also larger.

Full size image

T cell depletion reduces the frequency of ELISPOTs in PBL and LPL. Because cytokines can be produced by cells other than T cells, we used magnetic beads to deplete PBL and LPL of T cells and determined the frequency of ELISPOTs in both fractions. Attempts were also made to positively select T cells from LPL, but a combination of low cell yields and difficulty in removing the magnetic beads from the cells made this impossible. It is clear, however, that in the three children studied, T cell depletion markedly reduced the frequency of all cytokine-secreting cells in both LPL and PBL (Table 3).

Table 3 The effect of T cell depletion on cytokine secreting cells
Full size table

DISCUSSION

In this work we have used ELISPOT assays to determine the frequency of cells secreting IFN-γ, IL-4, IL-5, and IL-10 in blood of children with CMSE or CMA or controls and in duodenal lamina propria of children with CMSE and controls. ELISPOT assays are a well established method of measuring cytokine synthesis. They have the benefit that they allow cells secreting different cytokines to be compared with a common denominator, i.e. cell number, but have the disadvantage that one has very little idea how much cytokine each cell is secreting. This is the reverse problem of assays measuring cytokines in supernatants where one knows the exact concentration of cytokine but not how many cells are secreting the molecule. We also chose to investigate the number of spontaneous cytokine-secreting cells freshly isolated from untreated children, to try to determine what effects a milk-containing diet was having in vivo.

In blood, children with CMA and CMSE had increased frequencies of ELISPOTs for all cytokines tested compared with control subjects. Moreover, children with CMA had higher spontaneous IL-4-secreting cell numbers than children with CMSE, and the ELISPOTs were also considerably larger. It has previously been shown that lymphocytes from children who react immediately to cow's milk produced less IFN-γ when stimulated in vitro with β-lactoglobulin or after polyclonal activation than lymphocytes from children with delayed symptoms to milk(25, 26). It has also been reported that diminished in vitro production of IFN-γ may even precede the development of atopy(32, 33). Based on this we would have expected to find decreased IFN-γ ELISPOTs in CMA; however, this clearly is not the case. We consider it unlikely that malnutrition had any effect on our results. In rats protein malnutrition impairs IFN-γ production(34); however, the children with CMSE we studied were not overtly malnourished, although they were around the 3rd centile for weight. In addition, if malnutrition was playing a role one would expect diminished responses in children with CMSE, and in fact we found enhanced responses. We can think of no ready explanation for these differences, only to emphasize the problems of extrapolating in vitro data to the in vivo situation.

Although we carried out T cell depletion only on two children with CMSE and a single control subject, this procedure markedly reduced all cytokine ELISPOTs in blood and LPL. Thus the IL-4- and IL-5-secreting cells are probably T cells rather than a few contaminating basophils or eosinophils(34–36), and the IFN-γ-secreting cells are also likely to be T cells rather than natural killer cells(37). The cytokine-secreting cells in the blood may be T blasts stimulated by milk antigens in the Peyer's patches migrating back to the rest of the mucosa(38). It is of interest that, in children with milk-induced eczema, T cells activated in vitro with casein preferentially express the cutaneous lymphocyte antigen, which may enable them to home into skin(39). Perhaps, in children with CMSE without cutaneous manifestations, the activated lymphocytes express α4β7 integrin, which by interacting with mucosal addressin cell adhesion molecule on mucosal endothelium might allow these cells to extravasate into the lamina propria(40, 41).

In the duodenum, where we were able to investigate only control subjects and patients with CMSE, IFN-γ and IL-4 ELISPOTs were elevated in CMSE, IL-5 was unchanged, and IL-10 was reduced. Again, in the two patients with CMSE studied, T cell depletion of LPL virtually eliminated IFN-γ and IL-4 ELISPOTs, suggesting that these cytokines were being produced by T cells. Because in controls, IFN-γ-secreting cells are also dominant, our results demonstrate an exaggeration of the normal response in CMSE. This is different from inflammatory bowel disease, where there is reduction in IL-4-secreting cells(42). Previously, using a different technique of lymphokine plaque-forming cells, we have also demonstrated an increased frequency of IFN-γ-secreting cells in celiac disease and Crohn's disease(18, 43), both conditions in which there is accumulating evidence that exaggerated Th1-type hypersensitivity reactions are responsible for tissue injury. The results with ELISPOT in CMSE are also suggestive of a Th1-type response in the mucosa. Mouse models of inflammatory bowel disease have also shown that Th1 reactions to the normal flora are tightly regulated so that, in IL-2 and IL-10 knockout mice and severe combined immunodeficiency disease mice reconstituted with CD4 cells, inappropriate IFN-γ responses develop to the gut flora, resulting in epithelial cell hyperplasia(44–47).

Some comment should also be made concerning the very high frequency of T cells secreting IFN-γ in normal lamina propria. Lamina propria T cells are thought to be derived from Peyer's patches and mesenteric lymph node T cell immunoblasts responding to lumenal antigens(38) and have a surface phenotype consistent with recent prior activation, such as L-selectinlo, CD45R0+, CD103+, and CD69+(48–50). It has been suggested that Peyer's patch T cells preferentially develop into Th2-type T cells when responding to lumenal antigens(51, 52). If so, it would appear that the subsets which develop into Th1-type cells are migrating to the lamina propria, and that perhaps the Th2-type cells migrate into the periphery to mediate oral tolerance(53, 54). The production of IFN-γ by lamina propria T cells in health probably contributes to the induction of class II major histocompatibility complex molecules on enterocytes and the activated state of lamina propria macrophages(55). More importantly it suggests that the increased production of IFN-γ seen in the mucosa in Crohn's disease and celiac disease represents an exaggeration of the normal physiologic response rather than a novel type of response.

In summary therefore we have shown increased IFN-γ- and IL-4-secreting cells in the blood and intestinal mucosa of children with CMSE. The important fact that IFN-γ-secreting cells are about 10-fold more abundant than IL-4-secreting cells suggests that in CMSE there is an exaggerated Th1-type response, probably to CMPs.