Advertisement

Four distinct subtypes of non–IgE-mediated gastrointestinal food allergies in neonates and infants, distinguished by their initial symptoms

      To the Editor:
      Although most food allergies are IgE-mediated, there are a number of non–IgE-mediated gastrointestinal food allergies that affect mainly infants and young children.
      • Nowak-Wegrzyn A.
      • Murano A.
      Food protein-induced enterocolitis syndrome.
      • Sicherer S.H.
      • Sampson H.A.
      Food allergy.
      Because most such patients develop the allergy more than 2 hours after ingestion of the offending food and show negative skin prick tests and/or absence of serum specific IgE against the offending food, these allergies are thought to be cell-mediated. However, the precise pathogenetic mechanisms of these disorders remain poorly understood. Several investigators have described different subtypes of non–IgE-mediated gastrointestinal food allergies: food protein-induced enterocolitis syndrome (FPIES),
      • Powell G.K.
      Milk- and soy-induced enterocolitis of infancy: clinical features and standardization of challenge.
      food protein-induced proctocolitis syndrome (hereafter referred to as “proctocolitis”),
      • Lake A.M.
      Food-induced eosinophilic proctocolitis.
      food protein-induced enteropathy syndrome (hereafter referred to as “enteropathy”),
      • Savilahti E.
      Food-induced malabsorption syndromes.
      celiac disease, and allergic eosinophilic gastroenteropathies.
      Presumably because the main target organ of these syndromes is the gastrointestinal tract, patients with non–IgE-mediated gastrointestinal food allergies often exhibit similar symptoms, such as vomiting and diarrhea. However, it remains unclear whether these syndromes have the same pathogenesis and merely differ in severity, or whether the pathogenesis of each is distinct, meaning that they should be classified as separate clinical entities.
      We applied cluster analysis to the clinical and laboratory findings to characterize these non–IgE-mediated food allergies and determine whether they are made up of distinct clinical entities. A total of 176 patients with detailed clinical records who had been registered in the database of the Japanese Research Group for Neonatal, Infantile Allergic Disorders from 2007 to 2010 were enrolled. Among them, 136 patients fulfilled 3 of the Powell
      • Powell G.K.
      Food protein-induced enterocolitis of infancy: differential diagnosis and management.
      criteria: (1) a switch to therapeutic milk led to resolution of symptoms, (2) differential diagnosis from other disorders was possible, and (3) there was verified body weight gain. Definitive diagnosis was possible for 46 patients by oral food challenge tests that were performed after complete resolution of the initial symptoms (see this article’s Fig E1 in the Online Repository at www.jacionline.org). These 46 patients were subjected to further analysis. Details of food challenge test are available in this article’s Food challenge test, method section in the Online Repository at www.jacionline.org. Our total cohort included 15 patients who developed the most severe reactions, including ileus, shock, and developmental retardation. The clinical characteristics of those patients are summarized in this article’s Table E1 in the Online Repository at www.jacionline.org. Because of the medical and ethical justification, even though these patients fulfilled 3 elements of the Powell
      • Powell G.K.
      Food protein-induced enterocolitis of infancy: differential diagnosis and management.
      criteria, oral challenge tests were not performed. Thus, these patients were excluded from this cluster analysis of 46 patients. This study was approved by the Ethics Committee of the National Center for Child Health and Development.
      We omitted clinical and laboratory findings found only in a few patients and finally selected 5 variables: birth weight, age at first presentation (days after birth), severity of vomiting (ranked as 0, none; 1, 1-2 times a day; 2, 3-5 times a day; and 3, more than 5 times a day or bilious vomiting) and severity of bloody stool (0, none; 1, spotty; 2, intermediate; and 3, massive) at first presentation, and milk-specific IgE antibody titer (class 0-6). Unsupervised cluster analysis and discriminant analysis were performed by using SPSS version 18 software (SPSS, Inc, Chicago, Ill). The Wald minimum-variance hierarchic clustering method was performed by using an agglomerative (bottom-up) approach and Ward's linkage. The squared Euclidean distance was used as a proximity measure. Values were transformed by a maximum magnitude of 1. ANOVA, the Tukey-Kramer test, and the χ2 test were used for parametric continuous, nonparametric continuous, and categoric variables. As a result, the 46 definitively diagnosed patients were classified into 4 distinct clusters, and a dendrogram was generated (see this article’s Fig E2 in the Online Repository at www.jacionline.org).
      Stepwise discriminate analysis identified the 2 strongest discriminatory variables for cluster assignment: vomiting and bloody stool (Fig 1). Cluster 1 was the patient group with vomiting and bloody stool at initial presentation. Cluster 2 had vomiting but not bloody stool. Cluster 3 had neither vomiting nor bloody stool. Cluster 4 had bloody stool but not vomiting. One patient initially assigned to cluster 3 in fact had clear bloody stool, and was thus reassigned to cluster 4 in accordance with Fig 1. As a result, clusters 1 through 4 consisted of 14, 16, 5, and 11 patients, respectively.
      Figure thumbnail gr1
      Fig 1Tree analysis using 2 variables (vomiting and bloody stool at initial presentation) enables assignment of patients into 4 clusters.
      Table I presents the demographic data for each cluster. Cluster 3 showed a significantly lower birth weight and later onset of disease. Clusters 1 and 4 both had bloody stool, but they had normal birth weight and a somewhat earlier onset (median of 7 days after birth).
      Table IDemographic data of the patients (total = 46) whose diagnosis was confirmed by oral food challenge tests
      Cluster 1Cluster 2Cluster 3Cluster 4P
      Clinical characteristics(n = 14)(n = 16)(n= 5)(n = 11)value
      Birth weight (g)2642 (2410-3030)2745 (2223-3079)1008 (907-2491)2678 (2512-3170).03
      P < .05.
      Male/female (n)6/89/72/35/6.95
      Initial presentation
       Day of onset7.5 (3-23)16.5 (9.5-33.5)37 (8.5-132)7 (2-56).17
       Vomiting (%)10010000.000
      P < .05.
       Bloody stool (%)10000100.000
      P < .05.
       Fever (%)7.118.820.00.45
       (Laboratory data)
      n, Number with medical records.
      nnnn
       Blood eosinophil ratio (%)
      Normal range of blood eosinophils is 0% to 4%. However, it is known to rise to some degree in the neonatal period, especially in low-birth-weight infants.10
      1315 (3.0-23)147 (3.9-19.3)527 (3.2-39.3)1114 (4.5-25).63
       WBC (×103/mL)
      Normal range of WBC in neonatal period is 7.0 to 25.0 × 103/μL.
      1318.4 (13.7-22.7)1415.7 (11.4-21.9)521.8 (11.0-27.7)1113.1 (8.2-18.3).64
       Total IgE (IU/mL)
      Normal range of total IgE in infantile period is less than 20 IU/mL.
      145.2 (4.8-28.3)1611.4 (5.0-80.8)57.4 (5.5-653.5)105.0 (2.0-5.8).36
       Positive for milk-specific IgE (class ≥1) (%)14571637.5540119.28
       C-reactive protein (% positive, ≥0.5)134614505401030.47
       Stool eosinophil (% positive)850633307100.01
      P < .05.
      Diet (reaction to each milk, %)
       Cow’s milk14100161005100101001.00
       Breast milk83870250727.40
       Hydrolyzed formula90102020863.02
      P < .05.
      Oral food challenge test
       Onset of reaction (h)6 (1.8-12)10 (2-24)48 (24-60)24 (24-48).17
       Vomiting (%)85.781.309.1.000
      P < .05.
       Bloody stool (%)28.66.3072.7.001
      P < .05.
       Diarrhea (%)21.431.360.018.2.33
      WBC, White blood cell count.
      Data are shown as the median and the interquartile range.
      P < .05.
      n, Number with medical records.
      Normal range of blood eosinophils is 0% to 4%. However, it is known to rise to some degree in the neonatal period, especially in low-birth-weight infants.
      • Yen J.M.
      • Lin C.H.
      • Yang M.M.
      • Hou S.T.
      • Lin A.H.
      • Lin Y.J.
      Eosinophilia in very low birth weight infants.
      § Normal range of WBC in neonatal period is 7.0 to 25.0 × 103/μL.
      Normal range of total IgE in infantile period is less than 20 IU/mL.
      The laboratory data generated within the initial several days after onset showed that the peripheral blood eosinophil ratio was high in all clusters, with no significant differences among them. In contrast, eosinophils were found in the stool mainly of patients in clusters 1 and 4, in which all patients, by definition (Fig 1), had bloody stool. The presence of eosinophilia suggests that patients with non–IgE-mediated gastrointestinal food allergies tend to have a Th2-prone immune deviation at baseline, but some additional factors such as overproduction of eosinophil-attracting chemokines are probably necessary to induce immune responses involving eosinophils in the gut (see this article’s Fig E3 in the Online Repository at www.jacionline.org).
      A positive milk-specific IgE antibody titer was observed in 37% of the patients, with no statistically significant differences among any of the clusters. In addition, almost all symptoms at initial presentation as well as in oral food challenge tests began to manifest at more than 2 hours after ingestion of the offending food, whereas no patients developed typical IgE-mediated symptoms such as urticaria or wheeze. These results strongly suggest that the presence of milk-specific IgE antibody neither causes the gastrointestinal symptoms nor rules out a diagnosis of non–IgE-mediated gastrointestinal food allergy.
      One of the most notable findings of this study was the remarkably high reproducibility of symptoms provoked in the oral food challenge tests and those found at the initial presentation in all 4 clusters, even though the oral challenge tests were performed several months after the initial presentation (Table I). This observation suggests that the upper or lower gastrointestinal tract–specific hypersensitivity and perhaps the responsible immune cells remain in the same part of the gastrointestinal tract even after several months’ remission.
      Because the patients in clusters 1 and 2 had vomiting that was provoked at relatively early time points, they are likely to be diagnosed as having FPIES, although the bloody stool and eosinophilia seen mainly in cluster 1 patients were not emphasized in earlier reports.
      • Mehr S.
      • Kakakios A.
      • Frith K.
      • Kemp A.S.
      Food protein-induced enterocolitis syndrome: 16-year experience.
      • Hwang J.B.
      • Sohn S.M.
      • Kim A.S.
      Prospective follow-up oral food challenge in food protein-induced enterocolitis syndrome.
      The nearly simultaneous manifestation of vomiting and bloody stool suggests that FPIES may affect both the upper and lower gastrointestinal tracts.
      The main symptoms of the patients in cluster 3 were poor weight gain and diarrhea and were similar to those found in patients with enteropathy. The significantly lower birth weight and marked eosinophilia characteristically found in cluster 3 patients imply the involvement of immature gastrointestinal function in the pathogenesis of this syndrome.
      Bloody stool was the main symptom of the patients in cluster 4. Some patients in this cluster had no systemic manifestation other than bloody stool, whereas others also had diarrhea and/or poor weight gain. Therefore, these patients may be diagnosed as having proctocolitis or early onset of allergic eosinophilic gastroenteropathies, respectively. However, the pathogenetic similarity and/or disparity of proctocolitis and allergic eosinophilic gastroenteropathies need to be studied further.
      In our cohort, 3 children with exclusive breast-feeding have developed FPIES. This information is available in this article’s Breast-feeding and FPIED section in the Online Repository at www.jacionline.org.
      Elevated serum C-reactive protein levels were found in 30% to 50% of patients with non–IgE-mediated gastrointestinal food allergies. In addition, some patients developed a fever during oral food challenge tests, suggesting that TNF-α and other proinflammatory cytokines may be involved in the pathogenesis of these syndromes.
      • Chung H.L.
      • Hwang J.B.
      • Park J.J.
      • Kim S.G.
      Expression of transforming growth factor beta1, transforming growth factor type I and II receptors, and TNF-alpha in the mucosa of the small intestine in infants with food protein-induced enterocolitis syndrome.
      To confirm the results of cluster analysis, we performed the same analysis for the aforementioned 136 patients who fulfilled 3 of the Powell
      • Powell G.K.
      Food protein-induced enterocolitis of infancy: differential diagnosis and management.
      criteria (consisting of the 46 patients definitively diagnosed by oral food challenge and 90 patients not subjected to oral food challenge; Fig E1). We obtained exactly the same results: the patients were assigned to 4 clusters in accordance with the tree analysis shown in Fig 1. The patients’ demographics (see this article’s Table E2 in the Online Repository at www.jacionline.org), birth weight (see this article’s Fig E4 in the Online Repository at www.jacionline.org) and peripheral blood eosinophils (see this article’s Fig E5 in the Online Repository at www.jacionline.org) confirmed the earlier cluster analysis findings.
      In our ongoing cohort, 52% of the patients acquired tolerance to the offending food by 1 year of age, 88% by 2 years, and 94% by 3 years. Therefore, assuming that identification and elimination of the offending food had been done properly, it can be assumed that most patients outgrew their allergy by the age of 2 to 3 years. On the other hand, just like patients with severe IgE-mediated food allergy, patients with non–IgE-mediated gastrointestinal food allergies may develop severe reactions (Table E1). Thus, early diagnosis is very important, and refinement of the diagnostic method is truly necessary.
      Our findings clearly demonstrated that patients with these non–IgE-mediated gastrointestinal food allergies showed similar Th2-prone laboratory data (eosinophilia and presence of specific IgE antibody), but the disease entities of each cluster had distinct clinical features and may have different pathogenetic mechanisms.
      We express our sincere gratitude to all the members of the Japanese Research Group for Neonatal, Infantile Allergic Disorders. We also thank all the doctors, nurses, and technicians in the Division of Allergy, Gastroenterology, Pathology, Surgery, Interdisciplinary Medicine and Neonatology of the National Center for Child Health and Development for their hard work and invaluable comments.

      Food challenge test, method

      Generally, oral challenge tests were performed at 4 to 6 months of age. First, 4 mL milk/kg body weight was administered. If no reaction occurred, the dose of milk was increased daily until symptoms manifested. If the reaction had been evoked by a very small volume of milk in the initial presentation, the test was started using a lesser volume to avoid a serious reaction. Because of the medical and ethical justification for oral food challenge tests, patients with the most severe reactions were excluded from the initial cluster analysis. Their clinical characteristics are summarized in Table E1.

      Breast-feeding and FPIES

      Six of the 46 patients were exclusively breast-fed. Three of them were included in cluster 1 and can be diagnosed as FPIES. Those 3 patients showed a positive reaction to cow’s milk as well as breast milk even after their mothers stopped consuming milk products. These patients also developed symptoms when orally challenged with rice and/or soy. Therefore, these findings indicate that not only proctocolitis but also FPIES can develop even in children who are exclusively breast‑fed. A recent case report supports our findings.
      • Monti G.
      • Castagno E.
      • Alfonsina S.L.
      • Lupica M.M.
      • Tarasco V.
      • Viola S.
      • et al.
      Food protein–induced enterocolitis syndrome by cow’s milk proteins passed through breast milk.

      Fig E1.

      Figure thumbnail fx1
      A total of 176 patients with gastrointestinal symptoms who were suspected of having non–IgE-mediated allergy from 1999 to 2009 were registered by doctors of the Japanese Research Group for Neonatal, Infantile Allergic Disorders. Of them, 136 patients fulfilled elements 1 through 3 of the Powell criteria. Forty-six patients underwent food challenge tests and had a positive result, whereas the remaining 90 patients were not tested. Seventeen patients showed no reaction in the oral challenge tests. However, it was unclear whether this was because the patients had outgrown their allergy or because of misdiagnosis. Those 17 patients were excluded from further analysis in this study.

      Fig E2.

      Figure thumbnail fx2
      The 46 patients definitively diagnosed with non–IgE-mediated food allergies were analyzed for 5 variables by using an agglomerative (bottom-up) approach and Ward's linkage, and a dendrogram was generated.

      Fig E3.

      Figure thumbnail fx3
      Detection of accumulations of eosinophils in the stool mucus. The mucous part of the stool was thinly smeared on a glass slide and stained by using Hansel stain. The stool sample was taken from a patient in cluster 2 after a positive food challenge test. Representative images were found in a total of 13 patients ().

      Fig E4.

      Figure thumbnail fx4
      The birth weights in each cluster of the 136 patients who fulfilled 3 elements of the Powell criteria for a non–IgE-mediated allergy are shown.
      • Powell G.K.
      Food protein-induced enterocolitis of infancy: differential diagnosis and management.
      The birth weights in cluster 3 were significantly lower than in the other clusters. Moreover, 2 subgroups seem to be identified in cluster 3: a lower birth weight group and a normal birth weight group.

      Fig E5.

      Figure thumbnail fx5
      The peripheral blood eosinophil ratios in each cluster of the 136 patients who fulfilled 3 elements of the Powell criteria are shown.
      • Powell G.K.
      Food protein-induced enterocolitis of infancy: differential diagnosis and management.
      Although eosinophilia was found in all 4 clusters, the eosinophil ratios of the patients in cluster 3 were significantly higher than those of the patients in clusters 1 and 2.

      Table E1.

      Tabled 1Clinical features of most severe cases of non–IgE-mediated gastrointestinal food allergies
      These patients fulfilled 3 elements of the Powell criteria,E2 but oral challenge tests were not performed.
      Patient no.SexClusterComplicationDay of onsetDiet right before the onset of complicationsRemarks
      1F1Ileus8Cow’s milk 7 d
      2M1Ileus5Cow’s milk 3 d, breast milk 6 dRelieved by surgical operation
      3F1Ileus8Breast milk 9 dRelieved by surgical operation
      4F1Shock2Cow’s milk 2-3 timesMassive bloody stool, blood infusion required
      5F1Shock21Breast milk 18 dMassive bloody stool, disseminated intravascular coagulation
      6F2Ileus14Breast milk 2 d
      7F2Shock36Breast milk 30 dApnea, vomiting
      8M2Shock30Cow’s milk 50 mL by chanceVomiting
      9M2Shock241Soy food 2-3 timesVomiting and diarrhea, ICU admission
      10M3Ileus61Breast milk 45 dCholestasis
      11F3Shock22Cow’s milk 21 d, breast milk 21 dICU admission
      12F3Severe weight loss12Breast milk several monthsDevelopmental retardation
      13M3Severe weight loss46Cow’s milk 30 d, breast milk 30 dDevelopmental retardation
      14F4Ileus2Cow’s milk 6 d, breast milk 3 dStenosis of sigmoid colon
      15F4Ileus7Cow’s milk 10 d
      F, Female; ICU, intensive care unit; M, male.
      These patients fulfilled 3 elements of the Powell criteria,
      • Powell G.K.
      Food protein-induced enterocolitis of infancy: differential diagnosis and management.
      but oral challenge tests were not performed.

      Table E2.

      Tabled 1Demographics of the 136 patients who fulfilled 3 elements of the Powell criteria
      • Powell G.K.
      Food protein-induced enterocolitis of infancy: differential diagnosis and management.
      Cluster 1Cluster 2Cluster 3Cluster 4P
      Clinical characteristics(n = 38)(n = 43)(n = 24)(n = 31)value
      Birth weight (g)2823 (2501-3267)2581 (1779-3016)1363 (1023-2611)2778 (2512-3100).000
      P < .05.
      Male/female19/1928/1513/1112/19.16
      Initial presentation
       Day of onset6 (4-8)29 (7.5-52)16.5 (9.5-37.5)7 (2-35).01
      P < .05.
       Vomiting (%)10010000.000
      P < .05.
       Bloody stool (%)10000100.000
      P < .05.
      (Laboratory data)
      n, Number with medical records.
      nnnn
      Blood eosinophil ratio (%)3515 (3.5-21.0)409 (5.3-25.0)2026 (14.1-39.3)2817 (8.5-23.8).005
      P < .05.
      WBC (×103/mL)3218.7 (14.5-23.5)4013.8 (10.4-22.1)2315.9 (13.9-24.4)2713.9 (11.4-19.5).16
      Total IgE (IU/mL)325.2 (4.1-23.1)405.8 (4.0-17.8)2213.2 (5.5-122.9)285.0 (3.3-6.0).001
      P < .05.
      Positive for milk-specific IgE (class ≥1) (%)3141.93823.720502719.24
      C-reactive protein (% positive, ≥0.5)3661404520702733.69
      WBC, White blood cell count.
      Data are shown as the median and the interquartile range.
      P < .05.
      n, Number with medical records.

      References

        • Nowak-Wegrzyn A.
        • Murano A.
        Food protein-induced enterocolitis syndrome.
        Curr Opin Allergy Immunol. 2009; 9: 371-377
        • Sicherer S.H.
        • Sampson H.A.
        Food allergy.
        J Allergy Clin Immunol. 2010; 125: S116-S125
        • Powell G.K.
        Milk- and soy-induced enterocolitis of infancy: clinical features and standardization of challenge.
        J Pediatr. 1978; 93: 553-560
        • Lake A.M.
        Food-induced eosinophilic proctocolitis.
        J Pediatr Gastroenterol Nutr. 2000; 30: S58-S60
        • Savilahti E.
        Food-induced malabsorption syndromes.
        J Pediatr Gastroenterol Nutr. 2000; 30: S61-S66
        • Powell G.K.
        Food protein-induced enterocolitis of infancy: differential diagnosis and management.
        Compr Ther. 1986; 12: 28-37
        • Mehr S.
        • Kakakios A.
        • Frith K.
        • Kemp A.S.
        Food protein-induced enterocolitis syndrome: 16-year experience.
        Pediatrics. 2009; 123: e459-e464
        • Hwang J.B.
        • Sohn S.M.
        • Kim A.S.
        Prospective follow-up oral food challenge in food protein-induced enterocolitis syndrome.
        Arch Dis Child. 2009; 94: 425-428
        • Chung H.L.
        • Hwang J.B.
        • Park J.J.
        • Kim S.G.
        Expression of transforming growth factor beta1, transforming growth factor type I and II receptors, and TNF-alpha in the mucosa of the small intestine in infants with food protein-induced enterocolitis syndrome.
        J Allergy Clin Immunol. 2002; 109: 150-154
        • Yen J.M.
        • Lin C.H.
        • Yang M.M.
        • Hou S.T.
        • Lin A.H.
        • Lin Y.J.
        Eosinophilia in very low birth weight infants.
        Pediatr Neonatal. 2010; 51: 116-123