Eosinophilic Gastritis in Children

Clinicopathological Correlation, Disease Course, and Response to Therapy

Huaibin M Ko MD; Raffaella A Morotti MD; Oksana Yershov MS; Mirna Chehade MD; MPH


Am J Gastroenterol. 2014;109(8):1277-1285. 

In This Article


This study represents the largest clinical experience with HEG in children and EG in particular. EG is indeed rare, and although often classified as a separate entity by most investigators, it has been rarely studied. We found one previous pediatric study addressing EG that included 10 children.[10] It is possible that the disease is under-diagnosed. Indeed, our patients with EG had very variable symptomatology. In addition, gastric endoscopic findings were highly variable in our cohort, and were subtle or non-existent in some patients, highlighting the importance of obtaining biopsies even in the face of normal gastric mucosal appearance when EG is clinically suspected. In addition, we found that gastric eosinophilia could be patchy in EG. Some of the children in our cohort with HEG had eosinophilia in the fundic but not in the antral mucosa, the latter being the more commonly sampled site by pediatric gastroenterologists. These findings are in contrast to what has been reported on gastric eosinophilic involvement in children with EGE where eosinophilic infiltration was prominent in the gastric antrum.[3] Our findings highlight the importance of obtaining both antral and fundic biopsies when HEG is suspected.

Interestingly, peripheral eosinophil counts were normal in 25% of children in our HEG cohort and in 50% of those with EG, suggesting that this test is not a reliable diagnostic criterion. Therefore, absence of peripheral eosinophilia should not be a deterrent from performing an upper endoscopy with biopsies as a work-up for HEG/EG when clinically suspected. Although published data on EG are not available for comparison, data on EGE are variable, with reports of peripheral eosinophilia ranging from 23 to 86% of patients,[3,9,13] emphasizing the fact that a normal peripheral eosinophil count does not rule out GI eosinophilia. In addition, we found that peripheral eosinophil count was not a reliable marker of disease activity, and therefore not reliable for following up patients after therapy, as our patients with peripheral eosinophilia continued to have similarly elevated eosinophil counts regardless of their histological response to therapy.

We found no gender predilection in our pediatric cohort with HEG (male/female ratio of 0.9:1) or with EG in particular (male/female ratio of 0.8:1). This finding is contrary to EGE, where a slight male predominance (male/female ratio between 1.3:1 and 1.7:1) has been reported,[3,9,11,13] and EoE, where a striking male predominance (3:1) was found.[12,14] Most of our pediatric patients were young (almost half below 6 years of age), similar to findings in pediatric EGE.[3] This is different again from EoE, where the disease is seen in all pediatric age groups.[12] As described for EoE,[14] our cohort with HEG and EG in particular seemed to have race predilection toward Caucasians, despite the fact that our center serves a very diverse population. This race predilection has also been reported for EGE in children.[3] Interestingly, five of the six excluded children for history of solid organ transplant patients with history of previous immunomodulation were non-Caucasians.

In our cohort, HEG was frequently associated with EoE. This finding was of particular interest; incidental mild gastric eosinophilia has been reported in patients with EoE,[15] but concomitant presence of severe gastric eosinophilia with EoE has not heretofore been described as its own entity, and warrants further study. The association between HEG and EoE in our cohort was found to be either synchronous or metachronous. Associated EoE was more resistant to dietary restriction therapy than gastric eosinophilia, and even developed de novo in patients with HEG that was non-responsive to therapy.

Talley et al. examined histological and/or radiologically confirmed EGE (broadly defining EGE to include eosinophilic infiltration anywhere in the GI tract from esophagus to colon) in 40 adults and found many patients with multiple sites of disease.[13] Some of our children with gastric eosinophilia also had eosinophilia in contiguous organs such as the esophagus or the duodenum. However, eosinophilia in non-contiguous areas such as the colon was only seen in those children with history of immunomodulation following solid organ transplantation. These findings may indicate that a colonoscopy with biopsies may not be needed in the absence of lower GI symptoms, unless the patient has history of prior immunomodulation therapy.

PLE is characterized by loss of plasma proteins and blood in the GI tract, which can manifest clinically as anemia and edema. In our HEG cohort, there was a relatively high association with PLE (22%), a finding that has been previously described in 8.6% of children with EGE.[4,16] PLE has also been reported in adults with EGE to various degrees (2–30%),[9,13] the exact site of prominent eosinophilic infiltration is not specified in those reports. We found evidence of PLE in our cohort even without evidence of duodenal eosinophilia in some patients. The mechanism for protein and blood loss in these patients is unknown. It is possible that intestinal eosinophilia was potentially missed in these patients because of the inherent limitations of endoscopically obtained biopsies. These include the small size of biopsies, the limited depth of tissue that can be sampled, and the limited distance through the small intestine that the endoscope can reach. Alternatively, we tested the hypothesis that PLE may be secondary to intestinal mast cell rather than eosinophil infiltration, thought to increase intestinal permeability, hence disrupting the functional integrity of the intestinal barrier.[4] This hypothesis is supported by a separate report that observed an association between mastocytosis, upper GI erosions, and PLE.[17] We found this not likely to be the case, although our conclusion is limited by the small number of cases that were available for mast cell immunohistochemical staining.

Atopy and history of food allergies were prevalent in our cohort with HEG including those with eosinophilia limited to the stomach, similar to what has been observed in EoE and EGE.[4,11,12,13] In addition, multiple food sensitizations, as evidenced by positive skin prick tests and/or serum food-specific IgE, were present in our patients, as seen in patients with EoE and EGE.[12,13] However, foods to which patients were allergic or were merely sensitized were not necessarily the culprit triggers for the disease, as patients were already avoiding the foods they were allergic to, to avoid risk of anaphylaxis, and we saw no correlation between response to specific dietary restrictions and food sensitization results. In fact, test-based dietary elimination therapy was also previously tried in children and adults with EGE with minimal response.[4,9,13] Similar to EoE, the mechanism by which foods trigger EG and EGE is likely predominantly not IgE-mediated but rather cell-mediated, hence, the low predictive value for standard allergy tests such as skin prick tests and serum food-specific IgE level determinations.[18] Empiric dietary elimination therapy, consisting of removal of common food triggers established for EoE,[19] and very restrictive therapies, consisting of amino acid-based formula administration with a few foods, were effective in our cohort. These results indicate that EG may be amenable to dietary restriction therapies in children, similar to what has been observed in children with EoE and EGE.[4,12] This also includes patients with prior history of solid organ transplantation, as we saw response to similar dietary therapies in those patients as well. Our transplanted patients were maintained on tacrolimus, which has been associated with development of food allergy-induced gastroenterocolitis in a previous report.[20] Large multi-center systematic prospective trials using dietary restriction therapies for patients with EG are needed both in children and adults to confirm these findings.

In conclusion, EG in the pediatric population is highly variable in presentation both clinically and endoscopically, and may involve various, at times specific, areas of the gastric mucosa, highlighting the importance of obtaining both gastric antral and fundic biopsies when the diagnosis of EG is suspected. Peripheral eosinophilia is not a reliable marker for diagnosis or follow-up of these patients. Severe gastric eosinophilia can also be associated with PLE in the absence of duodenal eosinophilia on biopsies, and can be associated with eosinophilia in other parts of the GI tract, especially the esophagus. Associated EoE is more resistant to dietary restriction therapy than gastric eosinophilia. Children with HEG with or without concurrent eosinophilia at other GI sites are highly responsive to dietary restriction therapy. Standard allergy testing is not predictive of the food triggers in this disease, pointing toward a cell-mediated rather than an IgE-mediated mechanism. An empiric dietary restriction therapy consisting of removal of common food triggers established for EoE or a very restricted elemental diet is warranted in this population.