The Association Between Celiac Disease and Eosinophilic Esophagitis in Children and Adults

Michael J Stewart; Eldon Shaffer; Stephan J Urbanski; Paul L Beck; Martin A Storr


BMC Gastroenterol. 2013;13(96) 

In This Article


Celiac disease and EoE are two clinically, anatomically, and histologically distinct disorders of the upper gastrointestinal tract. An association between EoE and CD has been previously suggested. Contrary to what we expected to find in the current population-wide analysis that included a robust number of adults, this association may be solely a pediatric phenomenon. Although the overall number of patients with EoE and CD was low, no adults were diagnosed with both conditions, even though the majority of patients diagnosed with either condition were adults. Further, all three cases of EoE and CD were male. Although EoE is a predominantly male condition, CD is a predominantly female condition. In contrast to our findings, the Australian reports found that less than half of their patients with both diagnoses were males.[32,33]

This study confirms the association between EoE and CD in pediatric patients. The prevalence of EoE in our five-year pediatric CD cohort was somewhat lower than reported in two previous studies despite our pediatric CD cohort being much larger than the previous studies (1.2% vs. 3.2% and 4.0%).[32,33] It was, however, similar to the prevalence described in the New Your City CD cohort (0.97%).[35] It is difficult to compare these results directly, or to other published prevalence rates, and it should be stressed that extracting prevalence estimates from a limited cohort identified by only five-years of incidence data undoubtedly underestimates the true prevalence of a disease within a population. Thus, we have presented SIRs to assess the association between EoE and CD. These SIRs provide a conservative estimate of the degree of association as they relay on a patient being diagnosed with both conditions within the five-year study period. Our somewhat lower rate of concomitant diagnosis does raise the possibility that coexistent EoE and CD might have been under-diagnosed in our study population, especially given the high incidence of autoimmune gastrointestinal illnesses in general within our population.

Importantly, the disease incidence rates we present in this paper compare favourably to those published from elsewhere in North America. In a recent study from Olmsted County, Ludvigsson and colleagues report the average CD incidence of 17.4 cases per100,000 person-years.[20] This is comparable to the CD incidence we present in this paper, which ranged from 10.4 to 15.7 cases per 100,000 population. A similar paper also from Olmsted County reported the incidence of EoE to be 9.45 cases per 100,000 person years.[3] This is equally comparable to the EoE incidence rates we report that demonstrate a steady increase from 2.1 to 10.7 cases per 100,000 population over the five-year study period.

A limitation in this study relates to the clinicopathologic diagnosis of EoE. The clinical information was not available and diagnosis was based solely on established histopathologic features of EoE. The presence of tissue eosinophilia can be seen in the setting of other gastrointestinal diseases, although usually eosinophil density is greater than 15 per HPF.[40,41] At the least, gastric eosinophilia was eliminated as all three patients diagnosed with EoE and CD had concurrent gastric biopsies that were normal. While we have demonstrated improvement or normalization of duodenal biopsies after initiating a gluten free diet, we cannot comment on the response of esophageal eosinophilia to gluten restriction.

The retrospective design of the current study did limit the availability of certain data which could have added to our analysis, including the total number and location of biopsies obtained from each patient, the symptoms that prompted the endoscopic investigation, and the response to therapy. It is conceivable that a patient undergoing an upper endoscopy for one condition may be more likely to be indecently diagnosed with another condition. It is also possible that pediatric gastroenterologists perform more routine biopsies of normal appearing mucosa than their adult colleagues. Both of these issues could serve to generate detection bias when it comes to the diagnosis of both conditions and emphasizes the need of future prospective studies.

Another limitation of this study is that we only assessed data over a five-year period. We were unable to determine if any of the patients had a previous histological diagnosis of CD or EoE prior to 2004 or from another jurisdiction. This issue could potentially result in an overestimation of the true incidence, however, it would also serve to underestimate the degree of disease association.