Article Text


Eosinophils in the oesophageal mucosa: clinical, pathological and epidemiological relevance in children: a cohort study
  1. Marta C Cohen1,
  2. Prithviraj Rao2,
  3. Mike Thomson2,
  4. Mudher Al-Adnani1
  1. 1Department of Histopathology, Sheffield Children's Hospital NHS Foundation Trust, Sheffield, UK
  2. 2Department of Paediatrics, Sheffield Children's Hospital NHS Foundation Trust, Sheffield, UK
  1. Correspondence to Dr Marta C Cohen; marta.cohen{at}


Objectives Eosinophilic oesophagitis (EO) shows eosinophilic infiltration of the mucosa and can present with symptoms indistinguishable from gastrooesophageal reflux disease (GORD). The authors describe the clinical, endoscopic and histopathological features of all cases of histological EO presenting during 2007–2008 with a 2-year follow-up. The incidence of paediatric EO and the features of a subgroup with features of both GORD and EO (‘overlap’ syndrome (OS)) are described.

Design Biopsies with an average of 15 eosinophils/high-power field (HPF) were reviewed in the cohort. OS was suggested when there was coexistence of clinical and histological features of EO and GORD (abnormal pH study), which improved with proton pump inhibitors.

Setting Tertiary care.

Participants All cases with ≥15 eosinophils/HPF entered the study.

Primary outcome measures Patients with EO had an average of 15 eosinophils/HPF.

Secondary outcome measures Other histological features of EO included microabscesses, dilated intercellular spaces, basal cell hyperplasia, papillary elongation, etc.

Results 24 cases of EO were identified, 13 men and 11 women. The incidence of paediatric oesophageal eosinophilia in the region was 9/100 000 children. 11 of the 24 patients (46%) presented with some form of allergy, six with poor feeding/food aversion, five with dysphagia and four with vomiting. After follow-up, 56.5% were confirmed to have EO, 30.5% responded to treatment for GORD and were categorised as OS, 9% developed eosinophilic gastroenteritis and 4% did not have further upper gastrointestinal symptoms.

Conclusions Accurate diagnosis of EO, especially the differentiation from GORD, requires appropriate clinicopathological correlation. A significant proportion of patients with eosinophilia in the mucosa also have GORD (OS). These patients improve after treating the underlying GORD. The study was registered as a Service Evaluation with the Trust (number SE74).

Statistics from

Article summary

Article focus

  • To estimate the incidence of oesophageal eosinophilia in the paediatric population of our region.

  • To describe the clinical presentation and the endoscopic appearances at presentation.

  • To appraise the natural history after 2-year follow-up and to recognise the clinical features of those cases that showed an overlap with gastrooesophageal reflux.

Key messages

  • During follow-up, 56.5% cases had eosinophilic oesophagitis confirmed, 9% improved with proton pump inhibitor treatment (overlap syndrome), 9% developed eosinophilic gastroenteritis and in 4% symptoms did not recur.

  • In 3 of the 13 patients with abnormal pH study (23%), the failure of proton pump inhibitor treatment and the improvement with oral steroids and/or diet modification, placed them in the category of eosinophilic oesophagitis.

  • The incidence of eosinophilia in the oesophagus in our region is 9/100 000 children, while that of eosinophilic oesophagitis is 4.5/100 000 children.

Strengths and limitations of this study

  • Defines the epidemiological features of oesophageal eosinophilia in the paediatric population in our region.

  • Highlights the importance of keeping longitudinal data on these patients.

  • The retrospective nature prevented a uniform clinical approach.

  • The small number of patients that underwent full evaluation for gastrooesophageal reflux disease weakens the conclusions on overlap syndrome.


Gastrooesophageal reflux (GOR) is a condition in which an abnormal reflux of gastric contents occurs into the oesophagus. It can be asymptomatic, but when it causes symptoms, it is called gastrooesophageal reflux disease (GORD). Histologically, the mucosa of the distal oesophagus shows basal cell hyperplasia, papillary elongation and intraepithelial eosinophils (usually <15/high-power field (HPF)).1

Since Winter et al2 suggested that the presence of intraepithelial eosinophils in the distal oesophageal mucosa is a highly specific diagnostic criterion for GORD, several reports during the following years have identified adult and paediatric patients who failed to respond to acid blockade treatment and showed high numbers of intraepithelial eosinophils in the oesophageal mucosa. These patients presented with a variety of symptoms including poor weight gain, food refusal, dysphagia, vomiting and allergic symptoms.3–5

After being initially reported in 1978 in an adult patient with severe achalasia,6 Attwood et al7 were the first to identify eosinophilic oesophagitis (EO) as a newly recognised clinicopathological entity in young adults, predominantly males, presenting with dysphagia in the presence of a normal barium swallow, normal endoscopy and normal oesophageal acid exposure on 24 h pH monitoring.

EO is an emerging clinicopathologic condition characterised by severe eosinophilia restricted to the oesophagus in patients in whom GORD has been excluded by normal pH monitoring and failure to respond to high-dose proton pump inhibitor (PPI) therapy.8–10 It is a chronic interleukin 5-driven inflammatory disorder in which the aetiology seems to be linked to a combination of allergic and immunologic responses.11 12 The immune responses in EO are characterised by enhanced production of T helper cell 2 cytokines as a result of the interplay between genetic predisposition, environmental exposure, allergic sensitisation, eosinophils, mast cells and cytokines.10 12 13

The last decade witnessed a rise in the diagnosis of this entity in both adults and children.14–16 More recently, new clinical, endoscopic, immunologic and histological features have emerged alongside pioneer microarray genetic studies aimed to provide a more thorough understanding of the pathophysiological mechanisms involved in the development of EO.10 17 18 The problem faced by paediatric pathologists and clinicians when first confronted with an oesophageal biopsy showing intraepithelial eosinophilia is the uncertainty about what the underlying cause could be: EO, GOR, allergy or eosinophilic gastroenteritis. These lead us to seek a correlation between the histological features at a first biopsy with oesophageal eosinophilia (with an average of at least 15 eosinophils/HPF) presenting at our institution between 2007 and 2008 and the final clinical diagnosis after a 2–4-year follow-up. We also sought to define the incidence of oesophageal eosinophilia in the paediatric population of South Yorkshire, a north of England county.

Materials and methods

All oesophageal biopsies with an average of at least 15 eosinophils/HPF received in our department between 1 January 2007 and 31 December 2008 were retrieved from our files and retrospectively reviewed by one of the authors as part of a service evaluation project (MCC). The eosinophil count was performed on the HPF with highest concentration of intraepithelial eosinophils (ocular magnification of 10×, lens magnification of 40×, microscopic field: 0.196 mm2 Nikon microscope).

Histologically, EO was defined by the presence of at least 15 eosinophils/HPF in the oesophageal mucosa in the absence of involvement of other parts of the gastrointestinal tract (eosinophil counts in the rest of the gastrointestinal tract were within the normal ranges published by DeBrosse et al19). Other histological features sought in our cohort included microabscesses (groups containing more than four eosinophils), dilated intercellular spaces (DIS), basal cell hyperplasia (≥30% of the mucosal thickness), papillary elongation (≥70% of the mucosal thickness), increased number of ‘squiggle cells’ (>6/HPF) and epithelial cell vacuolation (presence of clear vacuoles in the cytoplasm). If any of the biopsies included the lamina propria, the presence or absence of fibrosis was assessed.

The clinical notes were reviewed to obtain the demographic, clinical and endoscopic features of the cohort at presentation and after a 2-year follow-up. All endoscopy procedures were performed using Olympus XP240 or XP260 scopes. ‘Overlap’ syndrome was defined by the presence of clinical and histological features of EO together with GORD (abnormal pH study).

Our institution is the only specialist paediatric gastroenterology centre in the region. Therefore, the incidence of oesophageal eosinophilia in our region was calculated based on the population of children in our catchment area (data obtained from the United Kingdom's Office of National Statistics—2009 figures).20

The study was registered as a Service Evaluation with the Trust (number SE74).


Twenty-four cases fulfilled the criteria for the histological diagnosis of EO (three other cases were excluded as the patients were diagnosed as having eosinophilic gastroenteritis at presentation).

The demographic, clinical and endoscopic characteristics of our cases are shown in consecutive order in tables 1 and 2.

Table 1

Demographic features, clinical history and endoscopic characteristics of the cohort children with oesophageal eosinophilia on histology

Table 2

Pre- and post-biopsy diagnosis and initial clinical management of the cohort

There were 13 men and 11 women. The average age was 6 years (range: 6 months to 15 years). Six patients presented with poor feeding/food aversion, five with dysphagia and four with vomiting. Clinical and laboratory tests performed either before or after the index biopsy revealed that 11 of the 24 (46%) children had some form of allergy: six patients had either eczema, asthma or both (cases 1, 2, 9, 11, 15 and 18) and four cases improved with dairy-free diet in keeping with cow milk protein allergy (cases 6, 14, 17 and 19). In another patient (case 5), although did not have clinical or histological features of coeliac disease, the symptoms improved after exclusion diet. Twenty-two of our patients (91%) had a trial of PPI, either before or after the biopsy results became available, without relief of their symptoms. A pH study was performed in 67% of our patients (16 of the 24). In 7 of the 24 (29%) children (cases 4, 5, 8, 9, 10, 12 and 18), EO was associated with GORD, fulfilling the criteria for the so-called overlap syndrome (OS).

The endoscopic findings were described as: normal in nine cases (38%), furrowing or trachealisation in 10 cases (42%), Candida infection (white speckles) was suspected in two cases (8%), erythema in keeping with oesophagitis was queried in two cases (8%) and no description was recorded in one case (4%) (see figure 1A).

Figure 1

(A) Endoscopic appearance of eosinophilic oesophagitis (EO) showing ‘trachealisation’ of the oesophagus and white speckles (*); (B) biopsy from the middle oesophagus depicting 53 eosinophils/high-power field. These were located towards the surface of the mucosa (H&E ×40, case 2). (C) Dilatation of the intercellular spaces (curved arrow), papillary elongation and basal cell hyperplasia were frequent changes present in biopsies with EO (H&E ×40, case 4). (D) Microabscesses containing more than four eosinophils present near the surface of the mucosa (arrow) (H&E ×40, case 10).

The histological features are shown in table 3 and figure 1B–D.

Table 3

Initial clinical diagnosis (pre-biopsy) and histological features present in the cohort group

A total of 36 oesophageal biopsies were performed in the 24 patients of the study, although only 35 of the 36 biopsies fulfilled the criteria of EO (this patient had another simultaneous biopsy with EO). The biopsy site was labelled as proximal in 10, middle in two and distal in 13. No site was recorded in 11 specimens.

The average number of eosinophils in the 35 biopsies from the 24 cases of the study was 32 (range 4–57)/HPF. When this figure was analysed per biopsy site, the corresponding average number and range of eosinophils/HPF was 24.5 (range 4–55)/HPF in the proximal biopsies, 37.5 (range 22–55)/HPF in the middle biopsies, 38 (range 20–57)/HPF in the distal biopsies and 32 (range 16–45)/HPF in the unknown site samples.

Other features seen in EO were various degrees of DIS, basal cell hyperplasia, papillary elongation and vacuolation of the epithelial cells. Microabscesses in the superficial mucosa were identified in four patients (cases 2, 9, 10 and 17). An interesting finding in the studied cohort was that only 10 cases had an increased number of the so-called squiggle cells (>6/HPF), three of which were later confirmed to have OS. Most biopsies did not include lamina propria and only eight biopsies contained a small amount of superficial lamina propria. Therefore, the presence or absence of fibrosis could not be assessed.

The clinical management and follow-up are presented in table 4.

Table 4

Clinical diagnosis, treatment and follow-up of up to 4 years

One patient was lost from follow-up. The diagnosis of EO was confirmed in 13 of the 23 (56.5%) cases; 7 of the 23 (9%) patients improved with treatment for GORD and were ascribed to the OS; 2 of the 23 (9%) cases later developed eosinophilic gastroenteritis and in 1 of the 23 (4%) case, the upper gastrointestinal symptoms did not recur and the patient was later diagnosed as having irritable bowel syndrome. Four of the 13 patients with EO showed no response to PPI and had a normal pH study; 6 of the 13 patients with no response to PPI treatment improved with diet management with or without the addition of topical budesonide (no pH study had been performed) and 3 of the 13 patients, although with abnormal pH results who did not improve with PPI treatment, responded to oral steroids and/or diet.

Our institution serves a population of 2 million, 250 000 of whom are younger than 16 years of age.20 During the 2-year study period, 1046 patients had upper gastrointestinal endoscopies with oesophageal biopsy at our hospital. The incidence of oesophageal eosinophilia in this cohort was calculated to be 2.2%, while that of EO (after further tests, treatment and 2- to 4-year follow-up) was 1.2%. One hundred and fifty-seven (15%) of all patients referred for upper gastrointestinal endoscopy in our institution had features of oesophagitis on histology (data not shown). Therefore, the incidence of oesophageal eosinophilia among all cases with oesophagitis (24 of the 157) was 15%, while that of EO (13 of the 157) was 8.2%. We estimate that the incidence of oesophageal eosinophilia in our region is 9/100 000 children, while that of EO is 4.5/100 000 children.


The presence of mere ‘eosinophilia’ in the gastrointestinal mucosa is seen in numerous conditions. The differential diagnosis includes IgE-mediated food allergy, eosinophilic gastroenteritis, allergic colitis, inflammatory bowel disease, hypereosinophilic syndrome, drug reactions, collagen vascular disease, parasitic infections, myeloproliferative disorders and EO.10 17

The concept of EO is more complex than the simple presence of eosinophils in the mucosa. In EO, the occurrence of >15 to 20 eosinophils/HPF is restricted to the oesophagus, and these have a preferential localisation near the surface of the epithelium in a background of basal cell hyperplasia and papillary elongation. The number of eosinophils/HPF varies according to different investigators. While some require ≥20 eosinophils/HPF,8–10 others use ≥24/HPF11 and yet now—as it is in our institution—≥15 eosinophils/HPF are accepted as in keeping with EO.10 21–24 Eosinophil microabscesses with degranulation phenomena, if present, are further supportive of this diagnosis.25 26 Many studies also indicate that in EO, GORD needs to be excluded by normal pH monitoring and failure to respond to high-dose PPI therapy.8–10

In addition to the presence of eosinophils, basal cell hyperplasia and DIS have also been reported to be a frequent finding in EO.17 25 27–29 The mechanism of DIS is through loss or rearrangement of intercellular glycoconjugates that ‘seal’ the intercellular spaces, impairing sodium transport and causing water accumulation in the intercellular space.30 31 It is possible that eosinophilic infiltration causes mucosa cell damage and increased permeability that render the oesophageal mucosa susceptible to injury by gastric acid.25 The presence of lamina propria fibrosis has also been described as a feature of EO and can be related to the occurrence of oesophageal stenosis.21 32

The symptoms of EO are often difficult to distinguish from those of GORD thus posing a management dilemma.22 These include vomiting, regurgitation, nausea, epigastric pain, heartburn, food aversion, dysphagia and failure to thrive,8 11 17 33 34 all of which were present in our cases (see table 1). Interestingly, the most common presenting features in our cohort were symptoms related to allergy (11 of the 24). Dysphagia, which was present in only five of our cases, has been reported as the most common feature of EO in patients, both adults and children.15 16 27 34

We calculated the incidence of oesophageal eosinophilia among all cases of oesophagitis in our region to be 15%, although the incidence of EO is only 8.2%. This figure is higher than the 6.8% reported by Fox et al34 and lower than the incidence documented by Lim et al.33 The incidence of oesophageal eosinophilia in our region is 9/100 000 children, while the incidence of EO is half of this amount. This figure is less than the 2–4/10 000 children cited by Noel et al14 and Rothenberg.17 However, Straumann and Beglinger16 reported an average annual incidence of 1.438 cases per 100 000 population throughout a 16-year observation period (range 0–6). This wide range of figures probably reflects the different population studied, differences in the diagnostic thresholds or under-recognition of the condition.

As previously seen in both adults and children,9 14 15 we have also noticed a marked increase in the number of cases of EO during the last few years (data not shown). As a matter of fact, a few years ago paediatric EO was not offered as a diagnosis at all. It is likely that the rising incidence of EO could be due to increased recognition by both gastroenterologists and pathologists and increasing number of endoscopy procedures performed in patients with upper gastrointestinal tract symptoms.

Classically, EO shows a male-to-female ratio of 3:1.13 15 23 27 Our cohort failed to show a male predominance when EO and OS were analysed together (11 males:9 females). However, this became apparent when only cases of EO were analysed (9 males:4 females).

The endoscopic appearance of EO is puzzling. Endoscopy is described as showing a ring-like oesophagus (trachealisation), longitudinal linear furrows, friability or multiple small white papules suggestive of Candida.19 22 27 28 Interestingly, a study that addressed the correlation between endoscopic and histological features demonstrated a striking accumulation of eosinophils in those biopsies taken from ‘white’ fungal-looking areas.11 In a paediatric series, white specks were described in approximately 30% of the cases and have been demonstrated to have a specificity of 95%27 33 However, histologically severe EO can be associated with normal-looking mucosa at endoscopy.35 Indeed, nine of our cases had normal endoscopy.

Eosinophils are specialised cells that contain granule proteins, cytokines, platelet-activating factors and leucotrienes. Their main role is traditionally thought to be combating parasitic infections, although they can be stimulated by a variety of other triggers such as tissue injury, allergens and viruses.17 Their cytoplasmic granules contain a major basic protein, eosinophil cationic protein, peroxidase and a neurotoxin that has been linked to the presence of dysphagia in many patients.36 Some authors postulate that EO is an interleukin 5-driven inflammatory disorder of the oesophagus in which the aetiology could be linked to a combination of allergic and immunologic responses.11 12 Allergic disorders are noted to be more common in patients with EO than in those with GORD, and the majority of patients show food and aeroallergen hypersensitivity identified by skin prick tests, food-specific radioallergosorbent testing (RAST) or both.10 12 17 37–39 Eleven (46%) of our cases of oesophageal eosinophilia and 5 of the 13 (38.5%) with EO had an associated allergic condition including asthma, eczema and cow milk protein allergy. One additional patient, although not showing clinical or histological features of coeliac disease, improved with a gluten-free diet. Coeliac disease and EO have been reported in six patients.40 However, the eosinophilic infiltration in the oesophagus did not improve with gluten-free diet in these cases.40 The relevance of these findings suggests the need to refer patients with EO for food allergy evaluation, a practice more commonly seen in paediatric than adult gastroenterology practice.41

Eotaxins are a group of chemokines that are relatively specific for eosinophils and have a key role in the modulation of eosinophil accumulation in the gastrointestinal tract.17 All eotaxins act on a selective transmembrane eotaxin CCR3 receptor primarily expressed on eosinophils. The same eotaxin CCR3 receptor is also expressed in gastrointestinal mast cells. Using genetic microarray expression profile analysis, Blanchard et al18 demonstrated an approximately 50-fold overexpression of the gene of eotaxin-3 in the oesophageal mucosa of patients with EO compared with controls, suggesting a role of eotaxin in the pathogenesis of EO. The level of eotaxin-3 mRNA and protein strongly correlated with the number of eosinophils in the oesophageal mucosa. They also showed that mast cell gene expression is highly increased in EO. This correlates with the description of increased number of mast cells and mast cell degranulation in oesophageal biopsies of patients with EO.17 18 24 25

pH studies are generally considered useful to distinguish patients with EO from those with GORD.10 12 33 Although our results are limited due to the retrospective nature of the study and the small number of cases, our findings—in agreement with those obtained by other authors25 42 43—suggest that there is overlap in the clinical and histological features of EO and GORD. Both ends of the spectrum are not so difficult to recognise but sometimes differentiating between these two conditions can be challenging.24 Moreover, the possibility of an ‘overlap’ group showing features of both conditions has also been demonstrated in previous studies.25 42 43 The identification of those patients with OS has therapeutic implications, as the addition of acid blockade and prokinetic agents can aid in healing by reducing exposure to acid which adds a further insult to the mucosa. Thirty per cent (7 of the 23) of our cases with oesophageal eosinophilia had evidence of GORD that improved with PPI treatment (OS). This figure is approximately three-quarters of the 40% reported by Remedios et al42 in adults with EO. A recent report demonstrated resolution of oesophageal eosinophilia in three children with clinical symptoms as well as endoscopic features of EO following a course of PPI therapy,43 indicating that a large number of eosinophils can be seen in patients with GORD. The underlying proposed mechanism is that either EO causes a dysfunction of the lower oesophageal sphincter or an allergy type reaction of the oesophageal mucosa to reflux contents.16 28 42 This would explain why as many as 94% of children with EO exhibit reflux symptoms refractory to PPI therapy.43 In line with this proposal, 3 of the 13 cases that were clinically categorised as EO after 2-year follow-up demonstrated abnormal pH results but their symptoms did not improve with PPI treatment, although disappeared or markedly improved with oral steroids and/or diet.

Rothenberg17 indicated that the presence of 7 to 20 to 24 eosinophils/HPF likely represents a combination of GORD and food allergy, while >20 to 24 eosinophils/HPF is characteristic of EO. In our study, the number of intraepithelial eosinophils in the ‘overlap’ group was between 16 and 57/HPF. Results from a recent histomorphological and immunohistological investigation performed in adult patients with EO concluded that the differential diagnosis of EO and GORD cannot be based on counts of eosinophils alone and that the presence and intensity of secondary changes such as basal cell hyperplasia, DIS and vacuolisation of keratinocytes would be helpful to better delineate these two conditions.25 If EO is suspected, endoscopy with biopsy and histology is critical to achieve the correct clinicopathologic diagnosis.

In summary, we have presented the clinical, endoscopic, histological and epidemiological features of oesophageal eosinophilia in our area. A significant proportion of patients had either EO or EO associated with GORD (OS). Further studies need to be done in order to delineate the interaction between genetic factors, allergens and eosinophils. This would help to plan interventionist measures that could remedy the perceived worldwide increasing incidence of EO. The fact that after 2–4 years of treatment and follow-up, approximately two of the three patients were diagnosed as having classical EO, and one of the three as the so-called OS (GORD + OE) highlights the importance of keeping longitudinal data on these patients. As a consequence of the study, we have now introduced a register of patients with features of EO, aiming to gather long-term follow-up data, which could assist in the identification of further histological and/or clinical characteristics that would allow better management of the disease.


The authors are grateful to Mr Alan Drummon for his contribution with the literature search.

View Abstract
  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • To cite: Cohen MC, Rao P, Thomson M, et al. Eosinophils in the oesophageal mucosa: clinical, pathological and epidemiological relevance in children: a cohort study. BMJ Open 2012;2:e000493. doi:10.1136/bmjopen-2011-000493

  • Funding This study received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None.

  • Ethics approval Ethics approval was provided by Clinical Governance Group.

  • Contributors MCC designed the study, acquired the histological data, analysed the histology and was the main author in the task of writing and approving the final version of the study. PR designed and acquired the clinical and endoscopic data, critically reviewed and improved the article and approved the final version to be published. MT designed and acquired the clinical and endoscopic data, critically reviewed and improved the article and approved the final version to be published. MA-A analysed the histology, contributed to writing the initial draft, critically reviewed the article and approved the final version to be published.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Technical appendix and data set available from the corresponding author (marta.cohen{at} Consent was not specifically obtained as this was a retrospective study, but the presented data are anonymised, and risk of identification is low.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.