Defining Eosinophilic Colitis in Children: Insights From a Retrospective Case Series

Behjati, Sam*; Zilbauer, Matthias; Heuschkel, Robert; Phillips, Alan; Salvestrini, Camilla; Torrente, Franco; Bates, Alan W*

Journal of Pediatric Gastroenterology & Nutrition:
doi: 10.1097/MPG.0b013e31818de373
Original Articles: Gastroenterology

Objectives: Although it is a well-described syndrome in infants, eosinophilic colitis is a loosely defined and poorly understood diagnosis in older children. The aims of this case series were to characterise colonic eosinophilia in children and to determine whether it represents a distinct clinicopathological condition.

Methods: We retrospectively reviewed symptomatic children older than 12 months with the principal diagnosis of colonic eosinophilia who presented between January 2000 and February 2007 (n = 38) and a further 10 children whose colonic biopsies were reported as histologically normal. The eosinophil density in all available gastrointestinal biopsies (n = 620) of these children was determined using a validated quantitative morphometric method. Patients were subdivided according to mean colonic eosinophil levels into 3 groups (marked, moderate, or minimal colonic eosinophilia). The following patient information was obtained and compared among patient groups: symptoms prompting endoscopy, atopic history, outcome, serum C-reactive protein and total immunoglobulin E (IgE) levels, erythrocyte sedimentation rate, blood eosinophil count, and endoscopic findings.

Results: In all 3 patient groups, there was a colonic gradient of decreasing eosinophil density from caecum to rectum. Upper gastrointestinal tract biopsies did not exhibit eosinophilia. Although a significant association (P = 0.03) between abnormal total IgE levels and moderate or severe colonic eosinophilia was found, there was no significant difference (P > 0.05) in other patient characteristics. Furthermore, follow-up data did not show a consistent relation between eosinophil density and progression of symptoms.

Conclusions: We find no association between “eosinophilic colitis,” defined as a histologically demonstrated marked colonic eosinophilia, and symptoms, history of atopy, inflammatory markers, or clinical outcome.

Author Information

*Department of Histopathology, UK

Centre for Paediatric Gastroenterology, Royal Free Hampstead NHS Trust, Royal Free and University College Medical School Pond Street, London, UK

Received 18 February, 2008

Accepted 9 September, 2008

Address correspondence and reprint requests to Alan W. Bates, MD, Consultant Pathologist, Department of Histopathology, Royal Free Hospital Pond Street, London, NW3 2QG, UK (e-mail:

The authors report no conflicts of interest.

Article Outline

Although eosinophilic colitis is a well-recognised syndrome in infants (1), it is a loosely defined disorder in children older than 12 months of age (2). The condition, first described by Kaijser in 1937 (3), is characterised by the histological hallmark of colonic eosinophilia (1). Allergic hypersensitive reactions have been implicated in its pathogenesis (2,4), and it has been suggested that eosinophils interfere with gastrointestinal motility (5), possibly through their effects on myenteric ganglion cells (6). The uncertainty surrounding the condition mainly stems from the lack of consensus on normal numbers of colonic eosinophils and on diagnostic criteria (2,7). Although eosinophilic colitis is encountered relatively frequently in paediatric gastroenterology practice, its true incidence is unknown. Reports suggesting geographical variation in numbers of colonic eosinophils in children add to the uncertainty (7,8). The relevant literature comprises case reports of eosinophilic colitis (3,9–13), investigations into normal numbers of colonic eosinophils (7,14,15), and case series of eosinophilic ganglionitis and eosinophilic gastroenteropathy (6,16).

Given our poor understanding of eosinophilic colitis, the finding of colonic eosinophilia in children with otherwise unexplained gastrointestinal complaints is particularly difficult to interpret and remains a challenge to manage. Should colonic eosinophilia be regarded as a pathological condition in its own right, that is, as eosinophilic colitis? Here we present the first case series of symptomatic children in whom colonic eosinophilia is the principal diagnosis to determine whether isolated colonic eosinophilia represents a distinct clinico-pathological entity.

Back to Top | Article Outline


The pathology database of this paediatric gastroenterology tertiary referral centre was screened for histopathology reports between January 2000 and February 2007 that commented on increased numbers of eosinophils in colonic biopsies. The reports were reviewed along with all case notes, previous histopathology reports, and biopsies in November 2007 giving a follow-up time of between 8 months and 7 years. Adults (ages 17 years or older) and infants (ages 12 months or younger) were excluded from further analysis, as were children who at the time of, or before, review (November 2007) had been diagnosed with gastrointestinal disease other than “eosinophilic colitis” (eg, inflammatory bowel disease or gastrointestinal parasites). All biopsy series were reviewed but only the first colonic series, related to 38 children with a report of colonic eosinophilia, were included in the initial analysis. A further 10 children were added to the study, randomly chosen from a list of 24 patients whose colonic biopsies had been reported as histologically normal and with no current or subsequent diagnosis of gastrointestinal disease. Upper gastrointestinal tract biopsies obtained on the same occasion as the colonic biopsies, which were available in 46 patients (96%), were also examined.

Back to Top | Article Outline
Histological Examination

The eosinophil density at each biopsy site was determined for each patient using an objective, quantitative morphometric method applicable to everyday pathology practice. An Olympus BH-2 microscope (Shinjuku-ku, Tokyo, Japan) with a high-power field (HPF) diameter of 0.5 mm2 (area of 0.196 mm2; magnification × 400) was used, and 3 HPFs were examined per biopsy. The method of examining 3 HPFs in each biopsy was adopted, because there was no statistically significant difference (P = 0.9854; median difference −0.05 eosinophils; range −5.8 to 1.8 eosinophils) in a comparison of the mean number of eosinophils in 3 HPFs with the mean number of eosinophils in 10 HPFs in 20 randomly selected gastrointestinal biopsies. In selecting HPF, areas of dense lymphocytic infiltration and artefact were avoided, and areas with preserved surface epithelium were included. According to a previously published method (7), eosinophils were counted where a portion of a cell containing eosinophilic granules was seen, irrespective of whether or not the nucleus was visible. The eosinophil density within the lamina propria and within the surface and crypt epithelium was determined, and the presence or absence of degranulation was documented. Other histological changes of note were also recorded, for example, the presence of neutrophils or eosinophilic crypt abscesses.

Back to Top | Article Outline
Review of Patient Characteristics

After collection of all histological data, and before further analysis, the following information was obtained from patients' medical records using a standardised data proforma: symptoms which led to endoscopy; medical history of asthma, eczema, allergic rhinitis (seasonal and perennial), or urticaria; treatment; outcome; and macroscopic findings on endoscopy. Furthermore, for each patient the age-specific results of the following blood tests were reviewed: serum C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), blood eosinophil count, and total serum immunoglobulin E (IgE) levels. CRP, ESR, and blood eosinophil count were only included in the study if performed on the day of or the day preceding endoscopy. Total IgE levels measured within 12 months of endoscopy were included in the study (median interval between total IgE measurement and colonoscopy was 2 months).

Back to Top | Article Outline
Patient Categorisation

To enable statistical analysis of patient characteristics in relation to eosinophil density, the patients were arbitrarily divided according to the mean colonic biopsy eosinophil count into the following patient groups: <10 eosinophils per HPF (group I, minimal tissue eosinophilia); 10 to <20 eosinophils per HPF (group II, moderate tissue eosinophilia); at least 20 eosinophils per HPF (group III, marked tissue eosinophilia; Fig. 1).

Back to Top | Article Outline
Statistical Methods

The Mann-Whitney U test was used for nonparametric independent data and the Wilcoxon matched-pairs signed-ranks for nonparametric paired data. When the number of pairs was less than 20, exact probabilities were calculated. Fisher exact test was applied to paired observations on 2 variables expressed in contingency tables. All tests were 2-sided. A P value of 0.05 or less was considered to be statistically significant. Statistical analysis was performed using a software package (SPSS 15.0 for Windows, Chicago, IL).

Back to Top | Article Outline


The age and sex distribution and eosinophil counts in 48 children are summarised in Table 1. The mean colonic eosinophil density in 38 children with a diagnosis of colonic eosinophilia was 16.4 eosinophils per HPF. All 10 children with colonic biopsies reported as histologically normal had a mean eosinophil density of <10 eosinophils per HPF (mean of 6.7 colonic eosinophils per HPF).

Table 2 lists all of the clinical symptoms documented in 3 or more children. Symptoms reported by only 1 or 2 patients included food avoidance, anaemia, offensive stools, belching, and nausea. No significant differences were found in symptoms among the 3 patient groups (P > 0.05); 40% of patients reported a history of 1 or more of the following: seasonal or perennial allergic rhinitis, eczema, asthma, or urticaria. There was no significant difference in allergic symptomatology among the 3 patient groups. In contrast, we found statistically significant differences in total serum IgE levels as summarised in Table 3.

Serum CRP levels and ESR were normal in all of the patients except for 2 in group III (1 patient with CRP of 13 mg/L and ESR of 43 mm/hour; 1 patient with raised CRP 43 mg/L and normal ESR), 1 patient in group II (CRP of 13 mg/L and normal ESR), and 1 patient in group I (CRP of 5 mg/L and normal ESR). The differences between patient groups were not statistically significant (P > 0.05). To assess whether the colonic eosinophilia occurred in the context of systemic eosinophilia, we reviewed the blood eosinophil count at the time of endoscopy. This was mildly raised in only 1 patient from group III (0.80 × 109 eosinophils/L, upper limit of normal 0.62 × 109 eosinophils/L).

Colonoscopic findings with histological correlation are presented in Table 4. All documented macroscopic abnormalities were mild and included the presence of colonic lymph follicles, erythema, or loss of vascular pattern; no overt ulceration or haemorrhage was seen in any case. Histological analysis of the colonic biopsies revealed that the crypt architecture was preserved in all but 3, in which minor distortion was found. Furthermore, eosinophilic crypt abscesses were seen in 5 biopsies (2 in group III and 3 in group II), and 5 biopsies contained a minimal neutrophilic infiltrate (3 in group II and 2 in group I).

The distribution of eosinophils along the colorectum was characterised in 30 complete colonic biopsy series, that is, cases including biopsies from caecum, ascending, transverse, descending, and sigmoid colon, as well as rectum: 11 from group I, 12 from group II, and 7 from group III. As illustrated in Figure 2, a gradient of decreasing eosinophil density from the caecum to the rectum with relative rectal sparing was present in all 3 patient groups. In all 3 patient groups, we found significantly (P < 0.02) lower eosinophil density in the distal colon (ie, descending colon, sigmoid colon, and rectum) compared with the proximal colon (ie, caecum, ascending colon, and transverse colon), and colonic eosinophil levels were lowest in the rectum (P < 0.05; comparing rectum with all other colonic biopsy sites). Figure 2 further shows that the ileal eosinophil density in patients from group III was significantly higher (mean count of 19 eosinophils per HPF) than in patient groups II (mean count of 12.1/HPF; P = 0.047) and I (mean count of 11.1/HPF; P = 0.048). There was no significant difference in ileal eosinophil density between groups I and II (P = 0.172).

The mean number of eosinophils counted in a total of 46 oesophageal biopsies was 0.5 per HPF (range 0–8 eosinophils/HPF) with no significant differences among patient groups (P > 0.05). In biopsies from gastric antrum (n = 45) and fourth part of duodenum (n = 46), the eosinophil density was 0.9 per HPF (n = 45; range 0–3.7 eosinophils/HPF) and 7.7 eosinophils per HPF (n = 46; range 1–22.3 eosinophils/HPF), respectively. There was no significant difference in gastric and duodenal eosinophil density among children in groups I, II, or III (P > 0.05).

A subgroup of 6 patients from groups II and III underwent more than 1 endoscopy because of persisting symptoms (Table 5) and required various treatments (detailed in Table 5). In this series, we found no correlation between the change in number of colonic eosinophils and clinical symptoms at follow-up endoscopy (Table 5).

Back to Top | Article Outline


In this retrospective case series, we aimed to determine the significance of isolated colonic eosinophilia in children experiencing otherwise unexplained gastrointestinal symptoms. Although colonic eosinophilia is a feature of various pathological conditions (15,17), the findings from our case series question whether this striking histological finding per se represents a distinct clinicopathological entity.

Most reports of normal colonic eosinophil levels in children have been on the basis of either rectal (18,19) or transverse colonic biopsies (20) and may not be representative of pancolonic eosinophil density. Two previous studies of pancolonic eosinophil density suggested that marked colonic eosinophilia may be a normal finding. DeBrosse et al (7) observed a mean colonic eosinophil density of 15 eosinophils per 0.28 mm2 HPF (54 eosinophils per mm2) in histologically normal colonic biopsies of children. Lowichik and Weinberg (14) found a mean colonic eosinophil count of 17 eosinophils per 0.19 mm2 HPF (89 eosinophils per mm2) in postmortem tissue of asymptomatic children who died in road traffic accidents. In our series, the mean colonic eosinophil count in children with the diagnosis of colonic eosinophilia was 16 eosinophils per 0.196 mm2 HPF (82 eosinophils per mm2) and 7 eosinophils per 0.196 mm2 HPF (34 eosinophils per mm2) in biopsy series were reported as histologically normal. The difference between studies may be partly attributable to geographical variation in the colonic eosinophil density of children (8).

Given the lack of a definition of normal colonic eosinophil density in children, we opted to analyse clinical and laboratory findings in relation to eosinophil density to determine whether findings in children with marked colonic eosinophilia (group III) are specific to this patient group. These children presented with nonspecific gastrointestinal symptoms and showed no significant difference in terms of systemic inflammatory markers or blood eosinophil counts at the time of endoscopy from children with lower colonic eosinophil counts. Furthermore, marked colonic eosinophilia was not associated with abnormal colonoscopic or additional abnormal histological findings. Degranulation of eosinophils and intraepithelial eosinophilia was significantly more common in children with marked colonic eosinophilia, but these changes are of unknown significance. This may reflect the increased number of eosinophils or indicate pathologically increased activity.

We found no association between colonic eosinophilia and eosinophilic oesophagitis, gastritis, or duodenitis in those patients who had upper endoscopic series. There was, however, a significantly raised eosinophil density in the ileal biopsies of children with marked colonic eosinophilia, perhaps a consequence of high eosinophil density in the neighbouring caecum, akin to the phenomenon of backwash ileitis described in ulcerative colitis. The previously reported histological finding of a gradient of decreasing eosinophil density from caecum to rectum was found to be preserved in moderate and marked colonic eosinophilia (7,9).

In his original article, Kaijser (3) proposed that eosinophilic colitis was an allergic phenomenon. Our observation that moderate and marked colonic eosinophilia was significantly associated with abnormal total serum IgE levels supports this hypothesis. Yet, we did not find a significant association between a history of atopy and colonic eosinophilia, which may be expected if colonic eosinophilia was driven by an allergic process.

In our study, we retrospectively followed up the subgroup of children who underwent repeat colonoscopies after the initial diagnosis of colonic eosinophilia. Our data document the various treatments used in our centre for the treatment of eosinophilic colitis, which include eosinophil-specific agents (eg, montelukast (21)) and systemic immunosuppression. Furthermore, the follow-up data in this subgroup of children showed no relation between change in eosinophil density and progression of symptoms, although the validity of this finding is limited by the treatment and the small size (n = 6) of the subgroup. Nevertheless, the finding raises the question that whether there is a relation between the degree of colorectal eosinophilia and severity of symptoms, which should be addressed in prospective longitudinal follow-up studies.

In summary, although the validity of our case series is limited by its retrospective design (22), our findings do not demonstrate that colonic eosinophilia necessarily represents a distinct clinicopathological entity currently termed “eosinophilic colitis.” Prospective follow-up and interventional studies are urgently required to clarify whether colonic eosinophilia does indeed cause gastrointestinal complaints in children, and whether the degree of eosinophilic infiltration is a useful marker of disease severity. Until such evidence from prospective studies becomes available, the clinical management of symptomatic children with the sole diagnosis of colorectal eosinophilia cannot be accurately defined. We recommend that management of these children be based on and targeted at symptom control, rather than at modification of colonic eosinophilia. Moreover, it would seem appropriate to consider further assessment of symptomatic children with colonic eosinophilia to identify possible upstream mechanisms that may explain their clinical presentation.

Back to Top | Article Outline


1. Maloney J, Nowak-Wegrzyn A. Educational clinical case series for pediatric allergy and immunology: allergic proctocolitis, food protein-induced enterocolitis syndrome and allergic eosinophilic gastroenteritis with protein-losing gastroenteropathy as manifestations of non-IgE-mediated cow's milk allergy. Pediatr Allergy Immunol 2007; 18:360–367.
2. Rothenberg ME. Eosinophilic gastrointestinal disorders (EGID). J Allergy Clin Immunol 2004; 113:11–28.
3. Kaijser R. Zur Kenntnis der allergischen Affektionen des Verdauungskanals vom Standpunkt des Chirurgen aus. Arch Klin Chir 1937; 188:36–64.
4. Gonsalves N. Food allergies and eosinophilic gastrointestinal illness. Gastroenterol Clin North Am 2007; 36:75–91.
5. Murch S. Allergy and intestinal dysmotility—evidence of genuine causal linkage? Curr Opin Gastroenterol 2006; 22:664–668.
6. Schappi MG, Smith VV, Milla PJ, et al. Eosinophilic myenteric ganglionitis is associated with functional intestinal obstruction. Gut 2003; 52:752–755.
7. DeBrosse CW, Case JW, Putnam PE, et al. Quantity and distribution of eosinophils in the gastrointestinal tract of children. Pediatr Dev Pathol 2006; 9:210–218.
8. Pascal RR, Gramlich TL. Geographic variations of eosinophil concentration in normal colon mucosa. Mod Pathol 1993; 6:51A.
9. Dunstone GH. A case of eosinophilic colitis. Br J Surg 1959; 46:474–476.
10. Moore D, Lichtman S, Lentz J, et al. Eosinophilic gastroenteritis presenting in an adolescent with isolated colonic involvement. Gut 1986; 27:1219–1222.
11. Sullivan S, Troster M. Eosinophilic colitis. Gut 1987; 28:506.
12. Persic M, Stimac T, Stimac D, et al. Eosinophilic colitis: a rare entity. J Pediatr Gastroenterol Nutr 2001; 32:325–326.
13. Lee JH, Park HY, Choe YH, et al. The development of eosinophilic colitis after liver transplantation in children. Pediatr Transplant 2007; 11:518–523.
14. Lowichik A, Weinberg AG. A quantitative evaluation of mucosal eosinophils in the pediatric gastrointestinal tract. Mod Pathol 1996; 9:110–114.
15. Pensabene L, Brundler MA, Bank JM, et al. Evaluation of mucosal eosinophils in the pediatric colon. Dig Dis Sci 2005; 50:221–229.
16. Whitington PF, Whitington GL. Eosinophilic gastroenteropathy in childhood. J Pediatr Gastroenterol Nutr 1988; 7:379–385.
17. Yang GY, West AB. What do eosinophils tell us in biopsies of patients with inflammatory bowel disease? J Clin Gastroenterol 2003; 36:93–98.
18. Winter HS, Antonioli DA, Fukagawa N, et al. Allergy-related proctocolitis in infants: diagnostic usefulness of rectal biopsy. Mod Pathol 1990; 3:5–10.
19. Machida HM, Catto Smith AG, Gall DG, et al. Allergic colitis in infancy: clinical and pathologic aspects. J Pediatr Gastroenterol Nutr 1994; 19:22–26.
20. Choy MY, Walker-Smith JA, Williams CB, et al. Activated eosinophils in chronic inflammatory bowel disease. Lancet 1990; 336:126–127.
21. Vanderhoof JA, Young RJ, Hanner TL, et al. Montelukast: use in pediatric patients with eosinophilic gastrointestinal disease. J Pediatr Gastroenterol Nutr 2003; 36:293–294.
22. Grimes DA, Schulz KF. Descriptive studies: what they can and cannot do. Lancet 2002; 359:145–149.

Allergic colitis; Colonic eosinophilia; Eosinophilic colitis; Eosinophilic gastrointestinal disorders

© 2009 Lippincott Williams & Wilkins, Inc.