Original Contribution - September 2000 - Volume 95, Number 9 - Pages 2285-2295

Enterocolitis in Children With Developmental Disorders

A. J. Wakefield, F.R.C.S.,^a,b A. Anthony, M.Sc., Ph.D., M.B.B.S.,^b S. H. Murch, Ph.D., F.R.C.P., F.R.C.P.C.H.,^b M. Thomson, MB.ChB., M.R.C.P., F.R.C.P.C.H.,^c S. M. Montgomery, Ph.D.,^c S. Davies, M.R.C.Path.,^b J. J. O'Leary, M.D., D.Phil., M.R.C.Path.,^b M. Berelowitz, F.R.C.Psych.,^e and J. A. Walker-Smith, M.D., F.R.C.P., F.R.A.C.P., F.R.C.P.C.H.^d

OBJECTIVE: Intestinal pathology, i.e., ileocolonic lymphoid nodular hyperplasia (LNH) and mucosal inflammation, has been described in children with developmental disorders. This study describes some of the endoscopic and pathological characteristics in a group of children with developmental disorders (affected children) that are associated with behavioral regression and bowel symptoms, and compares them with pediatric controls.

METHODS: Ileocolonoscopy and biopsy were performed on 60 affected children (median age 6 yr, range 3-16; 53 male). Developmental diagnoses were autism (50 patients), Asperger's syndrome (five), disintegrative disorder (two), attention deficit hyperactivity disorder (ADHD) (one), schizophrenia (one), and dyslexia (one). Severity of ileal LNH was graded (0-3) in both affected children and 37 developmentally normal controls (median age 11 yr, range 2-13 yr) who were investigated for possible inflammatory bowel disease (IBD). Tissue sections were reviewed by three pathologists and scored on a standard proforma. Data were compared with ileocolonic biopsies from 22 histologically normal children (controls) and 20 children with ulcerative colitis (UC), scored in an identical manner. Gut pathogens were sought routinely.

RESULTS: Ileal LNH was present in 54 of 58 (93%) affected children and in five of 35 (14.3%) controls (p < 0.001). Colonic LNH was present in 18 of 60 (30%) affected children and in two of 37 (5.4%) controls (p < 0.01). Histologically, reactive follicular hyperplasia was present in 46 of 52 (88.5%) ileal biopsies from affected children and in four of 14 (29%) with UC, but not in non-IBD controls (p < 0.01). Active ileitis was present in four of 51 (8%) affected children but not in controls. Chronic colitis was identified in 53 of 60 (88%) affected children compared with one of 22 (4.5%) controls and in 20 of 20 (100%) with UC. Scores of frequency and severity of inflammation were significantly greater in both affected children and those with UC, compared with controls (p < 0.001).

CONCLUSIONS: A new variant of inflammatory bowel disease is present in this group of children with developmental disorders.

Cite this article as: . Wakefield AJ, Anthony A, Murch SH, Thomson M, Montgomery SM, Davies S, O'Leary JJ, Phil D, Berelowitz M and Walker-Smith JA. Enterocolitis in Children With Developmental Disorders. Am J Gastroenterol September;95:2285-2295.

^aUniversity Departments of Medicine, ^bHistopathology, ^cPaediatric Gastroenterology, and ^dPaediatric Psychiatry, Royal Free and ^eUniversity College Medical School, Royal Free Campus, London, United Kingdom, and University Department of Pathology, Coombe Women's Hospital and Trinity College, Dublin, Eire

Introduction

We have recently described a characteristic pattern of intestinal inflammation in a cohort of children with developmental disorders (1). In these children, the majority of whom had autism, a period of initial normal development was followed by developmental regression and loss of acquired skills, sometimes occurring precipitously over a period of days to weeks. Long-standing intestinal symptoms, as described previously (1), were typical of this group of children. These symptoms had often started at around the same time as the behavioral changes.

Ileocolonic lymphoid nodular hyperplasia (LNH) was a consistent feature of this condition, an observation that has been reported subsequently in children with attention deficit hyperactivity disorder (ADHD) and non-IgE-mediated food allergy (2). There is an anecdotal impression that LNH is a common finding in children undergoing ileocolonoscopy, although this has not been subjected to systematic analysis in a controlled study. It cannot be assumed that LNH is a normal finding in children, as aymptomatic children are not subjected to ileocolonoscopy, and LNH may produce symptoms in its own right (3). Chronic intestinal LNH is a feature of either congenital or acquired immunodeficient states (4, 5, 6, 7, 8, 9, 10) and has been described in congenital B cell abnormalities (5, 6), and common variable immunodeficiency (7, 8). In its persistent acquired form, ileal LNH has been reported in association with infection with human immunodeficiency virus (HIV) before the development of AIDS (10).

The other consistent feature of the intestinal lesion was a mild-to-moderate colitis that lacked the specific diagnostic features of either Crohn's disease or ulcerative colitis (1). This combination of features, i.e., LNH and nonspecific colitis, indicates the possibility of chronic mucosal and/or systemic immune dysregulation. Systemic immunological abnormalities are not infrequent in children with autistic spectrum disorders (11, 12, 13, 14), although the origin and significance of the findings are uncertain. These immune changes, plus the presence of myelin basic protein (MBP) antibodies (15) and inhibition of macrophage migration to MBP (16), have led some workers to suggest that the behavioural syndrome may be associated with cerebral damage due to an autoimmune response to myelin or other structural components of the CNS (15, 16). As part of our initial study (1) we undertook cerebral magnetic resonance imaging, EEG, and biochemical analysis of cerebrospinal fluid; none of these investigations indicated cerebral inflammation that would be consistent with autoimmune demyelination, although a more subtle lesion remains a possibility. Alternatively, others have proposed that some forms of autism may arise from the toxic effects of intestinal products on the developing brain (17, 18, 19), a situation that may have some overlap with hepatic enephalopathy. An early childhood enterocolitis would be more consistent with the latter mechanism.

This study sought to describe some of the characteristic endoscopic and histopathological features of this syndrome in a larger cohort of children with developmental disorders and intestinal symptoms, and to compare the findings with those in developmentally normal children undergoing ileocolonoscopy.

Materials and Methods

This study involved the analysis of data from 60 consecutive children with developmental disorders (affected children) including those 12 children described in a preliminary report (1). The median age of the children was 6 yr (range 3-16 yr) and 53 were boys, consistent with the male bias of developmental disorders (20). Developmental diagnoses in the affected children were autism (50 patients), Asperger's syndrome (five), disintegrative disorder (two), attention deficit hyperactivity disorder (ADHD) (one), and schizophrenia (one). The latter child was the oldest in the cohort, at 14 yr of age. The remaining child in this cohort, a girl 8 yr of age, had dyslexia and learning difficulties: she underwent developmental regression from approximately 54 months of age, which was associated with mouth ulcers, conjunctivitis, and severe constipation. Her developmental status is currently under investigation.

Autism is a behavioral syndrome that consists of qualitative impairments in social interaction and communication, with restrictive, repetitive, and stereotypic patterns of behavior. Delays or abnormal functioning in at least one of these areas occurs before the age of 3 yr. Asperger's syndrome is a high-functioning autistic spectrum disorder, and disintegrative disorder is a regressive condition occurring at age >3 yr in a previously normal child. Loss of acquired skills may occur precipitously or over a period of months. The behavioral features are similar to those of autism but may be accompanied by loss of bowel and bladder control (20).

The majority of affected children were white (57), but two were of Middle-Eastern origin, and one child had an Indian father and a white mother.

All but three affected children had a developmental disorder that was associated with a clear history of regression, with loss of acquired skills after 1 year of documented normal development (general practitioner/health visitor records). In three cases, affected children who had been developmentally normal failed to progress beyond a certain point, but did not regress. All but one of the affected children had current intestinal symptoms consisting of abdominal pain, constipation, diarrhea (or alternating constipation and diarrhea), and bloating. The one affected child who did not have current intestinal symptoms was investigated at his parents' and general practitioner's request. Affected children were consistently fastidious in their eating habits, with a diet limited largely to cereals, potato crisps, and bread. Despite this, they typically seemed well nourished, with anthropomorphic indices within normal limits. Certain foodstuffs such as dairy products were reported by parents to produce deterioration in behavior, whereas withholding such foods apparently produced behavioral improvement—in particular, for aggression, eye contact, and sleep pattern. According to parental reports, recognizably undigested food was often seen in stools.

Investigations

Medical and developmental histories were taken, and a routine physical examination was conducted. A fasting, morning blood sample was taken for routine hematology and biochemistry. Sera were screened for antigliadin and antiendomyseal antibodies. Common enteric pathogens were sought by routine serology, microscopy, and culture. A review of the children was undertaken by an experienced child psychiatrist to confirm the developmental diagnosis using DSM-IV criteria (20).

Ileocolonoscopy and Histology

All children underwent routine ileocolonoscopy and mucosal biopsy. Ileal LNH was classified subjectively according to prominence and extent as mild (grade 1), moderate (grade 2), or severe (grade 3), or as grade 0 if the ileum showed no LNH (Fig. 1), according to the report provided by one of three physicians. Similarly, ileal appearances were graded in 37 developmentally normal children (median age 11 yr, range 2-13 yr). All of these children had been investigated for symptoms of possible inflammatory bowel disease, although the final diagnosis was neither Crohn's disease nor ulcerative colitis. Diagnoses in these children included idiopathic constipation (five patients), polyps (two), and LNH (five). In the majority (thirty), including those with constipation, the colonoscopy was reported as normal, and follow-up has not revealed any other abnormality.

Figure 1 Lymphoid nodular hyperplasia (LNH) of the terminal ileum in affected children showing representative grades of LNH: (A) none (score 0); (B) mild (score 1); (C) moderate (score 2); and (D) severe (score 3).

Mucosal biopsies were taken from the ileum, cecum/ascending colon, transverse colon, descending/sigmoid colon, and rectum. Hematoxylin and eosin-stained histological sections from all biopsies were reviewed in the routine pathology laboratory, followed by independent review and scoring on a standard proforma (Table 1). In those cases where there was disagreement between these two reports, sections were examined and reported by a third senior pathologist, whose arbitration provided the final score. In an identical manner, histological sections from the ileum and colon of children without developmental disorder were scored (median age 11.5 years; range 2-13). These included 22 consecutive ileocolonoscopic biopsy series that had been reported as normal after routine histopathology assessment. All children in this non-IBD control group had undergone ileocolonoscopy for investigation of intestinal symptoms and are included in the 37 endoscopic controls, as described above. To validate further the evaluation and scoring, 10 coded ileocolonic biopsy series (five affected children and five non-IBD controls) were reviewed at another institution by a senior pathologist in an observer-blinded fashion. Data from these independent assessments were compared.

Table 1. Histopathology Proforma Used for Scoring of All Biopsies From Affected Children (Autistic Enterocolitis), Non-IBD Controls, and Those With Ulcerative Colitis

In order to compare the site and degree of pathological changes in biopsies from affected children with those of a well characterized IBD, ileocolonic biopsy series from 20 children with established ulcerative colitis (median age 14 yr, range 8-15 yr) were examined and scored in an identical manner.

Selection Criteria

To avoid selection bias, children in all groups were chosen consecutively on the basis that they fulfilled the primary criteria; that is, developmental disorder with bowel symptoms, developmentally normal with bowel symptoms, developmentally normal with normal histology, or ulcerative colitis.

Ethical Approval

All clinical investigations were undertaken with fully informed, written consent from the parents. The initial phase of these studies (1) was approved by the Ethical Practices Committee of the Royal Free Hampstead NHS Trust. Thereafter, children were investigated according to clinical need after a formal referral from each child's General Practitioner or Consultant.

Statistical Analysis

The analyses were performed using SPSS 7.5 for Windows and EpiInfo 6.04b. The ² test was used to investigate differences in endoscopic and histopathological features between the affected children and the control groups. Where any cell in the analysis contained five or fewer subjects, two-tailed Fisher's exact test was used to assess significance, to adjust for the potential effect of small numbers. The Spearman rank correlation was used to investigate the relationship between grade of LNH (0-3) and absolute lymphocyte count among the affected children.

Results

Ileocolonoscopy

ILEUM. Complete ileoscopy, in which the terminal ileum was visualized and biopsied, was successful in 58 of 60 (97%) affected children and in 35 of 37 (95%) non-IBD controls. The frequency and grade of ileal LNH in these two patient groups is shown in Figure 2. Representative grades of ileal LNH are shown in Figure 1. The data demonstrate not only the consistent presence of LNH in the ileum of affected children 54 of 58 (93%), but also its relative infrequency in developmentally normal children undergoing investigation for similar intestinal symptoms (five of 35; 14.3%); this difference is statistically significant (p < 0.001). In affected children, 76% scored either moderate or severe; in non-IBD controls, all five children with ileal LNH scored either moderate (n = 4) or severe (n = 1). Of these five controls, symptoms that were the indication for ileocolonoscopy included chronic abdominal pain in four and change in bowel habit in one; endoscopically, there was no other demonstrable pathology to account for these symptoms.

Figure 2 Percentage of children showing ileal lymphoid nodular hyperplasia (LNH), comparing affected children with developmentally normal children , investigated for symptoms of inflammatory bowel disease, in whom the final diagnosis was neither Crohn's disease nor ulcerative colitis. LNH was significantly more common in affected children than in controls (p < 0.001).

COLON. Complete colonscopy was successful in all 60 (100%) of the affected children and all 37 (100%) non-IBD controls. Colonscopies were scored for either the presence or absence of the following: LNH (not graded), red halo sign surrounding follicles (21), loss of vascular pattern, mucosal granularity, mucosal erythema, and presence of ulcer(s). These features, which are recognized correlates of associated histopathological abnormality, were noted previously to be characteristic of this patient group (1). These features were recorded both for cases and for non-IBD controls; the data are shown in Figure 3. Colonic LNH was present in 18 of 60 (30%) affected children compared with only two of 37 (5.4%) non-IBD controls (p < 0.01). Similarly, of the other six endoscopic features described above, all but ulceration are significantly more common in affected children compared with controls (p < 0.05). Although the mean age of the developmentally normal comparison group was significantly greater than that of the affected children, the presence or absence of LNH was not related to age in either group.
 

Figure 3 Colonoscopic features in affected children   and non-inflammatory bowel disease controls   . Colonoscopies were scored for either the presence or absence of lymphoid nodular hyperplasia, red halo sign, loss of vascular pattern (LVP), mucosal granularity, mucosal erythema, and ulceration. All of the features except ulceration were statistically significantly more common in affected children than in controls (p < 0.01).

Histological Findings

Ileal and colonic biopsies from 60 affected children, 22 non-IBD control children, and 20 children with ulcerative colitis were examined initially by a clinical histopathologist, and were subsequently evaluated and scored by another pathologist using a standard proforma (Table 1). In the 60 affected children, there was discordance between the two reports in five cases: these were resolved by a third pathologist, whose independent review agreed with the first pathologist in two cases and the second pathologist in three cases. There was disagreement between pathologists concerning the interpretation of biopsies from only one of the non-IBD controls (which was resolved as showing mild inflammation in a cecal biopsy) and in none of the children with ulcerative colitis.

Ten ileocolonic biopsy series were reviewed and scored in an observer-blinded fashion at an independent institution. No indication was given of how many samples came from each patient group. Cases were clearly distinguished from controls by the blinded reviewer. Out of a possible total of 15 points, independent scores were identical for the same criterion in four of 10 cases (40%), within one point of each other in five of 10 cases (50%), and within two points of each other in one of 10 cases (10%) (Spearman rank correlation 0.79; p < 0.006). No reviewer scored systematically higher or lower than the other.

ILEAL HISTOLOGY. A total of 86 ileal biopsies were assessed and scored. This total comprised 52 biopsies from affected children (seven biopsies, consisting of fragments of villi only, were considered inadequate for evaluation), 20 from non-IBD controls, and 14 from children with ulcerative colitis. Reactive follicular hyperplasia (RFH) was identified in 47 of 51 (92%) biopsies from affected children and in four of 14 (29%) with ulcerative colitis. It was not present in any of the 20 biopsies from non-IBD controls. The differences between biopsies from affected children and from both non-IBD controls and those with ulcerative colitis are significant (p < 0.001 and p < 0.01, respectively). Qualitatively, ileal lymph nodes in those children with LNH showed marked expansion of lymphoid tissue in histological section, as described previously (1). Follicle numbers per biopsy were increased from 2-3 follicles per biopsy in normal ileal biopsies to 4-5 per biopsy in those with LNH: follicles were confluent with loss of follicle-to-follicle demarcation. In comparison with normal follicles, the germinal centers were grossly enlarged and reactive, as indicated by numerous tingible body macrophages. The outer margins of the T cell zone were not well defined as they were in normal ileal follicles, with the lymphoid compartment extending into, and apparently expanding, adjacent villi. There was also disruption, but not destruction, of adjacent crypts. Expansion of lymphoid tissue around crypts gave the impression of a decrease in crypt numbers, a histological appearance similar to that described by Fiber and Schaefer (3). In addition, neutrophils and lymphocytes were often seen infiltrating the epithelium overlying follicles: neutrophils were also seen infiltrating the crypt epithelium (acute cryptitis) in some cases. Overall, active ileitis (neutrophilic infiltration) was seen in four of 51 (8%) ileal biopsies from affected children. Aphthoid ulceration was seen in two of 51 (4%) biopsies. In the 20 non-IBD control ileal biopsies, none showed active inflammation. Only two (4%) biopsies from affected children scored positively for the presence of an increase in intraepithelial lymphocytes, and two (4%) for the presence of eosinophil infiltration of the lamina propria. These features were not present in either non-IBD or ulcerative colitis controls. These differences were not statistically significant because of small numbers.

COLONIC HISTOLOGY. A total of 380 colonic biopsies were examined and scored; these included 229 biopsies from affected children, 80 from non-IBD controls, and 71 from children with ulcerative colitis. The numbers and percentage of biopsies in each group showing pathological changes are shown in Table 2 and Figure 4, respectively. Only one cecal biopsy in the non-IBD controls showed evidence of inflammation, which was scored as mild. In contrast, a high percentage of biopsies from throughout the colon showed pathological changes in both affected children and those with ulcerative colitis. The differences between both affected children and those with ulcerative colitis and non-IBD controls, for the proportion of biopsies exhibiting pathological change are, for each site, significant (p < 0.001) (Table 2). Inflammatory changes in biopsy series from individual affected children, although distributed throughout the colon, were patchy. Some of the histological characteristics of this colitis are shown in Figure 5. Overall, chronic inflammation was identified in colonic biopsies from 53 of 60 (88%) affected children, compared with one of 22 (4.5%) non-IBD controls and 20 of 20 (100%) children with ulcerative colitis. The differences between both affected children and children with ulcerative colitis versus the non-IBD controls are statistically significant (p < 0.001).
There is no statistically significant difference between affected children and those with ulcerative colitis for the proportion of biopsies showing chronic inflammation (p > 0.1). An excess of intraepithelial lymphocytes scored positively in eight of 60 (13%) of affected children, 0 of 22 (0%) non-IBD controls, and 0 of 20 (0%) with ulcerative colitis. There are no statistically significant differences between the respective groups (p > 0.1). Eosinophil infiltration of the lamina propria was observed in 24 of 60 (40%) affected children, 0 of 20 (0%) non-IBD controls, and four of 20 (20%) with ulcerative colitis. The difference between affected children and non-IBD controls is statistically significant (p < 0.001). The differences between affected children and those with ulcerative colitis, versus those with ulcerative colitis and non-IBD controls, are not statistically significant (p > 0.1).
Subepithelial apoptosis/nuclear debris was present in 30 of 60 (50%) affected children, in eight of 22 (36%) non-IBD controls, and in 16 of 20 (80%) with ulcerative colitis. These differences are not statistically significant (p > 0.2). Although not scored prospectively, it was the clear impression of all reviewing pathologists that, whereas the presence of subepithelial apoptosis/nuclear debris was a feature of normal biopsies, its extent was much greater in inflamed biopsies (that is, those from either affected children or children with ulcerative colitis).

Figure 4 Distribution of histopathological changes in the ileum and colon (C/A = cecum/ascending, T = transverse, D/S = descending/sigmoid, and R = rectum) of affected children compared with controls whose biopsies were reported as normal (non-IBD controls) , and children with ulcerative colitis . The differences between both affected children and those with ulcerative colitis, and non-IBD controls, for the proportion of biopsies showing histological change at each site, are statistically significant (p < 0.001).

Figure 5 (A) Normal colonic mucosa from a non-inflammatory bowel disease control child. The surface epithelium is uniform, and the lamina propria shows loosely organized connective tissue and normal stratification of cellular density, which characteristically increases toward the epithelial surface. The demarcation between lamina propria and epithelial basement membrane is distinct (magnification ×40). (B) Colonic mucosa from an autistic child. There is mild disruption and lymphocytic infiltration of the surface epithelium. The crypt epithelium is infiltrated by lymphocytes and neutrophils. The superficial subepithelial basement membrane is indistinct compared with (A). The upper two-thirds of the lamina propria contains an excess of lymphocytes, plasma cells, and macrophages, with loss of stratification. The matrix of the lamina propria appears hyaline in nature (magnification ×40). (C) Crypt abcess formation in a child with autistic enterocolitis (magnification ×100). (D) Crypt distortion in a child with autistic enterocolitis; bifid crypts are seen to the left and right of the micrograph (magnification ×100). Micrographs (B-D) come from different affected children.

Table 2. Pathological Changes in Biopsies From Affected Children, Children With Ulcerative Colitis, and Non-IBD Controls