Collagenous Colitis in Children: Clinicopathologic, Microbiologic, and Immunologic Features

Abstract

INTRODUCTION Collagenous colitis was first described by Lindström in 1976 (1). Since then, more than 500 cases have been reported in the literature with a worldwide distribution. The mean annual incidence ranges from 1.1 per 100,000 inhabitants in Spain (2) to 1.8 per 100,000 in Sweden (3). There is a female-to-male predominance (4–6). Patients with collagenous colitis usually are middle-aged women. Most are in the fifth decade, with a mean age of 53 years. To date, only four children (ages 5–14 years) have been reported with this diagnosis (7–10). The etiology of collagenous colitis is unknown. It has been proposed that the mucosal injury could be attributable to an unknown luminal agent (11). The immune system, both at a local mucosal level and systemically, has been implicated in the genesis of this disorder (12). We describe here two children with collagenous colitis and microbiologic and immunologic peculiarities. We discuss the unexpected finding of the unusual pathogen Aeromonas hydrophila in stool cultures from both patients and its potential role in the pathogenesis of collagenous colitis. We have performed a comprehensive study of the immune system of the patients, not only systemically but also locally, by analyzing the characteristics of their colonic mucosa-associated lymphoid tissue. Finally, we review the literature regarding clinical features and management of collagenous colitis in children. We conclude that collagenous colitis in children shares features with the adult presentation and might be an end stage of different pathogenic processes in which A. hydrophila infection could be implicated. CASE REPORTS Case 1 A 15-year-old girl with congenital autoimmune hypothyroidism developed type I diabetes mellitus at the age of 3 months. At the age of 13 months, watery diarrhea started, and Aeromonas hydrophila was isolated in feces. After the oral administration of amoxicillin, the diarrhea temporarily resolved, reappearing 1 month later. A jejunal biopsy was performed that showed a mild increase of intraepithelial lymphocytes. In the next year, she had frequent episodes of watery diarrhea. Repeated stool examinations were negative for ova and parasites. During one episode of exacerbation of her diarrhea at the age of 2 years, A. hydrophila was again isolated. Treatment with oral amoxicillin resolved this infection. Her subsequent course was marked by frequent episodes, lasting for several weeks, of watery stools that did not contain blood or mucus. When the patient was 10 years old, her erythrocyte sedimentation rate was of 53 mm/h. A mild steatorrhea (8.4 g/d) was observed. Culture of the duodenal aspirate did not demonstrate bacterial overgrowth. The upper intestinal series showed normal morphology of the intestinal loops. Duodenal, gastric, and esophageal biopsy specimens were obtained. The duodenal mucosa showed findings similar to those of the previous biopsy. Patchy changes consistent with lymphocytic gastritis were found in the fundic mucosa. The esophageal biopsy specimen was normal. Administration of loperamide was maintained during the following year, but episodes of watery diarrhea persisted. At the age of 15 years, her weight and height were on the 25th and the 3rd percentiles for age, respectively. Fecal levels of fat were normal. Duodenal and gastric biopsies were done, and there was no villous atrophy, but the gastric mucosa (fundus and antrum) presented typical changes of collagenous gastritis. After this finding, a colonoscopy was performed with normal appearance of the mucosa. Rectal and colonic biopsy specimens showed characteristic changes of collagenous colitis: irregular thickening of the subepithelial collagenous band, with down growing of fibers, trapping of capillary vessels, a typical separation of the epithelium, and degenerative changes in the surface epithelium. In addition, as required for the diagnosis of collagenous colitis, there was a moderate plasmacytic infiltrate in the lamina propria (Fig. 1a).FIG. 1.: Colonic biopsy specimens from (a) Case 1 and (b) Case 2. The superficial epithelium shows degenerative changes and artifactual separation. A thick collagenous band is seen below (black arrows). A chronic pericryptal inflammatory infiltrate is seen (white arrows), composed mainly of plasma cells (hematoxylin and eosin, original magnification ×300).At this age, a systemic immunologic study was performed. Serum IgA anti-endomysial, IgA antitransglutaminase, IgA and IgG anti-gliadin, antinuclear, antimitochondrial, antiparietal gastric cells, antiadrenal and antihypophyseal antibodies, pANCA, cANCA, and rheumatoid factor study results were all negative. The peripheral blood immunophenotypical study result was consistent with a basal activated state, with an increase in CD5+ B cells (85% of B cells), HLA-DR+ T cells (27% of T cells), CD25+ T cells (13% of T cells), and CD95/Fashigh T cells (50% of T cells). Despite the latter finding, the relative lymphopenia (10% of leukocytes, 1200/mm3) and low CD4+ T cell counts (around 300/mm3) that were repeatedly observed were not attributable to an increased apoptotic rate of lymphocytes via Fas, as it was demonstrated by a normal response of her peripheral blood lymphocytes to the proapoptotic moAb CH11 (anti-CD95) (not shown). Although no autoantibodies were detected among a large panel of organ-specific and nonspecific antigens, including anti-endomysium, a weak oligoclonal serum IgM pattern was demonstrated by isoelectrofocusing (not shown),which is another feature of humoral immunologic activation. She was typed to be DRB1*04, DRB1*11, DRB4, DRB3, DQB1*0301, DQB1*0301 by SSP typing. Despite skin anergy to several common antigens, the in vitro mitogen-stimulated lymphoproliferation was normal, and there was no personal history of infections or familial history of autoimmunity or immunodeficiency. All in all, the picture was consistent with an uncharacterized immune abnormality that suggested a susceptibility for autoimmune diseases (hypothyroidism, diabetes). Case 2 A 2 years 3 months old girl with coarctation of the aorta, ventricular septal defect, and mitral stenosis, had had a resection of aortic coarctation and pulmonary artery band at the age of 4 months. She was referred to our unit because of diarrhea lasting for 3 months. She passed four or five watery stools per day that did not contain mucus or blood. Abdominal distention was frequently present. On physical examination, her weight and height were in the 25th and 10th percentiles for age, respectively. A cardiac systolic murmur was heard in the left sternal border, and the results of the remainder of the physical examination were unremarkable. Three complete blood counts, serum electrolytes, glucose, creatinine, urea, transaminases, albumin, and immunoglobulin results were within normal limits. Serum anti-endomysial, antitransglutaminase, antigliadin (IgA and IgG), antinuclear and antimitochondrial antibodies, rheumatoid factor, pANCA and cANCA were negative. The erythrocyte sedimentation rate was 33 mm/h. Stool examination for ova and parasites was negative, and A. hydrophila grew in a stool culture. The results of upper gastrointestinal series were normal. Upper gastrointestinal endoscopy was normal. Biopsy specimens of the duodenum revealed lymphoid nodular hyperplasia without other pathologic features. Colonoscopy of the entire colon did not reveal any mucosal abnormalities. Colonic biopsy specimens showed typical features of collagenous colitis with: plasmacytic infiltrate in the lamina propria, irregular thickening of the subepithelial collagen tissue, with trapping of capillary vessels, separation of the epithelial layer, and degenerative changes in the surface epithelium (Fig. 1b). In addition there were fairly prominent changes of lymphocytic colitis. The diarrhea resolved gradually with antibiotic treatment of the A. hydrophila infection, and a stool culture performed 1 month later was negative for bacterial pathogens. Stools in the next month had a normal appearance until a sudden episode of diarrhea occurred with up to eight watery bowel movements per day, without mucus or blood. A. hydrophila was subsequently isolated in feces, and antibiotic treatment once again was associated with an improvement in the watery diarrhea. However, it reappeared 3 weeks later, with negative stool cultures for bacterial pathogens, and resolved only after 7 weeks of bismuth subcitrate treatment (60 mg, two times in a day). A sigmoidoscopy performed 3 months later was normal, and biopsy specimens showed typical collagenous colitis features, although with a less prominent lymphocytic colitis. For the next 8 months the stools remained normal, but watery diarrhea reappeared, and once again, A. hydrophila was detected in stool cultures. Mucosal immunologic features The colonic mucosal lymphoid compartment of both patients was studied by flow cytometry after colonic biopsy specimens were obtained by endoscopy. Materials and methods for the isolation and phenotyping of colonic immune cells Lamina propria lymphocytes (LPL) and intraepithelial lymphocytes (IEL) were isolated as described (13). Briefly, the biopsy specimens were shaken in a buffer containing a reducing agent (DTT 1 mM) and a cation chelant (EDTA 1 mM, both reagents from Sigma, USA) at room temperature for 1 hour. This process induces the release into the medium of the cellular content of the epithelium (mainly IEL and enterocytes) without altering the lamina propria (14). After the IEL was collected, the biopsy specimens were washed and incubated with 1 mg/mL collagenase-V (Sigma), which liberates the LPL. Colonic biopsy specimens from five patients with hemorrhoids were also obtained, after written consent, as disease controls. The cells were stained with fluorochrome-conjugated monoclonal antibodies (Becton-Dickinson Immunocytometry Systems, BDIS, USA) specific for the different surface markers and analyzed in a FACScan flow-cytometer (BDIS). Results of the mucosal immunophenotype Case 1 Epithelium: The values for total IEL (9% of the total cellularity) and T cell-receptor γδ-bearing (TcR-γδ) IEL (3% of the IEL) were within the ranges observed in disease controls (6%–12% and 1%–6%, respectively). The only abnormality detected was an increase in the proportion of T cell-receptor αβ-bearing (TcR-αβ+) CD4+ CD25+ cells found in the IEL preparation (12% versus <5% in the controls) (Fig. 2).FIG. 2.: Differential features between the two patients' colonic intraepithelial lymphocytes (IEL). Biparametric representations (dot plots) of fluorescence intensity, as analyzed by flow cytometry. On the left column, Patient 1; on the right, Patient 2. The upper dot plots show the granularity (“side scatter” light dispersion or SSC) and CD45 (pan-leukocyte antigen) expression of the whole cell suspension obtained from the epithelium of each biopsy. This allows for the discrimination of IEL and the establishment of an analysis “gate” to further study surface markers on those cells. Patient 2 has an increased proportion of IEL (23%). The second row of dot plots shows the fluorescence of isotype-matched moAbs, which serve as negative controls. The third row shows the expression of CD4 versus the activation marker CD25. Patient 1 has an increased proportion of CD4+ cells (34%), and ⅓ are abnormally positive for CD25. The last row of dot plots shows the expression of CD4 and CD8, which confirms that the increase of IEL in Patient 2 is mainly attributable to an increased CD8+ IEL population.Lamina propria: The total number of LPL, as well as the CD4/CD8 ratio, and the proportion of B cells and plasma cells were all within normal ranges. Case 2 Epithelium: The proportion of IEL was increased (23% of the total cellularity, 6%–12% in disease controls) because of an increase in CD8+ T IEL (80% of the total IEL). The TcR-γδ IEL (8% of the IEL) and the CD4+ CD25+ T cells were within normal ranges (Fig. 2). Lamina propria: Unlike previous reports of increased LPL in collagenous colitis (15), the total number of LPL, as well as the CD4/CD8 ratio and the proportion of B cells and plasma cells were all within normal ranges. DISCUSSION Patients with collagenous colitis are usually middle-aged women, and most are in the fifth decade, with a mean age of 53 years. This condition has been rarely observed in childhood, and only four children younger than 14 years have been reported (7–10). Gremse et al. (16) reported another child with collagenous colitis, but the diagnosis was criticized by Ýardley et al. (17) because it did not fulfill the required histologic criteria. We agree with Ýardley's comments, and we have not included this patient in the current review of children with collagenous colitis. The clinical features of the reported children and the two others described here (Table 1) were similar to those reported for adult patients. Four patients were female and 2 male, and their ages ranged from 2 to 15 years. Watery diarrhea, with abundant and voluminous stools, was the primary symptom, sometimes accompanied by abdominal pain, such as occurred in two children. Dehydration did not occur in the patients reported by us or those reported by others. The absence of mucus and blood in the stools, a common characteristic of collagenous colitis, was observed in all children. Laboratory investigations of chronic diarrhea were normal except in one of our patients (Patient 1), who had steatorrhea. This finding has been previously reported in collagenous colitis (18). The erythrocyte sedimentation rate is frequently mildly elevated in collagenous colitis, such as occurred in our patients and in those reported by Perisic and Kokai (9) and by Busuttil (8).TABLE 1: Characteristics of children with collagenous colitisThe gross appearance of the colonic mucosa at endoscopy in children with collagenous colitis showed no abnormalities in five children, and only subtle endoscopic change (slightly congested mucosa) was observed in one (8). A previous report (19) has demonstrated that rectal biopsy in collagenous colitis can be normal in the presence of typical changes in the colon. The severity of the histologic changes declines from the proximal colon to the distal colon (20), and collagenous colitis may be patchy, especially in the specimens obtained from the rectosigmoid zone. Thus, a total colonoscopy should be done with biopsies of multiple sites to establish the diagnosis of collagenous colitis. In both of our patients, biopsies at different levels of the colon were obtained, and all the specimens had similar histologic findings. The presence of Aeromonas hydrophila in the stools from of our two patients at the onset of diarrhea episodes is an unexpected finding. This unusual bacterium causes gastroenteritis in humans (21), but these infections are not common in children in developed countries (22). In our institution, A. hydrophila was isolated in 4 of 467 stool cultures from children with diarrhea from 1999 to 2000, with one occurrence being from a girl (Patient 2) with collagenous colitis. Whether the association of A. hydrophila and collagenous colitis, previously unreported in children or adults, was casual, pathogenic, or secondary to the increased collagen bands, it deserves further comment. The cause of collagenous colitis is unknown. Järnerot et al. (11) have observed that ileostomy with diversion of the fecal stream induces clinical and histologic remission in collagenous colitis and that restoration of intestinal continuity results in clinical and histologic relapse. These observations and the demonstration of a cytotoxic activity in the feces from patients with collagenous colitis (23) indicate that a luminal agent might be involved in its etiology (11,12,23,24). The beneficial effect of various antibiotics (4,25), bile acid-binding resins (4), and bismuth (26,27) on watery diarrhea, as well as the sudden onset of the disease in many patients, support the possibility that a luminal agent, either bacteria or toxin might be pathogenic (12). In this sense, the A. hydrophila infection in our patients at the onset of watery diarrhea could have triggered the development of the disease. Some special characteristics of A. hydrophila could support this hypothesis. A. hydrophila possesses a collagen-binding protein (28,29) that allows this bacterium to attach to types I and IV of collagen (30). In normal colon, the basement membrane consists of collagen type IV (12). A. hydrophila could invade the colonic mucosa and bind to the normal subepithelial collagen layer. Subsequently, these bacteria could induce inflammatory changes in the mucosa of infected patients, as has been demonstrated (31–34). If this is the pathogenic mechanism, the collagen band observed in collagenous colitis would be a consequence of the mucosal inflammation and not the primary abnormality (24). In other words, A. hydrophila, by triggering chronic mucosal inflammatory changes (a constant feature of collagenous colitis (35,36)), could be a primary agent in the induction of the collagenous colitis. Alternatively, as an opposite mechanism, the presence of A. hydrophila might be simply secondary to the increased collagen band of collagenous colitis (consisting of collagen types I, III, IV and VI) (12). In this case, the bacteria would be merely responsible for an exacerbation of the symptoms, but not for the collagenous colitis itself, which would be driven by other forces, such as autoimmunity. There is no formal proof that infection is playing a role in the pathogenesis of collagenous colitis. Although a previous study by Mäkinen et al. (37) reported three patients with Yersinia enterocolitica infection preceding the diagnosis of collagenous colitis, most patients with collagenous colitis show no pathologic organisms in the stools (38,39). A possible explanation for the lack of an infectious pathogenic agent could be that patients with collagenous colitis have an altered immunologic response to different luminal agents, such as bacteria, toxins, alimentary antigens or others, that may trigger inflammatory responses in the mucosa. The increased association of autoimmune disease, in as many as 40% of these patients (12), would support this second pathogenic mechanism. This association is also observed in affected children. One girl described here (Patient 1) had congenital autoimmune hypothyroidism, type I diabetes mellitus, and collagenous gastritis and fulfilled the criteria for type III autoimmune polyglandular syndrome. Collagenous gastritis is an extremely rare entity, with only 14 cases published previously (40). Simultaneous occurrence of collagenous gastritis and collagenous colitis has been reported in only four of these patients (40–43). Two other children reported previously had juvenile scleroderma (7) and celiac disease (10). The features observed in the immunologic study of the colonic mucosa in our two patients support the possibility that and autoimmunity could both be involved in the pathogenesis of different types of collagenous colitis. A more prominent feature observed in our patients was the paradoxical immunologic disparity between them, despite their clinicopathologic and microbiologic similarity. Patient 1 had features of lymphocyte activation, suggesting autoimmunity (CD4+ CD25+ cells) as observed in autoimmune enteropathy (FL, GR, personal observation), whereas Patient 2 had an increase in CD8+ IEL, suggesting an effector response to a luminal infection. Patient 1 showed an autoimmunity-prone course, fulfilling the criteria for polyglandular autoimmune syndrome type III. However, Patient 2, who did not show any feature of autoimmune processes but only the Aeromonas infection, had a considerable increase of CD8+ IEL as the only abnormality detected in her colonic mucosa. This has been described in rectal biopsy specimens of collagenous colitis (44) and in response to mucosal infection (45). This was also the only alteration detected in the duodenum of an adult patient with collagenous colitis (FL, GR, personal observation). In summary, the contrasting picture of our two patients suggests that collagenous colitis could be the end stage of different pathogenic conditions, such as autoimmunity (triggered or not by an exogenous agent) or infections. Several therapeutic options have been attempted in children with collagenous colitis, including sulfasalazine, cholestyramine, loperamide, prednisone, gluten withdrawal, and bismuth, as shown in Table 1. In all of them, diarrhea persisted unchanged or with only mild improvement. One of our patients (Patient 2) received colloidal bismuth subcitrate for 8 weeks, and watery diarrhea resolved transitorily for 8 months, but reappeared afterward. Collagenous colitis persisted in the biopsies obtained at this time. Bismuth has been recently used as treatment of collagenous colitis with promising results in eight adult patients (26,27). After the conclusion of treatment, watery diarrhea disappeared and biopsies revealed normal colonic mucosa in all patients. The resolution of both symptoms and mucosal changes has been attributed by Fine and Lee (26) to the antidiarrheal, anti-inflammatory, and antibacterial properties of bismuth. However, the good histologic response to bismuth subsalicylate therapy achieved in these adult patients has not been confirmed in our patient treated with bismuth. In conclusion, collagenous colitis is rarely observed in childhood, but when it occurs, it has clinical features similar to those in adult patients. This entity, although rare, should be considered when caring for children with watery diarrhea. The observation of Aeromonas hydrophila infection in our patients supports the possibility that a luminal agent, of microbiologic origin, might be involved in the pathogenesis of (some forms of) the disease. The immunologic study of the colonic mucosa offered a different profile in our patients, suggesting that different etiologies (autoimmunity vs. infection) could underlay their disparate courses. Thus, collagenous colitis might represent an end stage of different pathologic processes, with a more defined autoimmune involvement in some patients and an infectious predominance in others. Bismuth, a promising therapy for collagenous colitis in adults, has not resolved colitis in one of our patients, and additional efforts are needed to achieve an effective treatment of this entity. Acknowledgements: The authors thank Ricardo Vazquez for technical assistance in the photography work.