ABSTRACT: Food-induced eosinophilic proctocolitis appears in the first 2 months of life with blood-tinged stools. Aside from occasional apparent pain on defecation and a few infants with moderate eczema, systemic features are absent. Indeed, aside from the diaper, the infant is generally described as well. Thus, in contrast to the infant with enterocolitis, features such as emesis, failure to thrive, significant anemia, and dramatic diarrhea are not seen. In contrast to infants with the other food-induced inflammatory diseases of the bowel, most symptomatic infants are exclusively breast fed. Proctoscopic examination reveals focal erythema, erosions, and/or the nodularity of lymphoid nodular hyperplasia of the rectosigmoid. Biopsies of the rectum and lower sigmoid reveal a characteristic infiltration of the mucosa and lamina propria with eosinophils, usually in excess of 6 to 10 per high powered field. Elimination of the offending protein from the diet of the infant, through use of an extensively hydrolyzed casein-based formula or the elimination of the protein from the diet of the mother of the breast-feeding infant, leads to clinical resolution of the bleeding within 72 to 96 hours. By 1 year of age the infants routinely tolerate an unrestricted diet, and the long term prognosis is excellent.
Associate Professor of Pediatrics, Johns Hopkins University School of Medicine, Lutherville, Maryland, U.S.A.
Address correspondence and reprint requests to Dr. Alan M. Lake, Associate Professor of Pediatrics, Johns Hopkins University, School of Medicine, 10807 Falls Road, Suite 200, Lutherville, MD 21093, U.S.A.
During the past 20 years, a group of infants have been described having bloody stools, but manifesting a more benign course than that seen in infants with classic enterocolitis. Over the years, this entity has carried a number of titles from allergic colitis, to benign dietary protein proctitis, to eosinophilic proctitis, to breast-milk–induced proctocolitis. Originally described in breast-fed infants (1), the disease has been reported in subsequent studies in infants receiving cow's milk formula or soy formula and in a few receiving hydrolysate formula (2–10). More than 50% of infants in published reports were exclusively breast fed, a somewhat unusual observation in the context of dietary protein-induced disease.
The infants typically have blood-streaked normal-to-soft stools at 2 to 8 weeks of age. The range of age can vary from 2 days to 3 months. In contrast to infants with enterocolitis, aside from the bloody diaper, the infant appears well. Indeed, most referring physicians initially assume that the infant has an internal anal tear. Perinatal risk factors such as prematurity, maternal health, and timing of introduction of breast or formula appear to play no role. The onset of bleeding is gradual, initially erratic over several days, then progressing to streaks of blood in most stools. Weight gain and growth are normal; the abdominal examination results are benign. Neither abdominal distension nor emesis is prominent. A fecal smear generally contains increased polymorphonuclear cells in the mucus layer. Of interest, Odze et al. (11) reported that fecal smears are routinely negative for eosinophils.
Laboratory studies have not been consistently reviewed. By definition, stool cultures are negative for pathogenic bacteria and the toxin of Clostridium difficile. In the two prospective studies by Odze et al. (11) and Machida et al. (2), mild peripheral eosinophilia was demonstrated, occasionally with mild hypoalbuminemia (2). Mild anemia develops in infants with continued bleeding, but is rarely significant at initial examination (1,8,9). Radiographic investigations are performed primarily to exclude necrotizing enterocolitis and the prestenotic enterocolitis of Hirschsprung disease. The mild degree of bleeding rarely justifies exclusion studies for coagulopathies.
Endoscopic evaluation of the rectosigmoid by fiberoptic instruments has yielded inconsistent descriptions of the character of inflammation (1–6). Focal erythema and friability are most commonly noted. Abnormal areas of mucosal inflammation are usually separated by normal areas of mucosa. Because of the age of the infant and the demonstration of disease in the rectosigmoid, systematic evaluation of the degree of inflammation in the upper colon or small bowel have not been reported. Dupont et al. (10) described linear erosions and ecchymoses, most prominent in the sigmoid, even when the rectum is spared. In the one infant in whom they completed total colonoscopy, no disease was noted above the descending colon. Focal erosions and nodularity, suggestive of lymphoid nodular hyperplasia, are also very common (2,6,10). In patients with more chronic manifestations at diagnosis, inflammatory features of ulceration in the rectum can be seen (1). In the limited number of infants in published reports, there is no correlation of nature of feeding (breast vs. formula) with the gross endoscopic findings.
In contrast to the variability in endoscopic features, the histopathology noted in biopsy specimens from the rectosigmoid is markedly consistent in these infants. There is moderate acute inflammation with striking eosinophilic infiltration of the colonic epithelium, lamina propria, and even muscularis. This infiltration can be very focal, even within a single biopsy sample (11). The number of eosinophils varies from 6 to more than 20 per high-powered field. Degranulated eosinophils are commonly noted and in the specimens with crypt abscess formation, both neutrophils and eosinophils are noted (1,5,11). There is no correlation of the degree of eosinophilia with the level of the biopsy within the rectosigmoid. Focal epithelial erosions are common, whereas features of chronicity such as goblet cell hyperplasia, forked crypts, and crypt abscess formation are unusual (1,5,11). The eosinophils are frequently in association with lymphoid nodules, surrounding and penetrating the outer rim of the lymphoid aggregates (11). Granulomas are not seen, but giant cells have been reported (12).
TREATMENT AND MANAGEMENT
By definition in these infants, the elimination of the offending protein from the diet leads to a clinical resolution of bleeding within 72 to 96 hours (1). This is true in breast-fed (with elimination of the protein from the mother's diet) and formula-fed infants. Extensively hydrolyzed casein-based formulas are usually used, with only a few infants requiring an L-amino acid formula. Although bleeding and fecal leukocytes clear within days, the endoscopic and histologic healing can take several weeks. Rechallenge of the infant with the offending protein in the first 6 months usually provokes recurrence of bleeding within 72 hours. In contrast to infants with enterocolitis, no peripheral blood leukocytosis is seen during the challenge. After 9 months of age, the infants routinely tolerate an unrestricted diet.
As noted earlier, a striking feature of this clinical entity is the frequency of occurrence in exclusively breast-fed infants. Table 1 summarizes the clinical and endoscopic features for 95 exclusively breast-fed infants who have been personally evaluated during the past 20 years. As expected, none of the infants was clinically ill, indeed, none were admitted to the hospital. Limited proctoscopic examination was undertaken in all infants, confirming focal erythema and frequent nodularity with superficial erosions. Nearly all of the infants with apparent painful defecation were noted to have rectal inflammation and a small internal anal fissure, compatable with anal spasm during defecation. Eosinophilia, most prominent in the lamina propria, was confirmed in the biopsy specimens. Table 2 summarizes the response to maternal protein elimination.
Initially, I asked all the mothers with breast-feeding infants to express and discard their milk for 72 hours while feeding the infant an extensively hydrolyzed casein-based formula. Once it became apparent that maternal intake of cow's milk protein was implicated in most symptomatic infants, I advised all mothers to continue nursing while strictly eliminating all cow's milk–based products from their diets for 5 days. If the infant remained symptomatic, sequential elimination diets for commonly implicated proteins were undertaken. In only five infants was more than one protein implicated. In 11 infants, bleeding persisted despite all efforts to manipulate the maternal diet. Of these, seven responded to use of extensively hydrolyzed casein-based formula, and four of the infants (all of whom also had eczema) required introduction of an L-amino acid–based formula. The maternal diets required strict elimination, with recurrence of bleeding often noted in cow's milk–sensitive infants after inadvertent maternal intake of only a pat of butter.
Ten of the 11 infants in whom no specific maternal dietary protein could be identified continued to breast-feed, even with intermittent persistent bleeding. Eleven other mothers could not maintain a strict diet, and thus their infants intermittently bled. These 21 infants were observed for more than 6 months and had persistent symptomatic proctitis. Six became mildly anemic despite routine iron supplements. None progressed to increased symptoms of concern, and serial biopsies therefore were not performed. All were weaned to a normal diet after 1 year with no concerns. All breast-fed infants with dietary protein eosinophilic proctitis tolerated a normal diet after 1 year of age. The only published prospective study of follow-up to the age of 10 years by Hill et al. (13) revealed no evidence for development of chronic inflammatory colitis in 13 infants with eosinophilic colitis. Of the 35 infants with breast milk protein–induced proctitis that we have observed for more than 10 years, none has developed inflammatory bowel disease.
The ability of a dietary protein to be transported antigenically intact in human breast milk was first demonstrated by Shannon in 1921 (14). More recent investigations have focussed on the presence of bovine β-lactoglobulin throughout lactation (15–17). This process of low-dose sensitization has also been implicated in anaphylactic reactions to breast milk proteins (18).
Why the protein induces an inflammatory response apparently limited only to the lower colon is unknown. In the formula-fed infants, this may reflect only a very mild subset of the infants with enterocolitis. Indeed, in enterocolitis, the maximal inflammatory response is usually in the rectum (6,10). In the breast-fed infant, it is tempting to suggest that the maternal dietary protein is complexed in some fashion with breast milk immunoglobulin (Ig)A or another immunologic component that is only cleaved from the protein antigen by some environmental feature of the rectosigmoid, such as microbial IgA proteases. The prominent role of the lymphoid nodules in antigen sampling suggest that this may be the site of initiation of the inflammatory response.
The pathogenic role of the eosinophil in this process has not been defined. The eosinophil is now considered a proinflammatory cell releasing a plethora of preformed and newly generated mediators (6). In early infancy, the eosinophil may be a critical component of the host response, independent of the presence of IgE or immediate hypersensitivity. Recognizing the potential role for breast milk IgA, alone or complexed to antigen, it is of interest to note that the human eosinophil has receptors that bind IgA and degranulate (19). To date, no one has measured tissue levels of eosinophil-derived inflammatory mediators in eosinophilic proctitis, although, in vitro, such mediators have been shown to stimulate chloride secretion from colonic epithelium (20). The histologic observation that the eosinophils are often clustered in proximity to the lymphoid aggregates below an epithelial erosion suggest their role in response to antigen or antigen-antibody complex uptake across the epithelium in the rectosigmoid. It is also conceivable that the eosinophil, in the infant, may play a more generic and less IgE-specific role in the inflammatory response, be it eosinophils in skin as in specimens of erythema toxicum, the esophagus in infantile acid reflux esophagitis, or the rectosigmoid in this condition.
No systematic investigations of the potential role for IgE-mediated hypersensitivity have been completed in infants with eosinophilic proctitis. In general, atopic features are rare, with only a few infants having eczema. Family histories for food allergy are reported in 25%(11). No data on cord blood IgE, radioallergosorbent testing, or skin testing have been reported.
1. Lake AM, Whitington PF, Hamilton SR. Dietary protein-induced colitis in breast fed infants. J Pediatr 1982; 101:906–10.
2. Machida HM, Smith AGC, Gall DG, et al. Allergic colitis in infancy: clinical and pathologic aspects. J Pediatr Gastroenterol Nutr 1994; 19:22–6.
3. Goldman H, Proujansky R. Allergic proctitis and gastroenteritis in children. Am J Surg Pathol 1986; 10:75–86.
4. Katz AJ, Twarog FJ, Zeiger RS, et al. Milk-sensitive and eosinophilic gastroenteropathy: Similar clinical features with contrasting mechanisms and clinical course. J Allergy Clin Immunol 1984; 74:72–8.
5. Winter HS, Antonioli DA, Fukagawa N, et al. Allergy related proctocolitis in infants: Diagnostic usefulness of rectal biopsy. Mod Pathol 1990; 3:5–10.
6. Odze RD, Wershil BK, Leichtner AM, et al. Allergic colitis in infants. J Pediatr 1995; 126:163–70.
7. Pittschieler K. Cow's milk protein induced colitis in breast-fed infant. J Pediatr Gastroenterol Nutr 1990; 10:548–9.
8. Wilson NW, Self TW, Hamburger RW. Severe cow's milk-induced colitis in an exclusively breast-fed neonate. Clin Pediatr 1990; 29:77–80.
9. Perisic VN, Filipovic D, Kokai G. Allergic colitis and rectal bleeding in an exclusively breast-fed neonate. Acta Pediatr Scand 1988; 77:163–4.
10. Dupont C, Badoual J, Lehuyer B, et al. Rectosigmoidoscopic findings during isolated rectal bleeding in the neonate. J Pediatr Gastroenterol Nutr 1987; 6:257–64.
11. Odze RD, Bines J, Leichtner AM, et al. Allergic proctocolitis in infants: A prospective clinicopathologic biopsy study. Hum Pathol 1993; 24:668–74.
12. Raafat F, Castro R, Booth JW. Eosinophilic proctitis with giant cells: A manifestation of cow's milk protein intolerance. J Pediatr Gastroenterol Nutr 1990; 11:128–32.
13. Hill DJ, Ford RP, Shelton MJ. A study of 100 infants with cow's milk allergy. Clin Rev Allergy 1984; 2:125–42.
14. Shannon WR. Demonstration of food proteins in human breast milk by anaphylactic experiments in guinea pig. Am J Dis Child 1921; 22:223–5.
15. Axelsson I, Jakobsson I, Lindbergh T, et al. Bovine B-lactoglobulin in the human milk: A longitudinal study during the whole lactation period. Acta Pediatr Scand 1986; 75:702–7.
16. Makinen–Kiljunen S, Palosuo T. A sensitive enzyme-linked immunosorbent assay for the determination of bovine B-lactoglobulin in infant feeding formulas and in human milk. Allergy 1992; 47:347–52.
17. Lovegrove JA, Osman DL, Morgan JB, et al. Transfer of cow's milk B-lactoglobulin to human sera after a milk load. Gut 1993; 34:203–7.
18. Lifschitz CH, Hawkins HK, Guerra C, et al. Anaphylactic shock due to cow's milk protein hypersensitivity in a breast fed infant. J Pediatr Gastroenterol Nutr 1988; 7:141–4.
19. Abu-Ghazaleh RI, Fujusawa T, Mestecky J, et al. IgA-induced eosinophil degranulation. J Immunol 1989; 142:2393–400.
© 2000 Lippincott Williams & Wilkins, Inc.
20. Resnick MB, Colgan SP, Patapoff TW, et al. Activated eosinophils evoke chloride secretion in model intestinal epithelia primarily via regulated release of 5`-AMP. J Immunol 1993; 151:5716–23.