Lower gastrointestinal bleeding (LGB), particularly in newborns, is a serious condition that raises parental anxiety and usually leads to rapid medical attention. Newborns in a clinically unstable condition are more likely to have diseases such as necrotizing enterocolitis, malrotation with midgut volvulus, Hirschsprung disease enterocolitis, intussusception, or Meckel diverticulum. An infant presenting with LGB who otherwise appears healthy should be examined for swallowed blood, hematochezia, allergic proctocolitis, anal fissures, or lymphonodular hyperplasia. Among these, allergic proctocolitis caused by food protein is a likely cause of LGB in an infant with eosinophilia (1–5).
Allergic proctocolitis is often seen in infants between 1 day and 3 months of age and presents as spots or streaks of blood and mucus in the stool and occasional mild diarrhea. It has been reported that bovine milk and/or egg protein can be transferred in adequate quantities in the milk of breast-feeding mothers to induce colitis (6), and bleeding occasionally abates with maternal diet modification accompanied with the elimination of milk and egg products. These signs and symptoms are considered to result from hypersensitive reactions to cow's milk and/or egg whites. However, there have been several cases of disease with signs and symptoms similar to those of allergic proctocolitis without an apparent influence by an allergic reaction.
In allergic proctocolitis tissue eosinophilic infiltration can be seen in the mucosa, muscularis, or serosa and may cause gastrointestinal bleeding. This condition clinically overlaps eosinophilic gastroenteritis, characterized by diffuse eosinophilic infiltration of the gastrointestinal tract and peripheral eosinophilia. It involves the stomach, duodenum, small bowel, and, rarely, the esophagus or colon (7). Given that allergic reactions to multiple foods are often seen in patients with eosinophilic gastroenteritis, the disease usually runs a chronic, debilitating course with sporadic severe exacerbations, usually requiring systemic corticosteroid administration. Whereas eosinophilic gastroenteritis can affect patients of any age, few cases in the literature have involved neonates (7).
In the present study we analyzed the cases of 2 neonates with gross eosinophilia who experienced LGB soon after birth. Inasmuch as pathologically “allergic proctocolitis” similar to LGB developed in these infants before their first feeding, we suggest a possible pathogenesis involving infiltrated eosinophils in the colonic mucosa in the absence of an allergic reaction. Given that the LGB in these infants was not caused by allergic reactions and their clinical signs and symptoms were atypical of eosinophilic gastroenteritis, we termed the disorder neonatal transient eosinophilic colitis, which is a new clinical entity.
PATIENTS AND METHODS
We examined 2 neonates who experienced fresh LGB soon after birth. There were no serious complications in their delivery. The amniotic fluids were clear. The family histories were negative for atopy and bleeding diathesis. The infants started to pass gross fresh bloody stool soon after birth, before their first feedings. They were in a clinically stable condition, and their abdomens were soft but slightly distended with gas. The skin color was not pale, and there were no petechiae. Abdominal radiographs showed nonspecific bowel gas patterns with no signs of obstruction, pneumatosis intestinalis, or necrotizing enterocolitis. The results of blood chemistry, liver function, C-reactive protein, serum immunoglobulin E levels, and coagulation screening tests were normal (Table 1). Their serum immunoglobulin M levels were not elevated; there were no signs of virus infections, including cytomegalovirus and Epstein-Barr virus; and their stool cultures were negative for abnormal microorganisms. Because the presumed diagnosis was hematochezia, their feedings were stopped, and vitamin K was injected intravenously in both cases; yet, the LGB persisted. Therefore, we decided to perform rectosigmoidoscopy in these infants. Informed consent for rectosigmoidoscopy was obtained from the parents of both infants.
A 2-day-old female neonate was referred to our hospital because of fresh blood in her stool. She had been born at 34 weeks and 4 days of gestation, weighing 2140 g. She had been delivered by cesarean section because of mild preterm asphyxia. Her Apgar scores were reported to be 6 and 9 points at 1 and 5 minutes, respectively. A few minutes after birth she had grossly bloody stool, which happened before feeding. Laboratory data revealed a total white blood cell count of 59,300/μm3 with 9014/μm3 eosinophils. To identify the cause of LGB in this infant, a rectosigmoidoscopy was performed on day 7 and revealed nodular lymphoid hyperplasia with a pale mucosal surface and massive oozing around the rectosigmoid colon (Fig. 1A, B). After sufficient hydration her general condition appeared to improve, and on day 8 the bloody stool disappeared. Repeat rectosigmoidoscopy on day 9 revealed no inflammation on the mucosal surface (Fig. 1C). However, biopsies revealed diffuse, massive infiltration of eosinophils in the lamina propria (Fig. 1D). On day 13 the total white blood cell count dropped to 24,400/μm3 with 3050/μm3 eosinophils. By 4 months of age she had grown well on a diet of breast milk without maternal diet modification and showed no signs of LGB.
A 7-day-old male neonate was referred to our hospital because of fresh blood in his stool. He had been born at 37 weeks and 1 day of gestation, weighing 2564 g. His Apgar scores were reported to be 8 and 9 points at 1 and 5 minutes, respectively. Thirty minutes after birth he had grossly fresh bloody stool, which happened before feeding. Laboratory data revealed a total white blood cell count of 17,000/μm3 with 1955/μm3 eosinophils. On day 8 a rectosigmoidoscopy was performed and revealed nodular lymphoid hyperplasia with frank bloody, friable, edematous colonic mucosa around the rectosigmoid colon (Fig. 2A). Biopsies demonstrated diffuse massive infiltration of eosinophils in the lamina propria (Fig. 2B). To confirm the involvement of eosinophils in the pathogenesis of his colitis, the serum ECP and PAF levels were examined and found to be elevated to 123 μg/L (normal, <14.7) and 13.1 μmol · L−1 · min−1 (normal, <6), respectively.
After sufficient hydration his general condition appeared to improve, and on day 8 the bloody stool disappeared. On day 12 the total white blood cell count dropped to 11,900/μm3 with 417/μm3 eosinophils. On day 9 an elemental diet was reintroduced to control any influence of allergic reactions on his colitis. On day 15 a hydrolyzed hypoallergenic formula was introduced, followed by breast milk, but there was no change in his physical condition. On day 22 rectosigmoidoscopy was performed again and revealed no inflammation on the mucosal surface (Fig. 2C). Histopathological analysis revealed diffuse but reduced infiltration of eosinophils in the lamina propria (Fig. 2D). The serum ECP and PAF levels were both decreased to 41.4 μg/L and 9.6 μg · L−1 · min−1, respectively. This infant was discharged on day 25, weighing 2.7 kg. By 3 months of age he had grown well on a diet of breast milk without maternal diet modification and showed no signs of LGB.
Eosinophilia is relatively common in newborns and is caused by an increased rate of hemocytopoiesis, especially in preterm infants, although most cases are symptomless. However, a closer examination of infants with bloody stool can often reveal abnormally elevated eosinophil counts. The present study describes 2 cases of neonates who showed LGB with massive eosinophilia on the first day of life before enteral feeding. We performed rectosigmoidoscopy to confirm the cause of LGB and found nodular lymphoid hyperplasia with a pale mucosal surface and massive bleeding around the rectosigmoid colon with diffuse eosinophil infiltration in the lamina propria, which is similar to the signs of allergic proctocolitis.
Eosinophils are highly sensitive unstable cells that readily release chemical mediators, such as ECP, PAF, eosinophilic peroxidase, major basic protein, and leukotrienes, even in the absence of an allergic stimulation. These chemical mediators are cytotoxic and can cause mucosal damage in the intestine. Thus, eosinophils are considered to be deeply involved in the pathogenesis of ischemic colitis and necrotizing enterocolitis. It is conceivable that neonatal stress during delivery, transient asphyxia, and other complications can stimulate eosinophils to release these chemical mediators, which introduced inflammation leading to LGB in these infants. To confirm the involvement of eosinophils in the pathogenesis of LGB, we measured the serum ECP and PAF levels, which are both major products of eosinophils, in patient 2 and found both to be elevated. Given that intravenous injection of PAF can cause necrotizing enterocolitis similar to the colitis seen in neonatal piglets (8), we suggest that the chemical mediator that is mainly released from eosinophils may be involved in producing the sort of inflammation and bleeding from the colonic mucosa observed in the present 2 cases.
Unlike patients with ischemic colitis and necrotizing enterocolitis, however, these infants lacked a severely distended abdomen and other serious symptoms. In fact, a few days after intravenous hydration therapy with the elimination of oral diet, they recovered from the LGB without any further medical treatment. Furthermore, another rectosigmoidoscopy after the resolution of the LGB showed persistent but reduced eosinophil infiltration according to the histology. The serum ECP and PAF levels were decreased to just above the normal range, suggesting that these infiltrated eosinophils in the mucosa were no longer activated.
Given the eosinophilia observed in these infants, the pathogenesis of eosinophilia remains unclear. It is possible that food-antigen sensitization in utero contributed to their condition. It has been suggested that eosinophilic colitis can be attributed to a transplacental sensitization of the colon by bovine milk antigens (9). However, it is unlikely that allergic reactions were involved in the LGB observed in these infants, inasmuch as enteral feeding had not been started, suggesting that the causes of bleeding were not antigen specific. Alternatively, the LGB could have simply been caused by an increased rate of hemocytopoiesis, as is often seen in premature infants. Thus, the eosinophilia observed in infants is often transient, possibly limiting the development of LGB, making it both less frequent and of shorter duration.
In the present study, we suggest 1 possible cause of LGB in early infancy, which may be associated with ECP and PAF released from the eosinophils. The prevalence of this condition is unknown. During a 12-month analysis in 2003, there were at least 6 infants with eosinophilia showing LGB before initial feeding in our hospital. Rectosigmoidoscopy was performed on 2 of these infants (reported here). These infants had a generally healthy appearance, and the extent of their disease was limited. Furthermore, all of them had negative results for microbial findings, and there was no evidence of viral infection or abnormal serum immunoglobulin M levels against viruses, such as cytomegalovirus and Epstein-Barr virus. This disorder is sufficiently common that practitioners should recognize it as a distinct clinical syndrome in infants with inflammatory colitis, and failure to appreciate this entity may lead to inappropriate therapeutic intervention. Because these changes are antigen nonspecific and are observed only transiently in neonates, we have termed this disease neonatal transient eosinophilic colitis, which represents a new category of eosinophilic gastroenteritis.
1. Dupont C, Kalach N, de Boissieu D, et al
. Digestive endoscopy in neonates. J Pediatr Gastroenterol Nutr 2005; 40:406–420.
2. Lake AM, Whitington PF, Hamilton AR. Dietary protein-induced colitis in breast-fed infants. J Pediatr 1982; 101:906–910.
3. Turck D, Michaud L. Lower gastrointestinal bleeding. In: Walker WA, Goulet O, editors. Pediatric Gastrointestinal Disease. 4th ed. Ontario: BC Decker; 2004. pp. 266–279.
4. Silber G. Lower gastrointestinal bleeding. Pediatr Rev 1990; 12:85–93.
5. Leung AK, Wong AL. Lower gastrointestinal bleeding in children. Pediatr Emerg Care 2002; 18:319–323.
6. Raine PA. Investigation of rectal bleeding. Arch Dis Child 1991; 66:279–280.
7. Stringel G, Mercer S, Sharpe D, et al
. Eosinophilic gastroenteritis. Can J Surg 1984; 27:182–183.
8. Ewer AK, Al-Salti W, Coney AM, et al
. The role of platelet activating factor in a neonatal piglet model of necrotising enterocolitis. Gut 2004; 53:207–213.
9. Sherman MP, Cox KL. Neonatal eosinophilic colitis. J Pediatr 1982; 100:587–589.