Diagnosis

The diagnosis is typically made by the identification of pneumatosis intestinalis (air in the bowel wall) and/or portal venous gas on abdominal radiograph, although in some cases of NEC, commonly in unfed patients, pneumatosis is not appreciated (Figure 94-1). In these situations, NEC may be diagnosed surgically or pathologically, or in some instances by ultrasound appreciation of portal venous air. Bell and colleagues suggested a classification scheme that differentiates suspected NEC (stage I) from proven NEC (stage II) and advanced NEC (stage III with peritonitis and/or perforation) (Table 94-1).⁴ In this scheme, stage I NEC includes mild systemic signs, abdominal distention with changes in feeding intolerance, but no confirmatory radiographic evidence. Stage II or proven NEC has similar symptoms or signs with pneumatosis and/or portal venous gas, and stage III demonstrates significant systemic signs with radiographic evidence of intestinal perforation (pneumoperitoneum). This classification scheme is useful in occasional circumstances, especially when one analyzes studies that evaluate NEC; readers should be wary of interventions that appear to influence stage I, suspected disease without altering definitive NEC.

Treatment

No specific treatment approaches have influenced the outcome of NEC, and as such, interventions are supportive and include fluid resuscitation, withholding feedings with gastric decompression, antibiotics to cover likely enteric pathogens, correction of acidosis, anemia, and thrombocytopenia as needed, and blood pressure support. Although blood cultures are positive in approximately 30% of NEC cases and are thought to reflect a breakdown in the mucosal barrier leading to bacterial translocation, intraluminal enteric bacterial pathogens are thought to contribute to the pathophysiology.⁴⁸ Antibiotic coverage in this condition typically includes ampicillin and an aminoglycoside or third-generation cephalosporin, although occasionally Staphylococcal species are heavy colonizers of the intestinal tract, and nafcillin or vancomycin should be considered. In situations with suspected or proved intestinal perforation, aggressive anaerobic coverage with clindamycin is often added. Although routine treatment usually proceeds for 7 to 10 days with NPO and antibiotics in uncomplicated, medical NEC, length of treatment has not been carefully studied, and one report has suggested earlier refeeding after ultrasonographic evidence of portal venous air has resolved.¹⁰

Surgical intervention for NEC is required in 30% to 50% of cases reported, although the approach and timing of these procedures remain controversial. Most physicians agree that intestinal perforation in a NEC patient requires surgery, but based on three clinical trials, patients treated with a bedside drain had similar rates of death and short-term intestinal function as those undergoing definitive laparotomy.9,64,77 In surgical trials for NEC treatment, almost 50% of patients treated with a drain never required a second procedure. In patients without perforation, but with worsening disease as manifested by abdominal discoloration and distention, persistent thrombocytopenia and acidosis, and respiratory failure, exploratory laparotomy is often undertaken to remove a discrete segment of necrotic bowel or to confirm the viability of enough remaining intestine to sustain life. Nonetheless, the timing and utility of these procedures in these complex patients have not been adequately studied.

Outcome

Approximately 30% of patients with radiologic evidence of pneumatosis intestinalis have mild disease and require a period of bowel rest, but no surgical intervention. Another 30% of patients eventually succumb to the disease, with most presenting acutely with rapid deterioration and death. Survivors from NEC have a significant risk for intestinal stricture; in some reports, as many as 25% develop partial bowel obstruction weeks or months following the initial presentation. Some patients develop short bowel syndrome from NEC; postsurgical patients have an incidence as high as 11% of short bowel syndrome, and these patients are an extremely difficult group for whom to care. Novel medical and surgical interventions have made only modest improvements on the morbidity and mortality associated with this dreaded complication. Of concern, accumulating data suggest that the neurodevelopmental outcome of NEC patients is significantly worse than their gestational age and birth weight–matched controls with similar respiratory disease. The morbidity includes mental retardation as well as an increased incidence of cerebral palsy, hypothetically related to white matter injury from cytokine mediators involved in the systemic inflammatory cascade. Further studies are needed to confirm this association and clarify the significance of these important results. Nonetheless, NEC is a major financial burden nationwide, with increased initial hospital costs (owing primarily to longer length of stay) of $60,000 for a case of medical NEC and up to $200,000/patient for surgical disease.⁸ Based on the current epidemiology, this projects an annual cost burden of $1 billion to the US health care industry without taking into account the long-term care issues associated with impaired survivors.

Pathology

Clues to the etiology are suggested by the pathologic changes observed in surgical specimens and autopsy material, including coagulation necrosis (suggesting some component of ischemic injury), inflammation (acute and/or chronic), and less commonly, ulceration, hemorrhage, reparative change, bacterial overgrowth, edema, and pneumatosis intestinalis.

Prematurity

Greater than 90% of NEC cases occur in premature infants; there is consistently a higher risk with lower gestational age and birth weight,⁵⁶ and as such, prematurity is the most consistent and important risk factor. Although there are many differences between preterm and full-term neonates, the specific underlying mechanisms responsible for this predilection of NEC in the premature condition remain incompletely elucidated. Studies in humans and animals have identified alterations in multiple components of intestinal host defense,⁹² motility,⁵ bacterial colonization,⁹⁵ blood flow regulation,⁷¹ and inflammatory response^13,27,67 that may contribute to the development of intestinal injury in this unique population.

Enteral Feeding

Because most cases of NEC occur after feedings have been introduced (>90%), enteral alimentation is a significant risk factor for disease in premature infants. Historical reports identified the onset of NEC several days following the first feed, but in studies of extremely low birth weight infants, NEC may be diagnosed several weeks after initiating enteral supplementation.¹¹ This change may reflect current neonatal practice that typically uses early trophic or hypocaloric feedings, characterized by small volumes and slow rates of increase, without a significant impact on the development of NEC. Although the precise relationship between enteral feedings and NEC remains poorly understood, studies have identified the importance of breast milk (versus formula), volume and rate of feeding advancement, osmolality, and substrate fermentation as important factors.

Breast milk feeding appears to reduce the incidence of NEC in human studies and in carefully controlled animal models.16,56 Breast milk contains multiple bioactive factors that influence host immunity, inflammation, and mucosal protection, including secretory IgA, leukocytes, lactoferrin, lysozyme, mucin, cytokines, growth factors, enzymes, oligosaccharides, and polyunsaturated fatty acids, many of which are absent in neonatal formula preparations (Table 94-2). Specific intestinal host defense factors acquired from breast milk such as EGF, PUFA, PAF-acetylhydrolase, IgA, and macrophages are effective in reducing the incidence of disease in animals,^17,27,55 and some have been effective in limited human trials.^20,29 Nonetheless, breast milk is not completely protective against NEC in premature infants; the largest prospective trial identified a reduction by 50% in most birth weight–specific groups, although there was not a statistically significant reduction in disease observed in a randomized subset from this cohort.⁵⁶ Because of ethical considerations, it seems unlikely that such an investigation will be accomplished, although there is renewed interest in evaluating donor milk samples and alternative human milk preparations in this context.⁸² Because most premature infants receive breast milk via the nasogastric route after artificial collection by mothers and subsequent freezing, it has been suggested that the lack of the normal maternal-infant physical interaction during feeding interferes with specific milk immunity, thereby reducing the protection against the neonate’s unique microbial flora. As discussed shortly, the particular microbial profile in the neonate’s intestinal environment may contribute to initiation of NEC.

Specific components of milk feedings have been implicated to cause mucosal injury in the high-risk neonate and subsequently stimulate the development of NEC. Studies have shown that hyperosmolar formulae resulted in disease and that addition of medication to feedings can markedly increase osmolality.⁹⁷ Animal studies have shown that short-chain fatty acids such as propionic or butyric acid can cause damage to developing intestine and that colonic fermentation leading to production of these acids by the host microflora may occur in situations of carbohydrate malabsorption.¹² This pathway may be especially problematic in the premature infant, partially deficient in lactase activity and other brush border enzymes. Finally, an intriguing new hypothesis suggests that bile acid accumulation may lead to mucosal injury in the unique environment of the preterm neonate.³⁵

Different approaches to feeding have been associated with the initiation of NEC. Early studies suggested that rapid volume increases with full-strength formula increased the incidence of disease, and protocols were designed to limit feeding advancement. Several studies have shown that early hypocaloric or trophic feedings are safe and improve gastrointestinal function in very low birth weight (VLBW) infants.⁸³ Feeding advancement has been evaluated, and the results suggest that judicious volume increase may be safer,⁶ although this remains controversial. It has been postulated that overdistention of the stomach with aggressive volumes may compromise splanchnic circulation, leading to intestinal ischemia. Nonetheless, there remains little clarity on the safety of differing feeding practices on the incidence of NEC, and additional trials will be needed to answer this challenging question.

Intestinal Ischemia/Asphyxia

Early observations on the pathophysiology of NEC suggested that profound intestinal ischemia led to intestinal necrosis in unusual clinical situations.⁹⁰ Similar to the “diving reflex” observed in aquatic mammals, it was hypothesized that in periods of stress, blood flow was diverted away from the splanchnic circulation resulting in bowel injury. Although early epidemiologic observations identified asphyxia as an important risk factor, subsequent studies have shown that the majority of NEC cases are not associated with profound impairment in intestinal perfusion. In animal models, studies have shown that the reperfusion following intestinal ischemia is required in the initiation of bowel necrosis; occlusion of the mesenteric artery for a prolonged period of time results in only mild histologic changes atypical for full-blown NEC.

Neonatal animals have been shown to have differences in the intestinal circulation that may predispose them to NEC. The basal intestinal vascular resistance is elevated in the fetus, and soon following birth, decreases significantly, allowing for rapid increase in intestinal blood flow that is necessary for robust intestinal and somatic growth.⁷⁴ It has been shown that this change in the resting vascular resistance is dependent on the balance between the dilator (nitric oxide) and constrictor (endothelin) molecules, and the myogenic response, and altered levels of these vasoactive mediators have been identified in human NEC samples.^72,73 Perhaps more relevant than basal vascular tone, studies have shown that the newborn has alterations in response to circulatory stress, resulting in compromised intestinal flow and/or vascular resistance. In response to hypotension, newborn animals (3- but not 30-day-old swine) appear to have defective pressure-flow autoregulation, resulting in compromised intestinal oxygen delivery and tissue oxygenation.^71,75 In addition, in the face of arterial hypoxemia, the newborn intestinal circulatory response differs from that of older animals. Although following modest hypoxemia, intestinal vasodilation and increased intestinal perfusion occur; severe hypoxemia causes vasoconstriction and intestinal ischemia and/or hypoxia, mediated in part by loss of nitric oxide production. There are multiple chemical mediators (nitric oxide, endothelin, substance P, norepinephrine, and angiotensin) that impact on intestinal vasomotor tone, and in the stressed newborn, abnormal regulation of these may result in compromised circulatory autoregulation, leading to perpetuation of intestinal ischemia and tissue necrosis.^65,66,76

The Possible Role of Packed Red Blood Cell Transfusions in Necrotizing Enterocolitis

More than 25 years ago, McGrady and colleagues noted a markedly increased relative risk of developing NEC in preterm patients who received packed red blood cell (PRBC) transfusions compared with preterm infants who did not.⁶¹ Subsequent studies have supported the notion that there is an association between PRBC transfusion and the development of NEC, particularly with PRBC transfusions that occur more than a few weeks after birth, and it has been suggested that these occur temporally within 48 hours after the transfusion has been completed. There are several clinical differences that are described between PRBC-transfusion–associated NEC with “routine” NEC, including PRBC-transfusion NEC having (1) lower birth weight and gestational age, (2) more significant anemia, and (3) more varied gestational age, but these have not been universal in all trials. Based on several case control and cohort studies, there have now been several reports supporting an association between PRBC transfusion and neonatal NEC, accounting for 20% to 30% of all NEC cases reported in these series.^21,49 Nonetheless, it remains unclear whether these transfusions clearly play a causal role in the initiation of gut injury.

Several mechanisms have been suggested to explain the phenomenon of PRBC transfusion–associated NEC. Similar to transfusion-associated lung injury (TRALI) that is observed in adults following transfusion in a small minority of patients, La Gamma and others have hypothesized that transfusion-associated gut injury (TRAGI) occurs in preterm infants, and the pathophysiology of this response is complex and could involve a variety of circulatory, immune, and host defense factors that are activated from stored packed red blood cells and/or PRBC products.⁵⁹ A second hypothesis suggests that premature infants who require transfusion are severely anemic and the PRBC transfusion results in intestinal blood flow changes that ultimately activate intestinal necrosis. The third hypothesis considers that stored PRBC preparations develop changes that might be risky to the developing intestine and intestinal microcirculation, including nitric oxide deficiency, increased bioactive platelet activating factor (PAF), reduced RBC deformability, and perhaps increased aggregation, adhesion, and thrombogenic effects of these banked aliquots. Nonetheless, if PRBC transfusions do initiate biologic changes leading to NEC, the etiology is not completely understood.

Because association studies inherently have a high risk for bias, Kirpalani and Zupancic reviewed the current understanding of PRBC transfusions and NEC. These investigators carefully considered the various studies and found that in the few randomized, controlled PRBC transfusion trials, there was a higher risk of NEC in the restrictive, or less transfused group, contradicting most of the observational studies.⁴⁷ Nonetheless, meta-analyses performed for case control studies and cohort studies, as expected, showed that PRBC transfusions were associated with a significantly higher risk of NEC. Interpretation of these opposite effects suggested that a confounder might independently lead to both transfusion and NEC, and it may be suggested that the clinical scenario of anemia, apnea and bradycardia, lethargy, and tachycardia could lead to the clinical decision to order a PRBC transfusion, but could also be the classic prodrome for developing NEC. Therefore, only in a randomized, controlled trial could these confounders be clearly elucidated, and it is universally accepted that preplanned randomized, controlled trials are necessary to clarify the clear role of PRBC transfusions in neonatal NEC.