Chapter 57. Necrotizing Enterocolitis

Necrotizing enterocolitis (NEC) is the most common life-threatening disorder of the gastrointestinal tract among neonates, primarily occurring in infants born prematurely. It is characterized by inflammation and patchy necrosis of the bowel wall that may rapidly progress to systemic sepsis, bowel perforation, and death. Despite decades of research, the underlying etiology remains poorly understood, preventive strategies are inconsistent, and treatment options are controversial.

The incidence of NEC is reported between 0.72 and 1.1 cases per 1000 live births.1,2 Its occurrence is inversely related to birth weight and gestational age, the most important and consistent risk factors in its development. Preterm infants comprise the overwhelming majority of cases, with near-term and term infants accounting for between 5% and 25%.3,4 For very-low-birth-weight (VLBW) infants born at less than 1500 grams, the incidence reported from large multicenter studies ranges from approximately 5% to 12% of VLBW live births.1,5,6 Reports from individual institutions, however, range widely, from as low as 0.8% to 22%.6-8 Timing of disease onset is inversely related to gestational age, with near-term and term infants typically developing disease in the first week of life, while onset is more common after the first or second week for earlier gestational ages as shown in eFigure 57.1.9

Mortality from NEC is related to gestational age, extent of bowel involvement, and surgical intervention. In 2000, the in-hospital US mortality across all weight categories was estimated to be 15.2% of all NEC cases.1 Among very-low-birth-weight infants, mortality from NEC varies from 12% to 30%,1,2 with rates as high as 50% for extremely low-birth-weight infants born at less than or equal to 1000 grams.10,11 Mortality increases with the need for surgical intervention to as high as 50% and is directly related to length of remaining intact bowel. Recent data from Blakely and colleagues11 reported 75% survival for infants with more than 80 cm of normal bowel at the time of surgical intervention, 46% for infants with 10 to 80 cm of normal bowel, and no survival in infants with less than 10 cm of normal bowel present.11 Most large multicenter studies have shown no significant change in NEC incidence over the past 3 decades.2,5,12,13 Furthermore, NEC mortality has increased. From 1986 through the early 1990s, while overall neonatal mortality was declining, NEC-associated mortality increased from 11.5 per 100,000 live births to 12.3 per 100,000 live births.14 As increasing numbers of extremely premature infants survive, the numbers alive to develop NEC increase, specifically among a population also more likely to die from the disease.14,15

Numerous demographic and clinical risk factors for NEC have been proposed, with gestational age and birth weight the strongest and most consistent.6,8,10,16-18 Associations between gender and race are less consistent once illness severity and prematurity are controlled.1,8,17although males are more likely than females to die from the disease.1,14Prenatal and postnatal factors have been variably implicated in increasing NEC risk for the preterm population. Concern regarding use of antenatal indomethacin (AI) for tocolysis is based on animal models demonstrating decreased mesenteric blood flow and altered prostaglandin-induced gastrointestinal defense mechanisms.19,20 The Cochrane Collaborative meta-analysis of randomized controlled trials comparing AI to other tocolytics (N = 4 studies) or placebo (N = 2 studies) found no significant relationship to NEC,21 although the number of NEC cases in each trial was small. A meta-analysis of 11 observational studies of AI22 also found no significant relationship between AI exposure and NEC (odds ratio [OR] 1.4; 95% confidence interval [CI], 0.9–2.3). The association was significant, however (OR 2.2; 95% CI, 1.1–4.2), when including only studies (N = 4) with recent (within 72 hours of delivery) AI exposure. The limitations of these studies preclude conclusions regarding the relationship of AI exposure and NEC. Use of antenatal corticosteroids has been reported to have a variable effect on NEC risk17,23 (discussed under “Prevention” in this chapter). Postnatal clinical conditions variably implicated in increased NEC risk include low Apgar scores, presence of a patent ductus arteriosus, placement of umbilical catheters, and therapies such as H2 blocker,16 indomethacin, and ibuprofen therapies.24-30

Occurrence of NEC in term or near-term infants commonly involves the presence of risk factors including congenital heart disease, gastroschisis, intrauterine growth retardation, birth asphyxia, sepsis, polycythemia, and exchange transfusions.3,31-34 Site of involvement is typically colonic rather than jejunal and ileal, as commonly seen in preterm infants.3

In 1975, it was first proposed that 3 critical factors required for the development of NEC were the presence of bacteria, feedings, and injury to the mucosa,35 with the concept of a susceptible host added later.36 Current hypotheses continue to revolve around these same concepts, although understanding of their interactions and role in NEC development has evolved (Fig. 57-1).37

Bacteria and the Intestinal Ecosystem

A role for bacteria in NEC development has been hypothesized since early in the recognition of NEC as a disease entity.35 Evidence for the role of bacteria includes the fact that NEC is never evident in utero or in stillborns, the occurrence of clustered outbreaks, and animal studies.38,39 Many organisms have been implicated, including both gram-negative (eg, Clostridia species, Escherichia coli,Klebsiella species) and positive bacteria (eg, Staphylococcal species), as well as viruses (eg, rotavirus).40 However, no single common organism has been identified as the sole cause of NEC.41,42

Recent evidence supports the hypothesis that rather than a specific infection, NEC results from an abnormal pattern of postnatal intestinal colonization occurring in infants in the neonatal intensive care unit.43 For the premature infant, admitted to the neonatal intensive care unit and often exposed to antibiotics, bacterial colonization is delayed, with limited species diversity, often of a pathogenic type.43,44 Gram-negative organisms predominate, with a paucity of symbiotic bacteria like Lactobacillus and Bifidobacterium. The absence of symbiotic commensal bacteria, increasingly recognized for their contribution to continued postnatal intestinal development and health,45-47 may allow for overgrowth and invasion from colonizing intestinal flora.

The Susceptible Host: Immaturity of the Innate Immune Response and Inflammation

The strong relationship of NEC to gestational age at birth supports the concept that a component of its pathogenesis lies in the unique characteristics of an immature gastrointestinal tract. Abnormalities in the intestinal colonization pattern are superimposed on an immature intestinal epithelium, which responds to bacterial inflammatory stimulation with excessive proinflammatory cytokine production.48,49 At least part of this response is related to the innate immune system. The response is initiated by bacterial cell wall products, referred to as pathogen associated molecular patterns (PAMPs), which bind to a family of pattern recognition receptors that include the toll-like receptors (TLRs), present on the intestinal epithelium. These receptors activate a signaling pathway that results in translocation of the nuclear transcription factor kappa B (NF-κB) to the nucleus, which initiates the inflammatory cascade, proposed to be the final pathway of intestinal injury. Toll-like receptors 2 and 4 are abundantly present on human fetal enterocytes.50 In animal models, TLR4 expression38 and NF-κB activation51 rapidly diminish after birth, which is believed to be an adaptive response limiting the risk of inflammatory response to synergistic bacteria or low amount of pathogens. In contrast, neonatal animal models of NEC demonstrate a marked increase in epithelial TLR4 and NF-κB expression and associated inflammatory mediators.38,52-54 Animals with diminished TLR4 function38,54 or that were pretreated with an NF-κB inhibitor51 experienced decreased incidence and severity of NEC.

A number of inflammatory mediators are produced once the NF-κB cascade is activated. These include tumor necrosis factor α, platelet-activating factor (PAF), the interleukins (interleukins 1, 6, 8, 12, and 18), nitric oxide, and oxygen-free radicals.55,56 In infants with NEC, cytokine levels are increased and correlate to disease severity.57 Among these many mediators, PAF has been shown to play a central role.58 Exposure of intestinal mucosa to increased local concentrations of PAF results in increased apoptosis and mucosal permeability, activation of secondary inflammatory responses, and bowel necrosis.58Administration of antagonists to the PAF receptor or the PAF-degrading enzyme acetylhydrolase reduces the occurrence and severity of NEC.59,60

Immaturity of Repair Mechanisms

Epithelial repair mechanisms maintain protection of the epithelial barrier in response to injury. In a healthy state, small intestinal epithelial cell turnover occurs at a rate of 20 to 50 million cells per minute,45 mediated by a balance between cell proliferation and apoptosis, programmed cell death. In injury, the integrity of the epithelial barrier is rapidly reestablished because of this enormous regenerative power.61In neonatal animals, epithelial cell turnover in response to injury is much slower than in mature counterparts.62 Epithelial repair is further diminished in NEC, with intestinal epithelium demonstrating increased apoptosis and decreased cell proliferation and migration to the site of injury, a process termed intestinal restitution. A number of factors have been implicated in altering the response to injury in the immature host,63 including the elevated expression and activation of TLR4 and NF-κB53,54,64; diminished levels of epidermal growth factor, an important modulator of epithelial cell proliferation and apoptosis18; and inflammatory mediators. including PAF,58 nitric oxide,65 and reactive oxygen species.66

Excessive apoptosis is a significant mode of cell death in response to injury for a variety of pathologic conditions.66-68 In NEC, enterocytes in the apical villi of resected tissue from patients with NEC demonstrated extensive apoptosis.56 Apoptosis is an early event in a neonatal rat model of NEC, preceding the onset of gross morphologic changes.69 Pretreatment with an inhibitor of apoptotic pathways reduced the incidence of apoptosis and experimental NEC. Reactive oxygen species,66 PAF,58 and nitric oxide and its oxidative by-product peroxynitrite56,65 have all been implicated in pathologically increasing apoptosis, resulting in excessive cell death and tissue necrosis, while epidermal growth factor decreases apoptosis in NEC, contributing to its protective effect.70

Immaturity of Barrier Function

The intestinal barrier is a combination of epithelial, immunologic, and mucosal factors to prevent penetration of noxious substances. Many of these barrier mechanisms are not fully developed at birth including gastric acid secretion, peristaltic motility, mucus production, immunoglobulin A production, defensin anti-microbial protein production by Paneth cells, and tight junction integrity.

Gastric acidity to prevent excessive ingestion of harmful bacteria and digestive enzymes to hydrolyze antigens is diminished in the neonate, particularly the preterm infant.71 The detrimental effect of decreasing gastric acidity is evident in data from the National Institutes of Child Health and Human Development’s Neonatal Research Network study of 11,072 very-low-birth-weight infants demonstrating a greater incidence of NEC in infants who received H2-receptor blockers.16 Peristaltic motility patterns, which facilitate expulsion of noxious substances, are not fully developed until near term.72,73 Goblet cell response to injury by increased mucus production is diminished, with mucin proteins that are less sulfated and viscous than in more mature counterparts.9,63 Production of secretory immunoglobulin A is underdeveloped in the newborn, although high concentrations in colostrum and early human milk may help compensate.74 Defensins, antimicrobial proteins produced by Paneth and other intestinal epithelial cells, are also diminished in the immature intestine.75,76 Permeability of intestinal epithelial cells is limited by tight junctions, comprising a family of proteins, including transmembrane proteins claudin, occludin junctional adhesion proteins, as well as zonula occludin proteins. Prematurity, pathologic insults, and nutritional status all effect expression and function of these proteins,9 resulting in intestinal hyperpermeability, or “leaky gut,” and predisposing to bacterial translocation and immune system activation.63

Intestinal Circulation and Ischemia

The role of ischemia in NEC is based on the pathologic hallmark of the disease, coagulation necrosis, which is the histologic consequence of preceding ischemia. The initial hypothesis that attributed responsibility to the diving reflex, or asphyxia and low flow, has not been borne out in epidemiologic studies or physiologic animal models.77,78 Rather, evidence now supports that perturbations in control of vascular tone result in vasoconstriction and subsequent ischemia.79 While many factors come into play, the potent vasoconstrictor endothelin 1 and vasodilator nitric oxide are factors most commonly implicated in the development of NEC. The low vascular resistance of the newborn is attributed to high constitutive expression of nitric oxide by the endothelial isoform of nitric oxide synthase (eNOS) compared to endothelin 1. In intestinal segments from infants with NEC, arterioles were vasoconstricted and failed to demonstrate a physiologic response to eNOS.80 Additionally, tissue concentrations of endothelin 1 are higher compared to healthy segments, and blockade of endothelinA receptors reversed vasoconstriction.81,82 Thus, disruption of endothelial cell function by ischemia-reperfusion, sustained low perfusion, or proinflammatory mediators alter the endothelin 1–nitric oxide balance in favor of constriction,79 perpetuating a cycle of further intestinal injury.

Substrate

Ninety percent of infants who develop NEC have received feedings,83 resulting in speculation that intestinal injury results from substrate within the gastrointestinal tract. Newborns’ absorption of carbohydrate and fat constituents in milk is not fully matured, and these undigested components may serve as substrate for proliferation of enteric bacteria. Their metabolic by-product, hydrogen gas, reducing substances, and organic acids, along with undigested casein and long-chain fatty acids remain intraluminal. Exposure of the intestinal epithelium to these substances is proposed to result in an “intestinal inflammation” and injury.83

In summary, it is now hypothesized that rather than one specific bacterium, the premature infants in the neonatal intensive care unit develop an inappropriate pattern of intestinal colonization, with limited diversity and paucity of synergistic bacteria. The innate immune response elicits a robust proinflammatory cascade in response to this colonization, with poor ability to repair intestinal injury. Immaturity of the mucosal barrier functions further increases susceptibility to injury from bacteria and toxic substance. The end result of mucosal injury to an immature gastrointestinal tract is NEC.

Differences in susceptibility to development of NEC maybe be explained in part by genetic variation.84 Twins offer a natural testing ground for genetic effects, and they comprise a significant component of the neonatal intensive care unit population. The largest retrospective cohort study of twin pregnancies (N = 1407 twin pairs) found monozygotic twins at increased risk for NEC compared to dizygotic twins (OR 4.05; 95% CI, 1.97–8.35), after adjustment for weight and age.85 In contrast, a retrospective cohort analysis of 450 twin pairs born at 32 or fewer weeks’ gestation did not find a genetic link for NEC within twins when evaluated by zygosity concordance and controlling for clinical covariants.86

Individual susceptibility to a variety of disease states has been linked to single nucleotide polymorphisms in genes encoding for proteins involved in disease pathogenesis. A number of single gene polymorphisms have been evaluated for their relationship to development of NEC, including inflammatory mediators, (interleukins 4, 12, and 18; tumor necrosis factor α; interferon γ),87-89vasoactive or angiogenic factors agents, (nitric oxide, vascular endothelial factor),90,91 pattern recognition receptors (TLR4, CD14, nucleotide oligomerization domains [NOD2 and CARD15])92,93, and thrombophilic agents (factor V Leiden, factor II, and methyltetrahydrofolate reductase). Others have looked more broadly at factors associated with prematurity.95 No clear single gene association to NEC is yet evident. Additional studies correlating genotype to protein phenotype and studies in larger populations are needed.

Onset of NEC may be subtle and insidious or catastrophic. Early signs can be nonspecific and common in premature infants, including temperature instability, signs of feeding intolerance such as delayed gastric emptying, and mild abdominal distention. Abdominal distention, bilious emesis or residuals, and bloody stool are more specific to NEC but can also be seen in other disease states. Abdominal distention, bilious emesis or residuals, and bloody stool are more specific to NEC but can also be seen in other disease states. As the disease progresses, abdominal findings become more severe. Abdominal distention may become marked, abdominal wall erythema may occur from necrotic bowel inflaming the abdominal wall, with development of a blue cast indicative of perforation and meconium in the peritoneal cavity. The clinical findings of NEC are not diagnostic and can be present in other systemic and gastrointestinal pathology. The differential diagnosis includes sepsis with ileus, spontaneous intestinal perforation (SIP), inspissated meconium syndrome, Hirschsprung disease with enterocolitis, viral gastroenteritis, volvulus, feeding intolerance, milk protein allergy, and metabolic disorders.

Infants with SIP are often included in studies of NEC because clinically the presentation is pneumoperitoneum, which is difficult to distinguish from stage III NEC. Spontaneous intestinal perforation is distinct from NEC, distinguished most clearly by histopathology. The mucosal margins of the perforation appear healthy, without evidence of coagulation necrosis.97 Differences in patient populations, risk factors, and presenting signs may be helpful in distinguishing the entities and point to a difference in their underlying etiologies. Compared to infants with NEC, infants who develop SIP are smaller, more immature, and more likely to have been treated with indomethacin, postnatal steroids, or both, in the first days of life.98,99 Two subpopulations of SIP have been identified: (1) extremely low-birth-weight infants who develop SIP on day of life 7 to 10 after exposure to indomethacin and glucocorticoids and (2) larger infants who develop early SIP on day of life 0 to 3 and were less likely to have been exposed to risk factors.100 Onset of SIP is often abrupt, with pneumoperitoneum the first radiographic finding and with no evidence of pneumatosis. Infants with SIP also have a more favorable prognosis, with less mortality than infants with surgical NEC.99

Diagnostic Evaluation

Laboratory findings in NEC are nonspecific and reflect the associated inflammatory process, sepsis and disseminated intravascular coagulopathy. They can include neutropenia, leukocytosis, thrombocytopenia, elevated C-reactive protein, metabolic acidosis, glucose instability, and hyperkalemia.While the severity of laboratory findings has been correlated with outcome, none alone has the predictive ability to confirm diagnosis or guide management.101-103 The distinguishing feature and radiographic hallmark of NEC is pneumatosis intestinalis (eFig. 57.2 and Fig. 412-3). Pneumatosis intestinalis represents the intramural accumulation of gaseous by-products from bacterial fermentation. Dissection of gas into the venous system results in air outlining the distribution of the portal venous collecting system, evident as linear densities over the liver on radiograph, and is termed portal venous air. The radiographic finding of pneumatosis alone is not related to disease severity; however, extensive pneumatosis or the presence of portal air suggests more severe involvement and carries a poorer prognosis.104-106 While the presence of pneumatosis intestinalis is pathopneumonic for NEC, it is important to note its absence does not rule out the diagnosis. Other radiographic findings include intestinal distention, which is the earliest and most common radiographic finding; diminished bowel gas, suggesting loss of motility; and fluid-filled loops, a persistent fixed, dilated loop suggesting ischemia, or pneumoperitoneum, indicating intestinal perforation. The diagnosis of free air in the peritoneal cavity is easily missed on a flat radiograph of the abdomen. For that reason, evaluation of NEC should always include either a decubitus or cross-table lateral radiograph. Recently, use of sonography for NEC diagnosis has been proposed, offering the advantage of real-time direct images of the abdominal structures and fluid in the peritoneal cavity.107 There are, however, no studies comparing the 2 modalities, and at this time standard radiography remains the mainstay of diagnosis.

Clinical staging of NEC (Table 57-1) is based on Bell criteria, a combination of clinical, laboratory, and radiographic findings.108,109Stage I includes nonspecific clinical manifestations that are suggestive of but not diagnostic for NEC and have no radiographic findings or signs of mild ileus. The diagnosis is established at stage II, with the appearance of pneumatosis intestinalis or portal venous air, with stage III indicating a more advanced stage and including intestinal perforation. Bell criteria were established in the presurfactant era of neonatology, a time when survival of extremely low-birth-weight infants was rare. As increasing numbers of these infants survive, the occurrence of other acquired neonatal intestinal diseases, particularly SIP, has increased and made diagnosis and staging of NEC based on Bell criteria problematic. Gordon and colleagues99 recently proposed a guideline in the diagnosis of acquired neonatal intestinal diseases (eTable 57.1) that discriminates between entities according to timing of onset; clustering of cases; and clinical, radiographic, and pathologic findings.

Pathologic Findings

Grossly, the bowel appears irregularly dilated, gray, or dark purple to black in areas of extensive hemorrhage and necrosis (Fig. 57-2).The most common site of involvement is the terminal ileum, although any part of the small or large intestines may be involved. Grossly, the bowel appears irregularly dilated, gray, or dark purple to black in areas of extensive hemorrhage and necrosis (Fig. 57-2). The involved segment is distended, friable, and congested, or it can be frankly gangrenous; intestinal perforation with accompanying peritonitis may be seen. The dominant histopathologic finding is coagulative, or ischemic mucosal necrosis (eFig. 57.3).110-112Both acute and chronic inflammation is also typically evident along with ulceration, bacterial colonization, and pneumatosis intestinalis (eFig. 57.3). Congestion and thrombi of the small mesenteric arterioles and venules is common, while thrombi in larger vessels is not. Perforation may occur in severely affected areas. Reparative changes, such as the formation of granulation tissue and fibrosis, begin shortly after the acute episode.

The mainstay of treatment for NEC is medical stabilization, including gastric decompression, bowel rest, cardiovascular support, and systemic antibiotics. There is little information regarding choice, duration, or outcomes among various antibiotic regimens.113,114 Broad spectrum antibiotics that cover bowel flora, such as ampicillin and gentamicin, are typically used. A third agent, such as clindamycin, is commonly added to improve anaerobic bacterial coverage. Concern over coagulase staphylococcus as a pathogen and emergence of resistant organisms has led to treatment with vancomycin and third-generation cephalosporins at some centers.115,116 Choice of antibiotics should be guided by local resistance patterns and disease severity. Many patients will be hypovolemic and require fluid resuscitation as well as correction of acidemia. The acidemia is often mixed respiratory and metabolic, and endotracheal intubation may be required to treat apnea, hypercarbia, and/or hypoxia. Coagulopathy and electrolyte abnormalities may rapidly develop and require prompt attention. As systemic signs of sepsis increase, maintaining intravascular volume with fluid resuscitation and/or blood transfusion is critical. Inotropic support may be required to maintain hemodynamic stability. Close monitoring of clinical, laboratory, and radiographic status is essential. Frequent serial physical examinations, with radiographs (including left lateral decubitus or cross-table lateral to identify free air) every 6 to 8 hours, and laboratory evaluations, including an initial blood culture and serial blood counts and acid-base status, are recommended.

While many infants can be managed medically, one third to one half will develop severe necrosis or perforation and require surgical intervention.6,10,17,117,118 Perforation continues to be the only absolute indication for surgical intervention,117 with relative indications including severe clinical deterioration, abdominal wall erythema, and some advocating inclusion of extensive pneumatosis and portal venous gas.105,106,117,119,120

Approach to surgical management of these infants is controversial, with laparotomy the traditional approach. As early as the 1970s, however, peritoneal drainage was introduced, with the initial intent of stabilization until infants were able to undergo laparotomy.121 Over time, laparotomy was increasingly performed after drain only if clinical deterioration occurred. In 2005, Blakely and colleagues11 reported results of a prospective cohort study of 16 academic centers within the National Institute of Child Health and Human Development Neonatal Research Network of 156 extremely low-birth-weight infants (⩽ 1000 g, N = 156) undergoing surgical intervention for NEC. Fifty-one percent (N = 80) were initially treated with drain, with only 23% of these having a subsequent laparotomy.11 Outcomes of infants receiving drains versus laparotomy vary depending on center and surgeon.11,122,123 Selection bias is present in many of these studies, with infants undergoing drain smaller and sicker than those undergoing laparotomy.124 The first randomized controlled trial of laparotomy versus drain125 found no difference in 90-day mortality between treatment groups (34% versus 35%). Questions and controversy remain, however, given the small number of infants in the 1000- to 1500-gram weight group (N = 10 drain and 17 laparotomy), lower survival in infants eligible but not randomized who received laparotomy (15%), and the need for long-term outcomes.126

For infants who survive NEC, morbidity is high. NEC is related to higher rates of hospital morbidities, including growth failure, chronic lung disease, and hospital-acquired infections.127-129 Occurrence of intestinal strictures is a common complication, occurring weeks to months later. Lengths of stay and hospital costs are significantly lengthened, particularly in surgical NEC. Long-term neurologic outcomes are also adversely affected. Infants who develop NEC are at higher risk of neurodevelopmental impairment, including a higher risk of cerebral palsy; visual, cognitive, and psychomotor impairment; and continued growth failure.128-132 Infants with NEC requiring surgery are at highest risk for neurodevelopmental impairment, with 2 to 3 times the risk compared to those managed medically.130,131 For these infants, there is also the risk of short-bowel syndrome and dependence on parenteral nutrition, which may require intestinal and/or liver transplantation depending on the amount of intestine lost.

Prevention of NEC is an area of priority for research37 and encompasses both antenatal and postnatal strategies133 (Table 57-2).

Antenatal Corticosteroids

A number of studies report antenatal steroids to be protective against development of NEC, even after controlling for birth weight, gestational age, and complications.134,135 Others report no difference136 or increased risk.17 A Cochrane Collaborative meta-analysis considering only randomized controlled trials found a significantly decreased risk (relative risk [RR] 0.46; 95% CI, 0.29–0.74) among 8 studies and 1675 infants.23 Use of antenatal steroids is not driven by NEC risk, although at this time, the evidence supports their use.

Human Milk

The American Academy of Pediatrics recommends human milk as the source of enteral nutrition for term and preterm infants.137 Controlled trials to evaluate the effect of maternal milk in preterm infants are difficult because of ethical issues regarding randomization. The randomized controlled trials that have been done relied on the use of donor milk, either as a supplement to mother’s milk or as the sole source of feeds. Among 5 randomized controlled trials that evaluated use of donor milk, none individually found a statistically significant difference in the rate of NEC, although each showed a trend toward decreased disease incidence.138-142 Meta-analysis of these trials demonstrated a statistically significantly higher incidence in the formula-fed group (RR 2.5; 95% CI, 1.2–5.1). The pooled estimate suggests that 1 extra case of necrotizing enterocolitis will occur in every 33 infants who receive formula rather than human milk.143 The protective effects of human milk are likely multifactorial and have been related to anti-inflammatory growth factors and innate immune components.144 Advantages to use of human milk reach beyond those related to NEC, and use of human milk in the preterm population is supported unless specifically contraindicated.137 Providing mothers with early prenatal and postnatal counseling on use of human milk has been shown to increase the initiation of lactation and neonatal intake of mother’s milk without increasing maternal stress or anxiety.145

Enteral Feedings

Most preterm infants who develop NEC have received enteral feeds,146 leading neonatologists to focus on manipulation of enteral feedings as a potential mechanism for prevention. Intercenter differences in NEC have been related to feeding practices, fueling speculations that enteral feeding regimens may impact disease onset.8,147,148 It has been proposed that NEC is related more to the variation in feeding practices than to any one practice alone.149 Reports from individual centers indicate a decline in the incidence of NEC with implementation of standardized feeding regimens.7,150-153 A meta-analysis of observational studies demonstrated that introduction of a standardized feeding regimen decreased the incidence of NEC (RR 0.13; 95% CI, 0.03–0.5) in low-birth-weight infants.149 The observational nature of the studies, significant variation among feeding regimens, and population characteristics limit conclusions on efficacy and justify controlled trials to verify effect.

Many aspects of feeding regimens have been evaluated in relationship to NEC, including use of trophic feeds, timing of initiation, rapidity of advancement, continuous versus bolus, and feeding site (transpyloric versus gastric). Trophic feedings were introduced in response to the prolonged withholding of enteral feedings, up to a month, in an attempt to prevent NEC.154 The effect of trophic feedings on NEC risk was analyzed in a meta-analysis of nine randomized controlled trials that included NEC as a primary or secondary outcome.155 None of the studies showed a significant effect on NEC risk individually, and there was no significant cumulative impact on NEC on meta-analysis (RR 1.16; 95% CI, 0.75–1.79). Only 1 study has compared trophic feedings to advancing feedings156 and found a significantly higher NEC rate in the advancing group than in the minimal feeds group (10% vs 1.4%, P < 0.03). The study had a low number of total NEC cases (N = 8), making point estimates of relative risk imprecise with wide confidence intervals and only a marginally significant relative risk (RR 0.14; 95% CI, 0.02–1.07). It remains unclear whether trophic feedings are beneficial relative to no feedings or advancing feedings in high-risk infants.155 Rate of enteral feeding advancement, rapid versus slow, on NEC risk was evaluated in three randomized controlled trials. No difference in NEC risk was found between rapid or slow advancement, defined as 10 to 20 ml/kg/day and 20 to 35 ml/kg/day respectively.157 Only one trial evaluated early (median of 2 days) compared to late (median of 5 days) initiation of enteral feedings,158 with no difference in NEC risk evident. Similarly, no differences are evident for NEC outcome with continuous versus intermittent bolus feedings159 or transpyloric versus gastric feedings.160

A review of these studies highlights the difficulties in clinical studies evaluating effect of feeding regimens on NEC. Studies were frequently single-center studies, resulting in low numbers of NEC cases. The imprecise point estimates and wide confidence intervals make it difficult to make definitive conclusions. Caretakers are typically not blinded, introducing the potential for diagnostic surveillance bias (eg, more x-rays in one group due to caretaker bias). Finally, effect of feeding regimen on NEC must be evaluated in the context of overall benefit to each infant. Alterations in feedings may have other effects on growth, catheter use, hospital-acquired infection, and length of stay, among others. As a result of these limitations, we are left with uncertainties regarding the effects of feeding regimens on NEC risk.

Probiotics

The World Health Organization defines probiotics as live microorganisms that, when administered in adequate amounts, confer a benefit to the host.161 Enteral administration of probiotics for prevention of NEC is based on data regarding human milk feedings that promote intestinal colonization with Lactobacillus and Bifidobacterium162 and also decrease the risk of NEC.143 Among the three randomized controlled trials of probiotics in preterm infants for which NEC was the primary outcome, one showed no significant difference in NEC incidence,163 although the trend was for protection, while 2 showed a significant decline.164,165 The meta-analysis of all randomized controlled trials for which NEC was a primary or secondary outcome estimated a lower risk of NEC in the probiotics group (RR 0.36; 95% CI, 0.2–0.65) compared to controls. Based on these results, 25 infants would need to be treated with probiotics to prevent one case of NEC.166 Questions regarding safety and protocol differences remain. None of the trials reported sepsis from organism in the probiotics supplement, although cases of Lactobacillus sepsis in immunocompromised children, including neonates, are reported in the literature.167-169 There was variation among studies regarding patient demographics; the type of probiotics used; and timing of onset, dose, and duration. Despite these issues, favorable results support larger clinical trials to evaluate efficacy and safety.170,171

Other strategies have been evaluated for prevention of NEC. Studies evaluating fluid management in the preterm population indicate that careful restriction of water intake so that physiological needs are met without allowing significant dehydration decreases the risk of NEC.172 Administration of oral IgG-IgA decreased the incidence of NEC in one randomized controlled trial of low-birth-weight infants (< 2000 g at birth),173 although the total number of NEC cases was small (N = 6), and the reduction in NEC risk was marginally significant with wide confidence intervals (RR 0.08; 95% CI, 0.0–1.39).174 Meta-analysis of additional studies evaluating enteral immunoglobulin (IgG alone or in combination with IgA), concluded that the evidence does not support the use of oral immunoglobulin for the prevention of NEC (RR 0.84; 95% CI, 0.57–1.25).174 No difference in NEC is evident with intravenous immunoglobulin.175,176 A meta-analysis of randomized controlled trials evaluating the effect of oral antibiotics found a significant decrease in the risk of NEC177 but was associated with development of resistant bacteria, precluding recommendation for use. Supplementation with the amino acid glutamine was found to be ineffective in decreasing NEC risk.178 Supplementation with arginine trended toward a protective effect for stage I or greater NEC, but results were not significant when cases were restricted to stage II or greater.179,180