Necrotizing enterocolitis

Abstract

Necrotizing enterocolitis is a devastating condition that affects premature infants, and remains one of the leading causes of death in neonatal intensive care units worldwide. Current approaches to the diagnosis and the treatment of NEC have changed minimally in the past 30 years, in part due to an incomplete understanding of the biologic mechanisms of the disease, and a lack of treatment options. Importantly however, recent research advances have improved our understanding of the pathogenesis of NEC, resulting in a broader approach to diagnosis, prevention and treatment. In parallel, an increased understanding of the long-term complications of NEC have shed light on the importance of early decision making for patients with this disease. This Disease Primers includes insights from an international team of clinician-investigators from Europe, Asia and North America, and which now seeks to answer some of the key questions in the NEC field, with a focus on understanding epidemiology, pathogenesis and optimal management, including the fact that many NEC infants require surgery, and to understand the morbidy that survivors experience. We also highlight how the future care of patients with NEC is likely to include a sophisticated diagnostic approach that includes novel molecular and imaging analyses, leading to the precise delivery of targeted molecular and cellular therapies, so as to improve outcomes for children with this disease.

Introduction

Necrotizing enterocolitis (NEC) is the leading cause of death from gastrointestinal disease in premature infants, and affects approximately 1 in 20 premature births¹. The development of NEC reflects an exaggerated pro-inflammatory response in the intestinal mucosa of the premature host to colonizing microbes within the lumen of the gut that is mediated by bacterial receptors including toll like receptor 4 (TLR4). The subsequent inflammatory response leads to disruption of the intestinal barrier, leading to translocation of bacteria into the circulation within the intestinal mesentery, leading to vasoconstriction and intestinal ischemia. The subsequent activation of the host immune system by the translocated microbes can lead to overwhelming sepsis, organ failure and death in the most severe cases.

The diagnosis of NEC relies on a combination of clinical, radiological and hematological parameters, and is limited by a lack of clear diagnostic criteria and disease definitions. In clinical practice, NEC is often classified using Bell criteria (Table 1) which is useful when describing the severity of disease in an individual patient.

Table 1:

Two diagnostic scoring systems for NEC

Stage	Systemic Signs	Abdominal Signs	Radiographic Signs	Surgical findings
Modified Bell
IA (suspected)	Including evidence of Temperature instability, apnea, bradycardia, and/or lethargy	Gastric retention, abdominal distention, emesis, and/or heme-positive stool	Normal or intestinal dilatation, mild ileus	Not included
IB (suspected)	Same as above	Grossly bloody stool	Same as above	Not included
IIA (definite NEC)	Same as above, plus mild metabolic acidosis and thrombocytopenia	Same as above, plus absent bowel sounds with or without abdominal tenderness	Intestinal dilatation, ileus, and/or pneumatosis intestinalis	Not included
IIB	Same as above, plus mild metabolic acidosis and thrombocytopenia	Same as above, plus absent bowel sounds, definite tenderness, without abdominal cellulitis or right lower quadrant mass	Same as IIA plus ascites	Not included
IIIA (Advanced, severely ill, no perforation)	Same as IIB, plus hypotension, bradycardia, severeapnea, combined respiratory andmetabolic acidosis, DIC, and/or neutropenia	Same as above, plus signs of peritonitis, marked tenderness, and /or abdominal distention	Same as IIA, plus ascites	Not included
IIIB (Advanced, severely ill, perforation	Same as IIIA	Same as above	Same as above plus pneumoperitoneum	Not included
Hackam clinical presentations
Classical NEC	Instability including abnormal heart rate, breathing pattern or blood pressure.	Abdominal tenderness	Pneumatosis, might proceed to pneumoperitoneum	Needed when perforated
Medical NEC that fails to improve	Instability	Abdominal tenderness	Pneumatosis	Needed, especially to protect the brain from ongoing inflammation
Abdominal tenderness in the presence of portal venous air	Instability	Abdominal tenderness	Portal venous air, no perforation	Needed, similar to perforation because portal gas is an indication for extensive necrosis
Staccato NEC	Rapid progression to overwhelming sepsis	Abdominal distention and bloody stools	Not included – clinical signs only	Urgent laparotomy or drainage/silo placement
NEC totalis	Variable clinical presentation	Variable clinical presentation	Clinical signs only	Vast majority of small bowel is necrotic

All signs and symptoms are required to accept the diagnosis of a particular stage, unless otherwise noted.

Treatment of NEC depends on the severity of disease and involves a combination of medical and surgical approaches. However, treatment is unsuccessful in many cases and NEC is fatal in around 25% of patients². Accordingly, NEC prevention strategies, namely, administration of donor breast milk and avoiding excessive use of antibiotics, have emerged as critically important in the overall management of patients with NEC, and which have significantly lowered the incidence of NEC in many centers.

Despite these features however, NEC remains a disease of incredible contradictions. The developmental risk factors for NEC are known but offer few insights that help to identify the biologic underpinnings of the disease. Moreover, treatment options for NEC are controversial and lead to varying degrees of disagreement in their application. While NEC may be a disease of contradictions, these contradictions can also provide unique opportunities to improve our scientific understanding, enhance care delivery, and hopefully improve the outcomes for patients.

This Primer summarizes the epidemiology, pathophysiology, diagnosis and treatment of NEC. This Primer also discusses quality of life issues faced by NEC survivors and their caregivers and discusses future research avenues.

Epidemiology

Limitations of epidemiological data

Reliable information on NEC incidence are sparse, primarily because of inconsistent case-definitions, and a lack of whole population data, with a particular paucity of information from low-income countries. The lack of robust diagnostic criteria introduces ascertainment bias, and the predominance of hospital-based information rather than whole population-based data imposes reporting bias. These are important considerations as reliable incidence data are essential to design interventional studies, assess the effectiveness of quality improvement initiatives, and evaluate changes within and between centres. Accordingly, we recommend that investigators and authors provide full details of the diagnostic criteria or case-definition used, the denominator population, any exclusions, the extent of coverage of the entire at-risk population, and commit to sharing patient-level data for future meta-analyses.

Incidence

In one systematic review of NEC incidence in high-income countries (14 studies from 12 countries in the final analysis³), the incidence of NEC was 2% to 7% among babies born below 32 weeks’ gestation and from 5% to 22% in those with a birth weight below 1000 g (known as extremely low birth weight, ELBW). The studies included in this meta-analysis varied in NEC case definition, quality, and risk of bias, and few covered the entire at-risk population. The most commonly used definition in the original studies was Bell stage ≥2 (Table 1), but some studies used Bell stage 1–3, Centers for Disease Control and Prevention and International Classification for Diseases criteria, and varying combinations of clinical and radiological signs. Since this systematic review, a 2-year prospective whole population study was carried out of severe NEC (defined as disease confirmed at laparotomy or autopsy, or resulting in death), using the UK National Neonatal Research Database, a repository containing data from admissions to all neonatal units in England, Scotland and Wales⁴. This study found the incidence of severe NEC in England was 3% in babies born below 32 weeks’ gestation.

At the population level, NEC is a rare disease (defined by The European Union as affecting <5 in 10,000 of the general population⁵). A Swedish birth cohort study including data from the National Patient Register, Swedish Medical Birth Register and National Cause of Death Register and covering the period 1987 to 2009, found an incidence of 3.4 per 10,000 live births⁶. Similar to some other studies, incidence was higher in boys than in girls (incidence rate ratio 1.22, 95% confidence interval 1.06–1.40). This study also identified evidence of seasonality with a peak in incidence in November and a trough in May. The authors of this study speculated that seasonal variation may indicate the involvement of environmental factors in NEC, noting that such factors are unlikely to occur in the same patterns in different populations. The effect of such environmental factors could include viral transmission, which has been linked to NEC development, but is otherwise unknown^7,8.

Variations in clinical practice compound difficulties in identifying true differences in NEC incidence between centres. Thus, as the incidence of NEC varies markedly in relation to gestational age (Figure 1)⁴, centres with a more active approach to resuscitation and provision of life-sustaining support for extremely preterm infants (as defined by less than 28w gestation), may have a higher rate.

Figure 1. Incidence of NEC in England resulting in death and/or surgery in babies born before a gestational age of 32 weeks.

Two-year whole population surveillance study from birth to neonatal unit discharge, the bands represent 95% Confidence Intervals. Reproduced with permission from reference (4)

Age at onset and mortality

In studies reporting NEC confirmed by histopathology following autopsy or surgery, the median age at presentation was 19 days (inter-quartile range 9.0–29.8 days), at a post-menstrual age of 31.2 weeks (inter-quartile range 27.9–32.7 weeks). Boys were older at presentation than girls (median age 27 days (IQR 18–32) versus 10 days (IQR 6–27))^9–12 although the reasons for this are unknown.

One systematic review comprised national, regional, or multicentre cohort studies of outcomes in infants with NEC in high-income countries from 2010–2018². This review demonstrated an overall mortality of 23.5% in neonates with confirmed NEC (Bell stage 2a+) (95% CI 18.5%−28.8%), 34.5% (30.1%−39.2%) for neonates undergoing surgery, 40.5% (37.2%−43.8%) for ELBW infants, and 50.9% (38.1%−63.5%) for ELBW infants with surgical NEC. This study also demonstrated a NEC-related infant mortality rate of 12.5 per 100,000 live births, based on death certificate data, recognizing that one of the main limitations of the review was the lack of an agreed definition for diagnosing NEC, rising to 2800 per 100 000 live births for babies born between 22- and 29-weeks’ gestation. Overall, NEC (defined as Bell stage ≥2a) is responsible for between 10% and 21% of deaths in preterm babies.¹³

Risk factors

Growth restriction is the most consistently reported endogenous infant risk factor for NEC¹⁴. One analysis of risk factors for NEC¹⁵ found that antenatal glucocorticoids, antenatal progesterone, and postnatal erythropoietin, lactoferrin, fluid restriction, arginine, human milk, and probiotics may reduce risk, whereas lower oxygen saturation targets, sepsis and use of formula may increase risk of NEC. Of note, strong evidence of causality is lacking for most of these factors and there is disagreement about their importance. Moreover, although formula feeds have traditionally been linked to NEC, a 2 year, whole population surveillance study found that half of all cases of NEC resulting in death and/or surgery had received only human milk feeds prior to disease onset⁴.

Moreover, the most recent Cochrane review assessing NEC as a secondary outcome also found that antenatal corticosteroids, and in fact any antenatal steroids, reduce NEC risk (Relative Risk 0.50, 95% CI 0.32 to 0.78)¹⁶. Several studies have reported differences in NEC incidence and severity between racial groups. For example, studies from the United States have consistently identified a higher incidence in infants of Black-African ancestry^17,18. However, these associations are highly susceptible to confounding by gestational age, co-morbidities, clinical practices and quality of care, among other factors. Studies also remain inadequate to draw firm conclusions regarding genetic determinants of NEC susceptibility or disease severity.

Mechanisms/Pathophysiology

NEC is one of the most challenging diseases to treat, in part because of an incomplete understanding of its pathogenesis. Historically, NEC was thought to develop owing to an immature immune system in premature infants that is incapable of tolerating formula and lacks the capacity to clear potential pathogens. More recent research using animal models of NEC and transcriptomic analyses of human tissue has demonstrated a central role for TLR4 activation within the intestinal epithelium.

A historical perspective

Studies performed ~30 years ago described NEC as a condition that was caused by immaturity of phagocytes, which was supported by findings of decreased phagocytic function¹⁹, and reduced respiratory burst²⁰ in circulating immune cells from infants with NEC. These findings led to the hypothesis that as a direct consequence of prematurity, the neonate fails to clear circulating bacteria, rendering them susceptible to sepsis and multi system organ failure. Further early studies using tissues from patients with NEC suggested that incomplete development of the mesenteric vasculature due to prematurity promotes vasoconstriction, resulting in the patchy ischemia that characterizes NEC^21,22. Moreover, these studies also suggested that the pattern of circulating cytokines in the blood of infants with NEC (such as IL-1, IL-6 and TNF) mirrored those seen in other inflammatory states such as sepsis, suggesting that a central pro-inflammatory process drives NEC development²³.

Although these discoveries were pivotal to our early understanding of NEC pathogenesis, they failed to account for the fact that NEC begins with abdominal symptoms (bloody stools, feeding intolerance) which then progress to systemic disease (hypotension, organ failure), rather than the other way around²⁴. Moreover, although it is attractive to consider impaired bacterial clearance as central to the development of NEC, these studies also failed to account for the fact that NEC is generally seen in the absence of specific pathogens²⁵, and that infants with NEC do not typically reveal increased susceptibility to non-bacterial infections above that seen in other premature infants²⁶, suggesting that a general defect in immune activity is unlikely to account for NEC development.

These gaps in our understanding of NEC began to close in part through the development of animal models that mirror the clinical condition²⁷ (Box 1). Indeed, studies in mice, rats and piglets have revealed that NEC develops from an abnormal (dysbiotic) intestinal microbiome, characterized by reduced microbial diversity, an increased abundance of pathogenic Gram-negative bacteria (especially Proteobacteria and Gammaproteobacteria), and a decreased presence of beneficial commensals such as Bifidobacterium, Firmicutes, and strict anaerobes^28–30. The abnormal intestinal microbiome then interacts with an immature and hyperinflammatory intestinal mucosa, resulting in bacterial translocation, mesenteric ischemia and intestinal injury via a combination of cell death and barrier disruption, leading to the characteristic intestinal injury of NEC (Figure 2)³¹. The corresponding histology of the involved intestine reveals epithelial injury and necrosis (Figure 3). Evidence in support of the importance of impaired bacterial-host signaling in the pathogenesis of NEC is provided below.

Box 1. Animal models of NEC.

One neonatal rat model of NEC²⁴¹ can be created by a combination of formula gavage and hypoxia beginning in the early newborn period of Sprague-Dawley rats, along with an inocula of Klebsiella and lasting for six or seven days. Mice models of NEC have also been developed,^35,242 which use mice at approximately one week of life, and involve administration of formula, hypoxia and some combination of either bacteria, lipopolysaccharide or cold exposure. Early rat studies identified a critical role for nitric oxide in the pathogenesis of NEC²⁴³, whose release could induce apoptosis of the cells lining the neonatal intestine, specifically the enterocytes, resulting in barrier injury²⁴⁴. Other groups have developed piglet models of NEC, in which NEC is induced by administering combinations of maltodextrin-rich formula to premature piglets delivered by cesarean section at 92% gestation^245–248.

Figure 2. Pathogenesis of NEC.

In the healthy infant, luminal bacteria interact with low levels of toll like receptor (TLR4) on the surface epithelium in a homeostatic manner. By contrast, in the setting of prematurity and intermittent hypoxia, along with various genetic factors, TLR4 expression is elevated and prone to increased activation, especially by a dysbiotic microbiome containing microbes that are enriched in lipopolysaccharide (LPS). TLR4 activation then leads to epithelial apoptosis, reduced proliferation, and an influx of proinflammatory Th17 lymphocytes resulting in bacterial translocation resulting in activation of TLR4 on the endothelium. Bacteria then activate TLR4 on the mesenteric endothelium resulting in a loss of endothelial nitric oxide expression which leads to vasoconstriction and intestinal ischemia. The combination of these factors results in intestinal damage, a systemic inflammatory response, andthe development of NEC. These processes are regulated by genetic and epigenetic factors.

Figure 3. Histology of NEC.

H&E stained ileum from an infant with either control bowel, obtained at the time of stoma closure (left) or during an acute episode of NEC (right), revealing epithelial disruption and loss of villi in the intestine of an infant with NEC. scale bar 100μm

Impaired bacterial-host signaling

Evidence for the role of bacteria in the early signaling events leading to NEC include the finding that NEC typically develops after bacterial colonization of the gastrointestinal tract has occurred³², and that patients with NEC often have enteric organisms (including Gram-negative bacilli, particularly members of the Enterobacteriaceae family such as Escherichia coli and Klebsiella pneumoniae, as well as Clostridium species³³) and lipopolysaccharide (LPS) in the circulation³⁴. The LPS receptor TLR4 mediates the interaction between bacteria and the premature host^35,36 and has a pivotal role in the pathogenesis of NEC. TLR4 expression in the intestinal epithelium is higher throughout the premature period in humans and animal models compared with the full term intestinal tract³⁷, revealing the non-immune role that TLR4 plays in normal gut development through modulation of Wnt-Notch signaling in intestinal stem cells³⁸. These findings predominantly reflect TLR4 expression in enterocytes, although this has not been carefully evaluated at the single cell level. Further studies have shown that TLR4 activation either by LPS or an endogenous TLR4 ligand on the lining of the intestine led to barrier injury³⁸, reduced mucosal repair³⁹, and epithelial death. The pathways involved include a combination of TLR4-dependent ER-stress induced apoptosis in intestinal stem cells via protein kinase-related PKR-like ER kinase⁴⁰, TLR4-dependent autophagy of enterocytes via ATG7⁴¹, TLR4-mediated necroptosis of enterocytes via RIPK⁴², and TLR4-mediated impaired enterocyte migration and restitution via the small GTPase Rho⁴³ and alpha 3 and beta 1 integrins⁴⁴; collectively, these processes permit the translocation of bacteria across the intestinal tract⁴². Bacteria translocation across the intestinal tract leads to TLR4 activation on endothelial cells, resulting in a loss of eNOS expression, mesenteric vasoconstriction and intestinal ischemia⁴⁵.

Evidence in support of a role for TLR4 activation in the pathogenesis of NEC in humans include increased TLR4 expression in the intestine of infants with NEC compared to those without NEC⁴⁶, although the precise cell types involved are unknown, and that breast milk is richly endowed with natural TLR4 antagonists⁴⁷. Moreover, mutations in genes encoding components of the Single Ig IL-1-related receptor (SIGGIR) signaling pathway (a negative regulator of the TLR4 signaling pathway) that increase TLR4 activation^48–50 are seen in patients with NEC at higher frequency than those without NEC⁵¹. The importance of bacterial signaling in NEC is highlighted by the fact that prior to NEC development in humans, the GI tract becomes enriched in LPS-expressing bacteria which can then trigger TLR4 activation²⁸.

A unifying hypothesis of NEC pathophysiology that integrates these findings is called the “cross switching hypothesis”. This hypothesis states that elevated expression of TLR4 in the premature infant induces a pro-inflammatory role upon colonization of the gastrointestinal tract after birth, leading to the development of NEC⁵². This concept is helpful in explaining NEC development³¹, and accounts for many of the features seen in this disease⁵³.

Additional biologic factors

The “TLR4 cross switching hypothesis” has emerged as a central, unifying paradigm to explain NEC development, but this hypothesis also leaves plenty of space for the investigation of parallel processes that contribute to the development of this disease. Interestingly, many of these additional processes are associated with exaggerated TLR4 signaling. For example, NEC is associated with a loss of goblet cells reported both in infants and mouse models (epithelial cells that secrete mucus to protect the intestine from mechanical damage⁵⁴). Importantly, TLR4 activation can inhibit goblet cell differentiation by activating Notch, therefore, hindering or preventing repair of the damaged epithelium. Accordingly, treatment of mice with the Notch inhibitor Dibenzazepine postnatally restored goblet cells and prevented NEC³⁸.

Other studies have focused on the potential role of Paneth cell dysfunction and antimicrobial insufficiency in the pathogenesis of NEC. Paneth cells release defensins and anti-microbial cytokines (such as IL-22 and IL-17a) in response to exposure to bacteria or bacterial antigens^55,56. These cytokines can kill or inhibit the growth of microorganisms, including bacteria⁵⁷. Fewer Paneth cells are found in the premature intestine compared with adults⁵⁸, resulting in an insufficiency of antimicrobial peptides, making the intestine more vulnerable to invading pathogens⁵⁹.

Other studies have focused on the involvement in impaired angiogenesis in NEC. Neonatal mice that lack vascular endothelial growth factor 2 (VEGFR2) signaling in the intestinal endothelium show abnormal development of the intestinal microvasculature, which creates a hypoxic intestinal environment and a predisposition to NEC. In these mice, pharmacologic inhibition of VEGF/VEGFR2 signaling pathways resulted in increased NEC severity and abnormal vascular development, indicating that VEGF/VEGFR2 signaling is required for both vascular development and NEC protection partially rescues vascular development and reduces NEC susceptibility⁶⁰. Other studies have examined the role of the cytoskeleton in NEC, and showed that NEC is associated with activation of RhoA-associated kinase 1 which weakens tight and adherens junctions leading to increased intestinal permeability. What causes the activation of RhoA-associated kinase 1 is unknown. Pharmacologic inhibition of ROCK1 improves intestinal permeability and decreases the severity of NEC in mice⁶¹.

An interesting line of investigation has opened up regarding the role of volatile organic compounds (VOCs, measured from breadth, stool or other biological samples) as non-invasive biomarkers to detect NEC, and which could discriminate between NEC and non-NEC stool samples^62,63. Other studies have shown that mouse and human NEC is associated with coordinated epigenetic remodeling especially across epithelial, immune and endothelial compartments, thus revealing novel mechanistic pathways while also potentially revealing novel biomarkers of disease^64–66. While these factors do not appear obviously interrelated, they serve together to illustrate the complex factors that together lead to the development of NEC.

To further refine our understanding of the factors leading to NEC, one study described a single-cell atlas of the human neonatal small intestine in patients with NEC, presenting a comprehensive view of the cellular changes and aberrant interactions that underlie the pathogenesis of this disease⁶⁷. This study demonstrated an abundance of proinflammatory macrophages, fibroblasts, endothelial cells, and clonally expanded T cells in the small intestine, along with a reduction in villus tip epithelial cells and upregulation of proinflammatory genes in remaining epithelial cells⁶⁷. Another study. performed single-cell RNA sequencing on the lamina propria infiltrates of NEC and non-NEC neonates with intestinal perforation to analyze the intestinal immune cell profile. This study revealed the enrichment of MTOR, TNF-α, and MYC signaling pathways in the T cells of patients with NEC, indicating upregulated immune responses associated with inflammation and cell proliferation⁶⁸. In this regard, some studies have suggested that inflammation in the placenta could also contribute to NEC development, as one matched-case control study revealed acute chorioamnionitis in 35% of patients with surgical NEC versus 15% of patients with medical NEC⁶⁹. This finding is supportive of earlier studies that indicate that chorioamnionitis is an independent predictor of NEC⁷⁰. However, a more recent study did not find a link between chorioamnionitis and NEC⁷¹, revealing this to be an area of ongoing controversy.

NEC epigenomics.

Epigenomics refers broadly to the epigenetic modifications that occurs across the entire genome and includes DNA methylation and histone modifications⁷². Epigenomic changes may be particularly relevant to the pathogenesis of NEC, as environmental factors that are involved or associated with NEC – such as the microbial flora⁷², probiotics⁷³, and formula⁷⁴ – all affect the epigenome. In one study, an epigenetic based array of the resected intestine from 46 infants with NEC revealed significant differences in methylation in survivors and non survivors of surgical NEC, with a particular effect on genes required for gut development and homeostasis⁷⁵. This finding is consistent with earlier studies that have compared NEC with non-NEC specimens⁷⁶. Intriguingly, intestinal cell methylation signatures can be detected in blood samples from neonates with NEC⁷⁷, suggesting that epigenetic markers may serve as novel biomarkers for this disease. Such studies potentially allow for the identification of epigenetically regulated pathways that may have a role in NEC development. In support of this possibility, a comprehensive epigenomic atlas of epithelial injury in a mouse model of NEC revealed epigenetic changes in the epithelial cells, and using a knock-down approach, showed that the epigenetic regulator Ezh2 is a critical regulator in maintaining epithelial stem cell homeostasis in this disease, providing a potential link between epigenomic signaling and NEC pathogenesis⁷⁸.

Mechanisms of systemic progression of NEC

The mechanisms by which NEC progresses to systemic disease are still incompletely understood, but are likely to be related to the translocation of enteric bacteria from the intestinal lumen across the damaged intestinal barrier and into the circulation⁷⁹, leading to activation of immune receptors on circulating leukocytes, and leading to sepsis⁸⁰. NEC can also lead to neurodevelopmental impairment in some individuals^81,82 through at least two well-defined pathways. In the first pathway, TLR4-dependent enterocyte injury in NEC leads to the release of pro-inflammatory molecules (including HMGB1) from the gut, which then travel to the brain where they activate microglia in a TLR4-dependent manner, leading to a loss of oligodendrocytes, a reduction of white matter, and neurodevelopmental impairment⁸³. These findings have also been observed in a mouse model of NEC⁸⁴. In the second pathway, TLR4 signaling in the intestinal epithelium lead to the release of a subpopulation of T lymphocytes that travel to the brain where they release interferons, which cause inflammation in peri-ventricular regions and brain injury that can be reversed using specific anti-lymphocyte monoclonal antibody-based strategies⁸⁵.Reduced cerebrovascular autoregulation is an additional mechanism that contributes to neurodevelopmental impairment in infants with NEC⁸⁶. The mechanisms involved likely include the exposure of the developing brain to both hypoperfusion injury that can lead to hypoxia, and hyperperfusion injury that can cause vascular injury and hemorrhage, particularly in the fragile germinal matrix vessels of extremely preterm infants.

Diagnosis and Prevention

Diagnosis

NEC can present insidiously or devastatingly acutely. The first signs of NEC are often feeding intolerance (spitting up non-digested formula or breastmilk) and abdominal distention, but abdominal symptoms are non-specific and vary with gestational age. Vomiting (which can be bilious) and diarrhea (which can contain blood) can also occur. Clinical signs such as abdominal distension, discoloration of the abdominal wall (reddish in those with peritonitis, bluish in those with perforation), and tenderness are also non-specific. Laboratory findings include metabolic acidosis, thrombocytopenia, increased c-reactive protein (CRP), and elevated (or reduced) leukocyte counts. However, no sensitive and specific routine laboratory tests are available for severe NEC (defined as the need for surgery).

The gold standard for diagnosis of NEC is abdominal radiography, which might reveal pneumatosis intestinalis (air bubbles in the intestinal wall), portal gas, free intra-abdominal air, and fixed or dilated bowel loops, and some combination of these features is generally considered diagnostic for NEC in the appropriate clinical setting (i.e. the premature infant who has received breast milk or formula feeds) (Figures 4–6). Pneumatosis intestinalis itself is considered the hallmark diagnostic feature of NEC, even when no other feature is present. Other signs may exist in other neonatal diseases, most notably spontaneous intestinal perforation (see below). These radiological signs are age-dependent; for instance, the likelihood of pneumatosis is approximately 25% in babies born at 24 weeks gestation but increases to 70% in infants born at 28 weeks⁸⁷. Due to variable sensitivity and specificity of abdominal radiography, along with poor interobserver agreement for the radiological signs of NEC, alternative imaging modalities such as abdominal ultrasound (AUS) have been explored. AUS can be used to identify pneumatosis, bowel necrosis and perforation with high sensitivity (generally >80%) but with low specificity (generally <70%), but large-scale studies confirming its reliability are lacking, so AUS is considered mainly a useful adjunct in early disease stages or when clinical and radiological findings are inconclusive.^88,89

Figure 4. Infant with Bell stage II NEC.

Representative posterior-anterior radiographic image of a premature infant with pneumatosis intestinalis of the right lower quadrant (arrows).

Figure 6. Portal venous gas in a premature infant with NEC.

Representative posterior-anterior (A) radiographic image of a premature infant with portal venous gas (arrows).

Diagnosing NEC is challenging in part because it can occur along a wide spectrum of severity. Common diagnostic criteria do not distinguish NEC from conditions like food protein-induced enterocolitis or intestinal diseases due to low flow states (e.g., congenital heart disease), complicating the creation of a unifying definition. Probably the most important condition to be distinguished from NEC, both clinically and in research, is spontaneous intestinal perforation (SIP) as these conditions have different treatment approaches. SIP typically occurs in the terminal ileum of a very low birth weight (<1,500g, VLBW) or an ELBW infant within a week after birth, generally prior to the introduction of enteral feeding. SIP is characterized by the presence of free air on abdominal X-ray, in the absence of pneumatosis.⁹⁰ In SIP, the perforation is small perforation (only a few mm wide) within otherwise healthy bowel, whereas intestinal disease is often widespread in NEC. Mortality rates are significantly higher in NEC when compared to SIP patients (~40% versus 20%)⁹¹.

Classification and Staging

The most commonly used classification for NEC is the (modified) Bell criteria, which integrates clinical and radiological manifestations (table 1), but was never intended as a diagnostic tool.⁹² Despite widespread use, the Bell criteria have been criticized for their lack of specificity, especially in the early stages of the disease (i.e. Stages 1–2), which can lead to substantial over- or underestimation.^93,94 Moreover, the criteria do not account for baseline risk factors, particularly gestational age, which, as previously mentioned, is a major risk factor for NEC. Additionally, Bell classification rely on abdominal X-ray for the detection of pneumatosis and includes subjective criteria such as abdominal distention. Therefore, clinicians should be aware of the limitation in using the Bell classification for diagnosing or staging of NEC, particularly the inability of these criteria to differentiate NEC from other neonatal intestinal diseases, such as SIP, gastric perforation, midgut volvulus, viral colitis or food-borne enterocolitis.

Other more recent classification systems, including the Vermont Oxford Network (VON) definition, the Centers for Disease Control and Prevention (CDC) definition, and the UK Neonatal Collaborative NEC Study group (UKNC-NEC) gestational age-specific case definition, also rely on a combination of clinical signs, symptoms, laboratory values, and imaging (Table 1).⁹⁵ Moreover, one study described five clinical presentations of NEC to account for its heterogeneous presentations and pathophysiological processes.⁹⁶ (Table 1). However, these newer classifications have not gained widespread traction due to their recent development and incomplete evidence for their clinical application. This also holds true for scoring systems aimed at predicting complicated NEC (NEC requiring surgery or leading to death).^97–99 During the 2018 meeting of the Special Interest Group NEC (SIGNEC), 57% of experts still considered the modified Bell stage as the gold standard.¹⁰⁰ Interestingly, machine learning models have shown that newer, less complex definitions might be more clinically useful than older, feature-rich ones¹⁰¹. For instance, in one study, machine learning algorithms predicted NEC with an F1 of 0.82 using only eight parameters (consisting of vital signs) obtained within the five days of life¹⁰².

Prevention

Screening for NEC on an individual level is not possible given the absence of reliable screening methods. Most available preventive measures focus on feeding, the gut microbiome and intestinal oxygenation, and many of them are readily available throughout the world.¹⁰³

Feeding regimens

The most important postnatal prevention strategy for NEC is the use of mothers’ own milk (MOM). The use of MOM can improve neurodevelopment and reduce NEC incidence¹⁰⁴. MOM contains numerous components such as secretory immunoglobulin A (IgA) and oligosaccharides that regulate gut barrier integrity and microbial colonization, often targeting the LPS/TLR4 cascade.^47,105,106 Fresh MOM is superior to pasteurized MOM or donor human milk (DHM) due to its higher content of macronutrients, immunoactive, and trophic factors¹⁰⁷. When mothers own milk is unavailable or insufficient, pasteurized donor human milk (DHM) or preterm formula are options for feeding preterm infants^108–110. Concern that preterm formula may increase the risk of NEC is paralleled by concern that use of pasteruised donor milk may increase the risk of neurodevelopmental impairment. Definitive evidence from large, randomized trials is awaited.

Within human milk-based feeding strategies, a 2025 systematic review and meta-analysis reported that, among very low birth weight infants receiving an exclusive human milk diet, human milk-derived fortifiers were associated with lower odds of both medical and surgical NEC than bovine milk-derived fortifiers (OR for Bell Stage ≥2 NEC: 0.65, 95% CI: 0.44–0.97; OR for surgical NEC: 0.51, 95% CI: 0.26–0.98)¹¹¹. However, this review included 15 observational studies employing diverse methods, and only 5 small RCT, hence definitive evidence remains lacking. An analysis of 59 studies showed evidence to support the recommendation for MOM or DHM and a human-milk based fortifier¹¹²

Current thinking on the role of DHM has thus evolved, and the evidence for cow-milk derived fortifiers and formula influencing the risk of NEC is unclear. A Cochrane review and meta-analysis comparing fortified and unfortified human milk found no evidence of an effect on NEC risk (13 RCT; 1110 infants; relative risk 1.37 [95% CI .72, 2.63])¹¹³. Moreover, another Cochrane review (including 2261 infants from 11 small trials) comparing formula and pasteurized DHM as either sole or supplemental diet¹⁰⁴ reported that the risk of NEC is halved through use of pasteurised DHM instead of formula. However, this Cochrane review cannot be considered definitive evidence as the criteria used to diagnose NEC in the included studies was very variable and included medical NEC, no data were presented on surgical NEC, almost half of the trials were conducted over 40 years ago when the patient population and clinical practice differed substantially from today, and 8 of the trials were judged to have high or uncertain risk of bias. Additionally, and of most importance, the review did not show significant differences in important functional outcomes that would be corroborate benefit from pasteurised DHM (mortality, invasive infection and neurodevelopment). The Canadian Domino Trial randomised very preterm infants to fortified pasteurised DHM or preterm formula to make up any shortfall in own mother’s milk¹¹⁴. This study identified no differences in neuro-impairment using the Bayley III cognitive composite score when infants were administered these feeding regimens for 90 days or to discharge when mothers milk was unavailable (adjusted scores, 92.9 in donor milk group vs 94.5 in formula group; fully adjusted mean difference, −2.0 [95% CI, −5.8 to 1.8). However, the babies in the pasteurised human donor milk group had substantially worse individual and cognitive composite scores indicative of neuro-impairment (<85) (donor: 27.2%; formula 16.2%; adjusted risk difference 10.6%; 95% CI 1.5% to 19.6%). No randomised controlled trial to-date has shown neurodevelopmental benefit from use of pasteurised DHM.

Absence of feeding for even a few days leads to intestinal mucosal atrophy, decreased mesenteric blood flow, and a pro-NEC microbiome in animal models and human infants¹¹⁵. Accordingly, early initiation of small volume enteral feeds is recommended within hours of birth, advancing as tolerated.^110,116 Standardized feeding protocols and minimising multiple periods of stopping and restarting feeds, are beneficial, cheap, and easy to implement. Individual centers tend to have their own feeding protocols that provide criteria for progressing and stopping feeds, and these protocols also include details regarding the amount and frequency of milk or formula given. These are often administered in consultation with nutritionists, pharmacists and feeding specialists within the neonatal intensive care unit.

There is also a clear need for high-quality randomised trials to assess the impact on NEC of pasteurised DHM, standard formula, hydrolysed formula, and human-milk and cow-milk derived fortifiers..^117,118 A current large, 2-randomisation adaptive trial is evaluating the impact of formula versus DHM as supplementary feeds, and routine cow-milk based fortification, on survival without NEC in babies <29w gestation (ISRCTN10443084).

Probiotic use

Probiotics are widely used for NEC prevention around the world, although use is not without controversy, due in part to concerns regarding the risk of sepsis¹¹⁹. Large meta-analyses suggest benefits of multi-strain over single-strain preparations. Indeed, in a network meta-analysis of 106 trials (25,840 preterm infants), the incidence of severe NEC (defined as Bell stage 2 or 3) was 3.7% lower with multiple-strain probiotics compared to placebo (risk ratio [RR] 0.38, 95% CI 0.30–0.50), with the absolute risk difference ranging from −4.1% to −2.9%¹²⁰. Importantly, efficacy in ELBW infants remains unproven and the optimal composition of the strains remains unclear.^121–123 Probiotics are speculated to reduce risk of NEC via restoring flora and through the release of their CpG-DNA-enriched DNA which inhibits TLR4 signaling by activation of TLR9.¹²⁴ In extremely rare cases, probiotics might have been associated with sepsis¹²⁵ (<0.04%, with 8 cases reported in over 20,000 exposed infants¹¹⁹).

Other approaches

Improving intestinal perfusion/oxygenation might reduce NEC incidence and severity. This can be achieved using appropriate hemoglobin and oxygen saturation thresholds. Moreover, delayed umbilical cord clamping might reduce incidence of NEC by preventing hypovolemia, avoiding fluctuation in cardiac output and giving extra time for lung aeration¹²⁶ One randomized trial is being carried out to investigate the benefits of physiological-based cord clamping versus time-based cord clamping in very preterm infants.¹²⁷

In addition, approaches to inhibit the TLR4 pathway are being evaluated, e.g. via small molecules such as C34, J11 or human milk fortification with human milk oligosaccharides.^128–130 However, until tailored evidence-based preventive strategies become available, NEC prevention bundles consisting of feeding practices (including active promotion of MOM), probiotics, antibiotics, and vasodilatory/oxygenation strategies should be constructed and validated on an international level.

Management

For decades, little has changed in the medical, or conservative management of established NEC. Medical management usually involves a range of treatments from mainly supportive care to few medical interventions, tailored according to the severity and progression of the disease.

Supportive Care

Nil Per Os.

All infants with confirmed NEC are managed initially with bowel rest and cessation of enteral feeds (nil per os, NPO), a practice intended to reduce intestinal stimulation and mitigate ongoing injury. In infants with clinical features suggestive of NEC who do not meet established diagnostic criteria (for example, those presenting with isolated feeding intolerance and abdominal distension in the absence of radiographic or systemic findings), bowel rest combined with empirical antibiotic therapy is frequently instituted for 12–48 hours while the diagnosis is clarified. Although this precautionary approach is widely adopted, criteria for initiating and discontinuing such management vary across centres, reflecting the absence of robust evidence to guide care in this early or equivocal phase. In clinical practice, this period is often described as “ruling out NEC” or placing the infant on “NEC watch”.

In these instances, or in infants with definitive NEC, enteral feeding is withheld to prevent further injury to the intestines. Total Parenteral Nutrition (TPN) and fluids are provided intravenously to ensure the infant receives adequate calories and nutrients and maintain hydration and electrolyte balance. Prolonged use of central venous catheters for TPN can lead to bloodstream infections, and is associated with cholestasis and liver disease. Therefore, re-initiating feeding after NEC should be carried out as soon as safely possible, but prospective trials are lacking to support the optimal timing.¹³¹ Some evidence suggests the safety of early refeeding (4 vs. 10 days), although further studies are needed¹³¹

Other supportive measures.

Vital signs and biochemical status are continuously monitored and corrected when deteriorating. Circulatory failure is treated according to local guidelines, and entails administration of fluid boluses and cardiovascular medication although there is little supporting evidence for the most effective medication at the various stages of NEC. Dobutamine improved intestinal perfusion in a few animal studies, and in small observational trials in circulatory compromised preterm infants and adults^133–135 but no difference in NEC incidence was demonstrated in a study comparing dobutamine to dopamine for low flow states in very preterm infants, (1% versus 3%).¹³⁶ Systemic symptoms include worsening of respiratory status, which often requires respiratory support. For optimal respiratory support and to prevent further accumulation of intestinal air, invasive ventilation may be indicated rather than continuous positive airway pressure (CPAP). In addition, broad-spectrum antibiotics are initiated to combat bacterial infection. Regimens typically include ampicillin and gentamicin, with metronidazole or clindamycin added if the infant fails to respond or in those who require surgery. However, evidence is insufficient to recommend specific antibiotics, administration routes, or treatment durations for infants with NEC.¹³⁷

If NEC Bell stage >1 is confirmed, repetitive abdominal radiological examination every 8–12h may assist in evaluating progress of the disease, and aid in the early detection of pneumoperitoneum or fixed bowel loop. Imaging protocols vary by institution¹³⁸. According to expert guidance, infants diagnosed with NEC stage ≥ II should also undergo routine pain evaluations using an appropriate neonatal pain scale, with analgesic treatment administered preemptively¹³⁹. Paracetamol is commonly used as first-line therapy; however, observational data from European NICUs indicate that opioids such as fentanyl and morphine are also frequently administered in clinical practice, particularly in more advanced disease stages²⁵⁴. Although these agents are widely used in preterm infants, robust evidence specific to NEC remains limited, and practice varies across centres.

Surgical management

Despite recent advancements in medical therapy for NEC, surgical intervention is still the only way to save critically ill patients with intestinal perforation or who have progressed to a state beyond which medical intervention could add further benefit. Among all infants with NEC, 30–35% undergo surgical intervention¹⁴⁰.

Thresholds for surgery

Intestinal perforation is the only absolute indication for surgery in NEC, although clinical deterioration despite full medical management, or failure to improve in the setting of maximal medical management, represent additional indications for surgery. Patients may have also reached the threshold for surgery if they have portal venous gas with abdominal distention and tenderness, persistent or worsening laboratory values including thrombocytopenia or acidosis, and physical findings of abdominal wall erythema.

Scoring systems have been developed to provide an objective approach to the requirement for surgery in infants with NEC. Abdominal wall erythema, acidosis and hypotension are important predictors of the need for surgery (when pneumoperitoneum is absent), and the presence of three or more clinical, radiographic or laboratory indicators strongly suggest the need for surgery¹⁴¹, and these findings have been confirmed in a multi-center case control study performed across nine neonatal intensive care units¹⁴². Other studies have developed prediction models to identify neonates with NEC who require surgery. For instance, a model incorporating CRP and lactate levels, presence of portal venous gas, reduced intestinal motility as determined by ultrasonography, white blood cell count and absolute lymphocyte count, and achieved excellent discrimination (AUC 0.946) with 91% sensitivity and 89% specificity¹⁴³. Additionally, the HASOFA score (hyperglycemia, hyperkalemia, hypotension requiring inotropes, acidemia, nSOFA score) achieved AUC 0.909 to predict the need for surgery, which improved when combined with abdominal wall erythema or portal venous gas¹⁴⁴. Importantly, most of these prediction models have not undergone external validation, and internal validation alone is generally considered insufficient for broad clinical acceptance¹⁴⁵.

Mortality after surgery reaches 30–40% and increases to 50.9% in ELBW infants^146,147. Since surgical intervention is an extremely invasive treatment for these patients, the timing of surgery is critical. Surgical decision making may be helped by ultrasound findings, especially if there is severe ascites or portal venous air in the setting of abdominal tenderness. In a large meta-analysis focusing on 11 articles comprising 748 infants, the presence of focal fluid collections, complex ascites, absent peristalsis, pneumoperitoneum, bowel wall echogenicity, bowel wall thinning, absent perfusion, bowel wall thickening and dilated bowel were associated with surgery or death in infants with NEC, whereas portal venous gas, pneumatosis intestinalis, increased bowel prefusion and simple ascites were not associated with surgery or death¹⁴⁸. This result likely reflects the variability in the sensitivity and specificity of abdominal ultrasonography in premature infants (i.e. quite sensitive and specific for complex ascites, but less so for pneumatosis and portal venous gas). Abdominal ultrasonography, X-ray and the clinical picture should be used in the surgical decision making process in infants with NEC.

Surgical approaches

A generally accepted surgical protocol for NEC patients has not yet been defined, owing in part to the variation in severity of NEC. The variability in surgical approach is compounded further by surgeon preference, which influences the timing and approach to surgery. The relatively standard surgical management for more hemodynamically stable patients is laparotomy with resection of grossly necrotic intestine and ostomy or primary anastomosis^149,150. About 80% of patients undergo enterostomy (in which the intestine is brought out through the abdominal surface) with mortality around 25%². Patients experience frequent complications postoperatively, including superficial or deep wound disruption, and metabolic impairments such as acidosis and hyponatremia, mostly due to high enterostomal output^149,151,152.

Primary anastomosis versus stoma creation

Primary anastomosis is carried out in about 20% of patients who undergo laparotomy¹⁵². One multicentre randomized controlled trial (known as the STAT trial) aimed to assess the value of primary anastomosis for surgical NEC compared with stoma formation. This trial recruited 80 patients and demonstrated that primary anastomosis was associated with earlier completion of parenteral nutrition than stoma formation, with no difference in mortality or rate of complications requiring further surgery between groups, although patients who had a stoma had more overall intestinal complications. The study is remarkable given the challenges in performing randomized controlled trials in infants with NEC. However, the study had several important methodological challenges that have prevented widespread adoption, including a very narrow patient population, lack of information on patients assessed but not enrolled in the study, a primary outcome (time on parenteral nutrition) that does not have a standard protocol, a long recruitment window (nine years), heavy concentration of patients in a few centers, and a lack of collection of growth data (which is important as parenteral nutrition duration is the primary positive endpoint). Taken together, primary anastomosis may have a role in selected cases.

Damage control approaches

In severe cases such as diffuse intestinal involvement and patchy perforated/necrotizing areas in the intestine, resection of all of the affected bowel would lead to the development of short bowel syndrome due to insufficient remaining bowel to allow for adequate nutritional absorption. To prevent this complication, other surgical approaches have historically been used, including the “Clip and drop” approach, whereby clearly necrotic bowel is resected while questionable bowel is left inside the abdomen after clipping either end to prevent spillage of stool inside the abdomen. Alternatively, the placement of a temporary silo could be used to allow for ongoing surveillance of the bowel, and thus potentially avoid vast area of intestinal resection¹⁵³. These approaches also include the so-called “damage control laparotomy”, in which the primary goals are resection of clearly necrotic intestine and stabilization with an open abdomen, followed by a delayed second look at which time definitive surgical treatment can be deployed¹⁵⁴. The use of indocyanine green is emerging as a novel technique to identify well-perfused areas of intestine from ischemic /necrotic areas during laparotomy, although it has not been extensively studied^155,156. The development of acute kidney injury (AKI) may significantly worsen the postoperative outcome while also worsening brain injury outcomes¹⁵⁷.

Peritoneal drainage (PD) can be used in hemodynamically unstable infants who are too vulnerable to tolerate laparotomy. PD is less invasive than laparotomy as it requires only small abdominal incisions for drain placement and local anaesthesia minimizing hemodynamical change¹⁵⁸. PD should be considered predominantly as a resuscitative measure, as this technique does not address the need to resect the necrotic intestine, and it is also associated with significant long term neurological impairment owing to the ongoing inflammatory response. For these reasons, initial PD is associated with subsequent laparotomy with resection of necrotic areas in up to 75% of cases^151,159,160.

Some trials – such as the NECSTEPS and NET trials – compared PD with laparotomy, and demonstrated that neither approach is inferior to the other^158,161 A distinctive characteristic of the NECSTEP trial was that subsequent laparotomy after PD was discouraged. By contrast, the NET trial consisted of ELBW infants and encouraged laparotomy 12 hour after PD if a patient’s condition worsened. The NET trial found a tendency of better survival rate with initial laparotomy compared to PD, although neither trials reached statistical significance in mortality or length of stay between surgical treatments^158,161. Importantly, the NECSTEP trial included patients who likely had SIP and not NEC as many patients did not have pneumatosis seen on x-ray, which reduced the generalizability of this study to NEC patients¹⁵⁸. Moreover, many patients who entered the NECSTEP trial were never randomized to either treatment, suggesting that the study participants were a more selected subset. By contrast, the NEST demonstrated that among infants preoperatively diagnosed with NEC, those who underwent initial laparotomy had a lower rate of death or neurodevelopmental impairment compared to those who underwent initial peritoneal drainage (69% and 85%, respectively) with frequentist adjusted relative risk of 0.81 (95% CI: 0.64 –1.04)^159,162. This finding supports our biological understanding of the disease, wherein intestinal necrosis drives the overwhelming septic response that patients experience, such that patients treated with initial laparotomy have improved outcomes. Considering that neurodevelopmental delay affects more patients treated surgically compared to those treated by medical and supportive management, this finding is meaningful¹⁶³. There is likely still a role for initial PD in patients who are extremely premature, weigh <700g, hemodynamically unstable, or in those with suspected SIP^159,164,165.

Recurrent NEC

Recurrent NEC is rare. In one meta-analysis (58 studies including 4260 patients who survived surgery for NEC), recurrent NEC occurred in approximately 8% of infants¹⁶⁶. Risk factors for recurrent NEC are similar to risk factors for NEC in general, and include prematurity, formula feeds (as opposed to breast milk) and symbiosis, yet are less clearly defined. The risk of recurrent NEC is extremely low once the infant reaches 40 weeks gestational age, therefore, a diet that is rich in human breast milk is preferred if possible until this timepoint, although there is no clear evidence regarding the role of formula versus human milk feeds to reduce recurrent NEC.

Quality of Life

The most frequent long-term complications of NEC are gastrointestinal problems of which the most serious are short bowel syndrome, and intestinal and liver failure, and effects on growth, neurodevelopment, and quality of life. Indeed, a questionnaire-based study involving NEC survivors and parents of children with NEC (including those with medical NEC) found that most individuals reported long-term complications of which digestive problems were the most frequent¹⁶⁷. The study was limited by being voluntary and lacking a comparator group, but an important finding was that parents and survivors would welcome receiving more information about NEC and its long-term outcomes at an early stage. Of note, most respondents did not report adverse effects on later quality of life. This finding similar to results from a small single centre study from Hong Kong that found comparable quality of life among survivors of NEC surgery without major medical illnesses and the normal population¹⁶⁸.

Gastrointestinal complications

Stricture formation (scarring resulting in intestinal narrowing) can occur in 10–30% of patients with NEC, including infants treated with medical and supportive management or surgery¹⁶⁹, and can occur 2–4 weeks after the initial diagnosis. Notably, CRP level <10mg/dL is a negative predictor of stricture development in NEC patients¹⁷⁰. The presence of a stricture requires surgery, either with a primary anastomosis or in a staged fashion with the creation of an enterostomy. Stricture are most commonly found in the colon (82%) or the ileum (18%), although some patients have strictures that extends from the terminal ileum into the ascending colon¹⁷¹. Strictures are suspected in children who had NEC and later develop evidence of intestinal obstruction, and can be confirmed with a distal contrast x-ray.

The American Pediatric Surgical Association found that 57% of over 300 survey respondents would recommend comfort care for infants with NEC resulting in a residual small bowel length of <30 cm¹⁷². However, a study from Boston Children’s Hospital showed that long-term survival can be excellent following severe surgical NEC and that 71% do not have severe neurodevelopmental disability¹⁷³. This indicates that a decision to recommend comfort care is best reached after consideration of multiple factors and not residual bowel length alone. A study of children with short bowel syndrome from the same group also found that infants with prior NEC weaned from parenteral nutrition earlier than those with other causes of short bowel syndrome¹⁷⁴. Overall, gastrointestinal complications occur in up to 35% of NEC survivors, with intestinal failure being the most common, especially following extensive bowel resection¹⁶⁶. Symptoms of short bowel syndrome include severe diarrhea and loss of appetite, leading to poor weight gain and short stature.

Brain injury and neurodevelopmental impairment

A systemic inflammatory response frequently accompanies acute NEC, with hypotension and shock common in those with systemic sepsis. The major risk factors for neurodevelopmental impairment for infants with NEC are degree of prematurity, and the severity of hypoxic, ischaemic, and inflammatory injury predominantly affecting cerebral white matter⁸¹. The extent to which subsequent nutritional deficiencies contribute to later neurodevelopmental and cognitive impairment is unclear. SIP is the second most common preterm acute intestinal pathology after NEC¹⁷⁵. SIP is characterised by localised disease and little systemic effect. Children with NEC have a greater prevalence of cerebral palsy and psychomotor and cognitive delay compared with those with SPI. Overall, around 38% of children surviving surgical NEC have evidence of major neurodevelopmental impairment at age two-years¹⁷⁶. One occurrence is periventricular leucomalacia, which is usually identified as the infant is recovering from the acute phase of the illness, and can be identified via the standard head ultrasound scans given to premature babies. Some recommend all infants with NEC undergo long-term neurodevelopmental assessment due to their increased risk of neurodevelopmental impairment and brain injury². This includes serial developmental assessments, neurologic examinations, and, when indicated, neuroimaging such as MRI¹⁷⁷. Brain MRI findings in infants with NEC include white matter abnormalities in approximately half of all cases by approximately 40 weeks post menstrual age¹⁷⁸, of which periventricular leukomalacia and diffuse white matter injury are the most common. Other MRI findings include gray matter abnormalities and cerebellar injury, with the risk of MRI findings increasing with the severity of the intestinal injury¹⁷⁸. Cerebellar injury occurs in a third of patients and include included cerebellar hemorrhage, siderosis and/or cerebellar volume loss, and a regression analysis showed that positive blood cultures and severe NEC were independently associated with higher risk of cerebellar injury¹⁷⁹. Few studies of surgically treated NEC have been carried out in low resource settings. However, a retrospective South African study covering 1992–95 identified a prevalence of significant neurodevelopmental delay (49%) on follow-up, which although high, is similar to that reported in high-income settings¹⁸⁰.

Other complications

NEC also increases the incidence of retinopathy of prematurity¹⁸¹, above and beyond that expected from prematurity alone¹⁸². Moreover, NEC associated bronchopulmonary dysplasia can occur with greater frequency and severity in patients with NEC, with a restrospective single cohort analysis indicating that BPD occurred in 77% of patients with NEC¹⁸³, in part due to shared inflammatory signaling between the lung and the gut^184–186.

Understanding of long-term outcomes of NEC would benefit from the establishment of population-based registers, with international collaboration necessary to accrue sufficient patient numbers to elucidate causal relationships, and research mechanistic pathways.

Outlook

Diagnosis

Traditional approaches to the diagnosis of NEC have relied on a combination of clinical evaluation, laboratory tests, and radiographic evaluation which includes both abdominal radiography and abdominal ultrasonography¹⁸⁷. Each of these diagnostic modalities is associated with varying degrees of specificity and sensitivity, which may delay the diagnosis of NEC and adversely affect disease outcomes¹⁸⁸. Indeed, physical findings occur in other septic conditions in neonates¹⁸⁹ and laboratory tests have low specificity for NEC¹⁹⁰, although the presence of progressive thrombocytopenia¹⁹¹ and high CRP (>7–8 mg/dL)^192,193 have diagnostic value in predicting more severe disease requiring surgery. Moreover, although abdominal distention and tenderness in the setting of pneumatosis intestinalis, pneumoperitoneum or portal venous gas on radiograph or ultrasound can establish a need for abdominal exploration, these radiographic findings can also occur in other abdominal emergencies, including intestinal volvulus, gastric perforation, or other infectious enteropathies⁸⁹.

In view of these limitations, there has been increasing enthusiasm to identify novel diagnostic approaches with greater sensitivity and specificity for NEC¹⁹⁴. These efforts have started to yield results¹⁹⁵. For instance, high serum levels of fatty acid binding protein (FABP) has emerged as an attractive candidate for the early diagnosis of NEC¹⁹⁶. Other researchers have used proteomic approaches in order to identify candidate biomarkers in the serum and the urine¹⁹⁷. The stool of neonates has also been examined in order to identify early markers for NEC, with a focus on changes in volatile organic compounds^62,198, fecal calprotectin¹⁹⁹, stool microbiome²⁰⁰, and proteomics²⁰¹. In addition to these biochemical and bacteriologic studies, investigators have also examined novel imaging techniques, including near-infrared spectroscopy (NIRS), a non-invasive technique that uses near-infrared light waves to measure tissue hemoglobin oxygenation, and which has been shown to provide diagnostic clues for NEC when tissue hemoglobin oxygenation is determined to be reduced^202,203. While the current approach to the diagnosis of NEC relies heavily on approaches that have changed little in the past 30 years, the outlook for NEC diagnosis likely includes a combination of new biochemical and imaging modalities that will allow for earlier diagnosis, and more effective treatment.

Biomarkers

Given the diagnostic uncertainty and the need to predict NEC development, much effort has been directed towards developing biomarkers. Predictive biomarkers could allow for targeted preventive measures, whereas biomarkers predicting disease course could enable case-specific therapies, avoiding unnecessary and potentially harmful interventions.

Biomarkers for NEC can be categorized into non-specific markers of inflammatory pathways (e.g., IL-6), enhanced non-specific biomarkers (e.g., Calprotectin and volatile organic components), and specific gut injury-associated biomarkers released due to ischemic or inflammatory injury to the intestine (e.g., I-FABP, or calprotectin).^{196,198,204–206}. One study estimated a sensitivity of 64% and specificity of 88% for I-FABP levels for the prediction of NEC, and a sensitivity of 86% and specificity of 92% for the prediction of complicated NEC within 24 hours after onset of disease¹⁹⁶. Biomarkers can be used alone or in combination, with algorithms combining clinical data and various biomarkers potentially providing more sophisticated information.

As previously mentioned, preterm infants who develop NEC show altered gut microbiota signatures with reduced diversity compared to other preterm infants of the same gestational age. Some researchers have suggested that a NEC-associated microbiota is present in meconium, although larger studies are needed to confirm this.²⁰⁷ In the future, analyses of the gut microbiome might therefore be of use to identify children at the highest risk for NEC, which might enable targeted use of e.g. probiotics (see below).

Cerebral and intestinal oxygenation measured using near-infrared spectroscopy (NIRS) has been studied to predict, diagnose, and prognosticate NEC. One study has suggested that NIRS can predict NEC development within a week after birth by using a splanchnic oxygen index of 30% as a cutoff, such that infants with values below 30% had 4.5 times higher odds for NEC development.²⁰⁸ Moreover, combining cerebral and splanchnic FTOE values with I-FABP levels could provide insight into the pathological chain of events in NEC.^209,210 However, larger studies are needed to confirm the routine use of NIRS in NEC.

Despite extensive research, no formal biomarker for NEC exists due to challenges such as the lack of established reference values, intra- and inter-individual variations, and the absence of large multicenter studies. Moreover, suitable samples (e.g., stool, urine, blood) might not always be available in critically ill neonates, and the turnaround time for detection of calprotectin and iFABP levels may not be fast enough to be clinically useful

Remote ischemic conditioning (RIC) as a means of NEC prevention

One development in the prevention and potential treatment of NEC is Remote Ischemic Conditioning (RIC), which involves brief, reversible cycles of ischemia and reperfusion in a limb to activate protective pathways and improve blood flow in distant organs like the intestine. RIC reduces intestinal injury in a preclinical model, appeared safe and feasible during a phase I trial,²¹¹ and its feasibility is being further investigated in a phase II trial.²¹² The timing and dosage of RIC for NEC remain unclear, as does the duration of its protective effects on neonatal intestinal microcirculation. RIC may be more effective when combined with hydrogen sulfide and nitric oxide donors, potentially enhancing and prolonging its benefits on intestinal perfusion and NEC outcomes by either vasodilatory or anti-inflammatory effects.^213,214

New and Investigational Therapeutic Treatments

The treatment of NEC includes a combination of antibiotics, cessation of feeds, and surgical resection of necrotic bowel. Despite expertise in the current treatment of NEC, the overall mortality in patients with NEC remains high², indicating an urgent need for novel therapies. The outlook for therapies for NEC includes a combination of prebiotics, probiotics, small molecules and stem cells. While each of these potential new therapeutic approaches are at the preclinical stage, there is reason for great enthusiasm for their ultimate translation to the bedside.

Lactoferrin

Lactoferrin is an iron-binding protein found in high concentrations in human colostrum with antimicrobial and anti-inflammatory properties. Aside from the potential ability of lactoferrin to prevent NEC, a pre-clinical trial in a murine NEC model suggested a reduction of NEC severity after lactoferrin administration by gavage, through upregulating intestinal epithelial proliferation.²¹⁵ Clinical trials are exploring lactoferrin supplementation as a therapeutic measure for NEC.^216,217

Granulocyte-Colony Stimulating Factor (G-CSF) stimulates the production of white blood cells and enhances immune response, potentially reducing infection and promoting healing. Investigational use of G-CSF in NEC treatment in a pilot study in preterm infants shows promise in boosting immune function and aiding in recovery,²¹⁸ but more studies are needed to establish its efficacy and safety.

Stem Cell Therapy

Stem cell therapy shows potential for NEC treatment – including mesenchymal stem cells and amniotic fluid stem cells – in large and small animal models^219–222, through mechanisms that remain incompletely understood. Challenges with stem cell therapy include determining the optimal timing and delivery method. Potential risks include immune rejection and tumorigenesis, although studies aim to mitigate these risks. Stem cell-derived products may offer safer alternatives than stem cells, although further research is needed to address extraction, dosage, and delivery methods. Although the concept is promising, extensive research is required to confirm safety and efficacy.^214,223

Epidermal Growth Factor (EGF)

EGF promotes intestinal cell growth and repair, potentially enhancing gut integrity and function. EGF has mainly been studied with the aim of preventing NEC; however, studies in preclinical models and early small clinical trials have also evaluated use as a treatment for NEC. EGF has shown potential in reducing NEC severity and promoting intestinal mucosa healing in both animal models and human trials.^224,225 Preliminary studies have shown that EGF it is well-tolerated, but long-term effects are still being studied.

TLR4 inhibition

Preclinical trials support the effectiveness of nucleotide-binding oligomerization domain-2 (NOD2) in mice with NEC.²²⁶ C34, a 2-acetamidopyranoside oligosaccharide, blocks TLR4 and effectively treats NEC in mice and piglets when administered orally or intravenously.¹²⁹ Pregnane X receptor (PXR), a xenobiotic sensor and signaling intermediate for certain host-bacterial metabolites, downregulates TLR4 signaling and delivery of PXR was associated with lower NEC severity in a murine model.²²⁷ Other factors shown to treat NEC in murine models through inhibition of the LPS/TLR4 cascade include brain derived neurotrophic factor (BDNF)²²⁸ and the aryl hydrocarbon receptor (AHR).²²⁹ Candidate AHR molecules can also be delivered to the mother during pregnancy and prevented NEC development in mice, raising the possibility that NEC prevention strategies could be delivered to the developing fetus²²⁹. Further trials are warranted to investigate feasibility, short- and long term safety and efficacy of inhibition of TLR4 in preterm human neonates.

LPS reduction

Intestinal alkaline phosphatase is an enzyme present in breast milk and is known to reduce intestinal inflammation by dephosphorylation of LPS.²³⁰ Preclinical observations have suggested a potential therapeutic role for alkaline phosphatase in the treatment of NEC, although this requires further investigation.^231,232

Prebiotics and novel probiotics

Prebiotics for NEC include molecules that promote the growth or function of commensal bacteria within the lumen of the intestine, and thus induce an anti-inflammatory environment within the intestinal mucosa²³³. Human milk oligosaccharides (HMOs) are prototype prebiotic, and comprise a family of non-digestible carbohydrates that enhance bacterial growth and prevent NEC in animal models^234–237. The mechanism of action of HMOs for NEC includes their prebiotic activity and their ability to induce nitric oxide release (a vasodilator)²³⁶, increase mucin expression²³⁷, and block the binding of LPS to TLR4²³⁵, thus attenuating the pathways that lead to NEC. Of note, breast milk containing higher levels of the HMO disialyllacto-N-tetraose (DSLNT) was more protective against NEC than breast milk that was relatively DSLNT-deficient²³⁸.

Probiotics have emerged as effective agents for the prevention of NEC in multiple large clinical trials with Bifidobacterium, Lactobacillus, and Enterococcus emerging as the most efficacious agents in reducing the mortality and incidence of NEC²³⁹. Probiotics are now being engineered further for greater functional activity through the development of bacterial-matrix constructs²⁴⁰.

The current approaches to the diagnosis and the treatment of NEC have changed minimally in the past 30 years, in part due to an incomplete understanding of the biological underpinnings of the disease, and a lack of treatment options. Recent advances in our understanding of the pathogenesis of NEC has led to the development of novel diagnostic approaches and have set the stage for a large number of potential therapies. The future care of patients with NEC is likely to include a sophisticated diagnostic approach including novel molecular and imaging analyses, leading to the precise delivery of targeted molecular and cellular therapies.

Figure 5. Infant with Bell Stage III NEC.

Representative posterior-anterior (A) and cross-table lateral (B) radiographic images of a premature infant with pneumoperitoneum (arrows).

Box 2. Clusters of NEC cases.

NEC occasionally presents in clusters, in which several patients are affected at one time, or individual cases that occurred within days or weeks of each other. NEC clusters are particularly impactful when they occur in units with either a relatively low baseline NEC incidence, or when they have not seen NEC for periods of many months. When NEC clusters are encountered, additional workup should be performed to search for the possible presence of a particular trigger. Triggers associated with NEC clusters may include bacterial causes. For example, specific E. coli subtypes and strains like Cronobacter sakazakii can cause NEC clusters²⁴⁹. Moreover, molecular studies reveal that NEC clusters are preceded by colonization with specific virulent bacterial strains that trigger TLR4 signaling in the immature intestine²⁸. Some evidence also suggests that viral infections can be triggers of NEC²⁵⁰, with a particular risk for rotavirus (OR 3.96), cytomegalovirus (OR 3.50) and norovirus (OR 11.95)⁸. Longitudinal virome analysis has identified specific viral signatures that precede NEC onset, including convergence toward reduced viral beta diversity in the 10 days prior to disease development²⁵¹. NEC clusters may also reflect changes in the care environment, and the adoption of new feeding practices or crowded nursery conditions have been linked to NEC outbreaks²⁵². Enhanced infection prevention and control measures, including hand hygiene education and contact precautions, have been associated with decreasing NEC outbreaks²⁵³. Nevertheless, in many cases, despite thorough workup, a specific trigger for the cluster is not identified, and the cluster disappears spontaneously for reasons that are not be understood.