Aim To determine (1) the incidence of neurodevelopmental impairment (NDI) in necrotising enterocolitis (NEC), (2) the impact of NEC severity on NDI in these babies and (3) the cerebral lesions found in babies with NEC.
Methods Systematic review: three independent investigators searched for studies reporting infants with NDI and a history of NEC (PubMed, Medline, Cochrane Collaboration, Scopus). Meta-analysis: using RevMan V.5.3, we compared NDI incidence and type of cerebral lesions between NEC infants versus preterm infants and infants with medical vs surgical NEC.
Results Of 10 674 abstracts screened, 203 full-text articles were examined. In 31 studies (n=2403 infants with NEC), NDI incidence was 40% (IQR 28%–64%) and was higher in infants with surgically treated NEC (43%) compared with medically managed NEC (27%, p<0.00001). The most common NDI in NEC was cerebral palsy (18%). Cerebral lesions: intraventricular haemorrhage (IVH) was more common in NEC babies (26%) compared with preterm infants (18%; p<0.0001). There was no difference in IVH incidence between infants with surgical NEC (25%) and those treated medically (20%; p=0.4). The incidence of periventricular leukomalacia (PVL) was significantly increased in infants with NEC (11%) compared with preterm infants (5%; p<0.00001).
Conclusions This study shows that a large proportion of NEC survivors has NDI. NEC babies are at higher risk of developing IVH and/or PVL than babies with prematurity alone. The degree of NDI seems to correlate to the severity of gut damage, with a worse status in infants with surgical NEC compared with those with medical NEC.
Trial registration number CRD42019120522.
- neurodevelopmental impairment
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
What is already known on this topic?
Necrotising enterocolitis (NEC) is a devastating disease in premature babies who may develop long-term sequelae.
Several NEC survivors suffer from some degree of neurodevelopmental impairment (NDI).
Some NEC survivors develop brain injury that is worse than that observed in their premature peers who did not have NEC.
What this study adds?
The incidence of NDI in NEC survivors is 40%.
NEC survivors are more likely than age-matched controls to develop cerebral palsy, blindness, and hearing impairment.
Babies with severe NEC that require surgery have poorer neurological outcomes than those with milder NEC that can be treated conservatively.
Necrotising enterocolitis (NEC) remains one of the most common and severe gastrointestinal emergencies in the neonatal period.1 NEC primarily affects premature neonates, occurs in approximately 7% of extremely low birthweight infants and requires surgery in the most severe cases.2–4 Understandably, NEC survivors are often burdened by short-term and long-term digestive morbidities, such as intestinal failure, parenteral nutrition–related liver disease or bowel obstruction.5 6 Furthermore, NEC survivors have been reported to be at risk of developing brain injury and neurodevelopmental impairment (NDI).6–9 In the past, this risk was attributed to the prematurity of their neonatal brain, but several recent studies have shown that babies who had NEC develop a brain injury that is worse than that observed in their premature peers who did not have NEC.10–12 The true incidence of NDI in NEC survivors remains unknown. Moreover, several studies have reported on different types of NDI that NEC survivors may develop, but to date no study has synthesised the accumulated data on the nature of NDI in NEC. Studies have reported that the risk of NDI seems to be significant in infants with advanced NEC who require surgical intervention.7 8 However, limited data are available on the neurodevelopmental sequelae of infants with NEC treated surgically compared with those treated medically.
In the present study, we systematically reviewed the literature (1) to determine the pooled incidence of NDI in infants with NEC, (2) to investigate the impact of NEC severity on the neurodevelopmental outcome in NEC babies and (3) to determine the type of cerebral lesions infants with medical and surgical NEC may develop.
Data sources and study selection
Both the systematic review and the meta-analysis were drafted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement.13 A librarian specialising in systematic reviews was consulted at the Gerstein Science Information Centre (University of Toronto, Canada). This study was registered with PROSPERO—international prospective register of systematic reviews.14
A systematic review of the literature was made using a defined search strategy. Three investigators (AM, AG and LM) independently searched scientific databases (PubMed, Medline, Cochrane Collaboration, Scopus) looking for all studies, published from 1975 to 2018, reporting neurodevelopmental outcome and/or cerebral lesions in human infants and children with a history of NEC (table 1). Reference lists were searched to identify relevant cross-references. Case reports, opinion articles, experimental studies and case series with less than 10 patients were excluded. All grey literature publications (ie, reports, theses, conference proceedings, bibliographies, commercial documentations and official documents not published commercially) were excluded. Full-text articles of potentially eligible studies were retrieved and independently assessed for suitability by the three investigators (AM, AG and LM). Any disagreement over the eligibility of a specific study was resolved through the discussion with a fourth author (AZ).
NEC was defined as Bell’s stage II or III, diagnosed clinically or radiologically.15 Surgical NEC was defined as NEC treated with a laparotomy, laparoscopy or peritoneal drainage. Medical NEC was treated with antibiotics, resuscitation and/or withdrawal of feeds. We considered NDI as a significant deviation or loss of neurodevelopmental function, resulting in below-average performance on neurological, cognitive or developmental assessment tools as reported by the investigators of each study. This definition includes motor deficits (such as cerebral palsy), sensory deficits (vision and hearing impairment), neurobehavioural and/or educational issues, as well as short-term or long-term sequelae.
The main outcome measures of this study were
The incidence of NDI overall (as reported by the investigators of each study) and the incidence by type of NDI.
Cerebral and neurological lesions in infants with NEC (intraventricular haemorrhage, periventricular leukomalacia and retinopathy of prematurity).
Categorical variable frequencies were compared using Pearson’s χ2 test or the two-tailed Fisher exact probability test, as appropriate. A meta-analysis was performed using comparative studies between infants with NEC and those without NEC with the following outcomes: NDI, cerebral palsy, hearing impairment, vision impairment, intraventricular haemorrhage, periventricular leukomalacia and retinopathy of prematurity. We also performed meta-analyses using comparative studies between infants with surgical and medical NEC, with the same outcomes when available.
Meta-analysis of comparative studies was conducted with RevMan V.5.3 (Review Manager (RevMan); Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration; 2014), using the random-effects model to produce risk ratio (RR) for categorical variables, along with 95% CIs. We produced I2 values to assess homogeneity and quantify the dispersion of effect sizes. A p value <0.05 was considered statistically significant.
Risk of bias for individual studies was assessed in duplicate (LM and GL) using the methodological index for non-randomised studies (MINORS).16 Differences between the two reviewers (LM and GL) were resolved through consensus and discussion with a third author (AZ). The total score for this 12-item instrument ranges 0–24 points with a validated ‘gold standard’ cut-off of 19.8.
For each outcome, we graded the quality of evidence as high, moderate, low and very low, using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) methodology.17 The quality of evidence was rated down in the presence of risk of bias, inconsistency, indirectness, imprecision and publication bias. For assessment of risk of bias in observational studies, we used the MINORS instrument. Inconsistency was determined according to heterogeneity. We produced I2 values to assess heterogeneity. I2 value of 0–40, 30–60, 50–90 and 75%–100% were considered as low, moderate, substantial and considerable heterogeneity, respectively. Imprecision was assessed using optimal information size, which was based on 25% relative risk reduction, 0.05 of α error and 0.20 of β error.18
Of 10 674 abstracts screened, 203 full-text articles were thoroughly examined, and 53 studies were included in the final analysis (figure 1). Of the latter, 21 were designed to address neurodevelopmental outcome and/or neuroimaging abnormalities in NEC survivors,11 12 19–37 and 11 of these were prospective studies.11 20 22 23 26–29 32 35 37
Incidence of NDI in NEC survivors
The overall incidence of NDI in NEC infants was 40% (IQR 28%–64%), as calculated from 31 studies (n=952/2403 infants with NEC; table 2).11 12 19–28 30 31 33 36 38–52 When we analysed the 17 comparative studies,11 12 21–23 25–27 31 33 39–41 45 47 50 51 we found that the risk of NDI was significantly higher in NEC infants (40%, 751/1876 patients) versus premature age-matched controls without NEC (29%, 7955/27 287 patients; p<0.00001; RR 1.59, 95% CI 1.42 to 1.79; figure 2A).
Types of NDI in NEC survivors
The most common type of NDI in infants with NEC was cerebral palsy, which was found to be more frequent in babies with NEC (17%) in comparison with premature age-matched controls (7%; p<0.0001; RR 1.84, 95% CI 1.38 to 2.45; figure 2B). Infants with NEC were also found to have higher incidence of blindness (3%) than age-matched controls (0.7%; p<0.001; RR 3.36, 95% CI 1.60 to 7.02; figure 2C). Moreover, hearing impairment was observed in 3% of NEC babies and in 1.5% of age-matched controls (p<0.0001; RR 2.00, 95% CI 1.41 to 2.83; figure 2D).
Three studies reported the occurrence of attention deficit hyperactivity disorder (ADHD) in infants with NEC, finding NEC to be a risk factor for developing ADHD.39 53 54 In particular, one study reported that ADHD was more frequent in NEC survivors (15%) compared with control preterms (8%).39 Moreover, we found studies reporting that NEC survivors had poor educational outcomes, such as low reading scores, low cognitive index, language impairment, low memory skills and problems with adaptive behaviour.10 26 55–60 As a result, 28% of NEC survivors required special education classroom and 21% required speech therapy.58
Impact of NEC severity on the developing brain
In the 10 studies that compared surgical versus medical NEC, we detected a higher incidence of NDI in infants with surgical NEC (43%) than in those with medical NEC (27%; p<0.00001; RR 1.60, 95% CI 1.34 to 1.90; figure 3).11 19 21 22 25 28 31 36 40 41 Furthermore, compared with those with medical NEC or without NEC, infants with surgical NEC were more likely to have cerebral palsy (surgical NEC: 24% vs medical/no NEC: 15%).25 Infants who underwent surgery for NEC were reported to have a delay in verbal skills and non-verbal performance compared with aged-matched control infants without NEC.26 36 Moreover, infants with surgical NEC had a higher risk of developing head growth failure and microcephaly compared with those with medical NEC.28 61 Only one study compared the severity of NDI across patients with surgical NEC and reported that infants who received an ostomy following bowel resection scored lower on intelligence testing than infants who had a primary anastomosis.30
Cerebral and neurological lesions in NEC survivors
Intraventricular haemorrhage (IVH) was more common in NEC babies (21%) compared with age-matched preterm infants (17%; p<0.001; RR 1.38, 95% CI 1.15 to 1.66; figure 4A). Likewise, the incidence of periventricular leukomalacia (PVL) was significantly increased in infants with NEC (10%) compared with preterm infants (5%; p=0.001; RR 1.87, 95% CI 1.29 to 2.71; figure 4B). Conversely, we did not find a difference in the incidence of retinopathy of prematurity between infants with NEC (45%) and premature peers (50%; p=0.09; RR 1.41, 95% CI 0.95 to 2.11; figure 4C). There was no difference in IVH incidence between infants with surgical NEC (25%) and those treated medically (20%; p=0.3; RR 1.49, 95% CI 0.70 to 3.20; figure 5).22 25 29 32
Brain imaging and functional studies
Two studies on head ultrasonography in infants with NEC reported that NEC was strongly associated with ventriculomegaly.62 63 Moreover, infants with advanced NEC were reported to have a higher risk of ventriculomegaly (45%) than infants with milder forms of NEC (1%).62 Infants with surgical NEC were more likely to have cystic PVL on head ultrasound compared with those who had no NEC (14% vs 7%).25 A study focusing on infants <1500 g reports an incidence of white matter injury on ultrasound as high as 86% in infants with NEC compared with 27% in weight-matched controls.64
In a study reporting on near-term structural MRI in preterm infants, 43% of infants with a history of NEC had brain abnormalities.65 On MRI, infants with surgical NEC or spontaneous intestinal perforation had significantly more white matter injury than infants with medical NEC (p=0.002) or age-matched controls (p<0.01).32 66 Moreover, preterm infants with surgical NEC had more white matter injury than infants with spontaneous intestinal perforation.34 MRI studies on premature infants showed that those with NEC had an impaired brain growth, as suggested by the smaller biparietal width brain pattern compared with controls.67 A study using diffusion-weighted MRI reported that the fractional anisotropy values throughout the white matter of babies with surgical NEC were lower than preterm controls, reflecting the severity of white matter injury.68
Our systematic review shows that since the first studies reporting neurodevelopmental outcomes of babies with NEC in the 1990s, the awareness and interest aimed at this aspect of NEC-related morbidity has increased. The papers collected in our study show that almost half of NEC survivors develop some degree of NDI that is worse than that of preterm infants without NEC. Moreover, babies requiring surgery for NEC have poorer neurological outcomes than those treated conservatively. This suggests that the degree of NDI seems to correlate to the severity of gut damage.
The pathogenesis of brain injury secondary to NEC remains unclear. In several septic perinatal conditions, such as early-onset sepsis or maternal chorioamnionitis, an association has been suggested between elevated levels of systemically disseminated inflammatory cytokines and cerebral injury or NDI.64 69 Similarly, NEC results in increased expression of several pro-inflammatory mediators.70 This has led to the hypothesis that NEC involves an infection-induced systemic inflammatory response with cytokine release, which in turns induces brain injury and NDI.71 In fact, elevated serum IL-6 in babies with NEC has been associated with poor growth and neurodevelopmental disability on long-term follow-up.27 Interestingly, there was no experimental study addressing this mechanism until recently, when in 2018 two research groups published three papers on the topic.72–74 Brunse et al reported that piglets with NEC developed systemic inflammation that led to blood–brain barrier disruption, resulting in region-specific neuronal degeneration.72 These results were confirmed by Niño et al, who demonstrated that, in a mouse model of NEC, activation of intestinal TLR4 signalling resulted in the release of high-mobility group Box 1 protein in the intestine, which activated the brain microglia and caused neurological dysfunction.74 More recently, our group reported that the levels of pro-inflammatory cytokines and the density of activated microglia and astrocytes were increased in the brains of mice with NEC and were positively correlated with the increase in the levels of pro-inflammatory cytokines in the gut and the severity of NEC damage, respectively.75 These laboratory findings are useful because they provide an explanation for the development of NEC-associated brain injury by suggesting the activation of the gut–brain axis is implicated in the pathogenesis and are also relevant for their clinical implications. In fact, these experimental findings would support the early resection of diseased intestine in babies with NEC. The management of infants with perforated bowel secondary to NEC includes a laparotomy with bowel resection or the insertion of a peritoneal drain.76 The latter was proposed as an alternative to surgery in fragile babies with NEC, especially in those with a low birth weight (<1500 g).77 78 Two randomised controlled trials comparing surgery versus peritoneal drain failed to show the advantage of one approach over the other.79 80 As a result, many surgeons prefer to insert a drain with the assumption that a laparotomy would be a stressful insult on an unstable and fragile infant with surgical NEC. The new evidence from the laboratory seems to assert the opposite, that is, leaving necrotic bowel in the abdomen may cause detrimental effects on the developing brain.81 A randomised controlled trial, the NEST trial, is currently underway to evaluate laparotomy versus drainage in infants with NEC with the primary endpoint being death or neurodevelopmental impairment at 2 years of age (NCT01029353).
In this systematic review, we have found that babies with NEC are prone to developing several types of NDI, such as cerebral palsy, visual, hearing and cognitive impairment, as well as neurobehavioural issues. Although it is difficult to establish a causative effect of NEC on the impairment of the different brain regions, it seems likely that NDI, including visual and hearing impairment, results from an insult to the central nervous system. In fact, we observed no differences in the incidence of retinopathy of prematurity in infants with NEC, suggesting that the type of blindness NEC survivors develop might occur through an independent mechanism, possibly via a cerebral injury. Hearing impairment in babies with NEC is proposed to result from impaired neural conduction in the brainstem auditory pathway, with NEC likely causing damage to myelination and synaptogenesis along the immature brainstem auditory pathway.82–84 Moreover, it has recently been reported that NEC survivors are more likely to develop long-term attention deficits that could lead to poorer educational outcomes and a need for specialised education.54 Although the underlying cerebral injury responsible for these deficits remains unclear, a link between the gut microbiota or traumatic brain injury and ADHD has been recently suggested.85 86
We acknowledge that this study has some limitations, partly due to its design and partly due to the condition studied. As with any systematic review, we relied on the quality of papers published in the literature and possibly accrued biases derived from the original articles. Specifically, the diagnosis and management of both NEC and neurodevelopmental impairment are very variable among clinicians, and can be measured in different ways, with different tools and at different time points. All these factors may introduce additional biases that cannot be avoided especially in retrospective non-randomised studies. The quality of the published studies in this area limits our ability to establish a link of causality between the severity of bowel damage and the degree of brain injury, thus leaving space for speculations and hypotheses. Nonetheless, laboratory studies are able to provide a more mechanistic explanation, but are limited by modelling NEC. With the increasing awareness of NDI in NEC survivors and the establishment of dedicated long-term follow-up clinics looking at this aspect of child development, we anticipate better designed clinical studies that will shed light on this association. In our meta-analysis, most studies included were retrospective (online supplementary file 1), and none of the studies reached the gold standard cut-off on MINORS, provided sample size calculations or reported a blinded evaluation of objective endpoints (online supplementary file 1). Furthermore, a GRADE Evidence Profile table was provided (online supplementary file 2). According to GRADE methodology, the quality of evidence was low for the outcomes reported (NDI, cerebral palsy, blindness, hearing impairment and neurological lesions). However, when independently assessed by two authors (GL and AZ) using A Measurement Toll to Assess Systematic Reviews (AMSTAR), the present systematic review and meta-analysis received a relevant score (online supplementary file 3). The PRISMA checklist was then completed. Lastly, we acknowledge that our review includes studies that were conducted over a broad period of time (1975–2018). This approach might introduce a bias, as the state of neonatal-perinatal medicine and the care of at-risk neonates has changed overtime. However, as the implementation of some practices may not have occurred concurrently and homogeneously in all institutions and countries, for this study we preferred to have a comprehensive approach.
This study shows that 40% of NEC survivors have NDI. NEC babies are at higher risk of developing IVH and/or PVL than babies with prematurity alone. In addition, blindness, hearing loss and cognitive impairment are significantly heightened in NEC survivors. The degree of NDI seems to correlate with the severity of gut damage, as shown by the worse status of infants with surgical NEC compared with those with medical NEC. Clinician awareness of the long-term outcomes in these children is important to ensure early detection, ongoing follow-up and support, and timeous treatment for these children. This study has solidified the importance of neurodevelopmental assessment at follow-up after treatment for NEC, as almost half of these babies go on to develop neurological sequelae.
AM and LM contributed equally.
Correction notice This paper has been corrected since it was published online. The affiliation for author
Giuseppe Lauriti has been updated.
Contributors Study conception and design: AZ, AM, LM. Data acquisition: AM, AG, LM. Analysis and data interpretation: AZ, LM, GL. Drafting of the manuscript: AZ, LM, AM. Critical revision: AZ, GL.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement All data relevant to the study are included in the article or uploaded as online supplementary information.