Necrotising enterocolitis is a life-threatening gastro-intestinal disorder of infancy wherein a compromised mucosal barrier of the gastro-intestinal tract leads to bacterial translocation, systemic inflammation, and localised perforation in susceptible individuals. Reference Kelleher, McMahon and James1 Necrotising enterocolitis is classically associated with the preterm infant. Reference Neu and Walker2 However, amongst term infants, cardiovascular disease is a major frequently implicated risk factor, accounting for 18% of necrotising enterocolitis in normal birth weight (>2500g) infants. Reference Velazco, Fullerton and Hong3 So-called cardiogenic necrotising enterocolitis has been postulated to possess a unique set of risk factors and pathophysiologic mechanisms, which differentiate it from that seen in prematurity. Reference Bubberman, van Zoonen and Bruggink4,Reference Carlo, Kimball, Michelfelder and Border5 The incidence of necrotising enterocolitis in infants with CHD is approximately 3% with particular lesions associated with a higher incidence. Hypoplastic left heart syndrome has a strong association with necrotising enterocolitis with rates of 6.1–9% in this population. Reference Lau, Cruz and Ocampo6,Reference ElHassan, Tang and Gossett7 The incidence of necrotising enterocolitis in infants with ductal-dependent lesions is reported to be 5%, with a significant proportion seemingly occurring in the post-operative period. Reference Lau, Cruz and Ocampo6 Several unanswered questions remain, including the role of pre-operative feeding in high-risk lesions with several studies suggesting that the practice may be safe, particularly with breast milk feeds. Reference Becker, Hornik and Cotten8–Reference Nordenström, Lannering, Mellander and Elfvin10 In the published literature to date, most retrospective series do not describe the rate of genetic abnormalities amongst participants. Reference Lau, Cruz and Ocampo6,Reference Becker, Hornik and Cotten8,Reference Cognata, Kataria-Hale and Griffiths9,Reference McElhinney, Hedrick and Bush11–Reference Kessler, Hau and Kordasz16 However, in infants with hypoplastic left heart syndrome, a higher rate of chromosomal anomalies has been described in infants who develop necrotising enterocolitis. Reference ElHassan, Tang and Gossett7 Furthermore, in the very low birth weight population, infants with trisomy 21 and atrioventricular septal defect have been shown to be at particularly high risk of necrotising enterocolitis, with 17.2% of such infants developing the condition in a prospective cohort, Reference Fisher, Bairdain and Sparks17 with this association not described in term infants.
Our aim was to review our institutional experience with necrotising enterocolitis in term, normal birth weight infants with CHD with a focus on their mortality outcomes both during and beyond their index admission. We sought to characterise the cardiac lesions most commonly encountered within our patient group and to highlight, where known, any associated genetic anomalies.
Materials and methods
A retrospective study was carried out at Children’s Health Ireland at Crumlin Hospital, Dublin. Ethical approval was gained for this study from the hospital ethics board. Requirement for informed consent was waived in this instance as all data collected were anonymized.
Infants with structural CHD or congenital cardiomyopathy confirmed on echocardiography and with a diagnosis of suspected or radiologically confirmed necrotising enterocolitis treated at our institution between January 2009 and December 2019. Patients were identified by searching our local hospital reporting database to The National Institute for Cardiovascular Outcomes Research or if necrotising enterocolitis was coded on discharge. Necrotising enterocolitis was defined according to Modified Bell’s Staging Criteria (Table 1). Reference Neu and Walker2 Where treating clinicians had not made use of this staging system at time of diagnosis, clinical examination findings, laboratory investigations, and radiology reporting were consulted to reach a consensus amongst the authors as to the appropriate stage. Infants were excluded if they were born less than 36 weeks’ gestation, or had a birth weight of less than 2.5 kilograms. Also excluded were those infants with known anomalies of the lower gastro-intestinal tract. Data were collected retrospectively from both electronic medical records and physical patient charts. Baseline characteristics were collected on gestational age at birth, patient sex, and birth weight.
Modified Bell’s criteria for necrotising enterocolitis
NEC = necrotising enterocolitis—table compiled from references. Reference Neu and Walker2,Reference Niño, Sodhi and Hackam18
Infants were included depending on the presence of structural CHD, cardiomyopathy, or arrhythmia. Structural CHDs were categorised as univentricular or biventricular. Ductal dependence at the time of development of necrotising enterocolitis or prior ductal dependence was recorded. Whether the lesion was detected antenatally was recorded. Medical records were reviewed for the presence or absence of genetic conditions including chromosomal abnormalities and specific molecular gene mutations (if known). Gene variant calling was based on the American College of Medical Genetics and Genomics 2015 criteria. Reference Richards, Aziz and Bale19
The following data were collected on infant feeding patterns; day of life when enteral feeds were commenced; whether the infant was fed enterally at time of development of necrotising enterocolitis; type of feed (breast milk, formula); and mode of feeding (nasogastric, oral). If infants were not enterally fed, need for parenteral nutrition or intravenous fluids was recorded and the proportion of each was quantified where possible.
Day of life at development of necrotising enterocolitis was recorded, in addition to whether it occurred pre or post-operatively. Haemodynamically instability prior to developing necrotising enterocolitis was recorded. This was considered to be present in the following circumstances; cardiorespiratory arrest requiring resuscitation; new or escalating ionotropic requirement; metabolic acidosis with pH < 7.25; pulmonary hypertensive crisis; and arrhythmia resulting in haemodynamic compromise.
The primary outcome was death within 30 days of development of necrotising enterocolitis. In this instance, whether necrotising enterocolitis was listed as the primary cause of death was also recorded. Secondary outcomes included long-term mortality beyond the 30-day period and need for surgical management of necrotising enterocolitis.
A sub-group analysis comparing infants with and without trisomy 21 was conducted. Outcomes within these two groups were analysed to assess whether there was a difference in mortality, Modified Bell’s Stage, or need for surgical intervention for necrotising enterocolitis between the two groups. Fisher’s t test was used for comparison of dichotomous variables. Significance was defined as a p value < 0.05. Statistical analysis was performed using GraphPad (Dotmatics. GraphPad. San Diego, California 2022)
Results
Baseline characteristics
Over the 10-year period, there were 25 patients who met the inclusion criteria. Mean gestational age at birth was 38 weeks (SD 1.39). Mean birth weight was 3.11 kg (SD 0.45 kg). Cardiac diagnoses and their frequency are recorded in Table 2. Infants with single-ventricle physiology comprised the largest group of defects, accounting for 32% of the total. The most common single defect was atrioventricular septal defect, with a total of n = 6 patients (24%), including two patients with an unbalanced atrioventricular septal defect. Patients with critical CHD requiring ductal patency to maintain circulation in the neonatal period accounted for 52% (n = 13). However, only five patients (20%) were on a prostaglandin infusion at the time they developed necrotising enterocolitis, with the remaining eight patients with a duct-dependent circulation developing necrotising enterocolitis in the post-operative period or beyond (32%). Antenatal diagnosis of cardiac defect was made in 56%.
Primary cardiac diagnoses
Genetic testing was not performed in all cases due to institutional practice during the time period of the review. Tests performed are listed in Table 3. Ten patients (40% of total) had a confirmed genetic diagnosis, the details of which are listed in Table 4. Infants with trisomy 21 comprised the greatest number (n = 5/25, 20%). Infants with trisomy 21 accounted for all patients with balanced atrioventricular septal defect (n = 4, 100%). One infant with phenotypic and documented trisomy 21 did not have a laboratory karyotype available for review. In this instance, initial genetic sampling may have been sent from the referring hospital.
Results of genetic testing
Cardiac defects in patients with genetic conditions
Timing of development of necrotising enterocolitis
The median age at development of necrotising enterocolitis was 17 days with an interquartile range of 5–61 days. A large proportion of infants developed necrotising enterocolitis within the first week of life (n = 9, 36%). In seven patients (28%), necrotising enterocolitis developed between the 2nd and 4th weeks of life. Approximately one-third (n = 9, 36%) of patients presented beyond this period with the latest presentation at 31 weeks of age. In those patients who presented beyond the neonatal period, a significant proportion occurred in the post-operative period or post-cardiac catheterisation (n = 4, 44%). The remainder (n = 5, 56%) were haemodynamically unstable prior.
In 40% of total cases, necrotising enterocolitis developed post-operatively or post-cardiac catheterisation (n = 10). The patients and procedures are listed in Tables 5 and 6. Patients with univentricular circulation accounted for the majority of these patients (n = 6, 60%). This most frequently occurred following first-stage surgical palliation (n = 4) and in one instance first-stage palliation with trans-catheter insertion of a PDA stent (n = 1). In 80% of cases (n = 8), there was pre- or post-procedural haemodynamic instability as defined above.
Characteristics of patients who developed necrotising enterocolitis in the post-operative period
Characteristics of patients who developed necrotising enterocolitis post-cardiac catheterisation
Feeding patterns
Nineteen patients were enterally fed at the time they developed necrotising enterocolitis. The feed type for these patients is listed in Table 7. The number of patients receiving breast milk feeds (including in combination with any formula) was low at four patients in total (21% of the enterally fed group). The majority of those infants who were enterally fed were on full enteral feeds (n = 11, 58% of enterally fed group). Four patients developed necrotising enterocolitis on advancement of feeds in the post-operative period.
Feeding patterns
Six infants were nil by mouth at the time of developing necrotising enterocolitis (n = 6/25, 24%), of whom three (n = 3/25, 12%) had never fed prior to developing necrotising enterocolitis. Those infants who had never been fed enterally all developed necrotising enterocolitis within the first week of life, 100% of whom (n = 3) were on prostaglandin infusion at the time. Three of the infants had been fed enterally previously all of whom had pulmonary atresia. Two were in the post-operative period. The final patient was diagnosed postnatally with pulmonary atresia and had fed enterally at birth, but feeds were held at diagnosis. Necrotising enterocolitis developed on the second day of life and this infant died with necrotising enterocolitis listed as the cause of death.
Outcomes
Death within 30 days of development of necrotising enterocolitis occurred in five patients accounting for 20% of the total. Of these, three patients (12% of total) had necrotising enterocolitis listed as the primary cause of death. Total mortality including at long-term follow-up was 40% (n = 10). Follow-up was variable between 0 and 10 years. Patients were classified according to Modified Bell’s Stage and are listed in Table 8. Death within 30 days and at long-term follow-up was increased in Bell’s stages 2 & 3. Patients with a univentricular circulation had total mortality of 62.5% (n = 5), and death within 30 days of necrotising enterocolitis of 37.5%.
Mortality rates by stage of necrotising enterocolitis
Sub-group analysis
We conducted a sub-group analysis comparing infants with and without trisomy 21 to assess differences in outcomes between the two groups using Fisher’s exact test (Table 9). We noted that while there were no significant differences in mortality between the two groups, infants with trisomy 21 were more likely to require surgical management of necrotising enterocolitis. In terms of baseline characteristics, infants with trisomy 21 had a different spectrum of cardiac diagnosis (Table 4) and none were duct-dependent.
Primary and secondary outcomes—comparison of infants with Trisomy 21 and without
p values derived using Fisher’s exact test.
Discussion
Diagnoses
As demonstrated in previous research on this subject, infants with a univentricular circulation accounted for a considerable proportion of the total number of term normal birth weight infants who developed necrotising enterocolitis at our institution, inferring that this group are at particular risk. Reference Lau, Cruz and Ocampo6,Reference ElHassan, Tang and Gossett7,Reference McElhinney, Hedrick and Bush11 Patients with ductal-dependent circulations in the neonatal period accounted for 52% of our total. However, many of these patients developed necrotising enterocolitis, not whilst on prostaglandin infusion, but in the post-operative period and beyond. A possible hypothesis for this is that abnormal haemodynamics including diastolic steal and episodic hypoperfusion may predispose these infants to necrotising enterocolitis.
The most common single lesion in our population was atrioventricular septal defect. The physiological effects of atrioventricular septal defects include left to right shunting, valvular dysfunction, and cyanosis. Reference Fisher, Bairdain and Sparks17 It is worth noting that two patients with atrioventricular septal defect in our group had concurrent LV hypoplasia conferring additional risk. Although atrioventricular septal defects have been shown to be strongly associated with necrotising enterocolitis in preterm infants, this has not been previously demonstrated in term infants of normal birth weight. Reference Fisher, Bairdain and Sparks17 Patient selection may partly explain why this association has not been demonstrated in prior work, with some studies excluding lesions not deemed to be high risk for necrotising enterocolitis with a focus on duct-dependent and univentricular lesions. Reference ElHassan, Tang and Gossett7,Reference Cognata, Kataria-Hale and Griffiths9,Reference Natarajan, Anne and Aggarwal13 Ireland, where this study was performed, has rates of live-born infants with trisomy 21 syndrome that are up to 3 times higher than European counterparts in part due to restrictions on termination of pregnancy for fetal anomalies. Reference Loane, Morris and Addor20 Trisomy 21 and atrioventricular septal defect are strongly associated and therefore the incidence of atrioventricular septal defect in our population may be higher when compared with previous literature on the topic.
Genetics
Infants with a confirmed genetic diagnosis accounted for 40% of our total series and infants with trisomy 21 alone accounted for 20% of the total patients. In addition to the cardiovascular effects of atrioventricular septal defect, infants with trisomy 21 are at an increased independent risk of pulmonary hypertension and myocardial dysfunction in both the neonatal period and beyond. Reference Smith, Levy, Franklin, Molloy and El-Khuffash21 While it is not routine practice at our institution to screen for immune deficiency in this group, increasingly appreciated is the substantial immune dysregulation associated with the condition which includes impairments in both innate and adaptive immune systems. This places these infants at an increased risk of serious infection and sepsis, while dysregulated cytokine release can lead to deleterious outcomes. Reference Huggard, Doherty and Molloy22 An excessive luminal inflammatory response to bacterial stimuli is a histological feature of necrotising enterocolitis, Reference Neu and Walker2 although there is no literature that demonstrates this specifically in infants with trisomy 21, there is a well-described increased incidence of congenital gastro-intestinal anomalies in this population. Reference Holmes23 Patients with major gastro-intestinal abnormalities were excluded from this series, however, there may be predisposing factors in the gastro-intestinal tract of patients with Trisomy 21 which makes them more susceptible. Therefore, the risk of developing necrotising enterocolitis may be multi-factorial.
The majority of previously published cohorts do not report their rates of chromosomal abnormalities. Reference Lau, Cruz and Ocampo6,Reference Becker, Hornik and Cotten8,Reference Cognata, Kataria-Hale and Griffiths9,Reference McElhinney, Hedrick and Bush11–Reference Kessler, Hau and Kordasz16,Reference Gong, Chen and Wang24,Reference Choi, Song and Kim25 El Hassan et al. in a large multi-centre cohort study of necrotising enterocolitis in infants with HLHS demonstrated that a larger percentage of infants who developed necrotising enterocolitis had chromosomal abnormalities (8.9% versus 4.7%, p,0.001) and those with chromosomal abnormalities had higher mortality (7.4% versus 4.5%).
Smaller studies have not noted this difference. A single-centre retrospective nested 2:1 matched case-control comparison of infants with CHD, both with and without necrotising enterocolitis, did not demonstrate a statistically significant difference in the rate of chromosomal abnormalities (25.9% versus 16.7% p = 0.31). Overall, patient numbers in this study (n = 81) were considerably smaller than the El Hassan cohort. Reference ElHassan, Tang and Gossett7,Reference Iannucci, Oster and Mahle26
In our series, patients with trisomy 21 were more likely to require surgical management of necrotising enterocolitis, suggesting that this is a vulnerable patient cohort. However, there was no difference in mortality. A limitation of this study is that it lacks a comparator group allowing for comparison of the background rate of chromosomal abnormalities in our CHD population.
Outcomes
The short-term mortality reported in this series is similar to previous series (20%). Reference Lau, Cruz and Ocampo6,Reference ElHassan, Tang and Gossett7,Reference McElhinney, Hedrick and Bush11 While some series have reported higher rates, these have tended to exclude suspected necrotising enterocolitis (analogous to Modified Bell’s Stage 1) from their analyses. Reference Bubberman, van Zoonen and Bruggink4,Reference Becker, Hornik and Cotten8 We have reported mortality according to Bell’s stage showing increasing mortality with stage. We have not identified any previous studies looking at long-term follow-up of infants with CHD who develop necrotising enterocolitis. Reference Bubberman, van Zoonen and Bruggink4,Reference Lau, Cruz and Ocampo6–Reference Becker, Hornik and Cotten8,Reference McElhinney, Hedrick and Bush11,Reference Motta, Scott and Mahony12,Reference Siano, Lauriti, Ceccanti and Zani14,Reference Kessler, Hau and Kordasz16,Reference Gong, Chen and Wang24 Notable in this series is the high mortality rates at long-term follow-up, particularly for those patients who developed stage 2 and stage 3 necrotising enterocolitis (67 and 56%, respectively). While these deaths cannot be attributed to necrotising enterocolitis, the high rates suggest that this is a particularly vulnerable group of patients, with the development of necrotising enterocolitis a poor prognostic indicator for long-term survival.
Feeding patterns and timing
A large proportion of infants developed necrotising enterocolitis within the first week of life. However, 55% of infants presented beyond the first week of life. Particularly high-risk periods appear to be the post-operative period, particularly in those with a univentricular circulation. Therefore, the ability to mitigate the risk of necrotising enterocolitis development through withholding feeds may be limited. Indeed, 28% of patients were nil by mouth at the time they developed necrotising enterocolitis, and 17% had never been enterally fed. Further work is necessary to determine whether feed advancement schedules both in the early neonatal period or in the post-operative period could help to ameliorate the risk of necrotising enterocolitis. Notable in this population is the low rate of human breast milk feeds (24% of the enterally fed group). Efforts to increase the rate of human breast milk feeds at our institution are ongoing.
Limitations
This study has several limitations, which must be clearly stated. Data are derived from a single centre. Findings described pertain to the only paediatric cardiology centre providing paediatric cardiac surgical care in Ireland so the findings are representative of the national experience. However, these findings may not be generalisable to other distinct populations. This study was conducted retrospectively and therefore is exposed to inherent biases as practices have changed over time. Follow-up periods varied from between 1 and 10 years. This study is presented as a series as it lacks a comparator group of patients to act as controls. Therefore, the interpretation of the findings in a broader sense is therefore limited and the study is not constructed to demonstrate risk factors for developing necrotising enterocolitis.
Conclusion
The high rate of trisomy 21 and atrioventricular septal defect noted in this series has not been previously described in term infants with CHD and necrotising enterocolitis. There is considerable overlap between the two conditions, and it is unclear which confers the greater risk. Atrioventricular septal defect confers numerous physiological effects, and there is a notable absence of infants with trisomy 21 and isolated ventricular septal defects who developed necrotising enterocolitis. Proposed mechanisms for the increased risk in trisomy 21 include immune dysregulation, cardiac dysfunction, and elevated pulmonary pressures. Infants with trisomy 21 had higher necrotising enterocolitis-related morbidity and were more likely to require surgical management of necrotising enterocolitis. While two patients with atrioventricular septal defects did not have trisomy 21, these patients had unbalanced defects. Overall, a high proportion of patients (40%) had a confirmed genetic condition including chromosomal anomalies and specific gene mutations. High mortality at long-term follow-up was noted in patients with Bell’s Stage 2–3. While the cause of death was attributable to necrotising enterocolitis in only a small number of cases, it suggests that the development of necrotising enterocolitis in infants with CHD could be considered a prognostic indicator of future adverse outcomes in this vulnerable cohort.
Acknowledgements
The authors would like to acknowledge the contribution of data management staff of the Children’s Heart Centre and the Paediatric Intensive Care Unit for their help in collating the data for this study.
Author contribution
All authors contributed to and reviewed the manuscript. Sean T Kelleher and John Coleman collected the data. Sean T Kelleher compiled the manuscript in draft form. Colin J McMahon reviewed and edited the manuscript and contributed to discussion points. Adam James devised the study, reviewed the data and manuscript, and contributed extensively to the discussion.
Financial support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Competing interests
None.
Ethical standard
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national guidelines on human experimentation (please name) and with the Helsinki Declaration of 1975, as revised in 2008, and has been approved by the institutional committees of Children’s Health Ireland at Crumlin.
Open access
