ASSOCIATION OF INFECTION IN THE NEONATE AND LONG-TERM NEURODEVELOPMENTAL OUTCOME

Abstract

Perinatal and neonatal infection and associated inflammatory response may adversely affect brain development and lead to neurodevelopmental impairment. Factors that predict the risk of infection and subsequent adverse outcomes have been identified but substantial gaps remain in identifying mechanisms and interventions that can alter outcomes. In this manuscript we describe the current epidemiology of neonatal sepsis, the pathogenesis of brain injury with sepsis, and the reported long-term neurodevelopment outcomes among survivors.

INTRODUCTION

Parents, neonatal providers, and researchers identified survival, sepsis prevention, and intact long-term neurodevelopmental function as core outcomes for neonates who require intensive care soon after birth.¹ These outcomes are related; sepsis during initial hospitalization increases risk for both in-hospital mortality and for adverse childhood neurodevelopmental outcomes.^2–6 For infants admitted to the neonatal intensive care unit (NICU), technologies that are life-saving can also increase the risk for unintended consequences, including bacterial infection.⁷ Improvements in preventative efforts have reduced the overall incidence of infection.^{8, 9} However, evolving microbiology with changes in resistance patterns,^{10, 11} and rising vulnerability of high-risk infants admitted to NICUs,¹² highlight the need for ongoing surveillance, better application of known therapies, and discovery of new interventions.¹³

EPIDEMIOLOGY OF NEONATAL BACTERIAL INFECTIONS

Bacterial infection defined as blood or cerebrospinal fluid (CSF) culture-confirmed infection, is described as early-onset (<72 hours after birth) and late-onset infection (≥72 hours after birth till the end of neonatal period at 28 days for term infants and usually until discharge among those continuously hospitalized) in neonates.^{9, 13, 14} This temporal distinction reflects the differing pathogenesis,^{15, 16} microbiology^{17, 18} and preventative strategies.^{8, 19, 20} Meningitis is often associated with bacteremia in early-onset sepsis¹⁷ but later may occur in a third of cases without associated bacteremia.²¹ Necrotizing enterocolitis (NEC) frequently occurs in the late-onset period²² and may be associated with bacteremia.²³ While few studies directly compare neurodevelopmental outcome among infants with early versus late infection, both presentations are associated with elevated risk of mortality and neurodevelopmental impairment (NDI).^{2, 3, 24}

Early-onset sepsis (EOS):

Based on cases reported in 2005–2014 to the Centers for Disease Control and Prevention¹⁴, EOS incidence is estimated at 0.77 per 1000 live birth (LB) with a stable trend over the years of surveillance.⁹ Among the 1,484 cases identified, Group B Streptococcus (GBS) was the leading cause of EOS followed by Escherichia coli and viridans group streptococci. Incidence of all-cause EOS is significantly higher in preterm infants where E. coli the most commonly detected pathogen.⁹ Case-fatality is substantially higher in preterm infants as well, with 165/1,484 (11%) deaths among EOS cases and 125/165 (76%) of these occurring in very-low birth weight infants (VLBW, birth weight <1500 grams). Multivariate modelling revealed a significantly increased risk for mortality with VLBW status (adjusted OR: 3.88; 95% CI: 1.47–10.22) and meningitis (adjusted OR: 3.53; 95% CI: 1.74–7.16).⁹ While more infants with E. coli infection died (23%) compared to GBS (7%), the difference was not significant when adjusting for gestational age.⁹ Similar incidence rates for EOS were reported in full-term infants in a multicenter study of academic centers that provide care for high-risk deliveries.¹⁷ In this study, however, incidence for all-cause EOS and E. coli infection among VLBW infants increased from 11.00 to 15.05 per 1000 LB (p 0.03), and from 5.07 to 8.68 per 1000 LB (p 0.008), respectively.¹⁷

Late-onset sepsis (LOS):

Unlike EOS, estimates for incidence of LOS often represent center-specific data with few nationally representative data sources^{18, 25–28} LOS epidemiology is also often reported separately for high-risk infants with prolonged admissions after birth^{18, 25} and community associated LOS,^{27, 29} with few studies that capture rates across all groups.³⁰ As many high-risk infants receive routine follow-up for underlying complications of prematurity, birth anomalies, or neonatal depression at birth, information for long-term outcomes associated with LOS is enriched for high-risk infants and may be difficult to generalize to all neonates with LOS.

LOS among infants with prolonged admission after birth:

Incidence of LOS is inversely proportional to gestational age²⁵ and directly proportional to the length of stay²⁶ making VLBW infants one of the highest risk populations for LOS. A reduction in incidence of LOS from 21.9% (2005) to 10.1% (2014) was reported for VLBW infants admitted to NICUs across the United States.⁸ Among extremely preterm infants (<28 weeks) admitted to 12 academic sites, LOS rates also decreased but were higher overall at 41% (2000–2005) and 34% (2006–2011) with an unchanged mortality rate (18%).¹⁸ The microbiology of LOS included gram-positive bacteria (>75% cases), most frequently Coagulase-negative Staphylococcus species (CoNS) and Staphylococcus aureus; gram-negative bacteria (17%), most frequently E .coli; and fungal organisms, most commonly of Candida species. Among late preterm infants (34–36 weeks) and term infants (≥37 weeks) without anomalies who required admission to a NICU, a cumulative rate of LOS, was reported at 6.30 episodes per 1000 NICU admissions for late-preterm and 2.7 episodes per 1000 NICU admissions, for term infants, with a distribution of pathogens in both groups similar to those reported in extremely preterm infants.^{26, 28} Compared to the higher mortality reported among preterm infants,¹⁸ mortality among term infants with LOS was lower at 4%, but was substantially higher than the <1% mortality among term infants without late-onset infection.²⁸

LOS among infants not continuously admitted after birth:

Data from an integrated health care system has been used to report the incidence of LOS among term infants without underlying medical conditions.^{27, 31} Among 224,553 full-term deliveries, 6232 (3%) infants were evaluated and 13% of those evaluated were found to have a serious bacterial infection defined as bacteremia, urinary tract infection (UTI), or meningitis. UTI episodes were the most frequent, followed by bacteremia and meningitis though the authors recognized potential for ascertainment bias with changing testing patterns with neonatal age.²⁷ The estimated incidence rate reported per 1000 full-term LB was 3.46 (95% CI: 3.34–3.60) for UTI, 0.57 (95% CI: 0.55–0.59) for bacteremia, and 0.07 (95% CI: 0.069–0.074) for meningitis. The microbiology of community acquired LOS, as opposed to that of healthcare-associated LOS among neonates readmitted to the hospital, is dominated by Gram-negative bacteria, specifically E. coli, with a lower proportion attributed to CoNS.^{18, 27, 29}

PATHOGENESIS OF INFECTION AND BRAIN INJURY

The association of perinatal and postnatal infection with neonatal brain injury has been described, particularly with the manifestation of cystic periventricular leukomalacia (PVL).³² With advances in neuroimaging techniques, there has been increased recognition of the incidence and importance of non-cystic, diffuse white matter injury.³³ While the association of fetal and neonatal infection and brain injury has been established, the degree, the timing, and the mechanism of involvement is still being debated.

The process of neonatal brain injury in the setting of infection is best described with neonatal meningitis, where there is direct penetration of bacteria across the blood brain barrier to infiltrate the central nervous system.³⁴ Interestingly, infection without direct bacterial invasion of the central nervous system can also lead to brain injury and NDI.³⁴ While the exact mechanism of how this occurs is still incompletely understood, the immaturity of the blood brain barrier in preterm infants and the inflammation generated by pathogens can activate a local inflammatory response and/or cause direct cytotoxic injury.³⁴

One mechanism suggested that inflammation associated with infection and hypoxic-ischemic injury occur upstream of the site of infection, instigating the cascade of events leading to brain injury.³³ These events can happen in isolation or combination, and one has the potential to potentiate the effects of the other.³³ In this proposed pathway, infection and hypoxia/ischemia lead to pro-inflammatory microglial activation; which leads to release of inflammatory cytokines such as tumor necrosis factor-α and interleukon-1β, release of reactive oxygen and nitrogen species, and a rise in glutamate levels with subsequent excitotoxicity.³³ Additionally, other types of microglia important to normal development may be converted to a pro-inflammatory state.³⁵ The combination of these factors results in oligodendroglial injury and/or subsequent inhibition of maturation and myelination, axonal damage, and neuronal loss.³⁶

PERINATAL INFLAMMATION AND NEURODEVELOPMENTAL OUTCOMES

Many studies have investigated the relationship of clinical and histological chorioamnionitis at the time of delivery and NDI in the newborn.^37–41 Most of these studies have focused on preterm infants and on cerebral palsy (CP) as an outcome and few report evidence for confirmed infection in the newborn.^{40, 41} In clinical practice, a diagnosis of chorioamnionitis increases the risk of culture-confirmed EOS, but most infants delivered to mothers with chorioamnionitis are not infected.⁴² However, immune activation and inflammation (with or without infection) in chorioamnionitis can result in injury to the developing brain that may lead to long-term sequelae.³⁹

A meta-analysis of studies published through September 2016 reviewed the association of chorioamnionitis and CP, and found differing results by study design - studies that investigated the risk of CP among infants with chorioamnionitis versus those that investigated the incidence of chorioamnionitis among infants with CP.³⁷ Among preterm infants, an increased risk of CP was associated with histological chorioamnionitis, but not with clinical chorioamnionitis. In contrast, an elevated incidence of clinical chorioamnionitis was found among patients with CP for all birth gestations, while the incidence of histological chorioamnionitis was increased only in term infants. The difference in associations between studies that look forward from exposure to backwards from outcome may be due to a second factor on the causal pathway which may be prematurity itself³⁷ or a genetic predisposition (e.g. polymorphism in the IL-6 gene).⁴³

Neurological deficits other than CP have been predominantly reported in preterm infants. In a 10-year prospective cohort study of infants <28 weeks gestational age, investigators found a higher adjusted odds for CP, autism, and epilepsy among infants with histological chorioamnionitis.³⁹ No effect on cognitive function was noted. In contrast, a study of infants born at 24–32 weeks gestation found an association of histological chorioamnionitis with motor or cognitive function at 2 years of age that was not significant after adjusting for key postnatal variables such as LOS.⁴¹ The role of study design, bias in selection, and varying study definitions have been proposed as an explanation for the inconsistency of findings across publications on the impact of chorioamnionitis on neurodevelopmental outcome.^{37, 38}

NEURODEVELOPMENTAL OUTCOMES AFTER NEONTAL INFECTIONS

Early-onset sepsis

Adverse outcomes have been reported among survivors of EOS with 6.3% of episodes resulting in survival with an oxygen requirement, hearing loss, or seizures.⁹ In contrast to studies on chorioamnionitis, few studies report the long-term neurological outcomes among infants with culture-confirmed EOS in the absence of meningitis. Published studies often focus on preterm infants and report an association of EOS elevated risk of mortality and abnormal cranial imaging.^{3, 24, 44}

In a cohort of 7,354 infants born <26 weeks in gestation, EOS occurred in 153 infants and 41% of the infants with EOS died prior to 2 years age.³ Infants with EOS have an elevated adjusted relative risk of death or NDI at 2 years of age of 1.23 (95% CI, 1.10–1.37), compared to infants without EOS. Impairment in both cognitive and motor scores were noted.³ In a 5-year follow-up study of 2,665 infants born at <28 weeks gestation, investigators found an increased risk of CP in survivors of EOS without an increased risk for impaired cognitive outcomes.²⁴

Meningitis

Meningitis, both early and late-onset, has been associated with poor neurodevelopmental outcome. Since the late 1990’s, neonatal mortality due to meningitis has remained stable at 10%.^44–50 Along with increasing rates of survival, up to half of surviving neonates are estimated to have NDI.^{4, 49, 51–54} The rate of moderate or severe NDI in surviving infants ranges from 15–25%.^{44, 51–53, 55–58} In a survey of physicians, five-year old survivors of neonatal bacterial meningitis had a 4- to16-fold increase in serious disability compared to hospital- and physician-matched controls without meningitis.⁵¹ Infants surviving bacterial meningitis have an increased risk of CP, cognitive impairment, behavioral problems, speech, auditory, and visual impairment,; neurologic complications such as hydrocephalus and seizures are also common.^{4, 51, 53, 58, 59}

Preterm infants with meningitis are at significantly increased risk of death and NDI.^{47–50, 60} The mortality rate from GBS and E. coli meningitis appears to be approximately equal.^{4, 45, 50, 58} with the risk for poor outcome not impacted by pathogen^{4, 53, 58};e seizures and delayed time to CSF sterilization are associated with worse outcomes.^{44, 61, 62}

Late-onset sepsis

Systemic infection associated with LOS is often associated with hemodynamic instability, inflammatory response, multi-system organ failure, and worse short and long-term outcomes, particularly for premature infants.^{4, 63}. Neuroimaging may offer insight into mechanisms of brain injury and adverse neurodevelopmental outcomes.^{32, 64, 65} An association between PVL and culture-confirmed infections (blood, CSF, or tracheal) in preterm infants less than 34 weeks gestation was noted to increase with recurrent infectious episodes.⁶⁵ Similarly, repeated culture-confirmed infectious episodes showed progressive white matter injury with serial MRI studies.⁶⁴

In a large cohort study, extremely low birth weight (ELBW, <1000 grams) infants across four different categories of infection (culture-negative clinical infection only, culture-confirmed sepsis only, both culture-confirmed sepsis and an episode of NEC, and culture-confirmed meningitis +/− sepsis) had higher likelihood of NDI at 18–22 months including CP, lower mental developmental index (MDI) score, lower psychomotor developmental index (PDI) scores, and visual impairment as compared to infants without any diagnoses of/treatment for infection.⁴ Hearing impairment was associated with sepsis and sepsis/NEC. This study also found an association between neonatal infection and poor head growth, which has been associated with impaired cognitive functioning and academic performance.^{4, 66} Similarly, in a multicenter Swiss study of preterm infants born 24–27 weeks gestation, culture-confirmed sepsis was associated with CP and NDI at 2 years age compared with infants with culture-negative suspected sepsis and those without any sepsis events.⁶⁷ These associations did not hold for culture-negative suspected sepsis cases, a group that included some infants with NEC. With longer follow-up, at 5 years, Mitha et al examined outcomes of infants born at 22–32 week gestation.²⁴ A higher risk of CP was found with LOS, alone and in combination with EOS, though it did not correlate with severe cognitive impairment.

Candidiasis, with or without bacteremia, has been associated with significant mortality and NDI.⁶⁸ In a multicenter study, infants with gram-negative, fungal, and combination (multiple episodes with different pathogens or a single polymicrobial episode) infection had a higher incidence of hearing impairment than those with no infections and those with CoNS infections.⁴ Other studies have reported an elevated risk of CP,⁶⁷ and cognitive delay⁶⁹ in infants with LOS CoNS infection.

Necrotizing Enterocolitis (NEC)

NEC often affects infants with other complications of prematurity that are also associated with poorer neurodevelopmental outcomes.^{70, 71} This pathologic milieu raises concern about the impact on the developing preterm brain,⁶³ and several authors have reported associations that vary with severity of disease, degree of systemic illness, and management. The pathophysiology of NEC-associated NDI likely is, akin to the pathophysiology of NEC itself, multifactorial.⁷⁰

In a systematic review of 11 observational studies (1989–2006), ELBW infants with NEC had a higher risk of NDI, including cerebral palsy, cognitive impairment, and severe visual impairment, than those without NEC (OR 1.82, 95%CI 1.46–2.27).⁷² This risk was higher in those requiring surgery. The largest study in this review, a multicenter retrospective analysis of ELBW infants from Neonatal Research Network sites⁷³ found that infants with NEC managed by surgery, had an increased risk for cystic PVL, low Mental Development Index (<70), low Phyisical Development Index (<70), and NDI at 18–22 month follow-up assessment compared with infants who had medically managed NEC. This association was also reported in a systematic review, in which the overall odds ratio for NDI was 2.3 higher in among infants with surgically managed NEC.^{74, 75} Comparisons of outcomes in patients with NEC-associated and non-NEC associated intestinal perforations (SIP) have yielded varied results, some with similar risk of NDI and others with more NDI in those with NEC-associated perforations.^76,77 Other factors associated with NEC and its management might also impact neurodevelopmental outcomes, including the method of surgical intervention.⁷⁸

Psychomotor impairment was found to be statistically associated with surgical NEC among infants born between 23–27 weeks; those who also had late-onset bacteremia were at increased risk for CP and microcephaly.⁷⁹ However, medical NEC, with or without late-onset bacteremia, did not increase risk for NDI.

SUMMARY

Efforts to improve sepsis related NDI outcomes among neonates include a multi-step approach consisting of (1) strategies to prevent neonatal infection, (2) early diagnosis and appropriate treatment of infection, (3) judicious follow-up and early intervention for at-risk infants, and (4) improved interventions to optimize functional rehabilitation of children with sequelae including adequate seizure management, timely neurosurgical and orthopedic interventions, if indicated and multidisciplinary support for deficits in motor, cognitive and behavioral development.

Challenges in describing the impact of infection on NDI include the inconsistency in definitions of exposures outcome. Interventions that target host-responses may have greater impact on long-term outcomes but are sparsely studied in the neonatal population. Rigorous studies with long-term developmental follow-up that include objective definitions of infection and consistent outcome measurements are needed to provide prognostic information.

Key Points.

• Neonatal bacterial sepsis remains a leading cause of neonatal mortality that is inversely proportional to gestational age.

• Systemic infection and inflammation can adversely affect early-life brain development resulting in long-term neurological impairments among surviving infants

• While most studies report an elevated risk of neurological impairment in preterm infants with neonatal infection, robust long-term outcomes beyond early childhood across all gestational ages is lacking

• Strategies to prevent infection, early management, close follow-up, and early intervention among affected children remain foundational to improving outcomes.

CLINICAL CARE POINTS:

1. Comparing long-term impairment in neonates with infection, is limited due to few prospective follow-up studies in term neonates and a lack of standard definitions for infections and outcomes.

2. Establishing follow-up that coordinates developmental assessment, neurological testing and early intervention may facilitate early identification and initiation of rehabilitation efforts

3. Novel interventions to target inflammatory pathways mediating injury may improve outcomes in infants with infection beyond survival