Background
As the survival rate of preterm infants has increased, the management of long-term complications, especially neurodevelopmental impairment, becomes important [
1]. The most common brain injury in premature infants is periventricular leukomalacia [
2], and its pathogenesis involves the effect of hypoxia, ischemia, and inflammation of the progenitor oligodendrocyte cells present during weeks 23–32 of gestation [
3].
Preterm infants are at a high risk for infection, and infection-induced inflammation was found to contribute to adverse neurodevelopment in several clinical and animal studies [
4‐
8]. The mechanisms of systemic inflammation-induced neuronal injury have not been fully elucidated, but systemic inflammation is known to disrupt the blood-brain barrier and incite a local inflammatory response in the brain [
7,
9]. The magnitude of systemic inflammation is possibly reflected in the level of blood inflammatory markers and these markers could indicate the high-risk preterm infants for brain injury and poor neurodevelopment at a later stage. A valid diagnostic biomarker needs to be designed for evaluation in a clinical setting of systemic inflammatory state [
10‐
12].. Furthermore, current neonatal intensive care increasingly utilizes non-invasive brain monitors, such as amplitude-integrated electroencephalography (aEEG), to detect acute brain injury, and a previous study showed that aEEG was associated with inflammatory markers and neurodevelopment [
13]. However, there have been few studies on the association between adverse neurodevelopment and inflammatory markers or aEEG findings during systemic inflammatory episodes that preterm infants experience during neonatal intensive care unit (NICU) admission.
The objective of this study is two-fold: first, to investigate whether preterm infants who experienced systemic inflammation induced by clinical infection and NEC are more susceptible to impaired short-term neurological and CA of 18 months neurodevelopmental outcomes in our study population, and second, to examine the association of laboratory markers including inflammatory cytokines and brain injury markers and aEEG at the time of systemic inflammation in the NICU and neurodevelopmental outcomes. We hypothesized that inflammation increases cytokine levels and causes brain injury, which are reflected as abnormalities in aEEG and adverse neurodevelopment.
Methods
Study design
This was a prospective cohort study. We recruited infants born before 30 weeks of gestation, or whose birth weights was less than 1250 g, and were admitted to the NICU of the Seoul National University Children’s Hospital from December 2013 to June 2017. We excluded infants with congenital anomalies and grade IV intraventricular hemorrhage (IVH).
The subjects were grouped according to the presence of systemic inflammation. During the NICU stay, systemic infection was clinically suspected if there were at least two newly developed signs of fever, apnea, difficulty in breathing, tachycardia or bradycardia, hypotension, irritability, lethargy, and feeding intolerance. For such infants, blood tests, including blood culture, complete blood count (CBC), and C-reactive protein (CRP), were performed subsequently for sepsis evaluation. If at least one of the following abnormal laboratory findings was revealed: (1) elevation of CRP > 1.0 mg/dL, (2) abnormal CBC of elevated proportion of immature neutrophils (I/T ratio ≥ 0.2), or platelets ≤100 × 109/L, and parenteral antibiotics were administered for more than 5 days, they were designated to the systemic inflammation group (I). I group would include subjects who revealed positive blood cultures or were diagnosed with stage II or higher necrotizing enterocolitis (NEC) when they fulfilled aforementioned criteria. If infants did not develop any septic episodes during the NICU stay or show negative laboratory results after the sepsis workup, they were designated to the control group.
In the I group, CBC with differential counts and CRP level were evaluated on day zero (the day of symptom onset), two, and six of the inflammatory episodes along with an aEEG. We performed aEEG within 4 h of septic workup. In addition, plasma and cerebrospinal fluid (CSF) samples were obtained for cytokine and brain injury marker analysis. The CSF tapping was performed for patients suspected of having sepsis or central nervous system infection. For the entire cohort, an aEEG was performed at 35 weeks of postmenstrual age (PMA), a brain MRI was performed before discharge from the NICU, and follow-up information was collected at a corrected age (CA) of 18 months.
To determine whether systemic inflammation is associated with abnormal neurologic or neurodevelopmental outcomes, we compared HC, aEEG, brain MRI, and Bayley scores between the I and control groups. In addition, to identify predictors of neurologic and neurodevelopmental impairment among infants who experienced systemic inflammation, we analyzed the correlation between biomarkers and neurodevelopmental outcomes. This study followed the guidelines set by Declaration of Helsinki and received ethical approval for human subject by the Institutional Review Board of Seoul National University Hospital, and written informed consent was obtained from the parents following a detailed description of the purpose of the study (IRB No. 1301–058-458).
aEEG
For infants in the I group, aEEG recordings were performed on day 0 (the day of onset of symptoms), 2, and 6 of the inflammatory episodes and at 35 weeks PMA. The infants in the control group were only assessed at 35 weeks PMA. A portable electroencephalogram monitor, Olympic Cerebral Function Monitor (Olympic Medical, Seattle, USA), was used. Continuous impedance was measured, and the electrodes were deemed to be well applied when the impedance was less than 10 kΩ. Continuous recording was performed over 3 h [
14‐
16]. The aEEG scoring system was used to evaluate a maturation score. Four components of the aEEG, including continuity, cycling, amplitude of the lower border, and bandwidth span, were evaluated and the summed score ranged from 0 to 13 [
14]. In addition, we classified the aEEG background activity as normal amplitude, the upper margin of band of aEEG activity > 10 μV and the lower margin > 5 μV; moderately abnormal amplitude, the upper margin of band of aEEG activity > 10 μV and the lower margin ≤5 μV; and suppressed amplitude, the upper margin of the band of aEEG activity < 10 μV and lower margin < 5 μV [
17]. We monitored the seizure spike at a rapid rise in both the lower and upper margins of the trace, which usually exhibits a ‘saw tooth’ pattern. We also evaluated the simultaneous raw EEG data [
18].
Cytokine and brain injury marker analysis (plasma and CSF)
Plasma samples were collected on the day of symptom onset (day 0). CSF samples were obtained when CSF tapping was performed for patients suspected of having sepsis or CNS infection. The analyses of IL-1 beta, IL-6, IL-8, Myelin Basic Protein (MBP), Tumor Necrosis Factor-alpha (TNF-alpha), S100B, Glial fibrillary acidic protein, and Enolase 2 were performed [
19]. The enzyme-linked immunosorbent assay kit (Cloud-Clone Corp, Houston, Texas, USA) and Quantikine® (R&D systems, Minneapolis, Minnesota, USA) were used for the quantitative determination of cytokines and brain injury markers.
Brain MRI
Brain MRI, including T2 and T1-weighted sequences, was performed at discharge. Experienced pediatric radiologists assessed the degree of white matter injury (WMI) by scoring five categories on MRI, including the nature and extent of white matter signal abnormality, periventricular white matter volume loss, cystic abnormalities, ventricular dilatation, and thinning of the corpus callosum [
20]. These five scores were summed and WMI on MRI was categorized as none (total score 5–6), mild (total score 7–9), and moderate to severe (total score 10–15). The radiologists were blinded to the clinical status of the patients.
Follow-up outcomes at a CA of 18 months
Body measurement data, including weight, length, and head circumference, were collected at a CA of 18 months. The diagnosis of cerebral palsy were collected. The diagnosis of cerebral palsy was made at the discretion of rehabilitation specialists [
21]. The patients were assessed using the Bayley Scales of Infant and Toddler Development - the 3rd edition (Bayley-III). The Bayley-III generates composite scores for five domains including cognitive, language, motor, social-emotional, and adaptive behavior [
22].
Statistical analysis
Statistical analyses were performed using the SPSS Statistics 20.0 software (SPSS Inc., Chicago, IL, USA). The chi-squared test was used to compare categorical data between the groups, while the Mann-Whitney U tests were used to compare quantitative data between groups. ANCOVA was used to analyze differences between groups in head circumference, aEEG maturation scores, WMI on MRI, Bayley scores, and incidence of cerebral palsy, while controlling for gestational age (GA), 1- and 5-min Apgar scores, respiratory distress syndrome (RDS), bronchopulmonary dysplasia (BPD), patent ductus arteriosus (PDA) operation, retinopathy of prematurity (ROP) (≥stage III), and IVH (≥Grade III). Pearson’s partial correlation analysis (adjusted for PMA of the inflammatory event and GA) was used to evaluate the relationship between the biomarkers and neurological outcomes including aEEG, brain MRI, and HC in the I group. Correlation analysis (adjusted for PMA of the inflammatory event and GA) was performed to assess the relationship between biomarkers and the results of neurological evaluation, including aEEG, brain MRI, and HC and Bayley scores at a CA of 18 months. P-values < 0.05, two-tailed, were considered statistically significant.
Discussion
Preterm infants who experienced systemic inflammation in the NICU had smaller head sizes and abnormal functional brain maturation at near-term age, more WMI at discharge, and had poor neurodevelopment at a CA of 18 months. In addition, the analyses to find associated factors for poor neurodevelopment revealed that elevated levels of I/T ratio, CRP, and plasma TNF-alpha and the presence of seizure spikes on aEEG at the initial phase of inflammatory episodes were correlated with neurodevelopmental outcomes. Our study is the first to comprehensively evaluate head growth, blood markers, aEEG, brain MRI, and neurodevelopment at a later stage in premature infants who experienced systemic inflammation.
The HC at 36 weeks of PMA was smaller in the I group compared to that in the control group. The group differences in HC were maintained at a CA 18 months after adjusting for GA. The HC at 36 weeks of PMA also correlated with Bayley scores at a CA of 18 months. Therefore, smaller head size at 36 weeks of PMA shown in the I group could be an indicator of poor neurodevelopment at a later stage.
Infants in the I group had significantly more WMI on the conventional MRI at discharge than those in the control group. A recent study using diffusion MRI reported differences in the white matter microstructure of the corpus callosum of 6-year-old preterm children in relation to the presence or absence of severe inflammatory conditions in the neonatal period [
23]. The authors concluded neonatal inflammation is one medical factor that may contribute to variation in long-term neurobiological and neuropsychological outcomes in preterm infants.
Compared to the control group after adjusting for GA, the I group had significantly lower cognitive, language, motor, and adaptive behavior scores. However, when the morbidities were adjusted as well, only the motor domain remained significant. Previous studies suggested that preterm infants with postnatal sepsis or NEC are more likely to have adverse neurodevelopment [
4,
5,
7,
24,
25]. In our study, infants with systemic inflammation without proven sepsis or NEC also had delayed maturation on aEEG, more WMI, and adverse CA of 18 months neurodevelopmental outcomes indistinguishable from infants with proven sepsis or NEC; this has been inconsistently shown in previous studies. Some studies reported that clinical infection without proven sepsis was also associated with poor neurodevelopment or brain white matter abnormalities [
4,
26], but in another study, the major impact on neurodevelopmental outcomes was confined to preterm infants with proven sepsis [
7].
We attempted to identify biomarkers associated with a high risk for neurodevelopmental impairment among infants with systemic inflammation. First, we analyzed whether the aEEG was associated with poor neurodevelopmental outcomes. The seizure spike on D0 aEEG had significant relationship with the motor and social-emotional domains. The increase in I/T ratio and CSF IL-1 beta, which are the indices of systemic and central nervous system inflammation, respectively, was significantly related to an increased seizure spike and low maturation of D0 aEEG. This finding implies that abnormal aEEG findings at symptom onset might be mechanistically linked to the inflammatory status in infants. Septic adult patients without central nervous system infection develop diffuse cerebral dysfunction and show characteristic EEG findings, known as sepsis-associated encephalopathy [
27]. Preterm infants also show acute electroencephalographic changes such as burst suppression during sepsis [
15].
Second, blood markers during inflammatory episodes were analyzed. It is well known that an increased I/T ratio and CRP and thrombocytopenia suggest infection [
28,
29]. D0 CRP was negatively correlated with the language and motor domains, and the D0 I/T ratio was negatively correlated with the motor and social-emotional domains. Several studies have shown that brain injury is associated with high inflammatory cytokine levels [
30‐
32]. We analyzed inflammatory cytokines and brain injury markers from D0 plasma and CSF. CSF IL-8, TNF-alpha, S100B, and Enolase 2 showed negative correlations with HC at 36 weeks of PMA; however, they were not correlated with HC at a CA of 18 months or Bayley scales. In a previous study, concentrations of inflammation-related proteins including IL-6, TNF-R2, IL-8, MCP-1, and ICAM-1 in plasma were associated with the risk of microcephaly at a CA of 2 years [
33].
Unlike other plasma markers, plasma TNF-alpha was correlated with scores of many domains in Bayley scales and HC at a CA of 18 months. On the other hand, CSF TNF-alpha was not associated with neurodevelopment and only correlated with HC at 36 weeks of PMA as with other CSF cytokines and brain injury markers. This finding might imply the critical role of plasma TNF-alpha in the mechanism of inflammation-associated brain injury in preterm infants. In the Extremely Low Gestational Age Newborn (ELGAN) study, elevated levels of TNF-alpha in blood collected on postnatal day 14 are associated with impaired mental and motor development at 2 years of age [
34]. From the same cohort, preterm infants who had sustained elevations of inflammation-related proteins including CRP, TNF-alpha, and IL-8 in the first postnatal month are reported to be more likely to have cognitive impairment at 10 years of age [
35]. While the ELGAN study collected cytokine samples independent of inflammatory events, our study analyzed the cytokines from samples obtained at the time of systemic inflammation. We think our approach has the potential to set an appropriate timing for collecting samples when we investigate the effect of systemic inflammation on brain injury and subsequent neurodevelopment because blood-brain barrier disruption is known to be a late or terminal event seen in large-scale activations of the immune system, such as sepsis [
9].
From the above results, aEEG findings and blood biomarkers during inflammatory episodes were associated with poor neurodevelopment. Interestingly, the neurodevelopmental outcomes were more closely associated with usual laboratory markers of I/T ratio or CRP rather than most of the inflammatory cytokines or brain injury markers investigated in this study. Notably, this is an observational cohort study; therefore, causation of study findings cannot be concluded, although there were some significant correlations between aEEG, CRP, and plasma TNF-alpha during inflammation and late-stage Bayley scores.
Our study has several strengths. We attempted to capture the point of significant systemic inflammation and conducted a comprehensive evaluation encompassing growth, body fluid biomarkers, neurophysiologic study, brain imaging, and CA of 18 months neurodevelopmental outcomes. We performed Bayley-III, which provides not only granular evaluation of cognitive, language, and motor development, but also social-emotional and adaptive behavior. This has enabled us to examine more diverse aspects of development in study infants. Limitations of this study are the sample size is modest from a single center, and aEEG and cytokine analysis during inflammation were partly performed in the I group. The D0 may not be the first day of an inflammatory episode for infants exposed to chorioamnionitis because sepsis workup was performed only when symptoms of systemic infection were present.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.