Physiologic brain dysmaturity in late preterm infants

Abstract

Neonatal EEG sleep was used to determine if differences are expressed between healthy late preterm and full term groups. Twenty-seven 24-channel multi-hour studies were recorded at similar post–menstrual ages (PMA) and analyzed for eight asymptomatic late preterm infants (LPT) compared with nineteen healthy full term (FT) infants as a preliminary analysis, followed by a comparison of a subset of 8 FT infants, matched for gender, race, and PMA. Z scores were performed on data sets from each group pair comparing each of seven EEG/Sleep measures for entire recordings, active (AS) and quiet sleep (QS) segments and artifact-free intervals. Six of seven measures showed differences between the 8 LPT and 8 matched FT cohort pair comparisons of >0.3; REMs, arousals during QS, spectral correlations between homologous centro-temporal regions during QS, spectral beta/alpha power ratios during AS and QS, a spectral measure of respiratory regularity during QS, and sleep cycle length. Quantitative neurophysiologic analyses define differences in brain maturation between LPT and FT infants at similar PMA. Altered EEG/Sleep behaviors in the LPT are biomarkers of developmental neuroplasticity involving interconnected neuronal networks adapting to conditions of prematurity for this largest segment of the preterm neonatal population.

INTRODUCTION

Preterm infants comprise 12.8% of all live births with the late preterm neonate (LPT) making up 72% of the overall preterm population (1). The birth rate for LPT (i.e. 34 0/7 weeks to 36 6/7 weeks) has been steadily increasing over the last quarter century. While the number of neonates less than 34 weeks has increased by approximately 10% since 1990, LPT has increased by nearly 25%. While more mature preterm infants have historically been considered healthy with comparatively low risk, there is now growing evidence that this population is not as healthy as previously considered. Compared with full term (FT) infants there is an increased mortality as well as higher risks for complications such as transient tachypnea of the newborn (TTN), respiratory distress syndrome (RDS), persistent pulmonary hypertension (PPHN), respiratory failure, temperature instability, jaundice, sepsis, feeding difficulties and prolonged neonatal intensive care unit (NICU) stays. Given their greater numbers compared to all preterm infants, LPT consume a significant amount of health care resources. In addition to increased mortality, LPT have an increased likelihood of neurological morbidities leading to adverse long-term neurodevelopmental consequences. Reliable neonatal biomarkers are needed that can more specifically define neurological status for LPT as a function of postnatal brain organization and maturation to help more accurately track neurodevelopmental outcome after the institution of neuroprotective interventions.

Functional brain organization and maturation of newborns have been assessed by visual and digital analyses of EEG/Sleep differences for over the last fifty years (2). This bedside neurophysiological test remains a reliable and comparatively low cost biomarker of brain function for different neonatal populations. Differences in EEG/Sleep organization and maturation have been described between more immature preterm (<32 weeks gestational age) and FT cohorts at matched post-menstrual ages (PMA) (3-9). Specific EEG/ Sleep behaviors suggest either an acceleration of brain maturation as expected for an older infant when compared to the full term neonate. Other measures suggest a delay in brain maturation expected for a preterm neonate (10, 11). In order to reconcile physiologic precocity or immaturity for any particular EEG/Sleep measure, an analytic approach comparing Mahalonobis distances were performed between preterm and full term cohorts across different physiologic groupings of multiple EEG/Sleep behaviors. A physiologic dysmaturity index (DI) was then defined based on seven measures of neonatal EEG/Sleep that best differentiated these neonatal cohorts in terms of state-specific neurophysiologic behaviors. DI was initially described for a healthy preterm group born less than 32-weeks gestational age for a comparison to EEG/Sleep behaviors of term infants at matched post-menstrual term ages. This study compares the DI of a LPT cohort to a FT group.

METHODS

Patient selection

The institutional Review Boards (IRB) at Rainbow Babies and Children’s Hospital, University Hospitals of Cleveland, Case Western Reserve University (Cleveland, Ohio) and Magee-Women’s Hospital, University of Pittsburgh (Pittsburgh, Pa) approved consent forms used to recruit subjects for this study. These 27 neonates were part of a larger database of 461 subjects and 1116 multi-hour recordings. Clinical, demographic and neurophysiologic information for this specific cohort was extracted from this larger database. Recruitment over a twenty-year period of all subjects in the database were part of five NIH-supported research grants, with IRB-approved consent forms, identified from neonatal populations admitted to the two neonatal intensive care units of these two hospitals. All EEG/Sleep studies were individually analyzed by visual and digital analysis methodologies.

From the total research group, eight LPT born between 34 0/7 and 36 6/7 weeks gestational age were identified. Selection was based on review of maternal, fetal and neonatal medical records, and recruitment occurred after consultations with the attending neonatologist. Infants chosen for this group were clinically asymptomatic throughout the study period. None were treated for severe respiratory distress syndrome, sepsis, encephalopathy, or seizures. Brazy Neurobiologic Risk Scores (NBRS) had been historically used for the entire preterm cohort in our database since this scoring system is designed for low birth weight infants to predict neurodevelopmental outcome. For the present study, since our study group was LPT, we only adopted the severity of illness score at FT ages to confirm healthy medical status for the LPT group to then compare with a healthy FT group without the intention of comparing later outcome. NBRS consists of a score of 1 to 4 for each of eight clinical items; blood pH, hypoglycemia, intraventricular hemorrhage, periventricular leukomalacia, seizures, infection, and the need for mechanical ventilation. Normal cranial ultrasounds were therefore described for all preterm neonates. Scores can be classified as low risk (≤4), moderate risk (5 – 7), and high risk (≥8). (12), and were all low risk for the LPT group. The PMA listed for each subject was the corrected age at term when the EEG/Sleep study was performed.

Nineteen appropriate for gestational age FT infants were identified from the database. Reviews of maternal and neonatal records as well as physical examinations were performed to verify the healthy status of this group. Nineteen neonates were initially identified whose EEG/Sleep measures were compared with the study group, matched over a wider gestational range of 38 to 42 weeks (mean group difference of 2 weeks), irrespective of gender or race. As with the LPT group, the PMA for each FT subject was the age when the EEG/Sleep study was performed. A group of eight FT newborns were then selected who were matched for gender, race, and a narrower PMA range (mean difference of 6 days). For the 8 LPT and 8 matched comparisons, all subjects were Caucasian.

EEG Sleep Recordings

Electroencephalographic/polysomnographic studies (EEG/sleep) were recorded on multi-modality 24 channel recording devices (Nihon-Kohden America, Inc., Foothill Ranch, Ca., model 4221), and carried out in an environmentally controlled setting in which sound, light, humidity and tactile stimulation were monitored, as discussed in previously publications (4). All infants were studied while sleeping prone or on their sides in an open bed which was their usual sleeping position in the nursery. Continuous recordings for 3 hours began after a diaper change and feeding at 0900 to 1000 and ended between 1200 and 1300 hours on the same day.

Off-line visual analyses of the all digitized neurophysiologic data were performed. Typed comments and start/end cursors were electronically recorded for each multi-hour recording. All notations were then tabulated to derive numbers and durations for specific EEG/Sleep measures; sleep state durations, arousals and rapid eye movements (REMs). One of six neonatal sleep state segments was assigned by a single neurophysiologist (MSS) according to conventional neonatal EEG sleep criteria (2, 13). Segments included 2 active (AS) and 2 quiet sleep (QS) segments as well as indeterminate sleep and waking intervals. For the purposes of this publication, each of the two AS and QS segments were combined into one AS and QS length in minutes. All minutes of all EEG/Sleep studies were scored by MSS, including start and end times of physiologic arousals during all consecutive minutes of the recording. Sleep state segments were digitally annotated on a display of the EEG recording with a start and end time. Artifact segments primarily containing excessive movement-induced artifact during which a sleep state could not be determined were identified. These epochs were later electronically removed to obtain artifact-free segments of all EEG/Sleep segments for each study, while preserving consecutive minutes of sleep that included transient arousal periods. Artifact-free segments of AS and QS permitted more accurate comparisons between groups for spectral power, correlations and respiratory regularity as well as comparisons of arousal indices and REM, since excessive periods of muscle activity and movements would alter these spectral measures. REMs were visually identified on the EOG channels and electronic cursors were placed on the record to obtain totals per unit time.

A neonatal research nurse provided clinical care for each infant during the recording session. Sleep, feeding, behavior, diaper changes, medication administration and technical comments (i.e. equipment malfunctions and environmental measures, etc) were documented in the computer database provided for these studies. Light and sound levels were continuously recorded and entered into the data files. No infants were given medications during the studies. No male children were circumcised prior to the study.

EEG Sleep Measures

A previously described dysmaturity index (DI), consisting of seven EEG/Sleep measures, has been historically used to quantify physiologic differences in brain organization and maturation between healthy preterm infants less than 32 weeks gestation and a full term group at matched PMA (5-8, 10, 11). The seven EEG sleep measures were arousal numbers, rapid eye movement (REM) numbers, percentage of quiet sleep, sleep cycle length, spectral beta/alpha EEG energy ratios, spectral EEG correlations between left and right hemispheric centrotemporal regions (i.e. T3C3/C4T4) and a spectral measure of respiratory regularity. These measures were statistically selected from initial physiologic groupings of thirty-four EEG/Sleep measures. Each of the seven measures best represent the interconnected neuronal circuitries expressed as precocious or immature neurophysiologic behaviors of a preterm cohort when compared to full term infants (4, 10, 11, 14) at similar PMA.

Analytic Methods

The ultimate goal for this study was to compare an effect size by Z-score comparisons between LPT and FT cohorts. Exploratory graphical methods demonstrated that distributions of measures for both cohorts were well represented by normal Gaussian distributions.

Standard Z-scores were calculated as follows: differences between the means of the cohorts over the pooled standard deviations. Sub-analyses were included to help verify true differences between cohorts rather than by chance. Effect sizes were also assessed for both total and artifact-free minutes during active and quiet sleep segments, identified by visual analyses. Z-score values of 0.3 or greater for the 8 LPT and 8 matched FT comparisons were considered supportive of our hypothesis for this publication that physiologic dysmaturity exists in healthy LPT in comparison to FT infants. This threshold for the Z score was chosen since a significant p-value of at least 0.05 could be predicted if greater numbers of subjects were recruited.

RESULTS

Demographic Features

Table 1 presents the 8 subjects in the LPT group compared with 19 and 8 matched FT comparison groups with respect to similar PMA at the time of the study and low risk NBRS of <4. Birth weight, height and head circumferences at birth and at the time of the study at term age are listed, including percentiles for these measurements. Along with the 19 full term cohort, 8 neonates were also included in Table 1, matched for gender, race, and a more similar PMA (i.e. mean 6 days).

Table 1.

Subject Demographics and Measurements.

Cohort	late-preterm	full term†	full term
Subjects (N)	8	8	19
Sex
Female	4	4	9
Male	4	4	10
Race
African American	0	0	4
Caucasian	8	8	15
Birth Statistics
GA (weeks)	35.5 ± 0.6	38.3 ± 1.1	38.7 ± 1.2
Length (cm)	43.4 ± 4.0	49.4 ± 2.9	49.9 ± 3.9
Weight (kg)	2.00 ± 0.61	3.47 ± 0.77	3.53 ± 0.79
Head Circumference (cm)	30.9 ± 2.1	34.6 ± 2.1	34.3 ± 2.4
Average Birth Percentiles for GA
Length (%)	21.1 ± 34.4	42.5 ± 35.1	49.2 ± 36.6
Weight (%)	16.2 ± 23.3	52.1 ± 32.1	58.2 ± 34.2
Head Circumference (%)	28.2 ± 33.2	51.3 ± 32.0	46.3 ± 33.5
EEG Study Age Statistics
PMA (weeks)	38.7 ± 0.7	39.5 ± 1.1	40.7 ± 1.6
Length (cm)	45.4 ± 4.1	49.6 ± 3.4	51.1 ± 4.2
Weight (kg)	2.32 ± 0.77	3.48 ± 0.71	3.65 ± 0.81
Head Circumference (cm)	32.7 ± 2.6	35.7 ± 1.5	35.9 ± 2.0
Average Study Percentiles
Length (%)	13.4 ± 21.5	35.2 ± 34.2	46.7 ± 37.3
Weight (%)	11.6 ± 20.0	40.6 ± 30.6	45.7 ± 34.3
Head Circumference (%)	24.5 ± 33.7	58.0 ± 25.3	53.1 ± 28.9

Values: count or mean ± SD

^†Full term subset matched to late-preterm group by gender, race, and PMA at EEG study.

Brazy Neurobiologic Risk Score (NBRS) was less than 4 for all subjects in all groups.

PMA=Post-menstrual age (weeks); GA=Gestational age (weeks); cm=centimeters; kg=kilograms

The LPT cohort had smaller measurements for head circumference, height and weight (means, standard deviations and percentiles listed in Table 1), both at birth and when compared with FT cohorts.

EEG/Sleep measures

Table 2 lists the effect sizes or Z-scores between cohorts for the 7 DI measures. Z-score comparisons are listed for the 8 LPT vs.19 FT neonates as well as for the 8 LPT vs. 8 matched FT. Five of the 7 DI measures showed effect size differences of 0.25 or greater between the 8 and 19 groups, including REM, % QS, spectral measures of correlation, beta/alpha ratios and respiratory regularity. Six of 7 of the DI measures showed effect sizes of 0.30 or greater for the 8 and 8 cohort comparisons, including REM, QS arousals, sleep cycle length, and spectral measures of correlation, beta/alpha ratio and respiratory regularity. When artifact-free epochs were compared, both the 8 and 19 and 8 and 8 comparisons had significant differences for spectral correlation, AS spectral beta/alpha ratios and AS respiratory regularity. However only the 8 LPT and 8 matched FT group comparisons resulted in significant differences during QS for spectral beta/alpha ratios and respiratory regularity.

Table 2.

Summary of late-preterm versus full term effect sizes [Z-Scores] for dysmaturity index parameters.

Cohort	late-preterm	full-term	full-term†
Gestational Age Range (weeks)	34 to 36-6/7	≥37	≥37
Subject count	8	19 [8 vs 19]	8 [8 vs 8]
Active Sleep REM index	3.9 ± 2.3	5.4 ± 2.5 [-0.59*]	5.4 ± 2.7 [-0.57*]
Quiet Sleep Arousals index	4.1 ± 2.2	4.1 ± 1.9 [0.02]	4.8 ± 2.2 [-0.30*]
Percentage Quiet Sleep	34.4 ± 8.5	36.9 ± 9.2 [-0.28*]	35.5 ± 11.7 [-0.10]
Sleep Cycle Length	62.0 ± 19.8	65.7 ± 22.0 [-0.17]	70.0 ± 28.7 [-0.32*]
Full Study (All Epochs)
Left/Right Pair Correlation	0.12 ± 0.04	0.17 ± 0.09 [-0.67*]	0.16 ± 0.07 [-0.80*]
Beta/Alpha Ratio Index	-0.12 ± 0.04	-0.15 ± 0.07 [0.36*]	-0.11 ± 0.07 [-0.25]
Respiratory Regularity Index	-0.93 ± 0.09	-1.00 ± 0.13 [0.57*]	-0.96 ± 0.13 [0.30*]
Arousal-free Quiet Sleep Epochs
Left/Right Pair Correlation	0.15 ± 0.07	0.21 ± 0.13 [-0.49*]	0.18 ± 0.09 [-0.32*]
Beta/Alpha Ratio Index	-0.21 ± 0.07	-0.20 ± 0.08 [-0.20]	-0.15 ± 0.06 [-0.97*]
Respiratory Regularity Index	-1.20 ± 0.11	-1.21 ± 0.14 [0.10]	-1.17 ± 0.14 [-0.28*]
Arousal-free Active Sleep Epochs
Left/Right Pair Correlation	0.18 ± 0.07	0.19 ± 0.12 [-0.12]	0.18 ± 0.06 [-0.01]
Beta/Alpha Ratio Index	-0.05 ± 0.05	-0.12 ± 0.08 [0.89*]	-0.09 ± 0.07 [0.56*]
Respiratory Regularity Index	-0.84 ± 0.10	-0.92 ± 0.14 [0.65*]	-0.90 ± 0.12 [0.58*]

Values are mean ± SD and [z-score]

^*magnitude of z-scores > 0.25

^†Full term subset matched to late-preterm group by gender, race, and PMA at EEG study.

Brazy Neurobiologic Risk Score (NBRS) was less than 4 for all subjects in all groups.

Effect sizes [z-scores] are the (difference between the means) / (pooled standard deviation)

DISCUSSION

The brain volume of a late preterm infant (LPT) at 34 weeks gestation is approximately 65% of the term brain (15). There will be a 5-fold increase in white matter volume between 35 and 41 weeks gestation. Important maturational changes occur within the brain during late preterm gestation that include increasing neuronal connectivity, dendritic arborization, and synaptic junctions, as well as maturation of neurochemical and enzymatic processes that take part in the regulation of brain growth and maturation. Although less common than younger preterm infants, LPT are more likely to develop periventricular leukomalacia than FT infants because of white matter vulnerability under adverse conditions. LPT are consequentially more vulnerable to more remote injuries in cortical and subcortical gray matter in regions where damaged white matter tracts innervate. Combined white and gray matter injuries lead to neurocognitive and behavioral deficits as the brain matures throughout childhood, as reported for LPT. Neonatal biomarkers are needed that can predict these deficits and quantify the effects of neuroprotective interventions.

Quantitative EEG/sleep studies in LPT have not been specifically reported, although LPT is historically referred to as a cohort who displays EEG/sleep activities that may resemble either preterm or full term patterns (2). This cohort is generally assumed to express specific physiologic behaviors that are intermediate between infants born before 34 weeks and those that are near or at term. This has been pointed out to be generally true for the control of breathing, heart rate and sleep state maturation (16, 17). There is a delay in maturation of integrated autonomic brainstem function throughout the neuroaxis which contributes to an increased risk for acute life threatening events including sudden infant death syndrome compared to term infants.

To our knowledge our study is the first to report altered quantitative EEG/sleep in LPT using visually scored and spectrally calculated neurophysiologic measures. While there have been maturational studies demonstrating differences in EEG/sleep in preterm infants <32 weeks gestation (2, 9, 18), we now suggest that LPT uniquely express altered functional neuronal networks that differ from FT infants. Our study suggests that the DI derived from an EEG-sleep study can be a meaningful physiologic biomarker of functional brain organization and maturation applicable to a LPT population. These physiologic differences may influence how the LPT will react to prenatal/postnatal illnesses, environmental stresses and neuroprotective interventions with long term neurodevelopmental consequences.

EEG/ Sleep behaviors are physiologic surrogates of multiple interconnected neuronal pathways that course through-out the neuroaxis within brainstem, diencephalic and cortical structures. Connectivities among specific pathways subserve state regulation (19-21). The ponto-medullary to basal-frontal pathways subserve respiratory activity. The penduculo-pontine geniculo-calacarine pathways are identified with REM behavior. The ascending reticular activating pathway subserves arousals. The cortico-thalamic pathways subserve quiet sleep (non-REM) expression and the cortico-cortical pathways are expressed as spectral beta/alpha energy ratios and electrode-pair correlations. Our study suggests that for LPT, these neuronal networks mature differently compared with a full term age, and express unique state-specific EEG/Sleep behaviors from its full term counterpart.

We noted different results in EEG/Sleep expression between LPT and FT groups when comparing the two groups pairs, suggesting that a greater range of PMA (mean of 2 weeks) as well as differences in race and gender for the 8 and 19 cohort comparisons influence specific neurophysiologic behaviors. Spectral correlation and beta/alpha ratios may be more robust measures reflecting less developed brain organization and maturation in LPT compared with FT for both group pair comparisons despite greater variations in PMA, gender and race for the 8 and 19 cohort pair. However, greater variations in PMA, gender and race may better explain differences in % QS and respiratory regularity (the latter calculated over the full study) rather than denoting specific differences between LPT and FT EEG/Sleep, since these findings are no longer significant in the 8 LPT and 8 matched FT comparison. When there is a closer match in PMA (mean of less than 6 days), gender and race with the 8 and 8 cohort comparison, other significant EEG/Sleep differences emerge. More mature (i.e. fewer) arousals but more immature (i.e. shorter) sleep cycle lengths are noted in the LPT group. QS-specific differences between LPT and FT are also observed. This tendency to observe differences for specific EEG/sleep measures between LPT and FT neonatal groups during QS has been previously observed in a cohort of healthy neonates over successive days of life after neonatal depression from a difficult parturition and delivery (22). In general, environmental stress alters sleep architecture, with greater percentages of QS. As a result of the greater stress of longer extrauterine life, the brain of the LPT adapts by expressing more discernible physiologic differences during QS, which is the segment of sleep which may be more responsive to stress factors. What is more, these differences were more easily demonstrated when muscle activity and body movements were removed during artifact-free sleep segments since artifact alters the brain generated spectral profiles expressed as power, correlation and respiratory regularity.

Our previously reported index of physiologic brain dysmaturity (DI) for a healthy preterm cohort (i.e. <32 weeks gestational age) at term compared to a FT cohort can be applied as a quantitative metric to represent the interconnected pathways expressed during sleep in LPT. This DI was statistically derived from winnowing down from 34 to 7 physiologic measures of EEG/Sleep that best differentiate brain organization and maturation between cohorts (4, 5, 11). The present study demonstrates that six of these seven measures also differentiate differences in these neuronal networks for LPT compared with FT cohorts, when controlling for PMA, race and gender for the 8 and 8 cohort pair comparison. We plan additional studies of LPT to further investigate time-dependent non-linear and neuronal connectivity relationships, as we previously reported in younger preterm cohorts, to study brain maturation using specific algorithms that quantify neuronal network complexity (23-25).

Previously described as ontogenetic adaptation (26), this biological process of developmental neuroplasticity has been redefined in terms of genetic and epigenetic variability at molecular, cellular and neuronal network levels of biological function (23). Quantitative neurophysiologic measures such as defined by the physiologic brain dysmaturity index represent functional neural network biomarkers of plasticity in developing brain circuitries. EEG/Sleep behaviors in an LPT cohort reflect this adaptive process of neuroplasticity unique to conditions of prematurity for this gestational age-range cohort. Such a metric when combined with genotypic/proteomic biomarkers can help define a functional endophenotype, with greater sensitivity to predict risk for different neurodevelopmental disorders as well as quantify the effects of neuroprotective interventions (27, 28). Furthermore, a genetic-neurophysiologic endophenotype when combined with neuroimaging biomarkers of the preterm brain (29) will further expand the clinician’s structural-functional perspective of the positive and negative consequences of developmental neuroplasticity in preterm populations.

We recognize that the interpretations of our study results are limited by the small sample size and selection strategy of subjects. Repeated assessments with larger sample sizes, matching for multiple demographic and clinical covariates are needed to validate how universally these biomarkers can be applied to all LPT. For example, we matched for race in the 8 LPT and 8 matched FT comparison, but all subjects were Caucasian, Also, while the PMA differences were smaller between the 8 LPT and 8 FT groups, the LPT group remained approximating one week less mature than the FT group. We also acknowledge that the LPT were smaller in weight, height and head circumference, reflecting both prenatal and postnatal influences on growth, suggesting that maternal, placental, fetal and neonatal factors potentially influence brain development through growth mechanisms during prenatal and postnatal periods. Finally, while we used the severity of illness scale, NBRS, as simply an inventory of postnatal diseases, this would be an score not suitable for prediction of neurodevelopmental outcome for LPT. Our results are nonetheless provocative, suggesting that altered neurophysiological maturation exists in LPT, expressed as quantitative measures of EEG/Sleep behaviors. These differences may help predict differences in outcome as well as responses to neuroprotective interventions in future studies.

Abbreviations

AS

Active sleep

DI

Dysmaturity index

FT

Full term infants

LPT

Late preterm infants

NBRS

Nursery Neurobiologic Risk Score

PMA

Post menstrual age

PT

Preterm

QS

Quiet sleep

REM

Rapid eye movement