Neurodevelopmental outcomes at 9-14 months gestational age after treatment of neonatal seizures due to brain injury

Abstract

Purpose:

Infants with brain injury are susceptible to developmental delays. Survivors of neonatal seizures are at risk for developmental delay, epilepsy and further neurological co-morbidities. Despite advances in neonatal critical care, the prevalence of adverse long-term outcomes and seizure recurrence remains unchanged. Our goal is to determine if early treatment of neonatal seizures with phenobarbital or levetiracetam is associated with worse neurodevelopmental outcomes in brain injured infants.

Methods:

We conducted a retrospective cohort study of 119 infants admitted between 2013 and 2017 who were at risk for developmental delay and assessed in our clinic. We compared brain injury infants with neonatal seizures to brain injury infants without neonatal seizures using Bayley scores, (BSID III) at 9-14 months gestational age. A comparison of Bayley scores between those exposed to phenobarbital and levetiracetam was conducted.

Results:

22 children with neonatal seizures scored lower than 53 children without seizures in all domains with significant values in composite scores for cognitive function (p = 0.003) and language (p=0.031). We found no difference in scores at 9-14 months between infants exposed to phenobarbital versus levetiracetam.

Conclusions:

Our results suggest that in infants with brain injury, the occurrence of neonatal seizures has an adverse effect on neurodevelopmental outcomes. The choice of antiseizure medication may not play a significant role in their outcomes.

Introduction

Neonatal seizures occur in one to four out of 1000 live births in North America and are a major predictor of future neurodevelopmental outcomes.¹ The incidence of ongoing epilepsy after neonatal seizures has been estimated to be 17.6-21.0%.^{2, 3} Several antiseizure medications are used in the neonatal ICU. These include phenobarbital, fosphenytoin, clonazepam, topiramate and levetiracetam. Phenobarbital is the most commonly used antiseizure drug for neonatal seizures.⁴ Levetiracetam is another antiseizure drug available in intravenous and oral form that is used off-label for the treatment of neonatal seizures.⁵ Analysis of current prescribing trends across the United States continue to show high utilization of phenobarbital but with an increase in levetiracetam use.⁶ Unfortunately neither drug is greatly effective at treating neonatal seizures. In a comparison of phenobarbital and phenytoin found both drugs equal and incompletely effective at treating neonatal seizures.⁷ Levetiracetam was associated with seizure improvement within 24 hours for 8 out of 23 neonates in a retrospective cohort study.⁸ Another recent study showed levetiracetam stopped seizures in 27 of 32 neonates with hypoxic ischemic encephalopathy who failed to respond to phenobarbital.⁹ Although this data is promising, prospective studies are needed to compare efficacy at stopping neonatal seizures.

Phenobarbital, a gamma aminobutyric acid (GABA) receptor agonist, has been associated with apoptotic neurodegeneration in infant rats at plasma concentrations between 25 and 35 μg/L. These concentrations are routinely achieved in humans with neonatal seizures.¹⁰ The long-term effects of phenobarbital on neurodevelopment are heavily debated. Antenatal exposure to phenobarbital before delivery was found not to impact neurodevelopmental outcomes compared to unexposed infants at 36 months of life.¹¹ A study measuring developmental delay in children with neonatal seizures using the Functional Independence Measure for Children score (WeeFIM) found similar degrees of delay in children sent home from the neonatal ICU on phenobarbital (58%) compared to children sent home without (52%).¹² Levetiracetam is a second generation antiseizure that targets the SV2 protein of the synaptic vesicle fusion complex, inhibiting neurotransmitter release. It is structurally related to the neuroprotective pyrrolidone compound piracetam. In rat models, it was found to reduce ischemic damage and suppress non-convulsive seizures.¹³ A retrospective study of infants who received antiseizure medications for neonatal seizures showed increased exposure to phenobarbital is associated with worse neurodevelopmental outcomes than levetiracetam at their 24 month assessment.¹⁴ However, many infants in this study were exposed to both medications and the severity of epilepsy after discharge from the NICU was not established.

Infants with brain injury are susceptible to developmental delays. Survivors of neonatal seizures are at risk for developmental delay, epilepsy and further neurological co-morbidities.¹⁵ Despite advances in neonatal critical care, the prevalence of adverse long-term outcomes and seizure recurrence remains unchanged.³ Our goal is to determine if early treatment of neonatal seizures with phenobarbital or levetiracetam is associated with worse neurodevelopmental outcomes in brain injured infants. Our study’s null hypothesis is the choice of initial treatment with phenobarbital or levetiracetam does not influence neurodevelopmental outcomes.

Methods

Study Design and Subject Enrollment.

This is a retrospective case controlled cohort design approved by the Institutional Review Board of the University of Florida (UF). We reviewed the medical records of all patients admitted to the Neonatal Intensive Care Unit of UF Health Shands Children’s Hospital from January 1, 2013 through September 1, 2016 who were assessed through the Early Developmental Assessment Clinic (EDAC) based in the Division of Pediatric Neurology at the UF College of Medicine. Patients were included in the brain injury with neonatal seizure group if they were assigned the ICD-9 code diagnosis of neonatal seizures (779.0), epilepsy (345.0-9), or other convulsions (780.39). Infants whose charts were coded with ICD-10 were included using the following codes: convulsions in the newborn (P90) or other convulsions (R56.9). All remaining charts were assigned to the brain injury group.

Inclusion/Exclusion Criteria.

Each chart was manually reviewed to screen for inclusion/exclusion criteria. All infants assessed in this study had brain injury. This included neonates who suffered intraventricular hemorrhage, ischemic and hemorrhagic stroke or some form of encephalopathy (anoxic brain injury, hypoxic ischemic encephalopathy, cardiopulmonary arrest, renal or metabolic encephalopathy) in the neonatal intensive care unit. Infants were included in the brain injury with neonatal seizure group if the chart review confirmed a diagnosis of neonatal seizures, and if the initial seizure treatment consisted of phenobarbital and/or levetiracetam. Exclusion criteria for the brain injury with neonatal seizure group included early infantile epileptic encephalopathies, congenital cerebral abnormalities, channelopathies, inborn errors of metabolism, initial treatment with any other antiseizure medication, hypsarrythmia on EEG, and/or a burst suppression pattern on EEG for term infants. Brain injury group without neonatal seizures included subjects with the same inclusion/exclusion criteria as above and were not exposed to antiseizure medications (phenobarbital or levetiracetam). Matched controls without risk factors for developmental delay were not obtained. This study reflects our experience with neonatal brain injury.

Seizure Characterization and Treatment.

EEG leads were placed using the international 10-20 system of electrode placement with silver chloride EEG electrodes. Seizures were identified as clinical only, clinical with electrographic correlate, or electrographic only. If clinical seizures were present, a continuous video EEG was obtained for 24-72 hours to evaluate for significant interictal discharges or electrographic seizures. All seizures (clinical, clinical with electrographic correlate or electrographic) were treated with either phenobarbital 20mg/kg IV bolus followed by maintenance dosing of 5mg/kg/day divided every 12 hours or levetiracetam 20-30mg/kg IV bolus followed by maintenance dosing of 20-30mg/kg/day divided every 12 hours. If seizures continued despite treatment with either antiseizure drug, additional boluses of phenobarbital or levetiracetam were administered. For the purposes of this study, if a child received both medications in the process of treating a seizure, the medication the child was on the longest was deemed their primary antiseizure medication. Patients were weaned off antiseizure medications at the time of discharge or soon after if they remained seizure free. This was done at the discretion of the neonatologist or pediatric neurologist.

Seizure and Neurodevelopmental Outcomes.

After discharge, patients were monitored for ongoing seizures by a board-certified pediatric neurologist (SG) in the EDAC. The presence of ongoing seizures was determined by parental reporting. If an infant was discharged on antiseizure medications and became seizure free, he/she was weaned off the antiseizure drug by 6 months of life. Infants were evaluated for developmental delays in the EDAC by board certified attending pediatric neurologists and an occupational therapist trained to administer the Bayley Scales of Infant and Toddler Development®, Third Edition (Bayley-III®). Evaluations of premature infants were corrected for gestational age and were performed between 9 and 14 month gestation age. Developmental delay was defined as a Bayley-III composite score two standard deviations below the mean. Composite scores for motor and language as well as the scaled subset scores in each domain (cognitive, fine motor, gross motor, expressive language and receptive language) were used for analysis. Only infants who were seen in follow up were included in the analysis.

Data Management.

Study data were collected and managed using REDCap electronic data capture tools hosted at University of Florida.¹⁶ REDCap (Research Electronic Data Capture) is a secure, web-based application designed to support data capture for research studies, providing 1) an intuitive interface for validated data entry; 2) audit trails for tracking data manipulation and export procedures; 3) automated export procedures for seamless data downloads to common statistical packages; and 4) procedures for importing data from external sources.

Statistical Analysis.

The statistical analysis of the data was performed by using the open source statistical software package R (version 3.3.3) with different statistical analysis methods including independent t test and the non-parametric Mann-Whitney U test. Specifically, in the descriptive analysis of continuous variables of demographics data, an independent t test was used. Categorical demographic data is reported as total number and percentage. Two sample comparisons for the 9-14 months’ group were conducted with a Mann-Whitney U test. A p-value less than 0.05 was regarded as significant in hypothesis testing.

Results

Patient Characteristics.

Figure 1 outlines the eligibility and selection for analysis of subjects for this study. One hundred and eighty-seven patients met eligibility criteria. A total of 119 patients were enrolled based on the inclusion criteria and availability of data. Eighty-seven infants with brain injury without seizures were included in the control group. Fifty-three infants in this group completed follow up at 9-14 months corrected gestational age. Thirty-two infants with brain injury and seizures were initially included in the neonatal seizure group. Twenty-two of these patients completed follow-up at 9-14 corrected months gestational age. Analysis of baseline characteristics (Table 1) indicates that subjects in the neonatal seizure group are more likely to have a lower Apgar score at 5 minutes, lower pH and high base deficits on venous or cord blood gas. Seizure characteristics such as etiology, number of clinical seizures, number of electrographic seizures, total duration of seizures in minutes, medications utilized and number of seizures after discharge are shown in Table 2. Maximal seizure burden was calculated as the maximal hourly seizure burden within continuous EEG recording and is measured in minutes per hour.¹⁷ There were no significant hospitalizations or surgeries after discharge in patients seen at 9-14 months. Baseline characteristics were similar between the levetiracetam and phenobarbital patients.

Figure 1:

Flow Diagram

Table 1:

Maternal and Neonatal Characteristics

	Brain Injury with neonatal seizure (mean ± SD)	Brain Injury without neonatal seizure (mean ± SD)	P value*
Number of patients	22	53
Characteristics Maternal
Age	27.8 ± 6.9	28.0 ± 5.4
Gravida	2.7 ± 4.2	2.8 ± 1.9
Para	2.0 ± 2.0	1.7 ± 1.2
Caucasian -no. (%)	10 (52)	28 (53)
Neonatal
Gestational age	35.2 ± 6.4	32.5 ± 5.6
Female -no. (%)	10 (45)	23 (43)
Apgar at 5 minutes	4.0 ± 2.6	7.7± 1.8	<0.01
Cord or VBG pH	7.2 ± 0.18	7.3 ± 0.13	<0.01
Base deficit –mmol/liter	12.4 ± 7.8	5.0 ± 5.8	<0.01
Birth weight (grams)	2418± 1514	1827 ±1048
Intubation (%)	19 (86)	36 (68)
Ventilation (%)	19 (86)	47 (89)
Length of stay	56.5 ± 58.1	56.8 ± 40.5
Type of Brain Injury (%)
Hypoxic ischemic encephalopathy/Anoxic brain	14 (64)	22 (42)
injury
Intraventricular hemorrhage (grade III-IV)	5 (23)	18 (34)
Ischemic/Hemorrhagic stroke	2 (9)	6 (11)
Other causes of encephalopathy (metabolic, renal, sepsis, hypovolemic shock)	2 (9)	10 (19)

^*Only significant p values are reported

Table 2:

Seizure Characteristics

	Brain Injury with neonatal seizure
Number of patients	22
Clinical seizures per patient (median, IQR)	1 (1-2.5)
Total minute duration of clinical seizures (median, IQR)	0.5 (0.2-4)
Electrographic seizures per patient (median, IQR)	2 (1.75-4)
Total minute duration of electrographic seizures (median, IQR)	4 (0.42-15.75)
Total minute duration of clinical and electrographic seizures (median, IQR)	4.17 (0.75-15.5)
Maximal seizure burden in minutes/hour (median, IQR)	0.12 (0.02-0.46)
Etiology
Hypoxic ischemic encephalopathy	13
Infectious	2
Hemorrhage	5
Unknown	2
First anti-seizure drug
Phenobarbital	15
Levetiracetam	7
Not responding to 1st drug	2
Second anti-seizure drug
Phenobarbital	1
Levetiracetam	1
Discharged with anti-seizure drug	10
Phenobarbital	10
Levetiracetam	0
Seizures at 6 months	2
Anti-seizure drug at 6 months	4
Seizures at 12 months	1
Anti-seizure drug at 12 months	1
Phenobarbital	1
Levetiracetam	0

IQR = interquartile range

Neurodevelopmental outcomes at 9-14 months.

Significant differences are present at 9-14 months between the brain injury without neonatal seizure group and brain injury with neonatal seizure group in language and cognition scores with the neonatal seizure group having significantly worse scores. Table 3 show the divergence between the groups. The composite cognitive scores were lower for the brain injury with neonatal seizure group with a median score of 77.5 compared to the brain injury group with a median score 90 (p =0.002). Similar differences were found in the composite language scores for the brain injury with neonatal seizure group with a median score of 86 compared to the brain injury without neonatal seizure group mean score of 100 (p = 0.01). There was no significant difference in scores between patients exposed to phenobarbital or levetiracetam, as shown in Table 4.

Table 3:

Bayley scores at 9-14 months gestational age comparing infants with “brain injury without neonatal seizures” to infants with “brain injury and neonatal seizure.”

Bayley Scores for 9-14 months gestational age
	Brain Injury with neonatal seizure	Brain Injury without neonatal seizures	P value
Composite score cognitive	77.5 (57.5-85)	90 (80-95)	0.001
Composite score language	86 (77-99.25)	100 (86-106)	0.012
Composite score motor	74.5 (47.5-88)	82 (64-91.75)	0.066
Scaled score cognitive	6 (3-7.75)	8 (6-9)	0.012
Scaled score receptive language	7 (5.25-9)	9 (7-10.25)	0.043
Scaled score expressive language	8 (6-10)	10 (9-12)	0.007
Scaled score fine motor	6 (3-8)	7 (5-8)	0.364
Scale score gross motor	3.5 (1-9)	7 (2-9)	0.252

All numbers are presented at median (interquartile range)

Table 4:

Bayley scores at 9-14 months gestational age of brain injury patients with neonatal seizures treated with phenobarbital versus levetiracetam

Bayley Scores for 9-14 months gestational age	Median (Interquartile range)		P value
	Phenobarbital	Levetiracetam
Composite score cognitive	70 (60-83.75)	85 (65-90)	0.530
Composite score language	86 (77-97)	86 (72.5-94.5)	1
Composite score motor	61 (49-92.5)	82 (58-86.5)	1
Scaled score cognitive	6 (3-7.75)	7 (3-8)	0.733
Scaled score receptive language	7 (5.5-9.75)	8 (5.5-8.5)	0.536
Scaled score expressive language	8 (6.5-9.75)	8 (5.5-10.5)	0.911
Scaled score fine motor	7 (3-9)	6 (2.5-6.5)	0.462
Scale score gross motor	4 (1-9.75)	3 (1-6)	0.953

Discussion

In our cohort of neonates with brain injury, we found that patients with seizures had more unfavorable cognitive and language outcomes compared to patients without seizures. Our population had a low seizure burden and no patient was in status epilepticus. The initial choice of antiseizure medication, phenobarbital or levetiracetam, was not associated with worse long-term developmental outcomes.

Several studies have shown the effects of neonatal seizures on brain injury and long-term outcomes. A prospective randomized control trial evaluated treatment of clinical and subclinical seizure identified by amplitude EEG versus treatment of clinical seizures without amplitude EEG. There was a correlation between duration of seizures and worsening changes on MRI of the brain. Patients in this study were full term neonates with hypoxic ischemic encephalopathy. A similar study using continuous EEG monitoring for neonates with hypoxic ischemic encephalopathy found increasing seizure burden correlated significantly with lower performance scores in cognition, language and motor skills on the Bayley III.¹⁸ A recent population based retrospective study based on claims data in the United States analyzed a cohort from 2009 and followed them up to 6 years. Of the 800 subjects with neonatal seizures in a cohort of 490,071 patients, 39.3% of them had some form on intellectual disability and 47.4% developed epilepsy. Data was based on claims reporting. Etiology for neonatal seizures in this study included other forms of brain injury besides, hypoxic ischemic encephalopathy such as perinatal complications, congenital infections, brain malformations, metabolic disturbances, cerebrovascular disease and hydrocephalus.¹⁹ Our data supports these prior studies and shows a correlation between neonatal seizures and worsening cognitive and language outcomes assessed by the Bayley III in neonatal brain injury patients from a variety of etiologies.

Our institution adopted a treatment algorithm allowing the neurologist or neonatologist to use either drug at a prescribed dose. Providers can choice either phenobarbital 20mg/kg IV bolus followed by maintenance dosing of 5mg/kg/day divided every 12 hours or levetiracetam 30-60mg/kg IV bolus followed by maintenance dosing of 30mg/kg/day divided every 12 hours. We recently changed our algorithm to reflect a recent study⁹ and findings at our institution showing seizure cessation at higher doses of levetiracetam in neonates with hypoxic ischemic encephalopathy. In a retrospective study treating neonatal seizures from various etiologies found levetiracetam was associated with seizure improvement in 8 of 23 neonates.⁸ High dose levetiracetam was found to have no effect on hemodynamics with high serum levels during rapid infusion.²⁰ Levetiracetam may be used as an alternative to phenobarbital for first line treatment in neonatal seizures. Large prospective comparison studies are needed to determine benefits of high dose levetiracetam on neonatal seizures and neurodevelopment.

A recent study compared the effectiveness of levetiracetam and phenobarbital for the treatment of seizures during infancy. Children treated with levetiracetam were free from monotherapy failure compared to phenobarbital. The findings suggest levetiracetam is superior to phenobarbital for infant seizures. Neonatal seizures were not included in this analysis. The choice of antiseizure medication was not randomized in the study. A higher proportion of infants in the phenobarbital group had developmental structural brain abnormalities.³⁰ Our study focuses on neonatal seizures in a population with brain injury and suggests no difference in treatment failure or neurodevelopmental outcomes for these infants.

The study is based on our experience with antiseizure medications in a population at risk for developmental delays. A major limitation of this study is not having matched groups for comparison. Thus, it is difficult to determine the effects of the individual medication on normal developmental in children seen in in the NICU. This study does show a trend towards worsening outcomes for our study population. Matched control studies would be needed to further test this hypothesis. There are likely many confounding variables that and biases within our study due to the small sample size, retrospective nature of the study and lack of randomization. Our data suggests that the presence of neonatal seizures may be a more significant risk factor for neurodevelopmental impairments than the selection between two commonly used antiseizure medications. However, a future prospective study could be more definitive regarding this point. The somewhat high dropout rate between eligible subjects and those who underwent analysis was due to multiple reasons including availability of follow-up data on some patients. This led to a smaller number of subjects in the neonatal seizure group than anticipated, representing a limitation of this study. It would be helpful to continue following this cohort over time to assess longer-term developmental outcomes.

Conclusion.

Brain injury infants with neonatal seizures had worsening of neurodevelopmental outcomes at 9-14 months gestational age irrespective of the anti-seizure medication used. Further studies are needed to determine whether the cause of this neurodevelopmental delay is an underlying etiology or the neonatal seizures themselves, and whether an alternative anti-seizure medication could be associated with better outcomes.