Main

The incidence of neonatal encephalopathy (NE) is about 1 to 2.5 per 1000 live term births, mostly as a result of perinatal asphyxia (1–3). NE is a significant cause of neonatal death and adverse neurodevelopmental outcome such as cerebral palsy, developmental delay, and epilepsy. The reported rate of epilepsy after NE due to hypoxic-ischemic encephalopathy ranges from 9 to 33% (4–6), and in one study, children with a history of hypoxic-ischemic encephalopathy had five times the risk of developing epilepsy when compared with those without (7).

Although children with NE are known to be at risk for epilepsy, there are few studies examining the neonatal risk factors for developing seizures beyond the newborn period. Past studies in small cohorts have examined severity of encephalopathy, neonatal seizures, and MRI as risk factors with conflicting results. Pisani et al. (4) found that severe (but not moderate) encephalopathy was associated with later epilepsy, whereas van Kooij et al. (6) found a 10% prevalence of epilepsy among a cohort of children who suffered moderate encephalopathy. Neonatal seizures were a risk factor in one study, although the effect was not significant after adjusting for the degree of encephalopathy (4). MRI injury is also risk factor for epilepsy (5,6). Small cohort size, inadequate neonatal imaging, or lack of long-term outcome data limit these studies, and counseling for parents of children with NE remains a challenge.

The objective of this study was to evaluate the neonatal clinical and imaging risk factors for childhood epilepsy in a single center cohort of children with a history of NE. All children were imaged using high-resolution MRI according to a standardized research protocol in the newborn period and evaluated longitudinally through childhood.

MATERIALS AND METHODS

This is a longitudinal cohort study of neonates who were admitted to the Intensive Care Nursery at the University of California San Francisco. Newborns met inclusion criteria of GA ≥36 wk at birth and any one of the following: umbilical cord arterial blood pH <7.1, umbilical cord arterial blood base excess >−10, or 5-min Apgar score ≤5. These broad inclusion criteria were chosen to encompass newborns with a wide range of injury and neurodevelopmental outcome and have been used in previous publications by our group (8–14). Newborns were excluded if there was evidence of intrauterine or perinatal infection, major anomalies of the brain or other major organ system, or evidence of congenital metabolic disease.

From December 1993 to June 2009, 234 infants ≥36 wk GA at birth were enrolled and imaged according to protocol. Thirty-one children were excluded for the following reasons: deceased during the birth admission (17 children) or withdrew from the study (14 children), leaving 203 subjects.

Clinical data.

Trained neonatal research nurses extracted birth delivery data including GA at birth, birth weight, and Apgar scores from medical records. Clinical seizures (spells identified by the attending physician as seizures and treated using anticonvulsant medication), electrographic seizures (events identified by the attending neurophysiologist as seizures), and status epilepticus (recurrent electrographic seizures considered by the neurophysiologist as “status epilepticus,” “continuous,” or “near continuous” seizures) were determined from chart review. Before 2008, video-EEG monitoring was performed at the discretion of the attending neurologist and neonatologist and performed as standard of care in all children with clinical seizures (minimum duration 30 min). Since 2008, video-EEG has been performed for all newborns treated with therapeutic hypothermia from the time of admission until 6–12 h after rewarming and at least 24 h after the last recorded EEG seizure. A pediatric neurologist measured the degree of encephalopathy in the first 3 d of life on a scale of 0–6, which considers feeding, alertness, tone, respiratory status, reflexes, and seizures (12).

MRI.

All neonates were imaged with MRI using a specialized neonatal head coil on a 1.5-Tesla Signa EchoSpeed system (GE Medical Systems) and using imaging sequences optimized for the neonatal brain (14,15). In cases where more than one MRI was performed during the hospital stay, the MRI closest to day of life 3 to 5 was used for analysis. A pediatric neuroradiologist blinded to the neonatal course prospectively scored the imaging as previously described (8). Injury was scored using a system that is strongly predictive of neurodevelopmental outcome after NE (8). The pattern of injury was described as “basal ganglia/thalamus predominant,” “watershed predominant,” or “normal” (11). We defined “mild-moderate injury” as a basal ganglia/thalamus score of 1 or 2 (abnormal signal in the thalamus and/or lentiform nucleus) or watershed score of 1–4 (abnormal signal in the anterior and/or posterior watershed zones), and “severe injury or near-total injury” as a basal ganglia/thalamus score ≥3 or a watershed score of 5 (more extensive involvement in either territory).

Epilepsy interview.

The parents of enrolled subjects were contacted by telephone in June 2010, and administered a structured seizure questionnaire that was developed for the purpose of this study. For the children with postneonatal seizures, parents were asked to provide information regarding age of seizure onset, frequency, and medication use. Epilepsy was defined as recurrent, unprovoked seizures, or a single seizure in the setting of abnormal EEG and initiation of medication. For the children who died after hospital discharge, the medical records were reviewed to complete the standardized seizure questionnaire. Epilepsy was graded according to modified Engel classification (class 0 = seizure free and off seizure medications for at least 6 mo; class 1 = seizure free for at least 6 mo while on medication or seizure-free off medication for fewer than 6 mo; class 2 = fewer than one seizure per month on medication; class 3 = one to four seizures a month on medication; class 4 = five to thirty seizures per month on medication; class 5 = thirty or more seizures a month) (16).

Neurodevelopmental follow-up.

A pediatric neurologist and a developmental psychologist who were blinded to the neonatal course examined the children at ages 3–6 mo, 1 y, 2.5 y, 4 y, and 8 y. Neuromotor function was evaluated using the neuromotor score, a 5-point scale where 0 is normal and 5 is spastic quadriparesis (8). The neuromotor score was classified as follows: 0 or 1 normal, 2 borderline, ≥3 abnormal. An age-appropriate neuropsychological test was administered to the child at each follow-up appointment. The Bayley Scales of Infant Development were used to test children at 1 y and 2.5 y (Bayley-II before 2008) (17,18). At the age of 4, the Wechsler Preschool and Primary Scale of Intelligence-Revised (WPPSI-R) was used (19,20). The Wechsler Intelligence Scale for Children IV (WISC-IV) (21) was administered for the 8-y follow-up. Scores from standardized tests (all with a mean of 100 and a SD of 15) were used to classify cognitive outcomes as follows: ≥85 on all subscales was considered normal, 70–84 on one or more subscales was considered borderline, and <70 (i.e. 2 SDs below the mean) on one or more subscales was considered abnormal. Comorbidities and most recent neurological and neurodevelopmental examinations were extracted from medical records and the structured seizure questionnaire.

Analysis.

Data analyses were conducted using statistical software Stata 10 (StataCorp LP, College Station, TX). The association of clinical predictors with time to onset of epilepsy was assessed using Cox proportional hazards regression and graphically displayed using Kaplan-Meier curves. Data were censored at the time of onset of epilepsy or at the time of the telephone interview in children without epilepsy. Differences between children with and without follow-up were assessed using two-tailed t test for continuous variables, Wilcoxon rank sum for nonparametric data, and χ2 or Fisher's exact test for categorical variables. Multivariable model included those predictors with p ≤ 0.1 in the univariable analysis. The p values <0.05 were considered significant.

The Committee on Human Research at the University of California, San Francisco, approved the protocol. Infants were studied only after informed voluntary parental consent.

RESULTS

Of the 203 subjects included in the study, we successfully contacted 127 families by telephone and were able to complete adequately adjudicate outcome based on medical records for 2 deceased children, for an overall follow-up rate of 64%. The study subjects are presented in Table 1. The children whose parents were not available to complete the survey were similar to those whose parents did complete the survey with no differences in sex, GA at birth, birth weight, cord gas pH or base excess, NE score, and MRI injury scores. However, those whose parents responded were significantly more likely to have had neonatal seizures (40% versus 25%, p = 0.04) and have been enrolled in the study more recently (median year of enrollment 2004 versus 1997, p < 0.00005). The median age of surviving study participants at the time of the follow-up survey was 73 mo (6 y; range, 12 mo to 16.5 y).

Table 1 General clinical characteristics among 129 children with neonatal encephalopathy
Full size table

Epilepsy features and outcome.

Of the 129 evaluated subjects, 13 (10%) developed epilepsy, 7 had febrile seizures, and 1 child had a single spell at age of 2 mo that was not clearly a seizure. The age of epilepsy onset ranged from 0 to 5 y; 7 (54%) had onset within the first year of life. Eleven children had abnormal neurodevelopmental outcome, one had borderline outcome and one was not assessed. The clinical characteristics of the 13 study subjects with epilepsy are presented in Table 2.

Table 2 Clinical seizure, MRI characteristics, and neurodevelopmental outcome for 13 children with neonatal encephalopathy and childhood epilepsy
Full size table

Perinatal risk factors for epilepsy.

Degree of encephalopathy was the only clinical feature associated with epilepsy (Table 3).

Table 3 Incidence rate and hazard ratios for epilepsy risk factors among 129 children with neonatal encephalopathy
Full size table

Neonatal seizures and risk of epilepsy.

All 13 children with epilepsy had a history of clinical and/or electrographic neonatal seizures. The frequency of epilepsy among the 52 children with a history of neonatal seizures was 25% (or 15.7 per 1000 person years). Among children with neonatal seizures, those with status epilepticus were most likely to develop epilepsy (5/6, 83% or 1237.1/1000 person-years). The single child who had not developed epilepsy was 21 mo at the time of the evaluation and had been treated with therapeutic hypothermia. Those children without status, but whose neonatal seizures EEG confirmed, were more likely to develop epilepsy when compared with those who had only clinical seizures [5/19, 26% or 93.5/1000 person-years versus 3/28, 11% or 14.5/1000 person-years; hazard ratio (HR), 4.0; 95% CI, 0.9–17.3]. The epilepsy-free survival data in those children with and without a history of neonatal seizures are presented in Figure 1.

Figure 1
figure 1

Kaplan-Meier estimates for epilepsy-free survival among 129 newborns with NE, with and without neonatal seizures. No seizures, solid line (n = 76); clinical seizures only, dashed line (n = 28); EEG confirmed seizures, dotted line (n = 19); status epilepticus, dash-dot line (n = 6).

Full size image

Brain injury and epilepsy.

All children with epilepsy had injury on neonatal MRI, and the severity of injury was associated with epilepsy. The children with severe or near-total brain injury were more likely to develop epilepsy (8/22, 36% or 54.7/1000 person-years) versus 5/65 (8% or 11.5/1000 person-years) for those with mild or moderate injury. Of the 16 children with severe or near-total brain injury and neonatal seizures, half developed epilepsy. Pattern of injury was also important: of the 30 children with basal ganglia/thalamus predominant pattern of injury, six (20% or 32.3/1000 person-years) developed epilepsy, whereas only seven (12%, 17.7/1000 person years) with watershed predominant pattern of injury developed epilepsy (p = 0.005). When compared with those children with mild/moderate brain injury, those with severe or near-total injury had a higher rate of epilepsy (HR, 5.5; 95% CI, 1.8–16.8; p = 0.003). Epilepsy-free survival by severity of brain injury is presented in Figure 2.

Figure 2
figure 2

Kaplan-Meier estimates for epilepsy-free survival among 129 newborns with NE, with and without brain injury. No injury, solid line (n = 42); mild/moderate injury, dashed line (n = 65); severe/near-total injury, dotted line (n = 22).

Full size image

Adjusted analysis.

In a multivariable analysis among those with moderate or severe encephalopathy, neonatal seizures and MRI injury, adjusting for NE, status epilepticus remained a highly significant risk factor for epilepsy (HR, 17.3; 95% CI, 2.7–110.0; p = 0.003), whereas severe/near-total injury (HR, 2.4; 95% CI, 0.7–8.4; p = 0.2) was not (Table 4).

Table 4 Adjusted risk of epilepsy among those newborns with MRI injury, neonatal seizures, and moderate or severe neonatal encephalopathy (N = 42)
Full size table

DISCUSSION

In this longitudinal sample of 129 newborns with NE, neonatal seizures and brain injury on MRI were strong risk factors for epilepsy, and the children with epilepsy all had adverse neurodevelopmental outcome.

These results support previous studies suggesting that epilepsy occurs in children with a more severe spectrum of NE. The 25% frequency of epilepsy after neonatal seizures is similar to previous contemporary cohorts that used broad clinical and/or electrographic definitions for neonatal seizures and lower than studies that required EEG confirmation of seizures (22–25). Although duration of follow-up is variable across studies, highest risk within the first year of life is a common feature, as is the high rate of associated neurodevelopmental disabilities (23,25–28).

Animal studies have begun to elucidate mechanisms by which neonatal seizures can induce persistent enhanced neocortical excitability (29). Our data also support evidence that severity of seizures is important, which is in keeping with studies by Clancy and Legido (22), who showed in a mixed cohort that a higher burden of neonatal seizures was a risk factor for epilepsy, and Pisani et al., (30) who showed a higher rate of epilepsy after status epilepticus when compared with recurrent seizures.

Severity and pattern of MRI injury are known risk factors for neonatal seizures (6,11), and we also show a clear relationship with epilepsy. Children with severe basal ganglia/thalamus pattern of injury and cortical involvement were at highest risk of developing epilepsy. There is controversy from animal studies regarding whether neuronal injury is required for acquired epileptogenesis in the immature brain, or whether seizures alone without neuronal injury can cause changes in excitability sufficient to result in unprovoked seizures beyond the neonatal period (31). Our finding, that epilepsy occurred only in those children with apparent brain injury on MRI, supports, but does not prove, the hypothesis that neuronal injury is, in fact, required to develop epilepsy.

Although our findings do not support a reduced risk of epilepsy in the cooled subjects, the data we present do not exclude this possibility. Since fall 2007, we have treated most newborns at risk for hypoxic-ischemic injury and moderate-severe encephalopathy with therapeutic hypothermia using whole body cooling. Therapeutic hypothermia, which has been shown to reduce death and disability at 18–22 mo in several randomized, controlled trials (32–35) also has an antiseizure effect in animal models (36,37), although its effect on seizures in human newborns is not clear. Among newborns treated with therapeutic hypothermia at our center, electrographic seizures are present in >30%. If neonatal seizures are suppressed in human newborns treated with hypothermia and neonatal seizures enhance long-term epileptogenesis, the risk of epilepsy in the cooled population may ultimately be reduced. The question of whether hypothermia reduces neonatal seizure burden and/or epileptogenesis should be further explored in larger data sets.

Although this is a large cohort with high-quality neonatal imaging and long-term follow-up, the data are limited in several ways. First, the survey response rate was only 64%. This limitation may have led to an overestimate of the prevalence of epilepsy, because the rate of neonatal seizures was higher in those children whose parents responded to the interview. However, response rate should not affect the relationships between perinatal risk factors, including seizures and MRI findings, and epilepsy. Second, it is possible that there was bias toward diagnosis of epilepsy in some cases of children with ambiguous spells and severe developmental impairment. This may have falsely increased the relationship between MRI injury and neonatal seizures and epilepsy. Third, since the study's inception in 1993, we have changed the guidelines for monitoring and treating newborns with encephalopathy. Before 2008, video-EEG was used at the discretion of the attending neurologist, usually for 30–60 min routine recordings. Since 2008, we have monitored all children with NE using continuous video-EEG for the duration of hypothermia and rewarming. To understand the true relationship between clinical versus electrographic seizures and epilepsy, we will need to examine a larger cohort of subjects with long-term conventional video-EEG monitoring.

CONCLUSIONS

Epilepsy is an outcome following brain injury and neonatal seizures in the setting of severe hypoxic-ischemic brain injury. These data add to our understanding of the pathogenesis of epilepsy after seizures in the newborn period and provide information for clinicians and parents planning long-term care for children with NE. For term infants with NE, but without both neonatal seizures and brain injury on MRI, parents can be reassured that the child is unlikely to develop epilepsy. Children with electrographic seizures (especially status epilepticus) and severe brain injury have a high risk of adverse neurodevelopmental outcome, including epilepsy. This high-risk population will be important for future studies examining neuroprotective and antiepileptic agents. Neonatal seizures and status epilepticus are potentially modifiable risk factors, and there is urgent need for studies to examine whether early monitoring and treatment will improve outcome.