The temporal evolution of electrographic seizure burden in

neonatal hypoxic ischemic encephalopathy

*yNiamh E. Lynch, yNathan J. Stevenson, yVicki Livingstone, *yBrendan P. Murphy,
zJanet M. Rennie, and *yGeraldine B. Boylan
*Department of Paediatrics and Child Health, University College Cork, Cork, Ireland; yNeonatal Brain Research Group, University
College Cork, Cork, Ireland; and zElizabeth Garrett Anderson Institute for Womens Health, University College London Hospitals,
London, United Kingdom
SUMMARY
Purpose: Hypoxic ischemic encephalopathy (HIE)
accounts for 60% of all neonatal seizures. There is emerg-
ing evidence that seizures cause additional injury to the
developing brain that has sustained hypoxic ischemic
injury. Temporal evolution of clinical seizure burden in
HIE has been characterized, with maximum clinical sei-
zure burden (the period of maximum seizure activity)
being observed between 12 and 24 h of age. The purpose
of our study was to investigate the distribution of electro-
graphic seizure burden (the accumulated duration of sei-
zures over a dened time period), following the initial
hypoxic ischemic insult.
Methods: Fifteen full-term newborns with HIE and sei-
zures, and a minimum of 48 h of continuous video
electroencephalography (EEG), were included in this
retrospective study. Medical records of the infants
were reviewed and details of clinical seizures and an-
tiepileptic drugs were recorded. The time of maxi-
mum seizure burden was dened as the midpoint of
an hour-long window, shifted in time by 1 s across the
full EEG recording, which contained the maximum
duration of seizures. The degree of temporal evolution
of seizure burden within this period was tested. Tem-
poral evolution was further analyzed by segmenting
the time series into two periods; the time between
the rst recorded seizure and the maximum seizure
burden (T
1
), and the time between the maximum sei-
zure burden and the last recorded seizure (T
2
). Sei-
zure burden, duration, and number of seizures per
hour were analyzed within each time period.
Key Findings: EEG was commenced at a median of 14 h of
age. Maximum electrographic seizure burden was
reached at a median age of 22.7 h. Time from rst
recorded seizure to maximum seizure burden (T
1
) was
signicantly shorter than time from maximum seizure
burden to last recorded seizure (T
2
) (p-value = 0.01).
Median seizure burden during T
1
was signicantly higher
than during T
2
(p-value = 0.007). There is temporal evolu-
tion of electrographic seizure burden in full-term new-
borns with HIE. There is a short period of high seizure
burden (T
1
) followed by a longer period of lower seizure
burden (T
2
).
Signicance: Understanding the temporal evolution of
seizure burden in HIE contributes further to our under-
standing of neonatal seizures, helps identify an optimal
therapeutic window for seizure treatment, and provides a
benchmark against which to measure the efcacy of new
and innovative forms of neuroprotection and antiepileptic
medication.
KEY WORDS: Newborn, Seizures, Seizure burden, Sei-
zure distribution, Temporal evolution, Hypoxic ischemic
encephalopathy, Electroencephalogram.
Peripartum asphyxia affects approximately 35 per 1,000
live births, with approximately 0.51 per 1,000 infants
subsequently developing symptoms of moderate or severe
hypoxic ischemic encephalopathy (HIE) (Levene et al.,
1985). The clinical manifestations of HIE evolve over time,
and reflect the evolution of the underlying brain injury.
Within the brain, following the initial hypoxic ischemic
insult there is an immediate decline in the level of adenosine
triphosphate (ATP), failure of the Na/K pump, cytoplasmic
calcium accumulation, and cell death (Brillault et al.,
2008). Activation of various proteins, and inflammatory
mechanisms, leads to further cell death through necrosis and
apoptosis in the following days and weeks (Volpe, 2008a).
The newborn brain is hyperexcitable, a feature important
for synaptogenesis (Sarnat et al., 2010). This hyperexcit-
ability predisposes the developing brain to seizures (Wirrell,
2005). c-Aminobutyric acid (GABA) is the predominant
inhibitory neurotransmitter in the brain, but it is paradoxi-
cally excitatory in the newborn brain. This is because
Accepted December 19, 2011; Early View publication February 6,
2012.
Address correspondence to Prof. Geraldine Boylan, Neonatal Brain
Research Group, Ground Floor, Cork University Maternity Hospital,
Wilton, Cork, Ireland. E-mail: g.boylan@ucc.ie
Wiley Periodicals, Inc.
2012 International League Against Epilepsy
Epilepsia, 53(3):549557, 2012
doi: 10.1111/j.1528-1167.2011.03401.x
FULL-LENGTHORIGINAL RESEARCH
549
activation of GABA
A
receptors results in membrane depo-
larization as opposed to hyperpolarization seen in mature
GABAergic synapses (Ben-Ari et al., 1997). This is as
a result of elevated intracellular Cl
)
, facilitated by the
Na
+
-K
+
-2Cl
)
cotransporter (NKCC1) (Dzhala et al., 2005).
Animal studies show that seizures in addition to a hypoxic
ischemic injury lead to depletion of high energy phosphates
and induce high levels of glutamate (Yager et al., 2002).
There is growing evidence that seizures in human newborns
with HIE exacerbate the initial hypoxic ischemic injury
(Miller et al., 2002; Glass et al., 2009; Ancora et al., 2010).
HIE grade reflects the degree of severity of the hypoxic
ischemic injury, and is a predictor of long-term outcome.
Newborns with grade 1 (mild) HIE are irritable and hyper
alert. They do not develop seizures. Newborns with grade 2
(moderate) and grade 3 (severe) HIE may develop seizures
(Sarnat & Sarnat, 1976; Levene et al., 1985). HIE accounts
for approximately 60% of all neonatal seizures (Levene &
Trounce, 1986). The evolution of clinical seizures over time,
the temporal evolution (TE), in HIE has long been recog-
nized. Rose & Lombroso (1970) described a peak incidence
of clinical seizures in the first 24 h of age in newborns with
presumed anoxic injury. Volpe (2008b) described clinical
seizures in HIE as becoming more severe and frequent
between 12 and 24 h of life, with status epilepticus often
being observed, and seizure cessation by 72 h of age.
Electrographic neonatal seizures have been defined as
clear ictal events characterized by the appearance of sudden,
repetitive, evolving, stereotyped waveforms that have a def-
inite beginning, middle, and end, and last for a minimum of
10 s (Clancy & Legido, 1987). It is recognized that neonatal
seizures can be difficult to diagnose using clinical features
alone in term neonates (Mizrahi & Kellaway, 1987; Scher
et al., 1993; Murray et al., 2008a). Amplitude integrated
EEG (aEEG) is a useful tool in seizure detection. However,
neonatal seizures can be difficult to assess using this modal-
ity, particularly when they are infrequent, brief, or of low
amplitude (Frenkel et al., 2011; Shellhaas et al., 2007;
Rennie et al., 2004).
The temporal behavior of electrographic neonatal sei-
zures across a diverse group of newborns with varying ges-
tational ages and seizure etiologies were investigated by
Clancy & Legido (1987) using EEG recordings ranging in
duration from 27 min to 3 h. Seizures were recurrent, but
brief, with a mean seizure duration of 137 s and mean inte-
rictal duration of 480 s. A study by McBride et al. (2000)
using continuous multichannel videoelectroencephalogra-
phy (EEG) from as soon as possible after birth in a high risk
group of newborns identified electrographic seizures within
1 h in 65% of newborns monitored. All infants who devel-
oped seizures did so within 21 h of commencement of moni-
toring. In newborns with HIE, mean seizure duration was
93 s, and electrographic seizures had ceased by 73 h of age.
Newborns with HIE had more seizures and a higher seizure
burden (the accumulated duration of seizures over a defined
time period) over the duration of the recording, when com-
pared to those with seizures secondary to stroke.
Treatment of neonatal seizures is difficult and controver-
sial, as there is evidence that traditional first- and second-line
antiepileptic drugs (AEDs), phenobarbital and phenytoin,
are poorly effective (Booth & Evans, 2004; Painter et al.,
1999). With increasing evidence that high seizure burden is
detrimental to the developing brain, a greater understanding
of temporal evolution would be extremely valuable (Miller
et al., 2002; Glass et al., 2009; Bjrkman et al., 2010;
Ancora et al., 2010; Glass et al., 2011).
Given the unreliability of clinical signs associated with
electrographic seizures, and because neonatal seizures can
be difficult to detect using aEEG, continuous multichannel
video-EEG remains the gold standard for monitoring sei-
zures in the newborn (Shah et al., 2012). Continuous EEG is
not, however, available in most neonatal intensive care units
(NICUs). As a result, there is relatively little information
regarding the temporal evolution of electrographic seizure
burden in newborns who have sustained a hypoxic ischemic
injury. In our institution, infants with neonatal encephalopa-
thy have continuous video-EEG monitoring for a minimum
of 48 h, from as soon as possible after birth, allowing us to
analyze the evolution of seizure burden over time.
Aims
The aim of our study was to ascertain whether there is a
pattern to the temporal evolution of seizure burden in new-
borns with HIE. We aimed to accurately map the evolution
of electrographic seizure burden over time from first elect-
rographic seizure recording in newborns who have sus-
tained a hypoxic ischemic injury and have developed
moderate or severe HIE. This cohort is unique, as data were
obtained from infants who did not receive therapeutic hypo-
thermia (TH). Because TH is now part of standard care for
infants with HIE, it is unlikely that these data will be avail-
able in the future. However, analysis of these data is essen-
tial to provide a benchmark against which we can assess the
effect of new and innovative forms of neuroprotection and
antiepileptic medication on evolution of seizure burden in
newborns with HIE.
Methods
Study population
Our institution recruits newborn infants with risk factors
for hypoxic ischemic encephalopathy for ongoing EEG
research studies. All studies have ethical approval from the
Clinical Research and Ethics Committee of the Cork Teach-
ing Hospitals. Newborns with suspected hypoxic ischemic
insult are recruited for continuous video-EEG from as soon
as possible after presentation to the NICU.
The majority of the infants in this retrospective study
were recruited between 2003 and 2006, prior to the intro-
duction of whole body therapeutic hypothermia (TH) in our
550
N. E. Lynch et al.
Epilepsia, 53(3):549557, 2012
doi: 10.1111/j.1528-1167.2011.03401.x
institution. Infants who received TH were excluded from
the study. A small number of infants recruited after 2006
and who were diagnosed with moderate HIE after 6 h of
age, which is beyond the age at which commencement of
TH would have a neuroprotective effect (Azzopardi et al.,
2009) were also included.
Inclusion criteria:
1 Term infants (>37 weeks of gestation).
2 Acute HIE.
3 Continuous EEG monitoring commenced within 36 h of
birth.
4 Continuous EEG monitoring for a minimum of 48 h.
5 Electrographic seizures.
Infants were considered to be at risk of acute HIE if they
met 2 of the following criteria: initial arterial or capillary
pH of <7.1, Apgar score at 5 min of <5, initial arterial or
capillary lactate of >7 mMol/L, or abnormal neurologic fea-
tures/clinical seizures. Our criteria ensured that infants with
a broad range of severity of neonatal encephalopathy were
recruited. Neurologic condition was assessed by a research
physician who was not involved in the clinical care of the
infant. Written, informed consent was obtained from the
parents of infants who fulfilled the criteria for EEGmonitor-
ing (Murray et al., 2009). A clinical grade of encephalopa-
thy was assigned to each infant using the modified Sarnat
score (Levene et al., 1985).
EEG recordings
EEG electrodes were applied to the scalp at F3, F4, C3,
C4, T3, T4, O1, O2, and CZ (according to the international
1020 system of electrode placement, as modified for neo-
nates). A NicOne EEG monitor (CareFusion NeuroCare,
Middleton, WI, U.S.A.) was used to record continuous
video-EEG recordings for approximately 72 h. Recordings
commenced as soon as possible after birth. Physiologic
measurements of heart rate, respiration, oxygen saturation,
and, where available, direct arterial blood pressure were
recorded from the infants intensive care monitor and were
recorded simultaneously with the EEG. EEGs were ana-
lyzed by two experienced electroencephalographers. An
electrographic seizure was defined as repetitive rhythmic
activity of >10 s duration, with a distinct beginning, middle,
and end (Clancy & Legido, 1987; Scher, 1993). All electro-
graphic seizures were annotated from the start of each sei-
zure on any channel to the end of the seizure on any
channel.
Data acquisition and analysis
The medical records of each infant were reviewed and the
time and description of the first clinical seizure recorded in
the medical records was noted. This was the first clinical
seizure diagnosed by the medical staff caring for the baby,
independent of information available from the EEG record-
ing. The first clinical seizure was defined as abnormal
movements or autonomic symptoms observed by the
medical or nursing staff and noted in the medical records as
seizure activity. Types of AEDs, as well as the time of
administration, were recorded.
Seizure annotations were converted into a time series,
where 1 represented the presence of a seizure and 0 the
absence of a seizure. This time series was calculated using a
sampling frequency of 1 Hz. The seizure annotation time
series was then cropped between the start of the first
recorded EEG seizure and the end of the last recorded EEG
seizure. The length of this period was recorded (seizure per-
iod). Within this seizure period the number of seizures and
the total seizure burden were recorded. The total seizure
burden is the total duration, in minutes, of all seizures
recorded in each infant during EEG monitoring (Clancy &
Legido, 1987; Bye & Flanagan, 1995; Murray et al.,
2008a).
These summary statistics are biased due to missing data
in some EEGrecordings. Missing data arose when recording
was discontinued to facilitate medical treatment or investi-
gations, or when EEG leads became detached. Where data
are missing, it is not possible to be certain whether or not
there was seizure activity. We assumed that the seizure
activity did not change within the missing data period. The
summary statistics were, therefore, averaged across the per-
iod of available data resulting in the seizure burden per hour
and number of seizures per hour. The seizure duration
(mean seizure duration per infant) was also averaged across
the period of available data.
For each infant, the distribution of seizure burden within
the seizure period was assessed using the skewness coeffi-
cient. A positive skewness coefficient indicates an accumu-
lation of seizure burden near the first recorded seizure,
whereas a negative skewness coefficient indicates an accu-
mulation of seizure burden near the last recorded EEG sei-
zure. To further investigate the temporal evolution of
seizure burden, the seizure annotation time series of each
newborn was then segmented into two periods: the
time between the first recorded seizure and the maximum
seizure burden (T
1
), and the time between the maximum
seizure burden and the last recorded seizure (T
2
). The time
of maximum seizure burden was defined as the midpoint of
an hour-long window shifted in time by 1 s across the full
EEG recording that contained the maximum duration of sei-
zure. Within these two periods, the seizure burden per hour,
the seizure duration, and number of seizures per hour were
measured.
Statistical analysis
Continuous data were described using the median, the
interquartile range (IQR), and the minimum and maximum
values. The symmetry of the temporal evolution of seizure
burden was described using the skewness coefficient, and
was tested (against a population median of 0) using the one-
sample Wilcoxon signed-rank test. Differences in summary
measures between the two time periods were investigated
551
Temporal Evolution-EEGSeizures in HIE
Epilepsia, 53(3):549557, 2012
doi: 10.1111/j.1528-1167.2011.03401.x
using the Sign test. Differences in summary measures
between HIE grades were investigated using the Mann-
Whitney test. The strength of the association between two
continuous variables was measured using Spearmans rank
correlation coefficient. All statistical analyses were per-
formed using PASW Statistics, version 18.0 (IBM SPSS
Statistics, Chicago, IL, U.S.A.). All tests were two-sided
and a p-value < 0.05 was considered to be statistically
significant.
Results
From 2003 to 2010, 107 newborns with HIE had continu-
ous video-EEG monitoring. The modified Sarnat score for
HIE was assigned as mild in 43, moderate in 34, and severe
in 30 newborns. Among the 64 newborns with moderate or
severe HIE, 31 received therapeutic hypothermia. These
infants were excluded from this study. In the remaining 33
newborns, who had moderate or severe HIE, 18 developed
seizures. Three infants were excluded due to EEG record-
ings of <48 h duration and, therefore, 15 newborns were
included in this study. Clinical characteristics of these
infants are described in Table 1. Thirteen of 15 infants
were recruited prior to 2006, when TH was introduced in
our institution. Nine of 15 infants had an instrumental deliv-
ery (vacuum or forceps), two of 15 were delivered by emer-
gency caesarean section, three of 15 had standard vaginal
deliveries, and one infant had shoulder dystocia. All of the
infants required resuscitation at birth. Placental histology
information was not available for any of the infants. None of
the infants developed hypoglycemia, severe electrolyte
imbalance, or sepsis during the EEG monitoring period. Six
infants had moderate HIE (grade 2) and nine had severe HIE
(grade 3) (Table 1).
Thirteen of 15 infants had an MRI scan of the brain within
the first 2 weeks of birth. None of the MRI scans showed
evidence of chronic injury. Six MRI scans were reported as
normal, two infants had basal ganglia abnormalities, and
five infants had abnormalities consistent with watershed
injury.
A total of 1,635 seizures were recorded in the 15 infants.
Figure 1 illustrates the variability in seizure burden between
infants. In many of the infants, there are prolonged periods
of quiescence that were not accounted for by missing data,
and do not appear to be temporally related to AED adminis-
tration. The time of the first recorded clinical seizure was
available in 12 of 15 infants. In 11 of 12 infants the first
recorded clinical seizure consisted of apnea, lip smacking,
or limb cycling. In one infant, clonic limb jerking was
recorded. The first clinical seizure was recorded before the
start of EEG monitoring in 9 of 12 infants. The median time
from first recorded clinical seizure to commencement of
EEG recording was 1.8 h (IQR )2.0 to 6.0). The median
time from first recorded clinical seizure to first recorded
EEG seizure was 2.8 h (IQR 0.85.3).
EEG studies commenced as soon as possible after the
infants were identified as meeting inclusion criteria for our
institutions EEG research studies and parental consent had
been obtained. As represented in Figure 1, EEG monitoring
was commenced at between 2 and 26 h of age and a median
age of 14 h (IQR 620). EEG monitoring continued for a
median of 70.7 h (IQR62.981.9). In 7 of 15 infants, electro-
graphic seizures were ongoing at the start of EEG monitor-
ing.
The summary statistics for seizure burden are shown in
Table 2. Seizures continued over a median period of 36.6 h
from first recorded seizure to last recorded seizure (IQR
17.852.7). The median age at first recorded electrographic
seizure was 17.1 h (IQR 12.521.3). There was time vari-
ability across which maximum seizure burden occurred
(between 13 and 42 h of age). The median age at which
maximum electrographic seizure burden was reached was
22.7 h (IQR 19.029.9). In 11 of 15 infants (four moderate
and seven severe) the maximum seizure burden was in
excess of 30 min of seizure per hour. This corresponds to a
definition of neonatal status epilepticus (Lawrence & Inder,
2010). The median age at which the last electrographic
seizure was recorded was 55.5 h (IQR 41.464.0).
Figure 2A illustrates the nonuniformity of distribution
of seizure burden, and the temporal evolution of seizure
burden in this cohort of infants with HIE. The median of
the skewness coefficient was positive and statistically sig-
nificant (coefficient 0.62, p-value = 0.02, Table 2), indi-
cating an accumulation of seizures near the first recorded
seizure. The time from first recorded seizure to maximum
seizure burden (T
1
) was significantly shorter than time
from maximum seizure burden to last recorded seizure
(T
2
) (p-value = 0.01, Table 2). Figure 1 illustrates that
maximum seizure burden occurs closer to the start of
EEG recording than to the end of EEG recording, as con-
firmed by a positive skewness coefficient. The median
time from first recorded seizure to maximum seizure bur-
den was 5.9 h.
Median seizure burden during T
1
was significantly higher
than median seizure burden during T
2
(p-value = 0.007,
Table 1. Clinical characteristics of neonates
included in the study (n = 15)
Gestational age (weeks) 40.7 (40.440.9)
Birth weight (kg) 3.3 (3.083.66)
Age at NICUadmission (mins) 25 (2050)
Gender (M:F) 9:6
Clinical HIE score
Moderate-to-severe 6:9
5-min Apgar score 6 (27)
First pH(n = 12) 7.06 (7.027.18)
Lactate (mMol/L) (n = 12) 10.8 (8.512.8)
Base excess (mEq/L) (n = 13) )12.2 ()16.8 to )11.8)
Data are median (interquartile range) or n.
552
N. E. Lynch et al.
Epilepsia, 53(3):549557, 2012
doi: 10.1111/j.1528-1167.2011.03401.x
Table 2). There was no significant difference in the number
of seizures per h between the two time periods.
Infants with severe HIE had a longer seizure period
(p-value = 0.007), as illustrated in Figure 2B. This arose
from a significantly longer T
2
period in severe HIE
(p-value = 0.01). Median seizure burden was significantly
higher during T
1
versus T
2
in newborns with severe HIE
(p-value = 0.004, Table 3), whereas the difference in sei-
zure burden during T
1
versus T
2
in moderate HIE was not
significant. Median seizure duration was significantly
longer during T
1
than during T
2
in infants with severe HIE
(p-value = 0.04, Table 3).
The use of AEDs was reviewed. Two infants, with sub-
clinical seizures only, received no medication. Thirteen of
15 infants received phenobarbital, at a median age of 12.3 h
(IQR 10.219.2). Eleven of 13 infants received phenobarbital
Figure 1.
A graphical representation of the seizure time series combined with clinical information. The start and end of EEG monitoring are
denoted by blue triangles. Annotated seizures are denoted by a vertical black line. The time of rst recorded clinical seizure is denoted
by a blue circle. Missing data are denoted by a red cross. Phenobarbital loading dose administration is denoted by a vertical red line,
phenytoin loading dose administration by a vertical magenta line, and rst dose of midazolam by a vertical green line. The maximum
seizure burden (peak in seizure burden) is denoted by a blue cross.
Epilepsia ILAE
Table 2. Comparison of timing, seizure burden, duration and frequency overall and during T
1
and T
2
(n = 15)
T
1
+ T
2
T
1
T
2
p-value
(fromSign test)
Median (IQR) Median (IQR) Median (IQR)
Seizure period (h) 36.6 (17.852.7) 5.9 (3.49.6) 29.0 (14.544.7) 0.01
Total seizure burden (min) 206 (152413) 92 (50133) 108 (66336) 0.30
Seizure number 84 (41190) 22 (848) 48 (22124) 0.42
Missing data (%) 11.7 (1.424.1) 0 (08.4) 4.8 (017.5) 0.15
Seizure burden (min/h) 8.9 (4.114.6) 17.1 (9.626.5) 6.3 (3.713.7) 0.007
Mean seizure duration (s) 206 (98331) 306 (192484) 177 (88404) 0.12
Number of seizures (per h) 2.7 (1.94.1) 4.2 (2.35.9) 2.2 (1.44.2) 0.12
Skewness* 0.62 ()0.09 to 1.28) 0.02*
T
1
: time from rst recorded EEG seizure to time of maximum seizure burden; T
2
: time from maximum seizure burden to time of last recorded clinical seizure.
*p-value = 0.02 from one-sample Wilcoxon signed-rank test.
553
Temporal Evolution-EEGSeizures in HIE
Epilepsia, 53(3):549557, 2012
doi: 10.1111/j.1528-1167.2011.03401.x
before maximum seizure burden was reached. Nine infants
received phenytoin as a second-line AED, at a median age
of 29.5 h (IQR 27.633.4). All nine received phenytoin after
maximum seizure burden had been reached. Eight contin-
ued to have electrographic seizures following its administra-
tion. Three infants received midazolam as a third-line AED.
All continued to have seizures following midazolam admin-
istration. There was no association between time of drug
administration and time that maximum seizure burden was
reached (Spearmans correlation coefficient [p-
value] = )0.17 [0.58] and )0.17 [0.67] for phenobarbital
and phenytoin, respectively).
Other medications that can influence the EEG and seizure
activity were reviewed. Two infants received magnesium
sulfate because of low magnesium levelsinfant 4 (Fig. 1)
at 45 h of age and infant 15 at 15 h of age. Two infants
received pyridoxineinfant 12 at 26 h of age and infant 6
at 80 h of age. Five infants received morphine at a median
age of 4 h.
Discussion
To our knowledge, this is the first quantitative description
of the temporal evolution of electrographic seizure burden
in full-term newborns with HIE. Electrographic seizure data
provide a more accurate picture of the true seizure burden in
newborns who often have subclinical or subtle seizures.
Recognition of a pattern of temporal evolution of seizure
burden in this historical cohort of newborns establishes a
benchmark against which to compare seizure burden in
newborns with HIE who receive therapeutic hypothermia,
and/or other novel forms of neuroprotection or antiepileptic
therapy.
The distribution of seizure burden is not uniform. The
time from first recorded seizure to maximum seizure burden
(T
1
) is significantly shorter than the time from maximum
seizure burden to last recorded seizure (T
2
), and the median
seizure burden during T
1
is significantly higher than the
median seizure burden during T
2
. This indicates a short per-
iod of high seizure burden followed by a longer period of
low seizure burden.
Hypoxic ischemic brain injury in the peripartum period
results in an evolving process of brain injury. The combina-
tion of reduced oxygen delivery and reduced blood flow
A
B
Figure 2.
Median seizure burden over time with and without temporal
evolution (TE). Graph (A) represents the entire cohort of new-
borns (moderate and severe HIE). The solid blue line repre-
sents evolution in seizure burden over time (temporal
evolution). The broken red line represents how seizure burden
would evolve if there were uniformity of seizure burden across
time, with no temporal evolution. Graph (B) compares distri-
bution of seizure burden in the newborns with severe HIE and
moderate HIE. The solid red line represents temporal evolu-
tion in severe HIE, and the broken red line represents how sei-
zure burden would evolve if there were no temporal evolution.
The solid blue line represents temporal evolution in moderate
HIE. The broken blue line represents how seizure burden
would evolve if there were no temporal evolution.
Epilepsia ILAE
Table 3. Comparison of timing, seizure burden, duration, and frequency over the two time periods by HIE
grading
T
1
+ T
2
T
1
T
2
p-value
(fromSign test)
Median (IQR) Median (IQR) Median (IQR)
Moderate (n = 6)
Time (h) 17.7 (10.834.6) 4.8 (2.510.8) 10.1 (4.628.8) 0.69
Seizure burden (min/h) 13.0 (4.116.6) 14.9 (8.833.6) 10.5 (3.327.8) 0.69
Seizure duration (s) 362 (168513) 395 (181779) 423 (149650) 1
Number of seizures (per h) 2.5 (1.23.8) 3.7 (1.55.6) 1.8 (1.05.0) 0.22
Severe (n = 9)
Time (h) 47.4 (35.460.6) 8.6 (3.811.4) 36.0 (26.756.1) 0.004
Seizure burden (min/h) 7.4 (5.214.0) 17.1 (9.525.0) 6.1 (4.310.5) 0.004
Seizure duration (s) 126 (97295) 306 (156377) 101 (84280) 0.04
Number of seizures (per h) 2.9 (2.04.8) 4.4 (2.66.9) 2.7 (1.85.1) 0.51
554
N. E. Lynch et al.
Epilepsia, 53(3):549557, 2012
doi: 10.1111/j.1528-1167.2011.03401.x
results in rapid depletion of energy metabolites, leading to
the failure of the ATP-dependent Na/K pump resulting in
Na
+
influx and leading to cell death by necrosis (Volpe,
2008a). Animal studies have shown that following the initial
insult, there is a latent phase during which oxidative metab-
olism normalizes, followed by a secondary failure of oxida-
tive metabolism (Gunn & Bennet, 2009). This corresponds
with an apparent clinical improvement in newborns with a
hypoxic ischemic injury. In our cohort, the median time of
first electrographic seizure recording is 17.1 h. This corre-
lates with Volpes described onset of secondary energy fail-
ure, which is usually apparent by 816 h of age and
maximal at 2448 h (Volpe, 2008a). The median age at
maximum electrographic seizure burden in our cohort was
22.7 h. The time of seizure onset we report corresponds clo-
sely to the findings of Filan et al. (2005) (1820 h) and is
earlier than reported in Scher et al. (2008) (34.8 h). TH may
alter the time of seizure onset. Wusthoff et al. (2011) inves-
tigated the incidence and timing of electrographic seizures
in term newborns who were undergoing TH. The mean time
of first seizure onset was 35 h in these infants. Future stud-
ies may help to clarify whether time of seizure onset is
related to time of initial injury. The focus of our study was
the evolution of seizure burden over time once seizures
begin.
The temporal and spatial characteristics of neonatal sei-
zures have been studied previously. Bye & Flanagan (1995)
quantified characteristics of 1,420 seizures in a cohort of 32
newborns and found that seizure variables were relatively
stable over time. Seizures in this cohort had a variety of eti-
ologies, however, with 9 of 32 infants having seizures sec-
ondary to HIE. Shellhaas & Clancy (2007) analyzed 851
seizures from a cohort of 121 newborns. Seizure burden was
quantified as a percentage of an infants EEG recording with
ictal activity, and was high, at 24.8%. Seizures were found
to be numerous but brief (mean 132 s). However, these data
were obtained from short EEG recordings of 23145 min.
Newborns in this cohort were of various gestational ages,
with a range of seizure etiologies. Analysis of temporal
evolution of seizure burden is possible only with long-term
continuous EEG. Our EEG monitoring periods were a mini-
mum of 48 h, allowing us to specifically investigate evolu-
tion in seizure burden over time in full-term newborns with
HIE. As in these previous studies, seizure burden was calcu-
lated as a total of seizures recorded from any channel. We
did not calculate seizure burden per EEG channel. It is rec-
ognized that prolonged seizures and high seizure burden are
damaging to the brain regardless of whether seizures are
focal or generalized (Holmes, 2002; Nagarajan et al., 2011).
However, further analysis of the spatial evolution of sei-
zures in conjunction with analysis of temporal evolution
will add important new information to our understanding of
neonatal seizures. In our cohort, seizure burden was high,
with infants having a median of 2.7 seizures per hour and
8.9 min of seizure per hour (Table 2). Seizure duration was
longer than that found in other studies (206 s). In these
studies, seizure duration ranges from 93137 s. (Clancy &
Legido, 1987; McBride et al., 2000; Shellhaas & Clancy,
2007). There is a trend toward seizure duration being shorter
during T
2
.
In this retrospective study, we found that there is a trend
toward a higher seizure burden among infants with severe
HIE due to a high seizure burden during T
1
and a longer
seizure period. This is in contrast to findings in previous
studies (Sarnat & Sarnat, 1976), which suggest that infants
with severe HIE have fewer seizures in the context of a
severely suppressed or isoelectric EEG. This may be
explained by brief recordings in other studies during the
period that we have identified as T
2
, where there are fewer
minutes of seizure per hour, but seizures continue over many
hours. However, the small number of infants in each sub-
group (six moderate vs. nine severe HIE) limits the statistical
analysis. This observation warrants further investigation.
Despite the small sample size, we have demonstrated a
consistent, statistically significant temporal evolution of
electrographic seizure burden. This pattern of evolution has
not been described previously and has not been studied in
animals or in humans. High initial seizure burden may be
explained by the excitability of the newborn brain and its
predisposition to seizures. The subsequent decline in seizure
activity may represent depletion of existing neurotransmit-
ters and the inability of the injured brain to replenish
depleted stores (Sarnat et al., 2010). It may also correspond
to the onset of apoptosis in previously epileptogenic neu-
rons. These findings warrant further animal studies and
observational studies in humans.
Not all infants were undergoing EEG monitoring when
the first clinical seizure was recorded. However, the first
clinical seizure is an unreliable time point, given that abnor-
mal movements are often misinterpreted as seizures, and
true seizures are frequently unrecognized (Murray et al.,
2008a). Furthermore, the median time from first recorded
clinical seizure to commencement of EEG recording was
short, at 1.8 h, and electrographic seizure burden in all cases
increased following start of EEG recording, implying that
maximum seizure burden was recorded in all infants, and
that the temporal evolution we report is accurate.
Some infants had periods of missing EEG data during
recordings either because of delayed application of EEG
leads or because of a temporary cessation in EEG acquisi-
tion during the recording period. Missing data in an observa-
tional study of this kind are difficult to avoid in the context
of ill newborns receiving intensive care in the NICU, who
require multiple interventions and investigations.
Although we cannot discount the possibility that some of
the newborns in this cohort had a suboptimal in utero envi-
ronment or experienced a degree of hypoxia in the antepar-
tum period that predisposed them to perinatal brain injury,
their clinical presentation and subsequent investigations
indicate that there was significant peripartum injury. The
555
Temporal Evolution-EEGSeizures in HIE
Epilepsia, 53(3):549557, 2012
doi: 10.1111/j.1528-1167.2011.03401.x
majority of infants were of normal birth weight, born by
either instrumental delivery or emergency caesarean sec-
tion, and required resuscitation and immediate admission to
NICU. There was a predominance of metabolic acidosis and
elevated lactate, which are associated with significant peri-
partum morbidity (Toh, 2000; Malin et al., 2010; da Silva
et al., 2000). Our findings are concordant with those of
Cowan et al. (2003), which suggest that events in the imme-
diate perinatal period are most important in neonatal brain
injury.
In this cohort, overall seizure burden may have been
influenced by the administration of phenobarbital, or sec-
ond-line AEDs. Given that AED administration is consid-
ered part of standard care for neonatal seizures, it is not
possible or ethical to study the temporal evolution of sei-
zure burden in the absence of medication in human new-
borns. The majority of infants in this cohort received
AEDs but the number of AEDs administered and the time
of administration varied greatly. However, the timing of
AED administration in these infants does not correspond
with the time points of interest in this study, in particular
the time at which maximum seizure burden was reached,
as indicated by weak and statistically insignificant correla-
tion coefficients. This may imply that the temporal evolu-
tion of seizure burden is independent of timing of AED
administration. Several studies, using both continuous
EEG and aEEG have illustrated inconsistent reduction of
seizure burden following administration of phenobarbital
and other second-line AEDs (Connell et al., 1989; Boylan
et al., 2002; Painter et al., 1999). Five infants received
morphine. There is no evidence to suggest that morphine
affects the incidence of seizures in newborns, although it
is recognized as having an effect on EEG background
(Bye et al., 1997).
We have illustrated that a period of maximum seizure
burden is reached within a median of 5.9 h of the first
recorded seizure. This suggests a narrow window for opti-
mizing therapy aimed at reducing seizure burden. This
study also provides reference data for the temporal evolu-
tion of seizures in HIE. Recognition of a consistent pat-
tern of temporal evolution may help confirm the diagnosis
of HIE, and rule out neonatal seizures due to other etiolo-
gies such as neonatal stroke, and inborn errors of metabo-
lism, more quickly than neuroimaging, which is not
usually performed within the first 72 h of life. The identi-
fication of temporal evolution in this cohort of infants
with HIE and seizures serves to emphasize the importance
of early, and continuous long-term EEG monitoring of
neonatal seizures.
Assessment of all aspects of seizure burden, and the effect
of medication or therapies on reducing this burden is impor-
tant, given the growing body of evidence showing that a
high seizure burden in HIE is associated with further brain
injury, independent of the initial hypoxic insult (Miller
et al., 2002; Glass et al., 2009; Ancora et al., 2010).
Conclusion
This study provides evidence that there is temporal evolu-
tion of electrographic seizure burden in full-term newborns
with HIE. There is a short period of high seizure burden
(T
1
), which reaches a peak at approximately 23 h after birth
followed by a longer period of lower seizure burden (T
2
).
Identification of a temporal evolution of seizure burden in
HIE is important for several reasons. It allows further char-
acterization of seizures in infants who have sustained a hyp-
oxic brain injury. It identifies a short period of intense
seizure burden near the beginning of seizure onset, which
may be an important therapeutic window during which
effective antiepileptic treatment would optimize reduction
in seizure burden.
Finally, it provides a benchmark against which the effect
of new and innovative forms of neuroprotection and antiepi-
leptic medication on evolution of seizure burden in
newborns with HIE can be assessed.
Acknowledgments
Dr Lynch is supported by the Denis O Sullivan Research Fellowship
from University College Cork, and the 2011 IICN/UCB Pharma Bursary in
Neuroscience. The research was supported by a translational award from
the Wellcome Trust UK (85249/z/08/z), and a Science Foundation Ireland
grant (10/IN.1/B3036).
Disclosure
None of the authors has any conflict of interest to disclose. We confirm
that we have read the journals position on issues involved in ethical publi-
cation and affirmthat this report is consistent with those guidelines.
References
Ancora G, Soffritti S, Lodi R, Tonon C, Grandi S, Locatelli C, Nardi L,
Bisacchi N, Testa C, Tani G, Ambrosetto P, Faldella G. (2010)
Acombined a-EEG and MR spectroscopy study in term newborns with
hypoxic-ischemic encephalopathy. Brain Dev 32:835842.
Azzopardi DV, Strohm B, Edwards AD, Dyet L, Halliday HL, Juszczak E,
Kapellou O, Levene M, Marlow N, Porter E, Thoresen M, Whitelaw A,
Brocklehurst P; TOBY Study Group. (2009) Moderate hypothermia to
treat perinatal asphyxial encephalopathy. NEngl J Med 361:13491358.
Ben-Ari Y, Khazipov R, Leinekugel X, Caillard O, Gaiarsa JL. (1997)
GABAA, NMDA and AMPA receptors: a developmentally regulated
mnage trois. Trends Neurosci 20:523529.
Bjrkman ST, Miller SM, Rose SE, Burke C, Colditz PB. (2010) Seizures
are associated with brain injury severity in a neonatal model of
hypoxia-ischemia. Neuroscience 166:157167.
Booth D, Evans DJ. (2004) Anticonvulsants for neonates with seizures.
Cochrane Database of Systematic Reviews. Issue 3. Art. No.: CD004218.
DOI: 10.1002/14651858.CD004218.pub2.117
Boylan GB, Rennie JM, Pressler RM, Wilson G, Morton M, Binnie CD.
(2002) Phenobarbitone, neonatal seizures, and video-EEG. Arch Dis
Child Fetal Neonatal Ed 86:F165F170.
Brillault J, Lam TI, Rutkowsky JM, Foroutan S, ODonnell ME. (2008)
Hypoxia effects on cell volume and ion uptake of cerebral microvascu-
lar endothelial cells. AmJ Physiol Cell Physiol 294:C88C96.
Bye AME, Flanagan D. (1995) Spatial and temporal characteristics of neo-
natal seizures. Epilepsia 36:10091016.
Bye AME, Naidoo D, Flanagan D. (1997) The effects of morphine and
midazolamon EEGs in neonates. J Clin Neurosci 4:173175.
556
N. E. Lynch et al.
Epilepsia, 53(3):549557, 2012
doi: 10.1111/j.1528-1167.2011.03401.x
Clancy RR, Legido A. (1987) The exact ictal and interictal duration of elec-
troencephalographic neonatal seizures. Epilepsia 28:537541.
Connell J, Oozeer R, Devries L, Dubowitz LMS, Dubowitz V. (1989) Clini-
cal and EEG response to anticonvulsants in neonatal seizures. Arch Dis
Child 64:459464.
Cowan F, Rutherford M, Groenendaal F, Eken P, Mercuri E, Bydder GM,
Meiners LC, Dubowitz LM, de Vries LS. (2003) Origin and timing of
brain lesions in term infants with neonatal encephalopathy. Lancet
361:736742.
da Silva S, Hennebert N, Denis R, Wayenberg JL. (2000) Clinical value of
a single postnatal lactate measurement after intrapartum asphyxia. Acta
Paediatr 89:320323.
Dzhala VI, Talos DM, Sdrulla DA, Brumback AC, Mathews GC, Benke
TA, Delpire E, Jensen FE, Staley KJ. (2005) NKCC1 transporter facili-
tates seizures in the developing brain. Nat Med 11:12051213.
Filan P, Boylan GB, Chorley G, Davies A, Fox GF, Pressler R, Rennie JM.
(2005) The relationship between the onset of electrographic seizure
activity after birth and the time of cerebral injury in utero. BJOG
112:504507.
Frenkel N, Friger M, Meledin I, Berger I, Marks K, Bassan H, Shany E.
(2011) Neonatal seizure recognition comparative study of continu-
ous-amplitude integrated EEG versus short conventional EEG record-
ings. Clin Neurophysiol 122:10911097.
Glass HC, Glidden D, Jeremy RJ, Barkovich AJ, Ferriero DM, Miller SP.
(2009) Clinical neonatal seizures are independently associated with out-
come in infants at risk for hypoxic-ischemic brain injury. J Pediatr
155:318323.
Glass HC, Nash KB, Bonifacio SL, Barkovich AJ, Ferriero DM, Sullivan
JE, Cilio MR. (2011) Seizures and magnetic resonance imaging-
detected brain injury in newborns cooled for hypoxic-ischemic enceph-
alopathy. J Pediatr 159:731735.e1. Epub 2011 Aug 11.
Gunn AJ, Bennet L. (2009) Fetal hypoxia insults and patterns of brain
injury: insights fromanimal models. Clin Perinatol 36:579593.
Holmes GL. (2002) Seizure-induced neuronal injury: animal data. Neurol-
ogy 59(9 Suppl. 5):S3S6.
Lawrence R, Inder T. (2010) Neonatal status epilepticus. Semin Pediatr
Neurol 17:163168.
Levene MI, Trounce JQ. (1986) Cause of neonatal convulsions. Towards
more precise diagnosis. Arch Dis Child 61:7879.
Levene ML, Kornberg J, Williams THC. (1985) The incidence and severity
of post-asphyxial encephalopathy in full-term infants. Early Hum Dev
11:2126.
Malin GL, Morris RK, Khan KS. (2010) Strength of association between
umbilical cord pH and perinatal and long term outcomes: systematic
review and meta-analysis. BMJ 13:340:c1471. Doi: 10.1136/bmj.
c1471.
McBride MC, Laroia N, Guillet R. (2000) Electrographic seizures in neo-
nates correlate with poor neurodevelopmental outcome. Neurology
55:506513.
Miller SP, Weiss J, Barnwell A, Ferriero DM, Latal-Hajnal B, Ferrer-
Rogers A, Newton N, Partridge JC, Glidden DV, Vigneron DB, Barko-
vich AJ. (2002) Seizure-associated brain injury in term newborns with
perinatal asphyxia. Neurology 58:542548.
Mizrahi EM, Kellaway P. (1987) Characterization and classification of neo-
natal seizures. Neurology 37:18371844.
Murray DM, Boylan GB, Ali I, Ryan CA, Murphy BP, Connolly S. (2008a)
Defining the gap between electrographic seizure burden, clinical
expression and staff recognition of neonatal seizures. Arch Dis Child
Fetal Neonatal Ed 93:F187F191.
Murray DM, Boylan GB, Fitzgerald AP, Ryan CA, Murphy BP, Connolly
S. (2008b) Persistent lactic acidosis in neonatal hypoxic-ischaemic
encephalopathy correlates with EEG grade and electrographic seizure
burden. Arch Dis Child Fetal Neonatal Ed 93:F183F186. Epub 2006
Nov 28.
Murray DM, Boylan GB, Ryan CA, Connolly S. (2009) Early EEG findings
in hypoxic-ischemic encephalopathy predict outcomes at 2 years.
Pediatrics 124:E459E467.
Nagarajan L, Ghosh S, Palumbo L. (2011) Ictal electroencephalograms in
neonatal seizures: characteristics and associations. Pediatr Neurol
45:1116.
Painter MJ, Scher MS, Stein AD, Armatti S, Wang Z, Gardiner JC, Paneth
N, Minnigh B, Alvin J. (1999) Phenobarbital compared with phenytoin
for the treatment of neonatal seizures. N Engl J Med 341:485489.
Rennie JM, Chorley G, Boylan GB, Pressler R, Nguyen Y, Hooper R.
(2004) Non-expert use of the cerebral function monitor for neonatal sei-
zure detection. Arch Dis Child Fetal Neonatal Ed 89:F37F40.
Rose AL, Lombroso CT. (1970) Neonatal seizure states. Astudy of clinical,
pathological and electroencephalographic features in 137 full-term
babies with a long- termfollow-up. Pediatrics 45:404425.
Sarnat HB, Sarnat MS. (1976) Neonatal encephalopathy following fetal
distress clinical and electroencephalographic study. Arch Neurol
33:696705.
Sarnat HB, Flores-Sarnat L, Trevenen CL. (2010) Synaptophysin immuno-
reactivity in the human hippocampus and neocortex from 6 to 41 weeks
of gestation. J Neuropathol Exp Neurol 69:234245.
Scher MS. (1993) Ictal and interictal electrographic seizure durations in
pretermand termneonates. Epilepsia 34:284288.
Scher MS, Aso K, Beggarly ME, Hamid MY, Steppe DA, Painter MJ.
(1993) Electrographic seizures in preterm and full-term neonates
clinical correlates, associated brain-lesions, and risk for neurologic
sequelae. Pediatrics 91:128134.
Scher MS, Steppe DA, Beggarly M. (2008) Timing of neonatal seizures and
intrapartum obstetrical factors. J Child Neurol 23:640643.
Shah DK, Boylan GB, Rennie JM. (2012) Monitoring of seizures in the
newborn. Arch Dis Child Fetal Neonatal Ed 97:F65F69.
Shellhaas RA, Clancy RR. (2007) Characterization of neonatal seizures by
conventional EEG and single channel EEG. Clin Neurophysiol
118:21562161.
Shellhaas RA, Soaita AI, Clancy RR. (2007) Sensitivity of amplitude-
integrated electroencephalography for neonatal seizure detection.
Pediatrics 120:770777.
Toh VC. (2000) Early predictors of adverse outcome in term infants with
post-asphyxial hypoxic ischaemic encephalopathy. Acta Paediatr
89:343347.
Volpe JJ. (2008a) Neurology of the newborn. 5th ed. Saunders Elsevier,
Philadelphia, PA, pp. 247305.
Volpe JJ. (2008b) Neurology of the newborn. 5th ed. Saunders Elsevier,
Philadelphia, PA, pp. 401461.
Wirrell EC. (2005) Neonatal seizures: to treat or not to treat? Semin Pediatr
Neurol 12:97105.
Wusthoff CJ, Dlugos DJ, Gutierrez-Colina A, Wang A, Cook N, Donnelly
M, Clancy R, Abend NS. (2011) Electrographic seizures during thera-
peutic hypothermia for neonatal hypoxic-ischemic encephalopathy.
J Child Neurol 26:724728. Epub 2011 Mar 29.
Yager JY, Armstrong EA, Miyashita H, Wirrell EC. (2002) Prolonged neo-
natal seizures exacerbate hypoxic-ischemic brain damage: correlation
with cerebral energy metabolism and excitatory amino acid release.
Dev Neurosci 24:367381.
557
Temporal Evolution-EEGSeizures in HIE
Epilepsia, 53(3):549557, 2012
doi: 10.1111/j.1528-1167.2011.03401.x