Perinatal stroke, a focal vascular brain injury occurring between 20 weeks’ gestation and 28 days of life,Reference Raju, Nelson, Ferriero and Lynch 1 occurs in 1:1500 to 1:5000 live births.Reference Laugesaar, Kolk and Tomberg 2 , Reference Lee, Croen and Backstrand 3 Neuroimaging methods of distinguishing perinatal stroke syndromes have been established,Reference Kirton, deVeber, Pontigon, MacGregor and Shroff 4 allowing advancement of syndrome-specific outcome research.Reference Kirton, deVeber, Pontigon, MacGregor and Shroff 4 - Reference Golomb, Garg, Carvalho, Johnson and Williams 6 Despite similar timing of brain injury, cognitive and behavioural outcomes vary widely, even in patients with comparable size and type of injury.Reference Westmacott, MacGregor, Askalan and deVeber 7 , Reference Ricci, Mercuri and Barnett 8 Although mean cognitive function testing is average to below average across groups, severe global difficulties with behavioural and cognitive dysfunction adversely impact the lives of 10% to 20% of children.Reference Westmacott, MacGregor, Askalan and deVeber 7 - Reference Ballantyne, Spilkin, Hesselink and Trauner 10 Explanations for these different outcomes are not understood and no modifiable factors have been identified.
Epilepsy occurs in 15% to 50% of perinatal stroke survivors.Reference Wusthoff, Kessler and Vossough 11 , Reference Wanigasinghe, Reid, Mackay, Reddihough, Harvey and Freeman 12 Epilepsy has been modestly associated with worse cognitive and developmental outcomes in two perinatal stroke studies.Reference van Buuren, van der Aa and Dekker 9 , Reference Ballantyne, Spilkin, Hesselink and Trauner 10 However, the role of epileptic activity during sleep has not been previously explored. Continuous spike-wave in slow wave sleep (CSWS) is an epileptic encephalopathy featuring electrical status epilepticus in sleep (ESES) associated with delays or regression in language, cognitive, and behavioural function.Reference Margari, Buttiglione, Legrottaglie, Presicci, Craig and Curatolo 13 Many authors use the terms CSWS and ESES interchangeably, and the two descriptions are likely a spectrum of similar pathology.Reference Kramer, Sagi, Goldberg-Stern, Zelnik, Nissenkorn and Ben-Zeev 14 - Reference Sanchez, Loddenkemper, Peters and Kothare 16 Original descriptions of ESES defined the electroencephalogram (EEG) abnormality as a bilaterally symmetric spike and wave discharge that occupies at least 85% of slow wave sleep and that commonly occurs in preschool children.Reference Patry, Lyagoubi and Tassinari 17 Others have since recognized that focal or unilateral ESES and/or discharges present for less than 85% of the recording may also contribute to neurocognitive dysfunction.Reference Margari, Buttiglione, Legrottaglie, Presicci, Craig and Curatolo 13 , Reference Kramer, Sagi, Goldberg-Stern, Zelnik, Nissenkorn and Ben-Zeev 14 , Reference Scheltens-de 18 , Reference Fernandez, Peters and Takeoka 19 The most recent, revised terminology on epilepsy by the International League Against Epilepsy does not list diagnostic criteria for CSWS/ESES,Reference Berg, Berkovic and Brodie 20 and spike-wave index cutoffs continue to be debated.Reference Scheltens-de 18 ESES and similar syndromes have been described in children with early brain injuries, particularly of the thalamus, including neonatal deep sinovenous thrombosis and hemorrhagic stroke.Reference Kersbergen, de Vries and Leijten 21 - Reference Losito, Battaglia and Chieffo 23
The objective of this study was to determine the association between the presence of ESES and developmental outcome based on formal neuropsychological scores in children with arterial perinatal stroke.
Methods
Population
In this retrospective, cross-sectional study, patients with perinatal arterial ischemic stroke were identified through the Alberta Perinatal Stroke Project (APSP). Established in 2008, the APSP is a population-based registry that has enrolled all prospectively diagnosed cases since 2008. Exhaustive review of >80 International Classification of Diseases 9 and 10 codes were combined with imaging analysis and in-person review to confirm all retrospective cases from 1992 through 2012. Perinatal stroke diagnosis required magnetic resonance imaging (MRI) confirming ischemic perinatal stroke classified by validated methods.Reference Kirton, deVeber, Pontigon, MacGregor and Shroff 4 Because differing stroke syndromes have different cognitive outcomes, for the current study, only patients with arterial stroke lesions were included. These were classified as either neonatal arterial ischemic stroke (NAIS) if they presented with acute neurological symptoms within 28 days of birth and had acute infarction on MRI, or arterial presumed perinatal ischemic stroke (APPIS) if presentation occurred later in infancy or childhood (typically with congenital hemiparesis) with remote arterial infarction on MRI. Inclusion criteria for the current study were MRI-confirmed NAIS or APPIS, EEG including sleep, and neuropsychological evaluation (4 years or older). The study was approved by the institutional ethics review board (ID 21522).
EEG
Patients were included if they had at least one clinical EEG performed after the neonatal period that captured sleep. All EEG reports were interpreted by the lead investigator (AM) who classified the original reports provided by epilepsy experts. A second, certified neurophysiologist and epilepsy expert (LB-E), who was blinded to neuropsychological outcomes, performed an independent review and classification of the original studies available. EEG reports at our centre often include a description of spike and wave discharges as a frequency of less than or greater than 75% or a qualitative description of “continuous” discharges in sleep. Because CSWS can refer to the clinical syndrome of epileptic encephalopathy with ESES documented on electrophysiology, for the purpose of this study, we classified EEG as either ESES or non-ESES. ESES was determined from EEG reports as either (1) a report of spike and wave discharges occurring in >75% of sleep or (2) the report of “continuous spike and wave” or “electrical status epilepticus in sleep.” A participant was included in the ESES group if at least one EEG met these criteria.
Neuropsychological Testing
A neuropsychological test battery was performed under the supervision of a clinical neuropsychologist where clinically indicated by the patient’s treating neurologist. Because these were clinical assessments, different versions of tests were used depending on the child’s age (e.g. Wechsler D. Wechsler Preschool and Primary Scale of Intelligence-III if younger than age 6 years and Wechsler D. Wechsler Intelligence Scale for Children, Fourth Edition, if age 6 and older), the needs of the child, or the specific referral question. Patients younger than 4 years of age were not included because of the limited number of tests available for that age range.
Domains of interest included in this study are overall intellectual functioning, language, reasoning, visual-construction, verbal learning and memory, visual learning and memory, executive functions, and behaviour. Table 1 contains a list of tests and subtests used to address functioning organized by cognitive domain. Percentile scores based on each test’s population norms were used as continuous dependent variables for each subset of tests. To reduce the influence of psychomotor speed and working memory on measures of intellectual abilities, the Canadian Wechsler General Ability Index was selected as a global estimate of intellectual functioning based on verbal and nonverbal skills. For some children, the Canadian Wechsler General Ability Index was estimated using either a two-subtest version (vocabulary plus matrix reasoning) with formulas provided by Sattler (2001)Reference Sattler 24 or a four-subtest version (vocabulary, similarities, matrix reasoning, and block design) with proration completed by the scoring program. All children that completed the Developmental Neuropsychological Assessment-2 word generation subtest were older than 7 years of age. Children older than 8 years of age completed the Delis-Kaplan Executive Function System letter fluency category fluency subtests.
Table 1 Neuropsychological tests used by cognitive domain of interest
BASC-2=Behavior Assessment System for Children, second edition; BRIEF=Behavior Rating Inventory of Executive Function; CBCL=Child Behavior Checklist; CMS=Children’s Memory Scale; CVLT-C=California Verbal Learning Test-Children’s Version; CVMT=Continuous Visual Memory Test; GAI=General Ability Index; DKEFS=Delis-Kaplan Executive Function System; NEPSY-II=A Developmental Neuropsychological Assessment; WISC-IVCDN=Wechsler D. Wechsler Intelligence Scale for Children, fourth edition, Canadian (ages 6-16:11); WJ-III=Woodcock-Johnson III Tests of Cognitive Abilities; WPPSI-IIICDN=Wechsler D. Wechsler Preschool and Primary Scale of Intelligence, third edition (ages 4-5: 11).
Statistical Analysis
Following confirmation of normal distribution (Shapiro-Wilk test), group comparisons between those with and without ESES (independent variable) were performed across neuropsychological test scores (dependent variable) using Student t tests. Mann-Whitney U test was used for group comparisons with non-normally distributed data. Bonferroni correction was performed to correct for multiple comparisons. Cohen’s d was used to determine the effect sizes between groups. A low, medium, and large effect size was set at d ≥0.2, 0.5, and 0.8, respectively. SPSS for Windows, v. 19.0, was used for analysis and the significance level was set at 0.05.
Results
Population
From the APSP database, 128 potentially eligible perinatal arterial ischemic stroke patients (73 NAIS, 55 APPIS) were identified (Figure 1). Of those, 37 (32 NAIS, 5 APPIS) were excluded because they were younger than age 4 years and 29 (12 NAIS,17 APPIS) because of a lack of adequate EEG data (neonatal or awake only). Of the remaining 62 patients (29 NAIS, 33 APPIS), 167 EEGs were reviewed, and 12 patients (19%) were classified as having ESES. Of those, three had severe developmental delays preventing formal neuropsychology evaluation. Of the 50 participants with non-ESES EEG, 22 had neuropsychology evaluations and were included. The final study population consisted of 34 participants (12 NAIS, 22 APPIS; 53% female). Sex did not differ between ESES versus non-ESES groups (p>0.8) and between NAIS and APPIS groups (p=0.8). All patients had anterior circulation stroke. Side of stroke did not differ between groups. The ESES group had more participants with bilateral stroke (3 ESES, 0 non-ESES; p<0.05). Clinical characteristics of the population are summarized in Table 2.
Figure 1 Study profile.
Table 2 Age of patients at time of EEG and neuropsychology testing
BH=bihemispheric epileptiform discharge; CLB=clobazam; CBZ=carbamazepine; F=female; FD=focal epileptiform discharges; FS=focal slowing; L=left; LEV=levetiracetam; L/R=both left and right; M=male; N=normal; NA=not applicable; NT=neuropsychological testing; OXC=oxcarbazepine; R=right; VPA=valproic acid;
${\bf{\bar{x}}} $
: mean.
*p<0.05.
EEG
A total of 167 sleep EEG recordings were scored. All EEGs were performed for the clinical indication of seizures, events suspicious for seizures, or follow-up of previous seizures. Most children had more than one recording (66%), with a mean of 2.7 studies. The mean age at time of EEG was 8.1 years (range, 0.2-16.4). ESES was diagnosed in at least one EEG of 12 children (19%) who had a mean age of diagnosis of 6.1 years (range, 2.1-10.3). Three of the 12 had bihemispheric ESES on EEG, with the rest being focal. Three patients with ESES had severe global developmental delay and could not participate in formal neuropsychology testing. They were excluded from the rest of the analysis. The majority (8/12) of children were taking at least one antiepileptic drug (AED) at time of ESES diagnosis. The medications included clonazepam (1), phenytoin (1), clobazam (2), oxcarbazepine (2), levetiracetam (1), and valproic acid (2). One patient was on dual treatment (clobazam and oxcarbazepine). Twenty-two children had EEG recordings that did not meet ESES study criteria but who had neuropsychology testing (mean age at EEG, 9.0 years; range, 0.2-16.4). The EEG characteristics are presented in Table 2. Patients with ESES were more likely to have the diagnosis of NAIS than APPIS (odds ratio, 6.8; 95% confidence interval [CI], 1.4-32.4; p=0.009) and to have their sleep EEG recorded at a younger age (mean, 6.1 years; 95% CI, 4.3-7.9 vs 9.0; 95% CI, 3.9-10.1, p=0.03).
Neuropsychology
The results of the neuropsychology assessments are illustrated in Figure 2 and Table 3. The median age at time of testing was 9.8 years (range, 4.4-16.7). Patients with ESES were significantly more likely to be taking prescribed AEDs at the time of neuropsychological testing (67% with ESES, 18% without ESES; p<0.01). Evaluation of all patients with perinatal arterial stroke revealed below-average scores in all cognitive domains compared with standardization values and externalizing behaviour, internalizing behaviour, and daily executive functioning were worse than peers (higher scores on these measures suggest more difficulties). The ESES group demonstrated marked impairments in all categories, scoring at the second percentile or worse. Medium to large effect sizes were noted in domains of letter fluency, receptive language, verbal learning, abstract reasoning, visual-constructional, visual memory, verbal-motor speed, externalizing behaviours, and daily executive functioning. Statistically significant lower scores (worse performance) in receptive language, verbal learning, verbal-motor speed, visual memory, and increased scores (more problems) in externalizing behaviour were noted in the ESES group as compared with non-ESES group. When correcting for multiple comparisons, differences in receptive language and externalizing behaviours remained statistically significant with worse scores in the ESES group.
Figure 2 Neuropsychology results. Boxplot demonstrating neuropsychology domain scores. *Statistically significant differences. †Those domains that remain statistically significant after Bonferroni correction.
Table 3 Neuropsychology testing results
M=median percentile ranks; IQR=interquartile range.
The values in this table represent percentile ranks, with the range being from 1 to 99.
*Statistically significant following Bonferroni correction. Higher percentile ranks represent better cognitive abilities, except for those measures† where higher percentile ranks represent more behaviour problems.
Discussion
Our data suggest that children with perinatal stroke and ESES have an increased risk of neuropsychological dysfunction, particularly in receptive language and externalizing behaviours. High effect sizes in these two domains as well as in visual memory and verbal-motor speed suggest that the differences in cognitive function may be broader than our sample size can determine with statistical significance. All participants had overall lower than average neuropsychological scores in most domains consistent with a broad spectrum of previously reported cognitive outcomes in perinatal stroke.Reference Westmacott, MacGregor, Askalan and deVeber 7 , Reference Ricci, Mercuri and Barnett 8
The first characterization of CSWS and ESES described a series of patients with potentiation of epileptiform discharges occupying >85% of the slow sleep recording. This pattern persisted on a total of three EEG recordings in 1 month.Reference Patry, Lyagoubi and Tassinari 17 The syndrome has been expanded in recent years to encompass patients with developmental delay or regression and marked activation of epileptiform discharges in sleep (<85%)Reference Margari, Buttiglione, Legrottaglie, Presicci, Craig and Curatolo 13 , Reference Kramer, Sagi, Goldberg-Stern, Zelnik, Nissenkorn and Ben-Zeev 14 , Reference Scheltens-de 18 , Reference Galanopoulou, Bojko, Lado and Moshe 25 or those with unilateral or focal potentiation of epileptiform discharges with sleep.Reference Fernandez, Peters and Takeoka 19 , Reference Paquier, Verheulpen, De and Van 26 Our findings of particular deficiencies in behaviour and language development are consistent with the historical descriptions of sleep-related epileptic encephalopathies as well as studies of both lesionalReference Kramer, Sagi, Goldberg-Stern, Zelnik, Nissenkorn and Ben-Zeev 14 , Reference Kersbergen, de Vries and Leijten 21 - Reference Losito, Battaglia and Chieffo 23 and nonlesionalReference Margari, Buttiglione, Legrottaglie, Presicci, Craig and Curatolo 13 , Reference Kramer, Sagi, Goldberg-Stern, Zelnik, Nissenkorn and Ben-Zeev 14 , Reference Kersbergen, de Vries and Leijten 21 , Reference Raha, Shah and Udani 27 ESES.
There has been additional recent focus on ESES in pediatric patients with early brain injuries. In the developing brain, thalamic injuries of heterogeneous etiologies have been associated with ESES.Reference Kersbergen, de Vries and Leijten 21 - Reference Losito, Battaglia and Chieffo 23 Disruption of thalamic oscillating rhythms seen in sleep and associated with spindles has been suggested as a potential pathological mechanism for ESES.Reference Sanchez, Loddenkemper, Peters and Kothare 16 Patients in this study had anterior circulation ischemic strokes, which do not result in direct lesions to the thalami. It is possible that secondary thalamic injury occurs through Wallerian degeneration and atrophy following cortical ischemic injury. It is also possible that disruption of the thalamocortical connections or epileptogenic mechanisms entirely remote from the thalamus may result in ESES susceptibility. Comparative studies across these populations may yield further insight toward underlying mechanisms and perhaps aid in better identifying children at risk.
Sleep-related epileptic encephalopathy may hinder brain development. Slow wave sleep has been implicated in recuperative function and memory consolidation.Reference Cirelli, Faraguna and Tononi 28 Slow waves are hypothesized to result in a depression or downscaling of synapses in sleep.Reference Tononi and Cirelli 29 ESES results in a decrease in slow waves and may reflect a disruption of this downscaling.Reference Bolsterli, Schmitt and Bast 30 Our results support the adverse effects of ESES on neurodevelopment, perhaps explaining in part the highly diverse neuropsychological outcomes seen in children with perinatal stroke, in which lesion timing, size, and location are often very similar.
Age at both EEG and neuropsychological testing are important considerations that could not be controlled for in this study because of its retrospective design. The mean age at time of documented ESES in our study was 6.1 years (95% CI, 4.3-7.9), which is similar to previously reported averages of 3.5 to 7.9 years.Reference Kramer, Sagi, Goldberg-Stern, Zelnik, Nissenkorn and Ben-Zeev 14 , Reference Fernandez, Peters and Takeoka 19 , Reference Sanchez, Takeoka and Tas E 22 , Reference Raha, Shah and Udani 27 A recent meta-analysis suggests that ESES may be seen earlier in lesional patients with a mean of 2.0 years.Reference Sanchez, Loddenkemper, Peters and Kothare 16 Furthermore, ESES severity can vary over time and from one recording to another.Reference Fernandez, Peters and Hadjiloizou 31 A potential period of vulnerability and a critical time for investigation may exist. Children in whom there is an index of suspicion may benefit from evaluation at several time points during the ages of highest risk. Clinical suspicion would be maximal in any child demonstrating regression or plateauing of their development. However, because many children with arterial perinatal stroke have motor deficits only, any early delays in nonmotor development would seem a reasonable indication for screening with sleep EEG. Neuropsychology evaluation of children with perinatal stroke focusing on language and behaviour domains might also help identify patients at risk.
In our study, participants with ESES were more likely to have a diagnosis of NAIS compared with APPIS. The presenting symptoms that differentiate these two syndromes may be association with future risk of epileptic encephalopathy. NAIS presents with symptomatic neonatal seizures in the vast majority, whereas most APPIS presents with motor asymmetry at 4 to 6 months of age.Reference Kirton, Armstrong-Wells and Chang 32 , Reference Kirton, Shroff, Pontigon and deVeber 33 In childhood stroke, acute seizures have been associated with the later development of epilepsy.Reference Fox, Glass, Sidney, Lowenstein and Fullerton 34 Stroke side did not differ between the two groups. However, bilateral strokes were more common in the ESES group, which may indicate more severe injury and increased risk of developing epileptic encephalopathy. Identification of clinical markers associated with epileptic encephalopathy after perinatal stroke could inform screening strategies. Future volumetric studies could further determine how stroke size may influence epilepsy, EEG, and neurodevelopmental outcome.
Successful treatment of ESES is often difficult, and multiple approaches have been reported. A recent pooled analysis of 575 cases reported EEG or developmental improvement with AEDs (49%), benzodiazepines (68%), corticosteroids (81%), and surgery (90%).Reference van den, van and Sagi L 35 With focal, well-circumscribed lesions, perinatal stroke patients may make excellent surgical candidates.Reference Battaglia, Veggiotti and Lettori 36 To date, minimal evidence has been published regarding specific treatment for lesional patients with ESES. Although cognitive impairments have been noted in patients with ESES, many patients continue to have significantly low cognitive function despite interventions.Reference Japaridze, Menzel, von, Steinmann and Stephani 37 Timing of potential interventions may be important in improving developmental trajectories.
Our results demonstrate that, at time of neuropsychological testing, patients with ESES were more likely to be taking AEDs. It is impossible to determine the effect of medications in this context. Patients with epileptic encephalopathy may be more likely to be taking medication because of active seizures. The effect of the medication on cognitive outcome in this context is also unknown. The medications may contribute to cognitive slowing or improve cognitive outcome secondary to improved seizure control.
One of the limitations of this study is the exclusion of younger patients resulting from limitations of neuropsychological testing. Three patients with ESES could not be tested but, based on their cognitive status, would lower neuropsychological outcome scores for the ESES group. Neuropsychology evaluation in our centre is usually performed when there is suspicion of delays and may explain the lower than average scores in all domains for all patients. The degree of cognitive difficulty in the perinatal stroke population overall may be less severe than reported here.
Additional limitations in this study include a variable time lag between EEG and neuropsychological assessment in some cases, the retrospective design, the absence of a control group (normative data used instead), and initial EEG interpretation by several different electrophysiologists. We were unable to control for seizure frequency or epilepsy diagnosis because data were not available. The small sample size also prevented analysis of medication effects and other treatment interventions. The difference in age at time of EEG between the ESES and non-ESES groups may reflect that EEGs were performed outside the window for ESES and patients with previous ESES might have been missed. This error would underestimate the difference in neuropsychology scores because patients with remote ESES would be included in the non-ESES group. This is more likely in patients with APPIS than NAIS who were slightly older at the time of their EEG.
This study was population-based with thorough capture of the stroke population, complete and detailed neuropsychological data, and precise definition of perinatal stroke syndromes using imaging classification. It contributes to an understanding of potential electrophysiology factors that may contribute to worse neurocognitive outcomes in patients with perinatal stroke and provides a clinically accessible tool to follow patients at risk. This knowledge justifies screening of at risk patients using sleep-deprived EEG to assess for ESES and neurocognitive assessments for planning appropriate learning and behavioural interventions. Future research should focus on the potential impact of treatment interventions.
Disclosures
The authors declare no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Statement of Authorship
AM compiled the data, performed analysis, and wrote the final manuscript. WQ analyzed data and contributed to figures and reviewing the manuscript. HLC and BLB compiled and analyzed the neuropsychology testing and reviewed the manuscript. LB-E compiled and analyzed the electrophysiology data and reviewed the final manuscript. AK supervised data collection and analysis and critically reviewed and revised the final manuscript.
