Cognitive outcome following unilateral arterial ischaemic stroke in childhood: effects of age at stroke and lesion location

Authors


  • Corrections added after online publication 20 August 2009: The positions of the last two authors’ names have been exchanged.

Dr Robyn Westmacott at Department of Psychology, Hospital for Sick Children, 555 University Avenue, Toronto, ON, Canada M5G 1X8. E-mail: robyn.westmacott@sickkids.ca

Abstract

Aim  Plasticity in the developing brain is a controversial issue. Although language and motor function often recover remarkably well following early brain injury, recent evidence suggests that damage to the developing brain results in significant long-term neuropsychological impairment. Our aim was to investigate the relationship among age at injury, lesion location and intellectual outcome.

Method  Using age-appropriate Wechsler scales of intellectual ability, we explored this issue by evaluating a large group (n=145) of children (89 males, 56 females) who experienced unilateral arterial ischaemic stroke during the perinatal period (diagnosed mean 73d, SD 29d), between the ages of 1 month and 5 years (mean 2y 10mo, SD 1y 9mo), or between the ages of 6 and 16 years (mean 11y 1mo SD 3y 6mo). The mean age at assessment was 8 years (SD 3y 10mo) in the perinatal group, 7 years 5 months (SD 2y 9mo) in the 1 month to 5 years group, and 12 years 5 months (SD 3y 9mo) in the 6 to 16 years group. The mean time interval between stroke and assessment was 8 years (SD 18d) for perinatal, 4 years 6 months (SD 1y 5mo) for 1 month to 5 years, and 1 year 4 months (SD 2y 9mo) for 6 to 16 years. The relationship between age at stroke and lesion location (subcortical, cortical, or combined) as it pertains to cognitive outcome was also examined.

Results  Measures of overall intelligence, verbal ability, working memory, and processing speed were significantly lower in children who had had a stroke than in the normative sample (all z>2.5, all p<0.01). The perinatal group performed more poorly than the other two groups on most cognitive measures, regardless of lesion location. The combined lesion location group performed more poorly than those with damage to either cortical or subcortical areas alone. Further investigation revealed different periods of peak vulnerability for subcortical lesions (perinatal) and cortical lesions (1mo–5y).

Interpretation  Lesion location modulates the relationship between age at stroke and cognitive outcome.

LIST OF ABBREVIATIONS
AIS

Arterial ischaemic stroke

PRI

Perceptual Reasoning/Organization Index

PSI

Processing Speed Index

VCI

Verbal Comprehension Index

WMI

Working Memory/Freedom from Distractibility Index

Despite the widely held belief that the increased plasticity of the young brain protects against the effects of injury, there is now considerable evidence that early brain injury results in significant long-term neuropsychological impairment. Although children are less likely than adults to exhibit specific lateralized cognitive deficits that map on to lesion location, an increasing body of research suggests that early brain injury – particularly before 5 or 6 years of age – often leads to widespread cognitive dysfunction across multiple domains.1 The most convincing evidence for this ‘early vulnerability hypothesis’ comes from children with diffuse brain injuries, such as cranial irradiation2 and traumatic brain injury,3 and children with chronic neurological conditions such as epilepsy.4 In these populations, brain damage early in infancy or childhood is more detrimental than brain damage later in childhood to overall cognitive ability,5 academic skills,6 verbal ability,7 attention,8 and executive function.9

In contrast, the literature on focal paediatric brain injury is considerably more mixed. Consistent evidence of the remarkable sparing of language function following early left hemisphere lesions has fuelled the ‘plasticity hypothesis’,10,11 although more recent studies have found that deficits in higher-level aspects of language processing (discourse, complex syntax, making inferences) often emerge later on in development in children with early focal lesions.12 Moreover, outside the domain of language, studies examining higher-level verbal, visual–spatial executive, and informational processing abilities also suggest that early focal brain lesions have adverse effects that emerge later in development and affect multiple cognitive domains.13 Attempts to examine the impact of age at focal brain injury on cognitive outcome directly have produced inconsistent findings, probably because of within-study and between-study heterogeneity in terms of aetiology, age at assessment, lesion size, lesion location, the specific cognitive skill in question and the cognitive measures used. Some have found that the first year or two of life is the period of highest vulnerability,14 others have found the poorest outcome to be when focal lesions occur between 1 month and 5 years of age,15 and still others have found no clear relationship between age at injury and cognitive outcome.1 These contradictory findings suggest that multiple factors interact with age at focal brain injury to impact specific cognitive outcomes.16

The aim of this study was to investigate the impact of age at focal brain injury on intellectual outcome in a large group of children with unilateral arterial ischaemic stroke (AIS). This population is ideal for studying the impact of focal brain injury on development because, for most children, the time of onset can be pinpointed exactly and many of these children have no other neurological comorbidity. Moreover, because AIS can occur at any time, from the prenatal period to adolescence, it is possible to compare groups of children with virtually identical lesions that occurred during different developmental periods. We explored the long-term cognitive impact of focal brain injury during the perinatal period (from 20wks’ gestation), early childhood (1mo–5y), and later childhood (6–16y), as well as the relationship among age at injury, lesion location (subcortical, cortical, combined, and intellectual outcome). We hypothesized that earlier age at focal brain injury is associated with poorer cognitive outcome than later injury, and that the relationship between age at injury and outcome varies depending on lesion location and extent.

Method

Participant population

Participants were identified upon enrolment in the Children’s Stroke Outcome Study at the Hospital for Sick Children in Toronto, Canada.17 Children were considered for participation if they had a history of AIS diagnosed before the age of 16 years, with a single unilateral infarct documented on magnetic resonance imaging (MRI) or computed tomography (CT). Cognitive assessments were carried out at least 6 months post-stroke in all cases. Exclusion criteria were bilateral or multiple lesions, recurrent stroke, cerebral sinovenous thrombosis, hypoxic–ischaemic encephalopathy, preterm birth (<36wks’ gestation), sickle cell disease, moyamoya disease, malignancy, epilepsy (either pre-existing or onset post stroke), and other neurological comorbidities. Children who were not fluent in English were also excluded. This study was approved by the Research Ethics Board at the Hospital for Sick Children and consent was obtained from all participants or their caregivers. Consent includes future contact for research purposes.

To ensure unbiased enrolment in the study, we implemented a telephone recruitment procedure for all children with unilateral AIS seen at our institution between 1994 and 2007 whose guardians had given consent to be contacted for research. Participants were stratified by age at stroke as follows: perinatal – stroke occurrence in the prenatal period or in the first 28 days of the postnatal period; 1 month to 5 years; and 6 to 16 years. Although there is no consensus regarding the cut-off that should be used to distinguish between early and later onset of childhood brain injury, 5 years of age is the cut-off point used most consistently in previous studies.18 Participants were also stratified by lesion location/extent as follows: subcortical – infarct restricted to basal ganglia and/or thalamus; cortical – cortical infarct with no subcortical involvement; and combined – infarct involving cortex plus basal ganglia and/or thalamus. The study neurologists reviewed the MRIs and CTs at clinic visits and coded stroke lesion characteristics (including size and location), referring to clinically generated neuroimaging reports. Similarly to previous studies of paediatric focal brain injury,18 lesion size was categorized in the following manner: small – infarct involving <1/10 of parenchymal volume; medium – infarct involving 1/10 to 25/100 of parenchymal volume; and large – infarct involving >25/100 of parenchymal volume. Volumes were determined by anatomical tracing using the ImageJ software package.

We determined the demographic and neurological characteristics of participants included in the study from a review of health records and structured parental interviews. Neurological status is indicated by scores on the Paediatric Stroke Outcome Measure19 obtained within 1 year of the neuropsychological assessment. Maternal education level was used as an indicator of socioeconomic status and rated on a five-point scale (1=did not complete high school; 2=completed high school; 3=some postsecondary training but not a diploma or degree; 4=completed a university or college programme; 5=professional or graduate school). The three age-at-stroke groups were matched on maternal education and neurological outcome (Paediatric Stroke Outcome Measure scores). Age at test was matched for the perinatal group and the 1 month to 5 years group, but the 6 to 16 years group was significantly older than the other two at the time of study (p<0.01). However, age at test was matched across the three lesion location groups. With respect to time since stroke, time between stroke and assessment was significantly longer in the perinatal group than in the 6 to 16 years group (p<0.05); however, the perinatal and 1 month to 5 years groups did not differ significantly on this factor. Similarly, there was no time since stroke difference across the three lesion location groups.

Neuropsychological testing

Cognitive assessment was performed using the Wechsler Intelligence Scale for Children, 3rd edition (WISC-III) or the 4th edition (WISC-IV). Both versions of the WISC provide index scores for overall intellectual ability (Full- scale IQ [FSIQ]), verbal ability (Verbal Comprehension Index [VCI]), non-verbal ability (Perceptual Reasoning/Organization Index [PRI]), auditory attention and mental manipulation (Working Memory/Freedom from Distractibility Index [WMI]), and visual–motor speed (Processing Speed Index [PSI]), with a mean score of 100 and a standard deviation (SD) of 15. Scores between 90 and 110 are considered to fall within the ‘average range’. Children aged 4 to 5 years completed either the Wechsler Preschool and Primary Scale of Intelligence Revised Edition (WPPSI-R) or the WPPSI 3rd edition (WPPSI-III). Both versions of the WPPSI provide index scores for overall intellectual ability (FSIQ), verbal ability (VIQ), and non-verbal ability (Performance [PIQ]), with a mean score of 100 (SD 15). Owing to the rarity of paediatric stroke, we chose to maximize our sample size by including participants who were tested over a period of 13 years and received different versions of the Wechsler battery. This is common practice in large-scale neuropsychological studies, and the rigorous process of test development (including convergent validity between new and old versions) minimizes the associated methodological confounds.

Statistical analysis

First, one-sample t-tests with an α of 0.05 were used to compare the groups’ WPPSI and WISC Index scores with the theoretical mean of the normative sample (i.e. mean=100, SD=15). In addition, two-tailed independent-samples t-tests were used to explore whether or not there were any significant differences based on sex or lesion laterality (right vs left). Next, a two-factor multivariate analysis of variance (MANOVA) was carried out on the five index scores (FSIQ, VIQ/VCI, PIQ/PRI, WMI, and PSI) to examine the main effects of age at stroke and lesion location and their interaction. Post-hoc t-tests were used for pairwise comparisons within each independent variable, controlling for type I error (Scheffé test). Single-factor MANOVAs and post-hoc tests were then carried out to examine the effect of age at stroke within each of the lesion location groups separately (subcortical, cortical, and combined). Dividing the sample by lesion location reduced the statistical power considerably, so effect sizes (η2) were calculated for these analyses as well. The η-values were interpreted as follows: 0.01=small effect size; 0.06=medium effect size; and 0.14=large effect size.

Results

Study cohort

A total of 222 children with unilateral AIS who met our inclusion and exclusion criteria were enrolled in the Children’s Stroke Outcome Study. Twenty participants could not be contacted because of phone number and address changes, nine declined participation because they lived more than 3 hours’ drive from Toronto, and 10 declined to participate because of lack of interest. Finally, 35 children were too young (i.e. <4y) to perform standardized tests of intellectual ability and three children were not testable owing to non-compliance or global developmental delay. A total of 145 children participated in the study. The demographics and neurological status of these children are shown in Table I. No participant had seizures that persisted beyond the acute stroke period, although 89 participants did have one or more seizures at the time of the stroke.

Table I.   Demographic and neurological characteristics of the participant group, by age at stroke
 All participantsPerinatal1mo–5y6–16y
  1. aMaternal education level is rated on a five-point scale (see text for details). bNeurological outcome is measured by the Paediatric Stroke Outcome Measure: The number given indicate the percent of subjects who had no neurological deficits in sensorimotor function, language, cognition or behaviour for each domain, 0=normal, 0.5=mild deficit, 1=moderate deficit, 2=severe deficit. Sensorimotor scores are presented for the side affected by the stroke. Because all participants had unilateral lesions, all had sensorimotor scores of 0 for the unaffected side. WISC, Wechsler Intelligence Scale for Children; WPPSI, Wechsler Preschool and Primary Scale of Intelligence; PSOM, Paediatric Stroke Outcome Measure.

Total number145465742
Males/Females89/5628/1834/2327/15
Maternal educationa, mean (SD)3.88 (0.80)3.96 (0.76)3.81 (0.71)3.86 (0.63)
Age at stroke, mean (SD)4y 5mo (5y)73d (29d)2y 10mo (1y 9mo)11y 1mo (3y 6mo)
Age at test, mean (SD)9y 1mo (3y 10mo)8y 0mo (3y 10mo)7y 5mo (2y 9mo)12y 5mo (3y 9mo)
Lesion location
 Subcortical64 93322
 Cortical4224 9 9
 Combined39131511
Lesion size
 Small83124031
 Medium33121110
 Large2922 5 2
Test version
 WISC-II51122019
 WISC-IV64172423
 WPPSI-R12 6 6 0
 WPPSI-III1811 7 0
Neurological outcome (% with no deficits)
31%32%30%30%

One hundred and fifteen children completed either WISC-III (n=51) or WISC-IV (n=64). The remaining 30 children were 4 to 5 years old at the time of the study and completed either the WPPSI-R (n=12) or the WPPSI-III (n=18).

Stroke participants compared with the normative sample

Consistent with previous studies of intellectual function after paediatric stroke, mean scores for the participant groups fell at the low end of the average range (Table II). All five index measures for the participant groups were significantly lower in comparison with data from a normative sample (all p<0.01). Each age-at-stroke group was compared with the normative sample (Table II). All five index measures were significantly lower in the perinatal group than in the normative sample. In the 1-month-to-5-years group, FSIQ, WMI, and PSI, were significantly lower than in the normative sample, whereas the 6 to 16 years group was significantly weaker only on PSI. Finally, each lesion location group was compared with normative data (Table II). PSI, but not any of the other index measures, was significantly lower in the subcortical group for. FSIQ, PIQ, WMI, and PSI were significantly lower in the cortical group, whereas all five index measures were significantly lower in the combined group.

Table II.   Intellectual outcome for children with unilateral arterial ischaemic stroke, stratified by age at stroke and by lesion location, mean (SD)
GroupAge at testYears since strokeFSIQVIQ/VCIPIQ/PRIWMIPSI
  1. Scores that are significantly lower than normative data are denoted as follows: ap<0.01; bp<0.05. FSIQ, Full-scale IQ; PIQ, Performance IQ; PRI, Perceptual Reasoning Index; PSI, Processing Speed Index; VCI, Verbal Comprehension Index; VIQ, Verbal IQ; WMI, Working Memory Index.

All9y1m (3y10m)4y9m (3y9m)94.74a (14.70)95.99a (14.41)96.48a (14.54)93.69a (15.09)92.14a (13.91)
Perinatal8y0m (3y1m)8y0m (3y1m)91.63a (14.30)92.20a (14.17)94.33b (13.31)88.10a (16.00)89.63a (14.44)
1mo – 5y7y5m (2y9m)4y6m (3y1m)95.42b (15.65)97.60 (15.20)96.37 (14.62)93.98b (14.92)93.31a (12.56)
6–16y12y6m (3y9m)1y6m (1y10m)97.21 (13.51)97.98 (13.04)99.00 (15.62)97.40 (13.67)93.17a (14.89)
Subcortical8y9m (4y3m)3y6m (3y0m)98.23 (14.68)99.25 (13.68)101.12 (13.72)96.49 (15.34)95.79b (13.39)
Cortical9y3m (3y4m)4y9m (4y0m)95.12b (12.86)96.10 (13.15)95.03b (14.26)95.58b (15.53)92.24a (13.77)
Combined9y4m (3y7m)6y6m (3y10m)87.95a (14.24)89.82a (14.74)90.09a (15.09)87.63a (12.81)84.38a (12.75)

Age at stroke effects

Levene’s test revealed homogeneity of variance across the three age-at-stroke groups. There was a significant main effect of age at stroke for FSIQ, VIQ, and WMI, but not for PIQ or PSI (Table III). Post-hoc tests revealed that VIQ was significantly lower in the perinatal group than in the 1 month to 5 years group, but there was no significant difference between these two groups on any other index measure. When compared with the 6 to 16 years group, FSIQ, VIQ, and WMI were significantly lower in the perinatal group. There was no significant difference between the 1 month to 5 years group and the 6 to 16 years group.

Table III.   Main effect of age at stroke on the five Wechsler index measures, mean (SD)
Index measurePerinatal 1mo–5y 6–16yFp
  1. aPost-hoc tests revealed a significant difference between the perinatal and the 6 to 16 years groups (p<0.05). bPost-hoc tests revealed a significant difference between the perinatal and the 6 to 16 years groups (p<0.05) and between the perinatal and 1 month to 5 years groups (p<0.05). FSIQ, Full-scale IQ; PIQ, Performance IQ; PRI, Perceptual Reasoning Index; PSI, Processing Speed Index; VCI, Verbal Comprehension Index; VIQ, Verbal IQ; WMI, Working Memory Index.

FSIQa91.63 (14.30)95.42 (15.65)97.21 (13.51)2.880.05
VIQ/VCI92.20 (14.17)97.60 (15.20)97.98 (13.04)2.090.05
PIQ/PRI94.33 (13.31)96.37 (14.62)99.00 (15.62)1.900.15
WMI88.10 (16.00)93.98 (14.92)97.40 (13.67)4.950.01
PSI89.63 (14.44)93.31 (12.56)93.17 (14.89)0.530.59

Lesion location effects

Lesion location was distributed differently across the three age-at-stroke groups. Basal ganglia/thalamic infarcts primarily occurred in the 1 month to 5 years and 6 to 16 years groups, whereas perinatal lesions were typically restricted to the cortex. This pattern is consistent with epidemiological studies of paediatric stroke, demonstrating a differential distribution of lesion locations in perinatal/neonatal stroke and childhood stroke.20

Levene’s test revealed homogeneity of variance across the three lesion location groups. There was a significant main effect of lesion location for all five index measures, even when the effect of lesion size was partialled out of the analysis (Table IV). Post-hoc tests revealed that all index measures were significantly lower in the combined group than in either the subcortical group or the cortical group. There was no significant difference between the subcortical and cortical groups.

Table IV.   Main effect of lesion location on the five Wechsler index measures after partialling out lesion size, mean (SD)
Index measureSubcorticalCorticalCombinedFp
  1. aPost-hoc tests revealed a significant difference between the subcortical and combined groups (p<0.01) and between the cortical and combined groups (p<0.05). bPost-hoc tests revealed a significant difference between the subcortical and combined groups (p<0.05) and between the cortical and combined groups (p<0.05). FSIQ, Full-scale IQ; PIQ, Performance IQ; PRI, Perceptual Reasoning Index; PSI, Processing Speed Index; VCI, Verbal Comprehension Index; VIQ, Verbal IQ; WMI, Working Memory Index.

FSIQa98.23 (14.68)95.12 (12.86)87.95 (14.24)4.690.01
VIQ/VCIa99.25 (13.68)96.10 (13.15)89.82 (14.74)4.130.02
PIQ/PRIa101.12 (13.72)95.03 (14.26)90.09 (15.09)3.640.05
WMIb96.49 (15.34)95.58 (15.53)87.63 (12.81)3.210.05
PSIa95.79 (13.39)92.24 (13.77)84.38 (12.75)5.660.01

Age at stroke and lesion location interaction

The two-factor MANOVA revealed a significant interaction between age at stroke and lesion location (F(5,104)=3.42, p<0.05). Single-factor MANOVAs and post-hoc multiple comparisons were carried out to examine the effect of age at stroke for each lesion location group separately. Within the subcortical group, there were significant main effects of age at stroke for FSIQ, VIQ, PIQ, and WMI, but not PSI (Table SI, supporting information published online). η2 values revealed medium effect sizes for FSIQ, VIQ, PIQ, and WMI, and a small effect size for PSI. Post-hoc multiple comparisons revealed that the perinatal group was significantly weaker than the 1 month to 5 years group and the 6 to 16 years group on all index measures except PSI. There was no significant difference between the 1 month to 5 years group and the 6 to 16 years group.

Within the cortical group, there were significant main effects of age at stroke for FSIQ, PIQ, and WMI, but not VIQ or PSI (Table SII, supporting information published online). Again, eta2 values revealed medium effect sizes for FSIQ, VIQ, PIQ, and WMI, and a small effect size for PSI. Pairwise comparisons revealed a trend toward poorer performance in the 1 month to 5 years group than the other two groups for all index measures except PSI. However, few of these post-hoc tests reached significance owing to small sample sizes. The 1 month to 5 years group was significantly weaker than the 6 to 16 years group on FSIQ and PIQ, but the analyses for VIQ and WMI did not reach significance. The 1 month to 5 years group was also significantly weaker than the perinatal group on FSIQ, but none of the other analyses reached significance.

Finally, within the combined group, there was no significant main effect of age at stroke (Table SIII, supporting information published online), although there was a trend towards poorer performance in the perinatal group and the 6 to 16 years group compared with the 1 month to 5 years group. η2 values revealed small effect sizes for FSIQ, VIQ, PIQ, and PSI, and a medium effect size for WMI.

Of note, when the perinatal group was divided into those with acutely diagnosed neonatal stroke and those with retrospectively diagnosed presumed perinatal strokes, there was a non-significant trend towards weaker performance in the latter group. However, our statistical analyses suggested that this reflected a greater proportion of subcortical lesions in the presumed perinatal stroke group.

Effect of lesion laterality

Planned t-tests, corrected for multiple comparisons, were carried out to compare children with right hemisphere lesions and those with left hemisphere lesions on the five index measures. There was no significant difference or noteworthy trend pertaining to lesion laterality on any measures. When the age-at-stroke groups and lesion location groups were examined separately, again there was no significant laterality difference.

Sex differences

Planned t-tests, corrected for multiple comparisons, were carried out to compare males and females on the five index measures. There was no significant difference or noteworthy trend pertaining to lesion laterality on any measures. When the age-at-stroke groups and lesion-location groups were examined separately, again there was no significant sex difference.

Effects of age at assessment and time since stroke

Correlational analyses, corrected for multiple comparisons, were carried out to examine the effects of age at assessment and time since stroke on the five index measures. There was no significant correlation or noteworthy trend between age at assessment and any of the IQ index measures. However, time since stroke was negatively correlated with FSIQ (r=−2.09, p<0.05), VIQ/VCI (r=−1.87, p<0.05), and PIQ/PRI (r=−2.01, p<0.05), and there were non-significant trends in the same direction with WMI (r=−1.68, p=0.075) and PSI (r=−1.50, p=0.086). Of note, when the age-at-stroke groups were examined separately, there was no significant correlation or noteworthy trend between time since stroke and any of the IQ indices.

Discussion

To date, this is the largest study of cognitive outcome following paediatric AIS. We evaluated the impact of age at time of stroke on overall intellectual ability, verbal skills, perceptual reasoning skills, working memory, and processing speed in 145 children with a history of unilateral AIS during the perinatal period, between 1 month and 5 years of age, and between 6 and 16 years of age, none of whom had epilepsy or other neurological comorbidities. Moreover, we compared cognitive test performance across children with subcortical strokes, cortical strokes, and combined cortical–subcortical strokes.

Several important conclusions can be drawn from our study. First, consistent with the existing literature,13 the stroke group performed significantly worse than the normative sample on all five index measures of cognitive outcome, although mean group averages remained at the low end of the average range. Second, paediatric stroke involving both cortical and subcortical regions was more detrimental to cognitive outcome than stroke affecting either cortical or subcortical tissue, and this effect remained significant even after partialling out the contribution of lesion size. Thus, the capacity of the young brain to reorganize successfully appears to be diminished with combined damage to cortical and subcortical areas. Third, earlier age at stroke was associated with weaker cognitive performance overall, but the relationship was modulated by lesion location. Specifically, subcortical lesions (basal ganglia and/or thalamus) in the perinatal period appear to be particularly detrimental to future cognitive outcome. Children in whom subcortical stroke occurred before the age of 28 days performed significantly more poorly than children in whom similar strokes occurred between 1 month and 5 years of age or between 6 and 16 years of age on measures of overall intellectual ability and on all four of the major cognitive domains. In contrast, in the case of cortical strokes, the period of greatest vulnerability appears to be between 1 month and 5 years of age. Cortical strokes sustained during this period were associated with significantly weaker overall intellectual ability compared with earlier or later cortical strokes. Cortical strokes between 1 month and 5 years of age were also associated with significantly weaker perceptual reasoning skills compared with later cortical strokes. These significant differences are particularly noteworthy given the low statistical power of these analyses. Finally, combined cortical–subcortical strokes were associated with weak performance on all the cognitive measures, regardless of age at stroke. With respect to sex differences, we found no significant difference between males and females on any cognitive measure. Similarly, we found no significant effect of lesion laterality on any cognitive measure. This was consistent with previous studies.1

Plasticity versus early vulnerability

Two competing views of brain plasticity and reorganization have emerged in the research literature as a result of conflicting findings. The traditional plasticity hypothesis is supported by evidence of better recovery of ‘essential’ motor and language functions in very young children with focal brain injuries compared with older children and adults.21 Recent functional neuroimaging studies have provided further evidence of this plasticity and have also started to delineate possible mechanisms for the cerebral reorganization associated with successful recovery/development of language and/or motor function.22,23 However, there is an increasing body of evidence to suggest that plasticity and reorganization after early brain injury are not always associated with positive or adaptive long-term outcomes for higher-level cognitive abilities, giving rise to the early vulnerability hypothesis.24 Attempts to investigate which of these hypotheses is correct have continued to produce inconsistent findings, leading some researchers to propose that the relationship between age at injury and cognitive outcome is non-linear and probably modulated by other variables such as lesion type, lesion location, and the specific cognitive skill in question.25 Our results provide support for this view. Overall, we found evidence of early vulnerability, in that children with stroke at a later age (≥6y) performed better on measures of general IQ, verbal skills, and working memory. However, periods of peak vulnerability were dependent upon lesion location. Overall, our findings are consistent with the argument that the damaged immature brain is limited in its capacity to compensate and support the development of higher-level cognitive skills.26 Because brain development is protracted throughout childhood, with the development of later-maturing areas dependent upon proper development of early-maturing areas,27 damage to one brain region can disrupt development in other brain regions. Moreover, it has been argued that successful cognitive development depends upon the functional integrity of different brain structures at specific time points, such that plasticity and reorganization may not be capable of compensating for damage to a particular brain region during a ‘critical period’.28 This cascade of neurobiological and cognitive events can result in cumulative or ‘emerging’ deficits in higher-level cognitive skills. Thus, although there is considerable evidence that the immature brain is more plastic than the mature brain, increased neural plasticity does not necessarily translate into better functional outcome, particularly for higher-level and later-developing cognitive skills.26

Different periods of vulnerability for subcortical and cortical infarcts

Our data suggest that subcortical strokes are most detrimental to intellectual ability and information processing skills when they are sustained during the prenatal or perinatal period. Cortical strokes during this period were less likely to result in cognitive deficits. However, we found that children who experienced cortical strokes in early childhood (between 1mo and 5y) performed more poorly across multiple cognitive domains than those in whom cortical strokes occurred earlier or later in development. These findings suggest an earlier period of peak vulnerability for subcortical infarcts than for cortical infarcts. Because brain development typically proceeds from inferior to superior, and posterior to anterior, with subcortical structures developing earlier than cortical structures,29 our findings are consistent with the notion that vulnerability is highest during periods of rapid development.24 However, it is also possible that very early subcortical damage is particularly detrimental to cognitive development because it disrupts later cortical and white matter maturation.30 This explanation for our findings is consistent with evidence that normal development of early-maturing brain regions is essential for normal development of later-maturing brain regions.28 Finally, it is possible that the impact of lesion location depends not only on age at stroke, but also on age at test. In light of evidence that the cognitive effects of cortical lesions may emerge over time with the progression of cortical maturation, perhaps the relationship between lesion location and age at stroke would change if participants were tested as adolescents or young adults. We did not find a significant impact of age at test on cognitive performance; however, long-term follow-up studies are needed to fully understand development trajectories in children with stroke.

Limitations and future directions

Although this is the largest study to date examining cognitive outcome after paediatric AIS, there are some limitations to acknowledge. First, the age at stroke by lesion location subgroups are not equal in size, although this was unavoidable given the known confound between these two variables. In our sample, the perinatal group was composed of children with acutely diagnosed neonatal stroke who had primarily cortical lesions and children with retrospectively diagnosed presumed perinatal stroke, many of whom had prenatal subcortical lesions. Children in the presumed perinatal stroke group did perform more poorly than children in the cortical lesion group, but it is unclear if this was because of the timing of their lesions or a preponderance of subcortical lesions. Examining a larger group of children with perinatal stroke, and making a more definitive distinction between prenatal and neonatal stroke, may provide further clarity.

It is also important to note that the interval between stroke and assessment was significantly longer in the perinatal group than in the other two groups. This is relevant because we also found a significant negative correlation between time since stroke and cognitive performance on several of the IQ index measures. This suggests that the longer interval between stroke and assessment for the perinatal group than for the other groups may have contributed to our finding of poorer cognitive performance in the former. However, it is also possible that this was simply a spurious correlation and that perinatal stroke is associated with poorer cognitive outcome regardless of the interval between stroke and assessment. This hypothesis seems plausible in light of the fact that, when the age-at-stroke groups were examined separately, there was no significant correlation between time since stroke and cognitive performance on any measure. Clearly, this is a question that requires further investigation in future studies, with longer follow-up periods for participants with later age at stroke. Other limitations of our study are that we did not include a comparison group and we did not take into account the effect that interventions (e.g. speech–language therapy, tutoring, etc.) may have had on cognitive outcome.

Finally, we examined several distinct aspects of intellectual ability, but our study is certainly not an exhaustive examination of cognitive development. Future studies should examine other cognitive abilities such as learning, memory, attention, and executive function, as these skills may be differentially affected by the impact of age at stroke and lesion location. It will also be important to consider the role of psychosocial factors in recovery and development following paediatric stroke. There is considerable evidence that these factors are important in recovery from diffuse brain injury,31 but little is known about their impact on the focal lesion population.

Acknowledgments

Gabrielle deVeber, the senior author, was supported by a Stroke Investigator Award from the Heart and Stroke Foundation of Ontario. We also wish to thank Anita Allen and Ivanna Yau for coordination of patient referrals, as well as Sandra Newton, Sarah Reiss, Vanessa Barry, Amy Wilkinson and Liza Kouzmicheva for technical support in data collection.

Ancillary