OUP user menu

Children with new-onset epilepsy: neuropsychological status and brain structure

Bruce Hermann, Jana Jones, Raj Sheth, Christian Dow, Monica Koehn, Michael Seidenberg
DOI: http://dx.doi.org/10.1093/brain/awl196 2609-2619 First published online: 23 August 2006


Abnormalities in cognition, academic performance and brain volumetrics have been reported in children with chronic epilepsy. The nature and degree to which these problems may be present at epilepsy onset or may instead become more evident over time remains to be determined. This study characterizes neuropsychological status, brain structure and their interrelationship in children with recent-onset epilepsy compared with healthy controls. Children (age: 8–18 years) with recent-onset idiopathic epilepsy (n = 53) and healthy controls (n = 50) underwent comprehensive neuropsychological assessment and quantitative volumetric measurement of segmented (grey and white matter) volumes of total cerebrum and lobar regions. Compared with controls, children with recent-onset epilepsy exhibit a pattern of mild diffuse cognitive impairment, regardless of epilepsy syndrome, as well as academic underachievement that in a subset of children antedates the first recognized seizure. There are no overall differences in MR morphometric analyses of total cerebral or lobar tissue volumes. Controls show a strong association between cognitive development and increasing cerebral tissue volume (especially white matter volume), an association that is absent in children with epilepsy. Children with a history of academic achievement problems exhibit the most abnormal cognitive function and have significant volumetric reductions in left occipital and parietal lobe grey matter. Patients with idiopathic epilepsy exhibit cognitive dysfunction and academic underachievement at the onset of the disorder, irrespective of epilepsy syndrome, and indications of antecedent neurocognitive impairment are present in a subset of children. Volumetric abnormalities are not yet apparent in the epilepsy group as a whole, but there are indications of an altered structure–function relationship in epilepsy, and the subset of children with prior history of academic problems have abnormal volume of posterior left hemisphere grey matter. These early abnormalities need to be integrated into lifespan models of the neuropsychology of epilepsy.

  • epilepsy
  • neuropsychology
  • MR volumetrics


Neuropsychological impairment is an important co-morbidity of chronic epilepsy (Elger et al., 2004). A long and rich history of research has characterized the relationships between impaired cognition and a variety of clinical epilepsy factors including aetiology, age of onset, seizure type and severity, duration, anti-epilepsy medications and other factors (Saling et al., 1993; Jones-Gotman, 2000; Helmstaedter and Kurthen, 2001; Jokeit and Ebner, 2002; Aldenkamp and Arends, 2004; Dodrill, 2004). In addition, modal cognitive profiles have been derived for several syndromes of epilepsy, and efforts have been undertaken to identify the shared versus unique cognitive abnormalities evident across epilepsy syndromes (Lassonde et al., 2000; Nolan et al., 2003; Elger et al., 2004).

The nature, timing and course of cognitive impairment in epilepsy are issues of substantial concern, particularly the degree to which chronic medication-resistant epilepsy may lead to progressive cognitive impairment (Pitkanen and Sutula, 2002). While evidence to this effect has been reported (Dodrill, 2004), the early cognitive substrate upon which subsequent chronic epilepsy may exert its effects is an important consideration. Controlled studies of children and adolescents with chronic but substantially shorter durations of epilepsy have demonstrated considerable neuropsychological impairment (Farwell et al., 1985; Schoenfeld et al., 1999; Roeschl-Heils, 2002; Smith et al., 2002; Germano et al., 2005), suggesting the influence of an early adverse neurodevelopmental impact on cognition. However, in order to develop a clearer perspective of the potential progressive and lifetime neuropsychological consequences of the epilepsies, it is important to characterize this very early cognitive substrate. To that end, investigation of children with new-onset epilepsy offers to contribute significantly to this literature.

To date, a very small number of studies have examined cognition in children with new-onset epilepsy (Bourgeois et al., 1983; Stores et al., 1992; Williams et al., 1998; Kolk et al., 2001; Oostrom et al., 2003). Three of the five studies identified cognitive impairments at epilepsy onset, and mixed results may be attributable, at least in part, to the variable age ranges, test batteries and epilepsy characteristics across studies. Especially interesting are reports of academic underachievement before and/or at the onset of idiopathic epilepsy (Oostrom et al., 2003; Berg et al., 2005; McNelis et al., 2005), suggestive of an antecedent neurobiological insult of uncertain aetiology. The neuropsychological characteristics of these children remain to be characterized, and the question remains whether children without such histories would differ from healthy controls.

One factor that may underlie cognitive pathology in children with epilepsy is structural brain abnormality. Quantitative MR volumetrics have been used to characterize the nature and pattern of brain abnormality in adults with epilepsy, especially temporal lobe epilepsy (Kuzniecky and Knowlton, 2002; Bernasconi, 2004; Koepp and Duncan, 2004; Cendes, 2005). Volumetric anomalies are of clinical consequence, as demonstrated by their relationship with impaired cognition (Baxendale et al., 1998, 1999; Martin et al., 1999; Burneo et al., 2003; Dow et al., 2004; Griffith et al., 2004; Hermann et al., 2004; Seidenberg et al., 2005). In contrast, there are very few volumetric studies of children with epilepsy. The studies to date have involved only children with chronic epilepsy, and the findings reveal abnormalities in cerebrum, cerebellum and hippocampus (Lawson et al., 1997, 1998, 2000a, b, 2002). A recent voxel-based morphometric investigation of children with chronic temporal lobe epilepsy reported a distributed pattern of abnormality in temporal and extratemporal lobe grey matter (Cormack et al., 2005), similar to that reported in adults with temporal lobe epilepsy (Woermann et al., 1999; Keller et al., 2002a, b; Bonilha et al., 2004, 2005; McMillan et al., 2004). Examination of the relationship between volumetric abnormalities and cognition are rare in the paediatric epilepsy literature and are limited to IQ (Lawson et al., 2002). To date, there has been no investigation of the morphometric status of children with new-onset epilepsy, the relationship between cognitive and academic achievement and volumetric abnormalities, or examination of patterns of brain development in children with epilepsy versus controls.

This paper represents the first examination of both neuropsychological status and quantitative MR volumetrics in children with new-onset epilepsy of idiopathic aetiology compared with healthy control children. The specific issues of interest include (i) the cognitive and quantitative MRI status of children with new-onset localization-related (LRE) and primary generalized epilepsy (PGE) compared with controls; (ii) the relationship between volumetric status and cognition in healthy controls and children with new-onset epilepsy; and (iii) identification of children with academic problems at seizure onset and/or antedating seizure onset and their clinical epilepsy, neurocognitive and neuroimaging features.

Material and methods


Research participants included patients with new-onset epilepsy (n = 53) and healthy first-degree cousin controls (n = 50), aged 8–18 years. Study participants were recruited from paediatric neurology clinics at two large Midwestern medical centres (University of Wisconsin-Madison, Marshfield Clinic). Criteria for the patients with epilepsy included (i) diagnosis of epilepsy within the past 12 months; (ii) chronological age between 8–18 years; (iii) no other developmental disabilities (e.g. autism, developmental delay); (iv) no other neurological disorder; and (v) normal clinical MRI. Epilepsy participants met criteria for classification of idiopathic epilepsy in that they had normal neurological examinations, no identifiable lesions on MR imaging and no other signs or symptoms indicative of neurological abnormality (Engel, 2001).

Control participants were age- and gender-matched first-degree cousins. Criteria for controls included no histories of (i) any initial precipitating event (e.g. simple or complex febrile seizures); (ii) any seizure or seizure-like episode; (iii) diagnosed neurological disease; (iv) loss of consciousness >5 min; or (v) other family history of a first-degree relative with epilepsy or febrile convulsions. All children were attending regular schools.

First-degree cousins were used as controls rather than siblings for the following reasons: (i) first-degree cousins are more genetically distant from the participants with epilepsy and thus less predisposed than siblings to shared genetic factors that may contribute to anomalies in brain structure and cognition; (ii) a greater number of first-degree cousins are available than siblings in the target age range; and (iii) the family link was anticipated to facilitate participant recruitment and retention over time.

The study was approved by the Institutional Review Boards at both institutions, and the recruitment procedures were identical. Clinic registry records were first used to identify new patients with epilepsy seen in the Departments of Neurology at the respective institutions. These cases underwent preliminary review by the study coordinator to ensure that they were a new-onset case and appeared to meet criteria for study inclusion. These cases were then staffed internally within each institution with a paediatric neurologist who verified patient eligibility. Then, monthly teleconferences were held for case review by paediatric neurologists and study personnel from both institutions where inclusion for the study was confirmed with preliminary diagnosis of epilepsy syndrome and seizure type. Eligible families and participants were then sent a letter introducing the study, and families were provided with a telephone number to immediately opt out of study participation if so desired. If families did not opt out of the study, the study coordinator contacted the family to answer questions, schedule participation and facilitate recruitment of available first-degree cousins.

The Institutional Review Board (IRB)-approved recruitment strategy for controls was to ask study participants and/or parents to identify potential first-degree cousin controls of the children with epilepsy and initially inquire into the family's interest in study participation. The parents of the participants with epilepsy provided the research coordinator with contact information for interested control families, and a similar recruitment process to that described above ensued.


On the day of study participation, families and children gave informed consent and assent, following which the children underwent comprehensive neuropsychological testing and MRI. Parents participated in a clinical interview and completed a set of questionnaires to characterize details regarding gestation, delivery, neurodevelopmental health history and seizure history of their child. All medical records pertinent to the child's epilepsy and treatment were obtained after signed release of information was garnered from the parent.

Neuropsychological assessment

Both the subjects with epilepsy and controls were administered a comprehensive test battery that included standard clinical measures of intelligence, language, immediate and delayed verbal and visual memory, executive functions, speeded fine motor dexterity and academic achievement. Table 1 provides details regarding the target cognitive domains, the specific abilities assessed within each domain, the test measures and the nature of the dependent measure (i.e. number correct, errors or time). The age range studied here was broad (age: 8–18 years), and particular attention was paid to the use of tests that would allow administration of identical items across the entire age range as opposed to administering different tests of particular cognitive abilities to children in different age ranges. For example, we assessed intelligence using the four-subtest Wechsler Abbreviated Scale of Intelligence (WASI), which involved administration of the same item content across the target age range as opposed to administration of the Wechsler Intelligence Scale for Children—III (WISC-III) to children ≤16 years of age and the WAIS-III to those > 16 years. This was the approach used to construct the entire battery.

View this table:
Table 1

Neuropsychological test battery

IntelligenceVerbalWechsler Abbreviated Scale of Intelligence (verbal IQ)a
Non-verbalWechsler Abbreviated Scale of Intelligence (performance IQ)a
LanguageConfrontation namingBoston Naming Testb
Expressive namingExpressive Vocabulary Testa
Receptive languagePeabody Picture Vocabulary Test-IIIa
Generative namingDelis–Kaplan Executive Function System (verbal fluency)a
MemoryVerbal memoryChildren's Memory Scale (word list—immediate)a
Children's Memory Scale (word list—delayed)a
Non-verbal memoryChildren's Memory Scale (dot location—immediate)a
Children's Memory Scale (dot location—delayed)a
Executive functionProblem solvingDelis–Kaplan Executive Function System (card sort)a
Response inhibitionDelis–Kaplan Executive Function System (Colour-Word Interference Test)a
Divided attentionDelis–Kaplan Executive Function System (switching fluency)a
InattentivenessConnors' Continuous Performance Test-II (omission errors)a
Motor functionSpeeded fine motor dexterityGrooved Pegboardc
Psychomotor speedWechsler Intelligence Scale for Children-III (Digit Symbol Test)a
  • aScaled scores; braw scores; cseconds.

Given the wide age and grade range of these subjects, raw test scores were converted to age-, gender- and grade-adjusted z-scores based on the control group. Adjusted norms for all the measures used in this study are not commercially available. Relationships between age, gender and education for each test index were determined for the controls using regression techniques with age and education as continuous variables and gender as a dichotomous variable. The resulting solution was then applied to the epilepsy subjects. This was the process used across all the test measures in the battery. This conversion has several advantages including the ability to place all test scores on a common metric so that relative performance differences across diverse cognitive abilities can be readily appreciated, and all raw scores are adjusted for factors known to affect psychometric performance. Mean cognitive domain z-scores were computed (intelligence, language, memory/learning, executive function, psychomotor speed) and served as the primary dependent measures. This data reduction process served to reduce the number of comparisons and reduce experiment-wise Type 1 error.

The Kolmogorov–Smirnov Test indicated that all mean raw cognitive domain scores and mean z-cognitive domain scores (the primary unit of analysis) for the epilepsy and control groups were distributed normally. Also examined were the raw score distributions for each of the 16 test measures within the epilepsy and control groups before conversion to z-scores. The Kolmogorov–Smirnov Test indicated that raw scores were distributed normally for 28 of the 32 test measures. Non-normally distributed raw scores occurred in the executive (CPT-omissions) and memory (visual memory) domains. To ensure that systematic bias was not introduced, these measures were transformed and the analyses described in the Results section were recomputed. Deviation from the primary analyses occurred in only 1 of the involved 18 analyses (a post hoc pair-wise comparison).

For safety reasons, the research assistants were aware of each participant's group membership (epilepsy versus controls). All tests were administered in a standardized fashion and all assistants met monthly with the investigating neuropsychology team to review scoring and protocol issues, address decision rules for scoring responses not addressed in test manuals and attend to other procedural concerns.

Subsequent to test administration, each child's IQ and academic achievement results were independently reviewed by a paediatric neuropsychologist blinded to the participants' group membership in order to obtain an independent determination of the presence of a learning disability. In addition, parents were queried in detail regarding (i) the presence and number of seizures or seizure-like episodes before the formal diagnosis of epilepsy, and medical records were also specifically reviewed for this information; and (ii) the presence and type of any special education services provided to children with epilepsy before diagnosis and first recognized seizures. This information was confirmed in further detail in follow-up structured phone interviews with the families.

MRI procedures

MR acquisition

Images were obtained on a 1.5 T GE Signa MR scanner. Sequences acquired for each participant included (i) T1-weighted, three-dimensional spoiled gradient-recalled (SPGR) acquired with the following parameters: echo time (TE) = 5, repetition time (TR) = 24, flip angle = 40, number of excitations (NEX) = 2, field of view (FOV) = 26, slice thickness = 1.5 mm, slice plane = coronal, matrix = 256 × 192; (ii) proton density (PD); and (iii) T2-weighted images acquired with the following parameters: TE = 36 ms (for PD) or 96 ms (for T2), TR = 3000 ms, NEX = 1, FOV = 26, slice thickness = 3.0 mm, slice plane = coronal, matrix = 256 × 192 and an echo train length = 8.

MRIs were processed using a semi-automated software package, that is, Brain Research: Analysis of Images, Networks and Systems (BRAINS) (Andreasen et al., 1996; Harris et al., 1999; Magnotta et al., 1999; Magnotta et al., 2002). MR processing staff was blinded to the clinical, sociodemographic and neuropsychological characteristics of the participants. The T1-weighted images were spatially normalized so that the anterior–posterior axis of the brain was realigned to the anterior commissure-posterior commissure (AC–PC) line, and the inter-hemispheric fissure was aligned on the other two axes. A piecewise linear transformation was defined, providing the ability to warp the standard Talairach atlas space (Talairach, 1988) onto the re-sampled image. Images from the three pulse sequences were then co-registered using a local adaptation of automated image registration software. Following alignment of the image sets, the PD and T2 images were re-sampled into 1 mm cubic voxels, following which an automated algorithm classified each voxel into grey matter, white matter, CSF, blood or unclassified (Harris et al., 1999). The brains were then ‘removed’ from the skull using a neural network application that had been trained on a set of manual traces (Magnotta et al., 2002). Manual inspection and correction of the output of the neural network tracing was conducted. The brain images were then volume-rendered using local utilities, producing tissue volumes for regions of interest within the brain.

Data were radio frequency (RF)-corrected for magnetic field inhomogeneity. This was addressed by creating a segmented image, the basis of which consisted of a large number of tissue samples randomly selected from the three co-registered image sets. The tissue plugs are used to generate a series of discriminate functions for classification; the spatial variation in signal intensity is modelled in the discriminate functions with a second-order polynomial. The surface contour data are then generated from the segmented image.

The PD and T2-weighted images were used to acquire multi-spectral information regarding the brain that allowed surface CSF to be quantified. This is difficult to do using only a T1-weighted image. It also allows separation of venous blood from grey matter. The T1-weighted image was spatially aligned along the inter-hemispheric fissure and the AC–PC line. No scaling was applied to the images. The T1-weighted images were re-sampled to 1.0 mm3. A 6 degree of freedom (d.f.) rigid registration is used to co-register the PD and T2-weighted images to the AC–PC-aligned T1. Because all the measurements were obtained in the image space of the subject and not normalized, intracranial volume (ICV) was used as a co-variate in the analysis. The variables of interest were total cerebral tissue volume and segmented volumes of cerebral grey and white matter and CSF. In addition, total lobar tissue volumes were derived (frontal, temporal, parietal and occipital). The reported series represents consecutively processed scans excluding those with artefacts (e.g. dental implants, braces), incomplete scan protocols and other acquisition errors, and scans of poor quality primarily due to movement or anxiety that prevented quantitative analysis.


Table 2 provides information regarding the sociodemographic characteristics of the epilepsy and control groups. As can be seen, there were no significant group differences in overall age, gender or years of education. There were significantly more left-handed children in the epilepsy group (χ2 = 7.1, d.f. = 1, P = 0.013).

View this table:
Table 2

Demographic and clinical characteristics

Epilepsy (n = 53) mean (SD)Controls (n = 50) mean (SD)
Age (years)12.7 (3.3)12.7 (3.2)
Gender (M/F)31/2223/27
Handedness (R/L)46/750/0
Years of education6.5 (3.3)6.4 (2.9)
Age of onset (years)11.5 (3.5)
Duration of epilepsy (months)10.0 (4.1)
Localization-relateda30 (57%)
Generalizedb23 (43%)
  • aIncluding LREs such as temporal, frontal, occipital, centrotemporal and NOS.

  • bIncluding PGEs such as childhood absence, juvenile absence and juvenile myoclonic epilepsy.

The participants with epilepsy had an average age of onset of 11.5 years and an average duration of epilepsy of 10.0 months. Regarding treatment, 75% (n = 40) were receiving monotherapy, 2% (n = 1) were receiving polytherapy and 23% (n = 12) were not taking anti-epilepsy medications.

Neuropsychological performance

Table 3 provides mean raw test scores for the epilepsy and control groups. The epilepsy subjects did not outperform the controls on any of the test measures and exhibited significantly poorer performance across measures of intelligence (verbal, performance and full scale), aspects of language (naming), attention (increased errors of omission but not commission), aspects of executive function (response inhibition but not card sorting) and speeded psychomotor abilities (digit symbol substitution, grooved pegboard). There were no significant differences between the groups across unadjusted measures of immediate or delayed verbal or visual memory or other aspects of expressive or receptive language.

View this table:
Table 3

Raw scores for the neuropsychological tests

DomainTestControls (n = 50) mean (SD)Patients (n = 53) mean (SD)
IntelligenceWechsler Abbreviated Scale of Intelligence (full scale)*142.6 (35.8)129.5 (40.0)
Wechsler Abbreviated Scale of Intelligence (performance)*65.8 (20.2)56.6 (25.1)
Wechsler Abbreviated Scale of Intelligence (verbal)*80.1 (19.2)72.9 (16.8)
LanguageBoston Naming Test*12.5 (1.5)11.7 (2.0)
Expressive Vocabulary Test121.0 (27.5)110.6 (26.8)
Peabody Picture Vocabulary Test-III157.7 (21.4)153.4 (25.4)
Delis–Kaplan Executive Function System (verbal fluency)29.0 (10.4)28.5 (10.9)
Verbal memoryChildren's Memory Scale (word list—immediate)35.9 (6.3)34.3 (7.9)
Children's Memory Scale (word list—delayed)8.5 (2.4)7.7 (3.0)
Visual memoryChildren's Memory Scale (dot location—immediate)20.6 (2.8)20.2 (3.2)
Children's Memory Scale (dot location–delayed)7.0 (1.3)6.9 (1.4)
Executive functionDelis–Kaplan Executive Function System (switching fluency)9.9 (3.4)9.3 (3.5)
Delis–Kaplan Executive Function System (colour-word interference)**64.8 (20.6)79.2 (31.7)
Delis–Kaplan Executive Function System (card sort—target sorts)10.3 (2.5)9.4 (3.4)
Connors' Continuous Performance Test (omission errors)*5.7 (8.3)10.2 (14.8)
Motor functionGrooved Pegboard (dominant hand)**68.5 (10.8)81.9 (23.1)
Grooved Pegboard (non-dominant hand)*77.5 (15.5)89.0 (29.8)
Wechsler Intelligence Scale for Children-III (digit symbol coding)**58.7 (16.8)47.9 (17.3)
  • *P < 0.05; **P < 0.01.

Figure 1 provides mean adjusted (age, gender, education) cognitive domain z-scores. Data were analysed using ANOVA (analysis of variance), and univariate effects were significant for intelligence (F = 9.5, d.f. = 1.100, P = 0.003), language (F = 4.69, d.f. = 1.94, P = 0.033), executive function (F = 15.3, d.f. = 1.89, P < 0.001) and speeded psychomotor abilities (F = 23.07, d.f. = 1.100, P < 0.001). In all cases, the epilepsy patients performed significantly worse than the controls. Memory performance showed a trend towards poorer performance in the epilepsy group (F = 3.01, d.f. = 1.100, P = 0.086).

Fig. 1

Performance of epilepsy and control groups across cognitive domains. The epilepsy group performs significantly worse than controls in intelligence, language, executive function and psychomotor speed, with a trend for memory.

To determine whether these general trends were present in both epilepsy syndrome groups, subjects with PGE and LRE were compared with controls using one-way ANOVA. There was a significant group effect across all cognitive domains including intelligence (F = 9.34, d.f. = 2.95, P = 0.001), language (F = 3.2, d.f. = 2.87, P = 0.043), memory (F = 2.6, d.f. = 2.96, P = 0.021), executive function (F = 10.4, d.f. = 2.93, P = 0.001) and psychomotor speed (F = 24.6, d.f. = 2.93, P < 0.001). Figure 2 shows mean adjusted z-scores for the groups across the cognitive domains. Post hoc pair-wise comparisons revealed that both the LRE and PGE groups performed significantly worse than controls across the domains of intelligence (P's < 0.006), executive function (P < 0.005) and psychomotor speed (P's < 0.001). The LRE, but not PGE, group differed from controls on the language and memory domains. There were no significant differences between the two epilepsy syndrome groups in any cognitive domain.

Fig. 2

Both the LRE and PGE groups perform significantly worse than controls in intelligence, executive function and psychomotor speed. LRE but not PGE differ from controls in language and memory. There were no significant differences between LRE and PGE groups in any cognitive area.

Use of adjusted z-scores for the neuropsychological tests allows calculation of an impairment index for each subject, that is, the proportion of test scores outside normal limits per subject. Defining abnormality as z < −2.0, Fig. 3 shows that, on average, only 1.5% of overall test scores were abnormal for the controls versus 9.25% of test scores for all epilepsy subjects (P < 0.001). When epilepsy syndrome was examined, both localization-related (8.7%) and primary generalized (10.0%) groups had greater impairment indices compared with controls (both P's < 0.001), but again there was no difference between the localization-related and primary generalized syndrome groups (P = 0.49).

Fig. 3

Neuropsychological impairment indices in controls, all children with epilepsy combined and children with epilepsy syndrome. Significantly more impairment was exhibited in all epilepsy groupings compared with controls, with no significant difference between LRE and PGE groups.

Academic achievement problems: cognitive status

In the epilepsy group, 26% presented with a history of academic problems (AP+) versus 4% of controls (χ2 = 15.9, d.f. = 1, P < 0.001). In the epilepsy group, children with academic problems (AP+) were not different from the group without academic problems (AP−) in epilepsy syndrome (P = 0.63), age (P = 0.38), gender (P = 0.45), handedness (X2 = 3.3, d.f. = 1, P = 0.07), grade (P = 0.25), age of onset of epilepsy (P = 0.24), duration (P = 0.12) or number of medications (P = 0.19). As would be expected, the AP+ epilepsy group performed significantly worse than healthy controls and AP− epilepsy children on measures of reading (P's < 0.001), spelling (P's < 0.002) and arithmetic (P's < 0.001) using the Wide Range Achievement Test-III (Wilkinson, 1993).

When compared across the cognitive domains (Fig. 4), the epilepsy AP+ group performed significantly worse than controls across all cognitive domains (all P's < 0.02) and significantly worse than epilepsy AP− children in the intelligence, language, executive function and psychomotor speed domains (all P's < 0.004) with a trend for memory (P = 0.052). The epilepsy AP− children continue to score significantly below controls in executive function (P = 0.032) and psychomotor speed (P < 0.001). Regarding the overall impairment index, both epilepsy groups continued to exhibit significantly (P < 0.001) higher impairment indices compared with controls (AP+: 12%, AP−: 8.2%), but the epilepsy groups did not differ significantly (P = 0.08). Similar trends were evident in academic achievement scores with the AP+ group performing significantly worse than both the controls and AP− group in word reading, spelling and calculation (all P's < 0.003), with the AP− and controls only differing in computation (P < 0.05) but not reading or spelling.

Fig. 4

Children with epilepsy with (AP+) and without (AP−) history of academic achievement problems. AP+ group performs significantly worse across all domains compared with controls and AP− group, with the exception of memory. AP− group differs from controls in executive and speed domains.

Quantitative volumetrics

Quantitative volumetrics were completed for a consecutive sample of 31 controls and 43 children with epilepsy. The groups were compared in total cerebral grey and white matter as well as total tissue volumes for frontal, temporal, parietal and occipital lobes via multivariate analysis of covariance (MANCOVA) with age and ICV as covariates. There was no significant overall group effect (F = 0.22, d.f. = 6.65, P = 0.97), nor significant univariate effects for any of the regions examined. MANCOVA was also computed between the controls and localization-related and primary generalized groups with age and ICV as covariates with the identical regions of interest. Again, there was no overall effect of group (F = 0.50, d.f. = 12.126, P = 0.91) and there were no significant univariate effects across the regions of interest (all P's > 0.25). Figure 5 depicts the mean volumetric measurements for the groups.

Fig. 5

Age- and ICV-adjusted total cerebral volumes and total lobar tissue volumes for controls and epilepsy syndrome groups.

Segmented cerebral tissue volumes and cognition

The broader relationship between cognition and quantitative volumetrics was examined by Pearson correlations. The primary finding (Table 4) was a significant relationship in the controls between improving cognitive scores and increased cerebral volumes, especially total cerebral white matter volume. These findings were observed across measures of intelligence (performance IQ and full-scale IQ), language (expressive and receptive vocabulary and lexical fluency), executive function (category switching) and motor speed. In contrast, these significant relationships were totally absent in the epilepsy group. Thus, while mean volumetric measurements are comparable between the epilepsy and control groups, there appeared to be a differential relationship between cognition and brain structure (white matter) in the control and epilepsy groups.

View this table:
Table 4

Correlation of cognition with volumes of total cerebral grey and white matter

Controls (n = 32)Epilepsy (n = 43)
Wechsler Abbreviated Scale of Intelligence (FSIQ)−0.070.56**−0.240.22
Wechsler Abbreviated Scale of Intelligence (PIQ)−0.160.38*−0.190.17
Expressive Vocabulary Test−0.060.47**−0.170.22
Peabody Picture Vocabulary Test−0.170.59**−0.240.28
Delis–Kaplan Executive Function System (letter fluency)−0.090.38*−0.210.05
Children's Memory Scale (dot location—long delay)0.210.30−0.39*−0.09
Delis–Kaplan Executive Function System (category switching)−0.250.49**−0.300.07
Delis–Kaplan Executive Function System (inhibition)0.38*−0.060.31*0.03
Grooved Pegboard (total)0.00−0.39*0.25−0.15
  • *P < 0.05; **P < 0.01.

Academic achievement problems: quantitative MRI correlates

Examination of lobar grey and white matter volumes using MANCOVA with age and ICV as covariates revealed significant univariate effects for left parietal grey (P = 0.018) and left occipital grey (P = 0.047) matter. Post hoc pair-wise comparisons revealed that the epilepsy AP+ group had significantly lower grey matter volumes compared with controls in parietal (P = 0.027) and occipital regions (P = 0.033) as well as significantly lower volumes compared with the epilepsy AP− group in parietal (P = 0.005) and occipital (P = 0.016) regions, with no differences between the controls and epilepsy AP− group (P's > 0.39).


The core findings of this study include the following: (i) mild diffuse neuropsychological problems are evident in children with new-onset epilepsy, regardless of epilepsy syndrome; (ii) academic difficulties are present at the time of diagnosis and appear to have existed before the first recognized seizure in a subset of children, suggesting an antecedent neurobiological abnormality; (iii) quantitative MR volumetrics do not differ between children with new-onset epilepsy, regardless of epilepsy syndrome; (iv) there is a differential relationship between cognitive performance and white matter volume in the epilepsy versus control groups; and (v) children with a history of academic problems at the onset of epilepsy demonstrate the most impaired cognition as well as volumetric reductions in left occipital and parietal grey matter compared with controls and children with epilepsy without academic problems. These points will be reviewed below.

Neuropsychological status at epilepsy onset

Impairments in neuropsychological status have been well described in children, adolescents and adults with chronic epilepsy (Lassonde et al., 2000; Elger et al., 2004). The findings presented here indicate that the neuropsychological substrate and academic status of children with new- or recent-onset epilepsy is adversely affected early in the course of the disorder, regardless of syndrome type. Children with new-onset epilepsy as a group exhibit a pattern of mild but diffuse cognitive impairment. Significantly poorer neuropsychological status was observed across the cognitive domains of intelligence, language, executive function and psychomotor speed, with a trend for memory function (Fig. 1), an effect that is quite comparable across LREs and PGEs (Fig. 2). Another reflection of the degree of cognitive dysfunction evident at the onset of epilepsy is the proportion of individual test scores outside normal limits (impairment index), defined as a z-score ≤ −2.0. Children with new-onset epilepsy had a significantly higher impairment index (9.3%) compared with controls (1.5%), again with no difference between the localization-related (8.7%) and primary generalized (10.0%) groups (Fig. 3).

All children were attending regular school, and academic achievement problems of a severity resulting in provision of supportive/remedial services were present in 24% of the children with epilepsy versus 4% of the controls. Of the children with epilepsy with a history of academic difficulties, services were provided before the diagnosis of epilepsy and the first recognized seizure in the majority, findings suggestive of an antecedent neurobiological abnormality. These findings support previous reports of academic difficulties in children with new or recent-onset epilepsy (Oostrom et al., 2003; Berg et al., 2005; McNelis et al., 2005), but extend previous findings by characterizing their neuropsychological and MR volumetric characteristics. Children with prior academic problems (AP+) exhibited particularly affected cognition with significantly poorer performance across all cognitive domains compared with controls and they were also more impaired than epilepsy children without learning problems (AP−) across all cognitive domains except memory (Fig. 4). It is important to point out that the cognition of epilepsy children without academic performance problems (AP−) differed from controls only in executive and motor/psychomotor speed domains. Thus, cognitive status of AP+ children is quite distinct and problematic, with the cognition of the AP− children still affected, but to a much milder degree and only in limited cognitive domains. The volumetric findings in AP+ children will be discussed below.

Quantitative MR volumetrics

Quantitative volumetric studies of children with epilepsy are few in number and to date involve children with chronic epilepsy. These studies have reported abnormalities in totalcerebral, cerebellar and hippocampal volumes (Lawson et al., 1997, 1998, 2000a, b, 2002), and recent voxel-based morphometry (Cormack et al., 2005) has revealed distributed abnormalities in grey matter in children with chronic temporal lobe epilepsy, not unlike that reported in adult patients (Keller et al., 2002a, b; McMillan et al., 2004). To date, we are not aware of an examination of MRI volumes at or near the onset of paediatric epilepsy. The current findings demonstrate that, as a group, children with idiopathic epilepsy do not differ from healthy controls in total cerebral volumes (segmented grey and white matter) at the onset of epilepsy nor do they differ in total lobar volumes (grey and white matter). Inspection of children with LRE versus PGE indicates that there are no differences across groups (Fig. 5). It cannot be ruled out that abnormalities may be evident in specific neuroanatomical structures (thalamus, hippocampus) or discrete areas of lobar regions as might be detected by other volumetric techniques such as voxel-based morphometry.

Despite the comparable MRI volumes in the epilepsy and control subjects, there was an interesting divergence between the groups in brain-behaviour relationships. In the control group, a robust association existed between increasing cerebral white matter volume and better cognitive performance in the areas intelligence, language, psychomotor speed and some aspects of executive function, relationships that were absent in the epilepsy subjects (Table 4). These findings could result from several factors. While there is no difference in white matter volume between epilepsy patients and controls, differences in the functional integrity of the white matter in children with epilepsy might be apparent using other neuroimaging techniques (e.g. diffusion tensor imaging), a hypothesis we are currently pursuing. It is also possible that white matter integrity may be completely normal in the epilepsy children but other factors related to epilepsy may affect cognition and disrupt the volume–cognition relationships. For example, EEG abnormalities such as generalized/focal slow waves or the presence and rate of interictal epileptiform discharges (Dodrill and Wilkus, 1978; Binnie, 2003; Koop et al., 2005) may affect cognition and disengage the white matter volume–cognition relationship.

Finally, the subset of children with epilepsy with neurodevelopmental academic problems (AP+) exhibited significant volumetric reductions in left parietal and occipital lobe grey matter, abnormalities that were evident when compared with both controls as well as epilepsy children without a history of academic problems (AP−). No other volumetric abnormalities were observed in other lobar grey or white matter regions. Thus, AP+ children appear to be a high-risk group given their academic difficulties, neurocognitive impairments and volumetric abnormalities.

Clinical and theoretical implications

There is increasing concern regarding the cumulative neurobiological burden associated with chronic epilepsy and the risk of progressive cognitive impairment (Pitkanen and Sutula, 2002). In addition, interest is growing in lifespan models of the neuropsychology of epilepsy (Hermann et al., 2002; Helmstaedter et al., 2003). Characterization of neuropsychological status at or near the onset of epilepsy provides insight into the cognitive substrate upon which the effects of medically intractable seizure, medications and other potentially adverse seizure-related factors may accumulate. It is becoming increasingly clear that this early substrate is not unaffected, the abnormalities being more diffuse than one might initially anticipate, with comparable effects across broad syndromes of epilepsy. An important clinical implication of these findings is that efforts are needed to identify and remediate these early neurobehavioural problems, especially in light of the known poor long-term psychosocial prognosis of persons with even remitted childhood epilepsy (Sillanpaa et al., 1998). It would seem that the children with academic problems at epilepsy onset would be an especially important group to target in that prospective research (Camfield et al., 1993) has demonstrated that a history of learning problems is a major marker for poor long-term psychosocial outcome.

In addition to the neuropsychological findings at epilepsy onset, reports have appeared indicating that psychiatric/behavioural and academic problems may antedate the diagnosis of epilepsy in both children (Austin et al., 2001; Dunn et al., 2002; Oostrom et al., 2003; Hesdorffer et al., 2004; Jones, 2004; Berg et al., 2005) and adults (Hesdorffer et al., 2000, 2006) with epilepsy. The mechanisms underlying these effects remain to be determined and remain an important issue for future research. The fact that these antecedent insults can be observed even in children with idiopathic epilepsy, without abnormalities in overall brain volumetrics, raises the possibility that factors associated with underlying epileptogenesis leading to the onset of seizures may play a role (Cortez et al., 2006).

Given these neurocognitive findings at or before onset of childhood epilepsy, the consequences of recurring seizures and treatment on normal cognitive development and brain growth remain to be characterized. Adverse neurodevelopmental trajectories may be most pronounced in younger children who are undergoing the greatest cognitive and brain growth compared with older children and adolescents, a hypothesis that remains to be tested.

More broadly, the implications of even static cognitive impairments in childhood may have lifespan implications. It has been demonstrated in general population research that childhood intelligence level (at age 11) is associated with the risk of adverse cognitive outcomes (e.g. dementia) decades later, with higher childhood intelligence associated with better cognitive outcomes/protective effects and vice versa for lower childhood intelligence (Whalley et al., 2000; Deary et al., 2004). Similarly, long-term prospective investigations of normal ageing have demonstrated that cognitive abnormalities in midlife may antedate or serve as harbingers of adverse cognitive outcomes decades later (Linn et al., 1995; Kawas et al., 2003). The status of mentation in older adults with unremitted epilepsy remains to be fully characterized, but the findings to date are not favourable (Martin et al., 2005). The degree to which early (childhood) or later (middle age) fixed cognitive abnormalities set the stage for greater than age-associated cognitive changes remains to be determined.


This project was supported by NIH NINDS RO1 44351, F32 MH64988-01A2 and MO1 RR 03186 (GCRC). We thank Michelle Szomi for overall project coordination; Kevin Dabbs, Katherine Bayless and Karen Wagner for MR processing; and Erica Johnson and Karyn Wagner for cognitive testing. We especially thank Drs Fred Edelman, Carl Stafstrom, David Hsu, and Jason Doescher for patient referrals. Dedicated to PDH.


View Abstract