Neurobehavioral comorbidities of pediatric epilepsies are linked to thalamic structural abnormalities




Neurobehavioral comorbidities are common in pediatric epilepsy with enduring adverse effects on functioning, but their neuroanatomic underpinning is unclear. Striatal and thalamic abnormalities have been associated with childhood-onset epilepsies, suggesting that epilepsy-related changes in the subcortical circuit might be associated with the comorbidities of children with epilepsy. We aimed to compare subcortical volumes and their relationship with age in children with complex partial seizures (CPS), childhood absence epilepsy (CAE), and healthy controls (HC). We examined the shared versus unique structural–functional relationships of these volumes with behavior problems, intelligence, language, peer interaction, and epilepsy variables in these two epilepsy syndromes.


We investigated volumetric differences of caudate, putamen, pallidum, and thalamus in children with CPS (N = 21), CAE (N = 20), and HC (N = 27). Study subjects underwent structural magnetic resonance imaging (MRI), intelligence, and language testing. Parent-completed Child Behavior Checklists provided behavior problem and peer interaction scores. We examined the association of age, intelligence quotient (IQ), language, behavioral problems, and epilepsy variables with subcortical volumes that were significantly different between the children with epilepsy and HC.

Key Findings

Both children with CPS and CAE exhibited significantly smaller left thalamic volume compared to HC. In terms of developmental trajectory, greater thalamic volume was significantly correlated with increasing age in children with CPS and CAE but not in HC. With regard to the comorbidities, reduced left thalamic volumes were related to more social problems in children with CPS and CAE. Smaller left thalamic volumes in children with CPS were also associated with poor attention, lower IQ and language scores, and impaired peer interaction.


Our study is the first to directly compare and detect shared thalamic structural abnormalities in children with CPS and CAE. These findings highlight the vulnerability of the thalamus and provide important new insights on its possible role in the neurobehavioral comorbidities of childhood-onset epilepsy.

Behavior problems, as well as impaired cognition, language, and social skills are common comorbidities of pediatric epilepsy, with enduring adverse effects on functioning (see review in Lin et al., 2012a). Population-based epidemiologic investigations have uniformly documented higher rates of attention-deficit hyperactive disorder (ADHD), anxiety disorders, and depression, as well as lower intelligence quotient (IQ) and poor social skills in children with epilepsy compared to children with other medical problems (Davies et al., 2003), healthy controls (Hoie et al., 2006), and siblings without epilepsy (Berg et al., 2007). These behavioral, cognitive, linguistic, and social comorbidities can collectively exert a negative effect on daily activities, such as relationships within the family, peer interactions, school performance, and extracurricular activities (Baca et al., 2011).

There is now substantial evidence indicating that childhood-onset epilepsy is associated with altered brain development (Hutchinson et al., 2010; Pulsipher et al., 2011; Tosun et al., 2011; Lin et al., 2012) and that these structural abnormalities are linked to neurobehavioral comorbidities (Daley et al., 2007; Caplan et al., 2008; Pulsipher et al., 2009; O'Muircheartaigh et al., 2011; Vollmar et al., 2011; Lin et al., 2012). In healthy growing children, cortical gray matter volume and thickness declines with increasing age (Giedd & Rapoport, 2010). In addition to age, the dynamic process of change in cortical thickness is associated with the maturation of intellectual abilities (Shaw et al., 2006). Against this background of expected developmental trajectory, Tosun et al., in a cross-sectional study, showed that pruning of the frontal and temporal regions is altered in children with complex partial seizures (CPS). Specifically, children with CPS with average IQ demonstrated greater cortical thickness than controls, whereas those with below average IQ scores show more thinning with age compared to typically developing children (Tosun et al., 2011). Behavior problems and linguistic deficits are also associated with abnormal cortical development in pediatric epilepsy. Smaller inferior frontal gyrus white matter volumes are related to the presence of a psychiatric diagnosis, mainly ADHD in children with chronic CPS (Daley et al., 2007). Furthermore, frontal lobe volumes are increased in children with new-onset epilepsy and comorbid ADHD relative to both children with epilepsy but without comorbid ADHD and to controls (Hermann et al., 2007). Structural abnormalities involving language-related cortex are associated with impaired language skill in children with epilepsy with average intelligence (Caplan et al., 2009).

Similar to the cortex, subcortical regions undergo age-related developmental changes with increasing volumes in childhood followed by decreasing volumes in adolescence (Sowell et al., 2002; Lenroot et al., 2007; Brain Development Cooperative Group, 2012). Although the link between age and subcortical volumes has not been extensively examined in childhood epilepsy, several lines of evidence suggest that children with epilepsy might have aberrant subcortical developmental trajectory compared to typically developing peers. First, volumetric reductions in the striatum and thalamus have been found in children with localization-related epilepsy (Cormack et al., 2005) and idiopathic generalized epilepsy (Chan et al., 2006; Pulsipher et al., 2009, 2011; Luo et al., 2011) compared to their healthy peers. Second, thalamic and striatal volumetric abnormalities are evident in children with new-onset epilepsy, thus implying a neurodevelopmental impact of childhood epilepsy on subcortical maturation prior to the onset of seizures (Pulsipher et al., 2011; Lin et al., 2012). Third, subcortical development is closely linked to cortical development in healthy children (Lenroot et al., 2007). Given our previous findings of abnormal age-related cortical development in childhood epilepsy (Tosun et al., 2011), we hypothesized that subcortical development might also deviate from normal.

Subcortical structural volumes are related to cognitive performance in pediatric epilepsy. In children with new-onset juvenile myoclonic epilepsy, baseline frontothalamic volume is associated with deficits in executive function (Pulsipher et al., 2009), and striatal enlargement predicts better executive function performances in children with new-onset benign epilepsy with centrotemporal spikes (Lin et al., 2012). These subcortical structures that are vulnerable to childhood-onset epilepsy are also implicated in pediatric psychiatric disorders. Specifically, structural and functional abnormalities in the thalamus and basal ganglia are found in children with ADHD (Xia et al., 2012), depression (Matsuo et al., 2008), internalizing symptoms (Herba et al., 2010), and anxiety disorder (Guyer et al., 2012).

Therefore, in the current study, we focused on subcortical regions in the corticostriatal circuit—including the caudate, putamen, pallidum, and thalamus—that have been implicated in both childhood-onset epilepsy and neuropsychiatric disorders. Among these subcortical regions, we targeted structures that would differentiate children with localization-related epilepsy (LRE) and idiopathic generalized epilepsy (IGE) from healthy controls (HC). Cryptogenic epilepsy with CPS and childhood absence epilepsy (CAE) were chosen as epilepsy syndromes that represent the two respective broad band epilepsy diagnoses, namely LRE and IGE. We predicted that children with CPS and CAE would have significantly smaller volume of the striatum and thalamus compared to the HC group. We also hypothesized that we would identify a different age relationship with striatal and thalamic volumes in the epilepsy and HC groups. We then examined the shared versus unique structural–functional relationships with behavior problems, intelligence, language, peer interaction, and epilepsy variables in these two epilepsy syndromes. In terms of behavioral problems, we focused on attention problems (e.g., daydreaming, difficulties with concentration), social problems (e.g., poor interpersonal skills and/or school performances, few hobbies), and thought problems (e.g., impaired or illogical reasoning, disorganized communication skills) because a recent meta-analysis (N = 2,434) showed that these problems were more common in childhood epilepsy compared to other pediatric chronic diseases (Rodenburg et al., 2005). In addition, several studies have shown that children with epilepsy have impaired social competence and peer interaction (see review in Hamiwka et al., 2011). Our previous study also linked linguistic skills with frontal lobe white matter volumes (Caplan et al., 2009). Given the integral connections among thalamus, basal ganglia, and prefrontal cortex (Draganski et al., 2008), and the evidence for the role of the thalamus in language (see review in Llano, 2013), we hypothesized that subcortical volumes will be related to linguistic skills. Therefore, among the children with epilepsy, we posited that smaller striatal and thalamic volumes will be related to more attention, social, and thought problems; lower IQ; impaired language skill; poor peer interaction; and severity of epilepsy. Together with previously described cortical findings, these predicted findings might provide a neuroimaging signature of the neurobehavioral comorbidities of pediatric epilepsy.



This study was approved by the institutional review boards of the University of California, Los Angeles (UCLA) and the University of Southern California (USC). Study participants included 20 children with CPS, 21 children with CAE, and 27 HC, aged 6–16 years. The primary study inclusion criterion for each epilepsy subject was that he/she had a diagnosis of epilepsy (CPS or CAE) and at least one seizure in the year prior to participation in the study.

Participants were recruited from tertiary centers (UCLA and USC) (N = 16, 40%) as well as community clinics (Los Angeles and Anaheim Kaiser Permanente) and the San Diego Chapter of Epilepsy Foundation (N = 24, 60%). A board-certified pediatric neurologist made the diagnosis of epilepsy (CPS or CAE), according to criteria of the International League Against Epilepsy, and all children with epilepsy experienced at least one seizure in the past year (Engel & International League Against Epilepsy, 2001). The diagnosis of CPS was made based on clinical criteria of having focal seizures with alteration of consciousness of unknown cause and normal findings on clinical magnetic resonance imaging (MRI). The diagnosis of CAE was based on clinical evidence of absence seizures induced by hyperventilation and electroencephalography (EEG) studies showing the typical 3-Hz generalized spike and wave discharges. We excluded children with other types of epilepsy (i.e., mixed seizure disorder and juvenile myoclonic epilepsy), prior epilepsy surgery, neurologic illness other than epilepsy, metabolic disorders, hearing impairments, intellectual disability, and bilingual speakers of American English who did not attend English-speaking schools or speak English at home.

Socioeconomic status was based on the parental occupational and educational status, using the Hollingshead two-factor index (Hollingshead, 1973). Hollingshead levels I and II were classified as high and levels III–V were classified as low. Table 1 presents the demographic characteristics of the study participants. The three study groups did not differ significantly on age, gender, ethnicity, and socioeconomic status.

Table 1. Demographic, comorbidity, and epilepsy variables of study groups
VariablesHC (n = 27)CPS (n = 20)CAE (n = 21)
  1. a

    CPS significantly lower than both CAE (t65 = 2.62, p = 0.01) and HC (t65 = 3.75, p = 0.0004).

  2. b

    Speech language quotient for CPS significantly lower than for HC (t65 = 2.21, p = 0.03).

  3. c

    Both CPS (t65 = 3.76, p = 0.0004) and CAE (t65 = 2.44, p = 0.02) significantly higher than HC.

  4. d

    Both CPS (t65 = 3.67, p = 0.0005) and CAE (t65 = 4.35, p = 0.0001) significantly higher than HC.

  5. e

    Both CPS (t65 = 3.49, p = 0.001) and CAE (t65 = 4.90, p = 0.0001) significantly higher than HC.

  6. f

    CPS significantly lower than both CAE (t65 = 1.96, p = 0.05) and HC (t65 = 3.81, p = 0.0003).

Chronologic age (year) (SD)10.9 (2.6)10.9 (2.7)9.6 (2.1)
Sex (%)   
Ethnicity (%)   
Socioeconomic status (%)   
Age of seizure onset (year) (SD) 6.9 (3.7)7.0 (2.1)
Duration of illness (year) (SD) 4.2 (2.4)2.6 (2.5)
Seizure frequency per year (log) 2.9 (2.7)5.2 (3.8)
Prolonged seizures (%) 259.5
Febrile seizures (%) 104.7
Antiepileptic drugs (%)   
None 09.5
Monotherapy 7061.9
Polytherapy 3028.6
Full scale IQa109.8 (12.9)94.5 (12.9)105.6 (15.9)
SLQb103.8 (16.3)93.2 (14.7)98.9 (17.2)
CBCL scores   
Social problemsc51.0 (2.9)58.6 (8.3)55.7 (7.9)
Thought problemsd51.7 (3.9)58.8 (6.7)59.8 (7.9)
Attention problemse51.9 (2.6)59.9 (7.8)61.7 (10.6)
Peer Interactionf48.5 (5.4)39.6 (9.8)44.4 (7.3)

The parents' reports and children's medical records provided information on seizure frequency, antiepileptic drugs (AEDs), age of onset, illness duration, as well as the number of febrile convulsions and prolonged seizures (i.e., >5 min) (Table 1). Of the 20 children with CPS, 7 had no focal epileptic activity on EEG conducted at the time of the initial epilepsy diagnosis; 3 had left, 3 had right, 7 had bilateral epileptic activity; and 12 had epileptic activity in the frontotemporal region. Seven children with CPS had background slowing, and one had secondary generalization. Of the 21 children with CAE, 2 had generalized tonic–clonic convulsions, and 1 had background slowing.

To include control subjects from ethnic and socioeconomic status backgrounds similar to the CAE group, we recruited the normal subjects from four public and two private Los Angeles schools. The study coordinator screened potential participants for neurologic, psychiatric, language, and hearing disorders through a telephone conversation with a parent. In children without epilepsy, we also excluded children with ADHD, depression, and anxiety disorder in the past or who met criteria for these disorders in a structured psychiatric interview once enrolled in the study.


Image acquisition

MRI data were obtained at UCLA with a 1.5-T GE Signa MR scanner (GE Medical Systems, Milwaukee, WI, U.S.A.). T1-weighted sagittal spoiled gradient (SPGR) images were collected using an imaging acquisition matrix of 256 × 256 × 124 and slice thickness of 1.2 mm. Imaging parameters were as follows: repetition time = 24 msec, echo time = 9 msec, flip = 22 degrees, field of view = 240 mm × 240 mm, number of excitations = 2.

Image processing

Image processing and analysis were performed using tools from the Oxford Center for Functional MRI of the Brain (FMRIB, FSL version 4.1.5; T1-weighted images were initially skull-stripped using the Brain Extraction Tool (BET) (Smith et al., 2002), and each image was visually inspected for proper brain extraction. Brain extracted images were then processed using FMRIB's Automated Segmentation Tool (FAST) (Zhang et al., 2001) to correct for spatial intensity variations and segment images into gray matter, white matter, and cerebrospinal fluid (CSF). Total intracranial volume was calculated by summing the volumes of gray matter, white matter, and CSF, as generated by FAST.

Subcortical gray matter structures, including caudate, putamen, pallidum, and thalamus, were automatically segmented using FMRIB's Integrated Registration and Segmentation Tool (FIRST) (Patenaude et al., 2011). FIRST is a segmentation tool that is based on a template originating from manually segmented images, with subcortical labels parameterized as surface meshes. This process first registered the T1 images to the template via 12 degrees of freedom affine transformation, and then 12 degrees of freedom registration to a standard MNI 152 template subcortical mask. This subcortical mask was then used to label and segment subcortical structures based on T1 image intensity. Image registration and subcortical segmentations were visually inspected to confirm accuracy of the results. Subcortical volumes obtained from automated segmentation have been shown to be highly comparable to manual labeling techniques (Patenaude et al., 2011).

Behavior, cognition, language, and peer interaction

Child Behavior Checklist (CBCL): Parents completed the 118 behavioral problem and 20 social competence items of the CBCL (Achenbach, 1991), which includes two major scales: a behavioral problem scale and a social competence scale. Although the CBCL generates broad band (i.e., externalizing and internalizing) and narrow band scales, in this study we targeted behavior problems that were more common in childhood epilepsy compared to other chronic pediatric illnesses: (1) attention problems, (2) social problems, and, (3) thought problems (Rodenburg et al., 2005). Individual item intraclass correlations (ICCs) are >0.90. Stability of ICCs over a 3-month period is 0.84 for behavior problems and 0.97 for social competencies. Test/retest reliability is 0.89. Tests of criterion-related validity for referred/nonreferred children support the validity of the instrument, and normative data are extensive (Achenbach, 1991). The cut point for borderline/clinically significant pathology in this study was based on the age and gender normed T score of 67 for the narrow band scales (Achenbach, 1991).

The three social subscales measure competencies in the following domains: activities (e.g., sports, hobbies); social (e.g., friendships, interpersonal skills); and school (e.g., performance, ability, school problems). In this article we focus on the peer friendship and interaction component, which has a cut point of 30. Scores below 30 are in the clinical range.

IQ: The Wechsler Intelligence Scale for Children-3rd edition (WISC-III) (Wechsler, 1991), administered to the study participants, generated Full Scale, Verbal, and Performance IQ scores.

The Test of Language Development (TOLD) (Newcomer & Hammil, 1988): The TOLD has three forms: the TOLD-2 Primary for children aged 4–8 years; the TOLD-2 Intermediate for children aged 8–12 years; and the TOAL for adolescents 12–18 years. Each form of the TOLD-2 consists of a series of subtests through which it assesses both vocabulary and grammar. We administered the TOLD Primary to 26.5%, the TOLD Intermediate to 57.3%, and the TOAL to 16.2% of the children (Newcomer & Hammil, 1988). Spoken language quotient (SLQ) derived from each of these tests was used as an independent variable.

Statistical analysis

We first compared volumes of caudate, putamen, pallidum, and thalamus among the three groups of children—CPS, CAE and HC of comparable age—using mixed models with repeated measures, with group (CPS, CAE, HC) as the intersubject and hemisphere (left, right) as the intrasubject variables. Four mixed models were estimated, one for each subcortical volume; age, gender, and total intracranial volumes were used as covariates. For those volumes that were significantly different between the groups, secondary analyses were conducted to determine which hemisphere was driving the findings. Then, to reduce the number of comparisons, only those specific subcortical volumes for which children with epilepsy were significantly different from controls were further analyzed. Within each group, we examined the association of these volumes with age, IQ, SLQ, CBCL behavior problem scores, and peer interaction scores, as well as epilepsy-related variables using Spearman correlation coefficients. The associations with the CBCL scores and epilepsy-related variables were studied only within the two epilepsy groups, due to the limited range of CBCL scores within the HC group. To accommodate nonlinear relationships between variables, we used rank-based Spearman correlations to study the associations. For the plots depicting the significant associations, we employed a loess procedure (as implemented in PROC LOESS of SAS v9.3, SAS Institute Inc., Cary, NC, U.S.A) to determine the best fit. The loess procedure fits weighted quadratic or linear least squares regression to localized subsets of the data. This is done at each point at which the regression line is to be estimated, thereby building a best fit point by point. The epilepsy-related variables that were studied included age of seizure onset, duration of illness (time from age of onset to participation in study); log seizure frequency in the year prior to recruitment to the study; and the number of AEDs subdivided into no AEDs, monotherapy, and polytherapy. All tests were two-tailed, and an alpha level of 0.05 was adopted for all inferences.


Subcortical volumes: group comparisons

A significant group difference was found for thalamus volumes, controlling for the effects of gender, age, and total intracranial volume (Fig. 1; F2,62 = 3.37, p = 0.04). The interaction between hemisphere and group was significant (F2,64 = 3.75, p = 0.03), and post hoc pair-wise analyses showed that left thalamic volumes were reduced in CPS (8036.9 ± 802.6; t64 = −2.84, p = 0.01) and CAE (8146.2 ± 899.7; t64 = −2.02, p = 0.05) compared to HC (8419.6 + 911.2). There were no statistically significant group differences in the other subcortical regions.

Thalamus volume asymmetry: group comparisons

Given that the thalamic volumes exhibited significant group differences, we then examined volume asymmetries (left – right) among the three study groups. A significant group difference for thalamic asymmetry was found, after controlling for age and gender (F2,62 = 3.93; p = 0.02). Post hoc pair-wise analyses revealed that the CAE group (89.6 ± 195.2 mm3) had significantly less leftward asymmetry compared to controls (294.8 ± 228.5 mm3); t62 = −2.78, p = 0.007), whereas the difference between the CPS group (176.5 ± 341.6 mm3) and controls approached statistical significance (t62 = −1.67, p = 0.10).

Figure 1.

Subcortical structural volumetric differences among children with complex partial seizures (CPS, red bar), childhood absence epilepsy (CAE, green bar), and healthy controls (blue bar). CPS and CAE group demonstrated significantly reduced thalamic volumes compared to controls. There were no significant volumetric differences in the caudate, putamen, and pallidum. Volumes are expressed in mm3, with error bars representing standard deviations. * denotes p ≤ 0.05, controls versus CPS and controls versus CAE.

Cognitive, linguistic, behavior, and social profiles

Table 1 presents the cognitive, linguistic, behavior, and social profiles of the three groups. Although there were no significant between-group differences in the demographic variables, the CPS group was significantly more impaired than the other groups in cognition and peer interaction, and both epilepsy groups exhibited significantly more behavior problems than the HC. In addition, the CPS group was more impaired than the HC on the language measure.

Association of volumes with comorbidities and epilepsy variables

As described in the 'Statistical analysis' section, since the left thalamus volumes were significantly different between groups, we investigated the association of only left thalamus volumes with age, IQ, SLQ, behavior, and social profiles. The left thalamus volume was positively associated with age in the CPS group (rs = 0.72, p = 0.0008) and CAE group (rs = 0.49, p = 0.03), while controlling for duration of illness (Fig. 2). Age was not related to thalamic volume in the HC group (rs = 0.22, p = 0.3). Post hoc analyses revealed that age was significantly correlated with duration of illness in the CAE group (CAE: rs = 0.64, p = 0.002) but not in the CPS group (rs = 0.08, p = 0.75). However, no significant relationship was found between thalamic volume and age of epilepsy onset, seizure frequency, and number of AEDs.

Figure 2.

Plot of left thalamus volumes versus age for controls, CPS, and CAE. The lines shown were the best fit lines determined by a loess (local regression) procedure.

In the CPS group, the left thalamic volume was significantly linked to Full Scale IQ (rs = 0.52, p = 0.03) and SLQ (rs = 0.58, p = 0.01) scores while controlling for age and gender, but these relationships did not reach significance in the CAE (Full Scale IQ: rs = 0.39, p = 0.10; SLQ: rs = 0.30, p = 0.20) or HC groups (Full Scale IQ: rs = 0.30, p = 0.14; SLQ: rs = 0.30, p = 0.16). In CPS and CAE, the left thalamic volume was significantly correlated with CBCL social problem scores (CPS, r = −0.52, p = 0.04; CAE, r = −0.46, p = 0.05). In CPS, left thalamus volume was also associated with CBCL attention problem (r = −0.54, p = 0.04) and peer interaction scores (r = 0.53, p = 0.04).


Four primary findings were derived from examining the subcortical structural integrity of children with CPS and CAE: (1) Among subcortical regions measured, both children with CPS and CAE exhibited significantly smaller left thalamic volume compared to HC; (2) in terms of developmental trajectory, greater thalamic volume was correlated with increasing age in children with CPS and CAE but not in HC; (3) with regard to the comorbidities, reduced left thalamic volumes were related to more social problems in children with CAE and CPS; and (4) smaller left thalamic volumes in children with CPS were also associated with poor attention, lower IQ and SLQ scores, and impaired peer interaction. These findings highlight the age-related vulnerability of the thalamus in children with epilepsy and its association with unique versus shared functional–structural relationships in CPS and CAE.

Reduced thalamic volume

Our study is the first to directly compare and detect shared thalamic structural abnormalities in children with CPS and CAE with similar age of seizure onset, duration of illness, mean chronological age, and demographic features, as well as MRI scanner and protocol. Several lines of evidence suggest that the thalamus is critical for seizure generation and propagation in both CAE and CPS. Animal models of CAE showed that the signature 3–4-Hz spike-and-wave discharges likely reflect aberrant thalamocortical oscillations (Hughes, 2009), and AEDs that specifically target T-type calcium channels in the thalamus effectively block this rhythm (Huguenard & Prince, 1992). Volumetric reductions in the thalamus have been documented in children with CAE (Chan et al., 2006; Luo et al., 2011). Functional neuroimaging and electrophysiology studies (Berman et al., 2010; Masterton et al., 2012) also implicate the thalamus in seizure generation and propagation. Thalamic volumes are also reduced in children (Cormack et al., 2005; Tosun et al., 2011) and adults (McDonald et al., 2008; Seidenberg et al., 2008) with CPS, and the degree of volume loss has been correlated with seizure duration (Keller et al., 2012). Furthermore, a number of studies showing microstructure (Widjaja et al., 2013) and metabolic abnormalities (Henry et al., 1993) also support thalamic involvement in CPS.

In the current study, we showed that thalamic volumes were selectively reduced in children with epilepsy, whereas basal ganglia volumes were preserved relative to healthy controls. Consistent with existing literature, thalamus abnormalities have been frequently found in pediatric epilepsy (Cormack et al., 2005; Chan et al., 2006; Pulsipher et al., 2009; Carney et al., 2010; Luo et al., 2011; Pulsipher et al., 2011), but the involvement of basal ganglia structures has been less common. In contrast to our findings, Luo et al. showed that basal ganglia as well as thalamic volumes were reduced in children with CAE, but age differences among the participants of these two studies may have contributed to the discrepant findings. The average age of the study participants in Luo et al. (2011) study (12.1 ± 5.1 years old) was older than our study cohort (9.6 ± 2.1 years old). Of interest, neuroimaging studies in adults have consistently demonstrated basal ganglia structural abnormalities in LRE (DeCarli et al., 1998; Dreifuss et al., 2001; Pulsipher et al., 2007; Bouilleret et al., 2008; Riley et al., 2011) and IGE (Seeck et al., 2005; Keller et al., 2011). The cross-sectional design of our study precludes examination of the origin, development, and trajectory of subcortical brain structural abnormalities. We, nevertheless, speculate that left thalamic abnormality may precede the onset of epilepsy or occur early in the course of the illness, whereas basal ganglia abnormalities might emerge with longer duration of epilepsy (i.e., in the older youth in Luo et al. and in adults with epilepsy).

It is unclear why the thalamic atrophy was found on the left rather than the right. The left hemisphere might be more vulnerable to developmental insults due to a more protracted maturation process. Supporting this explanation, gene transcription analyses of the embryonic human brain reveal less expression of genes that are critical for cortical development in the left hemisphere, and suggest that left hemisphere development lags behind that of the right hemisphere (Sun et al., 2005). In addition, lower cerebral blood volume and oxygenation in the left compared to the right hemisphere of newborns implies that a delayed maturation in the left hemisphere may enhance its plasticity to environmental stimuli as well as its vulnerability to insults (Lin et al., 2013). Finally, patients with temporal lobe epilepsy (TLE) arising from the left side exhibit greater anatomic abnormalities than patients with right-sided TLE (Kemmotsu et al., 2011).

Thalamic asymmetry

The significantly decreased leftward asymmetry in the CAE and trend in the CPS group might reflect the selective left thalamic volume reduction in our study. This finding further emphasizes that the left thalamus might be more susceptible than the right thalamus to the neurodevelopmental impact of childhood-onset epilepsy. Although previous studies have demonstrated mixed findings in terms of thalamic volumetric asymmetries in healthy children and adults, other than rightward asymmetry in Xie et al. (2012) the general trend corroborates our leftward asymmetry findings (Ahsan et al., 2007; McDonald et al., 2008; Brain Development Cooperative Group, 2012). A recent diffusion tensor imaging study also revealed that thalamic-prefrontal connectivity was greater on the left side compared to the right in healthy children between the ages of 8 and 17 years old (Alkonyi et al., 2011), the age range of the children in our study. Although the developmental origin of leftward thalamic asymmetry is unknown, several authors have speculated that hemispheric lateralization of language and cognition might underlie this finding (Toga & Thompson, 2003). In children with epilepsy, we showed that the normal leftward asymmetry is disrupted, with the CAE group showing significantly reduced asymmetry and the CPS group a trend for reduction.

Thalamic volume and age

Although cross-sectional, our study showed different age-related trajectories in the CPS, CAE, and HC groups. In contrast to normal subcortical developmental patterns of an inverted “U” curve, with volumetric increases in childhood followed by decreases in adolescence (Sowell et al., 2002; Lenroot et al., 2007; Brain Development Cooperative Group, 2012), we did not find an association between left thalamic volume and age in our healthy controls. The narrower age range and modest sample size of our healthy cohort might have obscured important normal developmental trajectories. Specifically, our mean age of 10.9 years (standard deviation = 2.6) in the HC group likely fell on the plateau of the inverted “U” curve, evident in larger normative studies. Yet, despite the similar age range in the two epilepsy and control groups, the statistically significant positive correlation between left thalamic volume and age in the epilepsy groups suggest a different developmental trajectory compared to typical developing children.

Relationships between thalamic volume and comorbidities

This is the first study to demonstrate involvement of the thalamus in behavior problems, lower IQ, impaired language, and poor peer interaction in children with epilepsy. Two of the three problem behaviors thought to occur more frequently in childhood epilepsy than other chronic pediatric illnesses (Rodenburg et al., 2005), social problems and attention problems, were associated with reduced left thalamic volumes. The role of the orbital frontal cortex in regulation of emotions and social decision making (Bachevalier et al., 2011), together with the connection between the thalamus and this brain region (Price et al., 2006) probably underlies our findings. In addition, the shared association between social problem scores and left thalamic volume in the CPS and CAE groups might also reflect the significantly smaller orbital frontal gyrus volumes in both children with CPS (Daley et al., 2007) and CAE (Caplan et al., 2009) compared to age- and gender-matched healthy controls. However, it is important to emphasize that the thalamus is composed of several nuclei with distinct thalamocortical connections (Draganski et al., 2008) serving multiple different functions, including sensory integration, language, memory, attention, and executive function (Jakab et al., 2012; Watanabe & Funahashi, 2012; Llano, 2013). Furthermore, previous studies suggest different patterns of thalamic involvement in IGE (Tyvaert et al., 2009; Wang et al., 2012) versus LRE (Langlois et al., 2010). Even though the overall thalamic volumes were linked to social problems and attention problems in both epilepsy groups, it is possible that specific nuclei may exert unique influences on these neurobehavioral comorbidities. Therefore, additional studies are warranted to determine the relationship between volume of specific thalamic nuclei and the comorbidities in children with epilepsy.

The relationship of IQ, SLQ, and peer interaction scores with left thalamic volume only in the CPS group probably reflects the significant differences between the CPS and CAE groups in these scores (Table 1). Despite similar social and attention problems scores in both epilepsy groups (Table 1), smaller left thalamic volumes were associated with higher social problem scores in CPS and CAE but with higher attention problem scores only in CPS. These findings imply different underlying subcortical mechanisms for the comorbidities in children with CPS and CAE. Similar findings are evident for the association between the comorbidities and cortical volumes (Daley et al., 2007; Caplan et al., 2009, 2010). These subcortical and cortical findings emphasize the need for prospective studies to determine how CPS and CAE differentially impair cortical and subcortical development while manifesting similar behavioral, cognitive, linguistic, and social deficits.

Limitations and Conclusions

Study limitations include the relatively small sample size, cross-sectional study design, and multiple statistical comparisons. Therefore, the findings of our study will require confirmation in a larger sample of children with epilepsy. However, using a conservative statistical approach for the analyses, we confirmed the study's hypotheses regarding smaller volume of the thalamus in both CPS and CAE, abnormal relationship of thalamic volumes with age in CPS, and the role of the thalamus in a wide range of CPS comorbidities but limited to social problems in CAE.


Supported by grants from the National Institutes of Health: K23 NS060993 (JJL) and NS32070 and MH 6718 (RC).


The authors have no conflicts of interests. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.


  • Image of creator

    Dr. Jack J. Lin uses advanced neuroimaging to examine neurobehavioral comorbidities of epilepsy.