Objective evaluation of language function is critical for children with intractable epilepsy under consideration for epilepsy surgery. The purpose of this preliminary study was to evaluate word recognition in children with intractable epilepsy by using magnetoencephalography (MEG). Ten children with intractable epilepsy (M/F 6/4, mean ± SD 13.4 ± 2.2 years) were matched on age and sex to healthy controls. Common nouns were presented simultaneously from visual and auditory sensory inputs in “match” and “mismatch” conditions. Neuromagnetic responses M1, M2, M3, M4, and M5 with latencies of ~100 ms, ~150 ms, ~250 ms, ~350 ms, and ~450 ms, respectively, elicited during the “match” condition were identified. Compared to healthy children, epilepsy patients had both significantly delayed latency of the M1 and reduced amplitudes of M3 and M5 responses. These results provide neurophysiologic evidence of altered word recognition in children with intractable epilepsy.
Progression of epilepsy may negatively affect language and other higher order cognitive functions, especially in young children. Impairments of receptive language function (language comprehension) in epilepsy patients are demonstrated by problems in phonological processing [
Approximately 20–30% of children with epilepsy continue to have disabling seizures despite high-dose medications (intractable epilepsy) or develop intolerable medication side effects. For some of those children, surgical intervention is undertaken to ameliorate and often cure the epilepsy [
Magnetoencephalography (MEG) is a diagnostic tool for evaluating patients with medically refractory epilepsy and localizing epilepsy focus [
MEG studies focusing on visual word recognition have been used in patients with intractable epilepsy to examine language processing and language lateralization [
The objective of this study was to use this previously developed paradigm of simultaneous auditory and visual word presentation [
The study was approved by the Cincinnati Children’s Hospital Medical Center (CCHMC) institutional review board.
Ten children with medically intractable epilepsy that were undergoing evaluation for epilepsy surgery were matched with a single healthy control by age (within one year) and sex. There were 4 females and 6 males in each group. The primary language for the subjects in both groups was English.
The age range of the epilepsy group was 10–17 years with a mean of
Characteristics of patients.
Nr. | Demographics | Diagnosis | Side | Epilepsy onset, years | Epilepsy duration years | Pathology | Average spike frequencyb | Antiepileptic drugsc | Neuropsychological exam | Seizure outcome | F/U duration (mo) | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Age, y | Sex | DH | Epilepsy focusa | VCI | PRI | WMI | PSI | FSIQ | WJ L-W | WJ Pcomp | Engel’s class | ||||||||
1 | 13 | F | R | Temporal | L | 6 | 7 | FCD 2A | 17 | Lamotrigine, topiramate, clonazepam | 79 | 94 | 94 | 103 | 88 | 79 | 76 | Class 1 | 24 |
|
|||||||||||||||||||
2 | 14 | F | R | Frontal, temporal | L | 6 | 8 | FCD 1B | 33 | Levetiracetam, ethosuximide, methosuximide | 67 | 82 | 52 | 68 | 60 | 65 | 75 | Class 1 | 28 |
|
|||||||||||||||||||
3 | 15 | M | R | Frontal | L | 13 | 2 | FCD 1A | 38 | Gabapentin, divalproex sodium | 93 | 102 | 80 | 70 | 84 | 113 | 104 | Class 1 | 26 |
|
|||||||||||||||||||
4 | 12 | M | R | Occipital, temporal | L | 7 | 7 | old ische-mic change | 46 | Levetiracetam, oxcarbazepine, zonisamide | 87 | 67 | 65 | 62 | 65 | 101 | 91 | Class 1 | 36 |
|
|||||||||||||||||||
5 | 12 | M | R | Frontal | R | 8 | 4 | FCD 2A | 0 | Topiramate, lamotrigine | 71 | 94 | 50 | 78 | 62 | 87 | 67 | Class 1 | 19 |
|
|||||||||||||||||||
6 | 11 | F | R | Temporal |
L | 6 | 5 | FCD 2B | 100 | Levetiracetam, lamotrigine, clonazepam | NT | 98 | 88 | 88 | 85 | 95 | 99 | Class 1 | 15 |
|
|||||||||||||||||||
7 | 14 | M | R | Frontal, parietal, temporal | R | 13 | 1 | NA | 14 | Oxcarbazepine, gabapentin | NT | NT | NT | NT | NT | 89 | 78 | No surgery yet | NA |
|
|||||||||||||||||||
8 | 10 | M | R | Parietal | L | 8 | 2 | FCD 1B | 0 | Levetiracetam, divalproex sodium | 115 | 102 | 102 | 91 | 105 | 107 | 97 | Class 1 | 47 |
|
|||||||||||||||||||
9 | 17 | M | R | Temporal |
L | 5 | 8 | FCD 1B | 5 | Levetiracetam | 73 | 73 | 71 | 75 | 68 | 93 | 92 | Class 1 | 19 |
|
|||||||||||||||||||
10 | 16 | F | R | Frontal |
L | 7 | 9 | NA | 0 | Lamotrigine, levetiracetam | NT | NT | NT | NT | NT | 105 | 96 | No surgery yet | NA |
|
|||||||||||||||||||
M |
13.4 ± 2.2 | 7.9 ± 2.8 | 5.3 ± 2.9 | 25.3 ± 31.2 | 87.4 |
92.7 ± 12.4 | 76.8 ± 19.2 | 78.5 ± 14.2 | 80 |
93.4 ± 14.3 | 87.5 |
26.8 ± 10.4 | |||||||
|
|||||||||||||||||||
Min | 10 | 5 | 1 | 0 | 67 | 67 | 50 | 62 | 60 | 65 | 67 | 15 | |||||||
|
|||||||||||||||||||
Max | 17 | 13 | 9 | 100 | 115 | 103 | 102 | 103 | 105 | 113 | 104 | 47 |
F: female; M: male; R: right; L: left; NT: not tested; NA: not applicable; DH: Dominant Hand; FCD: focal cortical dysplasia; VCI: Verbal Comprehension Index; PRI: Perceptual Reasoning Index; PSI: Processing Speed Index; WMI: Working Memory Index; FSIQ: Full Scale IQ; WJ L-W: Letter-Word Identification; and WJ Pcomp: Passage Comprehension.
aA major epilepsy focus is marked with a star
bNumber of spikes in 40 min of MEG.
cGiven the night before study (at least 12 h).
The age range of the healthy controls was 9–17 years with a mean of
All patients underwent presurgical evaluation in order to determine the epileptogenic area. The evaluation included seizure characterization by clinical semiology, video-EEG (VEEG) including overnight recording while the patients slept, magnetic resonance imaging (MRI) with epilepsy surgery protocol, ictal/interictal single photon emission computed tomography (SPECT), 2-deoxy-2[18F]fluoro-D-glucose positron emission tomography (FDG-PET), simultaneous MEG and EEG, and functional MRI (fMRI). None of the patients had electrographic status epilepticus of slow wave sleep (ESES). None of the patients had been previously diagnosed as Landau Kleffner nor did any of the patients fit the clinical profile for Landau Kleffner. The results of simultaneous MEG/EEG recording provided information about average spike frequency for each patient (Table
A total of 10 patients finished the full process of presurgical evaluation. Of those 10 patients, 8 patients underwent resective surgery. Seven patients (Engel Class I, 88%, 7/8) were seizure-free and 1 patient (Engel Class II, 12%, 1/8) had rare seizures after mean follow-up duration of 26 months (range 15–47 months). The pathology showed a focal cortical dysplasia in 7 patients and an ischemic change in 1 patient. Two patients did not have surgery because of incomplete evaluation or discordant test results. None of the patients experienced appreciable neurological deficits as a result of surgery.
The routine neuropsychological examination of epilepsy surgery candidates included the Wechsler Intelligence Scale for Children, Fourth Edition (WISC-IV) [
The stimuli consisted of 120 common nouns that were one to three syllables (mean 1.35 syllables) and three to eight letters (mean 4.88 letters) based on counts of Kucera et al. [
Visual representation of the audio-visual word presentation paradigm. Stimuli were presented simultaneously from visual (screen) and auditory (earphones) sensory inputs in two different condition (1) “match” condition, for which the visually and acoustically presented words were identical, and (2) “mismatch” condition, for which the visually and acoustically presented words were different. The participants were asked to compare visually and acoustically presented words and to press the response button only if they did not match.
A standard protocol for data acquisition as described in our previous studies [
MEG data were corrected with DC offsets based on the pretrigger time-window. An off-line low pass filter (30 Hz) and high pass filter (3 Hz) were applied to the averaged MEG data. The analysis window was 600 ms before the stimuli and 2,000 ms after the stimuli. This study focused on the responses from the “match” condition, as the “mismatch” condition was used only to ensure subjects’ attention to the stimuli. The latencies and the peak amplitudes of averaged MEG waveform were measured for each recognizable component with the DataEditor software (VSM MedTech Ltd., Port Coquitlam, BC, Canada). There were five major and consistent peaks, labeled M1–M5 at latencies of ~100 ms (50–120 ms), ~150 ms (150–200 ms), ~250 ms (250–300 ms), ~350 ms (300–400 ms), and ~450 ms (400–500 ms), respectively. The first of these peaks (M1 and M2) are usually described as representing processing of stimuli characteristics (e.g., encoding of physical stimuli features) separately for each sensory modality [
Study results were analyzed with SAS software version 9.1 (SAS Institute, Cary, NC). The amplitude and latency comparisons for ERF components M1, M2, M3, M4, and M5 were conducted with a mixed model analysis of variance (ANOVA) with two fixed factors:
In order to estimate the relationship between neurophysiological changes and disease characteristics, the amplitudes and latencies of the ERF components were correlated (1) with the age of epilepsy patients at disease onset (“age at epilepsy onset”) and (2) with the duration of disease (“epilepsy duration”). In order to detect possible relations between IQ and neurophysiological parameters in epilepsy patients, the IQ scores (VCI, PRI, WMI, PSI, and FSIQ), reading and language comprehension scores (WL L-W, WJ Pcomp), and the ERF parameters were analyzed (latencies and amplitudes separately at left and right hemispheres). In order to investigate the effect of spike frequency on ERFs, average spike frequency during the recording of word “match” condition was correlated with ERFs parameters (latencies, amplitudes). All correlations were calculated with Spearman’s rho.
The latencies of M1 and M2 components were delayed in both hemispheres in epilepsy patients compared to healthy controls (Table
Latencies (ms, mean ± SD) of M1–M5 components separately from left and right hemispheres in epilepsy patients and healthy controls.
M1 | M2 | M3 | M4 | M5 | ||||||
---|---|---|---|---|---|---|---|---|---|---|
Left | Right | Left | Right | Left | Right | Left | Right | Left | Right | |
Epilepsy | 114.1 |
109.74 |
181.7 |
185.6 |
264.0 |
263.6 |
371.5 |
371.5 |
474.2 |
475.4 |
|
||||||||||
Controls | 90.3 |
92.73 |
154.8 |
161.7 |
248.86 |
255.6 |
341.3 |
349.4 |
450.0 |
447.3 |
Amplitudes (fT, mean ± SD) of M1–M5 components separately from left and right hemispheres in epilepsy patients and healthy controls.
M1 | M2 | M3 | M4 | M5 | ||||||
---|---|---|---|---|---|---|---|---|---|---|
Left | Right | Left | Right | Left | Right | Left | Right | Left | Right | |
Epilepsy | 350.1 |
317.0 |
422.4 |
339.1 |
362.8 |
377.7 |
293.5 |
330.4 |
230.4 |
266.7 |
|
||||||||||
Controls | 447.7 |
448.3 |
487.5 |
477.1 |
544.6 |
493.3 |
421.5 |
353.6 |
343.4 |
266.7 |
MEG waveform and topographical map of neuromagnetic activation elicited by visual and auditory words. Five major neuromagnetic responses are clearly identified. They are named as M1, M2, M3, M4, and M5. In topographical maps, red color represents the incoming magnetic fields; blue color represents outgoing magnetic fields. Epilepsy patient (top) had delayed latencies of M1 and M2 components as compared with the healthy control subject (bottom). The amplitudes of M3 and M5 magnetic fields were significantly smaller in epilepsy patients than in healthy controls.
There was no significant correlation between spike frequency and ERFs parameters.
Patients’ IQ scores ranged from average to mildly impaired with overall group performance score in the borderline range (see Table
There was a trend toward significant positive correlation (
Graphic representation of correlation analysis results: (a) relationship between left hemisphere M5 component amplitude and age at epilepsy onset; (b) relationship between right hemisphere M3 component amplitude and Working Memory Index (WMI); (c) relationship between right hemisphere M5 component amplitude and WMI; (d) relationship between right hemisphere M3 component amplitude and Processing Speed Index (PSI). “
Neuromagnetic responses M1–M4 identified in this study are comparable with those we found in the healthy adult population [
Neuromagnetic responses to word stimuli can be separated into two groups: early (M1 and M2 components) and late (components M3–M5) [
There were no significant differences between the right and left hemisphere latencies or amplitudes. This implies that the utilized word recognition task produced bilateral language activation and cannot be recommended at its present form as a task for determining language dominance. Future studies focusing on specific components elicited during task presentation combined with source analysis of electromagnetic brain activity are advised to investigate possible task application for identifying hemispheric dominance for language.
In our study, the latency of M1 was significantly delayed in epilepsy patients as compared with the group of healthy control participants. M1 belongs to a group of early MEG responses, which are thought to reflect activation of the primary auditory and visual sensory cortices [
We observed reduced amplitudes of M3 and M5 components in patients with intractable epilepsy when compared with healthy controls. Both M3 and M5 components belong to a group of late MEG responses, peaking between 250 and 450 ms after stimulus onset, respectively. These responses follow primary cortex activations (reflected in responses M1 and M2) and represent phonological and semantic processing, engaging inferior frontal gyrus (Brocas’s area), superior/middle temporal gyrus, and angular/supramarginal gyrus (Wernicke’s area) [
The association between visual and auditory word recognition was confirmed in a number of MEG studies by Salmelin [
These results could have implications for epilepsy patients with regard to intervention options aimed at improving word recognition and language comprehension in general. Visual stimulation has been reported to produce changes in auditory brain response, depending on whether the visual stimulus is congruent (match) or noncongruent (mismatch). If the auditory and visual stimuli are congruent, the response to auditory stimuli (or participation of neuronal resources in it) increases [
There was a strong trend for a positive correlation between M5 amplitude and the age at epilepsy onset; this means that earlier age at epilepsy onset was associated with smaller M5 amplitudes. This suggests that early epilepsy onset has a negative impact on word processing (integration of both visual and auditory modalities) and is associated with insufficient neuronal resources participating in this stage of word recognition. Our observed result is in line with our previous study [
One can argue that combining the majority of left-hemispheric cases (80% of studied patients) with right-hemispheric epilepsy cases (20%) can bias the results towards stronger observed effect on language-related function. This would to certain degree apply to adult epilepsy patients. In children, however, the reorganization of language-related function occurs (specifically in those with the early-onset epilepsy) [
We did not find any significant effect of spike frequency on ERFs amplitudes or latencies. This is consistent with previous studies [
The experience at epilepsy centers worldwide shows that intractable epilepsy patients are a challenging group to study. Some of the drugs used could have an inhibiting effect on ERF responses by decreasing their amplitudes and delaying their latencies through activation of the GABA system [
Because the main purpose of this study was to demonstrate the application of a novel paradigm for evaluation of word recognition in children with intractable epilepsy, we did not focus on matching subjects to controls by IQ. However, this is an important issue that needs to be addressed in future studies. For example, to better answer the question whether epilepsy itself or the altered neural substrate leads to impairment in information processing/cognitive functioning, future studies should match the borderline IQ of epilepsy patients to control subjects without epilepsy.
In conclusion, the results of our preliminary study were as follows: when compared to healthy subjects, patients with intractable epilepsy had (1) delayed conduction times of encoding of both visual and auditory stimuli features (reflected in M1 latency delay) and (2) reduced neuronal resources required for integration of audio and visual streams, required for performing this word recognition task (reflected in M3 and M5 amplitude reduction). The effect of interventions can be studied with this paradigm by recording MEG before and after intervention. To our knowledge, this is the first study that described simultaneous written and spoken word stimuli presentation in pediatric epilepsy patients. A larger scale study, based on described methodology and focusing on cortical generators of registered magnetic activity, would further our understanding of neural origins of language comprehension in children with intractable epilepsy.
The authors have no conflict of interests relevant to this paper to disclose.
Elijah Kirtman provided technical assistance in MEG data acquisition. Tere Richards provided help with paper editing. The study was partially supported by a Trustee Grant to Dr. Jing Xiang from Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio, USA.