Clinical Neurophysiology 121 (2010) 1494–1501

Contents lists available at ScienceDirect

Clinical Neurophysiology

journal homepage: www.elsevier .com/locate /c l inph

Cognitive effects of topiramate revealed by standardised low-resolution brainelectromagnetic tomography (sLORETA) of event-related potentials

Ki-Young Jung a, Jae-Wook Cho b, Eun Yeon Joo c, Sun Hwa Kim c, Kyung Mook Choi c, Juhee Chin c,Kun-Woo Park a, Seung Bong Hong c,*

a Department of Neurology, Korea University Medical Center, Korea University College of Medicine, Seoul, South Koreab Department of Neurology, Yangsan Pusan National University School of Medicine, Busan, South Koreac Department of Neurology, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, South Korea

a r t i c l e i n f o a b s t r a c t

Article history:Accepted 9 March 2010

Keywords:TopiramateCognitive functionEvent-related potential (ERP)Standardised low-resolutionelectromagnetic tomography (sLORETA)Source localisation

1388-2457/$36.00 � 2010 International Federation odoi:10.1016/j.clinph.2010.03.013

* Corresponding author. Address: Department of NCenter, Sungkyunkwan University School of Medicingu, Seoul 135-710, South Korea. Tel.: +82 2 3410 359

E-mail addresses: sbhong@skku.edu, sbhongsmc@

Objective: To evaluate the effect of topiramate (TPM) on event-related potentials (ERPs) in patients withepilepsy.Methods: Neuropsychological tests and ERP study using auditory oddball paradigm were conductedbefore and after treatment with TPM in drug-naive epilepsy patients. To detect target brain regions inwhich ERP changed during the cognitive task, cortical current densities of ERP components were analysedusing standardised low-resolution electromagnetic tomography (sLORETA).Results: Neuropsychological tests (n = 18 patients) showed that TPM significantly decreased the score indigit span, Corsi block and Controlled Oral Word Association word fluency. Repeated-measures analysisof variance of ERP data (n = 13 patients) revealed that P2 amplitude was significantly increased at Fzelectrode following treatment with TPM. Statistical non-parametric map of sLORETA between pre- andpost-TPM ERPs revealed that current density of P200 component was significantly reduced by TPM inbilateral parieto-occipital, temporolimbic and dorsolateral right prefrontal regions.Conclusions: Our findings suggest that TPM affects selective brain regions which may be related to cog-nitive side effects.Significance: Source localisation of ERPs can be helpful in identifying target brain regions for the cogni-tive side effects of anti-epileptic drugs.� 2010 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights

reserved.

1. Introduction

Anti-epileptic drugs (AEDs) may be a cause of cognitive dys-function in some patients with epilepsy (Meador, 2006; Park andKwon, 2008; Trimble and Thompson, 1983). As AEDs are typicallytaken on a long-term basis, the impact of AED-related cognitiveimpairment on daily life is an important issue in the treatmentof epilepsy. Although topiramate (TPM) is one of the most effectiveAEDs, it has been reported to have a negative impact on cognition,and patients frequently describe themselves as having slowthoughts, difficulty in calculating and blunted mental reactions.Previous studies reported that the greatest changes were notedin impaired concentration and poor performance on verbal tests,as revealed by various neuropsychological (NP) assessments(Aldenkamp, 2000; Aldenkamp et al., 2000; Blum et al., 2006;

f Clinical Neurophysiology. Publish

eurology, Samsung Medicale, 50 Irwon-dong, Gangnam-2; fax: +82 2 3410 0052.gmail.com (S.B. Hong).

Lee et al., 2006; Martin et al., 1999; Meador et al., 2003, 2005;Thompson et al., 2000). Furthermore, withdrawal of TPM causessignificant improvement in frontal lobe function-associated mea-sures such as verbal fluency and working memory (Kockelmannet al., 2003). However, NP tests can provide only approximate esti-mations of the brain regions involved in cognitive dysfunction. Fur-thermore, temporal information about information processing bythe brain cannot be determined by such tests.

The event-related potential (ERP) recorded during an oddballtask reflects the brain activities underlying various cognitive func-tions such as attention and working memory, in which severalongoing cognition-related components such as N100, P200 andP300 are involved. P300, the most studied component, representsaspects of information processing, such as attention allocationand decision-making (Polich and Kok, 1995). P200 is another cog-nitive ERP component that is usually interpreted as a reflection ofevaluating the task relevance of stimulus items, which is achievedby suppressing irrelevant features or enhancing relevant features(Potts, 2004). N100 is an early sensory-perceptual relatedcomponent that is sensitive to attention. Thus, ERPs provide a

ed by Elsevier Ireland Ltd. All rights reserved.

K.-Y. Jung et al. / Clinical Neurophysiology 121 (2010) 1494–1501 1495

neurophysiological index of a subject’s cognitive function. ERPmeasurements during task performance may provide a more sen-sitive avenue for objectively assessing medication-related changesin cognitive function (Chung et al., 2002; Ozmenek et al., 2008).

ERPs have been used to assess pharmacological influences ofvarious drugs (including AEDs, that act on the central nervoussystem (CNS)) on attention-dependent information processing(Anderer et al., 2004; Anderer et al., 2002a,b, 2008; Barry et al.,2007; Chung et al., 2002; Higuchi et al., 2008; Ohlmeier et al.,2007; Ozmenek et al., 2008; Ruijter et al., 2000; Saletu et al., 2002;Sun et al., 2007). Subtle functional changes in the CNS induced byseveral old AEDs have been reported. P300 latencies and amplitudeswere significantly affected by old AEDs, such as phenobarbital, car-bamazepine and valproate (Chen et al., 1996; Enoki et al., 1996).Compared with placebo, an augmentation of the N160 component(corresponding N100 component of the visual ERP occurringbetween 130 and 180 ms) to match visual stimuli was reduced byphenytoin in healthy young adults (Chung et al., 2002). Valproatemay induce impairment of cognitive processing, as revealed byERP N270 and P300 (Panagopoulos et al., 1997; Sun et al., 2007).N270 is considered as a constant component of ERPs reflecting thecognitive activity in the human brain for processing conflict.

Contrary to old AEDs, only a few studies have reported effects ofnew AEDs on cognitive ERPs. Smith et al. (2006) reported that inhealthy subjects, P300 at Pz electrode was not significantly af-fected by TPM, while TPM blocked enhancement of positive-goingslow wave that followed the P300. Since previous studies had lim-ited ERP measurements from only a few scalp electrodes, the ef-fects of the AEDs could not be evaluated over the whole brain.Thus, target brain regions of AEDs could not be identified.

The method of source localisation may provide information foridentifying the generators of particular EEG activities (Michel et al.,2004). Low-resolution electromagnetic tomography (LORETA) pro-vides three-dimensional images of brain electrical activity, conse-quently giving information for brain regions that are involved inneurocognitive processes and are the targets of therapeutic drugaction (Pascual-Marqui et al., 1994). Thus, electrophysiologicalneuroimaging using LORETA may reveal those parts of the struc-tural network that make major contributions to the scalp-recordedERP component affected by AEDs. We hypothesised that informa-tion processing might be changed by treatment with TPM. Subse-quently ERP analysis by LORETA (ERP–LORETA) could revealbrain regions that may be related to ERP changes.

The purpose of the present study is to evaluate the effects ofTPM on cognitive ERPs in drug-naive patients with epilepsy, focus-sing particularly on temporal relationships and the target brain re-gion where TPM results in qualitative alterations in neuronalactivity related to cognitive tasks. Based on the literature, wehypothesised that sLORETA should reveal changes in cerebralsource activity in certain brain regions, particularly prefrontaland temporo-parietal areas, in relation to the cognitive effects ofTPM. To detect target brain regions in which ERPs change duringa cognitive task, depending on the temporal processing of cognitivefunction, cortical current densities of ERP components were ana-lysed using sLORETA (Pascual-Marqui, 2002).

2. Methods

2.1. Subjects and TPM schedule

Twenty-four consecutive drug-naive patients with partial epi-lepsy were enrolled from an outpatient epilepsy clinic. Exclusioncriteria included not being right-handed, a diagnosis of mentalretardation, use of other CNS-acting drugs, brain lesions other thanhippocampal sclerosis or having a progressive neurological or psy-chiatric disease.

Each patient received TPM monotherapy for 12–16 weeks. Be-fore the TPM treatment, all subjects underwent a physical exami-nation, routine blood tests, magnetic resonance imaging (MRI) ofthe brain, electroencephalogram (EEG), NP tests and an initialERP study. TPM was given at 50 mg/day for the first 2 weeks andat 100 mg/day for the next 2 weeks. From the fifth week, TPMwas gradually increased, up to a maximum of 400 mg/day, asneeded, to control seizures and was maintained at a fixed dosefor at least 2 weeks before the second ERP study was conducted.However, if a patient complained of intolerable adverse events be-cause of TPM administration, the dosage was decreased by 25 mg/day each week until the symptoms subsided. Blood testing for hae-matology and liver function was performed twice, before medica-tion administration and at the end of the study period. Thefrequency of seizures was checked by a self-recorded seizure diarywhen the patient visited the outpatient clinic.

All patients were informed of the procedure and informed con-sent was obtained from all subjects in accordance with the guide-lines of the institutional review board at Samsung Medical Center.

2.2. NP tests

A battery of NP tests, standardised for the Korean population,was administered. The first NP test (NP1) was given within 1 weekof the ERP1 study and the second NP test (NP2) was conducted atleast 6 months after NP1. The following NP tests were performed.

(1) The Korean version of the mini-mental state examination (K-MMSE) (Kang et al., 1997) and the short form of the KoreanWechsler Intelligence Scale (KWIS) to evaluate general cog-nitive function (Lim et al., 2000).

(2) Digit span forward and backward and Corsi block forwardand backward, which test attention and working memory.

(3) Digit symbol tests to assess psychomotor performance, sus-tained attention and visuo-motor coordination.

(4) The trail-making test (TMT, A and B types) to test attention,visuo-motor tracking abilities and mental flexibility.

(5) The Wisconsin Card Sorting Test (WCST) to test executivefunctions such as conceptual formation and cognitive set-shifting.

(6) The Stroop test of inhibitory control ability, mental vitalityand flexibility.

(7) Controlled Oral Word Association test (COWA) (Kang andNa, 2003) to measure phonemic word fluency (generatingwords beginning with the Korean characters , s, andsemantic word fluency (generating names of animals andsupermarket goods).

(8) The Korean version of the Boston Naming Test (K-BNT) (Kimand Na, 1997) for language function.

(9) Raven’s Coloured Progressive Matrices (RCPM) for non-ver-bal reasoning ability.

(10) The Korean version of the California Verbal Learning Test (K-CVLT) (Kim and Kang, 1997) for verbal memory and the Reycomplex figure test of visual memory (Meyers and Meyers,1995).

All NP tests were performed by a single examiner (CKM).

2.3. ERP study

Two ERP studies were carried out; the first (ERP1) was per-formed just before giving the first dose of TPM and the second(ERP2) was conducted at 12–16 weeks after TPM administrationwas started. EEGs were recorded with a SynAmps 32-channelsamplifier (Neuroscan, Herndon, VA, USA) with 30 Quick-cap elec-trodes. The reference electrode was set to linked mastoid elec-

Table 2Comparison of neuropsychological scores between pre- and post-TPM (n = 18).

Pre-TPM scoremean (SD)

Post-TPM scoremean (SD)

General cognitive functionK-MMSE 28.6 (0.76) 28.7 (0.53)FSIQ short form 105.1 (9.11) 102.1 (10.0)

Attention and frontal lobe functionDigit span forward 7.8 (1.91) 6.3 (2.67)*

Digit span backward 6.8 (2.31) 5.0 (1.28)*

Corsi block forward 8.5 (2.70) 7.0 (1.60)*

Corsi block backward 7.7 (1.89) 6.5 (1.39)*

Trail-making test A (s) 38.3 (15.19) 43.9 (22.81)Trail-making test B (s) 107.7 (79.80) 93.7 (34.14)Digit symbol test 57.3 (14.04) 57.3 (18.03)WCST perseverative errors 9.3 (2.47) 10.6 (4.16)Stroop word correct responses 112.0 (0.00) 111.9 (0.35)Stroop colour correct responses 100.8 (8.18) 104.0 (15.64)Raven’s Coloured Progressive Matrices 32.0 (2.76) 32.0 (3.85)COWA phonemic word fluency 27.3 (10.49) 20.4 (11.83)*

COWA semantic word fluency 27.1 (6.32) 22.9 (5.34)*

LanguageK-BNT 50.8 (5.06) 50.7 (5.14)

1496 K.-Y. Jung et al. / Clinical Neurophysiology 121 (2010) 1494–1501

trodes, the impedance was kept below 5 kO, and the band-pass fil-ter setting was 0.05–70 Hz, with a sampling rate of 1000 Hz. Fpzserved as ground electrode. Two electrooculogram (EOG) channels(placed on the left and right outer canthi) were added to confirmeyeball movement, while removing EOG artefacts.

To investigate the cognitive ERP, an auditory oddball paradigmwas used (Coles et al., 1986). Frequent 1000-Hz tones were used asstandard and rare 2000-Hz tones as targets. Targets were pre-sented in a random order, with a probability of occurrence of20%, and an interstimulus interval of 2000 ± 200 ms. The stimuliwere delivered binaurally through light headphones at an intensityof 90 dB SPL. The subjects were required to respond to the stimulusof the target tone by pressing a push button with the right indexfinger. Each subject received two blocks of an auditory oddball taskcontaining a mix of 200 tones. Only those trials with correct re-sponses were included for further analysis. EEG activity was aver-aged for target and standard stimuli separately for �200 ms to800 ms post-stimulus (1000 ms). Epochs exceeding ±100 V in theEEG or in the EOG were not included in the average waveform.Mean number of epochs was 270.9 ± 38.3 for standard stimulusand 77.3 ± 37.5 for target stimulus.

For every subject, the averaged ERP for each electrode site wasobtained for each of the two stimulus-presentation conditions. Thetime windows for N100 (75–200 ms), P200 (150–300 ms) and P300(250–450 ms) were determined by visually inspecting individualand grand averaged waveforms at midline electrodes (Fz, Cz andPz). Peak amplitude was referred to a 200 ms pre-stimulus baselineand latency defined from stimulus onset.

2.4. Statistical analysis

We used paired t-tests to compare patients’ NP test scores be-fore and after initiation of medication. Fisher’s exact test was usedfor categorised variables.

The latency and amplitude of the ERP component was analysedwith repeated-measures analysis of variance (ANOVA). The within-subject factors were stimulus-presentation condition (two levels;standard vs target), regions of interest (three levels: Fz, Cz andPz) and treatment condition (two levels: pre- vs post-TPM). Thesignificant level of statistical tests was set to 0.05. The Green-house–Geisser correction was used to evaluate F ratios to controlType 1 error in repeated measure design. Bonferroni post hoc testswere used to identify the sources of significant ANOVA.

2.5. Source localisation of ERP components and statistical non-parametricmapping

Source localisation of intra-cerebral current source and statisti-cal non-parametric mapping (SnPM) for ERP components that dif-fered between conditions was performed using the sLORETAsoftware (The KEY Institute for Brain-Mind Research, Universityof Zurich, Switzerland) (Pascual-Marqui, 2002). LORETA computes

Table 1Clinical characteristics of patients.

Clinical variables NP test (n = 18) ERP study (n = 13)

Age, yr, mean (range) 33.0 (20–57) 31.8 (22–59)Gender, % male 61 62Education, yr, mean (range) 13.1 (10–18) 14.6 (10–18)Duration of epilepsy, yr, mean (range) 12 (3–30) 13 (4–30)TPM dose at last visit, mg/day (range) 123.6 (50–400) 127.0 (50–400)Seizure frequency, #/monthBaseline (range) 6.8 (4–20) 8.8 (0.08–4)At second study (range) 0.3 (0.5–10) 0.8 (0.08–10)

NP, neuropsychological; ERP, event-related potential; and TPM, topiramate.

a unique three-dimensional electric source distribution by assum-ing that the smoothest of all possible inverse solutions is the mostplausible, consistent with the assumption that neighbouring neu-rons are simultaneously and synchronously active (Pascual-Marquiet al., 1994). sLORETA computes the sources, estimated on the basisof standardised current density, allowing more precise sourcelocalisation than the older version of the LORETA software. sLORE-TA images represent the electrical activity at each voxel in theMontreal Neurological Institute (MNI) brain template as the ampli-tude of the computed current source density (A/m2) distributionfor epochs of brain electric activity on a grid of 6239 voxels at5 mm spatial resolution (Pascual-Marqui, 2002).

Statistically significant ERP components revealed by repeated-measures ANOVA were subject to statistical non-parametricpaired-sample t-tests (SnPM) at each voxel to evaluate the effectsof TPM on ERP. Statistical significance was assessed non-paramet-rically with a randomisation test which corrects for multiple com-parisons (Nichols and Holmes, 2002). The corresponding alphalevel was 0.01. The voxels with significant differences were locatedin specific brain regions using sLORETA images. Based on t values,red colour meant increased activation and blue colour meant de-creased activation by TPM as compared to the base condition.

3. Results

3.1. Clinical characteristics

Six patients did not undergo follow-up NP testing; three werelost during follow-up and three changed their medications becauseof reduced appetite, paresthesia after TPM administration and the

MemoryK-CVLT total 51.1 (10.06) 52.8 (10.95)K-CVLT short delay free recall 10.2 (4.22) 10.5 (3.78)K-CVLT long delay free recall 11.0 (3.25) 11.5 (3.92)K-CVLT recognition 14.8 (1.31) 14.5 (2.56)RCPM copy 35.3 (0.46) 35.1 (1.49)RCPM immediate recall 21.1 (7.64) 19.7 (12.79)RCPM delayed recall 21.3 (8.94) 20.4 (10.43)RCPM recognition 20.0 (2.14) 20.6 (2.96)

TPM, topiramate; K-MMSE, Korean version of mini-mental status examination;FSIQ, full-scale intelligent quotient; WCST, Wisconsin Card Sorting Test; COWA,Controlled Oral Word Association test; K-BNT, Korean version of the Boston NamingTest; K-CVLT, Korean version of the California Verbal Learning Test; and RCPM,Raven’s Coloured Progressive Matrices.* P < 0.05.

Fig. 1. Grand average of the ERPs response to standard (A) and target (B) auditory stimulus pre- and post-treatment with topiramate (TPM).

K.-Y. Jung et al. / Clinical Neurophysiology 121 (2010) 1494–1501 1497

1498 K.-Y. Jung et al. / Clinical Neurophysiology 121 (2010) 1494–1501

administration of an additional AED during follow-up due touncontrolled seizures. Thus, a total of 18 patients completed bothsets of NP tests (NP group). Of 18 patients, five were excluded fromthe ERP analysis because their ERP data were obtained using a dif-ferent EEG machine. Thus, 13 patients (seven men) were includedin the final ERP analysis (ERP group). Six patients had complex par-tial seizures and seven had secondarily generalised tonic clonic sei-zures. No significant difference was observed in clinical features,including age, gender, duration of epilepsy, duration of educationor dose of TPM between NP group and ERP group (Table 1). Seizurefrequency significantly decreased with TPM monotherapy in bothgroups.

3.2. NP tests

No significant difference was found in the NP test between theNP and ERP groups. The mean daily TPM dose was 123.6 ± 86.0 mg(range 50–400), and the interval between pre- and post-TPM NPtests was 205.5 ± 31.7 days (range 182–276). Paired t-tests be-tween pre- and post-TPM treatment NP test scores showed thatTPM significantly decreased digit span forward (t = 3.692, p =0.001), digit span backward (t = 4.513, p < 0.001), Corsi block for-ward (t = 4.279, p < 0.001), Corsi block backward (t = 3.823, p =0.001), COWA phonemic word fluency test (t = 3.271, p = 0.005)and COWA semantic word fluency test (t = 2.688, p = 0.016) scores,indicating a decline in attention, working memory and verbal flu-ency. However, no significant difference was observed pre- topost-TPM treatment in the NP test scores of other tests (Table 2).

3.3. ERP and behavioral data

Subjects achieved a hit rate of 99.7% in ERP1 and 99.5% in ERP2(Wilcoxon signed-rank test, p = 0.59). The mean reaction time to atarget stimulus was 444.8 ms in ERP1 and 447.9 ms in ERP2(p = 0.95). The ERP waveforms and topographical distribution oftrials are shown in Fig. 1 and Table 3. N100 was predominant atFz, followed by a P200 with a topographical maximum at Cz anda P300 with a parietal maximum (Fig. 2). In standard trials, onlya frontal predominant N100 and a central predominant P200 wereelicited.

An analysis of the N100 latency between pre-treatment andpost-treatment along the three midline electrodes (Fz, Cz and Pz)revealed no main effect of treatment (F1,12 = 0.24, p = 0.63) (Table4). The interaction between treatment and location in N100 latencywas also not significant (F1.1,13.6 = 0.32, p = 0.61). N100 amplitudeshowed significant location effect (F1.28,14.08 = 30.78, p < 0.001).However, interaction between treatment and location did notreach statistical significance (F2,23 = 0.32, p = 0.73).

Table 3Mean amplitude and latencies of each ERP component at midline electrodes.

Channel N100 amplitude (uV)

Target Standard

Pre-TPM Post-TPM Pre-TPM Post-TPM

Fz �8.6 ± 2.7 �8.0 ± 2.9 �8.6 ± 3.2 �8.9 ± 3.2Cz �7.9 ± 2.2 �6.9 ± 2.3 �7.4 ± 2.4 �7.9 ± 1.7Pz �4.2 ± 2.7 �3.9 ± 1.5 �3.1 ± 1.6 �3.4 ± 0.8

P200 amplitude (uV)Fz 4.6 ± 3.9 7.2 ± 5.3 3.2 ± 4.9 7.8 ± 2.8Cz 5.7 ± 6.5 6.9 ± 5.5 6.6 ± 7.9 9.1 ± 4.6Pz 3.6 ± 4.9 3.2 ± 2.6 4.1 ± 4.3 4.0 ± 2.4

P300 amplitude (uV)Fz 8.5 ± 4.5 8.6 ± 5.0Cz 10.2 ± 4.1 10.4 ± 4.6Pz 11.4 ± 3.0 10.7 ± 4.4

An analysis of the P200 latency between pre-treatment andpost-treatment along the midline electrodes revealed no main ef-fect of treatment (F1,12 = 1.32, p = 0.28) (Table 4). The interactionbetween treatment and location was also not significant (F2,23 =0.23, p = 0.75). P200 amplitude showed significant location effect(F2,23 = 8.19, p = 0.002). Furthermore, interaction between treat-ment and location was statistically remarkable (F2,23 = 7.46,p = 0.003). Post hoc analysis showed P200 amplitude was margin-ally increased after treatment with TPM at Fz electrode (p = 0.07,95% confidence interval for difference �7.61 to 0.31).

ANOVA analysis of P300 latency between pre-treatment andpost-treatment along the midline electrodes revealed no significantmain effect (F1,12 = 1.05, p = 0.33). Neither location effect nor inter-action between treatment and location was also significant inP300 latency. P300 amplitude between pre-treatment and post-treatment along the midline electrodes revealed no significant maineffect of treatment (F1,12 = 0.02, p = 0.89). P300 amplitude showedsignificant location effect along the midline electrodes (F2,23 = 4.84,p = 0.02). However, the interaction between treatment and locationwas not significant (F2,23 = 0.71, p = 0.50).

3.4. Correlation between NP tests and ERP data

We searched for any correlation between change of NP tests andERP data before and after treatment with TPM. NP items (Table 2)which showed significant changes with TPM were subject to Pear-son product-moment correlation analysis with P200 amplitude asonly this amplitude showed significant ANOVA results (Table 4).We found that COWA semantic word fluency showed distinctivenegative correlation with P200 amplitude at Fz electrode (correla-tion coefficient = �0.76, p = 0.047). Other items of NP tests do nothave any important correlation with P200 amplitude.

3.5. SnPM analysis of sLORETA current source distribution

P200 component, which was statistically significant in re-peated-measures ANOVA, was subject to statistical non-parametricmapping, implemented in sLORETA programme. Following treat-ment with TPM, SnPM showed that current density of P200 wassignificantly decreased in bilateral parieto-occipital regions andright orbitofrontal and prefrontal regions, and increased in a smallarea of left temporal and cingulate regions (Fig. 3).

4. Discussion

In the present study, we found that TPM had a significant effecton particular brain regions responsible for cognitive functions,

N100 latency (ms)

Target Standard

Pre-TPM Post-TPM Pre-TPM Post-TPM

100.7 ± 8.8 96.7 ± 12.4 98.0 ± 8.4 95.0 ± 7.799.3 ± 9.3 97.7 ± 12.7 97.0 ± 9.7 94.3 ± 7.592.0 ± 35.8 93.0 ± 26.9 85.0 ± 18.1 87.0 ± 14.7

P200 latency (ms)177.3 ± 17.0 183.3 ± 9.9 185.7 ± 29.4 200.0 ± 21.4183.3 ± 17.2 183.7 ± 20.3 180.0 ± 24.2 190.0 ± 17.7173.0 ± 29.0 174.0 ± 27.0 180.7 ± 30.9 187.7 ± 24.4

P300 latency (ms)333.3 ± 44.6 342.0 ± 40.1325.7 ± 22.8 341.3 ± 33.4328.7 ± 40.0 334.3 ± 38.1

Fig. 2. Voltage topographic scalp mapping of each ERP component at pre- and post-treatment with topiramate (TPM).

K.-Y. Jung et al. / Clinical Neurophysiology 121 (2010) 1494–1501 1499

with a specific time course during cognitive tasks. sLORETA dem-onstrated that TPM significantly decreased current density mainlyin the bilateral posterior cortical areas and a part of frontal lobe.

TPM is considered one of the most effective anti-epilepticagents, but it has been associated with cognitive impairment as as-sessed by NP tests, particularly in language function, workingmemory and attention, in studies of healthy volunteers (Aldenk-amp, 2000; Aldenkamp and Bodde, 2005; Meador et al., 2005)and in patients with epilepsy (Blum et al., 2006; Martin et al.,1999; Meador et al., 2003). Likewise, NP tests in our study alsoshowed significant decreases in attention, working memory andverbal fluency after TPM treatment, although general intelligence,frontal executive function and verbal and visual memory functionswere preserved. Thus, certain functional brain regions may bemore affected by TPM than other areas.

In ERP study, only P200 amplitude significantly increased at Fzelectrode after long-term TPM administration. No considerablechanges were observed in N100 and P300 component. N100 isreflective of initial processing in a sensory perception which is re-lated to selective attention. The P300 has been proposed as an in-dex of multiple cognitive processes, including context updating orallocation of processing resources and decision-making (Polich andKok, 1995). In our study, P300 did not show any significant changewhich is in line with the previous study (Smith et al., 2006). P200has been reported to be related to evaluation of task-relevant stim-uli which is achieved by suppressing irrelevant features or enhanc-ing relevant ones (Crowley and Colrain, 2004; Kim et al., 2008).Thus, it is attributable to the preparation of the response, repre-senting a necessary step before a P300 can be elicited. P200 ampli-tude showed linear increase with age, particularly in anterior

Table 4Summary of analysis of variance.

Factor N100 P200

Amplitude Latency Amplitude

F P F P F

Tx <1 – <1 – 1.31Loc 30.78 <0.001 2.67 – 8.19Tx � Loc <1 – <1 – 7.46

Tx, treatment with topiramate; Loc, location of midline electrode; and Tx � Loc, interac

electrodes (Crowley and Colrain, 2004). In addition, it was reportedthat P200 amplitude was increased in harder tasks than easier ones(Kim et al., 2008). Therefore, increased P200 amplitude induced byTPM in the present study indicates that subjects may have diffi-culty in categorisation and classification of incoming stimulus ofoddball tasks. An increase of P200 amplitude is significantlycorrelated with semantic word fluency which is related to frontallobe function. However, as other NP tests pertaining to frontalfunction did not show significant correlation with P200 amplitude,further study is required to verify this relationship.

In sLORETA analysis, SnPM of the LORETA images showedsignificantly decreased current density associated with P200 inmultiple brain regions including prefrontal, temporolimbic andparieto–occipital association areas, after TPM treatment in patientswith epilepsy. Recently, a functional MRI study revealed significantunder activation in the inferior frontal cortex in a TPM-treatedgroup compared to a control epilepsy group (Jansen et al., 2006),which is in line with the present study. It is remarkable thatP200 is also associated with increased current density in smallareas of temporal and cingulate cortices. Because temporolimbicregion is one of the most common epileptic foci and has huge con-nection with other areas, increased current density in these regionsmight be relevant to epileptic activities, although it cannot be ex-plained clearly.

The results of the present study were limited by the technicalconstraints associated with sLORETA analysis. sLORETA cannotmake high-resolution images, and is known to occasionally resultin localisation errors, particularly sources located deeply in thebrain such as medial and inferior aspect of the brain (Kringset al., 1999). Furthermore, only a small number of electrodes

P300

Latency Amplitude Latency

P F P F P F P

– 1.32 – <1 – 1.05 –0.002 1.13 – 4.84 0.03 <1 –0.003 <1 – <1 – <1 –

tion between Tx and Loc.

Fig. 3. Voxel-wise statistical non-parametric map (SnPM) of sLORETA images on target ERPs of P200 between pre- and post-treatment with topiramate at the 1% significancelevel after correction for multiple comparisons. Red colour indicates a significant increase and blue denotes a significant decrease in current density following treatment withtopiramate. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this paper.)

1500 K.-Y. Jung et al. / Clinical Neurophysiology 121 (2010) 1494–1501

(particularly less than 32-channels) used in the present study maypreclude high-resolution topographical mapping of currentsources (Lantz et al., 2003).

5. Conclusions

Our study suggests that TPM affects selective brain regionswhich may be related to cognitive side effects. However, owingto the limited number of electrodes used in the present study,the functional regions involved require further confirmation fromstudies that use methods with higher spatial resolution.

6. Disclosure

The authors report no conflicts of interest.

Acknowledgements

This work was supported by Grant (2009K001257) from theBrain Research Center of the 21st Century Frontier Research Pro-gram funded by the Ministry of Science and Technology of theRepublic of Korea, by the Samsung Biomedical Research Institutegrant, #SBRI C-B0-237-1 and by a grant of the Korea Healthcaretechnology R&D Project, Ministry for Health, Welfare & Family Af-fairs, Republic of Korea (A090794).

References

Aldenkamp AP. Cognitive effects of topiramate, gabapentin, and lamotrigine inhealthy young adults. Neurology 2000;54:271–2.

Aldenkamp AP, Baker G, Mulder OG, Chadwick D, Cooper P, Doelman J, et al. Amulticenter, randomized clinical study to evaluate the effect on cognitivefunction of topiramate compared with valproate as add-on therapy tocarbamazepine in patients with partial-onset seizures. Epilepsia2000;41:1167–78.

Aldenkamp AP, Bodde N. Behaviour, cognition and epilepsy. Acta Neurol ScandSuppl 2005;182:19–25.

Anderer P, Saletu B, Saletu-Zyhlarz G, Gruber D, Metka M, Huber J, et al. Brainregions activated during an auditory discrimination task in insomniacpostmenopausal patients before and after hormone replacement therapy:low-resolution brain electromagnetic tomography applied to event-relatedpotentials. Neuropsychobiology 2004;49:134–53.

Anderer P, Saletu B, Semlitsch HV, Pascual-Marqui RD. Perceptual and cognitiveevent-related potentials in neuropsychopharmacology: methodological aspectsand clinical applications (pharmaco-ERP topography and tomography).Methods Find Exp Clin Pharmacol 2002a;24(Suppl. C):121–37.

Anderer P, Saletu B, Semlitsch HV, Pascual-Marqui RD. Structural and energeticprocesses related to P300: LORETA findings in depression and effects ofantidepressant drugs. Methods Find Exp Clin Pharmacol 2002b;24(Suppl.D):85–91.

Anderer P, Saletu B, Wolzt M, Culic S, Assandri A, Nannipieri F, et al. Double-blind,placebo-controlled, multiple-ascending-dose study on the effects of ABIO-08/01, a novel anxiolytic drug, on perception and cognition, utilizing event-relatedpotential mapping and low-resolution brain electromagnetic tomography. HumPsychopharmacol 2008;23:243–54.

Barry RJ, Johnstone SJ, Clarke AR, Rushby JA, Brown CR, McKenzie DN. Caffeineeffects on ERPs and performance in an auditory Go/NoGo task. ClinNeurophysiol 2007;118:2692–9.

Blum D, Meador K, Biton V, Fakhoury T, Shneker B, Chung S, et al. Cognitive effectsof lamotrigine compared with topiramate in patients with epilepsy. Neurology2006;67:400–6.

Chen YJ, Kang WM, So WC. Comparison of antiepileptic drugs on cognitive functionin newly diagnosed epileptic children: a psychometric and neurophysiologicalstudy. Epilepsia 1996;37:81–6.

Chung SS, McEvoy LK, Smith ME, Gevins A, Meador K, Laxer KD. Task-related EEGand ERP changes without performance impairment following a single dose ofphenytoin. Clin Neurophysiol 2002;113:806–14.

Coles MGH, Donchin E, Porges SW. Psychophysiology: systems, processes, andapplications. New York: Guilford Press; 1986.

Crowley KE, Colrain IM. A review of the evidence for P2 being an independentcomponent process: age, sleep and modality. Clin Neurophysiol 2004;115:732–44.

Enoki H, Sanada S, Oka E, Ohtahara S. Effects of high-dose antiepileptic drugs onevent-related potentials in epileptic children. Epilepsy Res 1996;25:59–64.

Higuchi Y, Sumiyoshi T, Kawasaki Y, Matsui M, Arai H, Kurachi M.Electrophysiological basis for the ability of olanzapine to improve verbalmemory and functional outcome in patients with schizophrenia: a LORETAanalysis of P300. Schizophr Res 2008;101:320–30.

Jansen JF, Aldenkamp AP, Marian Majoie HJ, Reijs RP, de Krom MC, Hofman PA, et al.Functional MRI reveals declined prefrontal cortex activation in patients withepilepsy on topiramate therapy. Epilepsy Behav 2006;9:181–5.

Kang Y, Na D. Seoul neuropsychological screening battery. Seoul: Human BrainResearch & Consulting Co.; 2003.

Kang Y, Na D, Hahn S. A validity study on the Korean mini-mental state examination(K-MMSE) in dementia patients. J Korean Neurol Assoc 1997;15:300–8.

Kim H, Na D. Korean–boston naming test (K-BNT). Seoul: Hakji Co.; 1997.Kim J, Kang Y. Korean-California verbal learning test (K-CVLT): a normative study.

Kor J Clin Psychol 1997;16:379–95.Kim KH, Kim JH, Yoon J, Jung KY. Influence of task difficulty on the features of event-

related potential during visual oddball task. Neurosci Lett 2008.Kockelmann E, Elger CE, Helmstaedter C. Significant improvement in frontal lobe

associated neuropsychological functions after withdrawal of topiramate inepilepsy patients. Epilepsy Res 2003;54:171–8.

Krings T, Chiappa KH, Cuffin BN, Cochius JI, Connolly S, Cosgrove GR. Accuracy ofEEG dipole source localization using implanted sources in the human brain. ClinNeurophysiol 1999;110:106–14.

Lantz G, Grave de Peralta R, Spinelli L, Seeck M, Michel CM. Epileptic sourcelocalization with high density EEG: how many electrodes are needed? ClinNeurophysiol 2003;114:63–9.

Lee HW, Jung DK, Suh CK, Kwon SH, Park SP. Cognitive effects of low-dosetopiramate monotherapy in epilepsy patients: a 1-year follow-up. EpilepsyBehav 2006;8:736–41.

Lim Y, Lee W, Lee W, Park J. The study on the accuracy and validity of KoreanWechsler intelligence scale short forms: a comparison of the WARD7 subtest vs.Doppelt subset. Kor J Clin Psychol 2000;19:563–74.

Martin R, Kuzniecky R, Ho S, Hetherington H, Pan J, Sinclair K, et al. Cognitive effectsof topiramate, gabapentin, and lamotrigine in healthy young adults. Neurology1999;52:321–7.

Meador KJ. Cognitive and memory effects of the new antiepileptic drugs. EpilepsyRes 2006;68:63–7.

Meador KJ, Loring DW, Hulihan JF, Kamin M, Karim R. Differential cognitive andbehavioral effects of topiramate and valproate. Neurology 2003;60:1483–8.

Meador KJ, Loring DW, Vahle VJ, Ray PG, Werz MA, Fessler AJ, et al. Cognitive andbehavioral effects of lamotrigine and topiramate in healthy volunteers.Neurology 2005;64:2108–14.

Meyers J, Meyers K. Rey complex figure test and recognition trial. Psychologicalassessment resources, Inc.; 1995.

Michel CM, Murray MM, Lantz G, Gonzalez S, Spinelli L, Grave de Peralta R. EEGsource imaging. Clin Neurophysiol 2004;115:2195–222.

K.-Y. Jung et al. / Clinical Neurophysiology 121 (2010) 1494–1501 1501

Nichols TE, Holmes AP. Nonparametric permutation tests for functionalneuroimaging: a primer with examples. Hum Brain Mapp 2002;15:1–25.

Ohlmeier MD, Prox V, Zhang Y, Zedler M, Ziegenbein M, Emrich HM, et al. Effectsof methylphenidate in ADHD adults on target evaluation processing reflectedby event-related potentials. Neurosci Lett 2007;424:149–54.

Ozmenek OA, Nazliel B, Leventoglu A, Bilir E. The role of event related potentials inevaluation of subclinical cognitive dysfunction in epileptic patients. Acta NeurolBelg 2008;108:58–63.

Panagopoulos GR, Thomaides T, Tagaris G, Karageorgiou CL. Auditory event relatedpotentials in patients with epilepsy on sodium valproate monotherapy. ActaNeurol Scand 1997;96:62–4.

Park SP, Kwon SH. Cognitive effects of antiepileptic drugs. J Clin Neurol2008;4:99–106.

Pascual-Marqui RD. Standardized low-resolution brain electromagnetictomography (sLORETA): technical details. Methods Find Exp Clin Pharmacol2002;24(Suppl. D):5–12.

Pascual-Marqui RD, Michel CM, Lehmann D. Low resolution electromagnetictomography: a new method for localizing electrical activity in the brain. Int JPsychophysiol 1994;18:49–65.

Polich J, Kok A. Cognitive and biological determinants of P300: an integrativereview. Biol Psychol 1995;41:103–46.

Potts GF. An ERP index of task relevance evaluation of visual stimuli. Brain Cogn2004;56:5–13.

Ruijter J, Lorist MM, Snel J, De Ruiter MB. The influence of caffeine on sustainedattention: an ERP study. Pharmacol Biochem Behav 2000;66:29–37.

Saletu B, Anderer P, Di Padova C, Assandri A, Saletu-Zyhlarz GM. Electrophy-siological neuroimaging of the central effects of S-adenosyl-L-methionine bymapping of electroencephalograms and event-related potentials and low-resolution brain electromagnetic tomography. Am J Clin Nutr2002;76:1162S–71S.

Smith ME, Gevins A, McEvoy LK, Meador KJ, Ray PG, Gilliam F. Distinct cognitiveneurophysiologic profiles for lamotrigine and topiramate. Epilepsia2006;47:695–703.

Sun W, Wang W, Wu X, Wang Y. Antiepileptic drugs and the significance of event-related potentials. J Clin Neurophysiol 2007;24:271–6.

Thompson PJ, Baxendale SA, Duncan JS, Sander JW. Effects of topiramate oncognitive function. J Neurol Neurosurg Psychiatry 2000;69:636–41.

Trimble MR, Thompson PJ. Anticonvulsant drugs, cognitive function, and behavior.Epilepsia 1983;24(Suppl. 1):S55–63.