Address correspondence and reprint requests to Dr. F. Rosenow at Department of Neurology, Philipps University, Rudolf-Bultmann-Str. 8, 35033 Marburg, Germany. E-mail: email@example.com
Summary: Purpose: Topiramate (TPM) is a novel drug with broad antiepileptic effect in children and adults. In vitro studies suggest activity as sodium-channel blocker, as γ-aminobutyric acid type A (GABAA)-receptor agonist and as non–N-methyl-D-aspartate (NMDA)-glutamate receptor antagonist.
Methods: With transcranial magnetic stimulation (TMS), we evaluated which of the mechanisms of action of TPM detected in vitro are relevant for the modulation of human motor cortex excitability. In a double-blind, placebo-controlled, crossover study design, we investigated the effect of single oral doses of 50 mg and 200 mg TPM on motor thresholds, cortical silent period (CSP), and on intracortical inhibition (ICI) and intracortical facilitation (ICF) in 20 healthy subjects.
Results: A significant dose-dependent increase of ICI was noticed after 200 mg TPM as compared with placebo at short interstimulus intervals of 2 to 4 ms. TPM had no effect on motor thresholds or the CSP.
Conclusions: We conclude that a single dose of TPM selectively increases ICI by GABAAergic and/or glutamatergic mechanisms without a relevant influence on measures, depending on ion-channel blockade or GABAB-receptor activity. The decrease of intracortical excitability (as measured by ICI and ICF) caused by TPM may correlate with its lack of proconvulsive potential in idiopathic generalized epilepsy, because drugs without this action or with less pronounced action may exacerbate seizures in this condition.
Topiramate [2,3:4,5-bis-O-(1-methylethylidene)β-d-fructo-pyranose sulfamate] (TPM) is a structurally novel compound that has demonstrated a broad spectrum of antiepileptic activities in both experimental and clinical studies in children and adults with focal and generalized epilepsies (1–4). Several pharmacologic mechanisms of action of TPM have been identified and are thought to contribute to its anticonvulsant action (5). These include (a) its activation-dependent sodium channel–blocking effect, which reduces the duration and frequency of action potentials during spontaneous epileptiform bursts of neuronal firing (6–8); (b) positive modulation of γaminobutyric acid type A (GABAA) receptors to increase chloride fluxes across the membrane of cultured cerebral cells, resulting in hyperpolarization and subsequent stabilization of seizure threshold (9,10); (c) its inhibition of the kainate and aminohydroxy-methylisoxazole propionic acid (AMPA) glutamate receptor subtypes (11,12); (d) its inhibition of L-type high voltage–activated calcium channels (13); and (e) the weak inhibition of the carboanhydrase isoenzymes II and IV (14). Furthermore, it has been suggested that TPM increases cerebral GABA levels in humans (15–17). This would suggest that unspecific GABABergic effects also could be relevant.
Transcranial magnetic stimulation (TMS) can be used to explore the effects of antiepileptic drugs (AEDs) on different mechanisms influencing the excitability of human motor cortex (18,19). Different TMS measures of cortical excitability are preferentially influenced by different classes of AEDs: (a) motor thresholds (MTs) are significantly increased by single doses of ion-channel blockers such as carbamazepine (CBZ), phenytoin (PHT), valproate (VPA), losigamone (LSG), and lamotrigine (LTG) (18,19); (b) the cortical stimulation–induced silent period (CSP) is lengthened in a dose-dependent fashion by intrathecal application of baclofen, a selective GABAB-receptor agonist (20). Therefore the CSP, at least in part, depends on GABABergic intracortical inhibition (21); (c) intracortical excitability, as measured by the paired-pulse paradigm, is decreased by GABAergic drugs such as gabapentin (GBP) and baclofen, as well as by glutamate-receptor antagonists such as riluzole and dextromethorphan (19,22,23).
We used single and paired-pulse TMS to investigate which of the mechanisms of action of TPM detected in vitro are relevant for the modulation of the excitability of the human motor cortex and, therefore, presumably for the antiepileptic effect of this compound.
SUBJECTS AND METHODS
Twenty healthy volunteers participated in the study (nine men and 11 women; mean age, 27.8 ± 5.8 years; range, 21–52 years). None of the subjects had a history of neurologic illness or was taking medication at the time of the study. All subjects were right-handed, except one man and one woman. The subjects were instructed not to take any neuro- or psychoactive drugs, alcohol included, for ≥24 h before the experiments. The study was conducted according to the Declaration of Helsinki, and approval was obtained from the Ethics Committee of the Philipps-University Marburg. All subjects gave their informed written consent.
Experimental procedures were based on those described by Kujirai et al. (24). Motor evoked potentials were recorded with surface electromyogram (EMG) from the abductor digiti minimi muscle (ADM) contralateral to the dominant side of motor cortex, with the active electrode near the motor point and the reference electrode placed on the metacarpophalangeal joint. The raw signal was amplified and bandpass filtered (20 Hz to 10 kHz). The EMG raw signal was digitized (analog/digital rate, 40 kHz) and recorded onto a PC by using a data-collection and -averaging program (Magnetix; Center for Sensorimotor Research, Munich, Germany) for offline analysis.
Both resting and active conditions were studied for all subjects. During the resting condition, EMG silence was monitored by visual and auditory feedback. During the active condition, subjects contracted the ADM voluntarily with ∼30% of maximal force, also monitored by audiovisual feedback. Between the trials, subjects relaxed for several minutes to avoid fatigue.
TMS was delivered through a focal figure-of-eight–shaped magnetic coil (each loop measured 90 mm in external diameter) connected to two Magstim 200 magnetic stimulators via a BiStim-module (all Magstim; Whitland, Dyfed, U.K.). Subjects were seated in an armchair with the head fixed in a plastic foam headrest. The coil was placed flat on the skull over the dominant motor cortex at the site optimal for contralateral ADM activation. The current induced in the brain beneath the junction of the coil flowed from posterior to anterior, approximately perpendicular to the assumed line of the central sulcus (left hemisphere in 18 subjects, right hemisphere in two subjects). This is thought to be the most effective way to activate the corticospinal system transsynaptically (25). This coil position was marked directly on the scalp to ensure accurate repositioning of the coil. In all double-pulse procedures, the interval between trials was randomly changed between 4 and 6 s; in single-pulse procedures, the intertrial interval was randomly changed between 8 and 10 s by the computer.
Measures of motor cortex excitability
Several TMS measures were used to investigate motor cortex excitability.
1The resting motor threshold (RMT) was defined as the lowest stimulator output intensity needed to induce a motor evoked potential (MEP) of >50 μV peak-to-peak amplitude in at least four of eight consecutive trials, with a step-by-step intensity resolution of 1% of the maximal stimulator output (26).
2The active motor threshold (AMT) was defined as the lowest stimulator-output intensity needed to induce a MEP of >50 μV in the moderately active ADM muscle (∼30% of maximal force) in at least four of eight consecutive trials.
3Intracortical inhibition (ICI) and facilitation (ICF) were tested at short interstimulus intervals (ISIs) of 2, 3, and 4 ms and longer interstimulus intervals of 10 and 15 ms by using the method described elsewhere (24,26). The effect of a conditioning stimulus delivered before a second stimulus (test stimulus) was investigated. The conditioning stimulus was set at an intensity of 90% of AMT that produces no changes of excitability in the spinal cord (24). The intensity of the test stimulus was adjusted to produce MEPs of ∼1.5 mV peak-to-peak amplitude at rest. Fifteen trials of the unconditioned control single test stimuli and 15 paired stimuli of each ISI were recorded, delivered 5 s apart in random order generated by the data-collection and data-averaging program. The curve average was used to define amplitudes. The conditioned response was defined as the mean amplitude of the conditioned responses belonging to the ISI, expressed as percentage of the mean amplitude of the unconditioned test responses. The amplitudes of the MEPs were measured from peak to peak.
4The cortical stimulation–induced silent period (CSP) was measured in 10 trials each at stimulus intensities of 100, 110, 120, 130, and 140% AMT in the moderately active ADM, as recommended by Tergau et al. (27,28). CSP duration was defined in the individual trials from the time of the first turning point of the stimulus-induced MEP to the first reoccurrence of voluntary EMG activity. This measure was selected because in our investigation, the MEP latency with this point had a lower variability (standard deviation) than did the MEP latency with the initial upward deflection. The time of the first turning point was determined by the PC software. CSP offset time was determined by a single investigator; therefore, interobserver variability was excluded (29). For each stimulus intensity, an average was calculated (23).
TPM has favorable pharmacokinetic properties. It has a high bioavailability, is rapidly absorbed, and maximal plasma levels are reliably reached within 2 h (30,31). There is a close correlation between plasma and cerebrospinal fluid (CSF) levels, and transport across the blood–brain barrier is not restricted by a saturable transport mechanism (32). For these reasons, all parameters of motor cortex excitability were measured before (baseline) and 2 h after a single oral dose of either placebo or 50 or 200 mg of TPM. Daily doses of 50 to 200 mg have been proven to be effective in the treatment of focal epilepsy by monotherapy trials (4). In our study, doses were given as identical amounts of white powder of indistinguishable bitter taste. Each subject participated in three sessions. Sessions were applied in a pseudo-randomized order established by one of the investigators. Subjects and the investigators performing the measurements and analyzing the raw data were blinded for the order of application until all data were acquired.
Stimulation sessions lasted ∼3.5 h, during which subjects received a total of ∼150 paired and 180 single stimuli. Stimulation sessions were separated by ≥3 days to guarantee return to baseline levels, and a maximum of 28 days was allowed between sessions. To avoid influence of the subjects' biologic rhythm on the measurement, the sessions took place at the same time of day.
The intensity of those side effects (SEs) associated with the long-term use of TPM was recorded in a semiquantitative fashion (0, not noticed, to 3, strong) by using a standardized questionnaire. A cumulative side-effect score was calculated and correlated with the dose and with the measures of cortical excitability to investigate whether effect and side effects can occur independently.
Because the data did not always show a normal distribution, the Wilcoxon signed rank test for matched pairs was used for statistical analysis. Before we conducted the experiments, we defined two primary outcome measures: the amplitude difference under the ISI3 condition and the duration of the CSP at 130% AMT stimulation. These measures were chosen because they reflect GABAergic inhibition, which was thought to be most likely involved in the effect of TPM on the human motor cortex. As fixed in the protocol, the comparison of 200 mg of TPM versus placebo was of main interest (confirmatory analysis). The level of significance was set to p < 0.05 for the confirmatory analysis. After Bonferroni adjustment for applying two tests, a value of p < 0.025 per test was regarded as statistically significant. Subsequently baseline values of all parameters were compared with the values measured 2 h after ingestion of placebo or 50 or 200 mg of TPM (explanatory analysis). In total 38 tests were performed (see Table 1). Thus adjusting for multiple testing also in the explanatory analysis, a level of significance of p < 0.001 per test was chosen.
Table 1. Explanatory comparison of TMS measurements at baseline versus 2 h after placebo, 50 mg, or 200 mg topiramate
Placebo (mean ± SD)
Topiramate, 50 mg (mean ± SD)
Topiramate, 200 mg (mean ± SD)
Values for motor thresholds (RMT, AMT) are given in percentage of maximal stimulator output. MEP amplitudes for different interstimulus intervals (ISIs) are given in percentage of the unconditioned MEP amplitude, and CSP durations are given in milliseconds. The Wilcoxon signed rank test for matched pairs was used for statistical analysis.
CSP, cortical silent period; MEP, motor evoked potential; TMS, transcranial magnetic stimulation; BSL, baseline; MED, median.
Bold indicates that this difference remains statistically significant after Bonferroni correction for multiple testing.
41.4 ± 4.2
41.6 ± 4.7
41.2 ± 5.4
41.1 ± 5.7
41.2 ± 5.5
40.4 ± 4.8
29.5 ± 4.4
29.2 ± 4.1
29.2 ± 4.2
29.0 ± 4.1
29.9 ± 5.3
29.4 ± 5.3
ISI 2 (%)
52.8 ± 22.3
52.1 ± 23.1
62.9 ± 23.6
54.6 ± 22.2
63.7 ± 27.8
49.1 ± 24.0
ISI 3 (%)
48.2 ± 25.4
57.9 ± 29.4
59.9 ± 23.6
61.3 ± 24.6
57.0 ± 28.4
49.6 ± 26.4
ISI 4 (%)
71.0 ± 21.6
70.3 ± 22.7
79.0 ± 20.3
82.8 ± 24.2
75.0 ± 20.9
66.1 ± 22.1
ISI 10 (%)
126.6 ± 30.6
133.2 ± 32.3
146.1 ± 28.9
125.5 ± 28.2
147.0 ± 32.2
124.0 ± 31.9
ISI 15 (%)
129.0 ± 21.7
128.8 ± 26.7
148.8 ± 46.3
134.9 ± 24.7
132.5 ± 28.3
119.5 ± 14.9
22.3 ± 7.5
21.1 ± 7.3
21.9 ± 8.6
20.5 ± 8.3
21.5 ± 11.8
22.7 ± 11.4
71.4 ± 22.7
67.9 ± 20.8
71.7 ± 26.6
68.4 ± 28.7
71.9 ± 23.5
64.1 ± 19.3
130.4 ± 34.3
127.8 ± 35.9
134.0 ± 39.6
127.0 ± 38.7
136.7 ± 32.0
129.7 ± 32.8
171.3 ± 33.3
175.3 ± 31.4
174.6 ± 39.5
166.7 ± 37.6
172.2 ± 27.1
169.3 ± 35.5
197.3 ± 33.2
213.3 ± 44.3
200.6 ± 44.2
193.6 ± 36.9
196.0 ± 34.2
198.7 ± 40.3
The ICI at an ISI of 3 ms increased significantly 2 h after a single dose of 200 mg of TPM as compared with 2 h after placebo (p = 0.0072). This was consistent with the general trend of the data (see later).
The CSP duration at 130% AMT stimulation (primary outcome measure) was not significantly different after a single dose of 200 mg TPM as compared with placebo (p = 0.2443).
In Table 1, the data at baseline and 2 h after placebo or 50 or 200 mg TPM are given as means ± standard deviations. The p values given are for the comparison baseline versus 2 h after placebo, 50 mg, or 200 mg TPM by using the Wilcoxon signed rank test for matched pairs. Figure 1 shows the same data as medians and 25% and 75% quartiles.
The resting and active motor thresholds (RMTs, AMTs) were not influenced by a single dose of 50 or 200 mg TPM, as compared with baseline (Table 1, Fig. 1).
Paired pulse TMS at short interstimulus intervals: intracortical inhibition
With short ISIs of 2–4 ms, ICI was increased 2 h after a single oral dose of 50 or 200 mg TPM as compared with baseline measures (Table 1). This increase occurred in a dose-dependent fashion (Fig. 1) and was most robust for the comparison of 200 mg with baseline at an ISI of 2 ms (p = 0.0007).
Paired pulse TMS at long interstimulus intervals: intracortical facilitation
With long ISIs of 10 and 15 ms, ICF was decreased 2 h after a single oral dose of 200 mg TPM as compared with baseline measures. However, this decrease was not statistically significant (Fig. 1, Table 1).
Cortical stimulation–induced silent period
The duration of the CSP expectedly increased with increasing stimulation intensities between 100 and 140% AMT. The rate of increase was similar between groups or when compared with baseline measures or placebo (Table 1, Fig. 2). At a stimulus intensity of 110% AMT, the CSP was shorter after 200 mg TPM as compared with baseline, and at a stimulus intensity of 140% AMT, the CSP was longer after placebo as compared with baseline. None of these differences was statistically significant (Fig. 2, Table 1).
As a rule, values obtained after placebo were not different from baseline values. Therefore we regarded minor differences observed for single measures [e.g., the comparison of baseline vs. placebo at an ISI of 3 ms (Table 1)] as incidental and not related to the general trend of the data.
After the administration of placebo, three subjects reported adverse effects of low intensity (average SE score, 0.2 ± 0.3); after 50 mg TPM, eight subjects reported mild adverse effects (average SE score, 1.4 ± 1.7). After administration of 200 mg TPM, all but one subject reported mild to moderate adverse effects (average SE score, 7.1 ± 3.7) such as vertigo, paresthesia, somnolence, slowing of mental processing, bradykinesia, ataxia, and taste perversion starting at ∼30 min after TPM intake and resolving within 6–24 h. Many of the side effects observed resembled those seen with long-term administration (i.e., slowing of mental processing). The intensity and frequency of these effects were clearly dose dependent, but did not interfere with the ability of the subjects to complete the study. There was no correlation between a cumulative SE score and the observed changes in measures of motor cortex excitability 2 h after ingestion of 200 mg TPM (Spearman correlation: p = 0.87; R = 0.06).
The principal new finding of this study is that a single oral dose of TPM selectively and dose-dependently increased ICI without affecting motor thresholds (RMT, AMT) or the duration of the CSP.
Does topiramate influence TMS measures of GABAA-receptor agonism and/or a non-NMDA glutamate receptor antagonism in human motor cortex?
ICI and ICF are thought to reflect the activation of inhibitory and excitatory cortical interneuronal circuits by the conditioning TMS pulse (19,24). Both measures are regulated in part by GABA via postsynaptic GABAA-receptor activation (33,34). However, glutamate-receptor blockers of the NMDA subtype, such as memantine, dextromethorphan, or less specifically, riluzole have similar effects on ICI and ICF (22,23,35). These glutamate antagonists as well as GABAAergic drugs such as lorazepam (LZP) mainly decrease ICF and to a lesser extent increase ICI without significantly influencing motor thresholds or the CSP (19,22,23,34,35).
TPM similarly decreases cortical excitability via ICI and ICF, suggesting that the GABAA-receptor activation demonstrated in vitro may be relevant for the anticonvulsant activity of TPM (9). However, in contrast to other GABAAergic drugs or NMDA antagonists, TPM, as a non-NMDA antagonist and GABAA agonist, appears to be the first compound that has a more robust effect on ICI than on ICF. This difference may be due to the inhibition of non-NMDA glutamate receptor subtypes shown in vitro or by an interaction of the multiple modes of action reported for TPM (2,10).
The differential effect of TPM and other drugs on ICI and ICF provides further evidence for the hypothesis that ICF and ICI are independent phenomena with different thresholds, different sensitivities to magnetic coil orientation, and based on different neuronal circuitry (24,33).
Does topiramate influence TMS measures of ion channel blockade in human motor cortex?
The fact that motor thresholds and MEP amplitudes were not altered suggests that a single dose of TPM does not alter neuronal membrane excitability, neuromuscular transmission, or the number of corticospinal neurons activated by TMS. Furthermore, the stability of these measures indicates that the effects observed are specific for TPM and not due to unspecific effects such as somnolence, which may alter MEP amplitudes (36). These findings are in contrast to the effect of AEDs that act mainly by blockage of sodium or calcium channels. Single oral doses of CBZ, LSG, and PHT all significantly increase the motor thresholds without influencing intracortical excitability (19,37). This indicates that the inhibition of voltage-gated sodium channels by TPM demonstrated in vitro is not relevant for its anticonvulsive effect in humans, as had previously been suggested (6,7). The lack of a single-dose effect on MTs in human motor cortex of TPM as compared with PHT or LTG may relate to different effects of TPM as compared with the classic ion-channel blocking AEDs on the cellular level. McLean et al. (6) studied sodium-dependent action potentials in cultured mouse spinal cord neurons. In this model, depolarization-induced spontaneous repetitive firing (SRF) was blocked completely by PHT or LTG, whereas TPM even at high concentrations blocked SRF in only 30% of the neurons, suggesting that sodium-channel blockade may not be the primary mechanism responsible for the anticonvulsant activity of TPM.
Does topiramate influence TMS measures of GABAB-receptor agonism in human motor cortex?
The CSP, at least in its later part, is attributed to the activation of GABAergic inhibitory interneurons in the cortex (38). Because the CSP follows a time course similar to that of the late GABAB receptor–induced component of the IPSP, it has been suggested that the CSP reflects activation of postsynaptic GABAB receptors (39). The fact that baclofen, a selective GABAB-receptor agonist, prolongs the CSP in a dose-dependent fashion if applied intrathecally supports this view (20). TPM, in contrast, did not prolong CSP significantly, suggesting that GABABergic mechanisms are not relevant to its effect on intracortical excitability. Our results support GABAA receptor–specific binding as demonstrated by in vitro studies (5,12,40). Conversely, magnetic resonance imaging studies of the human occipital lobe suggest that TPM increases brain GABA concentration (15–17), which would be expected to result in an unspecific GABAB-receptor activation (26). These contrasting findings may reflect regional or anatomic differences between the occipital lobe and the motor cortex.
Do TMS measures of cortical excitability reflect proconvulsive effects of antiepileptic drugs?
It is very interesting that different AEDs presumed or proven to act via GABAergic mechanisms such as LZP, TPM, gabapentin (GBP), vigabatrin (VGB), and tiagabine (TGB) have different to opposite effects on ICI and ICF: TPM increases ICI more than it decreases ICF; LZP and GBP increase ICI less than they decrease ICF; VGB slightly decreases ICI and decreases ICF (19); and TGB decreases ICI significantly and tends to increase ICF in a dose-dependent fashion (26). In short, the inhibitory effect on cortical excitability as measured by ICI and ICF decreases in this order from TPM and LZP to TGB. The tendency of these drugs to increase the frequency of myoclonic seizures and absences in patients with idiopathic generalized epilepsy (IGE) or in certain animal models of IGE (i.e., the lethargic mouse) increases in the same order (41,42). LZP and TPM are actually effective in the treatment of IGE (3,43), whereas TGB is most likely to induce myoclonic seizures and absences in patients with IGE or even focal epilepsies (42,44,45). Therefore the effect of these GABAergic AEDs on ICI and ICF appears to correlate with their proconvulsive effect in IGE. The finding that ICI is selectively and strongly decreased in juvenile myoclonic epilepsy (a form of IGE), especially if untreated, could be interpreted as supportive evidence for this hypothesis (21). Furthermore, of the ion-channel–blocking agents, CBZ and PHT have proconvulsive activity in IGE and do not influence ICI, whereas LTG slightly increases ICI (19) and is effective in the treatment of absence epilepsy and in animal models of IGE (42). In patients with IGE, there are presumably structural changes in the function of ion channels (21), and the effects of prolonged use of TPM on GABAAergic and GABABergic inhibition may be quite different over time compared with a single-exposure study. Therefore prospective studies will be needed to investigate whether proconvulsive activity in IGE can be predicted from TMS measures of ICI and ICF.
With a placebo-controlled design, we did not observe a constant placebo effect. It could, therefore, be argued that placebo-controlled study designs are not necessary in TMS research. However, for principal statistical reasons, it may be considered prudent to apply standard placebo-controlled study designs to the field of TMS in the future.
The reproducibility of the results of paired-pulse TMS studies has been investigated. A high variability across subjects and, less important, between sessions was reported (46). To address this issue, we chose to use a relatively large number of subjects. Because the data were not distributed normally, nonparametric data analysis was performed. To avoid misinterpretation due to multiple testing, we used only two predefined parameters for confirmatory analysis and Bonferroni adjustment for multiple testing for both confirmatory and explanatory data analysis.
Are TMS measures predictive of anticonvulsive efficacy of TPM?
There was no apparent correlation between the measures of cortical excitability (effect) and the severity of the side effects occurring. Similarly, in clinical studies of TPM, a substantial proportion of the patients experienced significant seizure reduction or even seizure freedom, frequently without developing relevant side effects. Conversely, some 30% have to discontinue TPM due to intolerable side effects, lack of efficacy, or both (4). Currently, there are no reliable predictors for seizure reduction. The question arises whether the effect of a single dose of TPM on cortical excitability (especially ICI) may be predictive of long-term anticonvulsive efficacy in the individual patient. If this were the case, prediction of efficacy could become a new clinical application for paired-pulse TMS.
Acknowledgment: This work was partially supported by the ULRAN-Foundation Professorship for Neurology/Epileptology, by Neurokard, Butzbach, Germany, and by the Janssen-Cilag GmbH, Neuss, Germany.