Effect of Vagal Nerve Stimulation on Interictal Epileptiform Discharges: A Scalp EEG Study


Address correspondence and reprint requests to Dr. R. Kuba at Department of Neurology, Brno Epilepsy Centre, St. Anne's Hospital, Pekařská 53, 656 91, Brno, Czech Republic. E-mail: robert.kuba@fnusa.cz


Summary:  Purpose: To investigate the effects of acute vagal nerve stimulation (VNS) on interictal epileptiform discharges (IEDs).

Methods: Fifteen epilepsy patients, all of whom had been treated with VNS for ≥6 months, entered the study. In each subject, the absolute number of IEDs was counted at the baseline period (BP), the stimulation period (SP), six interstimulation periods (IPs), and the prestimulation period (PP), by using an original paradigm. The number of IEDs at the BP and the PP was compared with the number of IEDs at the SP and IPs. The results were correlated with other variables (the duration of VNS, the value of the output current, the duration of epilepsy, the type of epilepsy, the effect of VNS, and the effect of extrastimulation).

Results: We observed a significantly higher reduction in the number of IEDs in the SP and all the IPs as compared with the BP. We noticed a significantly higher reduction in the number of IEDs in the SP and in the first IP as compared with the PP. The reduction of IEDs was greater in patients who responded to VNS (>50% reduction of all seizures) and in patients who responded positively to magnetic extrastimulation. There were no other significant results in the reduction of IEDs when comparing other variables.

Conclusions: Short-term VNS reduces IEDs significantly. The reduction is most prominent during the SP (i.e., when the pulse generator is active). The value of reduction of IEDs is higher in patients who respond to VNS and in patients with positive experiences with magnetic extrastimulation. These results can be useful in predicting the effect of VNS.

Vagal nerve stimulation (VNS) is an effective alternative therapy for patients with intractable epilepsy. Several studies with small and large groups of epilepsy patients have shown the positive effects of VNS on the reduction of seizure frequency in children and adults with both partial and generalized epilepsy (1–4). Although how VNS works is unknown, a number of studies using functional imaging techniques, such as single-photon emission tomography (SPECT) and positron emission tomography (PET), have demonstrated widespread changes in blood flow and metabolism in several cortical and subcortical regions during short-term VNS (5,6). These widespread changes in blood flow and metabolism in various cerebral structures have formed the foundation for the hypotheses explaining the VNS effect. These hypotheses have stressed the importance of subcortical structures in VNS. It has been postulated that the vagal nerve, as a major autonomic afferent nerve, may influence the cortical processes through a reticular activating system or through a projection from the nucleus of a solitary tract to the substantia innominata and zona incerta (7,8).

EEG is one way to investigate the influence of VNS on cerebral structures. Most animal studies have demonstrated the suppression of both ictal and interictal epileptiform EEG findings during stimulation (7,9–11). Although Hammond et al. (12) provided proof of the immediate effect of VNS on the termination of the seizure activity in men, the influence of VNS on interictal EEG findings has not been demonstrated (7,12,13).

This study investigates the influence of VNS on interictal epileptiform EEG findings at the scalp in several periods of stimulation and correlates them to several clinical variables.


Fifteen epilepsy patients (four women, 11 men) with an average age of 36.0 ± 7.1 years, who had had epilepsy for an average of 24.2 ± 9.5 years, participated in the study. VNS therapy was recommended by a special committee of the Brno Epilepsy Centre. All of the patients had been implanted with a vagal nerve stimulator ≥6 months before the beginning of the study, and intermittent stimulation of the vagal nerve started on the day of implantation. A VNS device (model 100 NCP; Cyberonics Inc., Houston, TX, U.S.A.) was implanted, strictly for clinical purposes. A standard surgical technique was used to implant the VNS device (14,15). Informed consent was obtained from each patient before the study, and the study received approval from the Ethics Committee of St. Anne's Hospital.

Stimulation characteristics

The VNS lasted from 6 to 25 months, with an average of 15.0 ± 4.8 months. After stimulation, parameters of chronic VNS were applied (Table 1): a 20-Hz frequency in 14 patients, 30 Hz in one patient; a pulse width of 250 μs in 12 patients, 500 μs in three patients; a 30 s/5 min on/off cycle was used in five patients, 21 s/3 min in five patients, 18 s/1.8 min in two patients, 30 s/1.8 min in two patients, and 30 s/3 min in one patient. The output current ranged from 0.75 to 1.75 mA, with an average of 1.25 ± 0.23 mA. The magnetic parameters were adjusted in all subjects for extrastimulation. The values of this output current were 25 mA higher that that of the intermittent stimulation in each subject; the other parameters for the extrastimulation were identical to those mentioned earlier (Table 1).

Table 1.  Patient demographic and stimulation characteristics
  Parameters of intermittent stimulation 
No./Sex (F/M)/AgeDuration of
epilepsy (yr)
Pulse width
On/off cycles
Output current
Magnet output current
  1. F, female; M, male; No., number.


Clinical characteristics

The group was divided into two major subgroups (Table 2): eight patients with temporal lobe epilepsy (TLE), manifested by complex partial seizures (CPSs); and seven patients with extratemporal epilepsy (EE). The EE group was further subdivided into three patients with focal frontal seizures (manifest as tonic or clonic partial seizures), one patient with generalized tonic–clonic seizures (GTCSs), and three patients with a combination of both partial and generalized seizures (GCTSs, atonic, and absence-like seizures). Etiologically, mesiotemporal sclerosis was established to be a cause of the epilepsy in four patients; diffuse changes of white matter (perinatal asphyxia, perinatal trauma, meningoencephalitis) were found in four patients; glioma, angioma, and posttraumatic changes occurred in one patient each; and in four patients, no lesions were revealed by magnetic resonance imaging (MRI; cryptogenic etiology). Previous surgery for epilepsy had been performed on seven patients: a partial anterior callosotomy in five, stereotactic amygdalohippocampectomy was done on one patient, and a tailored lesionectomy had been performed on one patient. Concerning the long-term outcome of VNS, a patient was considered responsive when the stimulation abolished (50% of all seizures, regardless of the seizure type, as compared with the preimplant seizure frequency. If the reduction of seizures was <50%, the patients were marked as nonresponsive. In our group, nine patients were considered to be responsive, and six, nonresponsive. The next important criterion of the effect of VNS is the direct suppression of seizures through the use of magnetic extrastimulation at the onset of the seizure, either by the patients themselves or by a caregiver. In nine patients, we noted the positive effect of magnetic extrastimulation in abolishing >50% seizures when used; in six patients, we did not observe any positive effects (Table 2).

Table 2.  Patient clinical characteristics, outcome of VNS, and effect of extrastimulation
No.Previous surgery
for epilepsy
Type of epilepsy/
types of seizures
Etiology according to MRIOutcome of long-term
stimulation of vagal nerve;
responsive, R;
nonresponsive, N
Effect of
by magnet;
positive, P;
negative, N
  • No., number; TLE, temporal lobe epilepsy; FLE, frontal lobe epilepsy; CPSs, complex partial seizures; SPSs, simple partial seizures; GEN, generalized seizures; GTCSs, generalized tonic–clonic seizures; MTS, mesiotemporal sclerosis.

  • a

     According to histology by previous surgery.

1Stereotactic amygdalo-
2NoFLE/clonic partial,
Postraumatic malatia,
5Anterior partial
FLE/tonic partial,
Diffuse leukoencephalopathy
 (perinatal asphyxia)
6Anterior partial
GEN/GTCSsDiffuse encephalopathy
 (perinatal asphyxia)
9NoFLE/tonic partial,
10NoTLE/CPSPreviously extirpated
11Anterior patial
GEN, FLE/tonic partial,
 atonic, absence-type
Diffuse leukoencephalopathy,
 cortical atrophy (perinatal
12Anterior partial
GEN, FLE/tonic and
 clonic partial, GTCSs,
 absence type
Diffuse encephalopathy
 (purulent meningoencephalitis)
 (prolonged visceral
 and psychic aura)
Astrocytoma grade IIaRP
15Anterior partial
GEN, FLE/tonic partial,
 absence-like seizures,
Cryptogenic, cortical atrophyNP

EEG analysis

The 21-channel Brain Quick system (Micromed), with a 10-20 system, was used for scalp recording. EEG was amplified with a bandwidth of 0.4–100 Hz, at a sampling rate of 128 Hz. No seizures were noted 24 h before the EEG analysis or during the entire EEG procedure in any of the subjects.

Interictal epileptiform discharges (IEDs) recorded during the baseline period (BP)

The VNS was switched off 1 h before the baseline EEG recording in all patients. During the baseline EEG, the VNS remained inactive. The baseline EEG lasted 20 min, in which the patient lay in supine position, with closed eyes. Vigilance was evaluated continuously by experienced nursing staff on the basis of the character of the EEG curve. Afterward, ten 30-s periods every other minute were subtracted from the baseline EEG recording (Fig. 1). The sum of the IEDs was counted in each of the ten 30-s “windows” by an independent visual analysis by two of the authors (both board-certified electroencephalographers; R.K. and M.G.) who were blind to the outcome of the long-term VNS. The interobserver reliability was assessed with Cohen's kappa scale.

Figure 1.

Baseline EEG (BP), subtraction of ten 30-second periods, every other minute

Only definite spikes, sharp waves, and spike–wave complexes were considered epileptiform abnormalities; their shapes were distinguished by morphology and/or amplitude from the background activity. Nonepileptogenic or uncertain sharp discharges were not considered significant. Both focal and generalized IEDs were always included in the analysis. No runs of spikes were present in the evaluated EEG recordings.

Recording of IEDs during the stimulation period (SP)

After the completion of the baseline phase, the VNS was activated manually by one of the investigators with a magnet. The value of the output current used for this stimulation was the same as that presented in Table 1 (magnetic output current); the pulse generator produced the same intensity of current as was used during extrastimulation of the patients themselves. The stimulation lasted 30 s in each subject, and was set beforehand on the stimulator through the programming wand. The activation of the VNS by a magnet was performed 10 times every 4 min. The sum of both focal and generalized IEDs was counted during SP in all 10 activations under the same conditions as described in the previous paragraph (Fig. 2).

Figure 2.

Stimulation paradigm, division to SP, IPs and PP

Recording of IEDs during the interstimulation (IP) and prestimulation (PP) periods

The period from the end of the SP (after the first 30 s after the activation of the pulse generator) to the next stimulation was divided into six 30-s periods (60, 90, 120, 150, 180, and 210 s after the onset of the stimulation period), which were considered the IPs 1–6). After 10 activations, the number of both focal and generalized IEDs was counted for each IP separately under the same conditions as described in the previous section, and the sum of each period was used for the final analysis. The period between 210 and 240 s after the onset of the stimulation immediately preceded the next stimulation, and was called the prestimulation period (PP; Fig. 2).

This paradigm allowed us to obtain the total sum of IEDs recorded and counted in ten 30-s epochs of the EEG recording in the BP, the SP, the IPs 1–6, and the PP in each patient (Figs. 1 and 2). The absolute number of IEDs in all 15 patients in each period was calculated, and the numbers of IEDs in the SP and all the IPs were compared with the numbers of IEDs in the BP and the PP. An analysis of variance (ANOVA) was performed to establish any possible statistical significance. A p value of <0.05 was considered to be statistically significant. The statistically significant results in the reduction of the absolute number of IEDs were correlated, by the use of the Wilcoxon test with several variables, to find any significant correlation. The following variables were used for statistical analysis: type of epilepsy (temporal vs. extratemporal), duration of epilepsy, duration of VNS, output current used for extrastimulation during the test, long-term outcome of VNS (responsive vs. nonresponsive), and the effect of extrastimulation on abolishing seizures (positive vs. negative).


Evaluation of absolute number of IEDs

The absolute number of IEDs was larger in the BP than in any other measured period (SP, IP, and PP) in 12 (80%) of 15 patients. We noted a decrease in the absolute number of IEDs during the SP as compared with the BP in 14 (93%) of 15 patients and in 12 (80%) of 15 patients as compared with the PP (Table 3).

Table 3.  Absolute number of IEDs in all measured periods in all patients
No. of
BPSP1st IP2nd IP3rd IP4th IP5th IP6th IPPP
  1. BP, baseline period; SP, stimulation period; IP, interstimulation period; PP, prestimulation period; IEDs, interictal epileptiform discharges.


The average reduction of IEDs was the highest during the SP (42% of IEDs), as compared with the BP (100% of IEDs), followed by a slight increase in IEDs in the next period (IP 1, 56% of IEDs). This was followed by another slight increase in IEDs, which remained relatively stable during the whole IP up until the PP (67–72%; Fig. 3). The percentage reduction of IEDs reached a high statistical significance in the SP, IPs, and PP, when compared with the BP (Table 4).

Figure 3.

The sum of IEDs in all investigated periods

Table 4.  Statistical analysis in percentage IEDs change in SP and IPs compared with BP and PP
31–60 s
2nd IP
61–90 s
3rd IP
91–120 s
4th IP
121–150 s
5th IP
151–180 s
6th IP
181–210 s
211–240 s
  1. NS, not significant; BP, baseline period; SP, stimulation period; IP, interstimulation period; PP, prestimulation period; IED, interictal epileptiform discharge.

Absolute number of IEDs1125467635805777754779807801
Compared with BP (percentage
  change to) statistics
 p Value 0.0030.0010.0000.0000.0020.0010.0180.002
Compared with PP (percentage
  change to) statistics
 p Value 0.0450.048NSNSNSNSNS 

The PP (211–240 s) is the phase occurring immediately before each stimulation (SP). For this reason, the PP was evaluated separately. As a basis for evaluation, we noted a statistically significant difference in the absolute number of IEDs during this phase compared with the SP and the period immediately following it (31–60 s after stimulation, IP 1). No significant difference was noted between the PP and other IPs (Table 4).

When the absolute number of IEDs during the SP was compared to that of other phases, all the differences were statistically significant except the 31-60 second period (1st IP) following the stimulation.

These results allow us to conclude that a single stimulation reduces the number of IEDs in the 4 minute period after stimulation, with the highest suppression in the number of IEDs occuring during the stimulation. This percentage reduction of IEDs is relatively stable, and always significant in comparison to the baseline EEG (BP), when the pulse generator is switched off. When the PP (211-240 seconds) is used as a baseline, the same significant percentage reduction is noticed in the SP and in the subsequent 30-second period (1st IP). The inter-observer reliability was assessed using Cohen's kappa scale (kappa = 0.74).

Comparing to other variables (Table 5)

Table 5.  Correlation between IED reduction and other variables
Type of epilepsyTemporal36%NS62%NS
 Extratemporal51% 67% 
Long term of outcome of VNSResponsive34%0.04462%NS
 Nonresponsive60% 67% 
Effect of magnet extrastimulationPositive40%0.03854%0.012
 Negative58% 80% 
Duration of epilepsy  NS NS
Duration of vagal nerve stimulation  NS NS
Output current used by stimulation
 (in the study)

As the most statistically significant results, the difference in the percentage reduction of IEDs between the SP and the BP (SP/BP) and similarly between the SP and the PP (SP/PP), were used for mutual comparison with other selected variables.

The statistical analysis did not reveal any significant relationship between the decrease of the absolute number of IEDs when comparing SP/BP and SP/PP and the duration of epilepsy, duration of VNS, or the value of the output current used in this study for stimulation. The reduction of IEDs in the periods used for evaluation, i.e. SP/BP and SP/PP, did not significantly differ between temporal and extratemporal patients. The SP/BP percentage reduction of IEDs reached 36% (BP=100%) in TLE, and 51% in EE. Similarly, SP/PP percentage reduction reached 62% (PP=100%) in TLE, and 67%in EE.

On the other hand, we noted a significantly higher reduction of IEDs in the patients who responded to VNS (as compared to those who did not respond to VNS) during the SP as compared to the BP. SP/BP percentage reduction was 34% (BP=100%) in the responsive group, whereas it was 60% in the nonresponsive group (p=0.044). The difference in SP/PP percentage reduction between these two groups did not reach a statistical significance.

A significantly higher suppression of IEDs was ascertained in the group of patients who experienced the positive effect of magnetic extra-stimulation in abolishing their seizures, as compared to the group who did not. SP/BP percentage reduction reached 48% (BP=100%) in the group of patients who experienced a positive effect from extra-stimulation, but only 58% in the group who did not (p=0.038). Similarly, SP/PP reached 54/80% in responsive/nonresponsive patients (PP=100%) (p=0.012).


It is generally accepted that VNS is an effective alternative treatment of intractable epilepsy in both adults and children (1, 2, 3, 4). The clear pathophysiological mechanism of its action remains unknown. The majority of vagal nerve fibres are visceral afferents, which are widely distributed throughout the central nervous system (16). Recently published theories suggest that this is the basis for the widespread change in cerebral blood flow during acute VNS in various cortical and subcortical regions using PET ([15O]H2O) and SPECT (99mTc-ethyl cysteinate dimer) (5, 6). The effect of VNS on EEG activity in experimental animals has been reported (17). Experiments on animals have shown that repetitive vagal stimulation can cause both synchronization and desynchronization of the EEG, depending on the stimulus frequency and current strength (7, 9, 16). High-intensity and high-frequency (>70 Hz) vagal stimulation produces desynchronization in cortical EEG in cats, and lower intensity stimulation at the same rate causes synchronization. Desynchronization was similarly tested with slower stimulation between 20–50 Hz, but with a high intensity of output current. All these effects depend on what type of vagal fibre is stimulated (16).

Our study showed the clear effect of acute VNS on reducing IEDs in humans during the stimulation period as compared to the baseline. The significant decrease in the number of IEDs during the SP, as compared to the PP, is most interesting. The PP, the time period from 210 to 240 second after the onset of 30-second stimulations, always preceded the next stimulation. The interval between each stimulation was 4 minutes, imitating the real inter-stimulus interval in chronic VNS in most patients. It is possible to use this period as a baseline for further evaluation of potential significant changes of IEDs. When looking at the IP, we can see a rather stable reduction in IEDs between 60 and 210 seconds after a single stimulation. This reduction of IEDs in comparison to the BP is highly significant. Previous studies evaluating the effects of VNS on interictal EEG have produced controversial results. In animal studies, McLachlan (7) applied potassium penicillin G solution to the left somatosensory cortex of a rat to obtain focal spikes. McLachlan demonstrated that electrical stimulation of the left vagal nerve reduced interictal spike activity. He noted that the initial effect appeared 1 to 2 seconds after the stimulus onset, occurred throughout the stimulation period, and persisted for an average of 60 seconds. Although it is an experimental model, we noticed a very similar time relationship between the stimulation and the maximal reduction of IEDs in humans. The most expressive IED decrease, as compared to the BP and the PP were in the SP (during stimulation) and early IP (31 to 60 seconds after stimulation). Mc Lachlan used a 20–50 Hz frequency range, the same as in our study (20 or 30 Hz) and found no difference in the response depending on the frequency value (7). Another experimental study, presented by Lockard et al. (18), did not find that VNS reduces interictal spikes in monkeys in which the seizures had been caused by topical instillation of cobalt on the neocortex. This negative finding was present despite the positive effect of long-term VNS on seizure reduction. The effect of VNS was demonstrated in another experimental epilepsy model (the pentylenetetrazol model in rats). A 60 minute VNS pretreatment decreased the seizure duration, the total number of seizures, and the number of tonic seizures after the administration of pentylentetrazol, as comparing to controls (19). When studying the effects of VNS on interictal EEG, human studies have consistently provided negative results. Hammond et al. (12) published data from nine patients, that presented the negative effect of a single stimulation of the vagal nerve in patients with both occasional and frequent interictal epileptiform activity. Moreover, there were no obvious effects of stimulation on interictal EEG in patients with long bursts of spiky activity. Hammond et al. (12) explained the negative results, as compared to the positive effects from some animal studies, by the differences in EEG recorded directly from the cortex in anaesthetised animals and from the scalp in awake humans. Salinsky and Burchiel's study (13) provided proof of the negative effect of VNS during the baseline, activation and postactivation condition in any of six studied individuals in different paradigms. From this point of view, our study is in contrast to published data. In our study, the IED reduction was present during acute stimulation (SP) in 93% of the patients when compared to the BP, and in 80% when compared to the PP. Moreover, the absolute number of IEDs was smaller in all of the measured periods (the SP, IP and the PP) than in the BP in 80% of the patients. The latter result confirmed the sustained effect of VNS on the reduction of IEDs. It has been theorized that VNS works as an antiepileptic by desynchronizing EEG. As mentioned above, the desynchronized effect on animal EEG was obtained using high intensity stimulation with a frequency between 20–50 Hz (9). We used a stimulation frequency within this range, and the intensity of stimulation used in this study was high (1.25 mA average).

It is well known that the use of magnetic extra-stimulation may abolish a seizure after its onset in some patients. This phenomenon has been studied in both animals (7, 9, 10, 11) and humans (12). Hammond et al. (12) clearly demonstrated the abrupt termination of EEG seizure activity in their experiment using a pulse generator manually activated at the onset of aura. Two major mechanisms are considered to explain this effect. First, vagal stimulation suppresses spikes and seizures in the cortex by an effect similar to arousal mediated through the reticular activating system (7). The second mechanism of described suppression could be mediated through a projection from the nucleus of the solitary tract through the parabrachial nuclei to the substantia innominata and zone incerta (8). From this point of view, our study provided very interesting data. The reduction in the absolute number of IEDs during the SP as compared to both the BP and the PP is significantly higher in patients who experienced the positive effects of magnetic extra-stimulation during their daily life. We can conclude that patients who experienced the positive effects of acute stimulation in suppressing the interictal epileptiform graphoelements, are probably good candidates for successful abolishment of ictal events through magnetic extra-stimulation. This fits well with the theory that both interictal and ictal discharges are VNS-influenced by similar mechanisms, as has been suggested previously (7). We found a similar relationship between the IEDs reduction during SP and the effect of long-term intermittent VNS. Patients who were considered to be responsive (>50% reduction of all seizures) have a significantly higher percentage reduction of IEDs in the SP. However, we did not find such a correlation when comparing the percentage reduction in the SP to the PP and the effect of VNS.

Long-term changes of interictal EEG during VNS and its correlation with the outcome showed that in patients with good outcomes the reduction of clustering and synchronization of epileptiform activity at the EEG is higher. (20). A different study demonstrated the progressive EEG changes in the form of clustering epileptic EEG activity and progressively increased periods of spike-free intervals (21). There is no clear relationship between the efficacy of VNS and the location of IEDs. Both temporal and extratemporal epilepsies are effectively influenced by VNS, with no significant differences (22). In agreement with the study by Mañon-Espaillat and Wernicke, we did not find any statistically significant differences between temporal and extratemporal patients in the percentage reduction of IEDs during the SP. Similarly, this percentage reduction was not significantly correlated with the duration of epilepsy, the duration of VNS, or with the value of the stimulation intensity.

Our study supports the fact that acute stimulation of vagal nerves reduces the number of IEDs. In comparison to Hammond et al. (12), we investigated more patients (15 in our study; 9 in Hammond's). The reason for the differences between our results is unknown. He used variable stimulation frequencies in the study with interictal EEG (1, 5, 10 and 50 Hz). We used only the frequencies habitually used in chronic stimulation in our patients (20 and 30 Hz). Moreover, we used the methodology with repeated evaluation of EEG tracings in pre-defined time-periods, which may strengthened the accuracy of our results.

Most recently, Olejniczak et al. (23) reported a case study of a patient treated by VNS who was investigated by invasive EEG prior to a temporal lobectomy. Stimulation at 30 Hz produced a significant decrease in the occurrence of epileptiform sharp waves in the left hippocampus. Although it was a case study, such precise proof of the effect of acute VNS effect on interictal activity recorded directly from the brain structures had not been published previously.

Our study, as in other experimental works (7, 12, 19) demonstrates the acute effect of VNS on the suppression of the interictal and ictal epileptic EEG finding. The evidence of the chronic effect of VNS was demonstrated by the study of Ristanovic et al. (24). They demonstrated persistent seizure control in patients who responded to VNS for 2 weeks during “end-of-service” period. We have to be aware of both the acute and chronic effects of VNS in the evaluating of the effect of VNS on the clinical state and on EEG.

Although our results of IED reduction after a single stimulation of the vagal nerve are highly statistically significant, these preliminary results should be tested on the larger groups of VNS treated patients. These results could be a logical basis for prospective investigation into whether the reduction of IEDs in defined periods could be a predictor of responsiveness to VNS and the positive effects of extra-stimulation on newly-implanted VNS patients.