JLE

Epileptic Disorders

MENU

Deep brain stimulation in epilepsy with particular reference to the subthalamic nucleus Volume 4, supplement 3, Supplement 3, December 2002

Figures

See all figures

About 30% of patients with epilepsy who are treated with antiepileptic drugs continue to have disabling seizures [1]. Resective surgery is considered when such seizures have a focal origin, and when presurgical evaluation clearly demonstrates that the epileptogenic zone, i.e. the volume of brain tissue at the root of the generation of seizures, can be removed without causing additional, unacceptable neurological or cognitive deficits. However, conventional resective surgery is unsuitable in at least one third of medically refractory patients because of the involvement of eloquent cortices, or because of the multifocal, bilateral or generalized expressions of the seizure disorder. Pharmacological advances are unlikely to decrease the number of such drug-resistant epileptic patients in the near future [2], so alternative therapeutic methods are under development.

Since the seventies, different structures of the human brain thought to modulate cortical excitability have been stimulated with the aim of improving severely impaired epileptic patients. Although the cortex plays a crucial role in seizure origin, accumulating evidence has pointed to the role of subcortical structures in the clinical expression, propagation, control, and sometimes initiation of epileptic seizures (see Vercueil and Hirsch; Semah in this issue). It is therefore not surprising that deep brain stimulation is gaining increasing interest in the field of epilepsy [3], inasmuch as chronic high frequency (130 Hz) electrical stimulation (HFS) of the ventralis intermedius thalamic nucleus (VIM), subthalamic nucleus (STN), and globus pallidus internus has been shown to be feasible, safe and effective in different types of movement disorders [4].

The aim of this article is to give a succinct overview of the different targets that have been stimulated in epileptic patients for therapeutic purposes, and to present our initial data on STN stimulation in epileptic patients.

From the cerebellum to the subthalamic nucleus

The cerebellum

Cooper et al. [5] were the first to suggest that chronic electrical neurostimulation of the human brain could modulate cortical excitability in epileptic patients, focusing their attention on the cerebellum, the role of which in epilepsy was suspected for many years. Their clinical and anatomical study published in 1976, showed that seizures were modified or inhibited by cerebellar stimulation in 10 out of their 15 patients, with no evidence of any adverse events [6]. Although encouraging, these results were not confirmed by further controlled studies [7, 8], so that chronic cerebellar stimulation in epilepsy was no longer considered. Although this stimulation was not properly a deep brain stimulation, as it consisted of a subdural stimulation of the cerebellar cortex, it raised the question of a distant control of cortical epileptogenicity by electric stimulation.

The anterior thalamus

Interest was then directed towards the thalamus, a structure closer in both anatomical and physiological aspects to the systems of control of cortical activity. Encouraged by their experience of cerebellar stimulation, Cooper and his group hypothesized that Òthe thalamic relay nucleus of this system might be a more direct sensitive funnel to which to apply neurostimulation, in order to affect limbic structuresÓ [9]. They therefore performed bilateral chronic stimulation of the anterior nucleus of the thalamus (AN) in six epileptic patients. Five showed a 60% reduction in seizure frequency, of whom three showed a decrease in paroxysmal activity. AN is a specific integration nucleus of the limbic system, its inputs coming from the cingular cortex. It also receives related fibres of the mamillo-thalamic bundle; the section of this bundle prevents seizures in epileptic guinea pigs [10]. In addition, electrical stimulation of the mammilary nuclei increased seizure threshold to pentylenetetrazol in rats [11]. More recently, the anticonvulsant effect of AN HFS has been reported in rats [12], and has generated clinical interest. In this respect, Hodaie et al. [13] have reported that bilateral 100 Hz stimulation of the anterior nucleus of the thalamus resulted in a significant decrease in seizure frequency in 5 patients suffering from primary or secondary generalized tonicclonic seizures, after a mean follow-up of 15 months, with no adverse events. Suppression of seizures appeared more pronounced in the two patients who suffered from partial epilepsy. Interestingly, as shown by Velasco et al. for the centromedian-thalamic nucleus stimulation [14], the observed benefits did not differ between stimulation-on and stimulation-off periods. Therefore, although possibly effective, deep brain stimulation of the AN requires further controlled studies.

The centromedian thalamus

At the end of the 80's, emphasis was put on the centromedian- thalamic nucleus (CM), an intralaminar nucleus which is part of the reticulothalamocortical system mediating cerebral cortex excitability [15]. The first CM stimulation study in epilepsy was performed by Velasco et al. [16], based not only on physiological background and a few previous experimental works [17], but also on anatomical grounds. CM is indeed a well defined anatomical structure in the center of the thalamus, and is thus easily localized by traditional stereotaxic techniques. The initial results were not confirmed by the controlled study of Fisher et al. [18], probably because of the difficulty in identifying good responders and in determining the correct stimulation parameters. Further studies conducted by the Velasco group, including long-term follow-up, confirmed that chronic CM stimulation was a potentially effective modulatory procedure for the control of intractable seizures, and especially generalized tonic-clonic seizures and atypical absences of the Lennox-Gastaut syndrome [14, 19]. These results might appear surprising since stimulation was applied at high frequencies (60-130 Hz), which are thought to produce functional inhibition in the basal ganglia and thalamic structures [4], and because lesion of the CM in the aluminum gel model in the monkey has been shown to increase seizure frequency or to counteract the antiepileptic effects of a previous lesion of the ventralis intermedius nucleus (VIM) [20]. This animal model is, however, a model of partial motor seizure, and the underlying mechanisms for the generation and control of seizures are different from those involved in atypical absences and tonic-clonic seizures. In addition, the CM is not a homogeneous structure, so that different effects can be expected, depending on the lesioned or stimulated sites. In this respect, Velasco and colleagues found that stimulation sites located inside the ventro-lateral portion of CM, and from which recruiting and desynchronizing direct current shift responses could be elicited, were predictors of a good outcome [14]. Unexpectedly, persistent antiepileptic effects were found 3 months or more after discontinuation of the stimulation (the Òoff effectÓ), and the authors have suggested that this could be due to functional (and possibly plastic) changes that developed during the stimulation procedure.

The basal ganglia

Since the beginning of the 80's, experimental evidence has accumulated on the Ònigral controlÓ of epileptic seizures [21]. In particular, inhibition of GABAergic neurons of the substantia nigra pars reticulata (SNR) was shown to suppress epileptic seizures in various animal models of epilepsy (See Deransart and Depaulis; Veliskova et al. in this issue). This inhibition of SNR neurons, which leads to disinhibition of superior colliculus cell bodies, may result either from activation of the direct striato-nigral pathway, or from inhibition of the indirect striato-nigral pathway that passes through the globus pallidus and the subthalamic nucleus [22]. A modulation of cortical excitability can thus be expected from pharmacological and electrical manipulations of many components, if not all, of the basal ganglia system (see Deransart and Depaulis in this issue). As a matter of fact, many clinical and neuroimaging data tend to show the involvement of the basal ganglia in the development and perhaps the control of epileptic seizures (see Vercueil and Hirsch; Semah in this issue). Paradoxically, the therapeutic relevance of such experimental, clinical, and neuroimaging findings was rarely considered until the end of the 1990's [4]. Chkhenkeli et al. [23] were the first to describe the beneficial effects of striatal stimulation in the case of pharmaco-resistant temporal lobe epilepsies. Although their study was based on a physiopathological hypothesis related to the balance between proconvulsant and anticonvulsant structures (i.e., the neostriatum), their work [24] showed clear antiepileptic effects during low frequency stimulation of the caudate nucleus in 57 patients. These data are somewhat in agreement with the concept of the nigral control of epilepsies, since activation of the striatum inhibits SNR via its GABAergic projections. Conversely, these authors could worsen the epileptic activity, particularly of hippocampal and amygdalar origin, by stimulating the caudate nucleus at high frequency. Although further studies are needed, these results highlight the ability of the basal ganglia system to modulate cortical epileptogenicity, as suggested by animal studies (see Deransart and Depaulis; Veliskova et al. in this issue).

The sub-thalamic nucleus

Emphasis has been put more recently on the STN, which is critical in the maintenance of SNR cells activity, and because pharmacological inhibition leads to suppression of epileptic seizures in various animal models of epilepsy (see Deransart and Depaulis in this issue). Such an antiepileptic effect has been reproduced using bilateral HFS (130 Hz) of the STN in a genetic model of absence seizures in the rat [25], suggesting that HFS could be used in the treatment of some forms of epileptic seizures. These experimental findings, together with our experience with STN HFS in Parkinsonian patients [4], led us to perform, in 1998, the first STN stimulation in a 5-year-old girl with pharmacologically resistant, inoperable epilepsy caused by focal centroparietal dysplasia [26]. Later, four additional patients also received STN stimulation at our institution.

Here we report the main findings of these five epileptic patients in whom STN HFS was undertaken at the Grenoble University Hospital from October 1998 to August 2001.

Chronic STN stimulation for intractable epilepsy: preliminary results in five patients

Patients

Five patients, from 5 to 38 years old (17.6 ± 12.7), presenting a severe, drug-resistant epilepsy, underwent surgery at our institution in 1998 (n = 1), 1999 (n = 1), 2000 (n = 1) and 2001 (n = 2). Their main clinical findings and epilepsy features are reported in table I. All patients and their family were clearly informed of the inherent risks of this new technique, offered as some hope, given the absence of any other new therapeutic solution. The type of epilepsy was considered to be inoperable in two patients after ictal scalp-EEG recordings, and in two other cases after stereotactic intracerebral EEG recordings (SEEG). The remaining patient (Case 3), had a right pre-central frontal resection after SEEG recordings, and was bilaterally implanted in the STN 22 months later because of the persistence of seizures of central origin.

Methods

The five patients were all implanted using the same technique as that used for STN implantation in parkinsonian patients [27]. A stereotactic frame (Radionics, Burlington, MA, USA) was positioned on the patient's head under general anaesthesia, and four titanium screws were implanted in the skull for later, non-invasive repositioning. Ventricular landmarks (anterior and posterior commissure, superior limit of the thalamus and midline) were obtained after ventriculography by direct injection of 7 Ml of Iopamiron into the frontal horn of the right ventricle. In each case these landmarks allowed determination of the subthalamic targets. Two days later, a magnetic resonance imaging scan was obtained using a stereotactic protocol allowing sagittal, coronal and axial T1 and T2 weighted sequences to be obtained. Four days later, chronic deep brain stimulation electrodes (Medtronic DBS lead 3389, four iridium platinum contacts, 1.27 mm in diameter, 1.5 mm long, spaced by 0.5 mm) were implanted under general anaesthesia. One to three electrodes were inserted in the right and left STN, except for Case 1 in whom only one electrode was inserted in the left STN. The electrodes were aimed at the STN target on the basis of our ventriculography- based coordinates and by the five-channel micro- electrophysiological method used routinely for electrode implantation in patients with movement disorders [28]. Additionally, a limited number of EEG needle electrodes were placed over the scalp bilaterally in all but 1 (Case 4) patients. At the end of the surgery, an X-ray teleradiography of the skull was performed that allowed us to localise the final position of each electrode, which was plotted into the Òproportional Guiot's schemeÓ (this schema is designed from the anterior and posterior commissure, the bicommissural line and the thalamic height). After implantation, the electrode leads were temporarily connected for a few days to a transcutaneous extension, to allow scalp-EEG and STN macrorecordings. These were acquired at a sampling rate of 256 Hz, using the audiovideo- EEG monitoring system employed at our institution for long term monitoring in epileptic patients (Micromed, Treviso, Italy).

After the test period (several days), scalp EEG electrodes were removed and the most relevant STN electrodes were connected to a dual channel stimulator (Kinetraª Medtronic, Inc., Minneapolis, MN, USA) placed subcutaneously in the right subclavicular area [29]. For patients who had three electrodes to connect, a Kinetra, plus an Itrell 2 generator (Itrell 2 Medtronic, Inc., Minneapolis, MN, USA) were used.

The settings of electrical parameters was performed after surgery and at each follow-up visit. The effect of different parameters was first studied on each contact, which was selected as the cathode, with the pulse generator case as the anode (monopolar stimulation), and permitted definition of the threshold and type of side effects. This allowed us, together with the stereotactic localisation method, to select the best electrode contacts to be stimulated. As shown in figure 1, the stimulated contacts were located in the subthalamic nucleus area in all patients. The average coordinates were: antero-posterior: 4.1 ± 0.7 1/12 o of intercommissural distance; verticality: - 2.5 ± 0.9 1/8 o of thalamic height; laterality: 10.2 ± 1.3 mm.

For chronic stimulations, we used a constant current electrical generator with monopolar pulses (intracerebral electrode as the cathode, the anode being an indifferent large electrode on the skin), a pulse width set at 90 µs or 60 µs, and a frequency set at 130 Hz. The intensity was gradually increased in order to avoid side effects such as dyskinesias, motor contractions, eye deviations and paresthesias.

Data analysis

The number and the type of seizures were noted daily by the patient or one of the family members. Data are represented as mean ± standard deviation. For the statistical analysis, the monthly average number of seizures which occurred preoperatively was compared to the monthly average number of seizures which occurred post-operatively by a non-parametric test (Wilcoxon test). Statistical significance level was set at P < 0.05.

Results

Effects of STN stimulation on seizure control

STN stimulation significantly reduced the occurrence of epileptic seizures in four patients (mean reduction of seizure frequency of 64.2%, range: 41.5% to 80.7%) (table II). Pharmacological treatment was kept unchanged in 3 patients, and was reduced in two others (Cases 1 and 2).

Good responders

Three patients clearly benefited from STN stimulation (Cases 1, 3, 4). They were all suffering from partial symptomatic seizures, which arose from the sensori-motor cortex (see table I). Stimulation was applied bilaterally in two cases, and unilaterally in 1. In one patient (Case 1), stimulation was more effective on seizures occurring in clusters and during the day than on those that occurred during sleep [26]. In this patient, vigabatrin was gradually stopped after the first 18 post-operative months to improve the child's alertness and attention at school. In another patient (Case 4), the improvement was particularly noticeable on seizures associated with falls, which ranged from 74.3 (± 24.8) preoperatively to 5.1 (± 11.03) after surgery (- 93.2%, P ^0.01). Consistent improvement in motor functions was noted in Cases 1 and 3 as assessed by the reduced post-ictal state. In these two patients, stimulation was applied intermittently during the last months of the follow-up period, and a further improvement in seizure frequency was observed in Case 1 [26], but not in Case 3 (figure 2). Interestingly, this latter case demonstrated a better improvement with bilateral stimulation than with unilateral stimulation, showing also, as previously reported [26], that with adjustment of the voltage, a better control of seizures could be obtained (figure 2).

Moderate responder

Patient 2 responded to continuous bilateral STN stimulation, although the percentage of improvement did not exceed 50%. This patient suffered from a severe myoclonic epilepsy (Dravet syndrome), characterised by seizures, as recorded on the scalp EEG, which bilaterally involved the anterior territories of the two hemispheres, including the central regions (see figure 3). Interestingly, plasma level of valproic acid gradually increased during the stimulation procedure, so that the drug could be reduced from 2 500 mg to 1 750 mg per day, with the same efficacy.

Poor responder

Patient 5 suffered from an autosomal dominant nocturnal frontal lobe epilepsy, whose seizures were proven to arise from the left insulo-opercular cortex after SEEG recordings. This patient, who experienced daily nocturnal seizures and presented several episodes of status epilepticus per month, did not show any improvement in seizure patterns after 6 months of continuous bilateral stimulation. After a 2 month stimulation period during which the deepest contacts of the electrodes could not be stimulated over 1.6 V, due to a low threshold for side effects, stimulation was applied at a higher intensity on the electrodes contacts located in the STN, without additional benefit.

Macroelectrophysiological recordings

In all patients, interictal spikes were recorded from different contacts of the STN electrodes. Most often they seemed to be concomitant with interictal spikes recorded from the scalp-EEG. However, some were recorded without apparent expression on the scalp, and vice versa. Whether STN spikes preceded or followed those recorded from the scalp EEG could not be specifically assessed. Seizures were also recorded in the 4 patients with simultaneous STN-scalp EEG recording. In all cases, a similar ictal pattern was recorded concomitantly at the STN and scalp-EEG electrodes (figure 5 and 4a). This pattern even appeared more prominently at the STN electrode in one patient (Case 1) during a long-lasting episode of epilepsia partialis continua (figure 4b).

Adverse events

Chronic STN stimulation was generally well tolerated in all five patients. However, surgical complications occurred in two patients, including infection of one of the generators in patient 3 (see legend of figure 2), and a postimplantation subdural haematoma in patient 2 who later underwent surgical treatment, without sequaelae.

Discussion

Our preliminary results, obtained in a small number of patients, strongly suggest that the electrical stimulation of the basal ganglia system may modulate seizure propensity in severely impaired, epileptic patients who are not suitable for conventional resective surgery.

Four of our five patients showed a clear reduction of seizure frequency during STN HFS, among which three showed a 67.1% to 80.7% reduction, that subsequently led to improvement in motor functions in two of them. These results are in agreement with those of two other studies: two of the five patients who underwent stimulation at the Cleveland Clinic Foundation had a reduction in seizure frequency of 60 and 80% after 16 and 10 month of follow- up, respectively [30]. The other three patients, including one with focal status epilepticus, did not show any improvement. An additional patient presenting Lennox- Gastaut syndrome and treated at the American University of Beirut, showed a complete suppression of generalized tonic-clonic seizures and reduction of ÒmyoclonicÓ seizures and atypical absences of over 75% after a 1 year follow- up [31].

These heterogeneous results should be interpreted in light of the widely differing seizure types. Nevertheless, they suggest that STN HFS can be effective not only in medically intractable partial epilepsies, but also in some of the most refractory generalized epileptic syndromes such as Dravet syndrome and Lennox-Gastaut syndrome. If we assume that STN HFS produces a Òfunctional inhibitionÓ, this wide range of potential candidates for STN stimulation is in agreement with experimental data, showing that pharmacological inhibition of the STN suppressed seizures in models of absence seizures [32], generalized tonic- clonic seizures [33, 34], and partial seizures with secondary generalization [34]. On the other hand, the effectiveness of STN HFS could depend on the type of seizure, as suggested by the observation that each of our three good responders to stimulation had an epileptogenic zone involving the central area, and as suggested by experimental data showing that tonic seizures, which involve brainstem circuits, are not suppressed by inhibition of the SNR (see Deransart and Depaulis in this issue). The characterisation of subgroups of patients that are more likely to benefit from this treatment will certainly remain an important issue in future studies.

Another important issue to address concerns the optimal stimulation parameters. It appears likely that bilateral stimulation is more effective, as demonstrated in animal experiments [25], and as suggested by one of our patients (see figure 2). In addition, experimental evidence suggests that an intermittent stimulation protocol is likely to be more effective than a continuous one, since the effect of stimulation tend to disappear with time [25]. However when the stimulation was switched from continuous to intermittent in the present study, a benefit was observed in only one patient, so that the question of discontinuous stimulation, whether cyclic or even random, remains opened. That these two patients did not experience seizure rebound when stimulation was turned off suggests that such discontinuous stimulation protocols can be safely conducted. We chose: a 22 hour on, 2 hour off ratio because it was felt to be safer in terms of the risk of rebound seizures. However, whether this regimen is measurably different from continuous stimulation is not known. One way to solve this problem would be to interrupt the seizures as soon as they occur, or even a few minutes before. In this respect, the fact that seizures can be recorded from the STN electrodes contacts, as shown in our patients and in those of Dinner et al. [35], opens the possibility of developing a closed stimulation system in which seizures are detected and stopped through the same electrode. This may have a significant advantage over vagal nerve stimulation, providing that seizure onset can be reliably anticipated, and that stimulators are capable of on-line EEG analysis. Recent studies have shown that non-linear analysis of EEG signals allows the prediction of seizures several minutes before they occur [36].

We have suggested [26] that the mechanisms of action of STN HFS in epileptic patients could be mediated through the Òfunctional inhibitionÓ of the STN, and subsequently of the SNR. The recording of epileptiform activity in the STN suggests that this structure, as well as the thalamic nuclei, is part of a cortico-subcortical network involved in the epileptogenic process. As an alternative, the effect could be mediated by the antidromic stimulation of the cortex (and thus of the epileptogenic zone), as suggested by Dinner and colleagues [34]. The fact that our best responders were suffering from a relatively well-circumscribed epileptogenic zone located in the central region of the cortex is in favour of this hypothesis. However, this hypothesis cannot by itself explain the anti-epileptic effects of STN HFS, and in particular its efficacy in generalized forms of epilepsy. Other mechanisms of action could also be hypothesised, such as possible changes in the resistance or sensitivity to antiepileptic drugs, as suggested in one of our patients.

Whether the optimal target in epileptic patients is the STN itself or, as suggested by experimental studies, the SNR, remains an important issue. The most effective targets in our patients were located in the inferior part of the STN, close to the SNR, in a region that is different from our STN target in parkinsonian patients, which has been shown to be located more laterally and more dorsally (see figure 1). This suggests that the circuits involved in the control of seizures would be different from those proposed for the control of movement disorders. This hypothesis requires the identification of electrophysiological markers in epileptic patients, as has been done for CM stimulation [14]. In contrast to the treatment of patients with movement disorders, specific difficulties are encountered in epileptic patients when targeting deep brain structures, particularly because of the absence of acute symptomatic improvement that can be monitored during surgery.

CONCLUSION

Our preliminary data, as well as recent data obtained in other groups, suggest that deep brain stimulation should be considered in the treatment of drug-resistant and inoperable epilepsies. In particular, CM, AN, and STN stimulations have been proven to be feasible, safe and possibly effective in patients suffering from different kinds of epilepsies. The fact that responders experienced a seizure reduction, but not a complete remission of seizures, means that surgical resection remains the gold standard treatment of drug-resistant epilepsies whenever this option is possible. However, in a reasonable number of refractory patients, such an option is not feasible, so that deep brain stimulation may offer a potential alternative. Nevertheless, although encouraging, published results do not provide a definite conclusion. Many issues remain unresolved, including patient selection criteria, optimal stimulation targets and stimulation parameters, continuous versus intermittent stimulation protocols and, above all, the mechanisms which subserve the antiepileptic effect of the stimulation. In this respect, additional animal studies are required, and controlled clinical trials will be necessary to further document and validate this new therapeutic method.