Deep Brain Stimulation in Epilepsy

12 Deep Brain Stimulation in Epilepsy


Alexander Ksendzovsky, Kareem A. Zaghloul


Abstract


Deep brain stimulation (DBS) has become a promising new treatment for epilepsy. Over the years, several targets have been investigated for stimulation through animal models and clinical trials. These targets include the cerebellum, thalamus, basal ganglia, and hippocampus. However, it is still unclear which is the optimal location for each seizure type, what the optimal stimulation parameters are, or what the underlying mechanisms are. In this chapter, we review each putative target for DBS stimulation.


Keywords: deep brain stimulation, epilepsy, cerebellum, anterior nucleus thalamus, centromedian nucleus thalamus, basal ganglia, hippocampus


12.1 Introduction


Epilepsy impacts approximately 70 million patients worldwide.1 Thirty percent of patients continue to have seizures despite medical therapy. In this group, continued seizures and polypharmacy have been associated with poor quality of life.2 In 2001, Wiebe et al established surgery as a viable treatment option for medically refractory epilepsy.3 Since then, many patients with uncontrolled seizures have undergone surgery with good outcomes. Unfortunately, approximately 75% of patients who have persistent seizures on anti-epileptic medications are not candidates for resective surgery.4,5 This group includes patients with seizures that arise from eloquent cortex, patients with multiple seizure foci, or patients with generalized epilepsy. For these patients, in addition to the ongoing medical management, other treatment options include vagus nerve stimulation (VNS) and responsive neurostimulation (RNS). However, these options have limited efficacy as only 8% of patients achieve freedom from seizures with multiple medications,3 7% with VNS,6 and 20% with RNS.7 Thus, novel treatment strategies for medically refractory epilepsy are needed.


Recently, several authors have investigated deep brain stimulation (DBS) as a potential surgical option for patients with medically refractory epilepsy. DBS has gained widespread acceptance as a safe and effective treatment option for movement disorders.8 However, it’s efficacy in treating medically refractory epilepsy is yet to be determined. This is partly because, unlike in Parkinson’s disease, the neural circuitry in epilepsy is not well defined. Moreover, at present, there is no current consensus on questions such as why stimulation works, which location to stimulate, which stimulation paradigms to use, or which seizure types respond to stimulation.


Because of its promise in modulating neural circuits, there have been a number of studies researching the use of DBS in several locations. In this chapter, we review the major DBS studies that describe the relevant circuitry, and preclinical and clinical trials related to epilepsy. Surgical targets include the cerebellum, thalamus, basal ganglia, and hippocampus.


12.2 Cerebellum


The cerebellum was the first target for DBS in epilepsy in humans. Originally, this target was chosen because of the general inhibitory nature of Purkinje cells that send projections to deep cerebellar nuclei.7 It was hypothesized that inhibition of Purkinje cells through stimulation would potentiate the inhibitory effect that the deep cerebellar nuclei have on the thalamus, thereby causing a decrease in the predominantly excitatory outputs from the thalamus to the cortex9,10 Cerebellar stimulation has been divided into two potential targets: the cerebellar cortex and the deep cerebellar nuclei.


12.2.1 Cerebellar Cortex Stimulation


Previous studies have revealed phase-locked oscillations in the cerebellar cortex, deep cerebellar nuclei, and thalamus11,12,13 during seizures. Along with the known inhibitory function of the Purkinje cells, this has made the cerebellar cortex an attractive target for DBS experiments.14 Cerebellar DBS was first investigated in early animal studies that began in the middle of the 20th century. In 1955, Cooke et al showed that cerebellar cortical stimulation decreased seizures in kindled cats.15 In 1976, Hablitz et al published their results using vermian cortical stimulation in cats to treat generalized epilepsy. In these experiments, both high- and low-frequency cortical stimulation led to a reduction in the number and amplitude of general cortical discharges.16


However, later animal experiments proved to be inconsistent. In 1980, Ebner et al used an aluminum gel primate model to characterize the impact of cerebellar stimulation on the activity of neurons within the seizure focus. They failed to find any statistically significant changes.17 Similarly, Hablitz and Myers et al failed to show any effect on penicillin-induced animal models.16,18


Similar to animal trials, human trials for cerebellar cortex stimulation also failed to show consistency. In the early 1970s, Irving Cooper and his colleagues implanted cortical cerebellar electrodes in 32 patients with varying seizure etiologies. They showed more than 50% seizure reduction in 56.2% of patients that sustained for an average of 18 months.19,20,21 Meanwhile, one year later, another group reported on cerebellar cortical stimulation in six patients. Five out of the six patients had decreased seizure frequency after stimulation.22 In a separate study, however, stimulating patients with generalized seizures showed significant reduction in only two of six patients.23 Krauss et al summarized the results of the subsequent two decades worth of open-label human trials. They showed that out of the 36 patients enrolled in these various trials, 91.6% has some seizure reduction but only 12 patients achieved seizure freedom.24


Despite this fluctuating data in animal and open-label human trials, several double-blind trials were attempted. Velasco et al studied five patients with heterogeneous seizure semiology. Cortical cerebellar electrodes were placed and were “ON” in three and “OFF” in two patients. In this study, there was only a 33% decrease in seizure activity in patients receiving stimulation compared to those not receiving stimulation.25 Three smaller, blinded studies also showed limited success. Of the total 14 patients receiving stimulation only two had any benefit.26,27,28


The past 40 years of cerebellar cortical stimulation question this technique as a means of treating epilepsy. It seems that mechanistically, the initial hypothesis that DBS stimulation of Purkinje cells would reduce the inhibitory drive on downstream deep cerebellar nuclei that would then effectively reduce seizure activity may be more complicated than anticipated. Indeed, several studies performed on sampled cerebellar tissue in epilepsy patients have demonstrated Purkinje cell degeneration. Such degeneration would suggest that the overall decrease in Purkinje cell volume may confound any effects of direct cerebellar stimulation on downstream nuclei.24


12.2.2 Deep Cerebellar Nuclei


Few studies have been performed examining the efficacy of deep cerebellar nucleus stimulation, and as with cortical stimulation, there are conflicting results. The deep cerebellar nuclei are directly connected to the thalamus, and are thus better positioned to modulate thalamic outflow. The deep cerebellar nuclei are divided into three functional and anatomical groups that project to distinct nuclei within the thalamus, which in turn project to distinct cortical regions (image Fig. 12.1). The three cerebellar nuclei are the lateral nuclei (dentate), interposed nuclei (globose and emboliform), and fastigial nuclei. The lateral nuclei project preferentially to the parafascicular (Pf) and ventral lateral thalamic (VL) nuclei. The interposed nuclei project primarily to the posterior thalamic nuclear complex (Po) along with VL. The fastigial nuclei send projections on to ventral medial thalamus (VM) and Pf. These projections are even more complicated as each set of deep cerebellar nuclei has some, though not preferential, projections to all the major thalamic nuclei along with the connections mentioned above.24 The thalamus in turn has broad projections to several areas of the cortex. VL projects primarily to the primary motor and sensory cortex while the Pf and Po target wide cortical areas including prefrontal cortex, primary sensory and motor cortices, cingulate gyrus, temporal lobe, frontal cortex, and the amygdala. Lastly, VM also projects widely to primary sensory and motor cortices, cingulate gyrus, temporal lobe, frontal cortex, and the amygdala. Despite these direct connections, however, the complexity of this network and redundant pathways likely account for the inconsistent results observed with deep cerebellar stimulation that is described below.


In the earliest animal study that used deep cerebellar stimulation, Dow et al demonstrated inhibition of cortical bursting activity in a focal cortical cobalt rat model of epilepsy.29 In 1972, Hutton et al directly compared cortical and deep cerebellar stimulation in a cat penicillin focal epilepsy model. They showed seizure reduction with both cortical and deep nuclear stimulation.30 Subsequently, Babb et al showed a decrease in seizure frequency and seizure duration in a cobalt model of hippocampal epilepsy31 with deep nuclear stimulation. Finally, a recent study in 2004 showed equivocal results with stimulation of the superior cerebellar peduncle (SCP) in amygdala-kindled rats. The investigators found that SCP stimulation potentiated limbic seizure initiation but decreased secondary generalization.32


The literature on deep cerebellar stimulation in humans is scarce, describes heterogenous seizure types, and is slightly conflicting as well. In 1976, Sramka et al reported on four patients with focal, generalized motor and myoclonic seizures that underwent dentate nucleus stimulation at 10 and 100 Hz. These four patients had a moderate improvement in seizure frequency, but the improvement was noted to be only temporary.33



12.3 The Thalamus


The thalamus has widespread cortical connections. It is the relay site for all sensory information except olfaction which has direct inputs to the cortex.7 The thalamus also modulates information from the cerebellum, basal ganglia, and limbic systems. Every thalamic nucleus except the reticular nucleus sends reciprocal cortical projections.26 For this reason, the thalamus has become a target for stimulation in epilepsy. Clinical trials have focused on the centromedian nucleus of the thalamus (CMT) and the anterior nucleus of the thalamus (ANT), and animal studies have focused on their mammalian equivalent, as detailed below.


12.3.1 Centromedian Nucleus


The CMT is part of the nonspecific thalamic system which consists of the intralaminar, paralaminar, and midline nuclei.34,35,36 The CMT is the largest of these nuclei and is located at the level of the posterior commissure.37 The nonspecific thalamic nuclei receive inputs from the reticular formation and are thought to play an important role in arousal.37 They have diffuse projections to other thalamic nuclei, the basal ganglia, and cerebral cortex (image Fig. 12.2). The CMT has been implicated in epilepsy from as early as 195138 and its larger size makes it amenable to surgical targeting.37


There is no direct animal corollary to CMT. However, the murine thalamic reticular nucleus (TRN)39 has similar projections to and from the reticular system and has been used as a surrogate for the CMT. The TRN houses mainly gammaaminobutyric acid (GABA)ergic neurons that are thought to be a relay between the corticothalamic and thalamocortical projections.40,41 Cortical and thalamic neurons send glutamatergic axons to the TRN while the TRN send GABAergic projections to other thalamic nuclei.40,42,43


Most animal and human trials have investigated the effect of CMT stimulation on multifocal, generalized focal motor and non-motor seizures. Pantoja-Jimenez et al investigated TRN stimulation in a pentylenetetrazol (PTZ) rat model of generalized epilepsy. High-frequency stimulation prolonged latency to tonic– clonic seizures and status epilepticus. Although the mechanism is not completely clear, Jiminez et al showed that modification of seizure-induced corticothalamic synchrony may play a role in the observed antiepileptic effects of TRN stimulation.39


In human subjects, much of the clinical evidence supporting CMT stimulation for epilepsy was pioneered by Velasco and colleagues. In 1987, Velasco et al reported on the first five patients with CMT stimulation for either generalized or multifocal refractory seizures. They showed 80% reduction in generalized tonic–clonic seizures and 60% reduction in generalized nonmotor seizures. One patient was seizure free and three patients were able to reduce medication.44 In an attempt to replicate these results, a separate group led by Fisher et al in 1992 performed a double-blind crossover trial of electrode implantation in CMT in seven patients. Unlike their predecessors, they did not show significant treatment differences.45 However, in an open-label follow-up study by the same group, half of the patients had 50% reduction in seizure frequency.45


In a larger study, Velasco et al reported on 15 patients who underwent CMT stimulation and were followed for 41.2 months. All patients had long-standing intractable seizures and were not good candidates for resective surgery. They divided the patient cohort into two groups: Lennox–Gastaut (LG) syndrome group and focal seizures with secondary generalization group. Patients in the LG group had 81.6% reduction in seizure frequency while patients in the second group had an overall reduction of 57.3% in seizure frequency.46 In a follow-up study specifically probing CMT stimulation for LG, Velasco et al again demonstrated an 80% overall seizure reduction rate at 18 months in 13 LG patients.47 Further analysis revealed that incorrect lead placement was associated with worsening seizure control. Patients with adequate lead placement enjoyed a seizure reduction of more than 87%, further validating CMT as a target for LG.47



Since these trials, there have been several attempts to replicate these results in patients with generalized epilepsy. In 2013, Valentin et al reported on 11 patients treated with CMT DBS for primary generalized and frontal lobe epilepsy refractory to medication or surgical resection. The trial was designed with 3 months of sham treatment, 3 months of stimulation, and 6 months of unblinded stimulation. Overall, all six patients with generalized seizures had more than 50% seizure reduction while blinded and five out of six patients had more than 50% reduction thereafter. In the frontal lobe epilepsy group, only one patient had more than 50% reduction during blinding and three had a therapeutic response after the blinding period was over.48


Taken together, these studies suggest that CMT DBS may be beneficial for patients with LG syndrome. Furthermore, investigation on stimulation of the CMT for generalized seizures involving the reticular system or thalamus may be more specifically considered for future trials.


12.3.2 Anterior Nucleus of the Thalamus


The ANT is located at a central point within the circuit of Papez.49 Papez circuit relays information from the hippocampus and subiculum to the mammillary bodies via the fornix. The mammillary bodies project to the ANT via the mammillothalamic tract. The ANT then projects to the cingulate gyrus which further sends projections to the parahippocampal gyrus and to the entorhinal cortex which finally project back to the hippocampus through the perforant pathway (image Fig. 12.2).50 The cingulate gyrus also sends projections to various higher cortical structures.7


Given that seizures frequently originate in the mesial temporal structures, and given the widespread connections between the cingulate and cortical regions, Papez circuit has been implicated in seizure propagation throughout the rest of the brain.50,51 Abnormalities such as magnetic resonance signal change and sclerosis were found within the components of Papez circuit in patients51 and in animal models of epilepsy.52 For these reasons, the ANT has become a promising target for DBS and has been studied in generalized, focal, and temporal lobe epilepsies.


The impact of ANT DBS on generalized epilepsy was evaluated in the pilocarpine and PTZ rat models of epilepsy. This data, however, was tested with inconsistent stimulation parameters, which may account for some of the discordant results observed later in human trials. Hamani et al used the pilocarpine rat model to compare anterior thalamotomy to high-frequency ANT stimulation. Rats in the stimulation group still developed status epilepticus but its latency was significantly prolonged. Interestingly, the thalamotomy group never developed seizures.53 In a follow-up study, in order to characterize necessary stimulation parameters, the same group showed that stimulation current, not frequency, was related to seizure latency change.54 Furthermore, Mirski et al showed that high-frequency stimulation (100 Hz) was necessary to raise the seizure threshold in PTZ rats, while low-frequency stimulation (8 Hz) actually lowered it.55 This lack of consistency in the literature regarding stimulation parameters for generalized epilepsy necessitated its exploration even further. Conovolan et al explored ANT stimulation parameters in chronically seizing pilocarpine rats. High-frequency (130 Hz) stimulation at 100 μA reduced seizures by 52% while higher current (500 μA) at the same frequency increased seizure activity 5.1 times compared to sham.56 This study further suggested that stimulation current was responsible for ANT-mediated seizure control in animal models.


Similarly, human trials for generalized epilepsy were performed with varying stimulation parameters that resulted in inconsistent outcomes. In 2002, Hodaie et al showed a seizure reduction of 54% after bilateral ANT stimulation with a followup of 14.9 months. In these patients, periods of ON and OFF stimulation (up to 2 months) did not change the seizure frequency, raising the possibility that the overall reductions in seizure activity may be related simply to the placement of the leads. Indeed, seizure reduction was seen prior to stimulation in most patients.57 Several years later, the same group reported long-term outcomes (average 5-year follow-up) in this patient cohort. After long-term stimulation, five patients had more than 50% seizure reduction.58 The following year, in 2007, Lim et al showed seizure reduction in four heterogeneous patients who also had a lesional effect. Unfortunately, they could not discern whether stimulation of the lesions were responsible for seizure reduction.59


Few studies have explored ANT stimulation in temporal lobe epilepsy (TLE). Zhong et al used an amygdala-kindled rat model to evaluate ANT stimulation. Bilateral low-frequency stimulation reduced the incidence of seizures and seizure severity.60 Given the uncertainties regarding stimulation frequency in previous trials, Stypulkowski et al evaluated ANT stimulation parameters in a penicillin sheep model of TLE. Only stimulation above 80 Hz reduced seizure activity and the activity returned after stimulation was turned off.61 This suggested that high-frequency stimulation was required for seizure control. Only one study assessed ANT stimulation in patients with TLE. Osorio et al showed an impressive 75.6% seizure reduction in TLE over the course of 36 months of ANT stimulation. This seizure reduction was associated with improved quality of life.62


Most of the successes observed with ANT stimulation were found in patients with focal epilepsies with and without generalization. The murine kainic acid model was used to explore the efficacy of ANT stimulation in focal cortical epilepsy. In a study by Takebayashi et al in 2007, unilateral high-frequency ANT stimulation significantly reduced seizure frequency and bilateral stimulation completely eliminated seizures.63 Unfortunately, these results were not replicated by Lado et al in chronic epileptic rats (after kainic acid-induced status epilepticus) where the results showed an increase in seizure frequency after high-frequency ANT stimulation.64 Despite this, human trials for focal epilepsies showed promising results. The first clinical trial for DBS in ANT was performed in 1987 in patients with focal cortical seizures by Upton et al. In this study, four out of six patients showed a statistically significant reduction in seizures after DBS lead placement.65 Furthermore, a separate group reported data in five patients with focal seizures with and without generalization. After 6 to 36 months of monitoring, four out of five patients had reduced seizure severity and reduced generalization, while only one patient had reduced seizure frequency.66


These promising results led to the first multicenter, randomized controlled trial for DBS in epilepsy. The Stimulation of the Anterior Nucleus of the Thalamus for Epilepsy (SANTE) trial was set in several institutions and enrolled 110 patients with medically refractory epilepsy (focal cortical seizures with generalization). Patients received bilateral ANT electrode placement and were randomized either to stimulation or no stimulation. The same stimulation parameters were used at all institutions (5 V, 90 μs pulses, and 145 pulses/second). The overall median seizure frequency at the end of the 3-month blinded period decreased by 14.5% in the control group and 40.4% in the treated group. After the blinding period ended, there was an overall seizure reduction of 56% over 2 years and 54% of patients had at least 50% decline in seizure frequency. Fourteen patients were seizure free at 6 months. Interestingly, patients with temporal lobe onset had better seizure reduction than patients with onsets in the parietal or frontal lobes. There was no significant change in mortality and the most common morbidity was surgical site infection (9.1%).67 Recently, a 5-year follow-up on the SANTE patients was reported. At 5 years, the mean seizure reduction increased from 43% at 1 year to 68% and 16% of the patients were seizure free. A significant improvement in quality of life was also seen at 5 years, when compared to 1 year after stimulation.68


After the initial optimistic 2-year results, regulatory bodies in Europe and Canada approved the use of ANT DBS for epilepsy. In the United States, however, the Food and Drug Administration (FDA) was more hesitant and required more compelling efficacy data. In addition, the FDA raised concerns related to individual participants in the SANTE trial who experienced significant increases in seizure activity and the relatively high infection rate.69 Given the long-term 5-year follow-up data, resubmission to the FDA for approval is currently being considered.70 The SANTE trial was a step in a positive direction for the use of DBS in epilepsy. As with SANTE, future trials should remain blinded, include a large cohort of patients, and maintain the same stimulation parameters in order to get a more accurate assessment of the efficacy of DBS.


12.4 Basal Ganglia


The caudate and subthalamic nucleus (STN) have recently emerged as targets for stimulation in epilepsy. The role of the basal ganglia, and in particular these two nuclei, in epilepsy stemmed from several decades of investigations exploring caudate and STN connectivity. Overall, these studies showed connections between the basal ganglia and superior colliculus that are thought to regulate cortical activity and epileptic discharges.71 These pathways and their modulation of cortical activity is known as the nigral control theory.71 The first two studies investigating this mechanism examined basal ganglia projections and provided context for how the basal ganglia could modulate seizures. Gale and Iadorola showed that the GABAergic projections to the substantia nigra originate from the striatum.72 In the context of epilepsy, these nigral projections are believed to play a role in GABA-mediated anticonvulsant activity.72,73 Subsequent studies found that the substantia nigra pars reticulata (SNpr) plays a central role in nigral control and maintains inhibitory control of a group of neurons described as the dorsal midbrain anticonvulsive zone (DMAZ) which is adjacent to the superior colliculus. These neurons have widespread projections to the cortex and are thus thought to modulate cortical activity.71 SNpr tonically inhibits the DMAZ via GABAergic inhibitory projections. When the SNpr is deactivated there is increased activity within DMAZ which leads to inhibition of epileptogenic cortical areas.72,74,75,76,77,78,79,80,81 Thus, by taking advantage of this pathway, one possibility would be that modulating DMAZ inhibition through the SNpr by stimulating the caudate within the dorsal striatum would lead to seizure control (image Fig. 12.3).


The SNpr also receives tonic excitatory input from the STN82,83,84 and phasic inhibitory input from the globus pallidus pars externa (GPe).83,85 Thus theoretically, inhibition of STN and phasic activation of GPe could activate nigral control of epilepsy through the DMAZ. Since this original discovery, several lesioning and activation studies of STN, striatum, GPe, and SNpr have directly shown that modulation of the DMAZ suppresses epileptic activity.71


Mar 23, 2020 | Posted by in NEUROLOGY | Comments Off on Deep Brain Stimulation in Epilepsy

Full access? Get Clinical Tree

Get Clinical Tree app for offline access