Role of Neuromodulation for Treatment of Drug-Resistant Epilepsy
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.302476
Source of Support: None, Conflict of Interest: None
Keywords: Deep brain stimulation, drug-refractory epilepsy, neuromodulation, outcome, responsive neurostimulation, vagal nerve stimulation
Drug-resistant epilepsy (DRE), defined as the “failure of adequate trials of two tolerated, appropriately chosen and used antiepileptic drug schedules (whether as monotherapy or in combination) to achieve sustained seizure freedom”,,, is a chronic debilitating condition accounting for nearly 30% of epilepsy cases. It's also clear now that DRE may be diagnosed quite early in the course of illness especially when there is a failure of the first anti-epileptic drug (AED).,, Currently the only effective treatment for DRE is surgery in both adult and paediatric population.,,, Class I evidence in the form of RCTs is available (for both adult and paediatric population) to support surgical over medical management.,,,
It is very important that neuromodulation should only be considered in those patients where a resection or disconnection surgery is not possible. These may include (1.) temporal resection surgery (e.g. temporal lobectomy with amygdalo hippocampectomy, (2.) extra temporal resections, lesionectomies) or a (3.) disconnective surgery (hemispheric surgery, temporal parietal occipital (TPO) disconnection, parietal occipital (PO) disconnection. The possible exception may be corpus callosotomy where a vagal nerve stimulation (VNS) may be considered as an option. In our centre, we prefer performing an endoscopic guided corpus callosotomy with anterior, hippocampal and posterior commissurotomy., Likewise, for hemispheric disconnections, we have introduced the procedure of endoscopic hemispherotomy.,,,
It is thus important that patients with DRE are evaluated properly at a centre routinely dealing with epilepsy surgery so that the above possibilities are excluded. The concept of neuromodulation is best understood if this is treated like “an additional drug” in a patient with DRE where all options of resection or disconnection surgery are exhausted. Thus, providing neuromodulation provides a “booster” option to control epilepsy. Likewise, a complete work up is required specially performing an epilepsy protocol MRI and video EEG (recording at least 3 habitual ictal events). The evaluation of MRI must be done by a radiologist experienced in dealing with epilepsy substrates as subtle lesions like a bottom of sulcus dysplasia may be easily missed out. Likewise, a video EEG is necessary to rule out non-epileptic events as these may occur even in persons with DRE (in our series up to 25%). Thus, the whole workup is not just to rule out an operable substrate for epilepsy but also to rule out PNES (psychogenic non-epileptic seizures) and other causes of DRE. The latter may include drug non-compliance, seizures precipitated by events like lack of sleep, fasting etc.,,,,,,,,,,,,,
Temporal lobe epilepsy and hemispherotomy have been seen to have the best long-term seizure-free outcomes (60-70%), with outcome after extratemporal epilepsy observed to be around 30-50%., The dilemma lies in the management of cases in whom seizures appear to be originating from multiple sources, or the EZ is localized to an eloquent cortex which cannot be sacrificed, or patients in whom multiple complex networks appear to be the cause of DRE. About 40-50% of patients fall under this category. These patients may be offered palliative surgery, in the form of neuromodulation, in an effort to decrease the seizure burden.
The term “neuromodulation” is essentially electrical stimulation of the nervous system to modulate or modify a specific function (as in movement disorders, pain, epilepsy). It can be delivered in different ways: through stimulation over the skin surface, peripheral nerve stimulation, cortical stimulation, or deep brain stimulation. The origin of the idea of using electrical current to stimulate the brain dates back more than a century to Sir Victor Horsley utilizing it to map neural functions to aid resection surgery in a patient with focal epilepsy. The field of cortical stimulation was further expanded by Penfield and Jasper at Montreal. Electrical stimulation for the treatment of human epilepsy was first described in the 1970s with the targeting of the cerebellum. Since then, the field has grown exponentially with rapid technological advances. Breakthroughs in neurophysiology, enhanced imaging modalities, robotics and computer devices have allowed rapid progress in this field. This is reflected in the explosion of research on the topic in the past two decades., Epilepsy has recently been postulated to be due to a maldeveloped network in which the normal neurophysiological parameters of the brain get abnormally re-organized. This recognition of has led to efforts to modulate the network to the patients’ benefit.
At present, there are multiple neuromodulator devices available in the physician's armamentarium. They may be classified in multiple forms. [Figure 1] They can be implanted inside the body or used to reduce cerebral excitability non-invasively. Based on the mechanism, they may be an open loop (stimulation only, no feedback) or a closed loop (feedback-based stimulation) device. Several intracranial areas and peripheral nerves have been targeted and shown promising results. When the stimulation site is the peripheral part of the cranial nerves, the stimulation ascends through brainstem nuclei and affects the cortex's excitability diffusely. However, the exact indication and the situation when it would be appropriate to use each device is not well defined. The underlying mechanism of action of each therapy, optimal stimulation settings and variables to predict the efficacy and effect on seizure frequency is still a work in progress. Here, we review the role of neuromodulation in DRE. We discuss the most common forms of neuromodulations used in DRE, their mechanism of actions, the evidence related to the process and the outcome.
Vagal Nerve Stimulation (VNS)
This was the first neuromodulator device approved for use by the Food and Drug Administration (FDA) of USA in 1997. It was initially approved for use in patients older than 12 years, but recently in 2017, approval has also been given for use in children more than 4 years of age with partial seizure with DRE. It's an invasive, open-loop device. A generator implanted in the chest wall intermittently stimulates the vagus nerve with pre-programmed current and timing. It is one of the most widely available techniques worldwide.
One proposed rationale of using VNS in DRE is that seizure activity in the brain is associated with venous hyperemia and electrical stimulation of the vagus nerve is associated with a decrease in the hyperemia and hence, the seizure abortion. Another hypothesis suggests that VNS causes desynchronization of the seizure network activity, modulate the neurotransmitter release with increase in gamma-aminobutyric acid (GABA) levels and decreased glutamate levels., More recently, vagal nerve afferents have been postulated to have a role in seizure modulation. Vagus nerve synapses intracranially on the nucleus of tractus solitarius (NTS), which in turn projects to the locus coeruleus (LC) and dorsal raphae. An increase in norepinephrine (NE) and serotonin has been observed which is thought to subserve the clinical effects of VNS. An animal study showed decreased responsiveness in rats in whom the LC was iatrogenically lesioned. Widespread network changes have been observed after VNS, most importantly in the limbic circuit, probably through the NTS and its projections to the amygdala.
The efficacy of VNS was demonstrated by two RCTs – EO3 and EO5,, which formed the basis for the subsequent FDA approval. They recruited patients older than 12 years with focal epilepsy with history of more than five seizures per month. Patients were randomized to receive high (3 mA, 20- to 50-Hz, “on” times 30–90s, “off” times 5–10 min) or low (up to 2.75 mA, 1–2 Hz, “on” times of 30 sec, “off” times 1–3 hour) stimulation. This was followed by a 3-month blinded evaluation. A mean seizure frequency reduction of 24.5% and 27.9% in the high stimulation group was seen in the EO3 and EO5 studies respectively as compared to 6.1% and 15.2% in the low stimulation groups. A point of criticism of these trials was that there was no real blinding due to the high incidence of stimulation related hoarseness of voice (as high as 66%), as even the control group was stimulated at a lower setting.
Long-term analysis has shown VNS to be efficacious in focal, generalized as well as syndromic epilepsy. A study assessing outcomes in 440 patients showed a responder rate (patients attaining >50% reduction in seizure frequency) of 36.8% at one year. This subsequently improved to 43.2% at two years and 42.7% at three years. A large meta-analysis was published in 2011 on VNS efficacy in epilepsy that included 74 clinical studies (3 blinded, RCTs; 2 nonblinded, RCTs; 10 prospective studies; and numerous retrospective studies) with 3321 patients with DRE. Its result showed an average 45% reduction in the seizure frequency with a 36% reduction in seizures at 3-12 months after surgery and a 51% reduction after >1 year of therapy indicating a better long-term neuromodulation in the seizure circuits with VNS than immediate suppression of seizure activity. Posttraumatic epilepsy and tuberous sclerosis were found to be the positive predictors of a favorable outcome after VNS in this metanalysis. Patients with generalized epilepsy were seen to respond better. Similar outcomes were seen in a Japanese database of 362 patients, with responder rate of 57.7% and 58.8% at two and three years respectively.
VNS has also been effective in children younger than 12 years of age. A study compared the outcomes of 86 children less than 12 years to adolescents. They found no difference in efficacy or complication rate between the groups. A recent retrospective analysis of VNS implanted in 99 children less than six years of age reported 55%, 60% and 52% responder rate one, two and four years after implantation. The overall complication rate was also comparable to adult series at 5.6%. Evidence of its therapeutic effect in children less than three years is also emerging.
There are case reports and studies suggesting benefit of VNS in status epilepticus, and epileptic syndromes like tuberous sclerosis,, Lennox-Gastaut syndrome, generalized epilepsy with febrile seizures plus, and absence epilepsy.
A significant improvement in drug-refractory depression has also been reported. FDA approval for the same was received in 2007. The effect of VNS on the rate of sudden unexpected death in epilepsy (SUDEP) is debatable with one study showing a significant decline in death rate at long-term follow-up up to 10 years as compared to the initial two years of follow-up. However, another study has failed to corroborate this finding.
Quality of Life (QoL)
A multicenter prospective trial (PuLsE) concluded that VNS with best medical therapy had significant improvement in health-related QoL, as compared to the medical management group alone. Overall improvement in attention, cognitive ability, memory, creativity and decision-making has been observed., An improvement in quality-adjusted life years of 5.96 (patients aged 1-11 years) and 4.82 years (in patients aged 12-17 years) has been reported. Epilepsy has a significant economic burden. Reduction in total health care costs by as much as 3000USD per patient per year and decreased emergency room visits has been found after VNS implantation, even in patients not having a significant reduction in seizures.,,
Many studies have shown that VNS implantation is a relatively safe operation. However, there are a few complications that the epileptologist should be aware of. It is an expensive procedure and may have mechanical equipment-related complications including lead fracture, or malfunction with interruption of the wire-electrode circuitry and need for battery replacement at regular intervals. Stimulation based side effects such as hoarseness of voice, coughing and laryngeal paresthesia are thought to be because of the efferent nerve fibers supplying the larynx. These are the most common side effects, the incidence of which has been reported to be as high as 60%.,,, Fortunately, they have been seen to improve with time and device parameter adjustment. Bradycardia has infrequently been observed due to the stimulation of the sinoatrial and atrioventricular nodes in the heart. Worsening of preexistent (50%) or development of new onset (57.9%) sleep apnea has been reported in small series of patients. Surgical complications such as infections (3-6%) and lead damage (3%) have rarely been seen.,,,,
Transcutaneous VNS (tVNS) has been introduced as a non-invasive technique, targeting either the auricular branch or the cervical branch of the vagus in the neck. A pilot study conducted on 10 patients showed reduction in seizure frequency in 50% of the patients. A controlled trial (n = 60) demonstrated significant reduction (p < 0.001) in seizure rate and intensity at twelve months follow-up. Improvement in mood and QoL was also noted.
A form of closed-loop VNS has been introduced in 2015, wherein the device detects tachycardia due to seizure onset and then delivers automatic stimulus., Another model detects bradycardia and prone positioning of the patient and differs the stimulation accordingly to offset the risk of SUDEP., Despite these ancillary benefits, significant improvement in seizure outcome of these devices over and above the regular VNS is debatable.,
Ongoing attempts to predict clinical response to VNS have shown interesting results. A study analyzed outcomes in 82 patients with and without cognitive deficits receiving VNS. They found patients without cognitive defects had improved seizure-free outcome, although secondary benefits like improved alertness were more in the group of patients with cognitive deficits. An improved synchronization of EEG waveforms has been seen in patients who respond to VNS. A recent review on the biomarkers of VNS response reported connectomics-based studies utilizing diffusion tensor imaging, magneto encephalography and functional MRI as the most promising avenues to predict seizure outcome. Further research in this field may improve preoperative decision making.
VNS device ushered in the modern age of neurostimulation. It is the only neuromodulation device approved for use in children. Epileptologists have extensive experience with the device and implantation is relatively simpler. It can be used to treat both focal and generalized epilepsy. Additional unique benefits include mood elevation and is the device of choice for patients with co-existent depression. Some disadvantages include poorer efficacy as compared to other modalities, MRI incompatibility and difficulty in device removal due to adhesion formation.
Deep Brain Stimulation (DBS)
In DBS the electrodes are directly implanted into the deeper epileptogenic targets in the brain to abort the episode. It is an invasive, open-loop neuromodulator device. Although many targets were studied for use by DBS, efficacy in anterior nucleus (AN) of thalamus was the first to be established by an RCT which led to the approval of the technique in Europe in 2010, followed by the USFDA in 2018. Direct targeting of the anteroventral AN near the mammillothalamic tract has been shown to be most efficacious.,
AN is well connected to the limbic circuit. It also sends projections to various cortical structures such as the orbito frontal, cingulate and mesial frontal cortices. Animal studies in the 1960s showed the effect of subcortical lesioning in achieving seizure control., Subsequently, high-frequency stimulation was seen to prevent seizure activity in rats. The first studies in human subjects were performed by Cooper and Upton et al. who showed chronic stimulation was efficacious in seizure control.,Low-frequency stimulation was seen to synchronize the EEG pattern and cause seizures. On the other hand, high-frequency stimulation was seen to disrupt the same and thus abort seizure activity. Stimulation induced changes in the ion channels, synaptic levels of neurotransmitters and subsequent glial changes leading to overall network modulation have been proposed as the mechanism of action.,,
A multicenter RCT (stimulation of the anterior nucleus of thalamus for treatment of refractory epilepsy – SANTE) included 109 adults. A 40.4% seizure reduction was reported in the stimulation group (14.5% in the control group) at the end of the blinded phase of evaluation (12 weeks). The responder rate at the end of one year was 43%, which improved to 54% at two years post-implantation. A criticism pointed out was that almost half (44.6%) of the study population had received prior VNS implantation, although recent reports have observed favorable outcomes in this subset of patients. Five year follow-up of the patients in the RCT revealed 69% seizure reduction and 68% responder rate, with 16% of patients reporting seizure freedom of at least six months. Post hoc analysis suggested improved seizure reduction for temporal lobe epilepsy in comparison to other lobes.,
A recent systematic review of DBS for DRE in children found 40 patients aged 4-18 years. Various areas were targeted, with eight patients receiving bilateral AN-DBS. Overall, 85% of patients showed seizure reduction. Case reports also suggest that DBS may have limited use in patients with refractory status epilepticus.
Quality of Life (QoL)
Significant improvement in the QoL has been reported at long-term follow-up. A relatively high rate of depression (14.8%) and memory decline (13.0%) was initially reported in the stimulation group of the SANTE trial. However, long-term studies with seven years of follow-up have allayed these concerns with objective measures reporting no significant decline in the two parameters. Additionally, an improvement in attention and executive function was noted.
Several other intracranial targets for DBS have been tried with varying success rates. DBS of the centromedian nucleus of thalamus (CMNT) seems to be effective in reducing generalized as compared to focal seizures. Good response has been seen in patients with Lennox Gastaut syndrome. In a study, only one of five patients with frontal lobe epilepsy showed a worthwhile reduction in seizures, as compared to all six of the patients with generalized epilepsy.
Targeting the hippocampus is indicated in patients with bilateral mesial temporal sclerosis, in MRI negative temporal lobe epilepsy, in patients with risk of memory deficits and in cases of previous temporal lobectomy with a contralateral relapse of seizures., Long-term follow-up in nine patients showed >95% seizure reduction in patients with a normal MRI, while 50-70% reduction was seen in substrate positive patients. In a recent RCT, seven out of eight patients in the control group had more than 50% reduction in seizures, and four of them became seizure free. Hippocampal DBS has been shown to be a safe procedure without any neuropsychological adverse effects. However, resection is still the procedure of choice in patients with a recognized unilateral seizure substrate.
Cerebellum was the oldest targets of DBS, although recent research targeting it is sparse. Electrodes are placed on the superomedial surface of the cerebellum. An RCT showed significant reduction in generalized seizure frequency in the stimulation group and overall reduction of seizure rate to 41% of baseline at the end of six months.
The most common adverse event noted has been paresthesia and implant site pain noted in 23% of patients on long-term follow-up. Implant site infection has been noted in 12.7% of cases. Lead mis targeting rate of 8.2% shows the importance of postoperative imaging to confirm lead position. Rate of hemorrhage has been comparable to the experience with implantation of DBS electrodes for other indications (2.4%).
DBS seems to be effective for both focal and generalized seizures, although maximum benefit seems to be in temporal lobe epilepsy. Although the efficacy seems to be slightly higher than VNS, DBS requires more frequent battery changes due to higher stimulation parameters and its more expensive. Other disadvantages are the risk of target mismatch and a more elaborate surgical procedure as compared to VNS.
Responsive Neurostimulation (RNS)
RNS is a form of invasive, closed loop device. It includes a set of recording electrocorticogram electrodes, the feedback of which is used to deliver electrical stimulus to the seizure focus upon the detection of epileptiform activity. The threshold at which the device responds is predetermined, and it does not require any further human setting. The stimulus thus delivered is expected to abort the seizure right at its onset. Additionally, the device provides valuable chronic electrophysiological data for research purpose. RNS has been approved by the USFDA in 2013 for adult patients with refractory focal epilepsy, with upto two identified seizure foci on preoperative evaluation.
It was observed by researchers in patients undergoing intracranial monitoring that stimulation in response to a seizure could reduce its frequency. They also reported that such a stimulus has to be delivered right at the onset of abnormal epileptiform recording. Acute stimulation has been observed to disrupt synchronization associated with epileptiform activity by upregulation of local GABA activity. Additionally, depletion of neurotransmitters has been reported at synaptic terminals causing depression of activity. Long-term effects have been postulated to be mediated via induced changes in gene expression with chronic stimulation.
The pivotal trial which established the efficacy of RNS included 191 patients assessed over a 12-week blinded period to receive stimulation vs no stimulation in response to seizure detection. The treatment group had a significant reduction in seizure frequency (n = 97, 37.9%), as compared to the control group (n = 94, 17.3%, P = 0.012). No significant difference was observed in the adverse effects and favorable outcome was seen in both the groups in the unblinded phase. Further follow-up demonstrated 44% seizure reduction at one year, and a significant improvement to 53% at the end of two years.
Subset analysis showed that at an average of 6.1 years of follow-up, the best response (70% median percent seizure reduction) was seen in patients with frontal and parietal seizure foci, while it was seen to be 58% with temporal neocortical onset and 51% with multilobar foci. Authors further demonstrated that eloquent cortex stimulation could be performed with no stimulation related adverse effects. Visible MRI abnormality was a predictor of better response (p = 0.02). There was no effect of prior surgery or VNS implantation on the seizure response. Another subset of patients with mesial temporal onset of epilepsy (n = 111, 72% of which had bilateral mesial temporal onset) were observed for a mean follow-up of 6.1 ± 2.2 years. Patients had a mean seizure reduction of 70% with 29% of them experiencing seizure free periods of longer than six months. More recently, a 9-year follow-up of all the patients (n = 230) enrolled in multicenter open label trials of RNS showed encouraging results. A significant median seizure reduction of 75% was reported, while the responder rate was 73%. Additionally, the SUDEP rate was found to be significantly reduced with RNS (p < 0.05). The efficacy of RNS in pediatric patients as an off-label therapy is restricted to case reports as of now.,
Quality of Life (QoL)
In the pivotal trial, in addition to the benefit with respect to seizure reduction, there were significant improvements seen in QoL following treatment. Analysis performed at two years in a subset of patients enrolled in the trial with temporal onset of seizures showed significant improvement in naming in patients with neocortical onset (p < 0.0001), while significant improvement in verbal learning was observed in patients with mesial temporal onset (p = 0.005). Long-term follow-up of nine years has also shown significant improvement in cognition and overall QoL. No definite adverse effect on mood has been reported.,
The rate of hemorrhage and infection has been found to be similar to other invasive neuromodulation devices., The risk of infection has been found to be 4.1%, with approximately half of them requiring explantation. The rate of hemorrhage has been seen to be 2.7%, with none causing any neurological deficit. A recent analysis of the manufacturer database between 2013 to 2020 reported overall 241 complications. 40% of these complications were infection related, while the second most common were lead breakage (12%).
RNS is a promising therapy for refractory partial seizures, especially in bitemporal epilepsy. The use of RNS is currently restricted to cases wherein two seizure foci are recognized, and thus requires extensive evaluation to identify the correct epileptogenic zone. It is the most expensive neuromodulator device available, although a longer battery life has been reported as compared to other modalities. Chronic long-term electrical recordings made available through this device are a unique advantage in increasing understanding of epilepsy networks. Recognition of the exact biomarkers of abnormal seizure activity would further improve the efficacy.
Trigeminal Nerve Stimulation (TNS) involves non-invasive bilateral stimulation of the supra orbital nerves. In an RCT (n = 50) patients either received high frequency (120 Hz) or low frequency (2 Hz) stimulation. A 30.2% responder rate was observed in the former as compared to 21.1% in the latter. Although this difference was not significant, significant within group improvement was seen with high frequency stimulation. Another recent RCT comparing TNS with medical management was able to show a 50% responder rate with TNS at 12 months follow-up. Overall, TNS is a relatively economical technique with easy applicability.
Repetitive Transcranial Magnetic Stimulation (rTMS) uses external magnetic fields to induce electrical currents which affect neurons to cause seizure reduction. Repetition causes longer lasting effects. An RCT of 21 patients with cortical dysplasia showed significant reduction in seizure frequency, with the beneficial effect maintained for two months. Headache was the most common adverse effect (25%).
Transcranial Direct Current Stimulation (tDCS) is a portable, non-invasive mode of treatment, wherein changes in cortical excitability are induced using weak direct current (2 mA) to influence resting membrane potentials. Twelve patients with mesial temporal sclerosis underwent tDCS and showed a responder rate of 83.33% at one month. In a randomized double-blind trial including 28 patients, 48% mean seizure frequency reduction was seen at two months.
Although these techniques are promising, further long-term experience in terms of sustained efficacy are required.
Concerns for low- and middle-income countries
There are several special considerations for the use of neuromodulation in developing countries. Even though these devices have been shown to reduce the long-term economic burden of epilepsy, the upfront cost of the device is still out of reach for many of the economically weaker section of patients. Additionally, some of the devices like RNS are currently not available in developing countries. There is a dearth of medical personnel in many of these countries. The availability of trained experts in neuromodulation is even rarer. Many patients travel from far off places for specialized treatment. Thus, regular follow-up for device programming is a big issue and is simply not possible for many. Lastly, there is a great potential for misuse of these devices. Patients as well as physicians need to be educated as to the risks and benefits of these techniques. Regulatory control to prevent implantation for improper indications is required. It is for all these reasons, that relatively more invasive palliative procedures such as corpus callosotomy are still used in several institutes.
Neuromodulation techniques are emerging as effective and relatively safe treatment options for DRE not amenable or not responsive to resection. Overall, these devices provide a significant added seizure reduction without typical medication side effects and their efficacy continues to improve over years. Several ancillary benefits have been noted including improvement of quality of life and overall reduced economic burden of epilepsy. However, the complete seizure freedom is rarely achieved using these techniques and there still are a subgroup of DRE patients who do not benefit from these therapies, leaving a scope for further advancement. Disadvantages such as hardware related complications, need for battery replacement and frequent parameter adjustment have been noted. The recent trend towards development of closed loop devices is an effort to offset some of these disadvantages. Currently there are no head-to-head trials comparing the different neuromodulation devices. The choice of device therefore depends on the type of epilepsy, whether the seizure focus can be identified, age of the patient, and other factors like availability and cost. Evolution in hardware and technology have continuously improved device longevity and performance. Advanced imaging and analysis of electrophysiological data including machine learning have shown promise in improving accuracy and predicting response to therapy. All this has led to the blossoming of neuromodulation for epilepsy into a promising area filled with exciting possibilities.
The paper has been partially funded by Dept of Biotechnology, Ministry of Science and Technology, India.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.