Stimulation of the cervical vagus nerve with high frequency and low voltage has been shown to induce synchronization on EEG, with increasing voltage leading to desynchronization (10). Vagal nerve stimulation (VNS) has also been shown to significantly increase the inhibition of circuits in the motor cortex by increasing the activity of GABA_A receptors, supporting the notion that neurotransmitter modulation is an underlying mechanism in seizure termination. The locus coeruleus (LC) and dorsal raphe nucleus (DRN) are two destinations for the nucleus tractus solitarius (NTS), a location where a considerable number of vagal afferents terminate. Consequently, the LC and DRN have been implicated in the mechanism of VNS. VNS has been shown to increase noradrenergic and serotonergic activity, from the LC and DRN, respectively, and reduce seizures in rat models (11).

In 1997, the Food and Drug Administration (FDA) approved the use of implantable vagus nerve stimulators for the treatment of drug resistant focal onset impaired awareness seizures in patients older than 4 years of age after an open label study demonstrated seizure reduction of >50% in 36.8% of patients 1 year after implantation, 43.2% of patients at 2 years, and 42.7% at 3 years (12). Reported adverse events were hoarseness (28%) and paresthesia (12%) at 1 year, hoarseness (19.8%) and headache (4.5%) at 2 years, and shortness of breath (3.2%) at 3 years. In a recent meta-analysis, 74 studies and 3,321 patients with intractable epilepsy were identified (13). VNS reduced the frequency of seizures by an average of 45%, 36% at 3–12 months and 51% after 1 year. The study also demonstrated the extension of VNS benefits to patients with generalized epilepsy, who experienced a reduction of 57.5%. Furthermore, patients with tuberous sclerosis and post-traumatic epilepsy benefited the most, with seizure reduction of 68 and 79%, respectively.

Novel advancements in VNS technology have allowed for the utilization of a closed loop stimulation approach in VNS using a heart rate-based seizure detection algorithm (14). Further, improvements in VNS programming, such as the ability to personalize therapy according to time of day, have enhanced the capability of VNS to be tailored to each patient's needs. Recent studies have compared efficacy between open and closed loop VNS (15, 16). One retrospective study found that closed loop VNS was associated with a greater reduction in seizure frequency compared to open loop VNS at 9 months following implantation (15); however, this difference was no longer present at 24 months after implantation. Another study comparing open and closed loop VNS in pediatric patients reported similar efficacy between the two in reducing seizure frequency (16). These findings suggest that open and closed loop VNS are both viable options for refractory epilepsy.

Trigeminal Nerve Stimulation (TNS) is a non-invasive, transcutaneous stimulation modality for medication-resistant epilepsy. Mechanistically, TNS exerts its effects through modulation of the trigeminal nucleus and its projections. Afferents from the trigeminal nerve synapse onto the NTS and LC (17), which are two brain regions that have been previously implicated in playing a role in seizure reduction and are targeted using VNS as well (11).

In 1976, pioneering work by Maksimow illustrated that application of pressure on infraorbital branches of the trigeminal nerve can inhibit generalized tonic clonic seizures if done prior to the beginning of convulsions (18). Following this, several studies supported the notion that TNS can reduce seizure frequency (19–21). The strongest evidence for safety and efficacy of TNS in refractory epilepsy stems from a double blind randomized controlled trial of 50 patients (22). At the conclusion of the 18-week blinded period, 30.2% of the stimulation arm and 21.1% of the control arm had a reduction in seizure frequency of >50% (responder rate). Additionally, the stimulation group had an increase in responder rate over the 18 weeks that was not seen in the control group. Side effects of TNS during the study period included skin irritation (14%), headache (4%), and anxiety (4%) (22). In a separate study investigating TNS safety, no short- or long-term cardiovascular side effects were seen with TNS (23).

In contrast to other neuromodulatory techniques, TNS does not require implantation of hardware. This is a key advantage of this modality and is ideal for poor surgical candidates. Furthermore, TNS has been linked with improvements in mood (22, 24) and could be beneficial for patients with co-morbid depression.

Responsive neurostimulation (RNS) is the initiation of stimulation in response to the detection of an epileptic event electrographically and is known as a closed-loop neuromodulation system. Of great significance to RNS, is the work of Psatta (25). His work demonstrated that responsive stimulation was more effective than continuous, open-loop stimulation in terminating epileptic activity in cat models. He also highlighted the temporal importance of stimulation; stimulation was more effective when applied at shorter intervals after detection of epileptic patterns (25). The later works of Motamedi et al. introduced the significance of spatial stimulation in seizure suppression. They found that stimulation of seizure onset zones was most effective in terminating after-discharges, supporting the notion that direct stimulation of epileptic foci enhances effectiveness in terminating seizure propagation (26).

RNS systems were approved by the FDA in 2013 for the treatment of drug resistant partial onset epilepsy in patients above the age of 18 years who are refractory to ≥2 trials of antiepileptic medications, have ≤2 epileptogenic foci, and experience significant impairment due to frequent seizure activity.

Closed-loop stimulation leads to the termination of synchrony in seizure onset zones to prevent propagation of the ictal stage. Electrical stimulation of these foci during seizure onset leads to alterations in cell membrane currents and hyperpolarization rather than desynchronization (27, 28). Axonal conduction, depression, and blockade are also hypothesized to be underlying mechanisms (29). Finally, chronic stimulation has been shown to alter gene expression, leading to cortical reorganization and synaptic plasticity that could further enhance the anti-epileptic effects (30).

A 9-year prospective study evaluated the efficacy and safety of RNS in patients with intractable focal onset epilepsy. Two hundred and thirty patients were recruited and 162 were able to complete the 9-year follow-up period (31). The percentage of seizure reduction was recorded at intervals of 6 months and median seizure reduction was 75%. Furthermore, 21% of patients achieved seizure freedom and greater than one third of patients achieved >90% reduction. Overall, seizure reduction percentages improved over time following implantation. Patients also reported improvement in their quality of life and perception of cognition, compared to temporal lobe resection and selective amygdalohippocampectomy.

Complications were no more significant than other treatment methods for epilepsy, including VNS, DBS, and surgical resection. Infection rate was 4.1% per procedure and 12.1% for the 1,895 patient implantation years. All infections were in the soft tissue and involved skin flora. 2.7% of patients experienced non-seizure related hemorrhage without neurological deficits. There were no exacerbations in depression, suicidality, or memory impairment (31).

In the early 1900's, Clarke and Horsley were the first to develop a stereotaxic apparatus that could facilitate targeted exploration of the deep ganglia and tracts (32). Although electrical stimulation was used in such operations, its main objective was to confirm the target areas prior to lesioning. Exploration and utilization of chronic DBS as a therapeutic rather than diagnostic intervention quickly ensued (33). Utilization of DBS in a wide range of conditions, such as essential tremor, Parkinson's disease, and pain, built the safety profile of this intervention, leading to a rapid decline in lesioning interventions. The success of DBS in movement disorders drew attention to its use in refractory epilepsy.

In 2018, the FDA approved the use of DBS for patients with refractory partial-onset seizures, with and without generalization, who are 18 years of age or older. Contraindications for DBS are minimal. DBS is contraindicated in patients who are incapable of operating the neurostimulator, have significant psychiatric contraindications, or are medically frail and unable to undergo surgical implantation (34).

The anti-epileptic mechanisms of DBS are largely unknown and complex. The proposed underlying mechanisms include inhibition, likely due to blockade of depolarization and voltage gated currents, or activation of GABAergic neurons (35–37). Shifting the focus from the effects of DBS on neurons, some studies highlighted the role of astrocyte activation in DBS by inducing local neuronal modulation (38). Electrical currents by DBS have been shown to induce electrotaxis- the migration of progenitor cells due to electricity- which can serve to promote neurogenesis and neuroplasticity that could alter neuronal pathways (39).

Most of the data regarding the safety and efficacy of DBS in epilepsy patients stems from the results of SANTE trial (40). The SANTE trial was a randomized, double blinded clinical trial investigating anterior nucleus of thalamus stimulation for epilepsy and consisted of a 3-month blinded study period. DBS implantation itself resulted in seizure reduction in both groups. Initial median seizure reduction was 33.9% in the active stimulation arm and 25.3% in the control arm in the first month following DBS implantation. However, this difference between groups expanded further during the final month of the blinded phase; seizure reduction diminished in the control arm to 14.5%, while the treatment arm exhibited a 40.4% reduction compared to baseline. Stratification by lobe of onset during a post-hoc analysis demonstrated a significant reduction rate for the temporal lobe (43.9% treatment vs. 29% control) that was not observed for other lobes (41). Additionally, patients with complex partial seizures benefited more than patients with simple partial seizures. A long-term follow-up study evaluating the safety and efficacy of DBS in the SANTE cohort demonstrated continued improvement of efficacy overtime (42); Specifically, a 41% and 69% median seizure reduction from baseline was noted after DBS at one and five years, respectively, in the SANTE cohort. At 5 years following surgery, differences in seizure reduction were present when stratifying by lobe of onset. Those with temporal lobe epilepsy experienced the greatest median seizure reduction (76%), while those with frontal lobe epilepsy experienced the least seizure reduction (59%).

Adverse effects over a follow-up period of 5 years included paresthesia and pain at the site of implants (20.9%), implant site infections (12.7%), and improper lead positioning (8.2%). A possible stimulation related adverse effect was depression (32.7%); however, 66% of patients had pre-operative depression (42).

The thalamus is a sensory relay center with widespread synaptic connectivity to many cortical regions. The role of this key subcortical structure in the propagation of abnormal epileptiform activity has long been described (43). Direct electrical stimulation of the thalamus allows for modulation of neural circuitry and ultimately interferes with seizure propagation. The specific anatomical and physiological profiles of various thalamic nuclei, detailed in the subsections below, make them attractive targets for DBS and RNS in epilepsy patients.

While DBS of the anterior and centromedian thalamic nuclei for epilepsy have been studied extensively, a novel promising target is the median pulvinar thalamic nucleus. Additionally, most studies evaluating the effectiveness and safety of RNS are limited to cortical stimulation, with less being known about subcortical targets.

While open loop ANT stimulation is the most established target for epilepsy, recent studies have shown promise with deployment of closed loop stimulation (Table 1). Because patients with medically resistant epilepsy who are candidates for neuromodulation may be suitable for open or closed loop stimulation, selecting the optimal therapy can be challenging. The primary advantage of closed loop stimulation is the ability to provide a more personalized approach to care by configuring stimulation in response to the patient's specific needs and epileptic activity. However, this is limited by our current insufficient understanding of optimal stimulation parameters and electrode selection, and thus prevents maximal efficacy with a closed loop approach (84). In comparison to open loop stimulation, closed loop stimulation carries a lower burden of stimulation, fewer stimulation related side effects, fewer cognitive and mood disruptions, and records chronic ambulatory EEG data. It is also useful in measuring seizure burden in response to changes in antiepileptic medications, behavior modification, and in characterizing neurobehavioral spells. With utilization of open loop stimulation, the complexities of precise seizure localization and seizure detection algorithm set up required for closed loop stimulation can be avoided (85). A drawback of open loop stimulation, however, is the lack of capability to personalize therapy. As both open and closed loop stimulation have been shown to be efficacious and safe in managing medically refractory epilepsy, treatment decision making should be guided by patient specific factors.