INTRODUCTION — Antiepileptic drugs (AEDs) are the primary treatment modality for patients with epilepsy. However, approximately one-third of patients with epilepsy continue to have seizures on medication [1]. Polypharmacy can reduce the number of seizures in many of these patients, but may not lead to complete seizure control, and medication side effects are often increased. Nonpharmacologic options include epilepsy surgery, ketogenic diet, and vagus nerve stimulator (VNS) therapy.

In 1997, the VNS Therapy System™ (Cyberonics, Inc) was approved by the Food and Drug Administration (FDA) as adjunctive therapy for adults and adolescents over 12 years of age whose partial-onset seizures were refractory to antiepileptic drugs. This topic reviews the relevant anatomy, possible mechanisms of action, and clinical results of VNS in patients with epilepsy. Other aspects of epilepsy treatment are discussed separately. (See "Overview of the management of epilepsy in adults".)

ANATOMY — The vagus nerve is primarily comprised of afferent nerve fibers, which have their cell bodies in the nodose and jugular ganglia and synapse bilaterally on the nucleus of the solitary tract (NTS) in the brainstem. From the NTS, vagal afferent pathways project on many regions in the brain, including pontine and midbrain nuclei, the cerebellum, thalamus, and cortex.

One vagal pathway, perhaps of particular relevance to epilepsy therapy, ascends to the forebrain via the pontine parabrachial nucleus [2]. This pathway transmits sensations of visceral origin to the ventroposterior parvocellular nucleus of the thalamus, which then projects to the insular cortex [3]. The parabrachial nucleus also projects to other thalamic nuclei, the amygdala, and the basal forebrain. These vagal projections to sites that are often found to generate seizures are likely relevant to the therapeutic effect of vagus nerve stimulation (VNS) in epilepsy treatment.

The locus coeruleus, another pontine nucleus, also receives afferents from the NTS. While receiving less dense vagal input than the parabrachial nucleus, the locus coeruleus may be essential to the antiepileptic effect of VNS, as suggested by animal studies using ablative and immunolabeling procedures [4,5].

Vagal efferent fibers originate in the dorsal motor nucleus of the vagus and the nucleus ambiguous. These innervate the heart, vocal cords, and other laryngeal and pharyngeal muscles, and also provide parasympathetic input to the gastrointestinal viscera [6]. Because the right vagus nerve provides more innervation to the cardiac atria than the left vagus nerve, electrical stimulation of the left vagus nerve is generally used in clinical practice to avoid adverse cardiac effects [7]. However, right-sided VNS has been reported safe in at least one case series, and in animal studies it appears equally effective against seizures as left-sided stimulation [8,9].

MECHANISM OF ACTION — The vagus nerve stimulator (VNS) is a battery-powered device similar to a cardiac pacemaker. Stimulating leads are placed around the left vagus nerve in the carotid sheath and are connected to an infraclavicular subcutaneous programmable pacemaker (figure 1). A wand attached to a laptop computer is waved over the device to program the stimulation intensity, frequency, duration, and other therapeutic parameters.

While vagus nerve stimulation therapy has been shown to be an effective treatment for epilepsy, the mechanism of its therapeutic effect remains unknown.

In the 1960's, studies in animals showed that repetitive vagus nerve stimulation (VNS) either synchronizes or desynchronizes cortical electrical activity, depending on the stimulus frequency and current strength, which, in turn, determine the relative activation of myelinated versus unmyelinated fibers [10-12]. Because epileptic seizures are characterized by hypersynchronized cortical activity, this observation that VNS can desynchronize cortical rhythms was one of the earliest findings that suggested a potential antiepileptic effect of VNS.

VNS studies in a variety of animal models were then undertaken and have demonstrated that VNS has multiple antiepileptic properties [13-20]:

• VNS can abort an ongoing seizure after seizure onset
• VNS is effective in acute seizure prophylaxis; ie, seizure-inducing insults (eg, strychnine administration) are less effective in inducing a seizure in the presence of VNS
• VNS is effective in chronic seizure prophylaxis, reducing seizure frequency in animal models of epilepsy
• VNS can inhibit epileptogenesis in animal models of seizure kindling

How VNS exerts its antiseizure properties remains unclear. Studies in animal models and in patients with epilepsy suggest some possible facets of underlying mechanisms:

• Low-intensity trains of VNS hyperpolarize (inhibit) pyramidal neurons in the rat parietal association cortex [21]. Low-intensity stimulation, which predominantly activates myelinated fibers, was more effective in this model in inducing long-lasting inhibitory effects than higher stimulus intensities, which also entrains nonmyelinated vagus fibers. Other studies also suggest that the unmyelinated C fibers are not activated in therapeutic VNS [14,22]. This may explain why efferent vagal symptoms do not more commonly complicate VNS.
• The cortical inhibition that is associated with VNS may be secondary to the release of inhibitory neurotransmitters, such as glycine and gamma-aminobutyric acid (GABA). A study using transcranial magnetic stimulation in five patients with VNS showed that effective vagus nerve stimulation was associated with a pronounced increase in cortical inhibition without an observed effect on cortical excitability [23]. This finding is consistent with enhanced GABA-A cortical activity. In another small case series, a therapeutic response to VNS was associated with normalized, ie, increased, GABA receptor density [24]. GABA-mediated neuronal inhibition is a therapeutic mechanism that also underlies several antiepileptic drugs. (See "Pharmacology of antiepileptic drugs", section on 'Drugs that affect GABA activity'.)

• The electrophysiological effects of VNS are not well characterized. Studies of interictal electroencephalogram (EEG) background rhythms in patients with VNS have had mixed results in regard to whether VNS is associated with a reduction in interictal epileptiform discharges in epileptic patients, and whether this change is associated with a therapeutic response to VNS [25-28].
• A number of neuroimaging studies have attempted to define which central structures are integral to VNS effect. Positron-emission tomography (PET) and single photon emission computed tomography (SPECT) studies have variably demonstrated regional areas of increased blood flow in the thalamus and temporal cortex with VNS [29-35]. Correlations between the extent of these changes in blood flow, particularly in the thalamus, and reductions in seizure frequency are also reported, although not consistently [31,33,34]. Functional magnetic resonance imaging (fMRI) is a more sensitive test than SPECT and PET for evaluating neural activity. Diffuse activation in multiple cortical and limbic regions, as well as the thalamus, have been seen in patients treated with VNS [35-39]. Regional activation varies with the pulse width of VNS [36]. In one study, thalamic activation appeared important to the antiepileptic effect of VNS [39].

EFFICACY — The first case series of patients treated with vagus nerve stimulation (VNS) was reported in 1990 [40]. This was followed by two pivotal trials of VNS in patients with partial epilepsy, the E03 study [41-44] and the E05 study [45].

Partial-onset seizures — The E03 and E05 studies were multicenter, blinded, randomized trials that compared two different VNS stimulation protocols for partial-onset seizures: active treatment or high stimulation (30 Hz, 30 seconds on, five minutes off, 500 microseconds pulse width) and an active control, low stimulation (1 Hz, 30 seconds on, 90 to 180 minutes off, 130 microseconds pulse width) [41-45]. Enrolled patients were at least 12 years of age with at least six seizures per month. Patients were treated with a mean of 2.1 antiepileptic drugs (AEDs) at study entry; these were not adjusted during the study period. The primary efficacy measure was the percentage change in seizure frequency during VNS treatment (measured over 12 weeks, two weeks after implantation) compared with a 12-week preimplantation baseline.

• In the E03 study, 114 patients with predominantly partial seizures were enrolled [44]. The high-stimulation group experienced a greater mean reduction in seizure frequency compared with the low-stimulation group (24.5 versus 6.1 percent). More patients in the high-stimulation group experienced a greater than 50 percent reduction in seizures (31 versus 13 percent).
• In the E05 study, 199 patients with complex partial or secondarily generalized seizures were enrolled [45]. The mean reduction in seizure frequency was greater in the high-stimulation group (28 versus 15 percent). A greater than 75 percent reduction in seizures was achieved in more patients in the high-stimulation group (11 versus 2 percent).

Patients in both studies, as well as 124 other patients who received VNS on a compassionate-use basis, were followed for an additional 12 months [46-48]. All patients received high-stimulation VNS. Median seizure reductions compared with baseline appeared even greater than in the controlled portion of the study: 34 percent at three months and 45 percent at 12 months. At 12 months, 20 percent of patients achieved seizure reduction of greater than 75 percent. These and other open-label studies suggest that the efficacy of VNS may improve over time [49-54]. However, these results should be interpreted with some caution; there were no control groups in these studies, and AED adjustments, permitted during in at least some of these observational studies, may have contributed to improved seizure control.

Other seizure types and patient populations — A number of open case reports and uncontrolled studies of VNS have been published, suggesting that the benefits of VNS extend to a broad range of patients in regard to age, neurologic comorbidity, epilepsy syndrome, and seizure type.

• Age groups — Experience with VNS in elderly patients with epilepsy is limited. The subset of 45 patients in the EO3 and EO5 studies who were 50 years of age or older experienced a benefit similar to the population as a whole [55]. However, only eight of these patients were older than 60 years, and it is doubtful that this can be considered an elderly population.

There is no evidence that efficacy of VNS is any different in children compared with adolescents and adults [56-59]. It has been used with success in children as young as 11 months.

• Epilepsy syndromes — Case series suggest that VNS is also effective in generalized epilepsy syndromes [53,58-61]. While some studies found that symptomatic generalized epilepsy is more responsive to VNS than idiopathic syndromes, others have reported the opposite or found no difference [53,60,62,63].

A relatively large accumulation of clinical experience with VNS in the Lennox Gastaut syndrome is due in part to the fact that seizures are often medically intractable, and surgical treatment options are limited [58,62-71]. VNS appears to be particularly effective in this patient group. Other refractory epilepsies with onset in childhood can also respond to VNS; these include epileptic encephalopathies (eg, Landau Kleffner syndrome, severe myoclonic epilepsy of infancy, Dravet syndrome), tuberous sclerosis complex, and typical absence epilepsy [53,57,58,60,62-68,72-75].

• Previous epilepsy surgery — VNS treatment has been reported to be successful in patients with previous epilepsy surgery [52,57,58,62,76,77]. The cumulative experience in this group, as documented in the VNS registry, was reported in 2004 [78]. The response was favorable but somewhat less so in the 591 surgical patients compared with the 2382 nonsurgical patients; median reduction in seizure frequencies at 3 and 12 months were 42.5 versus 47 percent and 46 versus 60 percent. While some case series have suggested that the outcomes with VNS may be more favorable in the post-callosotomy group, the VNS registry data suggest that at 12 months, response rates are similar in these two groups.
• On-demand stimulation — On-demand stimulation with the supplied magnet can be an effective adjunct to chronic stimulation for attenuating or interrupting seizures in some patients who experience auras [56]. In the E03 study, patients receiving high-stimulation (active) treatment also received an active magnet, while patients in the control group received an inactive magnet. Use of the active magnet terminated more seizures than the inactive magnet (21.3 versus 11.9 percent) [79]. Similar efficacy was seen in the E04 study. The effect of the magnet on reducing seizure severity or duration was similar in both groups. One case series in children reported that 19 of the 27 patients were able to abort or attenuate at least some of their seizures [59]. A similar percentage of patients in the E04 study reported that they were able to terminate seizures with the magnet [79].
• Status epilepticus — In at least two cases, one adult and one child, VNS apparently terminated prolonged, medically-refractory status epilepticus [80,81].

Nonseizure outcomes — Quality of life measures improve in most studies of VNS in both children and adults [57-59,71,82-86]. While this effect is greatest in those who achieve the highest reduction in seizures, there appears to be an effect that is independent of reduction in seizure frequency, and may relate to independent effects of VNS on mood, alertness, and other factors. As with seizure reduction, improvement in quality of life measures also appear to increase over time.

Evidence indicating that VNS may improve mood and other depressive symptoms in patients with epilepsy is inconclusive [87-89]. In some but not all studies, a positive effect of VNS on mood appeared to correlate with reduction in seizures. VNS is under investigation as a treatment for major depression in nonepileptic patients. (See "Treatment of resistant depression in adults", section on 'Vagus nerve stimulation'.)

VNS may improve daytime alertness and vigilance and reduce daytime sleepiness despite its potential to exacerbate sleep apnea (see 'Other adverse effects' below [67,83,90,91]. In one series, this was documented by longer mean sleep latency times and improved scores on a daytime sleepiness scale. Improved alertness occurred even in patients without significant reduction in seizure frequency [91]. Vagal stimulation of brainstem centers known to promote alertness (eg, parabrachial nucleus, locus ceruleus) may mediate this effect. There appears to be a differential effect of high versus low-intensity stimulus on sleep and alertness. Low-intensity stimulation (<1.5 mA) more consistently improves daytime alertness than higher intensity stimulation [83,91]. In one study, higher intensity stimulation was associated with reduced REM sleep [90].

Children with autism and mental retardation have demonstrated improved behavioral outcomes with VNS treatment of their epilepsy. Benefits included improved alertness, mental age, and performance on functional measures and some cognitive tests, as well as reduced autistic behaviors [57,63,69,72,74,82,92]. In some cases, these improvements appeared to be independent of seizure control. In adult patients, cognition does not appear to be positively or negatively affected by VNS treatment [93,94].

Reduction of AEDs can be of primary benefit to patients and may also impact the quality of life outcomes discussed above. In most cases, patients with VNS continue to require medical treatment [95]. However, some patients are able to reduce one or more drugs and/or doses, and a few patients are seizure free with VNS and no AED treatment [60,96]. There is limited information regarding the frequency of this outcome. Larger series indicate that most patients remain on the same number of AEDs after VNS, but don't specifically discuss dose alterations [54,57,95]. Smaller case series report higher percentages of AED dose and/or drug reductions, perhaps reflecting a more aggressive effort of the clinicians involved in achieving this goal [96].

SAFETY AND TOLERABILITY

Common side effects — In the E03 study, side effects that occurred in at least 5 percent of patients receiving high-stimulation vagus nerve stimulation (VNS) were [44]:

• Hoarseness (37 percent)
• Throat pain (11 percent)
• Coughing (7 percent)
• Shortness of breath (6 percent)
• Tingling (6 percent)
• Muscle pain (6 percent)

Hoarseness was the only side effect that occurred significantly more often with high stimulation than with low stimulation. In the E05 study, shortness of breath and pharyngitis, as well as voice alteration, occurred significantly more often in the high-stimulation group than in the low-stimulation group [45]. Lowering the pulse width of stimulation can alleviate symptoms and allow for higher stimulation intensities [97].

Long-term studies of vagus nerve stimulation (VNS) generally show improved tolerability over time [54]. Among 444 patients who continued VNS after participating in a clinical study, the most commonly reported side effects at the end of the first year post-implantation were voice alteration (29 percent), tingling (12 percent), dyspnea (8 percent), and cough (8 percent); at the end of two years, the prevalence of these complaints was 19, 4, 3, and 6 percent, and at three years, each of these was present in less than 3 percent. High retention rates (>70 percent) reflect the excellent tolerability of this treatment.

Cardiac events — Physiologic studies have generally found no clinically relevant effects of chronic VNS on cardiorespiratory function [98-100]. However, bradycardia followed by transient asystole lasting up to 45 seconds has been reported in association with the lead test conducted during VNS implantation in approximately 0.1 percent of cases [101-103]. Complete heart block due to atrioventricular nodal block was documented in three patients with no reported adverse effects [104]. In some cases, a rechallenge stimulus is uneventful, and the VNS has been implanted successfully without adverse consequences. More often, the procedure is aborted. In general, baseline cardiac conduction disorders are considered a contraindication to VNS.

Two case reports describe VNS-induced episodes of bradycardia and asystole occurring 2 and 9 years after device implantation [105,106].

Mortality — Neither sudden death nor overall mortality rates appear to be increased in patients receiving VNS. Rates of sudden unexplained death in epilepsy (SUDEP) were similar among a cohort of 1819 individuals with VNS to that of other cohorts of patients with medically refractory epilepsy [107].

Magnetic resonance imaging — The presence of any implanted pacemaker is widely regarded as a contraindication to magnetic resonance imaging (MRI). Potential problems include considerable heating at the lead tip, which has been documented in animal experiments, and programming changes of pacing parameters [108]. (See "Principles of magnetic resonance imaging", section on 'Precautions'.)

The VNS manufacturer's (Cyberonics, Inc) guidelines state that brain MRI operating at <2 Tesla and a specific absorption rate of <1.3 W/kg, conducted with a send and receive head coil, with the VNS turned off is safe [109,110]. Other head and body coils are considered unsafe. There are no guidelines for MRI of other body parts nor for machines operating at 3 Tesla (T), which are increasingly prevalent. The published evidence supporting these guidelines is limited:

• An ex vivo study of temperature changes associated with different MRI conditions found no clinically significant temperature changes under the guidelines outlined by the manufacturer [110]. Excessive heating was observed with the use of body coil during imaging of the neck and shoulder but not of the lumbar spine; however, the number of experiments in the latter case was limited. VNS device function was not affected by the MRI procedure.
• A survey reviewing the experience of 27 MRI scans in 25 patients with VNS revealed no significant complications [109]. Most, not all, were conducted under guidelines; one body coil was used; three were conducted with the stimulator on. The investigators concluded that the guidelines as written by the manufacturer are associated with safe performance of MRI.
• A functional magnetic resonance imaging (fMRI) study at 1.5T in four patients showed that Echo-planar MRI (EPI) scanning could be performed safely in epilepsy patients with an implanted vagus nerve stimulator [37].

From what is currently known, use of MRI outside the limits outlined by the manufacturer cannot be considered safe and is not recommended.

End of battery life — As the VNS generator battery is expended, seizure frequency may increase in some patients [111,112]. Others may note decreased or irregular perception of stimulation [112]. Fortunately, the end of battery service can be predicted with the current VNS model, allowing for elective generator replacement before the battery is fully depleted. Patients who experience efficacy with VNS should have the generator battery replaced before it is expended. Patients have been reported who initially responded to VNS, but did not regain seizure control with VNS re-implant after a period of seizure-worsening associated with an expended battery [111].

Other adverse effects — Surgical complications associated with VNS are infrequent. Early reports of electrode failure and lead fracture appear to be largely resolved with improvements in the device and enhancements in surgical techniques and postoperative care [113-116].

• Infection of the subcutaneous pocket that holds the VNS generator, usually with Staphylococcus aureus, complicates surgery in 3 to 7 percent of cases [117-119]. Systemic antibiotics along with explantation of the device and/or open wound debridement is usually required. (See "Infection of cardiac pacemakers and implantable cardioverter-defibrillators".)

• Children with severe mental and motor retardation on assisted feeding may aspirate if fed during VNS [58,115,120,121]. The association between aspiration and VNS may be disputed in some cases [120]. However, in others, magnet inactivation of the stimulator during meal time controls this problem.
• Sleep apnea is a relative contraindication for VNS [122-124]. VNS is associated with more frequent apnea and hypopnea episodes in sleep, but this appears clinically relevant only in those with preexisting sleep apnea. Lowering stimulus frequency can ameliorate VNS-related apnea and hypopnea [123]. Continuous positive airway pressure (CPAP) may be required for some cases [124].
• Unilateral vocal cord paralysis occurs in about 1 percent of cases, and is attributed to intraoperative manipulation of the recurrent laryngeal nerve [44,61]. Most of these recover. Two cases of delayed permanent vocal cord paralysis occurring several weeks after implantation appeared to be self-inflicted by patients who manipulated the device externally, presumably placing traction on the recurrent laryngeal nerve [125]. This phenomenon appears analogous to the "twiddler's syndrome" described in individuals with cardiac pacemakers. (See "Implantable cardioverter-defibrillators: Complications", section on 'Twiddler's syndrome'.)

• Other cranial nerve palsies that can complicate VNS implant include Horner's syndrome and facial paralysis [126].
• Some patients experience uncomfortable spasm of the left chest wall, which has been demonstrated to be due to collateral spread of stimulation to phrenic nerve, causing contraction of the left hemidiaphragm [127]. Contraction of the left anterior sternocleidomastoid muscle may also occur as a result of current stimulating adjacent structures [128]. These symptoms are often precipitated by assumption of certain postures or movement and are relieved by changing position.
• While gastrointestinal side effects might be expected with VNS, reports of this are infrequent. One patient experienced chronic diarrhea after VNS implant, but this is exceptional [129]. One case series documented clinically significant weight loss in 17 of 27 patients who had received VNS [130]. This observation requires further confirmation.
• Forced normalization refers to a phenomenon of psychiatric disturbances that emerge in some patients with long-standing, high-frequency seizures when their seizures are dramatically reduced. This has been described with VNS, usually with reduction in seizure frequency of greater than 75 percent [131,132]. Some but not all of these patients exhibited psychosis prior to VNS implantation. Psychosis in these cases usually responds to psychotropic medication; in some cases lowering the potency of antiepileptic treatment is required.
• Patients are cautioned by the manufacturer not to undergo short-wave, microwave, or therapeutic ultrasound diathermy, which in theory could cause the VNS generator or lead to heat up and cause thermal tissue damage. There are no documented cases of this complication in VNS-treated patients.
• Programmable shunt valves have been affected by use of the magnet [133]. VNS stimulation using the magnet should probably be avoided in individuals with programmable shunts.

CLINICAL APPLICATION — The Food and Drug Administration (FDA) has approved vagus nerve stimulator (VNS) therapy as adjunctive treatment for adults and adolescents over 12 years of age whose partial-onset seizures were refractory to antiepileptic drugs. Since the approval of VNS therapy for epilepsy, clinicians have actively debated its role [134-137].

While further controlled studies are needed to more fully understand the safety, tolerability, and efficacy profile of VNS in children and in patients with generalized seizures, VNS is often used in these cases as well. In general, VNS is considered a valid treatment option for patients with well-documented medically-refractory seizures, who are either opposed to intracranial surgery, are not candidates, or whose medically-refractory seizures were not substantially improved by prior intracranial epilepsy surgery [51,77,78]. Resective surgery for appropriate candidates is preferred over VNS because of the substantially greater potential for complete seizure remission. (See "Surgical therapy of epilepsy in adults".)

While the initial cost of VNS implantation is high ($15,000 to $25,000), economic analyses suggest that VNS is associated with subsequent reductions in epilepsy-related direct medical care that may offset the cost of VNS within the lifespan of the generator [135,138-140].

Identification of factors that accurately predict a clinical response to VNS has been elusive [52,141,142]. While some studies have suggested that VNS may be more efficacious in individuals who have had refractory seizures for a shorter duration, other studies find rather that seizures with an onset later in age (greater than one year) have a better response to VNS [57,59,60]. Earlier age of epilepsy onset is also a predictor of medical intractability [143]. Further studies are needed to identify predictive factors associated with a response to VNS.

Typical initiation stimulation parameters are outlined in the table (table 1). These are then adjusted over outpatient follow-up to maximize efficacy and minimize side effects. More research is needed to determine whether any initiation stimulation settings provide more benefit than standard initial settings [26,144]

SUMMARY AND RECOMMENDATIONS — Vagus nerve stimulation (VNS) is effective, safe, and well-tolerated in patients with long-standing, refractory partial-onset seizures.

• Controlled trials in individuals greater than 12 years of age with partial-onset epilepsy indicate that 30 to 40 percent achieve a greater than 50 percent reduction in seizure frequency. Open-label studies suggest that these benefits may increase over time and adjustment of parameter settings. (See 'Partial-onset seizures' above.)

• Other open-label studies suggest that younger children and individuals with other seizure types and epilepsy syndromes can also benefit from VNS. (See 'Other seizure types and patient populations' above.)

• Side effects that occur during stimulation in a minority of patients are usually mild to moderate in severity and diminish with time or reduction in stimulation intensity. These include voice alteration and tingling. (See 'Common side effects' above.)

• Serious adverse events are rare and include infections, cardiac arrhythmias, and aspiration. (See 'Cardiac events' above and 'Other adverse effects' above.)

• Magnetic resonance imaging (MRI) in patients with VNS is limited to brain MRI under specific conditions outlined by the manufacturer. Diathermy is also contraindicated in patients with VNS. (See 'Magnetic resonance imaging' above and 'Other adverse effects' above.)

• While the role of VNS in the management of epilepsy is debated, it is a valid treatment option for patients with medically refractory epilepsy who are not candidates for resective epilepsy surgery. (See 'Clinical application' above.)