Background

The vagus nerve, or tenth cranial nerve, is composed of approximately 80% afferent visceral fibers³ transmitting visceral sensory information from receptors in the heart, aorta, lungs, and gastrointestinal system to the central nervous system (CNS). Accordingly, it is not surprising that VNS can affect CNS activity. A small number of fibers pass directly to the spinal trigeminal nucleus and the reticular formation, but most vagal afferents originate from nodose ganglion neurons with projections primarily to the nucleus of the solitary tract, which has widespread projections to the cerebral cortex, basal forebrain, hypothalamus, dorsal raphe, and cerebellum,⁴ and important areas for epileptogenesis such as amygdala, hippocampus, and thalamus.

There is also a direct pathway with viscerotopic representation to the insular cortex via the parabrachial nucleus in the pons.⁵ This is probably the pathway that relays sensation from vagal stimulation to conscious perception.

Other axons from the parabrachial nucleus go to the thalamic intralaminar and midline nuclei, which have extensive and diffuse projections throughout the cerebral cortex. This pathway may represent the point at which vagal stimulation interacts with pathways traditionally believed to control cortical synchronization and desynchronization.

The parasympathetic nucleus lies beneath the floor of the fourth ventricle and receives afferent input from the hypothalamus and glossopharyngeal nerve (carotid sinus). The efferent output supplies involuntary muscles of the bronchi, esophagus, stomach, small intestine, part of the large intestine to the distal third of the transverse colon, and heart.

The motor nucleus of the vagus nerve is formed by the nucleus ambiguus in the reticular formation of the medulla and receives input from both cerebral hemispheres. This nucleus supplies the constrictor muscles of the pharynx and the intrinsic muscles of the larynx.

In addition to the anatomic studies of the vagus nerve showing the vast number of cortical projections, animal studies have demonstrated effects on the EEG with stimulation. Electrical stimulation of the vagus nerve in cats was first shown to cause EEG changes in 1938 by Bailey and Bremer.⁶ Further studies have shown that, depending on the stimulation intensity and frequency, the cortical EEG can be synchronized or desynchronized.⁷ Low frequency (1 to 16 Hz) and very high intensity synchronize (increase slow waves), and higher intensity and faster frequencies (greater than 70 Hz) desynchronize (arousal, rapid eye movement), as do high intensity and slower frequency (20 to 50 Hz).^8–10 This suggests that different afferent fibers are stimulated under different circumstances, affecting different pathways and connections, with ultimately differing effects on the cortical EEG.

Animal Studies

Based on studies showing that the vagus nerve has a wide “connectivity” to the CNS,^11,12 various animal models have been used to investigate the effect of VNS on epilepsy. Some of the early studies in cats, dogs, rats, and monkeys showed reduction or prevention of seizures using a variety of chemically induced seizures. Study of the stimulation parameters required showed maximal anticonvulsant effect with C-fiber stimulation.^13–16

In addition, McLachlan¹⁷ demonstrated reduction or abolition of interictal spike activity in an acute epilepsy model using rats. Secondarily generalized seizures were also reduced in duration, but only if VNS occurred within 3 seconds of the start of the seizure. After 3 seconds, VNS had no effect on duration. This was also borne out in studies by Woodbury and Woodbury,^14,15 who showed that the longer the delay in stimulating the vagal nerve, the longer the duration of a seizure. In the same study, it was also shown that the effect on interictal spiking was seen 1 to 2 seconds after the onset of VNS and could persist for 1 to 3 minutes beyond stimulation; therefore, the anticonvulsant effect appeared to outlast the stimulation. Koo¹⁸ showed that during stimulation at the frequency of 20 to 30 Hz, there is synchronization of the epileptiform activity during the time when the stimulus is on, and suggested that there may be desynchronization of EEG results during the time when the stimulator is off. This may contribute to the mechanism of intermittent stimulation of the VNS as opposed to continuous stimulation.

Other parameters for optimal stimulation were assessed and shown in animal studies to be a stimulation of 10 to 20 Hz with a pulse width of 0.5 to 1 millisecond.^14,15,19 In practice, human stimulation is usually started with a pulse width of 0.25 millisecond (250 microseconds).

Human Studies and Efficacy

The first human implantation was in 1988, with FDA approval following five clinical studies, enrolling about 500 patients. These studies looked at efficacy, safety, stimulation frequency, and tolerability of VNS as an adjunctive treatment in refractory partial seizures in adults.

The data show that patients treated with high stimulation rates (≤5 minutes of “off time”) had a mean decrease in seizure frequency of almost 30% versus a mean decrease of 15% in the low-stimulation-rate group (180 minutes off, used as an “active control”). Results also show that the responder rate of at least a 50% reduction in seizure numbers was 30% in the group with the high stimulation rate. The responder rate in the low-stimulation-rate group was 13%, which suggests that VNS works at both rates but the higher stimulation rates are optimal. The use of such “active control” strategies has been critical in minimizing potential placebo effects when interpreting the results of this modality.^20,21

Long-term results from the first four studies reported a 95% continuation beyond the first year of implantation, with 82% and 69% beyond 2 and 3 years, respectively. Reduction in seizure frequency remained the same or improved over time, with pooled results showing a 40% reduction in seizures at 36 months. Thus it appears that the effects of VNS are cumulative without increased adverse effects.

There have been a few small studies suggesting that rapid-cycle (7 to 14 seconds on, 30 seconds off) stimulation may be effective in the patient who has not responded to slower cycling rates. This has shown some added efficacy in patients with Lennox-Gastaut syndrome and tuberous sclerosis.^22–24

A few studies have included children also.^25,26 The numbers are small, but most children had a more than 50% reduction in seizure frequency and a significant number had more than 90% reduction.

A small number of patients with generalized seizures have been studied. These studies suggest that these patients may respond better, with a higher responder rate to VNS than patients with partial seizures. Of the 25 patients with generalized seizure, 11 had a more than 50% seizure reduction.

Since FDA approval of the stimulator device in 1997, over 400 children, adolescents, and adults have received implants and have been studied at our institutions. The age ranged from 2 to 68 years, and most had refractory mixed seizures and generalized seizures, with a few with partial-onset seizures. The children with generalized seizures have shown a greater than 50% reduction on seizure number in more than 50% of the group. The children with partial seizures have shown a 30% to 40% reduction in number in approximately one third of the group. In addition to reduction in seizure number, seizure severity and the postictal period are dramatically decreased in a large number of patients. Finally, the level of alertness and interaction improves significantly in many of these children according to parents’ impressions.

Mechanisms of Action

Chronic VNS induces many physiologic changes in the brain. Naritoku and colleagues²⁷ have reported increases in neuronal fos expression with VNS. The increased activity was seen in areas of the medulla, hypothalamus, thalamus, amygdala, and cingulate with connectivity to the vagus nerve. The increased fos activity suggests increased neuronal activity.

Positron emission tomography scanning has also shown activation of CNS structures with VNS. Regional blood flow changes showing increases in the ipsilateral anterior thalamus and cingulate gyrus were reported by Garnett and co-workers,²⁸ and in another study, Ko and associates²⁹ found increased blood flow in the ipsilateral putamen and cerebellum with contralateral increases in the thalamus and temporal lobe.

In addition to reports of increased blood flow, concentrations of inhibitory neurotransmitters and amino acids have been shown to increase. The recent demonstration that magnetoelectroencephalography (MEG) can feasibly be performed on patients with VNS devices in place³⁰ may suggest a direction to collect data to further understanding of the mechanism of action. A coherent understanding of the mechanism of understanding, which has not so far advanced beyond the vague concept of “desynchronization,”³¹ may help advance appropriate use of the devices.

Patient Selection

Selection criteria for VNS are broad and continue to evolve. Selected patients should have medically refractory epilepsy. VNS is not a first-line treatment for epilepsy and is thus reserved for patients who have already tried multiple treatments. In addition, a potentially curative surgical resection should be considered preferable to VNS, when possible. It is highly desirable that all patients undergoing consideration for vagal nerve surgery be evaluated by epileptologists and surgeons well versed in other surgical options. For many patients, stimulator implantation may be preferable to extratemporal surgery in an eloquent area, corpus callosotomy, or repeat craniotomy for those who have failed surgery before. The use of the stimulator in cases where a more risky surgical approach or a less effective procedure is the only option gives an alternative treatment option with low morbidity. The last inclusion criterion is that the patient’s body size allows implantation of the device.

Seizure type is not an inclusion criterion. Studies have shown that the efficacy of VNS in generalized seizures and Lennox-Gastaut syndrome is comparable with that in partial seizures.²²A clear definition of the best patients for use of this therapy is unknown. Initial studies focused primarily on patients with intractable partial epilepsy. Approximately 20% of the patients studied in this group had a 50% or greater reduction in the number of seizures, with another 50% of the patients having a significant decrease in seizure frequency of at least 20%. Many patients also reported a decrease in seizure intensity, which is more difficult to quantify.

Patients who have previously undergone left cervical or bilateral vagotomies are excluded. In addition, patients who have significant preexisting upper airway/pharyngeal, pulmonary, cardiac, or gastrointestinal problems, presence of a dysautonomia, history of vasovagal syncope, or concurrent brain stimulator should be approached with caution and may need more frequent follow-ups.

Once the patient has received the implant, the stimulator is activated at any time from surgery to 2 weeks postsurgery. There have been no adverse effects when activating the device on the day of surgery. At our institution, the device is routinely activated by the surgeon at the time of implantation with initial settings as follows: 0.25 mA output, 20 Hz, pulse width of 250 milliseconds, on time of 30 seconds, and off time of 5 minutes. These minimal settings allow the patient to acclimate to the stimulation. Follow-up for reprogramming can then take place as frequently as once a week to once every few months.

The current stimulator device gives options for telemetrically changing the current intensity in milliamps, pulse width, length of time for stimulus train, and off time or time between stimulus trains. Reprogramming the stimulus output is usually the first parameter to adjust, and it can be increased in increments based on the patient’s tolerance. Other parameters that are adjusted are stimulus on and off times, which can be shortened or lengthened. Changing the on time to as little as 7 seconds and the off time to as little as 30 seconds has been shown to improve seizure control in patients with Lennox-Gastaut syndrome who have shown no improvement with the usual settings.²²

Follow-up examinations serve not only as an opportunity to reprogram and interrogate the device, but to begin adjustment of medications. Tapering of medications usually begins within a reasonable period after improved seizure control is observed. If improvement in seizure control is not seen, continued reprogramming and medication adjustment can be done.

Adverse Effects

The most common adverse effects are intermittent voice alteration with the “on” period of stimulation. This has been described as a hoarseness or vibration of the voice, with the severity depending on the intensity of stimulation. Other, less common adverse effects that have been described by patients are pharyngeal paresthesias, coughing, increased drooling, and sensation of shortness of breath, all of which tend to improve or resolve shortly after activation or increase in intensity during reprogramming. Very debilitated patients with little subcutaneous fat have a risk of skin breakdown over the device, and this should be considered a relative contraindication. The manufacturer supplies information suggesting that peptic ulcer disease and a history of certain cardiac problems should also be considered relative contraindications, although stimulation of the vagus nerve itself has not been reported to have an autonomic effect on heart rate or acid secretion in the stomach. This lack of side effects presumably relates to a lower threshold for stimulation of the afferent as opposed to the efferent neurons coursing through the vagus nerve.

Surgical Technique

Implantation of the device is a fairly straightforward procedure and can be done in patients of all ages, including young children. The vagus nerve in the neck is located in the carotid sheath, usually between the carotid artery and the jugular vein. The nerve is usually invested in the tissue of the carotid sheath between the carotid and jugular vein, but it can be located anterior, lateral, or posterior to the carotid artery and is sometimes more intimately associated with the jugular vein than the artery. There are usually no macroscopic branches of the nerve in the lower cervical regions below the carotid bifurcation region where the exposure is made, which is important because placement of the coil electrodes around the nerve requires circumferential dissection of the nerve, and it is crucial not to devascularize or damage any branches of the vagus.

In animal experiments, the right vagus nerve tends to innervate the cardiac centers responsible for bradycardia to a significantly greater extent than the left. For this reason, the left vagus nerve is traditionally selected to be the side for stimulation, unless there is some significant anatomic contraindication such as a prior dissection in the region.

The patient is positioned supine with the neck in slight extension and with the head turned slightly to the right. If the neck is turned too severely, greater retraction is required on the sternocleidomastoid muscle to allow for the dissection, and a moderate degree of rotation gives the best exposure to a suitable length of the nerve.

For an anterior axillary incision, the arm is positioned with slight abduction to allow access to the anterior axillary fold, but without having the arm so far away from the body that it is difficult for the surgeon to stand near the patient. A rolled towel beneath the patient’s scapulas allows the left arm to fall backward slightly, giving good access to the border of the pectoralis muscle. The surgeon stands on the left side of the patient with the assistant either across the table or just to the surgeon’s right.

Two skin incisions are made (Fig. 113-1). The cervical incision is made in the direction of a skin crease centered over the anterior edge of the sternocleidomastoid muscle approximately two thirds of the way down from the angle of the jaw to the clavicle. It is helpful to keep the exposure low enough to be well below the carotid bifurcation to ensure the most straightforward dissection of the nerve. Adequate exposure of the nerve can be obtained through an incision 2.0 to 2.5 cm long in a thin patient, but in the presence of a thick neck or prior surgery a longer incision should be made. We prefer to divide the platysma in the direction of its fibers along the anterior border of the sternocleidomastoid muscle, and then follow this anterior border in the avascular plane down to the carotid sheath. A Cloward retractor, usually with the shortest blunt blades, gives excellent exposure of the carotid sheath, and the vagus nerve is identified within the sheath.

FIGURE 113-1 Positioning and incision location for vagal nerve stimulator placement. The neck is extended with the head turned slightly to the right for exposure of the left vagal nerve low in the neck. A short incision (2–3 cm long) is located at the anterior border of the sternocleidomastoid in the transverse direction of a skin crease. The left arm is abducted and extended so that the anterior axillary region can be well prepared. A 4-cm incision lateral and inferior to the pectoralis muscle allows the insertion of the pulse generator. Making this incision in the direction of natural skin creases also improves cosmesis. During induction of anesthesia and positioning, a handheld programmer can be used to program the stimulator (right).

It is important to pay attention to the thickness and nature of the carotid sheath fascia while opening it on either side of the nerve, because this fascia is used for placing anchoring sutures to hold in the electrodes after they are placed on the vagus nerve. The nerve needs to be dissected clear of the sheath for a distance of approximately 3 cm, and this is facilitated by passing a plastic vessel loop around it and using it to elevate the nerve while using sharp dissection to open the sheath of either side of the nerve to free it up (Fig. 113-2). Blunt dissection is then used to open up space within the carotid sheath inferiorly for another 2 cm, because this will be used for a relaxing loop of cable to be fashioned later.

FIGURE 113-2 Exposure of the vagus nerve by division of the carotid sheath on either side. A Cloward retractor with short smooth blades affords adequate exposure to isolate 3 cm of the nerve.

After the nerve is isolated, the subcutaneous pocket or subpectoral pocket is made (Fig. 113-3). Previously, an incision approximately 10 cm below and parallel to the clavicle was recommended, although the stimulator device can be easily implanted into the same pocket by making an incision just at the inferior and lateral border of the pectoralis muscle approximately 2.5 cm in length in the region of the anterior axillary line. We have further modified this technique to make the axillary incision more horizontal and slightly curved, to follow the natural skin creases. This results in improved cosmetic appearance of the scar. Dissection is carried down below all of the subcutaneous tissue and just above the fascia of the pectoralis muscle itself, if a subcutaneous pocket is desired, or deep to the pectoralis for subpectoral placement. The device is relatively elongated in shape, and with practice the surgeon can readily determine when the pocket is large enough to admit the stimulator device in relation to the size of his or her hand through an incision less than 2.5 cm in length. It is useful to have the pocket deep enough that the device is recessed 1 to 2 cm away from the skin incision.

FIGURE 113-3 A pocket deep to the pectoralis major muscle is created by sharp and blunt dissection, or in some patients with significant breast tissue or subcutaneous fat, a pocket superficial to the muscle may be preferred.

After the pocket is fashioned, the tunneling device can be passed in either direction between the pocket and the surgical incision. We usually do this from the pectoral pocket to the cervical incision above (Fig. 113-4), with care taken to avoid injury to the exposed deep structures and with the ideal level of emergence of the tunneler just below the platysma.

FIGURE 113-4 The electrode cable is tunneled the short distance from the pocket to a plane below the platysma but above the sternocleidomastoid. The one-pin contact of the Model 302 electrode fits snugly in the inner sheath of the tunneler.

At this point, the electrodes can be passed through the subcutaneous tunnel that has been created. It is important that the coil electrodes themselves not pass through the tunnel because they can be damaged with minor trauma. The electrode connectors can be passed into the plastic sheath of the tunneling device and then pulled from above to below without difficulty. The current tunneler allows placing the electrode contact within the inner tunneling tube, which is pulled through the outer tube.

The next step involves putting the coil electrodes onto the vagus nerve (Fig. 113-5). Although this is conceptually simple, it can be frustrating if an attempt is made to put them on in the wrong orientation. We have found the simplest technique is to pull the cable inferiorly somewhat so there is no redundant cable below the coil devices. Starting with the lowest of the three coils (which is really just a retaining coil, and not an electrode), fine threads coming from the two ends of the Silastic coil are grasped and the superior of these is passed from lateral to medial, going deep to the vagus nerve. This is helped by having the assistant retract the nerve gently upward using the plastic loop. If the nerve is allowed to lay in a middle groove of the slightly opened coil, bringing the upper thread laterally and the lower thread medially across that nerve completes one full revolution of the coil, and then releasing the coil causes it to snap back into a good position around the nerve. Usually one to one-and-a-half more revolutions of the helix need to be completed around the nerve, but this is relatively easily done. Each of the two other coils can then be placed around the nerve, and because the superior coils that contain electrodes are shorter and have fewer completed coils than the lower retaining coil, they often go in somewhat more easily. If the nerve is not dissected for an adequate length to begin with, the last coil may be more difficult because of the lack of adequate space.

FIGURE 113-5 Coiling the helical electrodes around the vagus nerve. A, A close-up view of the vagus, which is separated from the carotid artery and sheath but without elevation by the vessel loop. B, With the lowest anchoring helix already in place, the superior part of the middle electrode is passed behind the nerve from lateral to medial and is grasped only by the supporting threads. C, Crossing the ends of the helix over the nerve concludes one-and-a-half revolutions of the helix placement.

There have been some recent changes in the design of the cables (models 303 and 304 from the manufacturer), with the most recent model (304) returning to a flexibility similar to the earlier leads. Experience with the intermediate leads (303), necessitated minor changes in technique (such as wrapping the most superior leads first rather than the most inferior), but with a return to more flexible leads this change in order is less necessary. Each surgeon should experiment with the leads available to determine the best strategy and order for wrapping the helices.

Once the three cables are in place, the electrode wires leading away from the cables are secured in the neck. The wires leading away from the two electrodes are separate for a distance of several centimeters and then joined together at a bifurcation point. The separate cables can be slid down into the carotid sheath so that they are in line with the coil electrodes, which have already been placed. The point of bifurcation then becomes a good place to secure the cables using a small plastic wing anchor. This can be sutured to the deep fascia of the neck using a 4-0 braided nylon or silk suture and a fine needle. Locating this deep fascia around the carotid sheath is important because suturing these anchors to any muscular fascia causes them to move when the patient has neck motion, and the aim is to keep them anchored with as little motion relative to the vagus nerve as possible. It is also important to keep the anchors away from the large vascular structures, including the carotid artery and jugular vein, to prevent erosion into these structures with late vascular complications. Two such tie-downs should be placed, one at the lower and one at the upper limit of the exposure of the deep cervical fascia. A third tie-down should then be placed after some additional free cable is passed upward above the exposure but deep to the platysma, so that it can be fastened onto the fascia of the sternocleidomastoid muscle deep to the platysma and at a level where it will not interfere with the incision, either above or below the skin incision. All of these anchoring sutures should be completed before any manipulation of the more distal part of the electrode cable to ensure that the electrodes on the nerve do not inadvertently come off of their siting.

At this point, the connector cable can be placed into the connection socket on the stimulator device. In the current design (Model 103), only one cable is used, but for replacements with the older, two-contact electrode cables it is important to note which cable is positive and which negative. The positive cable is marked with a white band and goes into the stimulator at the level of a port marked with a plus sign. When inserting the cables, a set screw may block the socket if the cable connector does not insert smoothly. Simply backing out the set screws solves this problem. Once these are in place and the set screws are tightened, the electrical integrity and correct impedance of the nerve electrodes can be tested with diagnostics supplied by the computer (Fig. 113-6). This is done by placing the wand of the computer (along with the hand-held programmer, if this is used instead of a laptop computer for programming) in a sterile plastic drape, such as a clear x-ray cassette cover, and passing it onto the field. The device can then be tested and even activated. We turn on the device at the time of placement as noted above, but some clinicians prefer simply to leave it off for a few weeks to be certain that any postoperative symptoms the patient is having can be distinguished from symptoms related to stimulation.

FIGURE 113-6 Lead test follows fixing the cable to the generator and placement of the generator in the pocket. Normal electrical impedance tested by diagnostic routine of the generator serves as a final test of cable integrity.

The pulse generator is then inserted into the subcutaneous pocket and secured using a single 2-0 silk stitch to the pectoralis fascia, which keeps it anchored deeply into the pocket. It is useful to have the logo of the device facing toward the outside, which optimally orients the antenna for communication with the wand, especially in patients with thicker subcutaneous tissues. Once this is secured in place, both incisions can be closed with resorbable sutures to the deep fascia in the pocket and the platysma in the neck, followed by subcutaneous closure and subcuticular closure with paper strips to close the edges of the skin. Dry sterile dressings are placed, and we usually instruct patients to leave them on for 7 to 10 days to keep the incision completely dry (Fig. 113-7, Table 113-1).

FIGURE 113-7 Completion of surgery with dressings in place. In a pediatric patient, the stimulator makes a noticeable bulge in the contour of the skin with a subcutaneous pocket, but less so when the device is subpectoral.

Conclusion

Vagal nerve stimulation is a novel approach to treatment of refractory epilepsy in pediatric and adult patients. The spectrum of seizure type that has been shown to respond to chronic VNS includes both partial and rapidly generalizing seizures and epilepsy syndromes such as Lennox-Gastaut syndrome.