Right-sided Vagus Nerve Stimulation as a Treatment for Refractory Epilepsy in Humans


Address correspondence and reprint requests to Dr. J. Wheless at Texas Comprehensive Epilepsy Program, 6431 Fannin Street, MSB 7.044, Houston, TX 77030, U.S.A. E-mail: james.w.wheless@uth.tmc.edu


Summary: Purpose: We present three children who underwent right-sided vagus nerve stimulation (R-VNS). This treatment option for people with refractory epilepsy has not been described in children.

Methods: We reviewed our database of >350 patients implanted with vagus nerve stimulators and now describe our experience in three patients with R-VNS for the treatment of intractable seizures. All three patients improved dramatically with left-sided vagus nerve stimulation (L-VNS), but the devices had to be removed because of infection. The patients were thought to be at high risk for nerve injury if they were reapproached for L-VNSs; therefore R-VNSs were implanted.

Results: All three patients with an R-VNS had a reduction in seizures. Our first patient has had an R-VNS for 5 years; he has been seizure free for >2 years on R-VNS monotherapy. The second patient had an R-VNS for 8 months. His seizure control improved slightly, but not as dramatically as with L-VNS. The third child has had an R-VNS for >7 months and has cessation of his most disabling seizure type (generalized tonic–clonic seizures). None of the patients had cardiac side effects from therapeutic R-VNS. However, two of the three patients had respiratory events with R-VNS.

Conclusions: VNS is known to be an effective treatment in pharmacoresistant epilepsy. R-VNS should be considered if a patient has significant benefit from L-VNS but is unable to continue with L-VNS. R-VNS appears also to have antiepilepsy effects. Additionally, our case report suggests that in some patients, a differential response is found regarding seizure control with R-VNS or L-VNS, raising the question whether L-VNS failures should pursue a trial of R-VNS. Patients should be cautioned and monitored for reactive airway disease if they undergo R-VNS. More research is needed to compare the effects of right- and left-sided VNS on cardiac and pulmonary function in humans and to determine which has the best antiseizure effect.

Nonpharmacologic treatment options should be investigated for medically refractory epilepsy patients (1,2). If patients do not qualify for epilepsy surgery or are not good candidates, vagus nerve stimulation (VNS) is an option (3). VNS is approved in the United States as adjunctive therapy in patients older than 12 years with refractory partial-onset seizures (4); however, it is effective in younger children with refractory epilepsy (5).

Traditionally VNSs have been placed on the left side. The rationale for stimulating the L-VNS is based more on conventional wisdom than on sound scientific evidence. The VNS physician's manual states, “The NCP system is indicated for use only in stimulating the left vagus nerve in the neck area inside the carotid sheath,” and “The safety and efficacy of the NCP System treatment have not been established for stimulation of the right vagus nerve” (6). We report therapeutic right-sided VNS in three children without cardiac side effects. The patients improved dramatically with L-VNS, but the devices were explanted because of infection. All experienced a dramatic worsening of their seizure control. Reimplanting the L-VNS was not thought to be feasible from a safety standpoint (e.g., risk of injury to the vagus nerve). We hypothesized that R-VNS might be equally efficacious and as safe as L-VNS. We review our experience with R-VNS.


The same surgeon implanted the devices in all the children. We reviewed our database of slightly >350 patients and describe the clinical course of the three who underwent R-VNS.

Case 1

Our first patient, a now 11-year-old boy, has symptomatic mixed generalized seizures, moderate-to-marked cognitive impairment, hypotonic cerebral palsy, and dysphagia presumed secondary to in utero drug exposure. Seizure onset was at age 1 year, and he had initial daily generalized tonic, then myoclonic, astatic, and atypical absence seizures. At the time of his initial VNS placement, treatment with 10 antiepileptic drugs (AEDs) and the ketogenic diet had not resulted in seizure control. Interictal EEGs showed near-continuous generalized sharp-and-slow wave and spike-and-slow wave discharges at 0.5 to 2 Hz, with a shifting asymmetry, multifocal spikes, and diffuse slowing of the background. Magnetic resonance imaging (MRI) was normal. An L-VNS was placed on May 6, 1999 (age 6½ years; Fig. 1). VNS parameters used with all of his devices were 7 s on, 0.2 min off, 20 Hz, and 250-μs pulse width. Stimulation intensity varied between 0.75 mA and 1.5 mA, depending on the length of time the generator was in place before removal. Three weeks after implantation, the generator was removed because of infection with Staphylococcus aureus.

Figure 1.

Case one—Seizure and AED history. *= Start and end of ketogenic diet; L1, L2, L3 = Left VNS implantations; E1, E2, E3 = Left VNS implant removals; R1 = Right VNS implantation.

The seizures dramatically increased without VNS therapy to the preimplant baseline. Two other L-VNS implantations had the same fate, requiring removal. He was seizure free during periods of stimulation.

At this point, the patient had an L-VNS removed 3 times because of skin breakdown and infections. This resulted in a total of six surgical procedures in the usual placement site (left vagus nerve). He had no clinical evidence of damage to the left vagus nerve, but a significant concern existed that he was at high risk for injury if the left vagus nerve site was reapproached. This created a clinical conundrum, as he was seizure free and had improved cognition and quality of life with L-VNS. A decision was made to pursue R-VNS. A baseline electrocardiogram (ECG) was normal. On November 16, 2000, an R-VNS was placed with continuous ECG monitoring. A standard lead test and device diagnostics were performed in the operating room without problems. Holter monitoring performed the next day was normal. Repeated Holter monitoring during active VNS in February 2002 was normal (Fig. 2). His device was turned on 2 weeks after surgery, increased to 0.75 mA within 6 weeks, and he was seizure free at this time. At his last visit in October 2003, the patient was seizure free, off all medication. His last seizure occurred in November 2001.

Figure 2.

Case one—Stimulation history for the right-sided VNS implant. Stimulation parameters were set at 250 microseconds, 20 Hz, with a duty cycle of 7 seconds ON and 0.2 minutes OFF for the duration of the implant; *= Dates of Holter monitoring.

Case 2

Our second patient, a now 12-year-old boy, began having seizures at age 3.75 years. He initially had drop attacks, followed 1 to 2 weeks later by emergence of generalized tonic seizures. Since then, clinical semiology has developed consistent with generalized tonic-clonic (GTC), complex partial, and atonic seizures. The complex partial seizures are usually brief (10–30 s in duration) but occur multiple times a day; the GTC events were infrequent, and his drop attacks occurred daily to weekly. He has been treated with 11 AEDs and the ketogenic diet without seizure control. An L-VNS was placed in June 1999. His complex partial seizures decreased from 50 to 100 seizures per day to rare events (one per month), and his GTC and atonic seizures stopped. This dramatic improvement in seizure control was maintained until the early summer of 2002, when his seizure frequency increased again to multiple daily events. Investigation revealed a fractured lead wire. The patient's seizures improved dramatically after the VNS generator and left-sided lead wire were changed on August 23, 2002. An infection at the neck-incision site required L-VNS removal in October 2002. After the removal of the L-VNS, he experienced five to six generalized tonic seizures per night, 50 to 100 brief complex partial seizures per day, and daily to weekly atonic seizures. The patient had experienced significant improvement with L-VNS, but concern was expressed that he was at risk for nerve injury if he had a fourth operation on the left. An R-VNS was implanted on December 13, 2002. An ECG rhythm strip was normal during R-VNS placement and lead test. Holter monitoring performed December 16, 2002, was normal, and R-VNS therapy was initiated (7 s on, 0.2 min off, 20 Hz, 250-μs pulse width). The patient's drop attacks decreased significantly, and his GTC seizures decreased slightly in frequency and severity. R-VNS intensity was increased as tolerated to 1.5 mA; however, he continued to have multiple daily complex partial seizures. Exercise-induced reactive airway disease developed, which was characterized by wheezing, shortness of breath, and secondary tachycardia. This would occasionally occur at rest. An inhaled anticholinergic agent, ipratropium bromide, did not provide relief.

Because of persistent seizures and respiratory symptoms, he was admitted to our Epilepsy Monitoring Unit in August 2003 for repeated evaluation. Video-EEG monitoring confirmed that he was having ∼50 brief complex partial seizures per day. The complex partial seizure semiology consisted of an aura (sensation of body tremor) progressing to cessation of ongoing activity, staring, and eye blinks, lasting ∼10–20 s. His ictal EEG showed transient involvement from the entire right hemisphere with rapid progression to generalized patterns. The patient was taken to the operating room on August 21, 2003. The R-VNS was left in place, and output was set to 0 mA. An L-VNS was implanted, therapy initiated, and rapidly increased over a 48-h period to 1.75 mA, 7 s on, and 0.2 min off. Over the last 5 months since the surgery, the patient's seizures have again dramatically improved. He has had no drop attacks. His complex partial seizures have dramatically improved (fewer than one a day). His GTC seizures are briefer (most seizures have lasted <30 s) and occur only in sleep (Fig. 3). Using the magnet during a seizure is effective. The patient has had no respiratory symptoms since the surgery and L-VNS. He is taking no AEDs.

Figure 3.

Case two—Seizure and VNS history. L-VNS = left-sided VNS; R-VNS = right-sided VNS.

Case 3

Our third patient is an 8-year-old boy with intractable, symptomatic mixed seizure disorder since age 5 months secondary to a congenital brain malformation (bilateral pachygyria with thin corpus callosum). In addition, he has hypotonic cerebral palsy and moderate-to-marked cognitive impairment. He has undergone treatment with eight AEDs and the ketogenic diet without success. He had daily complex partial, astatic, and GTCc seizures. He underwent placement of an L-VNS on November 15, 2002 (age 7 years). The device was removed 11 days later because of infection, and another was placed on January 8, 2003. This device also became infected and required removal 12 days after implantation. In light of the multiple L-VNS infections, an R-VNS was placed on April 10, 2003. An intraoperative ECG during the lead test was normal. R-VNS therapy was initiated in the operating room (7 s on, 0.2 min off, 20 Hz, 250-μs pulse width, 0.25 mA). A 24-h ECG performed on postop day 1 was normal. At an office visit 3 months later, the patient's mother reported that he occasionally sounded raspy and once had wheezing after swimming. This was transient, lasting only a few days, and resolved without therapy. His ECG in July 2003 was normal, and his R-VNS was increased from 1 mA to 1.25 mA. During the patient's most recent visit on October 21, 2003, the patient's mother reported that he has not been having any GTC seizures since placement of the R-VNS. However, he continues to have almost daily myoclonic and astatic seizures. His VNS was increased to 1.75 mA (7 s on, 0.2 min off, 250 μs, 20 Hz) without side effects.


All three of our patients required R-VNS after experiencing removal of their L-VNS because of infection. All of our patients had their devices placed by an experienced neurosurgeon in the operating room, with standard operating precautions (i.e., prophylactic antibiotics) to minimize infections. However, even with the best clinical care, infections occur in 3–6% of all implants (7,8). The concern that further attempts at L-VNS were not safe, because of the possibility of nerve injury, led us to consider R-VNS in these patients who had shown improvement with L-VNS.

The use of R-VNS as a treatment of pharmacoresistant epilepsy has not previously been described. All three of our patients with R-VNS had some reduction in seizures. Our first patient became seizure free (case 1), and another has cessation of his most disabling seizure type (GTC, case 3). Our first patient had an identical degree of seizure control with L- or R-VNS, but our second patient showed a differential response between L- and R-VNS. It is difficult to compare the efficacy of L- and R-VNS in the third patient because the L-VNS was implanted for <2 weeks each time. A recent report of R- and L-VNS in rats demonstrated equal effectiveness in reducing generalized seizures, suggesting that the fibers necessary for seizure suppressant are not unique to the left vagus nerve (9). This study was performed after our current series in children, and we know of no reports comparing the efficacy of R- and L-VNS in humans. The differential response with regard to seizure control seen in our second patient raises the question as to which vagus nerve, when stimulated in a given patient, may lead to their best seizure control. The current recommendation is that L-VNS be performed. However, the differential response we observed raises the possibility that some patients may respond better to stimulation of one vagus nerve over the other. What is not clear is whether L-VNS gives the best response in all patients. Future studies, placing R-VNS in patients who have not responded to L-VNS (<25% seizure decrease from baseline) may be helpful to answer this question.

None of the three patients with R-VNS had any cardiac complications. To explain why the VNS is placed on the left, early reviews and clinical trials commented that autonomic neural input to the heart exhibited some degree of “sidedness,” with branches from the right vagus nerve predominantly, but not exclusively, innervating the sinoatrial (SA) node (cardiac pacemaker region), whereas branches from the left vagus nerve predominantly innervate the atrioventricular (AV) node (10,11). These authors reasoned that L-VNS would be less likely to cause significant cardiac side effects because of less innervation of the SA node. This asymmetric cardiac innervation by the left and right vagus nerves was based on stimulation studies performed in dogs (12–16). On average, R-VNS evoked a significantly greater bradycardia, although in a few animals, L-VNS induced sinus asystole. [This later finding may correlate with the rare sinus bradycardia or asystole observed in humans with L-VNS (17,18)]. Cricket et al. (19), in a review of the neural supply of the human heart, stated, “In the human heart these neural projections have yet to be clearly defined.” Subsequently, L-VNS was shown in pigs and humans to induce a significant negative inotropic effect on the left ventricular myocardium, independent of its bradycardic effect (20). Complicating this decision further (as to which vagus nerve is the safest to stimulate in humans) is a report by Matheny et al. (21) documenting that left-sided thoracic vagal stimulation can be used to cause cessation of the heartbeat, with termination resulting in a normal sinus rhythm. This immediate stimulation-induced asystole produced by thoracic L-VNS was subsequently verified by Lewis et al. (20) in 10 patients. These newer studies do not clarify which vagus nerve is the safest to stimulate in humans.

Despite the findings in canine models of VNS, reasons exist to believe that stimulation of the R-VNS in humans, by using the same surgical technique and parameters of L-VNS, would not cause significant cardiac effects. First, concerns about the cardiac effects of R-VNS are based on animal studies, which show inconsistencies based on the animal model (9,20,22,23).

Second, physiologic reasons may allow stimulation of the right vagus nerve in humans without producing significant cardiac effects. Activation of the unmyelinated, high-threshold C-fibers is necessary to induce bradycardia (24) but is not necessary for the efficacy of VNS (25,26), or probable to achieve with clinically used stimulation parameters (26–28).

Finally, it is recommended that surgeons place the electrode distal to the superior cervical and the inferior cardiac branches of the vagus nerve (29). This initial instruction was based on the hypothesis that an electrode placed below the inferior cardiac branch would not stimulate vagal cardiac fibers. This should be true regardless of the side of stimulation. With this strategy, initial trials did not reveal any cardiac effects of L-VNS (7,11,30,31). However, subsequent reports of ventricular asystole during the intraoperative lead test occurred (17,18), and as of October 2003, 47 reports existed of asystole or bradycardia occurring in adults among 24,640 patients (1:524) implanted with L-VNS (32). This may reflect individual anatomic variation of the inferior cardiac branch of the vagus nerve, resulting in inadvertent proximal electrode placement. Again, if this anatomic variation is the reason some L-VNS patients experience cardiac side effects, it is possible that the same rare event could occur with R-VNS.

Interestingly, although approximately one third of all VNSs are implanted in children younger than 16 years, no episodes of asystole or bradycardia have been reported in children or in our three patients (32). This may reflect developmental differences in the vagus nerve or the susceptibility to cardiac effects. Two of our three patients with R-VNS had respiratory side effects. Respiratory symptoms including dyspnea and worsening of asthma and bronchitis are rare, potential adverse events with L-VNS (33). Morris and Mueller (34) reported breathlessness in only 3.2% of patients with therapeutic L-VNS stimulation only during physical exertion and stimulation. L-VNS has been associated with a stimulation-dependent decrease in FEV1 (forced expiratory volume in 1 s) in one patient with obstructive lung disease (35). Long-term L-VNS did not induce any significant change in FEV1 in any patient (N = 8) without concomitant lung disease (35). The effect of R-VNS on respiration has not been studied in humans. Our experience suggests that bronchoconstriction can occur with stimulation of either vagus nerve, although it may be more common after R-VNS. None of our patients had prior history of reactive airway disease or recurrent pulmonary infections.

Animal experiments do not clarify which vagus nerve, when stimulated, produces the least respiratory effect and give differing results depending on the animal model studied (36–39).

Pulmonary studies in guinea pigs suggest that capsaicin-sensitive fibers may be responsible for producing bronchoconstriction (38). This mechanism could explain why our second patient did not respond to treatment with ipratropium bromide. This lack of response to ipratropium bromide also was noted after L-VNS by Lötvall (35). However, the exact mechanism of bronchoconstriction in our patients is not clear, as formal pulmonary-function studies were not able to be obtained.

In conclusion, R-VNS should be considered in a patient who has significant benefit from VNS but is unable to continue with L-VNS. Furthermore, our second patient clearly demonstrates that in some people with intractable epilepsy, a differential response may be seen with VNS. This patient had a dramatically better response to L-VNS than to R-VNS. However, this raises the possibility that some patients may have a better response to R- than L-VNS. This review of our experience with R-VNS raises several questions: (a) Should R-VNS be offered to patients who do not show a good response to L-VNS? and (b) Could bilateral VNS be more effective than unilateral VNS? Further studies addressing these questions will be performed in the future.