Vagus Nerve Stimulation in a Case of Epilepsy with CSWSS: Respiratory Side Effects during Sleep

We would like to alert the medical community to possible respiratory side effects of vagal nerve stimulation (VNS) during the treatment of nonsurgical refractory partial epilepsy. We observed hypopneic–polypneic episodes during sleep in a 9-year-old girl with epilepsy with continuous spike–waves during slow sleep (CSWSS) syndrome. These events were time-locked to each period of VNS stimulation and occurred without any modification of cardiac frequency. VNS may elicit a reduction of 25–50% in seizure frequency in both adults and children with intractable epilepsy 1. Adverse side effects and complications of VNS were few and mostly transient and were reported in adult patients only. They include hoarseness, neck pain, hypersalivation, cough, and shortness of breath (“air-missing”) sensations during physical exercise. These sensations might be related to the increase in end-expiratory volume observed in some patients receiving stimulation at high intensity. In contrast to the common observation that VNS modifies the respiratory pattern in animals, only one study has reported alterations in the respiratory pattern in humans 2. In the present case, the mother's pregnancy was marked by fetal death of a twin sister at 24 weeks of postconceptional age. Birth was uneventful at 36 weeks of gestational age. Systematic transfontanellar ultrasound analysis revealed bilateral diffuse periventricular leukomalacia. At age 2 years, she had right tonic partial seizures with loss of contact, and magnetic resonance imaging (MRI) evidenced a left parietal extended ischemic lesion. At age 5 years, her seizures changed and increased in frequency: during daytime, they were characterized by absence seizures associated with mild and inconstant atonia (up to one seizure every 2–3 min). Sleep EEG showed CSWSS. Over the course of the following year, she was treated unsuccessfully with seven different antiepileptic drugs (AEDs). A ketogenic diet also was ineffective. She was considered an ineligible candidate for surgery because of the extent of the left sylvian ischemic lesion. At age 6 years, a neurocybernetic prosthesis (Cyberonics, Houston, Texas U.S.A.) was implanted. The stimulation allowed a drug-free period of 2 months, during which there was an initial seizure-free period of 6 weeks. Then absences reappeared but were less numerous than before implantation. Subsequently she was given lamotrigine (LTG) with VNS. Despite the use of various AED combinations and changes in stimulation parameters over the next year, the absences persisted. At age 6½ years, while she was being treated with clonazepam (CZP) with VNS, a brief switching off of the stimulator resulted in the frequency of absence returning to preimplantation values (>100/day), indicating the relative efficacy of VNS in CSWSS. At age 7 years, the introduction of felbamate (FBM) and clobazam (CLB) with high-intensity VNS (intensity, 1.75 mA; frequency, 30 Hz; pulse width, 500 μs; and on-time/off-time 60 s/1.8 min) freed the child from seizures for 2 years until the present day. At age 8 years, an overnight polysomnographic recording included 20 EEG derivations, an electrooculogram, a submental electromyogram, a respiratory piezoelectric device encircling the ribcage, a nasal thermistance, and an electrocardiogram. The VNS signal was monitored by means of two electromyographic surface electrodes placed over the position of the vagal electrodes used for stimulation. The nasal thermistance moved during the recording and did not give reliable values over the whole recording period. Nevertheless, when both traces were recorded, the signal from the piezoelectric device was superimposable on that of the nasal thermistance, which indicated accurate ventilatory monitoring. The following reproducible markers were used in the data analysis: beginning of inspiration, peak inspiratory velocity (PIV), and peak expiratory velocity (PEV). From pilot evaluations, PIV and PEV did indeed reflect inspiratory and expiratory peak flows. The following time intervals were measured by using these markers: T1, interval between the beginning of inspiration and PIV; T2, interval between the PIV and PEV; and T3, interval between PEV and the beginning of inspiration. Instantaneous respiratory frequency was taken as (T1 + T2 + T3)/60. Breath-by-breath analysis was performed by using percentage changes, relative to control values, averaged from the three respiratory cycles immediately preceding the onset of the stimulus. Mean changes in ventilatory parameters were assessed by using values calculated within 10-s intervals. Eighty periods of VNS stimulation were examined for this purpose during sleep periods, free of body movement. Differences between control and stimulation periods were tested by using analysis of variance, followed by post hoc PLSD (Protected Least Square Differences) Fisher corrections (StatView+ package). Every period of VNS was tightly linked to a period of respiratory disturbance, which lasted for the entire period of stimulation. These respiratory modifications were characterized by an immediate decrease in the amplitude of both inspiratory and expiratory movements, as estimated from the piezoelectric trace and also as observed on nasal thermistance. At the same time, respiratory frequency increased (+16.0 ± 2.9%; p < 0.001) secondary to decreases in all the intermediate times (T1, –18.1 ± 3.7%; p < 0.003; T2, –14.8 ± 2.1%; p < 0.0001; T3, –20.4 ± 3%; p < 0.007). This early period lasted for ∼10 s. After this, the amplitude of respiration progressively returned to control values over the next 50 s. Nevertheless, respiratory frequency remained at a steady, increased value. It was only after a few seconds at the end of VNS stimulation that respiratory frequency and amplitude returned to prestimulation values. It should be noted that respiratory disturbances were observed throughout the sleep period, but were less easily distinguishable during wakefulness because of incessant movement of the child. In this child, the occurrence of CSWSS prevented any distinction between sleep stages within the EEG recordings, because interictal activity was not sufficiently suppressed to ascertain whether the child was in a rapid eye movement (REM) or non-REM (NREM) sleep period. No change was noticed in cardiac frequency (obtained from the RR interval), global EEG activity, or interspike interval that might be related to individual periods of VNS. Before implantation of the stimulator, daytime EEG presented a stereotypic pattern of paroxysmal activities. It consisted of a short run of theta waves intermingled with spikes over a 1-s period, followed by a period of 2–3 s relatively free of abnormalities. Then polyspikes occurred for 1 s, rapidly followed by spike–waves at 3 Hz. Spike–waves at a more variable frequency persisted for ∼2 s to several minutes, associated with a loss of contact. After implantation of VNS, in the different periods free from seizures, the stereotypic paroxysmal activities during daytime EEG were modified, being less numerous and more or less confined to the left hemisphere. The sequence of theta run followed by a period free of abnormalities persisted, but polyspikes or synchronized spike–waves at 3 Hz were no longer triggered. Instead, a period of spike–waves at very varied frequencies occurred for short periods, never exceeding 15 s. Our experience underscores the dual influence of VNS in a case of CSWSS syndrome. This is the first report of respiratory disorders tightly linked to VNS in a child with epilepsy. The changes in both respiratory timing and amplitude were very similar to those previously described in animals in response to electrical stimulation of small myelinated and unmyelinated vagal afferents 3. As these responses were elicited from the central cut end of the vagus nerve, they were assumed to result from direct interactions with respiratory control centers. They were marked by an earlier onset of both inspiration and expiration, resulting in shortening of both inspiratory and expiratory times, as we also observed. Moreover, the increase in respiratory frequency was associated with a reduction in the amplitude of the pneumogram, an alteration that appeared to be quite similar to that we observed at the beginning of VNS. Animal studies also raised the possibility that the reductions in respiratory flow and the velocity of thoracoabdominal distention might be due to a reflex restrictive respiratory episode. Indeed, the stimulation of the central end of vagus nerve increased the tonic inspiratory activity of the diaphragm, the intercostal, and the laryngeal muscles, which could result in stiffening the respiratory tract and the chest. Finally, the possibility of a direct activation of vagal efferents should not be definitely ruled out in the present case. Conversely, the present preliminary observation highlights the potential benefit of VNS in cases of epilepsy with CSWSS. Both the reduction in seizure frequency and the changes in EEG activities indicated that VNS improved the clinical status of the girl under study. As an additional argument, her symptoms immediately worsened during a temporary arrest of the stimulator. Nevertheless, it remains to be determined whether the improvements may be ascribed to VNS itself or to the combination of VNS and FBM. This deserves larger-scale evaluations to confirm the therapeutic efficacy of VNS in CSWSS syndrome. In addition, this case report appears sufficiently demonstrative to propose a systematic monitoring of respiratory parameters during overnight polygraphic recordings in patients treated with VNS. This should allow a detailed assessment of the risks of ventilatory disturbances, and their extent, within the range of the clinical applications of VNS. Extrapolating previous animal data to the present and other findings in humans raises the question whether the occurrence of discrete alterations of ventilatory parameters during VNS might provide a functional indicator of the activation of small myelinated and/or unmyelinated vagal fibers, which has been regarded as a contributing factor to the improvement of epileptic disorders. **We are greatly indebted to Dr. Plouin for her helpful comments and criticisms on this manuscript.