A Neural Prosthesis for Obstructive Sleep Apnea

Abstract

Obstructive sleep apnea (OSA) is generated by a collapse of the soft tissue and tongue in the oropharynx that can block the airway passages during sleep and affects nearly 4% of the population. The most common nonsurgical treatment for OSA, Continuous-Positive-Airway-Pressure (CPAP), is highly effective; however, it is not tolerated by a large number of patients. Surgical procedures have been developed but they are very invasive and not always effective. However, studies have shown that electrical stimulation of the hypoglossal nerve can clearly improve the patency of the airways in patients with OSA. These prosthetic devices have the potential to increase significantly the quality of life and prevent the development of the many severe comorbidities associated with OSA. However, there are limitations to their benefits and proper patient selection is essential.

Keywords

Electrical stimulation, Genioglossal, Hypoglossal nerve, Obstructive sleep apnea

Outline
Introduction 1321
Control of Airway Patency Through Tongue Muscles 1322
Prosthetic Design for Obstructive Sleep Apnea (OSA) 1323
Single Electrode Closed-Loop Prosthesis Design for Obstructive Sleep Apnea 1324
Obstructive Sleep Apnea Prosthesis With Selective Stimulation 1326
Clinical Implementation 1328
Conclusion 1328
Acknowledgment [CR]
References 1328

Acknowledgment

I would like to thank Saifur Rashid for editing this manuscript.

Introduction

Obstructive sleep apnea (OSA) affects 2%–4% of the adult population and is most commonly seen in middle aged, overweight men. A study at the University of Wisconsin showed that 4% of men and 2% of women, aged 30–60, have undiagnosed sleep apnea ( ). A 2004 report from MedTech Insight indicates that the prevalence of OSA in the United States is 28.2 million patients. Of these, 12, 8.8, and 8.1 million have mild, moderate, and severe OSA, respectively ( ). The degree of severity is determined by the apnea-hypopnea index (AHI). The AHI is the number of apneic and hypopneic episodes per night, and can reach up to 60 for patients with severe OSA. Of the 16.4 million patients with either moderate (AHP > 15) or severe OSA (AHI > 50), 15.6 million patients remain untreated. These patients develop upper airway occlusions ( Fig. 109.1 ) related to the prolapse of the tongue and its surrounding structure in the pharynx. This prolapse has been attributed to diminished neuromuscular activity in the upper airway dilating muscles ( ) during an occlusion as indicated by electromyogram (EMG) recording from the main protruder of the tongue, the genioglossus muscle. During wakefulness, OSA patients have an augmented genioglossus activity compared to patients without obstructive sleep apnea. However, this neuromuscular compensation may be lost during sleep thereby generating a collapse of the upper airway ( ).

OSA is also associated with arterial oxygen desaturation and subsequent arousal from sleep ( ). Referred to as the snorer’s disease, it has been suggested that OSA involves a gradual degeneration of the upper airways (UAW) mucosal receptors in addition to a progressive deposition of lipid tissue in the UAW lumen ( ). This gradual loss in muscle tone and inherently narrow UAW anatomy, both of which facilitate larger negative intraluminal pressure swings during inspiration, are considered the main factors that predispose individuals to OSA.

There are other several predispositions to OSA such as obesity ( ), ethnicity ( ), pharyngeal wall thickness, enlargement of the tongue, and retroposition of the mandible and/or hyoid bone ( ). The frequent and repeated nocturnal episodes of occlusion produce microarousals and lead to excessive daytime sleepiness (EDS). Aside from the chronic fatigue associated with EDS and increased risk for automobile accidents, OSA patients exhibit a greater likelihood for developing more serious long-term pathologic sequelae including hypertension, right-sided heart failure, arrhythmias, and stroke ( ).

Initial treatment of OSA involves a series of lifestyle changes followed by nonsurgical treatment options. The most common nonsurgical treatment is Continuous-Positive-Airway-Pressure (CPAP). It has been shown to be effective in reducing the symptoms of OSA in patients who use it on a regular basis ( ); however, there is a high rate of noncompliance (∼40%) ( ) mostly due to patient discomfort.

Surgical procedures for the treatment of OSA include opening one or more of the sites of breathing obstruction, via adenoidectomy, tonsillectomy, nasal polyp removal, airway abnormality correction, uvulopalatopharyngoplasty (UPPP), or, in the most severe cases, tracheotomy or surgical jaw reconstruction, nasal airway surgery, palate implants, tongue reduction, genioglossus advancement, hyoid suspension, maxillomandibular procedures, bariatric surgery, or combinations of these procedures. Many of these procedures are very invasive and not always effective. The most common surgery for OSA patients (UPPP) is initially successful but the success rate drops to 46% after a 13-month period ( ).

However, the physiology and anatomy of the UAW muscles suggest that the UAW patency could be preserved during sleep through electrical stimulation of the tongue muscles ( ). Stimulation of the hypoglossal nerve has been shown to coactivate the tongue protrudor and retractor muscles resulting in airway clearance ( ). This conclusion is based on results obtained from both animal and human experiments.

Control of Airway Patency Through Tongue Muscles

Although there are many muscles in the upper airways that affect patency, the most important ones are controlled by the hypoglossal nerve. The hypoglossal nerve innervates the geniohyoid (GH) muscle, the intrinsic muscles of the tongue, and the extrinsic muscles of the tongue, i.e., the genioglossus muscle (medial branch), the styloglossus (SG), and hyoglossus (HG) muscles (lateral branch) ( Fig. 109.2 ). The genioglossus (GG) and the GH muscles are the primary ones involved in dilation of the pharynx. Contraction of genioglossus provides tongue protrusion, hence widening the pharyngeal opening. Activation of the GH along with a tone present in the sternohyoid muscle pulls the hyoid bone ventrally, thus again dilating the pharynx. On the other hand, HG and SG are considered tongue retractor muscles.

Several methods have been tested to activate the UAW muscles with electrical stimulation in OSA patients such as (1) stimulation of genioglossus using submental transcutaneous stimulators, (2) direct stimulation of genioglossus with wire electrodes, and (3) direct stimulation of the HG nerve.

Transcutaneous stimulation of genioglossus has given inconsistent results and sometimes failed to prevent obstructions without causing arousals during sleep ( ). The poor efficiency of transcutaneous stimulations can be attributed to nonspecific activation of genioglossus because of the tissue present between the stimulator and the target muscle. Direct stimulation of the genioglossus using intraoral wire electrodes was effective in dilating the upper airways in OSA patients ( ). However, this method does not lend itself to an implantable functional electrical stimulation (FES) device since it uses EMG wire electrodes, which cannot be implanted chronically.

Electrical stimulation of the hypoglossal (XII) nerve directly has been investigated as an alternative mode of therapy to compensate for the increased airway collapsibility observed in OSA patients resulting from diminished or insufficient nocturnal activity of UAW dilators ( ). Stimulation of the HG by percutaneously inserted wire electrodes provided tongue protrusion with minimal discomfort in humans yet terminated only 23% of the apneic events ( ). The inefficiency of the HG nerve stimulation with wire electrodes could be due to the inappropriate placement of the electrodes resulting in the recruitment of retractor muscles before the protruder muscles of the tongue. It has been demonstrated in humans by another group that the flow of inspired air is doubled by stimulation of the main branch of the hypoglossal nerve ( ). Stimulation of the medial branch was nearly as efficient and was superior to stimulation of other branches. However, human experiments suggest that it might be easier to prevent an obstruction rather than opening the airways during an obstruction ( ).

In both animal and human experiments, subjects have shown significant improvements in UAW resistance (R _UAW) and stability (P _crit) in response to electrical stimulation ( ; ). P _critis the critical pressure in the UAW capable of inducing flow limitation. A low P _critindicates an UAW resistant to collapse. Reduction of the AHI is associated with a decrease in P _crit( ).

Although long-term studies in OSA patients have demonstrated significant decrease in AHI, an effective neuroprosthetic design for OSA has not yet been developed.

Prosthetic Design for Obstructive Sleep Apnea (OSA)

Since it has been shown that electrical stimulation of the UAW muscles which control the tongue can prevent the collapse of upper airways and decrease the AHI ( ), one could consider the design of an implantable prosthetic device to maintain patency by stimulation of the hypoglossal nerve. Three methods of stimulation have been considered. The first involves open-loop continuous stimulation (modulation) to maintain tone in the muscles thereby preventing a collapse. The fact that the stimulation levels required to open the airway are below the threshold for awakening the patients suggests that this method might work. Another method is also open loop but with intermittent stimulation whereby electrical stimulation is applied during inspiration at the natural respiratory frequency and duty cycle of the patient. The stimulation and the patient’s respiration is not synchronized, but the patient could learn to breathe only during the time when the stimulator is activated. Such a method was proposed and tested using a Bion ( ). The Bion is a small implantable stimulator that was injected with a syringe close to the medial branch of the hypoglossal nerve from the mouth. Continuous stimulation was applied to prevent the collapse of the upper airways. Although initially promising, this method was not pursued.

A third method involves synchronization of the stimulation with inspiration. The fact that synchronization is important is suggested by studies showing that once the collapse of the upper airways has occurred, it is difficult, even with a strong stimulus to restore patency ( ). This synchronization allows the stimulus to restore tone in the upper airways muscles when the upper airway is most sensitive to a collapse generated by the negative pressure generated during inspiration. A closed-loop synchronous prosthesis was developed by Medtronic. Fig. 109.3 shows the design of the device designed to relieve obstruction. A sensor is placed around the diaphragm (11, 12) to sense the respiration effort by impedance plethysmography. Stimulation is applied through an electrode positioned on the hypoglossal nerve (4). The stimulator (1) senses the effort and can synchronize the stimulation to the respiration signal ( ). Other sensing devices have been proposed and tested such as a sensor capable of detecting intrathoracic pressure and allowing the stimulation of the hypoglossal nerve to be synchronized with inspiration ( ).

A prototype of this device was tested in a chronic study involving eight patients with OSA. A C-shape electrode ( Fig. 109.4 ), stimulator and pressure sensor were implanted for 6 months ( ). Stimulation was synchronized with respiration and applied at 33 Hz, with 91 μs pulses. The results show that the stimulation was well tolerated and did not produce any significant adverse effects. The AHI decreased significantly during both random eye movement (REM) and non-REM sleep. Both the quality of sleep and the oxygen saturation were improved in these patients.

An acute study to test the device was carried out in 14 patients ( ). Electrical stimulation of the hypoglossal nerve was compared to direct muscle stimulation. Five patients were implanted with a C-shaped cuff electrode shown in Fig. 109.4 . Nine patients were implanted with fine wires in the genioglossus muscle. The ability of the stimulation to maintain patency in the airways was quantified by measuring the critical pressure capable of inducing flow limitations (P _crit) and the AHI. In both sets of patients, P _critwas significantly decreased by the application of the stimulation and the decrease in this critical pressure was also accompanied by a decrease in AHI. The results also indicated that the greatest improvement with hypoglossal stimulation was found in patients with low (subatmospheric) P _crit.

Taken together, these results suggest that electrical stimulation of the hypoglossal nerve can clearly improve the patency of the airways in patients with OSA. Given the fact that so many patients could benefit from this technology, it is unclear why such a device has not yet been deployed in large numbers. One difficult issue is the fact that synchronization may be important since improvements of the AHI were dependent on the degree of synchronization ( ). However, the detection of the respiratory signal requires an additional sensor and the synchronization has been difficult to achieve. This difficulty arises from the fact that the stimulation should be applied just before inspiration in order to maximize the effect of the stimulus. However, the respiratory cycle is not very regular, particularly during REM sleep, making the synchronization difficult.

Single Electrode Closed-Loop Prosthesis Design for Obstructive Sleep Apnea

Another approach to the design of a prosthesis for OSA is to use the same electrode for stimulation and recording. The hypoglossal nerve contains mostly motor efferents for the tongue muscles. Therefore, the activity of the nerve should reflect the attempts of the nervous system to open the airway during breathing (or during occlusion) and produce a detectable signal. In particular, hypoglossal activity should precede phrenic nerve activity to open the airways. Moreover, the activity should increase during an obstruction. To test that hypothesis, recording electrodes were placed on the hypoglossal nerve in two dogs and recording were obtained in a chronic preparation for over a year ( ). Obstruction was generated by applying a force to the submental region during sleep.

Fig. 109.5 shows an example of signals recorded during non-REM sleep. The applied force was applied with a ramping waveform. The corresponding increase in esophageal pressure is shown below the force waveform (Pes). The rectified and integrated HG activity is also shown and indicates that during each breath, a small spike of activity in HG activity is detectable just before inspiration. The amplitude of each spike of these increases significantly with obstruction. Therefore, the hypoglossal nerve activity is related to breathing and obstruction indicating that the electroneurogram (ENG) from the hypoglossal nerve could be used as a control signal for the stimulator activating the same electrode. This closed-loop system was implemented and tested successfully in two dogs ( ). The results are also shown in Fig. 109.5 . Following a single ramping obstruction maneuver with submental force, stimulation is applied with the same electrodes (Stim).