Epidemiology

Obstructive Sleep Apnea (OSA) afflicts approximately 3%–9% of women and 10%–17% of men between the ages of 30–70 years ( ). The disorder is characterized by recurrent episodes of decreased airway patency during sleep due to partial or complete collapse ( ), leading to a reduction of airflow (hypopnea) or absence of airflow (apnea). OSA severity is defined by the number of apneas and hypopneas occurring per hour of sleep (Apnea Hypopnea Index or AHI) with mild disease defined as 5–14 events/h, moderate disease as 15–29 events/h, and severe disease as 30 or more events per hour of sleep ( ). Left untreated, OSA induces repetitive bouts of hypoxemia with hypercapnia which subsequently evoke systemic inflammation, endothelial dysfunction, metabolic dysregulation, and sympathetic over-activation. These physiologic alterations increase risk for systemic hypertension, cardiovascular morbidity, glucose intolerance, and neurocognitive impairment ( ).

Mechanisms

OSA occurs when neuromuscular control of a collapsible oropharynx and/or retropalatal pharynx (velopharynx) is insufficient to maintain functional patency during sleep, a time of decreased upper airway muscle tone. Anatomic traits such as retrognathia, hypopharynx, nasal septal deviation, or polyps negatively impact airflow and pressure relationships, thereby contributing to upper airway collapsibility and instability. Dysfunction in neurochemical control of the upper airway and/or chest wall musculature also contribute ( ). This has led to hypotheses proposing the collective influence of four distinct pathways promoting the pathophysiologic mechanism of OSA. These include: (1) sleep itself (specifically a low respiratory arousal threshold); (2) anatomy (small airway size and high pharyngeal compliance); (3) reduced neural activation and reflex responses of muscles contributing to upper airway patency (i.e., inadequate upper airway muscle drive); and (4) loop gain (controls on overshoot and undershoot of ventilatory response) ( ). Interventions targeting these contributing pathways provide opportunities to interrupt the intermittent, cyclic reductions, or cessations of airflow in sleep apnea.

Current Therapy

First-line treatments primarily target the anatomical contribution to OSA. Continuous positive airway pressure (CPAP), delivered via nasal face mask, works by inflating and keeping open the collapsible portion of the upper airway. Secondarily, CPAP inhibits genioglossal muscle activity, reduces loop gain by lung inflation, and reduces sleep fragmentation by decreasing arousals triggered by apneic events ( ). Despite clinical effectiveness, adherence to therapy is often limited and attributed to mask discomfort, nasal and pharyngeal dryness, loss of partner intimacy, and claustrophobia ( ).

Alternatively, increased pharyngeal space can be achieved either with oral-appliance therapy or site-selective surgery on soft tissue or bony structures. Oral appliances, which can be customized to a person’s dentition, are designed to fit over the teeth and hold or advance the mandible and tongue forward. Although critical closing pressure of the airway is improved, clinical effectiveness of oral appliances is best in persons with mild to moderately severe OSA ( ). Furthermore, adherence is limited by patient complaints of continued snoring, dry mouth or excessive salivation, mouth or teeth discomfort, muscle tenderness, or temporomandibular joint irritation and inflammation ( ). Additionally, an effective oral appliance requires sufficient molar dentition in all four quadrants.

With the exclusion of nasal surgery for fixed obstruction, surgical modification of pharyngeal soft tissues or bony structures of the upper airway faces other challenges. Airway narrowing can occur in the velopharynx, the oropharynx, and/or the retrolingual pharynx (hypopharynx). This narrowing varies between persons, influenced by obesity as well as multiple skeletal dimensions such as width of the hard palate, length of the bony mandible, and orientations of the hyoid bone and maxilla ( ). With multiple potential sites of airway obstruction, surgical alternatives may provide inadequate relief for persons with moderate to severe disease ( ).

CPAP, oral appliance therapy, and surgery exert a therapeutic effect through structural enhancements of upper airway size and stability. In contrast, neuromodulation is a functional therapy, exerting its effects by reversing inadequate muscle activation. This makes hypoglossal nerve stimulation (HNS) uniquely different than CPAP, site-specific surgery, or oral appliances that address the anatomical or passive mechanical factors contributing to airway collapse ( ). Neurostimulation applied to the hypoglossal nerve (HN) represents a treatment paradigm shift and provides an opportunity to mitigate many of the issues of patient discomfort and stigma associated with previous treatment options.

Tongue Musculature and Hypoglossal Nerve (HN) Control

Four pairs of extrinsic muscles and four intrinsic muscles comprise the human tongue; the extrinsic muscles of the tongue originate from bone and insert into muscle, while intrinsic muscles have both origin and insertion points in muscle ( ). The palatoglossus muscle, an extrinsic muscle, is innervated by the cranial nerve (CN) X/XI complex; the remaining muscles of the tongue are innervated by the HN or CN XII as illustrated in Fig. 108.1 . The extrinsic muscles are primary controllers of tongue position for the various functions of swallowing, speech, and breathing. Two extrinsic muscles, the styloglossus (connecting the tongue to the base of the skull) and hyoglossus (connecting the tongue to the hyoid bone), are innervated by branches from the lateral division of CN XII and are mainly retrusors. The genioglossus muscle (connecting the tongue to the mandible), is an extrinsic muscle and the primary tongue protrusor, composed of oblique and horizontal compartments which receive about five to six nerve branches from the medial division of CN XII ( ).

Hypoglossal nerve innervation of upper airway musculature.

From Moore, K.L. & Agur, A.M.R. Clinically Oriented Anatomy, 6th edition, Wolters Kluwer Health, 2013.

Activation of the genioglossus muscle pulls the tongue forward, thus widening the pharyngeal airway ( ). Combined stimulation of both protrusor and retrusor muscles has been observed to stiffen the pharynx, contributing to maintenance of airway patency during sleep. Consequently, the point at which electrical stimulation is placed on the branches of CN XII produces distinct differences in tongue movement ( ).

Control of protrusive tongue movement occurs in neurons in the ventral motor nucleus of CN XII ( ). Electrophysiologic studies indicate that activity patterns in the motor nucleus are modulated by premotor respiratory neurons controlling inspiratory drive and tonic activity, thus closing the loop between tongue function and respiration ( ). Anatomic evidence exists for connection to the Kölliker-Fuse nucleus, an area in the brainstem respiratory central pattern generator responsible for coordination rather than initiation of a breath ( ).

Transcutaneous Electrical Stimulation (TES) of Upper Airway Muscles

The direct relationship between loss of genioglossus muscle activation during sleep and subsequent airway obstruction was first described in 1978 ( ). These studies led to the first effort to directly stimulate pharyngeal muscles in humans as a treatment for OSA. In 1988, skin surface electrodes were placed on the submental region and connected to a demand-type electrical stimulator triggered by tracheal breath sounds ( ). Initial reports suggested this approach decreased both frequency and duration of apnea episodes, decreased frequency of oxygen desaturation episodes, and promoted deeper sleep ( ).

Continued investigations into transcutaneous electrical stimulation (TES) of the submental area, the subhyoid area, and intraoral region provided additional insights, although efficacy outcomes were inconsistent. Transcutaneous stimulation applied during sleep did not appear to change upper airway size, prevent apneas or improve sleep architecture in one report ( ), although others reported a decrease in inspiratory resistance dependent on voltage and stimulation frequency, suggesting a widening effect on the supraglottic airway ( ). Other investigators observed that while an intraoral electrode placed at the floor of the mouth induced tongue protrusion and an increase in the size of the posterior oropharyngeal airway, electroencephalographic (EEG) arousals occurred each time submental stimulation broke an apnea episode ( ). Subsequent studies were unable to consistently replicate findings of improved sleep architecture, decreased snoring, reduced daytime sleepiness, improved oxygenation, and reduced AHI in all studies ( ).

Two reports from the same center illustrate current attempts to develop TES approaches. A feasibility report was generally supportive with outcomes of significantly reduced snoring, improved oxygenation, and decreased respiratory disturbance index and AHI when bilateral surface patch electrodes were placed halfway between the chin and the angle of the mandible over the submental areas ( ). Stimulation (mean current, 10.1 mA; mean frequency, 30 Hz) was manually applied for 10 min periods during stage N2 sleep when upper airway occlusion was observed during attended polysomnography (PSG). However, in the subsequent randomized clinical trial of a full night of transcutaneous stimulation applied during occlusive events, the total AHI, snoring, average, and nadir oxygen saturations did not change, although significant improvements were seen in the 4% oxygen desaturation index (4% ODI) ( ). When considering the a priori primary outcome of 4% ODI, 17 of 36 participants (47.2%) responded to the therapy, primarily those in the mild–moderate OSA category. Baseline ODI directly predicted response with approximately 10% of variance explained. Stimulation failed to improve AHI during rapid eye movement (REM) sleep ( ). TES may provide an additional way to screen for responsiveness to implantable neurostimulation therapy ( ).

Initial Demonstration of Hypoglossal Nerve Stimulation (HNS)

The next major step in the development of upper airway electrical stimulation occurred in , when Decker et al. performed the first trial of nerve stimulation in humans. HNs were directly stimulated during both wakefulness and sleep by current applied through fine-wire electrodes ( ). The study also assessed the amount of electrical current, applied to surface electrodes and then to fine-wire electrodes, required to induce an arousal from sleep. Key findings from that study revealed that (1) direct neurostimulation of the HN via percutaneous fine-wire electrodes resulted in genioglossus muscle recruitment and tongue protrusion; (2) when using fine-wire electrodes, the electrical current necessary to increase upper airway size was lower than that which induced arousal from sleep; and (3) neither surface stimulation nor fine-wire stimulation could consistently terminate obstructive apnea during sleep. CAT scan observation of pharyngeal changes during fine-wire stimulation in two OSA patients showed both increases and decreases in upper airway size with tongue protrusion, suggesting that electrode position on the various branches of the nerve determined whether upper airway size increased or decreased with neurostimulation ( ).

In 1997, a cuff electrode was surgically placed around the HN for testing during one sleep period ( ). Successful tongue protrusion occurred during stimulation of the distal branch of the HN that supplies the genioglossus muscle, and tongue retrusion occurred during stimulation of the main trunk of the HN. Another key finding was that neurostimulation did indeed open the airway despite occurrence of tongue retrusion ( ).

Simultaneously, biomedical engineers began exploring the feasibility and applicability of single or multielectrode cuffs, biocompatible hardware for implanting, development of both open-loop and closed-loop systems to trigger the neurostimulator, and determination of optimal stimulating parameters. Contributions from these investigations included observations that nonselective stimulation delivered by a multielectrode cuff on the HN yielded the greater benefit to airflow during expiration; during inspiration, improved airflow resulted when the whole nerve was stimulated along with selective coactivation of protrusors (genioglossus) plus retrusor muscles (hyoglossus/styloglossus) ( ).

Closed-Loop Hypoglossal Neurostimulation Devices

The first pilot study of an implantable hypoglossal neurostimulation device was published in 2001 ( ). The closed-loop stimulation device (Inspire I; Medtronic Inc., Minneapolis, MN) was tested in eight males with moderate-severe OSA (mean AHI, 52.0 with range 17.1–80.8, mean BMI, 28.4 kg/m ² ). The device consisted of (1) a half-cuff silicone-insulated, guarded, bipolar platinum stimulating electrode, placed unilaterally on an HN branch peripheral to the large inferior distal branch supplying the genioglossus muscle, (2) an implantable pulse generator (IPG) placed in a right-sided infraclavicular subcutaneous pocket, (3) a respiratory piezoelectric pressure sensor placed against the pleura through a drilled hole in the superior manubrium to sense the end of expiration and trigger the electrical stimulus to occur at the start of inspiration, and (4) a remote control unit to allow patient-controlled activation of the device for sleep. Stimulation parameter adjustments were made at one, three, and six-month follow-up to maximize inspiratory airflow and alleviate sleep apnea/hypopnea while avoiding EEG arousal. Significant decreases were noted in AHI during NREM and REM sleep for entire night PSG recordings, along with significant improvements in oxygenation. A nonsignificant trend toward an increased percentage of stage N3 sleep was noted. The implanted device was tolerated in nightly use, replacing CPAP for its participants, with no detected abnormalities in tongue structure or function. Device-related adverse events (AEs) were frequent and included IPG failure, intermittent sensor shut-down, transient asynchronous stimulation due to sensor signal artifact, and electrode breakage. These technical difficulties resulted in a decade-long hiatus of human trials as devices were redesigned to be tested in clinical trials.

The next human clinical trial report appeared in 2011. Safety and efficacy were reported for a second-generation, hypoglossal nerve stimulation system (HGNS; Apnex Medical Inc., St. Paul, MN), implanted into 21 participants at 4 Australian clinical sites ( ). The implanted device differed from that trialed a decade earlier, in that it had dual respiration sensing leads that were tunneled from the stimulator along the midline and then bilaterally along each costal margin. The sensor leads measured thoracic bioimpedence used to predict onset of inspiratory effort. The device was designed to deliver stimulation immediately prior to and then throughout the entire inspiratory phase at a preset time of day that could be modified by the patient. Stimulation parameters were titrated, based on tolerance and consistent abolishment of inspiratory flow limitation, and adjusted over time for patient comfort. Significant decreases in AHI were demonstrated at three- and six-month follow-up, as well as significant reduction in number of respiratory related arousals, total arousals, and oxygen desaturations ( ). Sleep architecture was significantly improved with reduced sleep latency, increased sleep efficiency, reduced percentage of stage N1 sleep ¹

1 Normal sleep is divided into nonrapid eye movement (NREM) and REM sleep. NREM sleep is further divided into progressively deeper stages of sleep: stage N1, stage N2, and stage N3 (deep or delta-wave sleep) ( ).

In 2012, publication of a two-part feasibility study appeared, reporting efficacy of a second upper airway stimulation (UAS) system (Inspire II UAS, Inspire Medical Systems, Maple Grove, MN). Refinement of selection criteria was proposed to identify optimal responders to neurostimulation therapy ( ). This second-generation device delivered stimulation through three platinum/iridium (PtIr) stimulating electrodes embedded in double overlapping polyurethane cuffs that circumferentially encircled the HN as illustrated in Fig. 108.2A and B . Initial selection criteria were rather broad: patients with moderate to severe OSA (AHI >20/h). Stimulation parameter adjustments were made during two- and four-month PSG evaluations to maximize inspiratory airflow and alleviate sleep apnea/hypopnea without EEG arousal. Six-month outcome data showed that AHI did not change significantly from baseline for the entire group in part 1 (n = 20). Overall, six patients (30%) met a predefined responder definition (reduction in AHI of at least 50% from baseline and AHI < 20 by six months) and showed significantly decreased ODI and improvements in sleepiness and sleep-related QOL. Analyses suggested that results were more predictable in those with baseline characteristics of BMI ≤ 32 kg/m ² , AHI ≤ 50, and absence of complete concentric palatal closure. Results also suggested an optimal placement of this cuff electrode on the distal medial branch of the HN. SAEs of this second-generation UAS device occurred in three persons (device-related infection, pain and swelling at the neck resolving with antibiotic administration, and inability to activate the tongue with amplitude in the allowable range) ( ).

(A and B) Inspire Medical’s 3-electrode stimulation lead (Courtesy of Inspire Medical Systems); (C and D) ImThera Medical’s 6-contact hypoglossal nerve cuff electrode

(Courtesy of ImThera Medical, Inc.).

Open-Loop Hypoglossal Neurostimulation Devices

A third medical device company focused on an open-loop approach, capitalizing on the activation of both protrusor muscles and retrusor muscles to increase pharyngeal stiffness as well as dilate the upper airway. A pilot study reporting 12-month outcomes (N = 13) evaluated an implanted neurostimulation device (aura6000; ImThera Medical, Inc., San Diego, CA) designed to target multiple muscle groups innervated by the HN ( ). The device consisted of (1) an IPG surgically placed in a subcutaneous pectoral pocket; (2) an 8 mm silicone cuff housing six independent electrodes, placed on the proximal portion of the HN before it branches; and (3) a patient remote control activation device set to initiate a preset 7-h period of stimulation, following a 45 min delay (unless halted earlier by the patient). The system did not require a respiratory sensing component. Instead, stimulation sequentially cycled from one electrode to the next, providing stimulation during both inspiration and expiration; no individual nerve fiber was subjected to continuous stimulation. Between two and four electrodes were activated per patient ( ). Device titration at one and three months during overnight PSG included adjustments in stimulation current, contact stimulation time, stimulation frequency, or cathodic phase duration on previously selected contacts. Clinical outcomes at 12 months reflected significant decreases in AHI and ODI and improved daytime sleepiness from baseline. Three of 13 subjects were nonresponders, one with an unusually large and long uvula, one suffering from predominantly central apnea, and one with the highest baseline AHI and BMI. Noteworthy AEs included a total of three persons experiencing either transient hemi-tongue paresis or transient dysphagia. Device-related AEs included malfunction of the external remote-control charging device experienced by all 13 patients with successful repair or replacement, broken leads in 2 patients, and one defective pulse generator ( ).