Electrical stimulation of the whole hypoglossal nerve in patients with obstructive sleep apnea

Abstract

Purpose

Electrical stimulation of the whole hypoglossal nerve (HGp-ES) has been demonstrated to enlarge the pharynx and improve pharyngeal stability and patency to airflow in all animals studied, but not in humans. The present study was undertaken to better understand the effect of HGp-ES on the human pharynx.

Methods

Eight patients with obstructive sleep apnea who had implanted stimulators with electrodes positioned proximally on the main truck of the hypoglossus were studied under propofol sedation. Pharyngoscopy and air flow measurements at multiple levels of continuous positive airway pressure (CPAP) were performed before and during Hgp-ES.

Results

HGp-ES that activates both tongue protrusors and retractors narrowed the pharyngeal lumen at the site of collapse (velopharynx in all subjects) from 1.38 ± 0.79 to 0.75 ± 0.44 cm², p < 0.05 (measured at mid-range of CPAP levels) and lowered airflow (from 8.88 ± 2.08 to 6.69 ± 3.51 l/min, p < 0.05). Changes in critical pressure (Pcrit) and velopharyngeal compliance were not significant, but oropharyngeal compliance decreased (from 0.43 ± 0.18 to 0.32 ± 0.13 cm²/cmH2O, p <0.05). No correlation was found between the pattern of change in luminal shape (determined as the ratio of a–p vs. lateral diameter when lowering CPAP) or changes in cross-sectional area and airflow during Hgp-ES.

Conclusions

Our findings indicate that human retractors dominate when stimulated together with the protrusors during HGp-ES. While co-activation of retractors may be beneficial, it should be limited. We speculate that exercises that augment protrusor force may improve the response to hypoglossal stimulation. The exclusion of patients with concentric pharyngeal obstruction should be re-evaluated.

Introduction

Obstructive sleep apnea (OSA) is characterized by recurrent episodes of upper airway obstruction during sleep, due to sleep-associated relaxation of pharyngeal dilator and stabilizer muscles [1, 2]. In addition to the many anatomical abnormalities implicated in the pathogenesis of OSA, the role of state-dependent mechanisms is clear, since apneas occur only during sleep [2, 3]. Numerous methods have been used to restore upper airway patency during sleep, but no single treatment modality has been shown to provide relief to all patients with this disorder, as all are only partially effective and/or may be poorly tolerated. Continuous positive airway pressure (CPAP) has been the mainstay of treatment since the early 1980s, but compliance remains problematic despite continuous technical improvements. A large number of oral appliances and operative solutions have been designed, but usually only partial improvement can be achieved, in selected patients [4–6].

A new approach, aimed at “keeping the tongue muscle awake during sleep” with electrical stimulation of the hypoglossal nerve, promises to provide a goal-oriented treatment modality for OSA. It addresses the primary cause of pharyngeal obstruction in OSA, namely, the sleep-related decrease in pharyngeal dilating forces [7]. An initial approach to stimulating the tongue muscles transcutaneously [8] was disappointing [9]. A more recent approach, however, based on direct tongue muscle activation during inspiration with an implantable stimulation device provided promising results in the Medtronic-sponsored phase I study [10].

It took almost a decade until a phase II study was initiated by Inspire Medical Systems. Although only few technical modifications were undertaken compared with the earlier study, the initial responses (in part 1 of the study) were surprisingly poor, with minor or no improvement in the first patients implanted with the hypoglossal (HG) stimulation device. Examining potential causes for this outcome led investigators to change patients’ selection criteria and move the location of the electrode during implantation from a proximal to a more distal site on the HG nerve, with the goal of excluding nerve fibers to tongue retractor muscles. These changes resulted in substantial improvement in the second part of this phase II study [11].

While avoiding retractor contraction may seem intuitively logical, both animal studies and an initial pilot study in humans suggested that stimulation of the whole HG nerve should be preferable, or at least equivalent to more selective tongue stimulation. Therefore, the current study was undertaken to better understand the effect of whole-nerve hypoglossal stimulation on airflow and the pharyngeal luminal cross-sectional area and shape when the stimulation electrode was placed in the proximal position of the hypoglossal nerve (HGp-ES).

Methods

Patients

Eight patients with moderate to severe OSA previously enrolled to part 1 of the Inspire phase II feasibility study that evaluated efficacy and safety of HGp-ES in OSA patients [11] participated in the present study. All patients were already implanted with the upper airway stimulation system (Inspire Medical Systems Inc., Maple Grove, Minnesota, USA), as previously described [11]. In part 1 of the Inspire phase II study, the electrodes were placed on the proximal, main trunk of the hypoglossus, thereby stimulating all tongue muscles, including tongue retractors. All patients were considered treatment failure, as they failed to reach the pre-defined reduction in apnea-hypopnea index (AHI) at the 6th month post implantation after repeated titration attempts to optimize hypoglossal nerve-electrical stimulation (HG-ES), and all stopped using the stimulation device long before the current study. The study was performed in the respiratory research laboratory of Bnai Zion Medical Center. The aims and potential risks of the study were explained, and informed consent was obtained from all subjects. The study was approved by the Human Investigations Review Board of Bnai Zion Medical Center.

Sedation

Propofol sedation was delivered by an anesthesiologist, using a loading dose of 2.5 mg/kg body weight and continuous drip of 6–12 mg/kg body weight/h. Patients were maintained under stable sedation that eliminated any reaction to pain and electrical stimulation, while maintaining adequate ventilation, as monitored by the pneumotachometer and pulse-oximetry. Adequate levels of continuous positive airway pressure (CPAP) enabled breathing without inspiratory flow limitation.

Recording procedures

The methodology used in the present study was previously described [12] and employed in several previous studies. Subjects breathed through a tight-fitting nasal mask and pneumotachometer, connected to a Validyne ± 2 cmH₂O pressure transducer (Validyne, Los Angeles, CA, USA). The pneumotachometer was connected to a digitized CPAP device. ECG and oxygen saturation were monitor continuously. Nasal pressure (Pn) was monitored with a catheter connected to a side port of the mask. Intrathoracic pressure (Pes) was measured with an esophageal balloon catheter (Ackrad Laboratories, Cranford, NJ, USA) and used to help identifying flow limitation, as well as respiratory phase during complete apneas. Analogue-to-digital acquisition of all parameters was performed for monitoring and data storage on a digital polygraphic data acquisition system (LabVIEW; National Instruments, Austin, TX, USA).

Pharyngoscopy

A flexible fiber optic endoscope (Olympus BF-3C40; Olympus, Tokyo, Japan; outside diameter 3.3 mm) was inserted through an adequately sealed port in the nose mask and positioned first above the site of collapse (velopharynx in all subjects) and thereafter above the oro-pharynx, providing adequate view on the pharyngeal cross-sectional area (CSA) at both levels. The video recordings were stored, accompanied by audio comments.

Electrical stimulation

HGp-ES stimulation was performed using the implanted device, under propofol sedation. Stimulation was adjusted by controlling amplitude, pulse width, and pulse frequency, independently of the respiration sensor, using an external programming device. Repetitive continuous trains of 10–20 s duration (3–5 breaths) were applied. Stimulation intensity was adjusted, using previously available amplitudes between threshold and pain voltage, obtained during wakefulness.

Experimental procedure

Patients were prepared with venous access, and placed in the supine position. Following induction of sedation, CPAP was applied via nose mask, and raised to the level that abolished flow limitation (holding pressure). Once stable sedation was reached, the endoscope and esophageal balloon were positioned, and the mouth was sealed. The primary site of collapse, determined visually during gradual reduction of Pn and verified by the concomitant cessation of airflow, was at the level of the velopharynx in all our patients. Accordingly, the endoscope was placed above the area of velopharyngeal collapse. Thereafter, CPAP was reduced to a level that produced moderate flow limitation, and stimulation parameters were titrated to obtain the best flow-mechanical results. Once the stimulation parameters were determined, Pn was raised back to holding pressure, and then lowered randomly for six breaths, to 4–6 levels encompassing the range of inspiratory flow limitation and the level below which airflow ceased. At each Pn level, the first 3 breaths were recorded as baseline data, followed by application of HGp-ES for 3 breaths. This protocol provided the flow/Pn and velopharyngeal CSA/Pn relationships before and during HGp-ES quasi-simultaneously, over the range of Pn associated with flow limitation, as previously described [12]. After each Pn drop and data acquisition, Pn was raised back to the holding pressure until stable baseline ventilation was observed. The same procedure was repeated after the pharyngoscope was lowered below the uvula, to visualize the oro-pharynx and record the oro-pharyngeal CSA/Pn relationship.

Data analysis

The flow/Pn relationship over the range of flow limitation was determined using least-squares linear regression, to provide the critical pressure (Pcrit) and upstream resistance (resistance above the site of collapse, Pn/flow, upstream resistance (Rus)) [13]. The video images of the pharyngeal lumen, recorded during evaluation of the flow/Pn relationship before and during stimulation, were digitized and viewed, and single images from the end of expiration, when intra-pharyngeal pressure was equal to Pn, were captured and stored. The velopharyngeal CSA in each digitized frame was outlined manually and calculated digitally, using a designated software [12, 14]. The Pes tube, marked every cm, provided both a landmark to ensure stable distance of the pharyngoscope from the area measured and a calibration reference. The CSA/Pn relationship (i.e., compliance of the velopharynx and oropharynx) were determined for the close-to-linear portion of this relationship (i.e., below the Pn of the bending point of the exponential relationship that characterizes the tube law of collapsible tubes), using least-squares linear regression [14]. Velopharyngeal antero-posterior (a–p) and lateral diameters were measured at each Pn level, and the a–p to lateral relationship (a–p/lat slope) was calculated, to assess the pattern of pharyngeal narrowing during lowering of Pn at the site of collapse. A slope of 1 indicates equal shortening of a–p and lateral diameters, while lower slope indicates larger reduction in the lateral diameter. Data are presented as mean ± SD. Paired t test was used for comparison of data before and during HGp-ES. p < 0.05 was considered as statistically significant.

Results

The anthropometric and baseline polysomnographic characteristics of the study subjects (n = 8; all male) are given in Table 1. Patients were predominantly middle-aged overweight men (6/8 BMI > 30), and all but one had severe OSA (AHI > 40 events/h).

Table 1.

Anthropometric and polysomnographic characteristics (without HGp-ES) of the study subjects

Patients (n = 8)	Mean ± SD	Range
AHI (events/h)	58.4 ± 20.1	19–77
Age (years)	53.5 ± 8.8	34–63
BMI (kg/m2)	32.1 ± 2.2	29–35
Apneas/total (%)	72.8 ± 19.9	41–99
Lowest SO2 (%)	68.3 ± 13.1	55–90

AHI, apnea-hypopnea index; Apneas/total, ratio of apneas of all events; Lowest SO₂, lowest oxygen saturation value recorded during the sleep study

Figure 1 illustrates the effect of HGp-ES on velopharyngeal CSA in one of the patients. It can be seen that the velopharynx became narrower during HGp-ES, both in the anterior-posterior and lateral direction. The intensity of stimulation applied was close to the pain threshold identified before sedation. Increasing slightly the stimulation intensity worsened pharyngeal narrowing. The fact that the airway narrowed despite the decrease in compliance in this patient indicates an increase in external pressure exerted by the contracting tongue. Both Pcrit and Rus increased in this patient during HGp-ES, i.e., the pharynx became more collapsible, and narrowed also above the area of collapse.

Fig. 1.

Illustration of the methodology used to obtain and calculate the data used in the study. a Pharyngoscopic images of the site of collapse, at the level of the velopharynx (VP), before baseline (BL), and during electrical stimulation of the proximal hypoglossus (HGp-ES). Delineation of the orifice provides the cross-sectional area (CSA) and the a–p and lateral diameters. Similar pictures were taken at several nasal (mask) pressure (Pn) levels. The esophageal-pressure tube with known diameter (right lower corner in the pictures) provided a calibration reference. b The slope of the CSA/Pn relationships (lt. panel) denotes pharyngeal wall compliance. In this patient, HGp-ES decreased compliance, i.e., stiffened pharyngeal wall. Airflow measured simultaneously enabled construction of the flow/Pn curve (rt. panel). The reciprocal of the slope provides the resistance upstream to the area of collapse (Rus). Intercept of the slope with the x-axis provides the critical pressure at which the pharynx collapses, Pcrit. c The diameter/Pn relationship (upper panel) enabled construction of the a–p vs. lateral diameter relationship (lower panel). The slope of this relationship was used to characterize the pattern of change in the shape of the site of collapse with decreasing Pn

Figure 2 presents individual and average results of pharyngeal wall compliance for the whole group. Baseline CSA/Pn slope of the velopharynx (i.e., compliance of the site of collapse, 0.26 ± 0.17 cm²/cmH₂O), remained nearly unchanged during HGp-ES (0.25 ± 0.20). In contrast, HGp-ES reduced the oro-pharyngeal compliance significantly (from 0.43 ± 0.18 to 0.32 ± 0.13 cm²/cmH₂O, p < 0.05), probably by stiffening the tongue. Pcrit and Rus tended to increase, but changes were not significant (Fig. 3).

Fig. 2.

Individual and mean ± SD response of velo- and oropharyngeal compliance to hypoglossal stimulation (HGp-ES). Although HGp-ES reduced oropharyngeal compliance, this beneficial effect probably did not have an important flow-mechanical effect, as the site of collapse was at the level of the velopharynx in all subjects. BL, baseline. *p <0.05

Fig. 3.

Individual and mean ± SD response of Pcrit (critical pressure) and Rus (upstream resistance) to HGp-ES. Both parameters tended to increase (i.e., deteriorate) during HGp-ES, but on the average changes were not statistically significant

However, when the effect of HGp-ES on CSA and flow at the site of collapse was calculated at the middle of the CSA/Pn and flow/Pn slopes (“semi-obstruction”), a significant reduction was found (Table 2). No correlation was found between the parameter used to characterize the pattern of obstruction at the site of collapse (a–p/lateral diameter slope, Fig. 1) and the magnitude of change of CSA (R = − 0.26) and flow (R = 0.1) during HGp-ES (Table 2).

Table 2.

Relationship between velopharyngeal antero-posterior to lateral diameter ratio (VP a–p/lat) at the site of collapse and HGp-ES induced changes in CSA and flow, at mid-range of Pn levels evaluated

patient	VP a–p /lat	CSA (cm2)			Flow (l/min)
		BL	HGp-ES	Δ CSA	BL	HGp-ES	Δ flow
1	0.65	1.4	0.9	−0.5	8.1	6.7	−1.4
2	0.35	0.51	0.56	0.05	9.5	12.5	3
3	0.53	2.4	1.7	−0.7	9.3	7	−2.3
4	0.61	1.2	0.54	−0.66	10.5	6.7	−3.8
5	0.09	0.2	0	−0.2	5.8	0	−5.8
6	0.55	1.7	0.8	−1.6	6.5	3.4	−3.1
7	0.39	2.6	0.7	−1.9	12.8	9.5	−3.3
8	0.24	1	0.8	−0.2	8.5	7.7	−0.8
mean ± SD	0.43 ± 0.18	1.38 ± 0.79	0.75 ± 0.44*	−0.59 ± 0.55	8.88 ± 2.08	6.69 ± 3.51*	−2.19 ± 2.43

^*p < 0.05 for comparison of baseline (BL) and HGp-ES (hypoglossal stimulation)

Discussion

The results of the present study indicate that electrical stimulation of the whole hypoglossal nerve (HGp-ES) which recruit all extrinsic and intrinsic tongue muscles, fails to enlarge and/ or prevent collapse of the pharynx in anesthetized patients with OSA. In fact, HGp-ES often narrowed the pharynx and reduced airflow. This finding was independent of the size of the pharyngeal CSA that was altered by CPAP. Also, narrowing was not due to distortion of the pharynx due to the unilateral HGp-ES, and occurred both at the level of the velo- and oro-pharynx. As HGp-ES contracts both tongue protrusors and retractors, our finding indicates that when stimulated simultaneously, the balance of forces of protrusors and retractors in human is either equal or the retractors dominate.

This finding was rather unexpected; unilateral HG-ES studies were performed in several animal species, and enlarged the pharynx in all [15–18]. In fact, in rats it was documented that co-activation of protrusor and retractors improved pharyngeal dilatation [19, 20]. Also, preliminary studies in few OSA patients reported a similar improvement in AHI with proximal (stimulation of tongue protrusors + retractors) compared with more distal (preferential tongue protrusors stimulation) placement of the stimulation electrode [21, 22]. As proximal placement was associated with bradycardia in one of the patients [23], the Medtronic phase I investigators choose to place the electrode in the more peripheral position, stating that this position was chosen because both sites provide similar responses [10]. Only in the later phase II study did it become clear that a peripheral electrode positioning is required to obtain adequate flow response, thereby apparently avoiding hypoglossus fibers that activate most of the retractors’ motor units [24]. The following Inspire phase III study was by far more successful [25], leading to implantation of many stimulation devices worldwide, with most satisfactory polysomnographic and clinical responses [26–28]. Distal placement of the stimulation electrodes is now a routine procedure when using the Inspire stimulation system, along with additional techniques to target specific muscle activation. The need for selective stimulation of specific areas of the tongue in order to optimize airflow during sleep [29] has also been implemented by the currently ongoing targeted hypoglossal nerve stimulation (LivaNova Inc.) which implants a cuff electrode array proximally and selectively stimulates specific sectors specific sectors of the main trunk of the hypoglossus nerve [30].

A notable finding in our study was that proximal stimulation of the hypoglossal nerve had an unfavorable flow-mechanical response. This finding is consistent with the notion that airway patency is determined by the relative balance of activity in tongue protrusor and retractor muscles, but that retractors overpower the protrusors when the whole nerve is stimulated. It might still be the case that synergy between these two muscle groups could accompany somewhat lesser degrees of retractor activation, which can combine with protrusors to stabilize tongue shape and position within the oral cavity and prevent its prolapse into the pharynx. Indeed, identifying specific defects in pharyngeal mechanics offers the potential for developing personalized approaches to electrical stimulation in patients with obstructive sleep apnea.

An additional observation noticed during the first part of Inspire phase II study was that concentric occlusion at the site of collapse, as seen during pre-implantation drug induced sedation endoscopy (DISE), appeared to be associated with poor results [11]. However, the visual assessment of DISE findings, including the presence of concentric occlusion, is still not well-standardized [31]. Using measurable parameters, we did not find a relationship between the geometrical pattern of pharyngeal collapse and the mechanical effect of HGp-ES in the present study. Concentric occlusion is still considered an exclusion criterion for Inspire hypoglossus nerve stimulator implantation [28, 32]. Although based only on 8 patients with electrodes placed on the main trunk of the hypoglossus, we believe that our findings justify re-evaluation of this parameter in patients implanted with the currently used distally placed electrode; if it turns out that concentric narrowing should not be a reason for disqualifying applicants, more patients can benefit from this novel treatment. For this purpose, however, a standardized methodology needs to be developed, as the method used in the current study is probably too cumbersome for clinical use.

The findings of the current study clearly demonstrate that results obtained in anesthetized animals need to be interpreted cautiously when applied to humans. Upper airway anatomy differs among mammals in several important aspects, and none of the animal models studied is prone to develop sleep apnea. Nevertheless, the results obtained with HGp-ES provide new insight to tongue physiology that may have practical relevance. Contrary to humans, other mammals need to stretch and protrude their tongues way out of the mouth many times every day in order to drink and lick, and often also for eating, leading to a balance of forces that favors tongue protrusion. The finding that retractors dominate in the human tongue and overcome the protrusors may contribute to the propensity of humans to pharyngeal collapse during sleep, when the wakefulness-associated increase in genioglossus activity declines [33]. If we expand on this idea further, our findings suggest that myofunctional therapy used to ameliorate OSA [34] should emphasize tongue protrusion exercises, and speculate that such exercises could also improve the response to hypoglossus stimulation.