Intraoperative identification of mixed activation profiles during hypoglossal nerve stimulation

Abstract

Study Objectives:

The effectiveness of hypoglossal nerve stimulation (HGNS) in the treatment of obstructive sleep apnea (OSA) depends on the selective stimulation of nerve fibers that innervate the tongue muscles that produce tongue protrusion (genioglossus) and stiffening (transverse/vertical) while avoiding fibers that innervate muscles that produce tongue retraction (styloglossus/hyoglossus). Postoperative treatment failures can be related to mixed activation of retractor and protrusor muscles, despite intraoperative efforts to identify and avoid nerve fibers that innervate the retractor muscles. This study describes a novel intraoperative protocol that more optimally identifies mixed activation by utilizing an expanded set of stimulation/recording parameters.

Methods:

This study was a case series in a university hospital setting of patients undergoing unilateral hypoglossal nerve stimulation implantation for obstructive sleep apnea. Data included electromyographic responses in the genioglossus and styloglossus/hyoglossus to intraoperative stimulation with an implantable pulse generator using unipolar (- - -, o-o) and bipolar (+-+) settings.

Results:

In a subset of patients (3/55), low-intensity unipolar implantable pulse generator stimulation revealed significant mixed activation of the styloglossus/hyoglossus and genioglossus muscles that was not evident under standard bipolar implantable pulse generator stimulation conditions. Additional surgical dissection and repositioning of the electrode stimulation cuff reduced mixed activation.

Conclusions:

A novel intraoperative neurophysiological monitoring protocol was able to detect significant mixed activation during hypoglossal nerve stimulation that was otherwise absent using standard parameters. This enabled successful electrode cuff repositioning and a dramatic reduction of mixed activation.

Citation:

Sturm JJ, Lee CH, Modik O, Suurna MV. Intraoperative identification of mixed activation profiles during hypoglossal nerve stimulation. J Clin Sleep Med. 2020;16(10):1769–1774.

BRIEF SUMMARY

Current Knowledge/Study Rationale: In patients who undergo hypoglossal nerve stimulation therapy for obstructive sleep apnea, a significant portion of postoperative treatment failures are related to mixed activation of tongue muscles that produce retraction, as well as those that produce protrusion. Novel intraoperative methods for detecting and avoiding such mixed activation are needed to optimize treatment outcomes.

Study Impact: This study describes the development and implementation of a novel intraoperative neurophysiological monitoring protocol for hypoglossal nerve stimulation in the management of obstructive sleep apnea. Use of this protocol led to the identification of significant mixed activation that was undetectable by standard methods and enabled intraoperative device repositioning that resulted in dramatic reductions in mixed activation.

INTRODUCTION

Obstructive sleep apnea (OSA) is characterized by upper airway collapse during sleep, causing oxygen desaturations and recurrent arousals. OSA is highly prevalent, affecting 13%–33% of men and 16%–19% of women, and is becoming increasingly common.¹ Patients with moderate to severe OSA (defined as apnea-hypopnea index [AHI] > 15 events/h) are at increased risk for a variety of health concerns, including cardiovascular disease and neurocognitive dysfunction.^2,3 First-line therapy for OSA is positive airway pressure (PAP), and with consistent use, PAP reduces OSA severity, as well as associated comorbidities^4,5; however, PAP adherence is a major barrier to effective therapy for many patients, with 46%–83% of patients reporting nonadherence.⁶

Hypoglossal nerve stimulation (HGNS) is a promising new therapy for PAP-intolerant patients with moderate to severe OSA.⁷ The HGNS system is composed of three surgically implanted components: a stimulation electrode cuff placed around the hypoglossal nerve, a pulse generator placed in the upper chest, and a thoracic respiratory sensor. During HGNS, when the thoracic monitor senses inspiration, a stimulus is delivered from the pulse generator to the electrode cuff, leading to stimulation of medial hypoglossal nerve (HGN) fibers. This creates tongue protrusion, which increases upper airway tone and opening, thereby preventing collapse during inspiration. Outcomes from HGNS have been encouraging, with 5-year data from the Stimulation Therapy for Apnea Reduction trial demonstrating long-term AHI reductions and sustained improvements in sleep symptoms. Despite this significant progress, however, therapy success rates of HGNS (defined as AHI < 20 events/h and > 50% reduction of AHI) remain around 66%.⁸ Accordingly, a major outstanding question in the field is how clinical response rates to HGNS can be optimized.

The effectiveness of HGNS in reducing upper airway obstruction in OSA depends on the ability to selectively stimulate HGN fibers that innervate the muscles that produce tongue protrusion while avoiding muscles that produce tongue retraction. Achieving this selective stimulation requires an in-depth knowledge of the complex functional neuroanatomy of motor innervation of the tongue by the HGN. The major extrinsic tongue muscles responsible for protrusion are the paired genioglossi (GG), which consist of an oblique compartment (GG_O) that produces depression of the tongue, and a horizontal compartment (GG_H) that produces protrusion of the posterior tongue.⁹ These muscles are predominantly innervated by the medial division of the ipsilateral HGN (GG_H by the inferior medial HGN and GG_o by the superior medial HGN) and can also receive significant crossed motor innervation from contralateral medial HGN branches.^10,11 The major extrinsic muscles responsible for tongue retraction are the styloglossus and hyoglossus (SG/HG), which are predominantly innervated by the lateral division of the HGN. Accordingly, during HGNS surgery, medial-division HGN branches are targeted for inclusion within the stimulation cuff, whereas lateral-division branches are avoided. Additionally, the first cervical nerve is often included, which innervates the geniohyoid muscle and assists with opening the hypopharyngeal airway by moving the hyoid bone anteriorly.⁹

When lateral-division HGN branches are mistakenly included in the HGNS electrode cuff, patients can exhibit mixed activation profiles (ie, a combination of tongue protrusion and retraction). Patients with significant mixed activation have inferior therapeutic outcomes compared with patients with unilateral or bilateral tongue protrusion.¹² Current standard protocols for achieving selective HGNS involve using intraoperative nerve integrity monitoring in conjunction with electromyography (EMG) to isolate medial-division HGN branches from lateral-division branches; however, despite these efforts, late takeoff, embedded lateral division branches are commonly missed and erroneously included in the stimulation cuff, which can produce mixed activation.⁹ Therefore, effective methods for intraoperatively identifying and avoiding difficult-to-detect lateral HGN branches during HGNS surgery are needed.

Here we describe a novel intraoperative protocol aimed at detecting mixed activation during HGNS that might otherwise go undetected using standard techniques. This protocol involves an expanded set of stimulation/recording parameters, including both unipolar and bipolar implanted pulse generator (IPG) settings at a range of stimulation intensities. We describe three cases of HGNS implantation where this protocol proved essential for detecting “hidden” mixed-activation profiles. Identifying these profiles intraoperatively enabled repositioning of the electrode cuff, which resolved mixed activation, as evidenced by unrestricted forward tongue protrusion and a corresponding decrease in EMG activity for SG/HG muscles.

METHODS

This study was approved by the Institutional Review Board at Weill Cornell Medicine. Fifty-five total patients with moderate to severe OSA undergoing unilateral (right) HG nerve stimulator implantation (Inspire Medical Systems, Maple Grove, Minnesota) at Weill Cornell Medicine were included in the study. The inclusion criteria were (1) demonstrable PAP noncompliance, (2) AHI ≥ 15 events/h, (3) body mass index < 35 kg/m², and (4) an absence of complete circumferential collapse of the velopharynx during preoperative drug-induced sleep endoscopy.

During HGNS implantation, EMG activity was recorded via electrodes placed in GG and SG/HG muscles, as previously described.¹³ EMG needles were placed directly into the tongue muscles of interest in the following manner:

1. Right-side GG: right midinferior section of the tongue (bipolar electrodes, tip exposed);

2. Left-side GG: left midinferior section of the tongue (bipolar electrodes, tip exposed);

3. Right-side SG/HG: right ventral-lateral side of the tongue (bipolar electrodes, tip exposed).

Intraoperative stimulation protocol

The main trunk of the HGN is exposed, and a bipolar stimulation probe is used to stimulate the lateral and medial nerve branches (0.1−0.3 mA) (Figure 1). As previously described,¹³ stimulation is delivered to four locations: (1) the lateral division branches of the HGN, (2) the superior section of the medial division branches of the HGN, (3) the inferior section of medial division branches of HGN, and (4) the first cervical nerve branch. During this stimulation, EMG responses are recorded from GG and SG/HG muscles (Figure 1). Additional stimulation at the breakpoint between the lateral and medial divisions of HGN is performed to identify late takeoff and hidden lateral nerve fibers based on EMG responses and direct visualization of muscle contraction in the operative field, especially in the situations of ambiguous EMG findings. These intraoperative data are used to direct initial placement of the electrode stimulation cuff around the desired HGN branches, and the remainder of the surgery proceeds with placement of the IPG and respiratory sensor leads.

Figure 1. Intraoperative neurophysiological monitoring protocol.

(1) The main trunk of the HGN is exposed, and a bipolar stimulation probe is used to stimulate four different HGN nerve sites A: Lateral (SG/HG), B: Superior medial (GG_O), C: Inferior medial (GG_H), D: C1 branches while recording EMG responses in GG and SG/HG muscles. Procedure as described previously.¹³ Additional stimulation at the breakpoint between SG/HG and GG_O is performed to identify late takeoff and hidden lateral nerve fibers. (2) Electrode cuff is targeted to include HGN branches that produced selective activation of GG, but not SG/HG. (3) Standard HGNS surgery is completed with placement of implanted pulse generator (IPG) in the chest and sensor lead in the intercostal space. (4) IPG is used to stimulate under three different polarities: Bipolar (+-+ ), unipolar (- - -), and unipolar (o-o) while measuring EMG activity and observing tongue motion. First, bipolar IPG stimulation is performed, progressing from high to low intensity. Next, unipolar IPG stimulation is performed, again progressing from high- to low-stimulation intensity. (5) If significant mixed activation is observed either by EMG (eg, SG/HG activity) or tongue motion, the electrode cuff is repositioned, and step 4 is repeated. Successful repositioning is confirmed by elimination of mixed activation measured by EMG and by direct intraoperative observation of unrestricted forward tongue protrusion in response to IPG stimulation. EMG = electromyography, C1 = first cervical nerve, GG = genioglossus, GG_H = horizontal compartment of the genioglossus, GG_O = oblique compartment of the genioglossus, HG = hypoglossus, HGN = hypoglossal nerve, IPG = implanted pulse generator, SG = styloglossus.

Next, an expanded IPG stimulation protocol is carried out involving both bipolar and unipolar settings at various stimulation intensities (Figure 1). Mixed activation was defined as (1) clinical evidence of ipsilateral tongue retraction directly observed intraoperatively and/or (2) significant polymorphic EMG activity on the SG/HG channel. The protocol begins with bipolar IPG stimulation at six stimulation intensities progressing from high to low ([+-+] 1.2 V, 1.0 V, 0.5 V, 0.3 V, 0.2 V, 0.1 V). During this IPG stimulation, both the tongue motion and the EMG activity are observed. If significant mixed activation is detected by visualizing tongue retraction and/or by characteristic fractionated EMG signal on SG/HG channel, further dissection is performed to isolate out the remaining retractor nerve branches, and the cuff is repositioned over the inclusion branches by identifying and dissecting out any retained distal retractor fiber, by further proximal separation of the exclusion fibers to prevent entrapment inside of the cuff, or by positioning the cuff more distally on the nerve. If unrestricted forward protrusion is observed without significant SG/HG activity by EMG, unipolar IPG stimulation is then performed under two settings (- - - and o-o), again progressing from high to low stimulation intensity (0.5V, 0.3V, 0.2V, 0.1V). Again, if mixed activation is observed either clinically or by EMG activity, the cuff is repositioned. If cuff repositioning occurs, the IPG stimulation protocol is repeated from the beginning. Successful repositioning is confirmed by elimination of mixed activation measured by EMG and by direct observation of unrestricted forward tongue protrusion in response to IPG stimulation. Once the protocol is completed without evidence of mixed activation, the incisions are closed, and the surgery is completed.

RESULTS

Of the 55 total patients implanted during the study period of 05/2018 to 11/2019, three patients (5%) displayed mixed activation under low-intensity unipolar IPG stimulation conditions that was otherwise absent under standard bipolar IPG stimulation. Demographic data and sleep study findings for each of the patients are illustrated in the Table 1. All 55 patients showed unrestricted forward (bilateral or right) protrusion in response to bipolar IPG stimulation at the time of postoperative device activation.

Table 1.

Patient demographics.

Case	Age (y)	BMI	Preoperative AHI	Preoperative O2 Nadir
1	50	27.2	50	77%
2	54	28.1	40	72%
3	60	35	87	66%

Demographic data for study patients include age, body mass index (BMI), preoperative apnea-hypopnea index (AHI), and preoperative oxygen saturation nadir (O₂ nadir).

In patient 1, bipolar IPG stimulation at 1.5–0.1V failed to demonstrate any activity of SG/HG (Figure 2A); however, when unipolar stimulation was applied at 0.1 V, clear mixed activation was present, as evidenced by tongue retraction with corresponding significant EMG activity of SG/HG (Figure 2B). Identification of this mixed activation led to electrode cuff repositioning with additional surgical dissection, exploration for retained retractor branches, and further separation at the proximal aspect of the fibers from the branches to be included in the cuff. After electrode repositioning, EMG mixed activation resolved as evidenced by depression of SG/HG activity and facilitation of GG activity (Figure 2C). Unrestricted forward tongue protrusion was also intraoperatively confirmed via direct visualization in response to IPG stimulation. On postoperative follow-up, the device was activated at a functional threshold voltage of 2.0 V under bipolar (+-+) settings, and unrestricted forward bilateral tongue protrusion was observed.

Figure 2. Patient 1.

(A) IPG stimulation under bipolar settings (0.3 V) did not clearly identify mixed activation. (B) IPG stimulation under unipolar settings (0.1 V) revealed mixed activation with significant SG/HG responses. (C) After further surgical dissection and cuff repositioning, mixed activation was significantly reduced under the same unipolar settings, as evidenced by a decrease in amplitude of SG/HG responses (post = 379 µV vs. pre = 520 µV) and an increase in amplitude of GG responses (post = 2,021 µV vs. pre = 173 µV). GG, genioglossus, HG = hypoglossus, IPG = implanted pulse generator, SG = styloglossus.

In patient 2, bipolar IPG stimulation at 1.5–0.1 V again failed to demonstrate mixed activation (Figure 3A); however, unipolar IPG stimulation at 0.3 V and lower revealed significant tongue retraction and clear SG/HG activity, indicative of mixed activation (Figure 3B). The cuff was repositioned by performing additional surgical dissection and separation of a retained distal retractor branch, as well as more distal placement on the nerve. After the cuff repositioning, the mixed-activation EMG pattern resolved (Figure 3C). Forward tongue protrusion was also confirmed by direct visualization. Upon postoperative follow-up, the device was activated at a functional threshold voltage of 1.0 V under bipolar (+-+) settings, and unrestricted forward right-tongue protrusion was observed.

Figure 3. Patient 2.

(A) IPG stimulation under bipolar settings (0.3 V) did not identify mixed activation. (B) IPG stimulation under unipolar settings (0.3 V) revealed mixed activation with large SG/HG responses. (C) After further surgical dissection and cuff repositioning, mixed activation was significantly reduced under the same unipolar settings, as evidenced by a decrease in amplitude of SG/HG responses (post = 491 µV vs pre= 4,095 µV), which led to GG responses being higher in amplitude compared with SG/HG responses (post GG = 675 µV vs post SG/HG= 491 µV). GG, genioglossus, HG = hypoglossus, IPG = implanted pulse generator, SG = styloglossus.

In patient 3, significant mixed activation was apparent only in response to unipolar IPG stimulation at 0.3 V (Figure 4A and Figure 4B). After further separation of proximal-exclusion nerve fibers and exploration for retained distal-retractor branch, the electrode cuff was repositioned over the inclusion-nerve branches. Thereafter, mixed activation was resolved as evidenced by decreased SG/HG-response amplitude and increased GG-response amplitude on EMG. Again, forward tongue protrusion was intraoperatively confirmed by direct visualization. Upon postoperative follow-up, the device was activated at a functional threshold voltage of 2.6 V under bipolar (+-+) settings, and unrestricted forward bilateral tongue protrusion was observed.

Figure 4. Patient 3.

(A) IPG stimulation under bipolar settings (0.3V) did not clearly identify mixed activation. (B) IPG stimulation under unipolar settings (0.3 V) revealed mixed activation (C) After further surgical dissection and cuff repositioning, mixed activation was significantly reduced under the same unipolar settings, as evidenced by a decrease in amplitude of SG/HG responses (post = 617 µV vs pre = 753 µV) and an increase in the amplitude of GG responses (post = 1,258 µV vs pre = 966 µV). GG = genioglossus = HG = hypoglossus, IPG = implanted pulse generator, SG = styloglossus.

DISCUSSION

Mixed activation during HGNS impairs tongue protrusion and worsens clinical outcomes.¹² Here we describe an expanded protocol for intraoperative neurophysiological testing that identifies significant mixed activation that is otherwise concealed by more traditional monitoring parameters. Identification of this mixed activation enabled further surgical dissection and successful repositioning of the electrode cuff, as illustrated by elimination of mixed activation, both clinically and electrophysiologically.

Mixed activation in HGNS is often due to the presence of “late takeoff” retractor nerve branches that are erroneously included in the electrode-stimulation cuff.⁹ The terminating branches of the HGN can be quite complex and variable between patients, making their identification by direct visualization alone technically difficult.¹⁴ Indeed, at least eight different major variations of these terminal branching patterns have been described.⁹ The neurophysiological monitoring protocol described here is able to detect the functional activity of such distal retractor branches and alert the surgeon to the fact that further surgical dissection may be necessary. In each of the patients described here, the hypoglossal nerve cuff was repositioned either by additional dissection of the proximal nerve with separation of retractor branches included in the stimulation cuff or by further dissection of the distal nerve with identification and exclusion of distal retractor branches.

Currently, intraoperative nerve integrity monitoring is widely used as means to guide selective electrode cuff placement.¹⁵ Standard protocols involve using a bipolar nerve integrity monitoring stimulation probe during nerve dissection to separate lateral HGN branches that produce tongue retraction from medial HGN branches that produce tongue protrusion. Even with this targeted approach, however, it is possible to miss significant amounts of mixed activation, which go unidentified until postoperative follow-up and therapy activation. It is now becoming clear that tongue-muscle activation during HGNS is highly dependent on stimulus polarity (bipolar vs unipolar) and intensity.¹³ Consistent with this, we find that in many cases mixed activation may not be visible under standard bipolar IPG configurations with high stimulation intensities that are traditionally used intraoperatively to assess tongue response to cuff stimulation. Indeed, in the three cases described here, using unipolar stimulation at low-stimulation intensities was essential for identifying mixed activation.

Why was mixed activation visible only under low-intensity unipolar IPG stimulation conditions in this subset of patients? We suspect that this phenomenon can be explained by differences in the spread of electrical current under unipolar versus bipolar IPG settings. Compared with bipolar stimulation, the broader electrical field generated by unipolar stimulation is more likely to stimulate all SG/HG retractor branches contained within the stimulation cuff. By contrast, bipolar stimulation is quite sensitive to even minor changes in cuff rotation/position, and therefore it may sometimes fail to elicit responses in lateral SG/HG branches erroneously contained within the stimulation cuff. Because of its broader electrical field, unipolar stimulation can also spread to nerve branches outside of the stimulation cuff (so-called far-field activation), but this is unlikely to be occurring at the low-stimulation intensities discussed in this article. This point is critical because the smooth-action potentials that may be generated by far-field nerve activation under high-intensity unipolar stimulation might drown out smaller polyphasic EMG signals on the SG/HG channel (ie, retractor activity). In other words, high-intensity unipolar stimulation might conceal retractor signals from detection. Instead, low-intensity unipolar stimulation may optimize the detection of these smaller polyphasic retractor signals on the SG/HG EMG channel.

The major limitation of this study is the small number of patients that exhibited hidden mixed-activation profiles in response to low-intensity, unipolar IPG stimulation (3/55). Additionally, no control group was included (ie, individuals who showed mixed-activation profiles under low-intensity unipolar settings who did not undergo cuff repositioning). These factors precluded rigorous analyses of the effects of cuff repositioning on treatment outcomes. Further work is necessary to determine whether the increased operative time required for expanded neurophysiological testing and cuff repositioning is truly clinically and/or cost-effective; however, there is a proportion of patients who have a suboptimal response to HGNS, and it has been reported that mixed activation can worsen clinical outcomes. Revision surgery is challenging and costly and should be avoided if possible. Therefore, it is logical to attempt to optimize selective stimulation for both bipolar and unipolar electrode configurations to the best of our ability during the initial implant surgery.

CONCLUSIONS

Based on our experience and data, we propose an expanded intraoperative IPG testing protocol to include bipolar and unipolar stimulation with both high and low intensities. Use of this protocol will maximize the surgeon’s chances of maximally detecting and avoiding unwanted mixed activation.

DISCLOSURE STATEMENT

All authors have seen and approved this manuscript. Joshua J. Sturm, Clara H. Lee, and Oleg Modik have no financial support or conflicts of interest to report. Maria V. Suurna is a post-approval and ADHERE clinical trial investigator for Inspire Medical and has received honoraria from the company. No payments or funding were received related to this research.

ABBREVIATIONS

AHI

apnea-hypopnea index

EMG

electromyography

GG

genioglossus

HGN

hypoglossal nerve

HGNS

hypoglossal nerve stimulation

IPG

implanted pulse generator

OSA

obstructive sleep apnea

PAP

positive airway pressure

SG/HG

styloglossus/hyoglossus