Transoral Versus Endoscopic Examination in Predicting Outcomes of Hypoglossal Nerve Stimulation for Obstructive Sleep Apnea

Abstract

Objectives/Hypothesis:

To examine the correlation between transoral and awake endoscopic examination and investigate their respective ability to predict outcomes of hypoglossal nerve stimulation (HGNS).

Study Design:

Retrospective cohort study at a US medical center.

Methods:

Subjects were adults with apnea-hypopnea index (AHI) >15 events/hr who underwent HGNS according to standard indications. Eligible subjects had diagnostic preoperative sleep studies, full-night efficacy postoperative studies, as well as postoperative video recordings of transoral examination and awake endoscopy. Recordings were independently scored by two blinded reviewers. Cohen’s κ coefficient, Student t test, and χ² analyses were performed.

Results:

Fifty-seven patients met all inclusion criteria. On average, patients were Caucasian, middle aged, and overweight. The mean preoperative AHI was 36.7 events/hr, which improved significantly to 18.3 events/hr following HGNS (P < .01). Overall, the response rate (defined as AHI reduction >50% and AHI < 20 events/hr) was 49%. There was slight correlation between transoral tongue protrusion and endoscopic tongue base movement (κ = 0.10). On transoral examination, patients with minimal/moderate tongue motion achieved a greater mean AHI reduction than patients with full motion (26.0 ± 18.0 vs. 12.8 ± 24.1, P = .02). In contrast, on awake endoscopy, patients with minimal/moderate tongue motion achieved a lesser mean AHI reduction than patients with full motion (8.7 ± 19.9 vs. 22.1 ± 22.7, P = .04).

Conclusions:

Transoral tongue protrusion bears an inverse relationship to HGNS success and correlates poorly with endoscopic tongue base movement. Endoscopic tongue base motion appears reflective of response to HGNS, with greater motion corresponding to greater AHI reduction.

INTRODUCTION

Obstructive sleep apnea (OSA) is a disorder characterized by recurrent collapse of the upper airway during sleep, leading to hypoxia and arousals. Patients with OSA are at increased risk for cardiovascular disease and neurocognitive dysfunction, among other health concerns.^1,2 Affecting 13% to 33% of men and 16% to 19% of women worldwide, OSA represents a significant public health burden.³ First-line therapy for OSA is positive airway pressure (PAP).⁴ PAP has been shown to decrease long-term cardiovascular risk and improve neurocognitive function.^5,6 Unfortunately, compliance is a barrier to effective therapy for many patients, with 46% to 83% of patients reporting noncompliance.⁷ Therefore, alternative therapies are essential.

Since its debut in 2014, hypoglossal nerve stimulation (HGNS) has presented a promising new treatment modality for PAP-intolerant patients.⁸ HGNS is composed of three surgically implanted components: a stimulation electrode on the hypoglossal nerve, pulse generator in the upper chest, and a thoracic respiratory sensor. When the thoracic monitor senses inspiration, a stimulus is delivered from the pulse generator to the hypoglossal nerve electrode. The resulting protrusion and stiffening of the tongue increase upper airway tone, preventing its collapse. Recent 5-year data from the Stimulation Therapy for Apnea Reduction trial demonstrated long-term reductions of apnea-hypopnea index (AHI) and improvement of sleep symptoms. Yet, the overall response rate (defined as AHI < 20 events/hr and >50% reduction of AHI) remains around 60%, leaving room for optimization.^9,10

It remains poorly understood why some patients benefit from HGNS, whereas others do not. Several studies have sought to identify patient characteristics associated with therapy success. Initial feasibility studies reported the greatest response rates in patients with AHI < 50 events/hr, body mass index (BMI) < 32 kg/m², and lack of complete circumferential palatal collapse on drug-induced sleep endoscopy (DISE). In 2019, registry data from 382 patients having completed a 12-month efficacy study showed a 69% responder rate with low BMI and female gender as favorable predictor variables.¹¹ Most recently, our group investigated therapeutic PAP level as a predictor for HGNS success, identifying a 92% success rate in patients with PAP values <8 cm H₂O, compared to 44% success in patients with higher PAP values.¹⁰

Currently, the standard for evaluation of tongue motion associated with HGNS is transoral examination. Intraoperatively, successful HGNS implantation is assessed by viewing the patient’s tongue motion transorally. During subsequent clinic visits, the transoral view is again used to evaluate tongue protrusion, direction, and stiffening. Although the transoral examination can be done expediently and without additional resources, it fails to view the tongue region implicated in OSA, the oropharyngeal tongue (i.e., tongue base). Transnasal endoscopy, both awake and under sedation, provides excellent examination of this region. Although tongue size on transoral exam does not appear to correlate with tongue obstruction on DISE,¹² there is no literature comparing transoral and awake endoscopic examination, particularly in the context of HGNS. In 2016, Heiser et al. assessed the tongue motion of 13 patients undergoing HGNS. They concluded that patients with bilateral or right-based protrusion demonstrated greater reductions in AHI compared to patients with mixed activation.¹³ In this way, awake tongue motion may be an important mediator of therapy success.

In this novel study, we tested two aims: to determine the correlation between transoral examination and awake endoscopy in postoperative HGNS patients and to compare outcomes across groups scored in regard to transoral and endoscopic tongue protrusion. We hypothesized that awake endoscopy would reveal different patterns of HGNS tongue movement than seen transorally and endoscopic findings would correlate more closely with therapy outcomes. Secondarily, we aimed to compare HGNS outcomes between groups categorized according to transoral tongue stiffening and direction.

MATERIALS AND METHODS

A prospective cohort study was approved by the Emory University Institutional Review Board (IRB00088402). Subjects were recruited from May 2016 to October 2018 at the sleep surgery clinic of the senior author (R.C.D.). A retrospective chart review was subsequently added to include patients from September 2015 to May 2016 to maximize sample size. Inclusion criteria were >18 years of age and HGNS for treatment of OSA. Indications for HGNS were AHI > 15 events/hr on the most recent diagnostic sleep study, PAP intolerance, and lack of complete circumferential palatal collapse on DISE. Patients were excluded if they were missing any of the following: preoperative diagnostic sleep study, full-night postoperative efficacy study, postoperative video recordings of transoral examination, and awake endoscopy.

Medical record extraction was performed by one of the authors (C.H.L.) and a trained research assistant. Data were manually entered into an Excel spreadsheet (Microsoft, Redmond, WA) and stored on a password-protected institutional server. Variables extracted from the medical record were age, sex, race, BMI, and past medical/surgical history. Variables extracted from sleep studies were AHI and 4% oxygen-desaturation index. Preoperative values were obtained by averaging values from the most recent diagnostic sleep study and all studies obtained within 3 years prior to HGNS. Postoperative values were extracted from efficacy studies, which were obtained when the senior author deemed a patient to be optimized on therapy. Efficacy studies using WatchPAT 200 (Itamar Medical, Caesarea, Israel) represent a full night of sleep at a single device setting. When possible, HGNS use was verified through Inspire Cloud (Inspire Medical, Golden Valley, Minnesota, MN, USA), software that allows remote monitoring of HGNS compliance.

Transoral examination and awake endoscopy was performed at clinic visits approximately 3 months following surgical implantation. With the patient seated in a reclined position (approximately 45°) in an examination chair, serial stimulation pulses were delivered at the therapeutic amplitude. Transoral video exams were recorded en face, with the patient breathing nasally and his/her mouth comfortably open. Subsequently, following topical decongestion and anesthesia, awake endoscopy was performed using a flexible fiberoptic laryngoscope, recorded to a digital recorder. Endoscopic exams generally involved two to four serial stimulations, visualizing the effect of HGNS pulses on the velum, oropharynx, tongue, and epiglottis.

Exams were independently reviewed by two of the authors (C.H.L. and G.B.M.). Prior to scoring, the reviewers performed five joint reviews to identify any discrepancies in interpretation of scoring criteria. Reviewers were blinded to patient outcomes through the use of deidentified patient data and lack of clinical familiarity with the patients. For the transoral view, three measures were assessed: degree of tongue protrusion, direction of tongue protrusion, and presence of tongue stiffening. For protrusion, each reviewer provided an ordinal score of 0 (minimal movement), 1 (moderate movement), or 2 (full movement). The direction of protrusion was determined to be right-based protrusion (RP), left-based protrusion (LP), bilateral protrusion, or mixed protrusion. RP and LP were further categorized as ipsilateral or contralateral based on the side of the implant documented in the operative note. Tongue stiffening based on elongation and flattening during stimulation was scored to be present or absent. For the endoscopic view, four anatomical locations were examined: velum, oropharynx, tongue base, and epiglottis. At each location, the degree of motion was determined to be 0 (minimal movement), 1 (moderate movement), and 2 (full movement). Discrepancies between the reviewers were resolved through mutual review and consensus.

The distribution of transoral protrusion and endoscopic movement scores stratified naturally into two groups: 0 or 1, representing minimal/moderate motion; and 2, representing full motion. After assessment for normality, the difference between groups was calculated with a Student t test for continuous variables and χ² test for categorical variables. A κ analysis was used to evaluate interrater reliability, as well as the correlation between final scoring of transoral examination and awake endoscopy (after mutual review and consensus).

RESULTS

Fifty-seven patients met all study criteria. Their demographics and baseline polysomnographic measurements are described in Table I. On average, patients were Caucasian, middle-aged, and obese. The most common non-OSA sleep disorders were insomnia (n = 20) and restless leg syndrome (n = 6). Twenty-seven patients had prior upper airway surgery, including 13 tonsillectomies, 11 nasal surgeries, seven palatal surgeries, and one maxillomandibular advancement. Of the seven patients with neurologic disease, three had trisomy 21, two had prior stroke without residual deficits, and two had mild dementia. Twenty-three patients had moderate OSA (AHI = 15–29.9 events/hr) and 34 had severe OSA (AHI ≥ 30 events/hr).

TABLE I.

Cohort Characteristics (N = 57).

Characteristics	Value
Demographics
Male sex, %	54
Caucasian race, %	86
Age, yr	62.2 ± 14.3
Body mass index, kg/m2	28.4 ± 4.1
Non-OSA sleep disorder, %	40
Prior upper airway surgery, %	49
Neurologic disease, %	12
Polysomnogram
Apnea-hypopnea index, events/hr	36.7 ± 17.3
4% oxygen desaturation index, events/hr*	30.1 ± 19.0

Values represent mean ± standard deviation, unless otherwise stated.

^*n = 50, not reported on all sleep tests.

Interrater reliability results are displayed in Table II. All measurements demonstrated >50% agreement. Substantial interrater correlation was observed with the transoral examination of tongue protrusion (κ = 0.78).¹⁴ Moderate correlation was observed with transoral tongue direction (κ = 0.53) and endoscopic epiglottic motion (κ = 0.62). Transoral tongue stiffening (κ = 0.24) and endoscopic tongue base motion (κ = 0.26) demonstrated fair interrater correlation. Finally, slight correlations were observed endoscopic velum motion (κ = 0.15) and oropharynx motion (κ = 0.03). Exploratory analyses revealed similar degrees of interrater reliability when scores were analyzed individually (i.e., without stratification into minimal/moderate and full movement groups).

TABLE II.

Interrater Reliability for Transoral and Endoscopic Examinations.

	Agreement, %	κ
Transoral
Protrusion	89	0.78
Direction	70	0.53
Stiffening	63	0.24
Endoscopic
Velum	58	0.15
Oropharynx	51	0.03
Tongue base	67	0.26
Epiglottis	82	0.62

A slight correlation was observed between transoral tongue protrusion and endoscopic tongue base movement (κ = 0.10). Fifty-eight percent of transoral tongue protrusion scores were in agreement with their endoscopic counterpart. The distribution of endoscopic motion scores at all anatomic sites is displayed in Figure 1.

Fig. 1.

Distribution of endoscopic motion scores at four anatomic sites. Videos were scored as 0 (minimal movement), 1 (moderate movement), or 2 (full movement), then naturally stratified into two groups. Dark bars indicate the number of scores in agreement with their transoral counterpart; light bars indicate the number of values in disagreement.

Following HGNS, the overall cohort achieved a significant AHI reduction from 36.7 events/hr to 18.3 events/hr (P < .01). The mean AHI reduction (preoperative – postoperative value) was 18.3 events/hr. Mean AHI reduction is compared across all scoring groups in Table III. On transoral examination, patients with minimal/moderate tongue motion achieved a significantly larger AHI reduction than the full motion group (26.0 ± 18.0 vs. 12.8 ± 24.1, P = .02). In contrast, patients with minimal/moderate endoscopic tongue base motion achieved a lesser AHI reduction than patients with full motion (8.7 ± 19.9 vs. 22.1 ± 22.7, P = .04). Between all other scoring groups, there were no significant differences in AHI reduction.

TABLE III.

Comparison of Apnea-Hypopnea Index Reduction (Events/Hour) Between Minimal/Moderate Motion (0–1) and Full Motion (2) Scoring Groups.

	Minimal/Moderate Motion (0–1)	No.	Full Motion (2)	No.	P Value	Power
Transoral Examination
Tongue	26.0 ± 18.0	24	12.8 ± 24.1	33	.02	0.42
Awake Endoscopy
Velum	19.3 ± 22.8	26	17.5 ± 22.7	31	.77	0.36
Oropharynx	18.7 ± 22.1	36	17.6 ± 23.9	21	.86	0.33
Tongue base	8.7 ± 19.9	16	22.1 ± 22.7	41	.04	0.35
Epiglottis	17.3 ± 17.9	22	19.0 ± 25.3	35	.76	0.41

Values represent mean reduction (preoperative value − postoperative value) ± standard deviation. Power to detect a difference of 10 events/hour.

The overall response rate (defined as AHI reduction >50% and AHI < 20 events/hr) was 49%. On transoral examination, patients with minimal/moderate tongue protrusion demonstrated a trend toward greater response rate than the full motion group (63% vs. 39%, P = .08). There were no significant differences in response rates between all endoscopic scoring groups. Notably, patients with minimal/moderate endoscopic tongue base motion achieved similar response rates with the full motion group (44% vs. 51%, P = .61).

Secondarily, there were no differences in HGNS outcomes between groups categorized according to transoral tongue stiffening or direction. The tongue-stiffening group (n = 32) and the nonstiffening group (n = 25) achieved similar mean AHI reductions (21.0 ± 23.5 vs. 14.9 ± 21.3, P = .31), as well as similar response rates (48% vs. 50%, P = .88). The direction of tongue protrusion, categorized as bilateral (n = 22), ipsilateral (n = 27), or mixed (n = 8), also did not reveal differences in mean AHI reduction (22.5 ± 18.8 vs. 11.4 ± 24.8 vs. 20.5 ± 19.3, P = .20) or response rate (55% vs. 48% vs. 38%, P = .70).

DISCUSSION

This is the first study to examine the relationship between transoral examination and awake endoscopy in the evaluation of HGNS. There was a weak correlation between transoral examination of tongue protrusion and endoscopic tongue base motion. Although the entire cohort achieved significant AHI reductions with HGNS, the greatest reductions were seen in patients with minimal/moderate tongue protrusion on transoral examination, and full tongue base motion on awake endoscopy.

Although data comparing awake endoscopy to transoral examination are not available, our observed findings are in accordance with previous studies comparing transoral examination with DISE. In their study of 138 patients with OSA (mean age = 42.9 years, BMI = 27.4 kg/m², AHI = 29.4 events/hr), Zerpa Zerpa et al. compared the incidence of severe upper airway collapse (defined as occluding >75% of the airway) identified during DISE between modified Mallampati indices (MMI) and Friedman stages.¹⁵ They found no statistically significant differences in retrolingual collapse between MMI groups (P = .58) or Friedman stages (P = .54). Similar to our study, the κ statistic was used to evaluate the concordance between endoscopy and transoral examination. The low values for MMI (κ = 0.007) and Friedman stage (κ = 0.019) suggest poor agreement between the two modalities, consistent with the findings in our study.

It is peculiar to note the superior outcomes in patients with minimal/moderate tongue protrusion on transoral examination, as one would expect a greater response in patients with full tongue motion. One plausible explanation for this finding is that most responders exhibit early success (i.e., at or near functional threshold), whereas nonresponders may be asked to continue increasing therapy amplitude in hopes to identify a therapeutic amplitude. Our extensive experience with HGNS titrations demonstrates that unlike PAP therapy, HGNS titrations are not linear. In this way, our finding that patients with greater transoral protrusion have less AHI reduction is consistent with our clinical experience. Another explanation for our paradoxical finding is that pure tongue protrusion is not ideal for airway dilation; rather, a balance of protrusion and retraction results in optimal airway dilation.^16,17 One final explanation stems from the muscular recruitment patterns from hypoglossal nerve stimulation. It is possible that some patients experienced tongue protrusion from the transverse vertical fibers without sufficient genioglossus recruitment. In this way, tongue base displacement would be absent despite the adequate tongue displacement on anterior/transoral view. Our results related to awake endoscopic motion dovetail on a recent radiographic study of HGNS success. Schwab et al. utilized awake computed tomography to identify anatomical differences between HGNS patients.¹⁸ Therapy responders were found to have more tongue displacement than nonresponders as well as less soft palate volume, greater increase in retroglossal airway size, and increased shortening of the mandible–hyoid distance with therapy activation. Specifically, the greatest difference in movement between responders and nonresponders was seen in the posterior inferior regions of the tongue. These findings reinforce the importance of tongue base displacement, not necessarily oral tongue displacement, for successful outcomes.

We acknowledge several limitations of our study. First, the interrater reliability is weak at several anatomical locations, including the endoscopic tongue base (κ = 0.26). This reflects the difficulty of obtaining uniform consensus on endoscopic exams, particularly at the level of the oropharyngeal lateral walls.¹⁹ Second, as individual patients had different tolerances to endoscopy, the degree of protrusion may reflect patient tolerance as opposed to physiologic effect. Nonetheless, patient tolerance of stimulation represents a practical limitation of this therapy. Third, we did not assess subjective measure as patient outcomes, which may have shown benefit not captured by polysomnography. Fourth, we had limited power. Finally, we excluded five patients due to lack of postoperative efficacy study and four patients due to lack of video recordings. Patients lost to follow-up may represent outstanding responders who do not feel a need to follow-up, or extremely poor responders who do not desire follow-up.

Our study also has several strengths. Our transoral and awake endoscopic examinations were prospectively recorded in a standardized fashion across all patients. We utilized full-night efficacy studies at a single setting as our dependent variable, providing accurate HGNS results.²⁰ When home sleep studies were performed, Inspire Cloud was used to verify HGNS use during these studies. Additionally, our study encompasses the practice of a single surgeon, ensuring consistency in surgical decision making, surgical technique, and perioperative care.

CONCLUSION

Our study indicates that transoral examination of HGNS correlates weakly with awake endoscopy. Awake transoral findings should not be used to infer therapy success; conversely, awake endoscopic examination of the tongue base may predict optimal HGNS response. These findings suggest that our current standard, transoral examination, cannot be used to deduce tongue base motion. Instead, direct examination of the tongue base should be performed via transnasal endoscopy to measure tongue displacement.