Abstract
Objective.
Pharyngeal surgery is a treatment option for patients with obstructive sleep apnea (OSA) unable to tolerate positive pressure therapy. This study aims to determine the association between palate shape as described by Woodson and pharyngeal surgical outcomes.
Study Design.
Exploratory analysis of retrospective cohort.
Setting.
Multicenter.
Methods.
Three blinded reviewers assessed palate shape using drug-induced sleep endoscopy (DISE) videos from a previously-assembled cohort of adults undergoing pharyngeal surgery. Palate shape scores were examined for association with surgical outcomes with univariate and multivariate analyses. Multivariate analyses included adjustment for consensus DISE findings determined previously.
Results.
Two hundred nine study subjects were included from 13 centers. Age was 53.7 ± 11.5 years, body mass index (BMI) was 30.3 ± 5.0 kg/m2, and 21% were female. In isolated soft palate surgery, greater GenuAP narrowing was associated with lesser odds of surgical response, whereas greater GenuLW narrowing was associated with greater odds of surgical response. These findings largely persisted after adjustment for key DISE findings, age, gender, OSA severity, BMI, and tonsil size. Other palate-shape findings were not clearly associated with surgical outcomes, although some palate-shape findings demonstrated trends toward an association with outcomes (P < .10).
Conclusion.
Greater GenuAP narrowing and GenuLW narrowing were associated with lesser and greater, respectively, odds of surgical response after isolated soft palate surgery. Palate shape and other palate shape level scores were not clearly associated with surgical outcomes. Larger studies may determine more precisely the association between palate shape and pharyngeal surgery outcomes.
Keywords: obstructive sleep apnea, palate surgery, pharyngeal surgery, sleep apnea surgery
Pharyngeal surgeries provide important alternatives for patients with obstructive sleep apnea (OSA) unable to tolerate positive airway pressure therapy. With various surgical options available, patient and procedure selection are key to optimizing outcomes. Prior literature has demonstrated that the following factors are associated with surgical outcomes: body mass index (BMI), tonsil size, Friedman tongue position, and drug-induced sleep endoscopy (DISE) findings.1–10
Studies incorporating cephalometric data and palate surgery have not shown clear association with outcomes.11 In the present study, we utilize a DISE-based assessment of the airway. DISE offers a dynamic view of the airway which is not appreciated with cephalometric data. Further, DISE captures collapse patterns akin to what is seen in natural sleep.12
In a previous multicenter cohort study of individuals undergoing pharyngeal surgery after DISE, Green et al5 showed that partial or complete lateral oropharyngeal wall collapse and complete tongue-related obstruction were each associated with lower odds of surgical response after pharyngeal surgery.
Palate shape (Figure 1) has been proposed by Woodson13,14 as a potential predictor of surgical response. Palate shape (vertical, intermediate, and oblique) is defined by airway openness at the levels of the hard palate (HP; posterior to the posterior aspect of the HP, approximated by the nasal septum), genu of the soft palate (independent anteroposterior and lateral components), and velum (soft palate distal to the genu). A previous multicenter cohort showed vertical palate shape and narrow HP score (especially with complete oropharyngeal lateral wall obstruction during DISE) were associated with unilateral hypoglossal nerve stimulation surgical outcomes.15
Figure 1.
These are pictorial demonstrations of the 3 palate shapes, oblique, intermediate, and vertical from left to right.15
This study aims to explore the association between palate shape and palate shape levels as described by Woodson and pharyngeal surgery outcomes.
Methods
This was an exploratory analysis of retrospective multicenter cohort study of adults undergoing pharyngeal surgery for the treatment of OSA. The study was approved by the Institutional Review Board (IRB) or similar committee of each participating center (University of Colorado IRB 19–0284, University of Minnesota IRB 1509M78221, University of Pennsylvania IRB 822006, Medical College of Wisconsin IRB PRO00034722, University of Pittsburgh PRO15020362, Vanderbilt University IRB 181496, University of Southern California IRB HS-14–00790, Thomas Jefferson University IRB 16D.238, Weill Cornell Medicine IRB 19–04020166, Wayne State University IRB PRO1901001962, Middlesex Hospital IRB FWA 00005937, Western University Health Science Research Ethics Board IRB 00000940, Henry Ford Health System IRB 9657, Medical University of South Carolina Pro00045593, Orange Regional IRB, no such approvals were required for European centers as de-identified data was sent which was exempt from review).
The study is an exploratory secondary analysis of a previously-assembled cohort,5 consisting of adults ≥ 21 years old, with OSA, no previous pharyngeal surgery (other than tonsillectomy and/or adenoidectomy), tonsils sized 0 to 2+ (Brodsky16 or Friedman et al1 classifications, according to center standard practice), who underwent preoperative DISE and preoperative and postoperative sleep studies.
Demographic information including age, gender, race, BMI, type of pharyngeal surgery, and pre- and postoperative apnea-hypopnea index (AHI) was previously obtained. DISE videos were stored in a Health Insurance Portability and Accountability Act-compliant database (Box.com; Box Inc). Consensus scores for VOTE classifications and primary structure of obstruction were determined in the previous study by 4 blinded reviewers.5
DISE videos were examined by 3 blinded reviewers(E. A. C., S. K., A. E. K.) to examine palate shape, according to the Woodson classification.13 Palate levels were each scored during the most open state (nonapneic episodes) during restful, rhythmic breathing determined by lack of swallowing and periodic soft palate motion with or without snoring, as in our previous similar study.15 The goal was determination of palate shape at a resting state, typically at end-expiration,13 and without dynamic movement associated with intrapharyngeal pressure changes with inspiration and expiration.
Three palate levels were scored by the anterior-posterior opening between the structure and posterior pharyngeal wall from 1 to 3, where 1 was the most open and 3 was the narrowest for that palate level. Anterior posterior variables included: HP, as approximated by the posterior aspect of the nasal septum; genu (GenuAP), as defined by the inflection point of the soft palate and velum; and velum, as defined by the soft palate distal to the genu. Additionally, transverse opening at the genu (GenuLW) was scored similarly. Of note, palate shape assessments are focused on airway opening (as opposed to DISE assessments that consider degree of airway narrowing).
Reviewers scored the first 10 videos as a training sample, with discussion of score differences, followed by a rescoring of these initial videos and the remainder of the videos. Subjects were excluded if, on initial review, 2 reviewers felt a DISE video could not be scored on at least 1 palate level. If 1 reviewer felt at least 1 palate level could not be scored, the video was reviewed and rescored independently. If after rereview, any single reviewer still felt a palate level could not be scored, the entire DISE video was excluded.
A priori criteria were used for obtaining consensus scoring of palate levels. For each palate level score, if at least 2 reviewers agreed on the score, the mode was assigned as the score for that level. In cases of no agreement, a score of 2 (the mean score) was assigned to that palate level.
Palate shape was determined from 3 individual palate-level consensus scores, as previously described.13,15 Vertical palate shape was assigned to subjects with the greatest airway narrowing at all levels, reflected by palate level scores of 3/3/3 (HP/GenuAP/velum). Oblique palate shape was assigned to subjects with a combination of HP score of 1, GenuAP score of 1 or 2, and any velum score. All other subjects were considered to have an intermediate palate shape.
Surgical response was the primary outcome and defined by at least a 50% reduction in postoperative AHI to below 15 events/h. This represents somewhat stricter criteria than the Sher criteria (reduction in postoperative AHI to below 20 events/h).17
As with original cohort,5 isolated palate surgery included traditional uvulopalatopharyngoplasty, modified palate surgery (eg, expansion sphincter pharyngoplasty, lateral pharyngoplasty, Z-palatoplasty), palatal advancement, or palate stiffening. Because the technique was often unreported or unclear, specific analysis of palate surgery technique was not performed.
Continuous variables were reported as means and standard deviation, and categorical variables were reported as frequencies and percentages. Continuous variables were compared using paired t-tests for within-group changes, and categorical variables were compared using Pearson chi-squared tests. Univariate analyses evaluating the association between palate shape variables included linear and logistic regression for continuous and categorical variables, respectively. Testing examined the overall cohort and subgroups defined by type of surgery and DISE findings. Analyses included comparison of individual palate level scores (eg, 1 vs 2 vs 3) and grouping (eg, 1/2 vs 3). Logistic regression models examined palate shape findings and surgical response in the overall cohort and subgroups, with adjustment for potential confounders: age, gender, BMI, tonsil size, OSA severity, and key DISE findings (as used in the previous study using this cohort).5 P < .05 were considered statistically significant, and P values .05 to .1 were considered as statistical trends. Interrater reliability of palate level scores was determined by the Scott/Fleiss κ statistic.
Results
The study included 209 of the 275 subjects in the cohort.5 Sixty-six subjects were excluded due to inadequate view of 1 or more palate levels (Table 1). Those who were excluded were younger and more likely to undergo nasal surgery, palate surgery (including possible tonsillectomy), genioglossus advancement, tongue radiofrequency, or hyoid suspension and less likely to undergo transoral robotic surgery, partial glossectomy, or epiglottis surgery (P < .05). Rates of exclusion varied by institution ranging from 0/5 at University of Pennsylvania to 34/43 subjects from the University of Southern California.
Table 1.
Demographic and Baseline Characteristic Differences Between Excluded Versus Included Subjects
| Included subjects, N (%) | Excluded subjects, N (%) | P value | |
|---|---|---|---|
| Sex (female) | 44 (21%) | 9 (14%) | .19 |
| Age (mean ± SD) | 53.7 ± 11.5 | 47.3 ± 12.0 | .001 |
| Race | .012 | ||
| White | 163 (78%) | 57 (86%) | |
| Black | 8 (4%) | 1 (2%) | |
| Asian | 3 (1%) | 7 (11%) | |
| Other | 1 (0.5%) | 0 (0%) | |
| BMI (mean ± SD) | 30.3 ± 5.0 | 29.7 ± 5.8 | .43 |
| Tonsil size 0 | 83 (40%) | 19 (29%) | .25 |
| Tonsil size 1+ | 77 (37%) | 27 (41%) | |
| Tonsil size 2+ | 49 (23%) | 20 (30%) | |
| Preop AHI (events/h) (mean ± SD) | 41.4 ± 24.3 | 40.3 ± 24.9 | .74 |
| Mild OSA (5–15) | 22 (11%) | 8 (12%) | .92 |
| Moderate (15–30) | 62 (30%) | 20 (30%) | |
| Severe OSA (>30) | 125 (60%) | 38 (58%) | |
| Change in AHI (mean ± SD)a | 20.7 ± 23.0 | 19.2 ±26.5 | .66 |
| Surgical responseb | 82 (39%) | 31 (47%) | .27 |
Abbreviations: AHI, apnea-hypopnea index; BMI, body mass index; h, hour; OSA, obstructive sleep apnea; SD, standard deviation.
Change in AHI represented as an absolute number, with both groups showing decrease in AHI.
Surgical response defined as at least a 50% reduction from preoperative AHI and reduction in events to less than 15 per hour.
The majority of subjects had moderate-to-severe OSA (AHI ≥15 events/h).
Overall, there was minimal to fair interrater reliability in scoring for all palate shape levels (HP = 0.237; GenuAP = 0.264; velum = 0.271, and GenuLW = 0.39).
Consensus scores for palate level and overall palate shape distribution are shown in Tables 2 and 3. The majority of subjects had HP, GenuAP, and GenuLW scores of 1 or 2 and an overall intermediate palate shape. Table 4 shows DISE findings, including VOTE scoring18 and primary level of obstruction for included subjects; a majority of subjects had primary obstruction related to the velum or tongue.
Table 2.
Palate Level Score Distribution for Included Subjects
| Score | Hard palate, N (%) | GenuAP, N (%) | Velum, N (%) | GenuLW, N (%) |
|---|---|---|---|---|
| 1 | 107 (51%) | 36 (17%) | 6 (3%) | 114 (55%) |
| 2 | 80 (38%) | 123 (59%) | 72 (34%) | 76 (36%) |
| 3 | 22 (11%) | 50 (24%) | 131 (63%) | 19 (9%) |
Table 3.
Palate Shape Distribution
| Palate shape | N (%) | Baseline AHI (SD) | Change in AHIa (SD) | Surgical response,b N (%) |
|---|---|---|---|---|
| Oblique | 88 (42%) | 40.6 (22.0) | 20.5 (19.9) | 32 (36%) |
| Intermediate | 113 (54%) | 41.1 (24.1) | 20.2 (24.3) | 48 (42%) |
| Vertical | 8 (4%) | 55.5 (44.1)c | 28.7 (36) | 2 (25%) |
Abbreviations: AHI, apnea hypopnea index; SD, standard deviation.
Changes in AHI are presented as absolute numbers with all groups having a decrease in AHI.
Surgical response defined as at least a 50% reduction from preoperative AHI and reduction in events to less than 15 per hour.
Baseline AHI for vertical palate shape compared to other palate shapes P = .09.
Table 4.
VOTE Classification Findings and Primary Structure of Obstruction
| Configuration | |||||
|---|---|---|---|---|---|
| Structure | AP | Lateral | Concentric | Degree of obstructiona | Primary Structure,b N (%) |
| Velum | 116 | 3 | 80 | 5/38/166 | 69 (35%) |
| Oropharynx lateral walls | 93/68/48 | 43 (22%) | |||
| Tongue Base | 63/63/83 | 79 (40%) | |||
| Epiglottis | 171/19/19 | 4 (2%) | |||
Note: Numbers in cells reflect number of study subjects with degrees of obstruction 0/1/2 for each structure. Shaded cells reflect combinations of structure and configuration that are not possible in the VOTE Classification. For the velum, the numbers are also presented for each configuration.
Abbreviation: AP, anterior-posterior.
Degrees of obstruction are defined by 0 is less than 50% obstruction, 1 is 50%−75% obstruction, and 2 is greater than 75% obstruction.19
Primary structure of airway obstruction was previously determined by prior study.5
Univariate analysis of individual palate shape level scores (Tables 5 and 6) and overall palate shape (Table 7) showed that there were no statistically-significant associations between palate shape and outcomes (change in AHI or surgical response) for the entire cohort.
Table 5.
Change in AHI and Palate Level Scores in Cohort
| Palate level score | Hard palate change in AHI (SD) | GenuAP change in AHI (SD) | Velum change in AHI (SD) | GenuLW change in AHI (SD) |
|---|---|---|---|---|
| 1 vs 2 vs 3 | 20.3 (20.1) vs 19(23.0) vs 28.1 (33.6) | 27.7 (22.5) vs 19.3 (22.1) vs 19.0 (24.9) | 14.6 (18.2) vs 24.1 (21.7) vs 19.0 (23.8) | 18.8 (22.3) vs 22.8 (22.8) vs 23.2 (27.7) |
| 1 vs 2 and 3 | 20.3 (20.1) vs 21.0 (25.7) | 27.7 (22.5) vs 19.2 (22.9) | 14.6 (18.2) vs 20.8 (23.1) | 18.8 (22.3) vs 22.9 (23.7) |
| 3 vs 1 and 2 | 28.1 (33.6) vs 19.8 (21.4) | 19.0 (24.9) vs 21.2 (22.4) | 19.0 (23.8) vs 23.4 (21.5) | 23.2 (27.7) vs 20.4 (22.5) |
| 3 vs 1 | 28.1 (33.6) vs 20.3 (20.1) | 19.0 (24.9) vs 27.7 (22.5) | 14.6 (18.2) vs 19.0 (23.8) | 23.2 (27.7) vs 18.8 (22.3) |
Note: Changes in AHI are presented as absolute numbers with all groups having a decrease in AHI. t tests for differences all P > .1.
Abbreviations: AHI, apnea-hypopnea index; SD, standard deviation.
Table 6.
Palate Level Scores and Surgical Response in Overall Cohort
| Palate level score | Hard palate, N (%) | GenuAP N (%) | Velum, N (%) | GenuLW, N (%) |
|---|---|---|---|---|
| 1 vs 2 vs 3 | 40 (37%) vs 35 (44%) vs 7 (32%) | 18 (50%) vs 49 (40%) vs 15 (30%)a | 3 (50%) vs 33 (46%) vs 46 (35%) | 45 (39%) vs 29 (38%) vs 8 (42%) |
| 1 vs 2 and 3 | 40 (37%) vs 42 (41%) | 18 (50%) vs 64 (37%) | 3 (50%) vs (39%) | 45 (39%) vs 37 (39%) |
| 3 vs 1 and 2 | 7 (32%) vs 75 (40%) | 15 (30%) vs 67 (42%) | 46 (35%) vs 36 (46%) | 8 (42%) vs 74 (39%) |
| 3 vs 1 | 7 (32%) vs 20 (37%) | 15 (30%) vs 18 (50%)a | 3 (50%) vs 46 (35%) | 8 (42%) vs 45 (39%) |
Note: Surgical response defined as at least a 50% reduction from preoperative AHI and reduction in events to less than 15 per hour.
P < .1.
Table 7.
Palate Shape and Outcomes (Change in AHI and Surgical Response) in Overall Cohort
| Palate shape | Change in AHIa (SD) | Proportion of surgical responseb |
|---|---|---|
| Oblique vs intermediate vs vertical | 20.5 (19.9) vs 20.2 (24.3) vs 28.7 (36.0) | 32 (36%) vs 48 (42%) vs 2 (25%) |
| Oblique vs intermediate and vertical | 20.5 (19.9) vs 20.8 (25.1) | 32 (36%) vs 50 (41%) |
| Vertical vs oblique and intermediate | 28.7 (36.0) vs 20.3 (22.4) | 2 (25%) vs 80 (40%) |
| Vertical vs oblique | 28.7 (36.0) vs 20.5 (19.9) | 2(25%) vs 32 (36%) |
Note: All P > .10.
Abbreviations: AHI, apnea-hypopnea index; SD, standard deviation.
Change in AHI represented as an absolute number, with both groups showing decrease in AHI.
Surgical response defined as at least a 50% reduction from preoperative AHI and reduction in events to less than 15 per hour.
Among study subjects undergoing isolated soft palate surgery (n = 76), Table 8 presents results of univariate analyses for the association with surgical response. A 1-unit increase in GenuAP score was associated with a decrease in the odds of surgical response (odds ratio: 0.44; 95% confidence interval [CI]: 0.20, 0.99). Compared to those with a GenuLW score of 1 or 2, a GenuLW score of 3 was associated with an increase in the odds of surgical response (odds ratio: 4.4; 95% CI: 1.03, 18.5). There was no association between surgical response and other palate shape level scores or overall palate shape. There was no association between palate shape level scores or overall palate shape and surgical response in other subgroups defined by procedure types or key DISE findings (data not shown).
Table 8.
Palate Level Scores and Surgical Response in Isolated Palate Surgery Subgroup
| Palate level score | Hard palate, N (%) | GenuAP, N (%) | Velum, N (%) | GenuLW, N (%) |
|---|---|---|---|---|
| 1 vs 2 vs 3 | 12 (36%) vs 14 (41%) vs 4 (44%) | 8 (57%) vs 19 (40%) vs 3 (20%)a | 0 vs 16 (52%) vs 14 (32%) | 9 (33%) vs 14 (36%) vs 7 (70%)b |
| 1 vs 2 and 3 | 12 (36%) vs 18 (42%) | 8 (57%) vs 22 (35%) | 0 vs 30 (40%) | 9 (33%) vs 21 (43%) |
| 3 vs 1 and 2 | 4 (44%) vs 26 (39%) | 27 (44%) vs 3 (20%)b | 14 (32%) vs 16 (50%) | 7 (70%) vs 23 (35%)a |
| 3 vs 1 | 12 (36%) vs 4 (44%) | 3 (20%) vs 8 (57%)a | 14 (32%) vs 0 | 7 (70%) vs 9 (33%)b |
Note: Surgical response defined as at least a 50% reduction from preoperative AHI and reduction in events to less than 15 per hour.
Abbreviation: AHI, apnea-hypopnea index.
P < .05.
P < .1.
Table 9 presents results of multiple regression analyses, with adjustment for key DISE findings (identified in Green at al5) and then with additional other potential confounders. The patterns seen in univariate analyses were generally unchanged, with similar point estimates for the associations between surgical response and both GenuAP and GenuLW, although with greater P values.
Table 9.
Logistic Regression Model for Surgical Response with Palate Shape and Subgroups Defined by Surgical Procedure
| Adjusted for DISE variablesa | Adjusted for DISE variables and other confoundersb | |||
|---|---|---|---|---|
| Procedure | Odds ratio | 95% CI | Odds ratio | 95% CI |
| Overall | ||||
| Hard palate | 1.0 | 0.67–1.6 | 0.95 | 0.61–1.5 |
| GenuAP | 0.65 | 0.40–1.0* | 0.66 | 0.41–1.1 |
| Velum | 0.69 | 0.41–1.2 | 0.61 | 0.34–1.1* |
| GenuLW | 1.3 | 0.78–2.1 | 1.3 | 0.75–2.4 |
| Moderate-to-severe OSA | ||||
| Hard palate | 1.1 | 0.68–1.7 | 1.1 | 0.66–1.7 |
| GenuAP | 0.65 | 0.40–1.05* | 0.66 | 0.40–1.1 |
| Velum | 0.76 | 0.43–1.3 | 0.69 | 0.38–1.2 |
| GenuLW | 1.3 | 0.77–2.2 | 1.4 | 0.75–2.5 |
| Isolated palate | ||||
| Hard palate | 1.2 | 0.59–2.5 | 1.0 | 0.44–2.5 |
| GenuAP | 0.44 | 0.18–1.0* | 0.38 | 0.13–1.1* |
| Velum | 0.58 | 0.22–1.5 | 0.42 | 0.13–1.4 |
| GenuLW | 2.0 | 0.89–4.4* | 1.6 | 0.57–4.6 |
| Palate + tongue | ||||
| Hard palate | 1.1 | 0.59–1.9 | 1.2 | 0.60– 2.4 |
| GenuAP | 1.2 | 0.50–2.2 | 1.2 | 0.60–2.7 |
| Velum | 1.0 | 0.46–2.2 | 1.2 | 0.47–3.0 |
| GenuLW | 0.83 | 0.38–1.8 | 0.58 | 0.23–15 |
| Palate + tongue resection | ||||
| Hard palate | 1.1 | 0.57–1.9 | 1.2 | 0.61–2.4 |
| GenuAP | 1.1 | 0.56–2.3 | 1.2 | 0.55–2.8 |
| Velum | 1.1 | 0.56–2.3 | 1.3 | 0.50–3.3 |
| GenuLW | 0.84 | 0.37–1.9 | 0.63 | 0.23–1.7 |
| Epiglottis ± other | ||||
| Hard palate | 0.55 | 0.17–1.7 | 0.62 | 0.076–5.2 |
| GenuAP | 0.20 | 0.054–0.74** | 0.036 | 0.0014–0.89** |
| Velum | 0.71 | 0.19–2.6 | 0.37 | 0.47–2.9 |
| GenuLW | 0.77 | 0.21–2.8 | 0.97 | 0.096–9.9 |
Abbreviations: BMI, body mass index; CI, confidence interval; DISE, drug-induced sleep endoscopy; OSA, obstructive sleep apnea.
DISE variables included: complete velum obstruction, any oropharyngeal lateral wall obstruction, complete tongue-based obstruction, and any epiglottis obstruction.
Other confounders included: Age, gender, BMI, OSA severity group, and tonsil size.
P < .1.
P < .05.
Discussion
This is the first DISE database study to explore the association between palate shape and outcomes of pharyngeal OSA surgery. A strength of this study is its generalizability, with involvement of 13 centers and multiple types of pharyngeal surgeries.
Rates of exclusion varied by institution. Differences between included and excluded groups are likely related to differences in videos between institutions, such that some sites were more likely to have videos excluded. The rates of exclusion and associated differences in populations between study sites may have created a potential selection bias.
GenuAP narrowing was associated with a decrease in the odds of surgical response after isolated soft palate surgery. This suggests that current soft palate surgery techniques (the procedures performed in this study including a combination of soft tissue resection and repositioning) may have a limited ability to open the airway in an anteroposterior direction at the genu. Interestingly, this study also showed a trend toward an association between GenuLW narrowing and a greater odds of response to isolated soft palate surgery; if this proves true in larger studies, this may indicate that these procedures open the transverse dimensions of the retropalatal airway more effectively. Another possible explanation is that narrowing at GenuAP may instead be related to pressure from the underlying oral tongue and tongue-based collapse instead of palatal collapse. Similarly, lateral collapse seen at the level of GenuLW, may indicate primary collapse of the velopharyngeal port that is more directly addressed with palate surgery. The interaction of tongue and palate (the junctional tongue) has been previously shown to be associated with surgical response.4,20 Prior work on this cohort showed complete tongue-based collapse was associated with lower odds of surgical response.5 Our findings with GenuAP and GenuLW persisted when controlling for complete tongue-based collapse.
Limitations
This study had important limitations. First, the DISE review was based on prerecorded videos. While this may be considered a strength (all reviews were blinded to the type of surgery and surgery outcomes), there are some limitations in the assessment by surgeons who did not perform the DISE. Palate shape is a 3-dimensional construct, such that the assessment may best be performed by the surgeon at the time of the endoscopy. Palate shape assessments are subjective and could vary based on depth of sedation and subject position. Second, it is possible that the selection of palate surgery technique was associated with palate shape findings. Additionally, as palate shape was not a feature being evaluated specifically at the time of DISE procedures, some videos failed to capture this anatomy adequately. Further, given limitations of DISE to evaluate exact bony landmarks such as the end of the HP, the end of HP was approximated by the posterior edge of the nasal septum as a readily available landmark. Though not often encountered, adenoid hypertrophy would also influence palate shape interpretations with narrower appearance at palate levels compared to patients without adenoid hypertrophy. The clinical implication of this difference is unclear.
Interrater reliability was minimal to fair (κ = 0.24–0.39), despite all reviewers having extensive DISE experience. Our interrater reliability is consistent with related literature.4,5,15,21 The interrater reliability speaks to the subjective nature of the scoring system. In its current form, palate shape assessment has limitations. We attempted to mitigate the subjective nature of the scoring with the use of 3 blinded reviewers and consensus scoring. As discussed below, the palate shape scoring system can be improved by incorporating objective measurements.
Our palate shape statistical analysis assumed independence of palate level scores. GenuLW is largely independent (for practical purposes) of the other levels, whereas the other palate level scores, progressing from the HP to GenuAP to velum, can only be equal or greater to the previous level score (for example, an HP score of 2 would be followed only by a GenuAP score of 2 or 3).
Future Directions
This study offers initial insight to potential associations of palate shape and surgical outcomes. For practitioners performing DISE, attention to palate patterns at the level of the genu may aid surgical decision making and counseling for patients considering sleep surgery, particularly isolated palate surgery. For example, for patients with significant narrowing at GenuAP, in conjunction with the overall clinical picture, the surgeon may opt to perform a multilevel procedure or favor hypoglossal nerve stimulation. Additionally, for patients undergoing palate surgery, palate shape findings seen on DISE, may influence trajectory of tissue repositioning. A prospective study designed to study palate shape would allow for a more-thorough examination, including the incorporation of objective endoscopic measurements at the different palate levels such as measuring anterior-posterior distance and cross-sectional area. Incorporation of physiological features such as critical closing or opening pressure may allow for more robust understanding of palate shape influences.
Our study included multiple types of pharyngeal surgeries, potentially limiting subgroup analyses. However, given the size of our sample, even if there is true statistical difference between some of the analyzed groups, these may not be clinically meaningful differences. Larger studies, with patients undergoing the same type of surgery, are needed to investigate potential associations more thoroughly, including between palate shape and DISE findings, such as complete concentric collapse related to the soft palate, oropharyngeal lateral wall collapse, and complete tongue-related obstruction.
Conclusions
In our exploratory analysis, GenuAP narrowing and GenuLW openness are associated with a decrease in the odds of surgical response after isolated soft palate surgery. Palate shape and other palate shape level scores were not clearly associated with surgical outcomes. Larger, prospective studies may determine more clearly the association between palate shape and pharyngeal surgery outcomes.
Funding source:
Research reported in this publication was supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under Award Number R01HL160993. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Competing interests:
B. Tucker Woodson: Inspire Medical Systems—research trial investigator; Nyxoah SA—research trial investigator, coordinating investigator; Medtronic—research trial investigator, consultant; CryOSA—advisory board member, stock options; Linguaflex—research trial investigator. David T. Kent: Laborie Medical Technologies Corp—research support; Inspire Medical Systems—research support; Invicta Medical Inc—consultant, research support; Nyxoah SA—scientific advisory board member, intellectual property interests, research support. Eric J. Kezirian: Inspire Medical Systems—research funding; CryOSA—consultant, medical advisory board; Nyxoah SA—consultant; huMannity Medtec—consultant, Berendo Scientific—consultant, intellectual property; Magnap—intellectual property. The remaining authors have no competing interests.
Footnotes
This manuscript was presented at the 2023 AAO-HNSF Annual Meeting & OTO Experience; September 30 to October 4, 2023; Nashville, Tennessee.
References
- 1.Friedman M, Ibrahim H, Bass L. Clinical staging for sleep-disordered breathing. Otolaryngol Head Neck Surg. 2002;127(1):13–21. doi: 10.1067/mhn.2002.126477 [DOI] [PubMed] [Google Scholar]
- 2.Kezirian EJ. Nonresponders to pharyngeal surgery for obstructive sleep apnea: insights from drug-induced sleep endoscopy. Laryngoscope. 2011;121(6):1320–1326. doi: 10.1002/lary.21749 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Thaler E, Schwab R, Maurer J, et al. Results of the ADHERE upper airway stimulation registry and predictors of therapy efficacy. Laryngoscope. 2020;130(5):1333–1338. doi: 10.1002/lary.28286 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Huyett P, Kent DT, D’Agostino MA, et al. Drug-induced sleep endoscopy and hypoglossal nerve stimulation outcomes: a multicenter cohort study. Laryngoscope. 2021;131(7):1676–1682. doi: 10.1002/lary.29396 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Green KK, Kent DT, D’Agostino MA, et al. Drug-induced sleep endoscopy and surgical outcomes: a multicenter cohort study. Laryngoscope. 2019;129(3):761–770. doi: 10.1002/lary.27655 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Friedman M, Ibrahim H, Joseph NJ. Staging of obstructive sleep apnea/hypopnea syndrome: a guide to appropriate treatment. Laryngoscope. 2004;114(3):454–459. doi: 10.1097/00005537-200403000-00013 [DOI] [PubMed] [Google Scholar]
- 7.Koutsourelakis I, Safiruddin F, Ravesloot M, Zakynthinos S, De Vries N. Surgery for obstructive sleep apnea: sleep endoscopy determinants of outcome. Laryngoscope. 2012;122(11):2587–2591. doi: 10.1002/lary.23462 [DOI] [PubMed] [Google Scholar]
- 8.Soares D, Sinawe H, Folbe AJ, et al. Lateral oropharyngeal wall and supraglottic airway collapse associated with failure in sleep apnea surgery. Laryngoscope. 2012;122(2):473–479. doi: 10.1002/lary.22474 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Hsu YS, Jacobowitz O. Does sleep endoscopy staging pattern correlate with outcome of advanced palatopharyngoplasty for moderate to severe obstructive sleep apnea? J Clin Sleep Med. 2017;13(10):1137–1144. doi: 10.5664/jcsm.6756 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Schwab RJ, Wang SH, Verbraecken J, et al. Anatomic predictors of response and mechanism of action of upper airway stimulation therapy in patients with obstructive sleep apnea. Sleep. 2018;41(4):1–12. doi: 10.1093/sleep/zsy021 [DOI] [PubMed] [Google Scholar]
- 11.Choi JH, Cho SH, Kim SN, Suh JD, Cho JH. Predicting outcomes after uvulopalatopharyngoplasty for adult obstructive sleep apnea. Otolaryngol Head Neck Surg. 2016;155(6):904–913. doi: 10.1177/0194599816661481 [DOI] [PubMed] [Google Scholar]
- 12.Park D, Kim JS, Heo SJ. Obstruction patterns during drug-induced sleep endoscopy vs natural sleep endoscopy in patients with obstructive sleep apnea. JAMA Otolaryngol Head Neck Surg. 2019;145(8):730–734. doi: 10.1001/jamaoto.2019.1437 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Woodson BT. A method to describe the pharyngeal airway. Laryngoscope. 2015;125:1233–1238. doi: 10.1002/lary.24972 [DOI] [PubMed] [Google Scholar]
- 14.Olszewska E, Woodson BT. Palatal anatomy for sleep apnea surgery. Laryngoscope Investig Otolaryngol. 2019;4(1):181–187. doi: 10.1002/lio2.238 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Kedarisetty S, Sharma A, Commesso EA, et al. Palate shape is associated with unilateral hypoglossal nerve stimulation outcomes. Laryngoscope. 2024;134:981–986. doi: 10.1002/lary.31018 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Brodsky L Modern assessment of tonsils and adenoids. Pediatr Clin North Am. 1989;36(6):1551–1569. doi: 10.1016/S0031-3955(16)36806-7 [DOI] [PubMed] [Google Scholar]
- 17.Sher AE, Schechtman KB, Piccirillo JF. The efficacy of surgical modifications of the upper airway in adults with obstructive sleep apnea syndrome. Sleep. 1996;19(2):156–177. doi: 10.1093/sleep/19.2.156 [DOI] [PubMed] [Google Scholar]
- 18.Kezirian EJ, Hohenhorst W, De Vries N. Drug-induced sleep endoscopy: the VOTE classification. Eur Arch Otrhinolaryngol. 2011;268(8):1233–1236. doi: 10.1007/s00405-011-1633-8 [DOI] [PubMed] [Google Scholar]
- 19.Hohenhorst W, Ravesloot MJL, Kezirian EJ, De Vries N. Drug-induced sleep endoscopy in adults with sleep-disordered breathing: technique and the VOTE Classification system. Oper Tech Otolaryngol Head Neck Surg. 2012;23(1):11–18. doi: 10.1016/j.otot.2011.06.001 [DOI] [Google Scholar]
- 20.Fleury Curado T, Pham L, Otvos T, et al. Changes in tongue morphology predict responses in pharyngeal patency to selective hypoglossal nerve stimulation. J Clin Sleep Med. 2023;19(5):947–955. doi: 10.5664/jcsm.10474 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Kezirian EJ, White DP, Malhotra A, Ma W, McCulloch CE, Goldberg AN. Interrater reliability of drug-induced sleep endoscopy. Arch Otolaryngol Head Neck Surg. 2010;136(4):393. doi: 10.1001/archoto.2010.26 [DOI] [PubMed] [Google Scholar]