Nonresponders to pharyngeal surgery for obstructive sleep apnea: insights from drug-induced sleep endoscopy

Abstract

Objectives/Hypothesis

To examine DISE findings in nonresponders to previous pharyngeal OSA surgery

Study Design

cross-sectional

Methods

Drug-induced sleep endoscopy (DISE) using propofol for unconscious sedation was performed in nonresponders to previous OSA surgery (including palate surgery with or without tonsillectomy and possible other procedures), defined by an apnea-hypopnea index >10 events/hour. Recorded findings included the presence and degree of obstruction in the palatal and hypopharyngeal regions, the contributions of specific structures (velum, oropharyngeal lateral walls, tongue, and/or epiglottis) to upper airway obstruction, and the degree of mouth opening.

Results

Thirty-three nonresponders underwent DISE examinations. Age was 46.2±11.8 years, and 9% (3/33) were female. On diagnostic sleep studies prior to DISE, the apnea-hypopnea index was 43.4±26.6 events/hour. During DISE, a majority of subjects demonstrated residual palatal obstruction, and almost all demonstrated hypopharyngeal obstruction. A diversity of individual structures contributed to upper airway obstruction, often in combination. Moderate to severe mouth opening occurred in one-third of subjects and was associated with narrowing of upper airway dimensions.

Conclusions

Residual upper airway obstruction in surgery nonresponders likely occurs due to multiple mechanisms, and DISE may enhance the understanding of them.

INTRODUCTION

Pharyngeal airway obstruction in obstructive sleep apnea (OSA) is often classified as occurring in the palatal (related to the palate, or velum) and/or so-called hypopharyngeal (actually corresponding to the hypopharynx and a portion of the oropharynx posterior to the tongue) regions. The most common OSA surgical treatment in the United States is isolated palate surgery, with or without tonsillectomy.¹ In cohort studies this procedure is associated with decreased mortality,^{2, 3} reduced cardiovascular deaths,⁴ and improvements in sleep-related quality of life.⁵ Palate surgery by itself, however, rarely eliminates OSA⁶ and achieves a meaningful reduction in OSA severity to no worse than mild disease (often termed a “response” or “successful” outcome) in as few as 5–38% of patients.^{7, 8} One proposed mechanism for imperfect outcomes after palate surgery is untreated airway obstruction occurring in the so-called hypopharyngeal region. In an attempt to improve surgical outcomes, hypopharyngeal procedures have been developed and are often performed in combination with palate surgery. Combinations of palate surgery with various hypopharyngeal procedures achieve this response in 35–62% of patients,⁹ suggesting improved but imperfect outcomes for a number of patients.

Although substantial effort has been devoted to identifying factors associated with outcomes after initial surgical treatment with palate and/or hypopharyngeal procedures,^{7, 9} few publications have considered patients who do not respond to initial pharyngeal surgery.^10–14 Examination of these nonresponders may determine the mechanisms for the lack of a response.

Drug-induced sleep endoscopy (DISE) is an unique OSA surgical evaluation technique that requires fiberoptic examination of the upper airway under conditions of spontaneous ventilation and pharmacologic unconscious sedation. In subjects without previous OSA surgery, DISE has been shown to be valid,^15–18 reliable,^{19, 20} and associated with palate surgery outcomes.^{21, 22}

The objectives of this study were to examine DISE findings in nonresponders to previous pharyngeal OSA surgery.

METHODS

This retrospective cohort study included consecutive subjects who were nonresponders to previous palate surgery (and tonsillectomy, whether performed concurrently or not) to treat OSA and who underwent DISE. Inclusion criteria included age > 18 years, apnea-hypopnea index > 10 events/hour on postoperative full-night diagnostic polysomnogram (performed according to standard clinical practice) following previous palate surgery and prior to enrollment, and inability to tolerate positive airway pressure therapy. Exclusion criteria included pregnant women and allergy to propofol or to components of propofol such as egg lecithin or soybean oil. Full-night diagnostic polysomnogram was repeated if not performed within one year of enrollment or if there was a change in body weight of more than 5 kg or a significant change in signs or symptoms of OSA. This study was approved by the University of California, San Francisco institutional review board, and all subjects provided written informed consent.

All subjects underwent DISE in the operating room. Subjects received an intravenous anticholinergic (glycopyrrolate 0.2–0.4 mg) to reduce secretions, while a topical vasoconstrictor/anesthetic combination (oxymetazoline/lidocaine) was applied to one nasal cavity. Subjects were positioned on the operating table in the supine position, with standard anesthetic monitoring, including pulse oximetry and three-lead electrocardiography. A continuous intravenous infusion of propofol was used as the sole agent to achieve unconscious sedation, defined as the absence of a response to verbal stimulation in a normal voice, similar to a Modified Ramsay Score of 5 or Observer’s Assessment of Alertness/Sedation Score of 2–3.^23–25 DISE was performed at the transition to unconsciousness (defined by a lack of response to loud verbal stimulation), based on previous research. Repeated assessment of the depth of sedation minimized the risk of oversedation. A subgroup had placement of a Bispectral index score (BIS) sensor (XP Quattro Sensor, Aspect Medical Systems, Inc., Norwood, MA) across their forehead to monitor the depth of sedation with the BIS VISTA system from frontal electroencephalogram activity. Awake flexible laryngoscopy was performed through the anesthetized nasal cavity to confirm adequate topical anesthesia at the beginning of the procedure, so that placement of the laryngoscope did not produce painful stimulation and arousal during propofol sedation. The initial infusion rate of propofol was 50–75 mcg/kg-min, and the rate was adjusted to meet this target level of sedation slowly, in order to avoid oversedation. The author performed all DISE exams and documented findings at the time of the exam, prior to any additional treatment.

DISE findings were summarized with three analyses. Analysis I was a global dichotomous (yes/no) assessment of obstruction at each of two levels: the palate/velum and the hypopharynx. Analysis II reflected the degree of palatal and hypopharyngeal obstruction. This was graded separately for each region subjectively and categorized in an ordinal fashion as <50 %, 50–75%, and > 75% obstruction; these were not quantitative but reflected a qualitative assessment of no/mild, moderate, and severe obstruction, respectively. Analysis III evaluated specific structures with a determination of which structure at the levels of the palate and hypopharynx was the primary factor in airway obstruction, if present, and a dichotomous evaluation of whether each of the individual structures contributed to airway obstruction. Structures were grouped as those at the level of the palate (the velum/soft palate, tonsils when present) and the hypopharynx (oropharyngeal lateral walls, tongue, and epiglottis).

During DISE, additional evaluation of mouth opening (for oral breathing) was recorded, classifying the degree of mouth opening as none (<1 cm distance between the maxillary and mandibular incisors), mild (1–2 cm), moderate (>2 – 3 cm), or severe (>3 cm). In cases of moderate or severe mouth opening, manual closure of the mouth to enable direct contact of the maxillary and mandibular incisors without mandibular advancement or neck extension was performed; any changes in the pattern of pharyngeal obstruction were recorded.

Over half (53%, 17/33) of the subjects underwent pre-DISE polysomnograms at the author’s institution. The recording montage at the author’s institution consists of C3/A2 and C4/A1 electroencephalograms, bilateral electrooculograms, a bipolar submental electromyogram, thoracic and abdominal respiratory inductance plethysmography, airflow (using nasal-oral thermocouple and nasal pressure cannula), finger pulse oximetry, electrocardiogram, body position (mercury switch sensor), and bilateral leg movements (piezoelectric sensors). Sleep stages and arousals were scored using standard criteria.²⁶ Apneas were defined as a complete or almost complete cessation of airflow (by thermocouple), and hypopneas were identified as a >30% reduction in airflow associated with a >4% oxygen desaturation, according to Medicare criteria.²⁷ Apneas associated with no evidence of effort on both thoracic and abdominal channels were considered to be central and otherwise as obstructive. The following polysomnogram results were recorded: apnea-hypopnea index (AHI, apneas plus hypopneas per hour of sleep), lowest oxygen saturation, the percentage of sleep time with oxygen saturation below 90%, and the percentages of sleep time spent in stages N3 and rapid eye movement sleep. Of the subjects not undergoing polysomnograms at the author’s institution, the majority (36% of total, 12/33) underwent studies at centers using recent consensus-based recommendations from the American Academy of Sleep Medicine, with a similar definition of apneas and a hypopnea defined by a decrease in airflow of at least 50% for at least 10 seconds with an associated oxygen desaturation of at least 3% or a sleep arousal.²⁶

Descriptive statistics were calculated for baseline subject characteristics, including body mass index (defined as weight in kg divided by the square of height in m). T-tests were used to compare age and body mass index among those with and without previous hypopharyngeal procedures and for those with previous genioglossus advancement vs. tongue radiofrequency.

DISE findings are reported for the entire cohort and for subgroups defined by the nature of previous surgical treatment. Fisher’s exact tests were used to compare DISE findings among those with and without previous hypopharyngeal surgery and for differences between previous genioglossus advancement and tongue radiofrequency. T-tests were used to examine association between DISE findings and both age and body mass index.

Pre-DISE polysomnogram results and DISE findings are presented for the entire cohort and for subgroups based on differences in the primary surgical treatment. . DISE findings were examined for an association with type of previous surgical treatment using chi-squared tests. All continuous measures are reported with mean ± standard deviation. P-values < 0.05 were considered statistically significant. Statistical analyses were conducted using Stata Version 10.0 (StataCorp LP, College Station, TX).

RESULTS

Thirty-three subjects underwent DISE examinations between 2004 and 2010. Age was 46.2±11.8 (range 25 – 69) years, and 9% (3/33) were female. Most (79%, 26/33) were non-Hispanic Caucasian, per subject report. Body mass index was 30.5±3.9 kg/m². On diagnostic sleep studies prior to DISE, the apnea-hypopnea index was 43.4±26.6 (range 11–120), with the following distribution across commonly-used apnea-hypopnea index cutpoints: 6% (2/33) with apnea-hypopnea index 10 – <15 events/hour, 30% (10/33) with 15 – 30 events/hour, and 64% (21/33) with >30 events/hour. The sleep study lowest oxygen saturation was 83.4±9.2%, and the subjects with oxygen desaturation below 90% during sleep (n=23) spent 19.3±21.8% with oxygen saturation below 90%. The percentages of sleep time spent in N3 and rapid eye movement sleep were 14.2±13.8 and 7.7±8.7%, respectively, during sleep study.

A complete DISE examination was performed in all cases, and all patients demonstrated upper airway obstruction during DISE. Propofol infusion rate required to achieve sedation was 125±24 (range 75–175) mcg/kg-min. Total propofol dose was variable, as the length of time for DISE evaluation varied widely. A subset (n=19) of patients underwent clinical evaluation of the depth of sedation using the Modified Ramsay Score (mean 4.9±0.8, range 4–6) and the Observer’s Assessment of Alertness/Sedation Score (mean 2.8±0.8, range 1–3); a smaller subset (n=9) underwent evaluation with the BIS monitoring system, with a mean score during unconscious sedation of 64±2.6 (range 59–69).

DISE findings are presented for the entire cohort and according to previous surgical treatment in Table I. Previous palate surgery was primarily uvulopalatopharyngoplasty (82%, 27/33), and other palate procedures included the uvulopalatal flap (n=4) and palatal implants (n=2). There were no differences in age or body mass index between subjects who had and had not undergone previous hypopharyngeal surgery or between those who had undergone genioglossus advancement vs. tongue radiofrequency (data not shown); because of sample size, it was not possible to evaluate for differences in gender or race/ethnicity.

Table I.

DISE findings according to previous surgical treatment.

	All previous palate surgery (n=33)	Palate surgery alone (n=17)	Palate surgery and genioglossus advancement** (n=7)	Palate surgery and tongue radiofrequency** (n=9)	Palate surgery and hyoid suspension*** (n=6)	P values for chi- squared tests of palate surgery alone vs. tongue region surgery and for genioglossus advancement vs. tongue radiofrequency
Analysis I. Global assessment of level(s) of obstruction
Palate	17 (53%)	7 (44%)	4 (57%)	6 (67%)	6 (100%)	0.36, 0.37
Hypopharynx	32 (97%)	17 (100%)	7 (100%)	8 (89%)	6 (100%)	0.50, 1.00
Analysis II. Degree of obstruction
Palate
<50%	16 (48%)	10 (59%)	3 (43%)	3 (33%)	0 (0%)	0.63, 0.63
50–75%	2 (6%)	1 (6%)	0 (0%)	1 (11%)	1 (17%)
>75%	15 (45%)	6 (35%)	4 (57%)	5 (56%)	5 (83%)
Hypopharynx
<50%	1 (3%)	0 (0%)	0 (0%)	1 (11%)	0 (0%)	1.00, 0.55
50–75%	6 (18%)	4 (24%)	2 (29%)	0 (0%)	2 (33%)
>75%	26 (79%)	13 (76%)	5 (71%)	8 (78%)	4 (67%)
Analysis III. Specific structures.
Primary structure contributing to obstruction at level of the palate
None	16 (48%)	10 (59%)	3 (43%)	3 (33%)	0 (0%)	0.36, 0.37
Velum	17 (52%)	7 (41%)	4 (57%)	6 (67%)	6 (100%)
Tonsils	n/a	n/a	n/a	n/a	n/a
Primary structure contributing to obstruction at the level of the hypopharynx
None	1 (3%)	0 (0%)	0 (0%)	1 (11%)	0 (0%)	0.84, 0.60
Oropharyngeal lateral walls	10 (30%)	5 (29%)	3 (43%)	2 (22%)	4 (67%)
Tongue	21 (64%)	11 (65%)	4 (57%)	6 (67%)	2 (33%)
Epiglottis	1 (3%)	1 (6%)	0 (0%)	0 (0%)	0 (0%)
Specific structures contributing to obstruction*
Velum	17 (53%)	7 (44%)	4 (57%)	6 (67%)	6 (100%)	0.48, 1.00
Oropharyngeal lateral walls	17 (53%)	8 (50%)	4 (57%)	5 (56%)	6 (100%)	0.72, 1.00
Tongue	29 (88%)	15 (88%)	6 (86%)	8 (89%)	4 (67%)	1.00, 1.00
Epiglottis	8 (25%)	6 (38%)	1 (14%)	1 (11%)	0 (0%)	0.22, 1.00

Note: Table reports raw results, followed by column percents.

^*Each structure is considered separately, with percentages expressed as a fraction of total DISE evaluations. Percentages sum to greater than 100% because it was possible for a patient to have more than one structure contributing to airway obstruction.

^**Includes subjects that also had undergone hyoid suspension.

^***Includes subjects that also had undergone genioglossus advancement or tongue radiofrequency.

During DISE, a majority of subjects demonstrated global obstruction (Analysis I) in the palatal region, and almost all subjects demonstrated hypopharyngeal obstruction, largely severe (>75%) obstruction (Analyses I and II). There were no differences among subgroups defined by previous surgical treatment in findings from Analyses I and II. Subjects with residual palatal obstruction were older (50.8±2.4 years vs. 39.4±9.8 years, p = 0.004) and had a trend towards higher body mass index (31.5±4.2 vs. 29.4±3.3 kg/m², p = 0.07) than those without; because almost all subjects demonstrated hypopharyngeal obstruction, it was not possible to evaluate similar associations.

There was a notable diversity of specific structures that contributed to airway obstruction during DISE (Analysis III), whether considering the primary structure within each region or individual structures. These were not systematically associated with previous hypopharyngeal procedure overall or with specific procedures, although no subjects with previous hyoid suspension demonstrated an epiglottic contribution to airway obstruction. Tongue contribution of obstruction during DISE was common, both without previous hypopharyngeal surgery and with previous genioglossus advancement, tongue radiofrequency, or hyoid suspension. Table II presents the combinations of individual structures contributing to palatal and hypopharyngeal obstruction. There were multiple observed combinations of involved structures, with no systematic differences based on previous surgical treatment (including type of palate procedure, data not shown due to small sample sizes) aside from those noted above.

Table II.

Combinations of individual structures contributing to airway obstruction at the levels of the palate and hypopharynx, according to previous surgical treatment.

	All previous palate surgery (n=33)	Palate surgery alone (n=17)	Palate surgery and genioglossus advancement† (n=7)	Palate surgery and tongue radiofrequency† (n=9)	Palate surgery and hyoid suspension†† (n=6)
Palatal Obstruction
None	16 (48%)	10 (59%)	3 (43%)	3 (%)	0 (0%)
Velum alone	1 (3%)	0 (0%)	0 (0%)	1 (11%)	0 (0%)
Velum + Hypopharyngeal Obstruction	16 (48%)	7 (41%)	4 (57%)	6 (%)	6 (100%)
Hypopharyngeal Obstruction
None	1 (3%)	0 (0%)	0 (0%)	1 (11%)	0 (0%)
Oropharyngeal lateral walls	4 (12%)	2 (12%)	1 (14%)	1 (11%)	2 (33%)
Tongue	11 (33%)	5 (29%)	2 (29%)	4 (33%)	0 (0%)
Epiglottis	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Oropharyngeal lateral walls + Tongue	9 (27%)	4 (24%)	3 (43%)	3 (33%)	4 (67%)
Oropharyngeal lateral walls + Epiglottis	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Tongue + Epiglottis	4 (12%)	4 (24%)	0 (0%)	0 (0%)	0 (0%)
Oropharyngeal lateral walls + Tongue + Epiglottis	4 (12%)	2 (12%)	1 (14%)	1 (11%)	0 (0%)

Note: Table reports raw results, followed by column percents.

Eleven subjects demonstrated moderate to severe mouth opening during DISE. Among these subjects, the structures contributing to airway obstruction at baseline were the velum (82%, 9/11), oropharyngeal lateral walls (73%, 8/11), tongue (73%; 8/11, including 2 of 3 subjects without a coexisting oropharyngeal lateral wall contribution), and epiglottis (18%; 2/11, both in cases with additional oropharyngeal lateral wall and tongue contributions). In all of these 11 subjects, manual closure of the mouth markedly improved airway dimensions due to a favorable displacement (improving airway dimensions) of the oropharyngeal lateral walls and tongue (laterally and anteriorly, respectively), although the airway obstruction did not resolve in 6/11 (55%). There was no similar change in epiglottic position with manual mouth closure for subjects with or without an epiglottic contribution. Four subjects with notable tongue displacement also had marked improvement in an obstructed palatal airway with mouth closure.

DISCUSSION

This is the largest study examining individuals who have not responded to OSA surgical treatment. The findings suggest that multiple mechanisms may contribute to nonresponse, including contributions from various upper airway structures (velum, oropharyngeal lateral walls, tongue, and epiglottis) and mouth opening that also narrows the pharyngeal lumen.

Similar to those without previous OSA surgery,^{19, 20, 28, 29} a diversity of factors may contribute to residual upper airway obstruction in surgical nonresponders. Differences between nonresponders and those without previous surgery (from previous studies) were that in the former, a higher proportion demonstrated contributions from the oropharyngeal lateral walls, tongue, and epiglottis and that a lower proportion had a contribution from the velum. As the subjects in the current study all had previous tonsillectomy and palate surgery, at a minimum, it is not surprising that residual contributions from the velum and tonsils would be less common.

The diversity of patterns of obstruction during DISE seen in this study is reassuring, as it may reflect the multifactorial nature of upper airway obstruction in OSA. The accuracy of propofol unconscious sedation (as in DISE) as a representation of natural sleep depends on multiple factors, including upper airway muscle tone, neuromuscular reflexes, and lung volumes--as well as outcomes of surgical treatment directed by DISE findings. During propofol unconscious sedation, normals have demonstrated decreases in genioglossus tone to 10% of maximum awake activity,^{30, 31} which is one-half to one-third of the level in normals at sleep onset³² but greater than during REM sleep in normals and subjects with OSA.³³ While unconscious sedation under propofol may not a perfect simulation of natural sleep, with identical effects on upper airway collapsibility, pharyngeal dilator muscle activity appears to lie somewhere between NREM and REM sleep. There have been no comparisons of lung volumes or upper airway neuromuscular reflex activity between propofol sedation and natural sleep.

The five previous studies examining site of obstruction in nonresponders to pharyngeal OSA surgery include one incorporating a combination of drug-induced sleep endoscopy and pharyngeal manometry,¹¹ two utilizing lateral cephalometry,^{10, 12} and two with airway pressure monitoring.^{13, 14} Woodson and Wooten examined 11 nonresponders and attempted to determine the primary region of collapse (palatal vs. hypopharyngeal) with two evaluation techniques.¹¹ Manometry demonstrated palatal and hypopharyngeal obstruction, respectively, in 8 (73%) and 3 (27%) of all subjects. The proportion demonstrating primarily palatal obstruction (vs. hypopharyngeal obstruction) included 5/6 with previous uvulopalatopharyngoplasty alone, 0/2 with transpalatal advancement pharyngoplasty alone,³⁴ and 3/3 with previous uvulopalatopharyngoplasty and hypopharyngeal surgery. Three patients with primarily palate obstruction on manometry also demonstrated hypopharyngeal obstruction during DISE, suggesting residual hypopharyngeal obstruction in 6/11 (55%). Riley et al. found a decreased posterior airway space and increased mandibular plane to hyoid distance on lateral cephalometry in a sample of 9 nonresponders to soft palate surgery,¹⁰ while Yao et al. compared preoperative and postoperative lateral cephalograms after combined palatal and hypopharyngeal surgery and showed that, in contrast to responders, nonresponders did not increase their posterior airway space, posterior uvular space (distance from the uvula to the posterior pharygneal wall), and mandibular plane to hyoid distance.¹² With airway pressure monitoring, Metes et al. examined 8 nonresponders to uvulopalatopharyngoplasty and showed residual palatal (vs. hypopharyngeal) obstruction in 6/8 (75%), with the primary region of obstruction unchanged, based on a similar preoperative evaluation.¹³ In a study of 22 nonresponders after uvulopalatopharyngoplasty with airway pressure monitoring, Farmer and Giudici found a primary site of obstruction in the palatal region for 4/15 (27%) and in the hypopharyngeal region in 11/15 (73%).

Residual palatal obstruction, even after soft palate surgery (uvulopalatopharyngoplasty. uvulopalatal flap, or palatal implants) was common, consistent with previous studies. The inability of these soft palate procedures to resolve palatal obstruction has led to the development of a number of palate-directed procedures, including transpalatal advancement pharyngoplasty, lateral pharyngoplasty,³⁵ expansion sphincter pharyngoplasty,³⁶ relocation pharyngoplasty,³⁷ and Z-palatoplasty.³⁸ Additional research can examine whether the latter group of procedures may be more effective in treating palatal obstruction and, possibly, a contribution from the oropharyngeal lateral walls.

Hypopharyngeal obstruction during DISE was nearly universal in this study, whether due to a contribution of the oropharyngeal lateral walls, tongue, or epiglottis. Hypopharyngeal obstruction was present even in subjects with previous hypopharyngeal procedures, suggesting that these procedures may not provide sufficient improvement in airway dimensions and/or that they treat structures that are not playing a specific role in hypopharyngeal obstruction for the individual subject. For example, genioglossus advancement and tongue radiofrequency are both directed primarily at the tongue and may not address the oropharyngeal lateral walls.

No subjects with previous hyoid suspension demonstrated an epiglottic contribution to upper airway obstruction during DISE. The epiglottis has multiple soft tissue attachments, but one of interest for OSA surgery may be the relationship to the hyoid bone via the hyoepiglottic ligament. Hyoid suspension may treat the epiglottis more specifically than other hypopharyngeal procedures due to anterior displacement and stabilization of the hyoid bone and, indirectly, the hyoepiglottic ligament. Additional studies may be able to determine this more systematically and consider which hyoid suspension technique, whether to the thyroid cartilage³⁹ or inferior border of the mandible,⁴⁰ is preferred. Other procedures such as supraglottoplasty or partial epiglottidectomy^{41, 42} may also be particularly beneficial in these cases.

One-third of all subjects in this study demonstrated substantial mouth opening during unconscious sedation with propofol, with apparent adverse effects on pharyngeal dimensions related to the velum, oropharyngeal lateral walls, and tongue base. This finding agrees with previous research showing that greater degrees of mouth opening during sleep are associated with poorer outcomes after primary palate surgery.⁴³ It is unclear whether mouth opening during sleep occurs due to nasal or pharyngeal obstruction or whether it can reverse with adequate treatment of obstruction (at least apparent adequate treatment without mouth opening). The implications for surgical treatment as well as non-invasive treatments such as chin straps are not entirely clear, and future investigations, ideally with larger sample sizes, is needed.

The importance of characterizing the patterns of obstruction, whether in primary or secondary treatment, lies in the association between treatment selection and outcomes. While previous studies have shown associations between DISE findings and outcomes for uvulopalatopharyngoplasty and mandibular repositioning appliances, future research—ideally with large, prospective cohorts—will enable an examination of hypopharyngeal surgery outcomes and the potential association between DISE findings related to hypopharyngeal obstruction (especially in the specific structures that may contribute to obstruction) and individual hypopharyngeal procedures.

In addition to those outlined above, this study has other limitations. Although it is the largest study of its kind, larger and more-detailed investigations will be useful. The subgroups were differentiated according to previous surgical treatment, but there are other differences including age, gender, race/ethnicity among subjects that cannot be examined in detail with a study of this size. Primary surgery was performed by multiple surgeons in multiple institutions, and therefore the procedures (i.e., palate surgery) are not standardized. As the subjects did not undergo DISE prior to primary surgery, there is no comparison of preoperative and postoperative findings to evaluate changes.

CONCLUSION

Residual upper airway obstruction in OSA surgery nonresponders likely occurs due to multiple mechanisms, and evaluation techniques such as DISE may enhance the understanding of these mechanisms. Oropharyngeal lateral wall prolapse was associated with poorer outcomes after secondary surgery and may explain a portion of nonresponders to both primary and secondary OSA surgery.