Customize your JAMA Network experience by selecting one or more topics from the list below.
Yalamanchili R, Mack WJ, Kezirian EJ. Drug-Induced Sleep Endoscopy Findings in Supine vs Nonsupine Body Positions in Positional and Nonpositional Obstructive Sleep Apnea. JAMA Otolaryngol Head Neck Surg. 2019;145(2):159–165. doi:10.1001/jamaoto.2018.3692
What anatomic mechanism underlying positional vs nonpositional obstructive sleep apnea (OSA) is suggested by drug-induced sleep endoscopy (DISE)?
In this cross-sectional study of 65 adults with OSA, anteroposterior airway obstruction was greater in the supine body position, whereas transverse airway obstruction was greater in the lateral body position in the overall study population and the positional and nonpositional obstructive sleep apnea subgroups. The exception was a lower degree of obstruction related to the oropharyngeal lateral walls in the lateral body position in the nonpositional OSA subgroup only.
Treatments that resolve anteroposterior velum- and tongue-related obstruction may have better outcomes in positional OSA.
The anatomic mechanisms underlying positional vs nonpositional obstructive sleep apnea (OSA) are poorly understood and may inform treatment decisions.
To examine drug-induced sleep endoscopy (DISE) findings in the supine vs nonsupine body positions in positional and nonpositional obstructive sleep apnea.
Design, Setting, and Participants
A cross-sectional study of 65 consecutive eligible adults with OSA undergoing DISE without marked tonsillar hypertrophy, including 39 with positional OSA (POSA) and 26 with nonpositional OSA (N-POSA) was conducted in a sleep surgery practice at a tertiary academic medical center.
Drug-induced sleep endoscopy performed in the supine vs nonsupine body position.
Main Outcomes and Measures
Drug-induced sleep endoscopy findings were scored separately for the supine and lateral body positions using the VOTE classification (velum, oroparyngeal lateral walls, tongue, epiglotis) and with identification of a single primary structure contributing to airway obstruction. Velum-related obstruction was separated into anteroposterior and lateral components.
The 65 study participants had a mean (SD) age of 52.4 (11.7) years, and 55 (84.6) were men. Mean (SD) body mass index (BMI, calculated as weight in kilograms divided by height in meters squared) was 27.2 (3.1), with only 9 (14%) of 65 participants having a BMI greater than 30. The supine body position was associated with greater odds of anteroposterior velum- (odds ratio [OR], 7.28; 95% CI, 3.53-15.01), tongue- (OR, 29.4; 95% CI, 12.1-71.5), and epiglottis-related (OR, 11.0; 95% CI, 1.3-92.7) obstruction in the entire cohort, with similar findings in the POSA and N-POSA subgroups. The supine body position was associated with a lower odds of oropharyngeal lateral wall–related (OR, 0.22; 95% CI, 0.07-0.70) obstruction in the N-POSA subgroup, whereas there was no increase in the overall sample or the POSA subgroup. The oropharyngeal lateral walls were a common primary structure causing obstruction, especially in the lateral body position.
Conclusions and Relevance
In a study population of primarily nonobese adults, DISE findings differed based on body position, generally corresponding to gravitational factors. Treatments that address velum- and tongue-related obstruction successfully may be more effective in the POSA population.
Obstructive sleep apnea (OSA) is a disorder of symptomatic, repeated upper airway obstruction. Positional obstructive sleep apnea (POSA) is commonly defined as an apnea-hypopnea index (AHI) that is at least twice as great in the supine position compared with nonsupine positions. Positional obstructive sleep apnea is common, with studies suggesting that 56% of individuals with OSA have POSA, with an additional 30% having a greater supine vs nonsupine AHI without meeting the threshold of being at least twice as great.1-3
The presence of positional changes has implications for treatment selection and outcomes. First, POSA may be treated with positional therapy, either alone or in combination therapy. Mandibular repositioning appliances4 and at least 1 surgical procedure (relocation pharyngoplasty)5 have been shown to be more effective in patients with POSA compared with nonpositional OSA (N-POSA). Weight loss has been shown to be associated with greater improvement in nonsupine AHI than supine AHI.6
Determining the anatomical mechanisms underlying POSA may explain some of the impact on treatment selection and outcomes, particularly for anatomical treatments such as surgery and oral appliances. Drug-induced sleep endoscopy (DISE) is a surgical evaluation technique that examines pharyngeal anatomy under conditions of unconscious sedation. Drug-induced sleep endoscopy characterizes structures that appear to contribute to upper airway obstruction and is generally performed in the supine body position. Previous research regarding DISE findings and body position has had conflicting findings, with 1 study showing changes in DISE only in POSA7 and another showing changes associated with body position but no evaluation of POSA vs N-POSA.8 The objective of this study was to examine DISE findings in the supine and lateral body position in individuals with POSA and N-POSA. Our hypotheses were that the supine body position was associated with greater degrees of anteroposterior airway narrowing and that the lateral body position was associated with greater degrees of transverse airway narrowing in both POSA and N-POSA study participants.
This was a cross-sectional study of consecutive OSA patients undergoing DISE at the Keck Medical Center of USC. Eligibility criteria included age 18 years or older, apnea-hypopnea index of 5 or more events per hour on a diagnostic home sleep apnea test or in-laboratory polysomnogram, with at least 30 minutes in supine and nonsupine body positions; no previous upper airway surgery other than nasal surgery or tonsillectomy; and absence of markedly enlarged tonsils (3+ or 4+ using the Brodsky classification).9 All study participants classified their own race and ethnicity according to options defined by the investigator’s institution; race and ethnicity are reported because of previous population-based studies suggesting that race and ethnicity are associated with OSA prevalence.
Drug-induced sleep endoscopy was performed using intravenous propofol sedation in a previously described technique.10 The study participant was placed in the lateral body position before initiation of the intravenous propofol infusion. The infusion was adjusted to maintain a level of sedation consistent with the transition from consciousness to unconsciousness, defined as the lack of response to verbal stimulation; bispectral index (BIS) monitoring was used as an aide to monitoring the depth of sedation, with a target score of generally 60 to 70 and, more importantly, consistent throughout the recorded DISE examination. At the transition to unconsciousness, the pattern of obstruction was determined. The study participant was then rolled to the supine body position, and the DISE examination continued (with a pause if there was arousal associated with movement), with a second determination of the pattern of obstruction in the supine position. All study participants were examined during adequate sedation for at least 2 minutes each in the nonsupine and then supine body positions, and the timing of the transition was documented in written form but not noted or otherwise incorporated in the recorded video. The study participant was examined first in the lateral body position because it is easier to move a study participant to the supine position without arousal than the reverse order. All of the DISE examinations were recorded. For this study, the video recordings were rereviewed separately for the lateral and supine body position in a random, blinded fashion by the senior author (E.J.K.).
The pattern of obstruction was scored using the VOTE classification (velum, oropharyngeal lateral walls, tongue, epiglottis),11 with an additional, subjective determination of the primary structure contributing to airway obstruction. The VOTE classification has been described previously. It requires a determination of the degree of obstruction (0: none; 1: partial; or 2: complete) related to the velum (palate), oropharyngeal lateral walls, tongue, and epiglottis. Velum-related obstruction requires scoring whether the configuration of obstruction was anteroposterior, lateral, or concentric (combination of anteroposterior and lateral). For this study, the anteroposterior and lateral components of velum-related obstruction were considered separately, with a concentric configuration considered as both anteroposterior and lateral. For example, partial anteroposterior velum-related obstruction would be considered as partial anteroposterior velum-related obstruction (velum AP) and no lateral velum-related obstruction (velum lateral). Complete concentric velum-related obstruction would be considered as complete velum AP and complete velum lateral. Because of the small sample size, the configuration of epiglottis-related obstruction was not considered separately in this study.
Statistical analyses were performed using Stata statistical software (version 10, StataCorp). Paired t tests (within-group, comparing body position) and 2 sample t tests (between group) were used to examine continuous variables; Shapiro-Wilk tests confirmed normality for AHI, BMI, and age (all P > .05). Fisher exact test was used to explore distribution of sex between OSA type subgroups. The degree of obstruction (0, 1, 2) was used to characterize each VOTE classification finding. The 65 participants each contributed 10 obstruction measurements (5 VOTE findings times 2 body positions), yielding 650 obstruction measurements. Because participants contributed repeated measurements (5 VOTE findings and 2 positions), the analysis used mixed-effects ordinal logistic regression, specifying a random effect for participant; variables in this analysis were VOTE finding (degree of obstruction), body position (supine vs lateral), and OSA type (POSA vs N-POSA). The initial model estimated and tested main effects for body position, VOTE finding, and OSA type. Subsequent models added and tested interaction terms to test: (1) if the effects of supine vs lateral body position on degree of obstruction differed over VOTE findings; (2) if the effects of supine vs lateral body position on VOTE findings differed by OSA type; and (3) if the effect of OSA type on VOTE findings differed over VOTE finding. Effects of each independent variable on obstruction measures are summarized as odds ratios (ORs) with 95% confidence intervals. Odds ratios greater than 1 indicate the variable was associated with a greater degree of obstruction than the referent level, and ORs less than 1 indicate the variable was associated with a lesser degree of obstruction than the referent level. Fisher exact tests were used to compare the primary structure contributing to airway obstruction, both for an association with body position and OSA type. P values <.05 and ORs that do not include 1.00 were considered statistically significant.
This study was approved by the institutional review board at the University of Southern California and was registered at ClinicalTrials.gov as NCT00695214. All study participants provided written informed consent.
There were 65 consecutive OSA study participants, with most (39) having POSA vs N-POSA (26). The POSA group included 32 men and 7 women with a mean (SD) age of 51.7 (11.4) years, and the N-POSA group inclded 23 men and 3 women with a mean (SD) age of 53.4 (12.2) years. There were no adverse events. There were no differences in key demographic, physical examination, or sleep study findings, other than a greater proportion (difference, 20.5%; 95% CI, 7.8%-33.2%) of Hispanic study participants in the POSA subgroup and a greater nonsupine AHI (difference, 17.3 events/h; 95% CI, 10.9-23.7) in the N-POSA subgroup (Table 1). Nine (14%) of the 65 study participants had body mass index greater than 30.
Overall VOTE classification scores are presented in Table 2, including the separation of velum-related obstruction into the anteroposterior and lateral components. For the 39 participants in the POSA subgroup, 36 (92%) and 21 (54%) had obstruction related to more than 1 structure in the supine and nonsupine positions, respectively. For the 26 participants in the N-POSA subgroup, 25 (96%) and 18 (69%) had obstruction related to more than 1 structure in the supine and non-supine positions, respectively.
Table 3 presents the main effects model results examining the association between body position, VOTE findings, and OSA type with degree of obstruction for each VOTE finding. Across VOTE findings and OSA type (POSA vs N-POSA), evaluations in the supine compared with lateral position were broadly associated with a greater degree of obstruction (OR, 2.61; 95% CI, 1.85-3.67) for VOTE findings. Compared with velum AP, all other VOTE findings were associated with a lesser degree of obstruction (for example, the odds of oropharyngeal lateral wall-related obstruction of a similar degree as for velum AP was 0.43, 95% CI, 0.26-0.69). Degree of obstruction did not significantly differ in POSA compared with N-POSA across body position and VOTE findings.
Table 4 presents the body position-VOTE finding interaction model results for the entire group and OSA type subgroups. Overall and within the POSA and N-POSA subgroups, the added interaction term of body position-by-VOTE finding was statistically significant, indicating that the effect of body position on degree of obstruction differed over VOTE finding. In the overall group (and controlling for OSA type), evaluations in the supine compared with lateral position were associated with greater degrees of obstruction in the velum AP, tongue, and epiglottis VOTE findings; whereas estimates indicated that the supine position may be associated with lesser degrees of obstruction than the lateral position in the velum lateral and oropharyngeal lateral walls findings, these differences were not statistically significant. The POSA and N-POSA study participants continued not to differ significantly on the degree of obstruction in this interaction model. Findings were similar with the POSA and N-POSA subgroups, with the exception of a lesser degree of obstruction in the supine vs lateral body position for velum lateral and oropharyngeal lateral wall in the N-POSA subgroup but not the POSA subgroup. In the N-POSA subgroup, the supine body position was associated with an OR of 0.22 (95% CI, 0.07-0.70) of greater oropharyngeal lateral wall-related obstruction, compared with the lateral body position. Because all 26 participants with N-POSA had a DISE of 0 in the epiglottis in the lateral position, and 3 of 26 had a DISE of 1 (remainder 0) in the supine position, the resulting OR was infinite. The 3-way interaction testing of the variable effects of supine vs lateral body position by VOTE finding differed in POSA vs N-POSA study participants, but was not statistically significant (P = .58).
Results for the primary structure contributing to airway obstruction are shown in Table 5. There were changes in the primary structure in the supine vs lateral body position in the entire group as well as the POSA and N-POSA subgroups independently (all P < .01). These changes were consistent with the results for the VOTE findings. For the POSA subgroup, most of the study participants with velum- or tongue-related obstruction in the supine position had a change in the primary structure in the lateral position, whether related to the velum (often concentric collapse), oropharyngeal lateral walls, or no obstruction (typically in participants with no OSA in the nonsupine body position on sleep study). No changes were seen in the primary structure for study participants with oropharyngeal lateral wall-related obstruction in the supine position. For the N-POSA subgroup, there were changes in primary structure for more than half of the participants. These changes were seen in those with primary tongue-related obstruction in the supine position, where many had primary oropharyngeal lateral wall-related obstruction or no obstruction in the lateral body position.
In this study, DISE findings associated with changes in body position were broadly consistent with the law of gravity, consistent with our hypotheses. Intravenous propofol sedation has been shown to result in decreases in genioglossus tone (and presumably in other muscles),12 so anteroposterior (anteroposterior velum-, tongue-, and epiglottis-related) airway narrowing would be expected in the supine body position and transverse (lateral velum- and oropharyngeal lateral wall-related) airway narrowing would be expected in the lateral body position. The 1 exception was that the changes with body position for lateral velum- and oropharyngeal lateral wall-related obstruction were not statistically significant overall and in the POSA subgroup.
These findings are consistent with (but extend beyond) previous DISE studies7,8 but differ from 1 small study of natural sleep endoscopy during administration of positive airway pressure.13 The present study is unique in evaluating the severity of obstruction (rather than presence/absence in 1 DISE study8), showing that the severity of structure-related obstruction may differ according to body position, even if the structure contributes to obstruction in both the supine and lateral body position. The present study also examines POSA and N-POSA separately (instead of distinguishing only those who still have OSA in the nonsupine position, as in 1 study8). Because it is common to have POSA with persistent OSA in the nonsupine, position, we believe our approach is more useful clinically (for example, when considering the combination of surgery with positional therapy). Finally, the present study also separates the anteroposterior and lateral components of velum-related obstruction, finding that these components of velum-related obstruction may also change with body position and in different ways.
Changes associated with body position were broadly similar for the POSA and N-POSA subgroups, consistent with 1 previous study8 but different from a smaller study.7 The only statistically significant difference was greater degrees of oropharyngeal lateral wall-related obstruction in the lateral vs supine body position for the N-POSA subgroup, although the finding related to lateral velum-related obstruction approached statistical significance. The lack of change in AHI in N-POSA may be related to a replacement of airway obstruction related to the velum (anteroposterior) and tongue in the supine position with airway obstruction related to the oropharyngeal lateral walls in the nonsupine position. This would be supported by this study’s findings of greater degrees of obstruction related to the lateral walls that results in more airway obstruction (and OSA) in the lateral position in the N-POSA subgroup. In contrast, the change in AHI in POSA suggests that there may be no such replacement, resulting in lesser degrees of airway obstruction in the nonsupine position.
In both subgroups, the oropharyngeal lateral walls were the primary structure contributing to airway obstruction in approximately half of all study participants in the lateral body position. In the supine position, there was a broad distribution of primary structures in the POSA subgroup, whereas 16 of 26 of the N-POSA subgroup had the tongue as the primary structure. If the anatomic mechanism underlying the POSA vs N-POSA differentiation relates to the lateral walls (oropharyngeal lateral wall- and, possibly, lateral velum-related obstruction), this may explain the better results seen in POSA for mandibular repositioning appliances4 and at least 1 technique of palate surgery (relocation pharyngoplasty).4,5 Although these interventions may treat the lateral walls to some extent, they may have a lower chance of resolving airway obstruction with greater degrees of obstruction related to the lateral walls.
This study has important limitations. First, DISE is performed during sedation and may not reflect natural sleep. Previous research has suggested that the use of propofol to achieve loss of consciousness is associated with many of the same changes that occur during natural sleep,12 but a previous study of body position and natural sleep endoscopy suggested that tongue-related obstruction did not change with body position.13 This study of natural sleep endoscopy focused on identifying a primary structure associated with airway obstruction rather than a VOTE classification-type analysis; in addition, they had an unexpectedly high (25%) proportion of study participants with primary epiglottis-related obstruction, so the results may not be generalizable.
Relatively few study participants were obese (BMI >30), and this also limits generalizability of the study’s findings. Multiple statistical analyses were performed, raising the potential for type I error; no correction was performed. The VOTE classification was used in this study because it has become widely adopted to report DISE findings, but other classification schemes may lead to different findings. In this study, only a single (blinded) review was performed, and future investigations may use multiple reviewers.
In addition, the present study did not consider other nonanatomic factors (such as arousal threshold, ventilatory control, or upper airway muscle responsiveness) that contribute to OSA that may also differ in POSA vs N-POSA. Also, all study participants were first examined in the lateral body position and then turned to the supine position; the result may be a deeper level of sedation during the supine examination and could account for no visualized obstruction in the nonsupine position even in some N-POSA study participants.
Drug-induced sleep endoscopy is routinely performed in only a single body position (often supine), but this study suggests there may be benefits to performing DISE in multiple body positions. One study has shown similarities in DISE findings between the lateral vs supine body position with head rotation to 1 side,14 so the latter may be an option. Positional OSA is receiving greater attention as a potential factor associated with outcomes of treatment other than positive airway pressure therapy, and the more detailed evaluation of DISE may determine the role of complementary therapies—or perhaps surgery with the use of the supine body position (to avoid the lateral pharyngeal wall-related obstruction that may not respond as well to surgery). Future research can examine in greater depth whether specific interventions are associated with differences in outcomes in POSA vs N-POSA and/or individual metrics such as supine vs nonsupine AHI.
In a study population of primarily nonobese adults, DISE findings were associated with differences in body position in both POSA and N-POSA, with greater anteroposterior airway narrowing (anteroposterior velum- and tongue-related) in the supine position. In the N-POSA subgroup, there was greater lateral (oropharyngeal lateral wall-related) obstruction in the lateral position. Treatments that resolve anteroposterior velum- and tongue-related obstruction may be more effective in the POSA population.
Corresponding Author: Eric. J. Kezirian, MD, MPH, USC Caruso Department of Otolaryngology–Head & Neck Surgery, Keck School of Medicine of the University of Southern California, 1450 San Pablo St, Ste 5100, Los Angeles, CA 90033 (firstname.lastname@example.org).
Accepted for Publication: October 24, 2018.
Published Online: December 20, 2018. doi:10.1001/jamaoto.2018.3692
Author Contributions: Dr Kezirian had access to all the data in the study and takes full responsibility for the integrity of the data and the accuracy of the data analysis.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Yalamanchili, Kezirian.
Critical revision of the manuscript for important intellectual content: Mack.
Statistical analysis: Mack, Kezirian.
Conflict of Interest Disclosures: Dr Kezirian reports research funding from Inspire Medical Systems; Advisory Board memberships for Nyxoah, ReVENT Medical, Pillar Palatal, Split Rock Scientific, Cognition Life Science, Berendo Scientific; and intellectual property ownership in Magnap.
Funding/Support: This publication was supported by grants UL1TR001855 and UL1TR000130 from the National Center for Advancing Translational Science (NCATS) of the U.S. National Institutes of Health.
Role of the Funder/Sponsor: The National Center for Advancing Translational Science had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Additional Contributions: We thank Yeini C. Guardia, AA, Research Associate, USC Caruso Department of Otolaryngology–Head & Neck Surgery, who performed medical record reviews, organized raw data, and edited DISE videos for review. She received salary compensation for her work according to standard institutional employment agreements.