Site of Collapse during Drug-induced Sleep Endoscopy Is Associated with Polysomnographic Endotypic Traits

Both the site of upper airway collapse during drug-induced sleep endoscopy (DISE) and pathophysiological endotypic traits are associated with non-CPAP treatment outcomes for obstructive sleep apnea (OSA). Reduced hypoglossal nerve stimulation (HGNS) treatment efficacy has been associated with complete concentric collapse at the level of the palate (CCCp), lateral wall collapse, lower arousal threshold, and poor dilator muscle compensation. However, these predictors may not be independent. Currently, the relationship between the site of upper airway collapse (structure) and pathophysiological endotypic traits (function) remains unknown. This retrospective cohort study examined 182 patients (median[95%CI], apnea-hypopnea index (AHI): 24.2[17.6,32.8], body-mass index (BMI): 27.8[25.2,30.5], age: 51.3[40.4,58.8]) who underwent in-laboratory polysomnography and DISE. All DISE studies were scored by one researcher, thereby avoiding inter-rater variability. Endotypic traits (loop gain, collapsibility, arousal threshold, and compensation) were estimated using routine polysomnography (Sands et al. AJRCCM 2018). Linear regression quantified differences in traits between DISE categories. Multivariable logistic regression quantified associations between DISE categories (dependent variable, with versus without a certain collapse type) and individual traits. Analyses were mutually adjusted for other endotypic traits. CCCp was independently associated with greater collapsibility (Δ collapsibility = 9.8[4.6,15.0]%, p<0.001with vs. without CCCp, odds ratio = 6.9[95%CI:2.2,22.1] per 2SD increase in collapsibility [SD=15.9%];), but a lower arousal threshold (Δ arousal threshold =-8.4[-15.6,-1.2]%; OR=5.4[1.2,24.2] per 2SD [SD=24.9%];). Conversely, complete tongue base collapse was associated with less-severe collapsibility (Δ collapsibility =-5.9[-10.2,-1.6]%, OR=5.0[1.4, 17.9];), but a higher arousal threshold (Δ arousal threshold = 7.6[1.6,13.5]%, OR=5.7[1.4, 23.5];). Complete lateral wall collapse was independently associated with reduced compensation (Δ compensation =-8.0[-14.5,-1.5]%, p=0.018), OR=3.6[1.2, 10.4] per 2 SD [SD=17.5%]; whereas epiglottic collapse was associated with greater compensation (Δ compensation = 8.1[1.0,15.3]%, OR=5.8[1.1,31.2]). Findings persisted with additional adjustment for AHI and BMI, except for collapsibility and tongue base collapse. Loop gain was not associated with any site of collapse. Different sites of upper airway collapse manifest distinctly different pathophysiological traits in OSA patients. The greater collapsibility and lower arousal threshold seen with CCCp and reduced compensation with lateral wall collapse may help explain reduced non-CPAP treatment efficacy in these populations.