Endotypic Mechanisms of Successful Hypoglossal Nerve Stimulation for Obstructive Sleep Apnea

Abstract

Rationale: Approximately one-third of patients with obstructive sleep apnea (OSA) treated with hypoglossal nerve stimulation (HGNS) therapy are incomplete responders, despite careful patient selection based on baseline characteristics and drug-induced sleep endoscopy.Objectives: Here we use polysomnographic endotyping to assess the pathophysiological mechanisms underlying favorable versus incomplete responses to HGNS therapy.Methods: Baseline polysomnography data of the STAR (Stimulation Therapy for Apnea Reduction) trial were included. Raw baseline polysomnographic data from 91/126 patients were available for analysis. Traits-loop gain, arousal threshold, collapsibility, and muscle compensation-were calculated from the baseline polysomnography data according to Sands and colleagues (AJRCCM 2018, SLEEP 2018). Logistic regression assessed apnea-hypopnea index (AHI)-adjusted associations between HGNS response (>50% reduction in AHI to <10/h at 1 yr) and OSA traits.Measurements and Main Results: Overall, HGNS treatment reduced AHI from 30.7 (24.9-39.9) to 8.5 (4.0-19.5) events/h (P < 0.0001; median [quartiles 1-3]); N = 53/91 were responders. In adjusted analysis, a favorable response to therapy was independently associated with higher arousal threshold (odds ratio [95% confidence interval]: 6.76 [2.44-23.3], P = 0.001), greater compensation (odds ratio: 4.22 [1.70-12.55] per SD, P = 0.004), and lower loop gain (in milder collapsibility, per significant interaction, P = 0.003). The higher arousal threshold was evident in responders before adjusted analysis. Predicted responders had an approximately fourfold lower treatment AHI versus predicted nonresponders (4.9 [2.7-8.5] vs. 20.7 [10.9-29.7], P < 0.0001; median [quartiles 1-3]); differences remained significant after cross-validation.Conclusions: Favorable responses to HGNS therapy are associated with the pathophysiological traits causing OSA, particularly a higher arousal threshold. Along with established criteria, individuals with favorable traits could potentially be prioritized for precision HGNS therapy.This analysis was a secondary analysis of the STAR trial registered with clinicaltrials.gov (NCT01161420).

Keywords: arousal threshold; pathophysiology; precision medicine; sleep-disordered breathing; upper airway stimulation.

Figure 1.

Two-trait cross-sectional “slices” of the five-dimensional multivariable model are shown. Each slice is a simplification of the full five-dimensional model; traits that are hidden from view are set to their mean value to facilitate presentation. Responders are indicated by green dots, nonresponders by orange dots. Open circles denote patients for whom the model simplification does not apply: for these patients, the hidden trait values are so far from the mean value that the two-dimensional prediction differs from the overall five-dimensional model prediction (these patients are better described by other views). Modeled response status is shown by the colored background (green for predicted responders and orange for predicted nonresponders). (A and B) Higher arousal threshold promotes hypoglossal nerve stimulation (HGNS) success independent of (A) apnea–hypopnea index and (B) collapsibility (Vpassive); arousal threshold is particularly influential in milder collapsibility (higher Vpassive). (C and D) Higher loop gain promotes HGNS failure independent of apnea–hypopnea index (C) but predominantly in those with milder collapsibility (D). (E) Greater compensation promotes HGNS success independent of apnea–hypopnea index. (F) Combination of greater compensation and lower loop gain favored HGNS success. (G) Combination of milder collapsibility (higher Vpassive) and higher apnea–hypopnea index favored HGNS failure.

Figure 2.

Reduction in apnea–hypopnea index in (A) predicted responders and (B) predicted nonresponders. Patients allocated to a different group after leave-one-out cross-validation are shown with dotted lines.