Scalable Sleep Apnoea Endotype Profiling and Estimated Oral Appliance Treatment Response in a Large Clinical Cohort

Authors:
Open Access
Publisher website
Google Scholar
Abstract
Background and Objective: Obstructive sleep apnoea (OSA) pathogenesis is heterogeneous with important contributions from impaired upper airway anatomy and at least three key non‐anatomical endotypes. Detailed physiology assessments in a relatively small number of well‐characterized participants indicate wide interindividual variability in pharyngeal collapsibility and that ~70% have impairment in one or more non‐anatomical endotype. In this study, we applied recently developed models that use standard clinical and polysomnography data to estimate OSA endotypes and oral appliance therapy response in a large clinical population. Methods: OSA endotypes and oral appliance therapy responses were estimated using standard sleep study outputs and clinical data (age and BMI) from > 6500 consecutive home‐based diagnostic studies with an Apnoea/Hypopnea Index (AHI) > 5 events/h (Sleep Profiler PSG‐Advanced Brain Monitoring). The proportion of participants who were estimated to have impairment in one or more OSA endotypes was calculated according to previously published physiological cutoffs. Results: On average, participants (39% female) had moderately severe OSA (AHI = 28 ± 23 events/h [Mean ± SD]), were middle aged (49 ± 16 years) and obese (BMI = 31 ± 7 kg/m 2 ). 29% were estimated to have only mild anatomical compromise. 23% had poor dilator muscles, 31% high loop gain and 29% had a low arousal threshold. 62% had severe (“bad”) impairment in one or more non‐anatomical endotypes. ~60% were predicted to respond to oral appliance therapy. Conclusions: These novel data highlight the heterogeneity of OSA in a large clinical OSA patient population and the feasibility for endotype‐informed precision medicine for OSA via application of novel physiology‐based algorithms using home sleep study recordings at scale.
Funding Information
  • National Health and Medical Research Council (1196261)

This publication has 25 references indexed in Scilit: