449 The Effects of Positional and Sleep Stage Dependency on Mandibular Advancement Device Treatment Outcome in Obstructive Sleep Apnea

Introduction Mandibular advancement device (MAD) responder phenotype are not well understood in patients with obstructive sleep apnea (OSA). Recent studies have reported the association between MAD treatment response and polysomnographic phenotypes using positional and sleep stage dependency, but with inconsistent findings. Thus, the study aims to investigate the relationship between the two phenotypes and MAD response. Methods This retrospective study recruited patients with OSA (apnea-hypopnea index [AHI] >10/h), who were 20 to 80 years old, treatment naïve, and received MAD treatment for more than three months from 2009 to 2017. AHIsupine/AHInon-supine ≥2 and <2 meant supine predominant (supine-p) and non-positional OSA, respectively. REM-AHI/NREM-AHI ≥2, ≤0.5, and between 0.5 to 2 indicated REM-predominant (REM-p), NREM-predominant (NREM-p), and stage-independent (SI) OSA, respectively. Three criteria defined successful MAD treatment (i.e., criterion 1: residual AHI <5/h with >50% reduction; criterion 2: residual AHI 50% reduction; criterion 3: reduction >50%). The association between the two phenotypes and the three treatment criteria was identified using multivariable logistic regression. Results A total of 218 patients with a median age of 52.5 years, body mass index (BMI) of 25.4 kg/m2, and AHI of 28.2/h were recruited. Supine-p OSA had lower waist circumferences than non-positional OSA. The REM-p group had lower AHI and more female than the NREM-p and SI group. Supine-p OSA had better response than non-positional OSA (criterion 1: 43.2% vs 34.1%; criterion 2: 63.6% vs 34.1%; criterion 3: 77.3% vs 51.2%). NREM-p OSA had lower response across all three criteria (REM-p vs NREM-p vs SI: criterion 1: 57.6% vs 0% vs 42.0%; criterion 2: 75.8% vs 16.7% vs 56.5%; criterion 3: 75.8% vs 33.3% vs 77.1%). The odds of MAD response for supine-p OSA was 3.78 (95% CI = 1.44–9.93) to 3.98 (95% CI = 1.58–9.99)-fold than non-positional OSA while the odds for NREM-p OSA were 0.06 (95% CI = 0.01–0.58) to 0.15 (95% CI = 0.03–0.67)-fold than SI OSA after adjusting demographics and clinical features affecting MAD response. Conclusion Positional and sleep stage dependency were associated with MAD response and could be indicators for personal-tailored OSA treatment. Support (if any) The Ministry of Science and Technology, Taiwan (MOST 109-2314-B-002-252)