Metabolic Subphenotypes of Obstructive Sleep Apnea: NHANES 2017-2020 (pre-pandemic)

BackgroundOSA and MetS have a bidirectional relationship but increasing evidence suggests metabolic heterogeneity in OSA, systematic phenotyping of metabolic drivers in OSA are lack. ObjectiveTo identify metabolic subphenotypes of OSA and elucidate potential pathophysiological mechanisms using population-level data. MethodsTo analyze the data related to OSA and MetS from 2,260 participants in the NHANES database (2017-2020, pre-pandemic) using PCA and machine learning. Results48.74% (1689/3465) participants were identified as OSA. PCA revaled that the first 6 PCs explained 85% of the variance in both overall and OSA participants, encompassing indicators of obesity, hypertension, dyslipidemia, and IR. Indicators with |loading value|[≥]0.6 in each PCs included obesity (Height, Weight, BMI, Waist, Hipline, W/H), blood pressure, blood lipids (CHOL, TG, LDL, HDL), and IR (GHB, GLU, VAI, LAP, TyG). Cluster analysis divided the overall participant into 6 clusters. Compared with Cluster 1 (32.92%), Cluster 2 (57.04%), Cluster 3 (57.21%), Cluster 4 (61.94%), Cluster 5 (47.69%) and Cluster 6 (50.80%) represented groups with higher prevalence of OSA, characterized by IR, isolated obesity, central obesity combined with hypertension, dyslipidemia and hypertension respectively. OSA participants were divided into 8 clusters. Cluster A, Cluster C, Cluster E, Cluster F were characterized by hypertension, dyslipidemia, isolated obesity and obesity combined with hypertension respectively. Cluster B had metabolic indicators better than average level. Clusters G and H were mainly characterized by IR. ConclusionMetabolic heterogeneity in the population is associated with the incidence of OSA. Metabolic characteristics of OSA populations may guide the treatment of OSA and its comorbidities.