0481 A Novel Phenotype Classification for Obstructive Sleep Apnea Based on Hypoxic Burden

Abstract Introduction Currently, we rely on the apnea-hypopnea index (AHI) to diagnose and quantify the severity of obstructive sleep apnea (OSA). However, AHI does not reflect the severity of respiratory events and does not capture the heterogeneity of patients with OSA. This study aimed to establish a novel phenotype classification based on a refined hypoxic burden (HB) and investigated its associations with a wide range of OSA-related health outcomes. Methods We retrospectively recruited 250 participants in each of four groups: normal (AHI 5), mild (5 ≤ AHI 15), moderate (15 ≤ AHI 30), and severe (AHI ≥ 30) OSA. To calculate HB, we adopted the method developed by Ali et al.,1 and we further divided HB into apnea-specific HB and hypopnea-specific HB, defined as HB calculated only based on apnea and hypopnea events, respectively. Hierarchical cluster analysis was performed based on the log scale of apnea-specific HB and hypopnea-specific HB to explore OSA phenotypes. We compared demographics and clinical outcome variables, including the prevalence of cardiometabolic disorders, comorbidity scores, and brain age index (BAI) based on sleep-EEG among these subtypes. Results After excluding participants with poor data quality, 954 participants were analyzed. These participants were classified into five clusters (Figure): subgroup #1 (n=177, ‘normal sleepers’), subgroup #2 (n=211, ‘mild hypopnea-driven HB’), subgroup #3 (n=225, ‘moderate HB’), subgroup #4 (n=225, ‘severe HB’), and subgroup #5 (n=116, ‘extreme HB with apnea-dominance’). Compared to subgroup #1, all other groups (subgroup #2-5) had a higher risk of hypertension and total comorbidity scores. By contrast, BAI was increased only in subgroups #4 and #5, with a significant correlation between apnea-specific HB and BAI. Different health outcomes were observed among subgroups even within the same AHI severity category. Conclusion This novel phenotype classification of OSA revealed significant associations with clinical outcomes, surpassing the limitations of the traditional categorization based on AHl. These insights could be used for risk stratification and the design of future OSA-related studies. Support (if any)