Obstructive sleep apnea (OSA) is a highly prevalent disorder, but reliance on the apnea-hypopnea index (AHI) alone may fail to capture its full physiological burden, particularly in milder cases. This study aimed to identify distinct physiological phenotypes in mild OSA using oximetry-based cluster analysis derived from polysomnography (PSG), and to determine whether these phenotypes exhibit differential patterns of sleep architecture and fragmentation. We conducted a cross-sectional study of 144 adults diagnosed with mild OSA, defined by an AHI between 10 and 15 events per hour. A cluster analysis was performed using conventional and novel oximetric parameters, including desaturation severity and duration. We compared sleep efficiency, sleep stages, and comorbidity burden between the identified groups. Two distinct phenotypes were identified. Cluster 1 showed significantly deeper and longer desaturations than Cluster 2, despite identical AHI levels. Patients in Cluster 1 demonstrated significantly lower sleep efficiency (78.7 vs. 86.9%), increased wakefulness after sleep onset (73.8 vs. 49.5 min), and reduced slow-wave sleep (15.4 vs. 18.4%). No significant differences were found in daytime sleepiness or the prevalence of cardiovascular and metabolic comorbidities. Mild OSA is a physiologically heterogeneous entity that extends beyond the granularity of the AHI. Oximetry-based clustering unveils a distinct phenotype characterized by significant sleep architecture impairment and hypoxic burden, independent of event frequency. These findings suggest that detailed oximetric analysis should be integrated into routine clinical assessment to identify high-risk patients who may warrant earlier therapeutic intervention.
Amorim et al. (Tue,) studied this question.