Distinct Hypoxemic Profiles of Obstructive Sleep Apnea in Southern Italy: The Living with OSA and CPAP Study.

Background: Obstructive sleep apnea (OSA) is a heterogeneous disorder associated with substantial cardiometabolic and neurocognitive morbidity. Although the apnea-hypopnea index (AHI) remains the conventional measure of OSA severity, it only partially reflects the underlying pathophysiological complexity. Growing evidence indicates that nocturnal hypoxemia may be a more powerful marker of adverse outcomes than event frequency alone. Therefore, this study aimed to identify distinct OSA phenotypes based on oximetry-derived features and to assess whether these profiles offer additional clinical insight beyond traditional AHI-based classification. Methods: This multicenter retrospective study, part of the Living with OSA and CPAP: The Apulia Region Experience project, included 1386 adults diagnosed with OSA across 15 sleep centers in Southern Italy. Standardized clinical, anthropometric, and polysomnographic (PSG) data were collected. Hierarchical clustering analysis was performed based on PSG oximetry-derived variables. Resulting clusters were compared across demographic, clinical, hypoxemic, and therapeutic features. Results: Three reproducible clusters emerged. Cluster 1 (mild-non-obese) included younger, leaner patients with lower AHI (22.9 ± 10.5 events·h-1), minimal desaturation (T90 5.6 ± 7.6%), and limited comorbidities. Cluster 2 (severe-obese-hypoxemic) represented the most critical phenotype, characterized by marked obesity (BMI 39.2 ± 8.2 kg·m-2), severe OSA (AHI 74.9 ± 17.9 events·h-1), profound nocturnal hypoxemia (T90 51.5 ± 28.2%), and a high prevalence of metabolic disorders (76%), requiring higher CPAP pressures and frequent oxygen supplementation. Cluster 3 (older-comorbid) comprised older males (63.7 ± 11.8 years) with moderate-to-severe OSA (AHI 44.8 ± 15.2 events·h-1) and multiple cardiometabolic comorbidities. Conclusions: Oximetry-derived variables identify distinct and clinically meaningful OSA phenotypes that extend beyond traditional AHI-based classification. Recognizing hypoxemia-driven subtypes could improve risk stratification and enable more personalized management strategies in clinical practice.