Plasma Profiling Reveals a Blood-Based Metabolic Fingerprint of Obstructive Sleep Apnea

RATIONALE: Obstructive sleep apnea (OSA) is a chronic, heterogeneous and multicomponent disorder with associatedcardiovascular and metabolic alterations. Despite being the most common sleep-disordered breathing, it remains a significantlyundiagnosed condition. The study of metabolism at the molecular level has emerged as a tool with great potential not only in theidentification of novel biomarkers, but also can contribute to the understanding of disease pathogenesis. We examined thecomplete plasma metabolome and lipidome of patients with suspected OSA, with the aim of identifying potential diagnosisbiomarkers and providing insights into the pathophysiological mechanisms underlying the disease. Additionally, we evaluated theimpact of continuous positive airway pressure (CPAP) treatment on the circulating metabolomic and lipidomic profile.

METHODS: Observational-prospective-longitudinal study including 206 consecutive subjects referred to the Sleep Unit forsuspected OSA. OSA was diagnosed as an apnea-hipoapnea index ≥15 events/h after a full overnight polysomnography (PSG).Patients treated with CPAP were followed-up for 6 months. Fasting blood samples were obtained at baseline and after the follow-up period at the same time of the day. Untargeted metabolomic and lipidomic analysis was performed in plasma using the goldstandard procedure: liquid chromatography coupled to mass spectrometry.

RESULTS: The study population was mainly middle-aged (median age 50.0 years), male (67.5%) and overweight-obese (median BMI 29.3 kg/m2). After adjustment for confoundingfactors (age, sex and BMI), a plasma profile composed of 33 metabolites was identified in OSA vs. non-OSA patients (Fig. A,B).The identified metabolites were mainly glycerophospholipids and bile acids. Hierarchical Clustering Analysis (Fig. C) andPrincipal Component Analysis (Fig. D) showed that metabolite levels were able to separate patients with OSA from those withoutOSA, depicting a differential metabolic plasma pattern of the OSA patient. This profile correlated with specific PSGmeasurements of OSA severity related to sleep fragmentation and hypoxemia (Fig. E). Multivariate machine learning analysisdisclosed a blood-based signature based on 4 molecules which provided an accuracy (95%CI) of 0.98 (0.95-0.99) for OSAdetection. CPAP treatment was associated with changes in 5 plasma metabolites previously altered by OSA.

CONCLUSIONS:This exploratory analysis of the circulating metabolome and lipidome reveals a molecular fingerprint of OSA, which was partiallymodulated after effective CPAP treatment. Our results suggest blood-based biomarker candidates with potential application inthe management of OSA and suggest the activation of adaptive mechanisms in response to OSA-derived hypoxia.