Abstract
Catathrenia is a rare sleep-related breathing disorder characterized by recurrent episodes of groaning during sleep, which leads to sleep fragmentation and daytime symptoms. Catathrenia could be misclassified as sleep apnea. Integrating omics data with machine learning techniques holds potential for diagnosis and advancing the understanding of its etiology. All participants included in the study underwent full-night polysomnography for evaluation. This study employed in-depth 4D-DIA proteomics to analyze salivary protein profiles in 22 catathrenia patients and 22 matched nonsnoring controls, revealing significant differences in protein characterization. Functional analysis of differentially expressed proteins indicated a predominant association with the extracellular matrix (ECM)-receptor interaction pathway. A machine learning pipeline was used for protein marker selection, and the findings were validated in a separate cohort of five catathrenia patients before and after mandibular advancement device (MAD) treatment. The study identified five potential protein markers with the highest area under the curve (AUC): HATL5 (airway trypsin-like protease 5), B2RBF5 (highly similar to Homo sapiens chitobiase, di-N-acetyl-CTBS), A0A7S5BZF8 (IGH Fragment), PRR27 (proline-rich protein 27), and BCAT2 (branched-chain-amino-acid aminotransferase). The relative abundance of these proteins was found to correlate with the Epworth sleepiness scale (ESS) and polysomnographic parameters related to groaning events. These proteins could be used as potential markers with validation in future studies. Catathrenia may be linked to ECM-related airway rigidity during sleep. Further investigation is required to elucidate the underlying mechanisms.