Multi-omics characterization of early chronic obstructive pulmonary disease

Abstract

Abstract Chronic obstructive pulmonary disease (COPD) is projected to become the third leading cause of death globally by 2030, accounting for 71.9% of chronic respiratory diseases cases in 2019. Early COPD (ECOPD) diagnosis heavily relies on clinically monitoring of lung functions, with a strong influence from smoking exposures, which may not align well with disease progression. As such, the GOLD 2022–2024 guidelines emphasize the discovery of biological markers over clinical symptoms for early detection. This study explores the biological characteristics of ECOPD in a cohort of 176 adults from China Pulmonary Health Study, consisting 88 healthy controls (HC) and 88 clinically diagnosed ECOPD, matched for age, gender and smoking history. While lung function tests revealed differences between HC and ECOPD, no significant distinctions were observed in routine blood tests. Proteomics analysis identified 377 plasma proteins common to both groups, with low-intensity proteins driving group-specific differences. Univariable logistic regression and gene set enrichment analysis identified 248 proteins associated with ECOPD, particularly those involved in inflammation process. Validation in an independent cohort confirmed the association of 15 proteins with ECOPD. Metabolomics analysis of the plasma identified 1788 metabolites, 137 of which were found linked to ECOPD. Machine learning models indicated that a multi-omics approach provided the best predication of lung function (R2 = 0.74), while proteomics alone effectively diagnosed ECOPD (AUC = 0.949). Similarity network fusion and clustering revealed two ECOPD subgroups: one by markers of inflammatory-immune response, and the other by the presence of those related to hemostasis or the vascular smooth muscle function. These findings underscore the potential of multi-omics integration in distinguishing ECOPD subgroups and predicting disease risk.