Integrative multi-omics identifies a four-gene diagnostic signature linking immune dysfunction and airway remodeling in COPD
Background Early diagnosis of chronic obstructive pulmonary disease (COPD) remains challenging due to the limited sensitivity of spirometry and imaging in early-stage disease and the lack of reliable blood-based biomarkers. COPD is increasingly recognized as a heterogeneous disorder driven by coordinated immune dysregulation and airway structural remodeling; however, clinically applicable molecular signatures that capture these processes are still lacking. Objectives To identify and validate a blood-based diagnostic signature for COPD and to explore the underlying immune and structural remodeling mechanisms. Design A multi-stage integrative study combining retrospective bioinformatics analysis with prospective experimental validation. Methods Public lung transcriptomic datasets were used to identify COPD-associated candidate genes through differential expression, co-expression network, and machine-learning analyses. Candidate genes were validated by RT-qPCR in peripheral blood samples from stable COPD patients and healthy controls. Exploratory immune, single-cell, virtual perturbation, and molecular simulation analyses were performed to assess potential biological relevance. Results Four lung-derived candidate genes, AC079767.4, CEP55, EMR3, and ATP6V0D2, were identified. In peripheral blood validation, CEP55 and ATP6V0D2 showed the strongest diagnostic performance, whereas AC079767.4 showed limited blood-based discriminatory ability. Functional and immune analyses suggested that these genes may be associated with intracellular pH regulation, vesicular acidification, cytokinesis, immune imbalance, and inflammatory pathways. Single-cell analysis indicated that CEP55 was mainly enriched in epithelial, endothelial, and smooth muscle cells, whereas ATP6V0D2 was enriched in macrophages and monocytes. Virtual perturbation and molecular simulation analyses provided hypothesis-generating evidence for possible involvement in antigen presentation, T-helper-cell-related pathways, and ligand–target interactions. Conclusion This study identified four lung-derived COPD-associated candidate genes through integrative transcriptomic, network, and machine-learning analyses. Among them, CEP55 and ATP6V0D2 showed the strongest peripheral-blood diagnostic performance and biological plausibility, supporting their prioritization as candidate circulating biomarkers for further validation. Exploratory docking and MD simulations suggested possible ligand–target interactions involving CEP55 and ATP6V0D2, but these computational findings require experimental pharmacological validation before any therapeutic relevance can be inferred.