Analysis of immune characteristics and inflammatory mechanisms in COPD patients: a multi-layered study combining bulk and single-cell transcriptome analysis and machine learning.

Objective: This study aims to investigate the potential roles and mechanisms of inflammatory genes in COPD. Methods: Transcriptome data from the airway epithelial tissues of COPD patients and normal individuals were downloaded from the GEO database. Differential gene expression analysis was performed using R software and its limma package, followed by GO, KEGG, and GSEA enrichment analyses. Inflammatory-related differentially expressed genes were screened based on literature data and analyzed for pathway enrichment using the Metascape database. Inflammatory-related COPD feature genes were selected using Lasso regression and random forest algorithms, and a COPD risk prediction model was constructed. Differences between the immune microenvironment of COPD and normal samples were analyzed using the ESTIMATE algorithm, the CIBERSORT method, and single-cell sequencing data. COPD patients were clustered using the ConsensusClusterPlus algorithm, and the pathway activity differences of different inflammatory subtypes were analyzed using GSVA. Potential traditional Chinese medicine monomer components capable of targeting key biomarker proteins were screened using the HERB database, and their binding potential was evaluated through molecular docking and molecular dynamics simulations. Results: A total of 495 significantly differentially expressed genes were identified, showing distinct expression patterns between COPD patients and healthy individuals. Functional and pathway enrichment analyses revealed significant enrichment of processes such as keratinocyte differentiation, arachidonic acid metabolism, IL-17 signaling pathway, and TNF signaling pathway in COPD. Fourteen inflammatory-related COPD genes were identified, which were significantly enriched in immune system processes and inflammatory responses. Using Lasso regression and random forest algorithms, seven feature genes were selected to construct a COPD risk prediction model, which demonstrated good accuracy. Immune cell infiltration analysis revealed a significant increase in monocytes, M0 macrophages, and eosinophils in COPD patients. Clustering analysis identified two inflammatory subtypes, with genes such as CLEC5A and CXCL8 significantly upregulated in the C2 subtype. Cinnamaldehyde, a potential traditional Chinese medicine monomer component, was identified to potentially exert anti-inflammatory effects by targeting the CXCL8 protein. Conclusion: This study reveals significantly enriched biological processes and pathways in COPD patients, identifies multiple inflammatory-related COPD feature genes, and finds that cinnamaldehyde may have potential therapeutic effects on inflammatory subtypes of COPD.