Identification of diagnostic biomarkers and immune cell profiles associated with COPD integrated bioinformatics and machine learning.

Abstract

This retrospective transcriptomic study leveraged bioinformatics and machine learning algorithms to identify novel gene biomarkers and explore immune cell infiltration profiles associated with chronic obstructive pulmonary disease (COPD). Utilizing an integrated analysis of metadata encompassing six gene expression omnibus (GEO) microarray datasets, 987 differentially expressed genes were identified. Further gene ontology and pathway enrichment analyses revealed the enrichment of these genes across various biological processes and pathways. Moreover, a systematic integration of two machine learning algorithms along with pathway-gene correlations identified six candidate biomarkers, which were validated in a separate cohort comprising six additional microarray datasets, ultimately identifying ADD3 and GNAS as diagnostic biomarkers for COPD. Subsequently, the diagnostic efficacy of ADD3 and GNAS was assessed, and the impact of their expression levels on overall survival was further evaluated and quantified in the validation cohort. Examination of immune cell subtype infiltration found increased proportions of cytotoxic CD8+ T cells, resting and activated NK cells, along with decreased M0 and M2 macrophages, in COPD versus control samples. Correlation analyses also uncovered significant associations between ADD3 and GNAS expression and infiltration of various immune cell types. In conclusion, this study elucidates crucial COPD diagnostic biomarkers and immune cell profiles which may illuminate the immunopathological drivers of COPD progression, representing personalized therapeutic targets warranting further investigation.