Identification and experimental validation of biomarkers for chronic obstructive pulmonary disease complicated with pulmonary arterial hypertension based on bioinformatics and machine learning

Objective To identify the key biomarkers for diagnosing chronic obstructive pulmonary disease (COPD) complicated with pulmonary arterial hypertension (PAH) using bioinformatics, and validate their clinical significance. Methods High-throughput sequencing data analysis was employed to identify differentially expressed genes (DEGs) in COPD-PAH. Functional enrichment analysis was then conducted to explore the biological functions of these DEGs. Machine learning methods, including least absolute shrinkage and selection operator (LASSO), random forest (RF), and support vector machine-recursive feature elimination (SVM-RFE), were utilized to screen 5 potential biomarkers. Single-cell analysis was performed to reveal the expression patterns of these key genes in macrophages. The clinical significance of these biomarkers was further validated using peripheral blood mononuclear cells (PBMC) data. A mouse model of COPD-PAH was established using hypoxia exposure. Sixteen mice (either sexes, 8 weeks old, weighing 20~22 g) were randomly divided into a hypoxia group [O2 (10.0±0.5)%, COPD-PAH, n=8] and a normoxia group (COPD, n=8). Immunofluorescence assay was used to label the key biomarkers, and their expression levels were quantified. Results A total of 28 DEGs (|Log2FC|≥2, P<0.05) were identified in COPD-PAH patients. Functional enrichment analysis indicated that DEGs in COPD were primarily associated with major histocompatibility complex (MHC) Ⅱ and cell division, and involved in lysosomes, oxidative phosphorylation, and cell cycle pathways (P<0.05). Machine learning identified 5 potential biomarkers (GRN, KLF4, SHTN1, LRP1, and GPNMB), and subsequent single-cell analysis revealed that these markers exhibited reverse expression patterns during disease progression. A nomogram model constructed based on PBMC data yielded an area under the curve (AUC) of 0.907 in diagnosing COPD-PAH. GRN, KLF4, SHTN1, LRP1 and GPNMB were significantly upregulated in the COPD-PAH group (P<0.05).Conclusion‍ ‍GRN, KLF4, SHTN1, LRP1 and GPNMB are identified as key biomarkers for the prediction and diagnosis of COPD-PAH, which providing new insights for the clinical and treatment of the condition.