Development of Diagnostic and Predictive Models for COPD Based on Anoikis Resistance.

Abstract

Background: Chronic obstructive pulmonary disease (COPD) pathogenesis involves persistent airway inflammation and remodeling, yet the role of anoikis resistance remains poorly characterized. This study aimed to identify anoikis resistance-related hub genes and evaluate their clinical utility in COPD phenotyping and prognosis.

Methods: Integrated bioinformatics analysis of the GSE11906 dataset identified anoikis resistance-related differentially expressed genes (DEGs). Functional enrichment, LASSO regression, and machine learning (RF, SVM, XGB, GLM) were employed to pinpoint core hub genes. Multi-level validation included external datasets (GSE19407), in vitro (CSE-stimulated 16HBE cells), in vivo (cigarette smoke-exposed mice), and clinical samples (PBMCs). Diagnostic and prognostic models were developed using logistic regression.

Results: Five core hub genes (UCHL1, ME1, SLC2A1, BMP4, CRABP2) were identified, with ME1, SLC2A1, and BMP4 consistently upregulated in COPD across models and strongly correlated with emphysema index (negative, R = -0.41 to -0.45) and airway wall thickness (positive, R = 0.40-0.45). These genes exhibited significant associations with peribronchial immune cell infiltration. Diagnostic models for emphysema-predominant COPD (AUC = 0.860) and disease staging (AUC = 0.882), along with a prognostic model for hospitalization duration (AUC = 0.867), demonstrated robust clinical performance.

Conclusion: ME1, SLC2A1, and BMP4 are pivotal anoikis resistance-related biomarkers in COPD, driving immune dysregulation and structural remodeling. The developed models enable precise phenotyping, severity stratification, and personalized prognosis prediction, advancing precision medicine strategies for COPD management.