Identification of Mucus-Associated Molecular Subtypes of Chronic Obstructive Pulmonary Disease: A Latent Profile Analysis Based on MUC5B-Associated Genes.

BACKGROUND Chronic obstructive pulmonary disease (COPD) is a disease with high heterogeneity, which is a major challenge in clinical individualized treatment. A mucus phenotype is one of the main characteristics of COPD. MATERIAL AND METHODS Gene expression profiles of lung tissue samples were from the Lung Genomics Research Consortium. MUC5B-associated gene signatures were obtained based on a nonlinear feature screening algorithm. These signatures were used to fit a latent profile analysis (LPA) model to identify COPD molecular subtypes and build a subtype classifier to verify the subtypes. Then, we explored the characteristics of cilium assembly and beating signatures, transcriptome features, immune infiltration among the 3 subtypes by xCell, single-sample gene set enrichment analysis, network perturbation amplitude, and weighted gene co-expression network analysis algorithms. An external dataset was used to verify the above COPD subtypes. RESULTS Three subtypes associated with mucus were identified by LPA and verified in an external dataset. Subtype 1 displayed higher T helper type 1 (Th1) and basophil infiltration, higher Th17/regulatory T cells (Tregs) ratio, a higher level of cilium assembly and beating, and lower mast cell and Treg infiltration. The subtypes 2 and 3 demonstrated higher macrophage M2 infiltration in lung tissue, while subtype 3 had higher neutrophil and eosinophil infiltration than subtype 2. CONCLUSIONS Overall, this work identified 3 mucus-associated molecular subtypes related to MUC5B expression, which deepens the understanding of airway mucus secretion in COPD and potentially provides valuable information for precision therapy.