Identification and Validation of a Two-Gene NK Cell-Related Risk Model for COPD: Integration of Single-Cell and Bulk RNA-Seq Analysis.

Abstract

Background: Chronic obstructive pulmonary disease (COPD) involves chronic inflammation with potential involvement of natural killer (NK) cells, but NK cell-related diagnostic markers remain limited. This study aimed to identify NK cell-related hub genes and construct a risk model for COPD.

Methods: This work was mainly based on multiple transcriptomic datasets, including single-cell RNA-seq data (GSE173896, 5 COPD vs 2 control) and bulk data (GSE38974 (23 COPD vs 9 control), GSE8545 (18 COPD vs 18 control), GSE11784 (22 COPD vs 72 control)). NK cell-related differentially expressed genes (DEGs) were identified. GO, KEGG, and LASSO logistic regression were applied to screen hub genes and build a risk score model. ROC analysis evaluated model performance. Immune cell infiltration was assessed via CIBERSORT.

Results: A total of 135 NK cell-related DEGs were identified. After cross-analysis, two hub genes, JUNB and TNFAIP3, were selected to construct the risk model, both significantly upregulated in COPD comparing with controls (p <0.05). The risk model showed relatively good performance, achieving AUCs of 0.928 (95% CI: 0.891-0.962) in training set and 0.754 (95% CI: 0.674-0.835) in validation set. High-risk patients showed increased infiltration of monocytes and macrophages M0, and all differential immune cells exhibited significant positive/negative correlation with the risk score.

Conclusion: Our two-gene NK cell-related diagnostic risk model shows good discriminatory ability for distinguishing COPD patients, providing insights into inflammatory and immune associations. The model holds promise as a potential non-invasive diagnostic tool and may inform personalized therapeutic strategies for COPD patients.