基于中性粒细胞胞外诱捕网和氧化应激相关基因的肺腺癌预后模型的构建与验证

Lung adenocarcinoma (LUAD) is a major cause of cancer-related morbidity and mortality globally, with challenges in prognosis and treatment due to its complex pathogenesis and heterogeneous tumor microenvironment (TME). Neutrophil extracellular traps (NETs) and oxidative stress play critical roles in tumor progression: NETs promote tumor cell adhesion, migration, and immune suppression, while oxidative stress induces DNA damage and activates pro-tumor signaling pathways. Moreover, oxidative stress is an important inducer of NETs, and their crosstalk shapes the LUAD immune microenvironment. However, systematic exploration of LUAD immunotherapeutic response prediction based on NETs and oxidative stress-related genes remains lacking.

The gene set related to oxidative stress was obtained from MSigDB. The gene set related to NETs was sourced from relevant literature. Transcriptomic and clinical data were integrated from The Cancer Genome Atlas (TCGA)-LUAD (training set) and GSE31210 (validation set). Weighted Gene Co-Expression Network Analysis (WGCNA) was employed to screen gene modules and characteristic scores related to NETs and oxidative stress signatures. Differentially expressed genes (DEGs) were screened, and prognostic model was established using univariate and LASSO Cox regression. Immune infiltration was analyzed using ESTIMATE algorithm, MCP-counter and ssGSEA methods. And we developed a nomogram incorporating clinicopathological features and RiskScore model, and performed drug sensitivity analysis. Finally, the biological role of CPS1 in lung cancer cells was investigated through CCK-8, wound-healing, and Transwell experiments.

22 co-expression modules were screened, among which the brown module showed significant correlations with NETs and oxidative stress signature scores. This module was intersected with DEGs, yielding 624 overlapping genes implicated in immune-relevant pathways (like leukocyte differentiation, neutrophil activation involved in immune response). A prognostic model was established utilizing 8 key genes (ADGRE3, ARHGEF3, CD79A, CLEC7A, CPS1, EPHB2, LARGE2, and OAS3). In the TCGA database, the model demonstrated robust prognostic discrimination (area under the curve (AUC) > 0.6), with high-risk patients exhibiting shorter overall survival (OS) (p < 0.05). Its stability was validated in GSE31210 (AUC > 0.6). The RiskScore showed negative correlations with immune infiltration (like T cells, CD8 T cells, and natural killer cells) as well as immune/stromal scores. A nomogram model combining RiskScore with N staging was developed and validated, demonstrating strong predictive accuracy through calibration and decision curve analyses. High-risk patients were more sensitive to drugs like BI-2536, BMS-509744, and Pyrimethamine. Finally, in vitro tests showed that CPS1 knockdown markedly decreased the viability, migration, and invasion of lung cancer cells.

The constructed prognostic model by NETs and oxidative stress-relevant genes effectively predicts LUAD prognosis, correlates with immune microenvironment characteristics, and guides drug sensitivity, providing novel insights for LUAD prognostic assessment and personalized therapy.