Multi-omics analysis reveals glutathione metabolism-related immune suppression and constructs a prognostic model in lung adenocarcinoma.

BACKGROUND

Metabolic reprogramming within the tumor microenvironment plays a pivotal role in tumor progression and therapeutic responses. Nevertheless, the relationship between aberrant glutathione (GSH) metabolism and the immune microenvironment in lung adenocarcinoma, as well as its clinical implications, remains unclear.

METHODS

We leveraged genome-wide association study (GWAS) data and applied genetic causal analysis to evaluate the causal relationships among plasma 5-oxoproline levels, lung adenocarcinoma (LUAD) risk, and 731 immune phenotypes. We incorporated single-cell RNA sequencing data from LUAD to compare transcription factor activity, cell communication networks, and CD8 T cell subset distributions across distinct GSH metabolic groups, followed by pseudotime analysis. Whole-transcriptome data from the TCGA database were analyzed for functional enrichment, immune infiltration, and immune functionality. Prognostic genes were identified using WGCNA and LASSO-Cox regression, and the expression was validated via qRT-PCR. Thereafter, immunotherapeutic efficacy and drug sensitivity were predicted using the TIDE platform and the oncoPredict package. A prognostic model was constructed to forecast patient survival, which was further validated in two independent GEO datasets.

RESULTS

Genetic causal analysis indicated a positive correlation between plasma 5-oxoproline levels and LUAD risk. ScRNA-seq analysis revealed an increased proportion of exhausted CD8 T cells in the high GSH metabolic group, accompanied by altered transcription factor activity and distinct cell communication patterns. Furthermore, whole-transcriptome data analysis demonstrated that patients with a high metabolic phenotype exhibited significantly diminished immune functionality and overall immune infiltration. Using WGCNA and LASSO-Cox regression, we ultimately identified three key genes (LCAL1, RHOV, and MARCHF4) and generated a gene risk score. This score effectively predicts both immunotherapy response and drug sensitivity. qRT-PCR confirmed the upregulation of MARCHF4 in LUAD cells. In addition, stratification by gene risk scores revealed significant differences in immune cell infiltration, immunotherapeutic response, and drug sensitivity. The nomogram model demonstrated strong predictive accuracy in both the TCGA cohort and two independent GEO validation datasets.

CONCLUSIONS

GSH metabolic reprogramming may suppress antitumor immunity by modulating transcription factor activity, remodeling cell communication networks, and regulating CD8+ T cells. The prognostic risk model developed herein effectively predicts immunotherapeutic response, drug sensitivity, and overall survival in patients with LUAD.