B21-04 A Prognostic Risk Model Based on the Tumor Microenvironment in Lung Adenocarcinoma

Abstract Rationale Lung adenocarcinoma (LUAD) is a predominant histopathological subtype of lung cancer, representing over 40% of cases and characterized by a persistently poor 5-year survival rate of approximately 21%. A more comprehensive understanding of the tumor microenvironment (TME) is essential for enhancing prognostic evaluation and informing treatment strategies. Methods We analyzed gene expression data from 492 LUAD patients to investigate TME heterogeneity. Using single-sample gene set enrichment analysis (ssGSEA) of 207 tumor and immune-related signatures, we performed consensus clustering to identify distinct molecular subtypes. For survival analysis, statistical analysis between Kaplan-Meier curves was performed using the log rank test. To develop a prognostic model, we identified differentially expressed genes (DEGs) between clusters and integrated them with hub genes identified through weighted gene co-expression network analysis (WGCNA). The final gene signature was refined using least absolute shrinkage and selection operator (LASSO) and Cox regression analyses. The resulting risk model was used to stratify patients, and its clinical relevance was evaluated through gene set enrichment analysis (GSEA), and mutation profiling. Results Consensus clustering identified two distinct subtypes: Cluster One, characterized by high expression of tumor proliferation and matrix remodeling signatures, and Cluster Two, enriched with immune features and associated with favorable prognosis. From 151 overlapping DEGs and hub genes, an eight-gene prognostic signature was constructed using LASSO and Cox regression analyses. Patients were stratified into high- and low-risk groups based on a defined cutoff. The high-risk group exhibited a poorer prognosis and showed upregulation of cell cycle-related pathways, including the G2/M checkpoint and DNA repair. This group was also significantly enriched for late-stage disease. Furthermore, the high-risk group had a significantly higher mutation burden in cancer-related pathways, particularly involving TP53. Conclusion This study delineates the TME-driven molecular heterogeneity in LUAD and establishes a novel eight-gene prognostic model. This model effectively stratifies patients into risk groups with distinct clinical outcomes, TME, and mutational profiles, potentially informing future clinical decision-making and therapeutic development. This abstract is funded by: National Natural Science Foundation of China