Integrative scRNA-seq and Scissor analysis reveal an endothelial cell-associated prognostic model and tumor microenvironment in lung adenocarcinoma

Vascular angiogenesis is a crucial mechanism that drives tumor growth and metastasis. Endothelial cells (ECs) play a pivotal role in this process by regulating intercellular interactions within the tumor microenvironment. This study aimed to identify marker genes in ECs using single-cell RNA sequencing (scRNA-seq) technology and to establish a prognostic model for lung adenocarcinoma (LUAD) patients based on these genes. Ten-fold scRNA-seq data for LUAD were retrieved from the Gene Expression Omnibus database. A prognostic model was developed based on endothelial cell-related differentially expressed genes using hierarchical density-based weighted gene co-expression network analysis (hdWGCNA) and least absolute shrinkage and selection operator (LASSO) regression. Differences in immune cell infiltration, drug sensitivity, and pathway enrichment across risk groups were assessed. The expression of model genes was validated in LUAD cells and tissues using quantitative real-time PCR (qRT-PCR). In addition, single-cell Scissor analysis revealed differences in cellular composition and spatial distribution between the risk-stratified subgroups; intercellular communication in the higher-risk subgroup was more complex and active, primarily involving adipose, epithelial, and immune-related cells, whereas overall communication in the lower-risk subgroup was comparatively weaker. Six genes (CX3CL1, MCAM, AKAP12, MGP, CA4, and CAV2) were identified through LASSO and Cox regression analyses to establish the prognostic model. SHapley Additive exPlanations (SHAP) analysis highlighted the significant contribution of CAV2 to risk prediction. Risk scores were inversely correlated with patient survival and were validated across multiple clinical subgroups. A nomogram incorporating clinical parameters and risk scores demonstrated strong predictive accuracy. Enrichment and immune analyses revealed distinct pathway activations and immune cell distributions between risk groups, with high-risk patients exhibiting greater tumor mutation burden and dedifferentiation characteristics. Drug sensitivity analysis indicated significant differences in therapeutic responses among risk groups. Experimental findings corroborated the downregulation of the model genes in LUAD samples. The endothelial cell-related prognostic model developed in this study precisely predicted survival outcomes in LUAD and revealed notable differences in biological features and treatment responses among patient subgroups, offering a theoretical framework for personalized therapeutic strategies.