Development and validation of a tumor microenvironment-related prognostic signature in lung adenocarcinoma and immune infiltration analysis*

Abstract

Abstract Objective Tumor-infiltrating immune cells and stromal cells in the tumor microenvironment (TME) significantly affect the prognosis of and immune response to lung adenocarcinoma (LUAD). In this study, we aimed to develop a novel TME-related prognostic model based on immune and stromal genes in LUAD. Methods LUAD data from the TCGA database were used as the training cohort, and three Gene Expression Omnibus (GEO) datasets were used as the testing cohort. The Estimation of STromal and Immune cells in MAlignant Tumor tissues using Expression data algorithm was used to analyze the immune and stromal genes involved in the TME. Kaplan-Meier and Cox regression analyses were used to identify prognostic genes and construct a TME-related prognostic model. Gene set enrichment analysis and TIMER were used to analyze the immune features and signaling pathways of the model. Results A TME-related prognostic model based on six hub genes was generated that significantly stratified patients into the high- and low-risk groups in terms of overall survival. The model had strong predictive ability in both the training (TCGA) and testing (GEO) datasets and could serve as an independent prognostic factor for LUAD. Moreover, the low-risk group was characterized by greater immune cell infiltration and antitumor immune activity than the high-risk group. Importantly, the signature was closely associated with immune checkpoint molecules, which may serve as a predictor of patient response to immunotherapy. Finally, the hub genes BTK, CD28, INHA, PIK3CG, TLR4, and VEGFD were considered novel prognostic biomarkers for LUAD and were significantly correlated with immune cells. Conclusion The TME-related prognostic model could effectively predict the prognosis and reflect the TME status of LUAD. These six hub genes provided novel insights into the development of new therapeutic strategies.