A risk model including eight tumor microenvironment-related genes for prediction of lung cancer prognosis

Abstract

Abstract Background The tumor microenvironment (TME) plays a crucial role in lung cancer development and outcome. In this study, we constructed a novel risk model using TME-related genes to predict the prognosis of lung adenocarcinoma (LUAD). Methods TME-related genes were collected from the literature, and the LUAD transcriptome profile and clinical characteristics from patients were retrieved from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) as the training and validation cohorts, respectively. In the training cohort, K-mean Cluster and Kaplan–Meier curve analyses were performed to examine the association of the TME-related genes with LUAD, while univariate Cox regression and LASSO Cox regression analyses assessed the key genes to construct a predictive risk model for LUAD prognosis. This risk model was then confirmed in the validation cohort using Kaplan–Meier and receiver-operating characteristic (ROC) curve analyses and then compared with other models and LUAD TNM stage. The interaction of this predictive risk model of genes with immune-related genes was also assessed using CIBERSORT, TIMER, and GEPIA. Results After screening 760 TME-related genes, we established a risk model containing ANGPTL4, FUT4, CDC25C, FLNC, KRT6A, NEIL3, HS3ST2, and DAAM2 that independently predicted LUAD prognosis in TCGA data. ROC curve and C-index confirmed the usefulness of this risk model, and a nomogram that integrated this predictive risk model with age and TNM stages was more effective in predicting LUAD prognosis. The risk model was further confirmed using GEO data. Furthermore, the risk model of genes interacted with 11 types of immune cells and three immune checkpoint molecules (LAG3, PDL1 and TDO2) in LUAD. Conclusion We constructed a predictive risk model and a nomogram that integrated the predictive risk model with age and TNM stage to predict LUAD prognosis. This predictive risk model of genes could interact with immune checkpoint genes. Future studies are required to validate these data.