作者
Shuoyan Liu,Feng Wang,Xuan Gao,Peiyuan Wang,Yujie Chen,Peng Chen,Hao He,Xiaofeng Chen,Derong Zhang
摘要
e20512 Background: The incidence of lung adenocarcinoma (LUAD) is increasing year by year worldwide. Thorsson et al. identified six immune subtypes (ISs) (C1 - C6) in 33 diverse cancer types, including LUAD. The tumor microenvironment and lung cancer cells of 6 ISs have complex and different interplay, and the consequence of the interplay determines the growth of tumor cells and the prognosis of patients. Methods: Clinical and molecular information of TCGA LUAD patients were downloaded from The Cancer Genome Atlas (TCGA) (N = 457). Kaplan-Meier survival analysis and Cox proportional-hazard analyses were performed in 4 independent GEO LUAD data sets (GSE31210, GSE37745, GSE50081, and GSE68465) and, where five immune subtypes C1-C4, C6 were stratified according to Thorsson et al.’s immune classification method. Results: Only 5 ISs were identified in the TCGA NSCLC cohort; the proportion of the C3 inflammatory IS (39.2%) was the most common IS observed in the TCGA LUAD cohort, whereas C1 was particularly dominant in LUSC (57.1%). The proportion of C3 was also the most common IS observed in GEO LUAD cohorts (GSE37745, GSE50081, GSE68465), except for one stage I-II lung adenocarcinomas cohort (GSE31210). The prognostic relevance of the C3 IS was associated with significantly longer overall survival and progression-free interval time of LUAD in the TCGA cohort. In contrast, the association was not significant in the TCGA LUSC cohort. C3 was also associated with significantly longer overall survival (p < 0.001) and disease-free interval time (p < 0.001) in the 4 GEO LUAD cohorts by multivariate analyses. To further evaluate prediction accuracies of C3, we compared the C3 model with the other 3 reported immune-related signatures. For all comparisons in these 4 GEO cohorts, C3 outperformed the other 3 signatures with superior overall survival and DFS predictive performances according to the time-dependent Concordance index. C3 presented higher levels of CD8 T, follicular helper T, M1 macrophage, CD4 memory resting, myeloid dendritic, neutrophils, and endothelial cells by using CIBERSORT and MCP counter. Low levels of M2 macrophages and cancer-associated fibroblast cells were also found in the C3 subtype. GSEA pathway analysis showed that the C3 subtype had a heightened activation in the antigen process and presenting pathway, interferon-gamma response pathway, T cell receptor signaling pathway, and neural killer cell-mediated cytotoxicity pathway than C1/C2. In contrast, the C1/C2 presented greater activation of pathways related to the cell cycles, MMR repair DNA repair, and p53 signaling pathways than the C3 subtype. Conclusions: The robust prognostic relevance of the C3 IS allows a more precise categorization of patients with relevant clinical and biological implications, which may be valuable tools to improve tailored therapeutic interventions in LUAD patients.