Characterizing heterogeneity of non‐small cell lung tumour microenvironment to identify signature prognostic genes. Issue 24 (12th November 2020)

• Record Type: Journal Article
• Authors: Mi, Kai; Chen, Fuhui; Qian, Zhipeng; Chen, Jing; Lv, Dongxu; Zhang, Chunlong; Xu, Yanjun; Wang, Hongguang; Zhang, Yuepeng; Jiang, Yanan; Shang, Desi
• Is Part Of: Journal of cellular and molecular medicine. Volume 24:Issue 24(2020)
• Page Start: 14608
• Page End: 14618
• Publication Date: 2020-11-12
• Subjects: non‐small cell lung cancer -- prognostic -- tumour microenvironment; Cytology; Medicine; Molecular Biology; Cytologie -- Périodiques; Médecine -- Périodiques; Biologie moléculaire -- Périodiques; Cytology -- Periodicals; Medicine -- Periodicals; Molecular biology -- Periodicals; 611.01805
• DOI: 10.1111/jcmm.16092
• Languages: English
• ISSNs: 1582-1838

Abstract

Abstract: Growing evidence has highlighted the immune response as an important feature of carcinogenesis and therapeutic efficacy in non‐small cell lung cancer (NSCLC). This study focused on the characterization of immune infiltration profiling in patients with NSCLC and its correlation with survival outcome. All TCGA samples were divided into three heterogeneous clusters based on immune cell profiles: cluster 1 ('low infiltration' cluster), cluster 2 ('heterogeneous infiltration' cluster) and cluster 3 ('high infiltration' cluster). The immune cells were responsible for a significantly favourable prognosis for the 'high infiltration' community. Cluster 1 had the lowest cytotoxic activity, tumour‐infiltrating lymphocytes and interferon‐gamma (IFN‐γ), as well as immune checkpoint molecules expressions. In addition, MHC‐I and immune co‐stimulator were also found to have lower cluster 1 expressions, indicating a possible immune escape mechanism. A total of 43 differentially expressed genes (DEGs) that overlapped among the groups were determined based on three clusters. Finally, based on a univariate Cox regression model, prognostic immune‐related genes were identified and combined to construct a risk score model able to predict overall survival (OS) rates in the validation datasets.