Xiaoxiang Gong1,2, Lingjuan Liu1,2, Jie Xiong1,2, Xingfang Li1,2, Jie Xu1,2, Yangyang Xiao1,2, Jian Li1,2, Xuemei Luo1, Dingan Mao1,2, Liqun Liu1,2. 1. Department of Pediatrics, The Second Xiangya Hospital, Central South University, Changsha 410011, Hunan, China. 2. Childern's Brain Development and Brain Injury Research Office, The Second Xiangya Hospital, Central South University, Changsha 410011, Hunan, China.
Abstract
BACKGROUND: Tumor microenvironment (TME) is closely related to the progression of glioma and the therapeutic effect of drugs on this cancer. The aim of this study was to develop a signature associated with the tumor immune microenvironment using machine learning. METHODS: We downloaded the transcriptomic and clinical data of glioma patients from the Chinese Glioma Genome Atlas (CGGA) databases (mRNAseq_693). The single-sample Gene Set Enrichment Analysis (ssGSEA) database was used to quantify the relative abundance of immune cells. We divided patients into two different infiltration groups via unsupervised clustering analysis of immune cells and then selected differentially expressed genes (DEGs) between the two groups. Survival-related genes were determined using Cox regression analysis. We next randomly divided patients into a training set and a testing set at a ratio of 7 : 3. By integrating the DEGs into least absolute shrinkage and selection operator (LASSO) regression analysis in the training set, we were able to construct a 15-gene signature, which was validated in the testing and total sets. We further validated the signature in the mRNAseq_325 dataset of CGGA. RESULTS: We identified 74 DEGs associated with tumor immune infiltration, 70 of which were significantly associated with overall survival (OS). An immune-related gene signature was established, consisting of 15 key genes: adenosine triphosphate (ATP)-binding cassette subfamily C member 3 (ABCC3), collagen type IV alpha 1 chain (COL4A1), podoplanin (PDPN), annexin A1 (ANXA1), COL4A2, insulin-like growth factor binding protein 2 (IGFBP2), serpin family A member 3 (SERPINA3), CXXC-type zinc finger protein 11 (CXXC11), junctophilin 3 (JPH3), secretogranin III (SCG3), secreted protein acidic and rich in cysteine (SPARC)-related modular calcium-binding protein 1 (SMOC1), Cluster of Differentiation 14 (CD14), COL1A1, S100 calcium-binding protein A4 (S100A4), and transforming growth factor beta 1 (TGF-β1). The OS of patients in the high-risk group was worse than that of patients in the low-risk group. GSEA showed that interleukin-6 (IL-6)/Janus kinase (JAK)/signal transducer and activator of transcription (STAT3) signaling, interferon gamma (IFN-γ) response, angiogenesis, and coagulation were more highly enriched in the high-risk group and that oxidative phosphorylation was more highly enriched in the low-risk group. CONCLUSION: We constructed a stable gene signature associated with immune infiltration to predict the survival rates of glioma patients.
BACKGROUND: Tumor microenvironment (TME) is closely related to the progression of glioma and the therapeutic effect of drugs on this cancer. The aim of this study was to develop a signature associated with the tumor immune microenvironment using machine learning. METHODS: We downloaded the transcriptomic and clinical data of glioma patients from the Chinese Glioma Genome Atlas (CGGA) databases (mRNAseq_693). The single-sample Gene Set Enrichment Analysis (ssGSEA) database was used to quantify the relative abundance of immune cells. We divided patients into two different infiltration groups via unsupervised clustering analysis of immune cells and then selected differentially expressed genes (DEGs) between the two groups. Survival-related genes were determined using Cox regression analysis. We next randomly divided patients into a training set and a testing set at a ratio of 7 : 3. By integrating the DEGs into least absolute shrinkage and selection operator (LASSO) regression analysis in the training set, we were able to construct a 15-gene signature, which was validated in the testing and total sets. We further validated the signature in the mRNAseq_325 dataset of CGGA. RESULTS: We identified 74 DEGs associated with tumor immune infiltration, 70 of which were significantly associated with overall survival (OS). An immune-related gene signature was established, consisting of 15 key genes: adenosine triphosphate (ATP)-binding cassette subfamily C member 3 (ABCC3), collagen type IV alpha 1 chain (COL4A1), podoplanin (PDPN), annexin A1 (ANXA1), COL4A2, insulin-like growth factor binding protein 2 (IGFBP2), serpin family A member 3 (SERPINA3), CXXC-type zinc finger protein 11 (CXXC11), junctophilin 3 (JPH3), secretogranin III (SCG3), secreted protein acidic and rich in cysteine (SPARC)-related modular calcium-binding protein 1 (SMOC1), Cluster of Differentiation 14 (CD14), COL1A1, S100 calcium-binding protein A4 (S100A4), and transforming growth factor beta 1 (TGF-β1). The OS of patients in the high-risk group was worse than that of patients in the low-risk group. GSEA showed that interleukin-6 (IL-6)/Janus kinase (JAK)/signal transducer and activator of transcription (STAT3) signaling, interferon gamma (IFN-γ) response, angiogenesis, and coagulation were more highly enriched in the high-risk group and that oxidative phosphorylation was more highly enriched in the low-risk group. CONCLUSION: We constructed a stable gene signature associated with immune infiltration to predict the survival rates of glioma patients.
Authors: Ana Rita Pombo Antunes; Isabelle Scheyltjens; Johnny Duerinck; Bart Neyns; Kiavash Movahedi; Jo A Van Ginderachter Journal: Elife Date: 2020-02-04 Impact factor: 8.140