Table1_Identification of the Immune Signatures for Ovarian Cancer Based on the Tumor Immune Microenvironment Genes.DOCX

Ovarian cancer (OV) is a deadly gynecological cancer. The tumor immune microenvironment (TIME) plays a pivotal role in OV development. However, the TIME of OV is not fully known. Therefore, we aimed to provide a comprehensive network of the TIME in OV. Gene expression data and clinical information from OV patients were obtained from the Cancer Genome Atlas Program (TCGA) database. Non-negative Matrix Factorization, NMFConsensus, and nearest template prediction algorithms were used to perform molecular clustering. The biological functions of differentially expressed genes (DEGs) were identified using Metascape, gene set enrichment analysis (GSEA), gene ontology (GO) and the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis. The copy number variations (CNVs), single nucleotide polymorphisms (SNPs) and tumor mutation burden were analyzed using Gistic 2.0, R package maftools, and TCGA mutations, respectively. Estimation of STromal and Immune cells in MAlignant Tumor tissues using Expression data and CIBERSORT were utilized to elucidate the TIME. Moreover, external data from the International Cancer Genome Consortium (ICGC) and ArrayExpress databases were used to validate the signature. All 361 samples from the TCGA OV dataset were classified into Immune Class and non-Immune Class with immune signatures. By comparing the two classes, we identified 740 DEGs that accumulated in immune-related, cancer-related, inflammation-related biological functions and pathways. There were significant differences in the CNVs between the Immune and non-Immune Classes. The Immune Class was further divided into immune-activated and immune-suppressed subtypes. There was no significant difference in the top 20 genes in somatic SNPs among the three groups. In addition, the immune-activated subtype had significantly increased proportions of CD4 memory resting T cells, T cells, M1 macrophages, and M2 macrophages than the other two groups. The qRT-PCR results indicated that the mRNA expression levels of RYR2, FAT3, MDN1 and RYR1 were significantly down-regulated in OV compared with normal tissues. Moreover, the signatures of the TIME were validated using ICGC cohort and the ArrayExpress cohort. Our study clustered the OV patients into an immune-activated subtype, immune-suppressed subtype, and non-Immune Class and provided potential clues for further research on the molecular mechanisms and immunotherapy strategies of OV.

卵巢癌（Ovarian cancer, OV）是一种致死率极高的妇科恶性肿瘤。肿瘤免疫微环境（Tumor Immune Microenvironment, TIME）在卵巢癌的发生发展中发挥关键作用，但目前学界对卵巢癌的肿瘤免疫微环境仍未完全阐明。因此本研究旨在构建一套完整的卵巢癌肿瘤免疫微环境调控网络。研究从癌症基因组图谱计划（The Cancer Genome Atlas Program, TCGA）数据库获取了卵巢癌患者的基因表达数据与临床信息。采用非负矩阵分解（Non-negative Matrix Factorization）、NMFConsensus以及最近邻模板预测算法进行分子分型；通过Metascape、基因集富集分析（Gene Set Enrichment Analysis, GSEA）、基因本体（Gene Ontology, GO）及京都基因与基因组百科全书（Kyoto Encyclopedia of Genes and Genomes, KEGG）通路富集分析，鉴定差异表达基因（Differentially Expressed Genes, DEGs）的生物学功能；分别使用Gistic 2.0、R包maftools以及TCGA突变数据，分析了拷贝数变异（Copy Number Variations, CNVs）、单核苷酸多态性（Single Nucleotide Polymorphisms, SNPs）与肿瘤突变负荷。利用基于表达数据的恶性肿瘤组织基质与免疫细胞定量算法（Estimation of STromal and Immune cells in MAlignant Tumor tissues using Expression data）以及CIBERSORT解析肿瘤免疫微环境。此外，本研究还从国际癌症基因组联盟（International Cancer Genome Consortium, ICGC）与ArrayExpress数据库获取外部数据，用于验证所构建的分子分型特征。TCGA卵巢癌数据集的全部361例样本被分为免疫型与非免疫型两类，且具有显著的免疫特征差异。通过对比两类样本，共鉴定出740个差异表达基因，这些基因显著富集于免疫相关、肿瘤相关及炎症相关的生物学功能与通路中。免疫型与非免疫型样本的拷贝数变异存在显著差异。进一步将免疫型样本划分为免疫激活亚型与免疫抑制亚型。三组样本的前20个体细胞单核苷酸多态性基因无显著差异。此外，免疫激活亚型的静息CD4记忆性T细胞、T细胞、M1型巨噬细胞及M2型巨噬细胞的浸润比例显著高于另外两组。实时荧光定量PCR（qRT-PCR）结果显示，相较于正常组织，卵巢癌组织中RYR2、FAT3、MDN1及RYR1的mRNA表达水平显著下调。最后，本研究通过ICGC队列与ArrayExpress队列验证了卵巢癌肿瘤免疫微环境的分型特征。本研究将卵巢癌患者划分为免疫激活亚型、免疫抑制亚型与非免疫型三类，为卵巢癌的分子机制研究及免疫治疗策略开发提供了潜在的研究方向。