Increasing evidence has indicated that hypoxia-related genes are associated with the malignant progression of ovarian cancer [19–21]. However, most of the research focused on exploring the function of single hypoxia-related genes in OC. There have been fewer reports investigating the significance of hypoxia-related signature sets in OC. In this study, we extracted the RNA transcriptome profiles of 15 hypoxia-related genes from TCGA-OV dataset. Then, we classified the 379 OC patients into two groups (cluster.A and cluster.B) based on the expression levels of the 15 hypoxia-related genes using the consensus clustering method. We explored significant differences in immune infiltration level between the cluster.A and cluster.B groups. Notably, previous studies also identified the relationship between hypoxia and immune infiltrates in cancer. For example, Wang et al. revealed that hypoxia can aggravate immune cell infiltration to promote the development of hepatocellular carcinoma (HCC) by inducing myeloid derived growth factor (MYDGF) [22]. Moreover, Xiao et al. established a four-gene tumor microenvironment-related model based on TCGA-GBM dataset. These four genes were reported to potentially be associated with the hypoxic phenotype of glioblastoma [23]. Furthermore, Gong et al. divided the breast cancer patients from TCGA database into two groups (cluster1 and cluster2) based on the expression levels of 13 hypoxia-related genes. The results indicated that the immune infiltrating state differed between cluster1 and cluster2, suggesting the hypoxia can regulate immune infiltrates in breast cancer [24]. Our study further supports the relationship between hypoxia and immune infiltrates in OC, but the potential mechanism behind this should be investigated. We hypothesized that patients in the cluster.A group were more sensitive to immune checkpoint inhibitors (ICIs) therapy, given their higher level of immunoinfiltration. We then compared the expression levels of immune checkpoint molecules between the cluster.A and cluster.B groups. Consistent with our expectations, the immune checkpoint molecules had higher expression levels in the cluster.A group than in the cluster.B group. We also found that the immune activation-related genes were upregulated in the cluster.A group, which again met our expectations. EMT is a biological process in which epithelial cells transform into mesenchymal ones through specific processes. Previous studies reported that hypoxia was involved in the specific processes of EMT [25–27]. Our research also supported this, in that patients in the cluster.A group with an activated hypoxic status showed higher expression levels of EMT-related genes.
To further confirm the hypoxic status, consensus clustering was performed to obtain two distinct hypoxia gene statuses (gene cluster.A and gene cluster.B) based on the 227 hypoxia-related DEGs. We found that the gene cluster.A group corresponded to the cluster.A group. The patients in the gene cluster.A group exhibited higher expression of immune checkpoint-related genes, EMT-related genes, and immune activation-related genes, as well as elevated immune infiltrates. Moreover, GO and KEGG functional enrichment analyses based on the 227 hypoxia-related DEGs further confirmed that hypoxia was associated with inflammatory response, immune response, and EMT of OC. To quantify the hypoxic state of patients and facilitate clinical interpretation, a PCA algorithm was used to calculate a hypoxia score for each patient. The patients with a high score corresponded to the cluster.A group and the gene cluster.A group. A high hypoxia score was associated with activated immune infiltrates, and the upregulation of immune checkpoint-related genes, EMT-related genes, and immune activation-related genes, suggesting that the hypoxia score calculated by the PCA algorithm can well quantify the hypoxic state of OC patients. All of the above results repeatedly indicate the important role of hypoxia in the immune infiltration and malignant progression of OC. The mutual validation of the analysis results led us to believe that the hypoxia score constructed based on the hypoxia-related genes would be beneficial for the clinical selection of patients sensitive to immune checkpoint inhibitors.
Cell mutations can be divided into germline mutations and somatic mutations. The former are found in all cells in the body, due to originally occurring in the sperm or egg of the individual’s parent. In contrast, somatic mutations are acquired mutations limited to a certain part of the body, which can occur due to environmental factors [28]. Non-synonymous somatic mutations of tumors may produce immune-stimulating neoantigens [29, 30]. Therefore, we hypothesize that hypoxia can elicit an immune response by inducing mutations in specific genes that produce neoantigens. To explore the genes that were differentially mutated between the groups with high and low hypoxia scores, the “maftools” package in R software was used. The results revealed that the mutation frequencies of BRCA1, PRUNE2, APOB, RELN, MDN1, ZSWIM8, MGAT4C, FAT2, and VWF were higher in the group with a high hypoxia score. From a search of the literature, no relevant research on hypoxia-regulating gene mutations of OC has been reported. Our results may provide a reference for future research in this field. Nevertheless, the potential immunogenicity of BRCA1 mutation has been demonstrated, so it may act as a biomarker for ICI therapies of patients with high grade serous ovarian cancer [31]. This provides some support for the reliability of our study.