Comprehensive Landscape of Ovarian Cancer Immune Microenvironment Based on Integrated Multi-Omics Analysis

Abstract

Epithelial ovarian cancer has a low response rate to immunotherapy and a complex immune microenvironment that regulates its treatment outcomes. Understanding the immune microenvironment and its molecular basis is of great clinical significance in the effort to improve immunotherapy response and outcomes. To determine the characteristics of the immune microenvironment in ovarian cancer, we stratified ovarian cancer patients into three immune subtypes (C1, C2, and C3) using immune-related genes based on gene expression data from The Cancer Genome Atlas and found that these three subtypes had significant differences in immune characteristics and prognosis. Methylation and copy number variant analysis showed that the immune checkpoint genes that influenced immune response were significantly hypermethylated and highly deleted in the immunosuppressive C3 subtype, suggesting that epigenetic therapy may be able to reverse the efficacy of immunotherapy. In addition, the mutation frequencies of BRCA2 and CDK12 were significantly higher in the C2 subtype than in the other two subtypes, suggesting that mutation of DNA repair-related genes significantly affects the prognosis of ovarian cancer patients. Our study further elucidated the molecular characteristics of the immune microenvironment of ovarian cancer, which providing an effective hierarchical method for the immunotherapy of ovarian cancer patients, and has clinical relevance to the design of new immunotherapies and a reasonable combination strategies.

Keywords: immune classification; immune microenvironment; immunotherapy; multi-omics; ovarian cancer.

Figure 1

The workflow of study design.

Figure 2

OC immunotyping based on IRGs. (A) Gene expression heatmap of OC immune subtypes. (B) Principal component analysis of the differences between immune subtypes. (C) The expression difference distribution box plot of the G1 gene set between subtypes was analyzed. (D) The expression difference distribution box plot of the G2 gene set between subtypes was analyzed. (E) The K-M curve of overall survival rate of immune subtypes in TCGA dataset was tested by log rank test. ****P < 0.0001.

Figure 3

Selection of representative genes. (A) The expression heatmap of representative genes in two gene sets. (B) The expression difference distribution box plot of the gSig1 and gSig2 gene sets between subtypes was analyzed. (C) The prognosis relationship between the two gene sets and OS. (D) Relationship between two gene sets and OS in patients with ovarian cancer. (E) Relationship between two gene sets and PFS in patients with ovarian cancer. **P < 0.01, and ****P < 0.0001.

Figure 4

Functional analysis of immune gene module. (A) Wayne map of the difference genes in three subtypes. (B) GO pathway enriched by C1 subtype specific genes. (C) GO pathway enriched by C2 subtype specific genes. (D) GO pathway enriched by C3 subtype specific genes. The thickness of the line indicates the number of shared genes; the color indicates FDR, and the node size indicates the number of enriched genes. (E) The upper panel represents the expression profile clustering heat map of EMT marker genes, and the lower panel represents the expression profile clustering heat map of ESC marker genes. (F) Box plot of expression and distribution of EMT marker gene in three subtypes. (G) The expression distribution of ESC marker gene in three subtypes was different by box line diagram. **P < 0.01, ****P < 0.0001, and ns, ‘No Significant’.

Figure 5

Analysis of immune characteristics of immune subtypes. (A) The box plot of immune score difference in each subtype was calculated by ESTIMATE. (B) The box plot of 10 immune components difference in each subtype was calculated by MCPcounter. (C) The expression differences of 8 common immune checkpoint genes among three subtypes. (D) The difference of 6 kinds of immune cell scores in each subtype. (E) The differences of 13 kinds of immune components in different subtypes. (F) Differences of enrichment scores of four immune pathways among three subtypes. (G) Differential expression of IFN-γ related regulatory genes in three immune subtypes. **P < 0.01, ****P < 0.0001, and ns, ‘No Significant’.

Figure 6

OC immunotyping based on IRGs in validation set. (A) Gene expression heatmap of OC immune subtypes. (B) The K-M curve of the overall survival rate of the immune subtypes in the validation dataset, measured by log rank test. (C) 10 immune components difference in each subtype. (D) The differences of 13 kinds of immune components in different subtypes. (E) The box plot of immune score difference in each subtype was calculated by ESTIMATE. (F) The expression differences of 6 common immune checkpoint genes among three subtypes. *P < 0.05, **P < 0.01, ****P < 0.0001, and ns ‘No Significant’.

Figure 7

Comparison of immune subtypes and existing subtypes. (A) The intersection of molecular subtypes and immune subtypes in TCGA. (B) Clustering heatmap of prognosis related immune genes in four subtypes. (C) Differential distribution of G1 gene set expression among immune subtypes. (D) Differential distribution of G1 gene set expression among immune subtypes. ****P < 0.0001.

Figure 8

Methylation analysis of immune subtypes. (A) Expression differences of methylation related genes in three immune subtypes. (B) Heatmap of genes with significant differences in methylation and transcriptional expression among subtypes. (C) GO enrichment analysis of hypermethylated genes in C3 subtype. (D) KEGG pathway analysis of hypomethylated genes in C3 subtype. (E) Correlation trend of methylation level and transcription expression of immune related genes.

Figure 9

(A) After A2780 cell lines were treated with 5-AZ at the concentrations of 10μM for 24 hours, the changes of gene expression levels were observed. (B) After HEY cell lines were treated with 5-AZ at the concentrations of 10μM for 24 hours, the changes of gene expression levels were observed. (C) K-M survival curve of differentially methylated genes associated with prognosis. (D) Relationship between prognosis related differential methylation genes and immunotherapy responsiveness. *P < 0.05, **P < 0.01, ****P < 0.0001, and ns, ‘No Significant’.

Figure 10

Genomic variation analysis of immune subtypes. (A) Differential high frequency mutation genes in C2 subtypes. (B) Correlation heatmap of co-mutated genes in C2 subtype. (C) Correlation of co-mutated genes in C2 subtypes. (D) Survival and prognosis of patients with co-mutated genes. (E) Significantly increased copy number genes in C2 subtype. (F) Significantly decreased copy number genes in C3 subtype. (G) Survival curve of differential copy number genes associated with prognosis.