Construction of mitochondrial signature (MS) for the prognosis of ovarian cancer

Abstract

Background: Ovarian cancer (OV) continues to be the most lethal type of gynecological cancer with a poor prognosis. During tumorigenesis and cancer advancement, mitochondria are key players in energy metabolism. This study focuses on exploring the mitochondria-related genes for the prognosis of OV.

Methods: RNA expression profiles and single-cell data were acquired from The Cancer Genome Atlas (TCGA), International Cancer Genome Consortium (ICGC), and Gene Expression Omnibus databases for screening and validating mitochondria-related differentially expressed genes (DEGs). After univariate Cox analysis, prognostic genes were carried out for modeling mitochondria signature (MS) based on 101 combinations of 10 machine learning algorithms. Functional enrichment analysis was performed on this prognostic gene set. Immune infiltration analysis was performed between MS groups. Validation for the prognostic model gene OAT was performed to identify the prognostic significance, combined with in vitro experiments to explore its expressions in OV cells. qRT-PCR assay was performed to examine the expression of OAT in human ovarian cancer cell samples and normal ovarian epithelial cells.

Results: A total of 21 prognostic mitochondria-related DEGs were identified for reliably constructing the model MS with excellent prognostic performance in OV. GO and KEGG analysis confirmed these genes were enriched in the generation of precursor metabolites and energy. It illustrated more lymphocyte infiltration in the high MS group than low MS group. OAT served as a novel biomarker for OV patients, showing poor survival in OV patients with high expression of OAT. qPCR assays confirmed its significantly high expression in human ovary cancer cell lines.

Conclusions: The MS offers tailored risk evaluations and immunotherapy treatments for each OV patient. MS model gene OAT has been recognized as a new oncogene for OV linked to immune escape.

Keywords: Biomarker; Immunotherapy; Ovarian cancer; Tumor microenvironment.

Fig. 1

Identification of mitochondria-related prognostic DEGs for OV. A Heat map of mitochondria-related DEGs. B. Forest map for 21 significant prognostic genes by univariate Cox analysis. C-D GO and KEGG enrichment histogram. E Circos map of the distribution of mitochondria-related prognostic genes in the human chromosome. F PCA plots of each OV dataset for batch effect removal. G. CNV plot of prognostic genes within chromosomal locations.

Fig. 2

Construction and validation of MS model. A Heat map of C index of 101 combination algorithms in nine validation datasets. B-K Survival analysis results of training and validation datasets of OV. L-P. Survival analysis results of pan-cancer immunotherapy datasets. Q-T Violin chart of IPS differences between MS groups showing the sensitivity to immunotherapy.

Fig. 3

Prognostic performance of MS model. A Bar chart comparing C-index of age, Stage and Grade in each dataset. B-C PCA plots and timeROC results of each OV dataset. D Comparison results of C-index values between MS model and other models reported in 32 literatures in each OV dataset.

Fig. 4

Immune infiltration results. A-F CIBERSORT and TIDE algorithms predicted the difference of TME in high and low MS groups. G Heat map of expression of immune regulatory genes in high and low MS groups. H Radar map of differences of immune-related enrichment pathways in high and low MS groups. I. Heat maps of immune infiltration differences evaluated by seven software between groups. J. Identification of pathological sections of high and low MS group samples from TCGA database.

Fig. 5

MS distribution and cellular interactions based on single cell data. A-C Single-cell analysis result of GSE235931. D-F Single-cell analysis result of GSE184880. G-H Violin plot of MS scores of each cell subtype by the GSE235931 and GSE184880 datasets. I-M Results of cell-cell interactions and communication differences between high and low MS groups

Fig. 6

Validation of OAT gene. A Correlation of MS score and OAT. B Survival analysis of patients with high or low OAT expression. C-D Immunohistochemical results of OAT in tumor and normal ovary tissues from HPA database. E Heat map of correlation between OAT and immune-related genes. F The mRNA expression of OAT in OV cells. ***p < 0.001