Identification of seven hypoxia-related genes signature and risk score models for predicting prognosis for ovarian cancer

Abstract

Ovarian cancer (OC) is the most common malignant cancer in the female reproductive system. Hypoxia is an important part of tumor immune microenvironment (TIME), which is closely related to cancer progression and could significantly affect cancer metastasis and prognosis. However, the relationship between hypoxia and OC remained unclear. OCs were molecularly subtyped by consensus clustering analysis based on the expression characteristics of hypoxia-related genes. Kaplan-Meier (KM) survival was used to determine survival characteristics across subtypes. Immune infiltration analysis was performed by using Estimation of Stromal and Immune cells in Malignant Tumors using Expression data (ESTIMATE) and microenvironment cell populations-counter (MCP-Counter). Differential expression analysis was performed by using limma package. Next, univariate Cox and least absolute shrinkage and selection operator (LASSO) regression analyses were used to build a hypoxia-related risk score model (HYRS). Mutational analysis was applied to determine genomic variation across the HYRS groups. The Tumor Immune Dysfunction and Exclusion (TIDE) algorithm was used to compare the effectiveness of HYRS in immunotherapy prediction. We divided OC samples into two molecular subtypes (C1 and C2 subtypes) based on the expression signature of hypoxia genes. Compared with C1 subtype, there was a larger proportion of poor prognosis genotypes in the C2 subtype. And most immune cells scored higher in the C2 subtype. Next, we obtained a HYRS based on 7 genes. High HYRS group had a higher gene mutation rate, such as TP53. Moreover, HYRS performed better than TIDE in predicting immunotherapy effect. Combined with clinicopathological features, the nomogram showed that HYRS had the greatest impact on survival prediction and a strong robustness.

Keywords: Biomarkers; Hypoxia; Ovarian cancer; Prognosis; TIME.

Fig. 1

Identification of hypoxia-related ovarian cancer subtypes. A The cumulative distribution function (CDF) curve in TCGA cohort. B The CDF Delta area curve in TCGA cohort. C Heat map of sample clusters when consensus k = 2 in the TCGA cohort. D C1 had longer overall survival (OS) than that in C2 in the TCGA cohort. E C1 had a better survival outcome in the GSE cohort. F Differences in hypoxia scores among subtypes in the TCGA cohort. G Differences in hypoxia scores among subtypes in the GSE cohort. H Molecular subtype comparison information. I Survival curves of reported molecular subtypes. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Fig. 2

Immune signatures in different molecular subtypes. A Differences in immune infiltration among molecular subtypes in the TCGA cohort. B Score differences of 10 immune cells among molecular subtypes were analyzed by MCP-Counter. C ssGSEA analysis of 28 immune cells scores among molecular subtypes. D Differences in immune checkpoint gene expression between C1 and C2 subtypes in the TCGA cohort. E Differences in TIDE scores between C1 and C2 subtypes in the TCGA cohort. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Fig. 3

Identification of differentially expressed genes between C1 and C2 subtypes. A Volcano plot showed differential expressed genes (DEGs) between C1 and C2 subtypes in the TCGA cohort. B Volcano plot showed DEGs between C1 and C2 subtypes in the GSE cohort. C Venn diagram of DEGs of the intersection of TCGA and GSE26712. D KEGG pathways of up-regulated DEGs. E KEGG pathways of down-regulated DEGs

Fig. 4

Establishment of a hypoxia-related risk scoring model. A Volcano plot showing candidate signature genes associated with OC prognosis. B Trajectories of candidate genes as lambda changes. C Confidence interval under lambda. D Distribution of LASSO coefficients of the hypoxia-related gene signature. E Receiver operating characteristic (ROC) curve of HYRS in TCGA cohort. F Survival curve of the high HYRS and low HYRS groups in TCGA cohort. G ROC curve of HYRS in GSE cohort. H Survival curve of the high HYRS and low HYRS groups in GSE cohort

Fig. 5

Clinical characteristics in different HRYS groups. A Differences in HYRS between different clinicopathological groups in the TCGA cohort, following by stage, grade, age, status, and cluster. B Survival curves between the high and low HYRS groups, which were divided based on clinicopathological patients in the TCGA cohort

Fig. 6

Genomic mutations in different HYRS subgroups in the TCGA cohort. A Comparison of homologous recombination defects, aneuploidy score, fraction altered, number of segments, and nonsilent mutation rate in the high group and low group (Wilcoxon test, *p < 0.05; **p < 0.01; ***p < 0.001; and ****p < 0.0001). B Somatic mutation in the high group and low group

Fig. 7

Pathway differences in different HRYS groups. A KEGG pathway in the high vs low HYRS group in TCGA cohort. B KEGG pathway in the high vs low HYRS group in GSE cohort. C Comparative analysis of metabolic pathway differences in TCGA and GSE cohorts

Fig. 8

Efficiency of HYRS model. A Survival and ROC curves of the high HYRS and low HYRS groups in the IMvigor210 cohort. B Survival and ROC curves of the high TIDE and low TIDE groups in the IMvigor210 cohort. C ROC curve of the HRYS group and TIDE group in IMvigor210 cohort. D Survival and ROC curves of the high HYRS and low HYRS groups in the GSE91061 cohort. E Survival and ROC curves of the high TIDE and low TIDE groups in the GSE91061 cohort. F ROC curve of the HRYS group and TIDE group. D ROC curve of the HRYS group and TIDE group in GSE91061 cohort

Fig. 9

Improvements in prognostic models and survival prediction. A Univariate Cox regression analysis of HYRS and clinicopathological features. B Multivariate Cox regression analysis of HYRS and clinicopathological features. C Nomogram model combined by HYRS and age. D 1-, 3-, and 5-year calibration curves of the nomogram. E The decision curve of the nomogram