A Hypoxia Signature for Predicting Prognosis and Tumor Immune Microenvironment in Adrenocortical Carcinoma

Abstract

Adrenocortical carcinoma (ACC) is a rare malignancy with dismal prognosis. Hypoxia is one of characteristics of cancer leading to tumor progression. For ACC, however, no reliable prognostic signature on the basis of hypoxia genes has been built. Our study aimed to develop a hypoxia-associated gene signature in ACC. Data of ACC patients were obtained from TCGA and GEO databases. The genes included in hypoxia risk signature were identified using the Cox regression analysis as well as LASSO regression analysis. GSEA was applied to discover the enriched gene sets. To detect a possible connection between the gene signature and immune cells, the CIBERSORT technique was applied. In ACC, the hypoxia signature including three genes (CCNA2, COL5A1, and EFNA3) was built to predict prognosis and reflect the immune microenvironment. Patients with high-risk scores tended to have a poor prognosis. According to the multivariate regression analysis, the hypoxia signature could be served as an independent indicator in ACC patients. GSEA demonstrated that gene sets linked to cancer proliferation and cell cycle were differentially enriched in high-risk classes. Additionally, we found that PDL1 and CTLA4 expression were significantly lower in the high-risk group than in the low-risk group, and resting NK cells displayed a significant increase in the high-risk group. In summary, the hypoxia risk signature created in our study might predict prognosis and evaluate the tumor immune microenvironment for ACC.

Figure 1

Establishment of hypoxia risk signature in ACC patients. (a) Prognostic genes identified by univariate Cox regression. (b) LASSO regression algorithm. (c) The hypoxia risk signature developed by multivariate Cox regression. (d)-(e) Comparison of Spearman's correlation coefficient among 3 hypoxia-associated genes.

Figure 2

Application value of the hypoxia-related signature in predicting the outcomes of ACC patients. (a) Heatmap of expressional profiles of 3 hypoxia-associated genes in the high/low-risk group in two cohorts. (b)-(c) the risk score and OS in patients in the high/low-risk group in two cohorts. (d) The death rate in the high/low-risk group. (e) Kaplan–Meier survival analysis of patients in the high/low-risk group.

Figure 3

Correlation between hypoxia-associated gene and stages of ACC. (a) Heatmaps of expressional profiles of 3 hypoxia-associated genes at different stages from TCGA cohort. (b) The expressional levels of hypoxia-associated genes in ACC patients at different stages.

Figure 4

Effect of the hypoxia risk signature in predicting the outcomes of ACC patients. (a)-(b) Evaluation of the independent prognostic effect of the hypoxia-related signature in TCGA cohort by univariate and multivariate Cox regression analyses. (c)-(d) ROC curves for assessing the effect of the hypoxia risk signature in predicting the outcomes of ACC patients in two cohorts.

Figure 5

Development of a predictive nomogram for ACC patients in TCGA cohort. (a) Nomogram for predicting prognosis of ACC patients in TCGA cohort. (b)–(d) Calibration plots for predicting probabilities of the nomogram at the 1, 3, and 5 years.

Figure 6

Hypoxia-associated signaling pathways screened by GSEA. (a) Gene sets enriched in the high-risk group performed by GSEA in TCGA cohort. (b) Gene sets enriched in the high-risk group performed by GSEA in GEO cohort.

Figure 7

Correlation of hypoxia risk signature with immunity microenvironment. (a) Presence of immune cellular infiltration in the high/low-risk group. (b) Heatmap of the genes for negative regulation of the cancer-immunity cycle in the high/low-risk group in TCGA cohort. (c)-(d) Percentages of resting and activated NK cells in the high/low-risk group. (e)-(f) PDL1 and CTLA4 expressional levels in the high/low-risk group.