A Novel Ferroptosis-Related Gene Signature to Predict Prognosis in Patients with Head and Neck Squamous Cell Carcinoma

Abstract

The clinical TNM staging system is currently used to evaluate the prognosis of head and neck squamous cell carcinoma (HNSCC). The 5-year survival rate for patients with HNSCC is less than 50%, which is attributed to the lack of reliable prognostic biomarkers. Ferroptosis-related genes (FRGs) regulate cancer initiation and progression. Therefore, we analyzed the correlation between FRGs and the clinical outcomes of patients with HNSCC. A typical prognostic model of FRGs for HNSCC was constructed using bioinformatics tools and data from public databases, including The Cancer Genome Atlas (TCGA), Gene Expression Omnibus (GEO), and GeneCards. The model was generated based on the following six FRGs: ATG5, PRDX6, OTUB1, FTH1, SOCS1, and MAP3K5. The accuracy of model prediction was analyzed systematically. The overall survival (OS) of the high-risk group was significantly lower than that of the low-risk group. The AUC for 1-year, 3-year, and 5-year survival were 0.645, 0.721, and 0.737, respectively, in the training set (TCGA cohort) and 0.726, 0.620, and 0.584, respectively, in the validation set (GSE65858). The multivariate Cox regression analysis revealed that the risk score was an independent prognostic factor for HNSCC. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses revealed that six FRGs were enriched in the ferroptosis pathway. A novel FRG prognostic signature model was established for HNSCC. The findings of this study reveal that FRGs are potential biomarkers for HNSCC.

Figure 1

Flow chart of the study plan. TCGA: The Cancer Genome Atlas; TICs: tumor-infiltrating immune cells; K-M: Kaplan-Meier; ROC: receiving operating characteristic; PCA: principal coordinate analysis; GO: Gene Ontology; KEGG: Kyoto Encyclopedia of Genes and Genomes; PPI: protein-protein interaction; HNSCC: head and neck squamous cell carcinoma, DCA: decision curve analysis.

Figure 2

Construction of ferroptosis-associated signature gene model. (a) Forest map of eight ferroptosis-associated genes correlated with the survival of patients with HNSCC was obtained using the univariate Cox regression analysis. (b) Forest maps of six ferroptosis-related genes (FRGs) were generated using the multivariate Cox regression analysis. (c–h) Kaplan-Meier survival curves of the high-expression and low-expression groups of six FRGs.

Figure 3

Validation of the prognostic model. (a, b) The Kaplan-Meier survival analysis of the high-risk and low-risk groups in The Cancer Genome Atlas (TCGA) and GSE65858 datasets. (c, d) Principal component analysis of the high-score and low-score groups in TCGA and GSE65858 datasets. (e, f) The area under the curve values for the 1-year, 3-year, and 5-year survival rates of TCGA and GSE65858 datasets. (g, h) The risk score distribution curve of high-risk and low-risk groups in TCGA and GSE65858 datasets. (i, j) Scatter plot indicating the association between survival time and risk score in TCGA and GSE65858 datasets.

Figure 4

The decision curve analysis (DCA) results showed a benefit in 1-year (a), 3-year (b), and 5-year (c) survival for patients in this prediction model.

Figure 5

Independent prognostic factor analysis. Results of the univariate Cox regression analysis to determine the association between overall survival rate and clinical characteristics in The Cancer Genome Atlas (TCGA) cohort (a) and GSE65858 datasets (b). The receiving operating characteristic curve was generated based on the risk score and clinical characteristics of TCGA (c) and GSE65858 (d) datasets. Results of the multivariate Cox regression analyses to determine the association between overall survival rate and clinical characteristics in TCGA (e) and GSE65858 (f) datasets.

Figure 6

Differential expression of six ferroptosis-related signature genes. (a–f) The expression levels of OTUB1, PRDX6, MAP3K5, ATG5, FTH1, and SOCS1 in the high-risk score and low-risk score groups.

Figure 7

Expression levels of six ferroptosis-related signature genes based on the age (a), P16 status (b), human papillomavirus-in situ hybridization result (c), tumor grade (d), tumor stage (e), and T-stage (f) of patients in The Cancer Genome Atlas (TCGA) cohort.

Figure 8

Biological function analysis of six ferroptosis-related signature genes. (a) Gene Ontology analysis. (b) Construction of protein-protein interaction network. (c) Kyoto Encyclopedia of Genes and Genomes analysis.

Figure 9

CIBERSORT analysis of The Cancer Genome Atlas (TCGA) cohort. (a) Histogram of relative proportions of 22 tumor immune cells (TICs). (b) Correlation among 22 tumor immune cells (TICs).

Figure 10

Association between the distribution of tumor immune cells and expression of ferroptosis-related genes (FRGs) in head neck squamous cell carcinoma. (a) ATG5 was positively associated with the proportion of CD4 resting memory T cells. (c) FTH1 was positively associated with the proportion of M0 macrophages. (b, d–f) FTH1 was negatively correlated with the proportion of resting dendritic cells, M1 macrophages, activated natural killer cells, and activated CD4 memory T cells. (g, h) MAP3K5 was positively associated with the proportion of naive B cells and resting mast cells. (i–l) SOCS1 was positively correlated with the proportion of naive B cells, activated CD4 memory T cells, and CD8 T cells but negatively correlated with the proportion of resting dendritic cells.