Identification Of Glycolysis-related Subtypes And Establishment Of Prognostic Model Of Head And Neck Squamous Cell Carcinoma

Objective:Head and neck squamous cell carcinoma is one of the malignant tumors with the worst prognosis.Glycolysis,as an emerging hallmark of cancer,is related to tumor progression,prognosis and treatment response.At present,there is no consensus on the relationship between glycolysis and HNSCC,so this study aims to explore the role and clinical significance of glycolysis-related genes in HBSCC based on bioinformatics.Methods : Using HNSCC data of TCGA and GSE65858 from GEO,independent prognostic genes related to glycolysis were obtained through gene enrichment set analysis,differential analysis,univariate and multivariate Cox analysis.Then consensus clustering analysis of glycolysis and prognostic differential genes were performed on the data set respectively.Principal component analysis,survival analysis,gene set variation analysis,immune cell difference analysis and other methods were used to compare the differences among different subtypes.In order to verify the effectiveness of the follow-up model,all patients were randomly and evenly divided into train group and test group.Lasso regression and multivariate regression were used to construct the prognostic risk model for the data of the train group.The risk score of each sample was calculated and the median of the train group was taken as the cut-off value.The samples of the train group and the test group were divided into high-and low-risk groups.Time-dependent receiver operating characteristic curves for 1-,3-,and 5-year survivals were used to assess the risk model,and the survival status and risk curves of high-and low-risk groups were analyzed and visualized.Then,Sankey diagram was drawn to show the construction process of the prognosis model and nomogram was established using clinical data.Finally,Western blot test was used to verify whether the key genes PYGL and STC1 were abnormally expressed in HNSCC clinical samples.Results:This study analyzed 768 HNSCC patients from TCGA and GEO databases.Two gene sets,HALLMARK＿GLYCOLYSIS and REACTOME＿GLYCOLYSIS,were significantly enriched in HNSCC through glycolysis-related gene set enrichment analysis in TCGA data.Then we obtained five glycolytic genes(EXT2,PGK1,SDHC,PYGL and STC2)by difference analysis,univariate and multivariate Cox analysis.These five genes were abnormally expressed in HNSCC and had copy number variation.Then the expression values of five genes in the merged database were obtained,and two glycolysis subtypes(A and B)were identified by consensus clustering.Subtype A had poor prognosis,high expression of five glycolysis-related genes and low immune cell content.Subtype B was on the contrary.There were significant differences between subtype A and B,with a total of 530 differential genes and 286 prognostic genes obtained by univariate regression.Then 2 prognostic differential subtypes(1 and 2)were obtained after re-clustering,subtype 1 had poor prognosis,high expression of 5 glycolysis-related genes while subtype 2 was the opposite.Risk score=(-0.26 ×expression of HLF)+(-0.18×expression of IL-34)+(-0.13 ×expression of MS4A1)+(-0.14 ×expression of SPINK6)+(0.15 ×expression of STC1).The prognostic risk model divided the patients into high-and low-risk groups.The train group,test group and total queue showed that the prognosis of the high-risk group was poor and the expression of five glycolysis-related genes was high while the low-risk group was the opposite.The survival rate of 1-,3-,and 5-years could be predicted according to different clinical characteristics of patients through the nomogram,which has a good survival predictive ability.Finally,we detected the expression of two key genes in HNSCC samples.The results of western blot and semi-quantitative analysis of 6 pairs of samples showed that compared with the control group,the expression of PYGL protein and STC1 protein were significantly up-regulated in cancer samples.Conclusion:1.In this study,two different molecular subtypes of HNSCC were identified based on glycolysis-related genes,and a risk model composed of HLF,IL34,MS4A1,SPINK6 and STC1 was constructed through screening and validation.2.This risk model can well distinguish between high-and low-risk group and predict the prognosis of HNSCC patients.3.Western Bolt confirmed that PYGL protein and STC1 protein were highly expressed in HNSCC tumor samples,which might provide potential targets and biomarkers for the treatment of HNSCC.