关键词: 头颈部鳞状细胞癌, 乳酸代谢, 免疫浸润, 生物信息学, LASSO回归分析

Abstract:

Objective To select the differential prognostic lactic acid metabolism-related genes （LRGs） of the head and neck squamous cell carcinoma （HNSCC） to construct the LRGs prognostic model of HNSCC， and to clarify the potential mechanism. Methods The HNSCC gene expression and clinical data were obtained from The Cancer Genome Atlas （TCGA） and Gene Expression Omnibus （GEO） Databases， the LRGs were identified through GeneCards Database， and R software was used to screen out the LRGs of HNSCC； univariate Cox regression analysis was used to identify prognosis-related genes； two different subtypes were identified based on the prognostis-related LRGs； Kaplan-Meier （K-M） curve analysis was used to compare the prognosis of the patients between two groups； CIBERSORT algorithm was used to perform the immuno-correlation analysis between two groups；multivariate Cox regression analysis and LASSO regression analysis were used to construct the prognostic model； receiver operating characteristic curve （ROC） and K-M survival curve were used to assess the relationship between LRGs and survival and prognosis of the HNSCC patients. The prognostic model was validated by GSE27020， GSE41613，and GSE65858 datasets.The experiment were grouped based on risk score，and immune-related analysis and tumor score analysis were performed. Results The TCGA Database differential analysis results showed that 1 196 LRGs were identified from HNSCC samples； univariate Cox regression analysis selected 27 differentially expressed genes （DEGs） associated with the prognosis of the HNSCC patients. Two different LRGs subtypes （Group 1 and Group 2） were identified according to the prognosis-related genes. The K-M survival curves results showed that the overall survival （OS） of the patients in Group 2 was significantly higher than that in Group 1， and the immune cell expression amount of the patients in Group 2 was also higher than that in group 1. The multivariate Cox regression and LASSO regression analysis results screened out 9 LRGs， including hypoxanthine phosphoribosyltransferase 1 （HPRT1）， amyloid precursor protein （APP）， glycogen phosphorylase L（PYGL），urokinase-type plasminogen activator（PLAU）， cannabinoid receptor 2 （CNR2）， stanniocalcin 2 （STC2）， nucleotide binding oligomerization domain-like receptor protein 1 （NLRP1）， integrin-linked kinase （ILK）， and forkhead box B1 （FOXB1）；the prognostic model was constructed.The K-M and ROC curve results indicated that the expression levels of above 9 genes were associated with the survival and prognosis of the HNSCC patients， providing good 1-year， 2-year， and 3-year survival prediction effect， and the area under ROC curve （AUC） values were all greater than 0.650. Furthermore， the predictive ability of the prognosis model was validated in GSE27020， GSE41613， and GSE65858 datasets. The patients classified based on the risk scores had distinguishable immune statuses. Conclusion The differentially expressed LRGs of HNSCC screened by bioinformatics methods are related to the survival and prognosis of the HNSCC patients； the prognostic model constructed by 9 LRGs can predict the survival status and treatment response of the HNSCC patients.

Key words: Head and neck squamous carcinoma, Lactic acid metabolism, Immune infiltration, Bioinformatics, LASSO regression analysis