Construction and validation of a metabolic-related genes prognostic model for oral squamous cell carcinoma based on bioinformatics

Abstract

Background Oral squamous cell carcinoma (OSCC) accounts for a frequently-occurring head and neck cancer, which is characterized by high rates of morbidity and mortality. Metabolism-related genes (MRGs) show close association with OSCC development, metastasis and progression, so we constructed an MRGs-based OSCC prognosis model for evaluating OSCC prognostic outcome. Methods This work obtained gene expression profile as well as the relevant clinical information from the The Cancer Genome Atlas (TCGA) database, determined the MRGs related to OSCC by difference analysis, screened the prognosis-related MRGs by performing univariate Cox analysis, and used such identified MRGs for constructing the OSCC prognosis prediction model through Lasso-Cox regression. Besides, we validated the model with the GSE41613 dataset based on Gene Expression Omnibus (GEO) database. Results The present work screened 317 differentially expressed MRGs from the database, identified 12 OSCC prognostic MRGs through univariate Cox regression, and then established a clinical prognostic model composed of 11 MRGs by Lasso-Cox analysis. Based on the optimal risk score threshold, cases were classified as low- or high-risk group. As suggested by Kaplan–Meier (KM) analysis, survival rate was obviously different between the two groups in the TCGA training set (P < 0.001). According to subsequent univariate and multivariate Cox regression, risk score served as the factor to predict prognosis relative to additional clinical features (P < 0.001). Besides, area under ROC curve (AUC) values for patient survival at 1, 3 and 5 years were determined as 0.63, 0.70, and 0.76, separately, indicating that the prognostic model has good predictive accuracy. Then, we validated this clinical prognostic model using GSE41613. To enhance our model prediction accuracy, age, gender, risk score together with TNM stage were incorporated in a nomogram. As indicated by results of ROC curve and calibration curve analyses, the as-constructed nomogram had enhanced prediction accuracy compared with clinicopathological features alone, besides, combining clinicopathological characteristics with risk score contributed to predicting patient prognosis and guiding clinical decision-making. Conclusion In this study, 11 MRGs prognostic models based on TCGA database showed superior predictive performance and had a certain clinical application prospect in guiding individualized.

Leveraging technology-driven strategies to untangle omics big data: circumventing roadblocks in clinical facets of oral cancer

“…Numerous databases, such as TCGA, GEPIA, GEO, KEGG, and DAVID among others, have made extensive gene expression data and associated clinical data available to researchers. Decoding these databases using various bioinformatics strategies can uncover pathways to arrive at treatment plans, develop new prognostic models with high precision from the standpoint of tumor cell biological behavior ( 62 ).…”

“…Zhang et al., 2020 ( 62 ) developed a model based on Metabolism-Related Genes (MRGs) to evaluate the prognostic outcome of oral cancer.…”

Transcriptome and Mendelian randomization were combined to screen and validate prognostic genes associated with lipid autophagy in oral squamous cell carcinoma

“…However, the main limitations of OSCC-related TNM system is its main focus on the anatomical extension of the disease. The prognosis can be modi ed by tumor-related factors, such as genetics, patient age, sex, race or comorbidities [4][5] .Consequently, it is urgently needed to construct the creditable prognostic approaches that can assist clinicians in selecting the suitable individualized therapeutic strategies, thus improving OSCC prognosis.…”