Abstract
Glioma is the most common and malignant cancer of the central nervous system, and the prognosis is poor. Metabolic reprogramming is a common phenomenon that plays an important role in tumor progression including gliomas. Searching the representative process among numerous metabolic processes to evaluate the prognosis aside from the glycolytic pathway may be of great significance. A novel prediction signature was constructed in the present study based on gene expression. A total of 1027 glioma samples with clinical and RNA-seq data were used in the present study. Lasso-Cox, gene set variation analysis, Kaplan-Meier survival curve analysis, Cox regression, receiver operating characteristic curve, and elastic net were performed for constructing and verifying predictive models. The R programming language was used as the main tool for statistical analysis and graphical work. This signature was found to be stable in prognostic prediction in the Chinese Glioma Genome Atlas Network and the Cancer Genome Atlas databases. The possible mechanism was also explored, revealing that the aforementioned signature was closely related to DNA replication and ATP binding. In summary, a prognosis prediction signature for patients with glioma based on five genes was constructed and showed great potential for clinical application.