Abstract
BACKGROUND
Glioblastoma (GBM) is one of the most common malignant brain tumors in adults and is characterized by high aggressiveness and rapid progression, poor treatment, high recurrence rate, and poor prognosis. Although super-enhancer (SE)-driven genes haven been recognized as prognostic markers for several cancers, whether it can be served as effective prognostic markers for patients with GBM has not been evaluated.
METHODS
We first combined histone modification data with transcriptome data to identify SE-driven genes associated with prognosis in patients with GBM. Second, we developed a SE-driven differentially expressed genes (SEDEGs) risk score prognostic model by univariate Cox analysis, KM survival analysis, multivariate Cox analysis and least absolute shrinkage and selection operator (LASSO) regression. Its reliability in predicting was verified by two external data sets. Third, through mutation analysis, immune infiltration, we explored the molecular mechanisms of prognostic genes. Next, Genomics of Drug Sensitivity in Cancer (GDSC) and the Connectivity Map (cMap) database were employed to assess different sensitivities to chemotherapeutic agents and small-molecule drug candidates between high- and low-risk patients. Finally, SEanalysis database was chosen to identify SE-driven transcription factors (TFs) regulating prognostic markers which will reveal a potential SE-driven transcriptional regulatory network.
RESULTS
First, we developed a 11-gene risk score prognostic model (NCF2, MTHFS, DUSP6, G6PC3, HOXB2, EN2, DLEU1, LBH, ZEB1-AS1, LINC01265, and AGAP2-AS1) selected from 1,154 SEDEGs, which is not only an independent prognostic factor for patients, but also can effectively predict the survival rate of patients. The model can effectively predict 1-, 2- and 3-year survival of patients and was validated in external Chinese Glioma Genome Atlas (CGGA) and Gene Expression Omnibus (GEO) datasets. Second, the risk score was positively correlated with the infiltration of regulatory T cell, CD4 memory activated T cell, activated NK cell, neutrophil, resting mast cell, M0 macrophage, and memory B cell. Third, we found that high-risk patients showed higher sensitivity than low-risk patients to both 27 chemotherapeutic agents and 4 small-molecule drug candidates which might benefit further precision therapy for GBM patients. Finally, 13 potential SE-driven TFs imply how SE regulates GBM patient's prognosis.
CONCLUSION
The SEDEG risk model not only helps to elucidate the impact of SEs on the course of GBM, but also provides a bright future for prognosis determination and choice of treatment for GBM patients.
Collapse