Medical Student Cooper Medical School Montville, New Jersey, United States
Introduction: Glioblastoma (GBM) remains the most lethal brain cancer with a five-year mortality rate of up to 80%. Genes with established implications in GBM pathogenesis include TP53, RB1, RTK, and RAS, among others. Many of these genes regulate nonspecific pathways promoting the succession of a wide berth of distinct cancers. The utility of weighted gene-co expression network analysis has been recently highlighted for its ability to resolve important regulatory networks in several cancers to date.
Methods: Level 3 bulk RNAseq gene expression data was obtained from The Cancer Genome Archive (TCGA) Firehose Legacy study (n=160). Gene and sample quality control included iterative filtering of genes with < 5 non-missing samples and hierarchical clustering to detect sample outliers. The final, filtered gene expression matrix consisted of 154 samples by 19051 genes. Gene co-expression networks were identified using R package ‘WGCNA’. Modules were resolved using unsigned topological overlap measure and a minimum allowed size of 20 genes. A soft-threshold power of 5 was selected, based on observed high similarity to the scale-free network and high mean connectivity. Hub genes were defined as the most connected gene in each module. Kaplan-Meier (KM) analysis and related statistical information were realized using R package ‘survminer’. For each hub gene, patients with above-median gene expression were compared to patients with below-median gene expression. Nominal log-rank P-values < 0.05 were deemed statistically significant.
Results: A total of 56 unique hub genes were identified. Low expression of Fas binding factor 1 (FBF1) correlated with greater overall survival (OS) probability (p=0.036). Low expression of fibronectin type III domain containing 3B (FNDC3B) correlated with greater OS probability (p=0.03). High expression of ATP binding cassette subfamily G member 2 (ABCG2) correlated with greater OS probability (p=0.02).
Conclusion : Gene co-expression analysis revealed three hub genes with prognostic value for glioblastoma.
