Background: Recurrence is still a major obstacle to the successful treatment of
gliomas. Understanding the underlying mechanisms of recurrence may help for developing new drugs to combat
gliomas recurrence. This study provides a strategy to discover new drugs for recurrent
gliomas based on drug perturbation induced gene expression changes. Methods: The
RNA-seq data of 511 low grade
gliomas primary
tumor samples (LGG-P), 18 low grade
gliomas recurrent
tumor samples (LGG-R), 155
glioblastoma multiforme primary
tumor samples (GBM-P), and 13
glioblastoma multiforme recurrent
tumor samples (GBM-R) were downloaded from TCGA database. DESeq2, key driver analysis and weighted gene correlation network analysis (WGCNA) were conducted to identify differentially expressed genes (DEGs), key driver genes and coexpression networks between LGG-P vs LGG-R, GBM-P vs GBM-R pairs. Then, the CREEDS database was used to find potential drugs that could reverse the DEGs and key drivers. Results: We identified 75 upregulated and 130 downregulated genes between LGG-P and LGG-R samples, which were mainly enriched in human papillomavirus (HPV)
infection, PI3K-Akt signaling pathway, Wnt signaling pathway, and ECM-receptor interaction. A total of 262 key driver genes were obtained with frizzled class receptor 8 (FZD8),
guanine nucleotide-
binding protein subunit gamma-12 (GNG12), and
G protein subunit β2 (GNB2) as the top hub genes. By screening the CREEDS database, we got 4 drugs (
Paclitaxel, 6-benzyladenine,
Erlotinib,
Cidofovir) that could downregulate the expression of up-regulated genes and 5 drugs (
Fenofibrate,
Oxaliplatin,
Bilirubin, Nutlins,
Valproic acid) that could upregulate the expression of down-regulated genes. These drugs may have a potential in combating recurrence of
gliomas. Conclusion: We proposed a time-saving strategy based on drug perturbation induced gene expression changes to find new drugs that may have a potential to treat recurrent
gliomas.