Cancer Medicine (Jan 2020)

Genome‐wide screening of abberant methylated drivers combined with relative risk loci in bladder cancer

  • Chuanjie Zhang,
  • Kangjie Shen,
  • Yuxiao Zheng,
  • Feng Qi,
  • Jun Luo

DOI
https://doi.org/10.1002/cam4.2665
Journal volume & issue
Vol. 9, no. 2
pp. 768 – 782

Abstract

Read online

Abstract Background To explore important methylation‐driven genes (MDGs) and risk loci to construct risk model for prognosis of bladder cancer (BCa). Methods We utilized TCGA‐Assembler package to download 450K methylation data and corresponding transcriptome profiles. MethylMix package was used for identifying methylation‐driven genes and functional analysis was mainly performed based on ConsensusPathDB database. Then, Cox regression method was utilized to find prognostic MDGs, and we selected 17 hub genes via stepwise regression and multivariate Cox models. Kruskal‐Wallis test was implemented for comparisons between risk with other clinical variables. Moreover, we constructed the risk model and validated it in GSE13507. Gene set enrichment analysis was performed using the levels of risk score as the phenotype. Additionally, we further screened out the relative methylation sites associated with the 17 hub genes. Cox regression and Survival analysis were conducted to find the specifically prognostic sites. Results Two hundred and twenty‐eight MDGs were chosen by ConsensusPathDB database. Results revealed that most conspicuous pathways were transcriptional mis‐regulation pathways in cancer and EMT. After Cox regression analysis, 17 hub epigenetic MDGs were identified. We calculated the risk score and found satisfactory predictive efficiency by ROC curve (AUC = 0.762). In the validation group from GSE13507, 17 hub genes remained higher predictive value with AUC = 0.723 and patients in high‐risk group. Meanwhile, Kruskal‐Wallis test revealed that higher risk score correlated with a higher level of TNM stage, tumor grade, and advanced pathological stages. Then, identified 38 risk methylated loci that highly associated with prognosis. Last, gene set enrichment analysis revealed that high‐risk level of MDGs may correlate with several important pathways, including MAPK signaling pathway and so on. Conclusion Our study indicated several hub‐MDGs, calculated novel risk score and explored the prognostic value in BCa, which provided a promising approach to BCA prognosis assessment.

Keywords