Identification of intrinsic subtype-specific prognostic microRNAs in primary glioblastoma

J Exp Clin Cancer Res. 2014 Jan 19;33(1):9. doi: 10.1186/1756-9966-33-9.

Abstract

Background: Glioblastoma multiforme (GBM) is the most malignant type of glioma. Integrated classification based on mRNA expression microarrays and whole-genome methylation subdivides GBM into five subtypes: Classical, Mesenchymal, Neural, Proneural-CpG island methylator phenotype (G-CIMP) and Proneural-non G-CIMP. Biomarkers that can be used to predict prognosis in each subtype have not been systematically investigated.

Methods: In the present study, we used Cox regression and risk-score analysis to construct respective prognostic microRNA (miRNA) signatures in the five intrinsic subtypes of primary glioblastoma in The Cancer Genome Atlas (TCGA) dataset.

Results: Patients who had high-risk scores had poor overall survival compared with patients who had low-risk scores. The prognostic miRNA signature for the Mesenchymal subtype (four risky miRNAs: miR-373, miR-296, miR-191, miR-602; one protective miRNA: miR-223) was further validated in an independent cohort containing 41 samples.

Conclusion: We report novel diagnostic tools for deeper prognostic sub-stratification in GBM intrinsic subtypes based upon miRNA expression profiles and believe that such signature could lead to more individualized therapies to improve survival rates and provide a potential platform for future studies on gene treatment for GBM.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Biomarkers, Tumor / genetics
  • Biomarkers, Tumor / metabolism*
  • Cohort Studies
  • Glioblastoma / diagnosis
  • Glioblastoma / metabolism*
  • Glioblastoma / mortality
  • Humans
  • Kaplan-Meier Estimate
  • MicroRNAs / genetics
  • MicroRNAs / metabolism*
  • Molecular Diagnostic Techniques
  • Oligonucleotide Array Sequence Analysis
  • Prognosis
  • Risk
  • Transcriptome

Substances

  • Biomarkers, Tumor
  • MicroRNAs