Biomolecules (Oct 2022)

Identification of a Fibroblast-Related Prognostic Model in Glioma Based on Bioinformatics Methods

  • Haofuzi Zhang,
  • Yutao Huang,
  • Erwan Yang,
  • Xiangyu Gao,
  • Peng Zou,
  • Jidong Sun,
  • Zhicheng Tian,
  • Mingdong Bao,
  • Dan Liao,
  • Junmiao Ge,
  • Qiuzi Yang,
  • Xin Li,
  • Zhuoyuan Zhang,
  • Peng Luo,
  • Xiaofan Jiang

DOI
https://doi.org/10.3390/biom12111598
Journal volume & issue
Vol. 12, no. 11
p. 1598

Abstract

Read online

Background: Glioma is the most common primary tumor of the central nervous system with a high lethality rate. This study aims to mine fibroblast-related genes with prognostic value and construct a corresponding prognostic model. Methods: A glioma-related TCGA (The Cancer Genome Atlas) cohort and a CGGA (Chinese Glioma Genome Atlas) cohort were incorporated into this study. Variance expression profiling was executed via the “limma” R package. The “clusterProfiler” R package was applied to perform a GO (Gene Ontology) analysis. The Kaplan–Meier (K–M) curve, LASSO regression analysis, and Cox analyses were implemented to determine the prognostic genes. A fibroblast-related risk model was created and affirmed by independent cohorts. We derived enriched pathways between the fibroblast-related high- and low-risk subgroups using gene set variation analysis (GSEA). The immune infiltration cell and the stromal cell were calculated using the microenvironment cell populations-counter (MCP-counter) method, and the immunotherapy response was assessed with the SubMap algorithm. The chemotherapy sensitivity was estimated using the “pRRophetic” R package. Results: A total of 93 differentially expressed fibroblast-related genes (DEFRGs) were uncovered in glioma. Seven prognostic genes were filtered out to create a fibroblast-related gene signature in the TCGA-glioma cohort training set. We then affirmed the fibroblast-related risk model via TCGA-glioma cohort and CGGA-glioma cohort testing sets. The Cox regression analysis proved that the fibroblast-related risk score was an independent prognostic predictor in prediction of the overall survival of glioma patients. The fibroblast-related gene signature revealed by the GSEA was applicable to the immune-relevant pathways. The MCP-counter algorithm results pointed to significant distinctions in the tumor microenvironment between fibroblast-related high- and low-risk subgroups. The SubMap analysis proved that the fibroblast-related risk score could predict the clinical sensitivity of immunotherapy. The chemotherapy sensitivity analysis indicated that low-risk patients were more sensitive to multiple chemotherapeutic drugs. Conclusion: Our study identified prognostic fibroblast-related genes and generated a novel risk signature that could evaluate the prognosis of glioma and offer a theoretical basis for clinical glioma therapy.

Keywords