Therapeutic Advances in Medical Oncology (Oct 2024)

Identification of a central network hub of key prognostic genes based on correlation between transcriptomics and survival in patients with metastatic solid tumors

  • Vladimir Lazar,
  • Eric Raymond,
  • Shai Magidi,
  • Catherine Bresson,
  • Fanny Wunder,
  • Ioana Berindan-Neagoe,
  • Annemilaï Tijeras-Rabaland,
  • Jacques Raynaud,
  • Amir Onn,
  • Michel Ducreux,
  • Gerald Batist,
  • Ulrik Lassen,
  • Fin Cilius Nielsen,
  • Richard L. Schilsky,
  • Eitan Rubin,
  • Razelle Kurzrock

DOI
https://doi.org/10.1177/17588359241289200
Journal volume & issue
Vol. 16

Abstract

Read online

Background: Dysregulated pathways in cancer may be hub addicted. Identifying these dysregulated networks for targeting might lead to novel therapeutic options. Objective: Considering the hypothesis that central hubs are associated with increased lethality, identifying key hub targets within central networks could lead to the development of novel drugs with improved efficacy in advanced metastatic solid tumors. Design: Exploring transcriptomic data (22,000 gene products) from the WINTHER trial ( N = 101 patients with various metastatic cancers), in which both tumor and normal organ-matched tissue were available. Methods: A retrospective in silico analysis of all genes in the transcriptome was conducted to identify genes different in expression between tumor and normal tissues (paired t -test) and to determine their association with survival outcomes using survival analysis (Cox proportional hazard regression algorithm). Based on the biological relevance of the identified genes, hub targets of interest within central networks were then pinpointed. Patients were grouped based on the expression level of these genes ( K -mean clustering), and the association of these groups with survival was examined (Cox proportional hazard regression algorithm, Forest plot, and Kaplan–Meier plot). Results: We identified four key central hub genes— PLOD3, ARHGAP11A, RNF216 , and CDCA8 , for which high expression in tumor tissue compared to analogous normal tissue had the most significant correlation with worse outcomes. The correlation was independent of tumor or treatment type. The combination of the four genes showed the highest significance and correlation with the poorer outcome: overall survival (hazard ratio (95% confidence interval (CI)) = 10.5 (3.43–31.9) p = 9.12E−07 log-rank test in a Cox proportional hazard regression model). Findings were validated in independent cohorts. Conclusion: The expression of PLOD3, ARHGAP11A, RNF216 , and CDCA8 constitute, when combined, a prognostic tool, agnostic of tumor type and previous treatments. These genes represent potential targets for intercepting central hub networks in various cancers, offering avenues for novel therapeutic interventions.