Prognostic gene expression signature revealed the involvement of mutational pathways in cancer genome

Background: Over the years, many efforts have been made to use the gene expression profiles of cancer types/subtypes to identify the prognostic genes with their potential clinical applications. However, one major challenge remains is to predict the common prognostic genes using simultaneously the dataset of multiple cancers, especially by considering the differences in survival, expression and the associated mutated pathways. Methods: Herein, we carried out a comprehensive examination for the prognostic genes and linked them to the mutational status of 29 cancers, so as to find independent prognostic genes and mechanisms. Additionally, their diagnostic value of them was also assessed. Results: our extensive analysis revealed: 1) the number of prognostic and diagnostic genes differs greatly across the cancers, 2) the potentially implicated 22 genes harbor the diagnostic as well as prognostic capacity, 3) the universal prognostic genes (CDC20, CDCA8, ASPM, ERCC6L, and GTSE1) were found to be involved in the spindle assembly checkpoint, 4) the prognostic genes were found to be statistically linked to the frequently mutated TP53-, MAPK-, PI3K- and AKT- related pathways. We also manually mined possible biological mechanisms for some of the statistical links in literatures. Conclusions: Taken together, we identified the prognostic genes and in addition we assessed their diagnostic capacity. Our analysis provides an important insight about the considerable overlapping between gene expression variation and the further associated altered mutational pathways across the cancer genome. We thus hypothesized that cancer related (mutated) genes are tightly connected and are capable to reshape the genome in multiple cancer types.