Mutational signatures and their association with cancer survival and gene expression in multiple cancer types

Different endogenous and exogenous mutational processes cause specific patterns of somatic mutations and mutational signatures. Although their biological research has been intensive, there are only rare studies assessing the possible prognostic role of mutational signatures. We used data from The Cancer Genome Atlas to study the associations between the activity of the mutational signatures and four survival endpoints in 18 types of malignancies. We further explored the prognostic differences according to, for example, the HPV status in head and neck squamous cell carcinomas and smoking status in lung cancers. The predictive power of the signatures over time was evaluated with a dynamic area under the curve model, and the links between mutational signature activities and differences in gene expression patterns were analyzed. In 12 of 18 studied cancer types, we identified at least one mutational signature whose activity predicted survival outcomes after adjusting for the established prognostic factors. For example, overall survival was associated with the activity of mutational signatures in nine cancer types and disease-specific survival in seven cancer types. The clock-like signatures SBS5 and SBS40 were most commonly associated with survival endpoints. The genes of the myosin binding protein and melanoma antigen families were among the most substantially dysregulated genes between the signatures of low and high activity. The differences in gene expression also revealed various enriched pathways. Based on these data, specific mutational signatures associate with the gene expression and have the potential to serve as strong prognostic factors in several cancer types. Different carcinogenic pathways have specific mutational signatures, which can provide details about gene defects related to cancer development and prognosis. Little is known, however, about associations between mutational signature activity and cancer outcome. Here, using data from The Cancer Genome Atlas (TCGA) database, the authors identify several mutational signatures that provide prognostic information for most cancer types. In particular, distinct cancer type-specific differences in gene expression were observed between low- and high-activity mutational signatures. The findings offer insight into a possible prognostic role for mutational signatures and could help further understanding of the etiology of mutational patterns in cancer.image