An integrated analysis of the epigenetic, genetic, and transcriptional patterns associated with outcome across cancer types

Abstract

Successful treatment decisions in cancer depend on the accurate assessment of patient risk. To improve our understanding of the molecular alterations that underlie deadly malignancies, we analyzed genomic profiles from 33,036 solid tumors with known patient outcomes. Contrary to expectations, we find that mutations in cancer driver genes are almost never associated with patient survival time. In contrast, copy number changes in these same genes are broadly prognostic. Analysis of methylation, microRNA, mRNA, and protein expression patterns in primary tumors define several additional prognostic patterns, including signatures of tumor mitotic activity and tissue de-differentiation. Co-expression analysis with a cell cycle meta-gene distinguished proliferation-dependent and -independent prognostic features, allowing us to construct multivariate survival models with improved stratification power. In total, our analysis provides a comprehensive resource for biomarker and therapeutic target identification, and suggests that copy number and methylation profiling should complement tumor sequencing efforts to improve patient risk assessment.