Integrative modeling of tumor DNA methylation identifies a role for metabolism

DNA methylation varies across genomic regions, tissues and individuals in a population. Altered DNA methylation is common in cancer and often considered an early event in tumorigenesis. However, the sources of heterogeneity of DNA methylation among tumors remain poorly defined. Here, we capitalize on the availability of multi-platform data on thousands of molecularly-and clinically-annotated human tumors to build integrative models that identify the determinants of DNA methylation. We quantify the relative contribution of clinical and molecular factors in explaining within-cancer (inter-individual) variability in DNA methylation. We show that the levels of a set of metabolic genes involved in the methionine cycle that are constituents of one-carbon metabolism are predictive of several features of DNA methylation status in tumors including the methylation of genes that are known to drive oncogenesis. Finally, we demonstrate that patients whose DNA methylation status can be predicted from the genes in one-carbon metabolism exhibited improved survival over cases where this regulation is disrupted. To our knowledge, this study is the first comprehensive analysis of the determinants of methylation and demonstrates the surprisingly large contribution of metabolism in explaining epigenetic variation among individual tumors of the same cancer type. Together, our results illustrate links between tumor metabolism and epigenetics and outline future clinical implications.