Expression and methylation profiles associated with recurrent mutations in stage II colorectal cancer.

Abstract

e14613 Background: Aberration of normal genetic and epigenetic patterns occurs at early stages of colorectal cancer (CRC) and accumulates throughout cancer progression. To characterize pre-metastatic tumors, a series of stage II, microsatellite stable, colon tumors and their paired mucosa were profiled on RNA expression and DNA methylation microarrays ( www.colonomics.org ). Our aim is to define molecular subtypes based on recurrent gene mutations, methylation and expression profiles, and explore if these molecular subtypes are associated to patients’ prognosis. Methods: We have sequenced exomes (Illumina Genome Analyzer) of a subset of 42 COLONOMICS normal-tumor paired samples (21 good and 21 bad prognosis). Variants identified in normal tissue were used to filter SNPs. DNA methylation (Infinium Human Methylation 450k) and RNA expression (Affymetrix U219) data from those samples were used in this analysis. Correlation was used to assess the association between tumor mutations and differentially expressed/methylated genes. Significant genes were subsequently used to perform tumor clustering. Results: Exome analysis revealed a mean of 150 somatic mutations per sample. From these, 12 variants were recurrently mutated (KRAS, TP53, etc) in more than 3 tumors that were used to define tumor subtypes based on gene methylation/expression patterns. We obtained 12 profiles that clearly identified the cluster of mutated samples. For some profiles, the cluster only includes those tumors with the mutation. Interestingly, some clusters included the mutated samples and additional tumors showing the same phenotype despite not having the mutation. For each mutation, the overlapping between the differently methylated/expressed genes ranged from 14 to 200 common genes. When combining data from the three platforms two main CRC molecular subtypes emerge; each of which shows molecular heterogeneity but no association with prognosis. Conclusions: Mutational status is associated with gene methylation and expression patterns in CRC patients. Although none of these clusters was associated with prognosis, different groups of tumors could be related to distinctive pathways, which may reveal useful as therapeutic targets.