DNA and RNA isolation and the GEP and WES analyses were performed as described previously (20 (link)). RNA samples with an RNA integrity number ≥6.0 were used for microarray analysis. Labeled samples were hybridized to the SurePrint G3 Human Gene Expression 8x60 K v2 Microarray (Agilent Technologies). Microarray analysis was performed in accordance with the MIAME guidelines. For DNA data analysis, somatic mutations were identified by comparing data from tumor and corresponding blood samples. Mutations in 138 known driver genes were defined as those identified as pathogenic in the ClinVar database. Vogelstein et al (21 (link)) demonstrated that 138 genes, when altered by intragenic mutations, can promote or drive tumorigenesis. A most of tumors including colorectal cancers contain two to eight of these ʻdriver gene’ mutations and the remaining mutations are passengers that do not contribute to tumorigenesis directly. Thus, these 138 driver mutations are accepted as relevant genes to the tumorigenesis (21 (link)). Single nucleotide variants (SNVs) of the total exonic mutations for each sequenced tumor included nonsynonymous, synonymous, and indel/frameshift mutations.