Figure 5

Human CRCs exhibit gene expression profile complexity consistent with significant tumor subclasses. Genes potentially able to distinguish cancer subtypes were identified from Affymetrix HG-U133 plus2 Genechip expression profiles by filtering for 3,285 probe sets that were top-ranked by raw expression and their differential regulation in at least 10 out of 100 human colorectal cancer tumors. Coordinately regulated transcripts and similarly behaving samples were identified via hierarchical tree clustering. Seven different gene clusters (C11-17) were identified that distinguished ten or more tumors from the other tumors. Gene clusters were found to be highly enriched for gene functions listed in Table 3. Data were processed using Robust Microarray Analysis (RMA) with expression value ratios depicted as the relative expression per probe set in each sample relative to the median of its expression across the 100 CRCs. A striking heterogeneity of gene expression was observed, including metallothionein genes in cluster C15 previously shown to be predictive of microsatellite instability (indicated by asterisk), and C17 represented by 734 probesets rich in genes associated with extracellular matrix and connective tissue, tumor invasion and malignancy. Tissue groups: AC, adult colon; CRC, human CRC. Staging: nAC, normal colon; Dukes A-D, human tumors obtained from individuals. Clusters C11-C17 labeled to the right of the heatmap correspond to the K-means functional clusters listed in Table 3.