Identification of the molecular mechanisms for dedifferentiation at the invasion front of colorectal cancer by a gene expression analysis.

PURPOSE The aim of this study is to identify gene expression signatures that accompany dedifferentiation at the cancer invasion front in colorectal cancer. EXPERIMENTAL DESIGN Two types of colorectal cancer were selected. Both types were well-differentiated adenocarcinomas at the superficial lesion. One type showed a dedifferentiated phenotype at the invasion front (type A, 13 samples); the other showed almost no dedifferentiated cancer cells at the invasion front (type B, 12 samples). Laser microdissection was combined with a cDNA microarray analysis to investigate the superficial lesions and the invasion front in colorectal cancers. RESULTS Eighty-three genes were differentially expressed between types A and B in the superficial lesions, and the samples of superficial lesions were divided correctly into two clusters by these genes. Interestingly, the samples of the invasion front were also divided into the two same clusters by these genes. The text mining method selected 10 genes involved in potential mechanisms causing dedifferentiation of cancer cells at the invasion front. The potential mechanisms include the networks of transforming growth factor-beta, Wnt, and Hedgehog signals. The expression levels of 10 genes were calculated by quantitative reverse transcription-PCR and 8 genes were confirmed to be significantly differentially expressed between two types (P < 0.05). The gene expression profiles of 8 genes divided 12 test cases into two clusters with one misclassification. CONCLUSIONS The molecular mechanisms constructed with 8 genes from three networks of transforming growth factor-beta, Wnt, and Hedgehog signals were found to correlate with dedifferentiation at the invasion front of colorectal cancer.