Differential gene expression profiles and identification of the genes relevant to clinicopathologic factors in colorectal cancer selected by cDNA array method in combination with principal component analysis.

The clinical outcome of patients with colorectal cancer frequently varies even if they are at the same clinicopathologic stage. Alternative superior tumor markers of colorectal cancer are needed for prediction of clinical outcome. To clarify the regulatory factors in colorectal cancers, we examined differential expression profiles using cDNA macroarray technique with surgically resected specimens obtained from the patients with colorectal cancer. The gene profiles by an average-linkage hierarchical clustering analysis were found to be almost separable into two groups: tumor group and normal mucosa group. The relationship between several clinicopathologic factors and cancer related genes were investigated by using statistical analyses including principal component analysis (PCA). c-myc-binding protein MM-1, and c-jun proto-oncogene were identified as possible markers of tumor histology and clinical prognosis and early growth response protein 1 (EGR1) was selected to play an important role in progression of clinical stage. We conclude that, with PCA method, we successfully selected the genes relevant to clinicopathologic factors using limited population of clinical samples.