Small and large-scale chromosomal aberrations, including amplifications, deletions, and rearrangements, are commonly seen in solid tumors. In human colorectal cancer (CRC), amplification of chromosome regions 20q, 8q, 13q, 7p, 12p, and loss of 18q, 8p, 17p, 4p, and 15q are frequently observed. These regions harbor oncogenes, tumor suppressors, and DNA repair genes that are known to be involved in the etiology of the disease. A combined, high-resolution gene expression and DNA copy number approach has been applied to test the hypothesis that a more comprehensive measure of the genetic damage in CRC will reveal novel molecular targets. Here, microarrays consisting of ∼35,000 spotted oligonucleotides have been utilized to examine both gene expression and DNA copy number changes in primary colorectal tumors from patients with stage II-IV disease. A colorectal cancer tissue array containing approximately 100 CRC samples has been constructed for use in validation studies. To date, 50 primary CRC samples in which the site of metastasis and other clinical information is available have been profiled. For expression studies, an unsupervised, class-discovery algorithm was utilized to search for molecular signatures that are associated with the clinical data. A subset of samples characterized by genes involved in cell communication, signal transduction, and metabolism was identified; however, no correlation with the clinical data was found for this subset of tumors. Using class comparison and hierarchical clustering, a statistically significant set of genes discriminated between primary tumors located in the proximal (cecum, ascending colon, and transverse colon) colon from those isolated from the distal (descending, sigmoid, and rectal) colon, but no gene signatures were associated patient age, disease stage, or CEA levels. The same microarray platform was utilized for array comparative genomic hybridization (aCGH) studies. Analysis of the aCGH results revealed several chromosomal changes that have not been previously reported. A combined analysis of the aCGH and expression data, followed by validation on tumor tissue arrays, is being carried out to identify candidate genes associated with colorectal cancer.

[Proc Amer Assoc Cancer Res, Volume 47, 2006]