Transcriptional gene expression profiles of colorectal adenoma, adenocarcinoma, and normal tissue examined by oligonucleotide arrays.

TLDR

Gene expression differences between normal colon tissue and neoplastic lesions, including adenomas and adenocarcinomas, have been previously documented. The study profiled mRNA from 18 adenocarcinomas and 4 adenomas, plus matched normal mucosa, using an Affymetrix GeneChip containing ~6,200 human cDNA and EST probes. Only a few transcripts were differentially expressed (19 up‑regulated, 47 down‑regulated), and hierarchical clustering of these genes cleanly separated adenoma, carcinoma, and normal samples, indicating that genome‑wide expression profiling can classify solid tumors.

Abstract

Using an oligonucleotide array containing sequences complementary to approximately 3200 full-length human cDNAs and 3400 expressed sequence tags (GeneChip, Affymetrix), mRNA expression patterns were probed in 18 colon adenocarcinomas and 4 adenomas. Paired normal tissue was available and analyzed for each of the tumors. Relatively few changes in transcript expression are associated with colon cancer. Nineteen transcripts (0.48% of those detected) had at least 4-10.5-fold higher mRNA expression in carcinoma compared with paired normal samples, whereas 47 transcripts (1.3% of those detected) had at least 4-38-fold or lower expression in the tumor tissue compared with the normal samples. Some of these differences were confirmed by reverse transcription-PCR. Many of these transcripts were already known to be abnormally expressed in neoplastic tissue in general, or colon cancer in particular, and several of these differences were also observed in premalignant adenoma samples. A two-way hierarchical clustering algorithm successfully distinguished adenoma from adenocarcinoma and normal tissue, generating a phylogenetic tree that appropriately represented the clinical relationship between the three tissue types included in the analysis. This supports the concept that genome-wide expression profiling may permit a molecular classification of solid tumors.