Comparison of gene expression patterns between avian and human ovarian cancers

Objectives. A putative model of spontaneous cancer has been described in the laying hen that bears significant similarities to human ovarian cancer. Our objective was to characterize and compare the patterns of gene expression in chicken and human forms of this disease. Methods. RNA from 20 localized and metastatic ovarian and oviductal chicken tumor samples was isolated, amplified using in vitro transcription, and hybridized against normal ovarian epithelium to a customized cDNA microarray constructed for these studies. Differentially expressed genes were identified for localized ovarian, metastatic ovarian, and oviductal (or tubal) cancer by class comparison using BRB-ArrayTools. Results were validated with semi-quantitative PCR. A gene list (prediction model) constructed with the class prediction tool was used in a human ovarian cancer microarray obtained from the GEO datasets (GSE6008) in order to compare these results across species. Results. Class comparison analysis between localized ovarian, metastatic ovarian and oviductal cancer yielded 41 different informative probes that coded for 27 unique genes. Localized ovarian samples clustered between metastatic ovarian and oviductal cancer samples. Using our chicken data as a training set and leaving oviductal samples out of the analysis, we created a prediction model that classified early stage and advanced stage human ovarian cancer gene expression arrays with 78% overall accuracy. Conclusions. Gene expression of spontaneous ovarian cancer in the chicken is comparable to gene expression patterns of human ovarian cancer. (C) 2010 Elsevier Inc. All rights reserved.