1. ¹Centre de Recherche du Centre Hospitalier de l'Université de Montréal (CHUM)/Institut du cancer de Montréal, Montreal, Canada
2. ²Division of Gynecologic Oncology/Université de Montréal, Montreal, Canada
3. ³Department of Medicine, Université de Montréal, Montreal, Canada
4. ⁴Department of pathology, Centre Hospitalier de l'Université de Montréal (CHUM), Montreal, Canada
5. ⁵McGill University and Genome Quebec Innovation Centre, Montreal, Canada
6. ⁶Department of Human Genetics, McGill University, Montreal, Canada
7. ⁷The Research Institute of McGill University Health Centre, Montreal, Canada
8. ⁸Department of Medicine, McGill University, Montreal, Canada

Correspondence: A-M Mes-Masson, CR-CHUM/ICM, 1560, rue Sherbrooke est, Montreal, Quebec, Canada H2L 4M1. E-mail: anne-marie.mes-masson@umontreal.ca

Received 30 July 2004; Revised 14 September 2004; Accepted 14 September 2004; Published online 30 May 2005.

Abstract

Tumors of low malignant potential (LMP) represent 20% of epithelial ovarian cancers (EOCs) and are associated with a better prognosis than the invasive tumors (TOV). Defining the relationship between LMPs and TOVs remains an important goal towards understanding the molecular pathways that contribute to prognosis, as well as providing molecular markers, for these EOCs. To this end, DNA microarray analyses were performed either in a primary culture or a tumor tissue model system and selected candidate genes showing a distinctive expression profile between LMPs and TOVs were identified using a class prediction approach based on three statistical methods of analysis. Both model systems appear relevant as candidate genes identified by either model allowed the proper reclassification of samples as either LMPs or TOVs. Selected candidate genes (CAS, CCNE1, LGALS8, ITG3, ATP1B1, FLIP, KRT7 and KRT19) were validated by real-time quantitative PCR analysis and show differential expression between LMPs and TOVs. Immunohistochemistry analyses showed that the two tumor classes were distinguishable by their expression of CAS, TNFR1A, FLIP, CKS1 and CCNE1. These results define signature patterns for gene expression of LMPs and TOVs and identify gene candidates that warrant further study to deepen our understanding of the biology of EOC.