Microarray expression identification of differentially expressed genes in serous epithelial ovarian cancer compared with bulk normal ovarian tissue and ovarian surface scrapings

The lack of reliable early detection of ovarian cancer and the absence of specific symptoms result in diagnosis of ovarian cancer at advanced stage in the majority of the patients. Through gene expression profiling we can identify important genes that may help understand the evolution from normal ovarian tissue to ovarian cancer. The gene expression profiles of 7 normal ovaries and 26 ovaries with serous epithelial ovarian cancer (SEOC) were examined by cDNA microarrays using supervised and unsupervised analysis, with sequential significance filtering. Real-time RT-PCR was used to measure and compare the expression levels of 5 selected genes: WAP four-disulfide core domain protein HE4 (WAP, up-regulated), secreted phosphoprotein 1 (SPP1, osteopontin; up-regulated), activin A receptor, type I (ACVR1, up-regulated), tumor necrosis factor (TNF superfamily, member 2; TNF, up-regulated) and decorin (DCN, down-regulated) in 4 epithelial scrapings and in 6 bulk-extracted normal ovaries. The gene expression profile of SEOC was not dependent on the stage of the disease at diagnosis. A supervised microarray data analysis identified a subset of 329 genes showing significant differential expression between SEOC samples and bulk normal ovarian tissue and ovarian surface scrapings, including several new genes such as TNFalpha and activin A receptor type I. The real-time RT-PCR for the up-regulated genes did not differ significantly between normal ovarian epithelial scrapings and bulk-extracted ovaries. However, decorin showed a statistically significant difference (P=0.0073) in expression between epithelial scrapings and bulk-extracted ovaries. Previously uncharacterized genes are associated with the malignant phenotype of SEOC. Bulk normal ovarian tissue may serve as control for SEOC tissue in gene expression profiling. Gene expression profiling and sequential statistical analyses of gene subsets can identify new genes and molecular pathways affecting development of SEOC. The genes of interest can be potential targets for future research of SEOC.