Gene expression was analysed in 68 stage I ovarian cancers and 15 borderline samples using microarrays. Tumors were obtained directly at surgery and immediately frozen in liquid nitrogen until analysis. Glass arrays containing 16000 genes were used. Data was interpreted by unsupervised and supervised analysis. All tumors had a wild-type p53. Unsupervised analysis identified eight major clusters one of which was statistically associated to overall survival, grading and histotype (cluster C) and another with grading and histotype (cluster D). Supervised analysis showed a clear association between histotypes and gene expression profile and distinguished the four major histotypes (serous, clear cells, mucinous, endometrioid). The data also discriminated between the different grades of the tumors. The subset of 15 borderline tumors were undistinguishable from grade 1 tumors. When patients, relapsing or not, were analysed, a subset of genes able to differentiate them was identified. The genes identified in the major clusters belong to general classes of lipid biosynthesis, transport and metabolism, immunoglobulins (antigen presenting and processing) and genes acting as metal, ion and anion transporters including ATPase activity coupled with ion transport. Keywords: disease state analysis
Overall design
Gene expression for 83 ovarian tumor samples was evaluated: 68 from patiens presenting stage I carcinoma and 15 with borderline desease. A common reference design was applied and each samples was cohybridizated with a universal reference. Also a dye swap control has been done so that for the two technical replicates of each sample an expression profile of raw data was obtained. These data were independently processed and combiend to have a unique normalized profile for each sample.
Less...
More...