Estrogen receptor status in breast cancer is associated with remarkably distinct gene expression patterns

To investigate the phenotype associated with estrogen receptor a (ER) expre ssion in breast carcinoma, gene expression profiles of 58 node-negative bre ast carcinomas discordant for ER status were determined using DNA microarra y technology. Using artificial neural networks as well as standard hierarch ical clustering techniques, the tumors could be classified according to ER status, and a list of genes which discriminate tumors according to ER statu s was generated. The artificial neural networks could accurately predict ER status even when excluding top discriminator genes, including ER itself. B y reference to the serial analysis of gene expression database, we found th at only a small proportion of the 100 most important ER discriminator genes were also regulated by estradiol in MCF-7 cells. The results provide evide nce that ER+ and ER- tumors display remarkably different gene-expression ph enotypes not solely explained by differences in estrogen responsiveness.