Robert S. Seitz, Sr.¹, Brian Z. Ring², Rodney S. Beck¹, Shannon M. Tarr³, David G. Hicks³, Noel C. Estopinal⁴, Douglas T. Ross^2
1Applied Genomics Inc., Huntsville, AL; ²Applied Genomics Inc., Sunnyvale, CA; ³Cleveland Clinic Foundation, Cleveland, OH; ⁴Comprehensive Cancer Institute, Huntsville, AL

Abstract #3642

cDNA microarray studies have revealed reproducible breast tumor subtypes distinguished by characteristic gene expression signatures (Perou 2000; Sorlie 2001). Principal among these are the 'luminal' and 'basal' types, so named because the gene expression pattern that distinguished them included normal luminal (keratin 8,18) as opposed to basal (keratin 5,17) markers. We have screened through hundreds of novel antisera targeted by gene expression data in order to characterize the biologic and clinical diversity of breast cancer in paraffin blocks from tumor cohorts. Several of these antibodies were targeted to genes that help distinguish gene expression-based subtypes. We show here that these antibodies a) stain the basal or luminal layer of normal breast ducts as predicted by the gene expression patterns, b) stain patients in a conserved manner across multiple cohorts, and c) are associated with different outcomes consistent with that predicted by their cognate genes in gene expression studies. These studies support the notion that gene expression based classification of carcinoma can be translated into immunohistochemical tools for biomarker discovery.

	(1.6 MB PDF)