Translation of gene expression profiles into a novel immunohistochemistry based classification of carcinoma

A90

Gene expression profiling combined with hierarchical clustering has identified significant biologic variation in tumor samples that can be used to classify patients in clinically significant ways. We have endeavored to translate this into a panel of immunohistochemistry antisera that can be used to classify tumor samples embedded in paraffin blocks. Several hundred novel antisera were generated, targeted by gene expression data and screened for utility in immunohistochemical classification of carcinoma samples. "Panels of Diversity" tailored to tissue type have been characterized and used to stain tissue arrays containing breast or lung cancer specimens with clinical follow-up. Individual antisera identified subsets of patient samples that were conserved in relative proportions and association with outcome between independent cohorts within tissue type and as well as between tissue types. These "Panels of Diversity" may be revealing a classification of carcinoma that transcends tissue of origin and may be useful for rational stratification of patients in clinical trials.