OncoScore™ Companion Diagnostic Platform

Hundreds of studies have examined the global gene expression patterns of human cancers. Building upon the standardized curation and controlled vocabulary of the Oncomine Platform, Compendia Bioscience has performed a meta-analysis of 450+ independent cancer gene expression datasets comprising 40,000+ human tumors and spanning 15 cancer types aimed at identifying and validating sets of co-expressed genes (gene expression modules) for each cancer type.

It is now clear that robust and validated gene expression modules exist for each cancer type, consisting of tens to hundreds of co-expressed genes that often vary dramatically across tumors within a given cancer type. Gene expression modules may correspond to cell lineage, pathway activation, deregulation of a cellular process, genomic alterations or tumor microenvironment.

We identified 161 modules across 15 cancer types. For each module we ascribed a functional annotation and prioritized representative genes. We propose that by measuring a small number of representative genes from each module within a cancer type (i.e., a stratification panel), the majority of the molecular heterogeneity of a cancer type can be represented and an individual tumor can be ascribed a subtype based on the set of module scores.