Molecular classification of cutaneous malignant melanoma by gene expression profiling

The most common human cancers are malignant neoplasms of the skin(1,2). Inc idence of cutaneous melanoma is rising especially steeply, with minimal pro gress in non-surgical treatment of advanced disease(3,4). Despite significa nt effort to identify independent predictors of melanoma outcome, no accept ed histopathological, molecular or immunohistochemical marker defines subse ts of this neoplasm(2,3). Accordingly, though melanoma is thought to presen t with different 'taxonomic' forms, these are considered part of a continuo us spectrum rather than discrete entities(2). Here we report the discovery of a subset of melanomas identified by mathematical analysis of gene expres sion in a series of samples. Remarkably, many genes underlying the classifi cation of this subset are differentially regulated in invasive melanomas th at form primitive tubular networks in vitro, a feature of some highly aggre ssive metastatic melanomas(5). Global transcript analysis can identify unre cognized subtypes of cutaneous melanoma and predict experimentally verifiab le phenotypic characteristics that may be of importance to disease progress ion.