Molecular classification of human carcinomas by use of gene expression signatures

Classification of human tumors according to their primary anatomical site o f origin is fundamental for the optimal treatment of patient's with cancer. Here we describe the use of large-scale RNA profiling and supervised machi ne learning algorithms to construct a first-generation molecular classifica tion scheme for carcinomas of the prostate, breast, lung, ovary, colorectum , kidney, liver, pancreas, bladder/ureter, and gastroesophagus, which colle ctively account for similar to 70% of all cancer-related deaths in the Unit ed States. The classification scheme was based on identifying gene subsets whose expression typifies each cancer class, and we quantified the extent t o which these genes are characteristic of a specific tumor type by accurate ly and confidently predicting the anatomical site of tumor origin for 90% o f 175 carcinomas, including 9 of 12 metastatic lesions. The predictor gene subsets include those whose expression is typical of specific types of norm al epithelial differentiation, as well as other genes whose expression is e levated in cancer. This study demonstrates the feasibility of predicting th e tissue origin of a carcinoma in the context of multiple cancer classes.