Molecular classification of cancer: Class discovery and class prediction by gene expression monitoring

Although cancer classification has improved over the past 30 years, there h as been no general approach for identifying new cancer classes (class disco very) or for assigning tumors to known classes (class prediction). Here, a generic approach to cancer classification based on gene expression monitori ng by DNA microarrays is described and applied to human acute Leukemias as a test case. A class discovery procedure automatically discovered the disti nction between acute myeloid Leukemia (AML) and acute Lymphoblastic Leukemi a (ALL) without previous knowledge of these classes. An automatically deriv ed class predictor was able to determine the class of new leukemia cases. T he results demonstrate the feasibility of cancer classification based solel y on gene expression monitoring and suggest a general strategy for discover ing and predicting cancer classes for other types of cancer, independent of previous biological knowledge.