Although cancer classification has improved over the past 30 years, there has been no general approach for identifying new cancer classes (class discovery) or for assigning tumors to known classes (class prediction). Here, a generic approach to cancer classification based on gene expression monitoring by DNA microarrays is described and applied to human acute leukemias as a test case. A class discovery procedure automatically discovered the distinction between acute myeloid leukemia (AML) and acute lymphoblastic leukemia (ALL) without previous knowledge of these classes. An automatically derived class predictor was able to determine the class of new leukemia cases. The results demonstrate the feasibility of cancer classification based solely on gene expression monitoring and suggest a general strategy for discovering and predicting cancer classes for other types of cancer, independent of previous biological knowledge.

Human T cell leukemias can arise from oncogenes activated by specific chromosomal translocations involving the T cell receptor genes. Here we show that five different T cell oncogenes (HOX11, TAL1, LYL1, LMO1, and LMO2) are often aberrantly expressed in the absence of chromosomal abnormalities. Using oligonucleotide microarrays, we identified several gene expression signatures that were indicative of leukemic arrest at specific stages of normal thymocyte development: LYL1+ signature (pro-T), HOX11+ (early cortical thymocyte), and TAL1+ (late cortical thymocyte). Hierarchical clustering analysis of gene expression signatures grouped samples according to their shared oncogenic pathways and identified HOX11L2 activation as a novel event in T cell leukemogenesis. These findings have clinical importance, since HOX11 activation is significantly associated with a favorable prognosis, while expression of TAL1, LYL1, or, surprisingly, HOX11L2 confers a much worse response to treatment. Our results illustrate the power of gene expression profiles to elucidate transformation pathways relevant to human leukemia.

A type 1 IFN pathway signature biomarker in blood is highly correlated with CDASI activity scores in DM, and may be a promising surrogate clinical trial end point. The correlation of serum IFN-β, but not IFN-α, with both a gene signature and CDASI suggests that IFN-β drives disease activity in DM.