Tailoring cancer therapy to specific tumour types maximizes efficacy while minimizing toxicity. Historically, cancer classification has been based on morphology, but cancers with seemingly identical morphological and histopathological features can progress and respond to therapy in radically different ways. A better method of classifying cancers was needed to help predict clinical outcome and make the most of the available therapy — the possible solution came from microarray technology.
The first evidence that gene-expression profiling could distinguish between cancer types came in 1999, from Todd Golub, Donna Slonim and colleagues. They chose two types of leukaemia as a test case: acute myeloid and acute lymphoblastic. The approach involved identifying a 'predictor class' of genes, based on their non-random expression patterns, and evaluating the prediction strength. In addition to distinguishing between the two types of leukaemia on the basis of expression-profile differences, the method could also predict their responsiveness to chemotherapy. The paper laid out a general analytical approach to cancer classification based on gene expression, which could be adapted to assign cancers to hitherto unknown classes.
A year later, Ash Alizadeh, Michael Eisen and colleagues used a similar approach to uncover gene-expression heterogeneity in diffuse large B-cell lymphoma (DLBCL), which is the most common type of non-Hodgkin's lymphoma. The expression profiles identified two distinct forms of DLBCL and correlated with the responsiveness of the tumours to treatment.
The next landmark example of how expression profiling can help to predict clinical outcomes came from the breast cancer field. In this case, specific molecular signatures (of genes involved in the cell cycle, invasion, metastasis and angiogenesis) were shown to accurately predict high likelihood of metastases and, therefore, poor overall prognosis, in the absence of other indicators. This study was the first to show that metastatic potential can be gleaned from the gene-expression data of the primary tumours. After further refinement, related breast cancer-profiling diagnostics have since become commercially available.
Although it might still be too early to see the effect of this technology in the clinic, an important feature of microarray analysis is its lack of bias, which allows microarray-based cancer classification to be systematic and not limited by our previous biological knowledge.
ORIGINAL RESEARCH PAPERS
Golub, T. R. & Slonim, D. K. et al. Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science 286, 531–537 (1999)
Alizadeh, A. A. & Eisen, M. B. et al. Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature 403, 503—511 (2000)
van't Veer, L. J., Dal, H. & van de Vijver, M. J. et al. Gene expression profiling predicts clinical outcome of breast cancer. Nature 415, 530–526 (2002)
About this article
Cite this article
Skipper, M. Profiling cancer expression. Nat Rev Cancer 6 (Suppl 1), S22 (2006). https://doi.org/10.1038/nrc1865