DNA arrays and cancer classification 27/8/03. By Richard Twyman DNA arrays can be used to classify different types of cancer that may be difficult to distinguish by other methods.

There are over 200 different types of cancer, each of which has a unique set of clinical characteristics, a specific treatment regime and a different chance of being cured. Unfortunately, it is sometimes difficult for even experienced specialists to tell the difference between particular cancer subtypes.

Acute myeloid leukaemia (AML) and acute lymphoblastoid leukaemia (ALL) cells look very similar, for example, but they respond to different therapies. A correct diagnosis is therefore essential for successful treatment. In the case of AML, drugs such as daunorubicin and cytarabine are favoured. In the case of ALL, patients respond better to drugs such as vincristine and methotrexate.

Traditionally, the correct diagnosis of leukaemia subtypes (class predicition) has relied on a combination of techniques. Smears or biopsies are studied by skilled clinicians to look for subtle differences in the cell shape. A cytogeneticist may be able to find particular types of chromosome rearrangement that are specific for each cancer. In certain cases, the different cells produce different enzymes and therefore stain in distinct ways, or they may produce a marker protein that can be identified using an antibody. However, none of these tests is 100 per cent accurate. Even if separate tests are carried out in separate laboratories, the results can be conflicting and in some cases inconclusive.

The DNA microarray is a new diagnostic tool in cancer classification. Arrays can be used to look at the expression patterns of thousands of genes simultaneously. This provides a much more robust diagnosis than one based on a single protein or the activity of one enzyme.

In the case of AML and ALL it has been found that accurate classification is possible by analyzing the expression of 50 genes on an array representing nearly 7000 genes in total. In one study, 36 out of 38 patients were classified correctly using this single test. In only two cases was the diagnosis uncertain.

In the above example, a DNA microarray was used to classify cancer subtypes that were already known to exist but were difficult to distinguish. Another application of the technology is the discovery of new cancer subtypes. For example, only 40 per cent of patients with non-Hodgkins lymphoma respond to current therapy. Until recently, it was impossible to tell which patients would respond until treatment was underway. However, DNA microarray analysis revealed two, previously unrecognised subtypes of the disease that matched the different clinical outcomes and drug responses.

Gene expression analysis is therefore likely to be used more widely in the future to classify cancers and choose effective therapies.

Image credit: Dr David Becker

Further reading

Golub TR, et al. Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science 1999, 286(5439): 531-7. Abstract