A CANCER, put simply, is a gang of rogue cells multiplying out of control. But each gang is different, so “cancer” is actually a term that embraces hundreds of specific ailments propelled by an even larger number of genetic and epigenetic traits. The old ways of characterising it, by the anatomical site of its debut (kidney, for example, or prostate gland) and the histology of its cells, seem increasingly out of date. Instead, thanks to genomics, researchers have unprecedented information on the molecular changes which propel it. The challenge is making sense of those data and putting them to use.
A pair of papers just published in Nature Genetics begin to do that. The Cancer Genome Atlas, a project launched by America’s National Institutes of Health, has assembled genetic data on thousands of tumours and made them available to anyone who wants to analyse them. These studies are early fruits of that endeavour.
One, led by Rameen Beroukhim of the Broad Institute, in Cambridge, Massachusetts, looked at cancers with an unusual number of copies of certain sections of their DNA. Dr Beroukhim examined 4,934 specimens from 11 traditionally defined types of cancer and found 140 regions of DNA that were sometimes either multiplied repeatedly or deleted altogether. Only 35 of these regions contained either genes known to suppress tumours or those, known as oncogenes, which when mutated encourage cancers to form.
The other study, led by Chris Sander of Memorial Sloan-Kettering Cancer Centre, in New York, proposed a way of categorising tumours by the genetic and epigenetic changes in their cells, rather than by anatomy and histology. Dr Sander developed an algorithm to examine 3,299 tumours from 12 traditionally defined types of cancer. He created two overarching groups: those with somatic genetic mutations (ie, mutations that have happened in a person’s body cells during his lifetime, and which are thus not inherited) and those that, like Dr Beroukhim’s, had an unusual number of copies of some sections of their DNA.
One surprise was that Dr Sander could draw such clear lines. Cancers with many somatic mutations rarely had many unusual copy numbers, and vice versa. Less surprising—indeed, the predicate of the whole exercise—was that within Dr Sander’s two main classes of cancer he recognised 31 subclasses. As cancer biologists have long suspected, tumours from the same type of tissue often had different genetic traits, while those from different tissues were frequently similar. For example, a type of lung cancer shared characteristics with a type of head and neck cancer.
This has practical implications. Both diseases might be good candidates for a specific combination of drugs that attack their specific combination of mutations. Increasingly, clinical trials of drugs are likely to span traditional categories of cancer by pooling those in different parts of the body that have similar genetic characteristics, and also to test more than one medicine at a time.
The work of Dr Beroukhim and Dr Sander would have been impossible until recently. Thanks to advances in genomics and computational biology, understanding of cancer looks set to grow rapidly. These studies are just the start. The Cancer Genome Atlas is part of a wider effort, the International Cancer Genome Consortium, that is cataloguing genetic abnormalities in 50 traditionally defined types of cancer. The consortium’s set of data is expected to triple over the next two years, and the trickle of results should turn into a flood. The hope is that this flood will then wash away ignorance about how cancers develop, and eventually wash the disease itself from the bodies of far more of those who suffer it.