Quite a lot: evolution is one of the defining features of cancer. In multicellular organisms cells cooperate in order to achieve a common goal: to preserve the organism. In a tissue containing a tumour the situation changes: mutations bring about a diversity of genotypes in which some cells will have a higher potential to reproduce and survive than others. In effect an ecosystem is created in which different individuals compete for a limited amount of resources, often to the point of destroying the ecosystem, that is, the organism.
This evolutionary nature of cancer has consequences that affect potential treatments. Since any given tumour will contain a number of different tumour cells, a therapy that is successful at eradicating one type of tumour cell might only make it easier for a different type to spread through. For a treatment to be successful its creators should consider how the ecosystem will be altered and what kind of evolutionary dynamics will be favoured.
Understanding the driving forces of cancer evolution is likely to be a necessary step to understand (and hopefully deal with) cancer. For that reason it would be interesting to see how models of what is beginning to be called computational evolution (nice article in Nature review Genetics, vol 7, september 2006, pp729-735) can be used to study evolution in the context of cancer.