Recently the New York Times has echoed research by Robert Moyzis (UC Irvine) and Henry Harpending (Utah) recently published in PNAS. This research claims that human evolution was, contrary to what it was previously thought, especially active between 50000 and 10000 years ago. These results are still quite controversial but (quoting from the NYT) Dr. Moyzis comes with a couple of reasons:

1) Human population started to grow in Africa and then elsewhere. Larger population size lead to more room for evolution to experiment with mutations

2) As a result of population expansion, some of this early humans migrated to different parts of the world where they had to confront different environments (climates, diseases…) to which they had to adapt differently.

This is interesting, plausible (which does not make it necessarily right) and also applicable to tumours. As tumours grow, the swell in the number of tumour cells make it very likely for mutations to increase the diversity of the population. Furthermore, the raise in population numbers and the space they take makes it also likely that subpopulations of the tumour will be subject to different microenvironments which should lead to different selection pressures. This would mean more diversity and and increased selection of phenotypic capabilities present at a given time in the tumour.