The loss of gene function would be expected to be close to random initially, but the cancer specific effects would occur because of hits on tumor suppressor genes. There are of course notable examples of gain of gene function too, that are selected for, but I am asking about the big picture here