An evolutionary paradigm for carcinogenesis?