Sir

The model of cancer metastasis suggested by René Bernards and Robert A. Weinberg in their Concepts essay (Nature 418, 823; 2002) is, in my view, a tautology. The suggestion that the same genes are exclusively responsible both for cancer-cell metastasis and for the emergence and proliferation of cancer cells is tantamount to saying “cancer cells that can proliferate do proliferate”. It would be a great loss if this type of idea caused a decline in research to investigate the existence of new genes involved in metastasis, a major factor in cancer mortality.

There is a finite probability that any cancer cell that can proliferate at one body site can also proliferate at other sites, if it can get there and stay there. How these two requirements are met is the real crux of the metastasis question. Bernards and Weinberg wish to dismiss the possibility that specific genetic changes, beyond those that govern proliferation, are required for successful cancer-cell relocation.

However, the argument Bernards and Weinberg used to arrive at this idea is, in my opinion, flawed. They argue that cells that acquire both proliferative and metastatic changes will be rare in primary tumours, thereby making it “difficult to imagine how metastasis can ever proceed”. This is a remarkable proposition in the context of a discussion of the initiation of cancer cells, which is itself an extremely rare occurrence. The authors' new concept — that metastasis is not due to a selected cell phenotype — would be better supported by a stochastic mechanism formulation.

Clinically, metastases range from presentation with large primary tumours to presentation without any identifiable primary tumour at all. It may be that the factors responsible for metastasis are cancer-cell proliferation plus myriad other small, unknowable variables that combine to create the conditions for relocation. This is a restatement of the tautology of Bernards' and Weinberg's argument.