Sir

The Concepts essay “A progression puzzle” by René Bernards and Robert A. Weinberg (Nature 418, 823; 2002), demolishing a popular view of cancer metastasis, is welcome. To take their argument even further, the concept that tumours evolve by sequential selection of gene changes suggests that both malignancy and metastasis are mechanistically rather simple processes, and follow from a complete abolition of the signals that normally prevent cells from surviving and growing in the wrong place.

Bernards and Weinberg argue that the gene changes that make cancer metastasis possible cannot be fundamentally distinct from the preceding gene changes that occur as the cell evolves to form a primary tumour, and conclude that “genes and genetic changes specifically and exclusively involved in orchestrating the process of metastasis do not exist”. If there is little difference between a primary malignant tumour and its metastases, the crucial issue becomes: what makes a tumour malignant (capable of metastasis)?

Applying the Bernards–Weinberg argument leads to the idea that malignancy must also be a short step from earlier stages in tumour development. It can be separated by at most one unique gene event from overproliferation, and hence there is probably no more than one gene and genetic change per tumour that is specifically and exclusively involved in malignancy.

Malignancy can be defined either as the ability of a tumour to spread into the surrounding normal tissue and grow there to form a primary malignant tumour, or as the potential to form metastases. The two definitions are almost interchangeable because the one almost invariably implies the other. Pathologists observed this over a century ago and it is a cornerstone of clinical histopathology — and strong support for Bernards' and Weinberg's view.

Applying the Bernards–Weinberg selectability argument to malignancy itself, the transition from non-malignant to malignant can involve at most one gene change that is unique to the malignant transition, because cells cannot accumulate mutations that provide malignancy-specific properties (the ability to spread) during their evolution before they begin to spread. At most, one critical gene change could occur at the point at which malignancy begins.

This leads to a very simple view of malignant tumours and metastasis: malignant cells are cells that can grow in alien environments, and by a process that is not subtle, but largely random accident, they spread, first locally and then eventually through the body to form life-threatening secondary tumours. Metastasis is merely a rare, stochastic event: the chance escape of a cell into the vasculature, its arrival at a suitable site and its growth there.

Much has been made of the need for cells to cross basement membranes, but damage by trauma, inflammation or necrosis can breach basement membranes, and even normal somatic cells can cross vessel walls (S. Koop et al. Proc. Natl Acad. Sci. USA 93, 11080–11084; 1996). Perhaps what is crucial is that when malignant cells are exposed to surrounding connective tissue they simply establish growth in the connective tissue space, instead of helping to repair the tissue organization by closing gaps or going into apoptosis (programmed cell suicide) when in the wrong place.