Bernards and Weinberg reply

We appreciate the interest that our Concepts essay (Nature 418, 823; 2002) has evoked. But Edwards misrepresents our thinking when he writes: “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)?”. In fact, we argued that there are various types of primary tumour, some of which are preordained to become metastatic, others not. Hence, the differences lie between various distinct types of primary tumour. We suppose that in some tumours, the particular combination of alterations/mutations that enables cells to create a robustly growing primary tumour cell population also incidentally empowers them to become metastatic. Also, we do not imply that a single genetic change per tumour is involved in malignancy, as Edwards concludes.

We did not say, as Sherley asserts,that “the same genes are exclusively responsible both for cancer-cell metastasis and for the emergence and proliferation of cancer cells”. Instead, we argued that there are multiple alternative genetic pathways that lead to the creation of a primary tumour, each path being defined by the identities of the particular genes that are altered during tumorigenesis. According to our thinking, some combinations of genes that lead to primary tumour formation create growths that are unlikely to metastasize. Other combinations yield tumours that have a high proclivity for metastasizing. In the latter case, the combination of genes that yielded the primary tumour happens to be able to confer invasive/metastatic ability even though these phenotypes were not selected during the clonal expansions that created the primary tumour mass.

We did not say, as Sherley asserts, that cells that acquire both proliferative and metastatic changes will be rare in primary tumours. Instead, we said that certain combinations of genetic alterations that are selected for the proliferative advantage they confer will, incidentally, also confer invasive/metastatic phenotypes. This is in no sense tautological. It is simply the statement of the possibly pleiotropic actions of certain cancer-associated genes.