At what point do tumour cells acquire their metastatic potential? Are certain primary tumours prone to metastasis, or is the ability to take up residence in a foreign tissue a characterstic of only a few cells that have managed to break free from their primary tumour host? A gene-profiling study, published by Todd Golub and colleagues in the January issue of Nature Genetics, begins to answer these questions.

The authors analysed the gene-expression profiles of 12 metastatic andenocarcinoma nodules from tissues such as lung, breast, prostate, colorectum, uterus and ovary, and compared them with expression profiles of 64 primary adenocarcinomas representing the same spectrum of tumours. They identified 128 genes that distinguished primary from metastatic adenocarcinomas. Metastasis was associated with the upregulation of a number of genes that regulate protein translation ( SNRPF , EIF4EL3 , HNRPAB and DHPS ). Other upregulated genes seemed to come from the non-epithelial component of the tumour, such as those that encode type I collagens, indicating the importance of the stroma in regulating metastasis. Analysis of additional tumours revealed a similar metastatic gene signature.

This metastasis-associated gene-expression pattern was also present in some primary tumours, so did this mean that these tumours were destined to metastasize? The authors found that patients with primary tumours that expressed the metastatic gene profile had significantly shorter survival times than cancer patients whose tumours did not. This means that some primary tumours already have the propensity for metastasis as early as the time of diagnosis.

The authors also looked for the metastasis-associated gene signature in other tumour types, and found that their pattern could be used to predict metastatic potential of small stage I primary breast adenocarcinomas, prostate adenocarcinomas and medulloblastomas. This indicates that there are generic gene-expression programmes associated with the metastatic process in different tumours. Notably, the gene-expression profile was not able to predict metastasis in patients with diffuse large-B-cell lymphoma. This might be because haematopoietic malignancies have special mechanisms for spreading throughout the blood vessels and lymphatic system.

Golub and colleagues admit that although their outcome predictor was statistically significant, it was still imperfect, and suggest that additional factors are involved in determining tumour behaviour. But the discovery of an expression signature that can be used to classify a subset of primary solid tumours as premetastatic will be useful not only in determining prognosis, but also in designing therapies to stop the spread of tumours.