What is the basis of the tissue tropism shown by metastatic cells? Joan Massagué and colleagues now report a gene-expression signature that distinguishes breast cancer cells that metastasize to bone from those that metastasize elsewhere.

Massagué and colleagues have previously shown that the human breast cancer cell line MDA-MB-231 has a poor-prognosis metastasis signature and have subsequently identified and validated another set of genes in this cell line that specifically mediates bone metastasis in the mouse. In this study, single-cell-derived progenies (SCPs) derived from MDA-MB-231 were introduced into the arterial circulation of immunodeficient mice. The development of bone metastases was followed by transducing the SCPs with a triple-modality reporter gene and tracking the metastatic cells with bioluminescence imaging and fluorescence microscopy. Bone was the main site of tumour growth, but the SCPs varied in the aggressiveness of their growth in bone and this correlated with whether or not they expressed the bone-metastasis gene set. A few SCPs grew in the adrenal glands and, if tail-vein injection was used, some SCPs grew in the lung. SCPs that grew well at one site did not necessarily grow well at other sites, consistent with the hypothesis that growth at metastatic sites is increased by genes that confer favourable tumour–stroma interactions at that site.

So are there differences in gene expression that can account for the variability in metastatic activity of the SCPs? The authors found 286 genes that differed more than twofold in their expression between SCPs. Classification of the gene-expression profiles showed that SCPs that had different primary metastatic tropisms — bone or lung — or were weakly metastatic formed distinct clusters. These three clusters were significantly closer to each other than to the profile of a normal human breast epithelial cell line. These data support the idea that distinct gene-expression patterns are responsible for variation in metastatic tropism.

To investigate the relevance of these findings to the behaviour of disease in humans, 63 primary breast cancers were examined for expression of 50 genes from the bone-metastasis gene set that are also present in the poor-prognosis signature. Hierarchical clustering did not distinguish between tumours that had metastasized to bone and those that had not. However, when the analysis was restricted to those tumours that were known to have metastasized, the 50-bone-metastasis gene set did distinguish a bone-metastasis cluster from a lung-metastasis cluster.

Confirmatory studies could lead to an accurate predicter of bone-metastasis tropism in primary breast cancers, which would be valuable in the effective management of breast cancer patients.