The contribution of gene expression profiling to breast cancer classification, prognostication and prediction: a retrospective of the last decade.

In the last decade, the development of microarrays and the ability to perform massively parallel gene expression analysis of human tumours were received with great excitement by the scientific community. The promise of microarrays was of apocalyptic dimensions, with some experts envisaging that it would be a matter of a few years for this technology to replace traditional clinicopathological markers in clinical practice and treatment decision-making. The replacement of histopathology by high-tech and more objective approaches to cancer diagnosis, prognostication and prediction was, at that time, a foregone conclusion. Ten years after the initial publications of translational research studies using microarrays, one cannot deny that this technology has changed the way breast cancer is perceived. It has brought the concept of breast cancer heterogeneity to the forefront of cancer research, and the fact that distinct subtypes of breast cancer are completely different diseases that affect the same anatomical site. Furthermore, it has led to the development of prognostic and predictive 'gene signatures', which are yet to be fully incorporated into clinical practice. Importantly, though, the prognostic and predictive power of microarrays has been shown to be complementary to, rather than a replacement for, traditional clinicopathological parameters. Here we endeavour to provide a fair and balanced assessment of what microarray-based gene expression analysis has taught us in the last decade and its contribution to breast cancer classification, prognostication and prediction.