Will systems biology translate into ever higher healthcare costs, or are there savings to be made?

It has been estimated that as much as $300 billion may have been spent on cancer research since the war on cancer was announced by President Nixon in 1971 [1]. As well as bringing about huge improvements in cancer care – an average of four extra years in life expectancy per cancer patient, totaling 23 million years for the US alone – this has had a huge impact on our understanding of biology. The impact has been both direct (increased understanding of cell cycle control, DNA replication and repair and so on) but also, and perhaps more importantly, indirect impact through an explosion in the availability of technologies that are being used to study life from the level of single molecules to the level of the entire genome. But did we really need the expensive, labour-intensive, openended use of genome-wide interrogation to develop four new multigene prognostic tests for breast cancer? This has led to yet more expense as clinical trials are performed to evaluate and compare each of the tests to the others. Is this worth it when a simple appraisal of breast cancer biology allowed the elaboration of a classical four protein immuno-histochemical test that appears to perform just as well. This question is particularly acute in the light of the huge debate that is raging on the cost of healthcare (e.g. [2]). Will systems biology add to these costs, or can it reduce them? The greatest transformation in cancer biology in the last two decades has been in our genome-wide approach to systems biology, where we now conceive of gene and protein networks, as opposed to individual genes or even pathways, specific to different cancers. Research on breast cancer, in particular, has been a trailblazer in terms of impact and innovation, with a terrifically positive outcome for many women who now are experiencing an 80% survival rate, as opposed to 80% mortality, five years from diagnosis. And yet the means by which this has been achieved – the identification of gene signatures that can be used to guide treatment for patients with breast cancer – highlights one of the great problems that still faces biologists and clinical scientists: How can we efficiently translate the information we get in the test tube, from a tissue culture or from a whole genome microarray into a meaningful understanding of the underlying biology? The key word here is ‘efficiently’. There are currently five distinct tests, which each analyse an almost completely distinct sets of genes and proteins, yet all of which are used to the same end: determining which patients with breast cancer belong to the 15% who need adjuvant chemotherapy to prevent early recurrence, and which belong to the 85% who can be spared the burden of a toxic and traumatic treatment that they do not need. Four of the five prognostic tests that are currently used to guide decision-making in breast cancer (Oncotype Dx, PAM50, Mammostrat and Mammaprint) have been developed through open-ended approaches, using empirical algorithms to detect patterns that correlate with disease. The other, IHC4, was developed on the basis of a knowledge of the biology of breast cancer: that it is growth factor (Her2) and hormone (estrogen and/or progesterone) dependent, and that it involves an increase in cell proliferation that can be measured by staining for the Ki67 antigen.