5 December 2008
In their study, Garman et al. retrieved gene expression data from 52 samples representing clinical stage I and II disease from a publicly available database, and correlated the patterns of gene expression with information on tumour recurrence in order to develop a prognostic model [Garman et al. (2008) PNAS 105(49):19431-19436]. The model they developed was based on examination of the expression pattern of 50 genes and was 90% accurate in predicting risk of recurrence in this initial data set. The model was further validated by examining its predictive power using two independent date sets consisting of 55 and 73 tumours and was able to correctly classify 69.1% of patients in the first cohort and almost all patients in the second. Following on from the identification of those at high-risk of recurrence, the researchers went on to investigate therapeutic strategies for this group. They used colon cancer cell lines to investigate the relationship between the high-risk gene expression phenotype and sensitivity to therapeutic agents, and demonstrated a correlation between the two; furthermore, treatment with certain agents was able to reverse the profile of a tumour from high risk to low risk. However, this was only demonstrated under laboratory conditions using cell lines and clinical trials would need to be undertaken in order to validate the observed effects.
Comment: Cancer cells usually exhibit many genetic aberrations and a greater understanding of these can be useful in classifying tumours, thereby aiding disease prognosis. In addition, a greater understanding of the biology of individual tumours can lead to targeted therapeutics as well as the identification of new targets for therapy. Using tumour genome expression profiling to guide prognosis and treatment has already been developed and approved for breast cancer (see previous news). Whilst extensive clinical evaluation of such risk prediction models is critical, these types of companion diagnostic tests are likely to become increasingly common and could potentially have an enormous impact on the pharmaceutical and biotech industries.