Generating an improved breast cancer subtype predictor based upon integration of gene expression data from multiple datasets

Breast cancer is a highly heterogeneous disease, determining the specific subtype is very important when selecting the most appropriate treatment. Research has been carried out over the past decade into the classification of breast cancer based on molecular subtypes in order to distinguish patients who will actually benefit from targeted therapies from those who will not, avoiding unnecessary treatment, side-effects and associated costs. However these new tests, one of which is currently being evaluated in a clinical trial, have recently been shown to be highly inconsistent.

We will assess the accuracy of these tests on a larger scale, to determine the predictive value and robustness of these methods in determining the same molecular subtype.