Use of a Combined Gene Expression Profile in Implementing a Drug Sensitivity Predictive Model for Breast Cancer

Chemotherapy targets all rapidly growing cells, not only cancer cells, and thus is often associated with unpleasant side effects. Therefore, examination of the chemosensitivity based on genotypes is needed in order to reduce the side effects.

Various computational approaches have been proposed for predicting chemosensitivity based on gene expression profiles. A linear regression model can be used to predict the response of cancer cells to chemotherapeutic drugs, based on genomic features of the cells, and appropriate sample size for this method depends on the number of predictors. We used principal component analysis and identified a combined gene expression profile to reduce the number of predictors.

The coefficients of determinanation (R²) of prediction models with combined gene expression and several independent gene expressions were similar. Corresponding F values, which represent model significances were improved by use of a combined gene expression profile, indicating that the use of a combined gene expression profile is helpful in predicting drug sensitivity. Even better, a prediction model can be used even with small samples because of the reduced number of predictors.

Combined gene expression analysis is expected to contribute to more personalized management of breast cancer cases by enabling more effective targeting of existing therapies. This procedure for identifying a cell-type-specific gene expression profile can be extended to other chemotherapeutic treatments and many other heterogeneous cancer types.