Drug combination through an integrated drug information

Abstract

Drug combination is that two or more active compounds combined in a single dosage form in order to enhance the effect of drugs. Combinatorial therapy is widely used in treating the complex disease like cancer because it is known as a promising method to improve the therapeutic efficiency and reduce the side effects. Thus, discovering beneficial drug combination has been getting more interests, so that methods using gene expression profiles have been developed recently. However, there were limitations in the previous approaches. First, it was ambiguous which information was important for prediction of efficient drug combination using gene expression data. Additionally, the studies did not choose the target diseases and genes arbitrarily rather than systematically, for example there have been only a few researches regarding complex diseases like cancer. To overcome these limitations, I focused on finding drug combination candidates suitable for treating cancer by prediction of combinatorial effects of drug pairs and which information mainly affects successful detection. In this research, it was the basic assumption that genes work as gene modules. The gene modules expressed at high level in cancer disease (cancer module) were obtained from gene expression profile data of cancer drug. In addition, gene expression when treating two drugs were predicted in the case that gene expression profile of each treatment was given. Then gene modules were identified from the predicted gene expression (prediction module). By comparing the prediction module to the cancer module, it was estimated whether two drugs were synergistic or antagonistic. Eventually, it was turned out that the properties of gene modules affected by drug combination are more important than similarity between gene expressions affected by each treatment for successful detection of useful drug combination.