Personalized drug repositioning using gene expression

Abstract

The use of existing medications for diseases they were not originally developed for is called drug repositioning. A popular drug repositioning method to find new drugs against specific cancer types is to search for drugs which are expected to bring back the gene expression activity of cancer cells to that of healthy cells (‘normalization’). One of the main research goals of this thesis was to investigate of this method could also be used on the gene expression profiles of individual tumors, enabling personalization of drug repositioning candidates for each patient. We initially had some success with this approach but this eventually lead to a systematic validation of the underlying principle using almost 10,000 tumor samples across 18 different tumor types. Unfortunately, the predictive power of the method was found to be much lower than previously reported and the part that remained could be nullified by correcting the gene expression profiles of the drugs for the downstream effects of reduced cell division. These results indicate that the current use of the method does not result in drug repositioning candidates specific for a tumor type but is only able to select generally cell-toxic drugs.