New computational approach identifies not so novel drugs to tackle ovarian cancer

Computer program predicts potential for existing anti-inflammatory drugs to fight ovarian cancer, matching known drugs with new diseases.

Nov 23, 2017

In a paper recently published in Oncogene, a team lead by Analisa DiFeo, assistant professor of general medical sciences at Case Western Reserve University School of Medicine (OH, USA), has developed a computer program, DrugPredict, to match disease gene expression data with existing FDA drug profiles. In doing so, the program attempts to find previously approved drugs to treat new diseases. Through this ‘drug-repositioning approach’, the group identified indomethacin and other anti-inflammatories as potential compounds to target epithelial ovarian cancer cells.

In recent years, there has been an increasing interest in data driven drug development, as access to computational capacity – as well as the pressure to identify new drugs – has increased. A number of attempts have been made at matching the actions of known compounds to disease, with limited success.

Previous work has focused on finding similarities between drugs or between diseases, based on factors such as structure, side effects, genotype or expression. However, these methods have often struggled to take account of the variety of factors involved in a mechanism of action.

With their new program, DiFeo’s group draws upon one of the most extensive sources of drug data, the FDA approved drug database, to couple the best features of drug and disease led approaches.

When a disease is imputed into the software, the program first identifies the associated genes being expressed, which are correlated to mouse analogues allowing a profile of the phenotype to be built. This phenotype can be matched to the profiles of existing approved FDA drugs, providing a ranked list of potential hits. Finally, the hits can be compared to highlight the drugs mode of action.

To test their program, the team examined potential drugs for the treatment of ovarian cancer, and, to their surprise, discovered that a number of commonly available anti-inflammatory drugs ranked highly among the potential drug hits The best performing of these, indomethacin, underwent a number of in vitro studies, showing it to be active.

This work could have a big impact in the treatment ovarian cancer, currently the fifth leading cause of cancer deaths in women.

More importantly, however, its success highlights the potential for this computational drug repositioning approach to be used in future drug development, taking advantage of existing big data to repurpose drugs in ways not previously considered.

“Traditional drug discovery process takes an average of 14 years and billions of dollars of investment for a lead anti-cancer drug to make the transition from lab to clinic,” commented Anil Belur Nagaraj, study first author and research associate at Case Western Reserve University School of Medicine.

“Drug re-positioning significantly shortens the long lag-phase in drug discovery and also reduces the associated cost.”

"The primary advantage of drug re-positioning over traditional drug development is that it starts from compounds with well-characterized pharmacology and safety profiles. This significantly reduces the risk of adverse effects and attrition in clinical trials" concluded Rong Xu, co-senior author and associate professor of biomedical informatics.


Nagaraj AB, Wang QQ, Joseph P, et al. Using a novel computational drug-repositioning approach (DrugPredict) to rapidly identify potent drug candidates for cancer treatment. Oncogene. doi: 10.1038/onc.2017.328 (E-pub ahead of print) (2017);

Benjamin Walden

Commissioning Editor, Future Science

No comments yet.