What if a cheap medicine sold over the counter turned out to be a cure against cancer or another deadly disease? Scientists have devised a new way of predicting such unexpected benefits of existing drugs, and they have confirmed two potential new therapies just to prove the point. "This promises new uses for drugs that have already been tested for their safety and offers a faster and cheaper way to new medicine," says Atul Butte, an expert in bioinformatics at Stanford University School of Medicine in California, who conducted the study.
There are plenty of examples of drugs originally developed to treat one disease that turned out to help another: Acetylsalicylic acid (aspirin) is not just a pain killer but is also used to reduce the risk of heart attack. And when a blood pressure drug called sildenafil was discovered to have an unexpected side effect, it went on to become the erectile dysfunction blockbuster now known as Viagra. Such crossovers can save drug developers a lot of time and money. Developing a single new drug on average takes more than a decade and costs about $800 million. Existing drugs have known safety profiles and are approved for human use, so they can be rapidly evaluated for new indications.
"But most repurposing of drugs is still due to chance observations or educated guesses," Butte says. In today's issue of Science Translational Medicine, he and his colleagues present a more efficient way of finding such new uses for old drugs: by bringing together data on how diseases and drugs affect the activity of the roughly 30,000 genes in a human cell. Researchers have collected information on which genes are activated or silenced in certain diseases and by certain drugs for many years. "Our hypothesis was, if a disease is characterized by certain changes in gene expression and if a drug causes the reverse changes, then that drug could have a therapeutic effect on the disease," he says.
To find such opposing pairs, Butte and colleagues used public databases and compared the data for 100 diseases with that for 164 drug molecules. They found candidate therapeutics for 53 of the diseases. Many matches had already been discovered and turned into therapies, but others were completely unexpected. For example, the analysis predicted that an epilepsy drug called topiramate would be active against inflammatory bowel diseases such as Crohn's disease. And the over-the-counter drug cimetidine, which inhibits acid production in the stomach and is used to treat heartburn, matched a certain type of lung cancer.
To confirm this latter link, the researchers investigated the compound in a mouse model of lung cancer. They showed that it slowed the growth of human lung cancer cells but not kidney cancer cells in these mice. Similarly, giving topiramate to rats with colitis reduced swelling and ulceration in the animals.
John Overington, a computational chemical biologist at the European Bioinformatics Institute in Hinxton, U.K., is not convinced that these two particular drugs will get very far. "Topiramate hits quite a lot of targets and has complex side effects, while the doses needed for functional effects for cimetidine seemed high," he cautions. But he praises the paper's main idea. "This is a really important concept; it is almost like they are looking for an antidote to a disease." Stefan Schreiber, an expert on the genetics of inflammatory bowel diseases at University of Kiel in Germany, agrees that the idea behind the papers is more important than the two drug candidates. "This is a proof of principle," he says. "The main point is that someone has taken all of this available genomic data and shown what you can do with it."
The opportunities are growing rapidly, Butte says. "When we started the project 5 years ago, we had data for a hundred diseases and 164 compounds, he says. "Today it would be about 1400 diseases and 300 molecules." Butte hopes that scientists and pharmaceutical companies will continue to make data publicly available. The unknown benefits of many more drugs are just waiting to be discovered, he says.