Discovering New Uses for Old Drugs
It’s widely estimated that it takes at least $1 billion to develop a new drug. More than 100,000 compounds are screened to find about 10,000 that are worthy of studying in preclinical laboratory cell and animal studies. Of these, only around 10 ever make it as far as clinical trials. One of these experimental agents is eventually proven effective and approved for use by the U.S. Food and Drug Administration (FDA) — eight to 14 years later.
Given that drug discovery and testing is prohibitively expensive, time consuming and prone to failure, the number of agents approved for use in the U.S. by the FDA has fallen from 50 in 2000 to just 24 in 2011.
Despite the odds, a team of scientists at Georgetown University Medical Center has found two new potential anti-cancer drugs for the small price of several federal research grants. And they did it, quickly, using mostly computers.
The team, led by Sivanesan Dakshanamurthy, Ph.D., has developed a novel method to uncover new uses for existing drugs, of which there are thousands. The latest count puts the number of clinically active agents, already approved and available worldwide, at about 27,000.
And, given the emerging realization that identical biological pathways can be active in different ways in different diseases, “the possibilities for repositioning existing drugs for new indications seems limitless,” says Dakshanamurthy, an assistant professor of oncology who works in the Experimental Therapeutics Program at Georgetown Lombardi Comprehensive Cancer Center.
Working backwards — taking a known drug and finding new uses — to discover potential new therapies could radically reduce drug testing and approval time, Dakshanamurthy says.
In the August 1 issue of the Journal of Medicinal Chemistry, Dakshanamurthy and seven other GUMC researchers described how they found that a drug used to treat hookworm has unexpected anticancer properties, and that a popular anti-inflammatory drug is active in both rheumatoid arthritis and hard-to-treat cancers.
“This is just so much fun — it’s like being a kid in a candy store,” he says. For this grown-up computational chemist/biochemist, the sweet reward will be to efficiently identify new and unexpected treatments for patients using drugs already on the market.
“Molecule of Best Fit”
What Dakshanamurthy and his colleagues developed is a novel rapid computerized system that maps the crystal structure of an approved drug and tests whether it fits into human protein crystal structures.
“Drugs work on a lock and key system. The lock is the protein and the drug is the key that turns it on or off,” he says.
To find these potential keys, Dakshanamurthy taps into the National Institutes of Health (NIH) Chemical Genomics Center (NCGC) database of 27,000 pharmaceutical structures, which the federal government has made available to help researchers pursue drug “repositioning.” To find the potential locks, he mines the protein database maintained by the NIH’s National Center for Biotechnology Information.
The scientists developed a comprehensive computerized prediction method called “train, match, fit, streamline,” or TMFS, to map new drug-target interactions and predict new uses. It measures 11 different variables, including the shape and topology of the drug and ligand, the contact points of the ligand and the target protein, chemical similarity, and the tightness by which the agent docks and binds to the protein. “This allows us to better predict the ‘molecule of best fit’,” Dakshanamurthy says.
Using TMFS, the researchers screened 3,671 FDA-approved and investigational drugs across 2,335 protein structures. The method turned up known lock-and-key, drug-ligand interactions with 91 percent accuracy, as measured by published studies, and it also uncovered several new uses for old drugs.
For example, TMFS predicted that the anti-hookworm drug mebendazole can inhibit the vascular endothelial growth factor receptor 2 (VEGFR-2) which is a protein that binds to and activates VEGF. VEGF promotes angiogenesis, or the creation of new blood vessels — a process necessary for tumor growth. Mebendazole could therefore be a possible VEGF inhibitor, joining other experimental anti-VEGF agents being tested for various cancers. Dakshanamurthy demonstrated that the drug does indeed affect the function of VEGFR-2.
“It may be now possible to repurpose this anti-parasitic drug to cut off the blood supply that enables many forms of cancer to grow and spread,” Dakshanamurthy says.
Working Backwards to Fill the Drug Pipeline
The group also discovered evidence that the architecture and chemical signature in Celebrex, a popular prescription medicine for pain and inflammation, could allow it to bind the cell adhesion molecule cadherin-11, a molecule known to be important in both rheumatoid arthritis and in certain cancers with no known targeted therapies and poor prognoses. Stephen Byers, Ph.D., a researcher who studies the function of cadherins in breast and colon cancer, then proved through laboratory experimentation that Celebrex does indeed bind to cadherin-11, disabling its function. Byers is a co-author on the Journal of Medicinal Chemistry study.
The scientists also predicted that tamoxifen, which is widely used to treat and prevent some breast cancers, would interact with multiple proteins related to various disorders, including Alzheimer’s disease, and cancers of the prostate and ovaries.
“The safety of these drugs is already known, so once an old drug is repurposed to a new indication, it can directly enter clinical trials,” Dakshanamurthy says. “Based on past drug approvals, I would expect testing of approved agents could take between two to four years.”
Dakshanamurthy is not the only scientist working to repurpose old drugs, but he says the TMFS system offers the hope that the process can be computerized, which is relatively cheap, compared to traditional laboratory methods.
“We need to fill up the drug pipeline and get it flowing smoothly and quickly,” he says. “This method may help to pump up the volume — and that is very exciting.”
By Renee Twombly, GUMC Communications