Skip to main content
Premium Trial:

Request an Annual Quote

Vanderbilt Applies DNA Biorepository, PheWAS to Speed Drug Repurposing Research


CHICAGO (GenomeWeb) – Vanderbilt University Medical Center is beginning to enroll about 440 subjects for a Phase II study of whether misoprostol — widely used to induce labor and to prevent ulcers — might also be effective for prevention of recurrent Clostridium difficile colitis.

While this sounds like just another trial to expand indications for an existing drug, this study is the first with human subjects to test a new method of drug repurposing based on genetic data. The process promises to drastically reduce the time and cost involved with achieving US Food and Drug Administration approval for new indications of proven generic drugs and other compounds coming off of patent protection in the near future.

"I would love to be able to have those of us who are interested in repurposing generics communicate to the world that it doesn't have to cost $1 billion and 15 years," said Jill Pulley, executive director of the Vanderbilt Institute for Clinical and Translational Research.

Pulley oversees the Vanderbilt Accelerating Drug Discovery and Repurposing Incubator, a federally funded effort to apply BioVU, the Nashville, Tennessee-based institution's massive data bank of more than 250,000 longitudinal medical records and DNA samples, and the Vanderbilt-developed phenome-wide association study technique to drug repurposing. (Vanderbilt just last week launched Nashville Biosciences, a company intended to leverage BioVU and PheWAS to accelerate drug and diagnostic discovery.)

"Repurposing starts with the human genome," Helen Naylor, assistant director of precision medicine at Vanderbilt's Center for Knowledge Management, said last month at the American Medical Informatics Association Informatics Summit in San Francisco.

The Vanderbilt Accelerating Drug Discovery and Repurposing Incubator is looking for relationships between genotypes and phenotypes, with the help of BioVU, Naylor explained.

"With BioVU being large, we're able to find genotypes that look like drugs," Pulley said in an interview. Not all do, but this method speeds the process of identifying certain candidates for this new type of repurposing work.

"We are using natural experiments of human beings walking around with SNPs that mimic the effect of drugs," Pulley continued. "When we find a SNP that mimics the effect of the drug, we can anchor on the known, established indications of the drug and then find additional uses."

"Not all of them do. There are plenty of them [for which] it's very hard to match to the effect of a drug, but in some cases, you can. That takes a lot of work."

There usually is not a direct observation of a SNP activating the same receptor a drug does. "There is a lot of other work to make sure it really does look like those people are walking around taking that drug and you can see what the effect is," Pulley said.

Two years into the work — and one year into the five-year NIH grant — Vanderbilt has created a dozen drug-indication pairings.

It still is a labor-intensive process, as Vanderbilt is working with about 20 different databases, not counting years of scientific literature, to identify potential new therapeutic uses for generic drugs. Humans have to "connect the dots in databases and in literature," Pulley said.

PheWAS helps. "One of the cool things is the PheWAS approach itself because it screens out so much of the noise that you have less to deal with," Pulley said. "We can't be checking 9,000 diseases every time, so PheWAS cuts off 8,000 of them to make it a manageable number."

The Vanderbilt drug repurposing program is able to limit its scope of work and its costs by focusing specific compounds. "We like drugs that either have generics or are coming off of patents in the next two to three years," Naylor said at AMIA.

"We also narrow it to diseases that have extreme unmet medical needs," Pulley later noted. Some are rare conditions, but many are subsets of heterogeneous conditions that are poorly treated. "When we explore in-depth into the phenotypes, we are almost always coming up with a subset of the disease," Pulley said.

Examples include depression and lupus. "There would never be, in our world, a drug that's used for all types of depression," Pulley said. "It's always a subset, and that's a really important part of what we're doing."

It helps if there is a known genetic link to the disease, but it is not the only factor the program considers in deciding what to target. "It helps with the gene product in the assessment of what that protein does in the context of disease and pathophysiology," Pulley added. "It really helps to have that there. But that doesn't mean that our enrichment strategy in the clinical trials that we're conducting is genotype-driven."

They are not all rare diseases. "Many are rarer diseases that that have no therapy after all, and we go after [those] full-steam, but some of them are subsets of disease that we think are think are the actual disease." It could be a subset of a heterogeneous disease.

She suggested that there about 250 drugs Vanderbilt might be able to repurpose with this genotype-phenotype matching. If the platform works, the university suddenly has a big pipeline for potential fast-track research.

For the misoprostol study, BioVU helped the laboratory of David Aronoff, the division chief of infectious diseases at VUMC, identify "SNP-based proxies" for the drug, Pulley explained.

"The indications from misoprostol show up in our data and then the new indication, or similar versions show up in our data. That was tested in animal studies multiple times in multiple ways," Pulley said. While the mice eventually died from severe C. diff. colitis, the misoprostol worked well enough for Vanderbilt to head toward a Phase II human trial, she said.

This whole process took a year rather than the typical 15.  

"You would never have the confidence to secure funding, investigator time, [institutional review board] approval, [and] FDA [Investigational New Drug program] approval without having all of these triangulated data sources," including genomes, Pulley said. "Within one year, there's now an FDA IND in place to conduct a Phase II human trial, and that part's important, too, because not only can you do it faster, we are able to skip Phase I work."

Vanderbilt set up a methodology involving PheWAS and evidence synthesis from Naylor and other information scientists. "It's a platform that's intended to pull ideas all the way to real-world clinical-approved use," Pulley said. That, of course, takes time, and Vanderbilt is in a proving stage now.

"But there are aspects that are very pragmatic and real-world and necessary for translation," Pulley noted. Those including finding regulatory pathways, how to commercialize, figuring out whether there is patent protection, not to mention how to make such a program sustainable once the NIH funding runs out.