Virtual repurposing can speed the discovery of new uses for existing drugs

The process for discovering drugs tends to be based on the same kind of pigeonholing that doctors use to treat disease: Parkinson’s is one disease, Crohn’s is something completely separate. A new approach, sometimes called virtual repurposing, offers a way to discover unknown connections between “unconnected” diseases that may lead to new treatments.

Drug discovery usually begins with a promising molecule. It is then tested in animal models and, if there are signs of effectiveness, in human clinical trials. At every step along the way, the data emerging from this work must be analyzed. It’s a slow process that can take a decade or more and cost, on average, more than $2.6 billion.

But recent advances in data analysis are making it possible to turn the process on its head. Virtual repurposing begins with existing data. Using new capabilities in biomedical informatics and machine learning, researchers comb through massive quantities of data in search of common genetic links between diseases that, based on symptoms, would seem to have nothing in common. If a shared genetic pathway is discovered, then it may be possible to treat one disease with medicines approved for the other.

My colleagues and I used this technique to uncover a possible link between Parkinson’s disease and the inflammatory bowel diseases, Crohn’s disease and ulcerative colitis.

We started by searching for common genetic mutations in a population at high risk for Crohn’s disease: Jews of Eastern European origin, known as Ashkenazi Jews. One gene we found, called leucine-rich repeat kinase 2 (LRRK2), was highly associated with Crohn’s disease, increasing the risk for the disease by 70 percent. That came as a surprise, because LRRK2 has been linked to Parkinson’s disease, a neurodegenerative disorder that appears to have little in common with IBD.

We confirmed the genetic link by finding a connection between the LRRK2 mutation and Crohn’s disease in patients with Parkinson’s disease in both Ashkenazi and non-Ashkenazi populations.

If the two diseases truly emanate from the same mutation, then people with IBD should be more likely to have Parkinson’s disease than individuals in the general population. We compared the health data of more than 140,000 individuals with IBD to a control group of 720,000 people without IBD, and found that people with IBD were 28 percent more likely to have Parkinson’s disease than those without inflammatory bowel disease, clear evidence of a significant correlation, not just a genetic link between the two disease categories. We confirmed this finding in separate health-data studies in Sweden, Denmark, and Taiwan.

How can a Parkinson’s-IBD connection be explained? Chronic inflammation is the primary driver of Crohn’s and inflammatory bowel disease, so it is possible that chronic inflammation in the gut may somehow play a role in Parkinson’s. That’s a novel theory, because researchers trying to solve the Parkinson’s puzzle have focused their work on the brain, not on inflammation in the gut.

To test this inflammation theory, we searched data on millions of U.S. patients to see if there was any correlation between the anti-inflammatory medications taken by many people with IBD and Parkinson’s disease. Our analysis found that people who were taking anti-inflammatory drugs called tumor necrosis factor (TNF) inhibitors for IBD were 78 percent less likely to later develop Parkinson’s disease than those with IBD who were not taking these drugs.

Equally interesting, people with Crohn’s or ulcerative colitis who were taking TNF inhibitors had a rate of Parkinson’s that was seven times lower than that of the general public, in spite of the fact that people with IBD are more likely to develop Parkinson’s.

Without conducting a single animal study or clinical trial, we used virtual repurposing to accumulate strong evidence that reducing inflammation with TNF inhibitors might play an important role in preventing Parkinson’s disease.

These insights suggest that it would be promising to explore using anti-TNF drugs for people who are at risk for Parkinson’s. Since the disease typically appears in the late 50s, the use of anti-TNF drugs before then may be able to slow the progression of — or even prevent — Parkinson’s symptoms.

While anti-TNF drugs are not harmless, and no one knows when and for how long they would need to be given to lower Parkinson’s risk, our findings point at intestinal inflammation as a new target for preventing Parkinson’s disease.

The data might work both ways. In addition to finding a possible new treatment for Parkinson’s, our findings also raise the possibility of a new treatment for IBD. Drugs that inhibit the LRRK gene, an approach being considered for treating Parkinson’s disease, may also be helpful in treating inflammatory bowel disease. In fact, a study in mice showed that LRRK2 inhibitors reduced inflammation in the colon.

Rather than starting with an animal study, we ended with one. The progression from using data analysis to discover a genetic link, then confirming a connection, and finally running a trial in mice as a test of drug repurposing was fast (completed in two years) and relatively inexpensive (costing just over $1 million).

Of course, there’s still plenty more work to be done. Our next step is to analyze genomic data to determine whether people who take anti-TNF drugs for other diseases, such as rheumatoid arthritis and psoriasis, but who do not have inflammation in the gut also have a lower risk of Parkinson’s, or whether the correlation exists only among people with inflammatory bowel disease who are taking anti-TNF drugs.

This is just one illustration of the growing power of genomics and medical informatics to promote the practice of precision medicine. It is now possible to harness existing databases to search across diseases and unearth troves of information that pinpoint common molecular pathways as new targets for fighting disease, as well as to identify FDA-approved drugs that can be tested in virtual repurposing trials as potential new treatments for diseases. This precision medicine approach to drug discovery should supplement traditional clinical trials as it is an efficient avenue of scientific exploration that has great potential to accelerate and scale up the discovery of new ways to effectively treat disease.

Inga Peter, Ph.D., is a professor of genetics and genomic sciences at the Icahn School of Medicine at Mount Sinai.