November 15, 2011 | A newly minted biotechnology company is offering to match the molecular genomic activity of previously approved medicines to that of known diseases to help drug makers find new uses for therapies. “We have two published studies demonstrating [our technology platform] can be used that way, and a third pending publication,” says Gini Deshpande, co-founder of Menlo Park, CA-based NuMedii.
The company’s matchmaking capability, which has been likened to an “online dating service,” is driven by a strong computational engine that reduces the search for new indications for existing drugs, says Deshpande. The company is based on technology developed in the Stanford lab of Atul Butte, who is a co-founder of NuMedii (along with Joel Dudley).
A multitude of other companies are trying to reposition approved drugs using more costly and time-consuming approaches, including cell based assays for screening an entire library of drugs, says Deshpande. NuMedii is less a fishing expedition than “educated fishing,” as one National Institutes of Health (NIH) scientist described it.
What distinguishes NuMedii is an annotated and normalized database developed at Stanford that holds genome-wide molecular profiles for more than 300 diseases, as well as the informatics know-how to use it to identify new uses, including those still in clinical development, says Deshpande. It is also the first computational company to test its hypotheses in a pre-clinical animal model context and publish the results, which could help drug makers “de-risk” their medicine repurposing work.
As reported in Science Translational Medicine, two predictions made by the technology—now licensed to NuMedii—have demonstrated preclinical activity in mouse studies (Dudley et al., 3:(96): 96ra76 and Sirota et al., 3:(96): 96ra77 ). One of the predictions helped reposition a generic anticonvulsant medication for the possible treatment of inflammatory bowel disease. The other aimed a generic anti-ulcer drug at lung cancer. A third prediction, also for a gastrointestinal indication, will soon be written up in the literature.
“The platform essentially marries the Stanford database with other drug, data, and knowledge bases,” says Deshpande. It can be tapped to find a disease whose molecular profile either counters or mimics that of a commercially available medicine. The NuMedii platform has potentially broad applications, she adds, but repositioning drugs with known safety profiles was the easiest starting point.
The timing couldn’t be better, given that the NIH has been actively pushing drug repositioning. Novartis, GSK, and Pfizer already have internal repositioning efforts. NuMedii’s approach would also be an efficient way for such companies to find first-ever treatments for a host of rare diseases, notes Deshpande.
NuMedii is in the midst of establishing partnerships with several pharmaceutical companies to demonstrate the breadth of the platform and enhance its value, says Deshpande. She’s hopeful the methodology will get used early in the drug development process to help companies “maximize the potential” of approved compounds in development by testing them for additional, multiple indications in parallel. NuMedii’s platform could also be used to detect potential blockbuster indications for current, low-volume niche products, she adds.
NuMedii faces more of an uphill battle in the clinical development realm, where computational technology is a novelty, says Deshpande. But the platform could prove valuable in guiding and prioritizing new drug development programs, together with physician input on real-world prescribing and dosing patterns. The technology can’t definitively predict a drug’s failure, given the volume of confounding factors (such as patient compliance), but it could capture its efficacy odds overall and in certain subsets of patients.
Ultimately, the platform could be used to identify potential biomarkers of disease and genetic receptivity to certain drugs, says Deshpande. “Preliminary publications have shown that these disease genomic databases can also be used for biomarker discovery and we hope to pair this approach with drug indication discovery over the next year.” •