Researchers at GlaxoSmithKline and their collaborators have demonstrated a method for suggesting alternative indications for existing drugs that uses data from genome wide association studies and a repository of drug information.
In a correspondence piece published last month in Nature Biotechnology, the researchers described their approach, which is based on data from the National Human Genome Research Institute's repository of GWAS data and Informa Healthcare's Pharmaprojects database of drug development projects.
Using the method, they found that out of 155 genes that could be mapped to GWAS traits and were also targeted by available drugs, 92 were associated with drugs that had indications that were different from the mapped GWAS traits, making those compounds potential repositioning candidates.
For example, denosumab, a monoclonal antibody marketed as Prolia by Amgen for osteoporosis and bone cancer, targets the TNFSF11 gene, which, in addition to its link to bone disease has also been associated with Crohn's disease by GWAS, the paper states.
Another example found that a Phase 3 drug being developed by ReceptoPharm, dubbed RPI-78M, which targets the IL27 gene and is currently indicated for adrenoleukodystrophy — an inherited disorder that leads to progressive brain damage, adrenal gland failure, and death — could potentially be repurposed for Crohn's and inflammatory bowel diseases.
These findings provide "evidence that GWAS data ... give insights into the biology of diseases" that could "lead to the immediate translational opportunities for drug discovery and development," the investigators wrote.
Philippe Sanseau and Pankaj Agarwal, computational biologists at GSK and co-authors on the paper, told BioInform that although some researchers have combined GWAS and drug data in studies of particular disease areas, their work marks the first time such an approach has been applied across multiple disease areas in a "systematic" fashion.
"The method is surprisingly simple," requiring only some simple Perl scripts that were used to analyze the data, said Agarwal, the director of systematic drug repositioning in GSK's computational biology division.
"We took all the genes that are associated genetically to a disease from NHGRI's website and we intersected it with all the drug targets that we could get from resources like Pharmaprojects," he explained.
Once that was done, "the key challenge was understanding if [the findings] meant anything," Agarwal said. "You've got this list of genes and drug targets but ... is it anything useful or ... was that just by chance?"
As the paper explains, the researchers began by creating a list of GWAS genes associated with disease traits from NHGRI's catalog of published GWAS studies, which includes 796 papers. Using a series of filtration steps, the researchers selected 991 genes "with recognizable HUGO gene names from Entrez Gene," of which 155 had an associated drug project, according to Pharmaprojects.
Next, the team compared the "disease indications pursued by the drugs with the GWAS trait" for each of the 155 genes to "identify matches and mismatches between the disease indications and the GWAS traits." Matches were drugs "with the same or closely related indication to the GWAS trait" and mismatches — or possible drug repositioning opportunities — were drugs "with indications different from the GWAS traits."
Using this approach, the team was able to uncover 123 mismatches, which corresponded to 92 genes mapped to 51 GWAS traits where drugs were used to treat one disease but not a second condition also associated with the same gene.
Using the denosumab example to illustrate, Agarwal explained that if the group found a protein that was genetically associated with both Crohn's disease and osteoporosis; and a drug that was already marketed for osteoporosis, "the hypothesis is that you could use the same drug for Crohn's disease."
The question, he added, is, "Is there any other data that supports the use of that drug for Crohn's disease?"
In the Crohn's disease example, the researchers found additional evidence that links differential expression of the TNFSF11 gene with an allele associated with Crohn's disease.
But although these results make it "tempting to speculate that denosumab could be tested for Crohn's disease ... more work is required to understand mechanistically the role of TNFSF11," the authors wrote.
In addition to several possible drug repositioning opportunities, the team's analysis revealed instances where the drug indication matched the GWAS trait or disease. In those cases, "we assumed that it was reinforcing or [providing] additional supporting evidence for the indication," Sanseau, who is the head of GSK's computational biology department, told BioInform.
One example discussed in the study showed that a class of statins — which are used to lower cholesterol — targets the HMGCR gene, which has SNPs that have been associated with low density lipopolysaccharide cholesterol.
Another drug, ustekinumab, which is marketed by Janssen Biotech as Stelara for treating psoriasis, was also shown to have the correct indication by the study, according to the Nature Biotech paper.
The next steps in this study would be to do additional "validation work," which, "depending on the drug and what stage the drug is in," could call for the use of animal models or "proof-of-concept studies" in humans for each of the potential new indications found in the study, Sanseau said.
"You cannot just do blanket validation, you have to look at the opportunities and decide what is the best validation strategy," he said.
The researchers warn in the paper that not all mismatches "will lead to successful drug repositioning opportunities."
An example is the NOS2 gene, which is the target for several inhibitors that have been developed for diseases like rheumatoid arthritis.
According to the researchers, GWAS studies have shown that SNPs within NOS2 are associated with psoriasis, suggesting that these inhibitors could also be used to treat the skin condition.
However, "at least one small study did not identify any clinical improvement in psoriatic subjects when a topical inhibitor of nitric oxide synthesis was applied," a finding that highlights some of the "limitations" of a GWAS-approach to drug repositioning, the paper states.
Other limitations of the approach are that it is restricted to diseases that have been studied using GWAS and it is "difficult to assess ascertainment bias in our study," the researchers wrote.