Researchers from Roche and Stanford University are developing a computational method that may be able to help drug and diagnostic makers better understand the genetic pathways related to drug response and disease mechanisms.
The method, which is currently only applicable to mice, but might eventually be developed for human use, uses haplotype data to compare genomic differences among mouse strains that could help predict drug metabolism. The researchers also believe the technique could help shed additional light on toxicology and pharmacodynamics.
The method has been in development by researchers at Roche since at least 2000, but the current proof-of-concept study on the metabolism of the drug warfarin marks the first time investigators have applied the technique to a problem with commercial implications. In the study, which is published in this month's Nature Biotechnology, the researchers used the computational method to trace one aspect of warfarin metabolism in mice to a genetic variation in CYP450 2C-type enzymes.
"It's basically an entirely new way of thinking about how to approach understanding how a drug is metabolized," Gary Peltz, head of Genetics and Genomics at Roche in Palo Alto, Calif., told Pharmacogenomics Reporter this week. "In the near future, what we'll be showing is a new way in which to understand how drugs act and how they induce toxicities" by modeling them in mice.
"Anything that you can measure in the mouse, you can potentially discover the genetic component in the [individual] variability."
Roche is "starting to experiment with ways to use" the technique, but all of the methods are published and the mouse SNP database used in the study is available on the Internet, so other investigators are free to use it, said Peltz. "It could markedly accelerate our ability to understand some key things about how drugs act."
In the study, the investigators showed that the accumulation of 7-hydroxywarfarin during warfarin metabolism is related to genetic variations in a haplotype block associated with CYP45 2C-type enzymes. The average size of haplotype blocks in the database employed for the study was about the same as a single gene, allowing for nearly single-gene resolution.
"Anything that you can measure in the mouse, you can potentially discover the genetic component in the [individual] variability," by looking for strain-to-strain genomic differences in silico, Steven Shafer, a co-author of the study and a professor of anesthesia at Stanford University, told Pharmacogenomics Reporter this week. "[That] presumably will help you understand the mechanisms you're looking at."
For a method that stands to inform pharmacogenomic developments, the researchers picked a likely subject for a proof-of-concept study. "We picked warfarin as an example for two reasons," said Shafer. "One is that warfarin is a very well-studied drug — for humans, we understand subject-to-subject variability. Another thing is that warfarin remains a clinically exceedingly relevant drug — a very commonly prescribed drug, and it's considered to be one of the most toxic drugs that are prescribed."
The technique's limitation is that in order to work, strain-to-strain variability should be accounted for by a single gene. Warfarin metabolism is associated with multiple genes, so in order to simplify the analysis, the researchers focused on the variability in the levels of the drug and nine particular metabolites in plasma from 13 mouse strains.
Peltz and colleagues are currently conducting studies in which they look for genetic factors that might confer susceptibility or resistance to drug-induced liver toxicity related to acetaminophen and troglitazone, he said. "We're also trying to reduce it from an in vivo experimentation to actually being able to do it in vitro, where you can do these assays in microtiter wells, and that will really markedly expedite the analysis."
Mice are genetically similar enough to humans that researchers can use mouse models to draw some conclusions about human drug response, said Shafer. "At the simplest level, if one sees high levels of variability in cytochromes, as opposed to high levels of variability in cytosolic hepatic enzyme systems, or in drug transporters, that can help to identify where the variability is likely to be in the human population," he said. Individual cytochrome enzymes do not map to human analogues one to one, but they come pretty close, he added.
There are at least three factors that would make it difficult to enable researchers to use this technique with human data, Peltz and colleagues wrote in a separate article last September in Trends in Genetics. These include the fact that short regions of linkage disequilibrium result from a large number of recombinations; humans are often heterozygous at specific loci, unlike inbred mice; and the fact that outbred animals simply have much greater genetic diversity.
In research published last week in the journal Anesthesiology, the same group of researchers that published the Trends in Genetics paper used the computational method to establish that the beta-2-alpha receptor gene could explain some of the individual variation in opioid-induced hyperalgesia. "Had it not been for that tool, one would never think to look at that particular gene as part of the mechanism," said Shafer.
— Chris Womack ([email protected])