PhD, MT (ASCP), Senior Editor
Researchers confess to the power of whole genome association studies, but say that this SNP-driven approach has limitations.
Heard any good science jokes lately?
Ever hear the one about the long-haired genetic researcher who went to the barber, asked for short hair-cut, but left with only a SNP (pronounced “snip”)?
Some researchers using the whole genome association (WGA) approach to identify gene candidates for human disease and/or drug targets have heard that old joke, especially the ones who use the single nucleotide polymorphism (SNP) as a marker.
A powerful new genetic screening approach of the high-resolution-loving, genomic-savvy basic or R&D scientist, WGA has only been around since the full human genome was published in 2002. But already, many have jumped on the bandwagon of this approach because of its statistically-solid, albeit enigmatic, experimental framework.
“If you are interested in making the connection between common variation and disease traits, then the WGA approach is the only game in town,” says David Goldstein, PhD, center director, Duke University Institute for Genome Sciences & Policy (IGSP), Center for Population Genomics & Pharmacogenetics, Durham, N.C. “All of the old battles about whether or not this approach would work have disappeared. And all of the folks that were on the naysayer’s side have just gone suddenly silent. There is just no debate about it—the whole framework works.”
Use of information
And WGA seemingly works so well that the approach is being used by academics and drug companies alike. “The primary use of the data will be to target disease,”
click to enlarge
This schematic demonstrates how WGA can be used to identify a disease gene from a whole genome scan. (Source: Genizon Biosciences)
says Eric Schadt, PhD, senior scientific director of genetics at Merck & Co., Inc., Seattle. “The charter here at Merck is to integrate information—genetic and other genetic expression profiling and proteomic—to come up with networks that are predictive of disease and drug response . . .,” says Schadt. “. . . and, with these networks, predict the target and the biomarkers that serve as biomarkers to diagnose disease and detect drug response.” So what Merck will do is make small molecules or other constructs to the hit the gene(s) target(s) identified in the WGA study.
But the whole purpose of a WGA study is not to simply find drug targets, but to look at the bigger picture, to peer into the biological world of disease. Schadt explains it this way: “What’s happening between that change in DNA and getting diabetes is a whole array of molecular networks being perturbed and twisted in a way that causes disease.” Schadt adds that Merck’s focus is “to generate the right kind of data to complement the public effort and to fill in the networks that actually cause disease.” For Merck, WGA is absolutely critical and used all of the time. They generate all of their WGA data using either Illumina, San Diego, or Affymetrix, Santa Clara, Calif., genotyping chips.
Another use for these WGA data is the generation of biomarkers. “We use the info in the DNA to come up with biomarkers that identify populations representing subtypes of a particular disease, such that subtype 1 would be given one drug and subtype 2 would be given another drug. The idea of coming up with the right drug for the right person at the right time is the overarching theme in the use of these data,” says Schadt.
Some WGA projects don’t exactly have this “personalized medicine” objective in mind just yet. In fact, some are just at the basic research level right now. “We are looking at a few different traits with WGA approaches,” says Goldstein. One set of genes he studies is related to host genetic determinants of control of human immunodeficiency type 1 virus (HIV-1) infection. The actual goal of this project is to try to understand genetic differences among people in the early control of the virus and use that as an indicator of how to best direct HIV vaccine strategies.
Results from Goldstein’s HIV WGA project, which is part of an ongoing series of studies, were recently published in Science magazine. One HIV gene that was determined through his WGA studies to have relevance in the control of infection is an RNA polymerase, which according to Goldstein, is also relevant as a potential target for drug or vaccine development.
Pharma and biotech are not involved in the funding of Goldstein’s projects. The research is entirely funded by the National Institutes of Health, Bethesda, Md. But why aren’t drug companies funding a project with the potential for a far-reaching impact on health care. “The way we are approaching that work is that as soon as we have clear genetic signals, we simply publish it and then it informs whatever development efforts are going on,” says Goldstein, who adds that “In some areas, we have partnered with drug companies to carry out studies that might help target selection.” In general, Goldstein’s attitude is simple. He tries to understand the genetics of a disease and then publishes his results, with which, he says, the drug industry does what it will.
Despite its popularity, the WGA approach still has some dissenters, especially since there has yet to be a drug whose target was identified using the approach approved by the US Food and Drug Administration (FDA). But there seems to be good news coming.
No drugs yet!
“So far I am not aware of any published result that came from a human genetic study and made it to a drug that made its way through the clinic,” says Schadt. “We do have several drugs in the pipeline that have originated this way that are making their way through the clinic.”
And there also appears to be drugs on the way from other companies as well, as Carsten Rosenow, PhD, senior marketing manager for Illumina explains. “I just came back from an FDA meeting
click to enlarge
This diagram represents how WGA can be used in a systems biology approach to identifying biomarkers that predict variations in drug response. (Source: Merck and Company)
and a few pharma companies there have looked at pharmacogenetics utilizing not high-density arrays but SNPs to look at individuals for adverse drug reactions,” he says. Illumina, with several iterations of the human beadchip, the largest of which contains one million genetic markers including a significant number of SNPs, has published a number of papers describing the use of this chip for performing WGA studies.
According to Rosenow, the whole goal that Pharma has after a identifying a SNP mutation in a gene that causes a disease phenotype is three-fold. First, they want to see how the gene is involved in the disease phenotype. Then, they want to develop a drug that targets this gene. And, finally, they want to see how the phenotype responds to the drug.
“For drug discovery [using the WGA approach], I don’t think the pharma companies really talk about it . . . it’s too early. The identified target goes into the black box where the pharma company doesn’t tell you which target they work on . . . then in eight to 10 years, they come out with a new drug against that target,” says Rosenow.
Another reason there are no drugs approved yet might not be so obvious and might actually go against the expectations some individuals have of WGA as a drug-target-finder. “We are finding real disease genes in the sense that we find genetic differences that have small effects on risk of disease . . . there’s no ambiguity about that,” says Goldstein. But, he explains, not every gene will be a drug target and that there are also cases of genes that are drug targets, but not involved in disease pathogenesis.
He gives the example of a group of genes formerly associated with common forms of epilepsy; the genes happen to code for a sodium channel. However, he says, when the same genes were carefully screened by WGA for association with epilepsy, it was found that they don’t contribute to disease. That does not make them bad drug targets, says Goldstein, just not associated with epilepsy. In other words, just because WGA data does not produce a drug target doesn’t mean it is useless.
“I think it’s going to be a slow process,” Goldstein says of the WGA approach to drug discovery. “It’s not an immediate outcome of a WGA study that there’s a new target and then a new drug for the disease that follows.” And it should not be. In fact, maybe the expectations of WGA were just too high to begin with. That seems to be the case for researchers like Demetrius Maraganore, MD, professor of neurology, Mayo Clinic College of Medicine, Rochester, Minn.
“I am quite disappointed at this early point with deliverables of WGA studies.” That’s what Maraganore had to say about WGA after he published results on Parkinson’s-related SNPs that could not be reproduced by other groups. “I am anticipating a series of disappointments for those who are using WGA to understand disease, to ultimately nominate targets for drug discovery,” says Maraganore.
The reason for his disappointment is rooted in the fact that there have been a number of papers published in the last year or two in Science and Nature in which researchers found a SNP or a few SNPs in a gene that were reproducibly associated with disease across two or three populations. However, Maraganore says that the effect sizes reported in those papers are trivial. For example, he explains that some of the effect sizes yielded odds ratios around 1.3 for a given SNP candidate. And with odds ratios this low, he says, it’s no wonder that there is not much news of drug companies using this approach to develop drugs against these targets.
According to Marganore, the main limitation of the WGA is the fact that the SNP is used as a marker, the piece that is actually measured, and upon which all WGA results are based. “I don’t think we will ever reach a point in time, with very few rare exceptions, where SNPs will make good diagnostic markers,” he says. “I think the concept of personalized medicine has some limitations because these SNP markers are so variable in frequency. Further, the LD (linkage disequilibrium) block structures are so variable across populations and the effect size for any given SNP is so nominal, that the likelihood we’ll ever be able to develop a panel of SNPs on a chip that will diagnose disease worldwide is a tall order.”
Despite his disappointment with WGA, Maraganore remains carefully optimistic: “We may one day get to the point where we can nominate and validate targets,” says Maraganore. Perhaps the WGA marker taking us there will not be a SNP. But if the ride will be taken on a SNP, researchers using WGA will certainly post a help wanted sign.
This article was published in Drug Discovery & Development magazine: Vol. 10, No. 9, September, 2007, pp. 16-20.
Filed Under: Drug Discovery