Skip to main content
Premium Trial:

Request an Annual Quote

Q&A: Duke's Thomas Urban Discusses Pharmacogenomics' "Coming of Age" Moment


According to Duke University's Thomas Urban and David Goldstein, after fifty years of genetic and genomic research, pharmacogenomics and personalized medicine have finally come of age.

The two published a perspective in Science Translational Medicine last month calling the present time a "promising and exciting phase" in the PGx field and discussing what we might see in the future with a focus on the potential of drug trials to advance the discovery of genetic variations that influence therapeutic outcomes and vice versa.

According to the authors, changes are afoot both in how pharmacogenomics studies take advantage of clinical trial data to drive their discoveries, and in how the pharmaceutical industry in turn takes advantage of PGx.

Additionally, the two write that they see the future of PGx focused much more on pharmacodynamic variants — those that influence a drug's target or related pathways, or cause off-target effects like adverse responses — than on the pharmacokinetic, or drug metabolism mainstays of the earlier PGx era.

Urban, an assistant professor at Duke's Center for Human Genome Variation, discussed the report with PGx Reporter late last month. Below is an edited transcript of the interview.

The title of your perspective implies that we are in a coming of age moment for pharmacogenomics. Why did you feel that this is the moment to mark this coming of age, rather than a few years ago or a few years hence?

In my mind, it's a confluence of lots of different elements. For one, the technology has matured to the point where the whole genome is available to us for inquiry, and the methods — statistical and analytical — and the interpretation of genetic information has matured to the point that we are able to do things the right way in ways we were not in the past.

And I also think renewed interest in pharmacogenomics in the pharmaceutical industry, at least as I perceive it, is going to lead to a lot of opportunities for genetic discovery that might no have been possible in the past

What about your and Dr. Goldstein's histories in this field — what have you been working on and how did you bring that to bear in putting together this report?

The watershed moment for us was a few years ago in 2009 when we had the opportunity to work, with what was then the Schering-Plough research institute and was later acquired by Merck, on a clinical trial of dual therapy for patients with chronic hepatitis C infection.

In that study we were initially interested in variations in therapeutic response, and through a GWAS we were able to identify variant in a gene called IL28B that has a strong effect on whether patients responded to the drugs.

Shortly thereafter, using the same cohort, we were also able to find a strong impact of variants in a gene called ITPA and their effect on hemolytic anemia from the same therapy.

That, I think, was sort of eye opening for us, that the impact of these variants was easily in the range of their being clinically useful, and that they were discovered using what are now standard techniques for analyzing human genomes but with sample sizes only on the order of 1,000 patients, which is really on the small side when people think about GWAS for common diseases for example.

Going through the paper – in the very beginning you cite a handful of what you imply are mistaken explanations for why there have been relatively few practical examples of PGx implementation clinically thus far. Are there any you feel have been particularly wrongheaded?

It's long been my view, and I think [Dr. Goldstein's] view, that this idea that clinicians are reluctant to take advantage of genetic information doesn’t seem to really hold up when you look at examples, like Abacavir hypersensitivity, where there is a clearly clinically relevant relationship between genotype and outcome. Clinicians are eager to use that information immediately even in the absence of prospective randomized controlled studies.

Similarly, with the IL28B example, we found that clinicians treating hepatitis C patients were immediately eager to learn their patients IL28b genotypes prior to treatment.

In some instances, there is resistance on the part of clinicians [to using pharmacogenomics in their practice], but it's not an irrational resistance. It's because the some of the PGx information out there isn’t actually going to help them very much. The important thing for us to do is to try to find more examples where the clinical utility is really smack-you-in-the head obvious.

You also say you see the future of pharmacogenomics going more in the direction of pharmacodynamics, and being less focused on drug metabolism, which is a category where there have been maybe more of these examples of true associations without strong clinical utility to draw them into medical use. Can you tell me more about why you see this shift in focus as the field moves forward?

One reason is that for pharmacokinetics, or drug metabolism and transport, variants in these genes have already been studied thoroughly at this point. This was sort of the earliest entry into doing pharmacogenomics, based on the fact that what little we knew about drugs [in the early years] was how they are metabolized. So we could go after genes that metabolize them and look for genetic variants that influence that. For that reason, there is not a whole lot of investigation that remains to be done.

But I think the other thing to think about is the fact that, especially in the past ten or twenty years, there has been a systematic tendency in the pharmaceutical industry to find out whether a certain investigational drug is a substrate for a drug metabolizing enzyme that's functionally polymorphic in the population from the beginning, and if it is, that candidate has less of a chance of going forward in development. So, essentially, the selection on drugs in development that occurs because of metabolism basically limits the impact that metabolism can have on drugs that make it into the clinic.

On the subject of how pharmacogenomics is going to be integrated into the pharma pipelines of the future, you talk about the Blockbuster drug model, which companies have had some reluctance to abandon, but you say that as part of the coming of age of PGx, it's much more widely recognized now that the "drug that will work for everyone" is a myth. But I'm wondering where you feel we really are in that process, because for every PGx drug trial success story aren’t there still handfuls and handfuls of therapeutics being pursued the old fashioned way?

I think you're right that proceeding according to the blockbuster drug model will continue to occur. But I think in a lot of instances, in particular in instances where there are multiple therapeutic options in the same disease area and known variability among patients, that being able to show that you can identify patients more likely to respond to your drug as opposed to others would provide an added value.

Hopefully we will see more sponsors investing in genetic investigation during development.

Who is driving this? Is the drug development industry just being realistic about the benefit to them of doing things this way, or are physicians demanding genomically-informed prescribing models? Is it coming from elsewhere?

I think to a large extent, my perspective is that the pharmaceutical industry sort of got burned by pharmacogenomics in the past, during the so-called candidate gene era where there were basically a lot of "discoveries" made that appeared to be important but were in fact false positives.

I think the field of human genetics has, almost uniquely in the natural sciences, cleaned up its act in the last five years or so, to where false-positive claims are no longer acceptable, because we know so much better how to prevent that from occurring. Primarily it's due to accurate attention being paid to the multiple testing problem, and the ability to correct for population stratification— things that either were not done before or couldn’t be done before the GWAS era.

And now that geneticists can be trusted, I think it makes it easier [for pharmaceutical companies] to want to partner, or to invest in genetic studies.

Do you think that there is an ideal way to incorporate PGx into drug development? In the paper you make recommendations, for example, the idea of genotyping and exome-sequencing all participants in drug trials at the Phase II level. Are there a lot of ways to do this?

I think the model we outline in the paper is what I would consider right now, in January 2014, the ideal way to go about things. However, depending on the particular instance — on how much variation there is in response, or what your sample size is, that might change as you cost-benefit analysis for any particular project. In some case you might want to only do GWAS in Phase II and consider exome sequencing in Phase III for example.