Name: Towia Libermann
Position: Associate Professor of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, 2000 the present
Director of the Genomics Center, Beth Israel Deaconess Medical Center and Harvard Medical School, 1999 the present
Director, Dana Farber/Harvard Cancer Center Cancer Proteomics Core, 2004 the present
Background: Founder, Tolerance Pharmaceuticals, 1999
Education: MS, Biochemistry Federal Institute of Technology, Zurich, Switzerland, 1980
PhD, Immunology, The Weizmann Institute of Science and Technology, Rehovot, Israel, 1986
At this week's IBC Early Efficacy Assessment, Towia Libermann was the chair of the session "Using Clinical Samples to Inform Preclinical Development," and his talk centered on the ongoing effort to revamp pharma for the genomic era. Pharmacogenomics Reporter spoke to him this week to glean a few of the thoughts he shared at the conference about "Integration of Functional Genomics and Proteomics Approaches Into Drug Development and Clinical Trial Design."
In your opinion, do you feel drug developers are using genomics and proteomics more and more in clinical trials?
Until recently, they probably didn't want to use them. They would obviously be happier with a drug that is being utilized by 100 percent of patients for a certain disease even if not everybody responds than to have to stratify and only maybe at the end be able to target 30 or 40 percent of the population. The big pharma companies are not really interested in that.
It's going to shift from pressure from the FDA and the public. They can't afford [for much longer] the blockbuster drugs that will be withdrawn because of people suffering adverse reactions, having lawsuits and everything else. In the long run, it definitely will change. At least you can see it nowadays in all the conferences when you see CEOs from various pharmaceutical companies, they're all talking about needing biomarkers for their drugs in order to see efficacy, prevent adverse reactions, and so on.
I think they all have recognized that they have to get into it. It's obviously something that adds additional costs and time, because you cannot immediately know who or how people will respond, and many times cannot really see it up front until you've done the clinical trial. So, many of these things, unless you know how to stratify up front, you probably have to do trials without stratification, and revisit it later and based on the analysis [by genotyping, proteomics, et cetera] identify why some patients responded one way and others another way. And then go back to the next trials.
Has the FDA pharmacogenomics guidelines changed pharma's opinion of these kinds of technologies?
I'm pretty sure it has changed. A lot of them are thinking about it and trying to implement it, but it's not really so clear from practical terms how to do it. [The CYP450 SNPs,] for example, are a small part of the SNPs that are involved in response to drugs, and if you want to use [all the possibly relevant SNPS] that are known, and trying to figure out which of them and what combination of them is responsible for certain reactions. That's a major undertaking, and no one knows how to do it from a practical point of view and such.
Can you tell me a little about your idea of how functional genomics and proteomics fit into drug development and clinical trial design?
So the idea is that right now, the way drug development is going, you have a target for a drug, then you develop the drug, and then you test the drug on some animal models, and at some point you go into clinical trials. And then the trial fails or doesn't fail, but even if it [succeeds], you have many times late-stage failures, as you've seen with Vioxx and Celebrex and other drugs going off the market even after they've been approved due to side effects.
The problem is that these drugs go off the market completely, even so, there's a high likelihood that only a small percentage of patients will actually have this adverse reaction to drugs. But because there's no good up-front stratification, it's very hard to know who will be a responder, a non-responder, and who will have adverse reactions.
So the idea, and the hypothesis, is that if you do much more legwork up front, you might be able to predict which subgroup of patients would be the best ones to target with particular drugs. And even with regard to the clinical trials, you might be able to design the trial more specifically, so that even the outcome of a clinical trial might be more positive, because many times the way clinical trials are designed they're often not optimally designed. The way groups of patients are mixed and so on, when you look overall at the data, you really don't see any efficacy of drugs, but if you really go into subpopulations, you would see a benefit. But that gets lost many times in the clinical trial, based on the trial design.
Is there any chance of reviving the cox-2 inhibitors?
I have no doubt about it, [they] will come back. First of all, I think several applications of cox-2 inhibitors one is for all these inflammatory diseases, arthritis, et cetera and on the other hand, there were a lot of trials going on with regard to the use of cox-2 inhibitors as anti-cancer agents. And several of these NCI trials halted due to the adverse reactions [on the market]. But again, if you are for example a cancer patient in the terminal stage, and you know you will have a two- to four-fold increase in risk for cardiovascular side effects, and you have a 100 percent chance of dying from cancer, I think you would probably be willing to take the risk.
And so for cancer, I'm sure [cox-2 inhibitors] will come back, just because there is no doubt that cox-2 inhibitors have strong anti-cancer effects, and seem to show some significant benefit for certain types of cancers. So again, you might want to know which patients are at highest risk for cardiovascular side effects, but once you know that, you should definitely treat the rest. With regard to arthritis and other inflammatory diseases where cox-2 inhibitors are being used, I'm pretty sure it will again be only a subpopulation. We don't know [how large the population is] at risk, but I have no doubt that there is a high percentage that are not at risk, and you need to figure that out.
For that you obviously need a relatively large study, but depending on how clinical trials have been designed, maybe the companies already have the samples to start analyzing that. If you know who responded and [who] didn't, if you collect serum and genomic DNA, et cetera, so that you could now start to do some genotyping retrospectively, you might be able to see why this population responded with an adverse reaction.