Originally Published IVD Technology January/February 2006
IN PERSON
Pursuing a common goal
Pharmacogenomics presents challenges for the pharmaceutical and diagnostics industries, but also opportunities for the two to work together.
 |
| Michael Murphy is president and CEO of Gentris Corp. (Research Triangle Park, NC). He can be reached at murphym@gentris.com. |
FDA’s release last March of a final guidance document for pharmacogenomics data submissions sent a strong message. In codifying its thinking on the development and evaluation of personalized medicine both in this document and in a later drug-diagnostic codevelopment paper, the agency made clear its support for emerging genetic tests.
Today, pharmacogenomics is often spoken of in terms of its “promise”—a term that encompasses both optimism and caution regarding its future. While the excitement surrounding recent personalized-medicine breakthroughs is warranted, a number of obstacles to adoption exist. Formal cooperation between IVD and pharmaceutical companies is still embryonic, and many important questions surrounding test reimbursement remain unanswered.
To learn more about how pharmacogenomics can be more effectively integrated into drug development, IVD Technology editor Richard Park spoke with Michael Murphy, president and CEO of Gentris Corp. (Research Triangle Park, NC). In this interview, Murphy discusses the growing need to outsource clinical pharmacogenomics trials and the importance of early participation in the creation of personalized therapeutics. He also talks about how a changing drug development model offers opportunities for the IVD industry to forge a strong relationship with both FDA and pharma.
IVD Technology: What are the latest trends in clinical and diagnostic pharmacogenomics?
Michael Murphy: For clinical pharmacogenomics, in the area of drug development, we’re seeing a lot more interest from pharmaceutical companies in incorporating pharmacogenomics into their clinical trials. This is, I think, a direct result of the FDA guidance documents published in March 2005 that provided examples of when pharmacogenomics should be used, and when it must be used, to demonstrate safety and efficacy of drugs in development.
On the diagnostics front, we’re seeing a consolidation of the industry. This is occurring both with diagnostic testing providers and platform companies that enable the clinical diagnostics laboratories to actually test patients.
What other indications of interest in pharmacogenomics have you seen from drug companies? Have they been reaching out to diagnostics companies?
Not so much reaching out to diagnostics companies. I would say that, prior to the release of the guidance document, there was a great deal of uncertainty about how their data might be interpreted by FDA, and when and if they had to be included in the regulatory submissions.
I don’t think that these companies suddenly had a change of heart about the utility and usefulness of pharmacogenomics. It was more a question of knowing what data were considered exploratory versus validated biomarkers, and where the company stood with FDA. That’s important for feeling comfortable with a science like pharmacogenomics and embracing it fully in your trials.
The resistance that we see from pharmaceutical companies is from the marketing groups. There is a misconception among them that pharmacogenomics will limit the marketplace for their drugs. Despite this perception, not only have several drug companies gotten their drugs approved with pharmacogenomics indications on the label, but there have been examples in which pharmacogenomics enabled a drug to get through the approval process. In these cases, there was a subset of patients who were having issues with the drug safety or efficacy, and it was difficult to figure out or predict who they were.
How do drug companies go about employing pharmacogenomics in their clinical trials? Are they more inclined to formulate a strategy in-house or to work with diagnostics companies to develop sets for their trials?
There are still some companies that are doing this in-house. However, I believe they recognize that once their task involves validated biomarkers, which should be conducted under good laboratory practices (GLPs), they probably need to outsource that.
Pharma’s expertise, even within its R&D and discovery groups, is in discovering new medicine and finding new targets for that medicine. But the GLP-based lab business requires a completely different set of skills to be compliant with FDA and the Centers for Disease Control. There is a growing need to outsource clinical pharmacogenomics testing when it requires data that can withstand the regulatory oversight focused on FDA submissions.
I don’t think we’re doing a good job right now of aligning pharmaceutical and diagnostics companies. There are still two separate silos, for the most part. Pharma companies have their own interests and conferences, and the diagnostics companies have theirs.
How can these two perspectives be reconciled?
The battle within companies that need access to pharmacogenomics is whether they should build it or buy it. Although everybody tries to build these tests internally, they ultimately realize that doing so requires very specialized expertise.
Many of these tests are high-complexity, esoteric tests that require a high level of understanding in molecular biology and clinical pharmacology. There is also a large regulatory component—understanding FDA’s rules regarding GLP, good manufacturing practice, and good clinical practice. All of those skill sets need to be in one place to provide the quality products that pharmaceutical companies can depend on during the regulatory submission process.
Regulatory Guidelines
Earlier this year, FDA also released a preliminary drug-diagnostic codevelopment concept paper. What are your impressions of these guidelines and documents?
The guidelines are welcome news for everybody in the industry. They show a high level of understanding within FDA about pharmacogenomics and what it can or cannot do as far as supporting a drug submission.
The genomics data submission guidance issued in March goes a long way toward helping pharmaceutical companies participate in pharmacogenomics. The concept paper goes one step further in outlining the direction FDA is headed in. I think it’s no coincidence that FDA has recently opened up its Office of Combination Products. One of the limitations for companies producing drug test combinations had been not being able to go to a single group.
Are efforts being made at FDA to develop pharmacogenomics guidelines that are more centered around diagnostics?
I know there is an interest, but I’m not sure about the level of participation. When a product that is considered the first in its class is approved, it goes through a de novo process and eventually triggers a special controls guidance document. And there has been at least one—stemming from the approval of the AmpliChip test by Roche Molecular Systems (Pleasanton, CA). So I guess you could argue that FDA is working on it, because it’s being forced to by de novo product approvals.
But we need to encourage FDA to be even more proactive and to provide guidance to the industry about, specifically, the kinds of tests that have the clinical utility to make them worthwhile products. To give you an example, many of the first approved tests around drug-metabolizing enzymes simply would predict a patient’s genotype, but wouldn’t tell the doctor what to do with that information.
We know from our own marketing research that doctors act very much in a quantitative manner—that is, they are looking for a yes or no decision when they order a diagnostic test. Unfortunately, these early tests don’t address that issue.
We need to have guidance from the Center for Devices and Radiological Health on how we can come up with tests that allow doctors to make more-quantitative decisions. Telling them that a patient falls into one of four subsets for metabolism doesn’t help them make that kind of decision.
The guidance should not only address regulations but should also provide incentive for manufacturers to go through the rigorous process of development and approval.
Development Challenges
What barriers to developing tests for pharmacogenomic applications do IVD manufacturers encounter?
Despite the fact that there is a large subset of historical data around certain genotypes, there are really no specific indications about how you can reference previously approved drugs. We know that many antidepressants are metabolized principally by a single enzyme in the liver called CYP2D6. But there is little information to advise companies about producing and labeling a 2D6 test appropriately for doctors. So basically we’re in a situation where the technology has gotten ahead of many other aspects of the in vitro diagnostic marketplace.
In other words, there is no hot list from FDA about which previously approved products it would like to see diagnostics built around. If FDA came to diagnostics manufacturers and said that it would like us to produce a test that will allow doctors to avoid a specific drug, then we would have something to go on. But right now, the relationship is very general and open. And there is a risk you’ll have a product approved with no marketable indication.
Would creating an official hot list be a helpful tool for streamlining the whole pharmacogenomics development process?
Yes, I think so. There have been a few drugs that have been discussed by FDA in which diagnostics would be useful. Larry Lesko, the director of FDA’s Office of Clinical Pharmacology and Biopharmaceutics, has been extremely helpful to the industry in this regard, pointing out, for instance, the need for a UGT1A1 gene test for the irinotecan cancer drug, a CYP2C9 test for the anticoagulant warfarin, and another CYP2D6 test for the Eli Lilly drug Strattera for attention deficit/hyperactivity disorder.
But those are only three examples. I can look in the Physicians’ Desk Reference and find close to 40 or 50 drugs for which we’re talking about pharmacogenetics. So we still have a lot of catching up to do. The hot list doesn’t exist, but it can and it should.
What training and education issues regarding pharmacogenomics do IVD manufacturers have to overcome?
This is a large question and often discussed among people in the industry. From an IVD manufacturing standpoint, the two main issues that need to be overcome to advance pharmacogenomics are physician education and reimbursement. These are both normal challenges for any new product, but particularly with pharmacogenomics because it relates to a specific area like clinical pharmacology. And there is little training in medical school for physicians around clinical pharmacology, let alone for molecular diagnostics.
We need to be able to translate a genetic test into something called a predictive phenotype, that doctors can act on. In this way, we can help demystify pharmacogenomics for physicians. Right now, there is not enough down-to-earth educational material for them. A number of companies are focusing on this particular area, so I think this will be addressed in the next couple of years. But it’s a large challenge, and even more so for some smaller companies.
The other area is reimbursement. I believe it’s almost impossible to develop a stand-alone IVD—both a test and a platform—that could be sold at a price that allows reimbursement to cover the cost of performing the test.
Much of the testing that’s going on now is being done via home brews, using general-purpose agents and research-only products. This is allowed under the Clinical Laboratory Improvement Act of 1988 as long as labs complete the validation work behind those tests.
But again, one of the reasons labs would opt for that kind of test is that, while it may not be the highest-quality product, it can be produced at a cost at which reimbursement still allows them to make a reasonable profit. That is definitely not true for any approved tests we have right now, and none that I know are coming in the future.
How can the IVD industry work to tackle these issues?
What you’re asking is how we can get companies that are competitors to work together for a common cause. And this is always a hot-button topic. One of the issues that we’re looking at is the multitude of platforms available to incorporate these tests. Everybody, of course, wants the customers to use their particular platform. So even if we are trying to move through the American Medical Association to establish current procedural terminology codes, the question is what the real cost of these tests is.
One thing that companies could do is to get together as an industry and determine fair and reasonable prices for pharmacogenomics. Then, they can measure this against their own cost of doing business and see if the numbers match up. Clearly, we don’t expect the Centers for Medicare and Medicaid Services to reimburse right now at the rate of $500–$700. Something more reasonable would be in the range of $150–$200 per gene.
I don’t know of too many IVD manufacturers positioned to offer a test and still make a profit at that reimbursement level. It’s a difficult issue, and I think it’s going to be difficult to get everybody on the same page.
Partnering with Pharma
How can IVD companies enhance their market positioning and penetration through pharmacogenomics?
IVD manufacturers can help position their product by being involved as early as possible in the use of that pharmacogenomics test. And the earliest use of the test that I’m aware of is during drug development. Once a pharmaceutical company understands that there is an important link between their drug and a particular test, IVD manufacturers position themselves well to partner with them.
So far, this hasn’t happened to a large extent, although I think it will in the future. Identifying such links would allow IVD manufacturers to make the argument for a drug that cannot be prescribed in the absence of a test—for which the test is as important as the drug itself. An example of this is the Her-2/neu test developed by Dako Corp. (Carpinteria, CA) in collaboration with Genentech Inc. (South San Francisco, CA) for their breast cancer drug Herceptin. Dako has made good penetration in the marketplace because it partnered early on with Genentech. IVD manufacturers need to take an active role and not just operate on the sidelines.
How do IVD manufacturers go about creating these types of relationships? Are pharmaceutical companies becoming more interested in developing partnerships?
Pharmaceutical companies want to focus on what they do best, which is finding new chemical entities that they can develop and take through the drug development process. They certainly don’t want to be in the laboratory business and, with few exceptions, don’t want to be involved with the process of getting a diagnostic device approved through FDA.
Pharma can achieve that only through partnerships. Some of the early work that needs to be done is not available to IVD manufacturers. It’s not possible, for example, for a pharmaceutical company to outsource its testing under GLP to some manufacturers.
Instead, it needs to find companies that can do laboratory work and diagnostic development under one roof. Doing so will ensure a smooth transition from clinical trial testing to approval of a diagnostic device.
The Promise of Pharmacogenomics
In the past year, a few high-profile drugs have been taken off the market. A number of pharmaceutical companies have indicated that the drug-development model has changed—from producing blockbusters that can saturate the market to making more-targeted therapeutics. Do you believe that this is the direction in which the drug industry will actually move?
It has become obvious to pharma that the blockbuster model is no longer sustainable and may be even, to a point, dangerous. A financial reliance on one or two key products can develop. If something should go wrong—even a very rare adverse event that causes the death of a few patients—it could be enough to get the drug labeled with a black-box warning or pulled from the marketplace.
Herceptin, which works in only about 25–30% of breast cancer patients, was approved, in part, because it is linked to a specific test. I believe that there is a misconception that pharmacogenomics can ruin a potential blockbuster. A blockbuster drug, by definition, probably works in the majority of patients who take it. And so, for example, a response rate of even 60% for an approved drug is considered fairly good.
If a drug works in most patients, then pharmacogenomics can’t help it or hurt it. On the other hand, if a subset of patients don’t respond, or have adverse reactions, to a drug, then pharmacogenomics can be the difference between having perhaps three “minibusters” that generate $300,000 a year and killing the drug altogether and starting over.
How can IVD manufacturers get the point across that pharmacogenomics is something that is beneficial to healthcare?
The first thing is that manufacturers have to prove it. There’s been a bit too much hype with regard to the promise of pharmacogenomics and what it’s going to do for the healthcare drivers—FDA, pharmaceutical companies, the diagnostics industry, the laboratory industry, and health insurers. We need to demonstrate that.
Right now, the low-hanging fruit is oncology, in which the decrease in patient quality of life warrants the risk of using pharmacogenomics testing to mitigate potential drug toxicity. For example, there is a test for patients taking mercaptopurine and azathioprine for acute lymphocytic leukemia. This is not yet an approved test, but I know from my personal work with Bill Evans, the CEO of St. Jude Children’s Research Hospital (Memphis), that this test has gone from the research bench into the clinic. It’s now being used on thousands of patients every year to make sure that the 1 in 300 who has a life-threatening adverse reaction is either not given this drug or given a reduced dose.
We need to collect more examples like this. We know that 106,000 people die each year from adverse drug reactions. Many of these adverse reactions could be prevented by genetic tests that would identify the patients as low metabolizers of a particular drug.
The technology is out there. We just need to work together to prove that we can deliver these tests in a cost-effective way.
A View of the Future
Do you believe that pharmacogenomics is the wave of the future for the IVD industry? Will it finally make the era of personalized medicine a reality?
I think pharmacogenomics will have a significant impact on the future of the IVD industry. But I also think we need to be very cautious about our expectations. After all, genetics plays a fairly limited role in drug response when you consider the large effect of environmental issues.
So it’s not the panacea, if you will, for the IVD industry or for the pharmaceutical industry. But I do think it will open up new markets. It will provide high-quality tests that IVD manufacturers can be proud of, tests that will decrease severe drug reactions—the fourth-leading cause of death in the United States.
We need to be very careful not to overstate the promise of personalized medicine. I do not believe, for example, that we will have a panel of 10 or 20 medications that we can select from for every patient. I think it’s more limited than that. Patients typically have a much smaller variety of phenotypes.
What sort of time frame do you envision for this type of testing? Is 5 or 10 years realistic?
In the area of cancer treatment, I believe that in 5–10 years we’ll know much more about which drugs to avoid, but not necessarily which ones will be efficacious to the patient. I do think that we’ll have a handful of efficacy markers and that we’ll be able to better profile the tumor itself and understand how aggressive it is and which panel of drugs it will better respond to.
I don’t know that even five years from now we’ll be able to go conduct whole- genome scanning and have a complete answer for cancer treatment. But we should be able to better understand how not to make cancer patients who are already sick even sicker by giving them drugs that they can’t tolerate. And that’s a good place to start.
Copyright ©2006 IVD Technology
