FDA Policy on Pharmacogenomic Data in Drug Development

AuthorJanet Woodcock
PositionM.D., Deputy Commissioner for Operations
Pages91-102

Page 91

    M.D., Deputy Commissioner for Operations, U.S. Food and Drug Administration ("FDA") (formerly Director, Center for Drug Evaluation and Research ("CDER"), U.S. FDA). This work is based upon a live presentation made on February 5, 2004, and does not necessarily reflect events and changes thereafter.

My presentation will address policy and developing public policy in the area of pharmacogenomics. To a great extent, this discussion so far has remained within the drug and device industrial community, the scientific community, and the FDA. I appreciate this opportunity to reach a wider audience.

How is any new science or new technology integrated into existing regulatory, legal, and policy frameworks? We ask these questions in the context of clinical medicine, insurability, and payment, and we ask them each time a new science or technology emerges.

This discussion centers on the new science of phramacogenomics and how it will be integrated into drug development and clinical medicine. First, it is important to understand why pharmacogenomics matters and has to be integrated into drug regulation. The major barrier to having really effective drugs is the variability in the way people respond to drugs and the inability to predict how they are going to respond. These factors drive the cost of developing drugs and cause many adverse reactions to them.

There are two kinds of variability that must be considered. One is variable effectiveness of drugs. Leaving aside antibiotics and drugs that are actually not directed at people--those are actually directed at organisms that get into people--often the measurable effect seen in randomized trials in populations is small. Therefore, sponsors have to conduct large studies of effectiveness. Often, erroneous conclusions are drawn that a drug does not work or that its efficacy is insignificant because the response rate in the population is small. The response may be variable in fact, and some people may respond very well--though the collective response across the entire population studied may appear unimpressive. Unfortunately, at the present time, it is difficult to predict which patients will be responders. Page 92

This same is true on the other side--the toxicity side. All of today's drugs have some associated risks; if you look at a drug tested against a placebo, you see that every drug has a consistent pattern of side-effects over and above what the people on placebo experience. In fact, given enough data, you can identify consistent patterns of both common and rare side-effects. Now, some of these side-effects actually are due to the pharmacologic action of the drug, and there is no easy way to get around that because it is related to the beneficial effect of the drug. So, for example, certain drugs that you take for asthma are going to make you jittery. That is the same physiological effect that is opening up your airways. But a lot of the side effects are medically termed idiosyncratic. What idiosyncratic means in this context is: We do not know the cause, but there was a cause. We just do not know anything about it, so it is idiosyncratic. The way we approach drug toxicity in development is to expose many people to the drug and catalogue what we see versus placebo; it is all very observational. We say: This is liver toxicity, or this is a kind of organ toxicity, without understanding the mechanistic cause. So, ultimately, we may decide that people with preexisting liver disease should not take a drug because it has liver toxicity while, in fact, those two things may be completely unrelated.

Pharmacogenomics, is the science of correlating drug responses to genetic data--meaning the generation of gene or gene expression data that correlate genes and observed drug responses. Pharmacogenomics with a focus on gene sequences and data is pharmacogenetics.1 There are many different kinds of data about pharmacogenetics or genomics that correlate with drug responses.

One is simple polymorphisms in genes--for example, single changes in genes that impact the production of enzymes that metabolize drugs in your body. In some cases, if a group of people is given a drug, some will have the normal level of the drug in the blood, others have no observable presence of the drug in the blood, while others have a very high level. Of course, polymorphisms are relevant to dosing. Individuals without detectable blood levels will not have any drug effectiveness. Those with extremely high levels from a normal dose often will be dangerously exposed to toxicity. So, simply by knowing the relevant polymorphisms, one can predict the drug exposure that people have. Unfortunately, we do not have easy ways to test for those drug metabolizing genes at the moment. Page 93

There are also gene expression patterns. If you take a drug, your genes will start expressing different RNA in response to getting that drug. We know that, if it is a liver toxic drug, your liver will start making little toxic RNA response messages. If one were to look at those liver cells, one could actually say, "Look, that liver cell is experiencing some stress or toxicity." So we could predict, sometimes, toxicity based on gene expression. Another big problem in therapeutics, contributing to the variability of therapeutic response, is the fact that diseases are lumped together for treatment purposes.

At this time, medical practice is predicated on observation. For example, we still collectively categorize lung cancer as we did one hundred years ago. We still are not sophisticated. We don't know what the actual molecular cause of that particular cancer is in that particular person because we don't look for it. Gene expression patterns are giving us this opportunity, and there are some breast cancer therapies actually targeted toward whether or not one is expressing certain genes. We also perhaps could monitor and guide therapy based on observed gene expression patterns. Again, we may determine that a patient's liver is looking a little toxic and, therefore, we ought to back off a particular therapy and monitor for toxic responses.

So, in summary, this new science of pharmacogenomics holds the potential to help us better predict effectiveness and avoid toxicity. Pharmacogenomics is being applied extensively in drug development right now to pick candidates products to move into clinical testing. This science has the potential to revolutionize the process and really help people by individualizing therapy. Patients do not want to know if a particular drug is the best one for people with their ailment, they want to know, "Is this the best drug for me?" Right now, we seldom can answer that question with genetic precision; drug selection generally is based on the mean responses of the disease population as measured. Pharmacogenomics could revolutionize both drug development and treatment. Imagine the possibilities for narrowing down treatment to populations most likely to benefit and eliminating populations likely to suffer adverse events.

The primary policy problem right now is that most of these genetic tests are not being evaluated in clinical studies, and they are not being seen by the regulatory agencies. Application in the official drug development regulatory process is stymied by concern about how these tests will be used by the marketing application reviewers. This could present a real lost opportunity for any person who wants to take medicine in the foreseeable future. Page 94

So, we need...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT