Genomic Research and Personalized Medicine: An Expert Interview With Francis Collins, MD, PhD

Francis S. Collins, MD, PhD

May 08, 2008

Editor's Note:
Medscape interviewed Dr. Francis Collins, Director of the National Human Genome Research Institute at the National Institutes of Health (NIH), about how close we are to achieving personalized medicine as well as its relevance to clinical practice.

Medscape: What are the most important developments in genomics that might affect clinical practice?

Dr. Collins: I think they fall into the categories of diagnostics, pharmacogenomics, and therapeutics.

In terms of diagnostics, we are already in a position with some heritable conditions where it's possible to determine who is at risk even before the disease has appeared, perhaps most dramatically in circumstances like the BRCA1 and BRCA2 genes. Women who carry mutations in one of those genes have a very high risk for breast and ovarian cancer, and there are now strong recommendations that women in such families ought to have the opportunity to undergo genetic testing in order to make decisions about surveillance or even surgical approaches to that high-risk situation.

Similarly, people in families with a very strong history of colon cancer can now take advantage of the kind of diagnostic testing that may identify individuals who have a risk as high as 60% for colon cancer and who can avoid a bad outcome by beginning colonoscopy at a very early age, and doing it every year instead of every 5 or 10 years. So those tests are with us now; they're not for some distant future time.

In addition, there is a lot of excitement now as we begin to identify genetic risk factors for an ever-longer list of diseases that includes diabetes and heart disease and the common cancers, all of which have yielded some very interesting clues in the last couple of years by taking advantage of a new strategy called genome-wide association studies. That has led to more than 100 new discoveries of genetic factors in common disease, all of which point to pathways that, for the most part, we did not anticipate were involved in those conditions. That makes it increasingly likely that we could offer to anybody who's interested the chance to find out their future risk for illness in a way that would empower them to practice better individualized prevention.

I think, though, that we still need better information about exactly what kind of interventions are appropriate, i.e., if somebody is found to be at increased risk for something like type 2 diabetes. Those data are being collected by research studies, and I think we can anticipate the availability of individualized, preemptive, preventive medicine becoming more and more possible for more and more individuals. And that should be a good thing. That's a focus on keeping people healthy instead of having our medical care system being more of a sick-care system than a healthcare system; we'd certainly like to make that transformation as much as we can.

The whole area of diagnostics associated with prevention is an exciting one. I should also say that diagnostics in the area of cancer are heavily dependent upon understanding the genome. Already, there are tests available to determine, say, for a woman with breast cancer, whether or not her tumor is likely to recur, in which case chemotherapy of an adjuvant sort is highly warranted; or whether there's a very low risk for that, and then chemotherapy could basically be skipped. That has been a very exciting development and has, I think, spared a lot of women from going through a really unpleasant and potentially dangerous experience.

You will see more of those kinds of diagnostic tests to assess what are the appropriate next steps to take for cancer, particularly now that we have The Cancer Genome Atlas project. This is a pilot effort to systematically and comprehensively characterize the genetic glitches in a long list of tumors in order to give us better information about how to prevent and cure cancers. So, that's the diagnostic side.

Then there's pharmacogenomics, the effort to try to make better predictions about getting the right drug at the right dose for the right person. Already in some instances, that is turning out to be extremely valuable. For example, in childhood leukemia, testing for a particular enzyme called TPMT allows you to be able to give the full dose of 6-mercaptopurine to produce a high likelihood of a cure of that leukemia without the fear of a toxic reaction. Such reactions sadly marred the experience for some kids who were very severely affected by that drug because they lacked the enzyme that normally metabolizes it. That is now preventable entirely by this kind of testing. Finding their way into the mainstream are a few other kinds of pharmacogenomic tests. For example, the commonly used drug warfarin, for blood clotting, now includes on the FDA label a suggestion that physicians be aware that genetic testing can help choose the proper maintenance dose of a drug that has a very narrow therapeutic window. Potentially in the next 2 or 3 years, that suggestion will become a recommendation based upon data that are being accumulated right now.

In regard to therapeutics, clearly this area has, in many ways, the greatest promise. However, it is also the part of the research enterprise that needs the longest lead time in order to go all the way through the process of identifying a promising therapeutic approach: testing it in an animal system, applying to the FDA for permission to conduct clinical trials, and then seeking long-term approval. That's often a many-years kind of pathway. But we are seeing that start to happen with an increasing number of discoveries that are based upon understanding the genomic basis of disease. Probably the best example is the drug imatinib for chronic myeloid leukemia, as well as drugs like gefitinib and erlotinib, that target therapy specifically to a protein that is only being made in the cancer cells and therefore ought to be less likely to cause side effects to the normal cells.

I believe that in the next 15 years the pharmacopoeia that we use for treating lots of diseases will be very heavily influenced by the things we're discovering right now about the molecular basis of disease. But that has the longest lead time, and so it won't happen overnight for many conditions. But it eventually will happen for Alzheimer's disease, for diabetes, for Parkinson's, for asthma and for hypertension. I think we will have new options coming out of these discoveries in the next decade or more.

Medscape: How can primary care and other physicians and caregivers keep abreast of important developments that might affect their practice?

Dr. Collins: That is a real challenge, because this field of genomic medicine is on an exponential curve, and yet many physicians have not had much in the way of training opportunities in genetics. It wasn't taught in their medical school class, it wasn't part of their residency, and they don't see a whole lot of it happening in their current CME exposures. Clearly, that needs to change as this gets more and more into the mainstream. After all, we do expect that in another 7 or 8 years, it will be possible for each of us to have our complete genome sequenced for $1000 or less, and there will be a strong interest in having that information determined and then placed in a private part of your medical record to inform decision making. Physicians will need to be prepared to know how to incorporate that, in a rational fashion, into prevention and treatment.

There are certainly outlets where this kind of information can be obtained. I would particularly point to an organization called the National Coalition for Health Professional Education in Genetics, or NCHPEG, which was founded by NHGRI together with the American Medical Association and the American Nurses Association. NCHPEG has as its mission the provision of user-friendly, case-based educational materials to busy practitioners -- that includes physicians, nurses, physician assistants, and nurse practitioners -- so that it is possible to quickly get up to speed on what the principles of genomic medicine are at that moment. Their Web site,, would be a place to go and look. Our own Web site for the Genome Institute, which is simply, has an increasing amount of information about genomic medicine for health providers. There is a particular site one can visit to learn about what genetic tests are available: It has a very useful listing of what kinds of tests are currently possible, both for rare diseases and common diseases, and what laboratories are offering them.

The Centers for Disease Control and Prevention (CDC) also has an ongoing effort called EGAPP, the Evaluation of Genomics Applications in Practice and Prevention. EGAPP is surveying, in an objective fashion, which of these various applications are ready for prime time and have demonstrated their utility in a research environment of a rigorous sort. That is also something to watch, and you can see more about that by going to the CDC Web site at

It will be important for all professional societies that representative practitioners join in this effort in a vigorous way. We need to make this kind of educational mandate a priority so that as decisions are made about CME programs, about the kinds of questions that ought to be mentioned in relicensure exams or certification exams, that this is not just some esoteric thing that happens in specialty clinics. This is going to be mainstream activity for primary care providers in the near future, and will increasingly, I think, be an empowerment for providers to be able to give better care.

Medscape: How close are we to having truly "personalized" medicine?

Dr. Collins: In a certain way, we've had personalized medicine for quite some time. If I need a blood transfusion, I sure want that unit of blood to be personalized for me and not just randomly taken out of the blood bank. So some of the things that we're very familiar with have required that kind of attention for a long time. A particular kind of personalized medicine that we've been aware of, but haven't utilized very effectively, could be one of the things to promote right now: namely, taking an effective 3-generation family history and incorporating that into recommendations about individualized preventive medicine. This is a problem, of course; busy practitioners who are given increasingly short time periods to interact with their patients have generally found it more and more difficult to spend the time taking such a pedigree and then trying to incorporate it into recommendations about healthcare. Often, even those who do this may find that the person sitting in front of them is a little unclear about their family medical history anyway.

One of the things that we have done to try to turn this around is to work with the United States Surgeon General to put together a free, Web-based family history tool. This tool can be used by patients at their own leisure, on their own time, to collect that family medical history from relatives. The tool, which is very user-friendly and private, produces a standard pedigree or a chart that can be printed out, and carried off to start a conversation with the healthcare provider about what in that family history might shed important light on things that should be monitored. That's a genetic test. It's a very reliable genetic test and it's even free; and yet we don't necessarily use it in many instances as effectively as we might. We could begin to personalize medicine right now by incorporating this kind of opportunity. That particular tool is available if you go to

So that's a start. And, of course, some of the things we've already talked about in terms of the ability for people to find out potentially high risk for certain kinds of cancers are certainly a form of personalized medicine that's here right now. But for people who don't have a strong hereditary predisposition to some illness, much of this is actually not quite here. However, it will be in the next few years, with the opportunity to find out individual risks for common diseases, with the pharmacogenomics revolution coming along, and with the chance to have complete genome sequencing.

It's fair to say that some of the enterprises that are out there in the private sector feel that it's time to start now. There's a large amount of direct-to-consumer advertising going on right now that will allow people who are interested to find out their genetic status at hundreds of thousands of variants in their DNA. Along with that will come some recommendations about what one's risk is for certain illnesses and even what to do about it.

This is going to be interesting because people who do decide to sign up for these efforts -- and of course you have to have your own resources to do so -- are going to bring those reports to their providers after they've stared at them a little bit and say, "What should I be doing about the fact that this test tells me my diabetes risk is 1.5 times the normal?" This will, I think, be an experience for healthcare providers who haven't tracked this very much to suddenly be put on the spot: How do you become an expert in genetics and a genetic counselor? Here again, we need to provide the resources for busy practitioners to be able to quickly get up to speed on that. This direct-to-consumer marketing effort may actually drive that process more quickly than many people are prepared for.

Medscape: What do you see as the major obstacles to achieving personalized medicine?

Dr. Collins: Many of the obstacles are basic scientific efforts that need to be conducted if we're really going to take advantage of this kind of individualized prevention. We need rigorous scientific studies to show that this information can, in fact, be incorporated into medical care in a way that improves outcomes. That is partly about making sure that we know what the quantitative risk associated with a particular variant is/ But it's also about assessing health behaviors in a way that represents the real world so that we can get a sense of whether people given this information do, in fact, take full advantage of it and modify their own plans for health maintenance in a way that improves the likelihood of staying healthy. This is going to require steps that we can't just skip over. In the pharmacogenetics /pharmacogenomics arena, we clearly need also to have prospective studies showing that information about DNA does, in fact, empower physicians to do a better job of prescribing and empowers patients to have a better outcome. We can't just assume, on the basis of retrospective studies, that this will be the case. Where I think we have a huge opportunity -- but also a huge challenge -- to take these discoveries about the basic molecular underpinnings of disease that are just now coming to light and turn them into truly novel approaches to therapeutics. It might really revolutionize the approach, in terms of both greater efficacy and lower toxicity.

On top of that, we have healthcare economics issues. Our system here in the United States is not particularly good at reimbursing for prevention. In fact, it's pretty lousy at that. If we're really serious about trying to incorporate this sort of individualized medicine into the mainstream, we really have to change our support systems for that so that people are not forced to pay out-of-pocket when, in fact, the system seems willing to pay for them once they fall sick. That seems to have things turned upside down, and we're going to have to face up to that if we're really serious about changing our healthcare system in favor of keeping people healthy.