Francis S. Collins, M.D., Ph.D.
Director, National Human Genome Research Institute
National Institutes of Health
"Consequences of the Human Genome Project for Medicine and Society"
November 1, 2001

Francis S. Collins: Before we get too carried away about what a momentous experience it is here to have the human sequence in front of us and how the world has changed. Let’s point out that in fact we are at a very early stage in genome mix or in understanding our instruction book. And I like the analogy a lot that where we currently are with Biology of the Human is where Mendeleyev was at the point that he put together this periodic table of the elements for my former scientific discipline of chemistry.

Actually it is not a bad parallel. We have our own periodic table now for human biology. It has perhaps 30,000 entries…a few more than the one for chemistry. And the 30,000 entries would be the human genes. It has isotopes which are the variations in those genes. Most genes have two or three common variants just like the periodic table for chemistry.

But, if you think about that analogy, I think it also makes the point that this is the beginning of understanding how this all works. The really interesting part of chemistry, after all, was the combinations of these atoms to make molecules and figuring out how all that worked and how chemical reactions occurred and so on. Well, likewise, for human biology, the really exciting part of this is to figure out how these genes make proteins that interact with each other in health and disease. A very complicated story indeed, that is going to be. Which will keep us busy for decades to come.

And so if there is anybody here in the student ranks who is a little worried that all the excitement of genomics had happened and that you missed it, don’t worry. We have sort of passed through the easy part and in some ways the boring part. And the exciting part of understanding really how this all works, of taking these basic facts about DNA sequences and turning them into real wisdom still lies ahead of us. And that is going to be a glorious opportunity for scientists over the coming decades.

If you are interested in playing a role in that, I am sure there are many faculty around this place that would be delighted to talk with you further about such opportunities and certainly I would encourage you if you are thinking along those lines, not to short change your training in the computational approaches. Because that is the way an awful lot of interesting observations are going to be made from here on with these huge data sets that really require very sophisticated algorithms in order to mine the nuggets of truth out of them.

Well, where is genomics going to go now? We have this draft of the human sequence. And I should be very straightforward here and say that it is not yet finished. We still have gaps in some of the sequences on some of the chromosomes that we are working very hard on to close. And by the spring of 2003 the genome really will be finished. We are on that track very clearly. Practically speaking, most of the questions people want to ask of the human genome can now be asked and answered quite nicely. The draft nature of the sequence isn’t getting in very many peoples’ way. But we don’t want to leave it there. This is after all, the book of life. Probably better not to have it be a draft any longer than necessary.

The things that we are now focusing very heavily on this next phase include quite a number of areas. And here they are displayed four of them. And I don’t have time to go through more than a couple of them. And I think in fact, I will focus particularly on this one of medical genomics. How do we apply all of this to an understanding of disease?

Well, again this gets me back to the point that while it is very interesting to determine the sequence of the human genome, that is a bit arbitrary. Whose genome anyway? As it turns out, it is the genome of an anonymous few people who live in the northeast who answered a newspaper ad and gave blood samples for this purpose whose identity we do not know and don’t want to know. But, after all, we are all 99.9 % the same at the DNA level, so maybe it doesn’t make a lot of difference. And it probably doesn’t make a lot of difference if what you are interested in is understanding where the genes are and something about their structure.

But, it makes a lot of difference to understand variation if what you care about is why I might be at risk for cancer and you might be at risk for heart disease. We have to understand variation on it’s own merits. So, there is another part of the genome project that is focused on that. And what we would like to do is to find the places where we differ if that is me, that might be you. There are 2000 letters of the code here and between two of us, and it wouldn’t matter which one of you I picked and what geographic area your ancestors came from, it would still be about .1% are the differences between us. By the way, that is a very interesting observation. That the variation in the human genome was largely there 100,000 years ago when there were 10,000 humans from which we are all descended living in Africa. So, when you find variations like the ones I am showing you here, it doesn’t really matter who you look in or what group you look in, you are probably going to find those same variants.

There are rare variants that may pop up in one group or another, but for the most part, this is shared variation across all groups. And the corollary of that is one cannot scientifically draw precise boundaries around particular groups and say they are different. And I think the corollary of that is that our concepts of ethnicity and race which are quite blurry when you start to really think about them, need to be blurry because scientifically there isn’t any justification for putting those in sharp terms. These are really cultural terms much more than they are scientific. And scientists are beginning to wrestle with how do we convey that message in a fashion that doesn’t actually make the mistake of endorsing such concepts as though they had scientific value when they do not.

Well, if you want to understand how variation contributes to a disease, the path that we are on is actually a pretty simple one. Once we get all the tools together and the technologies, this is what we are going to be doing for disease after disease, after disease over the next five to ten years. Collect a group of individuals who are affected with the disease. Let’s say it is diabetes. And a group of individuals who clearly aren’t. Study them clinically in a research protocol, very carefully so you know everything there is to know about their medical conditions. And then sample variants all across the genome. Not just for the genes that you guess might be responsible. But all of them.

The genome is a bounded set of information. It is in there somewhere. If you have a method that allows you to look at the whole thing, you are going to find an answer. So, here is the idea. You would test, then the affected individuals and the unaffected individuals for variants in all those 30,000 genes looking for one where there is a skew in the distribution.

Now here I have color-coated the variant to make it easier to see. But purple for gene A might be a T and green might be a C. And for gene A, that variant doesn’t seem to have anything to do with the disease because the proportion is the same between the affected and the unaffecteds. And that will usually be the case. Most of the genes aren’t going to be involved in predisposition to whatever disease it is you are studying. But, somewhere in that list of 30,000 you may find six or ten that look like gene B where there is a variant…here abbreviated as orange color which is over represented in the affecteds compared to the unaffecteds. That will be a profoundly useful observation because that tells you that the orange spelling of gene B in some way predisposes to this disease. And the changes are that gene B will be one of those you look at and go, gosh, I never would have thought that. Again, remember how ignorant we are so this is a way to shine a light on that ignorance and to help you find your way into the pathways that are really responsible.

This is going to be happening for most of the common diseases over the next decade. And the consequences for that for medicine are going to be very significant. So, let me now move to that. Whatever disease you are interested in unraveling, I think this strategy has a lot going for it. And what I just went over is going to suggest that over the next few years…and we have already done this for some conditions like breast cancer and colon cancer…we will identify the genes that are contributing to risk.

Now, that is all research up to that point. But after that point, it starts to be clinically applicable. You could after all then turn around and use that information to predict who is at risk even though they are not yet ill. And again, remember I suggested you might want to consider this for yourself. And realistically in a few years you will get asked. And most people when asked after they have thought about it a bit would tell you that they are interested particularly in knowing that information if there is something that they can do about it.

By the way, I could tell you right now who the people are in this room who are at highest risk of Alzheimer’s disease. We know how to do that. But that is not a test that is offered. And why is that? Because there is nothing currently that you can do if you are in the high-risk category. And extensive research has suggested that very few people want the information under those circumstances. I mean, you are at high risk for Alzheimer’s disease, there is nothing we can do about it, have a nice day. It doesn’t work for most people. They would rather not have that cloud over their head, waiting for the moment where they forget somebody’s name or drop their keys and wonder if this is the first sign of the disease.

So, that kind of diagnostics, unassociated with an intervention is probably all going to be of very limited value. Although there will still be some people who just want to know. And one could argue that they should have that chance. If on the one hand you have an intervention…a preventative medicine strategy, then people start to get a lot more interested. So, here is an example where that is already the case and this list will be growing rapidly.

There is a family with colon cancer. In fact, there are two siblings here who have had colon cancer in their fifties as you can see this person here and his sister over here…their mother had uterine cancer. A couple decades ago, I am not sure this family would have raised that much suspicion in a genetics clinic. Although probably they would have been advised that there is something going on here. But, this is now a circumstance where we understand in many of these families exactly what the problem is. And in this family, a particular misspelling has been identified in a known gene, a gene called MLH1. That makes it now possible to test the people in the family who are currently without disease but at high risk like these two children, the affected gentleman here, this daughter of the affected mother here, and these two siblings who don’t yet have cancer but are wondering about whether they are at risk. And in fact, of those various five people at risk, I know that two of them test positive. Now, is that something where you can do something? You bet. Because in that circumstance, colon cancer is a very slow process that begins as a benign polyp which can be seen through the colon scope and removed before it ever goes on to become an invasive cancer. So, these people at high risk need colonostopys starting at age thirty and done religiously every year. And the people who didn’t inherit that gene, don’t have to go through that because their risk is no higher than anybody else. Not to say that it is zero, but it doesn’t make them at higher risk and therefore appropriate for very intense scrutiny as if they had the positive test. So, this kind of thing is already happening but it is happening in families like this sort where this already a fairly strong history of a particular condition.

What is Pharmacogenomics? This is a term that was only invented about four or five years ago but now it is everywhere. Pharmacogenomics is the notion that we could not only predict risk of disease but we could predict likelihood of response to therapy by studying somebody’s DNA. The idea is that the variations in response to therapy may be largely genetically encoded. And so, if you could do a test first off, you might be able to say what is the right drug for this person instead of using a one-size fits all approach to an illness.

And there already are good examples where this seems to be looking pretty promising. This is for heart disease. What you are looking at here is the rate of progression of coronary artery disease in a group of individuals who have known heart disease. And first look at the yellow bars…what is being done here…these are being treated with placebos. So, there is no intervention in that group. But, what we have done is to break the group down by their genotype…that is their DNA spelling…at a particular gene called CETP which is involved in cholesterol metabolism. Now CETP like most genes that you have, you got two copies. The one you got from your mother, one from your father. The two different spellings here are being abbreviated B1 and B2.

The B1/B1 people have two copies of that spelling have the most rapid rate of progression if you don’t do anything. So, you can see you don’t want to be a high yellow bar in this diagram. That means your disease has progressed. Whereas, the people with two copies of B2 progress the most slowly and the people with one of each are in between. Now, look at the group that was treated with a standard drug, now very much on the market: Privastatin. This is not some drug of the future. This is a drug of the present. If you look to see what has happened to those treated, you can see that the B1/B1 people got a lot of benefit out of this. Their coronary artery disease was much slowed down whereas the B1/B2s got a little bit of benefit and the B2/B2s got no benefit at all.

This study is now in the process of being validated and I gather has been. And that means that you are not far away from seeing a recommendation made that before the physician prescribes Privastatin for somebody with coronary artery disease, you might want to know what their genotype is at CETP because if it is a B2/B2 individual, this is probably not the right therapy. And there are a long list of other drugs for which this kind of data is beginning to be accumulated. So, don’t be surprised if sometime in the next four or five years you are asked about a genetic test before the drug prescription is written.

Of course, where we really want to get to in this diagram and you notice the time is over here on the Y axis and we are moving from top to bottom, which means the stuff on the bottom is what we get to last. But, that is the stuff we most want to reach is the therapeutic consequences. Now, in some instances the gene itself will be the therapy. Gene therapy has had a pretty rocky road over the course of the last fifteen years. And I think it is fair to say that many of the early promises of the field were overblown and have not turned out to be correct. But, it is fair also to say that despite the tragedy of the young man who lost his life in a gene therapy trial a couple of years ago…and that was a great tragedy…there are encouraging signs in a couple of other trials that this approach is beginning to work. Notably for hemophilia and for certain type of immune deficiencies.

But, I think it is anybody’s guess where gene therapy will end up fitting in to the therapeutic armamentarium of a physician. And most of this is still going to be research for quite a while.

On the other hand, this other pathway over here where the gene gives you information that allows you to understand the biological defect and that allows you to design a drug that goes right to the heart of the problem, I suspect will be a more generalizable and more widely successful approach for many diseases where gene therapy presents major difficulties. And already we are beginning to see some success that have heartened people like me in terms of making the prediction that this strategy is going to work.

Perhaps the most heartening success of all is this one recently approved by the FDA a drug called Gleveck. Gleveck is used now for chronic myeloid leukemia. This is a type of leukemia where almost invariably, if you look at the malignant cells, they have a rearrangement of chromosome 9 and 22 where these two chromosomes have broken and then come back together again. And at the point where they joined, a gene called BCR which is on 22 and a gene called ABLE which is on 9 so they should be nowhere near each other, now all of the sudden, they are fused together. And that fusion gene makes a fusion protein which oddly enough is called BCRABLE and BCRABLE fusion protein has a binding site…this pocket right here which binds ATP and transfers a phosphate to a substrate which them starts a cascade that results in malignant transformation of those white cells which you will see in the blood smear with this type of leukemia has a very, over blown population of malignant white cells that are clogging up the circulation. And this is a terrible disease for which we have not had good therapies.

So, investigators, particularly Brian Durker at Oregon and people working at Novartis reasoned that if they could come up with a way of blocking an active site with a designer drug, they might be able to prevent this. So, they determined a three-dimensional structure of the proteins so they know exactly what the pocket was that they had to fit into and they built this thing called Glevack with good organic chemistry and made it fit into this…into the appropriate place. And then they gave this drug to thirty-two patients who had far advanced CML not expected to survive more than a few months. They had failed other forms of therapy. And in that first phase one trial, thirty-one of those thirty-two patients went into remission.

That is a dramatic result, which almost never happens in the first trial of a drug and quickly led through then more extensive trials and to approval of the drug by the FDA. One hopes to see many more examples where this genetic approach to developing new therapies ends up giving you something that is much more precise and more effective than what we might have arrived at by more empirical means.

In fact, this pipeline is getting pretty interesting. This is indication of just how many drugs like Gleveck are now coming through the pipeline. Over on the left here are the various types of targets named here in the middle and how many of them there are for each of these classes of molecular targets that cancer researchers have discovered over the course of the past twenty years. And over on the right here is an enumeration of how many different agents are now on trials either through the NCI or in Pharmaceutical companies and you can see it is pretty impressive. There are over a hundred cancer drugs now in clinical trials that are based upon a molecular understanding of the disease. And that compares with a handful a few years ago.

Well, I have gone on a bit here about the medical implications. And I hope you get a sense of the excitement of this because I think it is appropriate to be excited. But, there are also ethical, legal and social issues that need to be raised. And certainly in a place like this where there is a strong program in that particular area with people like Jim Childress, I don’t know that I need to necessarily introduce these topics to you. I suspect many of you have already thought about them.

But, let me just briefly enumerate what some of the issues are that I think are of most concern. From where I stand, particular concern has to be attached to this one. And it is certainly one in which numerous polls would indicate that the American public is worried about. Would it be safe for you to have that genetic test and find out that you are at risk for colon cancer as that family did, and not run the risk of losing your health insurance the next day or having your employer decide that you are no longer a good risk for a promotion because you might get sick? With that kind of discrimination on predictive genetic information is really unjust and unworkable. And it is something that threatens all of us. Remember we are all at risk for something. And so this is not about somebody else. This could be about you.

And basically the solution which has been proposed by virtually all of the ethicists and scholars who have looked at this is that we ought to take that information off the table when it comes to health insurance and the workplace. And there are now several bills in the United States Congress that aim to do that. But, they have not yet received sufficient attention and acquired enough momentum that I could tell you they are going to pass in the near future.

It is gratifying to see that both parties in both houses seem to agree on this. This is not a partisan issue but it has not reached high enough place on the priority list to actually have something done. This is, I think, something where Mr. Jefferson would have had something to say. Our laws and institutions, said Thomas Jefferson, must go hand in hand with progress of the human mind. Here is a golden opportunity to put that into practice.

We have a big educational challenge. If genetics is going to become mainstream in medicine which I think it will, most physicians are not yet ready for that. Perhaps those in medical school now are getting trained but those that are out in practice already may not have had much exposure at all to genetics. And the public is similarly confused about many of these issues.

We have to think about access. If this is high technology, will it be available to everybody or only to those with high levels of education and resources. That has to be a serious issue to deal with now and not later.

Will all of our learning about human variation in the effort to understand risks of disease shed a useful light on the issues of ethnicity and race and reduce human prejudice? Or will there be ways in which people misuse the data in order to try to under gird those prejudices? That also requires a lot of attention.

And perhaps hardest of all will we be able to set boundaries? Will there be places where we decide that this technology really should not go? And particularly people raise the question of the designer baby scenario where couples with lots of resources will take advantage of genetic technology to try to optimize the genes of their next generation. Is that something that we would consider a good use or a misuse of genetic technology? I think many people are uneasy about that scenario, me included.

Now, most of the scenarios that people have put forward in serious articles or in Hollywood movies like Attica are pretty unrealistic because they assume a degree of genetic determinism for things like intelligence or athletic ability or physical attractiveness that none of those features really deserve. Those are all things in which the environment is incredibly important and there is a lot of subjectivity as well.

But there are some serious issues here in terms of exactly what do we as a society want as far as setting boundaries in an area where we usually leave it up to couples to decide what to do about their own reproduction decisions. And I don’t think we have even begun to tackle that very difficult problem.

Well, if you will bear with me for one more minute here, I would like to do what I said in terms of making this risky projection about where we are going. And I will start in 2010, which is already pretty far off for a field that moves as quickly as this. So, don’t take these things too literally but it is my best shot here at guessing where we will be.

I think by 2010 we will have predictive genetic tests available for a dozen or so conditions. So, this hypothetical of my asking you do you want to know won’t be hypothetical anymore. And for several of those, there will be interventions available to reduce risk and I suspect those will be of considerable interest to lots of you. Pharmacogenomics where you get the genotype done before you write the prescription will be the standard of care for several drugs. I couldn’t tell you which ones. And there will be controversy. Will access be inequitable? Will the disparities between social-economic status groups and ethnic groups persist? Will we solve the discrimination issues with effective federal legislation or will we allow this to linger on? I certainly hope they will solve them long before 2010.

Okay, another ten years. I think the therapeutic consequences of genetics by 2020 will be all around us. Already even the case of Glevack, this is starting. But, it is going to be slower for some diseases than others. Nonetheless, by 2020, I think we will have a whole new list of designer drugs for lots of common diseases. And that will be good.

Gene therapy will be the standard of care for some conditions. I don’t know which ones. And probably, if you want to know, your genome can be sequenced not just a base pair here and there, but the whole blooming thing for about $5,000.00 by 2020. And probably many of us will have that done and we will carry that information around with us in some kind of storage medium hopefully in a fashion that is protected from various prying eyes.

But, more controversy will be present. There will be an intense debate which I am arguing need not wait until then, it should be discussed much sooner than that about the non-medical uses of genetics. And particularly the choices of characteristics that are more traits than they are risks of disease.

And I can’t imagine all this will happen without there being some backlash and some resistance and folks crying this isn’t natural. And we really should take technology out of this equation and go back to a simpler time. Look at the way these arguments are being raised about genetically modified foods. Many of them by very sincere people.

I think these are discussion we need to have. The whole purpose of the LC program…the ethical, legal and social issues program of the genome project was to try to have those discussions before there was a crisis. And I think that experiment which is what the LC program is where we have put 5% of our budget into these issues has been a good one. But, I can’t tell you how exactly it is going to turn out.

I would argue the best way to prepare for these intense debates about whether technology has started to run us or the other way around, is an educated public. And so that claims can be assessed for their accuracy and their reliability.

Well, I will take you one more ten-year period, although this is clearly getting way out on that limb. In 2030, health care will have, I think, a genomic’s basis in many parts of what is done. Not all. Don’t let me over state this case completely. Preventive and therapeutic strategies will be available for most diseases. We will have ways of detecting the beginnings of illness even before symptoms have appeared using various techniques that look at for instance, gene expression as having slipped out of balance even before you have noticed it yourself.

But, there will be other debates going on. If all this works, we are going to live longer. Social Security is going to have a harder time staying solvent which it is already having a hard enough time. I guess this is a good problem to have if it is for this reason. And I think at the more philosophical level, there will be those who will begin to argue that if we develop the skills…skills which we don’t yet quite have but might by 2030, to actually go in and change the human germ line…the part of DNA that gets passed to the next generation, should we limit the applications of that to the treatment of disease or should we re-engineer ourselves? Should we basically say that evolution got us so far, we can do better and see whether over the course of a very short time we could develop the human race into something even more intelligent and capable than what we currently see?

That particular scenario scares the Be Jesus out of me because it implies that somebody would know what an improvement was and that is obviously going to be a bit subjective. And it also implies that we would be so sure of the safety issues that we could undertake such a thing without concern that generations later we might discover we had done something wrong and then it would be a bit late to change. So, I personally hope that is something that we will not do. Although there is certainly people out there already beginning to advocate for some consideration for that.

Well, let me close with a quotation. It is not from a scientist, it is from a poet. It is probably familiar words to many of you because you have probably seen them. It is from T.S. Eliot and the Four Quartets. It is not, I suspect, written about the study of genomics, but it sounds like it could have been.

He says, "We shall not cease from exploration. In the end of all of our exploring will be to arrive where we started and know the place for the first time." I think that is very much the adventure that we are now on. This points out to you that we are still exploring and we are going to be exploring a long time. But, the place that we are exploring is pretty interesting. It is ourselves. Thank you all very much.