New Map Changes Search for Disease Genes
IRA FLATOW, host:
This is TALK OF THE NATION/SCIENCE FRIDAY. I'm Ira Flatow.
A bit later in the hour, we're going to be talking trash. But first, we're going to be talking DNA. And, you know, when it comes to our DNA, we humans are 99.9 percent the same. It's that remaining .1 percent--.1 percent is what makes us different from one another and what makes some of us more susceptible to certain diseases, respond differently to medicines--you know, people say, you know, `I only need a little bit of that medicine. It throws me off if I have to take a whole lot.' Well, it's that .1 percent that makes us all different.
And this week, an international team of scientists announced that they've completed a map of some of those differences. It's called a HapMap, H-A-P M-A-P, and it's essentially a huge catalog of patterns of genetic variations across different populations. And scientists are hoping that having this information will help them tease out the genes responsible for some common diseases such as diabetes, heart disease or Alzheimer's, and to help them tailor drug treatments to individuals. If we each have a little bit of a genetic difference, maybe we won't respond. You know, we don't all respond the same way to drugs. Maybe we'll have tailor-made drugs using the HapMap.
So we're going to start the hour talking with one of the lead researchers of the HapMap Project--their work appears in this week's issue of the journal Nature--and also with a scientist who is using this data to determine why people respond differently to anti-epileptic drugs. And if you'd like to get in on the conversation, you're more than welcome. Our number: 1 (800) 989-8255; 1 (800) 989-TALK. And, as always, you can surf over to our Web site at sciencefriday.com for more information.
Dr. David Altshuler is director of the Program in Medical and Population Genetics at the Board Institute of Harvard--at the Broad Institute of Harvard--excuse me--and MIT, and an associate professor of genetics and medicine at Harvard Medical School and Mass. General Hospital in Boston. He joins us from our member station WBUR in Boston.
Welcome to the program, Dr. Altshuler.
Dr. DAVID ALTSHULER (Director, Program in Medical and Population Genetics, Broad Institute): Thank you for having me.
FLATOW: You're welcome.
Dr. David Goldstein is the director of the Center for Population Genomics & Pharmacogenetics at the Institute for Genome Sciences & Policy at Duke University in Durham, North Carolina. He joins us today from the campus there.
Welcome to the program, Dr. Goldstein.
Dr. DAVID GOLSTEIN (Director, Center for Population Genomics & Pharmacogenetics, Duke University): Good to be with you.
FLATOW: Thank you.
Dr. Altshuler, I think we need a little bit of--how shall I call it?--biology 101 here. Can you give us an i--what are we talking about when we talk about the HapMap? How is it different from the map of the entire genome?
Dr. ALTSHULER: So, as you know, and people may have noticed, a number of years ago people described the sequencing of the human genome. And, in essence, you might have asked the question, `Well, whose genome is that?' Well, it turns out that since we are, as you already said, 99.9 percent identical, any two copies of the genome, it didn't matter so much to get a first look whose it was. And that was, essentially, somebody's genome. It turned out not the same person's in all places; a bit of a patchwork quilt. But we had a sequence of the human genome.
Now inheritance, the differences between us that run in families, that's explained not by the places we're all the same, but in the places that we vary between individuals. And so what this is a map of is of the common variations between individuals--if you will, what you inherited from your mom and your dad that was slightly different than what someone else might have had, and that might influence why in your family particular diseases or characteristics are found.
FLATOW: Do these differences tend to clump up in any certain place on the genome, or are they spread around?
Dr. ALTSHULER: Well, it's a very good question, a big part of the motivation of the project. It turns out that these DNA variations often travel together, as you put it, in clumps, which scientists call haplotypes. The relevance of this is actually fairly simple. If it turns out that there are, say, 10 genetic variations in a row that are always seen traveling together in the population--and that, in fact, turns out to be the case--then if we could measure just one of them, we'd get the information or a very good proxy of it for the other 10. And this increases the efficiency of doing the studies. It makes things more powerful than it might otherwise be, and in some sense, that was a motivation for the project, although I think we've all come to think of it as a more fundamental tool, simply, that if we want to correlate the genetic variations in our population with the medical outcomes, we need a map of those variations. We need an inventory. And that's what this really is.
FLATOW: Mm-hmm. Were you surprised to see that they were clumped into these HapMaps?
Dr. ALTSHULER: Well, it's actually an idea that goes back a long way in genetics. It goes back decades, actually. But early theoretical suggestions were that the clumping, if you will, would not be so significant as it would be, that helpful; it turns out empirical data, when we actually looked--and this is something that emerged over the last five years. There is more extensive clumping, which means more helpful. And so I think that--I wouldn't say that we were surprised; I would say we were gratified, because it was one of the motivations of the project. And, as we document in the paper, it turns out to be quite extensive.
FLATOW: Now let's talk about that 1/10th of 1 percent. And what exactly are you looking for? What kind of changes, and where do you find them in the DNA?
Dr. ALTSHULER: Well, it's a good question. Much of the variation people have focused on recently comes in the form of single-letter differences between copies of the genome. If you imagine that the genome sequence is a code, it has a series of letters, it's simply that one letter is swapped. Someone might have an A at a certain position--that's a code that we use in DNA. We talk about A's and T's and C's and G's. One person might have an A and someone else has a T. And so those make up a lot of the variation, and that's what the map consists of. It turns out there are, as in any text that might vary, more significant changes--a whole chapter moved to a different place or flipped on its end--and those are also important, but this map is largely focusing on those single-base differences.
Do they have a name that you call it, these little differences?
Dr. ALTSHULER: Well, they're called--you're--they're called SNPs, which stands for single nucleotide polymorphisms, but we refer to them as SNPs in the vernacular.
FLATOW: Fits right in there, I think.
Dr. ALTSHULER: Exactly.
FLATOW: The SNP. Dr. Goldstein, how will this HapMap change the way scientists hunt for disease genes? I know your--you concentrate on anti-epileptic drugs, correct?
Dr. GOLDSTEIN: That's right. I think that the biggest change is it'll make the work that we do a whole lot easier, a whole lot faster and a whole lot less expensive. So because these variant sites cluster together to a degree, you don't have to directly interrogate all of them in the laboratory. But the difficulty has been in the past that, in order to know how to make use of these associations among the variant sites, you have to know the pattern of association in advance.
And so to give an idea of just how much time it saves to have that information instead of having to determine it yourself when you do a study, about four years ago, we started looking at a gene that is responsible for making the target of some anti-epileptic drugs. And we wanted to try to relate genetic variation in that gene to how patients respond to the medicines. But the first thing we had to do was to look at what variation is in the gene, find out the variation that's there, and then look at how the different variant sites associate with one another so that we can represent variation in the gene.
That process took us two years, took my laboratory two years. It now takes literally five minutes at the press of several buttons, once you know which buttons, at the computer connected to the HapMap Web site. So they did all of that background work, which allows you then to find an efficient way to represent variation in a gene like that, which you can then go and ask, `Well, does it influence the way the patients respond to treatment?'
And to sort of illustrate the change in scale that's afforded by what HapMap has done, just a couple of years ago the studies that we were doing to look at how genetic variation influences response to medicines, we looked at three or four genes at a time. Our first study for two important anti-epileptic drugs looked at three genes. Right now, my lab is looking at 400 genes, making use of the data provided by the HapMap Project. So this work really has been accelerated dramatically...
Dr. GOLDSTEIN: ...by what the HapMap Project has provided.
FLATOW: Could any genetic-based disease be looked at in this way?
Dr. GOLDSTEIN: Any genetically influenced disease could be looked at in this way. What we don't know is whether we will find anything important. So what the HapMap provides is a remarkably valuable tool for relating, as in fact David already said, common genetic variation to traits that we're interested in, principally traits of clinical relevance, but we could also look at other traits, such as just normal height variation, weight...
Dr. GOLDSTEIN: \..variation in memory and things like that.
Dr. GOLDSTEIN: To the extent that the genetic differences among people that are important to variation in the trait that we are looking at are common, the HapMap data and the general approach that it embodies will definitely be valuable for finding those variants. For those cases where the genetic differences among people are due to more rare variants--so, for example, a variant that you only find maybe once out of every hundred people or out of every 200 people--those variants will be much more difficult for us to find.
FLATOW: Yeah. Dr. Altshuler, do you have to add more people to the database here to make it more accurate or to find other things?
Dr. ALTSHULER: Well, it's an interesting question. I want to touch briefly on what Dr. Goldstein was just saying, which is that there is, we know, in every individual a set of variations that are common, meaning that if you look at other people you'd see them again, and also variations that are rare; that is, more unique to that individual. And it's very natural, I think, to start with the questions that are most powerful and easy to answer at the present time. The common variations are more straightforward to look at. They're easier to identify, and it's much more straightforward to make sense of them, because if you see them many times in a study, you can get a good estimate of whether they're tracking with disease. Things that are rare are harder to make sense of. They might be very important. They certainly are very important. It's a matter of a stepwise approach.
And so I think that--you asked the question, `Will HapMap need to have more people added?' And I think the answer is, in time, absolutely, because we will, I think, need in the fullness of time to study all the genetic variation if we hope to and are motivated to understand the inherited basis of clinical traits. That doesn't in any way mean that we shouldn't start with that which we can do, and as my colleague was making clear, the tools are rapidly--have rapidly developed to allow us to ask about the common variation, and there's no doubt that, over time, we will move on to the harder and harder questions.
FLATOW: Mm-hmm. We have to take a short break. I don't want to get into another question because we're going to have to break away, and this is very interesting. So let me just remind everybody I'm talking with Dr. David Altshuler, director of the Program in Medical and Population Genetics at the Broad Institute of Harvard and MIT, and David Goldstein, director of the Center for Population Genomics & Pharmacogenetics at the Institute for Genome Science & Policy at Duke. As I say, we are going to take a short break. We'll come back, take some calls, if you have them, on the phone, talk more about where the HapMap may be taking us in the future, and our number: 1 (800) 989-8255; 1 (800) 989-TALK. We'll see you on the other side of the break.
I'm Ira Flatow, and this is TALK OF THE NATION/SCIENCE FRIDAY from NPR News.
(Soundbite of music)
FLATOW: You're listening to TALK OF THE NATION/SCIENCE FRIDAY. I'm Ira Flatow, and we're talking this hour about genetics with Dr. David Goldstein and Dr. David Altshuler. Our number, 1 (800) 989-8255. Talking about the publication of the HapMap.
Every time doctors that--I interview you folks and people who are talking about the human genome, I always ask about the so-called `junk' DNA. Did you find anything new about--I always think there's got to be a reason for it there. I mean, is--did you find anything new about the junk DNA? Maybe it's not so junk-filled?
Dr. ALTSHULER: The--I'll offer an answer and Dr. Goldstein might want to add to it. One thing that motivated this project, actually, was that earlier approaches to human genetics often focused on the 1 percent of the genome that was what encodes proteins, which was thought traditionally to be the functional parts of our genomes. And one of the motivations for this map was to look broadly, in an unbiased way, not assuming that that's where the important actors would be found. There are certainly examples emerging where pieces of DNA that are not encoding proteins are important because they turn genes on and off.
One of the striking findings in the current study, which certainly needs to be validated and extended, but if it's right will be quite interesting, is we looked for places in the genome that mattered to natural selection, that mattered to the evolution of our species. It turns out there's a way you can look at this data to get clues about that. And one of the striking findings was in a part of the genome; it had a signal in it that suggested this was really important to human evolution, and when we look at it, using all the tools of modern DNA sequence analysis, there's no hint of a gene there. And so we say--we scratch our heads and we say, `How is it that this piece of DNA mattered to our--the evolution of our species, but it doesn't have any of the hallmarks that we think of as functional?'
Well, if it turns out that it was important to evolution, with more study, what that'll tell us is there are segments of our DNA that do things that we can't tell, which means we're pretty ignorant, which is--I think where a lot of this work started, was assuming we're pretty ignorant.
FLATOW: That's not too broad an assumption. Dr. Goldstein, anything to add?
Dr. GOLDSTEIN: Yes. I think that, while the primary motivation, as we discussed earlier, for the project was to look at these patterns of association among variant sites, there's a lot of other things that you get out of what the HapMap Project has done. And one of the things is that we can start to look at all of these common genetic variant sites, and there's about 10 million of them, and we can try to make some assessments of which of them appear to be in the most important regions of our genome. And while it may not be such a clean division between parts of our genome that do things and parts that are junk, and it probably is more of a continuum, it is certainly the case that some regions are clearly pretty important and other regions less clearly important.
And some of the data that the HapMap Project has generated will help us to divide all of these variant places in our genome up into the apparently more important ones and the apparently less important ones, and allow us to concentrate more of our energies on the apparently more important ones and try to understand how those might be influencing human health. And one approach is, as David just referred to, asking questions about whether there's evidence of natural selection.
FLATOW: Is this--down the road, is this the pathway to personalized medicine by each of us having our own map of our genome and a HapMap that we might, you know, have on file someplace?
Dr. GOLDSTEIN: That's my guess, that the nearest-term return that we're going to see from these new data will be contributions in the personalization of medicines. I think that's because, while I expect that we will make progress with these sorts of approaches in understanding the genetic basis of disease directly, I have a strong suspicion that understanding why we respond to medicines differently is going to be an easier task. It will be more often the case that you have common genetic differences among people that have major impacts on how they respond to medicine. And I think that this is going to be a tool that really helps us track down those changes, differences among people, relatively quickly, and then the next job will be to try to understand how to make use of that information to optimize therapy for individual genetic makeup.
And to illustrate that this looks like it should be possible, in most cases where people have gone to the trouble to assemble large groups of patients where they look at how the patients have responded to medicine, and then they go and carry out a careful genetic study to try to identify genetic variants that influence the response to medicine, it's a typical outcome that you find such genetic variants that influence response to medicine. So that means that we really should be employing these approaches right across the board.
FLATOW: So this is not an effort--this is a little bit different, then--this is not a search for genes that may cause cancer or heart disease and things like that?
Dr. ALTSHULER: Well, perhaps--I think that there's two different benefits of this kind of research, both of which are important. I think that David likes to focus a little bit more on the drug treatment side; I tend to focus a little bit more on the fundamental basis of disease. I guess you could say that we're both glass-half-full kind of guys. We like to focus on different halves of the glass that are full. I would say that there are obvious examples of the sort that he described. To make it concrete, there was a recent paper published in the New England Journal of Medicine about a drug called Warfarin, which is used for blood thinning, and it's a drug that is actually quite challenging as a physician to dose for people, because it has what's called a narrow therapeutic index. It means that if you give too little it doesn't work, and if you give too much there's a bad side effect.
Well, it turns out in a recent paper in the New England Journal of Medicine, an it's been replicated or extended by other scientists--it looks like there are at least two genes that can predict on the order of up to three or four full differences in how much of this drug you'll need. And that's actually a very big difference in the dosing of this drug. So there--and we could give other examples. So there are clearly examples of that.
On the other hand, I think in the long run the greatest benefit, in my view, is going to come from understanding what actually causes the disease. It is, as David said, a longer path to clinical relevance, because the period of time between understanding the root cause of the disease and having a better treatment based on it might be decades. But I think that in the long run, if we want to find treatments that work very effectively, that don't just mitigate disease but actually are much more curative, we have to know what are the causes. And we don't know that today. If you asked me--I'm trained as a diabetes doctor, and if you asked me, `Why does one person get diabetes and another not in this country,' it turns out we know essentially nothing about that. We know that obesity tracks with it. We know that living in the modern world does, but we don't actually know whether it's the fat cell or the muscle cell or the pancreas that is fundamentally responsible. And so until we know that, development of drugs is, at best, haphazard.
FLATOW: And you think the HapMap can do that for you?
Dr. ALTSHULER: Well, I tend to be very optimistic about the idea that if we can identify these genes, they will help validate in the human being--not in a mouse or in a cell line, but in a human being in the clinic--what's actually influencing the disease. And the reason to focus on DNA, in my mind, in this quest is because classical studies, epidemiologic studies, tell us that on the order of half of the risk of most common diseases, half of the risk of whether or not you get Type II diabetes, is DNA differences you inherit from your parents, and the other half is environment, behavior, bad luck, etc. If it's half of the story, and we now have these powerful tools, it seems like a great opportunity to make progress.
Dr. GOLDSTEIN: If I could just...
FLATOW: Sure. Sure.
Dr. GOLDSTEIN: ...add to--I'm afraid that David and I are really a boring pair to have on a show together, because we always agree with one another. So...
FLATOW: It's Frick and Frack on SCIENCE...
Dr. ALTSHULER: We'll find a way to argue.
(Soundbite of laughter)
Dr. GOLDSTEIN: So I agree with everything David just said; just want to just sort of clarify that it's, in fact, not a choice, as I know David wasn't implying, but just to clarify--it's not a choice at all between using these kinds of approaches to study treatment response and using these kinds of approaches to study the underlying causes of genetic disease. We really definitely need to do both of those things, and I think that we will do both of those things. My own guess is that the time frame for delivery will be, in fact, a little bit different, and we'll get a short--you know, a nearer-term return from the study of treatment response. But we need to do both.
But to answer your earlier question, Ira, I think what will happen in terms of the differences between the genetic control of treatment response and disease--sometimes what we'll find is that the determinants of the exact type of disease that somebody has--exactly what kind of cardiovascular disease, exactly what kind of diabetes and so on--that will influence the best treatment. And in that case, the genetic differences among people that influence treatment response are, in fact, one and the same as those that influence predisposition to disease. In other cases, there are gene variants, and we know about many of these because we've found them--there are gene variants that, in fact, have nothing to do whatsoever with an individual having the disease. Nonetheless, they influence the way that the person responds to the medicine in a way that's independent of why they were sick in the first place. It's exactly those variants that are more specific to treatment response that will, in fact, be easier to find.
FLATOW: I have about--I only have about a minute or so left to talk with you, but there's one important question I wanted to bring up. And, Dr. Altshuler, you mentioned the powerful tool that you have. You also might look at it as a powerful sword, because it is information about people's susceptibility to disease which not only is a great diagnostic tool, but, you know, it's something that insurance companies want to know about, whether they should insure you or not.
Dr. ALTSHULER: Well, I'm glad...
FLATOW: This is not a new issue.
Dr. ALTSHULER: I'm glad you raised that, because I think that there is very appropriate concern about any type of information, not just DNA information, and how it's used for benefit and not harm. I would like to point out that there is, in fact, pending federal legislation that would provide privacy of genetic information and ensure that--do our best to ensure that it's used for people's benefit and not harm. That information has passed in the Senate and is being looked at by the House, and I think that we would all be very optimistic about the work we're doing, even more optimistic if that legislation or some version of it that was appropriate passed and could help protect people from potential misuse of the information.
FLATOW: All right. Well...
Dr. GOLDSTEIN: I myself don't see...
FLATOW: Go ahead.
Dr. GOLDSTEIN: ...how we're going to use this in the clinic without some kind of non-discrimination legislation being passed.
FLATOW: Yeah. Interesting story this week; I think it was IBM saying they would not be collecting that data from their employees, any kind of...
Dr. ALTSHULER: And at the announcement of the HapMap, the secretary of Health and Human Services participated, and he made the point in response to a very similar question that while this is a critical issue, it's an issue that's critical to our society more broadly--your credit card records, your personal...
Dr. ALTSHULER: ...and health information--and we need to protect this if we're going to reap the benefits of the information revolution, whether it be in DNA, in medical records or in other aspects of our lives.
FLATOW: Well, gentlemen, I know you're moving on to HapMap II, so we'll have you back when that one is published. I want to thank you both for taking time to talk with us. Dr. David Goldstein, director of the Center for Population Genomics & Pharmacogenetics at the Institute for Genome Sciences & Policy at Duke, and Dr. David Altshuler, director of the Program in Medical and Population Genetics at the Broad Institute of Harvard and MIT and also associate professor of genetics and medicine at the Harvard School and Mass. General Hospital. Thank you, gentlemen, for taking time to talk with us.
Dr. ALTSHULER: Thank you. Thank you very much.
Dr. GOLDSTEIN: Thank you.
FLATOW: And good luck to you.
Dr. ALTSHULER: Thank you.