| || |
| || |
| David Altshuler, MD, PhD |
Altshuler is a member of the Whitehead Institute, as well as a practicing endocrinologist at Massachusetts General Hospital. His research led to the discovery of a single nucleotide polymorphism in a gene that is implicated in type 2 diabetes. He is now involved with creating a haplotype map of the human genome.
How prevalent is diabetes?
Diabetes is a very common disease. Actually in America about 7% of people will get diabetes in their lifetime and so 1 out of 14. It's very common.
What kinds of physical problems do people with diabetes suffer from?
Initially diabetes has no symptoms and that's one of the reasons it's such an insidious disease, but over time it causes damage to blood vessels and this leads to four major problems, the worse of which is an increased risk of heart attack and strokes. That's really the lethal complication, but in addition, [there is] blindness [or] kidney failure. They can require dialysis or a transplant. [They can also have] nerve damage which can lead to numbness, tingling and then ulceration and infection of the limbs that can cause amputations. It's really a terrible disease.
How much do we still need to learn about diabetes?
The amazing thing is that people have studied diabetes for more than a hundred years-there's journal dedicated to it-and there's thousands and thousands of labs. [In] my medical training, I discovered that we actually couldn't tell for one patient vs. another-why did this one get diabetes and this one didn't. Why did this person have high blood pressure, this person a heart attack, and this person nothing? [We] don't know the answers to that question, and that means that we don't really understand the root cause of the disease. We know what happens. We know the blood sugar goes up. We know the nerves get damaged, the blood vessels, but we don't really know why or how.
How did you become interested in this line of research?
I went through scientific training before I became a doctor. I studied genetics. How traits run in families and things like that. And when I went through medical training, there were two things that really struck me. One was we didn't know the root causes of these common diseases that we saw every day in the clinic, like diabetes and high blood pressure and heart attacks, and secondarily, they ran in families. It was always the case that you would say to someone, "did anyone in your family have these problem?" And not every time but many, many times they say, "well, my uncle had it, my sister, my brother." Epidemiologic studies have shown that's not just anecdote. It's in fact the case that in people with a history of Type 2 Diabetes, their relatives have a much higher rate of diabetes than the public at large.
If the rate in the population of diabetes is 7%, what's the rate to the sibling of a person who already has diabetes? Now it's 30 or 40%. What's the rate of diabetes in the identical twin of somebody who already has diabetes? It's 90%. So the difference between 7% and 30% or 90% risk of diabetes is influenced strongly by your genes and that said to me, as somebody with the training in genetics, that if we can find those genes, we'd do a much better job of predicting- and we'd do a much better job of targeting therapies to the root causes.
Aren't there environmental and lifestyle choices that have an influence?
When people hear the word "genetics" they sometimes think that it's an absolute. That in other words if you get the gene for blue eyes you have blue eyes and if you get the gene for brown eyes, you have brown eyes. That's not how most diseases work. A few diseases work that way-very terrible genetic diseases where the genes absolutely are the cause. But for most diseases like diabetes and high blood pressure, the genes are just a partial contributor and each individual gene has a small role to play.
It's what we called in medicine a "complex trait." Complex because there maybe 5 or 10 or a hundred genes that play a role in the individual patient-plus the environment, behavior, or random bad luck so there's no one cause. There's many causes, but if we can understand them, we could probably do a better job of prediction and of treatment.
What role do genes play in a disease?
One way of thinking about how genetics influences these common diseases is by example of what we already know about heart attacks with regard to risk factors like cholesterol or high blood pressure. We know that if you have high blood pressure, it doesn't guarantee you'll have a heart attack but it increases the risk. If you have high cholesterol, it doesn't guarantee you'll have a heart attack but it increases the risk-each additional risk factor just takes you up one notch and that's what the genes do as well.
Some people have a genetic background [for which] they're predetermined to have a lower risk and maybe they can suffer more of these environmental insults than someone else who's just a little bit closer to the edge. And so if we knew what those genes were, we could maybe target prevention more affectively to those who are more likely to get it. Since we don't really know what's wrong, those genes are our clues to tell us not just that it happens but why it happens. Why do some people get the disease and others not? It's written in those genes if we could find them.
Why is it important to know what the genes are?
There are two reasons to study the genetic contributions to diseases, although we should be clear that genetics is just one part of what we should study. We should study the environment and behavior and everything else as well in the context of the genetics.
But I think genetics has two unique roles. One is when you're trying to disentangle cause and effect, it's always hard to know what came first-the old "chick and egg" problem. Did the high blood pressure cause the diabetes or the diabetes cause the high blood pressure? One thing about genes is you're born with them and so we know they're actually a root cause and not a secondary phenomenon caused by the disease, so that's one important thing genes bring us.
And the other is that because of all the advances in molecular biology over the last 30 years, when you discover a gene you don't just have a predictive tool. You actually have a mechanistic understanding of what's going on and a tool that can be used to study that disease in the laboratory in a cell line, in a test tube, in a mouse. And so it's both that it's a root cause and also it really enables the next researchers in a way that other kinds of discoveries don't.
Does a person need to have a mutated form of a gene to get a disease?
For the most part in diseases like diabetes and high blood pressure, we think that the genes that influence that are ones you're born with--passed down from your parents. In some other diseases, in particular cancer, there's a major role to be played by mutations that actually happen in your body after [birth]. They're not mutations you received from your mom or your dad. They're ones that occurred in some cell in your lung or in your skin that predisposed that cell to grow out of control and become a cancer.
Where would you start in your research to find a gene that caused diabetes?
One of the ways that scientists [look for] the genes that influence these common diseases are what are called "simple association studies." You find a variation in a gene. Some people have this version; some people have that version. And you ask: Is the rate of disease different in the two groups? But one of the challenges has been that these studies haven't been very reproducible and robust. It's just been a technical or methodological problem.
A couple of years ago, we did a study where we took a whole series of findings where someone had proposed that a gene played a role in diabetes but it hadn't really held up to scrutiny.
We [also] did a study where we took all of the examples at that time that had that character. They'd been proposed but not yet validated. In fact, there were not validated genes for diabetes at that time and we tested them all in an experiment where we had many more patients and some additional controls for possible false causes of a positive result. So it's really just an attempt to get more statistical power and a cleaner study and we did that. We took 16 previously published associations and only one of them interestingly could really be reproduced. But the one that was reproduced is a gene variant in a gene called PPAR Gamma. That one has turned out to be correct in that subsequent to our paper, multiple other papers have seen it again and again and so we really believe in the community that this one gene variant truly plays a role in the risk of diabetes in the general population.
How did you design your study?
One of the important things in all clinical research is to observe enough people that you can really tell what's actually happening, as opposed to "statistical noise," which is when things just go one way or another not because there's really an effect but because there's random chance that that happened.
In order to be sure that we weren't looking at random chance, we assembled a very large study, at least by the standards of the day; 4,000 patients were included in the study and we tested each of them for each for these gene variations. And what we found was that even though we started out with 16 [gene] candidates, only one of them reproducibly was seen over and over again in the patients we looked at to have an effect on diabetes. The effect was small, which is why it was hard to detect in the first place but it was reproducible, it could be seen again and again.
What is a SNP and how does it relate to your research?
In genetics, there's a lot of talk these days about what are called SNPs. SNPs stands for S-N-P, single nucleotide polymorphisms. It basically corresponds to a place in the genome where some people have one version and other people have another version of a letter-let's say one person has a T and one person has an A. SNPs are a hot topic right now in part just because most of the genetic variation in the human population is due to SNPs-probably 90% of all the variation between people is due to these single letter changes.
The 16 variants we [looked at] in our study were actually among these SNPs. But there's 10 million SNPs in the whole human genome. These are just 16 of them that had been suggested to be relevant and people had begun to test them. But we should be clear that there's a huge Atlantic Ocean full of SNPs to study. This is the study in one tidal pool when you think of how big the entire universe of possibilities is. There's a lot more to discover.
How did you end up with 16 variants?
In the long run what we'd like to be able to do is take all of the SNPs in the human population and systematically test them in large patient populations as to any of them play a role in diabetes and hypertension and schizophrenia, etc. But technically it's too big a problem right now. So what people have done is focused in on what are called candidate genes-a gene or a part of the genome that due to some other type of experiment has been implicated as playing a role in disease.
In the case of the 16 SNPs in our study, they were all genes whose biology seemed to play a role in blood sugar. It's known that diabetes is a disorder where the blood sugar's high and there's a lot know actually about the enzymes that make sugar or the systems that monitor how high your sugar is so people took those genes, found their variations, and asked if any of them play a role in disease.
In 16 cases, they had a clue the answer was yes but it wasn't really robust. So that's what we tried to do was look at a larger sample with some more controls to see if it really was true and could be seen and wasn't due to some false signal.
Where did you find all of the SNP information for your study?
The Human Genome Project [was a] large international project paid for by governments and by foundations. When this SNP research started studying the genetic variation, it was clear it would have a lot of implications for healthcare and for the pharmaceutical industry discovering which genes might play a role in disease.
And so in a very unique and successful public-private partnership, something called the SNP Consortium was formed. It was a partnership of academic research centers and the pharmaceutical industry to discover these variations and put them out on the Web for anyone to use without any proprietary benefit.
All of the information about human genome sequence and SNPs is on [various] web sites. There's one in the National Institutes of Health, there's one in England called Ensemble, there's one at the SNP Consortium. You can go to these websites, and anyone can just look up information about the human genome sequence, about its variation. Obviously this information has direct use in research. It's not that you can look up and find anything about any individual, but you can learn something about the landscape, which is very important information for doing research.
One of the fascinating things about this kind of research is to think for a second about how is it that 15% of the 6 billion people on the planet have one version of the PPAR Gamma gene and 85% have a different version. Where do they get it from? It's inherited.
It turns out that this mutation only happened once in human history. In other words, on some Tuesday in a sperm or an egg cell, a letter that was a T was miscopied into that sperm or egg as a G, and now 15% of the people on the planet are the descendents of that person who had that mutation and this affects their risk of diabetes. All common human genetic variation is like that. It was inherited from a single ancestor who lived long ago.
And one of the things that's been of great interest recently to geneticists is that when you look at the stretch of DNA, you realize it's not just that single SNP that was inherited from your great, great, great, great, great Grandpa Ugh who lived thousands and thousands of years ago. It's actually a whole stretch of that person's chromosome that was just passed down through the generations.
Those stretches which are called "haplotypes" actually are the "atomic unit" of human inheritance-atomic unit meaning the indivisible piece that's passed down through the generations and [which has] never been broken up actually or changed. The human population actually has very few of these haplotypes explaining all the genetic variation, as if we were a very small population, which is what we were at one point.
Even though there's 6 billion people today, not that long ago there was just one population, at least so we think. It lived in Africa and every one of the planet today-all the humans-were descendents of that small group of people. It's really astounding.
Is there an attempt to map these haplotypes?
Yes, there's something called the "haplotype mapping" project. We're just getting started which is just to figure out how to layer on top of the human genome sequence in the SNPs the particular different copies, the ancestral haplotypes that exist.
How did the Human Genome Project help your research?
The Human Genome Project has had all sorts of implications for science beyond just the data. One of the things that happened is that groups of scientists started working together in a much more collaborative manner than they did before. There's technology. There's computer science. There's clinical medicine. Typically, biology as a field has not worked in that kind of team manner.
But the Human Genome Project really sparked that and our study, which was done sort of under the umbrella of a large genome center, took on some of that character where we had people from many different skills and different disciplines working. I think it allowed us perhaps to do a larger study more quickly [and] efficiently and bring in some different methods.
Was what your methodology for your study?
This kind of research has multiple steps and in our case they took place in multiple continents. So the first thing you need is to identify patients with the disease, confirm their diagnosis by drawing their blood, check their blood sugar, and then extract DNA. That was all done by our collaborator Leif Rubin in Sweden.
The DNA from these thousands of patients was shipped to Cambridge, [where]we then used automated machines to very rapidly detect the different SNPs in each gene in every individual patient. Then our statistical collaborators went to work trying to figure out whether [there were] any consistent associations between any SNP and the clinical phenotypes of the patients.
In the end, [after] all the analyses, this one SNP in a gene [called] PPAR Gamma was left standing, as truly associated to diabetes risk. The very interesting thing about this gene is that through a totally unrelated line of research, it was discovered that this gene PPAR Gamma makes the target for a class of drugs that just came out on the market to treat diabetes.
Now, this would perhaps ideally happen in the opposite order, from a geneticist point of view. We discover the gene, and [then] someone makes a drug to it. In this case, the drug was made without knowing about the gene playing a role in diabetes. [This drug] changes the protein made by that gene to work slightly differently [and] makes the blood sugar go back down in people with diabetes.
We have also [created] knockout mice where the PPAR Gamma gene has been deleted-they in fact have different blood sugar levels and sort of a predisposition towards diabetes.
What was the end result of your work?
So at the end of the day, what we learned from this study was that although there were multiple genes that had been implicated in the inherited [form] of diabetes, when you looked at them in larger and larger samples, only one of them stood out as really being associated [with diabetes].
And that now allowed us, through the work of our lab and other labs to at least put that as a foundation for research on this gene and this disease. This gene in no way explains all of why someone gets diabetes. It's just a contributor. But we don't believe necessarily [that] there will be any gene that alone causes diabetes in the population at large. There will be many genes like this one that play a small role individually but when put together explain the risks of disease.
What would be your follow-up research?
In any scientific study, there are always more questions raised than answered and this study has led to a number of lines of research. One is just to understand how is it that this slightly different form of the PPAR Gamma gene causes diabetes. What actually is going on? There are a lot of labs working on that.
The second question is more clinical. In what way are the people who inherit the gene variation different than people in [the] general population who don't-that thus far, has actually been frustrating. People haven't really been able to figure it out. It seems at least in one preliminary study, [that] it might have to do with diet. There's some evidence that people [who] eat a high fat diet have a different response to that diet in terms of their obesity and blood sugar if they have one version of this gene vs. the other, so it may be an interaction between what you eat and your genetic background.
In addition, there's a fascinating question about the drug that's used, or actually multiple drugs that are used to treat diabetes that go through the same gene. Do the people who have a slightly different version of the gene respond differently to the drug than the people who don't? That would have direct clinical implications, maybe. Maybe you'd be better off starting on Drug A if you have one set of genetic variations and better off on Drug B if you have a different set of genetic variance. [This] could be very useful to identify which drugs will work best in which patients.
Will your study impact patient care at this point?
It's important to say that the research we're talking about here is still fairly far from the clinic even though we were studying patients in the clinic. [There] is a long path between a basic discovery like ours-that a gene plays a role in disease-and knowing how to advise patients. And today there's actually no specific advice we can give to patients.
I think that the study we did could help provoke [other] clinical studies. [A study might ask]: what if we treat patients with this gene variant in this way vs. not. Perhaps [we could] discover over time which groups do better and begin to target treatment. But those are studies that will have to begin now and will take years to run. That's an important message for genetic and genomic research-it's basic fundamental research for the most part that leads to the clinical studies that will ultimately change patient care.
[It] might take 30 more years. But how long would it take if you had no idea what was wrong? That's what the geneticists are really trying to do. It's not to say we can solve every problem, because we can't, but we can simply help find out what's really wrong and the next level of research to then figure out how to treat patients better. It's a long road.
What are your future expectations for this diabetes research?
One of the things that patients struggle with is that often they already have symptoms of disease that are permanent before they even know they have a disease like diabetes.
One of the hopes of genetic research is we can identify people before they have symptoms and begin the prevention. We don't know quite how to do this yet. It's going to take long clinical studies. But it stands to reason that you don't really want to start the prevention when the symptoms have already occurred. You want to prevent it before it's happened. That requires knowing early in life. There are lots of examples where simply knowing with a little more accuracy would make a huge difference in clinical care. Prevention is always a matter of balancing the costs of the prevention-not just dollar costs-but the risks of side effects and the personal inconvenience vs. the benefit. If you can tweak that a little bit just by finding the highest and lowest risk individuals, you might take a prevention that can't be recommended today and make it a wonderful prevention simply by targeting it just where it's going to make the best impact.
You are both a researcher and a clinical doctor: what are the benefits or drawbacks to this dual role?
One of the things that I found very gratifying about being both involved in this research and also taking care of patients is that they each play off each other. On the one hand research done in the absence of any application. In other words, why am I doing this? Why is society paying for it? And it's helpful for me at least to see the people who might benefit.
And on the flip side when you're taking care of patients, it can be frustrating, because there's problems that come in every day and you don't see progress. Seeing the research side reminds me that even if we can't solve that problem today in the clinic, there's good reason to think in five or ten or 15 years we will be able to. I find that I'm more optimistic with my patients than I used to be. I say to them, well, we can't do anything about that today, but when your kids come along our hope is that there's something we can do for them that we can't today. I think that's a good feeling to know that we can deliver care today and we can hopefully make better care tomorrow.