 Well, good morning and thanks. As you've heard until recently, our analytic tools allowed us to find gene variants or mutations that greatly affected a person's characteristic or their risk of disease. And non-genetic factors that have little effect on risk of disease in these kinds of these kinds of mutations, for example, hereditary polyposis in which individuals develop a number of polyps in their colon, and they have a greatly increased risk of colon cancer. So we are going to talk today about more recent tools, genome-wide association studies. And my job, as I understand it, is to give a little bit of an overview about causation models, thinking about genes and environment. And then I'll discuss an example of complex disease, etiology, cancer. Okay. So first, to look at causal models and inferences, inferences we can make from study populations. So what I'm trying to depict here is really a continuum of causation, where environment, rather than looking at nature or nurture, which historically people have concentrated on and had debates about nature or nurture, that really this is nature and nurture, that both are important, it's a matter of degree. So on the left side of the continuum, there's more a genetic effect where variation in the genes affects disease risk of the trait that we're studying. On the other end of the spectrum, our environmental factors, our nurture, and they have a greater impact on disease risk or on the characteristic we're studying. So the conditions that we've considered or found today, the kinds of genes that we found, are ones that are primarily due to genetic factors, such as Huntington's disease and cystic fibrosis, then environmental factors have little impact on risk. Now on the other end of the spectrum, are things that are largely environmental if we think of extreme exposures to ionizing radiation or airplane accidents or things like that. Those are things that are largely environmental in nature. Most of the common diseases fall somewhere on the spectrum. So for example, in cancer, there are a number of conditions called hereditary, cancer susceptibility syndromes, and those genes cause a greatly increased risk of cancer, but they're not deterministic. There's other factors as well. So the vast majority of what we're looking at then are somewhere in the middle of this spectrum. So just to take an example of height, that both genetic factors and environmental factors are nutrition, affect a person's ultimate height. So what we're really looking at when we study populations, it will cause a difference in height among individuals. Another one I'm trying to depict is that this is a population in which people are homogeneous genetically for genes related to height. So when we study both nutrition and height in this population and genetic factors, what we'll find is that it's nutrition largely that affects a person's ultimate height because there is no variation in genetics, so we aren't able to find that. In this case, in populations where, in which nutrition is adequate for development of height, for example, in the U.S., what we'll find, most of the variation in height in the population will be due to genetic factors. And finally, when in populations in which there is a lot of variation in nutrition that affects height and genetic variation, we will be able to find both factors. So the point of this small section is that we have a causal model here where both genes and environment affected height, but we can only find it, whether what we find depends on the population study. So if there's little or no variation in the genetic factors, we won't be able to find that. Likewise, if nutrition or environmental factors are somewhat homogeneous, we won't be able to find that. What this means practically is that you can find differences amongst studies or different populations that may indeed be true differences. Differences in study results from populations are sometimes viewed as errors or inconsistencies due to study design, or in fact, they may be real differences because of the nature of the populations that are studying. So I wanted to turn to an example of multifactorial models in complex diseases in cancer. Because I think when we've studied, we've seen people that have a greatly increased risk of cancer or other common diseases, and then we wonder what underlies the other risk. So what this is depicting, it actually is a group of, in this case, a group of women who have breast cancer. And we know that mutations in genes such as BRCA1 and BRCA2 cause a greatly increased risk of breast cancer and other cancers. But what is less clear, or we also know, that only 5 to 10 percent of breast cancers can be ascribed to a known inherited factor such as mutations in BRCA1 or BRCA2. And that's true for all cancers, that there's a small proportion that can be explained by strong risk factors. We go back to the end of the genetics end of the continuum. These others are sometimes, so what might be causing cancer in the 9 out of 10 women in this picture who don't have a known inherited factor? These are also often referred to as sporadic cancers, but they still likely have an inherited component to them. Most cancers come from random mutations that develop in the cells in the body. Cancer is inherently a genetic condition because it's due to genetic changes in the cells. But there are changes that accumulate over a person's lifetime. They're not inherited changes. So let's look at some of the mechanisms underlying both those changes and how they may be repaired or maintained. So just as a quick review that we're really talking about are changes in the double-strand or single-strand DNA. These are the nucleotides that Dr. Collins talked about. And in nucleotides or genes, genes are read in the cell to produce proteins. So if there's a mutation in a nucleotide, then that may result in an altered protein that then has a decreased efficiency or doesn't function at all. And go back to the BRCA-1 and BRCA-2 example. Many of the mutations that cause a greatly increased risk of breast cancer cause a protein that is shortened. So it's a protein that doesn't work properly in the cell. So those actually knowing the mutation, the type of mutation that someone has in those genes can predict something about their cancer risk. So thinking about the changes in the cells, that as I mentioned earlier, cancer is inherently a genetic condition that results from many changes in the cells. And this is depicting normal, this is cells life as it goes along and is replicating. And basically cancer results from cells result from many changes that allow uncontrolled cell proliferation. So the control, the normal control is lost. And what's thought is that then inherited mutations, people who have inherited cancer risk have a jumpstart on this process so that they require either able to accumulate mutations faster because of the changes or may have inherited an important mutation in this process. So cancer arises from a series of mutations in a single cell that then allow the cell to proliferate. Okay, so let's look at what might have gone wrong then for these nine women. So these are some of the factors that can affect cancer development and progression. And they're oncogenes, which are genes that promote cancer development when they're changed or mutated. Tumor suppressor genes actually would normally help control cell growth. And so when those are changed or mutated, the cell no longer has that control, but that control is altered. So that's another type of gene that may affect, will affect developing cancer. There's also a class of genes called cancer metabolism genes and carcinogen metabolism genes. If you think about the exposures that we're subject to every day and, for example, tobacco smoke that carcinogens can do their damage by getting into the cell and causing genetic changes. So the efficiency of getting those exposures into the cells and also getting the toxins out will affect the amount of damage that can be done in the cell. And finally there, that our body has immune surveillance for abnormal cells. And so some, many of the abnormal cells are found by the body and destroyed through immune surveillance. So I just want to look at one example of a tumor suppressor gene, which is DNA repair proficiency. So our cells actually, there's damage to our DNA occurring all the time. And estimates are about 1,000 to a million DNA modification events per cell per day. And that may sound like a lot, but when you think that we have 6 billion nucleotides in our cells, it's a small proportion of that. So clearly we have something that allows us, allows our bodies to repair this damage so that it doesn't accumulate and doesn't cause early cancer or early death. So this, so the damage to the cell can be done by both endogenous mechanisms. So through normal cell processes or also through exogenous mechanisms such as exposures. Exposures to ionizing radiation, exposures to tobacco smoke, different kinds of exposures. And normally these, the damage that's done from the endogenous exposures and exogenous exposures is repaired. And we've been talking about the DNA in the nucleus so far, but there's also DNA in the mitochondria in the cell. So we have a, so it's repaired and then normal metabolism goes on. Well, if these changes go on unrepaired, there's some pathology cell, that abnormal cell that has some pathology. And that can result, as we talked about in cancer, and can also result in two, two ways for the body to control, control damage. Kind of at the end point, senescence, the cell basically goes into a dormant state, a permanent, permanently dormant state and can't replicate. And then apoptosis is cell death, so programmed cell death. So it's a way, kind of an end way for the body to protect itself from the DNA damage. So actually I want to talk a little bit more about these repair mechanisms, because I think that those can illustrate some of the complexity that we talk about when we talk about multiple genes or multifactorial diseases, multiple genes having small effects. So for example, there's different kinds of DNA damage depending on the exposures. And there's two types are single strand damage, which is just a one strand of the double helix, and then there's also double strand damage. And if we think of the single strand damage, there's three types of repair. So if there's one base, if we think of the nucleotide, there's one base can be damaged. And there's, for example, endogenous, damage to endogenous exposures tends to be single base damage. And there's also damage to environmental exposures such as components of tobacco smoke. So there's under this rubric of repair, there underlies many pathways. I got the one minute signal so I'll speed up. So basically the complexity in DNA repair is under this area right here. So there are multiple pathways depending on the type of damage that's done. And I'll skip this slide. This was just to show actually some, a little bit of the complexity of genes that affect the lifespan and that a lot of those are due to DNA repair, dune DNA repair. So to conclude, what we learned about disease theology in population-based studies depends on the populations that we studied, as well as the measurements that we include in the study. So differing results may represent real differences in populations, or they may be due to design issues, that there may be differences in study design that affect the results. And finding the same results in several populations reinforces the finding that it may be real, which we hear a lot about in genome-wide association studies. And I've also tried here to illustrate the kind of a complexity of disease theology can lead to very complex pathogenic pathways, which result in many genes having small effects. And that these complex pathways account for most of the common diseases that we see in the population. And finally, that both nature and nurture are important to consider. Studying one without the context of the other may lead to erroneous or incomplete results. So what we'd like to do now is Emily will stay and have a conversation for the next 15 or 20 minutes before we have Dr. Brody come up. At your tables for the reporters are microphones, since we're doing this for a webcast. If you're going to ask a question, please turn your microphone on. The little red light comes on around the top of the microphone. And then when you're done, turn it off. But we definitely want to start a conversation around this stuff. Dr. Harris has started off trying to show us very much the complexity of the relationship between environment and genes. And especially as it goes down to the biochemical pathways that are life. So we'll start with your questions for Dr. Harris and then we'll move on. Hi. I have a question. Could you just tell us who you are? Sure. I'm Joanna Schaffhausen. I'm from ABC News. I work in the medical unit. I have a question about the model of the building up of the mutations leading to the development of cancer. It seems like with a lot of diseases, not just cancer, that they come in two forms. One, an early onset that appears to be really strongly genetic. And then the later onset, which seems to be much less strongly genetic, at least that we know of so far. Is it similar for things like Alzheimer's disease or Parkinson's, this idea that, you know, it's a building up of the mutations and the people with the genetics have this head start? Or would it be a different model for other types of diseases? Yeah. So yes, part of, for most diseases that are of adult onset, if you look hard enough, you can find instances with early onset severe disease that tends to be highly heritable. So for Alzheimer's disease, for instance, there are pedigrees where individuals are coming down with a disease in their 30s and 40s. They're inherited in a dominant fashion. And that are mutations generally in Precinellin 1 or Precinellin 2 or in the beta-amyloid protein. But most late onset Alzheimer's individuals do not have mutations in those genes. They tend to have a variation in ApoE called ApoE4. I don't know that in that instance you can draw, though, the mechanistic conclusion about a buildup of mutations. We know that cancer is coming about as a consequence of a buildup of mutations. With Alzheimer's disease, it's more like you're building up some kind of toxic product that gives rise to the tangles and the plaques that you see under the microscope, which appear to be sort of insoluble proteins. And presumably in the early onset form, that gets going very rapidly and leads to the pathology at an early age, whereas in older onset form it's not quite as bad and so it takes longer before the symptoms start to appear. But actually, can we back up? There was one slide here, Emily, that I thought we ought to dwell on for a minute, because while this is the National Cancer Institute slide, so it must be right, I'm actually going to take some issue with the caption here. So what they're talking about here by identifying one of the 10 people with breast cancer is to say that there are instances, kind of like your question, where breast cancer is strongly inherited. A woman who has a BRCA1 or a BRCA2 mutation is presumably what's being depicted here. But that's not to say that the other nine women with breast cancer have no hereditary contribution to the fact that they've developed the disease, because we know that in those other nine women, family history was probably a predictive factor of their risk collectively. It's just that in those other nine women, the genetic factors, the heritable factors were much weaker, and until recently, were not identifiable. I am aware there is a breast cancer genome-wide association study under review, which has identified genes that are playing a role in those other nine women at a very high level of statistical confidence. I don't know when that paper will come out, but you'll be hearing about it, I suspect, in the next month or two. And in a way, this is a bit of a fine distinction that is probably not going to hold up between highly heritable and not so heritable. This is going to be an indication of hereditability from things that are really deterministic, like high-intensity disease. If you have a mutation, you can get that on the screen. So things that have a very low odds ratio, if you can hear about odds ratio, if they want, they don't want to tweak your risk. They kind of will get everything over to you. With the breast cancer example, the women who don't have the, who don't inherit the BRCA genes, oh, Judy Foreman from the Boston Globe and other places, do the sporadic cases also end up with damage to the BRCA genes? We're in the house here, very early, who actually runs the database for BRCA1 mutations. Do you want to answer that question? What's the frequency of somatic mutations? So the answer is that the kind of mutations that we're used to where the DNA actually is, the DNA for the gene is damaged so that the gene doesn't function anymore, doesn't happen very often in sporadic breast cancers. There are cases where the genes turned off by something called epigenetic changes where the gene state is methylated and the gene is shut off, but if we go in and sequence it, it looks perfectly normal. So is that what you're getting at? So there's multiple pathways, obviously, to breast cancer. Even in the sporadic cases, they don't include necessarily the BRCA. Correct. Yeah. There are other tumor suppressor genes where they were found in inherited families, colon cancer is a great example, where mutations were found in inherited colon cancer families and then when we looked at the sporadic cases that most of the population, the same genes were mutated in tumors, but it wasn't the case for the BRCA1 and 2 gene. I'd like to reinforce something that Dr. Collins said about these so-called sporadic cases, that the point of showing that the talking about DNA repair is that that may be one of the many mechanisms that is affected by genetic variations, the efficiency of DNA repair that may underline the so-called sporadic cases and may put individuals at increased risk because they can't quite as efficiently repair DNA damage that causes, that leads to the mutations in cancer. We know that to be the case, for instance, in hereditary non-polyposis colon cancer, which is a fairly strongly inherited form of colon cancer, where the mutations are in the genes that make the proteins that normally act as the spell checker when you're copying DNA. And if your spell checker isn't working, you make more mistakes, you get colon and uterine cancer as a result. David Brown with the Washington Post. Can you say something about the usefulness of populations with extremely homogeneous genetic backgrounds and whether, first of all, how available they are, how you know, how you can test their homogeneity, and are they useful to attribute the function of the environment or are they useful for identifying the rare exception to the homogeneity? That's a great question, and I'm not sure we completely have defined the answer experimentally, but clearly populations that have been derived from a small set of founders and have remained as a group because of the absence of outbreeding, the Amish, for instance, or people who live in Finland who are often the subject of studies because of having descended from maybe only about 2,000 people that moved into Finland about 2,000 years ago. Those are populations that are particularly revealing for rare recessive conditions. Remember, a recessive disease comes about because both parents carry a rare mutation and then they each pass that on to the child who then has two copies of it and develops the disease in a circumstance where you have a limited number of founders the likelihood of having a child that has, in fact, inherited the same mutation from both parents goes up. That's the consequence of inbreeding. So there are lots of diseases of the recessive sort that have been characterized in populations like the Amish or the Finns or the Mennonites or the Ashkenazi, which reflect, I think, that population structure. I guess some of us thought maybe when it came to common disease, looking for variations that were going to predispose to diabetes, for instance, that it would also be more successful to look in a relatively homogeneous population because they would have less of a diverse spectrum of gene contributions. So far that hasn't really panned out. And I think it hasn't panned out because the things we are finding are, in fact, quite common variants. And that means even with a few thousand founders, you have essentially the same spectrum of genetic variation as if you had a few million. A few thousand is enough, basically, to be a pretty good snapshot of the rest of the population. In terms of environment, oftentimes the populations that are genetically constricted are also environmentally more homogeneous because they have stayed together in one place and they're likely to have, many times, same cultural practices. So that is a potential advantage of doing this kind of study, although it doesn't help you so much in terms of disaggregating genes and environment if they're both basically different than you'd see in a general population. I think the bottom line is for common diseases, it's not clear that it has made a huge difference to focus on those special populations as opposed to looking at a more outbred group. I think you did this, Teri Manolio. You did get the idea, though, that if you have limited genetic variability, you are more likely to pick up environmental influences on a disease than if you have much more genetic variability. So that is helpful. And likewise, if you have a group that lives in a very constrained environment, you're more likely to pick up the genetic contributions. Hi, Dr. Collins, could you go over again just briefly the notion of the SNPs traveling together in neighborhoods just boiling, boiled down? Oh, okay. Oh, sorry, Meredith Wadman with Nature. Yeah, it's coming. I was just giving a very quick overview. This will be repeated probably at least twice because I think Teri will say something about it, too.