 Thank you, and it's great to see some familiar faces in the audience. As you'll see, I haven't exactly left NIH, as I'm still using the template from the time that I was here in the Intramural Research Program, and largely I'm doing that because it's only fair. Much of what I'm going to be talking about today really was the work that I did here. I'm now a chair and spend a lot of time doing administration, but I will sort of point to some things that I'm looking forward to in the future. So today's lecture, first I have no financial interest to disclose, I'm going to talk about I think the importance of doing translation research, I'm going to build a case for that I think, in terms of applying what we're learning in genomics to public health problems. I'm going to give you an overview, kind of a tutorial, because I'm going to assume that many of you in the audience aren't that familiar with intervention research and public health, give you sort of an overview of the social and behavioral research approaches that we use. I'm going to talk about why do public health, even though we have to do interventions that may be of low or dose, and then give you some examples of translation research areas to sort of tease you, tantalize you, get you excited about the thought of doing this kind of research yourself, or at least being a part of an interdisciplinary team who would be doing this. And then I'll just share some take home messages. So why do start thinking about translation research now? And I think in 2003, when the sequence of the human genome was completed, there was a lot of concerns that we would wind up, how would we ever apply this optimally? Lots of concerns that this pile of data that we received would get lost in translation, we wouldn't know how to use it, and it would take years and years and years to figure that out. And some of that, I think, is looking like it may be true. Premature translation was the other concern. Direct-to-consumer tests were coming out, and there was a lot of concern that that was happening before the public was ready, before the science was ready. So the big question, the burning question, I think that all of us in translation research are sort of grappling with, is what is optimal application, and how do we do the research to support that? So the famous example is the direct-to-consumer marketing. I like to do a lot of what I call, let's talk about suppositions, because if you look at the popular press, you look at our colleagues in the field, a lot of them are talking about making big prognostications about what's going to happen with genomics and how it's going to really revolutionize health, and that term is actually used. And that sort of the public sector or private sector, excuse me, got on the bandwagon about that, and developed this direct-to-consumer marketing, which was highly controversial. And in one iteration of this, there was a consideration that direct-to-consumer marketing would be offered in Walgreen stores. Direct-to-consumer individuals could buy a testing kit off the shelf and do their own genetic testing and send that into a lab and get their results. And there was the FDA sort of said no, absolutely no, this can't happen. And this decision was based on no data, no data at all. Just a sense of great harm that was perceived based on ethical, legal, and social implications. I would argue that these kinds of decisions should be made based on data. So what is the data? And this was actually done in 2007, but the landscape hasn't changed when we look at what data there is out there to guide some of these decisions about translation. You see here this big blue area is bench-based science, human genomics, and this little blue area down here is anything that could even loosely smack of translation research. And a tenier, tenier part would be related directly to health services, delivery, and improvement of health, public health. So why is that? Why has translation lagged so slowly behind discovery? I think some of it is how we think about science and how it should be done and the sort of notion that it's a linear, a linear model, that we start with basic research, then we develop some kind of a treatment that can be an intervention, and then we test it under the most tightly controlled circumstances we can. Really not reflecting reality at all to prove that it can work. And then we then test it in a larger, less controlled situation to see if it's effective once it's rolled out. And then at the very end, we try to actually use it in the public. And in fact, our incentive structure for research is structured exactly the same way. If you see that we have these elements of T1, T2, T3, T4, now we're getting to T5s and T6s that talk about that this is the way, it reifies this assumption that this is how it should be. I think that actually translation research sort of challenges that and there's a resistance to that challenge. And it basically says let's start where there are gaps, let's start where there are problems, challenges, unmet needs. Think about how the discovery, anticipate how the discovery might fit those needs. I think adherence is a huge issue. Pharmacogenomics is sort of pushing on that. I think intervention research could be focusing on that as well. I think that we then use that to guide our basic science priorities, our treatment development, and our efficacy and effectiveness are sort of bundled together. So a few years ago, I think it was at the 10th anniversary of the completion of the sequence of the human genome, there was sort of a decision to sort of advance, start to think more thoughtfully about what was genomics and society research, a group of us here at NHGRI got together, intramural, extramural had long discussions and debates about let's define this field, and I think what you'll see is we came out of it sort of defining a discipline and then talking about all the methods that can be used in that arena. And you can see here that the disciplines that get tapped into are quite broad in the social sciences across a wide gamut of social sciences. And the methods too are quite diverse, ranging from qualitative methods, which I'll talk a little bit about today. I'm sorry, I'm going to get lost on my focus groups being... Am I not working yet? Qualitative surveys, structured interviews, archival research, community-based research I'll talk a little bit about. So a broad array of methods that we can use, those also take us to involve transdisciplinary methods just by virtue of the number of disciplines involved. Social science isn't monolithic any more than basic sciences. And then from that we get a number of outputs from basic knowledge, just as basic science does, all the way up to public policy. So before I get started, I want to make sure that we're all on the same page and using the same language around what is an intervention. And when we talk about an intervention, we're basically talking about anything that we do that's directed at a target group that's defined somehow to influence a desired outcome. And those outcomes can range from individuals' decision-making about healthcare options. It can be group or individual behavior change, things like smoking cessation, dietary change, improving exercise, adherence to medications. It can be attitude change or individual or group attitude change, changing beliefs, misconceptions about health because we know what people believe influences what they think the solutions are to become healthy, and then as well as public policy change. Also when we think about interventions, they span a spectrum from primary prevention to tertiary. So primary, secondary, and tertiary prevention. So with primary prevention, we're talking about healthy populations that we're working with and where the goal is to prevent illness or injury. Secondary, and the corollary to that in terms of genetics and translation research would be in susceptibility testing, tailoring interventions to those who are already healthy. Secondary prevention is really the early detection, testing surveillance of risk, and that really would be in the genomics arena would be predictive testing of high-risk groups, newborn screening being another element of that. And then the goal there is to identify disease early and do things to improve the health outcomes, long-term health outcomes. And then there's tertiary prevention, which is those who already have the condition and you're trying to help them manage the condition and be as healthy as they can living with that condition. And in the genetics arena, that would be assisting those who are already affected with the disease and particular rare diseases, where there's a lot of sequelae in the area of breast cancer, for example, long-term tamoxifen use. So I think that what distinguishes in many respects, at least what we think distinguishes social and behavioral science from many of the other sciences, when we design interventions or research studies, it's guided by theory. And the theories are numerous, as you can see here, and that they also map to different levels of influence. So we can talk about theories to when we're trying to influence an individual's health outcomes or health behaviors. We can talk about theories that involve intrapersonals, or dyadic, or small groups. We can talk about organizations, delivery systems, schools, churches are often targeted. And then we can talk at the societal level, which is policy. So this is all to say, this is sort of the toolkit. This is the science of social and behavioral research that we would then bring to these challenges of translating genomic applications. Now, again, when I say translation, I think what I probably need to always add at the head of that is anticipation of translation. So even if we don't have the perfectly figured out genomic application, we could still be figuring out ways that we would translate what we might learn. And I think that there's a bidirectional aspect of that where we're then shaping some of the basic science and the discovery that happens. So when we talk about models, when we talk about theory, what we typically do is we sort of have a category of outcomes, whatever that behavior changer, whatever that level of intervention is that we're targeting. We usually hypothesize there's some sorts of mechanisms that we're trying to influence that the intervention will likely target things like what people believe, their skill set, we give them incentives, and so forth. And then we also really consider the background, the context in which this is all happening. It all looks really simple when you look at it this way, but if you look at a particular, any particular theory and one that I've used a lot is information processing theories, how individuals respond to information they get and whether they believe it and process it and think about it and use it, you can quickly see that it gets very complicated and that typically what we try to do is measure every level of these kinds of, or at least the levels that we think are most relevant or the levels that we think are amenable to intervention. We typically try to measure those to understand how people are responding to the messages that we provide. So shifting now to thinking about why public health. So a couple years ago, I guess it's been, gosh, 10 years ago, wow. Anyway, a group of us got together to talk about priorities for translation research and what we, as a group of social scientists and social and behavioral scientists agreed was that if we were really going to optimize applications of genomics, the best things that we could do would be to, that prevention is key, that we should get our populations early. Dealing with chronic disease is expensive, difficult, and often doesn't result in improvements. That the major driver of public health in this country are risk behaviors that are poorly addressed, diet, physical activity and smoking being three big ones. That anything that we do, anything that genomics brings to these public health interventions has to be amenable to applications in a public health arena, which often has very little infrastructure and is really favors low cost solutions or primary care where the structure is quite busy, time limited, and so we have to keep that context in mind as we're shaping these new discoveries. The question will always be, under those circumstances, does genomics, whatever it is, add value to what's currently there? If it doesn't add value, the healthcare providers won't pay any attention to it and neither will the public health community. And then we also have to consider the fact that we're operating in a world of widespread health disparities and there's a lot of skepticism about whether or not any of these new translation will exacerbate those disparities, so we need to be very mindful of that and at best, anything that we do that we bring to this would actually reduce health disparities, that would be the ideal. But unfortunately, when we look at public health interventions, we take this efficacious intervention that we've tested under the most highly controlled circumstances and then we put it out in the real world and it often, it's up against some pretty big challenges and when we look at it, we're always then in this question of as we're testing new things and trying to develop new things of balancing efficacy against effectiveness and really what we want to do is find the sweet spot where we're actually maximizing the reach because the more people that we can give even a mildly effective intervention to, the greater the likelihood that we'll have a public and health impact and I want to make some examples of that. So if you think about the rare disease model and I think the one that is actually really relevant today to community settings is HNPCC, Genetic Counseling. That right now is listed as a top priority. Many of the health departments around the country are trying to consider how they can expand reach of genetic counseling and identify those who are at high risk for these rare cancers but that can be amenable to screening and improved health outcomes. So currently the approaches, and this may have changed a little, but the current approach in the clinical setting, clinical genetic setting is that these are relatively high dose approaches. So there's a genetic counseling session in which family history is taken and there's some counseling around whether or not genetic testing is appropriate. It's resource intensive in the sense that it requires a certified genetic counselor. We know in this country there are relatively few of those and they do not reach out into the rural settings of the U.S. so face-to-face sessions requiring the clients to come in which can meet with a lot of challenges for those individuals. So it would be very demanding to try to sustain this or scale this up across the nation because of the reasons I'm saying and it's also quite expensive. So it might be very efficacious and in fact I think that it would be regarded as an efficacious intervention but it has extremely low reach. So if you take a public health approach to this, the standard public health intervention is a low-dose intervention, less than an hour, generally even less than 30 minutes. This is resource light. It means that it can be implemented by a community health worker, a clinic staff, health educators. Much of the public health interventions rely on telephone, mail, internet, not face-to-face. It has to be sustainable so it has to be built into whatever is the existing infrastructure that is there to be used and it has to be inexpensive because most of these settings have very little infrastructure. Here effectiveness is the goal and reach is the most important benefit. So how can that work to our advantage? The current approach if we think it has really high efficacy, 80% will be improved in some way by being exposed to that intervention but the reach is really poor. Only 10% of those who are really, who might benefit from this service will ever see it or get it. We see that we can really have an effectiveness of only about 8%. Contrast that to a public health approach where we say we have efficacy that's actually quite low, maybe only 20% of the people benefit from it to a gold standard of some setting. But the reach is much broader, 50% will actually get exposed to it. We can see that we can actually, even with a less effective intervention, have a much broader effectiveness impact. So now I want to just turn to some examples of the research that I think, a lot of this is work that I've been doing and some of my colleagues in the audience and just sort of to give you a sense of the kind of areas where I think we could be beginning to do, anticipate ways to use genomics. So the three that I'm going to focus on today are sort of public understanding and use of genomic information, the potential for genomics to improve risk communication that can then increase the efficacy of health behavior change. And then using genomics to tailor kind of more of the precision, I'm going to refer to it as precision public health for this discussion, but the idea that we would be customizing our intervention approaches to individual genomic characteristics. So again, thinking about the suppositions, when I came to the NHGRI in 2003, there was just the sort of inklings that the direct-to-consumer testing was on the horizon. Larry Brody and I were having tea every Wednesday afternoon and in those conversations. He was a human genetics person. He was a social behavioral scientist. We sort of concocted this idea that, well, gee, if it's coming, why don't we develop something to address some of the suppositions that we were seeing coming up in the popular press. And those suppositions really were that the public really couldn't handle genetic information, and if they got genetic information without intense genetic counseling, they would be downplay the behavioral contributors and it would backfire the information to lead healthier lifestyles. That most of the individuals that would be targeted by these low probability susceptibilities would misunderstand them, either inflate them or downgrade them in ways that would also contribute to these negative outcomes. The big thing from the healthcare delivery system was that healthcare providers just would not be able to manage the tsunami of patients that would be coming in the door to seek out these results and that that would result in inappropriate use of healthcare physicians feeling pressure to ask for tests and give services that the patient really didn't need. These are all testable assertions. No one was doing it, so we launched the Multiplex Initiative. The Multiplex Initiative was to develop a prototype genetic susceptibility test that would look somewhat like, obviously way more simplified genetic susceptibility test, the 23andMe test, for example, look like. But we developed this test based on the evidence base that was out there. We brought in experts to guide us and read the literature and tell us which tests they could say with a straight face really was associated with even the small probability of increased risk for these common health conditions. We identified eight common health conditions as you can see here and 15 genes that at that time in 2006 had the evidence base, many of these have not changed, had the evidence base to be put into one of these tests. And the methods for this are described in this publication. You can read more about how we got to that test. We used at the time and still evidence-based approaches for risk communication. I'm just giving you an example here of this is the way that we use this information in the consenting materials so that individuals knew what they were getting into to study and we use these materials as part of their test result communications. Basically what it's showing here is that if you have the KCNJ-11 there's three possible what we call risk versions or variants and what your risk would be if you fall into one of these categories and using these sort of people designs to give you the trade-offs. Now obviously what you can see is that they're very, very small increases in risk only about six in a hundred between the no variants and the three. So very small probabilities. We also told them how common it was to fall into one of these categories to help them ground their risk assessments or personal risk assessments. This was a collaboration between that was funded by the that was rolled out and partially funded by the NCI Cancer Research Network and we worked specifically with Henry Ford Health System in Detroit because they had a demographically diverse population and we had a survey coordinator in Seattle that was part of the CRN. And we targeted healthy adults those who were ages 25 to 40 who didn't have any of these health conditions because we wanted to test this for that health promotion prevention, primary prevention domain. Here's what the feedback looks like that they received using public health approaches we mailed them their test results and made sure that they understood or tried to make sure that they understood was a kit a little by a multi sheets of paper that basically said we wanted to make sure that they understood that it wasn't just genes that these accounted for a very small probability. We followed it up with telephone counseling by a trained health educator with no background in genetics we trained them through multiple sessions. So here is the flow we identified through the Henry Ford Health System so we had a denominator, population based denominator. We conducted a baseline survey with these individuals we provided them access to a website they went on to the website and that's where they consented read the materials and then they scheduled a clinic visit to come in and get blood we could do now saliva but then we were adamant we need to do blood and we provided them with the results I think it was about eight weeks afterwards eight to twelve weeks afterwards and then three months later we did a follow-up survey and within a week of them getting their results someone was on the telephone with them trying to explain to them the test results. So our first supposition was that people if they are given genetic information will inflate or downgrade their beliefs about behavior what you can see here are the conditions across the bottom that we tested for the behavior bars are how much do you think your health behaviors influence these health outcomes the red bars are their behavior the blue bars are their genetics what you see across the board is behavior trumps genetics in these individuals minds variations in literacy I'm sorry I'm not showing the population but high school educated to college educated nice distribution of education level so this is not a highly literate population necessarily and what you can also see is that they can discern the nuances in genes and environment so for lung cancer for example our public health campaigns talking about smoking and association with lung cancer appears to be noting that behavior and there is a nice behavior there that's associated with lung cancer and a downgrading genetics but probably appropriately so this and this has been shown again and again in population based surveys of that genetic literacy or this notion that somehow information about genetics will trump behavioral contributors is false we also wanted to see this sort of this test this notion of the tsunami is there going to be so many people that will seek out the testing that it will overwhelm the system and what you can see here is that of the 1959 who we surveyed and were eligible only 612 went to the website testing this was free testing so this was even a model that isn't exactly replicable in the real world of that about half wanted testing and of the half who wanted testing only a slightly lower number actually about 90 less showed up to actually have their blood drawn so sort of a passive refusal and what you see here is that that accounts for about of the baseline that were approached and offered the testing only 14% got tested so the notion again that individuals would be really sort of excited about getting this kind of low probability information maybe inflated lastly did they go to their health care provider did they talk to their health care provider about this information and what you can see here is this is who they shared their test results with and I want to say that we kept the testing and the participation in the study completely separate from their health care delivery system so we didn't have any it was all done by this outside survey company that did the did the interviewing and we didn't include health care providers other than telling them the study was going on we didn't tell them anything about their patients or who was being tested or not so concerns about privacy this was all in patient empowered the patient was the one that would have to go talk to the health care provider what you can see here is that they talked to their family their spouse am I getting it yeah they talked to their spouse their family and they were only as likely to talk to the health care provider as they were to their friends so again not a sense of them charging the fort lastly we looked at health care utilization so we had this nice data that was available through the through the managed care organization on health care utilization actual utilization so we looked to see did did getting tested and getting test results increase health care utilization in the ways that would be concerning about costs what you can see here is that those who just completed the baseline survey are the dashed line and this is starting their utilization starting four quarters before testing this is where they were tested and then continuing four quarters after their testing what you can see here is that those who did the baseline only were the lowest utilizers and they stayed straight stayed flat and the ones who did the web only so they went on they thought about testing but they didn't get tested there was some bouncing that went on and what the take away is that the multiplex tested which were at the top were high utilizers to begin with and they stayed high utilizers afterwards so this idea that it was going to change it may just be that those worried well or healthy are the ones that are more attracted to this kind of testing but probably not at least going to bump up their utilization in the ways that we've been very concerned about so this is the kind of science that we can be doing in parallel or ahead of these kinds of discoveries so the second area that really has gotten a lot of attention in terms of where would genomics really revolutionize healthcare delivery is this idea that it's going to motivate behavior change that learning something personal about you is going to inflate or greatly increase your motivation to change health behaviors so here we see a theoretical model I'm just taking an amalgamation of the models that are often used in looking at motivation to say that genetic risk communication is somehow going to influence behavior change so what the thinking is is that it's going to increase perceptions of susceptibility because it's so personal this is your genome this is and that's going to be so relevant to you motivationally that you will want to take up precautions to improve your health the downside of this and the contradiction to this because the other concern is that it will demoralize people is that it will make it seem uncontrollable after all it's genetics you can't do anything about it and it will lower confidence to change and then you have a lot of other factors that you have to consider whether a person can understand the information their disposition their attitudes and beliefs that are in the background do they believe in genetics or not and then there are other motivations and then there's always context that we have to think about as well in terms of where is the testing being done how is the frame in which it's being received how's that influencing the frame and that somehow is going to influence motivation increase motivation or decrease motivation so the early studies were done in smoking cessation I was among a group that did those the first one though was Karen Lairman at Georgetown and she looked at CIP2D6 as giving that to smokers as a way to motivate them to quit smoking CIP2D6 is about how you process nicotine and most people fall into the category of high processors so almost everybody got the message that you were at risk that was the earliest studies because of ethical concerns and as you can see here there was no effect on cessation at short-term cessation or long-term cessation based on getting that genetic test my colleagues and I selected GSTM which is also involved in metabolizing nicotine and that is that the nice thing about that marker is that 50% of the population fall into sort of the high metabolizers and 50% of the population fall into the low but not high metabolizers and so you can have this nice comparison of does it do something harmful if you get the message that hey don't you know you don't have to worry as much does that diminish smoking cessation motivation and what you can see here is again providing that piece of information didn't have any effect on cessation at 6, 12 or continuous so basically there were a couple of since that time there have been several reviews and more recently Theresa Marto has done another Cochrane database to say this isn't working this giving a genetic risk information to an individual to get them to change their smoking their diet is not showing any effect across the studies I would argue you know some people say hey roll up the tent, go home, we're done this isn't going to work I think there's lots of reasons why this hasn't worked I wouldn't say I think it's maybe necessary to we're going to have to communicate about genetics I think to individuals for at least for a good long time but that doesn't necessarily mean that it's sufficient to get people to change their behaviors and there's lots of reasons why one of it is that just going back for a minute is that in both of these cases most people who show up for smoking cessation interventions it's the last ditch effort they're highly motivated they're looking for something novel I think in terms of behavior change generally that's what you're going to see so the idea is to try to identify these populations that really where motivation is not as high and see if we can get them to see if these risk communications what you see what we did when I first got to the genome institute was to look at whether we could identify younger members of families who were smokers and use work that I have been doing in the teachable moment to see if we could optimize the teachable moment by providing by including genomic risk information so in this case we looked at family members of those who'd been identified with patients who'd been identified with late stage lung cancers so these patients were likely going to die within a three months time frame and we approached their biological family members to ask them if they wanted who were smokers to ask them if they wanted to undergo genetic testing we decoupled it from a cessation intervention so we're not trying to get you to quit smoking we said that up right we just want you to see if you want to learn more about your risk with the thinking being that we would attract more people who had lower motivation and therefore perhaps have more room with our risk communications so this was a revolutionary at the time in that we were the first to be actually doing it online so people logged on got their test results online sent in a spit kit and that hadn't been done now it's standard but unfortunately again what we see here are the characteristics of those who logged on and didn't log on it's a small sample those who logged on had a 6.3 on a 7 point scale of motivation to quit smoking so again we were attracting compared to those who did not log on who were the lower motivated so again what we find is that we even in this situation we were attracting the most motivated those individuals who really probably were feeling like maybe this will do it maybe this will be the thing that will convince me and likewise we didn't show we offered so we at the end of the risk communication even though we said it's not a smoking cessation intervention we offered them smoking cessation help if they wanted it and here's the uptake we offered them nicotine replacement therapy and almost all of them took that didn't matter what their test result was whether they were in the lower risk or the higher risk category we offered them self help guides most of them took that this is sort of the standard of care for smoking cessation the only difference we saw and it wasn't significant was that more of those in the higher risk category signed on for something was fairly intensive six telephone counseling calls and those are associated with cessation rates but we didn't see any difference in cessation not surprisingly it really wasn't a cessation intervention so the take home message is that the motivated are showing up so using this as a tool for motivation maybe not the best investment but how do we get to the people who aren't motivated if we want to use this as a tool and so we have to think about what are the optimal context for that and when we think about that from a health standpoint as I mentioned earlier we're trying to go with the lowest dose we can because it's going to be the cheapest and have the broadest reach and we're trying to get people who are not as highly motivated because that's where we can have the bigger potential influence on so we're looking on that curve somewhere for what's the optimal place to time the genomic information what are the optimal populations and when it comes to smoking from younger to younger smokers like college age smokers and so we did we subsequently after that study started a study with college smokers in North Carolina several historically black colleges Duke and UNC to look at primary prevention among college smokers we had also done some work before I came to the genome institute looking at whether college smokers at a historically black college and they told us that they were only interested in finding out that they were at low risk which again supports that notion that they might be looking for ways to get permission to keep smoking so we knew we were going to have to target that and then we used our conceptual model was the protection motivation theory which is basically what are the motives that get people to seek out actions to like a smoking cessation program what we found in our baseline data and this was just an interim this study is still going on amazingly is that we were it does look like we were finding the sweet spot of those surveyed at baseline the mean desire to quit was only 3.8 on a 7 point scale so we're looking much better when we see the other population was the lowest was 5 that most of them have sort of the belief that they can quit any time so they're down playing addiction and the addiction properties of tobacco and that most of them are believing that the harms of smoking are far in the future so this seems like this could be a really good group to be thinking about risk communication so this is some of the qualitative data I don't have the outcomes yet that's still in process but what we see here is that I think it does give us some notions about where we might go forward so you see that some of what they were saying is on a scale of 1 to 10 I'm a 2 on worry smoking helps me with school with the stress I'm hearing these public health messages that are telling me that once I quit my lungs are going to repair within 2 years or something like that so I figure I can get through college and then in graduate school I can quit smoking and my lungs will be great so the sense that again these generic public health messages this may be another toe hold for our risk communications these generic health messages public health messages can be backfiring when we see there's variability in the population in terms of their understanding about some of these genetic susceptibility so the two leverage points that we were thinking about for genetic risk communication is that one young smokers don't really understand that those who are susceptible need less exposure to get the illness so there's an area of gene environment risk communication that we could be talking about genetic risk and that they also underestimate the potential for addiction and we know that there are lots of genetic risk variants out there that can give us insights into propensity for addiction so perhaps maybe rather than looking at risk communication in terms of a health outcome like lung cancer that does happen for most in older ages we should be looking at things that are more proximal in these kids' minds and we could be using lots of the social media strategies for doing that again bringing our translation research ideas to those environments could be very useful in terms of thinking about risk communication refinements so then the next area we think about is going even younger trying to get as early in terms of risk communication as possible and we all have heard about the obesity epidemic and we know where that's starting in kids and that we're seeing larger numbers of very young children who are already showing signs of markers for diabetes and hypertension so this is a major area thinking about where we could be bringing new innovation to it and colleagues of mine and I here at the genome did take that on one of the big concerns about genetic testing I mean primary prevention really should be targeting kids but the big concern is that providing genetic test results to kids has potential negative effects these are suppositions we still don't know if those really are true much of the literature says that it's sort of on balance that whether or not genetic risk information when you're young influences your life trajectory in terms of your hopes and expectations or labels these kids in some ways but we felt that we really couldn't embark on genetic testing with kids before we did some experimental studies to sort of gauge that propensity so we started out with thinking about whether or not genetic risk information provided to parents using sort of a communal coping theoretical framework the idea that moms and parents people will do things on behalf of loved ones that they wouldn't do for themselves so you're sort of getting in direct motivation if you're building on that really if you're targeting that relationship and so here we used moms thinking about getting risk messages about their kids risk for obesity and we use family history thinking that that's a relatively in fact the pediatric associations recommend that family history should be done in clinical encounters with moms and so here we have a family history risk assessment of whether based on whether the child has one parent or two parents who are overweight and so what you see here is that the child has one parent you see that same evidence-based risk communication approach they have almost a doubling effect of the child becoming overweight and if the child has two parents it quadruples or close to quadruples so this notion that this is a pretty powerful risk message much more powerful than our susceptibility variants were in the multiplex project and this is currently available it's not something that we have to wait for so what you see here is that also this sort of addresses this issue that there are a lot of challenges for doing translation research and some of the reason I think there's resistance to doing translation research is that the ball is moving that genetic discovery is moving and we don't know exactly where it's going to wind up and so we're all kind of hesitating to wait for that to happen that it's very hard to envision some of these future situations or at least we think it is I think if we work back from the problem it isn't so hard to envision but it can be and that the concepts that we're trying to deliver the context that we're trying to work in are technical and unfamiliar so how do we get around some of that and here at NHGRI Susan Persky who's in the audience has developed what we call IVETA an immersive virtual environment testing area and we use that to test out to move us beyond these hypothetical imagine that you have this risk factor what would you do which is terribly imprecise and not very often associated with outcomes actual outcomes we can improve upon that and then we can use that technology to rigorously evaluate some of our suppositions and then in this case since we were looking at a behavior like eating or food preparation we can avoid some of the challenges of trying to do that and the logistical challenges of doing that so here's the immersive virtual and testing area what you see here is our lovely model she's wearing a headset that is tracked by cameras that are mounted on the wall we can track her movements all around that room she's attached to you can see here she's attached to the computer so we can track physiological markers in addition to the movement and what she's seeing in this case is a buffet that is filled with foods that were selected based on school the slow go woe food public health food recommendations for kids and it has a mix of some of the foods that are slow meaning don't have them very much because they're not really good for you to go which are the healthier options have a mix of those and we can track her movements and what she selects her charge would be to fill a plate and our aims here are to explore one, whether concerns that genetic risk information could increase and maybe negatively in a bad way restrictive parenting practices if we tell moms that they're at high risk their kids at high risk is that going to make them overreact in ways that are going to be damaging to the child's eating behaviors which would be the worry the supposition that these can be harmful we also wanted to evaluate whether providing family history based this family history based risk information would improve, would result in improvements or have any effect on the food choices that the mothers made all based on calories and the content of this virtual plate so this is the design of the study, this in here we're seeing an experimental design so we have, we screened we recruited healthy volunteers and did a lot of sort of announcements around campus and off campus and schools and day care settings and we had a baseline web survey that sort of screened individuals for the mothers needed to be overweight we wanted all of the kids for all of the kids are all the mothers to get at least one risk information so at least one parent was overweight, we were looking for mothers of children ages four to six so that the children were young enough that it made sense that the mom would be filling a plate for her, for the child and then they came into the lab they had to practice the buffet in order to, we didn't want any learning effects so when they're in there filling their plate they have to know that if you push the gun here you get how many servings you get and so forth so they're skilled at it by the time we're going into the experimental part of the study so here they're randomized then to food safety information only so they're just told about how to keep from getting food poisoning so to speak behavioral risk information in the second group where they're told all the behavioral risk factors that contribute to a child becoming overweight and then the third group it's a combination of that behavioral risk factor information plus a family history risk assessment based on whether there's one biological parent who's overweight or two biological parents who are overweight then they do a, they do post information survey then they go into the buffet fill the plate and we can measure all the aspects of the plate the content of the plate the calories and so forth serving sizes and then we do a post buffet survey so here you see the results and what we find here is in terms of the calories on the plate is the outcome variable the behavioral risk factor arm compared to the food safety arm moms in the behavioral risk arm fill the plate with 35 fewer calories than moms in the food safety group and that was not significantly different moms in the behavioral plus family history arm filled a plate with 45 fewer calories than the food safety group and that was statistically significant the only at the P less than 5 level there were other things that were really interesting in terms of gender of the child and how that influenced but I'm going to just focus given time on the intervention effects when we looked at that just amongst the group to see what was just look at amongst the group that got the the family history message what we see is that with two overweight parents the restriction wasn't really happening that two overweight parents when the mother was in a house when the mother was a part of a duo where both parents were overweight she filled a plate with 71 or almost 72 calories more than if she was the only parent who was contributing to the risk of the child's overweight so there was a message coming through that it was the mothers who were being told it's your fault basically that were restricting not those that were being where in a sense the responsibility was being diffused across both of the parents and when we looked at that to see if that was indeed what was happening across the groups and again these are very small sample sizes so this is all very preliminary kind of hypothesis generating research what you see here is that this is the group where it was two parents it was 360 calories on the plate one parent 286 calories on the plate that difference was marginally significant at the .05 level if you go across and then we see the same thing with sweetened beverages which also contributes to overweight and childhood where that was more significant but that 38% were selecting sweetened beverage versus 14% for one when you look across the groups you don't see that happening in any of the other groups so indeed it was this family history mom being told it's your the contributing factor here that was was causing the restriction we've done some other analyses and some other papers on guilt and the role of guilt in that and I'll direct you to look at those publications but I think the thinking here is again the contextualization of that there isn't going to be one reaction to risk communication there's going to be a complex set of reactions to those to our risk communications and that our thinking about that needs to be guided by theory and there's going to be a fair amount of nuance in terms of required in terms of how we communicate these kinds of results so I think lastly the next area in terms of this is thinking about global health I mean there's been concerns in terms of the health disparities that what we're discovering won't really reach the poorest countries so the challenge here is to think about how do we do this kind of research in countries where literacy levels are very low and resources are even lower than they are here in many of our poorest public health settings so we started a project in Ethiopia to look at how we might use genetic information in a context that was greatly in need of that service and that was the context of podoconiosis, condition non-filarial elephantiasis that is in highland Ethiopia it's also in other volcanic areas around the world but in Ethiopia there's great poverty and individuals walk long distances so there's barefoot so this is a particular problem what we see here is that it's inflammatory response to soil irritants the sort of micro particles in the soil that come out of the volcanic rock and it clusters in families it's been shown to be genetic if you have a sibling it's your 5-fold increased risk of getting the condition but it's completely preventable if the child protects their feet from the soil and wears footwear consistently 50% of the population are under the age of 15 so providing shoes to everyone which would be the logical solution everyone can benefit from foot protection there's no infrastructure to do that so this idea of targeting those at high risk is one that is being bandied about in many other areas conditions that's what we're doing in cancer at public health level is something that we need to think about so targeting shoes to those at high risk is what we were thinking about doing and we were working with at the time Tom Shoes who was providing shoes to these high risk kids so we embarked on some pilot data the big issue here was that these families are highly stigmatized once a family member shows that they have this condition people are there banned from group functions they are can't go to church they can't they people won't marry into their families so very high level of stigma we started talking with the community organizations there was great resistance to the notion that we would introduce we would confirm what everybody already believed which was that it was genetic because by confirming that we would only exacerbate the stigma and that there wasn't really a belief that we could actually characterize this as a gene environment interaction and really diminish or lower stigma not a lot of belief in that so we started out looking at the study sites in a qualitative methodology so we did a very large qualitative effort over about three year period where we looked at different sites in these rural Ethiopian areas I'm sorry and looked at large numbers of individuals partly because there was such high rates of volunteering that we couldn't turn people away so these were different sites that these were the numbers of cases in those sites this was the duration of relationships so we wanted to vary that with the NGO we were distributing Tom's shoes to the kids and then we also wanted those that were more and less remote so varying these different communities to find out what some of the beliefs were we did 28 focus groups and 38 individual lots of we had 307 participations this is a qualitative effort on steroids most of them don't look like this so what we found out in our qualitative I'm just going to cut to the chase here was that if the individuals in the community believe that it was heredity contributing to POTO they diminish their beliefs they decrease their beliefs that it would be important to do anything preventive because it couldn't be prevented and they were much more likely to endorse stigmatizing behaviors and that that was actually associated with their beliefs about wanting more distance from individuals who had this condition they weren't willing to marry into these and they also if they were affected they stigmatized themselves and thought they themselves were less worthy there also was a lot of belief that it was contagious and that it was not hereditary but sort of contagious in the sense that it was hereditary in the sense that contagion would happen in the home amongst family members so in those cases they thought that shoes might be okay they might be greater importance and they had more empathy for the patients but they were so worried about contagion that it still contributed to stigma they didn't want to be around these families they didn't want to marry into them and if they had it they still felt stigmatized so what you tend to do then when you do these qualitative is you identify the themes and then you go in with a quantitative survey to verify that in fact what you've picked up in this qualitative methodology really bears truth at the population level so we did a survey with 1100 unaffected family members or unaffected households and almost 600 affected households to look and see if in fact these were prevalent beliefs and what you see here is that these differences the yellow being the affected those who come from families with podoconiosis and the gray bars being those who don't come from those households and what you see across the board is that our findings in the qualitative research were largely substantiated and you also see that there were quite large differences between the unaffected households where the unaffected households really looked like they were the ones that would benefit the most from an intervention so without that qualitative work and the quantitative paired up together the mixed methods approach is what we call it we might have not done the intervention targeted to the unaffected individuals we also saw that in terms of stigma that the individuals that our quantitative results did validate that there was stigma feelings of stigma 40% feeling ashamed inferior and that people were distancing them likewise the unaffected families were endorsing at fairly large levels that they were engaging in stigmatizing behaviors so again another sort of endorsement that if we just targeted the effectives with education around gene environment interactions and contributors to this condition we would miss a big element of the problem so what we did is we launched a what I'm going to call a quasi-experimental design study to evaluate what they were doing in usual care so it's also kind of comparative effectiveness approach what they were already doing in usual care to two other interventions a household skills building around how to wear shoes why it's important to wear shoes and boosters to encourage them to do that and then adding inherited susceptibility or not as an educational tool and our two outcomes were were they wearing in the prior seven days shoes every day and how much experienced or enacted stigma depending on the population or whether they were affected or unaffected they were engaging in we selected affected households using again population based strategies the ledgers that they had the NGO had and then identifying at random 100 households from each site and adult that had to be to be eligible the household had to have an adult who had an index child that was in this age group to participate in the study and then we identified two neighboring households within 500 meters of the affected household they had a child in the target age there was nobody in the household who had a blood relative with podoconiosis and there somebody some adult who said they were the cared for the child in that household was willing to participate so there was six communities I call it a quasi-experimental there's a lot of debate about this there was randomization but there were six communities so an N of six and you'll see that randomization doesn't always work very well in those situations and there were a lot of factors that we couldn't control obviously the three arms were what they were currently doing which was just an onsite when the shoes were being distributed an onsite education just to the affected household those two those two communities got just that intervention household-based skills training with lay health advisors who went out to the households and trained them on why it was important to wear shoes and how to wear shoes and then a public health campaign which I'll show you a little bit of in a minute to reinforce that message throughout what they call cabeles and then the two groups got this household-based skills plus the inherited soil sensitivity education so this was a six month intervention there was a booster session about midpoint as I said we did posters we had a local musician that did wrote a song that we played on a van that we drove around through the cabeles we had the lay health advisors going out to the households these are the kinds of posters that were out there the message being wear shoes everywhere all the time no matter what you're doing and the big thing for the genetic susceptibility was this gene environment interaction and trying to, as I said the concerns about stigma was so pronounced so what we did is we wanted to normalize this sort of susceptibility so we used the metaphor of sun exposure these individuals walk long distances in the sun there's no tree coverage in Highland Ethiopia as you can see in the walking in the people that are walking some folks are walking bare headed don't seem to be bothered at all by the sun some people have head wraps some people carry umbrellas and there's no judgments or stigma that is associated with any of that but it's reflecting a sensitivity a difference or a variation in sensitivity that we wanted to play up as this is the foot exposure is the same thing that resonated very well in our pilot testing it's not really unsophisticated but a good place to start an additive model of gene environment interaction where basically using a method that's been tested before a jar of marbles where your susceptibility fills the marble fills the jar up to a certain point so it takes less exposure to fill that jar totally that would be an additive model we used that but we took people so here we have basically essentially these jars if you have just by being exposed sort of area you have some risk but if you have the susceptibility you have even more risk and that really all you have to do and there is another figure here is put shoes on and it's like the umbrella it protects you from this exposure so here are the outcomes as you can see for our main outcome was affected observed shoe wearing and what we have here is the three groups the baseline I think it's always nice to say it was quasi-experimental baseline shoe wearing in the usual care group was 51% you compare baseline shoe wearing in this is just the health behavior 31% and in the gene environment 32% so this was just a fluke of the groups we selected then what we calculated at each of the follow up points was the change so the change in shoe wearing in the follow up so in this usual care group there was a decrease of shoe wearing but there were increases in shoe wearing in the other two groups but not significant increases and likewise same thing increases across so no effect on observed shoe wearing in the affected individuals was marginally significant I would say no so affected experience stigma what we see here is baseline levels of stigma in the usual care group about 1.78 higher in the HB group to begin with and higher in the GE group to begin with so again another fluke of this randomization thing decreases in stigma across the board from for each of the groups slightly not really no difference there either and that would have been experienced stigma reported but when we get to the unaffected individuals what we see here is more same baseline levels of stigma and then decreasing again stigma across the board but a higher decrease in stigma here again marginally significant but suggestive of the fact and I think the most important thing to hear is did it the point of it significantly decreasing stigma really isn't the issue it didn't increase stigma really concerned so it didn't produce the harm that was worried about another supposition that fell aside when the data was brought to it obviously we need to do more testing but I think what it says is that we can shape these messages in ways that will benefit health we don't have to assume that behavioral science can bring that kind of information to these shapings so the last little area I want to talk about is the sort of precision medicine kind of notions of customizing intervention design sort of notion that we're all zebras and that we need to move away from these one size treatments and look at individual characteristics and like I said that has been characterized largely as precision medicine but there are some of those out there who are saying let's think about this in terms of public health what implications does improving the sort of specificity or moving away from one size or what does that have for public health I'm beating a dead horse here to say that health behaviors are a major problem and that they don't seem to be getting any better we don't see despite all of our public health efforts being directed at these health behaviors we're not seeing that our interventions are really making a big difference and I think that many of us believe that that is at least in part because we're ignoring individual variability in how people respond to both our messages and to our intervention approaches and we can see here the problems are really intractable we still have even if we get people to change their behaviors we get people to quit smoking relapse rates are 80% even if we get people to lose weight they gain it back it's not to be fatalistic it's just that we really do have a lot of work that we could be doing and again the question is does this new genomic discovery bring anything to that challenge and that problem I'm going to skip this I think it's just basically the point is that there's a lot if we look at intervention trials people drop out of intervention trials for lots of reasons that I think point to their individual challenges in trying to accomplish our goals and interesting that when we think about public health messages around weight loss for example what we see is that it's sort of energy in energy out kind of stupid here's what you do you know you can take into account that there's lots of contextual factors here but largely we ignore that what we're learning from genomic discovery is that there's lots of genetic variance suggesting individual differences in this equation so in our response to calorie restriction appetite control eating in the absence of hunger and also how we respond to physical activity whether it gives us the mood boost or not how we overheat and so forth so lots of individual variation that has to be considered in the context of other public health challenges like we don't want to say well gee it doesn't matter that you're in a food desert that's not the point but it's a gene environment kind of interaction and there have been some folks who've been doing some thinking about this my favorite is Angela Bryant she continues to do these models of thinking about how we would customize interventions and what we need to learn about how people respond to exercise before we can start to tailor those or as part of tailoring those so what her arguments are is that if you take something like exercise behavior and you back up from that we all you know I've told you a little bit about motivation and the importance of motivation but that underlying motivation is a lot of what how people subjectively experience what it is they're being asked to do and how they respond to it physically and that all of those have genetic influences and in fact more and more we're even considering that exercise is going to have epigenetic influences on some of these things so it's bi-directional and should be considered and there's some examples of this older studies that have been done in terms of looking at whether or not individuals who are exposed we know are exposed to the same intervention in this case it's a Japanese study looking at women who have a common variant versus a risk variant in this trip 64R polymorphism and involved in appetite control and weight loss and what you see here is that if they have the common variant it looks like they are they walk this they restrict their calories and they get a nice you know nice correlation here of loss of weight and if they whereas those who have the risk variant we don't see that same association so that there is some validation of this notion that what people are telling us is I'm doing what you told me to do and it's not working for me that there's probably some truth to that and maybe we shouldn't be ignoring that as gee you're just not working hard enough or it's just a sort of the Protestant ethic approach similarly we see this with exercise the same sorts of things this is Angela's study where it's BDNF looking at the two different variants again hypothesis generating in my opinion you're looking at whether individual from a controlled exercise session on a bicycle everything being under control you see that some individuals are getting who are working harder for less positive affect or they're getting less positive affect from the full session and they're working harder to get it so again we see this sense of validating that perhaps we should be applying those same principles to our behavior change interventions similarly I think one of the things that's really interesting from the relapse prevention standpoint again I think probably my colleagues even at Emory say these are kind of they're small studies you know they're probably methodologically challenged yeah they are again I think that doesn't mean we don't keep trying to figure out make the studies more rigorous or improve the science but ask the question and in this case this is a nutrigenomics sort of modeling on the pharmacogenomics this was done in Greece very small study again looking at people who had failed repeated weight loss attempts which is really most of us and looking at patients in controls and customizing the dietary recommendations based on 19 genes across 7 categories all that were amenable to intervention so we could make a precise on their nutritional advice and what they found was which I think is very interesting it's no effect in the short term so maybe motivation is really all you need in the beginnings of an intervention but it isn't what keeps you able to sustain those changes afterwards and that maybe that's where customization because you can see here where the diet started the customized diet started to distinguish itself was after, was in the longer term so maybe it's the customization is going to help us with relapse prevention not so much with the initial treatment outcomes and all of this, again I say this is very, these are small studies I want to have a lot of caveats I don't think that this is the greatest science yet but everything that we do is going to have to be put into the model of some kind of comparative effectiveness so these are studies that take time they take resources and if we don't start doing it sooner we're going to really be up against a situation like what I showed in the beginning where we've got rolling out new applications that we have no idea of whether or not they work or not and I think we should, we have the knowledge and the power to avoid that outcome so the take home messages are that I think we should be doing translation research now in step with basic science discovery that there are lots of ways to do that lots of ideas I gave you just some examples from my own work but there are lots of other people out there doing really exciting stuff as well we have lots of conceptual models lots is the key word here and that we have to think about practicability to guide our research questions and where are the big problems and where is the discovery going to match bringing that match together early and using our full armamentarium of methods and obviously we have to work together as interdisciplinary teams basic scientists, social and behavioral scientists this is the environment you guys are the leaders in the field and you should be setting the example for doing that so thanks to my colleagues who helped in this work and our collaborators and friends and co-authors and you can reach me I want to put a plug in for the social and behavioral research branches alive and well and doing great but if you want to reach me I'm at Emory Now and there's my address so thank you very much for your attention