 Thanks very much and I appreciate the opportunity to be here. Can you hear me, okay? I Have been sitting here as you have listening to the amazing science the science that we're gonna continue hearing for the rest of the Day, so this is a kind of interesting break and step back To ask as this amazing science unfolds. What are the social implications of the expanding knowledge? Well, I want to start by going back to Something that Francis wrote almost a decade ago He was talking about what I would say are the two important Components of the human genome project and what is following from it. The first is amazing discovery The second is the understanding that there is a social Obligation that accompanies this amazing scientific discovery The notion that not only are we going to learn things find out things and they're going to be fascinating But they will lead us to the common good and the exciting news. I think is that we're getting closer all the time I'm going to talk about a range of social implications That flow and that are particularly relevant right now at this stage in genomic research And they come under three general headings. One is research participation that is a an increasing number of Participants in research human subjects in research and the ways in which the research that's going forward may raise questions for them non-medical uses of genomics and I'm going to focus particularly on the implications of genomics in law enforcement and then where I think we're all heading the prospects of Genomic research leading us to a new era in health care So let me first talk about Human subject participation in genomic research as you're all well aware We're now into the era of large data repositories And that means lots and lots of human subjects and that means a number of Complexities in terms of how we go about doing research that start with the creation of data repositories in order to have the Large enough number of samples to answer some of the questions that we're beginning to be able to answer This diagram this complicated diagram is taken from an article by Bill Lawrence and Francis Collins It's actually less complicated than the diagram was in the paper I just want to focus on a few issues one is that there are an important set of questions That need to be addressed between Some data being collected and then the decision to put that data through a data contribution agreement into a data repository We all understand the value of these data repositories They enable us to address questions that we couldn't address without the scope of data that's provided But there are issues that have to do with identification or Anonymization and to to do with what range of research will be done with the samples And then we have issues that have to do with how we're going to structure and store the data And then what uses we will permit and in each of these areas questions arise that have social implications In general There are These are the set of reasons or a set of issues that we need to think about in terms of concerns for Participants concerns about participant safety Concerns about the broader implications for participation clearly privacy is one of them as we Gather and collate large amounts of information on large numbers of people Whatever privacy issues we've encountered before Exponentially increase We have not only more people involved in their form more people potentially at risk for loss of Confidentiality but a large amount of data that has to be protected in a meaningful way We have non-medical uses of information and it's partly because we have increasing non-medical uses of information That we can raise some concerns about this research and I'm going to focus particularly on law enforcement Implications and research on questions that aren't strictly medical But I will also say that the whole notion of identity and what it is is likely to be impacted greatly by our increasing genomic knowledge and then finally participants Who have data put into data repositories where research? Then is done on important health questions may by virtue of having their data Be studied through a repository lose Access to potentially actionable health information, and I think that's a serious Issue that we'll have to consider but let me now reflect a bit on The use of DNA and law enforcement clearly there is uneasiness That's been expressed in a number of newspaper articles and statements about the growth of the use of DNA for Identification and the growth of DNA banks to enable such identification in the process of law law enforcement That uneasiness comes because some uses of this information Basically raised questions. We've seen DNA identification used in immigration decision-making We see calls for the routine collection of DNA at a rest Not conviction but arrest and that raises questions from a civil liberties perspective And there is the technical potential to identify a suspect not merely between via a match between Material collected at a crime scene and that individual but actually between the material and the individuals relative which Raises a lot of social ramifications These are important issues that we need to track they have uncertain implications for data Repositories because we're not sure at this point how absolute the protection might be that's conferred by a certificate of cadet Confidentiality in other words to what extent might human data in data repositories become subject to compelled disclosure For law enforcement purposes So there are a bunch of questions that are quite unresolved in terms of what controls we put on this use Let me quickly say of course That the use of DNA and law enforcement has very positive implications In particular the innocence project, which is now a nationwide project involving Units in several law schools around the country has been responsible for the exoneration of 200 individuals Who were wrongly convicted and who didn't have access to de identification DNA identification methods and with subsequent access Access have been able to be shown to be innocent Now this is a very welcome use of DNA Technology and it illustrates potential for profound social good It also carries with it its own train of uneasiness one Outcome of the use of DNA identification for exoneration has been to illustrate the fact to demonstrate very vividly That our justice system all doesn't always work in fact It's been estimated that above and beyond the 200 who've been exonerated to date There are probably thousands of others falsely convicted and in currently incarcerated So DNA exoneration is a really good thing, but it's also a process That has demonstrated flaws in our justice system And as a result and this is again another positive effect of this social use of genomic information There are changes that are going forward those changes are based on the flaws that have been identified It's very clear now as one analyzes the the cases of DNA exoneration That there have been flaws in criminal procedure around witness identification around the use of informants to identify suspects and Perhaps most concerning for our purposes around the handling of evidence by law enforcement Including handling of evidence in crime labs and it is worth noting that the standards for crime labs are generally lower than those for clinical labs so I think what we can take from this is a lot of issues of policy in terms of appropriate procedure and appropriate interface of the Use of DNA information with the criminal justice system a lot of Concern about what's been revealed and basically about the imperfection of social structures So that no matter how good the science is it lives within an imperfect human world That make us I think uneasy about the possible use of Research databases as part of this system Clearly there's another issue in terms of what we might call Social uses or non-medical uses of DNA information that we've seen a lot of publicity about and I'm showcasing in my slide here to To particular case examples that I've received a lot of media attention And that is the potential that genomic research will cause offense In both of these examples, I think we have situations in which researchers who were using genomic data with very good intentions They were doing what scientists do. They were trying to answer questions that arguably had scientific interest In one case the Havasupai case There was the claim that blood samples taken for research on diabetes were then shared with other researchers and Used for all sorts of questions that the Havasupai tribe found offensive and certainly would not have okayed one of those was Migration patterns another was patterns of intermarriage within the tribe And as the article here shows Huge lawsuit was brought to bear The first judgment has gone against the tribe. It's now on appeal It's gone against the tribe for a very understandable reason, which was the Individual consent form was pretty vague and would have appeared to allow a lot of different kinds of research But the tribal procedure that okayed the research was clearly focused on health issues of concern to the tribe The message here is strict legalities probably don't address all of the ethical concerns We do have to be concerned about what participants want us to do with their data What questions they would like their data to be used to answer? This is we're not in an era I think where we can say scientists know best about what questions matter in fact We can't do the research without the participants participation we need their samples and we need as a society of researchers and others to pay attention To what researchers part brought what research participants want to contribute to in the way of research questions the other example the example where studies of selection of a polymorphism Led to certain conclusions that probably the science couldn't support About potential differences in evolution of brains Illustrates also the social context in which genomic research occurs Genomic researchers really can't just think about the narrow Questions defined by genomic research, but the broader social implications of of their research Let me now talk about a very specific issue That has to do with Decisions we make in data repositories Because it speaks to another way in which I think there's an interesting Sensitivity that around Participants concerns and that is between anonymization and de-identification Anonymization has been used for a long time as a way to take a data sample that's identified And and use it for a purpose a scientifically valid purpose that wasn't previously Anticipated this is a really crucial issue because now in this genomic era We can see loads of ways in which data that's been Already collected for another purpose could be applied to a new purpose that couldn't even been anticipated at the time the data was was collected and The idea is that with anonymization You can do this ethically You you you guarantee that there'll be no linking back to the individuals who gave the samples But there are problems as dr. Gibbs and others have pointed out in fact Anonymization is no longer an absolute process. We used to think that we could do it. Absolutely. We now know that we can't And that's true not only because of the amount of genomic data We're collecting but also as we collect large genotype phenotype databases even the complexity of the phenotypic data May allow some degree of identification Also of concern and I think this is very relevant to the Havasupai case If you don't remove ethnic identifiers you have the potential for group harm clearly Researchers were interested in using the Havasupai samples to study migration Because they were labeled as Havasupai And that made them interesting but from the tribes point of view it also resulted in data That could be linked directly back to them which they did not want to have linked directly back to them Anonymization also and this is the final point prevents any possibility of return of results to participants This has received a lot of discussion in general I think there's an understanding that we should be cautious about returning results to participants when we don't know what they mean and In the early stages of investigation of any health problem It's likely that we won't know what the results are are going to mean. We're in the stage of gene discovery We're beginning to put together associations between gene variants or genomic structure variation and Certain health outcomes but in that stage we're still Understanding and we don't necessarily have information that we feel would be responsible to give back to participants But of course that will change as we succeed as we succeed in Understanding disease biology increasingly we will find ourselves in situations where we have information flowing from our research Including our data repository research that does have clinical meaning. What should we do about that? Well Hank really has just written a very interesting Discussion of these issues and I think he Asks us in a very important way to think very seriously about this issue In fact in his words the choice not to return clinically meaningful results Seems at least in the extreme situations immoral possibly illegal and certainly unwise and his example is Research that finds a deleterious msh2 mutation. This is clearly actionable genomic information. No question Someone who knows he or she has that mutation ought to be having Aggressive colon cancer screening starting from the early 20s something we would not offer to a person who didn't have that genetic risk The question of course is how often are we going to find that kind of information that's so significantly actionable that we Create a situation where researchers may have an obligation But what Hank is also saying is if you if that data flows from Research that was done out of a data repository If the researcher has no way to follow a path back that is to give the information back to the contributing researcher Then a very important Opportunity will be lost to do good, but also He points out an opportunity That if the story unfolds could be very damaging for research That is he's concerned not just with the legalities and the ethical components, but the practical Reality that if someone finds out after the fact that the researchers knew something that could have prevented a very Dramatically bad health outcome. That's not going to be a good thing for science. So we have to think forward so The example he gave had to do with families like this and the possibility that we will be finding Occasionally in our research gene mutations that have this kind of effect and have major implications for health care as we go into the increasingly Successful study of contributors to complex diseases We will undoubtedly find the gene variants that identify the kind of moderate risk family And then perhaps even gene variants that are significant for these individuals. These individuals have a grandfather with colorectal cancer We don't identify their risk as being on average above that of the general population But it's certainly possible that there are some polymorphisms in the family that if an individual inherited might have clinical implications might even lead to some differences in in what we would do and When we think about the range of issues that that arises from this kind of spectrum We realize that we have to think very carefully About how genomic information will or has the potential to improve health care Because it's only as we think prospectively about that That range of options that may flow from from the research that's going forward that I think we can begin to prepare for the kind of procedures both informing participants of actionable Genetic information, but then also preparing for the integration of genetic information into the health care system So I think it's fair to say that when we think about genomics moving in to the clinical realm That there are two general pathways that are going to be followed One is the identification of risk to inform preventive care and the other is the development of innovative therapies The the attention has been on the identification of genetic risk to inform preventive care And I'm going to give you a cautious read on the opportunities for that There are no question that there are opportunities the MSH to mutation story that I just discussed is an example I think those opportunities may not be as unlimited as is now being Projected and that means we have to think very carefully about that use of genomic information And then clearly we hope to move on to the stage of innovative therapy So I'm going to use this particular paper. It was one of a spate of papers that have come out recently on Gene variants associated with age-related macular degeneration To illustrate a point in the complexity of using genetic information as a risk predictor in clinical settings this particular paper looks at genes three genes and several variants in these three genes that Related in one way or another either by increasing or decreasing to risk of age-related macular degeneration and these genes Encompass the two biological pathways that we now know are involved Angiogenesis and the complement pathway and what you'll see is that the researchers in this study were able to to do combinations of different gene variants and to Estimate odds ratios and from there to estimate Lifetime risk of age-related macular degeneration based on a genotype constructed from variants in these particular three genes and They are showing that we can do exactly what is hoped Which is that we can identify some individuals who are at very high risk of age-related macular degeneration So this is a little bit like finding the BRCA one mutation carrier Amongst women at risk for breast cancer We can intuit that there's a lot of value to that information We don't have as clear a path to prevention for these individuals as we do for breast cancer But we're certainly very interested in thinking about how we might bring to bear Medical intervention to reduce risk of age-related macular degeneration in these folks. We also find folks that are at significantly reduced risk and the first point is that both of those categories are very rare so less than About 2% of the population have less than 1% lifetime risk of AMD and about 1% of the population Has a greater than 50% risk There are others that have increased risk not to that degree but potentially clinically important But the other really important message about this particular profile, which is just three genes We know we'll have profiles with many more genes eventually Is that most of the individuals have risks that are just a little above or a little below the general population? And the point here is that unlike the BRCA testing situation where we're picking out those that have mutations from the general population If we use genetic profiling we're always going to be creating a range of risk And so we're going to have a have to do devise thresholds We're going to have to think very carefully number one About what where we draw the threshold for actionable information and number two what we do with all the rest in Particularly what we do with people that have just marginally increased risks that we're not quite sure what to do with but informing about them about it might well generate worry or concern or Interventions of unproven value that might themselves cause more harm Than the benefit that flowed so these are these are serious issues as we develop this This potential for identifying genetic risk and just to carry that thought further as we think about where we draw the threshold We have to think about what we're going to do with the information right now when we think about prevention of age related macular degeneration Really all we can do is look at risk factors that we might address one of them is family history That's presumably or a major component of family history is what we're measuring in this genomic research Smoking obesity exposure to bright light all of these things are risk factors smoking Probably the most profound But I think we could say fairly firmly that we don't want to use the kind of genomic profiling that I just showed you As a motivator to decrease smoking It might decrease smoking and that would be good But what about all the folks at very low risk who now know they don't need to worry? In fact when you think about the use of risk information in clinical care It works best if it results in a binary decision Measurement of hypertension is useful because people who have it Need anti-hypertensives and people who don't don't when we have interventions that everyone Should be using smoking cessation being one of them Genetic risk is not really a useful way to get at that intervention now I don't mean to be pessimistic here in fact I'm very optimistic that the understanding of disease biology that is flowing from this genomic research on AMD May well lead to new ideas about prevention and at which point it's possible that genetic risk Profiling will be useful. It's also possible that those new insights about prevention will be broadly available Broadly applicable to the population We'll have to see so we have to think about thresholds for Clinically relevant risk and they have to be thought about in terms of what we're going to do with that information And that's going to be different for every individual case So it's going to be a complex analytic procedure as we go forward Understanding complex diseases better Now there isn't additional complexity and that is when we say actionable Genetic risk information what actions do we include? We've known for a long time about the association of a poe for a poe genotypes and specifically the Poe for alleles association with Alzheimer's disease at the present time We have no treatment available to reduce risk that may change and if it does the clinical implications of that information Will change in the meantime though there's been a research study going on the reveal study That's been looking at whether people who are who have a parent with Alzheimer's disease are interested in this information And it turns out some are So they're now reporting on their experience and they have a group of adult children of parents with Alzheimer's disease Who've had a poe for testing that is to determine whether they have one or two alleles? and Among that group that have undergone testing some have chosen not to have the results returned to them That's the no disclosure group some have found that they're a poe for negative some a poe for positive Now if no all of them have a family history of Alzheimer's disease by usual risk calculations all of them presumably have some degree of risk But the a poe for at least in some cases adds what's interesting about this data is the significant finding Which is people who find out that they're a poe for positive This is on self-report of actual behavior not just thinking about it, but whether they did something about it It looks like the folks that were a poe for positive got that result and acted on it in terms of purchasing Being much more likely to turn purchase long-term care insurance It's possible that there's an effect here and it's possible that there's an effect here with health and life insurance respectively, but we don't That didn't reach statistical significance Well clearly the social implications of this are quite considerable One of them is should we be including risk information that isn't actionable in the usual Medical sense that isn't part of health care in the usual sense into health care with the costs the resource uses that come with it Another question is is it okay for people to use genetic test results to determine whether or not they buy Long-term care insurance These folks are buying more because they quite legitimately Imagine that they're more likely to need that insurance And that means that they are in a sense gaming that the insurance system Is this okay from a societal perspective this is this a good thing or a bad thing? In fact, there are complexities to it in terms of how we fund long-term care In terms of how we use information this kind in health care Now let me get back to therapy What I want to argue is that there's a lot of complexity to using genetic risk information in health care Sometimes it's going to provide extraordinary benefit other times. It's going to be a very questionable use of Patients time and doctor's time and resources and may lead to lots of uncertainties or perhaps Untoward social effects and we're going to need to think about that as a society But what we hope for where I think the big gains will come though It's hard to say how soon they will come isn't the production of innovative therapy Genomic analysis is clearly already leading to an increased understanding of disease biology And then we can follow the pathway down and the question marks are there because for any given set of Disorders we don't know how much benefit we're going to derive But what we hope is that as we understand we develop hypotheses for innovative treatment that then we are able to develop Those treatments that we are able to confirm in clinical trials that the benefits there that we Hypothesized was there and we're going to successfully introduce into clinical practice all of these steps are difficult Success is not guaranteed, but this I think is where the big gains will come We can't say for any given disease that we're sure we will succeed in following this pathway But I think we can say with assurance that across important health conditions That the population suffers we are going to have successes by going down this pathway that genomic research takes us to And we really want therefore to encourage this pathway to look for the successes and maximize them as they come As we do so I found this reflection by Robert Califf useful He has an interesting reflection on the lessons to be learned from cardiovascular clinical trials now Cardiovascular medicine has been an area of extraordinary success over the past 30 years so he is speaking from an area of Applied research translational research and development of innovative therapy where there have been considerable Successes, but he says even within that context most effective therapies produce both benefit and harm We have to be mindful of that. We have to measure it Treatment effects are often modest. We need to be prepared for that sometimes unexpected And here's another really important point complex decision-making is prone to error That's another reason why we have to be careful how much We plan on doing nuanced stratification of genetic risk in the clinical setting because we may simply be introducing lots of opportunities for error and then finally long-term effects are difficult to predict and therefore important to measure so Let me reflect a bit on the challenges. I've said that as we think about this Wonderful research enterprise moving from answering interesting scientific questions To developing health benefits We need to think about two different categories under the category of identifying risk to inform preventive care We have to think very critically about when risk information is clinically useful How should the thresholds for action be drawn and who decides who's at the table when we decide what benefits matter? When we talk about development of innovative therapies I think we have a serious work to do in determining what incentives and regulatory systems will both Enhance innovation and make sure that the products that come to market are safe And in fact validated to do what they claim to do We also have an ongoing ethical concern in our society. How do we assure equitable access? I would argue that this concern has particular salience For the world of genomic research because of the extraordinary public support That has gone into the research and the large numbers of research participants now and in the future Who who will help us to get there to the benefits that we want to achieve? Obviously There are challenges because there are opportunities This is a glorious moment in genomic research and it's only going to get better Unprecedented opportunities to understand serious health problems molecular tools that Look like they will be important aids to developing new therapeutics. I want to emphasize the last two bullets though I think this is also a moment where there's extraordinary opportunity for interdisciplinary collaboration As the science goes forward the policymakers the legal scholars the ethicists People involved in health service and health resource decision-making should also be at the table thinking about this together All of the quote Elsie folks can't do a very good job if they aren't well informed by the science But I think we all do better if we work together and I will say that's a proud tradition of Genomics research that we try for that I think there's opportunities to do more and to do it better And I think as we do that we need to create meaningful opportunities for societal deliberation Because some of the opportunities involve serious trade-offs How much loss of privacy to get certain gains? How much exposure to genetic risk information or in order to gain certain health outcomes? And these are things that need to be decided together within society and that's a methodologic problem How do we do that? I'll just end with Knowledgements of all the wonderful people in my center and our funders the National Human Genome Research Institute and the National Institute for Child Health and Human Development Thanks Any questions from the audience for Wiley? I'll start with one. Can you maybe Richard? Thank you Wiley. I have one of those kind of alarmist questions So if there's a proliferation of very inexpensive genetic tests from third company third-party sources So you can send off your DNA and get back these all this data If the will the landscape shift because of that, you know, let's say we see in the next year a real Proliferation of that beyond our wildest dreams technically and expense wise Do you see that as being changing the whole landscape and more specifically? What are the checks and balances in that scenario? So you're talking about proliferation of genetic susceptibility tests that then get Marketed direct to consumers. Yes. Well, I think this is a great concern, but I I think we should And and I think there are two concerns one is that testing will be done that provides no value Potentially we certainly see tests on the market that we question the scientific value Of and I think the other is the risk that this will give genetic susceptibility testing a bad name I Think this careful rigorous approach that asks what are we going to do with the information? And how are we sure it will help people is really crucial to make sure that we use this technology? Wisely, but I think it's important to distinguish between what I would call the use of a genetic susceptibility testing medically in which case we really have to have the outcome Data we need to know that we're using risk information in a meaningful way and have the opportunity to improve improve health outcome from what I might call genetic Genetic testing as a consumer product So there are tests that are neutrogenomics. There's been an OMB analysis. We know that most of the claims don't make much sense However, it's not clear that it's really harmful to people to get tested Other than they you know, they paid some money for it In other words, it might be useful to think about the analogy of an exercise piece of exercise equipment People buy it with the notion that it might help them that might have improved health outcome Maybe it does maybe it doesn't but from a consumer product point of view what we want to make sure is that they don't Have a risk of a serious fall when they got on the machine that the machine works That it's not going to fall apart in a piece hit them So I think we have to accept probably in our society that there is going to be consumer Directed testing and some of it may be consumer products, and that's okay Nevertheless, I worry that we don't have enough regulatory oversight to ensure the truth in advertising piece That ought to be part of a consumer product In this era of high throughput whole genome association studies It strikes me that the results are that you get a p-value. That's really really low, which means it's believable But the relative risk is very low But 1.4 1.2 1.3 and that's not necessarily reflected in the press releases in the newspaper articles So are we confusing the general public by saying we found the diabetes gene or the obesity gene or things like that when the relative Risk is very very low Yes, so the point is that the rel that we can have Tiny p-values p-values that we all love and the relative risk can actually be low and the absolute risk is even less impressive Yes, I actually think significant associations with relative risks under two are a priori a very questionable Clinical significance, I don't think there's any question that there's misleading presentation of it Not just by media, but perhaps sometimes by scientists Who are trying to frame their research to show what's interesting about it? And I think we do have to think about think very carefully about how we frame data in part because it will gender unrealistic expectations genetic risk Profiling as I say will have clinical value But it will also generate a lot of indeterminate information of unclear clinical significance And I think when we talk about the gene for Crohn's disease without that complexity the complexity that you just pointed out About the low relative risk We're not creating the right environment for conveying that message