 Okay. Now that your glucose levels have been restored, let's get back to the agenda. We're going to hear next from Terry Minolio, the Director of the Division of Genomic Medicine, who's going to give you a report on a recent workshop on Stevens-Johnson Syndrome, Toxic Epidermal Acrolysis. Terry? Terry? Yeah. So, really, you must practice that. Put it on. There we are. That's the English pronunciation, I think. The necrolysis is the British one, so what do I do to get my slide to come up? Because I had it up here before. Oh, Eric. Would you say? Yeah, that's right. Well, it's always your fault, you know, no matter what happens. So, great. All right. This was a meeting that we held now about three months ago or so, sorry, six months ago. This was in March. And we came to this topic because of a series of meetings that we've held as part of the Genomic Medicine Working Group, which you'll be hearing about next. But just to remind you, we do have a working group of this council that focuses on genomic medicine implementation in particular. And if this isn't picking me up, or I'll just hold it. And one of those meetings then focused on international applications, sorry, so that's this one. The global leaders meeting that we held in January of 2014, where we brought together a bunch of groups that were implementing genomics around the globe. We were thrilled that we actually had parts of the world that we don't normally see in our meetings. And a particular group in Thailand really blew a lot of us away with a program that they were doing in pharmacogenetics and adverse reactions to drugs. So they had a number of things that they were talking about, and this is a difficult picture to look at after lunch, so sorry about that. But they were doing, working specifically in Stevens-Johnson syndrome, Toxic Epidermal Necrolysis, or SGS-10, which is a severe, probably the most severe, cutaneous adverse drug reaction, if not the most severe adverse drug reaction, in which essentially an immune reaction, autoimmune reaction, is set up that causes, can cause up to the entire skin surface to slough off, both externally and internally. So the respiratory tract, the GI tract, the corneas, it is just a horrible, horrible disease. I saw one case of it as an intern, and not something that you ever want to see again. And what happens is what's shown here is basically there's separation, there's death of the cell layer between the dermis and the epidermis, and the epidermis just comes off in sheets. So this was something that until about 12 years ago was believed to be totally idiosyncratic. Nobody had any idea what would cause it. It was just like lightning striking. Until this group in Taiwan started to look at cases because they seemed to have a higher prevalence incidence of it in their country. And in particular found that in 44 of 44 cases, a particular HLA allele was present in people who developed this condition after carbamazepine, which is an anti-epileptic. And it was this, the particular allele was only present in 3% of those who were tolerant to this medication and about 9% of the general population in that country. So this was a very strong indicator that this particular HLA allele was a risk allele for this condition. Subsequently, there was another allele in more European populations, HLA 01, and shown here are the associations with that, sort of a 26-fold odds ratio with 3101 for SGS-10, and then for other types of milder hypersensitivity, a 9-fold odds ratio. So still as a cardiovascular epidemiologist, which I started out as, we never saw odds ratios of 9, and certainly not of 26. So there are a number of alleles like this that are pre-exposed to this condition. And this led the FDA to issue a warning, basically saying that anybody who was going to get carbamazepine, who was of Asian ancestry, should be considered for screening for this allele. And this was criticized because it was basically noted that clinicians have to determine if the patient has ancestry across broad areas of Asia, and how is somebody supposed to know that? Wouldn't it be way better to just have the testing available and then say, you know, have a little pop-up in your medical record that says, caution this patient carries this risk allele when you're about to prescribe this drug, you know, consider prescribing something else or watching carefully? And what the type group showed us was something that they have done that is one step back from that, but still involves screening large numbers of people who are about to get these drugs, and they then produce a card that looks like a credit card that has a patient's name, the outcome of the assay, the date, the interpretation, and this particular person carried this risk allele, and they were high-risk from this. You turn over the card, and it says, you know, this person is at high risk, please contact us if you want more information, but consider not using this drug. And they give these out to their patients, they screen for about 10 different risk alleles for different drugs that are commonly associated with bad reactions in their population. And they then showed the cost effectiveness of this approach in Thailand, and you can see here in criminal cost effectiveness of about $7,000 per quality adjusted life year gained in certain populations of patients, the sort of standard threshold for cost effectiveness is about $50,000 in this country, $50,000 per quality adjusted life year gained, that's based on renal dialysis, and actually I think it's been inflated now up to even 100,000. So $7,000 is really good, and something that we would consider implementing. Recognizing that that cost effectiveness ratio depends on the prevalence of the allele, if the allele is much rarer in your population, you're going to have to test a whole lot more people. They estimated in Thailand you'd need to test about 350 patients to prevent one case. And they've shown then that since they implemented the screening, their incidence has come down really quite nicely, this is just in men and women here. Because this was cost effective, the government, which is the single payer of their health care system implemented universal screening, they've had declines as well as in Taiwan have been reported in Malaysia, Singapore, all have shown cost effectiveness as well. So a very exciting story, and one that really grabbed our attention when we heard about this, and so we kind of came back and said, all right, how much is the NIH doing in this area, what's our portfolio of research in SJS-10, because it was fairly clear to us that there were more variants that could be found and other ways to go about dealing with this. And unfortunately, it was kind of a barren desert, we had maybe one tree standing here that was a single grant in a very, very basic area. So talking with our colleagues at other institutes, it was pretty clear that this was an area that not much was happening. And one would think because this is a skin condition, obviously the skin institute would be heavily involved. It's actually not a skin condition, it's an immunologic condition, it's just the skin is the end organ. So they didn't have very much in it, but they were interested and wanted to get involved. Obviously the Institute of Allergy and Infectious Diseases in Immunology has a footprint in drug reactions as well. Certainly this one now being genomic or genetic at least, we would be involved. The NIDDK has recognized a number of alleles or helped to identify really alleles associated with severe drug induced liver injury. And so they have an interest in the area. And CATS, because this is a relatively rare condition, particularly in the US, their Office of Rare Diseases, is relevant. The Neurology Institute, because many of the drugs that cause this are used in patients who have epilepsy, neuropathic pain, et cetera, so they were quite interested. The Food and Drug Administration, because this is a pharmacosurveillance issue, one that they wanted to get involved in and have been involved in. And then there were other groups that had some lesser involvement. The I Institute, because there can be fairly severe sequelae, including blindness. Our clinical center does screening for some of these variants. The General Medical Sciences Institute is obviously involved in pharmacogenetics and has helped to identify some of these early associations. And NIMH were some of the same reasons the drugs are used in schizophrenia and other conditions. So we kind of got together a coalition of these groups and put together a workshop that was held in March. This was to review the current state, sort of where we are, what our knowledge is, look at the role of genomics and pharmacogenetics in this condition and how we can help to eradicate it. And then identify gaps and needs and priorities. The research, this is the planning group, ably led by my colleague Carolyn Hutter here at the Institute. And this is the group, handsome that they are. And we reviewed the current state of knowledge. By the way, all of this is on the NHGRI website. The sessions were video taped and archived. And then the slides are up as well. And you can see the incidence quoted. It's rare, about 1 per million of the general population, but about 100 to 1,000 times more common in users of high-risk drugs and about 1,000 times greater in HIV-AIDS. So here's an area where we really can have some relevance in that. These are some of the commonly implicated drugs, but there are others. Carbamazepine is anti-epileptic. This is used for gout. And there's concern that as its use increases, we're going to see many more cases of this. Because the risk alleles for this are not limited to the Southeast Asia, whereas the identified risk alleles for some of the others are. Tromethoprimsulpha is bacterium or septra. It's a commonly used antibiotic. And this I mentioned is an antiviral for HIV-AIDS. There's a high mortality, 30% to 50%, in fact, and higher at the ends of the age spectrum, as you might imagine. A high risk of recurrence if you're exposed to the same drug or a chemically similar drug. Risk seems to be higher in Southeast Asia. More than 80% are believed to be genetically mediated, though there do seem to be some that may be mediated by other means, rather than specific drugs in response to infections, viruses, and things. And there have been multiple variants identified since this initial identification. And as I mentioned, screening efforts have actually reduced incidence in several countries. We pulled together some data from the people at the workshop in terms of the frequency of the particular alleles that are associated with this. They're shown down here. 5701 is associated with the back of your sensitivity, which is one of the first major genetically mediated adverse reactions that has been essentially gotten rid of by genetic screening. 5801 is for alipurinol, and then 1502 for carbamazepine. And if you look at a scale of allele frequencies, you see that they're really only a couple of percent, at most, in people of Mediterranean ancestry. But if you go up to India, well more than 30% allele frequencies here and in other parts of the world, down in Singapore and Malaysia. So it really does vary, as does everything else, genetic in terms of ancestral populations. Shown here is sort of a timeline from Elizabeth Phillips and her group at Vanderbilt, showing that on the Class I side, this was the first association identified in 2002. The back of your wasn't Stevens-Johnson. It was a similar kind, but not quite a severe of adverse reaction. Then the Chung group with carbamazepine, and you'll note all of these HLA, many of them are for multiple drugs. So the allele sensitize you to multiple drugs that aren't even necessarily chemically related, and then a number on this side as well. Making you wonder what is it that's going on in this region of the genome. Obviously this is the part of the genome that presents non-self peptides to the immune system to be gotten rid of. And somehow that message gets messed up, and a drug does the messing up, but it's not at all clear how that happens. Since these initial identifications, there have been phenotype standardization projects. This is one that was published. You'll notice a name that you might recognize of a group that worked together to come up with a standardized phenotype. And then kind of a roadmap for identifying likely cases. This isn't an easy diagnosis to make, particularly early on. And then there have been algorithms developed for assessing drug causality. One of the challenges, particularly in AIDS patients, but in others as well, is that patients will sometimes be started on multiple drugs all at one time. And you don't know which drug it is that actually is doing this. And this is the algorithm there. And there's also a standardized algorithm for severity. So a number of tools for research now in place that weren't available even 10 years ago. And the number needed to test is shown here for a back of year. You only need to test an estimated 13 people because the predictive value is relatively high, as well as the prevalence is a little bit on the high side. Whereas when you get down to some of these other drugs with carbamazepine in the US, you need to test about 1,000 people. Fluoxicillin was a rare allele and a rare problem. But it was a serious one when it happened. So you need many more. So again, we need to know much more about which variants and in which drugs are cost effective and relevant to be tested. So in addition to our current state of knowledge, there's a lot of ignorance. We really don't understand the mechanism of the immune reaction at all. We had some HLA experts in the room, and they sort of scratched their heads, and they said, we should look at expression of HLA alleles. Because one of the challenges is that maybe one in 10, one in sometimes 20 or so of people who carry this allele will actually react to the drug. And thank goodness that they don't all react. But on the other hand, how do you then pick out the people who will? The mechanism of cellular damage is not understood. The high risk chemical features of the drugs, what is it? Is there a side chain that's causing a problem or whatever? Current treatments are really quite ineffective and immune-suppressive treatments just have not worked. And the treatment is largely supportive frequently in burn units. These people end up in the burn unit sometimes for weeks or months. And as I mentioned, severe sequelae and can include blindness, incidence estimates are really poor. This one per million is frequently quoted, but a recent study from the partners group suggested that it's much, much higher than that. Maybe even 375 cases per million. And these were not just looking through ICD codes. These had some degree of validation, although not as heavily validated as some other studies. And there are no in vitro tests for the causative drug. So as I mentioned, if you have somebody started on multiple drugs simultaneously, you have to stop them all and then kind of reintroduce them very gingerly and hope that you can pick up which one is causing the problem before all heck breaks loose. In addition, there are many drugs that get into clinical use with having this complication. And you don't really know it until you use it in millions of people. And if there are a way to identify this preclinically before those costs are spent in drug development, that would be a tremendous thing as well. I just show this here just to show that there are at least three and probably more models of what the immune mechanism here is. And no one really is sure which it is and how it works. So what follows is a number of very heavy text slides that are basically research questions and recommendations from the group. I'm going to flip through them relatively quickly. They are on the website. But I think things that we'd want to think about and where we're asking for your help is where should NHGRI go in this, recognizing that we need to do it in collaboration with other institutes. But what can we learn from ethnic specific associations? There are some associations that only show up in certain ethnic groups. Why does only a minority of risk allele carriers develop this on exposure to drugs? Can expression levels help us? Why are many of the same alleles implemented over and over and over again? What's the impact of rare variation in HLA? Obviously, if you have an allele that's named, you've seen it in enough times to get around to naming it. And the rare variation is this is the most variable region in the genome. What's the impact of that? Is it possible perhaps to get DNA samples from drug trials that have been halted? There have been a couple of really promising drugs in the diabetes field and that that have been stopped because of this complication. Couldn't we get samples from those patients and try to figure out why they're developing it? Could we use preemptive HLA typing in clinical care, maybe not just for this, but for a variety of conditions? Many people get HLA typing if they're going to be a donor, bone marrow donor or other things. Could it be useful in a broader sense in treatment of rheumatologic conditions, autoimmune related conditions? Could we use that information elsewhere as well as in pharmacogenetics? What impact might that have on clinical care outcomes? Can we look at retrospective EMR data across EMR linked biobanks to try to find out a bit more and identify more cases of this to undergo study? There were, as I said, several high priority research recommendations. Again, I'm not going to go through them in detail just to mention that there was a number of basic research questions, clinical research, low cost testing was a high priority. Pharmacogenetic and outcomes research, particularly cost effectiveness research, race ethnicity information, there's precious little known about American race ethnic groups. A group that hasn't been studied hardly at all is Asian Americans and Asians in other parts of the world where we know that Southeast Asians are at high risk in their native countries. Sort of overarching and facilitative research, can we combine with international efforts? There are a number of international efforts ongoing. The US is doing almost nothing in this area. Is there a way for us to work together using standardized case definitions and other things that I had showed you? Engaging burn units and specialists involved in the care of these patients. This is where they end up in the US and Canada and that's a good place to go to try to find them and to find investigators who want to do something about them. And we need some, there's infrastructure needs as well. And international consortium would address many of these research questions that have come up. Pharmacosurveillance is another whole area that we kind of leave to the FDA but to the degree that we can develop the phenotypes and other electronic record tools that can help them. We would like to try to do that. So I did want to kind of comment as we had talked a little bit earlier this morning about, when do we go genome wide and when do we go disease specific? And obviously NHGRI is on this continuum where we have a focus being genome wide more close to the nucleotide, close to the sequence. We tend to be resource building and tool building and paradigm setting. So when we get a little more specific it's because it's kind of the first thing in an area and we want to follow it forward and then use that as an example for other fields. Whereas things that are more disease specific or specific problem solving tend to us to be in the realm of the disease specific ICs. And so clearly this is something that really kind of falls on this boundary and we need to work together with those groups. So where we kind of came down is that a role that we could play is to stimulate high priority research. And it occurs to me if we have somebody on the phone they don't have my slides. So I'm very sorry. Is that Howard on the phone? Oh, I don't know who he is, but there actually I did something to Howard. So one thing that we could do is a large scale effort to identify risk alleles to include U.S. non-European groups because most of this work has been done in Asians and Europeans. Assess rare variation in expression of HLA, examine variable penetrance, look at biologic mechanisms of risk. All of this requires large numbers of patients carefully characterized particularly early on in our course. And assess the impact of preemptive genomic testing. And how we might do that, well the house that we need or the tools that we need would be facilitating comparison and harmonization of phenotyping with standardized case reports. We're working on that already with this group and have a subgroup that Carolyn and now Jeff Steering are leading to try to get a much more harmonized approach to reporting cases. And then working with the sister ICs that I mentioned to you earlier to invite applications and contribute to ongoing international collaborations. So that's an area that we will try to pursue. And also facilitate interactions of relevant investigators at various meetings. We're trying to call attention to the area. We've drafted a white paper for publication that will be submitted soon. We'd like to publish webpage-ish summaries and calls to action in the specialty journals because this is something that really doesn't belong to any given set of investigators and no given medical discipline. But it needs to go be notified to people in dermatology and neurology and people who use the drugs, rheumatologists, et cetera. So we're trying to get that word out. And working with patient advocacy groups, we had patient advocate groups at the meeting itself. And they were extremely helpful in identifying the fact that these patients suffer greatly. And they are very motivated to try to find the causes of this condition. We're working currently on announcing, again, in collaboration with multiple institutes, a program announcement that would just basically say this is a priority area for our institutes. And the details of that will be forthcoming. But that was one that we thought we could do relatively quickly. We'd like to pursue some feasibility, cost-effectiveness, and perhaps other pilot studies in the coming year and in fiscal 17, likely through administrative supplements. And then when our other institute partners are ready to do something further, maybe prepare for a large-scale risk-aleal finding effort. I think Jeff mentioned earlier that we have a kind of, in our thinking, a large effort, potentially, for genomics of adverse drug reactions that would be broader than SJS-10. SJS-10 can be a good paradigm-setting thing, but HLA seems to be relevant in many, many conditions and many adverse reactions. So we could broaden that beyond. So, and again, just to thank the people I mentioned, Uber, who helped put this together in our planning group. And then shift over to you and ask the questions that I hope were transmitted to you as we talked with you in our pre-council calls. Please advise us on what our role should be. I think right now we're playing a convening role. We also have some really cool findings that are genomic and that can drive people towards saying, gee, maybe this is something that doesn't just strike out of the sky, but it's something we can do something about. And then help us understand priorities for pursuing the workshops, recommendations for research and implementation. So we asked two council members to comment on this who have some history in this area. And I might ask Lon if you'd be willing to comment first and then Dan. Thanks, Siri. That was really helpful, I think, to frame things up. Very nice. There's this strange challenge, I think, in the drug discovery field where it doesn't matter what indication or what compound or what adverse event, most pass lead to HLA. And so at GlaxoSmithKline, we do a lot of preemptive type genetic research. We genome scan just about everybody that's in late stage trials before anything happens. But really where the value comes out is when you get down to HLA. And so I think, from my perspective, there clearly is immediate public health benefit to better characterize the cases and the diseases and try to predict better humanity, better the things coming through. But really the big challenge, I would love to see people engage mechanistically, trying to understand why this is happening despite whatever medicine treatment and combination you have. And as far as I know, and maybe Dan's or someone else's more up to speed on this, I don't think we've made much headway there. And maybe there's more and I'd love to hear more. But to me, I think both sides of that equation need attention, that is the screening and the more traditional genetic side. Maybe at a genomic scale, maybe not. But HLA mechanism is really the most compelling because if we could understand it, better we could predict in case we could. And it seems to be a really good example. It's like laring you in the face pointing to HLA. And so if we can understand it there, maybe we can understand it in other regions. Great, thank you. Dan? Well, I echo everything that Milan says. I think the tension that, one of the tensions we feel is this, there's a lot of enthusiasm for preemptive pharmacogenomic testing in routine clinical care. And at the end of the day, the HLA being related adverse reactions are spectacular when they occur. They're pretty rare and the drugs are not the most widely used drugs. So in the implementation space, we feel the tension that it would be lovely to have everybody's HLA haplotypes precisely documented and in their charts for when they might be exposed to one of those drugs. The issues are sort of cost-effectiveness of those kinds of approaches. And the fact that HLAB is one of the hardest regions of the genome to interrogate right now, I hope that that's going to go away. So I think it is a poster child for prevention though. I mean, the one prospective randomized clinical trial in the pharmacogenetics implementation space that's shown a spectacular positive result was on the back of your HLAB 50701. So that echoes what Milan was saying. And then the business of trying to figure out why this happens to some people and not others is a huge challenge. So this speaks to this idea of going back to the siloing or not. There has to be an interaction between the basic communities and the clinical implementation communities. So there's this drug, lupoxicillin, beef, star 57.01 related hepatitis. To do HLA testing to prevent one case, you have to test 10,000 people for a baccadar to the numbers out of 20 or 30 or something like that. So there's something else. And we've talked about what that else, something else, might be that needs to be thrown into the mix to figure out it's not just the HLA and the drug. It's HLA drug and something else. And there are ideas around that. And I think there are people working on that. This has implications not just for pharmacogenomics but for transplant genomics, for cancer susceptibility, for the better management of HIV infections. So I think that hopefully there's enough resources and enough smart people working on this that eventually we'll have an answer, eventually soon we'll have an answer. Thank you. Other comments? But I might just note that on the comment that you made, Lon, about how some people react and some don't. There are families where one person in the family has the variants, they react, and it gets the drug, and other people in the family get the drug, have the variant, very scary. And they have been out for three years and they never react. So why is that? The other challenge there is if a large proportion of people who have the allele would not react, you actually are shifting over to other drugs that may be more costly and could cost there, preventing one case. Again, if you could figure out who that case was of the 100, who is the allele carrier, you'd be more cost effective. I have two minds whether to make this comment, but I will. In Hong Kong, they implemented a carbamazepine screening program, and they indeed managed to see a real decrease in the incidence of carbamazepine-induced Stevens-Johnson syndrome, but an equal rise in the incidence of phenitone-related Stevens-Johnson syndrome. So the number of cases actually was awash. So that just reflects the fact that there's a public education or a physician education effort that needs to be made around this as well. It's not trading one drug in for another. So that speaks to what I'm saying. Especially one that's going to look similar. Yeah. Yeah. Yes, Bill. Yeah, as a pediatrician, someone interested in newborn screening, the numbers you present aren't so out of line with some of the things that they're screening for in a newborn screening, so I think they're a precedent for that. Well, I think, Terry, the last time I looked at rapid, cheap, but high-resolution HLA typing, it was sort of almost there or sort of there with next generation sequencing and costing maybe somewhere around $100. Do you know whether there's any push to prove that and make it cheaper and cheaper? Because $100 is not going to fly for newborn screening. Right. Yeah, I don't know. I know that the bone marrow field is what's been pushing this forward, and they've been quite effective at doing that. Dan, do you? I think this is an ongoing challenge. It is the hardest or one of the hardest regions that genome to enter into. I defer to some people like Jay to sort that out, of course. But I'm told, and I think that it makes intuitive sense, that some technology that does much longer reads should be able to sort this out. The problem is that it is a haplotype, and the short reads just won't do it perfectly. The imputation stuff you can get, and it works, we think, for Caucasian populations, but we don't know how well it works in other populations. We have evidence that it doesn't work as well in Africa. So all those methods are relatively indirect, and they're relatively imperfect. I think, yes, you're going to have a bone marrow transplant. You want to be HLA type. You don't want to be HLA imputed type. And I think the same applies to preventing really catastrophic drug reactions. You want the real data, not to say that it's imputed. So that's a long way of saying, I don't know. I guess one other thing I would say is that I think this does highlight one of the real issues when you have these cross-cutting problems in medicine where there is no institute that will take full responsibility for looking into this. And yet it's really interesting and really important. And so I think what you've done by pulling all these people together and having this meeting is fantastic, really interesting, and great initiative. I would love to see everybody from GM and NIAID right through to Minority Health Institute ought to be interested in this. To ask somebody whether they're Asian or not when you have Tiger Woods who's a Kalamazian, go after the genetics. Don't ask these silly questions. I completely agree, Bob. It really is a challenge when you have something that is this rare. It's very hard to get people to attend. I've never seen a case. It doesn't happen. So we will keep working on it. Actually, one of the first cases I saw actually is an interesting scientifically. And that was a young woman who had mononucleosis and got ampicillin. And somehow it's the combination of having mono and ampicillin that raises your risk much more than monoblon or ampicillin alone. What's going on there? And that's an interesting one. I was taught as a resident that the test for mono is you give ampicillin and they get a drug rash. And that's how you know you've got mono. But ours didn't usually go on to SGS10. Hey, shall I go ahead to the? Jay, did you want to solve our HLEB problem? I mean, I do want to say there are people working on this. And it would be great to bring them in. Peter Parham, my colleague at Stanford, and other folks at UCSF, we've got to be sort of HLA-typing program to figure out how to bring in next generation sequencing. And also, not only HLA, but CURE, often types. You actually need to look at both. And so I think it's a super great area to invest in. I echo what Bob said. I think that if it is a problem that no institute wants to actually grab hold of and looking at it from this point of view, makes it an HGRI problem. So we can leave. Well, and it's an HGRI opportunity. I mean, this is something we can really make a difference in. And it bridges lots of different parts of the strategic plan, from structure of genome to biology of genome to biology disease, proving health care, I mean, like post or child. Yes, yeah, absolutely. What's the current cost of HLA versus newborn screening out of curiosity? Oh, no. Well, in Iowa, we screen for, like, 60 things. And it costs $160 for all of them. And HLA is one for 100. Yeah, yeah. So it's still up there. Can I just add, because Bob rattled off a number of the institutes that would be interested here, it seems from the drug discovery perspective, where you're going to see these is when they arise really, really rarely. Because they made it through the discovery that you didn't see it in any clinical trials. Or when the risk benefit is a little bit more open. And that's why I think you see these sometimes more in oncology. And I would have put NCI on that very strongly as well. And that may not be SGS-10. But paths also lead to HLA for different. And NIDVK, because I think that there's hepatotoxicity that's also HLA-mediated. Oh, very much so. So J. Huffnigel was heavily involved with us. And he put together the whole drug-induced liver injury network. Yeah, so this is something I think we can get them. Hopefully, they'll get behind us. And, Lon, do you see things not make it through consequent HLA common reactivities? Is that? If there's any hint for most indications for type 2 diabetes, any hint of safety, it doesn't progress. So you never see it. So there's a lot of potential drugs that could have been discovered you'll never know, because they won't even make it into late-stage trials. That's why I raise cancer, because their risk benefit is more tolerable. And they do make it for it. So do you pick it up in animal testing? And can we get an animal model out of HLA animals? Yeah. Yeah, that's a tough one. OK, great. All right, I'd be happy to end now. All right, so then moving on to give you more of an overview of the genomic medicine working group, which was sort of the grandparent of this particular effort now. Telling you about a working group of you. So the advisory council has two major working groups that I'm aware of. One of them you just heard from before lunch, and now this one. And just to remind you who is on it. So this is the group. And then Eric and Laura and I are sort of the ex-officio folks from NHGRA. The GMWG was set up about four years ago, shortly after the strategic plan was released, to help us in charting this area, particularly to evaluate and implement genomic medicine, and review progress, identify research gaps, identify and publicize key advances. We decided we needed a series of meetings. We actually weren't sure it was going to be a series, but we knew we needed at least one, and figured there probably would be others. To help us explore this area, and also learn who is doing this work, bring them together, and form more of a community of effort. Facilitate collaborations and explore models for long-term infrastructure and sustainability. So as I mentioned, we've planned a series of meetings. These first three were in rapid succession in the first year or so after the strategic plan was released. And again, mainly to kind of develop collaborations and get people aware of what each other was doing, and not duplicate what each other was doing, because there was a fair amount of siloing and duplication. Our fourth meeting was focused specifically on physician education, because I think everybody recognized that that was an area that needed to be addressed. Our fifth was trying to develop federal strategies. That was one with the gentleman you're going to hear from next, and I shouldn't make it too first for a period. And really asking our other agencies, what should we be doing together in implementing genomic medicine. The global leaders meeting that I mentioned to you previously. And in October, we held one on genomic clinical decision support. And the report for that has gone in for publication. And then this most recent one in June was really kind of an overview. So the working group members said, you know, something that would be very helpful would be for us to take a step back and look at all of the programs all together and see how they fit together, where they could fit together better, and what the gaps might be. And just to kind of give you a feel for what's come out of these meetings, so our first meeting actually led to a separate workshop a few months later that we called ClinAction. And from that, the clinical genome resource, or the ClinGen, was born. And so that was kind of a direct descendant, basically, of that first meeting where people were saying, you know, we've got all these variants. We don't know what they mean. We all sit down in the room and try and hash it out. And it seemed to us that if people were doing that in separate rooms in 20 or 30 or 100 places around the country that we could organize and coordinate that effort and get it done much faster. The Emerge Pharmacogenetic effort also came in large part out of that first meeting where we recognized that pharmacogenetics was a ripe area and that we could be working in one of our large consortia to be able to implement it rapidly. Our second meeting was on collaborations, and that led to our Ignite program, where we really are trying to take centers that are expert in this area and then try to have them disseminate to centers that really don't do genomics at all, so family health clinics, community health centers, military, VA, other kinds of places. Our third meeting involved payers. We really wanted to work with payers to understand what kind of evidence they needed in order to be able to support this kind of reimbursement for these kinds of tests. Our fourth on education led to the Inner Society Coordinating Committee that you've heard about a few times since at this meeting. And the fifth one, I have dotted lines here not because I didn't want to imply that our fifth meeting produced CMS, FDA, or the Air Force, but what they did produce were closer collaborations with those groups in looking at ways that we can share information and collaborate to improve implementation where it's appropriate, not just willy-nilly. The fifth one also led to a trans-NIH working group within NIH. This is something we discussed with the Institute leaders at one of their strategic planning forums. And they basically said, we need to also know, everybody needs to know what's going on within NIH. And it was that group that we drew from when we put together the SGS10 planning group. Our sixth meeting, which was the Global Leaders Meeting, led to a global consortium called the Global Genomic Medicine Consortium, or G2MC. It is jointly hosted with the Institute of Medicine. And it will have another meeting in Singapore in November. And there are a series of areas that groups are collaborating in and also working with the Global Alliance for Genomics and Health, which is much more of a research effort. It's more of an implementation effort. You've heard about that before, so I won't go into it much with limited time. But out of that group came the SGS10 meeting that you just heard about. Our seventh meeting in clinical decision support led to a much stronger collaboration with the Institute of Medicine Genomics Roundtable. And there was kind of a plan hatched for a soup to nuts program in Genomics Clinical Decision Support at our meeting that is now being implemented in the Genomics Roundtable. So that's very exciting. And then our eighth meeting was just held in June. The objectives of it were, as often is, kind of, where are we going sort of meeting. So reviewing our portfolio, identifying gaps, identifying related programs for, among other centers, and ways that we can work together, research needs for ourselves and our partner agencies to pursue. And then we always try. And one of the reasons that we do these programs is large collaborative programs, is that they can have an impact on the field and sort of pushing it forward or pushing it in a direction that no single investigator can have by themselves, no matter how good they are. And then examine potential methods for assessing the impact of these programs. So we decided to kind of focus on the six programs that are the major genomic medicine portfolio parsing. You've heard these described before, so I'm not going to go into detail about them, but they're here. And our undiagnosed diseases, newborn screening, clinical sequencing, emerge the IGNITE program in dissemination and then the ClinGen resource. And then we also had a series of related programs that we wanted to hear from and have members in the room, some of them in the very basic realms, some in informatics, as you see here with GSIT. ENCODE was there, representatives from the ILM roundtable, some from the NCI clinical trials that are genomically driven, et cetera. So a large number that we tried to bring together and see where we had the opportunities. And then we asked for summaries from each of them. This is one from the clinical genome resource. It's basically a summary that describes who's involved, what the mission or the objectives are. Here's another one for newborn sequencing, a third one for the Centers of Menelium Genomics. And in these, we asked for just two pages to describe the objectives, what the funding period is, the working groups, the resources and tools that they've produced, publications, obstacles, and the approaches to meet those obstacles. And all of those are stored and available on the website for this meeting, which if you search Genomic Medicine 8, or if you just go Genomic Medicine Activities, the website for the meeting will come up and you can find these meetings summaries. And we're still referring back to them. They're very useful to have. One of the things we asked them to do, as I mentioned, was to identify their objectives. And then we kind of looked across them to see, well, what are objectives, particularly in the focus programs, that are really quite common across them? Because we would expect them to have some that would be in common. And as you might expect, these are very closely aligned with the goals of our division and the goals that the Genomic Medicine Working Group has outlined for itself. So integrating Genomic Data into patient care, incorporating actionable variants into the electronic medical record, particularly with using clinical decision support, educating clinicians and patients, assessing the outcomes, defining and sharing how best to do this in best practices, promoting interactions and collaborations, translating implementation outside highly specialized centers. And while this is Ignite's special role, actually several of the programs are doing this currently. There were also some that were pretty much targeted or unique to a given program. So the UDN is very heavily emphasizing improving genomic diagnosis and facilitating research in undiagnosed diseases, but Insight and Caesar do a little bit of that kind of work as well. Only Insight is looking at newborn care, electronic phenotypes are something that is unique to a merge, but is needed by all of these programs really. Identifying variants related to complex traits are showing a few of them as well. Pharmacogenetics is pretty much the domain of Emerging Ignite. Looking at penetrants is pretty much Emerge Alone, Standardizing Clinical Annotation, assessing actionability with mainly Insight and ClinGen, and then creating sort of genomics-enabled learning healthcare systems where we can actually improve care in a real-time way, something that Caesar Emerging Ignite are really focusing on. And we also identified barriers that face multiple programs. Lack of an evidence base was one we knew four years ago. It's one we have now, and it's probably one that will be with us for as long as we're in this area. The need for common data elements came up repeatedly as something that we want to try to encourage across the programs. Assessing the frequency and impact of variants, particularly in ancestrally diverse populations. Something that is a particular problem, and as Eric mentioned earlier today, that Ben's had an entire workshop on focusing specifically on why is it so hard to get information in ancestrally diverse populations when we must have it. Rapid evolution of evidence on variants, so how do you deal with the changing levels of evidence? A variant that you thought yesterday was benign, now a report comes out and it looks like, oh, it might not be so benign. A few more reports come out, and well gee, it's pretty clearly pathogenic. How do you deal with that in relating to patients and clinicians? The current limited usefulness and interoperability of clinical decision support system, so you build it in one hospital system and it works hopefully in that one hospital system but nowhere else. Regulations that impede return of results have really gotten in our way, in many ways, with the research efforts that we're doing, and we're working with our colleagues in the regulatory field to try to address those, but that's going very slowly and it is having an impact both on the research and the clinical care. Need for cloud computing is growing as the sequence data grow, so does the need for better ways to manipulate the data and transfer it to move it around when needed. Reimbursement policies and regulations, I mentioned earlier, I think, and a particular need for a bedside back-to-bench research. So how can we stimulate that virtuous cycle where we find something at the bedside and can take it back and really investigate it? We set up a series of panels. These were the panel areas and we asked them to then address a number of questions. How important is this topic in this area? What programs do we have addressing it? What are the gaps? What could be the synergies in the training opportunities? And then we asked each of the Genomic Medicine working group members to lead one of those and very much appreciated their efforts in doing that. There were a series of recommendations that got a lot of discussion. I may be heavily discussed is a little too strong, but I tried to pull out those that seem to be recurring just to kind of share them with you and again, sorry not to have gotten these out to you sooner, they are in the report and we'll post these slides so that if you want to refer back, you'll have them. But generating evidence was an area that everybody agreed was something we needed to do much more of data sharing and improved phenotyping, particularly standards for phenotype description that you could use from model organisms to humans so that you can go back to the bench when you've seen something at the bedside. A lot of interest in patient-oriented ontologies, so ways that patients can enter information on what their child has or what they have, what their symptoms might be and that could still be in translatable in ways that cases could be picked up around the world and linked together. Identifying and carrying out innovative studies, particularly engaging basic scientists more actively in planning our programs because we don't have that voice at the table very often when we have this table, but we need you to speak up more in terms of how we can do the science better. An interesting idea was to add family history to a large-scale sequencing effort so that we might end up with 20,000 people who had both a rigorous family history, not one that took weeks to put together, but something using some of the more up-to-date software tools that can be done relatively rapidly. If you had that on 20,000 people with their sequence information, imagine what you could learn about the added value of family history and how you need family history information to interpret particularly in rare sequence variants. Studying the impact and consequences of changes in variant annotation, facilitating implementation. Implementation commons was something that was very attractive to a number of the groups, putting tools into a common place where people could then use them and share their experience using them. Health disparities, and again, we discussed some of these last week and you'll be hearing about them more, but looking at specific health disparities, research questions related to genomics and implementation, developing dedicated programs for non-European ancestry populations and increasing patient engagement. Education and training is always an area that is needed, potentially joint training opportunities or best practices for clinician education. One of the things that we did with this meeting was to take, we had about 50 recommendations and we thought, well, why don't we ask people to sort of tell us what their top 10 is? So from the ones that I showed, I pulled out about 20 of them to report to you, but we then, through our, we Duke University was working with us on this workshop and so they sent out a little survey and said, just tell us what your top 10 are and rank them from one to 10. And then we kind of averaged those just to see which were those that tended to come to the top and you'll notice that up here, there are a couple that are really quite popular and then there's maybe a little bit more of a shelf down here at 10 or 11. So I just pulled these out to show you what the top ones were. And the top, number one was measuring outcomes of value to patients, clinicians, payers, and then a whole list of other stakeholders. So regulators, et cetera, et cetera. How can we learn what outcomes are important to these groups so that we can add them to our studies? It could not be very difficult to do and produce information that is much more useful to them. The family history tool came in second to us at 3.5, identifying types of evidence to collect and share across programs, accelerate rapid genotype explorations. One of the big concerns was that a lot of these explorations take a very long time and if you have a sick baby in the neonatal intensive care unit or if you have someone else who is really quite desperately you don't want that process to take months or years. Suggestion was made to consider cooperative sequencing groups to gather information about sequencing much like the cooperative oncology groups that the NCI has funded at the, basically the budget of this institute many, many times over per year. So we couldn't do anything quite that grand but at least something to consider. And then a number of others that I just sort of show here, facilitating coverage through evidence development, identifying payers needs, something that we are really struggling with. What is it that payers really want to see? Not that we're trying to get everybody to pay for these things, we want to understand what is appropriate to be paid and what isn't. And so how can we learn what evidence to generate for that? Post marketing studies, similar to pharmacovigilance studies but for genetic testing. So can you kind of gather information from people who've had genetic tests what their outcomes are and how they use that information? Some agreed upon nomenclature, variant definitions and allele identifiers that will help us with some of the trading information back and forth like the pharmacogenetic star allele system which is very cumbersome and very difficult to work with. HLA is another area where nomenclatures is a major challenge but are there other ways to work on that? Computable guidelines to put into clinical decision support systems, clinical trials of the added value of whole genome sequencing to more limited testing. So the question that we're doing in a merge right now we can afford basically a hundred gene panel and if we did a whole genome sequence and compared it, what would be the risks and the benefits? You might identify a whole bunch of stuff you don't know what to do with and you just run up costs without any added benefit. On the other hand, you might find somebody who's at risk for carbamazepine toxicity that you didn't know and they might be about to get that or a related drug. So our plans for follow-up immediately are to engage basic scientists. This came up over and over. We need basic scientists at the table and involved in developing our programs. So we've had a lot of emphasis on this sort of pathway from the bench to the bedside and we really need to go back and be sure that the function is explored and other things are explored in the laboratory. Areas that seemed like they would benefit from this included phenotyping that was compatible with model organisms to promote that kind of research. Variant, my pleasure was one, function to help us with clinical annotation. So those are the things that we are going to pursue and that will be sort of the focus of our ninth meeting, which will be held next April. Pursuing infrastructure needs, there were a lot of suggestions that were really kind of infrastructural, develop knowledge bases of what's going on in genomic medicine, a variety of other things. The implementation commons, common data elements, those are things we're going to have to struggle with a bit because they don't really seem to lend themselves necessarily and we would welcome your advice to funding opportunities where we say somebody, develop common data elements or increase patient engagement in that, but there are things we need to think about. And then comparative effectiveness research, I may have put this on anticipating the next talk to come, but we did hear the one, something that would be very useful would be whole genome sequencing versus targeted panels. Another would be whole genome sequencing with or without family history in large enough numbers to be able to draw some conclusion. So these are the people that were involved in putting this together and many thanks to them. And then back to you. So we welcome your comments on our recent activities and advice on the priorities that we've outlined. And I've asked Carol Bolt and Howard Jacob to comment. Carol, do you want to go first? So of course, these are all very broad ranging giant areas of activity that were discussed and outlined especially at the last genomic medicine meeting. And we've talked a little bit on our phone calls about how we can most effectively work together to move forward genomic medicine. And I think that I think there's a number of things in the priorities that came out and we're still discussing whether those are the right priorities. So really interested in people's comments about whether or not they feel like the ones that came out of the survey, which I forget, Teri, how many people responded to the survey? It was a small number. So we had about 88 at the meeting and 35 responded. Yeah, that's always a challenge, right? We hold these meetings and then you ask for feedback and you get feedback from a very small number of people and that ends up being your recommendations and priorities. So part of the benefit of bringing it out to the group like this is, is that a bias sense? Is there general agreement with these priorities or not? And just with respect to the planning for GM 9, you want me to comment on that as well? So as Teri said, one of the themes that emerged a number of times during GM 8 was this integration of basic science and how that can effectively promote and advance genomic medicine. And it goes back to what Eric was saying earlier this morning about the fact that we have all these variants come out in these genome sequencing projects and we don't know their biology. We don't know what they do. And to effectively integrate that information into genomic medicine, we have to functionalize them. And so this virtuous cycle that Teri had the slide on that, we want GM 9 to kind of focus on that virtuous cycle. How can we better integrate, especially the arc that goes from the patients back into the model organisms and how we develop the models themselves, right? When we talk about models, that means so many different things to so many different people and even being able to communicate effectively about what a model is and what it means is important to do. So that's going to be, currently that's going to be our focus for GM 9. Howard? Howard, did you want to talk? Yeah, so I think that's a great summary. I would only add one additional point to what Teri and Carol have mentioned and that is that connection at the basic research level. We need to also be able to do it and Teri did say this, but we have to do this much, much faster because as we're doing more and more of this clinical sequencing, it's right at the edge of care slash research and we're finding these variants of uncertain significance. So not only do we need to think how to functionalize it, but how do we do this fast enough to provide information back to the patient or to the physician in order to try to affect care? So that's all I had to add, but that was a good summary. Great, thank you. Thomas from the group? I just want to echo what Carol said. I think that we've spent GM one through eight on sort of looking at other communities and as we were planning for GM nine, there's one community that we really haven't engaged and that's our basic science colleagues and everybody around this table from the most clinical to the most basic recognizes that the potential for that virtual cycle, functionalizing, I like that word, variants of uncertain significance is a huge problem and there are many, many other, I'm sure, opportunities that the clinical data sets that we're generating should offer to the basic scientists and there are other things that basic science should be offering to us besides just function telling us whether a particular SNP is functional or not and that's why we have to have that interchange. Just that this is a great comment and this goes back to what I was talking about in terms of a model. So we're used to thinking about models of understanding the basic biology but there's also model systems to look at therapy outcomes. And so those can be completely different models but we tend to lump them all under the same thing and I think this meeting could help us start to tease some of those things apart so that there's really good communication. And I might note that we have that scheduled for April 19th and 20th, I believe. That's a Tuesday and Wednesday I think in 2016. Not quite sure if it'll be in Bethesda or if it'll be in a more central part of the country but we would hope that folks will put that on their calendars and Eric, I know you'd love to come to Bethesda so if we have it in Bethesda, you'll be there. Yes, we hold it in Houston, you won't be there. So great. Any other comments? Great, thank you very much.