 So, we've asked Paul Rittker to come and talk to us because as he points out he plays in the genomic space, but Paul is the father of a laboratory test that moved from the laboratory to clinical workflow and we were interested in hearing sort of the way in which that happened, the stumbling block CRP. Paul is the father of CRP. Or one of the fathers of CRP. Paul, is that a fair thing to say? It's a bastard, so it's okay. Success has many parents. No, sorry. So thank you for coming and sharing. Thanks. So, again actually by thanking Dan in particular and Teri for encouraging me to come and special thanks to Eric actually because some of the things I want to talk about I think are very relevant to your slides when you open with about how this translational piece occurs. That's okay, I can just do it forward. I'll figure it out. And I changed my talk like three times yesterday for a variety of reasons. One was because I realized I maybe wanted a few people living at the translational end of this. I am a cardiologist and so I kept, when you receive parentheses it's going to change the slides. So this was moving new biomarkers, genetic or non-genetics. That's the first bias of mine is that I don't think there's really a difference between a genetic biomarker and non-genetic biomarker. That's just a worldview I have. And a translational cardiologist perspective, I added a cautionary tale. You'll see why. Oh, and then I think I'm the only person who actually put a conflict slide up here. Pearl, you can demonstrate to the dean. I really did show it. So the other thing was is that I tailored this for the payers, but they're not here. They're playing golf. And so you'll see that some of this is really meant for them, but so be it. So a couple of broad observations for someone who's lived in this world. The first is misnomers. Prediction is not prevention. And we really often forget that. And I think we heard some of that yesterday. The more cynical part of me, and I can get much more positive at the end of my talk, I promise. The cynical part is many clinicians will not act even after there is hard evidence in knowing something new improves care. That's something to not get discussed yesterday, but it is just the nature of the beast. Guidelines, which were seemingly so important for payers. Guidelines usually lag clinical data by many years. I had 20 years in here in the earlier slide. I changed it to many. And rarely are evidence based, particularly those that claim to be so. And I want you to think about that a little bit. Physician obstacles to translation are large and very difficult to surmount. And the quote at the bottom comes from a cardiology friend of mine, senior mentor of mine, all changes for the worse, including change for the better. And many, many doctors. Yes, you do know who this is. Many, many doctors think that way. So Dan said, please come, Paul. I really did enjoy yesterday, and I want to show you that I have been involved in this field. And it's remarkable. I never get an opportunity to show it. Almost 20-year-old slide. So this is when you actually ran SNPs one at a time in your lab, and you only had to have 1,400 patients to get a paper and an email journal. This was our demonstration at Factor 5. Well, what became Factor 5 lied and was higher and people had venous thrombosis and arterial thrombosis. But I'm really a clinical trialist. And so this observation, we thought, wasn't that important. What we thought was interesting is could we design a randomized double blind placebo controlled trial to test a genetic hypothesis? So we think this was the first NHLBI funded pharmacogenic trial ever done. It was published in 2003, it began in the late 1997, 1998. The idea was to screen people for this new polymorphism, Factor 5 lied, and figure out whether or not they respond to warfarin differently. So we had people who were positive and negative for Factor 5 lied and recruited all over the country, followed them for recurrent venous thrombotic events. The punchline of the paper was this. We had a 64% reduction in recurrent venous thrombosis by giving a novel therapy, which by the way never made guidelines, which was low dose warfarin instead of giving full dose warfarin, very low side effect profile, very effective, therefore not in the guidelines. But the point I want to make to you was that the real trial was based on this. The NIH funded this because of the pharmacogenetics, and that particular pre-specified subgroup, Factor 5 lied, and all the prothrombotation, it just didn't matter. Both groups equally benefited from being on low dose warfarin for preventing recurrent venous thrombosis. So that was our first foray into what became much more commonly now pharmacogenetic clinical trials. I'm going to fast forward. We also run something called the Women's Genome Health Study. It's one of our many large-scale GWASs. We tend to do GWASs inside of randomized trials. One of the issues I want to leave you with is trials really matter, and I think NHGRI could not fund the trials. I know you don't have the budget for that, Eric, but you could think about funding the genetics inside the trials. So this is our WGHS, this NHLBI, it's 20,000 women. And the punchline of this one was on the left that taking the best-known 101 SNPs from the genetic risk score, looking at all the literature, we couldn't differentiate very well at all between high, medium, and low risk people. But what Jeff told you yesterday was crucial. So this is the genetic risk score, not doing much, but family history. Simple question, did mom and dad have a heart attack or stroke before 860? Discriminated very well. And so that tension between are the SNPs picking it up? Is it shared environment, or are we missing another piece of heritability? And so the family history piece, I really liked very much yesterday what you're trying to do in that business of adding that to clinical practice. Now this is what I was thinking I was going to present. How do you move a biomarker from the bench to the clinic in sort of a thoughtful way from our perspective? And it's a little like Passover, there's four questions, but is there evidence that individuals identified by the biomarker of interest, genetic or otherwise, are at high risk, even when other risk factors are acceptable? That's important. Why measure something new if it just is a surrogate for something else? So in the lipid world, that's advanced lipid testing versus a total cholesterol, HDL cholesterol ratio, am I really getting something new? But the second question is very rarely asked. I think this is the biggest problem for biomarkers in general, it'll be a big problem also for genetics is, is there evidence that individuals identified at increased risk due to the biomarker of interest benefit by receiving a therapy they otherwise would not have received? Why do a test if you're not going to do something? If all I'm going to do is recommend diet exercise and smoking cessation because you're at high risk for heart attack, I don't need a test to make that recommendation. Now because we're in a genetic setting, I added the third question, which is the obverse to that very important for side effect profiling. Is there evidence that individuals identified at increased risk due to the biomarker of interest benefit by avoiding a therapy the otherwise would have received? If I can predict the toxicity and we do these clinical trials where we treat everybody, maybe there's a way to weed out the problems. The last question I'm going to just is up there because I want to show you what we're doing. But it's not relevant for diagnostics. It's terribly relevant for therapeutics. Can you alter this pathway? Knowing a new pathway exists or a biomarker exists is really only partly interesting. The question is, can you change it? So I do inflammation biology. This is what we teach the Harvard Medical students, inflammation, ethosclerosis is the same as rheumatoid arthritis and psoriasis in terms of biology, but why does it affect the joints, the skin, or the aluminum and differentially? We publish papers like this. This is interleukin six levels predicting future myocardial infarction in otherwise healthy men, baseline levels predicting future vascular events. And I would argue some of you saw the paper in Lancet that the genetics community discovered now that IL-6 mattered because they had a Mendelian randomization. But these data are actually 16 years old. CRP, which is what Dan talked about, is that red molecule in the background. This is the cover of nature, whether or not CRP as a target for therapy, probably a misreading of the whole field. But I wanted to show you that this again is one of the things for Eric is the time it takes to move something from an idea to practice. So this is 1997, the study started in 1993. This was the original publication saying what's now called HSCRP predicts MI on the left and stroke on the right. We would go on to show this that was in men. We would do this in women. So generalizability across big NIH cohorts. This is a comparison of quintiles of CRP on the left measured a baseline against quintiles of LDL cholesterol on the right. And they're completely independent of each other. Fast forward. A meta analysis has done 54 prospective cohort studies, not one, not two, 54. And all of them show the exact same thing. If you know your CRP level, you can predict corny heart disease on the left. You can predict all vascular deaths on the right. And the most important piece of this is the biology. The incremental risk determined by knowing something about inflammation, adjusting for all usual covariates using a standardized model, the metric, the trivial risk for inflammation is at least as large as that of blood pressure and cholesterol. And the question is why aren't we actually actioning on this? This is here for Jeff again. The Reynolds risk score was developed with my statistical partner Nancy Cook. I kind of get credit for this. I shouldn't. She did all the heavy lifting because it was a math problem, not a clinical problem. But it's not just that we added CRP to Framingham. We also added family history. And as Jeff and I talked about, I think that's actually at least equally important in understanding what's going on. If you've never been to this website, please do. It's quite user friendly and it's meant to assist in this translational process of understanding why you might want to actually think about a new biology. But the first question is usually where fields stop. Usually people will develop something and say, this predicts risk. But as I said before, prediction is not prevention. You have to go to the second step, which is very rarely done. We were cognizant of that. And so we wanted to say to ourselves, what could we demonstrate to the clinical community why measuring this biomarker might matter for clinical outcomes, what our payers were talking about yesterday in terms of would you actually want to do it? And we felt very strongly that you needed to demonstrate clinical benefit from a therapy patients otherwise would not receive. So we had to design a program of studies saying who could we identify at risk with inflammation, as there is our example biomarker today, and could we then change practice. So we had observed, and again, data are very old, that people getting statins had very large risk reductions when inflammation was present. This is hard endpoint data that's on the left and that statins lowered this inflammatory biomarker, the CRP on the right. And so skipping through a lot of other stuff, we designed this clinical trial called Jupiter. And this is what I was referring to yesterday, and just as a side effect because the payers aren't here, the two that I spoke to did not know about the study. And that was instructive to me. So Jupiter is a randomized double blind placebo controlled trial asking the question, if I identify people who would never get a statin under any current guideline in primary prevention because their LDL cholesterol is already below the treatment target for therapy. So you could only get into this trial if you're untreated LDLs below 130, which is our target for treatment. The average LDL was only a hundred, but you otherwise were at risk because we screened for CRP and enrolled those people who had this pro-inflammatory response who therefore would have a high risk. And then they were randomly allocated to a potent statin and placebo, and it's not a surrogate endpoint trial. It's a hard endpoint trial. The question was, could we prevent heart attacks, strokes, and cardiovascular death among these people who simply would not get treated under any structure? So this was actually for the payers. 55% reduction in myocardial infarction among a group that A, people thought weren't even at risk, and B, the drugs would never work because the biology what didn't fit what it was all about. 48% reduction in stroke, 47-50% reduction in bypass surgeon angioplasty, and if you do some health care economics here you can very quickly figure out that's the big win in terms of costs. And a 20% reduction in all-cause mortality, which in fact the FDA, US FDA did not put in the label, which we thought was interesting because that's been the standard for what people thought really would matter. Now Canada did action on this. That's what's so interesting to me. Most of these patients were enrolled in the United States in Europe. Canada had a relatively small enrollment, but the Canadian guidelines changed almost immediately. So within four months of publication, six months when it came out, the Reynolds risk score in yellow was introduced in Canada and basically for the last three and a half years in Canada anybody with a high LDL, low HDL, or high CRP gets a statin. And that's been a very interesting thing to follow. What I wanted to talk about though is how genetics is very informative within trials though. So one of the controversies of our study was why did this work? These are very powerful LDL lowering drugs. They're also very powerful anti- inflammatory drugs. They do both. So no statin trial can deconvolute that. Well the genetics are very helpful. This is work from my colleague Dan Chasman and one of his postdocs, Audrey Chu, they were just published. And one of these is on the left really for Dan and for Eric in terms of what we think about in terms of statin pharmacogenetics. So the point I want to make on the diagram on the right is, yeah, we could, we could, we can predict LDL response by measuring a certain set of SNPs. And we've shown this repeatedly now and others have as well. But of course you just also measure the LDL, which might get you there just as fast. But the point I want to make is that the genetic determinants of the statin induced LDL reduction do not predict the statin induced CRP reduction. And we're able to show the genetic determinants of statin induced CRP reduction do not predict the statin induced LDL reduction. And that was instructive because of different pathways. So what we're trying to understand by using these trials is can we now take this and ask the fundamental question, can we lower inflammation to lower vascular risk? And I don't want to drag you through this too much. I want to come back to the genetics in a minute. But the way the clinical trial community thinks about this is how do you attack the problem. You have the biology, you have the genetics. And I want to show you two that share something in common. You want to do a clinical trial where you take the confounding out. Now this comes from someone who lived in observational epidemiology for years, became somewhat frustrated by it, and realize that maybe trials is the way to actually go if you can afford them. So you want to have a setting where you lower inflammation but you don't affect lipids, hemostasis, or thrombosis. You just affect the one pathway. The first trial uses something called LDM which is low dose methotrexate. This is the one that's been funded by the NHLBI. The study is called CERT, the Cardiovascular Inflammation Reduction Trial. We're taking secondary prevention patients on all usual therapies and rolling those with diabetes and metabolic syndrome as a surrogate for inflammation. And what they're getting is the standard of care for rheumatoid arthritis. But they don't have rheumatoid arthritis. They have coronary disease. We're trying to see whether lowering inflammation will lower their risk. And you might say that sounds very odd and gutsy. The reality is that there's seven observational cohorts. If you just scan the yellow column you see the hazard ratios are all below one for incident vascular events. And again it's the attempt to use trials based on this biology, based on this genics to find out can we alter care. I'm going to skip over this. This is just the NIH stuff. The primary aim is to test the inflammatory hypothesis of athertombosis by doing this trial. The second one funded by Novartis is up and running already uses interleukin-1 beta inhibition. Again an approach that lowers inflammation doesn't affect lipids. Here the genics is terrific. For those of you who know the IL-1 Genics story this notion of an endogenous, dangerous signal how modified LDL affects the NLRP3 inflammasome. That inflammasome is very important for CRP production and other things. That inflammasome of course converts caspase allowing pro IL-1 to go to IL-1 beta. You get IL-1 beta locally and we can now inhibit that directly. So now we have a cardiovascular event trial being based on genetic observations because it's now been shown that the earliest linkage here, the earliest trigger of the NLRP3 inflammasome in the vascular world is cholesterol crystals. And so if early cholesterol crystals are triggering this through this genetically inherited issue then we can actually try to inhibit this. And you can see that it goes through this pathway to IL-6 deliver and back to CRP to our biomarker. Some of you may or may not know this drug the pediatricians might. Canokinumab is a human monoclonal anti IL-1 beta antibody. It's approved in the US and Europe for one of the orphan drug act for rare genetic disorders that lead to pro hyper production of pro IL-1 beta but allows us to do a clinical trial trying to neutralize inflammation as we do this. And this is what the drug does. It lowers IL-6, it lowers CRP, it lowers fibrinogen, it doesn't do much of anything else. The question is will it lower vascular events. And so that trial is called CANTOS, the canokinumab anti-inflammatory thrombosis outcome study. We're running this around the world. There's some 17,000, well right now there's about 2,000 patients randomized. Eventually it'll be 17,200. Three different doses of the study drug against placebo and again a hard end point trial. You can imagine within this we have pre-consented everybody for all possible future genetics but don't have the money to do the genetics. So one of the other issues for NHGRI to think about is within these trials that are very expensive, small investments to leverage that might be something to think about. But I was talking to Jeff Ginsburg when we were in Washington together at another NIH meeting Wednesday. And thinking about this from a cardiologist's perspective about will genetic screening play a role in patient focused thrombosis care and will pharmacogenetics matter. And it's funny because I actually am someone who believes the answer is absolutely yes. But we haven't quite gotten there and trying to get there seems to be the critical issue for us. So I wanted to give you a few examples from the cardiovascular literature. The first is one that initially I thought wasn't going to help us very much. This has to do with the data about the SLC-101B1 polymorphism that the search investigators using again a clinical trial where they had placebo data were able to demonstrate that carriers had a far higher rate of myopathy than did intermediates and then did the non-carriers. And now this has turned out to be simvastatin specific. And Jeff made a very thoughtful point to me. He asked, well, Paul, what happened in your Jupiter trial because we had resuvastatin, which is a safer drug. And these are the data, Jeff. And all you have to notice is that there's nothing going on. And that's because this drug doesn't do it. And Jeff made the insightful comment that I said, well, who's going to pay for that test? And he said, well, maybe Ash is in a good which I thought was actually very interesting insight into how to get things to actually move forward because you're right, you might want to use this drug because it's safer if you happen to carry that particular issue. What I've been a little more skeptical of that is that doesn't seem to be how people think this is also from that trial. I asked the question yesterday of one of you about the commercialization of this process. This is the KIF-6 story. This was an assay that was highly promoted to the Cardio-of-Assad community based on sort of mixed evidence. And we finally did this within, again, a large randomized double-blind possible controlled trial. And the punchline on the left was the relative risk reduction with the statin among non-carriers was exactly the same as it was among carriers. There was no difference in the LDL reductions. There were no difference in the CRP reductions. There were no difference in the event rates. And there were no difference in the relative risk reductions. But the observational data has suggested there might be. And again, this is the tension, I think, between someone who moved from the observational world to the trial world trying to get rid of some of these issues. So, Eric, this was added for you about your density maps, because I wanted to end on a much more positive note in your question about the speed of translation to practice. And the first was don't be discouraged. It takes a long time to change practice even when randomized trials exist. And it's something that I think NHGR just has to realize you are making fantastic basic science discoveries. That's really exciting. You're in this process of translation. It'll happen. Number two, sure, there are bumps, potholes, and U-turns on the translational highway, but it's misspelled. But where else are you going to drive? That's what it should say. The third is something I've also changed my mind about. It would be nice in the cardiovascular community to have a killer app. But I don't think we actually need it. Because I don't think it's the average patient that we're really thinking about anymore. I mean, it'd be nice. But the flip side is if the cost of screening falls far enough and the Moore's law suggests that it will, you don't need a home run for all patients. You just need a clear benefit for some, even if they're rare individuals. And number four is a personal issue for me. It really matters for parents and for kids. Part of the reason I'm here is, and I don't mind saying this publicly, I have a child with long QT. It makes me think a lot about the personal implications of everything you all do versus the research implications of what people like I do. And that tension is very, very much about what this is sort of all about. So Eric, you ended with a quote yesterday. And I wanted to show you the quote that sits on my office wall. It's a little longer than your quote was because it wasn't quite as pithy. I have a big wall. Yeah, yeah. It's Harvard. It's not really that much bigger. It's a very small plaque. And Anna's going to read it. Many of you are familiar with this, but it really is relevant to what you're trying to do. It must be considered that there is nothing more difficult to carry out, nor more doubtful of success, nor more dangerous to handle than to initiate a new order of things. For the reformer has enemies in all those who profit by the old order and only lukewarm defenders and all those who would profit by the new order, this lukewarmness arriving partly from fear and partly from the incredulity of mankind who do not believe in anything new until they have had an actual experience of it. This is obviously Machiavelli talking about the politics of what it's like to be President Obama or anything else. But I think it's the same issue for all of us. Everyone in this room is dedicated, I think, to trying to bring innovation to practice. There are many, many, many reasons it's very hard to do. There's many, many people who actually don't want you to do it. But we all also understand it's the only way to actually go. So I'm very pleased to be here, Dan, and I think that NHGRI is one of those motors that will drive this process. So again, thanks for the opportunity to chat with you this morning. Thank you, Paul. We're running behind, as usual, but that's OK. So we have time for a couple of questions. David, and then Jean. It's about ignoring David. Very interesting talk. And I just was struck by your, I think, your third conclusion about the killer app and the average patient. So the data you showed us in the clinical trials really is data on averaging all patients. And the question is, even for an intervention that doesn't show an effect when you average all patients, there may be some patients in there in which that intervention would be terrific. And so how are we going to figure that out? So that's exactly why this is up here, because the extreme phenotype concept from the genetics world applied to the clinical trials world makes these GWAS within trials very valuable. Again, it'd be nice if I had a genetic test that said, give these people drug A and these people drug B. That's the average effect. Overall, I don't think it's going to happen. But if I can find 25 people out of this 15,000 who should never get a statin, either because they get absolutely no LDL reduction, so we're fooling them clinically, or because they have a clear toxicity that's going to hurt them and I can convert them to somewhere else. I think that's a great benefit because, again, at the end of the day, we are physicians and the patient in front of us matters. And there may be other patients who get an enormous LDL reduction, but I can tell you get no CRP reduction at all. I want them on the statin, but maybe I need to divert them into these other clinical trials that we're doing where you're going to inhibit the inflammation. So I really agree. And a part of the tension is the cost. Again, that's why the second half of that one is there. If the cost is low enough, then of course you don't need that big of benefit to justify the cost. And this is one of those rare fields where costs could absolutely plummet as we think this through. Two comments in a suggestion, Paul, and beautiful presentation. My wall has the piece from Churchill who says, the Americans always do the right thing after they've exhausted all other possibilities. And second is that I grew up with a cholesterol fight. In 1967, Bob Levy and Don Fredrickson published a very nice review. And it was pretty suggestive that cholesterol was something bad. Well, in 1994, there was finally a conclusive final trial that finished it. Prior to that, I had to personally respond to a writer in Washington who said that lowering cholesterol doesn't help. It may prevent cardiac disease, but it makes you violent. And so you die of violent death. I kid you not. The third point I want to make and more serious is that it seems to me that all clinical trials designed henceforth ought to really have a consent and an intention of collaborating with folks in genomics to do what you've done. I mean, this is a wonderful opportunity, and we ought to learn from it. So, Jean, I would say two things. You're absolutely right. The lipid thing, for those who remember this, is a very important memory because I think the controversy about screening for genetics, to me, is no different than the controversy we've lived through many other times and the lipid hypothesis. You remember the Atlantic Monthly articles and the whole thing. So you're absolutely right. Trials are tricky. Eight years ago, it was hard to get the pharmaceutical companies to agree to let you do a GWAS within them. Fast forward to today, it's quite easy. They've gotten over their anxiety about it. They don't necessarily want to pay for it because they haven't figured out what's in it but their fear of it diminishing their market share is not quite what it used to be. So again, one of my pitches for this was, you know, a modest amount of money from something like NHG or I to think about genotyping a trial on top of the hundreds of millions that go into the trial, it might be a very good investment. Paul, great remarks. I'm surprised you didn't mention the fact that CRP was discovered in 1930. But along those lines, I wondered if you could just give us your thoughts on why the road has taken so long for things like CRP and some of the examples you used. And then you look at Oncotype and the 21 gene predictor was published in 2004 and within a year or two we had a test on the market that was being rapidly taken up by oncologists. Boy, that's a six beer question. I think part of the answer here has to do with this intervention of one of the problems that CRP had was we were introducing a new biology and a new biomarker and a new pathophysiology and a new intervention simultaneously. It didn't come in pieces. And I think that was actually part of the problem. I was really hoping to talk to the payers today because I know exactly which payers did and didn't pay for this and when they didn't pay for it. And that was what was yesterday's little discussion about. On the other side, the genetics community had a better job because at least it's a zero one outcome. But the data we saw from one of yesterday about the thionopyridine spike and then fall is also very instructive and I stayed away from that example because I don't know what I really think. Part of me thinks it's a great idea and part of me doesn't. Those are all good studies with very different conclusions and that's one of the tensions in that field. I don't really know Jeff. It's tough. Can I add just one thing to that? I think that there are many examples that perhaps they're increasing examples more recently of much more rapid pace. And I'll give you the example. The IL-28B was discovered at Duke a year before it was on the cover of gastroenterology and being used by all the clinicians. So although there are some that lag for many years, it's often because of controversy, conflicting studies, et cetera, that there are many markers that are moving much quicker than that. So Howard next, but before Howard, I think part of it is also sort of what do you do with the result? I just want to comment because you're avoiding therapy and that's a great cost savings. So Dan's right, one of the problems this had was it led to the suggestion that maybe another 20 million people ought to be on a statin. And we think that's not a bad idea, but I'm not sure the payers thought that was remotely a good idea, is part of the unspoken thing here. Howard. So you've mentioned a little bit about the clinical trials are expensive and they're sitting there and just need the genetics piece performed. And I thought your name was familiar and looked back and we've actually published three papers together although we've never met. And that is the Harvard model. And that's because you made your DNA available to some stuff that Brian Gage and I were doing. So the NHGRI supporting that is a great thing. But there's also other funding sources, non-federal funding sources for doing these kinds of things. So what are your thoughts on how clinical trialists can make visible your, the potential collaborations like that? Well it's actually good news. I mean Jesse, so most clinical trials in the United States for better or worse these very large ones are done at one of about five centers and Duke is probably the largest of them. And the Duke group is very much in favor of banking everything as are all the major studies. Pearl knows this, again this shift in what I used to call genetic paranoia is pretty much gone. The companies no longer fear this the way they did just eight to ten years ago. And so many, many of these big trials now are routinely collecting, at least putting away whole blot or buffy coat. And usually you're doing some sort of back end genotyping. When the studies positive people tend to wait and see what the outcome is. But I think again Duke, Cleveland Clinic, Harvard there's a lot of these trials initiating and most of them now do have biobanks. And then we can go to people like you to get some other neat things done. Okay. I want to comment actually that this is also happening in cancer. So I actually just put together an application for a large trial, a large study actually a genome-wide association to look at response and immune-mediated adverse events in response to epilumimab. And BMS is giving us all of the samples that they have to be able to do that in addition to collecting samples from academic centers. So they're either going to give us data or samples and then we're bringing that together with samples from academic data centers to be able to power the study to do this, which for I don't know how many people know about this drug, but it has a lot of adverse events, not very many people respond. It's a very important study. There's a lot of evidence suggesting that immune-mediated drug immune immunotherapies may have inherited variation may contribute. So I just want to point out that sometimes when you ask the companies they'll actually give you the samples. Right. So again, you're absolutely right. The canikinumab trial I described, this interleukinone beta issue. Novartis needs it to be a win to get a label. But we were able to negotiate up front that even if it failed for that purpose, DNA would get handed over. We would then have to go to presumably the NIH to get the money to do the genotyping. But there's so much good biology buried into what are you doing when you inhibit a very specific part of the inflammatory system in otherwise stable patients that I think you're absolutely right. If we all did that we'd learn a lot quickly. And the beauty again of a trial, not to denigrate the observational epidemiology because but because the trials allow you to look at it in a structured setting against placebo where you can look at before, after and the drug effect. And so I think that both of these are very complementary. Okay, I think we can talk all morning. But let's go on. Pearl is going to talk about the clinical slash research interface working group. Maybe tell us what they do.