 Okay, so thank you very much for inviting me to be here. David Dimmick sitting over here is really the clinical director of the whole Genome Sequencing Lab, so everything that's all the hard questions go to David. So the overview of what I want to talk about is our Diagnostic Odyssey Program. This is a whole Genome Clinical Program. Talk about just one slide on the initial case and really spend a little bit of time talking about laboratory certification for the Sequencing Lab and something that I think is equally important is certification of the analysis, not just the wet lab side. Selection of patients and why we have a clinical board, consent and ethics, whole Genome Sequencing Analysis, the limitations around this, opportunities for these collaborations and so forth, and then whole Genome Sequencing in my opinion is simply another laboratory method. In absence of clinical data, it doesn't have a lot of utility, but it's the context in the clinical environment where we believe you can use it today and we also can talk about getting paid for it as we go forward. So this was our mission statement back in 1999, so this isn't new to us, although we surprised a lot of people with the initial case, was to try to figure out how to use Genomic Sequence in the clinic. That has been something we've been working towards since 1999. In 2004, our goal was to get Genomic Sequence in the clinic by 2014. So this is the index case that most of you have heard about. This is Nick Volcker, showed up at 23 months of age. This is how he presented into the hospital. This is not after surgery. This is presentation at the time. Long story short, after almost three years, a diagnosis was made using exome sequencing that the child had X inactivation of apeptosis, XIAP deficiency, and a decision was made to do a core blood transplant, which has effectively cured Nick. So that captured a tremendous lot of media's attention, much more than we were really anticipating was going to happen, and even with Francis Collins and so forth running around talking about it, it's created some good news and bad news. There has been some whiplash towards us around this, that it's been taking advantage of the patient and the family. We can discuss that if you like. But one of the issues that we've felt very strongly about a clinical program is it has to be for clinical utility. And because of that, we've set up an approach that we use for our clinical delivery. It's nominated, the patient's nominated by two physicians inside the division. And it's to end the diagnostic odyssey. I believe that this is a realistic, practical approach that you can use today. It's actionable. What do we mean by actionable? It might not necessarily mean therapeutic actionable, and I'll come back to that a little bit more at the end. We have found from some of our families that being able to make a diagnosis, even if it's negative, is of a potential benefit to that family member. A case review is part of the committee. There's consent that gets involved, genome sequencing, data analysis, and then follow-up counseling. So the guiding principles behind this is that it need that all reasonable clinical testing has been performed. And so the notion is that this is a diagnostic of last resort, if you will. It's likely to have genetic etiology. It's very rare. We hope that it's monogenic etiology, and the notion of that is it simplifies the search space. So having it being extremely rare and expecting it to be genetic, you'd expect it to only be found in that patient. Or one other family member, if it's a compound heterozygote, or in the case of Nick, was an X-linked. These have distinct and unique phenotypes that are more likely to have a definable result. The ability for genome-wide sequencing to assist and enhance a medical decision-making. Again, I want to emphasize it's to enhance the medical decision-making. It has not been, Nick Volcker case wasn't either, simply in and of itself sequencing. I don't believe that that's the methodology we're talking about. We need the ability to conclude a genome-wide sequencing study. So it's very easy to get in one of these. It's quite hard to get out. Because where do you in fact define that you've finished the analysis? And that is a major issue as you start talking about cost and where you go forward on this. We insist that the parents both must be available, appropriate tissue and DNA must be confirmed, and monetary cost benefit considerations. And the other advantage of having this clinical board is that they are detached from the decisions. They don't gain anything, either fame, fortune, or tomatoes by being involved in this. It also we think is important for helping to make sure that when we're practicing this on the edge that we can back up and say we've had other people be involved in the decision-making process and make sure that it's to the benefit of the patient. That's really what the goal is. So the genetic consent is another issue that we deal with. And genetics assessment is involved very early on, obviously, with an MD geneticist. Typically it's David Dimick or David Bick. A pedigree analysis with a genetic counselor, I think these are self-evident issues. The consent process, then, is discussions of data return. I know David Valley hates the term personalized medicine. But I think in our case, we believe that data return is also personal. We think that the patient and their family need to have a decision around what type of data they get back and we consider that to be part of personalized medicine. There's multiple consent counseling sessions, largely involving how do we deal with this consent process and what types of data with the family like back. The family has time to consider the testing and data return, followed by written consent. The average is between six and eight hours. It's two to three sessions. We're working on trying to streamline this. I recognize that this is not amenable for long term practice across common disease. That's not where we are. I think we still have much to do before we can get to that level. But we do believe in this environment, you can go forward on this. One of the things we spend a tremendous amount of time on is building a software package that is CAPCLEA certified. This has driven a huge amount of costs and it's a major issue. And I'll just give you a couple of little things that you may not have thought about. We use data from all types of sources. Those data sources have to be frozen, validated, and confirmed that the algorithms are coming up with the appropriate results that we've had previously. Many of you have been involved in some of these data sets. When you all change those data sets, we have to revalidate. And on top of that, the evolution of this continues to go. We have a continual argument about when do we use the new tools. And the new tools can't be used in the clinical environment until they've gone through clinical validation before they come forward. And that puts a real pressure on the bioinformatics team and a chronic recurrent cost of how do you stay up with this as it goes forward. But I will tell you, many of your most popular databases that you're using when there have been changes, there have been errors. And we're thankful that working with these groups, they've helped us be able to define that. So I think if you're interested in getting in this game, it's not just the wet lab. In fact, I would argue the wet lab is the easy part, because the wet lab is somewhat stable. But the analysis, I'll give you a classic example. We use Casava 1.8. Most people don't use that. The reason we use that was because that's what Illumina uses in their CAPCLIA laboratory. So when we get the data back from them, that's where we have to start. We can't change that. That's where we start. Okay, so I can answer questions around that. The limitations are, what does actionable mean? I think we can have a lot of debates around that today and tomorrow and for the next three months. But I think this is critical to how we decide what's there. Alignment versus de novo assembly, I think this is also a major issue. And it may well be that some of our really rare cases here, that it's going to be unique, strange events that we can't capture that we're missing. And that could be a major issue for us. So what do we need? Well, we need more data. The availability of genomic sequence. There's lots and lots of sequence out there. Most of it is not available. We'd love to have 462 new genomes. We'd love to be able to share. I think this is a space where for these diagnostic odyssey, if we're arguing that it's unique, simply searching across somebody else's genome is an advantage. And I think that's one place that we as a group can work together. There's very little clinical data. And I think these are major issues around as we're starting to think about this, not only family history, but clinical data at all. Questions of clinical utility and value, these are major drivers and they're major obstacles. And I'll talk a little bit about that as we go on. So this is what everybody argues about. The cost curve, can we actually bend the curve? And there's arguments about can we use genomics? And this is back to what Howard McLeod was talking about. This is really boring economics. But this is really what drives a lot of this. But if you go out and you look, the Congressional Budget Office, this is after the latest round of debates in the last presidential election, was talking about preventative medical care. This report says we actually drive costs up by trying to do preventative care. And so if we're going to try to bend this and use that argument, we have to be able to show that it actually changes the health care cost. Well, we're all familiar with this and I hate to walk into this because there's been a lot of people talking about this that are much smarter than I am. But just as a simple, silly example and an example that we've used to some of our local businesses in Milwaukee, what if you use Plavix and you go in and you have 300 patients? What's the economics on that? So this is a true case of a major employer in Milwaukee. And they have approximately 300 patients on Plavix. You could argue that approximately 100 of these will have a genetic abnormality. And up to 50 of these patients may have a complication. So you can do some simple math on this. 50 people at $50,000 for an event. It's $2.5 million just to look at that cost. If you were to say you could do whole genome sequencing and analyze one gene, we can come back and talk about the economics, but I'm not worried about the cost. The cost is continuing to follow. It is not expensive to analyze one gene. It would cost about $300,000 to survey these three people. The argument is let's just switch to another compound. I just picked this one. It could be a different one. But there's other ways you can save money on this that we don't stop to think about on the academic side. Plavix is coming off patent. So if you get a drop in price on Plavix, you may actually want to do the genetic testing in order to keep people on Plavix. Because keeping the people on Plavix would reduce cost. And if you go through the calculations here, a simple $100 per month drop for this particular company is $240,000 per year in clinical costs. And so you can add this up. And you would find that over five years that looking at these 300 patients would save $3.9 million. Here's the problem. You're saving money. Who's going to put the money in in order to save the money? And this is the challenge. And so as we go back and forth, there's major issues around this. And I want to ask a question that if we can answer this, it would help. It's not what is the cost of a genome? It's what is the value of a genome? And I think that has to be the driver, not only to our studies, but to other studies. And so you can leverage it. Somebody has to pay for the upfront costs. How do you amortitize that across the other studies? So one gene test at a time, it's not cost effective. If you run a clinical laboratory, it's not cost effective to do this one gene at a time. It's expensive to keep that up to date. It is much cheaper to do one platform or two platforms chronically and across the whole thing. But what's the value of a patient's genome sequence? I would argue that a whole genome sequence is, in fact, family history with data. And I'll tell you what I mean by that. Back to the point of, did you have your family history discussion at Thanksgiving? Even if I have family history discussions at Thanksgiving, I have zero confidence in my family's knowledge of what diseases are in the family. So not only do we not talk about it, but when we do talk about it, it's inaccurate. Yet we put this out there. So I was quite surprised. I used to be on the FarmGKB advisory group, so I went in, look, how many drugs are there, or compounds, with genetic information? Well, the answer to that would be 370. It's a lot of information that we're probably not leveraging. How many diseases or clinical phenotypes are out there? Well, the genome-wide association studies, which I, of course, have criticized up one side and down the other side, but I got to say, there's 237 traits that have been attached to the genome. And we argue about, well, what's the predictive value of these? It's not very important. Well, we've gone on to do some genetic studies in animal models. I'm not going to go into all of this because it's way beyond this. But we are finding, actually, that these genome-wide association genes are having significant effects in both blood pressure and renal disease. And so I think it's going to play a role. And there are some groups that are now starting to say, how about if we do risk-position of complex disease from family history with known susceptibility loci? So this is a very interesting paper that goes through and models this. And I'm simply going to say, if you have a SNP that's been associated with a common disease and you report a family history of a disease, I would say there's some level of confirmation. The value of that confirmation is unclear. But I think one of the things that could be done by this group is ask those fundamental questions. If you have it, what does it mean as one example? Now, you're all familiar with this, salt and hypertension? How many of you believe salt causes hypertension? It depends. It depends, OK? All right, so from Scientific American in July 8, 2011, it's time to end the war on salt. And the reason for that is, is if you go in and look at the epidemiology, across the epidemiology levels, it causes a 1.1 millimeter increase in systolic blood pressure and a 0.6 millimeter increase in diastolic blood pressure on average. Salt on average is what your kidneys are for. But we very well know that there are patients out there who benefit from a low salt diet. I'm going to argue that the SNP data is very much the same thing. On a population level, it might not be very useful. But on an individual basis, is it salt or is it completely useless? I think these are fundamental questions we need to address. So whole genome sequencing, it'll be a long time before it's used clinically. I've been told this over and over and over again. We've had our clinical program for one year. We've had 50-plus nominations come through in-house. We don't advertise this outside. This is not something that you'll see us running a web page on. This is in-house physicians who want to use this technology to answer their questions. No one will pay for it. So here's a letter from insurance group from February 23rd to David Dimmick, and it basically says, as we have discussed, the situations where you determine that on average the cost of routine testing will exceed the current contract price of whole genome sequencing, we will authorize whole genome sequencing as a first-line clinical test. So we've put forward whole genome sequencing 10 times. Four have given pre-approval. So I think, again, as the cost come down and the utility comes up, it's not the cost, it's the value. What is the value of that genome sequence? So we've put together a team that's involved in dealing with our patients, the clinical care team. We haven't modified at all. We have a sequencing team. We have a clinical molecular testing team. We have a data assessment team. We have an LC team. Our ethicists are involved in the decisions and the clinical phenotypes and family history and the bioinformatics and electronic medical records pieces all swirl around. I won't tell you that I'm happy with where we are with integration. There's huge needs that we have to bring together these different data and make good quality tools for making informed decisions. This is where we'd like to go. There's a huge amount of data that can be drawn in and really try to pull this together in a clinical context along with a person's genome. And I think that's very valuable as well as that evolves. As people make gene discoveries, it adds value. I'm going to argue that maybe what we're talking about might be a little bit too slow. Because I think it's a rapidly changing landscape. Now that whole genome sequencing is being offered by many groups, I think we're going to see a massive rush towards this. And it's going to come down to the same way that consumers start demanding these issues. The cost is still a little bit high. But I think as it comes down, there's going to be more and more of a driver towards us. And the more data there are, the more we have to deal with. And I think this is going to be a very aggressive pace. And I think we need to think about that as we have these discussions. I have a large group of people that are involved in working in a variety of the projects here. This is the overall team that was involved in dealing with Nick. And I just want to summarize by saying, the cost for data generation will fall. Within the year, I'm going to guess it's going to be about 2,000 for a whole genome sequence. And we need to start asking, what's the value of these individual gene chips? Should we basically figure out how to take these costs and go forward with this quickly? Data management and clinical decisions is the challenge. I will argue over the wet lab. And I, again, want to emphasize, I think having a certified software is important. I think it is just another lab value whose context with respect to clinical presentation is critical to the utility. Genome sequence by itself is not a diagnostic. It will never be a diagnostic by itself. We need the other pieces around it. But it does have value now. And I think it can be a family history with data. For example, somebody who doesn't have a good family history, those SNPs, I would argue, have a potential window that could be leveraged. The question is, what's the value of that window? It's cheaper to develop a single platform rather than having to maintain proficiency testing for 30,000 genes. It's much cheaper for one platform. Data return, in our view, is part of personalized medicine. There's lots of education needed. I think this is the biggest challenge that there is. How often are these tools and tests available and the physicians aren't taking them up? Plavix is a classic example. I was just presenting to a vascular surgery group this last week. One out of eight physicians has ever ordered a test and they ordered it three times and it was rejected twice. So the issue is, well, why are we offering it if no one's gonna pay for it? And I think these are critical issues. So the needs, access to results from whole genome sequence, those of us that have it, we should share these. It really enables the aggressive use of that data. It's critical. My rate limiting step for helping the patients today is not enough genomes. Access to whole genome sequence with clinical information, how do we do that? Well, one may be some type of a banking type structure where you're able to go through and query across a phenotype and a specific gene. Can we use that as a way of sharing data back and forth without necessarily sharing all of the data? What types of tools can we use around that? What is the value of having the whole genome sequence? I think if we can answer that question, a lot of the things that we're fighting for and against will be clear. Demonstration projects to show the cost effectiveness and utility that comes back with value. Tools for integrating with family history, I think that's essential. Follow-up validation, how much? How much will you require for your clinical practice to decide once you have that variant, you're gonna believe it? And that's a major issue as well. Decision support tools and more powerful electronic health records. I think we're not in a good position to be able to leverage that with the current technologies and lots of education. I'm happy to take questions, thank you. Oh, I have one last thing, one last thing. So you all recognize this as part of medical care. In this day and age, you would have been burned at the stake. You all look at this as part of medical care. I'm simply saying that I think sequence is the same level. We just don't know how to use it the same way. And although maybe it has been overhyped, maybe it has been something that people think that we've taken advantage of, but is there really a special case for whole genome sequencing? We make misdiagnoses every single day. Not because we're trying to, but due to lack of knowledge about a disease. Nick was misdiagnosed, not by his fault, not by anybody's fault, but that's what we knew at the extent of that. It's not actionable, it stresses the patient. How about end stage renal disease? We give people that information. 25% of the people are still alive five years later. How about stage four cancer? We do this all the time. Evidence-based medicine? Absolutely. But exactly how do we do the population versus the individual? And we drive all of this data and the large numbers. It's a critical question that again, I think this group could tackle and how do you show clinical utility around that? And I think on average, we are making progress. At our institution, what has changed, how we do medicine today is simply because of Nick and the fact that we did it and no one thought it would work and our administration has bought into it. But for me personally, Nick celebrated his seventh birthday on October 23rd and he went Halloween trick-or-treating for the first time. That's why we're trying to do this. We're making mistakes, but I think collectively we can all do this better. So that's what I wanted to finish with. All right, questions? Do you have experience with your program of finding changes that were not what you were looking for but you still consider actionable and has other testing been required that weren't relevant to the initial diagnosis? Wait, say the first part again. So who do we have? So you're like Nick, you sequenced for this specific immune disorder. Are there cases that you've done where you found other things in the genome that were not relevant to the diagnosis of interest but that you felt needed to be followed up? Yeah, so in the case of Nick and with all, Nick was a whole exome sequence and everybody else has been whole genome. One of the big challenges we have is how do you match a phenotype with the genes you're looking for? So we have a case right now that is driving us crazy where it appears to be immunological by some of the clinical teams view and a strokes are happening on the other side. So half of the clinical team thinks the strokes are the problem and half the clinical team thinks it's an immunological problem and the drugs are causing the stroke. There's a great example of what we're trying to deal with so which one is it? So as we go through and look at the genes, if you find a gene that has been implicated in one of those others, it helps lead you in the right direction. We're pretty much done with those lists and it may well be now that we're now stuck with, we've ruled out both of those two categories. There's one right now that might be but that is a major issue. So you do prioritize as you go through that but then when you run out of those you now are stuck with, okay, here's a variant of unknown significance and a gene of unknown function. So what do you do with that? So this is where having more genome sequence you could then say, okay, how many of those could we wash away that are rare or not rare? Does that answer your question? Yeah. Okay. But that was very interesting. Okay, how about if I get the question right? Yes. So do we have incidental findings? I mean, what are those? Right, we are not looking for incidental findings. However, however, we do have, and David can talk about this if you like, we do have as part of the data return, the families do have the right to ask for data, okay? So in those particular cases, we will go back in and look at those data and we will do validations and come back on that. Did I answer that right? Yeah. Okay. Discussion of the variant caller from Illumina. So as you know, there's been a thousand genome sequence by the thousand genomes project and I guess I would argue that we're still trying to perfect calling indels structural variants. If you call structural variants of five different programs, the intersections about 50% at best at this point. So, and indel is probably about 80% accurate. So by sticking with that manufacturer's caller, which is probably not one of the best ones out there. I mean, is that really necessary? Or as we move forward, do you imagine how we could implement things like GATK or other, I think it's a major challenge to decide what set of callers will be used because you're gonna be missing a lot of data. I mean, if you're not even calling indels or structural variants, then how can you go back to that patient and say they have no problems at all? Right, so it's a great question. So what do we deal with the fact of, first of all, there's no set analysis strategy that exists and there's variations in the different callers that come back, okay? So those are great questions. So the reason we started with the Cassava 1.8 is because that's part of their CAPCLEA certified delivery of data, okay? So that's where we needed to start with. So since we were using them to generate the data, that is part of their protocol, whether we agree with it or not agree with it. And one of the challenges of putting your clinical stuff out there and certifying it is you have to be able to certify that it's reproducible. So what you're using has to be reproducible, all right? So that doesn't mean there's not errors. That doesn't mean there's not things that are better, but it means that you can reproducibly go through this. So the challenge that we have right now as we go through with each iteration is bringing new software on. So if we now want to bring new software on, sure we can bring the new software on. We just simply go through now a whole certification process of making sure that GATK, when it makes a call, it calls it consistently. And then you make the decision around what do you do with that data that's consistent with it? It's not like in the research lab where you can pick and choose and take the best data. You have to settle on how you're gonna do it so you're at least consistent around that. And in no way says that there aren't better ways to do that, but those have to be continual iterations that you go through. Does that make sense? So that's just where we started. Now with the next iteration, we're rolling out more software and the next iteration will roll out more software and that's my point. Is that as we bring on our own wet lab group, this has been capically certified, we'll then have more control over what do we wanna do with the data analysis that comes out of that? It seems like a different call to anyone. Absolutely, I mean I don't wanna say for any stretch of the imagination that because we're doing it this way, that this has solved all the problems. We've simply put a stake in the ground and we're operating away from that stake. That's a moving stake and we have to continue to move on those. But I'm quite concerned, and I think that's probably the reality of where we're gonna be, that there's not ever gonna be one straight form analysis package that everybody's gonna agree on. I mean look at microwave analysis, this has been around for a long time, there's still big differences in how the analysis goes through. So I think another part that this group can do is make comparisons about what do these different tools mean and that of course is being done as part of the big genome groups as well. But those are also part of the decisions. If I can, since my name was taken in vain, I'll take 10 seconds to defend my point of view and then ask you a question. Yes. 10 seconds is just that medicine has always been personal. What it's not been is individual. So we ought to use the right clear term. The question is, the concern is how do you know you have the right answer when you think you have the right answer? There's lots of potentially harmful variants out there in the coding sequence, not to mention the potential regulatory sequence. So my concern, at least at the state of knowledge right now is how do we know we have the right answer without a lot of downstream biology? So in the case of NIC and in other projects that we have, we do the downstream biology. So that is one of the advantages of dealing with the immune system is that we have better ways to go in and do the downstream biology. So that was one of my points is, as you take on a case, one of the questions that we take in the case review is, if we think we find a variant, what are we gonna use to try to validate that piece? And so that is an essential piece and I think that's gonna be a major issue for a while and with some of them, it's gonna be even harder because if you don't have access to cell types, if you don't have access to tissues, what are you gonna use then for validation? So you can then start off and say, okay, we're gonna look at the known enterprise, okay? And then when we're done with that, we're done because we can't take it any farther. That might be one decision that would go into that process. You can't go off then if you can't validate it and do anything other than to say, okay, we're gonna look at all the 5,000 genes that are known for the cardiovascular system and stop there. That's one option that you could take. I'm not saying this is why, this is why I'm saying that we think that to end a diagnostic odyssey as a test of last resort, we've done everything we can possibly think of, we do run the risk of failing to find it but we've already failed to find it and so we're willing to say, hey, you know what? We can't do any worse. Let's not see, let's see if we can't make it any better but that's why I think this is not prime time for all common complex diseases for being part of making decision processes. I think we're a long way away from that. So Howard, I wanna come back to the question of the software which I think is huge and anybody could say whatever they want about how much a genome's gonna cost but the analysis of that genome is not gonna be included in the price and everybody talks about this but the analysis is key and de novo, we do everything on a reference which leaves out anything that's not in that reference and I think one of the things this group could do and put collective knowledge behind is what are best practices that we can do to begin to apply this and what are the validations that are best practice? What are, these are all really important questions regardless of what it can do. I mean, we as people who are practicing this should care about what people say about this, so. I agree, 100%. And best practices are gonna be one of the topics of the next meeting, standards and best practices. Question. Yeah, one question. Your description of the independent oversight and the consent, are these clinical oversight committees and clinical consent or is the IRB involved? It's clinical. Okay, great, thanks.