 But I decided to go the casual way and talk about stuff that might help in getting genomic medicine moving. This is from the point of view of where I sit at FDA, which is as the director of the personalized medicine staff, which resides in the office that regulates laboratory tests. It's now called OIR. It's got a much longer name, but a shorter acronym. And that's in the Center for Devices at FDA. We do do a lot of work with other centers as well. So I put together what I'd like to think is a short but profound list of things that might help in getting this to work and to work well for patients, and that is we need standards for genomics. We don't have standards, and this is a tremendous amount of information that's being churned out at many centers around the United States and the world, and we don't know if it's comparable one to another. We need quality systems and the things that are used to generate the information. And that means you actually have a plan. I'll talk about that in a little bit more. We actually would benefit hugely from curated databases of evidence, and that means when you have a finding in a whole genome that it actually gets into a database and outcomes and management and so on are recorded. Because if it's something that's rare or newly encountered, it's the only way other people are going to know, other than publishing a case study, that that even exists. And a lot of the databases that exist now are in fact fairly well populated, but mostly not curated. They're more like repositories. People just submit information. There's not necessarily a lot of control on what actually gets in there or the quality of what gets in there. So I'll talk about that a little bit more in a minute. And I think most of all, we need evidence to put in the databases. A lot of people are wanting evidence, but it doesn't seem like a lot of people are putting together programs to get evidence. So I think we would benefit a lot from that. I always hear it. I've been on these federal committees since 2001, and we've been talking about evidence for clinical utility since then, and I'm sure people talked about it for years before then, but I still haven't seen very many studies set up to actually create high quality evidence. So where are we now? FDA is working with the National Institute of Standards and Technology, and who is running the Genome in a Bottle Consortium, and we are creating standard reference materials for human genomes and microbial genomes. These will be materials that are highly characterized, and will continue to be characterized over time with new technologies and new information that we can gain. And these will be materials that can be purchased by anyone, and we'll have standard methods that go with them. So we're hoping that we're going to get at having the standards problem partially addressed, but we don't expect to have anything available for another year or so. We need to think about rational regulation of how these genomic systems should, and well how they should be regulated, and when I say genomics, I'm not just talking about sequencing, I'm talking about anything in genomics. I think a week or two ago there was a meeting that used to be called ISCA, it's now called ICCG or something like that, which started off around looking for copy number variations, I believe, in the human genome to look at underlying disease that can be detected by copy number variations. We need to think about rational regulation, what FDA can contribute, and what the clinical community can contribute. Because in my personal opinion, having 100 different labs all around the country, all doing their own thing, all spending their own money to validate their own system is probably not the most economical way to get at good testing, and it's probably got some holes in it that could be fixed with the right level of regulation. We need to look at databases and how we'll get at having high quality curated clinical databases, and we've kicked off a project at FDA where we've gotten funding from our critical path funders to actually contract out looking at all the databases in what the United States are everywhere, everywhere in the world to see, we're going to use people who actually do this for a living to find out who has what and how it operates and so on to see, we don't want to build a database, we want to help other people's databases be improved to the highest quality possible. So we're going to do a survey to see what's out there, and we've been working with NHGRI on this to try to work towards something that not only will inform, I think, patients and providers, but will inform the regulatory mechanism, you know, if it's in the curated database then it's probably right and we accept that kind of thing. So we're working towards that, and then we're following a lot what's being offered out there, and I don't know about if any of you are in the oncology field, but the area of testing for gene variants in oncology is absolutely exploded. There are, I don't know how many different companies that will test your somatic genome and tell you whether you have a K-RAS mutation and B-RAS mutation and so on, and then some of them provide you advice, some of them just provide you results, but we're following what's going on, and some of it's a bit unsettling, I can tell you. So we are hoping that the clinical community will help us to weed out the lousy players and not order things from them, but I'm telling you there are some lousy players out there. So what are our goals? We want widely available standards materials and standard methodology and welcome the participation. I think the genome and a bottle consortium, you can find it online, is working well together and lots of people are contributing towards building the kind of information we need to have to have good standards materials. What I would especially like and I think the FDA would especially like is instrumentation and software systems that are built under quality systems, and what does that mean? What are quality metrics? It means that you start out saying, what do I want to build? What do I want it to do? And then you say, how should I design it to do that? And how will I test it? So I can show that it does what I say it does. And you would be surprised, probably some of you, that almost anything you buy from a life sciences type company that provides instrumentation does not go through this process. They might go through it informally, but there will be in most cases no formal design control and they will have no way to show that they built, that they can routinely build what they think they're building or that it meets their needs. And I'm not knocking that, that's not a bad thing. In life science, you may want to rev a lot, you may want to do a lot of things in life science, but for clinical use, it's my view and the view of the FDA that a little more control is warranted. We really, really are, want to move towards the goal of useful databases so that we can build information and disseminate information that's actually based on real evidence that's been curated by people who know what they're doing. We want evidence to drive care and we need to define what actionable evidence is. In the case of cancer, actionable mutations has become the new buzzword, but nobody agrees on what that means and we need to define it. So to actually design studies that are intended to generate high quality evidence and not necessarily just a publication or something like that would be extremely, extremely welcome. I don't know how hard that is to get funded. We need education, really seriously education for interpreting the data that comes out of this and I'm sure everybody knows that decision support is included in this, how physicians will use this information and we did find on a website of a company who's offering stuff, you know, the question and answer for physicians about their genetic testing and the question from the physician was do I need to know about genetics and the answer is not really. I'm not sure that's true or advisable, but it does demonstrate sort of the direction some people are going is that the physician who's ordering this doesn't really need to know anything about it. We'll just give you the answer. Trust me, it's got to be right. It's genomics. So I think I don't know how education and good interpretation will happen, but it's pretty clear that we'll need it and I'll stop there. And as you'll notice, I didn't say anything about LDTs and I, so don't ask. No, the guidance is not, yeah, the guidance has not been published. Okay, let's see, we have many. So we have I think Brad, Mark, there was one behind Dan and Dan. I'm a little surprised you're getting into evidence and discussion of action ability and that sort of thing. That's part of our main mission at NIH as well. But can you expand on that a little bit? Yeah, from the regulatory viewpoint, if you say something's actionable, if you make a claim about it, then we need you to tell us what does that mean? It's very, very simple. I mean, we're not trying to establish what's actionable. We just need to know what do you mean when you say something is actionable? Do you mean there's a drug out there that may be targeted to that mutation? Do you mean? Yeah, I wanted to come back to a question that had come up earlier, which you didn't specifically discuss, which is also related to anticipated guidance from FDA related to clinical decision support related to their use as a device or their definitions of device. Could you update where that discussion is? I think the guidance is being drafted. I personally have run screaming away from clinical decisions of work, but David Litwak, my colleague, is working on that and on drafting that, so it's coming. And I can pretty much assure you that it's going to tell you what we think requires a submission, and what doesn't, and what we really don't even think is a device. I did a site visit review of FDA a year ago, and I know you guys have scientists who are sort of half scientists and half reviewers. I would imagine you'd need a huge army of people to handle this, to tackle this. Do you have expertise in place to sort of handle every gene or the whole genome? So the way regulation works is you don't really need expertise for every gene or genome, because regulation, while we deal with particular genes and genomes, it's not terribly different from one to the rest. The way we regulate IVDs is you look at analytical validity. That is how well does the system measure what it's intended to measure, and you look at clinical validity. Does the answer actually correlate well with the disease state or condition that you're looking at? And that's actually fairly easy to do in many cases without extensive knowledge, but you would be surprised. We do have a small army, and many of them are PhD-level scientists from the genetics field. Many of them are laboratorians who've come out of clinical laboratories and so on. So, yeah, and I mean, not to say the LDT word, but we actually don't get that many genetic tests come into FDA for regulation. So, and in fact, our scientific in CDRH, our scientists are not reviewers in most cases. That's more classical of the Center for Biologics. Mark Ritane, University of Chicago. I am an oncologist, and I share your concerns, and I do need to mention the LDT word. And so I think that there's different kinds of LDTs, obviously. There's the LDTs that a hospital will develop for its own patients and for which it may or may not get reimbursed on a piecemeal basis, and then we have LDTs being developed by companies, and many of which despite not bearing any resemblance to the safety and efficacy threshold of FDA have been deemed reasonable and necessary by CMS and therefore are being reimbursed and are being utilized by oncologists. And so the real problem out there is there's virtually no incentive for a company to seek FDA approval when they can make the same amount of money without FDA approval. And since oncologists are happy to order any tests that justifies treating a patient with a drug. And so I think that's the crux of the issue. And so somehow we need to create a system where maybe CMS would pay more when it's been deemed safe and effective, and so that you could still get, you know, if it's truly reasonable and necessary, that's fine, but safe and effective somehow should warrant a higher payment. Well, there's a huge disparity between whether you come to FDA or whether you don't now, you know, I totally agree with you. How you resolve that, there's probably multiple different paths, but I don't know that you want, that anybody really wants something that isn't safe and effective. Which I don't think is what you're saying, but... Well, no, I think patients don't know and physicians often don't care. Well, you probably know we've been working on this for a long time and have lots of ideas about it, but I couldn't agree more. There are perverse incentives for lots of things out there and not a lot of good incentives for people to sort of prove that their device does what it says it does. I wanted to just sort of highlight what I think is sort of a schizophrenia from where I sit when I look at the great mass that is the federal government and I understand that everybody at this table is part of that and you don't view yourself all as one part of one mass. But FDA labels for drugs include pharmacogenetic advice for dozens, I've lost count, in the germline and in the somatic genome. So I want to ask you and Steve to sort of comment on the schizophrenia that that creates on the part of practitioners who were told on the one hand the FDA label says that you should think about this or you should do it and it's always couched in sort of never completely coercive language and the fact that when people want to do that or when people want to act on it, there's this barrier. And I'm not saying that one, you know, there shouldn't be a barrier, but it does create this sort of how do you deal with that? How do you reconcile what it is you're putting in the label with what it is that those guys down the street are willing to pay for? You'll be surprised to hear, but we don't. No, I'm not surprised. Our directive is to determine safety and efficacy in the case of drugs without regard and safety and effectiveness in the case of devices without regard to who will pay for what when. So we actually cannot consider whether it will be paid for or what the price is. That is not part of our remit. So do we try to work together with CMS? We do. The statutory, it's not conflict, but differences make it so that it's not easy for us in reviewing safety and efficacy to provide all the information that would be required for a CMS reasonable and necessary decision. Thank you. You put you in. Can I just have quickly? For most of those particular issues where FDA has within a drug language this particular test is required, I think a huge portion of those are tier one codes and even if they're not, they're paid for. It is some of the newer codes where there is, we're now working through the contractors to see how we're going to pay for those. It may be an issue. But for most of the genomic testing that are being used to determine whether a particular drug is going to be effective or not in the oncology space in particular, most of those tests are paid for. Things like warfarin testing and so on, maybe what you're moving to. There is significant concern that those tests are being paid for and physicians are ignoring them. You do a HER2 test and you ignore the results. You do an oncotype DX and you ignore the results. You do a KRAS and you ignore the results. Those are all oncology tests. None of those are in the germline genome. The germline genome has 58 the last time I looked and I'm sure it's more than that now. CYP2D6, CYP2C19 and you don't pay for those. I'm just sort of highlighting the schizophrenia and trying to sort of get a handle on how maybe it's just that FDA has its own mandate and you have your mandate and that's the way things should be. That's pretty much what it is. They could be, you know, Congress would have, again, it's who's the Board of Directors. We have to operate under the laws that we have because our lawyers get really, really nervous when we step outside of that. I think you put your finger on the pulse here. I'm going to describe this a little bit more in the afternoon, but I think the discrepancies between the CMS approaches and the FDA approaches and other people approaches can be looked at as a discrepancy between clinical validity and utility. So for example, some of the gene drug labels that you have, I think the last time we checked there's more than 100 of these. There is no evidence of clinical utility. There are described gene disease associations and the FDA thought to put that as a service to the, you know, practitioners and the public. Whereas CMS uses a higher level of sort of evidence in terms of utility for reimbursement. So validity versus utility. Actionability sometimes falls in between and sometimes when we define actionability, we define it as only validity whereas actionability may be also viewed as utility. I think, you know, Blue Cross Blue Shield has their own processes. So I think a little bit of harmonization can go a long way in terms of understanding each other's definitions. Naomi, really the last word. So Liz, my understanding of how the FDA views drugs as being useful is something like it's likely to help a physician make a better decision as opposed to a more specific and definition of utility. Is that correct or is that change? You're talking about companion diagnostic? For diagnostics, I'm sorry if I misspoke. Whether this be imaging or laboratory, et cetera. I guess I would fall back on a recent guidance in which we talk about benefit risk, that the benefits of using the test should outweigh the risks of using the test. And because my group only really looks at IVDs, there are lots of tests that aren't IVDs, imaging and so on. And that's all in there is that there's some benefit that outweighs the risk. So that is a bit of evolution in the concept? That's always been there. That's always been the underlying thinking process, but we just published a guidance to sort of more formalize it in people's thinking. It's how do you know that something's safe and effective while you look at all the data and you determine that the benefit outweighs the risk. So a test, for example, could be not that great, but the benefit of using it might outweigh the risk of using it or outweigh the risk of not having a test at all. And there's a huge spectrum in that benefit risk assessment. Some things have to be really good, some things don't have to be that great. Okay, great. We ought to move on. Thank you very much, Liz. And I would just remind everyone that the purpose of bringing these groups together is really to find commonalities and ways in which the federal agencies and the various ones we've heard about before the break can work together to address some of the challenges that we're seeing in implementing genomic medicine. So we all recognize we've got a very challenging healthcare system. Some would say we don't have one. Until you try to change it, then you realize we do. But at any rate, let's try and keep the focus of the discussion on that if we can. Dr. Clancy.