 What I'd like to do is maybe Are you gonna change the subject completely or you okay? So what I'd like to do is sort of pursue this line of reasoning for a little or less discussion for a little while then if there are other Topics that we want to touch on In the in the discussion we can but I think this is we could probably spend the entire time talking about this because that's sort of the That's what we're here. So Rex go ahead. So I guess I wanted to make two points first is just a short reaction and While we may not want to be actually using some of that information clinically we should capture it somewhere It's really important that we not lose that because it all builds this prior probability argument that Less made and and I was struck in Gail your presentation talking about you know 44% yield on clinical exomes is is pretty impressive and you know taking that to the next conclusion of Crickets comment that we should just have a free-for-all. I mean if Hyperbole but but I'm just trying to make a contrast here. So, you know if the free-for-all produced the 44% Diagnostic rate you'd be quite happy with that and the question I guess Gail maybe this is for you is Is the reason that you were successful at the level of 44% because of all the prior probabilities that you were factoring in based on your clinical experience and can you Help us think that through For appendicitis the rate is 85% so what should it be for genetic testing or for exome sequencing? Yeah, tough question. So I'd like to think that There was some rationale. I mean I've looked at the diagnoses The only ones where it's been zero are in the severe autism kids So I see a lot of kids with autism I don't do exome in your average one because I think it's multifactorial We're not going to find anything even though companies say you might the really severe nonverbal ones that have severe intellectual Disability you might but even those have been negative it's been among all our clinicians and And You know we we don't do it on everybody Maybe it's selective for the kids whose insurance will pay for it, but I Think it's it's across the board, you know when we keep saying well It's going to go down because we're ordering it now We only took it on the worst ones that you've had for eight to ten years that you knew had something genetic You just couldn't find it and now it's sometimes it's out of frustration. They keep coming back You've done everything you can think of and well, this is the last thing I can do and even some of those Unexpectedly have been positive. So I Guess we need to wait a little more and see we're starting to move now into some of the other areas GI Rheumatology our he monk like neutropenia that I'm not as familiar with and Doesn't look good in like the Inflammatory bowel disease even the very young kids you got to be really young and severe for it to maybe be a single gene But we're trying to look at some of those families and select them based on severity family history Things like that. So I think at least in our case We still have a ways to go and when you take a lab like Baylor where they do thousands But they take whatever comes in it's kind of hard to compare So I don't know that I really answered it. I don't know why so There there are lots of people on my speaker list and some of them want it. No, no, no, that's okay Some of them want to address you specifically and I think we'll sort of stay on this thread for a while So Bob has been waiting patiently. Sure. So I think I'm just going to sort of extend what Bruce alluded to Which is that the the clinician their imperative is to decide what to do next with this patient Which doesn't always require Absolute bulletproof evidence And and so I think that's an important thing to remember and then the other point is that in in that decision process We have to ask whether the clinician is better off Is going to make is going to make a wrong decision more often without the information than with it Mike Yeah, again just extending some of the points in response to Howard's Presentation I think if you had presented to those nephrologist That your functional data implied a readily available therapeutic interventions such as ace inhibitors You would have seen a vast majority say we'll put that in the record and we'll we'll try it tomorrow Yeah, and and sort of following on again what Bruce raised in terms of the concern about misuse I mean it when Howard posed his question earlier this morning I think Liz said, you know, I would put something in the medical record and that was the way I felt I would put something in the medical record on probably suggesting that there is this variant that you know that looks Suggestive but but Bruce I wonder if you can advise you said sort of you know All the caveats on the world won't stop people from misusing and that's that's certainly true How do we make that judgment that Bob mentioned of you know, when is it likely to cause more harm than good? I guess all it depends on what the finding is and what the stakes are You know, I think you have to live with the fact that When you put a piece of data in the record, you can't control What eventually is going to be done with it and who's going to use it in one way or another? I think it's in my mind a strong argument to having Somebody in the loop who actually is trained in the area and and knows how to interpret this but at the end of the day I mean, this wouldn't be the only example of a piece of data in the medical record that gets misused. I mean I even crickets example of Imaging that happens all the time just because they say clinical correlation is Advised doesn't mean it's going to happen and there are lots of examples of nightmares with that So I guess my fundamental view is that we don't have to wring our hands too much any more than our colleagues in other areas do That we can't put this in a place because it will be misused I think we and all the people we train need to Realize that as black and white as this looks there are plenty of shades of gray that are not that obvious It's part of training it starts in you know early medical education It has to be a thread all the way through nothing you do I can tell you the most Clear black and white reports we've written I've seen be misused and you know you just at at a certain point you just have to live with the fact that The practice of medicine is like the making of sausage and we have to just do the best we can at every stage to use Information wisely. I don't think there's any body of evidence. That's going to be a threshold that You know guarantees this will never be used wrongly This was just in response to gales comment about what might the diagnostic success rate be if we sequence Not just the hardest cases the ones that have been in the works We did something of a concierge medical genetics practice data set MCW the diagnostic success rate there was 61 percent. So it's actually higher now arguably of course you could Got some of those diagnoses using the other cheaper test, but that's what it was Can I just make a comment so while Originally was like that. We have cases now that we move to much quicker someone comes to me I'll do an array and maybe a couple of simple things and then go to an exome if I'm going to do it And we don't have absolute criteria. So my colleague may do an exome on someone where if I saw the patient I wouldn't advice a versa in genetics, but I Glad to hear that that's great. Maybe we come learn from you The next person on the speaker's list is me So I just I just want to throw in the idea that While while there are data in the electronic record in the medical record that that people use or misuse they Should we factor in at some point the potential Extent of misuse or what would the consequences of misuse be so the misuse might be sort of ill-informed Changes in lifestyle for example in my own practice. We have I have a woman who has a variant which I am Sure does not cause the long qt syndrome at least in her her qt intervals are completely normal She faints and we know that her rhythm is normal when she faints and she showed up in clinic with a defibrillator Placed and we said what how did that happen and some doctor in the community did it and we talked to the office and The the office that the response was well. She has an abnormal genetic test. So we put in a defibrillator So that seems to me to be a sort of On the scale of things that can go wrong when you misuse medical data on one end So is that something that we should factor in and and and as I talk I think to myself I know what Bruce's answer is going to be can't you can't control what doctors what crazy things doctors do But but there is this potential of real real harm Yeah, there are some Examples that's one I guess where you could really alter the course of a person's life for the worst and and their kids Yeah, you know it comes in all different levels You know we have an undiagnosed disease program at our place and I didn't realize until this one day We saw a patient where we undiagnosed a disease what the term really? And this was a woman that came in With all sorts of problems one of which was she said she had spinal cerebellar ataxia and to me She didn't look like somebody with that and to make long story short. We it was an archaeological dig We found the report it said variant of unknown significance in SCN whatever it was and SCA and we called the lab this was seven years after the fact and it had been Reclassified as a benign variant and the clinician who sent the test Just you know cemented it in the record from then on did it hurt? I don't know really I mean there's no treatment for it She wasn't being treated differently, but she firmly believed she had it and I think there's probably a very large number of examples of that Sort of thing that people Misinterpreted what we're clearly labeled never mind, you know not labeled as variants of unknown significance Sometimes yes, maybe they'll end up with treatments or they'll undergo prenatal diagnosis I mean you can imagine a lot of really bad things, but How you know how bad is it that she believed erroneously what she had I don't know But it seems to me. It's not good medicine so the speaker was like right now looks like Mark Callum cricket and and Okay, yeah, I'm gonna Let's keep on going for a little while. Yeah, I just wanted to talk a little bit about culture And I think that one of the things that's been interesting is that the culture of genetics is that we are very comfortable in this Gray area Between the laboratory and the clinic in fact most of the discovery that's happened over the last 60 years has been based on this interaction between The basic scientists the laboratory and the clinicians going back and forth back and forth iterating constantly to try and develop new knowledge That's not a model that is really reflected in any other specialty I don't think And so the question would be getting back to the Presentation that Howard did of the Nephrology Society is that if this is in fact a cultural Issue How do we develop the culture of this type of bench-to-bedside and back? That we're trying to discuss at this because it's probably not going to be you know the geneticists And the genetic pathologist and the molecular geneticist solving the problem for the rest of medicine So that's kind of I guess a different Attack and Dan can say well, we're not going to talk about that right now But it's just something that as I look at these stories I say I live these stories every day every every one of our clinics is an undiagnosed diseases program And so we're comfortable with that, but that's not the reality for most of clinic clinical medicine So I'm not going to change the speaker order, but I would like people to start to think about Not sort of much where we are or we came from but where we should be going So so with that charge, I can actually Say that I disagree with mark I think fundamentally the whole of medicine is exactly like that It's just that most elements of medicine have not portrayed their test as a definitive end in its own right And I think that's what the problem has been with genetics is we've had this exceptionalism Where we make it as if it's a sort of binary test and there's no other test in which that's ever been the case It may have been set up like that, but in cardiology and oncology. These are that bench to bedside trip is a very regular Circuit and what happens is exactly what you see in in in a situation in which as you point out You have primary interpretation You have secondary post hoc interpretation and then you have an exchange between those two ends of the the causal chain To manage an individual patient. That's routine in clinical medicine. I think what's happened is that and it's I don't think it's genetics fault I think outside as Dan outlined there's this perception that the test is sufficient and necessary and I don't think we've conveyed from the community just what the The probabilistic analysis of the outcomes really is You know to tie it into to Bruce's comment some of this is availability of the data So if you look we did this in the med-seq study at imposter if you look at The ECG people will interpret because it's in every chart and you see patients. I have heart disease They'll interpret an extra atrial beat as if it's a risk of atrial fibrillation And people then make therapeutic decisions and situations of perfect aqua boys Based on that and in fact the data are the exact opposite. It's actually ventricular Premature beats that are more predict the likelihood of atrial fibrillation whereas sitting in the chart in the med-seq study was a Pit x2 Genotype that has a three to four acts risk of atrial fibrillation in a beautifully many beautifully Conducted population studies and so you get this sort of because one one test is available to everybody They know how they think they know how to interpret it It becomes Embedded in the therapeutic decision-making and another test because they've never seen it before is ignored or Con in on the opposite extreme is treated as if it's the be all and end all and then people are having Irrational therapies introduced and I think that's what we need to fix is exactly that perception and some of that I think is done through the availability of the testing and in education this cricket Maybe not a free-for-all, but certainly a much freer for all. Okay cricket So I'll add a couple of comments about that in terms of the free-for-all what I was implying actually and what I think is really Important is that we build a genetics community not just in genetics But in every discipline of medicine and until we have that we're not going to have that kind of thoughtful integration We're a colleague in nephrology can turn to someone say hey What does this really mean and I think that that's something that is very very very important for the future of genomic medicine In terms of burying things in a medical chart I would sanction what Callum said and maybe extend it a little bit simply by saying that when we look at Dare I say the word epic? There is a problem list. Okay. It is not a diagnostic list. It is a problem list and When Howard tells us that this person has a variant in a new gene, you know to me That would be akin to saying that the creatinine is a bit high. They have some renal dysfunction It's not a diagnosis. It's an elevated glucose. I mean we can sanction these Infra pieces of information so that they are front and center They are not lost, but they are also not overly interpreted And I think that most physicians are very comfortable with this and the reason I think it's important not to put them in some sequestered area is that we'd like to believe this field is just beginning and there will be no treatments and they may be relevant to that patient but until we have people familiar with the information see it not as a definitive destiny of the patient's medical outcome. I think that we're going to be missing the big picture Think a lot of the conversation to date has been what do I do to integrate using some Bayesian or intuitive method? The information that's out there, but in terms of the gaps Don't forget that medicine is becoming sort of a big data amenable science And I don't really think a lot of the conceptualization or computational data structures are out there Yet just to mention one example The level four variant so probably pathogenic well, does it actually mean we're let's say we're 80% sure It's 100% pathogenic or we're 100% sure it's 80% Penetrant and So I think you know computers are really dumb and we need to get data structures out there and and methods of collecting this big data To actually make it useful So less is next and then you Just following up a little bit on what Callum said and then previously Dan what you yourself said you meant you brought up the Mundane example of the appendicitis diagnostic error rate And I remember very clearly being taught in medical school by the surgeon that if in this case he wasn't removing 10 to 15 percent normal appendices that people would be dying from peritonitis and That kind of a thinking about which errors We actually want to make in order to allow us to help patients get better is a discussion We haven't much had in genomics based on Callum's point this naïve deterministic view and every time somebody makes a misdiagnosis or predicts an erroneous variant Pathogenicity assertion people are running around screaming with their hair on fire that we had you know committed the eighth deadly sin And that's just not the reality of medicine I think we need to move away from this obsessive focus that we've had for Three or four decades on all the harms that can be done by misdiagnosing people with genetics and move towards through functional data to find These pathophysiologic opportunities to use a variant to get a patient the ACE inhibitor if that's the right treatment We need to focus on benefit and not harm because harming patients happens every day in medicine And it's part of the errors we have to accept to make the decisions that allow us to get patients better So I was going to tell Howard that I thought that we should put the information in the zebra fishes medical record But I really agree with what Rex said Which was that we really should capture the information and maybe it's not ready for the medical record but we have to collect it and put it somewhere and When Gail talked about you know, who's who's interpreting the variant? Is it the laboratory? Is it the clinician and I think that wavering is really an answer that it was both And of course, it's it's collaborative so You know, I think what we have now is a situation where Often it's not so collaborative and all the lab really gets is sort of that top level diagnosis and you know, you have the benefit where The lab might say this is pathogenic in this gene and you look at the phenotype that's expected and say well They're supposed to have all these other things and I don't believe it So I think part of what we have to think about is to capture really all of that You know in ways that are categorical and coded and machine readable And not only are about one case by one case But really kind of sum up the sorts of phenotypes you see with you know in that setting So that it's all being collected and that can't happen is I think cricket is saying unless really there's you know broad participation by all the types of Clinicians and if that's the case then they need to be enabled to assist with this data collection and to round-trip it Celia Hi So I wanted to just make a comment about You know the difficulties that we're obviously appreciating with linking variants with disease and I think a lot of that may be related to the fact that we have to take the genomic Context into consideration even from Mendelian disorders Patients with the same mutation may present differently and I think of course it's very difficult You know in terms of taking all of that genomic information and integrate integrating it But I think that's one of the bioinformatic challenges that we have I think if we're able to really incorporate all of the variants in an individual it may give us much better predictive power in terms of understanding Why the patient present as as they do and also what these different variants may may mean And the other thing I wanted to just briefly comment was Bruce's Question about who cares And I think that it's in terms of looking at variants It's not only with regard to diagnosis, but you know from a clinical perspective It's really outcome that you really care about Not only whether you've diagnosed the patient But what is the long-term outcome and to me there seems to be a gap here as well in that We don't really have a systematic way of linking variants with outcome I don't think there are any databases out there where you can easily search research and Catalog that information. I may be wrong on that Maybe there's something that I'm not aware of but I don't think there's a general Mechanism in which that information can be easily deposited and and researched and I think ultimately what you want to know is not just diagnosis of the outcome, so I would like to make a plug for as a Sort of a centralized way in which we can start collecting that information and eventually maybe help to link variants with clinical outcome So conveniently enough the next person on my list is Aaron who has to say something about clingent I say something about clingent Right now so some of the things that we're spending a lot of time curating our first clinical validity, so You should back up and just tell tell us in 30 seconds what clingent is because so sounds like it Not everybody. Okay, so clingent is an NIH NHGRI mostly funded program To develop what we're calling sort of a central resource Defining the clinical relevance of genes and variants working really closely with ClinVar, so You know we're encouraging labs and researchers to submit data to ClinVar that summarizes their assertion on the relationship between a variant and disease And then within clingent we're actually spending time answering a few critical questions By doing curation one is on the clinical validity of a disease gene disease pair, so we've started doing a lot of that curation for example we've Looked at a lot of the testing panels that are out there and some for example with I think pancreatic cancer some other rare cancers at least 50% of the genes that are on those panels show Limited moderate to limited evidence of clinical validity yet. They're on those panels and the clinicians then who already don't necessarily understand how to use that data are getting results back from Variants and genes that show limited evidence for clinical validity Another thing that we're working on is I think actually you mentioned to Gail taking the ACMG guidelines and trying to standardize how we can use those for Assessing the the pathogenicity of variants so we have a launched a variety of pilot working groups that are taking that that model and testing it in Cardiology like Newton syndrome for example, and we have a cancer working group looking at p10 and a ton of other genes But my comment was going to be so a lot of work Which is trying to pull all of this together With both of those curation methods validity and the pathogenicity We are trying to develop frameworks for incorporating the functional data But right now the curators have to go into the literature and extract the data And like you said it that's really challenging to do because it's really not in any kind of standard format So it'd be incredibly helpful for us if we could figure out a model I don't know if it's working with the journal editors or just getting the community the basic scientists a custom to Putting that data and databases straight away that we could then tap into instead of having to wait for it to come out Through publications, but that would be incredibly helpful And then one other thing I come and I was going to raise was back to the education And I was wondering Mike if you want to dimension the maintenance of certification effort in clean gen so trying to develop modules for educating clinicians on how to Interpret the varying assessments that are out there Like I'll let you go even though you're not in my list. No, that's that's fine And I was gonna ask I was gonna actually gonna tweak Dan MacArthur just to sort of Make a comment about aggregating large data sets and this business of adding a column to x-act that we talked about last night So Yeah after after mark sure sure so just briefly the the efforts of clean gen we came quickly to the realization that You know, there are clinicians out there that could potentially contribute to this data and one of the motivations might be Their maintenance of certification requirements. So there's now a maintenance of certification Module which is being launched by the a BMG the medical genetics board whereby Individuals who have received variants of unknown significance as a clinical report for a patient We'll go back after a period of six to twelve months and Review that data and update it and then feed that back to clean gen So it's it's creating a loop that we think will be helpful So just you're gonna talk tomorrow. I think we're late today, but yeah I mean, I'll certainly discuss this more in a talk this afternoon, but So Dan and I chatted very briefly last night about the How great it would be if we could take the large scale Variant data that we currently have in the exact database Find a variant that you've seen in one of your patients and then look up to see whether those variants in the exact system Also have that particular variant. I think again as I'll mention a bit more this afternoon We should have more discussion then there's a there's a number of key challenges between between us and that goal One of them is consent most of the samples an ex aqua collected opportunistically. They're not consented for Releasing phenotype data in many cases. We just don't have the detailed phenotype data for those for those individuals and and then I think there's There's all sorts of issues around how we actually release that data in a way that doesn't enable Reidentification, so how do we how do we release enough phenotype data per variant that it's still useful to clinicians without Somehow allowing people to link that information together and and break break the privacy that that we you know Have consented those individuals under these are all very tough problems, and I think we should definitely discuss them more this afternoon Go So one of the things I've been thinking about as I've been listening to discussion is as a basic scientist We're very used to evolving information and evolving knowledge. So something that's a variant of unknown significance today Tomorrow we could put in a different category But you can imagine if you put that variant information in a patient's record today, and then tomorrow You know something new about it. It's that whole cycle of Updating our knowledge that's now captured in a patient record. I is that a barrier? I would assume that's a barrier in on the clinical cycles, you know from for basic researchers You know there's databases that capture that so you can imagine ClinGen is going to be updated all the time as new knowledge is published and released about the significance of variants But how that actually then gets back into a patient record and can influence subsequent patient care I don't actually know what that process looks like Yeah, I guess I could answer that one real quick again the problem list I think is the best place It's just that again, they have to understand what it means So if we take a pharmacogenomic variant and we're supposed to put it somewhere and I put it in the problem list theoretical risk based on exome sequencing that there could be You know slightly slower metabolism and then a neurologist doesn't give a patient Medication because they think for seizures because they think there's something bad here So I think again, it's how how do you write it in a way that's correct? But also that someone can quickly understand it, but I think the problem is can be updated And so we do that and I go back and change things and and I would assume Bruce You do that too that that's something that goes they show up in the ER if they show up You know in a different clinic. It's there and they usually will see it You're talking about seeing things that are in the problem list. Yeah, I mean well I think any clinician knows the problem list is a work of You know some fiction some fact So mark yeah, so So specific to this. I mean you're right Carol. This is a barrier. It's one that we know we have to overcome There are some innovations out there Probably the one in the genetic space from the laboratory perspective. That's the most advanced is the Harvard Partners gene insight where they actually do have the ability to create a tether to electronic health records where they can Send updated information again. It's still up to the clinical side to somehow implement that updated information, but I think we all recognize that We have to have the ability to update the information, but I think it raises a larger question Which is when a test like an exome is done in a clinical setting What what if any are the responsibilities of the laboratory or of the clinical side to Re-look at that information since we know it can be re-looked at and there's at least one case now In the liability Area where a laboratory is being sued on the basis of not going back and updating a variant and if that is Found in favor of the plaintiff that will completely change this discussion because it's not going to be a theoretical discussion about how Do we do it? It's if we don't do it. We're going to get sued. So on that note not really We're now two minutes into the break and we'll reconvene at 11 o'clock So we have an 18 minute break