Sep 27, 2017
Jane Ferguson: Hi, everyone. Welcome to Episode Four of Getting Personal: -Omics of the Heart." I'm Jane Ferguson, an assistant professor at Vanderbilt University Medical Center. This month, we have a special feature from early career member, Andrew Landstrom, who went to the Heart Rhythm Scientific Sessions in Chicago earlier this month and talked to some of the scientists who presented their research. So listen on for interviews Andrews conducted with Anneline te Riele, discussing the challenges and opportunities related to incidental findings in genetic testing, with Ernesto Fernandez, describing his research into whole exome sequencing and Long QT syndrome, and with David Tester, discussing novel variance and pathway analysis in Sudden Infant Death Syndrome.
Andrew : My name is Andrew Landstrom and I am from the Baylor College of Medicine Department of Pediatrics' section on Cardiovascular Disease. I'm here at the 2017 Heart Rhythm Society Scientific Sessions. Anneline, will you tell us a little bit more about yourself, and what brought you to HRS?
Anneline: Sure. So my name is Anneline Te Riele, I am a physician from The Netherlands. I finished my medical training in 2012 basically, in The Netherlands, and I started doing a PhD on ARVC in a combined project of our Netherlands patient as well as a group at Hopkins. So what brought me to HRS? I think of course the science. There's a lot of very good science. Actually, I think it's the best meeting for my purposes.
Andrew : Absolutely. So will you just start by telling us a little bit about the spectrum of genetic testing in the clinic and about both the opportunities and the challenges that it brings?
Anneline: Sure. So what we do in clinic, and I think this is really the challenge that we're facing currently, is we have moved from just testing on gene or one small panel of genes to bigger panels and then to whole exome or even whole genome sequencing. And I think the good part of that is that in certain cases, certain well-selected cases, you'll get a higher change of actually finding that gene that is responsible for disease.
On the contrary, it also leads to a lot of incidental findings. So findings that you were not expecting based on the phenotype of the patient and then you need to deal with those abnormalities that you've found and that brings on a lot of challenges as well for the family but also for us as physicians. Do we then need to screen those families, what do we do with this patient, do we treat them with medical therapies or drugs or do we give them ICDs? That kinds of question. So that I think is a virtually important part of what we're currently dealing with in clinical practice.
Andrew : It does seem to be a very widespread problem. And here in the US of course we have the American College of Medical Genetics guidelines about reporting a variance. How do you think that that plays into the increased genetic uncertainty here in the US at least?
Anneline: So that's a great questions. In 2013, the ACMG produced a guideline on which genes to report if you find these incidental findings. So 24 of these genes, and that's actually a big number, 24 of these genes are cardiovascular genes and that's mainly because changes in cardiovascular genes may detrimental effects down the line and really cause death or certain morbidities that are really important for the patient so we do need to deal with that.
And the problem with the ACMG guidelines and especially the pathogenicity guidelines is that they require two aspects. They basically require first that the variant was seen before in other cardiomyopathies or in this case other patients with disease. And that's really difficult for cardiomyopathy genes because these are large genes, they have a lot of novel or private mutations in there, so it's really hard to fulfill that requirement of having been seen before.
And the second thing is that the ACMG guidelines require functional studies as another proof of evidence of pathogenicity and of course, I think we would all like to do that in all of our patients, but it's just not feasible for financial purposes and all that. So that's a problem that we're facing. There are options and solutions but I think we'll talk about that later, but yeah, I think that's a problem that we're facing.
Andrew : So on the one hand you have the ability to make a diagnostic decision based on a clear finding, but oftentimes the threshold to calling it a clearly pathologic variant is very high and oftentimes it never rises to that so it becomes more genetic uncertainty.
Anneline: Yeah. I think that's basically right. And of course in an ideal world, we'll have certainty and say this is likely or this is definitely pathogenic, and this is likely or definitely benign, but in the real world, really, I think maybe even 80, 90% of the cases were in that gray zone in between and we need to deal with that.
Andrew : Yeah, yeah. And you had some great resources that both scientists and clinicians alike can apply to these unknown, uncertain variants that might clarify things at least a little bit, and what are these tools?
Anneline: So of course, from a traditional perspective, we have always looked at in silico predictive programs, we'll look at segregation data, and I think they're all very important, but they all have limitations, so for example, in silico predictive programs, they likely overcall mutations deleterious and segregation data is nothing more than evidence of pathogenicity of a locus to a disorder, not necessarily that variant, so the new things that are on the horizon, and a thing that could be the future of [inaudible 00:06:04] interpretation is collaborative project so really we should be collaborating, we should not be having our own little islands. The collaboration is the key here.
And collaborative efforts in the US have been for example, ClinVar and NHLBI funded effort, as well as ClinGen and ClinGen, or Clinical Genome, is perhaps the, at least it claims to be, the authoritative central resource to go back to that curates variants as being pathogenic yes or no. And I think these databases, ClinVar finally has a database entry, so the variants will be in ClinVar, but ClinGen provides an expert panel of individuals who will curate these variants as being pathogenic yes or no. I think that is a central resource that we should all be aware of. I know these are not the only ones, there are other collaborative efforts out there.
I mean, there are ways to connect clinicians, so for example, Match Maker Exchange is a website that you could use to enter your variant and the phenotype of the patient and you submit your own information and then you'll get matches in other databases, but not only your own match shows up. So if, say, two years later, another physician comes up and looks for the same variant, you'll get a pop up, which will actually be very nice for these clinicians to get in touch. So that's, I think, the feature ... future of variant interpretation is collaboration. That's basically my, I think my main important message here.
Andrew : I think that's absolutely right. I think this has become sort of a big data question that requires many perspectives, and a lot of resources to be able to curate accurately. What are some of the limitations of these tools that you've seen that kind of, you have to keep in mind in terms of trying to determine whether a variant is truly pathologic or not with a patient that you have sitting in front of you?
Anneline: So that is, I mean, of course, there's many limitations in the things that we currently do because there's so much that we don't know. But for example, to give you an example, ClinVar I think, is one central resource that we should all be aware of and if you go to ClinVar, there is actually data from two years ago, and I'm sure the numbers are high if we would look now, but if we look in ClinVar two years ago, we already saw that of the, say 120,000 variants that were in the database, 21% of these variants were called VUSes but if you look at these variants, 17% of the cases, the labs or the individual submitters of ClinVar didn't agree on the actual classification of that variant.
So the limitations that we all should be aware of is that there is not one single solution and you should look for evidence and really research your variants. So look at Popmap, look at what is out there, look the patient of course, look at the clinical phenotype, does it match what you think the gene should be doing or not, or is it completely unrelated? And then of course search these databases but be aware of the fact that there may be errors there.
Another thing I want to highlight too is that we typically go to population databases, so Exome Variant Server, ExAC, I think these are very popular databases that we use to look at the frequency of variants in a selected population. But really these databases may have sub-clinical disease patients, so I know ExAC has three NYBPC-3 mutations that are known to cause HCM, so this is something to keep in mind. There's not a gold standard truth if you open these databases, but you should have multiple pieces of information when interpreting your variant.
Andrew : And that's a good point. I think with a lot of these cardiomyopathies and channelopathies, particularly some of the more frequent ones, when you have a database of 60,000 people, at least a couple of them are going to have disease.
Anneline: Yeah. I think that is part of the problem. I mean HCM is pretty prevalent, I mean one in 500 individuals likely, I mean these are recent numbers, has the disease. So I think the cutoff of a minor allele frequency of five percent, which is in the ACMG guidelines, I think is way too high for this disease. So this is what the cardiovascular expert panel of ClinGen has done, so they ... This is, ClinGen, as you might know, Clinical Genome, is a one-on-one team of curators that know the framework of ClinGen and then there is disease experts that are very well accustomed with the disease and the genes associated with it. So they provide teams and these teams work together, and the cardiovascular expert group has recently published a modified, or customized, ACMG guidelines on how to deal with the intricacies of the cardiomyopathies and for example, NYH-7 which is the first genotype deposed in ClinGen or in ClinVar finally.
So they modify that cutoff, the minor allele frequency of five percent, which is the BA-1 ACMG guideline cutoff, they changed that to 0.1% and I think that's exactly what you were saying, that is important to keep in mind, some of the cardiomyopathies are way more prevalent so you should not consider that if you see it in a population database that you think that it's, then it's normal, it's not necessarily the case because this is a prevalent disease.
Andrew : Yeah, and particularly when commercial genetic testing companies all can't agree that a variant is bad, and we all can't agree that a healthy variant may or may not be good, there is definitely a lot of genetic uncertainty there.
Anneline: Exactly, exactly.
Andrew : Now, whole-exome sequencing certainly has its role clinically, even with that genetic uncertainty that we spoke about, but it has a clear role in genetic discovery as well.
Anneline: Sure.
Andrew : And you were part of a very recent paper, and you led a very long list of authors, speaking more about your collaborative approach to genetics research that evaluated a novel substrate for ARVC, is that correct?
Anneline: Yes. So this is something I'm actually pretty proud of. As you said, it's a collaborative effort, so it literally take a village to do these kind of studies and we're lucky enough to collaborate with a lot of people who are interested in the same topic. So what we did ... and I metnioned to you in the beginning, I come from the ARVC field ... So what we did is we had one ARVC patient that was discovered by whole-exome sequencing to carry an SCN5A variant and we, in and of itself, found that that was very interesting, because SCN5A, as you know, has been associated with Brugada syndrome predominantly but many other cardiomyopathies as well, so DCM, even ACM. There's been a lot of controversy about SCN5A in that matter.
So the computational data, the population data, it all pointed to the fact that this variant may be pathogenic, but we weren't really able to connect those dots just yet. So we then collaborated with the group in NYU with Mario Delmar, who did, first of all, functional studies on the sodium channel, but what was nice is that he was able to use his novel method of super-resolution microscopy which is a way in which we can look at the nano-scale structure of the cardiomyocytes, or really the small, small levels of molecules that you see in these cells. And what we did is we found that not only NAV1.5 which is the gene product of SCN5A but also [inaudible 00:13:53] which is an adherence structure molecule, which links the cells together was actually less present in our ARVC patient compared to the control. And this was in the IPS so cardiomyocyte molecule, which we corrected using CRISPR-Cas9 technology so I think at least in current practice, on of the best pieces of evidence that we can get.
So I think this shows that our SCN5A variant, I mean, in this case, probably really was pathogenic, but also in a pathophysiological standpoint, explains to us how SCN5A mutations, which are typically thought to be only affecting the sodium channel, can also lead to cardiomyopathy phenotype which has implications beyond the ARVC world, but also in DCM I think this is a nice finding of collaboration that I think ... I hope more people will look into this.
Andrew : Absolutely I think the trouble with SCN5A is exactly like you were saying, it's been implicated in Long QT, Brugada Syndrome, SIDS, [inaudible 00:14:57], now ARVC, and even nodal disease, like sinus syndrome and things like that. So the ability to show sort of mechanistically, that while you have a change in your sodium channel gating that you also have a change in the way that the cells can connect with each other and form contractile force is, I guess, key to your study.
Anneline: Yeah, yeah. I think this really, I mean, I'm hoping at least, it was also finally published in a journal that looks more into functional studies, so not necessarily only genetics, and I think we need to work closely not only on the genetic side, but look closely at the pathophysiological standpoint for gene discovery purposes because this will really explain to us why one gene is implicated in one disease, and also it points to possible directions to perhaps stop the disease process and treat these patients, which I think is vital in our clinical practice.
Andrew : So are SCN5A mutations in ARVC a common finding or are they rare?
Anneline: So they are pretty rare. I mean, we do find them every now and then and maybe they're modifiers. So what we did to follow up on that one individual, we check 281 ARVD patients who were screened just by regular screening, not by whole-exome but we did a targeted screening of SCN5A and we found five variants in these 281 patients, so that's two percent. I mean, it's still rare, but it is as rare as any other minor gene causing ARVC, but it is a rare feature, so I mean, I think it could be a player. And interestingly, the phenotype didn't change much. It wasn't really different from the ARVC patients without an SCN5A mutation which is reassuring.
What we also saw is that the prevalence of mutations in those with desmosomal mutations. So ARVC is, as you know, typically associated with diseases or mutations in the desmosome. It was more often seen in those without a desmosomal mutation. That was almost double as frequent as in those with a desmosomal mutation. So it does give us some direction to the fact that this may be a player out there. I mean of course it's not Plakophilin-2 which is the major player, I think, in ARVC, but I think it may cause a, at least a certain form of cardiomyopathy of arrhythmogenic cardiomyopathy that we need to be aware of.
Andrew : And how do you think your new discovery of SCN5A being associated with ARVC, how do you think that plays into the bigger discussion we were having about expansive genetic testing and what that may mean for a patient as far as diagnostic utility but also limitations of variant interpretation?
Anneline: That's a great question. So I think we should be cautious of saying this gene causes only this disease, and I think this is a common feature not only in ARVC but in a lot of cardiomyopathies and even in channelopathies. I think the concept of one gene causes one disease is outdated. We know that multiple genes have multiple effects and this SCN5A, of course the gene product is NAV1.5 which is the major alpha subunit of the sodium channels so it is really not the canonical function of SCN5A or NAV1.5 that causes cardiomyopathy here but it's a non-canonical function so I think we should be aware of the fact that gene products have different functions and that there can be overlap of the cardiomyopathies. So of course I think we should be screwing SCN5A in our ARVC patients and I'm hoping a lot of labs and a lot of physicians are already doing that, but it's really not the only thing that is associated with ARVC. So that's important to keep in mind.
Andrew : What do you think the next steps are for sort of broadening the implication of your finding?
Anneline: So what we are doing currently, and is a little bit of a sneak peek, because this data is not really out there yet, but we have, in this cohort, we found these five variants in 281 individuals, and we're currently working on one of these individuals to get another IPSO cardiomyocyte cell line and look into the functional components to that. And interestingly, this variant, that exact variant in that ARVC patient was also found in a Brugada Syndrome patient. So wouldn't it be nice to actually set them side by side and see what the differences are?
Of course this is a little bit of a future music, if you know what I'm saying, like this is something that we don't have just yet, but I think what we need to figure out is how epigenetic or environmental factors play into this field and to explain how one gene or one variant, even, can cause opposite functional effects in different phenotypes.
Andrew : What do you think is needed to help clarify some of the genetic uncertainty you see clinically?
Anneline: I think a lot of collaboration, a lot of money, quite frankly. I think we need to ... I mean, the functional data is really helping us not only for understanding that single variant, but also for gene discovery, and as I said, for treatment down the line, that is necessary, and I think the variant of uncertain significance, I mean, if we all live on our little islands and only do our little practices, then we're not going to go a lot further. So we need to work together to understand what your patient has in this variant, my patient had in that variant, and this is our phenotype, so we need to connect those dots to be able to make certain conclusions.
Andrew : Well, I'm all for collaboration, as well as additional money, that's good.
Anneline: Good.
Andrew : Well, thank you so much for spending time with us.
Anneline: Sure.
Andrew : And again, congratulations on a wonderful presentation.
Anneline: Thank you very much.
Andrew : I'm joined by Dr. Ernesto Fernandez from the Baylor College of Medicine to talk about his research project. Ernesto, I'm wondering if we can just start by introducing yourself and what your project is.
Ernesto: I am a second-year pediatric resident, I'm applying to a cardiology fellowship right now and I'm interested in, obviously, all aspects of pediatric cardiology. We're trying to figure out whether testing for Long QT genes or Long QT syndrome is actually warranted in otherwise healthy individuals. We're trying to see what the yield is on these testings, specifically whole-exome sequencing.
Andrew : And I think this project really hits on an important point, whereby, because we've been able to interrogate the genome more comprehensively with clinical testing, that we've run into more incidentally identified variants. And these variants can pop up in genes, like the genes responsible for Long QT syndrome. Talk a little bit more about these variants, what the implication is of finding these variants incidentally, and what your project hoped to target as far as the diagnostic value of these variants.
Ernesto: Yeah. So I guess the answer to your first question is that we are coming up with these marvelous new techniques of analyzing the genome and now we're using whole-exome sequence testing to look up is someone has any exome that's abnormal and this has caused a huge problem whereby we're now finding all these variants that we don't really know what they mean. We call them variants of undetermined significance.
Our study is basically premised by the fact that if you have no underlying suspicion for any arrhythmic disease, there's really no need or no indication to be referred for whole-exome sequencing testing, given that the most likely result is a variant that we don't really know what it means. And it's probably going to be benign.
Andrew : So on the one hand, you have a well-established gene panel that's being used for diagnostic purposes with you index of suspicion being high for Long QT syndrome versus something like a whole-exome gene screen where somebody may not be thinking about Long QT syndrome as a diagnosis and have low pre-test suspicion but then comes back with a variant found in these genes sort of incidentally. Is that sort of the dichotomy you're drawing?
Ernesto: Yeah. I think the best way of explaining it is through Bay's Theorem whereby if you have someone with a high index of suspicion when you start off to have sudden cardiac death, a family history of an arrhythmic disease, and you get a test for it, such as a gene panel for Long QT syndrome, and they come up with a positive test result, then you're going to say, "Oh. I should probably evaluate this further," whereas if you have someone who has some dysmorphism, they have delay, they might have seizures, but there's no family history of sudden cardiac death, no personal history of syncope, then there's really no need to send off this big gun, the whole-exome sequence, because you're likely to either get a normal variant or you're likely to get a variant that we don't know what to make of.
Andrew : So I think, Ernesto, that nicely summarizes the clinical question that you had in mind. What was your hypothesis going into the study, and how did you seek to approach that hypothesis, sort of experimentally?
Ernesto: So we came up with the hypothesis that if you have an incidentally identified variant within the whole-exome sequencing tests without any other clinical suspicion, it's likely to represent a benign finding. We went about by analyzing the data from the Baylor Miraca labs on the whole-exome sequencing data that they achieved, and we looked specifically at individuals who had gotten these tests and found to have a variant of undetermined significance, or had a pathologic variant for either one or all 17 of the genes for Long QT syndrome. We compared them to individuals who had known Long QT syndrome that had undergone genotype testing, and we [inaudible 00:25:21] these individuals from the literature. And we wanted to compare the whole-exome sequencing cohort to individuals who were otherwise healthy and had obtained a whole-exome sequence. So these are patients or individuals from the well-established ExAC database that are believed to be ostensibly healthy individuals.
Andrew : So if I understand you correctly, you're comparing this unknown cohort, that being the rare variants found in whole-exome sequencing, against a positive control cohort of pathologic cases versus a negative control cohort of healthy individuals derived from the ExAC database to look for whether those west variants are more similar to the cases or the controls. With regards to the west cohort, what was the prevalence of individuals with these incidentally identified variants, how many did you find?
Ernesto: So we actually found just about 49% of individuals had some variant in Long QT syndrome gene, and noted that about 12% of them had a mutation in the major causes of Long QT syndrome, and just over a third, or 36% had a mutation in the more rare causes of long QT syndrome.
Andrew : That's a pretty surprising finding. So you're saying that one in two individuals who get whole-exome sequencing sent for whatever reason, have a variant in a Long QT-associated gene?
Ernesto: That's what the data suggests.
Andrew : And where did you go from here?
Ernesto: So from there, we went onto compare the variant frequency between the case's cohort, those individuals with known Long QT syndrome, those individuals in our west cohort from the Baylor Miraca labs, and those individuals from the ExAC database who are otherwise healthy. So we noted that in our west cohort, there was about 13% of individuals who had a positive variant in the Long QT syndrome one through three genes, the major causes of Long QT syndrome. When we compare that to the ostensibly healthy individuals from the ExAC database, it was 12% in that study that had some variant in Long QT syndrome genes that are major causes of Long QT syndrome itself.
This was statistically similar, it was indistinguishable. And then when we compared it to the pathologic cases, it was actually about 50% of those cases who had a positive variant in a Long QT syndrome gene one through three.
Andrew : So there was a relatively low frequency of individuals who had variants in one of the big three Long QT genes in both controls and the west cohort, and was obviously much higher among individuals with a diagnosis of Long QT syndrome.
Ernesto: Yep. That's exactly what we found.
Andrew : And where did you go from here?
Ernesto: And then from there, we had a good idea that there was probably a big difference between cases and west, but we wanted to make sure, gene by gene, that there was no difference between our west cases and the ExAC database, the control cases. So we mapped each variant frequency by gene for the major causes of Long QT syndrome. There was no statistically significant difference between the west and the controls.
Andrew : So the gene frequencies between the controls and the west were indistinguishable and very much different, both of them, it would seem, to the pathologic cases.
Ernesto: Correct.
Andrew : And you then looked at the position of these variants, the actual amino acid residues, correct?
Ernesto: Yeah. So we looked at, for KCNQ1, KCNH2, and SCM5A, the three major causes of Long QT syndrome, one, two, three respectively, and we mapped out the amino acid positions where there was actually a mutation for each individuals. So the cases, controls, and pathologic cohorts. We determined the percent overlap between the west cohort and the controls and the percent overlap between the west cohort and the cases and noticed that for all three, there is a huge preference for west and control versus west and cases.
Andrew : So if you're a west variant you're more likely to reside in the residue also occupied by a healthy individual variant as opposed to a pathologic variant?
Ernesto: Yeah. Exactly.
Andrew : And so what did you do next? You retrospectively looked at some of the charts of the patients who were seen at Texas Children's Hospital, correct?
Ernesto: Mm-hmm (affirmative). So then we had 223 total individuals that had an incidentally identified variant within one of the major three genes, the Long QT syndrome genes. We looked at the reasons for their referrals and noticed that the vast majority of individuals were referred for some developmental delay, for some dysmorphism, for a non-cardiac cause, and then it was only about 23% of these individuals that actually had a reason for referral that was cardiac in nature. And less than on percent of individuals were referred for a solely cardiovascular reason. And we concluded that it's unlikely that these individuals were referred for a cardiac reason, as the data suggests, and that as a result, the index of suspicion for an arrhythmia is likely lower in these individuals.
Andrew : And what did you find when you looked at the charts of those individuals?
Ernesto: We had EKG data for a good number of them, and we excluded individuals who obviously had no EKG data, and we excluded individuals who had some congenital abnormality and then anyone with any other arrhythmia that would make the QTC interpretation more difficult, such as interventricular conduction defects.
We ended up with 62 individuals and 61 of them had a normal QTC, so there was no evidence of QT prolongation at all. There was one individual who was left who had borderline elevated QTC of 460, which was our cutoff for borderline elevation and this individual had actually been seen by pediatric cardiology at Texas Children's Hospital and found to have ... a history of syncope and it was found to be non-cardiogenic in nature.
Andrew : So matching the variant data which suggested that you had likely found background variation in the west, you found no evidence of Long QT syndrome in these individuals who had variants in Long QT genes.
Ernesto: That's correct. So, the overall percent was very similar between the healthy individuals and the west individuals. The variant frequencies were almost indistinguishable, and then the variant co-mapping for all, for both the west and the controls, was preferential to the western cases. So that kind of matched what we found in our study, that there was no clinical suspicion or clinical diagnosis of Long QT syndrome in these individuals who had been found incidentally.
Andrew : Well that sounds to me to be a pretty big finding.
Ernesto: Yeah. I think it's pretty important to get this information out there.
Andrew : So what do you think the take home message for your study is?
Ernesto: I think the take home message is if you don't have a suspicion of Long QT syndrome or of an arrhythmia, there's low likelihood that such a big gun test as the whole-exome sequence is likely going to change your mind.
Andrew : So Ernesto, what would you advise a cardiologist who maybe gets a patient in clinic with a chief complaint of a VUS in a Long QT associated gene picked up on west, what would you advise based on your study findings?
Ernesto: They're going to have to determine their own pre-test suspicion. They're going to have to get a good history and physical, probably get a baseline EKG to determine what the QTC intervals are, and if there's really no other clinical suspicion for Long QT syndrome, they're likely to be able to provide reassurance at that point in time.
Andrew : Ernesto, what do you think the next steps are for this project, and what do you think still needs to be done in the field to reinforce your conclusions?
Ernesto: I think my study is one of the early studies of this field, so getting more studies like this and other channelopathies, getting not just looking at Long QT one through three but looking at all of them, and in patients who've been evaluated at Texas Children's or any other institution would be helpful. And then moving forward to give more credence to the idea that if you have history that's reassuring and physical exam that's reassuring, then you probably don't need to have further testing.
Andrew : What do you recommend if your index of suspicion is high for Long QT syndrome, so maybe a QTC in the low 480s, maybe a family history of syncope or seizures, do you think whole-exome sequencing is the way to go?
Ernesto: Right now, that's probably not the best test, given all these incidental findings that we don't really know what to do with. There's other tests that are more high-tailored for those specific diseases, like Long QT syndrome panel among others, that are probably more likely to give you a positive post-test probability.
Andrew : So testing for the disease you're suspicious for as opposed to testing indiscriminately?
Ernesto: Yeah.
Andrew : So Ernesto, thank you so much for taking the time our of your day to speak with us.
Ernesto: Thank you, Andrew.
Andrew : I'm here with David Tester, senior research technologist working with Mike Ackerman at Mayo Clinic, and he just gave a wonderful talk on whole-exome sequencing and next-generation sequencing as an unbiased look to determine underlying causes of Sudden Infant Death Syndrome, or SIDS. So David, I'm wondering if you can introduce yourself and talk a little bit about your project.
Dave: Sure. I'm Dave Tester and I'm at the Mayo Clinic, again with Mike Ackerman. Dr. Ackerman and I have been together for about 18 years now, with a real focus on genetics of sudden cardiac death disorders. So this latest study was looking at whole-exome sequencing in a population of SIDS cases in collaboration with Dr. Elijah Behr at St. George's University in London.
And really the approach, what we were aiming for is really kind of two-fold. First we were looking to determine what is the yield of ultra-rare variance within genes that have been implicated in cardiovascular disorders? These would be the cardiac channelopathies and some of the cardiomyopathies such as ACM or ARVC, for example.
And the second thing that we were wanting to look at was can we use this to search for sort of novel candidate genes for Sudden Infant Death Syndrome susceptibility? And so we took that aim and really the main result was to show that about 14% of our SIDS cases had what we term potentially informative variants. And those are going to be variants that were within sort of the major channelopathy genes that are implicated in Long QT syndrome or CPVT as well as loss of function variants within the 90 ICC genes that we had examined.
Using the ACMG guidelines for determining the pathogenicity of variants, about 4.3% of our SIDS cases hosted an ACMG guideline predicated likely pathogenic to pathogenic variant. And most of those variants represent either a frame shift or splice site error variance really in minor cardiomyopathy genes and channelopathy genes. So there's still a lot of work that needs to be done in terms of looking at specifically missense variance within channel genes and that sort of thing, and really kind of functionally characterizing those to determine whether or not they truly are pathogenic or if they should remain variants of uncertain significance.
Andrew : And so you took a very complex disease like SIDS with probably a number of differens ideologies and found a pretty good percentage have suspicious variants, that 14% or so, and then 4% had variants that were so suspicious they would meet American College of Medical Genetics guidelines for being a possible or likely pathologic variant. Where do you think this study lies in sort of the continuum of identifying the genetic ideology of SIDS, and what do you think these findings sort of add to that overall picture?
Dave: Well I think these findings in general really just kind of show the complexity of SIDS. Whether or not SIDS is really truly genetic or not, or perhaps it just, if it's not monogenic, perhaps it's polygenic, and so those are some things that we should be considering and looking at. Now some of those questions might be able to be answer through our whole-exome sequencing data set that we have, and I think those are really going to be kind of the next phases.
We can also take and do some pathway analyses of the exome sequencing data, for example, and see our variance kind of lining up on certain pathways that may contribute to certain pathologies that could contribute to SIDS.
Andrew : And in your study, you had a few genes where the number of variants that were found in SIDS cases were higher than in your controls. Can you speak some more about what those genes may tell you in the context of pathway analysis for SIDS?
Dave: Yes. So there was ... There were not genes that came out with sort of a genome-wide significance level. But there were at least 400 genes that had a p-value of 0.05 over representation in SIDS versus our ethnic match controls and 17 of those genes have a p-value of 0.005 and we're really kind of focused on some of those that have a little bit higher p-value for us to assess. A few of those genes may represent biologically plausible candidate genes for SIDS and we were kind of actually going through and considering which ones we'd like to follow up on in terms of function. Some of these genes do play a role in, say, cardiorespiratory system and function of the heart as well as in the brain.
Andrew : So then given all these findings, and the fact that you may have some candidate genes and candidate pathways that might be interesting to look at further, what are the next steps that you think would help this project move forward, and what do you think the field of Sudden Infant Death Syndrome and Sudden Unexplained Death Syndrome needs to kind of move forward?
Dave: Well I think from a genetic standpoint, the study that we just complete was really on a large set of unrelated infants that had died suddenly. We did not have access to parental DNA and so moving forward in terms of the genetics, I think incorporating sort of a trio analysis I think would get at the question of sort of [inaudible 00:42:01] variance for example. The other things, in terms of genetic standpoint is perhaps looking at different genetic mechanisms. Whether these are copy number variance that may be missed by exome sequencing, perhaps some of the SIDS could be due to epigenetic abnormalities or even small chromosomal abnormalities that perhaps may not be detected on certain arrays on there being used. So I think going forward, kind of taking those approaches to look for sort of unique genetic variation.
Andrew : Well Dave, thank you so much for taking the time to speak with me and congratulations on a great project.
Dave: All right, great, thank you.
Jane Ferguson: Thanks to Andrew for highlighting the interesting precision medicine research presented at HRS and thanks to you all for listening. We'll be back with more next month.