 I changed my title so and I was asked to be provocative so like and for the people in a merge who don't know me it's obvious why I was asked but okay so and I'm taking my clinician hat here and also my ClinGen hat because I spend now we could have just called this a ClinGen meeting and you could have gotten three for the price of two a number of the members of the steering committee are here and I've just really become more and more convinced how difficult it actually is to classify variants and report them so I'm quite troubled by the potential difficulty in effectively classifying the variants on the list for reporting is incidental findings some have obviously extremely clear loss of function phenotypes but missense alleles as many of you in the lab know are much more difficult to classify the ACMG classification rules may approve our ability to classify variants but as you heard from Gail that still is not easy and takes a lot of work and in particular I've had the clinical experience and personal experience actually of having reclassification of variants from pathogenic to benign or V us and that is troubling and I think it's particularly troubling in the incidental finding setting our goal here is not to do harm and I did have the hypothesis you'll see I was wrong about this that some genes may be to note new to know the spectrum well and some genes have a very broad phenotype and this is definitely come up in our own reporting of incidental findings the table lists one phenotype but in fact when you go into OMIM there's five phenotypes and so what are you really telling the patient there at risk for and I have to say my complete bias start going into this was that incidental finding reporting was easier for cancer genes and cardiovascular genes and just to highlight some issues about reclassifying variants so this is a beautiful paper that came out in the New England Journal about misdiagnosis of cardiovascular pathogenic alleles in African Americans it was based on this genetics and medicine classification of variants of very large group of patients who not incidental findings for cardiovascular disease and I think at the previous meeting we talked to this about this but basically there were a number of patients with African American African American patients where the allele in fact once we had more population data is way too common to actually be responsible for the phenotype the patient had so I just I'm a clinician so I just wanted to give you three case examples and one happened yesterday so thanks Kate Nathanson for this so in 2004 when I was still doing adult genetics I saw an extremely nervous 29-year-old postdoc who had a single retinal hemangioma I sent off testing at what was considered the top lab and that's not the Baylor lab though the lab that's been used by the VHL NIH program for many years and I found I keep paper copies so I found the record and it was reported to have this intronic variant the report is completely straightforward in that this is pathogenic so there was no wiggle room in the original report the pathogenic this variants been shown to have a messenger RNA defect and that was all stated on the report so I told this unbelievably nervous person that they did have the HL and I have to say this had huge impact on him so this is not a minor diagnosis and he but he went on with his career and moved to Philadelphia and is now a faculty member and so 12 years later he came in to see Kate and she looked up his allele in ClinVar being a very conscientious clinician and now multiple labs called it benign and in fact it's found in something like 4.5% of Han Chinese and he was Chinese and he in fact has not developed any other signs of VHL so he probably did just have an isolated single retinal hemangioma and has spent you know 12 or 15 years with this diagnosis in our own CSER project we've had two incidental findings that have been reclassified both before the ACMG guidelines one very similar to the New England Journal paper a DSP allele that had been reported to be pathogenic which now based on exact data is certainly way too common and then we've had both symptomatic patient and an incidental finding with the same LDLR missense allele which was reclassified as a VUS now using the ACMG classification scheme although frankly I've spent a lot of time delving into this variant I still can't really tell if it's pathogenic or uncertain I guess that's why it's uncertain but this is really more a question of the reclassification based on better rules so what did I do for this talk I took the ACMG list I went through every one of them on OMIM which was quite interesting to do actually and I focused on the cancer and cardiovascular genes so one of my hypotheses was that some of these genes are too new so I looked at the date that the gene was first published and I just did before anything before 2000 or the date after 2000 I looked at in the literature I tried to look at what are the majority of the alleles I tried to look at other features I looked at segregation and I also looked at whether there was an expert panel in ClinVar so this is my main data okay so there's about comparable number of cancer and cardiovascular genes on the ACMG 59 it turns out the vast majority of them were identified before 2000s no offense to the human genome project but almost all of them were identified in that in the 1990s the cardiovascular genes do have more new genes but you'll see they're really not that new there are actually only six genes on the list that have an expert panel in ClinVar they're all cancer genes the primary the primary mutational spectrum being loss of function is much much more common in the cancer gene so I was right about that there's a small number of genes that have almost equal loss of function and mis-sense and then the genes where the primary mechanism of diseases mis-sense is much more common in cardiovascular and I think I think that is a potential problem with regard to age so these are the actual dates of all the genes that were discovered after 2000 you can see this line is 10 years ago so in fact the very few new genes on the list so it's not really an issue of newness we've had at least 10 years to learn about the spectrum of mutation in most of these okay so what are the other complications so they're most of them are dominant there is one x-linked fibrase and I just looked on the table again and it reckons only reporting in males although we all well those of us who do our clinicians there is clearly a female phenotype so we could argue about that there's one autosomal recessive UIH so do you report singles I think these are all things at labs even if the guidelines are thinking about the labs are going to struggle with each time what was really striking to me is how in OMIM is how many of these genes actually have both dominant and severe recessive phenotypes so again you've got a kind do when you're disclosing this do you have to explain walk okay also now you're a carrier for this rare severe phenotype there are couple that the one of the long QTs and I'm wrong about this slide is di-genic but it is also monogenic so but do you have to look for the other allele because that again is a more severe condition and a few of them really only have very few and this was just mentioned very few allele so TM 43 and SEH AF2 if you go into ClinVar each one of them only has one pathogenic allele and one has I think less than five pathogenic alleles the bigger issue that I'm just trying to get to is there's an incredibly wide range of quality of information so BRCA ones on that list okay there's thousands of documented alleles thousands of papers frankly and there's an expert panel and then you have genes that have only a few well documented alleles like Newt YH and a variety of others and then you have some extremely rare conditions you also have some other messy things like PMS to you could be calling a pseudo gene and you actually have to do a fair amount of work to know that it's not a mutation in a pseudo gene and then there are a number of these that have substantial genetic modifiers so the phenotype again is very different depending on the presence of a modifier so I have to be honest that's part of the reason I asked to give this talk I'm not clear we're doing an overall good by reporting incidental findings but I think even though we've known some of these genes for decades reporting variants in many of these genes continues to be quite difficult and I would suggest that perhaps we should consider really substantially simplifying the current recommendations I would consider and I I didn't make a suggestion to the to the committee so considering dropping even a few more of these very rare or recessing conditions unless they're really common but my main recommendation is I actually think we need a much clearer definition for the labs of exactly what should be reported for each of these genes in the setting of incidental findings so labs were not asked they're not being asked to look at the scene because of the patient's phenotype I think we need a very strict loss of function definition I know there's some BRCA1 variants that look all the world like expected pathogenic which the committee continues to use that terminology that in fact are not because there's a splice product that's that present in 10% of genes in 10% of transcripts that splices out those exons so it doesn't actually matter that you have a nonsense mutation in either of those two exons so even saying expected pathogenic can get you into trouble I think we actually have to have for each gene a very strict definition of what loss of function means I personally would recommend that for the genes that have missense mutations that an expert panel has a list of missense mutations and you don't report the others you don't have to sit there and classify each of the other ones and that if there any other specific complications like a common modifier that there be a clear rule you look for it or you don't and then I think for the clinicians receiving the results I think we need a fact sheet from the ACMG or another committee for each of these disorders that say if found in the setting of an incidental finding these are things to consider similar to what we have for newborn screening because I think right now it's actually I've disclosed some of these cardiovascular ones I think it's actually quite difficult and I'm sure the people that don't do cancer feel the same thing about the cancer ones thanks I actually in a dovetails with Wendy's presentation also I actually find it paradoxically really gratifying and really disappointing that 55 of the 56 genes on the original recommendations are still in the recommendations and that was always felt to be a first approximation recognizing many of the issues that you present but I think that the presentation you make that it's you know it's kind of it's hard to do these things I my general impression in medicine is that it's generally difficult to keep people from dying and that if we get up and complain to the world that golly gee whiz will occurs it's it's hard to do this I don't know that that's going to generate a lot of sympathy or really help us do what we need to do I'm okay with things being a little bit hard but I think the definition and the problem needs to be reframed and what do you want the positive predictive value of a secondary finding to be and what do you want this sensitivity to be right and I would love it to be a hundred percent sensitive and a hundred percent positive predictive value as would you and we don't get that and I can drive PPV close to a hundred and almost completely wipe out sensitivity and vice versa what do you want it to be I mean if because if you could frame it that way then we can tune these things to do whatever you in the community want it I'm gonna push back a little bit so I agree that we should certainly do things that are hard I think the issue of doing things that are really hard in the setting of incidental findings is quite different okay so I would prefer and many of you who are in Caesar know I've argued for four years that we spent too much time talking about incidental findings it up here here I am doing it I would prefer the lab put their effort into making the best diagnosis for my patient they can okay that's number one number two is I think it's not simply sensitivity and specificity what I'm trying to say is it's not as simple as saying here's a mis-sense allele that we know causes whatever longqt it's also the fact that the clinician's gonna look up that gene and discover oh well some of them have sudden death and some of them have forgot and some of them have whatever so I think if we're gonna continue to do this we need to have a much clearer and I frankly we've worked with oncologists who are not you know I mean they refer the patient to a cardiologist but I think we have to really arm the clinician which much better information about if this is found in an incidental finding this is what you need to know I do think the specificity I mean without sitting down and really doing numbers I think the specificity should be significantly higher than it is I just think that the genetics is much messier than we realize and you quickly get into issues about modifiers and things like that in an incidental finding which is not necessarily where most clinicians want to be spending their time but I actually think that gets back to the 55 still there I think the are all the things you say are true and I don't want to bring my hands about it or moan and wail I just actually want to just get to where we want to get and I think it's actually true for everything we're going to try and do in predictive medicine and I think that's a great reason for emerge and Caesar to be together because this is our challenge writ large and if it's hard with 59 genes it's gonna be impossible with 5,000 and so we have to just these are pretty straightforward questions how predictive do you want it to be and some of the genes and I think the some of the cardiomyopathy genes are candidates for being dropped for some of these reasons and I agree with you that the cancer ones I we're finding are generally less of a problem for us and I would just comment one thing the gale emailed me is it's not simply so I I just want to simple it so for example the genes that only have two pathogenic alleles if you want to leave that gene on the list then just put that gene that allele and stop having labs feel like they have to at least look at every other variant in that gene which is what they currently feel like they have to do so I think specificity of the list could also be helpful and we're actually doing yeah yeah