 Okay, so I don't have actionable in the title at least I'm gonna say it a lot So I'm gonna talk about the emerge consortium and in particular our experience with trying to determine what our Returnable variants so So the emerge network is a consortium of Institutions through a U01 mechanism funded by NHGRI where we took GWAS data 660k or million chips and phenotypes from electronic medical record and Through the consortium we all share genotypes and then we could get our phenotypes from the other sites So there are about 18,000 people in what we're now calling emerge one emerge was renewed this year to emerge two which added Geisinger and Mount Sinai to the existing sites and that the number of subjects is Increasing all the time and it's turned out to be a very very productive consortium Not just because of the data that we can access but also because of this interesting blend of people where we have You know the informatics people we have the statistical people we have the molecular people we have the ethicists We have the clinicians and we've all been able to really interact and learn from each other in this committee is a good example of that so the Consortium formed a consent and community consultation workgroup and that workgroup in turn Charged a return of results oversight committee to talk about what results were potentially Returnable so just to say that there's a lot of people in the original return of results workgroup And then the oversight committee actually added more people and what's interesting about the people who are here Is that there is a blend of clinical people and I have the clinical trains really hard to shoot it Anyway have the clinical training in the Right column But there's also the informatics people Molecular people statistical people And so it was an interesting blend though the one Disadvantage of that is that we all had to educate each other about our perspective Which took more time than I think if a bunch of docs had sat down But it but it did broaden out the discussion considerably and so I found it very valuable personally So these were a GWAS study. So it's a different kind of results So that we're looking mostly at common variants. So most of them aren't going to be high penetrance things that simplified our job There were however some incidental findings that we expected to find and have to deal with and then we expect future Issues when we're looking at copy number variation. We've actually already seen some deletions that are associated with bone marrow discrages and then We are now in emerge to employing a pharmacogenic array of many many many genes that will give us a lot more action ability So Okay, so these were the players in emerge one group health Marshfield Mayo and Northwestern and There there are about 18,000 people in emerge one. These are the new updated totals, which are going up But there's some variability among these groups. So group health for example that meet median age of the cohort is 74 For Marshfield, it's 48 There's a lot of variability every cohort had about 40% men which is typical except for Mayo, which had a peripheral vascular disease study and they had about 60% men most of the cohorts were very white with at least some representation of Non-caucasians at Northwestern and Vanderbilt. So not everybody was exactly the same We use the guidelines from the fabsitz paper on Reporting research results. Basically. It is an action ability standard of if there there is medical utility to the information It should go back to the patient I've highlighted in blue here that for the members of our committee were on that paper So it's no surprise that we were that we liked it And this paper I want to give credit to the original bookman paper Which set out the standards and and the fabsitz paper was really an update for newer kinds of data So now I'll give credit to Chris O'Donnell whose group nicely put together a beautiful paper on What are the variants on a GWAS chip that are used by clinical labs? And that saved us a ton of time and I highly recommend taking a look at that if you're interested in what's actionable on a GWAS chip This is anything that's used by a clinical lab So it's not necessarily a standard of what's medically relevant because some of these things here are autosomal recessive high penetrance early in life And so the person would know we decided to follow up on factor 5 Leiden and hemochromatosis We decided not to follow up on hereditary pancreatitis Where there is a variant on the chip that has an odds ratio around four or five for pancreatitis So The we were given a specific charge by the Ethics committee which was to look at what would change medical care We were asked to not look at something that impacts reproductive status i.e. we were told not to consider variant In the carrier state for a recessive disorder and then Very importantly, we're an oversight committee and all IRB decisions are local So we are advisory, but we are not telling people what to do And we tried to be very careful in our language to make say it's potential everything was potentially Returnable we didn't say anything what had to be returned everything's potentially returnable so we Struggled with a criteria for defining what is clinically actionable what actually gives direct benefit to the patient Medically not just information that the patient might like to have but actually would change medical care We are avoiding the clear issues for the purposes of this talk. None of these results were done in clear labs and Then we were charged with talking about how results might be returned and and probably most importantly Documenting what the results were that were found so it was interesting that ethics people didn't realize of course the GWAS people did that the first thing you find our chromosome Anomalies turn a syndrome a client-felter It's the very first thing you get because it jumps out on the QC when your genders don't match what they're supposed to match Or when you have extra X's in the presence of a Y So those were the first things that we dealt with Again, we're being very specific to the data we had and then the hemochromatosis and the factor 5 line Factor 5 line as it happens that variant is only present on the 1 million chip And that was run in a small minority of our samples because those were directed at non-crocation samples All right, so we went disease by disease And I can say one thing about how to shorten the process Which is that the more people that are at the table or on the call the longer the discussion is going to go and The more diverse the group the longer the discussion is going to go so Ultimately though, we decided that client-felter disease would be potentially returnable Because there are there is health screening that's done in client-felter's patients to avoid important diseases in particular testosterone is very useful in client-felters Notably in clinical care you come across incidental client-felters when you do bone marrow's and they're just Universally returned like nobody thinks twice about it if you come across a client-felter's you just give it back to them The patients are very favorable and their general response is well that explains a few things, you know So out of the 10 client-felters we found Four of them were known in the MR So it was really an advantage for us to be able to go back to the electronic medical record and say do these people know About this we don't have to return it if they already know about it And that's an advantage every study won't have and every IRB at every site allowed us to do that Happily for us the one mosaic client-felter's was among the known cases So we didn't have to deal with Separately, you know, if he's a mosaic does he need to know is that as actionable as and and I might add You can be a mosaic because you lost a Mua or mosaic because you lost a chromosome along the way the question is just where did you lose it? Did you lose it in your bone marrow and so you have a full client-felter's phenotype or did you lose it? You know in your whole body is not Is not xxy Of the six that were potentially returnable ie. They were client-felters, but we were Unable to find it in the electronic medical record because of course some of those people may have a diagnosis that just didn't find its way Into the medical record in effect one of the ones were calling known actually had turners in his medical record But we're we're sure they knew that it was client-felter's and that turners since it was a male but Of this we think and so it just says something about the medical records and also the doctors of those six One was already deceased and so that was you know not returnable one had documented genital urinary Abnormalities and was on testosterone and so even though they didn't have client-felter's They were doing an appropriate treatment The remaining four were all at Mayo and there was a local decision at Mayo to not return client-felter's Which we could talk about but I don't want to take a lot of time on because it's it was a long protracted discussion It never went to the IRB at Mayo though. It was Within the study group So when we turn to Turner's syndrome again There was a general feeling that most turners know their turners But that some turners could escape turnes is a little more complicated for a couple reasons But there are good reasons to know if you have Turner syndrome If you're young knowing about the infertility, of course is helpful if you're older. There are some complications Turners are treated with estrogen and they do have more medical screening interestingly the two patients that clearly had turners an XO and an X XQ minus so missing the long arm of one of the X's those were known in the EMR. We had eight other mosaic turners XO XX And and those can be acquired and many of these patients actually were quite older They four of them were from the group health study and they were in their 70s And so there is well documented loss of the X chromosome in the bone marrow So that can be just an acquired change that doesn't represent Turner syndrome at all Out of those eight three were deceased There are five of those eight had children documented in EMR only in one case though did actual births and pregnancies Appear in the EMR. So you don't know if children are adopted One of those eight patients though was documented to be infertile in the EMR And and none of these have been returned I'll come back to that I'm so factor five Leiden There was a sort of a sophisticated nuanced that homozygotes should be returned and heterozygotes not because the his the risk of a dbt or PE is so much higher in homozygotes and The interventions for heterozygotes are slight But there was some discussion and there was also a discussion of is there some age after which you would not tell the Homozygotes anymore because they would have presented by now Hemochromatosis Was the longest discussion so it's autosomal recessive It has a penetrance of 10% in males only 1% in females. It causes serious complications in death And the prevention is really actually quite simple of using phlebotomy to keep off the excess iron And screening by ferritin level very simple. There is not population Recommendation for population screening in hemochromatosis But the group felt very strongly that what we're doing isn't population screening if you know the genotype Is there a different level of burden to tell the person? So ultimately there was a mixed opinion on this one But the general sense was to return in males and not to return in females and of course that's what we had a Little debate about because we had some members of the group who are uncomfortable of with treating males and females differently But in this case the risks are truly different And I should say in hemochromatosis, of course if someone comes into clinic and says my brother has hemochromatosis We always test them. We don't you know, we don't go like well, maybe you don't need to know we just do it So coming back to what all the findings were I talked about The Turner's and the Klein filter for factor 5 Leiden three of the groups didn't have the one million chip at all and for the two groups that did have the one million chip They elected to not review the data. In fact, the the allele for factor 5 Leiden is quite rare in African-american populations and so the possibility to find in homozygote was quite low But nobody looked for For hemochromatosis group health is the only site that did not look and that's because the patients were so elderly that it Decided that it wouldn't be clinically relevant for them at that point for Marshfield they found 14 people who were homozygotes for the The major site six of those were male They found five of those had a diagnosis in the emr, which is an interesting proportion And in the rest there were no symptoms in the emr. They actually reviewed for heart disease other kinds of liver disease and Because they could not find symptoms. They decided not to return at Mayo They had about they had 14 5 also in the emr And they decided that they they have not decided about return yet They're still in the process at Northwestern They found six two in the emr. They felt that they could not return. This is the new gene cohort They don't they're not consented for return and they also don't really have contact with the subjects and finally for Vanderbilt They just have no ability to get back to the original subjects. It's a de-identified data set I skipped over for client filters the one returnable client filters that we had at group health We decided we would return it and the patient turned out to be deceased. So that saved that So the conclusions first of all Malia Fullerton Wrote up our experience and that paper was submitted to genetics and medicine. It came back for Very minor revision. So hopefully that will be in press in genetics and medicine and there's a lot more to the story That will be in that paper certainly we found that Incidental findings are going to remain an issue in research studies But I think that applies to a clinical genome when you're when you do an exome or a genome looking for a cancer gene And you find other things you really need to be prepared for how you're going to handle that information clinically actionable is a highly debatable term and and Importantly and I think we didn't really appreciate this going in it depends on the clinical context a lot So it depends if you're male it depends if you're female it depends on what age you are it will depend on what race you are and Many other factors. Do you do you still want to have children? If you don't want to have children then meaning to know if you're infertile is not a priority So I think that that's something that's going to be very hard to operationalize In emerged to we've divided this work into two committees. There's the CERC committee, which is community education and a couple other letters I should have looked up And the other is the actionable variant committee and so, you know, maybe you can get them to remain I didn't name it But I'm a co-chair with if the car cool-o and for the actionable variant committee We're going to be looking at risk source particularly for cardiovascular disease and then because we'll be getting this large amount of pharmacogenomic Data we'll be looking at those variants and those are our next steps. Thank you A couple of quick questions or comments Gail I may miss this at the beginning, but what was the consent status? and all the different sites about Actionability, yeah that you Barely missed it. It was on one of the slides But the fact was that most of the cohorts were not consented for return two of the cohorts were consented to return But for the other four sites each IRB indicated that if there was some Compelling reason that they would consider that and one of the things we had along they would consider actually returning results Despite what the consent form said We had a very long discussion about how you ask people if they want results without telling them They have results And so in in group health they actually went back and they just reconsented the whole cohort Well everyone who they were still able to get Went because that cohort started out as an Alzheimer's disease discovery cohort You know 15 years ago and nobody really thought that they'd find anything interesting like genetically in that cohort so I've reconsented several cohorts related to other studies because we didn't anticipate at the time of the original consent and and At UW that's our IRB's preferred process that you reconsent the whole cohort for return of potentially interesting Results with this giant disclaimer that you may not ever be giving them any but you just want to know And that way you're not stuck with well here are four people I'd like to tell how do I ask them if they want the information without tipping off that I have something useful for them So it's so we go with just try and reconsent everybody. It's not as efficient necessarily, but it it allows them to really make a choice Oh Yeah, so we're talking about hemochromatosis and and the fact that there is no recommendation for Population screening for hemochromatosis that the cost benefit is felt to not be worthwhile for screening What's not screening about this is we actually have the genetic data So that's the distinction I'm making and we're not spending any money to get the data We may spend money to do something with the data, but the data is sitting there The other comment that I'd like to add to that is that Bookman and fab sits both took the position and we took two that no one's even if you have the data You're not required to look at it You're not required to go digging through your data to find everything. That's potentially clinically relevant But if you if you are stumbling across it like the karyotypes that you can't ignore you find them Then then that's where your burden starts to to decide if you're going to return that information Does that answer your question? Okay One more question Terry Thanks, you made a very good point about Really that what's actionable and not depends on the clinical context, but then you said something about that it might depend on Resubdicity did I hear you correctly? And if so, could you explain well penetrance? Penetrance may be race specific. So as we move to variance that risk scores genetic risk scores, for example, those may become race specific and I you know, I'm particularly concerned about those Certainly the then the risk scores are primarily built around Caucasians because that's where our data is We don't have very much information. So so certainly there's no reason to believe that if You know most variants do translate across if they increase risk in this race They increase risk, but the amount that they increased risk by the time you're talking about odds ratio may not be the same at All and then the spectrum of variance is going to be different across Races so certainly for Turner syndrome. I'm not worried about race But for a risk score, it's I'm more concerned. Yeah I think we need to be real careful of that and I agree with you that the data may not be there But what what tends to happen that as people say well, we only have data in European ancestry people So we can't say anything about anyone else and I would disagree Right, I I totally agree with that statement. I and in fact, I've heard I was at a Seminole recently where someone said well, you know, none of this helps people who aren't white and I vehemently Disagreed with that statement. It's many of these things will translate But you you know, you have to decide what your cut-offs are if you're looking at a 10 or 15 percent increased risk That may be a five or even a 20 percent increased risk across a different background So we can't use it and even though it's rare if it's there, right for high penetrance alleles I don't think that that race is an issue. I think it's really for these low penetrance things. Yeah, I'll Put my stamp on that. Thanks for bringing that point up though. I didn't mean to muddle muddle that