 Well, I very much appreciate being invited to talk today. It's been very interesting. I think I'm here. Okay. Double. Okay. So, first, this is just for Dan, because he was asking how old Arno was. This is Arno actually in 1962, but Arno is 89 now, I think, plus or minus a year. So we have a long history of medical genetics at UW. All right. So what are the debates? What are the areas of concern that we have about consent and recontact and return and results? So there are different consent models, different access models. I'm not going to spend a lot of time on whether we should have narrow or broad access to the data, because I think it's pretty clear we all want broad access to data. Consent models can be full consent where the patient has an informed consent document that they sign versus an opt-out where they can withdraw from the study, but they don't have to sign consent or just anonymize data where you could potentially need neither of these. So a lot of that falls on recontact, and I'm going to come back to that. What is appropriate consent language is an area of concern, and broad consent versus narrow consent. So this has been a problem. IRBs want to push us to, well, we're just going to limit the use of this data to the aims of this specific study. But what we really want is a broad consent so we can look at lots of different phenotypes across the cohort. Recontact and return a result, what's the value of recontacting people? Is it worth the cost? What is an actionable result? If you're going to return results, what kind of results are you going to return? Can you ethically not return actionable, medically actionable data? Do you need to look for it or just wait until you stumble across it? Some people want all data, should they get it? Some subjects may not want their actionable data. Can they refuse it? And then what kind of results are you going to give adults and children? So I'm going to touch on some of these things. All right. So consent models will inform consent, anonymize versus opt out. And there are a couple negatives to the anonymize and opt out, but the primary one is recontact. So I'm going to come back to why I think recontact is important. Of course you can't return results. Some people may see that as a plus, actually. But you also can't incorporate genomics into their care. And so our next question as we move toward genomic medicine is, what is the impact of putting these genomic results that we have to use? And so if we can't do that, we really have limited the utility of our cohort. All right. So this is just to talk about why recontact is important. And the University of Washington CSERP project is a randomized control trial of adding exomes to usual care in patients with colorectal cancer and palposis. And the details of this aren't that important. You know, half usual care, half whole exomes. But what is relevant, wrong button, that in actually exome number two, we identified a subject with an HDMD class, a pathogenic mutation in an arrhythmia gene. Now we're actually not accepting HMD's view of that. And we're going through the literature ourselves because it's not an incredibly well-established mutation. But nonetheless, this is the sort of thing that we would want to return to a subject if we know about it. And that in fact, if it turns out to be a mutation that we feel is pathologic and highly penetrant, we would feel very uncomfortable not returning that kind of information to a subject. This is a different recontact issue. This is what we call the buff gene. And this is just to sex up our meeting here. This is our pro band. And she actually came into clinic for something completely different. But she said to the geneticist who she saw, you know, my family has this interesting thing. We all are very muscular and we don't work out. And I think we'd all like this gene, right? So this is the patient. These are her identical twins. This is her like 11-year-old. And this age is wrong, by the way. And this is the baby. And the baby's doing a hang, just so you can't appreciate the picture. And so we thought, well, this is an interesting phenotype. And we did the usual chip the whole family exome, a couple of people. And we got it down to a single variant. And it's not a novel variant. And so we'd like to find more people with that variant. And we went to the exome variant server to find out that what the frequency of the variant is. So it's less than 1%, but it's more than 0.05%. So someone in this room could have the buff gene or not. But as much as the exome variant server is fabulous for pulling up these allele frequencies, and even our genetic counselors use this tool all the time, you don't have the phenotypes at all, much less the phenotypes that the study never measured. And Debbie Nickerson has said, the number one question they get about this server is, can I find out what else this patient has with this genotype that I'm interested in, or these mutations in my genome interest? And the answer right now is no. And so it's hard to say that in the medical record, a doctor would note, boy, this patient looks unusually great. It's just not something we put in the medical record. So this is something that really does go back to the ability to recontact and collect more information on a subject that's in a database. All right, so informed consent, the study view, as opposed to the participant view, we really want to understand the subject's needs in a study. And I think that we actually don't do very well on that. I'll come back to that. We need to explain the data being generated and the kind of studies it might be used for in an accessible way. We do want to have broad study goals and broad data sharing. We do want to recontact. Many of us want to return the genetic results or feel ethically driven to do so. And those fall out to the primary result, the reason that they were chosen for the cohort, if that's relevant, if it's population-based, then it may not be. The incidental results, so that would be like the finding of an arrhythmia gene in my colon cancer patient. I am just taking a moment out to say, I hate the term secondary for this. I think that if your primary reason you're getting into test is that I am a woman and I might have hemochromatosis in my family, but I find I have a breast cancer mutation that is not secondary to me. So I like incidental there. And then the other thing that we're learning as we go on is that we need to address survivorship, particularly in our cancer patients, what happens to their genomic information if they die is not something we've routinely asked people, but a question that's coming up more and more as we do this. What's the subject's view? Well, the benefits to the subject completely depend on return. If there's no return of result, then there's no benefit to the subject other than the abstract or helping the world benefit that the IRB won't even let me list. So benefits can be prevention or early diagnosis. So again, a BRCA mutation would be a canonical thing for that where you could have screening or early more frequent MRIs or have your ovaries removed. Pharmacogenetics to maximize your drug efficacy and choices. Those are real benefits to the subject. We'll see if your risk of heart disease is increased by two is a benefit or not. I'm not clear on that yet. What are the risks? These are the risks that are independent of whether or not they get the results back. There's a privacy risk. And a big concern to most of us should be this possible redefinition of the common rule where genetic data would be defined as identifiable. I think that totally upends everything we're trying to do with data sharing. And I hope that all of you got into the comment period on that potential change in the rule, which I have to say I don't even think makes sense, much less is gonna be a real problem for this kind of a big cohort. The other risk which is return independent is a false sense of security. And so what we found by surveying people who are even told that they would not get genetic information back in their consent form is that they think that we would tell them if there was something wrong with their genome. And so making people really understand that that is not one of the things that's happening in the study is critical, even when it says right in the document. And I think we've all experienced this with women who have amniocentesis, back in the days when women had amniocentesis and they'd come out and they'd say my baby's perfect and you'd go like, eww, you know. So people really need to understand what the limitations are of what's being done. Risks that are dependent on return are of course the social and emotional risks of getting bad news. Even though the general law protects medical insurance, there are no protections for life and long-term care insurance. And so subjects need to understand that. Not that we have any evidence that insurers are using this information right now to define these insurance rates, but they legally can. They need to understand incomplete penetrance. We do not return incidental findings that are incidental. I mean that are VUSs. However, VUS misinterpretation in clinical genetics is a huge problem and we've seen many cases of patients having mastectomies or their thyroid removed. When they had a VUS, when the geneticist said this is a VUS, it's probably going to turn out to be a benign and we're just gonna have to follow it over time. And the surgeon says, breast cancer, breast cancer change, remove breasts. And so that's real harm that comes out of these tests. And then finally, what's the cost of follow-up? So if I tell this patient, G, you may have this arrhythmia gene. They may get an echo. They may get a 24 hour or two week averaged halter. They may get an implantable defibrillator. And those are real costs to the medical system. And so you really have to be convinced that the penetrance is there to warrant those kinds of costs. Okay, so this is a study that we did at Group Health in Seattle called the Glad You Ask Study. And we were required to re-consent the subjects at Group Health for the Emerge Study for broad data sharing. And so we interviewed the first 400 by mail and we asked them, did they think it was important that we were asking for their permission? And 90% of them said absolutely was important. The vast majority of the subjects re-consented but they felt it was very important that we specifically asked them if we could share their data broadly. We also found I should say that older patients were much more happy to share data broadly than younger people for a variety of reasons that include you pretty much know what your genome means by the time you're 80. So the acceptability of alternatives for kinds of consent models opt out versus notification that, you know, we, by the way, we shared it just so you know or no consent at all had very high rates of being found either completely or somewhat unacceptable, 40% unacceptable for opt out and 70% ish for the other models. Okay, what do people want? There is strong public support for return of results. I just fact-checked part of Gina Claude's New York Times article which will be coming out soon discussing return of genetic results for research subjects. Many people, as I said before, expect to be getting pathologic results returned to them even if they're consented that they won't get results back which is interesting. But my problem with this and I do this kind of research so I'm pointing the finger at myself is that if you pull in a focus group these are highly motivated people. If you get eight to 12 people in to talk to you for two hours in the evening about what they want out of research these are not the people who don't have enough money, they have to feed themselves. It's a very different selected group of highly educated people. So what do people want that are less engaged and we don't have that answer but I'm gonna try and give you a best case and maybe a worst case for that. So best case, this is a family study that we are conducting at University of Washington. Everyone in the study had been contacted within the last 10 years which is actually recent for us and by mail we sent them a consent form saying three things. One, do you want broad consent? Two, can we use your data for everything? And three, do you want our newsletter? And only one person said no. There were 21 people we couldn't find so 91% signed and said we want everything except we really don't want your newsletter which was interesting. Because a lot of people still said they take it but the ethicists are always telling us every study should have a newsletter but patients actually don't care. They have to be willing to say they don't care too which is interesting. Okay, so this is the worst case scenario and we're six months into trying to re-consent 1700 people from a VA study. It's an old study up to 20 years ago that we saw these people and it's DBGAP posting and actually cell lines from existing samples and so far I think these are the most interesting columns. 5% of people who we got the survey out of the whole study so we really only captured maybe 60% of people so far. Only 5% say no, they don't want to be posted on DBGAP. 7% were not interested per a phone call. A third of them we either haven't found yet or it's a lot of work to find people who haven't seen in 20 years actually but 50% of them and by far the majority of the people we did find are saying it's fine but the one exception is our African Americans who when they return the paperwork it's about at the same rate but over the phone we were calling them first to say you're gonna get this in the mail and we were asking them over the phone, did you want this? And disproportionately said no, I wouldn't sign that so don't send it to me. I will say we'd have a much better success rate if we were allowed to do a telephone consent, a recorded phone consent versus a lot of the problems getting the paperwork back from people who said on the phone it'd be fine. So I'm gonna skip over return of results to kids and adults but it's a problem. We can come back to them and ask me a question. Data access broad, problem minority populations we all know, Eric asked for two questions so here are my two just to be overly specific. I wanna know who needs colonoscopy screening at 40, 50 and 60 years old? Why are we screening everybody at age 50 and I might add people hate to be screened and they're very non-compliant with this screening. If we could be more targeted we would be more successful and we could have potentially saved money with this group that doesn't need it to later. And then secondly a little self-interest here because I am a physician who was self-diagnosed with celiac at 47 and my doctor thought it was an idiotic idea. There are many diseases like celiac that are underdiagnosed and would save us money and morbidity and can we predict those diseases? And the theme here is changing the medical system behavior and not the patient's behavior. I think it's easier for us to change what we're doing than them to change what they're doing so with that I'll just leave the review slide up. Thank you. Thank you. Discussion? So isn't family history the strongest predictor for who needs a colonoscopy? Well you know we have a universal recommendation that everyone gets colonoscopy at 50 with the modification that people get colonoscopy 10 years before the earliest cancer in the first degree relative but most people don't do that at 50 and many people probably don't need it till 60 and 70. I mean colon cancer is much more common after 60. So my idea is if you can target what the genetic profile is of someone who's gonna have an earlier or a later colon cancer you could refine a screening requirement and I am not limiting that to you know it could be a mammography, it could be other things. Not one size fits all. Right my point was only that family history has always been shown to be the strongest predictor of these outcomes, better than at least so far in the genetic variants. So family history can be part of the model, it doesn't need to be exclusively a genetic model but right now it's one size fits all. Couple of questions and a comment. When you mail those things out did you have a self-addressed stamped envelope to make it easy to respond or not? We did have a self-addressed stamped envelope, we were not allowed to like stick in a $5 bill that was considered coercive by the IRB. But it's been shown to get much higher risk because people actually read whatever you have in the envelope with the $5 bill. And secondly you were saying figure out who's got celiac, do you think using genetic studies we could identify markers of things that you can't measure? So you can't measure complex migraines, you can't measure the symptoms, it's flying by the seat of your pants to diagnose it and treat it. Do you think we could actually have patients who appear to have the phenotype since you can't measure it, it's a complicated phenotype and then you could have a way to measure who's got complex phenotypes based on genetic predictions. Right, so the beauty of the genome or if they could work out that HLA problem is that you can predict a lot of different diseases from the exact same data. But I think people who might not realize they have migraines, we do have patients who have had migraines for 10 years before they finally come in and say, you know I've had these headaches for 10 years. And that would benefit from earlier diagnosis and I'm particularly interested in that category of diseases that people have without realizing that they have a treatable disease and I think it would be very powerful to try and use genetic information to say, maybe at least you should be thinking about this diagnosis in this patient, maybe not give them the diagnosis, but at least say you should be thinking about this diagnosis in this patient based on their genetic profile because doctors have 15 minutes with their patient, they don't get through everything. And you had a comment, was that your comment? Actually, one other question. Well, when we started collecting DNA from CF patients, we were worried and continued to be worried about these incidental findings. And we actually have a part of our consent form where you can opt in for us to use their social security number to track them down if something shows up down the line and we don't have ready access to them. So we did this starting back 10 or 12 years ago. Is this still a viable option if this? I don't believe that my IRB would let me. And in fact, even at the VA where we use social security numbers for years because those are the patient record numbers, we were not even allowed to keep the last four of their social security number. So even in our HIPAA, identify our part of the database, we're not allowed any element of social security number, but that's a university specific ruling. So it may vary. Eric. Gail, I'll come back to your Celiac disease example. I worry a little bit that these individuals are gonna come into these cohort studies or into this program thinking they're healthy. We're going to sequence them and then we're gonna start to diagnose disease. We're gonna medicalize sort of normal variation. They're gonna come into it thinking they're healthy and they're gonna leave diagnosed with a handful of conditions because we use sequence information. Then we brought them in and then we find out they're, this and that's a little bigger or smaller than it, maybe it should be, but yet there's no obvious phenotypic manifestations. Yeah, so I appreciate you giving me the opportunity to clarify that my goal wasn't to make diagnosis in these cohort, but to use this cohort to learn if genetics is powerful for predicting undiagnosed disease. So even to build the model that says, here are the people who have diagnosed Celiac, here's their genetic profile. And then apply that to another population. So I'm not at all saying I wanna diagnose people in the cohort necessarily. I just wanna build a model that might help me diagnose people in the future. It's a totally unrelated topic is, I wonder, since you and your colleagues have expertise on interacting with participants in these studies, do you think we're using social media to our benefit? I mean, young participants in particular and even older participants to my discuss sometimes are very comfortable with Facebook, et cetera. And yet we don't seem to be using it at all. We're still mailing them things and having printed newsletters and phone calls, et cetera. I wonder if we need to expand into social media as we think about these cohort studies and keeping track of people. Well, having had a study advertised by an interview on Larry King Live, which was highly effective for, it was highly effective for getting white participants. And I mean, I think there are advantages to social media, but in a big cohort study, you have to decide who you want and you have to know how you got them. It's different than a Mendelian disease study where the ascertainment is less critical to the analysis. But I think understanding the ascertainment in a population based study of people you got through social media would be very complicated. Rory, I think you had it. With respect to the issues of feedback and the list of risks to participants that you described in terms of feedback, do you think that the benefits of feedback are actually not really for the participant, but actually for the researcher and their institution? And that in a way, there isn't a really honest debate about this, that this is actually more of a legal protection and is probably not to the benefit of participants, at least on average. Well, I think it depends on what you return. And I think it's not for our legal protection because if we said we're not gonna return anything ever and they consented to that, I don't see the legal recourse they'd have but I'm not a lawyer. I think there is a benefit to us because as a clinician, I feel like if something's actionable and highly penetrant, then I cannot sleep at night if I don't tell someone something that could significantly protect their health. I have real qualms about whether we'd be doing them any good by just like handing them their whole genome because there's so much in there that they don't understand and I have people coming to clinic now to have their 23s and me's explained to them, which their insurance company's paying for. And so I'm not sure the overall harm versus good of giving everything back but I think these very selected things that there has to be some real agreement about are very important. But the one thing that you make me wanna underscore is that patients really do need to be able to opt out of getting these results. There are patients who don't want them who won't enroll in your study if they're going to get them and ethically you shouldn't make someone get information they don't want because they're in a study. So I think that opt out of incidental findings is critical. And if patients... Sorry, they just illustrates my point because you were saying that the feedback should be because you wouldn't be able to sleep at night. And so now... No, no, I said that's the benefit to me. Yeah, so now there's gonna be feedback so they can't sleep at night. So I mean, I think that this is my concern that it's for the researchers benefit and for their institutions. It's not necessarily for the benefit of the participant. I think again it goes on what you're gonna return. So if you're going to return BRCA1 and 2 mutations and those patients could die of a cancer that was diagnosed late because you didn't give them that information. Those are the things you wanna return. You know, we argue a lot. We have 14 people in a room arguing about what's actionable and what's not. That's literally our process, which has been really fun. And it's what's the boundary of that? Does that include familial Mediterranean fever? Because even though those people are symptomatic, they clearly are under diagnosed and there's an easy treatment. And our attitude is very much what would you wanna know as a patient? What would the patient wanna know? And we actually allow a menu of choices for the patient. Do you want to know about pharmacogenetic traits? Do you want to know about cancer related traits? Do you wanna know about heart disease related traits so they can pick? And we try and get them to understand what they're consenting to. But it very much is what we think would change their health in a way that would be an improvement. We're not gonna return, you're gonna get Alzheimer's disease cause there's nothing you can do. So one thing is to try to explain to patients as much as possible what it is they're consenting to but still they might consent to it and then get results back that they don't understand. And so is there some way to build in or is it already built in a very best effort to be sure that whatever results go back to patients are put into some context that a patient can understand what a researcher's level of understanding is of what's happening in there and what's serious and what's not and when to blow things out of proportion and when not to because patients will tend to be very scared anyway. Right, so we're doing two things and we have the advantage of being clinicians. And it's a harder problem for people, for research teams that don't have clinicians or clinical geneticists. We do two things. One is we never return incidental findings at the visit where they're getting their colon cancer results because we want them to be able to focus that but we bring them back in person to a clinical, either a genetic counselor or a clinical geneticist to get the incidental findings and then it goes in their electronic medical record and we ask them before, when they come back for that visit and before they get anything we say, you're sure you want them. You have any questions about what we're gonna do right now and they can say no at that point and walk out but if they say yes and it goes into the medical record as a formal diagnosis or genetic test. I think Julie, did you have a comment? Yeah, can I just ask you a practical question about the re-consenting. Are IRB, when you put in a proposal to re-consent participants who are in a study but no longer are, the first question they ever ask is, did you ever tell them that you might be in touch with them in the future? And so I was wondering whether it seemed to me a consequence of that would be that when we're dealing with ongoing studies, if we ever think that we're gonna want to re-contact people in the future, maybe a year, two years, whatever after the end that we should just at least put a sentence in somewhere that said that we may be contacting you at a future time. And I was wondering if you had any experience with this and your experience with the IRB on getting this. Yeah, so we have some local experience with this. I will say when I was a brand new assistant professor and turned in my first IRB, they wanted me to say, I wouldn't re-contact the subjects and I would destroy their samples in 10 years. And I said, no, I'm not doing that. Thank you, but no. And every consent form I've ever had has said I may re-contact you in the future. Even when I had no idea why I would want to but it just seemed like a good idea. I have had colleagues who had to actually go back and re-consent because this was like a standard IRB thing, just destroy your samples in 10 years, they had to actually go back and re-consent to keep the samples after 10 years, which was ludicrous, but they were allowed to go back. But I would absolutely put in every consent form, broad data sharing, re-consent. And the other thing that we put in every consent form is that the genetic, the interpretation of the genetic data now will change over time. So we really need people to understand that. And as we were going through the CSER consent forms and comparing them, there was the one element I saw missing from the other sites. Consent forms is this communication that how I interpret the genome today is gonna be completely different than how I interpret it five years from now. And then the other thing I probably should have said, the question before is that the number of people who really have true incidental findings is going to be extremely small, we expect. It's gonna be probably less than 1% because these diseases just aren't that common. The highly penetrant, scary things. Just very quickly, when you said you put in the, you want people to put in the consent form that the interpretation will change, you must then say to the participant, and that means for you, Dr. What do you tell them that it means for them? What we're telling them is, just because we looked at this set of genes now doesn't mean that you couldn't have a genetic abnormality that we don't find in the genes we looked at because we don't know what most of the variation means right now. And so this test is not a test for all time that you've had your genome looked at. Boiled down, it's a little longer than that. Let's get one last comment, Thomas. So when you find something that you want to recontact them about, do you actually validate this with same sequencing? Yes, everything is, we're only, I don't believe that you have to only return clear results. That's a no-home-other argument, but everything we return is clear. And just one final, final comment is that maybe when putting together our report out of here is to think about getting a list together of some things that people are using and adding to consent forms that we might want to consider because newer things keep coming up and I think people have thought of things that others haven't and we probably want to be as comprehensive as possible, especially if we don't want to have to go back and reconsent for recontact and reconsent for other things. Right, or worse than that, which version of the consent form? Which version? Exactly, so I think we should think about that. All right, thank you very much.