Okay, so we are going to talk about health. Did you have a nice lunch? Yes. Good, terrific. Okay, so maybe the first question to answer is why MyHeritage is entering into the health domain. And I want to suggest two answers for that. The first answer that is relevant for all of you is that by entering to the health domain, we can bring new users, new people into genealogy that originally were not interested in genealogy. Believe it or not, that some people in the world that are not that much interested in genealogy or maybe are not or didn’t experience the joy of genealogy yet, but they’re really interested in their health. So by taking the DNA, the ancestry plus health DNA test that we offer, they come because they’re interested in health, but they will stay, they will join the relative matching and the other types of genealogy activities offered by the ancestry capabilities that we have now achieved. And I think one, I think really nice example we just heard yesterday about people that upload their data to Promethease, which is basically just for medical genetics or for health purposes, and now, we’re offering people the chance also to join to the MyHeritage database for regular genealogy purposes. So this way, kind of like we can reach to new audiences and new communities that maybe are less interested in genealogy, but at the end, we’re going to convert them and they’re going to be daft genealogists like us. So that’s the first answer. The second answer is that I think we have the moral duty, given all the data that you contributed, to tell people about their health. So really at MyHeritage, we have this quite massive resource that positions us quite uniquely to think about health of humanity. First, we have family trees, and we have billions of records of family trees. We also have in our DNA database over three million people that were tested, and these individuals were tested in a CLIA, CAP-certified lab, so the data is practically in clinical grade, although maybe they didn’t take a DNA, like a health test, we can still use this data to learn about medical conditions about the genetic basis of different diseases. And we also have surveys that many of you probably contributed to these surveys, and we already have 200,000 individuals that answered those surveys. If you don’t know the surveys that we have, so here’s kind of like an example of the list, or partial list of the surveys that we offer, all the way from demographic data, personal health, your family history, to even things about your sleep activities. We also just released a new survey, a really cool survey, in collaboration with researchers at MIT about moral machines. In this survey, we show you how basically driverless car can, different scenarios of what you should do in case of an emergency. Should you go straight ahead and hit a dog or steered away and maybe hit, I don’t know, a kid or something like that. And MIT researchers, we are going to collect the responses of people in order to understand the genetic basis of different ethical reasoning. So this is something that in the future we hope, down the road, can really give us some insights about the stuff like brain circuits, how we think about moral dilemmas and at the end, we can use this information to empower automatic machines, the AI of the future. But we have also many other types of surveys over here. Just another example that we have from a survey question is, here we ask you, has a doctor ever told you you had heart disease. And you can press if you had heart attack or if you suffered any other type of heart condition and we can collect this information. And the neat thing about it is that although this letter is collected over the internet, we’ve self reported data, not like medical, electronic health records, we can still replicate previous insights about the different basis of diseases. For instance, here we compared the BMI for individuals that answer our survey to the status, whether they reported they had heart attack or no heart attack. How do we collect the BMI? Because people report their height and their weight, so we can calculate their BMI. Now, of course, on the internet, everyone is a bit taller and a bit slimmer, also myself. (audience laughing) I’m six feet on the internet. Yet people are honest enough to answer this survey correctly so you can see that there is here an extra risk. So basically, if you had heart attack, you’re more likely to be slightly more obese or slightly more with higher BMI, and this is something that we know from previous epidemiological studies. So this is thanks to you and people like you that contributed to the surveys that we can collect this information. And using these data sets on the surveys, we had really wonderful resource that we can conduct studies. Here is a study that we published about a year and a half ago in Science, one of the top scientific magazines, using the MyHeritage and specifically, the Geni data. And we also publish, now, Science is extremely hard to get studies into this journal. It’s highly selective, yet the quantitative data was so good and so vast that we can actually get very interesting insights, especially about human longevity and migration patterns. We also published another paper in Science about a year ago, and this is really about how to use genealogy and the power of genealogy and genetic genealogy to capture criminals, like the Golden State Killer. So here we conducted a systematic analysis. So again, another paper in Science, thanks to the MyHeritage data and your contribution, and these studies generated immense interest in the community and also in the popular media, so we gave interviews to basically every major international media outlet. And also, it was selected by Science as one of the top breakthroughs of 2018, one of these studies. And also, had the pleasure to give a TED Talk about a year ago about, and really tell people why genealogy is so important for human health, and this TED Talk is going to be online in only three weeks, and I really hope that you will enjoy it. So we’ll let you know once it’s available. Okay, so today I really want to talk about the new health product and kind of like what is behind this health product and highlight some examples of reports that we have in this health product. So the MyHeritage Health, and I guess that you saw this slide already in several of the talks. Currently, we support 14 genetic risk conditions, and this include three polygenic risk scores, and I’m going to explain what are polygenic risk scores, and also 11 monogenic conditions, which are basically conditions that are dictated by a single gene in the genome, kind of like a strong one gene that dictates these conditions. On top of that, we also have 15 carrier status conditions. If you are a carrier, it doesn’t mean that you are sick with anything, I suspect you are probably healthy. But it means that if you go and you marry someone that’s also a carrier, you have a chance, one out of four, that your kid will have quite devastating genetic disorder. And we talk about cystic fibrosis, it can be Tay-Sachs if you’re Ashkenazi Jewish. It can be other types of conditions that are quite severe and we want to inform you in the case that you want to plan family, or maybe your kids, you are a carrier and you tell it to your kids that they are in the process of planning their family, because if you are a carrier, they have 50% chance of also being carriers. The process that we have basically in the US has physician oversight and genetic counseling, and I think this is a major component, because it enables us to contextualize the results to participants. We really hope to expand to Europe, but right now we still couldn’t identify a comparable service in Europe, but it’s something that we are highly interested in. We have here a new chip that we developed, a new array technology, that we developed and customized for this health report. And what you see right now, this is just the first glimpse that we can get from this array. We have many more, it’s like a treasure box that we can open and report many more conditions, and this is something that we’re going to see in the future, and we are actively working to expand this list with many, many more conditions. And I also want to highlight that for certain monogenic conditions, where the allele has high penetrance and severe consequences, we conduct Sanger sequencing to verify that you really have this condition and in a few slides, you’re going to see why it’s so important to have this Sanger validation. So I just want to first highlight one monogenic condition that I think people probably heard about it. Have you heard about Angelina Jolie and her BRCA status, this New York Times story? Yeah, raise your hand? Okay, so she discovered that she’s a carrier for BRCA1/2 mutation, and these genes, if you have a carrier, if you have these ultra rare variations in these genes, you have high likelihood to develop breast cancer if you’re a female. So this is the risk, but you can see over here, this is the breast cancer risk, and this is the age of the individual. So basically, the lifetime risk of a person can reach to about 80%, and she decide, she discovered that she’s a carrier, and she decide to take a preventive step of prophylactic surgery, which is not, it’s a mastectomy, it’s a quite aggressive step, but she felt this is the way that she can basically save her life and not to experience cancer, which people of this mutation usually develop triple-negative breast cancer that is quite deadly. We have the option to deliver this information to you if you want to know that. And our competitors at 23andMe, they test only three BRCA1 mutations, but we worked hard and we can actually test 15 BRCA mutations, and we can discover 70% more carriers in our test versus our competitors. Now, as I said, all the people that are positive to BRCA, they will go through Sanger sequencing, and do you know, from a chief science perspective, I have to tell you that this is a hustle. We already have the sample, we know that you’re a BRCA1/2 carrier on the array, why to do this extra step? Statistically, it complicates things, and also, it delays your results because we need another few more weeks to run this secondary test. So why this is important to do a Sanger validation? And let me explain that. I want first to define a term that called precision, which Gilad also defined it yesterday in his keynote. Precision is the rate, basically, that if I tell you that you are positive to BRCA1/2, that you are really positive, that it’s not a false positive, that you’re truly a positive and you are going to develop breast cancer. Now, mathematically, it’s basically the rate. Take all your true positives and divide them by the true positives plus the false positives, and we want the precision to be as close to 100%. If it’s 10%, it means that out of 10 people that I tell them they have a BRCA mutation, only one truly has, and all the other ones, I just told them the wrong information that probably created some stress. Maybe they even took some steps to do surgery or something like that, that is, so very bad idea to give them things with low precision. Okay, now let’s look at something. Let’s take an example. Let’s think that we have a very good test in terms of accuracy, and our test makes an error in only one out of 1,000 samples. Do you agree that this is a good accuracy for a test? Making an error in one out of 1,000, it’s nothing, right? But also, let’s assume that the allele frequency in the population is one in 1,000. And this number’s actually not far away from the real numbers that we have. Now, if you look at a precision, and you look at the number of true positives, this is this 1,000, because for each thousand individuals that we test, only one will be a carrier. Only one has the BRCA mutation. So we take the true positives and put it here, also the true positives we put it here, but it makes a mistake in every one in 1,000 regular samples. So what you get if you do the math is that even with this extremely good accuracy, this is one in 1,000, the precision is only 50%. So it means that if I test 10 individuals, five, and I tell them, these 10 individuals are carriers for BRCA mutation, only five are truly carriers. So this is what you get, basically. Now, this is something that we all learned in, if you take statistics as an undergraduate, you learn about this. But it was actually reported recently by The Guardian, so scientists in the UK analyzed this precision mechanism and said that, oh my god, this is like, it’s really bad because you get to precision that is very low if you just use arrays. But the thing is that we don’t just use the array, we have a Sanger validation step after we see a positive sample. So let’s say that Sanger also has one in 1,000, or something like that error rate. So now if you look at the precision, you take this number, since you did it twice, you have to be wrong in two independent tests. You get to a precision that is 99.9%, which is what we want. So this step, although it complicates from our logistic perspective, running the sample yet again, super important because it nearly eliminates through all the false positives that we have. So you get very good results. So if we tell someone that is a positive for a specific monogenic disease, highly severe disease that we test with Sanger, there is a slight, only a fraction, virtually no chance that this is a false positive because we ran it twice with two independent technologies. So this is for the BRCA mutation. And now I want to tell you about polygenic risk scores. And polygenic risk scores are one of the most hot topics right now in medical genetics. We see really a surge of studies and many advancements in the past two years about polygenic risk scores. And this is the report for a polygenic risk score that you have, and this is just for heart disease. And you notice in this report, basically, that we report the disease, basically, with some verbal cues, with some graphical cues, and also with some quantitative cues. Why is that? Because we know that different people like to consume information differently. Some people better understand graphical interpretation. Other people, they need really the verbal information, and other people need the quantitative information. So by scoring all of these, checking all these three options, we maximize their ability to really communicate the risk to individuals and explain to them, what is the test. And also we have some, kind of like in this report, other steps that you can take after you got the test. So what are polygenic risk scores? So the concept of polygenic risk scores is that the score is not affected by a single gene, but by many gene, by many SNPs that spread across the genome. What you see over here, this is basically many SNPs on the genome, in chromosome 1, chromosome 2, we just basically sorted them on the chromosomes, and then along the positions of the chromosomes. And on the y-axis, you see the contribution of the SNP to a specific disease. I think here we, I think I might have depression here or something like that, and what we can do, basically, is to look at all these highly significant SNPs and derive their effect on your ability to develop the disease of interest. So there is some, a certain weight for each SNP. And what we do, we take these weights and we take here the SNP status, whether you have zero copies of the risk allele, one copy or two copies. For the autosomal region, you have up to two copies, and then we’re basically going to take this effect plus the status of the SNP, multiply them together and keep doing that for all the SNPs, and for some of our reports, we do it for 300 SNPs. For some of our reports, we do it for over 40,000 SNPs, depends on the statistical model that we use and that generates the most robust results that we have. So we do this process, and then we can give you the score, and just as an illustration, this is, it looks like that. So from mathematical perspective, you start here, with some, let’s say for Type 2 diabetes, there is a general risk in the population of 30%, even before I inspected your genome, I didn’t look at your genome at all, you have, let’s say, 30% to develop type 2 diabetes. And then I start to take each one of your SNPs that are correlated with type 2 diabetes and look at the status of the SNP and then I look at what is the effect size of the SNP, and then I can just start to sum those SNPs together until I reach to your final risk, which, as you see, it’s larger than the 30%, so this person has higher risk for Type 2 diabetes. So this is the way that we can calculate a score for you. So the first disease that we decide to focus on was heart disease, and for me, it was highly important, because as I joined MyHeritage, about two months after I joined, my father passed away from his first and last heart attack. He was 68 years old. You never ask yourself, I knew that my father was a smoker and had high cholesterol levels. On the other hand, his BMI was kind of like normal and he didn’t exhibit other risk factors. Also, his father didn’t have heart attack until the age of over 90. So the question for me was, is this something I should be extra worried about, or maybe this was just a pure chance. What is my genetics going to tell me? Should I take any significant steps to reduce my risk, or maybe this was because he was a smoker and I’m not a smoker, so I should not be overly concerned about his heart attack in a relatively young age. So this is my story, but of course, there are many reasons why we want to focus on heart disease as our first polygenic risk score. Heart disease is the top cause of death in the Western world, so by looking at this risk, we can tell you something significant about one of the most hazardous conditions that you can think of. The risk is modifiable. If you go to the gym, and there is a nice gym here in this hotel, by the way, if you go to the gym, basically, three times a week, you watch what you eat, you make sure that your LDL levels are okay, you don’t smoke, you’re going to decrease your risk to develop heart disease. So by informing high risk individuals we can empower them to reduce their risk and this way, we can prevent this heart disease, in some cases, with these individuals. So this is not, so I want to emphasize, this polygenic risk score is not like a crystal ball that is going to say if you’re high risk, that for sure you’re going to have heart disease. There is something that you can do, there is some sort of empowerment going on here, because you can take it, you can take this information and you can decide to change different things in your life to mitigate the risk. There is also some support from the medical community of reporting the heart disease rate. So here is one example is this perspective by Evan Ashley, one of the top heart disease geneticist researchers from Stanford saying the importance of using these polygenic risk scores to stratify risk between individuals, and there is here another paper by another cardiologist, a geneticist, saying these polygenic risk scores are quite important. They can save individuals. So remember, here we do health. We need to be innovative but also humble and understand where is the medical community on certain things to make sure that we deliver value and we do not harm individuals. So we saw this support, and we thought, okay, that’s a good sign. This is the perfect polygenic risk score that we should develop. So to develop this polygenic risk score, we turned into a peer-reviewed study by Abraham et al, over here. And we wanted to use a peer-reviewed study because we knew that this study went through some rigorous review process by an independent group, published in a good journal, so the results are robust. In this study, the polygenic risk score is based on 46,000 SNPs, and the authors tested this polygenic risk score using five different cohorts, and this is great, because if they could replicate the results within five different cohorts, we thought, okay, once we take the results to the MyHeritage data, probably we can replicate them as well. And look at these results. Here I’m showing you the disease, sorry, the probability to develop heart disease as a function of age in males. Now, you see the two figures? These are high risk individuals, genetically, and these are low risk individuals from a genetic perspective. Now, of course, if you are a smoker, you have higher chance to develop heart disease, but look at the striking differences between the groups. People that are non-smokers in the high genetic risk perform much worse than smokers in the good genetic risk. So it means that these polygenic risk score can differentiate between high and low risk individuals better than the smoking status of the individual. If you have good genetics and you’re a smoker, you have less chance to develop heart attack, significantly less chance, whereas if you have bad genetics and you’re a smoker, very bad idea. Okay, now we tested. We took this polygenic risk score and tested it on the MyHeritage cohort. Now, polygenic risk score, it distributes like height, like normal distribution. Most people have the average score. They have a certain risk for heart disease, not too high, not too low. But we have some people that got great genetics, really good one, and we got some people that got really bad cards in their hand. And the odds ratios of each standard deviation going from the mean to the extreme increases the odds to develop heart disease by about 1.6, it increases, then increases more, and more, and more. And just to show you the effect, we stratified here individual at a high risk versus low risk genetically in females and males, and we look at the survey data when these individuals reported that they had heart attack, if at all. Now, you look at here at people who have the high risk, versus the low risk in males. There is almost a decade, even more than a decade, that they got heart attack difference on average. The heart of the people in the low risk group is kind of like a decade younger than the people in the high risk group. And we can stratify this risk, and also we can do it with females, and really tell these individuals to take better care of themselves. Here we stratified, again, the individuals and the MyHeritage data into different bins. We took the people in the lowest bin to develop heart attack, the lowest 2.5 percentile in this normal distribution, and we look at their survey results. And they have about half the risk, half the odds to have a heart attack, and then we look at people here, the top 2.5 percentile, and they have nearly four times the risk to develop heart disease. And this risk is similar to familial hypercholesterolemia, which is a quite, very genetic disease of high cholesterol. They exhibit the same risk. And I want to stress the point that 2.5, the top 2.5 percentile means that one out of 40 individuals that are tested with the MyHeritage Health will be in this group. In this room right now, I believe we have about 80 people. There are two people over here that they are in this group. And just to tell you what it means to four times the risk to develop heart disease, here you can see, I took a study about the connection between smoking and heart disease, and the odds, four times the odds to develop heart disease is comparable to a person that smokes 20 cigarettes a day. So it’s like we have people in this room that genetically, they have these set of combinations of SNPs that they behave as if they smoke a pack a day, although probably they, maybe they are not smokers at all. Still the heart experiences the same stress, the same risk as if the average person that smokes a pack a day. But the good news is that the risk is modifiable to some extent. Here is a study published by a Harvard group at New England Journal of Medicine. Now, in this study, they had a bit different definition about what is bad genetics and good genetics, but use a similar tool and in this study, if you have bad genetics and you have a bad lifestyle, meaning you don’t exercise, you eat crap and so on, then you have four times the risk to develop heart disease. But if you start to exercise good lifestyle, you go to the gym, you eat healthy, you maintain your BMI, you can cut your risk by half. It’s still more than the people with the best genetics and best lifestyle, but still, it’s not far. So you cut the risk by half. Now, just as a comparable, there is also a group of people that have good genetics and they have a bad lifestyle. They really ate all this greasy food, don’t watch their cholesterol, and they don’t exercise, but they got great genetics and nothing happens to them, or the risk is only twice. So yeah, people with the high risk group, we can modify the risk. They will never be like people with great genetics and great lifestyle, but we can still do something for them. We can inform them, and if they take an action, they can still significantly reduce the risk for heart disease. And if, just as a comparable, so we can move them from heart that experiences 20 cigarettes a day to heart that experiences 10 cigarettes a day, it’s better than nothing. It’s better than to be in the other group. And the result of self support, it’s not just us thinking that this is great, but the community showed that there is clinical utility in reporting to individuals about their genetic risk for heart disease. We saw that if people get this information, from a genetic perspective, not just a regular thing. You can go to your physician, and they can tell you, based on your BMI and your LDL levels, you have this risk for heart disease. But we see that if you tell people from, you take a genetic test, and you tell them the risk, after a genetic test, they actually take this risk much more seriously than people that just got the risk based on BMI and LDL. These people tend to adhere more to starting therapy, they tend more to talk with their family members and their family care physician about different options to maintain lower risk. We just see that people treat genetic information more seriously, and it moves them to take an action that benefits them. Okay, so now you might ask this question. You tell me, oh, Yaniv, so you can tell me about my risk for heart disease, but I already know my BMI, which is genetically controlled. I know my cholesterol, which is partially genetically controlled. Is this risk is going to inform me anything new? I already know my risk from other phenotypes, other risk factors. Do I get anything new? So I want to show you, the answer is yes, but I want to actually emphasize that with our type 2 diabetes report, which is more striking. So one of the most significant risk factors for type 2 diabetes is–
Sugar. Sugar and BMI, basically. The higher BMI that you have, you have the higher likelihood to have type 2 diabetes. So here we stratifying individuals based on their polygenic risk score for type 2 diabetes. Okay, again, this lower 2.5 lucky people that have very low likelihood to develop type 2 diabetes genetically, to this high risk group. And now, what you see over here, this is the BMI of individuals in each category. And then, whether they don’t have type 2 diabetes versus people that do have type 2 diabetes, and you can see the difference in the BMI. Now, look at this group. These people with great genetics got mixed in kind of like robust type 2 diabetes. They really need very high BMI to start developing type 2 diabetes, but as you move along to people with worse and worse genetics for type 2 diabetes, they need lower BMI to develop the disease. So it means that it’s not that genetic tests that we do look at the BMI, the genetic component of BMI individuals, what it means is that these people with the great genetics have some immunity. So the better genetics that you have, you have better threshold, you can tolerate more and more increase in your BMI before developing type 2 diabetes. And this is quite, this is exactly what we mean by precision medicine, that you go to your physician. If you are in this group and you are a bit overweight, your physician should tell you, it’s okay, probably you not develop type 2 diabetes because you’re in this group over here, and you really need to be above BMI of 35 to develop that. But if you’re in this group, man, you should start watching your BMI because even a slight increase in your BMI above the normal can significantly alter your risk to develop type 2 diabetes, and this is what we want to create in our health product, to inform you this type of decision so you can better plan your energy. We know energy is, it’s limited. You can only worry that much. So we want to focus our energy to worry about the conditions and the disease that are more likely to occur to you. Okay, one caveat that we have is that we have these polygenic risk scores only to Europeans. Unfortunately, if you’re not a European, and I’m also, I’m half Ashkenazi Jew, but my other half is from the Jewish community in Uzbekistan, so unfortunately, I cannot receive this polygenic risk scores, and I still don’t know whether my father heart attack means that I need to significantly change my life or not, in terms of protection to heart disease. Why is that? Just a second, it’s questions at the end. Why is that? Here you can see, this is a study. This is an independent study, not a MyHeritage study. But they took the same previous study that I described to you, the Abraham et al study that we also use, they took this study and they ran it again with different types of populations. The Abraham et al study studied the risk for heart disease based on Europeans. What this study found is that if you take this same kind of polygenic risk scores and try different ethnicities, whether it’s basically people from Africa or whether this is people from East Asia, and so on, you get a shift in this distribution as if everyone from Africa is highly likelier to suffer from heart disease, and many people who are from East Asia has higher likelihood to suffer heart disease. This is not because there is a risk in these communities that is higher for heart disease. This is because the model is not calibrated for people that are not Europeans. And I’ll skip this slide, but I just want to emphasize that, and this is something that we know from the scientific community, that and there was a study just published a few months ago in Nature Genetics, one of the top journals, saying that we have an issue in genetics, in basic science, since nearly all of our studies were done with people of European descent, we can only develop polygenic risk scores to this group and not to other groups. And since we at MyHeritage, we use only peer-reviewed data for our polygenic risk scores, we are basically bounded by the scientific knowledge that is out there, and since there is significant biases in these basic science studies, we can only right now deliver the polygenic risk scores only for Europeans, and this is a caveat that we hope to work in the future. It’s something that affects many people in my team that are only partially Ashkenazi Jews and they want to know their results for other and they cannot receive these results. And I think also, many members that you met from MyHeritage today, but this is what we can do right now and we hope to really remedy that in the future. So one question that is important to us is not just to deliver accurate results to individuals, but really to know whether they understand the results, whether they can comprehend what we told them. And we decided to test that in empirical study, and we used the same framework that was developed by a Harvard group, by Robert Green, the PGen study that they published, and the concept is that we show mock results to a group of individuals with the same demographic as you, same demographic as the MyHeritage users, and we ask them a list of questions, whether they understand or not the results. And these are just partial list of the questions. We tested with 100 people to get some statistical significance over here. So we showed them a report about a guy named Andrew, and then we asked about Andrew and so on, and for most questions, nearly everyone was able to correctly answer the results. Now, just to give you a control, the FDA requires at least 90% comprehension, and even with the most complicated question, we got over this number, meaning that people truly understand, most of them truly understand the report and can comprehend information. Okay, so let me summarize this talk. I tell you about why we are interested in moving into health. And this is a natural expansion for our genetic genealogy offering. We’re going to bring communities together and also we think that we have this moral duty to inform individuals due to this unique position that we have with the vast data sets that you contributed. Our report has both monogenic conditions, such as BRCA1, BRCA2, MUTYH that is associated with colorectal cancer, and also polygenic risk scores. These scores are not affected by a single gene, but by many SNPs across the genome that we have this framework to evaluate, process, and then deliver you the results. I showed you the results about heart disease and about all the concept behind this polygenic risk score, and the ability of individuals in the MyHeritage user base to comprehend this risk score. So with that, I’ll take any questions. Thank you very much. (audience applauding) Yes.
Here, I have the mic. Oh. So is the health product FDA certified? The health product, we use what’s called lab-developed test. We have a physician that prescribes the test, and it’s in a CLIA, CAP-certified lab. The FDA says that if you operate within this functionality, since it’s not a direct-to-consumer test, you don’t need, basically, a review process through the FDA. The FDA agrees to test like that, and we do have a rigorous review by the physician mentor that will work with every report before we give it to the individuals, go through a review process by the chief physician. We have to show all the data, basically, a very similar presentation that I did to you, we present it in a report, convinces, we convince this person about the accuracy and the validity of the data, and only then we can really give it to consumers. Are you working towards FDA certification? Right now, we really like this framework. We think it allows us the most of the innovation and the flexibility, and I want to emphasize that currently, the FDA doesn’t have a clear guide on how to develop polygenic risk scores, so we believe, and it’s not just us, but there are other companies in the market that operating through lab-developed tests, with an independent physician oversight, with genetic counselors, and this is something that the FDA doesn’t require, actually give you much better product. So it’s also safer and gives you the ability to innovate through this product. Here.
Here we go. I just want to comment that I really like the idea of the physician oversight. I think that’s really important. Terrific, thank you. As far as I understand, the physician oversight is only in America. So if I ordered my test and I didn’t have to go through a doctor, and I’d much rather not go through a doctor, personally, but is there any, are you going to get clearance through any of the European regulatory authorities for the test?
Yeah. So we want, the authorities, the regulations in Europe is much, it’s kind of like more a patchwork, why, because each country has its own thing. This is the reason why we don’t sell in France or in Germany, other places that it’s, we can really, cannot sell there. And we hope in the future, I think, first, another component is the genetic counseling to give to people. We just need to identify this telemedicine services that you can really have this phone conversation with a genetic counselor, not to tell you, okay, now drive away two hours to meet a person that nobody wants to do that. So this event will take some time, because in each country we have to identify these different genetic counselors. There are also very, very few genetic counselors in Europe. There’s only a handful in the UK. That’s why you need to take the telemedicine, because otherwise, maybe from London you have to go, I don’t know, somewhere else, or maybe from somewhere else to London to get your genetic counseling. So the other thing I wanted to ask was about the studies where it’s showing the risk. Are there likely to be other factors involved, so some environmental factors– Of course. Like somebody’s job could affect their risk score, and are you practically going to think about collecting data on environmental factors that could contribute to peoples’ risks? That’s an excellent question. We show you only the genetic component. We know that different types, you know, even you gave a great example about your job status. By the way, there is a great status showing that people, especially managers, before April 15th, the day of tax in the US, there is a spike in heart attacks. Yeah.
(audience laughing) So of course, environment plays a role. There are also examples of when your favorite soccer team loses, you see people admitted also to the emergency room with heart conditions, so yeah, there are environmental risk factors. We only give you, no, we don’t see your entire life. We cannot really monitor whether your favorite soccer team is losing, we want to give you the basic thing, what your genetics tells you, and this already explains quite a lot of, quite substantial risk, so with that, of course, you should take this information. This is like genealogy. This is just the first step. As always, DNA is just the first step. Take this information, take other things in your life that are going on, and think together about your risks, how to manage those risks. It’s well known that in addition to genetics, epigenetics is also very important for your chance of getting disease, and especially age-related disease. Would it be expensive, would it cost much money to develop tests for epigenetics? Yep, so epigenetics are different decoration that goes on the DNA, and kind of like, and changes the DNA. The thing is that, first, the science is less robust than the genetics, second, you have to remember, we don’t see these epigenetic modifications are in the heart issue. We don’t sample your heart. We sample your cheek. We don’t go in to do a biopsy of your heart. So I’m not sure that the epigenetic markers in your cheek are actually reflective of anything that is going on with your heart. So right now, we just stick with genetics and we want to squeeze as many disease that we can with the genetic information, and we have tons of data over there. We just started to scrape the surface, I believe, with this test. So, I’m not going to comment on ethnics people, because it’s obvious that the ethnics people are probably the most difficult to get these polygenic, the risk scores off, so I guess your only advice for those is to live a healthy lifestyle? It’s advice for everyone.
Exactly. But we hope to fill it. It will change in the future. It will just take some time. It will take time, exactly.
It will take time. We are basically, we need scientific, we took this decision that we only base on peer-reviewed studies for these polygenic risk scores, so since most of the studies are based on the UK Biobank and other European cohorts, we–
You have to– We wait until some efforts to collect more diverse populations. I’m also affected by that personally, so I don’t like it as well, and I’m impatient to actually see my results as well, calibrated results. But I wanted to ask you, for another, a health problem, and another, I think serious health problem is to live a very long life. So I remember there was recently a SNP identified that’s correlated with very old life expectancy. Which study are you talking about? I try to look it up, but I think it was, but my question is– Yeah, I think you–
Do you also predict longevity in your reports? Longevity, yeah. So A, there was a GWAS in longevity, published 2012 in Science. They identified two SNPs that explain 70% of the liability to being a centenarian. Unfortunately, the study was retracted because they had a problem with their Illumina array, which flipped some of these two SNPs. It was a huge embarrassment for everyone in the study, and it was retracted, published in PLOS One, and didn’t find anything, basically. So since then, I’m not aware of a single genome-wide association study with robust replication of SNPs that helped longevity. We published Science paper about longevity, one year and a half ago. I believe that longevity is a minute genetic component, in fact, only 15% of the variance in population is explained by genetics. So even if we can predict perfectly longevity from our genetic data, we can explain up to five years, on average. I think it was last year. I’ll look it up.
Sure. So maybe it’s junk science. It could be.
I think it’s probably that. We’re all going to die, currently. Oh my god. (audience laughing)
That’s sad to know. Let’s ask a difficult question. Are we going to solve that? We have a team. But I’m just, ’cause I’ve done my 23andMe test, and it’s given me lots of health risks, but I think what would, the take home message is, a healthy lifestyle, eat well, exercise more, don’t smoke. But what would help me also would be if there is a comparison of my risk for developing diseases that would be a financial strain on me, compared to say the average risk of developing those diseases that will cause a financial strain. Any plans for that? I’m not sure, to stratify the disease risk, but then, how you compare? What’s worse, breast cancer or Alzheimer? I’m not sure how we know what’s, what we do, basically, we give you the risk of different diseases, modifiable risk, automatically, and for Alzheimer and Parkinson, where the risk is not really modifiable, you have a key to select in order to see the risk. But I’m not sure, but you said, you know, and the interesting thing about, you should just live healthy life, the thing is, what is healthy life? Because again, energy is limited. For some people, healthy life means, is actually go to do breast cancer mammography more frequently, for other people, other females, maybe take your ovaries after you finish having kids because you’re at risk for ovary cancer. For again, and taking another from females, for other females, it’s actually, you are in high risk for osteoporosis. How about starting early on, do some exercise, basically lifting exercises, to make your bones stronger? So there are many things and many advice, and since we cannot tell you, okay, live a healthy life and just work in the gym all day and do mammography, you have actually to live your life also, and we appreciate that. So we want to actually give you, here are the top things that you should worry about, and it doesn’t mean that you will suffer from those top things, but actually to prioritize your energy in taking care of your health, and also life planning. If you’re in high risk for Alzheimer, okay, you should be, if you want, you should be aware of that, maybe it will help you to plan your life better. I know in America, you can get nursing home health insurance. It’s not available in the UK, I don’t know if it’s available anywhere else, but it means that if you have a high risk of Alzheimer’s disease, you can take out health insurance that will cover the cost of that– Might be.
So you don’t have to give, sell your house to take care of yourself in later life. That might be some sort, if it’s possible, and if you want to raise your premium by having high risk for Alzheimer, yeah, that might be a solution. There is a question over here. So if I buy the health subscription next year– Why next year, why not now? (audience laughing) So my understanding was it’s free for the first year, and so then if I buy the subscription next year for $99, what does that get me? Is that one report every month? What is that pipeline? What it means is that your reports are going to be aligned to the ability of my team to produce these reports. So basically, as we are working now on new reports, and we have great reports, and I cannot talk about them, but we have really interesting reports, so it means that as we get these reports, you will get them as well. So always, your genome will be aligned to the latest and greatest scientific knowledge that we can produce and so that’s what mean that you’re going to keep these reports going on. I would really like, I had the test. I would really like to see a combined chart like the one you had of the BMI and the heart risk, because I think that combines two really important factors and looking at that together was like, oh. I bet that provides something that neither one alone provide. I would just like to provide that input back. Yeah, and for some reports, we’re going to do that, so basically tell you, if you’re a smoker, you check, and then it will change your risk. But the thing is, it’s kind of like where to stop, because for many diseases, it’s not just you can be a smoker, but you can be in a high stress job, or you can be in a high stress situation, so the risk will change. So kind of like the question is, how much to move out from the genetics to these environmental factors. So by pulling for certain diseases, we will add this kind of like flavor that you can see your risk based on the major risk factor. So another question about penetrance, because I know the UK Biobank, they looked at things like hemochromatosis and people who have hemochromatosis that the penetrance in UK Biobank, I think there’s something like, I can’t remember what it was, something like 25%, so a lot of people have these markers but they never actually develop the disease. So is there any chance you might eventually be able to include information like that? So that I think certainly like Alzheimer’s, a large number of people have those genes, but– We do, that’s what we test. We test this condition, and we selected a condition with either very rigorous test based on the recommendation in ClinVar, right, by clinical geneticists, what’s their recommendation? We also consulted several people, several studies to make sure that we include the highly penetrant mutations. There is one that it was kind of questionable, and at the end, we decide that we don’t want to include it because want to reduce the risk of a false positive. But genetic information is evolving, and this is what you know if you have the subscription. It’s part of like, as this knowledge is getting updated, and if we can find these mutations on our array, we’re going to update the models because this is a nascent field, basically. Okay, thank you very much. (audience applauding)