Traditionally, cardiologists have focused on the same 5 cardiovascular risk factors. Preventive cardiology currently uses these factors to determine risk and then prescribes the same therapies to all those at risk. Can we use advances in technology and computation to more deeply characterize our patients, identifying the maximally effective individual therapies to prevent coronary artery disease?
Welcome. You're at the cutting edge of Cardiovascular sciences. I'm your host, John Cook. I'm professor and chair of the Department of Cardiovascular Sciences. And with me in the studio is Dr Lee Lai, who is also a junior faculty member in the Department of Cardiovascular Sciences. And she's going to be introducing our guest, Dr Pradeep Nada Rajan. Good morning, everyone. Today. It is my great honor to introduce our speaker. Doctor Prab Nad Rajan is the director of preventive cardiology and the Poor and the fire man in chair in vascular medicine at the Massachusetts General Hospital, Associate Professor of Medicine at Harvard Medical School, associate member of the Broad Institute of Harvard and MIT. He received his bachelors degree in molecular biology from the University of California Berkeley and received his MD from the University of California San Francisco. He also received his Master of Medical Science in biomedical informatics from Harvard, Harvard Medical School. Rajan complete his internship and residency in internal medicine at Brigham and Women's Hospital. Then complete his clinical and research fellowship in cardiovascular medicine at the Massachusetts General Hospital. Doctor Nat Rajan's research program span human genetic variations by medical informatics integrative genomics and genotype based deeper phenotyping to, to gain insights about cardiometabolic changes. Doctor Nat Rajan. Welcome to the show and looking forward to hearing about your multidisciplinary work. Welcome, Prade. Thank you so much. Let me just pull up my slides here. Um You know, before we get to the slides, Prade, maybe we can just have a little discussion to introduce you to the audience. And you're right now in charge of preventive cardiology at Mass General Hospital, you're doing research in the genetics of cardiovascular disease, which we're going to be hearing about in a moment. And, and you have a practice of preventive cardiology at Mass General. Tell us about your clinical practice. Uh Great, so wonderful to be with you folks and I love this educational forum um that you have here. I I, you know, my clinical practice, as you said is in uh preventive cardiology. So we have a group um of about 8 to 10 clinicians or M DS and then also research, nurses, exercise, physiologists, dieticians, um and preventive cardiology means something slightly different at each institution. But we house um cardiac rehabilitation as well as other cardiac rehab related programs, including supervised exercise therapy for peripheral artery disease. And then also longitudinal clinics that are focused on primary prevention as well as secondary prevention centered on sclerotic cardiovascular disease, identifying or helping to identify individuals for whom we want to think about pharmacotherapy that often involves risk stratification with biomarkers and imaging and then often, you know, evaluating our patients who have had premature events or recurrent events. Um despite um guideline, supported um medical therapies that all of that helps us inform. What are the important uh research questions in this domain. Um And one particular act I'll talk about especially on risk identification and management. We'll be using genetics, but I also do some clinical cardiovascular genetics really as it relates to coronary artery disease and lipid disorders. And we have a separate um cardiovascular genetics program and in that program, um us cardiologists in that group, we do see patients with genetic counselors and most of that as it relates to this domain is on familial hyperchloremia and we'll focus on a little bit of that too miss to, you know, there's another fellow in your, in your town, Boston Colin mcrae, I'm sure you're familiar with him and he has used a multi omics approach to define different phenotypes of cardiovascular disease. He believes that we are not properly treating cardiovascular disease because we're not properly phenotyping patients. And it sounds like you're ahead of the game. It sounds like you're trying to properly phenotype people and give them personalized therapies. Yeah. You know, I and I, I totally agree with Callum and I'll touch upon this in the talk. You know, we have been focusing on like the same five traditional cardiovascular disease risk factors um for at least a half a century. Um now we are able to um deeply characterize our patients at a much, much higher depth compared to before. That's, you know, thanks to advances in technologies as well as computation and the phenotyping, I think can come in a variety of different flavors. There's a lot of phenotyping that's already being collected by our patients about our patients from the electronic health records, from their smartphones and from their wearable. So I think there's a lot of opportunities there on the molecular side. What is at least um available at substantial scale now is genetics. But increasingly so all the other downstream molecular features transcripts metab prodi how all of this stuff will shape to inform the care of our patients. Still is actually fairly early days. There has been a hypothesis especially for complex diseases that can you like sub subdivide individuals for a complex disease like coronary artery disease. And if you subdivide will you be able to identify the best possible therapies um that has not yet borne out for coronary artery disease. That's because we know, I think we're still generating data. But you know, I think that is um that's what we hope to get towards. Ultimately, you know, we center preventive cardiology and coronary artery disease largely on risk prediction and then kind of give the same therapies for everybody at risk. What we'd love to do is kind of move beyond risk prediction to say this is gonna be the maximally effective approach to ultimately prevent coronary artery disease and that would be bespoken specific to you. Um And, you know, we are, we are on that trajectory, but we're actually still fairly early on that trajectory beyond the traditional modifiable risk factors of treating high blood pressure, if it's high, treating, you know, um, abnormal blood glucose, if that's elevated and then addressing risk related to those risk factors. I think that's, that's wonderful. And it's, it's interesting given the fact that it was in Boston, actually, that the traditional risk factors that we're talking about, you know, really were identified through the Framingham studies. And now, you know, you, you're on that trajectory, carrying it forward into a much finer phenotyping and that's, that's very cool. Um What got you interested in science and medicine? Yeah. You know, I think, um you know, when, when we all want to figure out how we're gonna contribute to the world, um my, my interest and focus has been on pro you know, problems that have been a around me and visible around me and that are important to people around me and then ideally important to the world. Now, coronary artery disease has been, you know, it had this dubious distinction of being the leading killer um in the U in the US for at least a century. Um at least says the epidemiology of cardiovascular diseases have evolved and it's retained that distinction um with cancer actually increasingly as a close second Um but, you know, in the world communicable diseases historically has been been the leading um killer her. However, over the last two decades, noncommunicable diseases has taken over and that's principally due to complications of ischemic heart disease. Um South Asians also are at, you know, very high risk for a sclerotic cardiovascular disease, type two diabetes. Um So these are things that were kind of around me all the time and clearly important problems. Clearly, nobody had solved. You know, it's not like it's not like we have solved them now, but we're all working towards um trying to address um this really important problem around us and you know, there are many ways that folks can contribute to this overall goal. Um I, you know, I've been excited over the last several years using lots of advances in assays, technologies and computation to really address this at a, at a scale and a depth at the target organism of humans. That was not previously possible. Great. Wonderful. Um Well, I think with that nice introduction, Prade, let's let you cut to your, your slides and tell us a little bit more about the polygenic risk scoring for coronary artery disease. Great. Um So, thanks so much. Yeah, this is gonna be one slice of, you know, I think opportunities but a slice that is um uh translation translatable potential. Um You know, on the near term polygenic risk scoring, these are my disclosures unrelated to this talk and you know, a year ago, actually less than a year ago, about six months ago, we put out a scientific statement on behalf of the American Heart Association about polygenic risk scoring. And you might say, you know, this, you know, this is seems like science fiction. Why are we talking about um something like this and why is it now important for a broader audience? What is unique is um for a biomarker like genetics because of these advances in technologies and the public's interest, genetic profiling broad genetic profiling is already out there among our patients and beyond, you know, this is a little bit of an outdated slide of the prevalence of direct to consumer genetic testing, but it's already available in tens of millions of participants. What is actually highly unique is that individuals are getting this, they're not getting it through their doctors. Whereas if you look at blood lipids, which is, you know, well described risk factor that we've known about for a half century, you know, you you you can't bypass your doctor or a doctor to get a blood lipid panel. However, you can get, you know, your whole genome sequenced. And increasingly, you know, the these platforms are returning health related information, including, you know, cardiovascular disease, risk predictors. And you know, there are many health care um entities that are developing large bio banks. So opportunities to synergize this kind of information with health care and these are growing at rapid scale. Um And so it provides us with the opportunity to, for uh a new way to or at least complement the way that we think about cardiovascular disease prevention. As I highlighted, you know, the epidemiology for much of the, you know, early 20th century has not been engrained and there has you, you'd say, oh, great. You know, it's, it, it has not been rising as far as deaths related to cardiovascular disease. And there's been a plateau, but it's not like we're coming back down to what it was before. And if anything over the last decade or so, the rates are rising again, wide variety of complicated factors and then again, still has this dubious distinction of being number one. And you know, we have been practicing preventive cardiology in its current form, you know, pretty similarly for the last 30 40 years with the stalled progress in epidemiology as the population gets older as things cost more, the costs related to cardiovascular disease continue to increase. So we need to think about, you know, what are the ways that we can substantially alter this? And this is important because cardiovascular disease affects a lot of people. So even making small changes, small increments actually can have a profound impact on both of these kinds of curves, as I was saying earlier on, you know, at the Framingham, uh hard study locally and and others seminal studies relating traditional cardiovascular disease risk factors with cardiovascular disease. And these, these have emerged to also be causal factors allowing such that, you know, if you, if one thing gets perturbed and then you rectify it, then you'll at, you know, you'll reduce the risk related to that. Now, these same risk factors have been using cardiovascular disease risk prediction since the seventies. And, you know, it's changed in a variety of different forms. Now it's the pool cohort equations. Um, and here's a mathematical exercise if you look at each of these individual risk factors. And um here it's just, you know, you know, in men, but really bend by age. Ok. So as people increase, if you're um you know, if you're on the older side, so in the seventies, the risk is always estimated to be high and these are 10 year risks for very young, you know, early middle age, at least here, the risk is always estimated to be very low and the sweet spot is basically solid, middle age for discriminating action, ability of who should be on pharmacotherapy and that's not unreasonable. You know, most events are happening later in life, not as many events are happening early in life. However, this approach doesn't allow us to identify the individuals earlier in life that you want to um invest um resources into not only prevent premature events, which this approach actually doesn't help us identify but also prevent later in life events. Now, genetics um has been invoked as a way to, you know, help identify these individuals early in life. As I was saying earlier on that, you know, the scope of uh genetics and coronary artery disease has been restricted to familial hyper cholesterol Lemi in the molecular basis, contributed to, you know, by many individuals, you know, a lot, a lot of work in the sixties and the seventies, ultimately, culminating in a Nobel Prize to Brown and Goldstein in the 19 eighties focused on the molecular characterization of FH, which is largely a dysfunction of the LDL receptor leading to higher LDL cholesterol because the liver is unable to clear it away. But the liver also thinks that there's not that much cholesterol because the um L receptor is dysfunctional, that leads to an upregulation of endogenous cholesterol synthesis. And so the liver is pumping out more cholesterol cholesterol is not getting cleared and you get into this vicious cycle. And that is a, you know, critical reason why these patients are at risk for premature cardiovascular disease events. And of and a key contributor, all these observations in the notion that LDL cholesterol is a causal factor for cardiovascular disease, even among patients that don't have FH what has emerged about trying to understand the clinical utility of FH has been large scale human genetic analysis. You know, it's always been controversial about whether you do FH genetic testing because you know, the lipids are readily available and you know, individuals will generally have very abnormal LDL cholesterol. However, for individuals that have a given amount of LDL cholesterol, if you have an FH mutation is the mechanism to getting there, that substantially discriminates risk, um go going down the road. Um So one that's, you know, important scientific observation because this cumulative exposure to LDL cholesterol, one is not always captured just in cross sectional and that cumulative exposure is a critical determinant for future risk for cardiovascular disease. Now, it's obviously unethical today to randomize patients with FH to placebo versus a cholesterol lowering medicine like a statin. However, in practice, there is heterogeneity practice on how these patients are managed. So one can go back and say, you know, if individuals with FH got a statin versus not, what is the change in prognosis and individuals who get a statin um who have FH have a substantial reduction in their cardiovascular disease risk. Now, you'd say, ok, maybe that's obvious because it's been shown obviously, with patients without FH they have lower cardiovascular disease risk if you put them on a statin. But that relative risk reduction um for the general population for a state is about 20 to 25%. Um whereas here, it's about 44% reduction suggesting that individuals with FH particularly benefit from a statin. And again, we want to move to an era where we're not only predicting risk but individuals who are gonna maximally benefit from the pharmacotherapy. Now, obviously FH is responsible for a, you know, minority of patients of our patients that we see who have not had events or at risk for events and also patients who do have events. However, you know, observed independent of traditional cardiovascular disease risk factors. That if you have a first degree relative also seminal work in Framingham, if you have a first degree relative who's had premature cardiovascular disease, you're at excess risk for having cardiovascular disease above and beyond those modifiable risk factors. And that lent its, you know itself to the notion that other genetic factors not just related to cholesterol may have a role in cardiovascular disease and particularly genetic factors that are much more common because this is a a common observation that is generally observed in the population that has formed the basis for our work in genome wide association studies for a variety of traits including coronary artery disease. The prince, you know, the basic principle is you take a bunch of patients, um a phenotype, a group that has coronary artery disease and a group that is disease free, broadly profile the genome and systematically for each of those genetic variants, see if there's a significant difference or enrichment in the leal frequency in cases to control. And that can help us pinpoint regions in the genome that have an important role for coronary artery disease. What is actually really cool about this approach is that it is a broad survey of the genome. We don't use any of our biases as far as traditional cardiovascular disease risk factors. And so actually has uncovered a lot of new pathways related to coronary artery disease. Now, many of them do exhibit plyo trophy or that they, in addition to having an enrichment for coronary artery disease, often do have an enrichment for a known traditional cardiovascular disease risk factor. But that's only about half of them. Whereas the other half we actually have no clue. Um at least on first past how they influence coronary artery disease. And it's a fair amount of work to unpack those mechanisms, but those provide tractable opportunities for new potential therapeutic targets. Um However, we can take the totality of that information actually to predict risk for a patient. So then knowing that there are all these regions in the genome that influence risk for coronary artery disease. Then you can go to your patient theory, who's been similarly profiled across their genome and say, ok, for the 300 regions that are associated, how many does a patient have uh of the risk snips or risk variants for coronary artery disease? And for each of those regions, it's, you know, basically 01 or two because there are two copies that you have. So a simple form of a genetic risk score could be 0 to 600 you know, because there are roughly 300 regions. Um And that's what the earliest iterations kind of look like. We can improve the score by weight, weighting each of those variants by the amount of enrichment that they have for their fold enrichment that they have for coronary artery disease cases relative to controls. And also we can use statistical methods that can actually use the full genomic survey because there are many of these variants that are not statistically significantly associated but are but are pretty close. And then we can use a variety of approaches to um uh to model in that kind of uncertainty. OK. And so this is um kind of one of the earlier iterations. Um you know, you, you might think that family history, your self reported family history is just capturing that information, an early surprise. And an observation is not, you know, these, these concepts are correlated, you know, self reported family history is correlated with a higher coronary artery disease, polygenic risk score. However, patients that don't have a family history of, of uh uh coronary artery disease also have an increased. Also, there are many of those individuals at high polygenic risk scores and they also have an increased risk for coronary artery disease. And that suggests that, you know, family history is is complicated, it often gets at heritable genetic factors, but also other nongenetic factors that pass to families that can relate to a wide variety of environmental exposures. Um anxieties stresses dietary patterns and physical activity patterns as well. Now, if you look at, you know, increasingly larger data sets and you look at the role of genetics and kind of compare with clinical risk factors. This is in about 100 and 50,000 individuals and a metric that is often used um for the value of a biomarker and the prediction for an event is the C statistic. And these are the traditional cardiovascular disease risk factors. This is a polygenic risk score for coronary artery disease and this is kind of on par or maybe even a little bit better than each of these individual risk factors. But obviously, we look at all these risk factors together today. So this is it all together. And then the genetics gives you some additional information and it suggests particularly in the scenario where you know, these genetic score scores are out there in the wild and patients are showing up in the clinic that this may not be information that we have to just ignore because it is actually providing some additional information and risk prediction and pretty similar to the risk factors that we consider today in the clinic. Now, that was a, that that study was done in the UK. We asked recently. OK, what about in the US and um US based tertiary health care systems that are geographically distinct comprise of different patients are clinicians just so good that they're already kind of picking up that risk signal. And this is, you know, leveraging that many of our health care institutions are generating genetic information that allow us to address this question. And we find that patients who have a high polygenic risk score for coronary artery disease still have an enrichment and a similar enrichment across each of these settings for coronary artery disease kind of beyond what we're doing today. Now, if you look at the traditional risk factors on the panel here at the LDL cholesterol for people who have high polygenic risk score, they only have a modestly in modest increase in LDL cholesterol. You know, it's about like 6 to 8 mg per deciliter. So really not clinically perceptible as opposed to somebody with FH who, you know, has very high LDL cholesterol. And if you use that 10 year risk calculator, the pool cohort equations and look at these statin eligibility categories, the distributions of those scores for who would be statin eligible between individuals who have a high genetic risk versus the rest. The scores are basically the same. There's nothing. So even though they exhibit plyo trophy, there are some association with traditional cardiovascular disease risk factors turns out that those associations are just actually pretty modest. And so the overall risk prediction based on clinical factors for people at high genetic risk versus not is pretty similar. So you can't really use the tools today to identify this risk signal. Now, um you know, in the last iteration of the guidelines, there's the introduction of this concept of a risk enhancing factor. These are additional factors that further increase the risk beyond the traditional risk factors. And they can be used to help support statin prescriptions in middle aged adults. You'll see here on the left of, you know, a, a wide variety of these, you know, things, some of these include, you know, family history. They also include lipo protein, little ac reactive protein, even things like South Asian ancestry and other inflammatory conditions. Now, we, we've established a risk signal that is actually very similar at minimum to each of these. And then we said, you know, because many individuals who are kind of in this intermediate range are of clinical risk where you would use. The risk enhancing factors are kind of often already prescribed statins. They're all considered statin eligible. However, many of those patients are obviously also not prescribed statin. So we said if you looked at that borderline intermediate range patients who had high genetic risk were not already prescribed a statin and then look at the totality of all patients who don't have cardiovascular disease. We estimate through this approach through this guideline supported approach that about 4% of patients may newly be recommended for a statin. Uh based on this sort of combination of clinical factors and genetic factors. What is a new concept or at least over the last five years that we and others have advanced is that, you know, this is a continuous distribution, individuals at the tail of a distribution have a much higher risk. The further and further you go. Actually, that is a major reason why the European Society of cardiology says that everybody who is a middle aged adult should get um a lipo protein little a screen because the individuals who are, you know, very high, which is a very small number of individuals would have a very high future risk for cardiovascular disease. Now, in a paper that we had about five years ago, we observed that individuals from the top fifth percentile. So one in 20 had a risk that were similar to patients with FH and that's about one in 300 individuals. Uh We have further improved our, you know, genetic risk scoring and we say, you know, for that same threefold risk. Now it's about one, almost one in five individuals who have a risk that you would only be able to determine based on genetics that has a similar risk related to FH a group where, you know, the singular factor is a major reason why we would promote pharmacotherapy. And you know, a major challenge with these scores that's not distinct from other biomarkers is that they perform differently across different ancestors and self reported races and ethnicity. However, with a lot of a lot more training data and non European populations and new methods, we actually substantially improve the performance of these scores that they um in many scenarios, not in all scenarios do perform similarly across each um individuals of African ancestry do have uh uh uh uh do, do you have a risk signal related to these polygenic risk scores? But it's not as good, that's actually pretty similar to most other traits and whether that's an expectation or not, that's, you know, ongoing work to assess out. Now, there have been some examples of what happens when you return this patients that the information back to patients and clinicians. This is uh done by colleagues at the Mayo clinic where they took patients in this borderline range, uh intermediate risk clinical range. And then, you know, gave them information regarding their clinical risks, but also gave their information regarding their genetic risk using an earlier form of the polygenic risk score and found that, you know, one when people were communicated their risk, their LDL cholesterol went down. Um and that I think is a testament towards risk communication um as far as optimizing risk uh profiles, but the individuals who had high genetic risk were more likely to have even lower LDL cholesterol. And you'd say, ok, you know, maybe that makes sense, you know, we give statins to people who are at high risk. What is pretty interesting when we've gone back to these completed clinical trials and then examine what is the relative benefit of this of statins among people with high genetic risk. We see a similar interaction as we did with individuals with FH and taking one step back when people have done this kind of analysis with clinical risk factors like diabetes, smoking hypertension, that relative risk reduction of about 20 to 25% for a statin is the same across those groups. It's just that those groups have higher absolute baseline risk for future risk for cardiovascular disease. So the absolute risk reduction is further magnified because it's the same relative risk. They're just starting at a higher risk. Now with the high polygenic risk score, they're starting at a higher risk, higher absolute risk, but the relative risk reduction is further magnified. Now here down to about 46%. So that further magnifies the absolute risk reduction and further shrinks the number needed to treat the number of individuals you'd have to treat with statin to prevent a future cardiovascular disease event. And again moves the needle beyond just simple risk prediction. To say this is a biomarker that may actually help identify that statins would benefit you. A recent observation is that this may actually have a role in predicting recurrent cardiovascular disease events. Still controversial controversial about the concept of using uh predictors for recurrent events because we are all pretty aggressive at um managing cardiovascular disease risk factors when people have had cardiovascular disease events. However, there are increasingly new medicines, increasingly more expensive medicines. In the last lines, there is this concept of using recurrent risk predictors and subs setting individuals to have very high risk atherosclerotic cardiovascular disease to help justify further driving down Elvi cholesterol in those scenarios. And interestingly, the top in that strategy is a polygenic risk score for coronary artery disease. And I'll tell you when we did this analysis or started this analysis, that was not something that any of us had expected. However, in retrospect, when you look at the traditional risk factors, as I said, we're all very good at constraining all of those to normal or in some cases less than normal like a LDL cholesterol. So as far as predictors, those at the population level become less important. Now, other things that are not managed, like obviously coronary artery disease, polygenic risk score that helps better identify individuals at recurrent risk. And then you can see some of these others and that may actually have a role not only in the allocation of more expensive medicines, but also in the design of future coronary artery disease clinical trials. Now, this last concept I'll leave you with is that a unique feature of genetics un unlike many of these other risk factors is that it is one available early in life. And if it's gonna be abnormal, it'll be abnormal early in life. Whereas many of these others like diabetes smoking, these are, you know, not features for most of our patients that are happening in childhood. These are usually manifested in middle age um across that spectrum. So here there's the possibility even before the onset of these risk factors, we may be able to use this as a singular biomarker to identify individuals at higher risk later in life. Now, if you look, you know, really across the age range of middle age, because that's where you know, best research infrastructure is to address these kinds of questions. And you look at the genetics versus the clinical factors for patients who are younger, you have a higher predictive capabilities of the of the genetic risk score relative to the clinical risk score. And then over time, the genetics becomes less important and the clinical risk factors become more common and they become more important in predicting cardiovascular disease events. And then if you look for the events that are happening earlier in life, very hard to pick those up with clinical risk factors. If clinical risk factors are very abnormal early in life, that's obviously a very powerful predictor. It's just doesn't happen as common. And so more of the events you can determine based on the genetics and it suggests a really separate use case, you know, you know, if it's available in middle age is a risk enhancing factor, but a separate use case that if available early in life and very abnormal, it may be a really important predictor of premature events that is not being captured by our current uh risk scores. And what's the importance of identifying risk early in life, particularly as it relates to LDL cholesterol, you know, when you lower LDL cholesterol, the amount of time that you lower LDL cholesterol further magnifies the benefits related to LDL cholesterol lowering. This is a beautiful study, you know, looking at completed statin clinical trials and for every year a participant is in the statin clinical trial, the relative risk reduction continues to increase and you can kind of map this out over time. And the, you know, ultimate test of this is based on human genetics, you know, perturbing, uh LDL cholesterol early in life by just, you know, 8 to 10 mg per deciliter early in life. And that lifelong that is similar to, you know, substantial lowering at middle age. But understanding the signals early on and having the substantial LDL cholesterol lowering early in life can really help address the attributable risk related to genetics. You know, whether it's f or polygenic risk score, this is kind of a Reprisal of that analysis that we did recently in multi ethnic populations. Now, as I said, you know, much of the research infrastructure to date has been on middle aged adults. You know, cohorts designed similarly to the Framingham heart study. Whereas there are some, this is another NHL B I cohort called cardiac um of patients that were ascertained early in life and have been followed at least for a few decades of their life. And among this cohort in cardia, there are, you know, not as many traditional cardiovascular risk factors that are abnormal. However, when you use the genetics, you can substantially discriminate um uh lifetime. You know, this is not obviously across our whole lifetime but long term risk for cardiovascular disease. And for many of these patients, this is all premature cardiovascular disease events that would otherwise be missed. Now, we've started to use this, you know, as you talked about John early on about synergize with clinical practice, we've started to use these scores in the clinic. Um they are available um through research means. But increasingly so there are all, there are also clinical avenues to get polygenic risk scores for our patients. As I said, I've scoped out avenues and drawing parallels to, you know, what, what do we do today that there can be a responsible use for this. Now, many of the scenarios today is where genetics is already being generated either by our patients through direct to consumer means and they're coming back with reports or they're part of um health care associated bio banks and they're often part of research studies and these reports are being generated back. So they're actually not. And so most of the times, at least in my clinical practice, if I'm seeing that it is coming through means where I am not actually ordering the test, but then interpreting in the context of the patient and their clinical risk factors. So I'll wrap up here. Um So you unique about polygenic risk scores, you know, only needs to be obtained once um and it's already available for many of our patients, there is a use case as a risk enhancing factor for middle-aged adults. It's well aligned with current guidelines may be used for um high recurrent cardiovascular disease risk. And that may help us distinguish how low to target for some patients who have cardiovascular disease. How low to target their LDL cholesterol may be helpful for um uh clinical trials. Um I'm particularly excited about this as a very early risk signal in identifying our patients early in life. Because I think that group we are certainly under treating and really only recognizing once we're kind of far along and often even using imaging tools to say, OK, let's wait until after sclerosis has arrived. But you know, there may be patients that we want to do this really early on. And then ultimately, you know, for prima for especially for primary prevention, you know, it's all about thinking about action, ability and risk communication. And so we are really building the foundational evidence to help with that kind of shared decision making. All this work is obviously not done in isolation with extended group of collaborators and just highlighting the, you know, talented folks that I have the privilege of working with um for this kind of work. All right. So I think we have some time for questions, really, a great scientific merit and a high clinical relevance, really enjoyed that talk. We're going to have some questions now and I just wanted to let the folks at home know that they can go to ask us questions. They can go to two different sources. They can go to pole dot com online. That's pollev dot com and enter and then you can respond to that activity. Or you can, you can join us by tweet by text, sorry, by text and just text the bay to the number. 37607. That's 37607, text to bay and your message and then you can ask us a question. Well, I'll get started. I got some clinical questions for you. I guess I'm a clinician and I'm interested in getting a polygenic risk score for my patient. You've shown us that there's a great benefit to the polygenic risk score. Letting the patient know about that is powerful because they're more likely to take their medicines. Their cholesterol goes down more. It also is helpful for the physician. They may want to be more aggressive in the treatment of the patient. And thirdly, you might want to start treatment earlier. So clearly, the polygenic risk score is useful. But two questions, where do I get it done? And are there differences between polygenic risk scores that are developed that are offered by different commercial vendors? What's your choice? Where do you go? Yeah, I, you know, I I hate to call out specific vendors because I do think this is, this is evolving and you know, we are very much involved in the development of polygenic risk scores. But really most importantly, the um um uh the equitable approach to polygenic risk scoring is I think particularly critical and important to us as is for the equity part is ensuring that we have the maximally performing genetic risk predictor across all groups. And that, that representation across all groups is critical. Now that overall approach is not necessarily solved. We are I think running um first uh with this technology before solving these important key aspects. Now, what I mean just to be a little bit more granular is that um you know, the genetic architecture for coronary artery disease in most traits have largely been well profiled in individuals of European ancestry and relatively less well in individuals of non ancestry. That is a major driver for why genetic risk scores perform best uniformly in individuals of European ancestry versus not that concept of differential performance. As I said earlier, is pretty is is not specific to genetics is that there are systemic reasons why that occurs in genetics and we're working towards its performance. So one, you know, when individuals do report polygenic risk scores today, they also try to um identify genetic ancestry, which is a a mathematical exercise in mapping often to continental ancestry using reference populations. However, increasingly so in the US, individuals don't cleanly map to continental ancestry. And that concept of continental ancestry is also continuously evolving. And so it doesn't often make sense these days to arbitrarily bind people. And so people are often admixed. So represent a mixture of genetic ancestors as well. And this doesn't mean that they're all like genetically distinct. This represents a very, very small fraction minority of the genome. You look across the human population. We're all like 99.5% genetic similar. This is just a small part of genetic ancestry. However, we leverage these kinds of genetic variation for risk prediction. So they're very important in the synthesis of these polygenic risk scores. So the long winded way of saying the reporting of this still has some complexities and that's why it is a little bit of a moving target and what, you know, we are working towards um trying to figure out how best to report these scores. So there's some companies and these are kind of Google companies. There is, I think the variation in the reporting and there's variation in the performance. However, uh the performance overall in many of these platforms are much better than what they used to be. I'd say like even five years ago, five years ago is a little bit of the wild west. And you would not always understand many of these companies today are often um benchmarking their scores, often in publicly available data sets like the UK Bio Bank and working on performance and others. So we have a good handle on performance and a good handle on reporting. So I think those that are out there that you can clinically order are reasonable. Um And ideally, most of most genetic testing will also provide some additional data on the approach that they have taken as well too. I think it would be, you know, if, if a patient was not of European ancestry, I think understanding how that's being represented back to patients is also important to understand. Um the, you know, as I said, a lot of our patients get direct to consumer genetic testing. Now there are some FDA cleared genetic tests that are allowed, you know, primarily 23 million has gone through this. And so there are some snips as it relates to pharmacogenomic pharmacogenomics and carrier testing that are suitable. However, the polygenic risk work part today is at least not FDA cleared. There are lots of complexities because these often represents like, you know, thousands, hundreds of thousands, if not millions of markers. Um so that FDA part um that is direct to consumer is not totally suitable. So a patient comes to you with this information, that is a poly genetic risk for from a direct consumer testing company. You can't technically that is not appropriate to use that on face value as a clinical test that may kick off a downstream clinical test. Um But as I said, there are a few of these options. However, I think over 2023 2024 there are continued to there will be more options out there as well. Your point about being surrounded, but these are kind of considerations. Your point about the difference in polygenic risk score between different populations, different ethnicities is a really good one. And clinicians have long been aware of certain differences in coronary disease between populations. And you alluded to one of those differences earlier in South Asians, South Asians, coronary disease is different. It looks different on an angiogram and tell us a little bit about that, the difference and perhaps a genetic basis for it. Yeah, this is, you know, obviously uh near and near this this concept. Um and it's, and it's, I think really interesting that South Asian ancestry is called out as a risk enhancing factor. It's interesting because, you know, the field of medicine has been trying to take out, you know, self reported race, ethnicity um as risk factors because they often obviously are, are complicated notions that are marking a wide variety of things often reflective of nongenetic factors, but may have some genetic factors that are contributing the major challenge from um a public health standpoint is that if you say that it's genetic ancestry, the risk of doing that is you say, OK, it's something necessarily inherent and it just kind of is a glaring example of the knowledge gap and not really getting at actually what the root issues are. Major challenge to actually um examining this is research infrastructure. There are very, very few cohorts, relatively of individuals of South Asian ancestry, particularly in the US. Most of this happens when there is globalization or westernization. This is also happening of the subcontinent as well. But overall, you know, in in the in the world, one in four individuals is of South Asian ancestry. But if you look at it in the US for research infrastructure, at least in the traditional epidemiological cohorts, it's only like a few 1000 individuals that are kind of in these epidemiological cohorts. A key message of these, you know, genetic studies is, you know, you obviously need representation from a wide variety of individuals to better understand the genetic drivers for cardiovascular disease. And if there are any alleles that are uniquely more common among certain groups versus others, and that also transcends obviously to nongenetic studies. So one such approach that the US has been taken to improve representation in these kinds of studies is the all of us research program. This started from President Obama's Precision Medicine Initiative. This is supposed to be our new Framingham aiming to get about a million individuals that has much better representation from populations that are not really um uh well represented in legacy research studies. And you know, we recently did an analysis comparing the initial phases, about 300,000 individuals set um or where data is available today. And it pretty closely mirrors the US census, meaning about 50 54% of European ancestry, actually substantial increase in the uh relative amount of individuals who identify as black or African American. However, the others Hispanic Americans, Asian Americans actually less than the US census that includes individuals who are South Asian. So that that's an even smaller fraction even though that is a growing group in the US. So we, you know, a lot of that research infrastructure is actually based in Boston at the Broad Institute where I'm at now, we have used that infrastructure to now develop a new cohort, new cohort infrastructure targeting specifically South Asians because we have this persistent knowledge gap and we need to develop the research infrastructure to address this knowledge gap. Now that um study will be actually launching in 2023 and is similarly designed that there can be in person recruitment but also remote recruitment and that we will be collecting information in addition to genetics that is relevant to that population that is not being captured in other research studies, including, you know, cultural practices, acculturation, um what people eat lifestyle practices and in manners that you know, make more sense in that community. So lots of outstanding questions. Um a you know, a glaring um tragedy that you know, many of us cardiologists see on a on a definitely on a daily basis and something that you know personally motivates me mainly on atherosclerosis, affecting the coronary arteries are the genetic variants different for peripheral arterial disease, carotid artery disease aortic disease. Yeah. Yeah. You know, so peripheral artery disease, atherosclerosis shares close overlap with coronary artery disease. But recently, you know, we did a pretty large genome wide association study for coronary for peripheral artery disease. And we did this primarily in the million veterans program. So in the va there's, you know, as, you know, high burden of cardiovascular disease and cancer and often, um, uh, atherosclerotic cardiovascular disease. So lots of it, uh veterans have peripheral artery disease and that allows us to look at this question, you know, that uh MVP has grown in size, you know, previously, then we did an analysis in about half a million, but it's, it's almost up to a million veterans to date, which is amazing. And so when we did this genetic association analysis, we did find a handful of regions that are associated with both coronary artery disease and peripheral artery disease. But it turns out there are actually lots of regions that are exclusively associated with peripheral artery disease. And as you know, peripheral artery disease is a highly morbid condition, high degree of morbidity, high degree of mortality, not always explained by high comorbid coronary artery disease. So I'm really excited about this approach as a tool to really address the novel risk drivers that are related to peripheral artery disease related. Late, you know, a couple of months ago, we published something on calcific aortic valve stenosis, you know, which many have likened to at least some mechanisms related to atherosclerosis, influencing the aortic valve. Now, calcific erotic valve disease is very, is very attractive to use these kinds of discovery tools for that condition because we have no demonstrable therapies that are efficacious for calcific erotic valve disease. As it stands today, it's a procedural disease. You know, patients either get catheter based therapies or surgical based therapies and we have no clue how to, you know, prevent its onset and identify patients who are at risk. So we actually found this is building up other work that has been out there. But again, using the large number of veterans that are in the million veterans program and many other data sets that have calcific Erin foel disease. And we find that there are many heritable variants in the germline genome that influence risk for calcific eric foel disease. And a minority of them are related to coronary artery disease. And so this, you know, provides now the trajectory to identify now, what are the novel drivers of co of calcific A valve disease one that has been described before. And you know, I mentioned it a little bit in the talk is lipo protein little A, an LDL liken protein that has a role in atherosclerotic cardiovascular disease has a role and calcific erotic val disease. And there are drug development programs that have been spurred based on the human genetic observations. Now, we have increased that catalog and have described many other attractive therapeutic targets for medicines to treat calcific erotic valve disease. And those that would be very specific to bowel disease that are distinct from atherosclerosis. Excellent. I think Dr Lai has a question. Um uh I think the uh I believe that your polygenic risk scoring is very powerful to predict this coronary heart disease. Uh but your, your scoring system is using mainly with genetic formation, right? But uh um the epigenetics is also a major contributor of the coronary R disease. So do you speculate uh in the future or incorporate the epigenetic information somehow uh in your risk scoring? Yeah. No, I think um so uh a a pro for the genetics, it's you know, available early, that information is gonna be unchanged, its relative contribution will change over time. Now, it's clear that there are a wide variety of other um molecular factors that are either downstream genetic products or reflect um an individual's personal history and exposures to a wide variety of things that may influence cardiovascular disease risk, epigenetics has proven to be one such example. Now, the practical challenge of using epigenetics today is that, you know, these are tests that are not widely available in human data sets to understand and prognostic capabilities as well as generalize in different contexts. That is a unique feature of the human genetics. And then also for these um biomarkers that change over time, I think we also need to understand in launching data sets how they change over time. What are the reasons that they change over time, how that influences prognosis? Um But as I said, you know, as the technologies continue to improve costs decrease and there are also substantial in um interest from NIH and NHL B I and better profiling these other omics. We'll continue to understand what is the clinical opportunity. And a lot of, you know, there are practical cost consideration as well because I think it is endless, all the different things that you could conceive of that influence risk for cardiovascular disease. And we would need to try to understand, you know, when would we order these tests if abnormal, how would you follow this test with the, with the genetics? That's kind of a one stop. And it's a singular bio marker that you can theoretically actually use to calculate polygenic risk scores for for anything. So I think, you know, as the costs go down and the opportunities continue to expand and the use cases are better delineated. This may be information that we just kind of have for everybody. And then we just kind of go back to that information and try to understand how it's useful for that particular patient. But I am excited about the prospect of using these other molecular features because it's it's key that they are also strong predictors for cardiovascular disease and have complimentary opportunities. But what you're doing is just amazing and uh really uh of merit and clinically relevant with polygenic risk score, you've shown us that it can improve the care of patients because you can start treatment earlier, you can start treatment more aggressively. Patients are more likely to take their therapy and respond to it. And it also is giving us insight into the mechanisms of coronary heart disease. I do see a problem however, with it and it's, it's not a problem you're making, it's really, it's a failure of the pharmaceutical industry to be able to keep up with this fast moving science. I mean, it's just you're segmenting the patient population into finer and finer groups, but we don't have therapies for, for those specific groups that are specific, we don't have specific therapies and I don't think the current platform, small molecule platforms are gonna be able to keep up, but there is a platform that might be able to keep up with your science. And that is RN A therapeutics and we have a center for RN A therapeutics here. And as you, as you well know, a therapeutics are a bit like writing software and you know, you can quickly come up with a construct that might be useful for a particular patient. Um What do you think about that idea about a more kind of a genetic basis for our therapy to keep up with the genetic segmentation of the phenotype? Yeah. No, I think the um the way that we have thought about action ability today is kind of uh leveraging what what is available and that, you know, clinical trial data sometimes is available. Sometimes sponsors and investigators have put in the hard work of one paying for and consenting and genotyping participants um to then be able to go back and do these sub group analysis. And that has been done for a handful of statin clinical trials, P CS K nine inhibitor trials who actually interestingly see the same interaction. Now, this is not something that anyone would have hypothesized based on a polygenic risk score that brings together lots of different features together. And you know, my guess is we're working on this, lots of people are working on this. Maybe you can create a better pharmacogenomic predictor that is not just based on, you know, information that is devoid of the clinical trial information. Interestingly, this has been also tested in other clinical trials like CE TP inhibitors and Aspirin clinical trials. Um And in those, you know, there isn't an interaction in those, it's just like kind of another factor that increases the absolute risk. And it suggests that there may be very specific strategies. One, you know, at least we have lots of different strategies to lower cholesterol, but that may not be the only approach. And as you highlight, you know, these are all genetic features and genetic drivers. So having genetic based therapies, particularly durable therapies like RN A based therapies may be important strategies through this and it may not be that, you know, you treat everybody with a high polygenic risk score, the same, maybe the mechanisms to get to a high polygenic risk score will be slightly different from one versus another. And we're working on better trying to tease that out. And I think as you do that, then you can think about bespoke strategies that are kind of earlier in life and kind of durable strategies. And you know, that is being advanced for LDL cholesterol, that may be a strategy in the polygenic risk score current forum. But I think one can build off of those concepts here. Wonderful. Well, we'll be following your work with great interest as you proceed. Very exciting. Thank you so much for joining us today. We've come to the end of our hour. So we'll say goodbye to you, Prade. Thank you for an amazing talk and keep up the great work. Thank you all for joining us today at the cutting edge of Cardiovascular sciences. Thank you. Thanks so much.
