Dr. Amitabh Pandey discusses the pros and cons of using genetic testing in predicting atherosclerotic cardiovascular disease.
Okay, so we are moving on to our last talk of the day. Uh uh it's one of those concepts of saving the best for last right doctor Pond. So I'm super excited to hear this final talk by Dr Amitabh Pandey. He uh incorporating apologetic risk scoring into my clinical practice is something I am definitely looking to get more educated on. So um I started off my day today, I start off the day to day as far as introductions, um introducing one of my mentors Dr trip in and now I get the book and the day by introducing a cardiologist who I've personally had the privilege of serving as one of his program directors and mentors in our cardiology fellowship program. DR Panda is a superstar. Not only was he one of our top clinical fellows to come to the program, but he was also one of our key l two scholars, kl two scholars which is a coveted NIH funded K award through the scripts research, transitional institute with a primary focus on genomics and atherosclerosis. He has a growing clinical practice here at scripts but still focus much and focuses much of his time on research. Early research endeavors have included areas of cellular inflammation and heart bear which have given away to his recent research efforts, efforts on atherosclerosis and genetics. He's also actively pursuing clinical applications of ai a really growing field. So dr Pond, he currently serves as an assistant professor of the molecular medicine of molecular medicine and the Department of Molecular Medicine at the Scripps Research Institute and is one of our general cardiologist who also works closely helping our fellows better use the epic E. H. R. To navigate the research. So it is with great pleasure that I introduce Dr Amitabh Pond who will be presenting the utility, apologetic risk scores and primary prevention of atherosclerotic cardiovascular disease. So, thank you very much. Thanks chris for that very kind introduction. You know I was just thinking about the other two talks that proceeding with Dr Einhorn Dr griffin and I was you know thinking you set the bar so high that I have to climb up to that now. So next time maybe put me ahead of everyone. So there's a lower bar for people to climb over here. Um So you can see here my disclosures. Um So you know genetic risk is something that we always try to get an idea of from our patients but really don't have a good way of quantifying that currently in our abilities to check with patients. So we have our current guidelines that we look at a variety of different factors trying to assess the most accurate way that we can their risk for advanced coronary artery disease or cardiovascular disease in general. And one thing that we always try to ascertain from that is family history, in a sort of indirect approach to identifying genetic risk factors that might be present in patients. The other problem that comes up is that we have patients of various ethnic diversity who come to see us in clinic. And so, as you can see here, there's a small adjustments and modifications to those primary prevention sort of guidelines that we have because the risk does change with different ethnic, um, ethnicities and different patients of different ethnic groups. So we have to keep that in mind as we go forward to identify their risk and adequately treat them for primary prevention. All right, so, I really want to take just a minute here to um identify how we're gonna look at genetic risk. And so for that, I'm apologizing in advance, but we got to do a little primer on genetics here. So hopefully I won't try to put you guys to sleep too much with this. Um so basically what we do is we know that individuals have very diverse background um from a genetic standpoint and we know everyone has regions of the genome that are preserved and regions that are prone to variation. And the question really becomes, how do we really go about identifying these genetic variants? And then in turn identifying which of these variants are clinically important for a specific disease state of interest. So we can do the first by looking at what we call genome wide association studies, Orjuela studies so that we can identify the variations that are present. And you can see that there are single DNA letter swaps that are uh occurring that are known as single nucleotide polymorphisms, or snips that are identified by the letters V. Here in the genome. So you can see a group of patients who are sitting here who have their full snip profiles completed their operations that have a variation that are conserved across multiple different individuals. And then there's ones that are more unique and diverse amongst those individuals. And trying to figure out how these actually convey risk for a specific disease process is where politics risk scores come into play. And so what we do is we complete genetic testing for monitoring risk currently for many diseases. So these include cystic fibrosis, um sickle cell disease, familial hypercholesterolemia. These are caused by single mutation in a single area that results in the disease phenotype and genotype. These are more so the holy grail of genetic diseases. Because you have a single point that you can go after and targeted therapies can then be generated because of those known variants that occur more commonly. Unfortunately are the diseases such as coronary disease that have actually a very diverse and apologetic risk variation that occurs in those patients. And so what we actually see is that they're influenced by a larger number of genetic um apologetic variants that occur throughout the genome. And they actually also encompass environmental and lifestyle factors typically when looking at that. And so using those US studies, we can identify where these changes are occurring and then try to gain insight into how to best go after those areas. So if you look at the diagram here, you can see that the snips uh for coronary disease are very diverse. They occur across the genome on multiple different chromosomes and multiple, multiple different areas of those chromosomes. So there is no specific area or arm of the chromosomes that are typically targeted. And so when we do these across patients, we typically get what's called a fingerprint or a sniper rate of these patients. And so that becomes very useful when we can compare them and identify areas that are more or less important. And so coronary disease, as you see here is very complex. In terms of its genetic variation, there is currently about 60 different genomic variants that occur across the chromosomes for C. A. D. And is dispersed everywhere. And so what we need to do is figure out a way to actually identify that and use that in an efficient manner. And so when we have these population based studies have been done in the past hundreds of thousands of patients that have been studied for a particular disease state, comparing those with the disease against those who do not have a disease, we start to learn more information about these patients and possible genetic factors that can come into those. When we do genetic sequencing on these patients, we are able to find those fingerprints those snipper raised and the specific profiles of those snips so that we can go back and identify and wait how those snips are actually contributing to risk. And so when you use these, that can then in turn be developed into apology at risk or because we have the weighting of all the individual snips. Now, it's important to remember that apologetic scores are a relative risk. The reason this occurs is because you are comparing two different cohorts to a single individual. So, if you look at the plot here, um of the normal distribution of uh apologetic risk force individuals who are at low risk you're going to take one individual comes to see you in clinic and this patient is going to be compared against the two cohorts of disease versus not disease that encompassed hundreds of thousands of patients. So comparing the one individual to those two cohorts will give you a relative risk similar flee for those an intermediate risk and those at high risk. Conversely oftentimes We'll hear about absolute risks in genetic risk. And those are things like Braca one mutations where we know that a specific Variation in a particular area of the genome will result in a disease state. So, Bracco, for example, has a 60-80% correlation with disease states and it's irrespective of population basis, purely based on that particular mutation. So that's important to keep in mind for apologetic risk force. The other thing that's important to keep mind is who the data is based upon. So current apology risk scores that are out there are all predominantly based off of a limited set of genomic data that we have from large cohorts of patients. And so if you look at the plot here, about 80% of those patients are of European descent. And so we have a much smaller representation of other ethnic minorities. And that becomes important when we're dealing with patients in clinic to be able to identify their risk based off of the prs. Um and and having to really incorporate that information in there as well when we're giving those risk assessments. Yeah, so the concept of apology at risk score has been circulating for a number of years now. Um and researchers, Both cardiologist as well as computational biologists, have been doing this for quite some time, but it really wasn't until 2018 that these scores were shown to have potential for broad scale clinical use. And so a group out of mass general looking at data from the U. K. Bio bank looking approximately at that point the UK bio bank was about 300,000 patients that had genetic and phenotype information available. They looked at four different disease states. So I looked at atrial fibrillation, they looked at diabetes type two, they looked at inflammatory bowel disease and they looked at breast cancer and they looked purely a genetic factors to identify if they could find a significant fold increase which was determined as a threefold increase in risk state. And predictor of that disease state based off of their apologetic risk scores that the algorithms that they had developed. And so this was really interesting because if you look for example at diabetes Um you find that there's a 6% chance of the population having diabetes based off of the apology at risk for for those in the high risk um numbers percentiles of the apology at risk for That compares to only 3% of the same high percentile greater than three full change and disease state in inflammatory bowel disease. So it's very specific to each disease state again, based on the data that we have. Right based off of what snips have we identified, how those related to the disease state and how are those influencing our recognition of genetic risk for those disease states. When they looked at this for corn er disease, first thing they found that that same three fold increase in the population was found to be at approximately 8% of the population. So taking all comers across the board, if you look at just genetic variation and look at those who have a high apology at risk for 8% of individuals are going to be fitting in that hype risk group. That is outstanding because these are patients who otherwise you would not have recognized as having any sort of cardiovascular risk that are coming in. And this speaks to the talks that we've heard earlier today about those patients who otherwise we're thinking are doing well. But that high genetic risk is something that we haven't been able to measure before. When we look at that threefold increase going across in the Parliament. Risk for calculations, we find that that score of about 7080 on a scale of 100 is where that risk really comes through another study that was done out of um the group at MGH, a different group actually um looking at three different uh large healthcare networks and the patients that were banked in those healthcare system. So partners out of boston University of pennsylvania and Mount Sinai system and they looked at the Association of the polytechnic Risk Force for coronary disease and it was significantly associated across all these different health care systems. Additionally, when you look at the data from these, we find some interesting trends that come out here. So the first thing that we did, our assuming that they did was that they looked at the accumulative rate of coronary disease from primary prevention of these groups. And we find that if you look at the high risk, apology at risk for a group versus the low quality at risk for group, there really is no difference amongst those. And the part where it gets very scary is the fact that you have patients who are eligible for status. And when you're looking at eligible for status for patients that you would treat with statin therapy based on guidelines. And you divide those then based off of the ch guidelines or the US preventive task force guidelines, You find that when you subdivide those groups um for stand eligibility and look at their policies at risk scores. Those with high political risk scores have no significant differences compared to those with low. So these are patients who you otherwise would not have started on statin therapy but have another risk factor that hasn't been considered in that. And that unconsidered risk factor is something that is potentially causing harm to these patients because it's a data point that we're not evaluating right now. And so that becomes something that we really need to change in the way that we practice medicine. Again, looking at another group mostly and colleagues who looked at reclassification of standard cohorts are used in cardiology. So they took two large cohorts, the eric study and participants in the Mesa trial and look at reclassifying those patients based purely on their genetic risk scores and policy at risk force. And you can see that there's no real change in the reclassification of these patients. There's a little bit of a question as to why this occurs based on the data that we saw from the UK bio bank studies. I think in part, there's two major factors here. One is that you're looking at significantly smaller groups of patients that were comparing here. UK Bio Bank was about 300,000 patients that were compared here. You're looking at about 10th of that. And additionally, Mesa is very ethnic diverse trial that was done specifically for that reason. And so 75% of the participants in Mesa would follow into about 20% of the current genetic data that we have. So I think that is playing a role in some of the changes that we're seeing here. But you can also see from other studies that are being done that there are data that are being reclassified. So son and colleagues here look at reclassification of another cohort of data and they found no significant change when you re classified it purely based off of the traditional risk factors versus just apology at risk for classifications for these patients. However, where it becomes very interesting is when you re classify them based not only on just one or the other criteria, what you actually combine both of those criteria. So if you take a combination of apology risk scores plus the classical traditional risk factors that we look at clinically, you actually get a significant change in re classification of those patients as compared to what you get from either one of those two classification models independently. So that brings us to the next point of how are these apologetic risk scores really best utilized? So I think ideally and where we need to get in clinical practice is being able to combine both the clinical risk factors that we currently look at and are doing for all patients who come in and using that in combination with polishing risk force for patients to identify those that are at high risk or intermediate risk with certain other clinical factors. And use this in combination so that we can really get a good understanding of where the risk for totality of primary prevention is coming out for these patients in terms of cardiovascular disease. And also by doing more of these studies will be able to get a threshold that can be achieved so that we can start therapies for these patients. So I want to shift a little bit towards therapies now that we're talking about that. So this is actually some really interesting data that came out of some of our pcs K nine inhibitor trial. So this is data from four years where they looked at um the participants who were in the trial and calculated their genetic risk based on apologetic discourse. And so when they did that, you can actually see the patients without multiple clinical risk factors or without high genetic risk were about the same. And that makes sense. Those who don't have high political risk factors or those that would not have high genetic risk, you would expect to follow along the same trend line, which actually does have when you incorporate either high multiple, high clinical risk factors or but they don't have high genetic risk factors, you get a small diversion of those lines, but it's roughly about the same. And so this again makes sense that, you know, one risk factor is there? We have the traditional political risk factors, but the genetics is not adding to that. When you look at just high genetic disease, this is where it gets really scary because you have a divergence of these lines and the number needed to treat for these patients, at least with PCSK nine inhibitors actually significantly reduces in that state. And so, looking at these for those patients that were identifying with high genetic risk, we can actually treat them with medications that we have currently to mitigate some of those risks in our clinical practices. Additionally, when you look at data from odyssey, again, looking at PCSK nine inhibitors, this goes to something that dr griffin just mentioned in his talk a little while ago, we're looking at traditional risk factors that are used in clinical practice today. LDL levels, framing hammer scores LP little A levels and you have patients who otherwise you would say are doing wonderfully based off of those metrics. But when you subdivide those groups who are, for example, LDL less than 100 mg per desk leader and you find that there's a significant percentage of those patients who have high genetic risk. Those patients aren't really being protected to the full capabilities that we have today. And so this becomes a recurring trend that you see throughout all the different classical risk factors that are listed here. But there is a significant portion of patients who have high genetic risk that we're basically missing at this point in time, because we're not really getting accurate quantitative representation of their genetic risk. So, coming back to coronary artery disease, Um, we know that from the data that we have that's been done in the past that the range of 70-80 as a minimum cut off for high um apologize at risk scores and high risk is um pretty well corroborated through multiple different data sets interestingly. It also converges on data for onset of a cumulative risk for coronary disease. So if you look at the patients and the blue line who are high um PRS risk scores of 80 to 100 Their cumulative risk starts at an earlier age as compared to those who are in the lowest quintile at 20-40%,, apologize, risk for which occurs significantly later. So this is something that we're missing as well and something that could gain benefit from in our clinical practices by utilizing apologize, risk force. This last plot is actually something that was very, very interesting when I first saw it come out and so this basically took multiple different individual risk factors for coronary disease and looked at their prediction ability in predicting patients who are going to end up with coronary disease. And you can see that their primary risk factors are grouped about the same. There's a little bit of variation for some of these. For example, cholesterol versus smoking and whatnot. But by and farther in the same range. And then when you look at politics, risk scores taken in isolation without any other clinical history or other metrics, they fall in about the same range as well, a little bit better than some of the other metrics. But in about roughly the same range. When you take these risk factors in aggregate and then add Polident risk course to that aggregate risk, you actually get a significant jump and the ability to predict who's going to develop coronary disease. And this is something that I think is prime for us to utilize in clinic because we have ways to be able to do this now through a variety of different methods and we'll touch on that here coming up. So I want to discuss before we get into really good news, A little bit of the downside might end on a bad note. Right? So limitations for our current use, apology at risk for our present. You know, it's not a perfect system yet. So we have a requirement of genotyping for snipper raise for evaluation of apology at risk force. The cost has been coming down significantly over the last few years, But it's still about $100 to complete this at minimum, which is not a trivial cost. And oftentimes it's not covered by insurance. There's also differences in sequencing for this particular snip array as compared to genetic testing and genetic studies that are done for other disease states, such as those that we do for familial hypercholesterolemia or dilated cardiomyopathy. These those require separate type of sequencing that's done for those, so that becomes a little bit of a challenge to complete in that respect. Additionally, there's the data that's currently there to generate the apology at risk scores is overwhelmingly from european ancestry because of databases that we have currently. So this does skew your apology at risk for um and the applicability of these policies at risk scores. And we have to keep that in mind when we're seeing our patients and additionally, we have to have better understanding and more widespread knowledge of these policies. Risk force a so they can be utilized, but also they can be understood by physicians and practitioners so that they can give these explanations to patients who can be invested in their health care. So let's talk a little bit about an example of where we can use these and how we can go about doing it. So there's a group over scripts research who are collaborators of mine who have really pioneered being able to bring these into clinical practice. The effort is led by Dr ali Turkmani over scripts research. Um He has really done some great studies and develop an app that is really wonderful and being able to use uh apology at risk for in a clinical setting. So what they did was they looked at the H. A. C. C. Guidelines and without any other therapies. Um you know, as we know the number of media treat for stand uh benefits has been seen to be 15 1. Um They looked at apology to restore risks for classifications and stratification and found that those who are at higher genetic risk for coronary disease based on PRS scores also have um a biochemical and molecular uh signal that indicates they would have increased statin efficacy in these patients as well. So when they looked again, divided these into the genetic risk categories, The high genetic risk group actually had a number of me to treat down 1-20 as compared to the 1-50. So, pretty impressive. They then went ahead and created a first class app that was called the Meijin Rank app that use some really innovative data visualization to be able to uh convey this information to patients in terms that they would understand. So all this talk about threefold increases and the really the math and statistics that go into politics at risk force was put on the back end and the visualisation was done in a sort of three color scheme representation. So red was you're a high risk yellow at an intermediate risk and green at a low risk so that patients can easily understand this and see this. And in addition they were also given their actual score so they could discuss with their physicians about how they could go about mitigating some of the risks that come from this. Um They've you know, done several iterations of the app to improve. It is now present on multiple different platforms and they've incorporated different aspects of risk factor modifications that can be employed between patients, physicians in a discussion of how to best counteract these genetic risks. And so some of the things that they've incorporated into the app or are working to do um currently are a physician results portal, inclusion uh integration into electronic health medical record, amongst other things that you can see here. And then they're also working on a new trial that they're about to launch the Peppers trial where they're going to look for broad adaptation of apology at risk scores. And so what the key points of this new trial that they're going to undergo are uh, you know, does apology at risk for information change physician clinician data making abilities and decision making process? Does the PRS information change patient perspectives on interventions? Are they more willing to do certain things because they know they have a higher genetic risk and our health care outcomes improved as a result of these interventions. So, um, that's uh what we have. I think I'm missing my conclusion side. But basically what we found is that we have, you know, significant um evidence that apology at risk scores are really um, you know, moving towards a field where we can have them incorporated into what we're doing every day, the exact way that we're able to develop this um and incorporate this into clinics still need some help. But I think the best strength of these really comes from the ability to be able to utilize these in conjunction with our traditional clinical risk factor models that we have in play and combining those is really going to be the way going forward to integrate genetics into those risk models. Thank you. That was that was amazing. Dr Pandey. I was really looking for a very uh concise breakdown of how about apologetic risk scoring. I I definitely am interested in getting more this better into my practice and that was a really nice um talk that really helped me out a lot. So thank you very much.