It's 10 a.m. I want to start on time and welcome everyone to the Horizon Society Research Fellowship Award Session. My name is Natalie. I'm a professor of medicine at Baylor College of Medicine, and it's my pleasure to chair this session with Dr. Hazi from Metro Health at Case Western, right? And I also want to give a shout-out to some of the current committee members who's also in the audience. If I call your name, please show your hands or stand up so the rest of the people can see you. We have Dr. Mona Raphael from Ohio State and Dr. Jeff Seffitz from Beth Israel. And Dr. Masahiko Takaki from Japan and Professor Randovasky from Ohio State, again, and Dr. Derek Dostal from University of Utah. Did I miss anyone? Oh, Dr. Luke Hoffman from Netherlands, yes, all the way back. All right. So today, our session will have two missions. Again, the first part is that we have three presentations from 2024 fellowship awardees. They received the award last year, so today, we'll hear their presentations and hear about the progress of their projects. And at the end, we'll give out award to the 2025 recipients, okay? Without further ado, our first speaker is Dr. Pavian Angustelala from University of California, Davis, and her research project is localization of a sodium channel regulator, intracellular fibroblast growth factors in housing and disease. Before I start, I would like to express my gratitude to the HIS for this fellowship award and for giving me an opportunity to present my work today on the localization of sodium channel regulator interacetyl fibroblast growth factor, IFGF, in healthy and diseased cardiomyocytes. So as we all well aware, voltage-gated sodium channel plays a crucial role in the initiation and the propagation of an action potential. With its activation underlies the upstroke of the action potential and its inactivation shaping the action potential duration. The distribution of the sodium channel in cardiac myocytes are not uniform, and past studies show that the cardiac dominant, NAV1.5, localizes to the intercalated disc and the lateral membrane, with some subpopulations in the t-tubes. And other NAV subtypes, including the NAV1.4 and the neuronal sodium channels, are also expressed in myocytes but at much lower expressions. And they also localize to distinct sites like the t-tubes. And these different NAV subpopulations, they were shown to have different electrophysiological properties due to a multitude of proteins that are interacting and regulating the sodium channel complex. And one regulatory protein that is of my interest is interacellular fibroblast growth factors, or IFGF, that is bound to the C-terminus of the sodium channel close to the calmodulin binding site. And why do these IFGF, consisting of IFGF 11 to 14, expressed in the cardiomyocytes are not well understood? IFGF can modulate the sodium channel function, as shown in the past studies, knocking out IFGF 13 in mouse, causing a hyperpolizing shift in the steady state in activation, as well as changing the fraction of the slow inactivation. IFGF 13 can also regulate the trafficking of the sodium channels to the membrane. So in this study, we aim to investigate the localization of various sodium channels and IFGFs in mouse ventricular myocytes in both healthy and deceased cardiomyocytes, and to also study their electrophysiological consequences. We perform immunofluorescence staining in isolated mouse ventricular myocytes. And as shown by other negative control here, when we only have the secondary antibody with no primary antibody, we observe very low unspecific signal. With NALT1.5, we see a high intensity at the lateral membrane and the intercalated disk with some in the T-tubules. And when we add the specific NALT1.5 blocking peptide, this immunofluorescence staining is almost completely abolished, suggesting a good specificity of the NALT antibody. A careful examination of the pixel intensity along the x-axis within different regions shows a higher peak intensity in the intercalated disk compared to the lateral membrane and the T-tubules. And this higher peak intensity is also broader in distance compared to the T-tubules. At the lateral membrane, when we look at the intensity along the y-axis, we also observe a higher peak intensity at the cell edge compared to the intracellular region. For other NALT subtypes, we see a clear striations for NALT1.4 and NALT1.1. And there is no peak intensity, no higher peak intensity at the intercalated disk and only similar peak compared to the lateral membrane and the T-tubules. Although expressed at a much lower expression, we also observe the localization of NALT1.8 at the intercalated disk. The immunofluorescence staining of IFGF shows low expressions of IFGF11 and 12. And that is in agreement with the mRNA expression level. And in mouse ventricles, the predominant IFGF is IFGF13. And we observe the localization of IFGF13 at the intercalated disk. For IFGF14, we see striations and no increase in the intensity at the intercalated disk. To study the disease condition, we use transgenic mice that over-express CAMK2 delta C to mimic the conditions found in various heart disease, including cardiac hypertrophy, dilated cardiomyopathy, and heart failure. And so to compare to the Y-type mouse myocytes, we see a redistribution of the IFGF13. The SIFGF12 and IFGF14 shows similar patterns as in Y-type mouse. For sodium channels, in CAMK2 over-expressed mouse, we still observe the localization of NALT1.5 at the lateral membrane and the intercalated disk, similar to Y-type, as well as the striations for the NALT1.4 and NALT1.1. For NALT1.8, we see an increase in the nuclear localization. So to summarize our findings so far, in the Y-type mouse myocytes, we observed the localization of the IFGF13 at the intercalated disk that is missing in the CAMK2 over-expressed mouse. And for NALT1.5, we see a localization at the lateral membrane and the intercalated disk in both Y-type and the CAMK2 over-expressed mouse. As I mentioned earlier, knocking out of IFGF13 causes a hyperpolarizing shift in the steady state inactivation. And past studies also showed a similar shift when CAMK2 delta-C is over-expressed in myocytes. This suggests the possibility that the loss of IFGF13 in the intercalated disk may have electrophysiological consequences. And so to study that, we measure the sodium current in mouse matricular myocytes in both Y-type, shown in black, and in the CAMK2 over-expressed mouse, shown in red. We also want to test whether there is a functional contribution from the non-cardiac sodium channels and so we use the TTX and the NALT1.8 specific blocker, A803467. We measure the TTX response curve in both Y-type and the CAMK2 over-expressed mouse, but we did not see any significant change in this response. The applications of the NALT1.8 specific blockers does not significantly reduce the peak sodium current in either Y-type or the CAMK2 over-expressed mouse. So to conclude, in our study, we observed the localization of NALT1.5 at the intercalated disk and the lateral membrane, and also at the TT boost with lower intensity. And at the intercalated disk in Y-type mouse myocytes, the NALT1.5 may co-localize primarily with IFGF13 and not other IFGFs. And the loss of IFGF13 at the intercalated disk in CAMK2 over-expressed mouse may have some electrophysiological consequences. And so for future directions, we plan to look at the co-localization between the IFGF and the sodium channels, in particular for IFGF13 and 14. And then we will look at expressions of IFGF and sodium channels in CAMK2 over-expressed mice and study the electrophysiological properties of the sodium channels. And with our collaborator, Dr. Gene Niborn, we plan to look at the localization of IFGF and sodium channels in IFGF13 knockout mice. So with that, I would like to thank my PI, Dr. Burse, and the lab members for their help and support, in particular Dr. Hegyi, Dr. Koh, and Dr. Ginsberg, and our junior specialists Megan and Shannon, and my undergrad, Leah. And this work was also supported by HRS and AHA. And thank you for your listening, and I'm now ready to take any questions. Very nice work. So overexpression is always a little bit dicey because it might have non-physiological effects. So what is your sense about how overexpression is causing this redistribution, and do you have any sense about whether that's a specific effect or just an effect of just overexpression? Yeah. I agree that with the over, especially with the CAMK2, that it can also regulate multiples of other proteins, so it's hard to pinpoint. So we're thinking that maybe we can apply specifically some AIP for the CAMK2 blockers to see whether that could be related, if there is any change after we apply the AIP in the white type compared to the CAMK2 overexpress. I think some studies also show that IFGF13 could also be phosphorylated, and that could change the interacting partner of the IFGF, so that could also be another possible reason. Thank you. Well, very nice work. So I'm wondering, have you also checked the... So, I think, like, that would not be my work, but, like, I think other people, like, in the lab. And I think it's more, like, industrial agents in the T-tubules, like, you will see that's where MK2 local is, too. Thank you for sharing this interesting work. Curious, what is the impact on the persistent or late current in these models and the role of FHF or FGF? The persistent current? Yes, and then, typically, in the model here. Yeah. So, I think MK2 over expressed miles, we also kind of see the increase in the late sodium current. And so, I think in our previous FGF13 knockout, I don't kind of look carefully at the late sodium current, but there have been some other studies that they kind of mentioned that there is an increase in the late sodium current, so it would be interesting to incorporate that into our FUJA work. Thank you. Thank you. And just as a follow-up, you showed that there is no difference with the selective blocker, whether in wild-type or non-wild-type. So, I'm just curious, like, what is the impact on the persistent or late current in these models and the role of FHF or FGF13 knockout? So, I think in our previous FUJA work, I don't kind of look carefully at the late sodium current, but there have been some other studies that they kind of mentioned that there is an increase in the late sodium current, so it would be interesting to incorporate that So, I'm not sure if you're talking about the NAT 1.8 specific blocker here? Precisely. So, yeah, I think it's still, like, very controversial, like, what's the role of the NAT 1.8. And so with this, it could be, we think that it could, instead of, like, contributing, like, to the sodium current, may regulate the sodium, like, NAT 1.5 current instead. So that, I think, one possible action of the NAT 1.8. That we not sure. Thank you. Thank you. Just a quick question. Very nice work. What do we know about how this relates to some more common disease forms? Do we have any sense of how uterine disease, ventricular disease, that we see more commonly? So, interestingly, like, in the human, like, the predominant IFGF is not IFGF 13, but it's IFGF 12. And they, like, some studies, they see that there is mutations in the IFGF 12 in Brugada syndrome patients. And also, some mutations in the sodium channel 1.5 that they, that can cause the reduction of the IFGF binding. And also, I'm not sure if it's in the long QT or the Brugada syndrome patients, but has some electrophysiological consequences. Thank you. All right. We'll go on to our next speaker. Our next speaker is Yang Zheng from University of Arizona. And the title is Mechanistic Roles of Cardiac SK Channel in Hereditary Calmodulopathy. Good morning, everyone. I would like to first thank the HRS for the fellowship award and the opportunity to talk about our study. Today, I'll be discussing the cardiac SK channel and our recent discoveries on its regulation. We don't need to be reminded of how vital and beautifully complex our hearts are. But when the heart rhythm got disrupted, the consequences extend far beyond the individual. Cardiac arrhythmia is not just a personal health crisis. It drives up the healthcare cost of the whole society. Annually, there is about 3 million sudden cardiac death cases in the US alone, with cardiac arrhythmia playing the major role. And among this, atrial fibrillation stands out as the most commonly diagnosed cardiac arrhythmia and one of the most dangerous. It increases the heart failure risk by four times and doubles the mortality. And the problem is growing. As you can see from this figure, the AFib-related death rate is rising steadily. And it's estimated that by 2030, there will be over 12 million US adults living with atrial fibrillation. That will be a significant burden to the patients, families, and the healthcare system. In order to treat cardiac arrhythmia more efficiently, we have to understand the molecular machineries that generate each heartbeat, and that is the ion channel. And here comes the one we want to focus on today, the small conductance calcium-activated potassium channel, or the SK channel. It belongs to the family of calcium-activated potassium channel, and it's closely related with voltage-gated and inwardly rectifying potassium channels. In the following three slides, I'll be introducing the SK channel and its two working partners. Here I'm showing you the cryo-EM structure of the human SK2 channel just published last week by Miao Jiang Group. As you can see from this figure, the SK channel is formed by four identical subunits that form a central pore. Like voltage-gated ion channels, SK channels are gated solely by intracellular calcium, and they control the action potential repolarization. SK channels sense calcium via a ubiquitous calcium sensor, calmodulin. Previously, SK channel upregulation and dysregulation have been linked to arrhythmia. And notably, SK channel inhibitors are currently in clinical trial for atrial fibrillation, making our study very timely. As we mentioned, SK channels sense calcium with the help of calmodulin. Calmodulin is a 148-amino acid protein that is highly conserved among all eukaryotes. It has two lobes, the L-lobe and C-lobe, each of which have two EF motifs that bind with the calcium. So each calmodulin binds with four calcium. Calmodulin is the master calcium sensor and regulates many key cardiac ion channels and proteins, including voltage-gated sodium channel, L-type calcium channel, reodinor receptor, and the SK channel. So it's central to the EC coupling and the function of a cardiomyocyte. Calmodulin mutation can cause lethal cardiac arrhythmia, and the patients are usually very young. So in the first half of the presentation, I want to talk about how some of the human calmodulin mutations affect the SK channel function. But before that, let me introduce another partner of the SK channel, PEP2. PEP2 is a critical phospholipid in cell membrane that regulates ion channel and signaling pathways in cardiomyocytes. As you can see from these figures, PEP2 has been demonstrated to directly modulate many cardiac potassium channels, including QR2.1, IKS, and the HERK channel. And growing evidence suggests that PEP2 is an important co-factor for the SK channel. What does that matter? Well, PEP2 dysregulation has also been linked to cardiac arrhythmia. So we have to understand PEP2 in order to dissect the SK channel function. Plus the PEP2 metabolism itself can serve as a potential antiarrhythmic therapy. Our overall goal is to gain insights into the mechanistic roles of cardiac SK channel in arrhythmias. We hypothesize that the SK channel gating is a tightly coordinated process involving both calmodulin and PEP2. And we perform the study with multiple experimental approaches, such as patch clamp, confocal microscopy, MD simulation, and in vivo electrophysiology. In this slide, I'm showing you the SK currents recorded from expression system HEK293 cells with patch clamp experiment. Apamine is a specific SK channel blocker, and we used 10 nanomolar here to review the currents. The black and red curves are recordings before and after apamine. So the difference in between will be specifically the SK currents. We tested mutations in the L-lobe, C-lobe, and both lobes of calmodulin, which significantly lower the calcium binding affinity. We used them as positive controls to show the inhibitive effect on the SK currents. And then we tested human calmodulin mutations that are associated with arrhythmias, such as CPVT, IVF, and Lankerton syndrome. And we found that those mutations was able to reduce the SK current. We then performed confocal microscopy to measure the SK channel expression to see if the decrease in currents is rooted from trafficking deficiency. On the left side, I'm showing you examples of two mutations, and on the right side is the quantification. We found that none of the mutation has a significant effect on the SK channel expression. So it's more likely that the inhibitor effect of the calmodulin mutation are through a direct regulation on the SK channel function or the SK channel activity. To elucidate variability caused by difference in calmodulin expression level, we engineered fusion protein that directly linked the SK channel with either calmodulin wild type or calmodulin mutation. This approach will ensure the binding stoichiometry between the SK channel and the calmodulin, enabling a more precise comparison. We first tested that those fusion proteins behave as expected. And then we did experiments on the human calmodulin mutations that were mentioned earlier. And now we have a more definitive confirmation that those mutation can decrease the SK current. Then we employed Nalkin mouse model to study a more physiological relevance of our findings. We did in vivo electrophysiological characterization in the F90L and N54I heterozygous. Giving those mutations association with CPVT and IVF, we used isoproterol to challenge the cardiac function. And judging by heart rate and the PR interval, it looks like the heterozygous are responding – not responding as well as the wild type. And we also found significance in the heart rate variability. One approach to measure the heart rate variability is to calculate the RSMMD, which is the root mean square of successive difference in RR interval. We found that in the heterozygous animal, the RMSSD didn't increase as expected. And all of that suggests that the heterozygous animal's heart, they have impaired adaptability to stress. And that parallel clinical observations. In addition, we found calcium mishandling in those heterozygous. The calcium transient was reduced, recorded from isolated cardiomyocytes. And there is a significant increase in the arrhythmia susceptibility in the Nalkin mass. We are currently performing more experiments such as confocal microscopy and patch gland in order to study the physiological relevance of those mutations to SK channel. Please stay tuned in for more interesting results. But now I'll switch gears a little bit and talk about the PEP2 regulation on the SK channel. We first verified the PEP2 effect on the SK channel with a previously established optogenetic system. In this system, the transcription factor CIBN is fused with a plasma membrane anchor. And then its partner protein CRY2 is fused with mCherry and PEP2 5-phosphatase. So with blue light, the CIBN and CRY2 will form dimer, recruiting the PEP2 5-phosphatase to the plasma membrane. Therefore, the PEP2 will be dephosphorylated and we're creating PEP2 depletion. And we found that removal of PEP2 with blue light was able to decrease the SK current, confirming the PEP2 effects on the SK currents. Then we want to study the mechanism. I won't go through much details here, but we generated a homology model of the human SK channel and we verified the stability and conductivity of the simulated channel. Then we did MD simulation to study the PEP2 effect. Here you can see the figure B showed us the inter-subunit sub-bridge formed by the arginine 395 and the glucamate acid 398. Interestingly, when PEP2 comes into the picture, it can form other sub-bridge with the arginine 395 and disrupt the inter-subunit one, which is crucial for the SK channel activation. And we think the sub-bridge is how PEP2 regulate the SK channel. So we did functional study to verify that simulated result. We created mutation on the SK channel, which would disrupt the sub-bridge between PEP2 and the SK channel. And we found out when we removed that sub-bridge, the regulation or the PEP2 effect on the SK channel was completely abolished, confirming our simulation. To clarify, we think that PEP2 form sub-bridge that disrupt the inter-subunit sub-bridge of the SK channel and that will open the intracellular gate of the SK channel and promote ion conduction. And here is how we think it works. So the calcium bind with the calmodulin and the calmodulin will go through conformational change. And then the unload of calmodulin will bind with the SK channel. And the PEP2 interaction with the SK channel will further promote conformational change of the SK channel that help the channel to activate. So PEP2, calcium, and calmodulin collaboratively activated the SK channel. We conclude that the gating mechanism of the SK2 channel is a highly regulated process involving precise coordination of calcium, calmodulin, and PEP2. And with that, I want to thank my mentor, Dr. Nipavan Chiamivimavad and Dr. Abneezer Yilma and our lab members and collaborators, especially Dr. Noman from Vanderbilt for the calmodulin knocking mice. And I want to thank my finding sources and thank you for your attention. Thank you very much. So great presentation. Thank you very much. I have a question for you on your patch clamp studies showing very high level of SK2 current. That's in the HEK cell, right? Which one? In the very beginning, the current HEK cell. Yes. Then you show no phenotypes when it's expressed in the mouse. Am I right? I just want to make sure. The problem with HEK cell study and calmodulin is calmodulin has three different types, one, two, and three. They all produce exactly the same protein. So in HEK cells, if you mutate one, and obviously SK2 doesn't work because it relies on calmodulin to gate its current. But in the whole heart, the other two calmodulins are completely normal. And so they produce normal calmodulin overwhelms the mutant calmodulin, and therefore it's very hard to demonstrate the phenotype. So I was wondering if your mouse has any phenotypes, and how do you explain this? Is my thinking about right? Yeah, I agree. I think the calmodulin, the different calmodulin genes are definitely, they are very different than the HEK cells, and that's kind of the reason we want to study it in the animal. At baseline level, actually, we don't have a lot of phenotypes. It's only when we challenge it with SO we see a little bit difference, but it's not like clearly VT or VF. And we are planning to do PCR to see how much of the calmodulin are mutated and how much of the SK channel are connecting with the mutated calmodulin. And I'm sorry, what was the question? No, I was just thinking about the absence of phenotype in the mouse model. It's probably because there are still a lot of normal calmodulin in the cell. Yes, yes. Thank you. Okay, thank you for your presenting excellent results. So you showed that mutation of the calmodulin reflects the heart rate variability in this slide. How do you think about the mechanism of the effect? I think the mechanism can be really complicated. Here we just want to see the contribution of the SK channel to anything happening in the heart. But the calmodulin, actually, as I mentioned, it regulates so many channels, like the L-type calcium channel, reactor receptor, and the calcium channel. So it's like, and it's an overall knock-in, it's not like a specific cardiac knock-in. So it can be like a lot of sources, it can be like from the nervous system, and it can be from other proteins. And right now, without more evidence, I won't try to dissect where the heart rate variability is exactly from. insights in the knockin mice. Yeah, that's actually a very interesting aspect and we are planning to do experiments on that, but right now we don't have any, yeah. Fair enough. Thank you for the comment, yeah. I have a question for you actually on this slide. So why do you see structure changes, the fractional shortening and ejection fraction? How long did you stimulate the mice with ISO and what is the reason of the changes, do you know? We stimulated, so the ISO, we did injection of the ISO and then we recorded for 10 minutes. The fractional shortening, so the calmodulin mutation we're seeing like the most difference here are from the F90L and the F90L we know that it will reduce the calcium binding affinity of the calmodulin. So I would expect like the heart is not responding as well to any kind of stress at the well top, yeah, that would be my answer. As I mentioned earlier, if I want to study like molecular mechanisms of all the ECG parameters I probably need to go through other channels and see the effect of the other like machineries, yeah. Yeah, no, it's very interesting. Every time you see changes you just try to work more, that's good. Yeah, yeah, exactly. Just to point out that the baseline, the fractional shortening is not different. So this is a response to isoproteinol's effect, so it's not like the ejection fractions go down but their ability to respond to stress is different. So maybe something related to calcium handling. Yeah, there has to be, it's calmodulin, yeah. Interesting. Right. Okay, so we'll go to the next one? Yes. Okay. So our third presentation is by Videshu Yogaswaran? Ben. Okay. And she'll be presenting Atrial Cardiomyopathy and Adverse Cardiovascular Outcomes in the UK Biobank. Oh, and I forgot, from, oh, University of Washington, sorry, I forgot about that. So, hi everyone, my name is Vid, I'm one of the ECHO Fellows at the University of Washington, and as mentioned, I'm going to be talking to you about atrial cardiomyopathy and adverse outcomes in the UK Biobank. So, as just mentioned, and as many of you are already familiar with, atrial fibrillation is the most common sustained cardiac rhythm disorder. And over the last few decades, there has been an increase in both the incidence and prevalence of AFib. The reasons behind these changes are multifactorial, there's been an increase in our life expectancy, and there's also been an increase in the burden of chronic medical conditions. Globally, there's also been a rise. If we look at some statistics from the Global Burden of Disease Study, in 1990, there was an estimated 30 million individuals living with atrial fibrillation. And in 2019, those numbers had doubled to 60 million individuals. Even then, those numbers are still an underestimation, because so many cases of atrial fibrillation go undiagnosed until patients present with symptoms or come in hospitalized for a stroke. Atrial fibrillation is associated with an increased risk of heart failure, ischemic stroke, and death. In a large meta-analysis, we've learned that treating AFib can decrease the risk of stroke by 64%, and many of our clinical studies have shown us that if we treat atrial fibrillation, we can decrease the risk of cognitive decline and heart failure. Therefore, it's essential that we develop new paradigms to identify those who are at highest risk for these complications. That's where atrial cardiomyopathy comes into play. Emerging research suggests that atrial cardiomyopathy is an independent risk factor for atrial fibrillation. Broadly, the consensus guidelines define this as adverse structural and functional changes to the cardiac atria that are capable of causing clinical manifestations. So if we think about the same risk factors that can cause atrial fibrillation, hypertension, obesity, sleep apnea, these can also cause cardiomyopathic changes. Those changes can lead to mechanical dysfunction, can contribute to a procoagulant state, and can lead to thromboembolic complications like stroke. They can also lead to electrical dysfunction and fibrosis that can lead to ectopy and atrial fibrillation. They can also dilate the atria and cause valvular dysfunction and heart failure. Several large epi and clinical studies have shown us that ECG and imaging measures of atrial cardiomyopathy can independently increase the risk of AFib and its downstream complications, even in patients who have no subclinical or clinical AFib. Now, most of these studies use insensitive measures, which is why we're limited, or measures that are challenging to reproduce. This includes echocardiographic LA diameter that only captures the left atria in one dimension and echocardiographic LA strain measures. Now, cardiogram MRI can help us overcome this limitation and is the gold standard for atrial volume and function assessment. CMR has a high spatial and temporal resolution, and in head-to-head studies that compare CMR and echo, echo actually underestimates LA size by almost 15 milliliters. Additionally, as some of you have seen clinically, CMR allows us to characterize tissue and scar. Early research evaluating these parameters have been performed in the multi-ethnic study of atherosclerosis, or MESA, a large cohort study based out of the United States. There, they observed that CMR-derived left and right atrial parameters were independently associated with an increased risk of AFib. In fact, at baseline, individuals that went on to develop AFib had a 17% larger left atrial volume measurement than those who did not. And across two studies, right atrial volume, left atrial volume, and LA ejection fraction and strain were independently associated with an increased risk of AFib after adjustment for clinical parameters. Now, even then, these studies were limited. They had low event numbers, and they had low sample sizes. Most of them were performed in cohorts of less than 500 individuals and have limited our ability to understand atrial cardiomyopathy as a generalizable risk measure. So we wanted to evaluate this further by determining the associations between high-quality left and right atrial measures of cardiac structure and function and the risk of not only AFib but also AFib-related complications, namely heart failure, ischemic stroke, and dementia. To do this on a large scale, we used data from the UK Biobank. The UK Biobank is a large longitudinal cohort study that began recruitment in 2006 in the UK. And to date, they have recruited more than 600,000 participants. The UK Biobank imaging sub-study began in 2014 and has performed more than 65,000 CMRs with a repeat study currently underway. Our main exposures are left and right atrial volume and function measures from CMR. Our outcomes, atrial fibrillation, ischemic stroke, heart failure, and dementia, which were identified from nursing interviews, EKGs, discharge codes during hospitalizations, and death records. And we used Cox regression with adjustment for clinically relevant risk factors. One of the research scientists in our group at the University of Washington Cardiovascular Health Research Unit, Jennifer Brody, applied a neural network that was published by Bayh and colleagues in 2018. This network takes the cardiac MRI images as input, learns the features via series of convolutions, and finally predicts an image segmentation, which can be used to automatically pull the measurements. We use this algorithm on those 65,000 available cardiac MRI to automatically obtain those measures in a matter of weeks. The key measures of interest are in B and C, so the two- and four-chamber views of the left atria and the four-chamber view of the right atrium. But when we went to look at those automated measurements, we learned that some of them are not compatible with life. You can't really have an LA min of 0.16. In fact, most of those early measurements were less than one milliliter. But at this time, despite the increase in AI algorithms that are being used to derive these images, there remains a shortage of guidance on how to approach actually quality controlling these images. So we went through the lowest and highest CMR measurements in 10 measurement increments, then up by one milliliter or 1%. Again, these are thousands of images. Until each group of measurements was about 50% accurate. To me, that meant, is this image in plane and is it actually capturing the atria? For example, some of the right atrial images I was quality controlling were actually capturing the aorta in a five-chamber view. After that, we randomly evaluated 50 measurements for each parameter and found that our quality control metrics were 95% accurate. This is an example of our LA exclusions. After quality control, here are the means and standard deviations for our six parameters of interest. In the middle two columns, I show the sex-specific ranges from the Society of Cardiac MRI Guidelines. Now, those guidelines use mostly smaller studies in specific racial and ethnic groups. And in the last two columns, I show the percent abnormal in our cohorts. So for the most part, even though those ranges were derived from very small cohorts that vary by race and ethnicity, may have used different MRI machines and different planes, the majority of measures were still in the reference ranges. So after that quality control method and excluding individuals with baseline AFib and those without available follow-up data, we ended up with a cohort of 51,693 participants with high-quality measures of left and right atrial structure and function. Our mean age was 65, it was 48% male, predominantly European, 97%. And the cohort itself was relatively healthy. Less than 1% had heart failure and only 2% had a prior history of MI. Our median follow-up time was four years, during which 964 individuals developed AFib, 266 had an ischemic stroke, 365 had heart failure, and 72 went on to develop dementia. And for the results of our main analyses, we observed that both left and right atrial measures of structure and function were associated with an increased risk of not only atrial fibrillation, but also ischemic stroke and heart failure. So all of these hazard ratios are reported per standard deviation increase in unit. And all volume measures were indexed to body surface area. Significant associations are bolded for all these analyses. So, for example, each increase in left atrial index minimum volume per standard deviation increase in unit was associated with a 55% increased risk of atrial fibrillation over time. And notably, we also observed a novel association between left atrial index minimum volume and the risk of new onset dementia. Now, these associations were relatively unchanged with secondary adjustment for Townsend Index, CRP, and ECG parameters of atrial cardiomyopathy, which were P wave duration and PR interval. Of note, the left atrial index minimum volume and dementia associations were no longer significant in this analysis. Similarly, for additional analyses for ischemic stroke, heart failure, and dementia results, they were relatively unchanged after we adjusted for intercurrent AFib, which is AFib that's developed during the study period. In fact, the association between left atrial index minimum volume and new onset dementia was slightly stronger. We also tested for effect modification and observed important sex-specific differences in risk. For example, for every increase in left atrial index minimum volume, women had a 17% increased risk of AFib and an 18% increased risk in heart failure compared to males. Now, these results were similar for our right-sided measures. We also looked at categorical associations with outcomes by evaluating these measures as quintiles. Here, participants in the highest LA quintiles had more than a three times increased risk of atrial fibrillation compared to those in the reference. When we looked at ischemic stroke, the highest quintiles of LA measures had a 1.5 times increase in risk compared to our reference. And for heart failure, the highest LA quintiles had more than a 2.5 times increase in risk compared to the reference quintiles. And although we see significant associations with our right atrial measures, similar to the prior analyses, they're not as pronounced. And with dementia, we did not find any significant associations when we evaluated these measures as quintiles. So overall, in this largest study to date evaluating high-quality atrial structure and function measures and their association with longitudinal outcomes, we found that both left and right atrial cardiomyopathy was independently associated with an increased risk of AFib and AFib-related outcomes, even after adjustment for intercurrent AFib. Notably, we observed important sex-specific differences in risk where women had a higher risk with each atrial measure when compared to males. Now, this highlights a role of personalized risk stratification in this patient population. More importantly, these high-quality measures of atrial cardiomyopathy may be of potential clinical use to screen individuals who may be at highest risk for adverse cardiovascular outcomes, especially women who, again, are at the highest risk for AFib-related complications. So despite its large size and the fact that this was studied in a well-phenotyped cohort, our analysis also has important limitations. The U.K. biobank is 97% European, and this may limit our generalizability to other racial and ethnic groups. Additionally, even though this is the most recent studies evaluating atrial cardiomyopathy in the U.K. biobank, the number of events for ischemic stroke and dementia are still relatively low, and with a median follow-up time of four years, this limits our ability to detect true associations with conditions that have a long natural history, like dementia and heart failure. And in this study, CMR was evaluated as a single point in time, so we're unable to assess how changes in these measures may actually impact these longitudinal outcomes, but again, there's a repeat study currently underway, so hoping to share those results soon. And then finally, this is still an observational study, so you can't exclude residual confounding. Currently, we're working on understanding the underlying genetic associations behind these findings and have some promising early results that we're hoping will be published by this summer. Additionally, as I've mentioned a few times, the U.K. biobank, 97% European, so it's important that we validate these findings in other racial and ethnic groups, especially in the large cohort studies. And then given the growing evidence of atrial cardiomyopathy as a novel risk factor, there's a shortage of studies that actually evaluate how this will perform as a screening tool in clinical populations. So our current work is looking at AIECG to see can we use transportable measures of atrial cardiomyopathy in real-life clinical populations. So thank you to my main mentors, Dr. James Floyd and Dr. Gnadzima Kuhn, my co-first author Jennifer Brody, the University of Washington Cardiovascular Health Research Unit, my many co-mentors, collaborators, prior mentors, and especially thank you to the Heart Rhythm Society postdoctoral grant for funding my work and helping start my career as an early physician scientist. I have a quick question. Very nice work. It's very important also, I think. So for the heart failure, do you look at both half-life and half-path? We did in this initial analysis. There's not that many events, so we weren't really empowered. There was maybe like 100 events of half-path, but we're hoping with the new update in the UK Biobank that we'll have more events next year. So 30 years ago, Moritz Alessi published a landmark paper in circulation showing that atrial fibrillation begets atrial fibrillation. And the idea here is that once you initiate AF, this causes structural and functional changes in the atria, which increase the risk for AF, which then cause more atrial cardiomyopathy. So I think an interesting question for you to think about is, what's causing what here? You know, the way you're ascertaining this for the UK Biobank, it's not clear whether patients with atrial cardiomyopathy got there because of AF, which hadn't yet been detected, which then led to more risk of AF and more cardiomyopathy or the other way around. So what are your thoughts on the causal relationships here? Yeah. I think that's a great point, and thank you for bringing that up. I think it's true that we have several studies that have shown us that atrial fibrillation can lead to atrial cardiomyopathy, but in our preliminary genetic analyses, we found that the variants that are associated with atrial fibrillation are distinct from the variants that are related to atrial cardiomyopathy, even though there is some overlap. So I do feel like there's two different things going on. Yes, one can cause the other, but it also seems like one can independently also lead to other things. And actually, I think one of the main limitations with the cohort studies is most of them have relied on clinical AFib diagnoses, but in MESA, they have some Zyopatch data, which again, it's only 14 days of monitoring, but I think in some analyses, they've shown that with subclinical AFib, in a very short amount of time, patients who have atrial dilation still independently have a risk of these AFib-related events, even if they have no evidence of subclinical AFib. Okay. It's a great presentation, but your presentation has a lot of mean and standard deviation in it. David Ouyang of our institution last year published a paper on single-point echocardiogram reading by the AI algorithm to predict atrial fibrillation. And we used that algorithm. We were able to predict atrial fibrillation in the near future, like one week. So you mentioned AI. Have you put your data to let AI to learn, and maybe it will come up a method of using this data to predict future atrial fibrillation? Sorry, can you repeat it? Your CMR data, have you used, can the AI read the CMR and predict if the patient has atrial fibrillation in the future? Yes. So far, in our prelim data, one of the research scientists in our group trained an AI ECG model to predict these atrial measures, and it does show that they were able to predict AFib outcomes, actually, almost as well as the direct CMR measures. You said AI ECG model, right? How about the CMR, the one that you are presenting? Oh, sorry, you're saying can the CM, we haven't used, we haven't trained AI to train a CMR model, but I think that would be interesting. Thank you for the interesting presentation. I actually have two questions. One is regarding the atrial myopathy. If you find the atrial myopathy, what are your thoughts on preventing the occurrence of AF later on? Yeah, it's interesting. There's a few studies, mostly in clinical journals, that show that there was a hypertension trial a few years ago that says if you're on Losartan compared to not being on Losartan, you actually had a decrease in LA size, but there haven't been a lot of studies that actually look at high-quality measures. There's also some animal studies that show that ARBs can decrease the levels of atrial fibrosis. Most of the work in this space appears to be preclinical, unless something's changed in the last few months, but it does seem like a lot of the medications we use for GDMT, and there's also some promising results with GLP-1s, can also be used to decrease atrial cardiomyopathy. It's just unclear on how this will actually translate to long-term outcomes. Thank you. And you mentioned that CMR can be used for detection of fibrosis and scarring in the atria. Is that done in this UK Biobank database? No. Their protocol was a 20-minute scan, so they don't have that available, but it is available in some of the other cohorts like MISA. Thank you. Yes, congratulations on this presentation. I have a question. So usually when we talk about atrial myopathy, we talk about it as a global phenomenon. So the right atrium and the left atrium are undergoing remodeling. My question is, can we isolate or did you try to isolate patients who had left atrial dilation without the right atrial dilation, for example, and assess which was more important for heart failure development and AFib development in the future? Yeah. We did adjust for each of the parameters, and I think that's an important question and something I'm trying to do over the next year. But when we adjusted, we did find that left atrial measures have a higher association with these outcomes, which I think is not unexpected. But I think the interesting thing to note is there's so few papers out there focusing on the importance of the right atrium, but the right atria were still important. It was just less than the left atria. Okay. Thank you. Thank you. I have so many questions I could ask, but I think we're running out of time. One thing I was going to say was it was very interesting how your initial AI screwed up, which is, I think, a lesson about AI. But the other thing I was going to mention, and we'll go on because we have to introduce the awardees, should we not be talking about atrial fibrillation at all? Should we just be talking about atrial myopathy? The CERT trial, if you recall, showed that there are patients with pacemakers that have stroke and then six months later have AFib. Are the two, should we be thinking about, rather than looking at people who are getting AFib and then thinking about CHAZ-VASc and anti-coagulation, should we be screening people for left atrial myopathy first? It's great work. All right, so the last part here is we're gonna present the fellowship winners for the following year, this year, that applied this year and received the award. So I'll mention the clinical recipients first. The first is Margarita Pujol Lopez from University of Arizona College of Medicine, and the research title is Left-Sided Electrophysiology Study of Resynchronization Mechanism in Left Bundle Branch Block, Learn Personalized CRT Randomized Clinical Trial. Here? Thank you. Oh, maybe quickly, okay. Yes. And you can just, yeah, stay here. And then the second clinical award will go to Jean-Jacques Nubia, University of California, San Francisco, and this is Sleep Disruption and Discrete Atrial Fibrillation Episodes. I think we should be clapping, shouldn't we, for both of them? Thank you. Oh, did the rest, okay. All right, so then for the basic awards, we have Ankar Rijav Shah from University of Utah, Salt Lake, and the title is Evaluation of Innovative Concepts for Left Bundle Branch Pacing Area, Area Pacing Procedures. And the final award is Min Jing Yang, Cedars-Sinai Medical Center in L.A., Effects of Mirabigran on Sympathetic Nerve Activity and Physical Activity in Postural Orthothetic Tachycardia Syndrome POTS. Thank you. Thank you all. We're very excited to see all your work in the future. Yes. Congratulations, and thank you everyone for stopping by. We'll take group pictures.

Horizon Society Research Fellowship

sodium channel regulator IFGF

cardiac SK channels

calmodulin mutations

atrial cardiomyopathy

UK Biobank

cardiovascular disease

electrophysiology

cardiology