testing session today. If you look at the title, Resynchronization Pacing, it's been called many things over the years. You can start with septal, then alternate site, then optimal site, and then select site, and then hisperkinesia, and now conduction system. I think we're going to get some answers to a lot of the questions that are still out there as far as which are the best patients, and most importantly, how are we really looking at effectiveness of these alternate pacing that we've been talking about for 25 years, and we're still talking about it. So without any further going on here, we'll start with our first speaker. Where'd you go? Oh, there you are. Dr. Marston, and her topic is Pathophysiology of Electromechanical Dyssynchrony, Novel Insights from Computational Modeling. All right. Great. So I have no disclosures. Appreciate the invitation. This is actually my first time at HRS. My background is primarily in blood flow simulation and tissue biomechanics applied to congenital heart disease. And I run a lab at Stanford where our primary goal is to develop computational tools and digital twin type of models that can aid in clinical decision making so that we can attempt to aid clinicians in moving from treatment plans that look like this. So this is an actual sketch from one of my colleagues related to interventional planning in a Fontan patient. And we want to move towards developing software tools such as the one on the right. So we develop software called SymVascular that enables personalized treatment planning. Typically we go through the following pipeline where we load in a patient's image data, typically C-tier MRI. We build a patient-specific anatomic model. And we can run finite element simulations of blood flow and consider traditionally that's been considering different surgical options. What you're looking at is a Fontan junction in a 3D model where we're running flow through the Fontan junction and we could consider different scenarios. But recently based on clinical needs we've been expanding our tool set towards whole heart modeling. And this has really been driven by clinical needs and requests from particularly from our surgical and EP teams at Stanford. And so now we've recently developed methods where we can use deep learning to create anatomic models of the heart for CHD patients. The picture on the left is a machine learning generated anatomic model of a patient with Tetralogy of Fallot. And that's using a signed distance field together with a deep learning network. And then we've recently implemented the physics into our solver to do active contraction of the myocardium to actually simulate blood flow in all four chambers with four valves and to simulate electrophysiology using both physics-based models and also neural network digital twin type of models that can be run in near real time. So what I'm going to do today is talk through a couple of specific examples of how we're using these tools in the EP world related to hypoplastic left heart syndrome. So of course you all know what this is better than I do. And we're really interested in kind of what are some of the EP related effects towards thinking about things like CRT in these patients. So the first thing we typically do is reconstruct the cardiac anatomy. And in this case we also reconstructed a torso model from the scans. We mesh that model so you can see on the right there we have a two chambered model with a hypoplastic left ventricle. And once we've done that together using that model together with the torso model we can place virtual leads on the chest and actually simulate a 12 lead ECG. And so you're seeing here a digitally produced 12 lead ECG from a model. And what does that model look like? It's a physics-based model. So we have a monodomain equation that basically propagates the electrical signal through the tissue. We have an ionic model which in this case is a Tantusher-Panflov model. And then we have a model that propagates the ECG signal through the chest to the lead locations. Our model includes a Purkinje system that's generated sort of automatically based on a fractal tree algorithm. And then we have a physics-based model. Of course considering it's important to understand that these physics-based models require a fair amount of computational resources to run them. And so in a recent study that really motivated the use of neural networks to develop sort of a digital twin of the physics-based model that can run in near real time. So the purpose of the slide that I'm showing you here is we wanted to be able to tune the model parameters very quickly in order to really closely match the ECG of this particular patient with HLHS. And so you can see in the table a list of conductances and conductivity as well as stimulation time. Those are needed as inputs to the model. And so one of the things we have to do in the modeling is parameter estimation. Basically we need to select the parameters to feed into the model such that we produce the ECG of the patient. And so this is almost like an inverse modeling problem. So what we did was we generated a database of 200 simulations from our physics-based models where we varied the seven parameters in the table. And that gave us a training set to use with this neural network. And this is a neural network, so without getting into all the details, but it's a neural network where we've disentangled different portions to account for both space and time. So it's a physics-inspired neural network, if you will. And so after training, we showed that this neural network model can be used very fast to do the parameter estimation and produce a very closely matched simulation to the ECG recording of this particular patient. So we're able to produce the clinical 12-lead ECG in a few minutes of compute time on a basically personal computer. Once we've calibrated the parameter, we then have a model we can use for modeling some hypothetical clinical scenarios. And so just to demonstrate this, we took that same model where we had calibrated the set of parameters and we simulated different clinical scenarios. This is really just a demonstration purpose, but of left-bundled branch block and right-bundled branch block. So you can see the outputs of a typical simulation. Once we had done that, we also wanted to ask a question about using these models as a controlled manner to test different mechanistic approaches. We wanted to look at the effect of growth on dyssynchrony in an HLHS patient. So we know that HLHS is associated with long QRS. This is a risk factor for heart failure. But the role of hypertrophy on electrical function is not entirely clear. And so our hypothesis was that RV enlargement would worsen electrical function in this patient, and that left ventricular enlargement may mitigate it. And so we ran a hypothetical scenario with our computational model. So we built a model for a 6-year-old patient with HLHS after the Fontan surgery. This time point 1 model is based on her MRI data. We identified a set of patient-specific EP parameters similar to what I just showed you. And then we grew the patient for about 6 years of time, and we were able to assess or we were able to differentially grow either the RV or the LV or both ventricles together. So this is a virtual patient, so we can do lots of things that we can't do obviously in reality. So then we could compare QRS duration and electrical dyssynchrony in these different growth scenarios. So first off I'll show you the two different leads, three different locations of the ECG simulated versus patient-specific. So we're matching this pretty well. And so after identifying the parameters that match the patient's data, we then grow the model. So what you're looking at are the three different columns represent both ventricles grown or just the LV or just the RV. And then you can see the three different leads, ECG signals, and then the activation times. Again on the top, both ventricles grown together. And then in green, just the LV is grown. And in blue, just the RV is grown. And again, representative of about 6 years of growth for the patient. And then our primary result is the plot shown on the right, which is intraventricular dyssynchrony. And so what you see is that growing either both ventricles together or the right ventricle increases dyssynchrony. But growing in isolation, just the left ventricle, which in this case is the hypoplastic one, actually reduces dyssynchrony in this hypothetical scenario. OK. So to summarize, we see effects of growth on dyssynchrony in a highly controlled manner using a computational model. And again, where we find RV enlargement increases dyssynchrony, whereas LV enlargement decreases dyssynchrony. And I've also shown you a demonstration that fast digital twins of cardiac EP are feasible by blending physics-based models with data-driven methods. And in this case, we used a large database of physics-based models to train a network that then provides a very fast analysis. And once we have that, we can accelerate analysis of clinical treatment planning scenarios in a variety of settings. So I will stop there and with some acknowledgments for funding and graduate students. In particular, we worked with Henry Chubb and Anne Dubin and Ellen Kuhl at Stanford on this work, and post-doc Mateo Salvador and student Oz Tignolari. All right. Thanks. Happy to take questions. So if you took a, let's say, a tetralogy of fall, and you take the ventricles and you put a patch in the septum, and you alter the alpha track of the right ventricle, could you then say, I want to start electrical activity from this location or that location, and how would that affect the ultimate geometry? Could that be something you could actually end up doing? Yes. So we could virtually change activation sites. We could also, in the future, apply different pacing lead locations in hypothetical scenarios. So one of the things we're working towards is, for example, optimizing CRT, either in single ventricle or tetralogy of fallot. Maybe one thing you're referring to we're working on, so a little bit out of the EP realm, but we're working on patch planning for biventricular repair in borderline single ventricle patients, and looking at sort of sizing for surgical patches. So could you do that from a patient's two-dimensional echo? Or you have, you said you got it from the MRI. We typically need to build an anatomic model from either MRI or CT, and preferably one where we have time-resolved data so that we know we're getting the mechanics right. For the EP part, we're just using a static model, so the model's not moving, but we also want to be able to link that with mechanics and functions. So in that case, we need activation and motion of the wall. So with a patient with HLHS, I could certainly go through the physics, but that math looks so easy, so I'll bypass that and jump to the clinical application. And perhaps as a means of preventing the need for CRT down the road, but with that LV model of growing the LV in a hypoplast, are there any clinical thoughts on should we be doing something to actually increase and augment LV growth in the hypoplastic leptar to prevent the RV from becoming so dominantly dilated? We have a nascent project exploring design of mechanical devices that might help induce growth in the hypoplastic ventricle. We've been exploring using those in rabbits thus far, and I don't have any results to show, so it's a little bit very kind of in the high-risk, high-reward category. But I also do think that it could be a consideration for the biventricular repair, sort of when people are making that decision about which pathway to send patients down. There could be implications on dyssynchrony down the road stemming from matrix. So you have correlated this with an ECG pattern, where really, when we're talking about resynchronization, we want mechanical contractility. So how would you then, if the ECG doesn't really reflect mechanical contractility, how else can you alter this model to really reflect the contraction of the heart, not necessarily the electrical activation of the heart? Can you alter that pattern? Yes. So we haven't gotten there in the work that I just showed you, but our goal is to link, to do electromechanical coupling. And right now, we have implemented these different physics all in the same solver. And so we can do cardiac contraction, and we can do electrophysiology, and we haven't quite coupled them together, but that's where we're headed with it. So then you would take the EP signal as input to the mechanics, and that would actually alter your activation of the myocardium regionally. Awesome. Cool stuff. Thank you for being here. Okay. Thank you. Thank you. Introduce the next one. Okay. Sorry. I was going to sneak one more in there. Oh, sure. So when you're tuning this to make a patient-specific model, you have several things, and you basically set the AI to, say, go and figure out what matches that EKG. Yeah. Do you restrain those to certain things? Are you finding that they're pretty much the same? How much variability is tolerated from that standpoint? We do constrain them to known physiologic ranges. And I think in one of the slides, there were envelopes shown for what were allowable ranges in the training dataset. And we are actively working on doing sensitivity analysis and uncertainty quantification. I don't have data to show you, but there is variability in the values. And in addition, there are a lot of pediatric values that are just not known very well. And so some of the time, we're kind of forced to use adult assumptions because those are the only things we can find in literature. So like one example, a non-EP parameter, but fiber orientation, we have to make assumptions based on adult values, but we're actively right now working on doing DTMRI on congenital patients post-explanted hearts, post-transplant with Dan Ennis at Stanford to try to actually quantify differences in fiber orientation from normal. I hope we can do similar, get similar data for some of those, like conduction velocity and similar. Cool. Sounds very complex, but very cool and hopefully helps us do better CRT. Excellent. Thank you. Thank you. I have the pleasure of introducing Dr. Jan Janicek from Prague as our next speaker, continuing in the now tradition of MD-PhDs talking with us today. Okay, thank you, Elizabeth and Peter, for the introduction, and I'm delighted to be here with you again, once more, and I'll start my talk. I have just these disclosures. So the focus of my topic should be looking at novel echocardiographic and advanced imaging techniques in planning CRT. And actually just to start and kick off, the indications in CRT in congenital heart disease are quite more diverse than in adult patients with ischemic or dilated cardiomyopathy, because we have more structural and functional diversity. And we have recent guidelines, how to indicate or what are the CRT indications. And these guidelines, if you look at them, are based basically on ventricular matrix, on gross electrical dyssynchrony in terms of bundle branch block patterns, and on heart failure symptoms. And they do not consider the specific electromechanical dyssynchrony patterns, which are or are not amenable to CRT. So that will be the focus of my next slides. So how to assess dyssynchrony in a CRT candidate. And it has two parts. It has an electrical dyssynchrony and a mechanical dyssynchrony. And both has to be put somewhat together to get the electromechanical dyssynchrony picture. And a simple look at the ECG may tell you whether it's potentially a CRT candidate or not. If you have a conduction delay within the failing ventricle, either right bundle branch block or left bundle branch block pattern, it might be a CRT candidate. There are nowadays more sophisticated methods available to depict non-invasively the electrical activation sequence. And one of those is a new emerging method of the ultra-high frequency ECG. And the ECG is based on the detection of very high frequency signals. And actually, what these signals represent is the phase zero overshoot of the action potential. So the ECG is sampling these very high frequency signals. And it detects only near field ECG signals and just is able to visualize ventricular activation sequence as projected to body surface. This is one example of a patient of ours, which is a patient with Epstein's anomaly, right bundle branch block pattern. And when we go through the imaging, so this is the electrode set up with five extended leads to cover the most of the right ventricular wall. And the second one is already the local activation timing. And you can see that on the right side, with the right-sided leads, there is considerable delay of the activation by, let's say, 100 milliseconds reflecting this right bundle branch block pattern. But as it is near field signals, you can really depict the size of that late activated parts of the myocardium. And by a proprietary software, we can project these activation times to all the respective lead positions. So the electrical part, and when we jump to the mechanical part. So assessment of mechanical desynchrony nowadays is mostly used by echocardiography. And we need both the old method of conventional echo, as well as the advanced spectral tricking imaging methods. The conventional one we use to establish global timing. So the timing of the failing and of the non-failing ventricle, timing of inflow and outflow. And we can relate to this timing the segmental motion events and thus detect wasted contraction that is not contributing to ventricular ejection. So this global timing gives us a global view on how this pattern of desynchrony works. But we may look by using spectral tracking more specific, detecting a specific classic pattern of desynchrony, performing, let's say, qualitative assessment of desynchrony. And this classic pattern of desynchrony has been correlated with CRT efficacy. And this classic pattern involves early septal contraction, accompanied by early lateral wall stretching. The peak septal contraction should be finished within less than 70% of ejection phase of the ventricle. And there is a rebound stretch of these early contracting segments by the late contracting wall contraction, which peaks after closure of the either aortic or pulmonary valve. So this is a pattern we can detect in patients with congenital heart disease. I will show you an example. This is not an easy example. It's a single ventricular patients of LV type with apical to basal desynchrony due to apical pacing for atrioventricular, complete atrioventricular block. And you can see exactly the classic pattern of desynchrony with early contraction at the apical segments, pre-stretch at the basal segments, eight length contraction of the basal segments with rebound stretch of the apical segments. And this is the result after resynchronizing such a ventricle by putting another lead on the base of the ventricle. And you can see the desynchrony pattern is gone. We can align both the electrical and mechanical activation and look for what has been discussed here before at electromechanical coupling or interaction. And this is, again, an example of the Epstein patient. And we can put this ultra-high frequency activation, electric activation sequence with the mechanical activation sequence. And here you can see how this correlates with peak systolic longitudinal strain on the right ventricle. So the timing and the delay is very similar to the electrical one. And again, once we resynchronize such a ventricle, so we achieve almost complete electrical and mechanical synchrony. By using and combining these two methods, we can get more insight into the pathophysiology of a desynchronous cardiomyopathy. This is an example, again, single ventricular patients, double inlet left ventricular type after TCPC. And this is the electrical activation sequence with early activation of the right wall of the ventricle and late activation of the left wall. And accompanied by the early onset of contraction reflected by the strain rate. And again, peak of contraction is reflecting by peak strain. And if you put this together and make a graphical representation of this, you will see how actually the electromechanical delay changes from early activated segments to the late activated segments. And how in the late activated segments, peak contraction is much more delayed than in the early activated segments. And thus, the electrical desynchrony is aggravated a lot by the mechanical behavior of those early versus late activated segments. So we can look at each specific patient in this way. Now CRT, the key element of CRT is improvement in contraction efficiency. We can again look at this with different methods using speckle tracking echocardiography. This is rather a simple method using strain rate and looking at the relation of stretch versus contraction during systole. And you can calculate a so-called systolic stretch fraction. And the lower the number, the better the contraction efficiency. But we can also use a new method like the myocardial work calculation. This is a non-invasive assessment of myocardial work from speckle tracking echocardiography that allows us to construct pressure strain loops. And the area under curve reflects the myocardial work. And we can then calculate constructive wasted work and work efficiency, both segmentally and globally. And this is again an example of the patient I referred previously, single ventricle LV type, apical paste patient. And this is before desynchronization. You can see the time to peak strain will I plot with very early peak strain in the apical segments close to the pacing lead and very late in the basal segment. And the respective myocardial work efficiency is very low efficiency, again, in the early paste segments. And the global work efficiency was 68% here. And after resynchronizing this ventricle, this picture improves. We can see almost equal activation times mechanically and improved work efficiency, both segmentally in the apex as well as globally. Now, I very much liked Alison's talk, which was on computer modeling. And we use another model to model desynchrone, which is the adapt model by the group of Maastricht. And this is rather a mechanistic model. It does not involve really cellular electrophysiology, but gives also quite good insights into the patterns of mechanical desynchrony as has been used and published for analysis of mechanical desynchrony, specifically in left-bound branch block patients. And what the model allows is to implement delays, contraction delays, for the free walls and for the septum. And we have used this model to construct a digital twin to a tetralogy of followed patients to study relative efficacy of different interventions, like pulmonary wife replacement, RVCRT, or combined PVR plus RVCRT, and a post-repair TOF. So this is a digital twin to an existing clinical scenario. And what you can actually compute or model is the pressures in the ventricle after PVR, RVCRT, or both methods, the strain curves and desynchronic patterns, the systolic stretch fraction, and wasted work ratio, just an example of parameters we can model here. On top of that, we can also model, for instance, the cardiac output at maximum exercise, and we can plot it against various degrees of pulmonary stenosis and pulmonary regurgitation. And the blue curve is CRT off, the green curve is CRT on. So you can see how, during various degrees of either pulmonary stenosis or regurgitation, there is a different gain from RVCRT dependent on the degree of the pulmonary valve pathology. So to conclude, the advanced imaging method allow for correlation of electromechanical activation sequence. They detect segmental contraction timing and classic pattern desynchrony as prerequisite for CRT efficacy, and we can calculate the contraction efficiency and myocardial work, and we may predict mechanical CRT effects by computational modeling. Of course, there are limitations still, because all the techniques showed had been mainly derived from adult dilated cardiomyopathy CRT patients. We have only case-based data in congenital heart disease, but the principle's results are obviously similar. So I would conclude with stay synchronized, please, all of you in the room. Okay. Always wonderful. Okay, so now one of the issues we look at, if you go back to the 90s with pressure-volume loops, it took too long. So when you're doing this now, how much extra time is being added to your implant of where you're putting the leads? You mean during implant? Well, I assume you use this during implant, right? During implant. Right. So as we mostly go for epicardial, we have to have an idea for the surges where to go grossly. And then we check in the OR by looking at the timing of the local activation signal from the lead or from temporary leads, whatever is placed there, and try to get a late electrically activated site. But it should fit with our pre-op idea on how the electromechanical desynchrony behaves. If there is no scar, the mechanical desynchrony should follow the electrical desynchrony grossly. So timing, 10 minutes, 15 minutes, how much extra? In the OR, you mean more time is consumed to find an electrical appropriate site? So kind of like the old days? For the activation. So like when we're doing surgical mapping in the old days, you're spending more time mapping it out than actually putting the leads on. Because people are going to want to know how much extra time is the patient going to be under anesthesia, and I got to do another case in 10 minutes, you know, so. Grossly. For the OR, to find the late activated spot, not more than 10 minutes. Excellent. Because we know where to go, basically. Thank you very much. Henry Chubb at Stanford, and as always, fantastic hearing from you and the true leader in looking at the imaging surrogates of electromechanical desynchrony. But what do you do clinically? Suppose you had a Fontan patient, YQ restoration, come in, EF at 30%. What do you look at on a clinical basis to help you decide whether or not you should crack open the chest and put on multi-site pacing? Yeah, we, nowadays we look at the ultra-high frequency ECG to see the activation sequence. We have the chance to do so. And then we go for echo analysis. And what we are looking at is the desynchrony pattern compatible with classic pattern desynchrony. Do we have segments who have peak contraction after closure of the respective, in that case, aortic valve? How much contraction is wasted, actually? And then we decide whether it's worth to go for the implant. We mostly don't go into the cat lab to test it nowadays, because we are quite sure using this method that either should help or should not. Thank you. One last question, if I may, if no one else has one. The utility of the imaging seems like there's so much more quantitative data that you presented, the global work, the strain fraction, there's another index, though in clinical terms it feels like we're mostly gestalting it. And looking at the speckle tracking pattern, but not using all the other quantitative data, does it take a lot of work to get that information? Or is it really there when they're doing, our imaging colleagues are doing speckle tracking and we're just not paying attention to it? I think doing that in-depth analysis of the systolic stretch fraction, you have to have some software to do it. It's not very complicated, it's MATLAB, we do it sometimes, but we don't do it in a case-based scenario. We just, if we see the desynchronic pattern, we see that there is contraction after closure of the valve, or very early contraction, not contributing to ejection. So then we know that there is wasted work for sure. So what kind of follow-up have you had now on these patients to determine that actually you're showing chronic improvement over time? So chronic follow-up, how long have you looked at these patients and what kind of follow-up studies are you doing to confirm that indeed you have the most beneficial contractility going on? Ideally, we look really as long as we have the patient under our follow-up. As soon as they transfer to adult care, we mostly lose them for that kind of follow-up. But as long as they are followed by ours, we do, let's say, yearly anecho-evaluation of the desynchronic. So just a show of hands, if you're not embarrassed, how many just do ejection fraction when they're looking at contractility? Good. So there are new echo parameters. As Jan has shown us, that's great. Any other questions? None? Okay. Thank you. Thanks. All right. And next we have Dr. Adam Keene from Indiana and Case Reserve, who is going to be talking about Hot or Not, Classic CRT, CSP, and Lot CRT. Right, excuse me, thank you for the organizers for allowing me to present today. I apologize for introducing you wrong, Riley Children's, I'm so sorry. Not a problem. Yes, so my presentation today is entitled Hot or Not, Classic CRT, CSP, or Lot CRT. This is going to be considerably less quantitative than we've just experienced over the past couple talks, but hope you'll bear it with me. The last time I was this underqualified in speaking to an audience about their own expertise was probably when I was nine at my father's Sherlock Holmes meeting describing the merits of Dr. Watson. So, nevertheless, I hope to ignite your intrinsic abilities into solving these riddles into congenital heart pacing, specifically in resynchronization. The objectives for my remarks include the following. We're gonna define CRT, CSP, and Lot CRT. We're gonna complete a brief literature review. I'm going to suggest an approach for pediatric and congenital heart pacing, and then finally we'll discuss the next steps. Ultimately, I hope to leave you with these three ideas. Numerous options exist both for standard ventricular pacing as well as for ventricular pacing strategies to overcome ventricular dysfunction. Despite its steep learning curve, conduction system pacing may prove to offer superior cardiac function for the chronically ventricularly paced individual. And finally, target ventricular lead placement to optimize the patient's cardiac function. So let's begin. CRT signifies cardiac synchronization therapy, most commonly using a pacing lead at multiple ventricular sites to improve the electrical and hopefully mechanical ventricular systole and diastole. Originally, this term was often used synonymously with biventricular or, quote, BIV pacing, involving pacing catheters in the subpulmonary ventricle and the coronary sinus. Given anatomic limitations and increased prevalence of epicardial pacing in the pediatric and congenital heart population, many different pacing lead configurations have been used, and of course, resulting in limited, though important, datasets. Wireless electrodes in the subsystemic ventricle are now in use and showing promise for those individuals not responding to traditional resynchronization, so-called YCR, excuse me, YCRT. CSP signifies conduction system pacing. As you can appreciate from the rotated diagram on the right, this includes his bundle and parahispundle pacing, as well as left bundle branch and left bundle branch area pacing. This method recruits some section of the cardiac conduction system distal to a pathological interruption to resynchronize the myocardial excitation and ultimately contraction. Finally, we have LOT and HOT CRT, which signifies left bundle branch and left bundle area pacing, or his bundle optimized CRT, where in a traditional system, the subpulmonary ventricular lead is a left bundle branch or a his bundle lead, and the coronary sinus lead is also used to support the electrical resynchronization effort. On the right is a diagram of a LOT CRT D system with the addition of a defibrillation coil, requiring, of course, a DF1 header, in contrast to the current headers demonstrating the improved electrical front. So to some, numerous options exist, both for standard ventricular pacing, as well as for ventricular pacing strategies to overcome ventricular dysfunction. On to a brief literature review. Searching for the headings of cardiac resynchronization therapy citations in PubMed shows nearly 12,000. Conduction system pacing or his bundle pacing or left bundle pacing takes off around 2015 with over 1,200 citations. LOT CRT or HOT CRT is just starting to make a ripple in the past few years. Of course, volume does not represent growth for those amateur economists out there. So here we have the year-to-year rate changes showing CRT stagnating since the late aughts, CSP decreasing since its peak in the mid-2010s, and LOT CRT just hitting the scene. This amateur economic illustration, of course, is somewhat irrelevant when we focus in on pediatrics. The blue lines show CRT for pediatrics and congenital heart disease, so there's really no reason to perseverate on further graphs in CSP and LOT CRT in this population. So the second take-home point I hope to convince you of, that the data's going to support, despite its steep learning curve, conduction system pacing may prove to offer superior cardiac function for the chronically ventricularly paced individual. Well, I'd love to go into deep detail, even just the highlights. We really don't have the time for that, so we're just going to do a quick snip of the critical pieces. Dr. Vijayaraman's multicenter retrospective cohort in 2023 showing echocardiographic improvement with CRT added to his bundle pacing in myopathic hearts, showing a greater than 7% decrease in death or congestive heart failure hospitalization for the left bundle branch area pacing cohort. Dr. Vijayaraman has led the way also in HOT CRT, a multicenter retrospective LOT CRT cohort eloquently demonstrated a stepwise improvement from the baseline to stepwise by VCRT to left bundle branch area pacing, and finally LOT CRT with the QRS duration measurements graphed on the left and demonstrated on the right. Regarding pediatric and congenital conduction system pacing, we first give due credit to Dr. Carpa, which is pioneering work in the early 1990s. I became interested in this modality many years later in my collaborations with Gopi Dandamudi about 10 years ago. Daniel Cortez, among others, shared this enthusiasm, and Jeremy Moore completed the largest retrospective cohort in adult congenital heart disease population comparing conduction system pacing with by VCRT at one year showing a non-inferior LVEF and a shorter QRS duration. In this context, Henry Chubb, Doug Ma, Ann Dubin, and Jeremy Moore published a comprehensive review on the topic leading, of course, up to the 2023 HRS guideline on conduction system pacing for the avoidance and mitigation of heart failure, which wonderfully and appropriately gives attention to the pediatric and congenital heart disease population. From the airplane view, we can see that for the adult population, though we do not have class one recommendations for CSP, class 2A and class 2B indications exist for his bundle or left bundle branch area pacing as well as lot CRT. For the pediatric patient with complete AV block, the data is not as specific, offering guidance insofar as optimizing the lead position in whichever method is deemed to be most successful. If there's congestive heart failure by VCRT or CSP, if not, avoiding the RV apex, of course, at all costs. For the congenital heart population, the story is a bit more complex, given the lesion-specific needs to optimize systemic ventricular function with evidence supporting CSP and multi-site pacing over traditional CRT for certain lesions. Which brings us back to Henry Chubb et al.'s review, which I consider wise and nuanced statement toward the end, which I'm going to shamelessly quote. Perhaps the greatest determinant of CSP modality in the long term in pediatric and congenital heart disease will be the difference in learning curves. In contrast to adult electrophysiology, even the largest pediatric and congenital heart disease centers rarely perform over 50 device implants per year. It has been suggested that approximately 40 implants are required for the learning curve to begin to plateau for his bundle pacing, with substantially smaller number of implants required to learn left bundle branch area pacing techniques. And so, may I suggest an approach for the pediatric and congenital heart pacing? Target ventricular lead placement to optimize the patient's cardiac function. CSP procedures are more difficult, though the potential benefits may be significant. Here we have an example of a lead revision CRT in a 15-year-old with an ASD and a VSD status post repair, as well as congenital AV block status post single chamber epicardial pacemaker system at two weeks of age. He underwent ASD and VSD repair at six months of age with a dual chamber pacemaker revision. At nine years old, he had an epicardial ventricular lead revision due to a fracture with an LVEF of 52%, despite the epicardial RV lead at the inferior base, as you can see. At 15 years of age, at this point under my care, the LVEF was noted to be 37%, although I had not yet done speckle tracking and strain. He began medical management with poor adherence. He subsequently experienced an atrial lead fracture in the setting of competitive soccer, leading to a VVIR reprogramming, of course, resulting in dizziness during activity. We move forward with a transvenous CSP revision resulting in a left bundle branch area pacing system. Notice the more distal lead location due to his VSD patch. The LVEF increased to 55%, and he was asymptomatic at four weeks post-op with subsequent return to competitive soccer. Next, an example of what I would consider mixed CRT in a 34-year-old with atrial and abdominal situs inversus, levocardia, L-looped ventricles, an anterior posterior pulmonary systemic outflow, and VSD status post-surgical patch. He was lost to follow-up for numerous decades. He presented with AF and AV block with a junctional slash ventricular escape cure restoration varying anywhere between 138 and 178 milliseconds. His sub-pulmonary ejection fraction was 52%, and subsystemic ejection fraction 49% by MR. Although I attempted a left bundle branch area pacing lead, I was not successful in adequately penetrating his septum, either because I couldn't get by his patch, or as you may have noticed, this bulge of tissue here. Either way, I was not successful in putting the lead that I had desired to place, and there was no clear CS approach. But given his anatomy, I paired with our congenital heart surgeon, and we completed a CRT system with a sub-xiphoid epicardial lead. On post-op X-ray, we see the anterior subsystemic LV between the transvenous and epicardial pacing leads. So the next steps. Exclude the impossible, and whatever left, however improbable, must be the truth. Needless to say, we are nowhere close to excluding the impossible at present. Prospective multicenter and equitable evaluation of CSP is dearly needed. This idea began five years ago, and as we learned this morning, we need to contact our learned colleagues for details, and to please contribute your patience. And this is just the beginning, as we seek to better understand the potential roles of YCRT, CSP, defibrillation systems, and then ultimately, how to incorporate leadless pacing in conduction system pacing. To some, numerous options exist, both for standard ventricular pacing, as well as for ventricular pacing strategies to overcome ventricular dysfunction. Despite its steep learning curve, conduction system pacing may prove to offer superior cardiac function for the chronically ventricularly paced individual, and finally, target ventricular lead placement to optimize the ventricular cardiac function. Thank you. So when you're trying to get a conduction system area pacing, which leads are you using? I think this might be a factor in why a lot of people have trouble getting where they quote, wanna go. Yeah, so like many of you who are performing these procedures, I started out with 3830s, and I actually use 3830s for all my implants, going back 15 years at this point. That, of course, has changed with new leads on the market, new catheters that can help direct us, whether they're deflectible for that individual who had an anterior LV. Of course, we had to reshape our catheter to get it to go anterior as opposed to directing it posterior to the septum. So it's a mixed bag is the honest answer. I think you need to use what works in your hands, and there's a lot of options out there. So one of the concerns I always have is there is no lead specifically designed to do conduction system pacing, period. Industry had that idea 15, 20 years ago, and it faded away because there was no volume for it. And I always have concerns about, yeah, you can take the 3830, you can go to the moon with it. You just keep torquing it, it's gonna go right through the heart, not a problem. So when you're doing that, how are you deciding I wanna do left-sided septal pacing versus more proximal septal pacing and try to get both parts of right and left bundles stimulated at the same time? So left bundle versus, quote, conduction system pacing. Are you making a decision at that time, or you just kinda say, look, it looks pretty good, the numbers look good, go from there? So I was much more what I would consider academic at the beginning of the process. Like many of you, I would often bring out an electroanatomical mapping catheter, and I would start mapping the conduction system. The process that we go through to operationalize this presently is to use the experience of the adult colleagues, and I really target the mid-septum, if you will, trying to go toward the left bundle branch, but am looking at my ECG characteristics, and if I'm happy with what I'm seeing, often it's one of those harken back to the whatever, late 70s, early 80s, I don't press my luck, and I sort of stop there. Please. Is it possible to use the microphone? Sorry. Hi, fantastic talk, thank you for that. My name is Saket, I'm an EP in Vancouver. I often find that larger the septum, more complex the anatomy, the harder it is for me to understand or decide that I'm accepting this as my QRS duration or pattern, especially as the septum is mostly just a patch. I find that the stem to the LVAD time can be very short, even though you're not actually engaging the left bundle, and also sometimes you can have a very displaced bundle where you don't get a typical AVR, AVL discordance. I'm just curious to know what your thoughts are regarding what would be the best parameters to target, because we don't often get an intact heart parameters. Thank you. Yeah, these are really important considerations, and I will say at the beginning of our experience, if you look at the adult data, they'll say, all right, stem to the peak of the R, as long as you get below 85, 75, you're gonna be great. And of course, there's some data from our pediatric experience that says that's not gonna do it, right? And the same thing that we learned when we were ablating accessory pathways or trying to assess the danger of antigrade wolf pathways or the typical criteria in terms of what constitutes AVRT versus AVNRT. It's different in a smaller heart or in a conduction system that is more rapid. So the honest answer again is, I'm gonna humbly say I don't have the answer. I can tell you what I do, but there's a lot of brilliant people in the audience, and we need to put all of this data together so that we can actually get criteria to tell you. What I will say is the shorter the stem to peak of the R wave has been shown, at least in my anecdotal practice, to show improved output and function, but we've got brilliant presentations who just showed here maybe we need to look a little more analytically at that as opposed to just ad hoc. Sean Mahan, University of Kentucky. So I finished my training right when all those publications were coming out in 2015, and just being in a combined peds adult program, it kind of seemed like the Wild West when conduction system pacing was coming out. Everyone was kind of figuring out their own protocol. And for kids in particular, when I check with my adult colleagues, if I have someone who I think is a candidate, what makes them leery? And there is a learning curve definitely, like how many turns you put in. Is there any imaging criteria as far as the thickness of the septum, especially when you have a young child where they're preteen or a teenager? There are definitely a couple case reports out there of perforations, and that are sometimes late complications, like after the implant. Yeah, so my criteria is by looking. I need to see the coil and see what it's doing to that septum. The whole, all right, you do seven turns. I mean, the reality is sometimes when you're doing this, you'll super coil the lead within your sheath, and you're not delivering any of that to the tip of the lead at all. And then you straighten out your sheath a little bit, and all that super coil unwinds. So I really need to see what it's doing into the septum. And then ultimately, obviously, once all the catheters are out, making sure it still looks good and so forth. But I don't have any criteria in terms of, all right, if they're seven years old, then usually five turns does it. If they're 12, then I add another one or anything along those lines. Now remember, there are data out there in the echo world as far as what is the septal thickness by age. That'll give you some idea of the beginning. And also remember, Mac Dick way back in the dark ages of the 70s, remember the 70s? No, half of you probably don't. He actually showed that the conduction system is like two millimeters away in a five, six, seven, eight, nine-year-old kid. So if you're gonna put a transvenous in, you don't have to go corkscrewing for 12 turns. You know, you get it in, like I said, you look at the dimension of that septum based on age and get you some idea. If you wanna stop, freeze it, and measure it, you can do that and give you an idea where you're going. I'll just make one other comment, if that's okay, about the depth of screwing in. No credit at all for me for this, but for Ron Rosas, I don't know if he's here, but he's got beautiful three-dimensional TEE imaging, which he does in real time. You can see that lead come across, you can see the position as the tip comes towards it. It's not something we've done at Stanford, but the images he's shown me are absolutely fantastic. So one other way of thinking about it. Thank you. Excellent. Thank you. So we are actually two minutes ahead as opposed to the last session. So thank you for attending and enjoy the conference.

resynchronization pacing

congenital heart disease

CRT

computational modeling

electromechanical dyssynchrony

conduction system pacing

echocardiographic techniques

ultra-high frequency ECG

3D modeling