Okay, ladies and gentlemen, dear colleagues, good afternoon. Despite this very strange COVID context, I'm very happy to virtually participate to this very exciting session on patient-tailored approaches in persistent AF. Here are my disclosures. So why performing a patient-tailored ablation in persistent AF? Let me start by the slide and to show you different type of patients. All these patients have a common point, they all have atrial fibrillation. Obviously, all these patients have various sex, ages, weights, and comorbidities, and of course, they all have various complexity of their atrial fibrillation. The question is, should they all be ablated exactly the same way? The answer for me is, of course, not. Let me show you here on this slide, the schematic biatrial ablation set using a wholly patient-tailored ablation approach in 13 patients that led to AF termination and freedom from AF at 18 months. The gray areas corresponded to the ablation areas that we performed in this patient using our ablation approach. And as you can see, the location and the extent of AF drivers is very variable, but I will come back to this approach later. So although BVA isolation is very effective and reproducible in paroxysmal AF, it is not sufficient in most persistent AF. So which ablation approach should we choose? Atrial segmentation with roof and mitral lines in addition to BVA isolation is not sufficient with, in the STAR-AF2 study, only 48% of success rate at 18 months with or without drugs after multiple procedures. What about chemical ablation in the vein of mortia? Let me show you on this slide the last data presented by Dr. Valderrabano at ACC a few weeks ago. Dr. Valderrabano is a pioneer of this technique and has a huge technical experience. In this study, authors have randomized persistent and long-standing persistent AF patients into two arms. The first one was a conventional ablation arm with BVA isolation in all patients plus minus posterior wall isolation, mitral line, CAFE. And the other group, the martial group, was conventional ablation plus vein of martial chemical ablation. And here is a follow-up. After one and multiple procedures, you can see that the martial group had better outcome. But you can observe also that in this complex patient, the success rates after multiple procedures remained moderate at one year with 62% of freedom from an arrhythmia. So what about panoramic mapping approaches for AF drivers detection? Sanjeev Narayan and colleagues, using an endocardial panoramic mapping method, were the first to demonstrate that localized rotational or focal AF drivers could be automatically detected and ablated with a clinical benefit in comparison to PVA isolation. Then the border group, using an extracardiac panoramic mapping method, drew the same conclusions. This data was very promising and these two teams brought back patient-tailored AF ablation approach in the field of persistent AF ablation after the CAFE era and after STAR-F2. So let me come back in 2004. You all know this paper from Winne-de-Manie, who published excellent long-term results with CAFE ablation. These results of CAFE ablation could be reproduced by some groups, including the one of Isabelle Deisenhofer in Munich, the one of Naoya Oketani in Japan, our group in Marseille, but also others. But as you know also, a lot of other groups could not reproduce CAFE ablation outcomes. And unfortunately, as you all know, the STAR-IF2 randomized trial did not demonstrate any benefit of CAFE ablation in addition to PVA. So it was of course a big disappointment in the EP community. And how could we explain such a difference between these two studies in terms of outcomes? Interestingly, if you look at the AF termination rate in the Winne-de-Manie paper, it was 95% and it was only 45% in the CAFE group of the STAR-IF2 study. So is AF termination associated with better outcomes? Verma and colleagues conducted a very interesting study that seems to indicate that despite a tendency, AF termination is not significantly associated with better outcomes. We have a tendency of superiority in the AF termination group in comparison to the no-acute AF termination, but there is no significant difference. But when we look at the literature, we can see that a lot of studies suggest that AF termination is associated with better long-term outcomes. Here, for instance, studies from the Border group that demonstrated that after a single or after multiple procedures, patients with AF termination had better outcomes. Here again, a meta-analysis with a lot of studies showing that patients in blue with AF termination had better freedom from any arrhythmia recurrence than patients without termination in pink. If we look at this study from O'Neill from the Border group in 2009, they have excellent long-term outcomes, they have excellent short-term outcomes with 85% of AF termination, but you can see that the procedures were long and extensive with 255 minutes of procedure time and 88 minutes of radiofrequency time. So, should we pursue AF termination, and should we perform an ablation set that is patient-tailored? That's the choice we made. Historically, we used to perform classical electrogram-based ablation, CAFE ablation, with quadripolar ablation catheters, but when we started to use multipolar catheters, as pentarray, HD grid or Orion catheter, we realized that we could gather much more information with multipolar catheter than with ablation catheter. Then, we realized that not only the fractionated morphology was important, and we defined a new target for ablation, named specific temporal dispersion of electrograms. Specific temporal dispersion of electrograms is a group of electrogram abnormalities, defined as non-simultaneous activation at multiple neighboring electrode locations, with or without fractionation. This is very important because, although the great majority of dispersed signals are fractionated, some dispersed signals are not fractionated, and a lot of fractionated signals are not dispersed. Using this definition, we conducted a prospective ablation of dispersion areas in patients without PVA isolation or any lines, in order to test the hypothesis that dispersion was a footprint of AF drivers. We compared this series of 105 patients to a control group of 46 patients ablated with a stepwise approach. You can see here that we observed a better acute efficacy, with 95% of AF termination, and only 60% in the stepwise group, with less extensive ablation, with shorter procedural time, less than 3 hours, 2 hours and a half, and shorter radio frequency time, around 49 minutes. Interestingly, the ablated left atrial surface was only 17%, so we can say that this kind of approach has a good acute efficacy and is not extensive. Let me come back on these 13 patients with a schematic beatrial ablation set that led to AF termination and freedom from AF at 18 months. And so you can see how much the location and the extent of the ablation areas, and so the AF driver's location and extent, is variable. Here you can see on the roof, on the anterior wall, here on the posterior wall, in some patients the posterior wall is completely clean, in some other patients we have to ablate in the right atrium. So this slide for me highlights the importance to adapt the ablation strategy to each patient. So what about the follow-up? At 18 months we observe a superiority of the dispersion-guided ablation in blue in comparison to the control group, with 85% of patients free from any arrhythmia, after 1.4% by patients with or without drugs. Juna Idzaman and Sanjeev Narayan did this very nice review a few years ago about localized rotational drivers. Different methods and tools have been used to describe the same mechanism. So personally I don't know exactly what a rotor is, but I know that some specific localized atrial areas with complex rotational and focal activities drive AFib. And when you ablate these areas, you contaminate AFib. So that's why we are not talking about rotor ablation, it could be rotors, but we prefer to talk about AF drivers and patient-tailored ablation approaches. So you can see on this table various acute and long-term efficacy with different level of mechanistic evidence to demonstrate that AF drivers are really AF drivers. For instance, in this study, evaluating the CARTOFINDER, authors performed PVA isolation followed by driver's ablation. The AF termination rate was 15% and they had 70% of freedom from AF. But as I just said, PVA isolation was performed before driver's ablation and there was no control group. So how authors could be sure that the 70% AF freedom was not due to PVA isolation? The main limitations of AF drivers-guided ablation approaches are the following. First, PVA isolation may be performed prior to any driver's ablation. And second, they might not have any control group. So have we demonstrated that spatio-temporal dispersion of electrograms is a footprint of AF drivers? We have a first level of evidence. We have performed theoretical demonstration with numerical simulation and animal studies with optical mapping. We also have demonstrated that by ablating these dispersion areas, we achieved a 95% of AF termination without any probabilistic ablation. We also have 18 months outcomes in comparison to a control group. But what about long-term outcomes? So let me present you the four-year follow-up of a series of patients ablated using dispersion-guided ablation. These data will be presented by Nicolas Mayot in a poster session at HRS 2020. You can see that we have 154 patients with 30% of long-standing persistent AF and 70% of persistent AF. They have structural heart disease in 46% of cases and their left atrium was dilated with 170 milliliters on average. The procedure times were quite short with 2.5 hours and the radiofrequency time was 49 minutes, the AF termination was 89.5%. At four years, we have 68% of patients free from any arrhythmia with or without antiarrhythmic drugs after 1.9 procedures. Interestingly, if we compare the AF termination rates of patients with IF or AT recurrence and in patients without recurrence, you can observe a higher AF termination rate in patients without recurrence. So what about the reproducibility of dispersion-guided ablation? More than 70 physicians from 20 countries came to visit us in Marseille to perform dispersion mapping for AF ablation and several publications have been done essentially by the Chinese group from the Shanghai Chest Hospital demonstrating a superiority in comparison to classical AF ablation. So I will quickly present this study in 142 comparable patients with persistent and long-standing persistent AF. They randomized patients into two groups, dispersion-guided ablation and a stepwise approach. They observed higher termination rates in the dispersion group and they also observed a better long-term outcome at one year. Although many obtained similar results, they also indicated that observing and analyzing multipolar signals in real time is challenging. So although some centers are able to reproduce our data, the subjectivity of visual analysis is of course a real issue and an automatization is absolutely needed. After several attempts and failures with classical signal processing techniques, we decided to work closely with data scientists and to extract and annotate a lot of intracardiac signals in order to create a huge database. Our engineers have created a sophisticated machine deep learning software with a high number of electrogram analysis features and this software provides a real-time analysis of electrograms in order to guide the ablationists in the epilab. In order to test this algorithm, we have performed a lot of studies. Let me quickly present some of them. We have some preclinical data based on a performance comparison between a panel of 28 expert electrophysiologists and the Volta algorithm in dispersion detection. It suggests that the performance of Volta algorithm exceeds the one of cardiac electrophysiologists when they have to make a decision with a limited amount of time. You have the rock curve on the right side of the slide. We also have performed a clinical study, the AI-FIB trial, including 300 patients with AF and ATAC. All these patients were ablated using the Volta algorithm with this interface, analyzing in real-time the penta-array HD grid or Orion bipolar electrograms. I'm not allowed to communicate on these results right now. These results are coming soon, but I can tell you that they are really promising. In conclusion, dear colleagues, the anatomical approaches in persistent AEF are easy to reproduce but may hinder or over-treat patients with moderate efficiency. Tailored approaches can be effective in case of high AEF termination rates. Spatiotemporal dispersion of electrograms is a signature of AEF drivers. Dispersion-guided ablation is effective and not extensive, but can be difficult to reproduce. Artificial intelligence seems promising for automatization and a better reproducibility. Thank you very much for your attention.

patient-tailored approaches

persistent atrial fibrillation

ablation

panoramic mapping

artificial intelligence