<v ->So, we've come a long way</v> in the last 20 years or so, some of which was just described nicely by John about the different ways of classifying stroke. And below here, you can see the different randomized trials and the RoPE study and the SCOPE study, and PASCAL finally. That's Blaise Pascal at the end that I'll explain to you in a bit. And you know, 20 years ago we all saw stroke patients, we all found PFOs, and we had no idea what to do with them. We didn't know who had a real PFO that was related to their stroke, who had part of the background noise of the 25%. There are 25 PFOs in this room. I'm gonna say whether you know it or not. But they're out there, and we didn't know what to do. And we've come a long way since then. I was also asked to address shared decision making, which of course is crucially important for this young population in general. And these are the questions I think that patients really are asking, which is, "What caused my stroke? "What is my risk of having another one? "And what can be done to reduce my risk?" And there are different variables that cause different directions of influence on these questions. So if this patient had a PFO, a "forkogenic" vertebral artery dissection, you probably wouldn't consider closing their PFO. You'd probably want to get the fork out and do something for the dissection, I guess. <v Messe>Depends if you're a cardiologist.</v> <v ->Okay.</v> Steve has an opinion about cardiologists. So, the real game is trying to identify the PFOs that are really important and the PFOs that clearly have nothing to do with the mechanism of stroke. Now, this is what we're fighting against. Let's just look on the left for a second. There are roughly 600,000 ischemic strokes in the United States per year. About 20% of them are in younger patients, about a third of them are cryptogenic. And in those cryptogenic patients, about half of them will have a discovered PFO. And that means about 18,000 in the United States per year and almost half a million or so worldwide. But just because you find a PFO in a cryptogenic stroke patient, it is not the same as making a diagnosis of paradoxical embolism. And here's the problem. If this is the universe of cryptogenic stroke patients. And there are a bunch of assumptions here, but just accept this for a moment. This is the universe of cryptogenic stroke patients. In those cryptogenic stroke patients, you will find in this example, 40% of them have a PFO. That's way too many. 25%, roughly, have a PFO in the general population. And I think we would all accept that this 60% in whom there's no PFO probably didn't have a PFO-related stroke. That's fair, right? But there have to be... Because these non-PFO-related cryptogenic stroke patients come from the general population, they need to have some PFOs amongst them. So, you have to give them some PFOs to make them match the general population. And if you do that. So, now here's a 25% prevalence of PFO amongst these non-PFO-related but still cryptogenic stroke patients. These are the, oops, I'm sorry. Can I go back? These are the excess PFOs here. These ones have nothing to do with the stroke because it's part of the non-PFO-related, and these are the excess. And the question really is, how can we tell these people apart? How can we disentangle the checked-box PFOs from the unchecked-box PFOs which are likely to have the pathogenic ones? And that's what the RoPE study was designed to help us figure out. I'm not gonna go through the details of how we did it. Many of you know about the RoPE score. It's available for free on MDCalc. It's an 11-point scale, 0 to 10. And the bottom line is the more points you get, the higher your RoPE score. If it's 10 or 9 or 8, the more likely you are to have a PFO-related stroke. And those tend to be younger patients without vascular risk factors, cortical infarct on imaging, and so forth. And then if you have a low RoPE score, you're older, and you've got a bunch of vascular risk factors, a deep stroke for example. So, what the RoPE score allows us to do is observe in cryptogenic stroke patients that in the higher RoPE score strata, you can see the prevalence of PFO is well over 70%. That's way too many. Right? And in the low RoPE score strata, 0 to 3, the prevalence is about 21, 22%, which basically matches the general population. And if you do Bayes' Theorem and a little bit of hypothesizing and some assumptions, these black bars here are what we're referring to as the PFO-attributable fraction. But you can think of it as the confidence that the discovered PFO in that stroke patient caused that stroke. It was a pathogenic one rather than an incidental one. So, this is the confidence of pathogenicity. And you can see here statistically in the 0 to 3s because the prevalence here matches the general population, the likelihood that those PFOs are pathogenic is essentially zero, statistically. The RoPE score, though, is not analogous to the CHA2DS2-VASc score. I've heard people say, "Well, if it's a high RoPE score, "you know it's a high-risk PFO." The CHA2DS2-VASc score, as you know, is a predictor of recurrence or the predictor of stroke in patients who have atrial fibrillation. But you should think of the RoPE score more like an EKG or Holter monitor. It makes the diagnosis. It helps you be confident more or less that it was a PFO-associated stroke rather than something else. And in fact, here are those same PFO-attributable fractions I showed you a second ago. The risk of recurrence goes in the opposite direction. Meaning the higher your RoPE score, the lower is your risk of two-year recurrence. And what that's telling us is that the cryptogenic stroke causes in the low RoPE score subjects, these are the ones who don't have PFO-related strokes, are more risky than are PFO-related strokes probably or statistically. And so just because it's a high RoPE score, it doesn't mean it's a risky one. And actually you can reassure patients that the risk of recurrence is relatively low. So, we did this review of all of these six randomized trials that have now been done. And I'm not gonna go through all of this in detail, but they were almost all just stroke patients. And various inclusion criteria. In the interest of time, I'm gonna race through this. There was some concern, by the way, when all the trials were being recruited, and it was in the '90's and 2000's and 2010's, that we were recruiting the low-risk subjects. 'Cause all the high-risk subjects were being closed off-label. And the concern was that we didn't have many big shunts, and we didn't have many atrial septal aneurysms. But actually that's not accurate. We had plenty of both of those in all of the randomized trials. This is a study-level meta-analysis. Again, I'm not gonna go through it in detail, but you can see that the relative risk reduction was on the order of 70% for PFO closure, in favor of PFO closure compared to medical therapy. An oxymoron. Most of you know that word. It's a figure of speech that combines contradictory or opposite ideas like jumbo shrimp, which will cause tight slacks if you eat too many of them. But here's another oxymoron. This is in the FDA labeling of the two currently approved PFO closure devices, which is cryptogenic stroke due to a presumed paradoxical embolism. Which is stupid. It's saying, "This is a stroke. "We don't know what caused it. "And by the way, it was caused by paradoxical embolism." So, that oxymoron is something that we tried to address when we came up with this notion of a PFO-associated stroke. Let's stop calling it cryptogenic stroke with PFO. Let's say this is or isn't a PFO-associated stroke. We don't say cryptogenic stroke with Afib. We say this is cardiogenic embolism secondary to atrial fibrillation. We think we're close enough now that we should be able to say PFO-associated stroke. And not only that, we've suggested that there's a range of confidence that you can have. Very high, high, medium, or low, possible, unlikely probable, highly probable based on these various characteristics that I'm going to go through in a moment in a little more detail. And this is what is referred to now as the PASCAL classification, which stands for the PFO-Associated Stroke Causal Likelihood. It's a terrible acronym, but I like the picture. There you are. So, this is from the original RoPE score. I wanted to point out these two numbers which are important. So in those patients who have a RoPE score of 7 through 10, the PFO-attributable fraction on average is 80%. 80%, remember that. And those subjects that have 0 to 6, the PFO-attributable fraction is 40%. I want you to remember that. 80 and 40. Keep that in your head. We're gonna come back to that. So, the SCOPE study was a follow-on to the RoPE study that we did together with international collaborators all around the world. It was the trialists from all six of the randomized clinical trials. David Kent, my colleague, is really the brains behind all of this. And we recognize that study-level meta-analyses have shown that device closure is associated with fewer strokes. But then guidelines stress the role of informed shared decision making in evaluating the individual probability of benefits. So we have to talk to individuals, our patients, based on these bunch of data, but study-level meta-analyses only describe the average result over a whole range of risk profiles. The individual participant data that we collected and we could put all together allowed us to, individually for each subject, give them RoPE scores and PASCAL classifications and classify them, therefore, based on the likelihood that their PFO was causally related. With this combined data set of all the six randomized trials, our primary outcome that we wanted to look at was recurrent stroke. And we looked at safety outcomes as well. And we had two pre-specified primary subgroup analyses that we wanted to look at. One is the RoPE score. We wanted to look at the high RoPE scores and the low RoPE scores. And the second is the PASCAL classification. The unlikelies, the possibles, and the probables. Keep in mind for a second. If you have a patient who has a PFO-associated stroke that you're 70% confident about, and what you're trying to do is prevent a recurrent stroke, the assumption is that the recurrent stroke is gonna be the same mechanism as the first one. That might not be true, but let's assume that for a moment. If you perfectly present that by closing the PFO and eliminating that possibility, the best result that you're gonna get from a clinical trial is a 70% relative risk reduction. You're not gonna prevent the recurrences from those other 30% mechanisms because this is a disease-specific therapy, PFO closure. So, the PASCAL classification takes the RoPE score into account, but also adds so-called high-risk PFO features, and puts them together and allows you to classify them into unlikely, possible, and probable. So the unlikely category are the ones who are the low RoPE scores. These are the older patients with vascular risk factors. And they have neither of the two high-risk PFO features. That's a large shunt and an atrial septal aneurysm. We can talk about those definitions, but let's push on for a moment. So if it's low RoPE score and they don't have either of those, then that's considered unlikely. If it's a high RoPE score, the bottom right, even if they don't have any of those high-risk features, that's called possible. If it's a low RoPE score and they have one or the other of those high-risk features, that's also possible. So, that's how you buy into the possible category, either just your high RoPE score or young RoPE score with a dangerous looking PFO. But if you have a high RoPE score and you have a high-risk PFO, then that's called probable. And with the combined data set of all the six randomized trials, we had a whole bunch of follow-up with a bunch of outcomes that we could test. And this was published recently in JAMA, by the way. If you looked at it when it first came out, you should download it again because JAMA made a mistake with one of our graphs, and it was updated in January of 2022. So if you want the correct graph, download it again. So, we had basically 4,000 subjects in the six randomized trials. I'm not gonna go through the Table 1. And here's the overall result. This is very similar to the study level meta-analyses. Not much difference there. Device closure clearly associated with lower risk of recurrence than medical therapy. And here's the primary outcome that we had pre-specified for stroke recurrence based on dichotomizing the RoPE score. And you can see that if you have a low RoPE score, the risk of recurrence is 41% reduced. So, relative risk reduction of about 41%. Sorry, 39%. Hazard ratio of 0.6. If you have a high RoPE score, young patients, no vascular risk factors, the hazard ratio is down to 0.21. Remember those numbers I asked you to remember, 80 and 40? So, 0.21 is a 79% relative risk reduction. That's spookily close to 80, which is what we would expect if this was a perfect therapy for an 80% likelihood population to have PFO-related strokes. And then a 39% relative risk reduction is also spookily close to 40%, which is the PFO-attributable fraction in that population. So, it's almost too good to be true, but these are the numbers that were actually found. When we then looked by PASCAL category. Come down to the bottom first. The probable ones. These are the young ones with high-risk PFO features. The relative risk reduction was 90%, a hazard ratio of 0.1. So, almost preventing almost all of the recurrent strokes. The unlikely category, the hazard ratio was essentially 1. So, it looks like there's no benefit to PFO closure in the unlikely category and the possible is somewhere in between. How about the risk? I'm gonna go quickly through this. Atrial fibrillation that's still persisting beyond 45 days is almost entirely seen in the unlikely group and almost not at all seen in the probable group. So, the benefit is almost completely in the probable category, and the risk is almost zero in the probable category. And that's essentially a perfect therapy for any disease that we're confronting. And then there's some halfway points in between. So, the conclusion is that there is a robust beneficial effect of closure overall. Stroke recurrence with medical therapy is about 1% per year. With device closure, that's reduced by, on average, 60%. But you can see we can do better than that with our PASCAL classification. That's what I said they affect very substantially. These are the same numbers I showed you a second ago. That's the absolute risk reduction. And just very quickly, we've published very recently the outcomes based on the PFO features themselves. Not PASCAL classification, but just whether there was or was not one or the other or both of the high-risk PFO features. This is if they have neither of them, that's one or the other, and that's both. So, you can see there's a huge treatment effect if they have both of those so-called high risk features. So, PASCAL appears to be a useful tool for clinical decision making. Combining multiple variables into a score may have advantages over conventional, one-variable-at-a-time subgroup analyses. You asked me to talk about what I think of TEEs. I think we have to do TEEs. I don't waste my time with TTEs. John, we can talk about that. Because if you find a PFO, you don't know how big it is, and you can't see the septum well on a TTE, so you've gotta do a TEE anyway. If you don't find a PFO, you don't know that you haven't missed one 'cause you only find half of them with a TTE. Very quickly that was my TEE editorial. "What caused my stroke?" This is what patients are asking. I think we're much better off now if we exclude the other causes, calculate a RoPE and PASCAL score, and then communicate the probability of PFO-related stroke. "What's my risk of having another one?" I think we're better off now at predicting the risk of recurrence with the SCOPE and PASCAL scores. "And what can be done to reduce my risk?" There is a treatment effect that we can now articulate to patients with a little more specificity by the PASCAL score. So, what do we know for sure? I'm really almost finished. Not all PFOs are pathogenic. RoPE and PASCAL scores can identify the likelihood of PFO relatedness. PFO closure is associated with fewer recurrent strokes, especially in PASCAL-defined subgroups. Devices are low risk but not no risk. It may never come to this, but this is where we need to be. Cardiologists and neurologists need to work closely together here. I'm gonna skip through these suggestions. You can read them if you like. <v Coylewright>Thank you very much, Dr. Thaler.</v> <v Thaler>Choose wisely.</v> Thank you. (audience applauds)

PFO

stroke patients

PFO closure

recurrent stroke

shared decision making