Paradoxical embolism via patent foramen ovale (PFO) is a rare cause of stroke, but it’s not uncommon to find PFOs in patients without traditional stroke risk factors (about 1 in 4 people in the general population have a PFO). How should patients with no other convincing cause of stroke be counseled, especially if invasive PFO closure is being considered? We talked to Dr. David Thaler, creator of the Risk of Paradoxical Embolism (RoPE) Score, about his research and experience with taking care of patients with cryptogenic stroke.
Why did you develop the RoPE Score? Was there a clinical experience that inspired you to create this tool for clinicians?
PFOs have interested and frustrated me for years. They’re so common in the general population, and we find them all the time in stroke patients, old ones and young ones. And paradoxical embolism is definitely a thing—there’s no question that it happens—but because the prevalence is so high in the general population, there’s also no question that a lot of the PFOs that we find are incidental. That’s where this started from in my mind: How can I disaggregate from within a cryptogenic stroke population those PFOs which are important and those which are ignorable? My friend David Kent, who is the first author on that paper, is not a neurologist, but does predictive models, and heterogeneity of treatment effect is sort of his thing—we talked about this problem and it fit really well with his approach to solving clinical problems. I couldn’t have solved it without him.
It was really my clinical frustration with young stroke patients, largely, who have PFOs, trying to figure out whether they were important or not, to make a diagnosis. It’s not a treatment issue; it’s an issue of what’s the diagnosis. I think this gets us closer to a real diagnosis of PFO-related stroke, or an ignorable PFO, and they’re still in that cryptogenic category. That’s what led me to it.
Who should the RoPE score be used for?
It’s meant to be for people who have no other compelling grade I cause for stroke in whom a PFO is found. If you have atrial fibrillation, for example, it doesn’t protect you from having a paradoxical embolus, which is the argument that a lot of the aggressive PFO guys make. If you have 90% carotid stenosis, you could still have atrial fibrillation as the cause of your stroke. It is possible to have more than one mechanism, but the RoPE study and score was developed from within a combined cohort of different studies of cryptogenic stroke patients. They all had slightly different definitions, but basically it’s all patients without high grade atherosclerotic stenoses and without afib and without mechanical valves and without lacunar strokes and without lots of other things, and you’re just left not knowing what it is.
Have you heard of cases where the RoPE score is misapplied or used incorrectly?
I haven’t heard of [people using it before ruling out other causes of stroke], but I wouldn’t be surprised if people tried to do that. Or if they use the RoPE score before they do an echo to decide if they should do an echo. I have not heard that, but I wouldn’t be surprised if someone said “Oh look, their RoPE score is so low that they don’t even need an echo.”
This is meant to be applied for an individual who has had a stroke, with no known cause, a PFO is found, and you want to sit with that individual patient and say “Based on the RoPE score, the likelihood that your PFO caused you stroke is 90% or 50% of 30% or 0%”—that’s a useful starting-off point for that conversation.
I’ll tell you where it has been misapplied—and this is where it gets a little bit subtle—is deciding whose PFO should get closed. This is part of the reason why there’s a little bit of urgency now to get it out there. We presented the first positive trial of PFO closure versus medical therapy, where it is now statistically significant that PFO closure with the device that we tested had fewer recurrent strokes than the medically treated group. And based in part on those data, the FDA approved a device for PFO closure, the first one ever, as of October.
Given that, and the indication that the FDA put on the device, very clear, is that patient selection is crucial, because 25% of the population have PFOs: the idea is not to have 25% of the population undergo PFO closure, the idea is not to have a cardiologist see a dizzy patient, decide that was a TIA and then look for PFOs and find that 1 out of 4 of those patients have a PFO and then start closing them. That’s irrational dispersion of this technology. We want rational dispersion, based on the way the studies were done and what seems to make clinical sense.
So what some people have been doing is use the RoPE score to decide who should get treated. Now, that sort of is intuitive, but I think the RoPE score is more adjunctive than predictive in that way. It’s not like a CHADS score—it doesn’t tell you who is likely to benefit from one treatment over another. It tells you who has the disease: PFO-related stroke. It’s more like an EKG or a Holter monitor for diagnosing afib. And EKG or Holter monitor doesn’t tell you the CHADS score. The CHADS score is something else: predicting stroke from within a population who have afib.
So the RoPE score tells you who has PFO-related stroke, more or less. The next step, who should get closed, is still open to discussion. But everyone agrees—including the FDA, including the company who makes the device, including those of us on the steering committee who are trying to decide how we can instill rational dispersion the device—everyone agrees that patients should be selected and identified by neurologists, together with cardiologists, and the RoPE score is part of it.
Now that the device is out there, people are trying to work their way through “How do I identify the right patients?” and I think the RoPE score has been an important part of the conversation, that people keep bringing up as part of the conversation. And though I say it doesn’t predict response to treatment, it does help to identify the right people.
Is the implication that patients with lower RoPE scores should be considered for closure? Or those with higher scores? With regard to recurrence.
It’s kind of complicated. The ones with lower RoPE scores probably did not have a PFO-related index event. That’s how it works. So the risk of risk of recurrence is higher in lower RoPE score patients. So all that’s telling us is that the non-PFO-related stroke mechanisms have a higher risk of recurrence than PFO-related stroke mechanisms. So who’s in the low RoPE score group? Those are people who have small vessel disease, who have 50% atherosclerotic stenosis in the carotid and a not-very protrusive plaque in the aorta that would not get you into the diagnostic category, but still a potential stroke mechanism. So those non-PFO-related stroke mechanisms tend to have a higher risk of recurrence than the PFO-related ones. So you might argue that we want the low RoPE score patients because they have so many events (for a trial). But they’re having a lot of events that would never be affected by PFO closure. Because they’re not PFO-related.
But then you say, let’s only do the high RoPE score ones. The problem with that is, if you only look at the high scores, the event rate is so relatively low that you need huge numbers to get statistical significance to show that there’s a benefit. Although you’d be treating the right kind of people, and trying to prevent the right kind of strokes. And the other thing we’ve learned over the course of the last 15 years since the study was started, one of the assumptions going into the study was that those patients who had PFO-related index events would have PFO-related recurrences. And we know that about a third of all recurrences in cryptogenic stroke patients are no longer cryptogenic. So if that’s the case, then a third of all strokes that occur in this kind of population, could never be prevented with PFO closure. So even if the device worked perfectly, 100% prevention of PFO-related recurrences, in the appropriate patients, the best reduction in recurrence that we would expect, would be about 60%. We wouldn’t expect 100%. Because we’re not going to prevent the other (one-third) of non-PFO-related strokes in this population.
The answer to the question is, no, I don’t think low RoPE score patients are the right ones [for PFO closure], because they probably did not have PFO-related index events to begin with.
You mention, briefly in the paper, PFO characteristics likes shunt size, hypermobile atrial septum and shunting at rest. Are there any plans for future studies to incorporate this into the RoPE score?
FIrst of all, we were just looking at PFO as present or absent as the outcome when we developed the RoPE score, so the shunt characteristics and the atrial septal aneurysms were not part of it. We did use those characteristics in lots of other ways looking at predictors of recurrence. We’re doing a presentation at the Stroke meeting in February about what we called our RoPE recurrence score.
So the first step was to come up with the patients who had PFO-related index events, that’s the RoPE score. We then did a paper (Thaler 2014) that came up with predictors of recurrence, and the predictors of recurrence differ by RoPE score. So the high RoPE score patients, the ones who more likely had PFO-related index events, had echo characteristics that predicted recurrence, as you would expect. So low RoPE score patients, which more likely had incidental PFOs, the echo characteristics were uninformative with regard to recurrence, because their PFOs were just hanging around, and didn’t have anything to do with the first stroke, and didn’t have anything to do with the second stroke either.
That was a pleasing observation, that the RoPE score was doing what it was supposed to be doing. Predictors of recurrence, including atrial septal aneurysms, hypermobile atrial septum, and shunt size, were also predictive of recurrence. But—and this is why science is so much fun—it was the smaller shunt sizes which predicted recurrence. Almost 3 times as often as larger shunts. Which is counterintuitive and contrary to what anybody expected and most people don’t believe it. But, it’s either true or it’s not true. And if it’s true, there is a way that I can explain it that sort of makes sense to me, but it’s really hard to prove. Or it’s not true, and it’s just a messy variable, which I don’t think is likely because if that were the case, it would have just been a null result, rather than highly informative but in the opposite direction.
So what’s your hypothesis? Does it have something to do with turbulence?
Sort of, that’s a good thought. Remember, a PFO is not really a hole, it’s more of a tunnel, and if you have a large shunt, then there’s a lot of blood whooshing through this tunnel all the time. And that would increase, one would think, the likelihood of a thrombus that happens to arrive in the right atrium (RA), from a DVT, or even an SVT. This is only a millimeter or two we’re talking about here—it doesn’t have to be a big chunk of anything. So I don’t think there has to be a discovered DVT, or even an identifiable atrial one. You cross your legs, you sit in a car for too long, you twist your ankle and you get a bruise, you get any other venous thrombus of a millimeter or two and that can paradoxically embolize into the right atrium, which i think is normal. I think we all do that all the time.
If you have a larger shunt, it certainly makes sense that something in the RA would be more likely to squeak across than if you have a smaller shunt. But if you have a small shunt, than this PFO might be closed a lot of the time, and might only be open if the RA pressure is increased by coughing or sneezing or bending over or hydration status or something. And it might be that in situ thrombus formation and embolization is a riskier mechanism than is paradoxical embolus. In other words, there might be more than one PFO-related stroke mechanism.
The whole discussion so far has been about PFO-related stroke versus non-PFO-related stroke. But within PFO-related stroke there might be more than one way for PFOs to be related to stroke. Paradoxical embolus, which is the name of the score, isn’t that confusing, in my head even though I say paradoxical embolus, I’m always thinking “PFO-related stroke”. Because we don’t really know what the mechanism is. One of them could be in-situ thrombus. And so that might be more common in people who have smaller shunts versus larger shunts. That makes more sense to me, and it’s consistent with the data, and I don’t know how to prove it.
So how do you tell a patient who has a small shunt that they might need a procedure to close it, if it’s so small?
I tell patients that we don’t know how to interpret shunt size. There are a lot of cardiologists who say, “Oh, it’s a huge one, we gotta close it”. The implication being if it’s a small one, then we shouldn’t close it. Which is not consistent with our data. I’ve made this joke in talks before, but the next study should be randomizing patients to PFO dilation, to see if we can reduce their risk of stroke. I don’t think cardiologists would want to do it (laughs) but that’s the way the data should drive it.
So what we’ve done is we’ve taken these predictors of recurrence in the high RoPE score group and we’ve come up with what we’ve called the RoPE recurrence score, which is essentially taking high RoPE score patients and adding predictors of recurrence and saying these are the ones that would most benefit from PFO closure. And we have all 3 of the clinical trials that have been published on this, we have all the individual patient data. So we could test the hypothesis that the RoPE recurrence score was predictive of the ones who were most likely to benefit. So there should be heterogeneity of treatment effect amongst those with high RoPE recurrence score compared to those with low RoPE recurrence scores. And when we did that, it seemed to be going in the right direction, but it was not significant. And I think there’s two reasons for that. One is the number of outcomes was pretty small, so there wasn’t much power to do it, and the second is that in the randomized trials that formed the model testing database [validation cohort] pts were pretty well selected, so the average RoPE score was about 7, and there wasn’t that much variability. So the scatter was probably not wide enough for us to see the differences in treatment effect based on the RoPE recurrence score. So we might need to come up with other ways to test that. But that could become the sort of CHADS score for PFO-related stroke, which you could use for prediction of treatment effect.
ABOUT THE CREATOR
David Thaler, MD, PhD is Neurologist-in-Chief at Tufts Medical Center and Chairman of the Department of Neurology at Tufts University School of Medicine. He has directed the Comprehensive Stroke Center at Tufts Medical Center since 1998. His clinical and research interests include stroke in the young and patent foramen ovale.