Does All AF After Stroke Warrant Anticoagulation? Experts Debate

When atrial fibrillation (AF) is detected after a cerebrovascular event, it should be treated with oral anticoagulation: that’s the general consensus of stroke experts speaking in a session on secondary stroke prevention at the recent International Stroke Conference in Dallas, TX,

But that’s not an absolute, Karen Furie, MD (Brown University, Providence, RI), said during a debate, highlighting uncertainties around what burden/duration of AF conveys a risk of future stroke, the trustworthiness of brief AF episodes detected on implanted devices, and the temporal relationship between subclinical AF and stroke.

“When I was first invited to take the position that we shouldn’t anticoagulate patients with atrial fibrillation, I have to tell you, it gave me some pause, but as I thought about it more, I realized that in fact there are patients who have some degree of atrial fibrillation by history or that we detect and it doesn’t make sense to anticoagulate them,” she said. “So I hope . . . you’ll appreciate that we have to be a little nuanced in how we approach this, and we certainly need more data on this topic.”

Yes, Any Burden of AF Should be Treated

Presenting the “pro” side of the debate, Roland Veltkamp, MD (Imperial College London, England), reviewed what’s known about how often AF is found using long-term cardiac monitoring after a stroke, how monitoring-detected AF affects use of oral anticoagulation, and the impact of AF on clinical outcomes, including recurrent stroke and mortality.

The rate of detection of paroxysmal AF after stroke varies depending on how an episode is defined, but for the purposes of his talk, Veltkamp limited it to AF lasting at least 30 seconds. It also depends on the type and timing of monitoring, the targeted population, and whether monitoring is done as part of a randomized trial.

Veltkamp pointed to data reviewed in recent guidelines from the European Stroke Organisation (ESO) on screening for subclinical AF after stroke or TIA of undetermined origin that show monitoring picks up AF in about 9% of patients over 30 days and 23% over 1 year. And rates were even higher in the EMBRACE and CRYSTAL-AF trials of patients with cryptogenic stroke, which used 30-day external event recorders and an implanted event recorders, respectively.

A meta-analysis published last year in Neurology confirmed that prolonged cardiac monitoring increases the detection AF after a stroke compared with conventional methods. Also, a day before the debate, meeting attendees heard about the results of the STROKE-AF study out to 3 years, affirming that more and more subclinical AF episodes are detected the longer patients are followed.

Moreover, data from that meta-analysis and the ESO guideline show that the detection of AF by long-term cardiac monitoring significantly increases the proportion of patients who receive oral anticoagulation.

The question, then, is whether these efforts ultimately improve patient outcomes. A meta-analysis in the ESO guideline indicates that mortality is unaffected, although there is some evidence from the Find-AF trial and other data showing that recurrent stroke risk is reduced.

“We get a clear hint that it is effective and does prevent stroke recurrence, presumably by raising the number of patients who are anticoagulated,” Veltkamp said.

Maybe Not All Patients Need Anticoagulation

Opening the “con” side of the debate, Furie said, “Fundamentally, we all agree that patients who have stroke and TIA and atrial fibrillation should, for the most part, be receiving anticoagulation. That is the large proportion of all patients that we’re talking about.”

But there may be “very few” situations in which consideration could be given to alternative treatments or interventions, she said, noting, “Any time you’re not anticoagulating a patient who’s already had symptomatic cerebrovascular disease and atrial fibrillation, and you’re deciding not to use the gold-standard evidence-based therapy, you need to have a strong rationale that’s well documented in the medical record.”

One reason may be related to uncertainty about how AF burden is defined and about what burden threshold actually conveys a risk of stroke. For instance, are two 60-minute episodes of AF equivalent to 10 episodes lasting 12 minutes each when it comes to stroke risk? “Evidence suggests that fewer episodes of longer duration may convey higher risk, but none of this has really been proven yet and that’s the direction the field is moving in.”

Though conventional wisdom would suggest that more AF is worse for stroke risk, evidence is lacking regarding the best cutoff, Furie said.

And when it comes to duration of AF episodes detected on implanted devices, prior studies have used a host of different thresholds, although they seem to cluster around 5 or 6 minutes. In the ASSERT trial, subclinical atrial tachyarrhythmia was defined as an atrial rate greater than 190 bpm for more than 6 minutes, and this was associated with greater risks of both clinical AF and ischemic stroke/systemic embolism.

That 6-minute threshold has persisted in the literature since then, Furie said, noting that it has been supported by other studies over the years. Still, other analyses have not shown a linear relationship between the duration of AF episodes and stroke/TIA risk.

“It begs the question as to whether these bins of time that we’ve been looking [at] and assigning risk to may in fact not be the correct approach for trying to calculate the risk of A-fib burden,” Furie said.

As for whether short, 30-second AF episodes detected on implanted devices can be trusted, she noted that there is the potential for both false and missed detections.

On top of those uncertainties, studies have shown there is a disconnect between subclinical AF episodes and future strokes. “Yes, you can find atrial fibrillation if you monitor patients. Yes, the patient may have had a stroke or TIA. But are the two things related?” she said, noting that in ASSERT, only 25% of patients had AF in the 30 days before a stroke. In TRENDS, the figure was 30%. “So did the A-fib cause the stroke? The jury is out.”

The Audience Weighs In

To get a sense of whether there were situations in which attendees would not anticoagulate in response to AF detected after a stroke, Furie polled the audience on two scenarios. The first was a 60-year-old man with multiple comorbidities who had two small-vessel strokes a year apart. An implantable cardiac monitor was placed in response to palpitations and 30 months later, there was a 1-minute AF episode recorded. Slightly more than half of the respondents (52%) said they would not anticoagulate in this situation.

The second case involved an 80-year-old man with hypertension and history of smoking who presented with a right cerebellar stroke 2 years after undergoing CABG. After the surgery, the patient had 24 hours of postoperative AF but had remained free of cardiac symptoms and had not had AF detected on annual ECGs. After the stroke, he had normal sinus rhythm on telemetry throughout the hospitalization. In this case, 74% of respondents said they would not anticoagulate.

“I claim victory that there is a burden of atrial fibrillation where the majority of people would opt not to anticoagulate,” Furie said.

Agree More Than Disagree

Veltkamp rebutted some of Furie’s arguments by pointing out that some of the studies she cited, including ASSERT, were in primary prevention populations.

Detecting AF, and then treating with anticoagulation, takes on greater importance after a patient has had a cerebrovascular event, he suggested. To back that up, he noted that in the AVERROES trial, using a direct oral anticoagulant over aspirin had a greater absolute benefit in a secondary prevention population than in a primary prevention population.

“I would suggest that we rather stick with the evidence that we have so far, and in case of doubt, go for anticoagulation,” Veltkamp said. Still, he said, “evidence from ongoing RCTs is needed to build an even stronger case for oral anticoagulation.”

Closing out the debate, Furie said, “I think you and I agree more than we disagree.”

The issue, she said, “is that the drugs we use to anticoagulate patients are expensive and they have some risk associated with them, so we have to be cautious in prescribing them in a reflexive manner just because a report comes back that 30 seconds of atrial fibrillation has been detected on an implanted device.”

Furie said she hopes that in the future, there will be an improved ability to quantify AF burden and identify patients with brief AF episodes who require anticoagulation for secondary stroke prevention.

“It’s heartening to me that our conception of how to monitor patients poststroke is something that continues to expand and evolve so that more patients will potentially benefit from appropriate institution of secondary stroke prevention medications.”