January 20th, 2012

Subclinical Atrial Fibrillation and the Risk of Stroke: An Interview


Editor’s Note: In this edition of The Expert Is In, NEJM editorial fellow Daniela Lamas interviews her father, Gervasio Lamas, Professor of Clinical Medicine at the College of Physicians and Surgeons, Chair of Medicine, and Chief of the Columbia University Division of Cardiology at Mount Sinai Medical Center in Miami Beach. The senior Lamas was the author of a recent NEJM editorial on the ASSERT study of subclinical atrial fibrillation (AF) and the risk of stroke.

Daniela Lamas: A 70-year old woman presents to your office for her routine cardiology follow-up. She has a history of hypertension, and had a pacemaker placed two years prior for sick sinus syndrome. Although she had sick sinus, manifested by sinus bradycardia with fainting spells, she had never experienced supraventricular tachycardia. She says she’s been doing well. Her pacemaker has not been checked recently, so you decide to interrogate the device. You review the impedance, threshold and percent of time spent in ventricular pacing. Although she has been asymptomatic, the interrogation reveals an episode of atrial rate to 200 bpm lasting 7 minutes.

Should this finding affect your management?

The results of the ASSERT trial suggest perhaps it should. In this large-scale study, investigators interrogated recently implanted pacemakers or defibrillators in 2,580 patients without a history atrial fibrillation. They recorded asymptomatic episodes of atrial high rate for the initial three months of the study. The patients were then followed for another two and a half years to see if they developed strokes or peripheral emboli. The study authors found that the patients with any episode of atrial tachyarrhythmia — defined as rate greater than 190 beats per minute for over 6 minutes — had more than double the annual risk of stroke or peripheral emboli. They also were much more likely to develop symptomatic atrial fibrillation or flutter during study follow-up.

While compelling, these data do not clearly point the way toward clinical action, wrote Gervasio Lamas, in his editorial that accompanied the ASSERT trial.

Daniela: Can you give us some background for this finding? Is ASSERT the first trial to show such an association between asymptomatic atrial high rate and risk of stroke?

Gervasio Lamas: The background that gives context to this finding comes from two different disease processes. In the neurologic literature, there are patients with what’s called cryptogenic stroke, which, as the name suggests, is a stroke whose etiology is somewhat mysterious. We’ve all encountered these patients. Neurologists have published long-term cardiac monitoring of these patients and found that a significant proportion of patients with cryptogenic stroke have asymptomatic atrial fibrillation or atrial flutter. Presumably, that was a cause of stroke in those patients.

The other side of the evidence comes from the cardiology literature. As a field, we have become very interested in the clinical significance of asymptomatic atrial fibrillation. This was not really possible until the advent of modern cardiac pacing and defibrillation. These devices have extensive programmable memories. Interrogating a device, either a pacemaker or ICD, is like downloading data from a long-term holter monitor.

Once we started identifying these high atrial rate events, we started trying to figure out what they meant. Two findings began to pop out. First, those patients with atrial high-rate events are a little sicker and older than those without. But of course that’s a confounder when we try to attribute cause-and-effect. Those patients tend to have more afib and more strokes and, in one study, were more likely to develop heart failure or die.

We have come to believe that these episodes certainly are significant, but is there a causal link between episodes of atrial high rate and adverse events? And if you reduce the cause, can you reduce the adverse effect?

Daniela: How well have we established cause and effect?

Gervasio: That’s where the studies TRENDS and ASSERT come in. TRENDS was an observational study of patients with implantable devices. Glotzer and co-authors studied the association between atrial high rate and stroke 30 days after the event. There was a strong trend toward doubling the risk of stroke for 30 days after a day with more than five hours of afib. It’s a little fuzzy, and was underpowered, but this study shows how we need to establish both the association and a temporal link.

It’s not plausible that six minutes of atrial high rate today would double your rate, without any further dysrrhythmia, of stroke a year from now. The assumption is that if you have this high rate today, you’ll have another one six months from now, and that will be the one preceding your stroke. We can’t say that in ASSERT, or in the ancillary study from MOST that also suggests this association, but we get a hint – though underpowered – that this is true in TRENDS.

ASSERT is not the first study that has tried to develop this link, but it is certainly the largest study and the one with the most statistical power. However, there are logical steps that still prevent us from saying ok, six minutes of an atrial high rate event equals warfarin or Pradaxa. We’re not quite there yet.

Daniela: Another hypothesis you propose, in your editorial, is that atrial high rate is marker for some underlying dysfunction that also predisposes to stroke. Could you elaborate on this?

Gervasio: It may be that patients who have an atrial high rate event are more likely to have myocardial fibrosis due to LVH or prior MI. Because they have higher left atrial pressure, they have these atrial high-rate events. But the cardio-embolic source, rather than being the atria, could be the ventricle. There are authors who propose that patients with a higher CRP are more likely to have AF, so maybe this is just an index of an inflammatory state. I don’t know.

Daniela: How would you translate these data into clinical recommendations? Should the 70-year old woman in your office be placed on anticoagulation?

Gervasio: So, in terms of this particular lady what would I do? I’d step back and look at her CHADS2 score. This lady is 70 years old, hypertensive but does not have diabetes, not in heart failure and has never had a stroke. She has a CHADS2 score of one, so her risk is quite low. On the other hand, the atrial high-rate event has put her on the high side of thromboembolic risk for a patient with a CHADS2 score of 1. I would be prone, as a clinician, to put this lady on aspirin.

But this is the kind of decision you have to discuss with your patient. It is really not well founded on clinical trial data. People on aspirin can have GI bleeds — and my own Murphy’s Law experience is that the flimsier the evidence, the more likely you are to have a GI bleed. Of course, all possible therapeutic moves have their dark underside.

There is one situation in which my approach to the patient would be quite different based on the pacemaker telemetry, that is, if this patient, rather than being so healthy, had had a prior cryptogenic stroke thought to be embolic in nature. Then, knowing what we do now about the clinical significance of a brief episode of atrial fibrillation, I would place the patient on warfarin or other anticoagulant drug instead of an antiplatelet agent.

Daniela: What further studies could help us establish this relationship?

Gervasio: This study identifies a group of patients in which the risk of stroke is high enough that you could potentially conduct a randomized trial. Patients with atrial high rate events and a low enough CHADS2 score could be randomized. But that would be, like all multi-center large-scale studies, years in the making.

What we have today is a new risk factor for stroke in pacemaker and ICD patients. And the question, of course, whenever you develop a new risk factor, is whether this is association or cause and effect, and is it modifiable? That is what future studies will have to address.


One Response to “Subclinical Atrial Fibrillation and the Risk of Stroke: An Interview”

  1. Jonathan Newman, MD, MPH says:

    Great discussion/interview. However, the CHADS2 risk score might underestimate this patients’ risk, as her CHADS(2)VAS(2)C score would estimate an annual stroke risk > 3%, suggesting more than aspirin might be warranted. Might be interesting to consider how the applications ASSERT might vary by risk calculator estimates.