Device-Detected AF: What Is the Stroke Risk and When Should Anticoagulants Be Used?

Daniel Singer, MD

Daniel Singer, MD

At the core of these concerns is the relationship between the length and frequency of AF episodes, and the stroke risk they pose.

Dr. Daniel Singer's lecture at AF2017 addressed whether patients with brief episodes of device-detected AF should be treated with anticoagulants.

With widespread use of devices that provide continuous heart rhythm monitoring for extended time periods, physicians face the question of whether patients with brief episodes of device-detected atrial fibrillation should be treated with anticoagulants. Guidelines have recommended that paroxysmal atrial fibrillation (PAF) should be viewed as equivalent to more sustained AF in terms of stroke risk and need for anticoagulants. But the studies supporting these guidelines were based on clinically detected AF, which generally reflected a higher burden of AF than device-detected AF.

“At the core of these concerns is the relationship between the length and frequency of AF episodes, and the stroke risk they pose,” Daniel E. Singer, MD (Massachusetts General Hospital, Boston, MA) told AF Symposium News. “We don’t yet have a precise description of this relationship, but studies over the past decade provide substantial insight into the problem. In brief, short episodes of AF (e.g., less than an hour on any given day) appear to pose much less stroke risk than longer episodes.”

“While the original trials of warfarin for AF found that PAF conferred the same risk as persistent or permanent AF, analyses from larger and more recent studies (ACTIVE A and AVER- ROES trials) have found that clinically categorized PAF poses about half the stroke risk as permanent AF,” said Dr. Singer. “But clinically quantified AF burden is highly inaccurate. Long-term continuous heart rhythm monitoring is needed.” The TRENDS study enrolled 2,486 patients with implanted cardiac devices (about one-half pacemakers) and a history of AF, and followed them for episodes of atrial tachycardia/atrial fibrillation (AF) and thromboembolic events. During an average follow-up of 1.4 years, only 24% had any AF, and most of these episodes were short. AF burden was defined as the number of hours of AF on any given day. The median AF burden was 5.5 hours. Forty thromboembolic events were observed during follow-up, although 17 were transient ischemic attacks (TIAs). Patients with ≥5.5 hours had an absolute thromboembolism rate of 2.4% per year, while patients with <5.5 hours had a rate of 1.1% per year, essentially the same as the rate for patients with no AF. The adjusted hazard rate for ≥5.5 hours of AF versus none was 2.2 (note P=0.06).

Singer-Slide-Graphics-3 (1).png

Dr. Singer added: “What is not widely appreciated is that in the TRENDS analysis, the increased rate of thromboembolism with high-burden AF decreased to the non-AF stroke risk by 30 days following the AF episode. That is, the effect of AF did not persist long after the episode.” (Figure 1) The ASSERT study followed 2,580 patients with implanted devices (95% pacemakers). Over an initial observation period of 3 months, 261 patients had at least >6 minutes of AF. During a mean follow-up of 2.5 years, 51 outcome events were observed, with 11 occurring among patients who had AF during the initial 3-month period. The adjusted hazard ratio was 2.5 comparing patients with >6 minutes of AT to the others. “Much of this effect was due to the quartile of patients who had >17.7 hours of AF,” explained Dr. Singer. “Most events occurring among those with AF in the initial 3-month period did not have AF preceding their event, prompting the hypothesis that AF permanently altered the left atrium to become prothrombotic.” (Figure 2)

The VA CareLink study linked Veterans’ Administration (VA) databases with Medtronic CareLink continuous heart rhythm records to identify 187 cases (98% ICDs) of ischemic stroke with 120 days of heart rhythm recordings pre-stroke. AF burden, using the TRENDS’ cutoff of 5.5 hours, during the 30 days immediately pre-stroke (‘case’ period), was compared to the same patients’ AF exposure on days 91-120 pre-stroke (‘control’ period). A total of 156 patients had no AF in either period and 15 had AF in both periods, but 16 had ‘informative’ discordant results. Thirteen of these 16 had AF in the prestroke period and no AF in the control period. “Analysis revealed that the adjusted odds for stroke were raised 4.2-fold in the presence of AF,” said Dr. Singer. “Additional analysis demonstrated that the odds were highest in the presence of AF and diminished to background risk by 30 days post AF.”

“These, and additional, studies argue strongly that stroke risk is a function of AF burden,” concluded Dr. Singer. “A conservative interpretation would be that AF episodes shorter than one hour on any given day pose little increase in stroke risk and do not merit treatment with anticoagulants despite the presence of a CHA2DS2-VASc score ≥2. The TRENDS and VA CareLink studies argue that anticoagulants are not needed 30 days after the end of an AF episode, while the ASSERT study suggests that there is residual increase in stroke risk long after the AF episode has ended. Randomized trials currently in process and in the planning stage may determine whether uninterrupted anticoagulation to prevent AF-associated stroke is beneficial for brief device-detected AF, and also whether intermittent anticoagulation coinciding with AF episodes is sufficient. It seems clear that the guidelines’ position that the full range of PAF is equivalent to sustained AF needs thoughtful revision.”

Tammy Griffin-Kumpey