TRENDS: Defining a threshold for stroke risk in device-detected AF

Susan Jeffrey

April 03, 2008

Chicago, IL - Preliminary findings from a study that analyzed data on atrial tachycardia and atrial fibrillation (AT/AF) generated from implantable cardiac devices showed a relationship between AT/AF burden and stroke risk, independent of other stroke risk factors.

Patients with >5.5 hours of AT/AF on any given day in the past 30 days had an increased risk for stroke of more than twofold compared with those who had no episodes.

"Our results suggest that device-detected AT/AF burden may be used to stratify thromboembolic risk, but further studies are needed to precisely identify an AT/AF burden threshold that merits medical intervention," lead author Dr Taya V Glotzer (University of New Jersey School of Medicine and Dentistry, Newark) concluded.

Results from the TRENDS study were presented here during a late-breaking clinical trial session at the American College of Cardiology 2008 Scientific Sessions.

Copious data

Atrial fibrillation increases the risk for stroke. Currently, Glotzer said, paroxysmal AF is thought to have the same stroke risk as sustained AF, but the risk posed by very short durations of AF is unknown.

Implantable devices such as pacemakers and implantable cardioverter defibrillators (ICDs) can record and store unlimited and accurate AT/AF burden data, she said. "These devices have tremendous memory," Glotzer said. "If you see a patient every three months, you'll see every beat of afib the patient has," she said. "We're seeing more and more of it, but we don't know what to do with it."

The aim of this study was to begin to understand the relationship between device-detected episodes of AT/AF and the risk for thromboembolic events (TE). In addition, they hoped to determine a threshold value of AT/AF burden associated with increased TE risk.

All patients included had a class 1 or 2 indication for implantation of either a dual-chamber pacemaker, a dual-chamber ICD, or a dual-chamber cardiac resynchronization (CRT) device, as well as one or more stroke risk factors, including diabetes, hypertension, CHF, prior stroke or transient ischemic attack (TIA), and age 65 years or older.

Patients were seen at three-month intervals, when device diagnostics were downloaded, and clinical evaluation was done every six months.

The primary end point for TRENDS was the occurrence of a thromboembolic event, including stroke, TIA, or systemic embolism (SE). Events were adjudicated by a panel of three neurologists. The threshold for AT/AF episode detection was an atrial rate of >175 beats per minute lasting for at least 20 seconds.

To account for varying lengths of follow-up among the patients, ranging from one to four years, they analyzed the data in 30-day "rolling windows," Glotzer explained. AT/AF burden was defined as the longest total duration of AT/AF in hours on any given day during the 30-day rolling window.

All windows were categorized into those with zero AT/AF burden, those with low burden, and those with high burden; low and high burden were separated by the calculated median burden value, which in the case of this patient population was found to be 5.5 hours.

The overall study group included 2486 patients, about half of whom received a pacemaker and half an ICD. The average age was 71 years, 60% were male, 60% had CHF, and 76% had hypertension.

The mean CHADS2 score, a stroke risk score that awards a point value to each of five risk factors, including CHF, hypertension, age, diabetes, and prior stroke or TIA, was 2.2 in this population, "which is pretty high," she said. Of the patients, 21% were on warfarin and 62% were on aspirin at baseline.

During an average follow-up of 1.4 years, or 3382 patient-years, there were 40 TEs: 20 strokes, 17 TIAs, and three SEs. The annualized rate of TEs was 1.2%, she said, which was "remarkably low."

Of the rolling windows, they found that 76% had zero burden of AT/AF; 12% had low burden, defined as episodes lasting <5.5 hours of AT/AF; and 12% had high burden, or >5.5 hours.

Comparing the annualized TE rates for the zero-, low-, and high-burden subset of windows, they found that although the low-burden windows were associated with the same TE rate as zero-burden windows, high-burden windows showed double the annualized TE rate vs zero-burden periods.

Annualized thromboembolic event rate by AT/AF burden

AT/AF b urden Annualized r ate (%/y) Annualized r ate e xcluding TIAs (%/y)
Zero 1.1 0.5
Low 1.1 1.1
High 2.4 1.8

Using a Cox proportional hazard model, adjusting for baseline stroke risk factors and time-dependent AT/AF burden and time-dependent antithrombotic therapy, they found an increased risk for TE for the high-burden subset vs the zero-burden group, "telling us that if a patient has 5.5 hours of AT/AF or more in the preceding 30 days, they are 2.2 times more likely to have a TE event than if they had no AT/AF detected," Glotzer said.

Risk for thromboembolic events with low and high AT/AF burden vs zero burden

AT/AF b urden Hazard r atio 95% CI p
Low 0.98 0.34-2.82 0.97
High 2.20 0.96-5.05 0.06

"Our results suggest that device-detected AT/AF burden >5.5 hours on any day during the preceding 30 days doubles the risk for thromboembolism, independent of known risk factors and independent of antithrombotic therapy use," she concluded.

In an interview, Glotzer emphasized that these results are preliminary. "I think we can say for sure that higher burden has a higher stroke rate, but what exactly the cut point is, I don't think we know that yet," she said. "Should you put them on Coumadin? We can't say that for sure."

Association with CHADS2 score?

Dr Bruce D Lindsay (Cleveland Clinic, OH) was a member of the panel participating in discussion after the presentation. Asked for comment, Lindsay called the TRENDS data "interesting," with "two unresolved issues.

"One is that the burden may change over time, so a patient with a low burden could change categories later on," he said. "The other issue that requires further study is whether the AF burden is a better discriminator than the CHADS2 score that we customarily use to assess stroke risk."

Lindsay asked Glotzer about this latter point during the presentation. Glotzer pointed out that so far they have only the overall CHADS2 score of 2.2 for the group. It's not clear yet whether individual scores for those who experienced an event might be higher than for those who did not, and they intend to look at this issue further.

They also intend to look more closely at reasons for the low stroke rate in this group, who, overall at least, had a relatively high CHADS2 score. "We have actually taken all the cardiac compasses of every patient who's had a stroke, and we're looking to match them to age- and sex-matched controls, to see whether there's something about their afib burden that is different," Glotzer noted.

Also during the presentation, session comoderator Dr Bradley P Knight (University of Chicago Medical Center, IL) asked about the decision to use the 30-day rolling window, which he said seemed, "relatively nonintuitive."

"The idea of the 30-day window came about because we thought that AF proximal to an event would probably be responsible for a stroke event," Glotzer pointed out. "AF six months ago probably wouldn't cause a stroke that happened today. The other thought was that if perhaps we would put some sort of an alert into a device to let us know that AF is occurring, 30-day windows would be a reasonable thing to look at."

Glotzer reports consulting fees/honoraria from Medtronic and Arrhythmias/EP and speakers' bureau relations hips with Medtronic, St Jude, Boston Scientific, and Arrhythmias/EP. Lindsay reports no conflict of interest.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: