High rate of undiagnosed AF in cryptogenic stroke

Susan Jeffrey

February 13, 2013

Honolulu, HI - Results of a large new trial show a high rate of undiagnosed atrial fibrillation (AF) among patients with ischemic stroke or transient ischemic attack (TIA) of unknown cause [1].

The trial, known as 30-Day Cardiac Event Monitor Belt for Recording Atrial Fibrillation After a Cerebral Ischemic Event (EMBRACE), supports prolonged monitoring to identify patients with cryptogenic stroke who may benefit from prophylactic anticoagulation, the researchers say.

Dr David J Gladstone

"We found in this study one in six patients over age 55 years with a cryptogenic stroke or TIA had previously undiagnosed paroxysmal atrial fibrillation, the vast majority of which had been untreated with anticoagulants, and so we believe that some of these patients may be candidates who may benefit from this strategy," Dr David J Gladstone (University of Toronto, ON) concluded.

"The results suggest that we can no longer rely on one or two Holter monitors," he added. "Prolonged monitoring is feasible and significantly more effective, has an incremental yield over 30 days, and resulted in a significant increase in the number of patients who were anticoagulated, which we believe represents a clinically meaningful change in patient management."

The results were presented here at the International Stroke Conference (ISC) 2013.

Unexplained stroke

Identifying AF is a key priority in secondary stroke prevention because of the important treatment implications, Gladstone noted. "The problem is that atrial fibrillation can be difficult to detect when it's paroxysmal and asymptomatic, and we suspect it may be systematically underdiagnosed in patients with unexplained strokes and TIAs," he said.

Approximately one in four strokes is considered cryptogenic, he noted. Currently, standard evaluation uses repeat Holter monitoring for 24 to 48 hours, but it has a relatively low sensitivity for AF, about 5%. If the AF is paroxysmal, it may be missed, resulting in what might be otherwise-preventable strokes. Longer-term ECG monitoring is challenging because patient adherence has been an issue, he said.

Previous trials have been helpful but were relatively small and heterogeneous in their findings. EMBRACE was a randomized trial to determine the diagnostic yield of 30 days of home-based cardiac monitoring vs repeat Holter monitoring to detect paroxysmal AF in patients with an ischemic stroke or TIA deemed cryptogenic after a standard diagnostic workup, including an initial negative result on Holter monitoring.

The investigators used an event-triggered loop recorder programmed to record AF episodes (Accuheart , Cardiac Bio-Systems). The devices had a maximum recording capacity of 30 minutes, he noted, "so please bear this in mind when I show the results, that the maximum recording time per episode was two and a half minutes."

The device was worn using a novel nonadhesive electrode belt that could be placed under clothing, and it provided the "theoretical" advantage of being more comfortable for patients to wear, Gladstone said. Patients were asked to wear the monitor for as much of the day and night as was possible for a target of 30 days or until AF was detected. Downloaded recordings were blindly interpreted centrally, and a report was sent to the enrolling physician.

The primary outcome was detection of one or more episodes of AF or atrial flutter lasting 30 seconds or more. "We acknowledge that there isn't really consensus on the optimal or appropriate cutoff of duration of atrial fibrillation," he noted, but this definition is consistent with American College of Cardiology/American Heart Association/European Society of Cardiology published definitions. In addition, it "was considered by our steering committee to represent a potentially clinically significant finding in this population for whom there was a high pretest probability of atrial fibrillation."

In all, 572 patients were randomly assigned from 16 centers participating as members of the Canadian Stroke Consortium. The mean age was 73 years, and 63% of patients had had a stroke and 37% a TIA as the index event; the median CHADS score was 3 in each group. More than 90% of patients were ambulatory and independent, so strokes were mild, and secondary stroke prevention would be important, he noted. Adherence was high; over 80% of patients were able to wear the monitor for at least three weeks.

On the primary end point, significantly more patients were found to have newly detected AF in the 30-day monitoring group vs those who used repeat Holter monitoring. "We found a 13% absolute difference in the detection between the 30-day group and the repeat-Holter group," Gladstone said. "This translates into a number needed to screen to identify one additional patient with atrial fibrillation of eight."

EMBRACE: Primary end point

End point
Repeat Holter m onitoring (%)
30-d m onitoring (%)
AF > 30 s (within 90 d)

In each group, 1% of patients had AF detected through other means, probably as part of routine clinical follow-up, Gladstone said.

Longer durations of AF—greater than 2.5 minutes—were detected in 10% of the 30-day monitoring group vs 2% of the repeat Holter monitoring group, and 20% of the 30-day group had episodes of any duration vs 4% with the repeat Holter monitoring, both statistically significant differences (p<0.001).

The prevalence of AF detected was similar whether the index event was a stroke or a TIA, and the yield was highest in the prespecified subgroup of elderly patients older than age 75 years; in that subgroup, one in five patients was found to have AF.

Most events were detected within two weeks of monitoring, Gladstone noted, but incremental events were detected out to four weeks.

Anticoagulant use was low at baseline, but if AF was detected, 72% of patients received anticoagulation; treatment increased in the 30-day monitoring group from 5% to 18% by 90 days, compared with 10% in the repeat-Holter monitoring group.

The main limitation of the study is the inability to document the total duration or maximum duration because of the finite memory storage capacity of the device, which recorded only 2.5-minute episodes, he said. "So this could include episodes of eight minutes or eight hours or a longer duration; we just don't know, but we do suspect that we did underestimate the total burden of atrial fibrillation, because patients stopped monitoring early if atrial fibrillation was detected prior to 30 days."

Further, the investigators aren't able to conclude causation from the detection of AF in these patients, he noted, "and the clinical significance of short episodes of atrial fibrillation is still uncertain."

"Short-duration monitoring is useful and should be considered as an initial screening test, but if negative, we propose the following recommendation: that prolonged monitoring should be considered in elderly patients with cryptogenic stroke or TIA, when AF is suspected but unproven, for those patients who would be considered appropriate anticoagulant candidates if AF were detected," he concluded.

More research is needed, he added, to determine the yield of monitoring beyond 30 days, and the minimum burden of AF required to confer an increase in stroke risk sufficient to warrant anticoagulation. The one- and two-year follow-up of this group is ongoing, he said.

"Kindling" effect?

During the discussion of this presentation, one attendee revisited the point of the unknown magnitude of stroke risk associated with lower burden of AF. Gladstone presented an additional slide at that point, showing that 41% of patients who had AF had just a single event recorded, 18% had two events, and 41% had three or more separate events recorded.

The duration and frequency of episodes determine the burden of AF, he said, but again pointed out that the study was limited by the memory storage capacity of the devices.

"But if we draw on the previous published literature, it does seem that short-duration events may be a marker for patients who have longer-duration events at other times," Gladstone said. "We know this from pacemaker studies, and there's a growing belief that even short episodes may be a valid stroke risk factor."

We think atrial fibrillation begets atrial fibrillation.

He pointed to results of the Asymptomatic AF and Stroke Evaluation in Pacemaker Patients and the AF Reduction Atrial Pacing Trial (ASSERT) trial, showing that detection of one episode of AF six minutes or longer within three months of monitoring was a significant independent predictor of confirmed diagnosis of AF and future stroke risk.

"We think atrial fibrillation begets atrial fibrillation, like the kindling model in epilepsy, applied to the heart, and if you have short episodes at one point in time, that might be associated with more burden if you continue to monitor the patients for longer duration," Gladstone said.

"The concept is that atrial fibrillation is a progressive condition, beginning with paroxysmal atrial fibrillation, and over time, progressing to persistent and permanent atrial fibrillation, so there may well be a role for early detection," he concluded.

EMBRACE was funded by the Canadian Stroke Network and the Heart and Stroke Foundation of Ontario . The authors have disclosed no relevant financial relationships.


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.