JOURNAL OF THE AMERICAN COLLEGE OF CARDIOLOGY
VOL. 65, NO. 3, 2015
ª 2015 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION
ISSN 0735-1097/$36.00
PUBLISHED BY ELSEVIER INC.
http://dx.doi.org/10.1016/j.jacc.2014.10.045
THE PRESENT AND FUTURE STATE-OF-THE-ART REVIEW
Device-Detected Atrial Fibrillation What to Do With Asymptomatic Patients? Carol Chen-Scarabelli, PHD,* Tiziano M. Scarabelli, MD, PHD,y Kenneth A. Ellenbogen, MD,z Jonathan L. Halperin, MDy
ABSTRACT Atrial fibrillation (AF) is the most common clinically significant arrhythmia and conveys an increased risk of stroke, regardless of whether it is symptomatic. Despite multiple studies supporting an association between subclinical atrial tachyarrhythmias (ATs) detected by cardiac implantable electronic devices and increased risk of thromboembolic events, clinical intervention for device-detected AT remains sluggish, with some clinicians delaying treatment and instead opting for continued surveillance for additional or longer episodes. However, the 2014 updated clinical practice guidelines on AF recommend use of the CHA2DS2-VASc stroke risk score for nonvalvular AF, with oral anticoagulation recommended for scores $2, regardless of whether AF is paroxysmal, persistent, or permanent. This paper reviews the epidemiology of AF and mechanisms of stroke in AF, and discusses device-detected AF and its clinical implications. (J Am Coll Cardiol 2015; 65:281–94) © 2015 by the American College of Cardiology Foundation.
A
trial fibrillation (AF) is the most common
AF, which is thought to affect as many as one-
clinically significant heart rhythm disorder
third of the U.S. population (3).
(1), with an estimated lifetime risk of 22%
to 26% or about a lifetime risk of 1 in 4 (2). It has
MECHANISMS OF AF
been diagnosed in >2.5 million people in the United States alone (3). In 2010, the incidence of diagnosed
The pathophysiology of AF is multifactorial and
AF in the United States was 1.2 million, and its
complex, including both genetic and neural mecha-
prevalence is projected to increase to >12 million
nisms. The main mechanism by which autonomic
cases by 2030 (4). In the European Union, there
activation triggers AF is activation of the sympathetic
were 8.8 million adults >55 years of age with AF
and parasympathetic nervous system, which likely
in 2010, with an expected increase to 17.9 million
interact with the pulmonary vein–left atrial (LA)
by 2060 (5). Globally, AF incidence in 2010 was esti-
junction to trigger atrial ectopy (7). Genetic mecha-
mated at 33.5 million (20.9 million men and 12.6
nisms linked to AF development include alterations
million women). Despite a higher incidence in
in potassium or sodium channels, connexin expres-
men, mortality associated with AF is greater in
sion or function (2), and microRNAs (8). Four major
women, doubling between 1990 and 2010 (6). These
mechanisms that promote focal ectopic firing and
statistics do not account for silent or undiagnosed
reentry substrate formation have been implicated in
From the *Veterans Affairs Ann Arbor Health Care System, University of Michigan, Ann Arbor, Michigan; yZena and Michael A. Wiener Cardiovascular Institute, Mount Sinai Medical Center, New York, New York; and the zDepartment of Medicine, Virginia Commonwealth University School of Medicine, Richmond, Virginia. Dr. Ellenbogen has served as a consultant for Medtronic, Boston Scientific and St. Jude Medical, and has received research, honorarium, and fellowship support from Medtronic and Boston Scientific. Dr. Halperin has served as a consultant for Bayer Healthcare, Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi Sankyo, Pfizer, Inc., Biotronik, Boston Scientific, and Medtronic. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Ellenbogen and Halperin contributed equally to this work as senior authors. Sumeet Chugh, MD, served as Guest Editor for this paper. Manuscript received June 29, 2014; revised manuscript received October 14, 2014, accepted October 14, 2014.
282
Chen-Scarabelli et al.
JACC VOL. 65, NO. 3, 2015 JANUARY 27, 2015:281–94
Device-Detected Atrial Fibrillation
ABBREVIATIONS
AF: 1) ion channel dysfunction; 2) calcium
conventional ECG group (18). Although anticoagulant
AND ACRONYMS
handling abnormalities; 3) structural remod-
prescription for AF was higher in the ICM group
eling (primarily atrial fibrosis); and 4) auto-
versus the routine ECG monitoring group (10.1% vs.
nomic neural dysregulation (2,8). These 4
4.6%) at 6 months, 97.0% of patients with detected
conditions not only trigger AF, but may also
AF were receiving oral anticoagulant agents by the
result from episodes of AF, supporting the
12-month follow-up (18).
AF = atrial fibrillation AHRE = atrial high rate episodes
AT = atrial tachyarrhythmia CIED = cardiac implantable electronic device
concept that “atrial fibrillation begets atrial
A similar study, the EMBRACE (30-Day Cardiac
fibrillation,” first reported in an early animal
Event Monitor Belt for Recording Atrial Fibrillation
study documenting atrial electrical remodel-
After a Cerebral Ischemic Event) study, compared
resynchronization therapy
ing in AF (9). Further advances in knowledge
new AF detection by noninvasive ambulatory ECG
CS = cryptogenic stroke
of the pathophysiology of AF have revealed
monitoring with either a 30-day event-triggered
that electrical remodeling in AF is not limited
recorder (intervention group) or a conventional 24-h
to the atria. More pronounced remodeling
monitor (control group) in 572 patients with CS
after brief episodes of induced AF has been
within the preceding 6 months, without a history of
monitor
documented in the pulmonary veins (10),
AF (19). The investigators reported a greater than
LA = left atrium/atrial
thereby extending the concept to “AF begets
5-fold increase (16.1% vs. 3.2%; p < 0.001) in AF
AF in the pulmonary veins”.
detection in the 30-day event monitor group, with a
AF AND STROKE
prescription (18.6% vs. 11.1%; p ¼ 0.01) among the
CRT = cardiac
ECG = electrocardiogram EGM = intracardiac electrogram ICM = implantable cardiac
LAA = left atrial appendage
subsequent significant increase in anticoagulation
TE = thromboembolic event(s)
30-day event monitor group. At the 90-day followAF is a major independent predictor of ischemic
up, 87% of patients with AF in the event monitor
stroke, resulting in a 5-fold increase in risk (1). Each
group and 100% of patients with AF in the control
year,
experience
group were on anticoagulant therapy (19). Thus,
strokes, of which 610,000 are first strokes and
both the CRYSTAL-AF and EMBRACE studies docu-
approximately 87% are ischemic. In the United States,
mented a significant increase in anticoagulant pre-
someone suffers a stroke every 40 s (that is, approx-
scription in CS patients with newly detected AF.
imately 90 people/h) (1). Among patients with AF, it
However, anticoagulation treatment rates are signi-
is estimated that every hour, 15 will have a stroke (11),
ficantly lower for patients without a prior history of
and such AF-related strokes impose a higher mortal-
stroke with newly detected AF on cardiac implantable
ity than strokes unrelated to AF (12). The prevalence
electronic devices (CIEDs). One retrospective study
of AF and associated stroke risk are highest among
reported a 50% incidence of pacemaker-detected
approximately
795,000
people
elderly patients, with stroke risk independent of
AF, yet 90% of thrombi in AF (25), it
in patients with nonvalvular AF (24), accounting
may be a significant strategy in stroke prevention
for >90% of thrombi (Figure 1). This thrombogenic
in nonvalvular AF.
tendency has led to targeted interventions to occlude the LAA in an attempt to reduce stroke risk in AF, as
DEVICE-DETECTED AF
reported in trials such as the PLAATO (Percutaneous Left Atrial Appendage Transcatheter Occlusion) trial
Subclinical atrial tachyarrhythmias (AT) can be
in 2002 (25) and, most recently, the PROTECT AF
detected by various cardiac monitoring methods,
(Watchman Left Atrial Appendage System for Embolic
including external surface monitoring (e.g., standard
Protection in Patients With Atrial Fibrillation) study,
12-lead electrocardiogram, ambulatory Holter moni-
which documented noninferiority of LAA occlusion to
tors, event monitors) and by CIEDs (e.g., implantable
systemic anticoagulation (26). In fact, the PROTECT
cardiac monitors, dual-chamber pacemakers, dual-
AF trial was credited with being the first trial to
chamber implantable cardioverter-defibrillators, car-
demonstrate involvement of the LAA in the patho-
diac resynchronization therapy [CRT] devices), many
genesis of stroke in AF (27).
of which enable remote monitoring. This review
Significant safety concerns for WATCHMAN im-
addresses only CIEDs, given their continuous moni-
plantation, including pericardial effusions and device
toring capability. AT commonly occurs in patients
embolization, were addressed in a subsequent trial,
with CIEDs and is associated with an increased risk
the PREVAIL (Watchman LAA Closure Device in Pa-
of thromboembolism (TE) (1). Several studies have
tients With Atrial Fibrillation Versus Long Term
correlated TE risk with the total duration or burden
Warfarin Therapy) trial, which documented a signifi-
of device-detected AT (32–34). However, there are
cantly lower rate of adverse events compared to the
presently no published randomized clinical studies
PROTECT AF trial (4.2% vs. 8.7%; p ¼ 0.004) (28).
investigating treatment of AT detected by CIEDs.
Although the third review by the Food and Drug
All cardiac rhythm recordings obtained from CIEDs
Administration Circulatory System Devices Advisory
require adjudication or review by a qualified clinician
Panel in October 2014 (29) resulted in a unanimous
to verify diagnostic accuracy. Retrospective review of
vote on safety of the device, analysis of the updated
device-derived data has confirmed that most of these
June 2014 PREVAIL dataset demonstrated new
tachyarrhythmias represent paroxysmal AF or atrial
ischemic strokes occurring more than 1 year after
flutter. However, false detection may occur due to
WATCHMAN device implant. Furthermore, neither
far-field R-wave (Figure 2) oversensing by the atrial
the first primary endpoint of the PREVAIL trial
lead (35–37) or runs of premature atrial complexes.
(composite 18-month rate of stroke, cardiovascular or
False negative or missed AF has been reported when
unexplained death, and systemic embolism) nor the
episodes of AF are very brief (35,37). In addition,
283
284
Chen-Scarabelli et al.
JACC VOL. 65, NO. 3, 2015 JANUARY 27, 2015:281–94
Device-Detected Atrial Fibrillation
antitachycardia responses of CIEDs are not spe-
Pacemaker Patients and the AF Reduction Atrial
cific for AF (35,37) and may be triggered by other
Pacing) trial, the positive predictive value of AHREs
forms of AT, including atrial tachycardia or atrial
for EGM-confirmed AF was examined in 2,850 sub-
flutter. Thus, intracardiac electrograms (EGMs) must
jects with implanted pacemakers. In 17.3% of cases,
be reviewed to verify the accuracy of the device di-
AHRE episodes at >190 beats/min lasting >6 min
agnostics. Device-stored data based solely on marker
were found to be falsely positive, due predominantly
channels, without EGMs, cannot be used to verify AF
to repetitive non–re-entrant ventriculoatrial syn-
due to the potential for diagnostic errors caused by
chrony
oversensing or undersensing by the atrial lead.
desynchronization arrhythmia (Figure 4). Repetitive
Furthermore, atrial tachycardia detection rate pro-
non–re-entrant ventriculoatrial synchrony is trig-
gramming and the duration of the post-ventricular
gered by retrograde ventriculoatrial conduction with
atrial blanking interval can also influence the num-
functional atrial undersensing. It results from retro-
ber of automatic mode-switching episodes in the
grade atrial activation during the post-ventricular
setting of AT (38). Although ICM are also susceptible
atrial refractory period and functional atrial non-
to false AF detection due to oversensing or missed AF
capture due to atrial stimulation during the absolute
detection due to undersensing, 2 ICM with AF algo-
refractory period, with the potential to trigger mode
rithms (Medtronic Reveal XT, Model 9529, Medtronic
switching (41–45).
(40),
also
known
as
atrioventricular-
Inc., Minneapolis, Minnesota, and, SJM Confirm Implantable Cardiac Monitor Model DM2102, St. Jude
REVIEW OF PUBLICATIONS ON
Medical, Inc., Sunnyvale, California) are currently
DEVICE-DETECTED AF
available on the market, with the Medtronic Reveal XT reported to have an overall accuracy of 98.5% in
Because the advent of dual-chamber devices and
AF detection (39) (Figure 3). Although atrial high rate
ventricular leads with atrial sensing capability, the
episodes (AHRE) have been used as a surrogate for
clinical implications of device-detected AT have been
AF, the data must be interpreted with caution. In the
considered in the context of anticoagulation for
ASSERT (ASymptomatic AF and Stroke Evaluation in
stroke prevention (46), but the question of what to
F I G U R E 1 Mechanisms of Stroke in Atrial Fibrillation
Cardioembolic sources, almost exclusively represented by left atrial appendage thrombi, account for >90% of embolic events. Noncardioembolic origin, more often embolic material dislodged from thoracic and or carotid plaques, account for the remaining 10% of events. Graphics source: National Institutes of Health/National Heart, Lung, and Blood Institute.
Chen-Scarabelli et al.
JACC VOL. 65, NO. 3, 2015 JANUARY 27, 2015:281–94
Device-Detected Atrial Fibrillation
F I G U R E 2 EGM Representative of AT/AF Due to Far-Field R-Wave Sensing
Far-field R-wave sensing on the atrial lead: AR on EGM 1 (atrial EGM) correlates with VS on EGM 2 (ventricular EGM). R-wave is sensed on the atrial channel, triggering false AT/AF.
Note: R-R interval is regular, therefore this is not AF.
Intracardiac electrogram (EGM) demonstrating sensing of the R-wave on the atrial lead, resulting in false detection of atrial tachycardia (AT)/atrial fibrillation (AF). AR ¼ atrial refractory event.
do about device-detected AF remains unsettled. The
the ASSERT trial are among the more comprehensive
MOST (MOde Selection) trial in patients with sinus
efforts to provide guidance. Although these studies
node dysfunction, the TRENDS study: A Prospective
varied with regard to rate thresholds for detection
Study of the Clinical Significance of Atrial Arrhyth-
of AT (including AT and AHRE), duration of epi-
mias Detected by Implanted Device Diagnostics, and
sodes, and follow-up, an association between these
F I G U R E 3 Implantable Cardiac Monitor Strips Exemplifying True and False AT/AF
False AT/AF
False AT/AF triggered due to PVC
True AT/AF
Note: irregular R-R intervals.
Examples of true and false detection of atrial tachycardia (AT)/atrial fibrillation (AF) by implantable cardiac monitors. False AT/AF detection due to irregular R-R intervals in a patient with frequent premature ventricular contractions (PVC).
285
286
Chen-Scarabelli et al.
JACC VOL. 65, NO. 3, 2015 JANUARY 27, 2015:281–94
Device-Detected Atrial Fibrillation
tachyarrhythmias and risk of ischemic stroke was
dual-chamber
consistently found.
pared to conventional dual-chamber pacing (49). By
minimal
ventricular
pacing
com-
The MOST trial, a 6-year randomized trial of DDDR
measuring AF with stored diagnostic data from
versus VVIR pacing in patients with sinus node
the pacemaker, the SAVE PACe trial showed a mod-
dysfunction, documented a 50% reduction in newly
erate reduction of the risk of persistent AF in patients
diagnosed AF with dual-chamber pacing compared to
with sinus node disease (49). These observations
ventricular pacing alone (hazard ratio: 0.50; 95%
motivated a prospective study (TRENDS) to investi-
confidence interval: 0.32 to 0.76; p ¼ 0.001) (47). A
gate the relationship between device-detected AT
subgroup analysis of 316 patients correlated AHRE
(including atrial flutter, AF, and atrial tachycardia)
with clinical outcomes, but found no significant as-
and stroke risk (50). The TRENDS study was a pro-
sociation of pacing mode (dual-chamber vs. single-
spective, observational study of 2,486 patients with
chamber ventricular pacing) on the presence or
1 or more risk factors for stroke who had implanted
absence of AHRE (48). The presence of AHRE (atrial
devices. Patients with either a low (#5.5 h on any
rate >220 beats/min for 10 consecutive beats) was an
single day within a 30-day period) or a high burden of
independent predictor of mortality (HR: 2.48), death
AT ($5.5 h) had a higher risk of stroke than those
or nonfatal stroke (HR: 2.79), and AF (HR: 5.93),
without AT (HR: 2.20; 95% CI: 0.96 to 5.05; p ¼ 0.06)
indicating that pacemaker patients with sinus node
(33). However, the difference in hazard ratio between
dysfunction and AHRE were more than 2.5 times as
the groups with low and high AHRE burdens was
likely to die or have a stroke, and were 6 times
not statistically significant (51).
as likely to develop AF than those without AHRE
The ASSERT trial evaluated whether detection of
(32). Limitations of this substudy were its retro-
asymptomatic AHRE predicted an increased risk of
spective design, small sample size, and that 81%
stroke and systemic embolism in pacemaker patients
(129 of 160) of patients with AHRE had prior su-
without a history of AF, as well as whether overdrive
praventricular arrhythmias (32). A prospective study
atrial pacing would reduce the risk of symptomatic
Search AV Extension and Managed Ventricular Pacing
AF (52). Although overdrive atrial pacing failed to
for Promoting Atrio-Ventricular Conduction (SAVE
reduce the risk of symptomatic AF, subclinical epi-
PACe) measured the time to persistent AF with
sodes of AT, defined as atrial rates $190 beats/min
F I G U R E 4 Example of Automatic Mode-Switching Due to Repetitive Non–Re-Entrant Ventriculoatrial Synchrony
Blue oval: P-wave falls in the post-ventricular atrial refractory period and is not sensed [functional atrial undersensing], with subsequent atrial pacing [in red rectangle] occurring in the absolute refractory period [functional atrial non-capture].
Electrogram documenting false detection of atrial tachycardia/atrial fibrillation due to retrograde ventriculoatrial conduction, with P-wave falling into the post-ventricular atrial refractory period (functional atrial undersensing), with subsequent atrial pacing during the absolute refractory period (functional noncapture).
Chen-Scarabelli et al.
JACC VOL. 65, NO. 3, 2015 JANUARY 27, 2015:281–94
lasting >6 min, were associated with an increased risk of ischemic stroke (HR: 1.76; 95% CI: 0.99 to 3.11;
T A B L E 1 CHADS 2 and Associated Stroke Risk in the NRAF and ASSERT Trials
p ¼ 0.05). Longer episodes of AT were associated with
NRAF
incremental stroke risk, but AT of 6 to 24 h (HR: 2.00;
Number of CVAs (94)
95% CI: 1.13 to 3.55; p ¼ 0.02) and episodes >24 h (HR: 1.98; 95% CI: 1.11 to 3.51; p ¼ 0.02) carried a similar risk. Stroke risk increased with the number
287
Device-Detected Atrial Fibrillation
n ¼ 1,733
ASSERT Adjusted Stroke Risk
n ¼ 259
Number of CVAs (11)
Adjusted Stroke Risk
CHADS2 ¼ 0
120
2
1.9 (1.2–3.0)
—
—
CHADS2 ¼ 1
463
17
2.8 (2–3.8)
68
1
2.11 (0.23–18.9)
—
CHADS2 ¼ 2
523
23
4.0 (3.1–5.1)
119
4
1.83 (0.62–5.4)
of subclinical AT episodes, with annual rates of TE ranging from 1.20 with a single episode to 1.93
Values are n or hazard ratio (95% confidence interval).
with $4 episodes. Subclinical AT was almost 8 times
ASSERT ¼ ASymptomatic AF and Stroke Evaluation in Pacemaker Patients and the AF Reduction Atrial Pacing; CHADS2 ¼ Congestive heart failure, Hypertension, Age 75 years or older, Diabetes Mellitus, and history of Stroke or transient ischemic attack; CVA ¼ cerebrovascular accident; NRAF ¼ National Registry of Atrial Fibrillation.
more common than clinical AF, which developed in 15.7% of patients with subclinical AT. The 2.5-fold greater risk of TE associated with subclinical AT was independent of the appearance of clinical AF and other stroke risk factors (52).
Resynchronization Devices) trial, which hypothesized that remote monitoring for AT with a predefined
Although the stroke risk on the basis of the CHADS 2
anticoagulation plan would prove superior to con-
(Congestive heart failure, Hypertension, Age $ 75
ventional methods for identification of AT with
years, Diabetes Mellitus, Stroke or transient ischemic
physician-directed anticoagulation (56). Patients with
attack) score appeared lower in the ASSERT trial than
ICDs or CRT defibrillators were randomized 1:1 to
that published by Gage et al. (53) (HR of 1.82 in the
either office visits or remote monitoring for detection
ASSERT trial vs. HR of 4.0 in the NRAF [National Reg-
of AT ($200 beats/min for 36 of 48 beats) and, when
istry of Atrial Fibrillation] trial, for CHADS2 score of 2)
AT was detected, an anticoagulation protocol was
(Table 1), the ASSERT trial was a prospective, ran-
initiated in the intervention group on the basis of
domized trial with lower-risk patients whereas the
the CHADS 2 risk score. Patients with CHADS 2 risk
NRAF was a registry trial with a higher-risk patient
score #2 were to initiate oral anticoagulation if AT
population. Furthermore, the NRAF trial had a greater
lasted $48 h, with discontinuation if there were
population of females, which was not factored into
no AT recurrences detected for 30 days. Those with
risk stratification with the CHADS2 score system.
CHADS2 scores of $3 to 4 would initiate anti-
However, female sex has been recognized as a risk
coagulation for device-detected AT $24 h in 2 days,
factor for cerebrovascular accident and is included in
with discontinuation if there were no AT recurrences
the currently recommended stroke risk stratification
detected for 90 days. Those with CHADS2 scores $5
scheme, CHA2DS2 -VASc (54). In addition to a higher
to 6, or with a history of TE were prescribed anti-
proportion of females, the NRAF trial population
coagulant therapy for any AT, without discontinua-
(in comparison to the ASSERT trial population) had
tion, regardless of AT recurrence. Exclusion criteria
a significantly greater incidence of congestive heart
included a history of stroke, transient ischemic
failure, prior cerebrovascular accident or transient
attack, or systemic embolism, and documented AF
ischemic attack, and lower utilization of aspirin ther-
or atrial flutter. A total of 2,718 patients were
apy (all p values