Is subclinical AF a risk factor for stroke?

Earlier this year, the Asymptomatic Atrial Fibrillation and Stroke Evaluation in Pacemaker Patients and the Atrial Fibrillation Reduction Atrial Pacing Trial (ASSERT) investigators reported data on 2580 hypertensive older pacemaker patients.9 These patients were aged 65 years or older and had recently been implanted with a pacemaker or defibrillator. Patients with a history of AF or atrial flutter and patients on oral anticoagulation with a vitamin K antagonist were excluded. Within the first 3 months of the trial, the authors detected subclinical AF (defined as any episode of atrial tachycardia (>190 beats per minute) for more than 6 min) in around 10% of the patients. Over an average follow-up period of 2.5 years, the risk of developing an ischaemic stroke or systemic embolism was 2.5 fold higher for patients with subclinical AF, compared with patients without. Similar results have been obtained in two previous, smaller trials: Glotzer et al reported that pacemaker patients who had at least one episode of atrial tachycardia with more than 220 bpm for at least 5 min had a 2.79 fold increase in the risk of stroke or death.7 In the TRENDS study, it was shown that patients with a daily burden of atrial tachycardia of more than 5.5 h have a 2.4 fold increase in the risk of thromboembolism, compared with patients with no atrial tachycardia. In contrast to ASSERT, patients with a history of AF were included in these trials. In all these trials, subclinical AF also increased the risk of clinical AF 5–6 fold, which suggests that subclinical AF could be regarded as a precursor to clinical AF. Whether the number of subclinical AF episodes, the ‘burden’ of AF (percentage of time spent in AF divided by total time) or the duration of the longest AF episode may be the best predictor for subsequent stroke, is currently still under debate.

Do patients with subclinical AF need anticoagulation?

The interesting question of whether patients with subclinical AF have to be anticoagulated currently remains unanswered. Patients with subclinical AF have rarely been included in previous anticoagulation trials. For instance, in the RE-LY trial, patients with paroxysmal AF were only included if they had two documented episodes of AF, but ECG from pacemakers of ICDs could only be used to document 1 episode.10 On the contrary, within other major anticoagulation trials (ROCKET-AF, ARISTOTLE), an inclusion of patients with subclinical AF was possible.11 ,12 However, no prospective randomised trial using anticoagulation in patients with subclinical AF has been investigated thus far. Data from the recently published WARCEF trial (Warfarin vs Aspirin in Reduced Cardiac Ejection Fraction), a study in patients with reduced ejection fraction, showed that using warfarin as compared with aspirin halved the number of ischaemic strokes.13 As patients with reduced systolic function are more prone to develop AF,14 an explanation for this intriguing finding could be that anticoagulation may have reduced thromboembolism in patients with undiagnosed subclinical AF. However, at the moment this is an unproven hypothesis and the WARCEF investigators are currently working on assessing this hypothesis (S Homma, JLP Thompson, personal communication). Whether patients with subclinical AF benefit from oral anticoagulation similarly to patients with clinical AF, or benefit at all, still remains to be determined. Future research has to assess whether the early detection of subclinical AF, followed by anticoagulation therapy, is a successful stroke prevention strategy.

Which patients should be screened for subclinical AF?

First, it has to be stressed that most data on subclinical AF have been obtained in pacemaker patients.7–9 Patients with sinus node disease are known to have an increased risk of AF,15 ,16 hence the prevalence of subclinical AF in a more general population may be smaller than in these pacemaker trials. There are only limited data available thus far, suggesting that there is a 2.5 fold relative increase of risk for subsequent stroke in patients with subclinical AF in all categories of the CHADS₂ score, an established clinical score for stroke risk in AF patients. This would result in the highest possible risk in patients with a high CHADS₂ score, for example, in patients after stroke/transient ischemic attack (TIA). These patients in particular, therefore, would potentially gain most benefit from the detection of subclinical AF.

Detection of subclinical AF after stroke

Patients with previous stroke/TIA usually have a CHADS₂ score close to 4 and the detection of AF after stroke therefore usually triggers the initiation of anticoagulation therapy. These patients are considered high-risk patients for stroke recurrence. There have been numerous studies on the detection of AF in stroke patients,17 and these have found AF in roughly 3% of stroke patients if 24 h Holter monitoring was performed.18–21 However, if the monitoring time was expanded to 5–7 days, AF could be diagnosed in up to >10% of stroke patients.22–24 Even longer continuous monitoring with the use of an implantable loop recorder (Reveal XT, Medtronic, Minneapolis, Minnesota, USA), currently investigated in the CRYSTAL-AF study (CRYptogenic STroke And underLying Atrial Fibrillation),25 may generate additional data for longer monitoring times and will, for the first time, provide data from a randomised controlled trial for intensified monitoring to detect AF. A growing area of research is the identification of patients likely to have AF, to increase pre-test probability and to reduce the number of patients needing longer monitoring times. For example, increasing numbers of chronic brain lesions on CT or MRI and acute cortical infarcts have been shown to be radiological indicators that could predict paroxysmal AF;26 key clinical information summarised in a score called Score for Targeting Atrial Fibrillation (STAF)27 may also help to identify those patients. However, results on the validation of the latter score are conflicting: where the group who developed the score were able to validate their results in an independent cohort,28 another group reported on the limited value of the score.29 A possible explanation for these conflicting findings might be that the derivation of the score was based on patients with persistent and paroxysmal AF, whereas the validation is mostly being used in patients with only paroxysmal AF.30

An alternative to the STAF score might be comprehensive echocardiographic parameters or biomarkers. Our group recently showed that a parameter which has been derived in patients with paroxysmal AF31 and combines the information of left atrial volume and left atrial function might be useful to select patients in whom prolonged cardiac monitoring might be useful. With the echocardiographic marker it could be possible to effectively reduce the number of stroke patients needing prolonged Holter monitoring by 48%, omitting only 8% of patients (who only had very brief episodes of AF of less than 1 min).32 Biomarkers (eg, natriuretic peptides) might be an attractive tool because they are more readily available than echocardiography and easier to investigate. For example, the use of prolonged Holter monitoring in patients with only a brain natriuretic peptide level >99 pg/ml could reduce the number of stroke patients needing prolonged monitoring by 42% omitting only 8% of paroxysmal AF cases.33 A third method for choosing patients who might develop paroxysmal AF during prolonged monitoring could be the number of atrial premature beats, which can easily be derived from 24 h-Holter ECG. It has been shown that more than 70 premature atrial beats in 24 h were associated with a 6.6 fold increase in the risk of paroxysmal AF34 during long-term follow-up. Larger studies are needed to determine which pre-selection strategy is the most feasible in daily practice and yields best results.