Risk Stratification for Stroke in Atrial Fibrillation: A Critique

Ammar M. Killu; Christopher B. Granger; Bernard J. Gersh


Eur Heart J. 2019;40(16):1294-1302. 

In This Article

Modifying Risk Factors for Stroke in Atrial Fibrillation

Despite their limitations, the available risk scores are clinically useful and easy to apply in determining an individual's basic risk of stroke. However, modifying factors should be considered. Further study into such factors and their incorporation into commonly used risk scores with prospective validation is required.

Left Atrial Function and Volume

Left atrial emptying fraction (LAEF) is a measure of LA mechanical function akin to ejection fraction for ventricular function assessment. Studies have demonstrated that LAEF is reduced in patients with AF. However, there is also suggestion that LAEF can predispose to AF and itself may be associated with an increased risk of thromboembolism.[62,63] Also, LA enlargement has been associated with an increased risk of thromboembolic stroke. However, given that atrial enlargement is commonly found in patients with AF it is hard to decipher the extent to which LA enlargement is an independent contributor. Furthermore, hypertension (the major risk factor for stroke) and diastolic dysfunction both lead to LA enlargement and are important confounders.[64,65] Thus, control of hypertension should be part of the overall strategy to reduce risk of stroke. This is emphasized in a recent study demonstrating that patients with AF and a blood pressure (BP) of 120–129/<80 mmHg had a lower risk of major cardiovascular events compared with those with BP <120/80 mmHg and ≥130/80 mmHg such that this should be the target BP goal.[66]

Left Atrial Fibrosis

Left atrial fibrosis as assessed by late gadolinium enhancement on cardiac magnetic resonance imaging has been postulated to be a contributing factor to the development and progression of AF with increasing severity noted in those with persistent AF. A recent study observed a J-shaped association of LA fibrosis with stroke or transient ischaemic attack in patients with AF suggesting that it is the atrial myopathy itself (rather than the rhythm) that contributes to the increased risk of adverse events.[67] Interestingly, patients with more atrial fibrosis in this study had significantly higher CHA2DS2-VASc scores than those with minimal fibrosis.

Left Atrial Appendage Morphology

Left atrial morphology has been sub-classified into four types based on the suggestion that morphology independently influences thromboembolic risk in patients with AF (Figure 6). A systematic review study by Lupercio et al.[68] including patients with AF found that those with the 'chicken wing' morphology were least likely to develop thromboembolic stroke. This may be related to a smaller appendicular opening and increased LAA emptying velocity, where a reduction has been associated with increased sludge formation and also stroke.

Figure 6.

Various morphologies of the left atrial appendage have been described, with the stroke risk variable according to morphology. The chicken wing morphology has been associated with the lowest risk of stroke, and the cauliflower the highest. Dr Joseph Maleszewski, Mayo Clinic, Rochester, MN, USA.

Additional Comorbidities

The CHA2DS2-VASc score has not been validated in people with AF and mitral stenosis (at very high risk of stroke), hyperthyroidism, hypertrophic cardiomyopathy (HCM), or active cancer.[69] Though historically there was concern for heightened risk of stroke in patients with hyperthyroidism and AF, recent data suggest that embolic risk is not necessarily increased independent of other risk factors.[70] As such, it may be reasonable to approach anticoagulation in such individuals based on their CHA2DS2-VASc score.[30] Approximately one in five individuals with HCM will develop AF—this in turn is associated with increased mortality.[71] Moreover, individuals with cancer have an increased risk of myocardial infarction and stroke, which may be double that of controls. In the presence of AF, stroke risk may be even greater. Therefore, individuals with HCM or cancer (or possibly hyperthyroidism) and AF should be deemed at high risk for thromboembolism irrespective of their CHA2DS2-VASc score.[69,71] It has long been known that coronary artery disease and ischaemic stroke share common risk factors. However, a recent study demonstrated an increased risk of ischaemic stroke/thromboembolism in AF patients with ≥50% coronary stenosis in at least one coronary artery such that this may be considered as an additional risk factor.[72]Chronic kidney disease is associated with elevated risk of AF and increased failure of sinus rhythm restoration measures.[73] Smoking is associated with increased incidence of AF, with >2-times increased risk of AF.[74] Untreated obstructive sleep apnoea has also been associated with an increased risk of AF and failure of therapy.[75] Whether such comorbidities independently influence the risk of thromboembolism in AF remains to be determined.

Atrial Ectopy and Clinically Silent Atrial Fibrillation

Given that AF is often asymptomatic, episodic, and of short duration, it can be difficult to detect in the absence of prolonged ambulatory monitoring. For example, in the ASSERT-II study of 256 patients attending cardiovascular or neurology clinics, sub-clinical AF ≥5 min had an incidence of 31% at 12 months.[76] At present, duration is not accounted for in risk scores and is generally viewed as a dichotomous variable—present or absent. The duration whereby AF becomes meaningfully significant is debatable: some suggest that a sustained period of ≥24 h, while others report a duration as short as 6 min.[77,78] However, it seems that longer episodes are associated with increased risk, likely representative of a more severe atrial myopathy. In a recent study, it was shown that a greater burden of AF is associated with a higher risk of ischaemic stroke independent of known risk factors in adults with paroxysmal AF.[79] It is now clear that permanent AF has modestly higher risk of stroke than paroxysmal AF.[80,81] One study including patients with pacemakers and AF demonstrated that stroke risk was dependent on both presence/duration of AF and CHADS2 score.[82] In a follow-up study of the Copenhagen Holter Study cohort, patients with frequent ectopic atrial premature beats (defined as ≥30/min or any run of ≥20 beats) and no history of AF were found to have a two-fold increase in stroke rate, even following adjustment for AF development. Additionally, the ASSERT (Asymptomatic AF and Stroke Evaluation in Pacemaker Patients and the AF Reduction Atrial Pacing Trial) study demonstrated that subclinical atrial tachyarrhythmias were associated with increased risk of ischaemic stroke or systemic embolism even without clinical AF.[77] The leading hypotheses as to why excessive premature atrial beats or short runs of atrial arrhythmias lead to stroke are that (i) it may be a surrogate of more severe comorbidities,[82,83] or (ii) be a precursor for AF and is associated with AF burden (though stroke was often seen prior to occurrence of AF). Indeed, a recent meta-analysis showed that subclinical AF strongly predicted the development of clinical AF. Furthermore, the risk of stroke was still elevated, though lower than that for patients with clinical AF.[84] As evidenced by the above, many uncertainties remain. While silent AF appears to confer increased risk of adverse events, whether or not the risk is negated by population screening measures and ad-hoc anticoagulation is undetermined. Even if anticoagulation in this population is beneficial, the threshold by which to commence treatment is unclear. Noteworthy, intermittent OAC based on occurrence of atrial arrhythmias in patients with implantable monitoring device was not shown to be beneficial.[85]