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

Atrial Fibrillation as a Risk Factor for Stroke

An abundance of studies have shown that patients with AF have elevated stroke risk. In a meta-analysis of five randomized controlled trials performed by the AF Investigators, the annual stroke rate was 4.5% in non-anticoagulated patients: however, the risk was not uniform and certain risk factors carried a higher-relative risk compared with others.[5] Furthermore, all of the risk factors were increased with age delineating why stroke risk is so age-dependent. Interestingly, patients without any risk factors have a much lower stroke risk, estimated at <1% per year.[5,11] As such, although AF can lead to left atrial appendage (LAA) thrombus and consequent stroke, the majority of the increased risk is likely secondary to associated comorbidities and not purely a manifestation of the arrhythmia itself but 'the company it keeps'. This was illustrated in case–control study of 110 patients with AF found to have a LAA thrombus on transoesophageal echocardiography. While patients with LAA thrombus had a higher CHADS2 score compared with controls (2.8 vs. 1.6), there was significant overlap suggesting that other important factors were present.[12] However, identification of associated risk factors form the basis for risk stratification scores aimed at determining which patients have sufficient stroke risk to have more benefit than risk from anticoagulation.

There are two main postulates regarding AF and consequent stroke. The first revolves around diminished LAA flow velocities during AF with secondary thrombus formation and thromboembolism. In this scenario, AF precedes stroke. As such, it follows that elimination of AF or exclusion of the LAA may prevent stroke. However, the fact that the actual stroke event does not always occur during (or immediately after) an AF episode in those with paroxysmal AF suggests that other mechanisms are at play.[13] The second postulate is that AF is primarily a marker of vascular disease burden which leads to increased stroke risk. The associated atherosclerotic risk factors result in vascular inflammation with associated diastolic and endothelial dysfunction. This results in an atrial myopathy with left atrial (LA) enlargement and secondary fibrosis resulting in electrical heterogeneity from which AF may arise. Coincident with this, endothelial dysfunction from the underlying vascular inflammation may lead to a hypercoagulable state and increased thrombus formation (Take-home figure). This is indicated by elevated levels of coagulation factors such as d-dimer, fibrinogen, von Willebrand factor, and platelet factor-4 in people with AF. For example, in a study of 591 patients, d-dimer levels were highest in those with AF, lowest in controls and intermediate in those treated with warfarin.[14] Furthermore, modulation of inflammatory markers has been associated with a reduction in cardiovascular events.[15] Associated vascular disease of the aorta may also contribute to stroke risk related to atheroemboli from aortic plaque.

Patients with elevated CHADS2 and CHA2DS2-VASc score seem to have an elevated stroke risk even in the absence of AF. In an analysis of 916 non-anticoagulated patients with stable coronary heart disease and without AF, patients with a score of ≥5 had comparable stroke rates to AF patients with moderate CHADS2 scores.[16] While this may be secondary to undetected ('silent') AF, it underscores that AF may be a manifestation of an overriding vascular disease as outlined above. Further evidence stems from the CHADS2 and CHA2DS2-VASc scores being predictive of myocardial infarction.[17] In a multicentre observational study of 3183 patients with acute coronary syndrome, CHA2DS2-VASc ≥2 was associated with a higher risk of adverse events compared with CHA2DS2-VASc <2 [hazard ratio (HR) 2.9].[18] In another study of 565 patients post-AF ablation, those with CHA2DS2-VASc 0–1 had a 1.1% event rate compared with 7.1% in those with a score ≥2.[19]

Take Home Figure.

Schematic representation of the postulated association between atrial fibrillation and stroke. On the left, atrial fibrillation is seen as the cause of stroke. On the right, atrial fibrillation is represented as an epiphenomenon, indicative of an underlying vascular process.