“Left Atrial Appendage Function and Stroke Risk”

Background

About 30-40% of ischemic stroke is of unknown cause.^{1, 2} Recently, biomarkers of atrial dysfunction, or “atrial cardiopathy”, have been associated with embolic stroke risk even in the absence of atrial fibrillation (AF), suggesting that the presence of AF is not required for left atrial thromboembolism to occur.³ Most left atrial thrombi occur in the left atrial appendage (LAA), but there is limited use of LAA dysfunction parameters such as LAA flow velocity and morphology to predict ischemic stroke risk. Here we review the literature on the association between LAA pathology and dysfunction and ischemic stroke, with a focus on patients with unexplained, or cryptogenic, stroke.

Cryptogenic Stroke

Each year, at least 200,000 people in the United States experience a cryptogenic stroke.^{1, 2} The classification of cryptogenic stroke, a subgroup of stroke of undetermined cause, was conventionally made when the initial stroke evaluation was unrevealing, regardless of whether the evaluation was exhaustive, routine, or incomplete.⁴ More recently, investigators have used the term “embolic stroke of undetermined source (ESUS)” in reference to a non-lacunar infarction occurring in the absence of a specific identifiable high-risk stroke mechanism, such as AF, valvular heart disease, or large artery stenosis.⁵ The idea that such infarctions are due to a distant source of embolus is supported by findings from the NINDS Stroke Databank indicating that undocumented cardiac or aortic sources of embolism were found in about two-thirds of patients with cryptogenic stroke on follow-up testing.² One of the most common mechanisms often found in a delayed fashion after cryptogenic stroke is paroxysmal AF, which is detected in up to 30% of patients with cryptogenic stroke on outpatient heart-rhythm monitoring.^{6, 7} However, more than half of cryptogenic strokes remain a mystery even after long-term heart-rhythm monitoring. To optimize secondary stroke prevention strategies, it is essential to identify other embolic causes, especially those that may require anticoagulant therapy

Relationship between the Left Atrium and Stroke

AF, with its implied intra-cavitary stasis in the setting of irregular atrial wall contraction, has been long considered a direct mechanistic explanation for left atrial thromboembolism. Such a mechanism is consistent with the finding from numerous randomized clinical trials that anticoagulation reduces the risk of ischemic stroke in AF.⁸ There is a challenge, however, in detecting episodes of paroxysmal AF, especially when they are subclinical. The Asymptomatic AF and Stroke Evaluation in Pacemaker Patients and the AF Reduction Atrial Pacing Trial (ASSERT) enrolled subjects 65 years of age or older with hypertension and no history of AF. This study showed that while episodes of subclinical AF were detected in about 35% at 2.5 years, only 16% of patients had clinically apparent AF. Subclinical AF was associated with increased stroke risk (hazard ratio, 2.50; 95% CI, 1.28 to 4.89; P=0.008).⁹ Importantly, there was lack of a temporal relationship between AF and stroke, with many patients showing AF for the first time after their stroke,¹⁰ challenging the concept that AF itself is the necessary and sufficient cause of stroke in patients with this dysrhythmia.

Recently, biomarkers of atrial dysfunction or “atrial cardiopathy” have been associated with stroke independently of AF. N-terminal pro-brain natriuretic peptide (NT-proBNP), has been shown to be a marker of atrial dysfunction, risk of incident AF,¹¹ and cardiac embolism^{12, 13}. ECG signs related to the left atrium, such as paroxysmal supraventricular tachycardia and P-wave terminal force in lead V₁ (PTFV₁) have been associated with the risk of incident stroke^{14, 15}, particularly of non-lacunar subtypes¹⁶ even in the absence of AF. Furthermore, increased left atrial size on echocardiogram has been shown to be associated with incident ischemic stroke risk¹⁷ and recurrent ischemic stroke subtypes related to embolism,¹⁸ again independent of AF. Together, these findings suggest that markers of left atrial pathology and dysfunction may be valuable in understanding stroke pathogenesis.

The Left Atrial Appendage as a Site of Thrombus Formation in Atrial Fibrillation

The left atrial appendage (LAA) is the remnant of the embryonic left atrium, while the smooth-walled left atrium originates from the primordial pulmonary vein and its branches. The LAA is largely considered to be nonfunctional. However, the LAA is an actively contracting structure, as demonstrated by flow velocity waveforms seen on pulse Doppler echocardiography and by its prominent muscular trabeculations.¹⁹ In addition, the LAA may play a role in modulating left atrial pressure. Studies have demonstrated that the LAA is more distensible than the left atrium and a relatively high concentration of atrial natriuretic factor (ANF) is contained within the LAA.^{20, 21} Therefore, the LAA may maintain left atrial pressure through the activation of stretch sensitive receptors and the actions of ANF, which increases heart rate, diuresis, and natriuresis.¹⁹

Regardless of its original function, the LAA has long been considered to be a prime site for thrombus formation in patients with AF. In a transesophageal echocardiography (TEE) study that included 317 patients with AF and a recent embolic event, about 20% had evidence of thrombus, all of which were in the LAA.²² Another pooled analysis of 23 studies that included 4792 patients with AF who underwent transesophageal echocardiogram, cardiac surgery, or autopsy, left atrial thrombi were detected in about 14% of patients, and the majority of these occurred in the LAA.²³ In almost all reports included in this review, the LAA was the site of thrombi in most patients.²⁴ Therefore, a better understanding of the anatomy and physiology of the LAA using various imaging modalities may be of value in advancing our understanding of stroke pathogenesis.²⁵

Left Atrial Appendage Morphology

The LAA varies significantly in shape, size, and orientation to adjacent cardiac structures.²⁶ A study of 500 postmortem hearts demonstrated that the LAA most commonly has 2 lobes (54%), followed by 3 lobes in 23%, 1 lobe in 20%, and 4 lobes in 3%.²⁷ The number of lobes has been shown to be an independent risk factor for the presence of thrombus.²⁸

Traditionally, TEE has been used to characterize LAA morphology.²⁹ More recently, cardiac computed tomography (CCT) and cardiac magnetic resonance (CMR) have been used to further describe LAA morphology. A retrospective study of 932 patients with AF who underwent CCT or CMR categorized the LAA into four distinct morphologies: chicken wing (48%), cactus (30%), windsock (19%), and cauliflower (3%).³⁰

Parameters of Left Atrial Appendage Function

TEE with Doppler echocardiography allows for functional assessment of the LAA. Using pulse Doppler 2D echocardiography with TEE, blood flow velocities of the LAA can be quantified. During sinus rhythm, four waves are observed. The initial wave occurs in late diastole and represents forward flow from LAA contraction and emptying. The second wave is a retrograde wave from LAA filling. Then there are additional and smaller forward and retrograde waves, which are thought to be due to ventricular relaxation and LAA elasticity.³¹

In a study using tissue Doppler imaging with TEE and transthoracic echocardiography (TTE) to characterize LAA myocardial function by measuring LAA contraction velocity that enrolled 141 patients (48 AF and 93 sinus rhythm), LAA contraction velocity was obtainable using TTE in 84% of patients and there was a strong correlation between LAA contraction velocity on TTE and TEE (r > 0.7, p<0.0001).³² Another study found that using TTE in 84 consecutive patients, LAA filling velocities were recorded in 78.1% of patients and LAA emptying velocities in 62.5% of patients that were shown to correlate well with those obtained by TEE.³³

TEE, however, is considered to be the gold standard for assessment of LAA thrombus formation.³⁴ This modality provides excellent visualization of posterior structures such as the left atrium and LAA, with high spatial resolution and 180 degree visualization of the LAA. Studies have demonstrated a sensitivity of 92% and a specificity of 98% with a negative and positive predictive value of 100% and 86%, respectively, for the detection of LAA thrombi.^{35, 36} However, TEE is a semi-invasive procedure that is not well tolerated by all patients and may not be as readily available as alternative modalities such as TTE. Visualization and evaluation of the LAA on TTE has improved with tissue harmonic imaging.³⁷ In the Comprehensive LAA Optimization of Thrombus (CLOTS) trial³⁸, a prospective multicenter study, 118 patients with AF were evaluated with TEE and TTE harmonic imaging with intravenous contrast. When compared to TEE, the latter modality demonstrated 100% diagnostic accuracy in the detection of LAA thrombi. Since TEE is not established as the standard of care in most patients with ischemic stroke, TTE, especially with the advent of harmonic imaging and use of intravenous contrast, pulsed Doppler, or tissue Doppler, may be a feasible, non-invasive, less expensive, and readily available imaging alternative to obtain LAA functional parameters such as LAA contraction velocity and flow velocity.

Left Atrial Appendage Function and Morphology in Relation to Thromboembolism

Currently, identification of patients’ risk of left atrial thromboembolism is based nearly exclusively on the surface electrocardiogram (ECG). However, a study that included 41 patients with AF after elective DC cardioversion showed that normal sinus rhythm on the surface ECG can co-exist with disorganized (i.e., non-sinus) LAA contraction.³⁹ In addition, a recent study included 201 patients with recent stroke and a history of AF who underwent TEE documenting that among patients who were in sinus rhythm (n = 24) at the time of TEE, 25% of them had a flow pattern in the LAA that is seen in active AF. This suggests an electromechanical dissociation between the surface ECG and echocardiographic LAA function during episodes of sinus rhythm.⁴⁰ The LAA could also be the origin of the arrhythmogenic foci in AF. In a study that enrolled 987 patients with atrial fibrillation undergoing repeat catheter ablation for atrial fibrillation, arrhythmias originating from the LAA, as opposed to the more usual pulmonary vein location, were noted in 27% of patients. When compared to non-successful ablation, successful ablation of LAA trigger resulted in a significant decrease in AF recurrence (15% vs. 74%, P<0.001).⁴¹ These findings support the value of direct echocardiographic or other measurements of LAA function as markers and contributors of the “arrythmogenic” potential of the left atrium.

Echocardiographic measurements of LAA function have been directly correlated with thromboembolic risk. Lower LAA velocities have been shown to correlate with ischemic stroke and thrombus formation (Table).^{32, 42-46}

Table.

Left Atrial Appendage (LAA) and Ischemic Stroke

LAA Parameter	Study	Findings
LAA Function	141 patients (48 AF and 93 sinus rhythm) who underwent TEE32	Mean LAA contraction velocity was lower in patients with LAA thrombus versus those without (10 +/− 4 vs. 22 ± 7 cm/s, p < 0.001), and similarly lower in those with AF and a history of stroke or transient ischemic attack than in those without (11 +/− 3 vs. 15 +/− 6 cm/s, P = 0.008). About one third of patients with a LAA flow velocity ≤ 11 cm/s had evidence of LAA thrombus.
	Cross-sectional study of 218 patients with AF44	Patients with stroke had lower LAA flow velocity (36 ± 19 vs 55 ± 20 cm/s, p <0.001).
	Post-hoc analysis from the Stroke Prevention in AF III (SPAF-III) trial that included 721 patients who underwent TEE45	Peak LAA anterograde flow velocity < 20 cm/s was independently associated with thrombus formation and risk of cardioembolism.
	Cross-sectional study of 360 patients with AF46	Lower left atrial appendage flow velocity was associated with ischemic stroke.
LAA Morphology	Cross sectional study of 932 patients with Atrial Fibrillation (AF)30	The prevalence of ischemic stroke among different LAA morphologies was: 4% in chicken wing, 10% in windsock, 12% in cactus, and 18% in cauliflower. Higher adjusted odds of stroke or transient ischemic attack with other morphologies as compared with chicken wing morphology: cactus (odds ratio [OR] 4.08, p = 0.046), windsock (OR 4.5, p = 0.038), and cauliflower (adjusted OR 8.0, p = 0.056)
	Cross-sectional study of 360 patients with AF46	Inverse association between chicken-wing morphology and stroke risk (adjusted OR 0.34, 95% confidence interval (CI) 0.14-0.84, P = 0.020).
	80 patients with AF who underwent cardiac computerized tomography.47	Cauliflower LAA morphology was an independent predictor of stroke (OR 3.4, p = .017)
	348 patients with AF undergoing ablation48	Association between non-chicken wing morphology and covert cerebral infarcts on brain MRI.
	1063 patients with AF49	Extensive LAA trabeculations is associated with ischemic stroke
LAA Orifice Area	Cross-sectional study of 218 patients with AF44	Patients with stroke were found to have a larger LAA orifice area (4.5 ± 1.5 vs 3.0 ± 1.1 cm2, p <0.001).
	Cross- sectional study of 360 patients with AF46	Larger LAA orifice area was also associated with ischemic stroke.
LAA Fibrosis	178 patients with AF50	LAA fibrosis was associated with reduced LAA flow velocities.

Besides LAA function, its morphology has also been associated with increased risk of stroke in patients with AF (Table). Specifically there are 3 known morphologic features of the LAA associated with ischemic stroke: shape, orifice size, and fibrosis. Studies have shown an inverse association between chicken-wing morphology with ischemic stroke^{30, 46} and covert brain infarcts (Table).⁴⁸ Furthermore, cauliflower LAA morphology was an independent predictor of stroke which is possibly related to extensive LAA trabeculations.⁴⁹ LAA flow velocity was highest among patients with chicken-wing as opposed to non-chicken wing morphology^{29, 51, 52} which may explain why chicken wing morphology has the lowest risk of ischemic stroke. In addition, larger LAA orifice area has also been shown to be associated with ischemic stroke (Table).^{44, 46}

Late gadolinium enhancement CMR may also be used to detect LAA structural dysfunction. Recent evidence showed that LAA fibrosis on CMR is associated with reduced LAA flow velocities⁵³, indicating that fibrotic changes of the LAA appendage are linked to stasis, thrombus formation and stroke risk.

The data reviewed in this section pertain primarily to patients with non-valvular atrial fibrillation rather than patients with valvular heart disease, in whom thrombus formation is more widespread in the left atrium than limited to the left atrial appendage. These findings, however, suggest that parameters of LAA function and morphology may improve stroke risk prediction tools in patients with AF, as well. For example, in patients with a CHADS score of 1 (in whom oral anticoagulation would generally be withheld), the prevalence of stroke history was about 5-fold greater in patients with non chicken-wing morphologies compared with chicken-wing morphology.⁵⁴ Therefore, the LAA is not only the site of thrombus, but also a marker and even contributor of left atrial electrical and mechanical function (Figure).

Figure.

Left Atrial Appendage Dysfunction as a Cause of Embolic Stroke

Effect of Left Atrial Appendage Closure on Thromboembolism

The evidence above supports the hypothesis that the LAA often serves as a site of thrombus formation. On this basis, clinical trials have evaluated the benefit of LAA closure in reducing stroke in patients with AF. The Watchman LAA System for Embolic Protection in Patients With AF (PROTECT AF) trial randomized 707 patients with non-valvular AF to either the Watchman device (n=463) or warfarin (n=244). The main outcome was stroke (ischemic or hemorrhagic), systemic embolism, or cardiovascular death. After a mean follow up of 2.3 years, the primary endpoint was reached in 3.0% of patients randomized to the Watchman device and 4.3% in the warfarin groups, respectively (RR 0.71; 95% CI, 0.44%-1.30%).⁵⁵ There was an increased rate of adverse events, mainly major pericardial effusions (~ 5%), in patients receiving the Watchman devise closure. This study was followed by the Prospective Randomized Evaluation of the Watchman LAA Closure device in patients with AF versus long-term warfarin therapy (PREVAIL) trial that randomized 407 patients to Watchman versus warfarin in a 2:1 manner. The primary endpoint again was stroke, systemic embolism, or cardiovascular death. After an 18-month follow-up, the primary endpoint was reached in 6.4% and 6.3% in the Watchman vs. warfarin group, respectively. Although these percentages were similar, the study did not achieve the pre-specified criterion for non-inferiority. However, the non-inferiority endpoint was reached when comparing the risk of stroke or systemic embolism between patients randomized to Watchman vs. warfarin (2.53% versus 2.00%). In addition, patients enrolled in the PROTECT-AF trial who underwent LAA closure had reduced mortality after 4 years of follow-up compared with warfarin therapy.⁵⁶ Furthermore, the complications observed in this procedure were lower than those seen in the PROTECT AF trial. Many questions remain about LAA closure, including whether it is as effective if compared with non-vitamin K antagonist oral anticoagulant (NOAC) drugs that have a lower risk of intracranial hemorrhage. However, the Watchman trials are nevertheless consistent with a protective effect against thromboembolism, which again supports the role of the LAA as the origin of thrombus formation in left atrial disease.

Left Atrial Appendage Dysfunction in Cryptogenic Stroke

Very little data exists on the relationship between LAA dysfunction and cryptogenic stroke specifically, and patients with cryptogenic stroke do not routinely undergo detailed assessment of the LAA. However, the considerations above support further investigation of the LAA as a cause of many currently cryptogenic strokes. Measurements of LAA function can be performed by TTE, which is widely available, noninvasive, and used in the evaluation of most patients with cryptogenic stroke. Including these measurements in the diagnostic evaluation of patients with cryptogenic stroke may help understand the recurrent stroke risk and potentially improve stroke prevention strategies.

The current AHA/ASA stroke prevention guidelines recommend antiplatelet therapy for secondary stroke prevention unless there is a mechanical heart valve or documented AF,⁵⁷ which means that cryptogenic stroke patients currently receive antiplatelet therapy as a routine. Patients with LAA dysfunction and cryptogenic stroke, however, may constitute a subgroup that may benefit from anticoagulation. In fact, while the Warfarin Aspirin Recurrent Stroke Study (WARSS) trial showed no difference between warfarin and aspirin in secondary stroke prevention in patients with stroke in the absence of an established cardioembolic cause, including known AF,⁵⁸ a post-hoc analysis showed that warfarin was substantially superior to aspirin in patients with elevated NT-proBNP, a biomarker that may partly reflect LAA dysfunction. In these patients, the 2-year rate of recurrent stroke or death was 16.6% in the warfarin group versus 45.9% in the aspirin group (hazard ratio 0.30, 95% CI 0.12-0.84; P=0.021).⁵⁹ Such findings support further research on the effectiveness of anticoagulant therapy in secondary stroke prevention for patients with cryptogenic stroke and evidence of LAA pathology and dysfunction.

Conclusion

The left atrium is increasingly implicated in the pathogenesis of cryptogenic stroke, and the LAA serves as the most common site of thrombus formation in the left atrium. Given the substantial benefit of anticoagulant therapy in patients with atrial disease in the form of AF, such therapy may ultimately prove beneficial for patients without AF but compelling evidence of LAA dysfunction. Future research is needed to identify optimal methods to assess patients for LAA dysfunction and to test the benefit of anticoagulant therapy in stroke prevention in these patients.