Stroke is a major cause of mortality and disability in adults, especially cardioembolic stroke, but the etiology can remain undetermined even after comprehensive study.1–3 Other causes of stroke4 include large-artery atherosclerosis, small-vessel occlusion, or other determined etiology. When echocardiography shows significant left ventricle dysfunction masses, cardioembolic stroke is relatively easy to diagnose. Nevertheless, most cardioembolic strokes are secondary to atrial fibrillation (AF),5 which is a well-established independent risk factor for this etiology.6

Approximately 10% of patients with acute ischemic stroke or transient ischemic attack will have newly diagnosed AF during their hospital admission for stroke7; an additional 12% of patients may be found to have AF when screened by continuous long-term electrocardiographic monitoring,8,9 which has been recently introduced in the recommendations for the management of patients with an acute ischemic stroke or transient ischemic attack without other apparent cause.10 Given the major burden of AF, it is possible that some cardioembolic strokes could be misclassified as being of undetermined etiology; if a cardioembolic source could be confirmed, this group of patients would benefit from anticoagulation therapy. Previous studies have suggested that unknown AF could be present in 20% to 28% of patients with stroke of undetermined etiology.11–15

It is well known that left atrial (LA) size predicts AF and stroke16,17 and is also related to stroke in patients without previous AF.18,19 Left atrial function, assessed by volumetric changes (left atrial ejection fraction [LAEF]) or by deformation imaging (LA strain [LAS] and LAS rate [LASR])20–23 has also been linked to paroxysmal AF.15,24–26 In patients with stroke of undetermined etiology, LAS (s-wave) has been related to paroxysmal AF.27 Additionally, in patients with persistent AF, LAS was a predictor of stroke28 even in patients with a low risk score for stroke.29 More recently, diagnosis of paroxysmal AF with continuous Holter monitoring has been related to LA function changes in patients with stroke of undetermined etiology15; patients with AF showed a more reduced LA function than those without AF.

The CHA2DS2-VASc (congestive heart failure, hypertension, age ≥ 75 [doubled], diabetes, previous stroke [doubled], vascular disease, age 65–74, and sex [female]) score30 includes clinical parameters to establish stroke risk in patients with AF. It is also a strong predictor of ischemic stroke in patients with coronary artery disease without AF.31,32

Considering the potential burden of AF in stroke patients and the large number of patients with stroke of undetermined etiology, we hypothesized that some strokes classified as being of undetermined etiology might have a cardioembolic source, with silent, undiagnosed, underlying AF; in this sense, the evaluation of surrogates of atrial disease such as LA size and function could depict a subgroup of patients at risk of having AF and a potential reclassification to cardioembolic stroke. Therefore, the aim of our study was to characterize LA function in patients admitted to hospital with ischemic stroke in order to identify LA function patterns that could identify potentially misclassified stroke of undetermined etiology that actually has cardioembolic source. Long-term monitoring could be useful in these patients, and the introduction of anticoagulation therapy, as appropriate, could potentially reduce their risk of stroke recurrence.

We included a retrospective cohort of in-hospital patients with ischemic stroke referred for transthoracic echocardiography after hospital admission between 2010 and 2013. The final diagnosis, based on the TOAST classification,4 was established by the treating neurologist.

Echocardiography Acquisition and Analysis

A comprehensive 2-dimensional transthoracic echocardiography study with conventional Doppler and tissue Doppler imaging was performed using the commercially available system Philips IE33 (Andover, Massachusetts, United States). All patients were in sinus rhythm during echocardiography acquisition. Two-dimensional echocardiographic images were taken at a rate above 50 frames/s. Images were digitally stored for off-line analysis in a workstation by 2 echocardiographers blinded to the final diagnosis of the stroke subtype. Left atrial and left ventricle dimensions were determined according to current recommendations.33 Left atrial volume was measured at the end of LA diastole and systole on apical 4-chamber views. The LAEF was calculated as [(LA diastole-LA systole)/LA diastole – LA systole) × 100. Left atrial longitudinal deformation was analyzed off-line from 2-dimensional echocardiography, using speckle tracking derived strain and strain rate, with a commercially available software package (Qlab version 7.1, Philips). From the apical 4-chamber view, 3 points in the LA (2 in the mitral annulus and 1 in the LA roof) were indicated and the endocardial border was manually traced using a point-and-click technique. The software divided the LA wall into 6 segments and the average was used for analysis. The following LA parameters were determined: LAS during ventricular systole, LASR during ventricular systole, LASR during early ventricular filling phase, and LASR during late diastole/atrial contraction (LASRa). The onset of the QRS complex on the surface electrocardiogram21–23 was selected as the reference point for strain analysis (Figure 1). The LASRa was chosen as the strain reference value because of its stronger relationship with AF onset in the general population compared with LASR during ventricular systole.26

Figure 1.

Left atrial strain profiles measured using Qlab (Philips). A: 2-dimensional echocardiographic apical 4-chamber view; the left atrial endocardial border was manually traced using a point-and-click technique; B: left atrial global longitudinal strain; C: left atrial longitudinal strain rate waves. LASRa, left atrial strain rate during late diastole/atrial contraction; LASRe, atrial strain rate during early ventricular filling phase; LASRs, left atrial strain rate during ventricular systole; LASs, left atrial strain during left ventricular systole.

(0.25MB).

We studied LA function in patients with different subtypes of ischemic stroke as defined by the TOAST classification. Patients with cardioembolic stroke had significant impairment of LA function (LAEF and LAS). None of the patients with stroke of undetermined etiology had previously known AF but almost half of them showed impaired LA function similar to patients with cardioembolic stroke, suggesting major LA disease that could have been the source of the ischemic stroke. Patients with other stroke subtypes did not have a definite pattern of LA function, although patients with stroke due to large-artery atherosclerosis also showed a slightly impaired LA function.

Previous studies12-15,17 have reported that up to 28% of patients with stroke of undetermined etiology develop AF during follow-up. These patients showed an increased risk of recurrent stroke unless they were administered appropriate treatment. Longer term Holter monitoring (30 days) is a new recommendation of recent guidelines.10 Like other authors,27 we have shown that the analysis of LA function could also detect or predict AF on admission and is more comfortable for patients, as images are already acquired for a routine echocardiographic examination and only additional post-processing analysis is needed. Long-term monitoring is expensive and has limited availability. Therefore, echocardiography is a cost-effective study to select patients who could benefit from long-term monitoring.

During the cardiac cycle, the LA acts as a reservoir, receiving pulmonary venous return during left ventricular systole; as a conduit, passively transferring blood to the left ventricle during early diastole; and as a pump, actively filling the left ventricle in late diastole. All of these functions participate in the atrial contribution to stroke volume.34 Left atrial ejection fraction is a marker of overall LA function and has been related to AF onset.15,24 In our study, LAEF was decreased in patients with cardioembolic stroke, as previously reported, as well as in patients with stroke of undetermined etiology.15 Left atrial ejection fraction < below 50% has been proposed as a cut-off point to rule out or suspect cardioembolic stroke,15 as well as to identify paroxysmal AF.20 In our population, only 1 patient with cardioembolic stroke had an LAEF>50%. Based on previous studies that related LAS to paroxysmal AF onset,16,26 we proposed the addition of LAS analysis to LAEF not only to rule out the diagnosis of cardioembolic stroke, but also to potentially identify patients with stroke of undetermined etiology that might be misclassified as cardioembolic stroke in order to extend the etiologic study with long-term monitoring and start anticoagulation therapy.

Of note, most patients had impaired reservoir function (depicted by LAS and LASR during ventricular systole), which was likely related to LA fibrosis35,36 and left ventricle function. In patients with stroke of undetermined etiology, LA strain during ventricular systole has recently been proposed as a useful tool to detect patients with occult paroxysmal AF.27 In our study, impairment of LAS during ventricular systole was similar in the groups with strokes due to undetermined etiology, large-artery atherosclerosis, and small-vessel occlusion and was more reduced in the group with cardioembolic stroke. The LASRa was selected as the LA function marker of occult AF because of evidence that it is superior to LAS during ventricular systole for the detection of paroxysmal AF in the general population.26 The LA booster function (LASRa) was lowest in the subgroups with cardioembolic stroke (−0.86±0.49 s-1) and stroke of undetermined etiology (−1.31±0.56 s-1), but was much less reduced in the large-artery atherosclerosis group (−1.5±0.47 s-1). The classification of patients according to LAEF and LASRa (Figure 2) showed that patients with cardioembolic stroke were clearly confined in the quadrant of low LAEF and low LASRa, suggesting a strong relationship between LA dysfunction and cardioembolic etiology. Previous studies have also described the association of decreased LAS and stroke among patients with persistent chronic AF28,29; our findings confirm this previously reported association and raise the question of a potential role of the analysis of LA function with LAS in patients with cardioembolic stroke to improve their risk stratification for recurrent stroke.

Patients with stroke of undetermined etiology had impairment of LAEF or LASRa and almost half of them had low values for both parameters. Patients with stroke of other determined etiology and the control group had a similar distribution of these values; finding this could be explained by the fact that these “other” etiologies of stroke (nonatherosclerotic vasculopathies, hypercoagulable states, hematologic disorders) are less related to systemic atherosclerosis. Regarding the remaining patients with other stroke subtypes, they were diffusely distributed in all the quadrants, suggesting an absence of relationship with LA function. Patients with small-vessel occlusion stroke did not show significant differences in LASRa compared with the control group. The finding that patients with stroke of undetermined etiology had significant impairment of LA function that differed significantly with the other subtypes (except cardioembolic stroke) indicates an association between LA dysfunction and stroke of undetermined etiology. Patients with stroke of undetermined etiology and impaired LA function (both LAEF and LASRa) could potentially have paroxysmal AF and, therefore, there may indeed have been a cardioembolic source of stroke.

Left atrial volume has been related to AF onset,16 as well as to the incidence of stroke in patients without known AF.18,19 In our cohort, LA volume did not show significant differences between groups, despite a trend to larger LA volumes in patients with cardioembolic stroke; in fact, this trend could be secondary to the presence of a proportion of patients with cardioembolic stroke who had previous known paroxysmal AF (36.4%). The similar LA volumes in patients with stroke of undetermined etiology and in healthy controls suggest that this parameter is less useful to reclassify the subgroup of patients with stroke of undetermined etiology with a misclassified cardioembolic stroke.

Study of LA function with transthoracic echocardiography is feasible and potentially helpful to discriminate those patients with ischemic stroke of undetermined etiology and significant LA dysfunction that might be misclassified as having cardioembolic stroke. Long-term monitoring could be appropriate in these patients, who could potentially benefit from anticoagulant therapy.

None declared.