Approximately 17% of ischemic stroke patients receive a diagnosis of embolic stroke with undetermined source (ESUS).1 ESUS patients have a higher risk of stroke recurrence, emphasizing the importance of understanding causation for effective prevention. While atrial fibrillation (AF) has emerged as the predominant mechanism in ESUS, the definitive cause remains undetermined. Potential causes may be identified by assessing atrial cardiomyopathy (ACM),2 even before AF onset, as a substantial portion of the ESUS population have clinical markers of ACM.1 Nevertheless, the correlation between ACM and clinical prognosis post-ESUS remains unclear. This study aimed to evaluate predictors, particularly ACM markers, of cardioembolic event occurrence in ESUS patients over long-term follow-up.

This observational and single-center retrospective cohort study enrolled consecutive patients who experienced ischemic stroke confirmed through brain imaging, excluding transient ischemic attack, at the Stroke Unit of the University Hospital of Saint-Étienne (France) from 2009 to 2012. ESUS was initially defined and confirmed as a negative comprehensive etiologic screening, which included a> 24-hour Holter electrocardiogram (ECG), transesophageal echocardiography, and intracranial artery imaging. The aim was to exclude any ischemic stroke with an identified cause, including patent foramen ovale.3 The study protocol was approved by the local ethics committee, and all patients were duly informed.

Markers of ACM were itemized at baseline through the CHA2DS2-VASc score, P-wave morphology, left atrial (LA) dilatation, increased burden of premature atrial contractions (PACs),4 and the presence of nonsustained supraventricular tachycardia.

Long-term follow-up data were collected from both former and current physicians. The occurrence of a cardioembolic event was a composite outcome that included diagnosis of AF and/or recurrence of ischemic stroke, in which a cardioembolic cause was identified based on the etiologic evaluation.

Patients who were lost to follow-up were censored at the time of their last assessment. Univariate and multivariate analyses were conducted using Fine-Gray proportional subdistribution hazard model regression analysis, including all-cause mortality as a competing event with cardiovascular events. All statistical analyses were performed using IBM SPSS Statistics.

Among the entire cohort of patients with an ischemic stroke, a potential cause was identified in 1094 patients. Out of these patients, 128 met the criteria for ESUS and were included in the study, representing 10.5% of patients with ischemic stroke overall (figure 1). The baseline characteristics of the ESUS population were as follows: the mean age was 61 years, hypertension was present in 68.8%, there was no history of stroke or transient ischemic attack in 87.5%, the median CHA2DS2-VASc score was 2, and LA dilatation was present in 26.6%. Upon discharge, antiplatelet therapy was prescribed in most patients (82.6%).

Figure 1.

The initial presence of atrial cardiomyopathy, characterized by left atrial dilation and/or increased premature atrial contraction burden, is a significant predictor of cardioembolic event occurrence during long-term follow-up. LA, left atrial; ESUS, embolic stroke with undetermined source; PAC, premature atrial contraction.

(0.37MB).

During the median 9.2-year follow-up period, 30 patients (25.0% of the whole cohort) died. The cumulative incidence of a diagnosis of cardioembolic events was 16.9%±4.5%. Among them, 13 patients were diagnosed with AF and 4 patients had a cardioembolic stroke.

Through univariate analysis, several predictors of cardioembolic events during follow-up were identified (table 1). Although not statistically significant, there was a trend suggesting an association between abnormal P-wave axis and reduced left atrial appendage outflow velocity and diagnosis of a cardioembolic source. Interestingly, none of the characteristics of the baseline ischemic stroke were associated with the occurrence of cardioembolic events. Independent predictors of the occurrence of cardioembolic events were the presence of LA dilatation based on baseline transthoracic echocardiography assessment (hazard ratio [HR], 2.87; 95% confidence interval [95%CI], 0.98-8.45; P=.05) and increased burden of PACs during a 24-hour Holter-ECG (HR, 4.13; 95%CI, 1.28-13.39; P=.02) as reported in table 1. At inclusion, 31.3% of the cohort had at least 1 of these independent predictors, defined as the presence of ACM. Patients with LA dilatation and/or increased PAC burden had a 5-fold higher risk of experiencing a cardioembolic event during the long-term follow-up (HR, 5.00; 95%CI, 1.84-13.60; P=.001) as illustrated in the figure 1.

The prevalence of ESUS in our cohort of ischemic stroke patients (10.5%) falls within the lower range reported in several studies (9% to 25%).1 This variance likely highlights the crucial need for a thorough etiological evaluation when assessing ischemic stroke, including transesophageal echocardiography to rule out patent foramen ovale. Ischemic stroke recurrence occurred in 10.2% of the cohort, which is lower than the reported incidence (4% to 5% per year).1 While the retrospective design of the study may introduce ascertainment bias, the lower prevalence of ESUS likely facilitated the initiation of appropriate therapies based on stroke etiology in many patients and reduced long-term complications.

ESUS is a heterogeneous condition with various underlying causes, including unknown peripheral atherosclerosis and cardioembolic sources. ACM is an emerging concept2 that has the potential to improve diagnostic management in ESUS patients. ACM assessment in clinical practice is constrained by the lack of standardized criteria. Recently, the ARCADIA trial did not demonstrate the superiority of an ACM-based therapy in the ESUS population with ACM, likely because of an inaccurate definition of ACM.5 Our study highlights an alternative definition of ACM, based on the presence of a dilated LA and an increased PAC burden, present in one third of the ESUS population at baseline, leading to a 5-fold higher long-term risk of cardioembolic events (figure 1). A randomized clinical trial assessing the benefit of an ACM-based therapeutic strategy as defined in our study is warranted.

In conclusion, the presence of ACM, defined as LA dilatation or increased PAC burden at baseline, is predictive of the diagnosis of a cardioembolic source during long-term follow-up in the ESUS population (figure 1).