Prevalence of left atrial septal pouch among patients with embolic stroke of undetermined source or stroke of known etiology: A retrospective study

Introduction

Strokes can be divided into two main groups: ischemic, which includes nearly 85% of strokes, and hemorrhagic, which includes the other 15%.^[1,2] In approximately 40% of cases, no origin is found,^[3] and they are referred to as cryptogenic strokes.

One of the first definitions of cryptogenic stroke was included in the classification of the Trial of Org in Acute Stroke Treatment (TOAST), which was proposed in 1993.^[4] The major concern about the TOAST classification is that a stroke can be classified as cryptogenic, simply because a complete workup was not performed to determine its etiology. For example, the TOAST classification does not consider results from the latest imaging modalities, such as magnetic resonance angiography (MRA) or rhythm monitoring for detecting atrial fibrillation (AF). MRA was first introduced in 1992,^[5] and currently, it has replaced or is used to complement color-coded Doppler ultrasound. With these advances in diagnostic technology, new classifications have been developed as the causative classification system (CCS) and the Chinese ischemic stroke sub classification (CISS).^[6,7] Consequently, in 2014, a strict definition was suggested for strokes without known origins, and this stroke category was called the Embolic Stroke of Undetermined Source (ESUS).^[8] An ESUS is a non-lacunar cerebral infarct, where a proximal stenosis or major thromboembolism cannot be identified in an extensive workup. The workup should include ECG-Holter monitoring to detect AF, echocardiography to search for cardiac sources of the embolism, and MRA and Doppler ultrasound to exclude intra- and extra-cranial arterial stenosis.^[8,9] The criteria for diagnosing embolic stroke of undetermined source (ESUS): (1)Ischemic stroke that is not lacunar detected by computed tomography (CT) or magnetic resonance imaging (MRI). Lacunar stroke is defined as a subcortical infarct ≤1.5 mm (CT) or ≤2 mm (MRI); (2)Absence of extracranial or intracranial atherosclerosis that causes ≥50% lumen stenosis in arteries that supply the region of infarction; (3) No major cardioembolic risk that could indicate the source of the embolism. Examples of risk factors: permanent or paroxysmal atrial fibrillation, sustained atrial flutter, intracardiac thrombus, prosthetic cardiac valve, intracardiac tumor, mitral stenosis, recent myocardial infarction (<4 weeks), left ventricular ejection fraction <30%, valvular vegetation or infective endocarditis, or any condition that requires a minimum diagnostic evaluation that includes rhythm monitoring for more than 24 h; (4)No other specific cause of stroke identified (e.g., arteritis, dissection, vasospasm, drug abuse).^[8]

Recently, a left atrial septal pouch (LASP) abnormality was suspected to be involved in ESUS. LASPs may be present in up to 25% of patients. A LASP results from an incomplete fusion of the primum and secundum septum, where a pouch is formed on the left side, without an atrial shunt. A LASP is thus defined as the fusion of septum primum and septum secundum at the caudal end of the foramen oval with a flap on the left side creating a pouch that opens towards the left atrium. A pouch may also form on the right side or even bilaterally.^[10,11] As reported regularly in case reports,^[12–19] a LASP can cause the formation of thrombi, and therefore, it could give rise to embolic events.

Several studies have investigated a potential association between a LASP and cryptogenic stroke, but they have reported contradictory results.

Tugcu et al.,^[20] Wayangakar et al.,^[21] and Strachinaru et al.^{[22, 23]} did not find an association between LASP and cryptogenic stroke. Sun et al.^[24] found an association between LASP and stroke. Wong et al.,^[25] Yilmaz,^[26] Holda et al.^{[27, 28]} and more recently Kapoor et al.^[29] all found an association between cryptogenic stroke and the presence of a LASP.

Those studies have had several weaknesses, including a small number of patients, a weak definition of cryptogenic stroke, based on the TOAST criteria, and comparisons with a control group without stroke rather than comparing to patients with strokes of known origin.

Here, we undertook a retrospective analysis of transesophageal echocardiograms (TEEs) in a large, monocentric cohort of patients that presented with an ischemic stroke. This study had three aims: (1) To determine whether a LASP might predispose to an ischemic stroke of undetermined origin. (2) To determine the LASP screening rate in our Cardiology department. (3) To determine whether the pouch dimensions might influence on the risk of developing an ESUS.

Material and method

Study groups

Between June 2012 and March 2021, 1714 patients underwent a TEE for a suspected stroke or transient ischemic attack (TIA) in our clinic of Cardiology. For the present study, we retrieved all these TEEs from our existing database. We excluded TEEs of patients with another final diagnosis (e.g., epilepsy, migraine, dizziness, medication, metabolic toxicity etc.). Next, two observers (the first author and an experienced echocardiographist) independently performed blinded reviews (i.e., without knowledge of the stroke classification) of the remaining 1152 patient TEEs to look for inter-atrial septum morphology that indicated a LASP. Due to incorrect imaging of the interatrial septum, 26 TEEs were excluded. The final analyses included 1126 TEEs (Figure 1). The Ethics Committee of Brugmann’s Hospital, Brussels, approved this study (CE 2021/23).

Figure 1

Flowchart outlines the selection/exclusion processes for the study. TEE: transesophageal echocardiogram; TIA: transient ischemic attack; ESUS: embolic stroke of undetermined source; LASP: left atrial septal pouch.

Next, we reviewed the medical files of all included patients to collect demographic data, medical histories, and information about the type of ischemic stroke. Strokes were evaluated with the ESUS criteria. An ESUS was defined as a non-lacunar cerebral infarct without proximal stenosis or a major identified thromboembolic cause, based on an extensive workup. The workup included rhythm monitoring for more than 24 h to detect AF, echocardiography to identify the cardiac source of the embolism, and an MRA and Doppler ultrasound to exclude intra- and extra-cranial arterial stenoses.

Patients were divided according to the type of stroke, as follows: patients with ESUS (ESUS+ group, N = 148) and patients with strokes of known origin (ESUS− group, N = 978). This study design was adopted to avoid a selection bias in the control group, and it allowed a direct comparison with previous studies on the patent foramen oval (PFO).^{[30, 31]}

Transesophageal echocardiography

When we examined TEEs, we looked carefully at the interatrial septum from several views. TEEs were performed with the injection of agitated saline at rest and during a Valsalva maneuver. Abnormalities included a PFO (considered present, when bubbles appeared in the left atrium within 3 cardiac cycles following opacification of the right atrium), an atrial septal aneurysm (defined as a septum excursion greater than 10 mm measured in TM mode), and a LASP.

A LASP was defined as a visible, incomplete fusion of the septum primum and septum secundum on the left side, in the absence of shunting across the inter atrial septum, based on a contrast injection (Figure 2). When contrast injection to exclude a PFO was not discriminating we classified the patients as having no LASP. The length and width of the LASP were also recorded.

Figure 2

Plane view of a left atrial septal pouch (arrow) during an injection of agitated saline in the right atrium. LA: left atrium; RA: injected right atrium.

Results

LASP as a predisposing factor for ESUS

LASP was present in 176 patients. Among these, 32 patients (21.6%) were in the ESUS+ group and 144 patients (14.7%) were in the ESUS− group. Results of univariate and multivariate analysis are shown in table 2. The prevalence of LASP was higher in the ESUS+ than in the ESUS− group and multivariate analysis shows that LASP is independently associated with ESUS (P = 0.019). Figure 3 shows the odd ratio (OR) for having a cryptogenic stroke, defined with ESUS criteria.

Figure 3

Graphic representation of odd ratio (OR). LASP: left atrial septal pouch; ESUS: embolic stroke of undetermined source; HTA: Hypertension.

Association between LASP dimensions and ESUS

The mean LASP length and width were 11.72 mm and 2.80 mm, respectively. The mean LASP dimensions were almost the same between the ESUS− group (length: 11.59 mm and width: 2.82 mm) and in the ESUS+ group (length: 12.33 mm and width: 2.74 mm). No significant difference was observed between groups (P = 0.398 for the length and P = 0.681 for the width). No thrombi were found in any of the LASPs.

Figure 4

Evolution of the rate of detecting a left atrial septal pouch, from 2012 to 2021.

Discussion

Left atrial septal pouch was first described on autopsied hearts^{[11,32, 33]} and its prevalence is high, reaching almost 50%. Holda et al. proposed that the fusion of septum primum with septum secundum is a lifelong continuous process leading to a spectrum of incomplete fusions (PFO, LASP and right atrial septal pouch) where the most incomplete fusion, represented by the PFO, is more prevalent at a younger age.^[11] These authors analyzed the anatomical features of these pouches on autopsied hearts and found that LASP has a complex shape with secondary diverticula that promote stasis and can lead to thrombi formation.^[11]

In addition to the pouch morphology, other factors have been proposed to favor thrombus formation.^[12] One essential factor is circulatory stasis, which is of course the case in atrial fibrillation but also in situations where left atrial pressure rises, like in mitral stenosis or heart failure. Another possible mechanism could be endothelial injury with inflammatory, as described in a histopathology of a resected thrombus.^[12]

As a consequence of this thrombogenic potential it is logical to think that LASP is not only an innocent structure but could be a cause of stroke. However, the clinical importance of LASP is still debated.

Previous studies that investigated the potential impact of a LASP on stroke risk have presented divergent results. Holda et al. performed a retrospective study to compare patients with stroke to controls without stroke. They found an association between the presence of a LASP and the development of a cryptogenic stroke.^[28] Wong et al. showed similar results in comparing patients with cryptogenic strokes, defined with the TOAST criteria, and patients with strokes of known origin.^[24] Recently, a retrospective study by Kappor et al. demonstrated that patients with LASP had a significantly increased risk of cryptogenic stroke. However, they also used the modified TOAST criteria and compared a stroke group to a control group without stroke.^[29]

In contrast, Tugcu et al.,^[20] in 2010, and Wayangankar et al.,^[21] in 2012, did not find any association between LASP and stroke, when they compared patients with stroke to controls without stroke. However, both studies included a small number of patients with cryptogenic stroke (respectively, 69 and 44 patients), defined with the TOAST criteria. Similarly, Strachinaru et al., in 2016, could not find an association between LASP and cryptogenic stroke, when they compared a cryptogenic stroke group, defined with the TOAST criteria, to a control group without stroke.^[22]

Our study included a large number of patients, and we found a strong association between the presence of a LASP and ESUS. Unlike most previous studies, we defined a cryptogenic stroke with precise criteria (ESUS criteria). Our definition included a more complete workup and updated tools for investigating the origin of the ischemic stroke. Furthermore, we compared two groups of patients with different types of strokes, which prevented a selection bias and reinforced our results.

Our overall LASP prevalence was 15.6%. Based on a review of the TEEs, we found that 61.9% of LASPs were not mentioned by the original TEE operators. However, we noted that this discordance progressively declined over time (P = 0.015); in 2012, 83.4% of LASPs were not mentioned in the original reports; but in 2021, only 40% of LASPs were not mentioned in the original reports. This increase in the rate of LASP descriptions indicated an increased awareness among cardiologists to this newly described structure.

Nevertheless, the prevalence of LASPs was far lower than the LASP prevalence reported previously for autopsied patients. Krishnan and Salazar^[32] reported a 39% prevalence in their series, Holda et al. a 47% prevalence^[11] and Klimek-Piotrowska et al.^[33] reported a 50% prevalence. This difference in prevalence observed between autopsy studies and in vivo studies could likely be explained by the pressure regimen on the foramen oval membrane in a living heart, which is very different from the pressure in a bloodless heart. Some of the pouches observed during autopsies probably did not display any signs in vivo.

Although previous reports have shown that thrombi can occur in these pouches,^{[12, 13, 14, 15, 16, 17, 18, 19]} we did not find any thrombi in the 176 LASPs that we reviewed (although we have already seen thrombus in a LASP in a patient not included in the present study because stroke was not the indication for TEE, Figure 5). One explanation could be that the LASP is a small, non-trabeculated pouch, and thus, thrombi are not retained for long times. Two recent studies^{[34, 35]} compared anticoagulation to antiplatelet therapy in patients with ESUS, but neither study could show superiority. A third study is still ongoing.^[36] These negative results could be explained by the heterogeneity of embolic source in ESUS patients.^[37] Nevertheless, the thrombus observations and our results have given rise to the question of whether it might be appropriate to prescribe anticoagulation therapy to some patients with a LASP that experienced an ESUS.

Figure 5

Thrombus in a LASP (arrow). LASP: left atrial septal pouch.

Our study was designed to evaluate cryptogenic stroke, which, by definition, excludes atrial fibrillation. This was done with a quite strong workup, even stronger than it was done in other studies, as in CLOSE study for example.^[38]

The role of AF and other factors (as coagulation abnormalities for example) in thrombus formation in a LASP could be an interesting subject to study.

Our study also demonstrated that the anatomical dimensions of the LASP were not associated with an ESUS. Therefore, special attention should be given to even small pouches during routine examinations. Future studies should confirm this result with larger patient cohorts.

Study limitations

The main limitation of this single-center study was its retrospective design. Thus, the TEE procedure was not designed to search for LASPs, in particular. However, all the TEEs in both groups were reviewed blindly by two independent observers, which prevented potential biases in data collection. Contrast injection to exclude a PFO was sometimes not discriminating and these patients where therefore classified as having no LASP, this could induce an underestimation of LASPs. Another potential limitation was the relatively small number of patients in the ESUS+ group (148 patients). However, to our knowledge, this was the largest ESUS+ cohort studied to date.

Conclusion

To our knowledge this is the first study on LASP, to date, to use ESUS definition for classifying a cryptogenic stroke. We found a higher prevalence of LASP in patients with ESUS compared to patients with strokes of known origin. Given this result, we recommend that patients with a cryptogenic stroke should be routinely screened for the presence of a LASP. Moreover, our results gave rise to the question of whether anticoagulation might be an appropriate treatment for some patients with a LASP that presented an ESUS. Future, large-scale, prospective studies should be conducted to address this issue.

We also found that the rate of describing LASPs was low among TEE operators. Therefore, we strongly recommend that TEE operators should be made aware of the importance of searching for LASPs in TEEs.