Comparison of right-to-left shunt characteristics in cryptogenic embolic ischemic stroke and non-cardioembolic ischemic stroke

Abstract Background: Patent foramen ovale (PFO) has been considered a potential mechanism of embolic stroke of undetermined origin. Objective: The aim of the present study was to identify the features of the right-to-left shunt (RLS) in patients with undetermined embolic ischemic stroke and compare them with those of patients with non-cardioembolic ischemic stroke. Methods: A retrospective study was conducted with 168 patients with stroke and RLS separated into the following two groups: the undetermined embolic stroke group (UES group) and non-cardioembolic stroke group (NCES group). All patients were assessed by transcranial Doppler to evaluate the presence and quantification of microembolic signals (MES) at rest and under Valsalva maneuver. Results: Of all patients evaluated in the current study, 96 were included in the UES group and 72 in the NCES group. In the UES group, 65 patients had RLS with ≥10 MES (67.7%), which was higher than that observed in the NCES group (51.4%, p=0.038). According to the moment of the cardiac cycle, 75 patients (78.1%) in the UES group had a positive test at rest compared to 42 (58.3%) in the NCES group (p=0.007). Conclusions: The current study demonstrated that almost 70% of patients with undetermined embolic stroke and PFO presented a large RLS and more than 75% had RLS at rest. These findings suggest that the size of the shunt should be taken into account when evaluating whether PFO could be a possible mechanism underlying cryptogenic stroke.


INTRODUCTION
Up to 30% of all ischemic stroke events are of undetermined etiology 1 . Paradoxical embolism, related to a right-toleft shunt (RLS) through a patent foramen ovale (PFO), is one of the possible mechanisms 2,3,4 .
PFO, a persistent congenital opening in the interatrial septum, is found in about 25% of the adult population 5 . In young patients with cryptogenic stroke, this prevalence can be as high as 56% 6,7 . In 2017, three large randomized controlled trials (RESPECT, CLOSE, and REDUCE) suggested that percutaneous closure of PFO could be more effective than medical therapy in preventing recurrent stroke in patients younger than 60 years, despite increasing the risk of atrial fibrillation/flutter 8,9,10 .
RLS diagnostic screening is performed with contrast transcranial Doppler (c-TCD) or transesophageal echocardiography (TEE), with advantages and disadvantages in each method. c-TCD is a highly sensitive test to assess RLS, but it cannot differentiate between cardiac and pulmonary shunts. On the other hand, c-TCD is not invasive and does not require patient sedation; therefore, the patient is more amenable to undergo an adequate Valsalva maneuver. Many studies have shown high sensitivity and specificity of c-TCD compared to c-TEE 11,12,13 . The aim of the present study was to use c-TCD to identify the features of RLS in patients with embolic cryptogenic ischemic stroke and compare with findings of patients with non-cardioembolic ischemic stroke.

METHODS
This study was a retrospective analysis of a prospective databank, including patients with ischemic stroke and RLS. These patients were selected from the outpatient clinics of the Hospital de Clínicas of the Universidade Federal do Paraná (UFPR) and of the Universidade Estadual Paulista "Júlio de Mesquita Filho" (UNESP). Patient's selection occurred between March 2014 and October 2016. The study was approved by the local Ethics Committee and the board waived the need for patient consent.
All patients underwent a brain scan (axial computed tomography or magnetic resonance image) to confirm the presence of a lesion consistent with a diagnosis of ischemic stroke. Patients were separated into the following two groups: undetermined embolic stroke group (UES group) and non-cardioembolic stroke group (NCES group).
Inclusion criteria for the UES group were as follows: 1) adult patients (≥18 years old) with non-lacunar stroke detected by CT or MRI; 2) absence of extracranial or intracranial atherosclerosis causing ≥50% luminal stenosis in arteries supplying the area of ischemia; 3) no major-risk cardioembolic source of embolism (confirmed by transthoracic echocardiography and electrocardiogram); 4) no other specific cause of stroke identified (e.g., arteritis, dissection, migraine/vasospasm, drug misuse) 14 .
Inclusion criteria for the NCES group was based on the ASCOD definition: a minimum investigation with echocardiogram (ECG) and transthoracic echocardiogram (TTE) showing absence of cardioembolic mechanism (C1 and C2) plus the presence of one potentially causal mechanism (1) or of uncertain causality (2): atherosclerosis (A), small vessel disease (S), other etiologies (O), or dissection (D) 15 . An exception for this criteria was the combination of PFO and atrial septal aneurysm (ASA), which is classified as C2 in the ASCOD classification. These patients were included in the statistical analysis if they had normal ECG and TTE, for being considered subject of interest in this study.
The following variables were analyzed: age, sex, and vascular risk factors (hypertension, diabetes mellitus, hyperlipidemia, and smoking habit). The RoPE score was determined and considered in both groups for PFO risk analysis 16,17 . A cutoff of 6 was chosen to identify patients with high risk of mortality and recurrent ischemic events 18 .
The main tool used for diagnosing RLS in all patients was c-TCD (Doppler-Box DWL, Singen, Germany). The studies were performed by trained neurologists with a standard protocol briefly presented as follows. Two 2-MHz pulsed Doppler transducers were fixed with a head frame (DiaMon DWL, Singen, Germany) and both middle cerebral artery (MCA) main stems were concomitantly insonated with the M-mode technique (32 sample gates in each channel) through the temporal window at a depth of 50 to 65 mm with two spectral gates (bigate technique) 8 mm apart. For the contrasting solution, we combined 8 mL of saline solution with 1 mL of the patient's blood and 1 mL of air in a 10-mL syringe. The test was repeated three times at rest and three times under provocative maneuvers.
The results from the bigate spectrogram and the M-mode test were combined for each MCA to improve the sensitivity for microembolic sign (MES) identification. All tests were recorded for 60 seconds with an interval of 3 minutes between the tests for posterior analysis.
The MES identification was based on the international criteria 19,20 . RLS shunt was considered present when at least one MES was detected on both spectral gate displays and a bright red color was observed on the m-mode of at least one monitored MCA. The quantification of MES was registered according to the passage during rest and under Valsalva maneuver (VM). The size of the shunt was estimated according to the quantification of microbubbles (MBs). Patients were classified into three levels: no MBs, negative test for RLS; ≤10 MBs, small or non-significant RLS; >10 MBs or significant RLS. Although there are some cut-off variations, the consensus seems to be that RLS is present if multiple MBs are detected (more than one) and is considered significant whenever more than 10 MBs are detected 21,22,23,24

RESULTS
Of the 168 patients with ischemic stroke and RLS evaluated in the current study, 96 were included in the UES group and 72 in the NCES group. Demographic variables are presented in Table 1. There were no differences in age and sex between groups. Diabetes was more frequent in the NCES group (9.4 vs. 20.8%; p=0.045), as well as hyperlipidemia (29.2 vs. 48.6%; p=0.015). A RoPE Score above 6 was found in 33.3% of the UES patients against 26.4% in the NCES group, but this difference was not statistically significant (p=0.397).
In the UES group, 64 patients (66.6%) were additionally investigated with TEE. Only eight patients (12.5%) had discordant tests results, with a negative RLS in TEE and positive RLS in c-TCD, which can be explained by increased sensitivity of the c-TCD. In this group, eight patients (12.5%) had an associated ASA. In NCES group, 66 patients (92%) were submitted to TEE and only two (3%) of them had ASA. In the UES group, prolonged cardiac monitoring for 24 hours was obtained in 27 (28.1%) patients; none of them presented clinically relevant arrhythmias.
In the UES group, 65 patients (67.7%) had RLS with ≥10 MES, which was higher than that observed in the NCES group (37 patients [51.4%], p=0.038). Considering the presence of MES at rest or under Valsalva maneuver, 75 patients (78.1%) in the UES group had a positive test at rest compared to 42 (58.3%) in the NCES group (p=0.007). Table 2 presents RLS characteristics in both groups.

DISCUSSION
The current study demonstrated that more than two thirds of patients with UES had large shunts. Also, most of them occurred at rest, which was significantly higher compared to patients with NCES mechanism. These findings suggest that, in patients with PFO, there is a link between features of RLS and stroke causality.
Previous studies demonstrated that the RoPE score could stratify the impact of the PFO as the mechanism of ischemic stroke and its recurrence likelihood 17 . This connection was not observed in our study. One possible explanation could be the fact that the RoPE Score is based on the information from the medical history of the patients, neglecting size and moment of the shunt and anatomical issues (association of ASA).
A higher frequency of diabetes mellitus and dyslipidemia was found in patients with NCES compared to patients with UES. This could be explained by the selection criteria for both groups, including patients with small and large vessel disease (A1, A2, S1 and S2) in the NCES group and excluding them from the UES group (no lacunar lesion and the absence of significant extracranial or intracranial atherosclerosis). The relationship between these two risk factors and stroke mechanisms was previously reported in some studies 25,26,27 . On the other hand, a lower prevalence of vascular risk factors in cryptogenic stroke has been very well documented 1,28 .
In the CLOSE trial, a reduction of recurrence was observed in patients with cryptogenic ischemic stroke and PFO treated by percutaneous closure. That trial selected patients with atrial septal aneurysm or large interatrial shunt, a subgroup considered at high risk of recurrence 8 . Meta-analyses including the main PFO studies confirm that patients with large RLS benefit more from PFO closure than those with small RLS 29,30,31 . In the current study, patients with PFO and undetermined embolic ischemic stroke presented larger RLS than patients with PFO and atherothrombotic ischemic stroke, suggesting that the quantification of RLS could represent a single marker for embolic stroke related to PFO.
Besides the size of the RLS, another important aspect to be considered is the presence of MES at rest and under  the Valsalva maneuver. In the pathophysiological process, the RLS at rest could represent a "continuous" opened PFO, where the paradoxical embolism could occur at any moment of the cardiac cycle, independent of the venous return increase 11,18,32 . This was previously considered in the decision to perform PFO closure 19 . This study had some important limitations. Patients were selected from an outpatient setting, after the acute phase of stroke. It is possible that we selected patients with a more benign profile by excluding those who died at the hospital or were disabled and could not show up for medical appointments. However, PFO is related to small embolic events more than to malignant infarctions 5 . Besides, for RLS diagnosis, the test sensitivity is the same whether the stroke is acute or chronic.
Patients included in the current study could not be considered as having embolic stroke of undetermined source (ESUS) because not all of them were submitted to at least 24 hours of Holter monitoring, which is an important limitation. Additionally, TEE was not performed in all patients to confirm the diagnosis of associated ASA or the size of the shunt. Meanwhile, c-TCD is a better method than transesophageal echocardiography to quantify the RLS 33,34,35 . Finally, we did not analyze the importance of the current findings in stroke recurrence, as our aim was to study the aspects of the RLS (intensity of the RLS and positivity at rest).
In conclusion, the current study demonstrated that 78% of patients with undetermined embolic stroke and RLS had positive test results at rest; almost 70% presented a large RLS. More studies are needed to better elucidate the link between RLS and cryptogenic stroke, but we believe that the features of the shunt should be taken into account for risk stratification and management.