A novel scoring system for stroke risk stratification in Japanese patients with low CHADS2 scores: Study using a transesophageal‐echocardiogram endpoint

Abstract Background Catheter ablation is an effective treatment for atrial fibrillation (AF), but it carries risk of perioperative thromboembolism even in cases with low CHADS2 scores. Here, we examined whether a combination of clinical variables can predict stroke risk factors that are assessed by transesophageal echocardiography (TEE). Methods The study population consisted of 209 consecutive AF patients with a CHADS2 score of 0 or 1 (58.7 ± 10.6 years old; persistent AF, 33.0%). All patients underwent TEE, and TEE‐determined stroke risk (TEE risk) was defined as cardiac thrombus/sludge, dense spontaneous echo contrast (SEC), and/or peak left atrial appendage (LAA) flow velocity <0.25 m/s. Results Transesophageal echocardiography risk was observed in 10.5% of the patients. In multivariate logistic analysis, persistent AF [odds ratio (OR): 11.5, CI: 3.14‐42.1, P = .0002], left atrial diameter (LAD) (OR: 1.10, CI: 1.01‐1.20, P = .0293), contrast medium defect (CMD) in the LAA detected by computed tomography (OR: 20.2, CI: 6.3‐65.0, P < .0001), and serum brain natriuretic peptide (BNP) level (OR: 1.00, CI: 1.00‐1.01, P = .0056) were independent predictors of TEE risk. A new scoring system comprising LAD > 41 mm (1 point), BNP > 47 pg/mL (1 point), CMD (2 points), and persistent AF (2 points) was constructed and defined as TEE‐risk score. The area under the curve (AUC) for prediction of TEE risk was 0.631 in modified CHADS2 score and it was 0.852 in TEE‐risk score. Conclusion Transesophageal echocardiography risk is predictable by TEE‐risk score, and its combination with CHADS2 score may improve the stroke risk stratification in AF patients with a low CHADS2 score.


| INTRODUC TI ON
Catheter ablation is an effective treatment for patients with atrial fibrillation (AF) and the number of procedures using catheter ablation has been increasing. One of the most important periprocedural complications is thromboembolism. Accurate risk stratification for stroke and appropriate anticoagulation management are therefore essential. The CHADS2 score is useful for stratifying the risk of stroke and thromboembolism in patients with AF and is widely used for assessment of the eligibility for anticoagulation therapy. It has a great advantage of not requiring an expensive medical examination, but its level of accuracy is reportedly modest. In a Japanese registry 1 in which AF patients without anticoagulation therapy were enrolled, the annual incidences of stroke were 0.54% in patients with a CHADS2 score of 0% and 0.93% in patients with a CHADS2 score of 1. In another Japanese registry, 2 patients with a low CHADS2 score (0 or 1) accounted for half of the total population of AF patients, suggesting that a low CHADS2 score group forms a considerable stroke source in the entire Japanese society despite the low annual incidence of stroke in Japan. Thus, for better stratification of stroke risk in AF patients, an index that improves risk assessment in patients with low CHADS2 scores is needed.
Several indices to complement the CHADS2 score for further risk stratification have been proposed. The indices include severity of fibrosis of the left atrium (LA) measured by magnetic resonance imaging, 3 renal insufficiency, 4 sleep apnea syndrome, 5 biochemical markers such as high-sensitivity troponin 6,7 and brain natriuretic peptide (BNP), specific morphology of the left atrial appendage (LAA), 8 and P wave abnormality on a 12-lead electrocardiogram. 9 Transesophageal echocardiography (TEE) is useful for detecting left atrial thrombus, dense spontaneous echo contrast (SEC), and reduced peak LAA flow velocity, which are known to be associated with a high risk of thromboembolic events in patients with AF. [10][11][12][13][14] However, TEE should not be recommended to all AF patients because of its invasive nature and possible complications.
Several surrogate markers based on left atrial dilatation or cardiac function have been proposed for risk stratification of cardiogenic stroke. [15][16][17][18][19][20] However, there has been no study in which those surrogate markers were analyzed in Japanese patients in whom catheter ablation was planned. The aim of this study was to determine simple and practical variables in addition to CHADS2 score that enable prediction of patients at high risk for stroke.

| ME THODS
The protocol for this research project was approved by the Ethics Committee of Sapporo Medical University (number 312-73) and it conforms to the provisions of the Declaration of Helsinki.

| Study population and clinical examinations
The study population consisted of 209 consecutive AF patients with a CHADS2 score of 0 or 1 who were scheduled to undergo catheter ablation during the period from 2009 to 2018 at our institute. Persistent AF was defined as continuous AF that was sustained for more than 7 days. Serum BNP and blood coagulation markers including fibrinogen (FBG), fibrinogen degradation products, d-dimer, and thrombin-antithrombin complex were measured at the time of admission. Enhanced computed tomography (CT) was performed 1 day before the catheter ablation, and three-dimensional geography of the LA was created after resecting the distal pulmonary vein and LAA to calculate the LA volume.

| Statistical analysis
Statistical values are shown as means ± 1SD. Patients were divided into two groups according to the presence or absence of TEE risk.
The Mann-Whitney U test was used for comparison of mean values in the two groups, and the chi-square test was used for comparison of prevalence. A P value less than .05 was considered statistically significant. Binocular logistic analysis was used to identify independent predictors of TEE risk. Variables with a P value less than .05 in univariate analysis were entered into multivariate models. Receiver operating characteristic (ROC) curves were traced for the prediction of TEE risk using LAD and serum BNP level, and cutoff values were determined. Comparison of the diagnostic accuracies of modified CHADS2 score and TEE-risk score for the prediction of TEE risk was performed using area under the curve (AUC) by ROC curve analysis.
The analyses were performed using JMP software (version 8.0.2; SAS Institute).

| Clinical characteristics of study subjects
The baseline characteristics of the study patients and comparison of clinical variables in the two groups with and without TEE risk are shown in Table 1. We enrolled 209 consecutive AF patients with a CHADS2 score of 0 (n = 97) or 1 (n = 112). The mean age of the patients was 58.7 ± 10.6 years, and 73.2% of the patients were male.

| Clinical parameters associated with TEE risk
The results of univariate and multivariate analyses using a binocular logistic regression model for prediction of TEE risk are shown in Table 2.
In univariate analysis, body mass index (odds ratio score and 0.852 in TEE-risk score. TEE-risk score had higher discrimination power as measured by AUC than that of modified CHADS2 score (P = .0025) (Figure 1). The optimal cutoff value of TEE-risk score for prediction of TEE risk was 3 points, and its sensitivity and specificity were 81.8% and 77.6%, respectively. Subgroup analysis was performed in 20 patients with CMD; the mean TEE-risk score tended to be higher in patients with TEE risk than in patients without TEE risk (5.5 ± 0.7 vs 4.2 ± 1.5, P = .059). TEE-risk score predicted the presence of TEE risk in patients with CMD (AUC: 0.742, sensitivity: 66.7%, and specificity: 90.9%) and the optimal cutoff value was 5 points.

| Prevalence of patients at high risk for stroke despite having a low CHADS2 score
It has been reported that cardiac thrombus/sludge, dense SEC, and reduced LAA peak flow velocity in TEE can be observed in a low percentage of patients with a low CHADS2 score of 0 or 1.  showed TEE signs of high cardiogenic stroke risk.

| Clinical significance of risk stratification for stroke prior to catheter ablation
Catheter ablation is a safe and effective treatment for patients with AF, but it carries the risk of periprocedural thromboembolism. Periprocedural major stroke is rare, but silent stroke does not appear to be rare. In recent studies, silent stroke was detected by brain magnetic resonance imaging after catheter ablation in 9.6%~27.2% of patients, 22,23 and Nagao

et al reported transient elevation of coagulation markers after catheter
ablation. 24 Optimal anticoagulation management to minimize periprocedural complications in patients undergoing catheter ablation has not been established. [24][25][26] Hence, we attempted to improve thrombotic risk stratification in patients scheduled for AF ablation by data including TEE and CT. Detailed stratification of AF risk is important before catheter ablation, but it may also be useful for determining the timing of discontinuation of anticoagulation therapy after catheter ablation.

| Surrogate markers for stroke risk detectable by TEE
Although TEE is useful for risk stratification in patients with AF, 10  TEE-risk score for indication of anticoagulation therapy. However, it is reasonable to use TEE-risk score for guiding further examinations (such as brain MRI for screening silent stroke and TEE) in patients with a low CHADS2 score. TEE-risk score >3 predicts TEE risk with sensitivity of 81.8% and specificity of 77.6%, which may justify TEE even when CHADS2 score is <2.

| Limitations
First, we used TEE risk as an endpoint in this study since SEC and reduced LAA peak flow velocity detected by TEE have been shown to be associated with future stroke in previous studies. 10 Third, LAA peak flow velocity is known to be different during sinus rhythm and AF, and the optimal cutoff value for prediction of stroke risk might be different depending on the cardiac rhythm, though we could not perform such subgroup analysis because of the small number of study subjects.

| CON CLUS ION
Even in patents with a low CHADS2 score of 0 or 1, approximately 10% of the patients have TEE signs of high cardiogenic stroke risk, which may be predicted by dilatation of the LA in TTE, serum BNP level, enhanced cardiac CT, and type of AF.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interests for this article.