The differences of atrial thrombus locations and variable response to anticoagulation in nonvalvular atrial fibrillation with ventricular cardiomyopathy

Abstract Objectives This study aims to research the clinical features of atrial thrombi in patients with nonvalvular atrial fibrillation (AF). METHODS This study included 191 patients of AF who had atrial thrombi. One hundred and twenty‐eight of them were assigned into nonventricular cardiomyopathy group (non‐VCM), and the remaining 63 into ventricular cardiomyopathy group (VCM). After atrial thrombi diagnosed, all patients had taken oral anticoagulant therapy. The resolution rates of thrombi within 12 months were compared between the two groups, as well as the locations of thrombi. Results Of all 191 patients, 161 had thrombi only detected in left atrial appendage (LAA), 20 in both left atrium (LA) and LAA, six in LA only, and four in right atrium only. More patients had thrombi out of LAA in the VCM group than in the non‐VCM group (30.2% vs 8.6%, P < .001). After propensity score matching, the atrial thrombi were resolved faster in the non‐VCM group than in the VCM group (mean time length: 22 ± 2 weeks vs 30 ± 3 weeks, P = .038), and the resolution rate within 12 months was higher in the non‐VCM group than in the VCM group (88.7% vs 61.4%, Log‐rank, P = .038). In Cox proportional hazards model, absence of ventricular cardiomyopathy was an independent predictor for the resolution of atrial thrombus (hazard ratio: 1.76; P = .035). Conclusions The patients of atrial fibrillation with ventricular cardiomyopathies have higher incidence of thrombosis in the body of left atrium or right atrium. And the resolution rate was lower in these patients.

some patients with AF, the atrial thrombi can take place out of LAA, and the outcomes of resolution may vary from different conditions. Consequently, we performed a retrospective study to research the clinical features and variable responses to anticoagulation in AF patients complicated with cardiomyopathy and ischemic heart disease. Participants in this study were assigned into two groups: ventricular cardiomyopathy group (VCM) and nonventricular cardiomyopathy group (non-VCM). Cardiomyopathy is a myocardial disorder that the heart muscle is structurally or functionally abnormal in the absence of some secondary disorders as defined by European Society of Cardiology working group 4 . In our study, the VCM group included patients with these cardiomyopathies and ischemic heart disease.

| Comparison of clinical features
The baseline characteristics were compared between the two groups mentioned above, including left atrial antero-posterior diameter (LAAPD), left ventricular end diastolic diameter (LVEDD) and left ventricular ejection fraction (LVEF) measured by echocardiography. And CHA2DS2-VASc, CHADS2, HAS-BLED scores were also collected as well as antithrombotic regimens before diagnosis.

| Follow up
After atrial thrombi diagnosed, patients had taken oral anticoagulants with either vitamin K antagonist or non-vitamin K oral anticoagulants (NOACs), which included rivaroxaban and dabigatran in this study. The patients were scheduled to outpatient clinic at 1, 3, 6, 9 and 12 months, and the same imaging method was repeated for spotting thrombi, until thrombi resolved.

| Comparison between the VCM and NON-VCM groups
Propensity score matching was used to compare the resolution rate of thrombi between VCM and non-VCM groups. The propensity score was generated according to potential covariates, including gender, age, type of AF (paroxysmal or persistent), BMI, history of stroke, history of congestive heart failure, history of hypertension, history of diabetes mellitus, CHA2DS2-VASc scores, LAAPD, LVEF, and concentration of NT-pro-BNP.
Nearest-neighbor matching without replacement was performed.
After matching, Multivariate analysis was used to evaluate independent risk factors affecting the resolution of atrial thrombi in both matching groups. Survival curves were compared with Kaplan-Meier method using a log-rank test, in order to estimate the time length for thrombi resolution.

| Statistical analysis
Statistical analyses were conducted using the SPSS version 26 (SPSS Corp., Armonk, NY, USA). Continuous variables are expressed as mean ± SD, and using the independent-sample T test. Categorical variables are expressed as numbers or percentages, using the Chisquare test or the Fisher's exact test. A P-value less than 0.05 was considered statistically significant.

| Clinical characteristics of participants
We had retrieved 1153 patients diagnosed with atrial fibrillation and atrial thrombus, and 962 patients were excluded because of valvular atrial fibrillation or cardiac surgery. One hundred and ninety-one patients of AF were eligible in our study. There were 128 patients in non-VCM group and 63 in VCM group. The patients in VCM group had larger LAAPD than in non-VCM group (48 ± 8 mm vs 43 ± 5 mm, P < .001), and LVEF was lower in the VCM group than in the non-VCM group (49% ± 15% vs 60% ± 6%, P < .001). More patients suffered congestive heart failure in VCM group than in non-VCM group (60.0% vs 26.2%, P < .001), and the NT-pro-BNP concentration was higher in VCM group than in non-VCM group (2309.1 ± 3093.8ng/mL vs 964.9 ± 1071.7ng/ mL, P = .001). The baseline clinical characteristics and examination results of these patients are shown in Table 1.

| Anticoagulant therapy
Before the atrial thrombi detected, 55.5% of patients (106/191) did not take any oral anticoagulants, including 31.4% only taking antiplatelet drugs ( The non-paroxysmal atrial fibrillation includes persistent, long-standing persistent and permanent atrial fibrillation in this study.

| Follow up of thrombi resolution
After anticoagulant therapy, the median time length for thrombi resolution was 25 ± 1 weeks. The atrial thrombi was resolved faster in non-VCM group than in VCM group (mean time length: 22 ± 1 weeks vs 32 ± 1 weeks, P < .001), and the resolution rate in one year was higher in non-VCM group than in VCM group (85.9% vs 61.9%, Log-rank, HR 2.13, 95% CI, 1.44 ~ 3.14,P < .001).
In the propensity score-matching analysis, 44 patients in the VCM group and 44 patients in the non-VCM group were matched and analyzed. There was no significance in these basic clinical characteristics of 88 patients after PS matching, as shown in atrial thrombi was resolved faster in non-VCM group than in VCM group (mean time length: 22 ± 2 weeks vs 30 ± 3 weeks, P = .038), and the resolution rate in one year was higher in non-VCM group than in VCM group (88.7% vs 61.4%, Log-rank, P = .038), as shown in Figure 3.
In univariate analysis of these 88 patients, there were less pa- was an independent predictor for atrial thrombus resolving.

| D ISCUSS I ON
Few studies had focused on the relationship between atrial thrombus and cardiomyopathy in atrial fibrillation, and the main findings from this retrospective study are as follows: 1. The atrial thrombi tend to arise out of LAA among atrial fibrillation patients with ventricular cardiomyopathy.

The atrial thrombi in patients with ventricular cardiomyopathy
were more difficult for resolution than those without ventricular cardiomyopathy.
3. Ventricular cardiomyopathy was an independent risk for atrial thrombi existing in AF patients.

| Difference of thrombosis locations in atrium
LAA is the most common location for thrombosis and a source of cerebral or systemic embolism in AF patients. 5,6 In patients under regular anticoagulation, the prevalence of left atrial thrombus varies from 1.5% ~ 3.0%. 7,8 In a recent meta-analysis, the results showed that left atrial thrombus was identified in up to 9.8% of overall patients with AF, and is associated with a 3.5-fold increased risk of stroke/systemic embolism. 2 As shown in past studies, during AF there is a decrease in LAA contractility and function, which is prone to thrombosis. 9,10 However, in our study, 15.7% of cases had thrombi formed in LA body or right atrium, and in patients with ventricular cardiomyopathy, this proportion was up to 30%. We supposed that it may be due to the atrial remodeling in patients with ventricular cardiomyopathy.
In our study, the mean anteroposterior diameter of LA was larger in the ventricular cardiomyopathy group than in the non-cardiomyopathy group, indicating more left atrial remodeling.
Moreover in order to elucidate whether this difference was just due to the enlargement of LA, we conducted the subgroup analysis. And we found that, whether in the subgroup of LA ≥45 mm or LA <45 mm, the proportions of non-LAA thrombi were both significantly higher in the patients with VCM than in non-VCM.
These results implied that, the cardiomyopathy may contribute to thrombi formation in the body of LA or right atrium rather than just enlargement of LA.
In some types of cardiomyopathies, such as dilated cardiomyopathy and ischemic heart disease, 11-13 the pathologic changes (ie fibrosis) were not only manifested in the ventricular myocardium, but also in the atrium. These lesions were possibly linked to the stasis in atrium by inflammation and (or) endothelial dysfunction, 14 which may promote thrombosis in the body of LA.
Based on the results, we inferred that, left atrial appendage closure may be unsuitable for some particular patients of AF. As in 2019 AHA/ACC/HRS guideline, left atrial appendage closure may be considered in some selected patients with AF at increased risk to prevent stroke. 15 However, it is a controversial measure for its efficacy. From this study, we found that LAA is not the only one part for thrombosis, while some thrombi could emerge in the right atrium or the body of LA, hence the occlusion of LAA may be unable to reduce thrombosis out of LAA. Note: The propensity score was generated according to potential covariates, including gender, age, type of AF (paroxysmal or persistent), BMI, history of stroke, history of congestive heart failure, history of hypertension, history of diabetes mellitus, CHA2DS2-VASc scores, LAAPD, LVEF and concentration of NT-pro-BNP.

| Anticoagulation therapy before atrial thrombus detected
In our study, over half of these patients did not take anticoagulants until the atrial thrombi detected. Nearly 25% of the patients were admitted before 2015, during which period, guidelines did not establish the status of CHA2DS2-VASc scoring for stroke risk evaluation, and still considered aspirin for some selective patients with AF. [16][17][18] In these patients without anticoagulating, over 60% were stratified into low risk with CHADS2 <2 points, which may contribute to the deficiency of anticoagulation. By revaluation through CHA2DS2-VASc scoring, up to 70% of the patients without pre-anticoagulation were stratified into high risk of stroke with CHA2DS2-VASc ≥2 points. Thus, CHA2DS2-VASc may promote anticoagulation by lifting the risk rating of stroke, thus to decrease atrial thrombosis.

| Resolution of atrial thrombi
As guidelines recommend, VKA is the most used treatment, 15,19 and in some trials, NOACs were also proved to be effective for treating atrial thrombi in patients with AF. [20][21][22] Wang et al conducted a meta-analysis and found that, the pooled resolution rate of left atrial thrombus from 16 warfarin studies was 63.7%, and two studies in direct-acting oral anticoagulants reported this resolution rates of 89.5% for dabigatran and 41.5% for rivaroxaban. 23 In our study, the total resolution rate in one year was about 74% after propensity score matching, and the resolution rate in one year was higher in non-VCM group than in VCM group. The difference of resolution rate may be related to the discrepancy of LA size and LVEF between the two groups. As a result, we conducted the univariate and multivariate analysis from the propensity score matching, and made most variates of baseline consistent. The result showed that the absence of VCM was the only independent predictor for thrombi resolution.
As discussed above, the left atrial diameter was greater in VCM group than in non-VCM group, and the atrial contractility declined in some patients with cardiomyopathy. In addition, the patients with cardiomyopathy in our study had lower mean LVEF and higher incidence of heart failure than patients without cardiomyopathy, which could further increase dysfunction of left atrium. 24,25 All these causes led to slow blood flow in circulation, especially the left atriums, which may increase the difficulty for thrombi resolution.

| Study limitations
This study was retrospectively conducted in a single center, and the sample size was relatively small, especially in the cardiomyopathy group. Hence, the limited size may overestimate the difference of atrial thrombus locations between the two groups. Moreover about 20% of the participants were diagnosed only by CT. As previous study indicated, CT had a poorer positive predictive value for thrombus in left atrial appendage compared with the gold F I G U R E 3 Kaplan-Meier analysis of thrombi resolution between two PSM groups, and there were 44 patients in each group. In one-year follow-up, there were more thrombi resolved in non-VCM group than in VCM group standard of TEE. 26 It may include some false-positive thrombi into this study.