Our study investigated the association among LAD, CHA2DS2-VASc score, and LAAFV in 716 patients with NVAF. To the best of our knowledge, this is the first study to demonstrate that patients with a larger LAD and a higher CHA2DS2-VASc score are prone to a decrease in LAAFV. Additionally, the results of the ROC curve analysis showed that the predictive ability of LAD and CHA2DS2-VASc score alone in predicting a decrease in LAAFV was limited. We therefore developed a model combining LAD and CHA2DS2-VASc score. The combined model had a significantly better discriminatory ability, suggesting that combined use of LAD and CHA2DS2-VASc score might be useful as a new surrogate to predict the decrease in LAAFV in patients with NVAF.
The LAA is a major thromboembolic source in patients with AF. As such, many studies have assessed the risk of stroke by analyzing LAAFV [13, 15]. A decrease in LAAFV has been well identified as a surrogate for cardioembolic risk in patients with NVAF. Several studies have shown that a low LAAFV is associated with a higher risk of stroke/thromboembolic events than a high LAAFV in patients with AF [13, 15, 17, 18]. Although TEE is a reliable method to evaluate LAAFV, it is relatively invasive and low yield. Furthermore, knowledge on the factors that influence LAAFV is limited. The LAA is adjacent to the left atrium; thus, the LAAFV is susceptible to LA remodeling. A previous study showed a significant negative correlation between LA volume and LAAFV [20]. In addition, a study by Schnieder et al. reported that LAD is inversely correlated with LAAFV [21]. Our study showed that LAD is negatively and linearly correlated with LAAFV, meaning that an increase in LAD parallels to a decrease in LAAFV. Additionally, the multivariate analysis demonstrated that LAD is an independent risk factor for the decrease in LAAFV after adjusting for other variables. For every additional unit change in LAD, the odds of a decrease in LAAFV in patients with AF increased by 1.098 times. In the subgroup analysis, as the LAD increased, the LAAFV decreased (P < 0.001). These subgroup analyses further validated the relationship between LAD and LAAFV at different levels. In a previous study on patients with non-valvular paroxysmal AF, LAD was an independent predictor of a decrease in LAAFV in patients with sinus rhythm (SR) during TEE [22]. Another study by Fukuhara et al. found that LA volume index could predict a decrease in LAAFV during SR in patients with AF, but a considerable proportion of patients with AF rhythm were excluded from this study [20]. Unlike the abovementioned studies, we did not distinguish between AF rhythm and SR during TEE in our study, suggesting that the conclusion from this study might be universal. In addition, we chose the LAD as the study target because it is easily obtained and more widely used than LA volume index. Notably, our study provided a specific cutoff value for LAD (42.5 mm) to predict the decrease in LAAFV, which is helpful for clinicians to evaluate of stroke risk in patients with NVAF.
The CHA2DS2-VASc score has been widely used to predict the risk of ischemic stroke in patients with AF. Recent guidelines recommend anticoagulant therapy in high-risk patients with a CHA2DS2-VASc score of ≥ 2 [6, 7]. The relationship between stroke/thrombus formation and LAAFV has been investigated in many studies, and an LAAFV of ≤ 0.4 m/s represents a risk of stroke/thrombus [13]. However, the relationship between the CHA2DS2-VASc score and LAAFV is still unclear. In the present study, the CHA2DS2-VASc score (beta = − 0.134, P = 0.034) was significantly associated with LAAFV according to the multivariate linear regression analysis. Our findings are inconsistent with previous study showing that the CHA2DS2-VASc score was an independent predictor of a decrease in LAAFV [23], although we found a strongly negative association between the CHA2DS2-VASc score and LAAFV. Possible explanations include the variations in the recruitment criteria and the fact that the CHA2DS2-VASc score served as a categorical variable. Nevertheless, the ROC curve analysis demonstrated that the AUC was 0.689, with a sensitivity of 74% and a specificity of 56% when using the CHA2DS2-VASc score to predict the decrease in LAAFV in patients with NVAF. The predictive power of the CHA2DS2-VASc score was modest; thus, we further sought to develop a combined model that might better predict the decrease in LAAFV as a surrogate for cardioembolic risk in patients with NVAF. In the present study, LAD was an independent risk factor for the decrease in LAAFV. Combined use of LAD and CHA2DS2-VASc score significantly increased the ability of these two parameters to predict the decrease in LAAFV compared with LAD or CHA2DS2-VASc score alone. In fact, LAD has been shown to be an independent risk factor for stroke/thrombus formation in patients with NVAF [24–26]. Therefore, a combination of LAD and CHA2DS2-VASc score could be used as a substitute to predict the decrease in LAAFV in patients with NVAF.
We also showed that LAAFV is related to BNP concentration and persistent AF, which is in agreement with previous studies showing that BNP concentration is significantly inversely correlated with LAAFV in patients with AF [20]. Persistent AF can lead to LA structural remodeling and is probably associated with a decrease in LAAFV [27].