We thank Chen et al. for their thoughtful comments. Our study prospectively demonstrated, on concomitantly performed transthoracic echocardiography (TTE) and transesophageal echocardiography (TEE), the incremental predictive value of E/e′ ratio and e′ velocity in identifying left atrial appendage (LAA) thrombi in patients with nonvalvular atrial fibrillation (NVAF) undergoing electrophysiology procedures aimed at restoring sinus rhythm. We showed that E/e′ ratio can identify patients at low risk for LAA thrombus formation. Our results confirm the findings of retrospective studies and other reports showing associations between markers of elevated left ventricular filling pressure, such as B-type natriuretic peptide and left atrial volume, and LAA thrombus formation.

Chen et al. accurately point out that our report did not include international normalized ratio (INR) data and duration of anticoagulation treatment. In our study, 178 patients (67%) were on oral anticoagulants. Among patients treated with warfarin ( n = 128), the mean INR in subjects with versus those without LAA thrombus was 2.26 ± 0.55 versus 2.06 ± 0.81 ( P = .20). In 98% of the patients, INR was tested within 3 days before TEE (in 82% on the same day). Fifty patients (19%) were on novel oral anticoagulants. Unfortunately, we do not have data on the duration of anticoagulation therapy before TEE. It is clear, however, that the majority of patients targeted with oral anticoagulation therapy were adequately anticoagulated, which explains the low prevalence of LAA thrombus in our cohort (6.4%). The remainder of the patients (33%) were not anticoagulated, because they were not candidates (low stroke risk) or therapy was not initiated yet (first presentation with NVAF) or because of patient or physician choice. Notably, in 64% of warfarin-treated patients, warfarin was held and “bridged” with unfractionated or fractionated heparin before invasive procedures. Thus, even these seemingly favorable INR data may understate the usual anticoagulation state of these patients; hence, we chose not to report the aforementioned INR information in the original article. Importantly, the complexity and variability of the anticoagulation milieu in the study cohort did not affect the validity of our findings, simply because TTE and TEE were performed concomitantly, under the same anticoagulation state. This is an important strength of our methodology.

Second, Chen et al. point out that we came up short in analyzing the association between LAA thrombus and other diastolic function parameters, such as E-velocity deceleration time, pulmonary vein systolic-to-diastolic velocity ratio, and the ratio of peak E-wave velocity to flow propagation velocity (E/Vp). In our investigation, we chose, a priori, to investigate E/e′ ratio and e′ velocity on the basis of their ease of use, measurement fidelity, and widespread use in daily practice in most laboratories. This cannot be said about other diastolic parameters. Moreover, having examined several parameters may have exposed the study to “chance findings” (type I error) due to multiple testing. From a practical perspective, other diastolic function parameters were beyond the scope or funding of our study, which was limited to abbreviated TTE focused on specific two-dimensional and Doppler imaging. We agree, however, that evaluating the predictive value of other diastolic parameters is a fertile ground for future research.

Third, Chen et al. critique our study for a lack of longitudinal observational follow-up to confirm the impact of left ventricular diastolic function on long-term risk for LAA thrombus and thromboembolic events. Our study was designed to identify patients at low risk for LAA thrombus who may forgo TEE when undergoing electrophysiology procedures. To that end, identifying LAA thrombus before an electrophysiology procedure, not at a future time, is an appropriate end point. The use of diastolic function parameters to predict stroke in patients with NVAF is certainly intriguing. However, following patients longitudinally for LAA thrombus formation and embolic events, though worth investigating, was beyond the scope of our investigation. In fact, the study design, population, and sample size were not appropriate to address long-term hard outcomes. Given the small sample size, extended follow-up will be needed to observe a meaningful number of events; during such time, diastolic function parameters may change, and a significant number of patients are likely to die given the natural history of cardiomyopathy processes causing abnormal diastolic indices. Additionally, the profile of patients undergoing TEE before electrophysiology procedures is different from that of the NVAF population at large; thus making general inferences on stroke risk from a selective patient population may be inappropriate. Morphologic characteristics of the LAA, as assessed by TEE, are important determinants of embolization risk but have no significant clinical role in predicting LAA thrombus, because TEE used to assess LAA thrombus morphology can readily determine the presence of LAA thrombus.

Although our study suggests that patients with normal E/e′ ratios and e′ velocities may be considered to forgo routine TEE before electrophysiology procedures aimed at restoring sinus rhythm, such change in the standard of care would require validation in a large multicenter investigation. Whether the role of diastolic function parameters can be extended to improve stroke prediction in patients with NVAF is an intriguing question that needs to be addressed in a large-scale multicenter trial or registry. Therefore, we join Chen et al. in their call for well-designed longitudinal studies.