Acute reverse annular remodeling during MitraClip^® therapy predicts improved clinical outcome in heart failure patients: a 3D echocardiography study

In this study, symptomatic heart failure patients with FMR considered inoperable according to a previous heart team decision, New York Heart Association (NYHA) class II–IV, and an ejection fraction <35% were accepted for TMVR in accordance with current guidelines [12]. Patients with primary degenerative mitral regurgitation (DMR) were excluded from the study. In total, 48 heart failure patients were enrolled (Fig. 1). In three patients, the MitraClip^® procedure was unsuccessful. In two of those 3 patients, the posterior leaflet could not be gripped by the device, and in one patient a thrombus was detected in the left atrial appendage at the beginning of the procedure so that TMVR was discontinued. Subsequently, 45 patients were included for final data analysis. The study was approved by the local ethics committee and all patients gave written informed consent before study inclusion.

MitraClip^® implantation was performed as previously described [13]. The procedure was conducted under general anesthesia, and was performed by the same two experienced interventionists using fluoroscopy and one imaging cardiologist providing 2D and 3D TEE images.

Echocardiographic data were acquired during TMVR before and after MitraClip^® implantation using TEE. Follow-up after 6 months was performed using transthoracic echocardiography (TTE). Echocardiography was performed using a commercially available echocardiography system (iE 33, Philips Medical Systems, Andover, Massachusetts) with matrix array transducers (TEE: X7-2t; TTE: X5-1) capable of generating both two-dimensional (2D) and 3D images. Pre-procedural echocardiographic chamber quantification, left ventricular function assessment, and mitral regurgitation evaluation were performed according to current recommendations [14‐16]. We graded the severity of MR as grade 1 (mild), grade 2 (moderate), grade 3 (moderate to severe), and grade 4 (severe), corresponding to the EVEREST criteria for quantification [13]. Echocardiographic image acquisition and presentation to the interventionist during the procedure was also performed according to current guidelines [17, 18]. Assessments such as 6-min walk distance and the Minnesota Living With Heart Failure Questionnaire were performed before and after TMVR. Additionally, N-terminal pro b-type natriuretic peptide (NT pro-BNP) levels were measured before TMVR and at follow-up.

Anatomical measurements were performed offline using dedicated software (MVQ QLAB version 8.1 Software 2010 (Philips, Andover, MA, USA)). After choosing an image at end-systole from 3D full-volume datasets and optimization in terms of size, contrast, and enhancement, the image was cropped to obtain a perfect en-face surgical view of the mitral valve. The cropped 3D image was then aligned along the transversal, horizontal, and sagittal planes by means of multiplanar reconstruction. The aligned image was used to mark reference points, such as the anterolateral, posteromedial, anterior, and posterior direction of the mitral valve annulus, and other anatomical landmarks, e.g., the aorta and the nadir of the mitral valve leaflets. Afterwards, semi-automated reconstruction was started, generating a virtual 3D model of the mitral valve apparatus. Adjustments of the mitral leaflet commissural points, the mitral leaflets, and the coaptation length were set. The coaptation points were set correctly with the help of the surgical view of the 3D image. The 3D model of the mitral valve now contained accurate measurements of the mitral valve geometry. All anatomical measurements were performed by the same well-trained, experienced investigator blinded to the numerical outcome of the measurements while adjusting.

Patients were allocated by outcome into two groups, the High and Low Responders. High Responders (HR) were defined as patients with a decrease in NYHA classification ≥1.5 at 6 months after TMVR. Patients with a change in NYHA classification <1.5 were defined as Low Responders (LR), as only marginal clinical benefits were evident. This allocation was based on a recent echocardiographic study with a similar population [10].

Normal distribution of continuous variables was examined using the D’Agostino-Pearson omnibus normality test. For normally distributed data (presented as mean ± SD), paired t tests were performed. For non-normally distributed data (presented as median and interquartile range), the Wilcoxon test was used. Comparisons of, e.g., mitral valve geometry before and after TMVR, were performed using t test or the Wilcoxon signed-rank test, depending on the distribution of data. Two-tailed p values <0.05 were considered to be significant; therefore, in case p value was equal or less than the chosen significance level, the null hypothesis had to be rejected. Categorical data were presented as frequencies and percentages. Graphics and statistical analysis were made using Excel for Mac 2011 (Version 14.1.0) and GraphPad Prism version 5.0b for MacOS X (GraphPad Software, San Diego, CA, USA).

In total, 45 patients (age 70 ± 11 years; 29 males) with FMR were successfully treated with TMVR, three of which received two clips. At baseline, 33 patients (73%) had severe (4+), 5 patients (11%) moderate to severe (3+), and 7 patients (16%) moderate (2+) MR. 34 patients (76%) had NYHA class III and higher. Baseline characteristics and interventional data are presented in Table 1. Atrial fibrillation was more common in the Low Responder group (84% vs. 45%; p < 0.005).

In patients who underwent TMVR, overall, MR was reduced significantly (p < 0.01, Fig. 2). After the intervention, no patient had a significant increase of mean pressure gradient (i.e., increase of mean pressure gradient ≥5 mmHg) or pericardial effusion. 2D values such as vena contracta (VC) and systolic pulmonary artery pressure (PAsys) were examined. PAsys improved significantly before and after TMVR in the HR group compared to the LR group (meanΔ −6 ± 10 mmHg; p < 0.03 vs. meanΔ −1 ± 20; p = 0.8). Changes of echocardiographic parameters regarding mitral regurgitation were not significantly different between both groups. Furthermore, there were no significant changes between the groups and at follow-up for atrial and ventricular dimensions, left ventricular (LV), and right ventricular function. All echocardiographic parameters are summarized in Table 2.

NYHA class improvement after TMVR occurred in both groups, but was significantly higher in the HR group as compared to the LR group (HR: 3.5 ± 0.5 vs. 1.6 ± 0.5; p < 0.001; LR: 3 ± 0.6 vs. 2.5 ± 0.7; p < 0.01). Consistent with these results, the results of the Minnesota Living with Heart Failure Questionnaire (55 ± 10 points vs. 34 ± 14 points; p < 0.01) increased within the HR group with a similar trend in the 6-min walk distance (290 ± 104 m vs. 362 ± 110 m; p = 0.2), whereas these parameters barely changed in the LR group (6-min walk distance: 364 ± 94 points vs. 372 ± 70 points; p = 0.9; MLHFQ: 36 ± 17 vs. 32 ± 21; p = 0.7). NT pro-BNP decreased in the HR group (2837 ± 1642 pg/mL vs. 1705 ± 1080 pg/mL; p = 0.06). Data for clinical response of both groups after 6 months are presented in Fig. 3.

This study focused on the analysis of 3D values using QLAB. MVQ results before and immediately after TMVR for both groups are presented in Table 3. Results obtained from 3D data immediately before and after clip showed that anterior-to-posterior diameter (DAP) (meanΔ −4 ± 3.5; p < 0.01), 3D circumference (C3D) (meanΔ −11.5 ± 8.7; p < 0.01), and 2D circumference (C2D) (meanΔ −11.8 ± 10; p < 0.01) significantly decreased in HRs as shown in Table 3. The Low Responder group did not show any significant changes in C3D (p = 0.07) or DAP (p = 0.1). We did not find any significant changes in anterolateral to posteromedial diameters (DAlPm) in either group (p = 0.07 in High Responder group and p = 0.09 in Low Responder group).

The diversification of mitral regurgitation based on (patho-) anatomical and (patho-) physiological aspects in degenerative (DMR) and functional mitral regurgitation (FMR) has tremendous implications for transcatheter interventions [19, 20]. In the EVEREST II cohort, 73% of all patients suffered from structural heart disease with DMR and less than 27% were heart failure patients with FMR [21]. Successful MitraClip^® implantation in a high-risk cohort consisting of heart failure patients with FMR [22] was first described in 2010. It could be demonstrated that in patients with FMR, not in those with DMR, MitraClip^® therapy leads to significant reduction in mitral valve orifice area during the procedure. There is only one other study demonstrating the correlation between changes in mitral valve anatomy during the procedure and clinical outcome [10]. In that study, an acute reduction of the 3D annulus area and the 3D sphericity index in 107 patients with FMR could be demonstrated. The acute reduction of the 2D DAP was associated with improved clinical response after 6 months. Our results are in accordance with the results of that study regarding the fact that changes in MV geometry occur only in patients with FMR. In addition to that, we could demonstrate significant changes in original 3D data (C3D). The new aspect we gained from our data is that reverse remodeling of the mitral annulus in heart failure patients with FMR only occurred in patients with great clinical improvement at 6-month follow-up, indicating a stronger influence on clinical outcome. Figure 4 impressively indicates an example of the changes of the mitral annulus in the High Responder group, while mitral regurgitation severity is equally reduced in both groups (also compare Fig. 2). A different study with a similar patient population showed the acute effects of TMVR on the mitral valve leaflets demonstrating that through clipping the anterior and posterior leaflet, leaflet stress is not increased [23]. However, the study could not show significantly reduced 3D parameters as shown in this study as well as previous ones [9, 10].

Under the premise that greater reduction in 3D mitral annular geometry yields greater clinical benefits for the patients, the goal of the intervention may shift from mere MR reduction to focusing the procedural goal to reduction of mitral annular geometry using 3D TEE.

As recommended in current guidelines, mitral valve disease is best characterized using 3D TEE [16, 24]. The complex relation between annulus, the saddle shaped ring, and the leaflets can only be described adequately using this technique, especially when it comes to the preparation for MitraClip^® interventions [25]. In our study, we used dedicated software for the constitution of a 3D model of the mitral valve apparatus that can very precisely detect minimal changes in mitral ring plasticity during the procedure (Fig. 4). Although these analyses were performed offline, it is possible to perform them during the procedure in order to give the interventionist additional information on the current geometry of the mitral annulus under influence of one, two, or even more clips, irrespective of the interventional result concerning the color flow mapping of regurgitation severity. Our results underline the importance of 3D TEE for diagnosis, preparation, and monitoring the mitral valve apparatus during MitraClip^® procedures as only this technique is capable to adequately describe this complex structure, as already described [26]. In our Qlab results (cf. Table 3), the 2D circumference of the mitral annulus was reduced in both groups, while the 3D circumference was only reduced in the High Responder group. These results demonstrate that only the combination of the reduction of the circumference in the dimensions is associated with high response to the MitraClip^® procedure. In a recent study, it was demonstrated that 3D changes of the mitral annulus only occur in patients with FMR immediately after Clip implantation. The team used a comparable software as has been used in this study in order to demonstrate a reduction in 3D annulus area, anterior-to-posterior diameter, and 3D tenting area [9]. In contrast to our study, they could not demonstrate a clinical benefit of these changes for patients with FMR. Most importantly, they could not demonstrate any difference in original 3D measurements of mitral annular circumference.

Our results demonstrate that the additional effect of TMVR leading to reverse remodeling of the mitral annulus leads to improved clinical outcome than reduction of MR severity alone. Besides the impressive improvement in New York Heart Association functional class, initial NYHA class was higher in patients with high response to the MitraClip^® procedure. This underlines our hypothesis that subjects with FMR and heart failure who suffer from severe NYHA class improve after successful mitral annulus remodeling. In contrast to other studies, we could not document any other changes of the cardiac chambers or left ventricular function that correlated with an improved clinical response to TMVR. Recently, it could be demonstrated that changes of left ventricular volumes are associated with improved ejection fraction in heart failure patients with FMR. These changes were associated with a decrease in NT pro-BNP and an increase in the 6-min walk distance [7]. In our population, no significant changes of heart chamber dimensions or ejection fraction could be detected. This might me due to the fact that we included patients with very low ejection fractions compared to other studies. Although the remodeling process seems to start with changes in mitral annulus area directly during the procedure as demonstrated by our results, changes in LV dimensions and LV function therefore might occur later in our population. In both groups of this study, left atrial area showed to be mild to moderately dilated and was not affected by the procedure. Previous studies have shown that dimensional changes in left atrial area do not occur as early as at 6 months after the procedure; however, left atrial global contractile function improves as early as at 3 months after the procedure [27]. Therefore, continuous investigation over a longer period of time (>6 months) may help analyze the left atrial area more precisely and help us understand this process. A different group could demonstrate that the amount of left heart chamber remodeling including the left atrium and the left ventricle depended on the magnitude of MR reduction [8]. We cannot reproduce these results, most likely due to the fact that we only included heart failure patients with FMR, while that group included both—patients with DMR and FMR with better LV function. Also, there were significantly more patients with negative left atrial remodeling and atrial fibrillation in the Low Responder group at baseline. The improved outcome of our High Responder group might partially be due to the fact that there were fewer cases of atrial fibrillation, and the remodeling of the mitral annulus in these patients might have a maintaining effect upon a regular sinus rhythm for improved life quality. This is underlined by the fact that in our High Responder group we had a significant reduction in BNP after 6 months, as already demonstrated by others [7]. In another study addressing this issue, an additional benefit of TMVR on right ventricular function and pulmonary artery pressure in patients with FMR [28] could be shown. We could also demonstrate an effect upon pulmonary artery pressure only in the High Responder group, but there were no changes in right ventricular dimensions and function in our study collective. This also might be due to the fact that these changes might occur later in our population consisting of end-stage heart failure patients.