Relation of Mitral Annulus and Left Atrial Dysfunction to the Severity of Functional Mitral Regurgitation in Patients with Dilated Cardiomyopathy

Methods 56 patients (58 ± 17 years, 42 men) with DCM and FMR and 52 controls, prospectively enrolled, underwent 3DTTE dedicated for mitral valve (MV), LA, and left ventricle (LV) quantitative analysis. Results Patients with FMR vs. controls presented increased MA size and sphericity during the entire systole, whereas MA fractional area change (MAFAC) and MA displacement were decreased (15 ± 5 vs. 28 ± 5%; and 5 ± 3 vs. 10 ± 2 mm, p < 0.001). In patients with moderate/severe FMR, MA diameters correlated with PISA radius, EROA, and regurgitant volume (Rvol), as also did the MA area (with PISA radius, EROA, and Rvol: r = 0.48, r = 0.58, and r = 0.47, p < 0.05). MAFAC correlated inversely with EROA and Rvol (r = −0.32 and r = −0.35, p < 0.05), with both active and total LA emptying fractions and with LV ejection fraction as well. In a stepwise multivariate regression model, decreased MAFAC and increased LA volume independently predicted patients with severe FMR. Conclusions Patients with DCM and FMR have MA geometry remodeling and contractile dysfunction, correlated with the severity of FMR. MA contractile dysfunction correlated with both LA and left LV pumps dysfunctions and predicted patients with severe FMR. Our results provide new insights that might help with better selection of patients for MV transcatheter procedures.


Introduction
Functional mitral regurgitation (FMR) is a common complication in patients with ischemic or nonischemic dilated cardiomyopathy (DCM), increasing mortality, and the risk of rehospitalization [1]. In patients with FMR, surgical mitral valve (MV) treatment showed suboptimal outcomes in terms of morbidity and mortality [2,3], while interventional treatment provided controversial results [4,5], suggesting that selection of patients is crucial for treatment effectiveness.
us, in order to select better the patients for the transcatheter mitral valve procedures, assessment of MV geometry should be mandatory [6].
Recent studies suggested an association between the anteroposterior diameter of the mitral annulus (MA) and FMR recurrence after MV interventional treatment [7]. However, the MA is a complex tridimensional structure, which cannot be characterized by a single annular diameter [8,9]. erefore, understanding the geometrical and functional changes occurring in the MA, as well as their relations with the LA and LV dilation and function, is essential in patients with DCM and FMR [10]. is is now possible with the progress of three-dimensional (3D) full-volume acquisition with high temporal and spatial resolution using transthoracic echocardiography (3DTTE), coupled with the development of specific software packages, which enable a detailed quantitative analysis of MA [11], LV, and LA geometry and functions.
Accordingly, we designed our study (i) to characterize MA geometry and dynamics changes, as the remodeling of the MA, by using 3DTTE in patients with DCM and FMR, in comparison with healthy controls; (ii) to assess the relationship between MA remodeling and severity of FMR; and (iii) to assess the relationship between MA dysfunction and severity of LV and LA dilation and dysfunction.

Study Population.
We conducted a prospective study, in which we enrolled 56 patients with mild-to-severe FMR due to ischemic and nonischemic DCM and compared them with 52 control subjects, similar in age, gender, and BSA. All subjects were in sinus rhythm and had complete 3DTTE datasets for MV, LA, and LV. Patients with more than mild aortic, pulmonic, or tricuspid valve disease, significant MV calcifications, or poor-quality image of the 3DE datasets were excluded. Control subjects were healthy volunteers free of any cardiovascular history or risk factors. Heart rate, blood pressure, height, and weight were measured in all subjects immediately before the echocardiographic examination. Body surface area (BSA) was calculated [12]. Etiology of DCM, NYHA class, and the presence of wide QRS (>120 msec) on ECG were recorded. e Local Ethics Committee approved the study, and all subjects provided an informed consent.

Echocardiography.
Echocardiography was performed in accordance with a standardized acquisition protocol, with a Vivid E9 system (GE Vingmed Ultrasound, Horten, Norway), equipped with M5S and 4V probes. Standard transthoracic echocardiography was performed to assess LV size and function and etiology and severity of FMR in patients with DCM. en, to obtain high temporal resolution of the datasets, separate 3D full-volume multibeat acquisitions of the MV, LV, and LA were obtained by combining six consecutive ECG-triggered subvolumes, during breath-hold and avoidance of patient or probe movement. Care was taken to encompass the MV in the full volume throughout the cardiac cycle and the entire LV and LA in the dedicated dataset. By using multislice display, the absence of stitching artifacts in the acquired dataset was carefully checked.
LA 3DTTE datasets were converted to DICOM format and analyzed using dedicated software package designed for volumetric analysis of the LA and recently validated against CMR (LA analysis 2.3, TOMTEC Imaging Systems, Unterschleissheim, Germany) [16]. At the beginning of the 3D analysis workflow, LA datasets were automatically sliced in 4-chamber, 2-chamber, and long-axis apical planes and a short-axis plane. Translating and rotating 4-chamber plane to obtain orthogonal nonforeshortened planes of the LA in all 3 apical views allowed rapid manual data set alignment. In each apical view, LA blood-tissue interface was manually initialized on 2 frames, identifying maximum and minimum LA volumes (LAVmax and LAVmin). A 3D surface of LA volume was then generated for each frame throughout the cardiac cycle resulting in a dynamic cast of the LA cavity. For each consecutive frame, the voxel count inside the 3D LA surface was used to measure LA volume, resulting in a smooth interpolated time-volume curve from which LAVmax, LAVmin, and preatrial contraction LA volume (LAVpreA) were obtained [17]. From LA volumes, the software automatically calculated total, passive, and active LA emptying fractions (LAEFs) [16] (Figure 2). MV 3DTTE datasets were converted to DICOM format and analyzed using dedicated software for MV analysis (TOMTEC, 4D-MV assessment 2.3, Unterschleissheim, Germany), as previously described [18]. Briefly, MA measurements were performed in early-systole-the frame after the MV closure (MVC); end-systole-the frame just before MV starts to open; and in mid-systole-the frame midway between MVC and end-systole. After adding anatomical landmarks, the software created a 3D model of the MA and MV leaflets, first at mid-systole (static analysis) and then during the cardiac systole (dynamic analysis) ( Figure 3). MA measured parameters were 3D and 2D (projected) areas and circumference; anteroposterior (AP) and anterolateralposteromedial (ALPM) diameters; sphericity index; height and nonplanarity angle to quantify its "saddle shape"; and Ao-AP angle, as the angle between the aorta and MA plane. In addition, MV commissural diameter, anterior leaflet area and length (ALA and ALL), posterior leaflet area and length (PLA and PLL), tenting height, area, and volume were also measured. For all parameters, values at MVC, mid-systole, and end-systole, their minimal value, and the time interval from MVC to moment of their minimal value were recorded and expressed as percentage (%) of the total duration of the systole [19]. MA diameters, area, and circumference were normalized to BSA. e software package also provided MA displacement, displacement velocity, and MA area fractional change during systole. e fractional change (difference between the maximal and minimal value divided by the maximal value and expressed as percentage) of the MA circumference, AP diameter, and ALPM diameter were also calculated [19].

Statistical Analysis.
Normal distribution of variables was checked by Kolmogorov-Smirnov test. Continuous variables were summarized as mean ± SD, and categorical variables were reported as percentage. Variables were compared between groups using unpaired t-test. Pearson's correlation was used to analyze the relationships between MR severity parameters and MA function and between MA function and LA and LV parameters. Multiple linear backward regression tests were used to identify the parameters that independently correlated with severe MR. All analyses were performed using SPSS version 20.0 (SPSS,     has previously reported the reproducibility for the echocardiographic methods, in unselected patients with MV disease [18].

General Characteristics.
e general characteristics of the study population are summarized in Table 1. Age, gender, blood pressure, and BSA were similar between groups. As expected, patients with FMR had larger LV volumes and lower LVEF than controls. Patients with DCM had an ischemic etiology in 40 cases (71%); they were in NYHA functional class I in 5 cases (9%), II in 26 cases (46%), and III in 25 cases (45%). 17 patients (30%) had mild, 25 (45%) moderate, and 14 (25%) severe FMR. PISArad was 6±3 mm, available in 42 patients (75%), while EROA was 0.17 ± 0.12 cm 2 and RVol 26 ± 19 ml, available in 33 patients (59%). Temporal resolution of the 3D datasets for MA analysis was 32 ± 7 volume/sec in patients with FMR and 35 ± 6 volume/sec in control subjects (p � 0.87). e number of systolic frames in each dataset ranged from 9 to 17, depending on heart rate or acquisition settings. Static analysis of the MA was feasible in all subjects. Dynamic analysis of MA was feasible in 54 patients with FMR (96%) and in 48 control subjects (92%). Table 2. Patients with FMR, when compared with controls, had larger MA dimensions. ey had 35% larger MA area, by 3DTTE. Moreover, they had larger ALA and PLA, but with a different extent of MV leaflet enlargement vs. controls, resulting in a lower ALA/PLA ratio. Meanwhile, patients with FMR vs. controls had increased MA sphericity and nonplanarity angle, but similar MA height. ey had increased MV tenting volume, area, and height, while Ao-AP angle was wider. Mitral annulus dynamic analysis is presented in Table 3, while dynamic graphic of MA dimensions and geometry is displayed in Figure 4. Patients with FMR, when compared with controls, had larger MA dimensions at every reference frame of the cardiac systole. ey had increased MA sphericity index, due to unequal increase in MV leaflets areas, and increased MV tenting volume, area, and height, throughout the entire systole. Meanwhile, they had decreased MA height and wider MA nonplanarity angle and MA Ao-AP angle than controls. Patients with FMR presented prolonged times to occurrence of minimum AP, ALPM, and commissural diameters and to occurrence of minimal MA area and circumference. ey also had prolonged times to occurrence of minimal MA sphericity, height, and minimal Ao-AP angle. Patients with FMR presented decreased fractional changes of the MA parameters than controls (Table 4).

MA Remodeling and Dysfunction and Severity of FMR.
Comparisons between LV, LA, and MA parameters in patients with trivial/mild, moderate, and severe FMR are shown in Table 5. Patients with moderate FMR had increased indexed LVEDV, lower LVEF, but similar LAVmax, than patients with trivial/mild FMR; meanwhile, they had increased MA dimensions and decreased MA area fractional change and MA displacement. Patients with severe MR had similar volumes and LVEF compared with patients with moderate MR, but increased LAVmax and MA dimensions, and decreased MA area fractional change.

Independent Predictors for Severe MR.
In a multiple backward linear regression model using indexed LVEDV; LAVmax; MA area; LVEF; active and total LAEFs; AP diameter fractional change; and MA fractional area change, independent predictors for severe MR were the LAVmax (p � 0.03), AP diameter fractional change (p � 0.02), and MA fractional area change (p � 0.003), respectively.

MA Dysfunction and the Severity of LV and LA
Remodeling. MA area showed similar correlations with LVEDV and LAVmax (r � 0.55 and r � 0.54, both p < 0.001). MA area fractional change correlated with both active and total LAEFs, as well as with LVEF, while MA AP diameter fractional change showed similar behavior; MA displacement correlated only with LVEF ( Figure 5).

Discussion
Our study showed that in patients with FMR there is geometric and functional remodeling of the MV, that can be summarized as follows: (i) increased MA size and sphericity, during the entire cardiac systole, was associated with a concomitant remodeling of the MV leaflets; (ii) MA was flatter, with increased MV tenting height, area, and volume during the cardiac systole; (iii) maximum MA contraction was not only decreased in amplitude, but also delayed in time; (iv) MA size and contractile dysfunction correlated with FMR severity; and (v) MA contractile dysfunction correlated with both LA and LV dysfunctions.

Mitral Annulus Geometry and Dynamics in Patients with
Functional Mitral Regurgitation. MA remodeling and dysfunction have been reported as mechanisms of both ischemic and nonischemic FMR [10,20,21]. In ischemic MR, MA dysfunction was related to local LV remodeling [22] and severity of regurgitation [20]. MA remodeling and dysfunction are more prominent in patients with LV dysfunction and FMR than in those without FMR [10], despite the fact that isolated MA dilation failed to explain the occurrence of MR in patients without dilated LV [23]. Meanwhile, LA remodeling was suggested to have a role in ischemic MR through an atriogenic mechanism that determines MV leaflet tethering [24]. However, the relationship between MA dysfunction and LV and LA remodeling and its relation to MR severity has been explored insufficiently in patients with FMR.
Patients with mild-to-severe FMR showed MA dilation throughout the entire systole, as previously reported [10,21]. However, our data are in contrast with those reported by Topilsky et al. [10], since our patients with FMR presented an increase of both AP and ALPM diameters of the MA and not of the AP diameter only. Still, an unequal increase in AP and ALMP size was noted in our FMR patients, which lead to a more spherical MA than in control subjects. In patients with moderate-to-severe FMR, the increase in AP and ALPM diameters was correlated with FMR severity. On the other hand, our data, showing the direct relation between MA size and shape and FMR severity, confirm and expand the data reported by Topilsky et al. [10], who showed that patients with LV dysfunction and FMR have larger AP size of the MA than patients with LV dysfunction without FMR. Moreover, although it is widely accepted that FMR is mainly due to regional and global remodeling of the LV, while the MV apparatus is normal, a study by Debonnaire et al. [25] contradicted the paradigm of the "normal mitral valve" and suggested a true remodeling of the entire MV leaflets occurring in patients with FMR. Our study showed the increase of both anterior and posterior MV leaflets areas, from   the MA to the point of their coaptation, in patients with FMR. e software package that we used for the study did not measure the entire leaflet area, as in the study by Debonnaire et al., but "an effective" leaflet area, from the base to the coaptation point of the MV. However, it seems that the increase in this "effective leaflet's size" is not symmetric, since the ratio between ALA and PLA became almost unitary in our patients with FMR. In a recent paper published by Stolfo et al. [7], the increase in the AP diameter of the MA was associated with a higher recurrence of MR after percutaneous repair of functional MR, in patients with advanced heart failure. Our data suggest that further studies should be performed, in order to better select patients for percutaneous repair of the MV, based on MA parameters, as well. MA was flatter in patients with FMR than in control subjects. Similar with the data reported by Topilsky et al. [10], our patients with FMR showed higher MV tenting volume, area, and height throughout the systole, but only MV tenting volume showed significant correlations with MR severity. Previous reports have shown that MV tenting volume was a significant predictor of MR severity [26] and of recurrence of MR after MV annuloplasty.
Beside MA enlargement and remodeling, patients with FMR also showed altered relation of the MA with the surrounding structures, presenting wider Ao-AP angle than control subjects. is has to be taken into account for annuloplasty techniques, as the angle Ao-AP proved to predict the occurrence of systolic anterior movement of the anterior mitral leaflet after MV annuloplasty [27]. Moreover, LV outflow obstruction is a potential complication of the new transcatheter mitral valve replacement procedure (MitraClip) that can be predicted during procedure planning since it is related not only to prosthesis design, but also to anatomy (septal thickness, LV size, Ao-AP angle, and anterior septal length) [28]. LV dyssynchrony has been reported as a mechanism of FMR, while cardiac resynchronization therapy (CRT) can reduce early systolic FMR in selected patients [29][30][31], but the detailed mechanism remains to be clarified. Delay in contraction of the MA, showed by us, might have a role in the early MV regurgitation during cardiac systole.
us, future studies might address how CRT affects the timing of MA contraction.

Relation between MA Geometry and Function and LA and LV Size and Function in Patients with FMR.
MA area fractional change was increasingly impaired in patients with FMR. We showed that the "contractile dysfunction" of the MA was associated with both LA and LV dysfunctions. In patients with DCM, Kwan et al. [32] reported that the extent of the decreased contraction of MA area and AP diameter was related to the extent of LV  dysfunction, in terms of LVEF. In addition, previous experimental models showed the relation between MA contraction and LA function [33], and this has been reported in normal individuals too, using 2D echocardiography [34]. Several groups [35][36][37] suggested that, in patients with atrial fibrillation, LA remodeling may play a role in the pathogenesis of FMR through a new mechanism, the "atriogenic leaflet tethering." Our results showed that in patients with dilated and dysfunctional LV, the dilation and dysfunction of the LA have an important role in the contractile dysfunction of the MA, beyond the role played by the LV dysfunction, and it is closely related to the severity of FMR. MA dysfunction, associated with LA remodeling, might be an additional "atriogenic" mechanism determining the worsening FMR from moderate to severe in patients with DCM, who had similar LV volumes and LVEF.

Predictors for Severe MR.
Otsuji et al. [23] showed that MA dilation fails to solely explain MR severity in patients with atrial fibrillation vs. patients with DCM and FMR. Our study confirms that neither the MA enlargement nor the LV dilation is a predictor for MR severity. Conversely, from all parameters measured for the LV, LA, and MA size and functions, only the LA maximal volume and the contraction of AP diameter and of MA area are able to predict severe MR. erefore, LA dilation associated with a decreased "contraction" of the MA might be an additional "atriogenic" mechanism that increases MR severity in patients with DCM, independently from the remodeling of the LV.

Limits of the Study.
In our study, we did not assess the complex relationship between papillary muscle position, orientation, and FMR severity; however, this has been extensively assessed in previous studies [38][39][40]. e software package, used to assess MA dynamics, performs MA measurements only during the systolic phase, and not during the entire cardiac cycle. However, it has been shown that, even in normal subjects, the most important conformational changes of MA occur during the systolic phase of the cardiac cycle and less in the diastolic phase [41]. Moreover, we aimed

Conclusions
Patients with dilated cardiomyopathy and functional mitral regurgitation have mitral annulus geometry remodeling and contractile dysfunction, which correlated with the severity of mitral regurgitation. Mitral annular contractile dysfunction correlated with both left atrial and left ventricular pump dysfunctions. us, these results enhance the association of atrial remodeling and dysfunction with the severity of functional mitral regurgitation severity, mediated through the mitral annular dysfunction. Our results may ensure further studies to use parameters of the mitral annular size and geometry, in order to help us to better select patients for transcatheter procedures, targeting the mitral annulus or the mitral valve.

Data Availability
All study information is available upon request from the corresponding author. For this purpose, the corresponding and first author of the paper may be contacted via the e-mail sorinamihaila1981@gmail.com.

Additional Points
Functional mitral regurgitation is a common and morbid complication in patients with ischemic or nonischemic dilated cardiomyopathy, and its management remains challenging. It is known that the mitral valve apparatus presents altered size, geometry, and dynamics in patients with functional mitral regurgitation, which is essential information in order to design new devices and to properly select patients to treat this disease. In patients with functional mitral regurgitation, the mitral annulus has increased size and sphericity and decreased height during the entire cardiac systole, coupled with a significant remodeling of the mitral valve leaflets. By using three-dimensional echocardiography, our study showed that the mitral annulus from patients with functional mitral regurgitation is larger, more spherical, and flatter during the entire cardiac systole. Moreover, the contraction of the mitral annulus is decreased and delayed in time and positively related to the functional mitral regurgitation severity. Finally, the contractile dysfunction of the mitral annulus related not only to the left ventricular systolic dysfunction but also to the left atrial pump dysfunction.

Ethical Approval
All procedures performed in studies involving human participants were in compliance with the ethical standards of the institutional and/or national research committee and with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards.

Consent
An informed consent was obtained from all individual participants included in the study.

Conflicts of Interest
e authors declare that they have no conflicts of interest regarding the publication of this paper.