The data used in this article will be shared upon reasonable request to the corresponding authors.

Our study represents a retrospective analysis of the Cedars-Sinai database of consecutive TEER procedures performed between January 1, 2013 and January 1, 2021 in adult patients for moderate-to-severe or greater MR accompanied by myocardial dysfunction and/or symptoms despite maximally tolerated medical therapy. Each intervention was undertaken following a Heart Team discussion, which considered overall patient status as judged clinically, standardized/formal surgical risk and operability, published scientific evidence, and patient preferences.

The inclusion criteria for the study were as follows: a) a diagnosis of chronic PMR, based on a morphologically abnormal valve apparatus as assessed during the intraprocedural TEE; b) the performance of an isolated, first-ever TEER; and c) the availability of a viewable preprocedural TTE.

The primary outcome was the composite of all-cause mortality or heart failure (HF) hospitalization during the first postprocedural year. Secondary endpoints included individual components of the primary outcome, as well as the persistence of significant functional impairment at 1 year, indicated by a New York Heart Association (NYHA) class III-IV, or above-moderate MR. A greater-than-mild MR at 1 month was also examined.

The study conformed to the Declaration of Helsinki and was approved by the Cedars-Sinai Institutional Review Board, which waived the need for informed consent.

MitraClip (Abbott Vascular Inc, United States) was the sole system employed in the registry. All procedures were performed under general anesthesia and used a trans-septal approach and femoral venous access. TEE, fluoroscopy, and right heart catheterization were used for guidance and monitoring. Technical success was defined as actual device deployment not accompanied by surgical intervention or major complications within the first 24hours.20

Echocardiograms were performed and interpreted by experienced sonographers and level III-trained echocardiologists, following accepted guidelines.21–23 The ultrasound system used was EPIQ (Philips, United States). Postprocessing used PICOM365 (SciImage, United States), QLAB 12.0 (Philips, United States), and TomTec Arena (TomTec Imaging Systems, Germany) for 2-dimensional (2D), 3-dimensional (3D), and speckle-tracking measurements, respectively.

All parameters were evaluated by multiple, focused, and zoomed views. For each continuous structural variable, a body surface area-indexed value was calculated. Regarding hemodynamic variables, either the highest or averaged values were considered based on rhythm regularity. To ensure reliability and consistency, all baseline continuous parameters were assessed by 2 study members (A. Shechter and M. Lee), who were blinded to patient history. In addition, selected mitral and left ventricular (LV) parameters were compared with those obtained by intraprocedural TEE and preprocedural cardiac computed tomography (CCT) exams, respectively. The latter were performed in patients simultaneously considered for valve replacement.

Mitral valve (MV)-related echocardiographic parameters included regurgitation severity, transmitral mean pressure gradient (TMPG), peak E wave velocity, presence and extent of mitral annular calcification (MAC), leaflet calcification, mitral annular diameter (MAD), and leaflet tethering/restriction. MR severity was evaluated by integration of qualitative and quantitative measures and graded as 0 (up-to-minimal), 1 (mild/mild-to-moderate), 2 (moderate), 3 (moderate-to-severe), or 4 (severe). The peak E wave velocity and TMPG were inferred from pulsed-wave (PW) or continuous-wave (CW) mitral inflow tracings, respectively. MAC was assessed semiqualitatively and described as above-mild when involving more the one-third of the annular circumference on the parasternal short axis view24 or protruding into the LV on apical views. Leaflet immobility was quantified based on leaflet closing angles on the parasternal long axis (PLAX) view. Anterior-posterior (AP) and medial-lateral (M-L) MAD lengths were measured at mid- and end-diastole in the PLAX and apical 4-chamber views, respectively.

Non-MV-related variables consisted of chamber function and dimensions, concomitant valvulopathies, pulmonary arterial systolic pressure (PASP), and LV global longitudinal strain (LVGLS). Left ventricular ejection fraction (LVEF) and left heart chamber volumes were calculated using the Simpson biplane method of disks, whereas global right ventricular (RV) function was assessed qualitatively. LV mass index was computed using the American Society of Echocardiography formula. Tricuspid annular plane systolic excursion (TAPSE) corresponded to the vertical displacement of the lateral tricuspid annular edge according to M-mode tracing in the apical 4-chamber view. Tricuspid regurgitation (TR) was quantified in a similar fashion to MR. PASP was calculated by combining the maximal CW-derived TR pressure gradient with the estimated right atrial pressure; the latter was dictated by inferior vena cava diameter and collapsibility as revealed in the subcostal views. LVGLS was calculated semiautomatically following manual adjustments of cardiac cycle and tracing borders as needed, by averaging endocardial strain measurements in the apical windows.

Intraprocedural pulmonary venous flow pattern (PVFP) improvement and normalization required any increase or the emergence of a value of ≥ 1, respectively, in the peak systolic/diastolic velocity ratio on any PV by PW interrogation.

Patient assessment was carried out at baseline, hospital discharge, and at 1 month and 1 year postprocedure. Data were extracted from an electronic medical chart, which was updated in real-time by medical providers and state authorities.

Variables are reported as frequencies and percentages or medians [interquartile ranges]. Selected continuous variables were assessed for correlation and change over time using the Pearson r coefficient and Wilcoxon test, respectively. Interobserver reliability regarding continuous TTE parameters was evaluated by the intraclass correlation coefficient (ICC).

To identify associations with outcomes, Cox and binary logistic regression multivariable analyses were performed that incorporated baseline TTE parameters with perceived or previously proven3,17 prognostic significance and a P value of <.1 in univariable models. Continuous variables were assessed both as such and as dichotomous, using the medians and 1st/4th quartiles of the cohort, as well as guideline-cited thresholds for intervention.25,26 Both LVGLS-inclusive/exclusive models were constructed.

The cumulative incidence of the primary outcome and its separate components as a function of TTE parameters identified by the regression models was further analyzed by the log-rank test and was graphically displayed using the Kaplan-Meier method.

To address potential confounders encountered by the “echo-only” regression models, “comprehensive” models were constructed for the primary outcome and its elements. In addition to preprocedural TTE parameters, these included baseline clinical parameters and procedural features showing differing frequencies among patients with and without TTE findings associated with the risk of the primary outcome (all as determined by the Pearson chi-square, Fisher exact, or Mann-Whitney U tests), as well as the year of TEER performance, device generation, and data availability regarding 1-month MR grade.

Cases with missing values were censored from the relevant calculations. Statistical significance was defined as a 2-sided P value of <.05. All analyses were performed using SPSS 24 (IBM Corporation, United States).

A total of 410 patients were included in the analysis and followed up for a median of 494 [151-1075] days (figure 1). These patients were characterized by a median age of 83 [76-88] years, a predominance of male sex (n=249, 60.7%), and a high burden of comorbidities, mostly hypertension (table 1). HF was highly symptomatic, as shown by NYHA class III-IV in 377 (91.9%) patients. Reflecting this profile, interventional risk was medium-to-high.

Figure 1.

Study flow chart. IQR, interquartile range; MR, mitral regurgitation; NYHA, New York Heart Association; TEER, transcatheter edge-to-edge repair; TTE, transthoracic echocardiogram.

(0.42MB).

Baseline TTE, performed 25 [IQR, 8-54] days prior to the procedure, demonstrated severe MR in most of the patients (n=349, 85.5%) (table 2). Overall, MR was attributed to degenerative disease in 403 (98.3%) patients, annular/leaflet calcification in 6 (1.5%), and a combination of the 2 in 1 (0.2%). Almost half of the patients (n=195, 47.6%) had an LVEF of ≤ 60% or an LV end-systolic diameter (LVESD) of ≥ 4.0cm. A similar proportion exhibited a left atrial volume index (LAVi) of ≥ 60mL/m2, and a little more than a third had MAC and a PASP of> 50mmHg. The median LVGLS was −15.9 [−18.9-(−12.3)]%. Importantly, the interobserver reliability was good for all continuous parameters (ICC> 0.87, P <.001). In addition, mid-diastolic MAD correlated with its end-diastolic counterpart, TEE parallel, and TEE 3D MV area (Pearson r ≥ 0.72, P <.001). CCT, undertaken in 79 (19.3%) individuals within 0 [IQR, 0-8] days of the preprocedural TTE, revealed higher indexed LV volumes that nevertheless correlated with the echocardiographic observations (Pearson r ≥ 0.77, P <.001).

Most procedures employed 1 to 2 first/second-generation devices and targeted the A2P2 segment (table 1 of the supplementary data). Almost all (n=401, 97.8%) were completed successfully, allowing for next-day discharge in most patients.

Immediately following clip deployment, MR regressed to mild-or-less in 317 (77.3%) patients and PVFP improved in 300 (85.5%). Later in the first postprocedural year, both LVEF, left heart dimensions, and PASP significantly decreased (table 2 of the supplementary data).

By 1 year, 61 (14.9%) patients had experienced the primary outcome of all-cause mortality (n=35, 8.5%) or HF hospitalizations (n=37, 9.0%). Of the 375 patients who survived to the first postprocedural year, 217 (57.9%) remained under active surveillance at Cedars-Sinai and had available data relating to functional status or residual MR (functional status, n=204/375, 54.4%; residual MR, n=184/375, 49.1%). Among the latter, 40 (18.4%) were found to be in NYHA class III-IV (n=24/204, 11.8%) or to demonstrate above-moderate MR (n=19/184, 10.3%) at 1 year. One-month MR grade, which as a continuous variable directly correlated with its 1-year counterpart (Pearson r=0.75, P <.001), was above-mild in 115 (36.6%) of the 314 individuals with viewable echocardiograms, the latter comprising 77.5% (n=314/405) of patients remaining alive at 1 month.

After multivariable analysis, an LVESD index (LVESDi) of ≥ 2.1cm/m2, corresponding to the 4th quartile of the cohort, emerged as the only preprocedural TTE parameter that was associated with the risk of the primary outcome, more than doubling its risk (hazard ratio [HR], 2.44; 95% confidence interval [95%CI], 1.09-4.68; P=.022) (table 3 of the supplementary data and table 3). An LVESDi of ≥ 2.1cm/m2 also conferred a higher probability of all-cause mortality (HR, 2.17, 95%CI, 1.28-4.88; P=.020) and HF hospitalizations (HR, 3.27; 95%CI, 1.38-5.75; P=.007) in separate analyses (tables 4 and 5 of the supplementary data), and was associated with higher rates and cumulative incidences of all above-mentioned outcomes (figure 2, table 6 of the supplementary data, and figure 1 of the supplementary data). Of note, the univariate link between increased LVESD and the cumulative incidence of the primary outcome was observed regardless of baseline LVEF and LVESD intervention cutoffs (figures 2 and 3 of the supplementary data).

Figure 2.

All-cause mortality and heart failure hospitalization. Increased baseline LVESDi was associated with a higher 1-year cumulative incidence of the composite of all-cause mortality or heart failure hospitalization (A) and of its separate components (B,C). Increased mid-diastolic M-L MAD was associated with earlier heart failure hospitalization (D). LVESDi, left ventricular end-systolic diameter index; M-L, medial-lateral; MAD, mitral annular diameter; TEER, transcatheter edge-to-edge repair.

(0.49MB).

Apart from LVESD, an increased LAVi (as a continuous variable) at baseline was also associated with a higher risk of all-cause mortality (HR, 1.02; 95%CI, 1.01-1.04; P=.012), and a mid-diastolic M-L MAD of ≥ 32.2mm, - with a higher risk (HR, 4.29; 95%CI, 1.55-6.90; P=.005) and earlier occurrence of HF readmissions. An above-mild MAC, while leading only to a trend toward a higher HF hospitalization risk, was nevertheless linked to excess HF hospitalizations (n=9/53, 17.0% vs n=28/357, 7.8%, P=0.040), most of which (n=6/9) were attributed to noncardiac causes.

Regarding functional status and residual MR, an LAVi of ≥ 60mL/m2, an above-mild MAC, and an above-moderate TR prior to TEER were associated with a higher odds of NYHA class III-IV or above-moderate MR persistence at 1-year postprocedure, and patients exhibiting a higher number of such characteristics had higher rates of the combined outcome (table 7 of the supplementary data and figure 3). All TTE parameters associated with the 1-year endpoints, except LAVi and M-L MAD, as well as an indexed mid-diastolic M-L MAD of ≥ 20.2mm/m2 and a mitral effective regurgitant orifice area (EROA) of ≥ 0.40cm2, conferred a higher risk of greater-than-mild MR at 1 month (table 8 of the supplementary data). Of note, patients with above-mild MR at 1 month experienced a steeper intraprocedural rise in the TMPG (2 [IQR, 1-3] vs 1 [IQR, 0-2] mmHg, P=.028), were treated by fewer clips (1 [IQR, 0-1] vs 2 [IQR, 1-2]; P=.437), and were marginally more likely to exhibit above-mild MR immediately after clip deployment (n=12/24, 50% vs n=30/91, 33.0%, P=.123) if they had above-mild (compared with up-to-mild) MAC at baseline.

Figure 3.

Functional status and grade of mitral regurgitation. A higher baseline parameter burden was associated with increased rates of significant functional impairment or MR at 1 year. The parameters were left atrial volume index of ≥ 60mL/m2, above-mild mitral annular calcification, and greater-than-moderate tricuspid regurgitation. MR, mitral regurgitation; NYHA, New York Heart Association.

(0.12MB).

Lastly, exploratory regression models that integrated baseline clinical variables and procedural aspects whose frequencies differed between the low and high LVESDi groups, year of intervention, device generation, and data availability largely reproduced the results of the TTE-only based analyses (tables 9-15 of the supplementary data). Additionally, they suggested a possible association between an increased mid-diastolic M-L MAD and both the primary outcome and all-cause mortality.

First, the single-center, retrospective design of the study and its lack of central adjudication may hamper the generalizability of results. However, our sample was relatively large, resembled a recently published American nationwide registry,37 and was assessed by experienced echocardiologists blinded to patient files, all of which potentially improved validity. Second, the lack of follow-up data on functional status and MR grade, as well as the low absolute number of outcome events, negatively impacted statistical power, making some analyses, and particularly those that included non-TTE variables, exploratory. Nevertheless, data availability was comparable to that of previous real-world registries12,13 and was similarly distributed in the various subgroups, somewhat limiting the probability of bias. In addition, the lack of follow-up data did not affect the regression models, suggesting that it did not influence the results. Furthermore, the better-documented 1-month MR grade, which has previously been associated with longer-term outcomes38 and is currently used to determine procedural success,20 correlated well with its 1-year counterpart and was associated with most of the prognostically meaningful TTE observations, thus supporting the latter's significance. Third, our study focused on 2D TTE measurements while disregarding 3D parameters,39 as the latter could be analyzed only in a fraction of patients (n=80, 19.5%). Notwithstanding possible inaccuracies in measurement, reliability and consistency were acceptable, as indicated by good interobserver/modality agreement. Further reinforcing validity were the multivariable analyses, which are among the most comprehensive to date. Fourth, baseline medical therapy was somewhat suboptimal, precluding the extrapolation of our findings to medically optimized populations. However, this represented patient tolerance and was consistent with the real-world setting of the study.40 Last, our observations may be less applicable to patients treated with the newest-generation devices and delivery systems or to rheumatic MR cases, as these were either underrepresented or not included in the study.