3D vena contracta area after MitraClip© procedure: precise quantification of residual mitral regurgitation and identification of prognostic information

Study population

The PMVR registry of the University Hospital Regensburg, Germany, comprises 102 patients, who underwent the procedure between 04/2012 and 12/2015 and were screened for eligibility. Inclusion criteria for this study were a standardised six-minute walk test before and after PMVR as well as stored 3D–TEE colour Doppler datasets before and during PMVR. For this investigation, we excluded 73 patients (unavailable six-minute walk test, 48 subjects; unavailable 3D–TEE colour Doppler dataset, 17 subjects; insufficient quality of stored echocardiography for VCA determination, 8 subjects), yielding 29 cases for this analysis.

Clinical parameters

Further information concerning the patients’ health status was derived from medical records. EuroScoreII and logEuroScore were calculated [12]. A six-minute walk test (6MW) was recorded before PMVR and 4 weeks after the procedure. The test was performed according to the current statement of the American Thoracic Society [13] indoors, along a flat, straight, enclosed, seldom travelled corridor with a hard surface by a trained nurse. The percutaneous-repair procedure was performed under general anaesthesia with the MitraClip System (Abbott Vascular, Lake Bluff, USA) as previously described [1, 14].

Echocardiography

Two-dimensional transthoracic echocardiography (iE-33 ultrasound system with S5–1 transducer; Philips Medical Systems, Amsterdam, The Netherlands) was performed in all patients before and 4 weeks after PMVR. Left ventricular volumes and left ventricular ejection fraction were calculated by Simpson’s rule according to recent guidelines [15]. MR was quantified in an integrative view according to the Endovascular Valve Edge-to-Edge REpair STudy (EVEREST) criteria [16]. Information on valve morphology, colour flow doppler, presence or absence of systolic pulmonary vein flow, regurgitant volume and regurgitant fraction was gathered according to recent guidelines [2–4] and combined to grade MR on a scale from mild to severe (I to IV) [16] to assure comparability to previously published registries [1, 17–21]. MR immediately after PMVR was graded from I to IV according to the recommendations of the German Cardiac Society (DGK, Additional file 1).

All patients underwent TEE for screening (“before PMVR”) purpose and during the catheter intervention providing a dataset immediately after Clip release (“immediately after PMVR”). In a subgroup, a follow-up TEE was performed 4 weeks after PMVR (“Follow up”). All images were acquired using an iE-33 ultrasound system equipped with a 3D–matrix array transducer (X7-2 t). Screening and follow-up examinations were done in conscious sedation using benzodiazepines. General anaesthesia was established for PMVR. The aetiology of mitral regurgitation was described as degenerative (DMR) or functional (FMR).

3D–colour Doppler datasets of the mitral valve were obtained from mid-oesophageal views. Seven electrocardiographically triggered sequential 3D–scans were composed for a 3D–colour full volume dataset. A post-hoc analysis was performed using commercially available software packages (Xcelera R3.2 L1, version 3.2.1.820–2011; Qlab, version 10.5, Philips Medical Systems, Amsterdam, Netherlands) according to current guidelines [2]. Tissue priority and tissue threshold were set to factory settings. All recorded 3D–colour full volume datasets were checked for lines of disagreement between neighbouring 3D subvolumes (“stitching artefacts”) within the VCA borders or “dropouts” within the dataset. Care was taken to identify blooming effects, rendering the Doppler signal larger than the laminar jet core itself [2]. If artefacts were present, the patients were excluded from further analyses (8 subjects). The median 3D frame rate came to 18 Hz, as recommended [22], with a narrow interquartile range (15 to 18 Hz; minimum 8 Hz in 1 subject). VCA was defined as the cross-sectional area of the narrowest portion of the proximal regurgitant jet through the closed mitral valve in early-to mid-systole [8, 23]. The datasets were manually cropped to provide a direct en-face view perpendicular to the jet direction. The Nyquist limit was stepwise reduced to a median of 41.1 cm/s (30.8;41.1) to visualise the colour flow regurgitant jet with maximum clarity as previously described [24, 25]. To identify the level of the narrowest portion of the regurgitant jet, the datasets were tomographically sliced. A manual planimetry of the colour Doppler signal was performed (Fig. 1). Care was taken to measure only the central laminar jet core [26] of highest, similar, transverse velocity and to exclude low-velocity eddies as recommended by previous publications and the current guidelines [2, 27]. As after PMVR there used to be multiple jets, VCA of each jet was determined separately and summed up as for complex mitral regurgitation [9].

Statistical analysis

Categorical data are expressed as percentages. Their differences were tested for significance by Pearson’s chi-squared test. The distributions of continuous variables were assessed for normality by Shapiro-Wilk test. If normally distributed, they are expressed as mean ± standard deviation. Significance of differences was tested by Student’s t-test for dependent or independent variables, respectively. Two-way analysis of variance (two-way ANOVA) was computed to analyse the influence of categorical independent variables on left ventricular volumes and ejection fraction, respectively. When normal distribution was rejected, median and interquartile range (P25; P75) are given and variables are shown as Turkey box plots. Mann-Whitney U test, Wilcoxon signed-rank test and Friedman’s test with consecutive Dunn’s multiple comparison test were performed, as appropriate. Effect size was approximated as Cohen’s d or Hedges’s g [28] using dedicated software [29]. To assess the reduction in VCA, the ratio (VCAr) was calculated as quotient (VCAr = VCA PMVR/VCA at baseline). The absolute area of VCA reduction (VCAdiff) was defined as difference (VCAdiff = VCA baseline–VCA PMVR). VCAdiff gives the VCA reduction in absolute numbers [cm²]. Six-minute walk change (6MWc) was calculated as difference (6MWc = distance after–before PMVR) [30]. Kendall rank correlation coefficient (τ) was calculated to measure the degree of correlation of non-parametric data.

All statistical analyses were performed using SPSS statistics version 22 (IBM, Armonk, New York, USA) and GraphPad Prism Version 6.00 (GraphPad software, La Jolla, California, USA). Statistical significance was assigned at a two-sided p-value of less than 0.05.

Power calculation analysis

For post-hoc power analysis G-Power [31] (version 3.1.9.2) was employed. A post-hoc power calculation was performed for the two main readouts (decrease in VCA and six-minute walking distance). It revealed sufficient power (β < 0.0001/0.01) for decrease in VCA/6 MW (Additional file 2).

Patients’ characteristics

Effect of PMVR on MR, 6 MW and NT-proBNP levels

6MW was significantly improved by PMVR. The effect was particularly pronounced in patients suffering from DMR (240.4 ± 80.3 m vs. 296.1 ± 63.0 m, Cohen’s d 0.97, p = 0.013, n = 10). In FMR, effect size was smaller and slightly missed significance (266.5 ± 84.4 m, Cohen’s d = 0.47, p = 0.053, n = 19). In consequence, PMVR achieved as early as 4 weeks after the procedure a significant improvement of 6 MW with a more distinct effect for DMR.

NT-proBNP blood levels, left-ventricular volumes and ejection fraction did not show a significant change (Table 4).

Reduction of VCA by PMVR and consistency between intra-procedural measurement and follow-up examination

VCA determination was independent of age, sex, left ventricular ejection fraction, NYHA class and EuroScore. Baseline VCA was significantly reduced from 0.89cm² to 0.17cm² after PMVR (Table 4). Median VCAr was 0.19 (0.09;0.42). In the subgroup of patients with follow-up 3D–TEE, VCA did not vary between the measurements immediately after PMVR and during follow-up (p = 0.999, Fig. 2). Their characteristics did not differ from the group missing a follow-up 3D–TEE examination with respect to sex, age or clinical manifestation of MR (Additional file 3).

Correlation of intra-procedural VCA measures and grading of residual MR

Common ordinal scaled MR grading (1 to 4) and VCA were compared for evaluating residual MR. There was significant but weak correlation (τ = 0.361; p = 0.01). As seen in Fig. 3, the VCAs of all ordinal scaled MR grades after PMVR spread widely.

VCA reduction as predictor of clinical outcome

The link of MR reduction (VCAr) and an improvement of the patients’ physical capacity (6MWc) is depicted by Fig. 4. Using the median VCAr of 0.19 as cut-off value, patients with a more pronounced procedural VCA reduction, mirrored by a VCAr below the median, had a significantly smaller 6 MWc. Contrary to expectations, data imply a more modest success for functional improvement in patients, whose technical success in VCA reduction was essentially more distinct.

To scrutinize possible underlying causes, two groups divided by the median VCAr were compared. Explorative data analysis did not yield significantly differing results between groups except for the absolute area of VCA change (VCAdiff) (Additional file 4). VCAr and VCAdiff were correlated with a negative Kendall rank correlation coefficient (τ = −0.51, p = 0.0001). Thus, as VCAr decreases, VCAdiff increases, which seems quite conclusive.

Based on these observations, we speculated, whether effects of very large VCA could drive our results. The 75% quantile of VCAdiff was used as cut-off to differ between low and high VCAdiff (75% quantile = 1.05 cm²). Patients, who were suffering from FMR and exhibited a VCAdiff below the 75% quantile, had a post-procedural increase in six-minute walking distance (+43.57 ± 49.53 m, n = 14). By contrast, patients with larger VCAdiff came to a decreased six-minute walk distance after PMVR (−12.20 ± 76.59 m, n = 5). The difference showed a strong effect (Hedges’s g = −0.977) slightly missing significance (p = 0.078).

There was no difference between small and large VCAdiff regarding changes in post-procedural left ventricular volumes or function compared to baseline (p for end-diastolic volume/end-systolic volume/ejection fraction = 0.97/0.68/0.45, 2-way ANOVA).

Thus, in large, potentially long-standing FMR, our data was indicative for a less beneficial effect than in smaller FMR with regard to the patients’ functional outcome.

Study sample and treatment success in the context of published literature

The baseline characteristics of the Regensburg PMVR Registry reported for the first time by our study were quite comparable to currently published results of several registries [1, 17–21] (Table 1) mirroring the real-life practice. The use in FMR was consistently reported higher than in the initial EVEREST-II trial [1]. Remarkably, in our data the effect of PMVR on patients’ functional capacity (6 MW) was more pronounced in DMR. NT-proBNP as marker of left ventricular wall stress and predictor of cardiovascular outcome [32] was elevated before and after PMVR without significant change in line with a current publication reporting no benefit of PMVR with regard to NT-proBNP levels in 144 patients [33]. Nevertheless, the field is still not settled due to contrary results in smaller samples [34]. Heterogeneous comorbidities might explain nonresponse of NT-proBNP after PMVR [35].

VCA as independent measurement for quantifying residual MR after PMVR

In patients with high grade MR and elevated surgical risk, PMVR is increasingly used leading to significant improvements in clinical outcome [1, 14, 19]. The edge-to-edge technique of PMVR alters the complex anatomy of the mitral valve in DMR [36] and FMR [10] and creates at least two neo-orifices. The remaining MR is split into often very eccentric regurgitant jets (Fig. 1) and common parameters of echocardiographic MR assessment as width of vena contracta and effective regurgitant orifice area are not applicable. The sparse existing guidelines (e.g. of the German Cardiac Society [5], Additional file 1) recommend a multimodal approach integrating parameters determined by echocardiography (visual grading of regurgitant jet), right-heart catheterization (v-wave) and left ventriculography (regurgitant volume) with echocardiography as mainstay. Though in real-life most often used during PMVR, the qualitative as well as quantitative assessment of the regurgitant jets using colour-Doppler is imprecise and tents to an overestimation of remaining MR in the situation of multiple jets [37]. Even before the launch of PMVR, VCA was known for several strengths in the assessment of high-grade MR [8]: In an in vitro model of MR, VCA provided the strongest correlation with known orifice area (r = 0.92, p < 0.001) compared to other echocardiographic measurements, which could be translated to a prospective study comprising 61 patients with at least mild MR of different aetiology: feasibility and reproducibility was established yielding satisfying interobserver agreement (r = 0.96; 0.05 ± 0.02 cm²) [38]. In the same year, a further prospective study including 57 patients with relevant MR of different aetiologies [39] reported feasible measurements in all patients within 2.6 ± 0.7 min of measuring time and ruled out significant interobserver variability (r = 0.97, 0.04 ± 0.09 cm²). VCA is reliable in multiple jet areas, too [9].

Considering VCA measurement after PMVR, Altiok et al. set the stage by using VCA determined by 3-D-TEE to analyse the procedural effects of PMVR in FMR [11] with acceptable feasibility and reproducibility. In 2017, these data were confirmed [10] by a retrospective study comprising 97 heart failure patients with severe MR undergoing MitraClip therapy reporting adequate inter-observer variability (r = 0.95, p < 0.001). Comparing VCA to the common ordinal scale of MR grading, our data show that within each MR grade VCA still spreads. It highlights the potential of VCA measurement to increase the resolution of residual MR grading.

The intraprocedural TEE has to face the inherent problem of anaesthesia, which changes cardiac pre- and afterload influencing mitral valve function [40]. Interestingly, VCA did not change in our dataset between the intraprocedural TEE in general anaesthesia and a follow-up examination, which has been done in conscious sedation using benzodiazepines 4 weeks later. These results might indicate a quite comparability of both examinations regarding particularly VCA measurement.

VCA as predictor of post-procedural outcome

The need for a reliable measurement of residual MR is underscored by its prognostic importance: data from the MitraSWISS registry revealed residual MR severity after PMVR as significant predictor of reduced survival after 2 years [18]. They suggested that MR should be reduced as far as possible. It has to be stressed that these analyses relied on ordinal scaled grading of residual MR by combined methods. In 2017, Alessandrini et al. measured VCA in FMR patients after PMVR. Dichotomised VCA (≥ 0.25cm², upper vs. lower and middle tertile of their sample) was associated with mortality during a median follow-up of 13.4 months (HR = 3.8, CI 1.9–7.8) [10]. Our data confirm prognostic implication of remaining MR. However, since MR severity is very dynamic, we hypothesised beyond previous published analyses, that there might be differences in outcome depending on pre-procedural MR anatomy. Therefore, we wanted to assess the association of decreasing VCA (ratio post/pre-procedural) and the patients’ outcome, which was measured as gain in 6 MW. Strikingly, we observed a more pronounced increase in 6 MW 4 weeks after PMVR in patients, whose VCA was less reduced by PMVR. This paradoxical result was predominantly driven by the negative outcome of patients suffering from FMR with a large absolute difference between their VCA before and immediately after PMVR (> 75% quantile, > 1.05cm²). These patients lacking functional improvement after PMVR had considerably large functional regurgitant jets measured as VCA 2.5-fold higher than the cut-off value defining the edge between moderate and high-grade disease [8, 39, 41]. It is tempting to speculate whether PMVR strictly decreasing VCA might be less beneficial in very-large, presumably chronic and long-standing FMR than at an earlier point of intervention as additionally suggested by recent data [42]. For FMR comprises a variety of entities as a result of cardiac remodelling [43–46]. Thus, our study might suggest with VCA an interesting, objective, measurable pre-procedural criterion for PMVR planning. Nonetheless, this issue awaits further, strongly required prospective evaluation as patient selection in FMR for PMVR is a central and current problem and valid parameters for this purpose are strongly needed.

Strengths and limitations

Strengths of our study include precise measurement of MR by the direct planimetry of VCA using 3D–TEE, which has already been shown to be accurate [8] and feasible in multiple jets [9] even after PMVR [10, 11]. However, data on DMR has been lacking to date. Furthermore, nor VCA neither another method has been used until now to assess prognostic implication of residual MR concerning functional treatment success. Thus, a novel approach was chosen and provided new insights. Furthermore, the recorded follow-up 3D–TEE examinations permitted an analysis of stability during short-term follow-up.

However, some limitations warrant consideration: reduction in regurgitant volume immediately after PMVR could similarly be discussed as another potential marker of later clinical outcome. Regurgitant volume is estimated as difference of stroke volumes measured at the LVOT and mitral valve level [47] or alternatively, by magnetic resonance imaging. Both methods are applicable even after MitraClip [48]. Unfortunately, we do not have this data to compare it with VCA reduction. This issue awaits future studies.

The retrospective design with a moderate-sized number of participants limits analytic options. Nevertheless, it is the first study in the field addressing this current and highly relevant issue by testing a clear and unambiguous hypothesis and using a precise measuring method as well as a quantifiable, relevant outcome variable. Thus, it allowed a thorough statistical analysis even in a medium-sized sample with appropriate statistical power.

Still, due to the small sample size, little effects can be missed and non-significant results do not rule out a potentially overseen small effect. Nonetheless, our study may report on significant effects yielded by PMVR. However, albeit significance was computed, some justified concerns about generalizability might remain because of the small sample size and should by answered by further research. To offer valid information at the moment, we provide also estimates of effect sizes beside p-values, which could facilitate a-priori power calculation for future prospective studies.

The study emphasised the importance of precise echocardiographic imaging in PMVR, although all large registries ignore new measurements of residual MR as VCA (Table 1). It is our hope that our results will help to design future prospective studies, which further elucidate the prognostic meaningfulness of residual MR particularly in outsized FMR.