Introduction

Secondary mitral regurgitation (SMR) refers to mitral regurgitation (MR) in the absence of structural abnormalities of the mitral valve complex and occurs most frequently in the setting of left ventricular (LV) dysfunction. The interaction between MR and LV dysfunction is complex, as both pathologies result in LV volume overload with subsequent disease progression. The presence of substantial SMR in patients with LV dysfunction is associated with worsened quality of life and increased mortality (1–4). Despite its recognized clinical importance, the diagnosis and treatment of SMR remains challenging. Treatment of heart failure (HF) with medical therapies or ventricular resynchronization may improve the severity of SMR and patient outcomes (5–7), although surgical repair of SMR has not been demonstrated to improve prognosis (8–10).

The COAPT (Cardiovascular Outcomes Assessment of the MitraClip Percutaneous Therapy for Heart Failure Patients with Functional Mitral Regurgitation) trial has demonstrated that transcatheter mitral valve repair (TMVr) through leaflet approximation reduces the rate of hospitalizations and improves survival in selected patients with HF and SMR compared with maximally tolerated guideline-directed medical therapy (GDMT) alone (11). While the COAPT trial demonstrated unequivocal evidence for clinical improvement with MR reduction, a similar randomized study, the MITRA-FR (Percutaneous Repair with the MitraClip Device for Severe Functional/Secondary Mitral Regurgitation) trial was negative, possibly due in large part to differences in the echocardiographic severity of MR and LV dysfunction in the enrolled patient populations (12). As the shape of the MR regurgitant orifice in SMR is elliptical rather than circular, the use of classic MR parameters such as proximal isovelocity surface area (PISA) or vena contracta (VC) can be problematic. Evaluating SMR severity after TMVr is especially challenging given the presence of multiple mitral orifices. A lack of consensus among experts has resulted in discordant recommendations from professional societies (13–16).

To address these issues, the COAPT investigators employed a specific echocardiographic core laboratory MR assessment algorithm to qualify patients for the COAPT trial. This methodology and its impact on trial outcomes have not been described. We therefore herein describe the echocardiographic methodology used in the COAPT trial and the echocardiographic features characterizing the eligible population; analyze the effect of intervention on LV remodeling and function; and identify echocardiographic characteristics that predict favorable outcomes after TMVr in HF patients with severe SMR.

Methods

Study design

The COAPT trial design has been published previously (17). In brief, COAPT was a multicenter, randomized, controlled, open-label trial of TMVr with the MitraClip device (Abbott, Chicago, Illinois) in patients with HF and moderate-to-severe (3+) or severe (4+) MR who remained symptomatic despite maximally tolerated guideline-directed medical therapies (GDMT). Patients had LV ejection fraction (EF) between 20% and 50%, left ventricular end-systolic diameter (LVESD) ≤70 mm, and absence of severe pulmonary hypertension (defined as pulmonary artery systolic pressure >70 mm Hg despite vasodilator therapy) or moderate or severe right ventricular failure. Patients were randomized to receive TMVr plus GDMT or GDMT alone.

Echocardiographic follow-up was performed at 1, 6, 12, 18, and 24 months after randomization. The primary effectiveness endpoint was all hospitalizations for HF within 24 months, assessed when all patients had completed at least 1 year of follow-up. All transthoracic echocardiograms were analyzed by an independent echocardiographic core laboratory (MedStar Health Research Institute, Washington, DC).

The protocol was approved by the investigational review board at each participating center, and all patients provided written informed consent. Abbott sponsored the trial and provided statistical support for the present analysis. The investigators had unrestricted access to the data and accept responsibility for the integrity of the present report.

Echocardiographic core laboratory analysis

Among patients who were identified as possible trial candidates at the sites, the echocardiographic core laboratory was responsible for confirming the presence of 3+ or 4+ SMR and other echocardiographic eligibility parameters from the screening transthoracic echocardiograms, rejecting randomization of ineligible subjects. For qualification purposes, MR severity was assessed as 3+ or higher following a pre-specified multiparametric algorithm created for the COAPT trial (Central Illustration, Online Table 1), adapted from the criteria recommended by the American Society of Echocardiography (ASE) 2003 guidelines (18,19). This algorithm consisted of 3 tiers of evaluation that were applied in a hierarchical manner; patients qualified for COAPT by meeting at least 1 of them. Follow-up MR severity was assessed using an integrative approach based on qualitative and quantitative data adapted from the ASE guidelines. MR was categorized as 0 (none), 1+ (mild), 2+ (moderate), 3+ (moderate to severe), or 4+ (severe) as detailed in Online Table 1 (19,20). All other echocardiographic parameters (Table 1) were analyzed based on ASE recommendations (19,21). Follow-up echocardiograms did not have PISA evaluation, as the presence of the MitraClip would prevent accurate, reproducible measurements of multiple regurgitant, eccentric jets with nonhemispheric proximal flow convergence. Therefore, MR regurgitant volume (RV) and regurgitant fraction (RF) at follow-up were obtained from Doppler hemodynamic and volumetric analysis. If more than 1 regurgitant jet was identified at follow-up, the vena contracta width of the dominant jet was measured or the vena contracta widths of multiple jets were added if more than 1 was considered significant.

Statistical analysis

Baseline characteristics were summarized with mean ± SD for continuous measures and proportions for categorical variables. Between treatment groups, variables were compared with the Student’s t-test for the continuous measures and proportional odds model for the categorical variables. For time-to-first event analyses, event rates were estimated by the Kaplan-Meier method and compared with Cox regression. Multivariable analysis was performed to identify the baseline variables that were independent predictors of the 24-month rate of death or HF hospitalization in each treatment group. The variables entered into these models are listed in Table 2. Changes in echocardiographic parameters over time were calculated as the difference between the baseline and follow-up visits. Subjects without an available follow-up echocardiographic image who had an adjudicated HF death prior to that visit were assigned the worst change from baseline to that visit. For all other subjects who had missing echocardiographic values due to other reasons (e.g., death not due to HF, withdrawals, missing echoes, and so on), multiple imputation with Markov Chain Monte Carlo was used. The imputations were done within treatment groups up to 12-months of follow-up. For the 18-month and 24-month visit, only eligible subjects were included in the analysis. Analysis of covariance was performed to compare changes over time adjusted for baseline values. A 2-sided p value <0.05 was considered statistically significant for all superiority tests. All statistical analyses were performed with SAS software, version 9.3 (SAS Institute, Cary, North Carolina).

Results

Screening echocardiographic assessments and study enrollment

Between December 2012 and June 2017, 1,576 subjects were screened at 78 centers in the United States and Canada, of whom 911 (57.8%) were ineligible. Principal echocardiographic exclusion criterion included <3+ MR severity, primary or mixed MR etiology, LVESD >70 mm, and LVEF <20% or >50%. Among the 665 enrolled patients (including roll-in subjects), 85.7% met the ≥3+ MR severity criteria based on the first tier of the multiparametric algorithm, while the remainder qualified based on tiers 2 or 3 (Central Illustration). Among those who qualified as tier 1, 41.5% met both the effective regurgitant orifice area (EROA) >0.3 cm² and pulmonary vein systolic flow reversal criteria, 54.9% met only the EROA criterion, and 3.5% met only the PV flow criterion.

A total of 51 patients were treated with TMVr as roll-ins and 614 were randomized (302 to TMVr plus GDMT, 312 to GDMT alone). Baseline clinical characteristics of the 2 treatment groups were well matched, as reported elsewhere (11).

Baseline echocardiographic characteristics

Overall, the feasibility to obtain most baseline echocardiographic measures was high. LV volumes and LVEF by Simpson’s rule, PISA-derived EROA, VC, and color Doppler MR were all obtained in >93% of cases. Pulmonary vein flow was assessible in 77.2% of cases and right ventricular systolic pressure (RVSP) in 86.0%. Grading of MR severity by the ASE-derived integrative approach was possible in all patients. However, significant limitations were encountered in the assessment of RV and RF by the combined 2-dimensional (2D) volumes Simpson’s rule/Doppler hemodynamics method. These calculations were feasible in only 42.3% of patients at baseline and were available for paired baseline and follow-up analysis in only 11.4% of patients. Furthermore, the volumetric analysis obtained from 2D and Doppler were frequently discrepant due to underestimation by the 2D Simpson’s method (22). Therefore, these variables were excluded from further analysis.

The mean LVEF was 31.3 ± 9.3%, left ventricular end-diastolic volume (LVEDV) was 192.7 ± 71 ml, RVSP was 44.3 ± 13.7 mm Hg, and 16.4% had tricuspid regurgitation (TR) of moderate or higher severity. MR was graded as 3+ and 4+ in 52.2% and 47.8% of patients, respectively. By PISA evaluation, mean EROA was 0.41 ± 0.15 cm². The device and control groups had similar baseline echocardiographic characteristics (Table 1), except for pulmonary vein flow (higher incidence of systolic flow reversal in the device group; p = 0.02). Baseline echocardiographic characteristics of patients who qualified in each of the 3 tiers are presented in Online Table 2.

Baseline characteristics and outcomes

Unadjusted analysis of baseline echocardiographic parameters was performed to identify predictors of first HF hospitalization or death. The salutary effect of TMVr compared with GDMT alone in reducing the time to death or HF hospitalization was consistent in all echocardiographic subgroups at 12 months (Online Table 3) and at 24 months (Figure 1). Reduced LVEF, greater EROA and RVSP, and the severity of TR were independent predictors of death or HF hospitalization within 24 months in patients randomized to GDMT alone. In contrast, only higher RVSP and Society of Thoracic Surgeons (STS) score were predictive of death or HF hospitalization in TMVr-treated patients (Table 2). The independent predictors of 24-month mortality alone are shown in Online Table 4.

Echocardiographic changes over time

At 30-day follow-up, only 7.4% of TMVr-treated patients had ≥3+ MR, an effect that was durable throughout the 24-month follow-up period (Online Figure 1). Some patients in the control group also had improved MR severity during follow-up (e.g., 34.3% had ≤2+ MR at 30 days), but much fewer than after TMVr at all time points (all p < 0.001). Compared with baseline, MR improved by ≥2 grades at 12 months in 84.1% of alive patients in the device group and in 15.9% in the control group (p < 0.001).

Changes in LV volumes, LVEF, and other parameters over time are shown in Table 3, Figure 2, and Online Table 5. LV end-systolic volume and LVEDV increased over time in both groups, but less so after TMVr. The LVEF also decreased over time in both groups, initially more so after TMVr compared with GDMT alone. However, by 12 months the reduction in LVEF from baseline was less in patients treated with TMVr compared with control.

Discussion

The potential implications of effectively applying the COAPT results to clinical practice are substantial. At least 5.7 million patients in the United States have HF (23), ∼15% of whom have 3+ or 4+ SMR (24). The COAPT trial demonstrates that such patients who remain symptomatic despite maximally-tolerated GDMT may benefit from MR reduction with TMVr, in terms of improved quality of life and exercise capacity, reduced HF hospitalizations, and prolonged survival (11). However, not all HF patients with SMR derive benefit from TMVr. Specifically, patients enrolled in the MITRA-FR trial, who had on average less severe MR and greater LV dilatation, had similar 12-month rates of death or HF hospitalization with and without TMVr (12). The present analysis, representing the formal COAPT trial echocardiographic substudy, is thus of direct clinical relevance in identifying those patients who are likely to benefit (and not benefit) from TMVr.

In this regard, the major findings from the present study are:

Accurately assessing SMR remains problematic even at experienced centers. Challenges include the diversity of regurgitant orifice shapes, dynamic changes throughout the cardiac cycle, suboptimal views and assessment technique, suboptimal reproducibility of MR quantification methods, and lack of uniformity in utilization of the multiple parameters for evaluation of MR severity. Assessment of MR severity is thus often subjective and variable. To minimize such imprecision, the COAPT trial implemented a tiered algorithm for MR severity qualification based on multiple quantitative echocardiographic Doppler parameters based on the general principles espoused by the ASE and ACC (15,20). To our knowledge, COAPT is the first randomized trial of MR therapies that utilized such well-defined criteria requiring verification by an independent echocardiographic core laboratory. This algorithm assigned a hierarchical value to individual standard parameters, and its implementation resulted in a fairly homogeneous study population with 86% of patients qualifying by tier 1 (EROA ≥0.3 cm² or pulmonary vein systolic flow reversal). Moreover, the benefits of TMVr were consistent in patients qualifying as having ≥3+ SMR with any of the 3-tiered criteria. Coupled with the requirement for LVESD ≤7 cm and LVEF of 20% to 50%, use of this hierarchy provided that the COAPT population had a well-defined and consistent severity of SMR and LV dimensions and function. As a result, 24-month clinical outcomes were consistently favorable after TMVr compared with GDMT alone in all echocardiographic subgroups examined, regardless of baseline LVEF, LV volumes, RVSP, or MR or TR severity, testifying to the uniformity of the enrolled population.

The criteria to identify severe SMR in COAPT were more restrictive than that used in the MITRA-FR trial, which required a single measure of EROA >0.2 cm² or RV >30 ml, adapted from ESC recommendations (12,13,16). As a result of the different definitions for MR severity required, patients in COAPT had substantially more severe SMR than those in the MITRA-FR trial (mean EROA 0.41 cm² vs. 0.31 cm², respectively). Many patients enrolled in MITRA-FR may not have had severe enough SMR to clinically benefit from TMVr. In this regard, baseline EROA in COAPT was an independent predictor of the 2-year risk of death or HF hospitalization in the control group, but not after TMVr. Also, of note, the COAPT trial capped the upper LV dimension that was acceptable to treat (LVESD ≤7 cm), whereas there was no such limit in MITRA-FR. As a result, the mean indexed LVEDV was substantially larger in MITRA-FR than COAPT (135 ± 35 ml/m² vs. 101 ± 34 ml/m²). It is thus likely that the prognosis of MITRA-FR patients was dictated relatively more by their LV dysfunction than MR severity, in contrast to patients enrolled in the COAPT trial, as proposed by Grayburn et al. (25). Thus, adoption of the COAPT criteria by heart teams and their recommendation by societal guidelines to select patients with severe SMR for TMVr should increase the likelihood that high-risk HF patients will be identified and benefit from MR reduction, while avoiding treatment in patients whose prognosis is dictated more by advanced LV dysfunction.

MR improvement after TMVr was sustained throughout 24 months, attributable to the fact that TMVr directly addresses the principal structural abnormality in SMR, that is, lack of leaflet coaptation. Notably, a modest proportion of control group patients experienced improvement of MR by 1 grade over time, possibly reflecting the dynamic nature of SMR, modifications in medical therapy, or regression to the mean. Ongoing studies from the COAPT trial are examining the impact of achieving ≤1+ versus 2+ versus ≥3+ MR whether by device treatment or control.

The LVs in both groups further dilated during follow-up, reflecting the natural history of the underlying cardiomyopathy. Such remodeling, however, was worse in the control group, suggesting that MR-mediated volume overload reduction with TMVr slowed the progression of the underlying disease process. Of note, the LVEF decreased more within 1 month after TMVr compared with the control group, consistent with the early effects of increased afterload after MR reduction. Nonetheless, patients’ symptomatic status and functional class markedly improved after TMVr at all follow-up intervals, including at 30 days (11), likely due to the immediate reduction in left atrial pressure with TMVr. Moreover, at 12 months and beyond, the reduction in LVEF was less pronounced in the TMVr group than with GDMT only, reflecting long-term benefits on LV remodeling. Of note, it was during this period (≥12 months) that the survival advantage with TMVr emerged. Nonetheless, although these findings are intriguing, interpretations of the absolute and relative temporal changes in LV volumes and LVEF within and between the groups must be tempered by the risk of survivorship bias. Within 2 years, 46% and 29% of patients died in the control and device groups, respectively, and it is likely that LV remodeling and progressive LV dysfunction were worse in patients in whom follow-up echocardiographic measures were not available either due to death or disability. To attempt to account for this bias, we utilized a 2-stage imputation method that was pre-specified in the statistical analysis plan to account for missing data. There is no perfect method, however, to adjust for this degree of missing data, and these results should thus be considered hypothesis generating. The different imputation methods used in the present versus prior HF studies may also explain differences in observed LV remodeling patterns (26–28).

Study limitations

First, inconsistencies between 2D-derived LV volumes and total stroke volume (SV), and Doppler-derived forward SV and RV and PISA-derived MR severity have been noted (29). Such observations reflect the use of different echocardiographic techniques to assess these different measures and inherent limitations of echocardiography in grading MR severity and other parameters of interest. Specifically, although PISA is the most widely used and reproducible method to grade MR severity, it assumes a single jet, a round flat orifice, and constant flow throughout systole, conditions that are often not present with SMR. LV volumes, total SV, and LVEF are most commonly measured by Simpson’s method, which underestimates LV volumes, especially with dilated ventricles (22). Forward stroke volume is measured by continuous wave Doppler at the LV outflow tract. These 3 measurement techniques are not interchangeable, and accurate direct measures of RV and RF were available in a minority of cases. Thus, assuming that total SV from Simpson’s method equals the sum of the forward SV from Doppler hemodynamics plus an assumed RV from PISA-derived MR severity is erroneous. For these reasons, each reported value should be considered on its own merits, and efforts to reconcile differences between the different methods should be avoided. Second, evaluating MR severity after TMVr is challenging. Multiple regurgitant orifices are present, further complicated by jet eccentricity. Simply adding measurements for each individual jet is not accurate. The parameters that are most reliable after TMVr include evaluation of color Doppler jet area and direction, pulmonary vein flow, and to a lesser extent, the VC. The post-TMVr MR grading scale utilized in the COAPT trial was based on these parameters, as detailed in Online Table 1B. Whether newer technologies such as 3-dimensional color Doppler offer advantages over those used in COAPT deserves further study. Further study is also required to validate the utility of post-TMVr MR severity as assessed herein or by recent consensus statements (30). Third, the relationship between echocardiographic characteristics and outcomes has been analyzed in a relatively simplistic manner. The present report focused on subgroup analyses pre-specified at study inception, defined in most cases by the observed median measures (which avoids bias); further exploratory analyses will be carried out in specific subpopulations and to identify nonlinear relationships. Finally, longer-term follow-up (currently planned for 5 years) is necessary to determine the durability and long-term impact of TMVr of SMR in HF.

Conclusions

In the COAPT trial, patients with HF and 3+ or 4+ SMR who remained symptomatic despite maximally tolerated GDMT benefited from MitraClip implantation, regardless of their degree of LV dysfunction, LV dilatation, pulmonary hypertension, severity of TR, or individual MR characteristics. Only higher baseline RVSP and STS score predicted the risk of death or HF hospitalization after TMVr. The improvement in MR severity was dramatic and durable through at least 24 months. Echocardiography is of critical importance in determining the etiology and severity of MR and LV dimensions and function to identify those patients most likely to benefit from percutaneous leaflet approximation, while avoiding treatment of patients less likely to benefit. Advanced echocardiographic expertise specific to this technology and population is imperative to ensure that the results of COAPT (and MITRA-FR) are translated to “real-world” clinical practice to provide clinical benefit to a population that is at extremely high risk of death and HF hospitalization.

Footnotes