Does the clinical effectiveness of Mitraclip compare with surgical repair for mitral regurgitation?

Surgical repair of the mitral valve has long been the established therapy for degenerative mitral regurgitation (MR). Newer transcatheter methods over the last decade, such as the MitraClip, serve to restore mitral function with reduced procedural burden and enhanced recovery. This study aims to compare the shortterm and midterm outcomes of MitraClip insertion with surgical repair for MR.


| INTRODUCTION
Mitral regurgitation (MR) is the most common valvular disorder in the United States. 1 Primary or degenerative MR is most commonly caused by mitral valve prolapse. 2 It is also caused by the degenerative changes seen in ageing, infective endocarditis, rheumatic heart disease, and chronic annular calcification. 2,3 Secondary or functional MR is caused by cardiac ischemia and heart failure. 1 Severe MR carries a high mortality rate of at least 5% per annum, 4 while contributing to significant health-related morbidity and healthcare burden. Curative management requires anatomical correction of the mitral leaflets to restore adequate leaflet coaptation, alongside annulus remodeling, which commonly leads to reverse left ventricular remodeling. This has been shown to be successful in primary MR 5 but remains unclear in secondary MR due to recurrence of regurgitation and need for reoperation, 6 which is mainly due to ventricular pathology that standard mitral repair techniques do not tend to address. Current guidelines recommend surgery for symptomatic patients with chronic severe primary MR (stage D) and left ventricular ejection fraction (LVEF) greater than 30% as well as asymptomatic patients with chronic severe primary MR and left ventricular dysfunction (LVEF 30%-60% and/or left ventricular endsystolic diameter ≥ 40 mm, stage C2). 7 Despite being the best available treatment, a large number of high-risk patients with severe MR are denied surgery due to frailty, old age, left ventricular dysfunction, and other associated comorbidities. 8 Percutaneous edge-to-edge repair has been developed as an alternative to open surgical repair for these high-risk patients.
The MitraClip system (Abbot Vascular) is the most widely used system that has been adopted in many cardiac centers globally. 9 It involves femoral venous and transseptal access to the left atrium, followed by the insertion of a clip through the mitral orifice into the left ventricle to hold the edges of the mitral leaflets and allow vertical coaptation. 10 MitraClip has been proven to be superior to medical therapy for the management of functional MR, 11 demonstrated in a single randomized controlled trial (RCT) and observational studies.

| Inclusion and exclusion criteria
Studies incorporating data from two distinct groups (MitraClip and surgical MVR) in patients undergoing intervention for any type of defined MR were included. Studies which highlighted mitral valve replacement in the surgical repair arm were included. Studies reporting outcomes from only one of either of the procedure groups without comparison were excluded. Only articles in the English language were considered. Any discrepancy in decisions regarding study inclusion were independently evaluated by two senior authors (M.Y.S. and T.A.) for concordance.

| Data extraction
The primary aims of data analysis were to highlight any differences in the burden of procedure and durability of repair between Mitraclip and surgical techniques. A standardized form for data collection for the input variables and outcomes was created and distributed among three authors. Data accuracy was cross-checked by two senior authors (M.Y.S. and T.A.) focusing on the following: • Preoperative patient characteristics: primary patient demographics, comorbidities, relevant echocardiographic features (grade of MR, LVEF %), and functional class.
• Operative characteristics: devices implanted, surgical access strategies, complexity of mitral repair, concomitant procedures, cardiopulmonary bypass times and aortic cross-clamp times.
• Early outcomes: (in-hospital or up to 30 days postsurgery) death, incidence of complications (including incidence of stroke and atrial fibrillation) and time spent in intensive care/hospital.
Durability of repair is a key measure of procedural performance we were aiming to compare in this meta-analysis. MR recurrence was measured as grades of regurgitation defined as failure to treat at follow-up reported in each individual study ( Table 2). In studies in which this cut-off was non-defined, a threshold of ≥ moderate was used to define incidence of MR recurrence, as per guidelines. 15

| Statistical analyses
The OR was used as the summary statistic for operative deaths and other binary outcomes. A random-effects meta-analysis was used to find an overall OR comparing surgical MVR with the MitraClip for 30-day operative mortality rates due to the expected heterogeneity between the studies. Similarly, a random effects meta-analysis was used to find an estimate of the overall HR for long-term survival comparing the two study arms. The inverse-variance weighting method was used to pool results from the studies. If studies did not present the HR but gave a Kaplan-Meier curve and numbers at risk for overall mortality, the method by Parmar 16 and the spreadsheet applying this method were used to give an estimate of and the standard error of the log HR. For continuous outcomes, reported means and standard deviations were used in a meta-analytical model to calculate standardized mean differences.
Heterogeneity was investigated using Cochrane's test and the I 2 statistic. Heterogeneity cut-offs were defined as follows: moderate (I 2 = 30%-60%), substantial (I 2 = 50%-90%). With an I 2 > 75% reflecting considerable heterogeneity. 17 Funnel plots were generated to assess for publication bias. Egger's test for small studies was conducted to rule out large effects from potentially nonsignificant studies. Meta-regression analysis was used to investigate the effects of covariates, including patient and operative characteristics. Statistical analyses were conducted using the Stata 13.0 software (Stata Corp.) and R 3.6.1 (R Core Team (2019); R Foundation for Statistical Computing).

| RESULTS
The initial systematic search yielded 1848 titles, from which 11 duplicates were excluded ( Figure 1A). The MitraClip device first received a certificate marking in 2008, with the first trial design published in 2010 and patient data regarding its use published by Feldman et al. 10 Therefore, a further 217 records before the dawn of MitraClip in 2010 were also excluded. Following the initial screening of titles and abstracts, 42 studies were selected for full-text review.
A single RCT was identified, 12 that provided more than one interval publications of its outcomes. 18−20 Based on this, the most recent publication was selected as the principal RCT. For the utilization of this RCT, if there was no reporting of required data in Feldman et al. the interval studies that did report relevant data were used for meta-analysis, 18,19 but multiple studies were never analysed simultaneously. This ensured maximal representation of the only RCT of Mitraclip versus surgical repair to date. One study 21 utilized patients from a national inpatient sample. While this study represented a high throughput of patient data, there was the potential of patient overlap with other studies in the analysis, as a result this study was excluded to try to maintain validity. Eventually, 12 studies 12, [22][23][24][25][26][27][28][29][30][31][32] were included in the final analysis (1 RCT, 11 retrospective analyses).
A funnel plot was constructed ( Figure 1B) which showed little evidence of publication bias, with only one study detected as an outlier. 25 However, no significant small-study effect in the overall study sample was identified through the Egger's test (p = .125). All subsequent analyses were repeated a second time to remove the outlying study from the evidence pool, although this was found to have no influence on the results (see Appendix 1).

| Valve pathology and operative strategy
Pathoanatomy of MR with regard to type of regurgitation is shown in  with a low level of heterogeneity (I 2 = 0.0%) ( Figure S1).

| Meta-regression
Given the perceived and measured heterogeneity between studies in the two treatment arms, meta-regression analysis was conducted to assess the influence of three covariates (age, EuroSCORE, and LVEF) had an influence on the measured outcomes. The decision to test these covariates was made since >40% of included studies reported significant differences in these covariates in their individual cohorts.
It was considered to assess the type of MR as a covariate, however only 33% of studies specified valve pathology, therefore it was decided there would be too little statistical power to include this in meta-regression.
Similarly, differences in age and LVEF were found not to influence reoperation (covariant meta-regression p = .903, .702, respectively). The considerable heterogeneity present in the meta-analysis for MR recurrence was also tested against the covariates. However, age, EuroSCORE, and LVEF were found not to significantly influence the measured outcome of MR recurrence (p = .624, .261, .820, respectively) (Table S1).

| DISCUSSION
The present study comprehensively analyses available evidence to evaluate the performance of an innovative device against established gold standards. In doing so, the study aims to add important value to the ongoing debate regarding the appropriate treatment of MR in high-risk populations. The statistical analysis has identified that whilst the MitraClip can enhance recovery in the short term, overall survival is not improved and moreover, the durability of valve repair is likely to be inferior to surgical management.
A previous meta-analysis conducted by Cardosa et al. 33 in 2016 evaluated differences in mortality between surgery and Mitraclip, finding a reduced incidence of cardiovascular mortality in patients receiving a Mitraclip. Our analysis builds upon these results, incorporating additional five studies that are more up to date, perhaps reflecting the differences in conclusions.
Studies have shown that up to 49% of patients with degenerative MR can be denied surgery by cardiac services. 8 The main patient characteristics that lead to this decision are age, poor left ventricular function, or multiple comorbidities. 34 The rationale in the design of The present study has also found that length of hospital stay was significantly shorter in patients receiving a MitraClip, with patients often being discharged within 1-2 days, 28 enhancing patient recovery while providing cost benefits. Perioperative mortality was not significantly different between patient arms, however, surgical repair often requires increased perioperative support, such as higher blood product requirements, ventilatory considerations, complex pain management, and cardiac rehabilitation. 24 This supports that the MitraClip and the development of similar devices will continue to have an important role in the multi-disciplinary process of the management of degenerative MR.
Overall, the durability of valve repair offered by surgery is not outweighed by the reduced procedural burden of MitraClip repair, especially given the similar short-term complication and survival rates between both treatments. Therefore, we conclude that, regardless of patient baseline characteristics, judging by perioperative mortality and stroke incidences, conventional surgery is non-inferior to the novel MitraClip approach. This finding challenges the current practice, as well as findings from a few previous studies, that older, more fragile, high-risk patients are more likely to benefit from a MitraClip procedure.

| Limitations
The present study provides useful insights into two treatment comparisons, although the results should also be taken with caution. The conclusiveness of this meta-analysis was restricted by the amount of studies available. Only one RCT was included in the analysis, with the remainder being smaller retrospective comparisons, therefore the quality of the data set could be impacted by selection biases due to their largely nonrandomized nature. Furthermore, the only RCT highlighted a proportion of patients who were operated on immediately after an index Mitraclip procedure, but data for these patients were not accessible to be crossed over into the surgical arm for re-operative results. MR recurrence, reoperation, and survival data were only pooled from four studies. Many of the meta-analyses demonstrated moderate to considerable heterogeneity levels, reflecting the variation in patient groups, primarily in age and comorbidity. Despite the use of meta-regression techniques to demonstrate little statistical effect of the variability of the studies on covariates and the outcomes measured, inherent differences at the patient level can potentially affect the outcomes. Above all, the value and validity of this preliminary finding would benefit from further comprehensive studies, especially RCTs with larger sample sizes.

| CONCLUSION
The