Introduction
Mitral regurgitation is the second most frequent indication for valve surgery in industrialised countries, with an overall prevalence of 9.3% in the general population who are above 75 years old [
1]. Traditionally, mitral valve replacement was the preferred option of treatment among surgeons as the valve repair was technically more demanding [
2]. However, there has been a surge in development of repair techniques post introduction of Carpentier’s techniques [
3]. Numerous qualified centres have reported excellent outcomes after mitral valve repair thanks to its reproducible clinical results [
4‐
7]. Therefore, mitral valve repair has been identified as the optimal intervention strategy to correct significant mitral regurgitation [
8]. However, the incidence of reoperation after initial mitral valve repair failure estimates at 4.5 to 8% at 10 years [
7]. The causes of incomplete repair can be categorised into procedure related and valve related factors [
9]. Procedure related factors encompass the technical failures of repair while valve related factors include the progression of native disease or new pathology such as degenerative, rheumatic, ischemic and endocarditis.
Although the incidence and causes of repair failure is well documented, little is known regarding the clinical outcomes post mitral valve re-repair (MVr) or mitral valve replacement (MVR). Thus, the optimal treatment strategy for an initial failed mitral valve repair remains unclear, thereby clouding surgeons’ decision making.
Hence, our present study aims to compare and analyse all available studies which report the clinical outcomes post MVr and MVR after a prior mitral valve repair.
Materials and methods
Search strategy
A literature search was performed using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines electronically utilizing Medline (PubMed), Cochrane and Scopus databases, from inception to 1st January 2020. A repetitive and exhaustive combination of the following ‘Medical Subject Headings’ (MeSH) were used: ‘mitral valve/surgery’, ‘mitral valve insufficiency’, ‘reoperation’, ‘recurrence’, ‘treatment outcome’, ‘treatment failure’, ‘survival analysis’ and ‘survival rate’. This study protocol was registered with PROSPERO #CRD42020160343. The full search strategy can be found in the supplementary materials (supplementary Table
1). Relevant articles were screened and systematically assessed with inclusion and exclusion criteria applied.
Eligibility criteria
The inclusion criteria included any retrospective cohort studies in which patients underwent surgical intervention with either a second repair (re-repair) or replacement after a prior mitral valve repair. Only studies that had a clear comparative data differentiation between the re-repair and replacement group were included, with the exception of studies with essential qualitative elements. Furthermore, only studies published after the year 2005 were included to prevent using outdated data. Additionally, any studies that were not written in the English language were excluded.
Data extraction and outcomes
Three reviewers (P.C, M. V, F.S) screened and assessed the studies independently for inclusion. The scientific papers were first screened by their titles and abstracts, where criteria were purposely broad to include all relevant studies. The full text review was performed on articles if the reviewer was unable to confirm the relevance of the study for inclusion.
Two authors (P.C, M.V) independently abstracted the details of the study population. The preoperative baseline characteristics extracted included the following: mean age, sex, history of diabetes, renal failure, prior atrial fibrillation and left ejection fraction.
Furthermore, relevant data of clinical outcomes was obtained from each study required for the generation of forest plots. Post-operative complications (such as stroke, atrial fibrillation, mitral regurgitation and congestive heart failure), requirement of 3rd mitral valve operation, re-operation due to bleeding, short term and long-term operative mortality were the main outcomes analysed in this study. Short term operative mortality was defined within a 30-day period. In addition, the meta-analysis included the Kaplan-Meier survival rates at 1,5,7 and 10 years.
Statistical analysis
The meta-analysis of the eligible studies was performed in line with recommendations from the Cochrane Community and the forest plots were generated through the means of Review Manager version 5.3 software (RevMan 5) [
10]. Since the clinical outcomes derived from the scientific journals were categorised under continuous data, the effect measures were estimated using odds ratio (OR). Odds ratio which was calculated using the Mantel-Haenszel method represents the odds of an adverse outcome occurring in the MVr compared to the MVR group. Heterogeneity (I
2) was graded as low (I
2 < 25%), moderate (25 < I
2 < 75%), or high (I
2 > 75%). All meta-analyses were carried out using random-effects models to account for statistical variability across the studies.
Furthermore, in presence of I2 > 25%, stability of pooled meta-analyses results were examined by the standard leave-one-out sensitivity analysis. This was conducted by removing the included studies one after another to validate the robustness of the results.
Quality of evidence and risk of Bias assessment
The 8 included retrospective cohort studies [
7,
9,
11‐
16] was all single centre studies (Table
1). The quality of these studies was assessed using Newcastle-Ottawa Scale [
17], seen in Table
2. Studies with a score more than or equal to 6 was considered to be of acceptable quality and included [
18].
Table 1
Summary of Included Studies
| 2018 | Retrospective cohort | 86 | 23 | 63 | Sakakibara Heart Institute | Operative mortality, postoperative morbidities, Long-term survival (mean follow-up period of 76.3 ± 55.0 months) |
| 2008 | Retrospective cohort | 13 | 9 | 4 | Service de Chirurgie Cardiovasculaire, Paris, France | Mechanisms of late failure, Long-term results were assessed on the basis of NYHA functional class, electrocardiogram and echocardiography |
| 2008 | Retrospective cohort | 43 | 21 | 22 | Service de Chirurgie Cardiovasculaire, Paris, France | Feasibility of Redo Mitral Valve Repair, Mechanisms of late Failure, Operative Mortality and Morbidity and Long-Term Outcomes. |
| 2017 | Retrospective cohort | 812 | 130 | 682 | US (Medicare Database) | Characteristics of Reoperation Cohort, Outcomes of Reoperation, Time to Reoperation, Hospital Mortality According to Hospital Annual Mitral Procedure Volume and Long-Term Survival |
| 2018 | Retrospective cohort | 40 | 23 | 17 | Shanghai Chest Hospital | Early mortality, Early major morbidities, Survival, reoperation for recurrent mitral valve pathology and echocardiographic data |
| 2018 | Retrospective cohort | 305 | 48 | 257 | University of Pennsylvania Health System | The primary outcome was operative mortality. Secondary outcome included postoperative complications and long term freedom from death. |
| 2006 | Retrospective cohort | 145 | 64 | 81 | Mayo Clinic Rochester | Indications for Reoperation, Predictors of Late Mortality, Predictors of Third Mitral Operation and Follow up data |
| 2007 | Retrospective cohort | 188 | 68 | 120 | The Cleveland Clinic | Mechanisms and Timing of Repair Failure, Freedom from reoperation, Incremental Risk Factors for Death after Mitral Valve Reoperation |
Table 2
Risk of Bias of Retrospective Cohort studies according to the Newcastle-Ottawa Scale
| * | * | * | * | – | * | * | * | 6 |
| * | * | * | * | – | * | * | * | 6 |
| * | * | * | * | – | * | * | * | 6 |
| * | * | * | * | – | * | * | * | 6 |
| * | * | * | * | – | * | * | * | 6 |
| * | * | * | * | – | * | * | * | 6 |
| * | * | * | * | – | * | * | * | 6 |
| * | * | * | * | – | * | * | * | 6 |
| * | * | * | * | – | * | * | * | 6 |
As recommended by chapter 14 of the online Cochrane Handbook version 5.1, the software GRADEprofiler (GRADEpro) was further utilised to validate the quality of evidence of the included retrospective studies (Supplementary Table
2) [
19]. Apart from the high risk of bias in confounding factors and patient selection that are typical of studies with retrospective nature, we determined that evidence provided by these studies are still of acceptable quality.
Discussion
To the best of our knowledge, this is the first systematic review and meta-analysis that compared the clinical outcomes between mitral valve replacement and mitral valve re-repair following failure of initial mitral valve repair surgery. Previously, there have been numerous reviews done on the outcomes of first mitral valve repair versus replacement [
4‐
6,
20] but none on a reoperation procedure involving the mitral valve.
In most aetiologies of initial mitral valve failure, it is commonly agreed by surgeons that mitral valve repair, rather than replacement, is the preferred surgical intervention due to the better reported clinical outcomes and freedom from reoperation and death [
4‐
6,
20]. However, in cases of failed initial mitral valve repair surgeons are often reluctant or hesitant to perform a re-repair owning to the first repair failure, and instead opt for replacement surgery [
11]. For instance, the study reports that although the feasibility of re-repair in their centre should have ideally been around 80% following the initial surgery, only 50% of the reoperations were reported to have been re-repair procedures as a result of surgeons’ reluctance to confront the hazards of a second mitral valve repair. Therefore, this meta-analysis was performed with the aim to throw light on and compare the clinical outcomes of MVR and MVr to see if the hazards and complications anticipated by the surgeons in fact hold true and, if not, hopefully provide a more definitive perspective.
The feasibility of mitral valve repair depends on Carpentier’s golden rules: the availability of sufficient leaflet tissue and its pliability [
11]. The general preference of repair over replacement is attributed to better preservation of left ventricular function and reduced valve-related complications. Additionally, repair surgeries preserve native valve tissue and avoid the use of chronic anticoagulation therapy, unlike in replacement procedures, favouring better recovery of cardiac function [
11]. Hence, this meta-analysis investigates whether these benefits persist after a redo mitral valve repair.
While looking at the clinical outcomes across the two groups, there were no significant statistical differences for incidence of stroke, congestive heart failure, requirement of 3rd mitral valve operation, mitral regurgitation and reoperation due to bleeding. These findings suggest MVr and MVR are associated with similar postoperative outcomes. Although mitral valve re-repair presented with a lower incidence of atrial fibrillation as compared to the valve replacement group, a larger patient pool needs to be used to substantiate this claim.
Six studies [
7,
11‐
15] reported no significant difference in incidence of short-term operative mortality (as seen in Fig.
2a) between the MVr and MVR groups. Similarly, there was no significant statistical difference in the long-term operative mortality (seen in Fig.
2b) between the two groups from 4 studies [
7,
9,
11,
14]. Interestingly, the 10 years Kaplan-Meier survival probability (Table
5) post reoperation favoured the mitral valve repair group in
Nishida et al. [
7] and
Zegdi et al. [
11] studies. However, this would require more long-term follow-up data with a larger patient cohort to make more definitive claims.
Table 5
Kaplan-Meier Survival Probability Rates
| 100% | 94% | 100% | 82% | 100% | 82% | 100% | 82% |
| 95% | 87% | 95% | 75% | 95% | 69% | 95% | 65% |
| 76.9% | 58.6% | – | – | – | – | | |
| 96% | 86% | 78% | 68% | – | 63% | – | 53% |
| 96% | 94% | 76% | 60% | – | – | – | – |
Mitral valve repair technique was dependent on the pathology of the mode of failure. The most common method of correction included triangular or rectangular resection and suture repair of the involved portion of the posterior leaflet supplemented by a standard-length flexible posterior annuloplasty band [
7,
16]. Anterior leaflet prolapse was corrected by means of chordal shortening, chordal transfer, or commissural annuloplasty [
16]. However, in recent times, surgeons tend to prefer the use of artificial polytetrafluoroethylene neochordae to repair these anterior leaflet lesions [
7,
16]. In cases of suture dehiscence or ring detachment, surgeons directly re-sutured if the valve leaflet was pliable [
7]. If not pliable, they reinforced it with the autologous pericardium. Edge to edge suturing was used frequently as an additional procedure to correct complicated regurgitation [
13].
On the other hand, mitral valve replacement was performed using biological or mechanical valves or mitral homograft. A trend was observed in a number of studies where surgeons preferred biological valves for MVR patients. In
Zegdi (Late) et al, [
9], 50% of the MVR patients had bioprosthesis, 25% had mechanical prostheses and the remaining 25% had mitral homograft. In
Zegdi et al. [
11], 59% of MVR patients had bioprosthesis, 36% had mechanical prostheses and remaining 5% had mitral homograft. In
Kilic et al. [
14], 92% of the MVR patients had mechanical prostheses while the rest had bioprosthesis. In
Dumont et al. [
17], 52% of MVR patients had bioprosthesis and the rest had mechanical prostheses.
Limitations
Despite the benefits of a pooled analysis, such as higher statistical power, there are several limitations of our current meta-analysis study. Firstly, the retrospective studies included in our meta-analysis carried inherent biases such as selection bias given their observational nature.
The decision to proceed with MVr or MVR depends on various factors including other comorbidities and factors such as age and sex, which lie outside the scope of our univariate analysis. Additionally, some centres might have had monetary considerations/restrictions which would have biased their choice of surgical intervention, along with the surgeon’s experience.
Another limitation we faced was the scarcity of randomised controlled trials available in literature comparing MVr and MVR procedures.
Therefore, further exploration and analysis needs to be performed with a larger patient cohort to provide a more substantial evidence for optimal treatment strategy following a failed mitral valve repair.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.