Outcome after redo-mitral valve replacement in adult patients: a 10-year single-centre experience

Abstract

The aim of this study was to investigate the overall outcome of adult patients undergoing redo-mitral valve replacement (redo-MVR) at our institution. Forty-nine patients (24 males) underwent redo-MVR with either bioprosthetic (n = 24) or mechanical valves (n = 25) between January 2000 and 2010. Median age of patients was 63 years (range 21–80 years), and the mean additive EuroSCORE was 12 ± 4. Median time to re-operation was 8.2 ± 6.6 years for first time redo-MVR and 6.4 ± 5.6 years for second-time redo-MVR. Indications included prosthetic endocarditis (n = 22), para-prosthetic leak (n = 12), structural valve degeneration (n = 8), prosthetic valve thrombosis (n = 6) and malignancy (n = 1). The mean follow-up was 47.5 ± 37.0 months (range 0.1–112.3 months). In-hospital mortality was 12% (n = 6). Mean hospital stay was 17 ± 11 days (range 8–50 days). Actuarial survival at 1 and 5 years was 81 ± 5% and 72 ± 6%, respectively. Three patients required re-intervention: two for prosthetic valve endocarditis and one for para-prosthetic leak. Multivariate analysis showed that overall survival was associated with the LVEF < 50% (P < 0.001), concomitant AVR (P < 0.001) and urgent surgery (P = 0.03).

INTRODUCTION

Since the first valve replacements in the 1950s, major advances have been made in mitral valve (MV) surgical technique, prosthesis design and peri-operative care [1]. Studies comparing percutaneous mitral valvuloplasty and MV repair with mitral valve replacement (MVR) have been published [2, 3]. In general, MV repair is associated with a better outcome than MVR at first-time surgery. However, MVR is still required in patients who have undergone previous MVR, failed MV repair or where MV repair is technically not feasible. Improved survival has inevitably meant that more patients require redo-MVR during follow-up. However, redo surgery may be associated with significant risk, which must be balanced against the benefits to the patient. To avoid the complications of redo-sternotomy such as injury to prior grafts and haemorrhage, right thoracotomy, mini-thoracotomy and port-access surgery have also been applied to redo-MV surgery. Another promising option is trans-apical trans-catheter MV-in-valve implantation, which might offer an alternate and safer approach for high-risk patients. The choice of prosthesis also has implications for potential re-intervention. Nevertheless, there is evidence that clinical outcomes following redo-valve surgery have improved which highlights the velocity of advancement in the field [4]. Contemporary studies that provide knowledge on the operative morbidity and mortality, survival and freedom from re-intervention of patients undergoing redo-MVR with current techniques and prostheses are thus required. In particular, it is necessary to identify the peri-operative variables associated with poor outcome in order to inform surgical decision-making and to offer patients the most appropriate interventions. This study reports a single centre's experience with redo-MVR in adult patients and aims to identify factors that contribute to poor outcome.

METHODS

Patient population

This retrospective study included all patients (n = 49) who underwent redo-MVR with either bioprosthetic or mechanical valves between January 2000 and January 2010. Patients were excluded if they had undergone alternative MV intervention (e.g. MV repair, mitral valvuloplasty, open or closed mitral commisurotomy) in the past without MVR or if they were <18 years of age at the time of re-operation. Patients were identified and data were collected from medical notes and bespoke electronic database. Patient consent was waived by the institutional board.

Surgical technique

On-table transoesophageal echocardiography was used routinely from 2005. Surgery was undertaken through a redo-median sternotomy and cardiopulmonary bypass (CPB) was established with aorto-bicaval cannulation. Where a patent internal mammary to left anterior descending artery graft was present, CPB was established via the femoro-femoral route or the femoral vessels were at least exposed before redo-sternotomy. Intra-aortic filters for cerebral protection (for example, Embol-X system) are not routinely used in our practice. Myocardial protection comprised antegrade cold blood cardioplegia and moderate hypothermia (32°C). Concomitant CABG and/or AVR were performed before MVR, while concomitant TVR was performed after MVR. The left atrium (LA) was opened after developing the inter-atrial groove. Extensive, careful debridement of the calcium (if present) was performed using sharp dissection with a scalpel taking care to avoid the circumflex artery (n = 3). Chords to posterior leaflet if present were left intact. This usually created enough space to fit at least a 25-mm valve. A mechanical or bio-prosthetic valve was then inserted with horizontal mattress pledgeted 2/0 Ethibond sutures. Sutures were placed from LV to LA. Homografts were not used in this series and bovine pericardial patch was used where an abscess cavity was present.

Statistical analysis

Statistical analysis was performed using Stata statistical software (Stata 11, StataCorp LP, College Station, TX, USA) and Microsoft Excel. Data were expressed as frequency with percentage, mean with standard deviation or as median with range, as appropriate. Kaplan–Meier survival analysis was used to evaluate time-dependent events in survival. Univariate analysis was performed to identify risk factors for in-hospital mortality and prolonged hospital stay (>10 days). Continuous variables were compared using the Mann–Whitney test, whereas categorical variables were compared using Fisher's exact test, as appropriate. Risk factors that indicated a relationship with in-hospital mortality and prolonged hospital stay were added into multivariate analysis using exact logistic or linear regression, as appropriate. Cox proportional hazards regression was used to control for covariates of overall survival. A P-value < 0.05 was considered statistically significant.

RESULTS

The mean age of the entire cohort was 63 ± 13 years (range 21–80 years) and the mean additive EuroSCORE was 12 ± 4. Other pre-operative and operative characteristics are given in Tables 1 and 2, respectively. The in-hospital mortality was 12% (n = 6). Causes of in-hospital mortality included cardiac (n = 2), multi-organ dysfunction (n = 2), stroke (n = 1) and respiratory failure (n = 1). Post-operative characteristics are listed in Table 3. Follow-up was complete for the 49 patients in this study for a mean of 47.5 ± 37 months (range 0.1–112.3 months). Univariate analysis demonstrated that in-hospital mortality was associated with pre-operative LVEF <50% (P = 0.04) (Table 4). Age, gender, indication for surgery, type of previous prosthesis and concomitant procedures had no impact on in-hospital mortality. The mean pre-operative EuroSCORE (P = 0.002) was greater in those with complications, while complications were more frequent in patients who underwent concurrent AVR (P = 0.02).

Table 1:

Pre-operative characteristics of the whole cohort

Pre-operative characteristic	Number (n = 49)
Gender
Male	24
Mean age in years (range)	63 ± 13 years (21–80 years)
LVEF <50%	14
Mean additive EuroSCORE	12 ± 4
Previous MVR
Once	43
Twice	5
Thrice	1
Bioprosthetic	24
Mechanical	25
Time to re-operation
First time redo-MVR	8.2 ± 6.6 years
Second time redo-MVR	6.4 ± 5.6 years
Concomitant procedures performed at the time of previous MVR
AVR	10
CABG	5
TVR	4
Indications for re-operation
Prosthetic valve endocarditis	22
Paravalvular leak	12
Structural valve degeneration	8
Prosthetic valve thrombosis	6
Malignancy	1
Haemodynamic pathology
Mitral regurgitation	37
Mitral stenosis	7
Mixed	5

LVEF: left ventricular ejection fraction; MVR: mitral valve replacement; AVR: aortic valve replacement; CABG: coronary artery bypass grafting; TVR: tricuspid valve repair.

Table 2:

Operative characteristics of the whole cohort

Operative characteristic	Number (n = 49)
Urgency of surgery
Elective	20
Urgent	20
Emergency	9
Cardiopulmonary bypass time	120 ± 56 min
Cross-clamp time	92 ± 32 min
Concomitant procedures	20
AVR	13
TVR	4
CABG	2
AVR + TVR	1
Median prosthesis size	29 mm (range: 25–33 mm)
Types of prosthesis
Bioprosthetic (%)	14 (29)
Mechanical (%)	35 (71)

AVR: aortic valve replacement; CABG: coronary artery bypass grafting; TVR: tricuspid valve repair; AVR + TVR: aortic valve replacement plus tricuspid valve repair.

Table 3:

Post-operative characteristics and complications

Post-operative characteristic	Number (n = 49)
In-hospital mortality (%)	6 (12)
Complications (%)
Re-exploration for bleeding	2 (4)
Sepsis	9 (18)
Supra-ventricular arrhythmias	18 (36)
Permanent pacemaker	8 (16)
Haemofiltration	6 (12)
Cerebrovascular event	4 (8)
Mean hospital stay in days (range)	17 ± 11 (8–50)
Follow-up in months (range)	47.5 ± 37 (0.1–112.3)
Mortality during follow-up (%)	14 (28)
Re-operation during follow-up	3
Endocarditis	2
Para-prosthetic leak	1

Table 4:

Univariate analysis for in-hospital mortality

	In-hospital mortality
Variable	Yes	No	P-value
Mean age (years)	64.7 ± 7.3	62.8 ± 13.6	0.87*
Gender
Male	4	20	0.41**
Female	2	23
EuroSCORE	13.2 ± 3.1	11.8 ± 3.7	0.31*
Parsonnet score	39.7 ± 17.9	31.1 ± 13.8	0.32*
Years from last MVR	5.5 ± 6.3	8.4 ± 6.5	0.29*
Surgical priority
Elective	0	9	0.67**
Urgent	3	17
Emergency	3	16
Salvage	0	1
Indication for surgery
Infective endocarditis	4	18	0.73**
Paravalvular leak	1	11
Valve degeneration	0	8
Valve thrombosis	1	5
Malignancy	0	1
MV pathology
Regurgitation	6	31	0.78**
Stenosis	0	7
Mixed	0	5
Pre-operative LV dysfunction
Normal/Mild	2	33	0.04**
Moderate/Severe	4	10
AVR
Yes	2	11	0.65**
No	4	32
TV procedure
Yes	1	3	0.41**
No	5	40
History of CAD
Yes	2	5	0.19**
No	4	38
Previous MVR prosthesis
Tissue	2	13	1.00**
Mechanical	4	30

MVR: mitral valve replacement; LV: left ventricle; AVR: aortic valve replacement; TV: tricuspid valve; CAD: coronary artery disease.

*Mann–Whitney U test.

**Fisher's exact test.

On univariate analysis (Table 5), hospital stay was longer in patients with a higher pre-operative EuroSCORE (P = 0.04). Age, gender, indication for surgery, type of prosthesis, concomitant procedures and other complications had no impact on hospital stay. The hospital stay was higher in the endocarditis group (23 days) versus the non-endocarditis group (14.2 days; P = 0.05). The following variables were included in the multivariate models: age, gender, pre-operative EuroSCORE, operative priority (elective, urgent, emergency or salvage), pre-operative LV function, MV pathology (stenosis, regurgitation or mixed), indication for surgery, type of previous MV prosthesis (tissue versus mechanical) and concomitant procedures. The dependent variables were in-hospital mortality, prolonged hospital stay and overall survival. There were no independent predictors of in-hospital mortality or prolonged hospital stay. Overall survival was significantly associated with pre-operative LVEF <50% (P = 0.001), concomitant AVR (P < 0.001) and urgent surgery (P = 0.03). Nine patients (40%) in the endocarditis group died when compared with five patients (18.5%) in the non-endocarditis group (P = 0.11). Eight more patients died after discharge from hospital during follow-up and actuarial survival was 81 ± 5 and 72 ± 6% at 1 and 5 years, respectively (Fig. 1). Causes of death included cardiac (n = 3), cerebral tumour (n = 1), multiple myeloma (n = 1), advanced colon cancer (n = 1), pneumonia (n = 1) and multi-organ dysfunction (n = 1). Three patients required re-intervention, two for prosthetic valve endocarditis and one for para-prosthetic leak (the initial indication for redo-MVR was endocarditis in all three).

Table 5:

Univariate analysis for prolonged hospital stay

	Prolonged hospital stay >10 days (n = 42)
Variable	Yes	No	P-value
Mean age (years)	62.3 ± 14.5	64.8 ± 7.9	0.87*
Gender
Male	14	6	0.47**
Female	18	4
EuroSCORE	12.2 ± 3.6	9.8 ± 2.2	0.04*
Parsonnet score	34.2 ± 15.6	24.4 ± 12.0	0.05*
Years from last MVR	5.5 ± 6.3	8.4 ± 6.5	0.29*
Surgical priority
Elective	6	2	0.23**
Urgent	12	7
Emergency	13	1
Salvage	1	0
Indication for surgery
Infective endocarditis	15	2	0.48**
Paravalvular leak	8	4
Valve degeneration	5	2
Valve thrombosis	4	2
Malignancy	0	0
MV pathology
Regurgitation	25	9	0.81**
Stenosis	4	1
Mixed	3	0
Pre-operative LV dysfunction
Normal/mild	8	4	0.43**
Moderate/severe	24	6
AVR
Yes	11	1	0.23**
No	21	9
TV procedure
Yes	2	1	1.00**
No	30	9
Previous MVR prosthesis
Tissue	8	5	0.22**
Mechanical	21	4

MVR: mitral valve replacement; LV: left ventricle; AVR: aortic valve replacement; TV: tricuspid valve.

*Mann–Whitney U test.

**Fisher's exact test,

Figure 1:

Kaplan–Meier plot and survival table for all patients who underwent redo-mitral valve replacement. Cumulative survival is plotted on y-axis.

DISCUSSION

Recent years have witnessed significant improvements in the clinical and functional outcomes of patients undergoing revision valvular surgery [4]. Despite this, redo-MVR remains a challenge. Taking account of the fact that patients who undergo MVR are surviving longer and are thus at an increased risk of prosthesis failure or valve-related complications, it is unsurprising that a greater number of patients are requiring redo-MVR. Consequently, there is a need for further information on the outcomes of this group of patients to help in surgical decision-making and aid in the choice between conservative, percutaneous or open surgical procedures. This study highlights a single-centre experience with redo-MVR over a 10-year period and has established the risk factors in-hospital mortality, prolonged hospital stay and overall survival.

Studies reporting mortality figures for revision valve surgery often do not discriminate between the anatomical position of the valve, with results regularly being combined for aortic, mitral and tricuspid valve replacements [5–10]. These studies have shown operative mortality rates between 1.3 and 17.4%. Other studies have also included patients who previously underwent MV procedures other than replacement (e.g. MV repair and mitral valvuloplasty) [5, 11]. Heterogeneity of outcome reporting in surgical trials is a major problem for cross-study comparisons, due to the propensity for outcome reporting bias and the inability to undertake meta-analyses. Our in-hospital mortality is in accordance with the recent literature [4, 7, 12, 13], despite 37% of our patients undergoing concomitant procedures. The cause of re-operation in bioprosthetic valves was structural deterioration in a third of our patients, which is in between 24 [4] and 46% [13] reported in other series. Overall, endocarditis was the commonest cause of re-operation, 60% for mechanical valves and 29% for bioprosthetic valves which is much >6% that is reported in the literature [8, 13]. This is likely due to improvements in valve technology and the nature of the local population in some parts of the South-West of England. Others have also shown that in the setting of redo valve replacement, endocarditis is a more frequent cause of re-operation in patients with mechanical when compared with bioprosthetic valves [9].

Like other groups [5, 12], we have also shown that the incidence of complications such as supra-ventricular arrhythmias, sepsis, acute renal failure requiring renal replacement therapy and stroke after redo-MVR is significant. This is likely to be related to the higher risk profile of the patients. For example, almost 30% of our patients had LVEF <50%, and the mean pre-operative additive EuroSCORE was 12 ± 4. The association of complications with higher additive EuroSCORE was demonstrated by univariate analysis in our series. We have also shown that the LVEF <50% is an independent predictor of mortality in the short-term as well as in the mid-term. Some authors recommend early intervention in relatively asymptomatic patients with failing MV prostheses because this may help in preventing irreversible myocardial damage, deteriorating the LVEF <50% and the inherent operative risk [5, 14]. Prolonged hospital stay is poorly described in studies of redo MVR [6, 11]. In this study, we have not identified any pre-operative variables predictive of prolonged hospital stay. It is quite conceivable that this is highly dependent on the immediate post-operative course and the occurrence of complications.

The factors that affect the potential outcome of patients who undergo redo-valve surgery are numerous and many confounding variables exist, which hinders the statistical rigour of univariate analyses. For this reason, studies of redo-MVR need to control for confounding variables using multivariate methods. The dependent variables examined in our study were in-hospital mortality, prolonged hospital stay and overall survival. Other studies have also reported multivariate outcomes for patients undergoing redo-MVR [4, 5, 7, 10]. Early mortality has been associated with older age [4, 7, 10], female gender [5], advanced NYHA class [5, 10], low left ventricular ejection fraction (<35), increased left ventricular end diastolic diameter (>50 mm), pulmonary oedema, urgent operations [4, 5, 13], concomitant procedures [7, 11] and previous myocardial infarction [13]. On review of the published results, pre-operative impairment of the LVEF remains the most consistent risk factor for early and overall mortality following redo-MVR. This highlights the need for early identification and intervention in patients with MV prosthetic dysfunction to avoid the deleterious and lasting effects of damage to the myocardium. The 5-year survival of 72.8 ± 6.5% in this series is comparable with figures of 65–89% [5, 6, 10, 14] published in the literature. Freedom from re-operation for prosthetic valve dysfunction has also been shown to be good with authors rarely reporting the need for re-intervention at medium-term follow-up [6]. This emphasizes that a good long-term outcome is achievable after redo-MVR.

Finally, transcatheter valve implantation for bioprosthetic valves (valve-in-valve) and minimally invasive techniques are emerging as attractive alternatives to conventional surgical valve replacement in patients requiring redo-MVR. For first-time replacement, experience is rapidly increasing with the implantation of bioprosthetic valves with the earlier-mentioned techniques. These implants are expected to degenerate over time and the same techniques may be applied to the redo-setting where the morbidity and mortality can be considerably higher than those for first-time valve replacement. Off-label implantation of devices for valve-in-valve is gaining popularity but results from larger series are awaited. This study is retrospective and must be interpreted in light of this limitation. However, studies that provide prognostic information and identify risk factors for short- and mid-term outcomes in patients undergoing redo-MVR are warranted to inform surgical decision-making. This study also does not provide any information on the functional outcomes and quality-of-life of patients in our cohort. A prospective evaluation using appropriate patient-reported outcome measures would be required for this. In conclusion, we have demonstrated that redo-MVR can be conducted safely and with good overall outcome. Hence, early intervention before ventricular dysfunction ensues should be considered.

Conflict of interest: none declared.