Article Text
Abstract
Context There is little evidence guiding heart valve prosthesis selection in patients with end-stage renal disease (ESRD) on dialysis.
Objectives To perform: 1) a systematic review of studies examining valve replacement in patients with ESRD on dialysis; and 2) a quantitative meta-analysis comparing survival and valve-related outcomes following valve replacement with bioprostheses versus mechanical prostheses in this population.
Data sources English studies published from 1990 onwards.
Study selection Studies were included in the meta-analysis if they compared bioprostheses with mechanical prostheses in patients with ESRD on dialysis.
Data extraction Extracted summary estimates included the hazard ratio (HR) for death, and the odds ratio (OR) for developing valve-related complications due to the use of bioprostheses versus mechanical prosthesis.
Results Twelve studies published from 1997 to 2010 were included in this review, of which 9 were used in the meta-analysis. No evidence of publication bias was detected. The aortic valve was the most common valve replaced in these studies (4339/6350), although 11 of the 12 studies also included mitral or multiple valve replacements. No difference in survival was observed between valve types (bioprostheses versus mechanical prostheses hazard ratio 1.3, 95% confidence interval (CI) 1.0-1.9, p=0.09). However, valve replacement with bioprostheses was associated with fewer valve-related complications compared to mechanical prostheses (odds ratio 0.4, 95% CI 0.2-0.7, p=0.002).
Conclusions A meta-analysis of the published literature demonstrates no survival difference following valve replacement with either bioprostheses or mechanical prosthesis in patients with ESRD on dialysis. Bioprosthetic valve replacement was associated with fewer valve-related complications. Although this meta-analysis cannot discriminate between the sites of valve implant, these data can likely be extended to include at least aortic valve replacement.
- Valvular disease
- prosthetic heart valves
- surgery-valves
- cardiac surgery
- surgery-valve
- surgery-coronary bypass
- mitral regurgitation
- mitral valve prolapse
- cardiopulmonary bypass
- coronary bypass surgery
- cardiovascular surgery
- statistics
- surgery (adult)
- valvular heart disease
- VHD
- surgery CAD
Statistics from Altmetric.com
- Valvular disease
- prosthetic heart valves
- surgery-valves
- cardiac surgery
- surgery-valve
- surgery-coronary bypass
- mitral regurgitation
- mitral valve prolapse
- cardiopulmonary bypass
- coronary bypass surgery
- cardiovascular surgery
- statistics
- surgery (adult)
- valvular heart disease
- VHD
- surgery CAD
Introduction
Heart valve disease is common in patients with end-stage renal disease (ESRD) who require dialysis.1 Notably, the rise in the number of patients treated with dialysis over the past decade has been accompanied with a rise in the number of dialysis patients referred for heart valve surgery.1
Previous guidelines from the American College of Cardiology and American Heart Association have recommended mechanical prostheses for patients with ESRD receiving dialysis who require heart valve replacement.2 This recommendation was based on the concern of accelerated calcification of bioprosthetic valves in patients with ESRD.3 However, several studies have subsequently described favourable outcomes following valve replacement with bioprostheses.4–12 Consequently, the updated guidelines no longer advocate any specific prosthesis type,13 and individualised prosthesis selection is encouraged.
Remarkably, there is little evidence guiding prosthesis selection in patients receiving dialysis. Published studies are retrospective and have involved small numbers of patients.4–9 11 12 The largest study included 5858 patients; however, valve-related complications were not described and follow-up was short.10
To our knowledge, no formal meta-analysis of the published literature has yet been made. This is probably owing to the varying clinical end points reported, which make quantitative synthesis challenging. However, statistical methods exist by which to extract hazard estimates from actuarial curves and these can be helpful in summarising this and other similarly disparate literature.14 Although studies describing outcomes following valve replacement with ESRD predominantly involved aortic valve replacement, these studies included mixed cohorts comprising mitral and multiple valve replacement.4–10 12
Despite the limitations of each individual study, a qualitative assessment of the published literature shows poor long-term survival for patients with ESRD.4–12 However, valve-related complications with mechanical prostheses appear more common in this population.4–12 We therefore performed a quantitative analysis of the published literature describing valve replacement in patients with ESRD undergoing chronic dialysis. The objectives of this study were to perform (1) a systematic review of studies examining valve replacement in patients with ESRD receiving dialysis; and (2) a quantitative meta-analysis comparing survival and valve-related outcomes following valve replacement with bioprostheses versus mechanical prostheses in this population.
Methods
Search procedure
We searched Medline, Google Scholar and the Cochrane Controlled Trials Registry, using combinations of the terms ‘chronic renal failure’, ‘dialysis’, ‘end-stage renal disease’, ‘valve replacement’, ’heart’, or ‘prosthesis’ both as text words and as MESH headings. Searches were restricted to English studies published from 1990 onwards, in order to include contemporary surgical cohorts. Abstracts were not included. Relevant studies were also identified from the reference lists of selected articles and from review articles.
Study selection
Studies were included in the meta-analysis if they compared bioprostheses with mechanical prostheses in patients with ESRD receiving dialysis. Studies were excluded if they involved patients with acute renal failure, or if they did not describe outcomes according to prosthesis type (table 1). The search strategy was repeated by two individuals and the inter-reviewer agreement was expressed as a κ value. Study quality was quantified by the Newcastle–Ottawa Scale for epidemiological studies (table 2).15
Search results
A total of 334 publications were retrieved from the initial search strategy (figure 1). Two hundred and eighty studies were excluded because they did not contain data pertaining to this study's objectives. Another 42 studies were excluded because they involved patients with acute renal failure or did not compare outcomes according to prosthesis type. In the remaining 12 studies, three primarily involved mechanical prostheses. These three studies are discussed as part of this article's systematic review section, but given the lack of a comparison group, they were not included in the meta-analysis.16–18
Outcome measures
Extracted summary estimates included the HR for death and the OR for developing valve-related complications due to the use of bioprostheses versus mechanical prostheses. Valve-related complications were defined according to pre-established criteria.19 Briefly, these included structural valve deterioration, non-structural dysfunction, bleeding, thromboembolism and prosthetic valve endocarditis.19
Statistical analyses
Publication bias was investigated by a funnel plot, in addition to the calculation of Kendall's τ coefficient (with and without continuity correction) and of Egger's regression intercept. The HR for death associated with valve replacement using bioprostheses was estimated from study-specific actuarial survival curves,14 and pooled according to a random-effects model. The OR for developing valve-related complications associated with the use of bioprostheses was also calculated using a random-effects model. Cochrane's Q and I2 statistics were used to evaluate heterogeneity between studies.20 A sensitivity analysis was conducted with the leave-out-one study method.21 A meta-regression was performed to determine sources of study heterogeneity. Statistical analyses were performed with Comprehensive Meta-Analysis 2.0 (Biostat, Englewood, New Jersey, USA) and statistical significance was set at p<0.05.
Results
Meta-analysis
Study characteristics
Nine studies were included in the meta-analysis. These nine studies were retrospective series conducted in the USA, Canada and Japan. Duration of follow-up ranged from 1.610 to 12.311 years. Four of these studies involved cohorts who predominantly received bioprostheses,4 7 8 12 whereas five studies involved cohorts who predominantly received mechanical prostheses.5 6 9–11 Each of the studies involved aortic valve replacement,4–12 although eight out of the nine studies also included patients who underwent mitral and multiple valve replacement.4–11 In these nine studies, there were 3741 aortic valve replacements and 598 aortic and mitral valve replacements in 6350 patients; therefore aortic valve replacement was performed in 68% of the total population.4–12
Overall study quality was acceptable (table 2). A common weakness of these studies was the lack of randomisation in assigning heart valve prostheses. Some studies stated that patients with expected poor survival were given bioprostheses.5 11 The completeness of follow-up was good across studies.4–12 However, late follow-up was available for only 46/72 (64%) of patients in the study conducted by Brinkman et al.5
A HR for death was calculated for eight of the nine studies.4 5 7–12 The exception was the study by Lucke et al, who described the outcomes of 19 patients, nine of whom received bioprostheses.6 A HR could not be calculated for this study because the actuarial survival curve was not separated according to prosthesis type.
Eight studies were used to calculate the OR for the development of valve-related complications according to prosthesis type.4–9 11 12 The remaining study did not describe valve-related complications.10
Assessment of publication bias
A funnel plot analysis revealed mild asymmetry. Kendall's τ, with and without continuity correction (both p=0.1) and the Egger's regression intercept (p=0.2) were not statistically significant; therefore, no evidence of publication bias was detected.
Survival
There was no difference in survival between valve types when assessed by a random-effects model (bioprostheses versus mechanical prostheses HR=1.3, 95% CI 1.0 to 1.9, p=0.09) (figure 2A). The HR for death due to bioprostheses in the individual studies ranged from 0.612 to 4.3.11 Sensitivity analysis showed that the summary estimate did not appreciably change when any one study was removed from the meta-analysis. Study heterogeneity was detected (Q-statistic 20, p<0.05, I2 67%) and appeared to be related to the proportion of aortic valve replacements (p=0.05) and also to the proportion of bioprostheses used in each study (p=0.05).
Valve-related complications
Overall, 125 valve-related complications19 were reported in the eight studies included in this meta-analysis.4–9 11 12 Bleeding was the most common complication, which was seen in 63 patients, of whom 52 received mechanical prostheses.4–9 11 12 The most commonly reported bleeding sites were cerebral and gastrointestinal.6 7 9 11 12 Thromboembolic events were reported in 35 patients, of whom 31 received mechanical prostheses. Notably, most of these studies did not further separate bleeding or thromboembolic complications according to the site of valve implantation.5–8 11 12 Chan et al reported that two thromboembolic events occurred in patients who received bioprostheses and who had undergone isolated aortic valve replacement and combined aortic and mitral valve replacement, respectively.4 The study by Tanaka et al included only patients who received aortic valve replacement.11 Prosthetic valve endocarditis was reported in 12 patients, in whom valve explant was subsequently performed in nine.4 5 11 Structural valve deterioration was reported in five patients who received bioprostheses.4 5 12 Brinkman et al reported two cases of structural deterioration requiring reoperation 15.2 months after aortic valve replacement and 54.6 months after mitral valve replacement.5 Chan et al reported a case of structural deterioration requiring reoperation 95.5 months after mitral valve replacement,4 and Kaplon et al also reported a case of structural deterioration requiring reoperation 10 months after mitral replacement.7 Lucke et al reported a case of aortic porcine valve failure after 156 months, which was managed without reoperation.6
Valve-related complications occurred less often after valve replacement with bioprostheses OR=0.4, 95% CI 0.2 to 0.7, p=0.002 via a random-effects model) (figure 2B). Removal of any one study did not significantly change the overall summary estimate. Study heterogeneity was not significant (Q statistic 10, p=0.2, I2 33%).
Systematic review
Three studies described clinical outcomes after mechanical valve replacement in patients with ESRD receiving dialysis. Ura et al described the outcomes of 12 patients with ESRD receiving dialysis who underwent valve replacement with the St Jude Medical valve.16 One death occurred within 30 days of surgery and two additional deaths were seen in late follow-up (mean 37.1 months). In another study, Kato et al presented their results with valve replacement in 27 patients; three of whom received bioprostheses.17 Over the follow-up period, four bleeding events occurred in the 24 patients who received mechanical prostheses, compared with one bleeding event in the three patients who received bioprostheses. Gultekin described the outcomes of 29 patients with ESRD receiving dialysis who underwent 30 (29 mechanical prostheses and one bioprosthesis) heart valve replacement operations.18 Thirty-day mortality was 3.4%, whereas 5-year actuarial survival was 47%.
Discussion
In this study, we summarised the literature describing prosthesis selection in patients with ESRD undergoing chronic dialysis. Of the nine studies included in this meta-analysis, eight compared survival with bioprostheses and mechanical prostheses,4 5 7–12 and eight compared valve-related complications for the two valve types.4–9 11 12
This meta-analysis showed no survival difference according to prosthesis type. Notably, two of the eight studies demonstrated a survival advantage in favour of mechanical prostheses over bioprostheses.4 11 The remaining six studies did not demonstrate a survival difference according to prosthesis type. This meta-analysis summarised overall crude survival from the individual studies; therefore, differences in survival may be influenced by unadjusted factors. Patients who received bioprostheses, however, were older,4 5 8 9 11 and had undergone dialysis for longer preoperatively9 than patients receiving mechanical prostheses. They were also more likely to have had a previous myocardial infarct or lower ejection fraction preoperatively,4 8 receive concomitant coronary artery bypass grafting,4 undergo double valve replacement9 and have a low indexed effective orifice area after valve replacement11 than patients who received mechanical prostheses. Therefore, survival should have been biased in favour of mechanical prostheses. Overall, the patients in each of the individual studies included in this meta-analysis were relatively young. Renal disease aside, current evidence also supports lowering traditional age limits for bioprostheses implantation.22
Valve-related complications occurred less often in patients who received bioprostheses.4–9 11 12 As with survival, crude estimates of valve-related complications were used in this meta-analysis, as adjusted values were not available for each study. Bleeding complications occurred most often, followed by thromboembolic complications.4–9 11 12 Valve-related reoperation was most often due to prosthetic valve endocarditis.4–9 11 12 Overall, four cases of structural valve deterioration requiring reoperation were reported, ranging from 10–96 months after the initial valve surgery.4–9 11 12 The small number of structural failures may be the result of the overall poor survival of patients in this population.1 Thus, bleeding and thromboembolic events were more commonly seen as these events can occur early after valve replacement.19
Limitations
Unadjusted summary estimates were used in this meta-analysis as adjusted measures were not available for each study. Although comorbid factors should have biased the results in favour of mechanical prostheses, it is possible that unknown confounders might have influenced outcomes. HRs were subsequently extracted from actuarial curves by methods that assume a constant rate of censoring.14 Also, valve-related complications were expressed as an OR; therefore, the timing of events could not be taken into account. It is important to note that the mean follow-up of each study was relatively short, with a range of 1.610 to 4.19 years; therefore, conclusions about the long-term performance of bioprostheses in patients with ESRD cannot be drawn. This relatively short period of follow-up is also influenced by the overall poor life expectancy of patients with ESRD.1 The influence of the different causes of renal failure cannot be determined. Specifically, the study by Tanaka et al included fewer patients with hypertension, diabetes and peripheral vascular disease than the other studies included in the meta-analysis, which included patients from North America. Also, the impact of prosthesis type and the location of implant (ie, aortic versus mitral) could not be determined since the outcomes of each study were not separated accordingly. The studies included in this meta-analysis, however, predominantly involved aortic valve replacement; and therefore, conclusions in this study can probably be extended to include at least aortic valve replacement.4–12 No prospective randomised trials were available for inclusion in this meta-analysis.
Conclusion
This meta-analysis establishes that the mid-term survival of patients with ESRD receiving dialysis is not related to valve type. Patients who receive bioprostheses, however, were less likely to develop valve-related complications; therefore, bioprostheses are recommended for patients with ESRD receiving dialysis requiring heart valve replacement. Although this meta-analysis does not discriminate between the sites of valve implant, these data appear relevant in patients with ESRD who require at least aortic valve replacement.
References
Footnotes
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.