Objective Bioprosthetic valves are being used with increased frequency for valve replacement, with controversy regarding risk:benefit ratio compared with mechanical valves in younger patients. However, prior studies have been too small to provide comparative estimates of less common but serious adverse events such as infective endocarditis. We aimed to compare the incidence of infective endocarditis between bioprosthetic valves and mechanical valves.
Methods We searched PubMed, Cochrane, EMBASE, Scopus and Web of Science from inception to April 2018 for studies comparing left-sided aortic and mitral bioprosthetic to mechanical valves for randomised trials or observational studies with propensity matching. We used random-effects model for our meta-analysis. Our primary outcome of interest was the rate of infective endocarditis at follow-up.
Results 13 comparison groups with 43 941 patients were included. Mean age was 59±7 years with a mean follow-up of 10.4±5.0 years. Patients with bioprosthetic valves had a higher risk of infective endocarditis compared with patients receiving mechanical valves (OR 1.59, 95% CI 1.35 to 1.88, p<0.001) with an absolute risk reduction of 9 per 1000 (95% CI 6 to 14). Heterogeneity within the included studies was low (I2=0%). Exclusion of the study with maximum weight did not change the results of the analysis (OR 1.57, 95% CI 1.14 to 2.17, p=0.006). A meta-regression of follow-up time on incidence of infective endocarditis was not statistically significant (p=0.788) indicating difference in follow-up times did not alter the pooled risk of infective endocarditis.
Conclusions Bioprosthetic valves may be associated with a higher risk of infective endocarditis. These data should help guide the discussion when deciding between bioprosthetic and mechanical valves in individual patients.
- valve disease surgery
- valvular heart disease
- systemic review
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Mechanical valves have been in use for valve replacement for several decades since the early 1950s.1 2 They have well-established efficacy and durability, but have heightened thrombogenicity requiring use of life-long anticoagulation and its associated risks. On the other hand, bioprosthetic valves do not require anticoagulation, but carry limited durability with most younger patients requiring repeat reoperation over their life time. In addition, there is increasing awareness of valve thromboses occurring on bioprosthetic valves.3–5
The 2017 American College of Cardiology/American Heart Association6 valve guidelines recommend bioprosthetic valves in patients >70 years of age and mechanical valves in patients <50 years of age with increasing uncertainty in the intermediate age groups. But despite these recommendations, there has been an increasing use of bioprosthetic valves in younger patients over the last few decades.7–9 The complex risk–benefit profile of reoperation, bleeding and thrombosis between mechanical valves and bioprosthetic valves has therefore led to continued uncertainty when there is clinical equipoise. Adding to this complexity, recent studies have suggested a possible increased risk of endocarditis with bioprosthetic valves10 11 which could further shift the risk:benefit ratio to favour mechanical valve replacement. We therefore performed a systematic review and meta-analysis of all available randomised controlled trials (RCTs) and propensity-matched studies to compare the risk of incident endocarditis and associated outcomes between mechanical valves and bioprosthetic valves.
Data source and search strategy
This systematic review and meta-analysis was reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. Literature search was performed with the help of an expert medical reference librarian (PJE). We searched PubMed, EMBASE, Cochrane, Scopus and Web of Science databases for RCTs and propensity-matched trials published from January 2000 to April 2018 using the following search terms: “infective endocarditis”, “mechanical valve”, “aortic valve”, “mitral valve”, “bioprosthetic valve”, “mortality” and their combinations. We limited the search to English language and studies including adult population only. We also searched ClinicalTrials.gov and reviewed the reference list of relevant articles. Conference abstracts, case reports and small case series were excluded. A detailed description of our search strategy from PubMed is included in online supplementary appendix 1. PRISMA search strategy is shown in online supplementary figure 1.
To be eligible, studies had to meet the following eligibility criteria: (1) (i) RCT or (ii) if an observational study, the analysis must have performed propensity matching to minimise bias in choice of prosthetic valve based on comorbidities; (2) age >18 years; (3) compare left-sided bioprosthetic valves to mechanical valves; (4) report the estimate of relative risk (RR) with 95% CI, or other measures of RR such as HR, OR or provide other forms of data from which OR could be calculated. Online supplementary table 1 shows the various covariates used for propensity matching in the observational studies.
Two reviewers (MAN and YNVR) independently reviewed the abstracts, study titles and selected full-length articles identified by the aforementioned search strategy. The reviewers also independently abstracted the study characteristics, designs, methods and other relevant outcomes. Any discrepancy between the first and second authors was resolved by consensus or by consulting with a third reviewer (VS). Risk of bias was assessed by a third reviewer (VS) using the Cochrane risk of bias tool (https://methods.cochrane.org/bias/resources/rob-2-revised-cochrane-risk-bias-tool-randomized-trials) for RCTs (online supplementary table 2A) and using the Newcastle-Ottawa Scale (NOS) (http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp) for non-randomised studies (online supplementary table 2B). The Cochrane risk of bias tool includes a fixed set of domains of bias that focuses on the various aspects of trial design, conduct and reporting. Within each domain, there is a list of signalling questions and the overall bias is stated as low, high or unknown risk of bias. The NOS is a tool available to assess quality of non-randomised studies. It judges a study based on three broad domains: selection of the study groups, comparability of the groups, and ascertainment of outcome and exposure. The scale rates the quality of studies as good, fair and poor.
This study included adult patients receiving left-sided mechanical valves or bioprosthetic valves as listed in table 1. We excluded studies of valve replacement with the indication for surgery being infective endocarditis as this may confound the risk of recurrent endocarditis and choice of valve. We excluded right-sided valves and included only studies with left-sided heart valves. When studies had more than two comparison groups, for example, two mechanical valve types (mechanical valve 1 and mechanical valve 2) compared against a bioprosthetic valve, we analysed the subgroups individually (mechanical valve 1 vs bioprosthetic valve and mechanical valve 2 vs bioprosthetic valve). When data for a variable of interest were not available, we manually extracted the numbers from Kaplan-Meier curve using methods previously described.12
Our primary outcome was risk of incident infective endocarditis between mechanical valves and bioprosthetic valves. Secondary outcome was 30-day mortality.
Patient and public involvement
This study does not involve direct patient participation and therefore this is not applicable to our study.
Categorical data were pooled using a random-effects model, with the pooled effect size represented as OR with a 95% CI limit. We also calculate risk difference (RD). To do this, the authors calculated the baseline risk (BR) in the control group (number of events in the control group/total sample size) and obtained pooled risk ratio (RR) with 95% CI using the formula OR/((1−BR)+(OR×BR)). We then calculated experimental event rate as EER=RR×BR. From this, we calculated the RD per 1000 as (1000×(EER−BR)) (https://handbook-5-1.cochrane.org/chapter_11/11_5_5_statistical_considerations_in_summary_of_findings.htm).
Publication bias was assessed visually using funnel plot. Cochrane’s Q-statistics were used to determine the heterogeneity of included studies for each outcome. Heterogeneity was calculated using prediction intervals as suggested by Borenstein et al. I2 values of <25%, 25%–50% and 50%–75% were considered as low, moderate and high heterogeneity, respectively. An exclusion sensitivity analysis was included to assess heterogeneity for the primary end point of interest. A meta-regression was performed when necessary to analyse the impact of moderator variables on outcomes of interest, especially the follow-up period on incidence of infective endocarditis. A p value of <0.05 was considered statistically significant. We also performed a sensitivity analysis that excluded moderate and high-risk bias studies. Analyses were performed by MAN using the software Comprehensive Meta-analysis (V.3.3). This study was exempt from Institutional Review Board approval at our institution. The study methods have been validated and published in a prior paper.13
The initial search identified 221 studies for abstract screening (online supplementary figure 1). After manually reviewing abstracts using the Rayyan software (https://rayyan.qcri.org/reviews), 63 studies were identified for full-text review, which were independently reviewed by two authors (MAN, YNVR). No disagreement on exclusion was noted after review by two authors. The final analysis included 1212 14–24 studies with one subgroup from a study,17 thus providing 13 comparison groups. Among the included studies, there were three randomised trials17 19 22 comparing mechanical valves to bioprosthetic valves while the remaining studies performed a propensity-matched analysis. Risk of bias was assessed using Cochrane risk of bias assessment tool for randomised trials (online supplementary table 2A) and using the NOS score for observational studies (online supplementary table 2B along with the online supplementary appendix). A majority of the propensity-matched observational trials had three or four green boxes in selection domain, one or two green boxes in comparability domain, and two or three green boxes in outcome/exposure domain which indicates good quality study selection. This indicates low risk of bias.
The overall study population consisted of 43 941 patients from 13 comparison groups (12 original studies and 1 subgroup) and about 60% were men with a mean age of 59±7 years. Mean follow-up time was 10.4±5.0 years with a maximum follow-up of 20 years.
Risk of incident infective endocarditis was analysed from 13 comparison groups. Incidence of infective endocarditis was significantly higher in patients who received bioprosthetic valves when compared with patients receiving mechanical valves (OR 1.59, 95% CI 1.35 to 1.88, p<0.001) (figure 1). Heterogeneity within the included studies was low (I2=0%). Funnel plot of included studies is shown in figure 2. Exclusion of the study by Brennan et al 16 with maximum weight did not change the results of the analysis with bioprosthetic valve still showing higher rates of infective endocarditis compared with mechanical valves (OR 1.57, 95% CI 1.14 to 2.17, p=0.006). Inclusion of only studies with lower risk of bias12 17 20 21 still showed higher rates of infective endocarditis with bioprosthetic valves compared with mechanical valves (OR 1.96, 95% CI 1.28 to 3, p=0.002). Calculated absolute RD was 9 (95% CI 6 to 14) indicating that if 1000 patients received mechanical valves, 9 cases of infective endocarditis would be prevented. A meta-regression of follow-up time on incidence of infective endocarditis (figure 3) was not statistically significant (p=0.788) indicating difference in follow-up times does not alter the pooled risk of infective endocarditis. In the original study by Hammermeister et al,17 rates of endocarditis were not statistically different between bioprosthetic valves and mechanical valves in aortic position using the log-rank test. This differed from our extracted study level summary estimate without accounting for censoring.
Thirty-day mortality was reported in six comparison groups. Thirty-day mortality was similar between bioprosthetic valves and mechanical valves (OR 1.51, 95% CI 0.97 to 2.35, p=0.069) (online supplementary figure 2). Inclusion of only studies with lower risk of bias15 18 21 did not alter the results with bioprosthetic valves having similar 30-day mortality to mechanical valves (OR 1.51, 95% CI 0.91 to 2.49, p=0.110). Absolute RD was 102 (95% CI −7 to 210) indicating no difference in risk of 30-day mortality between bioprosthetic and mechanical valves. Funnel plot showed low risk of bias and heterogeneity was low (I2=0%).
Current guidelines do not provide guidance on the risk of infective endocarditis between mechanical and bioprosthetic valves, and emphasise discussion of the risks of reoperation, bleeding and thrombosis to help guide valve choice in an individual patient based on age and patient preferences. It is unknown if prosthetic valve endocarditis rates differ between mechanical valves and bioprosthetic valves and therefore risk of endocarditis is not incorporated into this shared decision-making process. Our systematic review and meta-analysis therefore provides the most contemporary evidence on the differential risk of endocarditis between bioprosthetic valves and mechanical valves with a previously unrecognised higher risk of infective endocarditis with bioprosthetic valves. These data provide important information to help guide the risk:benefit ratio discussion in individual patients navigating the complexity of choice of valve prosthesis.
Choice of mechanical versus bioprosthetic valves
Much of the evidence base for guideline recommendations for selection of prosthetic valves comes from observational studies and only three small randomised trials.17 19 22 All three RCTs showed no difference in rates of infective endocarditis comparing bioprosthetic valves to mechanical valves. However, these three randomised trials were insufficiently powered for rarer but clinically important events such as infective endocarditis. In addition, the choice of valve is often not subject to clinical equipoise and may be influenced by baseline thromboembolic/haemorrhagic risk, socioeconomic status, medication compliance, life expectancy and comorbid conditions. However, at the present time even when there is clinical and patient equipoise, we remain uncertain whether mechanical valves or bioprosthetic valves remain the treatment of choice particularly in intermediate age groups reflected by current guidelines.6 25 Therefore, we have relied on high-quality observational research with propensity matching along with the randomised trials to help inform this decision. Our meta-analysis by combining and summarising information from all these studies shows that bioprosthetic valves may be associated with higher risk of infective endocarditis.
Risk of endocarditis with prosthetic valves
The meta-analytic findings are consistent with recent observational studies showing a heightened risk of endocarditis with bioprosthetic valves compared with mechanical valves.10 11 The most recent estimates of risk of prosthetic valve endocarditis estimate that around 5% or 1/20 patients develop prosthetic endocarditis at 10 years,11 26 with even more at risk in longer follow-up with continued valve degeneration. The possibility of a higher risk of infective endocarditis with bioprosthetic valves therefore has important implications for decision regarding valve choice since prosthetic endocarditis is a major event often requiring repeat surgery, implantable device removal and in-hospital mortality as high as 29%.27 The results of our analysis therefore are important to consider in light of an accumulating body of evidence suggesting an increasing trend in the use of bioprosthetic valves, especially in younger patients.7–9
Potential mechanisms of endocarditis with bioprosthetic valves
Our data do not allow an identification of the mechanism of heightened endocarditis in bioprosthetic valves, but a number of possibilities exist. Bioprosthetic valves undergo haemodynamic valve deterioration at a higher rate than often clinically detected with progressive elevation in Doppler gradients even when asymptomatic.28 This coupled with the higher baseline valve gradients compared with mechanical valves29 leads to increased shear stress on the prosthetic leaflets, endothelial injury and in setting of transient bacteremia potentially causes clinical endocarditis. There is also a growing body of literature indicating that bioprosthetic valve thrombosis may be more common than had been previously recognised when systematically assessed by CT scanning.30 Egbe et al studied bioprosthetic valve thrombosis in 11% of their patients of which >65% occurred more than a year after valve implantation.4 5 The resultant structural degeneration coupled with thrombi on the prosthesis may serve as a nidus for the organisms to adhere and cause prosthetic valve endocarditis.31 In addition, there may be fundamental differences in the biological risk of tissue prostheses as opposed to biologically inert mechanical valves, and our current data would not support using bioprosthetic valves to minimise risk of endocarditis in patients at high infectious risk as is increasingly being used.32 In fact, recent series have demonstrated a higher mortality and worse outcomes with bioprosthetic valve replacement in endocarditis33–36 which would corroborate our study findings of the higher infective risk with bioprosthetic valves.
Our study has some important limitations. Majority of the studies included in the analysis are observational in nature, and this is a major limitation of the meta-analysis preventing us from reaching solid conclusions. Despite propensity matching, we cannot exclude unmeasured confounding influencing the choice of prosthesis and subsequent risk of endocarditis, but this remains a limitation of all existing research in this field to date. The randomised trials and some of the observational studies that are pooled for this meta-analysis were not designed to measure infective endocarditis or embolic events as the primary outcome. We do not have data on duration of onset of infective endocarditis from the time of valve implant; the longer the valve is, the higher the rates of degeneration and infective endocarditis, and this is particularly important in the younger population. We cannot identify the effect of sex, age, valve location and the type of bioprosthetic valves on the outcomes of interest due the lack of individual patient-level data. Finally, the existing randomised trials are of insufficient statistical power to detect fewer common events such as endocarditis. Our study therefore provides the most contemporary evidence on endocarditis risk with bioprosthetic valves as compared with mechanical valves.
In this systematic review and meta-analysis of randomised and propensity-matched observational studies, bioprosthetic valve use is associated with a higher incidence of infective endocarditis at follow-up. These data should help in the shared decision-making process to guide choice of valve replacement in individual patients based on medical risks and patient preference. Further research to understand mechanisms and preventive strategies for endocarditis in bioprosthetic valves are urgently needed.
What is already known on this subject?
Bioprosthetic valves are being used with increased frequency for valve replacement, with controversy regarding risk:benefit ratio compared with mechanical valves in younger patients.
What might this study add?
In this systematic review and meta-analysis, we found that patients with bioprosthetic valves had a higher risk of infective endocarditis compared with patients with mechanical valves.
How might this impact on clinical practice?
Bioprosthetic valves may be associated with a higher risk of infective endocarditis. This should be an important part of the shared decision-making process in choosing between bioprosthetic and mechanical valves in individual patients.
MA-N and YNVR contributed equally.
Correction notice Since the online publication of this article, the author name 'Larry M Baddour' was updated to include the middle initial.
Contributors Conception or design of the work: MAN, YNVR, VS. Data collection: MAN, YNVR, VS, PJE. Data analysis and interpretation: MAN, YNVR, VS, MHMM. Drafting the article: all authors. Critical revision of the article: LB, HVS, RAN. Final approval of the version to be published: all authors.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data sharing not applicable as no datasets generated and/or analysed for this study.