Aortic valve replacement with mechanical vs. biological prostheses in patients aged 50–69 years

Abstract

Aims

The objective was to investigate the long-term all-cause mortality in patients aged 50–69 years after aortic valve replacement (AVR) with bioprosthetic or mechanical valves.

Methods and results

All patients aged 50–69 years who had undergone AVR in Sweden 1997–2013 were identified from the Swedish Web-system for Enhancement and Development of Evidence-based care in Heart disease Evaluated According to Recommended Therapies register. Subsequent patient-level record linkage with national health-data registers provided patient characteristics, vital status, and clinical outcomes. Of the 4545 patients, 60% (2713/4545) had received mechanical valves and 40% (1832/4545) bioprostheses. In 1099 propensity score-matched patient pairs, 16% (180/1099) had died in the mechanical valve group and 20% (217/1099) in the bioprosthetic group; mean follow-up 6.6 (maximum 17.2) years. Survival was higher in the mechanical than in the bioprosthetic group: 5-, 10-, and 15-year survival 92, 79, and 59% vs. 89, 75, and 50%; hazard ratio 1.34; 95% confidence interval (CI) 1.09–1.66; P = 0.006. There was no difference in stroke [subdistribution hazard ratio (sHR) 1.04; 95% CI 0.72–1.50, P = 0.848]; however, the risk for aortic valve reoperation was higher (sHR 2.36; 95% CI 1.42–3.94, P = 0.001), and for major bleeding lower (sHR 0.49; 95% CI 0.34–0.70, P < 0.001), in patients who had received bioprostheses than in those with mechanical valves.

Conclusion

Patients aged 50–69 years who received mechanical valves had better long-term survival after AVR than those with bioprostheses. The risk of stroke was similar; however, patients with bioprostheses had a higher risk of aortic valve reoperation and a lower risk of major bleeding.

Clinical Trial Registration

http://clinicaltrials.gov/show/NCT02276950.

ClinicalTrials.gov Identifier

NCT02276950.

Introduction

The standard treatment option for severe aortic valve disease, aortic valve replacement (AVR), is performed in ∼280 000 patients annually worldwide.¹ Current guidelines from the European Society of Cardiology² states that a mechanical valve should be considered in patients younger than 60 years of age and a bioprosthesis should be considered in patients >65 years of age. In patients aged 60–65 years, both valve types are considered acceptable options. According to the American Heart Association/American College of Cardiology (AHA/ACC) guidelines,³ it is reasonable with a mechanical valve in patients younger than 60 years of age, a bioprosthesis in patients >70 years of age and in patients aged 60–70 years, both valve types are considered reasonable options.

Interestingly, Chiang and colleagues recently published a large study that challenges current guidelines, showing similar long-term survival in patients aged 50–69 years with either mechanical valves or bioprostheses.⁴ These findings indicate that bioprostheses could be considered for patients down to 50 years of age and are supported by the study of McClure et al.⁵

Hence, the optimal prosthesis type for middle-aged patients remains controversial. We therefore conducted a population-based cohort study of all patients aged 50–69 years who had undergone primary, isolated AVR in Sweden between 1997 and 2013. The primary objective was to investigate the long-term all-cause mortality in these patients according to whether they had received mechanical valves or bioprostheses. The secondary objectives were to compare the rates of stroke, aortic valve reoperation, and major bleeding events.

Methods

Study design

This observational, nationwide, population-based, cohort study complied with the Declaration of Helsinki and was approved by the regional Human Research Ethics Committee, Stockholm, Sweden. The study population was obtained from the Swedish Web-system for Enhancement and Development of Evidence-based care in Heart disease Evaluated According to Recommended Therapies (SWEDEHEART) register.^6–8 This register holds information regarding the clinical and operative characteristics of all patients who have undergone heart surgery in Sweden since 1992.

All patients aged 50–69 years who had undergone primary isolated AVR in Sweden from 1 January 1997 through 31 December 2013 were included. The International Classification of Diseases version 10 procedure codes FMD00 and FMD10 were used to determine whether mechanical valves or bioprostheses, respectively, had been used. Further information regarding baseline characteristics was extracted from the National Patient Register (Supplementary material online, Table S1) and the longitudinal integration database for health insurance and labour market studies (maintained by Statistics Sweden). Individual level record linkages between all national health-data registers are possible in Sweden because a unique personal identity number is assigned to every Swedish citizen.⁹ The national registers are further described in Supplementary material online.

Outcomes

The primary outcome measure was all-cause mortality. Vital status, cause, and date of death were obtained from the national Cause of Death Register. Secondary outcomes included stroke, aortic valve reoperation, major bleeding events, and cardiovascular death. Information regarding post-operative stroke and major bleeding were ascertained from the corresponding primary diagnosis codes at hospital discharge from the National Patient Register (Supplementary material online, Table S2), and regarding aortic valve reoperation from the SWEDEHEART register.

Statistical methods

Patient characteristics are presented as frequencies and percentages for categorical variables and as means and SDs for continuous variables. The primary outcome measure was death from any cause. Person-times in days were counted from the date of surgery until the date of death or the end of follow-up (24 March 2014). Because information regarding stroke, major bleeding events, and cardiovascular death was available only until 31 December 2012, the follow-up period for the secondary outcome measures ended on 31 December 2012. Patients who had undergone surgery during 2012 and 2013 were therefore excluded from the secondary outcome analyses. The crude incidence rates and 95% confidence intervals (CIs) were calculated and the Kaplan–Meier method was used to calculate cumulative survival.

To reduce selection bias, logistic regression using all variables in Table 1 (including hospital) as covariates and bioprosthesis as the dependent variable was used to calculate a propensity score for each patient. A propensity score-matched cohort was constructed by 1:1 nearest neighbour matching without replacement, and a calliper width equal to 0.2 of the SD of the logit of the propensity score. Standardized differences for variables were calculated to investigate post-match balance. Baseline characteristics in patients who were, and were not, matched are shown in Supplementary material online, Table S3. Missing data were handled by estimating separate logistic regression models such as to maximize the number of included variables for each patient. In the propensity score-matched cohort, the relative risk of death, expressed as hazard ratios (HRs), in relation to prosthesis type was estimated using Cox proportional hazards regression with robust standard errors that allowed for intragroup correlation. We found no evidence of violation of the assumption of proportional hazards. The Cox model was stratified by calendar year of surgery and hospital. The mechanical valve group was the reference category in all analyses.

In the overall cohort, Cox regression was used to estimate the risk of all-cause mortality according to type of prosthesis. Crude and multivariable adjusted HRs and 95% CIs were calculated. The multivariable model included all variables listed in Table 1 and was stratified by calendar year of surgery and hospital. Body mass index was modelled as restricted cubic splines, age was included as a continuous variable, whereas all other variables were included as categorical terms. In the overall cohort, propensity scores were calculated and included in the multivariable Cox regression model as a continuous variable. Propensity scores were divided into quintiles and a separate Cox regression model was stratified by propensity score quintile.

Competing risk regression based on the Fine-Gray proportional subhazards model was performed¹⁰ and subdistribution HR and 95% CI were calculated to estimate the risk of the secondary outcomes according to type of prosthesis. The cumulative incidence function was used to graph the rates of the secondary outcomes, thus accounting for the competing risk of death.

We also performed separate analyses in propensity score-matched patients aged 50–59 years and 60–69 years, and in patients operated on before 2006.

Data were missing for the following variables that were used as covariates in the overall cohort multivariable analyses: renal function (12%), left ventricular ejection fraction (31%), and body mass index (14%). Multiple imputation by chained equations was used to handle missing data.¹¹ Under the assumption that the missing values were missing at random, the imputation models included 27 variables, including the event indicator and the Nelson–Aalen estimator of the cumulative baseline hazard.¹² Twenty-five data sets were imputed and estimates from these data sets combined according to Rubin's rules.

Data management and statistical analyses were performed using Stata 13.1 (Stata Corp LP, College Station, TX, USA) and R version 3.1.2 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Study population

We identified all patients aged 50–69 years who had undergone primary, isolated AVR and received either mechanical valves or bioprostheses in Sweden between 1997 and 2013. Patients who had undergone prior cardiac surgery or a concomitant procedure were excluded. A total of 4545 patients were included in the study. Of these, 60% (2713/4545) had received mechanical valves and 40% (1832/4545) bioprostheses. The use of bioprostheses increased from 17% in 1997–2002 to 58% in 2006–2013, while the mean patient age per calendar year remained relatively constant (Figure 1; Supplementary material online, Figure S1). The mean and maximum follow-up times were 7.3 (SD 4.7) and 17.2 years in the overall cohort, 8.8 (SD 4.6) and 17.2 years in the mechanical valve group, and 5.0 (SD 3.7) and 17.2 years in the bioprosthetic valve group. The total follow-up time was 32 989 patient-years, comprising 23 826 patient-years for the mechanical valve group and 9163 patient-years for the bioprosthetic valve group.

Patient characteristics

The baseline patient characteristics in the overall cohort are shown in Table 1. The mean age was 59.9 years in patients who had received mechanical valves and 63.7 years in those who had received bioprosthetic valves. Compared with the patients who received mechanical valves, the patients who had received bioprostheses were older and generally had more comorbidities, notably prior major bleeding events, impaired renal function, liver disease, alcohol dependency, and cancer. In the propensity score-matched cohort (1099 pairs), all baseline characteristics were well balanced, as shown in Table 2.

Survival

Kaplan–Meier-estimated survival curves in the propensity score-matched cohort are shown in Figure 2. The long-term survival was significantly greater in the mechanical valve than in the bioprosthetic valve group; in the overall unadjusted analysis (HR 1.67; 95% CI 1.44–1.94), in the overall multivariable adjusted model (HR 1.30; 95% CI 1.09–1.56) and in the propensity score-matched cohort (HR 1.34; 95% CI 1.09–1.66; P = 0.006) (Table 3). During follow-up, 16% (180/1099) patients died in the mechanical valve group and 20% (217/1099) in the bioprosthetic valve group. The 5-, 10-, and 15-year survivals in the propensity score-matched cohort were 92, 79, and 59%, respectively, in the mechanical valve group, and 89, 75, and 50%, respectively, in the bioprosthetic valve group. As shown in Table 3, these results are consistent with the analyses in the overall cohort (n = 4545). The mean and maximum follow-up times were 6.6 (SD 4.0) and 17.2 years, respectively, in the propensity score-matched cohort; 6.7 (SD 4.2) and 17.1 years, respectively, in the mechanical valve group; 6.5 (SD 3.9) and 17.2 years, respectively, in the bioprosthetic valve group. The total follow-up time was 14 423 patient-years, comprising 7324 patient-years for the mechanical valve group and 7099 patient-years for the bioprosthetic valve group.

Restricting the analyses to patients who had undergone surgery between 1997 and 2005 produced similar results as in the total study population; namely, an increased risk of death in patients who had received bioprostheses (adjusted HR 1.41; 95% CI 1.11–1.80; P = 0.005 in 824 propensity score-matched patients). We also performed separate analyses in propensity score-matched patients aged 50–59 and 60–69 years, P-value for interaction was 0.04 (Figure 3). In propensity score-matched patients aged 50–59 years, the long-term survival was significantly greater in the mechanical than in the bioprosthetic valve group (HR 1.67; 95% CI 1.06–2.61; P = 0.026, n = 574). However, in patients aged 60–69 years, there was no significant difference in long-term survival between the mechanical and bioprosthetic valve groups (HR 1.08; 95% CI 0.85–1.36; P = 0.539, n = 1502).

Secondary outcomes in the propensity score-matched cohort

Stroke

During a maximum follow-up of 15.8 years in the mechanical valve group and 15.9 years in the bioprosthetic valve group, 5.8% (54/939) in the mechanical valve group, and 6.1% (57/939) in the bioprosthetic valve group had strokes (Table 4). There was no significant difference in the incidence of stroke between patients who had received mechanical valves and those with bioprostheses (subdistribution hazard ratio, sHR 1.04; 95% CI 0.72–1.50, P = 0.848) (Figure 4).

Aortic valve reoperation

During a maximum follow-up of 15.9 years in the mechanical valve group and 16.0 years in the bioprosthetic valve group, 2.2% (21/939) in the mechanical valve group, and 5.2% (49/939) in the bioprosthetic valve group had undergone aortic valve reoperation (Table 4). The risk for aortic valve reoperation was significantly higher in patients who had received bioprostheses than in those with mechanical valves (sHR 2.36; 95% CI 1.42–3.94, P = 0.001) (Figure 4).

Major bleeding

During a maximum follow-up of 15.8 years in the mechanical valve group and 16.0 years in the bioprosthetic valve group, 9.6% (90/939) in the mechanical valve group, and 4.9% (46/939) in the bioprosthetic valve group had had major bleeding events (Table 4). The risk for major bleeding events post-operatively was significantly higher in patients who had received mechanical valves than in those with bioprostheses (sHR 0.49; 95% CI 0.34–0.70, P < 0.001) (Figure 4).

Cardiovascular death

During a maximum follow-up of 16.0 years in the mechanical valve group and 15.9 years in the bioprosthetic valve group, 5.2% (49/939) in the mechanical valve group, and 5.1% (48/939) in the bioprosthetic valve group died from cardiovascular causes (Table 4). There was no significant difference in the incidence of cardiovascular death between patients who had received mechanical valves and those with bioprostheses (sHR 1.00; 95% CI 0.67–1.50) (Figure 4).

Discussion

We found that patients aged 50–69 years who had undergone primary, isolated AVR and received mechanical valves had significantly higher long-term survival than those who had received bioprostheses in Sweden from 1997 to 2013. There was a similar risk of stroke in the two groups; however, patients who had received bioprostheses had a higher risk of aortic valve reoperation and a lower risk of major bleeding than those who received mechanical valves. A separate subgroup analysis of patients aged 50–59 years showed a significant association between bioprosthetic AVR and long-term all-cause mortality; however, we identified no significant difference in survival in those aged 60–69 years. We found no difference in cardiovascular death between patients who had received mechanical valves and those with bioprostheses. These results could indicate that factors other than valve choice explained the difference in survival between the groups, although this must be interpreted cautiously because the validity of diagnoses in the Cause of Death register is uncertain and rarely based on findings from autopsy.

Mechanical valves have longer durability compared with bioprostheses, but require life-long anticoagulation. Current guidelines from the European Society of Cardiology² recommend bioprostheses for patients aged >65 years and mechanical valves for those aged <60 years. Between 60 and 65 years, both bioprostheses and mechanical valves are considered acceptable options. In the guidelines from the AHA/ACC,³ a bioprosthesis is recommended >70 years of age, a mechanical valve <60 years and both valve types are considered reasonable options between 60 and 70 years of age. Our results support the current recommendations. Studies have shown an increasing use of bioprostheses in all age groups.¹³

Prior studies investigating survival and clinical outcomes following AVR with mechanical or bioprosthetic valves in middle-aged patients have reported contradictory results. Some studies have reported better long-term clinical outcomes in patients who received mechanical valves,^14–16 whereas others have reported no significant difference in long-term survival between middle-aged patients who received bioprostheses and those who received mechanical valves.^4,5,17

In an observational study of 4253 patients aged 50–69 years who underwent primary, isolated AVR in New York State between 1997 and 2004, Chiang and colleagues⁴ analysed 1001 propensity score-matched patient pairs who received mechanical or bioprosthetic valves. In contrast to our results, they found no difference in 15-year survival between the two groups. However, similarly to our results, they found no difference regarding the risk of stroke, a higher incidence of aortic valve reoperation, and a lower incidence of major bleeding in the bioprosthetic valve group.

There are several possible explanations for the difference in survival between the study from New York State and our nationwide study in Sweden. Specifically, routines for post-operative care and patient follow-up may differ between New York State and Sweden. In Sweden, the health-care system is based on publicly financed health insurance coverage that guarantees all citizens access to health services. Patients are generally followed by their family doctors or cardiologists. Anticoagulant treatment is usually monitored by general practitioners; however, some patients prefer self-management and home monitoring. The quality of anticoagulation control in Sweden has repeatedly been shown to be very high,^18,19 the proportion of time in the therapeutic range being well above the 70% recommended by the European Society of Cardiology Working group on Thrombosis Anticoagulation Task Force.²⁰ Thus, the high quality of anticoagulation management in Sweden may have favourably affected clinical outcomes in patients with mechanical heart valves. Furthermore, direct comparisons between the study by Chiang⁴ and the present study must be made cautiously because of some important differences in the patients' baseline characteristics. Notably, patients in the study by Chiang et al.⁴ had more comorbidities than did our patient cohort; namely, a higher prevalence of diabetes mellitus, atrial fibrillation, chronic pulmonary disease, and congestive heart failure. Also, their patients more frequently underwent surgery after urgent/emergency admission than did ours. One limitation common to both the study by Chiang et al.⁴ and ours was a lack of information regarding the specific prosthetic heart valves (model, manufacturer, and size) used; this may have implications for long-term results. Different types of prosthetic heart valves (e.g. bovine vs. porcine) may perform differently in the long term. Additionally, for a given valve size, the effective orifice area is larger in mechanical valves than in bioprostheses. Therefore, prosthesis-patient mismatch may occur more often in patients with bioprostheses, which may affect long-term survival.¹ Stassano et al. performed the only contemporary prospective randomized study comparing mechanical and bioprosthetic valves in 310 middle-aged patients.¹⁵ They found no difference in survival after 13 years. However, their study included older patients than our study (55–70 years; mean age 64 years), which could explain the lack of survival difference between the groups. Moreover, they found a significantly higher risk of valve failure and reoperation in the bioprosthesis group, and no significant difference in other major adverse prosthesis-related events.

In contrast to the previously mentioned studies, Brown et al. reported a survival benefit for mechanical valves in patients aged 50–70 years.¹⁴ They included 440 patients who received either a mechanical valve or a bioprosthesis. The patients were matched one-to-one according to age, sex, need for coronary artery bypass grafting and valve size. A limitation was that they were unable to balance the two groups for age and other important patient characteristics, which might have affected the results.

Study limitations

Because this was an observational study, the issue of selection bias was addressed primarily by matching using propensity scores and regression adjustment, both approaches having provided similar results. However, it is only possible to match or adjust for patient characteristics that are known and have been measured and recorded. There may be unmeasured or unknown factors or both that we were unable to control for (residual confounding). Another limitation was missing baseline data which potentially could have influenced our results.

Because some patients could have experienced adverse events outside of Sweden, we may have underestimated some secondary outcomes. However, all deaths that occur abroad are captured by the national Swedish Cause of Death register; thus, follow-up regarding the primary end-point was complete. Although the population-based design, which included all centres performing cardiac surgery in Sweden, increased generalizability, there may still be differences in pre- and post-operative care between Sweden and other countries with similar levels of healthcare. The particular strengths of this study include the large number of patients and the long and complete follow-up, which was made possible by linking information from several Swedish high-quality nationwide registers.

Conclusions

We found a better long-term survival in patients aged 50–69 years who had undergone primary isolated AVR and received mechanical valves than in those who had received bioprostheses in Sweden from 1997 to 2013. The two groups had a similar risk of stroke; however, patients who had received bioprostheses had a higher risk of aortic valve reoperation and a lower risk of major bleeding than those who had received mechanical valves.

Authors’ contributions

N.G. and U.S.: performed statistical analysis. M.H., A.F.-C., and U.S.: handled funding and supervision. N.G., M.H., and U.S.: acquired the data. N.G., V.J., M.H., A.F.-C., and U.S.: conceived and designed the research. N.G. and U.S.: drafted the manuscript. N.G., V.J., M.H., A.F.-C., and U.S.: made critical revision of the manuscript for key intellectual content.

Supplementary material

Supplementary Material is available at European Heart Journal online.

Funding

This study was supported by the Swedish Medical Society (Grant number SLS-330221 to U.S.), Karolinska Institutet Foundations and Funds (Grant number 40842 to U.S.), the Mats Kleberg Foundation (Grant number 2014-00017 to U.S), and a donation from Mr Fredrik Lundberg (to A.F.-C.).

Conflict of interest: none declared.

Acknowledgements

We thank the SWEDEHEART steering committee for providing data for this study.

References

Author notes