Long-term survival after Carpentier-Edwards Perimount aortic valve replacement in Western Denmark: a multi-centre observational study

Background This study describes the long-term survival, risk of reoperation and clinical outcomes of patients undergoing solitary surgical aortic valve replacement (SAVR) with a Carpentier-Edwards Perimount (CE-P) bioprosthetic in Western Denmark. The renewed interest in SAVR is based on the questioning regarding the long-term survival since new aortic replacement technique such as transcatheter aortic-valve replacement (TAVR) probably have shorter durability, why assessment of long-term survival could be a key issue for patients. Methods From November 1999 to November 2013 a cohort of a total of 1604 patients with a median age of 73 years (IQR: 69–78) undergoing solitary SAVR with CE-P in Western Denmark was obtained November 2018 from the Western Danish Heart Registry (WDHR). The primary endpoint was long-term survival from all-cause mortality. Secondary endpoints were survival free from major adverse cardiovascular and cerebral events (MACCE), risk of reoperation, cause of late death, patient-prothesis mismatch, risk of AMI, stroke, pacemaker or ICD implantation and postoperative atrial fibrillation (POAF). Time-to-event analysis was performed with Kaplan-Meier curve, cumulative incidence function was performed with Nelson-Aalen cumulative hazard estimates. Cox regression was applied to detect risk factors for death and reoperation. Results In-hospital mortality was 2.7% and 30-day mortality at 3.4%. The 5-, 10- and 15-year survival from all-cause mortality was 77, 52 and 24%, respectively. Survival without MACCE was 80% after 10 years. Significant risk factors of mortality were small valves, smoking and EuroSCORE II ≥4%. The risk of reoperation was < 5% after 7.5 years and significant risk factors were valve prosthesis-patient mismatch and EuroSCORE II ≥4%. Conclusions Patients undergoing aortic valve replacement with a Carpentier-Edwards Perimount valve shows a very satisfying long-term survival. Future research should aim to investigate biological valves long-term durability for comparison of different SAVR to different TAVR in long perspective.


Background
Aortic stenosis (AS) is the most frequent primary valve disease in Europe and North America [1]. Surgical intervention remains the gold standard in treating this condition [2]. Surgical aortic valve replacement (SAVR) with bioprosthetic is in Europe usually offered to patients older than 65 years of age with severe AS or severe aortic valve regurgitation and suitable for surgery, though it also is considered in patients younger than 65 years of age in preference of mechanical valves, depending on life expectancy, comorbidity, compliance and patient preference [1].
The future application of SAVR is however being questioned, due to the technological advancement of transcatheter aortic-valve replacement (TAVR) [20]. The primary objective of this registry-based retrospective cohort study is to investigate the long-term survival of all patients who underwent a solitary SAVR with CE-P valve bioprosthesis and secondary to explore the adverse clinical outcomes.

Methods
This registry-based study was approved by the Region Southern Denmark Data Protection Agency, Odense, Denmark, and the study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
We conducted a population-based retrospective cohort study analysis including all patients who underwent solitary SAVR with CE-P valve in Western Denmark based on data from November 1999 to November 2013. The primary endpoint of this study was long-term survival from all-cause mortality with valve size 19 to 29 mm in comparison with the Danish background population. Secondary endpoints were survival from major adverse cardiovascular and cerebral events (MACCE), risk of reoperation, cause of late death, valve patient-prothesis mismatch (VP-PM), PAOF, risk of AMI, stroke and pacemaker. We excluded patients operated with other valve types of aortic valve replacement (AVR) or had concomitant surgery.
The data was obtained November 2018 from the Western Danish Heart Registry (WDHR), The National Danish Patient Registry and the Program of clinical quality development of the Danish Regions. WDHR is a multi-centre prospective registry and is the most comprehensive registry regarding AVR in Denmark. The background population was obtained from Statistics Denmark and matched according to our cohort with a mean age of 73 years and a follow-up of 15 years. MACCE is defined as all-time cardiovascular events and early (≤30 days) cerebrovascular disease. Low-risk patients was defined according to guidelines, EuroSCORE II < 4% [1]. VP-PM was evaluated as effective orifice area index (EOAI) < 0.85 cm 2 /m 2 estimated with manufactural effective orifice area (EOA) measurements indexed with perioperative body surface area (BSA) knowing that the measurements reliability to predict EOAI are questioned. A total of 1613 people were registered with solitary SAVR in WDHR (Fig. 1). Six patients were duplicates, so the first surgery where kept, and the late will be included as reoperation. Three patients were removed due to invalid or inactive social security number. That leaves a population of 1604 patients whom all had an efficient follow-up.
Baseline characteristics were described using means, range, standard deviation, for continuous variables, Fig. 1 Flow chart of study population selection. WDHR, Western Danish Heart Registry; AVR, aortic valve replacement; CE-Perimount, Carpentier-Edwards Perimount median and quartiles for skewed variables and proportions for categorical variables. Time-to-death was calculated as the time in years from the date of operation to the time of death from all-cause or MACCE-cause. Survival was performed with the use of Kaplan-Meier estimates. The overall survival curve was compared to the background population graphicly by evaluating overlap of curves and confidence interval. Log-rank test was performed to compare equality of survivor function of subgroups. Cumulative incidence function of reoperation was performed with Nelson-Aalen cumulative hazard estimates. Only the first re-event for each patient during the study was used for analysis. Cox regression was used to detect the risk factors for death and reoperation of any cause reporting hazard ratio (HR) and 95% confidence interval (CI). The age between 65 to 70 was set as reference for age related cox analysis. The same applies to prosthesis size 25 in respect to valve size related cox analysis. A P-value of < 0.05 was considered significant. All statistical analyses were performed with the use of STATA software, version 15.

Reoperation
The risk of reoperation with a new CE-P or TAVR of any cause during the follow-up was 4% (n = 64) and 22% (n = 14) was due to endocarditis. The cumulative incidence of reoperation according to prosthesis size are presented in Fig. 5 and

Discussion
This multi-centre observational follow-up study evaluated up to 15 years of long-term survival after SAVR with CE-P valve in terms of mortality and clinical outcomes in 1604 Danish patients having a solitary SAVR. Our study found a 30-day all-cause mortality at 3.4%, and a 5-, 10-and 15year survival from all-cause at 77, 52 and 24%, respectively including a 7.5-year risk of reoperation < 5%.    This study's observation of all-cause mortality was consistent with the Danish matched population after early risk and until year 10, were the mortality takes another discrete yet noticeable hit until the end of the follow-up (Fig. 2). The early mortality was higher than predicted in comparison to mean ES2 of the population (meanES2: 1.45%, predicted number of deaths: 23). After patients surpass the general risk of the operation the first 30 days, the survival rate matches to the Danish population for 10 years. We may assume that the CE-P valves durability is being challenged by this point, which can also be supported by literature describing the development of SVD in CE-P at 9.24% after 10 years [10], and 18% after 15 years [7]. The Danish population is matched according to the mean age of our baseline (Median: 73 years, range: 20-91) which matches previous reports [3, 7-9, 19, 21, 22], and 12% of the cohort being < 65 years. SAVR at the age of 20-30 is very rare and the choice of valve should consider the risk of lifelong anticoagulation with mechanical valve versus the very high risk of reoperation with bioprosthetic. Young age and small valve sizes are well associated with accelerated development of SVD, why they are usually recommended mechanical valves, especially under the age of 60 [1,3,6,[8][9][10]. Our data suggests that age < 65 years is significantly associated with reoperation, however not with death (Table 2). However, the risk of death was significantly increased for smaller valve size 19-23 why the biggest valve possible including root-enlargement must be considered when the patient is in the range of 19-23 mm to avoid VP-PM. Furthermore, reoperation has a HR of death at 1.07 (P = 0.75) and older demographics (≥70 years) have a significant increased risk of death and decreased risk of reoperation (Table 2). This may suggest that ether the durability of the CE-P valve is not the primary cause of death in this age group, or and more likely that they're not suitable for reoperation because of development and progress of comorbidities. Furthermore, that the increased risk in mortality in small vales is presumable explained by the demographics rather than SVD development. Nevertheless, our findings could indicate that patients undergoing SAVR with CE-P can expect a survival matching the background population the next 10 years after overcoming early risks. A Danish study described the 10-year survival from all-cause mortality of CE-P at 36% resulting, which is significantly worse than our findings of 52% for roughly the same period [19]. This could be explained by their singlecentre approach, their inclusion of concomitant CABG procedures and the fact that we conducted our results November 2018 which prolongs our follow-up with 4.5 years. The surgical technique and postprocedural medical attention have improved tremendously over time, and therefore a prolonged period of 4.5 years will have a significant effect on a 10-year measurement. Additionally, we may assume that if patients are eligible for concomitant CABG, they have increased probability of comorbidity contributing to the risk of death. It does however interfere with their true long-term survival analysis and durability of the CE-P valve. It is noteworthy that our survival is consistent with other studies; overall 5-, 10-and 15-year survival rate in China (81.58 66. 19 and 57.33% [10]), overall 10-and 15-year survival rate in France (52.4 and 31.1% [9]), overall 5-, 10-and 15-year survival rate in a Canadian study (78, 55 and 34% [21]) and others [23]. We also investigated the survival from MACCE-cause mortality in order to more comprehensive describe the valve durability in combination with all causes of late death (Fig. 3). The cause of late death was 55.6% due to other than valve or MACCE-related causes, why one might argue that CE-P surpasses the general patient's life expectancy and therefore making it a reliable choice for SAVR and that future randomized studies could take MACCE into account when evaluating the durability in order to give a more widespread evaluation of the valves. VP-PM (HR: 2.19) and high-risk patients (HR: 6.62) had a significant impact on the risk of reoperation, while valve size was not an independent risk factor. VP-PM's impact on reoperation contradicts earlier report of CE-P using calculated EOA reference values to predict VP-PM, which were substantially higher than the values provided by the manufacture [24]. The prediction of VP-PM is controversial because of various calculations. However, one study comparing 4 methods argued that the best calculations are made by EOA measured in vivo by Doppler echocardiography [25]. The latest meta-analysis supported the association of VP-PM have higher risk for perioperative, 1-, 5-and 10-year mortality rates in comparison to those with non-significant or no VP-PM. In our study, 12.7% had EOAI < 0.85 why this could contribute to MACCE-cause mortality, though our estimates are based on manufactural EOA measurements indexed with CI Confidence interval, CIF Cumulative incidence function BSA. ES2 ≥ 4% were found as an independent risk factor for reoperation, which correlates with the selection-bias in this group. Patients with ES2 ≥ 4% have more comorbidities such as renal failure, previous AMI, COPD, poor lung capacity, endocarditis or the need of acute need of operation. This could also explain the prolonged hospital stay in high risk compared with low risk. The small valve sizes were not associated with reoperation, which may be explained by unsuitability for surgery or the small sample size. The operative risk of reoperation due to SVD in CE-P was earlier described at 0% for size 19 and 21 and 0.1% for all valve sizes [19], which may support our noncorrelation of small valves with reoperation. However, the only way to finally evaluate whether a correlation between small valves and the development of SVD is present, is by performing a prospective study with close echocardiographic follow-up, so no patient would experience SVD without notice. The cumulative incidence function of reoperation due to any cause was < 5% after 7.5 years (Table  3), which is more than previous reports of freedom of reoperation at 96-99.5% after 10 years [19,21,26]. This could be due to our broad range of age (20-91 years) [21]. Forcillo et al. divided their population into 3 different agecategories illustrating that younger patients had a higher risk of reoperation with a 10-year risk of 90% in < 60 years of age with no mention of the subgroup size [21], and Langanay et al. had an age range of 80-96 years [26]. The small sample size after 7.5 years inhibits further associations to be made. Other clinical implications such as stroke, AMI, permanent pacemaker or ICD and POAF is at focus, especially in the comparison to TAVR. We observed a 30-days stroke rate at 1.3% (21 patients; HR:1.35; CI:0.74-2.45; P = 0.32) being lower than the SAVR findings in the PARTNER 3 trail which was 2.4% (meanES2: 1.5%) [20]. The 30-day incidence of AMI was also lower in our study (0.4%) compared to literature (1.3-1.6%) [20,27]. Lower incidence was also found regarding pacemaker or ICD implantation (3.9%), as SAVR-reports once again reports higher incidences (4.1-7.4% -pacemaker only) [20,27,28]. Interestingly, our data matches in general the findings found in another CE-P valves cohort study (stroke 1%, AMI 1%, atrioventricular-block 4% [21]). The incidence of POAF (36.4%) is however higher than some reports (11%) [27], which might be due to the definition of POAF in regards to duration and the overall awareness in the ward, since it is comparable to other findings (32-39.5%) [20,29]. With survival matching the background population after overcoming the early risks, along with the lower risk of AMI, stroke, pacemaker/ICD compared to the literature, and only 36.2% of the mortality is caused due to major adverse cardiovascular and cerebral events, our data strongly suggest that the CE-P valve have favourable characteristics with moderated cardiovascular risks compared to other bioprosthetic valves i.e. Mitroflow valve [3]. This research was limited by its retrospective design and the absence of continuance echocardiographic evaluation of SVD [6,30]. Long-term SVD and its relations to long-term survival therefore still remains in question. The long-term data is also limited given that only 20% of the cohort had a flow-up for more than 10 years available. The VP-PM data is also limited by calculations based on manufactural EOA, and can therefore only be acknowledged as suggestive. Furthermore, of the 1604 patients, 754 died before this study was conducted, and postoperative evaluation of SVD would not have been possible and 8.2% of cause of death remains uncertain. Additionally, the cohort was database-dependent and therefore relays on correct SKS codes, though earlier studies of the WDHD have shown very good quality evaluation with errors lower than 3% [31]. Selection bias in this study was minimized by the universal health care system based on the Beveridge model.

Conclusion
We cautiously conclude that our findings match previous literature, and that CE-P is a well-tested valve with satisfying long-term durability, which do not affect the long-term survival seriously for patients undergoing aortic valve replacement. This support the idea for future comparative research of TAVR-valves may be randomized solely against Carpentier-Edwards Perimount, and not pooled with other inferior bioprosthetic valves [3].