Long-Term Outcomes of Bovine versus Porcine Mitral Valve Replacement: A Multicenter Analysis

Introduction Recent national guidelines recommending mitral valve replacement (MVR) for severe secondary mitral regurgitation have resulted in an increased utilization of mitral bioprosthesis. There is a paucity of data on how longitudinal clinical outcomes vary by prosthesis type. We examined long-term survival and risk of reoperation between patients having bovine vs. porcine MVR. Study Design. A retrospective analysis of MVR or MVR + coronary artery bypass graft (CABG) from 2001 to 2017 among seven hospitals reporting to a prospectively maintained clinical registry was conducted. The analytic cohort included 1,284 patients undergoing MVR (801 bovine and 483 porcine). Baseline comorbidities were balanced using 1 : 1 propensity score matching with 432 patients in each group. The primary end point was all-cause mortality. Secondary end points included in-hospital morbidity, 30-day mortality, length of stay, and risk of reoperation. Results In the overall cohort, patients receiving porcine valves were more likely to have diabetes (19% bovine vs. 29% porcine; p < 0.001), COPD (20% bovine vs. 27% porcine; p=0.008), dialysis or creatinine >2 mg/dL (4% bovine vs. 7% porcine; p=0.03), and coronary artery disease (65% bovine vs. 77% porcine; p < 0.001). There was no difference in stroke, acute kidney injury, mediastinitis, pneumonia, length of stay, in-hospital morbidity, or 30-day mortality. In the overall cohort, there was a difference in long-term survival (porcine HR 1.17 (95% CI: 1.00–1.37; p=050)). However, there was no difference in reoperation (porcine HR 0.56 (95% CI: 0.23–1.32; p=0.185)). In the propensity-matched cohort, patients were matched on all baseline characteristics. There was no difference in postoperative complications or in-hospital morbidity and 30-day mortality. After 1 : 1 propensity score matching, there was no difference in long-term survival (porcine HR 0.97 (95% CI: 0.81–1.17; p=0.756)) or risk of reoperation (porcine HR 0.54 (95% CI: 0.20–1.47; p=0.225)). Conclusions In this multicenter analysis of patients undergoing bioprosthetic MVR, there was no difference in perioperative complications and risk of reoperation of long-term survival after matching.


Introduction
Recent national guidelines recommending mitral valve replacement (MVR) for severe secondary mitral regurgitation have resulted in an increased utilization of mitral bioprosthetic valves. Similarly, a decline in the use of mechanical mitral valves has seen a compensatory increase in the use of bioprosthetic valves [1]. Given that bioprosthetic valves do not require lifelong anticoagulation, they have become an increasingly attractive option in the treatment of valvular heart disease. Currently, the two most popular bioprosthetics on the market are the porcine Moasic (Medtronic, Minneapolis, MN) valve and the bovine pericardial Perimount (Edwards Lifescience, Irving, CA) valve. Tere has been a signifcant investigation into the performance of various brands of bioprosthetic valves in the aortic position; however, in the mitral position, there still exists a paucity of data on how longitudinal clinical outcomes vary by prosthesis types [2][3][4][5][6][7].
Previous studies on the use of bioprosthetic valves in the aortic position seem to favor the durability of bovine pericardial valves over that of porcine. Additional studies call into question the durability of bioprosthetics in younger patients [8,9]. Surprisingly, investigations comparing the two bioprosthetic valves in the mitral position have been contradictory, with some data suggesting a possible longitudinal structural advantage to porcine valves in the mitral position and others favoring outcomes of bovine pericardial MVR; however, these studies were limited by either small sample size, limited number of surgeons, or were performed at a single institution [10,11].
Given this overall paucity of data and lack of heterogeneity in study populations, comparing outcomes of bioprosthetic valves in the mitral position is difcult. Furthermore, given the increased risk of structural valve deterioration (SVD) in bioprosthetic mitral valves when compared with valves in the aortic position, long-term data about MVR is especially important. Te purpose of this study was to examine long-term survival and risk of reoperation between patients having bovine vs. porcine MVR.

Data Source.
We examined data from the Northern New England Cardiovascular Disease Study Group (NNECDSG). Te NNECDSG is a multicenter, voluntary, regional collaboration of interventional cardiology, cardiac surgery, and structural heart programs in northern New England. Te collaboration started in 1987 with a mission to improve the outcomes of cardiac interventions. Members from the seven medical centers include surgeons, cardiologists, anesthesiologists, perfusionists, nurses, other providers, administrators, and researchers. Institutional review boards (IRBs) at all the medical centers have designated the NNECDSG as a quality improvement registry, and therefore, patient consent for data collection was not required. Registry data are validated against hospital billing data every two years for the complete capture of cases and to ensure the accuracy of data for vital status at discharge.

Patients.
A retrospective analysis of MVR or MVR + CABG from 2001 to 2017 among seven hospitals reporting to a prospectively maintained clinical registry was conducted. Te analytic cohort included 1,284 patients undergoing MVR (801 bovine and 483 porcine). We excluded patients with endocarditis and those who have had a prior MVR.

Data Collection.
Data were collected from each institution by trained data abstractors. Detailed data on demographics, past medical history, coronary anatomy and function, procedural indications and process, device type, access method, and hospital outcomes were collected using standardized reporting forms of the NNECDSG (https:// www.nnecdsg.org). In 2012, hospitals began to transition their data collection tool to use the STS Adult Cardiac Surgery Database.
To ascertain survival beyond hospitalization, the NNE was linked to the National Death Index using name, date of birth, social security number, and the state vital status data. Survival data are currently updated for all patients through 2018.

Study Endpoints.
Te primary end point was all-cause mortality. Secondary end points included in-hospital morbidity, 30-day mortality, length of stay, and risk of reoperation. Reoperation was defned as any redo mitral valve replacement surgery captured by the NNE.

Statistical Analysis.
Overall group comparisons of baseline demographics and medical histories, echocardiographic, and periprocedural and procedural characteristics were summarized using means and standard deviations, median and interquartile range, or percentages as appropriate. Diferences in categorical variables were evaluated by using a χ 2 test, while t-tests or Wilcoxon rank-sum tests were used for continuous variables.
To balance the diferences in patient and disease characteristics between the two groups, a 1 : 1, nearest-neighbor propensity match without replacement was used to create comparable groups for analysis. Te matching caliper was set to 0.01. Te nonparsimonious logistic regression model to predict propensity scores included the following variables: age, gender, body surface area, prior PCI, prior CABG, prior stroke, preoperative atrial fbrillation, peripheral vascular disease (PVD), diabetes, chronic obstructive pulmonary disease (COPD), congestive heart failure (CHF), prior dialysis or creatinine ≥ 2, white blood cell, ejection fraction, number of diseased vessels, left main (LM) artery stenosis ≥ 50%, coronary artery disease, recent myocardial infarction (MI), and priority at the surgery. We additionally performed an inverse probability weighted (IPW) analysis to estimate the average treatment efect in the study population as a sensitivity analysis. We verifed the performance of our propensity score by assessing standardized diferences of the means for covariates among our propensity-matched and inverse probability-weighted cohorts [12]. Standardized diferences of <0.1 indicate that the groups are comparable on a particular variable.
Survival curves showing long-term mortality were generated by the Kaplan−Meier method and compared across groups using a log-rank test. Cox proportional hazards regression was used to calculate hazard ratios (HR) with 95% confdence intervals (CIs).
Competing risk analysis, with death as the competing event, was performed for reoperation using the cumulative incidence function with the assessment of the diference between prosthetic valve types using Gray's test. Competing risk regression was used to calculate the HR and 95% CI for reoperation.
For all analyses, a two-sided p value <0.05 was considered to be statistically signifcant. Data were analyzed using the statistical software package Stata 17 (Stata Corp, College Station, TX).

Results
Overall, 1284 patients undergoing MVR with a bioprosthetic valve were identifed with 801 undergoing pericardial MVR and 483 porcine MVR. Te match cohort using 1 : 1 propensity score matching (PMS) had 432 patients in each group.
Tere was no diference in stroke, acute injury, mediastinitis, pneumonia, length of stay, in-hospital morbidity, or 30-day mortality in either the overall or matched cohorts (Table 3).

Discussion
We found no signifcant diference in any of our in-hospital primary outcomes regardless of the valve type. Tere was a small signifcant diference found in long-term mortality before adjustment. While the overall incidence of reoperation was higher in subjects who underwent MVR with a bovine pericardial bioprosthetic, this did not reach signifcance (as shown in Figure 2(b)). Tus, there appears to be no signifcant diference in mortality or risk of explantation regardless of the choice of the bioprosthetic material.
Tere is a myriad of choices for valve replacement surgery including mechanical, cadaveric, and bioprosthetic products on the market. On one hand, with no need for lifelong anticoagulation, bioprosthetic valves are an attractive option in patients at high risk for bleeding. On the other hand, when compared with their mechanical counterparts, bioprosthetic valves are far less durable. It is important when considering a bioprosthetic valve for MVR to consider not only if but why they fail. Previous studies of bioprosthetic valves in the aortic position have shown that they are prone to fail in multiple ways, including calcifcation, cusp tears, and pannus formation [13]. Due to the higher closing pressure of the mitral valve, bioprosthetic valves tend to be less durable in the mitral position when compared with the aortic position. In one study of 240 patients who underwent MVR, porcine valves were shown to be more prone to failure secondary to leafet tears, presumably due to the increased pressure. Teir bovine counterparts were more susceptible to failure from adhesion of the posterior leafet strut and leafet to the subvalvular apparatus [14].
Studies of the hemodynamics of each valve type bring into question which valve has superior clinical outcomes and better longevity. In one study, 145 patients with rheumatic valvular disease were prospectively randomized to either bovine pericardial (Perimount) or porcine (Mosaic) valves. Pericardial valves demonstrated better hemodynamics at one year with both peak and mean pressure gradients lower in the bovine pericardial bioprosthetic valve when compared with the porcine cohort, 2 mm hg and 4 mm hg [5]. Tis fnding suggests a plausible mechanism for the increased integrity and functionality of pericardial bioprosthetic valves. However, no diference in mortality was observed despite the diferences in the two groups. Tis trial was limited, however, by its short-term follow-up and lack of disease heterogeneity in the patient population.
When examining long-term survival, prior reports are conficting. Grunkemeier et al. [3] reported a lower rate of mortality within the porcine group vs. the bovine group (57% vs. 50%, respectively; p � 0.04) in a retrospective cohort of 312 patients undergoing MVR with a bioprosthetic valve. In the same cohort, the authors also reported a greater risk of valve explantation at 15 years in those with porcine MVR when compared with their bovine pericardial counterparts (22% ± 3.7% vs. by 15 years vs. 8% ± 3.8% by 8 years). However, in a propensity-matched cohort of 802 patients who underwent either porcine or bovine pericardial MVR in a single institution, there was no statistically signifcant diference in survival at both 10 and 15 years between the cohorts [2]. In another single-center retrospective review of 154 pericardial MVR and 120 porcine MVR, the actuarial survival rate was calculated at 10 years with those who underwent porcine MVR 96.4 ± 0.08% vs. 94.6 ± 0.09% (p < 0.06) [15]. In another retrospective cohort study by Kim et al. [16] of 241 bovine pericardial MVR vs. 68 porcine MVR patients, no signifcant diference in shortterm or long-term survival was seen. Tis continues the trend seen in more recent retrospective studies of overall survival. In our overall cohort, we did observe an increased mortality in the porcine group; however, these patients also were more likely to have preexisting CAD and COPD We saw no diference in risk of reoperation in both our unmatched and matched cohorts. Again, the existing literature is conficting regarding these outcomes. Previously reported studies have demonstrated good long-term durability of bovine pericardial valves in the mitral position such as that of Bourguignon et al. [17] who reported that the

Cardiology Research and Practice
"freedom from valve-related mortality was 61.7% ± 8.9% at 20 years." Tis longevity sets a high bar for long-term integrity, and early studies comparing pericardial vs. porcine MVR showed no efect of prosthesis type on the risk of structural valve deterioration [18]. Grunkemeier et al. [3] demonstrated that pericardial MVR had a lower overall risk of explantation at 10 years compared to porcine HR of 0.53 (0.34-0.81, p � 0.003); however, when stratifed to    [2] reported a high risk of reoperation for those who underwent MVR with a bovine pericardial bioprosthetic, and the rate of reoperation for any reason was 1.89 times higher in those who underwent MVR with pericardial bioprosthetics [2]. Even more strikingly, those who underwent MVR with a pericardial valve were 2.32 times more likely to undergo reoperation due to structural valve deterioration. Furthermore, the time to reoperation was shorter for those who underwent pericardial MVR, that is, 6.8 ± 2.3 years vs. 11.1 ± 2.3 years for porcine MVR (p < 0.001). Similarly, Ramen et al. [15] reported a lower risk of valve explantation due to SVD, NSVD, and all-cause reoperation in the porcine vs. pericardial bioprosthesis groups. Kim et al. [16] reported results consistent with these showing no signifcant diference in SVD between either pericardial or porcine prosthesis. In the past, there was concern that porcine valves had an increased risk of early reoperation compared with their pericardial counterparts [14,19]. However, observational studies investigating more recent generations of pericardial and porcine bioprosthetic valves have not reproduced this pattern of reoperation [19,20].

Limitations
Our study was limited by its retrospective nature and lack of randomization. Tis study also sufered from diferent techniques and care strategies that varied between institutions and providers. However, this heterogeneity ofers some benefts. Our larger sample size and variety of hospital environments may make our fndings more applicable to a broader range of patients. Unfortunately, our registry did not collect anticoagulation and antiplatelet data of our patient population, and therefore, the role of these medications in this cohort cannot be ascertained with certainty. Tat said, there was no diference observed in bleeding (p � 0.11) or stroke (p � 0.44) or in the overall cohort between porcine and pericardial valves and given the lack of long-term anticoagulation with bioprosthetic valves may suggest that no diference in short-term bleeding was observed. However, further investigation is necessary before drawing any conclusions regarding the role of anticoagulant and antiplatelet use in mortality and risk of reoperation in these patients. Furthermore, there was a lack of complete echocardiographic data for all of our patients' post-MVR. Terefore, determining SVD via standardized defnitions such as those proposed by the VIVID group could not be established [21]. Given this lack of echocardiographic data, our study relied on the surrogate end point of valve explant, and we were unable to track the specifc causes of valve explantation. Tis raises the question that if SVD in one of these groups may have been a greater cause of valve failure. However, SVD is only one of the innumerable factors that may lead to the need for valve explantation. Terefore, while SVD is an important metric of long-term valve function, we believe the overall rate of explantation more accurately describes the myriad efects and durability of a given prosthetic.

Conclusion
In this multicenter analysis of patients undergoing bioprosthetic MVR, there was no diference in perioperative complications, long-term survival, or risk of reoperation between bovine pericardial and porcine prostheses. Tese data suggest that surgeon preference can be used to determine the best bioprosthesis for each patient.

Data Availability
Te data that support the fndings of this study are available from the corresponding author upon request.

Disclosure
Tis research was previously presented at the 2021 American Heart Association Scientifc Session meeting, and the published abstract can be download from the following link: "https://www.ahajournals.org/doi/10.1161/circ.144.suppl_1. 11016." Cardiology Research and Practice 7