Comparison of Outcomes After Transcatheter Versus Surgical Repeat Mitral Valve Replacement

Introduction Repeat transcatheter mitral valve replacement (rTMVR) has emerged as a new option for the management of high-risk patients unsuitable for repeat surgical mitral valve replacement (rSMVR). The aim of this study was to compare hospital outcomes, survival, and reoperations after rTMVR versus surgical mitral valve replacement. Methods We compared patients who underwent rTMVR (n=22) from 2017 to 2019 (Group 1) to patients who underwent rSMVR (n=98) with or without tricuspid valve surgery from 2009 to 2019 (Group 2). We excluded patients who underwent a concomitant transcatheter aortic valve replacement or other concomitant surgery. Results Patients in Group 1 were significantly older (72.5 [67-78] vs. 57 [52-64] years, P<0.001). There was no diference in EuroSCORE II between groups (6.56 [5.47-8.04] vs. 6.74 [4.28-11.84], P=0.86). Implanted valve size was 26 (26-29) mm in Group 1 and 25 (25-27) mm in Group 2 (P=0.106). There was no diference in operative mortality between groups (P=0.46). However, intensive care unit (ICU) and hospital stays were shorter in Group 1 (P=0.03 and <0.001, respectively). NYHA class improved significantly in both groups at one year (P<0.001 for both groups). There was no group effect on survival (P=0.84) or cardiac readmission (P=0.26). However, reoperations were more frequent in Group 1 (P=0.01). Conclusion Transcatheter mitral valve-in-valve could shorten ICU and hospital stay compared to rSMVR with a comparable mortality rate. rTMVR is a safe procedure; however, it has a higher risk of reoperation. rTMVR can be an option in selected high-risk patients.


INTRODUCTION
Elderly and frail patients are more frequently submitted to reoperative cardiac surgery due to the aging of the population and the advancement of surgical techniques. At least 4% of patients who had a mitral valve repair or replacement will require repeat mitral valve surgery [1,2] . Despite the excellent results achieved after mitral valve repair [2] , re-repair may not be feasible in the second operation, and mitral valve replacement (MVR) is required [3] . Recent research showed marked improvement in repeat MVR outcomes, and the results were comparable to the primary MVR [4] . Although there is a marked improvement in the surgical outcomes of repeat surgical mitral valve replacement (rSMVR), several patients are not considered for surgery due to high surgical risk.
Repeat transcatheter mitral valve replacement (rTMVR) has emerged as a new option for managing high-risk patients. Early results of rTMVR were encouraging; however, the generalization of the technique to a lower-risk patient requires extensive studies [5] . In a benchmark study, Ejiofor et al. reported a 5% mortality for rSMVR after a previous mitral valve repair and 9% after a previous replacement. Long-term survival was lower in patients with prior replacement [6] . Studies comparing clinical and echocardiographic outcomes after rSMVR and rTMVR are limited, and no randomized trials were performed to compare both approaches [7] . The aim of this study was to compare hospital and echocardiographic outcomes, survival, and reoperations after repeat transcatheter versus surgical mitral valve replacement.

Design and Patients
We performed a retrospective study to compare patients who underwent rTMVR and rSMVR at Prince Sultan Cardiac Center, Riyadh, Saudi Arabia. The study included patients who underwent transcatheter mitral valve-in-valve (n=21) or mitral valve-in-ring (n=1) from March 2017 to July 2019 (Group 1). These patients were compared to patients who underwent rSMVR (n=98) with or without tricuspid valve surgery from April 2009 to October 2019 (Group 2). We excluded patients who underwent a concomitant transcatheter aortic valve replacement or other concomitant surgery and reoperative MVR without prior mitral valve surgery. The study flowchart is shown in Figure 1.
The Institutional Review Board of the Prince Sultan Cardiac Center approved the data collection for this study (Reference Number: R19022), and patients' consent to participate in the study was waived.

Data Collection and Study Outcomes
Data were collected via paper and electronic medical records review. Preoperative data included patients' demographics, comorbidities, risk stratification using EuroSCORE II, preoperative renal function, left ventricular ejection fraction (LVEF), left ventricular end-diastolic diameter (LVEDD) and left ventricular end-systolic diameter (LVESD), and pulmonary artery systolic pressure (PASP).
All patients underwent pre-and postoperative echocardiography. Echocardiographic measurements were collected preoperatively, pre-discharge, and after 6, 12 and 18 months.
Study outcomes included in-hospital complications, intensive care unit (ICU) and hospital stay, cardiac readmissions, mitral valve reoperations, survival, and changes in echocardiographic measurements.

Patient Assignment and Techniques
During the transcatheter mitral valve-in-valve era, patients were considered for this technique after heart team discussion. Patients who were eligible for surgery but refused surgical interventions were offered the transcatheter option (n=8). Patients with infective endocarditis, mitral valve vegetations, left atrial thrombus, and those with a mitral valve size <25 mm were not considered for rTMVR. All patients underwent rTMVR via a transseptal approach, and our transcatheter mitral valve-in-valve technique was previously described [8] . Surgical mitral valve replacement was performed via median sternotomy in all patients.
Postoperative anticoagulation was similar in both groups. It included warfarin and acetylsalicylic acid (AAS) for three months, followed by life-long AAS unless patients had other indications for warfarin.

Statistical Analysis Data Presentation
Stata 16.1 (Stata Corp, College Station, Texas, USA) was used for all statistical analyses. We performed an intention-to-treat analysis to simulate clinical trials. Continuous data were presented as the 25 th , 50 th (median), and 75 th percentiles. Normality was tested using the Shapiro-Wilk test, and the Wilcoxon rank-sum test was used to compare continuous variables. Chi-square test was used for categorical variables and, if the expected frequency was <5, Fisher's exact test was used. We used the McNemar's test to compare dependent categorical variables.

Regression Models
Negative binomial regression was used to test the effect of the group and EuroSCORE II on postoperative hospital and ICU stay. Logistic regression analysis was used to identify the factors affecting hospital mortality, and Hosmer-Lemeshow and area under the curve were used to test the quality of the model.
Mixed-effects linear regression analysis was used to compare changes in the echocardiographic measurements between the two groups (LVEF, PASP, and mean mitral valve pressure gradient). The measurements were recorded at fixed times, preoperatively, pre-discharge, after 1 year, and after 18 months. The model yielded two values, the baseline measurements and the degree of change. The significance of the change was evaluated over time and compared between the two groups. The mixed-effect model included group, time, and baseline value.

Time-to-Event Analysis
We compared three time-to-event variables (survival, reoperation, cardiac readmission) between the two groups. Kaplan-Meier method was used to plot the survival distribution for time-to-event variables, and the log-rank test was used to compare curves. Multivariable Cox regression was used to evaluate the effect of the surgical approach on time-to-event variables, and the proportional hazard assumption was tested using Schoenfeld residuals method.

Predictors of Hospital Outcomes
ICU and hospital stays were significantly longer in Group 2 and with a higher EuroSCORE II. The groups did not affect the operative mortality. Mortality was higher with a higher EuroSCORE II (Table 4).

One-Year Follow-Up
NYHA class improved significantly in both groups after one year compared to the preoperative value (P<0.001 for both groups). There was no difference in NYHA class between the two groups at 1-year follow-up (P=0.583).

Changes in Echocardiographic Measurements
The groups did not influence changes in LVEF, PASP, and mean mitral valve pressure gradient (Table 5) ( Supplementary  Figures 1-3).

DISCUSSION
Transcatheter mitral valve-in-valve replacement is an emerging new technology, which is considered as an alternative option to surgical reoperative MVR in patients with prohibitive or high surgical risk. The technique was listed in the European Society of Cardiology (ESC)/European Association for Cardio-Thoracic Surgery (EACTS) Guidelines (2017) as an alternative option for the management of degenerated bioprostheses in high-risk Table 5. Mixed-effects REML regression for the changes in left ventricular ejection fraction, pulmonary artery systolic pressure, and mean mitral valve pressure gradient.

Ejection fraction
Coef.   surgical patients [9] . We performed this study to compare rTMVR and rSMVR. Patients who underwent rTMVR were older and had higher PASP. Other preoperative variables, including EuroSCORE II, were comparable. There was no difference in operative mortality between the two groups, and the length of ICU and hospital stay was significantly shorter in rTMVR. We did not observe any significant difference in PASP, LVEF, and mean mitral valve pressure gradient changes over the follow-up between groups. Survival and cardiac readmission were similar in both groups; however, reoperation was significantly higher in patients who underwent rTMVR.

P-value
All patients in our rTMVR group had a transseptal approach, which played an important role in decreasing the ICU and hospital stay [10] . Additionally, this approach was associated with a lower bleeding rate than the transapical approach [11,12] . Computed tomography (CT) scan was not required for planning the transcatheter approach but was an essential part of the preoperative evaluation before rSMVR. No dye was used during rTMVR, and the ring of the mitral valve prosthesis was used to localize the valve. EuroSCORE II was comparable between groups, which can be explained by including 8 patients in the rTMVR group with low EuroSCORE who refused to undergo surgery.
We did report a significant difference in operative mortality, similar to the findings of Kamioka et al. [7] . They reported a 30-day mortality of 3.2% after rTMVR and 3.2% after rSMVR, which is lower than our results. Our mortality is within the range reported in the literature [13,14] . In the Society of Thoracic Surgeons' annual report, the in-hospital mortality in high-risk patients who underwent transcatheter mitral valve-in-valve was 7.2%. The 30-day mortality was 8.5% [15] , which is comparable to that of those who underwent transcatheter mitral in our results. In a meta-analysis of transcatheter mitral valve-in-valve procedures, the 6-month mortality was 23% [16] , and it was 13.5% in our study. The nonsignificant difference in hospital mortality in our series could be attributed to the comparable EuroSCORE II between groups, which was a significant predictor of mortality. Two-year survival was 74% and 90% in rTMVR and rSMVR groups, respectively. However, this difference did not reach statistical significance.
The mean mitral valve pressure gradient was not different between groups both at discharge and during follow-up. This includes patients who underwent a mechanical or bioprosthetic mitral valve replacement. The mean mitral valve pressure gradient reported in our series was comparable to several reports [5,7,11] .
Since the transcatheter procedure was valve-in-valve, a higher pressure gradient was expected. However, patient selection may contribute to the non-significant difference between the two groups. The transcatheter approach was not used in patients with small valves (<27 mm), making patient-prosthesis mismatch a low probability.
No studies to our knowledge have compared the long-term outcomes after rTMVR and rSMVR. In the present study, we found that both approaches improved clinical symptoms with no difference in survival and cardiac readmission between groups. However, patients who underwent rTMVR had a higher rate of reoperation. The high incidence of reoperation in this group could be attributed to the learning curve since most of these operations were required early. Five patients who underwent rSMVR required reoperation at a median follow-up of 36 months compared to 15 months in patients who underwent rTMVR. Conclusion about the potential earlier degeneration of transcatheter valves cannot be drawn from our study, and further studies are required.
Our study showed that the outcomes of rSMVR and rTMVR are comparable. Both techniques improved clinical outcomes and patients' symptoms. Patients who had left atrial thrombus and endocarditis, in addition to those with small implanted valves, should be considered for surgical MVR. A randomized trial is recommended to compare both approaches in patients who are considered to be at high risk for surgery.

Limitations of the Study
The main limitation of our research is the retrospective nature of the study. Patients assigned to each group were different, and the assignment was confounded by indication. However, we performed a multivariable regression analysis for the main variables that may affect the outcomes. Another limitation is the shorter follow-up period, which is attributed to the recent introduction of the transcatheter approach. The sample size is relatively small, but we created a restricted cohort study by applying rigid inclusion criteria for surgical and transcatheter approaches. Patients who had concomitant procedures, apart from tricuspid valve reintervention, were excluded. This was