Reintervention with Transcatheter and Surgical Aortic Valves: A Systematic Review and Meta-Analysis

Background: Despite expanding indications, data regarding the long-term durability of transcatheter heart valves (THV) are limited. Methods: We performed a systematic review and meta-analysis of all published studies with ≥5years of follow-up reporting aortic valve reintervention rates of transcatheter (TAVR) and surgical aortic valve replacement (SAVR). Randomized controlled trials (n = 4) and propensity-matched observational studies (n = 1) involving all surgical risk categories were included. The primary endpoint was the composite of aortic valve reintervention and death. Results: The meta- analysis included 4145 patients: 2101 underwent TAVR (mean age 81.7 ± 6.7 years, 54% male) and 2044 SAVR (mean age 81.8 ± 6.6 years, 54% male). All TAVR procedures were performed with early generations of THV. At a median follow-up of 5 years (range 5 - 6 years), TAVR had higher reintervention rates (odds ratio (OR) 3.33; 95% CI: [1.78, 6.24], p 2 = 0%), all-cause mortality (OR 1.45; 95% CI: [1.22, 1.75], p 2 = 44%) and the composite of reintervention and death (OR 1.47; 95% CI: [1.14, 1.91], p 2 = 64%). Rates of myocardial infarction, transient ischemic attack, stroke, endocarditis, and the composite of endocarditis and thrombosis were similar between the groups. Conclusion: Despite comparable short and medium-term results, TAVR with early-generation THV has higher rates of reintervention and the composite of reintervention and death. Further studies employing newer definitions of structural valve deterioration and bioprosthetic valve failure are needed to assess whether technological enhancements in THV technology will improve long-term outcomes.


Introduction
In patients with severe aortic stenosis and prohibitive or high surgical risk for aortic valve replacement (SAVR), multiple trials have shown that transcatheter aortic valve replacement (TAVR) is non-inferior to SAVR [1] [2]. These trials resulted in Class 1 evidence indication for TAVR in those with prohibitive or high surgical risk [3]. Recently, major randomized controlled trials have shown TAVR non-inferiority in intermediate and low surgical-risk patients [4] [5]. The expanding indications for TAVR to include lower-risk patients with longer life-expectancy highlight the importance of the long-term durability of transcatheter valves (THV).
There is limited evidence evaluating TAVR outcomes at long-term follow-up, despite the short and medium-term non-inferiority. Therefore, we performed a systematic review and meta-analysis of all studies comparing TAVR to SAVR with a minimum of five years of follow-up to determine the comparative rates of reintervention and death.

Literature Search
We performed the study according to the proposal for conducting and reporting meta-analyses of observational studies [6] and Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (PRISMA) [7]. We performed a computerized search through Medline, Embase, and Cochrane databases from January 2000 to November of 2020. The terms "transcatheter aortic valve replacement", "TAVR", "surgical aortic valve replacement" and "SAVR" were used in combination with "reintervention", "durability", "structural valve deterioration", "SVD", "bioprosthetic valve failure" and "BVF". Bibliographies of the retrieved studies were screened for relevant studies. Our search was limited to the English language.

Study Selection
We included randomized controlled trials (RCTs) and propensity-matched observational studies that compared outcomes with TAVR versus SAVR with clinical follow-up of at least five years. We excluded all non-randomized and singlearm studies, as well as those including only TAVR patients. Due to the exclusion of all studies with ≤5 years of follow-up, only TAVR with early-generation THVs were included (Corevalve, SAPIEN, and SAPIEN XT). Data from the Placement K. G. Buda et al. World Journal of Cardiovascular Diseases of Aortic Transcatheter Valves (PARTNER) 1B trial [8] was not included, as it did not report reintervention. Of the two publications reporting the results of the Nordic Aortic Valve Intervention (NOTION) trial at ≥5 years of follow-up, only the one with data on reintervention, death, and major adverse cardiovascular events (MACE) was included [9]. Similarly, we utilized the PARTNER 2A trial comparing SAPIEN XT and SAVR for clinical outcomes [10], excluding the propensity-matched analysis which included the non-randomized SAPIEN-3 registry [11]. Though five of the studies reported structural valve deterioration (SVD) or bioprosthetic valve failure (BVF), this data was excluded from our analysis due to: 1) use of standardized definitions in only four of the studies; 2) the dissimilarity in definitions; 3) definitions that were set after trials and not a priori; and 4) the lack of independent SVD and BVF adjudication.

Data Extraction and Quality Assessment
The data were reviewed and extracted by two independent investigators (KB, MM). Discrepancies were settled by consensus. The bias risk of the included studies was assessed using the New-Castle Ottawa Scale for cohort studies and the Cochrane risk assessment tool for RCTs [12].

Data Synthesis and Statistical Analysis
Statistical analysis was conducted using Review Manager Software (Version 5.4.1. Copenhagen: The Nordic Cochrane Centre, the Cochrane Collaboration, 2014). Categorical variables were reported as frequencies, while continuous variables as means with standard deviations (SD). Categorical variables were compared using Fisher's exact or Chi-square tests, while continuous variables were analyzed using the two-sample t-test. Tests were two-tailed, and a p-value of ≤ 0.05 was considered statistically significant. All reported baseline characteristics and outcomes are weighted by sample size.
Odds ratios (ORs) and mean differences (MD) with 95% confidence intervals (CIs) are presented as summary statistics. Statistical heterogeneity was assessed by I 2 statistics: I 2 statistic > 50% was considered substantial, and I 2 > 75% was considered considerable [13]. We used the Der-Simonian and Laird random-effects and random-effects generic inverse variance methods to calculate OR and MD, respectively, as we anticipated a high degree of clinical and methodological heterogeneity. Potential publication bias was assessed using the Egger test by visual examination of the funnel plots [14].

Study Selection and Study Criteria
The study selection process is described in Supplemental Figure S1. We ana-  (Table 1). All studies included patients with severe aortic stenosis. The average operative risk was considered high in two studies [15] [16] and intermediate to low in three studies [9] [10] [17]. The surgical risk was defined by the STS score.
Supplemental Table S1 has a complete list of study inclusion criteria and definitions. Bias assessment was determined using the New-Castle Ottawa Scale for observational studies and the Cochrane assessment tool for RCTs (Supplemental Table S2 and Table S3).

Patient Characteristics
The TAVR group included 2101 patients (mean age 81.7 ± 6.7 years, 54% male) and the SAVR group included 2044 patients (mean age 81.8 ± 6.6 years, 54% male). The median follow-up duration was 5 years (range 5 -6 years) for clinical outcomes. There were a higher percentage of patients with atrial fibrillation in the SAVR group, though the numeric difference was small. There were no other differences in baseline characteristics between the groups ( Table 2). Summary statistics are listed in Table 3.

Discussion
Our main findings can be summarized as follows: 1) TAVR with first and second-generation THV devices was associated with higher rates of reintervention, the composite of reintervention and death, and all-cause mortality relative to SAVR at ≥5 years of follow-up; 2) rates of myocardial infarction, transient ischemic attack, and stroke were similar; and 3) there was no difference in the incidence of endocarditis or the composite of endocarditis and thrombosis. TAVR has many advantages, the most obvious of which is its availability in patients with prohibitively high surgical risk. Similarly, there is mounting evidence on the benefits of TAVR across the spectrum of surgical risks. As a result of the expanding indications for TAVR to include lower-risk patients with longer life expectancy, the long-term durability of transcatheter valves is becoming increasingly important. In our study, patients who underwent TAVR with earlygeneration THV devices had a higher rate of the composite of death or reintervention and all-cause mortality.
Recent data suggest that TAVR durability depends heavily on valve-subtype.
The propensity-matched analysis by Pibarot et al. compared the outcomes of the third generation SAPIEN 3 and the second generation SAPIEN XT THV and found a lower rate of SVD and BVF in the SAPIEN 3 cohort [11]. Similarly, The Comparison of Transcatheter Heart Valves in High Risk Patients with Severe Aortic Stenosis (CHOICE) trial compared the Edwards SAPIEN XT with the Medtronic CoreValve and found a higher rate of SVD in the SAPIEN XT patients. Aortic valve reintervention, albeit clinically important, lacks sensitivity and specificity to detect structural valve deterioration. For example, a patient with severe perivalvular regurgitation, which was more common with first and secondgeneration THV devices, might have required reintervention unrelated to structural valve degeneration. Conversely, a patient with significant structural valve deterioration might have been denied surgery due to prohibitive surgical risk. Recognizing these limitations, standardized definitions of structural valve deterioration and bioprosthetic valve failure have been proposed by the Valve Academic Research Consortium (VARC-3). SVD is defined as a composite of ≥Stage 2 hemodynamic valve deterioration by echocardiography and/or SVD-related bioprosthetic valve failure (BVF). BVF is defined as: 1) symptomatic bioprosthetic valve dysfunction or severe Stage 3 hemodynamic valve deterioration; 2) valve reintervention; or 3) valve-related death (Supplemental Table S4 and Table S5). Recently, the European Association of Percutaneous Cardiovascular Interventions (EAPCI), the European Society of Cardiology (ESC), and the European Association for Cardio-Thoracic Surgery (EACTS) also proposed standardized definitions of structural valve dysfunction, including SVD, non-structural valve deterioration, and BVF [18].

Limitations
Our study has several limitations. First, it has limited power to detect differences K. G. Buda et al. World Journal of Cardiovascular Diseases in clinical outcomes due to the small number of studies (n = 5), the inclusion of observational studies (n = 1), and events with a significant degree of heterogeneity. We attempted to overcome this limitation by excluding non-propensity-matched and non-randomized studies and by using a random-effects model in our analysis. Second, it is unknown to what degree the need for permanent pacemaker placement (higher following TAVR) affected our outcomes. Third, valve type and surgical risk were likely significant confounding factors. There is evidence that technological (outer skirts) and procedural enhancements (sizing with CT) have improved outcomes with newer generation THV [11]. Fourth, the THVs included in this meta-analysis are no longer commercially available in the US.

Conclusion
During long-term follow-up, TAVR with early-generation THV devices has higher rates of reintervention and the composite of reintervention and death compared with SAVR, despite comparable short and medium-term results. Further studies employing newer definitions of SVD are needed to assess whether improvements in THV technology will improve long-term outcomes.   All studies with 7 stars or higher are considered high-quality studies. World Journal of Cardiovascular Diseases   Table 2 for definitions of stages.

Supplemental
Non-structural valve dysfunction  Any abnormality, not intrinsic to the prosthetic valve, resulting in valve dysfunction. Examples include residual intra-or para-prosthetic aortic regurgitation; leaflet entrapment by pannus, tissue, or suture; inappropriate positioning or sizing; dilatation of the aortic root after stentless prostheses or aortic valve sparing operations; prosthesis-patient mismatch; and embolization.
Valve thrombosis  Subclinical: Imaging findings of hypo-attenuated (CT) or hypo-echogenic (echocardiography) leaflet thickening and/or reduced leaflet motion with absent of mild hemodynamic changes and no symptoms/sequelae.  Clinically significant: 1) Clinical sequelae of thrombo-embolic event or of worsening bioprosthetic valve stenosis or regurgitation and hemodynamic valve deterioration Stage 2 or 3 (See Online Hemodynamic valve deterioration may be caused by structural valve deterioration but also by non-structural dysfunction including valve thrombosis and endocarditis. The assessment of valve leaflet morphology and structure is key to make differential diagnosis between the different etiologies of hemodynamic valve deterioration: SVD versus valve thrombosis or endocarditis. ‡ This criteria for hemodynamic dysfunction assume normal flow. Figure S1. Systematic review process.