Transcatheter aortic valve implantation versus surgical aortic valve replacement for pure aortic regurgitation: a systematic review and meta-analysis of 33,484 patients

We searched PubMed, Embase, Web of Science (WOS), Scopus, and the Cochrane Library Central Register of Controlled Trials (CENTRAL) from inception until 23 June 2023. We utilized the keywords aortic regurgitation, TAVI, and SAVR. The search strategy for each database is illustrated in Supplementary Table 1.

We used the population, intervention, comparator, outcomes, and study design (PICOS) selection criteria to determine the included studies. We included studies with the following PICOS criteria. (1) Studies including patients with pure AR. (2) Intervention is TAVI. (3) Comparator is SAVR. (4) Studies included in-hospital mortality or stroke among the reported outcomes. (5) RCTs or cohort studies. We excluded single-arm studies, studies with more than one publication, studies including AS patients or patients with mixed AR and AS, case reports, reviews, abstracts, and animal studies.

The articles retrieved through the systematic search were uploaded to the EndNote Reference Library, where duplicates were determined and removed. After duplicates had been removed, the titles and abstracts of the search results were uploaded to the Rayyan website [28] and screened for relevance by two authors. Potentially eligible studies were then retrieved for full-text screening. The final list of included trials was agreed upon by discussion between all authors. Disagreement amongst reviewers was resolved through consensus. The reference lists of the retrieved studies were manually screened for any additional eligible studies.

Extraction forms were constructed on Google Spread Sheets. Two authors extracted the data separately, and a third author solved disagreements. We extracted the following information for each study. (1) Summary of the included studies (the last name of the first author and the year of publication, study design, country, duration, sample size, details of each procedure, inclusion and exclusion criteria, implanted valve type, valve size, valve calcification, and duration of follow up). (2) Baseline characteristics including (Age, gender, body mass index (BMI), New York Heart Association classification of heart failure stage (NYHA), history of atrial fibrillation (AF), hypertension, diabetes, renal disease, liver disease, congestive heart failure, peripheral vascular disease, stroke or transient ischemic stroke, coronary artery disease, myocardial infarction (MI), previous percutaneous coronary intervention, and previous coronary artery bypass grafting surgery). (3) And outcomes including (mortality, major adverse composite cardiac events (MACCE), in-hospital stroke, MI, acute kidney injury (AKI), delirium, major bleeding, AF, pneumonia, sepsis, permanent pacemaker implantation (PPI), reintervention, and length of hospital stay (LOS)).

The Newcastle Ottawa scale (NOS) for quality assessment of non-randomized trials [29] was used for the quality assessment of the included studies. The NOS assigns a maximum of nine points for the three domains: (1) Selection of study groups (four points); (2) Comparability of groups (two points); and 3) Ascertainment of exposure and outcomes (three points). NOS’s total score of 0 to 3 indicates a high risk of bias, 4 to 6 indicates a moderate risk, and ≥ 7 indicates a low risk of bias. Two independent authors performed the quality assessment separately. A third author resolved any disagreement.

We used Review Manager (RevMan Version 5.4.1, The Cochrane Collaboration, 2020) for statistical analysis. To compare dichotomous outcomes, we used a risk ratio (RR) with a 95% confidence interval (CI) and the Mantel-Haenszel method; A p-value less than 0.05 was considered statistically significant. For continuous outcomes, we utilized the mean difference (MD) and 95% CI using the inverse variance method, and a p-value less than 0.05 was considered statistically significant. If we were comparing continuous outcomes with different measurement units, we used the standardized mean difference (SMD) and 95% CI instead of the MD; We only used this method when comparing the cost of procedures, as it was reported in US dollars and European Union euros. We used the inconsistency test (I²) to assess statistical heterogeneity and considered it significant when the I² statistic exceeded 50% or had a p-value less than 0.10. We used the random effect model in our analysis. We meta-analyzed results for in-hospital, 30 days, and 1 year after the procedures. We did subgroup analysis depending on the approach of TAVI (transfemoral and transapical) and depending on the country of origin of the included studies. In the case of heterogeneity, we did a leave-one-out test by excluding one study in each scenario in order to eliminate the heterogeneity. We could not test for publication bias using Egger’s test since we did not include at least 10 studies for any of the outcomes, which is necessary to obtain accurate results [30].

We evaluated the strength of evidence using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) scale. This scale assesses the certainty of the effect estimate. It has seven domains, including the number and design of studies, their quality, heterogeneity between pooled results, the direct effect of interventions, the precision of the confidence interval, and other considerations. GRADE categorizes evidence certainty into four levels: high quality, indicating that further research is unlikely to change the effects estimates; moderate quality, suggesting that further studies may affect confidence in effect estimation; low quality, indicating that additional research is crucial and likely to significantly impact the confidence of the effect estimation and may change the estimation; and very low quality, indicating uncertainty about the estimation.

Our search retrieved 1792 articles after removing the duplicates. Only 115 titles were eligible for full-text screening following the title and abstract screening. Finally, six studies were included [18‐21, 31, 32] with a total of 5633 patients for TAVI and 27,851 patients for SAVR (PRISMA flow diagram; Fig. 1). All included studies were retrospective cohort studies, and the majority (three studies) were conducted in the United States, followed by Germany (two studies), and one study conducted in China. All included studies involved only patients with isolated AR. Table 1 summarizes the included studies and their most significant findings. Age in the TAVI group ranged from 77 ± 12.62 years to 67 ± 6.77 years, versus 75.6 ± 5.7 years to 60.9 ± 14.1 years in the SAVR group. The detailed baseline characteristics of patients in the included studies are illustrated in Table 2.

NOS determined that all included studies posed a low risk of bias. Table 3 displays the detailed quality assessment domains of the included studies.

The results of the strength of the evidence according to GRADE are summarized in Supplementary Table 2. The GRADE system classified the strength of evidence as moderate for in-hospital stroke. Low for in-hospital mortality, AKI, major bleeding, and PPI. And very low for AF, delirium, and sepsis.

The present study aimed to compare the safety and efficacy of TAVI and SAVR for the treatment of AR. The meta-analysis revealed important findings regarding mortality rates, procedural complications, and healthcare system utilization. The mortality rates between the two procedures were comparable. TAVI demonstrated advantages over SAVR in terms of stroke, major bleeding, AKI, pneumonia, and shorter LOS. However, TAVI was associated with a higher risk of PPI.

Franzone et al. 2016 [23], Jiang et al. 2017 [22], and Takagi et al. 2020 [24] conducted single-arm meta-analysis studies of the feasibility of TAVI in AR patients. A small number of patients limited these studies and did not compare TAVI with SAVR. To the best of our knowledge, our study is the first to compare TAVI with SAVR in patients with pure AR.

The mortality rates between TAVI and SAVR were comparable. This finding aligns with previous research and supports the notion that TAVI is not inferior to SAVR regarding overall patient survival in AR [18, 19, 31] and also AS patients [33‐35]. In our cohort, TAVI patients were older and had higher comorbidity scores, which aligns with the current recommendations and practice directions that TAVR is assigned only to patients with higher risk who cannot undergo surgery [3, 5]. This may be why TAVR was associated with a lower number of deaths but did not reach statistical significance. Franzone et al. 2016 [23], Jiang et al. 2017 [22], and Takagi et al. 2020 [23] reported a 30-day all-cause mortality after TAVI of 8%, 12, 9%, and 9.5, respectively. Our study found that the rate of all-cause in-hospital mortality after TAVI was 3.1%. Although our study had a shorter follow-up period (in-hospital) compared to previous single-arm meta-analysis studies, we believe that the mortality rate after TAVI is decreasing due to the growing experience of the operators. The most recent studies report a mortality rate after TAVI of 3.7 and 2.01% among patients treated between 2018 and 2020 and 2016–2019, respectively [21, 31]. The source of heterogeneity in our analysis was Stachon et al. 2020 [20]; after excluding it, the overall RR favored the TAVI over SAVR (RR = 0.72; 95% CI: [0.59, 0.89], P = 0.003). Stachon et al. 2020 [20] reported an increased risk of in-hospital mortality with TAVI compared to SAVR. The different results may be because it was an early study conducted in 2008, which may resemble a limited experience of operators and less advanced technology. The sensitivity analysis after excluding Stachon et al. 2020 [20] suggests that TAVI may be associated with a decreased mortality rate than SAVR.

The decreased risk of postoperative stroke to half the incidence after SAVR is a very interesting finding of this meta-analysis, which suggests the superiority of TAVI in AR patients. Our results align with previous reports [21, 31].

Although TAVI has made significant progress over the past 20 years with improved devices and cardiologist experience, stroke remains a major concern associated with this procedure with AS. It can increase mortality rates and decrease the patient’s quality of life. Different reports have shown that the incidence of stroke within 30 days post-TAVI ranges from 0.6 to 6.7% with AS [9‐11, 36, 37]. Stroke after TAVI can have varying symptoms, from disabling or non-disabling stroke to silent stroke detected only by diffusion-weighted magnetic resonance imaging (DW-MRI) [38, 39]. 75% of AS patients experience debris embolization during TAVI of the calcified aortic arch and valve, posing a risk of stroke of any type [40].

Previous single-arm meta-analyses reported that stroke was an infrequent event in patients with AR who underwent TAVI. Franzone et al. 2016 [23] reported a 0% incidence of stroke, Jiang et al. 2017 [22] reported a 0.01% incidence of stroke, and Takagi et al. 2020 [24] reported a 2.9% incidence of stroke. However, these studies were limited by a small number of patients; our study also found similar findings, with only 1.8% of patients experiencing stroke in the TAVI group. AR is not always associated with calcification [41], so the decreased risk of stroke may be due to a different underlying pathology.

TAVI demonstrated advantages over SAVR, including bleeding, AKI, pneumonia, and shorter LOS, aligning with previous research [18, 19, 21, 32, 42, 43]. These benefits are consistent with the less invasive nature of TAVI, which avoids sternotomy and cardiopulmonary bypass. The shorter LOS associated with TAVI compared to SAVR suggests potential cost savings and improved resource allocation within healthcare systems. These findings can inform healthcare providers, policymakers, and administrators in making informed decisions regarding adopting and allocating resources for these interventions.

The increased risk of PPI after TAVI is not surprising, as shown in previous research on patients with AS and AR [18, 21, 35, 43, 44]. Franzone et al. 2016 [23] reported a PPI after a TAVI rate of 11%, Jiang et al. 2017 reported 11% [22], and Takagi et al. 2020 reported 11.6% [24]. The rate of PPI after TAVI in our study was near to the previous results (13.06%). It is assumed that injury to the superficial atrioventricular and left bundles during implantation is the direct cause of PPI after TAVI. Regardless of the condition, this injury is believed to be related to the difficulty of anchoring the valve. The superior anchoring mechanisms of the newer valves resulted in better PPI outcomes, but this improvement was not statistically significant [13]. This finding highlights the need for careful patient selection and diligent postoperative monitoring, particularly in patients at risk for conduction disturbances.

Healthcare providers need to assess the individual risk of each patient and involve them in the decision-making process. Factors such as age, comorbidities, anatomical considerations, and the surgeon’s expertise should be considered when determining the best procedure for a patient.

This study investigates the comparative efficacy of TAVI and SAVR in managing pure AR. The study’s strength lies in its clinical relevance, as it addresses a pressing issue in cardiology. It directly compares TAVI and SAVR in a meta-analysis for the first time, including six studies and a large number of pure AR patients. The study aimed to provide valuable insights into each intervention’s relative benefits and risks. This study investigated a large scope of clinical outcomes. Also, it provided a valuable subgroup analysis to deepen our understanding of each TAVI approach separately and if the results differ from one country to another. We were able to address the source of heterogeneity in many outcomes. The findings can potentially guide clinical decision-making and improve patient care by informing physicians about the most appropriate treatment option.

Considering the economic implications of TAVI and SAVR on healthcare utilization further enhances the study’s impact. Ultimately, this research fills a knowledge gap and advances our understanding of aortic valve disease management, making it valuable for clinicians and policymakers.

While the study has several strengths, it is also important to acknowledge its weaknesses. We have included various types of implanted valve devices, but determining the specific role of each valve is still necessary. Additionally, all the included studies are retrospective, which may limit our ability to control confounding variables adequately. The lack of RCTs to compare TAVI and SAVR could also affect the study’s robustness. While short-term outcomes are highly interesting, the included studies’ lack of long-term follow-up periods limits our understanding of intermediate- and long-term outcomes. Lastly, this study highlights the need for further research. The heterogeneity observed among the individual studies emphasizes the complexity and variability in outcomes associated with TAVI and SAVR. Future studies should aim to identify the factors contributing to this heterogeneity and explore additional efficacy and safety outcomes to provide a more comprehensive understanding of these interventions.

Future research directions should address the limitations of this study and further explore specific subgroups of patients. Large-scale prospective studies are needed to validate the findings and investigate the impact of evolving technologies and techniques. Further, longer-term follow-up studies with detailed efficacy and echocardiographic outcomes are warranted to confirm these results and assess potential differences in durability and valve-related complications. It would be highly important to analyze the performance of TAVI and SAVR in patients with different surgical risks to draw a definitive conclusion. Lastly, comparative cost-effectiveness analyses would also provide valuable insights for healthcare decision-makers.