Transcatheter vs surgical aortic valve replacement in low to intermediate surgical risk aortic stenosis patients: A systematic review and meta‐analysis of randomized controlled trials

Abstract Background Transcatheter aortic valve replacement (TAVR) is regarded as the most superior alternative treatment approach for patients with aortic stenosis (AS) who are associated with high surgical risk, whereas the effectiveness of TAVR vs surgical aortic valve replacement (SAVR) in low to intermediate surgical risk patients remained inconclusive. This study aimed to determine the best treatment strategies for AS with low to intermediate surgical risk based on published randomized controlled trials (RCTs). Hypothesis and Methods RCTs that compared TAVR vs SAVR in AS patients with low to intermediate surgical risk were identified by PubMed, EmBase, and the Cochrane library from inception till April 2019. The pooled relative risks (RRs) with 95% confidence intervals (CIs) were calculated for the data collected using random‐effects models. Results Seven RCTs with a total of 6929 AS patients were enrolled. We noted that TAVR significantly increased the risk of transient ischemic attack (TIA) (RR: 1.43; 95%CI: 1.04‐1.96; P = .029), and permanent pacemaker implantation (RR: 3.00; 95%CI: 1.70‐5.30; P < .001). However, TAVR was associated with lower risk of post‐procedural bleeding (RR: 0.57; 95%CI: 0.33‐0.98; P = .042), new‐onset or worsening of atrial fibrillation (RR: 0.32; 95%CI: 0.23‐0.45; P < .001), acute kidney injury (RR: 0.40; 95%CI: 0.25‐0.63; P < .001), and cardiogenic shock (RR: 0.34; 95%CI: 0.19‐0.59; P < .001). The risk of aortic‐valve reintervention at 1‐ (RR: 2.63; 95%CI: 1.34‐5.15; P = .005), and 2 years (RR: 3.19; 95%CI: 1.63‐6.24; P = .001) in low to intermediate surgical risk patients who received TAVR was significantly increased than those who received SAVR. Conclusions These findings indicated that low to intermediate surgical risk patients who received TAVR had low risk of complications, whereas the risk of TIA, permanent pacemaker implantation, and aortic‐valve reintervention was increased.


| INTRODUCTION
Aortic stenosis (AS) is the most frequent heart valve disease seen in elderly population, in which nearly 1/8 individuals aged 75 years or over suffer from moderate to severe AS. 1,2 The prevalence of AS in North America and Europe is 12.4%, and this implied that there are more than 291 000 candidates undergoing aortic valve replacement. 3 The outflow of blood from the heart of AS patients is shown to be impaired, which subsequently increases cardiac workload and causes heart failure and left ventricular hypertrophy. According to a previous study, nearly 25% of mortality rates were observed annually in patients with symptomatic AS comorbidities such as angina, syncope or heart failure. 4 So, effective treatment strategies are necessary for AS patients.
Surgical aortic valve replacement (SAVR) along with artificial prosthesis is regarded as a conventional treatment strategy due to its effective choice of intervention in operable cases with severe AS. However, very elderly patients, and patients with calcified aorta or scarring after undergoing cardiac surgery are not intolerant to SAVR. Therefore, transcatheter aortic valve replacement (TAVR) is used for inoperable or high surgical risk AS patients due to its less invasive nature. 5,6 Technological advances in valve replacement procedure produced easy repositioning and removal, and its minimally invasive nature permitted conduction of TAVR under local anesthesia, and is also associated with shorter hospital stay, low risk of bleeding and less post-interventional complications. 7,8 Previous meta-analyses have demonstrated that TAVR had comparable or better early and midterm outcomes in AS patients with high surgical risk. [9][10][11][12][13] However, whether these results are suitable for low to intermediate surgical risk AS patients remains inconclusive.
Several RCTs on the research topic have been conducted, but inconsistent results were obtained from these. Clarifying optimal treatment strategy for AS patients with low to intermediate surgical The current study was performed according to the guidelines of preferred reporting items for systematic reviews and meta-analysis statement. 14 The PubMed, EmBase, and the Cochrane library were systematically searched for articles published till April 2019. The following search terms were used as medical subject headings and free-language terms: "Transcatheter aortic valve replacement" OR "TAVR" OR "TAVI"AND "Surgical aortic valve replacement" AND "SAVR" AND "SAVI"AND "low to moderate surgical risk" AND "severe aortic stenosis". Furthermore, trials that have been completed but not published in clinicaltrials.gov website were also searched. If essential information was unavailable from eligible publications, then corresponding authors were contacted.
The studies were searched and selected independently by two authors following a standardized flow. Disagreements between them were resolved by contacting an additional author through reviewing of the original article. The study selection process was based on PICOS criteria. The inclusion criteria were as follows: (1)  Observational studies were excluded due to the possibility of confounding variables or bias in the pooled results.

| Data collection and quality assessment
The data and quality of the included trials were collected independently by two authors following a standardized protocol, and any inconsistencies between them were settled by group discussion till a consensus was reached. The collected information was as follows: first author or study group's name, publication year, country, sample size, age, sex, society thoracic surgeons (STS) risk, logistic Euro SCORE I (LES), diabetes mellitus (DM), prior stroke, peripheral vascular disease (PVD), prior percutaneous coronary intervention (PCI), prior MI, chronic obstructive pulmonary disease (COPD), New York heart association (NYHA) III or IV, valve type, and reported outcomes. The quality of the included studies was assessed by Jadad scale based on random sequence generation, allocation concealment, blinding, intention-to-treat analysis, and completeness of follow-up, and a scoring system of 0 to 5 was used for assessing the study quality. 15

| Statistical analysis
The investigated outcomes from each RCT were assigned as dichotomous data, and the relative risks (RRs) and 95% confidence intervals (CIs) were calculated by using the event number extracted from each trial before data pooling. After this, the summary RRs and 95%CIs for investigated outcomes were calculated using randomeffects model as the true effect that underlies varies among the included trials. 16,17 Heterogeneity was assessed by using I-square and Q statistics across the included trials, and P < .10 was considered as statistically significant heterogeneity. 18,19 Sensitivity analyses were conducted for studies that reported outcomes ≥5 to assess the impact of single study from overall analyses. 20 Moreover, subgroup analyses were conducted for studies that reported outcomes ≥5 based on sample size, mean age, STS score, percentage of DM, prior stroke, prior PVD, prior MI, prior COPD, percentage of NYHA III-IV, valve type, follow-up duration, and study quality. After this, P values between subgroups were calculated using interaction tests, which were based on Student's t distribution. 21 Publication biases for studies that reported outcomes ≥5 were assessed by funnel plots, Egger, 22 and Begg tests. 23 The P values for pooled results are two-sided, and P < .05 was regarded as statistically significant. All statistical analyses in this study were conducted using STATA software version 10.0 (Stata Corp., Texas).

| Literature search
In total, 1744 articles were retrieved from the initial search. Of these, 1687 were excluded after reviewing the titles and abstracts due to duplications or irrelevant topics. Further detailed evaluation was performed for the remaining 57 studies, and finally nine studies that reported seven cohorts were selected for this meta-analysis. 24

| Study characteristics
The general characteristics of included studies and patients were summarized in Table 1 Table 5). Moreover, no significant effect on PPI was observed if the mean age of patients ≥80.0 years, prior PVD ≥10.0%, prior MI ≥10.0%, prior COPD <10.0%, percentage of NYHA III-IV ≥50.0%, and patients who received balloon-expanding TAVR (Supplemental Table 6). No significant publication bias for PPB and PPI was observed (Supplemental Figures 22 and 24).  Tables 7 and 8). Finally, no evidence of publication bias for NOWAF and AKI was observed (Supplemental Figures 26 and 28). reported no prior stroke, studies that did not report prior PVD, studies that did not report prior MI, studies that did not report the percentage of NYHA III-IV, followed up for 30 days, and studies with low quality (Supplemental Table 9). Significant publication bias was inevident (P value for Egger: 0.118; P value for Begg: 0.462; Supplemental Figure 30).  Figure 8).

| Valvular endocarditis
The breakdown regarding the number of cohorts available for valvular endocarditis follow-up at 30 days, 1 year, and 2 years were two, three, and one cohorts, respectively. No significant differences between TAVR and SAVR with regard to the risk of valvular endocar- 1.63-6.24; P = .001; without evidence of heterogeneity) (Supplemental Figure 10).  Figure 11).

| Cardiogenic shock
Data on the effect of TAVR on the risk of cardiogenic shock at 30 days follow-up were available in two studies. The results showed that patients who received TAVR had a reduced risk of cardiogenic shock than those who received SAVR (RR: 0.34; 95%CI: 0.19-0.59; P < .001; without evidence of heterogeneity; Supplemental Figure 12).

| DISCUSSION
The current study included seven RCTs and used meta-analysis to provide solid supporting evidence. The summary results of this study indicated that TAVR demonstrated beneficial effects of PPB, NOWAF, AKI, and cardiogenic shock, whereas TAVR produced excess risk of TIA, PPI, and aortic-valve reintervention when compared with SAVR.
Finally, TAVR and SAVR showed no significant differences on the risk of all-cause mortality, cardiac death, stroke, major vascular complications, MI, valvular endocarditis, and coronary obstruction.
Numerous systematic reviews and meta-analyses were conducted on this topic; however, there are several inherent limitations in these studies [31][32][33][34][35][36][37][38] (Supplemental Table 10). Two studies showed association of TAVR with reduced risk of mortality, 31,32 while the remaining six studies showed no significant difference between TAVR and SAVR on the risk of mortality. [33][34][35][36][37][38] The risk of cardiac death between TAVR and SAVR showed no significant association. 32 TAVR showed association with less complications post-procedurally, which included bleeding, NOWAF, AKI, and cardiogenic shock, and was consistent with previous meta-analyses findings. 34,35,37 Moreover, the risk of PPI and aortic-valve reintervention was shown to be significantly higher in TAVR group.
Patients who received TAVR with high risk of conduction disturbances could explain these increased risk factors. 41 These reduced the risk that could be explained by minimally invasive approach when compared with traditional SAVR. Furthermore, the groups with the risk of valvular endocarditis and coronary obstruction showed no significant differences. However, these results were unstable as these outcomes were reported by smaller number of cohorts, which in turn produce broad confidence intervals, with no statistically significant differences.
However, there are several limitations in this study that should be acknowledged: (1) Some studies might have been missed as they were not included in the searched databases, and this might in turn produce inevitable publication bias; (2) a smaller number of cohorts were included in some subgroups, inducing variable results; (3) there might be bias due to TAVR, which induced substantial heterogeneity and affect treatment effectiveness of TAVR; (4) the causes of aortic valve re-intervention were not available from the included trials; (5) the treatment strategies after TAVR or SAVR were not available across the included studies, which could affect the prognosis of low to intermediate surgical risk AS patients; and (6) this study analysis was based on pooled data, restricting us from conducting a more detailed analysis.