Meta-Analysis Comparing Renal Outcomes after Transcatheter versus Surgical Aortic Valve Replacement

Background Acute kidney injury (AKI) is a common complication of aortic valve replacement. However, comparative on the incidence of (AKI) following transcatheter (TAVR) versus surgical valve replacement (SAVR) is sparse. Methods We performed a meta-analysis of the randomized controlled trials (RCT) and propensity-matched observational studies comparing (A) incidence of AKI and (B) incidence of dialysis-requiring AKI at 30 days after TAVR and SAVR. Results Twenty-six studies (20 propensity-matched studies; 6 RCTs) including 19,954 patients were analyzed. The incidence of AKI was lower after TAVR than after SAVR (7.1% vs. 12.1%, OR 0.52; 95%CI, 0.39-0.68; p<0.001, I2=57%), but the incidence of dialysis-requiring AKI was similar (2.8% vs. 4.1%, OR 0.78; 95%CI, 0.49-1.25; p=0.31, I2=70%). Similar results were observed in a sensitivity analysis including RCTs only for both AKI ([5 RCTs; 5,418 patients], 2.0% vs. 5.0%, OR 0.39; 95%CI, 0.28-0.53; p<0.001, I2=0%), and dialysis-requiring AKI ([2 RCTs; 769 patients]; 2.9% vs. 2.6%, OR 1.1; 95%CI, 0.47-2.58; p=0.83, I2=0%). However, in studies including low-intermediate risk patients only, TAVR was associated with lower incidence of AKI ([10 studies; 6,510 patients], 7.6% vs. 12.4%, OR 0.55, 95%CI 0.39-0.77, p<0.001, I2=57%), and dialysis-requiring AKI, ([10 studies; 12,034 patients], 2.0% vs. 3.6%, OR 0.57, 95%CI 0.38-0.85, p=0.005, I2=23%). Conclusions TAVR is associated with better renal outcomes at 30 days in comparison with SAVR, especially in patients at low-intermediate surgical risk. Further studies are needed to assess the impact of AKI on long-term outcomes of patients undergoing TAVR and SAVR.


Introduction
The introduction of transcatheter aortic valve replacement (TAVR) and the continuous improvement in the outcomes of surgical aortic valve replacement (SAVR) have revolutionized the treatment of patients with severe aortic stenosis in the last decade [1,2]. However, acute kidney injury (AKI) remains a common complication of both treatment modalities. Nonetheless, data on the incidence of AKI and dialysis-requiring AKI after TAVR vs. SAVR remain limited [3]. We performed a meta-analysis of the randomized clinical trials (RCTs) and propensity-matched (PSM) observational studies to compare renal outcomes following TAVR vs. SAVR (a) overall and (b) in subgroups of high-risk and lowintermediate risk patients.

Methods
Our review protocol was conducted in accordance with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) reporting guidelines (Supplementary Protocol). [4] The literature search was conducted in PUBMED, MEDLINE, EMBASE, EBSCO, and Cochrane (March 2, 2018) in order to identify eligible studies using the Medical Subject Headings search terms and text word search. We also did a manual search of the reference lists of relevant studies for additional publications and when multiple publications from the same study population were found, data from the most inclusive report was used. The data was reviewed independently from full-text articles by 2 of the authors (T.B. and K.S.). Disagreements were

Data Synthesis and Analysis
The data supporting this meta-analysis are from previously reported studies and datasets, which have been cited. The processed data are reported in the article and in the supplementary files. We performed our meta-analyses using Comprehensive Meta-Analysis version 3.0 (Biostat, https://www.meta-analysis.com). We used the random effects model with the Mantel-Haenszel (MH) method for each clinical endpoint and pooled estimates of odds ratio (OR) with 95% confidence interval (CI) were calculated. We used I 2 index, tau squared, and the Q-test P value to examine heterogeneity among individual study effect sizes. To reduce the risk of bias, we undertook independent pooling of data from RCTs and PSM observational studies. In order to formally assess publication bias we prepared funnel plots and Egger's linear regression test of funnel plot asymmetry (eFigures 1 and 2). All pooled estimates are displayed with a 95% confidence interval (CI). P values were considered statistically significant at less than 0.05. We also performed sensitivity analysis to investigate potential sources of inconsistency, including removal of nonrandomized studies. Forest plots were generated to show the relative effect size of TAVR and SAVR for each clinical outcome. Potential sources of heterogeneity were investigated using meta-regression techniques; factors analyzed in the metaregression included age, sex, diabetes, prior stroke, chronic renal insufficiency, vascular disease, and atrial fibrillation (eFigure 3). We followed standard protocol for performing meta-analysis as in our previous publication. The endpoint of interest in this study is renal outcomes which were not described in our previous papers making this study unique. [5,6]

Results
A total of 5,067 potentially relevant citations were identified and screened ( Figure 1). After removal of duplicated studies, we retrieved 76 full-text articles for evaluation, of which 26 satisfied the selection criteria. A total of 20 PSM observational studies and 6 RCTs were included in the meta-analysis ( Figure 1). All eligible studies were in the English Language. The baseline characteristics of the patients in the included studies are summarized in  (Figures 4(a) and 4(b)). No heterogeneity among these trials was observed (I 2 =0%).

Meta-Analysis Stratified by Surgical Risk.
A secondary analysis was performed to compare the pooled incidence of AKI and dialysis-requiring AKI among patients who are at high surgical risk and those at low-intermediate surgical risk.

Discussion
The major findings of the current investigation are as follows.
(1) TAVR is associated with lower rates of AKI compared with  SAVR, and this was consistent in the overall analysis, in a sensitivity analysis including RCTs only, and in subanalyses of high-risk and low-intermediate risk patients. (2) The risk of dialysis-requiring AKI appears to be comparable after TAVR vs. SAVR. However, the pooled incidence of dialysisrequiring AKI was significantly lower after TAVR than after SAVR in a subgroup of low-intermediate risk patients.
Patients with severe aortic stenosis are characteristically older and have many comorbidities including a high prevalence of chronic renal insufficiency. Cardiac surgery operations including SAVR are associated with significant risk of AKI and AKI requiring dialysis [33,34]. Transcatheter aortic valve replacement was introduced as an effective alternative to surgery in high-prohibitive risk patients but later expanded into young and lower risk patient cohorts. Nonetheless, both the preoperative work-up and the TAVR procedure itself carry a significant risk of AKI due to contrast medium usage, and the high prevalence of atherosclerotic risk factors among patients submitted for TAVR. Whether TAVR is associated with lower risk of AKI and AKI requiring dialysis than SAVR has not been well studied. In the pivotal PARTNER-1 trial, no difference in the rate of AKI was observed between TAVR and SAVR. [7] Subsequent RCTs showed lower rates of AKI after TAVR compared with SAVR. We hence performed a systematic review and a meta-analysis to synthesize the best available evidence on renal outcomes following TAVR and SAVR.
Our meta-analysis showed that TAVR is associated with about 50% reduction in the incidence of AKI compared with SAVR, but a similar rate of dialysis-requiring AKI between the two modalities overall. These findings have important prognostic implications and deserve more scrutiny for several reasons. (1) There is ample evidence that even AKI not requiring dialysis is associated with substantial negative impact on long-term outcomes [35][36][37][38][39][40][41]. (2) The risk of AKI and dialysis-requiring AKI may be more modifiable in patients undergoing TAVR. The advances in 3D echocardiography and the refinements in TAVR techniques have allowed the introduction of the 'Reno-protective TAVR' concept [42][43][44][45][46]. This concept along with the wide adoption of moderate sedation in TAVR procedures has the potential to further reduce postprocedural renal insufficiency although this has not yet been studied in a prospective fashion [47]. In contrast, the risk of AKI after SAVR may be more related to the patient risk profile than to modifiable procedural factors as surgical techniques in SAVR have not differed significantly in the last decade. (3) In our subanalysis of patients who are at low-intermediate surgical risk, TAVR was associated with a significant reduction not only in AKI but also in AKI requiring dialysis. In light of the continuous expansion of TAVR to lower risk populations, the impact of the TAVR on improving renal outcomes in these patients warrants more investigation [48].

Limitations
Our study has several limitations: (1) there are only few RCTs comparing TAVR with SAVR. Hence, we included observational studies in our meta-analysis. However, we limited our inclusion of observational studies to those with propensity score matched comparisons. While this can introduce heterogeneity into our analysis, our sensitivity analysis including RCTs only yielded similar results to the overall meta-analysis. (2) The definition of AKI varies among the studies, but those definitions were maintained the same in the same study for comparison between TAVR and SAVR, and hence the results are comparable.

Conclusions
TAVR procedure has significantly lower rates of AKI compared to SAVR but similar rates of AKI requiring renal replacement therapy. AKI has short and long-term effects on Journal of Interventional Cardiology outcomes and survival; hence every effort should be made to reduce the incidence of AKI.

Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.

Disclosure
All authors listed meet the authorship criteria according to the latest guidelines of the International Committee of Medical Journal Editors, and all authors are in agreement with the manuscript. The article is processing charges to be covered in the institutional membership at West Virginia University.

Conflicts of Interest
The authors declare that they have no conflicts of interest.

Authors' Contributions
Dr. Shah and Dr. Chaker contributed equally to this manuscript.