Impact of Renal Dysfunction on Mid-Term Outcome after Transcatheter Aortic Valve Implantation: A Systematic Review and Meta-Analysis

Background There is conflicting evidence regarding the impact of preexisting renal dysfunction (RD) on mid-term outcomes after transcatheter aortic valve implantation (TAVI) in patients with symptomatic aortic stenosis (AS). Methods and results Forty-seven articles representing 32,131 patients with AS undergoing a TAVI procedure were included in this systematic review and meta-analysis. Pooled analyses were performed with both univariate and multivariate models, using a fixed or random effects method when appropriate. Compared with patients with normal renal function, mid-term mortality was significantly higher in patients with preexisting RD, as defined by the author (univariate hazard ratio [HR]: 1.69; 95% confidence interval [CI]: 1.50–1.90; multivariate HR: 1.47; 95% CI: 1.17–1.84), baseline estimated glomerular filtration rate (eGFR) (univariate HR: 1.65; 95% CI: 1.47–1.86; multivariate HR: 1.46; 95% CI: 1.24–1.71), and serum creatinine (univariate HR: 1.69; 95% CI: 1.48–1.92; multivariate HR: 1.65; 95% CI: 1.36–1.99). Advanced stage of chronic kidney disease (CKD stage 3–5) was strongly related to bleeding (univariate HR in CKD stage 3: 1.30, 95% CI: 1.13–1.49; in CKD stage 4: 1.30, 95% CI: 1.04–1.62), acute kidney injure (AKI) (univariate HR in CKD stage 3: 1.28, 95% CI: 1.03–1.59; in CKD stage 4: 2.27, 95% CI: 1.74–2.96), stroke (univariate HR in CKD stage 4: 3.37, 95% CI: 1.52–7.46), and mid-term mortality (univariate HR in CKD stage 3: 1.57, 95% CI: 1.26–1.95; in CKD stage 4: 2.77, 95% CI: 2.06–3.72; in CKD stage 5: 2.64, 95% CI: 1.91–3.65) compared with CKD stage 1+2. Patients with CKD stage 4 had a higher incidence of AKI (univariate HR: 1.70, 95% CI: 1.34–2.16) and all-cause death (univariate HR: 1.60, 95% CI: 1.28–1.99) compared with those with CKD stage 3. A per unit decrease in serum creatinine was also associated with a higher mortality at mid-term follow-up (univariate HR: 1.24, 95% CI: 1.18–1.30; multivariate HR: 1.19, 95% CI: 1.08–1.30). Conclusions Preexisting RD was associated with increased mid-term mortality after TAVI. Patients with CKD stage 4 had significantly higher incidences of peri-procedural complications and a poorer prognosis, a finding that should be factored into the clinical decision-making process regarding these patients.


Introduction
As a rapidly evolving procedure, transcatheter aortic valve implantation (TAVI) has been shown to be a safe and effective alternative to surgical aortic valve replacement (SAVR) in high-risk or inoperable patients with symptomatic aortic stenosis (AS) [1][2][3]. These aging patients frequently have a high prevalence of chronic renal dysfunction (RD), which portends a poor prognosis in those who undergo SAVR [2][3][4]. However, the results from studies evaluating the impact of baseline renal function on outcomes after TAVI are conflicting [5][6][7]. In many TAVI studies, although higher mid-term mortality were observed in patients with RD, these differences were not found to be significant by multivariate analyses [6,[8][9][10]. In addition, the relationship between varying degrees of RD and mid-term prognosis has also not been elucidated.
Therefore, we conducted a meta-analysis of published studies to clarify the mid-term prognostic value of preexisting RD in patients undergoing TAVI.

Search Strategy
The PubMed online database and the Cochrane library were searched for articles published from January 2002 to April 2014. The following search strategy was used: (transcatheter OR percutaneous OR transfemoral OR transapical OR transsubclavian OR transaortic OR transaxillary) AND (aortic valve) AND (implantation OR replacement) AND (risk factor OR risk assessment OR predictor OR kidney disease OR renal insufficiency OR nephropathy OR creatinine OR estimated glomerular filtration rate OR dialysis OR hemodialysis OR hemodialysis). Reference lists of comparable articles were also retrieved to seek potentially relevant citations.

Study Selection
Two reviewers conducted the initial screening of titles and abstracts; full-length reports of identified studies were retrieved; and decisions were then made regarding eligibility according to pre-specified inclusion and exclusion criteria. Studies were included if they (1) reported the predictive value of the pre-procedure renal function or mortality outcomes in patients with RD compared with normal controls; (2) performed follow-ups for at least 6 months; and (3) were human studies and published in English. Studies were excluded if they were (1) abstracts, letters, editorials, or reviews and (2) duplicate publications. Studies with overlapping populations were handled by selecting the study that reported on the largest sample of patients undergoing TAVI, unless they used different definitions of RD or reported results in different analysis models.

Data Extraction
Data were extracted from relevant studies using a pre-specified data collection form that included the first author, journal, year of publication, baseline characteristics, definition of RD, valve type, follow-up duration, and number of complications and deaths. Complications were defined according to the Valve Academic Research Consortium criteria [11], including acute kidney injure (AKI), stroke, all-cause bleeding, and major vascular complications. The outcomes from 6 months to 3 years were defined as mid-term outcomes. The incidence of allcause mid-term mortality was the primary end point. TAVI-related complications were also the end points of interest.

Statistical Analysis
The hazard ratio (HR) of preexisting RD with regard to mid-term mortality after TAVI was extracted or calculated. The Generic Invers Variance method in the RevMan software, version 5.20 (The Nordic Cochrane Centre, Copenhagen, Denmark) was used for synthesis of the effect estimates. Heterogeneity was assessed by the Q-statistic and I 2 test. The fixed effects model was selected for the analysis without significant heterogeneity (I 2 <50% and a corresponding P>0.1); otherwise, the random effects model was used to obtain the combined effect estimates. Statistical significance was set at P<0.05 (two-tailed). Sensitivity analyses were performed using the STATA software version 12.1 (StataCorp, College Station, TX) to test the robustness of the results and the influence of potential effect modifiers. Publication bias was assessed by graphical inspection of funnel plots, Begg's tests, and Egger's tests. The "Trim and Fill method" was applied if there was any evidence of publication bias [13].
The present meta-analysis was conducted and reported according to the recommendation of the MOOSE group [14].

Study Selection
We identified 1096 citations in the initial screening (Fig. 1). After removing duplicates and screening at the abstract level, we retrieved 286 articles for a more detailed evaluation. While 239 studies were subsequently excluded, a total of 47 full-text articles were eligible for this meta-analysis, enrolling a total of 32,131 AS patients with renal function-specific data. No significant limitations were identified for the 47 trials, 3 of which were randomized comparisons [15][16][17], while others were observational cohort studies [5,6,. Although a few studies had overlapping patient populations, they provided different outcomes according to different definitions of RD, as defined by the author [16,18,20,23,40,42,43], baseline eGFR [10,27,49], or serum creatinine [6,15,17,34,35,38,47,56]. We thus assigned these studies to different groups for either univariate or multivariate analysis.

Study and Patient Characteristics
The baseline features of the included patients are presented in Table 1. Most studies were conducted in the general population, while a few studies performed TAVI in unique patient groups, such as octogenarians and nonagenarians [43], patients at very high risk (with a Euro-SCORE of more than 40%) [42], and patients with chronic lung disease (CLD) [15]. Four studies elucidated the impact of detailed CKD classification on outcomes after TAVI [10,29,48,49]. Patients with end-stage renal disease (ESRD) were included in 14 studies [5, 18, 22-24, 29, 38-41, 43, 47, 53, 55, 56] and were excluded in 7 studies [6, 10, 15-17, 44, 49]. In the remaining  studies, the number of patients with ESRD was unclear. Procedural characteristics and the main outcomes after TAVI are summarized in Table 2.

Association of Mid-term Outcomes with Varying Degrees of RD
Four studies included a detailed classification of CKD according to the baseline eGFR [10,29,48,49], and an additional 4 studies reported the mid-term mortality of patients on chronic dialysis [5,22,24,38].
A total of 9 studies that included 5,266 patients were eligible for the pooled analysis of baseline serum creatinine (for each increase of 1 mg/dl) with respect to mid-term outcomes [18, 37, [57] reported a similar finding with regard to baseline eGFR; specifically, each 10 ml/min decrease was found to be associated with a significantly higher risk of all-cause mortality after TAVI (univariate HR: 1.14, 95% CI: 1.07-1.22; multivariate HR: 1.14, 95% CI: 1.07-1.22).

Publication Bias
Although a subtle publication bias was observed in the funnel plot inspection comparing patients with RD (defined as decreased eGFR) with patients with normal renal function in the univariate model, the pooled estimates remained significant after implementing the "Trim and    Fill" method. In the rest of the analyses, funnel plots, Begg's test and Egger's test did not provide clear evidence for publication bias (S1-S5 Figs).

Discussion
The present study is the first to conduct pooled analyses (using both univariate and multivariate models) of the mid-term prognostic value of RD after TAVI. Preexisting RD, despite different definitions, was found to be associated with significantly increased mid-term mortality.
Although it has been clearly demonstrated that aging patients with symptomatic AS have a high prevalence of RD, only a few TAVI studies have treated RD as a component of the primary study question, and the results have been conflicting [6,7,10,29,44,48,49,58]. By conducting this meta-analysis, we have clearly shown a correlation between mid-term outcome and baseline renal function, as reflected by either baseline eGFR or serum creatinine.
Varying degrees of RD, as classified by advanced stages of CKD, were associated with significantly higher incidences of bleeding, AKI, and mid-term mortality after TAVI. Post-procedural stroke occurred more frequently in patients with CKD stage 4 compared with CKD 1+2. These findings were in line with previous TAVI studies focusing on the peri-procedural complications [8,44,59]. However, differences about the incidence of MVC were not significant in our study. At mid-term follow-up, patients with CKD stage 4 were noted to have a poorer prognosis compared with patients with CKD stage 3. This graded association was further confirmed when considering baseline serum creatinine (for each 1 mg/mL decrease), which was also strongly related to increased mid-term mortality. In previous studies, advanced stages of CKD have been shown to be independent risk factors in patients undergoing TAVI [10,29,49]. However, no stepwise increased adverse events was observed in patients with more severe CKD [10,29]. By pooling estimate effects from these individual studies, we found that patients with CKD stage 4 had significantly higher incidence of AKI and mortality rates compared with those with CKD stage 3. Because patients with ESRD have been excluded from many TAVI studies, only sparse data exist on the prognostic value of CKD stage 5 [29,60]. In the present study, pre-procedural chronic dialysis was also shown to be a strong risk factor for mid-term mortality after TAVI.
The presence of RD is an important factor contributing to poorer outcomes in patients undergoing TAVI. This phenomenon can be explained by several aspects. (1) Patients with RD were older, were frailer, and presented with a significantly higher Logistic Euroscore in previous studies [10,29,49]. In view of these data, RD may serve as a marker of unbalanced baseline risk profiles. Patients with RD frequently have a higher burden of severe morbidities, which may adversely affect their survival after TAVI. (2) RD modifies the natural history of AS, presumably through a pathophysiological mechanism that promotes calcium deposition on aortic leaflets, thereby worsening aortic stenosis and reducing cardiac output [61]. Severe AS with decreased flow to important organs is responsible for the onset of severe complications, which subsequently increase the mortality after TAVI [8,48,62]. RD was also found to be associated with disorders of primary hemostasis, in particular platelet malfunctions [63], which played an important role in the occurrence of peri-procedural bleeding and subsequently increased mortality [59].
(3) It is well known that one of the advantages of TAVI is the avoidance of cardiopulmonary bypass, which is one of the most important risk factors for post-procedural AKI [64]. However, the incidence of contrast-induced nephropathy (CIN) could conceivably increase as a result of the extensive use of contrast medium and multiple injections [65,66]. Although few studies have identified a significant association between contrast agents and AKI in the general population [67,68], when focusing on patients with RD, the incidence of CIN was found to be significantly enhanced. Among patients with CKD, the occurrence of CIN was strongly associated with a higher 60-day mortality [69], indicating that the nephrotoxic mechanisms of CIN were one of the major issues contributing to the mid-term mortality in such patients.
In patients with more severe kidney failure, a higher Logistic Euroscore and lower ejection fraction were more frequent [10,29]. Moreover, the incidence of post-procedural renal impairment was also significantly higher in patients with more severe RD, despite using a smaller dose of contrast medium [29]. These results could explain the graded association between the severity of RD and the stepwise increase in mortality after TAVI.
In view of these data, RD appears to be not only a marker of illness severity, but it also represents a risk factor for mid-term prognosis. Therefore, rigorous risk assessment, preventive therapies for bleeding and stroke, and timely detection of AKI would be crucial interventions that would improve the mid-term mortality after TAVI. Our study revealed higher incidence of peri-procedural complications and poorer outcomes in patients with CKD stage 4. However, this result also raises questions regarding whether these high-risk patients actually benefit from a TAVI procedure and which patients are at the highest risk of mid-term mortality.

Limitations
Several limitations exist in our study. (1) Because the present meta-analysis was based only on published studies, the possibility of potential publication bias cannot be completely ruled out.
(2) Although careful screening was conducted, the possibility of overlapping study populations could result in similar estimates. (3) Our meta-analysis was not conducted at the patient level, and only 5 studies treated RD as the primary study question. Even though the renal functionspecific baseline characteristics were not available, the effects of comorbidities could not be assessed. (4) The adjusted prognostic value of different degrees of RD on the mid-term mortality after TAVI was not assessed due to the scarcity of study data. (5) Most included studies calculated eGFR using the MDRD equation, which is affected by the considerable decline in muscle mass with age, severe cardiovascular disease, drugs, and diet, making it difficult to reflect the actual renal clearance in the cohort of elderly patients.

Conclusions
Preexisting RD, despite different definitions, was associated with significantly increased midterm mortality after TAVI. Varying degrees of RD were strongly associated with a stepwise increase in mid-term mortality rates. Given that patients with CKD stage 4 had a higher incidence of peri-procedural complications and a poorer prognosis, TAVI in such patients may present a significant challenge.