Author + information
- Received September 9, 2012
- Revision received January 14, 2013
- Accepted January 15, 2013
- Published online April 16, 2013.
- Ganesh Athappan, MD⁎,†,
- Eshan Patvardhan, MD⁎,
- E. Murat Tuzcu, MD⁎,
- Lars Georg Svensson, MD, PhD‡,
- Pedro A. Lemos, MD§,
- Chiara Fraccaro, MD, PhD‖,
- Giuseppe Tarantini, MD, PhD‖,
- Jan-Malte Sinning, MD‖,
- Georg Nickenig, MD¶,
- Davide Capodanno, MD, PhD#,
- Corrado Tamburino, MD, PhD#,
- Azeem Latib, MD⁎⁎,
- Antonio Colombo, MD⁎⁎ and
- Samir R. Kapadia, MD⁎,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Samir R. Kapadia, Cardiac Catheterization Laboratory, Cleveland Clinic, 9500 Euclid Avenue, J2-3, Cleveland, Ohio 44195
Objectives This study was designed to establish the incidence, impact, and predictors of post-transcatheter aortic valve replacement (TAVR) aortic regurgitation (AR).
Background AR is an important limitation of TAVR with ill-defined predictors and unclear long-term impact on outcomes.
Methods Studies published between 2002 and 2012 with regard to TAVR were identified using an electronic search and reviewed using the random-effects model of DerSimonian and Laird. From 3,871 initial citations, 45 studies reporting on 12,926 patients (CoreValve [Medtronic CV Luxembourg S.a.r.l., Tolochenaz, Switzerland] n = 5,261 and Edwards valve [Edwards Lifesciences, Santa Ana, California] n = 7,279) were included in the analysis of incidence and outcomes of post-TAVR AR.
Results The pooled estimate for moderate or severe AR post-TAVR was 11.7% (95% confidence interval [CI]: 9.6 to 14.1). Moderate or severe AR was more common with use of the CoreValve (16.0% vs. 9.1%, p = 0.005). The presence of moderate or severe AR post-TAVR increased mortality at 30 days (odds ratio: 2.95; 95% CI: 1.73 to 5.02) and 1 year (hazard ratio: 2.27; 95% CI: −1.84 to 2.81). Mild AR was also associated with an increased hazard ratio for mortality, 1.829 (95% CI: 1.005 to 3.329) that was overturned by sensitivity analysis. Twenty-five studies reported on predictors of post-TAVR AR. Implantation depth, valve undersizing, and Agatston calcium score (r = 0.47, p = 0.001) were identified as important predictors.
Conclusions Moderate or severe aortic regurgitation is common after TAVR and an adverse prognostic indicator of short- and long-term survival. Incidence of moderate or severe AR is higher with use of the CoreValve. Mild AR may be associated with increased long-term mortality. Therefore, every effort should be made to minimize AR by a comprehensive pre-procedural planning and meticulous procedural execution.
Transcatheter aortic valve replacement (TAVR) is a rapidly evolving technology that has been shown to be a durable alternative to surgical aortic valve replacement in patients with severe, symptomatic aortic stenosis considered a high or prohibitive operative risk. In the randomized PARTNER (Placement of AoRTic TraNscathetER Valve Trial) trial (1,2), TAVR exceeded expectations by decreasing mortality and improving quality of life in patients at prohibitive risk of surgical aortic valve replacement. Fueled by these results that had earlier been replicated in large prospective registries (3–5), there has been an exponential increase in TAVR procedures across the globe with speculations of its extension to a low-risk population. However, despite the progress made, there remain several potential TAVR limitations that need to be minimized before implementing this approach in low-risk patients.
Aortic regurgitation (AR) remains a frequent complication of TAVR, with yet unexplained determinants and clinical consequences. Depending on the method of assessment (angiography vs. echocardiography, quantitative vs. semiquantitative), the reported prevalence of AR after TAVR varies from 40% to 67% (6–8) for trivial to mild leaks and from 7% to 20% (3,6–8) for moderate to severe leaks. What constitutes clinically significant valve regurgitation after TAVR is not fully established and is currently a matter of utmost importance. Earlier reports suggested that mild AR is benign and well tolerated (9). However, a recent study (10) revealed that even mild paravalvular leak is an adverse prognostic indicator. The limited yet concerning evidence against AR emphasizes the need for further investigation and to identify variables that can be modified to minimize paravalvular leak.
The objective of our meta-analysis was to identify all currently published literature to establish the incidence of post-procedural AR, its impact on outcomes after TAVR, and its predictors.
We conducted a systematic review of the published literature on post-TAVR AR following the QUOROM (Quality of Reporting of Meta-Analysis) (11) and MOOSE (Meta-Analysis of Observational Studies in Epidemiology) (12) guidelines. We performed a computerized search to identify all relevant studies published from January 1, 2002, to May 5, 2012, in the PubMed database. We chose 2002 because that was the year that Dr. Cribier in Rouen, France, performed the first-in-human TAVR (13). The following search terms were used: TAVI, percutaneous valves, transcutaneous aortic valve, and transcatheter aortic valve. Citations were screened at the title and abstract level and retrieved as a full report if they reported on outcomes after TAVR. The search term regurgitation was then used in each retrieved paper to single out papers reporting data on post-procedural AR. Limiting the search parameters to the English language was applied subsequently. The full texts and bibliography of all potential articles also were reviewed in detail (G.A.) to seek additional relevant studies.
Studies were included if the following criteria applied: 1) reported data on post-TAVR AR severity or predictors of AR or mortality outcomes based on AR severity; 2) reported to have enrolled consecutive patients; 3) performed a minimum of 30 successful TAVR procedures; and 4) enrollment for TAVR was based on existing and accepted guidelines. When 2 similar studies were reported from the same institution or author, the most recent publication or the publication with the most information on post-TAVR AR was included in the analysis.
Studies were excluded if any of the following criteria applied: 1) duplicate publication, overlap of patients, subgroup studies of a main study; 2) lack of data on post-TAVR AR severity; 3) outcome of interests was not clearly reported or was impossible to extract or calculate from the published results; 4) they were studies on valve-in-valve procedure; and 5) a valve other than the CoreValve (Medtronic CV Luxembourg S.a.r.l., Tolochenaz, Switzerland) or Edwards valve (Edwards Lifesciences, Santa Ana, California) was used.
For the purpose of the current analysis, post-procedural AR included AR reported immediately after the procedure, AR at discharge, or AR at 30-day follow-up. AR: none = 0/4, mild = 1/4, moderate = 2/4, severe = 3 to 4/4.
Corevalve or Edwards Valve
A study that performed 80% or more of the TAVR procedures with either valve was subcategorized to the respective group. Studies with less than 80% predominance of either valve were excluded from the analysis of CoreValve versus Edwards valve. Individual data where provided were incorporated in the analysis irrespective of valve predominance.
Relevant information was collected and included but was not limited to first author, year and journal of publication, study design, inclusion exclusion criteria, definition of primary and secondary endpoints, number of subjects included, subjects undergoing successful TAVR, type of device and approach used, study population demographics, echocardiographic parameters post-TAVR, follow-up period, and primary and secondary outcomes.
Data Extraction: Mild AR
To date, only 2 studies (10,14) have reported data on the long-term outcome of mild AR post-TAVR. Therefore, to investigate the outcome of mild AR, we corresponded with trial authors and invited participation in our study. We contacted study authors who reported the outcome of AR post-TAVR. Authors from 4 groups responded. The responding authors analyzed their data and provided us with the hazard ratio (HR) of 1-year mortality for mild AR compared with no AR post-TAVR (4,14–16).
The primary endpoints evaluated were: 1) overall incidences of moderate or severe AR post-TAVR; 2) effect of post-TAVR AR on 30-day and long-term mortality; and 3) predictors of AR. Secondary endpoints of interest were AR incidence stratified by valve used (CoreValve vs. Edwards valve) and by individual grades (none mild, moderate, or severe).
DerSimonian and Laird's random-effects model was used to pool the estimates of post-TAVR AR from individual studies and subgroups. A random-effects model also was used to obtain a single pooled estimate of the HRs, odds ratios (ORs), and correlation from the individual studies. The effect across subgroups was compared using a Q test based on analysis of variance. The odds function, where appropriate, was converted to a correlation. The HRs were estimated from the survival curves when not reported by the method proposed by Parmar et al. (17). Statistical significance was set at p < 0.05 (2-tailed). Heterogeneity, which was anticipated to be significant, was assessed by a Q-statistic and I2 test. Significant heterogeneity was considered present for p values <0.10 or an I2 >50%. Sensitivity analysis was performed by deleting one study at a time and switching from a random-effects to a fixed-effects analysis. Data analysis was performed using Comprehensive Meta Analysis Software Version 2 (18).
Through a search by keywords, 3,871 reports were identified and reviewed at title and abstract level (Fig. 1). Initial evaluation identified 1,180 publications that were further evaluated using the search term regurgitation. This narrowed the selection to 158 potential publications. Manual search of the bibliographies further identified 8 relevant publications. When the inclusion and exclusion criteria were applied, 45 publications remained for assessment (3–5,10,14–16) (Online Refs. 1–38) of incidence and outcomes of post-TAVR AR and 25 publications remained for predictors of AR (15) (Online References 9,12,13,39–59). The 45 included publications had 2 randomized comparisons (10) (Online Ref. 25) and 43 observational studies (3–5,14–16) (Online Refs. 1–4,26–38).
All the studies included in the analysis were published between 2008 and 2012 (Table 1). Analysis was performed on 12,926 patients, the transfemoral/subclavian approach was used in 8,408 patients (65.1%), and the transapical/aortic approach was used in 3,995 patients (30.9%). The approach used was unavailable in 523 patients (4%) (14,16). The self-expanding CoreValve was implanted in 5,261 patients (40.7%), and the balloon-expandable Edwards valve was implanted in 7,279 patients (56.3%). Of the 45 publications, 42 studies (3–5,10,14–16) (Online Refs. 1–32,36–38) provided enough details on the primary endpoint of moderate or severe AR post-TAVR. The remaining 3 studies (Online Refs. 33–35) did not provide details on moderate AR but reported the incidence of severe AR post-TAVR.
The pooled estimate for overall incidence of moderate or severe AR was 11.7% (95% confidence interval [CI]: 9.6 to 14.1, I2 = 91.22, Q = 467.37) (Fig. 2). The Valve Academic Research Consortium (VARC) definitions were used in 12 papers (14,15) (Online Refs. 2,7,15,19,21,22,28,34,35,37). Pooled estimate of moderate or severe AR from papers that reported on the basis of VARC was 13.9% (95% CI: 9.8 to 19.3, I2 = 92.63, Q = 122.09) and without VARC definitions was 10.8% (95% CI: 8.5 to 13.7, I2 = 90.4, Q = 322.71). Papers that did not report on the basis of VARC failed to report the grading method or used the Sellers angiographic criteria or the American Society of Echocardiography/European Society of Endocrinology guidelines for transthoracic echocardiography (TTE). The incidence of AR reported by these reports did not differ significantly from reports using the VARC guidelines (Q = 1.35, p = 0.25).
The incidence of moderate or severe AR after CoreValve implantation was 16.0% (95% CI: 13.4 to 19.0, I2 = 74.81, Q = 59.56). The incidence of moderate or severe AR after Edwards valve implantation was 9.1% (95% CI: 6.2 to 13.1, I2 = 93.63, Q = 313.72). Moderate or severe AR was seen more often with the use of the self-expanding CoreValve (Q = 7.71, p = 0.005) (Fig. 3).
The overall pooled estimate was 1.6% (95% CI: 1.1 to 2.4, I2 = 80.48, Q = 194.63) for severe AR, 10.5% (95% CI: 8.4 to 13.1, I2 = 89.53, Q = 324.84) for moderate AR, 45.9% (95% CI: 40.8 to 51.0, I2 = 95.32, Q = 705.78) for mild/trivial AR, and 35.8% (95% CI: 30.0 to 42.0, I2 = 96.82, Q = 1,037.78) for none.
Our search for predictors of TAVR identified several publications on 3 major culprits (Table 2). The computed tomography-derived mean Agatston calcium score positively correlated with the development of post-TAVR moderate or severe AR with a pooled estimate of correlation being r = 0.47 (95% CI: 0.30 to 0.61, p = 0.001, I2 = 76.24, Q = 12.63). The other 2 predictive factors that were reported could not be pooled because of limited studies and variable reporting.
The overall 1-year mortality was unfavorable in patients with moderate or severe AR with an HR of 2.27 (95% CI: 1.84 to 2.81, p = 0.001, I2 = 26.02, Q = 10.81) (Fig. 4). The OR of 30-day mortality was increased in patients with moderate or severe AR post-TAVR, 2.95 (95% CI: 1.73 to 5.02, p = 0.001, I2 = 0, Q = 2.663). Mild AR post-TAVR was associated with significant mortality, with an HR of 1.829 (95% CI: 1.005 to 3.329, p = 0.048, I2 = 75.28, Q = 16.18) (Fig. 5). The analysis was performed on 1,620 patients across 5 individual studies.
We explored the robustness of our findings by omitting one study at a time or outlier studies and switching our meta-analysis model from a random- to a fixed-effects analysis. There was no change in the summary effects by either analysis other than for mild AR. The pooled HR for mortality with mild AR became insignificant on removal of the studies by Lemos et al. (14), Kodali et al. (10), Sinning et al. (15), and Fraccaro et al. (16) (Fig. 6). Nevertheless, there was a trend toward increased mortality.
AR after TAVR
The impact of mild AR on long-term outcomes has yielded conflicting results. In our pooled analysis, mild AR was associated with a 1-year hazard of 1.940 (95% CI: 1.090 to 3.452). However, the conclusions changed when studies were removed from the analysis set one at a time. This discrepancy in outcomes may be related in part to the challenges in identification and quantification of post-TAVR AR. Post-TAVR AR is frequently paravalvular, created by multiple eccentric jets that are nonparallel and irregular in shape (9,19). The eccentric jets in turn are frequently entrained along the LV wall with fanning of jets as they regurgitate. This makes the assessment of AR severity difficult and more subjective. Acoustic shadowing from the calcifications, reverberations, and Doppler attenuation from the prosthesis can further obscure significant regurgitant jets and result in underestimation of its severity (9). Various echocardiographic techniques (19,20) and grading schemes (grades 1 to 4, none to severe) have been proposed to measure the severity of post-TAVR AR; however, none of these techniques have been validated and the grading system has not been clearly defined (difference between mild, trivial, or trace, mild vs. moderate, moderate vs. severe). Therefore, the assessment of post-TAVR AR remains controversial and imprecise (21).
Sherif et al. (22) demonstrated an underestimation of the AR severity after implantation of the Core Valve by current echocardiography criteria compared with cardiac magnetic resonance. The VARC (23), which was developed to propose standardized consensus definitions to report post-TAVR complications, failed to adequately address paravalvular leak in its current report. The VARC did not propose standard terminologies or new diagnostic criteria for assessment of AR, but merely elaborated on the work done by others and summarized by Zoghbi et al. (20). The lack of core laboratory assessment of AR severity in the included studies and the absence of published standards for post-TAVR AR may have affected the quality of the reported data. This issue of imprecision is a major limitation in comparing echocardiographic studies performed in different laboratories and likely contributed to the discordant results on the outcome of mild AR. To develop standard criteria to achieve uniformity and accurate grading of paravalvular AR post-TAVR is therefore a pressing necessity.
To complicate the matter further, it is possible that AR is a surrogate for an underlying cause, such as severe valve calcification or sicker patients. The role of other confounding variables, such as pre-procedural AR, left ventricle function, and mitral regurgitation, remains unclear at present. However, the impact of moderate to severe AR is more apparent with a pooled 30-day OR for mortality of 2.95 (95% CI: 1.73 to 5.02) and a long-term HR of 2.27 (95% CI: 1.84 to 2.89).
CoreValve versus Edwards valve
Regardless of the valve type, post-TAVR AR was a frequent complication in reported studies. Nevertheless, implantation of the CoreValve carried a higher risk of post-TAVR AR (16.0% vs. 9.1%, p = 0.005). Concerns have been raised over the radial strength of the nitinol frame of the self-expanding CoreValve in highly calcific lesions (Online Refs. 13,26,51). Incomplete device expansion and resultant impaired apposition of the CoreValve to the native annulus and the left ventricular outflow tract have been implicated in these cases. Another possible cause for AR after core valve implantation is the extreme angulation between the left ventricular outflow tract and the ascending aorta, which reduces the ability of the self-expanding prosthesis to form a tight seal to close the paravalvular space. Balloon post-dilation and a greater oversizing as opposed to the Edwards valve may overcome the underexpansion of the CoreValve in these situations. However, balloon post-dilation has been associated with increased stroke risk (24). Another factor that is important for reducing AR when the CoreValve is used is the height of implantation. Because of the noncylindrical shape of the valve, the depth of implantation determines the effective diameter of the valve in the annulus. Particularly in larger annuli, the sealing of the CoreValve at the level of the virtual ring is dependent on a high implantation to take advantage of the diameter of the lower part of the valve.
Predictive factors and potential management strategies
Mismatch of the valve annulus and prosthesis diameter sizes, aortic root calcification, and suboptimal device implantation were identified as the major causes for post-TAVR AR in our search.
Undersizing of the prosthesis relative to the annulus size is the central cause for most paravalvular leaks after TAVR. Detaint et al. (Online Ref. 39) studied the effect of undersizing using the cover index [100 · (prosthesis diameter − transesophageal echocardiography [TEE] annulus diameter)/prosthesis diameter]. A low cover index was found to be an independent predictor of moderate or severe AR post-TAVR. This has been replicated in 2 other studies (15) (Online Ref. 40). Appropriate cover index for each valve may be different and requires more studies to clearly define this.
Therefore, precise annulus sizing by appropriate aortic imaging pre-TAVR is fundamental to prevent AR (25). The aortic annulus was initially sized by 2-dimensional measurements obtained from TTE, but there are several limitations to annulus sizing using TTE. TTE has been shown to underestimate the annulus size from anywhere between 1.4 mm and 1.7 mm (26) (Online Ref. 42). Similar underestimation may occur with TEE (1.2 mm) (26). Jilaihawi et al. (Online Ref. 47) showed that computed tomography–guided annular sizing reduced paravalvular AR compared with a TEE-guided approach in patients receiving an Edwards SAPIEN valve (7.5% vs. 21.9%). Willson et al. (Online Ref. 42) and Schultz et al. (Online Ref. 44) also showed the superiority of multidetector computed tomography annular measurements. A high correlation between multidetector computed tomography and 3-dimensional (3D) TEE measurements of annulus size have been demonstrated in experienced hands. Therefore, it is fair to conclude that 3D imaging for annulus sizing holds the key to better sizing the valves and decreasing paravalvular leaks.
Aortic valve calcium score
Aortic root calcium is thought to hinder uniform valve expansion and tight sealing. However, existing data on this topic are conflicting. In the analysis of the German TAVI registry on 1,365 patients, Staubach et al. (Online Ref. 57) found that the extent of aortic valve calcification did not influence the severity of post-TAVR AR, which is similar to that reported by Wood et al. (Online Ref. 56). Other investigators found that an Agatston score >3,000 predicted moderate or severe AR after initial release of the CoreValve (Online Refs. 41,50,52) and the need for post-dilation. Likewise, other quantitative calcium scores also have been shown to correlate positively with the occurrence of post-TAVR AR (Online Refs. 54,55). Despite the conflicting evidence, precise quantitation of the extent and location of calcification in the aortic root may allow the identification of patients with asymmetric heavy calcification that may increase the risk of AR.
Post-procedural AR is influenced significantly by the implantation depth. Valve positioning currently is based mainly on fluoroscopy and angiography with or without echocardiographic guidance. Choosing the correct fluoroscopic plane is critical. When misplaced high or low, the skirt of the prosthetic valve does not provide an adequate seal around the annulus, resulting in AR. The unequal geometry of the CoreValve with a narrow and tapered midsection further contributes to paravalvular leak from an inadequate seal when misplaced. Takagi et al. (Online Ref. 45) showed that low CoreValve implantation increased the OR of moderate or severe AR, 3.67 (95% CI: 1.01 to 13.35). Sherif et al. (Online Ref. 58) and Jilaihawi et al. (Online Ref. 59) showed that a 9.5-mm and 5- to 10-mm device depth, respectively, from the noncoronary cusp minimized the risk of moderate or severe AR for the CoreValve. Improvements in imaging technology to provide real-time 3D imaging, increasing experience and modifications in delivery systems, will likely improve the precision of valve deployment.
Clinical implications and future perspective
Moderate or severe AR post-TAVR is common but can be prevented by accurate annulus measurements with 3D techniques and adequate valve sizing. Accurate positioning of the valves may reduce the risk of paravalvular AR. Post-TAVR AR is difficult to quantitate with any single currently available imaging technique and should be accurately quantified using multimodality imaging with hemodynamic data. The next generation of transcatheter aortic valves designed to minimize paravalvular leak will have a major role to play in the future of TAVR.
The studies pooled in the analysis were observational studies with post-TAVR AR not being a primary outcome of interest. The included studies used different methods and grading schemes for assessment of AR severity, thereby introducing limitations in quality and completeness of data. There was significant heterogeneity across studies for all outcomes analyzed. Because of incomplete/unequal reporting of data, not all studies were pooled for all outcomes, which could lead to publication bias. Selection bias was introduced in the evaluation of the prognostic value of mild AR by our method of contacting authors. Precise distinction between paravalvular and valvular AR was not made. AR varies with time; therefore, our definition of post-procedural AR may not be precise. Expansion of the CoreValve over time was not taken into account, a fact that may have biased the results of post-procedural AR in favor of the Edwards valve (however, at present there are no data to indicate that expansion of the CoreValve caused by radial forces in the inflow portion reduces AR over time). Despite these limitations, the large sample size and robustness of our findings clearly demonstrate the need for ongoing critical evaluation of this problem.
Moderate or severe AR is common after TAVR and an adverse prognostic indicator of short- and long-term survival. Every effort should be made to predict and minimize post-procedural AR. Underestimation of paravalvular AR with currently used imaging modalities may be significant, and some patients with reported mild AR post-TAVR may have moderate or even severe AR. Sizing of the annulus is a key step to prevent post-procedural AR where sizing with 3D imaging is superior to 2-dimensional imaging techniques. Innovations designed to improve sealing, improvement in the range of available device sizes, accurate annular sizing, and precise positioning will help minimize AR after TAVR.
For a list of the additional references, please see the online version of this article.
Dr. Latib is a member of the Medtronic advisory board. All other authors have reported that they have no relationship relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- aortic regurgitation
- confidence interval
- hazard ratio
- odds ratio
- transcatheter aortic valve replacement
- transesophageal echocardiography
- transthoracic echocardiography
- Valve Academic Research Consortium
- Received September 9, 2012.
- Revision received January 14, 2013.
- Accepted January 15, 2013.
- American College of Cardiology Foundation
- Rodés-Cabau J.,
- Webb J.G.,
- Cheung A.,
- et al.
- Tamburino C.,
- Capodanno D.,
- Ramondo A.,
- et al.
- Moat N.E.,
- Ludman P.,
- de Belder M.A.,
- et al.
- Webb J.G.,
- Pasupati S.,
- Humphries K.,
- et al.
- Cribier A.,
- Eltchaninoff H.,
- Bash A.,
- et al.
- Sinning J.M.,
- Hammerstingl C.,
- Vasa-Nicotera M.,
- et al.
- Fraccaro C.,
- Al-Lamee R.,
- Tarantini G.,
- et al.
- Borenstein M.,
- Hedges L.,
- Higgins J.,
- Rothstein H.
- Zoghbi W.A.,
- Chambers J.B.,
- Dumesnil J.G.,
- et al.
- Leon M.B.,
- Piazza N.,
- Nikolsky E.,
- et al.
- Nombela-Franco L.,
- Rodés-Cabau J.,
- Delarochelliere R.,
- et al.
- Babaliaros V.C.,
- Liff D.,
- Chen E.P.,
- et al.