- Email: [email protected]
Impact of paravalvular aortic regurgitation after transcatheter aortic valve implantation on survival
International Journal of Cardiology 221 (2016) 46–51
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Impact of paravalvular aortic regurgitation after transcatheter aortic valve implantation on survival☆ Hisato Takagi ⁎, Takuya Umemoto, for the ALICE (All-Literature Investigation of Cardiovascular Evidence) Group Department of Cardiovascular Surgery, Shizuoka Medical Center, Shizuoka, Japan
a r t i c l e
i n f o
Article history: Received 21 April 2016 Received in revised form 1 July 2016 Accepted 2 July 2016 Available online 04 July 2016 Keywords: Meta-analysis Paravalvular aortic regurgitation Survival Transcatheter aortic valve implantation
a b s t r a c t Objectives: To determine whether ≥moderate paravalvular aortic regurgitation (PAR) after transcatheter aortic valve implantation (TAVI) independently impairs overall survival and how much the impact on survival is, we performed an updated meta-analysis pooling not unadjusted but adjusted hazard ratios (HRs). Methods: Databases including MEDLINE and EMBASE were searched through January 2016 using PubMed and OVID. Search terms included paravalvular or perivalvular; regurgitation, leak, or leakage; percutaneous, transcatheter, transluminal, transarterial, transapical, transaortic, transcarotid, transaxillary, transsubclavian, transiliac, transfemoral, or transiliofemoral; and aortic valve. Studies considered for inclusion met the following criteria: the design was an observational comparative study; the study population was patients undergoing TAVI; patients were divided into ≥moderate and ≤mild post-TAVI PAR; outcomes included ≥1-year all-cause mortality; and the adjustment method was a multivariate Cox proportional hazards analysis. An adjusted HR with its 95% confidence interval (CI) for ≥moderate post-TAVI PAR was abstracted from each individual study. Results: Our search identified 17 eligible studies including a total of 15,131 patients. A pooled analysis of all the 17 studies demonstrated a statistically significant 2.12-fold increase in mortality with ≥moderate PAR (HR, 2.12; 95% CI, 1.79 to 2.51; p b 0.00001). Exclusion of any single study from the meta-analysis did not substantively alter the overall result disfavoring ≥ moderate PAR. Although the statistical tests suggested funnel plot asymmetry, the corrected effect estimate from the trim-and-fill method demonstrated still a statistically significant 1.83-fold risk of mortality with ≥moderate PAR. Conclusions: ≥ Moderate post-TAVI PAR is associated with a 2.12-fold increase in overall (≥ 1-year) all-cause mortality. © 2016 Elsevier Ireland Ltd. All rights reserved.
1. Introduction The incidence of moderate or severe (≥moderate) paravalvular aortic regurgitation (paravalvular AR, PAR) after transcatheter aortic valve implantation (TAVI) for severe aortic stenosis (AS) has been reported to be 12–21% [1], which is approximately 6-fold higher than that after surgical aortic valve replacement (SAVR) [2]. A number of studies suggest that ≥moderate PAR is an independent risk factor of overall mortality. However, some investigators [3,4] demonstrated a significant N4-fold risk,
Abbreviations: AR, aortic regurgitation; AS, aortic stenosis; CI, confidence interval; HR, hazard ratio; LV, left ventricle; MR, mitral regurgitation; OR, odds ratio; PAR, paravalvular AR; SAVR, surgical aortic valve replacement; TAVI, transcatheter aortic valve implantation. ☆ All authors take responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation. ⁎ Corresponding author at: Department of Cardiovascular Surgery, Shizuoka Medical Center, 762-1 Nagasawa, Shimizu-cho, Sunto-gun, Shizuoka 411-8611, Japan. E-mail address: [email protected] (H. Takagi).
http://dx.doi.org/10.1016/j.ijcard.2016.07.006 0167-5273/© 2016 Elsevier Ireland Ltd. All rights reserved.
whereas other researchers [5,6] indicated a non-significant b2-fold risk. Only one previous (published in 2013) meta-analysis [7] of 9 studies [8–16] demonstrated a 2.27-fold risk of 1-year mortality for ≥moderate PAR. The meta-analysts [7], however, abstracted (then pooling in a meta-analysis) “adjusted” (using a “multivariate” analysis) hazard ratios (HRs) from only 3 studies [9,10,16] and “unadjusted” (using a “univariate” analysis) HRs from the other 6 studies [8,11–15]. One [16] of the 3 adjusted HRs was for not overall (including early) but only late (not including early) mortality. Furthermore, of the 6 unadjusted HRs, one [8] was for not all-cause but cardiovascular mortality, and another [13] for not ≥moderate (not including mild) but no/trace or mild (≤mild) (including mild) PAR. Unadjusted results of observational studies should be always interpreted with caution when they are included in metaanalyses because of greater potential biases compared with adjusted results. Particular concerns arise with respect to differences between patients in different groups (selection bias). It would usually be appropriate to analyze adjusted (rather than unadjusted) effect estimates, i.e. analyses that attempt to control for confounding [17]. To
H. Takagi, T. Umemoto / International Journal of Cardiology 221 (2016) 46–51
determine whether ≥ moderate post-TAVI PAR “independently” impairs overall survival and how much the impact on survival is, we performed an updated meta-analysis pooling not unadjusted but “adjusted” HRs.
47
in mortality with ≥ moderate PAR (HR, 2.12; 95% CI, 1.79 to 2.51; p b 0.00001; Fig. 1). To assess the impact of qualitative heterogeneity in study design and patient selection on the pooled effect estimate, we performed several sensitivity analyses (Fig. 2). First, excluding the study by Heinz et al. [4] with the greatest harm of ≥ moderate PAR (odds ratio [OR], 4.72; 95% CI, 1.83 to 12.19) generated still a statistically significant result disfavoring ≥ moderate PAR (HR, 2.07; 95% CI, 1.75 to 2.45; p b 0.00001). Second, eliminating the study by Miyazaki et al. [6] with the least harm of ≥ moderate PAR (HR, 1.09; 95% CI, 0.63 to 1.89) indicated still a statistically significant result disfavoring ≥ moderate PAR (HR, 2.19; 95% CI, 1.85 to 2.59; p b 0.00001). Third, excluding the study by Jerez-Valero et al. [24] with the greatest weight of 10.1% demonstrated still statistically significant harm of ≥ moderate PAR (HR, 2.18; 95% CI, 1.80 to 2.64; p b 0.00001). Fourth, eliminating 3 HRs for only late (not including in-hospital) mortality from the studies by De Carlo et al. [23], Ruparelia et al. [26], and Tamburino et al. [16] generated still statistically significant harm of ≥ moderate PAR (HR, 2.12; 95% CI, 1.75 to 2.57; p b 0.00001). Finally, exclusion of any single study from the meta-analysis did not substantively alter the overall result disfavoring ≥ moderate PAR (Fig. 2). To assess the impact of valve types among the studies on the pooled estimate, the effects of ≥ moderate PAR on mortality were explored separately in studies using exclusively CoreValve and SAPIEN. Subgroup meta-analyses indicated statistically significant increases in mortality with ≥moderate PAR for CoreValve (HR, 2.17; 95% CI, 1.63 to 2.90; p b 0.00001) and SAPIEN studies (HR, 2.27; 95% CI, 1.64 to 3.15; p b 0.00001), and there was no subgroup difference (p = 0.84; Supplemental Fig. S1). Additionally, a maximum likelihood random-effects meta-regression analysis was performed to determine whether the effect of ≥ moderate PAR was modulated by the proportion (%) of patients with CoreValve. The meta-regression coefficient (slope of the meta-regression line) was not statistically significant (−0.0015; 95% CI, −0.0070 to 0.0041; 2-sided p = 0.61; Supplemental Fig. S2), which suggested that the effect of ≥moderate PAR was not modulated by valve types.
2. Methods All adjusted observational studies, which compared overall survival in patients with ≥moderate versus ≤mild PAR after TAVI for severe AS, were identified using a 2-level search strategy. First, databases including MEDLINE and EMBASE were searched through January 2016 using Web-based search engines (PubMed and OVID). Search terms included paravalvular or perivalvular; regurgitation, leak, or leakage; percutaneous, transcatheter, transluminal, transarterial, transapical, transaortic, transcarotid, transaxillary, transsubclavian, transiliac, transfemoral, or transiliofemoral; and aortic valve. Second, relevant studies were identified through a manual search of secondary sources including references of initially identified articles and a search of reviews and commentaries. All references were downloaded for consolidation, elimination of duplicates, and further analysis. Studies considered for inclusion met the following criteria: the design was an observational comparative study; the study population was patients undergoing TAVI for severe AS; patients were divided into ≥moderate and ≤mild post-TAVI PAR; outcomes included ≥1-year all-cause mortality; and the adjustment method was a multivariate Cox proportional hazards analysis. An adjusted HR with its 95% confidence interval (CI) for ≥moderate post-TAVI PAR was abstracted from each individual study. Study-specific estimates were combined using inverse variance-weighted averages of logarithmic HRs in the random-effects model. Sensitivity analyses were performed to assess the contribution of each study to the pooled estimate by excluding individual studies one at a time and recalculating the pooled HR estimates for the remaining studies. Publication bias was assessed graphically using a funnel plot and mathematically using the adjusted rank-correlation test of Begg and Mazumdar [18] and the linear regression test of Egger et al. [19]. All analyses were conducted using Review Manager version 5.3 (available from http://tech.cochrane.org/ revman) and Comprehensive Meta-Analysis version 3 (Biostat, Englewood, NJ).
3. Results Our search identified 17 eligible studies [3–6,9,10,14–16,20–27] including a total of 15,131 patients (Tables 1 and 2). All but 2 [5,6] studies indicated statistically significant harm of ≥ moderate postTAVI PAR for overall all-cause mortality. A pooled analysis of all the 17 studies demonstrated a statistically significant 2.12-fold increase Table 1 Study characteristics. Study
Patient
Aortic valve Area (cm2)
Valve type (%)
Number Age (year)
Woman (%)
Barbanti 2014[20]
1376
81.6 ± 6.5
52.5
0.4 ± 0.2
Codner 2015 [3] D'Onofrio 2015 [21] de Brito 2015 [22] De Carlo 2013 [23] Duncan 2015 [5] Fraccaro 2012 [9] Gilard (FRANCE 2) 2012 [10] Heinz 2014 [4]
360 338 418 821 850 384 3195
82.1 ± 6.9 80.3 ± 6.7 81.5 ± 7.7 N/A 82 ± 7 80 ± 7 82.7 ± 7.2
56.4 46.4 52.2 N/A 48.0 51.8 49.0
0.6 ± 0.2 49.4 ± 16.8 0.45 ± 0.12 43.8 ± 14.9 0.64 ± 0.18 51.0 ± 16.4 N/A N/A N/A 0.75 ± 0.48 49.7 ± 17.0 0.7 ± 0.2 48.1 ± 16.5
110
Jerez-Valero 2014 [24] 1735 Kodali (PARTNER) 2434 2015 [25] Miyazaki 2015 [6] 514
Predicted mortality (%)
Gradient (mm Hg) Mean
Peak
CoreValve SAPIEN Others L-EuroSCORE
STS-PROM
51.7 ± 15.5
N/A
100
Median, 7.1 (IQR, 4–13) 7.5 ± 4.7 N/A 14.2 ± 11.5 N/A N/A 9.8 ± 8.8 14.4 ± 11.9
78.5 ± 24.4 71.7 73.1 ± 43.9 0 N/A 86.1 100 81 ± 27 52.4 81.3 ± 25.5 N/A N/A 32.6
Median, 83 53.6 (range, 58–97) 81 ± 7 51.1 84.45 ± 7.16 47.4
N/A 0.65 ± 0.20 46 ± 17 0.65 ± 0.19
79.5 ± 7.3
51.0
0.72 ± 0.21 52.5 ± 15.4
N/A
0 26.9 100 13.9 0 47.6 65.9
N/A 1.4 0 0 0 0 0 1.4
19.5 ± 11.2 17.3 ± 3.0 20.2 ± 13.8 N/A Median, 18.5 24.0 ± 15.6 21.9 ± 14.3 Median, 10 (range, 2–40)* 20.8 ± 13.9 26.37 ± 16.25
N/A N/A N/A
N/A 0
100
0
37.5
55.8
6.6 0.9†
Moat (UK TAVI) 2011 [14] Ruparelia 2015 [26] Sinning 2012 [15]
870
81.9 ± 7.1
47.6
N/A
N/A
80.9 ± 27.2 52.0
47.1
829 146
82.6 ± 8.2 80.5 ± 6.6
58.4 52.1
81.7 ± 24.5 N/A N/A 100
0
Tamburino 2011 [16] Toggweiler 2013 [27]
663 88
81.0 ± 7.3 83 ± 7
56.0 46.6
N/A 51.2 ± 16.2 0.67 ± 0.15 Median, 38.0 (IQR, 29.0–51.0) N/A 51.8 ± 17.0 0.62 ± 0.17 46 ± 18
83.7 ± 25.1 100 N/A 0
0 100
0
Median, 18.3 (IQR, 10.7–29.0) Median, 18.5 (IQR, 11.7–27.9) 22.7 ± 16.8 30.2 ± 18.0 23.0 ± 13.7 N/A
N/A 7.7 ± 5.2 11.42 ± 4.08 N/A N/A 8.9 ± 9.3 9.8 ± 7.3 N/A Median, 9.0 (IQR, 6.1–12.8)
*EuroSCORE II. †Unknown. AR = aortic regurgitation; IQR = interquartile range; L-EuroSCORE = logistic European System for Cardiac Operative; N/A = not available; risk evaluation; PARTNER = placement of aortic transcatheter valves; and STS-PROM = society of thoracic surgery-predicted risk of mortality.
48
H. Takagi, T. Umemoto / International Journal of Cardiology 221 (2016) 46–51
Table 2 Paravalvular aortic regurgitation. Study
Prevalence (%)
Relative risk for overall mortality Follow-up
No/trace Barbanti 2014[20] Codner 2015 [3] D'Onofrio 2015 [21] de Brito 2015 [22] De Carlo 2013 [23] Duncan 2015 [5] Fraccaro 2012 [9] Gilard (FRANCE 2) 2012 [10] Heinz 2014 [4] Jerez-Valero 2014 [24] Kodali (PARTNER) 2015 [25] Miyazaki 2015 [6] Moat (UK TAVI) 2011 [14] Ruparelia 2015 [26] Sinning 2012 [15] Tamburino 2011 [16] Toggweiler 2013 [27]
80.0 83.6 62.1 23.7 18.0 38.9 95.8 37.8 55.5 41.9 52.9 76.7 86.5 40.9 84.9 79.0 NA
Mild
29.6 67.0 60.8 47.4 45.7 34.5 43.9 38.0
45.7
Moderate
Severe
≥Moderate
NA NA NA NA 19.6 NA NA 15.7 9.1 NA NA NA NA NA NA NA
NA NA NA NA 1.6 NA NA 0.8 0.9 NA NA NA NA NA NA NA
20.0 16.4 8.3 9.2 21.2 13.6 4.2 16.5 10.0 14.2 9.1 23.3 13.5 13.4 15.1 21.0 10.2
Adjusted hazard ratio for ≥moderate PAR
1 year 681 ± 441 days 5 years Median, 343.5 (IQR, 74.3–721.5) days 1 year 5 years 9 ± 8 months 1 year Median, 377 (range, 2–1721) days 21 ± 17 months 1 year Median, 699 (IQR, 307–730) days Range, 11–46 months Median, 568 (maximum, 2677) days 1 year 19 ± 6 months 5 years
2.36 [1.56, 3.56] 4.26 [2.54, 7.15] 2.7 [1.3, 6.2] 2.76 [1.51, 5.04] 1.67 [1.05, 2.57]a 1.25 [0.95, 1.66] 2.19 [1.02, 4.67] 2.49 [1.91, 3.25] 4.717 [1.828, 12.195]b 1.74 [1.41, 2.15]c 1.96 [1.52, 2.51]d 1.09 [0.63, 1.89] 1.66 [1.10, 2.51] 2.30 [1.43, 3.70]a 2.4 [1.0–5.4] 3.785 [1.57, 9.10]a 2.98 [1.44, 6.17]
CI = confidence interval; HR = hazard ratio; IQR = interquartile range; N/A = not available; and PARTNER = placement of aortic transcatheter valves. a Not including in-hospital mortality. b Odds ratio. c Calculated from the HR for ≥moderate versus mild PAR and the HR for ≥moderate versus no/trace PAR. d Calculated from the HR for ≥moderate versus no/trace PAR and the HR for mild versus no/trace PAR.
To assess publication bias we generated a funnel plot of the logarithm of effect size (HR) versus the precision (reciprocal of standard error) for each study (Fig. 3). There was a trend toward statistically significant funnel-plot asymmetry (2-tailed p with continuous correction = 0.058 by Begg test [18]; 2-tailed p = 0.063 by Egger test [19]), which raised the possibility of publication bias. Because of this, we undertook a sensitivity analysis using the trim-and-fill method [28], which conservatively imputes hypothetical negative unpublished studies to mirror the positive studies that cause funnel plot asymmetry. The imputed 5 studies produced a symmetrical funnel plot (Fig. 4). The pooled analysis incorporating the 5 hypothetical studies showed still a statistically significant 1.83-fold increase in mortality with ≥moderate relative to ≤mild PAR (HR, 1.88; 95% CI, 1.57 to 2.24).
4. Discussion The results of our meta-analysis exclusively pooling “adjusted” HRs suggest that ≥ moderate post-TAVI PAR is associated with 2.12-fold “independent” risk of overall (≥ 1-year) all-cause mortality, which is compelling with data from a total of N 15,000 patients in 17 different studies. This result was robust in sensitivity analyses, even after correction of funnel plot asymmetry. In SAVR, the calcified native aortic valve leaflets and annular calcification are resected before the prosthesis is sutured onto the decalcified annulus. In TAVI, however, the native calcified leaflets are not removed and the prosthesis is deployed within the native annulus, crushing the native leaflets against the aortic wall. Thus, PAR is a common
Fig. 1. Forest plot of hazard ratios of ≥moderate (versus ≤mild) paravalvular aortic regurgitation (PAR) for overall all-cause mortality. CI = confidence interval; IV = inverse variance; PARTNER = placement of aortic transcatheter valves; and TAVI = transcatheter aortic valve implantation.
H. Takagi, T. Umemoto / International Journal of Cardiology 221 (2016) 46–51
49
Fig. 2. One-study-removed meta-analysis. CI = confidence interval; PARTNER = placement of aortic transcatheter valves; and TAVI = transcatheter aortic valve implantation.
phenomenon following TAVI, as a result of incomplete annular sealing of the transcatheter valve arising from the gaps between the prosthesis and native annulus [1]. In the meta-analysis by Athappan et al. [7] of 42 studies including a total of 11,528 patients, the pooled estimate for overall incidence of ≥moderate post-TAVI PAR was 11.7% (95% CI, 9.6% to 14.1%). In the meta-analysis by Kondur et al. [2] of 5 studies including a total of 3199 patients, the incidence of ≥moderate PAR at the end of the follow-up period was 8.4% in patients treated with TAVI and 1.5% in patient treated with SAVR, and TAVI was associated with a significant increase in incidence of ≥moderate PAR compared with SAVR (OR, 6.10; 95% CI, 3.66 to 10.16; p b 0.001). The previous meta-analysis (representing a total of 4791 patients) by Athappan et al. [7] pooling not only 3 adjusted but also 6 unadjusted HRs demonstrated unfavorable 1-year mortality in patients with ≥ moderate post-TAVI PAR with an HR of 2.27 (95% CI, 1.84 to 2.81; p = 0.001). The baseline clinical and echocardiographic characteristics
Fig. 3. Funnel plot of precision by logarithmic hazard ratio.
of the patients with no/trace, mild, and ≥moderate PAR, however, are different. A larger body surface area and a higher percentage of males are seen with a greater severity of PAR [25]. Additionally, patients with ≥ moderate PAR have larger ventricular dimensions and mass, and more baseline regurgitant valvular disease, worse ejection fraction, and lower stroke volumes [25]. Being quite distinct from the previous meta-analysis [7], we abstracted not unadjusted but adjusted (using a multivariate Cox proportional hazards regression analysis) HRs and then pooling in our meta-analysis. Furthermore, the present meta-analysis included approximately 2-fold and N3-fold number of studies and patients, respectively, compared with the previous meta-analysis [7]. Jerez-Valero et al. [24] observed no differences in survival at 2-year follow-up between patients with chronic ≥moderate AR (no significant changes in AR severity between baseline and after TAVI) and those
Fig. 4. Funnel plot with correction using the trim-and-fill method. Closes circles and a closed rhombus denote imputed studies and the corrected hazard ratio (with the 95% confidence interval) incorporating the 5 hypothetical studies, respectively.
50
H. Takagi, T. Umemoto / International Journal of Cardiology 221 (2016) 46–51
with ≤mild AR after TAVI, which strongly suggests that the acuteness of AR presentation plays a major role in the deleterious effects of ≥ moderate PAR after TAVI. The absence of AR at baseline was an independent predictor of 1-year mortality (HR, 1.49; 95% CI, 1.04 to 2.11: p = 0.02), i.e. the presence of AR at baseline was an independent protective factor of 1-year mortality, in the FRANCE2 registry [29]. Although chronic AR can remain asymptomatic for years even in the presence of left ventricular (LV) enlargement and dysfunction, acute AR caused by infective endocarditis or type A dissection of the aorta often leads rapidly to heart failure and shock [30]. Differences in clinical impact between these 2 entities may be explained by the presence of compensatory mechanisms in chronic AR, which are lacking in acute AR, finally leading to a rapid increase in end-diastolic pressure relative to regurgitant volume and a low forward stroke volume [24,31]. On the other hand, coexisting and untreated apparent (usually ≥ moderate) mitral regurgitation (MR) is associated with an increase in both early and overall mortality after TAVI for AS. A recent metaanalysis [32] of 8 studies including a total of 9356 patients and 14 studies including a total of 7405 patients respectively demonstrated a statistically significant increase in early (OR 2.17; 95% CI 1.50 to 3.14; p b 0.0001) and overall all-cause mortality (HR 1.81; 95% CI 1.37 to 2.40; p b 0.0001) in patients undergoing TAVI with apparent relative to unapparent coexisting MR. Because the rapid increase in enddiastolic pressures in acute AR usually leads to a worsening in MR to lower diastolic pressures, the degree of MR does not improve in patients with ≥moderate PAR after TAVI and tends to worsen despite the relief of the LV obstruction [24]. In the study by Sinning [15], pre-procedural AR did not correlate with PAR during TAVI (p = 0.65). In the study by Kodali et al. [25], however, patients with ≥ moderate post-implant PAR were more likely to have ≥moderate baseline AR (p b 0.001). Only 2 [6,20] of 17 studies included in the present meta-analysis reported the impact of baseline AR on survival. In a univariate analysis of the study by Barbanti et al. [20], ≥ moderate baseline AR was not associated with an increased hazard of 1-year all-cause (HR, 1.22; 95% CI, 0.91 to 1.64; p = 0.181) or cardiovascular mortality (HR, 1.10; 95% CI, 0.71 to 1.70; p = 0.668). In the study by Miyazaki et al. [6], though ≥ moderate baseline AR was the predictors of 2-year mortality at univariate analysis (HR, 1.58; 95% CI, 1.03 to 2.40; p = 0.033), multivariate analysis did not identify it as an independent predictor (HR, 1.29; 95% CI, 0.83 to 2.00; p = 0.24). Thus, baseline AR may be unassociated with impaired survival. Our meta-analysis must be viewed in the context of its limitations. First, mortality stratified by PAR severity was not a pre-specified primary outcome of interest in some of the included studies. Second, the included studies used different methods and grading schemes for assessment of PAR severity, thereby introducing limitations in quality and completeness of data [7]. Third, our results may be influenced by a publication bias. This risk was minimized through an exhaustive search of the available literature. Although the statistical tests suggested funnel plot asymmetry, the corrected effect estimate from the trim-andfill method [28] demonstrated still a statistically significant 1.83-fold risk of mortality with ≥moderate PAR.
5. Conclusions Despite the aforementioned limitations, we found, on the basis of a meta-analysis of adjusted (not unadjusted) HRs 17 observational studies including a total of N 15,000 patients, that ≥moderate post-TAVI PAR is associated with a 2.12-fold increase in overall (≥ 1-year) all-cause mortality.
Conflicts of interest None.
Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ijcard.2016.07.006. References [1] P.T. Chiam, S.H. Ewe, An update on complications associated with transcatheter aortic valve implantation: stroke, paravalvular leak, atrioventricular block and perforation, Futur. Cardiol. 9 (2013) 733–747. [2] A. Kondur, A. Briasoulis, M. Palla, A. Penumetcha, S. Mallikethi-Reddy, A. Badheka, et al., Meta-analysis of transcatheter aortic valve replacement versus surgical aortic valve replacement in patients with severe aortic valve stenosis, Am. J. Cardiol. 117 (2016) 252–257. [3] P. Codner, K. Orvin, A. Assali, R. Sharony, H. Vaknin-Assa, Y. Shapira, et al., Long-term outcomes for patients with severe symptomatic aortic stenosis treated with transcatheter aortic valve implantation, Am. J. Cardiol. 116 (2015) 1391–1398. [4] A. Heinz, M. DeCillia, G. Feuchtner, S. Mueller, T. Bartel, G. Friedrich, et al., Relative amplitude index: a new tool for hemodynamic evaluation of periprosthetic regurgitation after transcatheter valve implantation, J. Thorac. Cardiovasc. Surg. 147 (2014) 1021–1028 (1029.e1–e2). [5] A. Duncan, P. Ludman, W. Banya, D. Cunningham, D. Marlee, S. Davies, et al., Longterm outcomes after transcatheter aortic valve replacement in high-risk patients with severe aortic stenosis: the U.K. transcatheter aortic valve implantation registry, JACC Cardiovasc. Interv. 8 (2015) 645–653. [6] S. Miyazaki, E. Agricola, V.F. Panoulas, M. Slavich, G. Giustino, T. Miyazaki, et al., Influence of baseline ejection fraction on the prognostic value of paravalvular leak after transcatheter aortic valve implantation, Int. J. Cardiol. 190 (2015) 277–281. [7] G. Athappan, E. Patvardhan, E.M. Tuzcu, L.G. Svensson, P.A. Lemos, C. Fraccaro, et al., Incidence, predictors, and outcomes of aortic regurgitation after transcatheter aortic valve replacement: meta-analysis and systematic review of literature, J. Am. Coll. Cardiol. 61 (2013) 1585–1595. [8] N. Amabile, R. Ramadan, S. Ghostine, S. Cheng, A. Azmoun, F. Raoux, et al., Early and mid-term cardiovascular outcomes following TAVI: impact of pre-procedural transvalvular gradient, Int. J. Cardiol. 167 (2013) 687–692. [9] C. Fraccaro, R. Al-Lamee, G. Tarantini, F. Maisano, M. Napodano, M. Montorfano, et al., Transcatheter aortic valve implantation in patients with severe left ventricular dysfunction: immediate and mid-term results, a multicenter study, Circ. Cardiovasc. Interv. 5 (2012) 253–260. [10] M. Gilard, H. Eltchaninoff, B. Iung, P. Donzeau-Gouge, K. Chevreul, J. Fajadet, et al., FRANCE 2 investigators, Registry of transcatheter aortic-valve implantation in high-risk patients, N. Engl. J. Med. 366 (2012) 1705–1715. [11] K. Hayashida, M.C. Morice, B. Chevalier, T. Hovasse, M. Romano, P. Garot, et al., Sexrelated differences in clinical presentation and outcome of transcatheter aortic valve implantation for severe aortic stenosis, J. Am. Coll. Cardiol. 59 (2012) 566–571. [12] S.K. Kodali, M.R. Williams, C.R. Smith, L.G. Svensson, J.G. Webb, R.R. Makkar, et al., PARTNER trial investigators, Two-year outcomes after transcatheter or surgical aortic-valve replacement, N. Engl. J. Med. 366 (2012) 1686–1695. [13] P.A. Lemos, F. Saia, J. Mariani Jr., C. Marrozzini, A.E. Filho, L.J. Kajita, et al., Residual aortic regurgitation is a major determinant of late mortality after transcatheter aortic valve implantation, Int. J. Cardiol. 157 (2012) 288–289. [14] N.E. Moat, P. Ludman, M.A. de Belder, B. Bridgewater, A.D. Cunningham, C.P. Young, et al., Long-term outcomes after transcatheter aortic valve implantation in high-risk patients with severe aortic stenosis: the U.K. TAVI (United Kingdom Transcatheter Aortic Valve Implantation) registry, J. Am. Coll. Cardiol. 58 (2011) 2130–2138. [15] J.M. Sinning, C. Hammerstingl, M. Vasa-Nicotera, V. Adenauer, S.J. Lema Cachiguango, et al., Aortic regurgitation index defines severity of peri-prosthetic regurgitation and predicts outcome in patients after transcatheter aortic valve implantation, J. Am. Coll. Cardiol. 59 (2012) 1134–1141. [16] C. Tamburino, D. Capodanno, A. Ramondo, A.S. Petronio, F. Ettori, G. Santoro, et al., Incidence and predictors of early and late mortality after transcatheter aortic valve implantation in 663 patients with severe aortic stenosis, Circulation 123 (2011) 299–308. [17] B.C. Reeves, J.J. Deeks, J.P. Higgins, G.A. Wells, Chapter 13: Including nonrandomized studies, in: J.P. Higgins, S. Green (Eds.), Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 (Updated March 2011), The Cochrane Collaboration, 2011 (www.cochrane-handbook.org (accessed on 16.02.01)). [18] C.B. Begg, M. Mazumdar, Operating characteristics of a rank correlation test for publication bias, Biometrics 50 (1994) 1088–1101. [19] M. Egger, G. Davey Smith, M. Schneider, C. Minder, Bias in meta-analysis detected by a simple, graphical test, BMJ 315 (1997) 629–634. [20] M. Barbanti, A.S. Petronio, D. Capodanno, F. Ettori, A. Colombo, F. Bedogni, et al., Impact of balloon post-dilation on clinical outcomes after transcatheter aortic valve replacement with the self-expanding CoreValve prosthesis, JACC Cardiovasc. Interv. 7 (2014) 1014–1021. [21] A. D'Onofrio, M. Facchin, L. Besola, E. Manzan, C. Tessari, E. Bizzotto, et al., Intermediate clinical and hemodynamic outcomes after transcatheter aortic valve implantation, Ann. Thorac. Surg. (Oct 23 2015), http://dx.doi.org/10.1016/j.athoracsur.2015. 08.032 (pii: S0003–4975(15)01393–4. Epub ahead of print). [22] F.S. de Brito Jr., L.A. Carvalho, R. Sarmento-Leite, J.A. Mangione, P. Lemos, A. Siciliano, et al., Brazilian TAVI registry investigators, Outcomes and predictors of mortality after transcatheter aortic valve implantation: results of the Brazilian registry, Catheter. Cardiovasc. Interv. 85 (2015) E153–E162.
H. Takagi, T. Umemoto / International Journal of Cardiology 221 (2016) 46–51 [23] M. De Carlo, C. Giannini, C. Fiorina, F. Bedogni, M. Napodano, S. Klugmann, et al., Paravalvular leak after CoreValve implantation in the Italian registry: predictors and impact on clinical outcome, Int. J. Cardiol. 168 (2013) 5088–5089. [24] M. Jerez-Valero, M. Urena, J.G. Webb, C. Tamburino, A.J. Munoz-Garcia, A. Cheema, et al., Clinical impact of aortic regurgitation after transcatheter aortic valve replacement: insights into the degree and acuteness of presentation, JACC Cardiovasc. Interv. 7 (2014) 1022–1032. [25] S. Kodali, P. Pibarot, P.S. Douglas, M. Williams, K. Xu, V. Thourani, et al., Paravalvular regurgitation after transcatheter aortic valve replacement with the Edwards sapien valve in the PARTNER trial: characterizing patients and impact on outcomes, Eur. Heart J. 36 (2015) 449–456. [26] N. Ruparelia, A. Latib, N. Buzzatti, F. Giannini, F. Figini, A. Mangieri, et al., Long-term outcomes after transcatheter aortic valve implantation from a single high-volume center (the Milan experience), Am. J. Cardiol. (Dec 13 2015), http://dx.doi.org/10. 1016/j.amjcard.2015.12.014 (pii: S0002–9149(15)02363–2. Epub ahead of print). [27] S. Toggweiler, K.H. Humphries, M. Lee, R.K. Binder, R.R. Moss, M. Freeman, et al., 5-year outcome after transcatheter aortic valve implantation, J. Am. Coll. Cardiol. 61 (2013) 413–419.
51
[28] S. Duval, R. Tweedie, Trim and fill: a simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis, Biometrics 56 (2000) 455–463. [29] E. Van Belle, F. Juthier, S. Susen, A. Vincentelli, B. Iung, J. Dallongeville, et al., FRANCE 2 investigators, Postprocedural aortic regurgitation in balloon-expandable and selfexpandable transcatheter aortic valve replacement procedures: analysis of predictors and impact on long-term mortality: insights from the FRANCE2 registry, Circulation 129 (2014) 1415–1427. [30] S.H. Goldbarg, J.L. Halperin, Aortic regurgitation: disease progression and management, Nat. Clin. Pract. Cardiovasc. Med. 5 (2008) 269–279. [31] K.K. Stout, E.D. Verrier, Acute valvular regurgitation, Circulation 119 (2009) 3232–3241. [32] H. Takagi, T. Umemoto, ALICE (All-Literature Investigation of Cardiovascular Evidence) group, Coexisting mitral regurgitation impairs survival after transcatheter aortic valve implantation, Ann. Thorac. Surg. 100 (2015) 2270–2276.
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Impact of paravalvular aortic regurgitation after transcatheter aortic valve implantation on survival☆ Hisato Takagi ⁎, Takuya Umemoto, for the ALICE (All-Literature Investigation of Cardiovascular Evidence) Group Department of Cardiovascular Surgery, Shizuoka Medical Center, Shizuoka, Japan
a r t i c l e
i n f o
Article history: Received 21 April 2016 Received in revised form 1 July 2016 Accepted 2 July 2016 Available online 04 July 2016 Keywords: Meta-analysis Paravalvular aortic regurgitation Survival Transcatheter aortic valve implantation
a b s t r a c t Objectives: To determine whether ≥moderate paravalvular aortic regurgitation (PAR) after transcatheter aortic valve implantation (TAVI) independently impairs overall survival and how much the impact on survival is, we performed an updated meta-analysis pooling not unadjusted but adjusted hazard ratios (HRs). Methods: Databases including MEDLINE and EMBASE were searched through January 2016 using PubMed and OVID. Search terms included paravalvular or perivalvular; regurgitation, leak, or leakage; percutaneous, transcatheter, transluminal, transarterial, transapical, transaortic, transcarotid, transaxillary, transsubclavian, transiliac, transfemoral, or transiliofemoral; and aortic valve. Studies considered for inclusion met the following criteria: the design was an observational comparative study; the study population was patients undergoing TAVI; patients were divided into ≥moderate and ≤mild post-TAVI PAR; outcomes included ≥1-year all-cause mortality; and the adjustment method was a multivariate Cox proportional hazards analysis. An adjusted HR with its 95% confidence interval (CI) for ≥moderate post-TAVI PAR was abstracted from each individual study. Results: Our search identified 17 eligible studies including a total of 15,131 patients. A pooled analysis of all the 17 studies demonstrated a statistically significant 2.12-fold increase in mortality with ≥moderate PAR (HR, 2.12; 95% CI, 1.79 to 2.51; p b 0.00001). Exclusion of any single study from the meta-analysis did not substantively alter the overall result disfavoring ≥ moderate PAR. Although the statistical tests suggested funnel plot asymmetry, the corrected effect estimate from the trim-and-fill method demonstrated still a statistically significant 1.83-fold risk of mortality with ≥moderate PAR. Conclusions: ≥ Moderate post-TAVI PAR is associated with a 2.12-fold increase in overall (≥ 1-year) all-cause mortality. © 2016 Elsevier Ireland Ltd. All rights reserved.
1. Introduction The incidence of moderate or severe (≥moderate) paravalvular aortic regurgitation (paravalvular AR, PAR) after transcatheter aortic valve implantation (TAVI) for severe aortic stenosis (AS) has been reported to be 12–21% [1], which is approximately 6-fold higher than that after surgical aortic valve replacement (SAVR) [2]. A number of studies suggest that ≥moderate PAR is an independent risk factor of overall mortality. However, some investigators [3,4] demonstrated a significant N4-fold risk,
Abbreviations: AR, aortic regurgitation; AS, aortic stenosis; CI, confidence interval; HR, hazard ratio; LV, left ventricle; MR, mitral regurgitation; OR, odds ratio; PAR, paravalvular AR; SAVR, surgical aortic valve replacement; TAVI, transcatheter aortic valve implantation. ☆ All authors take responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation. ⁎ Corresponding author at: Department of Cardiovascular Surgery, Shizuoka Medical Center, 762-1 Nagasawa, Shimizu-cho, Sunto-gun, Shizuoka 411-8611, Japan. E-mail address: [email protected] (H. Takagi).
http://dx.doi.org/10.1016/j.ijcard.2016.07.006 0167-5273/© 2016 Elsevier Ireland Ltd. All rights reserved.
whereas other researchers [5,6] indicated a non-significant b2-fold risk. Only one previous (published in 2013) meta-analysis [7] of 9 studies [8–16] demonstrated a 2.27-fold risk of 1-year mortality for ≥moderate PAR. The meta-analysts [7], however, abstracted (then pooling in a meta-analysis) “adjusted” (using a “multivariate” analysis) hazard ratios (HRs) from only 3 studies [9,10,16] and “unadjusted” (using a “univariate” analysis) HRs from the other 6 studies [8,11–15]. One [16] of the 3 adjusted HRs was for not overall (including early) but only late (not including early) mortality. Furthermore, of the 6 unadjusted HRs, one [8] was for not all-cause but cardiovascular mortality, and another [13] for not ≥moderate (not including mild) but no/trace or mild (≤mild) (including mild) PAR. Unadjusted results of observational studies should be always interpreted with caution when they are included in metaanalyses because of greater potential biases compared with adjusted results. Particular concerns arise with respect to differences between patients in different groups (selection bias). It would usually be appropriate to analyze adjusted (rather than unadjusted) effect estimates, i.e. analyses that attempt to control for confounding [17]. To
H. Takagi, T. Umemoto / International Journal of Cardiology 221 (2016) 46–51
determine whether ≥ moderate post-TAVI PAR “independently” impairs overall survival and how much the impact on survival is, we performed an updated meta-analysis pooling not unadjusted but “adjusted” HRs.
47
in mortality with ≥ moderate PAR (HR, 2.12; 95% CI, 1.79 to 2.51; p b 0.00001; Fig. 1). To assess the impact of qualitative heterogeneity in study design and patient selection on the pooled effect estimate, we performed several sensitivity analyses (Fig. 2). First, excluding the study by Heinz et al. [4] with the greatest harm of ≥ moderate PAR (odds ratio [OR], 4.72; 95% CI, 1.83 to 12.19) generated still a statistically significant result disfavoring ≥ moderate PAR (HR, 2.07; 95% CI, 1.75 to 2.45; p b 0.00001). Second, eliminating the study by Miyazaki et al. [6] with the least harm of ≥ moderate PAR (HR, 1.09; 95% CI, 0.63 to 1.89) indicated still a statistically significant result disfavoring ≥ moderate PAR (HR, 2.19; 95% CI, 1.85 to 2.59; p b 0.00001). Third, excluding the study by Jerez-Valero et al. [24] with the greatest weight of 10.1% demonstrated still statistically significant harm of ≥ moderate PAR (HR, 2.18; 95% CI, 1.80 to 2.64; p b 0.00001). Fourth, eliminating 3 HRs for only late (not including in-hospital) mortality from the studies by De Carlo et al. [23], Ruparelia et al. [26], and Tamburino et al. [16] generated still statistically significant harm of ≥ moderate PAR (HR, 2.12; 95% CI, 1.75 to 2.57; p b 0.00001). Finally, exclusion of any single study from the meta-analysis did not substantively alter the overall result disfavoring ≥ moderate PAR (Fig. 2). To assess the impact of valve types among the studies on the pooled estimate, the effects of ≥ moderate PAR on mortality were explored separately in studies using exclusively CoreValve and SAPIEN. Subgroup meta-analyses indicated statistically significant increases in mortality with ≥moderate PAR for CoreValve (HR, 2.17; 95% CI, 1.63 to 2.90; p b 0.00001) and SAPIEN studies (HR, 2.27; 95% CI, 1.64 to 3.15; p b 0.00001), and there was no subgroup difference (p = 0.84; Supplemental Fig. S1). Additionally, a maximum likelihood random-effects meta-regression analysis was performed to determine whether the effect of ≥ moderate PAR was modulated by the proportion (%) of patients with CoreValve. The meta-regression coefficient (slope of the meta-regression line) was not statistically significant (−0.0015; 95% CI, −0.0070 to 0.0041; 2-sided p = 0.61; Supplemental Fig. S2), which suggested that the effect of ≥moderate PAR was not modulated by valve types.
2. Methods All adjusted observational studies, which compared overall survival in patients with ≥moderate versus ≤mild PAR after TAVI for severe AS, were identified using a 2-level search strategy. First, databases including MEDLINE and EMBASE were searched through January 2016 using Web-based search engines (PubMed and OVID). Search terms included paravalvular or perivalvular; regurgitation, leak, or leakage; percutaneous, transcatheter, transluminal, transarterial, transapical, transaortic, transcarotid, transaxillary, transsubclavian, transiliac, transfemoral, or transiliofemoral; and aortic valve. Second, relevant studies were identified through a manual search of secondary sources including references of initially identified articles and a search of reviews and commentaries. All references were downloaded for consolidation, elimination of duplicates, and further analysis. Studies considered for inclusion met the following criteria: the design was an observational comparative study; the study population was patients undergoing TAVI for severe AS; patients were divided into ≥moderate and ≤mild post-TAVI PAR; outcomes included ≥1-year all-cause mortality; and the adjustment method was a multivariate Cox proportional hazards analysis. An adjusted HR with its 95% confidence interval (CI) for ≥moderate post-TAVI PAR was abstracted from each individual study. Study-specific estimates were combined using inverse variance-weighted averages of logarithmic HRs in the random-effects model. Sensitivity analyses were performed to assess the contribution of each study to the pooled estimate by excluding individual studies one at a time and recalculating the pooled HR estimates for the remaining studies. Publication bias was assessed graphically using a funnel plot and mathematically using the adjusted rank-correlation test of Begg and Mazumdar [18] and the linear regression test of Egger et al. [19]. All analyses were conducted using Review Manager version 5.3 (available from http://tech.cochrane.org/ revman) and Comprehensive Meta-Analysis version 3 (Biostat, Englewood, NJ).
3. Results Our search identified 17 eligible studies [3–6,9,10,14–16,20–27] including a total of 15,131 patients (Tables 1 and 2). All but 2 [5,6] studies indicated statistically significant harm of ≥ moderate postTAVI PAR for overall all-cause mortality. A pooled analysis of all the 17 studies demonstrated a statistically significant 2.12-fold increase Table 1 Study characteristics. Study
Patient
Aortic valve Area (cm2)
Valve type (%)
Number Age (year)
Woman (%)
Barbanti 2014[20]
1376
81.6 ± 6.5
52.5
0.4 ± 0.2
Codner 2015 [3] D'Onofrio 2015 [21] de Brito 2015 [22] De Carlo 2013 [23] Duncan 2015 [5] Fraccaro 2012 [9] Gilard (FRANCE 2) 2012 [10] Heinz 2014 [4]
360 338 418 821 850 384 3195
82.1 ± 6.9 80.3 ± 6.7 81.5 ± 7.7 N/A 82 ± 7 80 ± 7 82.7 ± 7.2
56.4 46.4 52.2 N/A 48.0 51.8 49.0
0.6 ± 0.2 49.4 ± 16.8 0.45 ± 0.12 43.8 ± 14.9 0.64 ± 0.18 51.0 ± 16.4 N/A N/A N/A 0.75 ± 0.48 49.7 ± 17.0 0.7 ± 0.2 48.1 ± 16.5
110
Jerez-Valero 2014 [24] 1735 Kodali (PARTNER) 2434 2015 [25] Miyazaki 2015 [6] 514
Predicted mortality (%)
Gradient (mm Hg) Mean
Peak
CoreValve SAPIEN Others L-EuroSCORE
STS-PROM
51.7 ± 15.5
N/A
100
Median, 7.1 (IQR, 4–13) 7.5 ± 4.7 N/A 14.2 ± 11.5 N/A N/A 9.8 ± 8.8 14.4 ± 11.9
78.5 ± 24.4 71.7 73.1 ± 43.9 0 N/A 86.1 100 81 ± 27 52.4 81.3 ± 25.5 N/A N/A 32.6
Median, 83 53.6 (range, 58–97) 81 ± 7 51.1 84.45 ± 7.16 47.4
N/A 0.65 ± 0.20 46 ± 17 0.65 ± 0.19
79.5 ± 7.3
51.0
0.72 ± 0.21 52.5 ± 15.4
N/A
0 26.9 100 13.9 0 47.6 65.9
N/A 1.4 0 0 0 0 0 1.4
19.5 ± 11.2 17.3 ± 3.0 20.2 ± 13.8 N/A Median, 18.5 24.0 ± 15.6 21.9 ± 14.3 Median, 10 (range, 2–40)* 20.8 ± 13.9 26.37 ± 16.25
N/A N/A N/A
N/A 0
100
0
37.5
55.8
6.6 0.9†
Moat (UK TAVI) 2011 [14] Ruparelia 2015 [26] Sinning 2012 [15]
870
81.9 ± 7.1
47.6
N/A
N/A
80.9 ± 27.2 52.0
47.1
829 146
82.6 ± 8.2 80.5 ± 6.6
58.4 52.1
81.7 ± 24.5 N/A N/A 100
0
Tamburino 2011 [16] Toggweiler 2013 [27]
663 88
81.0 ± 7.3 83 ± 7
56.0 46.6
N/A 51.2 ± 16.2 0.67 ± 0.15 Median, 38.0 (IQR, 29.0–51.0) N/A 51.8 ± 17.0 0.62 ± 0.17 46 ± 18
83.7 ± 25.1 100 N/A 0
0 100
0
Median, 18.3 (IQR, 10.7–29.0) Median, 18.5 (IQR, 11.7–27.9) 22.7 ± 16.8 30.2 ± 18.0 23.0 ± 13.7 N/A
N/A 7.7 ± 5.2 11.42 ± 4.08 N/A N/A 8.9 ± 9.3 9.8 ± 7.3 N/A Median, 9.0 (IQR, 6.1–12.8)
*EuroSCORE II. †Unknown. AR = aortic regurgitation; IQR = interquartile range; L-EuroSCORE = logistic European System for Cardiac Operative; N/A = not available; risk evaluation; PARTNER = placement of aortic transcatheter valves; and STS-PROM = society of thoracic surgery-predicted risk of mortality.
48
H. Takagi, T. Umemoto / International Journal of Cardiology 221 (2016) 46–51
Table 2 Paravalvular aortic regurgitation. Study
Prevalence (%)
Relative risk for overall mortality Follow-up
No/trace Barbanti 2014[20] Codner 2015 [3] D'Onofrio 2015 [21] de Brito 2015 [22] De Carlo 2013 [23] Duncan 2015 [5] Fraccaro 2012 [9] Gilard (FRANCE 2) 2012 [10] Heinz 2014 [4] Jerez-Valero 2014 [24] Kodali (PARTNER) 2015 [25] Miyazaki 2015 [6] Moat (UK TAVI) 2011 [14] Ruparelia 2015 [26] Sinning 2012 [15] Tamburino 2011 [16] Toggweiler 2013 [27]
80.0 83.6 62.1 23.7 18.0 38.9 95.8 37.8 55.5 41.9 52.9 76.7 86.5 40.9 84.9 79.0 NA
Mild
29.6 67.0 60.8 47.4 45.7 34.5 43.9 38.0
45.7
Moderate
Severe
≥Moderate
NA NA NA NA 19.6 NA NA 15.7 9.1 NA NA NA NA NA NA NA
NA NA NA NA 1.6 NA NA 0.8 0.9 NA NA NA NA NA NA NA
20.0 16.4 8.3 9.2 21.2 13.6 4.2 16.5 10.0 14.2 9.1 23.3 13.5 13.4 15.1 21.0 10.2
Adjusted hazard ratio for ≥moderate PAR
1 year 681 ± 441 days 5 years Median, 343.5 (IQR, 74.3–721.5) days 1 year 5 years 9 ± 8 months 1 year Median, 377 (range, 2–1721) days 21 ± 17 months 1 year Median, 699 (IQR, 307–730) days Range, 11–46 months Median, 568 (maximum, 2677) days 1 year 19 ± 6 months 5 years
2.36 [1.56, 3.56] 4.26 [2.54, 7.15] 2.7 [1.3, 6.2] 2.76 [1.51, 5.04] 1.67 [1.05, 2.57]a 1.25 [0.95, 1.66] 2.19 [1.02, 4.67] 2.49 [1.91, 3.25] 4.717 [1.828, 12.195]b 1.74 [1.41, 2.15]c 1.96 [1.52, 2.51]d 1.09 [0.63, 1.89] 1.66 [1.10, 2.51] 2.30 [1.43, 3.70]a 2.4 [1.0–5.4] 3.785 [1.57, 9.10]a 2.98 [1.44, 6.17]
CI = confidence interval; HR = hazard ratio; IQR = interquartile range; N/A = not available; and PARTNER = placement of aortic transcatheter valves. a Not including in-hospital mortality. b Odds ratio. c Calculated from the HR for ≥moderate versus mild PAR and the HR for ≥moderate versus no/trace PAR. d Calculated from the HR for ≥moderate versus no/trace PAR and the HR for mild versus no/trace PAR.
To assess publication bias we generated a funnel plot of the logarithm of effect size (HR) versus the precision (reciprocal of standard error) for each study (Fig. 3). There was a trend toward statistically significant funnel-plot asymmetry (2-tailed p with continuous correction = 0.058 by Begg test [18]; 2-tailed p = 0.063 by Egger test [19]), which raised the possibility of publication bias. Because of this, we undertook a sensitivity analysis using the trim-and-fill method [28], which conservatively imputes hypothetical negative unpublished studies to mirror the positive studies that cause funnel plot asymmetry. The imputed 5 studies produced a symmetrical funnel plot (Fig. 4). The pooled analysis incorporating the 5 hypothetical studies showed still a statistically significant 1.83-fold increase in mortality with ≥moderate relative to ≤mild PAR (HR, 1.88; 95% CI, 1.57 to 2.24).
4. Discussion The results of our meta-analysis exclusively pooling “adjusted” HRs suggest that ≥ moderate post-TAVI PAR is associated with 2.12-fold “independent” risk of overall (≥ 1-year) all-cause mortality, which is compelling with data from a total of N 15,000 patients in 17 different studies. This result was robust in sensitivity analyses, even after correction of funnel plot asymmetry. In SAVR, the calcified native aortic valve leaflets and annular calcification are resected before the prosthesis is sutured onto the decalcified annulus. In TAVI, however, the native calcified leaflets are not removed and the prosthesis is deployed within the native annulus, crushing the native leaflets against the aortic wall. Thus, PAR is a common
Fig. 1. Forest plot of hazard ratios of ≥moderate (versus ≤mild) paravalvular aortic regurgitation (PAR) for overall all-cause mortality. CI = confidence interval; IV = inverse variance; PARTNER = placement of aortic transcatheter valves; and TAVI = transcatheter aortic valve implantation.
H. Takagi, T. Umemoto / International Journal of Cardiology 221 (2016) 46–51
49
Fig. 2. One-study-removed meta-analysis. CI = confidence interval; PARTNER = placement of aortic transcatheter valves; and TAVI = transcatheter aortic valve implantation.
phenomenon following TAVI, as a result of incomplete annular sealing of the transcatheter valve arising from the gaps between the prosthesis and native annulus [1]. In the meta-analysis by Athappan et al. [7] of 42 studies including a total of 11,528 patients, the pooled estimate for overall incidence of ≥moderate post-TAVI PAR was 11.7% (95% CI, 9.6% to 14.1%). In the meta-analysis by Kondur et al. [2] of 5 studies including a total of 3199 patients, the incidence of ≥moderate PAR at the end of the follow-up period was 8.4% in patients treated with TAVI and 1.5% in patient treated with SAVR, and TAVI was associated with a significant increase in incidence of ≥moderate PAR compared with SAVR (OR, 6.10; 95% CI, 3.66 to 10.16; p b 0.001). The previous meta-analysis (representing a total of 4791 patients) by Athappan et al. [7] pooling not only 3 adjusted but also 6 unadjusted HRs demonstrated unfavorable 1-year mortality in patients with ≥ moderate post-TAVI PAR with an HR of 2.27 (95% CI, 1.84 to 2.81; p = 0.001). The baseline clinical and echocardiographic characteristics
Fig. 3. Funnel plot of precision by logarithmic hazard ratio.
of the patients with no/trace, mild, and ≥moderate PAR, however, are different. A larger body surface area and a higher percentage of males are seen with a greater severity of PAR [25]. Additionally, patients with ≥ moderate PAR have larger ventricular dimensions and mass, and more baseline regurgitant valvular disease, worse ejection fraction, and lower stroke volumes [25]. Being quite distinct from the previous meta-analysis [7], we abstracted not unadjusted but adjusted (using a multivariate Cox proportional hazards regression analysis) HRs and then pooling in our meta-analysis. Furthermore, the present meta-analysis included approximately 2-fold and N3-fold number of studies and patients, respectively, compared with the previous meta-analysis [7]. Jerez-Valero et al. [24] observed no differences in survival at 2-year follow-up between patients with chronic ≥moderate AR (no significant changes in AR severity between baseline and after TAVI) and those
Fig. 4. Funnel plot with correction using the trim-and-fill method. Closes circles and a closed rhombus denote imputed studies and the corrected hazard ratio (with the 95% confidence interval) incorporating the 5 hypothetical studies, respectively.
50
H. Takagi, T. Umemoto / International Journal of Cardiology 221 (2016) 46–51
with ≤mild AR after TAVI, which strongly suggests that the acuteness of AR presentation plays a major role in the deleterious effects of ≥ moderate PAR after TAVI. The absence of AR at baseline was an independent predictor of 1-year mortality (HR, 1.49; 95% CI, 1.04 to 2.11: p = 0.02), i.e. the presence of AR at baseline was an independent protective factor of 1-year mortality, in the FRANCE2 registry [29]. Although chronic AR can remain asymptomatic for years even in the presence of left ventricular (LV) enlargement and dysfunction, acute AR caused by infective endocarditis or type A dissection of the aorta often leads rapidly to heart failure and shock [30]. Differences in clinical impact between these 2 entities may be explained by the presence of compensatory mechanisms in chronic AR, which are lacking in acute AR, finally leading to a rapid increase in end-diastolic pressure relative to regurgitant volume and a low forward stroke volume [24,31]. On the other hand, coexisting and untreated apparent (usually ≥ moderate) mitral regurgitation (MR) is associated with an increase in both early and overall mortality after TAVI for AS. A recent metaanalysis [32] of 8 studies including a total of 9356 patients and 14 studies including a total of 7405 patients respectively demonstrated a statistically significant increase in early (OR 2.17; 95% CI 1.50 to 3.14; p b 0.0001) and overall all-cause mortality (HR 1.81; 95% CI 1.37 to 2.40; p b 0.0001) in patients undergoing TAVI with apparent relative to unapparent coexisting MR. Because the rapid increase in enddiastolic pressures in acute AR usually leads to a worsening in MR to lower diastolic pressures, the degree of MR does not improve in patients with ≥moderate PAR after TAVI and tends to worsen despite the relief of the LV obstruction [24]. In the study by Sinning [15], pre-procedural AR did not correlate with PAR during TAVI (p = 0.65). In the study by Kodali et al. [25], however, patients with ≥ moderate post-implant PAR were more likely to have ≥moderate baseline AR (p b 0.001). Only 2 [6,20] of 17 studies included in the present meta-analysis reported the impact of baseline AR on survival. In a univariate analysis of the study by Barbanti et al. [20], ≥ moderate baseline AR was not associated with an increased hazard of 1-year all-cause (HR, 1.22; 95% CI, 0.91 to 1.64; p = 0.181) or cardiovascular mortality (HR, 1.10; 95% CI, 0.71 to 1.70; p = 0.668). In the study by Miyazaki et al. [6], though ≥ moderate baseline AR was the predictors of 2-year mortality at univariate analysis (HR, 1.58; 95% CI, 1.03 to 2.40; p = 0.033), multivariate analysis did not identify it as an independent predictor (HR, 1.29; 95% CI, 0.83 to 2.00; p = 0.24). Thus, baseline AR may be unassociated with impaired survival. Our meta-analysis must be viewed in the context of its limitations. First, mortality stratified by PAR severity was not a pre-specified primary outcome of interest in some of the included studies. Second, the included studies used different methods and grading schemes for assessment of PAR severity, thereby introducing limitations in quality and completeness of data [7]. Third, our results may be influenced by a publication bias. This risk was minimized through an exhaustive search of the available literature. Although the statistical tests suggested funnel plot asymmetry, the corrected effect estimate from the trim-andfill method [28] demonstrated still a statistically significant 1.83-fold risk of mortality with ≥moderate PAR.
5. Conclusions Despite the aforementioned limitations, we found, on the basis of a meta-analysis of adjusted (not unadjusted) HRs 17 observational studies including a total of N 15,000 patients, that ≥moderate post-TAVI PAR is associated with a 2.12-fold increase in overall (≥ 1-year) all-cause mortality.
Conflicts of interest None.
Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ijcard.2016.07.006. References [1] P.T. Chiam, S.H. Ewe, An update on complications associated with transcatheter aortic valve implantation: stroke, paravalvular leak, atrioventricular block and perforation, Futur. Cardiol. 9 (2013) 733–747. [2] A. Kondur, A. Briasoulis, M. Palla, A. Penumetcha, S. Mallikethi-Reddy, A. Badheka, et al., Meta-analysis of transcatheter aortic valve replacement versus surgical aortic valve replacement in patients with severe aortic valve stenosis, Am. J. Cardiol. 117 (2016) 252–257. [3] P. Codner, K. Orvin, A. Assali, R. Sharony, H. Vaknin-Assa, Y. Shapira, et al., Long-term outcomes for patients with severe symptomatic aortic stenosis treated with transcatheter aortic valve implantation, Am. J. Cardiol. 116 (2015) 1391–1398. [4] A. Heinz, M. DeCillia, G. Feuchtner, S. Mueller, T. Bartel, G. Friedrich, et al., Relative amplitude index: a new tool for hemodynamic evaluation of periprosthetic regurgitation after transcatheter valve implantation, J. Thorac. Cardiovasc. Surg. 147 (2014) 1021–1028 (1029.e1–e2). [5] A. Duncan, P. Ludman, W. Banya, D. Cunningham, D. Marlee, S. Davies, et al., Longterm outcomes after transcatheter aortic valve replacement in high-risk patients with severe aortic stenosis: the U.K. transcatheter aortic valve implantation registry, JACC Cardiovasc. Interv. 8 (2015) 645–653. [6] S. Miyazaki, E. Agricola, V.F. Panoulas, M. Slavich, G. Giustino, T. Miyazaki, et al., Influence of baseline ejection fraction on the prognostic value of paravalvular leak after transcatheter aortic valve implantation, Int. J. Cardiol. 190 (2015) 277–281. [7] G. Athappan, E. Patvardhan, E.M. Tuzcu, L.G. Svensson, P.A. Lemos, C. Fraccaro, et al., Incidence, predictors, and outcomes of aortic regurgitation after transcatheter aortic valve replacement: meta-analysis and systematic review of literature, J. Am. Coll. Cardiol. 61 (2013) 1585–1595. [8] N. Amabile, R. Ramadan, S. Ghostine, S. Cheng, A. Azmoun, F. Raoux, et al., Early and mid-term cardiovascular outcomes following TAVI: impact of pre-procedural transvalvular gradient, Int. J. Cardiol. 167 (2013) 687–692. [9] C. Fraccaro, R. Al-Lamee, G. Tarantini, F. Maisano, M. Napodano, M. Montorfano, et al., Transcatheter aortic valve implantation in patients with severe left ventricular dysfunction: immediate and mid-term results, a multicenter study, Circ. Cardiovasc. Interv. 5 (2012) 253–260. [10] M. Gilard, H. Eltchaninoff, B. Iung, P. Donzeau-Gouge, K. Chevreul, J. Fajadet, et al., FRANCE 2 investigators, Registry of transcatheter aortic-valve implantation in high-risk patients, N. Engl. J. Med. 366 (2012) 1705–1715. [11] K. Hayashida, M.C. Morice, B. Chevalier, T. Hovasse, M. Romano, P. Garot, et al., Sexrelated differences in clinical presentation and outcome of transcatheter aortic valve implantation for severe aortic stenosis, J. Am. Coll. Cardiol. 59 (2012) 566–571. [12] S.K. Kodali, M.R. Williams, C.R. Smith, L.G. Svensson, J.G. Webb, R.R. Makkar, et al., PARTNER trial investigators, Two-year outcomes after transcatheter or surgical aortic-valve replacement, N. Engl. J. Med. 366 (2012) 1686–1695. [13] P.A. Lemos, F. Saia, J. Mariani Jr., C. Marrozzini, A.E. Filho, L.J. Kajita, et al., Residual aortic regurgitation is a major determinant of late mortality after transcatheter aortic valve implantation, Int. J. Cardiol. 157 (2012) 288–289. [14] N.E. Moat, P. Ludman, M.A. de Belder, B. Bridgewater, A.D. Cunningham, C.P. Young, et al., Long-term outcomes after transcatheter aortic valve implantation in high-risk patients with severe aortic stenosis: the U.K. TAVI (United Kingdom Transcatheter Aortic Valve Implantation) registry, J. Am. Coll. Cardiol. 58 (2011) 2130–2138. [15] J.M. Sinning, C. Hammerstingl, M. Vasa-Nicotera, V. Adenauer, S.J. Lema Cachiguango, et al., Aortic regurgitation index defines severity of peri-prosthetic regurgitation and predicts outcome in patients after transcatheter aortic valve implantation, J. Am. Coll. Cardiol. 59 (2012) 1134–1141. [16] C. Tamburino, D. Capodanno, A. Ramondo, A.S. Petronio, F. Ettori, G. Santoro, et al., Incidence and predictors of early and late mortality after transcatheter aortic valve implantation in 663 patients with severe aortic stenosis, Circulation 123 (2011) 299–308. [17] B.C. Reeves, J.J. Deeks, J.P. Higgins, G.A. Wells, Chapter 13: Including nonrandomized studies, in: J.P. Higgins, S. Green (Eds.), Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 (Updated March 2011), The Cochrane Collaboration, 2011 (www.cochrane-handbook.org (accessed on 16.02.01)). [18] C.B. Begg, M. Mazumdar, Operating characteristics of a rank correlation test for publication bias, Biometrics 50 (1994) 1088–1101. [19] M. Egger, G. Davey Smith, M. Schneider, C. Minder, Bias in meta-analysis detected by a simple, graphical test, BMJ 315 (1997) 629–634. [20] M. Barbanti, A.S. Petronio, D. Capodanno, F. Ettori, A. Colombo, F. Bedogni, et al., Impact of balloon post-dilation on clinical outcomes after transcatheter aortic valve replacement with the self-expanding CoreValve prosthesis, JACC Cardiovasc. Interv. 7 (2014) 1014–1021. [21] A. D'Onofrio, M. Facchin, L. Besola, E. Manzan, C. Tessari, E. Bizzotto, et al., Intermediate clinical and hemodynamic outcomes after transcatheter aortic valve implantation, Ann. Thorac. Surg. (Oct 23 2015), http://dx.doi.org/10.1016/j.athoracsur.2015. 08.032 (pii: S0003–4975(15)01393–4. Epub ahead of print). [22] F.S. de Brito Jr., L.A. Carvalho, R. Sarmento-Leite, J.A. Mangione, P. Lemos, A. Siciliano, et al., Brazilian TAVI registry investigators, Outcomes and predictors of mortality after transcatheter aortic valve implantation: results of the Brazilian registry, Catheter. Cardiovasc. Interv. 85 (2015) E153–E162.
H. Takagi, T. Umemoto / International Journal of Cardiology 221 (2016) 46–51 [23] M. De Carlo, C. Giannini, C. Fiorina, F. Bedogni, M. Napodano, S. Klugmann, et al., Paravalvular leak after CoreValve implantation in the Italian registry: predictors and impact on clinical outcome, Int. J. Cardiol. 168 (2013) 5088–5089. [24] M. Jerez-Valero, M. Urena, J.G. Webb, C. Tamburino, A.J. Munoz-Garcia, A. Cheema, et al., Clinical impact of aortic regurgitation after transcatheter aortic valve replacement: insights into the degree and acuteness of presentation, JACC Cardiovasc. Interv. 7 (2014) 1022–1032. [25] S. Kodali, P. Pibarot, P.S. Douglas, M. Williams, K. Xu, V. Thourani, et al., Paravalvular regurgitation after transcatheter aortic valve replacement with the Edwards sapien valve in the PARTNER trial: characterizing patients and impact on outcomes, Eur. Heart J. 36 (2015) 449–456. [26] N. Ruparelia, A. Latib, N. Buzzatti, F. Giannini, F. Figini, A. Mangieri, et al., Long-term outcomes after transcatheter aortic valve implantation from a single high-volume center (the Milan experience), Am. J. Cardiol. (Dec 13 2015), http://dx.doi.org/10. 1016/j.amjcard.2015.12.014 (pii: S0002–9149(15)02363–2. Epub ahead of print). [27] S. Toggweiler, K.H. Humphries, M. Lee, R.K. Binder, R.R. Moss, M. Freeman, et al., 5-year outcome after transcatheter aortic valve implantation, J. Am. Coll. Cardiol. 61 (2013) 413–419.
51
[28] S. Duval, R. Tweedie, Trim and fill: a simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis, Biometrics 56 (2000) 455–463. [29] E. Van Belle, F. Juthier, S. Susen, A. Vincentelli, B. Iung, J. Dallongeville, et al., FRANCE 2 investigators, Postprocedural aortic regurgitation in balloon-expandable and selfexpandable transcatheter aortic valve replacement procedures: analysis of predictors and impact on long-term mortality: insights from the FRANCE2 registry, Circulation 129 (2014) 1415–1427. [30] S.H. Goldbarg, J.L. Halperin, Aortic regurgitation: disease progression and management, Nat. Clin. Pract. Cardiovasc. Med. 5 (2008) 269–279. [31] K.K. Stout, E.D. Verrier, Acute valvular regurgitation, Circulation 119 (2009) 3232–3241. [32] H. Takagi, T. Umemoto, ALICE (All-Literature Investigation of Cardiovascular Evidence) group, Coexisting mitral regurgitation impairs survival after transcatheter aortic valve implantation, Ann. Thorac. Surg. 100 (2015) 2270–2276.