Incidence, Causes, and Outcomes Associated With Urgent Implantation of a Supplementary Valve During Transcatheter Aortic Valve Replacement

Key Points

Question  What are the incidence, causes, and outcomes of transcatheter aortic valve replacement (TAVR) when a supplementary valve is needed urgently during the procedure?

Findings  In this cohort study of 213 patients undergoing a 2-valve TAVR (2V-TAVR), 2V-TAVR incidence decreased over time and was 1.0% overall. The use of 2V-TAVR was associated with certain patient and procedural characteristics, more periprocedural complications, and increased mortality at 30-day follow-up but not at 1 year.

Meaning  Because 2V-TAVR was associated with a worse outcome in this study, its role remains unclear, especially in patients at low surgical risk.

Importance  Transcatheter aortic valve replacement (TAVR) failure is often managed by an urgent implantation of a supplementary valve during the procedure (2-valve TAVR [2V-TAVR]). Little is known about the factors associated with or sequelae of 2V-TAVR.

Objective  To examine the incidence, causes, and outcomes of 2V-TAVR.

Design, Setting, and Participants  A retrospective cohort study was performed using data from an international registry of 21 298 TAVR procedures performed from January 1, 2014, through February 28, 2019. Among the 21 298 patients undergoing TAVR, 223 patients (1.0%) undergoing 2V-TAVR were identified. Patient-level data were available for all the patients undergoing 2V-TAVR and for 12 052 patients (56.6%) undergoing 1V-TAVR. After excluding patients with missing 30-day follow-up or data inconsistencies, 213 2V-TAVR and 10 010 1V-TAVR patients were studied. The 2V-TAVR patients were compared against control TAVR patients undergoing a 1-valve TAVR (1V-TAVR) using 1:4 17 propensity score matching. Final analysis included 1065 (213:852) patients.

Exposures  Urgent implantation of a supplementary valve during TAVR.

Main Outcomes and Measures  Mortality at 30 days and 1 year.

Results  The 213 patients undergoing 2V-TAVR had similar age (mean [SD], 81.3 [0.5] years) and sex (110 [51.6%] female) as the 10 010 patients undergoing 1V-TAVR (mean [SD] age, 81.2 [0.5] years; 110 [51.6%] female). The 2V-TAVR incidence decreased from 2.9% in 2014 to 1.0% in 2018 and was similar between repositionable and nonrepositionable valves. Bicuspid aortic valve (odds ratio [OR], 2.20; 95% CI, 1.17-4.15; P = .02), aortic regurgitation of moderate or greater severity (OR, 2.02; 95% CI, 1.49-2.73; P < .001), atrial fibrillation (OR, 1.43; 95% CI, 1.07-1.93; P = .02), alternative access (OR, 2.59; 95% CI, 1.72-3.89; P < .001), early-generation valve (OR, 2.32; 95% CI, 1.69-3.19; P < .001), and self-expandable valve (OR, 1.69; 95% CI, 1.17-2.43; P = .004) were associated with higher 2V-TAVR risk. In 165 patients (80%), the supplementary valve was implanted because of residual aortic regurgitation after primary valve malposition (94 [46.4%] too high and 71 [34.2%] too low). In the matched 2V-TAVR vs 1V-TAVR cohorts, the rate of device success was 147 (70.4%) vs 783 (92.2%) (P < .001), the rate of coronary obstruction was 5 (2.3%) vs 3 (0.4%) (P = .10), stroke rate was 9 (4.6%) vs 13 (1.6%) (P = .09), major bleeding rates were 25 (11.8%) vs 46 (5.5%) (P = .03) and annular rupture rate was 7 (3.3%) vs 3 (0.4%) (P = .03). The hazard ratios for mortality were 2.58 (95% CI, 1.04-6.45; P = .04) at 30 days, 1.45 (95% CI, 0.84-2.51; P = .18) at 1 year, and 1.20 (95% CI, 0.77-1.88; P = .42) at 2 years. Nontransfemoral access and certain periprocedural complications were independently associated with higher risk of death 1 year after 2V-TAVR.

Conclusions and Relevance  In this cohort study, valve malposition was the most common indication for 2V-TAVR. Incidence decreased over time and was low overall, although patients with a bicuspid or regurgitant aortic valve, nontransfemoral access, and early-generation or self-expandable valve were at higher risk. These findings suggest that compared with 1V-TAVR, 2V-TAVR is associated with high burden of complications and mortality at 30 days but not at 1 year.

Although transcatheter aortic valve replacement (TAVR) has matured considerably, various factors, such as valve malposition, sizing errors, and adverse aortic root anatomy, can potentially result in suboptimal valve performances. At times these suboptimal performances may be manageable with urgent implantation of a supplementary valve. Currently, existing data on 2-valve TAVR (2V-TAVR) associating factors and sequelae are scarce. We aimed to assess the incidence, causes, and clinical sequelae of 2V-TAVR in a large, multicenter cohort.

An investigator-initiated registry commenced in February 2019 as part of the Redo-TAVR registry described in detail elsewhere.^1,2 Briefly, the Redo-TAVR registry was designed to collect data on patients who underwent additional TAVR for transcatheter prosthesis dysfunction from 2010 through February 2019. A total of 37 centers from Europe, North America, and the Middle East contributed their patient-level data using a dedicated case report form and 21 298 TAVR procedures were identified. Data on 223 consecutive patients (1.1%) who underwent 2V-TAVR (attempted implantation of a supplementary TAV during a single procedure) were collected from 16 participating centers, and data from 12 052 patients (56.6%) who underwent 1V-TAVR (without use of a second device; control group) were collected from 15 participating centers from January 1, 2014, through February 28, 2019. After excluding patients with missing 30-day follow-up or data inconsistencies, 213 2V-TAVR and 10 010 1V-TAVR patients were studied (Figure 1). For each valve, data on implantation date, model, and size were collected. Baseline aortic valve area, mean and maximal gradients, and degree and mechanism of regurgitation were gathered from echocardiographic studies before the index procedure and at 30 days and 1 year later. Echocardiographic data from transthoracic echocardiography were reported by each site according to established guidelines.³ Follow-up data were obtained from outpatient visits, telephone interviews, and medical records at each participating center. Inconsistencies were resolved directly by communicating with the local investigators. The inclusion of patients was approved in each center by a local ethics committee and informed consent was not required. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Baseline demographics, clinical and echocardiographic features, and procedural and follow-up data were collected by the coinvestigators at each institution. Data collection and monitoring regarding the outcomes were assessed according to the Valve Academic Research Consortium 2 (VARC-2) definitions.³ The second valve was implanted at the discretion of the operator, and the indications were reported by each site by review of procedural records and/or angiographic images.

Principal end points were defined as mortality at 30 days and 1 year. Secondary end points included key VARC-2–defined outcomes with the exception of a modified definition for device success defined as successful valve delivery (in 2V-TAVR, of the additional valve) with correct positioning, intended performance (mean gradient <20 mm Hg and less than moderate aortic regurgitation), and freedom from a major cardiac structural complication (coronary obstruction, tamponade, or annular rupture). Patients were classified to be at low to intermediate vs high to prohibitive risk if they had a predicted 30-day surgical mortality of 8% or higher, respectively, as determined by the Society of Thoracic Surgeons (STS) mortality risk model. Valves were classified as early-generation (SAPIEN XT [Edwards Lifesciences], CoreValve [Medtronic], Lotus [Boston Scientific], or Portico [Abbott Laboratories]) or newer-generation (all other models), as repositionable (Evolut-R and Evolut-Pro [Medtronic], Portico [Abbott Laboratories], LOTUS and LOTUS Edge [Boston Scientific], Allegra [New Valve Technology]) or nonrepositionable (all other models), and as balloon-expandable (Edwards valve family [Edwards Lifesciences]) or self-expending (all other models) devices. Body surface area of all patients was calculated using the Mosteller formula.⁴

We evaluated the clinical outcomes of 2V-TAVR vs 1V-TAVR by applying 3 statistical approaches: propensity score (PS) matching, PS weighting, and multivariable Cox regression. All statistical analyses were performed in SAS software, version 9.4 (SAS Institute Inc) and R software, version 4.2.0 (R Foundation for Statistical Computing). The R mice package was used for multiple imputation, the R twang package was used for PS weighting, and the SAS macro was used for PS matching.

Results are presented as mean (SD) for normally distributed continuous variables, as median (interquartile range [IQR]) for non–normally distributed continuous variables, and as number (percentage) for categorical data. The 2-tailed t test was used to compare normally distributed continuous variables, and the Wilcoxon rank sum test was used for variables not normally distributed. The χ² and Fisher exact tests were used to compare categorical variables. The cumulative incidence of time-to-event outcomes was estimated by the Kaplan-Meier method, and the median duration of follow-up was calculated based on the method of reverse Kaplan-Meier. Differences between the groups were evaluated with the log-rank test. Cox proportional hazards multivariable analysis for mortality was used generating hazard ratios (HRs) with 95% CIs. The level of significance was defined by a 2-tailed P < .05 for all statistical tests.

We used a fully conditional specification using the multiple imputations via chained equations algorithm in the R mice package to handle missing data under the assumption of missing at random.⁵ In addition to the baseline covariates, we also included the outcome variables in the imputation model to avoid biased estimates. We used 40 imputations. Each of the 40 imputed data sets was separately analyzed with standard methods, and estimates were combined using Rubin rules and the bootstrap method.^6,7 Sensitivity analysis for multiple imputations using raw data without imputation or matching was performed.

Given the potential differences in baseline characteristics between patients undergoing 2V-TAVR and those undergoing 1V-TAVR, the 1 to 4 nearest-neighbor matching algorithm with a caliper 0.2 was conducted to identify a cohort of patients with similar baseline characteristics; thus, clinical outcomes of PS-matched cohorts were compared. In the PS matching, the PS was developed using a logistic regression model from a nonparsimonious approach, and all clinical variables (ie, age, sex, body surface area, New York Heart Association [NYHA] functional class, diabetes, estimated glomerular filtration rate, chronic lung disease, prior coronary artery bypass graft, left ventricular ejection fraction, and mean aortic valve gradient) as well as procedural data (procedural year, access, and first device type [early vs newer generation, balloon- vs self-expanding mechanism, or repositionable vs nonrepositionable]) were included in the analysis. In the PS weighting, patients undergoing 2V-TAVR were assigned a weight of 1, whereas patients undergoing 1V-TAVR were assigned a weight equal to the odds of undergoing 2V-TAVR, the PS divided by 1 minus the PS, which was estimated by the generalized boosted model. Covariate balance was confirmed by standardized mean difference of less than 0.1. The HRs estimated by the multivariable Cox proportional hazards regression models, adjusting for baseline covariates with a P < .10, were retained in the backward elimination.

The 213 patients undergoing 2V-TAVR had similar age (mean [SD], 81.3 [0.5] years), sex (110 [51.6%] female), and STS risk of death (mean [SD] risk score, 5.6 [0.2]) as the 10 010 patients undergoing 1V-TAVR (mean [SD] age, 81.2 [0.5] years; 110 [51.6%] female; and mean [SD] risk score, 5.8 [0.3]). Rates of frailty and most comorbidities were similar, excluding that of atrial fibrillation (70 [33.1%] vs 2487 [25.0%], P = .01) and left ventricular ejection fraction less than 35% (26 [12.4%] vs 779 [7.8%]; P = .04), which were higher in patients undergoing 2V-TAVR. Mean (SD) baseline aortic valve area (0.75 [0.03] cm² vs 0.71 [0.01] cm², P = .28) and mean (SD) gradient (41.8 [1.3] mm Hg vs 45.3 [1.4] mm Hg, P = .57) were also alike, but the rate of aortic regurgitation of moderate or greater severity was twice as high (63 [30.0%] vs 1687 [16.9%], P < .001) and that of bicuspid aortic valve was numerically higher (11 [5.5%] vs 327 [3.3%], P = .12) among patients undergoing 2V-TAVR (Table 1).

The incidence of 2V-TAVR decreased from 23 of 785 (2.9%) in 2014 to 26 of 2747 (1.0%) in 2018 (P = .01) (eFigure 1 in the Supplement). The proportion of procedures performed at earlier period (before 2016) was higher in the 2V-TAVR vs 1V-TAVR group (133 [63.0%] vs 4338 [43.0%], P < .001), as were the rates of alternative (nontransfemoral) TAVR access (29 [13.6%] vs 580 [5.8%]) and early-generation valve models (149 [70.5%] vs 5005 [50.0%, P < .001 for both). The application of valves with self-expandable mechanism was higher in the 2V-TAVR group (170 [81.4%] vs 6806 [68.1%], P < .001), yet that of repositionable valves was similar (68 [31.9%] vs 3263 [32.6%], P = .93).

The logistic regression model identified 3 patient characteristics and 3 procedural factors independently associated with 2V-TAVR: bicuspid aortic valve (odds ratio [OR], 2.20; 95% CI, 1.17-4.15; P = .02), aortic regurgitation moderate or greater severity (OR, 2.02; 95% CI, 1.49-2.73; P < .001), atrial fibrillation (OR, 1.43; 95% CI, 1.07-1.93; P = .02), alternative access (OR, 2.59; 95% CI, 1.72-3.89; P < .001), early-generation valve (OR, 2.32; 95% CI, 1.69-3.19; P < .001), and self-expandable valve (OR, 1.69; 95% CI, 1.17-2.43; P = .004) (eTable 1 in the Supplement).

After the PS matching was applied, the groups were well balanced, with no significant differences in baseline characteristics (Table 1). Standardized mean difference plot, baseline characteristics according to inverse probability weighting, and a summary of valve types in each cohort are available in eFigure 2 and eTables 2 and 3 in the Supplement.

In 165 2V-TAVR cases (80.0%), the incentive to implant an additional valve was residual aortic regurgitation after incorrect positioning of the primary valve (Figure 2), with 94 (45.4%) too high (too aortic) (of these, 30 had frank embolization of the valve to the aorta) and 71 (34.3%) too low (too ventricular) (of these, 3 had frank embolization of the valve to the left ventricle). Primary valve dysfunction most commonly presented as regurgitation, with 91 (53.2%) paravalvular, 32 (18.7%) mixed (intravalvular or paravalvular), 5 (2.9%) intravalvular, 43 (23.1%) undetermined or unknown location, and 4 (2.3%) annular rupture. Reasons that lead to valve malposition dysfunction were specifically identified and reported in 36 patients and included manipulation or postdilatation (n = 19), inappropriate fluoroscopic view or poor visualization (n = 11), pacing failure (n = 10), resuscitation (n = 3), adverse aortic root anatomy (n = 5), absence of calcification (n = 1), and sizing errors (n = 5). In 38 patients (18.4%), the spur was residual aortic regurgitation despite good primary valve position (26 paravalvular, 6 intravalvular, and 6 unknown location). There were 7 cases of annular rupture, 3 detected after and 4 before the second prosthesis was implanted (2V-TAVR as a bailout measure for the rupture).

Table 2 summarizes the procedural and 30-day outcomes. In the matched 2V-TAVR vs 1V-TAVR cohorts, device success was attained in 147 (70.4%) vs 783 (92.2%) (P < .001). The incidence of coronary obstruction (5 [2.3%] vs 3 [0.4%], P = .10), stroke (9 [4.6%] vs 13 [1.6%], P = .09), major bleeding (25 [11.8%] vs 46 [5.5%], P = .03), and annular rupture (7 [3.3%] vs 3 [0.4%], P = .03) was higher in patients undergoing 2V-TAVR vs those undergoing 1V-TAVR. Notably, exclusion of patients in whom annular rupture occurred before the second prosthesis was implanted (n = 4) did not change the difference in outcomes between the 2 cohorts (eTable 4 in the Supplement). The rates of acute kidney injury (19 [9.4%] vs 74 [8.8%], P = .84), major vascular complication (21 [10.3%] vs 96 [11.3%], P = .75), and new permanent pacemaker (40 [19.2%] vs 141 [16.6%], P = .52) were similar between the groups.

In the 2V-TAVR group vs 1V-TAVR group, the mean (SD) aortic valve gradient decreased to 10.4 (0.5) mm Hg vs 9.1 (0.1) mm Hg (P = .02) at 30 days and 10.1 (0.8) mm Hg vs 8.7 (0.1) mm Hg (P = .08) at 1 year. No difference was found in the rate of residual aortic regurgitation of moderate or greater severity at 30 days (19 [9.8%] vs 62 [8.0%], P = .64) or at 1 year (18 [12.4%] vs 59 [10.2%], P = .63) or in symptomatic status (NYHA III or IV at 30 days: 18 [8.5%] vs 68 [11.5%], P = .55; NYHA III or IV at 1 year: 10 [4.6%] vs 59 [9.6%], P = .33) (eFigure 3 in the Supplement).

Median follow-up was 991 days (IQR, 497-1693 days) overall, 879 days (IQR, 453-1427 days) in the 2V-TAVR group, and 1166 days (IQR, 402-1700 days) in the 1V-TAVR group. Predominantly driven by early event rates, mortality was higher after 2V-TAVR (Figure 3). The HR for mortality was 2.58 (95% CI, 1.04-6.45; P = .04) at 30 days, 1.45 (95% CI, 0.84-2.51; P = .18) at 1 year, and 1.20 (95% CI, 0.77-1.88; P = .42) at 2 years. Variables that were independently associated with mortality at 1 year after 2V-TAVR included alternative access (HR, 2.4; 95% CI, 1.1-5.2; P = .03), primary valve malposition (HR, 3.9; 95% CI, 1.4-7.1; P = .03), coronary obstruction (HR, 4.9; 95% CI, 1.2-20.6; P = .03), major bleeding (HR, 3.7; 95% CI, 1.8-7.7; P = .001), and acute kidney injury (HR, 3.7; 95% CI, 1.8-8.0; P = .001). Of note, 5 of 7 patients (71.4%) with annular rupture died in the hospital. Mortality was similar between patients with valve malposition and those with frank embolization of the primary valve (17 [13.2%] vs 5 [15.2%], P = .21 at 30 days; 28 [21.8%] vs 8 [24.2%], P = .27 at 1 year). In subgroup analysis according to procedural period, mortality rates after 2V-TAVR vs 1V-TAVR were 8.0% (95% CI, 1.6%-13.9%) vs 4.3% (95% CI, 2.2%-6.5%) (P = .20) at 30 days and 20.0% (95% CI, 16.8%-25)% vs 15.7% (95% CI, 6.6%-23.9%) (P = .70) at 1 year; 10.8% (95% CI, 5.1%-16.2%) vs 3.4% (95% CI, 1.8%-5.0%) (P = .02) if performed during earlier years and 20.3% (95% CI, 12.6%-27.3%) vs 12.8% (95% CI, 9.8%-15.6%) (P < .001) if performed more recently (2016-2019). These data as well as a sensitivity analysis for multiple imputations using raw data without imputation or matching are available in eTables 5 and 6 and eFigures 4 and 5 in the Supplement.

This cohort study is, to our knowledge, the largest and most updated 2V-TAVR series to date. The main findings are as follows. First, 2V-TAVR incidence has decreased over time from 3% in 2014 to 1% in 2018. Second, 2V-TAVR was associated with a higher burden of complications, morbidity, and mortality compared with matched patients undergoing 1V-TAVR. Third, malposition of the primary valve with residual paravalvular aortic regurgitation was the indication for most supplementary valve implants. Some patient characteristics (mainly bicuspid or regurgitant aortic valve) and procedural factors (TAVR access and type, although not repositionability) independently increased the risk. Fourth, alternative access, malposition of the primary valve, and certain periprocedural complications (eg, annular rupture, coronary obstruction, major bleeding, and acute kidney injury) were associated with increased mortality 1 year after 2V-TAVR.

Transcatheter aortic valve replacement has matured substantially through escalation of procedural volumes and operator experience, improved patient selection, and advances in technology. However, even with more stable delivery systems, better paravalvular sealing systems, and (in some devices) features that allow valve reposition, correct and successful valve deployment remains a challenge. Device failure may occur when the valve is implanted in positions that are too aortic or too ventricular, when malposition or malfunction can result in aortic regurgitation through or around the newly implanted device. If sufficiently severe, congestion, hemodynamic instability, or cardiogenic shock may ensue.^8,9 Implantation of a consecutive and overlapping valve that extends the seal around the annulus can restore normal prosthetic leaflet function and be effective, yet potentially hazardous.

An encouraging finding of this study was the temporal decrease in 2V-TAVR incidence from 2.93% in 2014 to 0.95% in 2018 (P = .01). This incidence is lower than the reported incidence in earlier studies (1.5% to 3.5%) and comparable to more recent data from the Society of Thoracic Surgeons/American College of Cardiology Transcatheter Valve Therapy registry that correspondingly found a temporal reduction in the 2V-TAVR rate.^10-14 Although it may partially reflect the temporal decrease in patients’ surgical risk (potentiating urgent surgery as a bailout technique option), it more likely outlines the learning curve and general advances in TAVR over the years. Higher risk was associated with anatomical (bicuspid) or functional (regurgitant) aortic valve features, which is consistent with previous studies^15-17 that found patients with aortic regurgitation or bicuspid aortic valve to be at risk for residual paravalvular leak and procedural failure. Absence of calcification, relatively large aortic annulus, regurgitant blood flow, and more complex aortic root anatomy likely explain this. Notably, patients with pure aortic regurgitation had similar 2V-TAVR incidence and outcomes compared with the rest of the group, presumably because the number of cases was small (n = 32) and most procedures were performed relatively recently and with newer devices (eTables 7 and 8 in the Supplement). Other predisposing factors included early-generation devices, possibly by reason of technological and experience restrictions in earlier years. The fabric sealing cuffs of newer-generation valves are generally longer, therefore facilitating more reliable sealing and being more forgiving in the event of malpositioning. In line with the FRANCE2 (French Aortic National CoreValve and Edwards) registry,¹⁰ where 2V-TAVR constituted 3% of CoreValve and 1.2% of SAPIEN valve TAVR, self-expandable valves predisposed for 2V-TAVR. Unfortunately, the use of repositionable valves did not moderate the 2V-TAVR risk. Similarly, repositionable valves increased the risk of valve embolization in a recent large cohort study.¹⁸ The strongest predisposing factor for 2V-TAVR was alternative procedural access, which was also associated with higher mortality, as it has repeatedly been the case in diverse TAVR populations and large randomized clinical trials.¹⁹

In 4 of 5 cases, the incentive to implant the second valve was incorrect positioning of the first valve, which resulted in residual aortic regurgitation. Retrospective review of the procedural records and images most commonly did not identify a specific cause but provided various reasons that may have been preventable in a sizable proportion of cases (eg, manipulations, poor visualization, and pacing failure), enhancing the importance of meticulous procedural planning and implementation.

Although 2V-TAVR attained device success in 70.0% of cases and achieved hemodynamic valve function similar to that of 1V-TAVR, it was associated with more periprocedural, life-threatening complications that indeed translated into higher mortality at 30 days. In subgroup analysis, mortality was explicitly higher in patients with primary valve malposition compared with 2V-TAVR with good position of the primary valve. It seems that most adverse effects were driven not by the supplementary valve implantation but by the misplacement of the first one.

Urgent surgery or conservative management are alternative treatment options to 2V-TAVR. Conversion to surgery occurred in 8 patients (3.7%) undergoing 2V-TAVR in whom 5 (62.5%) died in the hospital, whereas data on conversion to surgery was not available for the control TAVR group. In the last Transcatheter Valve Therapy registry report, the overall rate of conversion from TAVR to surgery in the US was 0.6% and decreasing,¹⁴ yet data are lacking on the outcome of urgent surgery after TAVR. Nevertheless, in the current era of patients at low surgical risk receiving TAVR, the threshold for conversion to surgery is inherently much different; therefore, the role of 2V-TAVR vs urgent surgery remains unclear, especially in low-risk patients.

This study has limitations. As an observational study, it lacked independent adjudication of events or an independent core laboratory imaging analysis. In addition, 2V-TAVR was performed based on the operator’s decision without prespecified criteria or technique. All procedures were performed by teams of experts in large TAVR centers and in a miscellany of patients. Outcomes may be less favorable in less experienced centers and nonselected patients, and risks should not be underestimated. Additional factors that were not captured in the study (such as operator experience, aortic root characteristics, and more detailed procedural information) may also be associated with 2V-TAVR risk and outcome. Data on alternative management strategies were similarly not captured; therefore, conclusions regarding 2V-TAVR vs medical management or open-heart surgery are beyond the scope of the study. A total of 4.0% of patients undergoing 2V-TAVR and 17.0% of patients undergoing 1V-TAVR were not included because of missing 30-day data. Although PS matching is a well-accepted approach in observational research to address differences in baseline characteristics, it cannot account for unmeasured bias. Last, the duration of follow-up is limited, and little is known about long-term durability of valves and coronary access difficulties after 2V-TAVR.

In this multicenter cohort study, the incidence of an urgent supplementary valve implantation during TAVR decreased with time and was 1% overall. Two-valve TAVR was most associated with malposition of the primary valve. Patients with certain baseline and procedural features were at higher risk. Although 2V-TAVR was generally effective, it was associated with a high burden of complications, morbidity, and mortality compared with 1V-TAVR.

Accepted for Publication: March 17, 2021.

Published Online: May 19, 2021. doi:10.1001/jamacardio.2021.1145

Corresponding Author: John G. Webb, MD, Centres for Heart Valve and Cardiovascular Innovation, St Paul’s and Vancouver General Hospital, 1081 Burrard St, Vancouver, BC V6Z 1Y6, Canada (webb@providencehealth.bc.ca).

Author Contributions: Drs Landes and Webb had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Landes, Sathananthan, Soendergaard, Van Wiechen, Leon, Kornowski, Webb.

Acquisition, analysis, or interpretation of data: Landes, Witberg, Kim, Codner, Buzzatti, Montorfano, Godfrey, Hildick-Smith, Fraccaro, Tarantini, De Backer, Soendergaard, Okuno, Pilgrim, Rodés-Cabau, Jaffe, Eitan, Sinning, Ielasi, Eltchaninoff, Maurovich-Horvat, Merkely, Guerrero, El Sabbagh, Ruile, Barbanti, Redwood, Van Mieghem, Finkelstein, Bunc, Leon, Kornowski.

Drafting of the manuscript: Landes, Montorfano, Hildick-Smith, Ielasi, Van Wiechen, Webb.

Critical revision of the manuscript for important intellectual content: Landes, Witberg, Sathananthan, Kim, Codner, Buzzatti, Godfrey, Hildick-Smith, Fraccaro, Tarantini, De Backer, Soendergaard, Okuno, Pilgrim, Rodés-Cabau, Jaffe, Eitan, Sinning, Eltchaninoff, Maurovich-Horvat, Merkely, Guerrero, El Sabbagh, Ruile, Barbanti, Redwood, Van Mieghem, Finkelstein, Bunc, Leon, Kornowski, Webb.

Obtained funding: Webb.

Administrative, technical, or material support: Sathananthan, Buzzatti, Godfrey, Hildick-Smith, Okuno, Maurovich-Horvat, Merkely, Van Wiechen, Webb.

Supervision: Landes, Sathananthan, Montorfano, Hildick-Smith, Fraccaro, Sinning, Maurovich-Horvat, Ruile, Barbanti, Van Mieghem, Bunc, Webb.

Conflict of Interest Disclosures: Dr Sathananthan reported serving as a consultant for Edwards Lifesciences and Medtronic during the conduct of the study. Dr Kim reported receiving personal fees from Abbott Laboratories, Boston Scientific, Edwards Lifesciences, Medtronic, and Meril outside the submitted work. Dr Hildick-Smith reported serving as an adviser or proctor to Edwards, Boston, and Medtronic. Dr Okuno reported receiving personal fees from Abbott Labories outside the submitted work. Dr Pilgrim reported receiving grants from Boston Scientific, Biotronik, and Edwards Lifesciences and personal fees from Boston Scientific, Biotronik, and HighLifeSAS outside the submitted work and serving as a proctor for Boston Scientific and Medtronic. Dr Rodés-Cabau reported receiving grants from Edwards Lifesciences and Medtronic outside the submitted work. Dr Sinning reported receiving personal fees from Medtronic, Boston Scientific, and Edwards Lifesciences during the conduct of the study. Dr Maurovich-Horvat reported being a shareholder in Neumann Medical Ltd outside the submitted work. Dr Merkely reported receiving speaker fees from Biotronik and Abbott Laboratories and grants from Medtronic and Boston Scientific outside the submitted work. Dr Guerrero reported receiving grants from Edwards Lifesciences during the conduct of the study and grants from Abbott Structural Heart outside the submitted work. Dr Redwood reported receiving speaker and proctor fees from Edwards Lifesciences during the conduct of the study. Dr Van Mieghem reported receiving grants from Abbott Laboratories, Boston Scientific, Medtronic, Edwards Lifesciences, PulseCath BV, and Daiichi Sankyo during the conduct of the study. Dr Bunc reported proctoring for TAVI, Edwards Lifesciences, Abbott Laboratories, Medtronic, and Meril. Dr Leon reported receiving grants from Edwards Lifescience to Columbia University as part of the clinical research contract and grants from Medtronic to Columbia University during the conduct of the study, nonfinancial support from the Edwards Advisory Board, and nonpaid and nonfinancial support from Medtronic Advisory board outside the submitted work. Dr Webb reported serving as a consultant to Edwards Lifesciences. No other disclosures were reported.