Balloon‐Expandable Versus Self‐Expanding Valves in Bicuspid Aortic Stenosis: Insights From the SWEDEHEART Registry

ABSTRACT

Background

Transcatheter aortic valve replacement (TAVR) is increasingly used in patients with bicuspid aortic valve (BAV) stenosis, but there is limited comparative data on balloon‐expandable (BEV) versus self‐expanding valves (SEV) in this population.

Aim

To compare clinical and hemodynamic outcomes between BEVs and SEVs in patients with BAV stenosis.

Methods

This observational cohort included all patients who underwent TAVR in Sweden between 2016 and 2022. Exclusion criteria included procedures for pure aortic insufficiency and valve‐in‐valve interventions. The analysis focused on Evolut, Sapien, Acurate, and Portico/Navitor valve families. A doubly robust approach was applied combining inverse probability of treatment weighting and multivariable regression. Sensitivity analyses were also conducted.

Results

Of 577 patients, 274 (47.5%) received a BEV. The majority in the SEV group received an Evolut valve (62%). The mean EUROSCORE II‐predicted mortality risk was 4.1% for BEV and 3.6% for SEV. BEVs were used more in patients with reduced ejection fraction (EF ≤ 40%) and larger aortic annuli. There were no significant differences between groups in periprocedural mortality, all‐cause mortality at a median follow‐up of 675 days, or device success. However, SEVs had higher technical success (aOR: 2.21, p = 0.006), lower postprocedural gradients (adjusted coefficient: −3.72, p < 0.001), and reduced risk of prosthesis‐patient mismatch (aOR: 0.10, p = 0.02). SEVs, though, had a higher incidence of paravalvular leakage (aOR: 7.5, p < 0.01).

Conclusion

Both BEVs and SEVs were feasible with similar clinical outcomes in BAV stenosis. SEVs had better hemodynamic outcomes but more paravalvular leakage. Randomized trials are needed to determine the optimal valve choice.

Abbreviations

BAV

bicuspid aortic valve

BEV

balloon expandable valve

PPI

permanent pacemaker implantation

PPM

prosthesis‐patient mismatch

PVL

paravalvular leak

SAVR

surgical aortic valve replacement

SEV

self‐expanding valve

TAV

tricuspid aortic valve

TAVR

transcatheter aortic valve replacement

1. Introduction

Since its introduction in 2002, the indications for transcatheter aortic valve replacement (TAVR) have continuously expanded. Originally reserved for older patients who were inoperable or at high surgical risk [1], TAVR is now also recommended for younger patients—those over 75 years old in Europe and over 65 years old in the USA—as well as those with intermediate or low surgical risk [2, 3]. This shift is supported by studies demonstrating that TAVR outcomes in tricuspid aortic valve (TAV) stenosis are comparable to, or even superior to, those of surgical aortic valve replacement (SAVR) [4, 5, 6]. However, its use in patients with bicuspid aortic valve (BAV) stenosis remains controversial, as these patients—except for those in the NOTION II study [7]—have been systematically excluded from most pivotal randomized controlled trials (RCTs) comparing TAVR and SAVR.

BAV stenosis has an estimated prevalence of 0.5%–1.4% in the general population [8, 9], though regional variability exists. For instance, in the VENUS‐A study evaluating the effect of TAVR in Chinese patients with severe aortic stenosis, 47.5% of the included patients had BAV stenosis [10, 11]. As younger adults with aortic stenosis increasingly undergo TAVR, the proportion of patients with BAV stenosis undergoing TAVR is expected to rise. This trend is further driven by the recent regulatory approvals, including the European Conformity (CE) mark for both the Evolut and Sapien valve families in BAV and the removal of prior restrictions on their use in BAV in the United States [12, 13].

Although TAVR is increasingly used in BAV patients, comparative studies between self‐expanding (SEVs) and balloon‐expandable valves (BEVs) remain scarce. This study aimed to compare clinical and hemodynamic outcomes between BEVs and SEVs in BAV patients using data from a Swedish nationwide registry.

2. Methods

2.1. Study Design and Data Sources

This study is an observational study utilizing data from the SWENTRY registry (SWEdish traNscatheter cardiac intervention regisTRY), which records all patients having undergone TAVR in Sweden since 2008. As a part of the SWEDEHEART (Swedish Web‐system for Enhancement and Development of Evidence‐based Care in Heart Disease Evaluated According to Recommended Therapies) [14, 15] registry, SWENTRY ensures complete nationwide coverage of the Swedish population.

2.2. Patient Selection

The study population comprised all patients who underwent TAVR in Sweden for BAV stenosis between January 1, 2016, and September 30, 2022. Patients undergoing TAVR for pure aortic insufficiency or valve‐in‐valve procedures were excluded. Additionally, cases with missing 30‐day mortality data were excluded, as were patients who did not receive one of the following transcatheter heart valves (THVs): Sapien family, Evolut family, Acurate family, or Portico/Navitor family (Figure 1).

FIGURE 1.

Patient selection process.

2.3. Study Endpoints

The study endpoints included all‐cause mortality at a median follow‐up time of 675 days, and periprocedural mortality, defined as death from any cause occurring within 30 days after the index procedure or beyond 30 days but during the index hospitalization. Other endpoints included device success and technical success during the index hospitalization, as defined by the VARC‐3 criteria [16] (Table S1 for definitions).

Additionally, the study evaluated the occurrence of prosthesis‐patient mismatch (PPM), more‐than‐mild paravalvular leak (PVL), stroke during the index hospitalization, the requirement for permanent pacemaker implantation (PPI), and post‐TAVR mean aortic valve gradients. Severe PPM was defined as a postoperative mean gradient that exceeds 20 mmHg or a Vmax over 3.5 m/s.

2.4. Statistical Analysis

Continuous variables are expressed as mean ± standard deviation, and variance was assessed using Levene′s test. Depending on the variance distribution, group comparisons of continuous variables were performed using either the Student′s t‐test or Welch′s t‐test. Categorical variables are reported as counts and percentages, with group comparisons conducted using the χ² test or Fisher′s exact test, as appropriate.

Missing data in the covariates (Table 1 and Figure S1) were assessed to be missing at random (MAR). They were addressed using a random forest imputation method [17], which has been shown to generate accurate estimates while preserving the data distribution. To minimize potential bias introduced by variables with a high proportion of missing values, these variables were assigned reduced weighting in the imputation process.

TABLE 1.

Baseline characteristics of patients with bicuspid aortic stenosis having undergone TAVR.

Characteristic	Overall N = 577a	BEV N = 274a	SEV N = 303a	p valueb	Missing
Age at TAVR	76.8 (7.8)	76.4 (8.4)	77.2 (7.3)	0.2	0
Sex				< 0.001	0
Male	344 (60%)	183 (67%)	161 (53%)
Female	233 (40%)	91 (33%)	142 (47%)
BMI (kg/m2)	26.4 (5.3)	26.5 (5.1)	26.3 (5.5)	0.6	1
NYHA functional status III or IV	417 (72%)	213 (78%)	204 (67%)	0.005	0
NTproBNP (ng/L)	3,084.7 (5,523.3)	3,270.4 (4,754.0)	2,951.7 (6,019.2)	0.5	124
EUROSCORE‐II predicted mortality (%)	3.8 (4.2)	4.1 (4.8)	3.6 (3.5)	0.2	0
Comorbidities
Hypertension	400 (69%)	179 (65%)	221 (73%)	0.048	0
Diabetes Mellitus	123 (21%)	65 (24%)	58 (19%)	0.2	0
CKD
(eGFR < 60 mL/min/1.73m2)	242 (42%)	113 (41%)	129 (43%)	0.7	1
Peripheral vascular disease	75 (13%)	40 (15%)	35 (12%)	0.3	0
Chronic pulmonary disease	89 (15%)	38 (14%)	51 (17%)	0.3	0
Atrial fibrillation	168 (29%)	93 (34%)	75 (25%)	0.02	0
Past history
Myocardial infarction within 3 months	18 (3.1%)	8 (2.9%)	10 (3.3%)	0.8	0
History of PCI	111 (19%)	48 (18%)	63 (21%)	0.3	0
Prior cardiac surgery	45 (8%)	22 (8%)	23 (8%)	> 0.9	0
Prior cerebrovascular incident	53 (9.2%)	31 (11%)	22 (7.3%)	0.09	0
Previous pacemaker	58 (10%)	30 (11%)	28 (9.2%)	0.5	0
Valve characteristics and echocardiographic features
Aortic valve area (cm2)	0.7 (0.2)	0.7 (0.2)	0.7 (0.2)	0.4	149
Mean aortic valve gradient (mmHg)	49.3 (14.1)	49.2 (14.9)	49.3 (13.3)	> 0.9	3
Maximum aortic valve gradient (mmHg)	80.7 (22.8)	80.0 (24.2)	81.4 (21.6)	0.5	72
Left ventricular ejection fraction				< 0.001	30
≥ 50%	365 (67%)	149 (58%)	216 (74%)
41%–49%	75 (14%)	47 (18%)	28 (9.6%)
≤ 40%	107 (20%)	59 (23%)	48 (16%)
Moderate/severe aortic valve insufficiency	53 (9.4%)	27 (10%)	26 (8.8%)	0.6	16
Moderate/severe mitral valve insufficiency	49 (8.8%)	24 (9.0%)	25 (8.7%)	0.9	22
Annulus diameter (mm)	25.7 (2.6)	26.1 (2.6)	25.4 (2.5)	< 0.001	0
Pulmonary hypertension	26.8 (21.3)	28.0 (21.1)	25.8 (21.3)	0.2	0
Anatomical features
Porcelain aorta	13 (2.3%)	7 (2.6%)	6 (2.0%)	0.6	1
Thorax deformity	19 (3.3%)	15 (5.5%)	4 (1.3%)	0.005	0
Unfavorable anatomy after CABGc	22 (3.8%)	10 (3.6%)	12 (4.0%)	0.8	0

Note: The values in bold represent differences between groups with p < 0.05.

Abbreviations: BEV, balloon expandable valve; BMI, body mass index; CABG, coronary artery bypass surgery; CKD, Chronic Kidney Disease; eGFR, estimated Glomerular Filtration Rate; HFmrEF, Heart Failure with mildly reduced Ejection Fraction; HFrEF, Heart Failure with reduced Ejection Fraction; NTproBNP, N‐terminal pro–B‐type natriuretic peptide; NYHA, New York Heart Association; PCI, Percutaneous Coronary Intervention; SEV, self‐expanding valve.

^a Mean (SD); n (%).

^b Two Sample t‐test; Pearson′s Chi‐squared test; Welch Two Sample t‐test; Fisher′s exact test.

^cRisk for injury of patent bypass grafts.

Missing outcome data (Table S2) were imputed using multiple imputations via chained equations (MICE) with predictive mean matching. Each imputed data set was analyzed separately, and the results from these multiple analyses were then pooled using Rubin′s rule.

To compare outcomes between BEV and SEV, a double robust adjustment [18, 19] methodology was employed. Inverse Probability of Treatment Weighting (IPTW) was used to adjust for confounding, with the weights estimated using energy balancing [20], while targeting the Average Treatment Effect as the estimand. The covariates included in the IPTW model encompassed baseline characteristics (year of TAVR, age at TAVR, sex, body mass index [BMI]), coexisting conditions (hypertension, diabetes mellitus, chronic kidney disease [eGFR ≤ 60 mL/min/1.73 m²], atrial fibrillation, chronic pulmonary disease, peripheral vascular disease), previous medical history (myocardial infarction in the past 3 months, previous PCI, prior heart surgery, previous cerebrovascular incident, prior pacemaker implantation), valve characteristics and echocardiographic features (aortic valve area, mean aortic valve gradient, maximum aortic valve gradient, aortic annulus diameter, systolic pulmonary artery pressure, ejection fraction, moderate/severe mitral insufficiency, moderate/severe aortic insufficiency), laboratory values and clinical features (NT‐proBNP, NYHA functional class III or IV, porcelain aorta, thorax deformity, unfavorable anatomy) as well as procedural characteristics (access site and urgency of TAVR procedure). Covariate balance was assessed through standardized mean differences (SMD) and the Kolmogorov‐Smirnov statistic with thresholds for excellent balance set at SMDs below 0.10 and Kolmogorov‐Smirnov statistics below 0.05. Variance ratios were also evaluated for continuous variables, with values below 2 indicating acceptable balance.

Following weighting, multivariable regression was performed to adjust for any residual confounding. Marginal effects were estimated using G‐computation, and confidence intervals were derived using the sandwich estimator for robust standard errors.

Event frequencies were calculated using both unadjusted and IPTW‐adjusted Kaplan‐Meier survival estimations, with p‐values obtained through the log‐rank test. Weighed Cox regression models were used for all‐cause mortality, and the proportional hazard assumption was tested via Schoenfeld residuals.

To ensure the robustness of our findings, we conducted sensitivity analyses by separately evaluating outcomes for the two CE‐marked valves: the Sapien and the Evolut valves. Additionally, we performed tipping point sensitivity analyses to identify the characteristics of a potential unmeasured binary or continuous confounder that could nullify our statistically significant results. This included examining the prevalence of the theoretical unmeasured confounder in both the BEV and SEV groups, as well as its hypothetical association (odds ratio/hazard ratio) with the outcomes.

All analyses adhere to the statistical significance definition, a two‐tailed α = 0.05. Analyses were conducted in R (R Foundation for Statistical Computing, version 4.3.1).

3. Results

3.1. Baseline Characteristics

A total of 577 patients were included in the study (Table 1), of whom 274 (47.5%) received a BEV. Among the SEV group, the majority received an Evolut valve (62%, n = 188), followed by Acurate valves (35%, n = 107) (Table 2). The mean age of the cohort was 76.8 ± 7.8 years, with no significant difference between the BEV and SEV groups (p = 0.2). The mean EUROSCORE II‐predicted mortality was 3.8 ± 4.2% with no differences between the groups (BEV: 4.1 ± 4.8% vs. SEV 3.6 ± 3.5%, p = 0.2). Male patients were more likely to receive a BEV (67% vs. 53% in the SEV group, p < 0.001), while body mass index did not differ significantly between groups (p = 0.6).

TABLE 2.

Procedural characteristics of the patients with bicuspid aortic stenosis undergoing TAVR.

Characteristic	BEV N = 274a	SEV N = 303a	p valueb	Missing
Access site			< 0.001	0
Transfemoral	261 (95%)	290 (96%)
Via subclavian artery	1 (0.4%)	13 (4.3%)
Transapical	8 (2.9%)	0 (0%)
Direct aortic access	4 (1.5%)	0 (0%)
TAVR urgency			0.13	2
Elective	226 (82%)	262 (87%)
Nonelective	48 (18%)	39 (13%)
Predilatation	108 (39%)	289 (95%)	< 0.001	0
Post‐dilatation	33 (12%)	146 (48%)	< 0.001	1
Prosthesis size (mm)			< 0.001	0
20	3 (1.1%)	0 (0%)
23	42 (15%)	13 (4.3%)
25	0 (0%)	40 (13%)
26	87 (32%)	28 (9.2%)
27	0 (0%)	61 (20%)
29	142 (52%)	72 (24%)
34	0 (0%)	89 (29%)
Valve type				0
Sapiens family	274 (100%)	Not applicable
Evolut family	Not applicable	188 (62%)
Acurate family	Not applicable	107 (35%)
Portico/Navitor family	Not applicable	8 (2.6%)
Contrast amount (mL)	60.72 (37.81)	74.10 (56.56)	< 0.001	0
Radiation dose (Gy)	3,137.52 (5,967.83)	2,448.39 (7,317.25)	0.2	10
Radiation time (minutes)	17.87 (12.25)	20.59 (12.11)	0.008	1

Abbreviations: BEV, ballon expandable valve; SEV, self‐expandable valve.

^a n (%); Mean (SD).

^b Pearson′s Chi‐squared test; Fisher′s exact test; Welch Two Sample t‐test.

Concerning coexisting conditions, there were no significant differences between the BEV and SEV groups in the incidence of diabetes mellitus (p = 0.2), chronic kidney disease (p = 0.7), peripheral vascular disease (p = 0.3), or chronic pulmonary disease (p = 0.3). However, hypertension was more prevalent in patients receiving a SEV compared to those receiving a BEV (73% vs. 65%, p = 0.048), while atrial fibrillation was more common among BEV recipients (34% vs. 25%, p = 0.02). No significant differences were observed between the groups regarding the incidence of myocardial infarction within the previous 3 months, prior percutaneous coronary intervention, previous cardiac surgery, or the presence of a pacemaker (Table 1).

The aortic valve area, mean aortic valve gradient, maximum aortic valve gradient, prevalence of moderate/severe aortic insufficiency, and prevalence of moderate/severe mitral insufficiency were comparable between the BEV and SEV groups. However, patients in the BEV group had a significantly higher prevalence of reduced ejection fraction, defined as EF ≤ 40% (23% vs. 16%, p < 0.001), as well as mildly reduced ejection fraction (EF 41%–49%; 18% vs. 9.6% for SEV). Additionally, they had a larger mean aortic annulus diameter (26.1 ± 2.6 mm for BEV vs. 25.4 ± 2.5 mm for SEV, p < 0.001).

3.2. Procedural Characteristics

The transfemoral approach was the preferred access site in both groups (95% in the BEV group vs. 96% in the SEV group). However, a higher proportion of patients in the SEV group underwent the procedure via the subclavian artery (4.3% vs. 0.4%). Transapical (2.9%) and direct aortic access (1.5%) were utilized exclusively in the BEV group (p < 0.001). Patients in the SEV group underwent predilatation and postdilatation more frequently than those in the BEV group (95% vs. 39% for predilatation and 48% vs. 12% for postdilatation, p < 0.001 for both). Patients receiving SEVs required higher contrast volumes (74.10 ± 56.56 mL vs. 60.72 ± 37.81 mL, p < 0.001) and experienced longer radiation times (20.59 ± 12.11 min vs. 17.87 ± 12.25 min, p < 0.001).

3.3. IPTW

IPTW achieved a good balance between the two groups, with SMDs below 0.10 across all covariates (Figure 2 and Table S3). However, the Kolmogorov‐Smirnov statistic exceeded 0.05 for the following covariates: hypertension, maximum aortic valve gradient, and annulus diameter. To further minimize confounding bias, these covariates were also included in the subsequent multivariable regression analysis.

FIGURE 2.

Standardized mean differences in the unadjusted and in the IPTW‐adjusted cohort for ballon versus self‐expanding valves (Panel A) and for Sapien versus Evolut valves (Panel B). [Color figure can be viewed at wileyonlinelibrary.com]

3.4. Endpoints

The median follow‐up time was 678 days (IQR 288–1208) for patients receiving a BEV and 675 days (IQR 296–1214) for those receiving a SEV. Periprocedural mortality was comparable between the two groups (1.5% for BEV vs. 2.0% for SEV; adjusted odds ratio [aOR]: 1.72, 95% CI 0.45–6.56; p = 0.4). All‐cause mortality for the follow‐up period was also comparable between the two groups with an adjusted hazard ratio of 0.8 (95% CI: 0.54–1.20, p = 0.3) (Table 3). Kaplan‐Meier survival curves up to 5 years (Figure 3) showed no significant differences in all‐cause mortality in either the unadjusted cohort (log‐rank p = 0.073) or the IPTW‐adjusted cohort (log‐rank p = 0.3). The patients receiving a SEV were 10% more likely to achieve technical success compared to those receiving a BEV (95% vs 89%, adjusted relative risk [aRR]: 1.10, 95% CI: 1.01–1.11, p = 0.03) (Table 3). Although exploratory, a higher percentage of patients in the BEV group experienced complications in the form of aortic and access site injuries requiring vascular intervention. Conversely, a numerically higher proportion of SEVs were associated with embolization into the aorta, necessitating the implantation of a second valve. Notably, no cases of annulus rupture were observed (Table S4). Device success was not significantly different between the two groups (aOR: 1.11, 95% CI 0.70–1.75; p = 0.7) (Table 3 and Table S5 for comparison of device success individual endpoints).

TABLE 3.

Outcomes in the unadjusted population and after adjustment with inverse probability of treatment weighting (IPTW).

Outcomec	BEV N = 274 1	SEV N = 303 1	Unadjusted OR/HR/estimatea (95% CI)	IPTW‐adjusted OR/HR/estimateb (95% CI)	p value for adjusted
Periprocedural mortality	4 (1.5%)	6 (2.0%)	1.36	1.72	0.4
			(0.39, 5.38)	(0.45, 6.56)
All‐cause mortality at median follow‐up of 675 days	64 (23%)	49 (16%)	0.71	0.8	0.3
(IQR 289–1208 days)			(0.49, 1.03)	(0.54, 1.20)
Technical success	245 (89%)	287 (95%)	2.12	2.13	0.03
			(1.14, 4.09)	(1.09, 4.18)
Device success	218	244	1.06	1.11	0.7
	(80%)	(81%)	(0.71, 1.60)	(0.70, 1.75)
Prosthesis‐patient mismatch	11 (4.0%)	2 (0.7%)	0.16	0.10	0.02
			(0.02, 0.60)	(0.02, 0.65)
More‐than‐mild PVL	4 (1.5%)	21 (6.9%)	5.03	7.47	< 0.001
			(1.88, 17.4)	(2.32, 24.1)
Permanent pacemaker need	28 (11%)	35 (12%)	1.10	1.11	0.7
			(0.65, 1.88)	(0.48, 1.72)
Stroke during index hospitalization	1 (0.4%)	2 (0.7%)	1.81	4.28	0.3
			(0.17, 39.2)	(0.31, 59.4)
Post‐TAVR mean aortic valve gradient	12.0	8.0	−4.1	−3.72	< 0.001
	(10.0, 16.0)	(6.0, 10.0)	(−5.2, −3.0)	(−4.85, −2.58)

Note: Results reported as number of events (%), OR/HR/estimate (95% CI) and p‐values. BEV serves as the reference group. The values in bold represent differences between groups with p < 0.05.

Abbreviations: BEV, Ballon‐expandable valves; CI, confidence interval; HR, Hazard ration; IPTW, inverse probability of treatment weighting; OR, odds ratio; PVL, Paravalvular leakage; SEV, self‐expanding valves; TAVI, transcatheter aortic valve implantation; VARC, Valve Academic Research Consortium.

^aGenerated with univariable logistic analysis, univariable linear regression or univariable Cox proportional hazards modeling.

^b Generated with IPTW‐adjusted logistic modeling, IPTW‐adjusted linear regression or IPTW‐adjusted Cox proportional hazards modeling.

^c Table S2 for outcome definitions.

FIGURE 3.

Kaplan‐Meier curves for the unadjusted (Panel A) and the IPTW‐adjusted cohort (Panel B). [Color figure can be viewed at wileyonlinelibrary.com]

SEVs were associated with a lower post‐TAVR mean aortic valve gradient, averaging 3.72 mmHg less than BEVs (95% CI: −4.58 to −2.58; p < 0.001) while patients receiving a SEV had lower incidence of severe PPM than those receiving a BEV (aOR: 0.10; 95% CI: 0.02–0.65; p = 0.02). However, SEV recipients had a more than sevenfold adjusted increase in risk of developing more‐than‐mild PVL compared to their BEV counterparts (6.9% vs. 1.5%; aOR: 7.47; 95% CI: 2.32–24.1; p < 0.001). Notably, the incidence of permanent pacemaker implantation post‐TAVR was similar between the two groups (12% for SEV vs. 11% for BEV; aOR: 1.11; 95% CI: 0.48–1.72; p = 0.7) as was the incidence of stroke during index hospitalization (0.7% for SEV vs. 0.4% for BEV; aOR: 4.28; 95% CI: 0.31–59.4; p = 0.3).

3.5. Sensitivity Analysis

3.5.1. Tipping Point Sensitivity Analysis

To negate the observed aOR of 2.13 for technical success, an unmeasured confounder would need to be present in 80% of patients receiving SEV but only in 40% of those receiving a BEV, with association with technical success of OR 3.1.

To fully explain away the observed aOR of 0.10 for severe PPM, an unmeasured confounder would need to be present in 5% in the SEV group and 95% in the BEV group with an OR of 19.66 between this confounder and PPM.

To nullify the aOR of 7.47 for PVL, an unmeasured confounder would need to be present in 95% in the SEV group and only 5% in the BEV group with an OR of 12.23 between the confounder and PVL.

To fully account for the observed adjusted difference of −3.72 mmHg in post‐TAVR mean aortic valve gradient, an unmeasured confounder would need to differ between the SEV and BEV groups by a scaled mean of 0.5 and have a strong inverse association with the outcome (effect size: −7.44 mmHg).

3.5.2. Subgroup Sensitivity Analysis

A sensitivity analysis comparing only the Sapien (BEV) and Evolut (SEV) valves yielded similar outcome trends to the overall BEV versus SEV comparison, except for the outcome of technical success. Specifically, there was no statistically significant difference in the incidence of technical success (94% Evolut vs. 89% Sapien; aOR: 1.83; 95% CI: 0.9–3.73; p = 0.09). Other outcomes were similar since Evolut patients had a lower risk of PPM (0.5% vs. 4.0% for BEV; aOR: 0.44; 95% CI: 0.21–0.93; p = 0.03), and, on average, 3.75 mmHg lower postprocedural mean aortic valve gradients (95% CI: −5.04 to −2.46; p < 0.001). However, Evolut recipients had a more than fivefold increased risk of developing more‐than‐mild PVL (5.9% vs. 1.5%; aOR: 5.27; 95% CI: 1.61–17.3; p = 0.006). No significant differences were observed in periprocedural mortality (p = 0.3), all‐cause mortality (p = 0.3), device success (p = 0.7), permanent pacemaker implantation (p = 0.5), or stroke during the index hospitalization (p = 0.6) (Table 4).

TABLE 4.

Sensitivity analysis.

Outcomeb	Sapien N = 274 1	Evolut N = 188 1	Unadjusted OR/HR/estimatea (95% CI)	IPTW‐adjusted OR/HR/estimate (95% CI)	p value for adjusted
Periprocedural mortality	4 (1.5%)	5 (2.7%)	1.84	1.95	0.3
			(0.48, 7.54)	(0.55, 6.84)
All‐cause mortality at median follow‐up of 661 days	64 (23%)	35 (19%)	0.88	0.81	0.3
(IQR 272–1167 days)			(0.58, 1.33)	(0.54, 1.20)
Technical success	245 (89%)	176 (94%)	1.74	1.83	0.09
			(0.88, 3.62)	(0.9, 3.73)
Device success	218 (80%)	152 (81%)	1.08	1.1	0.7
			(0.68, 1.74)	(0.66, 1.82)
Prosthesis‐patient mismatch	11 (4.0%)	1 (0.5%)	0.13	0.44	0.03
			(0.01, 0.67)	(0.21, 0.93)
More‐than‐mild PVL	4 (1.5%)	11 (5.9%)	4.19	5.27	0.006
			(1.41, 15.3)	(1.61, 17.3)
Permanent pacemaker need	28 (11%)	27 (15%)	1.38	1.31	0.5
			(0.78, 2.45)	(0.58, 2.05)
Stroke during index hospitalization	1 (0.4%)	1 (0.5%)	1.46	1.94	0.6
			(0.06, 37.4)	0.12, 32.6)
Post‐TAVR mean aortic valve gradient	12.0 (10.0, 16.0)	7.0 (6.0, 10.0)	−4.3	−3.75	< 0.001
			(−5.5, −3.1)	(−5.04, −2.46)

Note: Comparison of outcomes in the Sapien and Evolut families of transcatheter heart valves after IPTW adjustment. Results reported as number of events (%), OR/HR/estimate (95% CI) and p‐values. Sapiens family serves as the reference group. The values in bold represent differences between groups with p < 0.05.

Abbreviations: CI, confidence interval; HR, Hazard ration; IPTW, inverse probability of treatment weighting; OR, odds ratio; PVL, Paravalvular leakage; SEV, self‐expanding valves; TAVI, transcatheter aortic valve implantation; VARC, Valve Academic Research Consortium.

^aGenerated with univariable logistic analysis, univariable linear regression or univariable Cox proportional hazards modeling. †Generated with IPTW‐adjusted logistic modeling, IPTW‐adjusted linear regression or IPTW‐adjusted Cox proportional hazards modeling.

^b Table S2 for outcome definitions.

4. Discussion

This multicenter, observational study compared both clinical and hemodynamic outcomes between SEVs and BEVs in 577 patients with BAV stenosis undergoing TAVR. No significant differences were observed in periprocedural or all‐cause mortality between the two THV designs. While SEVs demonstrated slightly higher technical success rates, device success was comparable between SEVs and BEVs. A sensitivity analysis comparing Sapien to Evolut valves found no differences in either technical or device success. Overall, SEVs exhibited a more favorable hemodynamic profile, characterized by lower postprocedural gradients and a lower incidence of PPM. However, SEVs were associated with a higher risk of more‐than‐mild PVL. Stroke rates and the need for permanent pacemaker implantation were comparable between the two groups.

Mortality rates in our study are in line with other studies comparing BEV and SEV in patients with BAV stenosis [21, 22, 23, 24], supporting the viability of both valve types as options. Technical success rates were high, indicating low periprocedural complication rates and highlighting advancements in both THVs technology and procedural techniques. This is particularly important given the typically younger age of BAV patients. The lower technical success observed in the patients receiving a BEV may, though, be attributed to a higher incidence of periprocedural complications. In the sensitivity analysis, technical success was comparable between the Sapien and Evolut groups. This finding may be due to the smaller sample size, which could reduce statistical power and limit the ability to detect significant differences. Alternatively, this discrepancy may reflect true differences, with better technical success observed in the other two SEV platforms, potentially influencing the overall results in the broader SEV group. As such, further studies with larger sample sizes are needed comparing the individual valves.

Retrospective studies with earlier‐generation BEVs—such as the SAPIEN XT—reported an incidence of more‐than‐mild paravalvular leak (PVL) of nearly 15% in patients with BAVs [25]. However, with the advent of newer‐generation devices like the Sapien 3, the incidence of more‐than‐mild PVL in BAV has significantly improved, with rates now ranging between 0% and 5.2% [22, 23, 26, 27]. The incidence of PVL in SEV has been comparable higher in retrospective studies ranging from 3.4% to 11.3% [28, 29]. Even though patients receiving SEVs were more likely to undergo both pre‐dilation and post‐dilation of the aortic annulus—and the valves were more frequently oversized—the incidence of more‐than‐mild PVL remained higher compared to those treated with BEVs in our study. Our findings of increased risk for more‐than mild PVL with SEVs compared to BEVs in BAV stenosis patients align with previous studies [21, 22, 24]. This difference is likely multifactorial, stemming from the unique characteristics of bicuspid valves. For instance, in TAV stenosis, earlier studies have shown that the degree of native valve calcification is an independent predictor of PVL with SEVs (e.g., CoreValve) but not with BEVs (e.g., Sapien) [30, 31]. BAV anatomy is characterized by a higher volume of calcification compared to TAVs. Additionally, the distribution of calcification in BAV is distinct—it tends to be most prominent along the leaflet free edge and the non‐coronary cusp, often in an asymmetric pattern [32]. This unique calcification profile may prevent full and uniform expansion of SEVs, thus contributing to the increased risk of PVL. The elliptical shape of the BAV annulus may be another contributing factor since SEVs, with their lower radial force, may struggle to completely expand in certain areas, leading to gaps between the prosthesis and the native tissue.

While our study found a lower incidence of severe PPM in SEVs compared to BEVs, the long‐term impact of this difference remains an important clinical question. PPM has been associated with increased late mortality, impaired left ventricular remodeling, and reduced valve durability [33, 34, 35, 36], making it particularly relevant in younger BAV patients. SEVs, particularly those with a supra‐annular configuration, may generally offer superior hemodynamic performance and lower transvalvular gradients due to their structural design, which may explain the lower incidence of PPM observed in our study. These findings highlight the need for individualized valve selection in BAV patients, considering factors such as annular dimensions, calcification patterns, and long‐term hemodynamic performance.

Historically, both early‐generation and later‐generation SEVs, at least in tricuspid aortic valve morphologies, have been associated with higher PPI rates [37, 38] compared to BEVs. In our study, the incidence rates were comparable, and the use of SEVs was not associated with an increased risk of PPI. Mangieri et al. [22] and Deutsch et al. [23] similarly reported that, despite a higher frequency of valve post‐dilatation in patients receiving a SEV, there was no increased risk of PPI. Conversely, a recent study by Buono et al. [21] reported a lower incidence of PPI in BAV patients treated with BEVs compared to those treated with SEVs. This discrepancy may be attributed to differences in valve generations, as the SEV group in Buono et al.′s study included earlier‐generation devices (e.g., CoreValve), which are known to have higher PPI rates. Earlier studies comparing the need for PPI post‐TAVR in patients with BAV stenosis and TAV stenosis have found contradictory results. Studies using earlier‐generation THVs reported PPI rates as high as 25.5% for BEV and 26.9% for SEV [39, 40] while the prospective BIVOLUTX registry which employed the Evolut PRO device in patients with BAV reported PPI rates as high as 19% at 30 days and 25% at 1 year [29]. These rates are notably higher than those generally reported for TAV patients. The higher incidence of PPI in patients with BAV may stem from the asymmetric valve expansion caused by resistant calcified raphe and leaflet fusion. This asymmetry may lead to preferential expansion toward the non‐coronary cusp (in patients with right‐left fusion), leading to pressure on the atrioventricular (AV) node, increasing the risk of conduction disturbances. Consequently, BAV anatomy itself may inherently pose a higher risk for PPI due to these structural factors, potentially diminishing the additional influence of valve design on PPI rates.

Patients with BAV anatomy have traditionally undergone SAVR due to their younger age and typically low surgical risk. Additionally, concomitant aortopathy and coronary artery disease requiring intervention often necessitate surgical management. However, a significant proportion of BAV patients with isolated valvular heart disease may opt for TAVR as a less invasive alternative. Both BEVs and SEVs appear to be feasible and safe options for TAVR in these patients. However, the choice between BEV and SEV should be individualized based on patient‐specific factors. Key considerations include the need for future coronary artery access, the potential risk of PPM, particularly in patients with a small aortic annulus, and the anatomical suitability of each valve type for the patient′s specific BAV morphology. Weighing these factors carefully is essential for optimizing outcomes and ensuring long‐term procedural success. Nevertheless, the number of patients with BAV stenosis undergoing TAVR is expected to increase in the near future, highlighting the urgent need for randomized controlled trials comparing BEVs and SEVs.

5. Limitations

Detailed information on aortic valve characteristics, including Sievers classification and degree of calcification, was not available in the registry. With a median follow‐up of 675 days, our study does not capture long‐term outcomes such as structural valve deterioration or durability. This limitation highlights the need for future research with extended follow‐up to compare outcomes between BEV and SEV in this population. Despite robust adjustments residual confounding cannot be excluded given the observational design and absence of randomization. However, tipping point analyses suggest that any potential confounders would need to be highly imbalanced between the BEV and SEV groups and strongly associated with the outcomes to nullify our findings.

Conflicts of Interest

A.L has nothing to declare. J.S has nothing to declare. A.R has nothing to declare. H.H has nothing to declare. M.S has received fees for consultancy from Abbott Vascular, Medtronic, Anteris, W.L. Gore, Cardiomech and Smartcella outside of the present work. S.J has received grants and proctoring fees from Edwards and Medtronic. SK. has nothing to declare. K.S has nothing to declare. D.I has nothing to declare. S.V has nothing to declare. A.M has nothing to declare. O.A has received lecture fees from Medtronic and is a proctor for Abbott and Meril. P.P has received consulting fees from Abbott. T.R has received consulting fees from Cardirad Sweden AB, EPS Vascular AB and Boston Scientific.

Supporting information

supplementary_figure_1_missing_covariates.

Supplementary_figure_2.

Supplementary Table_1_definitions.

supplementary_table_2_missing_outcomes.

supplementary_table_3_balacing_variablesdocx.

supplementary_table_4_technical_success.

supplementary_table_5_device_success.