Impact of transcatheter aortic valve replacement on aortic root remodeling: a pre- and post-procedure CTA comparative analysis

Abstract

Background

Transcatheter aortic valve replacement (TAVR) induces anatomical changes in the aortic root; however, the hemodynamic relevance of post-procedural geometric remodeling remains incompletely understood.

Methods

We retrospectively analyzed 117 patients with severe aortic stenosis who underwent transapical TAVR using the self-expanding Xcor system (Saint Medical Technology, Nanjing, China) and had paired pre-procedural and 1-year post-procedural computed tomography angiography (CTA). Changes in aortic root dimensions and annulus angle were quantified. Hemodynamic outcomes were assessed by 1-year echocardiography. A composite adverse hemodynamic outcome was defined as mean pressure gradient ≥ 20 mmHg or Doppler velocity index < 0.35.

Results

At 1 year, TAVR resulted in statistically significant but modest expansion of the sinus of Valsalva (+ 2.6 mm, + 6.8%) and sinotubular junction (+ 1.6 mm, + 5.0%), with only 10.3% of patients exceeding predefined dimensional thresholds. Root dimensional expansion was not associated with adverse hemodynamic outcomes. In contrast, greater annulus angle change and higher baseline annular calcium volume were independently associated with increased transvalvular gradients. Patients with both risk factors had significantly higher rates of adverse hemodynamic outcomes (26.3% vs. 4.3%, p = 0.008).

Conclusions

Transapical TAVR using the self-expanding Xcor system induces predominantly hemodynamically modest aortic root remodeling at 1 year. Root expansion appears to represent adaptive remodeling without adverse hemodynamic consequences, whereas an exploratory analysis suggests that excessive annulus angle change combined with severe calcification may be associated with suboptimal hemodynamic performance, though these findings require validation.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12872-026-05903-9.

Introduction

Transcatheter aortic valve replacement (TAVR) has revolutionized the treatment of severe aortic stenosis, offering a less invasive alternative to surgical aortic valve replacement (SAVR) [1, 2]. The procedure involves the implantation of a stented bioprosthetic valve, which exerts a continuous radial force on the surrounding aortic root structures. This interaction leads to a cascade of geometric and hemodynamic changes, collectively termed aortic root remodeling [3, 10, 32]. While the immediate hemodynamic benefits of TAVR are well-established, the hemodynamic significance—as distinct from statistical significance—of long-term anatomical changes remains inadequately characterized.

Landmark randomized trials including PARTNER 3 and Evolut Low Risk have demonstrated excellent outcomes with TAVR across the risk spectrum, extending to low-surgical-risk patients with follow-up now exceeding 5 years [4–7]. These trials have established TAVR as a viable alternative to surgery even in younger, healthier patients. However, as TAVR is increasingly utilized in populations with longer life expectancies, understanding the long-term anatomical consequences of prosthesis-root interaction becomes increasingly important [8, 9].

Previous studies using pre- and post-procedural computed tomography angiography (CTA) have documented dimensional changes in the aortic annulus and sinus of Valsalva (SoV) [10, 11]. However, several critical questions remain unanswered. First, do the observed dimensional changes exceed thresholds of hemodynamic importance, or do they represent statistically detectable but clinically inconsequential variations? Second, what is the mechanistic basis linking geometric changes to hemodynamic alterations? Third, can anatomical parameters be integrated into actionable risk stratification frameworks to guide clinical decision-making?

The aortic annulus angle, which describes the orientation of the valve relative to the left ventricular outflow tract (LVOT), is a critical parameter potentially altered by TAVR. Prior studies have demonstrated that prosthetic valve angulation relative to the flow axis can create asymmetric jet profiles and increase transvalvular gradients [12, 13]. Understanding the relationship between post-procedural anatomy and hemodynamic performance is essential for refining device design, improving patient selection, and optimizing implantation techniques.

This study leverages a unique dataset of 117 patients with paired pre-procedural and 1-year post-procedural CTA scans from a multicenter registry. Using dedicated cardiovascular analysis software (CVPilot), we aimed to comprehensively quantify the dimensional changes in the SoV, STJ, and proximal AAo after TAVR, evaluate the change in the aortic annulus angle, determine the impact of these anatomical remodeling parameters on 1-year hemodynamic outcomes as assessed by echocardiography, and explore potential associations between anatomical remodeling parameters and adverse hemodynamic outcomes.

Methods

Study population

We retrospectively analyzed data from a multicenter registry involving 11 high-volume TAVR centers in China. From this registry, we identified 117 patients who underwent TAVR with the Xcor system (Saint Medical Technology, Nanjing, China) between May 2022 and May 2023 and had both high-quality, contrast-enhanced pre-procedural and 1-year post-procedural CTA scans available for analysis. All patients had severe, symptomatic aortic stenosis and were deemed to be at high or prohibitive risk for SAVR by a multidisciplinary heart team. The registry enrolled a total of 412 patients who underwent transapical TAVR with the Xcor system during the study period.

Exclusion criteria included: (1) suboptimal CTA image quality precluding accurate measurements (defined as motion artifacts affecting > 20% of the aortic root or inadequate contrast opacification < 250 HU in the aortic root); (2) incomplete echocardiographic follow-up data; (3) valve-in-valve procedures; (4) concomitant aortic root or ascending aortic surgery; and (5) death or re-intervention within the 1-year follow-up period. From the broader multicenter registry of 412 consecutively enrolled TAVR patients, 156 patients who had received the Xcor system and had at least one follow-up CTA were initially screened. Of these, 39 were excluded (17 for suboptimal image quality, 12 for incomplete follow-up, 6 for death/re-intervention, and 4 for other reasons), yielding a final cohort of 117 patients. The Xcor transcatheter heart valve system is a self-expanding, nitinol-based prosthesis delivered exclusively via the transapical approach. The device features a unique design incorporating supporting arms that anchor to the native aortic leaflets, providing enhanced positional stability independent of radial force alone [31]. All procedures were performed under general anesthesia without cardiopulmonary bypass via a left anterolateral mini-thoracotomy.

The study was approved by the institutional review boards of all participating centers, and the requirement for patient consent was waived due to the retrospective nature of the analysis.

CTA acquisition protocol

All CTA scans were performed using ≥ 64-slice multidetector CT scanners following a standardized protocol consistent with expert consensus recommendations [14, 15]. The contrast protocol consisted of 70–90 mL iodinated contrast (350–400 mgI/mL) at 4–5 mL/s, with bolus tracking triggered at 180 HU in the descending aorta. Retrospective ECG gating with tube current modulation was employed. All measurements were performed at 30–40% R-R interval (systolic phase) to ensure consistency and optimal visualization of the aortic root [16]. Reconstruction was performed with 0.5–0.75 mm slice thickness and 0.3–0.5 mm increment. Post-procedural scans were performed at 12 ± 2 months after TAVR.

CTA analysis

All CTA analyses were performed using the CVPilot software (TAVIMercy Data Technology Ltd., Nanjing, China), a Class II AI-driven medical imaging platform certified by the National Medical Products Administration of China. This platform has been validated against conventional analysis systems with excellent segmentation accuracy and measurement agreement [35]. To ensure reproducibility, images were independently analyzed by two experienced cardiovascular imaging specialists (Reader A: 8 years experience, > 500 TAVR CTA analyses; Reader B: 6 years experience, > 350 TAVR CTA analyses), both blinded to clinical and echocardiographic data. In cases of disagreement exceeding predefined thresholds (> 2 mm for diameters, > 3° for angles), a third senior reader (12 years experience) adjudicated to reach consensus.

Inter-observer reliability was assessed in a random subset of 30 patients (25.6%) who were re-analyzed by both readers after a 4-week interval. Intraclass correlation coefficients (ICC) were calculated for all primary measurements.

The following measurements were performed on both pre- and post-procedural scans. For aortic root diameters, mean, maximum, and minimum diameters were measured at three distinct levels using semi-automated centerline reconstruction: the sinus of Valsalva (SoV) at the level of the largest sinus dimensions measured cusp-to-commissure and cusp-to-cusp, the sinotubular junction (STJ) at the narrowest point superior to the sinuses, and the ascending aorta (AAo) at a standardized level 4 cm above the aortic annulus plane [17]. The aortic annulus angle, defined as the angle between the plane of the aortic annulus and the central axis of the LVOT, was measured in the coronal oblique view perpendicular to the annular plane. On post-procedural scans, prosthesis-related measurements including implantation depth (distance from the annular plane to the ventricular edge of the prosthesis frame) and prosthesis tilt (angle between the prosthesis axis and the LVOT centerline) were also obtained [18].

Based on clinical reasoning and prior literature [14, 19], we prospectively predefined dimensional thresholds as SoV expansion ≥ 3 mm (exceeding twice the inter-observer measurement variability), STJ expansion ≥ 2 mm, AAo expansion ≥ 2 mm, and annulus angle change ≥ 5°. Remodeling was quantified as the absolute and percentage change between the 1-year post-procedural and pre-procedural measurements (Fig. 1).

Fig. 1.

Anatomical Measurement Protocol and Representative CTA Images. A Pre-procedural CTA imaging protocol demonstrating the standardized measurement planes. B Post-procedural CTA at 1-year follow-up showing the implanted transcatheter aortic valve prosthesis within the native aortic root. C Annulus angle measurement methodology, illustrating the angle between the aortic annular plane and the left ventricular outflow tract (LVOT) central axis. The typical pre-TAVR angle of approximately 50° becomes more acute post-TAVR (approximately 46°)

Echocardiographic assessment

All patients underwent comprehensive transthoracic echocardiography at 1-year follow-up. Hemodynamic performance of the bioprosthesis was assessed by measuring the peak and mean transvalvular pressure gradients, peak aortic velocity, effective orifice area (EOA) by the continuity equation, and Doppler velocity index (DVI) [20]. All echocardiographic data were analyzed at a central core lab by readers blinded to CTA findings.

The primary hemodynamic endpoint was a composite adverse hemodynamic outcome, defined as MPG ≥ 20 mmHg and/or DVI < 0.35 at 1-year follow-up. This composite endpoint was chosen to capture both elevated transvalvular gradients and patient-prosthesis mismatch (PPM), which are directly relevant to aortic root geometry and prosthesis-annulus interaction. DVI < 0.35 is an established marker of severe PPM and provides a size-independent assessment of prosthetic valve function [20, 21]. This endpoint definition differs from the traditional VARC-3 composite that includes paravalvular regurgitation; we elected to focus on gradient-related and flow-related parameters given the primary study aim of evaluating geometric remodeling effects on forward flow hemodynamics. Paravalvular regurgitation is primarily determined by pre-procedural calcium distribution and device sizing rather than post-procedural geometric adaptation, making it less mechanistically relevant to our research question [21]. We acknowledge that this endpoint captures valve hemodynamic performance rather than comprehensive clinical outcomes as defined by VARC-3 criteria, which additionally incorporate device success and patient-centered outcomes [21].

Statistical analysis

Continuous variables were expressed as mean ± standard deviation or median (interquartile range) as appropriate. Categorical variables were expressed as numbers and percentages. Normality was assessed using the Shapiro-Wilk test.All 117 patients had complete CTA and echocardiographic data at 1-year follow-up; no imputation was required.

Inter-observer agreement was evaluated using intraclass correlation coefficients (ICC) with 95% confidence intervals, calculated using a two-way random effects model for absolute agreement. Paired t-tests (or Wilcoxon signed-rank tests for non-normal distributions) were used to compare pre- and post-procedural measurements. Repeated measures analysis of variance (ANOVA) was used to assess the overall significance of changes across different aortic root levels. Bonferroni correction was applied for multiple pairwise comparisons, with adjusted p < 0.017 considered significant for the three-level comparison.The proportion of patients exceeding predefined dimensional thresholds was calculated with 95% confidence intervals.

Multiple linear regression analysis was performed to identify independent anatomical predictors of 1-year MPG. The model included change in annulus angle, baseline annular calcium volume, change in SoV/STJ/AAo diameters, prosthesis implantation depth, prosthesis tilt, bicuspid morphology, and baseline LVEF. Variables with p < 0.05 in univariate analysis were entered into the final multivariate model using backward elimination. Logistic regression was performed for the composite adverse hemodynamic outcome.Given the limited number of endpoint events, we restricted the multivariate model to a maximum of two predictor variables to avoid overfitting, following the principle of approximately 10 events per variable.

The study sample size (n = 117) was determined by availability of paired high-quality CTA scans in the registry. Post-hoc power analysis indicated 80% power to detect a correlation coefficient of r = 0.26 at α = 0.05 for continuous outcomes. Given the limited number of adverse events (n = 11), the multivariate analysis should be interpreted as hypothesis-generating, requiring external validation.

An exploratory outcome classification was developed based on independently associated factors and internally validated using bootstrap resampling (1000 iterations). Discrimination was assessed by C-statistic, and calibration was assessed by the Hosmer-Lemeshow test.

A two-sided p-value < 0.05 was considered statistically significant. All statistical analyses were performed using SPSS version 26.0 (IBM Corp, Armonk, NY) and R version 4.2.0 (R Foundation for Statistical Computing).

Results

Baseline characteristics

The study included 117 patients with a mean age of 70.8 ± 4.1 years, of whom 56 (47.9%) were male. Bicuspid aortic valve anatomy was present in 70 patients (59.8%). The mean STS score was 8.3 ± 3.5%. Comorbidities included coronary artery disease in 50 patients (42.7%), diabetes mellitus in 31 (26.5%), hypertension in 82 (70.1%), and chronic obstructive pulmonary disease in 41 (35.0%). Mean left ventricular ejection fraction was 58.7 ± 12.0%. Pre-procedural CTA measurements showed mean SoV diameter of 38.0 ± 4.4 mm, STJ diameter of 32.2 ± 3.9 mm, AAo diameter of 34.3 ± 4.1 mm, annulus angle of 49.5 ± 7.9°, and annular calcium volume of 304 ± 168 mm³. Baseline patient characteristics are detailed in Table 1.

Table 1.

Baseline Patient Characteristics

Characteristic	All Patients (n = 117)
Demographics
Age, years	70.8 ± 4.1
Male sex, n (%)	56 (47.9)
BMI, kg/m²	22.9 ± 3.2
Risk Assessment
STS score, %	8.3 ± 3.5
Comorbidities
Bicuspid aortic valve, n (%)	70 (59.8)
Coronary artery disease, n (%)	50 (42.7)
Diabetes mellitus, n (%)	31 (26.5)
Hypertension, n (%)	82 (70.1)
COPD, n (%)	41 (35.0)
Cardiac Function
LVEF, %	58.7 ± 12.0
Pre-procedural CTA Measurements
SoV diameter, mm	38.0 ± 4.4
STJ diameter, mm	32.2 ± 3.9
AAo diameter, mm	34.3 ± 4.1
Annulus angle, degrees	49.5 ± 7.9
Annular calcium volume, mm³	304 ± 168

BMI body mass index, STS Society of Thoracic Surgeons, COPD chronic obstructive pulmonary disease, LVEF left ventricular ejection fraction, SoV sinus of Valsalva, STJ sinotubular junction, AAo ascending aorta

Continuous variables expressed as mean ± SD; categorical variables as n (%)

Inter-observer reproducibility

CTA measurements demonstrated excellent inter-observer agreement (Supplementary Table S1). ICC values for primary measurements were 0.94 (95% CI: 0.89–0.97) for SoV diameter, 0.92 (95% CI: 0.86–0.96) for STJ diameter, 0.95 (95% CI: 0.91–0.98) for AAo diameter, and 0.91 (95% CI: 0.84–0.95) for annulus angle. Intra-observer reproducibility was similarly excellent with all ICC values exceeding 0.90.

Aortic root remodeling at 1 Year

Comparative analysis of pre- and post-procedural CTA scans revealed statistically significant expansion of the entire aortic root at 1-year follow-up (Table 2; Fig. 2A). The mean SoV diameter increased from 38.0 ± 4.4 mm to 40.6 ± 4.5 mm, representing an absolute change of + 2.6 ± 2.1 mm (+ 6.8%, p < 0.001). The STJ mean diameter increased from 32.2 ± 3.9 mm to 33.8 ± 4.4 mm (+ 1.6 ± 1.8 mm, + 5.0%, p < 0.001). A significant but smaller expansion was observed in the proximal ascending aorta, with the mean diameter increasing from 34.3 ± 4.1 mm to 34.9 ± 4.0 mm (+ 0.6 ± 1.4 mm, + 1.7%, p = 0.012 [p = 0.036 after Bonferroni correction, which did not meet the adjusted significance threshold of p < 0.017]). Individual patient trajectories demonstrated substantial heterogeneity in remodeling response (Fig. 2B), with most patients exhibiting minimal dimensional changes despite statistical significance at the group level.

Table 2.

Pre- and Post-Procedural CTA Measurements

Parameter	Pre-procedural	1-Year Post-procedural	Absolute Change	Percentage Change (%)	p-value	Adjusted p-value*
SoV diameter, mm	38.0 ± 4.4	40.6 ± 4.5	+ 2.6 ± 2.4	+ 6.8	< 0.001	< 0.001
STJ diameter, mm	32.2 ± 3.9	33.8 ± 4.4	+ 1.6 ± 1.8	+ 5.0	< 0.001	< 0.001
AAo diameter, mm	34.3 ± 4.1	34.9 ± 4.0	+ 0.6 ± 1.4	+ 1.7	0.012	0.036
Annulus angle, degrees	49.5 ± 7.9	46.4 ± 7.9	-3.1 ± 3.8	N/A	< 0.001	-

SoV sinus of Valsalva, STJ sinotubular junction, AAo ascending aorta

Values expressed as mean ± SD. p-values from paired t-tests

*Bonferroni-adjusted p-values for multiple comparisons across three aortic root levels; significance threshold p < 0.017. AAo expansion did not reach statistical significance after correction

Fig. 2.

Aortic Root Dimensional Changes at 1-Year Follow-up. A Mean dimensional changes showing statistically significant expansion of the sinus of Valsalva (SoV, +2.6 ± 2.4 mm, +6.8%), sinotubular junction (STJ, +1.6 ± 1.8 mm, +5.0%), and ascending aorta (AAo, +0.6 ± 1.4 mm, +1.7%). Error bars represent standard deviation. B Individual patient trajectories demonstrating heterogeneity in remodeling response. Color coding indicates magnitude of change: gray (< 2 mm), light orange (2-3 mm), and dark orange (> 3 mm). The thick black line represents the cohort mean. C Distribution of patients relative to the predefined dimensional threshold (3 mm for SoV, indicated by dashed line). Only 10.3% of patients (n = 12) exceeded this threshold, with the majority showing changes within the expected measurement variability

Repeated measures ANOVA confirmed a significant overall effect of TAVR on aortic root dimensions (F = 28.7, p < 0.001), with the magnitude of expansion varying significantly across levels (p for interaction < 0.01).

Despite achieving statistical significance, only a minority of patients demonstrated threshold-exceeding dimensional changes (Table 3; Fig. 2C). Only 12 patients (10.3%, 95% CI: 5.4–17.2%) exhibited SoV expansion ≥ 3 mm, 18 patients (15.4%, 95% CI: 9.4–23.2%) showed STJ expansion ≥ 2 mm, and 8 patients (6.8%, 95% CI: 3.0–13.0%) demonstrated AAo expansion ≥ 2 mm. Among all 117 patients, 89 (76.1%) showed dimensional changes within clinically expected ranges at all three measured locations (SoV, STJ, and AAo).

Table 3.

Threshold Analysis: Proportion of Patients Exceeding Predefined Dimensional Thresholds

Parameter	Threshold	n (%)	95% CI
SoV expansion	≥ 3 mm	12 (10.3)	5.4–17.2%
STJ expansion	≥ 2 mm	18 (15.4)	9.4–23.2%
AAo expansion	≥ 2 mm	8 (6.8)	3.0–13.0%
Annulus angle change	≥ 5°	29 (24.8)	17.3–33.6%
Any dimensional change (SoV, STJ, or AAo)	Any above	28 (23.9)	16.5–32.7%
Any change including angular	Any above	38 (32.5)	24.1–41.8%

CI confidence interval

Thresholds were prospectively defined based on hemodynamic reasoning and measurement variability (SoV threshold set at twice the inter-observer SD)

Changes in aortic annulus angle

The aortic annulus angle underwent a significant change post-TAVR, becoming more acute from a mean of 49.5 ± 7.9° pre-procedure to 46.4 ± 7.9° at 1 year (mean change − 3.1 ± 3.8°, p < 0.001) (Fig. 3A). Twenty-nine patients (24.8%, 95% CI: 17.3–33.6%) experienced angle changes exceeding 5°, which we designated as the “high angular change” subgroup. When considering any threshold-exceeding change (dimensional or angular), 38 patients (32.5%) met at least one threshold criterion (Fig. 3B).

Fig. 3.

Annulus Angle Changes and Clinical Significance. A Distribution of annulus angle changes at 1 year. The histogram shows the observed changes (gray bars) overlaid with a normal distribution curve (blue solid line) and theoretical normal distribution (gray dashed line). The mean change was -3.1 ± 3.8°. Red dashed line indicates the threshold for clinically significant change (> 5° decrease), which occurred in 24.8% of patients (n = 29). Blue dashed line shows extreme increases (> 5°), observed in 4.3% of patients (n = 5). B Proportion of patients exceeding predefined dimensional thresholds across different parameters. Error bars represent 95% confidence intervals. Angular changes were more prevalent than dimensional changes, with 32.5% of patients meeting at least one threshold-exceeding criterion

Hemodynamic outcomes at 1 Year

At 1-year follow-up, the overall cohort demonstrated favorable hemodynamic performance (Table 4). The mean transvalvular pressure gradient was 10.4 ± 4.4 mmHg, with a peak velocity of 2.1 ± 0.4 m/s and effective orifice area of 1.8 ± 0.4 cm². The mean DVI was 0.48 ± 0.12; 8 patients (6.8%) had DVI < 0.35, indicating severe patient-prosthesis mismatch. The composite adverse hemodynamic outcome (MPG ≥ 20 mmHg or DVI < 0.35) occurred in 11 patients (9.4%). Of these, 5 patients met both criteria, 3 met only the MPG criterion, and 3 met only the DVI criterion.

Table 4.

Hemodynamic Outcomes at 1-Year Follow-up

Parameter	Value
Transvalvular Gradients
Mean pressure gradient, mmHg	10.4 ± 4.4
Peak pressure gradient, mmHg	19.8 ± 7.2
Flow Parameters
Peak velocity, m/s	2.1 ± 0.4
Effective orifice area, cm²	1.8 ± 0.4
Doppler velocity index	0.48 ± 0.12
Patient-Prosthesis Mismatch
DVI < 0.35 (severe PPM), n (%)	8 (6.8)
DVI 0.35–0.45 (moderate PPM), n (%)	21 (17.9)
DVI > 0.45 (no PPM), n (%)	88 (75.2)
Composite Outcome
Adverse hemodynamic outcome, n (%)	11 (9.4)
— MPG ≥ 20 mmHg only	3 (2.6)
— DVI < 0.35 only	3 (2.6)
— Both criteria met	5 (4.3)

DVI Doppler velocity index, PPM patient-prosthesis mismatch, MPG mean pressure gradient

Values expressed as mean ± SD or n (%)

Adverse hemodynamic outcome defined as MPG ≥ 20 mmHg and/or DVI < 0.35 (severe patient-prosthesis mismatch)

Correlation structure and association between remodeling and hemodynamics

To comprehensively characterize the relationships between remodeling parameters and hemodynamic outcomes, we first examined pairwise correlations among all measured variables (Fig. 4A). The correlation matrix revealed weak interdependence among dimensional remodeling parameters, with correlation coefficients between SoV, STJ, and AAo changes ranging from r=-0.03 to r = 0.18. Notably, dimensional expansion parameters showed minimal univariate correlation with mean pressure gradient (SoV change: r = 0.11; STJ change: r=-0.01; AAo change: r = 0.04), while annulus angle change also demonstrated weak univariate correlation (r = 0.06). These minimal correlations among predictor variables justified their inclusion as independent variables in subsequent multivariate modeling.

Fig. 4.

Correlation Structure and Multivariate Predictors of Hemodynamic Outcomes. A Correlation matrix of remodeling parameters and mean pressure gradient. Heat map colors indicate correlation strength (dark red: positive correlation; dark blue: negative correlation). Weak interdependence among dimensional parameters (r = -0.03 to 0.18) justified their inclusion as independent variables in multivariate modeling. B Forest plot showing standardized β coefficients from multivariate linear regression for 1-year mean pressure gradient. Significant predictors (red bars, p < 0.05) were annulus angle change (β = 0.24, p = 0.012) and baseline calcium volume (β = 0.19, p = 0.042). Dimensional expansion parameters (SoV, STJ) showed no significant association. Dashed vertical line represents the null hypothesis (β = 0)

In multiple linear regression analysis for 1-year MPG (Table 5; Fig. 4B), two factors remained independently significant after adjustment for prosthesis position and inclusion of all remodeling parameters. Greater change in annulus angle was associated with higher gradient (β = 0.24, p = 0.012, 95% CI: 0.05–0.43) (Fig. 5A), as was higher baseline annular calcium volume (β = 0.19, p = 0.042, 95% CI: 0.01–0.37) (Fig. 5B). The synergistic effect of these two risk factors is demonstrated in Fig. 5C.

Table 5.

Multiple Linear Regression Analysis for 1-Year Mean Pressure Gradient

Variable	Univariate β (95% CI)	p-value	Multivariate β (95% CI)	p-value
Change in annulus angle, per degree	0.29 (0.11–0.47)	0.002	0.24 (0.05–0.43)	0.012
Baseline calcium volume, per 100 mm³	0.23 (0.05–0.41)	0.014	0.19 (0.01–0.37)	0.042
Change in SoV diameter, per mm	0.08 (-0.10 to 0.26)	0.384	—	—
Change in STJ diameter, per mm	0.06 (-0.12 to 0.24)	0.512	—	—
Change in AAo diameter, per mm	0.04 (-0.14 to 0.22)	0.663	—	—
Prosthesis implantation depth, per mm	0.11 (-0.07 to 0.29)	0.231	—	—
Prosthesis tilt, per degree	0.09 (-0.09 to 0.27)	0.324	—	—
Bicuspid valve morphology	0.07 (-0.11 to 0.25)	0.445	—	—
Age, per year	0.05 (-0.13 to 0.23)	0.584	—	—
Male sex	0.04 (-0.14 to 0.22)	0.663	—	—

Model adjusted for prosthesis position (implantation depth and tilt). Final model R² = 0.18, F = 12.4, p < 0.001

The final multivariate model was restricted to two predictor variables given the limited number of endpoint events (n = 11), following the principle of approximately 10 events per variable to avoid overfitting

Fig. 5.

Association Between Anatomical Parameters and Hemodynamic Performance. A Scatter plot showing the relationship between annulus angle change and 1-year mean pressure gradient (β = 0.24, p = 0.012). Greater angular change was associated with higher gradients. Gray shaded area represents 95% confidence interval. Horizontal dashed line indicates the adverse outcome threshold (MPG ≥ 20 mmHg). B Relationship between baseline annular calcium volume and mean pressure gradient (β = 0.19, p = 0.042; slope = 0.029 mmHg/mm³). Data points are color-coded by risk category: green (low risk, 0 points), yellow (intermediate risk, 1 point), and red (high risk, 2 points). Vertical dashed line indicates the high-risk calcium threshold (400 mm³). C Box plots comparing mean pressure gradient across risk groups defined by combined anatomical features. The high-risk group (n = 19) demonstrated significantly higher gradients compared to low-risk patients (p = 0.008). Boxes represent interquartile ranges; horizontal lines indicate medians. D Scatter plot demonstrating the absence of association between SoV diameter change and mean pressure gradient (β = 0.08, p = 0.384, R² < 0.01). Vertical dashed line indicates the predefined expansion threshold (3 mm). This negative finding suggests that root expansion represents adaptive remodeling without adverse hemodynamic consequences

Importantly, none of the dimensional expansion parameters (SoV, STJ, AAo) were significantly associated with MPG in univariate or multivariate analysis (all p > 0.2) (Fig. 5D). As shown in the multivariate predictor analysis (Fig. 4B), after adjustment for confounders, STJ change (β = 0.06, p = 0.345) and SoV change (β = 0.08, p = 0.234) demonstrated no independent predictive value for hemodynamic outcomes. This null finding persisted after sensitivity analyses excluding outliers and adjusting for prosthesis type and size. Even patients with the most pronounced SoV expansion (top quartile, mean + 5.8 mm) did not demonstrate higher MPG compared to those with minimal expansion (bottom quartile, mean + 0.3 mm): 10.2 ± 4.1 vs. 10.8 ± 4.6 mmHg (p = 0.52). This suggests that geometric enlargement of the aortic root, within the ranges observed, may represent adaptive remodeling that accommodates the prosthesis without compromising hemodynamic function.

Exploratory risk associations

Based on the identified independently associated factors, we developed a simplified retrospective outcome classification approach (Table 6). Patients were assigned 1 point for annulus angle change > 5° and 1 point for baseline annular calcium volume > 400 mm³. This resulted in three risk categories: low risk (0 points, n = 69, 59.0%), intermediate risk (1 point, n = 29, 24.8%), and high risk (2 points, n = 19, 16.2%).

Table 6.

Exploratory Risk Associations for Adverse Hemodynamic Outcomes

Component	Description	Value	n (%)	Adverse Outcome, n (%)	MPG, mmHg
Risk Factors and Scoring
Annulus angle change > 5°	1 point		29 (24.8)
Baseline annular calcium > 400 mm³	1 point		38 (32.5)
Risk Categories
Low risk	0 points		69 (59.0)	3 (4.3)	9.2 ± 3.8
Intermediate risk	1 point		29 (24.8)	3 (10.3)	10.8 ± 4.2
High risk	2 points		19 (16.2)	5 (26.3)*	13.6 ± 5.1*
Model Performance Metrics
C-statistic		0.72 (0.58–0.86)
Bootstrap-corrected C-statistic		0.69 (0.54–0.84)
Hosmer-Lemeshow p-value		0.64
Sensitivity (high-risk)		45.5% (16.7–76.6%)
Specificity (high-risk)		86.8% (79.0-92.4%)

The wide confidence interval for C-statistic (0.58–0.86) reflects limited statistical power due to the small number of endpoint events. External validation is required before clinical implementation

*p < 0.01 vs. low risk group

Patients with both high-risk features had significantly elevated rates of adverse hemodynamic outcomes: 26.3% in the high-risk group compared with 10.3% in intermediate-risk and 4.3% in low-risk patients (p = 0.008) (Fig. 6A). The model demonstrated moderate discriminative ability for adverse hemodynamic outcomes with a C-statistic of 0.72 (95% CI: 0.58–0.86) (Fig. 6B). Calibration was adequate (Hosmer-Lemeshow p = 0.64) (Fig. 6C), and bootstrap internal validation showed minimal optimism (corrected C-statistic 0.69).

Fig. 6.

Exploratory Risk Associations for Adverse Hemodynamic Outcomes. A Adverse outcome rates by risk category. Patients with both high-risk features (annulus angle change > 5° and baseline calcium > 400 mm³) had significantly elevated adverse hemodynamic outcome rates (26.3% vs 4.3% in low-risk group, p = 0.008). Error bars represent 95% confidence intervals. Numbers in parentheses indicate patients with adverse outcomes out of total in each category. B Receiver operating characteristic (ROC) curve demonstrating model discrimination. The C-statistic was 0.72 (95% CI: 0.58-0.86), with bootstrap-corrected C-statistic of 0.69 (95% CI: 0.54-0.84). The red dot indicates the optimal threshold for identifying high-risk patients (2 points). C Calibration plot comparing predicted versus observed probabilities of adverse outcomes. Points represent the three risk categories with error bars showing 95% confidence intervals. The black diagonal line represents perfect calibration. Hosmer-Lemeshow test indicated adequate calibration (p = 0.64)

Discussion

This multicenter study, using paired pre- and 1-year post-procedural CTA analysis with rigorous measurement protocols, provides a comprehensive assessment of aortic root remodeling after TAVR. Our findings advance the field in four important ways: we demonstrate that while TAVR induces statistically significant root expansion, only a minority of patients exhibit changes exceeding predefined dimensional thresholds; we report the important negative finding that root dimensional expansion does not adversely affect hemodynamics; we identify the combination of excessive angular change and severe baseline calcification as key factors associated with suboptimal outcomes; and we present an exploratory outcome association analysis.

Statistical vs. hemodynamic significance of aortic root expansion

A central finding of our study is the discordance between statistical significance and hemodynamic relevance of root expansion. While mean increases in SoV (+ 2.6 mm), STJ (+ 1.6 mm), and AAo (+ 0.6 mm) diameters achieved statistical significance (p < 0.001 to p = 0.012), only 10–15% of patients exceeded predefined dimensional thresholds. Notably, after Bonferroni correction for multiple comparisons, the AAo expansion did not reach the adjusted significance threshold, further emphasizing the modest nature of these changes. This distinction is critical for hemodynamic interpretation. Our findings are consistent with prior reports on aortic root geometric changes following TAVR. Ng et al. [32] reported similar magnitudes of SoV expansion following CoreValve implantation using both echocardiography and CT. Madukauwa-David et al. [33] characterized aortic root geometry changes with both balloon-expandable and self-expanding devices, reporting that root geometry adapts to accommodate the prosthesis frame. Notably, our Xcor transapical cohort demonstrated SoV expansion (+ 6.8%) comparable to the self-expanding devices studied by these authors, despite the distinct delivery mechanism and anchoring system. The observed expansions fall within the measurement variability of CTA and are substantially smaller than changes typically associated with hemodynamic consequences such as coronary obstruction or sinus sequestration [19, 22].

Furthermore, when contextualized against natural variation and measurement precision, these changes may represent the upper boundary of normal physiological adaptation rather than pathological remodeling. The ICC values of 0.91–0.95 in our study indicate that the standard error of measurement for diameter assessments is approximately 1.0–1.5 mm, suggesting that a substantial proportion of the observed “expansion” may reflect measurement variability rather than true anatomical change.

Root expansion as adaptive remodeling: the importance of a negative finding

Perhaps our most clinically relevant finding is that root dimensional expansion—contrary to potential concerns—was not associated with adverse hemodynamic outcomes. This negative result held across all sensitivity analyses and subgroups. We propose that this reflects adaptive remodeling: the compliant aortic root accommodates the prosthesis frame through gradual expansion, thereby preserving optimal flow geometry and preventing the excessive radial compression that might otherwise elevate gradients.

This interpretation aligns with the concept of benign remodeling described in prior TAVR imaging studies, where modest post-procedural root geometric changes were not associated with adverse outcomes [10, 32, 33]. Within the ranges observed in our cohort and the limitations of a single-platform study, root expansion at 1-year follow-up was not associated with adverse hemodynamic outcomes. However, whether this finding extends to other device platforms, longer follow-up periods, or patients with different baseline anatomies requires further investigation.

Annulus angle change and hemodynamics: mechanistic considerations

In contrast to root expansion, the change in aortic annulus angle emerged as a significant predictor of higher residual gradients. This finding has plausible mechanistic underpinnings. Previous studies have demonstrated that prosthetic valve angulation relative to the flow axis creates asymmetric jet profiles, increases turbulent kinetic energy, and elevates energy loss across the valve [12, 13, 23]. The TAVR prosthesis, being more rigid than the native annulus, may “straighten” the LVOT-aortic root transition during deployment, particularly when the native anatomy features significant angulation.

Our finding that the combination of high angular change (> 5°) and severe calcification (> 400 mm³) identifies patients at particularly elevated risk (26.3% adverse hemodynamic vs. 4.3% in low-risk group) suggests these factors act synergistically—calcification may prevent optimal prosthesis seating and exacerbate angular distortion [24, 25]. This is consistent with prior reports demonstrating the impact of landing zone calcification on procedural outcomes [26].

We acknowledge that our study provides correlational rather than causal evidence for the link between angular change and hemodynamic outcomes. Direct flow assessment using 4D flow MRI would be required to confirm the hypothesized mechanism of turbulence-mediated gradient elevation. We did not perform computational fluid dynamics simulations, and our mechanistic interpretation relies on extrapolation from prior computational studies performed in different patient populations and with different prosthesis types.

Transapical approach and the Xcor system

The exclusive use of the transapical approach in our cohort warrants specific discussion. The Xcor system, with its supporting arm design, is delivered antegrade via left anterolateral mini-thoracotomy. This approach provides direct and coaxial access to the aortic valve, potentially allowing more precise deployment compared to retrograde transfemoral approaches, particularly in patients with horizontal aortas or challenging vascular anatomy [31]. The transapical route may also influence the pattern of radial force distribution on the aortic root compared to transfemoral devices, which could partly explain the remodeling patterns observed. Future comparative studies between transapical and transfemoral platforms are needed to clarify whether the access route independently influences aortic root remodeling.

Bicuspid aortic valve considerations

The high prevalence of bicuspid aortic valve anatomy (59.8%) in our cohort is consistent with the demographics of TAVR candidates in China and warrants specific discussion. Recent evidence from Ahn et al. [34] has demonstrated favorable outcomes of TAVR in bicuspid aortic valve stenosis, and a comprehensive state-of-the-art review by Barbanti et al. [37] has further summarized the advancements and remaining challenges of TAVR in bicuspid aortic valve disease, supporting the expanding applicability of transcatheter approaches in this population. The Xcor transapical system, with its supporting arm design providing enhanced positional stability, may be particularly suited for bicuspid anatomies where asymmetric leaflet calcification and elliptical annular geometry can complicate conventional device positioning.

Left ventricular remodeling and aortic root dynamics

Aortic root remodeling after TAVR should be considered within the broader context of cardiac structural adaptation. Left ventricular hypertrophy, which develops in response to the chronic pressure overload of aortic stenosis, undergoes reverse remodeling following successful valve replacement. Rozenbaum et al. [36] demonstrated that preprocedural LVH and its geometric patterns significantly impact mortality following TAVR, with eccentric hypertrophy conferring worse outcomes. The interplay between LV reverse remodeling, changes in LV outflow dynamics, and the aortic root geometric changes documented in our study represents a complex but important axis. It is plausible that LV mass regression following TAVR alters outflow tract geometry and flow profiles, which in turn could influence the hemodynamic significance of aortic root dimensional changes. Our study did not include serial LV mass measurements, and future studies integrating both aortic root and LV remodeling parameters would provide a more comprehensive understanding of post-TAVR structural adaptation.

Endpoint selection considerations

We elected to define our composite adverse hemodynamic outcome using MPG and DVI rather than paravalvular regurgitation (PVR). This decision was based on the primary study aim of evaluating how geometric remodeling affects forward flow hemodynamics. This focused endpoint allows direct assessment of the hypothesized mechanism linking anatomical changes to flow obstruction, avoiding confounding by factors unrelated to root geometry such as sealing performance. PVR data for the cohort are provided in Supplementary Table S2. DVI < 0.35, indicating severe patient-prosthesis mismatch, provides a size-independent assessment of effective prosthetic orifice relative to patient body size and is mechanistically linked to annular geometry and prosthesis expansion [20, 21]. Future studies examining the relationship between root remodeling and regurgitant outcomes would complement our findings.

Clinical implications and exploratory risk associations

Our findings have several practical implications for TAVR planning and follow-up. Regarding pre-procedural planning, patients with severely angulated LVOT-aorta anatomy (angle > 55°) and heavy annular calcification (> 400 mm³) should be counseled about modestly elevated risk of suboptimal hemodynamic outcomes. Although our study was limited to the transapical self-expanding Xcor platform, the mechanistic findings suggest that devices with greater conformability to native anatomy may offer advantages in high-risk patients, a hypothesis requiring prospective validation [27, 28]. Operators should be aware that aggressive positioning in angulated roots might exacerbate angular distortion, and careful attention to coaxial alignment during deployment may be particularly important in these cases [18, 29].

Regarding post-procedural surveillance, our data do not support the need for routine CTA surveillance to monitor root expansion, given its lack of clinical consequence. Patients identified as high-risk pre-procedurally may benefit from closer hemodynamic surveillance, though the clinical impact of the modest gradient elevations observed requires longer-term outcome data to fully characterize.

Limitations

This study has several important limitations. First, we analyzed only 1-year post-procedural CTA data; the temporal evolution of remodeling—whether changes occur acutely post-procedure, progressively over time, or reach a plateau—cannot be determined without early (30-day) CTA data. Second, our analysis was restricted to the transapical self-expanding Xcor system, and results may not be generalizable to balloon-expandable, mechanically-expanded, or transfemoral self-expanding platforms, which exert different radial forces and interact differently with the aortic root [27, 30]. Third, the retrospective design introduces potential selection and information biases, and the requirement for paired high-quality CTA scans may have selected for patients with better overall clinical status. Fourth, we inferred hemodynamic mechanisms from anatomical changes and echocardiographic gradients; 4D flow MRI would provide direct assessment of flow patterns and turbulence. Fifth, our exploratory outcome classification underwent only internal validation and requires external validation before clinical application. Sixth, the limited number of endpoint events (n = 11) constrains the statistical power for multivariate modeling; our findings should be considered hypothesis-generating rather than confirmatory. Seventh, the model’s C-statistic confidence interval (0.58–0.86) includes values near 0.5, indicating uncertainty in discriminative ability that warrants cautious interpretation. Finally, despite inclusion of prosthesis depth and tilt in our models, other factors potentially affecting gradients such as subclinical leaflet thrombosis and dynamic LVOT obstruction were not systematically assessed [21].

Conclusions

In this multicenter study with paired pre- and 1-year post-procedural CTA analysis, transapical TAVR using the self-expanding Xcor system induced statistically significant but predominantly hemodynamically modest expansion of the aortic root. Root dimensional expansion was not associated with adverse hemodynamic outcomes in this cohort, suggesting it may represent adaptive remodeling. In contrast, greater change in the aortic annulus angle, particularly when combined with severe baseline calcification, was associated with higher residual transvalvular gradients. An exploratory analysis suggests that the combination of these parameters may identify patients at risk of suboptimal hemodynamic performance, though these findings require validation in larger cohorts. However, given the limited endpoint events, single device platform, and moderate follow-up duration, these findings should be interpreted as hypothesis-generating. External validation in larger, multi-platform cohorts is warranted before any clinical application. The high proportion of bicuspid anatomy (59.8%) provides preliminary data on aortic root dynamics following transapical TAVR in this underrepresented population.

Abbreviations

AAo

Ascending aorta

AVA

Aortic valve area

AVR

Aortic valve replacement

BAV

Bicuspid aortic valve

BMI

Body mass index

BSA

Body surface area

CFD

Computational fluid dynamics

CI

Confidence interval

CTA

Computed tomography angiography

DVI

Doppler velocity index

EOA

Effective orifice area

HR

Hazard ratio

IQR

Interquartile range

LVOT

Left ventricular outflow tract

LVEF

Left ventricular ejection fraction

MDCT

Multidetector computed tomography

MPG

Mean pressure gradient

OR

Odds ratio

PPM

Prosthesis–patient mismatch

PVL

Paravalvular leak

PVR

Paravalvular regurgitation

ROC

Receiver operating characteristic

SAVR

Surgical aortic valve replacement

SoV

Sinus of Valsalva

STJ

Sinotubular junction

STS

Society of Thoracic Surgeons

TAVR

Transcatheter aortic valve replacement

TEE

Transesophageal echocardiography

VARC

Valve Academic Research Consortium

Authors’ contributions

Conceptualization: ZJ, YX, JLMethodology: ZJ, DL, HFData acquisition: DL, HF, HG, PW, ML, CL, ZF, SWFormal analysis: DL, HFSupervision: ZJ, YX, JLWriting – original draft: DLWriting – review & editing: All authorsDL, HF, HG and PW contribute equally to this work.All authors read and approved the final manuscript.

Funding

This work was supported by the Chongqing Natural Science Foundation (CSTB2022NSCQ-LZX0076) and the Chongqing Talent Project (CQYC20220303476).

Data availability

The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study protocol was approved by the Ethics Committee of West China Hospital (Approval No. HX-2022-17). This study was conducted in accordance with the principles of the Declaration of Helsinki. Due to the retrospective nature of the study and the use of anonymized data, the requirement for written informed consent was waived.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.