Impact of Anticoagulation on Valve Hemodynamics Following Transcatheter Valve-In-Valve Implantation

Abstract

Background

Transcatheter aortic valve implantation (TAVI) for prosthetic valve dysfunction is frequently associated with suboptimal echocardiographic outcomes. The effect of anticoagulation on preserving valve function is not well understood.

Methods

Patients who underwent the valve-in-valve TAVI for failed prosthetic aortic valves between 2015 and 2023 at a large academic medical center were retrospectively included. Data on anticoagulation use, along with clinical and echocardiographic outcomes, were collected. The end point of the study was the effect of anticoagulation use on the changes in mean gradient, peak velocity (V_max), and dimensionless index over time.

Results

One hundred thirteen patients (43 women) with a mean Society of Thoracic Surgeons score of 5.8 ± 5.5 were included in the study. Of these, 42% underwent valve fracture, and anticoagulation was used in 69.3% patients at discharge and 53.3% at 1-year follow-up. The mean gradient (β = 4.46, P < .001) and V_max (β = 0.292, P < .001) increased, whereas dimensionless index decreased over time in the overall cohort (β = –0.06, P = .001). However, there was a significant effect of anticoagulation use on changes in mean gradient (β = –4.36, P = .003) and V_max (β = –0.30, P = .003) but not on DI (β = 0.40, P = .104) at 1-year post-TAVI.

Conclusions

Anticoagulation postvalve-in-valve TAVI may lead to lower mean gradients and peak velocities at 1-year follow-up. More data are needed to understand its impact on nonflow-dependent indices and clinical outcomes.

Introduction

Bioprosthetic aortic valves now represent the most common form of valve replacement for aortic valve disease in both the United States and the Europe.^1,2 However, with the improvement in life expectancy and increasing use of bioprostheses at a younger age, many such patients require a second procedure for a degenerated aortic bioprostheses.

Transcatheter aortic valve implantation (TAVI) is now a well-established therapy for failed or degenerated prosthetic aortic valves—the so-called valve-in-valve (ViV) procedure—particularly in patients at high or prohibitive surgical risk.^3,4 However, despite favorable short and midterm clinical outcomes, ViV TAVI is often associated with suboptimal hemodynamic results compared to redo surgery, owing to the constraints of the previously implanted valve.⁵

Fracturing the surgical frame, either before or after transcatheter valve deployment, is one of the techniques developed to improve the hemodynamic performance of ViV TAVI.⁶ Despite these advancements, ViV TAVI is still frequently associated with a higher transvalvular gradient compared to native valve TAVI.⁷ Furthermore, many patients undergoing ViV TAVI are at high or prohibitive surgical risk and are not suitable candidates for a third valve procedure in the event of ViV failure. Therefore, preserving the valve function following ViV TAVI is of paramount importance in this population.

Abnormal flow patterns after ViV TAVI—due to neosinus formation, residual material from the previous valve frame, and relative patient-prosthesis mismatch—may contribute to leaflet thrombosis, thickening, and degeneration over time.⁸ Use of anticoagulation has been consistently shown to reduce the incidence of leaflet thrombosis after TAVI for native aortic valve stenosis. However, the routine use of novel oral anticoagulants after native valve TAVI is associated with increased bleeding and mortality, and is not recommended by professional society guidelines.³ However, short-term anticoagulation (3-6 months) with a vitamin K antagonist may be considered after surgical aortic valve replacement (SAVR) (class IIa) and TAVI (class IIb).³ The effect of anticoagulation on hemodynamic performance after ViV TAVI over time is largely unknown. Therefore, we aimed to study the effect of anticoagulation on hemodynamic performance after ViV TAVI over time using echocardiography.

Methods

Consecutive patients who underwent TAVI for a failed aortic prosthetic valve at the Robert Wood Johnson University Hospital (New Brunswick, New Jersey) between January 2015 and December 2023 were included in this study. All patients referred for symptomatic degeneration of prosthetic aortic valves are routinely evaluated by a cardiac surgeon and an interventional cardiologist at our institution. After reviewing patients’ comorbidities and imaging data, the heart team selects the most appropriate valve replacement modality for each patient.

Data were extracted from electronic medical records by trained medical professionals. Patients’ demographics, comorbidities, Society of Thoracic Surgeons scores, and size of failed prosthetic valve were recorded. Procedural details were extracted from operative notes and confirmed through review of fluoroscopic images by an interventional cardiologist. For the purposes of this study, the use of a noncompliant balloon larger than the internal diameter of the failed, fracturable surgical valve was defined as valve fracture. Use of warfarin or novel oral anticoagulants, regardless of dose, at discharge and follow-up was documented as anticoagulant use. An echocardiogram was routinely performed prior to discharge after TAVI in all patients. Subsequently, patients were encouraged to undergo clinical follow-up and echocardiogram at 1 month and 1 year post-TAVI. A board-certified echocardiographer, blinded to the anticoagulation status of the patients, reviewed the images and extracted the relevant data as per recommendations by Hahn et al.⁹ In the event that images were unavailable for review, the relevant data were extracted from the official echocardiogram report. Echocardiograms performed within 21 to 60 days and 9 to 15 months post-TAVI were considered acceptable to be inclusion as 30-day and 1-year follow-ups, respectively. The study’s protocol was approved by the Robert Wood Johnson Medical School Institutional Review Board.

Study end points

The primary end point was the effect of anticoagulation on the change in mean transaortic gradient over time. Secondary end points were changes in peak transaortic velocity, dimensionless index (DI) over time, and the incidence of moderate structural valve deterioration, defined as an increase of 10 mm Hg or more in the mean gradient, at 1 year between the anticoagulated and nonanticoagulated patients.

Statistical analysis

Continuous variables were presented as mean ± SD, and categorical variables as frequencies (%). Longitudinal data collected at discharge, 30 days, and 1-year follow-up post-TAVI were analyzed. Due to unequal time intervals (discharge, 30 days, and 1-year follow-up), time was used as a categorical variable. To account for individual-specific deviations in baseline values and changes over time, and the effects of time-invariant covariates, mixed-effects linear models were used to evaluate the interaction between anticoagulation use and time on the outcomes of interest. This model allowed for a random intercept and slope for each patient in addition to time-invariant covariates. Analyses were performed using Stata 17 software package (StataCorp LLC). A P value <.05 was considered significant.

Results

A total of 113 patients underwent ViV TAVI during this period and were included in the analysis. The mean age of patients was 77.7 years, and 38% were women. The mean Society of Thoracic Surgeons score was 5.8, and 53.9% suffered from atrial fibrillation at baseline. Only 7% ViV TAVI procedures were performed for failed transcatheter valves, and the rest were for failed surgical valves. Most of the failed surgical valves had a diameter of 21 mm or less (38.6%), followed by 22 to 23 mm (36.6%). The baseline patient characteristics are shown in Table 1. More than 90% of patients had TAVI performed via percutaneous transfemoral access using a balloon-expandable valve (Table 1). Valve fracture was performed in 43.7% of cases, exclusively after TAVI (Central Illustration).

Table 1.

The baseline characteristics of the valve-in-valve TAVI patients.

Characteristics	N = 113
Age, y	77.7 ± 9.5
STS score	5.8 ± 5.5
Women	38%
Hypertension	95.6%
Diabetes	38.9%
Prior coronary artery bypass	35.4%
Renal replacement therapy	22.1%
Atrial fibrillation/flutter	53.9%
Prior stroke	10.6%
Baseline pacemaker	13.5%
Previous valve type
Surgical	92.9%
Transcatheter	7.1%
Failure mechanism
Stenosis	66.4%
Regurgitation	15.9%
Mixed	17.7%
Surgical valve size
≤21 mm	38.2%
22-23 mm	37.3%
>23 mm	24.5%
Procedural details
Moderate sedation	56.6%
Transfemoral access	98%
Valve type used
Balloon expandable	90.3%
Self-expanding	9.7%
Valve fracture performed	43.7%
Anticoagulation use
Discharge	69.3%
30 d	72.7%
1 y	53.3%
In-hospital outcomes
Death	2.6%
Stroke	0
Pacemaker	3.1%

Values are mean ± SD or %.

STS, Society of Thoracic Surgeons; TAVI, transcatheter aortic valve implantation.

Central Illustration.

Impact of anticoagulation on valve hemodynamics after transcather valve-in-valve implantation. BEV, balloon-expandable valves.

Clinical outcomes

Three patients of 113 died (2.6%) before discharge. There were no postprocedure strokes. Three patients without a baseline pacemaker (3.1%) required pacemaker implantation after TAVI. An additional 2 and 7 patients died during the 30-day and 1-year follow-up periods, respectively. Within 1-year post-TAVI follow-up, 4 patients underwent balloon dilation of the aortic valve, and 2 patients underwent SAVR. Anticoagulants were prescribed to 69.3% of patients at discharge; however, 72.7% and 53.3% patients were on anticoagulants at 30 days and 1-year follow-up, respectively (Table 1).

Echocardiographic outcomes

Echocardiographic data were available for 110, 83, and 61 patients at discharge, 30 days, and 1-year follow-up, respectively. The means of peak aortic velocity, mean gradient, and DI at discharge, 30-day, and 1-year follow-up in patients with and without anticoagulation are shown in Table 2. For longitudinal analysis, patients on anticoagulation at 30 days and 1-year follow-up were considered as treated. The results of the mixed linear effects models are shown in Table 3 and Figure 1. The mean gradient and peak velocity increased from discharge to 1 year, whereas the DI decreased.

Table 2.

The average peak velocity, mean gradient, and dimensionless index at discharge, 30-day, and 1-year follow-up based on anticoagulation status.

	At discharge (n = 110)		30 d (n = 83)		1 y (n = 61)
	Anticoagulated	Control	Anticoagulated	Control	Anticoagulated	Control
Peak velocity, m/s	2.81 ± 0.06	2.71 ± 0.09	2.77 ± 0.06	2.79 ± 0.07	2.93 ± 0.08	3.10 ± 0.11
Mean gradient, mm Hg	18.25 ± 0.91	16.00 ± 1.13	18.22 ± 0.89	17.92 ± 1.12	19.82 ± 1.20	22.47 ± 1.67
Dimensionless index	0.37 ± 0.01	0.39 ± 0.01	0.36 ± 0.01	0.35 ± 0.02	0.33 ± 0.01	0.31 ± 0.01

Table 3.

Results of mixed linear models evaluating interaction between time and anticoagulation for mean gradient, peak velocity, and DI while adjusting for relevant covariates.

Variables	Mean gradient		Vmax		DI
	β ± SE	P value	β ± SE	P value	β ± SE	P value
Time since TAVI (at discharge)	1		1		1
30 d	1.304 ± 0.934	.163	0.081 ± 0.064	.205	–0.035 ± 0.016	.029
1 y	4.466 ± 1.187	<.001	0.292 ± 0.082	<.001	–0.067 ± 0.019	.001
Anticoagulation	2.163 ± 1.356	.111	0.178 ± 0.091	.051	–0.029 ± 0.018	.108
Anticoagulation (interaction with time)
At 30 d	–0.651 ± 1.233	.597	–0.096 ± 0.085	.260	0.040 ± 0.021	.064
At 1 y	–4.367 ± 1.479	.003	–0.300 ± 0.102	.003	0.040 ± 0.024	.104
Age at TAVI	–0.027 ± 0.072	.705	–0.003 ± 0.004	.472	0.001 ± 0.001	.721
Women	–1.746 ± 1.656	.292	–0.046 ± 0.110	.678	0.051 ± 0.019	.009
Failed valve’s diameter	–1.141 ± 0.337	.001	–0.074 ± 0.022	.001	0.010 ± 0.003	.008
Valve fracture performed	–1.551 ± 1.339	.247	–0.104 ± 0.089	.244	0.018 ± 0.015	.254
Ejection fraction	0.079 ± 0.042	.059	0.006 ± 0.002	.019	NA

DI, dimensionless index; TAVI, transcatheter aortic valve implantation.

Figure 1.

Primary and secondary study outcomes. (A-C) Margins plot for interaction between time and anticoagulation for (A) mean gradient, (B) peak velocity, and (C) dimensionless index. These models were adjusted for age, surgical size, surgical valve fracture, and ejection fraction (the dimensionless index model did not include ejection fraction). (D) Incidence of moderate structural valve deterioration at 1 year based on anticoagulation status (maroon, anticoagulated; blue, not anticoagulated).

However, there was a significant interaction between anticoagulation use and time. Patients on anticoagulation had significantly lower mean gradient (β = –4.367, P = .003) and peak velocity (β = –0.300, P = .003) at 1 year but not at 30 days. In this mixed-effects model, anticoagulation at 1 year was associated with a blunted increase in mean gradient and peak velocity compared with no anticoagulation in patients undergoing ViV TAVI. This association remained significant after adjusting for age at TAVI, sex, surgical valve size, ejection fraction, and surgical valve fracture at the index procedure, while the model also accounted for patient-level differences in baseline values and trajectories over time (random intercepts and slopes) (Table 3). Anticoagulation did not significantly interact with time on DI, although there was a trend toward higher DI with anticoagulation at 30 days (β = 0.040, P = .064) and 1 year (β = 0.040, P = .104). Margins plots derived from these regression models are shown in Figure 1A-C. The moderate structural valve deterioration at 1 year was significantly more common in the control group compared to THE anticoagulated group (28.57% vs 2.50%; P = .002), as shown in Figure 1D.

Discussion

In our single-center observational study, we found that the mean gradient and peak velocity increased, and DI decreased over time compared to the echocardiogram done at discharge after ViV TAVI. However, there was a significant interaction between anticoagulation use and time. Use of anticoagulation was associated with a smaller increase in mean gradient and peak velocity at 1 year. Furthermore, anticoagulation use was associated with less structural valve deterioration at 1 year. However, there was no significant effect of anticoagulation on DI over time (P = .06).

There is a steady increase in the use of TAVI for failed bioprosthetic aortic valves in the United States. ViV TAVI constituted only 1.1% of all TAVI cases in 2013 but increased to 4.25% in 2019, with a total of 10,012 cases by 2019.¹⁰ This trend is fueled by several observational studies and meta-analyses showing better short-term outcomes with ViV TAVI compared to redo SAVR.11, 12, 13 However, redo SAVR is generally associated with a lower transvalvular gradient, reduced patient-prosthesis mismatch, and possibly better mid- and long-term clinical outcomes.^5,11,14 Due to constraints of the surgical valve frame and frequent presence of inherent patient-prosthesis mismatch, ViV is associated with worse hemodynamic outcomes compared to native valve TAVI.⁷ Therefore, bioprosthetic valve fracture, a technique to disrupt the frame of the surgical valve to optimize the expansion of the transcatheter valve has been developed. It was used in about 21% of ViV TAVI procedures in the United States during 2020-2022; however, it was associated with higher in-hospital mortality and life-threatening bleeding.⁶ The risk of coronary occlusion, nonfracturable surgical valve frames, and the size of the left ventricular outflow tract and sino-tubular junction may be some of the reasons for not performing bioprosthetic valve fracture after ViV TAVI. Unlike prior reports, our study did not assess the immediate impact of valve fracture on postprocedural hemodynamics but instead examined longitudinal changes after discharge. Valve fracture was performed in 43.7% of patients; however, it was not associated with subsequent changes in mean gradient, peak velocity, or DI following TAVI (Tables 1 and 2).

We selected mean gradient, peak velocity, and DI as primary and secondary outcomes, as these parameters are less susceptible to measurement errors and image quality limitations. Furthermore, we accounted for time-invariant covariates like age, sex, failed bioprosthesis size, performance of valve fracture, and ejection fraction, while allowing for a random intercept (ie, baseline mean gradient or peak velocity) and a random slope (ie, changes in the outcome over time) for each individual patient. Several studies have reported hemodynamic outcomes after ViV TAVI and found a postprocedure mean gradient ranging from 17 to 19 mm Hg.^15,16 However, study results have been mixed on the changes in valve hemodynamics over time. The Placement of Aortic Transcatheter Valves (PARTNER) 2 and 3 ViV registry reported stable echocardiographic hemodynamics over time, whereas several other real-world studies reported an increase in transvalvular gradient over time.17, 18, 19 Both PARTNER 2 and 3 ViV registries excluded patients based on the size of the surgical prosthesis, as opposed to real-world registries, which included all eligible patients irrespective of the valve size. Similarly, 38.2% patients in our study had a surgical valve with a labeled diameter of 21 mm or less, and 17.6% of patients had a valve smaller than 21 mm. Wilbring et al¹⁷ reported outcomes of 77 consecutive patients who underwent ViV TAVI at the University Heart Center, Dresden, Germany. They found that the postprocedure mean gradient increased from 16.8 ± 7.1 to 19.9 ± 14.2 and 26.0 ± 12.2 mm Hg at the 1- and 3-year follow-up, respectively.

Current guidelines support the use of a vitamin K antagonist for 3 months in patients at low risk for bleeding after TAVI (class IIB).³ This recommendation is based on studies citing increased risk of restricted leaflet motion and leaflet thrombosis in the absence of anticoagulation,²⁰ resolution of valve thrombosis with anticoagulation,²¹ and significantly lower incidence of increase in mean gradient at 30 days and 1 year with anticoagulation compared to no anticoagulation on discharge after TAVI.²² However, routine oral anticoagulation after TAVI is not recommended after publication of the Global Study Comparing a Rivaroxaban-based Antithrombotic Strategy to an Antiplatelet-based Strategy after Transcatheter Aortic Valve implantation to Optimize Clinical Outcomes (GALILEO) trial,²³ which reported significantly higher major bleeding and deaths with use of low-dose rivaroxaban with aspirin compared to antiplatelet therapy alone. The Anticoagulation Versus Dual Antiplatelet Therapy for Prevention of Leaflet Thrombosis and Cerebral Embolization After Transcatheter Aortic Valve implantation study found no significant difference in leaflet thrombosis between edoxaban and dual antiplatelet therapy after successful native valve TAVI, although a trend favoring edoxaban was observed.²⁴ However, oral anticoagulation with novel oral anticoagulants compared to vitamin K antagonists in patients with indications for anticoagulation have shown comparable results and is supported by current guidelines.^4,25 Therefore, patients on long-term anticoagulation for other indications, like atrial fibrillation, provide an interesting opportunity to study the effect of anticoagulation on valve hemodynamics over time.²²

As opposed to a redo surgical valve replacement, where the previously placed valve is removed, during ViV TAVI, a transcatheter valve is implanted within a previously placed prosthesis. This may lead to unique flow characteristics due to the creation of a neosinus between TAVI valve leaflets and the surgical valve frame, and a residual anatomic sinus between the surgical valve and the aortic root.^26,27 In vitro modeling has shown significantly reduced flow and shear stress after ViV TAVI inside a surgical prosthesis, particularly in the noncoronary sinus, possibly contributing to leaflet thrombosis.²⁸ The RESOLVE registry (Assessment of Transcatheter and Surgical Aortic Bioprosthetic Valve Dysfunction With Multimodality Imaging and Its Treatment with Anticoagulation) identified overexpansion of balloon-expandable valves, deeper implantation depth of self-expanding valves, and low cardiac output as additional risk factors for leaflet thrombosis.²⁹ The multicenter Valve-in-Valve International Data registry reported an incidence of clinical valve thrombosis of 7.6% with 50% cases diagnosed within 3 months of the procedure³⁰ which is significantly higher than the TAVI for native aortic stenosis.³¹ Furthermore, the Valve-in-Valve International Data registry found that clinical valve thrombosis was significantly less common in patients on anticoagulation (1%) compared to those who were not (11.3%).

A few previous studies have evaluated the effect of anticoagulation on valve hemodynamics during follow-up. Del Trigo et al¹⁸ evaluated 1521 patients who underwent TAVI from a multicenter registry and found a mild but significant increase in the mean gradient over time. The lack of anticoagulation, a ViV TAVI procedure as opposed to native valve TAVI, higher body mass index, and use of a 23 mm TAVI valve were associated with a ≥10 mm increase in mean gradient during follow-up.¹⁸ Overtchouk et al³² evaluated the effect of anticoagulation on bioprosthetic valve dysfunction after TAVI, defined as a ≥10 mm increase or a new ≥20 mm mean gradient at follow-up, using data from the French TAVI registry. Anticoagulation, primarily prescribed for atrial fibrillation, was associated with a significantly lower risk of bioprosthetic valve dysfunction.³² Similarly, we found a significantly lower incidence of moderate structural valve deterioration at 1 year in patients who were anticoagulated compared to those who were not (Figure 1D).

Although these data are encouraging and support the use of anticoagulation in patients with suspected leaflet thrombosis or worsening gradient during follow-up,⁸ the routine use of anticoagulation after ViV TAVI still remains an unanswered question. Furthermore, ViV TAVI patients represent a heterogeneous group; more data are needed to identify specific subgroups—such as those with stented porcine valves,³⁰ small TAVI valve sizes,^32,33 nonfracturable surgical prosthesis with inherent patient-prosthesis mismatch, very high postprocedural gradient⁸—that might benefit from anticoagulation while carefully weighing the competing risk of bleeding.

Limitations

This was a single-center retrospective study; therefore, findings may not be reproducible at other centers. Although echocardiogram readers were blinded to the anticoagulation use during the data extraction, because of the nature of the study, observer bias cannot be completely excluded. Greater than 90% valves used in our study were balloon-expandable; therefore, the applicability of our results to self-expanding valve platforms may be limited. The majority of patients who received long-term anticoagulation in our study had atrial fibrillation. The presence of atrial fibrillation may impact the absolute value of mean gradient and peak velocity during echocardiographic assessment, representing a confounding by indication bias. However, we used the change in these parameters from baseline over time as our primary and secondary outcomes, using a mixed linear model allowing for random intercepts and slopes at the individual level rather than comparing 2 groups at different time points. Furthermore, the impact of an increase in echocardiogram-derived mean gradient and velocity on clinical outcomes after TAVI is not well-known. Particularly, studies have demonstrated significant discordance between invasive versus echocardiographic aortic valve gradient, in patients with ViV and balloon-expandable TAVI.34, 35, 36

Conclusion

In our single-center study, the use of anticoagulation was associated with lower transaortic mean gradient and peak velocity after ViV TAVI at 1-year follow-up. More data are needed to understand its impact on the need for repeat valve interventions and long-term clinical outcomes.

Declaration of competing interest

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.