Mixed Valvular Disease Following Transcatheter Aortic Valve Replacement: Quantification and Systematic Differentiation Using Clinical Measurements and Image-Based Patient-Specific In Silico Modeling.

Background Mixed valvular disease (MVD), mitral regurgitation (MR) from pre-existing disease in conjunction with paravalvular leak (PVL) following transcatheter aortic valve replacement (TAVR), is one of the most important stimuli for left ventricle (LV) dysfunction, associated with cardiac mortality. Despite the prevalence of MVD, the quantitative understanding of the interplay between pre-existing MVD, PVL, LV, and post-TAVR recovery is meager. Methods and Results We quantified the effects of MVD on valvular-ventricular hemodynamics using an image-based patient-specific computational framework in 72 MVD patients. Doppler pressure was reduced by TAVR (mean, 77%; N=72; P<0.05), but it was not always accompanied by improvements in LV workload. TAVR had no effect on LV workload in 22 patients, and LV workload post-TAVR significantly rose in 32 other patients. TAVR reduced LV workload in only 18 patients (25%). PVL significantly alters LV flow and increases shear stress on transcatheter aortic valve leaflets. It interacts with mitral inflow and elevates shear stresses on mitral valve and is one of the main contributors in worsening of MR post-TAVR. MR worsened in 32 patients post-TAVR and did not improve in 18 other patients. Conclusions PVL limits the benefit of TAVR by increasing LV load and worsening of MR and heart failure. Post-TAVR, most MVD patients (75% of N=72; P<0.05) showed no improvements or even worsening of LV workload, whereas the majority of patients with PVL, but without that pre-existing MR condition (60% of N=48; P<0.05), showed improvements in LV workload. MR and its exacerbation by PVL may hinder the success of TAVR.

Clinical Perspective

What Is New?

Valvular disease is a complex disease that also depends on the dictates of the left ventricle and the arterial system.

We developed a computational mechanics framework based on and correlated with clinically measured hemodynamic metrics and imaging in patients to noninvasively quantify the effect of: (1) mixed valvular disease on the left ventricle workload (global hemodynamics) and (2) mixed valvular disease on the flow structures (local hemodynamics) in the left ventricle and investigated the correlation of hemodynamics parameters with clinical metrics in both pre‐ and postintervention states.

What Are the Clinical Implications?

Paravalvular leak limits the benefit of transcatheter aortic valve replacement: It may increase left ventricle load, may potentially worsen mitral regurgitation, and worsens heart failure. Mitral regurgitation and its possible exacerbation by paravalvular leak will likely play an important hindering role in the success of transcatheter aortic valve replacement.

This represents an important finding, especially considering the expansion of transcatheter aortic valve replacement candidacy to lower‐risk and/or young patients.

Our findings suggest that beyond standard clinical indices for hemodynamic evaluation of the valvular disease, valvular and ventricular hemodynamics and their interactions should be quantified and considered for better management of patients with aortic valve disease.

Introduction

Transcatheter aortic valve replacement (TAVR) is an emerging minimally invasive intervention for patients with aortic stenosis (AS) across a broad risk spectrum.1 TAVR is increasingly used in lower‐risk (moderate valvular disease and/or young) patients as well. Although TAVR has provided positive outcomes and has remarkably reduced the mortality rate, there are risks associated with TAVR procedures. Many patients experience significant improvements after TAVR intervention, but in many others, the situation worsens and pre‐existing valvular diseases transform to even more‐extensive disease (eg worsened mitral regurgitation [MR] and heart failure).2, 3, 4

Despite recent improvements in the design of transcatheter heart valves and implantation techniques, paravalvular leak (PVL; leak around the prosthesis) remains a major complication and an independent predictor of short‐ and long‐term mortality.5, 6 As Pibarot7 noted, “Paravalvular regurgitation is considered the main Achilles’ heel of transcatheter aortic valve replacement.” Mild PVL occurs in 20% to 80% of patients, whereas moderate and severe PVL occur in 5% to 22% of them.7, 8, 9

Mixed valvular disease (MVD), mitral regurgitation (MR) from pre‐existing disease in conjunction with PVL following TAVR, is one of the most important stimuli for left ventricle (LV) dysfunction resulting in congestive heart failure, associated with worsening cardiac mortality.2, 7, 9, 10, 11, 12, 13 As indications and use of TAVR expand, we must advance our understanding of its interactions with diseases of other valves and the ventricular state.2, 7, 10, 13, 14, 15 However, despite the prevalence of MVD, the quantitative understanding of the interplay between pre‐existing valvular pathologies, PVL, LV, and post‐TAVR recovery is meager. This study was aimed to elucidate the relationship between MR, LV function, and TAVR.

“Cardiology is flow.”16 The main functions of the cardiovascular system are to transport, control, and maintain blood flow in the entire body. Abnormal fluid dynamics greatly alter this tranquil picture, leading to initiation and progression of disease17. These abnormalities are often manifested by disturbed flow, altered biomechanical forces, and, in some cases, an increase in heart workload. The hypothesis is increasingly appreciated that valvular disease is a complex disease that also depends on the dictates of the LV and arterial system.18, 19, 20, 21 There has thus been an emerging call by many for quantitative investigations of hemodynamics that take the interactive coupling of the valve, ventricle, and arterial system into account.18, 19, 20, 21

In this study, we developed a Doppler‐based, patient‐specific, lumped‐parameter modeling framework that takes interactions of the aortic valve, LV, and arterial system into account to investigate MVD and estimate LV workload noninvasively. We showed that effective quantification of MVD hinges on quantification of the load it imposes on the LV. We quantified the effect of MVD on LV workload (global hemodynamics) and investigated the correlation of LV workload with the metrics currently used in clinical practice in 72 MVD patients in both pre‐ and post‐TAVR states. In addition to global hemodynamics, we provide a mechanistic fundamental understanding about the effect of MVD on the 3‐dimensional flow structures in the LV and LV outflow tract (local hemodynamics). For local hemodynamics analysis, we developed a computational fluid mechanics and lumped‐parameter modeling framework based on and correlated with clinically measured hemodynamic metrics and clinical images in both pre‐ and post‐TAVR states.

Methods

Data Availability

The data and the code that support the findings of this study are available from the corresponding author upon request.

Study Population

We retrospectively and randomly selected 205 patients with severe AS who underwent TAVR from anonymized databases between 2013 and 2018 from the following institutions: St. Paul's Hospital (Vancouver, British Columbia, Canada; N=40); Massachusetts General Hospital (Boston, MA; N=40)22; and Hospital Universitario Marques de Valdecilla (IDIVAL, Spain; N=125). Selections were done by operators blinded to the objectives and contents of this study at each institution. Of these 205 patients, 72 had MVD (MR from pre‐existing disease and PVL following TAVR). The protocol was reviewed and approved by the ethics committees of the institutions. Suitability and eligibility for receiving TAVR were determined by the local clinical team. Data were acquired at 2 time points: preprocedure and 90‐day postprocedure. Valve type and size were planned before the procedure by the local clinical team according to preprocedural echocardiographic, tomographic, and angiographic parameters. Procedural techniques were at the discretion of the local senior interventional cardiologists.

Doppler echocardiography

Doppler echocardiography data included raw images, and reports were collected preprocedure and at 90‐day postprocedure. Echocardiograms and reports were reviewed and analyzed in a blinded fashion by 3 senior cardiologists using OsiriX imaging software (OsiriX version 8.0.2; Pixmeo, Bernex, Switzerland). The following metrics were measured: valve area, valve velocity, pressure gradient, stroke volume, cardiac output, and left ventricular outflow tract diameter, as recommended by the American Society of Echocardiography.

Data acquisition

Demographic and procedural data were collected from the TAVR database and patient medical records. Clinical outcome was evaluated using the New York Heart Association functional class and medical records, evaluated preprocedure and 90‐day post‐TAVR.

Statistical analysis

All results are expressed as mean±SD. Normal distribution was assessed with the Shapiro–Wilk test. Statistical analyses were performed using SigmaStat software (Version 3.1; Systat Software, Inc, San Jose, CA). The paired Student t test was used to detect any significant hemodynamic difference between pre‐ and post‐TAVR conditions.

Numerical study

Global hemodynamics (LV workload)

We developed a patient‐specific, lumped‐parameter model that considers interactions of the aortic valve, LV, and arterial system to estimate LV workload noninvasively (Figure 1; Tables 1 and 2) in both pre‐ and postintervention conditions. The model used a limited number of input parameters, all of which can be reliably obtained using Doppler echocardiography and a sphygmomanometer. Doppler echocardiography‐based parameters (eg, stroke volume, heart rate, ejection time, ascending aorta area, aortic valve effective orifice area, and aortic regurgitation effective orifice area) were measured in the parasternal long axis, parasternal short axis, and apical 2‐, 4‐, and 5‐chamber views of the heart.22 Other input parameters of the model were systolic and diastolic blood pressures measured using a sphygmomanometer. Note that the proposed method does not need catheter data for estimating LV workload. The model and submodels have already been used and validated against in vivo cardiac catheterization (N=118) and in vivo magnetic resonance imaging data (N=57).22, 23, 24, 25, 26 LV workload (global hemodynamics) was calculated in all 205 patients in both pre‐ and post‐TAVR.

Figure 1

Left ventricular function and hemodynamics. (A) Data acquisition. We developed a computational mechanics framework based on noninvasive clinically measured hemodynamic metrics (brachial blood pressure and Doppler echocardiography measurements) and computed tomography imaging to estimate local and global hemodynamics. (B) Schematic diagram of lumped parameter model. Inputs of the lumped model were all obtained from Doppler echocardiography measurements. This model includes several submodels: left ventricle, aortic valve, aortic regurgitation, and systemic circulation. (C) Schematic diagram of 3D flow model. Inputs of the 3D flow model were obtained from CT and Doppler echocardiography. We used CT images from patients to reconstruct 3D geometries of the LV and valves. These 3D reconstructions were used for investigating hemodynamic using computational fluid dynamics. Moreover, imposing correct boundary conditions to the flow model is critical because the local flow dynamics is influenced by down‐ and upstream conditions. These data were obtained from a lumped parameter model (part B). Cao indicates aortic compliance; CSAC, systemic arteries and veins compliance; CT, computed tomography; 3D, 3‐dimensional; ELV, left ventricle elastance; Lav, aortic valve inductance; LPVL, paravalvular leak inductance; LV, left ventricle; Rao, aortic resistance; Rav, aortic valve resistance; Rpda, proximal descending aorta resistance; RPVL, paravalvular leak resistance; RSA, systemic arteries resistance; RSV, systemic vein resistance; Rub, upper body resistance.

Table 1

Baseline Clinical and Echocardiographic Characteristics

	AS Patients (n=72, Mean±SD)
Patient description
Mean age, y	74.40±6.54
Female sex	(48%)
Mean weight, kg	71.71±13.92
Mean height, cm	161.86±10.13
Body surface area, m2	1.77±0.16
Body mass index, kg/m2	32.38±23.20
EuroScore II	7.04±7.68
STS mortality rate	6.62±5.33
Arterial hemodynamics
Systolic arterial pressure, mm Hg	139.0±22.5
Diastolic arterial pressure, mm Hg	79.0±11.7
Aortic valve hemodynamics
Stenotic aortic valve effective orifice area, cm2	0.58±0.16
Stenotic aortic valve type	Tricuspid: 70; bicuspid: 2
Maximum aortic valve pressure gradient, mm Hg	84.50±21.32
Mean aortic valve pressure gradient, mm Hg	51.52±13.60
Left ventricle hemodynamics
Ejection fraction, %	53.5±12.7
Stroke volume index, mL/m2	45.7±11.5
Heart rate, bpm	71.0±11.5
Associated cardiovascular lesions
Previous percutaneous coronary intervention	33%
Previous coronary artery bypass grafting	34%
Previous myocardial infarction	21%
Previous stroke	2%
Atrial fibrillation	29%
Cerebrovascular accident	9%
Peripheral vascular disease	29%
Hypertension	74%

AS indicates aortic stenosis; STS, Society of Thoracic Surgeons.

Table 2

Summarized Cardiovascular Parameters Used in the Lumped Parameter Modeling to Simulate All Cases

Description	Abbreviation	Value
COA and valve parameters
Effective orifice area	EOA	From echocardiography data
Energy loss coefficient	ELCo	(EOA)AA−EOA From echocardiography data
Variable resistance	Rav and RPVL	ρ2E_LCo^2Q
Inductance	Lav and LPVL	2πρE_LCo
Systematic circulation parameters
Aortic resistance	Rao	0.05 mm Hg·s·mL−1
Aortic compliance	Cao	Initial value: 0.5 mL/mm Hg Adjust for each degree of hypertension (proximal COA compliance)
Systemic vein resistance	RSV	0.05 mm Hg·s·mL−1
Systemic arteries and veins compliance	CSAC	Initial value: 2 mL/mm Hg Adjust for each degree of hypertension (systemic compliance)
Systemic arteries resistance (including arteries, arterioles and capillaries)	RSA	0.8 mm Hg·s·mL−1 Adjust according to the calculated total systemic resistance
Proximal descending aorta resistance	Rpda	0.05 mm Hg·s·mL−1
Upper body resistance	Rub	Adjusted to have 15% of total flow rate in healthy case
Output condition
Central venous pressure	PCV0	4 mm Hg
Input condition
Mitral valve mean flow rate	Qmv	From echocardiography data
Other
Constant blood density		1050 kg/m3
Heart rate	HR	From echocardiography data
Duration of cardiac cycle	T	From echocardiography data

COA indicates coractation of the aorta.

Local hemodynamics (blood flow dynamics)

We developed a fluid‐solid interaction and lumped parameter modeling framework to calculate 3‐dimensional blood flow dynamics in the LV (Figure 1; Tables 1 and 2). Because computed tomography images have higher resolution than Doppler echocardiography data, they were used for 3‐dimensional reconstruction of the LV for fluid‐solid interaction calculations. In addition, as described above, the lumped parameter model in this framework used few input parameters, all of which can be measured using Doppler echocardiography and a sphygmomanometer. Because of the massive computational load, the blood flow inside the LV (local hemodynamics) was computed and analyzed in 28 of 72 patients with MVD.

Please refer to Data S1 for the details related to the numerical study.

Results

Clinical Measure of Hemodynamics: Doppler Echocardiography Pressure Gradients

Clinical assessment of AS for management and intervention decisions is currently performed based on the symptoms and hemodynamics metrics that focus only on the aortic valve. Our Doppler echocardiography data showed (Figure 2) that the transvalvular pressure gradient in all patients with MVD (MR and PVL; Table 1) was universally significantly reduced by TAVR. Mean and maximum Doppler pressure gradients were reduced by 77% and 49%, respectively (N=72; P<0.05).

Figure 2

Changes in clinical Doppler echocardiography measurements in patients with MVD between baseline and 90 days post‐TAVR (N=72). Mean Doppler gradient measured was reduced 77% by TAVR.

Hemodynamics: Left Ventricular Workload and Fluid Dynamics

LV workload (global hemodynamics)

Despite the universal reduction in the transvalvular pressure gradient (N=72; Figure 2), TAVR reduced the LV workload in only 18 of the 72 MVD patients (25%), as Figure 3A shows. In 22 of the MVD patients, LV workload was not significantly reduced (<5% reduction) post‐TAVR, and in the other 32 patients, LV workload increased post‐TAVR. Furthermore, reductions in transvalvular pressure gradient were not always accompanied by improvements in clinical metrics such as New York Heart Association heart failure classification (Figure 3B) and ejection fraction (Figure 3C). Pre‐TAVR, untreated AS increased the burden on the LV attributable to the augmented flow resistance, which caused an LV pressure overload. Although the flow resistance and consequently the LV pressure decreased greatly post‐TAVR, the LV workload did not improve because PVL and MR contributed to a switch from a ventricular pressure overload to a ventricular volume overload. Figure 4 shows examples of LV pressure, volume, and workload in 4 patients who received TAVR: LV workload improved remarkably in Non‐MVD Sample#1 (38% reduction), slightly improved in MVD Sample#2 (12% reduction), whereas it did not improve in MVD Samples#1 and 3.

Figure 3

Changes in clinical assessments of LV and LV workloads in patients with MVD between baseline and 90 days post‐TAVR (N=72). (A) LV workload. (B) Heart failure classification. (C) Ejection fraction. LV indicates left ventricle; MVD, mixed valvular disease; NYHA, New York heart association; TAVR, transcatheter aortic valve replacement.

Figure 4

Examples of LV workloads in patients between baseline and 90 days post‐TAVR. (A) MVD Sample#1: pre‐TAVR: EF: 65%, brachial pressures: 65 to 130 mm Hg, aortic valve EOA: 0.66 cm2, no AR, no MR, LV stroke volume: 72 mL; post‐TAVR: EF: 60%, brachial pressures: 60 to 130 mm Hg, aortic valve EOA: 1.45 cm2, mild PVL, mild MR, LV stroke volume: 80 mL. LV workload did not improve by TAVR. (B) Non‐MVD Sample#1: pre‐TAVR: EF: 58%, brachial pressures: 56 to 86 mm Hg, aortic valve EOA: 0.6 cm2, mild‐moderate AR, mild MR, LV stroke volume: 76 mL; post‐TAVR: EF: 75%, brachial pressures: 70 to 123 mm Hg, aortic valve EOA: 1.1 cm2, no PVL, no MR, LV stroke volume: 60 mL. LV workload improved remarkably (38% reduction) by TAVR. (C) MVD Sample#2: pre‐TAVR: EF: 55%, brachial pressures: 74 to 180 mm Hg, aortic valve EOA: 1.95 cm2, mild‐moderate AR, mild‐moderate MR, LV stroke volume: 87 mL; post‐TAVR: EF: 65%, brachial pressures: 40 to 141 mm Hg, aortic valve EOA: 1.55 cm2, mild PVL, mild‐moderate MR, LV stroke volume: 103 mL. LV workload slightly improved (12% reduction) by TAVR. (D) MVD Sample#3: pre‐TAVR: EF: 65%, brachial pressures: 75 to 141 mm Hg, aortic valve EOA: 1.4 cm2, no AR, moderate MR, LV stroke volume: 85 mL; post‐TAVR: EF: 65%, brachial pressures: 60 to 123 mm Hg, aortic valve EOA: 2.3 cm2, mild‐moderate PVL, moderate‐severe MR, LV stroke volume: 100 mL. LV workload did not improve by TAVR. AR indicates aortic regurgitation; EF, ejection fraction; EOA, effective orifice area; LV, left ventricle; MR, mitral regurgitation; MVD, mixed valvular disease; PVL, paravalvular leakage; TAVR, transcatheter aortic valve replacement.

LV fluid dynamics (local hemodynamics)

PVL following TAVR substantially alters vortical structure in the left ventricular outflow tract and LV, creating disturbed flow (Figures 5, 6 through 7). The jets emerging from PVL orifices diverge within the LV. This unfavorable flow condition leads to high shear stresses (Figures 5, 6 through 7) on transcatheter aortic valve and mitral valve leaflets (Figures 5, 6 through 7; eg, peak time‐averaged wall shear stress in patient #28 with MVD: on transcatheter aortic valve leaflets=14.4 Pa; on mitral valve leaflets=15.5 Pa; peak time‐averaged wall shear stress in the same patient without PVL: on transcatheter aortic valve leaflets=0.8 Pa; on mitral valve leaflets=5.7 Pa). Moreover, PVL, characterized by multiple jets, enters the LV chamber, directly interacts with mitral inflow, hinders formation of normal fluid vortical structures in the LV, and may worsen the MR post‐TAVR (eg, Figures 5, 6 through 7). We observed similar flow characteristics in other patients with mild, moderate, and severe PVL. Interestingly, worsening of the average MR status post‐TAVR (measured by clinical Doppler echocardiography) confirmed our above findings about the local hemodynamics (Figure 8A and 8B). We found in 72 MVD patients: (1) MR status was individually improved in only 21 patients (29% of MVD patients); on average: pre‐TAVR, 2.4±0.5; post‐TAVR, 1.2±0.45 (Figure 8C). The average of the LV workload was also improved in this patient population (pre‐TAVR, 1.99±0.55 J; post‐TAVR, 1.37±0.29 J; 31% average decrease); (2) individual MR status became worse in 32 other MVD patients post‐TAVR; on average: pre‐TAVR, 1.15±0.38; post‐TAVR, 2.25±0.40 (Figure 8D). The average of the LV workload was not improved in this patient population as well (pre‐TAVR, 1.37±0.50 J; post‐TAVR, 1.6±0.53 J; +17% average increase); and (3) individual MR status remained unchanged in 18 other MVD patients post‐TAVR; on average: pre‐TAVR, 1.66±0.48; post‐TAVR, 1.66±0.48 (Figure 8E). The average of the LV workload was also not improved significantly in these patients (pre‐TAVR, 1.45±0.51 J; post‐TAVR, 1.31±0.36 J; 10% average decrease).

Figure 5

Flow modeling in an LV with moderate‐to‐severe PVL in patient #28. (A) Time‐averaged wall shear stress (TAWSS) during diastole. Total shear stress exerted on the wall throughout was evaluated using the TAWSS, which was obtained as . Here, T and τ are the period and instantaneous wall shear stress, respectively. (B) Velocity magnitude at mid diastole in different planes passing PVL. In patient #28, PVL, characterized by 2 jets, enters the LV chamber, directly interacts with the mitral inflow, and may worsen the MR post‐TAVR (MR status from Doppler echocardiography measurements: pre‐TAVR, mild to moderate; post‐TAVR, moderate to severe). This abnormal flow condition leads to high shear stresses on the mitral valve leaflets, LV wall, and TAV leaflets. LV indicates left ventricle; MR, mitral regurgitation; PVL, paravalvular leakage; TAV, transcatheter aortic valve; TAVR, transcatheter aortic valve replacement.

Figure 6

Flow modeling in an LV with mild‐to‐moderate PVL in patient #59. (A) Velocity vector during diastole. (B) Wall shear stress during diastole. In patient #59, PVL, described by 1 jet, interacts with the mitral inflow and may worsen the MR post‐TAVR (MR status from Doppler echocardiography measurements: pre‐TAVR, mild; post‐TAVR, mild to moderate). LV indicates left ventricle; MR, mitral regurgitation; PVL, paravalvular leakage; TAVR, transcatheter aortic valve replacement.

Figure 7

Flow modeling in an LV with mild PVL in patient #35. (A) Velocity vector during diastole. (B) Wall shear stress during diastole. In patient #35, PVL, defined by 1 jet, interacts loosely with the mitral inflow (MR status from Doppler echocardiography measurements: pre‐TAVR: mild; post‐TAVR: mild). LV indicates left ventricle; MR, mitral regurgitation; PVL, paravalvular leakage; TAVR, transcatheter aortic valve replacement.

Figure 8

Changes in mitral regurgitation status in patients with MVD between baseline and 90 days post‐TAVR (N=72). (A) In individual MVD patients. (B) Average in all 72 MVD patients. (C) Individual MR status was improved. (D) Individual MR status worsened. (E) Individual MR status remained unchanged. MR indicates mitral regurgitation; MVD, mixed valvular disease; TAVR, transcatheter aortic valve replacement.

Figure 9 shows changes in LV workloads versus mitral valve effective regurgitant orifice area (measured by Doppler echocardiography) between baseline and 90‐day post‐TAVR in patients with MVD (N=72). In all panels of this figure, patients with negative changes in LV workload benefited from TAVR (shaded regions). Figure 9A suggests that, in some patients (quadrant IV), although TAVR worsened the mitral regurgitation condition, it still improved LV workload. Figure 9B suggests that all patients with pre‐TAVR regurgitant mitral valve effective regurgitant orifice area >0.3 (moderate and worse) benefited from TAVR regardless of the conditions of other disease constituents. Interestingly, Figure 9C suggests that LV workload worsened in all patients who were found to have regurgitant mitral valve effective regurgitant orifice area >0.3 (moderate and worse) post‐TAVR.

Figure 9

Changes in LV workloads vs mitral valve effective regurgitant orifice area in patients with MVD between baseline and 90 days post‐TAVR (N=72). LV indicates left ventricle; MVD, mixed valvular disease; TAVR indicates transcatheter aortic valve replacement.

Discussion

In MVD, valvular (PVL and MR), and ventricular (eg, LV dysfunction and heart failure) diseases interact with one another, and these phenomena are independent predictors of short‐ and long‐term mortality following TAVR. We showed that reduction of transvalvular pressure gradient does not predict outcome of TAVR because of the effects of the PVL on MR and LV hemodynamics. Given that the benefit of TAVR has been unequivocally shown, an important concern is how to address pathologies in other valves—at the same time, before or after TAVR procedure, and medically or mechanically. Our study brings mechanistic insight to address this increasingly common clinical dilemma.

In MVD Patients: Mitral Regurgitation May Be Exacerbated by Paravalvular Leak Post‐TAVR

MR is a common entity in patients with AS—and perhaps it even arises from the long‐term effects of pressure overload from the stenotic valve. Concomitant MR, left untreated at the time of TAVR, has been associated with increased all‐cause mortality.2, 10, 11, 12 MR may increase patients’ vulnerability because any complication leading to fluid dynamics or workload instability may rapidly decompensate the patient's hemodynamic status, leading to a refractory heart failure and shock. We are already aware that many patients experience a significant improvement in the MR post‐TAVR, but, in many others, MR worsens.2, 10, 12 Substantial MR was reported to increase early (in‐hospital or 30‐day) mortality post‐TAVR.27, 28 It is crucial to identify patients in whom MR will not improve or will even progress post‐TAVR.29 In these patients, the increased risk of a double‐valve procedure may be justified.

Pre‐existing MR may be exacerbated by PVL

During filling of the healthy heart, the blood entering the LV through the mitral valve forms a vortex that minimizes energy dissipation and optimizes pumping efficiency.30 The vortical structure in the LV is altered in the presence of valvular and ventricular diseases. In such cases, vortex dynamics become less synchronized with the heart contraction than the healthy vortex ring is, and other vortices may emerge and interact with one another.30 In agreement with findings of Pibarot et al,6 our results suggest that PVL, characterized by multiple jets, enters the LV and directly interacts with mitral inflow (eg, Figures 5, 6 through 7) and worsens MR in MVD patients post‐TAVR. These modeling findings were confirmed by our clinical Doppler echocardiography data (Figures 8 and 9): From 72 MVD patients, individual MR status worsened in 32 patients and remained unchanged in 18 others post‐TAVR.

Additionally, we observed (Figures 5, 6 through 7) that PVL leads to elevated shear stresses on mitral valve leaflets. This can cause tissue inflammation, which can ultimately lead to mitral valve failure (34–37).

In Some MVD Patients: No Improvement in Left Ventricular Hemodynamics Post‐TAVR

Some patients, who underwent TAVR, experienced a significant improvement in terms of pronounced reverse LV remodeling and less congestive heart failure symptoms. However, the situation in some other patients worsened. LV workload is an effective metric of LV load and clinical state23, 25, 31 and represents the energy that the ventricle delivers to the blood during ejection plus the energy necessary to overcome the viscoelastic properties of the myocardium itself. LV workload is the integral of LV pressure and its volume change and was calculated as the area encompassed by the LV pressure‐volume loop. Our results revealed that, pre‐TAVR, AS increased the burden on the LV attributable to augmented flow resistance, which caused an LV pressure overload. Although the flow resistance and consequently the LV pressure decreased greatly postintervention, the LV load did not improve given that PVL contributed to an immediate switching from a LV pressure overload to a LV volume overload (Figures 3 and 4).

Patients With MVD Who Benefited From TAVR

Patients with AS and aortic valve regurgitation, representing up to 40% of the candidates of TAVR, derived great benefits from TAVR despite even developments of a mild PVL post‐TAVR that had a minimal influence on outcomes.32, 33 Our results showed that TAVR led to significant improvements in LV workload in MVD patients who had AS and aortic valve regurgitation before the procedure (N=21; Figure 10A). Such LV workload improvements are accompanied by improvements in clinical metrics such as NYHA heart failure classification (Figure 10B) and ejection fraction (Figure 10C).

Figure 10

Changes in clinical assessments of LV and LV workloads in patients with MVD with pre‐existing aortic valve regurgitation between baseline and 90 days post‐TAVR (N=21). (A) LV workload. (B) Heart failure classification. (C) Ejection fraction. LV indicates left ventricle; MVD, mixed valvular diseases; NYHA, New York heart association; TAVR, transcatheter aortic valve replacement.

Patients With PVL Following TAVR, But Without Pre‐Existing MR

We investigated 48 patients with post‐TAVR PVL, but without pre‐existing MR. TAVR reduced LV workload in 28 of these 48 patients (60%; P<0.05). In 8 of these patients, LV workload did not significantly reduce (<5% reduction) post‐TAVR, and in 12 others, LV workload increased post‐TAVR. Indeed, post‐TAVR, LV pressure decreased, but LV workload did not always improve because PVL contributed to a switch from LV pressure overload to LV volume overload. The same process happened in patients with MVD, but more intensively. In fact, PVL and MR had mechanical interactions with each other and potentially may contribute in worsening of MR. Most patients who suffered from MVD post‐TAVR (75% of N=72; P<0.05) showed no improvements or even worsening of LV workload, whereas the majority of the patients with PVL, but without a pre‐existing MR condition (60% of N=48; P<0.05), showed improvements in LV workload.

Limitations

This study was performed on 72 patients with MVD in both pre‐ and postintervention states. Future studies must be conducted on a larger population of MVD patients to further confirm the findings of this study. In addition, this study did not consider data collection immediately post‐TAVR. Future studies should be designed to include this in the data collection protocol. One limitation that may be associated with our simulations is modeling the transcatheter aortic valve leaflets to be rigidly close and mitral valve leaflets to be rigidly open throughout the diastolic phase. However, this study focuses on diastole, the left ventricular filling phase, when PVL occurs. Furthermore, the good agreement between the numerical simulations (in progress for our other study) and Doppler echocardiography velocity measurements, which include moving valve leaflets, shows that this limitation does not affect the conclusions of this study. Future numerical studies will consider the interactions between the fluid and deforming valve‐leaflet structure during the entire cardiac cycle and will investigate the effects of dynamical opening and closing of the aortic valve leaflets on vortex dynamics in the LV.

Conclusions

PVL limits the benefit of TAVR; it may increase LV load, may potentially worsen MR, and worsens heart failure. MR and its possible exacerbation by PVL will likely play an important hindering role in the success of TAVR. This presents an important finding, especially considering the expansion of TAVR candidacy to lower‐risk and/or young patients.

The findings of this study suggest that beyond standard clinical indices for hemodynamic evaluation of valvular disease (eg, Doppler echocardiography pressure gradients), valvular and ventricular hemodynamics and their interactions should be clinically quantified and considered to better conduct aortic valve management, treatment planning, and patient risk stratification.

Sources of Funding

Keshavarz‐Motamed and Khodaei were supported, in part, by Natural Sciences and Engineering Research Council (NSERC) Discovery Grant (RGPIN‐2017‐05349). Edelman was supported, in part, by a grant from the National Institutes of Health (GM R01 49039).

Disclosures

Keshavarz‐Motamed has research grants from NSERC, but there is no overlap with the work in this study. Edelman has research grants from Edwards LifeSciences, Boston Scientific, and Medtronic, but there is no overlap with the work in this study. The remaining authors have no disclosures to report.

Data S1. Computational Methodology.

Click here for additional data file.