Effects of MitraClip Therapy on Mitral Flow Patterns and Vortex Formation: An In Vitro Study

Abstract

MitraClip transcatheter edge-to-edge repair is used to treat mitral regurgitation (MR). While MR is reduced, diastolic left ventricular flows are altered. An in vitro left heart simulator was used to assess a porcine mitral valve in the native, MR, and MR plus MitraClip cases. Velocity, vorticity, and Reynolds shear stress (RSS) were quantified by particle image velocimetry. Peak velocity increased from 1.20 m/s for native to 1.30 m/s with MR. With MitraClip, two divergent jets of 1.18 and 0.61 m/s emerged. Higher vorticity was observed with MR than native and lessened with MitraClip. MitraClip resulted in shear layer formation and downstream vortex formation. Native RSS decreased from 33 Pa in acceleration to 29 Pa at peak flow, then increased to 31 Pa with deceleration. MR RSS increased from 27 Pa in acceleration to 40 Pa at peak flow to 59 Pa during deceleration. MitraClip RSS increased from 79 Pa in acceleration to 162 Pa during peak flow, then decreased to 45 Pa during deceleration. After MitraClip, two divergent jets of reduced velocity emerged, accompanied by shear layers and recirculation. Chaotic flow developed, resulting in elevated RSS magnitude and coverage. Findings help understand consequences of MitraClip on left ventricular flow dynamics.

INTRODUCTION

Mitral regurgitation (MR) is found in 1.7% of the general population and 9.3% of those age 75 years and older.³ MR occurs when leaflets of the mitral valve (MV) do not properly coapt during systole, resulting in regurgitant flow back into the left atrium. Transcatheter edge-to-edge repair (TEER) using the MitraClip device (Abbott, Menlo Park, California), which grasps and coapts the posterior and anterior mitral valve leaflets, is currently the only TEER device approved by the Federal Drug Administration (FDA) to treat patients with symptomatic moderate-to-severe (3+) or severe (4+) MR deemed prohibitive risk for MV surgery.

Blood flow through a native MV has minimal energy dissipation and greater efficiency in physiological, non-diseased cases.^26,34 However, mitral replacement and repair configurations, such as the double orifice formed after MitraClip placement, creates non-physiological anatomy that results in altered downstream flow patterns.^1,25,34,38 These abnormal or non-physiological flows were correlated to ventricular remodeling and some adverse effects.^25,35 Specifically, non-physiological flows such as altered vortex dynamics and generation of shear layers may contribute to ventricular adverse and reverse remodeling observed in patients.^{5,8,9,12,36,37} In several cases, patients treated with MitraClip have formed left ventricular thrombi or a thrombus appending the MitraClip.^{2,11,17,29,30}

The non-physiological double orifice configuration has been investigated over the past two decades. In 2001, Redaelli et al. performed a computational study on a highly simplified model of edge-to-edge repair.⁴¹ In 2009, Shi et al. explored edge-to-edge repair in porcine MVs under steady flow conditions in vitro.⁴² In 2018, Jeyhani et al. explored the effect of MitraClip repair in a highly simplified MV analog in vitro.²³ In 2020, Caballero et al. performed a patient-specific computational study on MitraClip and annuloplasty.⁷ More details on these studies will be discussed later. However, to our knowledge, flow analysis using particle image velocimetry (PIV) with MitraClip on a tissue valve under pulsatile conditions has not been performed prior to the present study.

The objective of this study is to assess the diastolic hemodynamic differences in the flow field in the vicinity of the valve post-MitraClip repair. The presence of the MitraClip at the center of the double orifice is expected to generate shear layers in the wake of the MitraClip during diastolic filling, which may damage blood cells and other blood components depending on the forces generated.⁴³ The structure of the shear layer in the anterior-posterior direction in relation to that in the septal-lateral direction is largely unknown, and the near-orifice vortex characteristics have not been characterized to-date.

MATERIALS AND METHODS

Mitral Valve

In this study, the fresh porcine mitral valve in Fig. 1a was tested. The valve was procured from a pig weighing approximately 250 lbs. from Bay Packing (Lancaster, Ohio). The MV had septolateral and intercommissural dimensions of 29 mm and 36 mm and was prepared with methods developed previously.¹⁵ Briefly, the valve was sutured to a plate that was cut to its annulus size, and an attachment piece was sutured to each papillary muscle for fixation to the papillary muscle control mechanism later.

FIGURE 1.

(a) Porcine mitral valve with MitraClip. (b) illustration of mitral valve chamber.

Valve States

The valve was modeled in the native unaltered state with induced primary MR and MR treated with a centrally placed MitraClip NT. Acute moderate-to-severe P2 prolapse was induced via cutting chordal connections to the P2 scallop until the mitral regurgitant fraction (MRF = regurgitant flow/forward flow³²) reached between 30 and 50%.^16,28 The models used yielded similar outcomes to our previous study (Gooden et al., 2020), where MRF and MV pressure gradient (MVPG) increased when MR was induced and decreased after MitraClip.¹⁵

Left Heart Simulator

The porcine valve was mounted into the specialized mitral valve chamber shown in Fig. 1b. Briefly, the valve plate was fixed, and the papillary muscles were attached to the papillary muscle control rods, which were moved such that there the chordae tendineae were taut. The valve was tested in an in vitro left heart simulator, shown in Fig. 2a, that functionally simulates all components of the left heart and integrates flow rate and pressure measurement points.^4,15,31 The simulator includes a fluid reservoir, a custom-made bladder pump for the left atrium, a flow probe (ME 20PXL, Transonic Inc., Ithaca, NY) measuring volumetric mitral flow rate, a specialized MV chamber with the MV and Millar pressure catheters (Millar Inc., Houston, Texas) for high fidelity MVPG measurement, a custom-made bladder pump for the pumping function of the left ventricle, an aortic valve chamber and valve, and a compliance chamber and resistance valve to simulate systemic compliance and control flow rate. Note that the MV chamber acts as the left ventricle chamber, which is similar to previous studies.^{4,10,14,15,21,24,32} More details on the setup are available in previous published work by our group.¹⁵ A custom LabVIEW program controlled flow and pressure data acquisition as well as timing of compressed air and a vacuum that regulated the bladder pumps.

FIGURE 2.

(a) Schematic of the left heart simulator. Note the location of the Millar pressure catheters are upstream at the mitral annulus and downstream immediately after the mitral leaflets in the ventricle. (b) Illustration of the two PIV imaging planes. Note that in this example, the pinched outlet represents the MitraClip, yielding the double orifice.

The valve in the aortic position was the 21mm Medtronic Hancock II bioprosthetic aortic valve. The working fluid used was 60–40 volume ratio of water and glycerin to achieve the density (1060 kg/m³) and kinematic viscosity of blood (3.88 cSt).

The setup allowed for high fidelity pressure and flow hemodynamic measurements of the valve to quantify parameters, including effective orifice area (EOA) and MRF, as well as PIV data collection explained later.

Imposed Hemodynamic Conditions

Heart rate was 60 beats per minute, systolic ventricular pressure was 120 mmHg, and atrial systole was 12% of the cardiac cycle. The cardiac output was 3 L/min as used in our previous because the valve acquired was smaller than that of a human.¹⁵

Particle Image Velocimetry

High-resolution time-resolved particle image velocimetry (PIV) was conducted to visualize and quantify the flow velocity field through the single orifice in the native and MR models and through the double orifice with MitraClip using a 2D2C PIV system. This included: (1) phase locked measurements to evaluate turbulence characteristics downstream the MV with 250 images for three phases of flow (acceleration, E-wave peak, and deceleration); (2) time series videos (recorded at 1000 Hz) for velocity and vorticity through the cardiac cycle. Note that in this manuscript, we use the terminology “downstream” to refer to the ventricular flow dynamics in the vicinity of the valve. The working fluid was seeded with fluorescent PMMA-Rhodamine B particles with an average diameter of 10 μm that fluoresced upon exposure to a laser sheet created by a pulsed Nd:YLF single cavity diode pumped solid state laser coupled with external spherical and cylindrical lenses. PIV data acquisition and processing were performed using DaVis PIV software (LaVision, Germany). Refraction was corrected, and velocity vectors were calculated using and advanced adaptive cross-correlation algorithms, with a 50% overlap multi-pass approach. Finally, post-processing was performed using adaptive median filtering.

Velocity vectors were calculated using adaptive cross-correlation algorithms used previously,¹⁹ where the initial interrogation window size of the multi pass scheme was at 32×32 pixels, which progressively reduced to 16×16 pixels. The interrogation window overlap was fixed at 50%.

The two planes analyzed in this study are the commissural plane and the anterior-posterior plane, as illustrated in Fig. 2b, and an example image of the physical set-up is shown in Fig. 3.

FIGURE 3.

Physical set-up of the PIV system. Placement of high-speed camera, laser, and mirror for alignment of the laser sheet in relation of the mitral valve. The inset gives a better view of the valve orientation.

Vorticity was calculated using Eq. 1.

$${\omega }_z=\frac{d{V}_y}{dx}-\frac{d{V}_x}{dy}$$ (1)

RESULTS

Regurgitation Fraction and Effective Orifice Area

Upon induction of MR, MRF increased from 12.44 ± 0.47% to 48.61 ± 1.36%, EOA increased from 1.58 ± 0.38 cm² to 2.82 ± 0.40 cm², and MVPG increased from 2.64 ± 1.02 mmHg to 3.20 ± 0.93 mmHg. With MitraClip, MRF reduced to 15.08 ± 0.51%, MVPG reduced to 3.00 ± 0.89 mmHg, and EOA reduced to 1.62 ± 0.17 cm², showing MitraClip is successful in greatly reducing regurgitation.

Vorticity and Velocity

The vorticity field with velocity vectors in the commissural plane is shown in Fig. 4. The velocity of the jet during peak flow in the native case was 1.20 m/s. With MR, the peak flow increased to 1.30 m/s. Presence of the clip resulted in two divergent jets, where the velocity of the jet from the posteromedial orifice remained high at 1.18 m/s, and the jet from the anterolateral orifice had reduced velocity of 0.61 m/s. Velocity immediately after the MitraClip is diminished in the center. A counterclockwise vortex is formed downstream the valve after the MitraClip is introduced, which is mostly apparent at peak flow and during deceleration. Note that this feature is new and unique to only the MitraClip configuration.

FIGURE 4.

Vorticity field in the commissural plane. PM and AL note the posteromedial and anterolateral papillary muscles, respectively.

The time series video through one cardiac cycle in the commissural plane for native, MR, and MitraClip cases are in Supplemental Videos 1, 2, and 3, respectively.

Vorticity in the anterior-posterior plane is shown in Fig. 5, where the posterior leaflet is the upper leaflet in the image. The velocity of the jet during peak flow in the native case was 1.51 m/s. With MR, the peak flow increased to 1.80 m/s. With MitraClip, the peak flow decreased to a near-zero velocity of 0.18 m/s, which is non-zero due to lack of perfect symmetry of the porcine mitral valve.

FIGURE 5.

Vorticity field in the anterior-posterior plane. AL notes the anterolateral papillary muscle. *Vector length is doubled compared to the commissural plane. **Vector length is 6x of those in the commissural plane.

The time series video through one cardiac cycle in the anterior-posterior plane for native, MR, and MitraClip cases are in Supplemental Videos 4, 5, and 6, respectively.

Vorticity Dynamics

High vorticity is observed with MR through the entirety of diastole (Supplemental Videos 1, 2, 4, and 5). With placement of the MitraClip, the vorticity magnitude lessens and spans less of the downstream area (Supplemental Videos 3 and 6). From the vorticity fields (Figs. 4 and 5), formation of a shear layer, especially at peak flow and deceleration, can be seen with MR, and the intensity diminishes with treatment with the MitraClip.

Focusing on the commissural plane during peak flow, the shear layer of the posterior orifice was 24 s⁻¹ at exit of leaflet tip and changed to 215 s⁻¹ 1.5 cm downstream the leaflet tip and 86 s⁻¹ 2.8 cm downstream the leaflet tip. The shear layer of the anterior orifice was 89 s⁻¹ at exit of leaflet tip and changed to 138 s⁻¹ 2.2 cm downstream the leaflet tip and 94 s⁻¹ 3.5 cm downstream the leaflet tip. Overall, vorticity at peak flow decreases from around 300 s⁻¹ to 200 s⁻¹.

Turbulence

Turbulence in the commissural plane is shown in Fig. 6. Through the cardiac cycle, maximum Reynolds shear stress (RSS) for the native case decreased from 33 Pa during acceleration to 29 Pa at peak flow, then increased to 31 Pa with deceleration. With MR, maximum RSS increased from 27 Pa in acceleration to 40 Pa at peak flow and continued to increase to 59 Pa during deceleration. With MitraClip, maximum RSS increased from 79 Pa during acceleration to 162 Pa during peak flow, then decreased to 45 Pa during deceleration.

FIGURE 6.

Turbulence in the commissural plane. PM and AL note the posteromedial and anterolateral papillary muscles, respectively.

Turbulence in the anterior-posterior plane is shown in Fig. 7, where the posterior leaflet is the upper leaflet in the image. Through the cardiac cycle, maximum RSS increased from 36 Pa during acceleration to 75 Pa at peak flow, then decreased to 48 Pa with deceleration. With MR, maximum RSS decreased slightly from 123 Pa in acceleration to 118 Pa at peak flow and continued to decrease to 72 Pa during deceleration. After MitraClip, maximum RSS increased from 42 Pa during acceleration to 57 Pa during peak flow, then decreased to 38 Pa during deceleration.

FIGURE 7.

Turbulence in the anterior-posterior plane. AL notes the anterolateral papillary muscle.

DISCUSSION

In this study, diastolic flow in the vicinity of a porcine mitral valve was studied under three different conditions: (1) native valve (unaltered), (2) with moderate-to-severe MR, and (3) after MitraClip implantation. The flow patterns changed significantly among the three groups, and an increase in turbulence was observed post-MitraClip.

Vorticity and Velocity

MitraClip presence led to a change in the diastolic flow characteristics downstream of the valve, mainly apparent in the (1) velocity magnitude, (2) vortex initiation and propagation, and (3) vorticity dynamics after MitraClip.

Velocity Magnitude

The peak velocity increased with induction of MR, which occurs in order to maintain the desired cardiac output. Velocity of > 1.20 m/s in this study is consistent with past experimental and clinical observations for MR.^22,23,44 Treating the regurgitant valve with MitraClip created two smaller orifices, which separated the flow trajectory into two pathways of decreased peak velocities, as shown in the commissural plane (Fig. 4). A near-zero velocity was observed in the anterior-posterior plane after MitraClip, which may be due to the lack of perfect symmetry of the porcine mitral valve.

Vortex Initiation and Propagation

Vortex formation for the native case occurs in the anterior-posterior plane (Fig. 5), with structures initiating at the leaflet tip during acceleration, becoming stronger as they move downstream during peak flow, and diminishing as they move further downstream during deceleration. These findings are consistent with previous studies.^1,22,23 Vortex formation with MitraClip occurs in the commissural plane (Fig. 4), with structure appearing and diminishing similar to the native case in the anterior-posterior plane. Flow visualization with MitraClip is also shown in Supplemental Video 7, where the vortex structures diminish downstream by early deceleration following the E-wave, and an additional set of small vortices emerge as a result of the A-wave and quickly diminish with valve closure. Overall, the presence and behavior of these structures may result in the change of MVPG.⁴¹ Studying and understanding these vortex structures is important because abnormal intraventricular flow and vortex structures have been known to play a crucial role in adverse remodeling, in addition to increasing energy dissipation and altering pressure distribution.^12,36

Vorticity Dynamics after MitraClip

High vorticity was observed with MR, which diminished after MitraClip placement (Fig. 4). This shows that the placement of a MitraClip changes the topology of flow structures and the magnitude of local rotation of the blood flow elements in the ventricle. The shear layers in the wake of the leaflets and edge of the MitraClip are also asymmetric, owing to the asymmetry the mitral apparatus and MitraClip placement.

Turbulence

RSS for the native and MR cases across all phases spanned 27–162 Pa. This is thought to be due to cycle-to-cycle variabilities. This can be seen by comparing each velocity image that composes the phase-locked turbulence field, which is shown for native and MR cases in both imaging planes for all phases in Supplemental Videos 8–19. On the other hand, elevation of RSS after introduction of the MitraClip is likely due to a combination of cycle-to-cycle variabilities (Supplemental Videos 20–25) and MitraClip presence that decreased EOA—a single outlet of 2.8 cm² to a double outlet totaling 1.6 cm². The substantially higher RSS in the commissural plane for the MitraClip is consistent with past studies,²³ where there is both an increase of RSS and greater area with elevated RSS. This shows increased viscous energy dissipation resulting from the altered flow patterns, which, in the setting of a pumping heart, increases the energy required by the myocardium in order to eject blood.³⁴

Elevated RSS as a result of altered flow with MitraClip may correlate to thrombus formation and abnormal pannus growth appending to MitraClip, which is present in some patients.^{2,11,17,29,30,45}

Comparison to Similar Studies

As mentioned previously, the double orifice configuration has been studied previously but using highly simplified approximations, imposing steady flow conditions, or not assessing the flow features measured in the present study. Details of studies of interest are summarized in Table 1, and results of each study at peak flow are summarized as follows for comparison to the present study.

TABLE 1.

Details of similar studies on the double orifice configuration.

Author	Study type	Flow conditions	Mitral valve	Subvalvular apparatus	Left ventricle	MitraClip analog
Redaelli et al.41	In silico	Pulsatile, peak flow 16 L/min	Circular 19 mm annulus (single orifice: converging tube)	Excluded	Approximated	Post-clip valve: dome with two circular outlets
Shi et al.42	In vitro, PIV in commissural plane	Steady, 15 L/min	32 mm porcine mitral valve	Excluded	Box Model	6 mm suture
Jeyhani et al.23	In vitro, PIV in commissural plane	Pulsatile, peak flow 10.5 L/min post-clip	27 mm straight tube (unspecified material)	Excluded	Approximated	Simulated MitraClip
Caballero et al.7	In silico	Pulsatile, peak flow 16.8 L/min pre-clip	Patient-specific with functional MR	Patient-specific	Patient-specific	Simulated MitraClip

Redaelli et al.⁴¹ found that before intervention, peak velocity was 0.94 m/s. After central edge-to-edge repair, two divergent jets emerged, and there is low-flow in the wake of the repair. The peak velocity of the jets were 1.00 m/s and 1.06 m/s. RSS was not assessed in this study. However, the valve model used in this study was quite small, did not incorporate the scallops of a human mitral valve, and was not able to faithfully represent a regurgitant mitral valve due to the lack of differences of diastolic flow required to maintain a set cardiac output between the pre- and post-repair configurations.

Shi et al.⁴² found that before intervention, peak velocity was about 1.00 m/s and maximum RSS was 11.5 Pa. After a central edge-to-edge repair, two 1.2 m/s jets emerged, with near-zero velocity in the wake of the repair and a recirculatory region downstream the valve. Maximum RSS was 28 Pa. In this steady flow study, however, the dynamic, time-varying events were not assessed.

Jeyhani et al.²³ found that before intervention, the peak velocity was 0.4 m/s, and RSS was 4 Pa. After central edge-to-edge repair, two divergent jets emerged, and there was near-zero flow in the wake of the simulated MitraClip. The peak velocity of the double orifice reached 1.6 m/s. Note that this peak flow rate was quite low compared to other studies. Maximum RSS was 36 Pa. Of note, the material of the MV model used in this study was not specified, and there was no comment on material properties compared to leaflet tissue. The MV model also did not resemble the mitral leaflets and was not able to faithfully represent a regurgitant mitral valve due to the lack of differences of diastolic flow required to maintain a set cardiac output between the pre- and post-repair configurations.

Caballero et al.⁷ found that before intervention, the peak velocity was 0.79 m/s. With one centrally placed MitraClip, two divergent jets merged, and there was no flow in the wake of the MitraClip. The peak velocity of the jets were 1.41 and 1.29 m/s. RSS was not assessed in this study.

Overall, focusing on the commissural plane, results from the present study are in good agreement with past studies on edge-to-edge repair. (1) Peak velocity before repair was around 1.30 m/s. This is higher than the studies summarized due to: (a) increased diastolic flow required to maintain a set cardiac output for the regurgitant valve and (b) the larger valve size used. (2) The jet velocities after repair were 1.18 m/s and 0.61 m/s in the posteromedial and anterolateral orifices, respectively. Contrary to the summarized studies, these velocities decreased, which may also be due to the increased diastolic flow and the larger valve size used. (3) Post-repair, two divergent jets emerge with low flow at the wake of the repair site, which is consistent with the studies summarized. (4) RSS increased from 40 Pa before repair to 162 Pa after repair. Although RSS was much higher in the present study compared to the studies summarized, the trend of increased RSS post-repair was observed.

Clinical Relevance

Studying the fluid dynamic implications after MitraClip aids in evaluating the performance and durability of the therapy.^27,42 Several patients treated with MitraClip have experienced thrombus formation, either within the left ventricle or appending the MitraClip.^{2,11,17,29,30} In this study, presence of the MitraClip was found to generate shear layers in the wake of the MitraClip during diastolic filling, and this elevated shear stress may be responsible for initiation of thrombus formation.⁴³ Also, velocities during diastolic filling contribute to MVG, where minimal MVG post-implant is desired.³³ Understanding how the velocity through each orifice is effected by MitraClip can also help optimize device usage with respect to resultant pressure gradients to improve patient outcomes. Overall, a true understanding of fluids in disease and remodeling is an area that is currently being studied but is advancing with coupling of clinical and experimental data. More patient-specific studies are underway to develop a comprehensive framework that touches base on true relevance.⁷

Study Limitations

The main limitation of this study is that the ventricular geometry was not anatomically correct. Therefore, measurements immediately after the valve hold true, while measurements further downstream likely differ from an anatomically correct model. Differences include vortex structures and velocities after the tip of each mitral leaflet as influenced by presence of the left ventricular wall and interventricular septum.⁴³ Hence, we expect various flow structures to occur within the ventricle. Another model limitation includes fixed PM placement. Use of 2D PIV is a limitation as well, which is why analysis was performed in 2 planes to capture more of the flow field. Although this is a limitation, 2D PIV is capable in giving the most prominent features. 3D PIV necessitates an independent 3D PIV system that was not available in the lab. Several heart valve studies have used 2D PIV for analysis.^{6,13,18,20,39,40} Lastly, optimal MitraClip placement was assumed to be central, and other clip locations were not explored.

Summary

The effect of MitraClip on diastolic left ventricular flow patterns was assessed in this study. Figure 8 summarizes findings. While MitraClip can effectively reduce regurgitation, two divergent jets of lower velocity emerge after MitraClip placement, accompanied by shear layers. Areas of recirculation and chaotic flow also develop downstream the MitraClip. These instabilities result in elevation of RSS with respect to value and downstream area. Ongoing studies are performed to further analyze mitral flow, including different configurations of MitraClip use.

FIGURE 8.

Changes of diastolic left ventricular flow patterns after MitraClip (shown at peak flow).

ABBREVIATIONS

MR

Mitral regurgitation

MV

Mitral valve

PIV

Particle image velocimetry

MRF

Mitral regurgitant fraction

MVPG

Mitral valve pressure gradient

EOA

Effective orifice area

RSS

Reynolds shear stress

TEER

Transcatheter edge-to-edge repair