Validating In Silico and In Vitro Patient-Specific Structural and Flow Models with Transcatheter Bicuspid Aortic Valve Replacement Procedure

Abstract

Introduction: Bicuspid aortic valve (BAV) is the most common congenital cardiac malformation, which had been treated off-label by transcatheter aortic valve replacement (TAVR) procedure for several years, until its recent approval by the Food and Drug Administration (FDA) and Conformité Européenne (CE) to treat BAVs. Post-TAVR complications tend to get exacerbated in BAV patients due to their inherent aortic root pathologies. Globally, due to the paucity of randomized clinical trials, clinicians still favor surgical AVR as the primary treatment option for BAV patients. While this warrants longer term studies of TAVR outcomes in BAV patient cohorts, in vitro experiments and in silico computational modeling can be used to guide the surgical community in assessing the feasibility of TAVR in BAV patients. Our goal is to combine these techniques in order to create a modeling framework for optimizing pre-procedural planning and minimize post-procedural complications.

Materials and methods: Patient-specific in silico models and 3D printed replicas of 3 BAV patients with different degrees of post-TAVR paravalvular leakage (PVL) were created. Patient-specific TAVR device deployment was modeled in silico and in vitro-following the clinical procedures performed in these patients. Computational fluid dynamics simulations and in vitro flow studies were performed in order to obtain the degrees of PVL in these models.

Results: PVL degree and locations were consistent with the clinical data. Cross-validation comparing the stent deformation and the flow parameters between the in silico and the in vitro models demonstrated good agreement.

Conclusion: The current framework illustrates the potential of using simulations and 3D printed models for pre-TAVR planning and assessing post-TAVR complications in BAV patients.

Keywords: 3D printing; Aortic valve; Bicuspid aortic valve (BAV); Paravalvular leakage (PVL); Patient-specific computational modeling; Transcatheter aortic valve replacement (TAVR).

Figure 1:

Structural simulation of TAVR procedure in one of the in-silico patient-specific models (Patient 3). (a) Uncrimped 29mm Evolut R stent inside the catheter. (b) Fully crimped stent being delivered at the aortic root of the patient. (c) 50% of the deployment step. (d) Fully deployed stent inside the aortic root. (e) Leaflet calcium distribution in all the patients. The LVOT calcification in patient 2 and the massive leaflet calcification in patient 3 are circled in red

Figure 2:

(a) Patient-specific flexible 3D printed model of one of the patient cases (Patient 1). The in-vitro model (b) was good representatives of the in-silico model (c). A self-expandable device (29mm CoreValve) (d) was manually deployed in the 3D printed models (e, f)

Figure 3:

Comparing the device deployment and performing a two-way validation between the in-silico and in-vitro models. (a) X-Ray image of the flexible 3D printed model of one of the patient (Patient 2) aortic roots with the deployed TAVR device in it. (b) FE model of the same patient. (c) Patient-specific surface mesh reconstructed from the CT scans obtained from the 3D printed model. (d) Corresponding FE model. The device eccentricity and the minimum and maximum diameters of the device was calculated at the annulus plane A, and at multiple other planes, P1- P5, drawn at the same levels in both in-vitro and in-silico models (c, d)

Figure 4:

(a) Custom patient-specific 3D printed rigid model of one of the patients (Patient 1), designed to fit the left-heart simulator, the patient-specific aortic root region is highlighted in the zoomed in part. (b) A 29mm CoreValve device manually deployed in the model. (c) The orientation of the device was adjusted according on the FE model. (d) The left heart simulator used in our flow study

Figure 5:

(a, b) Assessment of the anchorage of a 29mm CoreValve in all the in-silico patient models by analyzing the anchorage area (a) and anchorage force (b). (c) The in-silico and in-vitro device eccentricity index at the annulus level was compared

Figure 6:

Bland-Altman plot analyzing the level of agreement between the in-silico and in-vitro measurements. The upper and lower limit of agreement (Upper LOA and lower LOA) were defined by the mean difference ± 1.96 standard deviation

Figure 7:

Comparing the 29mm CoreValve deployment profile in all the in-vitro and in-silico patient models; (1^st and 2^nd column) Lateral views and (3^rd and 4^th column) top views of the patient-specific FE and 3D printed models at the end of the deployment. (5^th column) Overlay of the deformed devices obtained from the in-silico (black) and in-vitro (yellow) models

Figure 8:

CFD results of the TAVR simulation based on the devices that these patients originally received in order to perform clinical validation. (1^st and 2^nd column). Velocity streamlines through the patient-specific aortic root depicting the course and origin of the PVL flow jets from two different angles. (3^rd column) Freeze frames from the post-TAVR echo-Doppler color flow videos, indicating the location of the PVL jets in the patients

Figure 9:

(a) In-silico and in-vitro flowrates in all patients during one cardiac cycle. (Since the focus of our study was to analyze PVL flow during diastole, the original in-vitro flowrates obtained from the hydrodynamic studies were modified by setting the flowrates at the systolic parts to zero) (b) In-silico and in-vitro regurgitant volumes which were calculated by integrating the area under the flowrate curve during diastolic phase