Patient-Specific Predictive Simulations of Transcatheter-Edge-to-Edge Repair in Humans with Mitral Regurgitation

Introduction::
Mitral regurgitation (MR) is a highly prevalent cardiac disease affecting approximately 2% of the United States population. MR is typically categorized into two broad groups each covering a wide range of etiologies. Degenerative MR (DMR) is caused by primary myxomatous degeneration of the mitral valve (MV) tissue. In contrast, functional MR (FMR) is an insufficiency of an otherwise normal MV secondary to ischemic or non-ischemic cardiomyopathy and the associated left ventricular dilatation. MR of either type is also a major prognostic factor of mortality, with severe MR increasing the risk of mortality by 76%. Furthermore, MR remains severely undertreated despite its deadly outcomes, even in patients with low comorbidity, suggesting a critical need for treatment planning optimization.

Transcatheter edge-to-edge repair (TEER) techniques are significantly safer options for MV repair compared to traditional surgical approaches and have thus broadened the eligible patient cohort for MR treatment. However, the long-term durability and efficacy of TEER has been unpredictable, with major randomized controlled trials reporting diametrically opposite outcomes. These inconsistencies have largely been attributed to the vast heterogeneity of the MR patient population and emphasize the enduring challenge in appropriate patient selection for MV repair. Therefore, there remains an urgent, unmet clinical need for quantitative, patient-specific treatment optimization techniques based on preoperative standard-of-care clinical imaging that account for the highly variable patient population and combinatorial nature of TEER.


Materials and Methods::
First, we developed a finite-element model of the full, patient-specific MV apparatus. The MV chordae tendinae (MVCT) are not visible on transesophageal echocardiography (TEE), which is the standard-of-care imaging modality for the TEER procedure, and detailed distribution information does not exist for humans. Therefore, we analyzed 5 CT-imaged ex vivo human hearts and quantified the distribution of MVCT origins and insertions relative to the MV leaflets. Next, we obtained TEE image sets from 10 patients who underwent TEER with MitraClip (Abbott Inc.). From each patient’s preoperative images, we segmented the end-diastolic (ED) and end-systolic (ES) leaflet geometries and all identifiable MVCT origins in both the ED and ES states. Finally, we used the MVCT distribution map to integrate the chordal structure with the patient-specific leaflet geometry.

To model MV closure, we set the preoperative ED to ES annular and MVCT origin displacements as Dirichlet boundary conditions, assigned an established structural model for the human MV leaflet and an Ogden model for human MVCT, and applied a ventricular pressurization in Abaqus 6.14. Given that in vivo patient-specific material properties are impossible to obtain from TEE imaging and a full inverse model is highly computationally-intensive, we developed a simplified, tractable approach to tune the effective, patient-specific mechanical behavior by iteratively calibrating both the MV leaflet and MVCT prestrains. We utilized this fully calibrated MV model to simulate MV repair scenarios by implementing 3D MitraClip models onto the MV leaflets, assuming the annular displacement does not change immediately after MitraClip placement (Fig. 1A).


Results, Conclusions, and Discussions:: Our pipeline was able to predict the postoperative MV leaflet geometry to within 1.049 mm of the target (imaged) ES geometry, and the strain fields in both the circumferential and radial directions demonstrated close global and local correspondence to the target. Furthermore, we were able to predict a postoperative stress concentration on the posterior leaflet and increased radial strain in the center of the posterior leaflet (Fig. 1B & C), which may be drivers for subsequent repair failure. We also validated our assumption regarding the annular boundary condition and compared stress and strain results from simulations using the assumed preoperative annular displacement with the true postoperative annular displacement. We found almost no difference in leaflet stress or strain, suggesting that the preoperative annular displacement is adequate when simulating the immediate postoperative state (Fig. 2).

Given the complexity and inadequacy of current stratification methods for patient selection in MR treatment, there is a clear need for quantitative, patient-specific methods of predicting MV repair outcomes preoperatively using readily available clinical data. In this study, we have developed an anatomically faithful, patient-specific full MV apparatus model and a predictive TEER simulation pipeline based solely on preoperative standard-of-care imaging. Furthermore, we have demonstrated the capacity of this pipeline to predict the full postoperative MV functional state with close agreement to segmented data. The methods presented here thus lay the foundation for quantitative surgical planning, tailored patient selection, and ultimately, a more durable repair.




Acknowledgements (Optional): :

References (Optional): :