The Use of Biophysical Flow Models in the Surgical Management of Patients Affected by Chronic Thromboembolic Pulmonary Hypertension

Abstract

Introduction: Chronic Thromboembolic Pulmonary Hypertension (CTEPH) results from progressive thrombotic occlusion of the pulmonary arteries. It is treated by surgical removal of the occlusion, with success rates depending on the degree of microvascular remodeling. Surgical eligibility is influenced by the contributions of both the thrombus occlusion and microvasculature remodeling to the overall vascular resistance. Assessing this is challenging due to the high inter-individual variability in arterial morphology and physiology. We investigated the potential of patient-specific computational flow modeling to quantify pressure gradients in the pulmonary arteries of CTEPH patients to assist the decision-making process for surgical eligibility. Methods: Detailed segmentations of the pulmonary arteries were created from postoperative chest Computed Tomography scans of three CTEPH patients. A focal stenosis was included in the original geometry to compare the pre- and post-surgical hemodynamics. Three-dimensional flow simulations were performed on each morphology to quantify velocity-dependent pressure changes using a finite element solver coupled to terminal 2-element Windkessel models. In addition to transient flow simulations, a parametric modeling approach based on constant flow simulations is also proposed as faster technique to estimate relative pressure drops through the proximal pulmonary vasculature. Results: An asymmetrical flow split between left and right pulmonary arteries was observed in the stenosed models. Removing the proximal obstruction resulted in a reduction of the right-left pressure imbalance of up to 18%. Changes were also observed in the wall shear stresses and flow topology, where vortices developed in the stenosed model while the non-stenosed retained a helical flow. The predicted pressure gradients from constant flow simulations were consistent with the ones measured in the transient flow simulations. Conclusion: This study provides a proof of concept that patient-specific computational modeling can be used as a noninvasive tool for assisting surgical decisions in CTEPH based on hemodynamics metrics. Our technique enables determination of the proximal relative pressure, which could subsequently be compared to the total pressure drop to determine the degree of distal and proximal vascular resistance. In the longer term this approach has the potential to form the basis for a more quantitative classification system of CTEPH types.

Keywords: CTEPH; HPC-based computational modeling; biophysical flow modeling; computational physiology; patient specific computational modeling.

Figure 1

Pipeline for the generation of patient-specific models and simulations in this study (Patient 2, posterior view). (A) Post-surgical CT pulmonary angiography. (B) Segmentation of the stenosed model with surface interpolation and arrow indicating stenosis on the LPA. (C,D) tetrahedral mesh (C) with curvature refinement (D). (E) Prescription of boundary conditions at the outlets (two-element Windkessel model) and at the inlet (Dirichlet condition on flow velocity); (F) CFD simulation of blood flow visualized using streamlines colored by velocity magnitude.

Figure 2

Pressure fields at the wall at peak systole for Patients 1, (A,B), 2 (C,D), and 3 (E–G), in the stenosed and non-stenosed morphologies, posterior view. Patient 3a is illustrated in (E) and Patient 3b in (F).

Figure 3

Simulation results at peak systole. (A) Streamlines colored by velocity magnitude in Patient 2. (B) WSS magnitude in Patient 2. (C) Streamlines colored by velocity magnitude in Patient 3a; (D) WSS magnitude in Patient 3a. The mild stenosis case was employed for Patient 3a. The inserts show the flow behavior and WSS magnitude in the stenosed segment of the pulmonary arteries, highlighting the flow recirculation regions, and the helical flow (A,C) and the drop in the WSS magnitude (B,D).

Figure 4

(A,B) Pressure gradients from constant flow simulations in the non-stenosed (A) and stenosed model (B), with fitted parametric curve. (C,D) Pressure gradient over time in the non-stenosed (C) and stenosed (D) model. The solid line indicates the pressure gradient predicted by parametric modeling, while the dashed line represents an average of the gradients obtained in LPA and RPA from the transient flow simulation.