Hemodynamic comparison of differing anastomotic geometries using magnetic resonance velocimetry.

Background: Hemodynamic factors at the distal anastomosis play an important role in prosthetic graft performance. A new magnetic resonance imaging (MRI) technique was used to determine the effect of anastomotic geometry on hemodynamic flow patterns.

Methods: Four dimensional (4D) magnetic resonance velocimetry (4D-MRV) is a noninvasive method of analyzing pulsatile flow in three dimensions (3D). End-to-side anastomotic models were constructed by suturing 6 mm polytetrafluoroethylene (ePTFE) grafts to silicone tubing (4 mm i.d.). The models included straight ePTFE, precuffed ePTFE, and patched ePTFE configurations in a pulsatile system, which created flow consistent with physiologic flow rates and pressures. Blood was simulated by a solution of 40% glycerol in distilled water with trace gadolinium. The different models were imaged using MRV techniques in a three-dimensional (3D) coronal slab (0.5 mm thick coronal slices, in-plane field of view (FOV) 18 cm.) The data were reconstructed, resulting in an interpolated resolution of 0.35 mm in each coronal plane. The 3D flow fields were represented as isosurfaces, visualizing the internal geometry of the models with streamlines tangent to the velocity vectors identifying the path of the fluid. Volumetric flow rates for each time phase were calculated by integrating the flow through cross sections of each anastomotic model. Analysis of the flow patterns focused on the anastomotic regions prone to the development of intimal hyperplasia and graft failure as identified in the literature; the toe, floor, heel, and hood.

Results: Conventional end-to-side geometry resulted in uniform flow with a low angle of impingement on the recipient vessel floor. A small vortex at the anastomotic heel created minimal recirculation. The precuffed geometry resulted in a large recirculation vortex of chaotic, low flow that increased throughout the pulsatile cycle. Regions of low flow velocity were noted in a substantial portion of the precuffed anastomotic configuration. Flow separation distal to the toe occurred in both geometries, but was more apparent in the precuffed configuration. The patch model had flow characteristics similar to the straight end-to-side geometry.

Conclusion: Magnetic resonance velocimetry produces 3D, time varying velocity measurements with sufficient accuracy and resolution to analyze hemodynamics in anastomotic geometries. Flow structures in different graft configurations were effectively captured with marked differences noted between standard and precuffed anastomotic geometries. The findings support a conventional end-to-side anastomosis with a low incidence angle using a straight graft as producing favorable hemodynamics as compared to a cuffed configuration. The vein patch configuration closely approximates the conventional, straight anastomotic pattern. We believe the MRV technique has been sufficiently developed to warrant additional in vitro and in vivo studies providing insight into hemodynamic implications for the development of optimal prosthetic graft performance.