Optimization of Anastomotic Geometry for Vascular Access Fistula

Abstract

Arteriovenous fistula (AVF) is one type of vascular access which is a surgically created vein used to remove and return blood during hemodialysis [1]. It is a long-term treatment for kidney failure. Although clinical treatment and technology have both achieved great improvements in recent years, the vascular access for hemodialysis still has significant early failure rates after the insertion of AVF in patients [2]. Studies have shown that stenosis in the vascular access circuit is the single major cause for access morbidity. Majority of efforts to understand the mechanisms of stenosis formation, and its prevention and management have largely focused on understanding and managing this complication based on the pathophysiology, tissue histology and molecular biology; however these efforts have not resulted in significant progress to date. We believe that the major impact in this area will come from continued and accurate understanding of the hemodynamics, and by development of techniques of intervention to modulate factors such as flow rates, pressures and compliance of the circuit. The goal of this paper is to study anastomotic models of AV access using Computational Fluid Dynamics (CFD) and optimize them to minimize the wall shear stress (WSS). In order to achieve this goal, the commercial CFD software FLUENT [3] is employed in conjunction with a single objective genetic algorithm [4]. Computations for two types of AVF currently in use in clinical practice are performed. AVF with 25° angle/3–4mm diameter and 90° angle/3–5mm diameter are selected to conduct the optimization. A single-objective genetic algorithm is employed in the optimization process and a k-kl-ω turbulence model is employed in CFD simulations; this model can accurately compute transitional/turbulent flows. In order to optimize for the same flow conditions, a fixed boundary condition is used during the optimization process. Computations for 16 to 20 generations of the selected AVFs are obtained from the genetic algorithm solver. The maximum WSS in the two AVFs considered are 6997.8 and 7750 dynes/cm2; however, the maximum WSS in the shape-optimized AVFs are reduced to 3511.2 and 4293.9 dynes/cm2 respectively, which have decreased by 49.82% and 44.59% respectively. Thus, the probability of the formation of stenosis in AVFs and early failure rates of vascular access are reduced by using the optimized AVFs.