Does the end-to-end venous anastomosis offer a functional advantage over the end-to-side venous anastomosis in high-output arteriovenous grafts?

This study explores the hemodynamics, mechanics, and biologic response of end-to-end versus end-to-side venous anastomoses in a canine arteriovenous graft model. Femoral polytetrafluoroethylene grafts were implanted bilaterally in a paired fashion (n = 22). Detailed local hemodynamic measurements were made by use of color Doppler ultrasound imaging at 1, 4, 8, and 12 weeks after implant. Measurements included volumetric flow rate and Doppler-derived spectral window (percent window) as a measure of turbulence. Amplitude and velocity of vessel wall movement were also measured. Volume of perivascular tissue vibration quantitated kinetic energy transfer through the vessel wall. Volumetric flow rate (end to end, 1013 +/- 70 ml/min; end to side, 1015 +/- 72 ml/min), percent window (end to end, 6.6% +/- 0.6%, end to side, 5.6% +/- 0.4%) and volume of perivascular tissue vibration (end to end, 19.6 +/- 1.2 ml, end to side, 16.3 +/- 1.8 ml) were statistically equivalent in the two graft types (end to end vs end to side p greater than 0.05). Both graft types developed venous intimal-medial thickening of a similar magnitude: end to end, 0.35 +/- 0.05 mm, end to side, 0.43 +/- 0.09 mm, normal vein 0.070 +/- 0.004 mm (analysis of variance [ANOVA] p less than 0.001, p less than 0.01 for end to end or end to side vs control, end to end vs end to side p greater than 0.05 by Student-Newman-Keuls test). The best correlations with venous intimal-medial thickening were obtained from inverse percent window (r = 0.84, p less than 0.001) and volume of perivascular tissue vibration (r = 0.68, p less than 0.001). In the end to end configuration the relative amplitude of venous wall movement decreased, and the relative velocity of wall motion increased over time. We conclude that in the circumstances of this high flow arteriovenous graft model the end-to-end venous anastomosis does not significantly differ from the end-to-side venous anastomosis in terms of flow stability, turbulence, or kinetic energy transfer. The magnitude of the hyperplastic response is statistically equivalent for the two anastomotic types, but the pattern is somewhat different, possibly providing evidence for differences in stress distribution. Differences in the relative amplitude and velocity of vessel wall movement suggest that anastomotic geometry may affect the way in which kinetic energy is dissipated at the graft/vessel interface.