Continuous arteriovenous hemofiltration: an in vitro simulation and mathematical model.

In vitro and mathematical models of continuous arteriovenous hemofiltration (CAVH) have been developed. Human erythrocytes resuspended in normal saline containing 5% bovine albumin were used to perfuse the circuit from a gravity driven pressure source. Membrane hydraulic permeability was observed to decline from 31.2 x 10(-5) +/- 11.9 x 10(-5) cm/(min.mm Hg) before use to 12.3 x 10(-5) +/- 3.3 x 10(-5) (mean +/- SD) after use. This fall occurred during the first one to two hours whether perfused with blood or 5% albumin alone. Pressure-flow relationships of each circuit component, measured with 40% sucrose as a calibration medium, conformed to Poiseuille's equation. Use of high resistance blood access on the venous end of the circuit resulted in a low blood flow rate and high filtration fraction. The same access, when placed on the arterial end, produced both low blood flow rate and low filtration fraction. These results were a consequence of pressure distribution within the circuit as demonstrated by measurements of perfusion, prefilter, and postfilter pressures. The importance of negative pressure applied to the filter chamber in order to maintain favorable Starling forces, when the system was operated with a small bore arterial access, was demonstrated by similar methods. Enhancement of urea clearance by predilution was verified. Model simulations suggest that predilution will be of less benefit or even detrimental for other solutes which fail to distribute across the erythrocyte membrane. Comparison of results with predictions of a mathematical model demonstrated good agreement, but with some tendency to overestimate filtrate production. The latter was attributed to neglect of concentration polarization of plasma proteins in model development.