A new scintigraphic method to characterize ultrafiltration in hollow fiber dialyzers

Abstract

Ultrafiltration and pressure profiles in hollow fiber dialyzers with different hydraulic permeabilities have been investigated with a new scintigraphic method. Radiolabelled albumin macroaggregates, used as a nondiffusible marker molecule, were added to the blood in an in vitro circuit and circulated through cuprophan and polysulphon dialyzers. Since the marker molecule was too big to cross the dialysis membrane, its changes in concentration were assumed to occur in response to the variation of the blood water content (filtration or back-filtration). These changes in concentration, recorded by a gamma camera, were evaluated to establish the cumulative values of filtration and back-filtration and their relevant profiles along the length of the dialyzer. The achieved data were compared with the experimental values of ultrafiltration empirically measured and with the theoretical values predicted by a classic linear method. Two conditions were analyzed: A) the minimal filtration rate necessary to avoid back-filtration (critical filtration); and B) the condition of zero net filtration in which filtration equals back-filtration. The nuclear method proved to be extremely precise in predicting the ultrafiltration values and significantly more precise than the linear method, especially for the highly permeable dialyzer. The reason for that probably depends on the non-linear pressure and ultrafiltration profile observed with the scintigraphic pattern of the dialyzer. Viscosity changes and local variations in blood flow may in fact interfere with the pressure drop inside the hollow fibers and result in such a complex behavior. The other interesting aspect of this method is the possibility of accurate measurement of the amount of back-filtration that wouldn't be possible with simple calculations. In conclusion, the complex nature of the phenomena regulating the water fluxes in hollow fiber dialyzers requires more complex calculation than a simple linear model to achieve an accurate range of predictability.