Push/Pull Hemodiafiltration: Technical Aspects and Clinical Effectiveness

The use of a high-flux membrane, which eliminates larger molecular weight solutes with better biocompatibility, has steadily increased since the discovery of beta-2 microglobulin (beta 2m) amyloidosis in 1985. The long-term effects of a dialyzer membrane on morbidity and mortality are not completely understood. To examine the membrane effect as a factor of carpal tunnel syndrome onset and mortality, multivariate Cox regression analysis with time-dependent covariate was conducted on 819 patients from March 1968 to November 1994 at a single center. Two hundred and forty-eight of the patients were either switched from the conventional to high-flux membrane or treated only with a high-flux membrane. Fifty-one patients underwent a CTS operation and 206 died. Membrane status (on high-flux or on conventional) was considered as time-dependent covariate and risk was adjusted for age, gender, type of renal disease and calendar year of dialysis initiation. The relative risk of CTS was reduced to 0.503 (P < 0.05) and mortality 0.613 (P < 0.05) by dialysis on the high-flux membrane, compared to the conventional membrane. Serial measurements of beta 2m indicated significantly lower beta 2m to persist in patients on the high-flux membrane. The high-flux membrane decreased the risk of morbidity and mortality substantially. Larger molecule elimination was shown important not only for preventing beta 2m amyloidosis, but for prolonging survival of dialysis patients as well.

An effective therapeutic means to remove relatively large polypeptide uremic toxins seems to be a hemofiltration (HF) or hemodiafiltration (HDF) employing a larger-pore membrane, that is, a protein-permeable membrane. With either method, however, a significant amount of albumin will be lost into the ultrafiltrate or dialysate. Now, repetition of alternate short fore- and backfiltrations may prevent the development of the ultrafiltration-induced higher albumin concentration on the membrane surface (protein concentration polarization), where a single forefiltration time is shorter than the time needed for completion of protein concentration polarization. Since the albumin concentration on the protein-permeable membrane surface will be one of the determinants of albumin loss by convection, such HDF treatment may reduce protein loss into the dialysate. To examine this assumption, we alternately repeated short and rapid fore- and backfiltrations (push/pull HDF) through a protein-permeable membrane, each less than 1 second in duration and at each filtration volume of 15 ml, where a pyrogen-free dialysate was supplied. The present results indicated that the albumin amount lost by push/pull HDF was approximately one-third of that by conventional HDF. Nevertheless, the reduction rates of beta 2-microglobulin and myoglobin were significantly greater by push/pull HDF than by conventional HDF.