TMP revisited: the importance of plasma colloid osmotic pressure in high-flux dialysers

Daniel Schneditz
2011 Nephrology, Dialysis and Transplantation  
With high-flux dialysers and volume-controlled machines, transmembrane pressure (TMP) is no longer used to prescribe and control the adjustment of patient volume but still used to monitor filter function and membrane integrity. A drop in TMP at given operation characteristics indicates a leak or even a rupture in the membrane, while an increase in TMP indicates a loss in hydraulic permeability, most likely because of gradual clogging of the filter by formation of secondary membranes or
more » ... mbranes or obstruction of individual capillary fibers. Monitoring dialyser function is of special interest in different forms of haemofiltration and haemodiafiltration, where ultrafiltration rates are allowed to reach very high values exposing all components of the process to substantial strain. High ultrafiltration rates and excessive filtration fractions bear the risk of cell damage or activation [1] and filter failure [2] . Thus, there is increased interest to monitor driving pressures and to obtain objective criteria for safe operation of such treatment modes [3, 4] . In this issue of the journal, Ficheux et al. [5] describe that the ratio of ultrafiltration rate (Q uf ) to TMP in high-flux dialysers has a distinct maximum within a narrow range of ultrafiltration rates for a wide range of operating conditions. The Q uf /TMP ratio is interpreted as the ultrafiltration coefficient K uf of the dialyser. This maximum is easily identified during the actual treatment and the authors suggest that the ultrafiltration rate and/or the TMP associated with this maximum indicates a region for optimal ultrafiltration treatment. Indeed, this could be an interesting approach to improve convective techniques. The relationships between ultrafiltration, treatment (blood flows, dialysate flows and pressures), filter (flow resistance, filter hydraulic permeability and fibre geometry) and patient parameters (blood composition with regard to haematocrit and plasma protein concentration) are rather complex and the interested reader is referred to more detailed discussions of this subject [6-8]. These relationships have been examined using mathematical models [9] [10] [11] [12] [13] . We therefore assumed that such an approach could be helpful to better understand the Q uf /TMP maximum observed by Ficheux et al. [5]. The model to address this question was adapted from models published elsewhere [9,13]. The source code of this model that does not account for effects caused by concentration polarization and/or deposition of proteins is provided in the supplementary material and can be used with Berkeley-Madonna, a numerical integrator available from the Internet (Berkeley Madonna X, version 8.3.15, http://www.berkeleymadonna.com). For a given set of parameters comparable to those described in the technical note, the dialysate inflow pressure was varied to cover a TMP range from 0 to 300 mmHg, a common upper limit in convective techniques [3, 4] . The nominal filtration coefficient K uf of the dialyser was 100 mL/h/mmHg, as provided by the manufacturer, and assumed to represent the hydraulic permeability times the active membrane surface area of the dialyser. As expected, Q uf steadily increased with increasing TMP, but the increase was less pronounced for TMP exceeding 100 mmHg (Figure 1a, full line) . As a consequence, the slope (ΔQ uf / ΔTMP) of the Q uf to TMP relationship was not constant but continuously decreased from ∼83 to 7 mL/h/mmHg (Figure 1b, full line) . Interestingly, in spite of a constant K uf assumed in the model, the ratio of Q uf /TMP plotted versus Q uf (Figure 2a , full line) or TMP (Figure 2b , full line) showed a shape and a maximum comparable to that presented in the technical note [5] . The ratio of Q uf /TMP reached maximum values of 41 mL/h/mmHg at a Q uf of 90 mL/min and at a TMP of 102 mmHg. Thus, the experimental observation presented in the technical note was successfully replicated by the mathematical model describing internal filtration and backfiltration in a high-flux dialyser. Contrary to expectations, the simulations showed that there was a large discrepancy between the constant K uf of the dialyser (100 mL/h/mmHg), the slope of the Q uf to TMP relationship (Figure 1b , full line) as well as the ratio of Q uf /TMP (Figure 2, full lines) . A comparison of slopes ( Figure 1b) and ratios (Figure 2b) helps to resolve part of the problem. The slope of a functional relationship between two variables (ΔQ uf /ΔTMP in this case) is not identical to the ratio of these variables (Q uf /TMP). Slope and ratio are only identical for the linear case and in the absence of an intercept. The actual relationship between Q uf and TMP is nonlinear with a non-
doi:10.1093/ndt/gfq784 pmid:21273239 fatcat:ztdp2cvouvbhbp3c43gktqak44