Determinants of haemodialyser performance and the potential effect on clinical outcome

Size: px

Start display at page:

Download "Determinants of haemodialyser performance and the potential effect on clinical outcome"

Blaze Hill
6 years ago
Views:

1 Nephrol Dial Transplant 2001) 16 wsuppl 5x: 56±60 Determinants of haemodialyser performance and the potential effect on clinical outcome William R. Clark 1,2 and Claudio Ronco 3 1 Renal Division, Baxter Healthcare Corp., McGawPark, IL, 2 Nephrology Division, Indiana University School of Medicine, Indianapolis, IN, USA and 3 Nephrology Division, Opsedale San Bortolo, Vicenza, Italy Abstract The performance of a dialyser is determined by several factors. Many of these factors relate to the dialyser membrane, including mean pore size, pore size distribution, wall thickness, surface area, and adsorptivity. In this article, several of these factors are reviewed. The potential impact of these factors on the clinical outcome of chronic haemodialysis patients is also discussed. Keywords: convection; dialyser; ux; haemodialysis; membrane Fundamental properties of haemodialyser membranes A dialyser used for HD has a 2-fold purpose. First, its membrane acts as a semi-permeable barrier for the transfer of solutes primarily from blood to dialysate, but also in the opposite direction. Removal of low molecular weight uraemic toxins occurs almost exclusively by diffusion, while convection and adsorption generally assume an important role in the removal of larger compounds such as peptides and proteins. The second basis function of a haemodialyser is the ultra ltration of plasma water. Introduction When selecting a dialyser for a chronic haemodialysis HD) patient, a clinician must consider several factors, including performance, biocompatibility, water permeability ux), mode of sterilization, and cost. With respect to performance, the most logical manner to assess the solute removal capabilities of a dialyser is to use the depuration properties of the native kidney as a benchmark. The purpose of this article is to review the several important dialyser characteristics that have the greatest impact on its solute removal performance over a wide spectrum of uraemic toxins. Speci cally, mean pore size, pore size distribution, wall thickness, ux and adsorptivity are discussed. The potential impact of some of these factors on the clinical outcome of chronic HD patients is also highlighted. Correspondence and offprint requests to: William R. Clark, Hemodialysis Research Laboratory, Renal Division, Baxter Healthcare Corporation, Wishard HospitaluMyers Building D711, 1001 West 10th Street, Indianapolis, IN 46202, USA. Dialyser membrane water permeability A hollow ber dialyser membrane can be modelled as having straight cylindrical pores, all of the same radius r) and all with a directionality perpendicular to the owof blood and dialysate w1x. As discussed below, this model does not exactly replicate a clinical dialyser but is useful from a quantitative perspective nevertheless. The major determinants of plasma ultra ltrate owrate through the pores are the number of pores i.e. number per unit of membrane surface area), transmembrane pressure and pore size. However, this rate of ultra ltrate ow is actually dependent on the fourth-power of the pore radius i.e. r 4 ). As such, the membrane characteristic that most directly in uences water permeability is mean pore size. Water permeability has traditionally been used to categorize haemodialysers. In the past, this was largely driven by the existence of HD machines, both with and without ultra ltration control mechanisms, and the limitation that only dialysers of relatively low water permeability be used in conjunction with nonultra ltration control machines. However, the vast majority of machines in the market nowhave # 2001 European Renal Association±European Dialysis and Transplant Association

2 Haemodialyser performance ultra ltration control mechanisms so a more relevant dialyser classi cation scheme is one based on both water permeability and solute removal properties w2x. Such a scheme is proposed in Table 1. Although both conventional and high ef ciency classes are characterized by lowwater permeability, dialysers in the latter category have higher surface areas and are typically employed in high- owrate settings to maximize small solute clearances w3x. The larger mean pore sizes of the mid- ux and high- ux categories generally lead to enhanced removal of intermediate and large-sized uraemic compounds. Haemodialyser membrane diffusive permeability Membrane wall thickness is one important determinant of diffusive transport w4x. The relatively thin walled structure of cellulosic membranes usually 6±15 mm) is largely responsible for their particular suitability in the setting of diffusive HD. The other major determinant of dialyser membrane diffusive transport is porosity, also known as pore density. Based on the cylindrical pore model described above, membrane porosity is directly proportional to both the number of pores and the square of the pore radius r 2 ). Therefore, the smaller dependence of membrane porosity on pore size, relative to the case of water permeability, implies a relatively greater importance of pore number in determining diffusive permeability. That the major determinants of ux r 4 ) and diffusive permeability number of pores, r 2 and wall thickness) differ so signi cantly implies that the two properties can be independent of one another for a particular haemodialysis membrane w5x. Such is the case for cellulosic high-ef ciency dialysers, which typically have very high diffusive permeability values for small solutes but lowwater permeability. Pore size distribution effects on dialyser membrane properties A membrane represented by the cylindrical pore model described above deviates from an actual membrane used for clinical HD in that the latter actually has a distribution of pore sizes. Ronco et al. w6x have recently discussed the manner in which pore size distribution may differ between HD membranes and the manner in which this distribution in uences a membrane's sieving properties Figure 1). The membrane represented by curve A has a large number of relatively small pores, while the membrane represented by curve B has a large number of relatively large pores. Based on the relatively narrowpore size distributions, the solute sieving coef cient vs molecular weight pro le for both membranes has the desirable sharp cut-off, similar to that of the native kidney. However, the molecular weight cut-off for membrane A ;10 kda) is consistent with a high-ef ciency membrane, while that of membrane B ;60 kda) is consistent with a high- ux membrane. In addition, primarily due to the large number of pores, both membranes would be expected to demonstrate favourable diffusive transport properties. On the other hand, membrane C exhibits a pore size distribution that is unfavourable from both a diffusive transport and sieving perspective. The relatively small number of pores accounts for the poor diffusive properties. In addition, the broad distribution of pores explains not only the `early' drop-off in sieving coef cient at relatively lowmolecular weight, but also the `tail' effect at high molecular weight. This latter phenomenon is highly undesirable as it may lead to unacceptably high albumin losses across the membrane. Dialyser membrane properties in uencing non-diffusive removal Larger molecular weight uraemic toxins, particularly peptides and proteins, are cleared inef ciently by diffusion during HD. Solutes of this type may be removed primarily by convection or adsorption, or a combination thereof w7±9x. Ultra ltration rate and the dialyser membrane's sieving properties are the two most important determinants of convective solute removal w10x. As noted by several investigators w11,12x, typical clinical use of a high- ux dialyser dictates that the net ultra ltration rate i.e. the rate of patient weight loss due to due net plasma water removal) is not equivalent to the absolute ultra ltration rate in the dialyser. In fact, the rate of ltration from blood to dialysate in the proximal portion of the dialyser and the rate of `back ltration' from dialysate to blood in the distal part of the dialyser is signi cantly higher than the net ultra ltration rate. This internal ltration mechanism, physiologically analogous to Starling's ow, contributes signi cantly to overall convective solute removal by a high- ux dialyser. In fact, attempts 57 Table 1. Dialyser classi cation Class Surface area K UF mluhummhg) Urea clearance b2m clearance Conventional -1.5 m 2-12 Moderate Negligible High ef ciency )1.5 m 2-12 High Negligible Mid- ux Variable Variable Moderate High- ux Variable )30 Variable High

3 58 W. R. Clark and C. Ronco to enhance quantitatively the removal of middle molecules by this mechanism, either by manipulations in blood compartment or dialysate-side pressure, have recently been described w13±15x. Adsorption membrane binding) is another mechanism by which hydrophobic compounds like peptides and proteins may be removed during HD. Although adsorption during HD is a relatively poorly understood phenomenon, certain membrane characteristics play an important role. First, adsorption primarily occurs within the pore structure of the membrane rather than at the nominal surface contacting the blood only w16x. Therefore, the open pore structure of high- ux membranes affords more adsorptive potential than do low- ux counterparts. Secondly, synthetic membranes, many of which are fundamentally hydrophobic, generally are much more adsorptive than hydrophilic cellulosic membranes w17x. Effect of dialyser membrane ux on mortality in chronic haemodialysis patients Several recent retrospective studies have assessed the relationship between membrane type and mortality in the chronic HD population. However, in some of these studies, membrane composition and ux have been covariates. For example, using the United States Renal Data Systems USRDS) database, Hakim et al. w18x retrospectively analysed the effect of membrane on mortality in ;2400 patients. These patients were a randomly selected cohort of HD patients who had been on HD for at least 1 year and were receiving treatment in For analysis purposes, dialysis membranes were categorized according only to polymeric composition into unsubstituted cellulosic, substituted cellulosic `semisynthetic'), and synthetic classes. Therefore, the unsubstituted cellulosic class was homogeneous with respect to ux low permeability), while the ux of the dialysers in the other two groups could theoretically vary. The Cox model was used to determine the relative risk of death for each membrane group, with the unsubstituted cellulosic group having a reference value of 1.0. All results were adjusted for a battery of clinical parameters, such as demographic characteristics, cause of end-stage renal disease, major comorbidities and serum albumin concentration. Adjustments were also made sequentially for the delivered dose of dialysis urea KtuV) and the region of the country in which a patient was treated to account for geographic practice variations. The analysis in which adjustment was made for delivered dialysis dose showed that the relative risk of death was decreased signi cantly by ;25% with the use of both substituted cellulosic and synthetic membranes. After the added adjustment for geographic region, the decrease in the relative risk of death for the substituted cellulosic and synthetic groups was still 20%. This investigative group corroborated these ndings in a more recent analysis of patients dialysed during the same time period w19x. At the time these analyses were conducted, the vast majority of substituted cellulosic dialysers used in the United States were of relatively low water permeability K UF -12 mluhummhg), with negligible usage of high- ux cellulosic dialysers. On the other hand, the overwhelming majority of synthetic dialysers used at the time of the study were high- ux predominantly polysulphone) with minimal usage of low- ux lters. Speci cally, in the latter analysis w19x, relatively lowpermeability cellulose acetate accounted for ;80% of modi ed cellulosic usage, while high- ux cellulosic i.e. cellulose triacetate) usage represented -5% of cellulosic usage. Polysulfone represented )95% of the synthetic membrane usage, with the vast majority of utilization falling in the high- ux domain. Therefore, in the unsubstituted cellulosic vs substituted cellulosic comparison, dialyser ux was relatively invariant, such that mortality differences could be attributed reasonably con dently to differences in the membrane polymer i.e. biocompatibility). For the unsubstituted cellulose vs synthetic comparison, membrane ux and biocompatibility were covariates such that the Fig. 1. Pore size distribution and solute sieving coef cient pro les for three hypothetical dialysis membranes. The left diagram shows the number of pores as a function of mean pore size, while the right diagram shows the relationship between solute sieving coef cient and molecular weight. Reprinted with permission from Ronco et al. w6x.

4 Haemodialyser performance survival advantage conferred by the use of synthetic membranes may have been related to either or both) of these parameters. The potential role of membrane permeability rather than membrane composition) was explored in a recent investigation in which the USRDS database from the same time period was employed. Using middle molecule clearance as a surrogate for dialyser ux, Leypoldt et al. w20x determined vitamin B 12 clearances for 19 different dialysers used in the treatment of the USRDS database patients. These clearances were estimated from the combination of blood and dialysate owrates for a speci c dialyser in conjunction with that dialyser's manufacturer-derived vitamin B 12 KoA. This analysis suggested a signi cant inverse relationship between mortality and vitamin B 12 clearance. In fact, a 10% increase in the total vitamin B 12 clearance per week) was associated with a signi cant 5% decrease in the risk of death. By comparison, a 0.1 increase in delivered urea KtuV per treatment) was associated with a 7% increase in the risk of death. These recent USRDS ndings w18±20x suggest that, relative to lowpermeability unsubstituted cellulosic membranes, both enhanced biocompatibility provided by modi ed cellulosic and synthetic membranes) and higher permeability improve survival. Over a study period extending from 1968 to 1994, Koda et al. w21x retrospectively assessed 819 patients treated either with a low- ux unsubstituted cellulosic dialyser or a high- ux dialyser. The latter group consisted of a number of different membranes, including PMMA, polysulphone, PAN and cellulose triacetate. For the last 6 years of the study, a dialysate ultra lter designed to reduce the concentration of endotoxin and other bacterial products was used. The cellulosic group ns571) consisted of patients treated solely with that type of membrane, while the high- ux group ns248) included patients either treated solely with that type of membrane ns63) or that had switched to that type of membrane ns185). Patients treated either solely with or having switched to high- ux dialysis had an ;40% reduction in the relative risk of death compared with patients receiving conventional HD P-0.05). In the high- ux group, membrane utilization was as follows: PMMA, 32%; cellulose triacetate, 29%; PAN, 17%; and polysulphone, 17%. Finally, Woods and Nandakumar w22x recently attempted to isolate the effect of ux from membrane material in a retrospective analysis of 715 patients. These investigators compared the outcome of a group consisting of patients treated exclusively with low- ux polysulphone dialysers with a group treated for at least 3 months with a high- ux polysulphone dialyser. The general approach to dialyser prescription in this Singapore dialysis centre was to employ high- ux HD preferentially in patients with relatively greater comorbidity, including diabetic and elderly patients. Despite these differences in comorbidity unfavourable to the high- ux group, a Kaplan±Meier analysis suggested a signi cantly higher 5-year survival in this group Fig. 2. Kaplan±Meier analysis of the effect of membrane ux on survival in chronic haemodialysis patients. Reprinted with permission from Woods et al. w22x. Figure 2). Interestingly, the signi cant divergence between the two curves did not appear to occur until after ;4 years of treatment. Cox proportional analysis, excluding diabetic patients, indicated a signi- cant 65% decrease in the risk of death in the high- ux group relative to the low- ux cohort, while the same analysis including diabetics approached statistical signi cance Ps0.07). Summary An overviewof membrane characteristics in uencing dialyser performance has been provided. In addition, recent studies attempting to establish a relationship between membrane characteristics and survival of chronic HD patients have been reviewed. Emerging evidence from retrospective studies suggests that membrane ux is an important determinant of outcome in this patient population. A problem with some of these studies is the co-variation of membrane ux and composition. Data from prospective studies, such as those being performed currently in both Europe w23x and the US w24x, should provide further elucidation. References 1. Lysaght MJ. Hemodialysis membranes in transition. Contrib Nephrol 1998; 61: 1±17 2. Cheung A, Agodoa L, Daugirdas J et al. Effects of hemodialyzer reuse on clearances of urea and b2-microglobulin. J Am Soc Nephrol 1999; 10: 117± Keshaviah P, Luehmann D, Ilstrup K, Collins A. Technical requirements for rapid high ef ciency therapies. Artif Organs 1986; 11: 189± Lysaght MJ. Evolution of hemodialysis membranes. Contrib Nephrol 1995; 113: 1±10 5. Clark WR, Hamburger RJ, Lysaght MJ. Effect of membrane composition and structure on performance and biocompatibility in hemodialysis. Kidney Int 1999; 56: 2005± Ronco C, Ballestri M, Gappelli G. Dialysis membranes in convective treatments. Nephrol Dial Transplant 2000; 15 wsuppl 2x: 31±36 7. Klinke B, Rockel A, Abdelhamid S, Fiegel P, Walb D. Transmembrane transport and adsorption of beta2-microglobulin during hemodialysis using polysulfone, polyacrylonitrile, 59

5 60 W. R. Clark and C. Ronco polymethylmethacrylate, and cuprammonium rayon membranes. Int J Artif Organs 1989; 12: 697± Jorstad S, Smeby L, Balstad T, Wideroe J. Removal, generation and adsorption of beta2-microglobulin during hemo ltration with ve different membranes. Blood Purif 1988; 6: 96± Ronco C, Heifetz A, Fox K et al. Beta2-microglobulin removal by synthetic dialysis membranes: Mechanisms and kinetics of the molecule. Int J Artif Organs 1997; 20: 136± Henderson L. Biophysics of ultra ltration and hemo ltration. In: Jacobs C, Kjellstrand C, Koch K, Winchester J, eds. 4th edn. Kluwer Academic Publishers, Dortdrecht; 1996: 114± Ofsthun NJ, Zydney AL. Importance of convection in arti cial kidney treatment. Contrib Nephrol 1994; 108: 53± Ronco C. Back ltration: A controversial issue in modern dialysis. Int J Artif Organs 1988; 11: 69± Dellanna F, Wuepper A, Baldamus CA. Internal ltrationð advantage in haemodialysis? Nephrol Dial Transplant 1996; 11 wsuppl 2x: 83± Ronco C, Brendolan A, Lupi A, Metry G, Levin NW. Effects of reduced inner diameter of hollow bers in hemodialyzers. Kidney Int 2000; 58: 809± Ronco C, Brendolan A, Lupi A, Bettini MC, La Greca G. Enhancement of convective transport by internal ltration in a modi ed experimental dialyzer. Kidney Int 1998; 54: 979± Clark WR, Macias WL, Molitoris BA, Wang NKL. b 2 - microglobulin membrane adsorption: equilibrium and kinetic characterization. Kidney Int 1994; 46: 1140± Clark WR, Macias WL, Molitoris BA, Wang NHL. Plasma protein adsorption to highly permeable hemodialysis membranes. Kidney Int 1995; 48: 481± Hakim R, Held P, Stannard D et al. Effect of the dialysis membrane on mortality of chronic hemodialysis patients. Kidney Int 1996; 50: 566± Bloembergen W, Hakim R, Stannard D et al. Relationship of dialysis membrane to cause-speci c mortality. Am J Kidney Dis 1999; 33: 1± Leypoldt JK, Cheung A, Carroll C et al. Effect of dialysis membranes and middle molecule removal on chronic hemodialysis patient survival. Am J Kidney Dis 1999; 33: 349± Koda Y, Nishi S, Miyazaki S et al. Switch from conventional to high- ux membrane reduces the risk of carpal tunnel syndrome and mortality of hemodialysis patients. Kidney Int 1997; 52: 1096± Woods HF, Nandakumar M. Improved outcome for haemodialysis patients treated with high- ux membranes. Nephrol Dial Transplant 2000; 15 wsuppl 1x: 36± Locatelli F, Hannedouche T, Jacobson S et al. The effect of membrane permeability on ESRD: Design of a prospective randomized multicentre trial. J Nephrol 1999; 12: 85± Eknoyan G, Levey A, Beck G et al. The Hemodialysis HEMO) Study: Rationale for selection of interventions. Semin Dial 1996; 9: 24±33

HDx THERAPY. Enabled by. Making possible personal.

HDx THERAPY. Enabled by. Making possible personal. HDx THERAPY Enabled by Making possible personal. THE NEXT HORIZON IN DIALYSIS IS CLOSER THAN YOU THINK PHOSPHATE UREA HDx BY THERANOVA EXPANDS YOUR RENAL POSSIBILITIES The new HDx therapy (expanded HD)