Background

Despite technological advances in dialysis equipment and modalities, survival, morbidity, and quality of life of hemodialysis patients are still severely affected by acute intradialytic and long-term complications, possibly related to the treatment itself. Convective treatments, such as high-flux hemodialysis, hemodiafiltration, and hemofiltration are increasingly suggested as further improvements over standard diffusive hemodialysis. The membranes used for these techniques are high-flux semisynthetic and synthetic membranes. Characteristics of these membranes are high permeability, which allows convective removal of water and electrolytes and higher clearance of middle and large molecular weight solutes, and high biocompatibility, which minimizes the "inflammatory response" secondary to interactions between blood and the artificial material of the hemodialysis system.

Methods

With the specific aim of verifying the superiority of convective treatments in reducing morbidity and mortality, we performed a review of the published literature.

Results

Some epidemiological studies suggest that convective treatments reduce morbidity and mortality among dialysis patients. However, the results of the published prospective randomized controlled trials are conflicting. Moreover, since convective treatments are usually performed with synthetic biocompatible membranes, it is hard to separate the effect of convection from the effect of biocompatibility.

Conclusions

To finally assess the effect of high-flux membranes on morbidity and mortality, the results of two randomized, controlled clinical trials (HEMO study and MPO study) specifically designed with this aim are needed.

CARDIOVASCULAR STABILITY

In spite of technological advances, such as monitors with ultrafiltration control and bicarbonate dialysis, cardiovascular instability during the hemodialysis session is still an important clinical problem affecting a large percentage of hemodialysis patients. Although its pathogenesis is multifactorial, a role of membrane material and convection has been suggested.

Chanard et al³ reported that the incidence of hypotension, vomiting, cramps, and headaches was lower in patients dialyzed with high-flux membranes (AN69) compared to those in the control group (cellulose membrane). However, there was a lack of randomization and patients in the control group did not use machines with automatically controlled ultrafiltration.

In a prospective randomized multicentric trial, Locatelli at al⁴ compared treatments which differed both in terms of biocompatibility and permeability of membranes. During the 24 months follow-up, no significant difference in treatment tolerance and cardiovascular stability was demonstrated between the four treatment groups. As stressed in the paper, it is likely that significant differences in cardiovascular stability were not demonstrated because the incidence of intradialytic hypotension in the population as a whole was much lower than expected. In two prospective non-randomized studies^5,6 Altieri et al showed (although with some methodological drawbacks) a stabilizing haemodinamic effect of on-line predilution hemofiltration compared to ultrapure high-flux dialysis.

Taken together, these results do not support the hypothesis that convection and/or membrane biocompatibility decrease acute intradialytic clinical symptoms, at least in these study populations.

NUTRITIONAL ASPECTS

Malnutrition is a common feature among dialysis patients and is commonly associated with decreased body weight, depleted energy stores (fat tissue), and loss of somatic protein (low muscle mass). It has been stated that low plasma levels of serum albumin, transferrin, prealbumin, and other visceral proteins are clinical expressions of malnutrition. Furthermore, a strong association between low serum albumin and increased morbidity and mortality has been established. If protein malnutrition is secondary to increased protein catabolism, decreased protein anabolism, or a combination of both is still a matter of discussion. A role of membrane material as a whole (biocompatibility and/or convection) on protein malnutrition has been suggested by Gutierrez et al⁷. However, it must be underlined that Gutierrez findings on healthy volunteers were not confirmed in uremic patients (abstract; Lim et al, J Am Soc Nephrol 1:366, 1990).

It has been suggested that biocompatible high-flux membranes can positively affect the relationship between dietary protein intake (measured as PCR) and dialysis dose (expressed as Kt/V)⁹, i.e., that hemodialysis with a biocompatible high-flux membrane results in higher protein intake at every level of Kt/V. However, it must be noticed that these studies present major methodological drawbacks.

Locatelli et al⁴ did not demonstrate an influence of dialysis membrane or convection on any of the variables related to the nutritional status (body weight, serum albumin, serum transferrin, subscapular skinfold, triceps skinfold, mid-arm circumference, plasma cholesterol, or triglycerides). Similarly, a correlation between Kt/V and protein catabolic rate was not found. One possible reason for this lack of correlation could be the high levels of Kt/V reached in this study. Given that the relationship between PCR and Kt/V is curvilinear, the study participants probably fall into the plateau region, where the correlation between efficiency and nutritional status is lost. Therefore, in order to provide adequate statistical power, further trials, involving sicker patients with high prevalence of malnutrition, should be designed.

ANEMIA

Anemia in hemodialysis patients is a complex syndrome and many factors other than absolute or relative erythropoietin deficiency may contribute in its pathogenesis. Among these, bone marrow suppression, probably induced by the retention of toxic metabolites, may be present. Microbiological and pyrogenic contamination of dialysate and secondary enhanced production of cytokines also play a role in causing or aggravating anemia. Kobayashi et al¹⁰ reported the positive clinical results obtained in eight hemodialysis patients treated with a large-pore membrane (BK-F polymethylmethacrylate). This study presented several important limitations: lack of randomization, small sample size, and absence of a control group. Similarly, Villaverde et al¹¹ demonstrated that the switch from cellulosic to high-flux polysulphone membrane, in the absence of changes in the dialysis dose, improved the response to rh-EPO by about 14% in 31 hemodialysis patients. However, a number of drawbacks do not allow to draw definitive conclusions from this study either. Similar considerations apply to the study of Kawano et al¹², who reported their experience on 10 hemodialysis patients treated with a high-flux dialyzer (BK-F). In order to test the same hypothesis with a better designed study, Locatelli et al¹³ performed a multicentric, controlled, and randomized trial. To verify whether hemodialysis with high-flux membrane (BK-F polymethylmethacrylate) improves anemia in comparison with conventional hemodialysis with cellulose membrane, 84 patients were studied. An increase in hemoglobin levels was observed in the population as a whole, but this trend was not significantly different between the conventional and experimental group. In the experimental group, the tendency of hemoglobin levels to increase was present at each month during the follow up, possibly indicating an insufficient length of the observation period. It has recently been suggested that on-line treatments may have a stronger effect on anemia compared to conventional treatments or standard HDF techniques. These techniques may be more effective in achieving higher hematocrit levels by means of two mechanisms: higher removal of small, medium, and large molecules and reduced microbiological and pyrogenic contamination of the dialysate. Maduell et al¹⁴ followed for 1 year 37 patients after they had been switched from conventional HDF to on-line HDF. A significant increase in hemoglobin and hematocrit were observed during the on-line HDF period. This allowed a significant decrease in the rh-EPO doses. It should be noted that the patients also experienced an improvement in dialysis dose (with a 15% increase in Kt/V) during the on-line HDF period. Grillo et al (abstract; Nephrol Dial Transplant 14:207, 1999) studied 31 patients, who had been on on-line HDF for at least 9 months. They found a significant increase in hemoglobin levels and a consequent reduction in rh-EPO consumption (not statistically significant). Kt/V remained constant during the course of the study. Other studies could not confirm these observations. Ward et al¹⁵ prospectively compared two convective techniques (on-line HDF and high-flux hemodialysis) in 44 patients, who were followed for one year. Although the control of anemia was not a primary outcome, there was no significant change in hematocrit or hemoglobin over the course of the study. The average weekly dose of rh-EPO slightly increased, but this modification was independent of the dialysis technique. Wizeman et al¹⁶ also failed to confirm the possible effect of convection (on-line HDF) on the correction of anemia. In this prospective controlled study, 44 patients were randomized to undergo either low-flux hemodialysis or on-line HDF. To eliminate confounding factors, low-molecular efficacy (Kt/V = 1.8), treatment duration (4.5 h), and membrane (polysulphone) were matched. It has to be noted that the two groups differed not only for convection but also for levels of biocompatibility. After 24 months of follow-up, hematocrit levels and rh-EPO dose did not differ between the two groups. Given that the same ultrapure dialysate was used in both groups, it is possible to postulate that the main factor accounting for the effect of on-line HDF on anemia correction may be, in addition to the increased dialysis dose, a reduced inflammatory stimulus from the dialysate. Wizemann et al¹⁶ found a significant decrease in CRP during HDF in contrast to low-flux dialysis. However, as reported above, this acute effect of on-line HDF on CRP needs further elucidation. Confirming the importance of dialysate sterility on anemia correction, Sitter et al¹⁷ found a significant and sustained reduction of rh-EPO dose in patients switched from conventional bicarbonate hemodialysis with potentially microbiologically contaminated dialysate to a similar treatment modality using on-line-produced ultrapure dialysate. The switch also resulted in a lower bacterial contamination, with a significant decrease in CRP and IL-6 levels. In a multivariate analysis, IL-6 levels were shown to be strongly predictive of rh-EPO dose in both groups (treatment with conventional or ultrapure dialysate). Altogether, the possibility that at least on-line HDF/HF may achieve a better control of anemia is intriguing. However, results are conflicting, mainly because of differences in treatment modalities in control groups, small number of enrolled patients, and because anemia was not the primary outcome in any of the studies on on-line HDF/HF. Furthermore, on-line HDF often achieved higher dialysis dose than control treatments, further complicating the interpretation of these observations. The results of prospective randomized trials specifically designed to test this hypothesis are awaited.

DIALYSIS-RELATED AMYLOIDOSIS

In a retrospective analysis, Van Ypersele de Strihou et al¹⁸ demonstrated lower incidence of cystic bone lesions in patients treated with a synthetic membrane (AN69) compared to those treated with a cuprophane membrane.

Kuchle et al¹⁹ conducted a trial on 20 patients previously treated with low-flux cuprophane membranes who were randomized to continue the same hemodialysis regimen or to receive high-flux polysulfone membranes dialysis. After 6 years of follow up, no clinical signs of dialysis-associated amyloidosis were found in any of the 10 patients dialyzed with high-flux polysulfone membranes. On the contrary, eight out of 10 patients dialyzed with cuprophane membranes presented carpal tunnel syndrome and/or osteoarticular lesions. Using data from the Lombardy Registry, Locatelli et al²⁰ performed a historical prospective study of 6,440 patients who had started renal replacement therapy between 1983 and 1995. These authors found that the relative risk for carpal tunnel syndrome surgery (evaluated as a parameter of morbidity) was 44% lower for patients treated with hemodiafiltration or haemofiltration (RR = 0.56; 95% c.i. 0.34–0.92; P = 0.02).

In a study by Koda et al²¹, 248 patients were either switched from conventional to high-flux membrane or treated with a high-flux membrane only. Again, the relative risk of CTS resulted significantly reduced (0.503; P < 0.05).

On the other hand, in a retrospective analysis on 122 patients Bonomini et al²² did not demonstrate any difference in 2-microglobulin related morbidity between cellulose and synthetic membranes.

Using data from the Japanese dialysis patient registry, Nakai et al²³ performed a retrospective study on 1196 patients who were undergoing renal replacement therapy at the end of 1998 and who developed dialysis-related amyloidosis during 1999. The risk of developing dialysis-related amyloidosis (being the risk for conventional hemodialysis equal to 1) was found to be progressively lower for high-flux hemodialysis (0.489), off-line hemodiafiltration (0.117), and on-line hemodiafiltration (0.013).

Morbidity and mortality

The most relevant findings on the impact of dialysis membrane and flux on patients mortality are summarized in Figure 1. Chanard et al³ reported that the use of a high-flux biocompatible membrane (PAN) was related to positive effects on hospitalization index and number of hospitalization days compared to cuprophan membranes. Mortality did not differ between the two groups. Hornberger et al²⁴ performed a retrospective study comparing 146 patients on standard hemodialysis to 107 patients on high flux hemodialysis. The relative risk of mortality on high flux hemodialysis was 0.24 (95% confidence limits: 0.12–0.49). In a retrospective analysis of 715 patients, Woods and Nandakumar²⁵ identified two groups: 252 patients treated exclusively with low-flux polysulfone and 463 patients treated for at least 3 months and up to 5 years with high-flux polysulfone. The patients treated with high-flux polysulfone experienced lower mortality (21 vs. 36 per 1000 years; P < 0.01) and presented higher survival at 5 years (90 vs. 60%; P = 0.029). In a Cox proportional hazards model, high-flux dialysis was associated with a 70% reduction in death-risk when compared to low-flux.

Figure 1.

Impact of dialysis membrane and flux on patient mortality.

Full figure and legend (66K)

In a historical prospective study of 2410 patients from the United States Renal Data System (USRDS), Hakim et al²⁶ applied a Cox proportional hazards model to estimate the relative risk of mortality. After adjusting for dialysis dose, geographic distribution, and comorbidity, the relative risk of mortality in patients dialyzed with modified cellulose or synthetic membranes was 20% lower compared to patients treated with unsubstituted cellulose membranes. Analyzing a sample from the USRDS, Bloembergen et al²⁷ demonstrated that, after adjusting for dialysis dose and co-morbidity, all-cause mortality (RR 0.82; P = 0.002), and mortality due to infections (RR 0.69; P = 0.03) were significantly lower in patients dialyzed with synthetic or substituted cellulosic membranes compared to patients dialyzed with unmodified cellulosic membranes. Remarkably, no differences in mortality due to cardiovascular causes were observed. Koda et al²¹ reported that the use of high-flux membranes was associated with better survival of dialysis patients (relative risk for mortality = 0.613; P < 0.05). A recent retrospective study performed by Leypoldt et al²⁸ on a large data subset from the USRDS showed a 5% reduction in mortality (after correction for case mix, co-morbidity factors, and urea Kt/V) of the patients treated with membranes that increased the removal of vitamin B₁₂ by 10%. Given that the vitamin B₁₂ (molecular weight: 1355 Daltons) is considered a middle molecules marker, the authors concluded that the removal of middle (and small) molecules is independently related to the risk of mortality in dialysis patients.

In 6,440 patients treated in Lombardy between 1983 and 1995, Locatelli et al²⁰ found that the relative risk for mortality was not significantly lower when using convective treatments (RR = 0.90).

Port et al²⁹ performed a very large-sized retrospective study, evaluating 12,791 patients on hemodialysis from 1994 through 1995 with 1 to 2 years of follow-up. These authors reported that the relative risk for mortality was lower for synthetic membranes compared to other kinds of membranes (RR = 0.82, 95%, CI 0.72 to 0.93, P = 0.002). Among synthetic membranes, the relative risk for mortality was higher with low-flux than with high-flux membranes (RR = 1.24, 95%, CI 1.02 to 1.52, P = 0.04). These results suggest that both biocompatibility and flux membrane characteristics may play a role in improving survival of hemodialysis patients.

CONCLUSIONS

Several epidemiological studies suggest that convective treatments seem to reduce morbidity and mortality among dialysis patients. The results of the published prospective randomized controlled trials are conflicting. It is likely that the concept of dialysis adequacy has to be widened to include, besides urea kinetics, medium and high molecules removal as well as biocompatibility.

To assess the effect of dialysis membranes on mortality, two randomized, controlled clinical trials have been designed: the Hemodialysis (HEMO) study³⁰ and the Membrane Permeability and ESRD Patient Outcome (MPO) study³¹. The HEMO study began in the United States in 1995 and should be concluded by the end of 2001. Its 2 2 factorial design considers patient outcomes in relation to two different dialysis doses (an equilibrated Kt/V of 1.05 or 1.45) and two different levels of convection (low- or high-permeability dialysis membranes). The MPO study, which started in Europe in December 1998 and should be concluded by the end of 2005, is specifically designed to assess the effect of membrane permeability. Unlike the HEMO study, the MPO study only enrolled incident patients. Given the importance of this issue, the results of these two large trials are awaited with much interest.

References

References