Background

Biocompatibility and middle molecule clearance of hemodialysis (HD) membranes affect the survival, morbidity, and quality of life of uremic patients undergoing maintenance HD therapy. In particular, malnutrition [1] and dialysis-related amyloidosis due to the reference middle molecule β2-microglobulin (β2-M) [2] are the major long-term complications. A polysulfone (PS) dialyzer is the mainstay of HD treatment because of its high performance. However, TORAYLIGHT® NV dialyzer (NV; Toray Industries, Inc., Tokyo, Japan) is expected to have different characteristics from the conventional PS due to its newly developed hydrophilic-coated membrane [3]. We therefore designed this prospective, crossover study to evaluate the biocompatibility and low-molecular-weight protein permeability (small protein permeability) of NV compared with a conventional PS (APS; ASAHIKASEI Industries, Inc., Tokyo, Japan).

Methods

Eleven maintenance HD patients (7 males; mean age 67.0 ± 10.2 years) on regular thrice-weekly HD treatment for 4 h were enrolled in the study after they had given their written informed consent. The underlying renal diseases were diabetes mellitus (N = 3), glomerulonephritis (N = 2), and unknown (N = 6). The mean duration of dialysis treatment was 48.6 ± 45.0 (range 15 to 258) months. All of the patients underwent 12 consecutive HD treatments with NV-13U (1.3 m2) and then switched with APS-13SA (1.3 m2) as a control after a 2-week wash-out period.

Laboratory measurements

The platelet (PLT) count, white blood cell (WBC) count, and hematocrit (Hct) were measured at the beginning of dialysis and after 30, 60, and 240 min. To adjust for differences in the cell counts at the beginning of dialysis, we calculated the change ratio at each point using the following formula (1):

$$ {\displaystyle \begin{array}{l}\mathrm{Change}\ \mathrm{ratio}\ \mathrm{X}\ \min =100+100\times \left\{\left(\mathrm{Cell}\ \mathrm{count}\times \min \right)/\left(\mathrm{Cell}\ \mathrm{count}\ 0\ \min \right)-1\right\}\\ {}\mathrm{Cell}\ \mathrm{count}\ 0\ \min :\mathrm{Cell}\ \mathrm{count}\ \mathrm{at}\ \mathrm{the}\ \mathrm{beginning}\ \mathrm{of}\ \mathrm{dialysis}\\ {}\mathrm{Cell}\ \mathrm{count}\ \mathrm{X}\ \min :\mathrm{Cell}\ \mathrm{count}\ \mathrm{after}\ \mathrm{X}\ \mathrm{minutes}\ \mathrm{from}\ \mathrm{the}\ \mathrm{beginning}\ \mathrm{of}\ \mathrm{dialysis}\end{array}} $$
(1)

We measured the urea nitrogen, serum creatinine, inorganic phosphorus, single-pool Kt/V [4], and hemoglobin levels; the removal rate of β2-M; and the rate of variation of β2-M, α1-microglobulin (α1-M), and albumin. We also measured the levels of platelet factor-4 (PF-4), β-thromboglobulin (β-TG), fibrinogen/fibrin degradation products (FDP), d-dimer, and fibrinogen (FIB).

Dialysate samples were collected at the final two dialysis sessions at a speed of 1.0 L/h using a fixed-quantity pump from the dialysate discharge line for 0–60, 60–180, and 180–240 min after the start of HD. On the assumption that each value measured before dialysis was 100%, the results were evaluated for variability.

This study was conducted in accordance with the guidelines of the Declaration of Helsinki and approved by the ethics committees of Kamifukuoka General Hospital and National Defense Medical College Hospital.

Statistical analyses

The continuous values of variables are expressed as the mean ± standard deviation or as the median (25th, 75th percentile). The comparative p values for continuous variables were calculated by an analysis of variance (ANOVA). All of the statistical analyses were carried out using the JMP Pro Version 11.2.0 software program (SAS Institute, Cary, NC, USA), setting the significance level at a p value of less than 0.05.

Results

Biocompatibility

Reductions in the WBC count were seen with APS compared with NV at 60 min (NV 5.65 ± 1.60, APS 5.17 ± 1.65 × 103/μL, p < 0.05) and 240 min (NV 5.28 ± 1.38, APS 4.63 ± 1.20 × 103/μL, p < 0.005) after the start of HD (Fig. 1). The PLT count was not markedly different between the two dialyzers. The change ratios of the WBC count were significantly smaller with NV than with APS after 60 min (NV 96.3 ± 2.3, APS 92.5 ± 7.5%, p < 0.01) and 240 min (NV 91.3 ± 2.3, APS 84.6 ± 11.3%, p < 0.05) (Fig. 2).

Fig 1
figure 1

a WBC count and b PLT count. a Reduction in the white blood cell count were seen with the APS compared with the NV in 60 min (*p < 0.05) and 240 min (**p < 0.005) after the start of HD. b There were no significant differences between NV and APS

Full size image
Fig 2
figure 2

Change ratios of a WBC and b PLT. a The change ratios of the white blood cell count were significantly smaller with the NV than with the APS after 60 min (*p < 0.01) and 240 min (**p < 0.05). b There were no significant differences between NV and APS

Full size image

No significant differences were seen in the levels of Hct, PF-4, β-TG, FDP, or d-dimer or in the change ratio of FIB between NV and APS (Table 1). There were also no significant differences in the urea nitrogen, creatinine, inorganic phosphorus, single-pool Kt/V, or hemoglobin values between the two dialyzers (Table 2).

Table 1 Clinical parameters of Hct, PF-4, β-TG, FDP, d-dimer, and change ratio of FIB
Full size table
Table 2 Clinical parameters
Full size table

Small protein permeability

The intradialytic removal of α1-M was greater with NV than with APS in the first hour (NV 15.7 ± 4.0, APS 10.7 ± 3.0 mg, p < 0.001) and last hour (NV 6.47 ± 2.35, APS 3.23 ± 0.74 mg, p < 0.001) as well as over the total HD session (NV 32.0 ± 11.3, APS 17.7 ± 6.3 mg, p < 0.0001) (Fig. 3). In addition, the removal of albumin was greater with NV than with APS in the first hour (NV 340 ± 125, APS 198 ± 59 mg, p < 0.001) and last hour (NV 109 ± 46, APS 32.0 ± 3.8 mg, p < 0.001) as well as over the total HD session (NV 693 ± 283, APS 249 ± 90 mg, p < 0.001) (Fig. 4). There were no significant differences in the removal of β2-M between the two dialyzers (Fig. 5). However, the rates of variation during HD were greater with NV than with APS for β2-M (NV 45.5 ± 1.2, APS 40.1 ± 1.2%, p < 0.001), α1-M (NV 41.2 ± 9.9, APS 34.2 ± 18.5%, p < 0.05), and albumin (NV 31.6 ± 7.8, APS 18.1 ± 6.5%, p < 0.001).

Fig 3
figure 3

a Removal amount of α1-M and b total removal amounts of α1-M. Compared to APS. *p < 0.001 and **p < 0.0001 by paired t test. α 1 -M α1-microglobulin

Full size image
Fig 4
figure 4

a Removal amount of Alb and b total removal amounts of Alb. Compared to APS. *p < 0.0001 by paired t test. Alb albumin

Full size image
Fig 5
figure 5

a Removal amount of β2-M and b total removal amounts of β2-M. There were no significant differences between NV and APS. β 2 -M β2-microglobulin

Full size image

Discussion

The PS membrane has shown a very good biocompatibility due to its superior removal performance for small-molecule proteins such as β2-M, high clearance of small-molecule solutes such as urea and creatinine [5], and low rate of complications such as leukopenia or disorders associated with the complement system [6]. Therefore, the PS membrane is widely used for HD treatment in Japan [7]. The NV membrane was developed as an alternative to PS. With NV, the mobility of water adjacent to the membrane surface is enhanced using a new hydrophilic polymer to reform the membrane surface. NV has antithrombotic activity and less platelet and leukocyte activation than PS, as well as the ability to sufficiently remove small solutes and low-molecular-weight proteins [3]. Further, NV induces less interleukin-6 activity, which may reduce the risk of erythropoiesis-stimulating agent hyporesponsiveness [8]. In this study, we evaluated the biocompatibility and protein permeability of NV compared with those of APS as a conventional PS.

The variation of several parameters such as the WBC, PLT, Hct and levels of PF-4, β-TG, FDP, d-dimer, and FIB were compared to evaluate the biocompatibility of NV. PF4 and β-TG are substances contained within platelet α-granules, and changes in platelet PF4 or β-TG levels are useful indicators of platelet degranulation and activation during dialysis [9]. However, there were no significant changes observed in the coagulation factor values with either dialyzer. Although acute leukopenia is known to occur in all patients during the first 30 min of HD treatment [10], our study showed no reduction in the WBC during dialysis with either dialyzer. The platelet count usually decreases 5 to 15% at 15 to 30 min from the start of HD and recovers to the normal level by the end of the dialysis session [9, 11]. Oshihara [3] reported that the numbers of platelets adhering to the membrane surface were lower with NV than with the conventional PS (CX) membrane. In our results, the variations in the platelet counts with both membranes were stable, suggesting that both may prevent HD-associated thrombocytopenia.

It is known that WBC adhesion occurs with protein adsorption, and FIB adsorption promotes PLT adhesion on the dialysis membrane surface by blood exposure in HD treatment [12]. Platelet-derived microparticles are released from activated platelets during HD [13]. Although the biocompatibility of PS is generally regarded to be good, there are some reports that the PS dialyzer caused PLT depletion [14, 15]. On the other hand, the NV membrane has been reported that it not only inhibits the adhesion of blood components such as PLT and FIB but also acts as a highly biocompatible membrane to reduce blood cell stimulation because it is developed as a new hydrophilic polymer membrane [3]. With NV, the reduction rate of the WBC count was inhibited to a greater degree than with APS. However, the change ratio of FIB showed no significant difference between NV and APS. These results suggest that NV improves the biocompatibility compared with APS.

Dialysis-related amyloidosis is a major complication with long-term HD treatment and greatly affects patients’ quality of life. The removal of β2-M is therefore important to avoid this complication. However, there have been no reports of the small protein permeability of NV. We therefore designed this prospective crossover study to evaluate whether or not NV offers any advantages over APS. We found that the intradialytic removal of α1-M and albumin but not β2-M was significantly greater with NV than with APS. The rates of variation in the β2-M and α1-M levels during HD with NV were higher than with APS, but the differences were small. These results suggest that the removal performance of β2-M with NV and APS should be almost equivalent.

In addition, since the decrease in the serum albumin may cause low muscle mass [16], malnutrition [17], and mortality risk [18], it is important to select dialyzers with consideration of the laboratory data. NV removed significantly greater albumin than APS. Therefore, we should be cautious when using the NV membrane in elderly or malnourished patients, although the amount of albumin removed was less than 1 g (NV 693 ± 283, APS 249 ± 90 mg) in either dialyzer.

Because this was a short-term, prospective, crossover study involving only a few cases, large-scale and long-term future trials are necessary to corroborate our findings.

Conclusions

We designed this crossover study of the NV and APS membranes to evaluate the biocompatibility and small protein removal ability. The clinical characteristics of NV may reveal an improved biocompatibility and a comparable efficiency in small protein removal as compared to APS.