Differences in Bio-incompatibility among Four Biocompatible Dialyzer Membranes Using in Maintenance Hemodialysis Patients

Abstract

Background: Following the introduction of modified cellulosic and then synthetic membrane dialyzers, it was realized that the dialyzer bio-incompatibility depends on the membrane composition. We designed a prospective, randomized, cohort study of 6 months to determine several parameters of biocompatibility in maintenance hemodialysis (MHD) patients treated with four different membrane dialyzers. Methods: There were 60 MHD patients enrolled in the study. In baseline, synthetic low-flux dialyzer, polysulfone (PS) membrane was used in all patients for at least 3 months. Then the patients were randomly divided into three groups according to different dialyzer membranes. Synthetic high-flux dialyzer group, ployethersulfone membrane, cellulose triacetate (CTA) high-flux membrane, and synthetic low-flux dialyzer, polymethylmethacrylate (PMMA) membrane were used in 6 months. A new dialyzer was used for each study treatment, and there was no dialyzer reuse. The biocompatibility markers and solutes removal markers were detected repeatedly at different time points. Results: The blood levels of highly sensitive C reactive protein, interleukin (IL)-1β, and interleukin (IL)-13 showed no difference among different groups at al time points. However, the blood complement levels and white blood cell counts were significantly different among three groups. When the dialyzers changed from PS to PMMA membrane, C3a levels and white blood cell counts changed significantly (p < 0.05). Moreover, the changes of C5a levels were significantly different between group CTA and group PMMA in month 3 (p < 0.05). Conclusion: There were much more differences on bio-incompatibility among different dialyzer membranes.

Keywords:

• Hemodialysis
• dialyzer
• biocompatibility
• membrane flux

INTRODUCTION

Following the introduction of modified cellulosic and then synthetic membrane dialyzers, it was realized that the dialyzer bio-incompatibility depended on the membrane composition. This is based on the view that relatively bio-incompatible membranes (e.g., cellulose) as compared with more biocompatible membranes (e.g., polysulfone, PS) have a higher ability to activate complement and leucocytes which may increase the inflammatory activity.¹ The biocompatibility of these dialyzers is evaluated by their generation of the complement activation products C3a and C5a; production of proinflammatory cytokines such as interleukin (IL)-1β, tumor necrosis factor-α (TNF-α), or IL-6; and changes in peripheral blood mononuclear cells during hemodialysis (HD).² The high morbidity and mortality of maintenance hemodialysis (MHD) patients commonly were regarded as a result of dialysis membranes bio-incapability. However, in a recent Cochrane review on this subject,³ no evidence of benefit was found when biocompatible membranes were compared with cellulose/modified cellulose membranes in terms of reduced mortality or reduction in dialysis-related adverse symptoms. So it became apparent that not all adverse reactions during HD were solely due to dialyzer membrane composition, and that dialyzer design, membrane flux, anticoagulation with unfractionated heparins, dialysate water impurities, composition, and sterilants, such as ethylene oxide, could all cause sudden adverse effects during dialysis.⁴ Furthermore, acquired immunity disturbances in HD patients are general. They are caused by uremia per se, the HD procedure, chronic renal failure complications, and therapeutic interventions for their treatment.⁵

Table 1. Dialyzer characteristics.

Dialyzer	Membrane	Membrane area (m^2)	Sterilization	Kuf (mL × min^−1 × mmHg^−1)	KoA
F7	PS	1.6	Steam	6.4	787
130-DS	PES	1.3	γ-ray	36.0	729
130U	CTA	1.3	γ-ray	25.8	1003
B3-1.6A	PMMA	1.6	γ-ray	8.7	736

Note: PS, polysulfone; PES, ployethersulfone; CTA, cellulose triacetate; PMMA, polymethylmethacrylate; Kuf, ultrafiltration ratio of dialyzers; KoA, mass transfer area coefficient.

In view of these considerations, we designed a prospective, randomized, cohort study to determine several parameters of biocompatibility in MHD patients treated with four different membrane dialyzers. The bio-incompatibility of them was compared during 6 months.

MATERIALS AND METHODS

Study Design

In baseline, synthetic low-flux dialyzer PS membrane (F7, Fresenius Medical Care, Bad Homburg, Germany) was used in all patients for at least 3 months. Then the patients were randomly divided into three groups according to different dialyzer membranes. Synthetic high-flux dialyzer, ployethersulfone (PES) membrane (PES-130DS, NISSHO Corporation, Osaka, Japan), cellulose triacetate (CTA) high-flux membrane (FB-130U, NISSHO Corporation), and synthetic low-flux dialyzer, polymethylmethacrylate (PMMA) membrane (B3-1.6A, Toray Industries, Inc., Tokyo, Japan) were used in 6 months. A new dialyzer was used for each study treatment, and there was no dialyzer reuse. Details of the dialyzers are presented in Table 1.

Blood samples were collected from the afferent (arterial) line. Before the first dialysis session with each of the different dialyzers mentioned (baseline), blood samples were drawn at five time points in one session, including before HD, 15, 30, 60, 240 min during HD. After the patients were transferred to be treated by three different dialyzers, respectively, blood samples were drawn at the end of 3 and 6 months. There were same five time points for collecting blood samples as above.

Patients

There were 60 MHD patients enrolled in the study in one HD unit. For inclusion in the study, patients were required to be older than 18 and less than 65 years of age and have a stable HD prescription for more than 6 months. They must have been dialyzing through a native arteriovenous fistula (AVF) providing a blood flow of 300 mL/min. Exclusion criteria included non-compliance with their dialysis prescription, using glucocorticosteroid or immunodepressant, a hemoglobin <100 g/L or active infection. The treatments of hypertension and anemia were recorded. The study was approved by the Ethics Committee of the Beijing Friendship Hospital affiliated of Capital Medical University.

Dialysis Procedure and Materials

All patients were dialyzed three times per week, 4 h per session, according to their routine HD scheme. Treatments were performed using model DBB26 dialysis delivery systems (NIKKISO, Inc., Japan). Concentration support system of dialysate was used in our unit. The fresh bicarbonate dialysate flow was 500 mL/min during HD. The endotoxin levels of fresh dialysate from dialysis machines were 0.01–0.05 EU/mL, which was detected randomly every month by limulus amebocyte lysate method. According to the individual needs of the patients, blood flow and ultrafiltration rates were kept constant 250–300 mL/min and 500–1100 mL/h, respectively. All dialyzers were prerinsed with 1000 mL 0.9% NaCl before every session. Anticoagulation was achieved with unfractionated heparin administered as an initial loading dose followed by a constant infusion. The doses of heparin were unchanged during this study.

Measurements of Solute Removal

The change in solute concentration over the entire treatment was determined from pre- and post-dialysis blood samples. Pre-dialysis blood samples were drawn from the access needle immediately following needle insertion. Post-dialysis blood samples were drawn from the arterial blood line 20 s after reducing the blood flow to 50 mL/min to mitigate any access recirculation. Concentrations of urea, creatinine (Cr), uric acid (UA), phosphorus, calcium, total cholesterol, triglyceride, highly sensitive C reactive protein (hs-CRP), and albumin were determined by standard clinical laboratory methods. The concentrations of β₂-microglobulin and intact parathyroid hormone were measured by immunoradiometric assay (Immulite, Los Angeles, CA, USA). Asymmetric dimethylarginine (ADMA) levels were measured by enzyme-linked immunosorbent assay (ELISA, Adlitteram Diagnostic Laboratories, Inc., USA). Among these parameters, urea, Cr, and UA were regarded as small-molecule solutes, β₂-microglobulin as a middle molecule solute, and ADMA as a protein-bonded solute.^6,7

The pre- to post-dialysis change in solute concentration was calculated from

Table 2. Patients’ characteristics on month 0.

Groups^a	PES	CTA	PMMA
N	20	20	20
Male/female	10/10	10/10	10/10
Age (year)	53.8 ± 12.1	56.1 ± 13.5	50.1 ± 8.8
Time on HD (year)	8.25 ± 2.56	9.00 ± 3.34	8.37 ± 3.98
Protopathy [n, (%)]
Chronic glomerulonephritis	18 (90%)	16 (80%)	16 (80%)
Diabetes	1 (5%)	1 (5%)	1 (5%)
Hypertension	1 (5%)	1 (5%)	2 (10%)
ADPKD	0 (0%)	2 (10%)	1 (5%)
SBPpre-HD (mmHg)	119.8 ± 22.6	120.4 ± 12.8	129.7 ± 21.9
DBPpre-HD (mmHg)	70.3 ± 12.3	68.6 ± 17.8	76.4 ± 15.0
SBPpost-HD (mmHg)	106.5 ± 31.2	102.7 ± 18.2	113.8 ± 13.1
DBPpost-HD (mmHg)	66.5 ± 19.8	67.1 ± 17.0	74.3 ± 14.2
Dry weight (kg)	60.9 ± 10.1	64.2 ± 10.1	67.2 ± 14.3
Ultrafiltration volume (mL)	3,78 ± 1.09	3.41 ± 0.78	3.71 ± 0.82
Urea reduction ratio (%)	73.86 ± 6.86	71.99 ± 6.50	70.59 ± 7.33
Hemoglobin (g/L)	111.9 ± 13.2	112.9 ± 19.4	111.2 ± 18.1
Creatinine (μmol/L)	912.5 ± 178.9	871.8 ± 209.3	898.5 ± 199.8
Urea (mmol/L)	24.0 ± 5.5	22.1 ± 4.1	23.8 ± 5.0
Uric acid (mmol/L)	389.9 ± 61.3	368.8 ± 51.5	405.3 ± 77.3
Albumin (g/L)	38.2±2.4	38.8 ± 2.9	40.5 ± 2.2
Total cholesterol (mmol/L)	4.97 ± 1.46	4.92 ± 0.72	4.71 ± 1.22
Triglyceride (mmol/L)	2.33 ± 1.20	2.33 ± 1.98	2.62 ± 2.03
Calcium (mmol/L)	2.30 ± 0.16	2.30 ± 0.29	2.30 ± 0.22
Phosphorus (mmol/L)	1.93 ± 0.57	1.93 ± 0.41	2.01 ± 0.43
Ferritin (ng/mL)	310.2 ± 228.2	372.1 ± 205.9	287.2 ± 114.2
β2-microglobulin	27.53 ± 7.44	25.65 ± 6.65	25.01 ± 6.10
Intact parathyroid hormone (pg/mL)	336.1 ± 446.4	378.2 ± 361.7	270.0 ± 221.0
hs-CRP (mg/L)	0.58 ± 0.29	0.59 ± 0.42	0.78 ± 0.59

Notes: SBP_pre-HD, systolic blood pressure pre-hemodialysis; DBP_pre-HD, diastolic blood pressure pre-hemodialysis; SBP_post-HD, systolic blood pressure post-hemodialysis; DBP_post-HD, diastolic blood pressure post-hemodialysis; ADPKD, autosomal dominant polycystic kidney disease. There were no statistical significant differences among three different groups on all above variables (p > 0.05).

^aOn month 0, 60 maintenance hemodialysis patients using polysulfone (PS) membrane dialyzers were divided into three different groups according to three different dialyzers: PES, ployethersulfone; CTA, cellulose triacetate; PMMA, polymethylmethacrylate.

Table 3. Reduction ratios (%) of several solutes among different groups.

Solutes	Time	Group PES (N = 20)	Group CTA (N = 20)	Group PMMA (N = 20)
Urea	Month 0	73.65 ± 6.82	72.03 ± 6.56	70.65 ± 7.34
	Month 3	71.74 ± 8.04	70.96 ± 6.88	68.91 ± 6.19
	Month 6	71.36 ± 6.89	72.38 ± 5.45	69.88 ± 5.04
Creatinine	Month 0	64.53 ± 6.96	63.13 ± 7.20	63.13 ± 6.15
	Month 3	63.57 ± 8.02	64.22 ± 8.69	61.22 ± 5.42
	Month 6	61.35 ± 6.21	63.16 ± 4.67	62.28 ± 5.30
Uric acid	Month 0	74.70 ± 6.79	75.27 ± 6.40	72.04 ± 6.64
	Month 3	74.07 ± 7.42	74.82 ± 5.06	71.24 ± 5.98
	Month 6	73.31 ± 8.74	75.10 ± 7.48	71.28 ± 4.81
β2-microglobulin	Month 0	26.72 ± 25.71	12.50 ± 28.65	19.20 ± 16.23
	Month 3	10.18 ± 16.25	15.06 ± 21.60	16.44 ± 26.18
	Month 6^**	15.05 ± 12.56	12.43 ± 14.99	23.65 ± 22.11
ADMA^**	Month 0	23.48 ± 13.00	25.74 ± 9.30	26.84 ± 14.09
	Month 3	24.99 ± 19.79	22.45 ± 18.33	27.26 ± 15.75
	Month 6	27.04 ± 12.16	22.94 ± 10.21	27.22 ± 12.01

Notes: There were no statistical significant differences among three different groups on all above variables at different time points by using q-test (p > 0.05). However, the differences were statistically significant in group PES when the reduction ratios of β₂-microglobulin are compared by using paired-sample t-tests, month 0 versus month 3. There were no significant differences in paired-sample t-tests on other parameters among month 0 versus month 3, month 0 versus month 6, and month 3 versus month 6 (p > 0.05).

*Denotes p = 0.023, and month 0 versus month 6.

^**Denotes p = 0.056

where Cpre and Cpost are the pre- and post-dialysis solute concentrations, respectively. The post-dialysis concentrations of ADMA and β₂-microglobulin were corrected for body weight (BW) according to Bergström and Wehle⁸ before calculating the pre- to post-dialysis change in concentration:

where C_postc and C_post are corrected and uncorrected post-HD blood concentrations of solutions, respectively, BW_post is body weight post-HD, and ΔBW is the change in body weight during dialysis.

Plasma and dialysate samples were stored at −80°C until analysis.

Assessment of Dialysis Membrane Biocompatibility

Laboratory tests of dialyzer membrane biocompatibility included plasma level of complement factors of C3a, C5a, interleukin (IL)-1β, hs-CRP, blood cells counts, and anti-inflammation factor IL-13, and so on. Blood levels of complements and interleukin factors were measured by ELISA (Adlitteram Diagnostic Laboratories). Complement levels were detected at 0, 15, 240 min; blood cell counts were detected at 0, 15, 30, 60, 240 min; and other parameters were detected at 0 and 240 min. The post-dialysis concentrations of CRP, IL-1β, and IL-13 were corrected for BW.⁸

Statistical Analysis

Differences in plasma solute concentrations among dialyzers were assessed by analysis of variance (SPSS for Windows, version 13, SPSS, Chicago, IL, USA). Where significant differences were found (p < 0.05), differences between individual pairs of dialyzers (PES, CTA, and PMMA) were determined using the Student–Newman–Keuls correction for multiple comparisons. Paired-sample t-test method was used to compare differences of parameters between baseline and transferred dialyzers (between PS and PES, CTA, PMMA, respectively). Blood cell counts and complement factors at different time points were analyzed by repeated measures of Bonferroni method. Data were presented as mean ± SD.

RESULTS

Subject Characteristics

Table 2 shows the subject characteristics among different groups. Mean age of study patients was 53.3 ± 11.7 years, 50% were men, and the average time on dialysis therapy was 8.53 ± 3.29 years. Mean systolic blood pressure was 123.5 ± 20.1 mmHg, and mean diastolic blood pressure was 72.0 ± 15.2 mmHg before dialysis in midweek. Dialysis parameters included a mean dry body weight of 64.1 ± 11.9 kg, blood flow of 300 mL/min, mean ultrafiltration volume of 3.64 ± 0.91 L in midweek session, and mean urea reduction ratio of 72.1 ± 6.9% per dialysis session. Among selected laboratory variables, mean values were 112.0 ± 16.7 g/L for hemoglobin, 895.4 ± 192.8 μmol/L for plasma Cr, 23.4 ± 4.9 mmol/L for urea, 388.7 ± 65.1 mmol/L for UA, 38.5 ± 2.2 g/L for albumin, 4.87 ± 1.17 mmol/L fortotal cholesterol, 2.43 ± 1.75 mmol/L for triglyceride, 1.96 ± 0.47 mmol/L for phosphate, 2.30 ± 0.22 mmol/L for calcium, 327.1 ± 351.2 pg/mL for intact parathyroid hormone, and 0.65 ± 0.45 mg/L for hs-CRP. There were no significant differences on above parameters among three groups (Table 2).

Solute Removal

Table 3 shows the reduction ratios of different solutes according to different dialyzer groups. There were no statistically significant differences of urea, Cr, UA, and ADMA among different groups. The reduction ratio of β₂-microglobulin decreased significantly when PS membrane dialyzers were changed to PES membrane dialyzers (Table 3). There were no significant differences either among three groups at baseline or transfer dialyzers from PS membrane to CTA and PMMA membranes.

Dialyzers Membrane Biocompatibility

Inflammation factors

The blood levels of hs-CRP, IL-1β, and IL-13 showed no difference among different groups at all time points (Table 4).

Table 4. Blood levels of inflammation factors.

Inflammation factors	Time		PES group (N = 20)	CTA group (N = 20)	PMMA group (N = 20)
hs-CRP (mg/L)	Month 0	Pre-HD	0.58 ± 0.29	0.59 ± 0.42	0.78 ± 0.59
		Post-HD	0.46 ± 0.26	0.60 ± 0.36	0.52 ± 0.19
	Month 3	Pre-HD	0.72 ± 0.37	0.64 ± 0.48	0.56 ± 0.23
		Post-HD	0.59 ± 0.30	0.47 ± 0.24	0.53 ± 0.27
	Month 6	Pre-HD	0.60 ± 0.30	0.61 ± 0.29	0.49 ± 0.22
		Post-HD	0.48 ± 0.20	0.56 ± 0.32	0.54 ± 0.44
IL-1β (pg/mL)	Month 0	Pre-HD	42.74 ± 14.21	42.84 ± 13.76	37.51 ± 10.22
		Post-HD	35.64 ± 12.14	34.63 ± 14.08	31.33 ± 12.04
	Month 3	Pre-HD	40.12 ± 23.78	36.63 ± 13.85	36.36 ± 19.91
		Post-HD	31.26 ± 9.09	29.39 ± 12.88	32.35 ± 14.05
	Month 6	Pre-HD	33.56 ± 18.32	38.20 ± 19.50	37.27 ± 19.47
		Post-HD	30.91 ± 11.48	29.34 ± 20.87	28.40 ± 13.96
IL-13 (pg/mL)	Month 0	Pre-HD	38.06 ± 13.92	38.17 ± 15.60	32.71 ± 9.84
		Post-HD	40.95 ± 12.25	36.98 ± 10.88	34.98 ± 14.07
	Month 3	Pre-HD	39.17 ± 14.94	40.44 ± 12.25	39.55 ± 13.47
		Post-HD	42.99 ± 17.42	40.70 ± 18.94	38.58 ± 12.41
	Month 6	Pre-HD	37.29 ± 16.76	39.44 ± 12.74	36.77 ± 12.83
		Post-HD	40.54 ± 8.86	42.96 ± 17.83	36.51 ± 16.59

Notes: There were no statistical significant differences among three different groups on all above variables at different time points by using q-test (p > 0.05). When paired-sample t-tests were used to compare the differences on blood levels of inflammation factors between pre- and post-hemodialysis (HD), and on all parameters among month 0 versus month 3, month 0 versus month 6, and month 3 versus month 6, the significant differences were not found (p > 0.05).

Complements

The pattern on changes of C3a and C5a in three groups was similar. Blood levels of both C3a and C5a increased significantly at 15 min and decreased gradually at 240 min in most sessions (Figure 1). However, when the dialyzer was changed from PS to PMMA membrane, the range of C3a levels were declined significantly (p < 0.05). The difference of C5a levels was not significant among 0, 15, and 240 min in month 0, when all patients dialysis were done by PS membrane dialyzer (p > 0.05). The C5a levels raised more rapidly and at higher values significantly at 15 min in group CTA than in group PMMA in month 3 (p = 0.023) (Table 5).

Table 5. Blood levels of complements.

Complement		Time point	Group PES (N = 20)	Group CTA (N = 20)	Group PMMA (N = 20)	Within-subjects effects	Within-subjects effects	Between-subjects effects
						Minute	Group × minute	Group
C3a (ng/mL)	Month 0	0 min	117.42 ± 27.40	120.84 ± 30.01	117.92 ± 35.99	p = 0.001	p = 0.768	p = 0.444
		15 min	146.19 ± 48.31	168.20 ± 56.01	183.09 ± 57.02
		240 min	137.68 ± 49.91	149.45 ± 44.71	135.20 ± 43.08
	Month 3	0 min	108.14 ± 26.65	110.27 ± 21.24	96.11 ± 30.84	p < 0.001	p = 0.622	p = 0.098
		15 min	172.48 ± 50.20	154.49 ± 48.56	141.48 ± 50.70
		240 min	131.81 ± 43.51	132.29 ± 36.52	115.38 ± 37.49
	Month 6	0 min	96.88 ± 31.57	97.80 ± 27.45	85.05 ± 17.28	p < 0.001	p = 0.675	p = 0.376
		15 min	164.53 ± 37.62	155.89 ± 36.73	146.31 ± 41.07
		240 min	120.39 ± 36.78	115.99 ± 27.70	116.69 ± 32.95
Within-subjects effects	Minute	p < 0.001	p < 0.001	p < 0.001
Within-subjects effects	Month × minute	p = 0.954	P = 0.633	p = 0.483
Between-subjects effects	Month	p = 0.571	p = 0.058	p = 0.004^a
C5a (ng/mL)	Month 0	0 min	76.28 ± 15.55	76.47 ± 19.66	75.57 ± 20.27	p = 0.646	p = 0.773	p = 0.675
		15 min	85.21 ± 21.13	80.27 ± 23.17	84.49 ± 23.58
		240 min	82.50 ± 31.30	76.20 ± 17.83	75.51 ± 19.25
	Month 3	0 min	74.70 ± 18.70	87.16 ± 30.60	60.59 ± 19.07	p = 0.001	p = 0.324	p = 0.023^b
		15 min	98.85 ± 27.78	91.77 ± 26.11	85.30 ± 23.50
		240 min	83.96 ± 27.87	80.46 ± 30.94	72.74 ± 28.97
	Month 6	0 min	77.45 ± 20.86	89.56 ± 32.89	67.88 ± 17.87	p < 0.001	p = 0.561	p = 0.092
		15 min	103.5 ± 27.92	101.77 ± 28.63	94.61 ± 25.97
		240 min	93.02 ± 31.01	89.08 ± 34.30	81.72 ± 15.17
Within-subjects effects	Minute	p < 0.001	p = 0.043	p < 0.001
Within-subjects effects	Month × minute	p = 0.526	p = 0.887	p = 0.444
Between-subjects effects	Month	p = 0.118	p = 0.051	p = 0.415

Notes: Statistical method of repeated measures was used to compare the blood levels of C3a and C5a among different time points and different groups, respectively. In a certain group, the blood levels of C3a and C5a were significantly different among 0, 15, and 240 min (p < 0.05). There were no significant differences on the levels of C3a and C5a among different months both in group PES and in group CTA (p > 0.05).

^aIn group PMMA, there were significant differences on the C3a levels rather than C5a levels among month 0, month 3, and month 6 (C3a, p = 0.004, C5a, p > 0.05). Moreover, multiple comparisons on C3a levels in group PMMA show significant differences on month 0 versus month 3 (p = 0.015) and month 0 versus month 6 (p = 0.009), and no significant differences between month 3 and month 6 (p = 1.000). (These results were not listed in Table 5.) For blood C3a levels, the significant differences appear among 0, 15, and 240 min in every month (p < 0.001). There were no significant differences on C3a levels on a certain month among group PES, group CTA, and group PMMA (p > 0.05). For blood C5a levels, the differences were not significant on month 0 both among three time points and among three groups (p > 0.05). However, there were significant differences among 0, 15, and 240 min, both on month 3 and month 6 (p < 0.001).

^bFurthermore, the C5a levels on month 3 were significantly different among three groups (p = 0.023). By multiple comparisons, there were significant differences between group PES and PMMA (p = 0.023), and the differences were not significant in both group PES versus group CTA and group CTA versus PMMA (p > 0.05). (These results were not listed in Table 5.)

Blood cell counts

The parameter blood cell counts first decreased at 15 min, and then increased gradually during residual period in one dialysis session. There were no significant differences in red blood cell count and platelet count, neither among groups nor among the months 0, 3, and 6 (p > 0.05) (Table 6). However, there were significant differences in white blood cell count (WBC) when dialyzer was changed from PS to PMMA membrane (p < 0.05) (Figure 2). It shows that WBC in both months 3 and 6 decreased to a much lower value at 15 min and increased soon after to a much higher value and more rapidly than WBC in month 0 (Table 6).

Figure 1. Changes of blood levels on complement (C3a) (A); on complement (C5a) (B). The maintenance hemodialysis patients all were treated by polysulfone membrane dialyzers on month 0. They were divided into three different groups randomly equally according to different membranes as group PES, group CTA, and group PMMA. PES, ployethersulfone; CTA, cellulose triacetate; PMMA, polymethylmethacrylate. The number of patients in each group was 20. The patients were followed-up for 6 months. The blood levels of complement 3a (C3a) and 5a (C5a) were detected on months 0, 3, and 6. They were detected at 0, 15, and 240 min in one session on each month. The details of data are listed in Table 5.

Notes: Blood levels of both C3a and C5a increased significantly at 15 min and decreased gradually at 240 min in most sessions, except for C5a levels on month 0. The pattern on changes of C3a and C5a in three groups was similar.

*Denotes However, the range of C3a levels in group PMMA were declined significantly when the dialyzers changed from PS to PMMA membrane, both on month 3 and on month 6 (p < 0.05).

^**Denotes The changes of C5a levels were significantly different between group CTA and PMMA on month 3 (p < 0.05).

Table 6. Blood cells’ detection.

Blood cells	Time		PES group (N = 20)	CTA group (N = 20)	PMMA group (N = 20)
White blood cell (×10^9/L)	Month 0	0 min	5.15 ± 1.62	5.72 ± 1.58	5.60 ± 1.42
		15 min	4.50 ± 1.74	4.57 ± 1.13	4.44 ± 0.98
		30 min	5.22 ± 1.74	5.19 ± 1.16	5.64 ± 1.22
		60 min	5.74 ± 1.55	5.85 ± 1.14	5.97 ± 0.82
		240 min	5.74 ± 1.55	6.46 ± 1.53	7.41 ± 1.87
	Month 3	0 min	4.99 ± 1.75	5.58 ± 1.47	6.02 ± 1.36^a
		15 min	4.57 ± 1.45	4.77 ± 2.01	2.32 ± 1.11^b
		30 min	5.04 ± 1.62	5.07 ± 1.09	5.39 ± 2.38
		60 min	5.28 ± 1.71	5.67 ± 1.33	6.36 ± 2.55
		240 min	5.67 ± 1.80	6.03 ± 1.65	7.53 ± 2.31
	Month 6	0 min	5.28 ± 1.54	5.22 ± 0.93	6.45 ± 1.29^c
		15 min	4.67 ± 1.44	4.06 ± 0.93	2.20 ± 1.42^d
		30 min	5.17 ± 1.56	5.02 ± 0.96	4.74 ± 1.35^e
		60 min	5.15 ± 1.64	5.52 ± 1.30	6.82 ± 1.58^f
		240 min	5.49 ± 2.02	5.84 ± 1.38	8.44 ± 1.61^g,h
Red blood cell (×10^12/L)	Month 0	0 min	3.70 ± 0.64	3.76 ± 0.80	3.62 ± 0.52
		15 min	3.58 ± 0.67	3.72 ± 0.78	3.58 ± 0.51
		30 min	3.60 ± 0.67	3.67 ± 0.79	3.63 ± 0.51
		60 min	3.66 ± 0.73	3.79 ± 0.68	3.68 ± 0.53
		240 min	4.23 ± 0.92	4.34 ± 0.83	4.13 ± 0.58
	Month 3	0 min	3.66 ± 0.67	3.68 ± 0.88	3.61 ± 0.65
		15 min	3.58 ± 0.66	3.71 ± 0.90	3.58 ± 0.57
		30 min	3.59 ± 0.69	3.68 ± 0.92	3.67 ± 0.58
		60 min	3.73 ± 0.76	3.77 ± 0.92	3.73 ± 0.59
		240 min	4.16 ± 0.94	4.15 ± 0.97	4.12 ± 0.66
	Month 6	0 min	3.78 ± 0.66	3.79 ± 0.71	3.62 ± 0.61
		15 min	3.64 ± 0.66	3.74 ± 0.74	3.51 ± 0.45
		30 min	3.64 ± 0.63	3.81 ± 0.73	3.58 ± 0.49
		60 min	3.97 ± 1.08	3.86 ± 0.74	3.57 ± 0.55
		240 min	4.26 ± 0.86	4.32 ± 0.84	4.12 ± 0.60
Platelet (×10^9/L)	Month 0	0 min	164.90 ± 54.04	160.18 ± 50.69	156.05 ± 56.39
		15 min	162.39 ± 51.74	152.12 ± 48.79	153.53 ± 56.75
		30 min	161.50 ± 53.34	152.59 ± 43.48	152.89 ± 52.84
		60 min	167.39 ± 57.52	155.59 ± 47.83	156.95 ± 55.06
		240 min	191.33 ± 67.61	177.35 ± 53.08	186.58 ± 60.22
	Month 3	0 min	159.90 ± 53.67	163.11 ± 56.29	160.05 ± 62.45
		15 min	162.78 ± 57.51	155.71 ± 49.03	146.21 ± 59.13
		30 min	165.61 ± 61.39	157.24 ± 51.82	148.63 ± 58.72
		60 min	172.94 ± 69.84	160.94 ± 55.72	149.16 ± 56.52
		240 min	197.83 ± 76.69	178.94 ± 58.08	168.53 ± 61.80
	Month 6	0 min	161.14 ± 53.24	159.72 ± 43.33	151.11 ± 57.56
		15 min	165.33 ± 53.40	153.94 ± 46.04	146.42 ± 51.90
		30 min	167.72 ± 54.06	158.82 ± 48.82	147.95 ± 52.62
		60 min	175.93 ± 58.13	160.53 ± 46.98	152.37 ± 53.21
		240 min	196.61 ± 66.88	174.88 ± 50.87	172.74 ± 57.43

Notes: There were no statistically significant differences among three different groups on all above variables at different time points by using q-test (p > 0.05). There were no statistical differences on other parameters by using paired-sample t-tests among month 0 versus month 3, month 0 versus month 6, and month 3 versus month 6 (p > 0.05), except the white blood cell (WBC) count in group PMMA. When paired-sample t-tests were used to compare the differences on the WBC count, there were significant differences among month 0 versus month 3 at ^a0 min (p = 0.042) and ^b15 min (p < 0.01), month 0 versus month 6 in ^c-g0–240 min (0 min, p = 0.01; 15 min, p < 0.01; 30 min, p = 0.026; 60 min, p = 0.39; 240 min, p = 0.001), and month 3 versus month 6 at ^h240 min (p = 0.004) in group PMMA.

DISCUSSION

Twenty years ago, the early cuprophane dialyzers were recognized to cause pulmonary leuko sequestration associated with complement activation. This led to the term “dialyzer bio-incompatibility,” and spawned a multitude of both in vitro and in vivo studies to investigate the bio-incompatibility of different dialyzer membranes. It was commonly recognized that the biocompatibility improved from cuprophane membrane, to modified cellulose membranes, to synthetic membranes.⁹ Today, the cost of dialyzers is very similar, whether they are of high or low flux, modified cellulose or synthetic; therefore, biocompatibility is not such a contentious issue. In our study, there were no significant differences both on most bio-incompatibility markers, such as hs-CRP, IL-1β, and platelet counts, and on anti-inflammation factor IL-13 among PS, PES, CTA, and PMMA membranes.

However, the blood complement levels and WBC counts were significantly different among three groups, the two classical parameters to characterize biocompatibility in dialysis.^10–12 With all three dialyzers (PES, CTA, PMMA) tested in present study, a significant increase in complement C3a and C5a levels and a minor minimal transient leucopenia within the first 15 min of dialysis was observed. There was no statistically significant difference on the changes of C5a levels during one dialysis session by using PS membrane, although the difference was significant on C3a levels. Because of the powerful adsorption capability, both relatively smaller changes of C3a levels and pronounced changes of WBC counts on PMMA group were observed in present study. These results suggested there were some differences on the biocompatibility of four dialyzers (Table 7).

Figure 2. Blood cell detection of (A) white blood cell count; (B) red blood cell count; (C) hemoglobin; (D) hematocrit; (E) platelet count.The maintenance hemodialysis patients all were treated by polysulfone membrane dialyzers on month 0. They were divided into three different groups randomly according to different membranes as group PES, group CTA, and group PMMA. PES, ployethersulfone; CTA, cellulose triacetate; PMMA, polymethylmethacrylate. The number of patients in each group was 20. The patients were followed-up for 6 months. White blood cell (WBC) count, red blood cell (RBC) count, hemoglobin (Hb), hematocrit (Hct), platelet counts (PLT) in peripheral circulation were detected on months 0, 3, and 6. The above parameters were detected at 0, 15, 30, 60, and 240 min in one session on each month. The WBC decreased significantly at 15 min and increased gradually at 30, 60, 240 min in each session. The pattern on changes of WBC in three groups was similar. Notes: *However, the WBC declined much significantly at 15 min and raised up much more quickly at 30, 60, 240 min when the patients were treated by using PMMA membrane than other membranes (p < 0.05). The other parameters also first decreased at 15 min, and then increased gradually during residual period in one dialysis session. There were no statistical differences in RBC, Hb, Hct, and PLT neither among each group nor among the months 0, 3, and 6 (p > 0.05).

Table 7. Comparison of solute removal markers and biocompatible markers among four groups.

	PS	PES	CTA	PMMA
Solute removal markers
RRUrea (%)	NS	NS	NS	NS
RRβ2-microglobulin (%)	NS	↓^a	NS	NS
RRADMA (%)	NS	NS	NS	NS
Biocompatible markers
WBC (×10^9/L)	NS	NS	NS	↓^a
ΔC3a (ng/mL)	NS	NS	NS	↓^a
C5a (ng/mL)	NS	NS	NS	↓^b
hs-CRP (mg/L)	NS	NS	NS	NS
IL-1β (pg/mL)	NS	NS	NS	NS
IL-13 (pg/mL)	NS	NS	NS	NS

Notes: RR, reduction ratio; NS, there were no significant differences when one group is compared with other groups.

^aIt was decreased for certain markers when they compared to PS membrane (baseline level). When PS membrane dialyzer transferred to PES, reduction ratio of β2-microglobulin decreased significantly. When PS membrane dialyzer transferred to PMMA, the white blood cell counts declined more deeply and the blood complement levels of C3a increased more gently at 15 min.

^bThe blood complement levels of C5a at time point on month 3 were decreased significantly in PMMA group compared with other two groups (PES and CTA).

Three recent multicentered studies have suggested that high-flux dialyzers may convey a survival advantage. One German study looked at outcome in a group of type 2 diabetic patients, which progressed to end-stage kidney failure, and observed a stepwise increase in a cardiac composite end-point (comprising stroke, nonfatal myocardial infarction, and death from cardiac causes), both at 1 and 3 years of follow-up, with the risk steadily increasing from high-flux synthetic dialyzers to low-flux synthetic/semisynthetic dialyzers, with the highest risk associated with the use of low-flux cellulosic dialyzers.¹³ In another study, the HEMO study was re-analyzed according to dialyzer membrane flux,¹⁴ and this showed that there was a significant survival advantage for those patients randomized to high-flux dialyzers, who had been established on regular HD for more than 3.7 years, and also for those without established cerebrovascular disease. This survival advantage could either be due to improved removal of the so-called middle molecules, with β₂-microglobulin acting as a surrogate for other so-called azotemic toxins, or due to improved bio-incompatibility of the membrane. The use of high-flux membranes conferred a significant survival benefit among diabetic patients and patients with serum albumin ≤4 g/dL.¹⁵ Whether this improvement in patient outcomes is due to the dialyzer flux per se,^13,14 or improved bio-incompatibility, or to better dialysate water quality,¹⁶ or center treatment effects, remains to be determined.

In our study, there were no significant differences on the clearances of small-molecule uremic toxins (Cr, urea, UA) and protein-bonded small-molecule toxin ADMA (Table 3) among four dialyzer membranes. The findings of recent studies, showing no differences in ADMA elimination neither between hemodiafiltration and low-flux HD^17,18 nor among different membrane dialyzers,¹⁸ suggest that plasma proteins that bind ADMA are larger than 40 kDa.⁷ Whether ADMA levels reduced in the long-term treatment or modalities with increased convective transport remain to be established.

The difference of β₂-microglobulin clearance was significant, which was not associated with the flux of dialyzer membranes (Table 3). The removal of β₂-microglobulin was not increased, when low-flux synthetic membranes (PS membrane) transferred to high-flux modified cellulose membranes (CTA membrane). Interestingly, the removal of β₂-microglobulin was even decreased, when PS dialyzer transferred to high-flux synthetic membrane (PES membrane). It might be part of reason that patients with higher β₂-microglobulin level were divided into PES group (Table 3). However, there were no significant differences on the removal of β₂-microglobulin between high-flux membranes (PES and CTA membranes) and low-flux synthetic membrane (PMMA membrane) in month 3 and month 6. The middle molecules such as β₂-microglobulin have two clearance methods of adsorption and convection through the dialyzer membranes. It may be more important for β₂-microglobulin clearance by adsorption than convection in these dialyzer membranes.¹⁹ PMMA membranes have more powerful adsorptive ability than other membranes.^19–21 So the low-flux synthetic membranes with higher absorptive ability (PS/PMMA membrane) may have the similar capability to high-flux membranes (PES/CTA membranes) on β₂-microglobulin removal. It is obvious that the possibilities of up-to-date polymer chemistry allow for the development of large-pore membranes with different compositions and characteristics. It is known that various high-flux membranes differ in biocompatibility toward leukocytes and complement system, coagulator properties, and/or adsorptive capacity.²² Hence, although this question has rarely been addressed, it is conceivable that there are also differences in removal capacity.

Whether the biocompatibility and the removal capacity are important enough to be translated into differences in clinical conditions and outcome remains open for discussion and should be evaluated. As Cheung said, a certain bio-incompatible phenomenon can be further classified as beneficial or deleterious depending on its biological effects as well as its acute and chronic impacts on the dialysis patient.¹¹

Our study had several limitations. First, our study population was small, and the power to detect differences was, therefore, limited. Second, the follow-up time was limited, and we have no information on the clinical outcome affected by dialyzer membranes’ bio-incompatibility beyond 6 months. Third, although we used some markers of biocompatibility, there is as yet limited information on their prognostic significance. Fourth, one group using PS membrane might be set as the control group in order to decrease biases.

Declaration of interest: The authors thank Zhang Yu and Zhang Qi-Dong for their technical help. This study got financial support from Beijing Municipal Science and Technology Commission Funds (No. D09050704310903), Collaborative Project Funds on Fundamental and Clinical Research of Capital Medical University (No. 10JL26), the Research Fund for the Doctoral Program of Higher Education (PFDP) of Ministry of Education of the People’s Republic of China (20101107120003).