Abstract

Background. β 2 -Microglobulin (β 2 -M) is recognized as a surrogate marker of middle-molecule uraemic toxins and is a key component in the genesis of dialysis-associated amyloidosis. Few studies have evaluated the association of β 2 -M levels with clinical outcome in dialyzed patients.

Methods. The prognostic implication of serum β 2 -M levels for the survival of haemodialysis patients was examined in 490 prevalent haemodialysis patients (60.1 ± 11.8 years, haemodialysis duration of 87.4 ± 75.7 months, 288 males and 202 females; 24% diabetics). The patients were divided into two groups according to their serum β 2 -M levels: lower β 2 -M group ( n = 245) with serum β 2 -M <32.2 mg/L (the median serum β 2 -M) and higher β 2 -M group ( n = 245) with that ≥32.2 mg/L.

Results. During the follow-up period of 40 ± 15 months, there were 91 all-cause deaths, and out of them, 36 were from cardiovascular diseases. Kaplan–Meier analysis revealed that all-cause mortality in the higher β 2 -M group was significantly higher compared to that in the lower β 2 -M group ( P < 0.001). Multivariate Cox proportional hazards analyses showed that serum β 2 -M level was a significant predictor for all-cause mortality (hazard ratio, 1.05; 95% CI, 1.01–1.08; P = 0.005), and for non-cardiovascular mortality (hazard ratio, 1.06; 95% CI, 1.02–1.10; P = 0.006), after adjustment for age, gender, haemodialysis duration, the presence of diabetes, serum albumin and serum C-reactive protein.

Conclusion. These results demonstrate that the serum β 2 -M level is a significant predictor of mortality in haemodialysis patients, independent of haemodialysis duration, diabetes, malnutrition and chronic inflammation, suggesting the clinical importance of lowering serum β 2 -M in these patients.

Introduction

β 2 -Microglobulin (β 2 -M) is a low-molecular-weight protein (11 800 Da), which is produced by all cells expressing the major histocompatibility class I. Under physiological conditions, β 2 -M is generated at a constant rate, except in patients with systemic inflammation and haematopoietic neoplasia, and is eliminated from circulation through the renal pathway. β 2 -M is filtered by the glomerulus and is degenerated in the proximal tubules through a megalin-dependent pathway [ 1 ]. In patients with a reduced glomerular filtration rate, circulating β 2 -M levels are elevated. In dialysis patients, in whom the glomerular filtration rate is almost completely abolished, β 2 -M accumulates in the circulation far above its levels in normal subjects. Although it remains unknown whether β 2 -M molecule itself is a uraemic toxin, β 2 -M has been considered to be a surrogate marker of putative middle-molecule uraemic toxins, which are difficult to dialyze by use of low-flux membrane [ 2 ]. Furthermore, elevated concentration of circulating β 2 -M has been demonstrated to be a potential risk for the onset and/or development of dialysis-related amyloidosis [ 3 ]. Therefore, removal of circulating β 2 -M during haemodialysis has been considered to be beneficial. Several methods to reduce plasma β 2 -M levels have been attempted in dialyzed patients, such as the use of high-flux membrane [ 2 ], haemodiafiltration [ 4–6 ], ultrapure dialysate [ 2 ] and the absorptive affinity column for β 2 -M [ 3 ]. However, few studies have evaluated the association of β 2 -M levels with clinical outcome in dialyzed patients [ 7 ]. In this study, we examined the relationship between serum β 2 -M and survival of haemodialysis patients in a single dialysis centre, where high-flux membranes are used exclusively as a standard approach and the reuse of membranes is not done at all.

Subjects and methods

Subjects

There were 541 patients on prevalent haemodialysis (haemodialysis duration >3 months) at Shirasagi Hospital Kidney Center, Osaka, Japan, as of October 1999. Of these, patients with acute illness, significant infection or malignancy were excluded. Of the remaining patients, 490 prevalent haemodialysis patients provided informed consent prior to participation in the present study. The enrolled patients were treated with stable, regular haemodialysis, using bicarbonate dialysate. Our dialysis programme used high-flux membrane as a standard, defined as a β 2 -M clearance >20 mL/min [ 8 ], such as cellulose triacetate, polysulfone, and ethylenevinyl alcohol. Low-flux membrane was not used in our hospital for at least 2 years before 1999. In most patients, the membranes with a surface area of 1.5–2.0 m 2 were used. Kt/V was calculated according to the formula of Daugirdas [ 9 ]. Dialysate flow was fixed at 500 mL/min, being supplied by a central provider. Blood flow was between 160 and 220 mL/min and each dialysis session was between 3.5 and 4.5 h, in order to achieve the target total Kt/V of around 1.2 per session for three times weekly treatment, as a routine practice in most dialysis facilities in Japan. Dialyzer membrane surface was adjusted to further maintain the target total Kt/V of around 1.2. No reuse of dialyzer was performed in any patient. The clinical diagnoses of primary renal disease were chronic glomerulonephritis ( n = 244), diabetic nephropathy ( n = 120), hypertensive nephrosclerosis ( n = 32), polycystic kidney disease ( n = 19), toxaemia of pregnancy ( n = 6), gout ( n = 5) and others/unknown ( n = 64). This ratio of chronic glomerulonephritis, diabetic nephropathy and other diseases as primary renal disease was close to the data of the Japanese Dialysis Registry [ 10 ], in which 51.1% of primary renal disease in dialysis patients in 1999 was chronic glomerulonephritis in Japan.

In October 1999, data of the enrolled patients were collected and the observation study was conducted until 31 December 2003. The relationship between the baseline data and outcomes was analysed statistically. This study was approved by the ethics review committee of the hospital.

Biochemical assays and other measurements

Blood was drawn just before the start of a dialysis session in a non-fasting state, to measure serum albumin, creatinine, blood urea nitrogen, C-reactive protein (CRP) and haematocrit, using routine laboratory methods, twice a month as a routine clinical care as performed in most dialysis facilities in Japan. These values were a mean of eight measurements within 4 months. Serum β 2 -M concentration was measured by latex immunoassay [ 11 ] (Mitsubishi-Kagaku Medicine, Tokyo, Japan) at the time of enrolment.

Statistical methods

Data were summarized as the mean ± SD. Differences in the means between two groups were evaluated by unpaired Student's t -test. Data that showed skewed distributions were compared with the Mann–Whitney U -test. Categorical data were compared between groups by the chi-square test. Survival curves were assessed using the Kaplan–Meier analysis and evaluated by the log-rank test. Prognostic variables for survival were initially examined using the univariate Cox proportional hazards method. Multivariate Cox proportional hazards analyses were performed to identify independent factors associated with mortality. In multivariate analyses, age, gender, haemodialysis duration, diabetes and the factors that showed P -values <0.05 on univariate analysis were entered as possible factors associated with mortality. P -values <0.05 were considered statistically significant. All calculations were performed using statistical analysis software StatView 5 (SAS Institute Inc., Cary, NC, USA) on a Windows personal computer.

Results

Serum β 2 -M concentration and haemodialysis duration

Serum β 2 -M concentrations of a total of 490 patients were normally distributed (Figure 1 ), with a mean (±SD) of 32.5 (±7.2) mg/L. Serum β 2 -M concentrations according to haemodialysis duration are shown in Figure 2 . In patients whose dialysis duration was <5 years, serum β 2 -M concentrations correlated significantly with the duration of dialysis, tending to be higher in patients treated for 5 years compared to shorter durations. Compared to serum β 2 -M concentrations in patients with a haemodialysis duration of <1 year ( n = 41), those in patients with durations of 2– 3 years ( n = 47), 5–6 years ( n = 43), 6–8 years ( n = 49), 8–10 years ( n = 45) and 10–14 years ( n = 43) were significantly higher ( P < 0.05). In patients treated for >5 years, this relationship was reversed, with patients treated for >18 years ( n = 45) having significantly lower β 2 -M levels compared to those of patients treated for 1– 2 years ( n = 63), 2–3 years, 3–4 years ( n = 48), 4–5 years ( n = 39), 5–6 years, 6–8 years, 8–10 years, 10–14 years and 14–18 years ( n = 34) ( P < 0.01).

Fig. 1

Distribution of serum β 2 -microglobulin concentrations. Serum β 2 -microglobulin concentrations were distributed normally.

Fig. 1

Distribution of serum β 2 -microglobulin concentrations. Serum β 2 -microglobulin concentrations were distributed normally.

Fig. 2

Serum β 2 -microglobulin concentration according to haemodialysis duration. Serum β 2 -microglobulin concentrations increased until 5 years, reached a plateau level, and then decreased after 18 years. * Denotes P <0.05 versus β 2 -microglobulin concentrations with haemodialysis duration of less than 1 year. # Denotes P < 0.01 versus β 2 -microglobulin concentrations with haemodialysis duration of 1–2, 2–3, 3–4, 4–5, 5–6, 6–8, 8–10, 10–14 and 14–18 years.

Fig. 2

Serum β 2 -microglobulin concentration according to haemodialysis duration. Serum β 2 -microglobulin concentrations increased until 5 years, reached a plateau level, and then decreased after 18 years. * Denotes P <0.05 versus β 2 -microglobulin concentrations with haemodialysis duration of less than 1 year. # Denotes P < 0.01 versus β 2 -microglobulin concentrations with haemodialysis duration of 1–2, 2–3, 3–4, 4–5, 5–6, 6–8, 8–10, 10–14 and 14–18 years.

Clinical characteristics according to the low- and high-β 2 -M groups

Patients were divided into two groups according to the median value of the baseline serum β 2 -M concentration (32.2 mg/L), as follows: a lower β 2 -M group (serum β 2 -M < 32.2 mg/L, n = 245) and a higher β 2 -M group (serum β 2 -M ≥ 32.2 mg/L, n = 245). Table 1 shows the baseline clinical characteristics of the haemodialysis patients in each of the two groups.

Table 1

Baseline characteristics of the haemodialysis patients according to serum β 2 -microglobulin (β 2 -M) concentrations

  Lower β 2 -M group   Higher β 2 -M group  P -value  
 2 -M < 32.2 mg/L) ( n = 245)  2 -M ≥ 32.2 mg/L) ( n = 245)   
Age (years) 59.5 ± 10.8 60.9 ± 12.7 0.196 
Gender (male/female) 143/102 145/100 0.927 
Haemodialysis duration (months) 96.3 ± 85.3 80.4 ± 64.8 0.020 
Diabetes (diabetes/non-diabetes) 51/194 69/176 0.074 
Body mass index (kg/m 2 )  20.9 ± 3.0 20.4 ± 2.8 0.075 
Haematocrit (%) 30.7 ± 3.2 30.8 ± 3.7 0.655 
Albumin (g/dL) 4.1 ± 0.3 4.0 ± 0.4 <0.001 
Blood urea nitrogen (mg/dL) 69.2 ± 12.7 68.3 ± 12.0 0.405 
Creatinine (mg/dL) 10.7 ± 2.2 10.9 ± 2.3 0.292 
C-reactive protein (mg/dL) 0.10 (0.01–4.27) 0.16 (0.01–4.45) 0.002 
Kt/V 1.147 ± 0.245 1.149 ± 0.211 0.919 
  Lower β 2 -M group   Higher β 2 -M group  P -value  
 2 -M < 32.2 mg/L) ( n = 245)  2 -M ≥ 32.2 mg/L) ( n = 245)   
Age (years) 59.5 ± 10.8 60.9 ± 12.7 0.196 
Gender (male/female) 143/102 145/100 0.927 
Haemodialysis duration (months) 96.3 ± 85.3 80.4 ± 64.8 0.020 
Diabetes (diabetes/non-diabetes) 51/194 69/176 0.074 
Body mass index (kg/m 2 )  20.9 ± 3.0 20.4 ± 2.8 0.075 
Haematocrit (%) 30.7 ± 3.2 30.8 ± 3.7 0.655 
Albumin (g/dL) 4.1 ± 0.3 4.0 ± 0.4 <0.001 
Blood urea nitrogen (mg/dL) 69.2 ± 12.7 68.3 ± 12.0 0.405 
Creatinine (mg/dL) 10.7 ± 2.2 10.9 ± 2.3 0.292 
C-reactive protein (mg/dL) 0.10 (0.01–4.27) 0.16 (0.01–4.45) 0.002 
Kt/V 1.147 ± 0.245 1.149 ± 0.211 0.919 

There were no significant differences in age and gender between the two groups. The duration of haemodialysis was significantly longer in the lower β 2 -M group than that in the higher β 2 -M group (96.3 ± 85.3 versus 80.4 ± 64.8 months, P = 0.020). The prevalence of diabetes was lower and body mass index was higher in the former than that in the latter group, although the differences did not reach a statistical significance. There was no significant difference in Kt/V between the two groups ( P = 0.919).

Serum albumin levels were significantly higher in the lower β 2 -M group than those in the higher β 2 -M group (4.1 ± 0.3 versus 4.0 ±0.4 g/dL, P < 0.001), whereas CRP levels were significantly lower in the former group (0.10 versus 0.16 mg/dL, P = 0.002). There was no significant difference in the values of haematocrit, blood urea nitrogen or creatinine between the two groups.

Kaplan–Meier analysis

During the follow-up period of 40 ± 15 months, a total of 91 patients died. The death rate was near to the average of haemodialysis patients in Japan, according to the Japanese Dialysis Registry [ 10 ]. Kaplan–Meier analysis was performed to examine the univariate association between the two groups based on the β 2 -M concentrations and the outcomes of the cohort (Figure 3 ). Patients with higher β 2 -M concentrations exhibited a significantly higher death rate than those with lower β 2 -M concentrations (log-rank test, P < 0.001).

Fig. 3

Kaplan–Meier analysis of all-cause mortality of 490 haemodialysis patients, classified according to lower (<32.2 mg/L, n = 245) and higher (≥32.2 mg/L, n = 245) β 2 -microglobulin (β 2 -M) concentrations. Patients with higher β 2 -M concentrations (thick line) exhibited a significantly higher death rate compared to those with lower β 2 -M concentrations (thin line) (log-rank test, P < 0.001).

Fig. 3

Kaplan–Meier analysis of all-cause mortality of 490 haemodialysis patients, classified according to lower (<32.2 mg/L, n = 245) and higher (≥32.2 mg/L, n = 245) β 2 -microglobulin (β 2 -M) concentrations. Patients with higher β 2 -M concentrations (thick line) exhibited a significantly higher death rate compared to those with lower β 2 -M concentrations (thin line) (log-rank test, P < 0.001).

Univariate analysis with Cox proportional hazards models

Table 2 presents the univariate associations between mortality and other covariates. In addition to higher β 2 -M concentrations, increased age, the presence of diabetes, lower body mass index, lower serum albumin, lower serum creatinine and higher serum CRP concentrations were significant univariate predictors of increased all-cause mortality.

Table 2

Univariate Cox proportional hazards analysis of independent predictors of all-cause mortality in haemodialysis patients

 Hazard ratio 95% CI P -value  
Age (per 1 year) 1.09 1.07–1.12 <0.001 
Gender (male versus female) 0.93 0.62–1.41 0.741 
Haemodialysis duration 0.97 0.93–1.00 0.082 
 (per 1 year)    
Diabetes (diabetes versus 2.25 1.47–3.43 <0.001 
 non-diabetes)    
Body mass index 0.90 0.84–0.96 0.009 
 (per 1 kg/m 2 )     
Haematocrit (per 1%) 0.97 0.91–1.04 0.397 
Albumin (per 1 g/dL) 0.17 0.11–0.26 <0.001 
Creatinine (per 1 mg/dL) 0.70 0.64–0.77 <0.001 
Log C-reactive protein 2.44 1.81–3.28 <0.001 
 (per 1 log-unit)    
β 2 -Microglobulin  1.05 1.02–1.08 <0.001 
 (per 1 mg/L)    
 Hazard ratio 95% CI P -value  
Age (per 1 year) 1.09 1.07–1.12 <0.001 
Gender (male versus female) 0.93 0.62–1.41 0.741 
Haemodialysis duration 0.97 0.93–1.00 0.082 
 (per 1 year)    
Diabetes (diabetes versus 2.25 1.47–3.43 <0.001 
 non-diabetes)    
Body mass index 0.90 0.84–0.96 0.009 
 (per 1 kg/m 2 )     
Haematocrit (per 1%) 0.97 0.91–1.04 0.397 
Albumin (per 1 g/dL) 0.17 0.11–0.26 <0.001 
Creatinine (per 1 mg/dL) 0.70 0.64–0.77 <0.001 
Log C-reactive protein 2.44 1.81–3.28 <0.001 
 (per 1 log-unit)    
β 2 -Microglobulin  1.05 1.02–1.08 <0.001 
 (per 1 mg/L)    

CI, confidence interval.

Multivariate analysis with Cox proportional hazards models

Multivariate Cox proportional hazards analyses were performed to identify the independent predictors of mortality. As shown in Table 3 , β 2 -M concentration was a significant, independent predictor of mortality [hazard ratio (per 1 mg/L increase) 1.05 (95% CI 1.01–1.08), P = 0.005] in addition to age, gender, serum creatinine and serum CRP.

Table 3

Multivariate Cox proportional hazards analysis of independent predictors of all-cause mortality in haemodialysis patients

 Hazard ratio 95% CI P -value  
Age (per 1 year) 1.07 1.04–1.09 <0.001 
Gender (male versus female) 1.95 1.20–13.2 0.007 
Haemodialysis duration  (per 1 year) 1.02 0.98–1.06 0.370 
Diabetes (diabetes versus  non-diabetes) 1.56 0.98–2.50 0.063 
Body mass index  (per 1 kg/m 2 )  1.01 0.92–1.10 0.859 
Albumin (per 1 g/dL) 0.88 0.44–1.77 0.716 
Creatinine (per 1 mg/dL) 0.74 0.64–0.87 <0.001 
Log C-reactive protein  (per 1 log-unit) 1.45 1.03–20.5 0.034 
β 2 -Microglobulin  (per 1 mg/L)  1.05 1.01–1.08 0.005 
Global model significance   <0.001 
 Hazard ratio 95% CI P -value  
Age (per 1 year) 1.07 1.04–1.09 <0.001 
Gender (male versus female) 1.95 1.20–13.2 0.007 
Haemodialysis duration  (per 1 year) 1.02 0.98–1.06 0.370 
Diabetes (diabetes versus  non-diabetes) 1.56 0.98–2.50 0.063 
Body mass index  (per 1 kg/m 2 )  1.01 0.92–1.10 0.859 
Albumin (per 1 g/dL) 0.88 0.44–1.77 0.716 
Creatinine (per 1 mg/dL) 0.74 0.64–0.87 <0.001 
Log C-reactive protein  (per 1 log-unit) 1.45 1.03–20.5 0.034 
β 2 -Microglobulin  (per 1 mg/L)  1.05 1.01–1.08 0.005 
Global model significance   <0.001 

CI, confidence interval.

Multivariate analysis of cardiovascular and non-cardiovascular mortality with Cox proportional hazards models

Of the 91 patients who died, 36 patients died from fatal cardiovascular events, as defined previously [ 12,13 ]. Cardiovascular deaths included death from ischaemic heart disease ( n = 8), congestive heart failure ( n = 14), cerebrovascular disease ( n = 6) and sudden death ( n = 8). Non-cardiovascular deaths consisted of infectious disease ( n = 22), cancer ( n = 8), hepatic cirrhosis ( n = 3) and others ( n = 22). As shown in Table 4 , higher serum β 2 -M concentration was a significant, independent predictor of increased mortality from non-cardiovascular disease [hazard ratio (per 1 mg/L increase) 1.06 (95% CI 1.02–1.10), P = 0.006] after adjustment for other confounders, but not from cardiovascular disease.

Table 4

Multivariate Cox proportional hazards analysis of independent predictors of cardiovascular and non-cardiovascular mortality in haemodialysis patients

 Cardiovascular mortality Non-cardiovascular mortality 
 Hazard ratio 95% CI P -value  Hazard ratio 95% CI P -value  
Age (per 1 year) 1.09 1.05–1.13 <0.001 1.06 1.02–1.09 <0.001 
Gender (male versus female) 1.70 0.80–3.64 0.170 2.18 1.17–4.06 0.014 
Haemodialysis duration (per 1 year) 0.99 0.92–1.07 0.790 1.03 0.98–1.08 0.212 
Diabetes (diabetes versus non-diabetes) 2.56 1.28–5.13 0.008 1.09 0.58–2.05 0.800 
Body mass index (per 1 kg/m 2 )  – – – 1.02 0.91–1.15 0.696 
Albumin (per 1 g/dL) 1.69 0.55–5.23 0.360 0.59 0.24–1.45 0.246 
Creatinine (per 1 mg/dL) 0.76 0.60–0.96 0.020 0.73 0.60–0.89 0.001 
Log C-reactive protein (per 1 log-unit) 2.67 1.49–4.79 0.001 1.01 0.65–1.56 0.967 
β 2 -Microglobulin (per 1 mg/L)  1.03 0.98–1.09 0.252 1.06 1.02–1.10 0.006 
Global model significance  <0.001   <0.001  
 Cardiovascular mortality Non-cardiovascular mortality 
 Hazard ratio 95% CI P -value  Hazard ratio 95% CI P -value  
Age (per 1 year) 1.09 1.05–1.13 <0.001 1.06 1.02–1.09 <0.001 
Gender (male versus female) 1.70 0.80–3.64 0.170 2.18 1.17–4.06 0.014 
Haemodialysis duration (per 1 year) 0.99 0.92–1.07 0.790 1.03 0.98–1.08 0.212 
Diabetes (diabetes versus non-diabetes) 2.56 1.28–5.13 0.008 1.09 0.58–2.05 0.800 
Body mass index (per 1 kg/m 2 )  – – – 1.02 0.91–1.15 0.696 
Albumin (per 1 g/dL) 1.69 0.55–5.23 0.360 0.59 0.24–1.45 0.246 
Creatinine (per 1 mg/dL) 0.76 0.60–0.96 0.020 0.73 0.60–0.89 0.001 
Log C-reactive protein (per 1 log-unit) 2.67 1.49–4.79 0.001 1.01 0.65–1.56 0.967 
β 2 -Microglobulin (per 1 mg/L)  1.03 0.98–1.09 0.252 1.06 1.02–1.10 0.006 
Global model significance  <0.001   <0.001  

CI, confidence interval.

Discussion

In the present observational study, we examined the serum β 2 -M concentration in maintenance haemodialysis patients, and evaluated the prognostic significance of the baseline β 2 -M concentrations. We found that serum β 2 -M concentrations were a significant predictor for all-cause mortality, particularly for non-cardiovascular mortality, after adjustment for age, gender, haemodialysis duration, presence of diabetes, serum albumin and serum CRP. Our study examining prevalent haemodialysis patients demonstrated that lower serum β 2 -M concentrations were significantly preferable for better survival of haemodialysis patients.

Lowering serum β 2 -M concentrations has been emphasized in dialysis practice. One of the reasons is that the levels of serum β 2 -M serve as a surrogate marker of uraemic toxins of other middle molecules that have similar systemic or extracorporeal kinetics in dialysis patients [ 2 , 14 ]. Another reason is that β 2 -M is a precursor of amyloidosis in haemodialysis patients [ 2,3 , 15 ]. Reduction of serum β 2 -M has been demonstrated to be effective for the treatment of dialysis-related amyloidosis, through the use of haemodiafiltration, ultrapure dialysate, high-flux membrane and adsorptive affinity column for β 2 -M [ 2–7 ]. Reduction of serum β 2 -M has been demonstrated through the use of long dialysis and daily dialysis [ 16,17 ].

Some studies have reported the influence of middle-molecule clearance on dialysis patients’ survival [ 18 ]. As for the relationship between mortality and serum β 2 -M concentration, however, there have been no reports on this matter, except for the results of the HEMO study [ 7 ]. In the HEMO study, serum β 2 -M concentration correlated significantly with mortality, after adjustment for other confounders. However, serum β 2 -M levels predicted mortality in the subgroup of patients who had been on dialysis for <3.7 years (mean duration of dialysis in the HEMO study), but not in patients who had been on dialysis for a duration longer than 3.7 years. The HEMO study showed that the predictive value of serum β 2 -M was stronger in the subgroup of patients with detectable residual kidney function, although the predictive value was marginal in patients with anuria at baseline. They considered that the residual renal function with a shorter duration of haemodialysis, which significantly lowered the β 2 -M concentration, might be an important determinant of the predictive effect of β 2 -M concentration on all-cause mortality. However, in the present study in which patients with longer haemodialysis durations of 87.4 ± 75.7 months (7.3 ± 6.3 years) were examined, Cox proportional hazards analysis demonstrated that serum β 2 -M concentration was a significant predictor for all-cause mortality after adjustment for several confounders. Furthermore, a Cox proportional hazards analysis in patients with longer haemodialysis durations of >5 years ( n = 259), in whom residual kidney function was negligible, serum β 2 -M concentration was also a significant predictor for all-cause mortality (hazard ratio 1.071, 95% CI 1.009–1.137, P = 0.0234), after adjustment for several confounders. The result suggests that serum β 2 -M is a predictor for all-cause mortality even in longer term haemodialysis patients, without the effect of residual renal function. The reason for the difference seen between the HEMO study and the present study remains unknown. One explanation may be the fact that, in the present study, the haemodialysis duration (87.4 ± 75.7 months) was longer than that of the HEMO study, with mean haemodialysis duration of 3.7 years. Patients with lower serum β 2 -M concentration and longer duration of haemodialysis may live longer among our study patients, as suggested by the fact that serum β 2 -M concentration was significantly lower in patients with haemodialysis durations longer than 18 years. Another reason for the difference between the results of the HEMO study and the present study may be the fact that the low-flux membrane was not used in any of the patients in the present study during the study period, and that the high-flux membrane was never reused in our study, whereas in the HEMO study ∼40% of the patients used a low-flux membrane with β 2 -M clearance <10 mL/min and reuse of dialyzers was widespread in the HEMO study [ 19 ]. They demonstrated that the high-flux membrane, which significantly lowered β 2 -M concentration, was associated with lower risk of all-cause mortalities in a subgroup that had been on dialysis for >3.7 years [ 8 ]. The effect of sole use of high-flux membranes without reuse in the present study may reduce the middle-molecule uraemic toxins, leading to improved survival of patients with lower serum β 2 -M and longer haemodialysis duration in the present study. The high-flux membrane used in the present study may have some effect other than lowering serum β 2 -M, such as better biocompatibility [ 20 ] and adsorption of pyrogen [ 2 ], leading to lower mortality in patients with lower serum β 2 -M caused by use of a high-flux membrane.

Concerning the relationship between the serum β 2 -M concentration and haemodialysis duration, serum β 2 -M concentrations have been reported to be increased during the early phase of haemodialysis. Fly et al . reported that serum β 2 -M concentrations increased as the haemodialysis duration increased [ 6 ]. In their study, residual renal function was of importance as a determinant of serum β 2 -M concentrations during the early years of haemodialysis. In their analysis of the HEMO study, Cheung et al . also reported that the mean predialysis β 2 -M levels continued to increase during follow-up in both the low-flux and high-flux groups [ 7 ]. In the present study analysing a total of 490 patients whose mean haemodialysis durations were relatively long (87.4 ± 75.6 months), compared to the previous reports [ 6,7 ], we also found that serum β 2 -M levels continued to be higher up to 5 years of haemodialysis duration (Figure 2 ). This elevation may be due to the gradual decrease in residual urine as demonstrated by Fly et al . [ 6 ]. After 5 years, β 2 -M levels were at a plateau level. This result is consistent with that reported by Canaud et al ., who reported that serum β 2 -M levels reached a plateau levels after 10 years of dialysis treatment [ 21 ]. However, after 18 years, serum β 2 -M levels were lower (Figure 2 ). No previous studies have demonstrated lower levels of serum β 2 -M in patients with extremely long-term haemodialysis, except for an earlier report by Charra et al ., who reported lower serum β 2 -M levels at 20 years of haemodialysis treatment with use of a low-flux membrane, cuprophane [ 22 ]. The reason why serum β 2 -M levels are lower in long-term haemodialysis patients remains unknown. However, based on the findings of Kaplan–Meier analysis and Cox proportional hazards analysis in the present study, it may be considered that patients with higher levels of serum β 2 -M do not live longer, whereas patients with lower serum β 2 -M levels do. This may be the reason why serum β 2 -M levels were lower in patients with extremely longer duration of haemodialysis.

Our present study suggests that lowering serum β 2 -M levels is preferable for improving survival of haemodialysis patients. Compared to low-flux haemodialysis, which does not remove circulating β 2 -M, high-flux haemodialysis is associated with a significant clearance of circulating β 2 -M [ 2 , 16 ]. Although primary data analysis of the HEMO study did not prove a benefit of high-flux compared to low-flux membranes [ 19 ], several studies have demonstrated better outcomes in dialysis patients treated with high-flux membranes [ 23–26 ]. It was reported that the reduction ratio of β 2 -M per session was 20–30% higher with online haemodiafiltration than that with high-flux haemodialysis (72.7 versus 49.7%), and that regular use of online haemodiafiltration significantly reduced circulating levels of predialysis β 2 -M (median value 20 mg/L) [ 27 ]. Recently, high-efficiency haemodiafiltration patients were shown to have a significantly lower mortality risk than those receiving low-flux haemodialysis [ 28,29 ]. Taken together, removal of middle-molecule uraemic toxins, as represented by β 2 -M as a surrogate marker, by the use of high-flux membrane or haemodiafiltration may improve the survival of dialysis patients.

Our study, which demonstrated significantly improved survival in patients with lower serum β2-M may also indicate that removal of middle-molecule uraemic toxins is important in dialysis practice. Middle-molecular uraemic toxins may impact nutritional status or have immunosuppressive effects [ 30,31 ]. Indeed, in our present study, serum albumin levels were significantly lower and serum CRP were significantly higher in patients with higher serum β 2 -M, compared to those with lower serum β 2 -M. The HEMO study also demonstrated a negative correlation between serum β2-M levels and serum albumin. This may also indicate that serum levels of β 2 -M, and/or middle-molecular uraemic toxins, are associated with nutritional status and inflammation. Although poor nutritional status and chronic inflammation have been reported to be associated with cardiovascular mortality [ 32,33 ], they were also reported to be significantly related to non-cardiovascular mortality [ 34 ]. In an earlier study by Canaud et al . [ 21 ], it was reported that patients with high values of β2-M were observed in patients presenting with severe intercurrent disease, such as cancer. Taken together, these facts may explain why serum β 2 -M levels were associated significantly with non-cardiovascular mortality in the present study. In a recently published article by Cheung et al . [ 35 ], a significant association between serum β2-M levels and infectious mortality in haemodialysis patients were reported. In their article, they could not demonstrate significant association between serum β 2 -M levels and cardiovascular mortality in haemodialysis patients. Our results studying Japanese haemodialysis patients, who relatively live longer and have longer periods of haemodialysis due to less chances to receive kidney transplantation, compared to other countries [ 10 ], are consistent with the recent results reported by Cheung et al .

There were some limitations in this study. First, at the start of this study, we could not obtain data regarding serum lipids, smoking history, residual urine or history of cardiovascular or non-cardiovascular disease, which may affect survival of haemodialysis patients. Second, we could not include the information regarding medications, such as lipid-lowering drugs, calcium carbonate or calcium acetate and vitamin D, which could affect patient survival. In the analyses excluding these data, our data may reflect inadequate adjustments for these known risk or beneficial factors. However, serum β 2 -M levels in dialysis patients have not been reported to be associated with the above-noted clinical history or medications [ 3 , 7 , 36 ]. Although we analysed a cohort of prevalent haemodialysis patients who may have a limitation in elucidating the effect of β 2 -M on survival, we found that, after adjustment for several variables that were significant in univariate Cox proportional hazards analysis, serum β 2 -M levels remained associated significantly with both all-cause and non-cardiovascular mortality in multivariate Cox proportional analysis. Influence of β 2 -M on mortality should be examined in incident haemodialysis patients in future studies, to clearly confirm the results of our findings.

In conclusion, in an observational study on prevalent haemodialysis patients, we demonstrated that serum β 2 -M levels were a significant predictor of mortality in haemodialysis patients with relatively longer haemodialysis durations of 87.4 ± 75.7 months, independent of haemodialysis duration, diabetes, malnutrition and chronic inflammation, suggesting the clinical importance of lowering serum β 2 -M in these patients.

Conflict of interest statement . None declared.

References

1
Orlando
RA
Rader
K
Authier
F
, et al.  . 
Megalin is an endocytic receptor for insulin
J Am Soc Nephrol
 , 
1998
, vol. 
9
 (pg. 
1759
-
1766
)
2
Lonnemann
G
Koch
KM
Beta(2)-microglobulin amyloidosis: effects of ultrapure dialysate and type of dialyzer membrane
J Am Soc Nephrol
 , 
2002
, vol. 
13
 
Suppl 1
(pg. 
S72
-
S77
)
3
Kazama
JJ
Maruyama
H
Gejyo
F
Reduction of circulating beta2-microglobulin level for the treatment of dialysis-related amyloidosis
Nephrol Dial Transplant
 , 
2001
, vol. 
16
 
Suppl 4
(pg. 
31
-
35
)
4
Bolasco
P
Altieri
P
Andrulli
S
, et al.  . 
Convection versus diffusion in dialysis: an Italian prospective multicentre study
Nephrol Dial Transplant
 , 
2003
, vol. 
18
 
Suppl 7
pg. 
vii50–vii54
  
discussion (vii59–vii62)
5
Locatelli
F
Di Filippo
S
Manzoni
C
Removal of small and middle molecules by convective techniques
Nephrol Dial Transplant
 , 
2000
, vol. 
15
 
Suppl 2
(pg. 
37
-
44
)
6
Fry
AC
Singh
DK
Chandna
SM
, et al.  . 
Relative importance of residual renal function and convection in determining beta-2-microglobulin levels in high-flux haemodialysis and on-line haemodiafiltration
Blood Purif
 , 
2007
, vol. 
25
 (pg. 
295
-
302
)
7
Cheung
AK
Rocco
MV
Yan
G
, et al.  . 
Serum beta-2 microglobulin levels predict mortality in dialysis patients: results of the HEMO study
J Am Soc Nephrol
 , 
2006
, vol. 
17
 (pg. 
546
-
555
)
8
Cheung
AK
Levin
NW
Greene
T
, et al.  . 
Effects of high-flux hemodialysis on clinical outcomes: results of the HEMO study
J Am Soc Nephrol
 , 
2003
, vol. 
14
 (pg. 
3251
-
3263
)
9
Daugirdas
JT
Second generation logarithmic estimates of single-pool variable volume Kt/V: an analysis of error
J Am Soc Nephrol
 , 
1993
, vol. 
4
 (pg. 
1205
-
1213
)
10
Japanese Society for Dialysis Therapy
An overview of regular dialysis treatment in Japan as of Dec. 31, 2002
2003
Tokyo
Japanese Society for Dialysis Therapy
11
Bernard
AM
Vyskocil
A
Lauwerys
RR
Determination of beta 2-microglobulin in human urine and serum by latex immunoassay
Clin Chem
 , 
1981
, vol. 
27
 (pg. 
832
-
837
)
12
Kakiya
R
Shoji
T
Tsujimoto
Y
, et al.  . 
Body fat mass and lean mass as predictors of survival in hemodialysis patients
Kidney Int
 , 
2006
, vol. 
70
 (pg. 
549
-
556
)
13
Shoji
T
Shinohara
K
Kimoto
E
, et al.  . 
Lower risk for cardiovascular mortality in oral 1alpha-hydroxy vitamin D3 users in a haemodialysis population
Nephrol Dial Transplant
 , 
2004
, vol. 
19
 (pg. 
179
-
184
)
14
Vanholder
R
De Smet
R
Glorieux
G
, et al.  . 
Review on uremic toxins: classification, concentration, and interindividual variability
Kidney Int
 , 
2003
, vol. 
63
 (pg. 
1934
-
1943
)
15
Drueke
TB
Beta2-microglobulin and amyloidosis
Nephrol Dial Transplant
 , 
2000
, vol. 
15
 
Suppl 1
(pg. 
17
-
24
)
16
Tattersall
J
Clearance of beta-2-microglobulin and middle molecules in haemodiafiltration
Contrib Nephrol
 , 
2007
, vol. 
158
 (pg. 
201
-
209
)
17
Goldfarb-Rumyantzev
AS
Leypoldt
JK
Nelson
N
, et al.  . 
A crossover study of short daily haemodialysis
Nephrol Dial Transplant
 , 
2006
, vol. 
21
 (pg. 
166
-
175
)
18
Vanholder
R
Laecke
SV
Verbeke
F
, et al.  . 
Uraemic toxins and cardiovascular disease: in vitro research versus clinical studies
Nephrol Dial Transplant Plus
 , 
2008
, vol. 
1
 (pg. 
2
-
10
)
19
Eknoyan
G
Beck
GJ
Cheung
AK
, et al.  . 
Effect of dialysis dose and membrane flux in maintenance hemodialysis
N Engl J Med
 , 
2002
, vol. 
347
 (pg. 
2010
-
2019
)
20
Hakim
RM
Held
PJ
Stannard
DC
, et al.  . 
Effect of the dialysis membrane on mortality of chronic hemodialysis patients
Kidney Int
 , 
1996
, vol. 
50
 (pg. 
566
-
570
)
21
Canaud
B
Assounga
A
Flavier
JL
, et al.  . 
Beta-2 microglobulin serum levels in maintenance dialysis. What does it mean?
ASAIO Trans
 , 
1988
, vol. 
34
 (pg. 
923
-
929
)
22
Charra
B
Laurent
G
Calemard
E
, et al.  . 
Serum beta2-microglobulin (beta2-MG) decrease in hemodialysis (HD) patients: a long-term follow-up
Blood Purif
 , 
1988
, vol. 
7
 (pg. 
249
-
250
)
23
Koda
Y
Nishi
S
Miyazaki
S
, et al.  . 
Switch from conventional to high-flux membrane reduces the risk of carpal tunnel syndrome and mortality of hemodialysis patients
Kidney Int
 , 
1997
, vol. 
52
 (pg. 
1096
-
1101
)
24
Port
FK
Wolfe
RA
Hulbert-Shearon
TE
, et al.  . 
Mortality risk by hemodialyzer reuse practice and dialyzer membrane characteristics: results from the USRDS dialysis morbidity and mortality study
Am J Kidney Dis
 , 
2001
, vol. 
37
 (pg. 
276
-
286
)
25
Woods
HF
Nandakumar
M
Improved outcome for haemodialysis patients treated with high-flux membranes
Nephrol Dial Transplant
 , 
2000
, vol. 
15
 
Suppl 1
(pg. 
36
-
42
)
26
Chauveau
P
Nguyen
H
Combe
C
, et al.  . 
Dialyzer membrane permeability and survival in hemodialysis patients
Am J Kidney Dis
 , 
2005
, vol. 
45
 (pg. 
565
-
571
)
27
Wizemann
V
Lotz
C
Techert
F
, et al.  . 
On-line haemodiafiltration versus low-flux haemodialysis. A prospective randomized study
Nephrol Dial Transplant
 , 
2000
, vol. 
15
 
Suppl 1
(pg. 
43
-
48
)
28
Canaud
B
Bragg-Gresham
JL
Marshall
MR
, et al.  . 
Mortality risk for patients receiving hemodiafiltration versus hemodialysis: European results from the DOPPS
Kidney Int
 , 
2006
, vol. 
69
 (pg. 
2087
-
2093
)
29
Jirka
T
Cesare
S
Di Benedetto
A
, et al.  . 
Mortality risk for patients receiving hemodiafiltration versus hemodialysis
Kidney Int
 , 
2006
, vol. 
70
 pg. 
1524
  
author reply 1524–1525
30
Coyne
DW
Dagogo-Jack
S
Klein
S
, et al.  . 
High-flux dialysis lowers plasma leptin concentration in chronic dialysis patients
Am J Kidney Dis
 , 
1998
, vol. 
32
 (pg. 
1031
-
1035
)
31
Haag-Weber
M
Mai
B
Cohen
G
, et al.  . 
GIP and DIP: a new view of uraemic toxicity
Nephrol Dial Transplant
 , 
1994
, vol. 
9
 (pg. 
346
-
347
)
32
Wanner
C
Metzger
T
C-reactive protein a marker for all-cause and cardiovascular mortality in haemodialysis patients
Nephrol Dial Transplant
 , 
2002
, vol. 
17
 
Suppl 8
(pg. 
29
-
32
discussion 39–40
33
Yeun
JY
Levine
RA
Mantadilok
V
, et al.  . 
C-reactive protein predicts all-cause and cardiovascular mortality in hemodialysis patients
Am J Kidney Dis
 , 
2000
, vol. 
35
 (pg. 
469
-
476
)
34
den Elzen
WP
van Manen
JG
Boeschoten
EW
, et al.  . 
The effect of single and repeatedly high concentrations of C-reactive protein on cardiovascular and non-cardiovascular mortality in patients starting with dialysis
Nephrol Dial Transplant
 , 
2006
, vol. 
21
 (pg. 
1588
-
1595
)
35
Cheung
AK
Greene
T
Leypoldt
JK
, et al.  . 
Association between serum 2-microglobulin level and infectious mortality in hemodialysis patients
Clin J Am Soc Nephrol
 , 
2008
, vol. 
3
 (pg. 
69
-
77
)
36
Leypoldt
JK
Kinetics of beta2-microglobulin and phosphate during hemodialysis: effects of treatment frequency and duration
Semin Dial
 , 
2005
, vol. 
18
 (pg. 
401
-
408
)

Comments

0 Comments