Dr Takayasu Ohtake, Department of Nephrology, Immunology, and Vascular Medicine, Shonan Kamakura General Hospital, 1202-1 Yamazaki, Kamakura, 247-8533, Japan. Email: firstname.lastname@example.org
Atherosclerotic complications have a significant effect on mortality in patients undergoing hemodialysis (HD) therapy. However, anti-atherosclerotic and cardioprotective effects of on-line hemodiafiltration (HDF) remain to be elucidated. We prospectively compared the anti-atherosclerotic and cardioprotective effects in two randomly divided groups, i.e. on-line HDF group (n = 13) and conventional HD group (n = 9) for 1 year. Surrogate markers were brachial-ankle pulse wave velocity (baPWV), intima-media thickness (IMT) of carotid artery as an atherosclerosis marker, and cardiac functional surrogate markers included left ventricular mass index (LVMI), ejection fraction (EF), and LV diastolic capacity represented as E/A and deceleration time (DT). LVMI in on-line HDF patients showed significant regression after 1 year of treatment (131.9 ± 25.8 to 116.5 ± 24.7 g/m2, P = 0.03), while LVMI in HD patients did not show any significant change (148.0 ± 47.1 to 142.3 ± 35.5 g/m2). Levels of baPWV in HD patients showed a significant increase (11.4%) from basal levels, while on-line HDF groups showed no significant increase. Furthermore, HD patients showed significant worsening of LV diastolic capacity (E/A: from 0.87 ± 0.12 to 0.79 ± 0.08, P = 0.03), while it was not shown in on-line HDF patients. Ejection fraction and IMT did not show any significant change in both groups. Serum albumin, C-reactive protein, β2 microglobulin, blood pressure, and anti-hypertensive drug use did not change in both groups. On-line HDF showed a significant improvement in LVMI and prevented a significant worsening of baPWV or LV diastolic capacity compared with patients on conventional HD therapy.
Atherosclerotic complications including ischemic cardiac disease (IHD), stroke, and peripheral arterial disease (PAD) are serious problems and have a significant effect on mortality in patients on dialysis therapy (1–7). Since Lindner et al. (8) described that atherosclerosis is accelerated in dialysis patients, atherosclerotic problems in hemodialysis (HD) patients has been attracting much concern in nephrologists.
Hemodiafiltration (HDF), which was first reported in the literature in 1985 (9), has now been used as a dialysis modality. HDF can remove middle-sized uremic solutes more efficiently than diffusion-dependent conventional HD by combining convective clearance with diffusion. HDF therapy has some variations according to the differences of dilution methods such as pre-dilution, post-dilution, mixed-dilution (pre- and post-), and push-pull methods. Differences of substitution fluid also cause variations of HDF such as bottle-type HDF (off-line HDF), on-line HDF, and push-pull HDF. Direct infusion of dialysate into the blood line also as a substitution fluid is the most important methodological characteristic of on-line HDF. Therefore, purification of dialysate should be strictly controlled to perform on-line HDF.
Using ultra-pure dialysate not only as dialysate but also as substitution fluid, on-line HDF stabilizes the patients' circulation status during dialysis sessions better than the standard HD (10,11). On-line HDF has also been reported to improve anemia (12,13), skin lesions (13), amyloidosis (14), and bone complications (15). However, the anti-atherosclerotic effect of on-line HDF has never been reported in the literature.
Atherosclerosis surrogate markers such as left ventricular mass index (LVMI) (16), pulse wave velocity (PWV) (17), ankle-brachial pressure index (ABI) (18), and intima-media thickness of carotid artery (IMT) (19) are reported to be independent predictors of cardiac events and/or prognosis in HD patients. Therefore, the purpose of this study was to verify whether on-line HDF has anti-atherosclerotic and cardioprotective effects in patients with end-stage renal failure using several atherosclerosis surrogate markers including LVMI, PWV, ABI, and IMT along with left ventricular systolic and diastolic functional markers.
PATIENTS AND METHODS
Patients and study protocol
The potential subjects comprised stable outpatients who had been undergoing HD therapy 4 h per HD session three times a week in April 2007 at the dialysis center of our hospital. Patients aged over 80 years were excluded. Other exclusion criteria were patients with acute infection or hospitalization within 4 weeks before study entry, HD duration less than 6 months, functional failure of arteriovenous fistula with less than 5 mL/kg per min or more blood flow, malignancy, pregnancy, severely suppressed cardiac function (EF < 40%) and/or severe arrhythmia, and dialysis difficulty due to unstable intradialytic blood pressure status. After giving informed consent, 22 prevalent stable HD outpatients in our dialysis center were finally enrolled to this study in May 2007. The study protocol was approved by the local ethical committee in our hospital.
After stratification for age and dialysis duration, they were randomly divided into two groups; i.e. conventional HD group (nine patients) and on-line HDF group (13 patients). Randomization was conducted using a table of random numbers. Atherosclerotic surrogate markers and echocardiographic examination described below were evaluated at study entry and 1 year after the study entry. Clinical and laboratory parameters noted below were also registered at study entry and 1 year after the study entry. Dry weight was evaluated monthly by cardiopulmonary ratio on chest X-ray, blood levels of human atrial natriuretic peptide (hANP) after HD session, and the status of blood pressure control. Control of blood pressure in patients in both groups was done by resetting the dry weight and/or adding antihypertensive agents including calcium channel blocker and/or α-blocker. Angiotensin type II receptor blocker (ARB) and angiotensin converting enzyme inhibitor (ACE-I) were not newly added during the study period. Doses and types of ARB and ACE-I in patients who had already been prescribed ARB and/or ACE-I at study entry were not changed during the study period.
Method for HD and on-line HDF
High performance polysulfone membranes (APS-Ex, Asahi Kasei Kuraray Medical Co. Ltd, Tokyo, Japan) were used for all patients in both the HD and on-line HDF groups. The evaluation of dialysate purity was made by measurements of endotoxin concentration (Endospecy ES-24S kit, Seikagaku Biobusiness Corporation, Tokyo, Japan) and bacterial culture in our hospital every other week. Qb was set at 5.0 to 6.0 mL/min per kg body weight, and Qd was set at 500 mL/min of dialysis fluid (Kindary 3E, Fuso Pharmaceutical Industries, Osaka, Japan) in both groups. The composition of Kindary 3E (dialysate) is Na+ 140 mEq/L, K+ 2.0 mEq/L, Cl- 114.5 mEq/L, Ca (2+) 2.5 mEq/L, Mg (2+) 1.0 mEq/L, HCO3- 25.0 mmol/L, glucose 150 mg/dL, and CH3COO- 8 mEq/L. 40 L sterile dialysate was directly infused as substitution fluid in on-line HDF patients from the pre dialyzer chamber (pre dilution method) during 4-h on-line HDF sessions.
Echocardiography and atherosclerosis surrogate markers
All measurements of echocardiography and atherosclerotic surrogate markers were performed on the next day of second blood purification session of the week. These measurements and evaluation of records were blinded for treatment allocation.
Echocardiographic measurements were carried out for all patients according to the recommendations of the American Society of Echocardiography (20). Patients were examined by echocardiography to determine the ejection fraction (EF), left ventricular mass index (LVMI), peak early diastolic left ventricular filling velocity/peak atrial filling velocity ratio (E/A), and deceleration time (DT) within a week before initiation of the study, EF was used for the evaluation of LV systolic function, and E/A and DT were used for the evaluation of LV diastolic function, respectively.
Form ABI (Colin, Co., Ltd, Komaki, Japan) was used for measurements of brachial-ankle PWV (baPWV), ABI, and toe-brachial pressure index (TBI) as previously reported (18). The instrument simultaneously records the blood pressure in the four extremities, and can also calculate pulse wave velocity between brachial and ankle portion. ABI and TBI were calculated as the ratio of ankle and toe to brachial systolic blood pressure respectively, and were used for the assessment of the existence of atherosclerotic stenotic vascular lesions of lower extremities. The value of baPWV was used for the surrogate of arterial stiffness.
According to our previous report (21), both carotid arteries were examined by high-resolution B-mode ultrasound (Aloka, Co., Ltd, Tokyo, Japan) using a 7.5 MHz linear array transducer. This method could measure the arterial wall thickness in 0.1 mm increments. The carotid arteries were examined bilaterally at the following three sites: the common carotid artery (1 cm proximal to the carotid bulb), the carotid bifurcation (1 cm proximal to the division of flow), and the internal carotid artery (1 cm distal to the division of flow). The intima-media thickness (IMT) was defined as the distance between the leading edge of the lumen-intima echo of the near wall and the leading edge of the media-adventitia echo. Measurement of IMT was done blindly by two carefully trained sonographers. In a separate series of IMT measurements performed in other patients, the interobserver coefficient of variation was 8%. To enhance the reproducibility of measurements, standardized interrogation angles were used. The maximum value of the carotid intima-media thickness (max IMT) was recorded and used for analysis.
Clinical and laboratory parameters
Clinical parameters including age, sex, HD duration, cause of renal failure, comorbid atherosclerotic diseases (IHD, stroke, PAD), antihypertensive drug use including ACE-I and ARB, and statin use were registered at the study entry. Mean values of pre-HD or pre-on-line HDF systolic and diastolic blood pressures were calculated from the dialysis chart of three dialysis sessions in 1 week and registered. Laboratory parameters including serum urea nitrogen, creatinine, CRP, and β2 microglobulin (β2MG) were examined at the start of the first dialysis session of the week. All of these parameters were registered at study entry and 1 year after study entry in each group.
All data are presented as the mean ± SD. CRP underwent log transformation before statistical analysis. Comparison of mean values between study entry and 1 year later in each group was done with the two-tailed paired Student's t-test or Wilcoxon t-test. Comparison of mean values between two groups was done with the unpaired Student's t-test or Mann–Whitney U-test. Dichotomous variables were compared using the χ2 test and Fisher's test. Stat View 5.0 software for Windows (SAS Institute, Cary, NC, USA) and were used for data analysis on a personal computer and P < 0.05 was considered significant.
Patients and dialysate
As shown in Table 1, the HD group and on-line HDF group did not show any significant difference in clinical and laboratory parameters at study entry. Blood pressure control and the use of anti-hypertensive drugs including ARB and ACE-I were not different between the two groups. The atherosclerotic surrogate markers showed no difference except for baPWV. The mean value of baPWV in the on-line HDF group was higher than that of HD group (1789.1 ± 303.0 vs. 1568.3 ± 401.5 cm/s, P = 0.04). Cardiac functional parameters on echocardiography did not show any difference between the two groups. The endotoxin concentration of dialysate had been ascertained below undetectable levels and bacterial culture did not show any bacterial colony growth during all of the follow-up periods (data not shown).
Table 1. Basic characteristics of the patients at study entry
HD group (n = 9)
On-line HDF group (n = 13)
Data are expressed as mean ± SD. β2MG, β2 microglobulin; ABI, ankle-brachial pressure index; ACE-I, angiotensin converting enzyme inmhibitor; Alb, albumin; ARB, angiotensin type II receptor blocker; CGN, chronic glomerulonephritis; Cr, creatinine; DT, deceleration time, baPWV, brachial-ankle pulse wave velocity; E/A, peak early diastolic left ventricular filling velocity/peak atrial filling velocity ratio; EF, ejection fraction; HDF, hemodialfiltration; IHD, ischemic heart disease; IMT, imtima-media thickness; LVMI, left ventricular mass index; PAD, peripheral arterial disease; TBI, toe-brachial pressure index.
62.4 ± 7.7
58.6 ± 11.3
HD duration (months)
58.8 ± 64.4
64.5 ± 38.2
20.9 ± 3.8
21.8 ± 5.4
Blood pressure (mmHg)
150.3 ± 13.9
157.9 ± 12.1
83.6 ± 8.3
85.9 ± 13.9
Cause of renal failure n (%)
Comorbidity n (%)
Smoking status, n (%)
Drug n (%)
71.1 ± 13.9
70.3 ± 11.5
9.6 ± 2.4
11.3 ± 2.1
3.9 ± 0.2
3.8 ± 0.2
0.14 ± 0.27
0.09 ± 0.08
22.8 ± 6.3
25.5 ± 5.9
LDL cholesterol (mg/dL)
76.8 ± 25.9
82.7 ± 22.1
HDL cholesterol (mg/dL)
47.1 ± 14.7
44.8 ± 13.6
88.6 ± 50.3
74.2 ± 38.5
Cardiac functional parameters
63.2 ± 9.4
64.1 ± 4.1
148.0 ± 47.1
131.9 ± 25.8
0.87 ± 0.12
0.80 ± 0.21
244.7 ± 50.6
223.8 ± 49.7
Atherosclerotic surrogate markers
1568.3 ± 401.5
1789.1 ± 303.0
1.19 ± 0.07
1.22 ± 0.14
0.84 ± 0.25
0.82 ± 0.21
0.84 ± 0.16
0.78 ± 0.13
Comparison between HD and on-line HDF
Table 2 shows the change of several parameters in patients on conventional HD at study entry and 1 year later. The HD group showed a significant decrease in E/A ratio (from 0.87 ± 0.12 to 0.79 ± 0.08, P = 0.03), and significant increase in baPWV (from 1568.3 ± 401.5 to 1745.8 ± 492.6 cm/s, P = 0.013) (Fig. 1). DT, EF, LVMI, ABI, TBI, and IMT did not show a significant difference between study entry and 1 year later in the HD group. Laboratory parameters also did not show any difference between study entry and 1 year later.
Table 2. Changes of parameters in the conventional HD group
1 year after
Data are expressed as mean ± SD. *P < 0.05. β2MG, β2 microglobulin; ABI, ankle-brachial pressure index; Alb, albumin; Cr, creatinine; DT, deceleration time, baPWV, brachial-ankle pulse wave velocity; E/A, peak early diastolic left ventricular filling velocity/peak atrial filling velocity ratio; EF, ejection fraction; IMT, imtima-media thickness; LVMI, left ventricular mass index; TBI, toe-brachial pressure index.
In comparison with the HD group, the on-line HDF group showed a significant regression in LVMI from its basal levels (from 131.9 ± 25.8 to 116.5 ± 24.7 g/m2, P = 0.03) (Table 3 and Fig. 2). Furthermore, a significant decrease in E/A ratio and increase in baPWV shown in the HD group was not indicated in the on-line HDF group (Figs. 1 and 2). Carotid IMT in the on-line HDF group showed a tendency to decrease after 1 year of treatment, although it was not statistically significant (0.78 ± 0.13 to 0.72 ± 0.17 mm, P = 0.10). Control of blood pressure was not significantly different between study entry and 1 year later in each group (Tables 2,3). The absolute values of the pre-dialysis blood pressure also did not show any significant difference between the two groups.
Table 3. Changes of parameters in the on-line hemodialfiltration (HDF) group
1 year after
Data are expressed as mean ± SD. *P < 0.05. aBlood flow rate in HD at study entry. bBlood flow rate in 40 L pre-dilution on-line HDF. β2MG, β2 microglobulin; ABI, ankle-brachial pressure index; Alb, albumin; Cr, creatinine; DT, deceleration time, baPWV, brachial-ankle pulse wave velocity; E/A, peak early diastolic left ventricular filling velocity/peak atrial filling velocity ratio; EF, ejection fraction; IMT, imtima-media thickness; LVMI, left ventricular mass index; TBI, toe-brachial pressure index.
Serum levels of albumin, CRP, and β2MG did not change in both the HD group and the on-line HDF group between study entry and 1 year after treatment (Tables 2,3). Serum small solutes such as urea nitrogen and creatinine showed significant increase in the on-line HDF group 1 year later compared with their basal levels (Table 3).
Our present study clearly demonstrated that on-line HDF has anti-atherosclerotic and cardioprotective effects. A significant LVMI regression, prevention of worsening of LV diastolic functional abnormality and arterial stiffness were shown in on-line HDF group compared with the HD group. These effects were obtained by on-line HDF with 40 L pre dilution method. Furthermore, these effects were independent from bloodpressure control and/or drug use such as ARB and ACE-I.
Arterial stiffness and LVMI are the important predicting factors for prognosis in HD patients (16,17). Arterial stiffness is multifactorial, and many independent associating factors are reported, i.e. age (22), long HD duration (22), vascular calcification (22), blood pressure (23,24), leukocyte aggregates (25), glucose metabolism (26), high molecular adiponectin (27), and inflammation (24). LVMI is also multifactorial (23,28) and closely associates with arterial stiffness (29). Increased afterload by increased aortic/arterial stiffness promotes LV hypertrophy (LVH) by loss of Windkessel pressure compensation mechanism in the aorta. Therefore, it is theoretically acceptable that improvement in arterial stiffness might improve LVH. Although the exact mechanisms are unknown, 1 year on-line HDF treatment improved LVMI without changing arterial stiffness represented as baPWV.
The HD group showed progressive worsening of LV diastolic function, while the on-line HD group did not show significant worsening of LV diastolic function. This may be due to the difference of the change of LVMI between the two groups. Increase of LVMI promotes cardiac dysfunction in both a systolic and diastolic capacity. However, it is reported that LV diastolic function may be disturbed earlier than LV systolic function in HD patients (30). In general, one third of patients with heart failure have normal LV systolic function, and heart failure might be caused by single LV diastolic failure. Therefore, diastolic dysfunction (or diastolic failure) may be very important in consideration of the effect of heart failure on the prognosis of HD patients. Although LV systolic function (EF) did not change in both groups after the 1-year follow up period, LV diastolic function showed a significant decrease in the HD treated group in our present study, while it was not shown in patients with on-line HDF. Whether this difference in the change of LV diastolic function may lead to subsequent differences in LV systolic function between the HDgroup and the on-line HDF group may need a further follow up study.
In contrast to no change in the conventional HD group, serum BUN and creatinine levels significantly increased in the on-line HDF group 1 year later from their basal levels. On this point, Masakane reported an excellent manuscript about how much convective volume would be appropriate to achieve a good clearance of small and middle molecules in pre-dilution and post dilution on-line HDF (31). In pre-dilution HDF, it is certainly not evident that “more is better”. Plasma water may be diluted in pre-dilution HDF, thereby clearance may be attenuated. On the contrary, in post-dilution HDF, more will be probably better. However, convective volumes are limited by high trans-membrane pressure, membrane clotting and machine alarms. There may exist many conditions in performing HDF, especially on-line HDF. Selection of membrane, dilution method (pre-dilution or post-dilution), and convective volume may have to be carefully selected.
Clearance of middle molecules including β2MG is usually expected to increase with HDF therapy. However, contrary to our expectation, serum β2MG did not decrease in the on-line HDF group in our study. Although we do not have a clear explanation, the influence of residual renal function might be one of the reasons why serum β2MG did not decrease in the HDF group (32,33). We did not examine urinary volume and urinary β2MG excretion in the on-line HDF group. Therefore, we could not clarify this possibility.
The present study has some limitations. The numbers of patients were small and the follow up period was short (only 1 year) although we conducted a prospective randomized study. There was no cardiovascular event in patients in both groups during the follow up period. Therefore, we could not clarify whether cardioprotective and anti-atherosclerotic effects of on-line HDF confirmed preliminarily in this study consequently would lead to an improved patients' outcome.
Although the precise mechanisms remain to be elucidated, the present study indicated that on-line hemodiafiltration has anti-atherosclerotic and cardioprotective effects compared with conventional hemodialysis therapy. Left ventricular hypertrophy regressed after 1 year of on-line HDF treatment. Furthermore, on-line HDF also had more beneficial effects in arterial stiffness and LV diastolic function compared with conventional HD therapy. These anti-atherosclerotic and cardioprotective effects were independent from blood pressure control status and/or types of antihypertensive drugs. On-line HDF can remove middle-sized molecule or protein-bound uremic toxins more efficiently than conventional HD. These effects might influence the vascular endothelial function or vascular calcification, retard the progression of peripheral arterial atherosclerosis, and might finally provide cardioprotective effects. A larger scale study would be necessary to further confirm and clarify the mechanisms of cardioprotective and anti-atherosclerotic effects by on-line HDF.