Pediatric Nephrology

, 21:1179

Sodium modeling attenuates rises in whole-blood viscosity during chronic hemodialysis in children with large inter-dialytic weight gain

Authors

    • Division of Kidney Diseases, Department of Pediatrics, Children’s Memorial Hospital, Feinberg School of MedicineNorthwestern University
    • Feinberg School of Medicine, Northwestern UniversityChildren’s Memorial Hospital
  • Ellen R. Brooks
    • Division of Kidney Diseases, Department of Pediatrics, Children’s Memorial Hospital, Feinberg School of MedicineNorthwestern University
  • Craig B. Langman
    • Division of Kidney Diseases, Department of Pediatrics, Children’s Memorial Hospital, Feinberg School of MedicineNorthwestern University
  • Kenneth R. Kensey
    • Rheologics Inc
Original Article

DOI: 10.1007/s00467-006-0101-y

Cite this article as:
Fathallah-Shaykh, S.A., Brooks, E.R., Langman, C.B. et al. Pediatr Nephrol (2006) 21: 1179. doi:10.1007/s00467-006-0101-y

Abstract

Elevated whole-blood viscosity (WBV) is a risk factor for atherosclerosis and thrombosis. We analyzed WBV during hemodialysis (HD) in children and tested the hypothesis that sodium modeling (NaM) attenuates an increase in WBV. Each of six children underwent two control (C) and two NaM HD sessions, B and E. Rapid decline in sodium (Na) concentration occurred at the beginning of HD in B and at the end in E. We measured WBV at different shear rates (SRs) and documented the amount of fluid removed (FR), change in blood volume (BV), and hematocrit (Hct) before, during, and after HD. The percent increase of WBV in control sessions was significantly different at 2 h and 3 h during and after HD from baseline values. The mean percent change in WBV from baseline increased linearly over time during HD (R2>0.90). Hct, FR, and BV correlated with WBV (P<0.05). The effects of NaM on attenuation of WBV were statistically significant in three subjects with >5% inter-dialytic weight gain (IDWG) (P<0.05). WBV increased during HD in children. NaM appears to attenuate the rise in WBV in children with large IDWG.

Keywords

HemodialysisChildrenSodium

Introduction

Children with end stage kidney disease (ESKD) are at a high risk of developing cardiovascular disease [1]. Among those children who die while on maintenance dialysis the cause is cardiovascular in 21% to 57% [13]. Factors that may be responsible for the excess cardiovascular mortality in children and young adults with ESKD include hypertension and volume overload, a high calcium-phosphate product, and chronic inflammation [4].

Whole-blood viscosity (WBV) appears to be a predictor of cardiovascular disease [5]. Elevated WBV is associated with a higher risk of cardiovascular and cerebrovascular disease [611] and has been suggested as a risk of vascular thrombosis that may possibly affect hemodialysis access, such as arteriovenous fistula (AVF) [12,13]. Several investigators have measured WBV in adult patients prior to and following conventional hemodialysis (HD) and demonstrated an increase in WBV at HD completion [14, 15]. Thus, maintenance HD produces repetitive or cyclic increases in WBV, contributing to vascular dysfunction.

Sodium (Na) remains the major indicator of blood tonicity and also determines the distribution of water across the intracellular–extracellular boundary [16]. Na modeling is a simplified mathematical method used to describe quantitatively the fluid exchange in the body caused by changes in the extracellular Na concentration [17]. Theoretically, altering the dialysate Na in a step-wise manner permits a more gradual reduction in osmolality, a gradual adjustment in the Na pool, and results in a higher circulating blood volume, lower hematocrit, and avoidance of hypertonicity [18]. Such influences by the use of Na modeling may lower WBV, since WBV is affected by increases in hematocrit and plasma osmolality [5, 7].

In the present study we measured the change in WBV before, during, and after HD in a small group of children to evaluate whether Na modeling (NaM) attenuates the rise in WBV, as we had hypothesized. Additionally, we explored the changes in serum Na and osmolality as an explanation of changing WBV.

Patients and methods

Six children older than 8 years of age, who had been diagnosed with ESKD and on chronic HD for longer than 3 months, were enrolled. The subjects were clinically stable and had exhibited no evidence of additional systemic illness, uncontrolled blood pressure or infection for at least 1 month prior to participation. Informed consents/assents were obtained from all parents and adolescent subjects. Data were collected during visits that coincided with HD treatment sessions. A history and a physical examination were completed, including collection of data on blood pressure, weight, height and body temperature. The medical history included ethnicity, etiology and duration of chronic kidney disease and maintenance hemodialysis therapy.

During each of the study visits (scheduled once/week and 1 week apart) and following a 12-h fast, each subject underwent one of the following sessions: two control sessions (C) using a 0% Na gradient (Na dialysate concentration=145 mM throughout HD) and two Na modeling sessions, B and E, each assigned 1 week apart in a random order. In both B and E the Na gradient decreased from 145 mM to 135 mM at a rate of 1 mM decline every 30 min in a stepwise manner (Fig. 1). However, in B the Na dialysate concentration was lowered rapidly at the beginning, at a rate of 1 mM every 15 min (over the first 60 min of HD) while in E it was lowered rapidly at a rate of 1 mM every 15 min at the end of the HD session (last 60 min). The subjects were blind to the order of the sessions, which was randomly assigned by the biostatistician.
https://static-content.springer.com/image/art%3A10.1007%2Fs00467-006-0101-y/MediaObjects/467_2006_101_Fig1_HTML.gif
Fig. 1

Changes in Na concentration (mM) in controls and models B and E (sessions are referred to in the text) over the length of the hemodialysis session

To limit variations in fluid and electrolytes, all study-related HD sessions were scheduled early in the morning and only mid-week. Blood flow was set at 200–300 ml/min and the dialysate flow at 500 ml/min. A Fresenius hemoflow polysulfone series filter with a hollow fiber size F4–8 was used. Subjects were given heparin based upon a standard protocol. Heparin was stopped 1 h prior to the completion of each HD session. Blood was collected from the arterial port at each data collection interval after subjects had received the heparin.

WBV was measured over different shear rates (1–1,000/s), at 37°C without additional ex vivo anticoagulation, by a capillary scanning viscometer, Rheolog (Rheologics, Exton, Pa., USA), which detects very small changes (less than 5%) in WBV [19]. Scanning capillary viscometer measurements simulate in vivo conditions, at 37°C, and preclude the need for anticoagulation of blood samples. This procedure more accurately characterizes the rheological properties of whole blood, provides WBV data over a wide range of shear rates, and provided data within minutes. The amount of fluid removed during HD was recorded, and the blood volume (BV) and hematocrit were measured with a Crit-line III (Hema Metrics, Kaysville, Utah., USA). In addition, we measured the serum Na and osmolality. WBV, BV, Na, osmolality and Hct were measured at the following five time points in the hemodialysis session; immediately before HD and at hours 1, 2, 3 and within 10 min after HD at the shear rates (SRs) 1/s, 2/s, 5/s, 10/s, 50/s, 100/s, 150/s, 300/s and 1,000/s.

Statistical analysis

Linear curve fitting and the multiple comparison test of the means were performed (Fig. 2) with the curve-fitting and statistics toolboxes of Matlab (Mathworks, Natick, Mass., USA). Multiple comparison tests were used with one-way analysis of variance (ANOVA) and post-hoc analysis using Tukey. To determine the contributing factors to the rise of WBV during HD we used the Pearson’s correlation coefficient between WBV and Hct, amount of fluid loss and BV. To study the effects of sodium modeling in models B and E we performed the one-way repeated measures ANOVA, using Sigma Stat software (SPSS Inc., Chicago, Ill, USA). Descriptive statistics was performed on serum Na and osmolality measurements. One-way ANOVA was used to compare serum Na and osmolality between the different Na models. P<0.05 was considered statistically significant.
https://static-content.springer.com/image/art%3A10.1007%2Fs00467-006-0101-y/MediaObjects/467_2006_101_Fig2_HTML.gif
Fig. 2

WBV increased linearly during HD. ac Box plots of the percent increases from baseline (time=0) at SR=50/s, 300/s, and 1,000/s, respectively. The lower and upper lines of the “box” are the 25th and 75th percentiles of the sample. The distance between the top and bottom of the box is the interquartile range. The line in the middle of the box is the median value. The whisker lines extending above and below the box show the extent of the rest of the sample. The + sign at the top of the plot is an indication of an outlier in the data. Compared to baseline (t=0), percent change in WBV increases at hours 2, 3 and 4 (multiple comparison test of the means, P<0.001). df Mean percent increase (MPI) (filled circles) at SR=50/s, 300/s, and 1,000/s vs time, respectively. The dashed lines in d–f represent the linear relationship of the MPI during the course of the study [regression used y=ax+b (r2>0.97)]. The values of a and b are 10.79, 1.45; 11.13, 0.94; 9.74, 0.684; for d, e, and f, respectively

Results

A total of six children, 10–16 years of age (four male, two female) were enrolled. In Table 1 the duration of the dialysis sessions ranged from 3.5–4 h, as prescribed by the primary nephrologist of each subject. Baseline hemoglobin (Hb) levels ranged from 12–14 g/dl. Three subjects had an average inter-dialytic weight gain (IDWG) >5% from their baseline estimated dry weight over the four sessions (two controls, B and E), and three had an average IDWG <5%. IDWG was positively correlated with pre-dialysis sitting systolic blood pressure (BP) and diastolic BP (Pearson’s correlation coefficients r=0.67 and r=0.48, respectively; P<0.05).
Table 1

Subjects’ characteristics (H Hispanic, B Black, F female, M male, MPGN membranoproliferative glomerulonephritis, DK dysplastic kidney, OU obstructive uropathy, reflux reflux nephropathy. IJ tunneled internal jugular catheter, Qb blood flow during HD, IDWG % average weight gain from baseline dry weight, Hb hemoglobin measurement at enrollment)

Characteristic

Patient no.

1

2

3

4

5

6

Age (years)

14

11

16

14

14

10

Ethnicity

H

H

B

H

H

B

Gender

F

M

M

M

F

M

Etiology

MPGN type I

Unknown

DK

OU

Reflux

OU

Time on dialysis (years)

2

3

0.3

2

O.5

1.5

Access

Fistula

Graft

Fistula

Graft

Graft

IJ

Duration (h)

3.5

3.5

4

4

4

3.5

Qb (ml/min)

200

200

300

200

200

200

Filter

F5

F5

High flux

F4

F5

F6

Percent IDWG

5.9

6.2

5.7

3.4

4.6

4.9

Hb (g/dl)

13.2

13.5

12.8

12.1

12.2

14.0

Whole-blood viscosity increased linearly during conventional dialysis

For control sessions (C), the percent increases of WBV from pre-HD baseline (PI) at hours 2 and 3 and after HD were significant for SRs 50–1,000/s (multiple comparison test of the means P<0.001, Fig. 2a–c). Furthermore, the mean percent increase (MPI) from before HD of WBV correlated linearly with length of the HD sessions at all SRs (r2>0.97; Fig. 2d–f). Figure 2d–f demonstrates that WBV MPI decreased within 10 min after completion of HD, because the measurement was obtained after the subjects had received 200 ml of 0.9% saline solution.

Higher WBV is positively correlated with increase in Hct, fluid loss and decrease in BV

To determine the contributing factors to the rise of WBV during HD, we computed the Pearson’s correlation coefficients between WBV and Hct, amount of fluid loss, and BV. In all subjects the increase in WBV was positively associated with a rise in Hct at all SRs (r=0.60–0.75; P<0.001). Increase in WBV was also positively correlated with the amount of fluid removed in all subjects at SRs of 50–1,000/s (r=0.26–0.4; P<0.05). In addition, an increase in WBV was inversely correlated with the BV in all patients at SRs of 150–1,000/s (r=0.5–0.8; P<0.05 for all).

Sodium modeling and WBV

To study the effects of sodium modeling in models B and E we compared the WBV measurements of each subject with the mean of their control measurements at each SR, using one-way RM-ANOVA. Decrease in WBV measurements seen in model E were significant for SRs of 5–1,000/s in three of the subjects who had IDWG >5% (P<0.05). In the three subjects with less than 5% IDWG there were no significant differences in WBV between controls and NaM groups B and E across all SRs. To illustrate these differences, Fig. 3 demonstrates the effects of NaM E at the shear rate of 300/s in all six subjects.
https://static-content.springer.com/image/art%3A10.1007%2Fs00467-006-0101-y/MediaObjects/467_2006_101_Fig3_HTML.gif
Fig. 3

Sodium modeling E lowers WBV at SR=300/s in patients with IDWG of more than 5%. a–c and d–f represent the means of the WBV of the controls (dark columns) and the WBV of model E (light columns) in the three patients with IDWG of more than 5% and the three with IDWG of less than 5% at the shear rate of 300/s, respectively. The x-axis refers to the following time points: before HD, 1 h, 2 h, and 3 h after the start of HD, and after HD. The decrease in WBV over time was significant in patients with IDWG of more than 5% (P<0.05) but not in those patients with IDWG of more than 5% at all shear rates above 5/s

Serum Na and osmolality were assayed in two of three patients with IDWG >5% and in all three patients with IDWG <5%. Figure 4 and Table 2 show the mean and standard deviation of the serum Na and osmolaity of the different treatment groups. The percent ultrafiltration is also shown in Table 2. No statistically significant difference in serum Na and osmolality was found between the different treatment groups (control, Na modeling B and E using one-way RM-ANOVA). Figure 5 illustrates that the serum osmolality was decreasing in all patients over the course of each dialysis session, independent of their IDWG. However, the serum Na concentration remained stable, as shown in Fig. 6.
https://static-content.springer.com/image/art%3A10.1007%2Fs00467-006-0101-y/MediaObjects/467_2006_101_Fig4_HTML.gif
Fig. 4

Serum osmolality and sodium concentration are stable and not different between controls and Na modeling B and E. The mean and standard deviation of the serum osmolality (diamonds, milliosmoles per kilogram of water) and Na concentration (circles, millimoles) are plotted in controls and Na modeling B and E. One-way ANOVA shows no statistically significant difference

Table 2

Na and osmolality. Na values are given as mean ± SD mEq/l and osmolality values are mean ± SD mosmol/kg H2O. % Uf percentage of ultrafiltrate removed during dialysis from the actual dry weight

Patient no.

 

Control

NaM B

NaM E

1

Na

 

139.6±5.5

144.0±2.1

Osmol

 

295.0±14.5

295.0±4.7

% Uf

5.4

7.4

5.7

2

Na

139.2±3.5

141.0±2.2

140.4±3.4

Osmol

305.6±7.8

298.0±13.8

302.0±8.8

% Uf

5.1

7.4

7.1

3

Na

138.8±1.6

146.4±9.1

144.4±2.8

Osmol

303.0±5.7

299.0±6.6

303.0±5.0

% Uf

5.3

5.3

6.6

4

Na

141.9±1.5

143.0±1.7

145.9±4.6

Osmol

299.0±4.6

304.4±14.2

301.0±9.7

% Uf

4.1

3.2

1.7

5

Na

144.2±1.8

143.0±0.7

142.6±2.5

Osmol

310.0±10.8

305.8±16.4

301.2±12.6

% Uf

4.7

4.8

4.6

6

Na

137.1±2.6

137.6±2.5

141.0±2.5

Osmol

301.0±10.6

293.0±13.7

300.0±11.1

% Uf

5.6

4.8

4.3

https://static-content.springer.com/image/art%3A10.1007%2Fs00467-006-0101-y/MediaObjects/467_2006_101_Fig5_HTML.gif
Fig. 5

Serum osmolality is decreasing over the course of the dialysis session in all patients. The means and standard deviations of the serum osmolality (mosmol/Kg H2O) are plotted in two groups of patients with either large or small IDWG. The x-axis refers to the following time points: before HD, 1 h, 2 h and 3 h after the start of HD, and after HD. The decrease in osmolality over time was not statistically different between the two groups

https://static-content.springer.com/image/art%3A10.1007%2Fs00467-006-0101-y/MediaObjects/467_2006_101_Fig6_HTML.gif
Fig. 6

Serum Na remained stable over the course of the dialysis sessions. The means and standard deviations of the serum Na (mM) are plotted in two groups of patients with either large or small IDWG. The x-axis refers to the following time points: before HD, 1 h, 2 h and 3 h after the start of HD, and after HD. There was no statistically significant difference between the two groups (one-way ANOVA)

Discussion

Our results describe the changes in WBV during and after HD in children over a wide range of blood shear rates and with differing dialysate Na gradients. Others have previously reported an increase in WBV after HD in adults [14, 15]. We demonstrate that WBV increased linearly over the HD sessions in children too. We observed a significant correlation between increased WBV and hematocrit, the quantity of fluid removed, and decline in blood volume.

Importantly, we observed no changes in serum Na and osmolality by our experimental procedures for Na modeling when compared with absence of Na gradient modeling (control period).

To date, intra-dialytic viscosity changes have not been evaluated in children or adults, and studies in adults have tended to look at “before” and “after” hemodialysis values. Based on our limited data, we would suggest that inclusion of such observations gives a more accurate appraisal of the WBV changes that occur.

Thus, our results are consistent with previous data from adults demonstrating higher WBV and plasma viscosity after HD, which were correlated with an increased hematocrit and the degree of ultrafiltration (weight loss) [15, 20]. However, additional factors that may contribute, in part, to our observed changes WBV during HD include changes in plasma proteins, white blood cell (WBC) count, red blood cell (RBC) aggregability and deformability [5, 7, 21, 22].

Hemodialysis access failure is a major cause of morbidity in patients on hemodialysis. Various reports indicate that a high percentage of ESKD patient hospitalization is due to vascular access complications [2325]. The United States Renal Data System (USRDS) reports that hemodialysis access failure is the most frequent cause of hospitalization among ESKD patients, and, in some centers, it accounts for the largest number of hospital days [26]. In addition, vascular changes progress rapidly in adults receiving renal replacement therapy [27]. Atherosclerosis is exacerbated in ESKD in both young and older adults [2830]. WBV has been suggested as a risk factor in vascular thrombosis in patients with deep venous thrombosis [12]. Moreover, enhanced red blood cell aggregation has been shown to affect nitric oxide synthesis and thus control vascular smooth muscle tone and thereby vascular resistance [31]. Higher WBV during HD may contribute to thrombosis of the HD access and the atherosclerotic process, thereby producing ischemic heart disease [11, 13].

Sodium modeling has been shown to decrease both intra-dialytic and inter-dialytic morbidity in adolescent and adult hemodialysis patients, with no increase in adverse events when used over a period of a few weeks [32]. Benefits of NaM include reduction in the incidence of muscle cramps, improved sodium removal, and improved vascular stability. In our study NaM appears to have the additional benefit of reduction in WBV in children with large IDWG.

By decreasing dialysate Na to counteract the consequences of rapid removal of solutes, we suggest NaM maintains a stable plasma osmolality during the course of HD ultrafiltration. A stepwise decline in Na may have an advantage over linear and logarithmic modeling [33]. It is not uncommon for children with large IDWG to have large ultrafiltration. Thus, we expect a large increase in WBV. However, model E attenuated this increase to some degree. This should have long-term consequences if studied in a large group of children. Although the mechanism of the effect remains unknown and is not related to changes in serum Na or osmolality, we are continuing our study to look at other variables that affect WBV.

Our study has some limitations. These include a small sample size, which limits our ability to generalize the results. We also were not able to measure all variables that may affect WBV because of limited ability to withdraw larger volumes of blood in these children. Lastly, we could not predict IDWG but used the random nature of its magnitude to our advantage in the analyses presented.

In this pilot study, NaM attenuated the increase in WBV in subjects with large IDWG (>5%) but not in subjects with smaller IDWG (<5%). Such attenuation may benefit the children in the long term by decreasing their risk of access thrombosis and atherosclerotic events.

Acknowledgments

This study was supported in part by a grant from Rheologics, Inc., Exton, Pa., USA, and from the Zell Family and McNulty Family Foundations. We thank the research participants, their families, and the nursing staff of DaVita Children’s Dialysis Center for their kind assistance in this study.

Copyright information

© IPNA 2006