Fluid Dynamics During Hemodialysis in Relationship to Sodium Gradient Between Dialysate and Plasma

Abstract

Fluid shifts during hemodialysis involve changes in both extracellular and intracellular volumes. This study aimed to determine the effect of intradialytic sodium gradients (GNa⁺), that is, the difference between dialysate and serum sodium concentration, on dynamics of extracellular and intracellular volumes in a group of maintenance hemodialysis patients. Extracellular volume change (deltaECV) between predialysis and postdialysis periods was determined by whole-body bioimpedance spectroscopy; intracellular volume change (deltaICV) was indirectly derived as the difference between deltaECV and the change in body weight, corrected for intradialytically given fluids. A total of 200 bioimpedance measurements were performed in 32 dialysis patients. Extracellular and intracellular volume changes were −2.6 ± 0.9 L (range: −4.7 to −0.5 L) and −0.2 ± 0.7 L (range: −2.5 to +1.5 L), respectively. There was a significant correlation between deltaICV and GNa⁺; deltaICV = −0.12 * GNa⁺ + 0.26 (p < 0.001). In contrast, GNa⁺ was not correlated with deltaECV. We conclude that the sodium gradient between dialysate and plasma has a significant effect on the ICV during dialysis. Hemodialysis with GNa⁺ = 0 mmol/L should be sought to prevent ICV shrinking or swelling and to prevent excessive thirst, consequently high interdialytic weight gains, and ultrafiltration rates.

Fluid fluxes in patients during hemodialysis involve both the intracellular volume (ICV) and extracellular (ECV) volume, but the relative contribution of the two compartments to ultrafiltration volume (UFV) is poorly defined. Movement of water across cell membranes and the accompanying changes in cell volume are driven by osmotic gradients between ECV and ICV. The concentration of sodium in the ECV is a major determinant of effective ECV osmolarity. When effective ECV osmolarity exceeds that of the ICV, fluid exits from the ICV to the ECV compartment. During hemodialysis (HD), plasma sodium concentration (PNa⁺) may change as a function of the sodium gradient between dialysate and plasma. Whenever the dialysate sodium concentration (DNa⁺) exceeds PNa⁺, the latter will rise and vice versa. Frequently, dialysate sodium levels >140 mmol/L are used, which may exceed the PNa⁺ in the majority of maintenance HD (MHD) patients, resulting in a positive intradialytic sodium gradient (GNa⁺, defined as DNa⁺ minus PNa⁺).¹ The current study aimed to test the hypothesis that intradialytic GNa⁺ affects the fluid shift between ICV and ECV during dialysis.

Materials and Methods

We studied 32 MHD patients by whole-body wrist-to-ankle bioimpedance spectroscopy (WBIS) before and after two successive HD sessions. Pre-HD and post-HD body weights were recorded, as was the fluid administered during the dialysis treatment. All patients were dialyzed with Fresenius polysulphone membrane dialyzers (F80/F200; Optiflux 180/200). DNa⁺ was 142 mmol/L during all treatments. Predialysis PNa⁺ was measured by ion-specific electrode (Olympus Diagnostics, Center Valley, PA). Informed consent was given by all the patients, and the protocol was approved by the Institutional Review Board, Beth Israel Medical Center, New York, NY.

Estimation of Changes of ECV (deltaECV) and ICV (deltaICV)

Extracellular volume was measured by WBIS performed immediately before (pre-ECV) and 5 minutes after (post-ECV) the hemodialysis session. Measurements were taken on the nonaccess side of the body. WBIS was done in a standard fashion, with the patient lying supine on a flat nonconductive bed, using the Xitron 4200 device (Xitron Technologies Inc., San Diego, CA). ECV was calculated by using resistivity values provided by the manufacturer; for men, ρECV = 40.5 Ω cm; for women, ρECV = 39 Ω cm (Hydra ECF/ICF model 4200 Xitron Technologies Inc., San Diego, CA, 2001). The methodological details have been reported elsewhere.² The change of ECV during dialysis (deltaECV) was calculated as,

Based on the assumption that UFV equals the sum of changes of ECV and ICV (post minus pre),

The intradialytic change of ICV (deltaICV) was calculated as,

As the dialysis units routinely do not record UFV, it was estimated by adding the fluid administered (AdFl) during dialysis session (during the rinsing process) to the change in body weight (deltaWt).

Statistical Methods

Means, standard deviations (SD), 95% confidence intervals, and ranges of data are given. Linear regression analysis was used to assess the relationships between continuous variables. Standard errors (SE) of regression coefficient and intercept are reported. Correlation coefficients were estimated by the use of Pearson’s product moment method. Values of P < 0.05 were considered significant. Statistical analysis was performed with SPSS version 11.5 (SPSS Inc., Chicago, IL).

Results

A total of 200 dialysis sessions representing 100 pairs of successive HD sessions in 32 maintenance hemodialysis patients (22 male and 10 female) were analyzed. Distribution of the repeat paired measurements were as follows; 1, 14 patients; 2, 4 patients; 3, 4 patients; 4, 2 patients; 5, 3 patients; 6, 1 patient; 7, 1 patient; 9, 1 patient; 10, 1 patient; and 11, 1 patient. Baseline characteristics of the study population are shown in Table 1. Nine (28.1%) patients had type 2 diabetes, 20 had no diabetes, and diabetes information was not available in three patients. Patients had a body mass index (BMI) range of 14.5 to 55.5 and a pre-HD body weight range of 41.8 to 184.2 kg. Pre-HD weight was more than the dry weight, and all patients were treated to approach their dry weight (Table 1). Interdialytic weight gain expressed as percentage dry weight (IDWG %DW) was 3.6 ± 1.6%, which would rule out an overtly overhydrated state in our patients³ (data not shown).

Table 1: Baseline Patient Characteristics

The actual weight loss was less than the ultrafiltration volume not only because of administered fluid but also because many patients ate during dialysis (Table 2). The sodium gradient between the dialysate and the blood compartment (GNa⁺) varied from −1.0 to +11.0 mmol/L. There was no correlation between GNa⁺ and BMI, protein catabolic rate, creatinine, albumin, and hemoglobin levels. No intradialytic symptoms were collected in the present study; however, requirement of 0.2 to 0.7 L of intradialytic fluid administration (shown in Table 2) points against any significant hypotensive episode.

Table 2: Intradialytic Fluid Dynamics

Ultrafiltration volume was −3.1 ± 1.1 L (range: −6.2 to −0.4 L); intradialytic ECV change was −2.6 ± 0.9 L (range: −4.7 to −0.5 L), and ICV change was −0.2 ± 0.7 L (range: −2.5 to +1.5 L). The mean contributions of the ECV and the ICV to UFV were 94% (95% confidence interval: 90% to 98%) and 6% (95% confidence interval: 2% to 10%), respectively. Median GNa⁺ was 4 mmol/L; mean deltaICV was 0.05 L in those patients with GNa⁺ ≥4.0 mmol/L and −0.3 L with GNa⁺ below the median (difference, 0.35 L [95% confidence interval: −0.63 to −0.06 L]; p = 0.015). There was a significant correlation between deltaICV and GNa⁺: deltaICV = −0.12 * GNa + 0.26; the SE of the regression coefficient was 0.018 (p < 0.001), and the SE of the intercept was 0.087 (p = 0.003). With an r² of 0.18, 18% of deltaICV variability was explained by GNa variability.

As shown in Figure 1, about 50% of delta ICV points between 1 and 6 mmol/L GNa are uniformly distributed above and below deltaICV = 0, suggesting that ICV shifts largely independent of GNa level; in the presence of GNa above 6 mmol/L, virtually all deltaICV data points are below 0, indicating an ICV loss predominantly driven by the positive GNa. The regression line did not intercept at zero, and in some patients, particularly in those with a GNa⁺ below 2 mmol/L, a gain of ICV despite ongoing ultrafiltration was observed. The regression lines delta ICV versus GNa between patients with BMI below and above 30 kg/m² did not differ (data not shown). There was no relationship between deltaECV and GNa⁺ (Figure 2).

Figure 1.: Relationship between intradialytic change of intracellular volume and dialysate-to-plasma sodium concentration gradient (GNa⁺) (Pearson’s linear regression analysis). Depending on the sodium gradient, there is a gain or loss of intracellular volume.

Figure 2.: Relationship between intradialytic change of extracellular volume and dialysate-to-plasma sodium concentration gradient (GNa⁺) (Pearson’s linear regression analysis).

Discussion

The main finding of the study is the relationship between intradialytic sodium gradient (GNa⁺), that is, the difference between dialysate and predialysis plasma sodium concentration and ECV and ICV fluid dynamics during hemodialysis. A positive GNa⁺ (DNa⁺ > PNa⁺) resulted in a contraction of the intracellular compartment (cell shrinking) during dialysis, whereas a negative GNa⁺ (DNa⁺ < PNa⁺) resulted in a gain of ICV (cell swelling). Based on known factors regulating cell volume, it is likely that a rise in PNa⁺ is the major driving force responsible for fluid shifts between ICV and ECV during hemodialysis. In our study, the intradialytic loss of ICV was as large as 2.5 L, and the maximal ICV gain was 1.8 L. A positive GNa⁺ results in a rise of postdialysis PNa⁺, which subsequently stimulates thirst. Consecutively, both ICV and ECV are replenished by ingestion of free water in the interdialytic period. The volume shifts between ECV and ICV are accompanied by changes in cell volumes. According to the regression equation relating GNa⁺ to deltaICV (Figure 1), intradialytic shrinkage of cell volume occurs in the presence of a GNa⁺ > 2 mmol/L. The 82% of ICV shift, which is not determined by GNa, could be explained by changes of extracellular glucose concentration, food intake, rise of extracellular albumin concentration during ultrafiltration, modification of the Na⁺/K⁺ATPase activity induced by (a) changes of extracellular K⁺ concentration and (b) increased sympathetic activity during HD.

The pathophysiological consequences of repeated cycles of cell volume changes in the course of intradialytic and interdialytic periods are unknown. An array of metabolic cellular processes and gene expression patterns are linked to cell volume.⁴ It is tempting to speculate that rapid changes in cell volume during hemodialysis contribute to the morbidity of MHD patients. Cell volume regulation is key to immune function, hormone and transmitter release, excitability and contraction, cell migration, cell proliferation, and cell death. For example, cardiac swelling–activated chloride channels are involved in a variety of cell functions, such as regulation of membrane potential in excitable cells, cell proliferation and differentiation, cell volume regulation, and cell death through apoptosis. Cardiac swelling–activated chloride channels have been implicated in arrhythmogenesis, myocardial injury, preconditioning, and apoptosis of myocytes.^5,6

Extracellular volume and ICV were estimated by means of WBIS. WBIS is able to measure changes in ECV more accurately than the absolute volume. Although there is a poor correlation of the traditional bromide dilution measurements of ECV with that measured by WBIS, changes of ECV can be estimated for clinical use,^7,8 whereas technical limitations make estimation of intracellular volume by WBIS more difficult. In addition, data from our laboratory has shown that electrolyte changes after HD do not significantly influence the accuracy of ECV estimation by WBIS.⁹

Our findings are qualitatively in line with the only other study published on the relationship between dialysate sodium concentration (expressed as dialysate conductance) and changes of ECV and ICV, respectively.¹⁰ In this study, the contribution of ICV loss to the UFV was as high as 54%, whereas in our study ICV loss was 6% of UFV. Noteworthy, in the study of Jaffrin et al.,¹⁰ the patients were dialyzed with dialysate conductivity between 14.0 and 14.7 ms (which corresponds to a DNa⁺ between 143 and 150 mmol/L). Although the authors did not provide data explicitly on GNa⁺, a mean GNa⁺ of 6.6 mmol/L with a range 0 to 14 mmol/L can be approximated from the data on dialysate conductivity and plasma sodium reported in their paper. In our study, the mean GNa⁺ was 4.0 mmol/L (range: −1.0 to 11 mmol/L), and it can be expected that this greater GNa⁺ results in an even greater ICV loss. The authors report a significant positive correlation between conductivity difference between dialysate and plasma and the percent of ICV contribution to UFV (R² = 0.76); this finding supports the results obtained in the present study. In a previous study from our group,¹¹ in 10 patients the ECV contribution to UFV was 95%, with segmental bioimpedance and 70% with whole body bioimpedance. The wide range of BMI of our patients could have led to a bias in ECV estimation and hence in ICV calculation. However, similar to deltaICV, a comparison of regression lines delta ECV versus GNa between patients with BMI below and above 30 kg/m² showed no difference (data not shown).

In theory, in the absence of a GNa⁺, deltaICV should be zero. In our data set, the regression line of GNa⁺ versus deltaICV intercepts the x-axis (deltaICV = 0) at a GNa⁺ of ∼ +2 mmol/L. This could be due to the fact that not all patients in our study refrained from eating and drinking during dialysis. Depending on the amount of salt and water ingested, effects on ECV and ICV fluid dynamics may vary. We assume that in most cases, the food intake was hypotonic, with an excess of free water. Any intake of free water would ameliorate the effects of a positive GNa⁺ on ICV loss. This discrepancy also could be caused by the fact that whole-body BIS cannot supply reliable estimates of ICV due to recently described phenomenon of tissue anisotropy.^12–14 ECV estimates in whole-body BIS are based on the assumption that low-frequency current flows only in the extracellular fluid.^2,8 Recent studies have provided evidence that current at any frequency also flows through cells (electrical anisotropy of human tissues), particularly muscle cells, therefore making inaccurate ECV estimates.^11–14 Although it could be speculated that the ECV estimation by whole-body BIS is inaccurate, based on tissue anisotropy, it could be argued that such an effect would affect both pre-HD and post-HD measurements, making its effects on deltaECV insignificant.

In conclusion, depending on the sodium gradient between dialysate and the ECV, intradialytic swelling (with DNa⁺ < PNa⁺) or shrinking (with DNa⁺ > PNa⁺) of the ICV takes place. Cell volume shrinking and expansion during dialysis and in the interdialytic period may contribute to the morbidity in chronic dialysis patients; hemodialysis with zero sodium gradients should be sought for this and to prevent excessive thirst and consequently high interdialytic weight gain and ultrafiltration rates. Refinement of BIS technique in the future can improve the determination of ICV shifts, which would allow modulation of sodium gradient toward that end.