Volume and sodium overload in dialysis patients are related to hypertension, edema, and left ventricular dilatation.1 Preliminary data suggest that, apart from volume, sodium might also have an independent effect on blood pressure (BP) regulation in dialysis patients.2,3
During dialysis, the main determinant of sodium removal is the ultrafiltration volume. However, the contribution of diffusive mass transfer of sodium may also be significant.4 As such, the prescription of the dialysate sodium concentration has an important effect on sodium balance during dialysis. With the advent of dialysate conductivity monitoring, it has become possible to approximate sodium fluxes during dialysis by means of the assessment of intradialytic ionic mass balance (IMB), as well as plasma conductivity (PC) as a surrogate of plasma sodium.4–8 These developments offer the opportunity to study sodium balance in more detail. The main advantage of PC is that it can be used on a treatment to treatment basis without the need for blood sampling.
Although previous studies have focused on the effect of modifications of dialysis treatment, such as sodium profiling,4,5,9,10 in relatively small groups of patients, little is known about sodium balance on a treatment to treatment basis during dialysis with standard “physiologic” sodium concentrations (e.g., 140 mmol/L) at facility level. One study showed an increase in plasma sodium during dialysis in a significant subset of patients with the use of a dialysate sodium concentration of 140 mmol/L.11 Predialytic plasma sodium levels are generally considered to be stable in dialysis patients, assuming a fixed sodium set point in dialysis patients,12,13 which may be of importance when dialysate sodium prescriptions are individualized. However, long-term studies on sodium balance and variations in plasma sodium are scarce.
The aim of this study was to assess variations in PC as a surrogate marker of plasma sodium on a treatment to treatment basis at facility level. Moreover, the relation of inter- and intraindividual changes in PC with IMB, BP, relative blood volume (RBV), and interdialytic weight gain was assessed.
In the dialysis center of Maxima Medical Centre Veldhoven, The Netherlands, treatment data on PC, IMB, and changes in RBV are automatically stored in a database on a treatment to treatment basis. Data were retrieved from this database for a consecutive period of 6 months for each patient and were randomly collected in the period between August 2006 and November 2007. Moreover, in this database, pre- and postdialytic BP and body weight are recorded. The study was carried out according to the Declaration of Helsinki and approved by the hospital institutional review board.
Data from 73 chronic dialysis patients were recorded. Patient characteristics are displayed in Table 1. All patients were on thrice-weekly hemodialysis, except for two patients, who were treated with twice-weekly dialysis sessions. All patients were treated with Integra (Hospal-Gambro, Mirandola, Italy) device, by which PC, IMB, and changes in RBV are automatically assessed and recorded. Low-flux polysulfone dialysis membranes (F8HPS; Fresenius, Bad Homburg, Germany) were used. At the time of the study, prescription of the dialysate was sodium 140 mmol/L (conductivity 14.0 mS/cm), potassium 2.0 mmol/L, calcium 1.5 mmol/L, magnesium 0.5 mmol/L, bicarbonate 32 mmol/L, acetate 3.0 mmol/L, and glucose 1 g/L. Temperature of the dialysate was 36°C. Patients were ultrafiltrated until their clinically determined dry weight. Fifteen patients were treated with the Hemocontrol (Hospal-Gambro) biofeedback module, for which an equivalent dialysate conductivity of 14.0 mS/cm was also prescribed.
Ionic Mass Balance
Intradialytic IMB was estimated by Diascan (Hospal-Gambro) integrated in the Integra dialysis module.8,14 In short, Diascan measures IMB by constant measurement of the conductivity in the dialysis outlet (Cdout), and inlet (Cdin) IMB is reported directly by the dialysis module. By convention and also as such reported by the Integra device, a positive IMB reflects net ionic removal from the patient, and a negative IMB means net ionic transfer from dialysate to the patient.6 Diffusive IMB was assessed by the following formula:
Validation of IMB Measurements
In 22 patients, direct dialysis quantification by Quantiscan was performed during a single dialysis session. Quantiscan continuously samples a small amount of the spent dialysis fluid. Electrolytes were assessed in duplicate in the spent dialysis fluid: sodium (flame photometry and indirect ionometry), potassium, calcium, phosphate, bicarbonate, and chloride. Sodium balance was assessed as follows:
Cout and Cin are the sodium concentration in spent and fresh dialysate, respectively. Vout and Vin are the volume of spent and incoming dialysate. Vout is calculated as (dialysis flow rate + ultrafiltration rate) × dialysis time and Vin as dialysis flow rate (ml/min) × dialysis time (min). The same procedure was followed for the other ions.
PC was measured and reported directly by Diascan by measuring dialysance (D) in combination with measurements of Cdout and Cdin according to the following formula:
D is assessed every 30 minutes by measuring the increase in Cdout after a temporary increase in Cdin by 1 mS/cm according to the formula:
Equation (Uncited)Image Tools
Equations 1 and 2 indicate, respectively, the measurements before and after the temporary increase in Cdin.4,14 The relation between PC and plasma sodium given in the literature is as follows: plasma sodium = (PC × 10.4) − 9.57.6,7
Plasma sodium levels were assessed before dialysis, approximately once monthly, by indirect ionometry (Vitros 950). Plasma sodium levels were available in 332 treatments.
RBV, BP, and Residual Glomerular Filtration Rate
Changes in RBV were measured continuously by continuous optical assessment of changes in hemoglobin during the dialysis session (Hemoscan, Hospal-Gambro). Blood pressure was taken as part of the clinical routine in sitting position just before the start of dialysis treatment. Body weight was measured, also as part of the clinical routine, on a gravimetric scale. Residual Glomerular filtration rate was assessed by interdialytic urine collections as the mean of urea and creatinine clearance.
Intra- and interindividual variations in PC are reported by means of the range, SD, and coefficient of variation. The relation between predialytic PC, IMB, RBV, interdialytic weight gain, and BP was assessed by correlation analysis (Pearson's r). Measurements during the 6-month period were also pooled for each patient, and the mean of these data was used for analysis. Multiregression analysis was used where necessary. Regarding intraindividual variations, predialytic BP and interdialytic weight gain were compared between the session, with the highest and lowest value of PC and analyzed using a paired Student's t test. Correction for ultrafiltration volume and timing of the dialysis shift (after longest or short dialysis interval) were performed using univariate analysis, with lowest or highest plasma conductivity as fixed factor and timing of dialysis shift and ultrafiltration volume as covariates.
In total, 4,070 measurements were retrieved in 73 patients. The mean number of treatments analyzed was 56 (range: 45–63 treatments). Missing data were, among others, the result of technical problems with the data management tool. The mean of the pooled data collected during the 6-month period for the patient is summarized in Table 2, separately presented for patients on Hemocontrol and standard dialysis treatment. Diffusive mass balance was 18.3% ± 16.7% of convective mass balance. Although total IMB was positive (indicating net ionic transfer from the patient to the dialysate) in all patients using the pooled data, the averaged diffusive IMB was negative (indicating diffusive ionic transfer from dialysate to patient) in 33% of patients. In 14% of patients, the averaged PC increased during dialysis. As shown in Table 3, the only parameter that distinguished between patients with respective negative or positive diffusive IMB, or an increase or decrease in PC during dialysis, was the predialytic PC.
For intraindividual variations in predialytic PC, the coefficient of variation during the 6-month follow-up period was 1.3%. For postdialytic PC, the coefficient of variation was 1.0%. The mean (14.31 ± 0.19 vs. 14.21 ± 0.23) and coefficient of variation (1.2% vs. 1.3%) of predialytic PC was not significantly different between patients with (n = 15) or without diabetes mellitus (n = 58). Within patients, predialytic PC was 13.78 ± 0.29 before the treatment with the lowest predialytic PC and 14.65 ± 0.25 before the treatment in which the highest predialytic PC was recorded. The intra- and interpatient variations in predialytic PC can also be observed in Figure 1, A and B. The variation in the CV of predialytic PC is shown in Figure 1, C.
Although the observation period was not designed to assess seasonal variations in detail, we observed a significant difference between the various seasons, with the lowest values in autumn (14.21 ± 0.28 mS/cm), followed by summer (14.22 ± 0.28 mS/cm), winter (14.25 ± 0.30), and spring (14.27 ± 0.28 mS/cm) (p < 0.001). Regarding the timing of the dialysis shift, we also compared predialytic PC between treatments after the longest shift and other treatments. No significant difference in PC was observed: 14.24 ± 0.29 before the longest shifts and 14.25 ± 0.29 before the other shifts.
The variation in postdialytic PC was less when compared with predialytic PC (Figure 1, C). Mean SD for postdialytic PC was 0.09 mS/cm. Postdialytic PC was 13.85 ± 0.19 after the treatment with the lowest predialytic PC and 14.30 ± 0.20 after the treatment in which the highest predialytic PC was recorded.
Relation Between PC, IMB, and Hemodynamic Parameters
In the 4,070 measurements, predialytic PC was significantly related to diffusive IMB (r = 0.82; p < 0.001) (Figure 2). The same held true for diabetic patients (n = 15), who were analyzed in a subgroup (r = 0.86; p < 0.001). Total and diffusive IMB, as well as interdialytic weight gain, were not significantly different between patients with or without diabetes mellitus (data not shown).
The mean averaged PC during the 6-month period in the 73 patients was significantly related to total (r = 0.37; p < 0.01) and diffusive IMB (r = 0.91; p < 0.001), as well as to the change in PC during dialysis (r = −0.85; p < 0.001). Total IMB was strongly related to ultrafiltration volume (r = 0.88). Predialytic PC was significantly related to predialytic systolic BP (r = 0.35; p < 0.01) (Figure 3, A) but not to diastolic BP (r = 0.1; p = NS) or interdialytic weight gain (in the period before the highest and lowest PC) (r = −0.09; p = NS). Using multiregression analysis, the relation between predialytic systolic BP and predialytic PC (β = 0.37; p = 0.002) was independent of interdialytic weight gain and the number of antihypertensive agents used (β = 0.14; p = NS). Although the relation between predialytic PC and diffusive IMB was comparable (r = 0.84; p < 0.001) in the subgroup of patients on Hemocontrol, diffusive IMB was positive in all patients, whereas no increase in PC was observed.
Intrapatient Variations in Predialytic PC
When analyzing the effect of intrapatient variations of PC, significant differences in IMB, predialytic systolic BP (Figure 3, B), and diastolic BP were observed between the treatments with the lowest and highest predialytic PC, whereas interdialytic weight gain (in the period before the highest and lowest PC) was not significantly different (Table 4). There was no significant relation between the absolute difference in BP and the corresponding difference in PC between both treatments.
There were no significant differences in interdiaytic weight gain (2.6 ± 1.0 vs. 2.5 ± 1.0 L), systolic BP (148.6 ± 25.1 vs. 145.5 ± 24.3 mm Hg), or diastolic BP (80.5 ± 13.5 vs. 79.6 ± 14.4 mm Hg) in the period after the highest and lowest predialytic PC.
When using univariate analysis with correction for covariates, the difference between the treatments with lowest or highest predialytic PC value and predialytic systolic BP remained significant (F = 5.90; p < 0.05) independent of the season (F = 8.38; p < 0.01), timing of the dialysis shift (after long or short interval) (F = 0.27; p = NS), and interdialytic weight gain (in the period before the lowest or highest PC) (F = 3.03; p = 0.08). The same held true for the difference in predialytic diastolic BP between the treatments with the highest and lowest predialytic PC (F = 6.91; p = 0.01) when corrected for interdialytic weight gain (F = 4.13; p < 0.05) or timing of dialysis shift (F = 0.86; p = NS) and season (F = 2.91; p = 0.09).
Plasma Sodium: Variation and Relation With PC and IMB
Mean prediaytic plasma sodium was 140.0 ± 3.4 mmol/L. Mean intraindividual coefficient of variation was 1.4%. The mean difference between the lowest and highest plasma sodium concentration was 4.6 mmol/L.
In this study, the relation between predialytic plasma sodium and PC was highly significant (r = 0.77; p < 0.001) (Figure 4, A). The regression equation between plasma sodium and PC was as follows: plasma sodium = 26.98 + 7.94 × PC (in which a PC of 14.0 mS/cm would correspond to a plasma sodium concentration of 138.1 mmol/L).
Using the data of the 332 treatments for which plasma sodium was available, predialytic plasma sodium was significantly related to diffusive IMB (r = 0.57; p < 0.001) and total IMB (r = 0.26; p < 0.001) (Figure 4, B).
Validation Study: Relation Between IMB and Sodium Balance
As displayed in Figure 5, A, a highly significant relation was observed between IMB and sodium balance (using flame photometry) based on direct dialysis quantification (r = 0.97). The disagreement in absolute values appeared to increase at higher levels of IMB, as shown in Figure 5, B. However, the F value for the regression equation between sodium balance and IMB was very high (F = 307.6), with the following equation: sodium balance (mmol) = 57 + 1.29 × IMB. Using indirect ionometry to determine sodium, the correlation became somewhat less strong (r = 0.91), but the mean difference between both methods was also somewhat less as compared with the flame photometry measurements (76 ± 202 mmol). Although in multiregression analysis the relation between IMB and sodium balance was highly significant, the balance of the other ions, such as potassium, chloride, bicarbonate, and calcium, was not significantly related to IMB.
This study shows that both intra- and interindividual variations in PC, which were used as a surrogate for plasma sodium, have a significant effect on IMB (as a surrogate marker for sodium balance) during dialysis. With a dialysate sodium concentration of 140 mmol/L, diffusive ionic mass transfer from dialysate to patient was observed in a significant percentage of patients, whereas PC increased during dialysis in a lesser percentage of patients. Both inter- and intraindividual variations in predialytic PC were related to predialytic BP, independent of interdialytic weight gain, and thus appear to be of physiologic relevance.
In agreement with earlier studies in dialysis patients, we observed a significant interindividual variation in predialytic PC, as a surrogate for plasma sodium. Moreover, also significant intraindividual variations in PC were observed. In 10 nondiabetic dialysis patients who were followed for 12 months, Flanigan11 observed a variation in plasma sodium concentrations of ∼2%. This would appear in agreement with the mean coefficient of variation of 1.3% of PC in this study, which is in agreement with data from the normal population in the literature and likely reflects for the largest part “real” intraindividual and not method variation.15 However, the mean intraindividual difference in PC measurements between the lowest and highest PC measurements during the 6-month follow-up period was 0.9 mS/cm, corresponding to a plasma sodium concentration of ±9 mmol/L. This observation shows that during a follow-up period of 6 months, relatively large intrapatient variations in PC may occur. Although some of these values may be outliers, they still might have pathophysiologic relevance due to the relation with BP and IMB. We were not able to provide a definite explanation for the variations in PC. He et al.16 showed an effect of dietary sodium intake on plasma sodium levels. We also noted some seasonal variation in PC, although the study was not specifically designed for this purpose. The seasonal variation in PC is in agreement with the data of Chen and coworkers17 in peritoneal dialysis patients.
The variation in PC is in agreement with earlier observations on plasma sodium. In the study of Flanigan,11 the variation in predialytic plasma sodium concentrations was often in the order of 5 mmol/L during the 12-month follow-up period. In this study, the mean range in plasma sodium levels, which were however only available on a monthly basis, was 4.6 mmol/L.
Various authors suggested the presence of an individual sodium set point in dialysis patients, which may be used as a tool for individualized sodium prescription.6,13,18 Data on the presence of a fixed sodium set point in individual dialysis patients are somewhat conflicting. Although some authors did not observe changes in predialytic plasma sodium concentration after changing dialysate sodium prescription,18–20 we and others observed a change in predialytic PC or plasma sodium after a change in dialysate sodium prescription.5,9,21 The reason for this discrepancy is not clear, but it might also be hypothesized that patients who experience an increase in plasma sodium level after an increase in dialysate sodium managed to comply with the habitual fluid restriction at the cost of increased thirst. The results of this study suggest that plasma sodium levels, when used as a guide for individualization of dialysate sodium prescription, should be examined on a regular basis. However, it cannot be excluded that variations in other ions, such as potassium and bicarbonate, also play a role in the observed variations in PC, for example, potassium in a concentration of 2 mmol/L contributes only for 0.5% to dialysate conductivity.8
Predialytic PC was a strong determinant of diffusive IMB, and in a lesser degree to total IMB, during dialysis. Using the mean of 6 months, diffusive IMB was negative in nearly 33% of dialysis patients, suggesting diffusive ionic influx from dialysate to the patient. Because of the strong correlation between IMB and sodium balance,6 this might suggest net diffusive sodium influx from dialysate to the patient. However, these results should be interpreted with some caution, given the fact that sodium removal may be somewhat underestimated by IMB.6 Surprisingly, we did not observe a mean diffusive ionic influx in the patients treated with Hemocontrol, which is likely due to the fact that mean PC was higher in these patients. For this observation, we do not have a good explanation. In an earlier randomized crossover study, we observed no difference in IMB of PC when patients were treated with either Hemocontrol or standard dialysis.5
In this study, we compared IMB measurements with sodium balance using direct dialysis quantification. A highly significant relation was observed between IMB and direct dialysis quantification, with sodium measurements assessed by flame photometry. However, also in our study, some underestimation of sodium removal by IMB was observed. Mass balances of other ions, such as potassium or bicarbonate, were not significantly related to IMB. Access recirculation, which theoretically could also affect IMB measurements, was not assessed in the study. However, given the highly significant regression model between IMB and sodium balance, we believe that IMB might be used as an indicative marker for sodium balance, taking the caveats into account. In an earlier study, large differences in IMB were observed between treatments, which only differed in dialysate sodium prescription.4
Averaged over the 6-month period, PC increased during dialysis in 14% of patients, which is an additional argument for diffusive sodium influx in a significant minority of dialysis patients. In general, diffusive ionic influx seemed to occur when predialytic PC was <14.25 mS/cm (roughly corresponding to a plasma sodium concentration of 140 mmol/L).7,22 However, given the slight underestimation of sodium removal by IMB and the possible effect of other ions on IMB, the extrapolation of these findings should be interpreted with some caution. Interestingly, the number of patients in which PC increased during dialysis was less compared with the number of patients with diffusive ionic influx. Although the reason for this observation is not entirely clear, this might be due to the fact that decrease in PC may to some degree also be because of lowered plasma levels of other ions.
Although far less treatments in which predialytic plasma sodium levels were observed were available for analysis, diffusive ionic influx (negative IMB) from dialysate to patient seemed to occur when plasma sodium concentration was <140 mmol/L, which would appear in basic agreement with the relation between PC and diffusive IMB discussed above.
Because of the strong influence of ultrafiltration on ionic mass removal, net ionic mass removal was observed in nearly all patients. Admittedly, the clinical importance of the diffusive ionic mass transfer cannot be elucidated from this study. Lowering of dialysate sodium from 141 to 138 mmol/L at a facility level resulted in a small improvement in BP control without an increase in hypotensive episodes.23 Individualization of dialysate sodium concentration in patients with lower predialytic plasma sodium concentrations was shown to have additive value in terms of BP control.18
Apart from the effects on diffusive IMB, intraindividual differences in PC might be of physiologic relevance, given the significant relation with predialytic systolic BP. This relation was independent from interdialytic weight gain, suggesting an independent pressor effect of plasma sodium, although these observations should be confirmed in future studies, as a causal relationship cannot be deduced from the present observational data. Proposed pathophysiological mechanisms for this relation remain hypothetical and include, among others, an effect of sodium on sympathetic activity, nitric oxide metabolism, or digoxin-like factors.3,23 The effect of seasonal variations, which also might influence BP levels, did not explain the relation between PC and BP in our study. As plasma sodium may be decreased in patients with heart failure, one could argue, however, that also differences in cardiac function, leading to low predialytic systolic BP, could provide an explanation for these findings. We are not able to provide echocardiographic examinations for all patients during the study period. However, differences in cardiac function are unlikely to explain the effect of intraindividual differences in PC on BP: both predialytic systolic BP and diastolic BP were significantly different between treatments with the highest and lowest predialytic PC.
An important advantage of this study is that it comprises a large group of unselected patients, in which measurements of PC, IMB, and RBV were available for every treatment with a 6-month duration. A major limitation is the observational nature of the study. Another drawback of the study is the fact that PC, and not plasma sodium, measurements were used. The use of PC as a surrogate for plasma sodium seems justified in view of the strong relation between PC and plasma sodium,7 and due to the fact that measurements can be obtained automatically from the software of the dialysis module without the need for blood sampling during every dialysis treatment. From a theoretical point of view, both plasma sodium and PC are surrogate markers of plasma osmolality, which is the driving force for thirst. In this study, plasma sodium and PC were also highly significantly related. However, in view of the possible (minor) contributions of other ions to PC, our findings should be interpreted with caution.
Another drawback of the study is the absence of detailed information on volume status of the patient, alimentary intake, and the absence of ambulatory BP measurements.
In conclusion, diffusive ionic mass transfer is strongly dependent on predialytic PC. A dialysate sodium concentration of 140 mmol/L was associated with diffusive mass transfer from dialysate to patient and an increase in PC in a significant percentage of patients. Both intra- and interpatient variations in PC are significantly related to predialytic systolic BP, independent of interdialytic weight gain, which might suggest a volume-independent effect of sodium. The results of this study may provide additional support for the use of dialysate sodium levels at the lower physiologic range and, where necessary, individualization of dialysate sodium levels.
Supported by a research grant of Gambro Company.