Dialysate Sodium Lowering in Maintenance Hemodialysis

Visual Abstract

Abstract

Key Points

• Treatment to dialysate sodium 135 versus 138 mEq/L led to no difference in the rate of change in intradialytic hypotension, but symptoms were greater in the low arm.

• Use of a dialysate sodium concentration of 135 versus 138 mEq/L led to a small reduction in interdialytic weight gain, but had no effect on predialysis BP.

• Raising dialysate sodium concentration from 135 to 140 mEq/L reduced intradialytic hypotension and was associated with a marked increase in BP.

Background

Lowering dialysate sodium concentration may improve volume and BP control in patients on maintenance hemodialysis.

Methods

We randomized 42 participants 2:1 to dialysate sodium 135 versus 138 mEq/L for 6 months. This was followed by a 12-week extension phase in which sodium was increased to 140 mEq/L in low-arm participants. The primary outcome was intradialytic hypotension (IDH). Secondary outcomes included dialysis disequilibrium symptoms, emergency room visits/hospitalizations, interdialytic weight gain, and BP. Longitudinal changes across arms were analyzed using linear mixed regression.

Results

Treatment to dialysate sodium 135 versus 138 mEq/L was not associated with a difference in a change in the rate of IDH (mean change [95% confidence interval], 2.8 [0.8 to 9.5] versus 2.7 [1.1 to 6.2] events per 100 treatments per month); ratio of slopes 0.96 (0.26 to 3.61) or emergency room visits/hospitalizations (7.3 [2.3 to 12.4] versus 6.7 [2.9 to 10.6] events per 100 patient-months); difference 0.6 (−6.9 to 5.8). Symptom score was unchanged in the 135 mEq/L arm (0.7 [−1.4 to 2.7]) and decreased in the 138 mEq/L arm ([5.0 to 8.5 to 2.0]; difference 6.0 [2.1 to 9.8]). Interdialytic weight gain declined in the 135 mEq/L arm and was unchanged in the 138 mEq/L arm (−0.3 [−0.5 to 0.0] versus 0.3 [0.0 to 0.6] kg over 6 months; difference [−0.6 (−0.1 to −1.0)] kg). In the extension phase, raising dialysate sodium concentration from 135 to 140 mEq/L was associated with an increase in interdialytic weight gain (0.2 [0.1 to 0.3] kg) and predialysis BP (7.0 [4.8 to 9.2]/3.9 [2.6 to 5.1] mm Hg) and a reduction in IDH (odds ratio, 0.66 [0.45 to 0.97]).

Conclusions

Use of a dialysate sodium concentration of 135 as compared with 138 mEq/L was associated with a small reduction in interdialytic weight gain without affecting IDH or predialysis BP, but with an increase in symptoms. Raising dialysate sodium concentration from 135 to 140 mEq/L was associated with a reduction in IDH, small increase in interdialytic weight gain, and marked increase in predialysis BP.

Clinical Trial registration number

NCT03144817.

Introduction

In 2014, US opinion leaders recommended lowering dialysate sodium concentration from 140 to between 134 and 138 mEq/L as a strategy to improve volume and BP control.¹ A contemporary US study had shown that most in-center patients were prescribed a dialysate sodium concentration of 140 mEq/L despite a mean predialysis serum sodium concentration of 137 mEq/L.² Accordingly, dialysis would yield a net diffusive sodium gain in some patients, depending on the ultrafiltration volume. It has been recommended³ that dialysate sodium be less than (not just equal to) the serum sodium concentration in order to overcome the Gibbs–Donnan effect (sodium trapping by negatively charged proteins at the blood membrane interface) and in consideration of the margin of error between the delivered and prescribed dialysate Na concentrations.⁴

Lowering dialysate sodium concentration to below 140 mEq/L remains controversial. Whether it improves volume control or BP or any other outcome with modern dialysis technology is unproven, and such benefits, if they exist, must be weighed against higher risks of intradialytic hypotension (IDH) and symptoms.^5,6 There have been few randomized comparisons with hypotonic dialysate sodium concentrations. The largest trial to date, involving 99 participants in New Zealand, found treatment with dialysate sodium 135 versus 140 mEq/L to be associated with a 0.5 kg average reduction in interdialytic weight gain,⁷ but had no effect on BP, IDH, or left ventricular mass index. The trial involved self-care patients with an average treatment duration of 5 hours; accordingly, findings may not be broadly generalizable.

Given ongoing uncertainty, we conducted a pilot clinical trial randomizing patients on maintenance thrice-weekly in-center hemodialysis to a dialysate sodium concentration of 138 versus 135 mEq/L for 6 months, followed by a 3-month single-arm crossover period in which sodium was increased from 135 to 140 mEq/L, to evaluate the effect of dialysate sodium on feasibility and tolerability, interdialytic weight gain, BP, and other parameters.

Methods

Trial Design

The study was conducted at two Boston, MA, area dialysis units operated by Dialysis Clinic Inc. (DCI) (a medium-sized not-for-profit national dialysis organization). Patients on thrice-weekly in-center hemodialysis were randomized 2:1 to dialysate sodium 135 versus 138 mEq/L for 6 months, which was followed by a 3-month extension phase in which dialysate sodium concentration was increased from 135 to 140 mEq/L in the low-arm participants. Participants were recruited between April 20, 2017, and June 30, 2018, and randomized between May 24, 2017, and August 17, 2018. The last participant follow-up was on April 2, 2019.

Study Population

Patients were eligible if they (1) had dialyzed in an outpatient unit 3× per week for ≥90 days; (2) had hypertension (2-week averaged predialysis systolic BP ≥140 mm Hg or use of ≥1 antihypertensive agent); and (3) had been treated with a dialysate sodium concentration of 138 mEq/L for ≥12 weeks. Exclusion criteria were (1) intra- or postdialysis systolic BP <90 mm Hg in ≥20% treatments in the previous month; (2) inability to speak English, Cantonese, or Mandarin; (3) unable to provide informed consent; or (4) life expectancy <1 year.

Intervention

A dialysate sodium concentration of 138 mEq/L was chosen as the standard arm because it was the most commonly prescribed sodium concentration before study start. Patients randomized to the low arm had dialysate sodium lowered by 1 mEq/L every 2 weeks until 135 mEq/L or the lowest tolerated sodium above 135 mEq/L. After 6 months, dialysate sodium was then raised to 140 mEq/L for an additional 12 weeks in low-arm participants only. The rationale for using 140 mEq/L as opposed to returning to 138 mEq/L was a concern that a 3 mEq/L difference may not have been large enough to have effects and in consideration of Dialysis Outcomes and Practice Patterns Study data⁵ and expert opinion that a dialysate sodium <140 mEq/L may be of greater harm than benefit.⁶

Outcomes

The primary outcome was difference in change in the rate of IDH from baseline to the end of the randomized study across treatment arms. IDH was defined as intra- or postdialysis systolic BP <90 mm Hg⁸ (Supplemental Appendix “Detailed Methods”). Other safety outcomes included difference in change in dialysis disequilibrium symptoms between baseline and the end of the randomized study across study arms. Dialysis disequilibrium symptoms consisted of eight self-reported items of headache, disorientation, not thinking clearly, restlessness, muscle twitching, blurred vision, nausea, and drowsiness, each weighted 1 (never) through 5 (very often), with a total score ranging 8–40 (Supplemental Table 1). The difference in the number of emergency room (ER) visits/hospitalizations between study arms was also assessed. Feasibility was assessed as the number of participants dialyzing at sodium 135 mEq/L at study end. Secondary outcomes included change in interdialytic weight gain, predialysis BP, extracellular fluid volume (ECFV), and others, described subsequently.

Measurements

Weight and BP measurements were performed per usual care. Interdialytic weight gain was calculated as postdialysis weight of the previous session subtracted from predialysis weight of the current session. Participants were asked about interim ER visits or hospitalizations at study visits, and discharge summaries were obtained. Self-administered questionnaires (the Dialysis Thirst Inventory,⁹ Xerostomia Inventory,¹⁰ dialysis disequilibrium symptoms and the recovery question¹¹ bioimpedance, Critline monitoring and laboratory measurements were done at milepost visits of baseline, 6, 12, and 24 weeks after randomization and, for the extension study, at 12 weeks after switch to dialysate sodium of 140 mEq/L. The recovery question was presented with 5 response options of <2 to >12 hours. Pre- and postdialysis plasma sodium concentrations were measured by indirect potentiometry at the DCI central laboratory in Nashville, TN (Clinical Laboratory Certified Amendments), on two separate days within a 2-week window of the milepost visits, and the values were averaged. Volume was measured before the start of the midweek treatment using a multifrequency total body bioimpedance device (InBody S20 [InBody USA]) and reported as extracellular water/total body water, with normal range 0.36–0.40 and values >0.40 considered volume overload.^12,13 The change in relative blood volume from the beginning to the end of the midweek treatment was measured using Crit-Line. Treatment data autopopulate DCI's electronic health information system, Darwin, using a direct machine interface. Data that were not collected as part of routine practice were entered into a web-based electronic data capturing system (REDCap).

Randomization and Blinding

Randomization schedules were generated by a research coordinator uninvolved in this study through ordering study arm assignment into blocks of eight according to a random number generator. The randomization schedule was entered into REDCap and remained concealed to the study team. Dialysate sodium concentration change was prescribed by the principal investigator; the patient and the patient's nephrologist were not made aware of the change, although the dialysate sodium concentration could be accessed on dialysis machine display screens.

Statistical Analyses

Primary analyses were conducted on the basis of an intent-to-treat model, including all randomized participants. Change in continuous outcome variables (2-week averages) by randomized group was modeled using linear mixed-effects models with random intercepts, random slopes, and a linear function for time. Statistical significance of the differences in slopes (change from baseline) across treatment arms was tested with adjustment for baseline values. Event rates for binary outcomes, such as occurrence of IDH, were compared using Poisson regression with the number of treatments as the offset. Models were adjusted for variables in which there was baseline imbalance across treatment arms (race considered as Black versus non-Black), congestive heart failure, and systolic BP. In the extension study, parameters during the 12-week pre-extension period (when the dialysate sodium concentration was 135 mEq/L) were compared with those during the 12-week period with dialysate sodium concentration 140 mEq/L using linear mixed-effects models with a random intercept for subject and a fixed-effect covariate for the indicator variable denoting period. Paired t tests were used to compare outcomes that were measured once after sodium concentration was increased to 140 mEq/L. Analyses were performed using R (Version 3.1.0) and SAS (Version 9.4, Cary, NC).

A formal statistical power analysis was not performed. The target enrollment of 40 patients was, at the time this study was designed, approximately three-fold larger than a previous study that had shown statistically significant effects on interdialytic weight gain and BP at this difference in dialysate sodium levels.¹⁴ The study was approved by the Tufts University Institutional Review Board, and all participants provided informed consent. The trial was registered at ClinicalTrials.gov (NCT03144817) on May 9, 2017.

Results

Of 177 patients at the two participating units, 133 were ineligible or declined participation (Supplemental Figure 1). Of the 42 participants randomized, six were censored, four of whom had kidney transplant and two were transferred out of the participating unit. The mean±SD age at baseline was 61±15 years, predialysis BP was 150±16/78±10 mm Hg, and treatment duration was 3.7±0.3 hours (Table 1). The 135 mEq/L arm had fewer Black patients, more patients with congestive heart failure, and a higher incidence of IDH (15.7 [9.5 to 25.8] versus 8.6 [31.0 to 24.1] episodes per 100 treatments, respectively).

Table 1.

Baseline characteristics of a two-site randomized clinical trial examining the effects of dialysate sodium concentration

Characteristic	Total (N=42)	Dialysate Sodium 138 mEq/L (n=13, 31%)	Dialysate Sodium 135 mEq/L (n=29, 69%)
Age, yr	61±15	61±12	61±16
Male, n (%)	26 (62)	8 (62)	18 (62)
Race, n (%)
Asian	7 (18)	0 (0)	7 (27)
Black	9 (23)	3 (23)	6 (23)
White	23 (59)	10 (76)	13 (50)
Cause of kidney failure, n (%)
Diabetes	15 (36)	5 (39)	10 (35)
Hypertension	12 (29)	3 (23)	9 (31)
GN	8 (19)	2 (15)	6 (21)
Other	7 (17)	3 (23)	4 (14)
Dialysis vintage, mo	34 (17 to 71)	30 (14 to 71)	35 (23 to 67)
Comorbidity, n (%)
Myocardial infarction	4 (10)	2 (15)	2 (7)
Congestive heart failure	7 (17)	1 (8)	6 (21)
Peripheral vascular disease	1 (2)	0 (0)	1 (4)
Body mass index, g/m2	27.6±5.7	27.7±3.8	27.6±6.4
Duration of dialysis (hours per session)	3.7±0.3	3.7±0.3	3.7±0.4
Predialysis plasma sodium, mEq/La	136.7±2.5	136.8±2.1	136.7±2.6
Postdialysis plasma sodium, mEq/La	136.2±2.5	136.3±1.9	136.1±2.7
BP, mm Hg
Predialysis systolic	150±16	153±22	148±12
Predialysis diastolic	78±10	79±7	78±11
Homeb systolic	134±24	142±29	130±20
Homeb diastolic	71±13	71±16	71±11
Interdialytic weight gain (% of estimated dry weight)	2.9±1.4	2.9±1.0	2.8±1.5
Interdialytic weight gain, kg	2.2±1.0	2.3±0.7	2.2±1.1
IDH per 100 treatments	13.6 (8.7 to 21.3)	8.6 (3.1 to 24.1)	15.7 (9.5 to 25.8)
Time to recover from dialysis, h, n (%)
<2	16 (39)	5 (42)	11 (38)
2–6	15 (37)	4 (33)	11 (38)
7–12	7 (17)	3 (25)	4 (14)
>12	3 (7)	0 (0)	3 (10)

Continuous variables expressed as mean±SD or mean (95% confidence intervals) for the incidence of intradialytic hypotension. IDH, intradialytic hypotension.

Missing values: home BP (three standard arm and eight low arm).

^aMean of values measured on two separate dates during baseline.

^bBP obtained the morning after the midweek dialysis treatment.

Randomization Phase Comparing Dialysate Sodium of 135 and 138 mEq/L

The decrease in dialysate sodium concentration to 135 mEq/L was tolerated by 27 of 29 participants (93%); two remained at 136 mEq/L for the duration of the trial. The effects of the intervention as compared with control are summarized in Table 2. There was no difference in the incidence of IDH. Symptoms declined in the 138 mEq/L arm and were unchanged in the 135 mEq/L arm (Table 2 and Supplemental Table 2). There was no difference in hospitalizations/ER visits, which occurred in 5 of 13 patients (38%) in the 138 mEq/L arm (incidence rate [95% confidence interval]) (7.3 [2.3 to 12.4] per 100 patient-months) versus 9 of 29 (31%) in the 135 mEq/L arm (incidence rate [95% confidence interval] 6.7 [2.9 to 10.6] per 100 patient-months; difference 0.6 [−6.9 to 5.8] P = 0.86). The postdialysis plasma sodium concentration decreased in the 135 mEq/L arm and was unchanged in the 138 mEq/L arm. There was no difference in predialysis plasma sodium concentration; postdialysis plasma sodium concentration declined in the 135 mEq/L arm and was unchanged in the 138 mEq/L arm (Figure 1). Interdialytic weight gain decreased in the 135 mEq/L arm and was unchanged in the 138 mEq/L arm (Figure 2). There was no difference across treatment arms in rate of change over 6 months in predialysis systolic (Supplemental Figure 2) or diastolic (Supplemental Figure 3) BP, postdialysis sitting or standing BP, recovery question (Supplemental Figure 4), or other secondary outcomes.

Table 2.

Changes in parameters by randomization allocation

Parameter	Dialysate Sodium 138 mEq/L (n=13)					Dialysate Sodium 135 mEq/L (n=29)					Difference Between Study Arms
	Baseline (n=13)	Month 3 (n=12)	Month 6 (n=12)	Model Slope (Change per 6 mo)a (95% CI)	Within Group, P Valueb	Baseline (n=29)	Month 3 (n=27)	Month 6 (n=25)	Model Slope (Change per 6 mo)a (95% CI)	Within Group, P Valueb	Between-Group Difference in Slopes (95% CI)	Between Group, P Valuec
Primary outcome
IDHg per 100 treatments	8.6 (3.1 to 24.1)	6.1 (1.5 to 24.0) (n=11)d	13.4 (5.6 to 32.5)	2.8e (0.8 to 9.5)	0.10	15.7 (9.5 to 25.8)	26.8 (17.5 to 40.9)	24.7 (16.3 to 37.4)	2.7 (1.1 to 6.2)	0.02	1.0e (0.3 to 3.6)	0.95
Secondary outcomes
Dialysis disequilibrium symptomsf	12.6±6.2 (n=12)d	8.3±5.1 (n=9)d	8.1±5.9 (n=9)d	−5.3 (−8.5 to −2.0)	0.002	9.5±5.8	8.7±5.4 (n=25)d	10.3±7.4 (n=23)d	0.7 (−1.4 to 2.7)	0.51	6.0 (2.1 to 9.8)	0.003
Interdialytic weight gain, kg	2.3±0.7	2.0±0.8	2.2±0.9	0.3 (0.0 to 0.6)	0.05	2.2±1.1	1.8±1.1	1.9±1.1	−0.3 (−0.5 to 0.0)	0.02	−0.6 (−1.0 to −0.2)	0.003
Interdialytic weight gain, % of EDW	2.9±1.0	2.6±1.1	2.9±1.2	0.5 (0.0 to 0.9)	0.03	2.8±1.5	2.3±1.3	2.2±1.1	−0.4 (−0.7 to −0.1)	0.003	−0.9 (−1.4 to −0.4)	0.001
Predialysis plasma sodium, mEq/L	137±2	137±2	136±3	0 (−2 to 1)	0.57	137±3	135±3	136±3	−1 (−2 to 0.1)	0.08	−1 (−2 to 1)	0.55
Postdialysis plasma sodium, mEq/L	136±2	137±3	137±3 (n=11)d	0 (−1 to 2)	0.73	136±3	134±3	134±4	−2 (−3 to −1)	0.0004	−2.0 (−4 to 0)	0.02
Predialysis sitting systolic BP, mm Hg	153±22	150±30	155±26	0 (−10 to 9)	0.97	148±12	144±14	146±16	−5 (−12 to 1)	0.12	−5 (−16 to 7)	0.39
Predialysis sitting diastolic BP, mm Hg	79±7	78±13	80±9	0.0 (−4 to 4)	0.99	78±11	76±9	78±11	−3 (−6 to 0)	0.10	−3 (−8 to 3)	0.35
Postdialysis sitting systolic BP, mm Hg	148±19	147±28	146±17	−3 (−15 to 9)	0.62	137±17	131±20	131±23	−5 (−13 to 3)	0.23	−2 (−17 to 13)	0.79
Postdialysis sitting diastolic BP, mm Hg	78±8	77±11	76±9	−2 (−8 to 3)	0.40	73±11	69±11	70±11	−3 (−6 to 1)	0.14	0 (−7 to 6)	0.90
Lowest intradialytic MAP, mm Hg	84±12	82±12	84±14	−1 (−9 to 8)	0.89	79±14	73±14	75±14	−4 (−9 to 2)	0.19	−3 (−13 to 7)	0.53
Postdialysis weight, kg	78.5±14.2	77.1±14.3	77.8±14.0	−0.5 (−2.9 to 1.9)	0.70	78.7±18.1	80.0±18.9	81.3±19.3	1.0 (−0.6 to 2.6)	0.23	1.5 (−1.4 to 4.4)	0.32
Xerostomia Index	23.2±9.1 (n=12)d	20.7±5.1 (n=9)d	22.1±4.9 (n=9)d	−2.7 (−8.2 to 2.9)	0.33	25.0±9.0	23.8±10.2 (n=25)d	26.7±11.8 (n=23)d	1.8 (−1.7 to 5.2)	0.32	4.4 (−2.1 to 11.0)	0.17
Thirst Inventory	18.8±5.8 (n=12)d	17.0±5.1 (n=9)d	16.8±4.4 (n=9)d	−2.1 (−6.5 to 2.2)	0.34	16.3±6.3	15.5±7.7 (n=25)d	17.4±7.8 (n=23)d	1.3 (−1.4 to 4.0)	0.35	3.4 (−1.7 to 8.6)	0.19
RBVh slope	13.4±7.1	11.3±6.2 (n=12)d	13.5±10.4 (n=9)d	1.0 (−4.1 to 6.2)	0.69	12.0±6.2	10.4±5.6 (n=24)d	11.4±6.5 (n=23)d	−0.5 (−3.8 to 2.7)	0.74	−1.6 (−7.7 to 4.5)	0.61
Predialysis ECFV/TBWi	0.40±0.01	0.40±0.05 (n=12)d	0.40±0.01 (n=11)d	0.00 (−0.01 to 0.01)	0.99	0.40±0.01	0.40±0.01 (n=25)d	0.40±0.01	0.00 (−0.01 to 0.01)	0.85	0.00 (−0.01 to 0.01)	0.92

Descriptive statistics are shown in the first three columns of treatment group, as observed mean±SD, except intradialytic hypotension rate, which is a mean with 95% confidence interval. CI, confidence interval; ECFV, extracellular fluid volume; EDW, estimated dry weight; IDH, intradialytic hypotension; MAP, mean arterial pressure; RBV, relative blood volume; TBW, total body water.

^aModel slope is the rate of change of the parameter per 6 months. For example, for dialysis disequilibrium symptoms, a β coefficient of –5.3 indicates a decline in the symptom score of 5.3 over 6 months in the dialysate sodium 138 mEq/L arm. Models are adjusted for covariates for which there was imbalance between treatment arms at baseline, which are race (Black versus non-Black), congestive heart failure, and predialysis systolic BP.

^bWithin-group p value assesses whether the rate of change is different from 0 in the respective treatment arm.

^cThe between-group difference p value tests whether the model slopes from each treatment arm are different.

^dMissing data: The sample size is that shown at the top of each column unless indicated in brackets in the respective cell.

^eThe difference in rate of change over 6 months between treatment arms for this parameter is a ratio of slopes.

^fDialysis disequilibrium symptoms is a weighted score of eight items, which were occurrence during or after dialysis of headache, disorientation, not thinking clearly, restlessness, muscle twitching, blurred vision, nausea, and drowsiness, each weighted one (never) through five (very often) with total score ranging 8–40.

^gIntradialytic hypotension is defined as one or more occurrences of an intra- or postdialysis systolic BP of <90 mm Hg one or more times during treatment

^hΔ Relative blood volume slope is the change in relative blood volume (%) as measured by Crit-Line between the start and end of a hemodialysis session.

ⁱExtracellular fluid volume/total body water is a measure of overhydration by total body bioimpedance. Normal is 0.36–0.40; >0.40 is considered extracellular volume excess. It was measured before dialysis. See text for details.

Figure 1.

Pre- and postdialysis plasma sodium concentration over the randomized study by treatment arm. (A) Predialysis plasma sodiumconcentration. (B) Postdialysis plasma sodium concentration. Figures show side by side box plots with “Standard” as dialysate sodium 138 mEq/L (black) and “Low” as 135 mEq/L (gray). Solid dots represent the mean values, and the open dots are outliers. Numbers along the y axis show the number of participants at each time period.

Figure 2.

Interdialytic weight gain over the randomized study by treatment arm. The figure shows side by side box plots with “Standard” as dialysate sodium 138 mEq/L (black) and “Low” as 135 mEq/L (gray). Solid dots represent the mean values, and the open dots are outliers. Numbers along the x axis show the number of participants at each time period.

Predialysis plasma sodium concentration ≤129 mEq/L occurred 12 times (out of 345 measurements) in eight participants, of whom six were in the 135 mEq/L arm; two of these six participants had predialysis sodium concentration ≤129 mEq/L twice and one had three occurrences; both participants in the 138 mEq/L arm had a single occurrence of plasma sodium concentration ≤129 mEq/L. Postdialysis plasma sodium concentration ≤129 mEq/L occurred 17 times (out of 342 measurements) in ten participants, of whom nine were in the 135 mEq/L group, five of whom had more than one occurrence. There was a single occurrence in one participant in the 138 mEq/L arm.

Outcomes with Increasing Dialysate Sodium Concentration from 135 to 140 mEq/L

Four of the 29 participants in the 135 mEq/L arm dropped out before the extension study (three kidney transplant, one transfer). For the primary outcome (Table 3), raising dialysate sodium concentration from 135 to 140 mEq/L was associated with a 33% decrease in the rate of IDH. Results for the secondary outcomes were as follows: no change in dialysis disequilibrium symptoms and a significant increase in predialysis plasma sodium concentration (1.02 [0.42 to 1.61] mEq/L), postdialysis plasma sodium concentration (2.56 [1.57 to 3.55] mEq/L), predialysis systolic (7.0 [4.8 to 9.2] mm Hg) and diastolic (3.9 [2.6 to 5.1] mm Hg) BP, and interdialytic weight gain (0.20 [0.06 to 0.33] kg). The increase in predialysis systolic BP (Figure 3A) and interdialytic weight gain (Figure 3B) occurred early, at the first assessment (2 weeks) after the change in dialysate sodium concentration. Other outcomes were not statistically significant.

Table 3.

Within-participant comparison of parameters measured during the pre-extension randomized phase (dialysate sodium 135 mEq/L) and nonrandomized extension phase (dialysate sodium 140 mEq/L)

Parameter	During the Period of Dialysate Sodium 135 mEq/L	During the Period of Dialysate Sodium 140 mEq/L	Mean Differencea	P Value
Primary outcome
IDH per 100 patient treatments (incidence rate ratio)b	24.7 (19.5 to 31.2)	16.4 (12.2 to 22.0)	0.66 (0.45 to 0.97)	0.04
Secondary outcomes
Dialysis disequilibrium symptomsc	10.2±7.5	10.3±7.5	0.1 (−1.4 to 1.5)	0.90
Interdialytic weight change (% estimated dry weight)	1.7±1.1	1.9±1.2	0.3 (0.0 to 0.5)	0.02
Interdialytic weight change, kg	2.1±1.1	2.4±1.6	0.2 (0.1 to 0.3)	0.005
Predialysis plasma sodium, mEq/L	137±3	138±3	1 (0 to 2)	<0.001
Postdialysis plasma sodium, mEq/L	134±3	137±4	3 (2 to 4)	<0.001
Predialysis systolic BP, mm Hg	144±16	151±14	7 (5 to 9)	<0.001
Predialysis diastolic BP, mm Hg	76±10	80±10	4 (3 to 5)	<0.01
Postdialysis sitting systolic BP, mm Hg	133±23	135±20	3 (0 to 5)	0.07
Postdialysis sitting diastolic BP, mm Hg	70±10	72±12	2 (1 to 4)	0.001
Postdialysis weight, kg	80.5±18.8	80.5±19.2	−0.2 (−0.5 to 0.2)	0.36
Lowest intradialytic MAP, mm Hg	75±14	76±12	1 (−3 to 5)	0.51
Xerostomia Index	26.7±11.8	28.7±9.6	2.0 (−0.6 to 4.5)	0.12
Thirst Questionnaire	17.4±7.8	17.7±7.3	0.3 (−2.3 to 2.9)	0.84
Predialysis ECFV/TBWd	0.40±0.01	0.40±0.01	0.001 (−0.002 to 0.004)	0.45

Missing data: The sample size is 25 with exception of dialysis disequilibrium symptoms (n=22), pre- and postdialysis plasma sodium (n=24), and Thirst Questionnaire and Xerostomia Index (n=23). Data are presented as mean±SD or mean difference (95% confidence interval). ECFV, extracellular fluid volume; IDH, intradialytic hypotension; MAP, mean arterial pressure; TBW, total body water.

^aAnalyzed using mixed-effects regression with the exception of mean arterial pressure, which is a paired t test.

^bIntradialytic hypotension is defined as one or more occurrences of an intra- or postdialysis systolic BP of <90 mm Hg one or more times during treatment. Expressed as mean (SD) for continuous variables and rate per 100 patient treatments for count variable.

^cDialysis disequilibrium symptoms is a weighted score of eight items—headache, disorientation, not thinking clearly, restlessness, muscle twitching, blurred vision, nausea, and drowsiness, each weighted one (never) through five (very often) with total score ranging 8–40.

^dExtracellular fluid volume/total body water is a measure of overhydration by total body bioimpedance. Normal is 0.36–0.40; >0.40 is considered extracellular volume excess. See text for details.

Figure 3.

Change in predialysis systolic BP and interdialytic weight gain associated with raising the dialysate sodium concentration from 135 to 140 mEq/L in the nonrandomized extension phase. (A) Predialysis systolic BP. (B) Interdialytic weight gain. Figures show box plots with “Before Study Extension Date” representing the treatment period at dialysate sodium 135 (gray) and “After Study Extension Date” at dialysate sodium 140 mEq/L (black). Numbers along the x axis represent the number of participants with measures at each time period. Percent interdialytic weight gain is the difference between the predialysis weight of the current treatment and the postdialysis weight of the previous treatment divided by the estimated dry weight.

Discussion

In this randomized trial of patients on maintenance hemodialysis, a dialysate sodium concentration of 135 as compared with 138 mEq/L was tolerated by most participants but was associated with more symptoms, and while there was a small reduction in interdialytic weight gain, there was no change in predialytic BP, ECFV, IDH, or other outcomes. Raising sodium concentration to 140 from 135 mEq/L in the extension study was associated with a reduction in IDH, a large increase in predialysis BP, a modest increase in interdialytic weight gain, and no change in postdialysis weight or ECFV. These results suggest that broad use of a mildly hypotonic dialysate sodium concentration (135 mEq/L) in the setting of typical hemodialysis prescriptions in the United States would modestly affect interdialytic weight gain without suggestion of other substantial benefits and would increase dialysis disequilibrium symptoms while use of a mildly hypertonic sodium concentration (140 mEq/L) would reduce IDH occurrences but may substantially increase interdialytic BP.

There are only a few randomized comparisons of the range of dialysate sodium concentration studied here. A 2019 meta-analysis¹⁵ of randomized trials comparing two dialysate sodium concentrations (12 trials; N=239), categorized according to dialysate sodium concentration as hypotonic (<138 mEq/L), neutral (138–140 mEq/L), or hypertonic (>140 mEq/L), found that interdialytic weight gain was the sole parameter (of various BP and volume measures) on which dialysate sodium concentration consistently had an effect. The magnitude of effect correlated with the difference in sodium concentration across arms. This dose-response effect was shown in a subsequent meta-regression analysis.¹⁶ The present study's findings are consistent with the three trials that tested a hypotonic versus neutral dialysate sodium concentration. A trial from Brazil comparing 135 versus 138 mEq/L (N=38) found no effect on interdialytic weight gain,¹⁷ the Sodium Lowering in Dialysate Trial (SoLID) study comparing 135 versus 140 mEq/L (N=99) found a 0.5 kg greater reduction with the lower dialysate sodium concentration,⁷ a trial in China comparing 136 versus 138 mEq/L (N=64) found a 0.4 kg greater reduction in the 136 mEq/L arm (of unclear statistical significance as the between-group difference was not tested¹⁸), and the present study found a 0.6 kg reduction in the 135 versus 138 mEq/L arm. None of these studies found statistically significant effects of the lower sodium concentration on BP, post weight, or ECFV, although there was a trend toward a reduction in BP. Possibly, this would have been statistically significant with a large sample size, but the BP-lowering effect is not likely to be large. The implication of an increase or decrease in BP through altering dialysate sodium concentration is unclear. The optimal BP target is unknown, and the sole randomized comparison of two BP targets (a pilot study) showed signals of harm with treating to a predialysis systolic BP ≤140 versus ≤155 mm Hg.²⁷ Left ventricular mass index was not studied here but was shown to significantly decline in the Chinese study comparing 136 versus 138 mEq/L,⁷ but not in the SoLID study comparing 135 versus 140 mEq/L.¹⁸

The results regarding IDH incidence and other adverse effects are mixed. The SoLID study noted no increase in IDH; however, the rate of IDH was quite low (4 as compared with 13.6 per 100 treatments in this study).⁷ By contrast, Beduschi et al. described a nearly three-fold higher rate of IDH with the lower sodium concentration.¹⁷ While the IDH incidence was not different in the present study, the baseline difference in IDH across arms may have reduced the ability to detect an effect. Moreover, it is noted that raising sodium concentration from 135 to 140 mEq/L reduced the IDH rate by 33%, and dialysis disequilibrium symptoms were greater in the 135 versus 138 mEq/L arm. Two of 29 patients (who were selected for having a baseline IDH rate <20%) did not tolerate the dialysate sodium concentration of 135 mEq/L in this study, and five of 49 patients in the intervention arm of the SoLID study also dropped out because of symptoms.¹⁷ Reduced perfusion of the myocardium and brain has been observed with much smaller changes in BP than that defining IDH in this study.^19–21 The higher frequency of hyponatremia (plasma sodium ≤129 mEq/L) in the 135 mEq/L arm in our study is of uncertain clinical consequence, but another potential concern. A recent retrospective study of more than two million patients from 875 Fresenius dialysis clinics in 25 countries found the use of a dialysate Na concentration <138 mEq/L to be associated with mortality, independently of serum sodium or other confounding variables.²⁸ Considering the current data, we would suggest hypotonic dialysate be used in selected patients and with close monitoring.

A finding that also deserves discussion is the marked increase in predialysis BP that was seen with raising dialysate sodium concentration from 135 to 140 mEq/L, despite a modest change in interdialytic weight gain and no change in volume (by bioimpedance or postdialysis weight). While the extension period is not randomized, we believe this result is reasonably robust, given that the BP rise is seen immediately (Figure 2), and this is a within-patient comparison without patient drop-out. One possible explanation is nonosmotically active or water-free sodium storage in the skin and muscle, which is postulated to regulate BP independently of volume.²² Na²³-labeled magnetic resonance imaging studies have shown higher skin and muscle sodium content in patients dialyzing at 140 mEq/L than those dialyzing at 137 mEq/L¹¹ and that tissue sodium correlates with BP and the presence of left ventricular hypertrophy.^23,24 Predialysis plasma sodium concentration increased with the change from 135 to 140 mEq/L in this study, which raises another possible mechanism—direct effects of plasma sodium in stiffening the vascular endothelium.^25,26

This study has several strengths. Few randomized trials have tested a dialysate sodium as hypotonic as 135 mEq/L. We had complete data to ascertain IDH episodes and BP and volume parameters. There was limited drop-out. While the extension phase was not randomized, effects of the change in dialysate sodium concentration were seen immediately, which strengthens the case for causality. Studying the effects of 140 mEq/L in the extension phase is a strength because this remains a commonly prescribed dialysate sodium concentration in the present day and is the standard arm of an ongoing multinational randomized clinical trial with a primary outcome of cardiovascular events (ClinicalTrials.gov identifier: NCT02823821). The main limitation is the small sample size, which reduces the statistical power available to detect effects and precision in estimates and increases the potential for imbalance across treatment arms, which could also reduce the ability to detect differences that may exist. We did not have accurate data about antihypertensive medications. Possibly, antihypertensive medication use declined and masked a significant BP-lowering effect of a dialysate sodium concentration of 135 mEq/L. The primary nephrologist was not blinded to the intervention, which potentially introduces bias; however, the information is not usually accessed during rounds, so we are uncertain of the extent to which this was an issue. Symptoms declined in the 138 mEq/L arm and remained unchanged in the 135 mEq/L arm, which raises the possibility that the difference between arms reflects regression to the mean. Future studies should assess symptoms to delineate whether this is a true effect. The relatively small study population and the 6-month study duration do not allow for assessment of cardiovascular events or mortality or other important outcomes such as short and long-term cognitive function.

In conclusion, in a randomized clinical trial of patients on thrice-weekly maintenance in-center hemodialysis, lowering dialysate sodium concentration from 138 to 135 mEq/L as compared with 138 mEq/L was associated with more symptoms and a small reduction in interdialytic weight gain without effects on predialysis BP, postdialysis weight, ECFV, or the incidence of IDH. Raising sodium concentration to 140 from 135 mEq/L was associated with an immediate and marked increase in predialysis BP, increases in interdialytic weight gain and predialysis plasma sodium, and a reduction in IDH. These data do not support use of a mildly hypotonic over neutral dialysate sodium concentration as a strategy that would be widely effective and safe for improving volume control or BP, although a large trial with clinical outcomes is needed to be definitive about this.

Supplementary Material

cjasn-19-712-s001.pdf ^{(369.5KB, pdf)}

Disclosures

C.M. Hsu is supported by NIH/NCATS grant KL2TR002545 and 1K12TR004384. C.M. Hsu’s funder had no role in study design, data collection, reporting, or the decision to submit. Caroline M. Hsu also reports Ownership Interest: Amazon, Google, and Microsoft; Research Funding: the Paul Teschan Research Fund grant from DCI, and salary support to my institution from DCI. D.C. Miskulin and D.E. Weiner receive salary support from DCI. D.C. Miskulin also reports Research Funding: Reata Inc. and Regulus Inc. H. Tighiouart reports Ownership Interest: Ford, Citi, Merck, Oracle, Bank of America, Exxon Mobile, TJ Maxx, Cheniere Energy, Organon. D.E. Weiner reports Research Funding: All compensation paid to Tufts MC: Bayer (site PI), Cara (site PI), and Vertex (site PI); Advisory or Leadership Role: Co Editor-in-Chief, NKF Primer on Kidney Diseases, 8th Edition; Editor-in-Chief, Kidney Medicine; Medical Director of Clinical Research, DCI.; Member, ASN Quality and Policy Committees and ASN Representative to KCP; and Member, Scientific Advisory Board, National Kidney Foundation; and Other Interests or Relationships: Member, Adjudications Committee, ProKidney REACT Trial (George Institute CRO) and Member, Safety and Clinical Events Committee for “A Prospective, Multi-Center, Open-Label Assessment of Efficacy and Safety of Quanta SC+ for Home Hemodialysis” Trial (Avania CRO).

Funding

This study was funded by the Paul Teschan Grant award from DCI. Inbody generously donated the S20 bioimpedance monitor for the duration of this study. REDCap was supported by the Tufts CTSI via the National Center for Advancing Translational Sciences, National Institutes of Health, Award Number UL1TR002544.

Author Contributions

Conceptualization: Dana C. Miskulin.

Data curation: Dana C. Miskulin.

Formal analysis: Dana C. Miskulin, Hocine Tighiouart, Daniel E. Weiner.

Funding acquisition: Dana C. Miskulin.

Investigation: Dana C. Miskulin.

Methodology: Dana C. Miskulin, Hocine Tighiouart, Daniel E. Weiner.

Resources: Dana C. Miskulin.

Supervision: Dana C. Miskulin.

Validation: Dana C. Miskulin, Hocine Tighiouart, Daniel E. Weiner.

Visualization: Dana C. Miskulin.

Writing – original draft: Caroline M. Hsu, Dana C. Miskulin, Daniel E. Weiner.

Writing – review & editing: Caroline M. Hsu, Dana C. Miskulin, Daniel E. Weiner.

Data Sharing Statement

The individual participant data and data collection forms that underlie the results reported in this article, after deidentification, are available for sharing. Data will be available 3 months after publication and for 3 years. Researchers who provide a methodologically sound proposal (with any type of analysis) will be granted access to the data. Requests should be directed to dana.miskulin@tuftsmedicine.org.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/CJN/B861.

Detailed Methods

Supplemental Figure 1. Study flowchart of patient disposition.

Supplemental Figure 2. Predialysis systolic BP by randomization group.

Supplemental Figure 3. Predialysis diastolic BP comparison by randomization group.

Supplemental Figure 4. Time to recovery after dialysis by randomization group.

Supplemental Table 1. Dialysis symptom questionnaire.

Supplemental Table 2. Dialysis symptoms by randomization group.