Tissue sodium stores in peritoneal dialysis and hemodialysis patients determined by sodium-23 magnetic resonance imaging

Abstract

Background

Tissue sodium (Na⁺) content in patients on maintenance hemodialysis (MHD) and peritoneal dialysis (PD) was previously explored using ²³Na⁺ magnetic resonance imaging (²³NaMRI). Larger studies would provide a better understanding of Na⁺ stores in patients on dialysis as well as the factors influencing this Na⁺ accumulation.

Methods

In this cross-sectional study, we quantified the calf muscle and skin Na⁺ content in 162 subjects (10 PD, 33 MHD patients and 119 controls) using ²³NaMRI. Plasma levels of interleukin-6 (IL-6) and high-sensitivity C-reactive protein (hsCRP) were measured to assess systemic inflammation. Sixty-four subjects had repeat ²³NaMRI scans that were analyzed to assess the repeatability of the ²³NaMRI measurements.

Results

Patients on MHD and PD exhibited significantly higher muscle and skin Na⁺ accumulation compared with controls. African American patients on dialysis exhibited greater muscle and skin Na⁺ content compared with non–African Americans. Multivariable analysis showed that older age was associated with both higher muscle and skin Na⁺ and male sex was associated with increased skin Na⁺ deposition. Greater ultrafiltration was associated with lower skin Na⁺ in patients on PD (Spearman’s ρ = −0.68, P = 0.035). Higher plasma IL-6 and hsCRP levels correlated with increased muscle and skin Na⁺ content in the overall study population. Patients with higher baseline tissue Na⁺ content exhibited greater variability in tissue Na⁺ stores on repeat measurements.

Conclusions

Our findings highlight greater muscle and skin Na⁺ content in dialysis patients compared with controls without kidney disease. Tissue Na⁺ deposition and systemic inflammation seen in dialysis patients might influence one another bidirectionally.

KEY LEARNING POINTS

What is already known about this subject?

• Several small studies have suggested that maintenance hemodialysis (MHD) patients have increased muscle and skin sodium (Na⁺) content compared with healthy controls measured by ²³Na⁺ magnetic resonance imaging (²³NaMRI).

• Na⁺ deposition in tissues leads to local and systemic inflammation.

• Systemic inflammation is a risk factor for cardiovascular disease and protein-energy wasting in end-stage kidney disease patients.

What this study adds?

• Compared with a large, diverse group of controls with normal kidney function, patients on maintenance dialysis exhibit increased muscle and skin Na⁺ accumulation.

• Increased ultrafiltration volume in patients on peritoneal dialysis (PD) is associated with lower skin Na⁺ content.

• Muscle and skin Na⁺ concentrations and the levels of inflammatory markers (interleukin-6 and high-sensitivity C-reactive protein) are positively correlated in maintenance dialysis patients.

What impact this may have on practice or policy?

• This study provides further insights into the Na⁺ balance and accumulation in MHD and PD patients.

• The abnormally high tissue Na⁺ content and associated inflammation could be exacerbating cardiovascular disease and protein-energy wasting risk in dialysis patients, which could guide clinicians.

• Our results regarding the link between ²³NaMRI technique could be an effective tool for routine clinical use in the future for risk assessment and to guide therapy in patients on dialysis when prescribing dialysis treatment and modulating salt intake.

INTRODUCTION

Recent evidence has challenged the conventional two-compartment sodium (Na⁺) homeostasis model that was originally described in previous studies [1, 2]. Noninvasive quantification of interstitial Na⁺ in skin and muscle by ²³Na magnetic resonance imaging (²³NaMRI) allowed better recognition of this third compartment for Na⁺ in the body. Multiple studies have shown that muscle and skin interstitium house remarkable amounts of Na⁺ [3–7]. A disruption in the Na⁺ clearance from these tissue stores was found to be associated with salt-sensitive increases in blood pressure (BP) [8–10].

Patients with advanced kidney disease, especially patients on maintenance hemodialysis (MHD) and peritoneal dialysis (PD), are prone to positive Na⁺ balance and its associated complications. While ²³NaMRI represents an appealing tool to estimate tissue Na⁺ accumulation, data to date are conflicting and limited information is available regarding tissue Na⁺ stores in individuals with reduced kidney function [7, 11]. For example, Dahlmann et al. [11] did not report substantially increased tissue Na⁺ stores in their patients on HD compared with 27 age-matched controls. In a more recent study, Qirjazi et al. [7] reported increased tissue Na⁺ content in patients on MHD and PD compared with 10 healthy controls. Notably, they reported Na⁺ content in soleus muscle, whereas Dahlmann et al. [11] reported the complete calf muscle, leading to problems with standardization for additional studies.

In this study we hypothesize that end-stage kidney disease (ESKD) patients on maintenance dialysis will exhibit higher tissue Na⁺ deposition compared with controls without kidney disease. We also hypothesize that the Na⁺ content of muscle and skin is associated with inflammation in maintenance dialysis patients based on reports showing that increased Na⁺ content in the tissue microenvironment promotes immune cell activation. An additional aim of our study is to provide a range of normal values for skin and muscle Na⁺ content to develop an effective tool for routine clinical use in the future for risk assessment and to guide therapy. We also aim to define the demographic and clinical characteristics influencing muscle and skin Na⁺ stores.

MATERIALS AND METHODS

Subjects

This cross-sectional study included 33 MHD patients, 10 PD patients and 119 controls without kidney disease recruited at Vanderbilt University Medical Center and Veterans Affairs Tennessee Valley Healthcare System for multiple research studies between 2014 and 2020. All subjects signed written informed consent prior to their participation. HD patients >18 years of age who were on MHD treatment thrice weekly, with an adequate dose (single-pool K_t/V >1.2) for at least 6 months were included. Inclusion criteria for PD patients were to be on a stable peritoneal prescription (K_t/V > 1.7 or total creatinine clearance >50 mL/week/1.73 m²) for 3 months with glucose lactate–buffered PD solutions. Patients with active infectious or inflammatory disease (i.e. active infection or connective tissue disorder), liver disease, human immunodeficiency virus, active cancer or a history of hospitalization within 1 month of the study were excluded. Demographic and clinical data were collected from the subjects and chart review. BP was measured after 5 min of seated rest during the ²³NaMRI visit. Plasma samples were collected within 2 weeks of the ²³NaMRI scan in all subjects.

²³NaMRI

Multinuclear ²³NaMRI of the calf was performed using a Na⁺ knee coil in a 3.0 Tesla scanner (Philips Healthcare, Best, The Netherlands). The ²³NaMRIs were done after HD for MHD patients; the time between HD and the ²³NaMRI measurement was not standardized. Phantoms of 10, 20, 30 and 40 mmol/L NaCl aqueous solutions were used for signal calibration. The widest part of the left lower leg (calf region) was scanned using previously published methods [3, 4, 11] [repetition time (TR)/echo time (TE) 130/0.99 ms; field-of-view (FOV) 192 × 192 mm², in-plane spatial resolution 3 × 3 mm²; slice thickness 30 mm; flip angle 90°; scan duration 15 min 54 s]. In an identical FOV as Na⁺ imaging, multipoint Dixon imaging was acquired using the body coil (TR 200 ms, TE1 1.15 ms, TE2 2.30 ms; in-plane spatial resolution 1 × 1 mm²; slice thickness 5 mm; flip angle 90°; scan duration 4 min) for anatomical reference. High spatial resolution imaging was performed in an identical FOV including T₂-weighted with fat suppression (spectral attenuated inversion recovery, TR/TE = 3500/60 ms, spatial resolution 0.3 × 0.3 × 5 mm³) and T₁-weighted (TR/TE = 991/15 ms, spatial resolution 0.54 × 0.54 × 5.5 mm³) imaging. The ²³NaMRI protocol was fixed for all measurements, including the repeated scans.

Inflammatory markers

Plasma levels of interleukin-6 (IL-6) were analyzed using human IL-6 enzyme-linked immunosorbent assay (ELISA) kits (Quantikine HS Human IL-6 Immunoassay; HS600C; R&D Systems, Minneapolis, MN, USA) with a sensitivity of 0.09 pg/mL and an interassay coefficient of variation <4.9%. High-sensitivity C-reactive protein (hsCRP) levels were measured by a latex immunoturbidimetric assay on the Architect c16000 System (Abbott, Abbott Park, IL, USA).

Urine Na⁺ measurements

All urine specimens were centrifuged at 1500 rpm for 20 min at 4°C, supernatants were aliquoted and stored at −80°C within 24 h of collection. Na⁺ levels were measured using an IL 943 flame photometer (Instrumentation Laboratory, Bedford, MA, USA). To estimate the 24-h urinary Na⁺ excretion, we used the Tanaka equation (21.98 × XNa^0.392, where XNa is the spot urine Na⁺ concentration) [12].

Statistical analysis

All data are described as medians and interquartile ranges (IQRs) for continuous variables and as frequencies and percentages for categorical variables. Complete case analysis was used to handle missing data. Kruskal–Wallis and Wilcoxon rank-sum tests were used to assess differences between the three study groups and Spearman’s rank correlation was used to evaluate the association of continuous variables of interest. The trend line and confidence interval (CI) in each scatterplot were estimated with linear regression, except when displaying the correlation of skin and muscle Na⁺. In that setting, the trend line and CI were estimated with a Demming regression and quantile bootstrap, respectively. To create multivariable models that meet the model assumptions of regression, Box–Cox transformation was used. The associations between demographics (age, sex and race), clinical characteristics [body mass index (BMI), systolic BP (SBP) and presence of diabetes mellitus (DM)], inflammatory markers (IL-6 or hsCRP) and Box–Cox transformations of muscle and skin Na⁺ concentrations were explored. To guard against overfitting, the number of parameters (covariate terms) in each model was limited to one parameter for every 10 observations. For both muscle and skin Na⁺ outcomes, the Box–Cox transformation parameter was −1, which corresponds to the transformation $1-1/x.\mathrm{ }$The calculation used numerical integration. Residual diagnostics were generated that showed the model assumptions were met and that the analysis of the transformed Na⁺ was a reasonable approach. To summarize model results from the regression, partial effect estimates for each covariate were generated. The effects were estimated to compare the mean difference in outcome between the 25th and 75th percentiles of the covariate in question, while other covariates were held to the mode or median. For categorical covariates, the partial effects were estimated to compare the mean difference of each category to a reference group. Tests of association were performed with the linear model F-test. The agreement between the initial and follow-up measurements of skin and muscle Na⁺ was visualized with Bland–Altman plots [13]. Analyses were performed using R version 3.6.3 (R Foundation for Statistical Computing, Vienna, Austria) [14] with packages mcr [15] and blandr [16].

RESULTS

Study population

The demographic and clinical characteristics of the study population are shown in Table 1. Cardiovascular disease, DM and hypertension were more prevalent among patients on dialysis. None of the study subjects had a kidney transplant or were on any immunosuppressive therapy before or at the time of the ²³NaMRI scan.

Table 1.

Demographic and clinical characteristics of the study population at the time of baseline ²³NaMRI measurements

Parameters	Overall (n = 162)	PD (n = 10)	MHD (n = 33)	Controls (n = 119)
Age (years), median (IQR)	50 (38–59)	55 (48–61)	56 (45–65)	48 (37–56)
Male	73 (45)	3 (30)	22 (67)	48 (40)
African American origin	75 (46)	4 (40)	24 (73)	47 (40)
BMI (kg/m2), median (IQR)	27.7 (24.4–32.8)	23.8 (22–27.6)	28.9 (25.1–33.5)	27.7 (24.5–32.6)
SBP (mmHg), median (IQR)	128 (118–139)	136 (116–154)	138 (119–153)	126 (118–135)
DBP (mmHg), median (IQR)	76 (69–83)	76 (65–82)	71 (65–81)	77 (71–84)
Hypertension	49 (30)	8 (80)	30 (91)	11 (9)
DM	10 (6)	3 (30)	6 (18)	1 (1)
Coronary artery disease	13 (8)	2 (20)	10 (30)	1 (1)
Myocardial infarction	4 (3)	1 (10)	2 (6)	1 (1)
Stroke	3 (2)	0 (0)	2 (6)	1 (1)
Dialysis vintage (months), median (IQR)	42 (19–82)	12 (6–30)	55 (29–94)	–
Predialysis BUN (mg/dL), median (IQR)	–	–	47 (41–62)	–
Weekly Kt/V, median (IQR)	–	1.9 (1.9–2.2)	5.1 (4.3–5.3)	–
Interdialytic weight gain (kg), median (IQR)	–	–	2.2 (1.5–2.9)	–
Etiology of CKD				–
Hypertension	20 (12)	3 (30)	17 (52)	–
DM	6 (4)	3 (30)	3 (9)	–
Glomerulonephritis	7 (4)	3 (30)	4 (12)	–
Polycystic kidney disease	3 (2)	0 (0)	3 (9)	–
Othera	7 (4)	1 (10)	6 (18)	–
Medications
ACE inhibitors	11 (7)	1 (8)	8 (24)	2 (2)
Angiotensin-receptor blockers	17 (10)	8 (67)	5 (15)	4 (3)
β-blockers	30 (18)	5 (42)	20 (60)	5 (4)
Loop diuretics	7 (4)	4 (33)	3 (9)	0(0)
Thiazide diuretics	2 (1)	0 (0)	0 (0)	2 (2)
K+-sparing diuretics	1 (1)	0 (0)	1 (3)	(0)
Laboratory parameters, median (IQR)
Serum Na+ (mmol/L)	139 (138–141)	139 (135–142)	140 (138–141)	139 (138–140)
Serum creatinine (mg/dL)	0.9 (0.8–6.1)	10.1 (7.4–13)	9.2 (8.1–11.6)	0.8 (0.8–0.9)
hsCRP (mg/dL)	1.3 (0.5–4.5)	10.6 (0.6–12.8)	4.8 (1.6–11.1)	1 (0.3–2.7)
IL-6 (pg/mL)	2.1 (1.1–3.3)	9.9 (2.9–10.5)	5 (2.9–9.4)	1.6 (1–2.8)

Data are presented as n (%) unless stated otherwise. ^aOther includes focal segmental glomerulosclerosis, bilateral nephrolithiasis, posterior urethral valves and vasculitis. Some data were missing for hsCRP (n = 156 patients; 33 HD, 9 PD and 114 controls) and IL-6 (n = 141 patients; 18 HD, 9 PD and 114 controls). ACE, angiotensin-converting enzyme; CKD, chronic kidney disease; DBP, diastolic blood pressure.

Tissue Na⁺ stores in patients on dialysis

Median muscle and skin Na⁺ concentrations were higher in patients on MHD and PD compared with controls: muscle Na⁺ 21.8 mmol/L (IQR 18.6–25.1, P < 0.001) in MHD, 22.1 mmol/L (IQR 19.3–26.7, P < 0.001) in PD and 16.8 mmol/L (IQR 14.9–18.6) in controls; skin Na⁺ 17.9 mmol/L (IQR 13.2–21.3, P < 0.001) in MHD, 23 mmol/L (IQR 16.8–28.2, P < 0.001) in PD and 13.6 mmol/L (IQR 11.4–16.9) in controls (Figures 1 and 2). Muscle or skin Na⁺ content did not differ significantly among patients on PD and MHD (P = 0.672 for muscle Na⁺, P = 0.122 for skin Na⁺) (Figure 2).

FIGURE 1.

Anatomical and ²³NaMRI of the left calf of a 51-year-old African American male control, a 61-year-old African American male on MHD and a 63-year-old African American male on PD.

FIGURE 2.

(A) Muscle Na⁺ content in MHD, PD patients and controls. (B) Skin Na⁺ content in MHD, PD patients and controls. ^#P < 0.001 compared with controls.

After adjusting for demographics (age, race and sex) and clinical characteristics (BMI, SBP and the presence of DM), both MHD and PD groups were independently associated with higher muscle and skin Na⁺ deposition compared with controls (Table 2). Being on dialysis remained significant even after adjusting for IL-6 or hsCRP levels (Supplementary data, Tables S1 and S2).

Table 2.

Multivariable linear regression analysis exploring the associations between clinical characteristics and tissue Na⁺ content (N = 162)

	Muscle Na+		Skin Na+
Variables	Partial effecta (95% CI)	P-value	Partial effecta (95% CI)	P-value
Age (59:38 years)	1.48 (0.78–2.19)	<0.001	2.04 (1.23–2.84)	<0.001
Race (African American)	1.76 (0.72–2.80)	<0.001	−0.26 (−1.18–0.66)	0.605
Sex (male)	−0.28 (−0.57–1.13)	0.536	2.75 (1.57–3.93)	<0.001
BMI (33:24 kg/m2)	−0.41 (−0.98–0.16)	0.161	0.31 (−0.31–0.93)	0.332
SBP (139:118 mmHg)	0.22 (−0.29–0.74)	0.395	0.03 (−0.50–0.55)	0.959
DM	0.69 (−1.39–2.77)	0.517	0.24 (−1.86–2.35)	0.888
Study group (MHD)	2.36 (0.88–3.83)	<0.001	1.51 (−0.04–3.05)	0.039
Study group (PD)	4.16 (1.28–7.05)	<0.001	6.23 (1.98–10.48)	<0.001

^aPartial effects were estimated to compare the mean difference in outcome between the 25th and 75th percentiles of the covariate in question (provided in parentheses), while other covariates were held to the mode or median. For categorical covariates, the partial effects were estimated to compare the mean difference of each category with the reference group. Repeat ²³NaMRI measurements were not included in the analysis.

Tissue Na⁺ stores in controls

Eleven controls had hypertension (Table 1) and 95% CIs for muscle and skin Na⁺ content in controls (n = 119) were 16.7–17.8 mmol/L and 13.9–15.7 mmol/L, respectively. Tissue Na⁺ stores of controls with and without hypertension did not differ significantly [muscle Na, 16.7 mmol/L (95% CI 14.7–18.4) in controls without hypertension, 17.5 mmol/L (95% CI 16.7–19) with hypertension, P = 0.097; skin Na⁺ 13.4 mmol/L (95% CI 11.4–16.7) in controls without hypertension, 16.9 mmol/L (95% CI 14.4–17.7) with hypertension, P = 0.088]. After excluding the controls with other comorbidities such as diabetes, hypertension or any other cardiovascular diseases, the median muscle Na⁺ content in healthy controls was 16.8 mmol/L (IQR 14.8–18.7) and skin Na⁺ was 13.6 mmol/L (IQR 11.4–17). In the remaing 106 healthy controls, the 95% CIs for muscle and skin Na⁺ were 16.5–17.8 mmol/L and 13.7–15.6 mmol/L, respectively.

Factors influencing tissue Na⁺ content

Muscle and skin Na⁺ content based on sex, race and the presence of DM in each group are provided in Table 3. Notably, in the overall study population, subjects with DM exhibited higher muscle and skin Na⁺ content compared with subjects without DM. African Americans exhibit significantly higher muscle Na⁺ compared with non–African Americans in the overall study population and in MHD patients, but not in controls (Table 3). Male sex was associated with greater skin Na⁺ concentration in the multivariable linear regression analysis including controls (Supplementary data, Table S3).

Table 3.

Baseline muscle and skin Na⁺ concentrations based on race, sex and presence of DM

	Muscle Na+ (mmol/L)			Skin Na+ (mmol/L)
	African American	Non–African American	P-value	African American	Non–African American	P-value
Race
All	18.7 (15.9–23.5)	16.9 (14.9–18.5)	<0.001	14.5 (11.6–18.8)	14.6 (12.2–17.5)	0.850
PD	26.4 (22.1–30.8)	20.2 (18.1–23.8)	0.143	30.0 (22.3–35.9)	20.6 (15.9–23.3)	0.143
MHD	24.7 (19.4–26.5)	18.6 (14.5–21.0)	0.021	18.7 (13.8–21.4)	17.0 (13.2–18.2)	0.501
Control	17.1 (15.1–19.4)	16.7 (14.8–18.1)	0.269	12.8 (10.8–16.0)	14.4 (11.9–17.0)	0.054
Sex	Male	Female	P-value	Male	Female	P-value
All	17.9 (15.6–20.8)	17.3 (15.0–20.5)	0.442	17.0 (13.8–20.4)	12.8 (11.1–15.9)	<0.001
PD	19.9 (18.8–26.6)	22.8 (20.2–26.0)	0.917	24.3 (21.5–30.0)	22.4 (15.5–26.5)	0.458
MHD	23.0 (18.4–25.0)	21.0 (19.0–27.9)	0.550	18.3 (13.4–22.9)	17.2 (13.7–19.9)	0.601
Control	16.8 (15.5–18.4)	16.8 (14.6–18.7)	0.632	16.8 (14.0–18.9)	12.2 (10.9–14.6)	<0.001
DM	Yes	No	P-value	Yes	No	P-value
All	24.7 (20.0–26.3)	17.3 (15.2–19.7)	<0.001	21.1 (18.0–24.1)	14.2 (11.6–17.6)	0.002
PD	19.9 (18.8–22.3)	22.8 (20.2–28.7)	0.458	23.5 (21.1–23.9)	22.4 (15.5–32.6)	0.917
MHD	25.7 (21.4–28.7)	21.4 (17.2–25.0)	0.151	20.9 (18.0–26.6)	17.2 (13.1–20.7)	0.125
Control	25.2 (25.2–25.2)	16.8 (14.9–18.5)	0.097	12.8 (12.8–12.8)	13.6 (11.4–16.9)	0.795

Values are presented as median (IQR). Repeat ²³NaMRI measurements were excluded.

Increasing age correlated with higher muscle and skin Na⁺ content in the overall study population (Supplementary data, Figure S1). When the groups were analyzed separately, older age correlated with greater muscle and skin Na⁺ accumulation in patients on MHD but not in patients on PD (Tables 4 and 5). In the multivariable analysis, aging was associated with higher muscle and skin Na⁺ content (Supplementary data, Table S3).

Table 4.

Correlation between the clinical characteristics and baseline muscle Na⁺ content in each study group

Variables	Muscle Na+ (mmol/L)
	PD	P-value	MHD	P-value	Control	P-value
Age	−0.04 (−0.65–0.67)	0.907	0.51 (0.21–0.72)	0.002	0.17 (−0.00–0.34)	0.058
BMI	−0.04 (−0.7–0.68)	0.654	−0.29 (−0.63–0.11)	0.105	0.03 (−0.18–0.22)	0.734
SBP	0.09 (−0.11–0.28)	0.332	0.16 (−0.23–0.51)	0.378	0.14 (−0.54–0.66)	0.700
DBP	0.1 (−0.09–0.27)	0.297	−0.04 (−0.43–0.35)	0.819	0.02 (−0.7–0.78)	0.973
Predialysis BUN	–	–	−0.42 (−0.66 to −0.09)	0.016	–	–
hsCRP	0.16 (−0.73–0.88)	0.680	0.17 (−0.21–0.57)	0.357	0.02 (−0.17–0.19)	0.843
IL-6	0.4 (−0.47–0.93)	0.291	0.18 (−0.4–0.61)	0.467	−0.03 (−0.23–0.17)	0.747
Time between HD and 23NaMRI	–	–	0.15 (−0.24–0.53)	0.418	–	–

Values are presented as r (95% CI) and P-values. Repeat ²³NaMRI measurements were not included in the analysis.

Table 5.

Correlation between the clinical characteristics and baseline skin Na⁺ content in each study group

Variables	Skin Na+ (mmol/L)
	PD	P-value	MHD	P-value	Control	P-value
Age	0.38 (−0.33–0.87)	0.283	0.52 (0.16–0.79)	0.002	0.4 (0.23–0.54)	<0.001
BMI	0.03 (−0.67–0.74)	0.934	0.06 (−0.28–0.38)	0.757	0.18 (−0.01–0.35)	0.045
SBP	0.14 (−0.05–0.32)	0.116	0.14 (−0.29–0.53)	0.444	0.24 (−0.42–0.75)	0.498
DBP	0.13 (−0.06–0.31)	0.152	−0.11 (−0.47–0.27)	0.557	−0.1 (−0.74–0.65)	0.785
Predialysis BUN	–	–	−0.23 (−0.53–0.13)	0.203	–	–
hsCRP	0.16 (−0.73–0.88)	0.68	0.34 (0.01–0.63)	0.056	−0.04 (−0.2–0.16)	0.706
IL-6	0.4 (−0.47–0.93)	0.291	0.36 (−0.21–0.73)	0.14	0.15 (−0.04–0.33)	0.112
Time between HD and 23NaMRI	–	–	0.01 (−0.34–0.36)	0.967	–	–

Values are presented as r (95% CI) and P-values. Repeat ²³NaMRI measurements were not included in the analysis.

SBP was positively correlated with both muscle and skin Na⁺ in the overall study population (Supplementary data, Figure S2). However, SBP was not associated with muscle or skin Na⁺ content when groups were analyzed separately (Tables 4 and 5 and Supplementary data, Table S3).

BMI showed a slightly positive correlation with skin Na⁺ content in controls but did not show any other correlation with muscle or skin Na⁺ in other groups (Tables 4 and 5).

Higher ultrafiltration (UF) volume was associated with lower skin Na⁺ content but not with muscle Na⁺ content in the PD group (Figure 3). No significant correlation was found between tissue Na⁺ concentrations and urine output or total fluid output in patients on PD (Supplementary data, Figure S3).

FIGURE 3.

Correlation between (A and B) tissue Na⁺ concentrations and UF volume per day in the PD group and (C and D) UF volume per dialysis session in the hemodialysis group (Spearman’s correlation).

Higher urine output was associated with lower muscle Na⁺ content but not with skin Na⁺ among patients on MHD (Supplementary data, Figure S4). No significant correlation was found between tissue Na⁺ content and UF volume or total fluid output (Figure 3).

The time of the ²³NaMRI scan did not have a significant impact on the tissue Na⁺ results. Muscle or skin Na⁺ content did not differ significantly among patients who were scanned before 12 p.m. or after 12 p.m. (Table 6). The median time interval between the end of the HD session and the ²³NaMRI scan was 26.3 h (IQR 22.6–44.7). There was no correlation between the time after HD and muscle or skin Na⁺ content (Tables 4 and 5).

Table 6.

Muscle and skin Na⁺ concentrations based on the time of the²³NaMRI scan

	Muscle Na+ (mmol/L) skin Na+ (mmol/L)
Time of 23NaMRI scan	Before 12 p.m. (n = 69)	After 12 p.m. (n = 93)	P-value	Before 12 p.m. (n = 69)	After 12 p.m. (n = 93)	P-value
All subjects	18 (15.8–20.6)	17.1 (14.8–20.3)	0.155	13.7 (11.5–17.3)	15.1 (12.2–18.3)	0.232

Values are presented as median (IQR). Repeat ²³NaMRI measurements were excluded.

Tissue Na⁺ stores and inflammatory markers

Table 1 shows the hsCRP and IL-6 levels in each study group. Overall, our data showed a positive correlation between the muscle and skin Na⁺ depositioncontent and both inflammatory markers (Figure 4). However, the correlations between the tissue Na⁺ and hsCRP or IL-6 levels were not significant when the groups were analyzed separately (Tables 4 and 5 and Figure 5).

FIGURE 4.

Correlation between the tissue Na⁺ content and inflammatory markers: (A and B) hsCRP and (C and D) IL-6 when all subjects are combined. Higher plasma levels of both hsCRP and IL-6 were associated with increased muscle and skin Na⁺ concentrations in all subjects.

FIGURE 5.

Correlation between the tissue Na⁺ content and inflammatory markers: (A and B) hsCRP and (C and D) IL-6 in each group. Correlation coefficients and P-values are provided separately for each group in Tables 3 and 4.

Urinary Na⁺ excretion

The median spot urine Na⁺ concentration in controls was 37.3 mEq/L (IQR 24.9–72.5). The estimated 24-h urinary Na⁺ excretion in controls was 90.8 mEq/day (95% CI 77.5–117.8). We found no correlation between the tissue Na⁺ concentrations and 24-h urinary Na⁺ excretion (muscle Na⁺: r = −0.005, P = 0.965; skin Na⁺: r = 0.035, P = 0.745).

Repeat ²³NaMRI measurements

Sixty-four subjects (1 PD, 8 MHD and 55 controls) had repeat ²³NaMRI measurements. The median time interval between baseline and repeat measurements was 1.9 months. The characteristics of the study population, both at the time of baseline and follow-up measurements, are displayed in Supplementary data, Table S4. The degree of agreement between the baseline and follow-up tissue Na⁺ concentrations are provided in Figure 6 and Supplementary data, Figure S5. Bland–Altman plots showed that patients with higher baseline tissue Na⁺ stores exhibited greater variability in tissue Na⁺ stores on repeat measurements (Figures 6). Based on the results from the linear regression model, the time interval between the repeat and baseline measurements did not affect the agreement between the two ²³NaMRI scans [β = 0.02 (95% CI −0.08–0.12), P = 0.635 in the model for muscle Na⁺; or β = −0.07 (95% CI −0.21–0.06), P = 0.061 for skin Na⁺].

FIGURE 6.

Correlation between the baseline and follow-up (A) muscleand (C) skin Na⁺ measurements. (B and D) Bland–Altman plots showing the differences between the baseline and repeat ²³NaMRI measurements on the y-axis and the mean of baseline and repeat Na⁺ measurements on the x-axis. A total of 64 patients (1 PD, 8 MHD patients and 55 controls) had repeat NaMRI measurements. Notably, the agreement between the baseline and repeat measurement was greater for skin Na⁺. Lower mean muscle and skin Na⁺ content in the Bland–Altman plots seems to be associated with better agreement between the baseline and repeat measurements.

Correlation between muscle and skin Na⁺ content

Our data showed a positive correlation between the muscle and skin Na⁺ in all groups (Spearman’s ρ: r = 0.77, P = 0.014 in PD; r = 0.60, P < 0.001 in MHD; r = 0.47, P < 0.001 in controls) (Supplementary data, Figure S6). Notably, the correlation was stronger in the PD and MHD groups.

DISCUSSION

In this cross-sectional study we report that patients on MHD and PD have higher muscle and skin Na⁺ accumulation compared with controls without kidney disease. Overall, aging and being a dialysis patient were both independently associated with greater muscle and skin Na⁺ deposition and male sex was associated with higher skin Na⁺ content. In addition, African American patients on MHD exhibited higher muscle Na⁺ content compared with non–African Americans on MHD. We also found a positive correlation between tissue Na⁺ content and plasma levels of IL-6 and hsCRP in the overall study population, suggesting that tissue Na⁺ is regulated by multiple mechanisms.

Based on our cohort of 119 subjects without kidney disease, active infectious or inflammatory disease, we also report potential reference ranges for muscle and skin Na⁺ content of 16.7–17.8 mmol/L and 13.9–15.7 mmol/L, respectively. After excluding the controls with diabetes, hypertension or any other cardiovascular disease, the reference ranges based on the remaining 106 subjects were 16.5–17.8 mmol/L for muscle Na⁺ and 13.7–15.6 mmol/L for skin Na⁺.

The results of this study have several clinical implications. Accumulating evidence using the ²³NaMRI method shows that relevant amounts of Na⁺ are accumulated in the brain, muscle and skin [17]. People with ESKD constitute a particularly vulnerable population for Na⁺ storage due to their lack of ability to maintain Na⁺ and water balance. Accordingly, understanding the Na⁺ compartments in patients on PD and MHD may lead to a modulation in dialysis prescription and subsequent improvement in clinical outcomes.

There are conflicting reports in the literature regarding the extent of tissue Na⁺ content in maintenance dialysis patients compared with controls [7, 11]. While Dahlmann et al. [11] did not report increased tissue Na⁺ stores in German patients on HD, a recent study from Canada [7] reported increased tissue Na⁺ concentrations in patients on HD and PD compared with controls. However, the latter study reported Na⁺ content in the soleus muscle only instead of all muscles. Also, due to the relatively smaller number of HD patients, and especially controls, these studies were unable to explore muscle and skin Na⁺ with adjustment for demographic and clinical factors such as age, sex and race. They are also limited in terms of providing the so-called normal range that can be used as a reference to reliably compare data from patients with kidney disease, due to the small number of healthy controls with normal kidney function. With a greater number of HD patients and controls, our results validate the observations that patients on maintenance dialysis and PD with diverse clinical and demographic backgrounds have increased muscle and skin Na⁺ deposition. Our data also provide a reasonable reference range for future studies examining tissue Na⁺ accumulation in patients with kidney disease.

A significant finding in our study was that African Americans exhibited higher muscle Na⁺ content compared with non–African Americans among patients on MHD. This could be explained by various factors such as the natural tendency to retain more Na⁺ in African Americans [18], differences in epithelial Na⁺ channel activity, the renin–angiotensin–aldosterone system (RAAS) and several socioeconomic factors [19], all leading to increased Na⁺ deposition in the tissues. The clinical relevance of this increased muscle Na⁺ accumulation in African Americans in cardiovascular and kidney diseases should be explored. Our results related to the effects of age and sex on tissue Na⁺ content were also consistent with previously published studies [3, 4, 20].

UF volume is directly related to Na⁺ removal in patients on PD [21]. Furthermore, optimized UF volume and dialytic Na⁺ removal are associated with a reduction in the mortality of patients on PD [22, 23]. Our findings suggest that excess Na⁺ due to inadequate fluid and Na⁺ removal by PD could be partly retained as osmotically neutral Na⁺ in the skin, and tissue Na⁺ quantification by ²³NaMRI can be used as a marker for adequate solute removal. According to the three-pore model of conventional PD, Na⁺ is transported across the peritoneal membrane by convection (removal of Na⁺ coupled with water), diffusion (determined by diffusion gradient, volume and time) and peritoneal absorption (fluid and solutes absorbed to interstitial tissue and lymphatics), with convection being the most prominent Na⁺ transport mechanism. The diffusion time available for solute removal is of significance since short dwell time favors UF by free water movement across aquaporin-1 channels with no Na⁺ removal, while coupled-water removal of Na⁺ via the small pores requires longer dwells [24]. The inverse correlation observed between the UF volume and skin Na⁺ content in patients on PD suggests that vigorous UF, aiming to normalize not only the fluid balance but also the tissue Na⁺ content, should be considered in this population. Interestingly, a recent study showed a 4-fold increase in Na⁺ removal with a new zero-Na⁺ PD solution, suggesting another potential strategy to decrease tissue Na⁺ stores in patients on PD [25].

Recent evidence shows that muscle and skin Na⁺ could be involved in distinct pathways and thus have different determinants. Skin Na⁺ stores are regulated by immune mechanisms: macrophages sense hypertonicity in the skin and express tonicity enhancer-binding protein, which leads to vascular endothelial growth factor C release and subsequent hyperplasia of the lymph–capillary system in the skin [8, 10]. This facilitates the removal of Na⁺ and Cl⁻ from the skin interstitium. Disruption of this mechanism is associated with salt-sensitive hypertension [9]. Increased muscle Na⁺ content, on the other hand, was associated with insulin resistance in a recent pilot study from our group [5], suggesting that increasing Na⁺ accumulation in the muscle interstitium could be disrupting the insulin signaling pathway in the muscle either through direct or inflammation-related effects, leading to insulin resistance. In this study we showed that a positive correlation exists between tissue Na⁺ content and the inflammatory markers IL-6 and hsCRP. Although the associations were inconclusive in MHD and PD groups, likely due to the smaller sample size, the degree of associations will be more precise in a larger study. Recent evidence indicates that salt is an important driver of inflammation. High salt exposure leads to increased proinflammatory molecules in macrophages [26] and a proinflammatory T-helper 17 profile [27, 28]. The Na⁺-rich microenvironment in the skin or muscle interstitium may be a major culprit in the systemic inflammation seen in ESKD. Nevertheless, any causal inference about tissue Na⁺ accumulation and inflammation could only be hypothesized based on our findings.

It is also important to note that tissue Na⁺ is not regulated by a single mechanism, but rather is mediated by multiple mechanisms. In this study we observed that even after adjustment for demographics (age, sex and race), clinical characteristics (BMI, SBP, presence of diabetes) and inflammatory markers (IL-6 or hsCRP), patients on dialysis (either PD or MHD) were still associated with higher muscle and skin Na⁺ concentrations. This finding implies that factors other than those accounted for here, such as age and inflammation, contribute to the excess Na⁺ accumulation in these patients. We also found a negative correlation between predialysis blood urea nitrogen (BUN) and muscle Na⁺. Having a higher BUN is related to a better nutritional state (i.e. higher protein intake) which is associated with greater overall health and lower muscle Na⁺ in patients on MHD. However, with the current data we are unable to speculate about any causal relationship between protein intake, BUN and muscle Na⁺.

We also measured the reproducibility of the ²³NaMRI technique using repeat measurements from 64 subjects. The agreement between the baseline and follow-up measurements of the skin and muscle Na⁺ content was not perfect. This confirms the dynamic nature of this third compartment that is likely modified by multiple factors such as diet, medications, time of measurement relative to the dialysis session and dialysate Na⁺ concentration, which were not held constant among our study population. Our data show a stronger correlation between the muscle and skin Na⁺ in the PD and MHD groups compared with controls. A possible explanation for this finding could be that these two compartments operate through different pathways under normal conditions and become saturated in ESKD, causing the skin and muscle Na⁺ concentrations to converge.

Our study has several strengths and limitations. ²³NaMRI technology is still new and requires validation and generalization. Currently there are no established tissue Na⁺ targets for the ESKD population, as the currently available tissue Na⁺ data are still not enough to establish ‘normal ranges’ from a healthy population. Our cohort is the largest and most diverse control population reported so far, with 174 ²³NaMRI measurements from 119 controls, also being the first to include African Americans. Furthermore, our results are assumed to be valid regardless of the MRI vendor, as the ²³NaMRI signal intensity is calibrated to standard Na⁺ solutions to calculate the standardized tissue Na⁺ content for each scan, which should minimize the systemic differences due to different MRI scanners. On the other hand, the cross-sectional design of this study limits the ability to establish causal relationships between tissue Na⁺ content and other variables that are investigated. Although our sample size was relatively large, with 162 individuals, we had a relatively small number of patients on PD and MHD, which limited the power of our study to make inferences about these patients. The corrected R² in our full multivariable models with IL-6 was 0.350 and 0.360 for predicting muscle and skin Na⁺ content, respectively. This shows that about two-thirds of the variability in muscle and skin Na⁺ content is explained by other factors such as dietary Na⁺ intake, medication use, urinary Na⁺ excretion and inherent differences in Na⁺ regulatory mechanisms (RAAS or ionic transport mechanisms across the cell maintaining the Ca⁺⁺ and Na⁺ balance) that were not accounted for in this study. Parallel to this, as this study included data from five different studies, the time intervals between the blood draw for inflammatory markers, measurements of urine output, HD session and ²³NaMRI scans were not standardized in all patients. Hence the measurements may fluctuate from one HD session to another. One other important limitation is the absence of 24-h urine collection, which is the gold standard to assess dietary Na⁺ intake [29]. We used the Tanaka equation to estimate 24-h urinary Na⁺ excretion, which was previously shown to be an imperfect assessment [30]. This is important, as our findings in this study did not show a significant correlation between the dietary Na⁺ intake and tissue Na⁺, in contrast with what Braconnier et al. [31] and we [32] reported in other studies. In addition, the repeatability of the muscle and skin Na⁺ measurements using different ²³NaMRI scanners must be explored to further validate the reference levels reported here. Finally, we did not have a direct measurement of volume status in our study population. While the lack of any clinical findings related to volume overload in our inclusion criteria should minimize the impact of volume status, our results do not provide any insight into the relationship between volume overload and salt accumulation.

In conclusion, our findings highlight the relevance of the ²³NaMRI methodology and its potential implications in optimizing salt balance and potential removal strategies, especially for patients on dialysis. Compared with people without kidney disease, patients on MHD and PD both exhibit increased tissue Na⁺ stores. Aging is associated with higher muscle and skin Na⁺ content and males seem to exhibit higher skin Na⁺ deposition. The association between the inflammatory markers and Na⁺ stores suggests potential adverse metabolic consequences of tissue Na⁺ storage. While we show an acceptable correlation for within-subject repeatability, further studies incorporating additional data can provide more accurate information about the determinants of muscle and skin Na⁺ stores in humans.

SUPPLEMENTARY DATA

Supplementary data are available at ndt online.