Professional Documents
Culture Documents
Research Article
Keywords albumin, and spKt/V (p < 0.05 for each parameter). There was
Magnesium · Mortality · Hemodialysis · Risk factor a strong correlation between the baseline mean serum mag-
nesium concentration 1 year prior and the concentration 1
year later (r2 = 0.519, p < 0.001). After adjustment for con-
Abstract founding factors, multivariate Cox proportional hazards
Background: The relationship between magnesium and analysis showed hypomagnesemia to be an independent
mortality in hemodialysis patients has been evaluated in predictor of all-cause and cardiovascular mortality in chron-
several prospective studies, but few have assessed the risk of ic hemodialysis patients. Furthermore, subgroup analysis
all-cause mortality in elderly hemodialysis patients. The aim was performed, revealing that serum magnesium levels
of this study was to evaluate the association between mag- were still strongly associated with all-cause mortality and
nesium levels and the risk of cardiovascular and overall mor- cardiovascular mortality in patients older than 60 years, with
tality in elderly maintenance hemodialysis patients. Meth- HR values of 0.020 (95% CI 0.001–0.415) and 0.010 (95% CI
ods: This was a retrospective study, and patients undergoing 0.000–0.491), respectively. In addition, there were still sig-
maintenance hemodialysis were screened for eligibility at a nificant associations between the serum magnesium level
single dialysis center between July and December 2016. Pa- and all-cause mortality and cardiovascular mortality in elder-
tients were divided into two groups based on their magne- ly dialysis patients at the 6-month follow-up visit. Conclu-
sium levels: a low magnesium level group and a high mag- sion: Our study indicates that lower serum magnesium lev-
nesium level group. Associations between magnesium level els are strongly associated with cardiovascular and all-cause
and risk of cardiovascular and all-cause mortality were ana- mortality in maintenance hemodialysis patients, especially
lyzed with a Cox proportional hazards regression model. Re- in the short term and in those who are elderly. Factors affect-
sults: In total, 413 patients were included with a median fol- ing serum magnesium concentrations in hemodialysis pa-
low-up period of 12 months. We found that compared to tients should be investigated, and correcting hypomagnese-
patients with high magnesium levels, those with low mag- mia may benefit elderly hemodialysis patients.
nesium levels had significantly lower levels of hemoglobin, © 2019 The Author(s)
urea, creatinine, uric acid, phosphate, potassium, chloride, Published by S. Karger AG, Basel
© 2019 The Author(s) Jinghong Zhao, MD, PhD, or Dr. Jiachuan Xiong
Published by S. Karger AG, Basel Department of Nephrology, Kidney Center of PLA, Xinqiao Hospital
Army Medical University (Third Military Medical University)
karger@karger.com This article is licensed under the Creative Commons Attribution-
NonCommercial-NoDerivatives 4.0 International License (CC BY- Chongqing, 400037 (PR China)
www.karger.com/kdd E-Mail zhaojh @ tmmu.edu.cn or xiongjc @ tmmu.edu.cn
NC-ND) (http://www.karger.com/Services/OpenAccessLicense).
Usage and distribution for commercial purposes as well as any dis-
tribution of modified material requires written permission.
Introduction Biochemical Assays and Other Measurements
To measure serum albumin, calcium, phosphate, and Mg levels
using routine laboratory methods, blood was collected just before
Magnesium (Mg) is the fourth most abundant cation in the start of a dialysis session in a nonfasting state. Serum Mg was
the body and the second most important intracellular cat- measured by the isocitrate dehydrogenase enzymatic method
ion [1]. In recent years, the importance of Mg has been in- (Aqua-auto Kainos Mg Test Kit; Kainos Co. Ltd., Tokyo, Japan).
creasingly recognized due to the growing awareness that it The normal range of serum Mg concentrations in healthy subjects
is required as a cofactor in multiple enzymatic reactions and is 0.75–1.25 mmol/L. Mean values of measurements during the 6
months before January 2017 were used for analysis. To examine
that it plays an important role in neuromuscular processes variability in serum Mg concentrations, levels in most patients
[2]. Mg also functions in mineral bone metabolism, adenos- who had survived 1 year after enrollment in the study (during Oc-
ine triphosphate metabolism, neurotransmitter release and tober 2017 and March 2018, n = 351) were measured. Serum cal-
regulation of vascular tone, heart rhythm, and platelet-ac- cium concentrations were corrected by serum albumin concentra-
tivated thrombosis [3–5]. Changes in Mg homeostasis may tions, as follows: corrected calcium (mmol/L) = measured calcium
(mmol/L) + (40 – albumin [g/L])/40.
occur in patients with chronic kidney disease (CKD) and
end-stage renal disease (ESRD) [6]. Moreover, it has been Statistical Methods
reported that Mg deficiency is associated with vascular cal- Data are summarized as the mean ± SD. Differences in means
cifications, atherosclerosis, and cardiovascular disease between the two groups were evaluated by Student’s t test. Cate-
(CVD) and might increase the risk of sudden cardiac death gorical data were compared between groups by the χ2 test. p values
<0.05 were considered statistically significant. All calculations, in-
[7]. Overall, an understanding of the physiology of Mg han- cluding Cox proportional hazards models and Kaplan-Meier anal-
dling is important for CKD and ESRD patients. Recently, a ysis, were performed on a personal computer using statistical anal-
growing body of literature is associating hypomagnesemia ysis software SPSS 24.0 (IBM Co., Armonk, NY, USA).
with critical clinical endpoints, such as an increased risk of
CVD, arrhythmia, and mortality [8]. However, most of
these studies have focused on general dialysis patients with Results
a broad age range and long-term effects. Conversely, the as-
sociation between hypomagnesemia and mortality in elder- Baseline Characteristics of the Included Patients
ly maintenance hemodialysis (HD) patients, especially in In total, 413 patients were involved in this study. The
the short term, has not been well documented. In this study, average age of the patients was 50.4 ± 14.3 years, and
we investigated hypomagnesemia and the risk of mortality 57.4% were male. The HD duration was 43 months; 97%
in elderly maintenance HD patients. of the vascular access was via an arteriovenous fistula, and
diabetes patients accounted for 14.4% of the study popu-
lation. The mean BMI was 22.1 ± 3.3 kg/m2. Mg concen-
trations were normally distributed, with a mean ± SD of
Materials and Methods
1.01 ± 0.15 mmol/L, as shown in Figure 1. Mg concentra-
Patient Selection tions were also measured 1 year later to examine variabil-
This was a retrospective study. In total, 413 adult patients ity in serum Mg concentrations, and there was a strong
(mean age 50.4 ± 14.3 years) undergoing stable, regular HD at the correlation between the baseline mean serum Mg con-
purification center of the Department of Nephrology at Xinqiao centration and the Mg concentration 1 year later (r2 =
Hospital between July and December 2016 were screened for eli-
gibility. All data used in this study were collected from the database 0.519, p < 0.0001; Fig. 2). Patients were divided into two
of the Blood Purification Center’s Data Registry (BPCDR). Base- groups according to their baseline mean serum Mg con-
line data such as age, sex, body mass index (BMI), systolic blood centration, as follows: a low Mg level group (serum Mg
pressure, and serum levels of urea nitrogen, creatinine, albumin, <1.0 mmol/L, n = 199) and a high Mg level group (serum
blood hemoglobin, and intact parathyroid hormone were collect- Mg ≥1.0 mmol/L, n = 214). Table 1 shows the baseline
ed. In addition, the medications, including anti-hypertension, glu-
cose-lowering agent, calcium, antibiotics, proton pump inhibitor clinical characteristics of the HD patients in each of the
(PPI), and other commonly used drugs were analyzed. The pa- two groups. The levels of hemoglobin, urea, creatinine,
tients received dialysis three times a week, and the blood flow rate uric acid, phosphate, potassium, chloride, and albumin in
ranged from 230 to 350 mL/min during the procedure. A GAM- addition to spKt/V were significantly lower in patients
BRO AK 95 HD machine (Baxter International Inc., Lund, Swe- with low Mg concentrations than in those with high Mg
den) was used for HD, and the electrolyte concentration of the
dialysate fluid was as follows: sodium 140 mmol/L, potassium 2.0 concentrations (p < 0.05 for each parameter). There were
mmol/L, chloride 107 mmol/L, calcium 1.5 mmol/L, phosphate no significant differences in sex, age, duration of HD, vas-
0 mmol/L, Mg 0.5 mmol/L, and bicarbonate 35 mmol/L. cular access, or the presence of diabetes between the two
r = 0.519
1.2
50
1.0
0 0.8
0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
Baseline Mg, mmol/L
0.6
0.4 0.8 1.0 1.2 1.4 1.6 1.8
Baseline serum Mg concentration, mmol/L
Fig. 1. Distribution of baseline serum Mg levels.
Data are presented as n (%), mean ± SD, or median (range) as appropriate. Lower Mg group = Mg <1 mmol/L; Higher Mg group =
Mg ≥1 mmol/L. Mg, magnesium; AVF, arteriovenous fistula; BMI, body mass index; iPTH, intermittent parathyroid hormone; AST,
aspartate aminotransferase; ALT, alanine aminotransferase; ALP, alkaline phosphatase; PPI, proton pump inhibitor; CCB, calcium chan-
nel blocker; ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker.
HR 95% CI p value
HR, hazard ratio; BMI, body mass index; ALP, alkaline phosphatase; Mg, magnesium; iPTH, intermittent
parathyroid hormone.
All patients
Mg (per 1 mmol/L) 0.017 0.002–0.197 0.001 0.011 0.000–0.269 0.006
Mg group (low to high) 3.535 1.594–7.838 0.002 4.285 1.422–12.910 0.010
Age ≥60, 12-month follow-up
Mg (per 1 mmol/L) 0.020 0.001–0.415 0.012 0.010 0.000–0.491 0.021
Mg group (low to high) 4.358 1.474–12.882 0.008 5.525 1.224–24.934 0.026
Age ≥60, 6-month follow-up
Mg (per 1 mmol/L) 0.033 0.000–2.624 0.127 0.016 0.000–2.490 0.109
Mg group (low to high) 2.019 0.939–4.344 0.072 6.730 0.828–54.705 0.075
diovascular mortality, respectively. In addition, Ka- found that significant associations between the serum
plan-Meier analysis indicated significant differences in level of Mg and all-cause mortality (log-rank, p = 0.049)
all-cause mortality (log-rank, p = 0.003) and cardiovas- and cardiovascular mortality (log-rank, p = 0.038) in
cular mortality (log-rank, p = 0.012) between the two elderly dialysis patients remained during the 6-month
groups (Fig. 4a, b). Based on subgroup analysis, we also follow-up period (Fig. 4c, d).
Cumulative survival, %
90 log-rank = 11.096 90 log-rank = 7.992
p = 0.001 p = 0.005
Mg ≥1 mmol/L Mg ≥1 mmol/L
100 100
Cumulative survival, %
90 Cumulative survival, % 90
log-rank = 6.316 log-rank = 4.365
p = 0.012 p = 0.037
Mg ≥1 mmol/L Mg ≥1 mmol/L
Fig. 3. Kaplan-Meier analysis of Mg and mortality. a Overall mortality over a 12-month period. b Cardiovascular
mortality over a 12-month period. c Overall mortality over a 6-month period. d Cardiovascular mortality over a
6-month period.
Data are presented as n (%), mean ± SD, or median (range) as appropriate. Lower Mg group = Mg <1 mmol/L; Higher Mg group =
Mg ≥1 mmol/L. Mg, magnesium; AVF, arteriovenous fistula; BMI, body mass index; iPTH, intermittent parathyroid hormone; AST,
aspartate aminotransferase; ALT, alanine aminotransferase; ALP, alkaline phosphatase; PPI, proton pump inhibitor; CCB, calcium chan-
nel blocker; ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker.
90 90
Cumulative survival, %
Cumulative survival, %
80 log-rank = 8.572 80 log-rank = 6.328
p = 0.003 p = 0.012
70 Mg ≥1 mmol/L 70 Mg ≥1 mmol/L
100 100
90 90
Cumulative survival, %
80 Cumulative survival, % 80
log-rank = 3.860 log-rank = 4.327
p = 0.049 p = 0.038
70 Mg ≥1 mmol/L 70 Mg ≥1 mmol/L
Fig. 4. Kaplan-Meier analysis of Mg and mortality in elderly hemodialysis patients. a Overall mortality over a
12-month period. b Cardiovascular mortality over a 12-month period. c Overall mortality over a 6-month pe-
riod. d Cardiovascular mortality over a 6-month period.
presence of diabetes, which is consistent with the findings diabetes, lower albumin level, and lower phosphate level.
of previous reports. However, we also observed that pa- Furthermore, after adjustment for these confounders, a
tients with low serum Mg levels were significantly older lower serum Mg level remained a significant predictor for
and had significantly lower serum levels of albumin, phos- increased mortality according to multivariate Cox propor-
phate, and calcium than did those with high serum Mg tional analysis. Our results support that a lower level of
levels. These results may indicate that lower serum Mg lev- serum Mg is a significant predictor of mortality, indepen-
els are related to malnourishment in these patients. Selim dent of other parameters of malnourishment, including
et al. [14] reported an observable association between Mg the presence of diabetes, lower albumin levels, and lower
and mortality within a 5-year follow-up period, and Ago et phosphate levels. Hypomagnesemia may be influenced by
al. [21] showed that the same association could be observed medications, such as insulin, calcium, antibiotics, PPI, and
in a 1-year follow-up period. The results of our study indi- chemotherapeutic drugs. But in our study, there was no
cate that at the time of patient enrollment, the mean serum significant difference in medications between the two
Mg concentration over the previous 6 months can predict groups, which suggested that hypomagnesemia may be
survival, especially in elderly patients. We found that low- caused be the decreased intake or other reasons. This issue
er serum Mg levels can increase all-cause and cardiovascu- needs to be further addressed.
lar mortality in the short term. Other significant predictors Previous experimental research suggested that Mg
were increased age, longer duration of HD, the presence of plays a crucial role in cardiovascular function [22], and