Introduction
Chronic kidney disease (CKD) is an increasing and leading public health challenge worldwide, as the primary contributor to global mortality and morbidity [
1]. Nearly 10% of adults worldwide suffer from CKD, resulting in nearly 35.0 million years of healthy life lost and 1.2 million deaths every year [
2]. Therefore, it is of vital importance to recognize relevant and adaptable risk factors for the delay or prevention of CKD.
Magnesium (Mg), as the second most important intracellular positive ion and the fourth abundant cation in the human body [
3,
4], has been linked to many diseases, including hypertension [
5], cardiovascular disease [
6], diabetes [
7], anemia [
8], as well as CKD [
9]. Several epidemiologic studies have been conducted to dissect the connection between serum Mg status and CKD. For instance, the ARIC study suggested that lower serum Mg concentrations were independently linked to greater risks of CKD and end-stage renal disease (ESRD) among 13,226 middle-aged participants [
10]. In a cohort study involving 1,650 Western European CKD cases, Van Laecke et al. reported that lower serum Mg concentrations were linked to a more rapid decline in kidney function [
11]. The HANDLS study found a higher risk of fast renal function deterioration among 1,252 African-American and Caucasian participants with reduced dietary Mg consumption [
12]. Although the above-mentioned research has revealed that serum Mg is negatively correlated with the occurrence and prognosis of CKD, these studies are limited to European and American populations. In addition, Azem et al. [
13] revealed that either hyper- or hypo-magnesemia was substantially linked to a growing risk of CKD progression among 10,568 patients with CKD stage 3 and 4. Ortega et al. [
14] found that serum Mg was not a reliable indicator of cardiovascular events or overall mortality among 70 non-dialysis patients with advanced CKD. Several studies have also indicated an association between Mg status and survival in ESRD patients [
15‐
17]. Therefore, until now, the relationships between Mg and the onset, development, and prognosis of CKD are limited, inconsistent, and not well established, particularly in the Chinese population.
In this context, we sought to explore the relationships between serum Mg status and CKD in a nationally representative sample of Chinese adults using data from the China Health and Nutrition Survey (CHNS).
Methods
Study population and design
The CHNS is an ongoing national longitudinal survey to investigate health and the nutritional status of Chinese residents [
18]. A multistage random clustering process was utilized to sample 288 communities in 12 provinces of China by 2011 (i.e., Beijing, Heilongjiang, Liaoning, Henan, Shandong, Shanghai, Jiangsu, Hubei, Guangxi, Hunan, Chongqing, and Guizhou) [
19], which were stratified by income using State Statistical Office definitions. All participants voluntarily signed informed consent and this study was supported by the institutional review board from the Chinese Center for Disease Control and Prevention and the University of North Carolina at Chapel Hill. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline in conducting and reporting the study [
20]. The inclusion criterion was participants who had blood sample collection in the wave of 2009, and the exclusion criterion was individuals who were aged < 18 years old and had missing information. Finally, a total of 8,277 participants from the CHNS 2009 were eligible for inclusion.
Laboratory measurements
Fasting blood samples were collected in local laboratories according to consensus guidelines with strict quality control, then transported to a national central lab in Beijing for future testing [
18]. Serum Mg and creatinine were analyzed using the Xylidyl blue colorimetric method (Randox, UK) and a standard Picric acid method (Randox, UK), respectively. The measured serum Mg concentrations were substituted with the mean concentration ± 3×standard deviation for those with measured concentrations higher/lower than this value [
21]. Serum high-density lipoprotein cholesterol (HDL-C) and total cholesterol (TC) were measured using the enzymatic method (Kyowa, Japan) and the cholesterol oxidase-phenol plus aminophenazone method (CHOD-PAP; Randox, UK), respectively. Serum triacylglycerol (TG) was detected using the glycerol phosphate oxidase phenol 4-aminoantipyrine peroxidase (GPO-PAP) method (Kyowa, Japan). Blood hemoglobin (Hb) was examined using the Coulter volume scatter conductivity (VCS) hematology analyzer (Beckman Coulter, USA).
Ascertainment of CKD
The primary outcome of interest in the current study, CKD, was defined by an eGFR < 60 mL/min/1.73 m
2 [
22,
23]. The eGFR value was calculated from age (in years), sex, and serum creatinine (in mg/dL), using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation [
24]. Moreover, the Modification of Diet in Renal Disease (MDRD) equation was also adopted in our sensitivity analysis [
25].
Assessment of covariates
Information on age, sex, residential place, education level, self-reported diagnosed fracture/diabetes/hypertension, and lifestyles (e.g., smoking, and drinking) were obtained from a face-to-face interview. We divide the subject’s educational level into 4 categories (no formal education, primary school, middle school, high school or higher) [
26,
27]. Individual diet was repeatedly evaluated by a household food inventory survey, and 24-h dietary recalls on three consecutive days were performed to assess individual diets [
28]. Dietary nutrient intakes (e.g., caloric, carbohydrate, and protein) were estimated based on the compositions in the China food composition tables (FCTs) and individual dietary consumption data [
29]. Height and weight were measured using the standard scheme by trained examiners. Body mass index (BMI) was calculated as weight (kg) divided by the square of height (m). BMI cut-off points for overweight and obesity were 24.0 kg/m
2 and 28.0 kg/m
2 [
30] for Chinese adults, respectively.
Statistical analysis
Participants were classified according to the quartiles of serum Mg concentrations (Q1: below 2.16 mg/dL, Q2: 2.16–2.28 mg/dL, Q3: 2.28–2.43 mg/dL, and Q4: above 2.43 mg/dL). Differences between serum Mg quartile groups in socio-demographic and clinical characteristics were evaluated by the ANOVA or Kruskal-Wallis test for continuous variables and the Chi-square test for categorical variables, respectively.
Based on previous studies [
31‐
33], to calculate the odds ratios (ORs) and 95% confidence intervals (CIs) of CKD based on quartiles of serum Mg concentrations, 3 multivariable logistic regression models were constructed. Model 1 was a crude model. In model 2, we made age (years), sex (male or female), BMI (< 18.5, 18.5–23.9, 24.0-27.9, or ≥ 28.0 kg/m
2), residential place (rural or urban), educational level (no formal education, primary school, middle school, high school or higher), smoking (yes or no), alcohol consumption (yes or no), self-reported diagnosed hypertension, diabetes, or fracture (yes or no), and biochemical indicators (i.e., TC, HDL-C, TG, and Hb) adjustments. In model 3, we further adjusted for dietary consumption (i.e., protein, carbohydrate, and calorie intake), because of the collinearity problem, we did not adjust the variable of fat intake. Tests for trends across Mg quartiles were assessed by including the median values within each group as a continuous variable in separate models [
34].
The dose-response relationship of serum Mg with CKD was tested by a 3-knots (i.e., the 5th, 50th, and 95th percentile) restricted cubic spline regression models. Stratified analyses were performed by age (≤ 60 and > 60 years), sex (male and female), BMI (< 18.5, 18.5–23.9, 24.0-27.9, and ≥ 28.0 kg/m2), residential place (rural and urban), educational level (no formal education, primary school, middle school, high school or higher), smoking status (no and yes), alcohol status (no and yes), self-reported diagnosed hypertension, diabetes, and fracture (no and yes), and the significance of interaction was evaluated by the likelihood ratio test.
In order to verify that our findings are reliable, the relationship between quartiles of serum Mg and eGFR calculated from CKD-EPI was analyzed using linear regression models. Besides, we utilized the modified MDRD equation to calculate eGFR, then repeated the analyses between quartiles of serum Mg and incident CKD and eGFR. All statistical analyses were conducted using Stata 15.0 (StataCorp, College Station, Texas) software, and P < 0.05 (two-sided) was regarded as statistically significant.
Discussion
A similar “J”-shaped associations between serum Mg concentrations and CKD were discovered in this relatively large-scale, nationwide study among the general Chinese population. The correlation was independent of traditional risk factors, such as lifestyle factors, biochemical indicators, BMI, and dietary consumption. A variety of stratified analyses and sensitivity analyses demonstrated the robustness of our findings.
The concentrations of serum Mg (median/mean: 0.95 mmol/L) in the present study were slightly higher than those described in NHANES I study in the U.S. (median: 0.85 mmol/L) [
35], and from the EPIC-Norfolk cohort study in the UK (males: 0.82 mmol/L; females: 0.80 mmol/L) [
36], while were slightly lower than that reported in the rural areas of the southwest of China (median: 1.07 mmol/L) [
37]. The difference in serum Mg levels among these regions may be attributed to the inclusion of age limit, dietary preference, drinking water, and race specificity [
35,
38,
39].
Mg is frequently active in numerous pathophysiological changes and molecular mechanisms in the human body. As a critical cofactor, Mg could be helpful to exert biological functions in any reaction powered by adenosine triphosphate (ATP); as a phosphate binder, Mg could be helpful to decrease hyperphosphatemia and vascular calcification [
40,
41]; as a calcium channel antagonist, Mg could involve in regulating various activity accompanied by intracellular calcium concentration fluxes, such as contraction of muscles, release of insulin [
42]. Thus, Mg disorder, probably owing to an excessive dietary intake of Mg, pathological problems (e.g., abnormal renal wasting), or excessive use of drugs (e.g., diuretics) [
43], would give rise to serious consequences or dramatic effects on metabolism. Consistent with this notion, our current study showed a similar “J”-shaped relationship between serum Mg levels and CKD among 8,277 Chinese adults, which indicated that either serum Mg deficiency or surplus might be linked to CKD. The risk of CKD development significantly decreased with the increment of serum Mg in participants when serum Mg < 2.2 mg/dL, but if serum Mg ≥ 2.2 mg/dL, the relationship between them developed in the opposite direction. Similar to our study, in 10,568 patients with CKD stage 3 and 4, Azem et al. [
13] reported that either hyper- or hypo-magnesemia was significantly associated with an increased risk of CKD progression. In a cohort study with 142,555 hemodialysis patients, Sakaguchi et al. [
44]. found that the risk of all-cause mortality was much higher in sextiles 1–3 and 6 of Mg concentration, which suggested that there may be an optimal concentration of Mg. Therefore, it could be speculated that the concentration of Mg should be controlled at an appropriate range, which needed to be performed and verified in future studies. In addition, we found that the association between serum Mg and CKD was not significant in older individuals who were above 60 years, perhaps since serum Mg concentrations generally decrease with age due to uneven diet, chronic diseases or side effects of medication, while the risk of developing CKD increased with age [
45],then this relationship was weakened [
46].
The relevant mechanisms underlying the similar “J”-shaped relationships between serum Mg concentrations and CKD remain incompletely clear, and several possibilities could be proposed to decipher our findings. Firstly, Mg may involve the prevention of nephrocalcinosis, inhibition of phosphate-mediated apoptosis of tubular cells and calcification of renal arteries, and suppression of tubular calcium phosphate crystallization, which may protect against phosphate-induced kidney damage. In addition, tubular dysfunction and interstitial fibrosis could contribute to the loss of Mg [
47,
48]. Moreover, matured calciprotein particles have the capacity to induce vascular calcification, and recent studies have shown that Mg could prevent the maturation of calciprotein particles [
49], and these may be the basics that underlie the anti-calcification properties of Mg. Secondly, Mg exercises the functions of protein metabolism and energy synthesis. Therefore, Mg deficiency may induce the development of CKD by reducing protein synthesis and energy metabolism of renal reparative cells. Thirdly, the concentrations of antioxidants (e.g., selenium and vitamin C) are decreased in the context of Mg deficiency [
50], which was closely related to promoting oxidative stress. In addition, related studies conducted in vitro on endothelial cells have shown that a low Mg medium promotes inflammation and oxidative stress, and induces the expression of proatherothrombotic molecules such plasminogen activator inhibitor-1 and vascular cell adhesion molecule-1 [
51], and Mg deficiency was found to be linked to endothelial dysfunction [
52], therefore, we reasonably speculate that Mg may have protective effects on endothelium, and Mg deficiency could be responsible for endothelial dysfunction and vascular sclerosis of renal vessels, accelerating the progression of kidney damage. Finally, previous studies also have shown that chronic inflammation is crucial to the development of CKD [
53,
54], thus abnormal Mg levels (deficiencies or excesses) may aggravate inflammation-related renal injury.
The large sample size and population-based design of our study were its major strengths, allowing us to perform a series of stratifications and the results were reliable and generalizable to the general population in China. However, our research has certain limitations as well. First, we were unable to establish the causal link between serum Mg and CKD because of the cross-sectional nature of this study. It is still needed to affirm whether the reduced renal function is a crucial risk element for hypermagnesemia or not, maybe the hypermagnesemia results in the faster development of CKD or the compensatory protective elevation of serum Mg for delaying the progression of CKD. Secondly, serum Mg concentrations were assessed at a single time point, thus the measured data may not be objective to reflect the whole life-course activity, and there is no more relevant available data on Mg intake in this database, so we cannot conduct subsequent causal analyses. Thirdly, due to a lack of data on the severity of CKD, we could not determine whether these associations differed by the severity of CKD; while the prevalence of CKD in our study is comparable to that estimated in global analysis (8.0–16.0%) [
55]. Finally, although we have considered several potential confounders, including various lifestyles and dietary variables, additional unmeasured confounders cannot be fully excluded.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.