Abstract
Objective. To establish whether there is a relationship between serum magnesium (Mg) concentration and radiographic knee osteoarthritis (OA).
Methods. There were 2855 subjects in this cross-sectional study. Serum Mg concentration was measured using the chemiluminescence method. Radiographic OA of the knee was defined as changes consistent with Kellgren-Lawrence (K-L) grade 2 on at least 1 side. Mg concentration was classified into 1 of 4 quartiles: ≤ 0.87, 0.88–0.91, 0.92–0.96, or ≥ 0.97 mmol/l. Multivariable logistic analysis was used to test the association between serum Mg and radiographic knee OA after adjustment for potentially confounding factors. The OR with 95% CI for the association between radiographic knee OA and serum Mg concentration were calculated for each quartile. The quartile with the lowest value was regarded as the reference category.
Results. Significant association between serum Mg concentration and radiographic knee OA was observed in the model after adjustment for age, sex, and body mass index, as well as in the multivariable model. The multivariable-adjusted OR (95% CI) for radiographic knee OA in the second, third, and fourth serum Mg concentration quartiles were 0.90 (95% CI 0.71–1.13), 0.92 (95% CI 0.73–1.16), and 0.72 (95% CI 0.57–0.92), respectively, compared with the lowest (first) quartile. A clear trend (p for trend was 0.01) was observed. The relative odds of radiographic knee OA was decreased by 0.72 times in the fourth serum Mg quartile compared with the lowest quartile.
Conclusion. Serum Mg concentration may have an inverse relationship with radiographic OA of the knee.
Osteoarthritis (OA) is a progressive rheumatic disease, the incidence of which increases with age in most societies. In the United States, about 28% of people aged over 45 years are reported to have radiographic OA of the knee, but the proportion is slightly higher in African Americans1. A high prevalence of OA has also been observed in some Asian countries2,3,4,5,6,7. For example, about 20% of men and 36% of women were affected by radiographic OA of the knee among rural (Wuchuan) Chinese subjects aged between 59 and 84 years7. At present, the only effective therapy for late-stage OA is joint replacement. There is increasing interest in identifying effective conservative treatments, and high priority has been given to prevention. The importance of nutrition in the maintenance of joint health is now widely recognized, but more research is needed to establish whether nutritional interventions can be made to prevent the onset or progression of OA8.
It has recently been reported that the serum concentration of high-sensitivity C-reactive protein (hsCRP), a sensitive biomarker of low-grade systemic inflammation, correlated with radiological findings in the knee joint9. Lower serum concentration of magnesium (Mg), 1 of the most important micronutrients for human health, has also been shown to be independently associated with elevated hsCRP concentration10. Further, there is a strong relationship between Mg and immune responses11 that may be involved in the pathogenesis of OA12. It is plausible, therefore, that Mg deficiency is associated with OA, but there have been few studies that have examined whether serum Mg concentration is correlated with OA.
Hunter, et al13 reported that in a cohort of 229 female twin pairs, serum Mg concentration was significantly lower in co-twins with OA after adjustment for body mass index (BMI), but the authors acknowledged the potential limitation of the generalizability of twin studies. Qin, et al14 found that dietary Mg intake was inversely associated with radiographic OA of the knee in whites, but the association between serum Mg concentration and radiographic OA of the knee is not predictable. Nevertheless, both studies provide sufficient evidence for hypothesis generation. We designed a cross-sectional study to illuminate whether there is an inverse relationship between serum Mg concentration and radiographic OA of the knee.
MATERIALS AND METHODS
Study population
This cross-sectional study enrolled Chinese members of the general public who were undergoing health screening to explore the association between nutrition and the disease; the study design has been published15. Such screening checkups have become routine in China, and generally include anthropometric and basic clinical assessment (weight, height, blood pressure, and waistline measurements, etc.), biochemical (blood routine examination, hepatic function, renal function, trace elements test, etc.), and imaging tests (chest radiography, bilateral anteroposterior knee radiography, etc.). Registered nurses collected details of demographic characteristics and health-related habits, such as age, occupation, educational level, physical activity level, smoking status, alcohol drinking status, and diet using a standard questionnaire. Subjects were selected for this study according to the following inclusion criteria: (1) age ≥ 40 years; (2) availability of weight-bearing anteroposterior radiographs of both knees; (3) availability of blood biochemistry including serum Mg and fasting glucose concentrations; and (4) availability of data on all basic characteristics, including age, sex, BMI, smoking status, etc. Initially, this cross-sectional study recruited 5486 participants who underwent routine examinations including weight-bearing bilateral anteroposterior radiography of the knee and blood biochemical tests, such as blood glucose and serum magnesium concentration, at the Department of Health Examination Center of Xiangya Hospital, Central South University in Changsha, Hunan Province, China, from October 2013 to August 2014. Individuals with the following characteristics were excluded: radiographic evidence of non-OA joint disease such as osteochondroma or fracture (n = 67); missing data on certain characteristics or physical examinations, such as BMI, waistline, and blood pressure (n = 2); younger than 40 years (n = 539); or missing data on the records of behavior habits, such as smoking status, alcohol use, and physical activity level (n = 2023). A total of 2855 subjects were included in the final analysis. This research was approved by the ethics committee of Xiangya Hospital of Central South University. All participants gave written informed consent at the time of recruitment.
Blood biochemistry
All blood samples were drawn after a 12-h overnight fast and were stored at 4°C until analysis. The serum Mg concentration was measured using the chemiluminescence method. A weak positive association between serum Mg concentration and dietary Mg intake was observed in the study population (r = 0.07, p = 0.00). The fasting plasma glucose concentration was measured using the glucose oxidase enzyme method. Laboratory tests were undertaken using a Beckman Coulter AU 5800 (Beckman Coulter Inc.). Subjects with a fasting glucose ≥ 0.0 mmol/l or who were currently undergoing drug treatment for blood glucose control were regarded as having diabetes mellitus. The inter- and intraassay coefficients of variation were tested by low concentrations (2.5 mmol/l for glucose, 118 µmol/l for uric acid, and 0.60 mmol/l for serum Mg) and high concentrations (6.7 mmol/l for glucose, 472 µmol/l for uric acid, and 1.00 mmol/l for serum Mg) of standard human samples. The intraassay coefficients of variation were 0.98% (2.5 mmol/l) and 1.72% (6.7 mmol/l) for glucose, 1.39% (118 µmol/l) and 0.41% (472 µmol/l) for uric acid, and 1.86% (0.60 mmol/l) and 1.65% (1.00 mmol/l) for serum Mg. The interassay coefficients of variation were 2.45% (2.5 mmol/l) and 1.46% (6.7 mmol/l) for glucose, 1.40% (118 µmol/l) and 1.23% (472 µmol/l) for uric acid, and 1.87% (0.60 mmol/l) and 1.70% (1.00 mmol/l) for serum Mg.
Assessment of radiographic OA of the knee
All subjects were undergoing weight-bearing bilateral anteroposterior radiography of the knee. Two orthopedic surgeons, blinded to subjects’ clinical symptoms, assessed all radiographs independently using the Kellgren-Lawrence (K-L) radiographic atlas. Inconsistencies were resolved through discussion. OA was divided into 5 categories according to K-L grade: 0 = absence of OA, 1 = suspected OA, 2 = minimal OA, 3 = moderate OA, and 4 = severe joint OA16. A subject was diagnosed with radiographic OA of the knee if at least 1 knee joint was graded as K-L 2 or higher. The reliability of the 2 orthopedic surgeons’ assessments was judged at 2 different times using an ICC. The results indicated that both interrater and intrarater reliability were high (κ = 0.86 and 0.87, respectively).
Assessment of covariates
Blood pressure was measured using an electronic sphygmomanometer. Subjects with a systolic blood pressure ≥ 140 mmHg, diastolic blood pressure ≥ 90 mmHg, or who were currently using anti-hypertensive medication were regarded as having arterial hypertension. The weight and height of each subject were measured to calculate BMI. Participants were also asked about their average frequency of physical activity (never, 1 or 2 times per week, 3 to 4 times per week, 5 times or above per week) and average duration of physical activity (30 min, 30–60 min, 60–120 min, > 120 min). The current smoking status and alcohol drinking status (yes or no for each) and the use of nutritional supplements were established by direct face-to-face questioning. Participants taking any longterm nutritional product, such as calcium, vitamin, or mineral supplements, were considered nutritional supplementation users.
Statistical analysis
Quantitative data are expressed as the mean ± SD, and qualitative data are expressed as proportion (percentage). The serum Mg concentration was classified into 4 quartiles: ≤ 0.87, 0.88–0.91, 0.92–0.96, and ≥ 0.97 mmol/l. Differences in continuous data were evaluated by the 1-way classification ANOVA (normally distributed data) or the Kruskal-Wallis H test (non-normally distributed data), while differences in qualitative data were assessed by the chi-square test. We also established whether there were differences in age, BMI, and the proportion of men and women between participants and subjects who were excluded from this study using Mann-Whitney U test and chi-square test as appropriate.
OR with 95% CI for the association between radiographic knee OA and serum magnesium concentration were calculated for each Mg concentration quartile; the quartile with the lowest value was regarded as the reference category. To calculate the adjusted OR of each quartile of serum Mg concentration, a multivariable model was adopted for logistic analysis that included age (yrs), BMI (< 30, ≥ 30 kg/m2), sex (male, female), educational achievement (high school or above, lower than high school), smoking status (yes/no), physical activity level (continuous data), alcohol drinking status (yes/no), nutrient supplementation (yes/no), longterm use of aspirin (yes/no), diabetes mellitus (yes/no), and hypertension (yes/no). Tests for linear trends were conducted based on logistic regression using a median variable of Mg concentration in each category.
The adjusted OR of each 0.01 mmol/l increase in serum Mg concentration based on this multivariable model was also calculated. Interaction between serum Mg concentration and potential confounders (age, BMI, sex, educational achievement, smoking status, physical activity level, alcohol drinking status, nutrient supplementation, longterm use of aspirin, diabetes mellitus, and hypertension) were tested one by one. Then, subgroup analyses were conducted to assess whether the association between serum Mg concentration and radiographic knee OA was modified by the interaction terms that were statistically significant. All analyses were performed using SPSS 17.0; a p value < 0.05 was considered statistically significant.
RESULTS
The characteristics of the study population (2855 subjects) based on quartiles of serum Mg concentration are shown in Table 1, and the anthropometric and biochemical characteristics of the population according to OA status in Table 2. Significant differences were observed across all quartiles of serum Mg concentration in terms of female ratio (45.8%, 43.7%, 42.2%, and 40.1% for the first, second, third, and fourth quartiles, respectively), high school diploma ratio (42.5%, 51.3%, 49.2%, and 48.8% for the first, second, third, and fourth quartiles, respectively), and the ratio of diabetes (16.3%, 8.8%, 7.9%, and 6.8% for the first, second, third, and fourth quartiles, respectively). The prevalence of radiographic OA of the knee in the present cross-sectional study (age ≥ 40 yrs, with an average age of 52.26 ± 7.16 yrs) was 30.0%. There was no significant difference in terms of age (p = 0.12), sex (p = 0.14), or BMI (p = 0.96) between those included in the main analysis and those excluded as a result of missing health-related behavior data (n = 2023).
Characteristics among 2855 participants according to quartiles of serum Mg. Values are mean (SD) unless otherwise indicated.
Characteristics among 2855 participants according to OA status. Values are mean (SD) unless otherwise indicated.
A significant association between serum Mg concentration and radiographic OA of the knee was observed in a model after adjustment for age, sex, and BMI, and also in a multivariable model adjusted for age, sex, BMI, educational achievement level, smoking status, alcohol drinking status, physical activity level, nutritional supplements, use of aspirin, serum uric acid concentration, diabetes mellitus, and hypertension (Table 3). The OR (95% CI) for radiographic knee OA adjusted by age, sex, and BMI from the first to the fourth quartiles of serum Mg concentration were 1 (reference), 0.93 (95% CI 0.74–1.17), 0.95 (95% CI 0.76–1.19), and 0.75 (95% CI 0.59–0.95), respectively. A clear trend was observed (p for trend was 0.03). The relative odds of radiographic OA of the knee were decreased by 0.75 times in the fourth quartile of serum Mg concentration compared with those in the lowest, reference quartile. The multivariable-adjusted OR (95% CI) for radiographic OA of the knee from the first to the fourth serum Mg concentration quartiles were 1 (reference), 0.90 (95% CI 0.71–1.13), 0.92 (95% CI 0.73–1.16), and 0.72 (95% CI 0.57–0.92), respectively, and a clear trend was observed (p for trend was 0.01). The relative odds of radiographic OA of the knee were decreased by 0.72 times in the fourth quartile of serum Mg concentration compared with the lowest quartile. The adjusted OR of each 0.01 mmol/l increase in serum Mg concentration based on the multivariable model was 0.99 (95% CI 0.98–1.00, p = 0.04).
Multivariable-adjusted associations of serum Mg and radiographic knee OA (n = 2855). Values are adjusted OR (95% CI) unless otherwise indicated.
BMI (< 30, ≥ 30 kg/m2) and alcohol drinking status (yes/no) were found to be interacted with serum Mg. Table 4 shows the multivariable-adjusted associations identified in subgroup analyses stratified BMI (< 30, ≥ 30 kg/m2) and alcohol drinking status (yes/no); a clear trend was observed for an elevated serum Mg concentration in nonobese (BMI < 30 kg/m2, p for trend was 0.01) and non-drinkers (p for trend was 0.00). Sensitivity analysis including subjects who did not report their health-related behaviors (4878 participants) did not alter the nature of the associations detected. The OR (95% CI) for radiographic OA of the knee adjusted by age, sex, and BMI from the first to the fourth quartiles of serum Mg concentration were 1, 0.98 (95% CI 0.83–1.16), 0.91 (95% CI 0.77–1.09), and 0.79 (95% CI 0.66–0.95), respectively, and a clear trend was observed (p for trend was 0.00). The relative odds of radiographic OA of the knee were decreased by 0.79 times in the fourth quartile of serum Mg concentration compared with the lowest quartile.
Subgroup analysis of multivariable-adjusted associations of serum Mg and radiographic knee OA (n = 2855). Values are adjusted OR (95% CI) unless otherwise indicated.
DISCUSSION
This cross-sectional study identified an inverse relationship between serum Mg concentration and radiographic OA of the knee that was independent of some major confounding factors. To the best of our knowledge, ours is the first study to have examined the association between serum Mg concentration and radiographic OA of the knee in a large sample. The extent of the association that we found was nonetheless consistent with previous research13,14,17. Hunter, et al13 observed a significant decrease in serum Mg level in the co-twins with OA from 229 female white twin pairs aged from 24 to 79 years. The BMI-adjusted OR was 0.91 (95% CI 0.46–0.92). Qin, et al14 reported that the relative odds of radiographic knee OA were decreased by 0.52 times in the second quintile and 0.60 in the fourth quintile of Mg intake compared with those in the lowest quintile in 1447 whites aged over 45 years, but this significant difference did not exist in African Americans.
The potential benefits of Mg on chondrocyte have been investigated in vitro and in animal models18,19,20,21. Addition of Mg to culture medium reportedly reduces the toxicity of local anesthetics to chondrocytes18. Mg has also been proposed as a useful tool in cartilage tissue engineering because it reportedly enhances chondrocyte proliferation and redifferentiation19, as well as protecting against a substantial proportion of quinolone-induced chondrocyte damage in vitro20. Further, distal femur articular cartilage chondrocyte density is significantly reduced in some animal experiments if dietary Mg is restricted21. Our findings suggest that the relative odds of radiographic OA of the knee were decreased by 0.75 times in the fourth quartile (1.00 mmol/l) when compared with the lowest quartile (0.84 mmol/l). The mean dietary Mg intake in the fourth quartile was 381.42 mg/day, which slightly exceeds the recommended daily allowance of Mg (350 mg/day) in China. This suggests that sufficient intake of Mg-rich foods, such as whole grains, legumes, leafy green vegetable, and nuts, could be beneficial for patients with OA, although this would require further research.
Several cross-sectional studies have reported an association between dietary Mg intake and serum CRP concentration22,23,24,25,26,27,28, including a recent metaanalysis29. In addition, oral Mg supplementation has been shown to reduce CRP concentration in some randomized controlled trials30,31, and hsCRP was recently reported to correlate with radiological findings in the knee joint9. Although it is well recognized that inflammation plays a key role in the pathophysiology of OA32, there is still debate about the relationship between Mg supplementation and blood CRP concentration, and the association between CRP and the incidence of OA33,34,35,36,37. Our findings do not allow us to conclude that low serum Mg concentration increases the risk of OA because of its influence on systemic inflammation. Nevertheless, given the potential therapeutic effect of Mg supplementation and the prevalence and disease burden of OA, further prospective studies are warranted.
Our present study has several strengths. It is the first, to our knowledge, conducted on a large sample (2855 subjects) that aimed to directly correlate serum Mg concentration with OA. Second, the multivariable model was adjusted for a considerable number of potentially confounding factors, a step that greatly improves the reliability of the results. Third, the serum Mg concentration is a better indicator of the compartmentalization or renal handling of total body Mg than dietary Mg intake in normal diets38.
Limitations of our present study should also be noted. Its cross-sectional design precludes causal correlations, and further prospective research and interventional trials will be needed to establish the causal or effect association between serum Mg concentration and OA of the knee. Because no previous study that we know of examined the association between serum Mg and knee OA in a large sample, the value of this research should not be blotted out by the cross-sectional nature. Another limitation is that a single measurement of serum Mg concentration may not fully reflect the longterm Mg status39, and we cannot comment on likely serum concentration before OA developed, particularly because drugs may have influenced the concentration. Third, serum Mg concentration may not reflect the concentration in synovial fluid. Finally, a weak positive association between serum Mg concentration and dietary Mg intake was observed in the study population. However, this result was consistent with previous studies38,40,41,42,43,44. Thus, the serum Mg level may not be a good biomarker of dietary Mg intake.
Serum Mg concentration may have an inverse relationship with radiographic OA of the knee. High level of Mg could possibly exert a protective role in the control of radiographic knee OA.
Footnotes
Supported by the Hunan Provincial Innovation Foundation For Postgraduate (CX2014A005), the Fundamental Research Funds for the Central Universities of Central South University, the National Natural Science Foundation of China (NO. 81201420, 81272034, 81472130), the Provincial Science Foundation of Hunan (NO. 14JJ3032), the Scientific Research Project of the Development and Reform Commission of Hunan Province ([2013]1199), the Scientific Research Project of Science and Technology Office of Hunan Province (2013SK2018), the Doctoral Scientific Fund Project of the Ministry of Education of China (20120162110036).
- Accepted for publication March 24, 2015.
REFERENCES
In this issue
