Magnesium Intake, C-Reactive Protein, and the Prevalence of Metabolic Syndrome in Middle-Aged and Older U.S. Women

The WHS is a randomized, double-blind, placebo-controlled trial designed to evaluate the balance of benefits and risks of low-dose aspirin and vitamin E in the primary prevention of CVD and cancer (19). We randomized a total of 39,876 female health professionals ≥45 years of age who were free of coronary heart disease, stroke, and cancer (except for nonmelanoma skin cancer) at baseline (19). Of them, 98% provided detailed information on their diet by completing a 131-item semiquantitative food frequency questionnaire (SFFQ) in 1993 (20). We excluded from the analyses women with unreliable dietary data (≥70 items left blank in their questionnaire or energy intake <600 kcal or >3,500 kcal) and women with missing data for total magnesium intake. To minimize the effects of postmenopausal hormone therapy on systemic inflammation and the metabolic syndrome, we further excluded women who were current or ever postmenopausal hormone users, leaving 11,686 women for the analysis. Of these, 9,887 were free of diabetes at study entry and contributed complete data for all five components of the metabolic syndrome. The study protocol was approved by the Brigham and Women’s Hospital institutional review board, and the protocol adhered to the guidelines put forth in the Helsinki declaration and Belmont Accord for the duration of the study.

In the SFFQ, participants were asked how often on average they had consumed individual foods of a commonly used portion size during the previous year. Nine possible responses ranging from “never” to “six or more times per day” were recorded. Nutrient intakes were computed by multiplying the frequency of consumption of each unit of food from the SFFQ by the nutrient content of the specified portion size according to food composition tables from the Harvard Food Composition Database (21). Data on use of multivitamin supplements were taken into account to assess intake of supplemental magnesium. Total magnesium represents the sum of magnesium intake from both dietary and supplemental sources. Each nutrient was adjusted for total energy using the residual method (22). In populations of nurses and health professionals, this SFFQ has demonstrated reasonably good validity as a measure of long-term average dietary intakes (23). The Pearson correlation coefficient between magnesium intake assessed by SFFQ and 2 weeks of diet records was 0.76 (10).

The metabolic syndrome was defined according to the diagnostic criteria proposed by the Adult Treatment Program III of the National Cholesterol Education Program, as previously described (24). Women with three or more of the following conditions were typically defined as having the metabolic syndrome: 1) triglycerides ≥150 mg/dl, 2) HDL cholesterol <50 mg/dl, 3) blood pressure ≥135/85 mmHg, 4) obesity as defined by a waist circumference >88 cm (35 inch), and 5) abnormal glucose metabolism as defined by a fasting glucose ≥110 mg/dl. In the WHS, however, waist circumference was not reported until year 6 of follow-up. To assess the robustness of our definition for obesity using waist data, we repeated our analyses using a BMI cut point of 30 kg/m² or using a value of BMI that corresponded to the same percentile cut point for BMI as did a waist circumference of 88 cm. Because fasting glucose levels were not available, we instead used the diagnosis of incident type 2 diabetes during an average of 8.8 years follow-up as an alternative measure of baseline abnormal glucose metabolism. The validity of self-reported diabetes has been confirmed by a validation study, as previously reported (25).

Blood samples for the WHS were stored in liquid nitrogen and were thawed before assay. CRP was measured by a validated high-sensitivity assay (Denka Seiken, Niigata, Japan), as previously reported (26). Total cholesterol, HDL cholesterol, LDL cholesterol, and triglyceride levels were measured with direct measurement assays (Roche Diagnostics, Indianapolis, IN). This laboratory is certified for the measurement of lipids and is a core facility for ongoing standardization programs regarding the measurement of CRP. All samples were handled identically and analyzed in random order to reduce systemic bias and interassay variation. Blinded quality control specimens were analyzed simultaneously with the study sample. The average intraassay coefficient of variation (CV) for CRP was 7.8% (26), and the interassay CVs were 2.5−5.1% (27).

We first categorized total magnesium intake into quintiles and then calculated the median plasma concentrations of CRP according to each quintile of intake. We then calculated geometric means of plasma concentrations of CRP because it was not normally distributed. Geometric means were computed by regressing the natural logarithm of plasma CRP concentrations on magnesium intake and then taking an antilog of the resulting mean logarithmic value. Multiple linear regression models were used to control for potential confounding factors. The initial model was adjusted for age (continuous) and BMI (continuous). In multivariate models, we further adjusted for smoking status (current, past, and never), exercise (rarely/never, less than one time per week, one to three times per week, and four or more times per week), alcohol intake (rarely/never, one to three drinks per month, one to six drinks per week, and one or more drinks per day), total calorie intake (continuous), multivitamin use (never, past, and current), history of diabetes (yes/no), history of hypertension (yes/no), history of high cholesterol (yes/no), and parental history of myocardial infarction before 60 years of age (yes/no). The final multivariate model added dietary factors, including intakes of total fat, cholesterol, glycemic load, and folate (all categorized as quintiles). Tests of linear trend across increasing quintiles of intake were conducted by assigning the medians of intakes in quintiles treated as continuous variables. Furthermore, potential effect modifications were evaluated by subgroup analyses stratified by the prespecified factors, including BMI (< or ≥25 kg/m²), smoking status (never, past, or current), history of hypertension (yes/no), high cholesterol (yes/no), and diabetes (yes/no). To eliminate possible residual confounding, the number of cigarettes smoked per day (<15, ≥15 to <25, and ≥25) was adjusted in the subgroup analysis for current smokers, and BMI (continuous) remained in the models for both subgroups stratified by BMI (< or ≥25 kg/m²). Wald test was used to assess the significance of interaction terms.

We further sought evidence that magnesium intake might be associated with the components of the metabolic syndrome. We performed an analysis limited to the 9,887 study participants who had completed data on characteristics of the metabolic syndrome at study entry. First, we classified all participants as having none, one, two, three, four, or five components of the metabolic syndrome. We then compared the distribution of magnesium intake among individuals with or without each of the individual components of the syndrome as defined above. We also used logistic regression analysis to assess for evidence of association between magnesium intake and the metabolic syndrome. The odds ratios (OR) and 95% CIs for the prevalence of the metabolic syndrome were calculated. The logistic model adjusted for the same covariates as those in the CRP analyses, except for BMI, history of hypertension, high cholesterol, and diabetes, which were either used for defining the metabolic syndrome or highly correlated with the syndrome components.

The same analytic approach as above was used for analyses of dietary magnesium intake after excluding all magnesium supplements. All statistical analyses were conducted using SAS (version 8.0; SAS Institute, Cary, NC). All P values were two tailed (α = 0.05).

In the present study, dietary sources accounted for ∼96% of total intake of magnesium. The median intake of magnesium was 326 mg/day for our cohort of middle-aged women, which is close to the 10th Recommended Dietary Allowances of 320 mg/day for adult women (28). There was an ∼1.5-fold difference in total magnesium intake between the highest and lowest quintiles of the study population (median: 422 mg/day in the highest quintile vs. 252 mg/day in the lowest).

At baseline in 1993, women with high intake of magnesium were slightly older, were less likely to be current smokers or have a history of hypertension, and were more likely to exercise, use multivitamins, or have a history of high cholesterol or diabetes than women with low magnesium intake. High magnesium intake was also associated with a slightly lower BMI (Table 1). Women in the highest quintile of magnesium intake had a lower intake of total fat and cholesterol, but a higher intake of dietary carbohydrate, protein, fiber, and high–glycemic load foods.

Magnesium intake was inversely associated with plasma CRP concentrations (Table 2). In age- and BMI-adjusted analyses, the geometric mean CRP concentrations for increasing quintiles of total magnesium intake were 1.49, 1.37, 1.35, 1.34, and 1.33 mg/l (P for trend = 0.0004). After additional adjustment for smoking status, alcohol use, exercise, total calorie intake, multivitamin use, history of diabetes, hypertension, and high cholesterol, as well as parental history of myocardial infarction before 60 years of age, the mean CRP concentrations for ascending quintiles of magnesium intake were 1.46, 1.37, 1.35, 1.36, and 1.34 mg/l (P for trend = 0.02). Moreover, this association was more evident after additional adjustment for dietary factors, including total fat, cholesterol, glycemic load, and folate (P for trend = 0.0003). Although dietary fiber intake is highly correlated with magnesium intake (correlation coefficient = 0.62), further adjustment for fiber intake did not attenuate the association between magnesium intake and CRP concentrations. The inverse trend remained significant (the mean CRP for quintiles of total magnesium intake was 1.48, 1.38, 1.35, 1.34, and 1.33 mg/l; P for trend = 0.02).

As shown in Fig. 1, a significant inverse association between magnesium intake and plasma CRP concentrations was more apparent in overweight women with a BMI of ≥25 kg/m² (P for trend = 0.002) than those with a BMI ≤25 kg/m² (P for trend = 0.03). Likewise, this inverse association was pronounced among women who ever smoked (P for trend = 0.05 for current smokers and P for trend <0.0001 for past smokers) but not among those who never smoked (P for trend = 0.49). Tests of interaction were significant for smoking status (P = 0.0009) and overweight (P < 0.0001). In contrast, the inverse association between magnesium intake and CRP concentrations persisted across all subgroups stratified by history of hypertension or high cholesterol or diabetes without significant interactions (data not shown).

Table 3 displays the prevalence of individual metabolic syndrome components for 9,887 women free of diabetes at baseline according to quintiles of magnesium intake. The prevalence of each of the six components was lower in women in the highest quintile of magnesium intake than in women in the lowest quintile. All these associations reached statistical significance except for hypertriglyceridemia. When the waist circumference (>88 cm) was used for defining central obesity, the proportion of women with three or more characteristics of the metabolic syndrome was 24%. The prevalence of the metabolic syndrome was 7% lower in the highest relative to the lowest categories of magnesium intake (Table 3). Only 17% of participants were classified as having metabolic syndrome when a BMI cut point of 30 kg/m² was used as the obesity criterion.

We also observed significant decreases in prevalent risk of the metabolic syndrome across the quintiles of baseline magnesium intake in logistic models (Table 4). The multivariate-adjusted ORs of prevalent metabolic syndrome from the lowest quintiles to the highest quintiles of magnesium intake were 1.00 (referent), 0.91 (95% CI 0.78–1.06), 0.84 (0.72–0.99), 0.81 (0.68–0.96), and 0.73 (0.60–0.88) (P for trend = 0.0008). Further adjustment for dietary fiber or plasma CRP concentrations did not markedly change this inverse association. When obesity was defined using either a BMI cut point of 30 kg/m² or a value of BMI (≥25.69 kg/m²) that corresponded to the same percentile cut point for BMI as did a waist circumference of 88 cm, results were almost similar.

Additionally, results were not appreciably changed when we performed the same analyses restricted to dietary magnesium intake (without supplements). As magnesium intake from supplements alone contributed a small proportion of total magnesium intake (<4%), limited variation of supplemental magnesium intake did not allow us to have sufficient statistical power to perform a separate analysis of magnesium supplements.

In this large cohort of middle-aged and older U.S. female health professionals, we found a significant inverse association between magnesium intake and plasma concentrations of CRP. This inverse association appeared to be more pronounced among women who were overweight and those who ever smoked. We also found that magnesium intake was inversely associated with the prevalence of metabolic syndrome. These results suggest that a beneficial effect of magnesium intake on type 2 diabetes and CVD may be related to its roles in systemic inflammation and/or the development of the metabolic syndrome.

Although an inverse association between serum magnesium concentrations and CRP concentrations was observed in a cross-sectional study of 371 nondiabetic, nonhypertensive obese Mexicans (29), few studies have specifically examined the association between magnesium intake and systemic inflammation. As an acute-phase reactant reflecting low-grade inflammation, CRP levels have been recognized as an important risk factor for type 2 diabetes, hypertension, and CVD (18). Therefore, it is plausible that the beneficial effects of magnesium on these chronic diseases are partially mediated by improvement in low-grade inflammatory state.

Another finding of our study is that the potential effects of magnesium intake on CRP concentrations were greater among women who were overweight or ever-smokers. This result parallels our previous finding that magnesium intake is associated with lower fasting insulin concentrations among overweight women (5). Obesity and smoking status have been recognized as important predictors of high concentrations of plasma CRP. The extent to which magnesium intake influences insulin sensitivity or proinflammatory status may be magnified among overweight people or ever-smokers who are prone to chronic inflammation.

Growing evidence also suggests that CRP is a marker for the presence of metabolic syndrome (24,30–32). Metabolic syndrome comprises a constellation of metabolic abnormalities including visceral obesity, glucose intolerance, hypertension, and dyslipidemia (33,34). The evidence that magnesium favorably affects metabolic abnormalities such as insulin resistance, hypertension, and dyslipidemia, though not entirely consistent, has led us to hypothesize that magnesium intake is related to a lower risk of metabolic syndrome. Our findings provide some evidence for this hypothesis. This notion is also supported by a small cross-sectional study in which lower serum magnesium concentrations were strongly related to higher prevalences of dyslipidemia and hypertension, as well as the metabolic syndrome (35).

Our study strengths include the large size and the relatively homogeneous nature of our cohort, which reduced confounding by several variables, such as access to medical care, educational attainment, and socioeconomic status. The large number of serum samples assessed, detailed diet assessment, and adjustment for major risk factors all increase the validity of our results.

Nonetheless, several limitations in the present study merit consideration. First, the cross-sectional design precludes any causal inferences about the role of magnesium intake. Second, blood samples were assessed once only and dietary assessments were done using SFFQ. Measurement errors might have biased the results toward the null. Third, our evidence may be inadequate to support beneficial effects from magnesium independent of other highly correlated dietary nutrients, including fiber, calcium, and potassium. Nevertheless, consistency of our observed associations for magnesium may not be fully explained by other healthy lifestyle factors. More importantly, there is biological plausibility for the direct impact of magnesium intake on metabolic risk factors based on experimental evidence. Fourth, the lack of information on fasting glucose or insulin levels at baseline did not allow for a strict National Cholesterol Education Program definition for the metabolic syndrome. Because we used the diagnosis of incident diabetes during 8.8 years of follow-up as a surrogate for abnormal baseline glucose metabolism, misclassification of the metabolic syndrome may have biased our estimates. However, the findings of the present study were not altered after removing those with incident diabetes. Therefore, we believe it unlikely that this decision affected validity because magnesium intake was consistently related to several other metabolic components, including central obesity, dyslipidemia, and hypertension. Finally, because our study population included solely female health professionals who were mostly white, results from the present study may not be generalizable to those in the general U.S. population.

In conclusion, we found that magnesium intake was inversely associated with plasma concentrations of CRP and the prevalence of metabolic syndrome. These data support the potential benefits in primary prevention of type 2 diabetes, hypertension, and CVD by vegetables, whole grains, legumes, and nuts that are rich in magnesium, although future large clinical trials to confirm the efficacy of magnesium supplements are clearly warranted.

This study was supported by grant DK66401 (National Institute of Diabetes and Digestive and Kidney Diseases) from the National Institutes of Health.

Y.S. is supported by a scholarship from the Carson Family Scholarship Program in the Harvard School of Public Health.

We are indebted to the 39,876 dedicated and committed participants of the WHS. We acknowledge the contributions of the entire staff of the WHS.

P.M.R. is listed as a coinventor on patents held by the Brigham and Women’s Hospital that relate to the use of inflammatory markers in cardiovascular disease and diabetes.