Calcium/Vitamin D Supplementation and Cardiovascular Events

Vascular calcification and valvular calcification predict atherosclerotic risk^1,2 and are prevalent in chronic diseases such as diabetes,³ systemic lupus erythematosus,⁴ and chronic kidney disease,⁵ in which the risk of coronary events is high. Arterial calcification and valvular calcification are organized, regulated processes similar in many respects to bone formation and remodeling.^6–9 Because of this relationship, bisphosphonates have been proposed as antiatherosclerotic agents,¹⁰ and patients with coronary calcification commonly ask if they should reduce their calcium consumption.¹¹ The literature on this topic is scant and conflicting^12–16; even less is known about the relationship between vitamin D and coronary risk.¹⁷

We randomized 36 282 postmenopausal women to calcium plus vitamin D or to placebo in a fracture trial and report here the impact of 7 years of supplementation on cardiovascular outcomes, including time trends and analyses of subgroups.

Methods

Details of the study design have been published previously,¹⁸ as have the baseline characteristics.^19–21 Eligible postmenopausal women 50 to 79 years of age joined the Women’s Health Initiative hormone therapy and/or dietary modification trials between 1993 and 1998. One year later, they were invited to join the double-blinded calcium plus vitamin D trial; 91% joined the calcium/vitamin D trial at their first annual visit and 9% during the following year. Women provided written informed consent in a form approved by local institutional review boards and were randomly assigned to a calcium and vitamin D supplement (containing calcium carbonate, 500 mg as elemental calcium, with vitamin D₃ 200 IU twice daily) (GlaxoSmithKline Consumer Healthcare, Parsippany, NJ) or matching placebo. Concurrent calcium supplementation was permitted, as was vitamin D, up to 400 IU daily.

Clinical Outcomes

Weight, blood pressure, and waist circumference were recorded annually. Blood samples were collected at baseline, ie, at the time of enrollment into the hormone therapy and/or dietary modification trials, from all participants; in a random 6% sample, blood samples also were collected 1 and 3 years later.²²

Participants reported emergency room visits, overnight hospital stays, and outpatient coronary revascularization procedures semiannually. Medical records for all overnight hospitalizations and outpatient coronary revascularization procedures were scrutinized for potential outcomes of interest. Centrally trained physician-adjudicators classified outcomes on the basis of medical record review. Myocardial infarction was categorized through the use of an algorithm that included symptoms, ECG findings, and cardiac enzymes.²³ Confirmed angina required hospitalization for angina with confirmatory stress test or obstructive coronary disease by angiography.²⁴ Stroke required rapid onset of a persistent neurological deficit not due to trauma, tumor, infection, or other cause²⁵; strokes were coded as “other” if procedure related or if the adjudicator could not classify the event as hemorrhagic or ischemic. All deaths were centrally adjudicated; other outcomes were adjudicated on the basis of hospital record review by centrally trained, local adjudicators blinded to treatment assignment. Composite outcomes were defined during development of the analytical plan.

Statistical Methods

Statistical methods have been described.^20,21 In brief, hazard ratios with 95% confidence intervals (CIs) were calculated from Cox proportional-hazards models stratified by age, prevalent cardiovascular disease at baseline, and randomization status in the hormone and dietary modification trials. Subgroup analyses were planned a priori. Subgroup analyses were stratified by age, prevalent cardiovascular disease at baseline, and randomization status in the hormone and dietary modification trials. Consistency of treatment effects among subgroups was assessed by formal tests of interaction; tests for linear trend were used when appropriate. Nineteen subgroups were evaluated for coronary heart disease (CHD) and for stroke; subgroup results should be interpreted with caution because 1 significant finding would be expected by chance for each outcome based on a 0.05 nominal level of statistical significance. All reported probability values are 2 sided. Analyses were carried out by the coordinating center statistics unit using the SAS System for Windows, version 9 (SAS Institute, Cary, NC).

The authors had full access to and take responsibility for the integrity of the data. All authors have read and agree to the manuscript as written.

Results

Between 1995 and 2000, 36 282 women were randomized at 40 clinical sites; when the trial closed in April 2005, the mean duration of follow-up was 7.0±1.4 years. Baseline characteristics were balanced between treatment groups (Table 1) except for hypertension (P=0.03). At baseline, mean calcium intake (diet plus supplements, exclusive of study medication) was 1148±654 mg/d in the active treatment group and 1154±658 mg/d in the placebo group, close to the recommended intake of 1200 mg daily.²⁶ Vitamin D consumption was 365±265 IU/d in the active treatment group and 368±266 IU/d in the placebo group. Sixty percent of study participants took at least 80% of their study medication through year 6.

Intermediate Biomarkers and Risk Factors for CHD

Although blood samples were collected on the entire cohort, bioassays were performed in only a 6% random sample. At baseline, total cholesterol was 5.64 mmol/L, low-density lipoprotein cholesterol was 3.28 mmol/L, high-density lipoprotein cholesterol was 1.54 mmol/L, triglycerides were 1.81 mmol/L, glucose was 5.49 mmol/L, and insulin was 11.4 μIU/mL. Differences between mean percent change in the intervention group and mean percent change in the control group are shown from baseline to year 2 after randomization (Figure 1). Percent change from baseline differed significantly between treatment groups for low-density lipoprotein cholesterol (P=0.02), waist circumference and weight (P=0.03 for both), systolic blood pressure (P=0.01), and diastolic blood pressure (P<0.01).

• Download figure
• Download PowerPoint

Clinical Cardiovascular Outcomes

Myocardial infarction or CHD death was confirmed in 499 women assigned to active calcium/vitamin D and 475 assigned to placebo (hazard ratio, 1.04; 95% CI, 0.92 to 1.18). Stroke was confirmed in 362 women assigned to calcium/vitamin D and 377 assigned to placebo (hazard ratio, 0.95; 95% CI, 0.82 to 1.10; Figure 2). Among women taking at least 80% of study medication, the hazard ratio for myocardial infarction/CHD death was 1.05 (95% CI, 0.88 to 1.25) and for stroke was 0.97 (95% CI, 0.79 to 1.20) (data not shown). Risks of coronary revascularization, confirmed angina, hospitalized heart failure, transient ischemic attack, and composite outcomes also were similar in the 2 treatment groups (Table 2).

• Download figure
• Download PowerPoint

Temporal Trends

Hazard ratios with nominal 95% CIs for myocardial infarction/CHD death at 1-year intervals of follow-up were as follows: year 1, 1.13 (95% CI, 0.79 to 1.61); year 2, 1.10 (95% CI, 0.75 to 1.61); year 3, 1.00 (95% CI, 0.69 to 1.46); year 4, 0.92 (95% CI, 0.66 to 1.28); year 5, 1.00 (95% CI, 0.72 to 1.39), year 6, 1.11 (95% CI, 0.81 to 1.51), and year ≥7, 1.07 (95% CI, 0.80 to 1.42). The z score for trend, based on Cox proportional-hazards modeling with time-dependent treatment effects, was 0.22 (P=0.82), indicating no significant trend in risk over time.

Hazard ratios with 95% CIs for stroke were as follows: year 1, 1.09 (95% CI, 0.69 to 1.72); year 2, 0.62 (95% CI, 0.41 to 0.94); year 3, 1.22 (95% CI, 0.82 to 1.82); year 4, 1.07 (95% CI, 0.70 to 1.65); year 5, 1.01 (95% CI, 0.69 to 1.47), year 6, 0.71(95% CI, 0.50 to 1.01), and year ≥7, 1.11 (95% CI, 0.81 to 1.52). The z score was 0.55 (P=0.58).

Trends by Age

Cumulative hazard ratios for myocardial infarction/CHD death and for stroke were evaluated by age decade. For women 50 to 59, 60 to 69, and 70 to 79 years of age at baseline, hazard ratios with 95% CIs for CHD were 0.94 (95% CI, 0.70 to 1.27), 1.08 (95% CI, 0.90 to 1.30), and 1.05 (95% CI, 0.85 to 1.30), respectively (P=0.53 for interaction). Hazard ratios with 95% CIs for stroke were 0.90 (95% CI, 0.62 to 1.32), 0.97 (95% CI, 0.78 to 1.20), and 0.96 (95% CI, 0.76 to 1.20), respectively (P=0.72 for interaction).

Additional Subgroup Analyses

We evaluated several demographic and clinical characteristics to determine whether other subgroups of women were at lower or higher risk for myocardial infarction/CHD death with calcium/vitamin D (Figure 3) or for stroke (Figure 4). The hazard ratios for CHD (P=0.91 for interaction) and stroke (P=0.14 for interaction) did not differ by total calcium intake (dietary plus supplemental) at baseline. Similarly, hazard ratios did not differ by vitamin D intake at baseline (P=0.45 for interaction for CHD and P=0.12 for stroke). Hazard ratios also did not differ by ethnicity, although numbers of events were small among Hispanic, American Indian, and Asian women (P=0.54 for interaction for CHD and P=0.63 for stroke).

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

CHD risk with active calcium/vitamin D was inversely related to body mass index (P=0.04 for interaction); ie, women with higher body mass index were at lower CHD risk with active calcium/vitamin D supplementation, whereas those with lower body mass index were at higher CHD risk. Stroke risk with active calcium/vitamin D was lower among women with high cholesterol and those taking statins at baseline (P=0.04 for interaction for both). Stroke risk with active calcium/vitamin D was inversely related to the number of CHD risk factors; ie, women with fewer risk factors were at higher stroke risk with calcium/vitamin D supplementation (P=0.02 for interaction).

Discussion

Calcium/vitamin D supplementation neither increased nor decreased the risk for CHD or stroke in generally healthy postmenopausal women throughout the 7-year duration of this randomized trial. Neither total calcium intake (dietary plus supplemental) nor total vitamin D intake at baseline affected cardiovascular risk with calcium/vitamin D supplementation.

Possible explanations of this null finding include the following: (1) Background calcium use impaired our ability to identify a treatment effect; (2) the dose of vitamin D was inadequate; (3) poor adherence to study medication blunted any treatment effect; (4) concurrent postmenopausal hormone therapy interfered with treatment effects; (5) the trial was designed to evaluate the effects of calcium/vitamin D supplementation on fracture, not cardiovascular disease; or (6) calcium and vitamin D do not, in fact, affect cardiovascular risk.

A limitation of the trial was that women were allowed to continue their own calcium supplements because it would have been unethical to prohibit concurrent calcium use in a long-term, placebo-controlled trial. Baseline calcium consumption (diet plus supplements) was balanced between treatment groups, and no significant interaction between dietary or total calcium consumption at baseline and randomized treatment assignment was observed for either CHD or stroke.

Baseline vitamin D consumption (diet plus supplements) and regional solar irradiance²¹ also were balanced between treatment groups. Parathyroid hormone levels are maximally suppressed at 25-hydroxy vitamin D blood levels >75 nmol/L (30 ng/mL).²⁷ In our trial, despite consuming 365 IU vitamin D daily (supplements plus dietary vitamin D) at baseline, only 13% of women with incident fractures (n=1589) and 15% of matched controls had serum levels >75 nmol/L, consistent with the current view that 800 to 1000 IU daily may be needed to achieve optimal serum vitamin D levels.²⁸ Women assigned to active calcium/vitamin D supplementation would have been taking almost 800 IU daily, which may still have been insufficient. Nonetheless, women with higher vitamin D consumption at baseline were not at higher or lower risk for CHD or stroke if assigned to active calcium/vitamin D. Low vitamin D levels have been associated with acute stroke²⁹; because we measured serum vitamin D levels in fracture cases and controls, not stroke cases, we are not able to confirm this association.

At the end of the trial (mean follow-up, 7 years), 76% of participants were taking some study pills, and 59% were taking ≥80% of their study medication. Calcium/vitamin D supplementation did not alter CHD or stroke risk in sensitivity analyses, in which women were censored when they became nonadherent, reducing the likelihood that adherence affected study results. Use of postmenopausal hormone therapy, which increases the risk of stroke²⁵ and CHD,³⁰ was balanced in the treatment groups. Neither CHD nor stroke risk differed with calcium/vitamin D supplementation among women assigned to active hormone therapy in the randomized hormone trials, making it unlikely that postmenopausal hormone use affected study results.

Another limitation of our analysis is that this trial was designed to evaluate the effect of intervention on fracture, not cardiovascular disease. In fact, the number of myocardial infarctions/CHD deaths (n=974) and strokes (n=739) was greater than the number of hip fractures (n=374), so a reasonable treatment effect should have been readily detectable. Overall, the most likely explanation for our findings is that calcium/vitamin D supplementation did not modulate CHD or stroke risk.

Calcium and vitamin D had a mixture of favorable and unfavorable effects on intermediate outcomes. Systolic pressure rose 1.1±12.4% among calcium/vitamin D recipients during the 2 years after randomization and 0.7±12.4% among placebo recipients (P=0.01 for the treatment group difference in percent change from baseline to year 2). Diastolic pressure fell 0.2±12.4% in the active treatment group and 0.6±12.4% in the placebo group (P=0.007). These findings contrast with the National Health and Nutrition Examination Survey, in which dietary calcium consumption was inversely associated with an age-related increase in systolic blood pressure.³¹ Because of adjustments in concurrent medication dosage, we cannot be sure that the treatment group differences in our trial are due solely to calcium/vitamin D supplementation. Furthermore, in later years of the trial, the changes in blood pressure from baseline no longer differed between treatment groups.

Weight increased in both treatment groups during the 2 years after randomization (1.4±10.5% versus 1.7±12.0%), as did waist circumference (1.5±7.6% versus 1.8±8.4%), but these increases were smaller among women assigned to active calcium/vitamin D (P=0.03 for both). The relationship between weight and calcium/vitamin D consumption in other reports has been inconsistent,^32,33 but women taking calcium supplements gained less weight than nonusers in a recent, large, 10-year epidemiological study.³⁴

In both treatment groups, low-density lipoprotein cholesterol rose in the 6% of participants with measured biomarkers, but the increase was smaller among women assigned to active calcium/vitamin D (0.2±20.9% versus 2.6±20.7%; P=0.02). In a small 12-week trial, calcium/vitamin D supplementation had no effect on low-density lipoprotein cholesterol.³⁵ The modest difference seen in our trial may reflect changes in weight over the 2 years or adjustments in concomitant medications and cannot be definitively attributed to calcium/vitamin D supplementation.

We found several subgroups of women with lower hazard ratios for CHD or stroke with calcium/vitamin D supplementation. Women with higher body mass index appeared to be at lower risk for CHD with calcium/vitamin D supplementation (P=0.04 for interaction), but in view of the number of subgroups examined, this finding may be due to chance.

We also found that women with more coronary risk factors were at lower risk for stroke with calcium/vitamin D supplementation (P=0.02 for interaction) but think this may be due to chance as well, particularly because the number of women with ≥3 risk factors was very small. On the other hand, women with self-reported hypercholesterolemia and those who used statins were at lower risk for stroke if assigned to active calcium/vitamin D (P=0.04 for interaction for both). The clinical link between high cholesterol and statin use and the proposed, albeit controversial, effect of statin use on bone enhance the plausibility of this interaction.

Statins increase new bone formation in vitro and enhance trabecular bone formation in rodents³⁶ through blockade of the mevalonate pathway.³⁷ In women, as opposed to rodents, statins had no effect on markers of bone turnover,³⁸ bone density, fracture risk,³⁹ or progression of coronary calcification⁴⁰; thus, the relationship between statin use and fracture risk remains controversial. Overall, the relationship between consumption of calcium/vitamin D supplements and statin use remains quite unclear with regard to cardiovascular disease risk. Another placebo-controlled trial the size and duration of the Women’s Health Initiative is unlikely to be undertaken, although it is possible that trials of bisphosphonates or other agents may shed some light. Populations in those trials are not at particularly high risk for cardiovascular disease, however, so the number of atherosclerotic events may prove inadequate.

Calcium and vitamin D supplementation did not increase the risk for myocardial infarction, CHD death, stroke, coronary revascularization, hospitalized angina, heart failure, or transient ischemic attack. Thus, women taking these supplements need not fear adverse cardiovascular consequences while protecting their bone health.

Footnotes