Several lines of evidence link calcium metabolism with vascular disease. Serum calcium concentrations are positively associated with increased carotid artery plaque thickness1 and with increased incidence of myocardial infarction (MI) and death.2–4 An increased risk of cardiovascular events and death also has been reported in primary hyperparathyroidism, a condition in which serum calcium is elevated.5, 6 In patients with renal failure, calcium supplements accelerate vascular calcification and increase mortality in both dialysis and predialysis populations.7–9 Calcium supplements also have been associated with increased cardiovascular event rates in a randomized, controlled trial in older women10 and subsequent meta-analysis.11 The mechanism of the association between calcium metabolism and vascular disease is unknown. The process of vascular calcification is thought to be similar to bone formation,12, 13 and since calcium supplements increase bone mineral density (BMD),14, 15 they also may increase vascular calcification.
Abdominal aortic calcification (AAC) is an established risk factor for cardiovascular disease, with the presence of AAC detected by spinal X-rays associated with a higher risk of cardiovascular events and death.16–21 Recently, we and others have reported that AAC can be measured on vertebral morphometric images obtained by dual-energy X-ray absorptiometry (DXA) and that the presence of AAC independently predicted incident MI, stroke, and cardiovascular events, even after adjusting for traditional cardiovascular risk factors.22–25 AAC also has been inversely associated with BMD in many studies. However, in some studies, this relationship did not persist after adjusting for age or other covariates,26, 27 whereas in others, such adjustments did not explain the association.28–32 Similarly, AAC is an independent predictor of fractures in some31–33 but not all studies.34 Coronary artery calcification (CAC), another established risk factor for cardiovascular disease,35 also has been associated with low BMD.36, 37
We have recently completed randomized, controlled trials of calcium supplements in a cohort of healthy elderly women and a cohort of healthy middle-aged or older men.14, 15 We set out to investigate whether indices of calcium metabolism, dietary calcium intake, or use of calcium supplements were associated with AAC or CAC in these trials. We also investigated whether there was any association between AAC and BMD or fractures or between CAC and BMD.
A total of 1471 healthy postmenopausal women took part in a 5-year randomized, placebo-controlled trial of 1 g/day calcium supplementation, and 323 healthy middle-aged and older men took part in a 2-year randomized, placebo-controlled trial of 600 or 1200 mg/day calcium supplementation. The study design and results of both trials have been published previously.10, 14, 15 The women were over 55 years of age, were not taking agents for osteoporosis (including hormone-replacement therapy or vitamin D supplements in doses > 1000 IU/day), and were free from major ongoing illnesses, including hepatic, renal, or thyroid dysfunction, malignancy, metabolic bone disease, or serum 25-hydroxyvitamin D concentration of less than 25 nmol/L and had normal spine BMD for their age (Z-score > –2). The men were over 40 years of age, free from major ongoing illness (including coronary heart disease, hypertension, diabetes, renal or thyroid dysfunction, liver disease, malignancy, or metabolic bone disease), had an estimated 5-year cardiovascular risk of greater than 15% and a serum 25-hydroxyvitamin D concentration of less than 25 nmol/L, had normal spine and hip BMD for their age (Z-score > –2), and were not using lipid-lowering therapy or agents that might have an impact on calcium metabolism.
Abdominal aortic calcification scoring
Vertebral morphometry is the assessment of vertebral shape and size using lateral spinal X-ray plain films or DXA images. Vertebral morphometric images were obtained at baseline and 5 years in the women using a Lunar Expert densitometer (Madison, WI, USA) and at baseline and 2 years in the men using a GE Prodigy densitometer (Madison, WI, USA). Vertebral morphometric images were assessed for AAC as described by Schousboe and colleagues.23 The AAC-8 score was calculated from vertebral morphometric images as the sum of the total length of calcification of the anterior and posterior aortic walls in front of the L1–L4 vertebral segments.23 For each wall, the aggregate length of calcification is scored between 0 and 4 relative to vertebral body height. Thus absent calcification is scored as 0, and if the aggregate length of calcification is 1 or less vertebral body height, the score is 1. If the aggregate length of calcification is greater than 1 and 2 or fewer vertebral heights, the score is 2, and so on, to a maximum score of 4. The scores of each wall are summed to give an AAC-8 score that ranges between 0 to 8.23 For these analyses, each image was scored as to the presence (AAC-8 > 0) or absence of AAC (AAC-8 = 0). Full details of the method have been published previously, including details of the inter- and intrameasurer agreements.25
Coronary calcium scores (CAC)
After the study was completed, all men allocated to placebo or to calcium 1200 mg/day were invited to have a measurement of coronary calcium. As such, 163 men (80 in the placebo group and 83 in the calcium group) agreed and had coronary calcium scoring performed an average of 3.5 years after study entry using a 64-slice CT scanner (LightSpeed VCT, General Electric, Milwaukee, WI, USA). Scans were analyzed using the Agatston system38 with manufacturer-supplied software (Smartscore, General Electric). The coronary calcium score (Agatson score) and the calcium score for the proximal ascending aorta, including the aortic valve (aortic calcium score), were determined for each subject.
Cardiovascular event assessment
Ascertainment of cardiovascular events in these cohorts has been described in detail previously.10 Briefly, participants were seen 6-monthly during the trials, and adverse events were inquired after and recorded, but questions about specific symptoms were not asked. Standard definitions of MI and stroke were used,10, 25 and all self-reported events were verified independently from review of medical records.
Bone density, fracture, and other assessments
BMD of the lumbar spine, proximal femur, and total body was measured every 6 months using the Lunar Expert and GE Prodigy densitometers in the women and men, respectively. Patients were asked about the occurrence of fractures at their 6-monthly visits. All reported fractures were confirmed by radiographs. Osteoporotic fractures included all fractures except those of the head, hands, feet, and ankles and those resulting from major trauma. Dietary calcium was assessed using a validated questionnaire,39 and laboratory measurements were carried out as described previously.14, 15
The primary analyses were carried out on the women's trial because it was larger, had a longer duration of follow-up, and had a higher incidence of cardiovascular events. The analyses were repeated on the men's study to see whether similar findings were obtained.
Differences in baseline characteristics, coronary calcium scores, and prevalence of AAC between treatment groups were assessed using t tests or ANOVA for continuous variables, Wilcoxon tests for non–normally distributed continuous variables, and chi-square or Fisher's exact tests for categorical variables. Categorical modeling was used to assess changes in the presence or absence of AAC with time and whether such changes were related to treatment allocation. The cohorts were divided by the presence or absence of AAC at baseline, and differences between the groups in calcium-related variables and BMD were compared using t tests. The prevalence of AAC by quartile of the calcium-related variables was compared using chi-square tests with tests for linear trends. Categorical modeling was used to assess whether changes in the presence or absence of AAC with time were associated with the baseline calcium-related variables grouped by quartiles. Where significant between-group differences were present for BMD, potential confounders were adjusted for by ANCOVA. A mixed-models approach to repeated measures (ANCOVA) was used to assess whether changes in BMD measurement over time were related to the presence of baseline AAC, with treatment allocation and potential confounders included in the model. To assess the relationship between BMD and CAC, the cohort was divided into four groups by coronary calcium score (0, 1–100, 100–400, >400),35 and differences between groups compared by ANOVA. Finally, Kaplan-Meier analysis and Cox proportional hazards modeling were used to assess the contributions of treatment allocation and calcium-related variables on the incidence of cardiovascular events and the effect of baseline AAC on the incidence of fractures. All analyses were performed using the SAS software package (Version 9.1, SAS Institute, Cary, NC, USA) or Prism (Version 5, GraphPad Software, San Diego, CA, USA). All tests were two-tailed, and p < .05 was considered significant.
A total of 1471 women were randomized to placebo (n = 739) or calcium 1 g/day (n = 732). Then 1424 (718 placebo, 706 calcium) baseline images and 1014 five-year (525, 489) images were scored for AAC. Missing scores were due to scans not being undertaken (26 baseline, 444 follow-up) or scan quality (21 baseline, 13 follow-up). A total of 323 men were randomized to placebo (n = 107), calcium 600 mg/day (n = 108), or calcium 1200 mg/day (n = 108). Then 320 baseline images (106 placebo, 106 calcium 600, and 108 calcium 1200) and 298 two-year images (101, 101, 96) were scored for AAC. Missing scores were due to scans not being undertaken (0 baseline, 18 follow-up) or scan quality (3 baseline, 7 follow-up). The baseline characteristics of the cohorts have been published previously,10, 14, 15 and selected variables are shown in Table 1. The mean duration of follow-up of the women was 4.4 years and of the men was 2.0 years.
Table 1. Selected Baseline Characteristics of the Cohorts Divided by Treatment Allocation
Calcium 1 g/day (n = 706)
Placebo (n = 718)
Calcium 600 mg/day (n = 106)
Calcium 1200 mg/day (n =108)
Placebo (n =106)
Note: Data are mean (SD) or percentage. There were no significant differences between treatment groups for the women or the men.
Cholesterol was measured in all the men and a subgroup of 223 women.
Seven hundred and seventeen (50%) of women had no detectable AAC at baseline, whereas 20%, 17%, 6%, 4%, and 2% had AAC-8 scores of 1, 2, 3, 4, and ≥5, respectively.25 The mean baseline serum total calcium and phosphate concentrations and calcium-phosphate product were significantly higher in women with baseline AAC than in those without baseline AAC (p < .001, p = .04, and p = .003, respectively; Table 2). Figure 1 shows that, across quartiles of total serum calcium at baseline, the proportion of women with AAC increased linearly at baseline (p < .001) and at 5 years (p < .001). For serum phosphate and calcium-phosphate product, similar but weaker patterns were observed both at baseline and at 5 years (Fig. 1). When we divided the cohort into three groups by amount of AAC at baseline (AAC-8 = 0, AAC-8 = 1, or AAC-8 ≥ 2), there were linear increases in baseline serum total calcium and phosphate concentrations and the calcium-phosphate product with increasing AAC at baseline (Fig. 2). Similar relationships were seen between these variables and AAC at 5 years (data not shown).
Table 2. Baseline Biochemistry, Dietary Calcium, and Bone Mineral Density in the Cohorts Divided by the Presence or Absence of Baseline Abdominal Aortic Calcification
The mean baseline dietary calcium intake was significantly higher in women without baseline AAC than in those with baseline AAC (p = .03). This difference was mainly due to those in the highest quartile of calcium intake having the lowest prevalence of AAC (Fig. 1). Both findings persisted after adjustment for age differences between the groups. When we divided the cohort by amount of AAC at baseline, there were linear decreases in baseline dietary calcium intake with increasing AAC at baseline (Fig. 2) and at 5 years (data not shown).
At baseline, 49.7% of the cohort had AAC compared with 56.5% at 5 years (p < .001). This increase was not related to the quartile of baseline total serum calcium (p = .11) or serum phosphate (p = .53) concentration, calcium-phosphate product (p = .63), or baseline dietary calcium intake (p = .21). Of the women allocated to treatment with calcium, 52% had AAC at baseline compared with 48% allocated to placebo (p = .14), and at 5 years, the proportions were 59% and 54%, respectively (p = .18). Also, 71% of the women in each group had no change in AAC from baseline to follow-up. The increases in the proportion of women with AAC over time did not differ by treatment allocation (p = .24).
Forty-nine women had an incident MI, 57 had an incident stroke, and 105 had an incident cardiovascular event (ie, MI, stroke, or sudden death) during follow-up. The incidence of MI and cardiovascular events was related to calcium supplementation10 and baseline AAC.25 However, baseline total serum calcium did not predict MI (p = .45) or cardiovascular events (p = .84), and grouping the cohort by quartiles of baseline total serum calcium concentration produced similar results (p = .58 and p = .86, respectively). Likewise, baseline serum phosphate concentration, the calcium-phosphate product, and dietary calcium intake did not predict MI (p = .57, p = .45, and p = .84, respectively) or cardiovascular events (p = .28, p = .34, and p = .76, respectively), and grouping the cohort by quartiles of these variables produced similar results.
Table 1 shows that BMD was decreased significantly at the total femur (p = .002) and total body (p = .006) but not lumbar spine (p = .19) in women with baseline AAC compared with those without baseline AAC. However, after adjustment for age, baseline AAC was not related to BMD at any site (p > .09 for all sites). We then assessed whether baseline AAC was associated with changes in BMD over time. After adjusting for treatment allocation and age, there was no relationship between baseline AAC and changes in BMD at the total femur (p = .89) or total body (p = .25). At the lumbar spine, BMD increased by 1.4% over 5 years in women with AAC present at baseline compared with an increase of 0.2% in women without baseline AAC (p < .001). Next, we assessed whether increases in AAC were associated with changes in BMD. In 248 women, the AAC-8 score increased from baseline to 5 years, and in 712 women, it did not change. There was no difference in changes in BMD at the spine (p = .87), total femur (p = .59), or total body (p = .06) in women whose AAC-8 score increased compared with those whose score remained constant.
During follow-up, women with baseline AAC did not have an increase in the number of incident osteoporotic fractures (114 versus 99, p = .23) or total fractures (140 versus 130, p = .41) compared with women without baseline AAC, but they did have an increase in the number of incident hip fractures (14 versus 5, p = .039). After adjusting for age and treatment allocation, there were no differences in the incidence of hip (p = .13), osteoporotic (p = .55), or total fractures (p = .77) between women with and without baseline AAC.
Two hundred and sixty-three (82%) men had no detectable AAC at baseline, whereas 17% and 1% had AAC-8 scores of 1 and 2, respectively.25 In contrast to the women's results, there were no differences in baseline mean total serum calcium and serum phosphate concentrations or calcium-phosphate product according to baseline AAC status (p = .42, p = .86, and p = .99, respectively; Table 2). Figure 3 shows that the proportion of men with AAC tended to decrease across quartiles of serum calcium. This was not statistically significant at baseline (p for linear trend = .83) but was significant at 2 years (p = .003). There were similar but generally weaker relationships between AAC and baseline serum phosphate concentration and the calcium-phosphate product and dietary calcium intake to those seen in the cohort of women (Figure 3).
At baseline, 17.8% of the cohort had AAC compared with 26.2% at 2 years (p < .001). This increase was related to the quartile of baseline total serum calcium (p = .01) but not to the quartile of serum phosphate concentration (p = .80), calcium-phosphate product (p = 0.51), or baseline dietary calcium intake (p = .71). Also, 18.9% of men allocated to calcium 600 mg/day, 15.7% of men allocated to calcium 1200 mg/day, and 18.9% of men allocated to placebo had detectable AAC at baseline (p = .79), and at 2 years, the proportions were 26.5%, 24.2%, and 27.7%, respectively (p = .90). In addition, 88% to 89% of the men in each group had no change from baseline in AAC at follow-up. The increases in the proportion of men with AAC over time did not differ by treatment allocation (p = .90).
One hundred and sixty-three men (83 calcium 1200 mg/day and 80 placebo) had a coronary CT scan an average of 1.5 years after completion of the trial. Figure 4 shows that there were no differences between the treatment groups in coronary calcium scores (p = .88). There also were no differences between groups in the aortic calcium score (p = .75). Baseline total serum calcium serum phosphate concentrations, calcium-phosphate product, and dietary calcium intake were not associated with coronary calcium score (p = .37, p = .83, p = .72, and p = .18, respectively).
There were no differences in baseline BMD between men with AAC and men without AAC (Table 1), and baseline AAC was not associated with changes in BMD over time at any site (p > .30 for all sites). There also was no relationship between baseline BMD at any site and coronary calcium score (p > .68 for all sites).
In elderly women, the prevalence of AAC measured using vertebral morphometry at baseline and 5 years was positively related to the baseline total serum calcium and serum phosphate concentrations and the calcium-phosphate product, but changes in AAC over time and the incidence of MI and cardiovascular events were not related to these variables. In a smaller sample of middle-aged men with normal renal function, AAC and CAC were not consistently related to serum calcium or phosphate concentration. AAC tended to be least prevalent in women and men with the highest dietary calcium intake at baseline, but neither dietary calcium intake nor calcium supplementation was related to changes in the prevalence of AAC over time in women or men. Calcium supplementation also was not associated with CAC in men. After adjusting for age, baseline AAC was not consistently related to BMD at baseline, changes in BMD over time, or incidence of fractures. There also was no relationship between BMD and CAC in men.
We are not aware that relationships between serum calcium or phosphate concentration and aortic calcification have been described previously outside the setting of chronic renal failure, where vascular calcification occurs commonly and has been inconsistently related to calcium-phosphate balance.40 However, serum calcium and/or phosphate concentrations have been associated with increased carotid artery plaque thickness1 and increased incidence of cardiovascular events and death in large retrospective3, 4 and prospective cohorts2, 41 in the general population. Primary hyperparathyroidism is another condition with increased serum calcium concentrations and has been associated with increased risk of cardiovascular events and death,5, 6 although evidence for increased vascular calcification is inconsistent.42 Gender differences between the relationships between serum calcium concentration and vascular calcification warrant further study. These differences may have been due to chance or sample-size differences between the groups, or they might have been confounded by the age and renal function differences between the groups. The women were on average 17 years older than the men, with poorer renal function. It would be of interest to assess whether the same gender differences occur in age-comparable groups with similar renal function.
Previously, in this cohort of women, we reported that calcium supplementation was associated with an increased risk of cardiovascular events, principally MI.10 The mechanism for this increased risk remains unknown, but our findings do not rule out the possibility that this increase is mediated by vascular calcification. Our studies may have been too small or their duration too short to detect any changes induced by calcium supplementation, or any such changes may have been below the threshold of detection by the methods we employed. Changes in CAC in response to calcium supplementation may not have been detected because there was no measurement of CAC at baseline, with the single measurement occurring approximately 18 months after study completion. A recent abstract reported that in a 2-year study of 167 men, AAC scores measured by quantitative CT increased in the group assigned to milk fortified with 1000 mg of calcium and 800 IU of vitamin D relative to the control group, after adjustment for potential confounders.43 Further studies on the effect of calcium supplements on vascular calcification are needed. It also would be of interest to examine the effect of calcium supplements on biomarkers of vascular calcification, such as fetuin A, matrix γ-carboxyglutamic acid (Gla) protein, pyrophosphate, osteoprotegerin, and bone morphogenetic proteins.44
The effect of dietary calcium intake on vascular calcification has not been reported for individuals with normal renal function, but epidemiologic studies have suggested that high calcium intakes are associated with a lower prevalence of vascular disease.45–49 However, high calcium intakes have been associated with ischemic brain lesions on MRI50 and with vascular calcification in populations with renal failure.7, 51, 52 Our finding of decreased AAC in the highest quartile of dietary calcium intake is consistent with the epidemiologic studies. The findings that calcium supplements may increase the risk of cardiovascular events10 and that dietary calcium does not appear related to vascular events suggest that calcium supplements and dietary calcium intake have different biologic effects. Ingestion of equivalent doses of calcium from dairy products has a much smaller effect than calcium supplements on serum calcium concentration,53, 54 which may account for some of these differences. Another possibility is that the quantity of dietary calcium intake in these studies is generally much less than the combination of dietary and supplemental calcium intake, which also might contribute to the observed differences.
We did not find a relationship between vascular calcification and BMD or fractures. BMD was related to baseline AAC in women, but this was due to older women having lower BMD values and higher AAC values than younger women, consistent with some26, 27 but not all previous studies.28–32 Supporting this finding was the fact that baseline AAC did not predict subsequent changes in BMD over time, except at the lumbar spine, where the overlying aortic calcification may contribute to the measured BMD. Baseline AAC also was not associated with incident fractures after adjustment for covariates, again consistent with some34 but not all studies.31–33
There are several limitations to our study. Participants all were seen at one center, were healthy volunteers with low cardiovascular risk, and mainly were of European descent (>90%). Therefore, the results may not necessarily be generalized to other populations of different health status, age, and ethnicity. Follow-up assessment of AAC was not available in 30% of the women and 7% of the men. Interpretation of the results of the men's study is limited by the small sample size and shorter duration of follow-up, as well as the lack of a baseline measurement of CAC. CT scanning is a more sensitive technique for assessing vascular calcification. In our study, 71% of the women and 88% of the men had no change in AAC-8 over 5 and 2 years, respectively. Schulz and colleagues reported that AAC assessed using CT scan increased by 14% per year over 2 years in women of mean age of 65 years.31 Taken together, these results suggest that DXA may be a relatively insensitive tool for assessing AAC. It would be valuable to repeat our analyses using more sensitive tools for assessing vascular calcification in the aorta and at other sites.
In conclusion, calcium metabolism appears related to vascular calcification. Serum calcium and serum phosphate concentrations and the calcium phosphate product are associated with AAC in older women, but neither dietary calcium nor calcium supplementation were related to changes in AAC or CAC. Vascular calcification was not related to low BMD or fractures. Further studies exploring the relationships between factors involved in calcium metabolism and vascular calcification should be undertaken.
All the authors state that they have no conflicts of interest.