Abstract
Circulating total calcium or albumin-adjusted calcium is a risk factor for cardiovascular disease. As the biologically active ionized calcium is a physiologically more relevant measure and its association with cardiovascular disease is poorly understood, we tested the hypothesis that high plasma ionized calcium is associated with higher risk of myocardial infarction and ischemic stroke in individuals in the general population.
We included 106 774 individuals from the Copenhagen General Population Study, and defined hypocalcemia and hypercalcemia by the lowest and highest 2.5 percentiles, respectively, using the central 95% reference interval. Information on myocardial infarction and ischemic stroke was from registries and risks calculated using Cox regression and Fine and Gray competing-risks regression.
During a median follow-up of 9.2 years, 4932 individuals received a diagnosis of either myocardial infarction or ischemic stroke. Hypercalcemia was associated with subdistribution hazard ratios of 1.67 (95%CI: 1.05–2.67) for myocardial infarction, 1.28 (0.81–2.02) for ischemic stroke, and of 1.54 (1.10–2.15) for the combined endpoint compared to individuals with plasma ionized calcium within the reference interval; hypocalcemia was not associated with cardiovascular disease. In models using plasma ionized calcium as a continuous variable, the associations were nonlinear; above the median, each 0.1 mmol/L higher plasma ionized calcium was associated with a hazard ratio of 1.31(1.02–1.68) for myocardial infarction, 1.21 (0.95–1.54) for ischemic stroke, and of 1.28 (1.08–1.53) for the combined endpoint.
High plasma ionized calcium is associated with higher risk of myocardial infarction and ischemic stroke compared to plasma ionized calcium within the reference interval.
Introduction
Higher plasma calcium, either measured as total calcium or albumin-adjusted calcium, has been found to be a risk factor for cardiovascular disease (1). Similarly, the use of calcium supplements has been associated with increased risk of cardiovascular disease in secondary analyses of randomized intervention trials (2, 3). Plasma calcium can be measured as ionized (free) calcium, as total calcium (free plus protein-bound calcium plus complex-bound), or as albumin-adjusted calcium (total calcium adjusted for albumin concentration); it remains unclear which measurement should be preferred for clinical practice.
While ionized calcium may be the physiologically relevant measurement representing the biologically active fraction (4), total calcium and albumin-adjusted calcium are easier to measure because they are built into automated clinical laboratory instruments and are reasonably correlated to ionized calcium. However, many studies have found that total calcium or albumin-adjusted calcium tend to misclassify calcium status in a substantial proportion of individuals when ionized calcium is used as the gold standard (5–9). This, in part, may be due to concentrations of calcium binding plasma proteins and small anions being affected by chronic disease, which in turn affect plasma total calcium and albumin-adjusted calcium. Misclassifications of calcium status by total calcium or albumin-adjusted calcium could potentially lead to spurious associations between plasma calcium and cardiovascular disease. Thus, the use of ionized calcium, in contrast to other calcium measures, in epidemiological studies may be more informative in defining the specific role of calcium in cardiovascular disease.
The association of ionized calcium with cardiovascular disease has previously been investigated in one study of 974 middle-aged individuals, showing little evidence for an association with cardiovascular disease risk after adjustment for cardiovascular confounders (10). However, that study may have been too small to detect an association of modest size, and the question of a role of ionized calcium in cardiovascular disease is not yet answered unequivocally.
We therefore tested the hypothesis that plasma ionized calcium is associated with risk of myocardial infarction and ischemic stroke in a large-scale study of 106 774 individuals from the Copenhagen General Population Study.
Materials and Methods
Study Population
We used the Copenhagen General Population Study (CGPS), initiated in 2003 with ongoing enrollment. Briefly, individuals aged 20–100+ years were invited randomly from the Danish Civil Registration System to complete a questionnaire that was reviewed together with an examiner at the day of attendance, undergo a physical examination, and give blood for plasma measurements. The participation rate was 43%. We included 106 774 individuals of Danish descent who had plasma ionized calcium measurements. The study was approved by Herlev and Gentofte Hospital and Danish Ethical Committees, conducted according to the Declaration of Helsinki, and written informed consent was obtained from all participants. See the Supplemental Data for further information on methods.
Ionized Calcium
Ionized calcium was measured in serum using a Konelab autoanalyser (Thermo Fisher Scientific) with ion selective membrane electrodes (ISE). The total measuring range was 0.5–4.0 mmol/L; the day-to-day coefficient of variation was 2%. A pH corrected ionized calcium value was automatically reported; the adjusted value was calculated with the equation: Ca2+ at pH 7.4 = × ) which is valid between pH 7.2 and 7.6 (11). This correction is reasonable when the patient’s blood pH can be assumed to be close to 7.4, as in our sample with individuals from the general population, thus accounting for preanalytic factors causing smaller pH alterations such as time uncapped during automated analyses (12).
Endpoints
Information on diagnosis of myocardial infarction (ICD8 410 and ICD10 I21-22) and ischemic stroke (ICD8 433-434 and ICD10 I63.0-I63.6) from 1977 until December 2018 was obtained from the national Danish Patient Registry, while the date of death or emigration (n = 453) was obtained from the Danish Civil Registration System. For individuals with any registered cerebrovascular disease, records from general practitioners and hospital records were obtained, and the diagnosis of ischemic stroke was validated by 2 independent medical doctors, blinded to the test results as described previously. We used the endpoints separately and as a combined endpoint under the assumption that the mechanisms leading to ischemic stroke and myocardial infarction may be similar.
Covariates
The following lifestyle factors were self-reported: information on smoking, alcohol intake, physical activity, income, use of dietary supplements, antihypertensive treatment, hormone replacement therapy, menopause status, and use of diuretics. Body mass index (BMI), systolic and diastolic blood pressure was measured at the day of attendance. Plasma total cholesterol, high density lipoprotein (HDL) cholesterol, albumin, and creatinine were measured using Konelab autoanalyzer and standard hospital assays. The estimated glomerular filtration rate (eGFR) was calculated using the 2009 CKD-EPI Creatinine Equation (13). Non-HDL cholesterol was calculated as total cholesterol minus HDL cholesterol. Plasma 25-hydroxyvitamin D was measured using the DiaSorin Liaison 25-hydroxyvitamin D TOTAL assay.
Statistical Analyses
We used Stata v.13.1. P-values for trends across 3 groups were estimated using the Cuzick nonparametric trend test. All reported P-values are 2-sided. Plasma ionized calcium was included both as a categorical variable and as a continuous variable. As to the former, we divided plasma ionized calcium into 3 groups corresponding to the 2.5th and 97.5th percentile of the dataset, adjusted for month and year of measurement to account for any drift in the analysis over the years. Furthermore, in sensitivity analyses, we defined the reference interval as ionized calcium between 1.18–1.32 mmol/L, which is the reference interval commonly used in Denmark in order to ease interpretability for clinicians. Although arbitrary, it is a common practice to define the reference interval as the central 95% interval, and we hereby created a study-specific reference interval in which all individuals were recruited similarly and venipuncture samples were taken under the same circumstances and at the same time of day (14).
To examine the association between plasma ionized calcium and myocardial infarction and ischemic stroke, as well as the composite endpoint, we used Cox proportional hazards regression models with entry at examination and age as the time scale to estimate hazard ratios with 95% confidence intervals. We adjusted all analyses for age, as age is a major confounder for cardiovascular disease. Those emigrating or dying during follow-up were censored at their emigration or death dates. All estimates were adjusted for regression dilution bias using a nonparametric method as uncorrected associations of disease risk with baseline measurements underestimate the strength of the real association (15) (Supplemental Table 1). Multivariable adjustment was for covariates chosen a priori that potentially could confound the association between plasma ionized calcium and the endpoints. We performed multivariable adjusted, restricted cubic spline Cox regression for graphical representation using plasma ionized calcium and 4 knots, as well as univariate kernel density estimation for density graphs. Also, stratified and interaction analyses were performed using similar models. See the Supplemental Data for further information.
Since Cox regression may lead to biased effect estimates in the presence of a competing risk such as death or emigration that can preclude the occurrence of myocardial infarction or ischemic stroke, we calculated subdistribution hazard ratios for the endpoints accounting for the competing risk of death by any cause or emigration. This was done for ionized calcium trichotomized as described above by using multivariable adjusted competing-risks survival regression with the method of Fine and Gray (16).
Finally, plots using kernel-weighted local polynomial regression of ionized calcium on selected confounders were displayed as graphs of the smoothed values with confidence bands. This was done to allow visual interpretation of the association between ionized calcium and possible confounders such as age, eGFR, plasma vitamin D, and systolic blood pressure.
Results
Among the 106 774 individuals with plasma ionized calcium measurements available, after excluding individuals with a prior diagnosis of either myocardial infarction or ischemic stroke, a total of 104 549 for myocardial infarction, 105 615 for ischemic stroke, and 103 488 individuals for the combined endpoint were included in the prospective analyses with a median follow-up of 9.2 years (range: 0.003–15 years, interquartile range: 6.6–11.9). Among these, 2560 (2.4%) and 2704 (2.6%) received the diagnosis of myocardial infarction and ischemic stroke, respectively, and 4932 (5.3%) received either of the diagnoses. The median and the mean plasma ionized calcium were both 1.21 mmol/L (SD: 0.054 mmol/L). In the hypocalcemia group, median plasma ionized calcium was 1.11 mmol/L (range: 0.77–1.17 mmol/L), for the reference interval and hypercalcemia groups the corresponding values were 1.21 mmol/L (range: 1.10–1.37 mmol/L) and 1.36 (range: 1.30–2.01 mmol/L), respectively. Most characteristics were found statistically significantly different between groups (Table 1).
The association of baseline characteristics with plasma ionized calcium.
| Plasma ionized calcium | ||||
|---|---|---|---|---|
| Hypocalcemia | Reference interval | Hypercalcemia | P | |
| Plasma ionized calcium, mmol/L* | 1.11 (1.10-1.12) | 1.21 (1.18-1.24) | 1.36 (1.34-1.40) | |
| n | 2564 | 101 543 | 2667 | |
| Men, n (%) | 1385 (54) | 45 842 (45) | 786 (29) | 6 × 10-72 |
| Age, years* | 55.6 (45.7-68.5) | 58.0 (48.2-67.3) | 63.6 (54.8-71.1) | 2 × 10-66 |
| Ever smokers, n (%) | 1397 (54) | 59 104 (58) | 1645 (62) | 1 × 10-7 |
| Body mass index, kg/m2* | 25.8 (23.3-28.8) | 25.6 (23.2-28.4) | 25.7 (23.2-28.7) | 0.64 |
| Alcohol consumption, g/week* | 84 (36-168) | 96 (48-180) | 108 (48-180) | 1 × 10-6 |
| High income, n (%) | 1167 (46) | 43 743 (43) | 875 (33) | 6 × 10-21 |
| High physical activity in leisure time, n (%) | 204 (8.0) | 6825 (6.7) | 132 (5.0) | 1 × 10-5 |
| Systolic blood pressure, mmHg* | 140 (126-155) | 140 (126-155) | 142 (130-157) | 2 × 10-7 |
| Diabetes, n (%) | 142 (5.5) | 4243 (4.2) | 140 (5.3) | 0.66 |
| Use of diuretics, n (%) | 277 (11) | 6911 (6.8) | 266 (10) | 0.32 |
| eGFR < 60 mL/min/1.73 m2, % | 310 (12) | 9663 (9.5) | 402 (15) | 1 × 10-4 |
| Plasma non-HDL cholesterol, mmol/L* | 3.8 (3.0-4.5) | 3.9 (3.2-4.7) | 4.0 (3.3-4.8) | 6 × 10-14 |
| Plasma vitamin D < 50nmol/L, n (%)a | 204 (41) | 10 932 (45) | 234 (43) | 0.66 |
| Antihypertensive treatment at baseline, n (%) | 542 (21) | 19 777 (20) | 761 (29) | 5 × 10-12 |
| Hypertension at baseline, n (%) | 1412 (55) | 56 294 (55) | 1642 (62) | 2 × 10-6 |
| Any use of dietary supplements, n (%) | 1128 (44) | 49 022 (48) | 1489 (56) | 7 × 10-18 |
| Other chronic disease, n (%) | 329 (13) | 10 557 (10) | 405(15) | 0..003 |
| Hyperparathyroidism diagnosis prior to baseline, n (%) | 6 (0.2) | 147 (0.1) | 50 (1.9) | 6 × 10-44 |
| Hormone replacement therapy**, n (%) | 133 (11) | 6736 (12) | 264 (14) | 0.01 |
| Plasma ionized calcium | ||||
|---|---|---|---|---|
| Hypocalcemia | Reference interval | Hypercalcemia | P | |
| Plasma ionized calcium, mmol/L* | 1.11 (1.10-1.12) | 1.21 (1.18-1.24) | 1.36 (1.34-1.40) | |
| n | 2564 | 101 543 | 2667 | |
| Men, n (%) | 1385 (54) | 45 842 (45) | 786 (29) | 6 × 10-72 |
| Age, years* | 55.6 (45.7-68.5) | 58.0 (48.2-67.3) | 63.6 (54.8-71.1) | 2 × 10-66 |
| Ever smokers, n (%) | 1397 (54) | 59 104 (58) | 1645 (62) | 1 × 10-7 |
| Body mass index, kg/m2* | 25.8 (23.3-28.8) | 25.6 (23.2-28.4) | 25.7 (23.2-28.7) | 0.64 |
| Alcohol consumption, g/week* | 84 (36-168) | 96 (48-180) | 108 (48-180) | 1 × 10-6 |
| High income, n (%) | 1167 (46) | 43 743 (43) | 875 (33) | 6 × 10-21 |
| High physical activity in leisure time, n (%) | 204 (8.0) | 6825 (6.7) | 132 (5.0) | 1 × 10-5 |
| Systolic blood pressure, mmHg* | 140 (126-155) | 140 (126-155) | 142 (130-157) | 2 × 10-7 |
| Diabetes, n (%) | 142 (5.5) | 4243 (4.2) | 140 (5.3) | 0.66 |
| Use of diuretics, n (%) | 277 (11) | 6911 (6.8) | 266 (10) | 0.32 |
| eGFR < 60 mL/min/1.73 m2, % | 310 (12) | 9663 (9.5) | 402 (15) | 1 × 10-4 |
| Plasma non-HDL cholesterol, mmol/L* | 3.8 (3.0-4.5) | 3.9 (3.2-4.7) | 4.0 (3.3-4.8) | 6 × 10-14 |
| Plasma vitamin D < 50nmol/L, n (%)a | 204 (41) | 10 932 (45) | 234 (43) | 0.66 |
| Antihypertensive treatment at baseline, n (%) | 542 (21) | 19 777 (20) | 761 (29) | 5 × 10-12 |
| Hypertension at baseline, n (%) | 1412 (55) | 56 294 (55) | 1642 (62) | 2 × 10-6 |
| Any use of dietary supplements, n (%) | 1128 (44) | 49 022 (48) | 1489 (56) | 7 × 10-18 |
| Other chronic disease, n (%) | 329 (13) | 10 557 (10) | 405(15) | 0..003 |
| Hyperparathyroidism diagnosis prior to baseline, n (%) | 6 (0.2) | 147 (0.1) | 50 (1.9) | 6 × 10-44 |
| Hormone replacement therapy**, n (%) | 133 (11) | 6736 (12) | 264 (14) | 0.01 |
* Median (interquartile range).
Only measured in a subgroup of participants.
Other chronic diseases: any diagnosis of chronic obstructive pulmonary disease, diabetes type 1 or type 2 or cancer (non-melanoma skin cancer excluded) prior to examination date.
Hormone replacement therapy is any reported use of hormone replacement therapy or use of birth control pills after menopause among women. Only reported for women. eGFR: estimated glomerular filtration rate.
The association of baseline characteristics with plasma ionized calcium.
| Plasma ionized calcium | ||||
|---|---|---|---|---|
| Hypocalcemia | Reference interval | Hypercalcemia | P | |
| Plasma ionized calcium, mmol/L* | 1.11 (1.10-1.12) | 1.21 (1.18-1.24) | 1.36 (1.34-1.40) | |
| n | 2564 | 101 543 | 2667 | |
| Men, n (%) | 1385 (54) | 45 842 (45) | 786 (29) | 6 × 10-72 |
| Age, years* | 55.6 (45.7-68.5) | 58.0 (48.2-67.3) | 63.6 (54.8-71.1) | 2 × 10-66 |
| Ever smokers, n (%) | 1397 (54) | 59 104 (58) | 1645 (62) | 1 × 10-7 |
| Body mass index, kg/m2* | 25.8 (23.3-28.8) | 25.6 (23.2-28.4) | 25.7 (23.2-28.7) | 0.64 |
| Alcohol consumption, g/week* | 84 (36-168) | 96 (48-180) | 108 (48-180) | 1 × 10-6 |
| High income, n (%) | 1167 (46) | 43 743 (43) | 875 (33) | 6 × 10-21 |
| High physical activity in leisure time, n (%) | 204 (8.0) | 6825 (6.7) | 132 (5.0) | 1 × 10-5 |
| Systolic blood pressure, mmHg* | 140 (126-155) | 140 (126-155) | 142 (130-157) | 2 × 10-7 |
| Diabetes, n (%) | 142 (5.5) | 4243 (4.2) | 140 (5.3) | 0.66 |
| Use of diuretics, n (%) | 277 (11) | 6911 (6.8) | 266 (10) | 0.32 |
| eGFR < 60 mL/min/1.73 m2, % | 310 (12) | 9663 (9.5) | 402 (15) | 1 × 10-4 |
| Plasma non-HDL cholesterol, mmol/L* | 3.8 (3.0-4.5) | 3.9 (3.2-4.7) | 4.0 (3.3-4.8) | 6 × 10-14 |
| Plasma vitamin D < 50nmol/L, n (%)a | 204 (41) | 10 932 (45) | 234 (43) | 0.66 |
| Antihypertensive treatment at baseline, n (%) | 542 (21) | 19 777 (20) | 761 (29) | 5 × 10-12 |
| Hypertension at baseline, n (%) | 1412 (55) | 56 294 (55) | 1642 (62) | 2 × 10-6 |
| Any use of dietary supplements, n (%) | 1128 (44) | 49 022 (48) | 1489 (56) | 7 × 10-18 |
| Other chronic disease, n (%) | 329 (13) | 10 557 (10) | 405(15) | 0..003 |
| Hyperparathyroidism diagnosis prior to baseline, n (%) | 6 (0.2) | 147 (0.1) | 50 (1.9) | 6 × 10-44 |
| Hormone replacement therapy**, n (%) | 133 (11) | 6736 (12) | 264 (14) | 0.01 |
| Plasma ionized calcium | ||||
|---|---|---|---|---|
| Hypocalcemia | Reference interval | Hypercalcemia | P | |
| Plasma ionized calcium, mmol/L* | 1.11 (1.10-1.12) | 1.21 (1.18-1.24) | 1.36 (1.34-1.40) | |
| n | 2564 | 101 543 | 2667 | |
| Men, n (%) | 1385 (54) | 45 842 (45) | 786 (29) | 6 × 10-72 |
| Age, years* | 55.6 (45.7-68.5) | 58.0 (48.2-67.3) | 63.6 (54.8-71.1) | 2 × 10-66 |
| Ever smokers, n (%) | 1397 (54) | 59 104 (58) | 1645 (62) | 1 × 10-7 |
| Body mass index, kg/m2* | 25.8 (23.3-28.8) | 25.6 (23.2-28.4) | 25.7 (23.2-28.7) | 0.64 |
| Alcohol consumption, g/week* | 84 (36-168) | 96 (48-180) | 108 (48-180) | 1 × 10-6 |
| High income, n (%) | 1167 (46) | 43 743 (43) | 875 (33) | 6 × 10-21 |
| High physical activity in leisure time, n (%) | 204 (8.0) | 6825 (6.7) | 132 (5.0) | 1 × 10-5 |
| Systolic blood pressure, mmHg* | 140 (126-155) | 140 (126-155) | 142 (130-157) | 2 × 10-7 |
| Diabetes, n (%) | 142 (5.5) | 4243 (4.2) | 140 (5.3) | 0.66 |
| Use of diuretics, n (%) | 277 (11) | 6911 (6.8) | 266 (10) | 0.32 |
| eGFR < 60 mL/min/1.73 m2, % | 310 (12) | 9663 (9.5) | 402 (15) | 1 × 10-4 |
| Plasma non-HDL cholesterol, mmol/L* | 3.8 (3.0-4.5) | 3.9 (3.2-4.7) | 4.0 (3.3-4.8) | 6 × 10-14 |
| Plasma vitamin D < 50nmol/L, n (%)a | 204 (41) | 10 932 (45) | 234 (43) | 0.66 |
| Antihypertensive treatment at baseline, n (%) | 542 (21) | 19 777 (20) | 761 (29) | 5 × 10-12 |
| Hypertension at baseline, n (%) | 1412 (55) | 56 294 (55) | 1642 (62) | 2 × 10-6 |
| Any use of dietary supplements, n (%) | 1128 (44) | 49 022 (48) | 1489 (56) | 7 × 10-18 |
| Other chronic disease, n (%) | 329 (13) | 10 557 (10) | 405(15) | 0..003 |
| Hyperparathyroidism diagnosis prior to baseline, n (%) | 6 (0.2) | 147 (0.1) | 50 (1.9) | 6 × 10-44 |
| Hormone replacement therapy**, n (%) | 133 (11) | 6736 (12) | 264 (14) | 0.01 |
* Median (interquartile range).
Only measured in a subgroup of participants.
Other chronic diseases: any diagnosis of chronic obstructive pulmonary disease, diabetes type 1 or type 2 or cancer (non-melanoma skin cancer excluded) prior to examination date.
Hormone replacement therapy is any reported use of hormone replacement therapy or use of birth control pills after menopause among women. Only reported for women. eGFR: estimated glomerular filtration rate.
As to potential volunteer bias, individuals who participated were older and more often women than individuals who were invited but did not participate (Supplemental Table 2).
Ionized Calcium and Risk of Myocardial Infarction and Ischemic Stroke
The multivariable adjusted subdistribution hazard ratios taking into account the competing risk of death and emigration were 1.67 (95%CI: 1.05–2.67) for myocardial infarction, 1.28 (0.81–2.02) for ischemic stroke, and 1.54 (1.10–2.15) for the combined endpoint, for individuals with hypercalcemia compared to individuals with ionized calcium within the reference interval (Fig. 1). For hypocalcemia, the corresponding subdistribution hazard ratios were 1.04 (0.61–1.78), 1.20 (0.71–2.03), and 1.22 (0.83–1.80), respectively.
Cumulative incidence of myocardial infarction, ischemic stroke, and the combined endpoint. Cumulative incidence and subdistribution hazard ratio are from multivariable adjusted competing risk regression by the method of Fine and Gray, taking into account the competing risk of death or emigration and age is used as time scale. Estimates are adjusted for regression dilution bias. Red line: individuals with hypercalcemia. Black line: individuals within the reference interval. Green (dotted) line: individuals with hypocalcemia.
Above the median plasma ionized calcium value of 1.21 mmol/L, we found an approximately linear association between plasma ionized calcium and risk of myocardial infarction, ischemic stroke, as well as the combined endpoint with no apparent association for ionized calcium below the median (Figs. 2 and 3). In individuals with hypercalcemia, the multivariable adjusted hazard ratios were 1.71 (95%CI: 1.08–2.70) for myocardial infarction, 1.27 (0.81–1.99) for ischemic stroke, and 1.55 (1.12–2.16) for the combined endpoint compared to individuals with plasma ionized calcium within the study-specific reference interval. As the association was not linear over the whole range of plasma ionized calcium, we estimated hazard ratios for above and below the median per 0.1 mmol/L higher plasma ionized calcium, and found that each 0.1 mmol/L higher plasma ionized calcium was associated with a multivariable adjusted hazard ratio of 1.31 (1.02–1.68) for myocardial infarction, 1.21 (0.95–1.54) for ischemic stroke, and 1.28 (1.08–1.53) for the combined endpoint when including individuals with plasma ionized calcium above the median (Fig. 3).
Plasma ionized calcium and risk of myocardial infarction, ischemic stroke, and the combined endpoint. Solid lines are multivariable adjusted hazard ratios using a restricted cubic spline regression with 4 knots with a polynomial smoother. Dashed lines indicate 95% confidence intervals. Analyses are multivariable adjusted for year and month of measurement, sex, body mass index, alcohol intake, smoking status, use of diuretics, income, physical activity during leisure time, systolic blood pressure, estimated glomerular filtration rate, plasma non-HDL cholesterol, any use of vitamins, hypertension, chronic illnesses, and diabetes, and age is used as time scale. A kernel density plot of plasma ionized calcium is also depicted. Individuals with events prior to baseline were excluded from analyses (n = 3283 for the combined endpoint of myocardial infarction or ischemic stroke, n = 2222 for myocardial infarction and n = 1156 for ischemic stroke), thus, the number of individuals in each analysis differ. The graph was truncated at <1.05 mmol/L and >1.50 mmol/L (less than 20 individuals outside each extreme). The median of 1.21 mmol/L was used as reference.
Risk of cardiovascular disease associated with calcium status and per 0.1 mmol/L higher plasma ionized calcium, stratified by median plasma ionized calcium. Analyses are multivariable adjusted for year and month of measurement, sex, body mass index, alcohol intake, smoking status, use of diuretics, income, physical activity during leisure time, systolic blood pressure, estimated glomerular filtration rate, plasma non-HDL cholesterol, any use of vitamins, hypertension, chronic illnesses, and diabetes, and age is used as time scale. Estimates adjusted for regression dilution bias. N: total number of individuals. Analyses per 0.1 mmol/L higher plasma ionized calcium were stratified by the median of 1.21 mmol/L.
Ionized Calcium and Association with Confounders
In the age range of 40 to 70 years, plasma ionized calcium was higher with higher age; after the age of 70 years plasma ionized calcium seemed to decrease with age (Fig. 4). For eGFR, plasma ionized calcium was highest for those with eGFR in the range of 50–60 mL/min/1.73 m2, with lower plasma ionized calcium both in those with eGFR below and above this range. For plasma vitamin D and systolic blood pressure, plasma ionized calcium was marginally higher in individuals with higher plasma vitamin D and systolic blood pressure, respectively (Fig. 4). A one standard deviation higher albumin (53 µmol/L) was associated with -0.0038 mmol/L (-0.0041 to -0.0035 mmol/L) lower ionized calcium (data not shown graphically).
Polynomial regression of plasma ionized calcium and selected confounders. Graphs were created using kernel-weighted local polynomial regression of ionized calcium on selected confounders and displayed as graphs of the smoothed values with confidence bands.
Sensitivity Analyses
We investigated the association of high plasma ionized calcium and risk of the combined endpoint of myocardial infarction or ischemic stroke in strata of the included confounders. The only evidence of interaction was found with BMI with a regression dilution bias and multivariable adjusted hazard ratio of 1.95 (95%CI : 1.39–2.27) in individuals with BMI < 30 kg/m2 and of 0.65 (0.28–1.49) in individuals with a BMI ≥30 kg/m2 for hypercalcemia compared to individuals with plasma ionized calcium within the reference range (P for interaction: 0.005). All established risk factors for cardiovascular disease were associated with higher BMI except for plasma ionized calcium (Supplemental Table 3). The sensitivity of hypercalcemia was <4% and specificity >95% for all endpoints (Supplemental Table 4).
In other sensitivity analyses, we defined hypo- and hypercalcemia from the reference interval applied in clinical practice in Denmark (1.18–1.32 mmol/L) instead of the central 95% based on the data in the present study. Using these cut-offs, for hypocalcemia (n = 30 199) and for hypercalcemia (n = 4358) we found hazard ratios of 1.02 (95%CI: 0.84–1.24) and 1.61 (1.04–2.38) for myocardial infarction, versus individuals within the reference range (n = 69 992). Using this clinical reference range, for ischemic stroke, hypocalcemia was associated with a hazard ratio of 0.99 (0.82–1.20) and hypercalcemia with a hazard ratio of 1.55 (1.08–2.22). For the combined endpoint of myocardial infarction and ischemic stroke, corresponding hazard ratios were 1.01 (0.88–1.17) and 1.60 (1.21–2.10), respectively.
Excluding individuals with events within 2 years of the examination date yielded similar results as main analyses, the hazard ratios for myocardial infarction were 1.00 (95% CI: 0.55–1.82) and 1.42 (0.84–2.41) for hypocalcemia and hypercalcemia, respectively. For ischemic stroke and the combined endpoint, the corresponding hazard ratios were 0.99 (0.54–1.81) and 1.19 (0.72–1.96) and 1.12 (0.72–1.73) and 1.38 (0.95–2.0), respectively. We also performed analyses where we excluded individuals with BMI below 18.5 kg/m2 and results were similar to those presented.
Discussion
In this study of 106 774 individuals, high plasma ionized calcium was associated with higher risk of myocardial infarction and possibly ischemic stroke compared to plasma ionized calcium within the reference interval. To the best of our knowledge, our study is the only large-scale study to assess the association between plasma ionized calcium and cardiovascular disease.
Several mechanisms may explain the observed association between plasma ionized calcium and cardiovascular disease. Plasma calcium has been associated with atherosclerosis quantified by coronary artery calcium, with increased carotid artery plaque thickness, and intracranial atherosclerosis, changes that have been found to be predictive of future cardiovascular events (17–19). Thus, high plasma calcium may directly accelerate the atherosclerotic process. However, findings regarding the role of calcification in atherosclerotic plaque rupture risk have been contradictory (20, 21). Present studies suggest a complex relationship between microcalcification and plaque rupture; the early process of calcification may cause more rupture-prone plaques, whereas later, more densely calcified lesions may cause a more stable form of atherosclerosis with low probability of rupture. Although the exact mechanisms leading to calcifications are not clear, in vitro models suggest that extracellular calcium leads to differentiation of vascular smooth muscle cells into an osteogenic phenotype that may promote local mineralization (22).
Previous studies have shown conflicting results regarding the association of plasma calcium measures with cardiovascular disease. A meta-analysis of observational studies including data from 8 studies and approximately 31 000 individuals found a multivariable adjusted hazard ratio per standard deviation higher total calcium or albumin-adjusted calcium of 1.04 (95%CI: 1.01–1.08) for cardiovascular disease (1). However, the results differ according to measure of plasma calcium investigated; while total calcium was a risk factor for myocardial infarction and stroke in most studies (23–26), studies of albumin-adjusted calcium were largely negative (27, 28). As previously mentioned, one study of 974 individuals found no association between ionized calcium and cardiovascular disease risk after adjustment for cardiovascular confounders (10). However, that study included relatively few individuals, all of early middle-age, and thus had limited statistical power to exclude an association between plasma ionized calcium and cardiovascular disease. As that study only assessed the risk in a middle-aged population, any risk associated with plasma calcium in individuals of higher age were not addressed. In contrast, we included >100 000 individuals, spanning the age from 20 to 100 years, and we were able to detect a modest association, likely to be overlooked by smaller studies.
In support of our findings, meta-analyses of randomized clinical trials including >10 000 individuals found calcium supplementation allocation to be associated with increased risk of myocardial infarction in general or in subgroups of participants (2, 3). Moreover, 2 genetic case-control studies found that genetic variants in the calcium-sensing receptor (CASR) affecting total calcium and albumin-adjusted calcium may be associated with myocardial infarction (26, 29). Additionally, a Mendelian randomization study using 6 single nucleotide polymorphisms associated with plasma total calcium and >60 000 coronary artery disease cases found higher plasma total calcium to be associated with higher risk of cardiovascular disease and myocardial infarction (30).
Interestingly, although the number of stroke cases was higher, the risk of stroke was not as strongly associated with plasma ionized calcium as myocardial infarction. This could suggest that different mechanisms are implicated, or that calcium is a proxy for a risk factor which affects stroke risk differently than the risk of myocardial infarction. These are important results, as previous studies of this association have assessed total calcium or albumin-adjusted calcium, measurements that may not always reflect calcium homeostasis. The relation between total calcium and ionized calcium is influenced by plasma concentrations of proteins and small calcium-binding anions, or changes in the albumin binding of calcium (31). Thus, certain populations may be prone to misclassification of calcium status by cruder measurements of calcium, which could potentially lead to spurious associations between plasma calcium and cardiovascular disease. However, our results, using tightly regulated and biologically active ionized calcium, supports the notion that extracellular calcium plays a role in cardiovascular disease, and perhaps more so in myocardial infarction than in stroke. In addition to the use of ionized calcium, an important strength of our study was the large number of potential confounders included. The association between high ionized calcium and cardiovascular disease could potentially be confounded by treatment with diuretics, known to alter urinary calcium excretion (32), by other chronic illnesses known to increase both calcium and cardiovascular risk, or calcium supplementation. However, diuretics seem to have limited effects on plasma ionized calcium (32), and as to chronic illnesses including cancer, we had extensive self-reported and register-based information on the included individuals, and confounding by unrecorded severe illnesses seems unlikely. As to confounding by indication for calcium supplementation, calcium supplementation only marginally increases plasma ionized calcium (33), which does not support overtreatment with calcium as a likely cause of hypercalcemia in our study.
With our findings, high ionized calcium, in addition to high total calcium, is now shown to be associated with increased risk of cardiovascular disease. Whether the measurement of ionized calcium provides a benefit over total calcium, other studies may address in the future. As the sensitivity of hypercalcemia was low for all 3 endpoints, hypercalcemia is not likely a useful screening tool for cardiovascular disease and with the high false positive rate (approximately 96%), high ionized calcium should probably not prompt further clinical testing for cardiovascular disease unless additional risk factors are present.
Unexpectedly, we found that hypercalcemia was not associated with cardiovascular disease risk in individuals with BMI ≥ 30 kg/m2 compared with individuals with BMI < 30 kg/m2. We cannot explain this finding, but if left to speculate, the following explanations could be offered. First, as morbidly obese individuals may have a more symptomatic presentation of primary hyperparathyroidism than individuals with BMI < 25 kg/m2 (34), obese individuals with hypercalcemia could be more likely to be treated with parathyroidectomy, which may weaken the association in this subgroup. Second, competing common risk factors such as hypertension, diabetes, and hyperlipidemia in the obese may attenuate any association of hypercalcemia with cardiovascular disease risk, suggesting that high plasma ionized calcium has the largest relationship with risk in individuals with less severe cardiovascular risk profiles (35). Third, this could just be a chance finding.
Some limitations to our study should be noted. First, we were not able to directly account for primary hyperparathyroidism, a frequent cause of hypercalcemia in an outpatient setting. Primary hyperparathyroidism increases plasma calcium by increasing bone resorption, tubular calcium reabsorption, and renal synthesis of 1,25-dihydroxyvitamin D, which in turn increases intestinal calcium absorption (36). Primary hyperparathyroidism has been associated with premature valvular calcifications and left ventricular hypertrophy (37, 38), and parathyroid hormone (PTH) may itself exert a hypertrophic effect on cardiomyocytes or augment calcium entry into tissues independent of calcium concentrations (39, 40). Thus, if the observed hypercalcemia in our study is a consequence of hyperparathyroidism, the association between high calcium and cardiovascular disease could be due to parathyroid hormone and not ionized calcium per se. Second, we did not have total calcium measurements and only one measurement of plasma ionized calcium, and the latter may not reflect calcium concentrations of an individual over a longer time span. However, we have adjusted estimates for regression dilution bias in an attempt to account for underestimation of risk associations due to regression dilution occurring in long-term follow-up studies. As to the observed relation between ionized calcium and albumin, this is best explained as an artifactual measurement bias due to a diffusion potential at the reference electrode (See the Supplemental Data). Finally, as individuals who participated in the CGPS were healthier than individuals who were invited but did not participate, this may have biased our results. This bias is most likely towards the null hypothesis and does not invalidate our finding, but may make the results somewhat less generalizable to more sick populations.
Future studies are needed to replicate our findings of high plasma ionized calcium and higher risk of myocardial infarction and stroke. Also, the interplay between parathyroid and ionized calcium and cardiovascular disease risk needs to be addressed. However, as observational associations are prone to confounding and reverse causation, genetic variants robustly associated with plasma ionized calcium, not only total or albumin-corrected calcium, need to be identified and Mendelian randomization studies of genetically high ionized calcium and cardiovascular disease are warranted.
In conclusion, we found that high plasma ionized calcium was associated with higher risk of myocardial infarction and stroke compared to individuals with plasma ionized calcium within the reference range. To the best of our knowledge, our study is the only large-scale study to assess the association between ionized calcium and cardiovascular disease.
Supplemental Material
Supplemental material is available at Clinical Chemistry online.
Author Contributions
All authors confirmed they have contributed to the intellectual content of this paper and have met the following 4 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; (c) final approval of the published article; and (d) agreement to be accountable for all aspects of the article thus ensuring that questions related to the accuracy or integrity of any part of the article are appropriately investigated and resolved.
C.J. Kobylecki, statistical analysis; B.G. Nordestgaard, financial support, provision of study material or patients.
Authors’ Disclosures or Potential Conflicts of Interest
Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest:
Employment or Leadership
None declared.
Consultant or Advisory Role
None declared.
Stock Ownership
None declared.
Honoraria
None declared.
Research Funding
This work was supported by Herlev and Gentofte Hospital, Copenhagen University Hospital.
Expert Testimony
None declared.
Patents
None declared.
Role of Sponsor
The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, preparation of manuscript, or final approval of manuscript.
Acknowledgments
We thank Niels Fogh-Andersen for constructive comments and contribution to the manuscript, as well as staff and participants of the Copenhagen General Population Study.
