The smooth spline estimates the hazard ratio of the combined coronary heart disease event, according to annualized serum concentrations of 25(OH)D (nanograms per milliliter). Splines are adjusted for age, sex, and study site. Dotted lines represent 95% confidence intervals. Below each spline is the histogram of the distribution of serum 25(OH)D concentration.
eTable 1. Baseline participant characteristics by race/ethnicity
eTable2. Associations of annualized serum 25-hydroxyvitamin D concentration with hard incident coronary heart disease events
eTable3. Associations of raw serum 25-hydroxyvitamin D concentration with incident coronary heart disease events
eTable4. Associations of annualized serum 25-hydroxyvitamin D concentration with incident coronary heart disease events, after exclusion of 46 events occurring within the first 12 months of follow-up
eTable5. Associations of annualized serum 25-hydroxyvitamin D concentration with incident coronary heart disease events, among participants with “good,” “very good,” or “excellent” self-reported general health at baseline
Robinson-Cohen C, Hoofnagle AN, Ix JH, Sachs MC, Tracy RP, Siscovick DS, Kestenbaum BR, de Boer IH. Racial Differences in the Association of Serum 25-Hydroxyvitamin D Concentration With Coronary Heart Disease Events. JAMA. 2013;310(2):179-188. doi:10.1001/jama.2013.7228
Low circulating concentrations of 25-hydroxyvitamin D (25[OH]D) have been consistently associated with an increased risk of coronary heart disease (CHD) in white populations. This association has not been rigorously evaluated in other races or ethnicities, in which the distributions of 25(OH)D concentration and possibly other aspects of 25(OH)D metabolism differ.
To examine the association of serum 25(OH)D concentration with risk of CHD in a multiethnic population.
Design, Setting, and Participants
We studied 6436 participants in the Multi-Ethnic Study of Atherosclerosis (MESA), recruited from July 2000 through September 2002, who were free of known cardiovascular disease at baseline. We measured baseline serum 25(OH)D concentrations using a mass spectrometry assay calibrated to established standards. We tested associations of 25(OH)D with adjudicated CHD events assessed through May 2012.
Main Outcome and Measures
Primary outcome measure was time to first adjudicated CHD event, defined as myocardial infarction, angina, cardiac arrest, or CHD death.
During a median follow-up of 8.5 years, 361 participants had an incident CHD event (7.38 events per 1000 person-years). Associations of 25(OH)D with CHD differed by race/ethnicity (P for interaction < .05). After adjustment, lower 25(OH)D concentration was associated with a greater risk of incident CHD among participants who were white (n = 167 events; hazard ratio [HR], 1.26 [95% CI, 1.06-1.49] for each 10-ng/mL decrement in 25(OH)D) or Chinese (HR, 1.67 [95% CI, 1.07-2.61]; n = 27). In contrast, 25(OH)D was not associated with risk of CHD in participants who were black (HR, 0.93 [95% CI, 0.73-1.20]; n = 94) or Hispanic (HR, 1.01 [95% CI, 0.77-1.33]; n = 73).
Conclusions and Relevance
Lower serum 25(OH)D concentration was associated with an increased risk of incident CHD events among participants who were white or Chinese but not black or Hispanic. Results evaluating 25(OH)D in ethnically homogeneous populations may not be broadly generalizable to other racial or ethnic groups.
Low circulating concentrations of 25-hydroxyvitamin D (25[OH]D) have been consistently associated with increased risk of clinical and subclinical coronary heart disease (CHD).1- 11 Whether this relationship is causal and modifiable with vitamin D supplementation has not yet been determined in well-powered clinical trials, which are ongoing.12 However, experimental studies provide biological plausibility for a causal relationship. These studies demonstrate that 1,25-dihydroxyvitamin D, the active vitamin D hormone produced from 25(OH)D, potently suppresses the renin-angiotensin system, modulates immune cell function in a manner that may reduce chronic inflammation, and inhibits abnormal cell proliferation.13- 16 Based in part on these potential beneficial actions, 25(OH)D testing and supplementation have become increasingly common over the last decade.17
Most studies of 25(OH)D and risk of CHD have examined populations that are composed largely or entirely of white participants.1- 6,8,10,11 Results from these studies are frequently extrapolated to multiracial populations.12,18 This may not be appropriate because vitamin D metabolism and circulating 25(OH)D concentrations vary substantially by race/ethnicity.19,20
We tested associations of serum 25(OH)D concentration with incident CHD events in a large, community-based, multiethnic population of adults who were free of clinical cardiovascular disease at baseline. 25(OH)D status and CHD events were assessed using gold-standard methods. We evaluated whether associations of 25(OH)D with CHD differed among white, black, Chinese, and Hispanic participants.
The Multi-Ethnic Study of Atherosclerosis (MESA) is a multi-center, community-based prospective cohort study of clinical and subclinical cardiovascular disease.21 Each center’s institutional review board approved the study, and all participants provided informed consent. From 2000 to 2002, MESA enrolled 6814 adults aged 45 to 84 years from 6 field centers (New York and Bronx counties, New York; Baltimore and Baltimore County, Maryland; Forsyth County, North Carolina; Chicago, Illinois; St Paul, Minnesota; and Los Angeles, California). By design, MESA recruited a study population that was 38% white, 28% black, 22% Hispanic, and 12% Chinese. MESA excluded individuals who had prevalent clinical cardiovascular disease, defined as myocardial infarction, angina, stroke, transient ischemic attack, heart failure, atrial fibrillation, use of nitroglycerin, prior angioplasty, coronary artery bypass graft surgery, valve replacement, pacemaker or defibrillator implant, or any surgery on the heart or arteries.
We restricted our study population to 6476 participants for whom we measured serum concentrations of 25(OH)D at the baseline MESA examination. We excluded 6 participants with serum 25(OH)D concentration suggestive of high-dose vitamin D supplementation (>100 ng/mL; to convert to nanomole per liter, multiply by 2.496) and 34 participants without follow-up data, resulting in a final sample size of 6436 (94% of all MESA participants).
Sera samples were collected at the baseline MESA exam in 2000-2002, after an overnight fast. Samples were stored at −80°C until they were shipped to the University of Washington for analysis in 2011-2012. Total 25(OH)D (sum of 25-hydroxyvitamins D2 and D3) was measured using high-performance liquid chromatography–tandem mass spectrometry. Calibration was confirmed with National Institute of Standards and Technology’s standard reference material 972.22 The interassay coefficient of variation was 4.4% at 10.4 ng/mL with a lower limit of detection of 2.0 ng/mL for 25(OH)D3 and a coefficient of variation of 4.4% at 9.4 ng/mL with a lower limit of detection of 0.5 ng/mL for 25(OH)D2. 25-hydroxyvitamin D is stable during long-term storage at −80°C.23
At intervals of 9 to 12 months, a telephone interviewer contacted each participant to inquire about interim hospital admissions, cardiovascular outpatient diagnoses, and deaths. Self-reported diagnoses were verified via death certificate and medical record review, as previously described.24 Cardiovascular disease outcomes were adjudicated by the MESA events committee, which included cardiologists, physician epidemiologists, and neurologists.21 Incident CHD, used as the primary outcome for this study and other MESA studies,25 was assessed until May 2012 and defined as the first occurrence of any 1 of the following events: myocardial infarction, definite or probable angina, resuscitated cardiac arrest, or CHD death. Definite angina was defined as symptoms of typical chest pain or atypical symptoms, followed by 1 or more additional criteria, including coronary artery bypass graft surgery or other revascularization procedure, 70% or greater obstruction on coronary angiography, or evidence of ischemia by stress tests or by resting electrocardiogram. Probable angina required symptoms of typical chest pain or atypical symptoms, a physician’s diagnosis of angina, and medical treatment for it. As a secondary outcome, we used a more restrictive definition of CHD that included only incident myocardial infarction, CHD death, and resuscitated cardiac arrest, which we labeled hard CHD events.26
All covariates were ascertained at the baseline MESA examination, concurrent with 25(OH)D measurements. Participants completed self-administered questionnaires, interviewer-administered standardized interviews, and extensive in-person examinations yielding demographic and lifestyle characteristics, medical history, anthropometric measurements, and laboratory data. Race/ethnicity was characterized on the basis of participants’ responses to questions modeled from the year 2000 US Census. General health was self-reported on a questionnaire as excellent, very good, good, fair, or poor. Leisure-time physical activity was estimated as the total amount of intentional exercise performed in a usual week and measured in metabolic equivalent task–minutes. Participants were asked to report the average frequency of consumption of specific food items over the previous year using a 120-item food frequency questionnaire. Dietary vitamin D intake was estimated by multiplying the frequency and serving size for each food consumed by the vitamin D content of that food (Nutritional Data Systems for Research). Data regarding vitamin D supplement use were not available. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared. Diabetes was defined as a fasting blood glucose concentration of 126 mg/dL (to convert to mmol/L, multiply by 0.0555) or more or the use of insulin or oral hypoglycemic medications. Blood pressure was ascertained as the mean of the last 2 of 3 seated measurements. Total and high-density lipoprotein cholesterol were measured using the cholesterol oxidase method. Low-density lipoprotein cholesterol levels were calculated using the Friedewald equation. Intact serum parathyroid hormone was quantified using a 2-site immunoassay on a clinical analyzer.27 Glomerular filtration rate was estimated from serum creatinine using the Chronic Kidney Disease Epidemiology Collaboration equation.28 Chronic kidney disease was defined as a glomerular filtration rate less than 60 mL/min/1.73 m2.
Mean annual serum 25(OH)D concentration was estimated for each participant using a cosinor model previously developed and validated in MESA.20 The cosinor model assumes that 25(OH)D concentration follows a sine wave pattern across months of measurement. Using the observed mean amplitude of 25(OH)D excursion in the MESA population, each participant’s mean annual 25(OH)D concentration was estimated from the participant’s single baseline 25(OH)D measurement and the month during which it was obtained. Race is not included in the cosinor model because its inclusion did not improve prediction performance. Annualized 25(OH)D concentration was evaluated using clinically relevant categories that have been previously published: 30 ng/mL or more (reference group), 20-29 ng/mL, and less than 20 ng/mL.18,29,30 Because the distribution of serum 25(OH)D concentration and perhaps the functional form of its association with CHD risk was expected to vary by race/ethnicity group, we also evaluated annualized 25(OH)D using race-specific quintiles.
Participants were considered at risk of incident CHD events from the date of their baseline MESA examination until the first occurrence of the composite outcome or until their data were censored due to death from non-CHD cause (n = 217), loss to follow-up, or the end of available follow-up (May 2012). Functional forms of the race-specific associations of 25(OH)D with CHD risk were examined graphically using adjusted penalized smoothing splines with evenly spaced knots, among the inner 99% of annualized 25(OH)D concentrations.31 We used the Cox proportional hazards regression model to estimate the relative hazard of CHD, adjusting for covariates selected prior to analyses based on known or suspected biologic relationships. The first model included demographic data, including age, race/ethnicity, sex, and study site. The second model added likely potential confounding variables, including BMI (continuous variable), smoking status (never, former, current), educational attainment (high school, some college, completed college), total gross family income in the past 12 months, self-reported health status (poor, fair, good, very good, excellent), nutritional vitamin D intake (log-transformed continuous variable), and leisure-time physical activity level (log-transformed continuous variable). The third model added variables that may possibly confound or mediate the associations of interest, including diabetes status, chronic kidney disease status, systolic blood pressure, serum concentrations of high-density lipoprotein, low-density lipoprotein, triglycerides, parathyroid hormone, serum C-reactive protein (log-transformed), and the use of antihypertensive and lipid-lowering medications.
Approximately 5% or less of the study participants were missing data on education, income, physical activity, smoking status, serum C-reactive protein, systolic blood pressure, antihypertensive medication use, or lipid concentrations. For the regression analyses, these participants’ values were multiply imputed using chained equations.32 The multiple analyses over the imputations were combined using Rubin rules to account for the variability in the imputation procedure.33
Subgroup-specific hazard ratios (HRs) were calculated via linear combination of regression coefficients for main effect and cross-product terms. Interactions were tested by the Wald test of a product term for 25(OH)D concentration and race/ethnicity categories. Analyses completely stratified by race/ethnicity yielded similar results.
All analyses were conducted with Stata 11.2 (StataCorp). The nominal level of significance was defined as P < .05 (2-sided).
At baseline, the mean age was 62 years and 53% of participants were women. Mean (SD) serum 25(OH)D concentration was 25.5 (10.6) ng/mL and varied substantially by race/ethnicity: 30.1 (10.6) ng/mL for white participants (n = 2501), 26.7 (8.3) for Chinese participants (n = 784), 19.2 (9.0) for black participants (n = 1750), and 24.6 (9.4) for Hispanic participants (n = 1401) (P < .001). Participants with lower serum 25(OH)D concentrations were less likely to be white, were more likely to smoke or have diabetes or chronic kidney disease, and were characterized by larger body mass, higher serum parathyroid hormone concentration, and lower physical activity levels (Table 1). Race/ethnicity was associated with differences in baseline characteristics, particularly for the level of education completed, gross family income, self-reported health, prevalence of diabetes, and amount of intentional exercise (eTable 1 in Supplement).
During a median follow-up of 8.5 years (interquartile range, 7.6-8.6 years), there were 361 occurrences of the composite clinical end point (incidence rate: 7.38 events per 1000 person-years). The qualifying event was myocardial infarction for 139 participants, CHD death for 46 participants, resuscitated cardiac arrest for 13 participants, and angina for 163 participants (Table 2).
Unadjusted incidence rates and adjusted risks of incident CHD were higher with lower baseline serum 25(OH)D concentration (Table 3). However, we observed significant heterogeneity in the association of 25(OH)D with CHD risk by race/ethnicity, with P values for interaction < .05 across race/ethnicity groups for each model (Table 3). Lower serum 25(OH)D concentration was associated with significantly higher risks of CHD among white participants (fully adjusted hazard ratio [HR], 1.26 [95% CI, 1.06-1.49] per 10 ng/mL decrement in 25(OH)D concentration) and Chinese participants (HR, 1.67 [95% CI, 1.07-2.61]). However, there was no evidence of association among black participants (HR, 0.93 [95% CI, 0.73-1.20]) or Hispanic participants (HR, 1.01 [95% CI, 0.77-1.33]). Evaluation of the association of 25(OH)D with CHD risk using splines and race/ethnicity-specific 25(OH)D quintiles similarly revealed that significant associations were limited to white and Chinese participants (Figure and Table 4).
For the more restricted hard CHD outcome (n = 216 events), associations of low serum 25(OH)D concentration were of stronger magnitude for white and Chinese participants and were null for black and Hispanic participants (eTable 2 in Supplement). Similar results were observed when mean annual 25(OH)D concentration was replaced with untransformed 25(OH)D concentration and models were adjusted for season of blood draw, when participants with incident CHD events during the first 12 months of follow-up were excluded, and in analyses restricted to participants with good, very good, or excellent self-reported health statuses (eTables 3, 4, and 5 in Supplement).
In this multiethnic, community-based cohort of adults without clinical cardiovascular disease, low serum 25(OH)D concentration was associated with increased risk of adjudicated incident CHD events among white or Chinese participants but not among black or Hispanic participants. Differences in associations across race/ethnicity groups were consistent for both a broad and restricted definition of CHD and persisted after adjustment for known CHD risk factors.
We examined CHD events as our study outcome in part because low 25(OH)D concentration has been consistently associated with increased CHD risk in observational studies of white participants.2- 10 However, few of these studies included substantial numbers of multiracial participants. Among those that did, the number of multiracial participants was insufficient to test for racial heterogeneity (n = 332),5 the analyses were cross-sectional,7 or the outcome was limited to subclinical cardiovascular disease.9 Further analyses of racial differences in the associations of 25(OH)D with CHD are needed to confirm our results. Until such studies are available, results of studies testing associations of circulating 25(OH)D concentration with CHD or related outcomes in predominantly white populations should not be extrapolated to multiracial populations.
There are a number of possible explanations for the racial heterogeneity that we observed. Potential nonbiologic causes of heterogeneity include inadequate modeling in 1 or more racial groups, confounding that differs by race, and chance variation. However, results were robust evaluating race-specific 25(OH)D quintiles and splines, we are unable to discern characteristics that could conceivably cause differential confounding by race, and chance findings are unlikely—at least for black vs white comparisons—due to reasonable numbers of events in each race, the lack of any trend toward association in blacks, and the statistically significant global and black-vs-white P values for interaction. Therefore, our data suggest that biological differences explain much of the observed heterogeneity.
Prior studies testing associations of circulating 25(OH)D concentration with other health outcomes also generally support differing associations in black vs white populations. Specifically, associations of low 25(OH)D concentration with diabetes, fracture risk, stroke, and bone mineral density, observed in white individuals, were attenuated or absent among black individuals.19,34- 38 In addition, in a cohort study of the National Health and Nutrition Examination Survey (NHANES III)39 participants, low serum 25(OH)D concentration was associated with a significantly higher risk of all-cause mortality among white participants but not among the 3597 non-Hispanic black participants. In contrast, in the Health, Aging, and Body Composition (Health ABC)40 study (N = 2638), low 25(OH)D concentrations were associated with increased risk of all-cause mortality in both black and white participants. However, nearly one-third of the participants in the Health ABC cohort had cardiovascular disease at baseline, in contrast to the NHANES III study, in which fewer than 10% of participants had prevalent cardiovascular disease. Also, a majority of black participants in Health ABC had diabetes or hypertension, and all were 70 years or older. Similarly, significant associations of low 25(OH)D concentration with all-cause mortality were reported among both black and white participants in the Southern Community Cohort Study,41 although these associations were stronger in white participants. In this study, no data were presented on prevalent cardiovascular conditions. Together, these studies suggest that low circulating 25(OH)D concentration may be associated with a restricted subset of health outcomes in black populations. Alternatively, and perhaps more likely, confounding by chronic illness may inflate observed associations of 25(OH)D with broad health outcomes such as all-cause mortality in comorbid black and white populations, with true null results for black individuals apparent in studies in which confounding is minimized and outcomes are restricted to those most specifically linked with 25(OH)D deficiency.
In white populations, low 25(OH)D may lead to 1,25 dihydroxyvitamin D deficiency with resulting inappropriate activation of the renin-angiotensin system, dysregulation of immune cell functions, and failure to inhibit abnormal cell proliferation.13- 16 Differences in vitamin D metabolism may interrupt this sequence in black populations. Compared with white individuals, black individuals have higher circulating concentrations of 1,25-dihydroxyvitamin D, despite lower 25(OH)D concentrations.19 This suggests that increased activity of 25-hydroxyvitamin D-1-α-hydroxylase (CYP27B1) or reduced 1,25-dihydroxyvitamin D catabolism may allow black individuals to maintain adequate tissue-level 1,25-dihydroxyvitamin D in the setting of low circulating 25(OH)D, protecting against otherwise adverse cardiovascular effects of low 25(OH)D. Increased CYP27B1 activity in black individuals could be driven by increased parathyroid hormone or by racial differences in CYP27B1 affinity for substrate, and reduced 25-hydroxyvitamin D-24-hydroxylase (CYP24A1)–mediated vitamin D catabolism has been reported in black individuals with chronic kidney disease.19,42 Alternatively, vitamin D receptor affinity for vitamin D metabolites may vary by race.43 The distribution of at least 1 important vitamin D receptor polymorphism (FokI) is known to differ by race/ethnicity.44- 46 We recently observed that common genetic polymorphisms in the vitamin D receptor modify associations of serum 25(OH)D concentration with risk of a composite clinical outcome among older white adults, supporting the principle that genetic variation in the vitamin D receptor alters susceptibility to low 25(OH)D.47
To our knowledge, no previous study has evaluated associations of 25(OH)D with CHD events in Hispanic or Chinese populations. Our sample sizes for these groups were relatively small; thus we observed low numbers of events in these groups, particularly Chinese, so our results for these groups should be interpreted with caution. In addition, Hispanic MESA participants have diverse ancestry, and additional heterogeneity may be present within Hispanic participants.48 Further studies of 25(OH)D and health outcomes are needed in Hispanic and Chinese populations.
An important limitation of this observational study is the potential for confounding, as many unhealthy characteristics are linked with lower 25(OH)D concentrations. However, confounding is minimized in MESA because, by design, participants were free of self-reported clinical cardiovascular disease at baseline and because potential confounding variables were quantified in a high-quality manner. An additional limitation is that our study had only limited power to detect associations of small magnitude within individual racial and ethnic groups. As a result, it is possible that we missed potential nonlinear or threshold associations in black or Hispanic individuals. Serum 25(OH)D concentration was measured at only a single time point. Serum 25(OH)D concentrations have been reported to remain relatively stable over long periods,49,50 but misclassification of long-term 25(OH)D exposures could still bias results toward the null. We measured total rather than free or bioavailable 25(OH)D, and associations of free or bioavailable 25(OH)D with CHD may differ from associations of total 25(OH)D with CHD.51
Strengths of this study include the use of a large, multiethnic, community-based population; evaluation of adjudicated incident CHD events over relatively long follow-up time; use of an accurate 25(OH)D assay calibrated to National Institute of Standards and Technology standards; the estimation of mean annual 25(OH)D concentration from a single measurement to account for seasonal variation and reduce misclassification of the exposure; and measurement of exposure, outcomes and covariates in a standard and high-quality manner across race/ethnicity groups.
Well-powered clinical trials are needed to determine whether vitamin D supplements have causal and clinically relevant effects on the risk of CHD.12,52 Currently, at least 5 such trials are under way. One of these trials, the Vitamin D and Omega-3 Trial (VITAL),53 is targeting enrollment of a large multiracial study population, although power may be insufficient to determine whether effects vary by race even in this trial. Our study suggests that the risks and benefits of vitamin D supplementation should be evaluated carefully across race and ethnicity, and that the results of ongoing vitamin D clinical trials should be applied cautiously to individuals who are not white.
Corresponding Author: Ian H. de Boer, MD, MS, Division of Nephrology, Department of Medicine, Kidney Research Institute, University of Washington, PO Box 359606, Seattle, WA 98104 (email@example.com).
Author Contributions: Dr de Boer had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Robinson-Cohen, Hoofnagle, Ix, Tracy, Siscovick, Kestenbaum, de Boer.
Acquisition of data: Hoofnagle, de Boer.
Analysis and interpretation of data: Robinson-Cohen, Hoofnagle, Ix, Sachs, Tracy, Kestenbaum, de Boer.
Drafting of the manuscript: Robinson-Cohen.
Critical revision of the manuscript for important intellectual content: Robinson-Cohen, Hoofnagle, Ix, Sachs, Tracy, Siscovick, Kestenbaum, de Boer.
Statistical analysis: Robinson-Cohen, Sachs, Kestenbaum.
Obtained funding: Ix, Tracy, Siscovick, Kestenbaum, de Boer.
Administrative, technical, or material support: Hoofnagle, Tracy, Siscovick.
Study supervision: Hoofnagle, Kestenbaum, de Boer.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Hoofnagle reports receiving grant funding from the National Institutes of Health. Dr Ix reports receiving grant funding from the National Institutes of Health. Dr Tracy reports receiving grant funding and travel accommodations from the National Institutes of Health. Dr Siscovick reports receiving grant funding from the National Institutes of Health. Dr Kestenbaum reports receiving grant funding and payment for lectures from Amgen. Dr de Boer reports receiving grant funding from the National Institutes of Health and Abbott Laboratories. No other disclosures were reported.
Funding/Support: This study was supported by grants R01HL096875 and by N01-HC-95159 through N01-HC-95169 from the National Heart, Lung, and Blood Institute.
Role of the Sponsor: The funding organizations and sponsors had no role in the design and conduct of the study; the collection, management, analysis, and interpretation of the data; the preparation, review, or approval of the manuscript; and the decision to submit the manuscript for publication.
Additional Contributions: We thank the other investigators, the staff, and the participants of the MESA study for their valuable contributions. A full list of participating MESA investigators and institutions can be found at http://www.mesa.nhlbi.org.
Correction: This article was corrected on July 15, 2013, for typographical errors.