Vitamin D and Cardiovascular Disease: An Appraisal of the Evidence

Vitamin D Production and Homeostasis

Vitamin D (calciferol) is a term that refers to a group of lipid-soluble compounds with a 4-ringed cholesterol backbone. In the skin, pro–vitamin D is photoisomerized to vitamin D₃ (cholecalciferol) by sunlight and ultraviolet light. The other major source of vitamin D is intestinal absorption. Vitamin D₃ is then transported to the liver, where it is hydroxylated to 25-hydroxyvitamin D (25OHD), which comprises both 25OHD₂ and 25OHD₃. 25OHD then travels to the kidney, where it is further hydroxylated to 1,25-dihydroxyvitamin D [1,25(OH)₂D or calcitriol] (15), the physiologically active form of vitamin D (16, 17). The most representative measure of vitamin D status is the serum 25OHD concentration (18, 19). Serum 25OHD is an excellent marker of vitamin D sufficiency, because it reflects the total stored quantity from both endogenous and exogenous sources (18). As serum 25OHD concentrations decrease, parathyroid hormone concentrations increase and positively influence the conversion of 25OHD to 1,25(OH)₂D, which subsequently maintains normal intestinal absorption of calcium. Therefore, 1,25(OH)₂D is not representative of the total body storage of vitamin D, because serum calcium and 1,25(OH)₂D concentrations will be normal or slightly increased during vitamin D deficiency, owing to secondary hyperparathyroidism (18, 20).

Risk Factors for Vitamin D Deficiency

Risk factors for developing vitamin D deficiency, or lower serum vitamin D concentrations, include an age >65 years (21), dark skin pigmentation, obesity (from storage in adipose tissue (22)), kidney and/or liver disease (23), disorders affecting fat absorption (e.g., celiac disease, Crohn disease, ulcerative colitis, some types of bariatric surgery), and end-organ insensitivity to 1,25(OH)₂D. In addition, 25OHD deficiency is also known to be related to environmental variables that lead to decreased exposure to ultraviolet light, such as institutionalization, decreased outdoor physical activity, and frailty (24).

Potential Mechanisms for an Association between Vitamin D Deficiency and CHD

Vitamin D receptors (VDRs) have been identified in many tissues, including vascular smooth muscle cells (25), cardiomyocytes, and coronary arteries (26, 27). Given the presence of VDRs in the vascular system, including the coronary arteries, there are several biologically plausible pathways through which vitamin D could lead to improved cardiovascular health. Activation of the VDRs, for instance, has been shown to inhibit vascular smooth muscle cell proliferation, which is believed to be cardioprotective (28). Some studies have associated higher 25OHD concentrations and/or vitamin D supplementation with a systemic antiinflammatory state via effects on interleukins, C-reactive protein, and antiinflammatory cytokines—a milieu which, again, is believed to foster cardioprotection (29–31). Vitamin D may control blood pressure through its regulatory effects on the renin–angiotensin–aldosterone system (32). Limited research has suggested that vitamin D supplementation can decrease the incidence of impaired glucose tolerance and diabetes mellitus (33, 34), along with improving values for lipid parameters (35). Furthermore, the results of various studies have suggested a link between vitamin D and a lower likelihood of autoimmune conditions such as rheumatoid arthritis (36), diabetes (both type 1 and type 2) (37, 38), and multiple sclerosis (39).

Observational Data

Much of the excitement concerning a correlation between vitamin D deficiency and CHD stems from observational data. For example, Giovannucci et al. (9) followed 18 000 healthy male participants for 10 years. Individuals with vitamin D deficiency, defined as serum 25OHD concentrations ≤15 ng/mL (≤38 nmol/L), had a greater risk of a myocardial infarction than men with 25OHD concentrations ≥30 ng/mL (≥75 nmol/L) (9), with a relative risk of 2.42 (95% CI, 1.53–3.84; P < 0.001). The Framingham Offspring Study, a prospective analysis of 1739 individuals with 25OHD concentrations <15 ng/mL (<38 nmol/L), obtained an adjusted hazard ratio for a first-incident cardiovascular event of 1.6 (95% CI, 1.11–2.36; P = 0.01). Participants who had hypertension along with 25OHD deficiency had a hazard ratio of 2.1 (95% CI, 1.3–3.5; P = 0.003) for a first cardiovascular event (18). A metaanalysis of prospective but observational studies of 25OHD and cardiovascular events demonstrated a generally linear, inverse association up to 24 ng/mL (60 nmol/L) but revealed no further reductions (i.e., a threshold effect) at higher 25OHD concentrations (40).

Limitations of the Observational Data

Despite the considerable data demonstrating an association between vitamin D deficiency and poor cardiovascular outcomes, caution is advisable in interpreting the data from observational studies. Confounding by other lifestyle factors and a “healthy user” bias in nonrandomized studies may play a role in the current evidence suggesting an association. For example, age must be carefully controlled in analyses, because older age (21) increases both the risk of vitamin D deficiency and the risk of myocardial infarction (41). Decreased dietary intake and poor nutritional status can each lead to vitamin D nutritional deficiency. The lower intake could be due to other disease(s) or general malnutrition, either of which could increase the likelihood of CHD. Decreased exposure to ultraviolet light can be due to less outdoor activity/exercise and hence lead to an increased risk of CHD and low 25OHD. An additional confounding risk factor is obesity, which both increases the risk of CHD and lowers 25OHD concentrations because 25OHD can be sequestered in adipose tissues (22).

Few observational studies are able to adjust fully for these confounding factors. Therefore, all of these risk factors, which are more likely to be associated with low 25OHD, may confound the relationship between 25OHD and CHD in nonrandomized studies. As we pointed out earlier, vitamin D has been associated with a systemic antiinflammatory milieu. Although relevant pathways may include a beneficial interaction between vitamin D and C-reactive protein, interleukins, and/or cytokines (29–31), vitamin D deficiency has also been suggested to be a direct consequence of an inflammatory condition or state (42).

Randomized Controlled Trials

Few prospective randomized clinical trials evaluating the effects of vitamin D supplementation on CHD have been conducted, and currently none of the prospective trials have prespecified CHD as the primary outcome (43–45). Among the sparse randomized trials that have assessed CHD—or CHD risk factors—as a secondary or tertiary outcome, no correlation has been identified for CHD, and few have been found for CHD risk factors (Table 1) (46–58). In a study of 327 men and women older than 65 years, individuals who received vitamin D₃ actually had an increased risk of coronary death (P < 0.001) (58). In a double-blind, placebo-controlled, randomized clinical trial in the UK conducted among 2686 men and women 65–85 years of age, participants received 100 000 IU of supplemental vitamin D₃ every 4 months (equivalent to approximately 833 IU daily) for 5 years (46). No beneficial CHD effect could be attributable to vitamin D (46). Additional results from the Women's Health Initiative suggested that postmenopausal women who received 400 IU/day of oral vitamin D₃ combined with 1000 mg/day of calcium had no reduction in their risk of CHD events or stroke (50). In additional prospective trials (from subanalyses of the Women's Health Initiative), calcium and vitamin D₃ supplementation was not found to improve blood pressure (49) or coronary artery calcium scores (53). Furthermore, there was no decrease in incident hypertension (59) or any prevention of or improvement in the metabolic syndrome or diabetes (47). An 8-week prospective trial of 151 male and female vitamin D–deficient adults randomized to receive 50 000 IU of vitamin D₃ or placebo found no improvement in lipid parameters (48). Two small prospective trials that evaluated endothelial function produced mixed results, with one showing no effect (51) and the other showing a short-term improvement in stroke patients with well-controlled hypertension; however, the effect was not sustained by the completion of the 16-week study (52).

Several small prospective studies have shown some improvement in CHD risk factors (55) and inflammation (30, 56). Additional data have shown no effect on glycemic control, however (54, 57). A recent prospective randomized, double-blind, placebo-controlled clinical trial assessed the change in systolic and diastolic blood pressures in a healthy black population randomized to oral placebo or to 1000, 2000, or 4000 IU/day vitamin D₃ for 3 months (60). The results of this study revealed a 1.4-mmHg decrease in systolic blood pressure for each additional 1000 IU/day increase in the vitamin D₃ dose (P = 0.04). Although the investigators found no statistically significant effect of oral vitamin D₃ on diastolic blood pressure, their study did reveal a 0.2-mmHg decrease in systolic blood pressure for each 1-ng/mL (2.5-nmol/L) increase in 25OHD (P = 0.02). Despite this significant effect of oral vitamin D₃ on systolic blood pressure, there appeared to be a threshold effect, with individuals receiving 2000 IU/day and 4000 IU/day of vitamin D₃ having similar results. Furthermore, those with baseline 25OHD concentrations ≥20 ng/mL (≥50 nmol/L) experienced little benefit from supplementation, compared with participants with baseline 25OHD concentrations <20 ng/mL (<50 nmol/L), who had a 2.2-mmHg decrease in systolic blood pressure (P = 0.03). In addition, adjustment for baseline differences in blood pressure attenuated the study's findings. Whereas this trial was designed to assess change in blood pressure, it is noteworthy that most other trials were designed to assess bone health. Cardiovascular outcomes were not prespecified primary end points for most previous studies.

Metaanalyses of Randomized Trials

Several metaanalyses have evaluated both mortality and CHD risk with respect to their relationship to vitamin D supplementation. A metaanalysis of 18 randomized clinical trials including 57 311 individuals was published in 2008 (61). This analysis revealed a statistically significant 7% decrease in all-cause mortality in participants who received vitamin D supplementation (61); however, a subsequent metaanalysis by Rejnmark et al. (62), which included 24 randomized trials of patients receiving vitamin D supplementation, with or without oral calcium, showed similar results, but with the following important variation. Although the patients who received supplements of vitamin D and calcium had a similar 7% reduction in all-cause mortality, as was seen in the earlier study (61), the patients who received vitamin D alone did not have a significant decrease in mortality. These results raise a number of questions, including the role calcium may have in any potential beneficial effect related to vitamin D supplementation. In considering a role of supplemental vitamin D, a recent systematic review and metaanalysis identified randomized trials published through August 2010 in which patients were randomized to vitamin D supplementation or to no treatment (63). The outcome measures of interest included mortality, cardiovascular events, and CHD risk factors. A total of 51 studies were eligible. Of note is that the analysis was unable to identify statistically significant differences in any of the outcomes, including myocardial infarction, stroke, all-cause mortality, or such CHD risk factors as lipid fractions, glucose, and systolic and diastolic blood pressures. Most of the trials tested relatively low vitamin D doses, however.

Although the association between 25(OH)D deficiency and obesity is not new (22), the results of a recent large metaanalysis suggest that a higher body mass index leads to lower plasma 25(OH)D concentrations, implying a causative relationship. In contrast, lower 25(OH)D concentrations did not appear to lead to a high body mass index (64). If these findings are confirmed, strategies to decrease obesity could also lower the prevalence of 25(OH)D deficiency.

Current Recommendations

The Institute of Medicine (IOM) (44) and the Agency for Healthcare Research and Quality (AHRQ) (43) reviewed the literature related to vitamin D and health-related outcomes. Both concluded that although sufficient evidence existed to support a role for calcium and vitamin D related to skeletal health, evidence supporting effects on non–bone-related health outcomes was lacking (44, 45). The IOM recommended a dietary allowance of 600 IU/day for individuals 1–70 years of age, with 800 IU/day recommended for individuals older than 70 years (18, 61). The IOM and AHRQ reports have generated some controversy, and some investigators have stated that higher dietary allowances should be encouraged. The International Osteoporosis Foundation (IOF), for instance, recommends 800–1000 IU/day as the mean supplemental dose to achieve an appropriate plasma 25OHD concentration (65). The IOF adds that individuals at higher risk may require doses of up to 2000 IU/day to reach an appropriate concentration (65). The National Osteoporosis Foundation (NOF) recommends 400–800 IU/day of oral vitamin D₃ for adults <50 years of age and 800–1000 IU/day for those >50 years (66). In line with the IOF guidelines, the NOF has clarified that some people may need higher oral vitamin D₃ doses, with 4000 IU/day noted as an upper limit of safety (66). Similarly, the Endocrine Society's clinical guideline recommends at least 600 IU/day for adults 19–50 years of age and 600–800 IU/day for those older than 50 years. They also point out that doses of 1500–2000 IU/day may be required for all adults to consistently raise plasma 25OHD concentrations to >30 ng/mL (>75 nmol/L) (67). Although controversy surrounds the 25OHD concentration to use as a cutpoint for vitamin D deficiency or insufficiency, the IOM suggests that a serum 25OHD concentration of at least 20 ng/mL (50 nmol/L) will meet the vitamin D requirements for ≥97.5% of the US and Canadian populations (68). As we discussed above, results of the recent study assessing oral vitamin D₃ and blood pressure (60) likewise support a 25OHD concentration ≥20 ng/mL as being adequate; only individuals with 25OHD concentrations <20 ng/mL at baseline experienced a significant improvement in systolic blood pressure with vitamin D₃ supplementation (60).

Laboratory Testing

The current assays available for 25OHD testing include antibody-based methods and liquid chromatography. The methodologies for plasma 25OHD analyses, however, have changed greatly over the years. The early testing methods used competitive protein-binding assays, which were difficult to perform and lacked consistency. The introduction of early liquid chromatography techniques in the 1970s allowed, for the first time, the ability to detect 25OHD₂ and 25OHD₃ separately. As liquid chromatography assays were being refined in the 1980s, antibody-based assays were introduced. More recently, antibody assays have been modified to accommodate the automated multiwell plate format, which has made these assays quite popular. A notable drawback is the inability to distinguish between 25OHD₂ and 25OHD₃. It is also noteworthy that much of the past research related to plasma vitamin D concentrations were performed with antibody-based techniques. The variety of testing methods and questions about the reliability of testing also add to the complexity of interpreting previous research. Most recently, the liquid chromatography method has advanced substantially with the incorporation of a tandem mass spectrometer—the liquid chromatography–tandem mass spectrometry (LC-MS/MS) technique. This improvement produces data with very high specificity and sensitivity, along with outstanding reproducibility (44).

Because a majority of the previous data were produced with the antibody-based assays, it is important to point out the concern about inconsistencies between testing methods. Research that has incorporated interlaboratory comparisons suggests a high and concerning degree of variation (44). These results have led to external quality-assurance programs, including NIST reference standards (69, 70), which use a “validated” LC-MS/MS technique for calibration (44). A Standard Reference Material and a calibration solution are now available through the NIST to help assure the accuracy and reliability of 25OHD measurements (44).

A study by Lia et al. compared DiaSorin LIAISON results with those for the LC-MS/MS assay (selected as the nominal gold standard) obtained with same participant samples and found a vitamin D deficiency rate that was 16% to 29% higher, respectively, based just on the laboratory test used. Furthermore, the DiaSorin RIA has been shown to produce lower serum 25OHD values than the LC-MS/MS method (71). The LC-MS/MS assay, which is considered inherently more accurate (71) with its high sensitivity, high specificity, and better reproducibility, tends to produce values that are slightly higher than obtained with the RIA techniques (71). Some have argued that LC-MS/MS results may need to be adjusted downward according to a mathematical formula, whereas others have suggested that DiaSorin RIA results may need to be corrected upward (71). Regardless, the value of using consistent and reliable laboratory testing is therefore of paramount importance, and LC-MS/MS is currently the preferred laboratory assay.

For quality assurance, it is critically important that test measurements be performed on standardized samples, with the inclusion of NIST samples and split-replicate samples for laboratory assessment. Such protocols allow laboratories to compute means, SDs, and CVs. Samples should be protected from direct sunlight to ensure the accuracy and precision of assays (30, 44, 72). The new NIST reference standards offer hope that 25OHD measurements can achieve improved accuracy and reliability and thus diminish the variation between tests and laboratory centers seen in the past (44).

Future Research Directions

Although our understanding of vitamin D deficiency and its ramifications is rapidly expanding, there is still much to learn. Several large-scale randomized trials of moderate to high dosages of vitamin D supplementation in cardiovascular disease (CVD) prevention are being conducted in the US and throughout the world. As one example, the Vitamin D and Omega-3 Trial (VITAL) (J.E. Manson, principal investigator) is a randomized, double-blind, placebo-controlled clinical trial of 20 000 US men and women older than 50 years. It tests 2000 IU/day of oral vitamin D₃ and ω-3 fatty acid supplements in a 2 × 2 factorial design, with CVD and cancer as prespecified primary outcomes (73). Results are expected in 2017. While we await the results of VITAL and other ongoing randomized trials of vitamin D, it is also important to point out that many of the trials will analyze the effects of vitamin D₃ supplementation in a general population of patients with and patients without 25OHD deficiency. There may be value, therefore, in stratifying by baseline concentrations or subsequently analyzing results in vitamin D–deficient patients (74).

Although data have suggested that the VDR concentration may be inversely correlated with the degree of coronary artery atherosclerosis (26), the link to or relationship with the plasma 25OHD₃ concentration remains unclear. A recent study suggested that individuals with the highest plasma 25OHD₃ concentration and the lowest VDR abundance had the greatest degree of coronary artery atherosclerosis (27), but these findings need corroboration. If confirmed, they suggest a therapeutic-window phenomenon, that high 25OHD₃ concentrations may be detrimental above an upper threshold (18). Data have also suggested a strong link between 25OHD₃ deficiency and race/heritability (75–79). A recent study (76) indicated that individual differences in 25OHD concentrations have both genetic and environmental associations and that the relative contributions to CHD outcomes remains unclear. In this particular study, the variation attributable to genetics was predominantly demonstrated in the winter when ultraviolet exposure was minimal but was not apparent in the summer months, implying that environmental factors (mostly sun exposure) may compensate for vitamin D deficiency related to genetics (77).

As the review above indicates, we are clearly in need of well-designed and adequately powered prospective randomized trials with vitamin D supplementation and CHD or CHD biomarkers as primary outcomes. It will be important to determine whether supplementation makes a clinically meaningful difference for CHD and whether the baseline 25OHD concentration modifies the response. Such trials will help to elucidate whether low vitamin D concentrations represent a marker for other processes, are indicators of genetic predisposition to disease, or are causally related to risk. On the assumption that vitamin D supplementation truly is of value for preventing CVD, what is the optimal dose, and what role does calcium supplementation play in the equation? Furthermore, does a therapeutic-window phenomenon exist such that not only lower but also higher serum concentrations can be detrimental? In contrast, an examination of the completed randomized, prospective placebo-controlled trials (Table 1) shows that a major weakness of many of these trials is that they tested lower vitamin D doses than currently postulated to be of benefit for extraskeletal outcomes. What role does the VDR play, would a VDR agonist be of benefit, or are there ways to prevent VDR loss and hence delay the onset of coronary atherosclerosis? As we await the results of ongoing research, we eagerly await answers to these questions.