Vitamin D metabolism

A key concept in understanding vitamin D metabolism and vitamin D deficiency is to recognise that this compound is misnamed—it is not a vitamin (ie, an obligatory dietary component) but is a fat-soluble secosteroid produced in the skin from the action of ultraviolet light on 7-dehydrocholesterol. These principles are key to understanding who becomes vitamin D deficient in clinical practice—it is those who have little exposure of the skin to sunlight and those with dark skin. Obesity also reduces circulating vitamin D levels due to adipose tissue uptake of vitamin D, but typically does not produce severe vitamin D deficiency.3

While vitamin D can be obtained from the diet (in fish oils, egg yolk, butter, liver and in fortified foods), endogenous production is quantitatively much more important in most individuals. Cutaneous vitamin D₃ production is related to the intensity of ultraviolet B irradiation, so diminishes with increasing latitude. It is also diminished by skin pigmentation and by advancing age. When exposure to sunlight is sustained, there is increased production of inactive vitamin D metabolites, thus averting vitamin D intoxication. In plants, similar chemical processes result in the production of vitamin D₂, frequently used in supplements.4

Following its synthesis or ingestion, vitamin D and its metabolites circulate bound to vitamin D-binding protein, which is produced in the liver and usually has only approximately 5% of its vitamin D binding sites occupied. Hypoproteinaemic conditions are thus associated with reduced total vitamin D levels, although free levels may be normal. Vitamin D is not biologically active but undergoes 25-hydroxylation in the liver to form 25-hydroxyvitamin D [25(OH)D], which has some biological activity and is the principal circulating form of vitamin D. This transition is minimally regulated, in contrast to the second hydroxylation step in the kidney, which is regulated by parathyroid hormone, calcium, phosphate, FGF-23, calcitonin, growth hormone and IGF-1. This second hydroxylation produces 1,25-dihydroxyvitamin D [1,25(OH)₂D], which has a 1000-fold greater affinity for the VDR than 25(OH)D. With low 25(OH)D, there may actually be increases in 1,25(OH)₂D as homeostatic mechanisms act to maintain a stable vitamin D effect, and serum 1,25(OH)₂D is only reduced at very low levels of 25(OH)D (<10 nmol/l).5 Extrarenal synthesis of 1,25(OH)₂D occurs in some cells and may mediate the effects of 25(OH)D via an autocrine pathway . Both 25(OH)D and 1,25(OH)₂D are catabolised by 24-hydroxylation to inactive metabolites, under the regulation of 1,25(OH)₂D.

1,25(OH)₂D and 25(OH)D bind to the VDR, a member of the steroid hormone receptor super family, which then forms a heterodimeric complex with the RXR receptor. VDR has been said to be found in nearly all nucleated cells,1 although this has recently been challenged by Wang et al,6 who have questioned the specificity of antibodies against VDR used to date, and found that it is absent from cardiac muscle, aortic smooth muscle and endothelial cells. This is particularly relevant to the cardiovascular effects of vitamin D, although low vitamin D could still influence the vascular system indirectly (eg, via parathyroid hormone or effects on phosphate concentrations).

Vitamin D is classically regarded as a calcitropic hormone, in which its principal action is to regulate calcium absorption in the upper small bowel. In VDR-deficient animals, normal bone mineralisation can be restored by normalisation of serum calcium without replacing vitamin D, and normal calcium metabolism can be restored by selective expression of the VDR in the gut. There is also some human evidence that high calcium intakes abrogate the effects of vitamin D deficiency on calcium metabolism.7 8 1,25(OH)₂D also increases intestinal phosphate absorption, stimulates bone turnover through receptors in the osteoblast, and regulates its own production and degradation in the kidney. There is a growing amount of literature suggesting that it also acts on a variety of other tissues.1

Clinical correlates of circulating vitamin D levels

Epidemiological data are critical to the postulated link between low vitamin D and cardiovascular disease, so understanding the clinical determinants of circulating vitamin D levels is important in evaluating these data. 25(OH)D is the principal circulating vitamin D metabolite, and it is the entity that should be assessed when determining an individual's vitamin D status. Because ingested or endogenously produced calciferols are converted to 25(OH)D with very little regulation, serum levels of this metabolite accurately reflect both excess and deficiency states. The clinical correlates of 25(OH)D are entirely as would be predicted from the description of vitamin D physiology, set out above. Therefore, levels are higher in summer, in men, in supplement users and in those with fair skin, are positively related to serum albumin, time spent exercising and time outdoors, and are inversely related to obesity, smoking9 and age.3 10 11 Paradoxically, 25(OH)D does not always decline with increasing latitude,12 13 possibly because of sun avoidance in hot regions near the equator, and the widespread use of vitamin D supplements in the regions at high latitude.

What is a normal 25(OH)D?

The definition of the reference range for 25(OH)D, particularly its lower limit, has changed substantially over recent decades, and remains the subject of significant controversy. As for other analytes, normal values were initially defined from values found in apparently healthy populations, so the lower end of the range was frequently set between 25 and 35 nmol/l. This has since been progressively redefined to what is thought to be the optimal values of 25(OH)D, based on a variety of considerations. For skeletal health, the optimal value has variously been suggested to be that associated with the lowest levels of parathyroid hormone,14 with the greatest efficiency of intestinal calcium absorption,5 or with the lowest fracture risk in clinical trials.15 The appropriateness of each of these measures, and the definition of the optimal value based on these measures, have all been subject to disagreement. Others have turned to observational data showing associations between 25(OH)D and health measures, such as mortality, although this approach assumes a cause and effect relationship that may not be real. As a result, the lower band of the reference range for 25(OH)D has been suggested to be anywhere between 40 nmol/l16 and 120 nmol/l,17 with the consensus in Europe sitting at approximately 50 nmol/l and that in the USA at approximately 75 nmol/l.18 Others have defined degrees of 25(OH)D deficit, with ‘deficiency’ being less than 25 nmol/l (the range in which osteomalacia is sometimes observed) and 25–50 nmol/l representing ‘insufficiency’, although these terms are only used qualitatively in the present review. These differences have a profound effect on clinical practice, as acceptance of a higher 25(OH)D level as the lower limit of normal labels the majority of the population as being ‘vitamin D deficient’ and therefore in need of vitamin D supplementation. Therefore, in the USA in particular, an ‘epidemic’ of vitamin D deficiency has occurred and the supplementation of millions of apparently healthy individuals is now occurring, without any evidence from clinical trials to support the safety or efficacy of this practice. There is now a substantial industry investment in both the measurement of 25(OH)D and the provision of supplements, which provides vested interests that potentially influence ongoing debate on the critical scientific issues.19

Association of 25(OH)D with non-cardiovascular disease

A diverse range of pathologies has been found to be associated with lower vitamin D levels in observational studies. This includes cancer, infectious diseases, multiple sclerosis,20 diabetes and obesity. Some observational studies have demonstrated an association between low vitamin D and fracture, particularly hip fracture,21 although replacement with vitamin D alone has not been consistently shown to influence fracture frequency.22 Mortality has also been shown to be related to circulating 25(OH)D levels (figure 1).24 25

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Meta-analysis of observational studies of vitamin D and incidence of cardiovascular events and mortality. For the incidence studies, the pooled HR was1.54 (95% CI 1.22 to 1.95) and there was no heterogeneity across the studies. For mortality, the pooled HR was 1.83 (95% CI 1.19 to 2.80), although there was significant heterogeneity among these studies. RR greater than 1 indicate higher incidence in cohorts with lower vitamin D levels. Asterisks denote studies that demonstrated a statistically significant effect. From Grandi et al23 with permission.

While many of these associations are firmly established, causation is not. Individuals with any significant chronic illness are likely to spend less time exercising outdoors, and therefore manufacture less vitamin D in their skin. The frailty associated with illness is often also associated with keeping such individuals well covered up against the cold, further diminishing cutaneous sunlight exposure. Obese individuals show smaller increments in serum 25(OH)D after ultraviolet light exposure or oral dosing with vitamin D, suggesting that their lower vitamin D levels are related to the sequestration of the fat-soluble molecules in adipose tissue.3 Therefore, obesity is not likely to be a consequence of low vitamin D, and this conclusion is supported by the failure of vitamin D supplementation to influence body weight in randomised clinical trials.26 As a result, associations of low vitamin D with conditions known to be themselves associated with obesity (type 2 diabetes, hypertension) may simply be reflections of these patterns of disease association, rather than being causally linked. A further possible explanation for the link between 25(OH)D levels and various pathologies is the recent demonstration of a 40% decline in 25(OH)D levels in the days following surgery (knee arthroplasty).27 These changes were not fully reversed 3 months later. By this mechanism, many inflammatory conditions might cause reduced levels of 25(OH)D resulting in disease associations that will not be corrected by vitamin D supplementation. A common finding in observational studies is thus that cohorts with lower vitamin D levels have a higher prevalence of many risk factors for morbid events.28

Complementing the epidemiological data are a wealth of laboratory studies that document effects of 1,25(OH)₂D on biological processes, which could underpin an aetiological role for vitamin D in some of these conditions.1 29 There is evidence that 1,25(OH)₂D regulates cell differentiation and proliferation, suggesting it could play a role in cancer prevention. 1,25(OH)₂D has also been shown to regulate immune cell function, suggesting roles for it both in cancer prevention and in the body's response to infection.29 1,25(OH)₂D also influences hormone secretion, in particular that of insulin, providing a possible mechanism by which low vitamin D could contribute to the pathogenesis of diabetes.30 31

Association of 25(OH)D with cardiovascular disease

There is extensive evidence from laboratory studies to suggest that vitamin D might influence the cardiovascular system at many sites1 (figure 2), although these studies will need to be re-evaluated in the light of the recent suggestion that VDR might not be present in all these target cells.6 Its direct effects on the arterial wall may protect against atherosclerosis through the inhibition of macrophage cholesterol uptake and foam cell formation, reduced vascular smooth muscle cell proliferation, and reduced expression of adhesion molecules in endothelial cells.32–34 In addition, the juxtaglomerular cells of the nephron are a vitamin D target cell, and direct inhibition of renin expression by 1,25(OH)₂D has been confirmed in vitro.35 As a result, VDR-deficient mice develop high renin hypertension, cardiac hypertrophy and increased thrombogenicity.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Possible mechanisms for actions of 1,25-dihydroxyvitamin D [1,25(OH)₂D] on the cardiovascular system. PAI-1, plasminogen activator inhibitor 1; BNP,brain natriuretic peptide. From Bouillon et al1 with permission.

Observational studies in humans, have found an association of vitamin D insufficiency with increased arterial stiffness and endothelial dysfunction in the conductance and resistance blood vessels.36 There is also evidence of associations with hypertension, including a study in adolescents,37 although this could be confounded to some extent by the effects of obesity.38–40 A recent meta-analysis of observational studies reported an OR for hypertension of 0.84 (95% CI 0.78 to 0.90) associated with a 40 nmol/l higher blood 25(OH)D concentration (approximately 2 SD).41 Grandi23 and others42 have recently summarised the observational data linking circulating 25(OH)D levels with cardiovascular events (figure 1). From three cohort studies and one nested case–control study, two found a significantly increased risk of vascular events associated with low 25(OH)D. Cardiovascular mortality results were available from five cohort studies, three of which reported a significant association.

We have recently examined the associations of 25(OH)D with cardiovascular events in a prospective cohort study.43 A total of 1471 normal postmenopausal women had a baseline measurement of 25(OH)D and were then followed for 5 years. In women with baseline 25(OH)D less than 50 nmol/l, the HR for stroke was 1.7 and that for myocardial infarction was 1.5. However, there were significant differences in cardiovascular risk factors at baseline in the women with lower vitamin D levels. Fat mass was 10% higher, physical activity was less, and these women were more likely to have a past history of ischaemic heart disease or dyslipidaemia. As a result, their calculated cardiovascular risk was significantly higher at baseline than was that of women with 25(OH)D greater than 50 nmol/l. When adjustment was made for these baseline differences, there was no longer a significant excess of cardiovascular events during follow-up (HR for composite endpoint of myocardial infarction, stroke or sudden death 1.2, 95% CI 0.8 to 1.8), suggesting that the apparent increase in risk in those with low 25(OH)D is related to its associations with these other risk factors, rather than to a direct biological effect of vitamin D.

Vitamin D effects on cardiovascular endpoints in trials

Until recently, only one randomised controlled trial of any size had examined the effect of supplementation with vitamin D alone on cardiovascular event rates.44 Two thousand six hundred and eighty-six men and women with a mean age of 75 years at baseline were randomly assigned to receive 100 000 units of oral vitamin D₃ or placebo every 4 months over 5 years. The RR for events during follow-up were: ischaemic heart disease 0.94 (95% CI 0.77 to 1.15), cerebrovascular disease 1.02 (95% CI 0.77 to 1.36) and cardiovascular mortality 0.84 (95% CI 0.65 to 1.10). A number of other studies have assessed the effect of vitamin D co-administered with other interventions, usually calcium, and these data have recently been meta-analysed (figure 3).45 The RR for myocardial infarction and stroke, respectively, were 1.02 and 1.05, providing no suggestion of a vasculoprotective effect. However, this meta-analysis is dominated by data from the Women's Health Initiative, a study of a calcium plus vitamin D intervention. This trial permitted self-administration of calcium and vitamin D, and more than half of participants were doing this at randomisation with an increase to two-thirds subsequently. A recent re-analysis of this study has shown that personal use of calcium supplements significantly interacted with the cardiovascular effects of the study intervention.46 When the analysis was restricted to the 16 000 women not using personal calcium supplements at randomisation, the HR for myocardial infarction was 1.22 (1.00 to 1.50) and that for stroke was 1.17 (0.95 to 1.44), similar to the risks previously found with calcium monotherapy.47 Therefore, in the context of individuals receiving calcium supplements, there is no suggestion of a cardioprotective effect from the co-administration of vitamin D, and calcium supplements may mask the effects of vitamin D on cardiovascular events.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Meta-analysis of randomised controlled trials of vitamin D (with or without calcium) on myocardial infarction and stroke. No statistically significant effects were found. From Elamin et al45 with permission.

Very recently, the RECORD investigators published cardiovascular event data from the original study (intervention with calcium and/or vitamin D over 24–62 months) plus a further 3 years of post-trial follow-up.48 This shows no statistically significant effects, but the HR for vascular death in those randomly assigned to vitamin D was 0.91 (0.79 to 1.05) in the intention-to-treat analysis, and 0.76 (0.49 to 1.40) in analyses adjusted for treatment compliance. In contrast, the comparable figures for randomisation to calcium supplements in the same study were 1.07 (0.92 to 1.24) and 1.43 (0.75 to 7.61), consistent with possible differences in vascular effects between these two agents. The vitamin D results from RECORD are similar to Trivedi, and suggest either that there might be a small positive effect that was not detected because the trials were not primarily designed to assess cardiovascular endpoints and were underpowered (reflected in the wide CI), or that vitamin D supplementation does not prevent cardiovascular events. However, any effect is small, if it exists, and will require substantial trials to determine whether it is real. Such studies are underway (VITAL in the USA and ViDA in New Zealand, details in Scragg).49

Low 25(OH)D levels have also been found to be associated with hypertension; however, meta-analysis of 10 randomised trials showed a weighted mean difference in systolic pressure of −1.9 mm Hg (−4.2, 0.4) and no change in diastolic pressure following vitamin D supplementation (figure 4).42 Similarly, meta-analysis of the effects of vitamin D supplements on circulating cholesterol fractions have failed to demonstrate significant effects.45

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Meta-analyses of the effect of vitamin D supplementation on change in blood pressure. Data are the weighted mean difference and 95% CI for change in blood pressure from vitamin D supplementation versus placebo. Studies are arranged according to baseline blood pressure. The circle sizes are proportional to the study size. The black diamond represents the primary meta-analyses, and the other diamonds represent sensitivity and subgroup analyses. Dashed lines indicate studies for which SE were not reported. From Pittas et al42 with permission. * The 95% CI was estimated from the reported full range of changes in blood pressure. † The 95% CI is in the circle.‡ The 95% CI was estimated from the reported interquartile ranges of changes in blood pressure.

Clinical considerations

If a doctor is concerned about the vitamin D status of their patient, how should they proceed? It would seem logical to measure serum 25(OH)D and then treat with calciferol if necessary, with appropriate follow-up blood tests. However, measuring 25(OH)D can be difficult and expensive. There is substantial variation in results between assays and between laboratories,50–52 with some immunoassays giving results differing by up to 40% from the correct value.51 Furthermore, in some countries, a single measurement can cost substantially more than the annual cost of vitamin D supplements for an individual. These considerations have caused many clinicians dealing with high-risk groups (the frail elderly, veiled women, individuals with dark skins living in temperate climates) to treat without undertaking 25(OH)D measurement. Routine measurement of 25(OH)D greatly increases the cost of managing vitamin D status, adversely impacting on its cost-effectiveness.53

If treatment is required, regimens involving 500–1000 units per day or 50 000 units per month will usually achieve serum levels of 25(OH)D greater than 50 nmol/l. These doses have an unblemished safety record. Advocates of higher 25(OH)D levels encourage daily doses of 2000 or more units. The recent finding of increased falls and fractures with a supplement regimen that achieved mean levels of approximately 120 nmol/l,54 and the U-shaped relationship between 25(OH)D and mortality,24 55 cause many authorities to question the wisdom of pushing levels to more than 100 nmol/l.

At present, there is insufficient evidence to support vitamin D supplementation as a way of improving cardiovascular outcomes. However, many cardiovascular patients are frail and immobile and are at risk of markedly reduced vitamin D levels and osteoporosis.56 57 Supplementation of such patients is justified to prevent very low levels of 25(OH)D, with their sequelae of musculoskeletal pain, myopathy and accelerated bone loss.

Implications for interpretation of future research

The relationship between vitamin D and non-skeletal comorbidity has highlighted what perhaps may be a fundamental issue for research. The research in this field has largely been driven by the results of very large observational cohort studies. These datasets are now stored electronically, and often contain large numbers of variables for very large numbers of patients. Automated, inexpensive, laboratory assays are also now widely available, offering researchers opportunities to measure many analytes in these cohorts. Taken together, these studies have the power to detect small effects that are highly statistically significant, may potentially be clinically relevant, although may also be false positives or result from confounding. Such data are hypothesis generating because they do not provide evidence to support a causal relationship and require confirmation in suitably designed clinical trials. As with vitamin D and cardiovascular disease, it is usually premature to change clinical practice on the basis of observational studies alone, with or without supportive preclinical data.