The Role of Vitamin D in the Prevention of Type 2 Diabetes: To D or Not to D?

Abstract

Evidence on biological plausibility from mechanistic studies and highly consistent data from observational studies raise the possibility that optimizing vitamin D status may reduce the risk of type 2 diabetes. However, the observational nature of cohort studies precludes a definitive assessment of cause and effect because residual confounding or reverse causation cannot be excluded. Confounding is especially problematic with studies of vitamin D because blood 25-hydoxyvitamin D concentration is not only an excellent biomarker of vitamin D status, reflecting intake or biosynthesis, but also an excellent marker of good overall health. Results from underpowered trials and post hoc analyses of trials designed for nondiabetic outcomes do not support a role of vitamin D supplementation for prevention of type 2 diabetes among people with normal glucose tolerance. Whether vitamin D supplementation may have a role in the prevention of diabetes in high-risk populations remains to be seen. Adequately powered, randomized trials in well-defined populations (e.g., prediabetes) are ongoing and expected to establish whether vitamin D supplementation lowers risk of diabetes.

Optimizing vitamin D status may reduce the risk of type 2 diabetes. Trials in well-defined populations will establish whether vitamin D supplementation lowers the risk of diabetes.

Despite several advances in pharmacotherapy for type 2 diabetes, the increasing burden of the disease highlights the need for innovative and cost-effective prevention approaches. It is estimated that approximately one-third of adults in the United States are at increased risk of developing diabetes based on their having prediabetes, which is defined by the American Diabetes Association as impaired fasting glucose, impaired glucose tolerance, or abnormal hemoglobin A1c (HbA1c) (1). Epidemiologic data suggest that 9 of 10 cases of type 2 diabetes are attributed to modifiable lifestyle factors (2). In clinical trials, changes in physical activity and diet aiming at weight loss reduce the risk of diabetes (3). However, long-term weight control—especially outside of a clinical trial setting—is challenging. Moreover, even after successful weight loss, residual risk of diabetes remains elevated (∼40% to 50%) (2); therefore, complementary approaches to weight loss are needed. Over the last decade, suboptimal vitamin D status has emerged as a probable risk factor for type 2 diabetes, and vitamin D supplementation has been hypothesized as a potential intervention to lower diabetes risk. This review synthesizes the available evidence and highlights limitations and gaps in our knowledge.

Biologic Plausibility of an Association Between Vitamin D and Type 2 Diabetes

Type 2 diabetes is characterized by impaired pancreatic β-cell function, insulin resistance, and systemic inflammation (4, 5), and there is evidence that vitamin D modulates these mechanisms.

Vitamin D and pancreatic β-cell function/insulin secretion

Several lines of evidence support a role for vitamin D in pancreatic β-cell function and regulation of insulin secretion. In in vitro and in vivo studies, vitamin D deficiency impairs glucose-mediated insulin secretion in rat pancreatic β-cells (6–11), whereas vitamin D supplementation restores insulin secretion (6, 9, 10, 12). Vitamin D may have a direct effect on β-cell function mediated by binding of the circulating active form, 1,25-dihydroxyvitamin D [1,25(OH)₂D], to the vitamin D receptor, which is expressed in pancreatic β-cells (13, 14). Furthermore, mice lacking a functional vitamin D receptor show impaired glucose-stimulated insulin secretion, attributed to a reduction in insulin biosynthesis (15). Importantly, activation of vitamin D may occur within the β-cell by the 25-hydroxyvitamin D–1α-hydroxylase enzyme (CYP27B1), which is expressed in β-cells, thereby allowing for a paracrine effect of circulating 25-hydroxyvitamin D [25(OH)D] (16).

Regulation of insulin secretion by vitamin D appears to be independent of prevailing calcium concentration (6–10); however, because insulin secretion is a calcium-dependent process (17, 18), the effects of vitamin D may be indirectly mediated via regulation of calcium flux through the β-cell (19). Vitamin D also regulates calbindin, a cytosolic calcium-binding protein found in many tissues, including β-cells (13, 20), which controls the rate of insulin secretion via regulation of intracellular calcium (21). Finally, vitamin D may promote β-cell survival by modulation of production [e.g., through inactivation of nuclear factor–κB (NF-κB)] and effects of cytokines (22, 23). In humans, an association between blood 25(OH)D concentration and insulin secretion has been reported in some (24–27) but not all studies (28).

Vitamin D and insulin sensitivity

In peripheral insulin-responsive cells, vitamin D may enhance insulin sensitivity in several ways. 1,25(OH)₂D appears to directly augment insulin sensitivity by stimulating the expression of insulin receptors via binding to a vitamin D response element found in the human insulin receptor gene promoter (29–32). 1,25(OH)₂D may also enhance insulin sensitivity by activating peroxisome proliferator-activated receptor δ, a transcription factor implicated in the regulation of fatty acid metabolism in insulin-responsive tissues (33). Vitamin D may also influence insulin sensitivity indirectly via regulation of calcium homeostasis. Calcium is known to modulate intracellular processes in insulin-responsive tissues with a very narrow range of intracellular calcium levels needed for optimal insulin-mediated functions (34–38); therefore, vitamin D–mediated alteration in intracellular calcium concentration and calcium flux may impair insulin signal transduction leading to decreased glucose transporter activity (36, 38). Hypovitaminosis D also leads to increased parathyroid hormone concentration, which has been associated with increased insulin resistance (39). Vitamin D may also improve insulin sensitivity through the renin-angiotensin-aldosterone system. Angiotensin II is thought to contribute to insulin resistance in skeletal muscle via several mechanisms, including activation of NF-κB (40, 41). Renin expression and angiotensin II production were increased several folds in vitamin D receptor-null mice, whereas administration of 1,25(OH)₂D suppressed renin biosynthesis (42–44). Finally, vitamin D insufficiency is associated with increased fat infiltration in skeletal muscle, independent of body mass, which may contribute to reduced insulin action (45). In observational human studies, low vitamin D status [assessed by self-reported vitamin D intake or blood 25(OH)D concentration] has been associated with indices of insulin resistance, including measurements of fasting insulin and homeostasis model assessment (24–27, 46–54), but the association is not consistent (28, 51, 55).

Vitamin D and systemic inflammation

Systemic inflammation, via proinflammatory cytokines, plays an important role in the pathogenesis of type 2 diabetes mostly by promoting insulin resistance; however, pancreatic β-cell function may also be affected (56–58). Vitamin D can mitigate the effects of inflammation on diabetes risk in several ways. 1,25(OH)₂D may improve insulin sensitivity and protect against β-cell cytokine-induced apoptosis by directly modulating the expression and activity of cytokines (23, 59, 60). One such pathway may be through downregulation of NF-κB, which is a major transcription factor for tumor necrosis factor–α and other inflammatory mediators (61). Another potential pathway of the antiapoptotic effect of 1,25(OH)₂D on β-cells is through counteracting cytokine-induced Fas expression (62). Several other immune-modulating effects of vitamin D (e.g., blockade of dendritic cell differentiation, inhibition of lymphocyte proliferation, inhibition of foam cell formation and cholesterol uptake in macrophages, enhanced regulatory T-lymphocyte development) suggest additional pathways of protection against inflammation-induced diabetes risk (63, 64). In observational human studies, low vitamin D status has been associated with an elevated concentration of markers of systemic inflammation in some (54, 65–68) but not all studies (69–72).

Epidemiological Studies

Many cross-sectional studies have reported inverse associations between vitamin D status and glucose intolerance, including studies using data from the National Health and Nutrition Examination Survey (52, 73–76) and other large cohorts from the United States (77, 78) and Europe (79, 80). However, cross-sectional studies are only hypothesis generating because the directionality of the association cannot be established.

Over the last decade, several longitudinal observational cohort studies that examine the association between blood 25(OH)D concentration and incident type 2 diabetes have been published, and the results have been summarized in recent meta-analyses. Song et al. (81) combined data from 21 longitudinal cohorts (76,220 participants; 4996 incident diabetes cases) and estimated a 38% risk reduction for incident diabetes in the highest vs the lowest category of 25(OH)D concentration. The association did not differ by sex, duration of follow-up, cohort sample size, 25(OH)D assay method, or diabetes diagnostic criteria. A spline regression model showed that higher 25(OH)D concentration was monotonically associated with a lower diabetes risk, suggesting no apparent plateau. Ye et al. (82) included data from 22 longitudinal cohorts (89,698 noncases; 8492 diabetes cases) and reported that a 10-mg/dL lower 25(OH)D level was associated with a 22% higher risk of new-onset diabetes with moderate evidence of heterogeneity and no evidence of publication bias.

Despite variability among the various cohorts (e.g., baseline glucose tolerance status, age, ethnicity, latitude, vitamin D status, definition of diabetes outcome), the inverse association between 25(OH)D concentration and incident diabetes in longitudinal observational cohorts is highly consistent, and the evidence strongly indicates that 25(OH)D concentration is a strong biomarker of diabetes risk. However, the observational nature of these studies precludes a definitive assessment of cause and effect because residual confounding or reverse causation cannot be excluded. Confounding is especially problematic with studies of vitamin D because blood 25(OH)D concentration is an excellent biomarker not only of vitamin D status that reflects intake and biosynthesis but also of good health: a high 25(OH)D concentration is associated with young age, normal body weight, a healthy lifestyle (including good dietary and exercise habits), and other favorable behaviors that may be associated with lower diabetes risk independently of vitamin D. Adiposity, which is inversely correlated with blood 25(OH)D concentration, is likely the most important confounder in the cohort studies, but it is not clear whether obesity directly contributes to the development of hypovitaminosis D (via lower vitamin D intake, reduced biosynthesis due to less outdoor activity, vitamin D sequestration, volumetric dilution) or hypovitaminosis is involved in the development of obesity (83).

Observational studies using Mendelian randomization approaches, which offer the potential advantage that the reported genetic associations with phenotypes are unconfounded, have shown no association between certain alleles relevant to vitamin D physiology and incident type 2 diabetes (82, 84–86). However, Mendelian randomization studies center on certain assumptions that may not apply to vitamin D (82). Specifically, the tested alleles accounted for <5% of the variance in 25(OH)D concentration, which is not surprising given that 25(OH)D is determined by behavior (e.g., exposure to ultraviolet B light, diet) and other nongenetic factors (e.g., adiposity, age). Furthermore, Mendelian randomization studies did not predict amounts of bioavailable (e.g., free vitamin D) or biologically active [e.g., 1,25[OH]₂D] vitamin D and cannot distinguish between endogenous and exogenous sources of vitamin D. Indeed, variation in the DHCR7 gene, which is associated with lower vitamin D biosynthesis, is associated with type 2 diabetes risk, suggesting that lifelong exposure to endogenous vitamin D synthesis might be more important for prevention of diabetes than short-term vitamin D intake (85). Mendelian randomization studies may also be confounded by pleiotropic effects of genetic variants and are further limited by the assumption of a linear association between genetic variants, 25(OH)D, and diabetes risk, which may not hold. Despite their theoretical appeals, Mendelian randomization studies that predict 25(OH)D concentration from genetic polymorphisms may lead to unwarranted conclusions, and results should be interpreted with caution (82).

Randomized Clinical Trials

More than 30 trials have reported the effect of vitamin D [ergocalciferol (D₂) or cholecalciferol (D₃)] supplementation (with or without calcium) on glycemia and insulin sensitivity or incident diabetes in people without type 2 diabetes (i.e., normal glucose tolerance or prediabetes).

Vitamin D supplementation in people with normal glucose tolerance

Among trials in people with normal glucose tolerance, synthesized in a recent meta-analysis by Seida et al. (87), vitamin D supplementation had no effect on fasting glucose or HbA1c (88–99), insulin resistance (89–91, 94, 95, 99), or incident diabetes (90, 100). However, most trials were not designed for glycemic outcomes, and all trials except two (90, 100) had relatively short duration and were underpowered for glycemic outcomes. The two largest trials (90, 100) used relatively small vitamin D doses (400 or 800 units/d), which may not raise 25(OH)D concentration above a threshold required to change the pathophysiology of type 2 diabetes (101–103). Importantly, adherence to supplementation was suboptimal, which further limits drawing any conclusions. Overall, vitamin D appears to have no effect among people with normal glucose tolerance. This is not surprising because it is very difficult to demonstrate improvement in clinical variables (e.g., fasting glucose, HbA1c) that are in the normal range at the outset.

Vitamin D supplementation in patients at risk for diabetes (or prediabetes)

Thirteen trials have reported results on the effect of vitamin D supplementation on glycemic variables (89, 90, 103–112) or incident diabetes (105, 106, 108, 113) in patients at risk for diabetes. Many studies reported no effect of vitamin D supplementation, whereas some reported trends or significant improvements in glycemia. In a post hoc analysis of a trial designed to assess bone-related outcomes, daily supplementation with 700 units of vitamin D₃ and 500 mg calcium carbonate improved insulin resistance among those with prediabetes (impaired fasting glucose) at baseline (89). In a mechanistic trial among adults at risk for type 2 diabetes, short-term supplementation with vitamin D₃ improved β-cell function and attenuated the rise in HbA1c that occurs in this population over time (103). In a meta-analysis by Seida et al. (87), nearly significant improvements in fasting glucose and HbA1c were found among patients with prediabetes who received vitamin D supplementation. In a more recent meta-analysis that included 10 trials among people with prediabetes, vitamin D significantly reduced fasting glucose and HbA1c (114). Although mean glycemic improvements reported in the meta-analyses were small (e.g., absolute decline in HbA1c by ∼0.1%), such declines can have a large impact at the population level. For example, in the Diabetes Prevention Program study, the difference in HbA1c between the lifestyle arm and placebo was ∼0.15%, which was associated with a 58% decrease in incident diabetes (115).

The largest trial of vitamin D supplementation for the prevention of type 2 diabetes is the Tromsø study (Norway), which randomized 511 adults with prediabetes to 20,000 U/wk (∼2900 U/d) of vitamin D₃ or placebo and followed them for an average of 3.3 years for incident diabetes. In the vitamin D group, the rate of incident diabetes was lower than the placebo group throughout the study, but the difference was not significant (hazard ratio, 0.90; 95% confidence interval, 0.69 to 1.18). Despite its relatively large size, the Tromsø trial was underpowered to detect small but clinically significant reductions in diabetes risk; therefore, its results are inconclusive.

Overall, the available evidence from underpowered trials does not support a role of vitamin D supplementation in the prevention of type 2 diabetes; however, results from ongoing, large, long-term trials will be informative (Table 1). Two such trials are designed with incident diabetes as the primary outcome. The National Institutes of Health–supported D2d (Vitamin D and Type 2 Diabetes; 116) study tests the safety and efficacy of 4000 U/d vitamin D₃ supplementation for the prevention of diabetes among patients with prediabetes. The Diabetes Prevention with Active Vitamin D (DPVD) trial tests the effect of eldecalcitol (active vitamin D analogue) on diabetes incidence among people with prediabetes in Japan (117). Other large trials [the Vitamin D and Omega-3 Trial (VITAL; 118); D-Health (119)] test the effect of vitamin D supplementation on cardiovascular disease and cancer as primary outcomes in the general population and include incident diabetes as a secondary outcome.

Table 1.

Ongoing, Large, Long-Term Randomized Placebo-Controlled Trials of Vitamin D Supplementation With Incident Diabetes as an Outcome

Trial [Country]	Trial Registration Number	Type of Trial	Sample Size	Inclusion Criteria	Vitamin D Intervention	Treatment Duration, y	Incident Diabetes Primary Endpoint?	Definition of Incident Diabetes	Estimated Completion Date (Start Date)
D2d [United States] (116)	NCT01942694	Efficacy	2423	Age >30 y	D3 4000 IU/d	3	Yes	Two of three ADA diabetes criteria (fasting glucose >125 mg/dL; 2-h glucose >199 mg/dL; HbA1c >6.4%) or one ADA criterion with confirmation, assessed twice annually by the central laboratory	2019 (2013)
				BMI 22.5 to 42 kg/m2
				Two of three ADA prediabetes criteria (fasting glucose 100 to 125 mg/dL; 2-h glucose 140 to 199 mg/dL; HbA1c 5.7% to 6.4%)
DPVD [Japan] (117)	UMIN000010758	Efficacy	750	Age 30 to 90 y	Eldecalcitol 0.75 μg/d	2.8	Yes	HbA1c ≥6.5% plus one of the following: fasting glucose ≥126 mg/dL; 2-h glucose ≥200 mg/dL; random glucose ≥200 mg/dL	Not available (2013)
				IGT (2-h glucose 140–199 mg/dL) plus fasting glucose <126 mg/dL and HbA1c <6.5%
VITAL [United States] (118)	NCT01169259	Effectiveness	25,874	Age ≥55 y (women); ≥50 y (men)	D3 2000 IU/d, provided by mail	5	No	Self-reported (annual questionnaire, confirmed by review of medical records)	2017 (2010)
				Healthy
D-Health [Australia] (119)	ACTRN12613000743763	Effectiveness	21,315	Age 60 to 84 y	D3 60,000 IU/mo (∼2000 IU/d effective dose)	5	No	Self-reported (annual questionnaire) and data linkage to health records	2025 (2014)

VITAL is a 2 × 2 factorial design trial with omega-3 fatty acids and vitamin D as the active interventions; D2d and VITAL have completed recruitment. Enrollment in other trials is in progress. The final sample size for D-Health was reduced to 21,315 from 25,000.

Abbreviations: ADA, American Diabetes Association; BMI, body mass index; D2d, Vitamin D and Type 2 Diabetes; DPVD, Diabetes Prevention with Active Vitamin D; IGT, impaired glucose tolerance; VITAL, Vitamin D and Omega-3 trial.

Conclusions

The strong and consistent inverse associations between blood 25(OH)D concentration and incident diabetes reported in observational studies are supported by data on biological plausibility from mechanistic studies and raise the possibility that optimizing vitamin D status may reduce the risk of type 2 diabetes. Results from underpowered trials and post hoc analyses of trials designed for nondiabetic outcomes do not support a role of vitamin D supplementation for the prevention of type 2 diabetes among people with normal glucose tolerance. Vitamin D supplementation may have a role in the prevention of type 2 diabetes in high-risk populations; however, it is imperative that adequately powered randomized trials in well-defined populations (e.g., prediabetes) are completed to definitively test the hypothesis that vitamin D is a contributor to the pathogenesis of type 2 diabetes and has a role in prevention. It is important to note that because type 2 diabetes is a multifactorial disease, it is unlikely that vitamin D deficiency would prove to be a central cause or a major therapeutic target. Until definitive evidence is available, vitamin D supplementation should not be used to prevent type 2 diabetes.