Background

It is now known that insufficient serum 25 hydroxyvitamin D3 (calcifediol, vitamin D3) alters metabolite function causing perturbation of many cellular functions, including that of the endocrine pancreas [1]. Recently, there has been a resurgence of hypovitaminosis D3 in many populations [2], including young, pale-skinned adults [3] in addition to those with pigmented skin. Suggested recent recommendations of ideal vitamin D3 serum levels for metabolic health are >70–100 nmol/L (previously >50 nmol/L) [4, 5]. In parallel, there has been a world-wide increase in the prevalence of obesity [6]. Links between hypovitaminosis D3 and obesity have been reported when obesity is defined using body mass index (BMI) [7, 8] and waist circumference (waist) [9]. Large waist, a surrogate for abdominal obesity, is the key marker required for the metabolic syndrome (MetSyn) as defined by the International Diabetes Federation (IDF) [10]. Whilst there is overlap in total body and abdominal obesity, the diverse metabolic processes of different adipose depots and lean tissue may underpin dissimilar hypotheses for the mechanisms proposed for the inverse relationship of vitamin D3 and obesity.

This study was designed to 1) assess the relationships between vitamin D3 and anthropometric, metabolic syndrome and TIIDM markers, 2) determine whether whole body [BMI] or central [waist] adiposity was significantly related to vitamin D3, and if so whether one was related independently of the other when corrected for well-known influences in mixed-ethnicity adults.

Methods

Population and anthropometry

250 ambulant adults in Auckland, New Zealand were recruited into a body weight (weight) loss trial with primary criteria including BMI 28–50 kg/m2, age >18 y, not currently using weight loss agents nor participating in commercial weight loss programmes, and a desire to lose weight. Baseline data from this study were analysed for relationships with vitamin D3. Ethnicity was the surrogate for skin pigment. The lightly pigmented skin sub-group consisted of Caucasians and the variably pigmented skin sub-group included all other ethnicities which were New Zealand Maori, Pacific Peoples (Tongan, Samoan,) and Asian (East, South or Indian). For anthropometry, participants were lightly clad and measurements were taken in duplicate. Weight, height, waist and blood pressure were measured using standard methods as detailed in our previous publication [11]. Body fat percentage (fat%) was assessed indirectly by multi-frequency bioelectrical impedance analysis (SFB3 MFBIA, Impedimed, Australia). All participants provided written informed consent. Ethics approval for this study was obtained from the Auckland Ethics Committee, Auckland, New Zealand.

Laboratory samples

Participants attended our community clinic for collection of fasting blood. 200 women and 43 men provided evaluable samples. Vitamin D3 was analysed using Vit D25 pre-extraction with acetonitrile, double antibody radioimmunoassay (DiaSorin Inc Stillwater, MN, USA). Fasting plasma glucose (FPG), serum lipids and haemoglobin A1c (HbA1c) were measured using standard methods [11].

Statistical Analysis

Linear regression analysis was used to investigate the relationships of the demographic data, anthropometry and laboratory tests with vitamin D3. Multivariable regression was performed with vitamin D3 as the outcome variable. Explanatory variables were gender, age, ethnicity, season, and either BMI or waist as these were likely to be highly correlated. An analysis including both BMI and waist was also carried out to investigate if one variable contributed over and above the other. SAS 8.0 statistical software (Cary, NC, 2003) was employed for analyses.

Results

Baseline data shows an obese population with prevalences of MetSyn and TIIDM ≈ 40% and 5% in 234 participants with available FPG. Table 1.

Table 1 Baseline data from 250 female and male overweight and obese participants
Full size table

There were modest but significant inverse relationships of vitamin D3 with weight (p = 0.0009), BMI (p = 0.005) and waist (p = 0.03) but no relationship could be shown with fat %. Figure 1. Vitamin D3 could not be shown to be related to any of the non-waist MetSyn markers (MetSynM) or MetSynMcount (number of markers added together) or the presence of MetSyn (p > 0.05, all). Vitamin D3 and HbA1c, alone of the individual metabolic and MetSynM were weakly inversely related (p = 0.01) but this relationship was not significant after exclusion of the three HbA1c values >10% (p = 0.22). Abnormal FPG was related to hypovitaminosis D3, again only when the three values >10 mmol/L were included (r2 = 0.17, p = 0.005). Figure 1.

Figure 1
figure 1

The relationship between vitamin D3 and anthropometric and metabolic markers in 250 overweight and obese men and women.

Full size image

Multivariable regression showed an estimated decrease of 0.74 nmol/L (p = 0.002) in vitamin D3 per 1 kg/m2 increase in BMI with the total model explaining 22% of the variation in vitamin D3 levels and BMI explaining 3% of the variation. On replacing BMI with waist, there was a decrease of 0.29 nmol/L (p = 0.01) vitamin D3 per 1 cm increase in waist, with the total model explaining 21% of the variation in vitamin D3 and waist explaining 3%. When both BMI and waist were included neither could be demonstrated to contribute over and above the other (p = 0.25 and 0.67 respectively), nor could an association of vitamin D3 with gender (p = 0.52) or age (p = 0.52) be shown. However there was strong evidence of its association with ethnicity (p < 0.0001) and season (p < 0.0001).

Discussion

In the current study we showed that low levels of circulating vitamin D3 were inversely related to markers of TIIDM (large waist and raised HbA1c), rather than total adipose mass, non-waist MetSynM or MetSyn per se.

Of the three anthropometric variables that were significantly inversely correlated with vitamin D3 only BMI and waist were further investigated as weight is considered too crude a measure of obesity. Vitamin D3 showed inverse relationships separately, but of the same magnitude, with both BMI and waist when corrected for confounders. Neither could be shown to be related to vitamin D3 given the level of the other. This may indicate either similar mechanisms, or that different metabolic processes are occurring, coincidentally producing similar outcomes.

Hypotheses from the literature are discussed in light of our findings

Whole body obesity, as defined by BMI, has been associated with or contributes to low vitamin D3 status [8, 12]. Wortsman et al., found lower vitamin D3 in the serum of obese participants after experimental UV irradiation, deducing that "obesity-associated vitamin D insufficiency is likely due to the decreased bioavailability of vitamin D3 from cutaneous and dietary sources because of its deposition in body fat compartments" [12]. It was unclear from that trial which fat compartments were involved. In the current study MFBIA fat% did not correlate with vitamin D3, in contrast to that of Arunabh et al., [13] where DEXA was performed in women, BMI<24 kg/m2. Total body fat includes both peripheral adipose at the hip and thigh, with beneficial metabolic effects in both women and men [14], as well as less healthy upper body and central fat depots [15]. The opposing effects of these adipose depots could possibly weaken any correlation with vitamin D3. Furthermore, the lack of relationship of fat% with vitamin D3 may reflect influences of fat-free compartments of bone, muscle [16] and abdominal organs (liver, kidney, gut [17]).

The links between the metabolic syndrome and vitamin D3 are not clear. In the present study, apart from waist, none of MetSynM alone, MetSynMcount, nor the presence of MetSyn (defined by three of five positive markers [9]), was correlated with vitamin D3. This lack of relationship of vitamin D3 and MetSyn has been reported previously [18], and two studies of vitamin D3 in the morbidly obese report conflicting relationships [19, 20]. However, in the large USA NHANES dataset Ford et al., found that abdominal obesity as measured by waist alone, in addition to MetSyn, was related to low vitamin D3, notably affecting mixed-ethnicity participants equally [21].

Conversion of vitamin D3 to its derivative 1,25 vitamin D3 is complex and involves other hormones. 1,25 vitamin D3, via its receptor which is present in insulin-producing beta-islet cells, is known to be a potent regulator of cell proliferation and differentiation [22, 23]. However, there is evidence that low vitamin D3 itself is associated with TIIDM irrespective of 1,25 vitamin D3 [8]. The inverse relationship of vitamin D3 with high to extreme HbA1c [24, 25] and/or FPG [7, 8] may indicate that it is the long-term, severely abnormal (carbohydrate) metabolism of TIIDM [7, 26, 27] and muscle insulin resistance [28], that is associated with hypovitaminosis D3. HbA1c, a glycated protein, is a predictor of 2-hour glucose in oral glucose tolerance testing, [29] an indicator of chronic hyperglycaemia, protein glycation damage [30] and oxidative stress [31]. Many new, profound and interacting mechanisms link hypovitaminosis D with other correlates of the metabolic syndrome, including renin regulation [1]. Vitamin D-upregulated protein-1 reportedly modulates endothelial oxidative stress, macrophage and smooth muscle function, depending on the stage of atherosclerosis [32, 33].

Limitations of the present study include the cross sectional design where cause cannot be attributed. Lifestyle, body shape sensitivity [34, 35] or cultural reasons [36] for precluding skin exposure to view, and ultraviolet light for efficient vitamin D3 synthesis, may selectively affect obese people but were not examined.

In the current study low serum vitamin D3 was inversely related to weight and BMI, but not fat mass, and to markers indicative of TIIDM (large waist and raised HbA1c), rather than MetSyn per se. The link between hypovitaminosis D3 and metabolic disorders, including obesity, MetSyn, TIIDM and CVD requires further investigation, particularly for those most at risk of these combined conditions.