The Apparent Relation between Plasma 25-Hydroxyvitamin D and Insulin Resistance Is Largely Attributable to Central Adiposity in Overweight and Obese Adults^,

Abstract

Background: Research indicates that plasma 25-hydroxyvitamin D [25(OH)D] is associated with insulin resistance, but whether regional adiposity confounds this association is unclear.

Objective: This study assessed the potential influence of adiposity and its anatomical distribution on the relation between plasma 25(OH)D and insulin resistance.

Methods: A secondary analysis of data from middle-aged overweight and obese healthy adults [n = 336: 213 women and 123 men; mean ± SD (range); age: 48 ± 8 y (35–65 y); body mass index (BMI; in kg/m²): 30.3 ± 2.7 (26–35)] from West Lafayette, Indiana (40.4°N), were used for this cross-sectional analysis. Multiple linear regression analyses that controlled for multiple covariates were used as the primary statistical model.

Results: Of all participants, 8.6% and 20.5% displayed moderate [20.1–37.5 nmol/L plasma 25(OH)D] to mild (37.6–49.9 nmol/L) vitamin D insufficiency, respectively. A regression analysis controlling for age, sex, race, plasma parathyroid hormone concentration, season of year, and supplement use showed that 25(OH)D was negatively associated with fasting insulin (P = 0.021). Additional regression analyses showed that total and central adiposity but not peripheral adiposity predicted low plasma 25(OH)D [total fat mass index (FMI): P = 0.018; android FMI: P = 0.052; gynoid FMI: P = 0.15; appendicular FMI: P = 0.07) and insulin resistance (homeostasis model assessment of insulin resistance: total and android FMI, P <0.0001; gynoid FMI, P = 0.94; appendicular FMI, P = 0.86). The associations of total and central adiposity with insulin resistance remained significant after adjusting for plasma 25(OH)D. However, adjusting for central adiposity but not other anatomical measures of fat distribution eliminated the association between plasma 25(OH)D and insulin resistance.

Conclusion: Central adiposity drives the association between plasma 25(OH)D and insulin resistance in overweight and obese adults. The trial was registered at clinicaltrials.gov as NCT00812409.

Introduction

The prevalence of vitamin D insufficiency is widespread (1), with serum or plasma 25-hydroxyvitamin D [25(OH)D]⁵ concentrations falling below the current Institute of Medicine (IOM) recommendation in 40% of Americans (≥20 ng/mL, ≥50 nmol/L) (2). Traditionally, vitamin D is known for its essential role in regulating mineral homeostasis and bone metabolism (3). Maintaining adequate vitamin D status, as reflected by 25(OH)D, is critical for overall skeletal health (3). However, current evidence suggests the existence of nonskeletal effects of vitamin D in cardiovascular disease risk (4–6), cancer risk (7–9), and muscle (10–12) or immune (10, 13, 14) function. Low 25(OH)D concentrations are also associated with impaired glucose tolerance and an increased risk for type 2 diabetes mellitus.

Several population studies have shown that 25(OH)D is inversely correlated with HOMA-IR scores (15), fasting insulin (16), and glucose concentrations (17). A 2011 meta-analysis that examined the association between vitamin D status and risk of type 2 diabetes showed a 43% lower risk of developing type 2 diabetes in vitamin D-sufficient (>25 ng/mL) compared with -insufficient (<14 ng/mL) individuals (18). In addition, the progression from prediabetes to diabetes in a large prospective cohort study was reduced by 62% in the highest 25(OH)D quartile compared with the lowest quartile group (19). However, given that obesity is often associated with both low 25(OH)D concentrations (15, 17, 18, 20) and glucose intolerance (21–23), the association between 25(OH)D and insulin resistance could be confounded by adiposity. This relation between vitamin D status and adiposity may be driven by body fat distribution (24) and not simply body mass as suggested by others (25) as central adiposity has been shown to be a strong predictor of 25(OH)D independent of BMI (26). Central adiposity, particularly visceral fat, is also associated with impaired glucose tolerance (21) and the development of insulin resistance and type 2 diabetes (27) independent of age, sex, and BMI status (28–30).

The influence of adiposity and its anatomical distribution may confound the relation between 25(OH)D and glucose homeostasis. When controlling for BMI or total fat mass, the association between vitamin D status and insulin resistance continued to persist in some studies (31–35) but was lost in others (36, 37). However, BMI is a measurement of body mass relative to height and does not directly assess total adiposity or its anatomical distribution. Furthermore, as seen with gynoid compared with android adiposity (38, 39), assessing only total fat mass ignores the influence of regional adiposity on indexes of insulin-mediated glucose control. Therefore, using only BMI or total fat mass as a covariate in previous studies negated the opportunity to assess the influence of regional adiposity on the relation between plasma vitamin D status and insulin resistance. To determine whether adiposity and its anatomical distribution influence the apparent relation between plasma 25(OH)D and insulin resistance, a secondary analysis of data from a sample of middle-aged, overweight, and moderately obese participants were analyzed to determine 1) whether whole-body or regional adiposity influences plasma 25(OH)D, 2) the independent effects of adiposity and plasma 25(OH)D on indexes of insulin resistance, and 3) whether adjusting for measurements of adiposity or plasma 25(OH)D affects their relation to indexes of insulin resistance. We hypothesized that central adiposity would correlate with plasma 25(OH)D concentrations better than BMI or peripheral adiposity and would also correlate with insulin resistance independent of vitamin D status. Furthermore, we hypothesized that any significant relations between plasma 25(OH)D and insulin resistance would be lost after adjusting for central adiposity.

Methods

Subjects.

This secondary analysis used baseline (preintervention) data (n = 336) from a double-blind, placebo-controlled, community-based, randomized 36-wk intent-to-treat study conducted from 2007 to 2010 (40). Participants were recruited from the greater West Lafayette, Indiana area (40.4°N) with the following inclusion criteria: men or women aged 35 to 65 y, body mass <300 lb (136 kg), BMI (in kg/m²) between 26 and 35, serum low-density lipoprotein cholesterol <4.1 mmol/L, total cholesterol <6.7 mmol/L, TGs <4.5 mmol/L, fasting plasma glucose <6.1 mmol/L, blood pressure <160/100 mm Hg, no preexisting liver or renal conditions, not currently or previously (past 6 mo) consuming a weight-loss diet or other special/nonbalanced diets, no weight change (±4.5 kg) within the previous 6 mo, and <2 h/wk of habitual resistance or aerobic exercise training in the previous 6 mo. Participants taking medications for elevated blood pressure (n = 51), reduced high-density lipoprotein cholesterol or elevated TGs (n = 48), including statins, were included in the study because their medication use did not affect the final analysis. All subjects provided written informed consent and received monetary compensation for participating. The Purdue University Institutional Review Board approved the study protocol, which complied with the Helsinki Declaration as revised in 1983.

Body composition and anthropometric measurements.

Fasting-state body mass (±0.1 kg) (ES200L; Mettler) and height (±0.1 cm) were measured, and BMI was calculated. Whole-body and regional (android, appendicular, and gynoid) adiposity measures were determined in the supine position using DXA (Lunar iDXA and enCORE version 11.2; GE Healthcare). Android adiposity, which includes abdominal subcutaneous and visceral fat, was defined as the bottom of the lowest rib to the top of the iliac crest; gynoid adiposity was defined as a region twice the height of the defined android region below the iliac crest; and appendicular adiposity was defined as the summation of bilateral leg and arm fat masses. Although the Lunar software cannot differentiate between different compartments of adipose tissue (visceral compared with subcutaneous), previous studies have validated the assessment of adiposity by DXA compared to computed tomography (CT) that showed a low margin of error (41–43). Daily quality-assurance calibrations were conducted automatically in addition to using a standardized phantom on a weekly basis to ensure the accuracy and reproducibility of soft tissue analysis. Whole-body and regional fat masses are reported as absolute amounts (kg) and expressed as an index in relation to body height [fat mass index (FMI), kg/m²].

Blood collection and glucose tolerance assessment.

Blood samples were collected after a 12-h overnight fast in tubes that contained sodium heparin to obtain plasma (BD Vacutainer; Becton, Dickinson and Company) and placed immediately on ice for 30 min before being centrifuged (4°C for 10 min at 3000g). The supernatant was separated and stored in microcentrifuge tubes at −80°C for subsequent glucose and insulin analyses. For the 3-h oral-glucose-tolerance test, participants consumed a sugar solution that contained 75 g of dextrose, and blood samples were collected at 0 (fasting), 30, 60, 90, 120, 150, and 180 min after consumption. Plasma glucose concentrations were measured by enzymatic colorimetry using an oxidase method on a COBAS INTEGRA 400 analyzer (Roche Diagnostics). Plasma insulin concentrations were measured by an electrochemiluminescence immunoassay method on an Elecsys 2010 analyzer (Roche Diagnostics). Total AUCs for glucose and insulin were determined using the trapezoidal rule (44). HOMA-IR and the whole-body (composite) insulin sensitivity index (ISI) were calculated as previously described (45, 46). Radioimmunoassays (DiaSorin) were used to measure plasma 25(OH)D concentrations. All samples were measured in duplicate with a lower detection limit of 1.5 ng/mL or 3.7 nmol/L, and vitamin D status was classified according to IOM standards (47).

Statistical analyses.

We used multiple linear regression analyses to assess the influences of adiposity and plasma 25(OH)D on indexes of insulin resistance from plasma glucose and insulin measurements. The specific dependent variables of interest were fasting and 2-h glucose and insulin concentrations, glucose and insulin AUCs, and HOMA-IR and ISI. Initial analyses revealed positive associations between plasma 25(OH)D and supplement use, season of the year, and negative associations with race and plasma parathyroid hormone (PTH). We found both positive and negative associations between adiposity and sex depending upon the regional measurement of adiposity, whereas positive associations were shown for age and plasma PTH and negative associations with race. Therefore, age, sex, race, supplement use, season of year, and plasma PTH were used as covariates for all regression analyses. The relations between 1) plasma 25(OH)D and glucose tolerance indexes and 2) adiposity and glucose tolerance indexes were assessed both with and without adiposity (association 1) and plasma 25(OH)D (association 2) as covariates, respectively. The regression coefficient was signified by β and statistical significance by a P value <0.05.

Results

Participant characteristics.

Body composition, plasma glucose and insulin, plasma vitamin D status, and supplement use results are reported in Table 1. Despite being obese, the participants had normal glucose and insulin concentrations. Most participants (70.6%) also had plasma 25(OH)D concentrations above the IOM cut off of 50 nmol/L (sufficient).

TABLE 1.

Characteristics of middle-aged overweight and obese adults¹

	Total (n = 336)	Men (n = 123)	Women (n = 213)
Demographics
Age, y	48.3 ± 7.7	48.8 ± 8.0	48.0 ± 7.5
Race/ethnicity, n
Caucasian	321	120	201
Black	3	1	2
Asian	5	1	4
Hispanic	7	1	6
Supplement use, n	112	37	75
Adiposity measures
BMI, kg/m2	30.4 ± 2.8	30.7 ± 2.9	30.2 ± 2.7
Total fat mass, kg	35.5 ± 6.3	33.5 ± 6.2	36.7 ± 6.0
Total FMI, kg/m2	12.4 ± 2.5	10.5 ± 2.0	13.6 ± 2.1
Android fat mass, kg	3.6 ± 0.9	3.9 ± 0.9	3.5 ± 0.8
Android FMI, kg/m2	1.3 ± 0.3	1.2 ± 0.3	1.3 ± 0.3
Gynoid fat mass, kg	6.4 ± 1.5	5.3 ± 1.1	7.1 ± 1.2
Gynoid FMI, kg/m2	2.3 ± 0.6	1.7 ± 0.3	2.6 ± 0.4
Appendicular fat mass, kg	15.0 ± 3.5	12.2 ± 2.4	16.7 ± 3.0
Appendicular FMI, kg/m2	5.3 ± 1.4	3.8 ± 0.7	6.2 ± 1.0
Plasma glucose measures
Fasting glucose, mmol/L	5.3 ± 0.5	5.4 ± 0.5	5.2 ± 0.6
Fasting insulin, pmol/L	67.8 ± 6.6	76.2 ± 6.2	63.0 ± 34.8
2-h glucose, mmol/L	6.7 ± 2.0	6.9 ± 1.9	6.6 ± 2.0
2-h insulin, pmol/L	447 ± 391	468 ± 415	436 ± 377
Glucose AUC, mmol/L ⋅ 3 h	1250 ± 251	1300 ± 233	1220 ± 257
Insulin AUC, nmol/L ⋅ 3 h	73.5 ± 46.9	78.9 ± 46.8	70.4 ± 46.8
HOMA-IR	2.8 ± 1.8	3.1 ± 1.7	2.6 ± 1.9
ISI	5.0 ± 3.1	4.3 ± 2.6	5.4 ± 3.3
Vitamin D
Plasma 25(OH)D, nmol/L	59.9 ± 18.7	57.4 ± 17.0	61.2 ± 19.5
Moderately insufficient (20.1–37.5 nmol/L), %	8.6	8.9	7.5
Mildly insufficient (37.6–49.9 nmol/L), %	20.8	26.8	17.8
Sufficient (50–74.9 nmol/L), %	52.1	48.0	54.0
Beyond sufficient (75 nmol/L), %	18.5	16.3	20.7

¹Values are means ± SDs unless otherwise indicated. FMI, fat mass index; ISI, insulin sensitivity index; 25(OH)D, 25-hydroxyvitamin D.

Adiposity and 25(OH)D association.

Higher BMI and total FMI were associated with lower plasma 25(OH)D concentrations (Table 2). A similar trend was shown for central adiposity (android FMI). Other measurements of whole-body and regional adiposity, expressed in absolute amounts (Table 2) or as a percentage of body mass (data not shown), were not associated with plasma 25(OH)D.

TABLE 2.

Relations between adiposity and plasma 25(OH)D in middle-aged overweight and obese adults¹

	P	β
BMI, kg/m2	0.027	−0.119
Total FMI, kg/m2	0.018	−0.157
Total fat mass, kg	0.18	−0.074
Gynoid fat mass, kg	0.64	−0.031
Gynoid FMI, kg/m2	0.15	−0.120
Appendicular fat mass, kg	0.44	−0.053
Appendicular FMI, kg/m2	0.07	−0.156
Android fat mass, kg	0.18	−0.075
Android FMI, kg/m2	0.052	−0.106

¹Multiple linear regression controlled for age, sex, race, seasons of year, supplement use, and plasma parathyroid hormone (P <0.05). FMI, fat mass index; 25(OH)D, 25-hydroxyvitamin D.

Associations between adiposity and measures of glucose and insulin control.

Greater whole-body adiposity was associated with higher plasma glucose, insulin, HOMA-IR, and lower ISI (Table 3). Of the anatomical measures of fat distribution, only central adiposity was associated with glucose and insulin control. Adjusting for plasma 25(OH)D did not affect the associations between BMI, total fat mass, or central adiposity and indexes of insulin resistance.

TABLE 3.

Relations between adiposity and markers of glucose and insulin control in middle-aged overweight and obese adults¹

	Fasting plasma glucose		Fasting plasma insulin		HOMA-IR		ISI
	P	β	P	β	P	β	P	β
Not adjusted for plasma 25(OH)D
Total adiposity
BMI, kg/m2	0.004	0.158	<0.001	0.358	<0.001	0.337	<0.001	−0.287
Total fat mass, kg	0.032	0.123	<0.001	0.264	<0.001	0.272	<0.001	−0.228
Total FMI, kg/m2	0.05	0.133	<0.001	0.325	<0.001	0.316	<0.001	−0.323
Central adiposity
Android fat mass, kg	0.002	0.179	<0.001	0.323	<0.001	0.329	<0.001	−0.350
Android FMI, kg/m2	0.002	0.169	<0.001	0.316	<0.001	0.312	<0.001	−0.369
Peripheral adiposity
Appendicular fat mass, kg	0.41	−0.058	0.90	0.009	0.63	0.035	0.19	0.091
Appendicular FMI, kg/m2	0.23	−0.107	0.95	0.006	0.86	0.016	0.36	0.080
Gynoid fat mass, kg	0.42	−0.055	0.99	<0.001	0.70	0.027	0.40	0.058
Gynoid FMI, kg/m2	0.24	−0.101	0.95	−0.005	0.94	0.007	0.71	0.032
Adjusted for plasma 25(OH)D
Total adiposity
BMI, kg/m2	0.006	0.152	<0.001	0.347	<0.001	0.330	<0.001	−0.283
Total fat mass, kg	0.040	0.118	<0.001	0.256	<0.001	0.266	<0.001	−0.225
Total FMI, kg/m2	0.07	0.123	<0.001	0.310	<0.001	0.305	<0.001	−0.317
Central adiposity
Android fat mass, kg	0.002	0.174	<0.001	0.315	<0.001	0.323	<0.001	−0.347
Android FMI, kg/m2	0.004	0.163	<0.001	0.305	<0.001	0.305	<0.001	−0.365
Peripheral adiposity
Appendicular fat mass, kg	0.38	−0.062	0.98	0.001	0.68	0.029	0.18	0.095
Appendicular FMI, kg/m2	0.17	−0.120	0.87	−0.015	0.99	−0.001	0.30	0.092
Gynoid fat mass, kg	0.40	−0.058	0.94	−0.005	0.73	0.024	0.38	0.060
Gynoid FMI, kg/m2	0.20	−0.110	0.80	−0.022	0.95	−0.006	0.64	0.041

¹Multiple linear regression controlled for age, sex, race, season of year, supplementation, and plasma parathyroid hormone (P < 0.05). Comparable results were obtained for AUC and 2-h measurements (insulin and glucose). FMI, fat mass index; ISI, insulin sensitivity index value; 25(OH)D, 25-hydroxyvitamin D.

Associations between 25(OH)D and measures of glucose and insulin control.

Lower plasma 25(OH)D concentrations were associated with higher fasting insulin concentrations but not with fasting glucose concentrations, HOMA-IR, or ISI (Table 4). Adjusting for central adiposity (android FMI) but not other anatomical measures of fat distribution (gynoid and appendicular FMI) eliminated the relation between plasma 25(OH)D and fasting insulin (Table 4).

TABLE 4.

Relations between plasma 25(OH)D and markers of glucose and insulin control in middle-aged overweight and obese adults¹

	Fasting plasma glucose		Fasting plasma insulin		HOMA-IR		ISI
Measured plasma 25(OH)D	P	β	P	β	P	β	P	β
Not adjusted for adiposity, nmol/L	0.17	−0.078	0.021	−0.131	0.07	−0.104	0.26	0.064
Adjusted for android FMI, nmol/L	0.31	−0.058	0.08	−0.097	0.20	−0.070	0.58	0.029
Adjusted for gynoid FMI, nmol/L	0.15	−0.082	0.021	−0.133	0.07	−0.104	0.21	0.072
Adjusted for appendicular FMI, nmol/L	0.14	−0.084	0.021	−0.133	0.07	−0.104	0.19	0.076

¹Multiple linear regression controlled for age, sex, race, season of year, supplementation, and plasma parathyroid hormone (P < 0.05). Comparable results were obtained for AUC and 2-h measurements (insulin and glucose). FMI, fat mass index; ISI, insulin sensitivity index; 25(OH)D, 25-hydroxyvitamin D.

Discussion

Data from our study suggest that central adiposity but not other anatomical measures of fat distribution negatively influences both plasma 25(OH)D concentrations and markers of glucose and insulin control independent of vitamin D status in middle-aged overweight and obese adults. Previous clinical studies have shown an inverse association between low vitamin D status and insulin resistance that was independent of total adiposity (BMI and total fat mass) (31–35). However, this study shows that the inverse relation between vitamin D and insulin resistance is indeed confounded by adiposity, particularly central adiposity.

Many studies have consistently documented an inverse relation between obesity and 25(OH)D concentrations (25). BMI is a widely used measurement of adiposity and is often used to account for the confounding effects of obesity on nutrient–disease relations (25). However, more sensitive measures of adiposity obtained via DXA, CT, or MRI have shown that central adiposity (i.e., visceral fat) is a strong predictor of vitamin D status independent of age, sex, race, season, dietary intake, BMI status, and even UVB exposure (24, 26, 43, 48, 49). These results show that a direct relation exists between body fat distribution and plasma 25(OH)D concentrations.

Several possible mechanisms may explain the inverse relation between body fat mass and vitamin D status, including the sequestration of vitamin D into adipose tissue (50, 51) and increased lipogenesis caused by elevated PTH (and calcium-mediated signaling in adipocytes) when vitamin D status is low (52). However, although vitamin D deficiency is correlated with obesity (53), supplementing 7000 IU of cholecalciferol per day for 26 wk did not influence fat mass in obese subjects with 25(OH)D concentrations <50 nmol/L (54). This suggests that obesity negatively influences vitamin D status rather than low vitamin D status driving obesity. Our data extend this interpretation by showing that central adiposity and not peripheral adiposity is the major obesity-related determinant of vitamin D status.

Because adiposity is associated with low vitamin D status, it is critical to determine whether the reported negative association between vitamin D status and insulin resistance (15–18, 55) is independent of obesity and its positive association to insulin resistance (21, 56–60). Our data suggest that it is not. In support of our hypothesis, we found that adjusting for central adiposity eliminated the association between plasma 25(OH)D and fasting insulin. The loss of the relation between plasma 25(OH)D and insulin resistance was also seen after adjusting for general adiposity (BMI or total FMI). However, these associations were likely driven by the unique impact of central adiposity because only android FMI—and not peripheral adiposity indexes—affected the relation between plasma 25(OH)D and insulin resistance. These results are supported by 2 previous studies that assessed how adiposity influenced the relation between 25(OH)D and insulin resistance and showed that adjusting for visceral fat attenuates this relation in young adolescents (61) and healthy middle-aged adults of normal weight (16). Our study extends these observations by showing that it is central adiposity and not indexes of peripheral adiposity, which were previously not assessed, that is the underlying factor behind the relation between plasma 25(OH)D and insulin resistance. By assessing the influence of both central and peripheral adiposity on the relation between plasma 25(OH)D and insulin resistance in overweight and obese middle-aged adults, this study describes how certain fat depots may regulate both circulating plasma 25(OH)D concentrations and glucose homeostasis. Furthermore, given that a correction for plasma 25(OH)D in this study did not attenuate the relation between central adiposity and insulin resistance, our study shows that the impact of central adiposity on insulin sensitivity is independent of vitamin D status.

Although some animal studies show that the hormonally active form of vitamin D [1,25-dihydroxyvitamin D; 1,25(OH)₂D] regulates pancreatic insulin secretion (62–64), other data support our conclusion that vitamin D does not influence the development of diabetes. In vitamin D–sufficient nondiabetic (65, 66) and diabetic rats (65), 1,25(OH)₂D treatment did not alter insulin receptor expression, pancreatic insulin secretion, or insulin-stimulated glucose uptake. Similarly, vitamin D supplementation did not improve insulin secretion in vitamin D–sufficient adults (54, 67–69). In addition, 1,25(OH)₂D treatment did not influence the expression of insulin-sensitive genes or enhance glucose uptake in insulin-sensitive tissues of obese mice (70). Finally, a recent systematic review and meta-analysis of randomized controlled trials found no evidence that vitamin D supplementation could improve glucose homeostasis or prevent diabetes in either diabetic or obese populations (71).

The strengths of our study include the experimental design and outcomes of interest. Given the large sample size and profile of our participants [n = 336 middle-aged (35–65 y), overweight, and moderately obese (BMI: 26–35) men and women], our findings are generalizable to a large segment of the US adult population. Multiple indexes of insulin resistance were also assessed, including fasting plasma glucose and insulin concentrations, HOMA-IR, and ISI to assess hepatic glucose production, insulin-dependent and -independent glucose uptake, and whole-body insulin sensitivity. Indexes of fat mass were also used to more accurately assess body composition. Although BMI is often used to evaluate body weight in individuals of different heights (72, 73), BMI cannot distinguish between the individual contributions of fat mass and fat-free mass toward total body weight (74–76). As such, BMI lacks the sensitivity needed to specify the degree of adiposity of an individual since age, race, stature, and physical activity levels confound the relation between BMI and adiposity. However, only reporting fat mass or fat-free mass as a percentage of total body weight or in absolute terms (kg) fails to take into account the strong correlation between fat mass, fat-free mass, and body size (77). As originally proposed by Vanitallie et al. (78), normalizing fat mass by height, similar to BMI, allows researchers to interpret body composition more accurately than if expressed in absolute terms (kg) or as a percentage of body weight (79). Many clinical studies have supported the use of FMI (80–83) as an independent assessment of fat mass relative to total body size (84) and a more accurate assessment of obesity than BMI (79). There were admittedly some limitations to our study. As a secondary analysis of baseline data, our analysis could not provide causality but only reveal association. The use of DXA compared with CT or MRI to measure body composition precludes distinguishing between different types of fat depots, including subcutaneous and visceral fat. Although DXA and CT provide comparable results when assessing central adiposity (41–43), measurements of visceral fat may improve the specificity of these associations.

In conclusion, the data from our study show that the apparent relation between plasma 25(OH)D and insulin resistance is largely attributable to central adiposity. Although maintaining adequate vitamin D status is beneficial for bone (3) and potentially other nonskeletal outcomes (4–14), our results do not support the hypothesis that higher vitamin D status improves insulin-mediated glucose tolerance in middle-aged overweight and obese adults.