This study sought to determine the dose-response effect of vitamin D supplementation on fasting glucose, insulin and insulin resistance in healthy children. We conclude that vitamin D had no impact on these measures after 12 weeks.
Abstract
Context:
Vitamin D supplementation trials with diabetes-related outcomes have been conducted almost exclusively in adults and provide equivocal findings.
Objective:
The objective of this study was to determine the dose-response of vitamin D supplementation on fasting glucose, insulin, and a surrogate measure of insulin resistance in white and black children aged 9–13 years, who participated in the Georgia, Purdue, and Indiana University (or GAPI) trial: a 12-week multisite, randomized, triple-masked, dose-response, placebo-controlled vitamin D trial.
Design:
Black and white children in the early stages of puberty (N = 323, 50% male, 51% black) were equally randomized to receive vitamin D3 (0, 400, 1000, 2000, or 4000 IU/day) for 12 weeks. Fasting serum 25-hydroxyvitamin D (25(OH)D), glucose and insulin were assessed at baseline and weeks 6 and 12. Homeostasis model assessment of insulin resistance was used as a surrogate measure of insulin resistance. Statistical analyses were conducted as intent-to-treat using a mixed effects model.
Results:
Baseline serum 25(OH)D was inversely associated with insulin (r = −0.140, P = 0.017) and homeostasis model assessment of insulin resistance (r = −0.146, P = 0.012) after adjusting for race, sex, age, pubertal maturation, fat mass, and body mass index. Glucose, insulin, and insulin resistance increased (F > 5.79, P < .003) over the 12 weeks, despite vitamin D dose-dependent increases in serum 25(OH)D.
Conclusions:
Despite significant baseline inverse relationships between serum 25(OH)D and measures of insulin resistance, vitamin D supplementation had no impact on fasting glucose, insulin, or a surrogate measure of insulin resistance over 12 weeks in apparently healthy children.
Approximately 10% of adolescents ages 12–17 years meet the criteria for prediabetes (1), with overweight and obese teenagers having higher prevalence rates up to 30% (2). Cost-effective public health measures for prevention of prediabetes and type 2 diabetes mellitus are crucial because serious and costly comorbidities can present (3). One inexpensive nutritional approach that has generated considerable debate is vitamin D supplementation. Insufficient vitamin D intake is common in children, as 87% of Americans ages 2–18 years fail to consume the Institute of Medicine (IOM) Estimated Average Requirement for dietary vitamin D (400 IU/day), including fortified sources (4). The role of vitamin D in the pathogenesis of diabetes is an emerging area of extraskeletal vitamin D research. Vitamin D is involved in insulin secretion (5) and action (6, 7), and vitamin D status has been reported to be inversely associated with diabetes risk factors in pediatric (8–10) and adult (11) studies. In the 2010 Dietary Reference Intake report for calcium and vitamin D, the IOM dubbed the role of vitamin D in diabetes a “hypothesis of emerging interest,” and pointed to the paucity of randomized controlled trials (RCT) to establish causality and dose-response relationships (12).
RCTs examining the effect of vitamin D on diabetes-related outcomes have been conducted mostly in adults and have yielded equivocal results (13). To date, the five pediatric RCTs examining the effect of vitamin D supplementation on glucose homeostasis have also produced mixed findings. These studies were conducted in relatively small samples of obese children at high risk for cardiometabolic dysfunction (14–16) or with diagnosed metabolic syndrome (17). Additionally, a nonrandomized, noncontrolled vitamin D intervention study in obese adolescent females also exists (18). Supplementation regimens in these studies varied considerably by frequency, dose, and follow-up durations. Moreover, the studies used one treatment dose across a range of pubertal maturation stages. No dose-response RCTs have been published in healthy children to ascertain the effects of vitamin D supplementation on outcomes that assess diabetes risk. Therefore, the purpose of this study was to determine the dose-response effects of vitamin D supplementation on fasting glucose, insulin, and a surrogate measure of insulin resistance in black and white children in the early stages of puberty. Data from a large, multisite, randomized, triple-masked, dose-response, placebo-controlled trial over 12 weeks were used in this study.
Subjects and Methods
Study participants
The study design has been previously described, and was a secondary analysis of the main RCT (19). Apparently healthy children in the early stages of puberty (N = 323; males aged 10–13 years and females aged 9–12 years) participated in the Georgia, Purdue, and Indiana University (GAPI) vitamin D supplementation trial. This 12-week study involved eight strata defined by the factorial combinations of sex, race (black and white), and latitude (southern: The University of Georgia [UGA] in Athens, GA [34° N]; northern [40° N]: Purdue University [PU] in West Lafayette, IN, and Indiana University [IU] in Indianapolis, IN). Eighty-three black girls (43 northern; 40 southern), 82 black boys (42 northern; 40 southern), 78 white girls (38 northern; 40 southern), and 80 white boys (40 northern; 40 southern) were block randomized to receive one of five vitamin D3 doses: oral placebo (0 IU vitamin D) or vitamin D3 (400, 1000, 2000, 4000 IU/day). Subjects were tested in two cohorts during the winter (October–March) of 2009–2010 and 2010–2011, when 25-hydroxyvitamin D (25(OH)D) concentrations were likely at their seasonal nadir.
Screening and testing protocol
Procedures were approved by the Institutional Review Boards for Human Subjects at UGA, PU, and IU. All participants and their parents/guardians completed informed assent and consent forms, respectively. Screening proceeded in two phases. First, potential participants who met the age requirement were screened using a medical questionnaire to assess race/ethnicity, medication and supplement use, and medical history. We recruited apparently healthy children, defined as an absence of illness requiring medication. Neither overweight nor obesity classification, defined using body mass index (BMI)-for-age percentiles, were considered exclusionary criteria. Potential subjects then completed a sexual maturation self-assessment form. Those who classified themselves as early pubertal (ie, stages 2 or 3 for genitalia and breast development as described by Tanner (20)) were scheduled for the first testing session and randomly assigned to one of the five doses. Exclusion criteria included a self-rating of pubertal maturation other than stages 2 or 3, commencement of menarche, race/ethnicity other than non-Hispanic black or white (parents and grandparents were the same race as the child), diseases including diabetes, and medication use known to affect vitamin D or glucose/insulin metabolism. If participants were taking herbal or dietary supplements before enrollment, they were required to complete a 4-week washout period before their first testing session. Subjects were asked to refrain from taking herbal or dietary supplements during the study. Five testing sessions occurred and took place at baseline and 3, 6, 9, and 12 weeks.
Vitamin D supplements
Vitamin D3 tablets were formulated by Douglas Laboratories to contain 0, 400, 1000, 2000 or 4000 IU vitamin D3 and confirmed independently (Covance Inc.) as 0.184, 486, 1140, 1880, and 4710 IU vitamin D3, respectively. The manufacturer packaged the supplements in tablet form and labeled them with product code numbers unknown to the research staff and participants. A computer-generated randomization scheme was used to assign participants to the product code (ie, treatment). Subjects were instructed to consume one tablet per day, and supplement compliance calendars were provided. At each testing session, participants were asked to return their bottles and compliance calendars to the researchers, and pill counts were performed. A participant was considered compliant if pill bottles were returned at all five times points and more than 80% of pills were consumed. Adverse events of supplemental vitamin D were collected using a questionnaire to document clinical symptoms reported by the participants, along with serum and urinary calcium concentrations, and 25(OH)D to detect hypercalcemia, hypercalciuria, and vitamin D intoxication, respectively.
Anthropometry and pubertal maturation
Anthropometric measurements were conducted at baseline and 12 weeks using the Anthropometric Standardization Reference protocol (21). Weight was measured within 0.1 kg using an electronic scale. Standing height was measured within 0.1 cm using a wall-mounted stadiometer. Participants were measured in light outdoor clothing after the removal of shoes. Height and weight were used to calculate sex- and age-specific BMI percentiles using Centers for Disease Control and Prevention growth charts (22). Children were recruited at sexual maturity states 2 and 3 using valid self-administered questionnaires for genitalia or breast development (20).
Body composition
Percentage body fat, total body fat mass (kg), and fat-free soft-tissue mass (kg) were determined by dual-energy X-ray absorptiometry (Hologic Inc. Delphi-A [UGA], Hologic Discovery-W [IU], and Lunar iDXA, GE Medical Instruments Inc. [PU]) at baseline and 12 weeks by the same technician as described previously (19). Percent body fat cutoffs of 25% or more for males and 32% or more for females were used to delineate high-fat categories, whereas percent body fat values below 25% for males and below 32% for females were used as normal-fat categories. These cut-points align with the Cooper Institute's Fitnessgram Health Fitness Zone standards for body composition (23), which are based on total body adiposity and cardiovascular disease risk results from the Bogalusa Heart Study (24).
Biochemical analyses: serum 25(OH)D, glucose, and insulin
At baseline and 6 and 12 weeks, blood samples were collected following an overnight fast for analysis of serum 25(OH)D, glucose, and insulin. Serum samples were stored at below −70 C until analysis. Reference controls (from each kit) and internal controls (in-house pooled samples) were included with each assay run for quality control. All biochemical analyses were assayed using a block design, such that baseline and 6- and 12-week samples from the same subject were assayed at one time by the same investigator using the same kit.
Serum 25(OH)D samples were assayed using a two-step RIA (DiaSorin Laboratories) and run in duplicate. The interassay coefficient of variations (CVs) ranged from 5.6 to 8.4% and the intra-assay CVs ranged from 5.5 to 7.0%. Analytical reliability of 25(OH)D assay was further monitored through participation in the Vitamin D External Quality Assessment Scheme. Serum glucose concentrations were determined in triplicate using a microtiter modification of the enzymatic Autokit Glucose method (Wako Chemicals USA). The mean intra- and interassay CVs were 1.8% and 2.2%, respectively. Serum insulin was analyzed, in duplicate, by Human Insulin Specific RIA (HI-14K; Millipore). The mean intra- and interassay CVs were 3.5% and 5.3%, respectively.
Fasting glucose and insulin values were incorporated into the homeostasis model assessment of insulin resistance (HOMA-IR) as our fasting surrogate measure of insulin resistance, where HOMA-IR = [(fasting glucose [mg/dL] × fasting insulin [uU/ml])/405] (25). HOMA-IR correlates strongly with the gold standard euglycemic hyperinsulinemic clamp technique (26, 27).
Dietary intake
Three-day diet records, a valid and reliable method for estimating energy and nutrient intakes in children (28–30), were completed at baseline and 12 weeks by participants and their parents. The dietary record specifically inquired about consumption of calcium-fortified foods and nutritional supplements. Each record detailed two weekdays and one weekend day and included time of eating, type of food and amount, and preparation method. Diet records were analyzed using Food Processor SQL, version 9.7.3 (ESHA Research), were entered by two researchers, and were statistically compared for agreement.
Statistical analyses
Descriptive statistics were computed on the raw data and model diagnostics were examined for all fitted models to check for model violations. Three-way ANOVA was used to compare baseline variables across race, gender, and body fat groups and to assess interactions.
Linear mixed-effects models were used to implement repeated measures ANOVA (RM-ANOVA) to examine the effect of vitamin D supplementation on the three dependent variables: glucose, insulin, and HOMA-IR. These models assumed a general variance-covariance structure to capture within-subject correlation and heteroscedasticity, rather than the compound symmetry structure of classical RM-ANOVA. Models included main effects of treatment and time, and interactions between these factors. Secondarily, repeated measures analysis of covariance models were considered to control for possible confounding or effect-modifying effects of baseline covariates. In particular, sex, race, pubertal maturation stage, maturity offset, and percent body fat were considered for inclusion in the models and retained based upon a backward selection procedure, using α = 0.1 for retention in the model and following rules of hierarchy. Analyses were conducted on an intent-to-treat basis.
Secondary analyses used Pearson correlations to quantify associations between baseline and percent change values of biochemical variables and age, anthropometry, maturation, body composition, and dietary intake. Partial correlations were utilized to relate baseline 25(OH)D with dependent variables, controlling for race, gender, age, pubertal maturation, fat mass, and BMI. For all analyses, insulin, and HOMA-IR were analyzed on the natural log scale to correct for heteroscedasticity, and α = 0.05 was used to denote statistical significance. The mixed-effects model analyses were computed using SAS, version 9.2; other analyses were performed using SPSS, version 18.03.
Sample size (19) was chosen to detect differences in 12-week changes in 25(OH)D, one of the primary outcomes for the parent GAPI trial. Forty-eight subjects per group provided more than 90% power using an α = 0.05. Allowing for an attrition rate of 20% (based on our prospective childhood studies) and equal randomization to the five treatment arms, n = 64/group were recruited (N = 320).
Results
Participant characteristics
In the sample of 323 children, 299 completed the 12-week intervention (ie, 93% retention rate). Across all time points, approximately 12% (40 of 323) of participants failed to return their bottles for pill counts, but overall compliance did not differ across treatments. Based on Centers for Disease Control and Prevention BMI-for-age percentiles, 2% of participants were classified as underweight, 55% normal weight, 21% overweight, and 21% obese. Based on fasting glucose measurements, 5% of the children met fasting prediabetic criteria during at least one measurement. Baseline descriptive characteristics of the participants are presented in Table 1. The following group differences were statistically significant: males had more lean mass and were older, heavier, taller, and more mature than the females, along with having higher energy intakes than the females. Females had higher fat mass, percent body fat, BMI percentile, insulin, and HOMA-IR than their male counterparts. Racial differences included higher BMI percentile, insulin, and HOMA-IR for black subjects as well as lower 25(OH)D and calcium intake. Their white counterparts were older. Along with being heavier and having more fat mass, children in the high-fat vs normal-fat category were younger and had higher HOMA-IR (Table 1). Black children in the 34°N latitude compared to the 40°N latitude had higher serum 25(OH)D vitamin D, pubertal stage, and lean mass and lower percent body fat and dietary vitamin D intake (data not shown). There were no differences in baseline characteristics by dose (19).
Table 1.
Baseline Participant Characteristics Overall and Divided by Gender, Race, and Body Fat Categoriesa,b
| Overall (n = 323) | Male (n = 162) | Female (n = 161) | White (n = 158) | Black (n = 165) | Normal-Fat (n = 132) | High-Fat (n = 188) | Differencesc P < .05 | |
|---|---|---|---|---|---|---|---|---|
| Age (y) | 11.3 ± 0.07 | 12.0 ± .10 | 10.7 ± .10 | 11.6 ± .10 | 11.2 ± .10 | 11.6 ± .10 | 11.2 ± 0.09 | M > F, W > B, NF > HF Sex × %Fat |
| Anthropometry | ||||||||
| Weight (kg) | 47.4 ± .70 | 48.6 ± 1.0 | 46.1 ± .80 | 46.1 ± 1.0 | 48.6 ± 1.0 | 41.5 ± .75 | 51.6 ± .92 | M > F, HF > NF Race × Sex × %Fat |
| Height (cm) | 150.7 ± .50 | 152.7 ± .80 | 148.7 ± .70 | 151.3 ± .80 | 150.1 ±.70 | 151.2 ± .79 | 150.3 ± .69 | M > F Sex × %Fat |
| BMI-for-age (%) | 68.0 ± 1.6 | 64.4 ± 2.4 | 71.6 ± 2.2 | 61.3 ± 2.4 | 74.4 ± 2.1 | 47.9 ± 2.4 | 82.9 ± 1.5 | F > M, B > W, HF > NF Race × Sex × %Fat |
| Maturation | ||||||||
| Pubertal maturationd | 2.4 ± 0.03 | 2.4 ± 0.04 | 2.3 ± 0.04 | 2.3 ± 0.04 | 2.4 ± 0.04 | 2.4 ± 0.05 | 2.4 ± 0.04 | M > F |
| Body compositione | ||||||||
| Fat mass (kg) | 14.9 ± .40 | 13.7 ± .60 | 16.1 ± 0.6 | 14.2 ± .60 | 15.5 ± .60 | 9.1 ± .24 | 18.8 ± .51 | F > M, HF > NF |
| Percent body fat | 31.2 ± .50 | 28.2 ± .70 | 34.4 ± 0.7 | 30.5 ± .80 | 31.8 ± .80 | 22.9 ± .44 | 36.9 ± .51 | F > M, HF > NF |
| Lean mass (kg) | 30.1± .40 | 32.3 ± .60 | 28.5 ± 0.4 | 30.4 ± .60 | 30.6 ± .50 | 30.2 ± .57 | 30.5 ± .52 | F > M |
| Biochemical | ||||||||
| 25(OH)D (nmol/liter) | 70.0 ± 1.0 | 70.6 ± 1.3 | 69.1 ± 1.6 | 80.0 ± 1.2 | 60.1 ± 1.3 | 72.4 ± 1.6 | 67.6 ± 1.3 | W > B |
| Serum glucose (mg/dl) | 89.0 ± .40 | 89.3 ± .60 | 88.7 ± 0.6 | 88.5 ± .60 | 89.4 ± .60 | 88.5 ± .65 | 89.4 ± .51 | Race × %Fat |
| Serum insulin (uU/ml) | 20.0 ± .60 | 18.5 ± .80 | 21.4 ± 0.8 | 18.1 ± .80 | 21.8 ± .80 | 15.8 ± .59 | 22.2 ± .86 | F > M, B > W, HF > NF |
| Race × Sex | ||||||||
| Race × Sex × %Fat | ||||||||
| HOMA-IR | 4.4 ± .10 | 4.1 ± .2.0 | 4.7 ± 0.2 | 4.0 ± .20 | 4.8 ± .20 | 3.6 ± .13 | 5.0 ± .20 | F > M, B > W, HF > NF |
| Race × Sex | ||||||||
| Race × %Fat | ||||||||
| Sex × %Fat | ||||||||
| Race × Sex × %Fat | ||||||||
| Dietary intake | ||||||||
| Energy intake (kcal/day) | 2001 ± 31 | 2085 ± 46 | 1914 ± 43 | 2061 ± 44 | 1939 ± 45 | 2013 ± 51 | 1984 ± 41 | M > F |
| Vitamin D intake (IU/day) | 169.2 ± 7.1 | 178.8 ± 10.8 | 159.5 ± 9.2 | 173.0 ± 11.6 | 165.4 ± 8.1 | 162.1 ± 10.3 | 175.9 ± 9.8 | |
| Calcium intake (mg/day) | 901 ± 23 | 907 ± 33 | 894 ± 31 | 964 ± 34 | 837 ± 29 | 922.9 ± 36 | 890.7 ± 29 | W > B Sex × %Fat |
Abbreviations: B, black; F, female; HF, high-fat; M, male; NF, normal-fat; W, white.
Values are presented as means ± se.
Overall characteristics represent data collapsed across the five treatment groups. NF was defined as a percent body fat <25% for males and <32% for females. HF was defined as a percent body fat ≥25% for males and ≥32% for females.
Results of three-way ANOVA for sex, race, adiposity as %fat, and all interactions investigated. Differences shown are significant at α = 0.05.
Stages based upon secondary sexual characteristics as described by Tanner.
Body composition measures assessed using dual energy x-ray absorptiometry.
Relationships at baseline
Baseline serum 25(OH)D was negatively associated with weight and BMI percentile. Insulin and HOMA-IR were positively associated with weight, height, BMI percentile, fat mass, percent body fat, and lean mass (Table 2). Pubertal maturation was positively associated with insulin (r = 0.159, P = .005) and HOMA-IR (r = 0.148, P = .008). Baseline 25(OH)D was inversely associated with insulin (r = −0.140, P = .017) and HOMA-IR (r = −0.146, P = .012), after controlling for race, gender, age, pubertal maturation, fat mass, and BMI (Figure 1).
Table 2.
Relationships Among Biochemical Parameters, Anthropometry, and Body Compositiona
| Weight |
Height |
BMI Percentile |
Fat Mass |
Percent Body Fat |
Lean Mass |
|||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| r | P Value | r | P Value | r | P Value | r | P Value | r | P Value | r | P Value | |
| 25(OH)D | −0.177b | .002 | −0.81 | .152 | −0.163b | .003 | −0.139b | .016 | −0.085 | .143 | −0.111 | .054 |
| Serum glucose | 0.027 | .630 | 0.001 | .983 | 0.071 | .211 | 0.032 | .581 | 0.040 | .491 | 0.007 | .909 |
| Serum insulin | 0.498b | <.001 | 0.195b | <.001 | 0.473b | <.001 | 0.511b | <.001 | 0.413b | <.001 | 0.290b | <.001 |
| HOMA-IR | 0.472b | <.001 | 0.183b | .001 | 0.461b | <.001 | 0.486b | <.001 | 0.395b | <.001 | 0.273b | <.001 |
Data presented as unadjusted Pearson correlations and accompanying P values. Insulin and HOMA-IR were analyzed on the natural log scale to correct for nonconstant variance observed on the original scale.
P < .05.
Figure 1.
Partial associations between baseline 25-hydroxyvitamin D, fasting glucose, insulin, and HOMA-IR. A, Glucose and 25(OH)D. B, Insulin and 25(OH)D. C, HOMA-IR and 25(OH)D. Partial correlations were adjusted for race, sex, age, pubertal maturation, fat mass, and BMI. The figure shows residual associations among the variables of interest after removing effects of the adjustment variables. Insulin and HOMA-IR were analyzed on the natural log scale to correct for nonconstant variance observed on the original scale.
Vitamin D supplementation effects on 25(OH)D, glucose, insulin, and insulin resistance
Serum 25(OH)D ranged from 25.3 to 114.7 nmol/liter at baseline, with 15% of the sample having 25(OH)D lower than 50 nmol/liter (insufficiency), 6% having lower than 40 nmol/liter (Estimated Average Requirement), and 1% having lower than 30 nmol/liter (deficiency). The average baseline and 12-week 25(OH)D concentrations, respectively, were: 71.3 and 63.4 nmol/liter for the placebo group, 71.2 and 76.6 nmol/liter for the 400-IU group, 71.0 and 90.2 nmol/liter for the 1000-IU group, 65.7 and 101.8 nmol/liter for the 2000-IU group, and 69.9 and 147.3 nmol/liter for the 4000-IU group. After 12 weeks, 25(OH)D concentrations ranged from 24.2 to 237.4 nmol/liter, with 9% having serum 25(OH)D lower than 50 nmol/liter.
The effect of vitamin D supplementation on fasting glucose, insulin, and HOMA-IR over time was nonsignificant for all outcome measures (Figure 2) (glucose, F = 0.62, P = .762; insulin, F = 0.41, P = .918; and HOMA-IR, F = 0.46, P = .883. Glucose (F = 5.79, P = .0034), insulin (F = 11.59, P < .0001), and HOMA-IR (F = 12.41, P < .0001) all increased over time. Baseline covariates did not affect the results. Additionally, percent change over 12 weeks in 25(OH)D was not significantly associated with percent change in glucose (r = −0.032, P = .578), insulin (r = 0.001, P = .987), or HOMA = IR (r = −0.01, P = .924).
Figure 2.
Profile plots of estimated joint treatment means for fasting glucose, insulin, and HOMA-IR over 12 weeks of supplementation. A, Glucose. B, Insulin. C, HOMA-IR. Linear mixed model analyses using RM-ANOVA were used to examine the effect of vitamin D supplementation on the three dependent variables: glucose, insulin, and HOMA-IR. Profile plots represent the treatment × time interaction. Insulin and HOMA-IR were analyzed on the natural log scale to correct for nonconstant variance observed on the original scale. No baseline differences existed for glucose, insulin, or HOMA-IR.
Model diagnostics indicated substantial skewness for both insulin and HOMA-IR. Therefore, insulin and HOMA-IR were log-transformed in all statistical analyses. To check the influence of a small number of outliers on the results, analyses were repeated omitting any observations with normalized residuals greater than 3.5 in magnitude. Omitting such outliers had no qualitative effects on the results and negligible quantitative effects, so we report only the original analyses here.
Discussion
To our knowledge, this is the first study to examine the effects of vitamin D supplementation on fasting glucose, insulin, and a surrogate measure of insulin resistance in apparently healthy children entering the early stages of puberty. The primary finding was that vitamin D supplementation at various step-wise potencies did not alter serum glucose, insulin, or insulin resistance in these children over the course of the 12-week trial, despite the fact that basal inverse relationships existed between serum 25(OH)D and these measures of glucose, insulin and insulin resistance.
Vitamin D supplementation at 0, 400, 1000, 2000, or 4000 IU/day over 12 weeks did not affect the trajectory of glucose, insulin, or HOMA-IR outcomes in our sample of children. Other pediatric RCTs on this topic were conducted in older, more sexually mature children (14–16, 18). Differences in the stages of pubertal maturation are important to consider when interpreting these null results because the phenomenon of transient insulin resistance that occurs before and during the pubertal transition in pre-, early, and peri-pubertal youth, independent of adiposity status in both sexes, has been characterized in both cross-sectional and longitudinal studies (31, 32). The mechanisms underlying this reduction in insulin sensitivity is not fully understood, but the increase in fat mass, IGF-1, and the adrenal hormone dehydroepiandrosterone during this period of growth are thought to play a role (33). The significant increase in all three outcomes in this study suggests that this phenomenon was occurring organically in our children over the course of the trial. We observed that female, black, and high-fat children were more insulin resistant than male, white, and normal-fat children, respectively. We addressed these possible confounding variables statistically; however, none remained significant in the final models.
The baseline health status of participants in pediatric vitamin D trials may contribute to the differential findings among RCTs to date. Subjects in RCTs showing positive results had established metabolic syndrome (17) or were at high cardiometabolic risk, as obesity patients at weight management centers (14, 18). The authors of these studies suggest that having established glucose–insulin dysfunction or risk may contribute to the favorable effects of vitamin D supplementation on glucose homeostasis, perhaps via vitamin D repletion helping to address cardiometabolic pathogeneses (eg, β-cell function for insulin release, insulin sensitivity/action) before they are full-fledged. However, two other pediatric vitamin D trials in high-risk (obese) adolescents did not find positive results on fasting glucose or oral glucose tolerance test measures (15, 16). Although only 5% of our participants had evidence of prediabetes using fasting glucose measures, a significant subset (42%) of the overall cohort had BMI levels in the overweight (21%) or obese (21%) ranges, prevalences higher than national averages in youth ages 2–19 years (34, 35). Additionally, although our overall sample displayed high HOMA-IR values (mean = 4.4), which is higher than the estimated cutoff point (>4) to identify insulin resistance in children (36), vitamin D supplementation did not mitigate this insulin resistance.
Our study and the two other null pediatric studies (15, 16) differ from the aforementioned positive studies in two important ways: 1) vitamin D dose and 2) baseline vitamin D status. The studies with null findings used a 400–4000 IU daily vitamin D dose in subjects with vitamin D sufficiency at baseline (ie, 60–70 nmol/liter 25(OH)D on average), whereas the studies demonstrating positive vitamin D effects used higher vitamin D doses (ie, 4000 IU/day; 50 000 IU/week; 300 000 IU/week) in subjects with a lower baseline vitamin D (ie, 27–45 nmol/liter 25(OH)D). In that way, the studies with supplementation regimens designed to address inadequate vitamin D status appear more likely to produce positive vitamin D effects on diabetes risk factors in high-risk patients.
Most children (approximately 85%) in the GAPI study were considered vitamin D “sufficient” (>50 nmol/liter) at baseline based on IOM standards for skeletal outcomes in adults, though the range spanned 25–115 nmol/liter. Because populations living at higher US latitudes have lower 25(OH)D values, 50% of our participants were recruited from the northern latitude (40°N). Also, participants were tested only in the wintertime because of the seasonal nadir in serum 25(OH)D. Nonetheless, both strategies generated a sample with 15% having low baseline 25(OH)D concentrations. This prevalence is slightly lower than the 18% national prevalence estimate for children with 25(OH)D <50 nmol/liter (37). It remains to be determined whether the IOM 50 nmol/liter cutoff for 25(OH)D sufficiency is applicable to extraskeletal outcomes.
After controlling for adiposity measures at baseline, we observed significant inverse relationships between serum 25(OH)D, insulin and insulin resistance. Similar relationships have been previously demonstrated in children, including large, nationally representative samples (8, 9). Though these cross-sectional findings demonstrate that vitamin D status correlates with diabetes risk factors, the basal relationships did not translate into causality in our sample. Supplementation duration (12 weeks) did not differ from other pediatric vitamin D trials able to demonstrate significant effects (8 weeks (18), 12 weeks (17), or 6 months (14)) vs those with null findings (12 weeks (15, 16)). Short- and long-term adult vitamin D RCTs to date have also netted null findings for diabetes-related outcomes; although heterogeneity, variable risk of bias, and short-term follow-up duration of the available evidence are key limitations (13). Results from the ongoing Vitamin D and type 2 Diabetes study (38) may provide clarity into the role of vitamin D in diabetes development in adults.
Our study possesses strengths, including a rigorous study design. The study was randomized, placebo-controlled, and triple-masked. It included four treatment arms of sufficient dose range, two latitudes, a large sample size, both genders, and two races, and controlled for season of testing. To date, it is the most comprehensive study to assess the effects of vitamin D on fasting glucose and insulin measures in children. Additionally, we aimed to examine the effect of pertinent confounders (including age, race, sex, maturation, and adiposity) on the supplementation results over time. Our study also has limitations, such as the potential for error with self-assessment of sexual maturation. Also, we elected to use a surrogate measure of insulin resistance; however, HOMA-IR has been validated across a variety of physiological methods (26, 27) and across the five stages of sexual maturation (32).
In summary, we report findings from the first pediatric, dose-response vitamin D intervention on diabetes risk factors in apparently healthy children entering the early stages of puberty. Twelve weeks of vitamin D supplementation had no effect on glucose, insulin, or HOMA-IR, despite the presence of a significant subset of participants who were overweight, obese, and insulin resistant. Basal relationships observed among serum 25(OH)D, glucose, and insulin resistance persisted after controlling for adiposity, and alterations in glucose homeostasis during the course of the study were characteristic of those documented to occur during the pubertal transition. Though vitamin D supplementation is an important, affordable way to help achieve 25(OH)D sufficiency in children, vitamin D treatment does not mitigate alterations in glucose homeostasis during the pubertal transition in healthy children with sufficient baseline 25(OH)D.
Acknowledgments
We thank Ms. Ruth Gildea Taylor and Ms. Jessica Smith for their overall coordination of this project, Mr. Christian Wright and Dr. Lauren Atwell for their technical assistance, and the participants and their families for their commitment to furthering research.
This work was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development Grant RO1HD057126, by Allen Foundation grant 2.008.319, and by the University of Georgia Agricultural Experiment Station, HATCH projects GEO00647 and GEO00700.
ClinicalTrials.gov: NCT00931580.
Disclosure Summary: The authors have nothing to disclose.
Footnotes
- BMI
- body mass index
- CV
- coefficient of variation
- GAPI
- Georgia, Purdue, and Indiana University
- IOM
- Institute of Medicine
- HOMA-IR
- homeostasis model assessment of insulin resistance
- IU
- Indiana University
- 25(OH)D
- 25-hydroxyvitamin D
- PU
- Purdue University
- RM-ANOVA
- repeated measures ANOVA
- RCT
- randomized controlled trial
- UGA
- University of Georgia.
References
- 1. Bullard KM, Saydah SH, Imperatore G, et al. Secular changes in U.S. Prediabetes prevalence defined by hemoglobin A1c and fasting plasma glucose: National Health and Nutrition Examination Surveys, 1999–2010. Diabetes Care. 2013;36:2286–2293. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Li C, Ford ES, Zhao G, Mokdad AH. Prevalence of prediabetes and its association with clustering of cardiometabolic risk factors and hyperinsulinemia among U.S. adolescents: National Health and Nutrition Examination Survey 2005–2006. Diabetes Care. 2009;32:342–347. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Centers for Disease Control and Prevention. National Diabetes Statistics Report: Estimates of Diabetes and Its Burden in the United States. Atlanta, GA: US Department of Health and Human Services; 2014. [Google Scholar]
- 4. Fulgoni VL, 3rd, Keast DR, Bailey RL, Dwyer J. Foods, fortificants, and supplements: where do Americans get their nutrients? J Nutr. 2011;141:1847–1854. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Norman AW, Frankel JB, Heldt AM, Grodsky GM. Vitamin D deficiency inhibits pancreatic secretion of insulin. Science. 1980;209:823–825. [DOI] [PubMed] [Google Scholar]
- 6. Maestro B, Campion J, Davila N, Calle C. Stimulation by 1,25-dihydroxyvitamin D3 of insulin receptor expression and insulin responsiveness for glucose transport in U-937 human promonocytic cells. Endocr J. 2000;47:383–391. [DOI] [PubMed] [Google Scholar]
- 7. Maestro B, Molero S, Bajo S, Davila N, Calle C. Transcriptional activation of the human insulin receptor gene by 1,25-dihydroxyvitamin D(3). Cell Biochem Function. 2002;20:227–232. [DOI] [PubMed] [Google Scholar]
- 8. Ganji V, Zhang X, Shaikh N, Tangpricha V. Serum 25-hydroxyvitamin D concentrations are associated with prevalence of metabolic syndrome and various cardiometabolic risk factors in US children and adolescents based on assay-adjusted serum 25-hydroxyvitamin D data from NHANES 2001–2006. Am J Clin Nutr. 2011;94:225–233. [DOI] [PubMed] [Google Scholar]
- 9. Reis JP, von Muhlen D, Miller ER, 3rd, Michos ED, Appel LJ. Vitamin D status and cardiometabolic risk factors in the United States adolescent population. Pediatrics. 2009;124:e371–e379. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Ford ES, Zhao G, Tsai J, Li C. Associations between concentrations of vitamin D and concentrations of insulin, glucose, and HbA1c among adolescents in the United States. Diabetes Care. 2011;34:646–648. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Mitri J, Muraru MD, Pittas AG. Vitamin D and type 2 diabetes: a systematic review. Eur J Clin Nutr. 2011;65:1005–1015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Ross CA, Yaktine AL, Del Valle HB. Committee to Review Dietary Reference Intakes for Vitamin D and Calcium. Institute of Medicine; 2010. [Google Scholar]
- 13. Seida JC, Mitri J, Colmers IN, et al. Clinical review: effect of vitamin D3 supplementation on improving glucose homeostasis and preventing diabetes: a systematic review and meta-analysis. J Clin Endocrinol Metab. 2014;99:3551–3560. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Belenchia AM, Tosh AK, Hillman LS, Peterson CA. Correcting vitamin D insufficiency improves insulin sensitivity in obese adolescents: a randomized controlled trial. Am J Clin Nutr. 2013;97:774–781. [DOI] [PubMed] [Google Scholar]
- 15. Nader NS, Aguirre Castaneda R, Wallace J, Singh R, Weaver A, Kumar S. Effect of vitamin D3 supplementation on serum 25(OH)D, lipids and markers of insulin resistance in obese adolescents: a prospective, randomized, placebo-controlled pilot trial. Hormone Res Paediatr. 2014;82:107–112. [DOI] [PubMed] [Google Scholar]
- 16. Javed A, Vella A, Balagopal PB, et al. Cholecalciferol supplementation does not influence beta-cell function and insulin action in obese adolescents: a prospective double-blind randomized trial. J Nutr. 2015;145:284–290. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Kelishadi R, Salek S, Salek M, Hashemipour M, Movahedian M. Effects of vitamin D supplementation on insulin resistance and cardiometabolic risk factors in children with metabolic syndrome: a triple-masked controlled trial. J Pediatr (Rio J). 2014;90:28–34. [DOI] [PubMed] [Google Scholar]
- 18. Ashraf AP, Alvarez JA, Gower BA, Saenz KH, McCormick KL. Associations of serum 25-hydroxyvitamin D and components of the metabolic syndrome in obese adolescent females. Obesity (Silver Spring). 2011;19:2214–2221. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Lewis RD, Laing EM, Hill Gallant KM, et al. A randomized trial of vitamin D(3) supplementation in children: dose-response effects on vitamin D metabolites and calcium absorption. J Clin Endocrinol Metab. 2013;98:4816–4825. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Tanner J. Growth and Adolescence. 2nd ed Oxford: Blackwell Scientific Publications; 1962. [Google Scholar]
- 21. Lohman TG, Roche AF, Martorell R. Anthropometric Standardization Reference Manual. Champaign, IL: Human Kinetics; 1988. [Google Scholar]
- 22. Kuczmarski RJ, Ogden CL, Grummer-Strawn LM, et al. CDC growth charts: United States. Adv Data. 2000;314:1–27. [PubMed] [Google Scholar]
- 23. Going SB, Lohman TG, Falls HB. Body composition assessments. Fitnessgram Reference Guide. 2008; http://www.cooperinstitute.org/reference-guide Accessed February 17, 2016.
- 24. Williams DP, Going SB, Lohman TG, et al. Body fatness and risk for elevated blood pressure, total cholesterol, and serum lipoprotein ratios in children and adolescents. Am J Public Health. 1992;82:358–363. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC. Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia. 1985;28:412–419. [DOI] [PubMed] [Google Scholar]
- 26. Wallace TM, Levy JC, Matthews DR. Use and abuse of HOMA modeling. Diabetes Care. 2004;27:1487–1495. [DOI] [PubMed] [Google Scholar]
- 27. Gungor N, Saad R, Janosky J, Arslanian S. Validation of surrogate estimates of insulin sensitivity and insulin secretion in children and adolescents. J Pediatr. 2004;144:47–55. [DOI] [PubMed] [Google Scholar]
- 28. Taylor RW, Goulding A. Validation of a short food frequency questionnaire to assess calcium intake in children aged 3 to 6 years. Eur J Clin Nutr. 1998;52:464–465. [DOI] [PubMed] [Google Scholar]
- 29. Bergman EA, Boyungs JC, Erickson ML. Comparison of a food frequency questionnaire and a 3-day diet record. Journal of American Dietetic Association. 1990;90(4):1431–1433. [PubMed] [Google Scholar]
- 30. Crawford PB, Obarzanek E, Morrison J, Sabry ZI. Comparative advantage of 3-day food records over 24-hour recall and 5- day food frequency validated by observation of 9- and 10-year-old girls. J Am Diet Assoc. 1994;94:626–630. [DOI] [PubMed] [Google Scholar]
- 31. Goran MI, Gower BA. Longitudinal study on pubertal insulin resistance. Diabetes. 2001;50:2444–2450. [DOI] [PubMed] [Google Scholar]
- 32. Guzzaloni G, Grugni G, Mazzilli G, Moro D, Morabito F. Comparison between beta-cell function and insulin resistance indexes in prepubertal and pubertal obese children. Metabolism. 2002;51:1011–1016. [DOI] [PubMed] [Google Scholar]
- 33. Jeffery AN, Metcalf BS, Hosking J, Streeter AJ, Voss LD, Wilkin TJ. Age before stage: insulin resistance rises before the onset of puberty: a 9-year longitudinal study (EarlyBird 26). Diabetes Care. 2012;35:536–541. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Ogden CL, Carroll MD, Curtin LR, Lamb MM, Flegal KM. Prevalence of high body mass index in US children and adolescents, 2007–2008. JAMA. 2010;303:242–249. [DOI] [PubMed] [Google Scholar]
- 35. Ogden CL, Carroll MD, Kit BK, Flegal KM. Prevalence of obesity and trends in body mass index among US children and adolescents, 1999–2010. JAMA. 2012;307:483–490. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Reinehr T, Andler W. Changes in the atherogenic risk factor profile according to degree of weight loss. Arch Dis Child. 2004;89:419–422. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Mansbach JM, Ginde AA, Camargo CA., Jr Serum 25-hydroxyvitamin D levels among US children aged 1 to 11 years: do children need more vitamin D? Pediatrics. 2009;124:1404–1410. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Pittas AG, Dawson-Hughes B, Sheehan PR, et al. , D2d Research Group. Rationale and design of the Vitamin D and Type 2 Diabetes (D2d) study: a diabetes prevention trial. Diabetes Care. 2014;37:3227–3234. [DOI] [PMC free article] [PubMed] [Google Scholar]