Abstract
Context
Vitamin D may be important for prenatal programming of insulin and glucose regulation, but maternal vitamin D deficiency during pregnancy is common.
Objective
We examined associations of early vitamin D status with markers of fetal insulin secretion: cord blood insulin and c-peptide. We hypothesized that maternal 25-hydroxyvitamin D (25(OH)D) during pregnancy and cord blood 25(OH)D would both be positively associated with cord blood insulin and c-peptide.
Methods
We studied mother-newborn pairs from two cohorts: Project Viva (n = 862 pairs included) and Genetics of Glucose Regulation in Gestation and Growth (Gen3G; n = 660 pairs included). We analyzed associations of the cord blood hormones with maternal 25(OH)D using generalized additive models with nonlinear spline terms, and with cord blood 25(OH)D using multivariable linear regression models.
Results
The 25(OH)D levels were <75 nmol/L in >70% of mothers and 85% of newborns. Maternal and cord blood 25(OH)D levels were correlated (Project Viva, r = 0.58; Gen3G, r = 0.37). Maternal 25(OH)D had an inverted-U–shaped relationship with cord blood insulin and c-peptide in both cohorts. Cord blood 25(OH)D had a linear relationship with the cord blood hormones. In fully adjusted models, each 10-nmol/L increase in cord blood 25(OH)D was associated with higher cord blood insulin and c-peptide concentrations: 3.7% (95% CI, 0.09 to 7.5) and 3.2% (95% CI, 0.8 to 5.6), respectively, in Project Viva; 2.2% (95% CI, −0.1 to 4.6) and 3.6% (95% CI, 1.0 to 6.3), respectively, in Gen3G.
Conclusion
Vitamin D may play a role in regulating fetal insulin secretion, potentially affecting glucose regulation and growth.
In two separate cohorts of mother-newborn pairs with high prevalence of vitamin D insufficiency, we observed linear associations of cord blood 25(OH)D with cord blood insulin and c-peptide.
Vitamin D is a fat-soluble nutrient essential for many aspects of human health, and maternal vitamin D status during pregnancy has been associated with markers of fetal growth and development (1–3). The circulating form of vitamin D, 25-hydroxyvitamin D [25(OH)D], crosses the placenta, and the developing fetus relies on the mother for vitamin D (4, 5). However, vitamin D insufficiency is common in pregnant women, especially in regions with limited ambient UVB light (5).
In adults, vitamin D plays a role in regulating insulin secretion, and low vitamin D status has been associated with insulin resistance and impaired β cell function in several populations (6). Maternal vitamin D insufficiency during pregnancy has been linked to impaired fetal growth (2, 3), increased risk of gestational diabetes mellitus (GDM) (7), and adverse postnatal growth patterns and insulin resistance in the offspring (8). Some evidence suggests that low gestational vitamin D exposure may be associated with later development of type 1 diabetes (4). Thus, maternal vitamin D status may contribute to prenatal programming of the insulin axis and glucose regulation. However, other observational studies have not found associations of maternal 25(OH)D concentrations during pregnancy with markers of fetal growth (9) or fetal insulin production (9–11), and a recent review of randomized controlled trials that evaluated the impact of maternal vitamin D supplementation during pregnancy found only weak evidence of an association with fetal growth (12).
Cord blood 25(OH)D concentration reflects fetal vitamin D status in late pregnancy and is closely correlated with maternal 25(OH)D. However, maternal vitamin D sufficiency does not ensure fetal and neonatal vitamin D sufficiency, because cord blood levels are only about 60% to 80% of maternal levels (10, 13) and are influenced by maternal body mass index (BMI) (14). Maternal and cord blood 25(OH)D concentrations have been associated with measures of fetal growth. Neonates with higher cord blood 25(OH)D levels have higher adiposity (10, 14), and maternal 25(OH)D deficiency predicts lower birth weight and higher risk of being small for gestational age (3). Measures of fetal growth are also positively associated with cord blood insulin and c-peptide concentrations (15), which are indicators of fetal energy supply. Higher cord blood insulin or c-peptide concentrations have been associated among girls with slower growth in infancy and early childhood (16–18).
Although adequate vitamin D status is important for glycemic control in adults, the association of vitamin D with insulin secretion during fetal life is not clear. Because both vitamin D and adequate nutrient supply are important for fetal adiposity accumulation and weight gain, vitamin D may play a different role in insulin secretion and β- cell function during the prenatal period than in childhood and adulthood. The aim of this study was to examine associations of measures of prenatal vitamin D status with markers of fetal insulin secretion. We performed separate analyses in two prebirth cohorts of mother-newborn pairs based in high-latitude regions where there is high risk of 25(OH)D deficiency, which allowed us to replicate our findings in two similar but independent populations. We hypothesized that maternal circulating 25(OH)D levels during pregnancy and cord blood 25(OH)D levels would both be positively associated with cord blood insulin and c-peptide concentrations. We accounted for several potential confounders, including maternal prepregnancy BMI, which is inversely associated with vitamin D status but positively predicts birth weight (14, 19) and offspring insulin resistance (20); and gestational weight gain (GWG), a marker of fetal under- or overnutrition, which predicts fetal insulin secretion and growth (21, 22) and may also be associated with fetal vitamin D status. We also examined effect modification by child sex, based on evidence suggesting that influences on fetal hormone secretion may be sex specific (23, 24).
Methods
Study population and field data collection
We studied mother-infant pairs enrolled in two cohorts based in high-latitude regions with limited ambient sunlight during much of the year: Project Viva and Genetics of Glucose Regulation in Gestation and Growth (Gen3G). Project Viva recruited pregnant women during 1999 through 2002 at their initial obstetric appointment from eight offices of Atrius Health, a large multispecialty group practice in eastern Massachusetts. Gen3G recruited pregnant women during 2010 through 2013 at their first-trimester visit to the Centre hospitalier universitaire de Sherbrooke in Sherbrooke, Quebec, Canada. Detailed recruitment and retention procedures have been described previously for both cohorts (25, 26).
Data and blood samples were collected from the Project Viva and Gen3G cohorts in the first and second trimesters of pregnancy; at delivery, cord blood was collected from the umbilical vein, and additional data were collected on the mother and newborn. For this study, we used plasma samples from the first- (Gen3G: mean, 9.6 weeks) or second- (Project Viva: mean, 27.9 weeks) trimester study visits, data on potential confounders collected during pregnancy, and umbilical cord blood and birth outcome data collected at delivery. There are 2,128 liveborn singleton infants in the Project Viva cohort. For this analysis, we included 862 with available data on cord blood insulin and c-peptide concentrations and either second-trimester plasma 25(OH)D levels (n = 641) or cord blood 25(OH)D levels (n = 842). Compared with mother-infant pairs who were not included, a slightly higher proportion of included Project Viva mothers were multiparous (50% vs 55%) and white (64% vs 70%), and the included subset had a lower proportion of infants born preterm (<37 weeks, 8.8% vs 5.0%). There were no differences between included and excluded pairs in maternal education, age, BMI, gestational weight gain, smoking status, diagnosis of GDM, or child season of birth.
There are 854 mother-infant pairs in the Gen3G cohort with data at delivery, and we included 660 pairs with data on cord blood insulin and c-peptide concentrations and with first-trimester plasma 25(OH)D levels (n = 619) or cord blood 25(OH)D levels (n = 650). Compared with mother-infant pairs who were not included, a lower proportion of included Gen3G mothers smoked during pregnancy (15% vs 9%). There were no differences between included and excluded pairs in maternal race or ethnicity, BMI, age, gestational weight gain, parity, GDM diagnosis, season of visit, or child sex or gestational age at birth. Institutional review boards of institutions participating in both projects approved the study protocols, and mothers in both studies provided written informed consent for their own and their child’s participation.
Measurement of 25(OH)D and cord blood insulin and c-peptide
In Project Viva, trained phlebotomists collected maternal blood at the time of the routine nonfasting glucose challenge test (mean, 27.9 weeks’ gestation). Clinicians collected cord blood samples via syringe from the umbilical vein after delivery. Whole-blood samples were refrigerated for <24 hours, then centrifuged at 2000 rpm at 4°C for 10 minutes and aliquoted for storage in liquid nitrogen at −80°C (27). Project Viva obtained duplicate measurements of 25(OH)D [a combination of 25(OH)D2 and 25(OH)D3] for each sample using an automated chemiluminescence immunoassay and a manual RIA. The laboratory used US National Institute of Standards and Technology level 1 for quality control. As we have done previously (28–30), we used the average of the two values for each sample in our analysis to obtain more-stable estimates of 25(OH)D. Insulin and c-peptide levels were measured in the Project Viva cohort using a competitive electrochemiluminescence immunoassay (Roche Diagnostics, Indianapolis, IN). Day-to-day variability of the assay was <10%. In Gen3G, trained research staff collected nonfasting maternal blood samples at the first study visit (mean, 9.6 weeks’ gestation) and cord blood samples via syringe from the umbilical vein after delivery. Aprotinin (1 μL/mL of blood) was added to blood samples, which were then centrifuged at 2500g at 4°C for 10 minutes and aliquoted for storage at −80°C (26). In the Gen3G cohort, 25(OH)D2 and 25(OH)D3 concentrations were measured using liquid-liquid extraction followed by liquid chromatography–electrospray tandem mass spectrometry (Quattro micro mass spectrometer; Waters, Milford, MA), and total 25(OH)D was calculated as the sum of the two measurements. Intra- and interassay coefficients of variance were <10%. Insulin and c-peptide concentrations were measured in the Gen3G cohort using a multiplexed particle-based flow cytometric assay (Human Milliplex map kits; EMD Millipore, Billerica, MA). Intra- and interassay coefficients of variance were <15%.
Collection of covariate data
In Project Viva, mothers self-reported their race or ethnicity, smoking history, parity, age, and prepregnancy weight at the initial prenatal visit. We calculated prepregnancy BMI as weight (kg) divided by height (m2) using self-reported weight and height, which has been previously validated in this cohort (31). We obtained child sex from an interview conducted at the delivery visit. We calculated total GWG by subtracting self-reported prepregnancy weight from the last clinical weight measured within 4 weeks before delivery and categorized GWG as inadequate, adequate, or excessive, according to the 2009 Institute of Medicine guidelines (32). We determined gestational diabetes status using results of routine prenatal screening (nonfasting 50-g oral glucose load) done at 26 to 28 weeks of gestation, with a subsequent fasting 3-hour 100-g oral glucose tolerance test if the blood glucose level exceeded 140 mg/dL (23). All data collection instruments used in Project Viva are publicly available at https://www.hms.harvard.edu/viva/. In Gen3G, trained research assistants recorded data on maternal age, race or ethnicity, parity, smoking behaviors, prepregnancy weight, and visit season at the first study visit. We obtained child sex from delivery medical records. We calculated maternal prepregnancy BMI (kg/m2) from prepregnancy weight and height measurements taken by research staff at the first study visit. We calculated GWG by subtracting self-reported prepregnancy weight from weight recorded at the last routine visit before delivery and abstracted from medical records, and defined GWG using the Institute of Medicine categories of inadequate, adequate, and excessive. All women in Gen3G completed a fasting 75-g oral glucose tolerance test during the second study visit and were diagnosed with GDM according to International Association of Diabetes and Pregnancy Study Groups criteria (33).
Statistical analysis
We calculated correlations between maternal and cord blood 25(OH)D concentrations using Spearman correlation analysis. We examined associations of maternal and cord blood 25(OH)D and of cord blood insulin and c-peptide concentrations with categories of maternal and newborn characteristics using generalized linear models with global (type 3) P value testing to identify potential confounders. We assessed relationships of maternal and cord blood 25(OH)D with cord blood insulin and c-peptide using correlation analysis and generalized linear models with 25(OH)D categorized as follows: severely deficient (<25 nmol/L), deficient (25 to <50 nmol/L), insufficient (50 to <75 nmol/L), and sufficient (≥75 nmol/L) (34, 35). The relationship of maternal 25(OH)D with cord blood insulin and c-peptide appeared to be nonlinear and we therefore analyzed the associations using covariate-adjusted generalized additive models (GAMs) with nonlinear spline terms for maternal 25(OH)D. Model 1 was adjusted for season of delivery (Project Viva) or prenatal visit at which maternal 25(OH)D concentration was measured (Gen3G). Model 2 was adjusted for model 1 covariate plus child sex, maternal race or ethnicity, age at enrollment, parity, and smoking during pregnancy. Model 3 was adjusted for model 2 covariates plus maternal prepregnancy BMI. Model 4 was adjusted for model 3 covariates plus category of GWG.
The relationship of cord blood 25(OH)D with cord blood insulin and c-peptide met standard assumptions for linear regression and we analyzed associations using multivariable linear regression models. Model 1 was adjusted for season of delivery (Project Viva) or prenatal visit (Gen3G). Model 2 was adjusted for model 1 covariates plus child sex, maternal race or ethnicity, age at enrollment, parity and smoking during pregnancy. Model 3 was adjusted for model 2 covariates plus maternal prepregnancy BMI. Model 4 was adjusted for model 3 covariates plus category of GWG.
We decided not to adjust our models for diagnosis of GDM, given that development of GDM may be on the pathway between gestational vitamin D status and cord blood hormone concentrations. However, we did stratify our models by GDM status in a sensitivity analysis. Given interactions between other exposures and child sex when examining associations with cord blood hormones in our cohort (23, 24), we decided a priori to look for 25(OH)D by sex interactions and examined models separately for boys and girls in addition to looking at overall effects. We included interaction terms in the fully adjusted regression models to check for statistical interactions of 25(OH)D with sex and GDM status.
To improve the normality of residuals and linearity of the relationship between 25(OH)D status and the cord blood hormones, we natural log–transformed the values for cord blood insulin and cord blood c-peptide concentrations before analysis. We present geometric means or medians for these outcomes and regression results as a percent change in the outcome, calculated as: [% change = (exp(β) – 1) × 100], for a 10-nmol/L increase in 25(OH)D. We used hypothesis-driven models to look for trends and consistency of results across methods, and we interpreted all results in the context of our prespecified hypotheses. We performed all analyses using SAS, version 9.4 (SAS Institute, Cary, NC) and considered P < 0.05 statistically significant.
Results
Mean (SD) maternal 25(OH)D concentrations were 59 (20) and 64 (20) nmol/L and cord blood concentrations were 46 (19) and 53 (18) nmol/L in Project Viva and Gen3G, respectively (Table 1). Vitamin D insufficiency and deficiency were both highly prevalent: >80% of Project Viva mothers and >70% of Gen3G mothers had 25(OH)D levels <75 nmol/L during their pregnancy, and cord blood 25(OH)D levels were <75 nmol/L in 94% of Project Viva newborns and 88% of Gen3G newborns (Table 1). Correlations between maternal and cord blood 25(OH)D levels were 0.58 in the Project Viva cohort and 0.37 in the Gen3G cohort; the higher correlation in Project Viva is unsurprising given that maternal 25(OH)D levels were measured later in pregnancy (second trimester vs first trimester in the Gen3G cohort). The median ratio of cord blood to maternal 25(OH)D was 0.79 in the Project Viva cohort and 0.85 in the Gen3G cohort, with variation in the ratios by season. In both cohorts, higher maternal prepregnancy BMI was associated with lower maternal and cord blood 25(OH)D levels and higher cord blood insulin and c-peptide. In the Project Viva cohort, maternal and cord blood 25(OH)D levels were higher among mothers who were white, had a 4-year college degree and higher household income, were ≥30 years old at enrollment, and who did not smoke during pregnancy. Season of birth was a strong predictor of cord blood 25(OH)D, with higher levels among those born in summer and fall. In the Gen3G cohort, babies born to nulliparous mothers had higher cord blood 25(OH)D, and 25(OH)D levels were higher in mothers who did not have GDM. In both cohorts, cord blood c-peptide concentrations were higher in babies whose mothers were older and who had higher GWG and GDM (Table 1).
Table 1.
Maternal and Cord Blood 25(OH)D and Cord Blood C-Peptide Concentrations by Categories of Participant Characteristics for 862 Project Viva Mother-Child Pairs and 660 Gen3G Mother-Child Pairs
| Project Viva Cohort | Gen3G Cohort | |||||||
|---|---|---|---|---|---|---|---|---|
| Second-Trimester 25(OH)D (nmol/L) | Cord Blood 25(OH)D (nmol/L) | Cord Blood C-Peptide (ng/mL) | First-Trimester 25(OH)D (nmol/L) | Cord Blood 25(OH)D (nmol/L) | Cord Blood C-Peptide (ng/mL) | |||
| % | Mean or Geometric Meana | % | Mean or Geometric Meana | |||||
| Overall | 59 | 46 | 0.89 | 64 | 53 | 0.4 | ||
| Maternal race or ethnicity | ||||||||
| White | 70 | 62.0 | 50.9 | 0.89 | 98 | 63.8 | 53.2 | 0.42 |
| Black | 15 | 46.9 | 31.2 | 0.89 | N/A | |||
| Other | 15 | 50.4 | 37.2 | 0.91 | 2 | 55.2 | 44.4 | 0.41 |
| P | <0.0001 | <0.0001 | 0.93 | 0.09 | 0.07 | 0.93 | ||
| Maternal age at enrollment | ||||||||
| <30 y | 32 | 53.7 | 39.7 | 0.83 | 69 | 63.9 | 52.6 | 0.40 |
| 30–35 y | 41 | 59.8 | 48.4 | 0.92 | 26 | 63.0 | 52.8 | 0.44 |
| >35 y | 27 | 62.5 | 49.4 | 0.92 | 5 | 62.2 | 57.8 | 0.47 |
| P | <0.0001 | <0.0001 | 0.05 | 0.80 | 0.29 | 0.12 | ||
| Maternal parity | ||||||||
| 0 | 45 | 58.3 | 45.7 | 0.82 | 51 | 63.6 | 54.8 | 0.41 |
| ≥1 | 55 | 59.2 | 46.0 | 0.96 | 49 | 63.4 | 50.8 | 0.43 |
| P | 0.58 | 0.81 | <0.0001 | 0.93 | <0.01 | 0.27 | ||
| Maternal smoking during pregnancy | ||||||||
| No | 87 | 59.9 | 46.5 | 0.89 | 91 | 63.8 | 53.3 | 0.41 |
| Yes | 13 | 51.0 | 41.9 | 0.93 | 9 | 62.1 | 49.1 | 0.45 |
| P | <0.001 | 0.02 | 0.38 | 0.52 | 0.09 | 0.23 | ||
| Maternal prepregnancy BMI, kg/m2 | ||||||||
| <18.5 | 4 | 63.0 | 48.4 | 0.76 | 4 | 60.5 | 51.3 | 0.36 |
| 18.5–24.9 | 59 | 60.8 | 49.3 | 0.83 | 57 | 66.0 | 54.5 | 0.40 |
| 25.0–29.9 | 22 | 57.0 | 44.0 | 0.98 | 20 | 62.6 | 52.5 | 0.39 |
| ≥30 | 15 | 52.3 | 36.0 | 1.05 | 19 | 57.6 | 49.3 | 0.52 |
| P | <0.01 | <0.0001 | <0.0001 | <0.001 | 0.05 | <0.0001 | ||
| GDM during pregnancy | ||||||||
| No | 94 | 58.8 | 45.9 | 0.87 | 91 | 64.2 | 53.1 | 0.41 |
| Yes | 6 | 56.9 | 46.6 | 1.16 | 9 | 57.9 | 50.6 | 0.51 |
| P | 0.58 | 0.80 | <0.001 | 0.02 | 0.32 | <0.01 | ||
| Gestational weight gain | ||||||||
| Inadequate | 12 | 65.0 | 47.2 | 0.82 | 19 | 64.4 | 54.6 | 0.37 |
| Adequate | 30 | 57.1 | 46.1 | 0.85 | 35 | 64.4 | 54.4 | 0.43 |
| Excessive | 58 | 58.2 | 45.8 | 0.93 | 46 | 62.6 | 51.3 | 0.43 |
| P | <0.01 | 0.80 | 0.03 | 0.52 | 0.09 | 0.05 | ||
| Season of prenatal visit | ||||||||
| Winter | 25 | 59.3 | 45.7 | 0.90 | 23 | 55.6 | 62.6 | 0.44 |
| Spring | 27 | 55.8 | 51.4 | 0.93 | 29 | 61.6 | 50.6 | 0.44 |
| Summer | 22 | 61.8 | 49.9 | 0.87 | 24 | 71.3 | 45.7 | 0.41 |
| Fall | 26 | 58.7 | 41.8 | 0.88 | 24 | 67.4 | 53.5 | 0.38 |
| P | 0.06 | <0.0001 | 0.72 | <0.0001 | <0.0001 | 0.08 | ||
| Sex of child | ||||||||
| Male | 53 | 58.4 | 44.8 | 0.88 | 52 | 63.1 | 51.2 | 0.40 |
| Female | 47 | 59.2 | 47.1 | 0.90 | 48 | 64.1 | 54.1 | 0.44 |
| P | 0.58 | 0.07 | 0.54 | 0.51 | 0.10 | 0.03 | ||
| Category of maternal 25(OH)D, nmol/L | ||||||||
| <25 (severely deficient) | 3 | 19.7 | 23.5 | 0.76 | 1 | 21.3 | 36.8 | 0.30 |
| 25 to <50 (deficient) | 30 | 39.4 | 37.0 | 0.92 | 25 | 40.9 | 46.2 | 0.44 |
| 50 to <75 (insufficient) | 49 | 62.2 | 50.1 | 0.90 | 47 | 62.3 | 52.2 | 0.40 |
| ≥75 (sufficient) | 18 | 88.8 | 63.8 | 0.85 | 27 | 88.3 | 62.6 | 0.41 |
| P | <0.0001 | <0.0001 | 0.39 | <0.0001 | <0.0001 | 0.18 | ||
| Category of cord blood 25(OH)D, nmol/L | ||||||||
| <25 (severely deficient) | 13 | 36.4 | 17.3 | 0.86 | 4 | 51.9 | 19.4 | 0.36 |
| 25 to <50 (deficient) | 48 | 54.2 | 38.8 | 0.88 | 42 | 59.3 | 39.6 | 0.41 |
| 50 to <7 (insufficient) | 33 | 67.1 | 60.1 | 0.90 | 42 | 64.9 | 60.5 | 0.42 |
| ≥75 (sufficient) | 6 | 77.9 | 87.4 | 1.04 | 12 | 79.4 | 86.2 | 0.47 |
| P | <0.0001 | <0.0001 | 0.23 | <0.0001 | <0.0001 | 0.15 | ||
P values in bold type are statistically significant at P < 0.05.
Abbreviation: N/A, not applicable.
Cord blood c-peptide values were log-transformed before analysis and their geometric means are reported here.
The relationship of maternal 25(OH)D with both cord blood insulin and c-peptide appeared to have an inverted-U–shape in both cohorts. We examined plots resulting from GAM with nonlinear spline terms for maternal 25(OH)D and adjusted for season of delivery (Project Viva) or prenatal visit (Gen3G), child sex, maternal race or ethnicity, age at enrollment, parity, smoking during pregnancy, maternal prepregnancy BMI, and GWG. In Project Viva, cord blood hormone concentrations were slightly higher at maternal 25(OH)D levels in the range of 50 to 80 nmol/L (Fig. 1), but associations were not statistically significant and confidence bands were wide at the extremes as a result of few participants having very low or very high 25(OH)D levels. In the Gen3G cohort, there was a similar pattern, but the range of 25(OH)D levels associated with higher cord blood hormones was slightly higher, at approximately 60 to 90 nmol/L (P = 0.03 for both; Fig. 2).
Figure 1.
Spline plots for associations of maternal 25(OH)D with (a) cord blood insulin and (b) c-peptide concentrations among 626 mother-child pairs in Project Viva. Plots generated from covariate-adjusted GAMs with nonlinear spline terms for maternal 25(OH)D and linear terms for season of delivery, child sex, and maternal race or ethnicity, age at enrollment, parity, smoking during pregnancy, prepregnancy BMI, and GWG. DF, degrees of freedom.
Figure 2.
Spline plots for associations of maternal 25(OH)D with (a) cord blood insulin and (b) c-peptide concentrations among 602 mother-child pairs in the Gen3G cohort. Plots generated from covariate-adjusted GAMs with nonlinear spline terms for maternal 25(OH)D and linear terms for season of prenatal visit, child sex, and maternal race or ethnicity, age at enrollment, parity, smoking during pregnancy, prepregnancy BMI, and GWG. DF, degrees of freedom.
In both cohorts, the relationship of cord blood 25(OH)D with cord blood insulin and c-peptide appeared to be linear in bivariate analyses, confirmed by comparison of fully adjusted GAM with linear vs nonlinear terms for cord blood 25(OH)D. Therefore, we analyzed these associations using multivariable linear regression models. In both cohorts, models adjusted for season only (model 1) and additionally for child sex, maternal race or ethnicity, age at enrollment, parity, and smoking during pregnancy (model 2) did not show statistically significant associations of cord blood 25(OH)D with cord blood insulin or c-peptide concentrations, except for the models predicting cord blood c-peptide in the Gen3G cohort. However, after additionally adjusting for maternal prepregnancy BMI (model 3), each 10-nmol/L increase in cord blood 25(OH)D was associated with a 3.6% (95% CI, 0.0 to 7.3) higher cord blood insulin concentration and a 3.2% (95% CI, 0.8 to 5.6) higher c-peptide concentration in Project Viva, and with 2.2% (95% CI, −0.2 to 4.5) higher cord blood insulin concentration and 3.6% (95% CI, 1.0 to 6.3) higher c-peptide concentration in the Gen3G cohort. Additional adjustment for GWG (model 4) did not change these results (Table 2). When we stratified the models by child sex, cord blood 25(OH)D was associated with cord blood insulin and c-peptide in Project Viva and with cord blood c-peptide in Gen3G among male infants only; there were no statistically significant associations among girls (Table 3). However, the interaction between cord blood 25(OH)D and sex was not statistically significant in either cohort; therefore, we focus our primary results on both sexes combined.
Table 2.
Associations of Cord Blood 25(OH)D With Cord Blood Insulin and C-Peptide Among 842 Mother-Child Pairs in Project Viva and 650 Mother-Child Pairs in Gen3G
| Median (IQR) | % Difference (95% CI) per 10-nmol/L Increase in Cord Blood 25(OH)Da | ||||
|---|---|---|---|---|---|
| Model 1b | Model 2c | Model 3d | Model 4e | ||
| Project Viva cohort | |||||
| Cord blood insulin, μU/mL | 4.7 (2.6–8.3) | −0.2 (−3.3 to 3.1) | 1.8 (−1.7 to 5.4) | 3.6 (0.0 to 7.3) | 3.7 (0.09 to 7.5) |
| Cord blood c-peptide, ng/mL | 0.9 (0.6–1.3) | 1.7 (−0.4 to 3.9) | 1.7 (−0.6 to 4.0) | 3.2 (0.8 to 5.6) | 3.2 (0.8 to 5.6) |
| Gen3G cohort | |||||
| Cord blood insulin, μU/mL | 9.0 (6.7–12.3) | 1.9 (−0.5 to 4.3) | 1.5 (−0.8 to 3.9) | 2.2 (−0.2 to 4.5) | 2.2 (−0.1 to 4.6) |
| Cord blood c-peptide, ng/mL | 0.4 (0.3–0.6) | 2.7 (0.1 to 5.3) | 2.9 (0.3 to 5.6) | 3.6 (1.0 to 6.3) | 3.6 (1.0 to 6.3) |
Boldface type indicates statistical significance at an alpha level of 0.05.
Abbreviation: IQR, interquartile range.
From multivariable linear regression models. Cord blood insulin and c-peptide values were log-transformed before analysis and results are presented as % change in the outcome, calculated as: [% change = (exp(β) – 1) × 100].
Adjusted for season of delivery (Project Viva) or first-trimester blood draw (Gen3G).
Adjusted for model 1 covariate plus child sex, maternal race/ethnicity, age at enrollment, parity, and smoking during pregnancy.
Adjusted for model 2 covariates plus maternal prepregnancy BMI.
Adjusted for model 3 covariates plus GWG.
Table 3.
Associations of Cord Blood 25(OH)D With Cord Blood Insulin and C-Peptide Among Male and Female Newborns in Project Viva and Gen3G
| % Difference (95% CI) per 10-nmol/L Increase in Cord Blood 25(OH)Da | ||
|---|---|---|
| Cord Blood Insulin (μU/mL) | Cord Blood C-Peptide (ng/mL) | |
| Project Viva | ||
| Male (n = 429) | 6.6 (1.3 to 12.1) | 5.3 (1.9 to 8.9) |
| Female (n = 389) | 0.8 (−4.2 to 6.0) | 0.8 (−2.5 to 4.3) |
| P for cord blood 25(OH)D × sex interaction | 0.24 | 0.23 |
| Gen3G | ||
| Male (n = 324) | 2.4 (−1.0 to 5.9) | 4.1 (0.4 to 8.0) |
| Female (n = 303) | 2.3 (−1.0 to 5.7) | 3.5 (−0.2 to 7.4) |
| P for cord blood 25(OH)D × sex interaction | 0.43 | 0.60 |
Boldface type indicates statistical significance at an alpha level of 0.05.
From multivariable linear regression models adjusted for season of delivery (Project Viva) or first-trimester blood collection (Gen3G) and maternal race or ethnicity, age at enrollment, parity, smoking during pregnancy, prepregnancy BMI, and GWG. Cord blood insulin and c-peptide values were log-transformed before analysis and results are presented as % change in the outcome, calculated as: [% change = (exp(β) – 1) × 100].
We also performed a sensitivity analysis to evaluate whether a diagnosis of GDM would affect the associations of cord blood 25(OH)D with cord blood insulin and c-peptide. In both cohorts, the effect estimates were similar in women with and without GDM, but CIs were wider for the GDM groups, given the small sample sizes.
Discussion
In this study of mother-child pairs from two separate cohorts residing in high-latitude regions with high prevalence of vitamin D insufficiency, we found consistent positive associations of cord blood 25(OH)D with cord blood insulin and c-peptide concentrations. Maternal 25(OH)D status appeared to have nonlinear relationships with cord blood insulin and c-peptide, with slightly higher hormone concentrations related to moderate maternal 25(OH)D levels in the range of 50 to 90 nmol/L. Maternal prepregnancy BMI, which is a major predictor of cord blood insulin and c-peptide concentrations (36) and is also related to maternal and cord blood 25(OH)D concentrations (14, 37), was a strong confounder of these associations. One potential explanation for our findings is that mothers with higher BMI have lower vitamin D status but also transfer excess glucose and lipids to the fetus, which obscure any impact of vitamin D status on fetal insulin secretion. Consequently, after controlling for maternal BMI, we observed an association with vitamin D. Because 25(OH)D as well as insulin and c-peptide concentrations in cord blood have been associated with increased fetal growth, these biomarkers may all be indicative of a more robust fetal nutrient supply. Another marker of fetal under- or overnutrition is maternal GWG. Adjusting our models for GWG did not materially change the observed associations. Finally, it is also possible that more efficient placental nutrient transport results in greater delivery of both glucose (leading to higher cord blood insulin and c-peptide concentrations) and 25(OH)D to the fetus. However, we accounted for two factors that may affect placental nutrient transport function: maternal BMI (14) (via adjustment in our multivariable regression models) and GDM status (38) (using stratified models). The observed associations were strengthened after adjusting for maternal BMI and did not differ between women with and without GDM.
Many studies have examined maternal 25(OH)D status in relation to glucose control during pregnancy and birth outcomes. Interestingly, we observed that although maternal and cord blood 25(OH)D levels were moderately correlated, 25(OH)D measured in cord blood had a consistent positive association with cord blood insulin and c-peptide concentrations, but maternal 25(OH)D did not. Fetal 25(OH)D levels are entirely dependent on the mother’s 25(OH)D levels as well as the efficiency of placental 25(OH)D transfer, which may be influenced by maternal BMI due to placental abnormalities and/or reduced 25(OH)D bioavailability associated with obesity (14). Thus, cord blood 25(OH)D concentration is likely a more accurate measure of fetal vitamin D status. Our results suggest that fetal vitamin D status is directly related to fetal insulin secretion, whereas the relationship with maternal vitamin D status is more complex. It is also possible that the apparent inversion at the high end of maternal 25(OH)D levels is a result of unstable estimates owing to few mothers having high vitamin D status. Across the range of more prevalent 25(OH)D levels, the relationship appears to parallel the positive association observed between cord blood 25(OH)D and the cord blood hormones.
A major strength of our study was the ability to use comparable data from two cohorts to replicate our findings: We first conducted all analyses within Project Viva and then repeated our analyses using data from Gen3G. The results were very similar in the two cohorts, increasing our confidence that there is a true biological relationship between cord blood and possibly maternal 25(OH)D levels and cord blood insulin and c-peptide. In both cohorts, we had data available from a large sample of mother-child pairs and information on many potential covariates, and we were able to adjust our models for various maternal and child characteristics, including maternal BMI. Another strength was the availability of 25(OH)D status at two time points during the pre- and perinatal periods: that of the mother during pregnancy and the newborn at birth. This allowed us to demonstrate differences in the relationships of maternal and newborn 25(OH)D status with cord blood insulin and c-peptide.
Our study also had some limitations. Both 25(OH)D and insulin and c-peptide were measured concurrently in the same sample of umbilical cord blood. However, we are confident there is little potential for reverse causality in the observed association, because cord blood 25(OH)D levels are entirely dependent on the mother’s 25(OH)D status and other maternal characteristics that may influence placental transfer of 25(OH)D, such as maternal BMI. In contrast, the insulin and c-peptide measured in cord blood are produced by the fetus and would not affect the mother’s 25(OH)D status. We also observed similar associations of the cord blood hormones with maternal 25(OH)D measured in pregnancy, before measurement of the cord blood hormones. Cord blood insulin and c-peptide concentrations are both proxy measures of fetal β-cell function and insulin secretion, and insulin sensitivity and the efficiency of insulin clearance will also influence the levels in circulation. However, there is evidence that vitamin D affects insulin secretion directly: Pancreatic β cells express high levels of the vitamin D receptor (39) and have been shown to respond to 1α,25-dihydroxyvitamin D3, the active form of vitamin D (40). Vitamin D deficiency impairs insulin secretion via a direct impact on β cells in humans and in animal models (40). Generalizability of our results may be somewhat limited because the mothers in both cohorts are predominantly white and nonsmokers, and mothers in Project Viva generally had a high socioeconomic status and all had health insurance at the time of recruitment. All mothers in Gen3G have access to same health care system as the general Canadian population (i.e., universal governmentally supported).
Although we observed a linear association of cord blood 25(OH)D with cord blood insulin and c-peptide in both cohorts, there were few newborns in either cohort with sufficient 25(OH)D levels. Thus, we were unable to examine the relationship of increasing vitamin D status with the cord blood hormones in infants with sufficient vitamin D. It is possible that the direction of the relationship between cord blood 25(OH)D and cord blood insulin and c-peptide could change once the fetus reaches some threshold level of vitamin D, and this question deserves further study in a population with higher cord blood 25(OH)D levels. Research should also examine long-term outcomes to determine whether vitamin D status during gestation or measured in cord blood at birth results in long-term programming of glycemic control and related health outcomes.
In conclusion, among mother-child pairs from two cohorts with high prevalence of vitamin D insufficiency, newborns with higher cord blood 25(OH)D levels had higher cord blood insulin and c-peptide concentrations, but maternal 25(OH) status had a nonlinear relationship with cord blood insulin and c-peptide. Our results suggest that vitamin D may play a role in regulating fetal β-cell function, with a potential long-term impact on postnatal glucose regulation and growth.
Acknowledgments
Financial Support: Project Viva was supported by the US National Institutes of Health (Grants R01 HD034568, UH3 OD023286 to E.O.). Gen3G pregnancy and birth follow-up were supported by a Fonds de recherche du Québec en santé operating grant (Grant 20697 to M.-F.H.), a Canadian Institute of Health Research operating grant (Grant MOP 115071 to M.-F.H.), and by Diabète Québec (P.P.).
The study sponsors were not involved in the design of the study; the collection, analysis, and interpretation of data; writing the report; or the decision to submit the report for publication.
Clinical Trial Information: ClinicalTrials.gov nos. NCT02820402 (registered 1 July 2016) and NCT01623934 (registered 20 June 2012).
Disclosure Summary: The authors have nothing to disclose.
Glossary
Abbreviations:
- 25(OH)D
25-hydroxyvitamin D
- GAM
generalized additive model
- GDM
gestational diabetes mellitus
- Gen3G
Genetics of Glucose Regulation in Gestation and Growth Study
- GWG
gestational weight gain
References and Notes
- 1. Crozier SR, Harvey NC, Inskip HM, Godfrey KM, Cooper C, Robinson SM; SWS Study Group. Maternal vitamin D status in pregnancy is associated with adiposity in the offspring: findings from the Southampton Women’s Survey. Am J Clin Nutr. 2012;96(1):57–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Eckhardt CL, Gernand AD, Roth DE, Bodnar LM. Maternal vitamin D status and infant anthropometry in a US multi-centre cohort study. Ann Hum Biol. 2015;42(3):215–222. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Leffelaar ER, Vrijkotte TGM, van Eijsden M. Maternal early pregnancy vitamin D status in relation to fetal and neonatal growth: results of the multi-ethnic Amsterdam Born Children and their Development cohort. Br J Nutr. 2010;104(1):108–117. [DOI] [PubMed] [Google Scholar]
- 4. Weinert LS, Silveiro SP. Maternal-fetal impact of vitamin D deficiency: a critical review. Matern Child Health J. 2015;19(1):94–101. [DOI] [PubMed] [Google Scholar]
- 5. Bodnar LM, Simhan HN, Powers RW, Frank MP, Cooperstein E, Roberts JM. High prevalence of vitamin D insufficiency in black and white pregnant women residing in the northern United States and their neonates. J Nutr. 2007;137(2):447–452. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Chiu KC, Chu A, Go VLW, Saad MF. Hypovitaminosis D is associated with insulin resistance and beta cell dysfunction. Am J Clin Nutr. 2004;79(5):820–825. [DOI] [PubMed] [Google Scholar]
- 7. Lacroix M, Battista M-C, Doyon M, Houde G, Ménard J, Ardilouze J-L, Hivert M-F, Perron P. Lower vitamin D levels at first trimester are associated with higher risk of developing gestational diabetes mellitus. Acta Diabetol. 2014;51(4):609–616. [DOI] [PubMed] [Google Scholar]
- 8. Krishnaveni GV, Veena SR, Winder NR, Hill JC, Noonan K, Boucher BJ, Karat SC, Fall CHD. Maternal vitamin D status during pregnancy and body composition and cardiovascular risk markers in Indian children: the Mysore Parthenon Study. Am J Clin Nutr. 2011;93(3):628–635. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Farrant HJW, Krishnaveni GV, Hill JC, Boucher BJ, Fisher DJ, Noonan K, Osmond C, Veena SR, Fall CHD. Vitamin D insufficiency is common in Indian mothers but is not associated with gestational diabetes or variation in newborn size. Eur J Clin Nutr. 2009;63(5):646–652. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Godang K, Frøslie KF, Henriksen T, Qvigstad E, Bollerslev J. Seasonal variation in maternal and umbilical cord 25(OH) vitamin D and their associations with neonatal adiposity. Eur J Endocrinol. 2014;170(4):609–617. [DOI] [PubMed] [Google Scholar]
- 11. Casey C, McGinty A, Holmes VA, Patterson CC, Young IS, McCance DR. Maternal vitamin D and neonatal anthropometrics and markers of neonatal glycaemia: Belfast Hyperglycemia and Adverse Pregnancy Outcome (HAPO) study. Br J Nutr. 2018;120(1):74–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Roth DE, Leung M, Mesfin E, Qamar H, Watterworth J, Papp E. Vitamin D supplementation during pregnancy: state of the evidence from a systematic review of randomised trials. BMJ. 2017;359:j5237. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Kiely M, Hemmingway A, O’Callaghan KM. Vitamin D in pregnancy: current perspectives and future directions. Ther Adv Musculoskelet Dis. 2017;9(6):145–154. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Josefson JL, Feinglass J, Rademaker AW, Metzger BE, Zeiss DM, Price HE, Langman CB. Maternal obesity and vitamin D sufficiency are associated with cord blood vitamin D insufficiency. J Clin Endocrinol Metab. 2013;98(1):114–119. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Carlsen EM, Renault KM, Jensen RB, Nørgaard K, Jensen JE, Nilas L, Cortes D, Michaelsen KF, Pryds O. The association between newborn regional body composition and cord blood concentrations of c-peptide and insulin-like growth factor I. PLoS One. 2015;10(7):e0121350. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Meyer DM, Brei C, Stecher L, Brunner S, Hauner H. Maternal insulin resistance, triglycerides and cord blood insulin are not determinants of offspring growth and adiposity up to 5 years: a follow-up study. Diabet Med. 2018;35(10):1399–1403. [DOI] [PubMed] [Google Scholar]
- 17. Brunner S, Schmid D, Hüttinger K, Much D, Heimberg E, Sedlmeier E-M, Brüderl M, Kratzsch J, Bader BL, Amann-Gassner U, Hauner H. Maternal insulin resistance, triglycerides and cord blood insulin in relation to post-natal weight trajectories and body composition in the offspring up to 2 years. Diabet Med. 2013;30(12):1500–1507. [DOI] [PubMed] [Google Scholar]
- 18. Regnault N, Botton J, Heude B, Forhan A, Hankard R, Foliguet B, Hillier TA, Souberbielle JC, Dargent-Molina P, Charles MA; EDEN Mother-Child Cohort Study Group. Higher cord C-peptide concentrations are associated with slower growth rate in the 1st year of life in girls but not in boys. Diabetes. 2011;60(8):2152–2159. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Yan J. Maternal pre-pregnancy BMI, gestational weight gain, and infant birth weight: a within-family analysis in the United States. Econ Hum Biol. 2015;18:1–12. [DOI] [PubMed] [Google Scholar]
- 20. Sauder KA, Hockett CW, Ringham BM, Glueck DH, Dabelea D. Fetal overnutrition and offspring insulin resistance and β-cell function: the Exploring Perinatal Outcomes among Children (EPOCH) study. Diabet Med. 2017;34(10):1392–1399. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Parellada CB, Asbjörnsdóttir B, Ringholm L, Damm P, Mathiesen ER. Fetal growth in relation to gestational weight gain in women with type 2 diabetes: an observational study. Diabet Med. 2014;31(12):1681–1689. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Scholl TO, Sowers M, Chen X, Lenders C. Maternal glucose concentration influences fetal growth, gestation, and pregnancy complications. Am J Epidemiol. 2001;154(6):514–520. [DOI] [PubMed] [Google Scholar]
- 23. Oken E, Morton-Eggleston E, Rifas-Shiman SL, Switkowski KM, Hivert M-F, Fleisch AF, Mantzoros C, Gillman MW. Sex-specific associations of maternal gestational glycemia with hormones in umbilical cord blood at delivery. Am J Perinatol. 2016;33(13):1273–1281. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Switkowski KM, Jacques PF, Must A, Hivert M-F, Fleisch A, Gillman MW, Rifas-Shiman S, Oken E. Higher maternal protein intake during pregnancy is associated with lower cord blood concentrations of insulin-like growth factor (IGF)-II, IGF binding protein 3, and insulin, but not IGF-I, in a cohort of women with high protein intake. J Nutr. 2017;147(7):1392–1400. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Oken E, Baccarelli AA, Gold DR, Kleinman KP, Litonjua AA, De Meo D, Rich-Edwards JW, Rifas-Shiman SL, Sagiv S, Taveras EM, Weiss ST, Belfort MB, Burris HH, Camargo CA Jr, Huh SY, Mantzoros C, Parker MG, Gillman MW. Cohort profile: Project Viva. Int J Epidemiol. 2015;44(1):37–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Guillemette L, Allard C, Lacroix M, Patenaude J, Battista M-C, Doyon M, Moreau J, Ménard J, Bouchard L, Ardilouze J-L, Perron P, Hivert M-F. Genetics of glucose regulation in Gestation and Growth (Gen3G): a prospective prebirth cohort of mother-child pairs in Sherbrooke, Canada. BMJ Open. 2016;6(2):e010031. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Parker M, Rifas-Shiman SL, Belfort MB, Taveras EM, Oken E, Mantzoros C, Gillman MW. Gestational glucose tolerance and cord blood leptin levels predict slower weight gain in early infancy. J Pediatr. 2011;158(2):227–233. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Burris HH, Rifas-Shiman SL, Huh SY, Kleinman K, Litonjua AA, Oken E, Rich-Edwards JW, Camargo CA Jr, Gillman MW. Vitamin D status and hypertensive disorders in pregnancy. Ann Epidemiol. 2014;24(5):399–403.e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Burris HH, Rifas-Shiman SL, Camargo CA Jr, Litonjua AA, Huh SY, Rich-Edwards JW, Gillman MW. Plasma 25-hydroxyvitamin D during pregnancy and small-for-gestational age in black and white infants. Ann Epidemiol. 2012;22(8):581–586. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Bunyavanich S, Rifas-Shiman SL, Platts-Mills TA, Workman L, Sordillo JE, Camargo CA Jr, Gillman MW, Gold DR, Litonjua AA. Prenatal, perinatal, and childhood vitamin D exposure and their association with childhood allergic rhinitis and allergic sensitization. J Allergy Clin Immunol. 2016;137(4):1063–1070.e2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Provenzano AM, Rifas-Shiman SL, Herring SJ, Rich-Edwards JW, Oken E. Associations of maternal material hardships during childhood and adulthood with prepregnancy weight, gestational weight gain, and postpartum weight retention. J. Womens Health (Larchmt) 2015;24(7):563–571. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Rasmussen KM, Yaktine AL, Institute of Medicine, National Research Council, Committee to Reexamine IOM Pregnancy Weight Guidelines, eds. Weight Gain During Pregnancy: Reexamining the Guidelines. Washington, DC: National Academies Press; 2009. Available at: www.ncbi.nlm.nih.gov/books/NBK32813/. Accessed 18 April 2017 [Google Scholar]
- 33. Metzger BE, Gabbe SG, Persson B, Buchanan TA, Catalano PA, Damm P, Dyer AR, Leiva A, Hod M, Kitzmiler JL, Lowe LP, McIntyre HD, Oats JJ, Omori Y, Schmidt MI; International Association of Diabetes and Pregnancy Study Groups Consensus Panel. International Association of Diabetes and Pregnancy Study Groups recommendations on the diagnosis and classification of hyperglycemia in pregnancy. Diabetes Care. 2010;33(3):676–682. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Bischoff-Ferrari HA, Giovannucci E, Willett WC, Dietrich T, Dawson-Hughes B. Estimation of optimal serum concentrations of 25-hydroxyvitamin D for multiple health outcomes [published corrections appear in Am J Clin Nutr. 2006;84(5):1253 and Am J Clin Nutr. 2007;86(3):809]. Am J Clin Nutr. 2006;84(1):18–28. [DOI] [PubMed] [Google Scholar]
- 35. Ginde AA, Sullivan AF, Mansbach JM, Camargo CA Jr. Vitamin D insufficiency in pregnant and nonpregnant women of childbearing age in the United States. Am J Obstet Gynecol. 2010;202(5):436.e1–436.e8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Valsamakis G, Papatheodorou DC, Margeli A, Bakoulas V, Kapantais E, Papassotiriou I, Creatsas G, Kumar S, Mastorakos G. First trimester maternal BMI is a positive predictor of cord blood c-peptide levels while maternal visfatin levels is a negative predictor of birth weight. Hormones (Athens). 2014;13(1):87–94. [DOI] [PubMed] [Google Scholar]
- 37. Bodnar LM, Catov JM, Roberts JM, Simhan HN. Prepregnancy obesity predicts poor vitamin D status in mothers and their neonates. J Nutr. 2007;137(11):2437–2442. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Castillo-Castrejon M, Powell TL. Placental nutrient transport in gestational diabetic pregnancies [published correction appears in Front Endocrinol (Lausanne). 2019;10:5]. Front Endocrinol (Lausanne). 2017;8:306. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Neelankal John A, Jiang F-X. An overview of type 2 diabetes and importance of vitamin D3-vitamin D receptor interaction in pancreatic β-cells. J Diabetes Complications. 2018;32(4):429–443. [DOI] [PubMed] [Google Scholar]
- 40. Mathieu C, Gysemans C, Giulietti A, Bouillon R. Vitamin D and diabetes. Diabetologia. 2005;48(7):1247–1257. [DOI] [PubMed] [Google Scholar]
