Abstract
Introduction
Pregnancy is a critical period for fetal brain development and is sensitive to maternal micronutrient status. Maternal vitamin D [25(OH)D] concentrations have been associated with child neurodevelopmental outcomes, while biological aging processes reflected by epigenetic age acceleration (EAA) may contribute to these associations. This study examined associations between maternal vitamin D concentrations and neurodevelopmental outcomes in preschool-aged children and explored whether infant EAA mediated these associations.
Methods
We included 252 mother-child pairs from the APrON cohort. Maternal second trimester plasma 25(OH)D3 concentrations were quantified, and children's cognitive, language, motor, and behavioural outcomes were assessed at 3–5-years. In a subsample of 130 children, buccal epithelial cells collected at 3 months were profiled using the Infinium HumanMethylation EPIC v1.0 BeadChip. EAA was estimated using the Pediatric-Buccal-Epigenetic clock. Adjusted linear regressions examined associations between maternal 25(OH)D3 and children's outcomes, including models using quartiles and a 30 ng/mL threshold. Mann–Whitney U-tests compared outcomes according to a 20 ng/mL sufficiency threshold. In the subsample, exploratory mediation analyses examined the potential mediating role of infant EAA.
Results
No associations were found between maternal vitamin D and children's cognitive, language, or behavioural outcomes. Maternal vitamin D concentrations were positively associated with children's motor outcomes (B = 0.04; 95% CI = 0.004, 0.076, p = 0.03), although this small, nominal association did not survive correction for multiple comparisons (q = 0.18). This positive association was attenuated when vitamin D was modeled using quartiles or when “sufficiency” was defined as 30 ng/mL. No differences were found in neurodevelopmental outcomes based on vitamin D sufficiency defined as 20 ng/mL. No evidence of mediation via EAA was detected, although small indirect effects could have gone undetected due to the small sample size for this sub-analysis.
Conclusions
This study provides preliminary evidence that maternal vitamin D concentrations during pregnancy are associated with improved motor performance in children. A biomarker of biological aging, EAA, did not appear to play a role in this association. Examination of different forms of maternal vitamin D and statistical models that consider continuous measures and thresholds of “sufficiency” are needed to clarify associations between maternal vitamin D concentrations and children's neurodevelopment.
Keywords: children, infant epigenetic age acceleration, maternal nutrition, neurodevelopmental outcomes, pregnancy, vitamin D
1. Introduction
In utero brain development is highly sensitive to the maternal nutritional environment, with maternal nutrition playing a critical role in healthy fetal brain development and influencing long-term neurodevelopmental outcomes (1). While vitamin D has been recognized for its essential role in fetal skeletal development and bone health, emerging evidence suggests that fetal vitamin D is important for the normal development of the fetal brain through the regulation of neural stem cell proliferation, neuronal differentiation, and development of the dopaminergic system (2, 3).
Maternal vitamin D is obtained primarily through endogenous skin synthesis, while dietary intake contributes only a minor portion. In contrast, the fetus cannot synthesize vitamin D and is entirely dependent on placental transfer of maternal vitamin D (4, 5). Given the essential role of fetal vitamin D in normal brain development, maternal vitamin D deficiency may adversely affect fetal brain development (6).
Inconsistent findings are reported in research examining associations between sufficient maternal vitamin D during pregnancy and children's neurodevelopmental outcomes. Plasma maternal vitamin D concentrations have been associated with increased Full Scale (FSIQ), Visual (VIQ), and Nonverbal IQ (NIQ) scores in children 4- to 6-years of age (7), and fewer behaviour problems in children aged 2- to 4-years (8). Additionally, lower serum maternal vitamin D concentrations have been associated with an increase in clinically significant language difficulties in children 5- and 10-years of age (9), and with suboptimal fine- and gross-motor skill scores in children aged 2.5 years (10). In contrast, Darling et al. (10) reported no associations between serum maternal vitamin D concentrations and intelligence in 8-year-old children, and Gould et al. (11) found no associations between vitamin D and language outcomes or behaviour problem scores in 7-year-old children. Additionally, no association was found between serum maternal 25(OH)D concentrations and motor function in children 4-years of age (12).
These inconsistencies in the research literature may be attributable to numerous factors that influence maternal vitamin D concentrations, including vitamin supplementation, skin pigmentation, geographic location, clothing practices, sunscreen use, and the season during which samples are collected (4). Additionally, differences in the operational definitions of “sufficient” and “insufficient” vitamin D concentrations across epidemiological studies may account for some of the inconsistencies in the reported findings. In Canada, sufficient serum vitamin D concentrations are defined as ≥20 ng/mL in accordance with the Institute of Medicine's (IOM) guidelines, which are based on evidence related to bone health in the general U.S. and Canadian populations (13). In contrast, the Endocrine Society's previous clinical practice guidelines defined vitamin D sufficiency as serum concentrations as ≥30 ng/mL (14). However, these recommendations have since been revised and no longer endorse a single target concentration, reflecting the conclusion that current evidence is insufficient to support one single optimal serum target concentration for healthy adults (15).
The biological mechanisms through which maternal vitamin D may influence offspring neurodevelopment remain poorly understood. Emerging evidence suggests that vitamin D can induce epigenetic changes and that maternal nutritional status can alter the offspring epigenome (16, 17). Furthermore, fetal vitamin D deficiency has been implicated in aberrant epigenetic programming during early brain development, potentially contributing to adverse long-term neurobehavioural outcomes (18, 19).
One promising epigenetic biomarker is epigenetic age, which reflects age-related changes in the epigenome based on DNA methylation patterns at specific CpG sites (20, 21). Epigenetic age acceleration (EAA) is defined as the residual obtained by regressing DNA methylation-derived age (i.e., biological age) on chronological age (20). Estimated using epigenetic clocks, EAA provides an indicator of whether a tissue is aging faster or slower biologically than expected for its chronological age (20, 21).
Multi-modal studies integrating nutritional, epigenetic and neurodevelopmental data are needed to determine whether biomarkers of biological aging, such as EAA, mediate the association between prenatal nutrient exposure and child neurodevelopmental outcomes. Such investigations could provide insight into whether early-life nutrient perturbations influence neurodevelopment through biological aging-related pathways. However, to our knowledge, no previous study has examined EAA as a potential biological mediator of the association between maternal vitamin D concentrations during pregnancy and child neurodevelopment.
To better understand the relationship between maternal vitamin D during pregnancy and child neurodevelopment, this study investigated the associations between second trimester maternal vitamin D concentrations and cognitive, language, motor, and behavioural outcomes in preschool-aged children using data from a Canadian pregnancy cohort study. Maternal vitamin D concentrations were initially examined as a continuous exposure. In supplementary models, maternal vitamin D levels were categorized into quartiles to assess potential non-linear associations with neurodevelopmental outcomes. Additional analyses classified participants according to the recommended sufficiency thresholds (i.e., ≥20 and ≥30 ng/mL) to evaluate differences in child neurodevelopmental outcomes between groups. Finally, an exploratory sub-analysis was conducted to determine whether infant EAA mediated the association between maternal vitamin D concentrations during pregnancy and child neurodevelopmental outcomes.
2. Subjects and methods
2.1. Participants and procedure
The present study included 252 mother-child pairs from the APrON study, a longitudinal cohort study of pregnant women and their children living in Calgary and Edmonton, Alberta, Canada (22). Eligible participants were 16 years of age or older, <27 weeks' gestation at time of enrolment, and recruited between 2009 and 2012 (N = 2,189) (22). Inclusion criteria for the present study were as follows: (i) maternal vitamin D concentrations quantified in plasma samples collected during the second trimester of pregnancy, (ii) pregnant individuals had a singleton pregnancy, and (iii) children participated in a neurodevelopmental assessment that examined cognitive, language, motor, and behavioural skills at 3- to 5-years of age (see Supplementary Figure S1 for participant flowchart). For a sub-sample of 130 maternal-child pairs, buccal epithelial cell (BEC) cheek swabs were collected from infants at 3-months of age and profiled for DNA methylation (see Section 2.3 below). Mothers provided informed consent prior to the collection of biospecimens and their children's neurodevelopment assessments. Cognitive, language, and motor outcomes were assessed by trained psychometrists supervised by a registered psychologist (DD). Behavioural outcomes were assessed using a standardized parent-completed measure. This study was approved by the University of Calgary Health Research Ethics Board (REB14-1702) and the University of Alberta Health Research Ethics Biomedical Panel (Pro00002954).
2.2. Maternal vitamin D concentration measurement
Maternal venous blood was collected in the second trimester of pregnancy by a certified phlebotomist using an established protocol (23). Whole blood was fractionated into red blood cells, serum, and plasma, which were stored separately at −80℃. Vitamin D metabolites, including 25-hydroxyvitamin D2 [25(OH)D2], 1,25-dihydroxyvitamin D2 [1,25(OH)2D2], 25-hydroxyvitamin D3 [25(OH)D3], 3-epi-25hydroxyvitamin D3 [3-epi-25(OH)D3], and 1,25-dihydroxyvitamin D3 [1,25(OH)2D3], were extracted from maternal plasma by liquid-liquid extraction using hexane and ethyl acetate. These analytes were separated by liquid chromatography (Agilent 1290) using reverse-phase conditions and detected by tandem mass spectrometry (MS/MS) in positive-ion mode (Agilent 6495B MS/MS). The lowest limit of quantitation (LLOQ) for each of the vitamin D metabolites was 0.78 ng/mL. The quantitation assay was validated using pooled plasma samples. As described in the protocol (23), the standards from the National Institute for Standards and Technology (Standard Reference Material 2970) and Vitamin D External Quality Assessment Scheme procedures were followed. Concentrations of the forms 25-hydroxyvitamin D2 [25(OH)D2], 1,25-dihydroxyvitamin D2 [1,25(OH)2D2], and 1,25-dihydroxyvitamin D3 [1,25(OH)2D3] were below the LLOQ in all maternal samples (23). Maternal 25(OH)D concentrations are computed as the sum of 25(OH)D3 (calcifediol/calcidiol) and 25-hydroxyvitamin D2 [25(OH)D2], however 25(OH)D2 concentrations were 99.6% (n = 251) below the LLOQ for the preset sample. Thus, for the purposes of this study, the metabolite 25(OH)D3 (calcifediol) was used as our measure of maternal vitamin D.
2.3. Sub-sample with infant buccal epithelial cell samples
For a sub-sample of the present study (n = 130), buccal epithelial cell (BEC) cheek swab samples were collected when infants were 3-months of age. BECs were collected using sterile cytology brushes that were rubbed up and down against the infant's cheek 10 times. This process was repeated with two different brushes. Brushes were then air-dried upright for 10–15 min at room temperature, ensuring that the brush-end did not come into contact with any surface. Dried brushes were stored in a plastic bag at −80 °C until the DNA was extracted.
2.3.1. DNA extraction
BEC DNA was extracted by placing buccal cell cytology brushes in a cell lysis mixture and purified using the Gentra Puregene Kit. Cell lysis buffer and proteinase K were added to the samples and incubated at 56 °C for 2–4 h while occasionally mixing. The brushes were removed from the tubes and samples were centrifuged. Samples were processed for DNA purification using the Autopure method and the cell lysate program and left to air to evaporate excess ethanol, and a low-TE buffer was added to the samples. The samples were centrifuged, rehydrated, and transferred into microtubes for long-term storage at 4 °C in the Alberta Children's Hospital Genetics Laboratory.
2.3.2. DNA methylation microarray
Genomic DNA was bisulfite-converted and quantified by the Infinium HumanMethylation EPIC v1.0 (850k) BeadChip microarray (Illumina, San Diego, CA, USA). The 850k platform quantified DNA methylation status at >850,000 CpG methylation sites. In this platform, each CpG site of bisulfite-converted DNA is targeted by probes that differentiate between “methylated” and “unmethylated” loci through different dye colours (i.e., green and red). Beta (β) values, or methylation levels, were calculated using Illumina GenomeStudio software (Illumina, San Diego, CA, USA), which divides each methylation probe signal intensity by the sum of methylated and unmethylated probe signal intensities. Calculated β values range from 0 (i.e., genomic site is completely unmethylated across all cells in the sample) to 1 (i.e., genomic site is fully methylated across all cells in the sample). Raw data was pre-processed through the same pipeline as previous APrON studies (24).
2.3.3. Infant epigenetic age acceleration (EAA) estimation
The Pediatric-Buccal-Epigenetic (PedBE) clock was used to determine EAA. It is a validated tool for use with pediatric populations between the ages of 0 to 20 years and calculates biological age based on 94 CpG sites (20). The PedBE clock was developed to accurately estimate biological age in pediatric populations, as adult-derived epigenetic clocks do not generalize well to children because of the rapid changes in DNA methylation that occur during early development (20). The R code provided by the developers was used to generate PedBE estimates and is available online: https://github.com/kobor-lab/Public-Scripts/. EAA estimates were the residuals from a linear regression of PedBE-estimated biological age on the infants' chronological age and estimated BEC proportions. This is in line with the recommendations from previous research (20, 25), and estimated BEC proportion was derived using the EpiDISH package (26).
It is important to note that the infant age distribution at sample collection in this study was narrow (mean = 13.0 weeks, SD = 3.6 weeks). There was a range of ages of sample collection (range = 8.4 to 39.7 weeks), but 97.7% of samples were collected within 2SD of the mean (i.e., between 8.4 and 17.9 weeks). In samples with this type of limited age distribution, the correlation between chronological age and DNA methylation-derived biological age is typically weak because there is minimal variation in chronological age. Consequently, although epigenetic age acceleration (EAA) and epigenetic age deviation (EAD; defined as the difference between biological and chronological age) are conceptually distinct measures, they can yield highly similar estimates under these conditions. In this sub-sample, EAA and EAD were highly correlated (r = 0.97), indicating that the two measures were largely interchangeable.
2.4. Neurodevelopmental assessments
When children were an average of 4.3 years (SD = 0.48) they completed a neurodevelopmental assessment. Children's intelligence was assessed using the Wechsler Preschool and Primary Scale of Intelligence—Fourth EditionCDN (WPPSI-IVCDN) (27). It provides a Full-Scale Intelligence Quotient (FSIQ) score for children aged 2 years 6 months to 7 years 7 months. The FSIQ score is age-standardized with a mean of 100 (SD = 15, range: 40–160); with higher scores indicating better performance. The WPPSI-IVCDN FSIQ is a valid measure of intelligence in young children with high internal consistency (r = 0.96) (28).
The Developmental NEuroPSYchological Assessment—Second Edition (NEPSY-II) (29) Phonological Processing and Speeded Naming subtests were used to assess language development. The Phonological Processing subtest assesses phonemic awareness, while the Speeded Naming subtest assesses rapid semantic access to and production of names of colours and shapes (in preschool-aged children). Age-standardized scaled scores have a mean of 10 (SD = 3; range: 1–19); higher scores on these subtests indicate better performance. These NEPSY-II subtests have adequate to high test-retest reliability (r ≥ 0.60), as well as good construct overlap demonstrated through medium to high correlations with the Wechsler Intelligence Scale for Children—Fourth Edition (WISC-IV) Verbal Comprehension Index (VCI) (30).
Motor development was assessed using the Movement Assessment Battery for Children—Second Edition (MABC-II) (31). This test assesses manual dexterity, aiming and catching, and balance. Standard scores (mean = 10; SD = 3) for each area are summed to generate a standard Total score (mean = 10, SD = 3; range: 1–19); higher scores indicate better performance. The MABC-II Total score has demonstrated good test-retest reliability (r = 0.80) (32).
Behavioural problems were assessed using the Internalizing Problems and Externalizing Problems composite scales of the Behavior Assessment System for Children—Second Edition (BASC-II) Parent Rating Scale—Preschool for ages 2–5 (33). The Internalizing Problems scale includes the Anxiety, Depression, and Somatization scale scores, while the Externalizing Problems scale includes the Hyperactivity, Aggression, and Conduct Problems scale scores. Results are presented as T scores (mean = 50, SD = 10), with higher scores indicating more behaviour problems. The Internalizing and Externalizing composites scales have high internal consistency and test-retest reliability (34).
2.5. Covariates
Variables previously identified as associated with maternal vitamin D concentrations and/or child cognitive, language, motor, and behavioural development were considered for covariates. Potential covariates included the following maternal characteristics: mother's age at the birth of the child, education, pre-pregnancy body mass index (BMI), self-reported ethnic background, household income, marital status, delivery mode, parity, and season the maternal blood sample was obtained (35–41). Few women in the overall APrON cohort reported smoking or consuming alcohol during pregnancy, so these were not examined as covariates (42). Child characteristics included gestational age at birth and sex (43, 44). Dichotomous variables in models were coded as per previous cohort publications (45, 46) (for more details see Supplementary Materials).
2.6. Statistical analyses
All statistical analyses were performed in version 29.0 of SPSS (IBM Corp., 2024). Missing data was imputed for all variables using imputation techniques in SPSS 29.0. Only 0.15% of the total data was missing. The automatic imputation procedure in SPSS was used with fully conditional specification (FCS), with 10 iterations and 5 imputed datasets. Continuous variables were imputed using linear regression, while categorical variables were imputed using logistic regression. The imputation assumed that data were missing at random (MAR). Estimates across imputed datasets were pooled using Rubin's rules. Further detail on the imputation methods and model diagnostics is provided in the Supplementary Materials. For tests that did not support the use of multiple imputed datasets, either one imputed dataset was chosen at random for analysis or values were reported across all five imputed datasets, as appropriate. In the primary analyses, covariate-adjusted linear regressions were conducted with 25(OH)D3 concentrations examined as a continuous variable. The Benjamini-Hochberg procedure (47) was used to control the false discovery rate (FDR) and correct for multiple comparisons in the primary regression analyses. We report raw p-values and adjusted p-values (q-values) for all primary models.
Sensitivity analyses assessed the robustness of these associations when adjusting for additional maternal nutrient biomarkers including plasma concentrations of vitamin B12 (holotranscobalamin), phospholipid fatty acids [i.e., arachidonic acid (ARA) and docosahexaenoic acid (DHA)], total omega-3 long-chain polyunsaturated fatty acids (n-3 LCPUFA) [i.e., DHA + eicosapentaenoic acid (EPA) + docosapentaenoic acid (DPA)], ratio of DHA:ARA, serum ferritin, red blood cell folate, and energy-adjusted choline.
Supplementary analyses included covariate-adjusted linear regression models that grouped maternal 25(OH)D3 concentrations into quartiles (see Supplementary Table S1) and examined associations with neurodevelopmental outcome measures. In addition, maternal 25(OH)D3 concentrations were dichotomized as ≥30 ng/mL vs. <30 ng/mL (14) and group differences in neurodevelopment outcomes were examined. Separate supplementary analyses examined maternal 25(OH)D3 concentrations dichotomized as ≥20 ng/mL vs. <20 ng/mL (13). However, due to a small number of mothers who had <20 ng/mL of plasma vitamin D (n = 7), non-parametric Mann–Whitney U-tests were used to investigate differences between these groups on neurodevelopmental outcomes. The Mann–Whitney U-tests were performed as descriptive analyses only and used only complete cases (n = 247), as SPSS does not support the use of pooled imputed datasets for this analysis (IBM Corp., 2024). Therefore, participants missing data for any of the exposure, outcome, or covariate variables were excluded from the Mann–Whitney U analyses, resulting in a sample size of 247. It is important to note that maternal 25(OH)D3 concentrations were used in these supplementary analyses in lieu of a total measure; due the limits of the analysis method we were unable to compute total 25(OH)D [i.e., sum of 25(OH)D2 and 25(OH)D3]) as 25(OH)D2 concentrations were >99% below the LLOQ. Thus, in this study, the measure of maternal vitamin D [i.e., 25(OH)D3] could be considered the best “proxy” to a total measure.
The exploratory sub-analysis consisted of covariate-adjusted mediation models among participants with available EAA data (n = 130). Mediation analyses were conducted using the bootstrapping approach with 5,000 samples to estimate total and indirect effects with version 4.2 of the PROCESS macro for SPSS (48). In these analyses, continuous maternal vitamin D concentration was specified as the predictor variable (X), each neurodevelopmental measure was specified as the outcome (Y), and EAA was specified as the mediator (M). A power analysis from a simulation study indicated that a sample size of 126 would be adequately powered (0.80) to detect a mediation effect when there was a small-to-medium effect for path a (size of β path = 0.26; this is the path from maternal vitamin D to EAA) and a medium effect for path b (size of β path = 0.39; this is the path from EAA to the outcome) using the percentile bootstrap method of the PROCESS macro (49). Thus, the sub-analysis sample size of 130 was adequately powered to detect such an effect.
A directed acyclic graph (DAG) of the potential covariates considered is included in the Supplementary Materials (see Supplementary Figure S2). This study employed a common epidemiological approach to covariate selection in stepwise regression modeling (50, 51), whereby covariates were included in the primary, supplementary, and sub-analysis models if they were associated with the outcome at p < 0.20 (see Supplementary Table S2). Mediation analyses also adjusted for children's sex (Supplementary Figures S3–S8). All neurodevelopmental outcome scores were age standardized and thus child age was not included as a covariate in any statistical models. Associations with p ≤ 0.05 and confidence intervals that did not include 0 were considered statistically significant. Mann–Whitney U-test results were considered statistically significant at an asymptotic two-tailed p ≤ 0.05.
3. Results
3.1. Study population characteristics
Maternal and child characteristics for the study sample and sub-analysis sample are presented in Supplementary Table S3 (characteristics for the overall APrON cohort are presented in Supplementary Table S4). In the study sample, mothers' average age at childbirth was 32.38 ± 3.78 years and their average pre-pregnancy BMI was 24.52 ± 4.49. Most participants identified as having European ethnicity (n = 222, 88.1%), were married/cohabiting (n = 249, 98.8%), had a household income ≥$70,000 CAD (n = 218, 86.5%), held a university degree (n = 199, 79.0%), and gave birth vaginally (n = 197, 78.2%). About half of the mothers were primiparous (n = 136, 54%). Mothers had an average 25(OH)D3 concentration of 37.01 ± 10.21 ng/mL (range: 8.2–83.1 ng/mL). Season of maternal blood sample collection was evenly distributed throughout the year, although winter had the lowest representation (spring, n = 71, 28.2%; summer, n = 80, 31.7%; fall, n = 57, 22.6%; winter, n = 44, 17.5%). Children (51.2% male) had an average gestational age at birth of 39.32 ± 1.44 weeks, with most born full-term (≥37 gestational weeks) (n = 243, 96.4%). The mean ages and age ranges as well as the means and standard deviations of children's scores across the neurodevelopmental outcome measures suggest that the study sample demonstrated typical development (Supplementary Table S5).
3.2. Maternal vitamin D concentrations and child neurodevelopment
No significant associations were observed between continuous maternal vitamin D concentrations and children's WPPSI-IVCDN FSIQ, NEPSY-II Phonological Processing and Speeded Naming subtests, and BASC-II Internalizing and Externalizing Problems composite scores (Table 1). In contrast, maternal vitamin D concentrations were positively associated with MABC-II Total scores (B = 0.04, 95% CI = 0.004, 0.08, p = 0.03), although this association did not survive FDR correction (q = 0.18). We examined Q-Q plots of the primary linear regression models, which indicated that residuals followed a normal distribution (see Supplementary Figure S3). Additionally, case-complete analyses (n = 247) were completed and showed similar results to the original primary analyses (Supplementary Table S6).
Table 1.
Covariate-adjusted linear regressions of maternal 25(OH)D3 concentrations and neurodevelopmental outcomes (N = 252).
| Neurodevelopmental outcome measure | Unstandardized beta coefficient | Standardized beta coefficient | SE | 95% CI | p-value | q-value |
|---|---|---|---|---|---|---|
| WPPSI-IVCDN FSIQ | 0.10 | 0.07 | 0.08 | −0.07, 0.26 | 0.24 | 0.48 |
| NEPSY-II Phonological Processing | 0.004 | 0.02 | 0.02 | −0.03, 0.04 | 0.82 | 0.82 |
| NEPSY-II Speeded Naming | 0.02 | 0.08 | 0.02 | −0.01, 0.06 | 0.20 | 0.48 |
| MABC-II Total | 0.04 | 0.14 | 0.02 | 0.004, 0.08 | 0.03* | 0.18 |
| BASC-II Internalizing Problems | 0.05 | 0.05 | 0.06 | −0.07, 0.16 | 0.43 | 0.65 |
| BASC-II Externalizing Problems | −0.02 | −0.02 | 0.05 | −0.12, 0.08 | 0.72 | 0.82 |
Covariates included in the linear regressions: WPPSI-IVCDN FSIQ—maternal age, maternal race, marital status, child sex, maternal education, maternal pre-pregnancy BMI, household income; NEPSY-II Phonological Processing—maternal age, maternal race, marital status, child sex, maternal pre-pregnancy BMI, infant gestational age; NEPSY-II Speeded Naming—maternal age, maternal race, marital status, child sex, maternal education, household income, vitamin D sample season; MABC-II Total—maternal age, maternal race, marital status, child sex, maternal pre-pregnancy BMI; BASC-II Internalizing Problems—maternal age, maternal race, marital status, child sex, maternal education, delivery mode, parity; BASC-II Externalizing Problems—maternal age, maternal race, marital status, child sex, maternal education, maternal pre-pregnancy BMI, household income.
p < 0.05.
3.3. Sensitivity analyses
This association remained with a similar effect estimate and level of statistical significance in the sensitivity analyses adjusting for additional maternal nutrient biomarkers (Supplementary Table S7). No additional significant associations were identified in the sensitivity analyses.
3.4. Supplementary analyses
In analyses examining maternal vitamin D quartiles, in which the first quartile was the reference group, there were no associations observed between maternal vitamin D concentrations and children's MABC-II Total scores (Table 2). However, a positive association between the second quartile of maternal vitamin D concentrations and children's WPPSI-IVCDN FSIQ scores was found (i.e., relative to Q1, children of mothers in Q2 scored 11.8 points higher on the FSIQ). No associations were observed between maternal vitamin D concentrations dichotomized according to the 30 ng/mL threshold and child neurodevelopmental outcomes (Table 3). Similarly, the Mann–Whitney U-tests used for descriptive analysis only indicated no significant difference in neurodevelopmental outcomes between children of mothers who met the 20 ng/mL threshold and those who did not (Supplementary Table S8).
Table 2.
Covariate-adjusted linear regressions of maternal 25(OH)D3 concentrations categorized as quartiles and neurodevelopmental outcomes (N = 252).
| Neurodevelopmental outcome measure | Quartile | Difference vs. Q1 (95% CI) | p-value | Overall F (df) | Overall p-value | Partial eta squared |
|---|---|---|---|---|---|---|
| WPPSI-IVCDN FSIQ | – | – | – | 2.96 (3, 201) | 0.03* | 0.04 |
| Q1 | Reference | – | – | – | – | |
| Q2 | 11.79 (3.05, 20.5) | 0.008 | – | – | – | |
| Q3 | 0.30 (−7.61, 8.21) | 0.94 | – | – | – | |
| Q4 | 7.43 (−0.31, 15.17) | 0.06 | – | – | – | |
| NEPSY-II Phonological Processing | – | – | – | 0.75 (3, 231) | 0.53 | 0.01 |
| Q1 | Reference | – | – | – | – | |
| Q2 | −0.19 (−1.83, 1.45) | 0.82 | – | – | – | |
| Q3 | 0.08 (−2.04, 2.19) | 0.94 | – | – | – | |
| Q4 | −1.40 (−3.40, 0.60) | 0.17 | – | – | – | |
| NEPSY-II Speeded Naming | – | – | – | 0.53 (3, 153) | 0.67 | 0.01 |
| Q1 | Reference | – | – | – | – | |
| Q2 | 0.32 (−1.69, 2.33) | 0.75 | – | – | – | |
| Q3 | −0.57 (−2.65, 1.51) | 0.59 | – | – | – | |
| Q4 | 0.12 (−1.70, 1.93) | 0.90 | – | – | – | |
| MABC-II Total | – | – | – | 0.52 (3, 232) | 0.67 | 0.01 |
| Q1 | Reference | – | – | – | – | |
| Q2 | 0.22 (−1.30, 1.75) | 0.78 | – | – | – | |
| Q3 | −0.81 (−2.82, 1.19) | 0.43 | – | – | – | |
| Q4 | 0.56 (−1.34, 2.45) | 0.54 | – | – | – | |
| BASC-II Internalizing Problems | – | – | – | 0.75 (3, 175) | 0.52 | 0.01 |
| Q1 | Reference | – | – | – | – | |
| Q2 | 2.82 (−2.16, 7.79) | 0.27 | – | – | – | |
| Q3 | 0.28 (−4.65, 5.21) | 0.91 | – | – | – | |
| Q4 | 2.42 (−2.05, 6.89) | 0.29 | – | – | – | |
| BASC-II Externalizing Problems | – | – | – | 0.80 (3, 201) | 0.50 | 0.01 |
| Q1 | Reference | – | – | – | – | |
| Q2 | 3.10 (−2.49, 8.69) | 0.28 | – | – | – | |
| Q3 | 3.54 (−1.52, 8.60) | 0.17 | – | – | – | |
| Q4 | 2.13 (−2.82, 7.08) | 0.40 | – | – | – |
Covariates are the same as those included in the linear regressions of continuous maternal 25(OH)D3 concentrations and neurodevelopmental outcomes.
*p < 0.05.
Table 3.
Covariate-adjusted linear regressions of maternal 25(OH)D3 concentrations dichotomized as “sufficient” vs. “insufficient” at 30 ng/mL threshold recommended by the Endocrine Society's 2011 guidelines† and neurodevelopmental outcomes (N = 252).
| Neurodevelopmental outcome measure | Unstandardized beta coefficient | SE | 95% CI | p-value |
|---|---|---|---|---|
| WPPSI-IVCDN FSIQ | −0.12 | 1.97 | −3.97, 3.74 | 0.95 |
| NEPSY-II Phonological Processing | 0.10 | 0.46 | −0.79, 1.00 | 0.82 |
| NEPSY-II Speeded Naming | 0.29 | 0.41 | −0.51, 1.10 | 0.48 |
| MABC-II Total | 0.52 | 0.44 | −0.34, 1.38 | 0.24 |
| BASC-II Internalizing Problems | 1.25 | 1.36 | −1.41, 3.91 | 0.36 |
| BASC-II Externalizing Problems | 0.73 | 1.23 | −1.67, 3.13 | 0.55 |
Covariates are the same as those included in the linear regressions of continuous maternal 25(OH)D3 concentrations and neurodevelopmental outcomes.
The Endocrine Society's 2011 30 ng/mL threshold refers to total 25(OH)D; this total measure could not be calculated in the present study and thus maternal 25(OH)D3 concentrations were used in lieu of the total. These analyses should be a considered a historical sensitivity analysis to improve interpretation across the literature as the 2024 guidelines no longer endorse using this threshold.
3.5. Exploratory sub-analyses: potential mediation by EAA
In mediation models examining WPPSI-IVCDN FSIQ, BASC-II Internalizing and Externalizing Problems scores, and NEPSY-II Phonological Processing and Speeded Naming scores, no significant associations were observed between maternal vitamin D and infant EAA, between infant EAA and neurodevelopmental outcome, or in direct and indirect effects of the mediation models (Supplementary Figures S4–S9).
In these analyses, higher maternal vitamin D concentrations were positively associated with higher MABC-II Total scores (B = 0.05, 95% CI = 0.0009, 0.10, p = 0.05) (Supplementary Figure S9). Maternal vitamin D concentrations were not significantly associated with infant EAA (B = −0.02, 95% CI = −0.09, 0.06, p = 0.67), nor was infant EAA associated with MABC-II Total scores (B = 0.07, 95% CI = 0.05, 0.19, p = 0.24). There was no evidence that infant EAA mediated the relationship between maternal vitamin D concentrations and MABC-II Total scores (indirect effect = −0.004, 95% CI = −0.008, 0.0006).
In addition to these exploratory sub-analyses, several additional, related analyses were completed. First, we completed an analysis comparing the maternal and child characteristics for the sub-analysis sample to those not included in this sample (i.e., maternal-child pairs where the infant did not provide BEC cheek swabs that could be profiled for DNA methylation; n = 122). There were no significant differences in sample characteristics between these samples (Supplementary Table S9). Second, we re-ran the primary analyses using this sub-sample; many point estimates were similar, and the only change to significance was a drop in p-value for the previously noted association between maternal vitamin D concentrations and MABC-II Total scores (from p = 0.03 in the primary analyses) to p = 0.06 in these analyses (Supplementary Table S10). Third, we checked the results of mediation analyses in the complete cases only (n = 126; Supplementary Table S11) as well as across all five imputation datasets, for comparability (Supplementary Tables S12–S16); results were largely the same. Lastly, a post hoc sensitivity analysis that restricted the sub-analysis sample to individuals whose ages at sample collection were within 2SD of the mean (n = 127), was completed. The results of these mediation models were largely unchanged from the original sub-analyses (Supplementary Table S17).
4. Discussion
In this sample of 252 mother-child pairs from the APrON cohort higher maternal plasma vitamin D concentrations during the second trimester were associated with better motor performance among children aged 3- to 5-years. Although this association was small in magnitude (β = 0.14; i.e., each one standard deviation increase in maternal vitamin D was associated with a 0.14 standard deviation increase in MABC-II Total Test scores) and did not survive the correction for multiple comparisons (p = 0.03, q = 0.18), this association remained in sensitivity analyses adjusting for additional maternal nutrient biomarkers (i.e., vitamin B12, ARA, DHA, DHA + EPA + DPA, DHA:ARA, ferritin, folate, and choline), suggesting that the observed association was robust to adjustment for other nutritional factors. In contrast, no significant associations were identified when maternal vitamin D was categorized into quartiles or classified according to established thresholds (i.e., 20 and 30 ng/mL). These findings suggest that the nature of the association between maternal vitamin D and child neurodevelopment may differ depending on how vitamin D exposure is modeled. Specifically, analyses treating vitamin D concentrations as a continuous measure provided initial evidence of a positive relationship, whereby higher second-trimester vitamin D concentrations were associated with better motor performance. However, no significant associations were observed when vitamin D concentrations were examined using categorical cut-points or quartile-based groupings, and thus future research using a larger study sample is needed that comprehensively evaluates these potential associations using more advanced modeling techniques (e.g., restricted cubic splines, polynomial regression, etc.).
The mixed findings reported in previous epidemiological studies examining associations between maternal vitamin D concentrations during pregnancy and child neurodevelopmental outcomes may be attributed, in part, to differences in how maternal vitamin D status was modelled (i.e., continuous vs. categorical measures) and varying definitions of vitamin D “sufficiency” thresholds (10, 52, 53). Future research should further investigate how methodological decisions in modelling maternal vitamin D concentrations, including the selection of covariates and “sufficiency” cutoffs, influence observed associations with child neurodevelopment. Such work may help clarify inconsistencies in the literature and inform the development of more appropriate and sensitive statistical approaches for characterizing the relationship between maternal vitamin D status and child neurodevelopment.
The significant association observed between maternal vitamin D concentrations and motor outcomes in this study is consistent with a study where Morales et al. (38) reported that maternal second trimester 25(OH)D3 concentrations were positively associated with motor outcomes at 14-months of age. This is also similar to findings from the ECLIPSES study, in which maternal 25(OH)D3 concentrations <8 ng/mL during the third trimester were associated with poorer motor scores in 40-day old infants (59). However, these findings contrast with studies by Daraki et al. (12) and Laird et al. (52), which reported no associations between maternal vitamin D concentrations and motor outcomes in children aged 4- and 5-years, respectively. These inconsistencies across epidemiological research may be partly explained by differences in the form of vitamin D measured. For example, our study and studies reporting similar findings (38, 54) measured 25(OH)D3, whereas studies that did not observe associations measured total 25(OH)D, which includes both 25(OH)D2 and 25(OH)D3 (12, 52). Although 25(OH)D₃ represents the primary circulating form of vitamin D and is commonly used to assess vitamin D status, total 25(OH)D is considered the preferred biomarker because it reflects both endogenous synthesis and dietary intake (55). Nevertheless, differences in the form of vitamin D measured are unlikely to fully explain the variability across studies. Other methodological differences including timing of maternal vitamin D assessment during pregnancy, adjustment for confounding variables, neurodevelopmental assessment tool used, and age at assessment, may also contribute to discrepancies in observed findings (8, 56, 57).
In the present study, maternal continuous vitamin D concentrations were not associated with cognitive, language, and behavioural outcomes in preschool-aged children. Similarly, previous studies have reported no associations between maternal 25(OH)D concentrations measured during the second or third trimester and child IQ (58), verbal fluency (53), or behavioural difficulties (9). One potential explanation for these null findings is that most of the participating women in APrON had sufficient vitamin D concentrations during the second trimester of pregnancy (≥20 ng/mL: n = 245, 97.2%; ≥30 ng/mL: n = 190, 75.4%). Consequently, limited variability in maternal vitamin D status may have reduced the ability to detect associations with children's cognitive, language, and behavioural outcomes, particularly if association are more pronounced at lower vitamin D concentrations.
This study also conducted exploratory analyses that examined EAA as a biologically plausible mediator of the association between maternal vitamin D concentrations and child neurodevelopment. Infant EAA (which approximated epigenetic age difference), estimated from BECs using the PedBE clock, did not mediate associations between maternal second trimester vitamin D concentrations and cognitive, language, motor, or behavioural outcomes at age 3–5 in this sub-sample. However, future studies involving larger and more diverse populations, alternative tissues for DNA methylation profiling, different epigenetic clocks and other measures of biological aging, or different age ranges may identify potential mediation pathways. It is also notable that, consistent with findings from the primary analyses, maternal vitamin D concentrations remained positively associated with motor outcomes in the sub-analysis. These findings highlight the need for more research to clarify the relationship between maternal vitamin D status and child neurodevelopment, and to determine whether EAA or other biological aging biomarkers may serve as potential mediating mechanisms.
A major strength of this study was the comprehensive assessment of maternal vitamin D concentrations using multiple modelling approaches including continuous measures, quartiles, and dichotomized thresholds based on recommendations from the Institute of Medicine (13) and the Endocrine Society (14). This approach reduces the likelihood that findings were driven by specific modelling decisions or arbitrary thresholds. Additionally, neurodevelopmental outcomes were assessed using standardized measures, with cognitive, language, and motor outcome assessments administered by trained psychometrists. Together, these methodological strengths reduce measurement bias and enhance confidence in the validity of the observed associations.
A key limitation of this study was the predominance of participants of self-reported European ethnicity (n = 222, 88.1%), which limits the generalizability of the findings to more ethnically diverse populations. This consideration is particularly relevant given evidence that vitamin D deficiency (<20 ng/mL) is more prevalent among individuals who self-identified as Black (59). Higher melanin concentrations reduce ultraviolet radiation penetration, thereby decreasing cutaneous vitamin D synthesis and contributing to lower vitamin D concentrations (60). In this study, the small number of non-European participants may have limited statistical power to detect associations between maternal vitamin D concentrations and child neurodevelopmental outcomes, particularly, if vitamin D insufficiency is more common among these populations. Further, most of the maternal vitamin D forms (i.e., 25(OH)D2, 1,25(OH)2D2, 25(OH)D2, and 1,25(OH)2D3) were >99% below the lowest limit of quantification, and thus this study utilized maternal 25(OH)D3 (calcifediol/calcidiol) concentrations as our measure of maternal vitamin D. Based on the limitations of the analysis method, this could be considered to approximate a “proxy” or total measure of maternal vitamin D in the present sample. However, future investigations which use more sensitive analysis methods are needed to clarify the relationships between maternal vitamin D forms during pregnancy and children's neurodevelopment.
It is possible the unmeasured confounders may have influenced the associations between maternal vitamin D concentrations and child neurodevelopmental outcomes. For example, although few individuals reported smoking (2.2%) or consuming alcohol during pregnancy (3.6%) in the overall APrON cohort (41), these are well-documented risk factors for poorer neurodevelopmental outcomes in children that may also interact with other factors (e.g., socioeconomic status, lifestyle factors, etc.) (61, 62) and should be explored further in future research. Another limitation of this study was the narrow distribution of infant age at BEC sample collection (mean = 13.0 weeks ± 3.6) and subsequent lack of correlation between biological age and chronological age. In addition to the small sample size for our sub-analysis that examined potential mediation by PedBE EAA (n = 130), which was underpowered to detect small effects, it is possible that this limited age range impacted our power to detect potential associations between maternal vitamin D concentrations, EAA, and child neurodevelopmental outcomes. Future multi-model research that integrates longitudinal nutritional, epigenetic, and neurodevelopment data in larger samples is needed to examine whether associations with biological aging markers such as EAA may be observed at different ages or age ranges.
In conclusion, the present study highlights preliminary evidence of a small, nominal association (that did not survive multiple testing correction) between maternal vitamin D concentrations during pregnancy and motor outcomes in preschool-aged children. Given the limited clinical trial evidence regarding whether specific maternal 25(OH)D thresholds provide neurodevelopmental benefits (15), further research is needed to determine whether an optimal target concentration exists. Future studies examining different forms of maternal vitamin D, approaches to modelling vitamin D exposure, and varying definitions of vitamin D sufficiency may help to clarify the relationship between maternal vitamin D status and child neurodevelopment outcomes. Such evidence ultimately informs nutritional recommendations during pregnancy aimed at supporting optimal child neurodevelopment.
Acknowledgments
We are extremely grateful to all the families who took part in this study and the APrON team investigators, research assistants, graduate and undergraduate students, volunteers, clerical staff, and managers. We acknowledge the significant contributions of the APrON Study Team whose individual members are B.J. Kaplan, D. Dewey, C.J. Field, R.C. Bell, F.P. Bernier, M. Cantell, L.M. Casey, M. Eliasziw, A. Farmer, L. Gagnon, G.F. Giesbrecht, L. Goonewardene, D. Johnston, L. Kooistra, N. Letourneau, D.P. Manca, J. W. Martin, L.J. McCargar, M. O'Beirne, V.J. Pop, A.J. Deane, and N. Singha, and the APrON Management Team: N. Letourneau (current PI), R.C. Bell, D. Dewey, England-Mason, G., C.J. Field, L. Forbes, G.F. Giesbrecht, C. Lebel, B. Leung, and K. Ross.
Funding Statement
The author(s) declared that financial support was received for this work and/or its publication. An interdisciplinary team grant provided by Alberta Innovates Health Solutions (formally the Alberta Heritage Foundation for Medical Research) enabled the establishment of the APrON cohort. The collection and analysis of data presented in this manuscript was supported through funding provided by the Canadian Institutes of Health Research (CIHR) (MOP-123535, PG-201909, PS-156069) and The W. Garfield Weston Foundation (RES0054834).
Footnotes
Edited by: Pierre-Paul Vidal, Université Paris Cité, France
Reviewed by: Spyridon N. Karras, Aristotle University of Thessaloniki, Greece
Vittorio Ferrari, IRCCS University Hospital of Bologna Sant Orsola Polyclinic, Italy
Data availability statement
The data presented in this article are not available due to the ethic boards approvals and initial informed consents provided by participants. Therefore, we are unable to release any of the study data into a depository or other source. Requests for access to the dataset should be directed to the corresponding author.
Ethics statement
The studies involving humans were approved by the University of Calgary Health Research Ethics Board (REB14-1702) and the University of Alberta Health Research Ethics Biomedical Panel (Pro00002954). The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation in this study was provided by the participants' legal guardians/next of kin.
Author contributions
PC: Data curation, Visualization, Formal analysis, Conceptualization, Writing – review & editing, Writing – original draft. CF: Methodology, Investigation, Writing – review & editing, Resources, Project administration, Validation, Funding acquisition. FA: Funding acquisition, Validation, Resources, Writing – review & editing, Project administration, Methodology, Investigation. SM: Validation, Data curation, Software, Writing – review & editing, Investigation. JM: Investigation, Writing – review & editing, Validation, Methodology. GG: Funding acquisition, Writing – review & editing, Investigation, Resources, Project administration, Methodology. NL: Resources, Project administration, Funding acquisition, Methodology, Investigation, Writing – review & editing. MK: Resources, Investigation, Writing – review & editing, Methodology, Funding acquisition. DD: Methodology, Conceptualization, Investigation, Data curation, Writing – review & editing, Resources, Project administration, Funding acquisition, Supervision. GE-M: Validation, Data curation, Methodology, Project administration, Conceptualization, Supervision, Software, Visualization, Writing – review & editing, Funding acquisition.
Conflict of interest
The data presented in this article are not available due to the ethic boards approvals and initial informed consents provided by participants. Therefore, we are unable to release any of the study data into a depository or other source. Requests for access to the dataset should be directed to the corresponding author.
Generative AI statement
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher's note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fped.2026.1967139/full#supplementary-material
References
- 1.Heland S, Fields N, Ellery SJ, Fahey M, Palmer KR. The role of nutrients in human neurodevelopment and their potential to prevent neurodevelopmental adversity. Front Nutr. (2022) 9:992120. 10.3389/fnut.2022.992120 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Cui X, Gooch H, Petty A, McGrath JJ, Eyles D. Vitamin D and the brain: genomic and non-genomic actions. Mol Cell Endocrinol. (2017) 453:131–43. 10.1016/j.mce.2017.05.035 [DOI] [PubMed] [Google Scholar]
- 3.Eyles DW. Vitamin d: brain and behavior. JBMR Plus. (2021) 5(1):e10419. 10.1002/jbm4.10419 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Wagner CL, Taylor SN, Johnson DD, Hollis BW. The role of vitamin d in pregnancy and lactation: emerging concepts. Womens Health (Lond Engl). (2012) 8(3):323–40. 10.2217/WHE.12.17 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Kanike N, Hospattankar KG, Sharma A, Worley S, Groh-Wargo S. Prevalence of vitamin d deficiency in a large newborn cohort from northern United States and effect of intrauterine drug exposure. Nutrients. (2020) 12(7):2085. 10.3390/nu12072085 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Eyles D, Burne T, McGrath J. Vitamin D in fetal brain development. Semin Cell Dev Biol. (2011) 22(6):629–36. 10.1016/j.semcdb.2011.05.004 [DOI] [PubMed] [Google Scholar]
- 7.Melough MM, Murphy LE, Graff JC, Derefinko KJ, LeWinn KZ, Bush NR, et al. Maternal plasma 25-hydroxyvitamin d during gestation is positively associated with neurocognitive development in offspring at age 4–6 years. J Nutr. (2021) 151(1):132–9. 10.1093/jn/nxaa309 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Francis EC, Charron E, Li M, Chen L, Mayo R, Butler LS, et al. Third trimester maternal vitamin D and early childhood socioemotional development. Paediatric Perinatal Epid. (2021) 35(3):350–8. 10.1111/ppe.12736 [DOI] [PubMed] [Google Scholar]
- 9.Whitehouse AJO, Holt BJ, Serralha M, Holt PG, Kusel MMH, Hart PH. Maternal serum vitamin d levels during pregnancy and offspring neurocognitive development. Pediatrics. (2012) 129(3):485–93. 10.1542/peds.2011-2644 [DOI] [PubMed] [Google Scholar]
- 10.Darling AL, Rayman MP, Steer CD, Golding J, Lanham-New SA, Bath SC. Association between maternal vitamin D status in pregnancy and neurodevelopmental outcomes in childhood: results from the avon longitudinal study of parents and children (alspac). Br J Nutr. (2017) 117(12):1682–92. 10.1017/S0007114517001398 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Gould JF, Cuthbert AR, Yelland LN, Gibson RA, Smithers LG, Zhou SJ, et al. Association of cord blood vitamin D with child neurodevelopment at 7 years of age. J Paediatrics Child Health. (2024) 60(7):312–22. 10.1111/jpc.16590 [DOI] [PubMed] [Google Scholar]
- 12.Daraki V, Roumeliotaki T, Koutra K, Chalkiadaki G, Katrinaki M, Kyriklaki A, et al. High maternal vitamin D levels in early pregnancy may protect against behavioral difficulties at preschool age: the rhea mother–child cohort, crete, Greece. Eur Child Adolesc Psychiatry. (2018) 27(1):79–88. 10.1007/s00787-017-1023-x [DOI] [PubMed] [Google Scholar]
- 13.Ross AC, Manson JAE, Abrams SA, Aloia JF, Brannon PM, Clinton SK, et al. The 2011 report on dietary reference intakes for calcium and vitamin d from the institute of medicine: what clinicians need to know. J Clin Endocrinol Metab. (2011) 96(1):53–8. 10.1210/jc.2010-2704 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Holick MF, Binkley NC, Bischoff-Ferrari HA, Gordon CM, Hanley DA, Heaney RP, et al. Evaluation, treatment, and prevention of vitamin d deficiency: an endocrine society clinical practice guideline. J Clin Endocrinol Metab. (2011) 96(7):1911–30. 10.1210/jc.2011-0385 [DOI] [PubMed] [Google Scholar]
- 15.Demay MB, Pittas AG, Bikle DD, Diab DL, Kiely ME, Lazaretti-Castro M, et al. Vitamin d for the prevention of disease: an endocrine society clinical practice guideline. J Clin Endocrinol Metab. (2024) 109(8):1907–47. 10.1210/clinem/dgae290 [DOI] [PubMed] [Google Scholar]
- 16.Carlberg C, Raczyk M, Zawrotna N. Vitamin D: a master example of nutrigenomics. Redox Biol. (2023) 62:102695. 10.1016/j.redox.2023.102695 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Khoury E, Ashraf D, Shiu H, Mohammed HCN, Al-Dewik SGAA, Qoronfleh NI, et al. Epigenetics, vitamin status, maternal nutrition, and fetal development: a spotlight on the importance of precision nutrition. Dietetics. (2026) 5(2):19. 10.3390/dietetics5020019 [DOI] [Google Scholar]
- 18.Ideraabdullah FY, Belenchia AM, Rosenfeld CS, Kullman SW, Knuth M, Mahapatra D, et al. Maternal vitamin D deficiency and developmental origins of health and disease (DOHaD). J Endocrinol. (2019) 241(2):R65–80. 10.1530/JOE-18-0541 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Xue J, Ideraabdullah FY. Effects of maternal vitamin d deficiency on fetal epigenetic reprogramming. FASEB J. (2016) 30:912.1. 10.1096/fasebj.30.1_supplement.912.1 [DOI] [Google Scholar]
- 20.McEwen LM, O’Donnell KJ, McGill MG, Edgar RD, Jones MJ, MacIsaac JL, et al. The PedBE clock accurately estimates DNA methylation age in pediatric buccal cells. Proc Natl Acad Sci USA. (2020) 117(38):23329–35. 10.1073/pnas.1820843116 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Musci RJ, Raghunathan RS, Johnson SB, Klein L, Ladd-Acosta C, Ansah R, et al. Using epigenetic clocks to characterize biological aging in studies of children and childhood exposures: a systematic review. Prev Sci. (2023) 24(7):1398–423. 10.1007/s11121-023-01576-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Letourneau N, Aghajafari F, Bell RC, Deane AJ, Dewey D, Field C, et al. The Alberta pregnancy outcomes and nutrition (APrON) longitudinal study: cohort profile and key findings from the first three years. BMJ Open. (2022) 12(2):e047503. 10.1136/bmjopen-2020-047503 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Evanchuk JL, Kozyrskyj A, Vaghef-Mehrabani E, Lamers Y, Giesbrecht GF, Letourneau N, et al. Maternal iron and vitamin d status during the second trimester is associated with third trimester depression symptoms among pregnant participants in the APrON cohort. J Nutr. (2024) 154(1):174–84. 10.1016/j.tjnut.2023.10.029 [DOI] [PubMed] [Google Scholar]
- 24.England-Mason G, Merrill SM, Gladish N, Moore SR, Giesbrecht GF, Letourneau N, et al. Prenatal exposure to phthalates and peripheral blood and buccal epithelial DNA methylation in infants: an epigenome-wide association study. Environ Int. (2022) 163:107183. 10.1016/j.envint.2022.107183 [DOI] [PubMed] [Google Scholar]
- 25.England-Mason G, Merrill SM, Liu J, Martin JW, MacDonald AM, Kinniburgh DW, et al. Sex-specific associations between prenatal exposure to bisphenols and phthalates and infant epigenetic age acceleration. Epigenomes. (2024) 8(3):31. 10.3390/epigenomes8030031 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Zheng SC, Breeze CE, Beck S, Dong D, Zhu T, Ma L, et al. Epidish web server: epigenetic dissection of intra-sample-heterogeneity with online GUI. Bioinformatics. (2020) 36(6):1950–1. 10.1093/bioinformatics/btz833 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Wechsler D. Wechsler Preschool and Primary Scale of Intelligence. 4th ed. Toronto: Pearson Clinical Assessment Canada; (2012). [Google Scholar]
- 28.Syeda MM, Climie EA. Test review: wechsler preschool and primary scale of intelligence–fourth edition. J Psychoeduc Assess. (2014) 32(3):265–72. 10.1177/0734282913508620 [DOI] [Google Scholar]
- 29.Kemp S, Kirk U, Korkman M. NEPSY. 2nd ed. Toronto: Pearson Clinical Assessment Canada; (2007). [Google Scholar]
- 30.Brooks BL, Sherman EMS, Strauss E. NEPSY-II: a developmental neuropsychological assessment, second edition. Child Neuropsychol. (2009) 16(1):80–101. 10.1080/09297040903146966 [DOI] [Google Scholar]
- 31.Barnett A, Henderson SA, Sugden DA. Movement Assessment Battery for Children. 2nd ed. Toronto: Pearson Clinical Assessment Canada; (2007). [Google Scholar]
- 32.Griffiths A, Toovey R, Morgan PE, Spittle AJ. Psychometric properties of gross motor assessment tools for children: a systematic review. BMJ Open. (2018) 8(10):e021734. 10.1136/bmjopen-2018-021734 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Reynolds CR, Kamphaus RW. Behavior Assessment System for Children. 2nd ed. Toronto: Pearson Clinical Assessment Canada; (2004). [Google Scholar]
- 34.Altmann RA, Reynolds CR, Kamphaus RW, Vannest KJ. BASC-3. In: Kreutzer J, DeLuca J, Caplan B, editors. Encyclopedia of Clinical Neuropsychology. Cham, Switzerland: Springer International Publishing; (2017). p. 1–7. 10.1007/978-3-319-56782-2_1524-2 [DOI] [Google Scholar]
- 35.Basatemur E, Gardiner J, Williams C, Melhuish E, Barnes J, Sutcliffe A. Maternal prepregnancy bmi and child cognition: a longitudinal cohort study. Pediatrics. (2013) 131(1):56–63. 10.1542/peds.2012-0788 [DOI] [PubMed] [Google Scholar]
- 36.Bacharach VR, Baumeister AA. Effects of maternal intelligence, marital status, income, and home environment on cognitive development of low birthweight infants. J Pediatr Psychol. (1998) 23(3):197–205. 10.1093/jpepsy/23.3.197 [DOI] [PubMed] [Google Scholar]
- 37.Keshavarz P, Jandaghi P, Shafiee M, Islam N, Vatanparast H. Maternal vitamin d status among different ethnic groups and its potential contribution to adverse pregnancy and child outcomes. In: Fedotova J, editor. Vitamin D Deficiency. London: IntechOpen; (2020) 1–41. 10.5772/intechopen.90766 [DOI] [Google Scholar]
- 38.Morales E, Guxens M, Llop S, Rodríguez-Bernal CL, Tardón A, Riaño I, et al. Circulating 25-hydroxyvitamin d3 in pregnancy and infant neuropsychological development. Pediatrics. (2012) 130(4):e913–920. 10.1542/peds.2011-3289 [DOI] [PubMed] [Google Scholar]
- 39.Islam MM, Khan MN. Early childhood development and its association with maternal parity. Child Care Health Dev. (2023) 49(1):80–9. 10.1111/cch.13011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Tearne JE. Older maternal age and child behavioral and cognitive outcomes: a review of the literature. Fertil Steril. (2015) 103(6):1381–91. 10.1016/j.fertnstert.2015.04.027 [DOI] [PubMed] [Google Scholar]
- 41.Romero S, Lindström K, Listermar J, Westgren M, Ajne G. Long-term neurodevelopmental outcome in children born after vacuum-assisted delivery compared with second-stage caesarean delivery and spontaneous vaginal delivery: a cohort study. BMJ Paediatrics Open. (2023) 7(1):e002048. 10.1136/bmjpo-2023-002048 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Soomro MH, England-Mason G, Reardon AJF, Liu J, MacDonald AM, Kinniburgh DW, et al. Maternal exposure to bisphenols, phthalates, perfluoroalkyl acids, and trace elements and their associations with gestational diabetes mellitus in the APrON cohort. Reprod Toxicol. (2024) 127:108612. 10.1016/j.reprotox.2024.108612 [DOI] [PubMed] [Google Scholar]
- 43.Hua J, Barnett AL, Lin Y, Guan H, Sun Y, Williams GJ, et al. Association of gestational age at birth with subsequent neurodevelopment in early childhood: a national retrospective cohort study in China. Front Pediatr. (2022) 10:860192. 10.3389/fped.2022.860192 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Etchell A, Adhikari A, Weinberg LS, Choo AL, Garnett EO, Chow HM, et al. A systematic literature review of sex differences in childhood language and brain development. Neuropsychologia. (2018) 114:19–31. 10.1016/j.neuropsychologia.2018.04.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.England-Mason G, Anderson A, Bell RC, Subhan FB, Field CJ, Letourneau N, et al. Maternal pre-pregnancy bmi and gestational weight gain are associated with preschool children’s neuropsychological outcomes in the apron cohort. Children. (2023) 10(12):1849. 10.3390/children10121849 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Irvine N, England-Mason G, Field CJ, Letourneau N, Bell RC, Giesbrecht GF, et al. Associations between maternal folate status and choline intake during pregnancy and neurodevelopment at 3–4 years of age in the Alberta pregnancy outcomes and nutrition (apron) study. J Dev Orig Health Dis. (2023) 14(3):402–14. 10.1017/S2040174423000041 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Ser B Stat Methodol. (1995) 57(1):289–300. 10.1111/j.2517-6161.1995.tb02031.x [DOI] [Google Scholar]
- 48.Hayes AF, Little TD. Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach. 2nd ed. New York City: The Guilford Press; (2018). [Google Scholar]
- 49.Fritz MS, MacKinnon DP. Required sample size to detect the mediated effect. Psychol Sci. (2007) 18(3):233–9. 10.1111/j.1467-9280.2007.01882.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Mickey RM, Greenland S. The impact of confounder selection criteria on effect estimation. Am J Epidemiol. (1989) 129(1):125–37. 10.1093/oxfordjournals.aje.a115101 [DOI] [PubMed] [Google Scholar]
- 51.Greenland S, Daniel R, Pearce N. Outcome modelling strategies in epidemiology: traditional methods and basic alternatives. Int J Epidemiol. (2016) 45(2):565–75. 10.1093/ije/dyw040 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Laird E, Thurston S, Van Wijngaarden E, Shamlaye C, Myers G, Davidson P, et al. Maternal vitamin d status and the relationship with neonatal anthropometric and childhood neurodevelopmental outcomes: results from the Seychelles child development nutrition study. Nutrients. (2017) 9(11):1235. 10.3390/nu9111235 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Veena SR, Krishnaveni GV, Srinivasan K. Association between maternal vitamin D status during pregnancy and offspring cognitive function during childhood and adolescence. Asia Pac J Clin Nutr. (2017) 26(3):438–49. 10.6133/apjcn.032016.07 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54.Voltas N, Canals J, Hernández-Martínez C, Serrat N, Basora J, Arija V. Effect of vitamin d status during pregnancy on infant neurodevelopment: the eclipses study. Nutrients. (2020) 12(10):3196. 10.3390/nu12103196 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55.Jones G. Metabolism and biomarkers of vitamin D. Scand J Clin Lab Invest. (2012) 72(s243):7–13. 10.3109/00365513.2012.681892 [DOI] [PubMed] [Google Scholar]
- 56.Keim SA, Bodnar LM, Klebanoff MA. Maternal and cord blood 25(oh)-vitamin d concentrations in relation to child development and behaviour. Paediatric Perinatal Epid. (2014) 28(5):434–44. 10.1111/ppe.12135 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 57.Janbek J, Specht IO, Heitmann BL. Associations between vitamin D status in pregnancy and offspring neurodevelopment: a systematic literature review. Nutr Rev. (2019) 77(5):330–49. 10.1093/nutrit/nuy071 [DOI] [PubMed] [Google Scholar]
- 58.Gale CR, Robinson SM, Harvey NC, Javaid MK, Jiang B, Martyn CN, et al. Maternal vitamin D status during pregnancy and child outcomes. Eur J Clin Nutr. (2008) 62(1):68–77. 10.1038/sj.ejcn.1602680 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 59.Melough MM, Li M, Hamra G, Palmore M, Sauder KA, Dunlop AL, et al. Greater gestational vitamin d status is associated with reduced childhood behavioral problems in the environmental influences on child health outcomes program. J Nutr. (2023) 153(5):1502–11. 10.1016/j.tjnut.2023.03.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 60.Webb A, Kazantzidis A, Kift R, Farrar M, Wilkinson J, Rhodes L. Colour counts: sunlight and skin type as drivers of vitamin d deficiency at UK latitudes. Nutrients. (2018) 10(4):457. 10.3390/nu10040457 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 61.Polańska K, Jurewicz J, Hanke W. Smoking and alcohol drinking during pregnancy as the risk factors for poor child neurodevelopment—a review of epidemiological studies. Int J Occup Med Environ Health. (2015) 28(3):419–43. 10.13075/ijomeh.1896.00424 [DOI] [PubMed] [Google Scholar]
- 62.Wehby GL, Prater K, McCarthy AM, Castilla EE, Murray JC. The impact of maternal smoking during pregnancy on early child neurodevelopment. J Hum Cap. (2011) 5(2):207–54. 10.1086/660885 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data presented in this article are not available due to the ethic boards approvals and initial informed consents provided by participants. Therefore, we are unable to release any of the study data into a depository or other source. Requests for access to the dataset should be directed to the corresponding author.
