Maternal Vitamin D Supplementation during Pregnancy

Structured Abstract

Introduction

Maternal vitamin D status in pregnancy has been linked to many health outcomes in mother and offspring. A wealth of observational studies have reported on both obstetric outcomes and complications, including preeclampsia, gestational diabetes, mode and timing of delivery. Many fetal and childhood outcomes are also linked to vitamin D status, including measures of fetal size, body composition and skeletal mineralisation, in addition to later childhood outcomes, such as asthma.

Sources of Data

Synthesis of systematic and narrative reviews.

Areas of Agreement and Controversy

The findings are generally inconsistent in most areas, and, at present, there is a lack of data from high quality intervention studies to confirm a causal role for vitamin D in these outcomes. In most areas, the evidence tends towards maternal vitamin D being of overall benefit, but often does not reach statistical significance in meta-analyses.

Growing Points and Areas timely for developing research

The most conclusive evidence is in the role of maternal vitamin D supplementation in the prevention of neonatal hypocalcaemia; as a consequence the UK department of health recommends that pregnant women take 400 IU vitamin D daily. High-quality randomised placebo controlled trials, such as the UK-based MAVIDOS trial, will inform the potential efficacy and safety of vitamin D supplementation in pregnancy across a variety of outcomes.

Introduction

The classical role of vitamin D is in calcium and phosphate homeostasis, and it is widely known that that severe vitamin D deficiency (VDD) can result in rickets, osteomalacia and neonatal hypocalcaemia. Clinically, neonatal hypocalcaemia can result in seizures, and has been associated with softening and thinning of the skull (craniotabes) ⁴, and rarely, dilated cardiomyopathy ⁵. However, there is increasing evidence to suggest that VDD in pregnancy is associated with wide ranging clinical outcomes, including obstetric complications, preterm birth, and adverse offspring outcomes affecting the skeletal, immune, and respiratory systems⁶. As a result, a number of national and international guidelines recommend vitamin D supplementation during pregnancy or offer guidance on defining deficiency and sufficiency (Table 1), though there is considerable debate regarding the thresholds for vitamin D deficiency and sufficiency. Most recommend between 400 and 600 IU (10-15 micrograms) cholecalciferol daily throughout pregnancy, although this is not currently supported by the World Health Organisation (WHO)⁷. It is important to mention that the Endocrine Society recommends a safe upper limit for 25(OH)D intake in pregnancy of 10,000 IU/day for individuals over the age of 19 years at risk of vitamin D deficiency, though their recommendations for those not at risk remain at 600IU/day⁸.

Guideline	Countries covered by recommendation	Deficiency (nmol/l)	Insufficiency (nmol/l)	Sufficiency (nmol/l)	Dietary recommendation for vitamin D intake in pregnancy (IU)* - RI +
Scientific Advisory Committee on Nutrition (SACN)9 and UK Department for Health	UK	<25		≥25	400
Institute of Medicine (IOM) 10	USA and Canada	< 30	30-50	≥ 50	600
Endocrine Society Practice Guidelines 11	Worldwide	< 50	50-75	≥ 75	600
British Paediatric and Adolescent Bone Group 12	UK	< 25	25-50	≥ 50	Refers to SACN, 400
Global Consensus Recommendations on Prevention and Management of Nutritional Rickets 13	Worldwide	< 30	30-50	≥ 50	600
National Osteoporosis Society (UK) 14	UK	< 30	30-50	≥ 50	No recommendation
Canadian Paediatric Society 15	Canada	< 25	25-75	75-225	No recommendation
Working group of the Australian and New Zealand Bone and Mineral Society, Endocrine Society of Australia and Osteoporosis Australia 16	Australia and New Zealand	< 50		≥ 50 At the end of winter (level may need to be 10-20 nmol/l higher at the end of summer)	No recommendation
NORDEN Nordic Nutrition Recommendations 17	Nordic countries	<30		≥ 50	400
European Food Safety Authority 18	EU countries	< 50		≥ 50	600

* 400 IU cholecalciferol equivalent to 10 micrograms, ⁺RI – Recommended Intake in low-risk individuals

Here, we review the evidence basis for antenatal vitamin D supplementation to prevent obstetric complications, and the influence of vitamin D on fetal growth, with a focus on skeletal development.

Vitamin D in pregnancy: physiology and epidemiology

Vitamin D can be obtained through two pathways: via the diet or via endogenous synthesis. In the diet, it can be obtained as ergocalciferol (vitamin D₂) from plant sources, or cholecalciferol (vitamin D₃) from animal sources. However, the majority is formed endogenously within the skin from the action of ultraviolet B radiation to convert 7-dehydrocholesterol to pre-vitamin D₃. The main circulating form, 25-hydroxyvitamin D [25(OH)D], is produced through hydroxylation of pre-vitamin D in the liver. It is found either bound to vitamin D binding protein (VDP), albumin, or in the free form, and acts as a reservoir for conversion to the active metabolite, 1,25-dihydroxyvitamin D [1,25(OH)₂D]), primarily in the renal proximal tubular cells, but also within bone, the parathyroid gland and, in pregnancy, the placenta. The production of 1,25(OH)₂D is regulated in response to serum calcium and has a short half-life of 4-6 hours. Hepatic 25-hydroxylation, conversely, is not physiologically regulated and 25(OH)D has a half-life of approximately 2-3 weeks¹⁹. Serum 25(OH)D is consequently considered the best marker of vitamin D status²⁰.

The main role of circulating 1,25(OH)₂D is in calcium and phosphate homeostasis, which occurs in conjunction with parathyroid hormone (PTH). Low serum ionised Ca²⁺ stimulates PTH release, increasing renal calcium reabsorption in the distal tubule of the kidney, decreasing proximal tubule phosphate reabsorption, and increasing 1,25(OH)₂D synthesis. 1,25(OH)₂D then acts on the intestinal enterocytes to increase uptake of dietary calcium, simultaneously enabling PTH induced mobilisation of calcium and phosphate from bone mineral²¹.

During pregnancy, significant alterations to calcium and phosphate metabolism occur, allowing the accretion of calcium within the fetal skeleton, particularly during the third trimester²². It is quite remarkable that during pregnancy maternal serum ionised calcium concentrations remain stable, whilst intestinal calcium absorption increases ^23,24, and additional calcium is mobilised from the maternal skeleton²⁵. 1,25(OH)₂D is likely have an important role in these adaptations, as total 1,25(OH)₂D increases during the second and third trimesters^23,26, though this could be reflected by increases in vitamin D binding protein (DBP) in early pregnancy but not late pregnancy. Therefore, free 1,25(OH)₂D is increased in the third trimester relative to earlier in pregnancy^24,27,28. The increase in 1,25(OH)₂D appears not to be driven by PTH, which remains within the normal adult range throughout pregnancy²², though may be influenced by PTH-related protein (PTHrP), which is elevated in the maternal circulation from early pregnancy²⁶. It is also important to consider that the rise in free 1,25(OH)₂D from early pregnancy (when the calcium requirements of the developing fetus are minimal) may have another function, as serum levels of 1,25 (OH)₂D increase by approximately twofold during the first trimester ²⁹. Indeed there is evidence to suggest that 1,25(OH)₂D may have a role in immune tolerance in a pregnant woman, to prevent fetal rejection ³⁰.

The effect of pregnancy on 25(OH)D however is less well understood: some studies show a reduction through pregnancy²⁷, whilst others demonstrate no significant differences in 25(OH)D pre-pregnancy, within each trimester and during lactation²⁴. Maternal 25(OH)D in pregnancy is an important consideration as the fetus is entirely dependent on the mother for 25(OH)D. 25(OH)D readily crosses the placenta, partly through the megalin-cubulin endocytotic system³¹ and maternal and umbilical cord venous blood 25(OH)D are moderately-highly correlated (r=0.44 to 0.89), although umbilical cord concentrations are typically lower than that of maternal blood ^32–35.

Biochemically low levels of 25(OH)D are highly prevalent in pregnant women, albeit using various 25(OH)D thresholds for deficiency. Dark skin pigmentation and extensive skin covering (for example for religious or cultural reasons) are the strongest risk factors for vitamin D deficiency, though many Caucasian women are also deficient, particularly at higher latitudes. In a cohort of predominantly Caucasian women in the United Kingdom (UK), 31% had a serum 25(OH)D less than 50nmol/l, and 18% less than 25nmol/l³⁶; whilst in an ethnically more diverse UK population, 36% of women had a 25(OH)D <25nmol/l at pregnancy booking³⁷.

There is some evidence to suggest that the cutoffs of <25nmol/L as deficient and >50nmol/L as sufficient may not be relevant to pregnancy. Uniquely in pregnancy, the conversion of 25(OH)D to 1,25(OH)₂D may be optimised at a 25(OH)D concentration of around 100nmol/L³⁸. In this study, pregnant women were supplemented with 400IU/d, 2000IU/d or 4000 IU/d, the Institute of Medicine³⁹ recommended upper limit of intake in pregnancy, and was found to be effective in achieving vitamin D sufficiency throughout pregnancy without increased risk of toxicity in women or neonates.

Comparison of outcomes between trials is further complicated by the vast range of dosages of vitamin D given (200IU daily to 4400IU daily, or bolus doses ranging from single doses of 60,000IU to 35,000 IU per week^{3 40 41}).

Randomised controlled trials have clearly demonstrated that vitamin D supplementation in pregnancy can increase umbilical cord venous and neonatal serum 25(OH)D compared to placebo, however the amount by which maternal serum 25(OH)D increases with supplementation varies between individuals^38,42–48. In the MAVIDOS trial of 1000 IU cholecalciferol from 14 weeks gestation, factors significantly associated with the response to supplementation (maternal 25(OH)D at 34 weeks gestation) were season of delivery, baseline 25(OH)D at randomisation, and compliance with the medication (Figure 1)⁴⁹. Weight gain was negatively associated with 25(OH)D at 34 weeks, which is in agreement with previous observational findings suggesting that greater weight gain in pregnancy is associated with a reduction in 25(OH)D between early and late pregnancy⁵⁰, implying that those women who gain most weight during pregnancy may benefit from higher supplement doses. Interestingly BMI at baseline was not an independent determinant of late pregnancy 25(OH)D in either the treatment or placebo group, despite obesity being associated with biochemically low 25(OH)D levels⁵¹.

Figure 1.

Factors affecting response to 25(OH)D supplementation (1000 IU daily from 14 to 34 weeks) in the MAVIDOS trial. Adapted with permission from Moon et al ⁴⁹.

Consistent with findings in non-pregnant adults⁵², genetic factors play a role in the response to vitamin D supplementation during pregnancy. Single nucleotide polymorphisms (SNPs) in or near to genes in the vitamin D metabolism pathway appear to modify the response to vitamin D supplementation in pregnancy. Different SNPs were found to be associated with baseline 25(OH)D (rs12785878 in DHCR7, encoding 7-dehydrocholesterol reductase in the skin), compared with 25(OH)D achieved post supplementation (SNPs in genes encoding 25-hydroxylase and vitamin D binding protein) (Figure 2). Women with more “at risk” alleles may therefore require higher supplement doses to achieve vitamin D repletion in pregnancy, which may be of interest in the future era of personalised medicine².

Figure 2.

Associations between SNPs and change in 25(OH)D from 14 to 34 weeks of gestation following supplementation with 1000 IU/d cholecalciferol. Shown as β and 95% CI. The homozygous low-frequency allele was used as the reference group, with the β representing the change in 25(OH)D (nmol/L) per common allele (additive model). Models were adjusted for age, physical activity, smoking status, educational attainment, season of blood sampling, compliance, and pregnancy weight gain. Adapted with permission from Moon et al, 2017 ²

Vitamin D and obstetric complications

Numerous observational studies report associations between 25(OH)D status in pregnancy and a variety of obstetric complications, including gestational hypertension (GHT) and preeclampsia (PET), gestational diabetes (GDM), and timing and mode of delivery. However, the interpretation and comparison of these studies is limited by differences in the study design (e.g. prospective cohort, case-control), the timing of 25(OH)D measurements, ranging from first trimester to delivery, the definition used for both VDD and the pregnancy or delivery outcome, and the confounding factors adjusted for.

Gestational hypertension & preeclampsia

Previous studies have suggested that maternal calcium status might be important in the aetiology of PET, as calcium supplementation can reduce PET risk particularly in women with low calcium intake⁵³. This led to the interest in the role of calcitropic hormones, including vitamin D, in the development of PET. In addition, studies in humans have demonstrated an association between vitamin D insufficiency and endothelial dysfunction, implying that vitamin D has a role in regulating the conductance and resistance of blood vessels ⁵⁴. This provides another mechanism by which vitamin D insufficiency could be associated with PET; indeed a substudy of a RCT of vitamin D supplementation in pregnancy demonstrated that supplementation influenced mRNA level gene transcription of angiogenic factors (such as vascular endothelial growth factor, VEGF), in the human placenta ⁵⁵.

The outcomes of many observational studies reporting on GHT and pre-eclampsia are conflicting, with some demonstrating an inverse association between maternal vitamin D status and the risk of pre-eclampsia^56–64 and others no association ^58,65–72. In recent years, there have been several published meta-analyses of the relationship between maternal vitamin D status and PET risk, providing a variety of inconsistent results^6,40,73–78. The most recent systematic review and meta-analysis, published in the BMJ in 2017, showed that vitamin D had no significant influence on the risk of GHT or pre-eclampsia⁴¹. In the UK Health Technology Assessment meta-analysis (Harvey et al), again, no significant reduction in the risk of PET with higher vitamin D status was noted (Figure 3)⁶. In contrast, Aghajafari et al reported on 9 studies in another meta-analysis of vitamin D deficiency and pregnancy outcomes, and showed a significant association between pre-eclampsia and 25(OH)D insufficiency compared with the comparison group, with a pooled odds ratio of 1.79 (95% CI 1.25, 2.58)⁷⁵. However, this appeared subject to a particular analytical approach as the increased risk of PET in VDD was only observed in studies in which blood sampling was later than 16 weeks gestation and when VDD was defined as 25(OH)D<75nmol/l and not <50nmol/l⁷⁵. Tabesh et al., analysing a larger number of studies defining VDD as less than 50nmol/l, also demonstrated an increased risk of PET, which was not found when deficiency was defined as less than 38nmol/l⁷⁶. A recently updated Cochrane review of vitamin D supplementation and maternal and infant health outcomes by De-Regil et al included 15 trials, nine comparing the effects of vitamin D supplementation versus no supplementation or a placebo, and six comparing the effects of vitamin D in combination with calcium and no supplementation. Only two trials (n=219, combined) ^79,80 were included in the pre-eclampsia analysis, which suggested overall that women who received vitamin D supplements may have a lower risk of PET (though borderline), than those receiving no intervention or placebo (Risk Ratio 0.52 (95% CI 0.25, 1.05)) ⁴⁰. However, a more recently published trial, VDAART, showed no significant differences in the incidence of pre-eclampsia (as a secondary outcome) in a group of pregnant women supplemented with cholecalciferol (n=440, given 4400 IU 25(OH)D daily, and the placebo group, receiving 400 IU daily n=436) ⁸¹.

Figure 3.

Forest plot of the association between maternal vitamin D status and risk of pre-eclampsia (observational studies (adjusted), ES, effect size). Reproduced with permission from Harvey et al; Health Technology Assessment 2014 ³

Gestational Diabetes

As in the case of PET, conflicting findings have been reported for 25(OH)D status in case-control and prospective cohort studies of GDM risk, with some reporting lower^78,82–87, and others no difference in 25(OH)D^88,89 during pregnancy in women with GDM. Evidence suggests that vitamin D contributes to various factors associated with the propensity to develop impaired glucose tolerance, such as insulin sensitivity, β-cell function, and insulin secretion. Immune dysfunction, through inflammatory dysregulation (cytotoxic T-lymphocyte activation), links vitamin D deficiency with compromised maternal-fetal tolerance and increased risk of GDM⁹⁰.

Three separate meta-analyses of published studies all concluded that women with GDM had significantly lower mean 25(OH)D than normoglycaemic women^75,77,91 with the mean difference in 25(OH)D ranging from 3.9 to 7.4nmol/l. Furthermore, these meta-analyses suggested that the risk of GDM was increased by 40-60% in women with VDD ^75,77,91. However, since their publication, a meta-analysis by Harvey et al including eight observational studies ^{68,82,83,88,92–95} showed no overall relationship between maternal 25(OH)D status and risk of GDM. Similarly, in a Cochrane meta–analysis, two trials were included ^79,80 comparing vitamin D supplementation during pregnancy with no treatment/placebo, and no significant difference in the risk of gestational diabetes was seen between the two groups⁴⁰. In the most recent meta-analysis by Roth et al, vitamin D was shown to reduce the risk of gestational diabetes (risk ratio 0.61, 95%CI 0.34 to 0.83, 2643 participants) when all trials were included, but when only the trials meeting their strict eligibility criteria were included, no effect was seen ⁴¹.

Caesarean Delivery

There is increasing interest in the role of maternal vitamin D status in relation to mode and timing of delivery, as it has been suggested that vitamin D deficiency may reduce pelvic muscle strength and control ⁹⁶. However, again, findings are inconsistent. Several studies which assessed 25(OH)D in early pregnancy (at the time of GDM screening) or at delivery, reported an increased risk of Caesarean delivery in 25(OH)D deficient women ^96–98, whilst other studies measuring 25(OH)D in the first trimester demonstrated no increased risk^68,70. In trials of high dose 25(OH)D supplementation of 4000 IU daily versus 400 IU daily, primary Caesarean section births (with or without labour, for a maternal or fetal indication) were found to be significantly reduced in the high dose group ³⁸. Harvey et al (describing six observational studies), De Regil (including two trials), and Roth et al (including 16 trials) found no overall evidence in meta analyses to support the use of vitamin D supplementation to reduce rates of Caesarean section ^{6 40 41}.

Preterm Delivery and Newborn Health

Several investigators have reported on associations between low maternal vitamin D status and preterm births^74,98–100, with some presenting the hypothesis that systemic inflammation, which is associated with a low 25(OH)D status, may accompany the events leading to the initiation of preterm birth, including membrane rupture and uterine contractions ^101,102. It is therefore argued that vitamin D status in late pregnancy is a better predictor of preterm birth than early pregnancy vitamin D status ¹⁰³. Notwithstanding this, many studies have concluded that maternal 25(OH)D status is not beneficially related to preterm birth^{65,68,70,78,104–109}. Comparisons between these studies are complicated by the differing definitions of preterm, 25(OH)D deficiency, timing of measurement, and ethnicities of populations between studies. For example, Bodnar et al observed that only non-white mothers had an increased risk of preterm birth with low 25(OH)D at 26 weeks gestation¹⁰⁰, suggesting stratification of women by ethnicity in future intervention studies might be necessary.

In a meta-analysis of three trials ^43,79,80, De Regil et al found that women who received vitamin D supplements during pregnancy had a lower risk of having a preterm birth than those women receiving no intervention or placebo (3.3% versus 9.9%; average RR 0.36; 95% CI 0.14 to 0.93). The meta-analysis of 3757 participants from 13 trials by Roth et al, on the other hand, showed no effect of vitamin D supplementation on risk of preterm birth ⁴¹.

Health of the newborn is often of interest in trials of vitamin D supplementation in pregnancy. In one study of 4000 IU cholecalciferol versus placebo in pregnant Pakistani women, vitamin D supplementation had no effect on maternal outcomes or birthweight or size, but did have a beneficial effect on markers of newborn health, the Apgar score, at one and 5 minutes post-delivery (for example, mean Apgar at one minute in the supplemented group 7.10±0.66, versus placebo group 6.90±0.50, p=0.026) ¹¹⁰. In another randomised trial based in India, where women in the intervention group were given 25(OH)D supplementation according to baseline vitamin D status, a positive correlation was found between maternal vitamin D status and the Apgar Score of babies (r=0.325, p<0.001) ⁷⁹. However, the most recent meta-analysis of 1997 individuals from 5 trials showed that there was no reduction in risk of admission to the neonatal intensive care unit in the offspring of mothers receiving vitamin D supplementation ⁴¹.

In all obstetric outcomes, it must be acknowledged that 25(OH)D status is primarily determined by environmental factors. Therefore, confounding and reverse causality need to be considered; in addition, differences in the covariates included in multivariate models might explain the inconsistent findings. Examples include the observation that obese individuals have lower 25(OH)D status, and a higher incidence of GDM, GHT, PET, caesarean section and preterm delivery^111,112. Similarly, African-American women are more likely to require delivery by Caesarean section and to experience pre-eclampsia and preterm labour ¹¹³. Therefore, whether these relatively rare obstetric outcomes can truly be attributed to lower 25(OH)D and therefore could be prevented by vitamin D supplementation must be established through carefully designed intervention studies with large numbers of participants.

Neonatal Hypocalcaemia

Symptomatic neonatal hypocalcaemia is one of the fetal outcomes most reliably associated with maternal VDD. It is rarely reported in infants of white mothers, and most commonly occurs in infants of mothers with the most profound VDD, often from ethnic and cultural groups at highest risk.

Several intervention studies assessing the effect of vitamin D supplementation on umbilical cord and neonatal serum Ca²⁺ have been reported, though with inconsistent findings. Such differences might reflect the variations in supplementation used; the majority of studies that used high dose weekly or monthly oral supplementation reported positive effects of vitamin D supplementation on umbilical venous cord blood Ca^{2+ 44,114–116}, whereas the trials using daily oral vitamin D supplementation tended to demonstrate a null effect ^{41,44,48,115,117–119}. Nonetheless, vitamin D supplementation consistently reduced the incidence of symptomatic hypocalcaemia in the three studies in which this outcome was reported ^116,117,119. Hence, though the evidence for measurable biochemical changes to calcium homeostasis with antenatal vitamin D supplementation are variable, the clinical outcome of symptomatic hypocalcaemia, which is perhaps more important, is more consistent and justifies the routine use of antenatal vitamin D supplementation.

Vitamin D and Fetal Development

The fetus is dependent on the mother for accretion of approximately 30g of calcium to enable skeletal development, and rickets has been reported in infants born to mothers with VDD ^120–122. As a result, a subclinical role for vitamin D and/or calcium in fetal growth and bone development has been considered. Interestingly, maternal supplementation with calcium alone does not appear to have beneficial effects on fetal bone mineral accrual¹²³; indeed a recent study has suggested detrimental effects of maternal calcium supplementation on childhood growth in females in a Gambian population particularly accustomed to low calcium intake ¹²⁴.

A hypothesis is developing that vitamin D may have a role in the programming of fetal development – that is, the prevailing in-utero environment modifies later phenotype, a concept termed developmental plasticity ¹²⁵. This hypothesis has been applied to associations between maternal vitamin D status and offspring development in various contexts, with studies and trials in various populations showing positive associations between maternal 25(OH)D and birthweight, length and/or occipitofrontal head circumference ^{126,33,127,128,129,130} and sometimes postnatal growth ^{131 126}. However, this is balanced by a greater number of studies with null findings ^{32,83,88,132–139}. Evidence generally supports only the lowest levels of 25(OH)D being associated with poorer fetal growth outcomes ^{140, 127}. A 2017 meta-analysis of 27 different neonatal outcomes demonstrated vitamin D supplementation lead to an increase in birth weight and length at 1 year, a reduction in the risk of infants being small for gestational age, Risk Ratio 0.60 (95% CI 0.40 to 0.90), and a reduction in the risk of asthma or wheeze by age 3 years, Risk Ratio 0.81 (95% CI 0.67 to 0.98) ⁴¹.

Mixed evidence is available for the influence of vitamin D status in pregnancy and other fetal and childhood outcomes, including adiposity^{96,134 141} lean mass, and childhood asthma⁸¹, however the most detailed evidence is in the realm of skeletal development, which will be the focus for the latter part of this review.

Bone development

Observational studies

Some of the earliest data to suggest that vitamin D exposure might influence in utero bone mineral accrual used season as a proxy marker of vitamin D status. Namgung et al observed that in 71 Korean neonates, those born in summer months had 8% higher whole body bone mineral content (BMC), measured by dual energy x-ray absorptiometry (DXA), after adjustment for weight, than infants born in winter. In addition, in that cohort, neonatal 25(OH)D at delivery was positively correlated with whole body BMC ¹⁴². In contrast, the same authors found that infants in the USA who were born in the summer had lower whole body BMC than winter-born infants ¹⁴³. The authors proposed that these differences reflect the use of vitamin D supplementation in the two populations; the uptake of supplementation is low throughout pregnancy in Korea but standard practice after the first trimester in the USA, where differences in maternal 25(OH)D by season of birth were not observed ¹⁴². This would suggest that early pregnancy 25(OH)D status is crucial to vitamin D mineralisation, which is in contrast to several later studies with assessment of serum 25(OH)D. Data from the Southampton Women’s Survey mother-offspring cohort, in the UK, suggest higher neonatal BMC amongst summer than winter births (Figure 4).

Figure 4.

Neonatal whole-body bone mineral content (BMC) of neonates born in the summer (June-November) or winter (December-May) in the UK Southampton Women’s Survey. Data are mean and 95% CI.

Subsequent studies have used measurements of maternal or cord blood 25(OH)D as the exposure variable. In a Canadian study, 25(OH)D was measured in venous cord blood and used to divide the infants into two groups using a cut point of 37.5 nmol/l. Whole body and femur BMC relative to body weight were significantly lower in the 18 neonates with the lower cord blood 25(OH)D, compared with 32 infants with a 25(OH)D above this cut point ¹⁴⁴. Similarly, Viljakainen et al, using the mean of two maternal serum 25(OH)D measurements from early pregnancy and 2 days postpartum as the assessment of maternal vitamin D status, found neonatal tibial BMC and cross-sectional area measured by peripheral quantitative computed tomography (pQCT) were 14% and 16% higher, respectively, in infants born to mothers with 25(OH)D above the median for the cohort ¹⁴⁵. When a subset of these children were reassessed at 14 months of age, the difference in tibial BMC was no longer present, but tibial CSA remained significantly higher in those born to mothers with higher vitamin D status in pregnancy ¹⁴⁶. Conversely, when 125 Gambian mother–offspring pairs were studied, no significant relationships were observed between maternal 25(OH)D at either 20 or 36 weeks of gestation and offspring whole body BMC or bone area at 2, 13 or 52 weeks of age ¹³⁶. However, in contrast to the other studies, none of the mothers had a 25(OH)D below 50 nmol/l, consistent with the hypothesis that poorer skeletal mineralisation might only occur in fetuses of mothers with the lowest 25(OH)D levels.

There is some evidence to suggest that these relationships persist into childhood, although study findings are less consistent than in the neonatal period. Positive relationships between maternal 25(OH)D measured in late pregnancy and offspring whole body and lumbar spine (LS) bone area, BMC and aBMD at 9 years of age in 198 mother-offspring pairs in the Princess Anne Hospital Study (Southampton, UK) were reported by Javaid et al ¹³⁹. It was also noted that children born to women who consumed vitamin D containing supplements had higher whole body BMC and bone area, but not areal BMD, supporting the evidence for maternal vitamin D supplementation – this finding persisted after adjustment for socioeconomic status. These findings were replicated in the Southampton Women’s Survey (SWS), in which 1030 mother-offspring pairs had measurement of 25(OH)D at 34 weeks of gestation and whole body less head (WBLH) and LS DXA at 6-7 years. WBLH BMC, bone area, BMD and LS BMC were all significantly lower in children born to mothers with 25(OH)D < 25 nmol/l in late pregnancy, including after adjustment for maternal age, ethnicity, height, pre-pregnancy BMI, smoking in late pregnancy, social class, maternal educational attainment and duration of breast feeding ¹⁴⁷.

Similarly, Zhu et al found a positive relationship between maternal 25(OH)D status at 18 weeks gestation and bone mass in young adulthood in the Raine cohort in Western Australia. They showed, after adjustment for sex, age, height and body composition at age 20 years, maternal height and pre-pregnancy weight, maternal age at delivery, parity, education, ethnicity, smoking during pregnancy and season of maternal blood sampling, whole body BMC and aBMD were 2.7% and 1.7% lower at 20 years of age in offspring of mothers with 25(OH)D < 50 nmol/l compared to those above this level ¹⁴⁸.

In contrast, analyses using the UK Avon Longitudinal Study of Parents and Children (ALSPAC) cohort study do not support these findings. Initially, using 6995 mother-offspring pairs, Sayer et al reported a positive relationship between estimated maternal UVB exposure in late pregnancy and offspring WBLH BMC, bone area and BMD at 9.9 years of age ¹⁴⁹. However, further analysis in a subset of 3960 of the children for whom maternal serum 25(OH)D measurement was available in the first (n=1035), second (n=879) or third (n=2046) trimester did not reveal any significant associations between maternal 25(OH)D and offspring bone mineralisation ¹⁵⁰, though collinearity of the estimated maternal UVB measurement with age at assessment might have confounded the relationships reported in the initial study. More recently, the Netherlands Generation R cohort documented inconsistent relationships between pregnancy 25(OH)D and offspring bone measures¹⁵¹. Unfortunately the analytical models used make it difficult to conclusively compare these analyses with previous studies¹⁵²; however, these conflicting findings from 2 very large cohorts demonstrate the limitations of the observational approach, and the clear necessity for testing the hypothesis that maternal pregnancy supplementation with vitamin D will lead to improved offspring bone mass in randomised controlled trial settings.

Intervention studies

In 1983, the first intervention study to assess the effect of antenatal vitamin D supplementation on offspring bone mineralisation was published. Sixty-four women of Asian ethnicity living in the UK participated in a non-randomised study; 19 received a daily supplement containing 1000 IU vitamin D and calcium (of unknown strength) during the last trimester, whereas 45 received no supplementation. There was no significant difference in forearm BMC of the offspring at birth assessed using single photon absorptiometry ¹⁵³; however the study size, lack of randomisation and technology used limits the interpretation of the findings.

Three more recently published studies of gestational vitamin D supplementation report on postnatal bone outcomes, of which the largest is the Maternal Vitamin D Osteoporosis Study (MAVIDOS). MAVIDOS is a randomised double-blind placebo-controlled trial of antenatal vitamin D supplementation from 14 weeks of gestation until delivery conducted in three centres in the UK, reporting a primary outcome of neonatal bone mass ¹⁵⁴. 1134 women with a baseline 25(OH)D between 25 and 100 nmol/l were randomised to 1000 IU/day cholecalciferol or placebo; 965 remained in the study until delivery, and 736 infants underwent DXA of the whole body and/or LS. Although there were no differences in whole body or LS BMC, bone area or aBMD between the two groups overall, a significant interaction was observed between season of birth and maternal randomisation group, as shown in (p for interaction for BMC 0.04) ¹⁵⁵. Thus, whole body BMC and BMD were approximately 9% and 5% higher, respectively, in the children born in winter to mothers randomised to cholecalciferol compared to those randomised to placebo. This effect size is substantially larger than those observed between children with and without fractures ¹⁵⁶, and hence if persisting into later childhood is likely to be clinically relevant, particularly if it has an impact on peak bone mass, as peak bone mass achieved is a strong predictor of osteoporosis in later life ¹⁵⁷.

Two smaller intervention studies from India and Iran have also assessed bone mass in infants born to mothers randomised to vitamin D supplementation or placebo. Sahoo et al randomised 300 women to three groups, which received 400 IU/day cholecalciferol daily (“placebo”), 60 000 IU cholecalciferol every 4 weeks or 60 000 IU cholecalciferol every 8 weeks from the second trimester. All women also received daily calcium supplementation, though relatively small numbers of women and children (17% of the original cohort) underwent DXA at 12-16 months of age. The children in the placebo group were significantly older at DXA scan and had higher measurements of whole body BMC and BMD, but in multivariate analysis randomisation group was not a significant predictor of BMC or BMD ¹⁵⁸. Vaziri et al randomised 153 women to placebo or 2000 IU/day cholecalciferol from 26-28 weeks until delivery, but only 25 infants (16% of the original cohort) had DXA assessment. No significant difference in whole body BMC, BMD or bone area was found ¹⁵⁹, but as with the study by Sahoo et al, the small numbers included are unlikely to have sufficient power to detect a difference in the outcomes studied.

Evidence from observational studies does therefore suggest that achieving higher levels of serum 25(OH)D in pregnancy might have beneficial effects on offspring bone development, but further high quality RCTs are required to assess this. Moreover, long-term follow up of children born to participants of these trials is required. This will enable us to determine whether any effects observed in the neonatal period, such as the effect of gestational vitamin D supplementation on increased bone mineralisation in children born in winter, does persist beyond the neonatal period and whether this can influence peak bone mass obtained.

Maternal pregnancy vitamin D status, epigenetics and developmental plasticity

An individual’s skeletal health, most often described by measurement of BMD and BMC by DXA, is only partly explained by genetic factors^{160 161 162}, and there is increasing evidence that some of the residual variance in both BMD and fracture risk in adulthood might be explained by the influence of the environment on gene expression, both in utero and in early life ¹⁶³. Evidence is accumulating that this concept of developmental plasticity may apply in the case of the actions of vitamin D in utero on the developing fetus directly, or indirectly. It is widely recognised that genes effectively provide a library of information that can be read (expressed) differently in different cells and tissues according to function and need. Thus, in a single organism, although the genetic code contained in every somatic cell is the same, the genes expressed will vary widely from organ to organ and even from cell to cell, often in response to environmental cues ¹²⁵ – so a single genotype can give rise to many different phenotypes. Various experimental studies have clearly demonstrated that alterations to maternal dietary factors during pregnancy may lead to changes in offspring phenotype and gene expression ^{164 165}. These effects are likely to be underpinned by epigenetic mechanisms, processes by which gene expression is modified but without changes in the DNA code itself. Such epigenetic signals are essential in determining when and where genes are expressed. An understanding of these epigenetic processes has the potential to enable early intervention strategies to improve early development and later health; consequently the study of epigenetic biomarkers is a rapidly advancing field ¹⁶⁶. Epigenetic mechanisms include DNA methylation, histone modification and non-coding RNAs, the most widely studies of which is DNA methylation, the transfer of a methyl group to a particular genomic location, usually at the 5’ carbon position of cytosine adjacent to a guanine base, or CpG site. A CpG site can either be methylated or unmethylated in an individual cell, however, across a whole tissue where genes in cells may be methylated or unmethylated, a range of graded gene expression from 0% to 100% is possible ¹²⁵. Methylation at regions of the genome particularly rich in CpG sites, for example, at the 5’ end of genes in regions known as CpG islands, often near to the promoter of a gene, may have important influences on that gene’s expression^{167, 125,168}.

In terms of the links between maternal 25(OH)D status and offspring bone mass, there is evidence that this association may be mediated partly through umbilical cord calcium concentrations, and expression of a particular active placental calcium transporter was positively associated with neonatal bone mass¹⁶⁹, and with 1,25(OH)₂-vitamin D implicated in experimental studies¹⁷⁰. There is increasing evidence that epigenetic mechanisms such as DNA methylation may underlie such observations^125,171. In the SWS, perinatal DNA methylation at two loci of interest, CDKN2A, a key element in cell senescence¹⁷² and the retinoid-X-receptor-A(RXRA) gene were inversely associated with childhood BMC corrected for body size at four years ^{173 1}. RXRA forms a heterodimer with the vitamin D receptor and is essential in the nuclear action of 1,25(OH)₂-vitamin D. Methylation at one CpG site was related to an estimate of free 25(OH)D (figure 6). Further analyses are ongoing to establish the influence of maternal 25(OH)D on methylation of RXRA in a randomised controlled trial setting, MAVIDOS. In order to fully exploit these findings, it is essential to establish whether these associations persist into older childhood, to refine the bone phenotype, and to experimentally validate whether gestational supplementation with vitamin D might alter DNA methylation.

Figure 6.

Percent DNA methylation at RXRA & offspring size corrected bone mineral content (a); maternal 25(OH)D & RXRA DNA methylation (b). scBMC, size-corrected Bone Mineral Content. Adapted with permission from Harvey et al, 2014¹

In this section, we have mainly focussed on epigenetic mechanisms associating vitamin D status in pregnancy with offspring bone health. Research is ongoing into the links between maternal vitamin D and its influence on multiple developing body systems within the fetus, for example immunity and allergy, as investigated in the VDAART trial¹⁷⁴, and in maternal health, such as vascular pregnancy complications ⁵⁵.

Conclusions and knowledge gaps

Our understanding of the role of vitamin D during pregnancy and its role in maternal and offspring health has improved considerably over the last decade, though there is still a great deal of work to do. The UK Department of Health currently recommends routine antenatal vitamin D supplementation with 400 IU cholecalciferol daily throughout pregnancy for all women, independent of ethnicity and other risk factors for VDD ¹⁷⁵. There is good evidence to suggest that supplementation is important in reducing the risk of neonatal hypocalcaemia and increasing neonatal 25(OH)D status. There is evidence to suggest that certain maternal characteristics, including genetic factors, influence an individual’s response to supplementation, though currently these are not usually taken into account when advising on whether a higher level supplementation is necessary for particular individuals. Further research into this and into the way in which vitamin D status tracks through pregnancy, would aid in dosing recommendations. Avoidance of vitamin D toxicity by over-supplementation, which could cause harm, is also important – the Institute of Medicine suggests that toxic effects can occur above the threshold of 125nmol/L, and recommend avoidance of intakes above 4000IU/day.

In terms of obstetric outcomes, there is mixed and conflicting evidence on the benefits of vitamin D supplementation in pregnancy. As 25(OH)D status is primarily determined by environmental factors, confounding and reverse causality need to be considered when evaluating obstetric studies; in addition, differences in covariates included in multivariate models might explain the inconsistent findings. Whether these relatively rare obstetric outcomes can truly be attributed to lower 25(OH)D must be established through carefully designed intervention studies.

There is some evidence to support a positive relationship between vitamin D status and offspring birthweight and offspring bone mass, but conflicting evidence still exists. Intervention studies point towards a role for vitamin D in these offspring health outcomes, particularly in women who are deficient or borderline deficient in vitamin D. Research is ongoing into the mechanism by which vitamin D may be important in developmental programming of later diseases, such as osteoporosis. The programming hypothesis may present a potential approach to targeting and reducing the burden of diseases with a developmental origin across a population, rather than an individual level. Double-blinded RCTs with a lengthy follow-up periods, such as MAVIDOS, are required to further establish the associations between vitamin D status in pregnancy and offspring health. To guide future policy, research should evaluate several important factors. These include whether serum 25(OH)D levels improve maternal and offspring outcomes in populations with different diets or skin pigmentation, establish the timing of initiation of therapy and its most effective and safe dosage, and inform whether critical periods in pregnancy exist in the role of vitamin D and its impact on the future health of both mother and offspring. Experimental techniques, including epigenetic analyses, will be important in clarifying these underlying mechanisms.

Figure 5.

Neonatal whole body bone mineral content (BMC), bone area and bone mineral density (BMD) by intervention group and season of birth in the MAVIDOS trial. Data are shown as mean and 95% confidence interval. Winter is December to February, spring is March to May, summer is June to August and autumn is September to November.

Reproduced with permission from Cooper et al, 2016 ² and in accordance with a Creative Commons Licence; https://creativecommons.org/licenses/by/4.0/; https://doi.org/10.1016/S2213-8587(16)00044-9