|Home | About | Journals | Submit | Contact Us | Français|
Relationships between vitamin D exposure of the mother in pregnancy and skeletal development of the child are poorly understood.
(i) To establish whether background UVB levels in the third trimester of pregnancy are related to bone mineral content (BMC) of the child. (ii) To examine whether these relationships are explained by effects on height, fat or lean mass.
Prospective cohort study.
The Avon Longitudinal Study of Parents and Children (ALSPAC), a population-based birth cohort.
6955 boys and girls mean age 9.9 years.
Pre-specified analyses of relationships between background UVB levels in the third trimester of pregnancy, and total body less head BMC, bone area (BA), bone mineral density (BMD) and area-adjusted BMC (ABMC) as measured by DXA scans at 9.9 years.
Maternal UVB exposure was positively related to BMC (g) [9.6 (5.3, 13.8)], BA (cm2) [8.1 (4.3, 11.9)] and BMD (g.cm−2) [0.003 (0.001, 0.004)], but not ABMC (g) [0.69 (−0.22, 1.56)], suggesting an effect on bone size. Both height-dependent (cm) [0.18 (0.03, 0.32)] and height-independent (cm2) [4.1, (2.0, 6.2)] effects contributed to this association between UVB and BA. Although maternal UVB exposure was also related to lean mass (g) [163 (89, 237)], a positive association between UVB and BA persisted after adjusting for both height and lean mass [2.8 (1.0, 4.6)].
Maternal UVB exposure is related to bone size at age 9.9 independently of height and lean mass, suggesting vitamin D status in pregnancy exerts direct effects on periosteal bone formation in subsequent childhood.
Cohort studies from the United Kingdom, North America and Australia suggest that poor growth during fetal life and infancy is associated with decreased bone mass in adulthood (1-5). In a Finnish cohort, reduced trajectory of linear growth during intrauterine and early postnatal life was directly linked with an increased risk of hip fracture six to seven decades later (6). These associations may be explained by an adverse influence of factors such as maternal nutritional status on the in utero environment, which are postulated to affect both early skeletal development, and the trajectory of bone mass acquisition in subsequent childhood (7). To explore the role of early life exposures in the programming of bone development, several studies have sought to determine whether specific environmental exposures in utero are related to subsequent bone development. For example, maternal levels of 25 hydroxyvitamin D3 were found to be related to bone mineral content (BMC) in nine year old children in Southampton, UK (8). The suggestion that in utero exposure to vitamin D mediates programming effects on bone is consistent with evidence that vitamin D exposure in pregnancy is related to other aspects of the child's health, such as the risk of type I diabetes (9) and asthma (10).
Further evidence that vitamin D exposure of the mother in pregnancy exerts long-term effects on subsequent bone development of the child is provided by other studies in which rather than being assessed directly, vitamin D exposure was inferred from sunlight exposure, on the basis that UVB plays a major role in vitamin D3 synthesis, and is an important determinant of vitamin D nutritional status. For example, month of birth was reported to be related to bone mineral density (BMD) in adults (11), and maternal veiling was found to have an adverse outcome on bone mass in boys but not girls (12), providing further support for a role of maternal vitamin D status in programming skeletal development in childhood.
Since bone mass largely reflects bone size which is in turn closely related to height, the findings described may simply represent an association between maternal vitamin D exposure and longitudinal growth, as opposed to an influence on cellular aspects of bone cell function involved in bone modelling or remodelling. For example, several studies have documented relationships between season of birth and longitudinal growth in utero (13), in childhood (14-16), and in adulthood (17,18). Alternatively, since bone mass is known to be strongly related to body composition, any tendency for maternal vitamin D exposure to affect later bone mass accrual of their offspring might be an indirect consequent of effects on lean or fat mass. Consistent with the latter possibility, previous studies have reported associations between month of birth and subsequent risk of obesity (16,19).
In the present study, we used the Avon Longitudinal Study of Parents and Children (ALSPAC) to explore the relationship between maternal vitamin D exposure during pregnancy and subsequent bone development of the child. Due to the large number of participants (approximately 7000), which provides sufficient power to provide accurate estimates of effects of potential confounders such as height and fat mass, this gives an opportunity to examine whether vitamin D exposure in utero directly influences cellular processes which contribute to bone modelling or remodelling. To provide an estimate of maternal vitamin D status, we used UVB exposure over the last trimester of pregnancy as a proxy measure, derived from historical data collected at the Meteorological Office (UK) and the National Radiation Protection Board (UK), which we have validated against maternal vitamin D levels in a sub-group of the sample.
The ALSPAC study is a geographically based birth cohort study investigating factors influencing the health, growth, and development of children. All pregnant women resident within a defined part of the former county of Avon in South West England with an expected date of delivery between April 1991 and December 1992 were eligible for recruitment, of whom ~14,000 were enrolled (20) (http://www.alspac.bristol.ac.uk). ALSPAC is a relatively homogenous population in ethnic terms, approximately 97% of subjects being caucasian. 6995 children were available for analyses relating UVB exposure in pregnancy to results from DXA scans performed at age 9.9 years, which formed the basis for the current study. Ethical approval was obtained from the ALSPAC Law and Ethics committee and relevant local ethics committees. Data in ALSPAC is collected by self-completion postal questionnaires sent to parents, by linkage to computerized records, by abstraction from medical records, and from examination of the children at research clinics.
Birth weight and crown-heel length was obtained from a combination of medical records and measurements made by research staff, with at least 75% of measurements being taken within the first 3 days of life. At the age 9 research clinics, whole body DXA scans were performed using a Lunar Prodigy scanner with paediatric scanning software. At the same time, standing height was measured to the last complete millimetre (mm) using the Harpenden Stadiometer. Weight was measured to the nearest 50 grams (g) using the Tanita Body Fat Analyzer (model TBF 305). DXA measures included total body less head BMC (g), bone area (cm2) and BMD (g.cm−2). To provide a more complete adjustment of BMC for bone area, area adjusted BMC (aBMC) (g) was derived by adjusting BMC for bone area by linear regression (21). Whole body DXA scans were also used to provide measures of total fat and lean mass. The coefficient of variation for TBLH BMC was 0.8% based on 120 repeat scans.
Records of sunlight duration during inception of the cohort were obtained from a meteorological station in the county of Avon. UVB exposure was subsequently derived, using a linear regression model relating sunshine hours to UVB according to month, which we developed using contemporaneous sunlight and erythemal UV (eUV) recordings by the National Radiation Protection Board in Chiltern, Oxfordshire (approximately 60 miles North East from Avon), and Camborne, Cornwall (approximately 180 miles South West from Avon). For each subject, ambient eUV was imputed for the 98 days immediately prior to birth, which showed expected seasonal variation (Figure 1). Serum total 25-hydroxyvitamin D was measured at an average of 36.3 weeks gestation, in a subgroup of 355 study mothers; as expected, values [53.3 ± 31.5 nmol/l (mean and SD)] were strongly related to our estimate of background UVB in the third trimester [β = 0.020 (95% CI 0.017, 0.022), r2 = 0.40, P < 0.0001)].
The individual effect of ambient UVB on each outcome was investigated by linear regression, with effect sizes reported as a unit increase in outcome per standard deviation increase in exposure. Missing data within outcomes were considered missing at random (we have assumed that our estimates are unbiased, and so no multiple imputation methods were employed). Multivariate regression was used to examine composite models examining the effect of ambient UVB on multiple outcomes simultaneously, using wald tests to test the equality of estimates in addition to the Breusch-Pagan test of independence. All analyses were conducted using STATA 9.2.
Anthropometric and DXA-derived variables are shown in Table 1 according to gender. Study participants at the time of these analyses were a mean of 9.9 years of age. Boys were slightly taller than girls, and had a greater bone mass and lean mass, whereas fat mass was appreciably higher in girls. In univariable linear regression analyses between ambient UVB exposure during the third trimester and anthropometric data at birth, a strong positive relationship was observed with birth length whereas there was no association with birth weight (Table 2). In terms of relationships between UVB and height and weight at age 9.9, positive associations were observed for both these parameters.
Results for simple linear regression analysis between maternal UVB exposure and DXA measurements at age 9.9 are also shown in Table 2. There was a strong relationship between UVB and lean mass, but little association with fat mass. UVB was related to total body BMC at age 9, such that a one SD increase in UVB was associated with an increase in BMC of 9.6g (Table 2), equivalent to approximately 0.05 SD. Positive associations were also observed between UVB and bone area, and UVB and BMD. In contrast, little or no association was observed with aBMC.
The magnitude of effects of maternal UVB exposure on components of body mass was further explored with multivariate regression, regression estimates being calculated as standardised coefficients. UVB showed equivalent strong positive associations with BMC (0.052; 0.029, 0.075) and lean mass (0.050; 0.028, 0.073) (standardised regression coefficients with 95% confidence limits). There was no evidence of difference between these regression estimates, and bone mass and lean mass showed a high level of co-linearity (r = 0.85; residual error correlation from multivariate regression). In contrast, UV was only weakly associated with fat mass (0.014; −0.009, 0.037), whereas there was an intermediate association with weight (0.032; 0.010, 0.055).
Results from multivariate linear regression also suggested that the association between maternal UVB exposure and BMC is equivalent to that between UVB and bone area (0.049; 0.026, 0.072), with no evidence of any difference between the estimates of association, and very strong evidence of co-linearity between BMC and bone area (residual r = 0.98). Although UVB was strongly related to BMD (0.050; 0.027, 0.073), there was little or no association with aBMC (0.017; −0.006, 0.040), and there was weaker evidence of co-linearity between BMD and aBMC (residual r = 0.59).
Although maternal UVB exposure was associated with height, the strength of this relationship was somewhat weaker than that with bone area (Figure 2). Moreover, a positive albeit partially attenuated relationship was observed between UVB and height-adjusted bone area (hBA). Path analysis was subsequently performed to compare the contributions of longitudinal and periosteal bone growth to the association between UVB and bone area, using height and hBA to represent pathways for longitudinal and periosteal bone growth respectively. Standardised coefficients were calculated for the regressions between UVB and height and hBA, and between height, hBA and bone area (Figure 3). Based on the product of coefficients reflecting longitudinal and periosteal growth pathways, which were similar, we conclude that increases in longitudinal and periosteal bone growth make equivalent contributions to the stimulatory effect of maternal UVB exposure on bone mass accrual of the child. Similar conclusions were reached following equivalent analyses of the relationship between maternal UVB exposure and lower limb growth, based on lower limb bone area derived from sub-regional analysis of total body DXA scans, and measurements of lower limb length (derived from the difference of standing and sitting height; results not shown).
In further analyses intended to study the role of lean mass in mediating effects of maternal UVB exposure on bone mass, a high degree of co-linearity was observed between BMC and lean mass, bone area and height (residual r = 0.85 and 0.84 respectively), whereas the relationship between lean mass and hBA was somewhat weaker (residual r = 0.35). This suggests that UVB has the potential to influence hBA independently of lean mass. Further analyses in which UVB was found to be positively related to bone area after adjusting for height and lean mass were consistent with this conclusion (Figure 2).
We have found that ambient UVB levels during the last trimester of pregnancy are positively related to subsequent bone mass of the child as measured by whole body DXA scans at age 9.9 years, in around 7000 subjects from the ALSPAC birth cohort, such that a one SD increase in background UVB (approximately equivalent to a four week holiday in winter to a destination where UVB levels are similar to those in summer) is associated with a 0.05 SD increase in bone mass. In light of previous observations that each SD decrease in bone mass is associated with approximately a doubling of fracture risk (22), and assuming the effects that we observed persist into later adulthood, our findings suggest that for every SD increment in maternal UVB exposure there is an approximate 5% decrease in fracture risk of the off-spring. These results may help to explain the positive relationship previously reported between latitude and hip fracture risk (23). However, whether the relationship between hip BMD and fracture risk reported by Cummings et al can be applied to total body DXA scan results, which as described below may have a relatively complex relationship with fracture risk, is currently unclear.
Although pre-natal sunlight levels have previously been examined in relation to growth (24), we are not aware of any other study relating this measure to skeletal development as assessed by DXA. Our findings are consistent with results of a previous investigation of the relationship between BMD in adults and month of birth (11), which also provides an estimate for background UVB in pregnancy. A previous report that maternal veiling in pregnancy is related to BMD of the off-spring in boys (12) is also consistent with a role of prenatal maternal sunlight exposure in later skeletal development of the offspring as suggested by our results, although we saw no statistical evidence for a gender-specific interaction.
Presumably, the association between maternal UVB exposure and bone mass of the child which we found reflects the fact that UVB exposure is a major determinant of maternal levels of 25-hydroxyvitamin D3, by virtue of its role in stimulating synthesis from cholesterol precursors. As such, our results are also consistent with a previous report that maternal vitamin D levels were positively related to total body bone mass in 198 nine-year-old children from Southampton (8). Although the effect size which we found was somewhat smaller than that reported in the latter study, estimates of vitamin D exposure based on background UVB do not take into account variation in sunlight exposure related to differences in amount of time spent outside, or differences in dietary intake of vitamin D. However, we previously found that maternal dietary intake of vitamin D, as assessed by food frequency questionnaire in the last trimester of pregnancy, is unrelated to skeletal parameters of the child, possibly reflecting the fact that UVB exposure is a more important determinant of maternal 25-hydroxyvitamin D levels as compared with dietary intake of vitamin D (21).
That the measure of background UVB which formed the basis of the present study provides a reasonable instrument for maternal vitamin D status, is supported by the strong association which we observed with maternal levels of 25-hydroxyvitamin D in a sub-group of 355 ALSPAC mothers (see above under methods). We also examined childhood bone mass in relation to maternal 25-hydroxyvitamin D in this sub-group, but there was little evidence of an association. Since Avon and Southampton are at similar latitudes, and the respective study populations have similar characteristics, the explanation for any apparent difference in magnitude of effect of maternal vitamin D on skeletal development in childhood reported in these two investigations is currently unclear.
Precise understanding of how effects of maternal vitamin D on bone mass in pregnancy which we and others have found might influence fracture risk of the child in later life is also unclear. In particular, since DXA results are strongly influenced by skeletal size, associations between maternal vitamin D status and bone mass of the off-spring could potentially be explained by an effect on stature. Since the risk of hip fracture is increased in taller individuals (25), this scenario would imply that high levels of maternal vitamin D exposure increase rather than reduce fracture risk of the off-spring. This concern is highlighted by extensive evidence that maternal vitamin D exposure exerts a positive influence on skeletal growth of the off-spring (13-18).
Therefore, one of the main aims of the present study was to use the large size of the ALSPAC cohort, which provides sufficient power to estimate the effect size of potential confounders with reasonable accuracy, to tease out the role of possible height-independent effects of maternal vitamin D exposure on bone development of the offspring. This represents a novel aspect relative to previous studies in this area, such as that by Javaid which was considerably smaller, and had limited power to explore the relative contribution of height dependent and independent effects (8). As judged by the fact that UVB showed similar relations to total body BMC and bone area, and little association was seen with BMC after adjusting for bone area as reflected by aBMC, our results suggested that maternal vitamin D exposure influences later skeletal development of the child primarily by affecting bone size. Interestingly, the relationship between maternal UVB exposure and bone area was stronger than that with height, and although the association between UVB and bone area was minimally attenuated by adjusting for height a positive effect was still observed.
Therefore, maternal UVB exposure appears to influence subsequent bone growth in childhood at least partly independently of height, presumably reflecting an effect on periosteal bone growth (path analysis suggested longitudinal and periosteal bone growth make equal contributions to the association between UVB and bone area). This conclusion is significant, since in contrast to increased height, periosteal expansion is thought to confer considerable benefits in terms of geometric strength and reduction in fracture risk (26). Hence, our findings provide further justification for strategies intended to improve maternal vitamin D status in order to optimise skeletal health of the child, for example through vitamin D supplementation in pregnancy (27). However, an important caveat is that in dissecting out distinct influences on bone size, and determinants thereof such as periosteal expansion, DXA scans only provide a two dimensional projection of the skeleton. Therefore, our findings require confirmation using other techniques such as pQCT, which evaluate cortical bone geometry more directly.
Considerable co-linearity exists between the three body compartments namely fat, lean and bone mass, of which lean and bone mass are particularly highly correlated. Therefore, it is not surprising that maternal UVB exposure was associated with not only with bone mass but also with lean and to a lesser extent fat mass. In light of this high level of co-linearity, it was difficult to establish whether these findings represent a primary influence of maternal UVB exposure on bone mass with secondary effects on muscle, or vice versa. The effects of maternal UVB exposure on indices of body composition which we observed had a net effect of increasing BMI [0.026 (0.003, 0.049) for regression between UVB in pregnancy and BMI of the child], in-keeping with previous reports that month of birth is associated with the risk of obesity as defined by BMI (16,19).
Of the two pathways by which UVB influences bone growth, a high degree of co-linearity was also observed between lean mass and height, but not between lean mass and hBA. Consequently, the association between maternal UVB exposure and hBA may largely represent a direct influence on periosteal bone growth, rather than any indirect effect mediated by changes in lean mass. We are not aware of any previous reports as to programming of specific components of skeletal architecture in humans by early life events, and are keen to explore these changes in future studies, for example by use of pQCT to examine cortical geometry in more detail than can be achieved by whole body DXA. Further studies are also warranted as to the biological mechanisms by which maternal vitamin D exposure might exert long-term effects on periosteal bone formation.
In summary, we used the ALSPAC birth cohort to examine the association between estimated background UVB levels during pregnancy and bone mass of the child as measured at age 9.9 years. Maternal UVB exposure was positively related to bone mass, which appeared to reflect an influence on bone size. Although UVB levels were also positively related to height, which contributed in part to the association with bone size, approximately 50% of the influence of UVB on bone size appeared to be independent of height. Maternal UVB exposure was also found to be positively related to lean mass of the child, which may have mediated some of the effects of UVB on bone mass which we observed, particularly those involving longitudinal growth. On the other hand, associations between maternal UVB exposure and height-adjusted bone area in childhood, which are likely to represent an influence on periosteal bone growth, appeared to be independent of lean mass. Taken together, our findings suggest that maternal UVB exposure influences subsequent skeletal development of the child at least in part through a direct influence on periosteal bone formation, and may have long-term consequences for fracture risk.
We are extremely grateful to all the families who took part in this study, the midwives for their help in recruiting them, and the whole ALSPAC team, which includes interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists and nurses. The UK Medical Research Council, the Wellcome Trust and the University of Bristol provide core support for ALSPAC. Salary support for AS is provided by Wellcome Trust grant ref 079960. This publication is the work of the authors who serve as guarantors for the contents of this paper.
Neither author have any conflicts of interest to disclose