AIM: To model the pre and postnatal determinants of volumetric bone density and geometry in the child at 8 to 17 years.

BACKGROUND

Osteoporotic fractures constitute a huge public health burden, with a currently estimated annual cost of 1.7 billion in the UK(1). The peak bone mass achieved in early adulthood is a major determinant of osteoporosis risk in older age(2). Peak bone mass is partly inherited, but the nature of the inheritance from either parent in terms of site specificity, size, geometry and volumetric density is uncertain. Additionally, evidence is accruing that environmental factors early in life may have long term influences on childhood bone development to peak(3). Thus poor growth in early life is associated with reduced bone mineral content and increased risk of hip fracture in adulthood(4;5). Maternal lifestyle, body build, physical activity, diet and circulating 25(OH)-vitamin D status during pregnancy are associated with offspring bone mass(6;7). Animal studies have demonstrated similar effects(8;9) and suggested epigenetic processes such as DNA methylation as potential mechanistic explanations for these gene-environment interactions(10-12). However, in humans, it is not clear whether influences on bone mass in the neonate retain a persisting effect through childhood, and secondly whether these effects are mediated through bone size, geometry or true volumetric density. Existing data from mother-offspring cohorts have mainly utilised DXA(13;14), which, as a two-dimensional construct, does not allow measurement of true volumetric density, or differentiation of effects on cortical vs trabecular bone. Maternal behaviour before and during pregnancy is likely to influence how she feeds and looks after her child. In order to elucidate a model of pre and postnatal genetic and environmental determinants of childhood bone size, density and geometry, it is therefore necessary to address several issues:

1) Inheritance (genetic) of bone size, geometry and density from mother and father.

2) Environmental factors acting before and during pregnancy (e.g. maternal diet) which might act on fetal bone development in utero and potentially alter the long term growth trajectory.

3) Environmental factors observed before and during pregnancy which merely act as markers of later behaviour and child rearing (e.g. an influence of maternal diet in pregnancy on offspring bone development could act through an association between maternal diet in pregnancy and that of the child postnatally).

4) Postnatal environmental factors which may have temporary or permanent effects on bone (e.g. physical activity and calcium/ vitamin D exposure).

In this proposal we aim to use two large, well established mother-offspring cohorts to attempt to address these problems (SWS and ALSPAC).

Summary of Southampton Women's Survey (SWS): The SWS(15) is a unique ongoing prospective population cohort 12,583 non-pregnant women aged 20-34 years, resident in the city of Southampton and recruited through their general practitioners. Each woman was visited at home for an interview, which involved a questionnaire enquiring about diet, employment, physical activity, general health, alcohol consumption and cigarette smoking. Nutrition was assessed from a previously validated 100-item food frequency questionnaire. Detailed anthropometric measurements of the women (height, weight, leg length, skinfold thickness at the triceps, biceps, subscapular and supra-iliac regions) were made in order to assess body composition. From these 12,583 women, 3156 became pregnant and delivered infants within three years of initial interview. At early (11 weeks) and late (34 weeks) pregnancy further assessments were made. 1003 of these neonates underwent DXA of their whole body and lumbar spine within two weeks of delivery. Subsequent follow up at 4 (n=900, complete) and 6 (n=600, ongoing) years has included DXA (Hologic Discovery: whole body, lumbar spine and hip) and objective physical activity (combined accelerometer and heart rate monitor: Actiheart, Cambridge Neurotechnology, Cambridge, UK) assessment over 7 days continuously. Simultaneous recording of maternal physical activity by Actiheart is available at 4 years.

METHODS

The SWS 8-year follow-up is underway. Children are invited to attend for whole body, lumbar spine and hip DXA measurements (Hologic Discovery instrument). Anthropometry and grip strength measurements will also be made and a lifestyle/dietary questionnaire completed.

In order to model the influence of parental bone mass, and demarcate relationships by bone size, geometry and true volumetric density, we now propose to assess the parents of the SWS children, in addition to further refining the bone phenotype of the offspring. Interested parents will be invited to attend a research clinic at the Osteoporosis Centre at Southampton General Hospital. Here, a lifestyle questionnaire will be administered; height, weight, skinfold thickness and grip strength will then be recorded. Bone mass will be assessed in two ways: A DXA scan (Hologic Discovery) will be performed and measurements made of whole body, lumbar spine and hip bone area, bone mineral content and areal BMD in addition to total and regional lean and fat mass. Peripheral QCT at the tibia (4 sites to assess areas of predominantly cortical and trabecular bone) using a Stratec XCT 2000 machine. Measurements will be made of cortical and trabecular volumetric density, cortical thickness and cross sectional area, periosteal and endosteal circumference, bending strength and fracture load in the x and y axis. Muscle mass measurements will also be made. At this visit the 8 year old child will also undergo the same pQCT assessment yielding 500 mother-father-child triplets with DXA and pQCT indices generated on the same instruments. In a subset of 250 triplets, an Actiheart monitor will be fitted to each of mother, father and child, to be worn continuously for 7 days and posted back in a padded reply paid envelope at the end of this period.

ANALYSIS

1) SWS: All data will be checked for validity and outliers. Variables will be assessed for normality and transformed in necessary. The relationships between the exposures and the bone outcomes will be initially explored in univariate models. Associations found to be statistically significant (pless than 0.05) will be included in multivariate models to generate the set of maternal, paternal and childhood variables which best explain the bone outcomes at 8 years. Collinearity between parental and childhood environment, e.g. diet, will also be investigated to distinguish between true early life effects and inherited environmental factors.

2) ALSPAC: This will utilise existing data which will already be have been cleaned. Variables will be assessed for normality and transformed in necessary. The analysis will comprise two phases:

The initial analysis will again employ univariate models to explore the associations between parental and childhood factors and childhood bone size, areal density and size corrected density at 9 years. A multivariate model will then be generated (incorporating a variables where pless than 0.05 for the relationship with child bone outcomes) to compare with that derived from the SWS.

The second phase will include a similar process of univariate then multivariate modelling with qQCT derived indices of tibial size, geometry and volumetric density at 15 and 17 years. The aim of this analysis will be to confirm whether influences observed at 8 years in SWS persist after puberty.