Osteoporosis: A Lifecourse Approach


  • Nicholas Harvey,

    1. Medical Research Council (MRC) Lifecourse Epidemiology Unit, University of Southampton, Southampton General Hospital, Southampton, UK
    2. National Institute for Health Research (NIHR) Southampton Biomedical Research Centre, University of Southampton and University Hospital Southampton NHS Foundation Trust, Southampton, UK
    Search for more papers by this author
  • Elaine Dennison,

    1. Medical Research Council (MRC) Lifecourse Epidemiology Unit, University of Southampton, Southampton General Hospital, Southampton, UK
    Search for more papers by this author
  • Cyrus Cooper

    Corresponding author
    1. Medical Research Council (MRC) Lifecourse Epidemiology Unit, University of Southampton, Southampton General Hospital, Southampton, UK
    2. National Institute for Health Research (NIHR) Southampton Biomedical Research Centre, University of Southampton and University Hospital Southampton NHS Foundation Trust, Southampton, UK
    3. NIHR Musculoskeletal Biomedical Research Unit, University of Oxford, Nuffield Orthopedic Centre, Headington, Oxford, UK
    • Address correspondence to: Cyrus Cooper, MA, DM, FRCP, FMedSci, MRC Lifecourse Epidemiology Unit, University of Southampton, Southampton General Hospital, Southampton SO16 6YD, UK. E-mail: cc@mrc.soton.ac.uk

    Search for more papers by this author


It is becoming increasingly apparent that the risk of developing osteoporosis is accrued throughout the entire lifecourse, even from as early as conception. Thus early growth is associated with bone mass at peak and in older age, and risk of hip fracture. Novel findings from mother-offspring cohorts have yielded greater understanding of relationships between patterns of intrauterine and postnatal growth in the context of later bone development. Study of biological samples from these populations has helped characterize potential mechanistic underpinnings, such as epigenetic processes. Global policy has recognized the importance of early growth and nutrition to the risk of developing adult chronic noncommunicable diseases such as osteoporosis; testing of pregnancy interventions aimed at optimizing offspring bone health is now underway. It is hoped that through such programs, novel public health strategies may be established with the ultimate goal of reducing the burden of osteoporotic fracture in older age. © 2014 American Society for Bone and Mineral Research


Musculoskeletal disorders constitute a major public health problem.[1] Their costs approach 3% of gross national product globally,[2] and they have a major impact in terms of both acute and long-term disability, being the second greatest contributor to years lived with disability worldwide.[3] Indeed, whereas the average disability associated with other disease areas has risen by 33% from 1990, the increase secondary to musculoskeletal disorders is 45%, with such conditions accounting for 6.8% of total disability-adjusted life years.[4] Osteoporosis forms a major component of this burden: evidence from the U.S. Surgeon General's report of 2004, supported by data from the United Kingdom, suggests that almost one in two women and one in five men will have an osteoporotic fracture in their remaining lifetime from the age of 50 years.[5] The economic burden of osteoporotic fracture is substantial: in addition to the personal and societal cost incurred, estimated at 37 billion Euros annually among the 27 countries of the European Union,[6] prior and incident fractures accounted for 1,180,000 quality adjusted life years lost during 2010. With demographic changes currently occurring worldwide, costs are expected to increase 25% during the period 2010 to 2025.[6] The National Osteoporosis Foundation estimates that approximately 9 million adults in the United States suffer from osteoporosis and that by 2020 this figure will have risen to 10.7 million adults.[7]

The massive health burden imposed by osteoporotic fracture on individuals, healthcare systems and wider society is an indication of the urgent need for strategies aimed at optimization of bone mass and reduction of fracture risk. Although historically such approaches have targeted those most at risk in older age, it is becoming increasingly apparent that the fracture risk is accrued throughout the entire lifecourse, even from as early as conception.[8] Thus novel strategies might address optimization of bone mineral accrual through interventions very early in the lifecourse. In this article, we will describe insights gained from recent work investigating relationships between early growth and later bone mass, explore potential underlying mechanisms, and place this science within the context of public health, both in terms of existing policy, and the potential for novel programs aimed at optimizing the development of peak bone mass through interventions in early life.

Development of Peak Bone Mass: The Importance of the Early Environment

Bone mass (a composite of bone size and mineral density) increases through early life, childhood, and adolescence to reach a peak in early adulthood, the exact timing varying by site and sex.[9, 10] (Fig. 1). Mathematical modeling studies have suggested that the magnitude of peak bone mass achieved is a strong predictor of osteoporosis risk in later life, and that its contribution is at least as great as that of post-peak bone loss.[11] Thus optimization of peak bone mass represents a potential target for interventions aimed at reducing later fracture risk; evidence has accrued over the last two decades that such interventions might be usefully targeted very early in the lifecourse.[11]

Figure 1.

Development of bone mass over the lifecourse. Bone mass increases from the intrauterine period to a peak in early adulthood, with a decline thereafter. Modulation of the growth trajectory early in life may influence the magnitude of peak bone mass; intervention to reduce the rate of post-peak bone loss may be appropriate in older age. Reproduced with permission from Cooper and Melton.[10]

In adult cohorts where birth records exist, birth weight is positively associated with bone mass in both young and late adulthood[12, 13]; poor early growth has also been found to be a marker of increased risk of adult hip fracture.[14, 15] The observations linking birth weight to adult bone mass have been confirmed in a recent systematic review and meta-analysis indicating that overall[16] a 1-kg increase in birth weight is associated with a 1.41-g increase in hip bone mineral content (BMC) in adulthood. The associations between early growth and adult bone mass appear strongest for adult BMC rather than bone mineral density (BMD), suggesting that it is the size of the overall skeletal envelope which may be influenced more than volumetric density. Such associations are clearly likely to be of clinical importance because BMC predicts fracture risk comparably to BMD (indeed BMD from dual-energy X-ray absorptiometry [DXA] is partly determined by bone size),[17] and the diameter of a tubular bone strongly contributes to its bending strength.[18] Furthermore, the shape of the proximal femur, a determinant of fracture risk, has been associated with early growth, with higher weight at 1 year predicting a stronger femoral neck[19, 20] in adulthood.

Although such associations might suggest the direct pleiotropic effect of multiple genes on infant size and adult bone mass, there are several lines of evidence that support the role of early environmental influences: birth weight itself has a substantial environmental component, with intrapair differences in birth weight generally larger (and absolute mean birth weights lower) for monozygotic than dizygotic twins (showing the importance of differential placentation)[21-23] and the clear impact of factors such as maternal smoking[24] and gestational diabetes[25] on neonatal size; the concept of maternal constraint, beautifully illustrated with the Shire horse-Shetland pony crosses,[26] is described in Growth and Skeletal Development, Historical Perspective. In two twin studies, intrapair differences in birth weight correlated with adult hip BMC,[27, 28] with larger differences in monozygotic than dizygotic twins.[27] For both birth weight and adult bone mass, the documented contribution of specific genes appears modest,[29-31] and the secular increase in birth weight observed in many populations[32-34] over recent years is far too rapid to be attributable to change in the genetic code itself. Finally, experimental studies in animal models have clearly demonstrated the ability of alterations in maternal diet during pregnancy to cause long-lasting changes in offspring phenotype.[35-37] Although direct demonstration of the balance between gene and environment in humans is lacking, and newer techniques such as whole-genome sequencing may reveal previously undetected genetic contribution, such observations clearly establish the principle that environmental factors are able to play an important role in early growth and development.

Evolutionary Biology

Gene-environment interaction in the determination of later disease

The concept that environmental influences early in life might have long-term implications for adult health and disease has become well established over the last 20 years. Initial United Kingdom studies by David Barker and colleagues documented a high correlation between geographic areas with high rates of cardiovascular disease and those with high infant mortality five decades before.[38, 39] This pioneering work led to the hypothesis that many of the common chronic noncommunicable diseases of middle and older age might have their origins in a mismatch between the environment experienced in utero and that experienced in postnatal life (Fig. 2).[40] The concept of environment modifying phenotype through altered gene expression is ubiquitous in the natural world and is termed developmental plasticity; ie, the ability of a single genotype to generate multiple different phenotypes dependent upon the prevailing environmental milieu.[41, 42]

Figure 2.

Conceptual illustration of phenotypic diversion with age and potential for early environmental modulation. Adapted with permission from Hanson and colleagues.[42] NCD = noncommunicable disease.

Direct demonstration of the relative contributions of gene and environment to associations between early growth and later osteoporosis is lacking, and given the low overall coverage of current approaches of genome-wide association studies (GWASs), new technology for genetic characterization, such as whole-genome and exome sequencing, may yet reveal a much greater direct genetic contribution to individual phenotypes. However, it is likely that the distinction between purely genetic and purely environmental effects is somewhat artificial, and it has become increasingly apparent 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.[41] This variation in gene expression may be controlled by epigenetic processes,[36, 41] which allow gene expression to be switched on or off. Several epigenetic mechanisms have been characterized, including DNA methylation, chromatin histone modification, and noncoding RNA. At the level of the individual gene in a single cell, adding a methyl group to a target CpG site within the promoter region usually switches gene expression off; ie, methylation or demethylation lead to either 0% and 100% activity, respectively, at the individual gene site. 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.[41] Evidence is accruing that both genotype, and factors such as maternal diet and smoking during pregnancy, may influence the distribution of methyl marks in the offspring.[43] Indeed, experimental studies in animals have clearly demonstrated that perturbations of the environment experienced in pregnancy can lead to altered epigenetic marking, changed gene expression, and consequent phenotypic differences in the offspring, which can be maintained into the next generation.[35, 37, 44, 45] However, true inheritance of epigenetic marks, ie, beyond the second generation in which developing primordial germ cells may have been exposed to the initial environmental trigger, remains the subject of investigation.[46] Epigenetic processes may therefore permit an organism to fine-tune gene expression, enabling appropriate adaptation to the prevailing environment over one or two generations. In contrast, alterations to the genetic code, which may be preferentially selected over vastly greater time spans, may allow adaptation to much longer-term influences.

The approach to the characterization of epigenetic signals can be usefully modeled on that employed in the exploration of fixed genetic variation, but with the added complexity that epigenetic marking may be tissue specific and vary with time.[47] Thus epigenome-wide arrays are now available to screen around 500,000 individual CpG sites for methylation signals. Bioinformatic approaches to the analysis of these data yield associations between differentially methylated regions of interest and outcomes such as offspring bone mass and body composition,[48] which form the basis for exploration of candidate associations by measurement of methylation at specific sites using techniques such as pyrosequencing and mass-array. Emerging technologies are likely to yield greater returns: Next Generation methylation Sequencing (NGS) now permits coverage of the entire methylome with single CpG site resolution, such as in the ongoing Encyclopedia of DNA Elements (ENCODE).[49] Although currently expensive and not widely available, this technology has the potential to allow high-throughput discovery science, for example in cohort-based epigenome-wide association studies (EWAS), analogous to methods used previously to explore fixed genetic variation. Given the quantity of data generated by single-methyl resolution across the entire genome in a large population cohort, analytical approaches must be carefully thought out, usually based on sophisticated bioinformatic techniques, in order to minimize the false positive rate and to optimize interpretation of the results[50] (Fig. 3).

Figure 3.

Schematic representation of the investigation of methyl marking from epigenome to candidate. Next-generation methylation sequencing allows identification of individual methyl marks across the entire methylome. Array-based approaches permit identification of DMROIs across a wide genomic area. Refinement of candidate selection and investigation of individual CpG methylation may be obtained from techniques such as Sequenom and pyrosequencing. Functional significance may be elucidated from transcription factor binding, mRNA, and protein expression, and further validated using knockout models in cell lines and whole animals. DMROI = differentially methylated region of interest.

Intrauterine vitamin D exposure, epigenetic marks, and later bone health

Such investigations have revealed novel insights into the regulation and mechanistic processes involved in bone development in utero.[8] Several epidemiological studies have demonstrated positive relationships between maternal 25(OH)-vitamin D concentration in pregnancy and offspring bone mineral in childhood,[51-58] although such findings were not apparent in all studies,[59-61] with a large recent UK study finding no association.[59] Interestingly, this result contradicted an earlier finding from the same UK ALSPAC cohort, in which maternal UVB exposure (a major determinant of 25(OH)-vitamin D status) during late pregnancy was positively associated with offspring whole-body BMC and BMD at 9 years old.[56] The chance strong correlation in this cohort between maternal UVB exposure during pregnancy and the age at which the offspring underwent DXA assessment make these conflicting findings difficult to interpret with certainty.

Data suggest that association between maternal 25(OH)-vitamin D concentration and offspring bone mass might be mediated, at least in part, through placental calcium transport.[52] In the Southampton Women's Survey,[62, 63] and confirmed in a postmortem study using CT,[63] maternal vitamin D status was associated with femoral morphology of the developing fetus, even as early as 19 weeks gestation. mRNA expression of an active ATP-dependent placental calcium transporter, PMCA3, in placental tissue, was positively associated with offspring bone area and BMC of the whole-body site at birth.[64] The regulation of placental calcium transfer is poorly characterized in humans, and any mechanistic role of vitamin D remains to be elucidated, but members of the PMCA family appear to be regulated by 1,25(OH)2-vitamin D in animal studies.[65] Furthermore, the ongoing regulation of vitamin D metabolism may involve methylation of sites in the 1α-hydroxylase promoter region, suggesting a role for epigenetic processes in the vitamin D-parathyroid hormone axis.[66, 67]

Collection of umbilical cord samples from two UK cohorts has allowed the elucidation of relationships between epigenetic marking at candidate sites, identified through array approaches, and offspring bone size, mineralization, and density. Thus, in 66 mother-offspring pairs from the Princess Anne Hospital Cohort study, percentage methylation at two CpG sites in the promoter region of endothelial nitric oxide synthase (eNOS) in umbilical cord was positively related to the child's whole-body bone area, BMC, and areal BMD at age 9 years (r = 0.28 to 0.34, p = 0.005 to 0.02).[68] Replication of these results in a second cohort will be required. Furthermore, in the Southampton Women's Survey (Fig. 4), higher methylation at four of six CpG sites in the promoter region of retinoid-X-receptor A (RXRA) in umbilical cord was inversely correlated with offspring BMC corrected for body size at 4 years old (β = −2.1 to −3.4 g/SD, p = 0.002 to 0.047), results supported by findings from a second independent cohort.[69] RXRA forms a heterodimer with several nuclear hormones known to influence bone metabolism, including 1,25(OH)2-vitamin D, and here the maternal 25(OH)-vitamin D index (ratio of 25(OH)-vitamin D to vitamin D binding protein) was inversely related to methylation at one of these CpG sites.[69] Evidence of functional significance was obtained through altered response to transcription factor binding and further characterization of these processes is ongoing, but clearly replication in independent cohorts by other groups will be required to validate such findings.

Figure 4.

Percentage methylation at RXRA promoter region CpG sites and offspring bone mineral content corrected for body size. Reproduced with permission from Harvey and colleagues.[69] WB = whole body minus head; BMC = bone mineral content; scBMC = size-corrected BMC, ie, BMC adjusted for bone area, height, and weight.

It is apparent that epigenetic marking in early life is associated with later phenotypic variation. However, given the potential tissue specificity of epigenetic signals, the variation of such marks over time, and the difficulty in differentiating cause from effect, the exact characterization of epigenetic mechanisms in disease etiopathology is a complex process.[42, 70] Epigenetic marks identified in human cohorts through array and candidate investigation must be replicated in separate independent cohorts to robustly establish associations with later disease; experimental work using cell culture and animal models is required to document the detailed molecular processes, regulation, and functional consequences. A combination of such fundamental investigation and linkage to disease development will be essential to fully understand the role of epigenetic mechanisms in the causation of human pathology. In the meantime, whether the observed epigenetic marks are cause or consequence, if replicated, such signals may well present useful novel biomarkers for later adverse bone development.

Growth and Skeletal Development

Historical perspective

The net result of such processes and influences as described above is the trajectory of growth from conception into childhood through to achievement of peak bone mass. One of the earliest descriptions of growth through childhood was made by Count Philibert Gueneau de Montbeillard, charting his son's increase in height over 18 years from 1759 to 1777, and recorded by Georges-Louis Leclerc, Comte de Buffon, in the French Histoire Naturelle. A plot of absolute height, and height gain per year, demonstrated rapid change over the first 2 years of life but with progressive deceleration over this period, a relatively stable rate of growth until puberty followed by a rapid acceleration, and then deceleration over the later pubertal period. The growth patterns so documented by Montbeillard have been confirmed by subsequent investigators, in particular John Tanner, who undertook much of the seminal work in this area.[71-73] Assessing growth in postnatal life is relatively straightforward, not withstanding logistic issues of cohort inception and repeatability of measurements; attempting to generate the same sort of graphical representation of growth in utero is much more problematic. Until the advent of ultrasound scanning, this work inevitably had to rely on cross-sectional data. Indeed an early paper from Tanner, in which growth is described across the intrauterine, and into the postnatal, period, relied on separate prenatal and postnatal datasets. The prenatal data were cross-sectional, based on measurements of neonates born at different gestations, and the postnatal measurements were longitudinal. Cross-sectional measurements of size at different gestations, rather than longitudinal measurements within subjects, can lead to erroneous conclusions regarding growth trajectory, tending to overly smooth individual curves. More recently longitudinal data have been published, and these will be described in subsequent sections.

The pattern of growth described by Tanner suggests that linear growth velocity increases throughout the first half of pregnancy and then slows; velocity of weight gain appears to slow from around 32 weeks.[71-73] This growth velocity deceleration as delivery is approached is an important physiological happening, because unchecked growth might lead to a baby unable to exit through the birth canal. This phenomenon is known as “maternal constraint” and is poorly understood in mechanistic terms. It is, however, clearly a powerful process: a cross between a Shetland pony and a shire horse results in the birth of a small foal when the Shetland pony is the mother and a large foal when the shire horse is the mother[26]; interestingly, the respective offspring are the same size when fully grown, which is midway between the two parents. This graphic illustration from an animal model is consistent with human data (as described in Gene-environment interaction in the determination of later disease), suggesting that there is a substantial environmental contribution to birth weight.

Prospective assessments of early growth and postnatal bone size, mineralization, and geometry

The use of high-resolution ultrasound measurements during pregnancy have enabled much more detailed, and, importantly, longitudinal assessment of fetal growth to be made, and to be related to later outcomes such as bone mineral and body composition.[74-76] The UK Southampton Women's Survey (SWS)[77] is particularly well set up to investigate relationships between early growth and childhood bone mass. It is a unique, prospective, population-based cohort study of 12,583 initially nonpregnant women aged 20 to 34 years, and representative of the UK population. Women were assessed in detail at study entry in terms of diet, lifestyle, body build, physical activity, health, and medications, and venous blood was collected. Similar assessments were conducted in early (11 weeks) and late (34 weeks) gestation in those (n = 3159) who became pregnant. High-resolution ultrasound scans were obtained at 11, 19, and 34 weeks' gestation with measurements of indices such as crown-rump length, head, chest, abdominal circumferences, and femur length performed according to published guidelines by two trained operators. Offspring have been measured in detail at birth and then at 6 months, 1, 2, and 3 years. Consecutive samples of around 1000 children have undergone DXA (whole body and lumbar spine, and from 4 years onward, also both hips) and anthropometric assessments at birth, 4 and 6 years, with visits at 8 and 11 years ongoing. pQCT measurements of tibial bone strength are available at 6 years.

The choice of statistical methods with which to model growth continues to be widely debated, and different methods (eg, conditional regression, fitted linear or polynomial equations, latent growth curve modeling) all have their merits. The time points at which data are recorded may impose some restriction on the method chosen; eg, fitted polynomial equations are more appropriate for data recorded at many, and potentially overlapping, points across time. In the SWS analyses, where assessments are undertaken at discrete time points, conditional growth modeling based on variables converted to within group Z-scores was employed. The technique makes no assumptions about segmentation or shape of the growth trajectory, and uses regression-derived residuals to generate mutually independent variables describing growth velocity across successive periods, eg, 11 to 19 weeks, 19 to 34 weeks, and birth to 1 year.

Data from the SWS have informed three areas of investigation: (1) differential effects of growth on bone size and bone mineral density; (2) specific timing of relationships; and (3) effects on hip geometry. In a subset of 380 mother-child pairs with complete data for femur length and abdominal circumference at 19 and 34 weeks, together with DXA indices at 4 years, differential relationships were observed between growth measurements and childhood bone mass depending on adjustment for body size.[74] Thus the velocity of late pregnancy fetal abdominal growth was positively associated with childhood BMC after adjustment for body size (BMC adjusted for bone area [BA] height and weight: r = 0.15, p = 0.004) but not with skeletal size (BA: r = 0.06, p = 0.21). In contrast, the velocity of late pregnancy fetal femur growth was positively associated with 4-year skeletal size (BA: r = 0.30, p < 0.0001), but not with size-corrected BMC (r = 0.03, p = 0.51). Given that femur length is a component of skeletal size, it is not surprising that it is strongly associated with whole-body bone area at 4 years, and less so with BMC adjusted for body size (which also gives an indirect indication of volumetric mineralization). However, the associations between abdominal circumference growth and size-corrected BMC are intriguing, and suggest that changes in the subcomponents of the ultrasound measure, for example liver size or subcutaneous fat stores, might influence accrual of volumetric BMD. The involvement of adipose tissue is supported by findings from an earlier study in which umbilical cord venous serum concentrations of leptin were positively associated with DXA-assessed size-corrected BMC in the neonate,[78] and the known effects of leptin on bone formation; in contrast, umbilical cord venous serum concentrations of IGF-I, a potent osteoblast stimulus produced in the fetal liver, have been associated with neonatal bone size, but not density, in a similar study design.[79]

This work was extended to explore the temporal relationships between different phases of intrauterine and postnatal growth, and postnatal bone mass at birth and 4 years, using abdominal circumference, which can be measured at all three time points (11, 19, and 34 weeks' gestation; unlike femur length, which can be assessed from 19 weeks onward), as the primary growth measure.[76] This measure was also available postnatally at birth and 1, 2, and 3 years. Linear growth was assessed by femur length at 19 and 34 weeks and then postnatally by length (birth and 1 year) and height (2, 3, and 4 years old).

Relationships differed according to timing in pregnancy, with abdominal circumference growth in late pregnancy strongly related to bone mass at birth, but less so with bone mass at 4 years. In contrast, abdominal circumference growth in early pregnancy was more strongly related to bone mass at 4 years than at birth. For linear growth, the strongest associations with bone mass at 4 years were for growth in late pregnancy and in the first 2 years of postnatal life. Indeed, the proportion of children changing their position in the length distribution at each postnatal time point progressively decreased, consistent with the gradual settling on to a more sustained growth trajectory. This pattern of perturbation of growth in late pregnancy and early infancy, with gradual settling onto a longer-term trajectory, was confirmed in a study relating early linear growth to proximal femoral geometry at 6 years old in Southampton Women's Survey children.[75] Here, in 493 children assessed by DXA at 6 years, hip strength analysis was related to conditional measures of linear growth from 11 weeks gestation to 6 years postnatal life, with the strongest relationships for growth in late pregnancy and infancy. Figure 5 shows the standardized regression coefficients for linear growth at each time interval as predictors of narrow femoral neck section modulus (a measure of bending strength).

Figure 5.

Linear growth from late pregnancy into childhood and narrow femoral neck section modulus at 6 years old. Data are standardized regression coefficients showing the SD change in outcome per SD change in growth over each time period. The measures of linear size are crown-rump length (11 weeks gestation), femur length (19 and 34 weeks gestation), crown heel length (birth and 1 year postnatally), and standing height British Growth Foundation Z-scores (2 to 6 years postnatally). Reproduced with permission from Harvey and colleagues.[75] NN-Z = narrow neck section modulus.

Critical growth periods for later bone development

These observations support the earlier findings documented by Tanner,[71-73] and the models of Karlberg and colleagues,[80] but here for the first time, such associations have been documented using objective measures in a truly prospective cohort. These results would be consistent with the notion that there may be critical periods where growth velocity relates very strongly to longer-term measures of bone development, and thus offer potential opportunities for early intervention to optimize skeletal strength. Indeed, these findings have been supported by studies from birth cohorts elsewhere. In Generation R, a mother-offspring cohort in Rotterdam, Netherlands, fetal weight gain and catch-up in weight were associated with BMD at the whole-body site at 6 months.[81] Furthermore, children remaining in the lowest tertile of weight from birth to 6 months had a much higher risk of having low BMD at the whole-body site at 6 months of age. In a Norwegian cohort, fetal femur length was measured using 2D ultrasound in 625 pregnancies. Consistent with the pattern of altered growth velocity in late pregnancy, femur length Z-scores measured between 10 and 19 weeks were progressively less strongly correlated with later measurements (r = 0.59, weeks 20 to 26; r = 0.45, weeks 27 to 33; and r = 0.32, weeks 34 to 39; all p < 0.001).[82]

Early growth and risk of adult hip fracture

These studies consistently suggest that there is marked variation in growth trajectory across a period from late pregnancy into early childhood, and that development during these periods is of relevance to later bone health. Data from adult cohorts in which birth records exist clearly demonstrate the longer-term implications of such findings: in a large Finnish cohort, poor growth both in early and later postnatal life was associated with increased risk of adult hip fracture seven decades later[14, 15] (Fig. 6). This finding is complemented by results from the Hertfordshire cohort demonstrating positive associations between early weight and femoral cross-sectional area, independent of femoral neck length. Taken together, such results support a link between early growth, femoral geometry, and risk of adult hip fracture,[20] and critically, potential windows of opportunity for intervention early in life to optimize skeletal development.

Figure 6.

Poor growth in utero and childhood predict later risk of hip fracture: Helsinki and Hertfordshire Cohort Studies. Figure demonstrates data from the original Hertfordshire birth ledgers, together with hazard ratio for adult hip fracture by quartile of change in BMI Z-score from age 1 to 12 years in the Helsinki cohort. Figure based on data derived from Javaid and colleagues.[14] HR = hazard ratio; BMI = body mass index.

Emerging Research Questions: Etiology and Observation to Intervention

The United Nations (UN)[83] and the World Health Organization (WHO)[84] have recently published documents outlining the key role of the early environment in the determination of later common chronic non-communicable diseases. Such illnesses have been a major focus of UN and WHO policy over the last decade and reports published by both in 2011 clearly document the massive burden of noncommunicable diseases worldwide. Risk factors in adulthood such as cigarette smoking, alcohol, and obesity are clearly described, but so is the additional contribution made by adverse environmental factors during pregnancy. Both reports recommend that pregnancy is a critical period in which future risk of noncommunicable diseases may be accrued and thus should form part of any plans to address the incidence of these conditions.

The consideration of early life influences as part of high-level international guidance reflects the strength of the observational evidence as a basis for further investigation to identify actual potential interventions. In order to appropriately inform public health strategy, such observations must be tested using intervention studies. Key research issues include whether vitamin D supplementation during pregnancy will result in benefits to offspring musculoskeletal health; the dose and timing of supplementation required to achieve optimal results; and the elucidation of patient characteristics or early biomarkers that may be used to identify those who will benefit most from intervention. Traditional double-blind placebo controlled randomized trials are eminently suited to investigate such questions, an example of which is the ongoing Maternal Vitamin D Osteoporosis Study (MAVIDOS). In this UK trial women are randomized to either 1000 IU cholecalciferol daily or matched placebo from 14 weeks gestation until delivery of the baby, with whole-body bone mineral content of the infant the primary outcome.[85] A complementary approach is to address the ability of individuals to optimize compliance with medication and alterations to diet and health behavior appropriate to pregnancy. The role of maternal behavior, the mother's ability to control her own destiny, and her perception of that ability, are increasingly recognized to be major determinants of the potential to make such changes.[86] Evidence from a complex intervention study based in UK Sure Start centers (community hubs at which young women and families may access help and advice relating to pregnancy and childcare) suggests that a psychologically derived conversation-based intervention may help to empower women to comply better with supplementation and current national guidance on diet and lifestyle in pregnancy.[86-90] This intervention lends itself to a complex mixed-methods intervention, incorporating elements of the traditional randomized controlled trial, overlaid with quantitative analysis and qualitative data from individual interviews, and will form the basis for the next phase of the UK MAVIDOS trial. Observational studies clearly present limitations in the ability to derive causal influences, and intervention studies such as MAVIDOS are essential to establish the correct approach to intervention in pregnancy.[91]


Early growth represents the end-product of many influences, both environmental and genetic, together with the results of interaction between them. Our understanding of the implications of such interactions early in life for later health and disease has increased markedly over the last two decades. Furthermore, potential underlying mechanisms are being elucidated and tested for functional significance. Putative early biomarkers of later adverse health outcomes have been described, and the roles of early interventions, such as maternal vitamin D supplementation during pregnancy, are being tested in randomized placebo-controlled trials. In a condition such as osteoporosis, which is so common in the population, and associated with such a massive impact at the level of the individual, healthcare systems, and wider society, it is imperative that every avenue available is taken to ameliorate this burden. Only with such approaches will we be able to truly optimize bone development across the lifecourse and thus reduce the impact of osteoporosis and related fractures in older age.


NH has received consultancy fees, lecture fees, and honoraria from Alliance for Better Bone Health, AMGEN, MSD, Eli Lilly, Servier, Shire, Consilient Healthcare, and Internis Pharma. CC has received consultancy fees, lecture fees, and honoraria from AMGEN, GSK, Alliance for Better Bone Health, MSD, Eli Lilly, Pfizer, Novartis, Servier, Medtronic, and Roche. ED states that she has no conflicts of interest.


This work was supported by the Medical Research Council (UK), Arthritis Research UK, National Osteoporosis Society (UK), International Osteoporosis Foundation, NIHR Southampton Biomedical Research Centre, and NIHR Oxford Musculoskeletal Biomedical Research Unit. We thank them for funding this work.

Authors' roles: All authors contributed equally to authorship of the manuscript.