Introduction
Osteoporosis and low bone mass are major public health threats with great economic burden in the U.S. Maximizing peak bone mass in early life is an important strategy to prevent osteoporosis in later life. Bonjour et al. (2009) reported that an increase of peak bone mass in childhood by one standard deviation could reduce fracture risk as much as 50% in adulthood [
1]. Physical activity and calcium intake are two of the most important modifiable factors for bone health and their benefits to bone health could persist into young adulthood [
2,
3]. Ethnic-wise distribution shows that Hispanic adolescents are less physically active, and have lower calcium intake compared to their White counterparts [
4,
5]. Despite their importance, the role of physical activity and calcium intake with respect to genetic susceptibility to low bone mass has not been sufficiently studied in Hispanic children.
Studies that account for the full complexity of genetic and environmental factors influencing BMD have been mostly conducted in adults typically focusing on few candidate genes, with very few studies being conducted in children or capturing the effects of multiple genes [
6,
7]. Polygenic risk scores (PRSs), which could capture the genetic effects of multiple genes, could be a better approach to assess risk for osteoporosis. In one study of European children, significant interaction was found between physical activity and a PRS computed from selected SNPs identified from published genome-wide association studies (GWAS) [
8]. However, none of these studies have been conducted in a Hispanic population nor investigated genome-wide SNPs to identify significant genotypes that would interact with physical activity or calcium intake to affect BMD. Therefore, our aim was to examine if physical activity or calcium intake interact with BMD-associated PRS to affect BMD in an understudied and underserved population of Hispanic children aged 4–19 years old.
Discussion
Our study addresses the research gap to investigate the effects of interaction between PRSs and calcium intake and physical activity on BMD in Hispanic children. We found higher MVPA to be more beneficial for children with higher genetic predisposition to low bone density in comparison to children less prone to low bone density. No statistically significant interaction effects on BMD were found between genetic risk and calcium intake or dietary calcium/phosphorous ratio.
Most studies that investigated interaction between physical activity and genetic variants used a candidate gene approach in primarily adult populations [
6]. Studies have found that estrogen receptor-alpha (
ESRα) polymorphisms may modulate the effect of exercise on BMD at weight bearing sites in children [
29]. The vitamin D receptor (
VDR) genotypes were also attributed to bone metabolic response variance following resistance training in young adult males [
30]. One recent study in European children constructed a PRS based on GWAS-identified BMD loci in adults and found no evidence of effect of physical activity-PRS interactions on BMD, but did find beneficial effects on BMD of high impact physical activity on children overall with high genetic risk of low BMD [
8]. We, on the contrary, found significant evidence of interaction between MVPA and PRSs in relation to total body BMD and lumbar spine BMD. Moreover, our results suggest that higher MVPA may benefit the BMD of children genetically susceptible to low BMD, although it may not provide meaningful benefit to children less genetically susceptible to low BMD. It is possible that these genes influence the bones’ responses to physical activity or physical activity changes the way these genes are expressed. Larger sample sizes may be needed to detect the interaction effects between continuous MVPA variables and PRS on BMD, compared to binary MVPA. In individual SNPs analysis, significant effect of interaction between MVPA and
SLC8A1,
SEM1 loci and rs914153 were found to associate with BMD. Both
SLC8A1 and
SEM1 have been reported to associate with BMD or fractures in GWAS studies [
17,
31]. Exercise training was found to reduce
SLC8A1 expression in animals susceptible to heart diseases [
32] and
SLC8A1 was associated with hand grip strength and bone density [
17,
33]. Exercise training was also found in mice to affect ubiquitin-proteasome systems [
34], which has a role in skeletal muscle remodeling during exercise [
35]. As skeletal muscle and muscle strength could influence bone strength [
36], exercise might influence
SLC8A1 expression and interact with
SEM1 to affect ubiquitin-proteasome systems to further affect bone density. The SNP rs914153 is a novel locus that has not been reported to be associated with bone health or physical activity, so the mechanism is largely unknown. Given the scarcity of transcriptomic and epigenomic data from bone tissue [
37], further research is needed to determine the functional impact of these genetic variants.
We did not find any statistically significant interaction between PRS and calcium intake or dietary calcium/phosphorous ratio to affect total body BMD and lumbar spine BMD. Calcium intake and dietary calcium/phosphorous ratio are mainly below the recommendation levels in our study and the genetic variants in PRS may exert more influence on BMD with higher calcium intake. Also, calcium intake was found in randomized controlled trials in pre-pubertal children to mainly increase BMD at appendicular skeletal sites but not lumbar spine potentially due to different calcification proportion in different sites [
38,
39], which may partly explain the null interaction results on lumbar spine BMD. As PRSs only included several SNPs and our sample size is small, more genetic variants and larger sample size may be needed to detect the interaction effects. In individual variants analysis, we observed statistical interactions between dietary calcium/phosphorous ratio with
MAP 4 K3 and
SLC8A1 loci in relation to total body BMD and lumbar spine BMD respectively. As far as we know, no previous literature has reported the association between
MAP 4 K3 and bone health. However, the interaction is biologically plausible.
MAP 4 K3 encodes a kinase that is involved in the MAPK signaling pathway, which is critical to skeletal developments and maintenance and regulated by calcium [
40].
SLC8A1, which encodes a sodium/calcium exchanger, was found to be associated with heel BMD and total body BMD [
17]. Not all individual SNPs included in the PRS were found to interact with MVPA, calcium intake or calcium/phosphorous ratio probably due to involvements in different biological pathways or small effect sizes.
Our study has a few limitations. First, DXA-derived BMD is areal BMD (g/cm
2) rather than volumetric BMD; although it is commonly used and considered as the gold standard to assess pediatric bone health, areal BMD tends to be lower for smaller bones compared to larger bones, even when the volumetric BMD is the same [
24]. Also, even though total body excluding head is recommended, we used total body BMD as outcome due to SNPs associated with total body BMD may be more appropriate to assess variation from childhood to old age [
17]. Second, there are some challenges to measure dietary intake in children accurately including parents’ ability to report young children’s food intake, children’s reliability to recall food intake, and adolescents’ concerns about body image, which results in greater within-between subject variances in nutrient intake in children compared to adults [
41]. Although 2 days measurement of calcium intake were estimated to have good accuracy in adults [
42], more days may be required to capture the usual intake of calcium. Moreover, we did not capture the specific type of physical activity in accelerometers, especially, weight-bearing physical activity, which is well-known to be beneficial to BMD. But MVPA is more likely to capture the weight-bearing physical activity compared to light or sedentary physical activity. Our accelerometer may also deviate from the current recommendations [
43]. Also, the lumbar spine was not analyzed from a spine densitometry. Furthermore, although we account for main factors including age, sex, puberty stage, and BMI Z-scores, our results may not be comparable to all children studies on BMD due to the inconsistent covariates adjustment across studies and there could still be residual confounding [
44]. Additionally, the sample size is small and may suffer from false positive in the results. Future studies are needed to replicate and validate our findings. Also, limited number of SNPs are included and the conclusions could be different if different SNPs are included in the PRS. Finally, as we used cross-sectional data in the analysis, the results may not reflect causal processes and future research in longitudinal studies is needed.
Our study also has many strengths. To our knowledge, this is the first study that investigated interactions between dietary calcium and physical activity with PRS affecting BMD in Hispanic children. First, we constructed the PRS based on literature and our association results from GWAS and exome sequencing data, which could account for joint effects of multiple genetic variants and improve power for detection of interactions [
45]. Second, we created site-specific PRSs for interaction analysis due to genetic specificity to skeletal site [
20], which may provide insights on how environments associate with bone density in people with skeletal site-specific genetic risk. Third, we had physical activity data measured by accelerometers, which is an objective, nonreactive approach to estimate physical activity compared to self-reported questionnaires [
46].
In summary, MVPA is potentially more beneficial to bone health in Hispanic children who are at greater genetic risk for lower bone density. Future research on gene by environment interaction on bone health and functional studies to provide biological insights are needed.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.