Association of blood mercury levels with bone mineral density in adolescents aged 12–19

Bone mass increases rapidly in adolescence, and achieving higher bone mineral density (BMD) during this period can help prevent osteoporosis. However, the effects of metallic mercury on bone health remain controversial. Previous studies have discussed perimenopausal women and older adults, while the association of blood mercury with BMD in adolescents is yet to be studied. Date was collected from the National Health and Nutrition Examination Survey (NHANES) 2011–2018. Weighted multiple linear regression models were used to explore the association of blood mercury levels with BMD in adolescents, while smooth curve fittings and weighted generalized additive models were used to identify the potential nonlinear association. We found that blood mercury levels were negatively associated with BMD in adolescents, though not significantly, based on the results of statistical analyses of 2818 participants. Additionally, the trend in BMD with changes in blood mercury was different in male and female adolescents. We also found an inverted U-shaped association between blood mercury and BMD in male and Mexican–American adolescents. This suggests that increased blood mercury levels within a range may benefit bone health in male adolescents (inflection point: 5.44 nmol/L) and Mexican–American adolescents (inflection point: 5.49 nmol/L), while higher blood mercury levels may harm bone health. More prospective research is needed to confirm our findings.


Introduction
Osteoporosis, a chronic disease characterized by the destruction of bone microstructure and low bone mineral density (BMD) that affects millions of people worldwide, has been recognized as a global public health issue (Torres-Costoso et al. 2020). Adolescent bone mass increases rapidly, reaching approximately 95% of peak adult bone mass by the end of adolescence (Bland et al. 2021;Sioen et al. 2016). Therefore, achieving higher bone mass during adolescence is critical for osteoporosis prevention (Golden and Abrams 2014). Osteoporosis is also considered a pediatric disease due to the importance of adolescence for bone accumulation (Ciancia et al. 2022). Therefore, exploring the influencing factors of bone loss and BMD reduction is essential to protect highrisk groups (Alejandro and Constantinescu 2018). Therefore, early detection and prediction of BMD reduction in adolescence are necessary to prevent osteoporosis.
Mercury, a highly toxic heavy metal, seriously threatens human health, like lead and cadmium. Individuals are primarily exposed to mercury through occupational inhalation, dental amalgam, and food ingestion, especially from consuming contaminated seafood (Järup 2003). The toxicity of mercury to human organs has been widely reported, such as neurotoxicity, immunotoxicity, renal failure, and endocrine disorders (Rodríguez and Mandalunis 2018).
Numerous previous studies have suggested that heavy metals such as lead and cadmium harm bone health (Brito et al. 2014;Ma et al. 2021;Engström et al. 2012). In addition, the negative correlation between blood lead, blood cadmium, and BMD has been proved in some studies (Lu et al. 2021;Campbell and Auinger 2007). However, it is unclear whether and how mercury affects bone health. Some previous cross-sectional studies suggest that blood mercury may have a protective effect on BMD (Pollack et al. 2013;Cho et al. 2012;Kim et al. 2016). However, in another study, the protective effect was only found in specific populations (Tang et al. 2022). In addition, adults over the age of 20, older adults over 50, and perimenopausal women have been studied, and no study has discussed the association of blood mercury with BMD in the adolescent population. Therefore, this study aimed to explore the association of blood mercury levels with bone mineral density in adolescents using large population data.

Data source
The data analyzed in this study were derived from the National Health and Nutrition Examination Survey (NHANES) database. NHANES is an extensive, ongoing, cross-sectional survey aimed at collecting and assessing the health and nutrition status of adults and children across the USA. This survey was approved by the National Center for Health Statistics (NCHS).
We collected data from the NHANES database from 2011 to 2018. We limited the age of the participants to 12-19 (n = 5215), and after excluding missing values for total BMD (n = 1033) and total blood mercury (n = 1364), a total of 2818 (1458 boys, 1360 girls) participants aged 12-19 were included in this study (Fig. 1).
The NHANES project was approved by the NCHS Ethical Review Board. All participants in NHANES gave written informed consent to the survey. Adolescents under 18 were given informed consent by their parents or guardians, while those 18 and older gave their own consent.

Study variables
The exposure variable for this study was blood mercury levels. The blood mercury concentration was measured by inductively coupled plasma mass spectrometry (ICP-MS), an elemental analysis technique using quadrupole technology. In addition, total bone mineral density was measured using dual-energy X-ray absorptiometry as the outcome variable in this study. The covariates of this study were as follows: sex, race/ethnicity, education level, and moderate activities were included as categorical variables; age, body mass index, income to poverty ratio, total protein, total cholesterol, serum phosphorus, and serum calcium were included as continuous variables (Online Resources 1, 2 and 3 show the effect of each covariate on the exposure variable and the outcome variable). For further information on total blood mercury, total BMD, and covariates, visit wwwn.cdc.gov/ nchs/nhanes/.

Statistical analysis
All analyses in this study were carried out using weighted NHANES samples. We divided participants into quartiles (Q1-Q4) based on blood mercury levels. We built the following three models with reference to the statement of Strengthening the Reporting of Observational Studies in Epidemiology (von Elm et al. 2007): model 1: no adjustment for any variables; model 2: adjustment for sex, age, and race/ethnicity; and model 3: adjustment for covariates screened out. In order to examine the independent relationship between blood mercury levels and total BMD, we used weighted multiple-linear regression models. Smooth curve fittings and weighted generalized additive models were applied to determine the nonlinear association of blood  mercury with total BMD in stratified analyses. Furthermore, the threshold effect of blood mercury levels on total BMD was calculated using two-piece linear regression models. To avoid statistical bias caused by the direct exclusion of missing data, we applied multiple imputation methods based on five replications and the chained equations in the R MI program to impute missing values (Online Resource 4).
In this study, continuous variables were expressed as mean (standard deviation), while categorical variables were expressed as numbers (weighted percentages). P < 0.05 was considered statistically significant. All statistical analyses were performed using Empowerstats (www. empow ersta ts. com) software and R software (version 3.4.3).
The laboratory value units of some variables in this article were not expressed in the metric system. The relevant Système International (SI) was as follows: to convert cholesterol, phosphorus, and calcium to mmol/L, multiply values by 0.0259, 0.323, and 0.25, respectively; to convert protein to g/L, multiply values by 10. Table 1 presents the sociodemographic and medical weighted characteristics of the 2818 participants. Among these participants, 51.78% were male, 15.27% were Mexican American, 7.45% were Other Hispanic, 54.63% were Non-Hispanic White, and 12.72% were Non-Hispanic Black. The weighted mean age was 15.45 (2.25) years. Age, race/ethnicity, educational level, moderate activities, income-to-poverty ratio, total protein, total cholesterol, serum phosphorus, and total BMD all differed significantly across blood mercury quartiles (P < 0.05). Table 2 shows that blood mercury was positively associated with total BMD in model 1 [0.0017 (0.0005, 0.0028)] but negatively, albeit not significantly, in models 2 and 3 [model 2: − 0.0006 (− 0.0015, 0.0003); model 3: − 0.0006 (− 0.0016, In analyses stratified by sex, blood mercury was positively associated with total BMD in model 1 of both males and females. However, the association was insignificant in females [0.0011 (− 0.0006, 0.0028)]. The association of blood mercury with total BMD in model 2 and model 3 of male and female adolescents was negative but insignificant. In analyses stratified by race/ethnicity, blood mercury was positively associated with total BMD in three models of Mexican-American adolescents. However, in all three models, the association was not significant [model 1: 0.0029 (− 0.0001, 0.0058); model 2: 0.0005 (− 0.0019, 0.0029); model 3: 0.0005 (− 0.0020, 0.0031)]. Online Resources 6, 7 and 8 show the distribution of blood mercury levels for males, females, and Mexican Americans stratified by sex and race/ethnicity.

Association of blood mercury with total BMD
The analysis results of the complete dataset obtained by multiple imputations differed very little from the analysis results of the original dataset and did not affect the results of this study (Online Resource 5).
In addition, adjusted smooth curve fittings indicated a nonlinear association of blood mercury with total BMD after stratification by sex or race/ethnicity. In Fig. 2, the smooth curve trends differed for males and females. In males, total BMD Table 2 Association of blood mercury (nmol/L) with total bone mineral density (g/cm 2 ) Model 1, no covariates were adjusted Model 2, sex, age, and race/ethnicity were adjusted Model 3, sex, age, race/ethnicity, body mass index, income to poverty ratio, moderate activities, education level, total protein, total cholesterol, serum phosphorus, and serum calcium were adjusted In the subgroup analysis stratified by sex or race/ethnicity, the model is not adjusted for the stratification variable itself  Fig. 2 Association of blood mercury and total bone mineral density stratified by sex. Age, race/ethnicity, body mass index, income to poverty ratio, moderate activities, education level, total protein, total cholesterol, serum phosphorus, and serum calcium were adjusted increased with blood mercury levels until an inflection point, demonstrating an inverted U-shaped association (inflection point: blood mercury 5.44 nmol/L) (Table 3), while in females, total BMD decreased with blood mercury until the inflection point (inflection point: blood mercury 11.8 nmol/L) (Table 3). An inflection point was also observed in the smooth curve for Mexican Americans (inflection point: 5.49 nmol/L) ( Fig. 3; Table 3). An inverted U-shaped relationship of blood mercury with total BMD was found in male adolescents and Mexican Americans, but a U-shaped relationship in female adolescents.

Discussion
The sample for this study was a nationally representative adolescent population. This study found that blood mercury was negatively associated with total BMD, although not significantly. This negative association was also found in adolescent males and females in stratified analyses. However, a positive, albeit non-significant, association between blood mercury and total BMD was observed in Mexican Americans. In addition, blood mercury and total BMD followed an inverted U-shaped association in male and Mexican-American adolescents but a U-shaped association in female adolescents.
A study on the association of blood mercury with osteoporosis in postmenopausal women from South Korea showed that high blood mercury levels might reduce the risk of osteoporosis in postmenopausal women (Cho et al. 2012). The study suggested that high blood mercury levels may protect bone health in postmenopausal women. However, a previous study of premenopausal women reported that the association of blood mercury with bone mineral density was not significant in premenopausal women (Pollack et al. 2013). The protective effect of high blood mercury levels on bone was also found in men. A cross-sectional study of Korean men over 50 found that high blood mercury was associated with decreased odds of reduced BMD in the femoral neck of Korean men (Kim et al. 2016). However, this effect was not significant in subgroup analysis. A cross-sectional study of adults over 20 observed a positive association of blood mercury with bone mineral density, especially in the femur (Tang et al. 2022). In addition, different results were found in different races. Previous findings were controversial and contradicted our findings. We suspected that gender and race differences might be significant. Previous studies had included Korean and American populations, the latter of which were ethnically diverse. Furthermore, differences between men and women may result in differences in the association between blood mercury and BMD. Therefore, the association of blood mercury levels with BMD remains uncertain, and these clinical studies cannot explore the specific mechanisms of how mercury affects bone metabolism.
We found a possible negative but insignificant association of blood mercury with total BMD in adolescents compared to the abovementioned studies. However, in a stratified analysis, although not significantly, blood mercury was positively associated with total BMD in the Mexican-American population. We have two conjectural explanations for this phenomenon. First, statistically, the small sample size of Mexican Americans appeared responsible for the difference with other races (n = 574, Table 3 Threshold effect analysis of blood mercury level on total bone mineral density using two-piecewise linear regression model Sex, age, race/ethnicity, body mass index, income to poverty ratio, moderate activities, education level, total protein, total cholesterol, serum phosphorus, and serum calcium were adjusted In the subgroup analysis for males, females, and Mexican American, the model is not adjusted for sex or race/ethnicity, respectively Association of blood mercury and total bone mineral density stratified by race/ethnicity. Sex, age, body mass index, income to poverty ratio, moderate activities, education level, total protein, total cholesterol, serum phosphorus, and serum calcium were adjusted weighted percentage 15.27%). Biologically, genetic factors appeared to affect the association of blood mercury with BMD. In addition, it is worth mentioning that the non-linear relationship between blood mercury and total BMD differed between adolescent males and females. The inverted U-shaped curve of blood mercury versus total BMD in adolescent males suggests that blood mercury may be protective of bone within a range of levels but detrimental beyond this range (inflection point: blood mercury 5.44 nmol/L). On the other hand, the U-shaped curve of blood mercury versus total BMD in adolescent females suggests that high levels of blood mercury may have protective effects on bone (inflection point: blood mercury 11.8 nmol/L). This protective effect was also found in perimenopausal women and middle-aged and older men, as previously reported in observational studies (Cho et al. 2012;Kim et al. 2016). However, the distribution of blood mercury levels for females (Online Resource 7) revealed a small proportion of the population with high blood mercury levels (95% percentile: 10 nmol/L), implying that the sample size of an adolescent female with blood mercury above the inflection point was small [blood mercury > 11.8 nmol/L: n = 48 (48/1360)], which might lead to bias. In addition, an inverted U-shaped curve of blood mercury versus total BMD was also discovered in Mexican-American adolescents (inflection point: blood mercury 5.49 nmol/L), indicating that blood mercury was beneficial to Mexican-American bone within a range of levels. However, more prospective studies are required to validate our findings.
In osteoblast-like cell models, the researchers found that extracellular methylmercury and intracellular mercury affect gap junction channels by depleting calcium stores (Schirrmacher et al. 1998). Suzuki et al. found that shortterm mercury exposure in fish scales enhanced osteoclast activity resulting in increased blood calcium (Suzuki et al. 2004). This finding could be one explanation for our findings. In contrast, a mercury exposure experiment in Girella punctata found that short-term mercury exposure reduced osteoclast activity and may be involved in protecting osteoblasts (Yachiguchi et al. 2014). As is well known, estrogen deficiency is associated with an increased risk of osteoporosis and decreased bone mineral density (Xu et al. 2022;Zhu et al. 2021;Riggs 2000). Previous studies have found that mercuric chloride can exert estrogen-like effects to regulate the activity of osteoblasts (Suzuki et al. 2004;Wang et al. 2005;Jin et al. 2002). In addition, a study found that males appear to be more sensitive than females to exposure to methylmercury during early development (Vahter et al. 2002). This is consistent with what our study found in the sex-stratified analysis. However, the physiological mechanism by which mercury affects bone metabolism remains unclear, so more studies are needed to prove our findings.

Limitations
The limitations of this study are as follows: first, this study is cross-sectional, so causality cannot be determined. Second, the laboratory and body measurement data in NHANES were only assessed once, and there were missing data and questionnaires, which all contributed to bias. Third, there are some possible confounding factors in this study that were not adjusted, which might lead to possible bias. Finally, the findings of this study cannot be generalized to other populations because our study population consisted of adolescents aged 12 to 19.

Conclusion
The study found that the association of blood mercury with total BMD in adolescents differed by sex or race. Increased blood mercury levels within a range may benefit bone health in male adolescents (inflection point: 5.44 nmol/L) and Mexican-American adolescents (inflection point: 5.49 nmol/L), while higher blood mercury levels may harm bone health. However, more extensive and prospective research is needed to confirm our findings. informed consent, details of which are available at wwwn.cdc.gov/ nchs/nhanes/.

Consent for publication
The data used in the analysis of this study have been included in this article. The raw data can be obtained from the NHANES official website (wwwn. cdc. gov/ nchs/ nhanes/).

Competing interests
The authors declare no competing interests.