Research Studies
Study Tracker
Association between relatively low fluoride exposure and bone mineral density in children and adolescents: Insights from NHANES and a wistar rat model of fluorosis.
| Ecotoxicology and Environmental Safety | Zhang R, Luo M, Wu L, Yan S, Li A, Mo Z, Zafar G, Wu D, Li X, Gao N, Wang X, Pei J. | 311:119828.
Children NHANES Bone Density Bones/Joints Serum-F Skeletal Fluorosis Urine-F Animal Study Human Study
View PDF

Abstract

Full-text study online at
https://www.sciencedirect.com/science/article/pii/S0147651326001570?via%3Dihub

Highlights

  • Integrated NHANES and animal studies of fluoride–BMD in children/adolescents.
  • Multi-region analysis identified extremities and pelvis as most fluoride-sensitive.
  • Linear, nonlinear, trend tests, and subgroup analyses for intergrated insights.

This study employed an integrated approach combining a population-based study with animal experiments to evaluate the effects of fluoride exposure on bone mineral density (BMD) in children and adolescents. Population data were obtained from the National Health and Nutrition Examination Survey (NHANES) 2015–2016 cycle, and a Wistar rat model of fluorosis was established for the animal study. Fluoride concentrations in drinking water, blood, and urine were measured using the ion-selective electrode method. BMD in the population was measured using dual-energy X-ray absorptiometry (DXA), whereas the BMD in rats was measured using digital X-ray imaging. Multivariate linear regression and restricted cubic spline (RCS) models were used to assess the association between fluoride exposure and BMD. Results from the population study showed that water fluoride concentration was inversely associated with BMD in the left arm and left leg. Additionally, urinary fluoride concentration was inversely associated with BMD in the left leg, right arm, and right leg. The RCS model further revealed a nonlinear association between water fluoride concentration and BMD in the left arm. In the animal experiments, urinary fluoride concentration was inversely associated with BMD in the left arm, right leg, and pelvis. Additionally, the RCS model indicated nonlinear associations between urinary fluoride concentration and BMD across the left arm, right leg, and pelvis. This study suggests that fluoride exposure is associated with BMD in multiple skeletal regions in children and adolescents.

    Keywords

    Fluoride; Bone mineral density; Children; Adolescents

    1. Introduction

    Bone mineral density (BMD) is a key indicator for assessing skeletal health (Wei et al., 2021). Adolescence is a critical period for bone mass accumulation, during which BMD not only preserves the structural integrity of the skeleton and supports bone health, but also acts as a protective barrier for vital organs, while laying the foundation for the body’s long-term calcium storage (Sun et al., 2023). The most immediate consequence of reduced BMD is the onset of osteoporosis, which often leads to fractures, severely compromising an individual’s health and quality of life (Liu et al., 2024). A variety of factors, including genetics, nutrition, lifestyle behaviors, insufficient physical activity, medication use, and metabolic factors, have been identified as contributors to reduced BMD (Abood et al., 2020, Qu et al., 2021, Bragança et al., 2023, Cao et al., 2024, Hansen et al., 2014, Yoo et al., 2021). In addition to these established risk factors, emerging evidence suggests that exposure to certain environmental toxins may also be associated with lower BMD, with excessive fluoride intake being a notable example.

    Fluorine (F), the most electronegative and chemically reactive halogen in the periodic table, is widely distributed in nature in various forms due to its unique physicochemical properties (Wang et al., 2020). Consequently, humans can be exposed to fluoride through multiple environmental sources, including drinking water, food, and air. Prolonged exposure to excessive fluoride can result in fluorosis (Yadav et al., 2019). It is estimated that fluorosis affects approximately 260 million people across more than 20 countries, posing a significant public health challenge (Wang et al., 2021). Fluorosis can adversely affect multiple organ systems, including the skeletal (Dey Bhowmik et al., 2023), renal (Zheng et al., 2024), digestive (Zhao et al., 2022), endocrine (Lu et al., 2022), nervous (Zhao et al., 2024), immune (Qiao et al., 2024), cardiovascular (Yang et al., 2024), and reproductive systems (Cheng et al., 2024). Among these, bone tissue serves as the primary organ for fluoride accumulation and is particularly vulnerable to its toxic effects (Li et al., 2020). Excessive fluoride intake can adversely affect osteoblasts, osteoclasts, cartilage tissue, and bone mineralization, ultimately leading to the development of skeletal fluorosis, which is characterized by osteosclerosis, periosteal soft tissue calcification, osteomalacia, osteoporosis, and degenerative changes in joints and cartilage (Qiao et al., 2021).

    A cross-sectional study conducted in Kaifeng City, China, found that low-to-moderate fluoride exposure was associated with lower BMD in adult women aged 18–60 years, possibly through RUNX2 promoter methylation. In this population, among women with urinary fluoride concentrations exceeding 1.6 mg/L, BMD was negatively associated with urinary fluoride concentration (B: -0.14; 95 % CI: -0.26, -0.01) (Gao et al., 2020). Similarly, a cross-sectional study of 722 women in rural Henan Province found that BMD was negatively associated with fluoride exposure in women aged 50–54 years (B: -0.063; 95 % CI: -0.129, -0.002) (Sun et al., 2020). Furthermore, a prospective cohort study in Sweden among postmenopausal women reported that high consumption of drinking water with fluoride concentrations below 1 mg/L was associated with an increased risk of fractures, and this risk was positively correlated with both dietary fluoride intake and urinary fluoride concentration. When comparing the lowest and highest quartiles of urinary and dietary fluoride, the multivariable-adjusted hazard ratios for fracture incidence were 1.50 (95 % CI: 1.04, 2.17) and 1.59 (95 % CI: 1.10, 2.30), respectively (Helte et al., 2021). The epidemiological evidence suggests that fluoride exposure, even at low-to-moderate doses, may be associated with reduced BMD in adults. However, these studies were conducted primarily in adult women.

    The skeleton undergoes continuous growth and mineral accrual during childhood and adolescence until peak bone mass is attained (Cao et al., 2022). During this critical period of rapid skeletal development, growth processes are highly sensitive to external perturbations, and disruptions in endochondral ossification or bone remodeling may impair normal bone formation, thereby increasing the risk of skeletal structural defects (Fan et al., 2011). Furthermore, children and adolescents are particularly vulnerable to the adverse effects of fluoride exposure (Yousefi et al., 2018). This underscores the need to shift research focus from adults to children and adolescents.

    The U.S. Environmental Protection Agency (EPA) has established a maximum contaminant level (MCL) of 4.0 mg/L for fluoride in drinking water to prevent skeletal fluorosis (Wang et al., 2019). Exposures above this threshold have been shown to significantly disrupt bone metabolism. Given that children and adolescents are particularly sensitive to fluoride and BMD during skeletal development is highly susceptible to adverse influences, it is critical to determine whether relatively low-dose fluoride exposure (below the MCL) affects BMD in different skeletal regions in this population. Accordingly, the present study evaluates the association between fluoride exposure and BMD in children and adolescents using data from NHANES with subsequent validation in a rat model of fluorosis.

    2. Materials and methods

    2.1. Population study

    2.1.1. Participants

    This study utilized data from the National Health and Nutrition Examination Survey (NHANES) from the 2015–2016 cycle. NHANES, a program conducted by the Centers for Disease Control and Prevention, is designed to assess the health and nutrition status of a nationally representative, noninstitutionalized sample of people of all ages living in the United States. The survey incorporated data obtained from questionnaires, in-home interviews, and physical examinations at mobile examination centers where blood and urine samples were collected. The NHANES protocol was approved by the National Center for Health Statistics (NCHS) Ethics Review Board, and all participants provided written informed consent. A total of 9971 participants were included in the 2015–2016 cycle of the NHANES. Among these participants, plasma fluoride concentration was measured in 2145 individuals. Of these, BMD measurements in different skeletal regions-including the head, left and right arms, left and right legs, left and right ribs, thoracic spine, lumbar spine, pelvis, trunk, subtotal (total body minus head), and total body-were available for 1564 participants. After excluding 139 participants with missing data on body mass index (BMI) or the ratio of family income to poverty, 1425 participants were included in the final analysis of plasma fluoride concentration. Drinking water fluoride concentration was measured in 3987 participants, with complete BMD data available for 1729 individuals. After excluding 153 participants with missing BMI or ratio of family income to poverty, 1576 participants were included in the final analysis of drinking water fluoride concentration. Urinary fluoride concentration was measured in 2408 participants, and complete BMD data were available for 1717 individuals. After excluding 160 participants with missing BMI or ratio of family income to poverty and one participant with an extremely high urinary fluoride concentration, the final analytical sample for urinary fluoride concentration included 1556 participants. The participant selection process is shown in Figure S1.

    2.1.2. Fluoride measurement in the population

    Fluoride concentrations in urine samples were measured using an ion-selective electrode (ISE). Urine samples were processed, stored, and shipped to the Laboratory Sciences Division of the National Center for Environmental Health at the Centers for Disease Control and Prevention in Atlanta, Georgia, for subsequent analysis. The limit of detection (LOD) for urinary fluoride was established at 0.144 mg/L, and values below the LOD were substituted with LOD/?2. Plasma fluoride concentrations were measured using an ion-selective electrode and the hexamethyldisiloxane (HMDS) method. Fluoride concentrations in household water samples were similarly measured using the ion-selective electrode method. Both plasma and water samples were processed, stored, and shipped to the College of Dentistry at the University of Georgia in Augusta, Georgia, for analysis. The LODs determined for plasma fluoride and water fluoride were 0.25 nmol and 0.10 mg/L, respectively; values below these thresholds were also replaced with LOD/?2 as appropriate.

    2.1.3. BMD measurement in the population

    Whole body scans were acquired on the Hologic Discovery Model A densitometers (Hologic, Inc., Bedford, Massachusetts), using software version APEX 3.2. All data were subsequently analyzed with Hologic APEX 4.0 software.

    2.1.4. Covariates

    The covariates included in the analysis were age, gender, BMI, race, and the ratio of family income to poverty. The ratio of family income to poverty was calculated by dividing the annual household income by the specific poverty guideline applicable for the survey year.

    2.1.5. Statistical analyses

    All analyses applied survey weights from the mobile examination center (MEC) visit to account for the clustered sample design, survey non-response, over-sampling, post-stratification, and sampling error. This approach ensured that the sample more accurately reflected the characteristics of the U.S. general population. Statistical analyses were performed using R software (version 4.4.1). Baseline characteristics were summarized according to variable type. Continuous variables were described using means and standard deviations (SDs), whereas categorical variables were presented as numbers (percentages). Participants were categorized into quartiles based on water fluoride, plasma fluoride, and urinary fluoride concentrations. For continuous variables, univariate analysis of variance (ANOVA) was used to assess between-group differences, while chi-square (X²) tests were used for categorical variables. Weighted linear regression models were then employed to examine the associations between fluoride exposure and BMD in different skeletal regions. For skeletal regions showing statistically significant associations between fluoride exposure and BMD, trend tests were performed by modeling the median value of each quartile as a continuous variable in the regression models. Subgroup analyses stratified by age, gender, and race were conducted to further explore the associations between fluoride exposure and BMD. The results are shown as coefficients (B) and 95 % confidence intervals (95 % CIs). Finally, weighted restricted cubic spline (RCS) models were applied to assess the nonlinear association between water fluoride, plasma fluoride, and urinary fluoride concentration and BMD. All regression analyses were adjusted for age, gender, BMI, race, and the ratio of family income to poverty. Two-sided P values  < 0.05 were considered statistically significant.

    2.2. Animal experiment

    2.2.1. Animal experimental design

    A total of 150 SPF-grade Wistar rats, aged 3 weeks, purchased from Beijing Viton Lihua Laboratory Animal Technology Co., Ltd. (License No.: SCXK (Beijing) 2016–0011), were used as experimental subjects. The rats were randomly assigned to groups based on the concentration and duration of fluoride exposure. Initially, all rats were divided into three groups based on exposure duration: 1 month, 3 months, or 6 months. Within each duration group, the rats were further randomized into five subgroups based on fluoride concentration, with each subgroup consisting of ten rats and equal numbers of males and females. The control group drank distilled water, whereas the fluoride-treated groups drank fluoridated water with 10, 25, 50, and 100 mg/L of fluoride ion, and the corresponding concentrations of sodium fluoride (NaF) were 22, 55, 110, and 220 mg/L, respectively. All animals were housed in an SPF-grade animal facility at Harbin Medical University. Environmental conditions were maintained at a temperature of 23 ± 3°C with a relative humidity range of 40–70 %, and a 12-hour light-dark cycle was employed. The rats were provided with standard growth and reproduction chow and allowed to eat and drink ad libitum during the experimental period. The study protocol and animal procedures were approved by the Animal Ethical Committee of Harbin Medical University (approval No. hrbmuecdc20220204).

    2.2.2. Fluoride measurement in rats

    Blood samples were collected and allowed to stand at room temperature for 30 min. Subsequently, the samples were centrifuged at 1000 rpm for 10 min, and the pale yellow serum layer was separated for analysis. Fluoride concentrations in the serum were measured using the fluoride ion-selective electrode method according to the standard method (WS/T 212–2001, China). Before euthanasia, 24-hour urine samples were collected from rats housed in metabolic cages. The samples were centrifuged at 3000 rpm for 5 min, and the supernatant was collected for analysis. Urinary fluoride concentrations were measured using the fluoride ion-specific electrode method (WS/T 89–2015, China).

    2.2.3. BMD measurement in rats

    Animal X-ray imaging was performed using an SR-1000S system (Shantou Ultrasonic Instrument Research Institute Co., Ltd., China). Image analysis was conducted using ImageJ software (version 1.54p). A 1 mm diameter steel wire and a carbon filament were used as calibration standards. X-ray images were saved in DICOM format and standardized using ImageJ. The images were inverted at a resolution of 2304 × 2800 pixels and subsequently converted to 8-bit TIFF format, allowing for 256 grayscale levels. In the processed images, the grayscale values corresponding to the carbon standard and steel wire were set to 0 and 255, respectively, rendering the carbon standard invisible and the steel wire standard fully white. All rat X-ray images were processed using the same workflow, and grayscale values were used as surrogate indicators for BMD. Regions of interest (ROIs) were defined according to the criteria used in the NHANES BMD measurements. Whole-body BMD measurements included background values, whereas other skeletal regions required manual ROI delineation to minimize background interference. To reduce variability associated with manual selection, each ROI was delineated three times, and the average grayscale value was used as the surrogate BMD value for that region. As grayscale values were used in place of absolute BMD measurements, all results are reported in Grayscale Value (GV).

    2.2.4. Statistical analyses

    Statistical analyses of the animal experiments were performed using SPSS (version 27.0), and graphical representations were generated using GraphPad Prism (version 9.0). The Shapiro-Wilk test was used to assess the normality of all datasets. For normally distributed data, generalized linear models were used to estimate the main effects and interactions of time and dose. For non-normally distributed data, stratified analyses and nonparametric tests were used to assess the main effects of time and dose, with pairwise comparisons executed using the Kruskal-Wallis H test. Linear regression models were employed to assess the associations between serum fluoride and urinary fluoride concentrations and BMD in different skeletal regions. For skeletal regions showing statistically significant associations between fluoride exposure and BMD, trend tests were performed by modeling the median value of each quartile as a continuous variable in the regression models. Subgroup analyses stratified by age and sex were conducted to further explore the associations between fluoride exposure and BMD. RCS models were employed to assess nonlinear associations between serum fluoride and urinary fluoride concentrations and BMD. All regression analyses were adjusted for age and gender. Two-sided P values < 0.05 were considered statistically significant.

    3. Results

    3.1. Population study results

    3.1.1. Baseline characteristics

    Demographic characteristics and BMD parameters of all participants are presented in Table S1. Table S2 shows demographic characteristics and BMD parameters of the study population according to quartiles of water fluoride concentration. Statistically significant differences across quartiles were observed in BMI, left arm BMD, left leg BMD, right arm BMD, right ribs BMD, thoracic spine BMD, trunk bone BMD, subtotal BMD, and total BMD. Table S3 shows demographic characteristics and BMD parameters according to quartiles of plasma fluoride concentration. Statistically significant differences were observed in left arm BMD, left leg BMD, right arm BMD, right leg BMD, right ribs BMD, thoracic spine BMD, lumbar spine BMD, trunk bone BMD, subtotal BMD, and total BMD across quartiles. Table S4 shows the demographic characteristics and BMD parameters according to quartiles of urinary fluoride concentration. Statistically significant differences were observed in gender and race distribution across quartiles.

    3.1.2. Multiple linear regression models assessing the association between fluoride exposure and BMD in different skeletal regions in the population

    Associations between fluoride exposure and BMD in different skeletal regions are shown in Fig. 1. In weighted multivariable linear regression models, water fluoride concentration was significantly inversely associated with BMD in the left arm and left leg. Specifically, each 1 mg/L increase in water fluoride concentration was associated with a decrease of 0.03 g/cm² in the left arm BMD and 0.07 g/cm² in the left leg BMD (Fig. 1A). No significant associations were observed between plasma fluoride concentration and BMD in any skeletal region (Fig. 1B). Urinary fluoride concentration was significantly inversely associated with BMD in the left leg, right arm, and right leg. Each 1 mg/L increase in urinary fluoride concentration was associated with decreases of 0.014 g/cm² in the left leg BMD and 0.011 g/cm² and 0.014 g/cm² in the right arm and right leg BMD, respectively (Fig. 1C).

    Fig. 1

    1. Download: Download high-res image (341KB)
    2. Download: Download full-size image

    Fig. 1. Forest plot of the associations between fluoride exposure and BMD in different skeletal regions in the population. (A) Associations between water fluoride concentration and BMD in different skeletal regions. (B) Associations between plasma fluoride concentration and BMD in different skeletal regions. (C) Associations between urinary fluoride concentration and BMD in different skeletal regions. B, regression coefficient; 95 % CI, 95 % confidence interval.

    .

    3.1.3. Trend tests and subgroup analyses assessing the association between fluoride exposure and BMD in the population

    For skeletal regions in which statistically significant associations between fluoride exposure and BMD were observed, trend tests and subgroup analyses stratified by age, gender, and race were performed to further assess the associations between fluoride exposure and BMD. The results for water fluoride and urinary fluoride are presented in Table 1 and 2, respectively.

    Table 1. Association between water fluoride concentration and BMD in the population.

    Exposure Left Arm BMD
    B (95 % CI), P value    		 Left Leg BMD
    B (95 % CI), P value
    Quartiles of water fluoride (mg/L)
    Quartile 1 [0.07–0.15] 0(ref.) 0(ref.)
    Quartile 2 (0.15–0.36] -0.006(-0.028,0.016) -0.011(-0.034,0.012)
    Quartile 3 (0.36–0.67] -0.024(-0.050,0.002) -0.047(-0.098,0.004)
    Quartile 4 (0.67–2.68] -0.028(-0.050,-0.006)** -0.069(-0.128,-0.009) *
    P for trend 0.029* 0.027*
    Stratified by gender
    Female -0.020(-0.045,0.006) -0.033(-0.100,0.034)
    Male -0.031(-0.058,-0.004)* -0.092(-0.158,-0.026)*
    Stratified by age
    Children (8–11 years) -0.014(-0.043,0.015) -0.062(-0.132,0.008)
    Adolescents (12–19 years) -0.039(-0.069,-0.009)* -0.065(-0.129,-0.001)*
    Stratified by race
    Mexican American -0.001(-0.019,0.016) -0.006(-0.044,0.033)
    Non-Hispanic Black -0.010(-0.048,0.029) -0.029( – 0.102,0.044)
    Non-Hispanic White -0.054(-0.085,-0.023)** -0.123(-0.197,-0.049)**
    Other Hispanic 0.005(-0.038,0.048) 0.011(-0.084,0.107)
    Other Race – Including Multi-Racial -0.043(-0.108,0.021) -0.115(-0.256,0.025)
    In the subgroup analysis, the analysis was not adjusted for the stratification variable itself. *P < 0.05, **P < 0.01, ***P < 0.001. B, regression coefficient; 95 % CI, 95 % confidence interval.

    Table 2. Association between urinary fluoride concentration and BMD in the population.

    Exposure Left Leg BMD
    B (95 % CI), P value    		 Right Arm BMD
    B (95 % CI), P value    		 Right Leg BMD
    B (95 % CI), P value
    Quartiles of urine fluoride (mg/L)
    Quartile 1 [0.102–0.306] 0(ref.) 0(ref.) 0(ref.)
    Quartile 2 (0.306–0.524] -0.011(-0.033,0.011) -0.003(-0.018,0.012) -0.010(-0.032,0.013)
    Quartile 3 (0.524–0.799] -0.011(-0.043,0.021) -0.009(-0.028,0.010) -0.015(-0.048,0.018)
    Quartile 4 (0.799–4.448] -0.022(-0.043,-0.000)* -0.012(-0.028,0.004) -0.019(-0.043,0.006)
    P for trend 0.032 * 0.077 0.077
    Stratified by gender
    Female -0.004(-0.027,0.018) -0.011(-0.031,0.010) -0.005(-0.030,0.020)
    Male -0.020(-0.041,0.001) -0.010(-0.023,0.004) -0.018(-0.038,0.002)
    Stratified by age
    Children (8–11 years) -0.017(-0.040,0.006) -0.011(-0.029,0.006) -0.017(-0.040,0.007)
    Adolescents (12–19 years) 0.002(-0.018,0.022) -0.001(-0.013,0.011) 0.003(-0.017,0.023)
    Stratified by race
    Mexican American 0.007(-0.040,0.055) 0.006(-0.015,0.028) 0.007(-0.036,0.050)
    Non-Hispanic Black -0.004(-0.033,0.024) 0.003(-0.014,0.020) -0.009(-0.038,0.019)
    Non-Hispanic White -0.027(-0.055,0.002) -0.021(-0.043,0.001) -0.023(-0.050,0.004)
    Other Hispanic 0.010(-0.053,0.073) 0.001(-0.032,0.035) 0.008(-0.048,0.063)
    Other Race – Including Multi-Racial -0.023(-0.087,0.042) -0.011(-0.046,0.024) -0.021(-0.074,0.033)

    In the subgroup analysis, the analysis was not adjusted for the stratification variable itself. *P < 0.05, **P < 0.01, ***P < 0.001. B, regression coefficient; 95 % CI, 95 % confidence interval.

    Compared with participants in the lowest quartile of water fluoride concentration, those in the highest quartile exhibited significant inverse associations between water fluoride concentration and both left arm BMD and left leg BMD, and the P values for trend were also statistically significant. When the lowest quartile of urinary fluoride concentration was used as a reference, a significant inverse association was observed between urinary fluoride concentration and left leg BMD in the highest quartile, with a statistically significant P value for trend.

    In subgroup analyses stratified by gender, age, and race, significant inverse associations between water fluoride concentration and BMD were observed among males, adolescents aged 12–19 years, and non-Hispanic whites for both left arm and left leg BMD.

    3.1.4. Nonlinear association between fluoride exposure and BMD in the population

    RCS models were applied to assess nonlinear associations between water fluoride concentration and BMD in the left arm and left leg, as well as between urinary fluoride concentration and BMD in the left leg, right arm, and right leg. The results are presented in Fig. 2. As shown in Fig. 2A, a significant nonlinear association was observed between water fluoride concentration and BMD in the left arm. An inflection point was identified at 0.68 mg/L. Below this threshold, BMD in the left arm decreased with increasing water fluoride concentration, whereas above 0.68 mg/L, the rate of decrease was attenuated.

    Fig. 2

    1. Download: Download high-res image (325KB)
    2. Download: Download full-size image

    Fig. 2. Nonlinear association between fluoride exposure and BMD in the population. (A) Nonlinear association between water fluoride concentration and left arm BMD. (B) Nonlinear association between water fluoride concentration and left leg BMD. (C) Nonlinear association between urinary fluoride concentration and left leg BMD. (D) Nonlinear association between urinary fluoride concentration and right arm BMD. (E) Nonlinear association between urinary fluoride concentration and right leg BMD.

    3.2. Animal experiment results

    3.2.1. Fluoride exposure in rats

    Serum fluoride concentrations are shown in Figure S2A. Statistically significant differences in serum fluoride concentrations were observed across the 1-, 3-, and 6-month time points within each fluoride exposure group (H = 13.304, P = 0.001; H = 9.413, P = 0.009; H = 19.134, P < 0.001; H = 15.243, P < 0.001; H = 24.368, P < 0.001). In the control group, serum fluoride concentrations at 3 months were significantly higher than those at 1 month. In the 10 mg/L fluoride exposure group, serum fluoride concentrations at 6 months were higher than those at 1 month. In the 25, 50, and 100 mg/L fluoride exposure groups, serum fluoride concentrations at both 3 months and 6 months were significantly higher than those at 1 month, exhibiting a gradual time-dependent increase, which was particularly pronounced in the 50 mg/L and 100 mg/L groups. Compared with the control group, a significant increase in serum fluoride concentrations was observed at 1 month with increasing fluoride exposure levels. For both the 3-month and 6-month measurements, except for the 25 mg/L exposure group, serum fluoride concentrations increased significantly with escalating exposure levels (H = 38.587, P < 0.001; H = 31.518, P < 0.001; H = 41.042, P < 0.001).

    The results for urinary fluoride concentration are presented in Figure S2B. No statistically significant differences in urinary fluoride concentrations were observed over time within the control group. However, significant differences were observed across the 1-, 3-, and 6-month time points in the 10, 25, 50, and 100 mg/L fluoride exposure groups. (H = 2.395, P = 0.302; H = 18.866, P < 0.001; H = 6.991, P = 0.030; H = 24.103, P < 0.001; H = 9.822, P = 0.007). In the 10 and 25 mg/L fluoride exposure groups, urinary fluoride concentrations at both 3 and 6 months were significantly higher than those at 1 month. Similarly, in the 50 mg/L fluoride exposure group, urinary fluoride concentrations at both 3 and 6 months were significantly higher than those at 1 month; however, concentrations at 6 months were lower than at 3 months. In the 100 mg/L fluoride exposure group, urinary fluoride concentrations at 3 months were significantly higher than those at 1 month. Compared with the control group, urinary fluoride concentrations significantly increased at all time points (1 month, 3 months, and 6 months) as fluoride exposure levels increased, demonstrating a clear dose-dependent upward trend (H = 42.840, P < 0.001; H = 40.347, P < 0.001; H = 43.371, P < 0.001).

    3.2.2. BMD in different skeletal regions in rats

    The definitions of skeletal regions of rats are presented in Figure S3 (total BMD was derived from the entire X-ray image and therefore is not presented separately). Figure S4 presents the effect of fluoride exposure dose and duration on BMD in different skeletal regions without gender stratification. Figure S5 and Figure S6 present the effects of fluoride exposure dose and duration on BMD in different regions in female and male rats, respectively. In 1-month-old rats, significant differences in BMD across fluoride exposure groups were observed in different skeletal regions when analyses were not stratified by gender. Notably, except for the left and right arms, the 50 mg/L fluoride exposure group exhibited significantly lower BMD than the other dose groups in most regions. In contrast, BMD in the 100 mg/L fluoride exposure group increased, although BMD in the left and right arms remained significantly lower than in the other dose groups. In 3-month-old rats, no statistically significant differences in BMD across fluoride exposure groups were observed in different skeletal regions, except for total BMD. In 6-month-old rats, the 25 mg/L fluoride exposure group consistently exhibited the highest BMD across regions, whereas the 50 mg/L group showed the lowest BMD, with statistically significant differences observed in several skeletal regions. In female rats, the lowest BMD values were generally observed in the 50 mg/L and 100 mg/L fluoride exposure groups at 1 month. At 3 months, the 100 mg/L group exhibited the highest BMD across most regions, whereas at 6 months, the 50 mg/L group showed the lowest BMD, and the 25 mg/L and 100 mg/L groups exhibited the highest BMD across regions. In male rats, the 50 mg/L and 100 mg/L fluoride exposure groups exhibited the lowest BMD across most regions at 3 and 6 months, while the 25 mg/L group displayed the highest BMD at 6 months. When not stratified by gender, regional BMD within each fluoride exposure group generally exhibited either an initial increase followed by a decrease or a gradual increase with prolonged fluoride exposure, with the exception of total BMD. These patterns were broadly consistent between male and female rats.

    3.2.3. Multiple linear regression models assessing the association between fluoride exposure and BMD in different skeletal regions in rats

    Associations between fluoride exposure and BMD in different skeletal regions are shown in Fig. 3. In multivariable linear regression models, no significant associations were observed between serum fluoride concentration and BMD in any skeletal region (Fig. 3A). In contrast, urinary fluoride concentration was significantly inversely associated with BMD in the left arm, right leg, and pelvis. Specifically, each 1 mg/L increase in urinary fluoride concentration was associated with a decrease of 0.139 GV in the left arm BMD, 0.092 GV in the right leg BMD, and 0.102 GV in the pelvis BMD, as shown in Fig. 3B.

    Fig. 3

    1. Download: Download high-res image (386KB)
    2. Download: Download full-size image

    Fig. 3. Forest plot of the associations between fluoride exposure and BMD in different skeletal regions in rats. (A) Associations between serum fluoride and BMD in different skeletal regions. (B) Associations between urinary fluoride and BMD in different skeletal regions. B, regression coefficient; 95 % CI, 95 % confidence interval.

    3.2.4. Trend tests and subgroup analyses assessing the association between fluoride exposure and BMD in rats

    For skeletal regions in which statistically significant associations between fluoride exposure and BMD were observed, trend tests and subgroup analyses stratified by age and sex were conducted to further assess the association between fluoride exposure and BMD. The results are presented in Table 3.

    Table 3. Association between urinary fluoride concentration and BMD in rats.

    Exposure Left Arm BMD
    B (95 % CI), P value    		 Right Leg BMD
    B (95 % CI), P value    		 Pelvis BMD
    B (95 % CI), P value
    Quartiles of urine fluoride (mg/L)
    Quartile 1 [0.419–4.898] 0(ref.) 0(ref.) 0(ref.)
    Quartile 2 (4.898–12.874] -7.298(-12.777,-1.819)* -2.020(-7.781,3.742) -2.077(-7.512,3.358)
    Quartile 3 (12.874–38.006] -10.552(-16.073,-5.031)*** -8.787(-14.593,-2.982)** -10.169(-15.646,-4.692)***
    Quartile 4 (38.006–96.553] -11.399(-16.872,-5.927)*** -6.861(-12.615,-1.107)* -6.542(-11.970,-1.114)*
    P for trend < 0.001*** 0.024* 0.027 *
    Stratified by sex
    Female -0.013(-0.148,0.122) 0.074(-0.066,0.214) 0.051(-0.078,0.179)
    Male -0.217(-0.334,-0.099)*** -0.196(-0.314,-0.077)** -0.196(-0.313,-0.079)**
    Stratified by age
    1month -0.458(-0.696,-0.219)*** -0.230(-0.430,-0.030)* -0.138(-0.351,0.075)
    3month -0.039(-0.150,0.072) -0.012(-0.132,0.109) -0.045(-0.160,0.069)
    6month -0.218(-0.394,-0.043)*** -0.232(-0.447,-0.017)* -0.236(-0.433,-0.039)*

    In the subgroup analysis, the analysis was not adjusted for the stratification variable itself. *P < 0.05, *0.01, ***P < 0.001. B, regression coefficient; 95 % CI, 95 % confidence interval.

    Using the lowest quartile of urinary fluoride concentration as the reference group, the second, third, and highest quartiles of urinary fluoride concentration were significantly inversely associated with left arm BMD, with a statistically significant P value for trend. In addition, the third and highest quartiles of urinary fluoride concentration showed significant inverse associations with BMD in the right leg and pelvis, with a statistically significant P value for trend.

    In subgroup analyses stratified by gender and sex, significant inverse associations between urinary fluoride concentration and BMD of the left arm, right leg, and pelvis were observed in male rats. In both 1-month-old and 6-month-old rats, urinary fluoride concentration was significantly inversely associated with BMD in the left arm and right leg. Furthermore, in 6-month-old rats, a significant inverse association was observed between urinary fluoride concentration and pelvis BMD.

    3.2.5. Nonlinear association between fluoride exposure and BMD in rats

    The nonlinear association between urinary fluoride concentration and BMD in rats is shown in Fig. 4. A statistically significant nonlinear association was observed between urinary fluoride concentration and BMD in the left arm, right leg, and pelvis. An inflection point was identified at a urinary fluoride concentration of 38 mg/L, BMD in the left arm, right leg, and pelvis decreased with increasing urinary fluoride below this threshold. Conversely, when urinary fluoride concentration exceeded 38 mg/L, an increase in BMD was observed in these regions.

    Fig. 4

    1. Download: Download high-res image (156KB)
    2. Download: Download full-size image

    Fig. 4. Nonlinear association between fluoride exposure and BMD in rats. (A) Nonlinear association between urinary fluoride concentration and left arm BMD in rats. (B) Nonlinear association between urinary fluoride concentration and right leg BMD in rats. (C) Nonlinear association between urinary fluoride concentration and pelvis BMD in rats.

    4. Discussion

    This study employed an integrated approach combining the population-based study and animal experiments to systematically evaluate the association between fluoride exposure and BMD in children and adolescents. In the population study, significant inverse associations were observed between water fluoride concentration and BMD in the left arm and left leg. In addition, urinary fluoride concentration was significantly inversely associated with BMD in the left leg, right arm, and right leg. A nonlinear association was observed between water fluoride concentration and BMD in the left arm, characterized by an initial steep decline followed by a more gradual decrease. In the Wistar rat model of fluorosis, urinary fluoride concentration showed significant inverse associations with BMD in the left arm, right leg, and pelvis. Furthermore, a nonlinear association was observed between urinary fluoride concentration and BMD in the left arm, right leg, and pelvis, exhibiting an initial decrease followed by an increase. These findings suggest that fluoride exposure may adversely affect bone metabolism in children and adolescents, with particular sensitivity observed in the arms, legs, and pelvis.

    In this study, we observed both significant linear and nonlinear associations between drinking water fluoride concentration, urinary fluoride concentration, and BMD in specific skeletal regions. In contrast, no significant associations were observed between plasma or serum fluoride concentrations and BMD. Previous studies have demonstrated that the half-life of fluoride in blood is remarkably short, typically lasting only a few hours (Adkins et al., 2022, Singh et al., 2025). This characteristic suggests that blood fluoride concentrations are strongly influenced by short-term or recent intake, resulting in substantial variability in single measurements and limiting their ability to accurately reflect a long-term fluoride exposure (Helte et al., 2021). As a result, drinking water concentration and urinary fluoride concentration may serve as more reliable biomarkers for assessing fluoride exposure than plasma or serum concentrations. Drinking water fluoride concentration directly reflects environmental exposure sources and generally remains relatively stable over time (Chen et al., 2013). Additionally, approximately 75 % of the body’s total fluoride burden is excreted through the kidneys as urinary fluoride, making urinary fluoride a consistent and reliable biomarker for evaluating overall fluoride exposure status (Deng et al., 2025).

    Results from weighted linear regression models adjusted for relevant covariates in the population study indicated that for each 1 mg/L increase in water fluoride concentration, BMD decreases by 0.03 g/cm² in the left arm and by 0.07 g/cm² in the left leg. Additionally, for each 1 mg/L increase in urinary fluoride concentration, BMD decreased by 0.014 g/cm² in the left leg, 0.011 g/cm² in the right arm, and 0.014 g/cm² in the right leg. In animal experiments, we further explored the effects of fluoride exposure duration and dose on BMD across different skeletal regions in rats. Among 1-month-old rats, without stratification by gender, BMD in all regions exhibited an overall decreasing trend as fluoride exposure duration increased. Moreover, results from multivariate linear regression analysis adjusted for covariates demonstrated significant inverse associations between urinary fluoride concentration and BMD. Specifically, each 1 mg/L increase in urinary fluoride concentration was associated with decreases of 0.139 GV in the left arm BMD, 0.092 GV in the right leg BMD, and 0.102 GV in the pelvis BMD. A cross-sectional case-control study conducted among adults aged 20 years and older in Lüliang, Shanxi Province, China, identified urinary fluoride exposure and the G allele mutation of the ADAMTS14_rs4747096 gene as potential risk factors for reduced BMD, with a significant interaction between these two factors (Qin et al., 2023). Additionally, a cross-sectional study involving 907 farmers aged 18–60 in Tongxu County, China, reported that excessive fluoride intake was associated with decreased BMD in adults, with this association partially mediated by reductions in total antioxidant capacity (T-AOC) and mitochondrial DNA copy number (mtDNAcn) (Ba et al., 2024). These findings are in agreement with the results from our study.

    To further explore the association between fluoride exposure and BMD in the population study, subgroup analyses were conducted. The results revealed a significant inverse association between water fluoride concentration and BMD in the left arm and left leg in males, adolescents aged 12–19 years, and non-Hispanic whites. Although no statistically significant association was observed between urinary fluoride concentration and overall BMD across subgroups, the inverse associations between urinary fluoride concentration and BMD were more pronounced in the left leg, right arm, and right leg in males, children aged 8–11 years, and non-Hispanic whites. In the rat model, when analyses were not stratified by sex, no statistically significant differences in BMD were observed across skeletal regions in 3-month-old rats. However, among the 3-month-old male rats exposed to 50 mg/L and 100  mg/L fluoride exposure groups, the lowest BMD was observed in all regions. In contrast, in 6-month-old male rats, BMD in most skeletal regions was lowest in the 50 mg/L and 100 mg/L fluoride groups. Further analyses revealed a significant inverse association between urinary fluoride concentration and BMD in male rats. In both 1-month-old and 6-month-old rats, urinary fluoride concentrations were strongly inversely associated with BMD in the left arm and the right leg. Additionally, in 6-month-old rats, a significant inverse association was observed between urinary fluoride concentration and pelvis BMD.

    These findings suggest that fluoride exposure is associated with a more pronounced reduction in BMD in males. This phenomenon may be attributed to the effects of estrogen produced in female organisms. Research indicates that estrogen binds to estrogen receptors, promoting the expression of osteoprotegerin (OPG) while inhibiting the action of receptor activator of nuclear factor kappa-B ligand (RANKL). This interaction suppresses osteoclast formation and bone resorption activity (Cheng et al., 2022). Additionally, estrogen activates the Wnt/B-catenin signaling pathway, enhancing osteogenesis and upregulating the bone morphogenetic protein (BMP) signaling pathway, which promotes the differentiation of mesenchymal stem cells from pre-osteoblasts to osteoblasts (Wang et al., 2023). Fluoride-induced activation of osteoblasts synergizes with estrogen-mediated responses to encourage bone formation over resorption. In contrast, male organisms, lacking sufficient levels of estrogen, experience significant bone loss due to enhanced osteoclast activation that far exceeds any increase in bone formation induced by osteoblasts.

    We also observed that the impact of fluoride exposure on BMD varied across different age groups in the population. To further explore these age-related differences, an animal model was established to investigate the effects of fluoride exposure on BMD at distinct developmental stages. The actual human ages corresponding to 1-, 3-, and 6-month-old rats are approximately 10, 14, and 18 years old, respectively, which align with childhood and adolescence (Sengupta, 2013; Zhang et al., 2021; Liu et al., 2019). Our results demonstrated significant inverse associations between urinary fluoride concentration and BMD in both the left arm and right leg of 1- and 6-month-old rats. Additionally, in 6-month-old rats, a significant inverse association was observed between urinary fluoride concentration and BMD in the pelvis. However, no statistically significant association was observed in any skeletal region of 3-month-old rats. This discrepancy may be explained by the fact that 3 months corresponds to a critical period of sexual maturity in rats, during which growth hormone (GH), insulin-like growth factor-1 (IGF-1), and sex hormones (such as estrogen and testosterone) peak (Zych et al., 2019, Katic et al., 2024). These hormones are potent regulators of bone formation and play essential roles in maintaining BMD (Dixit et al., 2021, Vico and Vanacker, 2010). The enhanced endogenous growth-promoting environment during this period may temporarily mitigate the metabolic disturbances caused by fluoride exposure, resulting in a protective window where no immediate association between fluoride exposure and BMD is apparent.

    The impact of fluoride exposure on BMD exhibits variability across different race groups. The adverse effects of fluoride on BMD are particularly pronounced among non-Hispanic whites, which may be influenced by genetic polymorphisms that differ between ethnicities. A systematic review suggests that genetic polymorphisms in the COL1A2 (Collagen type 1 alpha 2), CTR (Calcitonin receptor gene), ESR (Estrogen receptor), COMT (Catechol-o-methyltransferase), GSTP1 (Glutathione S-transferase pi 1), MMP-2 (Matrix metallopeptidase 2), PRL (Prolactin), VDR (vitamin D receptor), and MPO (Myeloperoxidase) genes are associated with varying patterns of susceptibility to different types of fluorosis among individuals residing in the same community and exposed to similar environmental factors (Pramanik and Saha, 2017).

    We also employed RCS models to evaluate the nonlinear association between fluoride exposure and BMD. In the population study, a significant nonlinear association was observed between water fluoride concentration and left arm BMD, with an identified threshold at 0.68 mg/L. When the water fluoride concentration was below 0.68 mg/L, left arm BMD decreased with increasing water fluoride concentration. Conversely, when the water fluoride concentration exceeded 0.68 mg/L, the rate of BMD decline became attenuated. In the animal study, below the threshold of 38 mg/L, increases in urinary fluoride concentration were associated with decreases in BMD across these regions: left arm, right leg, and pelvis. In contrast, when urinary fluoride concentrations exceeded 38 mg/L, BMD exhibited an increasing trend. This phenomenon may be partly explained by the bidirectional effects of fluoride exposure on osteoclasts. In a previous study conducted by our group, sixty male C57BL/6 mice randomly assigned into three groups treated with deionized water containing NaF at concentrations of 0 mg/L, 50 mg/L, and 100 mg/L for three months revealed an inverted U-shaped relationship between levels of fluorine exposure and osteoclast activity; higher doses of fluorine slightly reduced osteoclast formation while concurrently increasing overall BMD (Yao et al., 2019). In addition, fluoride exposure has also been shown to increase osteoblast activity. In one study, 36 female Sprague – Dawley rats were treated with different concentrations of NaF (0, 55, 110, and 221 mg/L) for 3 months, and fluoride was found to promote hypoxia-inducible factor-1B signaling. This in turn triggers autophagy and activation of the canonical Wnt/B-catenin signaling pathway, which ultimately leads to abnormal activation of rat osteoblasts (Zhu et al., 2022). Therefore, we speculate that the decrease in BMD before the turning point may be due to the dominant role of bone resorption, while after the turning point, bone formation may play a dominant role.

    In both population-based study and animal experiments, we observed significant associations between fluoride exposure and BMD in regions of the extremities. The heterogeneous effects of fluoride on BMD across skeletal regions may be partly attributable to differences in bone composition. The skeleton of the extremities is predominantly composed of cortical bone, whereas the skull, ribs, and vertebrae contain a higher proportion of trabecular bone (Hoshi-Numahata et al., 2023, Iaquinta et al., 2019). Compared with cortical bone–rich regions, trabecular bone–rich regions exhibit higher bone turnover rates (Fredericson et al., 2023). This higher turnover may partially offset fluoride-induced interference with bone mineralization, resulting in less pronounced changes in BMD in trabecular bone–rich regions.

    The strengths of this study are as follows: First, it integrates a population-based study with animal experiments to systematically evaluate the impact of fluoride exposure on BMD in children and adolescents, providing insights into the more pronounced adverse effects on BMD in males and non-Hispanic whites. Second, by assessing the association between low fluoride exposure and BMD across different skeletal regions, the study identifies the extremities and pelvis as the most sensitive regions to fluoride-induced bone alterations.

    This study has several limitations. First, the cross-sectional nature of the NHANES component inherently limits causal inference. Because the cross-sectional study assesses exposure and outcome at a single time point, it is unable to establish the temporal order between fluoride exposure and alterations in BMD (Chua et al., 2022). The cumulative effects of fluoride within the body and the dynamic changes of BMD cannot be fully captured by measurements obtained at a single time point (Zhang et al., 2025). Although our animal experiments complemented the population-based findings by establishing Wistar rat models of fluorosis with different exposure durations, a fully longitudinal animal experimental design was not implemented due to considerations of animal welfare, practical feasibility, and time and resource constraints. Therefore, future prospective cohort studies are warranted to further validate the effects of fluoride exposure on BMD in children and adolescents. Second, the fluoride concentrations used in the animal experiments were higher than those observed in the NHANES population. This was primarily based on interspecies differences in fluoride pharmacokinetics and the study’s objective of systematically characterizing dose-response patterns through low-, medium-, and high-dose groups. Due to differences in exposure patterns and anatomy/physiology, the animal findings should be interpreted as complementary mechanistic evidence supporting the associations observed in the NHANES analysis, rather than as a basis for direct quantitative extrapolation to low-dose human exposure (Lee et al., 2013). Third, subjective errors inherent in manually delineating BMD regions during animal studies cannot be eliminated. Even when a single operator averaged three manual selections of the same area, human variability persisted due to “ambiguous criteria for identifying bone tissue boundaries” (e.g., distinguishing between trabecular structures and soft tissue). Finally, given the substantial amount of missing data for dietary calcium, vitamin D, and protein, including these variables as covariates would have markedly reduced the analyzable sample size and compromised statistical power. Therefore, these dietary factors were not included as covariates in our models. However, incomplete covariate adjustment may result in residual confounding, potentially affecting the interpretation of the true association between fluoride exposure and BMD.

    5. Conclusion

    In the population-based study, we observed significant inverse associations between water fluoride concentration and BMD in both the left arm and left leg. Additionally, urinary fluoride concentration was significantly inversely associated with BMD in the left leg, right arm, and right leg. A nonlinear association was also observed between water fluoride concentration and BMD in the left arm. In animal experiments, urinary fluoride concentration exhibited significant inverse associations with BMD in the left arm, right leg, and pelvis. Moreover, urinary fluoride showed nonlinear associations with BMD in the left arm, right leg, and pelvis. These findings suggest that fluoride exposure may adversely affect bone metabolism in children and adolescents, with the arms, legs, and pelvis being particularly sensitive to these effects.

    CRediT authorship contribution statement

    Shirui Yan: Validation, Methodology. Lei Wu: Validation, Methodology. Minghan Luo: Validation, Methodology, Investigation, Data curation. Junrui Pei: Writing – review & editing, Writing – original draft, Supervision, Resources, Project administration. Rui Zhang: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Methodology, Formal analysis. Ailin Li: Validation, Methodology. Gazala Zafar: Validation, Methodology. Zhe Mo: Validation, Methodology. Xiaowei Wang: Validation, Methodology. Nian Gao: Validation, Methodology. Xinxiao Li: Validation, Methodology. Di Wu: Validation, Methodology.

    Funding

    This research was funded by the National Natural Science Foundation of China [82273749] and [82473747].

    Declaration of Competing Interest

    The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

    Appendix A. Supplementary material

    Data availability

    Data will be made available on request.

    References