Urinary fluoride and dental fluorosis in relation to kidney and liver function in adolescents and young adults in the United States.

Background:

Higher fluoride in plasma has been associated with lower eGFR among adolescents in the National Health and Nutrition Examination Survey (NHANES); however, whether fluoride exposure may contribute to decreased kidney function or fluoride may accumulate in plasma from lower kidney filtration could not be parsed. We examined the presence of dental fluorosis (DF; reflecting chronic fluoride exposure during tooth development) among adolescents and young adults and urinary fluoride (UF) levels among adolescents in relation to kidney and liver parameters in the United States (US).

Methods:

Participants were 1,031 adolescents (aged 12–19 years) and 734 young adults (aged 20–29 years) from NHANES 2015–2016. DF was derived from Dean’s Fluorosis Index (DFI), and defined as any DFI score indicating very mild, mild, moderate, or severe fluorosis, while scores of normal and questionable were categorized as no fluorosis. Kidney or liver function parameters included estimated glomerular filtration rate (eGFR), serum uric acid, the urinary albumin to creatinine ratio, blood urea nitrogen (BUN), gamma glutamate transferase, aspartate aminotransferase, alkaline phosphatase, alanine aminotransferase, and albumin. We conducted survey-weighted linear regression adjusted for covariates to examine associations between fluoride exposure and kidney or liver function.

Results:

Adolescent participants were 15 years old and adults were 24 years old on average. The median (IQR) UF concentration among adolescents was 0.48 (0.48) mg/L. Approximately 74% of adolescents and 70% of adults had DF with varying degrees of severity (ranging from very mild to severe). Each 1 mg/L increase in UF was associated with an approximately 5 mL/min/1.73 m² lower eGFR among adolescents (B = -4.73, 95% CI: -8.35, -1.12, p=0.010). Higher UF was also associated with higher serum uric acid (mg/dl) (B=0.17, 95% CI: 0.01, 0.33, p=0.040) among adolescents. DF was associated with lower eGFR among adolescents (B = -3.72, 95% CI: -7.10, -0.33, p=0.031) and adults (B = -3.90, 95% CI: -6.49, -1.31, p=0.003). In addition, having DF was negatively associated with BUN among adolescents (B = -0.83, 95% CI: -1.44, -0.22, p=0.007). No other significant associations were observed for liver markers.

Conclusion:

Chronic fluoride exposure during tooth development and recent fluoride exposure in adolescence are cross-sectionally associated with a lower rate of kidney filtration among adolescents and young adults in the US. Prospective US-based studies are needed to determine whether these associations are causal.

Fluoride is widely recognized for its ability to prevent tooth decay [1]. It has been routinely used as an additive in community water supplies since the 1940’s [2] and topical dental treatments since the 1950’s [3]. Aside from water intake, fluoride exposure can occur through consumption of produce grown in fluoride-rich areas and/or with fluoridated water [1], air particles, tea, certain pharmaceuticals and dental products, including toothpaste and mouthwash [4]. Community drinking water provides a major source of fluoride intake in the United States (US) with 62.8% of the population receiving fluoride-supplemented water [5].

The kidneys filter much of the fluoride that is within the blood [6]. While adults excrete approximately half of their consumed fluoride, children can preserve as much as 80% of it in their bones and teeth [1]. Both the kidneys and liver readily accumulate fluoride [7, 8]. However, there remain unanswered questions regarding the potential impact of relatively low fluoride exposures on these vital organs. Chronic high and relatively low fluoride exposure in animal models have caused both kidney and liver damage [9,10,11]. Excess fluoride exposure has also been associated with chronic kidney disease, disruption of kidney enzyme pathways, and reabsorption problems linked to damage to kidney tubules in humans and animals [12]. There is currently a breadth of literature detailing the effects of high fluoride exposure on kidney and liver function [7]; however, many studies have not expanded on potential effects of exposure near the recommended level of 0.7 mg/L in the US population [13].

Malin et al. (2019) [14] found that adolescents in NHANES 2013–2016 who had higher fluoride in their blood plasma had lower estimated glomerular filtration rate (eGFR). However, whether fluoride exposure may contribute to a lower kidney filtration rate or fluoride may accumulate in plasma from lower kidney filtration could not be parsed. Therefore, this study included two different fluoride exposure biomarkers: urinary fluoride (UF) and dental fluorosis (DF), to help determine whether associations of fluoride exposure with kidney or liver function reflect potential effects of fluoride exposure on these outcomes. UF is routinely used as a proxy for relatively recent fluoride exposures [15]. Additionally, since permanent tooth development spans prenatal development up until age 8 [16], DF can provide a proxy for chronic prenatal and childhood fluoride exposure during tooth development (but not adolescent or adult exposures) [17]. This study examined associations of UF and DF with indicators of kidney and liver function among adolescents and adults living in the US.

Participants

Participants were from the 2015–2016 cycle of the National Health and Nutrition Examination Survey (NHANES) as UF was only measured in that cycle. There were 9,455 individuals that attended both an interview and mobile examination center (MEC) visit during which biological samples were collected, and DF was assessed. Of these, 1,031 adolescents (12–19 years old) and 734 adults (20–29 years old) had complete data for fluoride exposure biomarkers, covariates, and at least one kidney/liver parameter. UF data was only available for individuals below 20 years of age. We excluded any individuals with daily protein intake = 0 or >400 gm, as a lack or excess of protein intake can impact kidney and liver function. Five adult participants age 20–29 years with an eGFR, below 60 mL/min/1.73 m² were excluded from the analysis, as these values are indicative of potential kidney dysfunction [14]. We did not include children under 12 years of age because kidney and liver parameters, other than the albumin-creatinine ratio, were only measured for individuals aged 12 years or older (see Fig. 1 for a participant selection flow diagram). The current study was determined to be nonhuman and therefore exempt from review by the University of Florida’s Institutional Review Board (Protocol #: NH00042178).

Participant selection flow diagram

Full size image

Fluoride measures

UF and DF served as fluoride exposure biomarkers. UF concentration was measured in participants aged 12–19 years using an ion-selective electrode (ISE) at the Inorganic and Radiation Analytical Toxicology Branch (IRATB) of the Division of Laboratory Sciences [18]. This method had a lower limit of detection (LLOD) of 0.144 mg/L [18]. DF, the visual changes accompanying the hypomineralization of the outermost layer of the tooth, is associated with excess fluoride exposure during tooth maturation [19]. DF per tooth (in up to 28 teeth) was measured by licensed dentists during the MEC visit using the Dean’s Fluorosis Index (DFI). NHANES examiners did not assess any teeth from the primary dentition, partially erupted and missing teeth, and any teeth with over half of the visible surface area concealed due to dental treatment or disease (other than fluorosis). DFI codes included: normal (DFI = 0), questionable (DFI = 5), very mild (DFI = 1), mild (DFI = 2), moderate (DFI = 3), severe (DFI = 4), non-fluoride opacity (DFI = 8) and if the tooth was missing, not fully erupted, or 1/2 or more of the tooth was replaced with a restoration, covered with orthodontic band, or destroyed by caries and cannot be assessed (DFI = 9) [20]. Questionable was recoded to DFI = 0.5. A binary fluorosis variable was created using the second most affected tooth if the top two most affected teeth were not equal. A “yes” for fluorosis was defined as the second most affected tooth having a non-zero and non-questionable category score while a “no” was defined as the tooth having a DFI score equal to 0 or 0.5 [21]. DFI codes 8 and 9 were considered unmeasurable assessments of fluorosis and were categorized as missing data for the purposes of this study.

Kidney and liver parameters

Kidney parameters include serum uric acid (mg/dl), the urinary albumin to creatinine ratio (mg/g) and estimated glomerular filtration rate (eGFR) (mL/min/1.73m²). Serum liver parameters, include gamma glutamate transferase (U/L), aspartate aminotransferase (U/L), alkaline phosphatase (IU/L), alanine aminotransferase (U/L), and albumin (g/dL). Additionally, blood urea nitrogen (mg/dL), can reflect kidney filtration impairment when elevated, and liver disease or malnutrition when low [22].

For adolescents aged 12–19 years, eGFR was estimated using the CKiD U25 equation [23]:

The CKiD U25 equations were developed for individuals under 25 years of age with mild to moderate chronic kidney disease (CKD) [23]. These equations demonstrate enhanced accuracy in estimating glomerular filtration rate (GFR) within the CKD range across pediatric and young adult populations, thereby offering a more reliable foundation for clinical decision-making [23].

Serum creatinine was measured using the Jaffe rate method (kinetic alkaline picrate), with a calibration traceable to an isotope dilution mass spectrometry (IDMS) reference method [24] and using an enzymatic method [25]. For adult participants aged 20–29 years, we measured eGFR utilizing the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation. This equation is recommended by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) to assess eGFR among adults aged 18 and over [26]. Serum uric acid (SUA; mg/dL) level was ascertained via an endpoint method that was timed [27]. The National Center for Health Statistics (NCHS) calculated the albumin creatinine ratio (ACR; mg/g) as urinary albumin (ug/mL) divided by urinary creatinine (mg/mL), multiplied by 100 and rounded to the nearest hundredth [28]. Albumin concentration in urine was measured using a fluorescent immunoassay method [29]. Urinary creatinine concentration was measured using an enzymatic method [30]. Blood urea nitrogen (BUN; mg/dL) [31, 32] concentration was assessed with an enzymatic conductivity rate method. Aspartate aminotransferase (AST; U/L) concentration was measured using an enzymatic rate method [33] while alanine aminotransferase (ALT; U/L) and alkaline phosphatase (ALP; IU/L) were measured using a kinetic rate method [34, 35]. Serum albumin was measured using timed and digital endpoint methods [36, 37] and gamma-glutamyl transferase (GGT; U/L) was measured via an enzymatic rate method [38].

Covariates

We selected covariates a priori according to their associations with fluoride exposure/metabolism and kidney/liver function [14]. These included age, race/ethnicity (Non-Hispanic White, Non-Hispanic Black, Non-Hispanic Asian, Mexican American, Other Hispanic, Other Race/Multiracial), sex, the ratio of family income to poverty, and body mass index (BMI) utilized as a continuous variable. NHANES calculated BMI as weight in kilograms divided by height in meters squared, and then rounded to one decimal place [39]. NHANES provides total protein intake measured in grams over a 24-hour period. This value is derived from participants’ first day dietary recall data and reflects the cumulative protein consumed from all foods and beverages reported [40]. A 24-hour dietary recall questionnaire was used to evaluate protein intake in all models with kidney/liver outcomes.

Statistical analyses

We conducted univariate analyses of covariates, UF, DF and indicators of kidney and liver function. Covariate-adjusted logistic regression was conducted to examine associations between UF concentrations and DF among a subsample of participants (n = 1,015; weighted N = 27,990,302). Covariate adjusted survey weighted linear regression was conducted to examine associations between fluoride exposure and indicators of kidney and liver function. The assumptions of normality, homoscedasticity, and linearity were met in models evaluating associations between fluoride exposure (UF and DF) and outcomes such as eGFR, SUA, BUN, and serum albumin. For other outcome variables where these assumptions were violated, a natural log transformation was applied to address skewness, including for ACR, ALT, AST, ALP, and GGT. We also adjusted for serum cotinine (ng/mL) in these models in sensitivity analyses. Serum cotinine is a biomarker of nicotine exposure that can impact both kidney/liver function and fluoride metabolism [41, 42]. Serum cotinine was measured using isotope dilution high-performance liquid chromatography/atmospheric pressure chemical ionization tandem mass spectrometry (ID HPLC-APCI MS/MS) [43]. Additionally, we conducted supplemental analyses examining associations between DF and eGFR (calculated with the CKD-EPI equation) among adults aged 18–29 years (rather than 20–29 years) and for adults aged 25 years and over. We also conducted supplemental analyses calculating eGFR using the “Bedside IDMS (isotope dilution mass spectroscopy)-Traceable Schwartz” equation (i.e., the updated Schwartz formula) [44, 45] instead of the CKiD U25 equation in models examining associations of UF or DF with eGFR among adolescents ages 12–19 years. We also applied the CKiD U25 formula for participants under the age of 25 years in supplemental analysis in addition to applying it for adolescents ages 12–19 years in primary analyses. Lastly, we tested interactions of fluoride exposure with sex, the ratio of family income to poverty, race/ethnicity and BMI in all primary models as these variables can modify associations of fluoride exposure with health outcomes [46,47,48,49]. All analyses utilized MEC weights to adjust for oversampling of subgroups, nonresponse, sampling variability, and post-stratification to create a generalizable analysis for the US-based population [50]. For all analyses with kidney and liver parameter outcomes, we reweighted MEC weights to the dietary sample that included protein intake as a covariate using the approach by Malin et al. (2019) [14]. Data was analyzed using STATA MP version 18.5 and R version 4.3.2.

Participants were 15 years old for adolescents and 24 years old for adults on average (see Table 1 for demographic characteristics). There was a similar distribution of males and females. Most participants were non-Hispanic White; approximately 12% of adolescents and 13% of adults were non-Hispanic Black and 14% of adolescents and 12% of adults were Mexican American. The mean (SD) BMI was 24.11 (6.11) kg/m² for adolescents and 27.90 (7.28) kg/m² for adults. For adolescents, the mean (SD) of daily protein intake (gm) was 74.66 (40.39) while it was 88.08 (45.51) for adults (Table 1).

The median (IQR) UF concentration among adolescents was 0.48 (0.48) mg/L. Approximately 70% of participants exhibited very mild or mild DF, whereas fewer than 2% of adolescents and 4% of adults presented with moderate or severe forms. Overall, the prevalence of DF was approximately 74% among adolescents and 70% among adults, with varying degrees of severity (Table 1). Higher UF among adolescents was positively associated with having DF (OR=1.66, 95%CI: 1.08, 2.54, p=0.024). Descriptive statistics for kidney and liver parameters are summarized in Table 2. The median (IQR) eGFR was 99.78 (22.91) mL/min per 1.73 m² for adolescents and 121.29 (21.97) mL/min per 1.73 m² for young adults.

Associations between UF and indicators of kidney and liver function are presented in Table 3.

UF was negatively associated with eGFR. Specifically, each 1 mg/L higher UF among adolescents was associated with a 4.73 mL/min/1.73m² lower eGFR (B = 4.73, 95% CI: -8.35, -1.12, p=0.010). Higher UF was also associated with higher serum uric acid (mg/dl) (B=0.17, 95% CI: 0.01, 0.33, p=0.040). Findings remained consistent after adjusting for serum cotinine (see Table S1). No significant associations were observed between UF and liver function parameters. None of the interactions between UF and sociodemographic variables tested were statistically significant. Associations between UF and eGFR, and DF and eGFR were consistent when the updated Schwartz formula was applied to estimate eGFR (Table S2).

Associations between DF and indicators of kidney and liver function are presented in Table 4.

Having DF was negatively associated with eGFR among adolescents and adults (B = -3.72, 95% CI: -7.10, -0.33, p=0.031) and (B = -3.90, 95% CI: -6.49, -1.31, p=0.003) respectively. In addition, having DF was negatively associated with BUN among adolescents (B = -0.83, 95% CI: -1.44, -0.22, p=0.007). Associations between DF and indicators of kidney and liver function were consistent after adjusting for serum cotinine (Table S3). Associations between DF and eGFR were generally consistent when the CKD-EPI formula was applied for participants aged 18–29 years, when the CKiD U25 formula was utilized to calculate eGFR for participants under aged 25 years, or when the CKD-EPI formula was applied for participants aged 25–29 years only (Table S4 – S6). None of the interactions between DF and sociodemographic variables tested were significant.

This study examined associations of UF levels and presence of DF with indicators of kidney and liver function in a nationally representative US sample of adolescents and young adults. UF levels observed among adolescents in this study were relatively low and typical for US youth [47]. Consistent with findings from other studies, the prevalence of DF (ranging from very mild to severe) was high, ranging from 70% to 74% in adults and adolescents, respectively [51]; although most cases of fluorosis were very mild or mild. Adolescents with higher UF were more likely to have dental fluorosis. Other studies have yielded consistent findings. For example, a case-control study conducted in India observed that children with DF tended to have higher urinary fluoride levels [52]. Similarly, higher plasma and tap water fluoride concentrations have been associated with higher odds of having DF among youth in NHANES 2013–2016 [53].

Higher UF and having DF were associated with changes in kidney and liver function indicators. A 1 mg/L higher UF concentration in adolescents was associated with an approximately 5 mL/min per 1.73 m² lower eGFR. Furthermore, having DF was associated with an approximately 4 mL/min per 1.73 m² lower eGFR among adolescents and adults. Findings were generally consistent after adjusting for serum cotinine, when eGFR was calculated using the updated Schwartz equation among adolescents, or when different age subgroups were examined among adolescents and adults. While determining causality is limited in cross-sectional studies, Malin et al. (2019) [14] also observed that adolescents in NHANES 2013–2016 who had more fluoride in their blood plasma tended to have lower eGFR. If worse kidney function was contributing to an accumulation of fluoride in the blood, we would expect lower eGFR (i.e., lower kidney filtration rate) to be associated with less fluoride excreted into the urine. However, we observed the opposite, such that adolescents with higher UF concentrations tended to have lower eGFR. This provides potential evidence that fluoride exposure may be contributing to lower kidney filtration rate. Moreover, given that DF reflects excess fluoride exposure during prenatal and childhood tooth development, findings of this study may also prospectively suggest that excess fluoride exposure in utero or during childhood contributes to lower kidney filtration rate by adolescence or even young adulthood. However, more research is needed to determine whether these associations are causal or whether magnitudes of association observed in this study may reflect clinically meaningful reductions in eGFR.

Consistent with this study, prior studies have also found that higher UF concentration and/or DF were associated with lower kidney filtration rate [52, 54]. For example, Ando et al. (2001) [54] observed that youth aged 10–15 years old in fluorosis areas in China had lower urinary inorganic phosphate along with higher UF when compared to people in non-fluorosis areas in China and Japan. Furthermore, a study of over 1600 adults in China found that higher UF levels were cross-sectionally associated with higher SUA, AST and BUN [55]. Additionally, a study by Khandare et al. (2017) [52] found that 8–15 year old children residing in endemic fluorosis areas in India had higher serum ALP and creatinine along with considerable reductions in GFR, reflecting kidney damage. However, a Mexico-based cross-sectional study of 374 children aged 5–12 years found that higher UF was associated with higher eGFR [56]. Furthermore, a prospective study of 438 children aged 8–12 years in Mexico City found no association of UF with eGFR; although they observed a trend of an association of higher UF with lower eGFR among children with high adiposity [46]. Both studies included different age ranges, smaller sample sizes, and different measurement techniques for UF detection and eGFR calculation which could have contributed to these differences in findings. Specifically, unlike the current study, both used specific gravity-adjusted UF concentrations and calculated eGFR using the Creatinine-Cystatin C-Based CKiD Equation or the 2012 CKiD Cystatin C-based Equation [57, 58].

We also observed lower BUN among adolescents with DF. While increased BUN could indicate kidney filtration impairment, certain dietary conditions, and other disorders, low BUN may suggest malnutrition or liver disease [22]. None of the associations between fluoride exposure and other liver parameters were statistically significant. The significant associations observed for kidney parameters as opposed to most liver parameters may be because fluoride is renally cleared and accumulates more in the kidneys than the liver or any other organ system [6]. Consistently, animal studies show that fluoride exposure can contribute to kidney damage, even at low levels [59, 60]. Potential mechanisms for fluoride-induced kidney damage include increased oxidative stress, inflammation and mitochondrial dysfunction [61].

There are several strengths of the current study. It utilized nationally representative data with a large sample size and survey weights making it generalizable to the US population of adolescents and young adults. Additionally, by employing both UF during adolescence and DF measurements in adolescence and adulthood, we examined recent low-level fluoride exposures and chronic fluoride exposures from in utero development and childhood, respectively. Further, we adjusted for multiple variables that can impact fluoride exposure/metabolism and kidney/liver function in our models. Additionally, we utilized objective measures of kidney and liver function. However, this study has some limitations. First, a limited number of participants (n = 5) with an eGFR below 60 mL/min/1.73 m², suggestive of chronic kidney disease, were excluded from the final analysis. Therefore, we could not assess the clinical relevance of the study findings in participants with possible kidney disease. Older-aged adults and those more at risk of developing kidney disease should be studied in the future to better understand the clinical significance of these results. Second, NHANES did not provide urinary specific gravity measurements for the 2015–2016 cycle, and therefore, we could not adjust for dilution which can influence fluoride concentration measurements. UF can also be influenced by dietary patterns throughout the day, so a single spot urine sample may not capture typical daily fluoride exposure levels. Nevertheless, exposure misclassification would be more likely to bias associations toward the null rather than to create a spurious finding. Future studies employing 24-hour UF measurements, or morning fasting samples, would better account for daily fluctuations in an individual’s diet and drinking patterns. Third, geocoded data in NHANES are restricted from public access. Therefore, we were unable to consider whether associations of UF or DF with kidney or liver function differed according to geographic location. Lastly, in 2015, the US Public Health Service (USPHS) lowered the recommended fluoride concentration in community drinking water from a range of 0.7–1.2 mg/L to 0.7 mg/L [13]. While community water systems generally fell close to the updated USPHS recommendations between 2016 and 2021 [62], some locations may have taken longer to adapt to these changes. As such, UF data may partially reflect exposure to fluoride in community drinking water at the previously recommended range. Furthermore, DF data would reflect chronic prenatal and childhood fluoride exposure at that previous range too [13]. Future research on fluoride exposure and kidney and liver function should include individuals that consume tap water fluoridated at the current recommended level.

Chronic fluoride exposure during tooth development and recent fluoride exposure in adolescence are cross-sectionally associated with a lower rate of kidney filtration among adolescents and young adults in the US. Prospective studies should address whether these associations may be causal.

Data are available in a public, open access repository. All data is publicly available at [https://www.cdc.gov/nchs/nhanes].

NHANES:

National Health and Nutrition Examination Survey

UF:

Urinary fluoride

DF:

Dental fluorosis

US:

United States

eGFR:

Estimated glomerular filtration rate

CKD-EPI:

Chronic Kidney Disease Epidemiology Collaboration

IDMS:

Isotope dilution mass spectroscopy

ISE:

Ion-selective electrode

IRATB:

Inorganic and Radiation Analytical Toxicology Branch

LLOD:

lower limit of detection

DFI:

Dean’s Fluorosis Index

SUA:

Serum uric acid

ACR:

Albumin creatinine ratio

BUN:

Blood urea nitrogen

GGT:

Gamma glutamate transferase

AST:

Aspartate aminotransferase

ALP:

Alkaline phosphatase

ALT:

Alanine aminotransferase

NCHS:

The National Center for Health Statistics

BMI:

Body mass index

We thank the participants and staff of the NHANES 2015–2016 cycle for making this research possible. We also thank the nephrotoxicology expert who provided consultation for this study.

This work was supported in part by funding from the National Institute of Environmental Health Sciences (NIH/NIEHS): R00ES031676. The funder had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

Authors and Affiliations

1. College of Public Health and Health Professions, University of Florida, Gainesville, FL, 32611, USA

   Durdana Khan, Alexandra Mattia,  Zhilin Wang & Ashley J. Malin

2. College of Medicine, University of Florida, Gainesville, FL, 32610, USA

   Durdana Khan,  Zhilin Wang & Ashley J. Malin

Contributions

D.K: Writing—original draft, methodology, investigation, data curation, formal analysis, and writing—review & editing. A.M: Writing—original draft, methodology, investigation, data curation, formal analysis, and writing—review & editing. Z.W: Methodology, investigation, formal analysis. A.J.M: Writing—original draft, writing—review & editing, supervision, funding acquisition, resources, methodology, conceptualization.

Corresponding author

Correspondence to Ashley J. Malin.

Ethics approval and consent to participate

This study was determined to be nonhuman and therefore exempt from review by the University of Florida’s Institutional Review Board (Protocol #: NH00042178).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.