Early childhood fluoride exposure and preadolescent kidney function

Abstract

Background:

Early-life renal maturation is susceptible to nephrotoxic environmental chemicals. Given the widespread consumption of fluoride and the global obesity epidemic, our main aim was to determine whether childhood fluoride exposure adversely affects kidney function in preadolescence, and if adiposity status modifies this association.

Methods:

Our study included 438 children from the PROGRESS cohort. Urinary fluoride (uF) was assessed at age 4 by diffusion analysis; outcomes studied included estimated glomerular filtration rate (eGFR), blood urea nitrogen (BUN), selected kidney proteins and blood pressure measured at age 8–12 years. We modeled the relationship between uF and outcomes, and adjusted for body mass index (BMI), age, sex, and socioeconomic status.

Results:

The median uF concentration was 0.67 ug/mL. We observed null associations between 4-year uF and preadolescent eGFR, although effect estimates were in the expected inverse direction. A single unit increase in ln-transformed uF was associated with a 2.2 mL/min decrease in cystatin C-based eGFR (95% CI: −5.8, 1.4; p = 0.23). We observed no evidence of sex-specific effects or effect modification by BMI status. Although uF was not associated with BMI, among children with obesity, we observed an inverse association (ß: −4.8; 95% CI: −10.2, 0.6; p = 0.08) between uF and eGFR.

Conclusions:

Low-level fluoride exposure in early childhood was not associated with renal function in preadolescence. However, given the adverse outcomes of chronic fluoride consumption it is possible that the preadolescent age was too young to observe any effects. Longitudinal follow-up in this cohort and others is an important next step.

Introduction

Fluoride is a naturally occurring ion that can be ingested through diet, drinking water, supplements or dental products¹. Fluoride has been widely used as a tooth decay preventative², however the dose must be titrated as excess fluoride can lead to dental or skeletal fluorosis^3,4 (i.e., dental mottling, osteoporosis, sclerosis and osteomalacia), as well as other systemic health effects, including subclinical liver and kidney damage⁵. The primary source of fluoride exposure varies in different countries. In the US and Canada, fluoride is added to drinking water distribution systems, and the primary exposure is via drinking water^6,7. However, in Mexico, fluoride is naturally occurring in some areas⁸, and is not added to drinking water distribution systems as a public health measure. In Mexico City (the location of the cohort in the study herein) the primary fluoride source is mainly from fluorinated salt in the diet⁹.

Fluoride is both a tubular and glomerular nephrotoxicant affecting both the reabsorption and filtration systems of the kidney^10–13. Low-level fluoride nephrotoxicity is characterized by oxidative stress resulting in disrupted calcium metabolism, glucose and ion transport, and water homeostasis^13–19. Additionally, fluoride can induce mitochondrial damage²⁰, which in turn contributes to chronic kidney disease (CKD) and metabolic disorders²¹. Fluoride nephrotoxicity is characterized by a reduced ability to concentrate urine as well as increased N-acetyl β-d-glucosaminidase (NAG) and kidney injury molecule- 1 (KIM-1) excretion^6,22. Recent cross-sectional studies of children have shown evidence of potential fluoride nephrotoxicity characterized by elevated levels of KIM-1 or altered estimated glomerular filtration rate (eGFR)^23,24.

Additionally, preadolescence is a critical stage for both kidney function and risk of overweight and obesity. While early life fluoride exposure may alter kidney function, the subsequent onset of obesity-induced compensatory kidney hypertrophy can compound nephron damage and accelerate functional decline²⁵. Thus, early life nephrotoxic exposures and child adiposity are risk factors for CKD that may act synergistically via complementary pathways to impair adolescent kidney function.

Fluoride exposure often co-occurs with heavy [or trace] metals exposure, and fluoride additives may contain metal contaminants such as lead (Pb)²⁶; however, the early-life impacts of fluoride and Pb co-exposures are poorly understood. Fluoride-metal co-exposures are particularly relevant given that animal models demonstrate that perinatal exposure to fluoride leads to reduced ability to concentrate urine²⁷, and we hypothesize that this potentiates dehydration stress and potential nephrotoxicant effects; lead is a predominant heavy metal exposure in Mexico City²⁸.

Few studies have examined the renal consequences of children’s environmental fluoride exposure, and prior studies were cross-sectional or case-control^23,24,29. Furthermore, possible effect modification by overweight and obesity of these associations has not previously been explored. This study aimed to assess longitudinal associations of early childhood fluoride exposure with preadolescent eGFR, BUN, select urine proteins and blood pressure. We assessed potential effect modification of fluoride and kidney outcomes by sex or adiposity status. Lastly, we examined potential interaction effects of fluoride as a co-exposure with concurrent blood lead levels (BLLs).

Methods

Our study included mother-child pairs participating in the Programming Research in Obesity, Growth, Environment and Social Stressors (PROGRESS) study. PROGRESS is a prospective, ongoing, NIH-funded birth cohort study in Mexico City. Pregnant women receiving care through the Mexican Social Security System (IMSS) were enrolled between July 2007 and February 2011. Healthy women were eligible for enrollment if they were 18 years or older, pregnant at fewer than 20 weeks’ gestation, had access to a telephone, and planned to reside in Mexico City for the following three years. Women were excluded if they had heart or kidney disease, used steroids or anti-epilepsy drugs, or consumed alcohol on a daily basis. All participants provided informed and written consent. The study protocols were approved by the institutional review boards of the Icahn School of Medicine at Mount Sinai and the National Institute of Public Health in Mexico.

Mother-child pairs were followed after delivery and every 6 months until children were 2 years old, and at biannual visits since then. Of 948 enrolled women who delivered a live birth, 581 preadolescents attended the 8–12 year-old visit, when fasting blood samples were collected in BD Vacutainer tubes by a trained phlebotomist, and serum was separated according to the standard protocol. During the 4-year-old visit, children provided a urine sample that was immediately aliquoted. Serum and urine samples were stored at −80°C until subsequent laboratory analyses. The study participants for this analysis were restricted to the 438 children who had complete measures of 8–12 year-old serum cystatin C, 4-year urine fluoride concentrations, and covariates of interest.

Urinary Fluoride:

Urine samples were collected under non-fasting conditions and not standardized with respect to collection time. Child urine fluoride (CUF) concentrations were analyzed at the Oral Health Research Institute at The University of Indiana using diffusion analysis.^30,31 This methodology employs a Thermo Scientific Orion 9609BNWP combination fluoride electrode for which the lower limit of detection is 0.02 mg/L. To account for variations in urine dilution, CUF concentrations were adjusted for specific gravity (SG) using the following equation: CUF_SG (mg/L) = CUFi * (SGm-1)/(SGi-1) where UF_SG (mg/L) is the SG adjusted fluoride concentration, CUFi is the observed fluoride concentration, SGi is the SG of the individual urine sample and SGm is the median SG for the sample³².

Blood Lead Measurements:

Children’s venous blood was collected at the 4-year visit. All blood specimens were drawn in trace metal free tubes, refrigerated at 4°C until shipment to the laboratory where they were frozen at −20°C until analysis. Lead concentration was measured by external calibration using the Agilent 8800 ICP Triple Quad (Agilent, Santa Clara, CA) in MS/MS mode in the trace metals laboratory at the Icahn School of Medicine at Mount Sinai. The limit of detection was <0.2 μg/dL and the instrument precision (given as %RSD) was approximately 5%. Blinded quality control samples obtained from the Maternal and Child Health Bureau and the Wisconsin State Laboratory of Hygiene Cooperative Blood Lead Proficiency Testing Program showed good precision and accuracy.

Quality control (QC) was implemented by analyzing and verifying the initial calibration; continuing calibration verification standards; NIST traceable mixed-element standard solution at two concentration levels, and procedural blanks. 2% of samples were also subject to repeat analysis.

Serum Cystatin C and Creatinine Measurements:

Cystatin C was measured using the Quantikine Enzyme-Linked Immunosorbent Assay (ELISA) Human Cystatin C Immunoassay (R&D Systems, Minneapolis, MN). Optical density was measured by a microplate reader (Synergy HT, BioTek Instruments, Inc., Winooski, VT) set to 450 nm. The limit of detection was 31 pg/ml. Creatinine was measured using Creatinine FS reagent and Respons 910 (both by DiaSys, Holzheim, Germany), based on Jaffe’s kinetic method without deproteinization, with a detection limit of 0.2 mg/dL. To study alternate methods of adjustment for urine concentration, a subgroup analysis of 74 participants with urine creatinine was performed instead of urine specific gravity.

eGFR and BUN Calculations:

eGFR was calculated using creatinine or cystatin C according to previously validated formulas for children. These included the Schwartz equation appropriate for Jaffe-based creatinine measurements, wherein eGFR_Schwartz = (K*height in cm) / (SCr in mg/dL), where K =0.7 for adolescent boys, and 0.55 for adolescent girls, and children (male or female) <13 years old)³³, and the 2012 CKiD cystatin C-based equation, wherein eGFR_CysC2012 = 70.69*(Cystatin C^)−0.931 and cystatin C is in mg/L. As cystatin C-based eGFR is preferrable for children and adolescents³⁴, the analyses presented in this manuscript use eGFR_CysC2012. Analyses with eGFR_Schwartz are available in supplemental tables. We calculated blood urea nitrogen (BUN) with the following formula: serum urea (mg/dL)/2.14; serum urea was determined through the Urease - GLDH enzymatic UV test.

Urine Protein Biomarkers

Urinary protein concentrations of a1-microglobulin, epidermal growth factor (EGF), osteopontin (OPN), neutrophil gelatinase-associated lipocalin (NGAL), albumin, renin, kidney injury molecule-1 (KIM-1), and clusterin were determined from spot urine samples. Protein concentrations were assessed at the Mount Sinai Hospital using a Luminex-multiplex system. Concentrations were determined using an ELISA panel by Milliplex xMAP technology (EMD Millipore, Billerica, MA) on Luminex-200 multiplex immunoassay system.

Blood pressure

Children’s resting BP was measured using an automated oscillometric device (Ambulatory BP 90207 monitor, Spacelabs Healthcare, Snoqualmie, WA) as previously described³⁵. After 3–5 minutes of rest, BP was measured using the automated Spacelabs system with a sized cuff, taking three measurements. We used average of the 3 measurements in analyses.

Covariates:

Child weight and standing height were measured with a professional digital scale and stadiometer (InBody230, InBody, Seoul, Korea) with the head in the Frankfort plane. Weight and height were used to calculate BMI categories according to the WHO z-scores for age and sex, defined as underweight (z-score ≤ −2), normal weight (−2 > z-score ≥ 1), overweight (1 < z-score ≤ 2) and obesity (z-score > 2)³⁶. Information on socioeconomic status (SES) and environmental tobacco exposure were collected at the 2^nd trimester visit using a standardized questionnaire. Household environmental tobacco smoke exposure was dichotomized as yes/no based on the mother’s report that at least one household member smoked during the pregnancy. The SES index during pregnancy was calculated based on the 1994 Mexican Association of Market Intelligence and Public Opinion Agencies (AMAI) rule 13 * 6. The index classifies families into 6 levels based on 13 questions related to the characteristics of the household. The majority of families were middle or low SES. Thus, the 6 resultant levels were further collapsed into 3 SES categories: lower, medium, and higher. Gestational age at delivery was calculated using maternal report of last menstrual period (LMP) and confirmed with Capurro physical examination at birth.³⁷ When the physical examination assessment of gestational age differed by more than 3 weeks from the gestational age based on LMP, the physical exam was used in lieu of the LMP-determined gestational age.

Statistical Analysis:

We assessed the distribution of relevant variables for normality. Urine fluoride and blood lead levels had a non-normal distribution and were therefore transformed via natural logarithm. Four values of specific gravity-adjusted urinary fluoride data that were below the limit of detection (LOD) were replaced with the value of the LOD divided by the square root of two. Potentially influential data points were assessed using Cook’s Distance. We used multiple linear regression models to assess the association between urinary fluoride and eGFR, cystatin C, serum creatinine, BUN, blood pressure, and selected urinary proteins at age 8–12 years in separate models. Covariates in adjusted models included child age, sex, maternal SES, and child BMI z-score selected a priori. For urine protein outcomes, models also included specific gravity. We tested for sex-specific effects via assessing interaction in a linear regression model, as well as in stratified models. Similarly, we assessed potential effect modification using interaction terms in the full model (BMI X Fluoride) as well as in models stratified by categorical BMI status. Additional analyses tested for an interaction between 4-year BLL and urinary fluoride exposure.

Children born preterm or low birth weight are at increased risk of reduced kidney function³⁸, therefore we performed two sensitivity analyses that excluded participants who were: 1) preterm (gestational age less than 37 weeks) or 2) low birth weight (less than 2.5 kg). We also conducted a sensitivity analysis additionally adjusted for prenatal indoor smoke exposure as a covariate.

Results

Demographic characteristics of the 438 preadolescents with complete data at 8–12 years of age are shown in Table 1. The average age was 10 years and ranged between 8 and 12 years. Half of the children (50%) were boys. A majority of participants were lower SES. Approximately half were normal weight, and 25% and 24% were considered overweight or obese respectively. Demographic characteristics of the preadolescents in this analysis did not differ significantly from the 581 preadolescents who participated in the study visit. The median urinary fluoride concentration was 0.67 ug/mL, a level consistent with prior reports of relatively low-level fluoride exposure in Mexico City.³⁹ The average serum creatinine was 0.43 mg/dL and the median eGFR_Schwartz was 174 mL/min/1.73m²; levels were within the expected physiological ranges for this age⁴⁰. According to eGFR_Schwartz, no participants had an eGFR below 60 mL/min/1.73m² (according to the cystatin C-based equation, eGFR_CysC2012, four children had values <60 mL/min/1.73m²). The average BLL assessed at 4 years of age was 1.7 ug/dL. Approximately 8% of child BLLs at age 4 exceeded the CDC guideline level of 5 ug/dL. Urinary fluoride concentrations were associated with child age (B=0.1, p=0.04), and we observed no associations with child sex, BMI Z-score, prenatal tobacco smoke, SES or cross-sectional BLLs.

Table 1:

Participant demographics for n=438 children in the PROGRESS cohort.

	Arithmetic mean (SD) or median [25th and 75th percentiles]
Age, years	9.6 (0.66)
Male, %	50.5%
Prenatal SES
Lower (n= 225)	51.4%
Medium (n = 169)	38.6%
Higher (n = 34)	10.0%
Prenatal exposure to secondhand smoke	30.1%
Preadolescent BMI
Normal Weight (n = 226)	51.6%
Overweight (n = 109)	24.9%
Obese (n = 103)	23.5%
Urine fluoride at age 4 (SG-adjusted, ug/ml)	0.67 [0.51, 0.86]
Blood Lead Levels at age 4 (ug/dL)	1.7 [1.3, 2.5]
Preadolescent Kidney Parameters
Cystatin C (mg/L)	0.71 [0.61, 0.84]
BUN (mg/dL) (n=355)	12.2 (3.1)
Creatinine (Serum) (mg/dL) (n=356)	0.43 (0.08)
eGFRSchwartz (mL/min/1.73m2) (n=356)	174 [152, 204]
eGFRCysC2012 (mL/min/1.73m2)	97 [84, 112]
Systolic Blood Pressure (mmHg)	111.9 (10.5)
Diastolic Blood Pressure (mmHg)	70.1 (7.8)

SD, standard deviation; IQR, interquartile range; BMI, body mass index; BUN, blood urea nitrogen; eGFR_CysC2012, estimated glomerular filtration rate from 2012 cystatin C equation.

Associations between SG-adjusted urinary fluoride at age 4 years and preadolescent eGFR

Estimates for the association between urine fluoride concentrations at age 4 and preadolescent eGFR_CysC2012 were in the expected inverse direction but did not reach statistical significance (Table 3). Among all participants, a single unit increase in ln-transformed urine fluoride was associated with a 2.2-unit decrease in eGFR_CysC2012 that did not reach statistical significance (95% CI: −5.8, 1.4; p = 0.23). We did not observe evidence of sex-specific associations (p for interaction = 0.82). Similarly, we did not observe differences in sex-stratified models, (boys: ß: −2.2, 95% CI: −6.5, 2.1; p = 0.3; girls: ß: −2.5, 95% CI: −9.0, 4.1; p = 0.5).

Table 3.

Associations between 4-year fluoride and eGFR_CysC2012, overall adjusted model, stratified by sex, stratified by BMI and sensitivity analyses.

	Estimate (95% CI)	P-value	Fluoride Interaction Term P-Value
Stratified by Sex:
Male (n= 221)	−2.2 (−6.5, 2.1)	0.32	Referent
Female (n= 217)	−2.5 (−9.0, 4.1)	0.46	0.82A
Stratified by BMI Categories:
Normal Weight (n= 226)	−1.4 (−5.5, 2.6)	0.49	Referent
Overweight (n= 109)	3.1 (−6.6, 12.7)	0.53	0.29B
Obese (n= 103)	−4.8 (−10.2, 0.6)	0.08	0.56C
Sensitivity Analyses:
Term and NBW Only (n=316)	−1.8 (−5.9, 2.4)	0.40

^A:Term assessing interaction between fluoride and sex, female compared to male.

^B:Term assessing interaction between fluoride and BMI category, overweight compared to normal weight.

^C:Term assessing interaction between fluoride and BMI category, obese compared to normal weight.

Covariates included age, sex, BMI, and SES.

We observed no relationship between fluoride and BMI (ß: 0.03, 95% CI: −0.6, 0.6; p = 0.92) or between fluoride and BMI Z-score (ß: −0.003, 95% CI: −0.2, 0.2; p = 0.98). Among normal weight and overweight participants, a single-unit increase in ln-transformed urine fluoride was associated with decreases in eGFR_CysC2012 that did not reach statistical significance (ß_{normal weight}: −1.44, 95% CI: −5.5, 2.6; p = 0.49; ß_overweight: 3.06, 95% CI: −6.6, 12.7; p = 0.53). Among obese participants, a single-unit increase in ln-transformed urine fluoride was associated with a 4.8 unit decrease in eGFR_CysC2012 that was marginally significant (95% CI: −10.2, 0.6; p = 0.08). Tests for interaction between 4-year urine fluoride and categorical adiposity were not statistically significant (p_F*Overweight = 0.29; p_F*Obese = 0.56).

In the sensitivity analysis that excluded either participants born preterm or at low birth weight (n = 377), a single unit increase in ln-transformed urine fluoride was associated with a 2-unit decrease in eGFR_CysC2012 that did not reach statistical significance (ß: −1.8, 95% CI: −5.9, 2.4; p = 0.40). For the sensitivity analysis (n = 435) that included secondhand tobacco smoke exposure as an additional covariate, similar to the main findings, a single unit increase in ln-transformed urine fluoride was associated with a 2.3-unit decrease in eGFR_CysC2012 that was not statistically significant (ß −2.3, 95% CI: −5.9, 1.3; p = 0.21).

The subanalysis of 74 participants that used urine creatinine for adjustment for urine concentration (as opposed to specific gravity) showed similar trends of a negative association with eGFR_CysC2012, although confidence intervals were wider due to the smaller sample size (ß −1.6, 95% CI: −6.9, 3.8; p = 0.56).

Associations between children’s 4-yr-old SG-adjusted urinary fluoride and additional kidney function parameters

In separate models adjusted for age, sex, categorical BMI, and SES (Table 2), we observed no evidence of association with preadolescent BUN, cystatin C, or systolic blood pressure. A unit increase in ln-transformed urinary fluoride was associated with a slight increase in serum creatinine of 0.01 (95% CI: −0.004, 0.03; p = 0.14) and a modest increase in diastolic blood pressure (ß: 0.98, 95% CI: −0.4, 2.4; p = 0.16). Additionally, in separate models adjusted for age, sex, categorical BMI, specific gravity, and SES, we observed no evidence of association with urine kidney-specific proteins such as a1-microglobulin, EGF, OPN, NGAL, albumin, renin, KIM-1, and clusterin (Supplemental Table 3).

Table 2.

Adjusted associations between 4-year fluoride and additional kidney function parameters.

Kidney parameter*	Estimate (95% CI)	P-value
Blood Urea Nitrogen (n = 353)	−0.28 (−0.26, 0.83)	0.30
Cystatin C, serum (n = 438)	16.7 (−11.2, 44.3)	0.24
Creatinine, serum (n = 354)	0.01 (−0.004, 0.03)	0.14
Systolic blood pressure (n = 419)	−0.31 (−1.6, 2.2)	0.75
Diastolic blood pressure (n = 419)	0.98 (−0.4, 2.4)	0.16
eGFRSchwartz (n = 354)	−3.63 (−11.1, 3.8)	0.34
eGFRCysC2012 (n = 438)	−2.2 (−5.8, 1.4)	0.23

Models were adjusted for child’s age, sex, categorical BMI, and SES.

Fluoride - lead co-exposure and preadolescent eGFR_CysC2012

Similar to our prior findings⁴¹, we observed a null association with children’s 4-year BLL and preadolescent eGFR_CysC2012 (ß: −0.18, 95% CI: −3.9, 3.6; p = 0.92). In a fluoride × Pb interaction model (Table 4), the interaction term was marginally significant (p = 0.10).

Table 4.

Associations between blood lead level (BLL) at age 4 and preadolescent eGFR_CysC2012, as well as interaction between 4-year BLL and urine fluoride.

	Estimate (95% CI)	P-value	Interaction* Term P-Value
BLL (n=432)	−0.18 (−3.9, 3.6)	0.92	0.10

^*The interaction term between 4-year BLL and urine fluoride.

Covariates included age, sex, BMI, and SES.

Discussion

We examined longitudinal associations between early childhood urinary fluoride levels with subsequent preadolescent eGFR, BUN and blood pressure. Overall, we observed an inverse relationship between fluoride and subsequent eGFR, although it did not reach statistical significance. Similarly, we did not observe evidence of effect modification by BMI category or sex. Although we did not see an interaction with fluoride, we note that within the strata representing children with obesity, we observed an inverse association between urinary fluoride and eGFR that was stronger than in the other strata with lower BMI, although it did not reach statistical significance. We also observed evidence of a lead × fluoride interaction at age 4 that was marginally associated with decreased eGFR.

These findings suggest that chronic low-level fluoride exposure in early childhood may not contribute to clinically-apparent adverse effects to preadolescent kidney function. Since fluoride tends to accumulate in the bones¹², with a half-life that is measured in years, this may act as a depot for chronic exposure that can affect the kidney in the longer term. Periods of life in which there is rapid bone turnover (adolescent growth spurt for example or osteoporosis in the elderly) may be sensitive time periods in which fluoride renal toxicity is exacerbated through bone remodeling. It may be that bone accumulation of fluoride throughout childhood and adolescence contributes to adverse renal effects observed later in life; whereas fluoride exposure in early childhood (or at any one specific time point in early youth) is not sufficient to produce adverse effects. Also, hyperfiltration related to adiposity may be a separate mechanism from the maladaptive alterations in glomerular pressure that lead to nephron-specific GFR increases and cellular injury. These maladaptive changes may also be less likely to appear in our study population of preadolescents, which may help explain our findings. Future studies employing longitudinal repeated measures and more sensitive kidney function biomarkers are needed to explore this possibility.

Prior studies have examined fluoride exposure and children’s kidney function and reported mixed findings.^23,24,42,43 We previously conducted a cross-sectional study including both plasma and water fluoride concentrations (mean plasma fluoride: 0.40 μmol/L; mean tap water fluoride: 0.48 μmol/L) with kidney function parameters among 1985 adolescents in the United States aged 12–19 years. We found that a 1-μmol/L increase in plasma fluoride was associated with a 10.36 mL/min/1.73m² lower eGFR as well as higher serum uric acid²³. Our observation of significant associations between fluoride exposure and kidney function parameters among US adolescents, but not among children in Mexico in the current study could support the possibility of a cumulative effect of fluoride exposure. Interestingly, a cross-sectional study of 210 Chinese children aged 10–12 observed a dose-response relationship between exposure to high concentrations of fluoride in drinking water (at concentrations over 2 mg/L) and two protein markers of renal dysfunction, NAG and gamma-GT activity^5,42. A second cross-sectional study of 642 children in China aged 10–15 years (mean urinary fluoride: 30 μM F/mM Cr) reported that exposure to fluoride from coal pollution was associated with reduced glomerular function (assessed by urine inorganic phosphate).⁴³ Water or coal-associated fluoride levels may better approximate chronic fluoride exposure during childhood or adolescence than urinary fluoride measurements which can be influenced by daily behaviors (e.g. swallowing toothpaste, recent tea or bottled water consumption) that may not necessarily reflect long term fluoride intake. Additionally, our findings may be inconsistent with the aforementioned studies because fluoride concentrations in the present study were relatively lower and we assessed different outcome measures, including cystatin C-based eGFR which performs better as an indicator of eGFR among obese children³⁴ (whereas prior studies reported creatinine-based eGFR). Finally, a cross-sectional study of 374 children in Mexico aged 7–11 (median urinary fluoride: 2.2 μg/mL) reported an association between urinary fluoride with increased eGFR.²⁴ This could potentially reflect hyperfiltration from chronic fluoride exposure similar to that which has been observed among youth exposed to certain metals.⁴⁴ Alternatively, it may be a consequence of fluoride metabolism whereby children with higher eGFR excrete bodily fluoride in urine more readily. While we did not have concurrent fluoride measures for this study, we aim to explore these potential mechanisms in future studies.

Our results from analyses stratified by BMI are intriguing since previous studies have shown that overweight and obesity are associated with decreased kidney function in preadolescents.^45,46 Indeed, we observed an inverse association between fluoride exposure and eGFR in obese children, but a positive association among overweight children. These findings are important in the public health context of Mexico, where prevalence of childhood overweight and obesity is over 30%.⁴⁷ Interestingly, a cross-sectional study showed that fluoride exposure (as measured by urinary fluoride at age 7–13) was associated with higher BMI z-score⁴⁸. In the PROGRESS cohort, we found no direct association between 4-year urinary fluoride and preadolescent BMI z-score. BMI is a potential mediating factor that should be explored in future studies; also, the increase in childhood and adolescent obesity in recent decades presents an important opportunity for longitudinal assessment (such as the one presented here) for hypothesis testing. Our findings highlight the relevance of follow-up of these children and assessment of longitudinal change in BMI in relation to kidney function.

This study has many strengths. PROGRESS is a well-established cohort study with simultaneous exposures, as well as well-characterized summary data on covariates and demographics. Four-year urinary fluoride was measured years prior to eGFR, and thus quantification of exposure cannot be biased with respect to kidney function assessment. To our knowledge, our study is also the first longitudinal assessment of children’s urinary fluoride levels and eGFR. This study is also unique in that it does not involve a group at risk for fluorosis, but rather concerns low-level population-wide fluoride exposure. Renal function was assessed using multiple measures of glomerular function, including a cystatin C-based equation, which is a preferred biomarker for pediatric populations such as our cohort. Creatinine-based equations, such as both the previous and the revised Schwartz equation, are more subject to variations in muscle mass. Furthermore, changes in other biomarkers (such as serum creatinine, BUN, or the select urine proteins that were assayed) may not be observable until substantial kidney damage has already occurred³⁴.

The lack of available data on prenatal fluoride levels precludes the assessment of risk during the perinatal period, a distinct window of vulnerability for the developing kidney. There is also a lack of available data on plasma fluoride levels, which may be a better biomarker of exposure in some populations. An advantage of plasma fluoride may be its ability to approximate bodily fluoride absorption, as opposed to urinary fluoride, which is more representative of bodily fluoride excretion. Young children also tend to excrete less fluoride than older children because more fluoride is absorbed in bone due to the rapidly growing skeletal system⁴⁹. An additional limitation is the single time point of eGFR, which is before the adolescent growth spurt (longitudinal follow-up will include adolescence). Finally, the potential effects of fluoride may not manifest by preadolescence, but rather could take decades. Use of repeated measures and longitudinal follow-up regarding the relationship between fluoride exposure and kidney function are warranted.

Conclusions

While we did not find significant associations between urinary fluoride and eGFR, this is the first longitudinal study of fluoride exposure in children, and trends found within strata of children with obesity deserve further study. Subclinical effects may be more prominent as the children age. Future studies should consider repeated measures of both exposure and kidney function parameters.