Impact of sample mass on the reliability of toenail fluoride for exposure assessment.

Toenails are often used as non-invasive biomarkers in epidemiological research due to their practical advantages, such as simpler collection and storage procedures. Compared to fingernails, they are generally less prone to external contamination [1, 2] and have a slower growth rate, making them well-suited for evaluating long-term exposure to environmental and chemical agents, such as trace-elements and non-metals [2], including fluoride [1].

Fluoride, the negatively charged ion of the element fluorine, is widely present in nature and in the human diet. Due to its well-established role in preventing dental caries [3], fluoride is commonly added to oral healthcare products like toothpastes and mouthwashes and is applied topically in clinical settings. It is also added to drinking water, salt, and milk as a public health initiative in some regions [4]. However, excessive fluoride intake is associated with adverse effects on mineralized and non-mineralized tissues [5,6,7], and with potential adverse health outcomes, including reduced child IQ [8, 9] and disruptions to thyroid function [10]. These concerns highlight the growing need to better understand and monitor fluoride exposure across different populations.

After ingestion, fluoride is primarily absorbed in the stomach and small intestine, a process influenced by factors such as gastrointestinal pH, nutrient interactions and overall dietary composition [11]. Once absorbed, fluoride is distributed via the bloodstream, with plasma concentrations affected by factors such as bone turnover, renal function and the body’s acid–base balance [12]. Due to its high affinity for mineralized tissue, fluoride is primarily deposited in hard tissues (mainly bone) where it is strongly —though not irreversibly— bound. In contrast, only ~ 1% is deposited in soft tissues. Circulating fluoride is also incorporated into the nail’s matrix, and its content in nails depends on factors that influence the nail’s growth rate and length [13]. Fluoride detection in fingernails and toenails has been reported to reflect exposure over a period of approximately 3–4 months [13, 14]. This has supported the use of nails as a biomarker of longer-term exposure, particularly in association to dental fluorosis [15, 16].

Compared to other biomarkers of fluoride exposure, toenails are considered biomarkers of recent exposure [17]. While contemporary biomarkers, including blood/plasma, saliva, and urine, capture shorter-term fluctuations, they reflect different periods of exposure, with blood/plasma reflecting exposure within the last 3–6 h, saliva over the last few hours (as it follows plasma levels) and urine over the preceding 24 h [18]. Historical biomarkers such as bone and teeth, provide a longer window but require invasive sampling [18]. Compared to hair, also a keratinized matrix, toenails are less prone to external contamination and provide stronger, often more consistent, correlations with fluoride intake [19].

A range of techniques has been employed for fluoride analysis in nails, including neutron activation analysis [20, 21], acid digestion/extraction methods [22, 23], gas chromatography [24], colorimetric spectrophotometry [25], and the most frequently reported, hexamethyldisiloxane (HMDS)-facilitated diffusion method [14, 26]. However, the potential impact of variations in methodologies and protocols for analyzing fluoride in the nail matrix using HMDS-facilitated diffusion techniques requires further research aiming for standardization [27].

The HMDS-facilitated diffusion method is used to release fluoride by acid-induced dissociation of HMDS, an organosilicon compound [28]. In the presence of a strong acid such as sulfuric acid (H₂SO₄), HMDS reacts with hydrogen fluoride (HF) generated from the sample or standard, forming the volatile compound trimethylfluorosilane. This compound diffuses into an alkaline trap, where the fluoride ion is released by exchanging with a hydroxyl ion from the trap, and is then measured using an ion-selective electrode. For fingernail specimens, studies have reported fluoride analysis through HMDS-facilitated diffusion using a minimum of 5.0 mg of nail collected from adults [14]. However, to date, no studies have investigated the impact of low toenail sample mass, defined in this study as less than 5.0 mg, on the reliability of fluoride analysis using the HMDS-facilitated diffusion method. There is also limited understanding of whether analytical protocols using the HMDS-facilitated diffusion method introduce variability in fluoride extraction efficiency (%FE) from toenails, which evaluation could advance methodological standardization. Therefore, the objectives of this study were (1) To determine the minimal mass required for reliable fluoride analysis in toenail samples, and (2) to compare the %FE from toenail samples using two different protocols for HMDS-facilitated diffusion.

Study design

A methodological study was conducted to determine the minimum toenail mass required for reliable fluoride analysis and to compare the fluoride extraction efficiency of two different laboratory protocols.

Sample collection

Toenail specimens with mean pooled mass of 49.5 ± 26.8 mg (range: 22.4–104.6 mg) were collected from a convenience sample of 11 female participants with mean age of 30.7 ± 8.7 years. The participants were residents of the city of Toronto, Canada, and five (45.5%) were between 15 and 38 weeks of pregnancy. Pregnant and non-pregnant participants were included based on previous research indicating no differences in nails’ composition due to pregnancy [29]. All analyses were conducted using the same group of participants in order to eliminate potential variations in exposure and pinpoint differences in methodological procedures. Participants were instructed to remove any toenail polish or varnish and to clip their toenails after a bath or shower. Then, participants placed either big toenail clippings or other toes’ clippings in separate plastic collection bags. All participants signed a written informed consent form prior the study procedures and authorized the use of their samples for study purpose. Toenail specimens were stored and shipped to Indiana University School of Dentistry’s fluoride research laboratory for analysis. Study procedures were approved by the York Research Ethics Board in Toronto, Canada.

Sample preparation

Pooled toenail specimens were cleaned by sonication in deionized water (diH2O) for 2 min and dried at 37 °C for 4 h (Fig. 1). Toenails from each participant were randomly selected and cut using ethanol-cleaned lab scissors, weighed and grouped into specific masses ranging from 1.0 to 5.0 mg (± 0.5 mg) to create replicate samples (Table 1) for each toe type (big toes and other toes), depending on the availability of specimens’ masses. Only four participants provided toenail clippings from both types of digits.

A total of 98 toenail replicate samples were prepared for fluoride analysis. The average number of replicate samples (± SD) analyzed per participant was 9 (± 5) with a range of 4–20. The range of mass of toenail samples was 0.80–5.50 mg (Table 1).

Determination of fluoride content

Figure 1 summarizes the study procedures. The present study used the procedure described by Martinez-Mier et al. (2011). First, each toenail specimen was placed on the bottom of a disposable 60 × 15 mm Petri dish, and 2.0 mL of diH2Owere added into each dish. After applying petroleum jelly to the inside of each Petri dish lid, 50 uL of 0.05 N sodium hydroxide (NaOH) solution was placed in five equal drops on each dish lid. Each dish was then immediately tightly sealed. After burning a small hole into each lid with a soldering iron, 1.0 mL of HMDS-saturated 3 N sulfuric acid (H₂SO₄) was pipetted in each hole and sealed immediately with petroleum jelly. After overnight diffusion at ambient temperature, fluoride was released and trapped in the NaOH. The trap was recovered and buffered with 25 uL of 0.1 M of acetic acid. The recovered solution was adjusted to a final volume of 100 uL with diH2O. The concentration of fluoride was then measured using a fluoride-ion selective electrode (with a limit of detection of 0.019 mg/L) coupled to a pH/ISE meter (Orion™ Fluoride Electrode and Dual Star™ pH-meter, Thermo Scientific, Waltham, MA, USA). A calibration curve using tenfold serial dilutions of Orion™ ISE standard solution traceable to the National Institute of Standards and Technology (ranging from 0.001 to 2.00 µg/mL) was constructed following the same procedure. Millivolt readings from the samples were recorded and the unknown fluoride content from toenail samples were determined using the equation that explained the relationship between the log of the fluoride concentration of the standards and their corresponding millivolt readings (R² > 0.99).

Quality control procedures

Assay precision was evaluated using the daily calibration standards over four days of analyses. The mean intra-assay precision (coefficient of variation [CV] between daily duplicates across all standards) was 5.1%. The inter-assay precision (CV for each standard across the four analytical days) was 8.2%.

Fluoride extraction efficiency and protocol comparison

Toenails specimens from four randomly selected participants were used to compare fluoride extraction efficiency (%FE) using two protocols for HMDS-facilitated diffusion (Fig. 1). Only nail specimens from other toes (excluding big toe) were used to minimize potential variability in fluoride content by toe type. After sonication with diH2O and drying as described above, six replicate samples weighing approximately 3.0 (± 0.5) mg were pooled from each participant. Then, three replicate samples from each participant were randomly assigned for fluoride analysis using Protocol A and three to Protocol B, yielding 12 replicate samples assigned per protocol. Each replicate sample underwent three sequential analyses (initial analysis followed by two re-analyses), resulting in 36 fluoride measurements per protocol. The two protocols use HMDS-facilitated diffusion but differed in the amount of reagent and water: Protocol A used 1.0 mL of HMDS-saturated 3 N H₂SO₄ and 2.0 mL of diH2O (see section on determination of fluoride content), while Protocol B used 3.0 mL of HMDS-saturated 3 N H₂SO₄ and 3.0 mL of diH2O. %FE was calculated as the amount of fluoride (mg) extracted in each analysis divided by the sum of fluoride amount (mg) extracted across all analyses, multiplied by 100. The fluoride amount (mg) was determined by multiplying the sample weight (g) by its fluoride content (µg/g) and converting to milligrams (mg). For each analysis, samples were dried at 37 °C overnight and weighed to account for changes in toenail mass following HMDS-saturated acid and water exposure.

Statistical analyses

Descriptive statistics were used to summarize the fluoride content in toenail samples, including measures of central tendency (median) and dispersion (interquartile range [IQR] and range). To assess variability in fluoride content by toe type (big toes versus other toes), a mixed-effects regression model with inverse-transformed fluoride as the outcome and robust standard errors was fitted, accounting for within-participant replicates. To assess within-subject variability and establish a minimum toenail mass threshold for reliable fluoride measurement, intraclass correlation coefficients (ICCs) were calculated for 0.50 mg mass increments using the variance components of linear mixed-effects models, specifying participant as a random intercept. Interpretation of ICC values followed Koo and Li [30], where values < 0.5 indicate poor, 0.5–0.75 moderate, 0.75–0.9 good, and > 0.90 excellent reliability. Bootstrapped 95% confidence intervals were estimated using 1,000 resamples in the mixed-effects modelling. Change-point analysis with cluster bootstrapping [31] was conducted to detect the mass at which a significant shift in toenail fluoride content occurs as a function of toenail mass, accounting for participant-level variability. The 95% confidence interval for the change-point was estimated using 1000 cluster bootstrap replicates. Normality of the %FE distribution for both protocols was tested using the Shapiro—Wilk test (p > 0.05). The variance in %FE between protocols was compared using an F-test for equality of variances. Statistical significance was set at p < 0.05. All analyses were performed in R (version 4.4.2) and Stata (version 19, Stata Corp).

Determination of minimal mass required for reliable fluoride analysis in toenails.

The median (IQR) fluoride content across all toenail samples (n = 98) was 2.07 (1.53) µg F^–/g, ranging from 0.82 to 9.22 µg F^–/g. When comparing the fluoride content in toenails according to toe type, nails from big toes (n = 25) showed a higher median (IQR) [1.98 (1.50) µg F^–/g] than those from other toes (n = 25) [1.55 (1.27) µg F^–/g]; however, this variability by toe type was not statistically significant (p = 0.32).

The distribution of fluoride content in toenail versus replicate sample mass for each participant is presented in S1 File. Mixed effects models were run to determine total and between-subject variance, and calculated intraclass correlation coefficients (ICCs) for 0.5 mg increments of toenail mass. It was observed that with a minimum toenail mass of 2.5 mg, fluoride measurements showed very high to excellent reliability (ICC = 0.90, 95% CI: [0.71, 0.95], Fig. 2). Similarly using a change point analysis fluoride content significantly shifted at a toenail mass of 2.2 mg (95% CI: [1.70, 2.70 mg]), appearing unrelated to mass above that mass.

Comparison of fluoride extraction efficiency from toenail samples using two different protocols with HMDS-facilitated diffusion.

Table 2 presents the fluoride content and fluoride extraction efficiency (%FE) for each set of analysis or re-testing of the same samples. Median fluoride content and %FE varied across analyses for both protocols, with Protocol A showing a narrower range of %FE values compared to Protocol B. Specifically, the variance of %FE for Protocol A (var = 32.0) was significantly lower than the variance of %FE for Protocol B (var = 137.7), with F = 0.23 and p = 0.02.