Pharmacokinetics of 38% Silver Diamine Fluoride in Children

Abstract

Purpose:

The purpose of this study was to measure serum levels and characterize the pharmacokinetics of silver and fluoride in healthy children receiving SDF treatment for dental caries lesions.

Methods:

Children (3–13 y.o. with ≥1 caries lesion) were recruited at the University of California, San Francisco Pediatric Dental Clinic from August 2019 through March 2020. Blood was obtained at one randomly selected timepoint up to 168 hours after SDF application. Serum fluoride and silver were measured, and population pharmacokinetic modeling was used to estimate pharmacokinetic parameters and simulate silver concentration versus time profiles in cohorts of children (15–50 kg).

Results:

55 children completed the study. Serum fluoride had no discernable temporal pattern. Silver concentrations were best described by a one compartment model with first-order absorption and elimination, and weight as a covariate. Simulated 15 kg children had higher predicted peak silver concentrations than simulated 50 kg children (22.0 ng/mL [95% CI: 19.4–24.6] vs. 12.8 ng/mL [95% CI: 11.3–14.3]), and a longer predicted silver half-life (15.5 days [95% CI: 12.5–18.5] vs. 4.0 days [95% CI: 2.7–5.3]).

Conclusion:

Intermittent topical SDF application in children is safe and serum concentrations of fluoride and silver pose little risk of toxicity.

Trial Registration:

NCT04186663.

38% silver diamine fluoride (SDF) is a topical agent used to arrest dental caries.¹ The penetration of silver ions leaves a permanent black stain on carious tooth structure that poses esthetic concerns for some parents, but is preferred by many when compared to alternative treatments that may require sedation or general anesthesia.^2–5 The efficacy^6,7 and clinical safety^8,9 of SDF in children is well-documented. Furthermore, SDF is a minimally-invasive, easy-to-use, and cost-effective treatment that has expanded access to dental care for children.^10,11,12 The American Academy of Pediatric Dentistry (AAPD) has outlined criteria for pediatric patients who might benefit from SDF treatment.¹³ When using SDF to treat dental caries, the AAPD recommends initial application, follow-up at 2 to 4 weeks with reapplication as needed, and additional reapplication at recall appointments based on clinical evaluation of caries arrest.¹³ For sustained arrest, when lesions are not restored, biannual reapplication is recommended.¹³

SDF is applied topically in milligram quantities to teeth, after which small amounts can be swallowed and absorbed into systemic circulation. Two pharmacokinetic (PK) studies of SDF application to noncarious teeth in adults found no appreciable elevation of serum fluoride above pre-treatment baseline levels, and the absorbed silver was cleared from the body with an estimated half-life of approximately 2 days.^14,15 Based on adult PK parameters for silver, a physiologically-based pharmacokinetic (PBPK) model of SDF in children was developed. The model predicted that younger children would have higher plasma silver concentrations for a given dose, and that silver exposure decreases as a function of age or weight to adult levels.¹⁶ To date, the pharmacokinetics of SDF in children have not been directly studied.

While SDF is classified as a class II medical device for treatment of dental hypersensitivity, it is primarily used “off-label” for arrest of dental caries, especially in the pediatric population. While clinical PK data on SDF is available in adults, none are available in children likely due to methodological challenges associated with repeated blood sampling over a long duration of time in this population. Nonetheless, it is important to understand the kinetics of silver and fluoride in children to aid in characterizing the safety profile of SDF in one of its intended use populations.¹⁷ Additionally, pharmacokinetic data are needed to obtain FDA approval for SDF treatment of dental caries in children.

The objective of this study was to measure serum levels and characterize the pharmacokinetics of silver and fluoride after 38% SDF treatment of dental caries lesions in healthy children. This study was designed to be minimally invasive by leveraging single time point blood sampling and population PK modeling.

Methods

Participant recruitment

The study was conducted under Investigational New Drug authorization 124808 from the US Food and Drug Administration. The study was approved by the Western Institutional Review Board (20191756), Trial Registration NCT04186663.

Children were enrolled from August 2019 through March 2020 from the University of California, San Francisco (UCSF) Pediatric Dental Clinic. Participants were 3–13 years old, healthy, and not taking prescription or over-the-counter medication, except “as needed” inhalers or allergy medication. Each child had at least 1 carious enamel or dentin lesion. Participants were excluded if they received SDF treatment within the past 3 months, had oral mucositis or ulcerative lesions, or had known sensitivity to silver or fluoride.

Procedures

Procedures were performed at UCSF Pediatric Dentistry Clinic and UCSF Benioff Children’s Hospital Pediatric Clinical Research Center. Screening, review of medical history, and informed consent were obtained in the parent or caregiver’s primary language (English, Spanish, or Cantonese), with professional medical interpreters used for non-English consents. Assent was obtained from all participants 7 years or older. REDCap (Research Electronic Data Capture hosted at UCSF) was used to randomly assign participants to one of seven timepoints for blood sampling: 2, 4, 6, 24, 48, 96, or 168 hours post-SDF application. Timepoints were selected to characterize the absorption and elimination of silver and minimize the invasiveness of the study for participants. Participants were instructed to not brush their teeth on the morning of the visit, and they were provided a non-fluoride toothpaste to use in the time interval between SDF treatment and their blood draw.

The test product was aqueous SDF (Ag(NH₃)₂F), CAS Registry No. 33040-28-7, 38.3–43.2% in purified water, 5.0–5.9% (w/v) fluoride and 24.4–28.8% (w/v) silver (Advantage Arrest^®, Elevate Oral Care LLC, West Palm Beach, FL, USA). The product was from a single lot and certified by the manufacturer. It was stored according to the manufacturer’s instructions. SDF (1–2 drops) was dispensed from the manufacturer’s multi-use bottle into a plastic dappen dish. Teeth were brushed with a soft toothbrush to remove debris. Difficult to access lesions were brushed with an orthodontic prophy angle to remove remaining debris in the cavitations. Affected areas were isolated with cotton rolls and dried with air before application of SDF. A mouth prop and saliva ejector were used to aid in maintaining isolation. SDF was applied to the carious lesion(s) with a dental applicator brush, and the teeth were isolated for about one minute. A water rinse, with high-volume evacuation, was performed. Using an analytical balance (Veritas L123i Precision Balance), an estimate of the amount of SDF applied was calculated as the difference in weight of the brush and dappen dish before and after SDF application. Participants were asked not to eat for two hours after SDF application.

For each participant, a 6 mL blood draw was completed at UCSF Benioff Children’s Hospital Pediatric Clinical Research Center at the randomly assigned timepoint in fluoride-free collection tubes. The actual time of the blood draw was recorded. Serum was obtained by centrifugation (1400 g × 10 min at 4°C) and stored at −80°C until analysis. Samples were analyzed by the Environmental Health Laboratory and Trace Organics Analysis Center at the University of Washington as described in detail by Lin et. al. (2019). In brief, serum fluoride and silver concentrations were determined using fluoride ion selective electrode and inductively coupled plasma-mass spectrometry, respectively. Based on the analytical reports by the University of Washington Environmental Health Laboratory and Trace Organics Analysis Center, the limits of detection for this study were 1 ng/mL (0.001 parts per million) for fluoride and 0.1 ng/mL for silver.

Data analysis

There was no discernable temporal pattern for serum fluoride concentrations; therefore, pharmacokinetic analysis on fluoride was not performed. Serum silver concentration versus time data were analyzed simultaneously using population pharmacokinetic analysis with nonlinear mixed effects modeling using Phoenix WinNonlin software (version 8.3, Certara, Princeton, NJ USA). Model development was performed using the first-order conditional estimates method with extended least squares estimation with η-ε interactions. To establish the structural base model, data were fit to a one compartment model with first-order absorption and elimination. The estimated PK parameters were apparent volume of distribution (V/F) and apparent oral clearance (CL/F). The initial values for these parameters were obtained using the naïve pooled data.

To select an error model, additive and multiplicative error models were tested, and the additive model was selected based on the diagnostic plots. Given that age and weight are highly correlated, and weight is used by dentists for administering local anesthesia, weight was selected as a continuous covariate to evaluate for correlation with the PK parameters (V/F and CL/F). The inter-individual variability (IIV) in PK parameters of silver was evaluated by using an additive error model as described by: P_i = Ptv·[weight/mean(weight)]·dPdWeight·exp(η_i), where P_i is the parameter value (either V/F or CL/F) of the i^th individual, Ptv is the typical value of the population parameter, dPdWeight is the weight covariate on the parameter, and η_i is the random variable for the i^th individual, which is normally distributed with mean 0 and variance ω². For a covariate to be retained in the model, its inclusion had to result in a decline of 3.841 for one parameter, or 5.991 for two parameters, in the objective function value (ΔOFV) at α = 0.05.

The model was internally validated using a bootstrap analysis in Phoenix WinNonlin, in which datasets were resampled with replacement from the original datasets and refit to the model (n=1000). To better understand the effect of weight on silver PK in theoretical cohorts of children with varying weights (15, 20, 30 and 50 kg), the final population PK parameters and interindividual variability estimates were used to simulate silver concentration versus time curves for 100 children per cohort following application of 33 mg SDF. Peak concentration (C_max), time to peak concentration (T_max), CL/F, V/F, area under the curve (AUC), and elimination half-life (t_1/2) from these 100 simulations per cohort were estimated using non-compartmental modeling in Phoenix WinNonlin and summarized as mean values and 95% confidence intervals.

Results

Participants

59 healthy participants were enrolled in the study and 55 children completed the study. Of those who did not complete the study, 3 children were not able to complete the blood draw and 1 child was lost to follow-up after SDF application. Participants were 45% female and were racially and ethnically diverse. The average age was 7.7 ± 2.9 years (47% of participants were 3–6 years old). An average of 33 ± 8 mg of SDF was applied (range: 10–55 mg) to 7.3 ± 2.6 teeth (range: 2–15 teeth). The participant demographics and SDF application details are summarized in Table 1.

Table 1.

Participant demographics and topical 38% silver diamine fluoride treatment in a PK study of children treated for dental caries.

Characteristic*	Healthy Child Participants (N = 55)
Age (years)	7.7 ± 2.9
3–6	26 (47%)
7–10	21 (38%)
11–13	8 (15%)
Weight (kg)**	32.6 ± 16.5 (15.0–73.6)
Gender
Male	30 (55%)
Female	25 (45%)
Race
White	11 (20%)
Asian	11 (20%)
African American	8 (15%)
Unknown or not reported	25 (45%)
Ethnicity
Hispanic or Latino	24 (44%)
Not Hispanic or Latino	25 (45%)
Unknown or not reported	6 (11%)
Amount of SDF applied (mg)	33 ± 8 (10–55)
Number of teeth treated	7.3 ± 2.6 (2–15)

^*Reported as mean ± standard deviation (range) or count (%)

^**Weight was missing for 1 participant (n = 54)

Fluoride and Silver Pharmacokinetics

Serum samples from 6 to 10 children were collected at each timepoint. Given some protocol deviations in the blood draw times, actual sampling times for each participant were used in the pharmacokinetic analyses.

Following SDF application, the serum fluoride concentrations ranged from 6 to 36 ng/mL (0.006 to 0.036 ppm) (Figure 1). As baseline blood samples were not obtained, baseline serum fluoride concentration corrections were not made to the post-SDF application fluoride concentrations. The average serum fluoride concentration was slightly higher in children during the first 6 hours after SDF application (17.5 ± 7.1 ng/mL) compared to subsequent sampling timepoints (12.2 ± 3.4 ng/mL for 24, 48, 96, 148 hours).

Figure 1.

Observed serum fluoride concentrations following application of 38% SDF to children. Each dot is the measured serum fluoride concentration from a child at the actual sampling time.

Following SDF application, the serum silver concentrations ranged from 1.4 to 46.2 ng/mL (Figure 2). The coefficient of variation (CV%) at each timepoint ranged from 46% (24 h) to 148% (48 h). A one compartment model with first-order absorption and elimination best described the silver concentration versus time data following application of SDF. Due to the paucity and variability in the silver concentrations at early timepoints reflecting the absorption of silver, the absorption rate constant for silver was fixed at 23.7 day⁻¹. This value was based on published adult silver pharmacokinetic data.¹⁵ Adopting a fixed absorption rate constant resulted in estimated peak silver concentrations at ~4 hours that agreed with the observed trend in the pediatric data. Inclusion of intraindividual variability (IIV) for the apparent volume of distribution (V/F) and apparent oral clearance (CL/F) improved the model. The CV% for V/F and CL/F decreased from 13% to 7% and from 39% to 16%, respectively (data not shown).

Figure 2.

Observed serum silver concentrations following application of 38% SDF to children. Each dot is the measured serum silver concentration from a child at the actual sampling time.

Based on an additive error model, weight was a significant predictor for V/F (ΔOFV: −8.99) and CL/F (ΔOFV: −11.6) and combined for V/F and CL/F (ΔOFV: −16.9). The parameter estimates of the final population PK model are presented in Table 2. All relative standard error values of the parameter estimates were below 30%, indicating that the estimates obtained from the data were of relatively good precision (Table 2). The population estimates of the apparent volume (V/F) of silver was 2279 L and apparent oral clearance (CL/F) was 387 L/day.

Table 2.

Population pharmacokinetic parameter estimates of silver following SDF application to children.

Parameters	Model Parameter Estimates (RSE%)
ka (day−1)	23.7 FIXED
V/F (L)	2279 (6.8%)
CL/F (L/day)	387 (16%)
Interindividual variability (IIV)
V/F IIV (%)	61 (13%)
CL/F IIV (%)	89 (28%)
dVdWeight	0.37 (17%)
dCLdWeight	1.66 (7.9%)
Residual error	0.074 (2.6%)

ka: absorption rate constant

V/F: apparent volume of distribution of silver

CL/F: apparent oral clearance of silver

IIV: Interindividual variability of the indicated parameter

dVdWeight: weight covariate on V/F

dCLdWeight: weight covariate on CL/F

RSE: relative standard error

Bootstrap analysis was used to verify the reproducibility and/or robustness of the final population PK model for silver. The estimated model parameters of the final population PK model were comparable to those following a bootstrap analysis (data not shown). Individual plots of the visual predictive check were created as each study participant received a different dose based on the amount of SDF applied (Supplemental Figure 1). The diagnostic plots for the final model suggest that the model adequately described the serum silver concentration versus time data following SDF application (Supplemental Figure 2).

To explore differences in silver exposures in children, the population PK parameters and interindividual variability estimates were used to simulate serum silver concentrations at a fixed 33 mg SDF amount for 100 children in each cohort (15, 20, 30 and 50 kg). C_max, T_max, exposure (area under the curve [AUC]), and t_1/2 were summarized by cohort (Table 3). The simulated peak concentration and AUC were highest in the smallest children and decreased with increasing weight (Figure 3; Table 3). Simulated 15 kg children had a predicted peak concentration that was 1.7-fold higher than simulated 50 kg children (22.0 ng/mL [95% CI: 19.4–24.6] vs. 12.8 ng/mL [95% CI: 11.3–14.3]) and an exposure that was 7.5-fold higher (382 ng·day/mL [95% CI: 317–447] vs. 51 ng·day/mL [95% CI: 38–63]). The simulated half-life was 15.5 days (95% CI: 12.5–18.5) in 15 kg children and decreased to 4.0 days (95% CI: 2.7–5.3) in 50 kg children.

Table 3.

Estimated pharmacokinetic parameters of silver based on simulated silver concentration vs. time profiles for cohorts of children ranging from 15 to 50 kg (100 simulations per cohort).

Silver PK Parameters	Simulated Pediatric Cohort
	15 kg	20 kg	30 kg	50 kg
SDF Applied (mg)	33	33	33	33
Tmax (hr)	5.9 (5.7 – 6.1)	5.6 (5.4 – 5.8)	4.9 (4.7 – 5.2)	4.4 (4.2 – 4.7)
Cmax (ng/mL)	22.0 (19.4 – 24.6)	21.7 (18.4 – 24.9)	16.9 (14.6 – 19.1)	12.8 (11.3 – 14.3)
V/F (L/kg)	136 (117 – 155)	122 (103 – 142)	89 (77 – 101)	73 (61 – 86)
CL/F (L/hr/kg)	0.52 (0.41 – 0.63)	0.63 (0.51 – 0.74)	0.88 (0.71 – 1.05)	1.32 (0.93 – 1.72)
AUC (ng·day/mL)	382 (317 – 447)	260 (188 – 333)	106 (84 – 128)	51 (38 – 63)
t1/2 (day)	15.5 (12.5 – 18.5)	12.1 (8.8 – 15.4)	5.9 (4.6 – 7.2)	4.0 (2.7 – 5.3)

Reported as average (95% confidence interval)

C_max: peak concentration of silver

T_max: time of peak silver concentration

V/F: apparent volume of distribution of silver

CL/F: apparent oral clearance of silver

AUC: area under the curve of silver

t_1/2: elimination half-life of silver

Figure 3.

Average simulated serum silver concentration vs. time curves in cohorts of children following application of 33 mg of SDF (100 simulations per cohort). Cohorts: 15 kg (gray solid line), 20 kg (dotted line), 30 kg (dashed line), and 50 kg (black solid line).

Discussion

This is the first study to evaluate the pharmacokinetics of silver in healthy children receiving SDF treatment for dental caries lesions. Single timepoint blood sampling was used to measure serum levels of silver and fluoride, and population PK modeling was used to minimize the invasiveness of the study. The population PK parameters and interindividual variability estimates were subsequently used to examine differences in silver exposures via simulations in cohorts of children.

SDF treatment was well tolerated by participants and no adverse effects were observed or reported. This is consistent with other studies that have demonstrated the safety of this treatment.^8,9

The average dose of SDF applied per child in this study was 33 mg (~26 μL), which contains approximately 1.5 mg of fluoride and 7.0 mg of silver. After SDF treatment, participant serum fluoride concentrations (6–36 ng/mL) fluctuated around previously reported baseline levels in teenagers (3.04–15.4 ng/mL),¹⁸ and were similar to baseline levels in an adult PK study on SDF (10–50 ng/mL).¹⁵ Reported fluoride varnish (approximately 3–5 mg fluoride applied or less than half of a standard 0.5 mL packet) and fluoride gel treatments (approximately 40 mg fluoride applied or 3 mL) result in higher serum fluoride concentrations compared to SDF (varnish: 60–120 ng/mL and gel: 300–1443 ng/mL).^19,20 These results suggest that a single SDF application does not appreciably increase serum fluoride levels in children.

Participant serum silver concentrations ranged from 1.4–46.2 ng/mL, similar to adult peak concentrations in Vasquez et al. (3–29 ng/mL)¹⁴ and higher than adult peak concentrations in Lin et al. (0.13–2.2 ng/mL).¹⁵ The amount of SDF applied, number teeth treated, and whether treated teeth were carious or noncarious differed between this pediatric study (10–55 mg SDF applied to 2–15 carious teeth) and the adult studies (approximately 4–12 mg SDF applied to 3–5 noncarious teeth).^14,15 The increased serum silver concentrations in some children may in part be due to the larger amount of SDF applied in this study. It could also be due to differences in body size, liver size, and hepatic blood flow,²¹ and may be impacted by differences in biliary excretion or the amount swallowed/absorbed in children versus adults.

Based on the simulations of silver exposure in cohorts of children (15–50 kg) (Table 3), the predicted peak silver concentration in children (12.8–22.0 ng/mL) was within range of peak concentrations observed in adults (3–29 ng/mL).¹⁴ Consistent with predictions from the PBPK model of SDF in children,¹⁶ the highest peak silver concentrations were seen in the smallest children (Figure 2), with silver exposure 7.5-fold higher in 15 kg children compared to 50 kg children. The estimated time to peak silver concentration was comparable in children and adults (4.4 to 5.9 hours vs. 5.3 ± 5.8 hours, respectively),¹⁵ and the estimated elimination half-life of silver was longer in children (4 to 15.5 days) compared to adults (1.9 ± 1.1 days).¹⁵ Given the estimated long half-life of silver, limited duration of the PK studies, and high degree of variability in silver concentrations in children and adults, uncertainty remains around the true half-life of silver in children and adults.

A clinical outcome in humans who ingest too much silver is argyria, a permanent bluish-gray discoloration of the skin with no associated adverse health effects.²² To avoid this cosmetic effect, the United States Environmental Protection Agency (US EPA) set the lowest observed adverse effect level (LOAEL) to 1 g total intravenous (IV) dose, which is approximately 90- to 500-fold above the amount of silver applied in this study. The EPA also estimates that a 25 g oral dose is roughly equivalent to the 1 g IV dose²², so the safety margin could be as high as 2,000- to 13,000-fold. These estimates assume a worst-case scenario in which the entire amount of SDF applied is ingested and absorbed. We know, however, that the amount absorbed is far less than the amount applied. SDF can remain on the tooth, be suctioned during isolation, soaked into a cotton roll, or rinsed-off, with a small fraction of SDF being swallowed and a small fraction of that being subsequently absorbed. Given these factors, the safety margin is likely larger. Additionally, low SDF dosage and intermittent use further decreases risk.

Limitations of this study include use of a fixed absorption rate constant based on adult data¹⁵ (due to insufficient data and variable silver concentrations at the early timepoints in children), small numbers of participants per timepoint, short study duration that did not allow precise estimation of the silver half-life, and an inability to determine the amount of silver applied or absorbed. Determination of the absorption rate constant and full characterization of silver pharmacokinetics would require each participant to receive multiple early blood draws and weekly blood sampling over the span of 4 or 5 half-lives (i.e., over 2 months). A pediatric study following this design would pose logistical challenges. The amount of SDF applied was estimated as the difference in weight of the brush and dappen dish before and after SDF application. It is possible that the amount of SDF applied could be underestimated if the brush absorbed saliva during SDF application to carious lesions. Although the amount of SDF applied cannot be accurately determined, the pediatric serum silver concentrations were comparable to adult serum silver levels. As with the adult PK studies, we were unable to confirm the amount of silver absorbed by measuring the amount of silver excreted. Silver has a long half-life and is primarily eliminated via biliary excretion. It would be impractical to collect fecal samples for 2 months or longer to accurately measure excretion.²³

Despite the limitations, these data are useful for estimating the safety of SDF treatment and establishing a basis for future studies of SDF in children. The recruitment of a large population of ethnically diverse children across a broad age spectrum increases the generalizability of these results. Additionally, the use of standard in-office methods of SDF application to carious lesions in therapeutic doses strengthens the clinical applicability of these results. Importantly, because participants had a high burden of dental caries necessitating treatment of multiple teeth, our data support the safe use of SDF for treating multiple caries lesions in children.

Conclusions

Based on this study’s results, the following conclusions can be made:

1. Serum concentrations of fluoride and silver, after intermittent topical application of SDF, has little risk of toxicity in children.

2. SDF application in children may result in minor serum fluoride increases that are not appreciably different from baseline fluoride levels. SDF application causes a lower serum fluoride increase than fluoride varnish treatment.

3. The PK of silver is affected by the child’s weight. Smaller children (i.e., 15 kg) have higher predicted peak silver concentrations, higher silver exposure, and longer silver half-lives compared to larger children (i.e., 50 kg).

4. Additional studies in pediatric patients are needed to obtain more accurate estimates of silver absorption and elimination half-lives following SDF application.