The study protocol was approved by the University of Gent ethical committee (n° B670201836852) and 20 healthy human volunteers participated after written informed consent. Inclusion criteria were: being of an age > 18 years, ability to perform all 2 legs of the study, having no orthodontic or removable prosthetic appliances. Exclusion criteria were pregnancy, self-reported suffering from reduced salivary flow and known hypersensibility to the toothpaste ingredients. Participants had to be recruited outside university staff and students according to ethical committee specifications. A sample size of 20 was calculated based on a pilot study involving the applied quantity of toothpaste (Fig. 1), resulting in a power of 0.9 with α = 0.05.
The study protocol consisted of a two-legged crossover design. The volunteers brushed with one commercially available NaF toothpastes (Sensodyne, GSK Europe, Genval, Belgium). Brushing was performed at libitum using a power toothbrush (Oral-B® Sensitive Clean, Oral-B, Procter & Gamble, Marktheidenfeld, Germany) and a manual toothbrush (Tepe soft, TePe Munhygienprodukter AB, Malmö, Sweden) with a 5-day washout between legs. Foodstuffs with a known high fluoride content was to be avoided the day before and the day of the experiment. A fluoride-free toothpaste was given to the volunteers to be used the day before and the morning of the trial.
Volunteers were instructed to donate a salivary sample at baseline, after 2 minutes of toothbrushing, and, following a 10 second rinse with 10 ml of deionized water, after 3, 5, 10, 20 30, 40 and 60 minutes. Saliva was collected by drooling in a disposable vessel until a volume of 2.5 ml was reached. Fluoride concentration of saliva samples was measured as quickly as possible, if analysis on the same day could not be done, samples were kept frozen at -18°C overnight.
Saliva analysis was performed on duplicate samples according to the protocol developed by Cury and al [5] for in-vitro determination of fluoride in toothpaste. 2 aliquots of whole saliva + toothpaste slurry were pipetted first and subjected to acid digestion with 2N HCl at 45°C for 1 hour, then neutralized with 2.5 ml of 2N NaOH. This is the total amount of fluoride potentially available (total fluoride, TF). 2 more aliquots were analyzed after centrifugation without acid digestion as ionizable fluoride (IF) by only adding 0.5 ml of an equinormal solution of NaOH and HCl. All aliquots were buffered by 0.5 ml of TISAB 2 (Fischer Scientific, Chelmsford, USA) and assayed for fluoride using direct potentiometry with an ion-specific electrode (ISE, Mettler-Toledo, Greifensee, CH). The electrode was calibrated before measurements using appropriate standards prepared with a NaF solution. The mV-concentration graph was plotted by semi-logarithmic regression using GraphPad Prism software (GraphPad, La Jolla, USA) and concentration of the unknown solution was obtained by interpolation.
The protocol was validated internally by always analyzing duplicate samples and calculating correlation coefficients.
For each volunteer (mean of duplicates), a fluoride-time curve was established. The maximum concentration (Cmax) was established from the curves and the Area under the Curve without baseline was established by triangulation (Prism V6.0, GraphPad, La Jolla, CA, USA). Logarithmic decay curves were established from the moment brushing was terminated until the end of the experiment on the shifted fluoride measurements based on a power function. A curve fit through a non-linear least squares regression (R version 3.6.1 package nls.multstart 1.0.0) was performed on the fluoride-time data according to the equation: F = e^(a tµ) – 1,
ln(y + 1) = a tb, with y being fluoride concentration (mg/L), t time (min) and a as the overall scaling factor and b the exponent, indicating the rate of the concentration decay. Parameters a and b were compared using non-parametric statistics.
For the statistical analysis, AUC and Cmax values were normalized by logarithmic transformation and compared using parametric one-way ANOVA. Applied toothpaste quantities were compared Wilcoxon tests and correlated between the manual and power toothbrush using Spearman-Rank correlation.