Protective effect of titanium tetrafluoride and silver diamine fluoride on radiation-induced dentin caries in vitro

This in vitro study evaluated the protective effect of titanium tetrafluoride (TiF4) varnish and silver diamine fluoride (SDF) solution on the radiation-induced dentin caries. Bovine root dentin samples were irradiated (70 Gy) and treated as follows: (6 h): 4% TiF4 varnish; 5.42% NaF varnish; 30% SDF solution; placebo varnish; or untreated (negative control). Microcosm biofilm was produced from human dental biofilm (from patients with head-neck cancer) mixed with McBain saliva for the first 8 h. After 16 h and from day 2 to day 5, McBain saliva (0.2% sucrose) was replaced daily (37 °C, 5% CO2) (biological triplicate). Demineralization was quantified by transverse microradiography (TMR), while biofilm was analyzed by using viability, colony-forming units (CFU) counting and lactic acid production assays. The data were statistically analyzed by ANOVA (p < 0.05). TiF4 and SDF were able to reduce mineral loss compared to placebo and the negative control. TiF4 and SDF significantly reduced the biofilm viability compared to negative control. TiF4 significantly reduced the CFU count of total microorganism, while only SDF affected total streptococci and mutans streptococci counts. The varnishes induced a reduction in lactic acid production compared to the negative control. TiF4 and SDF may be good alternatives to control the development of radiation-induced dentin caries.

www.nature.com/scientificreports/ Fluorides have been used to prevent RCLs. Annual application of sodium fluoride (5%) varnish or 38% silver diamine fluoride (SDF) solution are able to reduce the emergence of new RCLs by 64 and 71%, respectively 14 . The application of fluoride products on the teeth has been suggested throughout the period of radiotherapy [15][16][17] , however, there are few studies on this subject. Wu et al. 18 showed in vitro that dentin (exposed to 68.25 Gy of radiation) treated daily with 5% NaF varnish, effectively showed a restored surface and increased microhardness compared to untreated irradiated dentin.
On the other hand, previous studies have shown a promising effect of titanium tetrafluoride (TiF 4 ) varnish to control enamel de-remineralization compared to NaF varnish 19,20 . The application of TiF 4 induces deposition of an acid-resistant layer containing titanium oxide and hydrated titanium phosphate, able to provide mechanical protection, and higher fluoride uptake compared to NaF, increasing the acid resistance of dental hard tissue 21 .
A recent study also showed a promising effect of TiF 4 varnish on the prevention of dentin carious lesions formation compared to NaF varnish, under microcosm biofilm model produced on previous sound dentin 22 . We do not know if TiF 4 varnish would have the same protective effect if applied on previous irradiated dentin, as well as under conditions simulating patients with HNC. The same is assumed for SDF.
Due to the need for finding alternatives to minimize the development of radiation-induced dentin caries, the aim of this work was to evaluate: (1) the protective effect of TiF 4 varnish compared to NaF varnish and SDF solution on irradiated root dentin with respect to demineralization; and (2) the antimicrobial action of TiF 4 varnish compared to NaF varnish and SDF solution on microcosm biofilm produced from biofilm collected from patients subjected to head and neck radiation.

Results
TiF 4 varnish and SDF solution significantly reduced the integrated mineral loss compared to NaF varnish, placebo varnish and negative control (p < 0.0001). With respect to lesion depth, no protective effect of fluorides was found (Table 1). Figure 1 shows the TMR image and the lesion profile of a representative dentin sample per group.
In agreement with viability assay, TiF 4 was the only one able to reduce the CFU counts of total microorganisms (p = 0.003). No effect of fluoride treatments was seen on Lactobacillus sp. SDF significantly reduced the total Table 1. Mean and standard deviation of the integrated mineral loss (ΔZ, vol% μm) and lesion depth (LD, μm) presented by the demineralized dentin samples. Different letters in the same column indicate significant differences among the treatments. ANOVA and Tukey-Kramer test (ΔZ p < 0.0001; LD p = 0.3986).  www.nature.com/scientificreports/ streptococci (p < 0.0001) and mutans streptococci (p = 0.0001) CFU numbers compared to placebo varnish and negative control, while TiF 4 and NaF did not ( Table 2). All varnishes (TiF 4 , NaF and placebo) significantly reduced lactic acid biofilm production compared to negative control and SDF (p < 0.0001), which in turn did not differ from each other (Fig. 3).

Discussion
Our recent study has shown irrelevant antimicrobial effect of TiF 4 varnish, but a significant ability to reduce carious lesions development in sound dentin under microcosm biofilm model 22 . One other recent study evaluated the effect of fluorides on irradiated dentin, but not under cariogenic conditions 18 .  Table 2. Mean and standard deviation of the colony forming unit (CFU) counting (log 10 CFU/mL) for total microorganisms, total streptococci, mutans streptococci and Lactobacillus sp. Different letters in the same column indicate significant differences among the treatments. ANOVA and Tukey-Kramer test (total microorganisms p = 0.003; total streptococci p < 0.0001; mutans streptococci p = 0.0001 and; Lactobacillus sp. p = 0.23).

Treatments
Total microorganisms Total streptococci Mutans streptococci Lactobacillus sp.  www.nature.com/scientificreports/ Considering the high prevalence of HNC and the side-effects of radiotherapy, efforts to find a good alternative to avoid the development of radiation-dentin caries lesions are extremely relevant. Accordingly, our study confirmed the anti-caries effect of TiF 4 varnish compared to NaF varnish, as shown by Dos Santos et al. 22 , which was comparable to those found for SDF. An annual application of SDF is able to prevent 71% root caries lesions development 14 . However, tooth discoloration induced by SDF application is of major concern, while no significant staining potential by TiF 4 varnish was found 23 .
Interestingly, carious lesions created under conditions simulating irradiated patients were 1.5-2-folder greater compared to those induced on sound dentin under a similar microcosm biofilm 22 and, despite it, TiF 4 varnish still had a protective effect. This result is also supported by an in vivo study 24 , which further showed penetration of F and Ti into the dentin caries lesion. According to Tveit et al. 25 , a titanium-rich layer remained on the dentin surface even after 3 weeks of TiF 4 application in vivo.
Microcosm biofilm induced using human biofilm from irradiated patients also presented a bacterial load 1.2-1.4 times higher compared to microcosm biofilm growth under health conditions on dentin samples 22 . Signori et al. 26 have discussed that the bacteria source (saliva vs. biofilm, caries vs. free-caries patient) has no influence on the potential of microcosm biofilm to induce demineralization under sucrose exposure. However, in the present study, the bacterial load may have had a significant influence on radiation-induced dentin caries. We thus speculate that a high bacterial load could have impaired the anti-caries effect of NaF, found in our previous work 22 , but not demonstrated here.
Therefore, we are planning to conduct future studies analyzing the biofilm behavior and the degree of dentin demineralization, comparing the sources of bacteria (biofilm from irradiated vs. non irradiated patients) and the quality of dentin substrate (irradiated vs. non irradiated), to better address the effect of both factors on the caries development.
We did not apply methods to demonstrate morphological dentin alterations induced by radiation and its relation to the caries lesions formation. We have support from previous studies that showed morphological dentin alterations (presence of cracks on surface and dentin tubules occlusion) after irradiation, by using scanning electron microscopy-SEM 18,27 . The damage of dentin surface induced by irradiation can have interfered on the CaF 2 deposition produced by NaF varnish, and consequently on the F − release from this reservoir, justifying its lack of antimicrobial and anti-caries effects in the present study either.
In contrast to the report of Dos Santos et al. 22 , the present study found that TiF 4 was able to reduce total microorganism CFU, however, with no effect on the numbers of cariogenic species. A microbiome study of microcosm biofilm growth under health and radiation conditions are needed, since other species not studied here may be part of the microcosm biofilm, contributing to the sudden caries lesion development in irradiated dentin 28 and justifying the antimicrobial effect of TiF 4 varnish found on total microorganisms. SDF, on the other hand, had an antimicrobial effect on total streptococci and mutans streptococci, in agreement with the literature 29,30 . Zhao et al. 30 demonstrated that SDF hinders dentin collagen degradation. SDF also induces the formation of products such as CaF 2 , Ag 3 PO 4 and NH 4 OH, which interact with dentin hydroxyapatite forming fluorapatite 31 , providing an acid-resistant structure. A clinical trial also confirms that SDF is more effective in stopping caries lesions in dentin when compared to NaF 32 . Therefore, the anti-caries effect of SDF may be due to its antimicrobial action and also to its chemical interaction with dentin surface. It is likely that the interaction with the tooth surface is more important than its antimicrobial effect, since SDF did not reduce lactate production, despite it had significantly decreased the cariogenic microorganism CFU counting.
In principle, we expected to see low lactate production by the biofilm after SDF application, since this fluoride significantly decreased mutans streptococci CFU counting 33 , which was not observed. We speculate that other aciduric bacteria could be presented in the biofilm, which in turn were not affected by SDF. Also, SDF did not reduce the biofilm thickness. The high amount of extracellular matrix, involved in the biofilm thickness, could have retained more acid in the biofilm, justifying our findings. This hypothesis needs to be confirmed in future studies.
NaF varnish has previously shown to reduce the production of lactic acid under microcosm biofilm 34 , monospecies biofilm 35 and multispecies biofilm 36 . Both NaF and TiF 4 significantly reduced biofilm thickness and lactic acid production, despite the fact that they were not different from placebo varnish, in disagreement with the findings of Dos Santos et al. 22 This finding deserves further attention. Somehow component(s) of our varnish (except F) might have acted as glycolytic enzyme inhibitor(s), reducing the amount of lactate produced by the biofilm.
The caries protective effect of TiF 4 seems to be more due to dentin surface modification than to the antimicrobial effect. Interestingly, the TMR profile of TiF 4 -treated dentin showed an intermediated highly mineralized layer (around 90 μm depth), which might contain F and Ti. For both TiF 4 and SDF, their chemical interaction could have improved mineral gain especially in the intermediate layer (arrows in Fig. 1), but they were not able to impair bacterial acid penetration and, consequently, no reduction in lesion depth was seen.
In conclusion, both TiF 4 varnish and SDF solution were similarly able to reduce the development of radiationinduced dentin caries in vitro. The result of the present study needs to be confirmed by randomized clinical trials in patients affected by HNC and treated with radiotherapy.

Methods
Tooth sample preparation and treatment groups. The bovine teeth were collected from cattle slaughtered in the food manufacturing industry (Frigol S.A, Lençóis Paulista-SP, Brazil). The study was approved by Ethics committee on animal research (CEUA, Number: 004/2018, Bauru School of Dentistry, University of São Paulo, Bauru, Brazil) following the guidelines of the CONCEA (National Council for Control of Animal Experimentation). No animals were harmed in order to conduct this study. One hundred and eighty bovine root dentin samples (4 mm × 4 mm) were prepared 22  www.nature.com/scientificreports/ an energy of 6 meV (Varian, Clinac 6EX, USA) in total dose of 70 Gy and, thereafter, sterilized using ethylene oxide for 4 h under a pressure of 0.5 ± 0.1 kgF/cm. Before sterilization, the average surface roughness was measured by using a contact profilometer (Mahr Perthometer, Göttingen, Germany) and the software MarSurf XCR-20 (Mahr Perthometer, Göttingen, Germany) (5 readings for the calculation of the mean), for samples randomization into the groups (n = 36/group, n = 12 for each biofilm assay): 4% TiF 4 varnish (pH 1.0, 2.45% F); 5.42% NaF varnish (pH 5.0, 2.45% F); 30% SDF solution (pH 8.5, 3.54% F); placebo varnish (pH 5.0) or untreated (negative control) 37 . The dentin roughness may influence biofilm formation 22 and, therefore, it was applied to standardize the initial conditions of the groups (mean: 0.34 ± 0.03 μm). Before treatment, 2/3 of the dentin surface was protected using nail polish to allow to have 2 control areas (untreated and non-demineralized areas).
The F and placebo varnishes contained the same artificial resin as base and ethanol as solvent; SDF solution contained hydrofluoric acid, silver nitrate, ammonium hydroxide and deionized water. The treatments were applied using microbrush on the samples surfaces for 6 h. During this period, dentin samples were stored in remineralizing solution 37 . The treatments were then removed using cotton swab and acetone solution 22,37 and the nail polish was reapplied at the same sites, before biofilm formation.

Microcosm biofilm formation.
A mixed solution containing thawed inoculum (compound from human biofilm from 2 donors who received a total radiation dose of 70 Gy in the head and neck region, mixed with 1% saline solution in the proportion 2 g: 1 mL, and diluted for freezing in 30% of glycerol 26 ) and McBain saliva (proportion 1:50) was added to each well containing a treated dentin samples (24-well microplate, 1.5 mL/well), and incubated for 8 h (5% CO 2 and 37 °C) 22,38 . Thereafter, the medium was removed and fresh McBain saliva containing 0.2% sucrose was added to the wells for further 16 h 22,37 . The medium was replaced daily for more 4 days and incubated under the same conditions described above 22,37 . Demineralization analysis: Transverse microradiography (TMR). After 5 days of biofilm growth, dentin samples (except those from the lactic acid assay) were cleaned, transversally sectioned, polished and submitted to microradiograph exposure (20 kV and 20 mA, Softex, Tokyo, Japan) as previously described 22 . The developed plate was analyzed using a transmitted light microscope fitted with a 20× objective. Two images per sample were obtained using data acquisition (version 2012) and interpreted using calculation (version 2006) software from Inspektor Research System (Amsterdam, Netherlands). The mineral content was calculated based on the work of Dos Santos et al. 22 , assuming 50 vol% of mineral content for sound dentin and that the lesion depth ends when dentin contains around of 47.5% of mineral volume. The integrated mineral loss (ΔZ, vol% μm) and lesion depth (LD, μm) were calculated for the mean of the 2 images per sample.
Biofilm viability analysis. Biofilm was stained using nucleic acid marker (2 µM SYTO 9 green fluorescent nucleic acid stain, Thermo Fisher Scientific, USA) (v = 10 μL/well) for 15 min in a dark environment 22 . Confocal laser scanning microscopy (Leica TCS SPE, Mannheim, Germany) and Leica Application Suite-Advanced Fluorescence software (LAS AF, Mannheim, Germany) were used to analyze the biofilm surface. Three images (275 μm 2 ) were captured and analyzed using BioImage L 2.0 software. The percentage of live bacteria-% and the biofilm thickness-μm were obtained.

Analysis of lactic acid production.
Dentin samples with 5-day microcosm biofilm were incubated in a buffered peptone water (BPW) (Synth, Diadema, Brazil) supplemented with 0.2% sucrose (v = 1 mL/sample) for 3 h, under 5% CO 2 and 37 °C 22 . Lactate concentrations were evidenced via enzymatic method (Enzymatic assay for d-and l-Lactic acid-Ref. 8240; R-Biopharm, Darmstadt, Germany) following the manufacturer's guidelines. The absorbance was measured at 340 nm using a microplate reader (Fluorstar Optima-BMG Labtech, Ortenberg, Germany) and the values converted to g/L.
Statistical analysis. The biofilm assays (viability, UFC counting and lactic acid assay) were performed in triplicate with four data points for each replicate (n = 12). Dentin samples from all biofilm analysis, except from the lactic acid assay, were analyzed by TMR (n = 24). Data were statistically compared using GraphPad Prism software for Windows (GraphPad Software, San Diego, USA). The normal distribution and homogeneity were checked using Kolmogorov-Smirnov and Bartlett tests, respectively. All data were compared using Analysis of Variance (ANOVA) followed by Tukey-Kramer test. The level of significance was set at 5%. www.nature.com/scientificreports/ old with 20 teeth and 1 female: 57 years old with 24 teeth) who received the total head and neck 3D radiotherapy (final dose: 70 Gy), 5 months previously to the study, and met the inclusion criteria: (1) low salivary flow (stimulated saliva flow < 1 mL/min and non-stimulated saliva flow < 0.3 mL/min), (2) without acute gingivitis, (3) not using antibiotics or (4) being submitted to professional fluoride application in the last 3 months neither. Biofilm was collected from the cervical area of all roots without active caries lesions, by using a periodontal curette. Biofilm collection and storage were performed as described by Signore et al. 26 .