Enhanced effectiveness of silver diamine fluoride application with light curing on natural dentin carious lesions: an in vitro study

This study aimed to compare the mean mineral density difference (mMDD) and surface morphology of 10- and 60-s silver diamine fluoride (SDF)-applied dentin carious lesions and to study the effect of an additional 20-s light curing (LC) on SDF-treated teeth. Forty primary molar blocks with natural dentin carious lesions were measured for baseline lesion depth and mineral density using Image-Pro Plus software. The samples were randomly distributed into 4 groups; 38% SDF applied for 1) 10-s (10SDF), 2) 60-s (60SDF), 3) 10-s + LC (10SDF + LC), 4) 60-s + LC (60SDF + LC) and an additional control group to assess the outcome of pH-cycling only. Then all the groups underwent a 7-d bacterial pH-cycling. The dentin carious lesions’ mMDD was determined by digital subtraction radiographic analysis. The surface morphology and elemental profile were assessed by scanning electron microscopy and energy-dispersive X-ray spectroscopy. The mMDD of the dentin lesions was analyzed using two-way ANOVA, generalized linear models analysis. Light curing was the only factor that affected the mMDD (p = 0.007). The mMDD in the 10SDF + LC and 60SDF + LC groups were significantly higher than those without light curing (p = 0.041 and 0.041, respectively). The 60SDF + LC group demonstrated a significantly higher mMDD than the 10SDF group (p = 0.010), while that in the 10SDF + LC group was similar to the 60SDF group (p = 1.00). Scanning electron microscopy revealed denser mineral content layers, which were likely silver and chloride, in the 10SDF + LC and 60SDF + LC groups than in the 10SDF and 60SDF groups, respectively. In conclusion, shortened application time with light curing enhanced SDF remineralization similarly to the conventional method.


Introduction
Silver diamine fluoride (SDF) has become a popular caries-control method for arresting dentinal caries in children because it is easy to handle and more cost-effective in community settings [1,2]. This colorless silver compoundcontaining solution is photosensitive, with 253,900 ppm silver ions and 44,800 ppm fluoride ions [3] and a pH of 9-10 [4]. Because SDF arrests caries, prevents dentin hypersensitivity [5], and its use does not require caries excavation [6], it is considered an appropriate treatment for young or uncooperative children, elderly, and specialneeds patients. Furthermore, aerosol-generating procedures should be avoided to reduce the risk of airborne infection. This makes SDF a safer and more attractive choice in managing cavitated dentin lesions.
SDF arrests dental caries due to its bactericidal effects, interrupts demineralization, promotes remineralization in demineralized enamel or dentin, and inhibits dentin collagen degradation [4]. When SDF is exposed to light, the lesion turns dark [2] and the surface hardness increases [1]. Lesion depth also decreased after SDF application, which might have been due to calcium and phosphate aggregation on the lesion's surface [7].
Light might play a role in the caries arrest by SDF because it is light-sensitive. Crystal and Niederman [2] hypothesized that SDF-treated dental caries in primary anterior teeth was arrested more than in posterior teeth because they could be cleaned easier and were exposed to more light. They also hypothesized that light increases the remineralization efficacy of SDF. Furthermore, a study [8] investigated the effect of a dental curing light on SDFapplied lesions by comparing the silver ion penetration using energy-dispersive X-ray spectroscopy (EDS) in dentin caries lesions applied with SDF and then lightcured. They concluded that the dental curing light induced more silver precipitation in the dentin lesions, especially infected dentin, and increased the lesion's surface hardness. The American Academy of Pediatric Dentistry (AAPD) reported that the typical SDF application time ranged from 10 s to 3 min. However, the AAPD recommends that SDF should be applied for 1 min per tooth [9], which may be too long for young, uncooperative children. Thus, light curing the applied SDF may accelerate silver precipitation, resulting in more rapid caries arrest and decreased application time.
Digital subtraction radiography (DSR) is a method to detect the progression of dental caries, i.e., remineralization and demineralization, by superimposing digital radiographs from before and after the treatment; unchanged structures are subtracted, revealing only changes in mineralization. The advantages of this method are that it can detect carious lesion remineralization with greater sensitivity and reproducibility compared with conventional radiographs, and the whole tooth can be analyzed without being sectioned [10][11][12]. DSR is a very sensitive method that can detect subtle changes of only 5% in demineralized or re-mineralized structure density, while conventional radiographs reveal only a 30-60% change in mineralization [13].
Currently, there is no report demonstrating the effect of light curing (LC) on evaluating the surface and remineralization of SDF-applied carious lesions as revealed by DSR. Therefore, the aim of this in vitro study was to evaluate the remineralization effects of SDF application on natural dentin carious lesions in primary molars using different application times with additional LC compared with its application without LC, as analyzed by DSR, EDS, and scanning electron microscopy (SEM). The null hypothesis was that the remineralization effects were not different between groups regardless of the additional LC or SDF application time.

Sample size calculation and tooth selection
The study protocol was approved by the Ethics Committee (HREC-DCU 2020-008) and Institutional Biosafety Committee (DENT CU-IBC 007/2020).
Currently, there is no report on the efficacy of LC on dentin carious lesion remineralization after SDF application. Therefore, the sample size determination was performed based on our pilot study (4 groups, n = 5/group), comparing the mean mineral density difference (mMDD) after SDF application among 4 groups with and without light curing of 10 s and 60 s application times. Using the G*power program, with a 0.304 partial eta-squared value, a two-way ANOVA statistical test determined that at least 6 samples/ group would be required for this study (α = 0.05, β = 0.20).
Due to the small sample size, the sample size was increased to 10 specimens/group to compensate for potential errors in specimen preparation and data analysis. Therefore, 40 carious primary molars were used in this study.
SEM-EDS analysis was conducted on the specimens. The sample size calculation for elemental profiling (EDS) was performed based on an ex vivo study by Mei et al. [7], and determined that 16 carious primary molars; i.e., 4 teeth per group were required. The surface morphology of 50% of the EDS samples was evaluated by SEM.
The inclusion criterion for both analyses was extracted primary molars diagnosed with irreversible pulpitis with a pulp exposed portion and another portion, in which the caries involved only the outer 1/3 to middle 1/3 of the dentin depth on the occlusal or occluso-proximal surface. The teeth were extracted due to the parents choosing a space maintainer instead of pulpectomy. The exclusion criteria were teeth with buccal and lingual lesions, restorations, developmental defects, arrested caries, and craze lines. Furthermore, the specimens were excluded if the occlusal or occluso-proximal caries involved more than the middle 1/3 of the dentin depth radiographically.

Specimen preparation
Forty primary molars with carious dentin were collected from the dental department at Nongjik hospital in Pattani, Thailand, and stored in 0.9% sodium chloride at 4 ºC until used. Subsequently, the teeth were cut in half along the buccolingual axis using a slow-speed cutter (Isomet 1000; Buehler Ltd., Lake Bluff, Illinois, USA). The portions with the pulp exposure were discarded and only the non-pulp exposed portions were used as experimental samples. The samples were rinsed with deionized water in an ultrasonic bath and dried. Clear nail polish was applied on the specimens, except for the dentin lesion. Finally, the specimens were fixed in 1 × 1 × 0.9 cm (W ×L ×H), self-curing acrylic resin blocks, and polished with 600-, 800-, and 1,000-grit silicon carbide paper (Fig. 1).

Radiographic assessment
Using an X-ray alignment system, baseline digital radiographs were taken to generate reproducible projection geometry (Fig. 2). A phosphor digital imaging plate No. 0 (CS 7600; Carestream Health, Rochester, New York, USA) in a protective sleeve was stabilized in a customized film holder. Each dentin block was put on the imaging plate with the buccal aspect of the tooth facing toward the X-ray tube. A 15-mm thick, rectangular, plexiglass plate was placed on the plexiglass supporter, acting as a soft tissue equivalent [14]. A custom device for locking the collimator was used to position the X-ray collimator perpendicular to the tooth and receptor [12]. The radiographs were taken using an intraoral X-ray system (Kodak 2200; Carestream Health, Rochester, New York, USA) set at 70 kVp, 7 mA, and 0.119 s exposure time. The same machine and settings were used throughout the experiment by the same investigator (J.K.).

Permuted block randomization
After the digital radiographs were taken, they were saved as Tagged Image File Format with a resolution of 792 dots per inch. Image-Pro Plus software version 7.0 (Media Cybernatics; Rockville, Maryland, USA) was used to evaluate the baseline lesion depth (LD) and mineral density (MD) of each group's samples. The baseline LD and MD were used to allocate the specimens into four groups by permuted block randomization. LD was determined by dividing the distance from the dentinoenamel junction (DEJ) to the deepest part of the dentin lesions by the total depth of the dentin as measured parallel from the DEJ to the dentin-pulp junction. MD was measured by identifying two areas of interest (AOIs) at the same lesion depth on each specimen; the deepest part of the dentin lesion and the adjacent sound dentin as a control. The sizes of the measurement windows were set at 15 × 15 pixels for the outer 1/3 dentin lesion and normal dentin, and 30 × 30 pixels for the middle 1/3 dentin lesion to cover a larger area (Fig. 3). The MD measurement was described as the mean density/intensity value. The baseline MD of each specimen (MD baseline ) was determined by the difference between the MD at the deepest part of the lesion (MD lesion ) and the MD at the adjacent normal dentin (MD control ). The Fig. 1 Specimen preparation. Forty primary molars with carious dentin were cut in half along the buccolingual axis. The non-pulp exposed portions were used as experimental samples. Clear nail polish was applied on the specimens, except for the dentin lesion. The specimens were fixed in 1 × 1 × 0.9 cm (W × L × H), self-curing acrylic resin blocks specimens were sterilized with hydrogen peroxide gas and divided into 4 groups according to their baseline LD and MD using permuted block randomization to equally distribute the different LDs and MDs in the groups: Group 1: SDF applied for 10 s (10SDF). Group 2: SDF applied for 60 s (60SDF). Group 3: SDF applied for 10 s + LED light curing 20 s (10SDF + LC).
The SDF was applied without caries excavation of the lesions. An additional 10 specimens were used as the negative control, with deionized water applied on the lesions for 60 s to assess the bacterial pH-cycling outcome (demineralization versus remineralization).
The specimens were immersed in artificial saliva (KCl, MgCl 2 , CaCl 2 , K 2 HPO 4 , KH 2 PO 4 , sodium carboxymethylcellulose, sorbitol, sodium benzoate, and deionized water) at pH ~ 7 for 1 h before SDF application on the carious surface to mimic the oral condition. Each application was performed in a dark, sterile laminar flow cabinet, using a micro-brush soaked with 5 µL 38% SDF (Saforide; Toyo Seiyaku Kasei Co., Ltd., Osaka, Japan). An LED dental curing light with an output intensity of 520 mW/cm 2 and a wavelength range of 450-470 nm (Demi Plus; Kerr, Orange, California, USA) was used in the LC groups. The light intensity was calibrated before curing 20 specimens and the light source contacted the tooth surface during light curing.

Bacterial pH-cycling
Due to the antibacterial property of SDF, the demineralization solution was prepared with two cariogenic bacteria, Streptococcus mutans (S. mutans) ATCC25175 and Lactobacillus casei (L. casei) IFO3533 as previously described [15,16]. The bacteria were cultured on tryptic soy agar plates at 37 °C, 5% CO 2 for 24 h. One colony of each culture was transferred to tryptic soy broth containing 0.5% yeast extract and incubated at 37 °C, 5% CO 2 for 16 h. The incubated broths of each bacteria were diluted in tryptic soy broth containing 0.5% yeast extract, 2% sucrose, and 1% glucose to achieve optical densities of 0.1 as measured by a spectrophotometer at 540 nm. The cultures were mixed at a 1:1 ratio (≈1.37 × 10 8 CFU ml −1 of S. mutans and ≈7.8 × 10 7 CFU ml −1 of L. casei) to make a dentin-demineralizing solution with an average pH of 6.31 ± 0.1 [17].
The specimens were immersed in the demineralizing solution for 4 h and in artificial saliva without fluoride for 20 h. The specimens were rinsed with deionized water before changing the solution and again at the end of the pH-cycling. The specimens were sterilized with hydrogen peroxide gas after bacterial pH-cycling for 7 d [16] and re-assessed for

Digital subtraction radiographic analysis
The samples were radiographed post-bacterial pH-cycling to obtain post-treatment radiographs. They were superimposed with their baseline radiographs using Image-Pro Plus software to create digital subtraction radiographs (Fig. 5). The baseline AOI windows on the dentin lesion and normal dentin were recorded and used to measure the MD of the same area on the subtraction radiographs. To calibrate the MD gray value of the subtracted radiographs, each normal dentin MD derived from the subtraction program of pre-and post-treatment radiographs (gray value) was standardized to an intermediary gray value of 128 (adjusted MD S-control ) [18]. The MD of the lesion on the subtracted radiographs was calculated (adjusted MD S-lesion ) using the rule of three. MD subtraction was the difference between the adjusted MD S-lesion and the adjusted MD S-control . A positive MD subtraction value represented remineralization, whereas a negative value represented demineralization.
MD subtraction = adjusted MD S-lesion -adjusted MD S-control Before the subtraction analysis, J.K. was calibrated for measuring mineral density on the images by an oral and maxillofacial radiologist (S.P.) using a random 20% of the subtraction radiographs to ensure reproducibility and reliability by intra-class correlation coefficient (ICC). Each investigator separately and blindly measured the MD subtraction of the samples. When there was any difference in the measurement, the two investigators discussed and agreed on the measurement. The intra-class correlation coefficient value for the MD subtraction was 0.925. The mean MD subtraction in each group was calculated, and this value was defined as the mean mineral density difference (mMDD).

Statistical analysis
Statistical analysis was performed using SPSS software version 22.0 (IBM, Armonk, New York, USA). The Shapiro-Wilk test was used to assess the distribution of the data. One-way ANOVA was used to compare the baseline LD and MD between the groups. The mMDD between the groups determined by the subtraction method was analyzed using two-way ANOVA, generalized linear models (GLMs) with Bonferroni post hoc test. The significance level was set at p < 0.05.

Surface morphology and elemental profiling
The sterilized post-treatment blocks were taped to an aluminum stand to undergo elemental profiling with SEM-EDS after 7 d of bacterial pH-cycling. Four samples were randomly selected from each group for assessing the posttreatment elemental profile using an Energy Dispersive X-ray Spectroscopy (EDS) system (INCAEnergy, Oxford Instruments, High Wycombe., UK) with a scanning electron microscope (SEM, JSM-IT300, JEOL, Japan) at a 15 kV operating voltage. Two of the four samples were randomly selected to assess their surface morphology by SEM (Quanta 250; FEI Company, Netherlands). Thus, 8 specimens were kept dry in a desiccator cabinet before being evaluated. After the elemental profile of the surface lesion was assessed, the specimens were coated using gold sputtering and taped to the aluminum stand to evaluate the surface lesion morphology after treatment at 1000x, 10,000x, and 30,000x magnification with 20 kV.

Results
None of the 40 teeth initially evaluated using the inclusion/exclusion criteria were excluded. The mean mineral density of pre-and post-treatment among four groups is shown in Table 1. The baseline MD of each group was not significantly different from other as determined by oneway ANOVA (p = 0.081). The pre-and post-treatment mineral density of all the groups is shown in Table 1.
The pH-cycling process demonstrated demineralization because the negative control mMDD was -2.12 ± 4.08 (data not shown).
Two-way ANOVA, GLMs was used to evaluate the effects of the two factors on the mMDD of the carious dentin lesions, LC and SDF application time. The results indicated that there was no interaction between LC and application time on mMDD (p = 0.920) (Fig. 6). When each factor was evaluated separately, LC tended to influence the mMDD (p = 0.065). However, the application time did not (p = 0.249).

Fig. 5
Two-dimensional radiographs of before and after treatment, and digital subtraction radiographs (DSR); red asterisks indicate dentin carious lesions, yellow dashed lines indicate the re-mineralized areas. DSR of light cure (LC) groups demonstrate distinct re-mineralized area than those without LC Because there was no significant interaction effect between LC and application time on the mMDD, the interaction between the two factors was eliminated, and the data were re-analyzed. The results indicated that LC significantly affected mMDD (p = 0.007), represented by the difference in mMDD and standard error of -4.27(1.58). However, the application time did not significantly affect mMDD (p = 0.084), giving the difference in mMDD and standard error of -2.73(1.58). Moreover, post hoc analyses were conducted using Bonferroni correction ( Table 2).
The mMDD in the LC groups was significantly higher than those without LC ( Table 2). The 10SDF + LC group demonstrated a significantly higher mMDD compared with the 10SDF group (p = 0.041). Furthermore, the 60SDF + LC group presented a significantly higher mMDD than the 10SDF and the 60SDF groups (p = 0.010 and 0.041, respectively). Comparing application times, the mMDD in the 10SDF and 60SDF groups were not significantly different (p = 0.502). In addition, the mMDD in the 10SDF + LC group and 60SDF group were similar (p = 1.00) ( Table 2).
The EDS analysis presented higher silver and chloride peaks in the 10SDF + LC and 60SDF + LC groups than those without LC. The surface morphology of the 10SDF group had a few spherical grains sparsely aggregated on the lesion's surface (Fig. 7). Densely packed spherical grains    . 6 The interaction between two factors, light and application time, on mean mineral density difference (mMDD). The mMDD in the LC groups was significantly higher than those without LC. However, the application time did not influence the mMDD mixed with square-shaped particles tightly adhered to the inter-tubular dentin were seen in the 60SDF group. Comparing the LC and non-LC specimens, the 10SDF + LC group demonstrated more spherical grains and square-shaped particles than the 10SDF group. A similar pattern was observed comparing the 60SDF + LC and 60SDF groups.

Discussion
This study evaluated the remineralization effects of SDF with different application times and additional LC on natural dentin carious lesions in primary molars compared with its application without LC, as analyzed by DSR, EDS, and SEM. The results demonstrated that LC was the only factor related to lesion remineralization and enhanced precipitation of silver ions after SDF application. These results indicated that light improves the remineralization efficacy of SDF. Digital subtraction radiography was used to analyze the post-treatment mineral change. To establish the reproducibility and the same quality of the digital radiographs, the X-ray alignment system, digital receptor, imaging plate scanner, and exposure factors were the same throughout the study. Before comparing the subtracted radiographs, the normal dentin MD value of each radiograph was standardized to an intermediary gray value (128) of an 8-bit radiograph. Thus, this method did not require an aluminum step wedge [18]. The distinct opaque density of silver was not shown in the post-treatment images, indicating that no silver was present. We observed gray to white areas in the lesions in the subtraction images confirming that the SDF application induced frank remineralization. The light source in the present study was an LED light curing device with an output intensity of 520 mW/cm 2 , which is the highest intensity used in our pediatric dentistry clinic, and a wavelength range of 450-470 nm that triggers silver ions to precipitate into metallic silver [19]. Because the silver ions in SDF partly combine with specific amino acids in collagen in the exposed dentin and are reduced to metallic silver, the demineralized dentin turns dark [20]. When the silver particles are exposed to light, the shape of the particles tends to change from spherical to plate-like triangular, square, or hexagonal [21]. Concordantly, the SEM images of the SDF applied with LC groups illustrated more square-shaped particles on the dentin surface lesions compared with the non-LC groups. This strongly suggests that the particles observed were silver, consistent with the high quantities of silver observed from the elemental profile. However, the post-treatment radiographs did not demonstrate the total radiopacity due to silver as seen in amalgam restoration. The effect of the proportion of silver and fluoride content on mineral gain requires further investigation.
The amount of chlorine found in this study was higher than that of phosphorus. Therefore, the silver compound formed after SDF application could be silver chloride, rather than silver phosphate and may be combined with hydroxyapatite. The morphology of silver chloride particles, i.e., budding cubic particles, [22] was seen in the SDF60 + LC group. Furthermore, these particles are likely not composed of silver phosphate because silver chloride has a lower solubility than silver phosphate (8.9 × 10 -5 and 6.5 × 10 -4 g/100 ml, respectively) [23]. Silver phosphate dissolves when immersed in artificial saliva, releasing free silver ions to enhance antibacterial activity [24] and react with the chloride ions in artificial saliva to form silver chloride [23], which could play a major role in hardening the lesion surface [25]. Several studies also detected silver chloride after SDF application [3,25,26].
Seto et al. [27] concluded that the metallic silver obliterates the porous surface of the carious lesions and reinforces the structure of the remaining organic matrix, increasing the lesion's hardness. Our results demonstrated that the deepest part of almost all of the carious lesions in each group gained mineral density, i.e., re-mineralized. However, this study could not identify the specific lesion layer that was remineralized because the evaluations were performed using DSR, which provides two-dimensional results.
In the present study, the effects of light curing the SDFtreated teeth on SDF penetration were not evaluated. However, a previous study [8] investigated SDF penetration in dentin carious lesions with and without a 40 s light curing after SDF application. The study classified three different layers of dentin carious lesions using a scanning electron microscope and a stereomicroscope; infected dentin, affected dentin, and sound dentin. Their results demonstrated that SDF penetrated 60 ± 10 µm into the sound dentin in the light-cured group (test group) while penetrating 130 ± 50 µm in the group without light-curing (control). The zone with the most silver precipitation was infected dentin, which was 2.6-fold greater in the test group compared with the control group. The affected dentin caries zone presented no difference in silver precipitation in both groups. The authors concluded that SDF treatment with light curing induced more silver ion precipitation in infected dentin, causing less SDF penetration into sound dentin. Infected dentin is rich in bacteria and is where demineralization occurs. This agrees with Mei et al. [7] who reported mineral deposition 150 µm from the surface of SDF-applied dentin carious lesions in primary incisors without light curing. In contrast, Lau et al. did not find a difference in the silver penetration depth of sound dentin between light-cured and non-light-cured specimens. However, they prepared their specimens with different kinds of burs or laser ablation. Furthermore, there were no caries lesions or pH-cycling in their method [28].
The application time of SDF has been widely investigated. Despite a 1-min recommendation by the AAPD [9] and the 3-4 min recommendation by Saforide's manufacturer, application times ranging from 10 s to 3 min were used in the clinical study review by Crystal and Niederman [2]. Although Duangthip et al. [29] applied SDF for only 10 s, they demonstrated a better caries arrest rate than using 5% NaF at the same frequency. Therefore, we chose to use SDF applications times of 10 and 60 s based on Duangthip et al.'s results and the AAPD recommendation. Our results revealed that the mMDD in the 10SDF and the 60SDF groups were not significantly different, suggesting that application time did not affect remineralization. This finding agrees with the results from a clinical study that found no relationship between the length of application time and the effectiveness of SDF treatment [30]. Notably, the AAPD suggests that when a reduced application time is used, careful monitoring is needed to evaluate caries arrest and consider the need for re-application [9].
The strength of this study is the use of natural carious primary molars with active lesions, the bacterial pH-cycling model that contains cariogenic bacteria, and artificial saliva that simulated the environment in the oral cavity. Unlike when using chemical pH-cycling, SDF can exhibit its bactericidal effect in the bacterial pH-cycling model [31]. The demineralizing solution had an average pH of 6.31 ± 0.1, which is lower than the critical pH of dentin (pH 6.5) [17].
The limitation of this study is that it was performed in vitro. Although our method mimics the oral environment to some extent, the results may not be extrapolated to the clinical situation. Furthermore, EDS analysis could be performed only at the surface of the lesions because specimen preparation required nail varnish coatings to prevent demineralization in unwanted areas during the bacterial pH-cycling. Therefore, the elemental profile of the lesions beneath the surface could not be assessed. The findings in the present study have practical implications for managing dentin caries in young or uncooperative children and specialneeds patients to reduce chair time. However, clinical studies with long-term follow-ups are suggested to confirm these results before clinical use to improve clinical outcomes by reducing SDF application time through using additional LC.

Conclusion
Light curing enhances the remineralization efficacy of SDF, as shown by DSR, EDS, and SEM analyses. The remineralization due to applying SDF on the lesions for 10 s with 20 s LC (total time of 30 s/cavity) was not significantly different from those treated using a 60 s application time with or without LC. The results could be valuable for managing dentin caries in young or uncooperative children and special-needs patients by reducing the SDF application time by 50% of the AAPD recommendation while maintaining an adequate SDF remineralization effectiveness.