There is limited evidence on the effect of different toothpastes containing either sodium fluoride or low concentration fluoride in bioactive glasses on root caries. To simulate the caries process, artificial saliva rather than simulated body fluid was used in this study. The immersion solution was changed at each tooth brushing time point to simulate the effect of saliva exchange under the in vivo conditions. Furthermore, each sample for the study groups was derived from the same tooth. Interestingly, the amount of mineral density was different from the baseline to ARCLs in each group (Table 3). This means that there were variations in different parts of the same tooth.

The mineral density increases at the surface of ARCLs in Groups 1–3 demonstrated the ability of toothpastes to precipitate minerals and potentially remineralise the surface of ARCLs. This outcome directly corresponded to the surface mineral density changes in the line profiles. The mineral density increase was also observed within the lesion surface in all three toothpaste groups, which supported the previous study [24]. The study reported an increase of 22% in the mineral density by using the 1,450ppm fluoride toothpaste with remineralisation solution alone for a period of 45 days [24]. The pH-cycling model was designed to evaluate the effect of fluoride in either reducing demineralisation or enhancing remineralisation [26]. Chen et al., (2023) showed the significant increase of hardness within root carious lesions following the use of toothpaste treatment with 13 days pH-cycling. Regarding the Knoop hardness, the penetration was up to 0.1 mm depth, which was lower than the real mineral density change (0.4 mm depth). Although the BG with 540 ppm F was proven to show the highest hardness in all toothpaste groups, the mineral density was low in comparison to the toothpastes either containing 5,000 ppm or 1,450 ppm F. Interestingly, the hardness measurements and mineral density were consistent in deionised water, 1,450 ppm, and 5,000 ppm F groups [17]. However, each technique used in that study had limitations in relation to the structural changes within the ARCLs according to formation of apatite using X-Ray diffraction analysis (XRD) following the use of different toothpastes.

In this current study, the XMT showed the increase of mineral density for all toothpaste groups during the 13 days pH-cycling in accordance with the previous study. In addition, the artificial root caries was used rather than natural one in this study. Artificial root caries without bacteria and enzymes could produce lesions most closely resembling natural lesions. Although the artificial root caries is a dead sample, the similarities between artificial and natural root caries in the demineralising system may indicate that this is not necessarily a major drawback in attempts to obtain a working model [27]. Qi et al., (2021) used the similar composition pH-cycling model for 14 days to prepare the artificial occlusal carious lesions by XMT assessment [22]. The authors reported the mineral loss with an approximately 0.3 mm thickness from surface to subsurface, which was similar mineral loss in this study by 5 days at the pH 4.8 demineralisation. As remineralisation involving in the pH-cycling model increases the hardness and mineral density, the demineralisation rate can be reduced. Therefore, pH 4.8 demineralisation solution only is recommended to form ARCLs.

As fluoride is a well-known remineralising agent, it can be speculated that this could be related to the acid resistant and less reactive fluorapatite layer [28]. This might suggest the precipitation of apatite mineral onto the ARCL surface as a result of hydroxyl ions being substituted (at least partly) by fluoride. This could explain the formation of either fluoride-substituted apatite or fluorapatite. The 19F-MAS-NMR signal between − 101.0 and − 107.0 ppm was used to identify the fluoride environment. The − 105.3 ppm chemical shift in the Group 1 indicated the origin of fluoride from the tooth itself as this tooth did not receive any fluoride application.

Gao et al., (2016) showed that 19F-MAS-NMR chemical shift plotted against the F percentage in the apatite. The 100% highly fluoridated apatite (fluorapatite) presents at -103.5 ppm chemical shift, whilst − 105.0 ppm would indicate a 40% fluoridated apatite [29]. The 19F-MAS-NMR spectra obtained for Group 2 at -105.1 ppm represented as fluoridated apatite. However, the formation of CaF2 would not be expected with toothpaste containing low fluoride as seen in the Group 2. In this study, the slight shoulder at -108.6 ppm seems to be an artefact. It can also be speculated that this might be formed following the use of toothpaste containing high fluoride prior to tooth extraction. In this respect, Chen et al., (2023) reported that the concentration of fluoride was less than 10 ppm after the toothpaste treatment with rinsing [17]. However, the slurries were diluted with water using the toothpastes containing 5,000 ppm F, 1,450 ppm F and BG with 540 ppm F and rinsed out after treatment in this present study. In addition, Mohammed et al., (2013) indicated the formation of CaF2 at over 45 ppm fluoride concentration. The presence of CaF2 has also been reported to exhibit anti-caries effect by forming a physical barrier on the enamel surface, thereby slowing the demineralisation process, and serving as a reservoir for fluoride [30, 31].

The 1,450 ppm and 5,000 ppm F toothpastes have intense peaks at -103 ppm. Mohammed et al., (2013) reported that the 19F MAS-NMR presented a sharp peak when enamel block is immersed in 11 ppm F [32]. The fluoride concentration for these two toothpastes in this study after rinsing was around 1–2 ppm [17], which resulted in fluorapatite as the major chemical species at low fluoride concentrations. Low fluoride levels found in saliva can significantly reduce enamel demineralisation, and those found in dental plaque fluid have a potential remineralisation effect [33].

Ekstrand et al., (2013) previously reported that the effect of 5,000 ppm fluoridated toothpaste was significantly more effective for arresting root carious lesion progression and promoting remineralisation compared to the 1,450 ppm fluoride toothpaste [7]. The mean numbers of hard lesions were 2.13 (1.68) in the 5,000 ppm fluoride toothpaste and 0.61 (1.76) in the 1,450 ppm fluoride toothpaste (p < 0.001). This current laboratory-based study also indicated the high mineral density in the toothpaste containing 5,000 ppm fluoride when compared to the 1,450 ppm one.

The present study demonstrated a decrease of mineral density in the subsurface for all toothpaste samples (Fig. 2). Previously, Ten Cate and Arends (1981) reported the blockage of surface layer pores as a result of fluoride-enhanced deposition within the surface [34]. Farooq et al., (2015) reported that BG toothpaste can occlude the dentinal tubules after simulated tooth brushing using SEM [35]. Once the remineralisation occurred at the beginning of pH-cycling, the pores of root dentine surface might have been obliterated. Therefore, the subsurface lesions would have been failed to remineralise further, since the formation of fluoride-substituted apatite was unable to pass through the surface to reach the subsurface. This was supported by the BSE images for high minerals at the edge of dentinal tubules however not within the dentine tubules. However, the SEM and BSE images also showed that few irregular white particles were noticeable on the subsurface of each sample rather than embedded in the dentinal tubules. It can be speculated that this could also be related to the polishing paste during the cutting and polishing process for the SEM analysis.

XMT was in 15 × 15 × 15 µm3 voxels, avoiding the errors in determining the sample thickness. In this respect, Davis et al., (2018) reported the XMT could be a useful method to evaluate demineralisation and remineralisation [36]. However, the XMT is unable to distinguish the mineral element to investigate the density change(s). The 19F MAS-NMR can indicate the formation of fluoridated apatite or fluorapatite, however the technique only analyses the powdered samples. It should be noted that each technique used in this study demonstrated some limitations, however they also complemented and supported each other with respect to the mineral density changes and fluorapatite formation. However, the system fails to distinguish the position of fluorapatite formation between the surface and subsurface. In addition, this study compared the SEM images for each group, however these images were not recorded at baseline and after the development of ARCLs due to the destructive nature of this technique. Theoretically, these samples would also have been exposed to fluoride environment in the oral cavity prior to extraction, which might have caused the variation of fluoride ions within the study teeth.

The small sample size could another limitation of this study. It should be noted that the samples are their own controls reducing the need for a large sample size required to account for variability in baseline mineral concentration since the XMT scanner was set at 15 μm voxel size/resolution (3D) for 24 h. The carious lesions for each sample were approximately 3 mm x 4 mm, therefore these high-resolution images can capture the full range of pore sizes in a representative elementary volume for the lesions. This unique XMT employed time-delay integration (TDI) to avoid ring artefacts and facilitate high signal-to-noise ratio imaging. Conventionally Transverse Microradiography (TMR) would have be used for quantifying the mineral density, however this technique requires a 100 μm thin tooth section. In addition, the grinding samples can produce undefined artefacts at the demineralised areas. The TMR technique involves the destruction of samples and scans of the same section which were unable to be detected before and after the intervention.

In addition, the XMT results were based on hundreds of slices to reconstruct the 3D images. This system can measure the same sample from the same slice at the start and end of the study which would provide a real comparison. The line scans were representative of the hundreds of slices. However, the variation in the mineral content of the teeth before the experiment would not have been seen in a conventional TMR study. Zain et al., (2020) also provided the evidence for the mineral changes in demineralised dentine for two teeth only using the same non-destructive 3D-XMT [37]. In conclusion, the TDI XMT takes 10 times longer than the conventional XMT, which would reduce the number of teeth, that can be evaluated.

In future laboratory-based studies, unerupted extracted wisdom teeth without any exposure or extracted teeth with root caries can be considered to mimic real life situation. In addition, the 19F-MAS-NMR might be used to assess the fluoride levels. Studies conducted without rinsing would also be interesting, since rinsing following tooth brushing is not recommended for the fluoride retention [17].