Interaction between Hexametaphosphate, Other Active Ingredients of Toothpastes, and Erosion-Abrasion in Enamel in vitro

It is generally accepted that the application of the combination of stannous and fluoride ions is standard in the prevention of dental erosion [Schlueter et al., 2010; Lussi and Carvalho, 2015; Ganss et al., 2016; West et al., 2021]. In the search for active agents which further enhance these anti-erosive effects, sodium hexametaphosphate (HMP) as a chelator of divalent cations was found to have promising properties [Kura et al., 1974; De Kort et al., 2009]. It is used in various products, for example, to remove limescale in dishwashers [Schwartz and Gilmore, 1934]. Toothpaste manufacturers often utilize HMP for its cleaning enhancing effects [He et al., 2007]. By binding to calcium, it also has the ability to form layers on the enamel surface, which are claimed to prevent erosive tissue loss [Hooper et al., 2007]. Added to toothpastes-containing stannous and fluoride ions as anti-erosive agents, it is supposed to act as a stabilizer. Stannous ions have the highest affinity to hydroxyapatite in their divalent form [Claessens and Kolar, 2000], and HMP should prevent oxidation processes which transform the ion to its less active tetravalent form. As HMP can potentially bind calcium ions, it is not only able to form layers on tooth surfaces but it might also bind calcium ions dissolved by acids in a stable chelate complex. As a result, the addition of this molecule to toothpastes could carry the risk of increasing erosive tissue loss as shown in a previous study [Schlueter et al., 2016], by removing calcium ions from the equilibrium. Various properties of HMP can therefore possibly influence the anti-erosive but also the combined anti-erosive/anti-abrasive effect of stannous- and fluoride-containing toothpastes.

There is a study suggesting concentration-dependent anti-erosive effects of HMP on hydroxyapatite discs in a pH-stat model [Do Amaral et al., 2016]. However, the impact of the HMP concentration on the erosive tissue loss on human enamel has not been investigated, yet. Fluoride and stannous ions combined with HMP have been studied, contained in various commercially available toothpastes [Hooper et al., 2007; Barlow et al., 2009; Schlueter et al., 2016]. However, due to the complex formulation of the toothpastes, it is not clear whether the effects found can be attributed to HMP, to fluoride and stannous ions, or to other ingredients. Therefore, the aim of this in vitro study was to investigate the impact of different HMP concentrations on anti-erosive or anti-erosive/anti-abrasive properties of experimental preparations containing fluoride or the combination of stannous and fluoride ions (F/Sn) on enamel. The null hypothesis was that the concentration of HMP has no impact on tissue loss, independent of other active agents added to a standardized fluoride-free toothpaste suspension.

Three hundred and twenty enamel specimens were prepared from previously impacted human third molars. Different dental practitioners extracted the teeth for medical reasons and immediately put them anonymously into saturated thymol solution until further use. All donors gave verbal informed consent. The collection and the use of the teeth in the study were approved by the Local Ethics Committee (Ethik-Kommission Albert-Ludwigs-Universität Freiburg, No. 352/16) and followed the Declaration of Helsinki (2013) as well as relevant guidelines and regulations. Longitudinal enamel slices were cut (Exakt Trennschleifsystem, Exakt Apparatebau GmbH und Co. KG, Norderstedt, Germany) from the crown and the natural enamel surfaces were ground flat and polished (SiC Paper #1200, Struers ApS, Ballerup, Denmark; Polyesterfilms SiC P1200 and P2500, Exakt, Norderstedt, Germany; PET polishing film SiC, 3 μm, Microdiamant GmbH, Kempten, Germany), removing approximately 150 μm of the outer enamel. The slices were further cut (diamond coated separating disc 911HP, Komet Dental, Gebr. Brasseler GmbH & Co. KG, Lemgo, Germany) to obtain rectangular enamel surfaces of at least 2 × 3 mm. Specimens with cracks, fluorosis, demineralisations, scratches, or those with more than 0.5 μm deviation from flatness on a 2-mm trace were discarded. The specimens were divided into 10 groups at random (n = 32 each). Specimen holders designed for a brushing machine were equipped with eight specimens each, resulting in four holders per group using light-curing resin (Technovit 7230 VLC, Heraeus Kulzer, Wehrheim, Germany). One half of each enamel surface was covered with the same light-curing resin to remain untreated as reference area.

A 0.5% citric acid solution was prepared with 5 g/L citric acid monohydrate (Sigma-Aldrich, St. Louis, Belgium). The solution was used with native pH (2.4–2.5).

A mineral-salt solution was used for intermediate storage and for preparation of toothpaste suspensions (common mineral-salt solution). It contained 0.4 g/L H₃PO₄ (Ortho-Phosphoric acid 99%), 1.0 g/L NaHCO₃ (sodium hydrogen carbonate), 1.5 g/L KCl (potassium chloride), and 0.22 g/L CaCl₂ (calcium chloride, Carl Roth GmbH + Co. KG, Karlsruhe, Germany), according to previous studies [Gerrard and Winter, 1986].

In addition, a concentrated mineral-salt solution with a 25% higher concentration of the mineral salts was prepared with the same ratio between components as in the above-mentioned mineral-salt solution. It contained 0.5 g/L H₃PO₄ (Ortho-Phosphoric acid 99%), 1.25 g/L NaHCO₃ (sodium hydrogen carbonate), 1.88 g/L KCl (potassium chloride), and 0.275 g/L CaCl₂ (calcium chloride, Carl Roth GmbH + Co. KG, Karlsruhe, Germany). The pH of both mineral-salt solutions was adjusted to 6.7 with NaOH.

Three different solutions with the active agents fluoride (F) and stannous ions (Sn) and the stabilizer sodium gluconate were prepared. One contained only 31 g/L sodium gluconate (Glu), one contained 31 g/L sodium gluconate and 11 g/L sodium fluoride (Glu/F), and one contained 31 g/L sodium gluconate, 11 g/L sodium fluoride, and 17.1 g/L tin (II) chloride dehydrate (Glu/F/Sn). The solutions were used with native pH, which were 6.9 for Glu, 7.4 for Glu/F, and 4.8–4.9 for Glu/F/Sn.

Sodium HMP (crystalline, 96%, Sigma-Aldrich, St. Louis, Belgium) solutions were prepared in three different concentrations: 33 g/L, 233 g/L, and 433 g/L with pH values of 6.9, 6.4, and 6.2, respectively. The pH was not adjusted. For the toothpaste suspensions, 3 g of each of the active agent solution (Glu, Glu/F, or Glu/F/Sn) was combined with 3 g of each concentration of the sodium HMP solutions to yield nine test solutions. A tenth solution only containing sodium gluconate was used as negative control (NegContr). To each of these ten solutions, 24 g of concentrated mineral-salt solution was added and each was mixed with 10 g of fluoride-free toothpaste (Lavera Neutral Zahngel, Laverana GmbH & Co. KG, Wennigsen, Germany) to a homogeneous suspension. The ratio of toothpaste to solution was 1:3. The target values in the toothpaste suspensions were 0.23% sodium gluconate, if present 0.8% sodium fluoride (373 ppm F⁻), and 1.3% tin (II) chloride dehydrate (800 ppm Sn²⁺) as well as 0.25%, 1.75%, and 3.25% sodium HMP (equivalent to concentrations in the non-diluted toothpaste: 0.93% gluconate, 3.3% sodium fluoride [1,492 ppm F⁻], 5.1% tin (II) chloride dehydrate [3,200 ppm Sn²⁺], and 1%, 7%, and 13% sodium HMP).

All solutions and suspensions were prepared twice daily (before each application) except for the common mineral-salt solution and the citric acid, which were prepared once daily. The toothpaste suspension was stirred during the whole preparation procedure and until use in order to ensure homogeneity. All products were purchased at Merck KGaA (Darmstadt, Germany) if not otherwise stated.

In a cyclic erosion and erosion-abrasion model lasting 10 days, the specimens of all groups were eroded six times a day for 2 min with the 0.5% citric acid solution at 25°C. Twice a day, after the first and the last erosive challenge, the 32 specimens per group were immersed in one of the ten toothpaste suspensions for 2 min. During this immersion time, half of the specimens of each group (n = 16) was additionally brushed for 15 s in a brushing machine (Zahnbürstsimulator ZM-3.4, SD Mechatronik GmbH, Feldkirchen-Westerham, Germany; linear movement pattern, 25 oscillations, 13 mm travel path, 60 mm/s travel velocity, load 200 g) equipped with manual toothbrushes (elmex 39 toothbrush medium, CP GABA GmbH, Hamburg, Germany). In-between cycles and over-night, the specimens were stored for at least 1 h in the common mineral-salt solution. The solutions were applied in a shaking water bath (Shaking water bath 1,086, GFL Gesellschaft für Labortechnik mbH, Burgwedel, Germany) at 35 horizontal movements per minute. The specimens were rinsed for 30 s after application of citric acid and for 60 s after application of toothpaste suspensions or mineral-salt solution with tap water. At weekend and after the last experimental procedures, the specimens were stored in a wet chamber at 4°C until analyses.

After carefully removing the coverages of the reference areas, the tissue loss was measured with an optical profilometer using the measuring software FRT Acquire and analysed with the software FRT Mark III (Fries Research & Technology GmbH, Bergisch-Gladbach, Germany). Three parallel traces were measured on each specimen orthogonally to the border between reference and experimental area. The traces were made at 0.2 mm intervals and had a length of 2 mm, with 1 mm on the reference and 1 mm on the experimental area (200 Pixel/2 mm, 1,000 Hz). Two parallel regression lines were constructed through the outermost 0.5 mm of the reference and the experimental side of each trace. The vertical distance between the two regression lines was determined. The mean value of the distance of the three traces was defined as tissue loss of the specimen and given in µm.

The surfaces of all specimens were analysed with energy dispersive X-ray spectroscopy (EDX). Three eroded and three eroded-abraded specimens from each group were additionally selected randomly for evaluation of the surface morphology by scanning electron microscopy (SEM). For EDX and SEM, the specimens were air-dried and for SEM sputtered with gold (Sputter coater SCD 050, Bal-Tec AG, Balzers, Liechtenstein; sputter time: 60 s; amperage: 60–65 mA). All SEM analyses were performed with the secondary electron detector and the EDX analyses with an EDX detector of the JSM-IT100 (Jeol GmbH, Freising, Germany) according to the analysis software JSM-IT100 Vers. 1.09 (Jeol GmbH, Freising, Germany). All settings were kept constant (15 kV acceleration voltage, 10 mm working distance, 0° tilt angle and probe current adjusted between 45 and 50 to obtain a count rate of approximately 1,000 counts per second). The relative amount (percentage by weight, wt%) of the elements carbon, oxygen, silicon, phosphorus, calcium, and tin on the surfaces were measured with standardless EDX at 1,000-fold magnification. SEM images of representative areas were taken at 2,500-fold magnification. The brightness and contrast of the SEM images were adjusted with the software GIMP version 2.10.14.

According to previous studies, a sample size of 16 per treatment condition (combination between active agent used and erosion or erosion-abrasion) was chosen. Samples size was calculated for previous studies with SPSS (IBM, Armonk, NY, USA) under the assumption that a clinically relevant difference of 3.5 μm should be detected (standard deviation: 3 μm, α = 0.05, 1-β = 0.8, t tests for independent samples).

For descriptive analysis, mean values and standard deviations were computed. Mean tissue loss and EDX values (only phosphorus and tin) were analysed for group differences using one-way ANOVA. In subsequent pairwise comparisons, a correction was made for multiple testing using the Tukey method. To investigate the influence of the HMP concentration on the mean tissue loss for erosion and erosion-abrasion in more detail, the association between HMP concentration and mean tissue loss was presented graphically. The increase in tissue loss is not linearly dependent on the HMP concentration but is the square root of the concentration. Hence, to investigate the influence of the HMP concentration on the mean tissue loss, a linear regression analysis with the square root of the HMP concentration and group (no F, F, or F/Sn) as covariates was carried out without the negative control. To quantify the effects of tin on tissue loss under erosion and erosion-abrasion conditions, a linear regression analysis was performed including all HMP concentrations. The level of significance was set to 0.05. Scatter plots were used for graphical presentation. All computations were done with STATA (Version 17.0, College Station, TX, USA).

An overview of the profilometrically measured tissue loss is shown in Table 1. The erosive conditions induced a tissue loss of 10.9 ± 2.0 μm (mean ± standard deviation) in the control group. All groups containing 0.25% HMP reduced tissue loss statistically significantly in comparison to the control by 28–54% (p < 0.026), irrespective of any active agent added (no fluoride, fluoride [F], or stannous in combination with fluoride ions [F/Sn]). With higher concentrations of HMP only 1.75% HMP/F/Sn was significantly more effective than the control by 28% (p = 0.022). None of the others reduced tissue loss significantly, 3.25% HMP without F and F/Sn even increased it by 35% (p = 0.002).

Under erosive-abrasive conditions, the mean tissue loss of the control group was 22.2 ± 1.6 μm and all combinations of HMP with active agents lost their efficacy. The tissue loss increased significantly in all groups with 3.25% HMP by 15–30% (p < 0.011) and in all groups with no other active agent than HMP by 23–30% (p < 0.001).

Considering only the active agents (no F, F, or F/Sn) in one common analysis adjusting for group, a significant positive association between tissue loss and HMP concentration was observed (p < 0.001 for erosion and erosion-abrasion). The common regression coefficient showed an increase of tissue loss with increasing square root of the HMP concentration by 4.2 μm in the erosion groups and by 2.5 μm in the erosion-abrasion groups. This is shown in Figure 1a for each active agent separately.

The percentages by weight of the element phosphorus on the surfaces ranged between 17.22 ± 0.34 wt% (0.25% HMP/F/Sn) and 18.08 ± 0.50 wt% (1.75% HMP/F) in the erosion groups and between 17.35 ± 0.43 wt% (0.25% HMP/F/Sn) and 18.07 ± 0.50 wt% (3.25% HMP/F) in the erosion-abrasion groups. Statistically significant differences were found neither when groups without Sn were compared with each other nor when groups with Sn were compared with each other (p ≥ 0.05). Even though significant for 0.25% HMP/F/Sn (p = 0.013), the differences between groups were, all in all, very small and of minor relevance.

In the groups treated with Sn, the tin retention on the surfaces ranged between 1.35 ± 0.18 wt% (3.25% HMP/F/Sn) and 1.75 ± 0.30 wt% (0.25% HMP/F/Sn) in the erosion groups and between 0.63 ± 0.22 wt% (3.25% HMP/F/Sn) and 1.51 ± 0.26 wt% (0.25% HMP/F/Sn) in the erosion-abrasion groups. Within these groups, statistically significantly more tin was retained on specimens treated with 0.25% HMP/F/Sn under both erosive and erosive-abrasive conditions (p < 0.001). However, no differences were found between 1.75% HMP/F/Sn and 3.25% HMP/F/Sn (erosion: p = 0.998; erosion-abrasion: p = 0.755).

The negative association between the tin retention and the tissue loss within the F/Sn groups with a regression coefficient of −2.6 was not statistically significant under erosive conditions (p = 0.059). However, with additional abrasion, more tin retention led to significantly less tissue loss with a regression coefficient of −3.5 (p = 0.003) (shown in Fig. 1b). The tissue loss is thus reduced by 2.6 μm per 1 wt% tin under erosive conditions and by 3.5 μm under erosive-abrasive conditions.

The SEM images of the erosion groups show amorphous deposits on the surfaces even in the control group. While in the groups with only HMP deposits cover the surfaces homogeneously, in the groups with F and F/Sn, the underlying eroded surfaces can partially be seen.

The abraded surfaces are smoother. In groups without F and F/Sn, hardly any structure is left on the roughened surfaces, especially in the 3.25% HMP group. F and F/Sn seemed to be able to preserve at least parts of the prism structure of the enamel (shown in Fig. 2).

The erosive and abrasive cycling was supposed to mimic exogenous erosive challenges by regular consumption of erosive drinks. The study conforms with recommendations for in vitro erosion and erosion-abrasion studies [Shellis et al., 2011; Wiegand and Attin, 2011; Schlueter et al., 2016].

Concentrations of the active agents fluoride, stannous ions [Schlueter et al., 2013, 2016], and sodium HMP [Wefel et al., 2002] in toothpastes slurries were chosen according to commercially available formulations. In order to avoid premature interactions with the ingredients of the toothpaste, separate solutions containing the active agents were prepared and added to the toothpaste in the last preparation step shortly before applying the suspensions to the specimens.

Sodium gluconate was added to stabilize the stannous ions in their divalent form. Like sodium HMP, it is a chelator with the ability to complex polyvalent cations like calcium [Gyurcsik and Nagy, 2000; Phadungath and Metzger, 2011]. In order to allow comparison between the various groups and to focus on the effect of the three active agents, fluoride ions, stannous ions, and HMP, sodium gluconate was added to all solutions.

The different pH values of the sodium gluconate solutions (no F = 6.9, F = 7.4, and F/Sn = 4.8–4.9) might be considered as shortcoming, as the pH value itself has an effect on the anti-erosive effects of fluoride and stannous ions. At lower pH, more calcium-fluoride-like precipitates are formed [Scholz et al., 2019], which have a slight anti-erosive potential [Huysmans et al., 2014]. However, adjusting the F/Sn solution was not possible because the stannous ions are hardly stable in their divalent form at a neutral pH [Hefferren, 1963]. Lowering the pH of the other groups would potentially have led to more erosive tissue loss in the groups without F and F/Sn.

Under erosive conditions without additional abrasive forces applied, sodium HMP has most likely properties such as adhesion to enamel, which may reduce the demineralizing effect of the acid. With its numerous functional groups, its persistence is particularly high, as it can bind to more than one Ca²⁺ ion, and the probability of each functional group loosening simultaneously is low [Van Dijk et al., 1980]. The resulting layers are stable enough to withstand rinsing with water and impede acids from diffusing to the surface [Da Camara et al., 2014]. Consequently, demineralization-inhibiting effects have been shown for HMP in a pH-stat model [Do Amaral et al., 2016] and in different in vitro caries models [Van Dijk et al., 1980; Da Camara et al., 2014; Mohammadipour et al., 2019]. The present study shows for the first time that HMP in a concentration of 0.25% without F⁻ or Sn²⁺ is able to reduce erosive tissue loss on human enamel. However, HMP lost its anti-erosive efficacy at a concentration of 1.75% and tissue loss even increased with 3.25% HMP. A potential for polyphosphates to disaggregate calcium ions from the hydroxyapatite crystal lattice has long been discussed [McGaughey and Stowell, 1977]. The reason for this might be a process called ligand-promoted dissolution [Shellis et al., 2014]. When chelating anions from a solution attach to polyvalent cations arranged in a crystal lattice, complexes are formed. The anions provide electrons weakening the bonding between the cations and the crystal lattice. Although these complexes might still adhere to the surfaces, detachment is facilitated [Furrer and Stumm, 1986; Shellis et al., 2014]. HMP is also a chelating anion and able to form strong complexes with Ca²⁺ ions [Changgen and Yongxin, 1983]. Therefore, it is likely and in accordance with the findings of the present study that ligands of HMP promote dissolution of Ca²⁺ ions from the enamel surface. Whether HMP dissolves Ca²⁺ ions from the surface or stabilizes them seems to depend on its concentration, as 0.25% HMP reduced tissue loss under erosive conditions while 3.25% HMP increased it. It has to be born in mind that steric effects play an important role when chelating anions form complexes with cations. Whether both ligands of a chelator can bind to the cation depends on its arrangement at the surface [Perry et al., 2004, 2005]. It might be speculated that the roughness and the higher surface reactivity of erosively altered enamel facilitate ligand-promoted dissolution. Therefore, the results of this study should not be transferred one-to-one to sound enamel.

Sodium fluoride itself has a certain anti-erosive effect mainly attributed to formation of calcium-fluoride-like precipitates on tooth surfaces, which for one thing directly protect the underlying enamel surface, then again release fluoride ions in case of an erosive challenge [Huysmans et al., 2014]. In vitro studies found ranges of 28–49% [Ganss et al., 2016], 19–35% [Ganss et al., 2011], and 27% [Moretto et al., 2010] tissue loss reduction on human or bovine enamel for different toothpaste formulations. The combination of fluoride with HMP might affect demineralization through additional mechanisms. On the one hand, higher retention rates of CaF⁺ and Ca²⁺ on the enamel surface are suggested, which might be released during an acid impact [Conceição et al., 2015]. On the other hand, the addition of fluoride to HMP increases its complexing ability [Changgen and Yongxin, 1983]. Synergistic effects could be shown in a pH-stat model in which the combination of 1% HMP with 1,100 ppm fluoride led to a lower dissolution rate compared to each active agent on its own. However, as tissue loss was not measured, potential effects of a ligand-promoted dissolution by HMP were not investigated [Do Amaral et al., 2016]. In an in situ study, the actual erosive tissue loss was reduced by adding 9% HMP to 1% sodium fluoride compared to fluoride only [Conceição et al., 2015]. In contrast, the present study found no significant differences when fluoride was added to any of the HMP concentrations under erosive conditions. The contrary findings might probably be explained by the different concentrations of HMP and fluoride as well as by the impact of salivary pellicle.

Stannous ions in combination with fluoride ions (F/Sn) are known for their anti-erosive potential in solutions [Hove et al., 2007; Ganss et al., 2010] as well as in toothpastes [Schlueter et al., 2013]. Toothpastes-containing F/Sn reduced erosive tissue loss by ranges of 53–66% [Ganss et al., 2016] up to 55–78% [Ganss et al., 2011] in vitro. Various mechanisms are responsible for its efficacy. In the presence of fluoride and stannous ions, tin-rich precipitates are formed on the enamel surface, which resist erosive challenges [Ganss et al., 2008]. In addition, the amount of calcium-fluoride-like precipitates is also increased compared to an application of fluoride without stannous ions [Wegehaupt et al., 2012]. Furthermore, stannous ions are incorporated in the superficial enamel layers under erosive conditions thereby enhancing the acid resistance of the surface [Schlueter et al., 2009]. Key to these anti-erosive effects is the presence of stannous ions in their divalent form. As these ions are reactive and thus rather unstable, their stabilization plays a critical role for maintaining their efficacy [Mühlemann and Schmid, 1985]. As a chelating anion, HMP has the potential to complex and therefore to stabilize the divalent form of stannous ions [Wefel et al., 2002]. However, the findings of different studies are contradicting. On the one hand, two studies showed anti-erosive effects for toothpastes containing stannous fluoride and HMP in vitro [Hooper et al., 2007] and in situ [Barlow et al., 2009]. As the active agents, however, have not been studied separately, the effects could be attributed solely to the stannous fluoride. On the other hand, an in vitro study found better anti-erosive effects for an SnF₂ gel without HMP than for a toothpaste containing NaF/SnF₂ and HMP [Schlueter et al., 2016]. As HMP is able to form complexes with divalent cations, it is conceivable that in some cases it complexes the available stannous ions and removes them from the equilibrium, most likely if the concentration exceeds a defined value. Beyond this value the stannous ions are not available anymore for an interaction with the enamel surface, with the consequence that the tissue loss increases. The EDX analyses of the present study corroborate this assumption, as the tin retention decreased with higher HMP concentrations (shown in Table 1; Fig. 1b). Thus, F/Sn combined with 1.75% and 3.25% HMP lost its anti-erosive efficacy in contrast to its combination with 0.25% HMP under erosive conditions. Furthermore, the ratio between dissolved calcium ions and free stannous ions might also be of relevance – perhaps by competing for binding sites at the HMP. The amount of calcium ions dissolved might depend on the erosivity of the acids used for erosive demineralisation. Further studies including different types of acid and/or different concentrations might give deeper insights into this issue.

Brushing of sound enamel has hardly any effect, even when performed regularly over a lifetime. Erosively altered enamel surfaces are, however, highly susceptible to mechanical wear, thereby abrasive forces accelerate erosive tooth wear [Shellis and Addy, 2014]. HMP might further enhance tissue loss, as the complexes it forms with Ca²⁺ are only loosely attached to the enamel surface [Furrer and Stumm, 1986]. While under erosive conditions, these precipitates act as a chemical barrier and decrease tissue loss, they are most likely worn off by abrasive forces. Thereby, not only calcium which is complexed by HMP is lost from the enamel but also the underlying enamel is exposed to the following erosive challenges. Accordingly, in an in vitro study using 1% citric acid, the difference between erosion and erosion-abrasion was greatest with a toothpaste containing HMP [Schlueter et al., 2016]. The present study corroborates these findings as groups without F or F/Sn, and groups with 3.25% HMP lost more dental hard tissue than the negative control under erosive-abrasive conditions. Therefore, ligand-promoted dissolution by HMP has the potential even to enhance erosive-abrasive tissue loss. Considering all results, the null hypothesis that the HMP concentration has no impact on tissue loss, independent of the active agents used, has to be rejected.

The study clearly shows that there is an impact of concentration of HMP on tissue loss, which has to be born in mind for the development of oral care products. However, one shortcoming of the study is that no proteins were added to the storage solution, which might have an impact on results. It could be that the proteins (mucins or other proteins from saliva and pellicle) react with the phosphate groups of the polyphosphate, potentially changing the number of binding sites. This on the one hand could reduce the calcium-binding potential and following, the effect of the polyphosphate could potentially be attenuated. On the other hand, the polyphosphate could be retained within the pellicle, what could lead, by its complexing properties, to a higher retention of calcium within the pellicle and following, directly at the tooth surface [Do Amaral et al., 2016]. Ex vivo or in situ studies would give more information on this issue.

Conclusively, adding HMP in specific concentrations to toothpastes used as an anti-erosive therapeutic approach bears the risk of enhancing tissue loss by ligand-promoted dissolution. Calcium ions can be complexed by HMP and detached from the hydroxyapatite crystal lattice of the enamel. Stannous ions might also be complexed by HMP inhibiting their protective effects. These complexations of both calcium and stannous ions seem to be concentration dependent. While under erosive conditions, fluoride and stannous ions were able to compensate the negative effects of HMP at least partially, no beneficial effects of any active agent combination were found with additional abrasion, at least under in vitro conditions.

The teeth used for this study were extracted for medical reasons, and all donors gave verbal informed consent. In accordance with the Local Ethics Committee, written informed consent was not required. The collection and the use of the teeth in the study were approved by the Local Ethics Committee (Ethik-Kommission Albert-Ludwigs-Universität Freiburg, No. 352/16) and followed the Declaration of Helsinki (2013) as well as the statement of the German Ethics Committee (2003) for the use of human body material in medical research.

The authors have no conflicts of interest to declare.

Funded by the Division for Cariology, Department of Operative Dentistry and Periodontology, Center for Dental Medicine, Medical Center – University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany.

Benedikt Luka contributed substantially to the design of the study, supervised the acquisition and analysis of the data for the work, interpreted it, and drafted the work. Andrea Duerrschnabel contributed to the design of the study, acquired the data and critically revised it. Sina Neumaier contributed to the design of the work, performed all practical procedures, acquired the data for the work, contributed to its interpretation, and critically revised it. Nadine Schlueter conceived and designed the work, supervised the acquisition, analysis and interpretation of the data for the work, and contributed substantially to the drafting of the work. Kirstin Vach analysed and interpreted the data and contributed to the drafting of the work. All authors approved this version of the manuscript and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Nadine Schlueter and Kirstin Vach contributed equally to this work.

Due to legal reasons data are available on request. Further enquiries can be directed to the corresponding author.