Curvature-dependent shear bond strength of different attachment materials for orthodontic lingual indirect bonding

To evaluate the shear bond strength (SBS) of different attachment materials used for lingual bonding, the influence of artificial aging and the radii of curvature of the enamel surface on SBS, 192 third molars were photographed to determine the radius of curvature of the oral surface. After phosphoric acid etching a cylindrical test piece was bonded to the oral enamel using a mold that was filled with a chemically curing (Maximum Cure, Transbond IDB Premix) or a dual-curing (Nexus NX3, RelyX Unicem2) attachment material. SBS was tested after 24 h, 500 thermal cycles or 90 days at 37 °C with a universal testing machine. Computed tomography scans were performed to determine the bonded surface and calculate SBS. Values ranged from 8.3 to 20.9 MPa. RelyX Unicem2 showed the highest SBS values at baseline, 500 thermal cycles and after 90 days (p < 0.001). Ninety days of wet storage significantly reduced SBS of Maximum Cure (p = 0.028). The radius of curvature correlated positively with SBS (rs = 0.204, p = 0.005). The SBS of all attachment materials was sufficient for clinical use, even after artificial aging. RelyX Unicem2 showed almost twice as high SBS values as the other attachment materials.

www.nature.com/scientificreports/ a cylindrical stub of polymerized AM (adhesive cylinder) was protruding orthogonally from the lingual/palatal tooth surface. The dual curing AMs, Nexus NX3 (NX, groups 3, 7, 11) and RelyX Unicem2 (RXU, groups 4, 8,12), were applied by syringes with mixing tips (Fig. 2B). The first fraction of material coming out of the mixing tip was always discarded. The second fraction of the AM was then poured into the silicone mold and cured by light for 20 s with an intensity of 1.492 mW/cm 2 using a MiniLED (ACTEON, Düsseldorf, Germany). Directly after bonding, all specimens were put in interim storage in distilled water at room temperature. Then, groups 1-4 were stored in distilled water at 37 °C for 24 h, groups 5-8 underwent 500 thermal cycles (5/55 °C, dwelling time: 2 min; eGo-Kälteysteme GmbH, Regensburg, Germany) and groups 9-12 were stored in distilled water at 37 °C for 90 days. The composition of the attachment materials is presented in Table 1.
Shear bond strength measurement. Prior to the shear bond strength measurements, the specimens were stored in distilled water at room temperature for 1 h. Shear bond strength was measured with the universal testing machine Instron 5965 (Instron Deutschland GmbH, Pfungstadt, Germany). The samples were fixed in a magnetically lockable device at the base of the testing machine. A metal blade containing a rectangular hole measuring 6 mm × 6 mm, which was used to shear off the adhesive cylinders from the teeth, was fabricated according to DIN 13990:2017-04 17 and fastened to the movable part of the testing device. The tooth surface carrying the polymerized AM was carefully positioned parallel to the blade with the AM adhesive cylinder reaching through the rectangular hole (Fig. 3). The force was applied to the interface between enamel and AM by moving the blade in an occlusal-gingival direction with a crosshead speed of 1 mm/min until failure. The maximum force was recorded in Newton.
All specimens were then photographed under a microscope (Bresser, Rhede, Germany) with 10 × magnification and screened for cracks. Due to the varying degree of enamel curvature, the bases of the cylinders that were bonded with the flexible silicone mold, did not have a uniform diameter of 3 mm. To accurately determine the area bonded to the enamel surface all adhesive cylinders were collected after the force measurement and digitized with an industrial computer tomograph (Zeiss Metrotom 800, Zeiss, Oberkochen, Germany). The resulting dataset was converted into the STL format using the software VGStudioMax 3.2 (volumegraphics, Heidelberg, Germany). To calculate the exact bonding area, every cylinder was analyzed using the software Meshmixer (Autodesk, Mill Valley, USA). First, the area of the adhesive cylinder surface that had been bonded to the enamel (Fig. 4A,C) was compared to the respective microscope image displaying the counterpart area of  www.nature.com/scientificreports/ bonding on the corresponding tooth (Fig. 4B,D). On this image the different shades on the surfaces of the etched teeth were distinguished to identify the outline of the surface that the adhesive cylinder was bonded to. This way, it was possible to determine, if a reconstruction of the adhesive cylinder surface as a result of chipping during the SBS test was necessary. In these cases, the marginal ridge of the area of the adhesive cylinder was thus virtually reconstructed ( Fig. 4C,D) and the area filled before determination of the surface area. In case of bubbles present on the surface of the adhesive cylinder, the bubble area on the bonding surface was also calculated (Fig. 4E) and subtracted from the entire bonding surface. Thus, the area of the bubbles was not included in the effective bonding area used for SBS calculation. The bonding strength was calculated for every specimen with the formula: here R represents the SBS in Megapascal (MPa = N/mm 2 ), F constitutes the maximum force in Newton (N) before failure, and A is the effective bonding area in square millimeters (mm 2 ). The adhesive remnant index (ARI) was evaluated as follows: 0 = no remaining AM on the bonded enamel surface, 1 = AM remaining on less than 50% of the bonded enamel surface, 2 = AM remaining on more than 50% of the bonded enamel surface, 3 = AM remaining on 100% of the bonded enamel surface 19 . A random sample from each AM group was sputter-coated with a 20 nm layer of gold using the Leica EM ACE200 system (Leica Mikrosysteme GmbH, Vienna, Austria) and viewed with a JEOL scanning electron microscope at 30 × and 250 × magnification (JSM-IT 300LV, JEOL Germany GmbH, Eching, Germany).

Reliability of measurements.
Twenty specimens were randomly selected, and the assessment of ARI was performed a second time by the same investigator and additionally by a second experienced investigator. To calculate intra-and interrater reliability the intraclass correlation coefficient (ICC; two-way mixed, absolute agreement) selected was used.
Statistical analysis. Data were analyzed using SPSS version 24.0 (IBM Armonk, NY, USA). For descriptive statistics of the SBS data, medians and interquartile ranges (IQR) as well as means and standard deviations (SD) were calculated for every treatment group. Normal distribution was tested with a Shapiro-Wilk-Test and visual assessment of histograms. Non-parametric Kruskal-Wallis-tests followed by post hoc Dunn-Bonferroni-tests and Mann-Whitney-U-tests using the Bonferroni correction were applied to check, if the differences between groups were statistically significant. To test the correlation between SBS and the radii of curvature Spearman's rank correlation coefficient was calculated. The Chi 2 test was applied to evaluate the ARI. p values of p ≤ 0.05 were interpreted as statistically significant.
Ethical approval. Approval for the collection and use of human teeth extracted for medical reasons was given by the ethics committee of the University of Regensburg, Germany (Approval No. 12-170-0150).  The ARI was related to the different AMs utilized (Chi 2 < 0.001). In both chemical curing groups (MC, IDB) an ARI score of 1 was detected most frequently (72.9% and 70.8%, respectively). In the NX group the most frequently encountered ARI score was 0 (56.3%), followed by 1 (41.7%). In the RXU group, an ARI of 1 was The areas occupied by gas bubbles were determined on the CT scan and then subtracted from the total area. This way, the effective bonding area was calculated. Scale bars: 1 mm. www.nature.com/scientificreports/ encountered most frequently (64.6%), followed by an ARI of 2 (12.5%) and 3 (4.2%). The detailed results are presented in Table 3.
Scanning electron microscopic images of the bonding surface on the lingual/palatal side of the third molar crowns after the SBS measurement are displayed in Fig. 5 for all experimental AM groups. The circular bonding area was easily discernible in all specimens. The samples representing the IDB and the RXU groups show remnants of the adhesive cylinder. No fractures, fissures or other forms of damage were found on the enamel. Instead, the typical wavelike pattern, possibly representing Hunter-Schreger bands, was observed on the enamel surface. These patterns were not confined to the bonding area but were present all over the lingual/palatal sides of the molar crowns.

Reliability of measurements.
The intra-as well as the interrater reliability were both excellent with an ICC of 0.969.

Discussion
The aim of this in vitro investigation was to compare the shear bond strength (SBS) of four different orthodontic attachment materials (AMs) that are commonly used for indirect bonding of lingual appliances. Moreover, the influence of different aging methods and the curvature of the enamel surface on SBS was examined.
The first hypothesis ("There is no difference in SBS between the different AMs. ") had to be rejected, as the results of this study clearly show significantly higher SBS values between human enamel and RXU in comparison to the other AMs. RXU is a dual cure self-adhesive resin cement normally used for prosthodontic purposes, such as the cementation of restorations and posts 20,21 . Two studies reported that, in comparison to Transbond XT (3 M, Monrovia, USA), RXU did not show sufficient SBS for bonding labial brackets 22,23 . In those studies, the enamel surface was not etched with 35% phosphoric acid before bonding the brackets with RXU. The functional acidic monomers that are contained in self-adhesive resin cements such as RXU are considered to be weaker in comparison to selective enamel etching using phosphoric acid, which results in a lower bonding strength to enamel. As a consequence, additional enamel etching is recommended, for example, for cementing veneers to enamel 20 . In the present study, enamel etching was performed for 30 s before bonding the adhesive cylinders, resulting in higher SBS values for RXU.
In accordance with our study, Scribante et al. found significantly higher maximum force values until debonding of RXU in comparison to MC and IDB for customized gold-alloy base brackets that were indirectly bonded 24 . Enamel surfaces of premolars were pretreated with sandblasting (27 μm aluminum oxide) and 37% phosphoric acid (30 s). Remarkably, they even described a significantly higher rate of enamel fractures after debonding for  MC and IDB are both chemical curing attachment materials, NX3 and RXU are dual curing. Moreover, RXU is also self-adhesive. Although the statistical analysis indicates significantly higher SBS values for dual curing attachment material, this result has to be viewed with caution. A closer look on the data reveals that this is mainly caused by the much higher SBS values found in the RXU groups. Especially after 90 days of storage the RXU-related SBS values are twice as high when compared to the other AMs. Therefore, the third hypothesis ("The way of curing (chemical or dual cure) does not influence the SBS. ") can be largely accepted. The functional acidic monomers contained in the self-adhesive RXU enhance surface demineralization of the enamel 20   www.nature.com/scientificreports/ used 6 . The highest value of SBS was found in the labial customized group with a resin as customized base 6 . Sung et al. reported a significantly higher SBS for limited resin custom bases in comparison to extended resin and gold-alloy custom bases 26 . SBS values in the present study were generally higher than in the above-mentioned studies. Considering the experimental setting for SBS testing, the force application point will be different as soon as the shapes of the brackets and their bases differ, especially their thickness. The curvature of most customized lingual brackets makes a reproducible force application even more difficult within the available experimental setups. Consequently, different ratios of direct force and moment of force application have an influence on the outcome. Therefore, it was our approach, in accordance with the DIN 13990:2017-04 17 , to have a closer look at the performance of the bonding quality only between lingual/palatal enamel and the different AMs. The bonding strength of the interface between different types of brackets and AMs needs to be considered separately, as there are too many factors that have an influence on SBS. Additional customized resin interfaces or different metal surfaces and their possible pretreatments are good examples for that. As a result, the SBS values presented here can hardly be compared to those of other studies. Nevertheless, SBS values of at least 5 MPa are generally considered to be sufficient for multi-bracket-treatment 8,9 . Restricted to the interface between enamel and AM, all the tested materials showed sufficiently high SBS values.
In the present study, thermal cycling did not have a significant influence on SBS. Nevertheless, the second hypothesis ("Different aging methods do not have an influence on SBS. ") had to be partly rejected, as SBS values of MC were significantly lower after 90 days of wet storage at 37 °C. In general, in-vitro investigation of the longterm performance of the tested materials is also of importance for the assessment of their stability and clinical applicability 27 . In contrast to materials utilized in restorative dentistry, the requirements for a reliable bonding in fixed orthodontic therapy is time-limited, as the treatment lasts on average about 2 years 28 . Therefore, a smaller number of thermal cycles was considered to be sufficient. In accordance with the DIN 13990:2017-04, all specimens in the thermal cycling group underwent 500 cycles 17 . Thermal cycling imitates daily intraoral temperature fluctuations that occur when patients eat/drink cold and hot food/beverages. This results in inner stress in the AM and over time leads to degradation. Inner stress can be a result of differences in thermal expansion and shrinking between filler particles and the resin 29 . Furthermore, microleakage or even debonding can be caused by differential thermal expansion of the AM and the tooth structure 29 . Apart from thermal cycling, the second most applied method for artificial aging of resins is wet storage 27 . Wet environment can lead to degradation of restorations 29 and therefore also have an influence on the AMs that are used to bond brackets. In the present investigation, MC showed significantly lower SBS values after 90 days of wet storage at 37 °C, which might be caused by weakening of the attachment material due to degradation of the fillers, softening, and hydrolysis 29 . Compared to IDB and NX3, the SBS values were lower but not significantly. RXU still showed very high SBS values after wet storage aging. When the functional acidic monomers have reacted, RXU becomes more hydrophobic and the tendency to absorb water, expand, and hydrolytically degrade is minimized 20 . In summary, thermal cycling and/or wet storage are essential parts of testing orthodontic attachment materials. However, no previous study investigating SBS of attachment materials for lingual brackets has taken this aspect into account.
The fourth hypothesis ("There is no correlation between SBS and the radii of curvature. ") was rejected as well. The results of the present study showed an influence of the radii of curvature on SBS in terms of smaller radii of curvature (rather curved surfaces) correlating with lower SBS values and larger radii of curvature (rather flat surfaces) with higher SBS values. According to the DIN 13990:2017-04, it is recommended to use rather flat (radius of curvature ≥ 12.5 mm) specimens to test SBS. Since lingual brackets are bonded to the lingual/palatal surfaces, that vary considerably in terms of their curvature, it seems prudent not to use flat surfaces in these special cases. As the results of the present study suggest, it seems to be important to determine and consider the radii of curvature when brackets are bonded to curved surfaces for SBS testing. To date, this aspect has not been considered by any other investigation dealing with orthodontic SBS testing. However, this aspect may be of clinical relevance as the clinician might consider the application of an attachment material with higher SBS values when bonding to very curved surfaces. This might be especially crucial for the rebonding of brackets after accidental debonding during active treatment.
After the adhesive cylinder had been sheared off, the ARI was determined. An ARI of 0 means that no AM was left on the enamel surface, which results in less clinical chair time to remove the adhesive, but at the cost of a higher risk of fracture 26 . With exception of NX3 an ARI of 1 was most common, which is in accordance with Scribante et al. 24 . However, as this is the first study on the SBS only between enamel and AM, the present ARI results can hardly be compared to those from other studies and the clinical relevance is therefore limited. Nevertheless, the results show that RXU was the only AM the bonding was strong enough to cause two of the resin cylinders to crack within themselves, leading to the whole enamel surface still being covered with RXU. This is in line with the much higher SBS values of RXU.
SBS testing is often criticized for several weak points. Due to the large number of different experimental setups and testing procedures, the results are very often difficult to compare. Therefore, it is essential to adhere to standardized protocols as far as possible if a specific research question is to be answered validly. In general, shear tests show several limitations compared to tensile tests, especially micro-tensile bond strength, such as a higher stress concentration e.g. due to deviations of the moduli of the materials involved [30][31][32] . This may well influence results and fracture patterns. However, shear tests are less complex, easier to perform, less costly, and take into account the surface configuration of the substances being tested. Due to this and the near-clinical loading situation, shear tests with a detailed description of the specimen setup and test configuration are good to use as comparative measurements. Furthermore, the long-term aspect regarding artificial aging should always be considered. This way, SBS testing provides a good possibility to compare the performance of different materials and to use these results as a basis for clinical trials. In this study, only the bonding between enamel and AM was investigated. Further research is required to consider all the other aspects, such as different bracket shapes www.nature.com/scientificreports/ and materials or customized resin pads. For this purpose, a standardized procedure for shearing customized brackets off will be crucial.

Conclusions
Within the limitations of this study, the following conclusions can be drawn: • All investigated AMs showed sufficiently high SBS values concerning the interface between enamel and attachment material. • SBS between RXU and enamel was significantly higher than between enamel and the other AMs tested.
• After 90 days of wet storage SBS values of MC were significantly lower than after 24 h.
• The curvature of the bonded surface correlates negatively with SBS.

Data availability
The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.