Bacterial Adhesion on Dental Polymers as a Function of Manufacturing Techniques

The microbiological behavior of dental polymer materials is crucial to secure the clinical success of dental restorations. Here, the manufacturing process and the machining can play a decisive role. This study investigated the bacterial adhesion on dental polymers as a function of manufacturing techniques (additive/subtractive) and different polishing protocols. Specimens were made from polyaryletherketone (PEEK, PEKK, and AKP), resin-based CAD/CAM materials (composite and PMMA), and printed methacrylate (MA)-based materials. Surface roughness (Rz; Ra) was determined using a laser scanning microscope, and SFE/contact angles were measured using the sessile drop method. After salivary pellicle formation, in vitro biofilm formation was initiated by exposing the specimens to suspensions of Streptococcus mutans (S. mutans) and Streptococcus sanguinis (S. sanguinis). Adherent bacteria were quantified using a fluorometric assay. One-way ANOVA analysis found significant influences (p < 0.001) for the individual parameters (treatment and material) and their combinations for both types of bacteria. Stronger polishing led to significantly (p < 0.001) less adhesion of S. sanguinis (Pearson correlation PC = −0.240) and S. mutans (PC = −0.206). A highly significant (p = 0.010, PC = 0.135) correlation between S. sanguinis adhesion and Rz was identified. Post hoc analysis revealed significant higher bacterial adhesion for vertically printed MA specimens compared to horizontally printed specimens. Furthermore, significant higher adhesion of S. sanguinis on pressed PEEK was revealed comparing to the other manufacturing methods (milling, injection molding, and 3D printing). The milled PAEK samples showed similar bacterial adhesion. In general, the resin-based materials, composites, and PAEKs showed different bacterial adhesion. Fabrication methods were shown to play a critical role; the pressed PEEK showed the highest initial accumulations. Horizontal DLP fabrication reduced bacterial adhesion. Roughness < 10 µm or polishing appear to be essential for reducing bacterial adhesion.


Introduction
Composites, PMMA, and thermoplastic polymer materials such as polyaryletherketones (PAEKs) represent the state of the art for the fabrication of temporary and fixed dental prostheses [1,2]. With regard to manufacturing processes in dental technology, additive (AM) and subtractive manufacturing (SM) techniques are increasingly replacing conventional manufacturing protocols. It has been shown that the manufacturing methods have an influence on the mechanical properties [3,4]. Thus, it can be assumed that the fabrication may also influence bacterial adhesion.
Oral biofilm formation follows a sequential colonization that begins with the adhesion of early colonizers such as streptococci to the pellicle-coated surface and is strongly influenced by material-dependent surface properties [5,6]. Plaque and biofilm formation can lead to caries formation and periodontal inflammation [7,8]. Secondary caries is still considered the main reason of dental restauration failure [9]. Relevant parameter for the
The blue and nonfluorescent Resazurin is reduced by viable and metabolically active cells/bacteria via dehydrogenase enzymes into the violet and fluorescent pigment Resorufin. Quantification was performed by measuring fluorescence or absorbance, which is linearly proportional to the number of living bacteria in the sample. Surface characterization: Roughness (R z , R a ) on all specimens was measured with a laser scanning microscope (KJ VK-X100K, Keyence, J, class 2 laser 658 nm, 50× magnification; measuring range 2500 × 1900 µm, cutoff λs 0.8 µm λc 0.08 µm, resolution h = 0.005 µm, w = 0.01 µm, 50 × 12 nm, repeatability σ h: 0.02 µm, w: 0.05 µm). Contact angles with two liquids differing in hydrophobicity (Millipore water, Diiodomethane) were determined using the sessile drop method and a computer-aided contact angle measurement device (DSA25, Krüss, Hamburg, Germany, drop volume of 1 µL). Surface free energy (SFE) was calculated. SEM micrographs (Phenom, FEI Company, Eindhoven, The Netherlands) were used to assess the surfaces prior and after polishing (untreated specimens, resolution 1024, quality high, WD 10 µm, magnifications up to 1500×, n = 2).
The blue and nonfluorescent Resazurin is reduced by viable and metabolically active cells/bacteria via dehydrogenase enzymes into the violet and fluorescent pigment Resorufin. Quantification was performed by measuring fluorescence or absorbance, which is linearly proportional to the number of living bacteria in the sample.
All specimens were cleaned with 70% v/v ethanol for 10 s and then rinsed with distilled water for 10 s. Specimens were transferred into sterile 48-well plates, fixed, and incubated with 1 mL of phosphate-buffered saline solution (PBS, Sigma-Aldrich, St. Louis, MO, USA). The intrinsic fluorescence of the specimens was determined.
Pellicle formation was initiated by incubating the sample-containing plates with 1 mL of artificial saliva per well at 37 • C in a thermos-shaking-device (OrbitalShaker; Thermo Forma, Marietta, OH, USA). After 2 h, saliva was carefully removed, and each specimen was incubated with 1 mL of microbial suspension (either S. mutans or S. sanguinis) and 15 µL of Resazurin (Resazurin, Sigma-Aldrich, St. Louis, MO, USA) for 2.5 h. Specimens were gently rinsed twice with 1 mL of PBS to remove unbound bacteria. Fluorescence was measured using an automated multi-detection reader (Fluostar Optima, BMG Labtech, Offenburg, Germany).
Statistical analysis was performed with SPSS 28.0 (IBM, Armonk, NY, USA). Normal distribution of data was verified using the Shapiro-Wilk test. Means and standard deviations were calculated and analyzed using one-way analysis of variance and the Bonferroni test for post hoc analysis. Between-subject effects were investigated. The level of significance was set to α = 0.05. Pearson correlation (PC) between the individual parameters was determined.

Bacterial Adhesion
The Shapiro-Wilk test confirmed a normal distribution of the data in~80% of cases. One-way ANOVA confirmed significant influences (p < 0.001) for individual parameters treatment and material, as well as their combinations. The highest impact on the adhesion of S. mutans was found for the combination of material and treatment (η 2 = 0.879, p < 0.001), followed by material (η 2 = 0.860, p < 0.001) and treatment (η 2 = 0.589, p < 0.001). For the adhesion of S. sanguinis, material (η 2 = 0.600, p < 0.001) had the highest impact, followed by the combination of material and treatment (η 2 = 0.595, p < 0.001), while treatment had a significantly lowest impact (η 2 = 0.317, p < 0.001).

S. mutans
Mean relative fluorescence intensities varied between 36 for TC and 2043 for MA-V. Post hoc analysis identified that PEKK and TC showed significantly (p < 0.037) lower fluorescence intensities correlating with less bacterial adhesion compared to printed specimens (MA-V, MA-H, and A) and pressed polyetheretherketone (P, P plus), as well as IM-E and the veneering composite (Table 2). Table 2. Relative fluorescence intensities (mean and standard deviation; -: not determined) with bacteria (S. mutans and S. sanguinis) on different materials with different surface treatments (P0, P1000, and P4000).  The lowest fluorescence intensities indicating lowest microbial adhesion were identified for PEKK for all unpolished materials, while TC showed the lowest fluorescence intensities correlating with lowest bacterial adhesion for polishing protocols. However, no significant (p = 1.000) difference between milled specimens was observed. MA printed specimens (MA-V, MA-H) featured significantly (p < 0.001) higher fluorescence intensities indicating higher bacterial adhesion in comparison to the other materials.

S. sanguinis
Mean relative fluorescence intensities ranged between 83 for AKP and 1216 for MA-V. Post hoc analysis revealed significantly (p < 0.001) lower fluorescence intensities indicating lower biofilm adhesion for AKP compared to BC, as well as both pressed PEEK (P, P plus) and MA printed specimens (MA-V, MA-H). Pressed PEEK specimens showed significantly (p < 0.001) higher fluorescence intensities correlating with higher bacterial adhesion compared to both control groups (glass and Sinfony) and milled specimens (TC, AKP, PEKK, and PEEK) as well as injection-molded and printed PEEK specimens. AKP featured lowest relative fluorescence intensities indicating lowest biofilm formation for all polishing protocols. MA-V showed significantly (p < 0.001) higher fluorescence intensities indicating higher adhesion of bacteria compared to all other materials, while no significant (p = 1.000) differences were observed between MA-H and pressed PEEK (P and P plus) specimens Table 2.

Influence of Fabrication Methods
Post hoc analysis revealed that milled PEEK specimens showed no significant (p = 1.000) difference in fluorescence intensities indicating the adhesion of S. sanguinis when compared to injection molded specimens (IM-E, IM) and printed PEEK (A). Pressed specimens (P, P plus) showed significant (p < 0.001) more S. sanguinis adhesion than the other manufactured PEEK specimens. Post hoc analysis of S. mutans adhesion revealed no significant (p = 0.145) differences for different fabrication methods of PEEK. Printed MA specimens which were printed in horizontal direction (MA-H) promoted significantly (p < 0.001) less biofilm compared to specimens with vertical printing orientation (MA-V).

Effects of Polishing
Post hoc analysis with homogeneous subsets revealed significant (p < 0.001) differences between the polishing procedures P0 and P1000, as well as between P0 and P4000, regarding relative fluorescence intensities, indicating differences in microbial adhesion.
ANOVA analysis showed significant (p < 0.002) differences in relative fluorescence intensities indicating differences in bacterial adhesion as well as significant (p < 0.001) differences in surface roughness (R z and R a ).

SEM Figures
In the initial condition P0, the specimens exhibited a wavy and slightly rough, but mostly homogeneous surface. IM-E showed isolated cracks, and P Plus exhibited a heavily damaged surface. After the first machining P1000, the specimens mostly showed grinding marks and occasional scratches and cracks. IM-E had isolated holes. P Plus showed a homogeneous surface with punctual brightening. After the second processing step P4000, mostly homogeneous surfaces with isolated scratches were found. P Plus and MA-H also showed additional grinding marks (Figure 2).

Correlations
A highly significant (p = 0.010, PC = 0.135) correlation between relative fluorescence intensities and R z was identified in the S. sanguinis model, while no significant (p = 0.138) correlations between R z and relative fluorescence intensities was identified in the S. mutans model. Regardless of the individually applied microbial model, no significant correlations between R a (p = 0.273) and SFE (p = 0.0667) and fluorescence intensities were identified.

Type of Material
The first null hypothesis suggesting that the type of material has no influence on bacterial adhesion has to be rejected. The investigated materials showed different adhesion of streptococci. For about 50% of the materials, microbial adhesion was considerably lower or about twice as high as for the reference materials. Exceptions were unpolished AM MA samples which exceeded microbial adhesion to the reference materials manyfold.
PAEKs: For the various PAEK materials, significant differences in microbial adhesion were identified. Similar results were previously reported for other thermoplastic materials [36,37]. Despite the same manufacturing method, the SM-processed PAEK materials showed different bacterial adhesion. These differences were probably due to the composition of the various PAEK materials. PAEK systems vary in their degree of crystallinity and their rate of crystallization, both depending on the individual technique blanks and semi-finished products. PEEK is generally amorphous or semi-crystalline, but it can also reach a very high degree of crystallinity (up to 40%). Processing temperatures during pressing or SM may impair the surface quality. Surface defects on SM PEEK could only be removed after polishing with 4000 grit sandpaper. In contrast to composites, the type and amount of filler in thermoplastic materials had only a small influence on bacterial adhesion, which might be attributed to the small quantities involved. Nevertheless, microbial adhesion on PEKK might have been influenced by nano spikes, as reported in previous studies [38,39].
Composite: The resin-based CAD materials and the laboratory reference composite showed a low tendency to bacterial adhesion. These observations can be explained by the composition of the filled resin-based materials. Minimizing resin matrix exposure might reduce biofilm formation on the surface of resin composites [17]. A negative correlation between inorganic filler content and biofilm formation was previously identified [10]. The highly filled composite showed higher bacterial adhesion than the unfilled PMMA, but in the range of the less highly filled printed materials. Moktar et al. revealed that a nanohybrid filler system and a modified resin matrix structure cause considerably less biofilm formation [15]. Further research should, therefore, address the influence of filler content on bacterial adhesion even in thermoplastic formulations.
AM materials: The printed samples showed higher S. mutans accumulation compared to the controls. For printability, smaller and fewer fillers, as well as a modified matrix, are required, which presumably impacts the adhesion of microorganisms. Unpolished AM MA specimens showed significantly higher adhesion of microorganisms compared to polished specimens, which is most likely caused by uncured resin remnants on the surface of the specimens. Post-polymerization or washing procedures have been shown to influence the in vitro cytotoxicity and, therefore, might also affect bacterial adhesion [40,41]. This study also observed a tendency for increased bacterial adhesion on printed MA samples that were subjected to the polishing protocol P4000.

Fabrication of the Material
The null hypothesis that fabrication methods do not influence bacterial adhesion has to be partly rejected. The data of the current study indicated that the effect of fabrication methods is strain-dependent, as S. mutans adhesion was not affected by fabrication methods. For S. sanguinis, significantly higher adhesion was identified on pressed PEEK specimens than on specimens processed by the other fabrication methods (milling, injection molding, and printing). As a result of polishing, processing effects were significantly altered. Relevant differences in microbial adhesion between milled, injection-molded, and pressed PEEK were identified. Artefacts such as air inclusions could be observed for pressed and injection-molded specimens. These superficial defects might explain the higher bacterial adhesion (especially for pressed specimens). Industrial fabrication of blanks increases mechanical properties [3] and, probably, the surface quality. It is suggested that there is a reduced risk of porosities when the material is fabricated under optimal conditions [20]. A comparable effect was also observed for industrially fabricated resin composite blocks for dental CAD/CAM applications [42]. All materials processed with CAM featured similar levels of bacterial adhesion compared to the other manufacturing processes. Nevertheless, previous studies did not report a noticeable difference in the adhesion of Staphylococcus aureus and Candida albicans to pressed or milled PEEK samples [26].
Temperature management and cooling processes are crucial for the structure and quality of semi-crystalline materials, as they influence the composition of the crystalline and amorphous phases. The processes in which the material is formed during heating (fused filament fabrication and pressing) are likely to be more affected. When shaping was carried out in the CAM process, a low level of bacterial adhesion was observed. This phenomenon could have been caused by modification of the surface due to crystallization effects. Thus, the temperature control during processing and, for example, the preheating of the muffle could also have an effect on the structure of the thermoplastic material. Since the temperature control in the blank plays an important role, the shape and size of the blank are also relevant. In addition, the type of mold used (plaster or metal) could lead to interactions and/or changes in the surface, thus affecting the adhesion of S. sanguinis. Higher adhesion of S. sanguinis was identified in specimens produced by the compression molding process than in those processed using injection molding. During extrusion, the solidification of the filament results in a partial alignment of the polymer chains, whereby, starting from the crystallization nuclei, molecule chains become structured and form ordered regions. The deposition in filament production is in the middle range between CAD and press production. The influence of the adhesion mechanisms of the different bacterial strains was evident in the adhesion of S. mutans.

AM Systems
Comparing different building angles, horizontally printed MA specimens promoted less biofilm formation compared with vertically printed specimens. The reason for this could be that, in vertical production, a number of layers and transitions are on the measuring surface of the specimens, whereas, in horizontal production, only one layer is exposed on the surface. Mokhtar et al. observed that bacterial adhesion on MA materials was influenced by fabrication methods (AM, SM, and heat/cold curing), with AM showing the best results [15]. Results were heterogenous, since more C. albicans was observed on AM specimens compared to SM ones [43]. Although a significant effect of the building angle on surface roughness/topography was previously reported, no relevant difference in the adhesion of Candida albicans was observed [24].

Surface Free Energy and Roughness
The influence of the superficial roughness on the adhesion of microorganisms has been intensively discussed in the literature [25,26,44,45], and it always represents the requirement for a smooth surface [46]. It is expected that microbial adhesion can modify fillers and reduce the microhardness of the surface. The release of material components, especially ethylene glycol dimethacrylate or triethylene glycol dimethacrylate, can also promote the growth of further bacteria [47,48]. R a : The null hypothesis that bacterial adhesion is not influenced by surface roughness R a could be confirmed. R a showed no correlation with bacterial adhesion, although the materials provided significantly different initial values in R a in the unpolished state and after both polishing steps. One reason for this is that, especially in the low roughness range of R a , the application of the pellicle caused a leveling and equalization of the surfaces. For most materials, lowest R a roughness was achieved with the P4000 polishing protocol. Exceptions were AKP and IM-E, which hardly showed any R a changes after the various polishing regimes, and A, which was even rougher after polishing. These observations can be explained by the distinct type of fabrication, as the materials showed smooth surfaces directly after production. For material A, this was certainly due to the filament printing process, where the processing parameters, such as temperature and the type of die through which the molten material is pressed, determine the surface quality. For the AKP specimens, a different polymer matrix compared to other PAEKs was expected. For IM-E, the relatively high filler content for PEEK influenced the surface finish.
The previously introduced R a roughness threshold of 0.2 µm was highly exceeded by all materials of the current study, due to the acquisition of surface roughness data. In comparison to mechanical scanning, the surface quality, such as reflecting, light-absorbing, or absorbing surfaces, can result in significant differences in the roughness of various materials. For surface roughness measurements, discrepancies of about 75% of the value obtained between the stylus method and other systems have been reported [49][50][51]. The samples all fell within the range of the clinically relevant reference veneering material. These considerations confirm that the threshold value is dependent on the measurement approach applied. R z : The hypothesis that bacterial adhesion is not affected by surface roughness R z could only be partially confirmed. A positive correlation between R z and microbial adhesion was shown for S. sanguinis. This observation might partly confirm earlier studies, which reported a positive correlation between surface roughness and bacterial adhesion [10,12,17,25,26]. In most cases, polishing resulted in a decrease in R z values, with the lowest values identified after a 4000-grit polishing regime. Overall, this polishing regime caused significantly less adhesion of S. sanguinis and S. mutans. SFE: The null hypothesis that SFE did not influence bacterial adhesion could be confirmed. SFE was not correlated with either polishing or bacterial adhesion. In contrast to previous studies, it could not be shown that the surface energy of the materials affects the adhesion of the bacteria [52]. SFE was not affected by processing, as shown for MA or PEAK materials, but might have been influenced by filler or matrix components. Protein formation and bacterial adhesion decrease in materials with a high surface energy [53,54]. It is well known that the formation of pellicles reduces the surface energy on the tooth surface [52,53]. Due to the formation of pellicles, the free surface energies of the different surfaces approach each other [53]. The correlation between SFE and bacterial adhesion appears to be strain-specific; hydrophilic bacteria such as S. mutans with a high surface energy tend to be hydrophilic surfaces [55][56][57].

Conclusions
Resin-based materials, composites, and PAEKs showed noticeable differences in bacterial adhesion. Fabrication methods were shown to play a critical role for bacterial adhesion; the pressed PEEK exhibited the highest initial accumulation, whereas the milled PAEK samples provided similar lower adhesion. Horizontal DLP fabrication reduced bacterial adhesion in comparison to vertical processing. For DLP printed MA and pressed PEEK, the surface finishing significantly reduced bacterial adhesion, making it essential for clinical application. The processing method and the polish affected the quality of the surfaces; Apium and Ultraire already have very smooth surfaces due to processing and, thus, low bacterial adhesion. Roughness values < 10 µm or polishing appear to be essential for reducing bacterial adhesion, whereas SFE did not show any influence.

Data Availability Statement:
The data presented in this study are available upon request from the corresponding authors.