Breast cancer is the most diagnosed cancer and the one with the highest death rate (Arnold et al., 2022). According to the Global Cancer Observatory, in 2020, over 2 million new cases around the world of breast cancer were diagnosed, resulting in over 30% deaths within the first year (Global Cancer Observatory, 2020). Most women diagnosed with this type of cancer undergo a surgical treatment called mastectomy, which consists of the partial or total removal of one or both breasts. Consequently, the result of this surgery can compromise them physically, emotionally, and socially. For this reason, reconstruction options such as augmentation mammaplasty are used, which helps to reduce the sequelae of mastectomy, allowing to improve satisfaction, self-esteem, body image, and in general, the quality of life of patients (Atisha et al., 2008).

Augmentation mammaplasty is a cosmetic surgery that consists of increasing or restoring the size of the breasts using silicone implants, which can be filled with a silicone gel or a sterile, aqueous saline solution. It is one of the most common surgical procedures performed today, since it is performed both for reconstructive surgery in post-breast cancer treatment and for pure cosmetic reasons (Abdul-Al et al., 2020). According to the International Society of Aesthetic Plastic Surgery (ISAPS), breast augmentation continues to be one of the most popular surgery worldwide, with over 1.5 million interventions that represented 13.1% of all surgical procedures in 2021 (The Int ernational Society of Aesthetic Plastic Surgery, 2021). Although reconstruction offers considerable benefits for the recovery of patients with breast cancer, various complications have been described for these procedures, such as bruising, rupture or emptying of implants, capsular contracture, and bacterial infections (Pittet et al., 2005; Maxwell and Gabriel, 2012). Since the introduction of silicone implants, new designs have been made to increase the biocompatibility of the implants and reduce postoperative problems (Holzapfel et al., 2013). Five different generations of implants have been developed so far (Bergmann et al., 2014). However, these still experience complications (Teck Lim et al., 2013); therefore, to minimize postoperative complications, it is necessary to modify the implants available on the market in order to improve their properties and biocompatibility.

There are different types of breast implants in terms of filling, size, shape, and surface topography. In the case of surface topography, implants with smooth and textured surfaces are commercialized. The smooth surface allows implants to move within the chest, simulating the breast’s natural movement. However, this displacement capacity prevents the implants from adopting a stable position, causing a lateral deviation of these and their settlement in the lower part of the pocket; can stretch lower pole over time (Calobrace et al., 2018). Furthermore, the smooth surface has been associated with high rates of fibrosis. However, these implants are still being used because they provide a perfect circular shape of the breasts, making them visually more natural (Shin et al., 2018). On the other hand, textured implants have higher control of implant movement through an adhesive effect. The roughness of the surface induces the formation of a capsule of collagen tissue around the implant that facilitates its fixation. These implants are intended for reducing the incidence of capsular contracture, and although there are studies that support the effectiveness of these implants, there is still controversy regarding their benefits (Adams, 2009; Chang and Hammond, 2018).

In the context of a patient’s quality of life, a viable alternative to modify the surface of an implant is the use of cyclodextrins (CDs). These molecules are a family of cyclic oligosaccharides with a hydrophilic outer surface (C-OH groups) and a hydrophobic apolar cavity (C-O-C and C-H bonds) (Mura, 2015). They do not present toxicity, have good solubility, are easily modifiable, and have a high biological availability (Zhang et al., 2019). In addition, they are available to generate controlled release systems due to their characteristic cavity capable of forming reversible complexes with various drugs (Sherje et al., 2017). Additionally, CDs can be crosslinked with bi- or multifunctional agents to form polymers (Concheiro and Alvarez-Lorenzo, 2013), which consist of a three-dimensional network suitable for drug delivery applications (Hernández-Montelongo et al., 2014). These systems substantially increase the stability of drugs and decrease side effects in the body, resulting in a specific, safe, and effective medication to reduce the development of possible postoperative complications. For these reasons, CD-based polymers have been used to functionalize different biomedical materials for drug release applications, such as mesh implant for parietal reinforcement, titanium hip prostheses (Vermet et al., 2017), porous bioceramics (Taha et al., 2014), paper points for periodontal pockets treatment (Chai et al., 2014), and others.

Currently, there are no reports in the literature regarding the application of cyclodextrin biopolymers as coatings on silicone breast implants for drug delivery. However, there are some reports about other type of polymers that have been used, for example, Kim et al. (2019) (Kim et al., 2019) that used patterned coating dots of poly (lactic-co-glycolic acid) (PLGA) to release tranilast for the fibrosis suppression. In the same sense, Kang et al. (2019) (Kang et al., 2020) heat-induced polymerization of 2-methacryloyloxyethyl phosphorylcholine on silicone breast implants for the inhibition of capsular formation and inflammation. Moreover, Gosau et al. (2013) (Gosau et al., 2013) coated silicone material used for breast augmentation with different copper concentrations to study the anti-adherence and bactericidal effects on S. epidermidis.

Based on the above, this work focuses on a novel functionalization tool for commercial silicone breast implants, with smooth and textured surfaces, by β-cyclodextrin/citric acid and 2-hydroxypropyl-β-cyclodextrin/citric acid in situ polymerization as a preventive treatment for mammaplasty postoperative risks. Firstly, a pretreatment of oxygen plasma oxidation with subsequent chitosan grafting was evaluated. Then, polymerization stability on samples was studied and physicochemically analyzed. Three molecules that could be useful in augmentation mammaplasty surgery, were used as model to test the obtained samples as controlled local drug delivery system for mammaplasty: pirfenidone (PFD) proposed for capsular contracture prevention (Veras-Castillo et al., 2013; Guimier et al., 2022), Rose Bengal (RB) that possesses cytotoxic properties in cancer cells (Loya-Castro et al., 2018; Demar et al., 2021), and KR-12 peptide, which is antimicrobial and has antibiofilm properties to prevent bacterial infections (M et al., 2019; Ajish et al., 2022; Liu et al., 2019). Previous works have reported the successful complexation CDs and these molecules: PFD with β-cyclodextrin (Ji et al., 2016; Panigrahi et al., 2023), RB with hydroxypropyl-cyclodextrins (Fini et al., 2007) and β-cyclodextrin (Khushbu and Jindal, 2022), and KR-12 with β-cyclodextrin (Consuegra et al., 2013). The kinetic release mechanism of the used model molecules was studied by the Korsmeyer-Peppas, and monolithic solution mathematical models. Furthermore, the antibacterial effect of KR-12 was tested against Staphylococcus epidermidis and Pseudomonas aeruginosa bacteria, and the cytotoxicity of functionalized samples was studied using epithelial cells. It is of significance to emphasize that the KR-12 peptide constitutes the most concise antimicrobial motif within the human cathelicidin LL37 (Wang, 2008; Ramos et al., 2011). Remarkably, it not only demonstrates antibacterial activity but also possesses immunomodulatory properties on epithelial cells (Ramos et al., 2011; Lei et al., 2023). The conducted assays have corroborated the heightened attributes of functionalized samples when compared to unprotected implants.

In this study, samples were obtained from the surface of smooth (S) and textured (T) commercial breast implants (Figure 1A). As polydimethylsiloxane (PDMS) is the basis for both breast implant silicone shell (Daniels, 2012), the surface of these types of implants is hydrophobic. Therefore, before functionalizing the samples, a surface pretreatment of oxygen plasma oxidation with a subsequent chitosan grafting (0.1% or 1% w/v) was carried out (Figure 1B). After that, in-situ polymerization (BCD or HPBCD) in the presence of citric acid was performed (Figure 1C). The oxidation step is intended to integrate silanol groups on the implant surface. Consequently, the attachment of CHI would result in the formation of a coating through hydrogen bond interactions between hydroxyl groups and the partially amidated CHI, as the solution was prepared in an acidic medium (Chessa et al., 2016). Likewise, CHI generate an electrostatic bond between the oxidized surface and the copolymers (BCD or HPBCD with citric acid), which are also partially negatively charged due to their unreacted carboxylic groups of the copolymerized citric acid (Hernández-Montelongo et al., 2018): oxidized surface (−)/chitosan grafting (+)/cyclodextrin polymer (−).

The MB absorbing capacity was used to evaluate the pretreatment step of samples. This dye is positively charged and can be used to detect cyclodextrin/citric acid polymers, which are negatively charged (Hernández-Montelongo et al., 2018). Figure 2 shows representative MB absorbance spectra of samples with or without pretreatment and functionalization: plasma oxidation (OX) + CHI grafting (0.1% or 1%), followed by the cyclodextrin polymerization (BCD or HPBCD) on the implants surface (S or T). The control samples did not present an MB signal, confirming the absence of polymer. In the case of samples directly polymerized with no pretreatment presented some MB absorbance peaks: 0.35 for both S-BCD and S-HPBCD, 0.54 for T-BCD, and 1.5 for T-HPBCD. On the other hand, all samples with the pretreatment but without the polymerization process presented practically undetectable absorbance peaks. Nevertheless, a considerable MB absorbance increase was observed across in all the samples subjected to the pretreatment step and polymerization process. It is important to highlight that in all cases, samples with HPBCD presented higher MB absorbance values than BCD. Although it is known that BCD forms a 1:1 complex with MB, while HPBCD has a 2:1 inclusion complex (Kacem et al., 2013; Martin et al., 2013), which means twice HPBCD molecules are needed as BCD to encapsulate the same amount of MB. The higher MB absorbed in HPBCD samples is explained by the higher degree of polymerization that the HPBCD molecule generates as compared to native BCD. The hydroxypropyl groups of HPBCD improve the reactivity of the–OH groups present in the cyclodextrin, facilitating the esterification reactions with the cross-linking agent (Blanchemain et al., 2007; El Ghoul et al., 2010).

Figure 2 also shows the absorbance spectra of the samples treated with different CHI solution. Smooth samples treated with CHI 1% showed higher MB absorbance than smooth samples with CHI 0.1%: 1.1 and 1.5 maximal absorbance values for S-OX-CHI(0.1%)-BCD and S-OX-CHI(0.1%)-HPBCD, respectively. These values increased up to 3 and 3.3 for S-OX-CHI(1%)-BCD and S-OX-CHI(1%)-HPBCD, respectively, as shown in Figures 2A, B. In the case of textured samples, the increase of CHI concentration did not reflect the same proportional increase in MB absorbance as in smooth samples: 3.3 and 4.1 absorbance values for T-OX-CHI(0.1%)-BCD and T-OX-CHI(0.1%)-HPBCD, respectively, increased up to 3.7 and 5.1 for T-OX-CHI(1%)-BCD and T-OX-CHI(1%)-HPBCD, respectively, as can be seen in Figures 2C, D. These results suggest that the roughness of textured samples plays a key role in the polymerization. The protrusion in the surface could work as anchoring of the cyclodextrin polymers (Hernández-Montelongo et al., 2014). A higher amount of polymer was forming the films in the textured surfaces, even with CHI 0.1%. Due to the higher amounts of obtained using the concentration of CHI 1%, this was chosen as a condition for the pretreatment step.

To detect changes on the surface after the chemical treatments, the wettability and roughness of the samples were measured (Figure 3). Figure 3A shows the water contact angle of the smooth and textured samples. Both control samples, smooth and textured, are considered hydrophobic (angles of 93° ± 5° and 121° ± 10°, respectively), due to the–Si-CH3 groups present on the surface of the PDMS surface (Mor et al., 1990). After the plasma oxidation, the contact angle decreased to 74° ± 5° and 72° ± 8° on the smooth and textured samples, respectively. These results indicate a hydrophilic behavior due to the formation of silanol groups (–Si-OH) on the surface (Peterson et al., 2005). Thus, this allowed the CHI to bind to the activated PDMS surface by hydrogen bridges. After the CHI grafting, samples remained hydrophilic, and the contact angle was practically the same for both types of surfaces (77° ± 5° for the smooth and 77° ± 10° for the textured).

In the case of polymerized samples, the contact angle increased to 88° ± 4° and 87° ± 5° on the smooth surface, BCD and HPBCD, respectively; while in the textured samples, the angles were 73° ± 12° for BCD and 84° ± 7° for HPBCD. The hydrophilicity of polymerized samples comes from the cyclodextrin functional groups interaction (COOH, OH) with the water and the residual carboxylic acid groups present in the cross-linking agent (citric acid) (Blanchemain et al., 2007; El Ghoul et al., 2010).

Figures 3B, C show the smooth and textured average roughness values, respectively. The smooth samples showed values in a range of hundreds of nanometers, and the textured in the scale of tens of microns. After each chemical treatment step the roughness increased; obtaining similar tendency in both surface types. The smooth surface sample increased from 63 ± 38 nm (control) to 348 ± 45 nm and 298 ± 40 nm for the samples polymerized with BCD and HPBCD, respectively. In the case of the textured samples, the roughness increased from 12.5 ± 5 µm (control) to 25 ± 9 μm and 21 ± 10 µm for the samples polymerized with BCD and HPBCD, respectively.

In order to make the reading easier, the functionalized samples were relabeled as S-BCD*, S-HPBCD*, T-BCD* and T-HPBCD* to substitute S-OX-CHI(1%)-BCD, S-OX-CHI(1%)-HPBCD, T-OX-CHI(1%)-BCD, and T-OX-CHI(1%)-HPBCD, respectively. The chemical analysis was performed by ATR-FTIR to directly identify the surface modification (Figure 4). The bending vibrations of the fingerprint functional groups of PDMS were detected. The Si-(CH₃)₂ (786 cm^–1), symmetric bending of Si-CH₃ (1263 cm^–1), stretching of CH (2,960 cm^–1) from CH₃, Si-O-Si stretching at 1065 cm^–1 and 1007 cm^–1 were observed in the spectra of both controls and functionalized samples (Bodas and Khan-Malek, 2006). Figures 4B, D show the spectra of the cyclodextrin functionalized samples. Characteristic functional groups of cyclodextrin polymers were detected: O-H stretching of the hydroxyl and carboxyl groups (3,435 cm^–1) and the C=O stretching (1736 cm^–1) of ester groups resulted from the esterification and the residual carboxylic acids of the cross-linking agent (Zhao et al., 2009). It is essential to mention that these two signals were not detected in both substrate controls. In the case of the smooth samples (Figure 4B), the intensity of O-H and C=O functional groups was higher for S-HPBCD* than S-BCD*, which indicates that there were more esterification reactions using HPBCD molecule than BCD. In the textured samples (Figure 4D), T-BCD* and T-HPBCD* showed both O-H and C=O groups with similar intensity, confirming the key role of roughness for the in-situ polymerization.

Figure 5 shows the top view and cross-section SEM images of the control and functionalized samples. Regarding the smooth implants, the top view of S-Control exhibits a completely flat surface (Figure 5A), while the functionalized samples exhibited textures and folds attributed to the biopolymers, Figure 5B for S-BCD* and Figure 5C for S-HPBCD*. The layer thickness, obtained by the cross-section, was 14 ± 3 μm and 20 ± 3 µm for S-BCD* and S-HPBCD*, respectively. The images of textured samples showed a rough surface with structures of approximately 200 μm; valleys with a deep down to 100 µm (Figures 5G, J for the T-Control). In the case of functionalized samples, the valleys were filled, exposing partially the surface peaks. T-HPBCD* (Figure 5 I and 5 L) showed a better coverage of the substrate surface than T-BCD* (Figure 5 H and 5 K), suggesting a higher polymerization degree using HPBCD than BCD.

Table 1 presents the degree of functionalization and degradation parameters (Equations 1 and 2). Results indicate that the degree of functionalization on textured samples was higher than smooth ones, confirming the key role of the initial control surface roughness. Additionally, the functionalization degree was higher using HPBCD than BCD on both substrates, in agreement with previous characterization analyses.

The hydrolytic degradation results are shown in Figure 6. The polymeric films on textured samples were more stable than smooth ones, around 11 and 7 days, respectively. Degradation rates were higher in samples functionalized with HPBCD than BCD. This is because HPBCD samples presented a higher amount of polymer forming the layers, then, the polymer was degraded faster. Remarkably, for all cases, the polymer was not totally degraded after a month in the hydrolytic assays fluid for all the cases.

To evaluate the drug delivery functionality of samples, they were tested with PFD, RB, and KR-12 in PBS batches at 37°C under stirring. Controls and functionalized samples were loaded with highly concentrated solutions of PFD, RB, and KR-12. The obtained drug release profiles are shown in Figure 7. The maximum released amount of each drug, reached in the equilibrium time, is reported in Table 2. Results show that both pristine samples (S-Control and T-Control) were able to release considerable amounts of PFD, demonstrating the chemical compatibility between the chemical structure of PFD and silicone substrates (Wu et al., 2021). In the case of functionalized samples, those polymerized with HPBCD presented higher amounts of released PFD than samples polymerized with BCD. Independently of the type of substrate, BCD and HPBCD samples released around 38% and 62% higher than controls, respectively. In the RB release profiles, both controls released negligible amounts of RB, and in the case of polymerized samples, the released mass depended on the type of cyclodextrin and substrate. These results mean that functionalized samples released in the order of thousands of times higher mass of RB than controls. Moreover, it is important to highlight that, for both molecules, PFD and RB, functionalized samples reached higher equilibrium times than controls. In the case of the antimicrobial peptide release profiles, the controls released same amounts of KR-12. Functionalized smooth samples released similar values between them, and functionalized textured samples reached also similar values. In this experiment series, the equilibrium time was not reached even at 400 h. This could be due to the high molecular weight of the KR-12 peptide (1,570.95 g/mol) in comparison with RB (973.67 g/mol) and PFD (185.22 g/mol) since high molecular weight molecules present, except for specific surrounding environments, lower diffusion coefficients (Bogdan et al., 2011).

To attain a deeper perception of the mechanisms that govern the release of PFD, RB, and KR-12 from the functionalized samples, release profiles were fitted to the Korsmeyer-Peppas model (Wu et al., 2021). The diffusion coefficients were obtained using the monolithic solution model (Siepmann et al., 2012). Both models were fitted to the normalized experimental data (Supplementary Figure S3). The obtained kinetic parameters from them are presented in Table 3. According to the r ² _adj obtained in all samples, both models fit well to the experimental data. In the case of the Korsmeyer-Peppas model, the kinetic constant k _KP , which represents the release rate of a molecule (Bruschi and Bruschi, 2015), presented similar values among all samples, suggesting a similar kinetic rate between them. On the other hand, the release exponent n = 0.5, or n ≤ 0.5 corresponds to a Fickian diffusion release during which the solvent penetration is the rate-limiting step; 0.5 < n < 1.0 is related to non-Fickian release, this means that drug release followed both diffusion and erosion mechanisms; and n = 1 corresponds to zero order release, where drug release is independent of time applications (Hernández-Montelongo et al., 2014). According to this, the release of PFD from the four types of samples (S-BCD*, S-HPBCD*, T-BCD*, and T-HPBCD*) was controlled by both diffusion and erosion mechanisms. In the case of the other molecules, RB and KR-12 were mainly controlled by diffusion. These results can be explained by the molecular weight and size of molecules. In a determined volume of the cyclodextrin polymers, it is expected to find more smaller molecules (PFD) than bigger ones (RB and KR-12). Then, the erosion of the polymer releases more small molecules than big ones. That is why erosion was significant in the PFD experiments and not in the RB and KR-12 assays.

Regarding the monolithic solution model, the obtained diffusion coefficients (D) from Table 1 are plotted in Figure 8 for an easier data analysis and correlation. The results indicate that the diffusivity of molecules from the smooth samples, independently of the cyclodextrin polymer, showed similar values (from 1.3 to 3.5 μm²/h). However, in the case of textured samples, the diffusivity depended on the molecular weight (Bogdan et al., 2011). Therefore, lower molecular weights presented higher diffusivity and vice versa. The surface area of samples can explain this phenomenon. The high roughness of textured samples, 100 times higher than smooth samples, also represents a higher surface area for textured samples than smooth ones; then, the mass diffusion should be favored. That is why the PFD molecule showed the highest diffusion coefficient in the textured samples (132.6 and 121.9 μm²/h for T-BCD* and T-HPBCD*, respectively); RB presented the medium values (82.4 and 46.4 μm²/h for T-BCD* and T-HPBCD*, respectively), and KR-12 exhibited the lowest values (8.9 and 3.9 μm²/h for T-BCD* and T-HPBCD*, respectively). We suggest that in the case of the smooth samples, the diffusion coefficients were independent of the molecular weight because the polymer on the top surface substrate had less steric conformation restrictions for movement and diffusion. In all cases of smooth samples, D values were in the range of 1.3–3.5 μm²/h.

After physicochemical characterization and drug release studies, the antibacterial effect of the KR-12 was evaluated. It has been reported that KR-12 possess antimicrobial and antibiofilm properties against S. epidermidis (Gram-positive) and P. aeruginosa (Gram-negative) bacteria (M et al., 2019; Ajish et al., 2022). In that sense, first, the minimum inhibitory concentration (MIC) of peptide in solution was obtained for both bacteria, which are two of the most virulent strains in breast implant infections (Chessa et al., 2016; Mesa et al., 2021). The results were 100 μg/mL for S. epidermidis and 200 μg/mL for P. aeruginosa, respectively. The enhanced resistance of the P. aeruginosa against KR-12 can be explained by the outer cell membrane, typical of Gram-negative bacteria. Afterwards, the antibacterial effect of the implants loaded with KR-12 was tested (Figure 9). In the case of S. epidermidis strain, all funcionalized samples, with the two types of implants and the two polymers, showed almost 100% of antibacterial rate in comparision with the unfunctionalized samples. In the case of the P. aeruginosa strain, the antibacterial effect was mainly observed in the functionalized soft implants (S-BCD* and S-HPBCD*), with also 100% of antibacterial rate in comparision with the control samples. However, textured samples presented a low antibacterial rate for P. aeruginosa bacteria: 75% for T-BCD* and 20% for T-HPBCD*. This can be explained due to their low diffusivity values (Figure 8); for these samples, MIC values of KR-12 for P. aeruginosa were not totally released at 24 h of bacterial culture.

Concerning the cytotoxicity of the implants against mammalian cells, due to the viability of Vero cells after 24 and 48 h of treatments (Figures 9C, D, respectively) is possible to see that the implants showed no cytotoxic profile, when they were compared with the positive control (+). In fact, during the first 24 h, samples increased the cell density up to 1.5–2 times. This phenomenon can be explained by the fact that cells tend to preferentially grow on rough substrates (Gentile et al., 2010; Ferrari et al., 2019). In the case of the control (+), where no motifs, neither geometric nor biochemical for cell adhesion are present, cell viability values for both the control and the samples were notably higher. In the case of 48 h, samples started to exhibit a reduction in the cell density, but the values were higher than for the positive control (+). The same behavior was also observed in samples loaded with KR-12 peptide.

The functionalization of breast implants through cyclodextrin in-situ polymerization was obtained by an oxidation pretreatment step with O₂ plasma and CHI 1% w/v grafting. CHI facilitated an electrostatic bond between the oxidized surface and the BCD or HPBCD polymers. Among the samples, the textured sample polymerized with HPBCD exhibited the highest polymerization yield and stability. The polymer thickness on the top surface varied with the substrates: smooth samples had a 14–20 µm thickness, while textured samples had a thickness of approximately 100 µm. The diffusivity values (D) for the smooth samples ranged from 1.3 to 3.5 μm²/h. For the textured samples, the diffusivity depended on the molecular weight: PFD (185.22 g/mol) showed the highest D values (121.9–132.6 μm²/h), RB (1017.64 g/mol) exhibited medium values (46.4–82.4 μm²/h), and KR-12 (1,570.95 g/mol) displayed the lowest values (3.9–8.9 μm²/h). Notably, KR-12 demonstrated successful antibacterial activity against S. epidermidis with an antibacterial rate close to 100%. The antibacterial rate against P. aeruginosa bacteria varied, showing dependence on the diffusivity values. Furthermore, the functionalized samples were tested with epithelial cells and confirmed to be non-cytotoxic. Although the performed in vitro assays have confirmed the improved properties of functionalized samples compared to unprotected implants, it is advisable to conduct further research involving biological molecular and in vivo experiments to gain a deeper understanding of the therapeutic effects of the controlled release molecules in this drug delivery system.