Implant functionalization with mesoporous silica: A promising antibacterial strategy, but does such an implant osseointegrate?

OBJECTIVES: New strategies for implant surface functionalization in the prevention of peri-implantitis while not compromising osseointegration are currently explored. The aim of this in vivo study was to assess the osseointegration of a titanium-silica composite implant, previously shown to enable controlled release of therapeutic concentrations of chlorhexidine, in the Göttingen mini-pig oral model.
MATERIAL AND METHODS: Three implant groups were designed: macroporous titanium implants (Ti-Porous); macroporous titanium implants infiltrated with mesoporous silica (Ti-Porous + SiO2 ); and conventional titanium implants (Ti-control). Mandibular last premolar and first molar teeth were extracted bilaterally and implants were installed. After 1 month healing, the bone in contact with the implant and the bone regeneration in the peri-implant gap was evaluated histomorphometrically.
RESULTS: Bone-to-implant contact and peri-implant bone volume for Ti-Porous versus Ti-Porous + SiO2 implants did not differ significantly, but were significantly higher in the Ti-Control group compared with Ti-Porous + SiO2 implants. Functionalization of titanium implants via infiltration of a SiO2 phase into the titanium macropores does not seem to inhibit implant osseointegration. Yet, the importance of the implant macro-design, in particular the screw thread design in a marginal gap implant surgery set-up, was emphasized by the outstanding results of the Ti-Control implant.
CONCLUSIONS: Next-generation implants made of macroporous Ti infiltrated with mesoporous SiO2 do not seem to compromise the osseointegration process. Such implant functionalization may be promising for the prevention and treatment of peri-implantitis given the evidenced potential of mesoporous SiO2 for controlled drug release.

INTRODUCTION

Alterations in manufacturing processes and surgical techniques in implant therapy together with decreasing costs have rendered implant therapy for oral rehabilitation in partially and fully edentulous patients common. Additionally, high success levels (>95%) and good predictability have motivated patients to choose treatment with implants (Buser et al., 2017; Esposito et al., 2014; Naert et al., 2002; Quirynen et al., 2014).

Developments in implant dentistry were mainly related to enhance the rate of implant osseointegration as well as to address conditions with impaired bone quality at the bone‐implant interface (Joos et al., 2006; Nkenke & Fenner, 2006; Rupp et al., 2018). In order to accelerate the process of osseointegration and improve the strength of the established bone‐implant interface, implant surface characteristics have been investigated extensively (Meirelles et al., 2008; Pellegrini et al., 2018; Smeets et al., 2016; Wennerberg & Albrektsson, 2010). Optimization of the bone‐implant interface has been obtained by incorporating inorganic phases within or onto the titanium oxide layer, or by increasing the roughness (Cardoso et al., 2014; Dohan Ehrenfest et al., 2011; Kim et al., 2010; Wennerberg & Albrektsson, 2010). Despite that, patients still lose implants as a result of mechanical or biological complications. Summarized research findings point out two aspects as main causes of implant failure: occlusal overload and peri‐implantitis (Naert et al., 2012; Renvert & Quirynen, 2015; Schwarz et al., 2018).

Research is currently focusing on the development of effective treatment protocols to prevent or to bring to a halt the peri‐implantitis disease in patients receiving dental implants (Jepsen et al., 2015; Qin et al., 2018; Renvert & Polyzois, 2018; Sanz & Giannobile, 2018; Schwarz et al., 2015; Tonetti et al., 2015). Whereas the surface characteristics of the implant components influence not only the biocompatibility, but also the bacterial adhesion and colonization, the mechanisms restricting the formation of dense layers of micro‐organisms, so called biofilms, on these surfaces have been studied, with the rationale to decrease the initial bacterial adhesion and minimize the subsequent inflammation of the peri‐implant tissues (Lang et al., 2000; Mabboux et al., 2004; Norowski & Bumgardner, 2009; Sardin et al., 2004; Sennhenn‐Kirchner et al., 2009; Zhou et al., 2017). Among the possibilities to prevent or treat the increasing peri‐implantitis incidence and at the same time reduce the risk of antibiotic resistance development related to systemic drug administration are implant surface modifications that prevent adhesion of pathogens at the implant surface (e.g., nanostructured TiO2), the local administration of antimicrobial or antibiofilm drugs at the implant site (e.g., Zn‐ or Ag‐modified TiO2, chlorhexidine‐grafted titanium, chlorhexidine releasing hydroxyapatite coatings), or the physical removal (debridement) of biofilms (Campoccia et al., 2013; Hallström et al., 2012; Qian et al., 2012; Xu et al., 2017). In that context, we previously designed a dental implant composed of a porous titanium‐silica (Ti/SiO2) composite material and containing an internal reservoir that can be loaded with antimicrobial compounds. The antimicrobial compounds can diffuse in a controlled manner through the porous implant walls, thereby reducing microbial biofilm formation on the implant surface (Braem et al., 2015; De Cremer et al., 2017). Hence, these implants allow controlled release of compounds from the implant inside towards the tissues outside, favoring prevention and treatment of biofilm formation because of the local treatment while avoiding the occurrence or progression of peri‐implantitis. Proof‐of‐concept for the drug delivery functionality and concomitant antimicrobial activity of chlorhexidine released from this composite material has been established in vitro, and this is in a biofilm preventive as well as curative setup using the oral bacterial pathogen Streptococcus mutans (Braem et al.,  ). Yet, to further prove the usefulness of this novel material in the clinic, its influence on osseointegration with and without a functional drug release needs to be addressed as well.

Therefore, the present experiment evaluated the osseointegration potential of such porous Ti/SiO2 composite implant material in the absence of any drug release. Implant osseointegration was compared to the osseointegration of the macroporous Ti implant as such, as well as to a commercially pure dense Ti implant. The experiment was performed in the mandible of the Göttingen mini‐pig and implant osseointegration was evaluated by histomorphometry.

MATERIALS AND METHODS

Animals and surgery

Six 36‐month‐old female Göttingen mini‐pigs (Ellegaard, Göttingen minipigs A/S, Dalmose, Denmark) were used as experimental animals, due to their known physiology and similarities with human physiology (Stadlinger et al., 2012). The Ethics Committee for animal research of KU Leuven approved the study (P074/2015), accordingly to the regulations and guidelines of the Belgian animal welfare.

In order to allow implant installation, the last premolar and first molar on both sides of the lower jaw were extracted in a veterinary hospital under general anesthesia, following standard protocol (Figure 1). The animals were pre‐anesthetized with butyrophenone neuroleptic sedative (1 ml Stresnil i.m. Aurischter, Annadale, Australia), followed by anesthesia (1 ml/10 kg body weight i.v. of a cocktail of Zoletil 100 (Virbac, Barneveld, the Netherlands) with xylazine (Vexylan, CEVA, Brussels, Belgium). Anesthesia maintenance was performed using gas anesthesia with 1.5% isoflurane (Isoflo, Abott, QC, Canada). Post‐operatively, the animals received analgesia for 3 days by buprenorfine 0.005 mg/kg body weight i.m. (Temgesic, Schering‐Plough, Brussels, Belgium) and antibiotic therapy during 5 days by amoxycillin (Duphamox LA, 0.1 ml/kg body weight i.m., Norbrook Laboratories Ltd., Newry, Ireland).

FIGURE 1

Clinical illustration of tooth extractions (a–c) and implant installation (d–f)

Three implant groups were designed as follows:

Group 1 (n = 6): custom‐made porous titanium screw‐shaped implant (Ti‐Porous; Figure 2a) with dimensions of Ø 3.5 × 11 mm, fully interconnected open porosities of 30% and pore window size: 0.5–2.0 μm.

FIGURE 2

Design and dimensions of the implants. (a) Custom‐made porous titanium implant with or without mesoporous SiO2 diffusion barrier (group 1: Ti‐Porous; group 2: Ti‐Porous + SiO2), developed by MTM KU Leuven (Braem et al., 2015; De Cremer et al., 2017). (b) Commercial dense c.p. titanium implant (Astra Tech Osseospeed 3.5S, Denstply Sirona, Mölndal, Sweden, group 3: Ti‐Control). Scale bar: 1 mm

Group 2 (n = 7): custom‐made porous titanium‐silica composite screw‐shaped implant (Ti‐Porous + SiO2; Figure 2a), with dimensions of Ø 3.5 × 11 mm. The composite material consists of a sol–gel derived mesoporous SiO2 diffusion barrier (pore diameter of 6.4 nm) integrated in the macroporous Ti load‐bearing structure.

Group 3 (n = 10): commercial dense c.p. titanium screw‐shaped implants (Astra Tech OsseoSpeed 3.5 S, Dentsply Sirona, Mölndal, Sweden) with dimensions of Ø 3.5 × 8 mm (Ti‐Control; Figure 2b).

Two months post‐extraction, healing was evaluated under sedation through clinical evaluation and intraoral digital radiographs (Soredex, Tuusula, Finland) with following parameters: 70 kV, 8 mA and 0.16 s. Four months post‐extraction, the installation of the implants was performed (identical analgesia, anesthesia and postoperative care protocol as was applied for the tooth extractions; Figure 3). Either one or two implants were installed in each quadrant. The allocation of the implant types to the operation sites was predefined taking into account an even distribution of the different implant types over the different positions (left vs. right quadrant; mesial versus distal position). After crestal incision, pilot holes were placed with precision drills at regulated speed. Wider drills were used to expand the implant holes (Ø 2.0 mm; Ø 3.0 mm; Ø 3.35 mm), followed by over‐preparation of the cavity using a Ø 3.85 mm bur. In this way, a circumferential gap of nearly 0.35 mm around all implants was created. Surgery was performed under abundant saline irrigation. The screw‐threaded part of the implants was placed equi‐osseus with the vestibular bone as a reference. Implants were inserted manually, without the use of a torque wrench. Because of the created defects around the implants, the implants did not have primary stability. Therefore, submersion underneath the gingiva of the implants was performed to allow proper healing. Given the height of the cover screws and the length of the implants of groups 1 and 2, and given the striving towards non‐tension primary closure, the buccal periosteum was relieved using a curved blade in the apical‐horizontal direction. Flaps and surgical wound margins were sutured. Furthermore, in order to avoid adverse mechanical forces affecting the implant osseointegration process because of mastication, a soft diet was applied from implant installation onwards.

FIGURE 3

Evaluation of healing 1 month post‐implantation. (a and b) Representative intraoral X‐ray images of implant positioning of experimental and commercial implants, respectively. The schematic below indicates the position of the implants in the lower jaw bone. (c) Clinical observation of the implant site

After 1 month post‐implantation, healing was evaluated under sedation through clinical evaluation and intraoral digital radiographs, similarly as described above (Figure 3). Subsequently, the animals were euthanized by an intravascular injection of an embutramide‐mebenzoniumjodide‐tetracaine‐HCL solution (1 ml/5 kg body weight, T61, Intervet, Mechelen, Belgium) into the ear vein until cardiac arrest occurred.

Sample processing and analysis

The implants with the surrounding jaw bone were harvested. The bone blocks were fixed by immersion in a CaCO3−buffered formalin solution (4%), dehydrated in an ascending series of ethanol concentrations for 18 days and embedded separately by infiltration of a benzoylperoxide (0.018%)–methylmetacrylate solution.

X‐rays of the bone samples with implants were taken to confirm the presence of the implants. These X‐rays were also used as a reference to define the cutting plane for separating the bone block into two halves containing each one single implant. Subsequently, the embedded bone blocks were mounted on a precision diamond saw (Leica SP 1600, Leica Microsystems, Nussloch, Germany). The cutting orientation was defined parallel to the implant and perpendicular to the jaw. Slices of approximately 700 μm thickness were obtained. Two or three sections per sample were selected for analysis. The sections were micro‐ground under running tap water and polished to a final thickness of 120–130 μm (Exact 400 CS grinding device, Exact Technologies Inc., Norderstedt, Germany). Finally, sections were stained with a combination of Stevenel's blue and Von Gieson's picrofushin red that allowed the visualization of mineralized bone (red) and demineralized tissue (blue‐green). Additionally, representative sections were analyzed by scanning electron microscopy (SEM), as described before (Braem et al., 2014). To this end and prior to histological staining, sections were ground with a 4000 grit SiC grinding paper. To produce a thin conductive film on the surface, the sections were coated with Au‐Pd with a sputtering device.

Morphological and histomorphometric analyses were performed by light microscopy (Leitz Laborlux S, Wetzlar, Germany), with magnifications of 40×, 100× and 400×, or by SEM (XL30‐FEG, FEI, The Netherlands), performed under standard vacuum conditions at 10 mm working distance and 20 kV accelerating voltage using backscattered electron imaging. The histomorphometric evaluation was conducted with an ultra‐sensitive color video camera (AxioCam MRC5, Zeiss, Gottingen, Germany) mounted on an amplification glass with amplification of 10× (WILD Heerbrugg type MDG17, Switzerland) and using a color image analysis software (Axiovision 4.0, Zeiss) and a customized script particularly programmed for the needs of our analysis. The quantified variables were:

Bone‐to‐implant contact (BIC), calculated as the proportion of the implant regions in direct contact with the bone over the total length of the implant embedded in bone tissue (Figure 4a,b).

FIGURE 4

Illustration of the assessment of BIC and BV, shown for a commercial Ti‐Control implant. (a) Total length of the implant embedded into bone (indicated by the yellow line); (b) Summation of the lengths of the implant portions in direct contact with the bone (indicated by the yellow lines). Bone‐to‐implant contact (BIC) is then defined as the ratio of the latter over the total length of the implant. (c) A predetermined region of interest (ROI) for bone volume (BV) analysis is extending either 500 or 800 μm from the implant surface. For the current example, the ROI is extending 800 μm from the implant surface (delignated by the red line); (d) Subtracted image, solely visualizing the ROI. BV is then defined as the percentage of the ROI surface area occupied by bone. Staining with Stevenel's blue and Von Gieson's picrofushin red. Scale bar: 1 mm

Bone volume (BV), calculated as the percentage area occupied by bone within the predetermined areas of interest. The BV was calculated for two areas of interest defined by two circumferential regions extending 500 and 800 μm from the implant surface, respectively. The rationale of this analysis was to identify to what distance from the implant surface the bone formation might have been influenced by the specific implant surface. To assess BV, the structures were semi‐automatically depicted using customized scripts running in the Axiovision images analysis software (Figure 4c,d). Values were expressed as a percentage of the total area of the region of interest.

Statistical analysis

Statistical analysis was performed using the software package Stat Plus, a statistical analysis program for MAC OS Version v6, for all the parameters. The F‐test for variance, and t‐test to compare the means, assuming equal (homoscedasticity) or unequal (heteroscedasticity) variances, were adopted. Means and standard deviations are given for each implant group. The results were verified using the one‐tailed t‐test (with 95% confidence intervals), with a significance level <0.05.

Ethics statement

The Ethics Committee for animal research of KU Leuven approved the study (P074/2015), accordingly to the regulations and guidelines of the Belgian animal welfare.

RESULTS

Wound healing, both post‐extraction and post‐implantation was uneventful. However, out of the 23 installed implants, two, two and one implants were lost in implant groups 1, 2 and 3, respectively, resulting in an overall implant osseointegration success of 78%. Per group, the success rate was 66.66%, 71.42% and 90%, respectively. Clinical inspection after 1 month of healing showed that some healing abutments had punctured the mucosa and were exposed to the oral cavity. This was observed ad random over the different groups. The gingiva and mucosa surrounding these exposed abutments was evaluated as healthy (Figure 3c).

Histological analysis did not reveal signs of infection. The filling of the defect with newly formed bone was incomplete after 1 month of healing in all specimens. Bone formation occurred from the base and the lateral walls of the defects. The presence of hematopoietic and pre‐osteoblastic cells could still be observed, as result of the ongoing bone regeneration process with the neoformation of bone trabeculae. Despite the fact that the implants have been installed without primary stability, BIC was observed in some samples though to a limited extent (Figure 5a–f). Yet, in the event of BIC, both Ti‐Control as well as experimental Ti‐Porous and Ti‐Porous + SiO2 samples showed a close bone apposition (Figure 6).

FIGURE 5

Representative histological illustrations of bone healing surrounding the implant (1 month healing). (a) Bone trabeculae arranged in a tridimensional network with some parts in contact with the implant surface. The gap between the implant and the host bone is partially filled with newly formed woven bone. Scale bar: 200 μm. (b) Initial signs of tissue organization including cubical osteoblast‐like cells interlaced into a network of recently released bone matrix. Scale bar: 50 μm. (c) Ridges of osteoid and woven bone and areas of immature marrow. Scale bar: 200 μm. (d) Bone trabeculae formation prevalently started from the lateral walls of the host bone and was directed towards the implant surface, with soft tissue interposed. Fibrous tissue is running parallel to the implant surface. Scale bar: 200 μm. (e) Bone apposition on the surface of the newly formed bone on the lateral wall of the defect. Scale bar: 200 μm. (f) Well‐organized tissue including the presence of trabecular mineralized bone connected to areas of non‐mineralized bone tissue. The presence of blood vessels suggests a higher level of tissue organization. Scale bar: 200 μm

FIGURE 6

SEM images of the bone/implant interface (1 month healing). Bone‐to‐implant‐contact (BIC) could only be observed in limited zones for all implant conditions, but showed a close bone apposition for both (a) Ti‐Control and (b) experimental Ti‐Porous + SiO2 samples. Scale bar: 200 μm

Table 1 summarizes the histomorphometric results for the different implant groups. Because several implants were lost upon exposure in the mini‐pig oral cavity, these data refer to a mean (±standard deviation) for four, five or nine implants for the Ti‐Porous, Ti‐Porous + SiO2 resp. Ti‐control groups. BIC did not differ significantly for Ti‐Porous versus Ti‐Porous+SiO2 implants. In contrast, significantly higher BIC values were measured for Ti‐Control implants compared with Ti‐Porous + SiO2 implants. Regarding the amount of bone regenerated in the peri‐implant gap at a distance of 500 and 800 μm from the implant surface (BV), no significant differences were observed for both gap distances between the Ti‐Porous and Ti‐Porous + SiO2 implants. Likewise BIC, Ti‐Control implants revealed significantly higher values for BV compared with Ti‐Porous + SiO2 implants.

TABLE 1

Mean values (%) and standard deviation of the histomorphometrical results for BIC and BV for the groups Ti‐Porous, Ti‐Porous + SiO2 and Ti‐Control (the data result from a total number of 4, 5 resp. 9 implants)

Parameter	Ti‐Porous	Ti‐Porous + SiO2	Ti‐Control	p‐value* for Ti‐Porous versus Ti‐Porous + SiO2	p‐value* for Ti‐Control versus Ti‐Porous + SiO2
BIC	1.15 (±2.62)	00.99 (±2.14)	17.92 (±10.55)	0.438	0.00001
BV 500	31.43 (±22.04)	24.60 (±18.52)	61.65 (±10.61)	0.273	2.764E−8
BV 800	46.30 (±22.14)	38.12 (±21.26)	70.10 (±7.45)	0.199	1.662E−7

Abbreviations: BIC, bone‐to‐implant contact; BV, bone volume, within a circumferential region extending either 500 or 800 μm from the implant surface.

p < 0.05.

DISCUSSION

Implant dentistry research has focused on the optimization of dental implants, aiming to obtain faster and better results by modulating the biological process of bone healing. This can be obtained through different ways of alterations, such as the implant surface topography (macro‐scale, micro‐scale and nano‐scale), the implant shape, and its chemical composition that can significantly affect protein adsorption and modulate cell adhesion, proliferation, differentiation and rate of mineralization in the early stages of bone healing. But implant modification is now gaining interest from another perspective, namely the perspective of preventing and tackling peri‐implantitis (Bumgardner et al., 2011; Johnson & García, 2015; Karoussis et al., 2018; Shahi et al., 2017; Shi et al., 2015). Braem and co‐workers established proof‐of‐concept for the drug delivery functionality of a mesoporous SiO2 barrier incorporated into a high‐strength macroporous Ti carrier (Braem et al., 2015; De Cremer et al., 2017). This novel and promising implant design possesses an internal reservoir which can be refilled, thereby ensuring controlled release of antimicrobial or antibiofilm compounds over sustained period of time.

Prior to evaluating the release of antibacterial compounds in a peri‐implantitis set‐up in vivo, evidence should first be provided that such an implant with specific surface characteristics osseointegrates as other implants do. Therefore, the present experiment was conducted for evaluation of the effect of SiO2 functionalization of a macroporous implant on the peri‐implant bone healing in vivo in the Göttingen mini‐pig, by comparing to an identical macroporous implant without modification with SiO2, and to the well‐documented c.p. titanium implant. It was hypothesized that the integration of SiO2 in the implant does not hamper implant osseointegration.

Regarding the osteotomy defect, filling of it with newly formed bone was incomplete in all specimens in this stage of healing of 1 month post‐implantation. These results are in agreement with previous work, investigating early healing periods in animal models (Botticelli et al., 2005; Rossi et al., 2012). Despite the absence of primary implant stability and thus the presence of implant (micro)movements, histology illustrated bone regeneration activity around the implants in all samples, regardless of the implant group analyzed, with the presence of blood cells, cubical osteoblast‐like cells interlaced into a network of recently released bone matrix and trabecular mineralizing and mineralized bone connected to the parent bone. Given the size of the peri‐implant defect and the short time span for healing, only limited zones with bone‐to‐implant contact were observed, but with close bone apposition for all samples.

Besides the bone tissue in contact with the implant, the amount of bone in the implant's surroundings contributes to the solid anchorage of an implant into the bone. For this reason, not only BIC but also BV in the closer (BV500) and broader (BV800) implant vicinity was calculated. No significant differences were found for BIC and BV500/800 for Ti‐Porous + SiO2 versus Ti‐Porous, suggesting that the functionalization of the Ti‐Porous implant with a mesoporous silica did not affect the peri‐implant bone response, neither negatively nor positively. Likewise, Inzunza and co‐workers illustrated that the viability and proliferation of the osteoblast‐like cells is not altered in contact with a mesoporous SiO2 coating on titanium (Inzunza et al., 2014). Moreover, in combination with bioactive glass, which is known to improve osseointegration around titanium (Braem et al., 2014), such coatings can also accelerate the formation of bone tissue in the implant periphery (Covarrubias et al., 2016). As the mesoporous SiO2 phase enables the controlled release of drugs into the implant surroundings (Braem et al., 2015) and as it was shown in the present study that SiO2 does not affect the osseointegration of the implant, such “hybrid” implant designs are promising for further exploration and use in a peri‐implantitis set‐up.

It should be taken into account that cell behavior and the consequent bone formation around an implant are not solely depending on specific isolated implant surface properties. The cellular response should be considered as a complex phenomenon that depends on several factors acting simultaneously, such as shape, chemical composition and macro/microtopography of the implant as well as the characteristics of the host bed. When comparing Ti‐Control with Ti‐Porous + SiO2 implants, it was observed that BIC and BVs values were significantly more elevated for the control group in this early healing stage. One of the important variables influencing and directing the healing process of bone in the early stages after implantation is the implant design, in the perspective of the osteoconductive property of titanium. The implant macro‐design differed between the control and the experimental group with respect to the screw thread design, neck design (microthreads) and apex design (bone window allowing tissue interlocking and providing anti‐rotational resistance to forces). These differences may have co‐influenced the biological and mechanical micro‐environment leading to a differential peri‐implant bone healing response. In an attempt to exclude the implant macro‐design differences between the control and experimental implants, a larger‐than‐clinically‐advised osteotomy was created. The implant cavity was prepared at wider dimensions compared to the implant diameter. This notwithstanding, it seemed that the implant neck microthread and body thread pitch, width and depth of the control implant were still favoring the bone‐to‐implant contact and bone volume in the implant's surrounding. In line with these thoughts, the difference between both groups (Ti‐Control versus Ti‐Porous + SiO2) was larger when analyzing BV500 compared to BV800. Repeating the present study while adopting longer healing periods and/or while using a more appropriate control implant design (which was not possible for the present implant prototype with an internal reservoir for drug release) is under consideration. Of note is that, in the classical surgical protocol of close bone‐to‐implant contact at installation, resident bone first resorbs prior to new bone formation (Rossi et al., 2014; Slaets et al., 2006; Slaets et al., 2007). A small gap around the implant during the early phases of healing may thus accelerate bone regeneration.

CONCLUSION

Overall, based on the findings and the limitations of the present study, it can be concluded that the functionalization of a macroporous titanium implant with SiO2 does not negatively affect the peri‐implant bone healing response, as observed in the jaw bone of Göttingen mini‐pigs. At the same time, the importance of the implant macro‐design, in particular the screw thread design in terms of number, width, depth and pitch, in a marginal gap implant surgery protocol was emphasized. Subsequent studies will explore the use of such Ti/SiO2 implants in preventing and treating peri‐implantitis given their potential for controlled release of antibiofilm compounds (Braem et al., 2015; De Cremer et al., 2017).

CONFLICT OF INTEREST

The authors declare that there is no conflict of interest in this study.

AUTHOR CONTRIBUTIONS

Study design: KV, KT, JV, JAM, JM, BPAC, AB. Study conduct: KV, MFM, GRC, JA, PT, WJdS, KDC, EG, AB. Data collection: KV, MFM, PT, WJdS, AB. Data analysis: KV, MFM, WJdS, AB. Data interpretation: KV, KT, MFM, JV, AB. Drafting manuscript: KV, KT, MFM, WJdS, KDC, AB. Approving final version of manuscript: all authors.