Ceramic TiO2 scaffolds were fabricated by polymer foam replication as described by Tiainen et al. [16]. TiO2 slurry was prepared by gradually dispersing 65 g of TiO2 powder (Kronos 1171, Kronos Titan GmbH, Leverkusen, Germany) cleaned with 1 M NaOH into 25 ml of sterilised H2O and stirred at 5000 rpm for 2.5 h. The slurry was adjusted to pH 1.5 during the stirring with 1 M HCl. Cylindrical polyurethane foam templates (60 ppi, Bulbren S, Eurofoam GmbH, Wiesbaden, Germany) were soaked in the slurry and squeezed to remove excess slurry. The polymer sponge was burned out before sintering at 1500 °C for 20 h. After sintering, the scaffolds were immersed with a slurry containing 40 g of powder dispersed in 25 ml of sterilised water. Excess slurry was removed by centrifugation and the second coating was sintered at 1500 °C for 20 h. The finished scaffolds were 7.0-7.5 mm in diameter and 5 mm in height. Prior to use, the scaffolds were steam sterilised at 121 °C for 20 min.
Animals
Six female minipigs (Göttingen minipig, Sus scrofa, Ellegaard AS, Dalmose, Denmark) aged 27-32 months and weighing 42-51 kg were used. The animals were kept in a centre for large experimental animals at the Veterinary Teaching Hospital Rof Codina in Lugo, Spain. The animals were kept on a soft diet and subjected to oral hygiene by mechanical cleaning once every 3 weeks during the experimental study. The Regional Ethics Committee for Animal Research of the University of Santiago de Compostela approved the protocol (Ref. AE-LU-001/002/14).
All surgical procedures were performed under general anaesthesia and sterile conditions in an operating room using propofol (2 mg/kg/i.v., Propovet, Abbott Laboratories, Kent, UK) and 2.5–4% of isoflurane (Isoba-vet, Schering-Plough, Madrid Spain) for the entire period of the surgery. Lidocaine 2% with epinephrine 1:100.000 (2% Xylocaine Dental, Dentsply, York, PA, USA) was infiltrated locally to reduce intra-operatory bleeding and provide local analgesia. The animals were premedicated with acepromazine (0.05 mg/kg/i.m., Calmo Neosan, Pfizer, Madrid, Spain) and pain controlled with the administration of morphine (0.3 mg/kg/i.m., Morfina Braun 2%, B. Braun Medical, Barcelona, Spain). During anaesthesia, a veterinarian continuously monitored the animals.
Experimental design
Mandibular premolars were extracted 4 weeks before implant placement. Mucoperiosteal flaps were bilaterally reflected and teeth carefully removed after tooth separation. Primary healing was accomplished by mattress sutures. Prophylactic administration of cefovecin sodium (8 mg/kg body weight S.C. S.I.D., Convenia®, Zoetis, Spain) was given postoperatively.
Four weeks after extraction, full thickness flaps were raised on the buccal and lingual side and the alveolar crest was levelled to create a flat surface for osteotomies. In each mandibular quadrant, two implant osteotomies were prepared. Thereafter, peri-implant defects were made with a trephine bur, 8 mm in diameter and 5 mm in height (Meisinger Bone Management® Systems, Hager & Meisinger GmbH, Neuss, Germany). The defects were placed with a minimum of 10 mm between the osteotomy centres, and at least 6 mm from neighbouring teeth. The defects were made so that the buccal portion of the defect wall was missing. Dental implants (OsseoSpeed TX 3.0 S 11 mm, Dentsply) were placed in the centre of the defect with the coronal 5 mm of the implants exposed.
The treatment allocation was assigned by a split mouth design where sham and TiO2 alternated between anterior and posterior location for every other animal. For the sites allocated to treatment with the TiO2 scaffold, a cylindrical block scaffold was perforated in the centre, creating a donut-shaped scaffold to fit around the implant and shaped to fit the defect with burs or scissors. The scaffold was secured in the defect by press-fit installation before the flap was passively adapted and primary wound closure was obtained (Fig. 1). After 12 weeks of healing, the animals were euthanized using a lethal dose of sodium Pentothal (40–60 mg/kg/i.v., Dolethal, Vetoquinol, France) and the mandibles were dissected and fixed in formalin.
Micro-computed tomography scanning
Microcomputed tomographic imaging (microCT) of the samples was performed (SkyScan 1172, Bruker microCT, Kontich, Belgium) with a resolution of 7 μm, using 100 kV and 100 μA, with 3 average images every 0.4 degrees for a 360-degree rotation with an aluminium and copper filter. The data were reconstructed using NRecon software (Bruker microCT, Kontich, Belgium) with ring artefact correction of 10%.
The defect fill volume was analysed for a volume of interest (VOI) slightly smaller than the scaffold, to avoid possible errors at the interfaces of the scaffold’s outer and inner border (Fig. 2). The VOI was a cylindrical donut-shaped area with a height of approximately 2.3 mm (including the first ten upper threads of the implant) and a diameter of 6.0 mm concentric with the implant. Individual thresholds was set to measure the defect fill of bone and scaffold. Percentage of defect fill was analysed in CTan (Bruker microCT, Kontich, Belgium).
Histological preparation
The tissue blocks were dehydrated in different graded ethanol series (70-100%), and infiltrated with four different graded mixtures of ethanol and infiltrating resin, glycometacrylate (Technovit 7200® VLC, Heraus Kulzer GMBH, Werheim, Germany). The samples were then polymerised, first under low intensity UV light for 4 h, followed by high intensity UV light for 12 h. Finally, the samples were placed in an oven at 37 °C for 24 h to assure a complete polymerisation. Longitudinal sections in bucco-lingual direction were obtained by cutting with a band saw and mechanically micropolished (Exakt Apparatebau, Norderstedt, Germany) with silicon carbide papers until a thickness for approximately 50 μm was obtained. The slides were stained with Levai Laczko for both histological examination and histomorphometric analysis.
All sections were observed using light microscopy and a PC-based image capture system (BX51, DP71, Olympus Corporation, Japan) and histometrically analysed. Quantitative histology was performed by a masked examiner using PC-based image analysis programs: Cell-sens 1.13 (Olympus Corporation, Japan) and Image Pro-Premier 9.0 (Media cybernetics, Bethesda, MD, USA).
Histological analyses
The images were painted with a digital tablet (Cintiq, Wacom, Japan) and Photoshop CS (Adobe, USA). The region of interest was the buccal and lingual rectangles defined from the top of the implant to the bottom of the defect created and from the surface of the implant to the lateral portion of the defect. The fraction of possible bone fill within each rectangle was calculated, excluding scaffold material in the TiO2 group (Fig. 3a).
Vertical linear measurements were made for two distances: 1) Implant shoulder to the first bone contact (first bone-to-implant contact, FBIC) and 2) Implant shoulder to the first bone contact 500 μm lateral to the implant (FBIC500) (Fig. 3b). Percentage of bone-to-implant contact (%BIC) was calculated for the first 5 mm from the implant top.
Horizontal bone growth within the defect was measured from the implant surface to the first bone contact in the buccal and lingual direction. Measurements were done in millimetre increments from the top of the implant to defect bottom 5 mm apically. The defect dimensions were set as reference points, hence when no bone was seen the value was set at a maximum 3 mm (Fig. 3c).
All measurements were done for both the buccal and lingual side of the implants using ImageJ.
(ImageJ 1.52a, National Institutes of Health, USA).
Linear measurements of soft tissue dimensions were performed for the shortest distance from implant shoulder to the oral cavity using ImageJ (Fig. 4a). The area of soft tissue above the implants was measured in Photoshop CS6 (Fig. 4b). One sample was excluded from the analysis due to missing soft tissue following the histological preparation.
Statistics
Comparison across the two groups was performed using a paired t-test between contralaterals when normality was assumed and Wilcoxon Signed Rank Test when normality test failed. All statistical analyses were performed using SigmaPlot 14 (Systat Software, San Jose, CA, USA). Statistical significance was set at the 0.05 level.