Total 36 titanium implants (ASTM Grade IV) with a length of 10 mm, an outer diameter of 3.75 mm, a pitch height of 0.6 mm, and a turned surface were used. In this study, turned titanium screws were used as a control group. Electrochemically oxidized implants incorporating Mg (Mg-implants) in the surface titanium oxide layer were test group (Fig. 1). The Mg-implants were prepared using the micro arc oxidation (MAO) process at high forming voltages and current densities at galvanostatic mode in a magnesium containing mixed electrolyte system.

Chemical surface analysis was investigated using XPS (ESCALAP 250, VG Scientific Ltd). The XPS spectra were recorded using normal Al Kα radiation (1486.8 eV) with probing beam size of 200 µm. The outmost surface of implants was etched with Ar ion with ion energy of 5 KeV and beam current of 0.3 µA for 150 seconds, corresponding to 2 nm in thickness, resulting in removal of surface contaminants. The surface oxides of turned-implants and Mg-implants were consisted mainly of TiO₂. XPS survey spectrum of the Mg-implants revealed the presence of the magnesium elements, Ti, O, C and some traces like P and S (Fig. 2a). The relative atomic concentration of Mg was approximately 7.6% at the as-received surface and 9.3% after Ar⁺ sputter cleaning for 150 seconds. 15% of C at the as-received surface was rapidly decreased after sputter cleaning. This may indicate the C is a surface contaminant. The mean surface porosity was 23.7%. The pore size was ≤ 1.5 µm in diameter. The pore density (pore population/scanning area) was 3.25. The mean oxide thickness was 3.4 µm in test Mg-implants as measured with a scanning electron microscope (S-3000N; Hitachi Science System, Tokyo, Japan) (Fig. 2b) on cross-section view and 17 nm in controls as measured with Auger Electron Spectroscopy (AES). The surface roughness measured with Optical Interferometer (MicroXam™) revealed 0.68 (± 0.2) µm Sa.

Six male beagle dogs, 24 months of age and 13 kg weight, were selected. All subjects were healthy and had no periodontal disease.

Bilateral mandibular premolars were extracted under general anesthesia. The anesthesia protocol for all surgical and follow-up procedures included premedication with acepromazine IM (Sedaject; Samu Median Co., Yesan, Korea), followed by 1:1 ratio mixtures of ketamine hydrochloride (Yuhan-Ketamine; Yuhan Co., Seoul, Korea) and xylazine hydrochloride (Rompun; Bayer Animal Health Co., Seoul, Korea). Initial dosage was 5 cc, and additional dosage 3 cc for maintaining anesthesia. Local infiltration was also performed with 2% lidocaine (Lidocaine HCl; Huons Co., Seoul, Korea).

After a full thickness mucoperiosteal flaps were raised adjacent to the mandibular premolars, teeth were hemisected under copious irrigation with a small fissure bur and disc. Extraction was performed with elevators and forceps. Flaps were closed with single interrupted 4-0 Gore-Tex sutures (Gore-Tex® Suture; W. L. Gore & Associates, Inc., Newark, USA).

Postoperative care protocol included antibiotics (Kanamycin-100; Samyang Pharma-Chem. Co., Seoul, Korea) and pain control (Hipyrine; Samyang Pharma-Chem. Co., Seoul, Korea) for 5 days.

Three months after extractions, surgical placement of implants was performed under sterile conditions. After adequate anesthesia, an incision was made through the mucosa at the crest of the alveolar ridge and a full thickness mucoperiosteal flaps were elevated in each mandibular quadrant over the healed extraction sites. Three implant sites were prepared per mandibular quadrant. Implant sites were gradually drilled with a round bur, 2.0 mm twist drill, pilot drill, and 2.85 mm twist drill (Nobel Biocare, Göteborg, Sweden). During surgical sequence, low rotary speed and profuse saline irrigation were used. The implants were inserted with a custom-made fixture mount via a connector and finally tightened manually with a screwdriver. A distance of approximately 7 mm between implant centers was maintained. The implants were placed in such a way that the fixture margin coincided with the bone crest. Cover screws were connected and the mucoperiosteal flaps were repositioned with mattress and single interrupted 4-0 Gore-Tex sutures. Each beagle dog received 3 control and 3 test implants which were inserted in the left and right side, respectively.

After 4 weeks, the surgical site was reopened for healing abutment connection and impression taking. Non-rotational fixture level impression copings (Neoplant® Neobiotech, Seoul, Korea) were connected to implants and secured with screws. Impression copings on each side of the mandible were splinted together using Duralay pattern resin (DuraLay; Reliance Dental Mfg Co., Worth, USA). Polyether impression material (Impregum; ESPE Co., MN, USA) was used for pick-up impression. After healing abutment connection, flap closure was accomplished with single interrupted 4-0 Gore-Tex sutures, leaving the healing cap exposed.

Fixture analogs were connected to the impression copings. Master casts were poured in type IV dental stone (Die-Keen® Heraeus Kulzer, Inc. New York, USA). Individual crowns were waxed on the abutments to proper contacts and contours. Total 36 single crowns were invested (Investment; Whipmix Co., Louisville, USA) and cast in type III gold (DM 76; Wooridongmyung Co., Seoul, Korea) (Fig. 3a). After devesting, extraoral occlusal adjustments were performed.

In 2 days after impression, the gold crowns were placed on the implants and intraoral occlusal adjustment performed to eliminate any direct occlusal contact on closure in premolar areas. All lateral occlusal interferences were removed. Gold crowns were subsequently connected to the abutments by means of abutment screws with a torque of 20 Ncm (Fig. 3b).

During the study period, three times weekly tooth brushing with a soft brush and 0.2% chlorhexidine gel (USP, Maryland, USA) cleaning were performed. For 2 weeks after each surgery, chlorhexidine swabbing of the surgical wounds was performed instead of tooth brushing to minimize mechanical trauma. In addition, for the first 2 weeks after surgery, the animals received a soft diet in an attempt to reduce mechanical trauma that could negatively influence healing process.

During the experiment, the several parameters were measured: (1) radiographic evaluation of peri-implant radiolucency, (2) measurement of marginal bone levels and calculation of changes over time, and (3) implant stability as measured with resonance frequency analysis (RFA, Osstell; Integration Diagnostics Ltd., Göteborg, Sweden).

For the radiographic interpretations, standardized periapical radiographs were used. The radiographs were taken with a customized occlusal index fabricated by affixing a siloxane putty impression material. For obtaining a parallel image, individual extension cone paralleling (XCP) device was made (Fig. 4). Standardized exposure parameters and processing procedures were applied. Each radiograph was then analyzed for changes in bone levels. Bone level changes were evaluated mesial and distal to each implant measuring the distance in millimeters between the top of implant head and the most coronal point of direct bone-to-implant contact.

RFA measurements were carried out by screwing a transducer (Type F1L5; Integration Diagnostics Ltd., Göteborg, Sweden) to the top of the fixture. Measurements were repeated on each implant by rotating the transducer and orienting it perpendicular and parallel to the long axis of the mandible.

Comparisons were made between implant stability quotient (ISQ) values obtained at the time of fixture installation, 2nd stage surgery, 4 weeks after loading, and 10 weeks after loading.

At the time of 10 weeks after loading, all subjects were sacrificed with an overdose of sodium-pentothal and perfused with a fixative through the carotid arteries. Implants and their surrounding tissues were removed en bloc, and immersed in 10% neutral buffered formalin (Accustain® Sigma-Aldrich®, Inc., Steinheim, Germany) for 1 week. After dehydration in a graded series of ethanols, the specimens were infiltrated, and embedded in methyl methacrylate (Technovit® 7200 VLC-resin, Kulzer, Friedrichsdorf, Germany) for nondecalcified sectioning according to Donath's protocol.20 Transverse sections, 100 µm in thickness, were taken along the long axis of the fixture by using a band saw with a diamond blade (Exakt Appratebau, Norderstedt, Germany), representing the mid portion of each implant site. Each section was ground to approximately 20 µm or less utilizing an EXAKT Micro Grinder, and polished to an optical finish. The specimens were stained with 1% toluidine blue.

A gross histological observations and morphometrical measurements were performed. Leitz Aristoplan™ microscope (Leitz, Wetzlar, GmbH, Germany) and Microvid™ equipment connected to an IBM computer were used. Measurements were carried out directly in the eye-piece of the light microscope using a 10× magnification objective and a zoom of 2.5×. The morphometric measurements were performed blindly. The mean percentages of bone-to-implant contact (BIC) and bone area in each thread (left and right side) were calculated. New bone formation around the implant within the threads was regarded as an R area. Mirror images (S area) of the R areas were also evaluated for status of original bone remodeling (Fig. 5). Marginal bone level as evaluated from the distance between the top of the implant head and the most coronal part of direct bone-to-implant contact.

Associations between the implant type and resulting data were evaluated by analysis of covariance (ANOVA) with repeated measurements. For histomorphometric analysis, the Mann-Whitney U test for comparison in-between implants was performed. And Wilcoxon signed rank test were used for paired comparisons (R area Vs S area) within each animal. With Mann-Whitney U test as well as Wilcoxson signed-rank test, two sided test was used and P ≤ .05 was considered statistically significant.

The changes in marginal bone levels as determined radiographically are presented in Fig. 6. An increase in the bone level indicated bone loss around the implant. Both groups showed gradual bone loss. Comparing the bone levels at baseline and 10 weeks after loading, a statistically significant differences were found according to time lapse (P = .000). Overall bone loss was 1.5 mm in the control group and 1.3 mm in the Mg-implant group. However, there was no statistically significant difference between test implants and controls (P = .46).

The RFA values varied as a function of time. Initial mean ISQ value was 60.5 in the control group and 60.8 in the Mg-implant group (Fig. 7). Mean ISQ was slightly decreased in the control group at second stage surgery. However, these values were almost same at the 10 weeks after loading (62.9 Vs. 62.8 for the control group and Mg-implant group, respectively). The statistical analysis showed that the differences between the implants at implant installation and at 10 weeks after loading (P = .000). However, repeated measured ANOVA could not demonstrate the statistically significant difference between Mg-implants and controls through overall measuring time.

Bone shape, quantity and quality varied the most prominently in the anterior installation site. Implants of both groups were surrounded by mature bone within the threads and in direct contact with the implant surface. Intimate bone-implant contact was established, although it was occasionally interrupted by soft tissue. Bone filled almost the entire area within the threads of the control implants (Fig. 8a). Mg-implants had a thin bone collar in contact with their surfaces, whereas the areas within the threads were not occupied completely by bone (Fig. 8b). No signs of inflammatory reaction were detected in the tissues surrounding the implants.

The mean BIC% in the Mg-implants was 54.5 ± 8.4%. The mean BIC% in the turned implant was 45.3 ± 12.2%. These differences between the Mg-implant and control implant were statistically significant (P = .005) (Table I). Mean BIC in anterior installation site regardless of implant types was 48.1%. These were 46.1% and 39.3% in middle and posterior installation site, respectively. The mean BIC decreased from the anterior site to the posterior site. However, these data did not show a significant difference (P = .35).

The results from histomophometric analysis in bone area are presented in Table II. New bone formation in control implants was 72.8 ± 7.3% (R1 area), and 56.8 ± 8.5% (R2 area) in Mg-implants (Table II). Comparing the bone area inside the threads, there was significantly more bone within the threads in the control group than in the Mg-implant group (P = .000). The bone area in S1 (original bone remodeling in control implants) showed a significantly higher than in S2 (Mg-implants) (P = .023).

First bone contact of Mg-implants was 0.69 ± 0.47 mm, whereas 0.80 ± 0.35 mm in turned implants. However, these data were not statistically significantly different between Mg-implants and controls (P = .817). First bone contact in sectioned specimens was shorter than the data from radiographic measurements.