Design and fabrication of the modular endoprosthesis

Details of the fabrication of the endoprosthesis have been published previously . Two designs were used: condyle replacement (Design 1); and body replacement (Design 2). The device, made of titanium alloy (Ti6Al4V), comprises a two-piece modular system for Design 1 and a three-piece system for Design 2. For Design 1, the first module contained the stem that was to be cemented into the mandibular stump. The second module was shaped to replace the ascending ramus and condyle. The condylar head was designed to articulate against the natural disc and fossa. In Design 2, the stems were cemented to the mandibular stumps at either end and the body of the module, about 70% of the original height of the mandible, made up the third part. The stems were made to fit as closely as possible into the medullary space of the mandibular stump. They were rounded off at the ends to prevent shearing forces generated at the stem–cement interface.

The modules were held together with screws (one screw for Design 1 and two screws for Design 2). A ‘stud’ was later inserted adjacent to the screws to prevent screw loosening in both Designs 1 and 2.

Animal surgery

The material and methods for animal surgery and death have been reported elsewhere . Briefly, 16 young, adult male Macaca fascicularis monkeys, weighing approximately 3.5 kg, were used; eight each for Design 1 and Design 2. The monkeys had all molars erupted except for the third molars.

Permission for the study was granted by the Institutional Animal Care and Use Committee of SingHealth. The animal laboratory has been certified by the International Association for Assessment of Laboratory Animal Care.

For condyle replacement (Design 1), following administration of general anaesthesia, the right mandible was approached from an intraoral route in the first monkey, but due to inadequate access, a submandibular approach was used in the subsequent seven monkeys. Ostectomy was done just posterior to the second molar. The coronoid process was left in situ after coronoidectomy above the level of the sigmoid notch. The condyle was dissected free from its muscle attachments and the mandibular ramus segment, consisting of the ascending ramus and condyle, was rotated and removed, leaving the disc in the fossa. Finger palpation was performed to check the position of the disc in the fossa. Space was created to install the stem in the marrow cavity of the mandibular stump using a fissure bur. The first module, consisting of the stem, was cemented into the marrow space with a PMMA bone cement, Palacos ^® (Biomet Inc., Warsaw, USA). The ascending ramus and condyle module was inserted after ensuring that the condyle had been seated within the fossa. It was connected to the stem module with a screw. The wound was closed in layers with 4×0 Vicryl ^® and 5×0 PDS (Ethicon, Somerville, USA) sutures.

In the body replacement group (Design 2), a segment containing the first and second permanent molars and attached gingiva was resected via an intraoral approach. Bleeding from the inferior mandibular canal was stopped easily. Space was created to fit the stems of the prosthesis. The stems were cemented using polymethy-metacrylate (PMMA) cement (Palacos ^® , Biomet, Florida, USA) and the body was connected to the stems with two screws after cement setting. The wounds were closed.

The monkeys had an uneventful recovery and were put on a soft diet for the first 2 weeks. They received 2 mg/kg Rimadyl ^® (Pfizer, New York, USA) subcutaneous (s.c.), for analgesia for 2 days and 5 mg/kg Baytril ^® (Bayer, Leverkusen, Germany) and 15 mg/kg amoxicillin s.c. for 5–7 days. For the two designs (Designs 1 and 2), with eight monkeys in each group, half of the monkeys were killed after 3 months and the other half after 6 months. A perfusion fixation method was used. Under general anaesthesia, the animal’s thorax was opened by cutting the sternum, followed by exposure of the heart. A 16-gauge intravenous catheter was inserted into the left ventricle and used for rinsing the circulation with 300–500 ml of Hartman’s solution followed by 800 ml of a mixture of 2.5% paraformaldehyde and 2% glutaraldehyde. The right atrium was cut to allow the blood and fluid to flow out. The operated side of the mandible was then harvested together with the endoprosthesis.

Preparation of specimens

The harvested mandibles were scanned using a cone beam 3-D dental imaging system I-CAT ^® (Imaging Sciences International, Hartfield, PA, USA) to determine the position of the endoprosthesis in the mandible. The specimens were sectioned and the parts containing the stems of the endoprosthesis were used for further scanning. These were placed in a micro-computed tomography (micro-CT) scanner, SkyScan ^® 1072 (SkyScan, Aartselaar, Belgium) and scanned by transmitted X-rays with a focal spot of 6 μm and a pixel size of 18 μm. The X-rays were absorbed and transformed into light by a phosphorous screen. This light was detected by a charged couple device (CCD) camera consisting of a 2-D 512 × 512 array of pixels. The digitized signal was transferred to a PC where the micro-CT images were recorded and reconstructed. After scanning, the regions of the stems from the base to the tip of the stems were selected and virtual cross sections with a 32-bit dynamic range were reconstructed perpendicular to the long axis of the stem. This was converted into 8-bit bitmap (BMP) images according to the selected density window. The advantage of the 8-bit BMP images is that they can be visualized on the computer monitor. Every pixel in an 8-bit BMP image has a colour or grey value between 0 and 255.

Colour 255 was assumed to be white (void space), whereas 0 is black or the densest part of the image. Different colours in the reconstructed cross-sections of the micro-CT image can be correlated to a given mineral content.

Analysis of images and grey values

The images of the micro-CT scans were analysed using CT analyser software (SkyScan, Aartselaar, Belgium). To prevent inter-operator variability in the measurement of the BMD, only one person used the software. At the buccal, lingual and inferior regions of the stems, a circular region of interest (ROI) was chosen manually with a diameter of 0.52 mm consistently in the mid-region of the stems. The location of the ROIs was always 1.5 mm from the stem to allow space for the cement; however, if the ROI was seen to be in the region of the cement, it was moved just adjacent to the cement ( Fig. 1 ). The ROI was then extrapolated through all the relevant slices, thus forming a volumetric ROI or volume of interest (VOI).

ROI at the stems of the endoprosthesis.

To express grey values as mineral content, calibration with bone phantoms with a known value of calcium density was used, in this case hydroxyapatite (HA) Ca ₁₀ (PO ₄ ) ₆ (OH) ₂ was chosen, as the chemical composition of this compound is similar to the mineral part of bone. The bone phantoms used gave a value of 3500 Hounsfield Unit (HU) for 0.25 g/cm ³ of calcium and 5750 HU for 0.75 g/cm ³ of calcium.

Selection of the grey scale values using the binary image was set manually and kept constant for all the images in each specimen. The bone density was calculated using the software, according to the formula derived after calibration with the bone phantoms:

Control group

It is important to know the baseline reading of BMD to be able to compare the effects of surgery. The contralateral left sides of the mandibles in this study were used concurrently to place bone plates for another study; it was considered unsuitable to use them as controls for this BMD study. The unoperated contralateral left mandibles from three monkeys, which had been killed 6 months after undergoing mandibular reconstruction with the modular endoprosthesis (from another study to be published) were used as controls. The regions of the third molar and the second premolar (about 10 mm length), which correspond to where the stems of the endoprosthesis would be inserted, were selected and scanned with a microCT scanner, SkyScan ^® 1072 (SkyScan, Aartselaar, Belgium). ROI in the buccal, lingual and inferior portions were manually selected using the same CTanalyser software. The bone mineral densities of these regions were calculated.

Statistical analysis

The mean BMD was calculated for each region. A χ ² analysis was done to compare the different regions separately. A rank-based statistical analysis was implemented to test non-parametric hypotheses arising from a two-factor crossed design with unequal number of observations per cell in SAS Version 9.1 . The test assumed independent observations, a fixed number of levels, and several observations per cell. The two factors compared were: time group with three levels, control, 3 months and 6 months; and design type with two levels, body replacement and condyle replacement. The null hypotheses were as follows. No significant differences exist in the lingual, buccal, and inferior bone density measurements between: baseline and 3 months post implantation; baseline and 6 months post implantation; and 3 months and 6 months post implantation. No significant difference exists in the lingual, buccal, and inferior bone density measurements between the body and condyle replacement designs. No significant interaction effect exists between time and design in the lingual, buccal and inferior bone density measurements.