Six mongrel dogs that weighed 280 to 320 newtons were used. A porous-coated segmental prosthesis was inserted into the mid-diaphyseal region of both femora of all animals. The left limb (the control side) did not receive a bone graft, whereas the right limb (the experimental side) was treated with a corticocancellous onlay graft. The animals were followed daily for any signs of fracture or infection until they were killed at twelve weeks postoperatively. The protocol was accepted by the institutional animal-care committee.
Design of the Prosthesis
The prosthetic implant, developed for this study, was a bistemmed segmental diaphyseal replacement made of titanium alloy (Ti-6Al-4V). The porous surface was made of commercially pure titanium beads sintered to the titanium-alloy substrate. The beads were screened to a diameter of 187 to 250 micrometers; the resulting mean pore size was 200 to 300 micrometers. The body of the modular two-component prosthesis was sixty millimeters long and eighteen millimeters in diameter, with a conical coupling in the middle of the body. The intramedullary stems were ten millimeters in diameter and were fluted for secure fixation with cement. The proximal stem was thirty millimeters long. The distal stem was forty millimeters long and was curved to accommodate the proximal curvature of the distal end of the canine femur.
Operative Procedure (Fig. 1)
The femoral diaphysis was exposed with a lateral approach through the fascia lata between the vastus lateralis and biceps femoris muscles. The femoral periosteum was elevated circumferentially, and the insertions of the adductor muscles were incised along the linea aspera. A six-centimeter segment of the diaphysis was resected with a reciprocating saw. The field was irrigated continuously to minimize thermal necrosis of bone. The implant was secured with intramedullary cement. On the control side, the wound was closed in three layers with absorbable sutures. On the experimental side, bone-grafting was performed. The resected femoral cortex was cut longitudinally into eight strips, each strip was cut in half, and the strips were evenly distributed around both the proximal and the distal junction between the prosthesis and the femoral cortex. Each three-centimeter-long strip of cortical bone was placed evenly between the host bone and the prosthesis. Cancellous bone was obtained from the medullary cavity and the metaphyseal area both distally and proximally. It was measured in terms of both volume (thirteen cubic centimeters) and weight (range, 8.5 to 11.4 grams) and was then evenly distributed under and between the strips of cortical bone. The strips were secured with sutures.
Load-Bearing
Load-bearing was measured before the operation and at four, seven, and ten weeks postoperatively. Each dog was led by a leash over a force-plate. The approach distance was at least four meters, which allowed the establishment of a normal gait. An observer recorded each successful foot-strike, which was defined as complete contact of either the left or the right hindfoot within the margins of the surface of the force-plate. At least six successful runs were obtained for each hindlimb.
Radiographic Analysis
Anteroposterior and lateral radiographs were made preoperatively to check for abnormal findings and to ascertain that the intramedullary diameter was sufficient for insertion of the prosthesis. Standard radiographs were also made at one, two, four, six, nine, and twelve weeks postoperatively with the animal sedated with ketamine (eight milligrams per kilogram of body weight, administered intramuscularly) and xylazine (0.8 milligram per kilogram of body weight, administered intramuscularly). The size of the callus, new-bone formation, and the contact area at the bone-prosthesis interface were measured directly from radiographs with an image-analyzer software package (Bioquant System IV; R and M Biometrics, Nashville, Tennessee) and a sonic digitizer (Summa Sketch II Plus; Sammagraphics, Seymour, Connecticut)8,9.
Measurement of the Capsule
After the animals were killed, the muscle attachments and other soft tissues were carefully dissected from the femur, and the attachment sites of the soft tissues were identified. The length of the capsule from the end of the dense fibrous tissue on the host cortex to the nearest edge of the porous coating on the prosthesis was measured with calipers after dissection of the soft tissues. The border between the periosteum and the fibrous capsule on the host bone could be visualized clearly because of the pink color of the tissue of the fibrous capsule and the sudden change in the thickness of the tissue.
Volumetric Analysis of Bone Tissue with Computed Tomography Imaging
Periprosthetic bone formation was analyzed with computed tomography scans. The quality of the imaging data was not affected by the presence of the titanium prosthesis14,15. All scans were made with a Somatom-Plus-4 scanner (Siemens Medical Systems, Iselin, New Jersey). Specimens were aligned with the longitudinal axis of the table. A spiral computed tomography technique was used with a two-millimeter slice thickness. The overall length of the scan was ninety millimeters. A bone-optimization technique was used with a scanning power of 130 milliamperes, a field of view of eighty millimeters, and a matrix of 512 by 512 pixels. The resulting imaging data were recorded from sixteen bit (512 gray levels) to eight bit (256 gray levels). The computed tomography scans were analyzed with a three-dimensional reconstruction software package (Analyze 7.5; Biomedical Imaging Foundation, Mayo Foundation, Rochester, Minnesota) on a Silicon Graphics workstation (Mountain View, California). Voxel intensity was reported as eight-bit gray levels rather than as Hounsfield units, to provide a more generalized description of bone and callus density. Gray-level thresholding was done in increments of five, from seventy-five to 175 of 256 possible gray levels. A gray level of zero corresponded to the density of air, and a gray level of 255 corresponded to the density of titanium. Periprosthetic bone was defined as mineralized tissue surrounding the extracortical area of the prosthesis between its proximal and distal interfaces with the host bone. Periprosthetic bone was rendered in three dimensions with use of voxel gradient shading algorithms. Bone volume was determined by summation of all rendered voxels from the image surface.
Histological Analysis
The prosthesis was separated at the coupling joint, and the distal bone segment was used for histological analysis. The specimens were fixed in 70 percent ethanol solution, dehydrated in increasing concentrations of ethanol, defatted in acetone, and embedded in polymethylmethacrylate (Technovit 9100; Kulzer GmbH, Wehrheim, Germany). Sections taken from the area between the prosthetic shoulder and the bone and ten millimeters distally and proximally (Fig. 2) were ground to a thickness of 100 micrometers. Areas of new-bone formation and the rate of formation were studied with a multiple fluorochromic labeling technique. The animals were given bone labels biweekly: ninety milligrams of xylenol orange per kilogram of body weight, thirty milligrams of calcein blue per kilogram of body weight, and thirty milligrams of alizarin complexone per kilogram of body weight were administered intravenously, and thirty milligrams of oxytetracycline per kilogram of body weight was administered intramuscularly. Unstained bone specimens were viewed under ultraviolet light to allow identification of labeled new-bone formation.
Contact microradiographs were made with a high-resolution film (Industrex SR, Kodak-Industrie, Challon sur Saune, France) exposed with thirty-five kilowatts and twenty milliamperes and a target-to-specimen distance of twenty centimeters with an exposure time of forty-five seconds. The films were developed for three minutes in Kodak film developer (model D-19 developer; Eastman Kodak, Rochester, New York).
Biomechanical Testing
The proximal part of the femur, which included the proximal segmental portion and stem, was used for torsional testing. In order to eliminate the confounding strengthening effect of the cement fixation from the prosthesis-cement-bone system, the cement was removed before torsional testing, as described previously15 (Fig. 2). Radiographs were used to verify that the cement had been removed completely.
After the cement had been removed, the bone-prosthesis specimens were mounted on a materials testing machine (MTS Bionix 858; MTS, Eden Prairie, Minnesota) that had been adapted for torsional testing. The proximal part of the femur was secured to the actuator through a universal joint to ensure pure axial rotation of the specimen. Torsional load versus angular deformation was continuously recorded during testing. A nonphysiologically slow rate of rotational deformation of 15 degrees a minute was used. Each test was carried out in external rotation to a maximum of 50 degrees. The slope of the initial linear portion of the curve was measured as an index of torsional stiffness. Ultimate strength was defined as the maximum torque applied during testing.
Statistical Analysis
Statistical comparison of the experimental and control sides was performed with a paired t test. Time-related data were analyzed with repeated-measures analysis of variance. A difference was considered significant when the p value was less than 0.05. The results are given as the mean and the standard deviation.
All dogs had a transient decrease in body weight of approximately 5 percent during the postoperative period. There were no operative complications. One animal was excluded from the study after analysis of the serial radiographs showed loosening of the prosthesis on the control side four weeks postoperatively. All of the dogs appeared to be free of pain. Intake of food and liquid was normal. Red and white blood-cell counts and serum-electrolyte and protein levels were within normal ranges.
Load-Bearing
All animals were able to stand unassisted on the first postoperative day. The animals bore weight on both involved limbs within two days postoperatively. On the control side, dynamic load-bearing at four weeks was significantly decreased to a mean of 84 percent of the preoperative value (p < 0.05). The levels returned to the preoperative values by seven weeks. With the numbers available, we could not detect a significant difference between the mean values for preoperative and postoperative dynamic load-bearing on the experimental side (Fig. 3) or between the experimental and control sides with regard to load-bearing at any time during the study.
Radiographic Analysis
As mentioned, one prosthesis on the control side was loose distally at four weeks. The loosening was associated with accelerated formation of periosteal callus, and this animal was excluded from the study. The mean total area of the callus was significantly larger (p < 0.02) on the experimental side than on the control side throughout the entire study period (Fig. 4-A). The difference was even more significant (p < 0.005) when the callus was expressed as a percentage of the total surface area of the prosthesis (Fig. 4-B). The mean callus area increased significantly at two weeks (p < 0.05) and leveled off between two and six weeks, with the maximum area present at four weeks, on both the experimental and the control side. Between six and twelve weeks, the mean callus area decreased significantly on the experimental side (p < 0.05) but remained the same on the control side. Although the reduction in callus area was greater on the experimental side, at twelve weeks the mean callus area was still 161 percent of the original area covered by the corticocancellous bone graft immediately after the operation. At twelve weeks, the total callus area and the percentage of the total prosthetic surface area were significantly larger on the experimental side than on the control side (p < 0.01 and p < 0.002, respectively). On the control side, only occasional bone-prosthesis contact was seen. On the experimental side, the bone-prosthesis contact area increased until four weeks postoperatively, after which it remained stable. Of note is the fact that the decrease in callus area did not result in a decrease in bone-prosthesis contact area.
Measurement of the Capsule
The adductor magnus muscle was securely attached to the linea aspera and the posterior portion of the prosthesis on both the control and the experimental side. On the experimental side, the quadriceps and adductor muscles and a dense fibrous capsule were circumferentially adherent to the proximal and distal bone-prosthesis junctions. This dense fibrous capsule was particularly well formed in the anterior and distal region at the junction between the bone and the prosthetic shoulder; the capsule covered the cortical grafts and extended onto the prosthesis and the host bone. In contrast, on the control side, the muscles and a thin layer of fibrous tissue were attached to the extracortical porous surface of the prosthesis but not to the host bone. The mean length (and standard deviation) of the fibrous-capsule attachment was significantly greater (p < 0.05) on the experimental side (43 ± 4 millimeters) than on the control side (11 ± 2 millimeters) (Table I).
Volumetric Analysis of Bone Tissue with Computed Tomography Imaging
The strips of cortical bone graft were still present at twelve weeks, with new bone formed circumferentially on the extracortical part of the prosthesis. The greatest degree of extracortical bone formation was posterior to the prosthesis and the host bone. The mean periprosthetic bone volume was significantly greater (p < 0.05) on the experimental side (4240 ± 820 cubic millimeters) than on the control side (622 ± 605 cubic millimeters) (Table I).
Histological and Histomorphometric Analyses
Histological and histomorphometric analyses of cross sections showed a clear difference in bone area between the experimental side and the control side. The difference was significant (p < 0.02) in the shoulder region of the prosthesis and also ten millimeters proximal to the junction, within the porous coating (p < 0.002) (Table I). Parts of the original autogenous cortical grafts could still be seen in the shoulder region in all experimental specimens. However, most of the original cancellous graft chips had been resorbed. Resorption had occurred superficially in the remaining parts of the cortical bone strips. New-bone labeling was the most intensive superficially around the graft, but some label could be seen in vascular channels inside the graft. However, large areas of the grafts remained necrotic. Direct bonding between new bone and the porous surface of the prosthesis was not seen. The bone envelope covered more than half of the circumference of the prosthesis in all of the grafted specimens; in three of the five specimens, the shoulder region of the prosthesis was totally surrounded by the bone envelope. Ten millimeters proximal to the shoulder region, the bone envelope diminished 38 percent. Only occasional formation of bone and no ingrowth of bone was seen in the control specimens.
Biomechanical Testing
Both the mean torsional stiffness (p < 0.007) and the mean maximum torque at failure (p < 0.05) were significantly greater (eighteen and five times greater, respectively) on the experimental side (Table II). The mean stiffness of the control specimens was only 6 percent that of the experimental specimens.
Our study was designed to determine if modified corticocancellous onlay grafting enhanced extracortical bone-bridging around the porous metallic surface of a metal segmental replacement prosthesis. Cancellous bone is an excellent inducer of new-bone formation. When it is combined with cortical strips, the two types of bone graft might enhance extracortical bone formation by contributing mechanical strength and acting as a scaffold for guided regeneration of tissue. We used resection and reconstruction of the diaphysis to simulate the treatment of a malignant bone tumor. While cement achieves immediate fixation, the long-term stability of this interface is still unknown. Biological fixation with extracortical bone attached to the prosthesis would be ideal to enhance long-term fixation. The bone bridge may be capable of protecting the bone-cement interface by transmitting bending and torsional loads, thereby reducing load transfer between the bone and the cement. Artificial membranes made of polylactic acid, Gore-Tex (W. L. Gore and Associates, Flagstaff, Arizona), and polyurethane have been used to guide this kind of tissue regeneration3,7, but the biological response to such materials is still unpredictable and long-term data are not available. We previously used cancellous and morseled cortical bone grafts in the same model in an attempt to enhance extracortical new-bone formation, but relatively rapid resorption of the graft material followed by a decrease in the size of the callus was a major problem8,15. Since the turnover rate of cortical bone is much lower than that of cancellous bone, large autogenous cortical bone grafts may persist as a mixture of new bone and originally grafted necrotic bone1. We believe that onlay cortical bone strips have a better chance to create a bone bridge between the host bone and the surface of the prosthesis and to maintain it over a long period of time.
When we used the same model with crushed corticocancellous bone graft (rather than intact cortical strips combined with cancellous grafts as in the present study), the stiffness and the ultimate strength of the bone-prosthesis interface were 49 and 32 percent of the current values, respectively15. Also, unlike the previous study15, the present study showed that, at twelve weeks, the extracortical bone area was larger than that immediately after the operation and the contact area between the newly formed bone and the surface of the prosthesis was not reduced throughout the entire remodeling period. On the basis of our short-term results, therefore, it seems that onlay graft may be suitable for guided tissue regeneration and maintenance of the osseous envelope.
Whereas callus formation and extracortical bone-bridging across the cortex and the surface of the prosthetic shoulder were significantly greater when graft had been used, there was a thin layer of fibrous tissue between the titanium surface and the ingrown bone. Despite this thin fibrous layer, the mechanical strength of the interface provided by grafting with cortical bone strips was significantly greater than the mechanical strength when bone graft had not been used.
Enhanced soft-tissue formation may also be important for functional recovery. At four weeks, the animals were able to bear more weight on the side on which the graft had been used, which suggests earlier functional recovery. Rapid functional recovery is especially important for patients who have had a tumor removed because the normal healing process is often delayed by radiation and chemotherapy.
Implantation of a metal prosthesis may result in the release of metal debris into surrounding tissues. Ward et al. showed that the dense fibrous capsule growing into a porous surface effectively seals in the fluid that accumulates within the periprosthetic capsule, thus minimizing the transport of particulate debris12. The results of the present study show that bone-grafting supported the formation of extracortical bone tissue and fibrous tissue.
In conclusion, the short-term results in our segmental resection dog model demonstrated that extracortical bone and soft-tissue formation around a porous-coated surface can be achieved reliably with onlay bone-grafting. This finding is in strong contrast to the lack of such formation in the absence of bone graft. Compared with the results of our previous study in which we used loose crushed corticocancellous bone graft15, the short-term results with the current method appear superior.