Animals
Twenty-six neutered male goats that weighed 45 to 55 kg were used in this study. The study protocol was approved by the Institutional Animal Care and Research Advisory Committee and complied with the procedures detailed in the
Guide for the Care and Use of Laboratory Animals, published by the National Institutes of Health. All animals were acclimated to the laboratory environment. Physical and radiographic examinations indicated that the femoral tibial joints were normal. Eight tibiae from four additional animals of similar breed, size, age, and gender from a concurrent study were used as biomechanical and histomorphological normal controls.
Tibial Plateau Fracture Model
With the animals under general anesthesia, both hind limbs were prepared for surgery under sterile conditions. A specifically designed jig that accepted an 8-mm drill-bit was positioned over the lateral tibial plateau under fluoroscopic guidance. The jig was positioned and secured to the distal aspect of the femur with use of two 3.2-mm drill-bits placed through guide ports within the jig. A small incision was made through the jig port, exposing the lateral tibial eminence. With use of an 8-mm drill-bit placed through the jig port, a cylindrical defect that was 8 mm in diameter and 10 mm deep was created within the plateau (
Fig. 1 ,
A ). The jig was removed. A 5-mm osteotome placed into the defect was used to fracture the subchondral plate and articular cartilage along the defect margins, creating a 5-mm-thick osteochondral fragment (
Fig. 1 ,
B ). Through a small arthrotomy, the articular fragment was displaced into the defect with use of a periosteal elevator (
Fig. 1 ,
C ). With use of the same elevator, the fragment was reduced to its normal anatomical position under direct visualization (
Fig. 1 ,
D ). Identical procedures were performed on the contralateral limb.
Treatment Groups
Each defect was randomly assigned to be filled with either calcium phosphate cement (a-BSM) or corticocancellous autograft reamings obtained from the defects. Hemostasis prior to grafting was maintained by packing the defects tightly with sterile gauze. The calcium phosphate cement was supplied as a powder in 2.5-mL plastic injection bulbs. The cement was prepared by injecting 0.8 mL of sterile saline solution per gram of powder into the bulb. After gentle mixing, the bulb injection cap was removed and the cement was expressed from the bulb into a 3-mL syringe. The needle hub of the syringe was removed, and the calcium phosphate cement was slowly injected until it completely filled the defect. Autograft was delivered into the defects with use of a similarly prepared syringe. Both materials were gently tamped with use of an 8-mm steel dowel to ensure complete filling of all defects. With gentle tamping, approximately 2 g of cement (2.7-mL volume) was required for each defect. Similar volume amounts of autograft were used. Articular congruency was confirmed, and the joint capsule and skin were then closed. Postoperative radiographs were made. Sterile compressive bandages were applied to the involved limbs, and the animals were allowed to recover from the anesthesia. Analgesics were administered for five days postoperatively. Each animal was randomly assigned to one of the following study intervals: twenty-four hours (five animals), three weeks (five), six weeks (five), six months (five), one year (three), and eighteen months (three). A previous study
22 suggested that calcium phosphate cement would be replaced with new bone within one year; therefore, the smaller group size was chosen for the last two groups to allow the progression of any degenerative joint changes that may have occurred to be followed and to provide additional information on the long-term stiffness of augmented tibial plateau fractures in this model after several bone-remodeling cycles.
Clinical Course
The animals were allowed unrestricted exercise and were monitored for any changes in limb usage or gait during the study. Postoperative knee-joint effusion was determined by a single investigator who took circumferential measurements of the knee daily for two weeks and once weekly thereafter. The passive range of motion of the knee was measured weekly with use of a goniometer. Tibial radiographs were made weekly for six months and then once per month for the remainder of the study. To assess new-bone formation activity, fluorochrome bone labels were administered intravenously at the time of surgery and at two weeks and twenty-four hours before the animals were killed with an overdose of a barbiturate.
Necropsy
The knee joints were disarticulated, photographed, and assigned a score, ranging from 0 (normal findings) to 5 (severe changes), for gross evidence of degenerative changes, which included pannus over the articular cartilage, cartilage thinning or erosion, condylar flattening, osteophytes, and degree of subsidence of the fracture fragment. The tibiae were then stripped of soft tissues, wrapped in saline-solution-soaked gauze, and prepared for immediate mechanical testing.
Mechanical Testing
Mechanical testing was performed at room temperature. The tibial diaphysis of each specimen was cut to a standardized length of 11 cm and was potted in aluminum channels with use of resin. The embedded specimens were rigidly positioned within an adjustable clamp that was fastened to a servohydraulic testing system (Bionix 858; MTS, Minneapolis, Minnesota). The clamp allowed the tibial articular surface to be aligned perpendicular to the loading indentor. An 8-mm cylindrical indentor was used to apply a compressive load at 0.3 mm/sec directly over the fracture fragment to a maximum load of 130 N. The linear portion of the load displacement curves were analyzed to determine the stiffness values of the fractures at twenty-four hours to eighteen months postoperatively.
Histological Processing and Analysis
After mechanical testing, the lateral tibial plateau was sectioned from the proximal part of the tibia and was fixed in 70% ethanol. The specimens were processed undecalcified and embedded in methylmethacrylate, and two coronal sections, 5 mm apart, were obtained from the middle of the defect near the center of the plateau. From these sections, slides were cut, ground, and polished with use of a cutting-grinding system (EXAKT Apparatebau, Norderstedt, Germany) to a final thickness of 30 to 35 µm. The slides were stained with use of Sanderson rapid bone stain (Surgipath Medical Industries, Richmond, Illinois), preserving the fluorescence labeling. Within the area of the original defect, the percentage of total new-bone volume, fibrous tissue volume, and remaining calcium phosphate cement were measured with use of a custom-designed image-analysis system (Carl Zeis, Thornwood, New York) interfaced with a light-fluorescence microscope at a magnification of 4× and 20×. The articular cartilage and subchondral bone characteristics were assigned a histological score, according to a modification of the system of Mankin et al.
25 .
Fracture Fragment Subsidence
The whole-mount sections from each specimen were placed into an optical comparator (Micro-vu, Windsor, California), and the image was magnified by a factor of 20. At the margins of the articular fracture, the distance from the intact articular surface to the articular surface of the fragment was digitally measured to determine the amount of fragment subsidence. These measurements were made by two investigators without knowledge of the treatment group, and all values for each specimen were averaged.
Statistical Analysis
The joint circumference and knee range of motion were evaluated with use of a paired Student t test. The histological and stiffness measurements were analyzed with use of two-way analysis of variance and Tukey multiple-comparisons methods. A p value of <0.05 was considered significant. The data were expressed as the mean and the standard deviation.
Clinical Course
All animals completed the study. Postoperatively, most animals exhibited decreased range of motion of the knee, which slowly resolved over two weeks. There was no apparent difference between the treatment groups with respect to weight-bearing or gait. The joint circumference measurements were similar among the treatment groups; however, joint effusion persisted longer in several limbs treated with autograft, which were subsequently found at necropsy to have marked fragment subsidence and degenerative changes.
Gross Pathology
At twenty-four hours, four of five specimens from the limbs treated with autograft had marked subsidence of the fracture fragment on gross examination. In contrast, two specimens from the limbs treated with cement showed only minimal subsidence grossly, and the remaining specimens had articular congruency. By six weeks, the specimens from the limbs treated with autograft that had marked fragment step-off had synovial pannus, generalized cartilage thinning with condylar flattening, and moderate degenerative changes. Near the fracture margins, fibrillation of the articular cartilage was observed. The joints that had not sustained substantial fragment subsidence (mostly the tibiae treated with cement) appeared grossly normal, although linear cartilage defects along the margins of the fracture remained visible at all time-periods. At each subsequent time-interval, degenerative changes tended to increase in severity with the degree of fragment step-off and time. With the exception of the specimens retrieved at twenty-four hours, the groups treated with autograft had more extensive gross degenerative changes and higher scores (4 to 5 points) than the groups treated with cement (1 to 2 points) (p < 0.05) that were retrieved at all of the other time-periods. Several specimens from the limbs treated with autograft in the six-month, one-year, and eighteen-month groups that had complete fragment collapse demonstrated partial collapse of the adjacent posterior condyle. The articular cartilage of the femoral condyles opposite the tibiae with moderate to marked subsidence showed cartilage thinning and fibrillation.
Fracture Fragment Subsidence
As observed grossly and in the photomicrographs shown in
Figs. 2-A, 2-B,3-A, 3-B, and 3-C , the prevalence and amount of fracture fragment subsidence was significantly increased in the specimens from the groups treated with autograft compared with the groups treated with cement at all time-periods (p < 0.05) (
Table I ). Time did not influence the amount of subsidence in either group, suggesting that fragment subsidence occurred early in the postoperative period, as shown in the specimens from the tibiae treated with autograft that were retrieved at twenty-four hours (see
Figs. 2-A and 2-B ).
Histological Analysis of Defects
Groups Treated with Autograft
The specimens from the defects treated with autograft that were harvested at twenty-four hours contained a mixture of compressed graft bone and hematoma. All defects showed complete filling with autograft bone (
Fig. 2-B ). By three weeks, fluorochrome labeling indicated active new-bone formation mixed with some graft resorption. This resulted in a defect total bone volume (mean and standard deviation) of 15.0% ± 8%. Numerous osteoclasts were observed within the subchondral bone along the fracture margins. Much of the graft had been remodeled and replaced with new bone by six weeks, and the subchondral fracture-healing appeared complete. At six months, defect bone volume was 35.35% ± 12%, with most autograft bone having been remodeled and replaced with new trabecular bone. Two specimens from the group treated with autograft demonstrated partial graft resorption with residual voids within the defect area (
Fig. 3-B ). At twelve and eighteen months, the new bone occupying the original defect area showed little evidence of remodeling activity and had reached normal bone volume and marrow proportions (mean bone volume, 36.21% ± 8%).
Groups Treated with Cement
At twenty-four hours, the defects appeared to have been completely filled with the cement (94.0% ± 5% of the defect volume) (
Fig. 2-A ). By three weeks, much of the cement had been resorbed around the periphery of the defect, leaving a layer of fibrous tissue separating the cement and the osseous margins (
Fig. 4 ). The defect volume of cement and fibrous tissue was 64.25% ± 12% and 25.5% ± 5%, respectively. The cement surface was covered by numerous multinucleated cells, presumably osteoclasts, apparently resorbing the cement (
Fig. 5 ). The fractured subchondral plate was being remodeled. By six weeks, most of the cement had been resorbed and replaced by fibrous tissue (35.42% ± 8%) mixed with new bone (12.57% ± 4%). Remnants of cement were encased within bone trabeculae, which were undergoing remodeling. The fractures of the subchondral plate were healed. By six months, new bone occupied an increased proportion of the defect area (38.57% ± 14%) with a reduction in the amount of fibrous tissue (6.92% ± 3%) (
Fig. 3-A ). The amount of cement at six months was reduced to 4% ± 1% of the total defect area, with all of the cement embedded within bone trabeculae. The new trabecular bone architecture and orientation was comparable with that of the intact controls. By one year, the original osseous margins of the defect were not discernable, and there were only occasional remnants of cement present. The percentage of new-bone volume (40.51% ± 5%) was slightly increased compared with the controls (32% ± 4%). By eighteen months, the defects contained trabecular bone in normal proportion (35.52% ± 12%) and orientation. One specimen from this group demonstrated a persistent void within the original defect area. No evidence of stress-shielding was observed, and there was no evidence of an inflammatory response at any time.
Histological Analysis of Articular Cartilage
Specimens from the group treated with autograft that were retrieved at twenty-four hours and had marked fracture step-off demonstrated matrix deformation of articular cartilage at the fracture margins (
Fig. 2-B ). In contrast, the specimens with articular congruity or only slight incongruity showed more sharply defined edges with little cartilage matrix deformation. By three weeks, chondrocyte clusters were apparent at the fracture margins in most specimens but were more prevalent in those with moderate to marked step-offs. In the specimens from the tibiae treated with cement that had articular continuity, repair fibrocartilage was present within the cartilage clefts. The repair fibrocartilage showed evidence of integration with original cartilage only at the cartilage-subchondral bone junction; however, nearer the articular surface, small clefts within the articular surface remained throughout the study. Specimens from both treatment groups with minimal-to-moderate fragment step-offs had matrix deformation with incomplete integration of fragment cartilage by six weeks (
Fig. 6 ). No specimen demonstrated reestablishment of the tidemark across the fracture. The staining intensity of the cartilage matrix overlying the fracture fragment was decreased in relation to the degree of subsidence, with most staining loss occurring primarily within the superficial layers. Over time, the joints with large step-offs had a generalized decrease in matrix-staining intensity, with cartilage thinning and focal fibrillation of cartilage adjacent to the lesion. Histological scores indicating degenerative changes were significantly greater (p < 0.05) in the groups treated with autograft (scores of 6 to 8 points) compared with the specimens from the groups treated with cement (scores of 3 to 4 points) at all time-periods. The severity of degenerative changes tended to increase over time in both groups.
Mechanical Testing
The fracture fragment margins were grossly visible and were used to ensure accurate alignment of the 8-mm indentor directly over the fragment. Several specimens from the group treated with autograft that demonstrated complete fragment collapse into the defect were not tested mechanically. The stiffness data from these specimens were not considered to represent healing of the defect and the subchondral plate as the data were not comparable with those from specimens with no or only moderate fragment subsidence. At all of the time-points examined, there was no significant difference in stiffness between the two treatment groups. The stiffness of the treated fractures from both groups was significantly less than that of the intact control tibiae during the first six weeks of healing (p < 0.05). By six months, however, the stiffness values of the fractured plateaus were approaching those of the controls. Two specimens in the group treated with autograft that were retrieved at one year had complete fragment collapse and were not tested mechanically, leaving insufficient data for statistical comparison. At eighteen months, the stiffness values of both treatment groups were comparable with those of the intact controls.
The goal of repair of displaced intra-articular fractures is the restoration of normal joint function
1,2,6 . Any articular step-off has the potential to result in posttraumatic joint degeneration
8,11-14 . Numerous methods have been described for the reduction and stabilization of depressed intra-articular fractures of the tibial plateau
1-5,8,9 . Most techniques use internal fixation through open, limited open, or percutaneous approaches. In many instances, morselized autogenous or allogenic bone graft is packed within voids beneath the reduced fracture fragments to help to maintain anatomic reduction and augment the internal fixation. However, neither autograft nor allograft provides adequate mechanical stability in these instances, and maintenance of anatomical reduction relies primarily on the internal fixation hardware
15-17,26 . In addition, procurement (harvesting) of the autogenous bone graft has related donor-site morbidity, and allograft has the potential for disease transmission and immunological response
27,28 .
Bone-graft substitutes have mechanical and handling characteristics that may be advantageous in the treatment of intra-articular depression fractures. Two studies have demonstrated the clinical utility of the use of preformed porous hydroxyapatite blocks to help to maintain an anatomical reduction of a tibial plateau fracture
26,29 . The blocks require tedious shaping in order to fit within the bone voids. Porous hydroxyapatite is resorbed very slowly if at all and, therefore, tends to interfere with the radiographic interpretation of fracture-healing and potentially may act as a stress-riser
26,30 .
Recently, calcium phosphate cements that undergo isothermic setting have become available, and studies from Europe have documented the clinical efficacy of injectable calcium phosphate cements in fractures of the wrist, tibial plateau, and calcaneus
31-33 . The compressive stiffness of these cements has ranged from 26 to 60 MPa, which is comparable with or exceeds that of cancellous bone. As the cements are moldable, they can be easily injected and can be used to fill irregularly shaped bone defects. A potential drawback to these new biomaterials is the slow osteoclastic resorption and replacement with new host bone. Although the mechanical characteristics of the cement-bone composite remains intact during remodeling, the slow incorporation (four years or more) may interfere with assessment of fracture-healing
17-21 .
Other calcium phosphate cements that are resorbed and replaced with new host bone within weeks or months rather than years have been developed
22-24 . One of these new cement formulations, a-BSM, was evaluated in the current study. This cement has extended working time and remains in a paste-like consistency until it is injected into a bone void and hardens at body temperature. During hardening, a-BSM converts to apatitic calcium phosphate, similar to the mineral phase of bone, within twenty minutes and achieves complete nanocrystalline conversion and mechanical strength within four hours following implantation
19 . The reported mechanical strength of the cement formulation under investigation is 15 MPa
23 . The relative insolubility after hardening would prevent a-BSM from simply dissolving. Rather, the amorphous crystalline structure appears to elicit a resorptive response by osteoclasts.
In the large animal model used in this study, the a-BSM maintained anatomical reduction of the fractures, compared with autogenous bone-packing, by limiting the amount of fragment subsidence. The cement underwent relatively rapid cellular resorption, with most of the cement resorbed within six weeks. Because the rate of bone formation was not at the same rate as cement resorption, fibrous tissue temporarily occupied the space between the new bone and residual cement. However, by six months, the proportion and architecture of new bone that had formed within the defect following cement resorption was similar to control metaphyseal bone and remained within normal values until the completion of the study. The bone-remodeling and fracture-healing rate of the goat model used in the current study is slightly more than double that of humans
34 . This finding suggests that, for a similar location and defect size in human bone, the a-BSM may require eight months or more to be resorbed and replaced by new trabecular bone. In a recent report from Europe, biopsy specimens obtained at second surgeries from patients who had implantation of a calcium phosphate cement similar to that used in the current study had only small residues of cement encased within cancellous bone at twelve months
35 . In the present study, neither autograft nor augmentation with cement restored the acute stiffness of the subchondral plate and articular cartilage to that of the intact controls. However, there were sufficient mechanical properties within the defects augmented with cement to prevent fragment subsidence with immediate postoperative weight-bearing. During the period from six weeks to six months, sufficient restoration of new bone within the defects of both groups restored subchondral stiffness to within the range of the control values. The increase in stiffness corresponded with the increased proportion and maturity of new bone present within the defects.
The prevention of fragment subsidence during early fracture-healing that was seen in the specimens from the group treated with cement reduced the prevalence and severity of degenerative changes compared with that seen in the specimens from the group treated with autograft. The reduced articular cartilage step-off in the defects treated with cement facilitated improved filling and integration of fibrocartilage into cartilage clefts at the fracture margins and maintained more normal appearing cartilage over the fracture fragment itself.
The results of the current study suggest that the calcium phosphate cement a-BSM may serve as a suitable alternative to autologous bone-grafting for filling bone voids associated with displaced fractures of the tibial plateau. The use of this cement may reduce the amount of surgical hardware required to maintain anatomical reduction and perhaps allow earlier joint rehabilitation.
Note: The authors are grateful for the assistance provided by Dr. Nama Doddi.