Journal of Orthopaedic Research 9809419 Raven Press, Ltd., New York 0 1991 Orthopaedic Research Society
Remodelling of Bone Around Intramedullary Stems in Growing Patients Gordon W. Blunn and Mary E. Wait Department of Biomedical Engineering, Institute of Orthopaedics, Brockley Hill, Stanmore, Middlesex, England
Summary: The remodelling of bone around intramedullary cemented stems, used for the fixation of massive distal femoral replacements in young patients, has been studied. Several types of remodelling have been identified, and this is common to all of the retrieved specimens that have been in situ for periods of 6 to 18 months. Transection of cortical bone and the insertion of a massive prosthesis leads to (a) the formation of a pedicle of bone or bony bridge that grows from the transection site over the shaft of the prosthesis, (b) bone resorption at the transection site, (c) the development of an inner porotic cortex surrounding the intramedullary stem and subperiosteal bone formation, and (d) a shell of bone around the acrylic cement. The tibial component that allows continued growth of the proximal tibial physis consists of a press fit metallic plateau with a stem that slides into an ultra-high-molecular-weight polyethylene sleeve inserted below the growth plate. A sclerotic layer of bone forms adjacent to a soft-tissue interface around both the sleeve and the metallic component. Remodelling of bone around intramedullary stems is attributed either to a redeveloping blood supply or to load adaptive changes. Key Words: Bone remodelling-Intramedullary fixation-Interface-ChildrenAdolescents-Prosthesis-Ultra-high-molecular-weight polyethylene.
Limb salvage after tumour resection has been made possible by a combination of early and improved diagnosis of tumours, advances in chemotherapeutic methods (16), and the development of various surgical treatments, such as autogenous bone grafting, allografting, and replacement by custom-made prostheses. Such treatments should meet the objectives of eradicating the tumour, with results as good as those achieved by amputation, and restoring the best possible long-term function of the limb. Massive allograft and autogenous grafts have a high incidence of nonunion, infection, and fracture (21). For prosthetic replacement, although in-
fection can lead to a high early failure rate, secure fixation is the most important requirement for longterm clinical success. As a large number of tumours occur in patients who are still growing, a custommade extending prosthesis has been developed. The design of this prosthesis, which consists of cemented and uncemented components, has been described previously (15). In total joint replacement where cement is used, a major complication has been found to be loosening of the cement at the bone interface-a problem that is particularly evident in active children and adolescents (5,7). An additional problem associated with intramedullary cement fixation in these patients is that the femur and the medullary cavity increase in girth as the patient grows. Therefore, it is important to consider the success of intramedullary cement fixation of massive prostheses in growing patients. This article
Received November 3, 1989; accepted September 5 , 1990. Address correspondence and reprint requests to Dr. G . W. Blunn at Department of Biomedical Engineering, Institute of Orthopaedics, Brockley Hill, Stanmore, Middlesex, HA7 4LP, England.
809
G . W . BLUNN AND M . E . WAIT
810
reports on the remodelling of bone and on the bone/ prosthesis interface around cemented and uncemented components, as observed in retrieved growing distal femoral prostheses. Fixation of cemented femoral components used in total hip replacement relies on secure anchorage of the stem in the intramedullary cavity. The results presented in this article may equally apply to fixation of these stems, particularly in children and adolescents. MATERIALS AND METHODS
Prostheses Of the seven retrieved limbs, three had a MkII Stanmore growing distal femoral prosthesis. The other four had a nonextending femoral component (Table 1). All femoral components were cemented in place using either Simplex P or C polymethylmethacrylate bone cement. Titanium (Ti318) alloy was used for the construction of the stem and shaft. This shaft was attached to a cobalt, chrome, and molybdenum alloy knee joint. Because these prostheses were custom-made, the stems and shafts of the femoral components were of varying lengths. In one case a large proportion of the distal femur was resected, and in order to gain sufficient anchorage in the remaining proximal bone a dual divergent stem was used (Case 3). All of the cases except one (Case 4) had passively sliding tibial components consisting of threaded ultra-high-molecular-weight polyethylene sleeves (UHMWPE) and metallic components. The sleeves
were either 15 mm or 13 mm wide. For insertion, a sleeve was screwed into a hole tapped in the tibial intramedullary cavity to a position below the epiphyseal plate. The tibia was resected and grooved so that flanges on the metallic plateau located in the grooves and prevented rotation. The metallic component was able to slide in the polyethylene sleeve as the tibia grew (Fig. 1). Prostheses were clinically stable at the time of retrieval and became available either after amputation of the limb due to recurrence of the tumour (one case) or upon death of the patient due to secondary metastases (six cases). All of the patients had been ambulatory and weight-bearing without any aids, but due to deterioration of their general condition prior to death or to excessive pain in the limb with the prosthesis (one case), they were not weight-bearing immediately before retrieval. All patients were placed on a course of chemotherapy (methatrexate-folic acid) before and after surgery.
FEMUR
TELESCOPIC SHAFT
TABLE 1. Patient details Case
Sex
Mos.
Diagnosis
1"
M
6
osteosarcoma
2
M
I
osteosarcoma
36
F
9
osteosarcoma
4d
F
14
osteosarcoma
5"
M
15
osteosarcoma
6'
M
15
osteosarcoma
7=
M
18
osteogenic sarcoma
__
a
Retrieval of treated and untreated limb. Dual divergent stem. Nonextending femoral component. Nonextending femoral and tibial components.
J Orthop Res. Vol. 9, No. 6 , 1991
Reasons for removal lung/chest metastases (deceased) local recurrence (amputation) pulmonary emboli (deceased) chest metastases (deceased) chest metastases (deceased) chest metastases (deceased) multi. metastases (deceased)
TIBIAL COMPONENT
POLYETHYLENE SLEEVE
TIBIA
FIG. 1. Diagrammatic representation of a Stanmore growing distal femoral replacement.
REMODELLING BONE AROUND INTRAMEDULLAR Y STEMS
811
Preparation of Histological Sections Retrieved limbs were fixed in formal saline solution; the soft tissues were removed; and, in order to aid Orientation, small grooves running in a proximal-distal direction were made on the anterior and medial aspects of the bones using a slitting wheel. Bones were radiographed and photographed before being cut into 5-mm-thick serial slices using an Exotom cutoff machine. Slices were then contactradiographed and photographed, and selected slices were processed for histology. Undecalcified sections were prepared. For decalcified histology, selected slices were incubated in saturated EDTA solution at pH 7. Slices were then dehydrated in alcohol and double embedded in celloidin and paraffin wax. Sections of 5 pm were cut and the wax removed and stained in haematoxylin and eosin. RESULTS
Remodelling of Bone Around Cemented Femoral Components Radiographs of the retrieved femora showed that in every case bone had resorbed at the transection site adjacent to plateau of the prosthetic shaft; this resorption was apparent after 6 months. With the exception of Case 1, all of the retrieved specimens had a pedicle of bone extending distally over the medial-proximal aspects of the shafts of the prostheses. Bone was not directly attached to the metallic shaft, as evidenced by a radiolucent gap between the shaft of the prosthesis and the bone. The pedicle was attached to the shaft of the femur in an area above the transection site (Fig. 2). Contact radiographs of serial slices taken through the femur of Case 1 showed considerable remodelling of bone around the stem. This remodelling occurred after 6 months. The original cortex in the distal slices had become porotic on the anteriorlateral side. Concentric rings of new bone were laid down by the periosteum. Comparison with slices taken at the same level on the retrieved, untreated femur (Case 1) indicated that this subperiosteal reaction resulted in an increase in femoral circumference (Figs. 3, 4). The increase became less evident in sections towards the stem tip. The subperiosteal reaction coupled with cortical porosis was more evident in the longer-term cases. Examination of sec-
FIG. 2. Medial-lateral and anterior-posterior radiographs of retrieved femur (Case 6 ) showing pedicle of bone growing over the posterior aspect of the shaft of the prosthesis (P). Note the resorption of bone at the transection site.
tions taken from the femur at the midstem level of Case 5 showed remodelling of the cortex to form a new marrow cavity on the medial side of the femur. This cavity was strutted by trabeculae, which were in contact with the outer cortex and with a shell of bone that surrounded the acrylic cement (Fig. 5 ) . This radiograph also illustrates the formation of “picket fence” bone on the outer posterior cortical surface. Newly formed bone was continuous with the bony pedicle that extended over the shaft of the prosthesis. In Case 3 a bony shell also extended around the acrylic cement that anchored the dual divergent stem into the proximal femur (Fig. 6 ) . Slight cortical remodelling was evident in the diaphysis, and a predominant shell of bone was adjacent to the acrylic cement in the more proximal metaphyseal slices.
J Orthop Res, Vol. 9, N o . 6 , 1991
G . W . BLUNN AND M . E . WAIT
812
FIG. 3. Radiographs of slices taken at the same level from the femur with the prostheses (left) and from the untreated limb (right).
This shell of bone developed only around the cement in cancellous regions and was joined to the cortex by struts of trabecular bone (Fig. 7). In this case the proximal femur was prepared by handreaming, and the acrylic cement was finger-packed. This method appeared to have resulted in poor penetration of the cement into cancellous bone, and
AREA hdl
700
although the patient did not show symptoms of either clinical or radiographic loosening, there appeared to be a radiolucent line between the shell of bone and the acrylic cement (Fig. 7). Radiographic Appearance of Retrieved Tibiae A radiographic study of the retrieved tibiae showed that a region of sclerotic bone usually de-
-
600-
500-
5
0
10
DISTANCE FROM PLATEAU (cml FIG. 4. Histograph showing the increase in area of the slices taken from the femur with the intramedullary stem compared with the untreated limb from Case 1.
J Orthop Res, Vol. 9,
NO. 6,1991
FIG. 5. Slices taken from midstem level of Case 5, showing shell of bone (S) surrounding acrylic cement with remodelled cortex (C), extent of femur before insertion of the intramedullary stem (arrows), and subperiosteal bone formation (B).
REMODELLING BONE AROUND INTRAMEDULLAR Y STEMS
813
FIG. 6. Radiograph of retrieved prosthesis and associated bone (Case 3), showing the divergent stem and the shell of bone (S) that surrounds the cement in the cancellous region.
veloped around the tapped polyethylene sleeves. This sclerotic bone was in apposition to and contoured the polyethylene sleeve. The bone was apparent in the case that was retrieved after 6 months (Case '); Only Of the polyethylene sleeve were contoured by sclerotic bone. EXamination of tibiae retrieved after 15 months
FIG. 8. Radiograph of retrieved tibia (Case 8) showing shell of sclerotic bone that has developed around the screwthreaded polyethylene sleeve. Sclerotic bone is also found under the tibia1 Plateau (arrow).
showed that the polyethylene sleeve was completely invested with bone (Fig. 8). After 6 months sclerotic bone had also developed under the tibia1 plateau. However, there appeared to be a small radiolucent space between the metal and the bone. Slices taken through the cancellous region of the tibia of Case 5 showed that the sclerotic bone surrounding the polyethylene sleeve was honeycombed (Fig. 9). This sclerotic bone was joined to the outer cortical bone by a network of branched trabeculae. There appeared to be no radiolucent space between the sclerotic bone and the sleeve. More distally the bone surroundingthe polyethylene sleeve appeared to be denser, and struts of trabeculaejoined the sclerotic bone with the cortex (Fig. 10). Histology of the Bone/Implant Interface FIG. 7. Radiograph of a slice (Case 3) showing the shell of bone strutted to the cortex. A radiolucent space is evident between the bone and the cement (arrows).
The pedicle of bone that grew from the shaft of the femur in an area adjacent to the transection site
J Orthop Res, Vol. 9, No. 6, 1991
814
G . W . BLUNN AND M . E. WAIT
FIG. 10. Radiograph of a slice distal to that in Fig. 9 showing denser bone around the sleeve.
FIG. 9. Radiograph of a slice through the tibia showing honeycomb appearance of bone around the sleeve, which is joined to the cortex by a network of branched trabeculae.
increased in size by direct ossification. The bone was separated from the metallic shaft by a layer of soft tissue measuring -30 pm thick composed predominantly of fibroblasts and well-oriented collagen fibres running parallel to the shaft of the prosthesis. In the region close to the plateau, this soft-tissue layer became more vascular and was continuous between metal and bone at the plateau (Fig. 11). Dead bone characterised by empty lacunae remodelled in areas adjacent to the vascular tissue at the transection site, and blood vessels migrated into dead cortical bone from this area. Histological sections of the acrylic cement interface in the diaphyseal region showed that bone surrounding cement did not form a complete shell and that a soft-tissue layer occurred between the bone and the cement (Fig. 12). Sections taken through the plateau interface of the tibia in Case 2 showed a layer of soft tissue adjacent to the titanium alloy. This layer was composed of fibroblasts and highly oriented collagen
J Orthop Res, Vol. 9, No. 6 , 1991
FIG. 11. Histological section taken through the transection site (Case 2) showing vascular soft tissue (S) with numerous blood vessels (V) between the bone and the prosthesis. Dead bone (D) below the transection site is being remodelled in areas adjacent to the vascular tissue and around blood vessels that migrate into dead bone. Haematoxylin and eosin, x50.
REMODELLING BONE AROUND INTRAMEDULLAR Y STEMS
815
FIG. 12. Section through the shell of bone surrounding acrylic cement in the diaphyseal femur. The bone surrounding the cement does not form a complete shell, and a soft-tissue layer (arrow) occurs between the bone and cement (C). Also note the extensive porotic inner cortex filled with marrow (M)that has developed in areas where cortical bone has resorbed. Haematoxylin and eosin, x 19.5.
fibres. An incomplete layer of bone occurred adjacent to this tissue: trabeculae were strutted from this layer and ran predominantly in a vertical direction. These trabeculae were usually thicker at the base of the antirotation groove (Figs. 13, 14). With
time the layer of bone under the plateau became complete and thicker (Fig. 15). In addition, a fibrocartilage interface was established. Fibrocartilage first appeared within the antirotation groove in specimens. Sections taken through the sleeve inter-
FIG. 13. Histological section through the tibial plateau on the medial-lateral plane showing soft tissue (S) adjacent to the prosthesis, the antirotation groove (A), the position of the metallic tibial stem (T), and remodelled bone (B) beneath the soft tissue. Haematoxylin and eosin, x4.5.
J Orthop Res, Vol. 9, N o . 6 , 1991
G . W . BLUNN AND M . E. WAIT
816
FIG. 14. Light micrograph of the interface around the antirotation groove (Case 1, duration 6 months) showing the position of the metallic antirotation lug (A) and the soft tissue ( S ) that occurs between the metal and newly remodelled bone. Sclerotic trabeculae are seen at the base of the antirotation groove. Haematoxylin and eosin, x 13.5.
face from various specimens showed that a softtissue layer was always present between the polyethylene and the shell of bone (Fig. 16). This layer measured up to 500 Frn thick. Birefringent polyethylene wear particles were not evident in this layer, which was composed predominantly of fibroblasts. DISCUSSION Massive prostheses usually rely on intramedullary fixation using acrylic cement, although bone ingrowth into porous surfaces is used in a number of designs (10). In these noncemented prostheses, resistance to torsional forces is achieved by extracortical fixation using plates and screws. Designs by Chao and Sim (6) use porous coatings over the shafts of prostheses. Pedicle formation and bone ingrowth are encouraged by bone grafts that result in enhanced torsional resistance. The present study demonstrates that the pedicle forms in the majority of retrieved cases but that connective tissue inter-
J Orthop Res. Vol. 9, No. 6, 1991
venes between the bone and the smooth metal shaft. The pedicle forms on the medial-posterior aspect of the femur in young patients, as demonstrated by our research, and in older patients with distal femoral replacements (2) and proximal tibia1 replacements (14). This aspect of the femur has been demonstrated in laboratory static loading tests to be the side that is under compressive load (22), and pedicle formation may therefore be a direct result of load-adaptive bone formation. In all cases in this study the bone at the transection site had resorbed, resulting in bone below the shoulder of the prosthesis becoming less dense because of stress protection. Burrows and colleagues (4) first reported resorption of bone due to nonphysiological loading conditions in a femur with a massive Stanmore prosthesis. Initial resorption at the transection site is accompanied by the development of a highly vascular soft-tissue interface between the shoulder of the prosthesis and resorbed bone. Blood vessels from this tissue invade necrotic cortical bone adjacent to the transection site. During surgery the blood supply to diaphyseal cortical bone is destroyed when the intramedullary cavity is reamed and packed with cement (12,19). Cortical bone adjacent to the shoulder of the prosthesis remodels and revascularises with the invasion of blood vessels from this vascular soft tissue. Rhinelander (1 1) observed diaphyseal cortical remodelling when a tight-fitting intramedullary nail blocked the reentry of the nutrient artery into the medulla of canine femora. Remodelling in the posterior femoral cortex was attributed to the path of longitudinally directed branches of the nutrient artery. Remodelling of the diaphyseal cortex in patients with growing prostheses may be a consequence of the redevelopment of the blood supply. Subperiosteal bone is laid down in areas adjacent to the remodelled cortex and is probably a result of stressadaptive bone formation. A shell of bone surrounding acrylic cement, particularly in areas where there is poor penetration into cancellous bone, was evident in this investigation. A shell of bone surrounding acrylic cement has been identified around the stems of retrieved Charnley total hip replacements (9). In addition, a shell of bone has been found around the distal nonosseointegrated stems of cementless canine total hip replacements (1). This shell of bone has been attributed by others working on sheep models to be typical of a loose implant (13). In the sheep model,
REMODELLING BONE AROUND INTRAMEDULLAR Y STEMS
817
FIG. 15. Histological section through the interface of the tibia1 plateau (Case 6 , duration 15 months) showing a complete, thick, irregular layer of lamellar bone below the soft-tissue interface. Haematoxylin and eosin, x25.
concentric layers of fibrous tissue adjacent to the cement were surrounded by a shell of dense bone, particularly where interdigitation with cancellous bone was poor. Using finite element analysis, for-
mation of this shell was attributed to a response to the hoop stresses generated during pistoning of the implant (3). Slight pistoning of the femoral component may occur around these massive prostheses,
FIG. 16. Histological section at the interface of the sleeve showing the shell of remodelled bone (B) and the soft tissue (S) between the polyethylene and the bone. Haematoxylin and eosin, X180.
J Orthop Res, Vol. 9, No. 6 , 1991
818
G . W . BLUNN AND M . E. WAIT
as the cement is surrounded by a layer of soft tissue and the plateau is not in contact with bone. Load must be transferred through a layer of vascular reactive tissue found between the plateau and the bone of the passively sliding tibial component, because the metallic stem does not bear on the end of the polyethylene sleeve. This interface is similar to those that occur under porous coated tibial trays of cementless condylar knee replacements, where ingrowth of bone is infrequently observed and a connective tissue layer is found (18). Although a shell of bone does contour the screw thread, the sleeve is not osseointegrated, and a layer of connective tissue is found adjacent to the polyethylene. Polyethylene and polysulfone have been used as porous coatings on the surface of femoral components of total hip prostheses (8). Animal studies have revealed that bone grows into and is maintained in porous polyethylene and porous polysulphone in exactly the same fashion as in porous metals (17). The reason that bone is not in direct contact with the polyethylene, as in porous surfaces, may be due to the surface topography of the threaded sleeves. The effect of the vascular reactive tissue on load transfer is speculative. In laboratory tests using embalmed femora with a silicone rubber membrane, Wright and associates (22) demonstrated that some load was transferred across this membrane at the plateau. However, the magnitude of longitudinal surface stresses induced in the intramedullary stem was greatly increased. Using a finite element model, Weinans and co-workers (20) showed that load transfer would be concentrated in particular spots when a soft-tissue membrane is present between the cement and bone in total hip arthroplasty. If such load transfer occurs around stems of major prostheses, the mechanical properties of the bonecement interface may well be exceeded. Therefore, the maintenance of the plateau-bone contact and the transfer of load across the transection site would be an ideal situation. Clearly these conditions are not achieved with current designs, and therefore the incidence of revision operations for aseptic loosening from failure of the cement-bone interface can be expected to increase in the future. Acknowledgment: The authors acknowledge the help and advice given by Professor J. T. Scales and Professor P. S. Walker. They are also indebted to the following surgeons: Mr. H. B. S. Kemp, FRCS, Mr. R. S. Sneath, FRCS, Mr. D. R. Sweetnam, FRCS, Mr. R. J. Grimer, FRCS, and Mr. S.R. Cannon, FRCS, who made the retrieval of specimens possible.
J Orthop Res. Vol. 9, No. 6, 1991
REFERENCES 1. Bobyn JD, Pillar RM, Binnington AG, Szivek JA: The effect Df proximally and fully porous coated canine hip stem design 3n bone remodelling. J Orthop Res 5:393-408, 1987 2. Bradish CF, Kemp HBS, Scales JT, Wilson JN: Distal femoral replacement by custom-made prostheses. Clinical follow up and survivorship analysis. J Bone Joint Surg [Br] 69B:27&284, 1987 3. Brown TD, Petersen DR, Radin EL, Rose RM: Global mechanical consequences of reduced cementhone coupling rigidity in proximal femoral arthroplasty: A three-dimensional finite element analysis. J Biomech 21:115-129, 1988 4. Burrows HJ, Wilson JN, Scales JT: Excision of tumours of humerus and femur with restoration by internal prostheses. J Bone Joint Surg [Br] 57-B:148-159, 1915 5 . Chandler HP, Reineck FT, Wilson RL, McCarthy JC: Total hip replacements in patients younger than 30 years old. J Bone Joint Surg [Am] 63A:1426-1434, 1981 6. Chao EYS, Sim FH: Modular types of tumour endoprostheses for limb salvage. In: Enneking WF, ed. Limb Salvage in Musculoskeletal Oncology. New York: Churchill Livingstone, 1987:227-233 7. Dorr LD, Takei GK, Conaty JP: Total hip arthroplasties in patients less than forty-five years old. J Bone Joint Surg [Am] 65A:474-479, 1983 8. Klawitter JT, Bagwell JD, Weinstein AM, Saur BW, Pruitt JR: An evaluation of bone growth into porous high density polyethylene. J Biomed Mat Res 10:311-323, 1976 9. Malcolm AJ: Pathology of cement low friction arthroplasty in autopsy specimens. In: Older J, ed. Implant Bone Znterface. Berlin, Heidelberg, New York: Springer-Verlag, 1990:77-82 10. Plenk H Jr, Salzer M, Locke H: Extracortical attachment to bioceramic endoprostheses to long bones without bone cement. Clin Orthop 132:252-265, 1978 11. Rhinelander FW: Circulation in bone. In: Bourne GH, ed. Biochemistry and Physiology of Bone. New York, London: Academic Press, 1972 12. Rhinelander FW, Nelson CL, Stewart RD, Stewart CL: Experimental reaming of the proximal femur and acrylic cement implantation-vascular and histologic effects. Clin Orthop 141:75439, 1979 13. Rose RM, Martin RB, Orr RB, Radin EL: Architectural changes in the proximal femur following prosthetic insertion: Preliminary observations of an animal model. J Biomech 17:241-250, 1984 14. Scales JT, Wait ME: Intramedullary fixation of custommade prostheses with bone cement. In: Coombs R, Fnedlaender, G, ed. Bone Tumour Management. London: Buttenvorth & Co., Ltd., 1987:123-134 15. Scales JT, Sneath RS, Wright KWJ: Design and clinical use of extending prostheses. In: Enneking WF, ed. Limb Salvage in Musculoskeletal Oncology. New York: Churchill Livingstone, 198752-6 1 16. Simon M: Current concepts review: Causes of increased survival of patients with osteosarcoma: current controversies. J Bone Joint Surg [Am] 66A:30&310, 1984 17. Spector M: Bone ingrowth into porous polymers. In: Williams DF, ed. Biocompatibilify of Orthopaedic Implants. Boca Raton: CRC Press, 1982:55-88 18. Thomas KA, Cook SD, Thomas KL, Haddad JR: Tissue growth into retrieved noncemented human hip and knee components. In: Saha S, ed. Biomedical Engineering V. Recent Developments. New York: Pergamon Press, 1986:198203
REMODELLING BONE AROUND INTRAMEDULLAR Y STEMS 19. de Wall Malefijt J, Slooff TJJ, Huiskes R, de Laat EAT, Barentez JO: Vascular changes following total hip arthroplasty. The femur in goats studied with and without cementation. Acta Orthop Scand 59543-649, 1988 20. Weinans H, Huiskes R, Grootenboer H: The mechanical effects of fibrous tissue interposition at the cement bone interface in THA. In: Transactions of the 34th Annual Meeting of the Orthopaedic Research Society, 1988;13:502
819
21. Wilson PD Jr, Lance EM: Surgical reconstruction of the skeleton following segmental resection for bone tumours. J Bone Joint Surg 47A:1629-1656, 1965 22. Wright KW, Meswania JW, Scales JT: Load transmission characteristics of intramedullary stedpolymethyl methacrylate fixation in endoprostheses after tumour resection. In: Enneking WE, ed. Limb Salvage in Musculoskeletal Oncolo g y . New York: Churchill Livingstone, 1987
J Orthop Res, Val. 9,No. 6, 1991
Remodelling of Bone Around Intramedullary Stems in Growing Patients Gordon W. Blunn and Mary E. Wait Department of Biomedical Engineering, Institute of Orthopaedics, Brockley Hill, Stanmore, Middlesex, England
Summary: The remodelling of bone around intramedullary cemented stems, used for the fixation of massive distal femoral replacements in young patients, has been studied. Several types of remodelling have been identified, and this is common to all of the retrieved specimens that have been in situ for periods of 6 to 18 months. Transection of cortical bone and the insertion of a massive prosthesis leads to (a) the formation of a pedicle of bone or bony bridge that grows from the transection site over the shaft of the prosthesis, (b) bone resorption at the transection site, (c) the development of an inner porotic cortex surrounding the intramedullary stem and subperiosteal bone formation, and (d) a shell of bone around the acrylic cement. The tibial component that allows continued growth of the proximal tibial physis consists of a press fit metallic plateau with a stem that slides into an ultra-high-molecular-weight polyethylene sleeve inserted below the growth plate. A sclerotic layer of bone forms adjacent to a soft-tissue interface around both the sleeve and the metallic component. Remodelling of bone around intramedullary stems is attributed either to a redeveloping blood supply or to load adaptive changes. Key Words: Bone remodelling-Intramedullary fixation-Interface-ChildrenAdolescents-Prosthesis-Ultra-high-molecular-weight polyethylene.
Limb salvage after tumour resection has been made possible by a combination of early and improved diagnosis of tumours, advances in chemotherapeutic methods (16), and the development of various surgical treatments, such as autogenous bone grafting, allografting, and replacement by custom-made prostheses. Such treatments should meet the objectives of eradicating the tumour, with results as good as those achieved by amputation, and restoring the best possible long-term function of the limb. Massive allograft and autogenous grafts have a high incidence of nonunion, infection, and fracture (21). For prosthetic replacement, although in-
fection can lead to a high early failure rate, secure fixation is the most important requirement for longterm clinical success. As a large number of tumours occur in patients who are still growing, a custommade extending prosthesis has been developed. The design of this prosthesis, which consists of cemented and uncemented components, has been described previously (15). In total joint replacement where cement is used, a major complication has been found to be loosening of the cement at the bone interface-a problem that is particularly evident in active children and adolescents (5,7). An additional problem associated with intramedullary cement fixation in these patients is that the femur and the medullary cavity increase in girth as the patient grows. Therefore, it is important to consider the success of intramedullary cement fixation of massive prostheses in growing patients. This article
Received November 3, 1989; accepted September 5 , 1990. Address correspondence and reprint requests to Dr. G . W. Blunn at Department of Biomedical Engineering, Institute of Orthopaedics, Brockley Hill, Stanmore, Middlesex, HA7 4LP, England.
809
G . W . BLUNN AND M . E . WAIT
810
reports on the remodelling of bone and on the bone/ prosthesis interface around cemented and uncemented components, as observed in retrieved growing distal femoral prostheses. Fixation of cemented femoral components used in total hip replacement relies on secure anchorage of the stem in the intramedullary cavity. The results presented in this article may equally apply to fixation of these stems, particularly in children and adolescents. MATERIALS AND METHODS
Prostheses Of the seven retrieved limbs, three had a MkII Stanmore growing distal femoral prosthesis. The other four had a nonextending femoral component (Table 1). All femoral components were cemented in place using either Simplex P or C polymethylmethacrylate bone cement. Titanium (Ti318) alloy was used for the construction of the stem and shaft. This shaft was attached to a cobalt, chrome, and molybdenum alloy knee joint. Because these prostheses were custom-made, the stems and shafts of the femoral components were of varying lengths. In one case a large proportion of the distal femur was resected, and in order to gain sufficient anchorage in the remaining proximal bone a dual divergent stem was used (Case 3). All of the cases except one (Case 4) had passively sliding tibial components consisting of threaded ultra-high-molecular-weight polyethylene sleeves (UHMWPE) and metallic components. The sleeves
were either 15 mm or 13 mm wide. For insertion, a sleeve was screwed into a hole tapped in the tibial intramedullary cavity to a position below the epiphyseal plate. The tibia was resected and grooved so that flanges on the metallic plateau located in the grooves and prevented rotation. The metallic component was able to slide in the polyethylene sleeve as the tibia grew (Fig. 1). Prostheses were clinically stable at the time of retrieval and became available either after amputation of the limb due to recurrence of the tumour (one case) or upon death of the patient due to secondary metastases (six cases). All of the patients had been ambulatory and weight-bearing without any aids, but due to deterioration of their general condition prior to death or to excessive pain in the limb with the prosthesis (one case), they were not weight-bearing immediately before retrieval. All patients were placed on a course of chemotherapy (methatrexate-folic acid) before and after surgery.
FEMUR
TELESCOPIC SHAFT
TABLE 1. Patient details Case
Sex
Mos.
Diagnosis
1"
M
6
osteosarcoma
2
M
I
osteosarcoma
36
F
9
osteosarcoma
4d
F
14
osteosarcoma
5"
M
15
osteosarcoma
6'
M
15
osteosarcoma
7=
M
18
osteogenic sarcoma
__
a
Retrieval of treated and untreated limb. Dual divergent stem. Nonextending femoral component. Nonextending femoral and tibial components.
J Orthop Res. Vol. 9, No. 6 , 1991
Reasons for removal lung/chest metastases (deceased) local recurrence (amputation) pulmonary emboli (deceased) chest metastases (deceased) chest metastases (deceased) chest metastases (deceased) multi. metastases (deceased)
TIBIAL COMPONENT
POLYETHYLENE SLEEVE
TIBIA
FIG. 1. Diagrammatic representation of a Stanmore growing distal femoral replacement.
REMODELLING BONE AROUND INTRAMEDULLAR Y STEMS
811
Preparation of Histological Sections Retrieved limbs were fixed in formal saline solution; the soft tissues were removed; and, in order to aid Orientation, small grooves running in a proximal-distal direction were made on the anterior and medial aspects of the bones using a slitting wheel. Bones were radiographed and photographed before being cut into 5-mm-thick serial slices using an Exotom cutoff machine. Slices were then contactradiographed and photographed, and selected slices were processed for histology. Undecalcified sections were prepared. For decalcified histology, selected slices were incubated in saturated EDTA solution at pH 7. Slices were then dehydrated in alcohol and double embedded in celloidin and paraffin wax. Sections of 5 pm were cut and the wax removed and stained in haematoxylin and eosin. RESULTS
Remodelling of Bone Around Cemented Femoral Components Radiographs of the retrieved femora showed that in every case bone had resorbed at the transection site adjacent to plateau of the prosthetic shaft; this resorption was apparent after 6 months. With the exception of Case 1, all of the retrieved specimens had a pedicle of bone extending distally over the medial-proximal aspects of the shafts of the prostheses. Bone was not directly attached to the metallic shaft, as evidenced by a radiolucent gap between the shaft of the prosthesis and the bone. The pedicle was attached to the shaft of the femur in an area above the transection site (Fig. 2). Contact radiographs of serial slices taken through the femur of Case 1 showed considerable remodelling of bone around the stem. This remodelling occurred after 6 months. The original cortex in the distal slices had become porotic on the anteriorlateral side. Concentric rings of new bone were laid down by the periosteum. Comparison with slices taken at the same level on the retrieved, untreated femur (Case 1) indicated that this subperiosteal reaction resulted in an increase in femoral circumference (Figs. 3, 4). The increase became less evident in sections towards the stem tip. The subperiosteal reaction coupled with cortical porosis was more evident in the longer-term cases. Examination of sec-
FIG. 2. Medial-lateral and anterior-posterior radiographs of retrieved femur (Case 6 ) showing pedicle of bone growing over the posterior aspect of the shaft of the prosthesis (P). Note the resorption of bone at the transection site.
tions taken from the femur at the midstem level of Case 5 showed remodelling of the cortex to form a new marrow cavity on the medial side of the femur. This cavity was strutted by trabeculae, which were in contact with the outer cortex and with a shell of bone that surrounded the acrylic cement (Fig. 5 ) . This radiograph also illustrates the formation of “picket fence” bone on the outer posterior cortical surface. Newly formed bone was continuous with the bony pedicle that extended over the shaft of the prosthesis. In Case 3 a bony shell also extended around the acrylic cement that anchored the dual divergent stem into the proximal femur (Fig. 6 ) . Slight cortical remodelling was evident in the diaphysis, and a predominant shell of bone was adjacent to the acrylic cement in the more proximal metaphyseal slices.
J Orthop Res, Vol. 9, N o . 6 , 1991
G . W . BLUNN AND M . E . WAIT
812
FIG. 3. Radiographs of slices taken at the same level from the femur with the prostheses (left) and from the untreated limb (right).
This shell of bone developed only around the cement in cancellous regions and was joined to the cortex by struts of trabecular bone (Fig. 7). In this case the proximal femur was prepared by handreaming, and the acrylic cement was finger-packed. This method appeared to have resulted in poor penetration of the cement into cancellous bone, and
AREA hdl
700
although the patient did not show symptoms of either clinical or radiographic loosening, there appeared to be a radiolucent line between the shell of bone and the acrylic cement (Fig. 7). Radiographic Appearance of Retrieved Tibiae A radiographic study of the retrieved tibiae showed that a region of sclerotic bone usually de-
-
600-
500-
5
0
10
DISTANCE FROM PLATEAU (cml FIG. 4. Histograph showing the increase in area of the slices taken from the femur with the intramedullary stem compared with the untreated limb from Case 1.
J Orthop Res, Vol. 9,
NO. 6,1991
FIG. 5. Slices taken from midstem level of Case 5, showing shell of bone (S) surrounding acrylic cement with remodelled cortex (C), extent of femur before insertion of the intramedullary stem (arrows), and subperiosteal bone formation (B).
REMODELLING BONE AROUND INTRAMEDULLAR Y STEMS
813
FIG. 6. Radiograph of retrieved prosthesis and associated bone (Case 3), showing the divergent stem and the shell of bone (S) that surrounds the cement in the cancellous region.
veloped around the tapped polyethylene sleeves. This sclerotic bone was in apposition to and contoured the polyethylene sleeve. The bone was apparent in the case that was retrieved after 6 months (Case '); Only Of the polyethylene sleeve were contoured by sclerotic bone. EXamination of tibiae retrieved after 15 months
FIG. 8. Radiograph of retrieved tibia (Case 8) showing shell of sclerotic bone that has developed around the screwthreaded polyethylene sleeve. Sclerotic bone is also found under the tibia1 Plateau (arrow).
showed that the polyethylene sleeve was completely invested with bone (Fig. 8). After 6 months sclerotic bone had also developed under the tibia1 plateau. However, there appeared to be a small radiolucent space between the metal and the bone. Slices taken through the cancellous region of the tibia of Case 5 showed that the sclerotic bone surrounding the polyethylene sleeve was honeycombed (Fig. 9). This sclerotic bone was joined to the outer cortical bone by a network of branched trabeculae. There appeared to be no radiolucent space between the sclerotic bone and the sleeve. More distally the bone surroundingthe polyethylene sleeve appeared to be denser, and struts of trabeculaejoined the sclerotic bone with the cortex (Fig. 10). Histology of the Bone/Implant Interface FIG. 7. Radiograph of a slice (Case 3) showing the shell of bone strutted to the cortex. A radiolucent space is evident between the bone and the cement (arrows).
The pedicle of bone that grew from the shaft of the femur in an area adjacent to the transection site
J Orthop Res, Vol. 9, No. 6, 1991
814
G . W . BLUNN AND M . E. WAIT
FIG. 10. Radiograph of a slice distal to that in Fig. 9 showing denser bone around the sleeve.
FIG. 9. Radiograph of a slice through the tibia showing honeycomb appearance of bone around the sleeve, which is joined to the cortex by a network of branched trabeculae.
increased in size by direct ossification. The bone was separated from the metallic shaft by a layer of soft tissue measuring -30 pm thick composed predominantly of fibroblasts and well-oriented collagen fibres running parallel to the shaft of the prosthesis. In the region close to the plateau, this soft-tissue layer became more vascular and was continuous between metal and bone at the plateau (Fig. 11). Dead bone characterised by empty lacunae remodelled in areas adjacent to the vascular tissue at the transection site, and blood vessels migrated into dead cortical bone from this area. Histological sections of the acrylic cement interface in the diaphyseal region showed that bone surrounding cement did not form a complete shell and that a soft-tissue layer occurred between the bone and the cement (Fig. 12). Sections taken through the plateau interface of the tibia in Case 2 showed a layer of soft tissue adjacent to the titanium alloy. This layer was composed of fibroblasts and highly oriented collagen
J Orthop Res, Vol. 9, No. 6 , 1991
FIG. 11. Histological section taken through the transection site (Case 2) showing vascular soft tissue (S) with numerous blood vessels (V) between the bone and the prosthesis. Dead bone (D) below the transection site is being remodelled in areas adjacent to the vascular tissue and around blood vessels that migrate into dead bone. Haematoxylin and eosin, x50.
REMODELLING BONE AROUND INTRAMEDULLAR Y STEMS
815
FIG. 12. Section through the shell of bone surrounding acrylic cement in the diaphyseal femur. The bone surrounding the cement does not form a complete shell, and a soft-tissue layer (arrow) occurs between the bone and cement (C). Also note the extensive porotic inner cortex filled with marrow (M)that has developed in areas where cortical bone has resorbed. Haematoxylin and eosin, x 19.5.
fibres. An incomplete layer of bone occurred adjacent to this tissue: trabeculae were strutted from this layer and ran predominantly in a vertical direction. These trabeculae were usually thicker at the base of the antirotation groove (Figs. 13, 14). With
time the layer of bone under the plateau became complete and thicker (Fig. 15). In addition, a fibrocartilage interface was established. Fibrocartilage first appeared within the antirotation groove in specimens. Sections taken through the sleeve inter-
FIG. 13. Histological section through the tibial plateau on the medial-lateral plane showing soft tissue (S) adjacent to the prosthesis, the antirotation groove (A), the position of the metallic tibial stem (T), and remodelled bone (B) beneath the soft tissue. Haematoxylin and eosin, x4.5.
J Orthop Res, Vol. 9, N o . 6 , 1991
G . W . BLUNN AND M . E. WAIT
816
FIG. 14. Light micrograph of the interface around the antirotation groove (Case 1, duration 6 months) showing the position of the metallic antirotation lug (A) and the soft tissue ( S ) that occurs between the metal and newly remodelled bone. Sclerotic trabeculae are seen at the base of the antirotation groove. Haematoxylin and eosin, x 13.5.
face from various specimens showed that a softtissue layer was always present between the polyethylene and the shell of bone (Fig. 16). This layer measured up to 500 Frn thick. Birefringent polyethylene wear particles were not evident in this layer, which was composed predominantly of fibroblasts. DISCUSSION Massive prostheses usually rely on intramedullary fixation using acrylic cement, although bone ingrowth into porous surfaces is used in a number of designs (10). In these noncemented prostheses, resistance to torsional forces is achieved by extracortical fixation using plates and screws. Designs by Chao and Sim (6) use porous coatings over the shafts of prostheses. Pedicle formation and bone ingrowth are encouraged by bone grafts that result in enhanced torsional resistance. The present study demonstrates that the pedicle forms in the majority of retrieved cases but that connective tissue inter-
J Orthop Res. Vol. 9, No. 6, 1991
venes between the bone and the smooth metal shaft. The pedicle forms on the medial-posterior aspect of the femur in young patients, as demonstrated by our research, and in older patients with distal femoral replacements (2) and proximal tibia1 replacements (14). This aspect of the femur has been demonstrated in laboratory static loading tests to be the side that is under compressive load (22), and pedicle formation may therefore be a direct result of load-adaptive bone formation. In all cases in this study the bone at the transection site had resorbed, resulting in bone below the shoulder of the prosthesis becoming less dense because of stress protection. Burrows and colleagues (4) first reported resorption of bone due to nonphysiological loading conditions in a femur with a massive Stanmore prosthesis. Initial resorption at the transection site is accompanied by the development of a highly vascular soft-tissue interface between the shoulder of the prosthesis and resorbed bone. Blood vessels from this tissue invade necrotic cortical bone adjacent to the transection site. During surgery the blood supply to diaphyseal cortical bone is destroyed when the intramedullary cavity is reamed and packed with cement (12,19). Cortical bone adjacent to the shoulder of the prosthesis remodels and revascularises with the invasion of blood vessels from this vascular soft tissue. Rhinelander (1 1) observed diaphyseal cortical remodelling when a tight-fitting intramedullary nail blocked the reentry of the nutrient artery into the medulla of canine femora. Remodelling in the posterior femoral cortex was attributed to the path of longitudinally directed branches of the nutrient artery. Remodelling of the diaphyseal cortex in patients with growing prostheses may be a consequence of the redevelopment of the blood supply. Subperiosteal bone is laid down in areas adjacent to the remodelled cortex and is probably a result of stressadaptive bone formation. A shell of bone surrounding acrylic cement, particularly in areas where there is poor penetration into cancellous bone, was evident in this investigation. A shell of bone surrounding acrylic cement has been identified around the stems of retrieved Charnley total hip replacements (9). In addition, a shell of bone has been found around the distal nonosseointegrated stems of cementless canine total hip replacements (1). This shell of bone has been attributed by others working on sheep models to be typical of a loose implant (13). In the sheep model,
REMODELLING BONE AROUND INTRAMEDULLAR Y STEMS
817
FIG. 15. Histological section through the interface of the tibia1 plateau (Case 6 , duration 15 months) showing a complete, thick, irregular layer of lamellar bone below the soft-tissue interface. Haematoxylin and eosin, x25.
concentric layers of fibrous tissue adjacent to the cement were surrounded by a shell of dense bone, particularly where interdigitation with cancellous bone was poor. Using finite element analysis, for-
mation of this shell was attributed to a response to the hoop stresses generated during pistoning of the implant (3). Slight pistoning of the femoral component may occur around these massive prostheses,
FIG. 16. Histological section at the interface of the sleeve showing the shell of remodelled bone (B) and the soft tissue (S) between the polyethylene and the bone. Haematoxylin and eosin, X180.
J Orthop Res, Vol. 9, No. 6 , 1991
818
G . W . BLUNN AND M . E. WAIT
as the cement is surrounded by a layer of soft tissue and the plateau is not in contact with bone. Load must be transferred through a layer of vascular reactive tissue found between the plateau and the bone of the passively sliding tibial component, because the metallic stem does not bear on the end of the polyethylene sleeve. This interface is similar to those that occur under porous coated tibial trays of cementless condylar knee replacements, where ingrowth of bone is infrequently observed and a connective tissue layer is found (18). Although a shell of bone does contour the screw thread, the sleeve is not osseointegrated, and a layer of connective tissue is found adjacent to the polyethylene. Polyethylene and polysulfone have been used as porous coatings on the surface of femoral components of total hip prostheses (8). Animal studies have revealed that bone grows into and is maintained in porous polyethylene and porous polysulphone in exactly the same fashion as in porous metals (17). The reason that bone is not in direct contact with the polyethylene, as in porous surfaces, may be due to the surface topography of the threaded sleeves. The effect of the vascular reactive tissue on load transfer is speculative. In laboratory tests using embalmed femora with a silicone rubber membrane, Wright and associates (22) demonstrated that some load was transferred across this membrane at the plateau. However, the magnitude of longitudinal surface stresses induced in the intramedullary stem was greatly increased. Using a finite element model, Weinans and co-workers (20) showed that load transfer would be concentrated in particular spots when a soft-tissue membrane is present between the cement and bone in total hip arthroplasty. If such load transfer occurs around stems of major prostheses, the mechanical properties of the bonecement interface may well be exceeded. Therefore, the maintenance of the plateau-bone contact and the transfer of load across the transection site would be an ideal situation. Clearly these conditions are not achieved with current designs, and therefore the incidence of revision operations for aseptic loosening from failure of the cement-bone interface can be expected to increase in the future. Acknowledgment: The authors acknowledge the help and advice given by Professor J. T. Scales and Professor P. S. Walker. They are also indebted to the following surgeons: Mr. H. B. S. Kemp, FRCS, Mr. R. S. Sneath, FRCS, Mr. D. R. Sweetnam, FRCS, Mr. R. J. Grimer, FRCS, and Mr. S.R. Cannon, FRCS, who made the retrieval of specimens possible.
J Orthop Res. Vol. 9, No. 6, 1991
REFERENCES 1. Bobyn JD, Pillar RM, Binnington AG, Szivek JA: The effect Df proximally and fully porous coated canine hip stem design 3n bone remodelling. J Orthop Res 5:393-408, 1987 2. Bradish CF, Kemp HBS, Scales JT, Wilson JN: Distal femoral replacement by custom-made prostheses. Clinical follow up and survivorship analysis. J Bone Joint Surg [Br] 69B:27&284, 1987 3. Brown TD, Petersen DR, Radin EL, Rose RM: Global mechanical consequences of reduced cementhone coupling rigidity in proximal femoral arthroplasty: A three-dimensional finite element analysis. J Biomech 21:115-129, 1988 4. Burrows HJ, Wilson JN, Scales JT: Excision of tumours of humerus and femur with restoration by internal prostheses. J Bone Joint Surg [Br] 57-B:148-159, 1915 5 . Chandler HP, Reineck FT, Wilson RL, McCarthy JC: Total hip replacements in patients younger than 30 years old. J Bone Joint Surg [Am] 63A:1426-1434, 1981 6. Chao EYS, Sim FH: Modular types of tumour endoprostheses for limb salvage. In: Enneking WF, ed. Limb Salvage in Musculoskeletal Oncology. New York: Churchill Livingstone, 1987:227-233 7. Dorr LD, Takei GK, Conaty JP: Total hip arthroplasties in patients less than forty-five years old. J Bone Joint Surg [Am] 65A:474-479, 1983 8. Klawitter JT, Bagwell JD, Weinstein AM, Saur BW, Pruitt JR: An evaluation of bone growth into porous high density polyethylene. J Biomed Mat Res 10:311-323, 1976 9. Malcolm AJ: Pathology of cement low friction arthroplasty in autopsy specimens. In: Older J, ed. Implant Bone Znterface. Berlin, Heidelberg, New York: Springer-Verlag, 1990:77-82 10. Plenk H Jr, Salzer M, Locke H: Extracortical attachment to bioceramic endoprostheses to long bones without bone cement. Clin Orthop 132:252-265, 1978 11. Rhinelander FW: Circulation in bone. In: Bourne GH, ed. Biochemistry and Physiology of Bone. New York, London: Academic Press, 1972 12. Rhinelander FW, Nelson CL, Stewart RD, Stewart CL: Experimental reaming of the proximal femur and acrylic cement implantation-vascular and histologic effects. Clin Orthop 141:75439, 1979 13. Rose RM, Martin RB, Orr RB, Radin EL: Architectural changes in the proximal femur following prosthetic insertion: Preliminary observations of an animal model. J Biomech 17:241-250, 1984 14. Scales JT, Wait ME: Intramedullary fixation of custommade prostheses with bone cement. In: Coombs R, Fnedlaender, G, ed. Bone Tumour Management. London: Buttenvorth & Co., Ltd., 1987:123-134 15. Scales JT, Sneath RS, Wright KWJ: Design and clinical use of extending prostheses. In: Enneking WF, ed. Limb Salvage in Musculoskeletal Oncology. New York: Churchill Livingstone, 198752-6 1 16. Simon M: Current concepts review: Causes of increased survival of patients with osteosarcoma: current controversies. J Bone Joint Surg [Am] 66A:30&310, 1984 17. Spector M: Bone ingrowth into porous polymers. In: Williams DF, ed. Biocompatibilify of Orthopaedic Implants. Boca Raton: CRC Press, 1982:55-88 18. Thomas KA, Cook SD, Thomas KL, Haddad JR: Tissue growth into retrieved noncemented human hip and knee components. In: Saha S, ed. Biomedical Engineering V. Recent Developments. New York: Pergamon Press, 1986:198203
REMODELLING BONE AROUND INTRAMEDULLAR Y STEMS 19. de Wall Malefijt J, Slooff TJJ, Huiskes R, de Laat EAT, Barentez JO: Vascular changes following total hip arthroplasty. The femur in goats studied with and without cementation. Acta Orthop Scand 59543-649, 1988 20. Weinans H, Huiskes R, Grootenboer H: The mechanical effects of fibrous tissue interposition at the cement bone interface in THA. In: Transactions of the 34th Annual Meeting of the Orthopaedic Research Society, 1988;13:502
819
21. Wilson PD Jr, Lance EM: Surgical reconstruction of the skeleton following segmental resection for bone tumours. J Bone Joint Surg 47A:1629-1656, 1965 22. Wright KW, Meswania JW, Scales JT: Load transmission characteristics of intramedullary stedpolymethyl methacrylate fixation in endoprostheses after tumour resection. In: Enneking WE, ed. Limb Salvage in Musculoskeletal Oncolo g y . New York: Churchill Livingstone, 1987
J Orthop Res, Val. 9,No. 6, 1991
