The options for reconstruction after resection of a bone tumor about the knee include the use of a custom total knee prosthesis, an osteoarticular allograft or an allograft-joint replacement composite, an arthrodesis with intercalary bone-grafting, or conversion to a rotationplasty. For some patients, an arthrodesis is the best option because of the patient's age, weight, or desired level of activity or because the extent of resection of bone or soft tissue makes an arthroplasty impossible. An arthrodesis also may be indicated as a salvage procedure after a prosthetic replacement or an osteoarticular allograft has failed.
Several authors have reported on the technique of resection arthrodesis; the defect has been reconstructed with use of various combinations of autogenous grafts from the fibula, tibia, or iliac crest2-4,7. In these studies, a rate of non-oncological complications of as high as 40 per cent has been reported7; the complications have included non-union, fatigue fracture of the graft, loss of fixation, and inadequate soft-tissue coverage. These problems, along with the limited availability of autogenous bone, have led to interest in the use of allografts for arthrodesis. Several centers9,11-13 have reported satisfactory early results in a small number of patients who had a resection arthrodesis of the knee with an allograft. This study is a review of the results for patients whom we have managed with this technique.
*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. No funds were received in support of this study.
†Akron City Hospital, 55 Arch Street, Suite 3A, Akron, Ohio 44304.
‡Department of Orthopaedic Surgery, University of Florida, Box 100246 JHMC, Gainesville, Florida 32610.
Forty-two patients had a resection arthrodesis of the knee with use of an intercalary allograft fixed with a long intramedullary nail. Three patients were excluded from the study because they died fewer than six months after the operation: two died from metastatic disease and one, from a complication related to chemotherapy. Of the thirty-nine remaining patients, twenty-five were male and fourteen were female, and the ages at the time of the operation ranged from ten to fifty-five years (average, twenty-six years). The cause of the bone loss was resection of an osteosarcoma in twenty-six patients, a giant-cell tumor in five, a malignant fibrous histiocytoma of bone in three, a chondrosarcoma in three, a parosteal osteosarcoma in one, and an undifferentiated sarcoma in one. The tumor involved the distal part of the femur in thirty-one patients and the proximal end of the tibia in eight (Table I).
The decision to perform a resection arthrodesis was based on the anatomical extent of the lesion, as determined from the radiographic staging studies, and extensive discussion with the patient (and family) regarding the various options for reconstruction and their functional consequences.
Thirty-one patients had a primary (one-stage) resection arthrodesis with an intercalary allograft. In sixteen, an extra-articular resection of the knee (en bloc resection of the entire knee joint along with the involved segment of bone) was necessary because a previous operation or a pathological fracture had contaminated the knee joint or because tumor extended into the pericapsular structures or the knee joint. An arthrodesis was recommended for these patients because the extent of the soft-tissue resection left no available muscle to power an arthroplasty and neither a rotationplasty nor an amputation was acceptable to the patient. The other fifteen patients who had a primary resection had an intra-articular resection, with the plane of dissection passing through the knee joint. A resection arthrodesis was done in these patients because of the extent of the soft-tissue resection or because it was the patient's choice. The operative margin was wide (the plane of dissection was through normal tissue) in twenty-nine patients and marginal in two.
Eight patients (Cases 9, 10, 11, 14, 22, 28, 32, and 36) had had a resection of the tumor and another type of reconstruction, two to five years earlier. The reconstruction had failed without evidence of recurrence of the tumor. Three of these patients (Cases 9, 10, and 11) had a persistent non-union after a resection arthrodesis with autogenous cortical and cancellous bone grafts. Two patients (Cases 14 and 22) had a symptomatic, unstable osteoarticular allograft of the distal part of the femur. In one patient (Case 36), two osteoarticular allografts at the proximal end of the tibia had fractured, and in another (Case 28), multiple fractures and symptomatic patellofemoral disease had developed after curettage and cementing for a large giant-cell tumor of the distal part of the femur. One patient (Case 32), who had a long-stem custom total knee replacement that had become loose and associated with infection, was managed with a two-stage procedure: the prosthesis was removed, antibiotics were given locally and systemically, and then an arthrodesis was done.
The preoperative planning and the operative technique were similar for all of the patients. Full-length (hip-to-ankle) anteroposterior and lateral radiographs were made of the involved limb, along with a ruler and a magnification marker. The planned extent of the resection was marked on the radiographs, and the length of the allograft to be used and the diameter of its intramedullary canal were measured.
In patients who had little or no skeletal growth remaining, the goal was to leave the limb approximately one to two centimeters shorter than the contralateral limb to facilitate the swing phase of walking. In patients who had more than one year of potential skeletal growth remaining, the goal was to lengthen the limb by two to four centimeters to accommodate future growth of the contralateral limb. The average length of the allograft was nineteen centimeters (range, thirteen to thirty centimeters).
Use of a fluted intramedullary knee arthrodesis nail (Sampson nail; Kirschner Medical, Timonium, Maryland) was the primary method of fixation. (It should be noted that since 1994 we have used a long Russell-Taylor locking nail [Smith and Nephew Richards, Memphis, Tennessee] for fixation, as the Sampson nail is no longer available.) The diameter, length, and curvature of the nail were chosen on the basis of the radiographic planning studies and the goals regarding the length of the limb. The nominal diameter of the nail was determined by the smallest diameter of the intramedullary canal of either the femur or the tibia, as measured on the anteroposterior or lateral radiographs. The diameter of the intramedullary canal of the tibia was usually the limiting factor, and the eleven or twelve-millimeter-diameter nail was used most often. The length of the intramedullary nail was calculated so that it would extend from the base of the greater trochanter to approximately two centimeters proximal to the ankle joint. A nail that was somewhat shorter than this distance was usually chosen, as the Sampson nail can be lengthened by as much as eight centimeters intraoperatively, with the use of extenders that screw into the ends of the nail. The proximal two-thirds of the nail is curved to match the curve of the femur and so that the knee fuses in slight flexion. The distal third of the nail is straight to match the straight intramedullary canal of the tibial shaft. The proximal curvature was measured from the full-length lateral radiograph, and the radius of the curvature usually was fifty or sixty inches (127 or 152 centimeters).
All of the allografts were procured and processed under sterile conditions by the University of Florida Musculoskeletal Tissue Bank, according to the guidelines established by the American Association of Tissue Banks, and were stored at -80 degrees Celsius. The allograft was matched primarily by size, if possible, so that the diameters of the intramedullary canals of the allograft and host bone were within several millimeters of each other at the level of the planned diaphyseal junction after the arthrodesis. Matching according to right or left side was not found to be essential, as an allograft from either side could easily be shaped to fit the defect.
Operative Technique
After the tumor had been resected or the graft or prosthesis employed for the previous failed reconstruction had been removed, the selected allograft was thawed and prepared for use. Transverse osteotomies were performed at both allograft-host junctions. The intramedullary canals of the allograft and femur were usually reamed to a diameter that was 1.0 to 1.5 millimeters more than the diameter of the selected intramedullary nail, and the canal of the tibia was reamed to a diameter that was 1.5 millimeters more. The Sampson nail was used in thirty-eight patients, and a custom locking (Russell-Taylor) nail was used in one. The proximal end of the intramedullary nail was driven retrograde up through the reamed femur, through the piriformis fossa, and out through an incision in the buttock.
The allograft was placed on the distal end of the nail, and the fit of the proximal allograft-host junction was modified, if necessary, to achieve uniform contact at the junction. The lower limb was placed in 5 to 10 degrees of external rotation, with the hip in neutral rotation. Marking of the anterior cortex of the femur and tibia before resection helped to position the distal part of the reconstructed extemity in the desired amount of rotation. The distal allograft-host junction was reduced and modified as necessary for uniform contact, and the nail was driven distally into the distal end of the tibia. The junctions were examined for rotational and bending stability, and the final position of the intramedullary nail was verified on intraoperative radiographs.
In the three patients who had a failed osteoarticular allograft, the existing allograft-host junction had already healed. The articular surfaces and the previous internal fixation devices were removed, and the technique for insertion of the nail was modified accordingly.
Each allograft-host junction was characterized by the type of bone at the junction, as the healing process differs13. A metaphyseal junction was defined as one between the metaphyseal end of the allograft and the metaphyseal bone of either the distal part of the femur or the proximal end of the tibia, depending on the site of the tumor. A diaphyseal junction was defined as that between the diaphysis (that is, cortical bone) of the allograft and the remaining part of the diaphysis of the proximal part of the femur or the distal end of the tibia. A mixed junction resulted from the use of an allograft that was only a segment of diaphyseal bone and contained no metaphyseal bone at either end. This type of junction was between the diaphyseal bone of the allograft and the metaphyseal bone of the remaining part of the femur or tibia. Thirty-five patients had one metaphyseal and one diaphyseal junction, and four patients (Cases 3, 10, 11, and 24) had one diaphyseal junction and one mixed junction. Thus, the thirty-nine patients had a total of seventy-eight junctions.
In the early part of the series, autogenous bone was commonly taken from the iliac crest and was grafted at the junctions and along the allograft. In three patients (Cases 1, 2, and 7), autogenous non-vascularized fibular bone graft was placed next to the allograft. As these types of supplemental bone grafts increased the complexity of the operation and did not seem to enhance healing or function, they were not used in the remaining patients. In nineteen patients, the operative report noted that local bone (from preparation of the host bone) or material obtained from intramedullary reaming was placed around the allograft-host junctions.
Local rotational muscle flaps with split-thickness skin grafts were used as needed to obtain both tension-free closure and soft-tissue coverage of the allograft. Twenty-three patients had a muscle flap consisting of the medial or lateral gastrocnemius (twenty-one patients), biceps femoris (one patient), sartorius (one patient), or a combination of these muscles.
In two patients (Cases 13 and 18), rotational instability was noted during the operation. The unstable junction was supplemented with a plate, with the screws angled to avoid the nail.
The type of immobilization used postoperatively varied over the course of the study. A cylinder cast or a thigh-to-foot orthosis was used initially, and only a soft dressing was used more recently. The patient was allowed toe-touch weight-bearing as soon as the incision had healed and then progressed to full weight-bearing over the next four to six months. Follow-up radiographs of the limb were made at three to four-month intervals for at least the first year and then at four to six-month intervals as part of the patient's routine oncological follow-up. Twenty-seven patients also received two courses or more of chemotherapy postoperatively.
The allograft-host junction was considered to be healed when the patient had no symptoms related to the junction site and either the junction line was no longer visible radiographically or it was bridged with periosteal bone on anteroposterior and lateral radiographs14. In the three patients (Cases 14, 22, and 36) in whom a previous allograft had been retained, the time to healing of the existing diaphyseal junction was measured from the time of the original operation. All of the radiographs were independently reviewed by two of us (S. D. W. and R. A. V. G.), and any differences in the estimated time to union or the type of union were resolved by repeat review of the radiographs. Functional evaluation was carried out with the system of the Musculoskeletal Tumor Society5 (Table II). Additional follow-up information was obtained, as needed, from the patients or their local physicians.
Thirty-four of the thirty-five metaphyseal junctions healed in an average of seven months. Healing resulted in a characteristic transverse zone of increased radiodensity in the allograft, usually within two centimeters of the junction site6,14 (Fig. 1). In the one patient who had a metaphyseal non-union (Case 23), a diaphyseal non-union also developed and the intramedullary nail broke. A second nailing with bone-grafting at both junctions was followed by healing.
All four of the mixed junctions healed, but at two, the allograft had settled about one centimeter into the host metaphyseal bone. No shortening was observed at any of the other types of junctions.
Thirty-two of the thirty-nine diaphyseal junctions healed at an average of thirteen months (range, six to forty-two months). Two patterns of radiographic healing were seen at these junctions. One pattern consisted of a variable amount of callus that extended from the host bone across the junction onto the allograft and then slow disappearance of the osteotomy line. This healing pattern was seen at all but five of the diaphyseal junctions that healed. In the second pattern, a large amount of callus formed and extended across the osteotomy site and along the surface of the allograft. The osteotomy line became less visible but did not completely disappear (Fig. 1). In all five of the patients who had this pattern of healing, there was a zone of radiolucency around the intramedullary nail near this junction, suggesting motion.
In one patient (Case 35), motion developed at the diaphyseal junction within three months after the operation. This was treated with fixation with a plate and autogenous bone-grafting, and the junction healed.
The functional ratings were similar among the patients in whom both junctions healed. These patients had a stable lower extremity and did not use external supports. They usually reported some limitation of function during recreational activities but were generally satisfied with the reconstruction.
We could not identify any difference in the prevalence of union or the time to union between patients who had and those who did not have an autogenous bone graft added to the allograft-host junction. There was also no difference in union between patients who had and those who did not have chemotherapy postoperatively.
Seven patients (Cases 5, 6, 10, 23, 30, 32, and 39) had a non-union at the diaphyseal junction that led to additional treatment. Two of these patients eventually had an above-the-knee amputation. In one (Case 5), the intramedullary nail broke three years after the operation. In the other (Case 30), a bone graft and plate had been added but the junction did not heal. Neither patient wished to return for additional attempts at reconstruction, and both had an amputation done by their local physician.
Two patients with a chronic non-union eventually had replacement of the allograft and repeat fixation. In one of these patients (Case 6), both the intramedullary nail and the allograft broke eight years after the operation. The other patient (Case 10) had a radiographically apparent non-union and almost no symptoms for eight years; increased pain and resorption of the allograft then developed. In both of these patients, the revision allograft was successful.
The original allograft was retained in the three remaining patients who had a non-union. One (Case 23) was managed with repeat intramedullary fixation and bone-grafting, as described; one (Case 32), with onlay autogenous bone-grafting and no supplemental fixation; and one (Case 39), with plate fixation and bone-grafting. All of these junctions healed.
Five patients (Cases 1, 6, 12, 20, and 25) had a fatigue fracture more than one year after the metaphyseal junction had healed. In each, a radiolucent line extended either partially or completely across the metaphysis of the allograft proximal to the transverse zone of increased radiodensity. In the first patient (Case 1), this radiolucency was thought to represent a tumor, but biopsy showed only repair tissue. This same radiographic appearance of the fracture was similar in the remaining four patients. Four of the five fractures healed with formation of callus on the surface of the allograft and gradual disappearance of the fracture line. The fifth patient (Case 6) had displacement of the fracture when the intramedullary nail broke at the proximal non-union site.
Ten patients had pain in the distal part of the leg, usually at the level of the distal end of the nail. Pain was more common when the nail did not extend all the way into the distal metaphysis of the tibia, and it was usually transient or intermittent. However, one patient (Case 35) had persistent severe pain, which resolved after the initial nail was removed and a longer nail was inserted. Symptomatic proximal migration of the intramedullary nail occurred in two patients (Cases 13 and 36). The proximal extender segment was removed and the rest of the nail was left in place in both. Substantial distal migration of the nail did not occur in any patient.
Six patients (Cases 20, 27, 29, 31, 34, and 38) were thirteen years old or less at the time of the operation. In all of these patients, the lower extremity was lengthened by as much as three centimeters at the time of the arthrodesis. One patient (Case 31) died of metastatic disease; one (Case 38) died in an accident; and three (Cases 20, 27, and 34) reached skeletal maturity, had good function despite the stiff knee, and had a limb-length discrepancy of fewer than three centimeters. At the time of the most recent follow-up, the remaining patient (Case 29) was sixteen years old and had a limb-length discrepancy of 7.0 centimeters, for which a shoe-lift was used; lengthening of the limb was being considered.
Four patients needed a reoperation for an immediate postoperative complication. One patient (Case 16) had a fasciotomy for a compartment syndrome of the leg, one (Case 37) had drainage of a hematoma, and two (Cases 26 and 30) had excision followed by closure of the skin for a small area of marginal wound necrosis. There were no deep infections. Three patients (Cases 13, 29, and 35) had transient peroneal-nerve palsy, which we believe was probably caused by the dissection needed for the resection of the tumor rather than by the reconstruction.
In one patient (Case 14), a Garden stage-III fracture of the ipsilateral femoral neck occurred fourteen months after the arthrodesis. The fracture did not involve the exit hole for the nail in the piriformis fossa. The fracture was treated with a closed reduction and fixation with three percutaneous screws that were passed anterior to the intramedullary nail. The fracture healed, and nine years later there was no evidence of aseptic necrosis.
One patient (Case 21) had a local recurrence in the remaining proximal part of the tibia eighteen months after an intra-articular resection of the distal end of the femur. A mid-thigh amputation was done, but pulmonary metastasis developed and the patient died. One patient (Case 38) who had metastasis died in an accident eighteen months after the operation, as already mentioned. Metastatic disease developed in seven other patients and leukemia, in one; six of these patients died, twelve to twenty-nine months after the resection arthrodesis. At the time of the last orthopaedic follow-up examination, the allograft-host junctions had healed in all of them and they had no restriction of activities because of the lower extremity.
The choice of reconstruction after resection of a tumor about the knee is based on many factors, including the type and extent of the tumor, the amount of tissue that must be removed to achieve the desired margin, the experience of the surgeon with the various reconstructive procedures, an understanding of the advantages and disadvantages of the reconstructive operations, and the needs and preferences of the patient. An arthrodesis offers the potential advantage of being a stable, durable, and predictable reconstruction that enables unrestricted activity but at the expense of losing motion of the knee. Once the decision has been made to perform an arthrodesis, the technique selected should be associated with a high rate of success and few complications, as this group of patients will probably receive chemotherapy postoperatively and some will have a limited life expectancy.
Our previous experience, as well as that reported by others2-4,7, has shown that the use of autogenous bone grafts and an intramedullary nail can result in successful fusion but that the prevalence of non-oncological complications is high. In the present series, an allograft was used in an attempt to circumvent some of these problems. Allografts are available in a wide range of sizes and shapes, and they make it possible to avoid the complications related to an autogenous donor site, such as pain, infection, additional operative time to obtain the graft, and the potential for contamination of the donor site with tumor. However, there is a risk of transmission of infectious disease, and the immunobiology of the healing and incorporation of allografts is still not completely understood.
Benevenia et al., in a follow-up of Makley's initial series, reported on nineteen patients who had been managed with an allograft stabilized with an intramedullary nail. They reported good results, but they found non-oncological complications in eight of the nineteen patients. Others4,9 have reported similar results.
In our series of thirty-nine consecutive patients who were managed with an allograft and arthrodesis, a non-union developed in seven patients: six had a diaphyseal non-union and one had both a metaphyseal and a diaphyseal non-union. There was radiographic evidence of instability at all of these allograft-host junctions, with either increased periosteal new bone attempting to bridge the junction or progressive radiolucency between some portion of the intramedullary nail and the bone, or both.
Stable fixation was more difficult to achieve when a short segment of bone remained after the resection or when the differences among the diameters of the intramedullary canals of the tibia, allograft, and femur could not be equalized by reaming. Because the diameter of the nail is chosen on the basis of the diameter of the intramedullary canal of the smallest bone (usually the tibia), it was often difficult to achieve a uniform interference fit along the entire nail as it passed through the tibia, allograft, and femur. This is less of a problem at the metaphyseal junction, which has a large surface area that can be impacted for added stability. Stability was more of a problem at the diaphyseal junctions, especially in the proximal end of the femur when the resection extended proximal to the femoral isthmus.
When rotational or axial instability is noted intraoperatively, the junction should be supplemented with a short plate, as was done in two of our patients. In one other patient, a plate was added when instability of the junction became apparent three months after the operation. The addition of a plate did not seem to have adverse effects on the healing of the junction. For supraisthmal, diaphyseal junctions in the proximal part of the femur, we now either supplement the fixation at the junction with a short plate or use a nail that can be locked into the proximal end of the femur.
In three of the seven patients who had a non-union, the intramedullary nail eventually broke and an additional operation was needed. We believe that early intervention with bone-grafting and supplemental fixation, if needed, should be considered when there is radiographic evidence of instability, such as increasing radiolucency along the nail, especially when there is also pain or tenderness at the junction site. Persistence of the osteotomy line in the absence of either clinical symptoms or radiographic signs of instability does not necessarily indicate a non-union, as we observed gradual disappearance of the osteotomy line over more than two years in several patients.
Three patients had an amputation. One had a local recurrence and the other two had, in our opinion, a potentially salvageable problem with the reconstruction; however, both chose to have an amputation done by their local physician rather than to return for additional reconstruction.
A fatigue fracture occurred in five patients. All of the fractures were in the metaphyseal bone of the allograft and occurred through or just proximal to the zone of increased radiodensity that was seen at the healing metaphyseal junction. As the metaphyseal junction heals, new bone is laid down on allograft trabeculae. This process is rapid but extends only a short distance into the allograft6. These fatigue-type fractures seemed to occur at the end of this revascularization zone, which may be an area of relative weakness between the revascularized, remodeled metaphysis of the allograft and the rest of the allograft. When the intramedullary fixation was stable, none of these fractures became displaced and all healed with formation of periosteal new bone on the surface of the allograft. Operative treatment was not needed once it was recognized that the resorption and new-bone formation characteristic of this process did not necessarily represent a local recurrence or infection.
The use of postoperative chemotherapy or local, iliac-crest, or fibular bone graft did not seem to have an effect on the occurrence of or the time to healing of the allograft-host junctions. Although rotational muscle flaps were used in more than half of the patients, the effect of these flaps on healing of the junction is unknown.
The use of allografts has been associated with an increased risk of infection. Lord et al. reported a rate of infection of 12 per cent (thirty-three of 283 patients), which was believed to be related to a prolonged operative time, multiple operations, and complications associated with the wound. A deep wound infection did not occur in any patient in our series. This may be due to the liberal use of pulsating lavage irrigation intraoperatively, the use of local muscle flaps to obtain a tension-free closure, and early operative treatment of superficial wound problems.
We believe that resection arthrodesis continues to be a valid reconstructive procedure for tumors about the knee. The use of allograft bone resulted in a stable, functional reconstruction in most patients. The patients' acceptance of the reconstruction was high, despite the loss of motion of the knee. Most of the problems and complications in this series were related either to the complexity of the resection of the tumor (rather than the method of reconstruction) or to problems with fixation of the allograft.
Benevenia, J.; Makley, J. T.; Locke, M.; Gentil, A.; and Heiner, J.: Resection arthrodesis of the knee for tumor. In Complications of Limb Salvage. Prevention, Management, and Outcome, pp. 69-74. Edited by K. L. B. Brown. Montreal, International Symposium of Limb Salvage, 1991.
Campanacci, M., and |and |Costa, P.: Total resection of distal femur or proximal tibia for bone tumours. Autogenous bone grafts and arthrodesis in twenty-six cases. J. Bone and Joint Surg.,61-B(4): 455-463, 1979.61-B(4)455
1979
D'Aubigné, R. N., and |and |Dejovany, J. P.: Diaphyso-epiphysial resection for bone tumour at the knee. With reports of nine cases. J. Bone and Joint Surg.,40-B(3): 385-395, 1958.40-B(3)385
1958
Dunham, W. K., and |and |Calhoun, J. C.: Resection arthrodesis of the knee for sarcoma. Preliminary results. Orthopedics,7: 1810-1819, 1984.71810
1984
Enneking, W. F.: Modification of the system for functional evaluation of surgical management of musculoskeletal tumors. In Limb Salvage in Musculoskeletal Oncology, pp. 626-639. Edited by W. F. Enneking. New York, Churchill Livingstone, 1987.
Enneking, W. F., and |and |Mindell, E. R.: Observations on massive retrieved human allografts. J. Bone and Joint Surg.,73-A: 1123-1142, Sept. 1991.73-A1123
1991
Enneking, W. K., and |and |Shirley, P. D.: Resection-arthrodesis for malignant and potentially malignant lesions about the knee using an intramedullary rod and local bone grafts. J. Bone and Joint Surg.,59-A: 223-236, March 1977.59-A223
1977
Garden, R. S.: Low-angle fixation in fractures of the femoral neck. J. Bone and Joint Surg.,43-B(4): 647-663, 1961.43-B(4)647
1961
Gebhardt, M. C.; McGuire, M. H.; and Mankin, H. J.: Resection and allograft arthrodesis for malignant bone tumors of the extremity. In Limb Salvage and Musculoskeletal Oncology, pp. 567-582. Edited by W. F. Enneking. New York, Churchill Livingstone, 1987.
Lord, C. F.; Gebhardt, M. C.; Tomford, W. W.; and |and |Mankin, H. J.: Infection in bone allografts. Incidence, nature, and treatment. J. Bone and Joint Surg.,70-A: 369-376, March 1988.70-A369
1988
Makley, J. T.: The use of allografts to reconstruct intercalary defects of long bones. Clin. Orthop.,197: 58-75, 1985.19758
1985
[PubMed]
Mankin, H. J.; Doppelt, S.; and |and |Tomford, W.: Clinical experience with allograft implantation. The first ten years. Clin. Orthop.,174: 69-86, 1983.17469
1983
[PubMed]
Reinecke, M. C.; Springfield, D. S.; and Vander Griend, R. A.: Resection arthrodesis of the knee using intercalary allograft. Read at the Annual Meeting of The American Academy of Orthopaedic Surgeons, Atlanta, Georgia, Feb. 6, 1988.
Vander Griend, R. A.: The effect of internal fixation on the healing of large allografts. J. Bone and Joint Surg.,76-A: 657-663, May 1994.76-A657
1994