Knee Arthrodesis Abstract Arthrodesis is one of the last options available to obtain a stable, painless knee in a patient with a damaged knee joint that is not amenable to reconstructive measures. Common indications for knee arthrodesis include failed total knee arthroplasty, periarticular tumor, posttraumatic arthritis, and chronic sepsis. The primary contraindications to knee fusion are bilateral involvement or an ipsilateral hip arthrodesis. A variety of techniques has been described, including external fixation, internal fixation by compression plates, intramedullary fixation through the knee with a modular nail, and antegrade nailing through the piriformis fossa. Allograft or autograft may be necessary to restore lost bone stock or to augment fusion. For the carefully selected patient with realistic expectations, knee arthrodesis may relieve pain and obviate the need for additional surgery or extensive postoperative rehabilitation. I n 1958, Charnley and Lowe 1 re- ported a 98.8% fusion rate in 171 adults who underwent compression arthrodesis of the knee. The authors used one transfixion pin in the distal femur and another in the proximal tibia, with compression clamps placed medially and laterally (Figure 1). Most of the patients had tubercu- losis, osteoarthritis, or rheumatoid arthritis; none had a previously at- tempted knee arthroplasty. Howev- er, with the development of total knee arthroplasty (TKA), the failed TKA became an indication for knee arthrodesis. 2 Using Charnley clamps for the procedure, Hagemann et al 3 in 1978 reported a fusion rate of only 64% in 14 patients with failed TKA. Lack of adequate bone contact and poor stability were cited as the main causes of failure. Unsatisfactory fusion rates in post-TKA patients initiated develop- ment of new techniques to improve the reliability of fusion. To improve stability and rates of fusion, both ex- ternal and internal fixation methods were attempted. The external fixa- tion constructs used were unipla- nar, 4 modified biplanar frames with transfixion pins and half pins, 5,6 or circular frames. 7-9 Internal fixation devices included an anterior ten- sion-band plate, 10 double plating, 11 antegrade locking nail, 12-14 combined intramedullary rod and plate fixa- tion, 15 and modular intramedullary nails. 16,17 No single technique has proved to be superior in all situa- tions; each has its relative advantag- es and disadvantages. Indications and Contraindications In the late 1800s and early 1900s, knee arthrodesis primarily was per- formed for sepsis, articular tubercu- losis, and instability caused by polio- myelitis. 1 Today, the most common indication for fusion is a failed, unre- James H. MacDonald, MD Sanjeev Agarwal, FRCS (Orth) Matthew P. Lorei, MD Norman A. Johanson, MD Andrew A. Freiberg, MD Dr. MacDonald is in private practice in Annapolis, MD. Dr. Agarwal is Orthopaedic Surgeon, Leeds Orthopaedic Program, Leeds General Infirmary, Leeds, UK. Dr. Lorei is in private practice in Bryn Mawr, PA. Dr. Johanson is Chair, Orthopaedic Surgery, Drexel College of Medicine, Philadelphia, PA. Dr. Freiberg is Chief, Arthroplasty Service, Massachusetts General Hospital, Boston, MA. None of the following authors or the departments with which they are affiliated has received anything of value from or owns stock in a commercial company or institution related directly or indirectly to the subject of this article: Dr. MacDonald, Dr. Agarwal, Dr. Lorei, and Dr. Johanson. Dr. Freiberg or the department with which he is affiliated serves as a consultant to or is an employee of Zimmer. Reprint requests: Dr. MacDonald, Orthopaedic and Sports Medicine Center, 108 Forbes Street, Annapolis, MD 21401. J Am Acad Orthop Surg 2006;14:154- 163 Copyright 2006 by the American Academy of Orthopaedic Surgeons. 154 Journal of the American Academy of Orthopaedic Surgeons visable TKA, which may be second- ary to persistent infection, massive bone or soft-tissue loss, or irrepara- ble damage to the extensor mecha- nism. 18 Septic arthritis and osteomy- elitis are the next most common indications after failed TKA for knee arthrodesis. An uncommon complication fol- lowing TKA is loss of the extensor mechanism. A number of techniques have been described for reconstruc- tion, including extensor allograft, which achieves variable results. 19 Knee fusion may be considered as a salvage method for this condition, particularly when extensor mecha- nism reconstruction has failed. Patients with highly aggressive or malignant periarticular tumors may also be treated by fusion after resec- tion when allograft, metal prosthe- ses, or rotationplasty are not appro- priate. Fusion is most commonly considered after tumor resection when the extensor mechanism is lost or in the presence of infection. 20 Occasionally, knee arthrodesis is indicated for young patients with unilateral posttraumatic degenera- tive joint disease who maintain jobs that require heavy manual la- bor. 12,21 Rarely seen indications are painful ankylosis, paralytic condi- tions with severe deformities, and neuropathic joints. Despite exten- sive soft-tissue release, a chronical- ly painful and stiff knee is unlikely to regain useful range of motion fol- lowing total knee replacement. In patients with neuropathic joints, ar- throplasty may fail because of asep- tic loosening or a poor functional re- sult. In these circumstances, fusion may be indicated for salvage. 22 Pa- tients who experience significant disability from instability also are candidates for elective arthrodesis. The primary contraindications to knee fusion are bilateral involve- ment or an ipsilateral hip arthrode- sis. 2 When planning surgery, it is im- portant to check the ipsilateral hip for arthritis; a fused knee will trans- fer more stress to the hip and ankle. The decision to proceed with a knee arthrodesis should be individu- alized, dependent on the clinical sit- uation and the expectations of the patient. The functional limitations imposed by arthrodesis should be fully understood preoperatively by the patient. Alternatives to Knee Arthrodesis When primary TKA or revision TKA are not good options, resection ar- throplasty may be considered as an alternative to knee arthrodesis in some patients. This procedure in- volves removal of all components and cement when an arthroplasty is in place as well as thorough débride- ment. The bone ends are shaped to provide maximum contact in exten- sion; however, no attempt is made to achieve fusion. Postoperative sup- port is provided by a cast followed by bracing, when required, and patients are encouraged to flex and extend the knee as is comfortable. In one report of 26 patients, 28 knees with failed, infected TKA were managed by resection arthroplasty. 23 Systemic signs of infection were eliminated in all knees; local signs of infection were eliminated in 89% of the knees. After surgery, 58% of the patients (15/26) were able to ambu- late without assistance, 3 knees had spontaneous fusion, and 6 were later converted to an arthrodesis because of instability. In the successful group, the average postoperative range of motion was 36°. The authors con- cluded that patients’ degree of satis- faction following a resection arthro- plasty is directly related to their degree of preoperative disability. Transfemoral amputation is an option for the severely damaged knee not suitable for primary or revision TKA, resection arthroplasty, or fu- sion. General Principles Systemic problems, such as diabetes mellitus, rheumatoid arthritis, chronic renal failure, peripheral vas- Figure 1 Charnley compression clamp. (Adapted with permission from Nelson CL, Evarts CM: Arthroplasty and arthrodesis of the knee joint. Orthop Clin North Am 1971;2:245-264.) James H. MacDonald, MD, et al Volume 14, Number 3, March 2006 155 cular disease, and corticosteroid use, are evaluated preoperatively because of their potential effect on wound healing. In addition, a complete pe- ripheral neurologic examination is performed and any flexion deformi- ty, malalignment of the limb, and loss of range of motion are noted. The examination includes standard anteroposterior and lateral knee ra- diographs along with standing hip- to-ankle radiographs for assessing limb alignment and position of ex- isting implants. Assessment of existing scars is important; a solitary longitudinal scar can be used for the surgical inci- sion in knee arthrodesis. Because most of the blood supply to the skin overlying the knee is derived from fascial perforators from the medial side, the surgical approach should use the most lateral longitudinal parapatellar incision when previous multiple incisions are present. 24 An existing transverse scar can be crossed perpendicularly without compromise of the skin flaps; how- ever, plastic surgery consultation is recommended when the viability of the overlying skin is in question. A preoperative trial period in a knee immobilizer or a cylinder cast, simulating a knee fusion, will dem- onstrate for the patient potentially problematic activities, such as the use of public transportation and the increased energy required for ambu- lation. Surgical Technique At surgery, previous incisions are clearly marked before application of sterile adhesive drapes. Skin flaps should not be undermined more than necessary and should be handled with care to minimize damage to cu- taneous circulation by forceps or forceful retraction. Stiffness may be a problem when removal of a knee prosthesis is required at the time of débridement or fusion. Many exten- sile approaches have been described and may be used to gain adequate ex- posure. 25 The application of sound fracture fixation principles can improve the likelihood of successful arthrodesis. The elements needed to achieve bony union are good viable bone contact, adequate blood supply, and rigid fixation. The cruciate ligaments and me- niscal remnants are removed, as are prosthetic components and any bone cement. The final position for knee fusion should be in the “neutral range”: 7° ± 5° of valgus and 15° ± 5° of flexion. 26 TKA cutting guides may be used to achieve this alignment. To maintain limb length, minimal bone resection is performed when the fusion is being done after TKA. Bone graft is required when there is marked loss of bone stock from the distal femur or the proximal tibia. Large bone defects can be managed by use of a vascularized fibular rota- tory graft. In a series of 13 patients so treated, fusion was achieved in 12 (mean follow-up, 51 months). 27 The patella may be left alone, used as a graft to fill defects, excised to relieve tension on the incision, or fused to the trochlear groove. Good apposi- tion of the bone ends is critical to achieving fusion. Careful attention to hemostasis helps to reduce the in- cidence of wound hematoma and breakdown. A slightly foreshortened, fused leg allows easier clearance during the swing phase of gait. When the leg is too short, a shoe lift may be used. The presence of active sepsis may decrease the fusion rate; therefore, it is necessary to eradicate infection before the arthrodesis procedure. Conventional External Fixation The advantages of external fixa- tion compared with internal include limited exposure and decreased blood loss during application. Exter- nal fixation does not involve perma- nent hardware that could act as a po- tential nidus for infection, and there is no risk of spreading an indolent in- fection in the knee farther into the tibial or femoral canal as there is with internal rod or plate fixation. External fixation provides intraoper- ative flexibility, allowing alignment of the fusion site into a position of mild flexion and valgus, compared with intramedullary fixation. Final- ly, when repeated débridements are required, the fixator frame can be disconnected from the pins during the procedure and then reassem- bled. Disadvantages of external fixa- tion include the risk of pin-tract in- fection and pin loosening before suc- cessful fusion. In addition, a risk of neurovascular injury during pin in- sertion exists. The apparatus is more cumbersome postoperatively com- pared with internal fixation. Further, weight bearing cannot advance as quickly, and there is a risk of stress fracture through the pin sites. Final- ly, in some series, a lower fusion rate is reported for external fixation com- pared with intramedullary fixation after failed TKA (Table 1). External fixation is preferred, however, for patients with good bone stock in whom pseudarthrosis is less likely, for those in whom blood loss or fat emboli associated with in- tramedullary nailing would not be well tolerated, and for those at par- ticularly high risk for infection with the use of internal fixation devices. In a comparative study of 26 patients undergoing knee fusion, all 13 with intramedullary nailing achieved fusion, whereas 6 of 13 with an external fixator required re- vision surgery and intramedullary nailing for successful fusion. 12 How- ever, the study groups were not ran- domized, and the external fixation group had a higher percentage of vir- ulent organisms. The surgical technique used for all fixators follows similar princi- ples: a longitudinal incision is made, and the bone is cut from the distal femur and proximal tibia perpendic- ular to the longitudinal axis of the Knee Arthrodesis 156 Journal of the American Academy of Orthopaedic Surgeons limb prior to fixator application. Compression of the apposed bone surfaces is applied both during sur- gery and in the postoperative period. The patient is instructed to remain in partial weight bearing until evi- dence of early healing is seen on follow-up radiographs; then is al- lowed to proceed to full weight bear- ing. No additional support other than the external fixator is needed. After frame removal, a lightweight cast may be applied for 6 to 8 weeks until union is achieved. Uniplanar External Fixation A variety of external fixator con- structs is used to achieve successful Table 1 Modes of Fixation and Rates of Knee Fusion* Study No. of Cases Predominant Indication for Fusion Mode of Fixation Fusion Rate † (%) External fixators Charnley and Lowe 1 171 None post-TKA Charnley clamp 99 Figgie et al 26 27 All post-TKA, all rheumatoid, 22 septic 16 ex fix, 2 dual plating, 1 IM nail, 8 casts 70 Rand et al 7 28 All post-TKA, 25 septic Ace-Fischer fixator (DePuy, Warsaw, Indiana) 71 Hak et al 6 36 22 septic TKA, 9 loose TKA, 3 posttraumatic OA, 1 Charcot neuropathy, 1 TB 19 single plane ex fix, 17 biplanar 58 single plane, 65 biplanar Knutson et al 28 91 All post-TKA, 66 septic, 25 aseptic mechanical failures 49 Hoffmann frames, 40 Charnley clamps; also IM nails, staples, Rush pins 50 Plating Pritchett et al 10 26 9 posttrauma, 6 post-TKA, 5 DJD, 2 neuropathic, 4 unstable Tension-band plating 100 IM nails Vlasak et al 12 32 All post-TKA, 18 septic, 8 loose aseptic, 6 nails post-ex fix failure 19 IM nail, 13 ex fix 100 IM nail, 38 ex fix Enneking and Shirley 13 20 All posttumor resection IM nail with cortical struts 80 Donley et al 14 20 8 septic TKA, 1 loose, 3 giant cell, 2 failed ex fix, 4 unstable IM nail 85 Puranen et al 29 33 15 post-TKA, 5 unstable, 4 post- trauma, 2 septic, 2 TB, 5 fusion nonunion IM nail 88 Modular nails Waldman et al 16 21 All post-TKA, septic Modular IM nail (Neff nail) 95 Arroyo et al 17 21 16 tumors, 5 failed TKA Modular IM nail (Neff nail) 90 Combined modalities Knutson et al 30 20 All post-TKA, 15 septic, 4 loose, 1 unstable 10 IM nails, 10 ex fix 85 Fahmy et al 31 20 10 failed Charnley clamp, 5 post-TKA, 2 posttrauma, 3 neuropathic joints Charnley clamp with IM nail 100 * Studies reporting on a minimum of 20 cases. † The fusion rate indicated is the initial fusion rate and does not include knees that required other procedures (eg, bone grafting) to achieve union. DJD = degenerative joint disease, ex fix = external fixation, IM = intramedullary, OA = osteoarthritis, TB = tuberculosis, TKA = total knee arthroplasty James H. MacDonald, MD, et al Volume 14, Number 3, March 2006 157 arthrodesis. Cunningham et al 4 re- ported on uniplanar unilateral frame use in 10 patients; successful union was achieved in 9. Because anterior placement increases rigidity, the fix- ator is applied anteromedially or anterolaterally, depending on the position of the skin incision. Two or three 6-mm Schanz pins are used on either side. Although the uniplanar external fixator provides the bend- ing stiffness and the compression re- quired for fusion, medial-lateral sta- bility is less optimal. Biplanar External Fixation Biplanar fixators use transfixion pins in the femur and tibia, supple- mented by additional Schanz pins placed anteriorly (Figure 2). The an- terior pins may be placed through the arthrodesis incision or through separate incisions. 5 In one series, bi- planar fixators used in 17 patients achieved a fusion rate similar to that obtained with uniplanar fixators in 19 patients. 6 However, complica- tions with the biplanar fixator in- cluded a 39% primary nonunion rate (reduced to 25% after revision sur- gery), a 17% infection rate at pin site, one stress fracture through a pin site, and one persistent infection, re- sulting in above-knee amputation. Circular External Fixation Circular frames increasingly are used in knee arthrodesis because of ease of application, excellent stabil- ity, and versatility in terms of limb alignment and ability to bear weight in the early postoperative period. Advantages include continuous axi- al compression and the ability to easily make adjustments for align- ment. These frames can be applied in the presence of active infection, and bone grafting is not required. 32 The main disadvantage of the circu- lar external fixator is the bulkiness of the frame. Typically, two rings each are used for the tibia and femur; alternative- ly, one ring and one arch may be used for the femur instead of two rings. The rings are connected by threaded rods, which can be used to provide compression across the fu- sion site. Preoperatively, the frame is assembled and sterilized. The posi- tion of the rings is planned based on radiographic assessment. A gap of 10 cm between the distal femoral ring and the proximal tibial ring al- lows access to the wound when needed. During surgery, compres- sion can be applied by the fixator across the fusion site and the amount of compression can be ad- justed postoperatively. Weight bear- ing as tolerated is allowed immedi- ately after surgery. David et al 8 reported solid fusion at a mean of 28 weeks in all 13 pa- tients in whom the Ilizarov external fixator was used for failed knee ar- throplasty. Patients were treated for an average of 19 weeks in the fixator, followed by 9 weeks in a plaster cyl- inder cast. Recurvatum deformity was corrected in one patient by add- ing a hinge, a modification that exemplifies the versatility of the sys- tem. Oostenbroek and van Roer- mund 32 reported on 15 infected TKAs treated with arthrodesis using the Ilizarov ring fixator; union was achieved in all but one patient. Aver- age patient age was 75 years, average time in the frame was 28 weeks, and mean leg-length discrepancy as a re- sult of bone loss was 4 cm. The com- plications reported were osteitis in two patients, frame instability in two, nonunion in one, and, as a re- sult of poor bone stock, fracture of the femur and tibia after frame re- moval in one. In another study, a hy- brid frame was used successfully in two patients with an infected knee arthroplasty after a single-stage Figure 2 Biplanar fixator. (Adapted with permission from Brooker AF, Hansen NM Jr: The biplane frame: Modified compression arthrodesis of the knee. Clin Orthop 1980;6 0:163-167.) Knee Arthrodesis 158 Journal of the American Academy of Orthopaedic Surgeons débridement and fixator applica- tion. Average time to union was 10 weeks. 33 However, 19 patients treated by a ring fixator had a fusion rate of only 68% after an average of 4.5 months and an overall complication rate of 84%. 9 The problems included super- ficial pin-tract infection in 55% and nonunion in 32%. Despite the high rate of pin-tract infection, no actual pin loosening was observed. It is im- portant to note that a circular or uni- lateral fixator is preferable to inter- nal fixation in the presence of highly virulent or polymicrobial infec- tion. 34 Plate Fixation Compression plate osteosynthe- sis, used for stabilization of the knee fusion, can also achieve rigid fixa- tion of the fusion mass. Compres- sion between the distal femur and proximal tibia may be obtained, en- hancing the potential for successful fusion and subsequent alignment of the lower extremity. In the early postoperative period, external im- mobilization typically is not re- quired, and partial weight bearing may be started. A drawback to inter- nal plate fixation is that recurrent or new infection following plate osteo- synthesis may be difficult to man- age. In these cases, removal of hard- ware is required, followed by an alternative means of stabilization to gain a solid knee fusion. Pritchett et al 10 reported on 26 pa- tients whose knee arthrodeses were managed by a single tension-band plate technique. The plate was ap- plied on the anterior aspect across the fusion site; by positioning the knee in slight flexion, the plate ab- sorbed the tension stress while the compressive forces were transmitted through the bone. Six patients had a failed prior TKA. Three other pa- tients had an active infection in the joint, with drainage and inflamma- tion; serial débridement had been used to manage the patients with in- fected TKA before arthrodesis. Bone graft and external support were not used. An anterior midline approach was used for the arthrodesis and the plating, resulting in fusion clinical- ly in 6 to 12 weeks and radiological- ly by 16 weeks, without major com- plications. Patients were allowed immediate partial weight bearing. In another report, dual plates were used in 11 patients after failed TKA. 11 One plate was applied on the medial and the lateral side, and each was contoured to approximate the anteromedial and anterolateral sur- face of the tibia, respectively. Two patients had bone grafting, with the patella either fixed to the trochlear notch or used as bone graft. A cylin- der cast was applied with weight bearing as tolerated until healing was evident on radiographs. All pa- tients had fusion at an average of 5.6 months, although one patient had a femoral stress fracture and an- other, persistent infection. Intramedullary Fixation With a Long Nail Since it provides rigid fixation, an intramedullary nail can be used to stabilize the fusion site until bone union. This is accomplished best with insertion of an interlocking nail antegrade from the piriformis fossa, down the length of the femur, across the knee joint and into the distal tibia. Intramedullary nailing (Figure 3) produced a fusion rate of 95% com- pared with 64% for external fixation in one series. 35 Intramedullary nail- ing also is advantageous because it is a surgical technique familiar to many surgeons. Nailing is especially Figure 3 Anteroposterior (A) and lateral (B) views of a long antegrade intramedullary nail. (Adapted with permission from Windsor RE: Knee arthrodesis, in Insall JN, Scott WN [eds]: Surgery of the Knee, ed 3. New York, NY: Churchill Livingstone, 2000, pp 1103-1116.) James H. MacDonald, MD, et al Volume 14, Number 3, March 2006 159 useful in patients with substantial bone loss because intercalary bone graft can easily be placed alongside the rod. Internal fixation allows pa- tients to more easily mobilize post- operatively, and the strength of the construct allows early weight bear- ing. Disadvantages of intramedullary nailing include prolonged surgical time and increased blood loss. 14 An- other limitation is that the compres- sion exerted across the fusion site can be applied only with weight bearing. When the canal diameters of the femur and tibia are different, achieving optimum fixation with a uniform diameter device may be dif- ficult, although custom implants with dual diameters can be designed. Contraindications to intramedullary nailing include the presence of an ip- silateral hip replacement prosthesis, malunited femoral shaft fracture, gross deformity of the femoral shaft, and active infection. Persistent infection following nailing is a potential hazard, but many studies have shown a low risk of infection after intramedullary nailing for knee arthrodesis. 29,36,37 Control of infection is important before nailing. In a series of 20 pa- tients (8 failed TKAs), the average time necessary to clear infection after prosthesis removal was 12 months. 14 For antegrade nail insertion with an interlocking nail, the patient is positioned supine, and the ipsilater- al hip is elevated for easier exposure. The desired length of nail is deter- mined preoperatively from long- standing anteroposterior radiographs of the extremity. Adequate débride- ment and preparation of the distal fe- mur and proximal tibia is performed. Because the tibial canal is narrower than the femoral canal, the tibial ca- nal may be reamed first to judge the best nail diameter. The piriformis fossa is entered, and a ball-tipped guide wire is inserted, as is done in antegrade femoral nailing for femo- ral fracture. Following reaming, the nail is inserted with the bow of the nail kept anteromedial to provide valgus and flexion at the knee. The nail is advanced to within 2 cm of the ankle joint, with locking screws used to prevent proximal migration or rotation of the nail. A study of 33 patients (15 failed TKAs) treated by an antegrade long nail reported union in 29 patients at 3 to 4 months; 29 17 of these patients required less assistance in walking than previously after arthrodesis. In addition, intramedullary nailing may be used as a salvage procedure after failed external fixation. 12 The average time between removal of the fixator and insertion of the nail in the series of Vlasak et al 12 was 26 weeks, and no patient developed an infection from the previous pin sites. Following nailing, the axis of tib- ia and femur coincide, and, despite positioning the curvature of the nail, there is loss of normal knee valgus. However, the shor tening resulting from bone loss after knee prosthesis removal may obviate the need for flexion. Therefore, a position of full extension is recommended for fu- sion following failed TKA; this posi- tion is thought to provide a more nearly normal gait. 38 As a result of the altered mechanical axis of the limb, a possibility of increasing stress on the ankle joint exists, but this has not been found to be a prob- lem. Potential complications of in- tramedullary fixation include migra- tion and breakage of the intramedul- lary device, neurovascular injury, tibial fracture during insertion, and delayed union. 35 In addition, fracture of the tibia distal to the nail has been reported. Complications specific to a long intramedullary nail include glu- teal pain and proximal migration of the nail. 39 Vascularized Fibular Graft and Allograft In cases that involve large bone defects or segments of avascular bone, vascularized fibular grafts have been used with femoral nailing for knee arthrodesis. 27 Management op- tions include shortening the extrem- ity or using allografts, vascular bone grafting, or amputation. The nail is inserted first. The ip- silateral fibula is harvested next, ei- ther on a pedicle or as a free graft, and is used to bridge the defect. Fem- oral defects ≥10 cm usually need free transfer of fibula, whereas smaller femoral defects and most tibial de- fects can be managed by fibular graft on a pedicle. 27 The fibular graft is used as an onlay graft, or is placed in a trough proximally and distally, and is fixed with a proximal and distal plate and/or screws. In a series of 13 patients, 8 with tu- mor resection, 4 with failed infected TKA, and 1 with severe complicated rheumatoid arthritis, 12 had a suc- cessful fusion. 27 One patient had an amputation for persistent infection. This procedure is important to con- sider in cases with large bone defects but is technically demanding, with an average operating time of 8.3 hours and an average blood loss >2 L. 27 Intercalary allograft is a recon- struction option for large defects, such as those following tumor resec- tion. In a study of 39 patients, the graft was stabilized after placing it in the defect with an intramedullary nail passed in a retrograde tech- nique. 40 To create rotational stability, two patients had an additional plate fixation at the graft-host junction. Union was obtained on both ends of the graft in 32 patients, and 5 non- unions healed following bone graft- ing. Three patients had above-knee amputation—two for chronic non- union and one for local recurrence. Intramedullary Fixation With Modular and Nonmodular Nails Two categories of nails can be in- serted through the knee: modular and nonmodular. Examples of mod- ular intramedullary nails (Figures 4 and 5) include the Wichita and the Knee Arthrodesis 160 Journal of the American Academy of Orthopaedic Surgeons Neff. 17 A single incision is used. Fol- lowing bone preparation, the femo- ral part of the nail is inserted retro- grade into the femoral canal, and the tibial half of the nail is inserted an- tegrade into the tibial canal. The two ends are then connected at the level of the knee joint by a conical couple and secured with locking screws. Different types of modular nails are available, and it is critical that surgeons become familiar with the details specific to each before inser- tion. Modularity of femoral and tib- ial components improves fit in situ- ations in which a discrepancy between the canal diameters exists. A curved femoral rod allows place- ment of the knee in slight valgus, when desired. When there is no ma- jor bone loss, patients are allowed to bear weight as tolerated immediate- ly after surgery. In cases of major bone defects, weight bearing is de- layed until radiographic signs of healing are evident. In a series of 21 patients (6 with failed TKA) treated with a modular nail, fusion was achieved in 90% at an average of 8.4 months. 17 The au- thors reported a 38% complication rate, which included three stress fractures, three peroneal nerve pal- sies, one wound infection, and one reflex sympathetic dystrophy. For extensive bone loss around the knee, 2- and 4-cm spacers are available, and bone grafting can be done at the same time. Another series reported successful fusion using a modular nail in 20 of 21 patients at a mean of 6.3 months. 16 An example of a nonmodular nail is the Huckstep intramedullary nail, a solid titanium nail with a quadrilat- eral cross-section design and multi- ple locking holes throughout its en- tire length. 41 The required length is measured and cut preoperatively. Be- cause of the ability to use multiple locking screws, a shorter length of nail provides adequate stability. It is possible to position the knee in slight valgus and flexion with this device. In addition, the nail can be used in the presence of the femoral stem of an ipsilateral hip replacement. The nail is inserted retrograde into the femur and guided antegrade into the tibia, using the screw holes for purchase. When the middle of the nail is at the level of the knee joint, tibial interlocking is done, using the visible screws to align the jig. Then the apposing ends of the femur and tibia are impacted together, and fem- oral interlocking is done using the jig. 41 Four to six screws are inserted in both the femur and the tibia. Be- cause the nail fits loosely in the fe- mur, it is possible to position the knee in slight valgus. No external support is required postoperatively, and weight bearing as tolerated is al- lowed. In one series of 31 patients with infected TKA and 2 with Charcot joints, union was achieved in 30 of 33 knees (91%), with an average time to union of 5.2 months. 41 Prob- lems associated with the nail in- clude screw backing out, screw breakage, and nail breakage, the lat- ter reported in 4 of 22 patients treat- ed with the Huckstep nail. 42 Regardless whether a modular or nonmodular nail is chosen, the ad- Figure 4 Modular nail. (Adapted with permission from Arroyo JS, Garvin KL, Heff JR: Arthrodesis of the knee with a modular titanium intramedullary nail. J Bone Joint Surg Am 1997;79:26-35.) Figure 5 Anteroposterior (A) and lateral (B) radiographs demonstrating the Neff nail in situ. James H. MacDonald, MD, et al Volume 14, Number 3, March 2006 161 vantage of the nails inserted through the knee is the ability to gain in- tramedullary fixation without an in- cision near the hip and without plac- ing the hip at risk should sepsis occur. The major problem with these nails is the difficulty in remov- al, which may require an anterior cortical window and burrs in order to cut the nail into multiple pieces. Results of Knee Fusion Fusion rates in knee arthrodesis range from 38% to 100%, depending on the method used (Table 1). In- tramedullary nail fixation typically has a higher fusion rate, especially after failed TKA, whereas infected knees have a lower fusion rate, as do knees with rheumatoid arthritis. The fusion rate after TKA is lower than that of knees that have not un- dergone TKA; this lower rate most likely is a result of increased bone loss, poorer bone apposition, persis- tent infection, or limb shortening. 9 A Swedish study that included mul- tiple surgeons with multiple tech- niques treating 91 failed TKAs, dem- onstrated that the fusion rate in community practice may not be as high as that reported by some cen- ters. Of 91 knees, fusion occurred in the 10 in which an intramedullary rod was used, whereas external fixa- tion for failed TKA had a failure rate ≥50%. 28 Complications and Disadvantages The most frequent complication of arthrodesis is pain resulting from nonunion. Causes of nonunion in- clude deficient bone stock, inade- quate fixation, persistent infection, and lack of solid osseous contact. 17 Established nonunions should be treated as in fracture care: atrophic nonunions with iliac crest bone graft or vascularized fibula graft, hyper- trophic nonunions with more rigid fixation. A failed external fixator may be converted to an intramedul- lary rod; however, this conversion should occur in two stages to avoid spreading pin-tract contamination through the medullary canal. Other complications, regardless of the method of immobilization chosen, include intraoperative fracture, deep vein thrombosis, continued drainage from infection, new infection, wound dehiscence, and peroneal nerve palsy. Supracondylar fractures have been reported in above-knee fu- sion; these fractures can be managed with antegrade locking nails. A po- tential long-term complication of knee arthrodesis is a biomechanical effect on the hip, spine, and ankle as a result of an altered gait. Conversion to TKA Patients should understand preoper- atively that, in most cases, a knee ar- throdesis cannot be converted to TKA at a later date. Many authors have reported high complication rates associated with converting a fusion to TKA. Henkel et al 43 de- scribed seven cases of TKA after formal knee fusion; two patients un- derwent secondary refusion for in- fection and ligamentous instability, respectively, six of seven underwent revision surgery—for arthrolysis (three patients), a regional soft-tissue flap for skin necrosis (two), and neurolysis of peroneal nerve (one). The mean range of motion achiev- ed was 74° at a mean follow-up of 56 months, and the mean Hospital for Special Surgery (HSS) score im- proved from 54 to 68. A similar report of 30 patients showed final flexion achieved was 75.8°, with an average extensor lag of 9°. 44 There were no revisions for loosening at 5.3 years mean follow- up; 53% of patients had skin edge necrosis. The HSS score improved from 60 to 73. Hu 45 reported a 53% complication rate in 17 cases of TKA following formal knee fusion; two patients were converted back to fu- sion for patellar tendon loss and in- fection, respectively. The mean range of flexion achieved was 84°. However, a recent study of 36 pa- tients undergoing a posterior stabi- lized knee prosthesis at an average of 24.5 years after knee fusion showed an improvement in HSS knee score from a mean of 60 to 83.2. 46 The av- erage patient age was 39.2 years; the average follow-up was 7.7 years. Summary Techniques for obtaining stability for knee fusion have evolved to meet a challenging common indication, the failed TKA. Studies demonstrate good results with intramedullary fix- ation, either with a long antegrade nail through the piriformis fossa or with a coupled nail inserted through the knee. Circular fixators also are gaining popularity as experience with these devices increases. Causes of fusion failure include persistent infection, inadequate bone contact, insufficient immobilization, or inad- equate bone stock. Multiple tech- niques are available for knee arthro- desis, each with its own advantages and disadvantages. The treating sur- geon should be well versed in the ad- vantages and disadvantages of these techniques before selecting the ap- propriate method. References Citation numbers printed in bold type indicate references published within the past 5 years. 1. Charnley J, Lowe HG: A study of the end-results of compression arthrode- sis of the knee. J Bone Joint Surg Br 1958;40:633-635. 2. Nelson CL, Evarts CM: Ar throplasty and arthrodesis at the knee joint. Orthop Clin North Am 1971;2:245- 264. 3. Hagemann WF, Woods GW, Tullos HS: Arthrodesis in failed total knee re- placement. J Bone Joint Surg Am 1978;60:790-794. 4. Cunningham JL, Richardson JB, Sori- ano RMG, Kenwright J: A mechanical assessment of applied compression and healing in knee arthrodesis. Clin Orthop Relat Res 1989;242:256-264. Knee Arthrodesis 162 Journal of the American Academy of Orthopaedic Surgeons 5. Brooker AF Jr, Hansen NM Jr: The bi- plane frame: Modified compression arthrodesis of the knee. Clin Orthop Relat Res 1981;160:163-167. 6. Hak DJ, Lieberman JR, Finerman GAM: Single plane and biplane exter- nal fixators for knee ar throdesis. Clin Orthop 1995;316:134-144. 7. Rand JA, Bryan RS, Chao EYS: Failed total knee arthroplasty treated by ar- throdesis of the knee using the Ace- Fischer apparatus. J Bone Joint Surg Am 1987;69:39-45. 8. David R, Shtarker H, Horesh Z, Tsur A, Soudry M: Arthrodesis with the Ilizarov device after failed knee arthro- plasty. Orthopedics 2001;24:33-36. 9. Garberina MJ, Fitch RD, Hoffmann ED, Hardaker WT, Vail TP, Scully SP: Knee arthrodesis with circular exter- nal fixation. Clin Orthop 2001;382: 168-178. 10. Pritchett JW, Mallin BA, Matthews AC: Knee arthrodesis with a tension- band plate. J Bone Joint Surg Am 1988;70:285-288. 11. Nichols SJ, Landon GC, Tullos HS: Arthrodesis with dual plates after failed total knee arthroplasty. J Bone Joint Surg Am 1991;73:1020-1024. 12. Vlasak R, Gearen PF, Petty W: Knee arthrodesis for failed total knee re- placement. Clin Orthop 1995;321: 138-144. 13. Enneking WF, Shirley PD: Resection- arthrodesis for malignant and poten- tially malignant lesions about the knee using an intramedullary rod and local bone grafts. J Bone Joint Surg Am 1977;59:223-236. 14. Donley BG, Matthews LS, Kaufer H: Arthrodesis of the knee with an in- tramedullary nail. J Bone Joint Surg Am 1991;73:907-913. 15. Stiehl JB, Hanel DP: Knee arthrodesis using combined intramedullary rod and plate fixation. Clin Orthop 1993; 294:238-241. 16. Waldman BJ, Mont MA, Payman KR, et al: Infected total knee arthroplasty treated with arthrodesis using a mod- ular nail. Clin Orthop 1999;367:230- 237. 17. Arroyo JS, Garvin KL, Neff JR: Arthro- desis of the knee with a modular tita- nium intramedullary nail. J Bone Joint Surg Am 1997;79:26-34. 18. Wiedel JD: Salvage of infected total knee fusion: The last option. Clin Orthop Relat Res 2002;404:139-142. 19. Leopold SS, Greidanus N, Paprosky WG, Berger RA, Rosenberg AG: High rate of failure of allograft reconstruc- tion of the extensor mechanism after total knee arthroplasty. J Bone Joint Surg Am 1999;81:1574-1579. 20. Wolf RE, Scarborough MT, Enneking WF: Long term followup of patients with autogenous resection arthrode- sis of the knee. Clin Orthop 1999; 358:36-40. 21. Windsor RE: Knee arthrodesis, in In- sall JN, Windsor RE, Scott WN (eds): Surgery of the Knee. New York, NY: Churchill Livingstone, 1993, pp 1103- 1116. 22. Kim YH, Kim JS, Oh SW: Total knee arthroplasty in neuropathic arthropa- thy. J Bone Joint Surg Br 2002;84:216- 219. 23. Falahee MH, Matthews LS, Kaufer H: Resection arthroplasty as a salvage procedure for a knee with infection af- ter total arthroplasty. J Bone Joint Surg Am 1987;69:1013-1021. 24. Vince KG: Revision knee arthroplasty technique. Instr Course Lect 1993; 42:325-339. 25. Younger AS, Duncan CP, Masri BA: Surgical exposures in revision knee arthroplasty. J Am Acad Orthop Surg 1998;6:55-64. 26. Figgie HE III, Brody GA, Inglis AE, Sculco TP, Goldberg VM, Figgie MP: Knee arthrodesis following total knee arthroplasty in rheumatoid arthritis. Clin Orthop Relat Res 1987;224:237- 243. 27. Rasmussen MR, Bishop AT, Wood MB: Arthrodesis of the knee with vascular- ized fibular rotatory graft. J Bone Joint Surg Am 1995;77:751-759. 28. Knutson K, Lindstrand A, Lidgren L: Arthrodesis after failed knee arthro- plasty: A nationwide multicenter in- vestigation of 91 cases. Clin Orthop 1984;191:202-211. 29. Puranen J, Kortelainen P, Jalovaara P: Arthrodesis of the knee with in- tramedullary fixation. J Bone Joint Surg Am 1990;72:433-442. 30. Knutson K, Lindstrand A, Lidgren L: Arthrodesis after failed knee arthro- plasty: A report of 20 cases. J Bone Joint Surg Br 1985;67:47-52. 31. Fahmy NR, Barnes KL, Noble J: A technique for difficult arthrodesis of the knee. J Bone Joint Surg Br 1984; 66:367-370. 32. Oostenbroek HJ, van Roermund PM: Arthrodesis of the knee after an in- fected arthroplasty using the Ilizarov method. J Bone Joint Surg Br 2001; 83:50-54. 33. VanRyn JS, Verebelyi DM: One stage debridement and knee fusion for in- fected total knee arthroplasty using the hybrid frame. J Arthroplasty 2002;17:129-134. 34. Rand JA: Alternatives to reimplanta- tion for salvage of the total knee ar- throplasty complicated by infection. J Bone Joint Surg Am 1993;75:282- 287. 35. Damron TA, McBeath AA: Arthrode- sis following failed total knee arthro- plasty: Comprehensive review and meta-analysis of recent literature. Orthopedics 1995;18:361-368. 36. Wilde AH, Stearns KI: Intramedullary fixation for arthrodesis of the knee af- ter infected total knee arthroplasty. Clin Orthop Relat Res 1989;248:87- 92. 37. Bose WJ, Gearen PF, Randall JC, Petty WI: Long-term outcome of 42 knees with chronic infection after total knee arthroplasty. Clin Orthop 1995;319: 285-296. 38. Vahvanen V: Arthrodesis in failed knee replacement in eight rheuma- toid patients. Ann Chir Gynaecol 1979;68:57-62. 39. Levine M, Rehm SJ, Wilde AH: Infec- tion with Candida albicans of a total knee arthroplasty: Case report and re- view of the literature. Clin Orthop 1988;226:235-239. 40. Weiner SD, Scarborough M, Vander Greind RA: Resection arthrodesis of the knee with an intercalary allograft. J Bone Joint Surg Am 1996;78:185- 192. 41. Lai KA, Shen WJ, Yang CY: Arthrode- sis with a short Huckstep Nail as a sal- vage procedure for failed total knee ar- throplasty. J Bone Joint Surg Am 1998;80:380-388. 42. Abe S, Tateishi A, Tokizaki T, Takeyama S, Nakano H, Matsushita T: Mechanical failure of unsupported Huckstep intramedullary nail in ma- lignant bone tumor reconstruction. Clin Orthop Relat Res 2001;393:272- 278. 43. Henkel TR, Boldt JG, Drobny TK, Munzinger UK: Total knee arthro- plasty after formal knee fusion using unconstrained and semi-constrained components: A report of seven cases. J Arthoplasty 2001;16:768-776. 44. Kim YH, Kim JS, Cho SH: Total knee arthroplasty after spontaneous os- seous ankylosis and takedown of for- mal knee fusion. J Arthroplasty 2000;15:453-460. 45. Hu C: Results of total knee arthro- plasty following takedown of formal knee fusion. J Arthroplasty 1996;11: 732-737. 46. Kim YH, Oh SH, Kim JS: Conversion of fused knee with use of a posterior stabilized total knee prosthesis. J Bone Joint Surg Am 2003;85:1047- 1050. James H. MacDonald, MD, et al Volume 14, Number 3, March 2006 163 . March 2006 155 cular disease, and corticosteroid use, are evaluated preoperatively because of their potential effect on wound healing. In addition, a complete pe- ripheral neurologic examination is performed. a knee immobilizer or a cylinder cast, simulating a knee fusion, will dem- onstrate for the patient potentially problematic activities, such as the use of public transportation and the increased energy. mass. Compres- sion between the distal femur and proximal tibia may be obtained, en- hancing the potential for successful fusion and subsequent alignment of the lower extremity. In the early postoperative