Revision arthroplasty is the typical treatment for a failed total knee arthroplasty, but it is not always feasible in knees that have an infection or a Charcot arthropathy2. Thus, alternative techniques must be used. These include antibiotic suppression, débridement, resection arthroplasty, arthrodesis, and amputation12. Arthrodesis is indicated when revision arthroplasty is not advisable because of a poorly controlled infection with a virulent organism, problems with soft-tissue coverage, severe loss of bone stock, a deficient extensor mechanism, extreme ligamentous instability, or a high functional demand in a young patient.
Techniques of arthrodesis include external fixation, intramedullary nailing, and fixation with a plate. Of these techniques, intramedullary nailing has been shown to result in the highest rate of fusion4,12. However, the popular method, in which a long curved Küntscher nail is inserted through the piriformis fossa to the distal part of the tibia, is technically demanding. In addition, a long operative time, a large amount of blood loss, and migration of the nail are common5,6.
Since 1988, we have used a technique in which a Huckstep nail (Downs Surgical, Sheffield, England) is advanced in a retrograde fashion through an incision about the knee. The purpose of the present paper was to describe the technique, outcome, advantages, and complications of arthrodesis with the Huckstep nail after a failed total knee arthroplasty.
*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.
†Department of Orthopaedics, National Cheng Kung University Medical Center, Number 138 Sheng-Li Road, Tainan, Taiwan.
‡Po-Cheng Orthopaedic Institute, 100 Po-Ai 2nd Road, Kaohsiung, Taiwan. E-mail address: w2212@ms14.hinet.net.
We retrospectively reviewed the records of thirty-six patients who had been managed at our institution, between 1988 and 1994, with arthrodesis of the knee with a Huckstep nail because of a failed total knee arthroplasty. One patient died from unrelated causes less than one year after the operation, and two patients were lost to follow-up. Thirty-three patients were followed clinically and radiographically for an average of forty-seven months (range, eighteen to ninety-four months) after the arthrodesis, and they form the basis of this report.
Postoperatively, clinical and radiographic evaluation as well as measurement of the erythrocyte sedimentation rate and the level of C-reactive protein were performed for all patients at two-month intervals until union. Union was defined clinically as stability or no motion of the knee on examination and radiographically as evidence of bridging osseous trabeculae or cortical union of more than 75 per cent of the total cortical apposition, according to the method of Fern et al.
The average age of the patients at the time of the arthrodesis was 68.2 years (range, sixty-one to seventy-eight years). There were nineteen women and fourteen men. The right knee was involved in fifteen patients and the left, in eighteen. The underlying diagnosis before the primary arthroplasty was osteoarthrosis (twenty-seven patients), rheumatoid arthritis (four), and post-traumatic osteoarthrosis (two). The average time-interval between the most recent arthroplasty and the arthrodesis was 4.9 years (range, two to eight years). The indication for the arthrodesis was recurrent infection after a revision arthroplasty (fifteen patients), infection with gross ligamentous instability (eight), infection with poor soft-tissue coverage (five), a Charcot joint (two), and a failed previous attempt at arthrodesis of an infected knee with use of external fixation (three).
Associated medical problems included diabetes (seven patients), a cardiovascular disorder (four), a previous stroke (three), cirrhosis of the liver (two), Parkinson disease (two), and chronic nephritis (one).
All total knee prostheses that had been placed before the arthrodesis were of the non-constrained resurfacing type.
Infections were classified as purulent, chronic, or quiescent (Table I). An infection was considered to be purulent when the patient had a fever as well as distention of the knee and there was gross pus on aspiration. An infection was defined as chronic when there was either a discharging sinus or local heat and erythema but no pus on aspiration. Four patients (Cases 16, 17, 23, and 29) who had negative cultures were considered to have a chronic infection because the histological finding of more than five polymorphonuclear neutrophils in five fields at a magnification of forty times, as well as the elevated erythrocyte sedimentation rate and level of C-reactive protein, were strongly suggestive of infection. An infection was termed quiescent when the patient had a history of infection but no infection was evident at the time of the arthrodesis, as demonstrated by a normal erythrocyte sedimentation rate.
Gram-positive organisms, mostly Staphylococcus, were found in eighteen knees, and gram-negative or mixed infections were found in seven. Cultures of specimens from the other knees revealed negative findings. A two-stage procedure was used for the eight knees that had a grossly purulent infection. Initially, the prosthesis and the cement were removed, the knee was thoroughly debrided, and cement impregnated with gentamicin or vancomycin (depending on the organism) was inserted. Arthrodesis was performed only after the infection appeared to be controlled clinically, the erythrocyte sedimentation rate and the level of C-reactive protein were normal, and there was no growth on culture of fluid aspirated from the knee. The average time from the removal of the prosthesis to the arthrodesis was 5.1 weeks (range, two to fourteen weeks). The twenty-five knees that did not have a grossly purulent infection were treated with removal of the implant and the arthrodesis as a single-stage procedure.
The Huckstep nail is a solid titanium reconstruction nail with a quadrilateral cross section. Transverse interlocking holes 4.5 millimeters in diameter are spaced fifteen millimeters apart along its entire length, and four-millimeter-diameter fine-threaded tapered screws are inserted through these holes. A 12.5-millimeter-diameter nail (the largest size available) was used in all knees.
The femur was carefully measured on preoperative radiographs, and the nail was cut to a suitable length with use of a diamond-edged saw before the operation. A fourteen-hole, 22.5-centimeter-long nail was used in the twenty-four patients in whom the involved femur was less than forty centimeters long, as measured from the tip of the greater trochanter to the line of the knee joint. A sixteen-hole, 25.5-centimeter-long nail was used in the nine patients in whom the involved femur was more than forty centimeters long.
The arthrodesis is performed with the patient in the supine position. A tourniquet is placed on the thigh and is inflated to a pressure of 350 millimeters of mercury (46.7 kilopascals). The arthrodesis is performed through a medial parapatellar approach, as the total knee arthroplasty was. The implant is removed, and a thorough débridement and synovectomy is done. The distal part of the femoral canal then is reamed to a diameter of thirteen millimeters and to a depth of twenty-two to twenty-five centimeters from the knee. This level is distal to the tip of the stem of most standard-size femoral components, so the nail can be used in patients who have an ipsilateral total hip prosthesis. The tibia is reamed from the plateau to a depth of twelve to thirteen centimeters and to a diameter of thirteen millimeters.
A Huckstep nail of appropriate length is inserted gently into the femoral canal in a retrograde manner until only the last screw-hole is visible (Fig. 1, A). The nail is rotated so that the screw-holes are oriented anteroposteriorly and the heads of the screws are medial to the tibial tuberosity and crest. The knee is extended, and then the nail is advanced into the tibia, with use of the exposed screw-holes for purchase, until half of the length of the nail is in the tibia (Fig. 1, B). The targeting jig for the interlocking screws is applied, and four, five, or six drill-bits are used to fix the nail to the tibia temporarily (Fig. 1, C). Because two holes usually are visible in the gap between the femur and the tibia, the jig can be aligned precisely. Also, as the nail is short and stiff, interlocking can be done without fluoroscopy and with use of the jig alone for guidance (Fig. 2). The intervening drill-bits are removed, and the femur is compressed against the tibia. The femoral fit usually is not tight, which allows the knee to be positioned in valgus angulation and slight flexion. Peripheral cortical contact is achieved, and the bone is trimmed as necessary to allow better bone-to-bone contact. The femur and the tibia are held in compression, and four, five, or six drill-bits are placed, again with use of the screw-targeting jig, into the screw-holes on the femoral side (Fig. 1, D). The stability of the site of the arthrodesis is checked, and the drill-bits are replaced with screws of appropriate length as determined with a depth gauge (Fig. 1, E). Four, five, or six screws are placed in both the femur and the tibia.
In the thirty-one knees in which the patella was suitable for bone-grafting, it was fixed with one screw into the femur. Bone graft was taken from the tibial and femoral condyles. Polymethylmethacrylate beads containing gentamicin (Septopal) were placed subcutaneously around the site of the arthrodesis in twenty-three patients. The chain of beads was pulled out at approximately two weeks after the operation because later removal is technically more difficult.
Postoperatively, the patients were allowed weight-bearing as tolerated. External supports were not used routinely. Antibiotics were selected according to the results of sensitivity testing and were given parenterally for an average of 3.8 weeks (range, two to six weeks). The antibiotics were then given orally for the balance of the six-week course.
The angle of the knee fusion was measured on anteroposterior and lateral radiographs made at the time of the latest follow-up (Table II).
The average duration of the operation was 104 minutes (range, sixty-five to 155 minutes) (Table I). None of the knees had an intraoperative fracture, a perforation of the femoral cortex, or a peroneal nerve palsy. Because a pneumatic tourniquet was applied, intraoperative blood loss was negligible. The estimated postoperative blood loss averaged 468 milliliters (range, 150 to 720 milliliters). Seven patients received blood replacement of 500 to 1000 milliliters postoperatively because of anemia, cirrhosis-related coagulopathy, or cardiovascular problems.
Thirty (91 per cent) of the thirty-three knees had clinical and radiographic evidence of union at an average of 5.2 months (range, two to ten months) after the arthrodesis (Table I). Union was not seen radiographically in three patients. The nail broke two years postoperatively in one patient (Case 9) who had had a Charcot joint, and he was managed with repeat nailing and a bone graft. Three years after the second operation, the patient had union and was doing well. In the other two patients (Cases 11 and 21), who had had an infection at the time of the arthrodesis, a non-union was diagnosed on the basis of pain in the distal part of the thigh and a radiolucent zone around the nail. One (Case 21) had had an infection with Staphylococcus aureus, and the other (Case 11) had had a mixed infection with Staphylococcus aureus and Proteus mirabilis. At 3.5 and 4.5 years after the operation, neither patient had had a recurrence of the infection and both declined an additional procedure despite occasional pain in the thigh.
Only one patient (Case 18), who had rheumatoid arthritis and was being managed with immunosuppressive therapy, had a recurrence of infection (with Staphylococcus aureus). The infection was treated successfully with operative débridement, with the nail left in place, and the knee eventually fused.
Screws backed out in five knees (Table I). At least one screw broke in three knees, but the stability of the knee was not affected. Three patients needed removal of at least one screw that had backed out because the screw had become prominent and had resulted in irritation of the skin. The screws that did not back out were not associated with tenting of the skin and irritation. The only nail that was removed was the one that broke.
The angle of the knee fusion averaged 6.0 degrees (range, 0 to 10 degrees) of flexion in the sagittal plane and 3.0 degrees (range, 0 to 6 degrees) of valgus in the frontal plane (Table II; Fig. 3). All patients could walk at least inside the house and could bear full weight on the limb, but four patients had mild pain in the involved knee. The three patients (Cases 9, 11, and 21) with a non-union had pain over the middle part of the thigh, probably secondary to motion of the nail in the distal aspect of the femur. None of the patients who had a successful fusion reported any pain in the thigh. All patients reported inconvenience caused by the stiff knee, particularly when they used the toilet.
The average limb-length discrepancy, measured clinically, was 2.6 centimeters (range, two to four centimeters). All patients had a discrepancy that was either corrected with a shoe-lift or was so slight that the patient considered the shoe-lift to be unnecessary.
The purpose of the present study was to describe the technique, outcome, advantages, and complications of arthrodesis with a Huckstep nail after a failed total knee arthroplasty. Because we retrospectively reviewed the records of the patients in the present study, the functional evaluations before and after the operation were not standardized. In addition, no other group was available for a direct comparison because arthrodesis of the knee is always performed with a Huckstep nail at our institution. Most of our patients who have a failed total knee arthroplasty secondary to infection are managed with a staged reimplantation, but those who have a recurrent infection, gross ligamentous instability, poor soft-tissue coverage, or poor bone stock are considered candidates for arthrodesis.
The Huckstep nail was developed in Australia, in 1967, as a reconstruction nail for severely comminuted femoral fractures9. As far as we know, Huckstep was the first to use this nail in an arthrodesis of the knee. He reported on two patients who had an arthrodesis of the knee with a sixty-centimeter-long nail locked with seven to eleven screws for treatment of a failed total knee arthroplasty. The senior one of us (K.-A. L.) made the decision to use this nail in a retrograde manner after a discussion with Dr. Huckstep at a meeting in 1988.
Goldberg et al. reported the results of arthrodesis with the Huckstep nail to salvage total knee replacements that had failed because of infection. Four of their seventeen patients had a non-union after an average duration of follow-up of 22.6 months, but all of the patients were reported to be satisfied with the result. The rate of non-union in their study, which was much higher than that in our study, may be related to the fact that the patella was not used as a bone graft. It is of note that they inserted a long Huckstep nail in an antegrade manner, a technique that is similar to arthrodesis with use of a long straight Küntscher nail.
It has been recommended that, when a long intramedullary nail (such as a Küntscher nail) is used for arthrodesis, it should be inserted to within two, three, or four centimeters of the ankle5,11. This placement necessitates reaming of the tibia, without tourniquet control, beginning at the greater trochanter and extending distally past the isthmus, and it has been associated with blood loss of more than 1500 milliliters5,6.
Specific contraindications for intramedullary nailing, as described by Knutson and Lidgren, include the presence of a prosthesis in the proximal aspect of the ipsilateral femur, gross malunion of a fracture of the femoral shaft, and marked curvature of the femur. One patient in our series later needed a total arthroplasty of the ipsilateral hip after the arthrodesis of the knee. A femoral component with a sixteen-centimeter-long stem was successfully inserted without removal of the Huckstep nail.
The design of the Huckstep nail and its screw-targeting jig allow the screws to be placed reliably without use of fluoroscopy (Fig. 2). We found the jig to be easy to use, but we acknowledge that many surgeons would prefer to have the position of the screws visually confirmed with fluoroscopy. Multiple interlocking screws provide immediate axial and rotational stability and allow full weight-bearing without the need for walking aids. No nail migrated either proximally5 or distally13 in the present series, and gluteal pain was not an issue as the nail does not traverse that region.
An additional advantage of this technique is the ability to place the knee in the desirable position of slight valgus and flexion (Fig. 3); such a position is very difficult to achieve with use of a long intramedullary nail. Preservation of the normal valgus angulation at the knee theoretically lessens the strain on the ipsilateral hip.
Cheng and Gross performed arthrodesis with a short locking stainless-steel intramedullary nail, with use of an outrigger targeting rod to guide the insertion of the screws, in two patients; both had a successful result. The use of the Neff femorotibial nail system (Zimmer, Warsaw, Indiana) in arthrodesis of the knee recently was described in detail by Arroyo et al. Compared with our technique, their system appears to offer the advantage of modularity; the nails used in the femur and the tibia can have a different diameter. A potential disadvantage is the care that must be taken with regard to the reaming for the conical couples; if the couples are too deep, it may be difficult to join them. In addition, compression is difficult to achieve and the control of rotation is less firm because it relies on the fit in the isthmus rather than on interlocking screws. We have no personal experience with the use of the Neff femorotibial nail system.
A meta-analysis of five reports on the results of different operative techniques of arthrodesis after a failed arthroplasty of the knee in fifty-six patients revealed an overall rate of fusion of 95 per cent for intramedullary nailing compared with only 64 per cent for external fixation4. In another meta-analysis, the overall rate of success of arthrodesis with intramedullary nailing was 91 per cent in ninety-eight knees12. The time to union has averaged 5.5 months (range, 2.5 to eleven months)6. It should be noted that these meta-analyses involved a small number of patients.
Infection of any kind is known to decrease the rate of fusion. Damron and McBeath suggested that the rate of fusion in knees with a gram-positive infection is better than that in knees with a gram-negative or mixed infection. In the present study, the non-unions occurred in a patient who had a Charcot joint, one who had a mixed infection, and one who had a gram-positive infection, but the number of knees with a gram-negative infection was too small for statistical comparison. A one-stage arthrodesis was adequate for patients who had a low-virulence or non-purulent infection. Other authors have reported the same finding7,11. Knees with a grossly purulent infection are treated with a two-stage procedure. Initially, the prosthesis is removed, the knee is debrided, and antibiotic-impregnated cement is inserted. The arthrodesis is attempted only when the infection appears to be controlled, as indicated clinically and by a normal erythrocyte sedimentation rate and level of C-reactive protein as well as no growth on culture of fluid aspirated from the knee. When this protocol is used, infection does not preclude a successful fusion.
The rate of fusion associated with a resurfacing type of implant has been shown to be higher than that associated with a hinged implant because more bone stock is present12. All prostheses in our series were of the resurfacing type, which may account for the high rate of fusion despite the infections and for the relatively small limb-length discrepancies.
The most troubling drawback of arthrodesis with a Huckstep nail is the difficulty of retrieving a nail in the event of a malunion or a recurrent infection. In our series, the only nail that was removed was one that had broken in a patient (Case 9) who had a non-union; the nail was removed by reversing the original procedure. In knees that have a non-union, the nail can be cut and removed in pieces or it can be backed up the femur. We have not removed any nail after a solid fusion has been achieved. Fenestration of the femur may be necessary for removal. A high-speed drill should be available so that the nail can be cut if necessary.
In summary, insertion of the Huckstep nail in a retrograde manner allows use of a tourniquet, which minimizes intraoperative loss of blood. The rate of fusion is high (91 per cent in our study). In addition, the technique provides immediate axial and rotational stability, which permits early weight-bearing without use of external supports, and allows the surgeon to place the knee in a slightly flexed and valgus position. There is no gluteal pain, and the Huckstep nail can be used in the presence of a standard-size prosthesis in the ipsilateral hip. As a surgeon gains experience with the technique, the procedure may be done without fluoroscopy, thus minimizing the exposure of the patient and the surgeon to radiation.