A fifty-three-year-old woman presented with pain and deformity of the left elbow. She had rheumatoid arthritis and had been treated with oral prednisolone and methotrexate for more than fifteen years. The range of flexion of the elbow was from 30° to 90°. Radiographic findings were compatible with severe rheumatoid arthritis.
Because the pain, deformity, and instability were intolerable to the patient, total elbow arthroplasty was carried out with a Gschwend-Scheier-Bähler (GSB)-III prosthesis (Zimmer, Warsaw, Indiana). Postoperatively, the wound did not heal, and, after four weeks, Enterobacter cloacae was grown on culture of specimens from a fistula. The infection was controlled, without removing the implant, by debridement, continuous irrigation, coverage of the fistula with an anconeus muscle flap, and administration of culture-specific intravenous antibiotics.
After one year without additional problems, the patient fell and sustained a fracture of the humerus just proximal to the tip of the prosthetic stem. Contrary to our recommendation, she chose to have initial cast immobilization followed by the use of a fracture brace. However, the fracture remained unstable and gradually created a cutaneous fistula; again, Enterobacter cloacae was grown on culture of specimens from the fistula. Six months after the fall, the humeral and ulnar components were removed and the bone defect was filled with a polymethylmethacrylate bone cement spacer containing piperacillin and gentamicin. Six months later, the spacer was removed and a Coonrad-Morrey total elbow prosthesis (Zimmer) was implanted. This prosthesis was selected because it has a longer humeral stem. The humeral and ulnar components were stabilized with bone cement. As it was our intent to achieve osseous union at the fracture site, care was taken not to fill the nonunion site with bone cement. Kirschner wires were inserted to increase stability at the nonunion site, and the bone defect was filled with iliac crest bone graft. Cultures of specimens from the nonunion site showed no growth.
Two years later, there was no clinical evidence of infection; however, the humerus remained ununited and the humeral stem had penetrated the anterolateral cortex (Fig. 1). Because the bone defect was extensive and the bone quality was poor, we proposed using a strut allograft to treat the persistent nonunion, but the patient refused for religious reasons. Another option was plate fixation; however, we were concerned that a plate would not provide enough stability and could also increase the risk of a new fracture proximally. Therefore, we customized a commercial hollow intramedullary nail to serve as a sleeve that would nest over the stem of the humeral component (Figs. 2-A and 2-B). We verified the length, diameter, and shape of the Coonrad-Morrey long humeral stem with the manufacturer in order to select an appropriately sized inner diameter of a hollow intramedullary nail. Then we measured the length of the humerus. The distal part of the intramedullary nail was cut to an appropriate length and prepared prior to surgery. Specifically, the straight nail system (Nakashima Medical, Okayama, Japan), which had an outer diameter of 10.5 mm and an inner diameter of 8.0 mm, was cut to a length of 170 mm. The prepared nail was inserted into the humerus in an antegrade fashion with use of a deltoid-splitting approach. Then, another incision was made on the anterolateral aspect of the upper part of the arm to expose the nonunion site. The radial nerve was identified and protected. There was a large bone defect proximal to the nonunion site. However, the humeral component was not loose and was tightly integrated with the distal part of the humerus. After the nonunion edges had been freshened, the nail was nested over the humeral stem under direct vision. The distance for which the nail and the stem overlapped was approximately 8 cm. Autologous cancellous bone chips harvested from the iliac crest were then used to fill the bone defect. After the nail had been impacted around the stem, two proximal interlocking screws were inserted. The humerus was shortened approximately 10 mm as a result.
Postoperatively, the arm was immobilized in a u-shaped plaster slab and a sling for a week. Union was achieved after four months. Three years postoperatively, solid consolidation had been achieved with no evidence of loosening around the nail or the elbow implant (Figs. 3-A and 3-B). The elbow and arm were pain-free, with a range of active elbow flexion of 10° to 120°.
Sanchez-Sotelo et al. reported that a periprosthetic humeral fracture after a total elbow arthroplasty could be treated successfully with implant revision and allograft bone struts1. Union was achieved in ten of their eleven patients, and they concluded that this method treated the fracture effectively while maintaining joint function. In Japan, however, some patients refuse allograft bone grafts for various reasons. Under such circumstances, another option is plate fixation with or without an autologous fibular strut graft. Because our patient could barely walk as a result of pain and weakness of the lower limbs due to the rheumatoid arthritis, we were reluctant to use fibular autograft. In addition, the humerus was very osteopenic and we were concerned that plate fixation would lead to another fracture proximal to the plate.
Therefore, we treated the nonunion with a customized intramedullary nail and autologous bone graft. A similar method has been reported for the treatment of periprosthetic femoral fracture following total knee and hip arthroplasties6,7. However, to the best of our knowledge, this method has not been reported for the treatment of humeral shaft fracture, or nonunion, following total elbow arthroplasty. The humeral stem of the prosthesis was nested tightly into the hollow nail to achieve stability during the operation. We call this maneuver the "docking" method. Postoperatively, the humerus united and good elbow function was preserved. We believe that this method is a reasonable option for the treatment of an intractable periprosthetic fracture in a patient with osteoporotic bone.
This method has several drawbacks. First, it is necessary to modify a hollow intramedullary nail to achieve a tight fit over the humeral stem. Second, as the intramedullary nail and the stem are only press-fit to each other, rotational stability may not be adequate. Furthermore, the intramedullary nail is in contact with the humeral stem of the implant. Micromotion between the nail and the stem may create metal debris particles, which could lead to osteolysis locally or even systemic toxicity.