A locking compression plate as an external fixator for treating infected nonunion of the humeral diaphysis

Between June 2010 and August 2014, a total of seven patients with infected nonunion of the humeral diaphysis were treated at our department using an LCP as an external fixation. Cases without infection and those with infected nonunion of proximal or distal humeral fracture were excluded from this study. The average age of the patients at presentation was 40.9 years (range 26–61 years). Five patients were male and two were female. Of these seven cases, six were closed humeral diaphyseal fractures, and one was an open fracture (Gustilo grade II). The AO/OTA classification of these patients is listed in Table 1. One case was accompanied by radial nerve injury. The initial operations of all patients (open reduction and internal fixation) were carried out in their respective local hospitals. The anterolateral approach was applied in six out of seven patients; the posterior approach was used in the case accompanied by radial nerve injury, which was repaired by end-to-end anastomosis. Erythema, swelling, purulent discharge of the affected upper limb, and implant failure were observed in all patients before revision surgery. Six out of seven patients presented with definite infection caused by known bacteria as diagnosed by germiculture. The mean time from fracture to revision surgery was 7 months (range 3–15 months). The patient data are described in Table 1.

All patients were retrospectively evaluated clinically and radiographically. The functional results were evaluated according to the Disabilities of the Arm, Shoulder and Hand (DASH) scoring system [19]. Complications evaluated included nonunion, deep infection, pin tract infection, implant failure, and limb shortening. Radiographs were obtained at regular intervals: at the time of admission, immediately postoperatively, at postoperative 1, 3, 6, and 12 months, and at the final follow-up. Union was defined as the presence of a mature, bridging callus of three to four cortices seen on radiography, with the simultaneous absence of implant loosening or breakage and absence of pain during weight-bearing. Nonunion was defined as the absence of radiological signs of union 9 months postoperatively, without any tendency toward progressive union. Deep infection was defined according to clinical symptoms (such as erythema, swelling, and presence of purulent discharge) and laboratory examination results (such as total leukocyte count, C-reactive protein level, and erythrocyte sedimentation rate).

Under general or regional anesthesia, six of the seven patients were placed in the supine position with an arm table, and one patient was placed in the contralateral decubitus position because the posterior approach was used in the first surgery. The involved limb was prepared and draped in the usual standard sterile fashion, and the quondam incision was chosen (anterolateral in six cases, and posterior in one). All surgeries were performed by the same orthopedic surgeon (Chongqi Tu). The radial nerve was first explored and marked with a catheter. Care was taken not to damage the radial nerve during the entire procedure. In cases with an LCP still in situ, the LCP was removed. A thorough debridement was then performed, including resection of avascular bone, excision of sinus tracts and infected scarred soft tissue, collection of representative tissue culture samples, and opening of the medullary canal using a drill. Cortical bleeding was taken as an acceptable sign of vital tissue procurement [20]. After complete debridement, the nonunion site was reduced and aligned. Both fracture ends were pruned for better reduction and bony contact, regardless of the occurrence of bone defection. The fracture site was stabilized using temporarily placed Kirschner wires or 3.5 mm screws; otherwise, reduction was temporarily maintained by the assistant with two Kocher clamps.

Once the fracture was acceptably reduced, an LCP was applied as an external fixator. A suitable LCP position was carefully considered before implanting screws. The LCP was placed over the anterolateral side in the six patients that were operated on via the anterolateral approach, whereas the LCP was placed over the posterior side in the patient in whom the posterior approach was used. To ensure that the plate was matched to the bone, we performed temporary positioning by drilling one 2.0 mm Kirschner wire into the cortex of the shaft at the most proximal hole of the LCP and another similarly at the most distal hole. A stack of evenly folded towels was then provisionally placed as a spacer under the plate, separating the plate from the skin (Fig. 1). To optimize mechanical stability, the LCP was placed as close to the bone as possible, while still allowing for potential swelling of the soft tissue. Because mechanical stability decreases as the distance between the plate and bone increases [16, 21], we chose relatively long plates with at least three 4.5 mm screw holes on both sides of the fracture site, except in cases where the salvaged LCP was reused. Generally, for the patients with an LCP in situ, the removed LCP was soaked in a container with povidone iodine for more than 30 minutes and then reused as the external fixator, with new screws being applied. For those without an LCP in situ, a 4.5 mm LCP (Weigao Orthopaedic Device Co., Ltd., Shandong, China) was applied as the external fixator. The first screw was implanted in the most distal hole, ensuring bicortical fixation was achieved; extreme care was taken when implanting the distal screws. We operated carefully under direct vision to avoid damaging the radial nerve. The second screw was then implanted in the most proximal hole. When alignment was deemed satisfactory fluoroscopically, the remaining screws were implanted in the usual fashion. The position and orientation of screws were reassessed under fluoroscopy.

Screw tracts were sterilized with iodophor or 75 % alcohol three times per day. Intravenous antibiotics were administered for 3–4 weeks according to antimicrobial susceptibility testing for tissue specimens or pus harvested intraoperatively (vancomycin in one case; cefoperazone sodium and tazobactam sodium in six cases); no bacterial growth was detected in one case. Third-generation cephalosporin oral medication was then administered for 4–6 weeks. The patients were encouraged to perform initial pendulum and elbow flexion-extension exercises from postoperative day 2. Functional use of the limb for light tasks was allowed when patients could tolerate it, usually from 6 weeks postoperatively.