We observed a revision-free survival of 80.7% (95% confidence interval, 71.7–90.8) at 5 years and 75.9% (95% confidence interval, 65.6–87.8) at 10 years (Fig. ). When revision secondary to allograft failure alone was considered, the 5- and 10-year survival rate was 87.4% (Fig. ).
The Kaplan-Meier 5- and 10-year survivorship rates with revision for any reason were 80.7% (95% confidence interval, 71.7–90.8) and 75.9% (95% confidence interval, 65.6–87.8), respectively.
The Kaplan-Meier 5- and 10-year survivorship rates with revision secondary to allograft failure only were 87.4% (95% confidence interval, 75.9–96.1) and 87.4% (95% confidence interval, 75.9–96.1), respectively.
Sixteen patients underwent revision for failed reconstructions. The average time to failure was 42 months (range, 1–68 months) (Table ). Five failures occurred secondary to infection, comprising 7.1% of the study group. Two of these patients had a history of infection, two had local wound problems at the time of revision surgery requiring muscle flap or skin grafting, and two required above-knee amputation as definitive treatment. Two of the infected revisions contained large segment bulk allograft, two contained femoral head allograft, and one contained femoral head and segmental allograft. Of these 16 failures, eight related to the allografts. Seven of the failures were complications secondary to resorption, including instability, implant loosening, and component failure. Only one patient had allograft nonunion and subsequent rerevision was needed at 42 months (Fig. ). Of the eight allograft failures, three occurred with large segmental allografts and were secondary to allograft nonunion, allograft fracture, or knee dislocation secondary to allograft resorption. Five failures occurred in the smaller femoral head allografts and all of these were secondary to resorption (Fig. ). At 30 months followup, the patient had allograft resorption with subsequent tibial tray fracture. Three nonallograft failures occurred secondary to tibial loosening in two patients and patellar loosening in one; all allografts in these cases were intact, united, stable, and did not require replacement or revision at the time of revision surgery.
Summary of the 16 failures requiring revision
Fig. 4A–C The radiographs show (A) a periprosthetic femur fracture that was treated with (B) revision TKA using a distal femoral allograft. (C) The patient’s radiograph 42 months after revision shows a symptomatic distal femoral allograft nonunion (more ...)
Fig. 5A–B A 63-year-old patient experienced failure secondary to allograft resorption 18 months after femoral head allografting to the tibial plateau. (A) Allograft resorption is evident in the medial tibial plateau. (B) A radiograph obtained 30 months (more ...)
For the 40 patients with a minimum 5-year followup, the average postoperative Knee Society score was 87 (range, 37–100) compared with 49 (range, 0–84) preoperatively. The average range of motion (ROM) increased from 73° preoperatively to 101° postoperatively. Two patients had a decrease in Knee Society score compared with preoperatively.
For the 32 patients (33 knees) with adequate radiographs and minimum 5 years followup, two had progressive radiolucencies and two had asymptomatic allograft-host nonunion. The remaining 28 patients (29 knees) had no progressive radiolucencies at last followup.
Thirteen patients (13 knees) had 15 complications not requiring revision (Table ). Two patients had ligamentous laxity develop that was treated with bracing; neither of these patients wanted a repeat operation. Two patients had a flexion contracture develop. One was treated with manipulation with improvement in ROM; the other had severe rheumatoid arthritis and declined additional treatment. Intraoperative patellar fracture occurred in two patients and intraoperative tibial fracture in one patient. These were treated adequately without internal fixation and by protected weightbearing. A superior pole of the patella fracture occurred 3 years postoperatively in one patient that healed with knee immobilizer treatment. The fracture healed and was stable intraoperatively at a future revision. One patient with pulmonary embolus had no long-term complications after appropriate anticoagulation. A distal femoral stress fracture through host bone (nonallograft) occurred 2 years postoperatively and healed with a brief period of immobilization. Chronic swelling suspicious for low-grade infection developed in one patient; he currently is asymptomatic on long-term antibiotic suppression. Three patients had inadequate soft tissue coverage requiring skin grafting in one case and skin grafting with a muscle flap in two cases. The patient requiring skin grafting had an infection develop 11 months postoperatively that was treated with above-knee amputation. One of the two patients requiring muscle flap coverage who had an infection in the immediate postoperative period had an above-knee amputation. The third patient had a good result with pain relief, increased ROM, and activity level. Seven of the 15 complications occurred in patients eventually having rerevision.
Complications of revision TKA with allograft
For the 11 patients who died before 5 years followup, the Knee Society score had improved an average of 36 points from preoperatively to last followup. All allografts were radiographically intact at last followup, Nine additional patients were lost before 5 years followup at an average of 42 months. These patients had, on average, a 24-point increase in Knee Society score at last followup. Three of these nine patients had clinical worsening of the Knee Society score compared with preoperatively.