Purpose: We studied the effect of rhBMP-2 in patients with open leg fractures to determine the impact on the number of revision procedures and on late bone healing or nonunion.

Material and method: Four hundred fifty patients with an open tibial shaft fracture that could be treated with a stratified nail (Gustilo-Anderson) were included in the study. Patients were randomly assigned to three treatment arms: control, with rhBMP-2 0.75 mg/ml, and with rhBMP-2 1.5 mg/ml. The proteins were carried on a biodegradable collagen sponge. The rhBMP impregnated sponge was placed on the wound in contact with the fracture after reduction and nailing. A dynamic or locked nail was used, with or without reaming.

Results: Follow-up data were available for 93% of the patients at 12 months after nailing. Compared with the control group, the number of reoperations for delayed healing was lower in the rhBMP-2 groups (p = 0.0017). Results were better in the 1.5 mg/ml group (−44%, RR=0.56, 95CI = 0.40-0.78, p=0.0005). The number of major reoperations (bone grafts new nailing) was considerably reduced (−49%, p = 0.0264). Between the 10th and 52nd week, the proportion of patients with a healed bone was significantly higher in the 1.5 mg/ml group than in the control group. At six months, 58% of the patients treated with 1.5 mg/ml had healed, compared with only 38% in the control group. Mean delay to healing was significantly lower in the 1.5 mg/ml group compared with controls (Kaplan Meier, p=0.022) and mean delay to healing in 50% of the patients was 145 days, compared with 184 days. Rate of infection was similar in the three groups, but there were significantly fewer infections in the 1.5 mg/ml group patients with a grade 3 fracture than in controls (p=0.0219). There was also a lower rate of fixation material failure in the 1.5 mg/ml group (p=0.0174). Anti rhBMP-2 antibodies (< 6%) or anti-collagen bovine antibodies (< 20%) were observed without presence of anti-human collagen antibodies and without any clinical expression or apparent effect on the clinical outcome.

Conclusion: At the dose of 1.5 mg/ml, rhBMP-2 associated with centromedullary nailing significantly improved outcome, with fewer reoperations for late healing and fewer major reoperations. Fracture healing was accelerated and rate of infection was lower in patients with the most severe fractures.