We retrospectively reviewed the charts and radiographs of 44 patients who underwent surgery for acute traumatic tibial plateau fractures between September 2005 and June 2006. We excluded four patients who were lost to followup, leaving 40 patients for the study cohort. Surgery was performed within 7 days of the injury in all cases by a fellowship-trained trauma surgeon (DGL). The average age of the patients was 53 years (range, 17–83 years). Nineteen patients were younger than 60 years old. There were 22 males and 18 females in the study. We classified the patients into two groups: those who received intraoperatively the osteoinductive agent BMP-2 (BMP-positive group; n = 17) and those who did not (BMP-negative group; n = 23). The minimum followup in the BMP-positive group was 12 months (average, 18 months; range, 12–25 months). The minimum followup in the BMP-negative group was 13 months (average, 18 months; range, 13–26 months). All information was obtained from the medical records; no patients were recalled specifically for this study. We had prior approval from our Institutional Review Board.
We used a Type I collagen sponge (Medtronic, Sofamor Danek, Memphis, TN) as the carrier of the recombinant human BMP-2 for insertion into the fracture site per the manufacturer’s guidelines. A 7.5- × 10-mm absorbable collagen sponge of Type 1 derived from bovine Achilles tendon was used a surgical matrix. A standardized manufacturer’s protocol was used to constitute BMP with sterile water. Freeze-dried, cancellous fibular allograft was trimmed to the required shape and was placed in the metaphyseal bony defect. After hemostasis was achieved, the collagen sponge with BMP was used as an onlay, covering the allograft and bridging the fracture fragments. Early on in the series, we did not use BMP in the treatment of tibial plateau fractures. Rather, after en bloc elevation of the articular cartilage, we used subchondral bone void fillers. The choice of the bone void filler depended on the quality of the metaphyseal bone and the size of the defect after elevation of the articular cartilage. Allograft fibula was our void filler of choice when a sizable structural defect was present. Norian bone cement (Norian Corp, West Chester, PA) and demineralized bone matrix were used more frequently early on in the series. Since the indications of different bone void fillers evolved over the course of the study period, it is difficult to assess the true isolated effect of BMP after controlling other variables, including the choice of other void fillers and the quality of the bone; however, we have reported the subchondral bone integrity when BMP was used. Of the 17 patients in the BMP-positive group, four patients received demineralized bone matrix, two patients received calcium phosphate bone cement, and 11 patients received fibular allograft for structural support.
All patients were followed up at 1, 3, 6, and 12 months and yearly thereafter. Radiographs were made at every followup visit. We recorded union, additional surgeries, and other complications, including reoperation rates for removal of HO around the knee, from the medical charts. We also noted the use of other bone-grafting agents, such as calcium phosphate bone cement, demineralized bone matrix, allograft fibula, and autografts.
Three of us (SB, DH, MW) independently assessed radiographs for the presence and site of HO and subchondral bone settling. The amount was quantified using length and breadth measurements of the fragments on magnification-standardized anteroposterior radiograph film. We noted fragments of HO located on the lateral side in the iliotibial band area but not on the medial side of the tibia because the surgical approach through which the collagen-soaked BMP sponge was placed was from the lateral side and any HO there would not likely relate to the sponge. We deemed this appropriate because there was very little chance, if any, the BMP would have seeped to the medial side. To identify postoperative subchondral bone settling we used joint line congruity as a marker. That is, we compared the articular surface depression between the immediate postoperative radiographs and the radiographs at the most recent followup. A line was drawn along the tibial shaft axis. Two more lines were drawn along the articular surface of the fractured condyle and the unfractured condyle to intersect the first line. In the immediate postoperative radiographs, the latter two lines typically overlapped. These lines were then drawn on the late postoperative radiographs. The subchondral bone settling was measured by the difference between the two lines along the articular surface for each condyle. In case of bicondylar fractures, the preoperative and postoperative radiographs were overlaid to generate the readings.
We used Fisher’s exact test to compare the differences in HO formation, reoperation rates, and the maintenance of articular congruity between the BMP-positive and BMP-negative groups. To determine whether the age, use of BMP-2, surgical approach, and use of other bone-grafting agents predicted HO, we performed multivariable log-binomial regressions.