Voir la version française

Introduction

Treatment of complex displaced fractures of the proximal humerus is challenging. Management of a comminuted fracture in poor quality bone with a conventional fixation device is difficult,13 not to mention the potential risk of post-traumatic necrosis of the humeral head,10 nonunion or malunion of the tuberosities.2 In such situations, Neer advocated the use of hemiarthroplasty11. But what was a success in Neer's hands was not in others' hands and many surgeons were disappointed by a high rate of nonunion of the tuberosities.2 Percutaneous fixation devices do not allow early mobilization and often have poor holding power in osteoporotic bone. The ideal solution would be a device that allows anatomical reconstruction, provides good stability at the fracture site, allows early mobilization, and preserves bone stock. Actually, intramedullary systems meet these requirements; locking nails offer good stability, and fixation of the tuberosities is provided by the proximal locking screws. The good results achieved by Cuny with his Telegraph nail5 have not been confirmed by other users.1 There is another intramedullary implant called "Bilboquet" which has been designed by Doursounian who reported his preliminary results in 2006. In view of the device advantages,9 we decided to start using it in 2000. The aim of this study is to present our results with the "Bilboquet" in the treatment of complex proximal humeral fractures.

Figure 1: Duparc Classification.

Figure 2: Bilboquet implant (Stryker France).

MATERIALS & METHODS

We have studied retrospectively the first series of 33 Bilboquet devices implanted in 32 patients. There were 26 females and 6 males. Mean age of patients at injury was 67 years (range, 22-88 years). Fractures were classified by both the Neer12 and the Duparc classification systems (Fig. 1). For the preoperative radiological evaluation, standard radiographs and a CT scan were used for clear identification of the fracture type. Postoperatively, one patient died of problems which were unrelated to the fracture or the surgery, 2 were lost to follow-up, and an additional 2 patients had such a bad general condition that it precluded evaluation of the outcome. This left us with 28 shoulders (27 patients) available for analysis. There were 14 three-part fractures according to the Neer classification, and 14 four-part fractures including 4 CT2, 7 CT3 and 3 CT4 according to the Duparc classification. The Bilboquet implant (Stryker France) designed by Doursounian was used in all the patients. The name "Bilboquet" was inspired by the old Janik-Gosset technique. The Bilboquet consists of two components: one male, one female. The male component is a stem with a short neck and a 135° neck-shaft angle; the stem is fixed in the diaphysis with bone cement. The female component is a sort of cylindrical staple with 5 spikes for impaction into the cancellous bone of the humeral head (Fig. 2). A Morse taper connection is used for assembly of the two components. Should avascular necrosis (AVN) occur later on, the humeral head can be resected and the female component (staple) removed; the Morse taper of the humeral stem can accommodate a prosthetic humeral head.

The same surgical protocol was used in all the patients: the patient was placed in a beach-chair position and the procedure was performed using a deltopectoral approach. The tuberosities were identified and tagged with sutures. They were slightly retracted to afford access to the cancellous bone of the humeral head. Great care was exerted to maintain vascularity of the femoral head and preserve the integrity of all attachments. The cartilage was never exposed. The appropriate staple size was that which provided optimal coverage of the cancellous bone. The staple was impacted in a central position under image intensifier control. The humeral shaft was widely exposed to allow insertion of the humeral stem. Then, the staple and the stem were connected. A trial reduction was performed to check soft tissue tensioning, reduce the tuberosities, and evaluate length and version of the humerus. The humeral shaft was exposed again. During insertion of the humeral stem, care was taken to achieve the proper retroversion and height as previously determined with the trial. Then, the implant was cemented. The two components were connected and the tuberosities secured with sutures which were passed around the prosthetic neck as described by Boileau.3 Postoperatively, the limb was immobilized with elbow close to the body, and gentle passive motion was started on day 3. Patients were reviewed at a mean of 14 months. Clinical evaluation was based on Constant score, range of motion, and patient satisfaction. For the radiographic evaluation, several parameters were analyzed: height position of the tuberosities, staple position relative to the center of the humeral head (Fig. 3), healing status, and absence of AVN.

Figure 3: Staple position is accurately determined A: Morse taper axis D: Diameter intersects the outline of the articular surface at 90° AD distance > 3 mm = staple is off-centered

Table 1: Results and AVN per type of fracture (includes 1 CT3 and 1 CT4 converted to hemiarthroplasty)

Table 2: Results according to the presence or absence of AVN AFE: Active Forward Elevation

Figure 4: Sample case of AVN

Figure 5: X-rays showing the outcome of a three-part fracture treated with the Bilboquet

Results

Complications:

There were four complications in this series. Dissociation of stem/staple taper connection occurred three days after the procedure in a multiply injured patient who suffered an epileptic seizure. The patient was reoperated on and developed such a massive periarticular ossification that now his shoulder joint is almost fused. Infection in one patient was successfully treated by lavage and appropriate antibiotic therapy with the implant in situ. In two patients, the development of avascular necrosis in four-part fractures (one CT3, one CT4) required conversion to hemiarthroplasty; therefore, these patients were not included in the final analysis.

Mean Constant score was 59.7 (range, 13-87) and mean weighted score was 84.1% (range, 19-120%). Mean ROM values were 113° (range, 70-160°) active forward elevation, 30° (range, -20-80°) external rotation (with elbow close to body), and internal rotation to the L1 level. Fourteen patients claimed to be very satisfied with their operations, 10 were satisfied, 2 were disappointed. Weighted Constant score and active forward elevation were significantly higher in three-part fractures than in four-part fractures. There was no difference in pain, activity, and strength scores between the various types of fracture. However, the Duparc classification showed a significant difference between CT2 and CT3 (Table 1). Both Constant score and ROM were higher in patients who had been admitted to rehabilitation centers, but the difference was not significant.

Radiographic results:

Avascular necrosis (AVN) was radiographically demonstrated in 4 cases (Fig. 4) and suspected in 2. In three-part fractures, it was demonstrated in 1 patient and suspected in 1. In four-part fractures, it was demonstrated in 3 patients and suspected in 1. According to the Duparc classification, there was no AVN in CT2 fractures. AVN clearly affected the functional results (Table 2).

Nonunion of the greater tuberosity occurred in one patient only. Malunion occurred in two patients due to healing of the greater tuberosity in too low a position without this affecting the Constant score. In most patients, near anatomical reconstruction was achieved (Fig. 5). The staple was judged off-centered in 10 patients, too low in 6, and too high in 4. AVN was more frequent when the staple had been improperly positioned too low.

Discussion

The benefits of this type of implant in complex proximal humeral fractures include a relatively good functional outcome with a Constant score of 60 and a weighted score of 84%, mean active forward elevation of 113°, a fairly high satisfaction rate, and proper healing of tuberosities in all the patients but one.

These results are similar and even better than those reported by Doursounian6, but due to our short follow-up we cannot predict the rate of avascular necrosis in the medium and long run. Another point is that these cases were done at the beginning of our experience, during the learning curve. Now we have standardized the surgical protocol, an image intensifier is routinely used, and the indications are clearly defined. The use of the image intensifier is critical for determination of the correct position of the staple in the vertical plane: well centered, not too high or too low. Centering of the staple in the horizontal plane should be determined from CT scan images to ensure correct version of the humeral head. Better tuberosity healing is achieved with the Bilboquet than with hemiarthroplasty.2 This may be attributed to retention of a well-vascularized humeral head as opposed to placement of a metal head in hemiarthroplasty. In total, there were 6 radiographically demonstrated necroses and 2 suspected ones; four patients were not reoperated on, two cases were converted to hemiarthroplasties. Avascular necrosis significantly prejudices the outcome. In this respect, contrary to what is reported in the literature, we believe that in the presence of AVN, the Bilboquet will not ensure a good outcome even though an anatomical reconstruction has been achieved.8 One of its major advantages is that it allows conversion to hemiarthroplasty where necessary. But the decision must be made quickly, before the spikes of the staple damage the articular surface of the glenoid. One of the drawbacks of the Bilboquet is the difficulty to remove the implant should infection occur. We had one infection in this series that did not require removal of the implant. When removal is necessary, treatment is the same as that for an infected prosthesis. It is the reason why it shoud be reserved for the aged. Another important point is the classification of fractures. The Neer classification allows to differentiate between three-part and four-part fractures. In the Duparc classification, four-part fractures are further divided into 4 subtypes, which refines assessment and makes it possible to differentiate CT2 fractures which have a good outcome in the absence of avascular necrosis from CT3 and CT4 fractures which have a poorer prognosis. This of course must be taken into account when selecting the best treatment method.

Conclusion

Longer follow-up is needed to better evaluate the efficacy of this implant. Nevertheless, it can be said for sure that it provides stable fixation and good healing of the tuberosities, and that it yields satisfactory short-term results. Should avascular necrosis occur, it will be readily managed with a standard procedure. Three-part fractures and CT2 four-part fractures in the elderly appeared to be the indications of choice for the Bilboquet. As for CT3 and CT4 four-part fractures, due to the potential risk of avascular necrosis associated with these fractures, we preferably used hemiarthroplasty.