Comparison of extended anterolateral approach in treatment of simple/complex tibial plateau fracture with posterolateral tibial plateau fracture

We retrospectively analysed the patients with posterolateral tibial plateau fracture treated by the extended anterolateral approach at our hospital between January 2013 and December 2015. The exclusion criteria were as follows: open fracture, > 3 weeks between the injury and the initial operation, presence of a pathological fracture, vascular or nerve injury, and knee arthritis history.

A total of 46 patients were included in the study and were followed up for 23–45 months (mean, 31.9 months). Twenty patients were male, and 26 were female, aged 29–77 years, with an average of 53.9 years. According to Schatzker classification, 24 patients of Schatzker type II fracture were divided into group A, and 18 patients of type V fracture and 4 patients of type VI fracture were divided into group B. According to the three-column classification, 6 cases involved the posterior column, 18 cases involved the lateral and posterior columns, 6 cases involved the medial and posterior columns, and 16 cases involved three columns. The causes of injury were tumbles in 28 cases (60.7%), traffic accidents in 14 cases (30.4%), falls in 2 cases (4.3%), and direct impact in 2 cases (4.3%). The injury mechanism was flexion valgus injury in 30 cases (65.2%), flexion varus injury in 4 cases (8.7%), extension injury in 6 cases (13%), and flexion injury in 6 cases (13%). Twenty-two patients had a fibular fracture, and 6 had associated injuries (3 vertebral fractures, 1 skull fracture and pelvic fracture, 1 pelvic fracture, and 1 ankle fracture) (Table 1).

Routine preoperative examinations consisted of plain radiography and CT scan. The injured limb was fixed with plaster or calcaneal pin traction temporarily, and the surgical treatment time was based on the condition of the soft tissue. The injury mechanism, surgical approach, and treatment method of the fracture were determined by the fracture characteristics [15].

We set the patient in the supine position on an ordinary operation table. In group A, we usually used a single extended anterolateral approach to reduce and fix the lateral plateau fracture with a posterolateral fracture. In group B, the tibial plateau fracture included the medial and lateral plateaus’ fractures, so we used two incisions consisted of a medial incision and an extended anterolateral approach to make the treatment. In group B, we did the medial incision first. Usually, the medial plateau fracture was a metaphyseal fracture, and the articular surface was intact without any compression; there was no need to open the medial joint capsule. We only need to make sure that the medial tibial cortex is anatomically reduced. After the X-ray confirmed that the medial plateau was reduced, we used a pre-bending reconstruction plate (Synthes GMBH, Zuchwil, Switzerland) to fix the medial tibial plateau, and then the following extended anterolateral approach would be made to reduce and fix the lateral plateau fracture with a posterolateral fracture.

We made the extended anterolateral approach as follows. Firstly, we put the knee joint in modest flexion with a bump underneath the knee joint. A 10- to 15-cm-long curvilinear incision centred over Gerdy’s tubercle was made. After the skin incision, subcutaneous dissection was made approximately 2 cm along the skin incision. Then, a fascial incision was made along the same line as the skin incision. The iliotibial band was split in the middle along the direction of the fibres and sharply elevated from Gerdy’s tubercle anteriorly and posteriorly. Then, the fascial incision was extended down to the crural fascia. The dissection was extended posteriorly by taking down the extensor muscles from the lateral surface of the lateral plateau to the point right in front of the fibular collateral ligament (FCL). By this point, we flexed the knee joint further up to 90° to relax the fibular collateral ligament and the common peroneal nerve. Then, a retractor was placed on the inferior part of the FCL to retract the FCL posterolaterally. The interval between the FCL and the posterolateral surface of the lateral plateau (para-FCL space) was developed by dissecting loose soft tissues from the posterolateral plateau. If we did not use arthroscopy treatment, we directly opened the joint capsule and retracted the lateral meniscus superiorly to make visualization of the joint surface. If we used arthroscopy treatment, we made two arthroscopy incisions and did not open the articular capsule, and then we made the visualization of the joint surface with the arthroscopic monitor.

The next step depended on the type of fracture being treated. When the fracture involved compression of the joint surface and a displaced lateral split, the split portion of the fracture was opened like a book to gain access to the compressed fragments. These fragments were then reduced to the joint surface. After the C-arm X-ray checked the reduction, we used Kirschner wires for temporary fixation. Then, the split was closed, reduced, and lateral locking plate (Synthes GMBH, Zuchwil, Switzerland) was placed. The plate was placed as posteriorly as possible and the transverse arm stretched to the supra-fibular-head space. In such situation, at most two rafting screws can support the articular surface of the posterolateral column. When a split fracture was nondisplaced or hinging it open would require extensive dissection of the intact bone and periosteum, we used the “open-window and reduction rod” technique for fracture reduction. We always used this method when arthroscopy was assisted. A cortical window was created on the anterolateral tibial metaphysis 3–5 cm below the articular surface. The depression of the articular surface could be elevated using a rod to restore the congruence of the articular surface. We confirmed the fracture was reduced on the C-arm X-ray or arthroscopic monitor. After reduction, Kirschner wires and then the plate (Synthes GMBH, Zuchwil, Switzerland) were placed as previously introduced. We would check reduction quality, implant location, and screw lengths by C-arm X-ray or arthroscopic monitor at the end of surgery again.

Postoperative drainage routinely continued for 24 h, and antibiotics were routinely administered for 48 h to prevent infection. Passive joint function activity was performed immediately after the operation, and partial weight training was started 2 weeks after the operation under the doctor’s guidance. Patients were followed up every 3 months, and knee joint pain, joint activity, and plain radiography were tested each time. The operation time, blood loss, fracture healing time, fracture reduction, knee function score, knee range motion, and postoperative complications of each patient were noted. According to the standard clinical and radiological criteria to judge the healing time of the fracture, the Rasmussen score system [16] was used to evaluate knee joint function, and the method proposed by Biggi et al. [17] was used to evaluate fracture reduction. Whether the posterolateral fracture fragment was fixed was evaluated by CT scan (Fig. 1).

The operation time, blood loss, fracture healing time, knee Rasmussen score, and range motion were compared between group A and group B by the independent t test. The fracture reduction and the posterolateral fracture fragment fixation were compared between group A and group B by the chi-square test (Fisher’s exact test). P < 0.05 was considered significant. SPSS software (22nd edition, SPSS, Chicago) was used to record and analyse the study results.

The operation time was 55–150 min (average 124 min) in group A and 110–300 min (average 175 min) in group B; the amount of bleeding was 20–400 ml (mean 118 ml) in group A and 50–500 ml (mean 190 ml) in group B. Both made significant differences (P < 0.05). In group A, patients were treated with a single anterolateral plate (Fig. 2); In group B, patients were treated with medial and lateral plates. Twelve patients received arthroscopy treatment (Fig. 3) including six patients in each group. Patients with lumbar fractures were treated conservatively, patients with ankle fracture were treated with surgery, and patients with pelvic fractures underwent staged surgery. All patients had achieved bony union at the last follow-up. The average fracture healing time was 4.58 months (3–6 months) in group A and 5.54 months (4–8 months) in group B, and there was no significant difference (P > 0.05). Fourteen patients removed the internal fixation at the last follow-up.

All patients experienced no significant influence on daily life. They could walk on the floor and up and down the stairs and engage in light physical labour. In group A, the reduction was excellent in 20 cases (83.3%), good in 2 cases (8.3%), and poor in 2 cases (8.3%), and the excellent and good rate was 91.7%. Four posterolateral fracture fragments were not completely fixed (16.7%). The knee Rasmussen score was 22–28 (average 26.8), and knee range motion was 100–120° (average115.5°). In group B, the reduction was excellent in 6 cases (27.3%), good in 12 cases (54.5%), and poor in 4 cases (18.2%), and the excellent and good rate was 81.8%. Eight posterolateral fracture fragments were not completely fixed (36.4%). The Rasmussen score was 20–28 (average 23.5), and the knee range motion was 95–115° (average 106.6°) in group B. The excellent and good rate of reduction had no significant difference (P > 0.05) between the two groups, but the excellent rate of reduction had a significant difference (P < 0.001). The unfixed rate of the posterolateral fragment showed no significant difference nor did the Rasmussen score or knee range motion (P > 0.05) (Table 2).

One patient in group B suffered postoperative wound infection. The lateral plate was removed, and negative vacuum drainage was performed. After three sessions of negative vacuum drainage, the wound healed, and finally, the fracture healed. No internal fixation failure, no postoperative compartment syndrome, and no other postoperative complications occurred.

This was a retrospective study, and all patients were from our hospital, so the collected data had a greater chance of bias. MRI examinations were not done in most patients; thus, we might have failed to accurately describe the ligaments of the knee joint and the soft tissue injury. Finally, we had no control group for comparison with other approaches.