Bicondylar Tibial Plateau Fractures Involving the Posterior Fragment: Injury Mechanism and Fracture Characteristics

Background: Bicondylar tibial plateau fracture (BTPF) involving the posterior fragment is the most complex intra-articular fracture to treat. Although the fracture characteristics of BTPFs have been reported by many researchers, due to the sophisticated kinematics of the knee, reports focused on the injury mechanism are still scarce. The current injury mechanism is insufficient to explain the various fracture characteristics and to guide surgery. This study used a three-dimensional (3D) simulation method to investigate the injury mechanism and fracture characteristics of BTPFs involving the posterior fragment. Methods: Ninety-six BTPFs involving the posterior fragment were included. A computer-assisted 3D technique, which enabled us to move the femoral condyles to match the displaced fracture fragments of the tibial plateau, was applied to analysis the injury mechanism for all cases. The 3D and two-dimensional (2D) morphology of each fracture were reviewed thoroughly, and the main fracture lines were mapped and superimposed on a template. Results: After simulation and quantitatively analysis, three main types (extension, flexion-internal rotation, flexion-external rotation) and six sub-types (valgus and varus in each main type) of injury mechanism had been classified according to the degree of knee motion in three planes. In the extension type, femoral condyles compress the anterolateral tibial plateau and central medial plateau, resulting in an anterior-to-posterior fracture line. In the flexion-internal rotation type, the compressed areas are located on the posterolateral and anteromedial plateau, forming an oblique anteromedial-to-posterolateral fracture line. In the flexion-external rotation type, the compressed areas are located on the anterolateral and posteromedial plateau, exhibiting an anterolateral-to-posteromedial fracture line. Conclusions: Different injury mechanisms result in different fracture characteristics. A thorough understanding of the injury mechanisms underlying complex BTPFs involving the posterior fragment is meaningful not only for fracture characteristics reorganization but also for surgical decision making. The 3D simulation method employed in this study may be a useful supplemental method for investigating the mechanisms underlying fracture injuries.

historically been observed infrequently, but with the development of imaging methods such as computed tomography (CT), posterior fractures of the tibial plateau are increasingly being recognized and noted [1][2][3]. Bicondylar tibial plateau fractures (BTPFs), which are classified as Schatzker type V or VI, are complex articular fractures with multiple fracture lines and fragments, many of which involve posterior fracture fragments [4][5][6][7][8]. Yang [6] reported 93 posterior fractures out of 149 BTPFs, Zhu [7] reported that 44.32% of BTPFs involved posterolateral fractures, and Barei [4] identified 66 posteromedial fractures among 146 BTPFs. Posterior fracture fragments, especially in complex BTPFs, are difficult to stabilize by conventional surgical approaches and are always associated with an inferior functional outcome for patients [5].
Treatment decision making for complex tibial plateau fractures requires two aspects to be considered: the three-dimensional (3D) characteristics of the fracture and, more importantly, the mechanism of injury. Currently, the injury mechanism of BTPF is believed to consist of extreme axial force load on the articular surface [9], and posterior fractures of the tibial plateau results from force in a position of flexion [5,7,10]. However, the mechanism of injury is a matter of speculation based on surgeons' experience and biomechanical experiments, which cannot always explain the various fracture characteristics.
Tibial plateau fractures result from the impact of femoral condyles with the knee joint in various positions in conjunction with a certain deforming force. Reconstructing the position of the knee joint at the time of fracture can provide the opportunity to understand the injury mechanism more precisely and reliably. In this study, we used a computer-assisted 3D technique to simulate the knee position during fracture by moving the femoral condyles to match the displaced articular surfaces of the tibial plateau fracture, and then we analysed the motion of the femur. We hypothesized that the simulation could reveal the injury mechanism of BTPF involving the posterior fragment.

Subjects
This retrospective study was performed with the approval of the institutional review board (IRB Protocol #2019-036-1). One hundred and eighty-three patients with BTPFs treated between December 2016 and November 2019 were identified by searching an orthopaedic database maintained at a level-I trauma centre. Fifty-nine bicondylar fractures without posterior involvement were excluded. Twenty-eight additional fractures had insufficient CT record data available in their Digital Imaging and Communications in Medicine (DICOM) files, and those fractures were also excluded. Patients under 16 years old or with pathological fracture, open proximal tibial fracture, previous knee surgery and/or existing knee ligament malfunction, existing knee deformity, insufficient-quality CT images or axial CT images with a slice thickness > 3 mm were excluded.
Ultimately, a total of ninety-six BTPFs with posterior part fractures, classified as Schatzker types V and VI, were included in this retrospective review.
After the DICOM files of fractures were imported into Mimics software (19.0, Materialise, Leuven, Belgium), 3D models of the tibia and femur were created separately with the "CT Bone Segmentation" command. Then, the tibial and femoral 3D models were adjusted to the fully extended position as the reference position with standard orientations by using the "Pan" and "Rotate" functions as necessary.
For comparative analysis, the right knee 3D models were flipped horizontally to obtain standard left knee models (Fig. 1).

Knee position simulation and analysis
To simulate the knee position at the time of fracture, we used the "reposition" function to move the femoral 3D object to ensure that the articular surfaces of the femoral condyles matched the geometric shapes of the displaced articular surfaces in the tibial plateau fracture. The femur and tibial contour lines in the 2D views were also adjusted to achieve optimal matching in all slices (Fig. 2). The final tibial position was determined by consensus between 2 senior orthopaedic surgeons (S.P. and Y.H.) supervised by the corresponding author (A.P.).
The motion of femur was analysed by the software, and the rotation values in the three planes were recorded and compared. The 3 items in "Rotation" represent the motion of the femur in different planes relative to the standard extension position (reference position) at the time of the fracture injury. A positive value in the sagittal plane represents flexion, and a negative value represents hyperextension. In the axial plane, a positive value represents external rotation, and a negative value represents internal rotation. In the coronal plane, a positive value represents varus, and a negative value represents valgus.

Fracture mapping
The method of two-dimensional (2D) fracture mapping was first described by Cole and colleagues [11] and then modified by Molenaars et al. [12] for fracture line depiction. In the current study, we used the 3D top-view of the fracture instead axial CT slice of the fracture to comprehensively describe the articular fracture lines in fracture mapping.
Top-view images of 3D models of BTPFs were imported into Adobe Illustrator software (2019, Adobe Systems Incorporated, San Jose, CA) on a 2D standard template of an intact left tibial plateau. Specific tibial plateau landmarks such as the medial and lateral tibial plateau, the tibial tubercle, and the fibula were matched to ensure proper rotation and alignment (Fig. 3).
Main fracture lines were drawn on the template by consensus between 2 senior orthopaedic surgeons (S.P. and Y.H.) based on the axial, sagittal, coronal, and 3D model images; disagreements were resolved by discussion with the corresponding author (A.P.). In every case, the main fracture lines are depicted and overlapped, yielding a 2D diagram with variations in density relative to fracture frequency.

Data analysis
All statistical analyses were performed using SPSS software (version 25.0; IBM). The data on age were subjected to one-way analysis of variance (ANOVA), and the data on sex, fracture side, and Schatzker classification were compared using Fisher's exact test. The injury mechanism "Rotation" data from all three planes were compared among the different types and sub-types using the Kruskal-Wallis and Nemenyi tests. A P < 0.05 was considered statistically significant. The fracture characterization and main fracture lines of each type were descriptive in nature.

Result
We reviewed ninety-six CT scans of BTPFs involving the posterior part: thirteen were Schatzker type V and eighty-three were Schatzker type VI tibial plateau fractures. The study subjects were composed of seventy male patients and twenty-six female patients, with an average age of forty-four years (range, twenty to sixty-six years).

Injury mechanism classification
This injury mechanism classification was based on the interpretation of the knee position at the time of fracture, the main types were classified by the motion in sagittal (extension or flexion) and axial plane (internal or external rotation), and the sub-types were classified by the motion in coronal plane (valgus or varus). Three main types (extension, flexion-internal rotation, flexion-external rotation) and six sub-types (valgus and varus in each main type) of injury mechanisms were found in this series of BTPFs involving the posterior part. The baseline demographics are summarized in Table 1 Table 2). The knee position motion data from all the three planes were summarized in Table 3.

Extension injury mechanism
In the extension position, the knee is under a "lock" pattern, with no obvious additional rotation observed between the tibial and femur. When an axial force and a valgus or a varus force act on the knee, the femoral condyles impact the anterior part of lateral plateau and central part of the medial plateau, leading to the formation of the main fracture lines, which were oriented in an anterior-toposterior pattern, and the main fracture fragment. Under a valgus force, the anterolateral tibial plateau fracture pattern is mainly a compressed fracture and lateral displacement; with load transfer, the fracture segment of the posterolateral plateau is split and displaced backward. Under a varus force, the medial plateau fractures are split, and in severe cases, transverse lines can be found extending in the distal direction from the middle part (Fig. 4).

Flexion-internal rotation injury mechanism
In this injury mechanism, the knee is in the flexion and internal rotation position, and the femur exhibits external rotation. The lateral femoral condyle rolls back, impinging the posterolateral quarter of tibial plateau, and the medial femoral condyle moves to the anterior quarter of the medial plateau. In this study, we simulated the position of the knee at the time of the fracture using a computerassisted 3D technique that enabled us to move the distal femoral condyles to match the displaced fracture fragments of the tibial plateau. After simulation for ninety-six BTPFs involving the posterior fragment, we analysed the fracture line orientation, fracture morphology and the motion of the knee in three planes. Three main injury mechanism types and six sub-types were identified and found to exhibit distinct fracture characteristics. We believe that these findings are meaningful for a greater understanding of the complex fractures.
The kinematics of the knee are complex and sophisticated [13,14]; with different knee positions, the femoral condyles impinge different areas of the tibial plateau, resulting in different fracture characteristics (Fig. 3). In BTPFs, the femoral condyles impinged both the lateral and medial plateau, leading to fracture in both locations, so the fracture locations in the lateral and medial plateau are interrelated, not isolated [15,16]. With the extension injury mechanism, the knee joint is extended and stable at the time of injury, and rotation is negligible [17]; in this scenario, the lateral and medial impingement areas are located in the anterolateral and medial central parts of the tibial plateau [18], respectively. Under axial force, impingement results in the classic BTPF, with an anterior-to-posterior fracture line. With flexion motion, the lateral and medial femoral condyles exhibit different trajectories in the axial plane, which form flexion rotation positions, and the maximum ranges of internal and external rotation can reach 45 degrees [19], so the flexion mechanism can divide into two types (internal and external rotation). When the knee is injured during internal rotation and flexion, the lateral femoral condyle rolls back and externally rotates (the knee is internally rotated in relation to it) [20][21][22], and the lateral condyle compresses the posterolateral part of the tibial plateau, while the medial condyle moves forwards and compresses the anteromedial part of the tibial plateau [18,22,23]. A fracture line oriented from anteromedial to posterolateral is observed in flexioninternal rotation-type fractures. By contrast, when the injury occurs during external rotation and flexion, the femoral condyles compress the anterolateral part and posteromedial parts of the tibial plateau, respectively [18,23]. An anterolateral-to-posteromedial fracture line orientation is found in this type of fracture.
Posterolateral fractures are common in BTPFs, and their morphology has been described in many studies [6,8,10,24]; however, few studies have focused on the mechanism of injury. Currently, posterolateral fracture is believed to result from axial and valgus loading with the knee in flexion, and the femur tends to move in the posterior direction [25]. In the current study, 21 BTPFs with posterolateral compression fracture involvement resulted from this injury mechanism with the knee internally rotated. The compression morphology of posterolateral fracture is a result of the direct impingement between the lateral femoral condyle and an osteoporotic posterior lateral tibial plateau.
A different posterolateral fracture morphology, the split type, was observed in 68 BTPFs; this type results from two other injury mechanisms, extension and flexion-external rotation injury mechanisms.
With these two types of injury mechanisms, the lateral femoral condyle moves anteriorly and impinges on the anterolateral part of the tibial plateau; the fracture extends to the posterior part, which results in the split morphology of posterolateral fractures. This finding is consistent with the research of Zhu et al. [7], who also observed these two types of posterolateral fractures with different injury mechanisms in BTPFs and reproduced them in biomechanical experiments. However, in their research, rotational motion was not taken into consideration as a mechanistic factor.
Posteromedial fracture accounts for nearly one-quarter of the surface area of the plateau [4,26], and the morphology usually belongs to the split type due to the concave geometry and bone mineral density of the medial plateau [14,16]. Three types of fracture lines in the medial plateau were observed in this series of BTPF cases, corresponding to the three injury mechanisms. With the extension injury mechanism, the medial femoral condyle impinges on the central part of the medial plateau [14], and the anterior-to-posterior fracture line commonly has a horizontal side branch in medial plateau. With the flexion-internal rotation injury mechanism, the femoral condyle impinges on the anterior part of the medial plateau [14,18,19], resulting in an oblique fracture line (directed from anteromedial to posterolateral) and a large posteromedial fragment. These fracture characteristics have been depicted via a "fracture mapping" method by Molenaars [12], and Yang [6] reported 42 posteromedial plateau fractures with these characteristics. With the knee flexion-external rotation injury mechanism, the femoral condyle compresses the posterior part of the medial plateau [14,18,19], resulting in small fragments and parallel fracture lines. Previous research has deduced that the mechanism involved in this fracture pattern is flexion and internal rotation of the medial femoral condyle (external knee rotation) [4,27], which is consistent with the current findings.
Patzold et al. [28] described the difference in medial plateau fracture line orientation in 81 BTPFs and classified the fractures based on the fracture line angle (sagittal fracture line, coronal fracture line, or no fracture line in the medial plateau). As described by Patzold, in the axial plane, a coronal fracture line (60°-120° relative to the sagittal plane) in the medial plateau fracture type was accompanied by a sagittal fracture line in the lateral plateau, which is consistent with the fracture line morphology resulting from the extension injury mechanism in the current study. The sagittal fracture line type consists of two types of fracture line orientations, 0°-30° and 150°-180°, which can result from flexion-internal and external rotation injury mechanisms, respectively.
The posterior fracture fragments resulting from different injury mechanisms may be fixed with different approaches and strategies by reversing the injury mechanism. For BTPFs with an extension injury mechanism, a semi-flexed position (20-40° flexion) and valgus or varus traction during the reduction procedure are recommended to counter the initial injury mechanism. The posterolateral fragment can be fixed to the anterolateral fragment through an extended anterolateral incision via a single 3.5 mm lateral rafting locking plate [29,30]; a displaced posteromedial fragment always needs a posteromedial approach [3] with a buttress plate to achieve reduction [4,31]. This study has some limitations that must be considered. First, the injury mechanisms found in this study are all based on BTPFs with posterior involvement, which are the most complex fractures to understand, the most challenging to devise surgical strategies for, and always inferior in prognosis.
Therefore, not all tibial plateau fractures can be explained by these mechanisms; we may further investigate other types of tibial plateau fractures in the future. Second, the findings of our research were based on 3D simulations, and one may argue that the interpretation of injury mechanisms is subjective. We believe that further biomechanical investigations will strengthen the evidence presented herein. Third, for this series of fractures caused by high-energy trauma, the ligaments were difficult to observe and reconstruct by MRI due to the serious displacement of the fragments, and so this study did not investigate the mechanisms of ligament injury.

Conclusion
In this study, we investigated the mechanisms of injury in 96 BTPFs involving the posterior part of the joint by using 3D techniques to simulate the knee position at the time of fracture; three main types and six sub-types of recurrent injury mechanisms were ultimately found. Along with these different injury mechanisms, surgeons should pay attention to the differences in fracture morphologies and the main fracture location. Additionally, given the different characteristics of the posterior fragment, the surgical approach and strategy should be chosen according to the injury mechanism. The 3D simulation method employed in this study may be a useful supplemental method for investigating the mechanisms of fracture injuries.      The impingement areas and fracture line in the three main injury mechanism types. The top view images of the 3D tibial models (Fig. 3-A column) were imported on a template. The main fracture line depicted in the template (Fig. 3-B column). Fracture lines mapping in    Fracture characteristics of BTPFs involving the posterior fragment in flexion-external rotation injury mechanism type.