The effect of coronal splits on the structural stability of bi-condylar tibial plateau fractures: a biomechanical investigation

Bi-condylar tibial plateau fractures are challenging traumas due to their complex fracture geometry and accompanying soft tissue injury [1, 2]. These fractures mainly occur in young patients as a result of high-energy trauma and generally require open reduction and internal fixation [3, 4]. Healing complications for these fractures have been reported from 14% up to a staggering 42% [5‐10]. The main goal of operative treatment is the patient’s return to daily activity and functionality, which can be achieved through accurate reconstruction of the knee joint and the anatomical axes [2]. For planning and achieving suitable treatment, it has been recognized that providing a three-dimensional representation of the fracture by CT imaging plays a critical role [1, 11‐14]. This is particularly important for articular fractures in the coronal plane as they are difficult to detect on bi-planar radiographs and complicated to characterize by two-dimensional fracture classifications such as Schatzker or AO/OTA [15]. The AO/OTA and Schatzker classifications are the most common taxonomies of tibial plateau fractures due to their simplicity, while they disregard injury patterns observed in the third dimension [16].

The clinical relevance of these coronal fractures was identified by Barei et al. as a fracture pattern that separates a posteromedial fragment from the tibial plateau [11] and which has a prevalence of almost 50% in complex tibial plateau fractures [1, 11, 17]. The detection of this fracture line is clinically important because lateral locking plates, which are a common fixation method for this fracture, may not effectively stabilize the posteromedial fragment and supplemental implants may be required [5, 6, 11, 18‐20]. To adequately describe the personality of complex tibia plateau fractures, various three-dimensional classification schemes such as the “three-column” concept were developed to refine the traditional planar classifications as a guide for surgical planning [21]. Recently, an extension of the Schatzker classification was introduced in which the fracture type and the mechanism of injury were described based on plain radiographs as well as CT data were utilized to provide complemental third-dimensional information about the location of the main fracture planes [22]. Although the clinical relevance of the posteromedial fragment and the dependency of treatment plans on identifying fracture locations has been recognized, there is still a lack of understanding regarding the biomechanical implications of the posteromedial fragment, in particular with respect to its adequate stabilization [1]. Previous biomechanical studies on bi-condylar tibial plateau fractures have largely been based on the model developed by Horwitz et al. [23] to simulate a Schatzker Type VI fracture. As this model is based on a coronal projection of the fracture, it completely ignores any coronal fracture lines and thus the presence of a posteromedial fragment. Yet, it remains the most frequently employed biomechanical model [23‐35] on which recommendations for fracture fixation of complex tibia plateau fractures are based upon.

Due to the aforementioned negligence of articular fracture lines, establishing a clinically relevant, biomechanical fracture model is required to resolve controversies regarding the ideal fixation method for this complex trauma. The aim of this study was to develop a biomechanical model for bi-condylar tibial plateau fractures, which incorporates a coronal fracture line. We hypothesized that our novel coronal fracture model would exhibit inferior mechanical stability compared to the traditional Horwitz model.

This biomechanical study was performed on synthetic bone analogues which were osteotmized to produce two different fracture models: the traditional Horwitz fracture model (Fracture H) and a novel fracture which was based on a systematic review of CT images [1] and included a coronal fracture line (Fracture C). The fracture models were fixed with locked plating constructs and were quasi-statically as well as cyclically loaded to determine the mechanical stability as measured by stiffness, fragment movement and failure loads.

In static loading, the axial stiffness of Fracture C was 43% lower than that of Fracture H (Table 1, p = 0.013). During cyclic steps, depending on the load level, Fracture C was on average 47–55% laxer than Fracture H (Fig. 7).

To elucidate the effect of the coronal fracture line on fragment stability, relative displacements and rotations were evaluated for the key fracture fragments. Although fragment displacements were evaluated during the whole cyclic testing procedure, we report the displacements after 2500 cycles at the 500-N load level corresponding to 20% of the maximum knee contact force (Fig. 8). For both Fractures H and C, displacements on the medial sites were predominantly in the longitudinal direction followed by frontal and sagittal displacements. In particular, displacement at the medial anterior-shaft gap of Fracture C was almost exclusively in longitudinal direction and exceeded the 5-mm clinical failure criteria (Fig. 8a). Both coronal gaps of Fracture C were mainly displaced in the sagittal direction followed by the longitudinal and frontal movements. The average sagittal displacements exceeded the 2-mm failure criteria at both locations in the plateau (Fig. 8b).

As the displacement measurements indicated that the medial-anterior and medial-posterior fragments of Fracture C moved in different directions, relative rotations of these fragments were analyzed (Fig. 9). Analysis of the sagittal rotation revealed that the coronal split mainly destabilized the medial-anterior fragment which showed larger rotations compared to its corresponding posterior one as well as the whole medial segment of Fracture H (Fig. 9a). Regarding the frontal rotation, unlike the movement of the medial segment in Fracture H, the coronal split resulted in reversing the rotation of both medial fragments toward the posterior direction (Fig. 9b). Finally, for the longitudinal rotation, the coronal split destabilized the medial-posterior fragment, demonstrating a more than threefold increase in internal rotation when compared to its anterior counterpart as well as the medial component of Fracture H (Fig. 9c). For both Fractures H and C, interfragmentary movements of the lateral-shaft and medial–lateral gaps were almost negligible compared to those of the medial-shaft and coronal gaps. Moreover, the movements of these two gaps and relative rotations of lateral fragments were similar in both fracture models.

Shapiro–Wilk’s test (p > 0.05) as well as visual inspection of the histogram, normal Q–Q, and box plots indicated that for both Fractures H and C, the parameters of mechanical stability (i.e., static stiffness, failure load, and failure cycles) were approximately normally distributed. According to the clinical failure criteria, Fracture C failed earlier and at lower load levels compared to Fracture H which tolerated 60% higher load levels (p = 0.016). Failure cycles of Fracture H samples were almost 2.5 times that of Fracture C (Table 1, p = 0.017). The survival analysis revealed that the survival rate of Fracture H was almost 2.5 times that of the Fracture C group. While none of the Fracture C samples survived the clinical failure criteria until 2500 cycles or 500-N load, half of Fracture H specimens survived 2500 cycles (Fig. 10, p = 0.006).

Articular fracture lines in complex tibial plateau fractures which result in a coronal split have been widely ignored in previous biomechanical studies and fracture classifications [11]. Thus, we developed a coronal fracture model based on 3D fracture morphologies [1] and assessed this model biomechanically. Our study revealed that the presence of a coronal split generates an unstable posteromedial fragment which is inadequately stabilized with unilateral plating from the lateral aspect. The coronal split reduced the stiffness and decreased the strength of fracture fixation and resulted in increased fragment displacements during loading.

The biomechanical assessments revealed that Fracture H was almost two times stiffer and tolerated 1.5 and 2.5 times higher load level and cycles, respectively, as compared to Fracture C (Table 1). Also, the survival rate of the Fracture H group was nearly 150% higher than that of Fracture C. Thus, the coronal fracture line remarkably destabilized the fracture fixation construct.

To further understand the effects of coronal fracture lines, the rotations of fracture fragments relative to the tibial shaft and interfragmentary displacements were analyzed. Interfragmentary movements (especially on the articular surfaces) are considered an important clinical parameter because an articular subsidence of more than 2–3 mm remarkably increases knee joint pressure and may result in osteoarthritis [41, 42]. Moreover, malalignment is an important factor in evaluating treatment outcomes of tibia plateau fractures and rotations of joint lines more than 5° are considered fixation failure [37, 43, 44]. Thus, interfragmentary displacements of the medial and plateau fracture gaps were analyzed to assess a risk of bone non-union and osteoarthrosis, respectively. The relative rotations of medial fracture fragments with respect to the tibial shaft are also indicative of limb malalignment. To the best of our knowledge, this was the first study in which 3D displacements of fracture fragments have been evaluated for complex tibial plateau fractures. To build upon previous investigations regarding bi-condylar tibial plateau fractures, which only evaluated the facture fragment subsidences [23‐31, 33, 45] or the frontal intra-articular gap movements [35], we presented a comprehensive 3D kinematic evaluation of all fracture fragments to illuminate the effects of coronal fracture lines.

During cyclic loading, the medial-anterior segment of Fracture C mainly moved in the longitudinal direction and exceeded 5-mm clinical failure which is indicative of a high risk for bone non-union. At the same time, the medial-posterior portion separated from the anterior one in the posterior direction with sagittal displacements exceeding 2 mm which indicates a risk of knee osteoarthritis (Fig. 8).

Due to the vertical nature of the coronal fracture line, the medial segments of Fracture C were subjected to higher shear forces [11]. Therefore, compared to Fracture H, the medial segments of Fracture C, especially medial-posterior one, demonstrated higher instability (Fig. 9). From sagittal rotational disparity between the two medial fragments, it can be concluded that coronal fracture line results in an articular step-off which may lead to knee osteoarthritis. Since sagittal rotations occurred due to screw bending, the coronal split mainly destabilized the anterior fragment perhaps due to the inclusion of fewer screws in comparison to its posterior counterpart. Regarding the frontal rotations, the contact between the load applicator and the coronal edges of the medial fragments in Fracture C resulted in reversing the movement of these fragments compared to Fracture H. For Fracture C in the 500-N cyclic load level, the diagonal screw which is the only connection between the medial fragments and mainly passes through the medial-posterior segment, lost its connection with the anterior one. Thus, medial-anterior fragment rotated in frontal direction, while the medial-posterior part remained stable. Importantly, longitudinal rotations demonstrate that the medial-posterior and medial-anterior fragments of Fracture C tilted away towards the posterior direction of the transverse plane. This separation resulted in higher loading on the medial-posterior part and consequently more longitudinal rotation compared to the medial-anterior one. It should be noted that for Fracture C there is a risk of malalignment in frontal and transverse planes due to exceeding the clinical failure criteria for sagittal and longitudinal rotations of the medial-anterior and medial-posterior fragments, respectively.

These biomechanical findings demonstrated that disregarding the coronal fracture line will result in overestimation of structural rigidity of tibial plateau fracture constructs. Also, the relative displacement of fracture fragments was considerably affected by the presence of a coronal split. While the medial fragment of Fracture H displaced mainly in the frontal plane, the coronal fracture resulted in rotational instabilities of both medial fragments. Therefore, with the presence of the coronal split, there is a risk of non-union at the tibial medial side as well as malalignment of the tibial plateau and eventually osteoarthrosis. With kinematic evaluations in mind, these consequences are likely to occur at lower load levels in Fracture C than in Fracture H. Noticeably due to higher relative rotations for the medial-posterior fragment and interfragmentary movements over 5 mm for medial anterior-shaft gap, an additional medial fixation could provide higher stability for Fracture C. These conclusions agree with previous clinical studies that have demonstrated higher clinical failure rates at the medial side of complex tibial plateau fractures [1, 2]. Also, it has been clinically [2, 11, 17, 19, 20] or biomechanically [45] observed that lateral locking implants may not adequately stabilize the posteromedial fragments of bi-condyla tibia plateau fractures and a supplementary medial implantation will be required.

Considering the clinically observed importance of posteromedial fragment for bi-condylar tibia plateau fractures, developing a biomechanical fracture model including the coronal fracture line is highly demanded [11]. Contrary to previous experimental studies, which applied loads only on the medial tibial plateau [23, 29, 32‐34, 45], our load applicator was designed to simultaneously apply axial forces on both tibial plateau surfaces and distribute it 40% laterally and 60% medially, corresponding to physiological conditions [35, 36]. This load applicator with the adjustable position of the medial and lateral indenters can be used even for cadaveric samples with various plateau widths. Moreover, using artificial femur condyles provided a more physiological contact pressure on the tibial plateau as compared to previous biomechanical studies [35]. Furthermore, like some of previous investigations [23, 26, 31], the fracture model of this study presented a highly unstable situation by removing the central triangle of bone from the proximal tibia. Comparing mechanical behaviors of the Horowitz fracture model among previous experimental studies [23, 26, 29, 31, 33, 35] is difficult due to some distinctions regarding sample types (composite or cadaveric bones), loading and boundary conditions as well as loading scenarios. In this regard, only the study of Lasanianos et al. [35] was almost similar to ours. The static stiffness of Fracture H was found to be 304 \(\pm\) 95 N/mm which is in the same range of their reported value (400 \(\pm\) 64 N/mm) [35]. The mentioned research group obtained higher static stiffness values for specimens fixed with lateral locking plates. A reason for this could be that the fracture model of their study was stiffer due to preserving the middle fracture gap. Also, a different distal boundary condition was assumed in their experimental setup. Failure load of Lasanianos et al.’s fracture model was almost three times that of Fracture H reported in the current research, since we assumed the clinical failure criteria which is in contrast with their mechanical failure criteria. Among previous biomechanical studies, only the investigation of Yoo et al. [45] considered a Horwitz fracture model consisting of the posteromedial fragment. However, their fracture model was different than our coronal fracture model developed based on the clinical review of Pätzold et al.’s study [1]. Fracture C includes a sagittal articular fracture line in addition to the proximal fracture gap that is located laterally compared to that of Yoo et al.’s fracture model. They reported failure load of the Sawbone samples fixed with the tibial less invasive stabilization system (LISS) to be 1680 ± 179 N that is almost 4.5 times that of Fracture C. This difference could be due to their stiff fracture model in which the middle fracture gap was preserved as well as the lateral articular split was not included. Additionally, they utilized a hemispherical impactor for exclusive loading on the medial tibial plateau, although our loading was performed with a dual applicator which simultaneously applies the axial force on both tibial plateaus. Moreover, their failure criteria were different than our clinical ones. Regarding the loading scenario, an incremental cyclic loading until failure was considered in this study. Unlike previous studies, in which the testing protocol did not include the incremental cyclic loading [23‐35], the maximum load level of our gradual fatigue test increased stepwise by 50 N every 500 cycles. We believe that this progressive cyclic loading can simulate daily living activities of patients during the healing process, since after surgery incremental weight bearing on the injured limb is recommended.

Naturally, this study had some limitations which should be taken into consideration. First, artificial fourth-generation Sawbone samples were used for mechanical testing. Due to their material properties, their failure behavior might potentially differ from failure in human bone specimens. That is why we focused the failure analysis on relative movements between fragments rather than catastrophic failure of the fracture fixation construct. Also, Sawbone provided consistency in material, geometry, and mechanical properties, which increases the power to detect differences between groups. Second, the two fracture models were compared only under axial and bending loading conditions. The load applicator used in this study provided the sagittal rotational degree of freedom and consisted of femoral components of unilateral knee joint replacements which applied physiological pressure to the tibial plateau surface with distribution of 60% on the medial side and 40% laterally. In addition, the loading scenario only included axial knee joint forces, although for more realistic loading, the effect of muscle forces should also be considered. Moreover, instead of simulating the ankle joint, specimens were fixed distally to the testing machine directly, resulting in a somewhat overconstrained loading condition. Furthermore, to evaluate displacements and rotations, the fracture fragments were assumed as rigid parts. This assumption could be used due to high stiffness of the Sawbones which makes local deformations of segments negligible. Additionally, we simulated a hypothetical, unstable bi-condylar tibial plateau fracture in both fracture models by removing the central triangle of the proximal tibia as well as the side gap from the medial tibia. Last but not least, only one particular implant configuration with single lateral plating has been considered. With a more stable configuration like double medial and lateral platting which stabilizes the posterior fragment, the destabilizing effect of the coronal fracture is most likely less pronounced.

The outcomes of this study emphasize that it is mandatory for biomechanical simulations regarding complex tibial plateau fractures to be based on a clinically relevant fracture model such as ours (Fracture C) due to its ability to mimic native mechanical behavior more accurately than the traditional Horowitz model. The observed instability on the medial side of the coronal fracture model suggests that lateral plating alone provides insufficient mechanical fracture stabilization. We intend to pursue future research in this endeavor to propose the best fixation method for our novel fracture model.