Plate osteosynthesis is a well-established technique applied for fixation of bone fractures with use of metallic plates and screws to stabilize the reduced fracture, promote bone healing, and restore function (Perren, 2002). Trauma fixation devices are implemented to treat approximately 5%–10% of all bone fractures, corresponding to 320,000 and 1.1 million cases in Germany and the United States, respectively (GlobalData_MediPoint, 2016). Approximately 90% of these devices are internal fixators. Despite the impressive evolution of existing implants and guidelines throughout the recent decades (Perren, 2002; Augat and von Ruden, 2018), failure rates of up to 18% have been reported for plating of long bone fractures over the observed follow-up periods of approximately 12 months (Banovetz et al., 1996; Button et al., 2004; Kregor et al., 2004; Vallier et al., 2006), usually requiring reoperation. Additionally, non-unions occur in 5%–10% of the cases (Zura et al., 2016; Ekegren et al., 2018), increasing the burden on both the patients’ quality of life and the healthcare system (Hak et al., 2014; Rupp et al., 2018). Mechanics plays a fundamental role in the occurrence of fixation failures and healing complications (Sommer et al., 2004; Rizk and Al-Ashhab, 2015). Insufficient construct stability can lead to loss of reduction or to failure of the implant, the bone, or their interface. Implant malfunction or failure is one reason contributing to the high complication rate with up to 29% of the cases (Korner et al., 2005; Koso et al., 2018; Yetter et al., 2021), suggesting the need for improvements in the design and use of the implants (Gueorguiev and Lenz, 2018). Overloading failure may occur in unsuitably treated cases, where the fracture reduction, selection and configuration of the implants are inadequate and lead to insufficient construct strength, even in correctly performed fixations via too early and aggressive weightbearing due to the recommended rehabilitation protocol or poor patient compliance (Figure 1A). This risk may be substantial, especially in case of load bearing fixations such as bridge plating of the tibia or femur. Similarly, in a recent preclinical study, plate bending was observed in plated ovine tibia shaft fractures (Figure 1C) compared to the direct post-operative state (Figure 1B). Plate deformation could be observed radiologically already 1 week post operatively, suggesting early implant overloading.

Finite element (FE) modelling has emerged as a valuable tool for prediction of construct behaviour in various fields, including traumatology, orthopaedics and biomechanics (Taylor and Prendergast, 2015), providing valuable insights into the biomechanical performance of various implant designs and configurations (Mischler et al., 2020). Although validated FE modelling has been proven to be effective in predicting construct behaviour, most studies have relied on literature-based material properties and ignored post-elastic behavior, which can introduce uncertainties and limit the accuracy of the predictions (Lewis et al., 2021). This indicates the need for experimentally determined material properties to ensure accurate and reliable simulations of osteosynthesis constructs’ behaviour.

Towards the overarching goal to describe in vivo implant failures, the first step is to understand the construct behaviour in a more controlled in vitro setting. Therefore, the main aim of this study was to investigate the elastic and plastic biomechanical behaviour of plate fixation constructs using experimental testing and to establish validated specimen-specific FE models predicting construct stiffness, yield and maximum load. A secondary aim was to compare the predictive ability of the FE models using either experimentally determined elastic and plastic implant material properties or literature-based properties.

It was hypothesized that the FE models can more accurately predict fixation construct behaviour with use of experimentally determined elastic and plastic implant material properties versus use of literature-based properties.

FE simulations are widely used in orthopaedics, mainly to analyse the biomechanical behaviour and stability of bone-implant constructs (Taylor and Prendergast, 2015; Lewis et al., 2021). However, the models often lack validation and only consider the elastic properties of the implant (Lewis et al., 2021). The validation approach employed in this study, involving the development of an elasto-plastic FE model with experimentally determined implant properties and its validation using in vitro cadaveric construct test, has demonstrated promising results in accurately predicting the post-plastic behaviour of osteosynthesis constructs. By incorporating experimentally determined values of elastic and plastic properties of locking plates, the models’ predictive capabilities have been enhanced compared to literature-base properties, offering a more accurate prediction of the mechanical behaviour of the fracture fixation construct.

The FE simulations were able to predict the construct stiffness, yield load, and maximum load of the in vitro cadaveric experiments highly accurately, proving the validity of the modelling approach and thus confirming the initial main hypothesis. Kopec et al. investigated broken implants and hypothesized that these failed most likely due to mechanical overloading resulting from either repetitive or excessive limb loading (Kopec et al., 2021). They concluded that either the implant design was not fully suited to withstand the mechanical forces and moments, or the underlying physiological activity was excessive. Besides fatigue plate fractures, plastic implant deformation is another failure type seen in patients with good bone stock quality, leading to hardware failure if the applied loads exceed the implant capacity, especially in load bearing fixations of unstable comminuted fractures where the implant entirely carries the load (Perren et al., 2014). Similarly, the osteotomy gap sizes in the present study were large enough to avoid interfragmentary far cortex contact of the bone, thus the construct behaviour was mainly determined by the implant bending. In such load bearing scenarios, knowledge of the accurate implant material properties is crucial to accurately model the behaviour of osteosynthesis constructs when predicting failure. Taylor et al. described the general difficulty of collecting the material properties of all materials and brands (Taylor and Prendergast, 2015), thus literature values are often used for modelling. The importance of implant material properties might change when investigating different failure modes related rather to the bone component of the fixation construct, such as screw pull-out, cut-out, or perforation, or when analysing construct stiffness in the elastic region only. However, this study provided evidence for the importance of determining and using accurate material properties as the results were highly sensitive to the implemented values.

Implant material properties evaluated via uniaxial tensile testing of dog bone shaped samples were highly reproducible. When comparing the results to the used values in the literature, differences could be observed. Others used solely elastic properties, usually ranging between 195–210 GPa for stainless steel (Chen et al., 2010; Antoniac et al., 2019; Beirami et al., 2021; Takahashi et al., 2021) and 96–110 GPa for titanium alloy (Chen et al., 2010; Antoniac et al., 2019) LCPs. For implant grade stainless steel (1.4441), the Young’s modulus is often reported higher in literature as 200 GPa or 210 GPa versus the 183.8 GPa found in this study. Only a few previous studies on implant-related FE modelling investigated the post-elastic behaviour of osteosynthesis plates and thus, literature-based values are scarce. Reports by Zhou et al. and Mohandes et al. used yield stresses for titanium and stainless steel that corresponded with the minimum yield strength provided by the manufacturer but were lower compared to our results (Disegi, 2008a; Disegi, 2008b; Zhou et al., 2017; Mohandes et al., 2021). Simulating properties with underestimated values may suffice for worst-case analyses but not for accurate modelling of construct behaviour. When using literature-based implant material properties in our analyses, the concordance correlation coefficient, the standard error of estimate, and the relative standard error were lower compared to the results with experimental-based values, suggesting that the construct behaviour could be predicted with higher accuracy with the latter approach. Moreover, the measured properties were well in line with specifications of implant manufactures for stainless steel and titanium alloy LCPs (Disegi, 2008a; Disegi, 2008b) that were found to be very close to the values obtained by tensile testing, further underlining the accuracy of the tests.

Several limitations apply to the current study. The evaluated implant material properties were based on a single raw material profile bar of stainless steel or titanium. Even though this resulted in a small variance between the samples, the properties might vary between different suppliers and manufacturing series. To accurately determine the range of the different properties, multiple series from different suppliers would need to be analysed. Furthermore, the cadaveric experiments consisted of seven specimens of which only two were instrumented with titanium plates. However, the aim of the study was not to compare the differences in construct behaviour between the two implant materials but to validate the FE model. The fixation constructs were tested in compression-bending only and it was not investigated whether the behaviour in other loading modes could be predicted accurately. No post-elastic material behavior of the bone was incorporated in the simulations. However, no bone damage was observed around the screws after testing. Furthermore, a tied interface between screw and bone without the existence of press-fit forces was assumed. However, MacLeod et al. have demonstrated that the load-deformation characteristics of a construct exhibit minimal sensitivity to the utilization of a tied interface between the screw and bone, thereby justifying the chosen modeling approach (MacLeod et al., 2012). It is important to acknowledge that further studies and refinements to the FE model would be necessary to account for a broader range of implant failures such as fatigue failure caused by healing disturbances and incorporating musculoskeletal models to model more realistic in vivo loading before translating the models to the clinics. Nonetheless, this validation will contribute to the development of more reliable and effective treatment strategies for bone fracture management.

In conclusion, FE models of cadaveric ovine tibia shaft fracture fixations using relative stability via bridge plating with experimentally determined elasto-plastic material properties of stainless steel and titanium LCPs were able to predict construct stiffness and failure with high accuracy. The findings that the use of these properties significantly enhanced the predictive performance of the FE model compared to literature-based values emphasize the importance of carefully determining the appropriate elasto-plastic implant material properties either from the manufacturer or by experimental testing. Consequently, these models can be used to determine the forces needed to cause plastic deformations—i.e., overloading failure—in osteosynthesis constructs. In the future, these validated models could be used to predict the level of loads that cause the observed plate deformations in vivo or guide preoperative planning and postoperative rehabilitation protocols. However, to maximize prediction accuracy, potential errors at each step of the modeling workflow should be eliminated in an isolated manner, as it was presented here for the implant material properties.