Biomechanical comparison of five cannulated screw fixation strategies for young vertical femoral neck fractures

Abstract Vertical femoral neck fractures in patients younger than 65 years of age often require hip‐conserving surgeries. However, traditional fixation strategies using three parallel cannulated screws often fail in such patients due to an unfavorable biomechanical environment. This study compared different cannulated screw fixation techniques in patients via patient‐specific finite element analysis with linear tetrahedral (C3D4) elements. Forty vertical femoral neck fracture models were created based on computed tomography images obtained from eight healthy participants. Five different fixation strategies: alpha, buttress, rhomboid, inverted triangle, and triangle were assessed in walking status. Biomechanical parameters including stiffness, interfragmentary motion in two directions (detachment and shearing), compression force, and maximal implant stress were evaluated. The mean relative coefficient of strain distribution between the finite element analysis and experiment was from 0.78 to 0.94. Stiffness was highest (p < .05) in the buttress group (923.1 N/mm), while interfragmentary motion was lowest (p < .05) in the alpha group. Maximal stress was highest (p < .05) in the buttress group and lowest in the alpha group. Shearing values were significantly lower in the alpha group than in the rhomboid group (p = .004). Moreover, Shearing values were significantly higher (p = .027), while detachment values were significantly lower (p = .027), in the inverted triangle than in the triangle group. Clinical significance: Our results suggest that alpha fixation is the most reliable and biomechanically efficient strategy for young patients with vertical femoral neck fractures. Regular and inverted triangular fixation strategies may be suitable for fractures of different skeletal constructions due to antidetachment/shearing abilities.


| INTRODUCTION
Vertical femoral neck fractures in patients younger than 65 years of age often require hip-conserving surgeries 1 ; however, such procedures remain challenging for orthopedic surgeons due to their high-energy nature, 2 general vulnerabilities in the vasculature, 3 and an unfavorable biomechanical environment. Anatomic reduction, thorough stable fixation, and primary healing are necessary for reducing the risk of avascular necrosis and non-union in these patients. However, the optimal fixation strategy for this fracture type remains controversial. 4 According to a web-based survey conducted in 2014, 4 sliding hip screws and cannulated screws are the two most commonly utilized fixation devices for vertical femoral neck fractures. Recently, sliding hip screws have been criticized 5,6 due to their association with an increased risk of osteonecrosis. Thus, cannulated screws remain the most promising and commonly used devices 1 because of their minimal invasiveness, easy handling, and ability to induce dynamic compression. Unfortunately, due to the adverse nature of vertical femoral neck fractures, traditional methods utilizing three cannulated parallel screws are associated with high rates of mechanical failure (19%) and osteonecrosis (14%) in patients with Pauwels type III vertical femoral neck fractures. 7 Given that these methods are widely accepted amongst surgeons, it remains necessary to identify more efficient screw fixation strategies that can reduce the rates of such complications.
Recently, researchers have successfully modified fixation techniques by adding a fourth screw or augmenting the fixation system with a buttress plate. 8,9 However, few studies have systematically compared biomechanical outcomes among the available fixation methods. To provide guidance for clinical practice, the present study aimed to compare different cannulated screw fixation techniques in patients with vertical femoral neck fractures and to illustrate the detailed biomechanical properties of these techniques via patientspecific finite element analysis (FEA).

| Models establishment
Eight healthy volunteers ranging in age from 20 to 55 years without any history of hip fracture, metabolic bone disease, or general comorbidities were recruited for the present study (Table 1). Mimics software (Version 19.0; Materialise) was used to develop patientspecific three-dimensional (3D) models with a modified Pauwels angle of 70°based on 0.625 mm thick computed tomography (CT) images.
The models were then osteotomized using 3-Matic software (Version 11.0; Materialise). We assessed the effects of five different internal fixation strategies in each model ( Figure 1): (1) three inverted parallel screws plus one off-axis screw, arranged in an "alpha" configuration (group ALP [G-ALP]); (2) three inverted parallel screws plus one buttress plate strengthening the calcar (G-BUT); (3) four parallel screws arranged in a "rhomboid" configuration (G-RHO); (4) three parallel screws with an inverted triangular construction (G-ITR); (5) three parallel screws with a triangular construction (G-TRI).
To control the essential confounding variable of screw position, all the cannulated screws were implanted according to the same standard criteria, which has been well-studied and established in previous studies. [10][11][12][13][14] For the parallel cannulated screws, the directions were along the femoral neck axis which was automatically calculated in MATLAB (The MathWorks). 15 The parallel screws were positioned dispersedly, 10,11 at 2.5 mm to the cortex of the femoral neck, 12,13 and 5 mm distal to the subchondral bone in the femoral head. 14 The off-axis screw in G-ALP was implanted at 5 mm proximal to the most prominent part of the great trochanter (to prevent soft tissue irritation due to screw protruding) and targeted at the inferior femoral head-neck junction (to provide more favorable bone mass for screw purchase).
The constructs were all created in SolidWorks2017 (DS Solid-Works Corp.) using 6.5-mm cannulated screws (Stryker) and a

| FEA validation
The donor was 160 cm in height, 50 kg in weight, and absent of any All cannulated screws were assigned as titanium (Ti-6L-4V), with Young's modulus (E) of 110,000 MPa and Poisson's ratio of 0.3. 21,22 Thread-bone interfaces were tied while shaft-bone and fracture interfaces were assigned as slide contact with a frictional coefficient 23 of 0. 46

| Patient-specific FEA simulation
Following the validation experiment, patient-specific FEA was performed in all the 40 models with C3D4 meshes using the abovementioned procedure. There were approximately 500,000 elements (from 430,826 to 542,864) and 100,000 nodes (from 93,492 to 117,661) in each model. Additionally, tie contacts were assigned to the plate-screw and screw-bone interfaces in the G-BUT models. All the models were subjected to 237.7% body weight in line with the femoral mechanical axis. Parameters including stiffness, IFM, compression force, and implant stress were comprehensively analyzed.
Stiffness was calculated by dividing patient-specific load by the displacement of the applying node. The IFM of each node was calculated using a previously described formula ( Figure 3E Randomized block one-way analysis of variance test was used for comparison among five groups and paired t test was used for comparison between just two groups.

| RESULTS
In the validation study, we observed a significant linear correlation in the strain distributions between the FEA and VIC-3D results  Figure 6A). In addition, several sites of stress concentration were detected in G-BUT, including at the curvature and locking plate-screw junction ( Figure 6B).

| DISCUSSION
In the preliminary validation study, the correlation coefficient (R = 0.78-0.94) for the results of the FEA and VIC-3D analyses was satisfactory and within the range of values when compared to previous studies (R = 0.74-0.96). 28,29 The models with C3D4 performed no worse than those with C3D10 in terms of principal strain distribution, showing similar correlation coefficient (R-value). Quadratic tetrahedral elements (C3D10) modeling can obtain more accurate values owing to denser nodes compared to linear tetrahedral (C3D4), which was confirmed in our test with the value of the slope closer to one. The results of differences between G-ALP and G-ITR in stiffness and IFM were the same for the two element types which was consistent with the mechanical experiment. Besides, the C3D4 model has been commonly used in many studies, 9,[30][31][32] and it was proved to be similar to C3D10 models in accuracy under axial deformation (error: 2.7% vs. 2.8%). 33 ))/IFM(G-ITR))) has a significant (p < .05) linear correlation with DIM% (DIM/IFM), which means that as detached proportion of IFM became larger, the biomechanical benefits of G-ITR became increasingly prominent. (B) Analysis of the bone structures of the patients suffering from the greatest detached and shear force. The femur on the left has a typically thin femoral neck (390.66 mm 2 ) with the lowest neck-shaft angle (137.55°) which was better to be fixed by ITR whereas the one on the right has a thick neck (928.00 mm 2 ) with the highest neck-shaft angle (137.55°) which was better to be fixed by G-TRI. (C) The derived formula used to demonstrate the phenomenon that the detached force increased as the thickness of femoral neck and the neck-shaft angle decreasing. M, moment; W, section factor. Note that the angle between anatomic and mechanical femoral axis was assumed as a mean value of 7°. IFM, interfragmentary motion [Color figure can be viewed at wileyonlinelibrary.com] ordinary compression force. The detachment force in the superior part of the fracture surface may be transferred to the plate-screw junction via the proximal locking screw. This force can also be ascertained based on the residual stress concentration at the platescrew junction ( Figure 6B). Unfortunately, the stress concentration of the plate-screw junction represents a major mechanical pitfall of this technique, given that it is associated with an increased risk of breakage and subsequent fixation failure 35 (Figure 6C-F). Furthermore, applying a buttress plate may exert detrimental effects on blood supply to the femoral neck due to additional dissection. This may endanger the inferior retinacular artery, which plays an important role in perfusing the femoral head following femoral neck fractures. 36,37 Moreover, application of a buttress plate in patients with femoral neck fractures (especially those with the subcapital type) increases the risk of impingement 35,38 as the hip flexes, potentially leading to significant complications such as osteoarthritis.
Given these complications, augmentation using a buttress plate may not be ideal in patients with femoral neck fractures. Nonetheless, buttress plates may be helpful in patients with osteoporotic or comminution when adequate bone purchase and compression force cannot be achieved using cannulated screws.
To further investigate the biomechanical properties at the off-axis screw, we compared IFM between G-RHO and G-ALP, decomposing IFM into shearing (SIM) and detachment (DIM) components. In accordance with previous hypotheses, 39,40 we observed compromised compression force in G-ALP due to the lack of parallelism, although these decreases in compression force did not compromise DIM in our study. In contrast, anti-shearing ability was significantly greater in G-ALP than in G-RHO, indicating that the biomechanical effect of the off-axis screw is to improve anti-shearing stability. This unique biomechanical advantage can be explained as follows: First, the offaxis screw is more likely to be perpendicular or angulate upward to the fracture plane, which can neutralize the sliding effect 6 caused by three angulated parallel screws. Thus, the technique may confront shearing forces more efficiently. Second, bone quality is much better around the calcar than around Ward's triangle of the femoral head, leading to better bone purchase for techniques utilizing an off-axis screw. Third, the off-axis screw acts as a lever to transfer the bending moment from the femoral head to the calcar, thereby enhancing cortical support. 41 Furthermore, in contrast to the RHO technique, the off-axis screw used in the ALP technique will not increase stress at the lateral wall, which can reduce the risk of iatrogenic fractures. In accordance with previous findings, the biomechanical properties observed in the present study indicate that techniques utilizing an off-axis screw are associated with improved union rates and reduced rates of avascular necrosis when compared with traditional techniques utilizing three cannulated screws (Figure 9). 41,42 In the present study, we also compared mechanical differences between G-ITR and G-TRI. As reported in previous studies, 43,44 there were no significant differences in stiffness or IFM between these two groups. However, our detailed biomechanical analysis ( Figure 5) revealed that the G-TRI technique was associated with a greater ability to resist shearing forces, while the G-ITR technique was associated with a greater ability to resist detachment forces. As expected, the two lag screws placed superiorly and inferiorly are better at resisting tensile and shearing forces, respectively. Further regression analysis revealed the advantages of G-ITR given increases in SIM and vice versa. Interestingly, we observed that patients with higher detach- Apart from the abovementioned cannulated screw techniques, dynamic hip screws (DHS) with an anti-rotational screw are also a common clinical strategy for vertical femoral neck fractures. This device has the advantage of providing angular stability and has been proved to be stiffer than cannulated screws in a previous biomechanical study. 45 However, the implantation of DHS is more complicated than cannulated screws and it requires an invasive procedure with dramatic damage to the blood supply, soft tissue dissection, and large bone volume loss. Clinical studies have found a statistically higher operative time, incision size, intraoperative blood loss, 46 and subsequently higher avascular necrosis rate 5,6 in DHS compared with cannulated screws. Consequently, cannulated screws with the advantage of being minimally invasive, easy handling, and the ability to induce dynamic compression remain the most promising fixation strategy.
The present study possesses some limitations. First, all models were developed using linear elastic materials and did not incorporate bone plastic deformation or screw loosening processes, which are known to lead to mechanical failure in older patients with osteoporotic fractures.
In addition, underestimations of IFM values were attributed to the intact fracture surface and accurate anatomic reduction employed in our models. Our study also focused only on initial stability rather than that during the bone healing process. Furthermore, the thread of the implants was simplified in this study, but it has been proven to have little effect on the outcome. 47 Despite these limitations, our findings may aid orthopedic surgeons in selecting the most appropriate fixation strategy in clinical practice.
In conclusion, our findings indicate that the techniques utilized in G-ALP and G-BUT provided the most stability with the least IFM in vertical femoral neck fractures. Given its unique biomechanical characteristics, relatively lower implant stress, and decreased likelihood of surgical dissection, G-ALP may be more reliable than G-BUT. However, augmentation using a buttress plate may be helpful in patients with comminuted fractures when adequate compression cannot be achieved using fragments alone. A regular triangular screw configuration can be used to ensure maximum anti-shearing ability and may represent the most appropriate choice for patients with a thick femoral neck and high NSA values, while an inverted triangular configuration may be the most appropriate choice for patients with a thin femoral neck and lower NSA values.