Biomechanical Evaluation of Oblique Lumbar Interbody Fusion with Various Fixation Options: A Finite Element Analysis

Objective The aim of the present study was to clarify the biomechanical properties of oblique lumbar interbody fusion (OLIF) using different fixation methods in normal and osteoporosis spines. Methods Normal and osteoporosis intact finite element models of L1–S1 were established based on CT images of a healthy male volunteer. Group A was the normal models and group B was the osteoporosis model. Each group included four subgroups: (i) intact; (ii) stand‐alone cage (Cage); (iii) cage with lateral plate and two lateral screws (LP); and (iv) cage with bilateral pedicle screws and rods (BPSR). The L3–L4 level was defined as the surgical segment. After validating the normal intact model, compressive load of 400 N and torsional moment of 10 Nm were applied to the superior surface of L2 to simulate flexion, extension, left bending, right bending, left rotation, and right rotation motions. Surgical segmental range of motion (ROM), cage stress, endplate stress, supplemental fixation stress, and stress distribution were analyzed in each group. Results Cage provided the minimal reduction of ROM among all motions (normal, 82.30%–98.81%; osteoporosis, 92.04%–97.29% of intact model). BPSR demonstrated the maximum reduction of ROM (normal, 43.94%–61.13%; osteoporosis, 45.61%–62.27% of intact model). The ROM of LP was between that of Cage and BPSR (normal, 63.25%–79.72%; osteoporosis, 70%–87.15% of intact model). Cage had the minimal cage stress and endplate stress. With the help of LP and BPSR fixation, cage stress and endplate stress were significantly reduced in all motions, both in normal and osteoporosis finite element models. However, BPSR had more advantages. For cage stress, BPSR was at least 75.73% less than that of Cage in the normal model, and it was at least 80.10% less than that of Cage in the osteoporosis model. For endplate stress, BPSR was at least 75.98% less than that of Cage in the normal model, and it was at least 78.06% less than that of Cage in the osteoporosis model. For supplemental fixation stress, BPSR and LP were much less than the yield strength in all motions in the two groups. In addition, the comparison between the two groups showed that the ROM, cage stress, endplate stress, and supplemental fixation stress in the normal model were less than in the osteoporosis model when using the same fixation option of OLIF. Conclusion Oblique lumbar interbody fusion with BPSR provided the best biomechanical stability both in normal and osteoporosis spines. The biomechanical properties of the normal spine were better than those of the osteoporosis spine when using the same fixation option of OLIF.

anterior surgery are complex and the incidence of complications is high. During surgery, the large blood vessels in the abdomen can be damaged, which causes hemorrhaging and can even be life-threatening; if the vas deferens is damaged, retrograde ejaculation may occur [5][6][7] . Using the recently developed oblique lumbar interbody fusion (OLIF) technique, the target segment can be reached from the retroperitoneal space through the front of the psoas major muscle, avoiding direct damage to the spinal canal and nerve, with minimally invasive characteristics, such as small incisions, less bleeding, and short hospital stay [8][9][10][11] .
Oblique lumbar interbody fusion was introduced by Silvestre in 2012 12 ; it not only has advantages in surgical approach but also enables the implantation of a larger cage, increases the intervertebral fusion rate, and reduces the risk of subsidence of the cage. Since its introduction, it has been welcomed by spinal physicians and widely used in the treatment of degenerative diseases of the lumbar spine [13][14][15] . Although OLIF is popular, there are few studies on the biomechanical stability of various fixation options of OLIF, and it is not clear which fixation option has the best biomechanical properties. In addition, there are increasing numbers of patients with osteoporosis of the spine, with a rising trend in younger age groups. However, the biomechanical stability provided by different fixation methods of OLIF for osteoporosis has not been reported.
At present, there are two research methods used for spine biomechanics: studies of in vitro specimens and finite element studies. Studying in vitro specimens is a basic biomechanics research method. However, experiments take a long time because of the difficulty of obtaining specimens. Moreover, tissues such as muscles and ligaments are prone to fatigue, and the method has poor repeatability [16][17][18] . Many scholars have begun since the 1990s to use the finite element method to study biomechanics. The finite element method can establish different surgical models using computer software and simulate the transient postoperative state by imposing different boundaries and loading conditions. Not only can the experimental conditions be artificially controlled, but the operation can be repeated. A disadvantage of the finite element model (FEM) is that it disregards the roles of soft tissue. However, the experimental results obtained by the validated FEM are still of certain significance. Brekelmans et al. 19 first applied the finite element techniques in the field of orthopaedics. Liu et al. 20 took the lead in establishing a three-dimensional FEM of the lumbar spine. With the development of computer technology, the accuracy of FEM is improving, and its contribution has been recognized by numerous scholars. Considering the repeatability and operability of the finite element method, we chose to use it to research the biomechanical properties of OLIF with different fixation methods.  In this study, the three-dimensional finite element method was used to establish the following fixation models: OLIF stand-alone cage (Cage), cage with lateral plate and two lateral screws (LP), and cage with bilateral pedicle screws and rods (BPSR) for normal and osteoporosis spines. By comparing the surgery segmental ROM, cage stress, endplate stress, supplemental fixation stress, and stress distribution, we aimed to: (i) determine which fixation method in the normal model had the best biomechanical performance; (ii) determine which fixation method in the osteoporosis model has the best biomechanical performance; and (iii) compare the difference in biomechanical properties of normal and osteoporotic models when using the same fixation method.

Construction of an Intact Lumbar Finite Element Model
In this study, a 28-year-old healthy male volunteer (weight 65 kg, height 173 cm, and without lumbar disease) underwent CT scanning, with slice thickness of 0.625 mm. A total of 570 CT images were processed using commercial software (Mimics 20.0; Materialize, Leuven, Belgium) create a solid model. After repair, Hypermesh (Altair Technologies, Fremont, CA, USA) was used to mesh the solid model of the bony and ligamentous structures. Finally, Abaqus (Simulia., Providence, RI, USA) was used for material property definitions, model assembly and FEM analysis.
The FEM (Fig. 1) included L 1 -S 1 vertebrae, intervertebral discs and the ligaments system. The vertebral body included cortical bone, cancellous bone, and posterior bone. The thickness of the cortical bone was 1 mm 21 . The intervertebral discs were separated into annulus fibrosus, nucleus pulposus, and superior and inferior endplates. The discs were defined to be composed of 44% nucleus pulposus and 56% annulus fibrosus based on histological data 22 , and the thickness of the endplate was 0.5 mm 23 . The ligaments included the anterior longitudinal ligament, the posterior longitudinal ligament, the ligamentum flavum, the interspinous ligament, the supraspinal ligament, the capsular ligaments, and the intertransverse ligament. They were set as truss elements (T3D2) subjected only to tensile load. The FEM was meshed using the tetrahedral and hexahedral elements, except for the ligaments. There were 1,352,850 elements and 268,908 nodes. The material properties of components are shown in Table 1 22,[24][25][26][27] . For "osteoporosis spine", the elastic modulus of cortical bone and cancellous bone decreased by 33% and 66%, respectively, compared with the normal intact model 11 . Group A was the normal model and group B was the osteoporosis model.

Construction of Surgical Finite Element Model
The L 3 -L 4 level was defined as the surgical segment, and the annulus fibrosus, the nucleus pulposus, and the cartilage endplate were removed from the left side. Then, the supplemental fixation models of stand-alone cage (Cage), cage with lateral plate and two lateral screws (LP), and cage with bilateral pedicle screws and rods (BPSR) were constructed, respectively, for the normal and osteoporosis intact models. There were eight FEM. The normal FEM with various fixation options are shown in Fig. 2. The interbody cage and supplemental fixations were tetrahedral mesh. The cage was modeled based on Oracle cage (DePuy Synthes). It had 8 lordosis, was 40 mm long, 22 mm wide, 11 mm high anteriorly and 8 mm posteriorly, and it was made of polyetheretherketone. The lateral plate and two lateral screws were modeled based on a doublemedical plate, and the diameter of the lateral screws was 6.5 mm and 35 mm long. The bilateral pedicle screws and rods were modeled based on the EXPEDIUM 5.5 System (DePuy Synthes). The diameter of the pedicle screw was 5.5 mm and the length was 50 mm. The diameter of the rod was 5.5 mm, and the length exceeded the distance between the upper and lower pedicle screws. The lateral plate, screws, and rods were made of titanium alloy (Ti6Al4V). The material properties of the implant components are listed in Table 1.

Boundary and Loading Conditions
All the surgical models assumed that there was no sliding between the cage and the contact surfaces of the upper and lower vertebral bodies, and frictional contact was set between the articular process and the structure, and between the screws and the screw holes. The friction coefficient was set as 0.2. The lower surface of the S 1 vertebral body was fixed. So that the finite element research was closer to reality, an axial compressive preload of 400 N was added to the superior surface of L 2 to simulate the physiological compression, and a torsional moment of 10 Nm was imposed to simulate the motion of flexion, extension, left bending (LB), right bending (RB), left rotation (LR), and right rotation (RR). The ROM of the L 3 -L 4 segment was recorded and compared with the intact . Cage stress, endplate stress, supplemental fixation stress, and the stress distribution were compared in different surgical modes.

Finite Element Model validation
The L 3 -L 4 segment ROM for different motions of the normal intact model was compared with previous outcomes of Yamamoto 28 to verify the validity of the model. It was determined that the L 3 -L 4 ROM of the normal intact model was similar to that in the previous study, verifying that the normal intact FEM was valid.

Normal Model
For all surgical constructs, the L 3 -L 4 ROM was decreased compared with the intact in all motions. Cage provided the minimal reduction of ROM, which was 83.81% in flexion, 98.81% in extension, 84.07% in LB, 87.98% in RB, 88.55% in LR, and 91.24% in RR of the intact spine. The ROM of LP was slightly less than that of Cage. It was 79.73% in flexion, 74.55% in extension, 64.58% in LB, 66.13% in RB, 63.25% in LR, and 67.37% in RR of the intact spine. BPSR showed the maximum reduction of ROM compared with the intact; it was 50.38% in flexion, 43.94% in extension, 54.24% in LB, 51.90% in RB, 61.14% in LR, and 59.52% in RR of the intact spine.

Osteoporosis Model
For osteoporosis FEM, the ROM of Cage was 94.88%, 94.83%, 92.94%, 92.04%, 93.40%, and 97.30% of the intact spine in flexion, extension, LB, RB, LR, and RR, respectively. The LP ROM was less than that of Cage, greater than that of BPSR, and was 86.59% in flexion, 87.15% in extension, 70.00% in LB, 78.38% in RB, 72.82% in LR, and 78.10% in RR of the intact spine. BPSR also provided the maximum reduction of ROM compared with the intact for all motions, and it showed 45.61%, 45.81%, 59.41%, 51.05%, 62.27%, and 62.16% of the intact spine in flexion, extension, LB, RB, LR, and RR, respectively. Table 2 describes the L 3 -L 4 segment ROM for the normal and osteoporosis models.

Normal Model
For normal FEM, Cage provided the greatest cage stress and BPSR provided minimal cage stress; the cage stress of LP was slightly less than that of Cage, except for flexion and  Fig. 3A; stress was distributed in the periphery of cage for all motions.

Osteoporosis Model
The cage stress trends of the osteoporosis FEM were consistent with those of the normal model. The cage stress of BPSR was 65.91%, 80.10%, 59.71%, 52.24%, 71.36%, and 65.29% less than that of Cage in flexion, extension, LB, RB, LR, and RR, respectively. The cage stress distribution is shown in Fig. 3B; it was distributed in the periphery of cage. Table 3 describes the cage stress of normal and osteoporosis models.

Endplate Stress
Cage subsidence often occurs in the superior endplate after the operation, so we recorded the stress of the L 4 superior endplate (Table 4).

Normal Model
In the three fixation option models, Cage had the greatest endplate stress and BPSR had minimal endplate stress in all motions. Endplate stress of LP was slightly less than that of  Figure 4B shows the endplate stress distribution; it was also distributed in the endplate periphery.

Supplemental Fixation Stress
The supplemental fixation stress included LP stress and BPSR stress (Table 5).  Fig. 5B depicts the LP and BPSR stress distribution of osteoporosis models; it is also concentrated at the junction of the screw and the vertebrae.

Discussion
O blique lumbar interbody fusion evolved from lateral lumbar interbody fusion (LLIF) 29 . It has been popular since its introduction due to advantages including preservation of the normal structure of the posterior lumbar, less bleeding, and shorter hospital stay 8,30 . Although OLIF has been successful in the clinic, there are still the risks of cage subsidence, postoperative segment instability, and fracture for patients who undergo posterior supplemental fixation 31,32 . In cases of osteoporosis, these risks may be greater. However, it is not clear which fixing method of OLIF has the best effect. Consequently, we established various fixation options for FEM of normal and osteoporosis spines to determine their biomechanical stability.
Regardless of the approach, the purpose of lumbar fusion is to stabilize the lumbar spine by reducing the ROM of the target segment 33 . In the present study, the results indicate that all the fixation options enhanced the construct stability compared with the intact spine in normal models. Our research showed that in the normal spine group, the degree of stability was greatly different among supplemental fixation models. Cage provided the minimal reduction of ROM in all motions; it was 82.30%-98.81% of the intact model. After the addition of LP, the ROM of the L 3 -L 4 segment was reduced compared with Cage, but the degrees of reduction of bending and rotation were better than those for flexionextension. BPSR showed the greatest reduction of ROM in all conditions, and it was 43.94%-61.13% of the intact spine model. The reason for this result is that LP is fixed on the lateral, which has little restriction on the ROM in flexion and extension. BPSR is fixed at the posterior facet joint, which has greater restriction on ROM in various directions. These findings concur with those of with previous LLIF biomechanical studies [34][35][36] . However, in our study, the degree of various fixation options' ROM reduction was smaller compared with previous studies. Possible factors contributing to these differences include individual lumbar differences and

524
ORTHOPAEDIC SURGERY VOLUME 13 • NUMBER 2 • APRIL, 2021 the osteoporosis model was larger for all motions. The reason for this difference is that: the elastic modulus of the osteoporosis model was less than that of the normal model; the pullout strength of the internal fixation was weak; and the ROM limitation was not as good as that of the normal model. Therefore, we speculate that the effects of the B Fig. 4 Continued 525 ORTHOPAEDIC SURGERY VOLUME 13 • NUMBER 2 • APRIL, 2021 operation for elderly osteoporosis patients are not as good as for normal patients when the same fixation method is applied.
The greater the stress of the cage and the endplate, the greater the risk of subsidence of the cages into the endplate and the adjoining vertebral bone over time 37 . Malham et al. showed that the radiographic subsidence rate of OLIF is approximately 8% 38 . In our study, for a normal spine model, the BPSR option had minimal cage stress and endplate stress. It was indicated that the BPSR risk of cage subsidence was lower compared with LP and Cage. In addition, the cage stress and endplate stress were the greatest after only implanting the stand-alone cage, indicating that the Cage risk of cage subsidence was the highest compared with LP and BPSR. After lateral plate fixation, the cage stress and endplate stress were obviously decreased compared with the Cage, except for flexion-extension. These findings were comparable to those of previous finite element analysis 33 . However, in the previous finite element study, LP only reduced the cage stress and endplate stress in bending while changing little for other motions. Various types of LP may contribute differently to postoperative stability. Because there is no uniform type of LP at present, we suggest that clinicians select this fixation method with caution. For osteoporosis, we observed similar results to the normal spine model; that is, Cage has the greatest cage stress and endplate stress, and BPSR has the lowest cage stress and endplate stress. Due to the risk of cage subsidence of osteoporosis being higher than normal, the supplemental fixation method with the minimal cage stress and endplate stress should be selected. We suggest the application of OLIF with BPSR to treat osteoporosis. In addition, we observed that with the same fixation method, the cage stress and endplate stress of the osteoporosis model was greater than that of the normal model. Due to the pullout strength of the internal fixation in the osteoporotic model being weak, when external forces exist, the supporting force it provides is small. Therefore, the cage and the endplate are subjected to greater stress. This is also the reason why cage subsidence is more likely to occur in the case of osteoporosis than in a normal spine.
In the present study, in addition to comparing the ROM, the cage stress, and endplate stress, we recorded supplemental fixation stress and stress distribution. Chen et al. 39 showed that the yield strength of titanium alloy was 897-1034 Mpa. Our results showed that bilateral pedicle screws and rod stress and lateral plate and lateral screws stress were far less than the yield strength both in normal and osteoporosis FEM. Implant lateral plates or bilateral pedicle screws are feasible and safe to use. The stress distributions of LP and BPSR were concentrated at the junction of the screw and the vertebrae. To avoid screw breakage or ejection, we suggest that postoperative overactivity be avoided. Moreover, our results show that the supplemental fixation stress in all motions of the osteoporosis model is greater than that of the normal model with the same fixation method, which indicates that the safety of normal patients is better than for osteoporosis patients when using the same fixation option of OLIF.
To our knowledge, this study is the first to apply a three-dimensional FEM to establish an osteoporotic spine model to research the single-segment biomechanical stability of OLIF with different fixation methods. Cage and LP provide less stability than BPSR in the osteoporosis FEM, and are comparable to those of the normal FEM. The biomechanical properties of osteoporosis models were not as good as those of normal models when using the same fixation method. Because there are few finite element studies on osteoporosis, our results need to be further confirmed.
This study has some limitations. First, although finite element analysis has many advantages over in vitro experiments in the study of spinal biomechanics, the FEM cannot perfectly mimic the human body; for example, the paravertebral soft tissue cannot be precisely recreated and the function of muscles is ignored, which is a common problem faced by all finite element studies. Second, the osteoporosis FEM was constructed by decreasing the elastic modulus of cortical bone and cancellous bone by a certain proportion in normal FEM, ignoring the differences of individuals. Finally, the calculation results of the FEM only reflect the situation directly after the operation. It is not indicative of long-term postoperative status. Despite some limitations, in this study, the validity of the FEM has been verified, and the different types of fixation have been studied under the same experimental conditions. Therefore, the results of this study are still instructive for clinicians.

Conclusion
Bilateral pedicle screw and rod fixation had advantages in surgical segment ROM, cage stress, and endplate stress compared with Cage and LP fixation for OLIF both in normal and osteoporosis spines, and had the best biomechanical properties. The biomechanical properties of the normal spine were better than those of the osteoporosis spine when using the same fixation method.