Distribution characteristics of stress on the vertebrae following different ranges of excision during Modified Anterior Cervical Discectomy and Fusion: A correlation study based on finite element analysis

Abstract

Background

Modified Anterior Cervical Discectomy and Fusion with specific resection ranges is an effective surgical method for the treatment of focal ossification of the posterior longitudinal ligament (OPLL). Herein, we compare and analyse the static stress area distribution by performing different cuts on an original ideal finite element model.

Method

A total of 96 groups of finite element models of the C4–C6 cervical spine with different vertebral segmentation ranges (width: 1–12 mm, height: 1–8 mm) were established. The same pressure direction and size were applied to observe the size and distribution area of stress following various ranges of excision of the C5 vertebral body.

Results

Different cutting areas had similar stress aggregation points. As the contact area decreased, the stress and the bearing above area increased. The correlation of stress area variation was highest between the 1–2 MPa and 6 MPa–Max regions (Rho = − 0.975). In the surface visualisation model fitting, the width and height were of different ratios in different stress regions. The model with the best fitting degree was the 1–2 MPa group, and the equation fitting (Rho = 0.966) was as follows: Area = 908.80 − 25.92 × Width + 2.71 × Height.

Conclusion

Modified Anterior Cervical Discectomy and Fusion with different resection ranges exhibited different stress areas. In a specific resection range of the cervical spine (1–12 mm, 0–8 mm), area conversion occurred at a threshold of 4 MPa. Additionally, the stress was concentrated at the contact points between the vertebral body and the rigid fixator.

Background

Ossification of the posterior longitudinal ligament (OPLL) is a rare ectopic ossification caused by genetically driven metabolic abnormalities [1,2,3]. For example, blood glucose and lipid levels can indirectly influence systemic soft tissue ossification. Insulin resistance disrupts glucose homeostasis and alters osteoblast function and bone mineral density by affecting the insulin-like growth factor 1 (IGF-1) signaling pathway [4, 5]. However, further studies are needed to elucidate the pathways and metabolomic mechanisms involved [6,7,8].

The posterior longitudinal ligament features a unique physiology. Its thickening or abnormal ossification can lead to nerve compression and the resultant symptoms [9], which tend to worsen with disease progression.

Several guidelines exist for classifying the posterior longitudinal ligament, including the direct morphological classification proposed by Japanese scholars, which consists of the following types: continuous, segmental, mixed, and localized [10]. This classification is currently used in clinical practice settings. Further, while the Asia Pacific Spine Society believes that different ossifications require different surgical plans [11], specific types of OPLL have yet to be considered.

Typically, focal OPLL is localized adjacent to the intervertebral space; ACDF is often the best solution. This surgical technique sacrifices the mobility of only one cervical segment and is widely used to treat focal OPLL. Unfortunately, for patients with focal OPLL located outside the intervertebral space, ACDF has significant limitations.

A rare case of central-focal OPLL in the vertebral body has been reported; here, the C5 vertebral body was partially removed with the ossification mass using the Modified Anterior Cervical Discectomy and Fusion (Mod ACDF) technique [12] (Fig. 1). This could mitigate the shortcomings of the anterior cervical corpectomy decompression and fusion (ACCF) and posterior cervical approaches. The surgery’s reliability was confirmed using magnetic resonance imaging (MRI) 16 months postoperatively (Fig. 2). In a subsequent study [13], the safety of the Mod ACDF vertebral body model in the static state was verified through a finite element analysis(FEA) and nine groups of different variables. The average Von Mises stress and displacement levels of the Mod ACDF were between those of the ACDF and ACCF techniques. Also, with the Mod ACDF, a different bone mass was associated with the risk of vertebral collapse.

Fig. 1

Mod ACDF technique. The skin is incised and the soft tissue is dissected, avoiding major nerves and blood vessels. After incising the anterior longitudinal ligament, the C4/5 disc tissue is removed. The disc is nipped using a nucleus pulposus forceps, and the upper and lower endplates are removed by curettage. The appropriate size of the C5 vertebral body is excised directly using an ultrasonic bone knife until the OPLL ossification mass is fully exposed. The bone mass is carefully separated and removed. An appropriately sized cage is placed between the C4 and C5 vertebrae for fusion, and a four-welled titanium plate is fixed to the vertebrae

Fig. 2

MRI of the patient who underwent Mod ACDF at the 16th month postoperative follow-up. The physiological curvature of the cervical spine remained, and the CL curvature angle of the C5 vertebra was 36.2°. There was no mass in the cervical canal

These studies had some limitations. Because this type of OPLL was scarce, the excision protocol of the Mod ACDF must precisely match the mass’s size. In previous studies, only one incision was performed at a single vertebral body level, with a cross-sectional area of 8 × 9 mm. We must determine if the ossified mass’s size affects this surgical technique’s implementation. Therefore, we investigated local static stress distributions related to different excision sizes. Our results will inform future improvements to existing surgical approaches. An effective FEA system is essential for simulating and summarizing the stress distributions caused by different excision size ranges to achieve timely and early intervention.

Methods

Finite element model of the cervical spine

The study’s protocol was approved by the institutional review board of Fuzhou Second Hospital (Approval Number: 2023001). The study participant provided written informed consent, and all methods adhered to relevant guidelines and regulations. The study participant was a healthy 52-year-old male weighing 70 kg and measuring 168 cm in height. He had no history of cervical vertebrae-related diseases. Computed tomography (CT) images of the entire lower extremity and gait data (based on optical motion capture) were collected for later analysis.

Computed tomography

The CT scan covered the skull base to the seventh cervical vertebra. Three-dimensional CT images of the cervical vertebra were generated with a reconstruction thickness of 1 mm and a reconstruction interval of 1 mm and then saved using DICOM. The DICOM file was imported into Mimics Medical (version 21.0; Materialise Mimics, Leuven, Belgium), and a model was constructed based on the default image orientation. After adjusting the image threshold, all vertebrae except for C4–C6 were isolated and removed. All abnormal connections were eliminated, and the resulting 3D cervical spine model was exported as an STL file. This STL file was brought into Geomagic Wrap (version 2021; Geomagic, Research Triangle Park, North Carolina, USA). Following grid redrawing, polishing, and other processes, curved 3D images of cervical vertebrae were generated and saved as STP files.

Subsequently, the cortical bone was modeled as a 1 mm thick outer shell surrounding cancellous bone, with the bony endplate defined as a 1 mm thick layer covering the surface of the vertebral cortical bone. In nonsurgical regions, a 0.8 mm thick cartilaginous endplate was applied over the lateral side of the bony endplate. The intervertebral disc is connected to a cartilaginous endplate. It comprises a gel-like nucleus pulposus surrounded by a tough annulus of fibrous and connective tissue fibers. The nucleus pulposus makes up 40–60% of the disc’s mass, giving it distinct material properties [14, 15]. We used SOLIDWORKS (version 2020; Dassault Systems SOLIDWORKS Corp, Waltham MA) to open STP files, create models of the spine’s different parts (endplate, articular cartilage, intervertebral disc, polyether ether ketone(PEEK) cage, titanium plate, and screws), set three-dimensional parameters for the spinal parts (Fig. 3), and assemble the cervical vertebrae. Lastly, we used the Mod ACDF model to treat the completed cervical vertebrae.

Fig. 3

Screw size: diameter (a) = 4.8 mm and length (b) = 7.6 mm. Titanium plate size: width (c) = 14.5 mm, length (d) = 26.8 mm and thickness = 3.4 mm. PEEK cage size: length (e) = 15.1 mm, width (f) = 11.0 mm and height (g) = 3.7 mm

Mod ACDF technique

During the ACDF technique, the C4/5 intervertebral disc and the targeted intervertebral space’s upper and lower cartilaginous endplates were removed. Parts of the C5 vertebral body were also removed, forming a rectangular excision block within the boundaries of the C4/5 space cage. The screw was placed 38% away from the lower endplate of the C5 vertebral body. The removal range was set as follows: width, 1–12 mm, and height, 1–8 mm, with a data interval of 1 mm, resulting in 96 groups of models.

A cervical cage (5 × 14 × 16 mm; Libeier) was positioned in the C4/5 intervertebral space, 1 mm from the front edge of the vertebra. The cage was wedged securely between the bones without gaps to prevent errors in force application. Additionally, a plate (1 × 20 × 16 mm) was placed across the anterior aspect of the C4 and C5 vertebrae and fixed with four screws at the head and tail ends, ensuring that all screws were enclosed in cancellous bone (Fig. 4), (Fig. 5).

Fig. 4

FEA results of the Mod ACDF model group

Fig. 5

The FEM ranges of motion and facet contact force in all directions in this study were compared to those in previous studies

Finite element analysis

In this study, we used ANSYS Workbench (version 2022 R2, ANSYS, Pennsylvania, USA) to define each part’s material parameters (refer to Table 1) and establish mutual contact points. The analysis focused on the cortical bone’s outermost layer and modeled the cervical spine’s axial stress distribution under static conditions using 96 different models. Each model represented one of seven stress intervals (0–1 MPa, 1–2 MPa, 2–3 MPa, 3–4 MPa, 4–5 MPa, 5–6 MPa, 6 MPa–Max) by adjusting the stress interval. The final results were stored on the cortical bone’s surface, and each stress interval’s distribution area was measured to determine the distribution.

Table 1 Spinal structure and instrumentation material properties

Table 2 Mesh convergence test results

Table 3 Area correlation analysis under the influence of different forces

Table 4 Correlation between width-height of vertebral excision and different stress distributions

Mesh convergence analysis

To obtain accurate data, we conducted a mesh convergence analysis to study the impact of mesh size on the predictions of the finite element model. We used tetrahedral elements for the mesh. We evaluated four different mesh sizes (2, 1.5, 1, 0.5 mm) and compared the computation time and peak Von Mises stress at the upper edges of C6 and C7. After comparison, we selected a mesh size of 1 mm as the optimal size for this study. This size is balanced between shorter computation time and a peak Von Mises stress change rate of less than 5%. Using this mesh size resulted in a model comprising 667,554 nodes and 366,367 elements (Table 2).

Validity test

The lower endplate of C7 in the unresected cervical spine model was completely fixed to verify its validity. We applied an axial force of 73.6 N and a 1 Nm torque to the upper endplate of C3, following the right-hand rule for the direction of the torque. The cervical spine model was tested for flexion, extension, left and right bending, and rotation. Finally, we compared the results of this model with those of previous experiments [16,17,18].

Boundaries and loads

The screws, cage, and vertebrae were positioned in direct contact, and all components were securely attached to the vertebrae. All the nodes on the lower surface of the C6 vertebral body were fixed in all directions. However, C4 was not restricted. The material properties in this study were considered to be continuous and uniform. Surface friction was applied between the joints, between the disk and vertebral body, and between the disk and endplate (coefficient of friction: 0.1). A 200 N preload was applied to the C4 upper surface in each model, and a finite element analysis (FEA) was conducted. Different contour levels were set in the final stress display interface, and the respective areas of different stress intervals in various model groups were calculated.

Statistical analysis

The statistical analyses were conducted using SPSS 23.0 (IBM, New York, USA). There were two independent variables (length and width of the excised area) and seven dependent variables (area of the different stress areas of the remaining vertebrae). Normally distributed data are presented as means ± standard deviations, and non-normally distributed data are expressed as medians (quartiles). Spearman correlation analysis was used to examine the correlation between dependent variables, and multiple linear regression was employed to investigate the relationship between independent and dependent variables to identify the optimal function. Origin 2022 (OriginLab, Massachusetts, USA) was utilized for surface visualization processing. The equations are shown based on the fitting results. A correlation coefficient (Rho) greater than or equal to 0.8 was considered a high correlation, 0.8 to 0.5 a moderate correlation, 0.5 to 0.3 a low correlation, and less than 0.3 a weak correlation.

Results

Model validation

A comprehensive non-pathological model of the cervical spine was created. Using the same interference method, the ranges of motion and Von Mises stress in the model were compared with those from previous studies. This confirms the effectiveness and reliability of the finite element model established in this study, with the error within a controllable range.

Finite element analysis

The two screws on the C5 vertebral body were confirmed to be positioned at the same level. The shortest distance between the two screws was 5.6 mm, and the distance between their centers was 6.4 mm. Additionally, the distance between the center of the screw and the upper endplate of the C5 vertebra was 8.1 mm, while the total height of the vertebral body was 13.7 mm (Fig. 6).

Fig. 6

(a) The distance between the centre of the two screws was approximately 6.4 mm. (b) The shortest distance between the two screws was approximately 5.6 mm. (c) The distance between the screw and the upper endplate of the C5 vertebra was approximately 8.1 mm. (d) The total height of the C5 vertebral body was 13.7 mm

The stress concentration points were found to be similar between different cutting ranges. The stress was mainly concentrated at the corner of the nail eye and the cutting rectangle. The maximum stress was focused around the two sides of the nail eye. The stress-bearing area increased as the contact area decreased (Fig. 7).

Fig. 7

(a) The stress on the C5 vertebral body is mainly distributed around the rigid structure and the posterior vertebral body when the minimum volume is removed. (b) The stress on the C5 vertebra is distributed more uniformly when the maximum volume is removed

In the 96 groups of models, the contact areas were as follows:

• For 0–1 MPa: 2719.0 (58.8) mm², accounting for 59.8 (0.8) % of the entire C5 vertebral body surface area.

• For 1–2 MPa: 757.00 (181.5) mm², accounting for 16.8 (3.7) % of the entire C5 vertebral body surface area.

• For 2–3 MPa: 382.00 (54.3) mm², accounting for 8.5 (1.0) % of the entire C5 vertebral body surface area.

• For 3–4 MPa: 226.0 (19.8) mm², accounting for 5.0 (0.4) % of the entire C5 vertebral body surface area.

• For 4–5 MPa: 143.5 (29.5) mm², accounting for 3.2 (0.7) % of the entire C5 vertebral body surface area.

• For 5–6 MPa: 72.0 (50.8) mm², accounting for 1.6 (1.1) % of the entire C5 vertebral body surface area.

• For 6 MPa–Max: 196.0 (134.8) mm², accounting for 4.3 (3.0) % of the entire C5 vertebral body surface area.

Correlation analysis

The comparison of different pressure ranges (0–1 MPa, 1–2 MPa, and 6 MPa–Max) showed that there was no significant correlation (P > 0.05) between certain areas. The groups were then divided into high, moderate, and low correlation groups based on a predetermined Rho. The findings were visually represented in a surface diagram divided into seven intervals. The distribution area was highest for the 0–1 MPa range when the model’s cut height and width were 1 and 10 mm, respectively. In the 1–2 MPa range, the distribution area was closely associated with the width, showing a downward trend with minimal correlation to height. In the 2–3 MPa range, the model group with a width and height of 1 mm and 1 mm, respectively, had the highest area, and an increase in height caused a decrease in area. The width for the 3–4 MPa and 4–5 MPa ranges demonstrated an increasing trend, reaching peaks at specific points. The area change was negatively correlated with height for specific ranges. Lastly, the area increased to particular widths in the 5–6 MPa range. The sensitivity to changes in the high area was low, and the area reached its highest point at coordinates (9, 8 mm). For widths greater than 9 mm, the area decreased more sharply. In the 6 MPa–Max range, the entire surface showed a relatively smooth trend. The width was significantly correlated with the area ratio, and an increase in height positively affected the distribution area. Additionally, the area was at its lowest at point (1, 1 mm) and at its highest at point (12, 8 mm).(Table 3), (Table 4), (Fig. 8)

Discussion

The existing biomechanical data suggests that the complete spinal bone unit is an essential prerequisite for maintaining the normal load-bearing physiological function of the spine [19, 20]. Diseases or surgical procedures can sometimes disrupt this balance, such as the anterior approach used to treat OPLL in the cervical spine. ACDF has been suggested for treating issues near the intervertebral space, such as degenerated discs and ossified ligaments, using advanced clinical technology for a maximum of two segments. While ACDF has successfully treated OPLL that affects more than two segments without removing vertebrae, it is still not recommended for treating multi-segment OPLL [21]. For continuous OPLL of the cervical spine affecting three vertebrae, ACCF is a relatively classical surgical radical treatment method that removes the vertebrae-ossification complex and uses titanium mesh for maximum support [22, 23].

However, this technology can cause severe damage. Due to the characteristics of titanium mesh, its contact with the endplate is limited to a few sharp points. This limited contact can cause a concentration of Von Mises stress, increasing the risk of complications of fusion device settlement after ACCF compared to ACDF. Moreover, ACCF is no longer performed for continuous OPLL involving more than three vertebrae. Sun et al. reported that excising the anterior vertebral structure and floating the anterior vertebrae complex for treating cervical OPLL involving more than three vertebrae has gained clinical recognition [24]. Biomechanical tests of this new surgical technique have confirmed that the total range of motion changes observed were similar to those following traditional ACDF. This demonstrates that the novel technique produced various spinal array changes within an acceptable range for spinal biomechanical levels [25].

The Mod ACDF technique involves directly removing the central OPLL of the vertebrae using a static unit load, which falls between conventional ACDF and ACCF. This technique features aggressive intraoperative resection of the vertebrae, using only a PEEK cage to fill the vertebral space. The spinal vertebrae primarily bear the load on the static vertical axis. This invasive technology directly intervenes in the longitudinal bearing structure of the vertebral body, altering the standard mechanical structure of the human body. In the Mod ACDF technique, the cage bridges the bony endplate of the C5 vertebra directly above the deficient vertebra. This reduces the force area of the C5 vertebra and the cage under constant higher load bearing. As per PASCAL’s theorem (P = F/S), this increases the upper surface pressure of the C5 undergoing partial vertebral osteotomy, a significant factor in the concentration of Von Mises stress. While concerns regarding cage subsidence and vertebral collapse have been raised, previous studies with a follow-up of 16 months have addressed these concerns. It was found that under normal bone mass and weight of 200 N, approximately 9.4 × 20 × 7.8 mm of the vertebral body was removed in the resting state [12].FEA simulation analysis revealed that the stress concentration in the vertebral body after Mod ACDF was 4.4 times higher than after conventional ACDF, which is 43% lower than after ACCF. Additionally, the risk of vertebral collapse after Mod ACDF was 5.6 times higher than after conventional ACDF, which is 31% lower than after ACCF [13].

Notably, several study limitations should be considered. Previous studies have only discussed one possible resection method. OPLL does not always appear in the standard location in clinical practice. When OPLL is in a non-standard position, it becomes necessary to expand the resection range of the vertebral body accordingly. In the Mod ACDF technique, the size of the vertebral body to be excised depends on the location and size of the ossification mass. A deeper ossification mass may require a wider resection. However, decisions based on bone size often carry hidden dangers, such as excessive removal of bone, which could lead to complications. Therefore, to simulate the types of OPLL encountered in the future, we need to simulate the differences in the excision methods using FEA tools.

This study used a correlation analysis model based on finite element theory to address a specific problem. We simulated the excision range in a surgical technique called Mod ACDF by using vertebral bodies C4 to C6 and matching different excision schemes. We could identify statistical differences by creating seven types of vertebral body surfaces with varying stress regions. Our results showed a strong correlation between the width of the excision and the distribution of stress, particularly in areas with stress levels above 5 MPa. These high-stress areas were concentrated at contact points between the cortical bone and the screws and titanium plates, as well as around the notch and the posterior vertebral body.

However, areas with stress levels between 1 and 3 MPa did not show consistent distribution predictions. Our findings underscore the importance of considering stress concentration as a significant risk factor for vertebral compression. We observed that stress tends to concentrate at the bottom of notches, especially at the lower two corners, similar to how stress concentrates along cracks in intact objects according to classical physics [26,27,28].

It was observed that the best fitting interval for the remaining vertebral body was within the 0–1 MPa region (Rho = 0.679). However, the fitting effect was unsatisfactory in the 3–4 MPa (Rho = 0.462) and 4–5 MPa (Rho = 0.425) regions. There was also a weak correlation between the 0–1 MPa region and all other regions except the 5–6 MPa region. Additionally, when fitting the equation, the coefficient for the height excised in the 0–1 MPa region was three times that of the width excised. In the 1–2 MPa region, the coefficient for widths and heights reversed, with the width showing a negative correlation and the height showing a positive correlation; there was a 12-fold coefficient difference. In the 2–3 MPa region, the width and height were negatively correlated with the distribution area, and the coefficient of the width was four times that of the height. The width-height coefficient in the 3–4 MPa region was similar to that of the 2–3 MPa region, with the width negatively correlated with the distribution area. In the 4–5 MPa region, the width and height showed a positive correlation, and the coefficient of the width was six times that of the height. Finally, the width coefficient was three times and 12 times that of the height in the 5–6 MPa and 6 MPa–Max regions, respectively.

The final area correlation chart showed a consistent trend when comparing areas between 0 and 4 MPa within the same group. Conversely, the opposite trend was consistently observed when comparing areas between 4 MPa and the maximum value within the same group. This means that the two types of areas either increased or decreased together. Therefore, 4 MPa can be considered the threshold at which the area changes. This result suggests that in a specific cervical resection range (1–12 mm, 0–8 mm), a larger vertebral resection can cause the rigid fixed structure to bear more stress. The vertebral area subjected to different stresses will change at approximately 4 MPa, consistent with the surface diagram.

The Mod ACDF is the most effective surgical approach for treating OPLL at specific locations. Its selection strategy offers greater flexibility. Our FEM simulation study has predicted the stress area ratio under various resection strategies, providing a theoretical foundation for the future surgical treatment of patients using the Mod ACDF technique.

Limitations

The FEA is commonly used in engineering to solve mechanical problems numerically. In biomechanics, particularly in the study of the spine, FEA remains crucial. However, this study had some limitations. To simplify the surgical technique, only a rough shape was used to summarize the resected volume. This may have led to inconsistent stress distribution compared to actual operations and failed to account for the bone’s heterogeneity and different surface morphologies [29]. The material properties used in this study were all linear. Despite this, many studies have pointed out that linear material models cannot fully replicate the characteristics of soft tissues like intervertebral discs, which can lead to biased results [30]. Because the model was based on a single patient’s CT images, individual differences could change the results and limit their generalizability. This study only simulated the area of stress distribution in various intervals and provided theoretical support for the future development of the Mod ACDF technique.

The critical stress threshold for changes in the area was approximately 4 MPa. Future studies must further subdivide groups to confirm this threshold value. Additionally, because the maximum stress limit within each model group varied, no additional subdivisions beyond 6 MPa were included, which may have caused the deviation in the 6 MPa–Max group.

Conclusion

The study found that the Mod ACDF technique, used within different resection ranges, produced different stress areas. In a specific resection range in the cervical spine (1–12, 0–8 mm), the conversion of stress areas occurred at a threshold of 4 MPa. The concentrated stress areas were found at the contact points between the vertebral body and the rigid fixator. Our results indicate that biomechanical changes can be a theoretical basis for using modified ACDF to treat OPLL of different sizes.

Fig. 8

In the visualisation surface diagram, the coefficient of the width in all the groups, except in the 0–1 MPa group, was higher than that of the height. The findings of the area percentage model were consistent with those of the total area model. 0–1 MPa (Rho = 0.679): Area = 2832.65 − 6.40 × Width − 19.33 × Height. 1–2 MPa (Rho = 0.966): Area = 908.80 − 25.92 × Width + 2.71 × Height. 2–3 MPa (Rho = 0.910): Area = 459.47 − 9.53 × Width − 2.02 × Height. 3–4 MPa (Rho = 0.462): Area = 241.20 − 2.44 × Width + 0.25 × Height. 4–5 MPa (Rho = 0.425): Area = 119.93 + 3.09 × Width + 0.68 × Height. 5–6 MPa (Rho = 0.822): Area = 31.27 + 6.42 × Width + 2.22 × Height. 6 MPa–Max (Rho = 0.926): Area = 44.18 + 25.78 × Width + 2.47 × Height