A previously validated 3-dimensional, nonlinear ligamentous cervical segment (C3–7) finite element model was used for the present study (Fig. 1)10). The geometrical data of the current multi-segmental cervical model were reconstructed from CT scans of a 26-year old man with no pathologies. The information of the material properties were selected from the literature (Table 1)5–7,10,16,24,25). The model included vertebral bodies, bony posterior elements, intervertebral discs, and six major groups of ligaments: the anterior longitudinal ligament (ALL), the posterior longitudinal ligament (PLL), the ligament flavum, the facet capsular ligament, the interspinous ligament and the supraspinous ligament10). The vertebral body consists of an outer shell of high-strength cortical bone reinforced internally by cancellous bone with a thickness of 0.5 mm (on average)10). The bony posterior element includes the pedicle, lamina, transverse processes, spinous processes and articular processes (facet joint)10). The eight node brick elements are used for body structures whose material properties were assumed to be homogeneous and isotropic3,7,10,25). The origins and insertions of the six ligaments were obtained from a morphological study10). The spinal ligaments adopt the nonlinear load displacement property for the physiological nonlinear behavior of the ligaments6,10). The intervertebral disc is modeled as a fiber-reinforced structure surrounding the incompressible nucleus pulposus10). The reinforcement structure (annulus fibers) is modeled by truss elements with the modified tension-only properties with an orientation of approximately 25°3,10). The facet joint is oriented at 45° from the horizontal plane, where the initial surface gaps between each facet region are assumed to be 0.5 mm based on CT imaging10). The interaction of the facet joints facilitates increased contact force with the narrowing of the initial gap distance between the upper and lower facet surfaces3,10,25). The segmental angular measures used to create the lordotic curve for the model were as follows: C3–4, 4.35°; C4–5, 1.87°; and C5–6, 3.94°8,10). The general-purpose FEA package ABAQUS (Habbitt; Karlsson and Sorensen, Inc., Providence, RI, USA) was used in this study. This study was approved by the institutional review board (Nos. 1512-126-729 and 1607-003-771).

The present study simulated a single-level ACDF at C5–6 without the use of a plate/screw system because C5–6 is the most common surgical level14). The interbody graft was anteriorly implanted after removal of the ALL, the intervertebral disc and the PLL16). We assumed that ideal fusion is the complete incorporation of the interbody graft to the cranial and caudal vertebra. In the model, the interbody graft was completely united with the cranial and caudal vertebrae (Fig. 1). The height of the interbody graft was identical to the original disc height (anterior, 7.36 mm; posterior, 4.46 mm). The property of the interbody graft varies when bone fusion is achieved because the incorporated interbody graft undergoes remodeling21). Therefore, the FEA variables were the property of the incorporated interbody graft and the area of contact surface between the vertebra and interbody graft (Graft-area). The property of the interbody graft ranged from cancellous bone (Young’s modulus 100 or 300 MPa) to cortical bone (10000 MPa)10,16). The morphology of the interbody graft was simplified as a cylindroid shape, and the Graft-area was increased from 10 to 200 mm² in intervals of 10 mm² (Fig. 2). A compressive follower load of 73.6 N and a pure moment of 1.0 Nm were applied on both the flexion and extension motions10,16,17), and the interspinous distance was measured between the spinous processes of C5 and C6 in flexion and extension finite element models (interspinous motion) (Fig. 3). This method is identical to that used in plain radiographs23). The measurements were repeated in three different interbody grafts with properties of 100 MPa, 300 MPa, and 10000 MPa. Similar to the criteria used with plain radiographs, interspinous motion between flexion and extension models of less than 2 mm was considered solid fusion11). To compare the present results with existing commercial interbody cages, the areas of three different polyetheretherketone (PEEK) cages were analyzed14). The areas inside the cages where bone graft filled in (Inside), the PEEK contact surface (PEEK) and the total contact surface (Inside+PEEK) of the MC+^® cage (LDR Medical, Troyes, France), Solis™ cage (Stryker, Kalamazoo, MI, USA), and C7 cage (Medyssey, Jecheon, Korea) are presented in Table 2.

Song et al.22) assessed anterior cervical fusion using CT scans in 254 surgically confirmed cases and suggested that ExGBB was more reliable than InGBB for detecting and assessing cervical fusion. The sensitivity, specificity, positive predictive values and negative predictive values of ExGBB to determine solid bony union were 100%, 86.5%, 70.8%, and 100%, respectively, whereas those of InGBB were 78.8%, 53.2%, 35.5%, and 88.5%, respectively, after the use of a stand-alone cage for ACDF22). This result implies that InGBB is not sufficient for solid fusion and that ExGBB may be necessary for solid bony fusion in most patients22). ExGBB may form sequentially or simultaneously with InGBB during the solid fusion process, and we hypothesized that ExGBB may form to provide the strength required for solid fusion in some patients1,19,22). The formation of ExGBB strengthens the fusion mass; however, it is impossible to control whether ExGBB forms anterior or posterior to an interbody graft15). ExGBB encroached the spinal canal in 21/27 (78%) patients after ACDF with a stand-alone cage and 6/31 (19%) patients after ACDF with an autologous iliac bone and plate fixation15). Although the clinical significance of ExGBB has not been reported, solid bony union without the formation of ExGBB toward the spinal canal may be an ideal condition15).

The present study demonstrated that less Graft-area was required for cortical bone than for cancellous bone. In the two illustrated cases, the same cage (MC+^® cage, depth×width=14×17 mm, inside area 92.27 mm²) and bone substitute were used. However, the first case (Fig. 5A–C) exhibited delayed fusion 24 months postoperatively with the formation of ExGBB. Because the properties of InGBB with a lucent line were lower than those of cortical bone, the necessary Graft-area might be greater than 100 mm², and the formation of ExGBB was inevitable.

The second case was a typical example of successful solid fusion after ACDF. Fusion was achieved within 12 months. CT scans revealed the complete formation and remodeling of InGBB with cortical bone without the formation of ExGBB (Fig. 5D). Although there was a difference between the Graft-area (100 mm² with 10000 MPa cortical bone) and inside area (92.75 mm²) of the cage, the area and property of InGBB in the second case appeared to be sufficient for solid fusion. Although the surgical level was the same, the following biological characteristics of the two patients were not equal: age and gender. Old age might delayed the fusion process and result in the formation of InGBB with weak bone quality. Therefore, the Graft-area with lower property of InGBB might not be sufficient, and ExGBB was formed. These two cases demonstrate that both Graft-area and InGBB properties influence the formation of ExGBB.

Cervical interbody cages have evolved based on clinical outcomes, fusion rates, the incidence of subsidence and complications2). However, the issue of ExGBB has not been addressed in this evolution. Although the designs of the example cages analyzed in the present study are similar to those of mainstream cages, the areas inside the commercial cages are smaller than the minimal Graft-area for solid fusion with only InGBB. Although the biomechanical effect of the cage-vertebral endplate was not assessable, it seemed that the Graft-area was insufficient for solid bony union even when the area of PEEK was considered, potentially leading to the formation of ExGBB.

Increasing the area inside the cage to increase the contact area of InGBB may be a solution, but decreased cage stiffness should be considered. If the area of InGBB cannot be increased with current designs, osteointegration of the cage itself into the vertebra through the bioactivation of surface materials or expansion of the contact area of the cage may increase areas of fusion mass and reduce the requirement for ExGBB2). With the currently available cages, the application of a plate/screw system might reduce the formation of ExGBB15).

This study has several limitations. First, solid bony union with complete remodeling of bone graft was not always a necessary condition for successful clinical and radiological outcomes. Because of various factors such as supporting muscles, ligaments, and the biomechanical effect between the cage-vertebral endplate and ExGBB, successful outcomes have been reported after ACDF with a stand-alone cage19,22). In the present study, we used the liberal criterion of 2 mm11) based on the above-mentioned factors. However, the cut-off for successful clinical and radiological outcomes in practice may differ from that in the present study, as shown in case 2. A postoperative radiological analysis is required to define the exact Graft-area for solid fusion.

Second, InGBB and ExGBB may be formed simultaneously during the fusion process. Therefore, ExGBB may be formed even in the presence of adequate InGBB contact area because of the insufficient strength of InGBB until the remodeling of InGBB is complete (Figs. 4 and 5). However, we could not simulate such scenarios with the current FEA.

Third, the clinical significance of ExGBB was not addressed. The formation of ExGBB might be the process toward a successful arthrodesis. Moreover, the influences of the biological factors of patients and graft materials were not considered in FEA. Therefore, the hypothesis of the present study and suggestions for future innovation may be appropriate only after confirmation of the clinical significance of ExGBB15). The present result was obtained from FEA, and these limitations of FEA must be considered when interpreting the present results. Nonetheless, solid fusion with the formation of ExGBB may be an unpredictable radiological outcome. The present study may prove beneficial for highlighting the issue of ExGBB and suggesting one causative factor for the formation of ExGBB with suggestions for refinement.