Biomechanical role of osteoporosis affects the incidence of adjacent segment disease after percutaneous transforaminal endoscopic discectomy

A 3D lumbosacral model was reconstructed based on the imaging data obtained from an adult male volunteer (L3–S1, three segments of the intervertebral disc (IVD), six facet joints, and six ligaments) without any history of lumbar diseases using computed tomography (CT). The structures in the current model could not be clearly differentiated using CT and have therefore been constructed based on our anatomical observations [13, 19]. In the current study, the vertebra included a cortical shell (thickness 0.8 mm), a trabecular core, superior and inferior endplates (thickness: 0.8 mm), and posterior structures. IVD comprised the annulus fibrosis and inner nucleus, which occupied 44% of the cross-sectional area of the disc, whose location was slightly backward to the disc center, and comprised facet joints including capsule and facet cartilage surfaces (0.25 mm) and ligaments including the anterior longitudinal ligament, posterior longitudinal ligament, ligamentum flavum, intertransverse ligament, interspinous ligament, and supraspinous ligament [18, 20, 21].

The L4–L5 segment, which is one of the most common sites of LDH, was selected for the simulation of PTED. PTED was simulated on the right side. To imitate tears in the annulus fibrosis, a 5-mm incision was made and to represent discectomy, the nucleus in this segment was deleted [22, 23]. Moreover, facetectomy and ligamentum flavum excision are inevitable in PTED, and therefore, sufficient space was created to insert instruments and remove the nucleus. In this, one third of each of these structures were removed [21, 22, 24, 25]. The schematic diagram of current models is shown in Fig. 1, and the magnetic resonance imaging (MRI) date of patients subjected to PTED is shown in Fig. 2.

The reconstruction of osteoporosis FEM models has been described in the scientific literature. The Young modulus in different parts of the current model has been changed, but morphological features remained the same [14, 15, 17]. The Young modulus for cortical, endplate, and posterior structures decreased to 67% of its original value. The Young modulus in cancellous bone decreased to 34% [13‐15].

Boundary and loading conditions were uniformly applied in the three models in this study. To simulate the complicated spinal motion, five different loading conditions, including flexion, extension, left and right bending, and axial torsion, were applied by setting a 10-Nm moment on the superior surface of the lumbosacral model and a 800-N vertical compressive force in the basic loading condition [9, 17].

To reduce the computed error, tetrahedral elements were selected to fill the current model because they seemed suitable for models with complex surfaces; moreover, mesh refinement was set in distortion areas. In current models, the contact type was set as “bounded”; however, for facet surfaces, it was set as “frictionless” [9, 17]. Regarding the material properties, the vertebra and facet cartilages were defined as isotropic, homogeneous materials; the Mooney–Rivlin hyper-elastic material was set in the annulus fibrosis; and the nucleus was defined as an incompressible semifluid material. Moreover, “tension-only” cable elements were used for defining the ligaments (Tables 1 and 2) [13, 18]. In addition, for simplification, the complete lumbosacral model was defined as model 1, the model that underwent PTED with normal BMD as model 2, and the model with osteoporosis as model 3.

The intradiscal pressure in the L4-L5 segment was validated by comparing with a previously published, well-validated in vitro biomechanical study, and the disc compression values were validated by comparing with another in vitro research. To calculate the intradiscal pressure, 300, 1000, and 2000 N of vertical compressive pressure was applied on the superior surface of model 1 and 1200 N of vertical compressive pressure was applied to assess disc compression [22, 25].

At 300, 1000, and 2000 N of vertical compressive force, the intradiscal pressure in our model was 0.2, 0.78, and 1.49 Mpa and that previously reportedly was 0.3 ± 0.09, 0.9 ± 0.26, and 1.85 ± 0.46 Mpa, respectively [22]. Disc compression values in the L3–S1 segment in the current model were 1.7, 1.5, and 1.1 mm and those from a well-validated research were 1.6 ± 0.55, 1.6 ± 0.5, and 1.3 ± 0.5 mm in L3–L4, L4–L5, and L5–S1 segment disc, respectively [25]. Thus, the values in the present study were within one standard deviation from those reported in the well-validated studies except the intradiscal pressure under 300-N compressive pressure (Fig. 3). Considering the only exception listed immediately below the standard deviation and equaling to the lower value in the original data, we believe that the current model can make a credible representation of the lumbosacral spine in actual body positions [22].

Notably, the trend in biomechanical characteristics increased in model 2 in most loading conditions, except for the von Mises stress in the annulus in extension condition and when the endplates were in left lateral bending. Meanwhile, except for the shear stress of the annulus and the von Mises stress in the endplates in the flexion condition, most of the biomechanical characteristics increased by < 5%.

However, this trend dramatically changed with respect to the differences between model 3 and the other two models. Compared with that in model 2, the von Mises stress in endplates was lower in flexion and left lateral bending condition in model 3, and this value in the annulus in extension condition increased in model 3 but was lower than that in model 1. In addition, all the other biomechanical characteristics increased dramatically in model 3 as compared with those in models 1 and 2 (Fig. 4).