Biomechanical effects of endoscopic ULBD in a reduced‑stiffness lumbar spine model: focus on endplate stress and hypothesis‑generating risk indicators (a finite element analysis) Authors: Wei Wei, Zihao Liu, Xing Zhou, Lin Zhao, Mindong Lu, Zhanxiang Wu, Sheng Zhou, Jie Liu Published: 2026 ABSTRACT This finite element study aimed to compare the biomechanical effects of endoscopic unilateral laminotomy for bilateral decompression (Endo-ULBD) on lumbar spinal stability and load distribution in both healthy and osteoporotic spine models, with a specific focus on identifying biomechanical changes in a reduced‑stiffness bone model that may serve as hypothesis‑generating risk indicators, recognizing that the model does not simulate full osteoporotic pathology. A validated L4–L5 finite element model was developed. Virtual Endo-ULBD was performed on both healthy and reduced‑stiffness (osteoporosis‑simulating) models. Segmental range of motion (ROM) and von Mises stress in the intervertebral disc and endplates were evaluated under flexion, extension, lateral bending, and axial rotation, with a 500 N compressive preload and 10 N·m pure moments. Nonlinear mechanical characteristics of spinal tissues and the potential modulating effect of paraspinal musculature were discussed as key contextual factors for interpreting in vitro results. The model was successfully validated against established biomechanical data. After Endo-ULBD, ROM percentage increases ranged from 2.80 to 16.90% in the healthy model and from 3.00 to 17.10% in the reduced‑stiffness (osteoporosis‑simulating) model (depending on loading direction), with all absolute ROM values remaining below 10°. The reduced‑stiffness (osteoporosis‑simulating) model showed slightly higher absolute ROM values than the healthy model both preoperatively and postoperatively, consistent with reduced bone stiffness. The percentage increases after Endo‑ULBD were similar between the two models. Disc stress showed increases ranging from 3.88 to 25.78% in the healthy model and from 2.47 to 10.48% in the reduced‑stiffness model. There is no established threshold to determine whether these increases are biomechanically significant. Endplate stress was markedly elevated in the reduced‑stiffness model (average 13.5% increase), suggesting a hypothesis‑generating concern for fracture risk. Endo-ULBD suggests preserved segmental stability under the tested loading conditions, with only modest changes in ROM in both healthy and reduced‑stiffness (osteoporosis‑simulating) models. These findings are computational predictions and require clinical validation. The ROM increase observed in our reduced‑stiffness (osteoporosis‑simulating) model (similar to that of the healthy model) should not be directly extrapolated to clinical osteoporosis, as our model does not include the degenerative changes (disc degeneration, facet hypertrophy, ligament thickening) that typically coexist with osteoporosis in elderly patients. Future models that incorporate both bone quality deterioration and age‑related soft‑tissue degeneration are needed to determine whether osteoporotic patients truly have a lower risk of iatrogenic instability after Endo‑ULBD. For now, the present finding is hypothesis‑generating only. However, the marked rise in endplate stress in osteoporotic bone suggests a potential biomechanical mechanism for increased fracture risk. It is important to note that our reduced‑stiffness (osteoporosis‑simulating) model represents reduced bone stiffness (elastic modulus) only, without simulating microarchitectural deterioration such as trabecular loss or cortical thinning. Therefore, the absolute magnitude of clinical risk cannot be directly inferred from these computational results. Instead, the findings should be considered hypothesis‑generating and require validation using more sophisticated models and clinical studies. It is important to note that our reduced‑stiffness (osteoporosis‑simulating) model represents only reduced bone stiffness and does not simulate microarchitectural heterogeneity (trabecular thinning, cortical porosity). These findings advance current knowledge by defining a bone quality‑specific biomechanical risk profile for Endo‑ULBD, suggesting that while segmental stability is preserved, osteoporotic spines are at higher biomechanical risk for endplate-related complications. This suggests that bone quality may be a relevant factor for further investigation in future clinical studies, but no surgical recommendations can be derived from this computational work. Introduction Degenerative lumbar spinal stenosis (DLSS) is a prevalent spinal disorder characterized by low back pain and radicular symptoms, primarily resulting from degenerative changes in the intervertebral discs and facet joints[1]. As a leading cause of disability worldwide, its incidence continues to rise with aging populations[2, 3]. Surgical intervention is indicated when conservative treatments fail[4]. While traditional open decompression effectively relieves neural compression, it often damages the posterior ligamentous complex and may compromise spinal stability. Instrumented fusion, though effective, carries risks such as implant loosening, cage subsidence, prolonged surgery, and significant tissue trauma[5]. In recent years, spinal endoscopic techniques have gained popularity as minimally invasive alternatives that achieve adequate decompression while preserving spinal stability[6]. Endoscopic unilateral laminotomy for bilateral decompression (Endo-ULBD) enables bilateral nerve root decompression through a unilateral approach, sparing the spinous process, contralateral lamina, and most posterior musculoligamentous structures[7–9]. Clinical studies have confirmed its efficacy in treating lumbar spinal stenosis, offering a viable alternative to conventional procedures[9, 10]. However, existing research has not yet defined how bone quality (e.g., osteoporosis) modulates the biomechanical consequences of Endo-ULBD, nor has it translated objective changes in endplate stress into biomechanically informed considerations that require clinical validation for osteoporotic patients—a critical gap given the high comorbidity of osteoporosis and DLSS in the elderly. Osteoporosis (OP), a metabolic bone disease marked by reduced bone mass and deteriorated microarchitecture, significantly increases fracture risk [11]. Globally, osteoporosis contributes to approximately 8.9 million fractures annually [12, 13]. In elderly patients, osteoporotic changes frequently coexist with disc degeneration, where diminished bone quality may accelerate spinal degenerative processes [14]. To our knowledge, no previous finite element study has specifically investigated the influence of osteoporosis on Endo-ULBD. Finite element analysis (FEA) is a powerful computational tool for simulating mechanical behavior in complex anatomical structures. With advances in computational power and modeling techniques, FEA has been increasingly applied to study lumbar degenerative pathologies. However, current FE models of spinal degeneration often lack biomechanical evaluation under osteoporotic conditions. Therefore, this study aims to go beyond confirming the stability-preserving nature of Endo-ULBD and instead compare the biomechanical effects of the procedure in normal and osteoporotic lumbar spine models with a focus on: (1) identifying bone quality‑specific changes in load distribution (e.g., endplate stress); (2) discussing these biomechanical changes as hypothesis‑generating indicators that could inform future clinical studies; and (3) discussing the limitations of linear FE modeling and the role of paraspinal musculature in interpreting study results. We hypothesize that Endo-ULBD will maintain segmental stability in both conditions but may lead to altered load distribution—particularly elevated endplate stress in osteoporosis—that increases the risk of stress‑related spinal complications. The novelty of this work is threefold. First, while previous finite element studies on ULBD focused almost exclusively on normal bone, we directly compare healthy and reduced‑stiffness (osteoporosis‑simulating) model to establish a bone quality‑specific biomechanical risk profile for Endo‑ULBD. Second, we explicitly discuss objective changes in endplate stress as potential biomechanical indicators that could be investigated in future clinical studies for risk assessment for osteoporotic patients (endplate fracture, vertebral body collapse, accelerated disc degeneration)—an unaddressed gap in the existing literature. Third, we provide a transparent discussion of the methodological limitations of linear FE modeling (absence of viscoelasticity, musculature, microarchitectural heterogeneity) and explain how these simplifications affect clinical interpretation, thereby enhancing the clinical applicability of our findings. This study thus provides a biomechanically informed, hypothesis‑generating framework that may guide future research into surgical considerations for osteoporotic patients. Materials and methods The study protocol was approved by the Fourth People’s Hospital of Tianshui City’s Institutional Review Board(SSDW-2123-23S1), and all procedures conformed to the ethical principles of the Declaration of Helsinki. Written informed consent was obtained from all participants whose imaging data were used. Design This was a computational finite element (FE) study designed to evaluate the biomechanical properties of the lumbar spine following endoscopic unilateral laminotomy for bilateral decompression (Endo-ULBD). The study compared two conditions: a normal bone model and an osteoporotic bone model. The workflow consisted of three main phases: (1) development and validation of a baseline L4–L5 FE model; (2) virtual simulation of the Endo-ULBD procedure on both models; and (3) comparative analysis of segmental range of motion (ROM), intervertebral disc stress, and endplate stress under physiological loading conditions. Model development: imaging source and software The baseline geometry of the L4–L5 functional spinal unit was derived from computed tomography (CT) images of a healthy adult male volunteer (age 32 years) without a history of lumbar pathology. It is important to note that the patient images shown in Fig. 1B (a 72‑year‑old female with DLSS and osteoporosis) are for illustrative purposes only and were not used to construct or parameterize the finite element model. The model was based exclusively on the healthy 32‑year‑old male volunteer described above. Imaging was performed using a Siemens dual-source spiral CT scanner (Siemens AG, Munich, Germany), producing 473 contiguous axial slices with a thickness of 0.625 mm, stored in DICOM format. The initial mesh was refined in Geomagic Wrap through feature removal, noise reduction, manual editing (e.g., removal of spurious triangles, filling holes), and smoothing operations (using a low‑pass filter to preserve anatomical details while eliminating surface irregularities) to achieve anatomical accuracy.Fig. 1CT imaging and three-dimensional reconstruction of the lumbar spine. A: Representative three-dimensional CT reconstructions of the L4–L5 segment from the healthy volunteer (32‑year‑old male) who served as the source for the finite element model described in this study. B: Illustrative images from a 72‑year‑old female patient with degenerative lumbar spinal stenosis (DLSS) and osteoporosis who was treated with Endo‑ULBD. These patient images are shown for clinical context only and were not used to construct the finite element model. Using Mimics 21.0 software, the vertebral bodies were reconstructed based on CT data. The software interface displays the coronal view (upper left), sagittal view (lower left), axial view (upper right), and the three‑dimensional reconstructed model (lower right). CT imaging and three-dimensional reconstruction of the lumbar spine. A: Representative three-dimensional CT reconstructions of the L4–L5 segment from the healthy volunteer (32‑year‑old male) who served as the source for the finite element model described in this study. B: Illustrative images from a 72‑year‑old female patient with degenerative lumbar spinal stenosis (DLSS) and osteoporosis who was treated with Endo‑ULBD. These patient images are shown for clinical context only and were not used to construct the finite element model. Using Mimics 21.0 software, the vertebral bodies were reconstructed based on CT data. The software interface displays the coronal view (upper left), sagittal view (lower left), axial view (upper right), and the three‑dimensional reconstructed model (lower right). The following software packages were used in the modeling pipeline: Materialise Mimics 21.0 for initial segmentation and 3D reconstruction; Geomagic Wrap 2021 for mesh optimization and surface smoothing; SolidWorks 2021 for assembly and component design; ANSYS Workbench 17.0 for material assignment, meshing, and finite element analysis. Finite element model construction The DICOM data were imported into Mimics, where threshold-based segmentation was used to isolate bone tissue and generate a preliminary 3D model of the L4 and L5 vertebrae in STL format. The initial mesh was refined in Geomagic Wrap through feature removal, noise reduction, and surface smoothing to achieve anatomical accuracy. The vertebral models were imported into SolidWorks and assembled in their anatomical alignment. The cortical shell and cancellous core of each vertebra were modeled as separate components. The intervertebral disc was constructed to include the nucleus pulposus, annulus fibrosus, and superior/inferior endplates. The facet joint cartilage was also modeled as a separate interface. All components were assembled without geometric interference [15]. Material properties and meshing Material properties were assigned based on published literature to represent physiological spinal tissues (Table 1). The osteoporotic condition was simulated by reducing the elastic modulus of both cortical and cancellous bone according to established relationships between bone mineral density and mechanical properties. It is critical to emphasize that this approach captures only the reduction in bone stiffness (elastic modulus). It does not incorporate microarchitectural features of true osteoporotic bone, including trabecular thinning, trabecular perforation, cortical porosity, or regional variations in bone density. Thus, the model represents a reduced‑stiffness condition rather than a full representation of osteoporosis. This simplification is a deliberate limitation of the present study and should be considered when interpreting all results involving the reduced‑stiffness (osteoporosis‑simulating) model. The rationale for using uniform modulus reduction is as follows. First, this approach is widely accepted in lumbar spine finite element studies for parametric comparison between healthy and osteoporotic conditions[16–18]. Second, it is based on established linear or power‑law relationships between bone mineral density (BMD) and Young’s modulus[19]. Third, our primary objective was to isolate the effect of reduced bone stiffness on Endo‑ULBD biomechanics, not to capture all microarchitectural features of osteoporosis. Fourth, we explicitly acknowledge the limitations of this simplification.Table 1Material parameters assigned to components of the lumbar finite element model.ComponentYoung’s modulus (MPa)Poisson’s ratioCross-sectional area (mm2)Cortical boneShim et al., 2008 Normal12,0000.3–Morgan et al., 2003 Osteoporotic80400.3–Goel et al., 1995 Severe osteoporotic64320.3–Keller, 1994Cancellous bonePolikeit et al., 2004 Normal1000.2–Liu et al., 2015 Osteoporotic340.251–Argoubi & Shirazi-Adl, 1996 Severe osteoporotic27.20.251–Sharma et al., 1995Endplate6700.401–Sharma et al., 1995Annulus fibrosus4.20.45–Sharma et al., 1995Nucleus pulposus10.499–Sharma et al., 1995Articular cartilage100.401–Goel et al., 1995Anterior longitudinal ligament200.30160Goel et al., 1995Posterior longitudinal ligament700.30122Goel et al., 1995Ligamentum flavum500.30161[14]Capsular ligament250.30167.6[14]Values are given for normal, osteoporotic, and severe osteoporotic conditions where applicable. Note: The osteoporotic and severe osteoporotic material properties represent only a reduction in elastic modulus. Microarchitectural features such as trabecular thinning, trabecular perforation, and cortical porosity are not simulated. Table 1Material parameters assigned to components of the lumbar finite element model.ComponentYoung’s modulus (MPa)Poisson’s ratioCross-sectional area (mm2)Cortical boneShim et al., 2008 Normal12,0000.3–Morgan et al., 2003 Osteoporotic80400.3–Goel et al., 1995 Severe osteoporotic64320.3–Keller, 1994Cancellous bonePolikeit et al., 2004 Normal1000.2–Liu et al., 2015 Osteoporotic340.251–Argoubi & Shirazi-Adl, 1996 Severe osteoporotic27.20.251–Sharma et al., 1995Endplate6700.401–Sharma et al., 1995Annulus fibrosus4.20.45–Sharma et al., 1995Nucleus pulposus10.499–Sharma et al., 1995Articular cartilage100.401–Goel et al., 1995Anterior longitudinal ligament200.30160Goel et al., 1995Posterior longitudinal ligament700.30122Goel et al., 1995Ligamentum flavum500.30161[14]Capsular ligament250.30167.6[14]Values are given for normal, osteoporotic, and severe osteoporotic conditions where applicable. Note: The osteoporotic and severe osteoporotic material properties represent only a reduction in elastic modulus. Microarchitectural features such as trabecular thinning, trabecular perforation, and cortical porosity are not simulated. Material parameters assigned to components of the lumbar finite element model. Values are given for normal, osteoporotic, and severe osteoporotic conditions where applicable. Note: The osteoporotic and severe osteoporotic material properties represent only a reduction in elastic modulus. Microarchitectural features such as trabecular thinning, trabecular perforation, and cortical porosity are not simulated. All materials were modeled as linearly elastic and isotropic for computational feasibility—a deliberate simplification that acknowledges the nonlinear, viscoelastic, and time‑dependent material properties of in vivo spinal tissues (e.g., intervertebral disc, ligaments). Specifically, the annulus fibrosus, nucleus pulposus, endplates, articular cartilage, and all ligaments were assigned constant Young’s moduli and Poisson’s ratios (Table 1) without strain‑rate dependence, viscoelastic creep, or hyperelastic behavior. This simplification was adopted because (1) our primary objective was to compare relative changes between healthy and osteoporotic conditions under identical static loading, where linear‑elastic models have been widely used in previous ULBD finite element studies, and (2) computational resources for non‑linear, time‑dependent simulations of multiple loading conditions across two bone quality models were prohibitive. However, we acknowledge that this approach does not capture the viscoelastic response of spinal tissues under cyclic or sustained loading, such as prolonged standing or repetitive motion [20]. The model was meshed in ANSYS Workbench using tetrahedral elements. A systematic mesh convergence study was conducted to ensure solution independence (see Mesh Convergence Study section). The final element sizes (determined from the convergence study) were: cortical bone 2.0 mm, cancellous bone 2.0 mm, endplates 0.5 mm, annulus fibrosus and nucleus pulposus 1.5 mm, and articular cartilage 1.0 mm. Final element and node counts are provided in supplement Table 1. These values are the outcome of the convergence study, not arbitrarily chosen from a reference[21]. All ligaments (anterior longitudinal, posterior longitudinal, ligamentum flavum, capsular, interspinous, supraspinous) were modeled as tension‑only spring elements with stiffness values derived from published experimental studies[22, 23]. The cross‑sectional areas for each ligament are provided in Table 1. Within the physiological range of motion, the force‑strain relationship of ligaments is approximately linear; therefore, linear stiffness was used as an approximation. We acknowledge that this simplification does not capture the non‑linear toe region or viscoelastic behavior of ligaments, which is a limitation of the present model. Hereafter, this condition is referred to as the ‘reduced‑stiffness (osteoporosis‑simulating) model’ to avoid implying full representation of clinical osteoporosis. Mesh convergence study To ensure solution independence, a systematic mesh convergence study was performed. Four sequentially refined tetrahedral meshes were generated for the L4–L5 functional spinal unit: very coarse (element sizes 150% of final), coarse (120%), fine (100%, final mesh), and very fine (70%). The percentage change in two output quantities—peak von Mises stress on the L5 endplate and segmental range of motion (ROM) under extension loading (10 N·m moment + 500 N preload)—was monitored between successive refinements. As shown in Figure S1 (Supplementary Materials), the relative error in peak endplate stress decreased from 75.1% (very coarse mesh) to 2.49% (fine mesh), and ROM error decreased from 71.8 to 1.79%, both below the predefined 5% convergence threshold. Convergence was defined as a relative change < 5% between the fine and very fine meshes. The fine mesh achieved a change of 2.1% for peak stress and 1.3% for ROM, well below the 5% threshold. Therefore, the fine mesh was selected for all subsequent simulations. Supplement Table 1 lists the final element and node counts for each component. This level of mesh density meets the minimum reporting requirements for computational solid mechanics publications[24, 25]. Boundary and loading conditions Throughout computational simulations, complete constraint conditions were applied to the inferior surface of the L5 vertebra[26]. A 500 N compressive preload, representing approximately two-thirds of standard body mass, was distributed across the superior surface of L4 [27]. Evaluation of segmental kinematics was conducted by applying 10 N·m pure moments to the superior L4 node, reproducing flexion–extension, lateral bending, and axial rotation movements[28]. ROM quantification involved six repeated measurements at distinct reference markers on the L4 superior surface, with model trajectory patterns continuously monitored. Maximum von Mises stress values within disc tissues were computationally derived by the ANSYS solver without requiring manual quantification. Virtual surgical simulation The Endo-ULBD procedure was simulated according to established surgical principles and previously published finite element studies on ULBD[29, 30]. To ensure reproducibility, all resection parameters were defined quantitatively, as summarized in Table 5 (new table). The simulation steps were as follows: ①Ipsilateral decompression: On the ipsilateral (approach) side, a laminotomy was performed with the following dimensions: longitudinal length (cephalocaudal) = 12 mm, transverse width (mediolateral) = 8 mm, starting from the medial border of the ipsilateral pedicle and extending to the base of the spinous process. The laminotomy removed the inferior half of the L4 lamina and the superior half of the L5 lamina (full thickness, approximately 3–4 mm). A medial facetectomy was then performed, removing the medial 30–40% of the ipsilateral inferior articular process of L4 and superior articular process of L5, corresponding to approximately 30% of the total facet joint volume. This exposed the lateral recess and the traversing nerve root. The ligamentum flavum was completely resected from its superior attachment (deep surface of the lamina) to its inferior attachment (upper margin of the lower lamina) on the ipsilateral side. ②Contralateral decompression: Using the spinous process base as a pivot, the surgical trajectory was angled 25–30° contralaterally. Contralateral decompression was achieved by undercutting the base of the spinous process at its junction with the contralateral lamina. The undercut depth was 8 mm from the midline, and the thickness of bone removed was approximately 3 mm (cortical bone only, preserving the cancellous core). This maneuver exposed the contralateral lateral recess. The contralateral ligamentum flavum was removed in its entirety. The decompression extended to the medial border of the contralateral pedicle, but the contralateral facet joint was completely preserved (no resection). The supraspinous ligament, interspinous ligament, and the spinous process itself were all preserved in the simulation. ③Verification: The resected bone volumes were measured from the virtual model: laminotomy volume = 245 mm3, ipsilateral facetectomy volume = 128 mm3, spinous process undercut volume = 92 mm3. These values are consistent with clinical reports of endoscopic ULBD[31]. It is important to emphasize that these resection parameters represent a standardized, idealized decompression in a model without pre‑existing severe degenerative changes (e.g., facet hypertrophy, ligamentum flavum thickening, disc collapse). In clinical practice, elderly osteoporotic patients with advanced lumbar spinal stenosis often require more extensive facet resection (exceeding 40%) and ligamentum removal to achieve adequate neural decompression. Therefore, our simulated resection likely represents a lower‑bound (best‑case) scenario in terms of structural preservation. Data extraction and analysis Segmental ROM was calculated as the angular displacement of L4 relative to L5. The von Mises stress within the intervertebral disc and on the superior endplate of L5 was extracted for each load case. Since the normal and reduced‑stiffness (osteoporosis‑simulating) models shared identical geometry, direct comparison of absolute values and percentage changes was performed. Percentage change was calculated as: (postoperative value—preoperative baseline)/preoperative baseline × 100%. Biomechanical stability was defined as a postoperative ROM increase of less than 20% compared to the intact condition and an absolute ROM value below 10°, consistent with previously published criteria. Endplate von Mises stress was extracted from the superior endplate of L5 (the cephalad endplate of the L5 vertebra) for each loading case. Values were recorded as the maximum von Mises stress within the endplate element set. Model validation The intact (preoperative) FE model was validated by comparing its predicted ROM under pure moment loading with published experimental data from Shim et al. (2008). The results showed close agreement across all motion directions (Table 2), confirming the validity of the modeling approach for subsequent surgical simulations. It is important to note, however, that only the healthy (normal bone) model was validated against published data. The reduced‑stiffness (osteoporosis‑simulating) model, created by proportionally reducing the elastic modulus of bone, was not independently validated against experimental or clinical reference data because such data for osteoporotic lumbar functional spinal units under identical loading conditions (500 N preload + 10 N·m pure moments) are not available in the literature. This lack of pathological validation is a limitation that should be considered when interpreting the results specific to the osteoporotic condition. A P‑value > 0.05 was considered to indicate no significant difference between the model and the reference data.Table 2Validation of the intact finite element model: comparison of range of motion (ROM in degrees) between the present study and previously published experimental data.Operating conditions in FEPresent researchShimP valueFlexion5.12 ± 0.055.48 ± 0.880.34Extension2.88 ± 0.082.79 ± 0.420.44Left bending4.42 ± 0.044.45 ± 1.010.21Right bending4.38 ± 0.024.45 ± 1.010.11Left rotation3.62 ± 0.353.80 ± 0.990.09Right rotation3.99 ± 0.083.80 ± 0.990.32P‑values calculated using one‑sample t‑test comparing present model mean (n = 6) to Shim et al. mean. Table 2Validation of the intact finite element model: comparison of range of motion (ROM in degrees) between the present study and previously published experimental data.Operating conditions in FEPresent researchShimP valueFlexion5.12 ± 0.055.48 ± 0.880.34Extension2.88 ± 0.082.79 ± 0.420.44Left bending4.42 ± 0.044.45 ± 1.010.21Right bending4.38 ± 0.024.45 ± 1.010.11Left rotation3.62 ± 0.353.80 ± 0.990.09Right rotation3.99 ± 0.083.80 ± 0.990.32P‑values calculated using one‑sample t‑test comparing present model mean (n = 6) to Shim et al. mean. Validation of the intact finite element model: comparison of range of motion (ROM in degrees) between the present study and previously published experimental data. P‑values calculated using one‑sample t‑test comparing present model mean (n = 6) to Shim et al. mean. Results Model validation The intact L4–L5 finite element model demonstrated good agreement with published biomechanical data. Under pure moment loading, the predicted range of motion (ROM) values closely matched those reported by Shim et al. (2008) across all six loading directions (flexion, extension, left/right lateral bending, left/right axial rotation). No significant differences were found between the present model and the reference dataset (all P > 0.05), confirming the validity of the modeling approach (Table 2). However, these ROM increases should be interpreted within the context of an idealized decompression model without severe pre‑existing degenerative changes. In patients with marked facet hypertrophy or ligamentous thickening, the required resection may be larger, potentially resulting in greater postoperative ROM increases than reported here. Segmental range of motion after endo-ULBD Endo-ULBD led to a modest increase in segmental ROM in both the healthy and reduced‑stiffness (osteoporosis‑simulating) models under all loading conditions (Table 3). It should be noted that while the healthy model was validated against literature data, the reduced‑stiffness (osteoporosis‑simulating) model lacked independent validation. In the healthy model, postoperative ROM percentage increases ranged from 2.80% (left bending) to 16.90% (right bending). In the reduced‑stiffness (osteoporosis‑simulating) model, the range was 3.00% (left bending) to 17.10% (right bending). The two models showed similar percentage increases, with the reduced‑stiffness (osteoporosis‑simulating) model exhibiting slightly higher values in some loading directions (e.g., right bending). All absolute ROM values remained below 10°, consistent with biomechanical stability criteria (Figs. 2, 3).Table 3Range of motion (ROM, in degrees) and percent change before and after Endo-ULBD in healthy and reduced‑stiffness (osteoporosis‑simulating) model.Loading conditionModelPreoperative (deg)Postoperative (deg)Percent change (%)FlexionHealthy3.674.019.30Osteoporotic3.824.199.70ExtensionHealthy4.655.2312.50Osteoporotic4.825.4212.40Left bendingHealthy3.243.332.80Osteoporotic3.363.463.00Right bendingHealthy2.422.8316.90Osteoporotic2.512.9417.10Left rotationHealthy2.312.6715.60Osteoporotic2.42.7815.80Right rotationHealthy2.782.997.60Osteoporotic2.883.118.00Percent change calculated as (postoperative—preoperative)/preoperative × 100%. Positive values indicate an increase. Values are mean ± SD.Fig. 2Workflow of the finite element model development and loading simulation.A: Coronal and sagittal contours of the lumbar spine with mask creation for initial skeletal reconstruction.B: Generation of patch boundaries comprising contour lines and bounding curves.C: NURBS surface fitting applied to the vertebral geometry.D: Modeling of disc components: superior endplate, annulus fibrosus, and nucleus pulposus. E: Complete intact (theoretical) finite element model with smooth surfaces of the disc and vertebral bodies. F: Degenerated model representing osteoporotic condition with proliferative bony changes. G: Application of a 500 N compressive preload to the superior surface of L4. H: Segmental range of motion (ROM) testing under 10 N·m pure moment applied to the superior node of L4 to simulate flexion, extension, lateral bending, and axial rotation.Fig. 3Distribution of von Mises stress on the superior endplate of the L5 vertebra in healthy and osteoporotic (OP) models under six loading conditions. Stress contours are shown for flexion (FL), extension (EX), left bending (LB), right bending (RB), left rotation (LR), and right rotation (RR). The color scale represents von Mises stress in MPa. For exact numerical values, see Table 6.Stress distribution remains generally uniform without evidence of focal stress concentration. Table 3Range of motion (ROM, in degrees) and percent change before and after Endo-ULBD in healthy and reduced‑stiffness (osteoporosis‑simulating) model.Loading conditionModelPreoperative (deg)Postoperative (deg)Percent change (%)FlexionHealthy3.674.019.30Osteoporotic3.824.199.70ExtensionHealthy4.655.2312.50Osteoporotic4.825.4212.40Left bendingHealthy3.243.332.80Osteoporotic3.363.463.00Right bendingHealthy2.422.8316.90Osteoporotic2.512.9417.10Left rotationHealthy2.312.6715.60Osteoporotic2.42.7815.80Right rotationHealthy2.782.997.60Osteoporotic2.883.118.00Percent change calculated as (postoperative—preoperative)/preoperative × 100%. Positive values indicate an increase. Values are mean ± SD. Range of motion (ROM, in degrees) and percent change before and after Endo-ULBD in healthy and reduced‑stiffness (osteoporosis‑simulating) model. Percent change calculated as (postoperative—preoperative)/preoperative × 100%. Positive values indicate an increase. Values are mean ± SD. Workflow of the finite element model development and loading simulation.A: Coronal and sagittal contours of the lumbar spine with mask creation for initial skeletal reconstruction.B: Generation of patch boundaries comprising contour lines and bounding curves.C: NURBS surface fitting applied to the vertebral geometry.D: Modeling of disc components: superior endplate, annulus fibrosus, and nucleus pulposus. E: Complete intact (theoretical) finite element model with smooth surfaces of the disc and vertebral bodies. F: Degenerated model representing osteoporotic condition with proliferative bony changes. G: Application of a 500 N compressive preload to the superior surface of L4. H: Segmental range of motion (ROM) testing under 10 N·m pure moment applied to the superior node of L4 to simulate flexion, extension, lateral bending, and axial rotation. Distribution of von Mises stress on the superior endplate of the L5 vertebra in healthy and osteoporotic (OP) models under six loading conditions. Stress contours are shown for flexion (FL), extension (EX), left bending (LB), right bending (RB), left rotation (LR), and right rotation (RR). The color scale represents von Mises stress in MPa. For exact numerical values, see Table 6.Stress distribution remains generally uniform without evidence of focal stress concentration. Stress distribution in the intervertebral disc and endplates Postoperative von Mises stress within the intervertebral disc increased slightly in both groups under all loading conditions (Table 4, Fig. 4). In the healthy model, disc stress increases ranged from 3.88 to 25.78%, and in the reduced‑stiffness (osteoporosis‑simulating) model from 2.47 to 10.48%. The magnitude of these increases varied by loading direction, with the largest being 25.78% (healthy model, extension). However, no validated threshold exists to determine whether such increases are biomechanically or clinically significant. The observed absolute disc stress values remained within the range of previously reported physiological disc stresses under comparable loading conditions (e.g., 1.0–3.0 MPa for lumbar discs under 10 N·m moments) (Table 5, Fig. 5).Table 4Disc von Mises stress (MPa) and percent change before and after Endo-ULBD in healthy and reduced‑stiffness (osteoporosis‑simulating) model (mean ± SD).LoadingModelPreoperativePostoperativePercent change (%)FlexionHealthy1.28 ± 0.521.38 ± 0.237.81Osteoporotic1.39 ± 0.221.45 ± 0.124.32ExtensionHealthy2.25 ± 0.192.83 ± 0.5125.78Osteoporotic2.48 ± 0.422.74 ± 0.3910.48Left bendingHealthy1.55 ± 0.411.63 ± 0.355.16Osteoporotic1.62 ± 0.021.66 ± 0.182.47Right bendingHealthy1.29 ± 0.741.42 ± 0.3610.08Osteoporotic1.35 ± 0.211.45 ± 0.527.41Left rotationHealthy1.29 ± 0.371.34 ± 0.523.88Osteoporotic1.38 ± 0.231.44 ± 0.204.35Right rotationHealthy1.35 ± 0.281.51 ± 0.5211.85Osteoporotic1.52 ± 0.381.62 ± 0.336.58Percent change calculated as (postoperative—preoperative)/preoperative × 100%. Positive values indicate an increase. All changes are increases. The largest increase (25.78% in healthy extension) is highlighted in bold. There is no established biomechanical or clinical threshold for what constitutes a ‘negligible’ versus ‘significant’ disc stress increase after decompression. Therefore, these percentage changes should be interpreted as descriptive only.Fig. 4Equivalent (von Mises) stress distribution within the intervertebral disc under six loading conditions. All images are shown at 600 dpi for clarity. The color scale (blue to red) represents von Mises stress in MPa, with blue indicating low stress (0 MPa) and red indicating high stress (maximum values as indicated on the color bar). Comparison between healthy (left column for each loading condition) and osteoporotic (OP, right column) models after Endo‑ULBD. Stress contours are displayed for flexion (FL), extension (EX), left bending (LB), right bending (RB), left rotation (LR), and right rotation (RR). Note that the stress distribution remains homogeneous in both models, with no focal stress concentrations or abnormal disc bulging. The color bar range is consistent across all panels to facilitate direct visual comparison.Table 5Quantitative parameters of the virtual Endo-ULBD simulation.ParameterValueIpsilateral laminotomy Longitudinal length (cephalocaudal)12 mm Transverse width (mediolateral)8 mm LocationFrom medial border of ipsilateral pedicle to spinous process base Bone thickness removed3–4 mm (full thickness)Ipsilateral medial facetectomy Percentage of facet joint resected30–40% (medial portion) Resected volume128 mm3Contralateral undercutting Undercut depth from midline8 mm Bone thickness removed3 mm (cortical only) Resected volume (spinous process base)92 mm3Ligamentum flavum resection IpsilateralComplete from superior to inferior attachment ContralateralComplete Preserved structuresSupraspinous ligament, interspinous ligament, entire contralateral facet joint, spinous ProcessFig. 5Schematic representation of loading and boundary conditions. A 500 N compressive preload is distributed on the superior surface of L4. A 10 N·m pure moment is applied to the superior node of L4 to simulate flexion, extension, lateral bending, and axial rotation. The inferior surface of L5 is fully constrained (all degrees of freedom fixed). Table 4Disc von Mises stress (MPa) and percent change before and after Endo-ULBD in healthy and reduced‑stiffness (osteoporosis‑simulating) model (mean ± SD).LoadingModelPreoperativePostoperativePercent change (%)FlexionHealthy1.28 ± 0.521.38 ± 0.237.81Osteoporotic1.39 ± 0.221.45 ± 0.124.32ExtensionHealthy2.25 ± 0.192.83 ± 0.5125.78Osteoporotic2.48 ± 0.422.74 ± 0.3910.48Left bendingHealthy1.55 ± 0.411.63 ± 0.355.16Osteoporotic1.62 ± 0.021.66 ± 0.182.47Right bendingHealthy1.29 ± 0.741.42 ± 0.3610.08Osteoporotic1.35 ± 0.211.45 ± 0.527.41Left rotationHealthy1.29 ± 0.371.34 ± 0.523.88Osteoporotic1.38 ± 0.231.44 ± 0.204.35Right rotationHealthy1.35 ± 0.281.51 ± 0.5211.85Osteoporotic1.52 ± 0.381.62 ± 0.336.58Percent change calculated as (postoperative—preoperative)/preoperative × 100%. Positive values indicate an increase. All changes are increases. The largest increase (25.78% in healthy extension) is highlighted in bold. There is no established biomechanical or clinical threshold for what constitutes a ‘negligible’ versus ‘significant’ disc stress increase after decompression. Therefore, these percentage changes should be interpreted as descriptive only. Disc von Mises stress (MPa) and percent change before and after Endo-ULBD in healthy and reduced‑stiffness (osteoporosis‑simulating) model (mean ± SD). Percent change calculated as (postoperative—preoperative)/preoperative × 100%. Positive values indicate an increase. All changes are increases. The largest increase (25.78% in healthy extension) is highlighted in bold. There is no established biomechanical or clinical threshold for what constitutes a ‘negligible’ versus ‘significant’ disc stress increase after decompression. Therefore, these percentage changes should be interpreted as descriptive only. Equivalent (von Mises) stress distribution within the intervertebral disc under six loading conditions. All images are shown at 600 dpi for clarity. The color scale (blue to red) represents von Mises stress in MPa, with blue indicating low stress (0 MPa) and red indicating high stress (maximum values as indicated on the color bar). Comparison between healthy (left column for each loading condition) and osteoporotic (OP, right column) models after Endo‑ULBD. Stress contours are displayed for flexion (FL), extension (EX), left bending (LB), right bending (RB), left rotation (LR), and right rotation (RR). Note that the stress distribution remains homogeneous in both models, with no focal stress concentrations or abnormal disc bulging. The color bar range is consistent across all panels to facilitate direct visual comparison. Table 5Quantitative parameters of the virtual Endo-ULBD simulation.ParameterValueIpsilateral laminotomy Longitudinal length (cephalocaudal)12 mm Transverse width (mediolateral)8 mm LocationFrom medial border of ipsilateral pedicle to spinous process base Bone thickness removed3–4 mm (full thickness)Ipsilateral medial facetectomy Percentage of facet joint resected30–40% (medial portion) Resected volume128 mm3Contralateral undercutting Undercut depth from midline8 mm Bone thickness removed3 mm (cortical only) Resected volume (spinous process base)92 mm3Ligamentum flavum resection IpsilateralComplete from superior to inferior attachment ContralateralComplete Preserved structuresSupraspinous ligament, interspinous ligament, entire contralateral facet joint, spinous Process Quantitative parameters of the virtual Endo-ULBD simulation. Schematic representation of loading and boundary conditions. A 500 N compressive preload is distributed on the superior surface of L4. A 10 N·m pure moment is applied to the superior node of L4 to simulate flexion, extension, lateral bending, and axial rotation. The inferior surface of L5 is fully constrained (all degrees of freedom fixed). Quantitative endplate stress data are presented in Table 6. In the healthy model, postoperative endplate von Mises stress showed minimal changes across all loading conditions, with percent increases ranging from 1.6% (flexion) to 3.0% (extension), all below 3.1%. In contrast, the reduced‑stiffness (osteoporosis‑simulating) model exhibited substantially larger increases, with the most pronounced elevations observed under axial rotation (18.3% increase, from 1.99 MPa preoperatively to 2.39 MPa postoperatively, averaged across left and right rotation) and extension (15.7% increase, from 2.29 to 2.65 MPa). The average endplate stress increase across all six loading conditions in the reduced‑stiffness (osteoporosis‑simulating) model was 13.5%. These quantitative data confirm the redistribution of load from the disc to the endplates after decompression in osteoporotic bone, a finding that is not observed in the healthy model.Table 6Endplate von Mises stress (MPa) before and after Endo-ULBD in healthy and reduced‑stiffness (osteoporosis‑simulating) models under six loading conditions (mean ± SD).Loading conditionModelPreoperative (MPa)Postoperative (MPa)Absolute change (MPa)Percent change (%)FlexionHealthy1.82 ± 0.111.86 ± 0.100.042.20Osteoporotic1.79 ± 0.141.93 ± 0.120.147.80ExtensionHealthy2.34 ± 0.182.41 ± 0.150.073.00Osteoporotic2.29 ± 0.212.65 ± 0.190.3615.70Left bendingHealthy1.56 ± 0.091.60 ± 0.080.042.60Osteoporotic1.52 ± 0.111.68 ± 0.100.1610.50Right bendingHealthy1.58 ± 0.101.62 ± 0.090.042.50Osteoporotic1.54 ± 0.121.70 ± 0.110.1610.40Left rotationHealthy2.05 ± 0.142.12 ± 0.130.073.40Osteoporotic1.98 ± 0.162.35 ± 0.150.3718.70Right rotationHealthy2.08 ± 0.152.14 ± 0.140.062.90Osteoporotic2.00 ± 0.172.36 ± 0.160.3618.00Values are presented as mean ± standard deviation (n = 6 repeated measurements). Percent change calculated as (postoperative—preoperative)/preoperative × 100%. Positive values indicate an increase. Endplate stress was measured on the superior endplate of L5. The average percent change across all six loading conditions in the reduced‑stiffness (osteoporosis‑simulating) model is 13.5% (range: 7.8%–18.7%). Healthy model changes are all ≤ 3.4% Table 6Endplate von Mises stress (MPa) before and after Endo-ULBD in healthy and reduced‑stiffness (osteoporosis‑simulating) models under six loading conditions (mean ± SD).Loading conditionModelPreoperative (MPa)Postoperative (MPa)Absolute change (MPa)Percent change (%)FlexionHealthy1.82 ± 0.111.86 ± 0.100.042.20Osteoporotic1.79 ± 0.141.93 ± 0.120.147.80ExtensionHealthy2.34 ± 0.182.41 ± 0.150.073.00Osteoporotic2.29 ± 0.212.65 ± 0.190.3615.70Left bendingHealthy1.56 ± 0.091.60 ± 0.080.042.60Osteoporotic1.52 ± 0.111.68 ± 0.100.1610.50Right bendingHealthy1.58 ± 0.101.62 ± 0.090.042.50Osteoporotic1.54 ± 0.121.70 ± 0.110.1610.40Left rotationHealthy2.05 ± 0.142.12 ± 0.130.073.40Osteoporotic1.98 ± 0.162.35 ± 0.150.3718.70Right rotationHealthy2.08 ± 0.152.14 ± 0.140.062.90Osteoporotic2.00 ± 0.172.36 ± 0.160.3618.00Values are presented as mean ± standard deviation (n = 6 repeated measurements). Percent change calculated as (postoperative—preoperative)/preoperative × 100%. Positive values indicate an increase. Endplate stress was measured on the superior endplate of L5. The average percent change across all six loading conditions in the reduced‑stiffness (osteoporosis‑simulating) model is 13.5% (range: 7.8%–18.7%). Healthy model changes are all ≤ 3.4% Endplate von Mises stress (MPa) before and after Endo-ULBD in healthy and reduced‑stiffness (osteoporosis‑simulating) models under six loading conditions (mean ± SD). Values are presented as mean ± standard deviation (n = 6 repeated measurements). Percent change calculated as (postoperative—preoperative)/preoperative × 100%. Positive values indicate an increase. Endplate stress was measured on the superior endplate of L5. The average percent change across all six loading conditions in the reduced‑stiffness (osteoporosis‑simulating) model is 13.5% (range: 7.8%–18.7%). Healthy model changes are all ≤ 3.4% In contrast to disc stress, endplate stress increased significantly and quantitatively in the reduced‑stiffness (osteoporosis‑simulating) model following Endo-ULBD, with the most pronounced elevations observed under axial rotation (18.3% increase vs. preoperative) and extension (15.7% increase vs. preoperative)—loading conditions that are most clinically relevant to daily spinal motion. While endplate stress in the healthy model showed only minor changes (≤ 3.1% variation across all loading conditions), the osteoporotic condition demonstrated marked stress elevation across all motion planes, with average endplate stress increasing by 13.5% relative to preoperative baseline. This indicates a redistribution of axial and rotational load from the intervertebral disc to the vertebral endplates in osteoporotic bone after decompression—a biomechanical shift that is not observed in healthy spinal segments. Facet joint contact force and stress Table 7 summarizes the facet joint contact forces before and after Endo‑ULBD. In the healthy model, facet contact forces increased by 3.3%–5.4% across loading conditions, while in the reduced‑stiffness (osteoporosis‑simulating) model, increases ranged from 4.8–6.7%. Absolute forces were slightly lower in the reduced‑stiffness (osteoporosis‑simulating) model (e.g., extension: 69.4 N preoperatively vs. 73.5 N postoperatively) compared to the healthy model (78.6 N vs. 82.1 N), consistent with reduced bone stiffness. These small increases suggest that the preserved contralateral facet joint partially compensates for the ipsilateral medial facetectomy, without causing abnormally high contact stresses.Table 7Facet joint contact forces (N) before and after Endo-ULBD in healthy and reduced‑stiffness (osteoporosis‑simulating) models under six loading conditions (mean ± SD).Loading conditionModelPreoperative (N)Postoperative (N)Absolute change (N)Percent change (%)FlexionHealthy32.4 ± 2.133.8 ± 2.31.44.30Osteoporotic28.7 ± 2.530.2 ± 2.41.55.20ExtensionHealthy78.6 ± 4.282.1 ± 4.53.54.50Osteoporotic69.4 ± 5.173.5 ± 4.84.15.90Left bendingHealthy45.2 ± 3.346.9 ± 3.11.73.80Osteoporotic39.8 ± 3.841.9 ± 3.52.15.30Right bendingHealthy44.8 ± 3.046.3 ± 3.21.53.30Osteoporotic39.2 ± 3.641.1 ± 3.41.94.80Left rotationHealthy52.3 ± 3.755.1 ± 3.92.85.40Osteoporotic46.1 ± 4.249.2 ± 4.03.16.70Right rotationHealthy51.9 ± 3.554.4 ± 3.82.54.80Osteoporotic45.6 ± 4.048.5 ± 3.92.96.40Values are presented as mean ± standard deviation (n = 6 repeated measurements). Percent change calculated as (postoperative—preoperative)/preoperative × 100%. Positive values indicate an increase. Contact forces represent the sum of ipsilateral and contralateral facet joint forces at the L4–L5 level. Table 7Facet joint contact forces (N) before and after Endo-ULBD in healthy and reduced‑stiffness (osteoporosis‑simulating) models under six loading conditions (mean ± SD).Loading conditionModelPreoperative (N)Postoperative (N)Absolute change (N)Percent change (%)FlexionHealthy32.4 ± 2.133.8 ± 2.31.44.30Osteoporotic28.7 ± 2.530.2 ± 2.41.55.20ExtensionHealthy78.6 ± 4.282.1 ± 4.53.54.50Osteoporotic69.4 ± 5.173.5 ± 4.84.15.90Left bendingHealthy45.2 ± 3.346.9 ± 3.11.73.80Osteoporotic39.8 ± 3.841.9 ± 3.52.15.30Right bendingHealthy44.8 ± 3.046.3 ± 3.21.53.30Osteoporotic39.2 ± 3.641.1 ± 3.41.94.80Left rotationHealthy52.3 ± 3.755.1 ± 3.92.85.40Osteoporotic46.1 ± 4.249.2 ± 4.03.16.70Right rotationHealthy51.9 ± 3.554.4 ± 3.82.54.80Osteoporotic45.6 ± 4.048.5 ± 3.92.96.40Values are presented as mean ± standard deviation (n = 6 repeated measurements). Percent change calculated as (postoperative—preoperative)/preoperative × 100%. Positive values indicate an increase. Contact forces represent the sum of ipsilateral and contralateral facet joint forces at the L4–L5 level. Facet joint contact forces (N) before and after Endo-ULBD in healthy and reduced‑stiffness (osteoporosis‑simulating) models under six loading conditions (mean ± SD). Values are presented as mean ± standard deviation (n = 6 repeated measurements). Percent change calculated as (postoperative—preoperative)/preoperative × 100%. Positive values indicate an increase. Contact forces represent the sum of ipsilateral and contralateral facet joint forces at the L4–L5 level. Intradiscal pressure Intradiscal pressure (IDP) measured at the nucleus pulposus showed minimal changes after Endo‑ULBD (Table 8). In both healthy and reduced‑stiffness (osteoporosis‑simulating) models, IDP increased by 1.5–7.0% depending on loading condition, with the largest increases occurring under axial rotation. All absolute IDP values remained within previously published physiological ranges for lumbar discs (0.36–0.54 MPa). These findings corroborate the disc von Mises stress results and indicate that Endo‑ULBD does not substantially alter intradiscal hydrostatic pressure.Table 8Intradiscal pressure (MPa) before and after Endo-ULBD in healthy and reduced‑stiffness (osteoporosis‑simulating) models under six loading conditions (mean ± SD).Loading conditionModelPreoperative (MPa)Postoperative (MPa)Absolute change (MPa)Percent change (%)FlexionHealthy0.52 ± 0.030.54 ± 0.030.023.80Osteoporotic0.49 ± 0.040.52 ± 0.040.036.10ExtensionHealthy0.38 ± 0.020.39 ± 0.020.012.60Osteoporotic0.36 ± 0.030.38 ± 0.030.025.60Left bendingHealthy0.44 ± 0.030.45 ± 0.030.012.30Osteoporotic0.41 ± 0.030.43 ± 0.030.024.90Right bendingHealthy0.43 ± 0.030.44 ± 0.030.012.30Osteoporotic0.40 ± 0.030.42 ± 0.030.025.00Left rotationHealthy0.47 ± 0.030.49 ± 0.030.024.30Osteoporotic0.44 ± 0.040.47 ± 0.040.036.80Right rotationHealthy0.46 ± 0.030.48 ± 0.030.024.30Osteoporotic0.43 ± 0.040.46 ± 0.040.037.00Values are presented as mean ± standard deviation (n = 6 repeated measurements). Percent change calculated as (postoperative—preoperative)/preoperative × 100%. Positive values indicate an increase. Intradiscal pressure was measured at the center of the nucleus pulposus. All absolute pressures remain within published physiological ranges for lumbar discs under comparable loading. Table 8Intradiscal pressure (MPa) before and after Endo-ULBD in healthy and reduced‑stiffness (osteoporosis‑simulating) models under six loading conditions (mean ± SD).Loading conditionModelPreoperative (MPa)Postoperative (MPa)Absolute change (MPa)Percent change (%)FlexionHealthy0.52 ± 0.030.54 ± 0.030.023.80Osteoporotic0.49 ± 0.040.52 ± 0.040.036.10ExtensionHealthy0.38 ± 0.020.39 ± 0.020.012.60Osteoporotic0.36 ± 0.030.38 ± 0.030.025.60Left bendingHealthy0.44 ± 0.030.45 ± 0.030.012.30Osteoporotic0.41 ± 0.030.43 ± 0.030.024.90Right bendingHealthy0.43 ± 0.030.44 ± 0.030.012.30Osteoporotic0.40 ± 0.030.42 ± 0.030.025.00Left rotationHealthy0.47 ± 0.030.49 ± 0.030.024.30Osteoporotic0.44 ± 0.040.47 ± 0.040.036.80Right rotationHealthy0.46 ± 0.030.48 ± 0.030.024.30Osteoporotic0.43 ± 0.040.46 ± 0.040.037.00Values are presented as mean ± standard deviation (n = 6 repeated measurements). Percent change calculated as (postoperative—preoperative)/preoperative × 100%. Positive values indicate an increase. Intradiscal pressure was measured at the center of the nucleus pulposus. All absolute pressures remain within published physiological ranges for lumbar discs under comparable loading. Intradiscal pressure (MPa) before and after Endo-ULBD in healthy and reduced‑stiffness (osteoporosis‑simulating) models under six loading conditions (mean ± SD). Values are presented as mean ± standard deviation (n = 6 repeated measurements). Percent change calculated as (postoperative—preoperative)/preoperative × 100%. Positive values indicate an increase. Intradiscal pressure was measured at the center of the nucleus pulposus. All absolute pressures remain within published physiological ranges for lumbar discs under comparable loading. Discussion The present study employed a validated finite element model to investigate biomechanical alterations following Endo-ULBD in both healthy and osteoporotic lumbar spines. While our findings confirm the well-accepted core advantage of MIS decompression—minimal increases in segmental ROM and preserved spinal stability—this work moves far beyond mere confirmation by defining a bone quality-specific biomechanical risk profile for Endo-ULBD, translating objective stress changes into clinically meaningful outcomes, and addressing critical methodological limitations (e.g., nonlinear spinal biomechanics, paraspinal musculature) that limit the clinical applicability of prior FE studies. The principal novel findings are as follows: (1) Endo-ULBD results in only a modest increase in segmental ROM percentage increases ranged from 2.80 to 17.10% depending on loading direction, underscoring the inherent stability-preserving nature of the procedure; (2) The reduced‑stiffness (osteoporosis‑simulating) model exhibits a similar postoperative increase in ROM compared to the healthy model; (3) Postoperative disc von Mises stress increases from preoperative values by 2.47% to 25.78%, depending on loading direction and bone stiffness condition. Because no validated threshold exists for what constitutes ‘safe’ versus ‘unsafe’ disc stress increase, we cannot conclude that the procedure does or does not cause disc overloading. The absolute stress values remain below published tissue failure limits, but this does not guarantee clinical safety under cyclic or fatigue loading. These findings should be interpreted as descriptive biomechanical outputs, not as a safety certification; and (4) Osteoporotic spines experience a marked, quantifiable elevation in endplate stress (average 13.5%) after Endo-ULBD, which suggests a potential biomechanical mechanism for increased fracture risk of stress-related spinal complications. These findings advance current understanding by linking bone quality to the biomechanical consequences of MIS decompression and providing hypothesis‑generating biomechanical patterns that require clinical validation before any surgical recommendations can be made—an unaddressed gap in the existing literature. Before discussing the clinical implications of the similar range of motion increase observed in the reduced‑stiffness (osteoporosis‑simulating) model, it is essential to restate that our model does not simulate the degenerative disc disease, facet hypertrophy, or ligamentous thickening that typically coexist with osteoporosis in elderly patients[32, 33]. These degenerative features increase spinal stiffness and could substantially alter the biomechanical response to decompression. Therefore, the finding of a similar postoperative ROM increase in our reduced‑stiffness (osteoporosis‑simulating) model should not be overinterpreted as a direct prediction of clinical outcomes; rather, it highlights the need for future models that incorporate both osteoporotic bone changes and age‑related soft‑tissue degeneration[34]. The quantified resection parameters used in our simulation (laminotomy 12 × 8 mm, medial facetectomy 30‑40%, contralateral undercut 8 mm) are consistent with previously reported endoscopic ULBD techniques [35], supporting the clinical relevance of our model. The modest increase in ROM observed following Endo-ULBD under our idealized resection parameters (30–40% medial facetectomy, preserved contralateral facet and midline ligaments) indicates that this technique may preserve passive segmental stability in the absence of severe pre‑existing degenerative changes. However, in elderly osteoporotic patients with marked facet hypertrophy, ligamentum flavum thickening, and severe lateral recess stenosis, adequate decompression may require more extensive facet resection, which could lead to greater postoperative ROM increases than those reported here. Therefore, our findings likely represent a best‑case estimate rather than a universal prediction. However, this finding should not be interpreted as a guarantee of clinical stability, which also depends on muscle function, healing response, and patient activity levels. The surgical principle of ULBD is to achieve adequate neural decompression while preserving midline osteoligamentous structures, such as the supraspinous and interspinous ligaments, along with a substantial portion of the contralateral facet joint. Our results align with clinical studies reporting preserved outcomes and low rates of postoperative instability after microscopic or endoscopic ULBD[36]. The ROM increase observed in our models validates the biomechanical rationale for this technique, highlighting its capacity to preserve spinal stability even in the challenging context of osteoporosis[37]. Before interpreting the results, a key methodological limitation must be restated: our reduced‑stiffness (osteoporosis‑simulating) model is derived from a young, healthy spine (32‑year‑old male) with only the elastic modulus of bone reduced. It does not simulate the trabecular loss, cortical thinning, or heterogeneous density distribution that characterize real osteoporotic bone. Consequently, all findings related to the ‘osteoporotic’ condition in this study specifically reflect the effects of reduced bone stiffness, not the full pathological entity of osteoporosis. Clinical extrapolation should therefore be made with caution. We must first acknowledge a critical limitation: our model does not include any age‑related degenerative changes such as disc degeneration, facet joint enlargement (hypertrophy), or ligament thickening. These features are common in elderly osteoporotic patients and typically increase segmental stiffness, thereby reducing range of motion. In our model, however, the similar postoperative ROM increase in the osteoporotic condition cannot be explained by these absent degenerative changes. Instead, the observed reduction in ROM increase may reflect an artifact of the simplified modeling approach (e.g., uniform reduction of bone modulus without accounting for concurrent soft‑tissue stiffening) or may indicate that bone quality alone has a complex, non‑linear effect on segmental kinematics that requires further investigation [38]. Therefore, the finding that osteoporotic spines show a similar ROM increase after Endo‑ULBD should not be interpreted as evidence that osteoporosis confers protection against iatrogenic instability. As expected from reduced bone stiffness, the reduced‑stiffness (osteoporosis‑simulating) model exhibited slightly higher baseline and postoperative ROM compared to the healthy model. The percentage increase after Endo‑ULBD was similar between the two models, indicating that bone quality does not substantially modulate the destabilizing effect of the procedure. In real clinical practice, the net effect depends on the balance between osteoporotic bone softening (which may increase motion) and degenerative stiffening (which may decrease motion). Future models that incorporate both bone quality deterioration and age‑related soft‑tissue changes are needed to clarify this interaction[39]. We initially considered whether the similar ROM increase in the reduced‑stiffness (osteoporosis‑simulating) model could be explained by concurrent degenerative stiffening, but this interpretation would be logically flawed. Our model was derived from a young healthy spine with no degenerative changes; osteoporosis was simulated only by reducing bone stiffness. Therefore, the observed difference in ROM increase between healthy and reduced‑stiffness (osteoporosis‑simulating) models cannot be attributed to ‘pre‑existing degenerative stiffness’ because no such degeneration exists in the model. Instead, this finding may reflect a non‑linear or counterintuitive biomechanical effect of uniformly reducing bone modulus on segmental kinematics under the applied loading conditions. Alternatively, it could be an artifact of our linear‑elastic material assumptions or the absence of musculature. Importantly, this result does not imply that osteoporotic patients in clinical practice have lower postoperative ROM increases; real osteoporotic patients often have concurrent degeneration that could independently affect ROM. Thus, this finding should be considered model‑specific and hypothesis‑generating, not a clinical prediction [40, 41]. After correcting the error in our original interpretation (the disc stress actually increased slightly, not decreased), we re‑evaluate this finding. Our results show postoperative increases in disc von Mises stress ranging from 3.88 to 25.78% in the healthy model and 2.47 to 10.48% in the reduced‑stiffness model, with the largest increase (25.78%) occurring in the healthy model under extension. A critical question is whether such increases are biomechanically meaningful. To our knowledge, no published study has established a definitive threshold for ‘clinically significant’ disc stress elevation following spinal decompression. In vitro studies have reported ultimate tensile strength of the annulus fibrosus in the range of approximately 4–6 MPa, and compressive failure of the disc–vertebra complex typically occurs at stresses exceeding 5–10 MPa, depending on loading mode and degeneration status. Compared with these failure limits, the absolute disc stress values in our study (maximum 2.83 MPa postoperatively in the healthy model under extension) are well below reported failure thresholds. However, this comparison is limited by differences between von Mises stress (a yield criterion for ductile materials) and actual tissue failure modes, as well as by our linear‑elastic material assumptions, which do not capture nonlinear, viscoelastic, or fatigue behavior. Therefore, while the absolute disc stresses we computed do not approach reported failure limits, we cannot conclude that the observed increases are ‘negligible’ or that Endo‑ULBD ‘does not cause disc overloading’ in the sense of long‑term degenerative risk. Cyclic loading, fatigue, and patient‑specific factors (e.g., pre‑existing disc degeneration) could alter the biomechanical response. A more accurate statement is that within the limitations of our static, linear‑elastic model, the calculated disc stresses remain below published failure thresholds, but the absence of validated thresholds for ‘safe’ versus ‘unsafe’ stress increases precludes definitive clinical interpretation. Future studies employing hyperelastic and viscoelastic disc models under cyclic loading are needed to assess fatigue‑related disc overload risk. This corrected finding aligns with the stability‑preserving nature of Endo-ULBD and does not alter the main conclusions of the study[42]. We postulate that by relieving central and lateral recess stenosis, the procedure reduces pathological pressure within the spinal canal, potentially facilitating a more physiological distribution of loads across the three-joint complex (the disc and the two facet joints). The small changes in disc stress suggest that the procedure does not adversely affect disc biomechanics, but its clinical relevance for pain or degeneration remains uncertain [43, 44]. However, the concurrently significant increase in endplate stress within the osteoporotic group unveils a critical biomechanical risk. Osteoporotic vertebral bone, characterized by a reduced elastic modulus and inferior load-bearing capacity, is highly susceptible to stress concentration. While disc loading remains within a modest range, a greater proportion of the axial load may be transferred directly to the vertebral endplates [45]. This elevated endplate stress is a primary biomechanical precursor to complications in the context of reduced bone stiffness. However, because our model lacks microarchitectural deterioration (trabecular loss, cortical thinning), the calculated stress changes should be interpreted as indicating a potential hazard rather than directly predicting clinical complication rates. In real osteoporotic bone, the actual risk of endplate fracture or vertebral collapse may be even higher than our model suggests due to additional structural weaknesses, but this remains to be tested [46]. This finding underscores a vital caveat: although Endo-ULBD appears mechanically safe in terms of segmental stability [47, 48] , the inherent compromise in bone quality associated with osteoporosis presents a distinct risk factor for stress-related complications due to altered load transfer [49]. In contrast to disc stress, endplate stress increased significantly and quantitatively in the reduced‑stiffness (osteoporosis‑simulating) model following Endo-ULBD. As shown in Table 6, the most pronounced elevations were observed under axial rotation (18.3% increase, from 1.99 to 2.39 MPa) and extension (15.7% increase, from 2.29 to 2.65 MPa). A critical caveat regarding the interpretation of our stability results is the absence of paraspinal musculature in the model. In vivo, the lumbar spine is stabilized not only by passive osteoligamentous structures but also by active muscle forces. Elderly osteoporotic patients often have reduced muscle mass and strength (sarcopenia), which compromises their ability to dynamically stabilize the spine. Our model, by excluding muscles, assumes an ideal scenario where passive structures alone determine segmental kinematics. This may lead to an overestimation of the stability‑preserving effect of Endo‑ULBD, because in a sarcopenic patient, the same bone resection could result in greater instability due to inadequate muscular compensation. Therefore, our conclusion that Endo‑ULBD causes only minimal increases in ROM should be viewed as a best‑case estimate, and clinical application should consider the patient’s muscle function status. Our results provide robust biomechanical evidence supporting the use of Endo-ULBD in patients with lumbar spinal stenosis, including those with osteoporosis. The technique’s minimal impact on segmental stability makes it a viable option for patients who may be poor candidates for more destabilizing procedures or instrumented fusion. However, the identified increase in endplate stress in osteoporosis demands heightened surgical prudence. Surgeons should exercise meticulous care to preserve the structural integrity of the facet joints and posterior elements during decompression, as these structures are crucial for sharing axial load and protecting the vulnerable endplates. In cases of severe osteoporosis, a more conservative bone resection might be warranted. Furthermore, if concomitant interbody fusion is planned, extreme caution is advised during endplate preparation and graft insertion to mitigate the risk of subsidence. If these findings were to be validated in clinical studies, a hypothetical biomechanical goal could be to achieve adequate decompression while preserving load‑bearing structures, but this remains to be tested. A further limitation that deserves emphasis is our single‑segment focus. The L4–L5 level was selected because it is the most common site of degenerative lumbar spinal stenosis requiring surgical decompression. However, we did not evaluate the biomechanical consequences at adjacent segments (L3–L4 and L5–S1). Even minor increases in ROM at the operated level can theoretically alter load distribution to adjacent discs and facet joints, potentially accelerating adjacent segment degeneration. In osteoporotic patients, whose vertebrae have reduced fracture resistance, such compensatory changes might be particularly deleterious. Because our model did not include these adjacent levels, we cannot comment on whether Endo‑ULBD increases the risk of ASD. Clinical studies have reported low rates of adjacent segment problems after endoscopic decompression, but long‑term follow‑up in osteoporotic cohorts is lacking. Our computational findings therefore do not provide any evidence for or against ASD risk, and this question remains open for future investigation. A major challenge in interpreting our disc stress results is the absence of published biomechanical or clinical thresholds that define what magnitude of postoperative stress increase is ‘negligible,’ ‘modest,’ or ‘clinically significant.’ Unlike implant loosening (defined by micromotion thresholds) or endplate fracture (associated with specific stress concentrations relative to bone strength), the intervertebral disc’s tolerance to increased von Mises stress after decompression has not been quantified. Factors such as loading rate, cycle count, disc degeneration grade, and individual patient activity levels all modulate the clinical relevance of a given stress increase. Therefore, our reporting of percentage changes (e.g., 25.78% increase) is purely descriptive. We strongly caution readers against interpreting these values as evidence of safety or risk without external validation. Limitations and future directions This study has several inherent limitations of finite element analysis. First, the model simplifies the complex, nonlinear, and time-dependent material properties of biological tissues. Second, the simulations did not account for the dynamic stabilizing effect of surrounding musculature, which could influence segmental ROM and intradiscal pressure in vivo. This omission is particularly relevant for our study population because elderly patients with osteoporosis frequently suffer from sarcopenia—age‑related loss of muscle mass and strength. Paraspinal muscles (erector spinae, multifidus, etc.) provide active stabilization, generate compressive preloads, and control intervertebral motion during daily activities. Without these muscle forces, our model likely underestimates the total compressive loads on the vertebral bodies and endplates, while also overestimating segmental ROM because the stabilizing moments of muscles are absent. Consequently, the biomechanical stability observed in our model (i.e., minimal ROM increase after Endo‑ULBD) may be more favorable than what would occur in a real osteoporotic patient with impaired muscle function. Future studies should incorporate subject‑specific muscle forces, for example using electromyography‑driven musculoskeletal models coupled with finite element analysis, to better represent the in vivo condition. Third, the osteoporotic condition was simulated solely by uniformly reducing the bone elastic modulus, based on a young, healthy spine. This approach does not incorporate the heterogeneous deterioration of trabecular bone microstructure (e.g., trabecular thinning, trabecular perforation, loss of connectivity) or cortical porosity, nor does it account for regional variations in bone density. As a result, our model may be associated with reduced bone stiffness rather than a true osteoporotic condition, and it cannot capture local stress concentrations that arise from microarchitectural defects. This simplification may lead to an underestimation of the actual fracture risk in osteoporotic bone. All conclusions regarding clinical risk in osteoporosis must be interpreted within this constraint. Fourth, while the healthy model was successfully validated against published experimental data, the reduced‑stiffness (osteoporosis‑simulating) model lacked independent validation. Our approach to simulate osteoporosis by uniformly reducing the elastic modulus of bone is based on established relationships between bone mineral density and mechanical properties, but we could not directly compare our reduced‑stiffness (osteoporosis‑simulating) model’s biomechanical outputs (e.g., ROM, endplate stress) against reference data because such data for osteoporotic lumbar segments under the same loading conditions are not available in the literature. This absence of pathological validation means that we cannot be certain whether the observed differences between healthy and reduced‑stiffness (osteoporosis‑simulating) models are solely attributable to bone quality changes or are partly influenced by unvalidated modeling assumptions. Consequently, our findings regarding the osteoporotic group should be considered exploratory and hypothesis‑generating. Future studies should seek to validate osteoporotic finite element models using cadaveric experiments or clinical imaging‑based inverse analyses. Fifth, our analysis was restricted to a single functional spinal unit (L4–L5). While this allowed for detailed evaluation of local biomechanical changes at the decompressed level, it entirely omitted the assessment of adjacent segment effects. In clinical practice, any alteration in segmental kinematics—even a small increase in ROM—can potentially lead to adjacent segment degeneration (ASD) over time. This risk may be amplified in osteoporotic patients due to reduced bone quality and altered load transmission. Our single‑segment model cannot determine whether the observed postoperative ROM increases at L4–L5 cause compensatory hypermobility, increased intradiscal pressure, or facet joint loading at the adjacent levels (L3–L4 and L5–S1). Therefore, the findings of this study should not be interpreted as evidence that Endo‑ULBD is safe with respect to adjacent segment degeneration. Future studies should employ multi‑segment models (e.g., L3–S1) to specifically investigate adjacent segment biomechanics after Endo‑ULBD, particularly in osteoporotic spines. While we provided quantitative resection parameters, actual surgical decompression extent may vary based on patient anatomy and surgeon preference. Our model represents a single standardized resection pattern; sensitivity analyses varying the resection size were not performed but would be valuable in future studies. Furthermore, while Fig. 1B shows a clinical example of an elderly osteoporotic patient, our finite element model was not derived from this patient. The model was constructed from a young healthy volunteer, and the osteoporotic condition was simulated by parametric reduction of bone stiffness. Therefore, patient‑specific features such as individual variations in anatomy or degenerative changes were not represented. Sixth, the use of linear‑elastic material models for soft tissues (intervertebral disc, ligaments, cartilage) represents a significant biomechanical simplification. In reality, the annulus fibrosus exhibits anisotropic, hyperelastic, and viscoelastic behavior; the nucleus pulposus behaves as a poroelastic gel; and ligaments show non‑linear stress‑strain relationships with toe regions. Our linear‑elastic models cannot capture phenomena such as stress relaxation, creep, hysteresis, or rate‑dependent stiffness. Consequently, our results for disc stress and segmental ROM reflect instantaneous, small‑strain responses to static loads, not the time‑dependent behavior that occurs during daily activities (e.g., walking, lifting, prolonged sitting). This limitation is particularly relevant for assessing long‑term risks such as disc degeneration or endplate fracture, which may be influenced by cumulative creep and cyclic loading effects. Future studies should employ hyperelastic (e.g., Ogden or Neo‑Hookean) and viscoelastic (e.g., Prony series) material models to more accurately represent the time‑dependent response of spinal tissues after Endo‑ULBD. Seventh, our decompression model assumes an idealized surgical scenario. The resection parameters (12 mm × 8 mm laminotomy, 30–40% medial facetectomy, 8 mm contralateral undercut) were derived from standard endoscopic ULBD techniques applied to a non‑degenerated spine model. In real clinical practice, elderly osteoporotic patients with degenerative lumbar spinal stenosis frequently exhibit severe facet hypertrophy, thickened ligamentum flavum, disc space collapse, and osteophyte formation. In such cases, achieving bilateral decompression may require larger bone resection, including more extensive facetectomy (sometimes exceeding 50%) and more aggressive undercutting, despite the surgeon’s intention to preserve stabilizing structures. Consequently, the actual postoperative range of motion increase and endplate stress elevation in these patients could be substantially higher than our model predicts. Our results should therefore be viewed as hypothesis‑generating and applicable primarily to less degenerated or early‑stage stenosis, not as a direct simulation of complex, severely degenerated osteoporotic spines. Future studies should create models based on CT scans of elderly osteoporotic patients with advanced degenerative changes and simulate the extent of decompression actually required in those cases. Future research should focus on developing more sophisticated, patient-specific models that incorporate heterogeneous bone density distributions derived from quantitative CT and integrate the effects of muscle forces. Longitudinal analyses simulating bone remodeling after surgery could provide insights into long-term adaptations. Most importantly, these computational findings need to be correlated with prospective clinical studies that focus on patient-reported outcomes, radiographic signs of instability, and the incidence of endplate-related complications in osteoporotic populations undergoing Endo-ULBD. From a clinical perspective, the 13.5% average increase in endplate stress observed in the reduced‑stiffness (osteoporosis‑simulating) model—particularly under axial rotation (18.3%) and extension (15.7%)—may have direct implications for postoperative outcomes. These loading conditions are commonly encountered during daily activities such as twisting while standing or rising from a chair. In patients with compromised bone quality, such stress elevations could theoretically increase the risk of endplate fracture, interbody cage subsidence if fusion is performed, or accelerated disc degeneration due to altered load transmission. However, because our model does not simulate trabecular microarchitecture, these implications remain hypothesis-generating and require clinical validation. Conclusion In conclusion, this finite element analysis suggests the well‑accepted stability‑preserving nature of Endo‑ULBD in a healthy bone model and in a model with reduced bone stiffness (simulating certain aspects of osteoporosis), with segmental ROM percentage increases ranging from 2.8 to 17.1% in the healthy model and 3.0 to 17.1% in the reduced‑stiffness (osteoporosis‑simulating) model (depending on loading direction). However, it is important to note that our reduced‑stiffness (osteoporosis‑simulating) model represented only reduced bone stiffness without microarchitectural deterioration, and the model lacked independent validation. Therefore, the following conclusions should be interpreted as hypothesis‑generating. The study suggests preliminary, hypothesis‑generating biomechanical patterns related to bone stiffness reduction for Endo‑ULBD: in our stiffness‑reduced model, osteoporotic spines showed a marked elevation in endplate stress (average 13.5%) after decompression. This finding indicates a potential biomechanical mechanism that could increase the risk of endplate fracture, vertebral body collapse, cage subsidence, and accelerated disc degeneration in osteoporotic patients. However, because our model did not simulate trabecular loss or cortical porosity, the absolute magnitude of clinical risk cannot be directly inferred from these computational results. The similar ROM increase observed in the reduced‑stiffness (osteoporosis‑simulating) model (compared to the healthy model) is a model‑specific finding and should not be interpreted as clinical evidence that osteoporotic patients have a lower risk of iatrogenic instability. Real osteoporotic patients often have concurrent degenerative changes that were not simulated here. Given the limitations of linear FE modeling (absence of viscoelasticity, musculature, microarchitectural heterogeneity, and adjacent segments), our results may provide conservative estimates of endplate stress changes, but the actual in vivo response could differ. These computational results are not validated for clinical prediction and must not be used directly to guide patient care. Collectively, these findings suggest that Endo‑ULBD may be associated with preserved segmental stability in reduced‑stiffness bone, but the potential for elevated endplate stress warrants further investigation. Future prospective studies should examine whether preoperative bone quality assessment, surgical technique modifications (such as minimizing facet resection), or postoperative load management could mitigate endplate‑related complications. The present computational results alone do not support specific surgical recommendations.