1 Introduction
The human spine is a marvel of natural engineering, and at its core lies the intervertebral disc (IVD)—a sophisticated structure essential for movement, load-bearing, and shock absorption. This critical component functions as a complex, viscoelastic composite: a soft, hydrated nucleus pulposus (NP) resists compression and distributes pressure, while a tough, collagen-rich annulus fibrosus (AF) provides circumferential strength.
[1,
2] This synergy creates a system that manages immense mechanical stresses through its fluid dynamics and fibrous architecture.
[3,
4]Unfortunately, disc degeneration disrupts this delicate balance, diminishing water content and altering mechanical properties.
[4] This breakdown leads to abnormal stress distributions, spinal instability, and pain, underscoring the critical need for effective restoration strategies.
[5] While artificial intervertebral discs (AIDs) have been developed to preserve motion, most prevalent designs—such as ball-and-socket or metal–polymer sandwiches—prioritize kinematics over biomechanical fidelity. They often fail to replicate the natural disc’s vital viscoelasticity and shock-absorbing functions. Although newer biomimetic designs (e.g., M6, fiber-reinforced hydrogels) show promise in mimicking annulus-nucleus structure and swelling behavior,
[6,
7] significant gaps remain. Few studies have thoroughly investigated how these viscoelastic AIDs influence stress distribution within the implant and surrounding tissues.
[8] Moreover, manufacturing limitations and a lack of sophisticated computer modeling have hindered detailed mechanical analysis.
[9,
10]To address these challenges, this study introduces a novel biomimetic AID composed of a 3-dimensional (3D) lamellar scaffold integrated with an osmotic hydrogel matrix, designed to closely emulate the structure and properties of the natural IVD. Our research aimed to experimentally evaluate the mechanical performance of this prototype under axial compression, shear, and torsion loading and utilize finite element analysis (FEA) to visualize internal stress distributions and validate its biomechanical behavior. By bridging the gap between biomimetic design and rigorous mechanical evaluation, this work offers a meaningful step toward the development of AIDs that not only restore motion but also truly recapitulate the spine’s natural biomechanics.
2 Materials and methods
2.1 Fabrication of the biomimetic AID
A biomimetic AID was designed based on the dimensions and multilamellar structure of the human C6–C7 disc. A 3D lamellar scaffold, representing the AF, was fabricated using a fused deposition modeling machine and medical-grade thermoplastic polyurethane (TPU; Tecoflex®LM-63D; Lubrizol, Cleveland, Ohio) with a tensile modulus of 23–33 MPa. The NP was simulated using a polyacrylamide hydrogel. A pregel solution was prepared by mixing 10 wt% acrylamide (A8887; Sigma, Saint Louis), 10 wt% N,N-methylenebisacrylamide (146072; Sigma) as a crosslinker, and 0.1 wt% ammonium persulphate (248614; Sigma) as a photoinitiator. The solution was degassed in a vacuum chamber, after which 40 wt% nano-Al2O3 (702129-100G; Sigma) was added as a crosslinking accelerator. The flexible TPU scaffold was then placed in a mold, and the pregel solution was poured and infused into it. The assembled AID was ultraviolet-cured for 30 min and stored in a humid box for 24 h to allow the reaction to stabilize. The final AID had nominal lateral and sagittal diameters of 25 and 18 mm, respectively, and a nominal height of 7 mm under preload (Fig. 1A).
2.2 Mechanical testing
Fifteen AID samples were fabricated for mechanical characterization. Creep tests were conducted on a material testing system (ElectroUniversal Testing Machine, AGS-X-300kN; SHIMADZU, Tokyo, Japan) in a 0.15 M NaCl bath maintained at 37 ℃.
2.2.1 Axial compression, compression shear, and axial torsion (Fig. 2A–C)
Samples were mounted between custom stainless steel fixtures (conforming to standard F2346-05) to simulate a spinal motion segment. Axial compression tests were performed at a strain rate of 5 mm·min-1 up to a load of 2kN. Failure was defined as a force drop of > 5% at constant displacement or a thickness reduction of > 50% compared with the postpreconditioning thickness. Strength was defined as the peak load before failure. Compression-shear testing was initiated with the device’s Z-axis rotated +45° about the Y-axis. For torsion tests, a combined compressive force (F) and moment (M) were applied, with bending moment minimized.
2.2.2 Dynamic fatigue test
In the fatigue tests, the AIDs were axially loaded (150–1500N) at 5 Hz. A diurnal cycle, consisting of 16-h loading and 8-h recovery at 50N. “Failure” was defined as 1 or more of the following
[10]: a decrease (> 20%) in thickness compared with the beginning of the loading period; a thickness decrease (> 5%) at the start of 1 loading period relative to the previous one; a thickness decrease (> 50%) compared with the thickness after preconditioning; or visible extrusion of gel particles.
2.2.3 Compressive creep
Before and after fatigue tests, a compressive creep test was carried out. An axial load of 150N was applied for 7200 s (2h) in a saline bath at 37 ± 0.5 ℃, and strain was monitored. The constant load of 150N was applied using a linear ramp loading regime for 5 s as described previously.
[11] The last creep curve was compared with the range of natural IVD responses, as described by the creep model used by Skrzypiec et al.
[12]2.3 Finite element analysis
2.3.1 Model development
Finite element models of the AID and its clamping apparatus were reconstructed using 3D reverse engineering technology in Geomagic Studio 2014 software to simulate an in vivo motion segment. The model, consisting of the biomimetic disc and fixtures, was meshed with 273,436 nodes and 263,500 IsoMesh Hex8 elements. A binding contact was defined between structural grids to ensure uninterrupted stress transmission. The meshed model was processed using Hypermesh 13.0 software (Fig. 1).
Mesh sensitivity analysis was performed by refining the mesh from approximately 120,000 to 400,000 elements. The results for peak stress in the annulus and nucleus changed by less than 3% between the medium (about 260,000 elements) and fine meshes, confirming that the chosen mesh density provided converged and reliable results.
2.3.2 Material properties
The bone tissue of the vertebrae, endplates, and the interspinous cartilage was described by isotropic and linear-elastic material properties.
[13–
15] Material properties were assigned based on literature values and measured data (Table 1). Vertebral bone, endplates, and cartilage were modeled as isotropic, linear-elastic materials. The AID’s AF was modeled as a hydrogel matrix reinforced with 3 concentric TPU layers, assigned an isotropic, linear-elastic modulus of 20 MPa in the circumferential direction.
[14,
16,
17] The NP was modeled as an isotropic, viscoelastic hydrogel with a compressive modulus of 0.5 MPa. Articulating surfaces used surface-to-surface contact elements (penalty algorithm, friction coefficient = 0.1), mirroring natural disc kinematics.
[18,
19] The penalty stiffness was tuned to minimize penetration while maintaining numerical stability. The stainless steel (SS304) clamping apparatus was assigned an elastic modulus of 195 GPa and a Poisson’s ratio of 0.247.
2.3.3 Boundary and loading conditions
Simulations replicated the 3 experimental mechanical tests: axial compression, compression shear, and axial torsion. A compression preload of 150N was applied in all stages.
[20,
21] Subsequently, for compression shear, a 150N shear force was applied; for torsion, a 6° rotational motion was applied. In all simulations, the load was applied to the superior fixture, while all 6° of freedom were constrained for the nodes on the inferior fixture surface.
2.4 Statistical analysis
For all material parameters, statistical differences were determined with a one-way analysis of variance. For creep duration analysis, the Bonferroni pairwise comparison was applied.
3 Results
3.1 Mechanical properties
The AID samples did not fail under compressive loads up to 2kN (corresponding to 50%–60% strain). The stress–strain curve exhibited a J-shaped profile, with a toe region extending to about 10% strain, followed by a linear region. The compressive modulus, derived from the linear region, was 10.0MPa at a strain rate of 5 mm·min
-1 (Fig. 3A). The compression–shear stiffness was 90N·mm
-1 (Fig. 4A), and the torsional stiffness was 0.23 N·m·degree
-1 (Fig. 5A). To evaluate the viscoelasticity, the time-dependent creep tests were conducted. Under a constant axial load at 150N, the prosthesis exhibited viscoelastic strain values ranging from 6.1% to 14.6% over 2 h, demonstrating a trend of progressively decelerating deformation. The creep strain of the postfatigue artificial IVDs is lower (the difference is not statistically significant), ranging from 4.9% to 11.7%. Both prefatigue and postfatigue creep strain values fall within the natural disc range (Fig. 1B). Such a curve aligned well with the response range reported for natural cervical IVDs under similar loading (Fig. 1A).
[12]3.2 Finite element analysis
Finite element analysis was employed to quantify internal stress distributions and biomechanical behavior under load, specifically measuring layer height (H) changes and bulging in the AF.
3.3 Axial compression (Fig. 3)
The superior surface of the AID sustained an average stress of 1.64 MPa (peak: 2.31 MPa), while the inferior surface bore 0.38 MPa (peak: 0.84 MPa) (Fig. 3E and F). The reinforcing fiber layer sustained a radial stress of 1.67 MPa, and the hydrogel NP core bore a stress of 0.78 MPa. The artificial disc exhibited a radially decreasing stress profile, which was primarily localized within the inner scaffold. The lamellar scaffold effectively restricted the hydrogel matrix from horizontal buckling.
3.4 Compression shear (Fig. 4)
Under shear loading, the superior surface sustained an average stress of 0.22MPa (peak: 0.38MPa), and the inferior surface bore 0.13MPa (peak: 0.26MPa) (Fig. 4E and F). The fiber layer and NP bore radial stresses of 0.44 and 0.39MPa, respectively. The fixture connection minimized stress concentration.
3.5 Axial torsion (Fig. 5)
In the axial torsion test, the superior surface sustained an average stress of 0.35 MPa (peak: 0.84 MPa), and the inferior surface bore 0.32 MPa (peak: 0.77 MPa) (Fig. 5E and F). The fiber layer and NP bore the highest stresses at 1.90 and 0.93 MPa, respectively. The fixture was displaced by 1.52 mm, but prevented excessive buckling.
3.6 Stress profile of the macroscale disc model
Figure 6A–C correlates the stress contour plot with the regional stress distribution in the AID under different loading conditions. A ripple-shaped contour can be observed for all normal stress components. The magnitude of transverse normal components increases consistently when approaching to the NP, which is considered to be caused by the radial constraint from the lamellar AF. The stiffness of NP hydrogel is much lower than that of AF. When the axial compression load is applied, the composite ring is barely stretched, thus causing less constraint onto the hydrogel core. While under the compression shear or axial torsion load, the lamellar scaffold is stretched. In this case, the stress magnitude concentrated at the annulus fibrosis, indicating the critical constraint at the radial direction, as illustrated in Figure 6D. A ring of stress concentration along the interfacial region can also be seen, and it appears more conspicuous as the overall strain increases.
4 Discussion
The development of AIDs has historically progressed slowly, one of the challenges primarily attributed to the complex physiological and biomechanical nature of the native disc.
[22] Clinical adoption has been limited by issues such as particulate wear debris and a fundamental mismatch in mechanical properties between existing prostheses and the natural IVD, often leading to adjacent segment degeneration. In response to this need, our study presents a novel biomimetic AID prototype that integrates a 3D-printed TPU lamellar scaffold—mimicking the AF—with a viscoelastic polyacrylamide hydrogel nucleus. This design philosophy is grounded in the understanding that the natural IVD is a fiber-reinforced, lamellar, and viscoelastic composite structure. Our results demonstrate that this composite model successfully replicates the key biomechanical and viscoelastic properties of the human cervical disc.
A significant hurdle in AID design is that synthetic hydrogels, while biocompatible and hydrous, are often mechanically compliant and fragile compared with the tough, resilient native tissue.
[23,
24] The prevailing principle for creating tough hydrogel systems involves incorporating mechanisms for efficient energy dissipation while retaining high elasticity.
[25] Our design addresses this by utilizing the 3D-printed TPU scaffold to provide high tensile strength and structural integrity, while the hydrogel matrix contributes energy dissipation through its viscoelastic behavior. This synergistic interaction was evident in the characteristic J-shaped stress–strain curve under axial compression, a hallmark of natural soft tissues. The initial toe region represents the uncrimping and alignment of collagen fibers in the natural annulus, which in our model is replicated by the gradual engagement of the reinforcing scaffold fibers. The subsequent linear region reflects the dominant elastic response of these straightened fibers. Furthermore, the prosthesis exhibited compression shear and axial torsion stiffness values (90N·mm
-1 and 0.23 N·m·degree
-1, respectively) that satisfy the basic mechanical requisites for cervical disc replacement. Crucially, the time-dependent creep behavior of our AID closely approximated that of natural cervical IVDs, underscoring its capacity for critical shock-absorbing functions. It is noteworthy that the artificial IVD remained intact and free from delamination after periodic 5 million fatigue loading. Although the AIDs did become stiffer in the loading process, with a decrease in creep strain (Fig. 1B). Nevertheless, the mechanical property values remained within the reference ranges of its natural counterpart. The excellent durability of the AID can be attributed to the tough and self-healable hydrogel, which automatically repairs in response to damage.
Our work advances the field beyond previous biomimetic efforts. For instance, van den Broek et al.
[10] developed a promising prototype with a hydrogel core and fiber encapsulation. However, its manual fabrication introduced structural inconsistencies that precluded reliable FEA and precise customization. Gullbrand et al.
[1] developed a layered scaffold consisting of 3 different components to mimic natural spinal motion segments. In our previous work, we applied linear motors to weave glass fibers into a scaffold structure and then injected a self-healing, tough nanocomposite hydrogel.
[26] This scaffold, made of 3D braided fibrous textiles and a hydrogel, mimicked the 3D tetragonal fibrous structures of natural AF and natural NP, respectively. FEA of the artificial IVD from mesoscale and macroscale analysis indicates the coherent load transfer through both the interconnections within the fiber mesh and the fiber–matrix interface. In contrast to the complex structures, we developed a simplified lamellar scaffold–matrix compsite to elucidate and verify the underlying mechanical significance of native annulus-nucleus structure. In addition, our use of computer-aided design and 3D printing ensures high reproducibility, allows for patient-specific customization of structural parameters (e.g., lamellar spacing and fiber orientation), and facilitates direct integration with computational modeling. This represents a significant methodological improvement for the iterative design and analysis of AIDs.
The FEA results provide critical insight into the internal mechanical environment of our AID, a feature difficult to assess experimentally. The model revealed a stress distribution pattern consistent with findings in natural IVDs, where the highest compressive stresses are concentrated in the hydrostatic NP, and stress within the annulus decreases radially outwards.
[19,
27] This validates the biomimetic performance of our design. However, the analysis also indicated that the average and peak stress values under load were higher than those reported in some cadaveric studies. This discrepancy may be attributed to 2 factors: the application of a 150N load, which represents an upper-limit scenario (e.g., head-helmet system) rather than pure physiological loading; and the absence of posterior spinal elements (facet joints, ligaments) in our isolated segment model. In an intact spine, these posterior structures share a significant portion (up to 64%) of the axial load, thereby reducing the stress borne by the disc itself.
5 Limitations
Despite promising results, this study has several limitations. First, the in vitro model lacked surrounding musculoskeletal structures, which are essential for replicating the full in vivo loading environment. Future work should incorporate these elements into more complex FEA models and utilize spinal motion simulators for experimental validation. Then, the current analysis focused on short-term mechanical performance. Long-term fatigue testing, wear analysis of the hydrogel–scaffold interface, and comprehensive biocompatibility assessments are essential next steps to evaluate the implant’s durability and safety. In addition, the fully fixed boundary condition of the lower fixture, while matching the experimental setup, may introduce higher interfacial stresses compared with the compliant support of vertebral bone in vivo. Finally, while the material properties were based on literature values, further characterization of the hydrogel’s permeability and the scaffold’s anisotropic properties could refine the FEA models. Future studies should also incorporate multisegment spinal models with realistic vertebral mobility and posterior elements (facet joints and ligaments) to evaluate coupled motions and load-sharing across adjacent levels, providing a more comprehensive simulation of the in vivo biomechanical environment.
6 Conclusion
In conclusion, to our knowledge, this study is the first to integrally combine experimental biomechanical testing with detailed FEA of the internal mechanical state in a fully biomimetic, 3D-printed AID. Our prototype, characterized by its annulus-nucleus structure, successfully emulates the viscoelastic, shock-absorbing, and load-bearing capabilities of the natural cervical disc. This work not only provides a feasible design paradigm for a new generation of AIDs but also establishes a robust framework for their computational evaluation and iterative optimization, paving the way for more successful clinical applications in the future.
© The Author(s) 2026. Published by Wolters Kluwer Health, Inc. on behalf of Higher Education Press.
PDF
(411KB)
