Animal models in spine research: applications, advantages, limitations, and future perspectives

Needle puncture-induced degeneration models

The needle puncture-induced degeneration model is one of the most widely used and standardized approaches for experimental IVDD^[15,16]. By mechanically disrupting the annulus fibrosus, needle puncture alters intradiscal pressure and tissue integrity, leading to nucleus pulposus injury, ECM breakdown, and progressive structural degeneration. This approach has been successfully applied in multiple species, including rats, rabbits, goats, and pigs, and is widely used to investigate degenerative mechanisms and evaluate regenerative interventions^[57].

A major advantage of the needle puncture model is its high degree of experimental controllability and reproducibility. The severity of degeneration can be modulated by varying needle diameter, puncture depth, rotation, and retention time^[58]. In particular, the ratio of needle diameter to disc dimensions is an important determinant of the extent of degeneration^[59]. In mouse models, the target caudal intervertebral disc is typically exposed or localized after anesthesia, followed by insertion of a fine needle through the annulus fibrosus into the nucleus pulposus. Depending on the experimental protocol, the needle may be rotated and retained briefly before withdrawal. This procedure induces progressive degenerative changes that can be monitored by imaging and histological assessment [Figure 3A]. Typical findings include reduced T2-weighted MRI signal intensity, loss of disc height, decreased proteoglycan content, depletion of nucleus pulposus cells, and disruption of annulus fibrosus organization^[60,61].

Figure 3. Animal models of intervertebral disc degeneration. (A) Needle puncture-induced intervertebral disc degeneration models: After anesthesia and fixation, the target caudal intervertebral disc in mice is identified, and a 27-30 G needle is inserted through the annulus fibrosus into the center of the nucleus pulposus. The needle is then rotated, retained briefly, and withdrawn, thereby causing annulus fibrosus injury and nucleus pulposus disruption to induce intervertebral disc degeneration; (B) Lumbar spinal instability models: After anesthesia, the posterior lumbar structures of the experimental mouse are exposed through a posterior midline approach. Under a microscope, the bilateral inferior articular processes are resected, and the supraspinous and interspinous ligaments are transected to disrupt posterior column stability and establish a lumbar motion segment instability model; (C) Mechanical loading models: Continuous axial loading is applied using spring-loading devices, Ilizarov external fixators, or customized compression systems to simulate a chronic mechanical overload environment; (D) Chemical-induced intervertebral disc degeneration models: Specific enzymes, inflammatory cytokines, or chemical reagents are injected into the intervertebral disc to disrupt the local microenvironment and rapidly induce intervertebral disc degeneration. NP: Nucleus pulposus.

At the molecular level, puncture-induced injury is accompanied by substantial remodeling of the intradiscal microenvironment. Expression of matrix-associated molecules such as aggrecan and collagen type II alpha 1 chain (COL2A1) is typically reduced, whereas catabolic enzymes, including matrix metallopeptidase 3 (MMP3), MMP13, and members of the a disintegrin and metalloproteinase with thrombospondin motifs (ADAMTS) family, are increased during disease progression^[62]. Needle puncture models have also been widely used to investigate different forms of regulated cell death and stress responses, including apoptosis, autophagy dysregulation, pyroptosis, and ferroptosis^[63]. Studies using these models have linked reduced glutathione peroxidase 4 (GPX4) activity, lipid peroxidation, and disturbances in iron metabolism to IVDD progression, thereby supporting further investigation of ferroptosis-related therapeutic strategies^[64,65].

Despite their reproducibility and practical advantages, needle puncture models primarily represent an injury-induced form of disc degeneration. They therefore differ from the naturally evolving course of human IVDD, which typically develops over years under the combined influence of aging, genetic susceptibility, abnormal mechanical loading, metabolic dysfunction, and chronic inflammation. In contrast, puncture models initiate degeneration through an acute structural insult and subsequently reproduce many downstream degenerative changes. They are therefore particularly suitable for studying disease progression after structural injury and for evaluating therapeutic efficacy, but are less appropriate for investigating the earliest etiological events underlying spontaneous human IVDD.

Because of their technical simplicity, relatively short experimental duration, and consistent degenerative phenotype, needle puncture models remain widely used in studies of inflammation, matrix remodeling, cell death, and regenerative therapy^[65]. They are especially valuable for proof-of-concept evaluation of interventions such as stem cell transplantation, biomaterial implantation, exosome-based therapy, and gene delivery^[66-69]. However, the suitability of a specific species depends on the therapeutic strategy and experimental endpoint. Small-animal models, particularly mice and rats, are advantageous for mechanistic studies, genetic manipulation, dose exploration, and early efficacy assessment, whereas rabbits and larger animals offer greater anatomical accessibility for intradiscal delivery, longitudinal imaging, tissue-level repair assessment, and longer-term evaluation. These advantages must be balanced against higher costs, longer study periods, and greater technical demands. Accordingly, model selection for exosome- and gene-based therapies should be guided by the intended mechanistic, anatomical, and translational objectives rather than by reliance on a single universally preferred model.

Instability-induced degeneration models

Spinal stability is essential for maintaining the biomechanical environment and tissue homeostasis of the intervertebral disc. Together, the intervertebral disc, facet joints, spinal ligaments, and paraspinal muscles form the functional spinal unit (FSU), which collectively accommodates axial compression, flexion, extension, rotation, and shear forces while preserving segmental motion and structural integrity^[70]. Disruption of this coordinated mechanical system, whether through facet joint degeneration, ligamentous injury, or damage to posterior stabilizing structures, can result in segmental instability and abnormal loading of the intervertebral disc. Sustained biomechanical disturbance may subsequently promote cellular dysfunction, ECM degradation, and progressive structural deterioration of disc tissues. Spinal instability is therefore considered an important contributor to the development and progression of degenerative lumbar spinal disorders.

Based on this concept, instability-induced models have been developed to reproduce disc degeneration secondary to chronic alterations in the spinal mechanical environment^[71]. Among these, the mouse lumbar spine instability (LSI) model has been increasingly used in mechanobiological studies. The model is typically established through posterior disruption of stabilizing structures, most commonly bilateral resection of the inferior articular processes combined with transection of the supraspinous and interspinous ligaments [Figure 3B]. This procedure destabilizes the lumbar motion segment without directly injuring the intervertebral disc. Over the following weeks, progressive degenerative changes may develop, including loss of disc height, depletion or phenotypic alteration of nucleus pulposus cells, annulus fibrosus disorganization, and endplate remodeling.

A key distinction between the LSI model and needle puncture-based approaches is that degeneration arises indirectly from altered spinal biomechanics rather than from direct structural injury to the disc^[72]. This feature makes the LSI model particularly useful for investigating how sustained mechanical instability contributes to degenerative changes across the spinal motion segment. It may therefore more closely reproduce aspects of clinically relevant degeneration associated with abnormal loading, posterior element dysfunction, and progressive disruption of spinal homeostasis, although it does not fully reproduce the multifactorial natural history of human IVDD.

Recent studies using LSI models have provided important insights into the molecular mechanisms through which abnormal mechanical loading drives disc degeneration. Spinal destabilization has been associated with activation of mechanosensitive and inflammatory signaling pathways, including piezo-type mechanosensitive ion channel component 1 (PIEZO1), Yes-associated protein/transcriptional coactivator with PDZ-binding motif (YAP/TAZ), Wingless-related integration site (Wnt), and nuclear factor kappa B (NF-κB) signaling^[73-76]. These pathways may contribute to inflammatory mediator production, cellular senescence, altered cell fate, and increased expression of matrix-degrading enzymes, thereby accelerating degenerative remodeling. Importantly, instability-induced changes are not confined to the intervertebral disc. Endplate remodeling, facet cartilage degeneration, osteophyte formation, and alterations in paraspinal muscle composition have also been reported, supporting the concept that lumbar degeneration can involve multiple components of the FSU rather than the disc alone.

Because it reproduces biomechanical disturbance together with degenerative changes in several spinal tissues, the LSI model provides a valuable platform for studying mechanobiological regulation, inter-tissue crosstalk, and coordinated degeneration of the FSU. It is particularly useful for investigating how altered spinal mechanics influence cell fate, matrix homeostasis, and tissue remodeling over time.

Mechanical loading-induced degeneration models

Although IVDD is influenced by multiple factors, including genetic susceptibility, aging, metabolic dysfunction, and inflammation, abnormal mechanical loading is widely recognized as an important contributor to disease initiation and progression^[77]. Intervertebral discs are continuously exposed to complex mechanical forces, including compression, tension, torsion, and shear. Within the physiological range, these stimuli are essential for maintaining ECM homeostasis and normal disc function. Excessive or prolonged loading, however, can disrupt cellular mechanotransduction, promote matrix catabolism and inflammatory signaling, and impair cell viability, ultimately contributing to progressive structural and functional deterioration of the disc^[78]. Compared with models based on direct structural injury, mechanical loading models are particularly valuable for investigating how sustained biomechanical disturbance contributes to IVDD development^[79,80].

Mechanical loading models were initially developed using rodent caudal intervertebral discs^[81]. The caudal spine of rats and mice is readily accessible and can be subjected to controlled loading using external fixation systems, making it suitable for studies of load magnitude, duration, and frequency^[82]. Axial compression is commonly applied using spring-based devices, Ilizarov-type external fixators, or customized loading systems to reproduce conditions of sustained mechanical overload [Figure 3C]. After prolonged loading, the affected discs typically exhibit progressive degenerative changes, including reduced disc height, decreased T2-weighted MRI signal intensity, and ECM degradation^[83]. Histological findings may include proteoglycan loss within the nucleus pulposus, disruption of annulus fibrosus organization, and endplate remodeling or sclerosis^[84]. Together, these changes reproduce several important features of mechanically driven disc degeneration.

The development of mechanobiology has further expanded the use of these models for investigating how disc cells sense and respond to abnormal mechanical cues. Rather than acting as passive load-bearing cells, nucleus pulposus, annulus fibrosus, and endplate cells express a variety of mechanosensitive molecules that convert physical stimuli into intracellular biochemical signals^[85]. Mechanotransduction pathways involving PIEZO1, PIEZO2, transient receptor potential vanilloid 4 (TRPV4), integrins, and YAP/TAZ have therefore received increasing attention in IVDD research^[86]. Among these, PIEZO1 has emerged as an important mechanosensitive ion channel capable of mediating calcium influx in response to excessive mechanical stimulation. Sustained PIEZO1 activation has been associated with mitochondrial dysfunction, reactive oxygen species (ROS) accumulation, inflammatory signaling, cellular senescence, and altered cell survival^[86-90]. These findings show how mechanical loading models can link abnormal biomechanical stimuli to downstream molecular and cellular responses.

Mechanical loading models have also been widely applied to investigate stress-related cell injury and death during IVDD. In addition to apoptosis, excessive mechanical stress has been linked to ferroptosis and pyroptosis, as well as to dysregulation of autophagy and mitochondrial homeostasis^[91]. Ferroptosis has attracted particular attention because of its close association with oxidative stress, lipid peroxidation, and iron metabolism^[92]. Sustained compression has been reported to disturb iron homeostasis, increase lipid peroxide accumulation, reduce GPX4 activity, and promote oxidative damage in disc cells^[93]. Studies focusing on regulatory molecules such as solute carrier family 7 member 11 (SLC7A11) and GPX4 further suggest that ferroptosis may contribute to nucleus pulposus cell loss under mechanically adverse conditions. Mechanical loading models therefore provide a useful platform for examining how biomechanical stress interacts with oxidative injury and regulated cell-death pathways during degeneration.

Beyond cell injury, these models have also contributed to the study of progenitor cell homeostasis and regenerative capacity within the intervertebral disc^[94]. Advances in single-cell RNA sequencing have identified heterogeneous cell populations with progenitor-like characteristics and potential roles in tissue maintenance and repair^[95]. Abnormal mechanical environments may compromise the maintenance, differentiation, and reparative capacity of these cell populations^[96]. Mechanosensitive signaling pathways, including Hippo-YAP/TAZ, Wnt/β-catenin, and Notch signaling, have been implicated in the regulation of progenitor cell fate and tissue remodeling under mechanical stress^[97,98]. Mechanical loading models are therefore useful not only for studying degenerative mechanisms but also for examining how the biomechanical environment influences endogenous repair potential.

Overall, mechanical loading models provide a controllable experimental system for investigating the contribution of abnormal biomechanics to IVDD. Their main strength is the ability to manipulate the magnitude, duration, and pattern of mechanical stimulation while avoiding direct injury to the disc itself. This makes them particularly suitable for studies of mechanotransduction, stress-induced cellular responses, cell fate regulation, and tissue homeostasis. However, the loading conditions generated by external devices remain simplified representations of the complex and dynamic mechanical environment of the human spine. Accordingly, these models should be regarded as complementary platforms for dissecting mechanically driven mechanisms of degeneration rather than as complete reproductions of the natural course of human IVDD.

Chemical-induced degeneration models

In addition to mechanical injury and abnormal loading, inflammation and dysregulated ECM metabolism play important roles in the initiation and progression of IVDD^[99,100]. On this basis, a variety of chemically induced degeneration models have been developed to reproduce specific pathological features of the degenerative disc microenvironment. These models typically involve intradiscal administration of enzymes, inflammatory cytokines, or other bioactive agents to perturb matrix homeostasis, alter cellular function, and induce relatively rapid degenerative changes^[100,101] [Figure 3D].

Enzyme-induced models represent some of the earliest chemically induced approaches used to reproduce IVDD. These models rely on intradiscal administration of proteolytic or matrix-degrading enzymes to disrupt ECM integrity and induce structural deterioration of the disc^[102]. Papain, trypsin, and chondroitinase ABC (C-ABC) are among the most commonly used reagents. Papain rapidly degrades proteoglycans and collagenous matrix components, resulting in dehydration and volume loss of the nucleus pulposus. Experimental studies have shown that papain injection can produce reduced disc height, decreased T2-weighted MRI signal intensity, and increased histological degeneration scores within several weeks^[57,102]. However, the relatively rapid and extensive tissue destruction induced by papain differs substantially from the gradual progression of human IVDD, limiting its use in studies aimed at reproducing chronic degeneration.

Compared with papain, C-ABC has gained increasing attention because it preferentially degrades chondroitin sulfate-containing glycosaminoglycans that are abundant within the nucleus pulposus^[57]. This results in progressive proteoglycan depletion and loss of matrix hydration, thereby reproducing an important component of disc degeneration^[103]. The severity of degeneration can be modulated by adjusting the concentration and dose of C-ABC, allowing the generation of graded degenerative phenotypes. Because matrix depletion can be induced in a relatively controlled manner while preserving the overall architecture of the disc, C-ABC models are particularly useful for evaluating regenerative strategies, including cell-based therapies, injectable hydrogels, and other tissue-engineering approaches^[104,105].

With increasing recognition of the role of inflammation in IVDD, cytokine-induced models have also become widely used experimental systems^[106,107]. Clinical and experimental evidence indicates that degenerative discs are characterized by persistent inflammatory activity, with cytokines such as tumor necrosis factor alpha (TNF-α), interleukin-1 beta (IL-1β), IL-6, and IL-17 contributing to matrix catabolism, cellular dysfunction, and disease progression^[107,108]. Based on these observations, intradiscal administration of pro-inflammatory cytokines, particularly TNF-α and IL-1β, has been used to reproduce selected features of the inflammatory microenvironment associated with IVDD^[109,110]. In contrast to enzyme-induced models, which primarily target matrix integrity, cytokine-based approaches better suit investigation of inflammatory signaling, cellular stress responses, and inflammation-driven remodeling of the disc microenvironment.

IL-1β and TNF-α are among the most frequently used cytokines in these models. IL-1β activates signaling pathways such as NF-κB and mitogen-activated protein kinase (MAPK), increases the expression of matrix-degrading enzymes including MMP3, MMP13, and ADAMTS5, and suppresses matrix-associated molecules such as aggrecan and COL2A1^[111]. TNF-α similarly promotes inflammatory signaling and matrix catabolism and has also been associated with increased expression of neurotrophic mediators such as nerve growth factor (NGF) and brain-derived neurotrophic factor (BDNF), making TNF-α-based models useful for studies of inflammation-associated discogenic pain. Combined cytokine stimulation has also been explored as a means of generating a more sustained and complex inflammatory environment than that produced by a single cytokine alone^[108].

Chemically induced models are also useful for investigating stress responses and regulated cell-death pathways associated with IVDD. Pro-inflammatory stimulation can promote oxidative stress, mitochondrial dysfunction, lipid peroxidation, and disturbances in iron metabolism. In IL-1β-based models, for example, increased ROS production and impaired antioxidant defense have been associated with ferroptosis-related changes^[112,113]. Inflammatory stimulation may also activate the NOD-like receptor family pyrin domain containing 3 (NLRP3) inflammasome, promote caspase-1 activation and gasdermin D (GSDMD) cleavage, and thereby contribute to pyroptotic cell death^[114,115]. These models therefore provide useful experimental platforms for studying the relationships among inflammation, oxidative stress, ferroptosis, pyroptosis, autophagy dysregulation, and cellular senescence.

Despite their experimental flexibility, chemically induced models have several inherent limitations. Most rely on artificial, simplified stimuli and therefore do not reproduce the multifactorial etiology of human IVDD. Some reagents induce rapid matrix destruction or intense inflammatory responses that differ from the chronic and progressive nature of naturally occurring degeneration. In addition, differences in reagent type, concentration, injection volume, treatment frequency, and experimental species can substantially influence disease severity and reduce comparability across studies. Chemically induced models are therefore best suited to investigating defined pathological processes or testing specific therapeutic mechanisms rather than reproducing the entire natural history of IVDD.

Overall, chemically induced models provide efficient and controllable approaches for studying matrix degradation, inflammation, oxidative stress, and related cellular responses in IVDD. Enzyme-based models are particularly useful for reproducing matrix depletion and evaluating regenerative interventions, whereas cytokine-based models are better suited to studies of inflammatory signaling, cell fate regulation, and discogenic pain. Future strategies that combine chemical stimulation with other disease-relevant factors, such as abnormal mechanical loading, aging, or genetic susceptibility, may provide more physiologically representative models for studying the complex pathogenesis of IVDD.

Genetically engineered models

Although needle puncture, mechanical loading, and chemically induced models can reproduce selected pathological features of IVDD, these exogenously induced models are less suitable for directly investigating the genetic determinants and cell-specific regulatory mechanisms involved in disease initiation and progression. With advances in genetic engineering, particularly Cre-LoxP-based conditional targeting and CRISPR/Cas9-mediated genome editing, genetically engineered models have become increasingly important for studying intervertebral disc development, tissue homeostasis, and degeneration. Unlike conventional induction-based approaches, these models enable targeted manipulation of specific genes in defined cell populations or developmental stages, providing a powerful way to dissect gene function and validate candidate therapeutic targets.

Early genetically engineered IVDD studies focused largely on genes involved in ECM organization and maintenance. Proteoglycans and collagen fibers are major structural components of the intervertebral disc, and their balanced synthesis and turnover are essential for preserving tissue hydration and biomechanical function^[116]. Deficiency of aggrecan (ACAN) can lead to impaired disc development and premature degenerative changes, including reduced nucleus pulposus hydration, proteoglycan depletion, and loss of disc height^[117]. Similarly, disruption or mutation of COL2A1 compromises type II collagen production and may result in abnormal annulus fibrosus organization and impaired matrix integrity^[118]. These findings provided early genetic evidence that disruption of ECM homeostasis can directly contribute to disc degeneration.

Transcriptional regulators are equally important for maintaining disc cell phenotype and matrix homeostasis. SRY-box transcription factor 9 (SOX9), for example, is a key regulator of chondrogenic differentiation and ECM synthesis and controls the expression of cartilage-associated genes, including COL2A1^[119,120]. Conditional loss of SOX9 has been associated with reduced ACAN and COL2A1 expression, altered cellular phenotype, and progressive disruption of intervertebral disc architecture^[121]. Such models illustrate that IVDD may arise not only from external injury or mechanical stress but also from disruption of intrinsic developmental and homeostatic regulatory programs.

Despite their mechanistic value, genetically engineered models have several limitations. Global gene deletion may cause severe developmental abnormalities or embryonic lethality, making it difficult to distinguish developmental defects from adult-onset degeneration. Conditional and inducible systems are therefore increasingly preferred because they allow greater spatial and temporal control of gene manipulation. In addition, single-gene perturbations cannot fully reproduce the multifactorial nature of human IVDD, which reflects interactions among genetic susceptibility, aging, mechanical stress, metabolic disturbance, and inflammation. Generating and maintaining genetically modified animals can also be time-consuming and costly, limiting their use in large-scale or routine studies.

Overall, genetically engineered models have substantially expanded the ability to investigate the molecular and cellular basis of IVDD. They are particularly valuable for defining cell-specific gene functions, identifying causal regulatory pathways, and distinguishing developmental abnormalities from degenerative processes. Integration of conditional genetic models with lineage tracing, single-cell omics, and spatial transcriptomics is expected to provide a more detailed understanding of the cellular dynamics and regulatory networks that shape disc degeneration and repair.

Spontaneous degeneration models

Although experimentally induced models have contributed substantially to IVDD research, most rely on external interventions, such as structural injury, abnormal mechanical loading, chemical stimulation, or genetic manipulation, to accelerate degenerative changes. By contrast, human IVDD typically develops gradually over many years under the combined influence of aging, genetic susceptibility, mechanical stress, metabolic alterations, chronic inflammation, and declining endogenous repair capacity. Spontaneous degeneration models are therefore of particular interest because they allow age-associated degenerative changes to develop without direct experimental induction and may reproduce aspects of the natural history of human IVDD more closely than acute injury-based models.

A defining feature of spontaneous degeneration models is the progressive development of disc degeneration with age in the absence of deliberate local injury. These models enable longitudinal investigation of the transition from early disturbances in tissue homeostasis to progressive matrix loss, cellular dysfunction, and structural deterioration. They are particularly useful for studying aging-related mechanisms, chronic changes in the disc microenvironment, and the long-term effects of therapeutic interventions. However, their translational relevance remains dependent on species-specific differences in disc biology, lifespan, biomechanics, and patterns of age-related degeneration.

Sand rats are among the best-established spontaneous models used in IVDD research. These animals develop progressive age-associated changes in the intervertebral disc, including reduced nucleus pulposus cellularity, ECM deterioration, and endplate calcification^[122]. Their gradual degenerative phenotype makes them useful for investigating mechanisms associated with aging and chronic disc degeneration^[123]. Nevertheless, limited availability, specialized breeding requirements, and relatively restricted experimental resources have constrained their broader use.

Aged rodents have also become increasingly important for studying naturally occurring, age-associated disc degeneration. Wild-type mice and rats develop progressive structural and molecular changes in the intervertebral disc with advancing age, including loss of disc height, proteoglycan depletion, increased inflammatory activity, and accumulation of senescent cells^[116,124]. These alterations are generally more pronounced in older animals and provide a useful platform for examining processes such as cellular senescence, mitochondrial dysfunction, chronic low-grade inflammation, and impaired regenerative capacity during IVDD progression^[125]. Compared with rapidly induced models, aged rodents are particularly valuable for studying mechanisms linked to the gradual decline in disc homeostasis over time.

Despite these advantages, spontaneous degeneration models have several practical limitations. The development of stable degenerative phenotypes requires prolonged observation, resulting in substantially longer experimental timelines than those of induced models. Disease severity and progression can also vary among individual animals, reducing experimental consistency and increasing sample-size requirements. In addition, long-term housing, maintenance costs, and ethical considerations may further limit their use, particularly in larger species. For these reasons, spontaneous models are often used as complementary systems alongside induced models rather than as replacements.

Overall, spontaneous degeneration models provide valuable opportunities to investigate age-associated changes, chronic tissue remodeling, and progressive loss of disc homeostasis under conditions that do not require direct experimental injury. Their greatest strength lies in modeling the temporal evolution of degeneration, whereas their major limitations include long study duration, inter-individual variability, and species-specific differences. Combining spontaneous models with appropriately selected induced models may therefore provide a more comprehensive strategy for studying IVDD mechanisms and evaluating long-term therapeutic responses.

Autologous nucleus pulposus transplantation models

The autologous nucleus pulposus transplantation model is a well-established experimental approach for investigating the inflammatory and neurobiological mechanisms associated with LDH. Early studies by Olmarker et al. demonstrated that exposure of nucleus pulposus tissue to neural structures can induce inflammatory changes and nerve root dysfunction even in the absence of substantial mechanical compression^[126]. In this model, autologous nucleus pulposus tissue is typically harvested from the caudal or lumbar intervertebral discs and placed adjacent to the lumbar nerve root or dural sac [Figure 4A]^[127,128]. Following transplantation, local inflammatory responses may develop rapidly, characterized by macrophage recruitment, increased production of pro-inflammatory mediators, nerve root edema, and pain-related behavioral changes. Because sustained mechanical compression is absent, this model is particularly useful for isolating the biological effects of exposed nucleus pulposus tissue on neural and immune tissues.

Figure 4. Animal models of lumbar disc herniation. (A) Autologous nucleus pulposus transplantation models: After anesthesia, the experimental animal is placed in the prone position, and autologous nucleus pulposus tissue is harvested from the caudal or lumbar intervertebral disc. The target lumbar nerve root (L4-L5 or L5-L6) is then exposed, and the harvested nucleus pulposus tissue is transplanted around the nerve root or dural sac. The incision is subsequently closed, followed by postoperative observation and relevant assessments; (B) Nerve root compression models: After anesthesia, the experimental animal is placed in the prone position, and the lamina and lumbar nerve root are exposed through a posterior midline lumbar incision. A silicone sheet, polyethylene tube, or other compression device is then placed around the nerve root to establish a sustained mechanical compression model. The incision is closed after surgery, followed by postoperative observation and evaluation of relevant outcome measures. NP: Nucleus pulposus; DRG: dorsal root ganglion.

Under physiological conditions, the nucleus pulposus is largely separated from systemic immune surveillance by the surrounding annulus fibrosus and cartilaginous endplates. Disruption of this anatomical isolation during disc herniation exposes nucleus pulposus components to the perineural environment and can initiate innate immune activation and recruitment of inflammatory cells. Studies using transplantation models have implicated mediators such as TNF-α, IL-1β, IL-6, and C-C motif chemokine ligand 2 (CCL2) in the development of neuroinflammation, nerve root sensitization, and pain-related responses following nucleus pulposus exposure^[129]. These findings have helped establish inflammation as an important component of LDH pathophysiology and have provided an experimental basis for evaluating anti-inflammatory therapeutic strategies.

The principal strengths of the autologous nucleus pulposus transplantation model include good experimental reproducibility, relatively standardized surgical procedures, and the ability to examine inflammation-driven neural responses independently of persistent mechanical compression. However, because the model does not reproduce the complete process of annular rupture, tissue extrusion, and sustained nerve root compression observed in many patients with LDH, it is best regarded as a model of the biological and inflammatory consequences of nucleus pulposus exposure rather than a complete representation of clinical disc herniation.

Nerve root compression models

Nerve root compression models are designed to reproduce the mechanical component of LDH-associated nerve injury^[130]. In these models, researchers apply sustained compression to the lumbar nerve root using materials or devices such as silicone sheets, vascular clips, suture ligation, or customized compression systems [Figure 4B]. These approaches have been widely used to investigate the consequences of persistent mechanical compression on neural function and pain behavior. Experimental studies have shown that nerve root compression can induce ischemia, edema, demyelination, and axonal injury, together with behavioral manifestations such as mechanical allodynia and thermal hyperalgesia^[131-133].

However, mechanical compression alone does not fully explain the heterogeneity of symptoms observed in patients with LDH. The severity of radiographic nerve root compression does not always correlate closely with pain intensity or neurological dysfunction, indicating that additional biological factors contribute to symptom development^[134]. Increasing evidence supports an important role for inflammation and immune activation in modulating nerve root sensitization and pain after disc herniation. Accordingly, combined models that integrate mechanical compression with exposure to autologous nucleus pulposus tissue have been developed to reproduce both the mechanical and inflammatory components of LDH. These models may provide a more representative platform for investigating the complex interactions between neural compression, local inflammation, and pain-related responses.

Melatonin deficiency-based and biomechanics-based models

Melatonin-related abnormalities and altered spinal biomechanics represent two historically important hypotheses in scoliosis research. Early studies showed that pinealectomy in juvenile chickens can induce progressive spinal curvature with several radiographic features resembling those of human AIS, providing experimental support for a potential role of melatonin signaling in scoliosis development^[135,136]. However, subsequent studies demonstrated that melatonin deficiency alone does not consistently induce scoliosis across species, suggesting that impaired melatonin signaling is unlikely to act as an independent cause of AIS and may instead interact with other developmental or biomechanical factors^[137].

Biomechanical models have therefore been developed to investigate how altered loading influences spinal growth and deformity progression. Bipedal models, for example, are established by modifying normal quadrupedal posture to increase axial loading and reproduce selected features of the upright mechanical environment of the human spine^[138]. Neither pinealectomy nor bipedalization alone consistently produces scoliosis with high penetrance, whereas combining melatonin-related abnormalities with altered mechanical loading can increase the incidence and stability of scoliosis-like deformities^[138-140]. C57BL/6J mice, which exhibit low endogenous melatonin production because of defects in melatonin-synthesis pathways, have been used in combination with bipedalization to investigate interactions between melatonin signaling, spinal growth, and biomechanical loading [Figure 5A]. These models are particularly useful for exploring how neuroendocrine and mechanical factors may interact during scoliosis development, although they do not reproduce the full complexity of human AIS.

Figure 5. Animal models of scoliosis. (A) Melatonin deficiency-based models: Melatonin deficiency is established either by pinealectomy or by using animals with naturally low melatonin levels, followed by bipedalization to simulate abnormal vertical loading. The combined effects of melatonin deficiency and altered mechanical loading can induce AIS-like scoliosis; (B) Biomechanics-based models: Local mechanical intervention models, including rib traction, asymmetric muscle traction, and asymmetric spinal loading, induce scoliosis through sustained asymmetric mechanical stress and are used to investigate the mechanobiological mechanisms underlying AIS progression; (C) Genetic models: Based on AIS susceptibility genes identified through GWAS and sequencing analyses, mouse and zebrafish models are generated using gene-editing technologies. Phenotypic assessments of spinal curvature, growth and development, and neuromuscular function are then performed to elucidate the genetic factors and molecular mechanisms associated with AIS. AIS: Adolescent idiopathic scoliosis; GWAS: genome-wide association studies.

In addition to bipedal models, several approaches apply asymmetric mechanical forces directly on the growing spine, including rib tethering, asymmetric muscle traction, and spinal tethering^[141,142]. These models are commonly used to investigate the effects of asymmetric loading on vertebral growth, disc morphology, and curve progression [Figure 5B]. Among them, porcine tethered scoliosis models are widely used because the size and growth characteristics of the porcine spine allow clinically relevant instrumentation and longitudinal assessment. By applying unilateral or asymmetric tethering during growth, these models can induce vertebral wedging, asymmetric disc growth, coronal curvature, and axial rotation^[143-145]. Modified tethering techniques, including posterior or less invasive approaches, have further improved model consistency and reduced procedure-related morbidity.

Porcine tethering models are particularly valuable for studying growth modulation and for evaluating non-fusion corrective strategies, including vertebral body tethering and growth-preserving instrumentation^[146,147]. However, the deformity is produced by externally imposed mechanical asymmetry and therefore does not reproduce the spontaneous onset, genetic susceptibility, neuromuscular regulation, or other biological features associated with AIS. Moreover, some tethering paradigms may more closely resemble mechanically induced or early-onset scoliosis than idiopathic adolescent scoliosis. These models should therefore be viewed primarily as platforms for studying mechanically driven curve progression and testing corrective interventions rather than as complete models of AIS pathogenesis.

Overall, melatonin-related and biomechanics-based models have contributed substantially to understanding how neuroendocrine signaling and asymmetric mechanical loading may influence scoliosis development. Their phenotypic characteristics, reproducibility, and clinical relevance vary considerably across species and modeling protocols. Accordingly, these models are best used to address specific mechanistic or translational questions rather than to reproduce the entire pathophysiological spectrum of human AIS.

Genetic models

Advances in genome-wide association studies (GWAS) and high-throughput sequencing have led to the identification of an increasing number of genetic loci and candidate genes associated with AIS^[148] [Figure 5C]. Animal models based on these genetic findings provide valuable tools for investigating the molecular pathways that contribute to spinal development, growth regulation, and deformity formation. Frequently reported AIS-associated genes include LBX1, ADGRG6 (GPR126), PTK7, PAX1, BNC2, and SOX9. In parallel, genes implicated in syndromic forms of scoliosis, such as CHD7 and FBN1, have provided additional insight into biological pathways relevant to spinal deformity^[149-152].

Using global knockout, conditional knockout, transgenic, and other genetic manipulation strategies, researchers have generated a range of animal models exhibiting spinal curvature, abnormal skeletal growth, vertebral developmental defects, or neuromuscular abnormalities. These models support a key role for genetic susceptibility in scoliosis pathogenesis while also highlighting the heterogeneity of the biological mechanisms involved. However, the phenotypes produced by individual gene perturbations often reflect specific developmental pathways and should not necessarily be regarded as direct reproductions of human AIS.

Zebrafish have become particularly useful for genetic studies of scoliosis because of their rapid development, high experimental throughput, and amenability to genetic manipulation^[153]. Mutations or deficiencies involving PTK7, KIF6, and other genes related to ciliary function have been shown to produce progressive spinal curvature in zebrafish^[154-156]. These models provide tractable systems for investigating how abnormalities in ciliary biology, cerebrospinal fluid dynamics, skeletal development, and related signaling pathways contribute to scoliosis-like phenotypes.

Osteoporosis-associated vertebral compression fracture models

Osteoporosis-associated vertebral compression fracture models are widely used to investigate fracture development and repair under compromised bone-quality conditions. Osteoporotic phenotypes are commonly induced through ovariectomy (OVX), glucocorticoid administration, aging, or combined approaches, resulting in reduced bone mass, deterioration of trabecular architecture, and decreased vertebral mechanical strength^[157-160]. Vertebral compression fractures are subsequently generated by applying controlled axial loading to the vertebral body until structural failure occurs [Figure 6A].

Figure 6. Animal models of vertebral fracture and spinal fusion. (A) Osteoporosis-associated vertebral compression fracture models: Bilateral OVX is performed to induce osteoporosis, followed by 12 weeks to establish stable bone loss. The target vertebral body is then exposed, and axial compression is applied using a customized device to induce a vertebral compression fracture. Fracture healing is evaluated by micro-CT, histology, and biomechanical testing; (B) Vertebral bone defect models: The target vertebral body is exposed through a posterior approach, and a standardized bone defect is created using a trephine or high-speed burr. Experimental materials are implanted into the defect, and bone regeneration is subsequently assessed by micro-CT and histological analysis; (C) Posterolateral fusion models: The bilateral L4-L5 transverse processes are exposed and decorticated, followed by implantation of test materials into the fusion bed. Fusion outcomes are evaluated using X-ray, micro-CT, and histological analysis; (D) Interbody fusion models: The target lumbar segment (L4-L5) is exposed, followed by discectomy and endplate preparation. Bone graft materials are implanted into the intervertebral space, and interbody fusion is assessed by X-ray, micro-CT, histological staining, and biomechanical testing. OVX: Ovariectomy; CT: computed tomography.

These models reproduce several clinically relevant features of OVCFs, including loss of vertebral height, trabecular disruption, deterioration of bone microarchitecture, and subsequent remodeling responses. Because the magnitude and rate of compressive loading can be experimentally controlled, fracture severity can be adjusted according to specific study objectives, providing relatively good reproducibility across standardized protocols. Such models are widely used to investigate fracture healing, bone remodeling, and the interaction between osteoporosis and vertebral repair, as well as to evaluate interventions such as bone cement, injectable biomaterials, tissue-engineered scaffolds, and anti-osteoporotic therapies^[161].

Overall, osteoporosis-associated vertebral compression fracture models provide a useful preclinical platform for studying vertebral injury in osteoporotic bone and for assessing therapeutic strategies aimed at improving fracture stabilization, bone regeneration, and skeletal quality. However, the experimentally induced fracture pattern and healing environment remain simplified relative to the heterogeneous clinical presentations of human OVCF, and these differences should be considered when interpreting translational outcomes.

Vertebral bone defect models

Vertebral bone defect models are widely used to investigate bone regeneration and to evaluate tissue-engineering strategies. Standardized defects are typically created within the vertebral body using high-speed burrs, trephines, or bone drills, allowing the size and location of the defect to be controlled according to the experimental design^[162,163] [Figure 6B].

Compared with vertebral compression fracture models, bone defect models offer several practical advantages, including reproducible defect geometry, controllable injury severity, and convenient implantation of biomaterials or cell-based constructs. These characteristics make them particularly suitable for evaluating the osteogenic and regenerative performance of bone graft substitutes, bioactive ceramics, tissue-engineered scaffolds, and stem cell-based therapies^[164-166]. Rabbits and sheep are commonly used for vertebral bone defect studies because their vertebral dimensions provide sufficient surgical access and space for implantation, while also permitting radiological, histological, and biomechanical assessment of bone repair^[167-171]. Large-animal models such as sheep are especially valuable when clinically relevant implant dimensions and load-bearing conditions are required.

Nevertheless, vertebral bone defect models primarily reproduce localized osseous loss rather than the complex fracture patterns and systemic skeletal abnormalities encountered in many clinical conditions. They are therefore best suited to studying bone regeneration and biomaterial performance rather than the full pathophysiology of vertebral fracture.

Posterolateral fusion models

Posterolateral fusion models are widely used to investigate the biological processes underlying spinal arthrodesis and to evaluate bone grafts and regenerative strategies. In these models, the bilateral posterolateral elements, typically including the transverse processes, are surgically exposed and decorticated, followed by implantation of autologous bone, allogeneic bone, or synthetic bone substitutes to promote new bone formation and bridging fusion between adjacent vertebral segments^[172,173] [Figure 6C].

These models offer relatively standardized surgical procedures and well-established outcome measures, including radiographic assessment, histological analysis, manual palpation, and biomechanical testing. They are therefore frequently used to evaluate the osteogenic and fusion-promoting effects of biomaterials, bone graft substitutes, growth factors, and cell-based therapies^[174,175]. However, posterolateral fusion occurs outside the intervertebral disc space and does not fully reproduce the axial loading environment or implant-endplate interactions encountered in interbody fusion. Accordingly, this model is particularly useful for studying bone formation and posterolateral arthrodesis, but its findings should be interpreted cautiously when extrapolated to other forms of clinical spinal fusion.

Interbody fusion models

Interbody fusion models more closely reproduce the surgical environment of clinical spinal fusion than posterolateral approaches. These models are typically established by removing the intervertebral disc and implanting a fusion cage, bone graft, or other interbody construct into the disc space [Figure 6D]. Because the implant is positioned within the load-bearing intervertebral space, these models allow evaluation of fusion progression, osseointegration, implant stability, and biomechanical performance under conditions that more closely resemble clinical interbody fusion. They are therefore widely used for the assessment of fusion cage design, bone graft substitutes, biomaterials, and tissue-engineering strategies^[176-178].

With advances in three-dimensional printing and patient-specific implant design, interbody fusion models have become increasingly important for preclinical evaluation of novel spinal implants and regenerative constructs. Their principal advantage lies in the ability to assess interactions among the implant, vertebral endplates, bone graft material, and local mechanical environment. However, these models generally require more technically demanding surgical procedures and are often performed in medium- or large-animal species, resulting in greater experimental cost and complexity.

Emerging models for bone regeneration and biomaterial evaluation

Advances in regenerative medicine have increasingly expanded the application of vertebral fracture and spinal fusion models beyond conventional studies of bone healing and arthrodesis. Osteoporotic fracture, vertebral bone defect, and spinal fusion models are now frequently combined with tissue-engineering approaches to evaluate stem cell-based therapies, bioactive scaffolds, injectable hydrogels, gene delivery strategies, and other regenerative interventions.

At the same time, improvements in micro-CT, biomechanical testing, histomorphometric analysis, and multi-omics technologies have enabled more detailed assessment of bone regeneration across structural, functional, cellular, and molecular levels. These integrated approaches allow investigators to evaluate not only the extent of new bone formation and fusion, but also the underlying biological responses, material-tissue interactions, and remodeling processes.

Accordingly, contemporary vertebral fracture and spinal fusion models increasingly serve as translational platforms for both mechanistic investigation and preclinical evaluation of regenerative therapies. Their value lies in combining clinically relevant skeletal injury or fusion environments with multidimensional outcome assessment, thereby providing a more informative framework for comparing emerging biomaterials and bone-regenerative strategies.

Contusion injury models

Spinal cord contusion models are among the most widely used experimental approaches for traumatic SCI. They reproduce several key features of clinically relevant blunt spinal cord trauma, including an initial mechanical impact followed by progressive secondary tissue damage. In experimental settings, the spinal cord is typically exposed by laminectomy and subjected to a standardized impact using devices that allow controlled adjustment of injury severity through parameters such as impact force, displacement, or drop height^[179] [Figure 7A].

Figure 7. Animal models of SCI. (A) Contusion injury models: After anesthesia, the animal is placed in the prone position, and laminectomy is performed at the target thoracic level, typically T9-T10, to expose the spinal cord while preserving the dura. A standardized impactor, such as the NYU or Infinite Horizon Impactor, is then used to deliver a controlled contusion injury. After surgery, animals receive routine postoperative care, and injury severity and recovery are assessed using locomotor tests, histological analysis, and immunological evaluation; (B) Compression injury models: After laminectomy and spinal cord exposure under anesthesia, a calibrated aneurysm clip is applied to the target segment for 30-60 s to induce compression injury. Neurological impairment and recovery are subsequently evaluated using the BBB locomotor score, histological analysis, and imaging examinations; (C) Complete transection and hemisection models: In the complete transection model, the exposed spinal cord is completely severed at the target level, with a small segment of tissue removed when necessary to prevent reconnection. This model is widely used to study axonal regeneration, neural tissue engineering, and cell-based therapies. In the hemisection model, only one half of the spinal cord is transected while the contralateral side remains intact, making it particularly suitable for investigating neural plasticity, axonal regeneration, neural circuit reconstruction, and functional recovery. NYU: New York University impactor; IH: Infinite Horizon impactor; BBB: Basso, Beattie, and Bresnahan; MRI: magnetic resonance imaging; SCI: spinal cord injury.

Following contusion injury, animals typically develop pronounced motor deficits accompanied by acute pathological changes such as hemorrhage, edema, vascular disruption, and neuronal loss. Secondary injury subsequently evolves over time and may include inflammatory cell infiltration, axonal degeneration, demyelination, glial scar formation, and development of cystic cavities^[180-182]. Because this model captures both the initial mechanical insult and the subsequent secondary injury cascade, it is widely used to investigate neuroprotective mechanisms, cell-based therapies, biomaterial-based interventions, and strategies to promote neural repair and functional recovery.

Recent advances in high-throughput and single-cell sequencing have further expanded the application of contusion models by revealing dynamic changes in the cellular composition and molecular state of the injured spinal cord. These studies have identified time-dependent responses involving microglia, infiltrating macrophages, reactive astrocytes, oligodendrocyte-lineage cells, and other neural and immune populations^[183]. Such findings have improved understanding of the complex cellular interactions that shape secondary injury and repair after SCI.

Compression injury models

Spinal cord compression models reproduce neural injury associated with sustained or progressive mechanical compression, as may occur in vertebral fracture or dislocation, spinal canal stenosis, and space-occupying lesions. In contrast to contusion models, which primarily reproduce a brief mechanical impact, compression models allow the duration and magnitude of spinal cord deformation to be experimentally controlled. Common approaches include vascular clip compression, balloon compression, and screw- or device-based compression. Among these, vascular clip models are widely used because they provide relatively standardized and reproducible injury severity. In a typical procedure, a calibrated vascular clip is applied to the exposed spinal cord for a defined period, producing neuronal loss, axonal disruption, vascular compromise, and ischemic injury [Figure 7B]^[184,185].

Chronic compression models have also been developed to investigate slowly progressive spinal cord dysfunction, particularly in the context of degenerative cervical myelopathy. These models typically use expandable materials or gradually advancing compression devices to impose mild but persistent pressure on the spinal cord over an extended period^[186,187]. The resulting pathological changes may include progressive axonal degeneration, demyelination, vascular dysfunction, and neurological decline, thereby reproducing selected features of chronic compressive myelopathy. Such models are particularly useful for studying the mechanisms of gradual neurological deterioration, the effects of prolonged compression, and functional recovery after surgical decompression.

Complete transection and hemisection models

Spinal cord transection models are widely used to investigate axonal regeneration, neural plasticity, and circuit reconstruction after SCI. Depending on the extent of tissue disruption, these models are generally classified as complete transection or hemisection injuries [Figure 7C].

In complete transection models, the spinal cord is fully severed, completely interrupting ascending and descending neural pathways. The clearly defined lesion boundary and absence of residual neural continuity make this model particularly useful for determining whether regenerating axons can traverse the injury site and for evaluating neural grafts, biomaterial scaffolds, and other strategies designed to promote structural reconnection^[188]. However, complete transection is relatively uncommon in clinical SCI and therefore has limited value for reproducing the broader spectrum of traumatic spinal cord injuries encountered in patients.

Hemisection models, in contrast, disrupt one side of the spinal cord while preserving partial neural continuity on the contralateral side^[189,190]. This configuration allows investigators to examine interactions between injured and spared pathways and to study compensatory plasticity during functional recovery. Because residual neural circuits can undergo structural and functional remodeling after hemisection, these models are particularly valuable for investigating neural plasticity, axonal sprouting, and circuit reorganization.

More recently, transection and hemisection models have been combined with neural tracing, optogenetic, and chemogenetic approaches to define the contribution of specific neuronal populations and pathways to post-injury remodeling and functional recovery. These techniques have expanded the utility of transection models for mechanistic studies of neural circuit repair after SCI.

Pyogenic spondylodiscitis and spinal tuberculosis models

Pyogenic spondylodiscitis and spinal tuberculosis models are commonly established by direct inoculation of pathogenic microorganisms into the vertebral body or intervertebral disc [Figure 8A]. Depending on the experimental objective, inoculation can be performed within vertebral cancellous bone, the endplate region, or the intervertebral disc to induce different patterns of osseous and disc involvement. Following inoculation, animals may develop progressive inflammatory changes, bone destruction, disc space narrowing, and extension of infection into adjacent soft tissues. Disease severity and progression can be evaluated using radiological, histological, and microbiological methods. These models provide useful platforms for investigating host-pathogen interactions, tissue destruction, and therapeutic responses in spinal infection.

Figure 8. Animal models of spinal infection. (A) Pyogenic spondylodiscitis and spinal tuberculosis models: The target vertebral body is exposed under anesthesia, and a bone defect is created for inoculation with either pyogenic bacteria or Mycobacterium tuberculosis. Infection establishment, bone destruction, and inflammatory responses are subsequently evaluated using imaging, histological, and microbiological analyses. Apart from differences in the inoculated pathogen and disease course, the overall modeling procedures are largely similar; (B) Implant-associated infection models: Spinal implants, such as titanium screws or rods, are placed near the lamina or pedicle, followed by local inoculation of a bacterial suspension around the implant. Bacterial adhesion and biofilm formation on the implant surface gradually establish a stable implant-associated infection. Infection severity and tissue pathological changes are assessed by imaging, colony counting, histological analysis, and biofilm evaluation. Micro-CT: Micro-computed tomography; MRI: magnetic resonance imaging; HE: hematoxylin and eosin staining; TB: toluidine blue staining; CFU: colony-forming unit; SEM: scanning electron microscopy; CLSM: confocal laser scanning microscopy.

Staphylococcus aureus is the most commonly used pathogen for establishing pyogenic spondylodiscitis models, reflecting its clinical importance in vertebral and disc infections. Other microorganisms, including Escherichia coli, Pseudomonas aeruginosa, and methicillin-resistant Staphylococcus aureus (MRSA), have also been used to reproduce different infectious conditions. Depending on the pathogen, inoculum, and animal species, acute or subacute lesions can develop over days to weeks and may be characterized by inflammatory cell infiltration, bone destruction, disc space narrowing, and abscess formation^[194,195]. Because of their relatively rapid and reproducible disease course, these models are well suited to studies of bacterial virulence, host inflammatory responses, infection-associated bone loss, and the efficacy of systemic antibiotics, local drug-delivery systems, and anti-infective biomaterials.

Spinal tuberculosis models are generally established by inoculating Mycobacterium tuberculosis (M. tuberculosis) or related mycobacterial strains into the vertebral body. Rabbits are frequently used because their vertebral size facilitates surgical inoculation and subsequent imaging and pathological assessment^[196,197]. Compared with pyogenic infection models, however, spinal tuberculosis models require longer experimental periods and more stringent biosafety measures because of the slow growth and pathogenicity of M. tuberculosis. To improve experimental safety and feasibility, some studies have explored attenuated or alternative mycobacterial strains, although the pathological fidelity of these approaches should be interpreted in relation to the specific organism used^[198].

Following infection, animals may gradually develop chronic granulomatous inflammation, caseous necrosis, progressive vertebral destruction, collapse, and, in advanced cases, spinal deformity or neurological compromise. Imaging modalities such as micro-CT and, in selected settings, positron emission tomography (PET)-CT can facilitate longitudinal assessment of lesion development and bone destruction. These models are used to evaluate antituberculous therapies and to investigate immune processes involving macrophages, T cells, granuloma formation, and host-directed therapeutic responses^[199,200].

Fungal and parasitic spinal infection models have also been described using related inoculation strategies. However, these models remain relatively uncommon, and standardized experimental systems have not yet been established for most of these pathogens.

Implant-associated infection models

Implant-associated infection is an important complication of spinal instrumentation and is characterized by the ability of microorganisms to adhere to implant surfaces and form biofilms. Biofilm formation can substantially reduce bacterial susceptibility to antimicrobial agents and facilitate persistence by limiting effective host immune clearance. To reproduce these features experimentally, titanium screws, rods, plates, or other implant materials are placed in or adjacent to the spine and subsequently inoculated with selected bacterial strains^[201,202] [Figure 8B]. Following inoculation, bacteria can adhere to the implant surface and develop organized biofilm communities, resulting in persistent local infection and surrounding tissue inflammation.

Compared with planktonic bacteria, biofilm-associated microorganisms exhibit altered metabolic states and increased tolerance to antimicrobial treatment, making implant-associated infections particularly difficult to eradicate. These models are therefore widely used to investigate biofilm formation, host responses to infected implants, and the interactions between microorganisms and biomaterial surfaces. They also provide useful preclinical platforms for evaluating antibacterial coatings, local drug-delivery systems, surface modifications, and novel implant materials designed to reduce bacterial adhesion and biofilm development^[203].

However, outcomes can be strongly influenced by factors such as bacterial species and inoculum, implant material and surface properties, implantation site, and the timing of bacterial challenge. Standardization of these parameters is therefore important for improving reproducibility and for comparing anti-infective strategies across studies.