Endoscopic spine surgery (ESS) has emerged as a minimally invasive alternative to traditional open surgery, offering several advantages that include reduced soft tissue damage, lower hospitalization rates, and faster patient recovery [1,2]. ESS facilitates earlier postoperative mobilization and a quicker return to work. Moreover, its reoperation rate is comparable to other minimally invasive techniques, such as microdiscectomy. ESS is also a valid option for elderly patients with surgical risks who suffer from spinal stenosis. However, ESS is not without risks, with overall complication rates reported as 9.26% [3], reaching up to 21% in some series. The leading countries in ESS, based on publication volume, are China, South Korea, and the United States, yet the technique is experiencing significant growth in Europe [4]. Despite increasing recognition of ESS and its benefits, surgeons face adoption challenges, mainly due to limited funding for specialized training, a lack of hands-on courses, and the high cost of equipment, factors that collectively increase procedural expenses [5]. These training limitations represent a critical obstacle to the safe implementation of ESS. The factors mentioned above substantially contribute to the difficulty of overcoming the ESS learning curve, necessitating innovative training approaches to ensure safety and competence. The technically demanding nature of ESS, combined with its limited visualization and confined working space, requires structured, progressive training before clinical application. In this context, developing effective training models becomes fundamental. Given the ethical, logistical, and financial limitations associated with cadaveric models, large animal models have become essential tools for the development and training of ESS techniques. Model selection is typically based on anatomical similarity to the human spine and surgical feasibility.

The choice of the porcine model is supported by multiple anatomical and practical advantages. Among the available options, the porcine model is the most commonly used, particularly due to its comparable vertebral architecture and intervertebral foramen morphology [6]. Anatomical studies have shown that the porcine lumbar spine exhibits vertebral dimensions, intervertebral-disc height, and spinal-canal structure that closely approximate human characteristics, providing a realistic training environment for endoscopic procedures.

Moreover, the porcine model offers practical advantages over other animal alternatives. Ovine and bovine models are also employed for selected spinal or cranial procedures, though with more restricted applicability [7]. The porcine model permits practice of both main endoscopic approaches—transforaminal and interlaminar—offering versatility in surgical training. Its availability, cost-effectiveness, and favorable ethical considerations make it the most practical option for structured training programs.

Despite known anatomical differences, the porcine model retains its validity for training. Although anatomical differences—such as reduced disc height and narrower interlaminar windows—have been reported in the porcine spine [8,9], these variations can be accommodated through technique adaptation, rendering the model highly reproducible and cost-effective. Previous anatomical studies have documented these discrepancies [10], yet there remains a need to standardize the porcine model specifically for monoportal endoscopic decompression and to assess its practical suitability for procedural training.

Therefore, the aim of this study was to validate the porcine lumbar spine as an experimental model for monoportal endoscopic decompression, assessing its anatomical fidelity, technical feasibility, and applicability for structured surgical training programs.

Thirteen pigs (6 male, 7 female) with a mean animal weight of 44.7 ± 5.3 kg (range: 33.8–61.7 kg) underwent lumbar endoscopic decompression using the monoportal technique, utilizing two approaches: transforaminal at L4-L5 and interlaminar at L5-L6. The actual body weight at the time of surgery ranged from 33.8 to 61.7 kg, as some animals gained weight during the acclimatization period, while others arrived slightly below the target range. Surgical data, including operative times, approach-specific details, complications, and necropsy findings, are summarized in Table 1.

Comprehensive surgical data revealed a mean total surgical time of 134.7 ± 32.3 minutes. The transforaminal approach (TF) required a mean approach time of 53.3 ± 18.5 minutes, with a mean coagulation time of 20.1 ± 33.1 seconds and a mean saline usage of 4.8 ± 3.6 Liters. The interlaminar approach (IL) was performed with a mean approach time of 51.5 ± 17.6 minutes, a mean coagulation time of 4.1 ± 10.7 seconds, and a mean saline usage of 5.5 ± 3.6 Liters. Post-mortem examination confirmed accurate vertebral level access in all instances. Nine animals (69.2%) were able to stand on all fours after waking from anesthesia.

There were no intraoperative monitoring incidents. Surgical complications were observed in 4 cases (30.8%).

Additionally, epidural hematomas were identified in three other animals at necropsy. These findings were clinically silent, as none of the animals exhibited postoperative neurological deficits.

Only three animals (23.1%) were euthanized prior to the study endpoint according to humane endpoint criteria (Cases 1, 4, and 13). In Case 4, the animal developed paraplegia without any intraoperative complication; necropsy did not reveal structural abnormalities. This case had a prolonged time on the surgical table due to anesthetic management, and the deficit was attributed to possible positioning-related ischemia.

A Spearman’s rank correlation was performed to evaluate the relationship between surgical case number (i.e., the chronological order of procedures) and total surgical time. The analysis revealed a statistically significant negative correlation (ρ = −0.73, p = 0.004), indicating that total surgical time decreased as the number of procedures performed increased. These findings demonstrate a clear learning curve effect, with increased experience leading to improved surgical efficiency (Fig 5).

This study demonstrates the feasibility and anatomical compatibility of using the porcine model for monoportal endoscopic lumbar spine surgery. In this context, the anatomical landmarks and spatial organization of the porcine lumbar spine proved sufficiently analogous to the human anatomy to support fundamental procedural steps such as transforaminal and interlaminar access. The porcine lumbar spine has been widely validated as a suitable model for endoscopic spine surgery training and research due to its anatomical similarities to the human lumbar region. Studies such as those by Amato et al. [6] and Gotfryd et al. [8] have shown that the overall vertebral architecture, neural decompression possibilities, and pedicular instrumentation techniques in pigs closely resemble those in humans. However, several anatomical peculiarities must be considered [9,10].

An additional anatomical factor relevant to surgical planning is the orientation and distribution of lumbar nerve roots. Unlike in humans, where lumbar roots follow a caudally oblique trajectory, in pigs the nerve roots emerge from the spinal cord in a more horizontal plane, nearly perpendicular to the cord. In addition, the porcine spine exhibits longer spinous processes, a thinner ligamentum flavum, and a narrower anteroposterior vertebral body diameter, which may necessitate adaptation of surgical instruments and techniques. One crucial consideration is the length of the spinal cord in pigs, which extends to the sacral region, and that they typically have six lumbar vertebrae. This anatomical characteristic increases the risk of spinal cord injury during lumbar procedures, highlighting the need for meticulous surgical technique and precise anatomical knowledge.

Surgeons training with this model must develop a precise understanding of these neuroanatomical differences to perform safe and reproducible procedures.

In the human L5-S1 interlaminar approach, the endoscopic sheath is used to retract and protect the dural sac and expose the intervertebral disc, a fundamental maneuver for decompression. However, this technique is not feasible in the porcine model due to the caudal extension of the spinal cord, which significantly increases the risk of spinal cord injury and paralysis. Therefore, for porcine lumbar endoscopic training, the transforaminal approach is safer and more anatomically comparable than the interlaminar approach.

Additionally, for interlaminar procedures in this model, we recommend performing them at higher lumbar levels to improve training safety.

Our experience confirms that these anatomical differences can present technical challenges during endoscopic lumbar procedures, particularly in animals weighing less than 40 kg. In our series, the only animal below this threshold (33.8 kg) required a more complex approach due to reduced access and anatomical constraints, resulting in delayed recovery and evidence of intramedullary hematoma on necropsy. This aligns with previous literature emphasizing the importance of specimen size in ensuring adequate working space and reproducibility of surgical maneuvers. Based on our findings, we recommend the use of pigs weighing more than 40 kg for endoscopic lumbar decompression studies, as larger animals provide vertebral dimensions and tissue planes more comparable to those encountered in human surgery, reducing the risk of technical complications and facilitating the learning process.

Interestingly, in one case of paraplegia that required early euthanasia based on humane endpoint criteria (case 4), necropsy did not reveal any structural cause for the neurological deficit. The authors hypothesize that this deficit may have resulted from the intraoperative positioning, specifically prolonged hip flexion during the surgical procedure. In this case, surgical table time was extended due to prolonged anesthetic induction and stabilization.

All procedures in our study were confirmed at the correct vertebral level by post-mortem examination, supporting the reliability of the porcine model for anatomical and technical validation. Nevertheless, the inherent anatomical differences—such as narrower disc spaces and lower disc heights—should be accounted for when extrapolating results to clinical practice or using the model for training purposes.

The 30.8% intraoperative complication rate and 23.1% early euthanasia incidence encountered reflect the authentic technical challenges of monoportal endoscopic spine surgery in living biological tissue. Spinal cord injuries (Cases 1, 13) resulted directly from the porcine spinal cord’s caudal extension into lumbar levels, while Case 4 paraplegia likely represented positioning-related ischemia. To establish this model as a safe training platform, we propose comprehensive technical safety optimization strategies applicable to both transforaminal and interlaminar approaches:

TECHNICAL TIPS FOR SAFETY OPTIMIZATION:

These evidence-based recommendations, derived directly from our complication analysis, substantially enhance procedural reproducibility and trainee safety across both endoscopic approaches.

One of the most relevant findings in our study was the significant learning curve observed, reflected by a steady reduction in operative time as the number of procedures increased. This trend, supported by statistical analysis, highlights the utility of the porcine model not only for anatomical simulation but also for progressive acquisition of technical skill. The model effectively reproduces the challenges of real surgical scenarios and allows iterative refinement of technique, making it especially valuable for surgeons at early stages of endoscopic training.

This study utilized 13 porcine specimens, demonstrating a statistically significant learning curve effect. However, this limited sample size precludes robust definition of the precise number of procedures required to achieve surgical competency or generalization of these findings to multi-surgeon training scenarios. The observed variability in operative times, particularly case 6 (90 minutes), likely attributable to a smaller and younger animal (36.3 kg) highlights the influence of individual biological factor on statistical interpretation. While this outlier steepens the observed correlation slope (ρ = −0.73), its exclusion maintains statistical significance (ρ = −0.62, p = 0.02), confirming the general trend but underscoring the need for larger cohorts to better characterize learning process heterogeneity. A prospective power analysis was not performed; future studies with n > 30 across multiple surgeons are necessary to establish definitive competency thresholds. Despite these limitations, the consistent downward trend validates the porcine model for progressive skill acquisition during early training phases.

Although objective assessments of motor function were systematically performed, this study is limited by its short postoperative follow-up period and the relatively small sample size, which may have restricted the detection of delayed-onset complications or long-term neurological sequelae.

Finally, while the porcine model does not fully replicate all aspects of human spinal anatomy, it remains a robust and reliable platform for developing and testing endoscopic decompression techniques. From a translational perspective, training in a porcine model bridges the gap between conceptual learning and clinical application. It provides surgeons with exposure to real biological tissues and hemodynamic responses, which are absent in most benchtop simulators. Therefore, this model not only enhances anatomical familiarity and technical confidence but also promotes the development of tactile and procedural competence that are essential for clinical translation.

While total operative time reduction effectively demonstrated technical proficiency acquisition across the 13-case series, this study represents an important methodological limitation by relying solely on chronometric assessment. More comprehensive skill evaluation requires Objective Performance Metrics (OPMs) including target localization accuracy (measured as mm deviation from planned entry point), quantitative bone removal assessment (volume of unnecessary resection), and critical structure visualization time (seconds to clear dural sac/nerve root identification). The porcine monoportal model’s distinct training value transcends anatomical similarity already documented in prior literature. Unlike cadaveric models lacking biological responses or benchtop simulators absent tactile complexity, this platform uniquely provides biological realism through real tissue handling and dynamic hemostasis challenges, procedural authenticity via endoscope maneuvering through living anatomical planes, and translational bridging as critical intermediate between conceptual simulation and clinical application. These characteristics position the porcine model as an essential platform for developing the composite competencies required for safe monoportal endoscopic spine surgery implementation.

Future studies should focus on establishing objective performance metrics and validating the correlation between training in this model and outcomes in human procedures. Additionally, anatomical mapping to further refine approach trajectories and reduce variability will improve standardization and educational reproducibility.

With appropriate anatomical knowledge and careful case selection, this model provides a high-fidelity, ethically justifiable alternative for surgical training and preclinical validation.

The porcine model has proven to be a valuable and anatomically viable platform for practicing monoportal endoscopic lumbar spine surgery. Despite minor anatomical differences compared to humans, it allows for safe and reproducible execution of both transforaminal and interlaminar approaches with satisfactory visualization and spatial orientation. As such, it represents an essential intermediary between artificial simulators and clinical practice, offering biological realism without compromising safety.