PDF
Citation
Print
INTRODUCTION
Traumatic spinal cord injury (TSCI) causes irreversible primary damage, making mitigation of secondary injury during the acute and sub-acute phases a central focus of both research and clinical care [1]. Existing literature suggests that spinal cord edema elevates the intraspinal pressure (ISP) beneath the rigid dura, potentially resulting in a self-perpetuating cycle of hypoperfusion and ischemia analogous to subdural compartment syndrome [2-5].
Current intensive care guidelines for cervical TSCI recommend maintaining mean arterial pressure between 85–90 mm Hg to ensure adequate spinal cord perfusion [6]. This mirrors the intensive care approach to traumatic brain injury (TBI) [7,8], in which sustaining cerebral perfusion pressure (CPP) in the face of elevated intracranial pressure (ICP) is the primary goal. Pioneering work from St. George’s Hospital, London, demonstrated a correlation between spinal cord perfusion pressure (SCPP) and neurological outcomes [9-14], fueling interest in interventions aimed at optimizing SCPP [15-17]. Notably, expansion duroplasty improves SCPP by reducing subdural ISP [18].
In our previous study on thoracic TSCI, we observed a statistically and clinically comparable elevation of subdural ISP in both TSCI and sham-operated (stand-alone laminectomy) pigs. This suggests that the TSCI itself did not cause a substantial elevation of the subdural ISP [19]. The main hypothesis of the present study was that ISP elevation due to TSCI was either missed or that there were alternative contributors to ISP elevation after TSCI and laminectomy. This study aimed to test this hypothesis by investigating whether (1) the ISP increases more within the spinal cord compartment than in the subdural compartment, (2) ISP elevation is due to cerebrospinal fluid (CSF) reconstitution, and (3) ISP elevation is driven by epidural pressure resulting from post-laminectomy soft tissue edema and/or the accumulation of blood components.
METHODS
Study overview
This study was conducted using an anesthetized porcine model of contusion TSCI. Three sub-studies were designed to test the three hypotheses concerning the mechanisms of ISP elevation. All experiments were terminated, and longitudinal ISP measurements were the sole outcome parameters. The detailed methodology of the model was described previously [20].
Experimental design
Anesthesia and intensive care protocol
Female Danish Landrace pigs (38–42 kg) were fasted for 6 hours prior to sedative premedication and transported to the surgical laboratory. After arrival, total intravenous anesthesia was induced using propofol, fentanyl, and midazolam. Prophylactic antibiotics were administered before surgery. Arterial blood pressure was monitored throughout the follow-up period. A standardized hourly checklist with correction protocols ensured that the animals remained within normal physiological parameters throughout the follow-up [20].
Surgical procedure
Following a midline skin incision and subperiosteal dissection of the paravertebral musculature, a full 2.5-level thoracic laminectomy, centered at the Th8 level, was performed. The width of the laminectomy was adjusted to accommodate the width of the injury-causing device’s falling rod.
Injury protocol
Spinal cord contusion was induced by releasing a 75-g rod from a height of 75 mm using a no-touch electromagnetic system. The rod remained in contact with the spinal cord for 5-minute after the impact to ensure sustained compression.
Follow-up time
The follow-up duration was reduced compared to our previous 72-hour protocol: sub-study 1 included 15 hours, whereas sub-studies 2 and 3 were limited to 10 hours. This adjustment was based on earlier experiments showing that the subdural ISP generally reached a maximum plateau around the 10 hours mark, before dropping again around 16 hours post-injury, regardless of injury severity [19]. No changes that would expectedly delay or increase the immune response responsible for edema of the spinal cord and epidural tissue or delay or accelerate the replenishment of CSF were observed in the model. Hence, it is likely that the subdural ISP reaches a maximum plateau after 10 hours. Thus, sub-studies 2 and 3, which measured only the subdural ISP, were limited to 10 hours of follow-up. Sub-study 1 measured the spinal cord compartment pressure and epidural pressure, which had not been previously measured for longer periods of time. To detect any temporal mismatch between compartment pressures, the follow-up time was extended to 15 hours.
Sub-study 1
To measure the compartmental differences in the ISP, pigs (n=5) were instrumented with pressure probes in the epidural, subdural, and spinal cord compartments before TSCI induction. All three pressures were recorded continuously over a 15-hour follow-up period. A small opening in the dura allowed the placement of the subdural probe. Subsequently, a targeted myelotomy was performed using a cannula to create a tract for the insertion of the spinal cord compartment probe. Ultrasonography confirmed the correct placement and cranio-caudal alignment of both probes at the planned site of injury (Supplementary Fig. 1). The probes were then retracted by 5 mm, TSCI was induced, and the probes were reinserted and secured using sutures. A third probe was placed in the epidural compartment, appropriately aligned, and secured with sutures. The surgical wound was closed in layers (muscle, fascia, and skin) to prevent CSF leakage. Finally, the external parts of the probes were secured to the skin using an adhesive tape to minimize the risk of displacement during the follow-up period. An anatomical overview of the laminectomy, probe placement in different compartments, and injury site is shown in Fig. 1.
Sub-study 2
To evaluate whether the observed increase in the subdural ISP was due to CSF reconstitution, a CSF drain was placed adjacent to the subdural probe at the trauma site (n=3). Following TSCI induction and surgical closure, the drain was opened and aspirated at 4.5-hour intervals during a 10-hour follow-up period, aiming to restore pressure to the post-injury baseline. Each aspiration was followed by a 30-min drainage period with the system open to atmospheric pressure (0 mm Hg). To evaluate potential morphological changes in the subdural, spinal cord, and epidural compartments, magnetic resonance imaging (MRI) was performed before euthanasia. A prone T2-weighted sequence was acquired using a 3.0 Tesla MRI system (GE Medical) with a spatial resolution of 0.39×0.39×4.4 mm).
Sub-study 3
To evaluate whether epidural pressure resulting from soft tissue edema and collected blood components contributed to elevated subdural ISP, the surgical wound was intentionally left open during the 10-hour follow-up period in both sham-operated (n=3) and TSCI pigs (n=3). All pigs were instrumented with subdural pressure probes at the trauma site, as described in sub-study 1. The wounds were covered with sterile surgical drapes to minimize fluid loss by evaporation. As part of the hourly intensive care protocol, the wounds were inspected for blood accumulation and regularly cleaned using low-pressure suction.
Data collection
The spinal cord compartment ISP (sub-study 1) and subdural ISP (all sub-studies) were measured using identical systems of Pressio ICP monitoring probes (Sophysa) connected to Pressio ICP monitors. Epidural ISP (sub-study 1) was measured using probes and a monitor from Raumedic (Helmbrecht). Pressure signals were transmitted to MP70 Intellivue monitors (Philips) and then relayed via the RJ45-to-USB interface to a laptop for acquisition using the ICM+ software (Cambridge University) [21]. All probes were zeroed prior to insertion in accordance with the manufacturer’s instructions.
Analysis
Data management
ISP data from the first 30 minutes were excluded for all animals to avoid artificially high values produced by the initial 5-minute of spinal cord compression, placement of probes and/or CSF drain, adjustments, and surgical wound suture. All analyses were performed in STATA 18 (StataCorp.).
MRI analysis
Unblinded analyses of the MRI scans were conducted in Image J (version 8, 2024) [22] by the principal and secondary investigators. Regions of interest (ROI) were manually drawn around the cord for every MRI slice of the injured segment for each estimated spinal cord area. Five sequential ROI were drawn around the unaffected spinal cord, both cranially and caudally, after laminectomy.
Statistical modeling and presentation
Sub-study 1
The raw data for each animal were plotted and presented as line plots. Mixed model data were computed with spinal cord compartment pressure as the dependent variable, and compartments and time as explanatory variables. Diagnostic plots of the residuals and fitted values were computed (Supplementary Fig. 2). The mixed model was presented as a margin plot of 95% CI of the predicted mean pressure at different follow-up time intervals. The global intercept, main effects of explanatory variables, and interaction estimates with p-values and 95% CIs are provided in Table 1.
Sub-study 2
The mean areas of the unaffected and injured ROIs were computed and presented with 95% CIs. A paired samples t-test between the mean areas was computed and presented with 95% CI.
Sub-study 3
A second mixed model was constructed with the subdural ISP as the dependent variable, and time, injury condition (TSCI versus sham operation), and wound condition (open versus closed) as the explanatory variables. The model also included data from sub-study 2 and five sham-operated animals and five TSCI animals (same injury protocol as in the present study) from our previous study [19]. Before the acceptance of the model, diagnostic plots of the residuals and fitted values were computed (Supplementary Fig. 3). The mixed model was presented as a margin plot (95% CIs) of the predicted mean pressure at different follow-up time intervals. The global intercept, main effects of the explanatory variables, and interaction estimates with P-values and 95% CIs are provided in Table 2.
RESULTS
Sub-study 1
Raw ISP
All five animals exhibited a net increase in the subdural and spinal cord ISP from baseline. Epidural ISP also increased in four out of five animals (Fig. 2). The pressure trajectories across the three compartments were nearly identical for each animal, with the exception of Animal 3, where the epidural ISP diverged. In Animal 2, the signal noise from the subdural probe required surgical reopening and replacement of the probe, leading to the exclusion of the first 4.5 hours of data. Notably, reopening resulted in a steep pressure drop across all compartments (pressures at time 4.5 hours were markedly lower than those in the other animals at the same time) (Fig. 2). A sixth animal was excluded because of complete probe malfunction and because no replacement probe was available at that time. Transient ISP spikes were observed in association with routine care procedures. In animals 1 and 4, sudden minor drops in pressure were observed across all compartments.
Mixed model
We constructed a mixed model (Fig. 3) with the spinal cord compartment ISP as the dependent variable. A global intercept of 3.64 mm Hg (Table 1) corresponded to the predicted mean spinal cord compartment ISP during the first 1.5 hours. At baseline, subdural and epidural ISP were 1.20 mm Hg (P=0.2) and ISP 5.79 mm Hg (P<0.0001) lower, respectively, than the spinal cord ISP (Table 1). Overall, a significant main effect of time confirmed that the spinal cord ISP increased over time (Table 1). The time-by-compartment interaction was low and not statistically significant (Table 1). The spinal cord and subdural compartment ISP were statistically similar and followed similar trajectories throughout the 15-hour period (Fig. 3). Although significantly lower, epidural ISP followed a similar trajectory.
Sub-study 2
CSF-drain
In all three animals, subdural ISP increased over the 10-hour follow-up period relative to baseline. The CSF cannot be drained passively or by aspiration. Neither attempted aspiration nor passive opening of the drain affected subdural ISP.
MRI
Magnetic resonance imaging revealed soft tissue edema and minor hemorrhage within the epidural compartment, accompanied by spinal cord swelling and compression. Air pockets were also observed (Fig. 4). The mean cross-sectional area of the injured spinal cord was 11.2 mm2 (95% CI, 8.4–13.9]), compared to 14.4 mm2 (95% CI, 11.0–17.7]) in the un-injured cord, yielding a mean difference of –3.2 mm2 (95% CI, –6.3 to –0.10). This corresponds to a 28.7% reduction in the mean cross-sectional area associated with the TSCI. CSF was displaced around the injured cord (Fig. 4).
Sub-study 3
Raw ISP
All animals in sub-studies 2 and 3 and those pooled from our previous study [19] exhibited a net increase in subdural ISP from baseline to the end of the follow-up period (Fig. 5). Sudden drops and transient spikes in the ISP were observed, similar to sub-study 1. The spikes coincided with intensive care procedures such as lung recruitment and airway suctioning.
Mixed model
A mixed model (Fig. 6) was constructed to estimate the mean subdural ISP, incorporating the injury condition (TSCI vs. sham) and wound condition (open vs. closed) as fixed effects. A significant main effect of time showed that the subdural ISP increased over time (Table 2). Statistically significant TSCI-by-time interactions were observed in the final 3 hours, demonstrating that TSCI led to an additional increase in subdural ISP of 1.9–2.2 mm Hg compared to that in sham-operated animals (note the separation of the open and closed wound groups during the final hours of follow-up) (Table 2, Fig. 6). In contrast, leaving the wound open led to a significantly greater reduction in subdural ISP (from 2.2 to 5.5 mm Hg) compared to the closed-wound condition. The effect size of the open-wound condition increased over time (Table 2).
DISCUSSION
In sub-study 1, the ISP increased progressively across all compartments after laminectomy and TSCI. No significant difference was observed between subdural and spinal cord ISP. However, epidural ISP remained significantly lower throughout the observation period, with a nearly constant margin. These findings clearly argue against the first hypothesis that ISP increases more within the spinal cord than in the subdural compartment. As such, the results of sub-study 1 support the use of subdural ISP as a reliable proxy for spinal cord pressure in experimental and clinical contexts.
In sub-study 2, MRI analysis showed that the CSF was displaced around the injury site, most probably explaining why drainage attempts were unsuccessful and did not affect the subdural ISP. These findings strongly indicate that CSF restoration does not account for elevated ISP levels, which is the second hypothesis.
MRI analysis also showed that laminectomy combined with TSCI resulted mostly in spinal cord compression from epidural soft tissue swelling and minor hematoma, rather than expansive spinal cord swelling. This suggests that the epidural ISP is larger than the subdural ISP. Correspondingly, sub-study 3 showed that the subdural ISP increased comparably over time in both sham-operated and TSCI animals. The ISP was significantly higher in TSCI animals than in sham-operated animals only during the final 3 hours. However, by leaving the wound open, the subdural ISP was reduced by 2.5-fold magnitude over time. Therefore, the progressive elevation of subdural ISP observed in closed-wound animals most likely reflects the rising epidural pressure.
In summary, converging evidence from sub-studies 2 and 3 strongly suggests that ISP elevation is driven by epidural pressure resulting from post-laminectomy soft tissue edema and/or the accumulation of blood components, supporting the third hypothesis. In partial contradiction, sub-study 1, which provided a direct measure of epidural pressure, was found to be significantly lower than subdural and spinal cord compartment pressures. However, the parallel trajectories of pressure in the spinal cord and the subdural and epidural compartments suggest biomechanical interdependence. Additionally, there was a steep drop in all compartment pressures when Animal 2 in sub-study 1 was re-opened for probe replacement. As expected, the pressure originating below the dura precludes these observations.
These observations suggest that although the epidural pressure measurements in sub-study 1 appeared technically stable, they may have been systematically inaccurate, i.e., precise but not physiologically representative of the epidural pressure above the injury site. One plausible explanation is that following wound closure, the large and heterogeneous epidural space may have allowed the probe to migrate into localized regions of lower pressure. This is supported by the MRI findings from sub-study 2, which demonstrated air pockets, fluid accumulation, and minor hemorrhage within the epidural compartment. A key limitation of sub-study 1 was that the epidural probe was not securely affixed to the dura, which may have contributed to probe migration and/or inconsistent positioning. Furthermore, sub-study 3 did not include concurrent epidural pressure measurements, limiting our ability to determine the effect of the open wound directly on epidural pressure. In general, it should be noted that the sample size of each sub-study (n=3–6) was small, and the statistical power was limited.
One clinical study involving both TSCI patients and controls reported a mean subdural ISP of 22.5 mm Hg (standard deviation [SD], 5.1 mm Hg; n=10), significantly higher than the mean CSF pressure of 7.8 mm Hg (SD, 3.3 mm Hg; n=12) in healthy individuals evaluated for normal pressure hydrocephalus (the control group) [23]. Interestingly, epidural ISP was measured in two TSCI patients and found to be significantly lower than the subdural ISP, consistent with our findings. However, although we observed a progressive increase in epidural ISP over time, the measurements remained constant [10]. They also reported that the supine position led to increased ISP [24], supporting epidural pressure transmission to the subdural compartment following laminectomy. The ISP spikes observed in conjunction with the lung recruitment maneuvers in the present study could thus be explained by thoracic expansion, which increased the epidural pressure on the cord.
In our previous sham-controlled animal trial, directly comparing sham-operated controls to TSCI, we did not observe values as elevated as those in human TSCI patients [19]. Similarly, the ISP values measured in the present study measured ISP values considerably lower than those reported in clinical studies. However, it should be noted that not all patients seem to develop elevated ISP [10,18,23].
Our previous studies have demonstrated visible injury and glial activation [19,20], indicating an inflammatory response. We assumed that the identical experimental conditions in the present study elicited a similar response. However, the lack of histological verification of TSCI is a limitation. Additionally, visual inspection confirmed a comparable and significant spinal cord contusion.
Animals were sedated throughout the study period. Anesthesia is used therapeutically in traumatic brain injury to reduce metabolism and intracranial pressure [7]. A previous human study found no correlation between anesthesia and ISP levels [18], and our earlier work also reported no association between infusion levels and ISP [19] in porcines. Given that ISP is measured directly at the injury site where the damaged tissue swells against the dura, it has been suggested that ISP is largely unresponsive to such interventions [10].
In human trials, all TSCI cases involved cervical injuries [10], whereas the present porcine model involved thoracic injuries. The cervical spine has a larger cross-sectional area, potentially allowing swelling to fill the subdural compartment more readily, thereby elevating the ISP to higher levels. Although the possibility of a different ratio between subdural pressure and external forces highlighted in our study cannot be excluded, the MRI from sub-study 2 demonstrated the absence of CSF around the injured cord, suggesting near-complete obliteration of the subdural space. This implies that swelling was sufficient to eliminate the CSF buffer, even in cases of thoracic injury.
Anatomical differences alone may not fully explain ISP discrepancy. Human studies have predominantly involved complete injuries [10], which are expected to provoke more severe inflammation and edema within the injured segment than incomplete or sham conditions. Injury completeness may correlate with the expansive force of subdural swelling and, consequently, with the ISP magnitude. In our previous porcine study, the ISP did not differ significantly between near-complete injuries (average injured-to-normal spinal cord ratio of 93%) and mild or sham controls [19]. Although injury completeness and spinal level may influence the ISP, further comparative studies are required to disentangle their respective contributions.
The temporal profile in our study differed markedly from that of clinical TSCI. In our design, ISP monitoring commenced immediately after injury, whereas in human trauma studies, delays occur owing to transport, admission, diagnostics, and hospital logistics. Consequently, ISP monitoring in patients often begins multiple hours after injury, allowing secondary injuries to develop over time. For example, one clinical study initiated ISP monitoring within 72 hours and continued for up to 7 days [10]. This suggests that patients exhibiting an initially or sustainably high ISP may have reached such levels after 15 hours, which was the maximum follow-up duration in the present study. Even in our previous 72-hour study, ISP values remained lower than those typically reported in human cohorts. These observations reinforce the importance of long-term follow-up when modeling ISP dynamics and evaluating therapeutic interventions in animal models. However, compared with other porcine TSCI trials, the ISP measured up to 120 hours post-injury was similar to or lower than that in our studies [25-28].
These arguments bridge the gap in the applicability of real-life human TSCI. However, the inherent artificial aspects of the model impose general limitations on its applicability. In addition to the above-discussed aspects of anesthesia, injury location, injury completeness, and temporal aspects, the model is limited by the animal model and artificial injury. Injury differs from real-life human TSCI in several ways. First, the injury was induced to an exposed cord following laminectomy. Second, the injury was blunt and there were no sharp bone fragments to compress the cord. Third, although the blunt device was left behind after the impact to mimic compression from the bone, the 5-minute period was very short. Fourth, the injury was uni-local, which, in comparison with multi-local injuries in real life, may elicit a smaller immune response, with implications for ISP dynamics.
The development of elevated ISP following TSCI is likely to be a multifactorial process influenced by intrinsic factors, such as trauma type and energy, spine and spinal cord anatomy, immune profile, and inflammatory response, as well as extrinsic factors, including patient positioning and external pressure, after laminectomy. Accurate ISP measurements also critically depend on correct probe placement [29]. This complexity may partly explain the variability in ISP levels observed within and between clinical and porcine studies. Our findings suggest that epidural pressure was a major contributor to the observed increase in the incidence of ISP.
Our findings demonstrate that subdural ISP closely approximates the spinal cord compartment pressure, validating its use as a surrogate in both experimental and clinical settings. The hypothesis that ISP increases more within the spinal cord compartment than in the subdural compartment was not supported. CSF restoration does not appear to contribute to ISP elevation. Instead, epidural pressure following laminectomy emerged as a driver of elevated ISP in both TSCI and sham groups. In the present study, the effect was 2.5-fold greater than that of TSCI alone. Although direct clinical translation is broadly constrained by model limitations, our findings underscore the importance of further investigations of post-laminectomy epidural tissue and pressure dynamics in the pathophysiology of elevated ISP. Caution should be exercised against direct pressure on the area of laminectomy in patients with postoperative TSCI.
XML Download