Neurophysiological and Spinal Cord Perfusion Changes at Dynamic Neck Positions in a Compressive Spinal Cord Injury Rat Model


 Background: Cervical spondylotic myelopathy (CSM) is a degenerative condition of the spine that caused by static and dynamic compression of the spinal cord. However, the pathophysiological changes at dynamic neck positions remain poor. This study investigated the interplay between neurophysiological and haemodynamic responses at dynamic neck positions in a chronic compressive spinal cord injury (CCSCI) rat model. Methods: Behavioural tests including Basso, Beattie, and Bresnahan scores and an inclined plane test were used to evaluate the motor function recovery. Combined examination of dynamic motor and somatosensory evoked potentials (DMEPs and DSSEPs, respectively) was performed regularly to evaluate the dynamic motor and sensory conduction of the cervical cord. At 4 weeks post-injury (wpi), dynamic magnetic resonance imaging (MRI) and dynamic laser Doppler flowmetry (LDF) were used to demonstrate the interstructure and spinal cord blood flow (SCBF) at the compression site at dynamic neck positions. Hematoxylin and eosin (HE) staining was performed to assess the cords' pathological changes.Results: Behavioural tests and combined DMEPs and DSSEPs examination showed that spinal cord neurological function and dynamic neural conduction deteriorated gradually within a 4-week compression period. The DMEPs were mainly deteriorated upon flexion, while DSSEPs were upon all neck positions after the compression. At 4 wpi, dynamic MRI showed increased T2-weighted image (T2WI) signal intensities. Also, dynamic LDF demonstrated decreased SCBF at the spinal cord compression site. Both of them altered especially upon cervical flexion. The dynamic change in SCBF was significantly correlated with the change in DMEP amplitude upon flexion. Conclusions: This exploratory study revealed that changes in axonal conduction in the motor and somatosensory tracts of the spinal cord were significantly related to chronic compression time and neck position. Furthermore, spinal cord ischaemia may be intimately related to motor conduction dysfunction upon flexion in CCSCI models. These results indicated the potential for therapies targeting dynamic spinal cord perfusion to prevent progression and functional loss in CSM.


Introduction
Cervical spondylotic myelopathy (CSM) is a degenerative condition of the spine that leads to static and dynamic compression of the spinal cord. [1] Dynamic compression of the spinal cord is recognized as an important pathogenic factor for CSM. [2] It has been proposed that dynamic injury may occur through instability, [3] an increase in the range of motion, [4] and minor trauma in the setting of pre-existing degenerative cervical myelopathy. [5] Numerous clinical studies have reported the impacts of dynamic neck positions on CSM. Morphologically, Muhle et al.[6] used the dynamic magnetic resonance imaging (dynamic MRI) to demonstrate that the prevalence of spinal stenosis increased with exion and extension compared with the prevalence in a neutral position. Further study reported extension shortens the cervical cord and decreases the space available for it with ligamentum avum buckling [7], which could potentially aggravate posterior compression, causing pincer effects. [8,9] On the other hand, although the spinal canal diameter may increase in exion, some studies have reported that in this position, the tension forces applied to the spinal cord cause ventral spinal cord compression against osteophytes and discs, worsening eventual ventral compression. [8,9] Neurophysiologically, CSM patients' dynamic somatosensory evoked potentials (DSSEPs) have been shown to deteriorate signi cantly upon extension and exion. [10,11] We further revealed that the percent changes in DSSEP amplitude at dynamic neck positions were related to preoperative radiographic characteristics, such as the presence of cervical segmental instability, compression degrees, and patterns of intramedullary T2WI hyperintensity. [12] Pathophysiological changes in the spinal cord at dynamic positions were demonstrated by cadaveric studies, in which the lateral columns and the anterior horns were deformed by mechanical stress produced by spondylotic bars during exion. [8] Notably, the tissue of the spinal cord is highly vascularized and extremely sensitive to hypoxia induced by static or dynamic compression. The exion position causes anterior compression of the cord, where the anterior spinal artery (ASA) traverses. Thus, it has also been suggested that ischaemia of the spinal cord as a result of neck exion can cause neurological deteriorations and cervical myelopathies. [13][14][15] However, due to the lack of appropriate animal studies, the mechanisms behind the changes in neurological functions at dynamic neck positions are still not clear.
In recent years, numerous attempts have been made to establish a CSM in cases of chronic compressive spinal cord injury (CCSCI) model in different animals, employing various methods, such as tumour induction [16], the use of penetrating hydrogels [17] or urethane polymers [18], the use of spinal hyperostotic mice (twy/twy) [19], and plastic screw implantation. [20] Among these approaches, hydrogels and polymers are expected to be applicable materials that lead to chronically progressive injury.
Previously, we successfully established a CCSCI rat model using a water-absorbable polyurethane polymer sheet [21], which provided the foundation for this study. We have found that various changes, including decreased microvascular density, [22,23] varying degrees of ischaemia-hypoxia, [24,25] BSCB disruption, [25,26] and neuron loss, [27] occurred at different time points after static chronic spinal cord compression. Nevertheless, neurophysiological changes and their relationship with perfusion alterations at dynamic neck positions were remained largely unknown. In this study, we used CCSCI rat models to study the changes in dynamic MRI, dynamic motor evoked potentials (DMEPs), DSSEPs, and spinal cord blood ow (SCBF) at different neck positions and further analysed the correlations between them.

Study design
All experimental procedures were approved by the Research Ethics Committee of Sun Yat-sen University, Guangzhou, China and conformed to all relevant regulatory standards. A total of 46 male adult Sprague-Dawley (SD) rats (250-400 g) were randomly allocated to the sham (n = 16) and CCSCI (n = 30) groups.
The animals in the CCSCI group underwent implantation of a water-absorbing polymer sheet into the cervical spinal canal, which expanded over time to induce chronic compression of the cord. Neurological functions were evaluated by behavioural tests from 1 day post-injury (dpi) to 4 weeks post-injury (wpi), and DMEPs and DSSEPs tests from 1 to 4 wpi. At 4 wpi, the structural changes of the cord at various neck positions were evaluated by dynamic MRI scans. The perfusion status of the cord at dynamic positions was assessed by laser Doppler owmetry (LDF). The ultrastructure of the cord was evaluated by routine histology. (Fig. 1A) Induction of spinal cord chronic compression Each rat in the sham and CCSCI groups was anaesthetized with 10% chloral hydrate (300 mg/kg) (Guangzhou FISCLAB Environ. Sci-Tech. Co., Ltd., Guangzhou, China). Following exposure of the spinal process and laminae of C4-C6 from the posterior, the ligamentum avum and C5 lamina were removed to access the epidural space. In the CCSCI group, the polymer (1→4)-3,6-anhydro-a-l-galactopyranosyl-(1→3)-β-D-galactopyranan) sheet (1 mm × 3 mm × 1 mm) was implanted into the C6 epidural space on the dorsal part of the spinal cord. Spinal cord compression was achieved by expansion of the polymer caused by liquid absorption. [22,28] This polymer sheet can absorb liquid in the spinal canal to expand its volume sevenfold (approximately 2.3 mm × 4.2 mm × 2.2 mm). In the sham group, the C5 lamina was removed without insertion of the polymer sheet. Following surgery, the incision was closed in layers with complete haemostasis. To prevent dehydration, animals received a subcutaneous (s.c.) injection of lactated Ringer's solution (200 µL) immediately after surgery. All rats were administered an intramuscular injection of penicillin G (80 U/g) during surgery to prevent infection, and carprofen (4-5 mg/kg, Rimadyl, P zer) was injected subcutaneously 2 days post-surgery for further pain relief as needed. All surgeries were performed by the same experienced investigator.

Neurological Function Evaluation
To evaluate motor function recovery after CCSCI, the Basso, Beattie, and Bresnahan (BBB) locomotor scale [29] and an inclined plane test (IPT) [30] were conducted at 1 dpi and 1, 2, 3 and 4 wpi. BBB scores ranged between 0 and 21; a score of 0 re ected complete paralysis and a score of 21 indicated normal locomotion. Lower scores (0-7) denoted isolated joint movements with little or no hindlimb movement; intermediate scores (8-13) indicated intervals of uncoordinated stepping; and higher scores (14)(15)(16)(17)(18)(19)(20)(21) signi ed forelimb and hindlimb coordination. For the IPT, rats were placed horizontally on a smooth, tilted board. The board was initially placed in a horizontal orientation (0°), and the angle of the board was increased by 5°-10° after each attempt. The maximum angle at which the rats remained on the board for 10 seconds was recorded. The evaluation was conducted by two investigators blinded to the group assignments.

Histological evaluations
In brief, rats in different groups were euthanized with an overdose of intravenous sodium pentobarbital and transcardially perfused with 0.9% saline followed by 4% paraformaldehyde in 0.1 M phosphate buffer (PFA) at the endpoint of the study. The C5-C7 spinal cord was harvested, xed overnight with 4% formaldehyde in phosphate-buffered solution at 4°C, and embedded in para n. A series of 25-µm thick spinal cord sections were used for immunohistochemical staining. Slides were placed on a plastic rack and vessel for microwave epitope retrieval.

Electrophysiological evaluations
The functional integrity of the spinal cord among the model rats at dynamic positions was evaluated by DMEPs and DSSEPs at 1 to 4 wpi. An electrophysiological monitoring system (Nicolet Endeavor CR) was used to elicit and record transcranial electrical stimulation motor evoked potentials (TES-MEPs) and SSEPs. The animals were evaluated under general anaesthesia with 10% chloral hydrate (300 mg/kg) (Guangzhou FISCLAB Environ. Sci-Tech. Co., Ltd., Guangzhou, China) intraperitoneally. Their scalps and the posterior portion of their necks were shaved and aseptically treated using iodine. Two electrodes were placed 2 mm posterior to bregma and 3 mm to the left and right of the longitudinal midline; these locations correspond to the left or right primary motor cortex (C3 and C4), respectively, and served as DMEP stimulation electrodes and a reference electrode alternatively. Another two electrodes were placed at the midline of the skull 2 mm anterior to bregma (Fz) and at the midpoint of the ears (Cz) for DSSEP recording. Two electrodes were placed around each side of the sciatic nerve along the course of the biceps femoris muscle for DSSEP stimulation. Two were placed at each side of the extensor digitorum communis in the forelimb and tibialis anterior in the hindlimb for DMEP recording. A ground electrode was placed on the back of the subject subcutaneously. Rats were positioned neutrally at approximately 40° extension and then at approximately 40° exion of the cervical spine using an angle-adjustable stereotaxic frame for dynamic MEP and SSEP measurements. Constant current stimulation at the skull was used for the generation of MEPs. Single-trial MEPs were obtained with a current intensity of up to 16 mA and a pulse width of 50 ls at a frequency of 350 Hz for a 1-minute duration. MEPs were recorded using subdermal needle electrode pairs from the extensor digitorum communis in the forelimb and tibialis anterior in the hindlimb.
Constant current stimulation around sciatic nerves with a magnitude of 6 mA, duration of 0.02 ms, and frequency of 3.43 Hz was used to elicit DSSEPs. Cortical SSEPs were recorded from the skull at Cz-Fz. We averaged 256 SSEP trials to improve the signal-to-noise ratio. SSEP signals were ltered using a bandpass lter of 10 Hz to 250 Hz. A sensitivity of 20 lV/div and a time base of 5 ms/div were used to display the SSEP responses.
Onset latency and peak-to-peak amplitude of the responses at dynamic positions were measured at the endpoint of the study. The onset latency was measured from the delivery of the stimulus to the rst positive or negative de ection (N1 in Fig. 4) from baseline. Peak-to-peak amplitude was de ned as the maximum amplitude between the largest positive and negative peak (N1-P2 in Fig. 4). Each MEP/SSEP test was repeated 3 times, and their average value was taken. We de ned an immeasurable SSEP/MEP as a waveform that could not be identi ed by averaging over 500 sweeps. SSEP and MEP responses recorded from each limb were classi ed separately. For the nal analysis, the lowest SSEP or MEP amplitude among the four limbs was utilized as the de nitive data point.

MRI evaluations
Cord compression was evaluated by MRI using a 3.0-T MR imager (Siemens Trio). The animals were anaesthetized by iso urane inhalation, and a surface coil was placed over the animals' cervical spine region to acquire anatomical T2-weighted images (T2WIs). T2WIs were acquired with the following parameters: echo time [31]/repetition time (TR) = 35/2500 ms (T2W) and 115/2500 ms (PDW), slice thickness = 1 mm, interslice distance = 1.1 mm, and number of excitations (NEX) = 4. A total of 15 axial slices covering C3-C7 of the cervical spinal cord were acquired at each disc and body level. The image slice planning was the same as that in anatomical axial images, with 15 slices covering the cervical spinal cord from C3 to C7.
The sagittal diameters of the spinal canals and spinal cords and the transverse diameters of the cords and canals were measured with Osirx (Pixmeo, Geneva, Switzerland), which is a standard software for the MR equipment packages. MR signal intensity was analysed in regions of interest (ROIs) at the compression site (or C6 levels in the sham group) of the spinal cord with Osirx. We used the T2WI intensity signals of paraspinal muscles as calibration references in each MR image and quanti ed the relative signal intensities of the ROI in cervical grey matter in both sham and CCSCI rats. The relative signal intensity was calculated by dividing the signal intensity of the ROI in cervical grey matter by that in paraspinal muscles.

Dynamic LDF measurements
We measured the SCBF and oxygen saturation (SO 2 ) of the rats at 10 minutes and 4 weeks after implanting the compression material in the CCSCI group or exposing the dorsal dura mater in the sham group. For SCBF and SO 2 measurements, an OxyFlo (ADInstruments Pty Ltd, Castle Hill, Australia) LDF probe was attached to the stereotaxic frame and positioned in contact with the dorsal dura mater to the right side of the central vein at the C3 and C6 segmental levels of the spinal cord. Upon neutral positioning, after adjusting the SCBF at C3 to the same position (580 PU) for both the sham and CCSCI models, we monitored the SCBF at the C6 (compression) level in the sham and CCSCI models. The rats were then positioned upon maximum cervical extension and cervical exion postures for at least 5 minutes before undergoing the same recording process. Each recording process started with an adjustment of the SCBF data at the C3 level, and the recording time at C6 or the injury level lasted for at least 1 minute. Output signals were recorded in Powerlab (ADInstruments Pty Ltd, Castle Hill, Australia) continuously throughout the experiment and averaged every 3 seconds.

Statistical analysis
Comparisons of the ultrastructure of the spinal cord on the injured and non-injured sides were performed using paired t-tests. Comparisons of the neural conduction situations of spinal cords among the animals categorized by DMEP and DSSEP responses and spinal cord perfusion status demonstrated by LDF at different neck positions were performed using one-way analysis of variance (ANOVA) and post hoc tests. The level of signi cance was set at p < 0.05. All data analyses were performed using SPSS 15.0 analysis software (SPSS Inc., Chicago, IL, USA).

Results
Neurological dysfunction after the induction of CCSCI Following incomplete CCSCI, progressive neurological deterioration occurred in a sizable number of patients; similar results have also been observed in animal models. [32] Here, we used the BBB score and IPT outcome to evaluate the changes in neurological function in a rat model of CCSCI. In our study, from 1 dpi to 4 wpi, the BBB scores and IPT angles of the CCSCI rat models gradually decreased and were signi cantly lower than those of the sham models after one and two wpi. (Fig. 1B, C)

Structural and cellular characteristics in sham and CCSCI spinal cords
Haematoxylin and eosin (HE) staining revealed histological changes at the injury epicentre under different levels of magni cation (Fig. 2). The spinal cord was intact in the sham group. Polygonal Nissl bodies inside neurons in the grey matter anterior horn were large and dense. In a transverse section of a CCSCI model, the compressed spinal cord exhibited structural damage, including fragmentation of neuronal nuclei, pyknosis, neuropil damage, degradation of the extracellular matrix, interstitial oedema, cytoplasmic reduction, and cavity formation. The dorsal funiculi suffered the most severe breakage. Neuronal cell bodies in the anterior horn were round and small. Nissl body chromatolysis was evident inside the neurons. The insults to the cord circulation in the CCSCI models included central canal enlargement and intratissue bleeding in the grey matter. The longitudinal neurological conduction pathways of the CCSCI models were severely disrupted, as neuronal loss in the grey and axon breakage in the white matter were observed in sagittal sections.

Dynamic motor de cits in CCSCI models evaluated by DMEPs
DMEPs were tested at 1, 2, 3 and 4 wpi, following the protocol described in the Methods section. The number of rats with abolished waves and the latency and amplitude data of the remaining rats are summarized in Table 1. The latency and amplitude of DMEP peaks in the sham group at all three positions remained unchanged at all testing time points. After chronic compression injury, the MEP response was abolished in one rat at all neck positions. At the rst 2 wpi, the average DMEP latencies and amplitudes of the CCSCI models were slightly affected upon neutrality and extension but were signi cantly diminished upon exion (t-test, p < 0.001) compared with the average DMEP latencies and amplitudes of the sham models. At 4 wpi, the DMEP latencies and amplitudes of the CCSCI rats were signi cantly diminished at all neck positions compared with the average DMEP latencies and amplitudes of the sham rats. It should be noted that the CCSCI models had the worst DMEP latencies and amplitudes (ANOVA post hoc test, p < 0.001) and the most abolished DMEP waves (Chi-square, p < 0.05) upon exion compared with the values at the other two positions at all time points after compression. (Fig. 3)

Somatosensory de cits in CCSCI models evaluated by DSSEPs
After a few minutes of testing the DMEPs, DSSEPs of rat all models were tested. The number of rats with abolished waves and the latency and amplitude data are summarized in Table 2  Dynamic MRI showed intramedullary T2WI hyperintensity of the spinal cord upon exion To explore the inner structural changes in CCSCI rat models at dynamic neck positions, we performed dynamic MRI on the cervical cord region of our models at 4 wpi. Compared with the sham model, compression was clearly evident at the posterior side of the spinal cord at the C6 region of CCSCI rat models (Fig. 5a-h) signi cantly from neutral to extension positions, both of these measures decreased signi cantly upon exion in the CCSCI models (t-test, p < 0.001). (Fig. 6a-e) We also calculated the relative change in SCBF (R-SCBF) and oxygen saturation (R-SO 2 %) upon extension or exion by dividing each model's SCBF and SO 2 % at extension or exion by that at the neutral position. The R-SCBF at 4 wpi of CCSCI rat models were 0.98 ± 0.14 and 0.73 ± 0.11 upon extension and exion respectively. The R-SO 2 % at 4 wpi of CCSCI rat models were 0.96 ± 0.15 and 0.56 ± 0.09 upon extension and exion respectively. Both the R-SCBF and R-SO 2 % at exion were signi cantly lower than that at extension (t-test, p < 0.001) for the CCSCI model at DSSEPs were all severely reduced and were identical at all neck positions. The DMEPs were always signi cantly deteriorated upon exion after the compression, but were not damaged signi cantly upon neutrality and extension until 4 wpi. Dynamic MRI data at 4 wpi revealed that the sagittal diameter of the canal was compromised by more than a quarter (26.3%) and the spinal cord by more than half (53.7%) in CCSCI models and did not vary among different postures. The T2WI signal intensities of CCSCI models' cervical grey matter at exion were signi cantly greater than those at neutral and extension, suggestive of severer ischemia at exion. So we looked into the CCSCI models' perfusion at the compression site, and found it was signi cantly deteriorated at 4 wpi, especially upon cervical exion. The dynamic change in SCBF was signi cantly correlated with the change in DMEP and DSSEP amplitudes upon exion. These  positions also showed signi cant abnormalities compared with the sham group but was still signi cantly better than that at the exion position. We assume this was because dorsal compression had much less in uence on the anterior horn and motor conduction pathways. CSM patients suffering from hypertrophic ligamentum avum generally present ataxia, which is mainly caused by dorsal funiculus disruption, but present full strength in all 4 extremities. [41] Ueta et al. [34] noted that SEPs were lost rst during posterior, circumferential, and lateral acute transient compressions, while MEPs were lost rst during anterior compression in pigs, which coordinated with our ndings.
To better understand the mechanisms of varied DMEP and DSSEP performance at dynamic neck positions, we performed dynamic MRI of both sham and CCSCI rats. The relative T2WI signal intensities of the ROI in cervical grey matter in both the sham and CCSCI rats were analysed. The T2WI intensity was the greatest upon exion compared with that at neutral or extension positions. This nding was in line with many clinical studies. [42,43] While some authors assumed that cervical canal expansion at exion permits better visualization of intramedullary hyperintensity on T2WI, [42,43] others concluded that hyperintensities were more related to spinal cord ischaemia. [44] The signi cance of these lesions in CSM patients is also controversial in the literature. [45] Some studies have reported that the pattern of intramedullary hyperintensity lesions is an important criterion. [46,47] In this study, the spinal canal remained constant at all three positions; also, we found a signi cant hyperintense pencil-like lesion pattern on sagittal T2WIs and the so-called "snake eyes" or "owl's eyes pattern" with hyperintense signal conversion on axial T2WIs of CCSCI models upon exion (see Fig. 5D, H). Previous studies reported that these typical "pencil-like" lesions on sagittal orientation and "snake eyes" lesions on axial orientation were important MRI characteristics of spinal cord infarction, indicating partially symmetrical small infarcts in the anterior horns. [44,48] Intramedullary T2WI hyperintensity represents a variety of histologic changes, including oedema, ischaemia, demyelination, gliosis, microcavities, and cavities. Similarly, histological analysis in this study showed that the compressed spinal cord exhibited structural damage in its anterior horn, including fragmentation of neuronal nuclei, pyknosis, neuropil damage, degradation of the extracellular matrix, interstitial oedema, cytoplasmic reduction, and cavity formation. It has been reported that under chronic focal spinal cord compression, there was a decrease in blood ow in the compressed segment, which contributed to the pathogenesis of myelopathy.
[18] Our previous study showed that 4 weeks of chronic compression led to a severe ischaemia-hypoxia environment, as demonstrated by increased HIF-1α expression, [25] and signi cantly decreased microvascular density, as demonstrated by micro-CT and immunohistochemistry data. [39,49] Thus, the increased T2WI hyperintensity in our study was more likely to represent transient ischaemia upon exion. We assume that the CCSCI models, whose spinal cord circulation was already compromised, suffered more perfusion decrease upon exion than sham mice. The increased T2WI hyperintensity at exion in CCSCI models is suggestive of a transient decrease in blood perfusion of the cord, which still needs to be further con rmed.
LDF is a non-invasive method for blood ow measurement, which makes it preferable for measuring microcirculatory alterations of the spinal cord at different neck positions. The technique uses tissue backscattered light to qualitatively assess the blood ow rate and has been a useful method for quantifying perfusion of the spinal cord in several animal and human studies. [50][51][52] Brain et al. [53] rst reported ischemia as one of the crucial pathophysiological mechanisms in CSM. Pathological changes such as vessel wall thickening and hyalinization in the anterior spinal artery and parenchymal arterioles [53], and decreased number of vessels as indicated by the decreased laminin staining have been identi ed in chronic compressed cervical cords [54]. Kurokawa et al[18] previously reported an approximately 20% reduction in blood ow in the compressed segment (C5-C6) compared with that of the rostral (C3-C4) segment in CCSCI rat models using hydrogen clearance to measure blood ow. In our study, the reduction value was 33% in neutral position. The SCBF data in the two studies were not comparable because Kurokawa et al [18] did not quantify their compression degree. Moreover, we rstly report a 27% reduction of SCBF upon exion compared with neutral position in CCSCI rat models at 4 wpi.
Since the vasculature of the spinal cord is organized such that its ventral aspect and grey matter are supplied centrifugally by the central artery and the anterior part of the vasocorona, both of which arise from the ASA, ventral spinal cord compression caused by exion position could probably exacerbate the ischaemic condition by compressing the ASA. We have previously reported that CSM patients with a "snake-eye" appearance of T2WI intramedullary hyperintensity suffered more severe exion DSSEP deterioration in the clinic, probably caused by decreased SCBF at exion. [12] Our LDF data directly indicated the perfusion loss in the compressed spinal cord during cervical exion, as the studies mentioned above have suggested.
Finally, our data also indicated that motor functional loss was signi cantly related to decreased SCBF in CCSCI rat models at a exed neck position. These ndings are in general agreement with those of Bennett [55] , who examined the effect of focal spinal cord ischaemia induced by segmental dorsal and ventral rhizotomy on MEPs and SSEPs in the cat spinal cord. He observed that MEPs more sensitively re ected spinal cord ischaemia than SSEPs but he did not measure SCBF. Fehlings [57] Histological abnormalities in all dogs with combined compression and ischaemia were more pronounced than in dogs with either compression or ischaemia alone. [57] Thus, DMEP deterioration upon exion in our CCSCI rat models was also more likely to be due to the combinatorial effects of ischaemia and direct neuronal compression.

Conclusions
In     The alteration of instant spinal cord blood ow and oxygen saturation level in different positions after CCSCI, and its correlations with electrophysiological changes. The maps of instant blood ow and oxygen saturation level were evaluated by using a laser Doppler owmetry at the C5/6 segment of the cervical cords upon neutral (a), extension (b) and exion (c) positions in Sham and CCSCI groups. Spinal cord blood ow (SCBF) (d) and oxygen saturations (e) at the compression sites of the CCSCI rat models were signi cantly lower than that of sham models at all three neck positions (p<0.001). Both of them also deteriorated signi cantly upon exion (p<0.001). Spearman correlation analysis demonstrating that the R-SCBF was not correlated with relative amplitude of DMEPs (RA-DMEP) or relative amplitude of DSSEPs (RA-DSSEP) upon extension(f, g). It signi cantly correlated with the RA-DMEP (Spearman correlation, R=0.72, p<0.001) (h), but not with RA-DSSEP upon exion (i). *, **, *** Indicated statistical signi cance (p < 0.05, 0.01, 0.001 respectively) in the difference between the Sham and CCSCI group using t test. #