Introduction

Patients with intramedullary spinal lesions, such as spinal cord tumors and syringomyelia, often require surgery. However, surgical interventions can inadvertently damage the spinal cord tissues surrounding the lesion, resulting in motor paralysis, sensory paresthesia, and neuropathic pain. Although this is a devastating condition, no reliable method is currently available to predict its occurrence or to evaluate the intensity of symptoms. Therefore, the underlying mechanisms of intramedullary spinal lesions should be determined to improve the clinical outcomes.

Voxel-based morphometry (VBM) is a novel neuroimaging analysis technique based on magnetic resonance imaging (MRI). Using the statistical parametric mapping approach, VBM enables investigators to evaluate focal anatomical differences in the brain [1, 2]. VBM has been used to assess the cerebral structural characteristics in patients with chronic low back pain, fibromyalgia, and migraine [3,4,5,6]. However, no study has yet investigated the potential cerebral structural changes in patients who have undergone spinal cord surgery. In this study, we analyzed the cerebral structural characteristics in 49 patients who had undergone spinal cord surgery using VBM, and examined the relationship between cerebral structural differences and post-surgical chronic myelopathic pain.

Methods

Participants

In all, 49 participants [29 males and 20 females; average age, 55.3 ± 14.5 years (mean ± SD)] who had undergone surgery at our hospital for spinal intramedullary lesions between January 2002 and April 2014 were enrolled in this study. Among these, 18 participants had ependymoma, 4 had astrocytoma, 3 had subependymoma, 6 had hemangioblastoma, 6 had cavernous hemangioma, 3 had spinal syringomyelia, and 9 had other lesions (subpial neurinoma, intramedullary lipoma, intramedullary arachnoidal cyst, and dural arteriovenous fistula). Postoperative duration was 181–6174 (mean, 1884) days. Based on the numeric rating scale (NRS) score, participants were divided into two groups: the Pain group including those with NRS score of ≥3 and Control group including those with NRS score of <3 (Table 1a). After agreeing to participate in the study, each participant completed the painDETECT questionnaire, which is a widely accepted screening tool for neuropathic pain [7]. In addition, participants in the Pain group were further classified into two subgroups: below-level pain (BLP) (+) group including those with BLP and BLP (−) group including those without BLP, according to the recently updated international spinal cord injury classification [8, 9]. BLP refers to the neuropathic pain that is perceived to be greater than three dermatomes below the neurological level of the surgical intervention [10].

Table 1 Characteristics of each patient group.
Full size table

The state of paralysis and motor function were evaluated with the modified McCormick scale (grade range I–V: I = normal gait, II = mild gait disturbance not requiring support, III = gait with support, IV = assistance required, and V = wheelchair needed) [11, 12]. Geographic data of the subject date are shown in Table1. All participants were right handed. In the BLP(+) group, twelve participants had pain in both sides, one in the right, and five in the left. In the BLP(−) group, three participants had pain in both sides, nine in the right, and seven in the left. There was no significant difference in the laterality between the groups (Kruskal-Wallis test, p < 0.05).

MRI scans and data analysis

MRI scans were obtained using a 3.0 T MRI (Discovery MR750/GE Healthcare, Tokyo, Japan). Each image volume contained 45 axial slices covering the entire brain (T1WI 3D IR-FSPGR, voxel = 0.94 × 0.94 × 1.00 mm). Brain MRI results were obtained in all but two participants. All MRI images were processed using MATLAB (Mathworks, Natick, MA) and the SPM12 (Welcome Trust Centre for Neuroimaging, UCL Institute of Neurology, London, UK). VBM was used to morphometrically evaluate all images. Two sample t-test between the two groups and a correlation analysis between NRS score and gray matter volume were performed with age and total gray matter volume as covariates. A cluster was considered statistically significant with a threshold of cluster size familywise error-corrected p < 0.05.

Statistical analysis

Mann–Whitney U tests and Kruskal-Wallis tests were used for the statistical comparison between groups. The Spearman’s test was used to assess the correlation coefficients. P-value of <0.05 was considered statistically significant. Data are presented as mean ± standard deviation unless otherwise indicated.

Results

Among the 49 participants, 11 were included in the Control group and 38 in the Pain group (Table 2a). No significant difference was observed in age or postoperative days between the groups. participants in the Pain group were further divided into two subgroups: the BLP(+) and the BLP(−) groups (Table 2b). NRS and painDETECT scores were higher in the BLP(+) group than that in the BLP(−) groups; however, the difference did not reach statistical significance. Age was significantly higher in the BLP(+) group than that in the BLP(−) group (Table 2b). The painDETECT score, which reflects the neuropathic pain intensity, was significantly correlated with NRS score (Fig. 1a; R = 0.849, p = 4.71 × 10−14), indicating that participants in the Pain group were mostly afflicted with neuropathic pain. There was no significant correlation between NRS score and the modified McCormick Scale, and no significant differences among groups in the Kruskal-Wallis test (p < 0.05) (Fig. 1b).

Table 2 Group comparison.
Full size table
Fig. 1: Corralation analysis between NRS and painDETECT/modified McCormick scale.
figure 1

a Correlation between NRS score and painDETECT score in patients who had undergone intramedullary spinal surgery. b Correlation analysis between NRS score and the modified McCormick Scale in patients who had undergone intramedullary spinal surgery. There was no significant difference between NRS score and modified McCormick scale.

Full size image

Comparative analyses between the pain and Control groups revealed that gray matter volumes in the left supplementary motor area (SMA), left primary motor area (M1), and left posterior cingulate cortex (PCC) were significantly greater in the Pain group than that in the Control group (Fig. 2 and Table 3). Comparative analyses also showed that gray matter volumes among these areas were all significantly greater in the BLP(+) subgroup than those in the Control group (Table 3 and Fig. S1).

Fig. 2: The gray matter volume is increased in specific cerebral areas in patients in the Pain group than that in the Control group.
figure 2

Regions with increased gray matter (M1, SMA, and PCC) in the Pain group are superimposed on a normalized structural cerebral image. The color bar represents the t-score. Coordinates (X, Y, and Z values) are given in the Montreal Neurological Institute (MNI) space.

Full size image
Table 3 Brain regions with increased gray matter volume in the Pain group and the BLP group each compared with the Control group.
Full size table

Correlation analyses using the VBM data and NRS score revealed a significant correlation between NRS score and gray matter volume in the left M1, left PCC, and right superior parietal lobule (SPL) [Table 4, Fig. 3a and (Fig. 3b; R = 0.580, p = 1.93 × 10−5)].

Table 4 Brain regions with a significant correlation between the increase in the gray matter volume and NRS.
Full size table
Fig. 3: VBM correlation analysis between the NRS score and gray matter volume.
figure 3

a The regions showing the positive correlation between the NRS score and gray matter volume are superimposed on a normalized structural cerebral image. The color bar represents the t-score. Coordinates (X, Y, and Z values) are given in the Montreal Neurological Institute (MNI) space. b. Correlation analysis on the relative value of the gray matter volume at x = 15, y = 45, z = 60 (MNI coordinates) and NRS.

Full size image

Concerning the correlation between the VBM data and post-surgical duration, there was no significant difference with a threshold of cluster size familywise error-corrected p < 0.05.

Discussion

In this study, we attempted to identify cerebral structural characteristics using VBM in patients who had undergone intramedullary spinal surgery. The gray matter volume in the left SMA, left M1, and left PCC was found to be significantly increased in participants with post-surgical chronic myelopathic pain (Pain group). A subgroup analysis revealed that the presence of BLP is closely associated with increased gray matter volume in these areas. Furthermore, the gray matter volume in the left M1, left PCC, and right SPL was found to be significantly correlated with NRS score. These results provide evidence that structural differences occur in specific cerebral areas in patients with post-surgical chronic myelopathic pain. The post-surgical duration after spinal intramedullary surgery was ranged from 181 to 6174 days, and this duration was thought to be long enough to develop myelopathic pain [13]. Post-operative duration did not have any impact on the type of pain. The post-surgical myelopathic pain occurred within 180 days after surgery and did not resolve over time.

VBM indicates that the left SMA, left M1, and left PCC are all associated with post-surgical chronic myelopathic pain; however, the causal relationship between the structural changes in these areas and myelopathic pain remains to be determined. Nevertheless, previous studies have shown the potential involvement of these three cerebral areas in regulating somatic pain. The SMA region of the cerebral cortex that contributes to movement control is also considered to be a part of the pain matrix that functions to integrate sense and body movements [14, 15]. The M1 is the primary region of the motor system commanding voluntary movements in association with other motor areas. Interestingly, previous studies suggest that this area is significantly associated with chronic pain [5, 14, 16]. The PCC plays a role in emotion comprehension and the default mode network in the brain as part of the pain matrix [17, 18]. Conversely, the correlation between the SPL gray matter volume and NRS is controversial. The SPL is mainly involved in spatial orientation recognition, visual processes, and sensory input from the hand. Based on our literature review, no data suggested the potential involvement of SPL in processing the pain signal. Nevertheless, our results suggest that the SPL is involved in amplifying pain signal, at least in those with post-surgical chronic myelopathic pain.

Several previous studies have investigated the pain matrix [19]. Studies using functional MRI show that nociceptive pain is related to the thalamus, anterior cingulate, insula, primary, and second somatosensory areas[20, 21]; whereas, non-nociceptive pain processing is related to the hippocampus, medial prefrontal cortex, basal ganglia, post cingulate cortex, and inferior parietal lobule [19, 22]. Subjects in present study have post-surgical myelopathic pain, suggesting the involvement of the latter brain region.

Several previous studies have shown changes in the gray matter volume using VBM in patients with pain [2, 6]. Jutzeler et al. and Ung et al. showed an increased gray matter volume in the left primary/secondary somatosensory area, M1 and anterior cingulate cortex, and decreased gray matter volume in the primary somatosensory area and thalamus in patients with chronic back pain [5, 23]. Burgmer et al. reported that patients with fibromyalgia showed decreased gray matter volume in the right prefrontal cortex, left amygdala, and right anterior cingulate cortex [3]. Ivo et al. reported that patients with low back pain had decreased gray matter volume in the right dorsolateral prefrontal cortex, right thalamus, and right middle cingulate cortex [4]. The reasons for discrepancies among previous studies and ours are unclear; however, they are most likely derived from differences in the study design, especially in differences in the patient population. participants in the Pain group in this study were all suffering from post-surgical chronic myelopathic pain, whereas participants in other studies were inflicted with chronic disorders, such as low back pain and fibromyalgia. Moreover, in present study, gray matter volume change was seen in the left dominant counterparts. So far, there is no consensus on the difference in the laterality; however, some studies have suggested that the neurodegeneration caused by aging and disease may preferentially affect the language- and motor-dominant left hemisphere [23]. Further studies are required to understand how differences in the type of pain would affect structural changes in the brain differently.

A previous study has shown a correlation between the pain intensity and gray matter volume in the M1 in patients with spinal cord injury [24], indicating that the deafferented pain induces gray matter thickening in the M1. In accordance, our data also showed a correlation between the NRS score and increased gray matter volume in the left M1 and in the left PCC and SPL (Fig. 3b). Given that increased gray matter in particular brain regions reflects the intensity of myelopathic pain, it is tempting to speculate that these changes are potentially involved in the development of central sensitization, a condition that triggers chronic and amplified pain.

Several factors limit the generalization of our findings. First, as discussed above, the causal relationship between the cerebral structural changes and degree of pain remains to be elucidated because this is an observational study, and lacks adequate controls (including pre-operative brain MR images). Second, the age- and gender-matched control group was not included in this study. Therefore, some results in this study must be cautiously interpreted. Third, other factors that could potentially affect the degree of pain, such as psychological factors and medication, were not evaluated. In addition, neither recruitment of the patrticipants or the analysis was not blinded-manner.

In conclusion, this is the first study to identify the cerebral structural changes in patients who have undergone intramedullary spinal surgery. Our data revealed increased gray matter volume in specific cerebral areas in patients with post-surgical chronic myelopathic pain. Although further studies are required to confirm our findings, this study suggests that VBM is a valid tool to objectively evaluate post-surgical chronic myelopathic pain in patients who have undergone spinal cord surgery and therefore has important clinical implications.

Data archiving

The datasets generated during the current study are available from the corresponding author on reasonable request.