Resting-state functional magnetic resonance imaging indices are related to electrophysiological dysfunction in degenerative cervical myelopathy

The age-related degenerative pathologies of the cervical spinal column that comprise degenerative cervical myelopathy (DCM) cause myelopathy due spinal cord compression. Functional neurological assessment of DCM can potentially reveal the severity and pathological mechanism of DCM. However, functional assessment by conventional MRI remains difficult. This study used resting-state functional MRI (rs-fMRI) to investigate the relationship between functional connectivity (FC) strength and neurophysiological indices and examined the feasibility of functional assessment by FC for DCM. Preoperatively, 34 patients with DCM underwent rs-fMRI scans. Preoperative central motor conduction time (CMCT) reflecting motor functional disability and intraoperative somatosensory evoked potentials (SEP) reflecting sensory functional disability were recorded as electrophysiological indices of severity of the cervical spinal cord impairment. We performed seed-to-voxel FC analysis and correlation analyses between FC strength and the two electrophysiological indices. We found that FC strength between the primary motor cortex and the precuneus correlated significantly positively with CMCT, and that between the lateral part of the sensorimotor cortex and the lateral occipital cortex also showed a significantly positive correlation with SEP amplitudes. These results suggest that we can evaluate neurological and electrophysiological severity in patients with DCM by analyzing FC strengths between certain brain regions.


The demographic and clinical information
Thirty-four patients with DCM and 21 HCs participated in this study (Table 3).There was no significant age difference between patients with DCM and HCs (p = 0.995).In addition, there was no significant sex difference between patients with DCM and HCs (p = 0.94).The mean total JOA score in DCM patients was 11.0 ± 2.4.The medications taken by each patient ware also shown in Table 3.
We did not perform the CMCT measurement in patients who had a cardiac pacemaker or intracardiac lines or metal in body because of the contraindication 13,[43][44][45][46][47] .As a result, we measured CMCT from 33 patients with DCM (Table 3).We could not examine SEP amplitudes for 14 patients because SEP could not be measured in patients with very severe paralysis.Therefore, we measured SEP amplitude to 20 patients with DCM (Table 3).

Comparison of the fALFF between patients with DCM and the healthy control group
We compared fALFF in the < 0.1 Hz low-frequency band to that in a full-frequency band of fMRI signals between patients with DCM and healthy controls (HC).This measure reflects the magnitude of spontaneous regional brain activity 12,50 .The patients with DCM showed significant increase in fALFF in the occipital pole and in the lateral occipital cortex superior division compared with those in the HC (Table 1).Furthermore, patients with

Discussion
In the present study, we attempted to apply rs-fMRI to explore predictive measures for evaluating the severity of DCM.Previous rs-fMRI studies have reported several correlations between brain activity indices, such as FC, ALFF, and regional homogeneity, and clinical scores/scales or neurological symptoms 31,35,37 .To our best knowledge, however, this is the first rs-fMRI study to show a correlation between the electrophysiological indices and FC strength in patients with DCM.
This study showed a significant increase in fALFF in the occipital pole and lateral occipital cortex superior division in the patients with DCM.A previous study also noted that patients with DCM showed an increase in fALFF in the occipital cortex 33 .Interactions between visual and proprioceptive information have an important role in motor programming 56,57 .Proprioception is impaired in patients with DCM 58 , and severe DCM leads to severe movement disorders.Therefore, this increase in fALFF in the occipital cortex suggests that the visual system compensates for the lack of proprioceptive information.We also found a significant decrease in fALFF in the cerebellar hemisphere, superior parietal lobule, frontal pole, right central opercular cortex, thalamus, and the inferior temporal gyrus posterior division.These brain areas are thought to taken part in sensory information processing and/or cognitive function.A previous study revealed that patients with DCM have sensory and   www.nature.com/scientificreports/cognitive deficits 42 , therefore we think that the decrease in fALFF in these areas reflects such functional deficits in patients with DCM.
CMCT is an objective measure representing the condition of the central motor pathway.In the present study, FC strength between the primary motor cortex and the precuneus correlated positively with CMCT (Fig. 1).The role of the primary motor cortex in the execution of movements is crucial 59 .In addition, the precuneus, a part of the default mode network, is related to performing a variety of highly integrated cognitive tasks involving visuo-spatial imagery, retrieval of episodic memory, and self-processing operations 60,61 .As patients with severe DCM have severe movement disorders, it can be assumed that patients with severe DCM need to functionally compensate for impaired motor functions to maintain their motor ability 33,62 .The role of the precuneus in integrating sensorimotor and other endo/exogenous information is also crucial 63 .That is, the large-scale brain network may interact with the primary motor cortex via the precuneus, and this interaction may be involved in the cognitive compensation for motor dysfunction.Klöppel et al. reported that the interaction between executive and cognitive motor areas, i.e., the supplementary motor area and superior parietal lobule, respectively, was involved in compensation for motor deficits of patients with pre-symptomatic Huntington's disease 64 .In addition, another previous study revealed that sensorimotor network-default mode network interactions were involved in the compensatory function of motor performances in the stroke rehabilitation process 65 .Considering these findings, brain regions involved in cognitive functions may assist processing in motor areas and compensation for motor deficits.Therefore, the increased FC strength between the primary motor cortex and the precuneus in patients with DCM may reflect cognitive compensation of motor dysfunction in these patients.
SEP amplitude is an objective measure for the assessment of the condition of the sensory tracts or sensory loss.FC strength between the thalamus and the precentral gyrus anterior division correlated with SEP amplitude in this study (Fig. 2).A previous study showed that a postoperative DCM group manifested decreased FC between the right thalamus and the bilateral primary motor cortex 66 .The primary motor cortex is crucially involved in executing voluntary movements.In addition, the thalamus works as relay hub that directs information between different subcortical areas and the cerebral cortex, especially in the sensory area.Both are parts of the sensorimotor network, which processes bodily sensations and sends signals to the motor cortex to execute appropriate motor responses.We think that the positive correlation between FC strength and SEP amplitude reflects functional compensation for the sensory and motor disability in patients with DCM 67 .
A correlation of FC strength between the sensorimotor lateral and the cingulate gyrus anterior division with right SEP amplitude was also found.Moreover, this was the only FC to show a significant group difference (Fig. 3).The anterior cingulate cortex is the central part of the salience network, which is thought to be associated with allocating attentional resources to a salient stimulus such as pain 68 .It is also well known that acute pain activates the salience network and the sensorimotor system 69 .In patients with DCM, the more severe the spinal cord injury, the more impaired is signal transmission from the body to the brain.As a result, severe sensory deficits may decrease FC strength between the sensorimotor and anterior cingulate cortex, thus causing dysfunction of sensory information integration in the sensorimotor network and the salience network in these patients 70 .Moreover, the significant difference in this FC strength between the patients with DCM and the HC further supports the notion that alteration of this FC is related to sensory deficits in these patients.
We also showed that FC strength between the sensorimotor lateral and lateral occipital cortex correlated with SEP amplitude (Fig. 4).Sensory deficits occur in not only exteroception but also in deep sensation 58,71 .Impairment of deep sensation probably causes visual dependence on motor control.Therefore, our finding reflects the visual information-dependent compensatory mechanisms for sensorimotor deficits of the upper limbs in patients with DCM 72,73 .
Although we observed significant correlations between several FC strengths and CMCT or SEP amplitude in DCM patients, we did not find significant differences between the DCM and HC among them except for one (FC between sensorimotor lateral and cingulate gyrus anterior division).The interpretation of the results is difficult because the relationship between the electrophysiological indices and FC strengths has never been investigated in humans or animals.
We speculate there are two reasons for the lack of difference in FC strengths between the DCM and HC.One reason is the physical status during rs-fMRI scans, and the other is that tract damage indirectly affects FC strengths.DCM patients felt little neurological deficit during rs-fMRI scans because they did not perform active movements of the upper and lower limbs.Therefore, spontaneous brain activities, particularly FC, of patients The changes in FC strengths in patients with DCM are thought to reflect the compensatory mechanisms for maintenance of motor and sensory functions depending on visual information and cognitive functions.Therefore, changes in FC strengths in patients with DCM may be too small to show significant differences to HCs, even though they show significant correlations to physiological measures.Further studies for not only humans but also animals are necessary in order to test these speculations.

Limitations
Several patients with DCM had been administered analgesia.Previous studies revealed that analgesic agents acting on the central nervous system affect resting brain activity 74,75 .Therefore, analgesics may have influenced the present study results.Information on the analgesics administered in each patient is shown in Table 3.
Although SEP amplitude is a continuous variable, the SEP amplitudes were zero in 14 cases due to severe sensory dysfunction.Therefore, the "floor effect" may have affected the results of the correlation analysis.Intraoperative SEP examination was conducted under anesthesia.From the viewpoint of how alterations of SEPs in patients with DCM affect their brain activities, this is also the limitation in this study because the conditions of patients differ between the rs-fMRI scans and SEP measurements.However, in clinical, SEPs measured under anesthesia are widely used.Moreover, as mentioned above, SEP examination places a burden on patients.Considering them, rs-MRI without such a burden is the reasonable method to develop alternative SEP measures which detect sensory tract damage, and we believe that it is reasonable to compare rs-fMRI measures in awake rest and SEPs under anesthesia.
In this study we only revealed correlations between FC strength and preoperative electrophysiological measures.As is widely known, correlation does not imply causation.Moreover, we did not validate the accuracy of FC as a predictive measure of the severity of DCM.Therefore, it will be necessary to evaluate the capability of FC to assess DCM with larger samples in future studies.

Conclusion
We showed significant correlations between FC strengths among certain brain regions and the preoperative and intraoperative neurophysiological indices of CMCT and SEP amplitude in patients with DCM.Our findings indicate the feasibility of rs-fMRI and FC analysis to provide novel predictive measures for assessment of the neurological severity of DCM.

Participants
We recruited 34 patients with DCM (11 females, mean age: 65.6 years, range: 33-85 years) at our institution from May 2020 to June 2021.Inclusion criteria were that they (1) agreed to participate in this study; (2) had a medical indication for cervical decompression surgery; and (3) had symptoms caused by DCM 6,7 .Exclusion criteria were that they (1) refused to enroll; (2) had traumatic spinal cord compression; and (3) had a history of brain diseases such as cerebrovascular disease or tumors.We also recruited 21 age-and sex-matched HC, (7 females, mean age: 65.6 years, range: 33-83 years).The characteristics of the patients with DCM are shown  3. We used the Japanese Orthopedic Association (JOA) score to assess DCM severity.The JOA score consists of six domains: motor function of upper extremity/lower extremity, sensory deficit of upper extremity/ lower extremity/trunk, and bladder dysfunction 76 .The Ethical Review Board of Yamaguchi University Hospital approved all protocols for this study (approval no.: H30-207-2), which was performed in accordance with approved guidelines and in compliance with the principles of the Declaration of Helsinki.All participants provided written informed consent before undergoing any study procedure.

Electrophysiological assessment
We conducted an electrophysiological examination to assess objective neurological function and deficit in each patient with DCM.We measured preoperative CMCT in the upper and lower limbs and intraoperative posterior tibial nerve SEP.

Measurement of CMCT
We examined the CMCT in all of the patients with DCM.We placed self-adhesive surface recording electrodes on the target muscles according to the belly-tendon method.The motor evoked potentials (MEPs) were recorded on the left side during voluntary contraction from the abductor digiti minimi (ADM) and abductor hallucis (AH) muscles 72 .We delivered transcranial magnetic stimulation through a round coil of 14-cm outer diameter (Magstim Co. Ltd., Spring Gardens, Whitland, UK) positioned flat on the scalp and centered over the Cz in the International 10-20 system at an intensity set at 20% above the MEP threshold during voluntary contraction.
We first recorded the electromyogram (EMG) during maximum voluntary contraction.Then, the patients with DCM were told about the amplitude of their raw EMG signal, and they worked to maintain a level 10-20% of that recorded during the maximum voluntary contraction.With this procedure, we confirmed that patients could maintain 10-20% of the maximal force of contraction during the MEPs measurements.A supramaximal electric stimulation was delivered to the ulnar nerve at the wrist and the tibial nerve at the ankle during which compound muscle action potentials (CMAPs) and F waves were recorded.Reference ground plates were located on the forearm or lower limb.We examined 16 serial responses and measured the shortest latency of the F waves.CMCT was calculated as MEPs latency − (CMAPs latency + F wave latency − 1)/2 (ms).All CMCT measurements were made with a Nicolet Viking 4 instrument (Natus Medical Incorporated, San Carlos, CA, USA).CMCT-ADM and CMCT-AH were measured in all patients 39,78 .

Measurement of SEP
We examined intraoperative posterior tibial nerve (PTN) SEP in 20 patients with DCM.For intraoperative neuromonitoring, patients were placed under total intravenous anesthesia maintained with propofol 2.5-3.5 mg/ mL delivered via a target-controlled infusion technique and remifentanil 0.1-0.3mg/kg/min.Rocuronium 0.6-0.9mg/kg (muscle relaxant) was administered only at the induction of anesthesia.The bispectral index score (BIS) derived from a frontal electroencephalogram (Aspect Medical Systems, Newton, MA, USA) was used to monitor anesthetic depth.An anesthetic depth associated with 40 < BIS < 60 was maintained by adjusting the rate of propofol infusion.
Stimulation needs to be delivered to the posterior tibial nerve at the ankle.Surface electrodes were placed on the skin over the nerve where it passes posterior to the medial malleolus.The cathode was applied midway between the medial border of the Achilles tendon and posterior border of the malleolus, whereas the anode electrode was applied away from the nerve 3 cm distal to the cathode.The electrical pulse delivered should have sufficient intensity to cause plantar flexion of the toes of 1-2 cm.The stimulator was set at intensity 30-40 mA, stimulus rate 4.7 Hz, and duration 0.5 ms according to the Guidelines of the International Federation of Clinical Physiology 79 .We used the same stimulation system with all patients.PTN-SEPs were recorded in a 60-80-ms time window following the stimulation.The electrode for scalp recording was placed at site Cz of the international 10-20 system, and the reference electrode was located at the Fz site on the forehead, and electrical impedances at these sites were < 5000 Ω. SEP amplitude was defined as the peak-to-peak amplitude of N1 and P1 (I-PTN).The amplifier settings for I-PTN-SEP recording included 20-3000-Hz bandpass filtering and 60-Hz notch filtering, with 200 trials averaged 44 .All SEP tests were performed with a Neuromaster MEE-1200 (Nihon Kohden, Tokyo, Japan).Senior electrophysiologists with 10 years of experience in evoked potentials study analyzed the recordings for presence of the main peaks P1-N1.

MRI data acquisition
MRI scanning was by a 3.0 Tesla MRI scanner (Siemens Magnetom Prisma, Siemens Healthineers, Erlangen, Germany).During scanning, participants were instructed to simply rest with their eyes open, stare at one point, not to think of anything, and try not to fall asleep.We also confirmed whether the participants had not fallen asleep during the resting scan by direct questioning.To acquire the functional images, a 10-min scan was performed using a gradient-echo echo-planar pulse sequence with the following parameters: repetition time = 2500 ms, echo time = 30 ms, slice thickness = 3.

Data quality assessment
To check data quality, we confirmed no differences in the number of excluded fMRI volumes and size of body movements in each direction (x, y, z, pitch, roll, yaw) between the two groups by using two-sample t-tests.The thresholds were set at p < 0.05 for these tests.

rs-MRI data analysis
Preprocessing Data pre-processing was done using the CONN-fMRI toolbox (version 20b) The first five volumes of each participant's fMRI data were discarded, and then the following steps were performed: slice timing correction, realignment, normalization to the Montreal Neurological Institute stereotactic template with resolution of 2 × 2 × 2-mm cubic voxels and smoothing with 8-mm full-width at half-maximum Gaussian kernel.
In addition to image processing, to remove non-neural artifacts, components associated with fMRI signals in white matter/cerebrospinal fluid were extracted with the component-based noise correction procedure 82 , after which six motion parameters were regressed from the data.Subsequently, scrubbing of high-motion timepoints was also performed, and finally, a 0.008-0.09-Hzbandpass filter was applied 83,84 .

fALFF analysis
For each participant, we calculated the fALFF value 54 in each voxel of preprocessed rs-fMRI data using the CONN-fMRI toolbox and SPM12.ALFF value is a measure of BOLD signal power within the low frequency band (typically 0.01-0.1 Hz), and is defined as the root mean square of the BOLD signal at each voxel after temporal filtering 55 .fALFF value is a relative measure of BOLD signal power within the low frequency band compared to the entire frequency spectrum, and is defined as the ratio of root mean square of the BOLD signal at each voxel after vs. before temporal filtering 54 .Although these indices have been developed to characterize regional spontaneous brain activity, fALFF has been proven to be more gray matter-specific and sensitive to BOLD signal 14 .

FC analysis
We conducted seed-to-voxel correlation analyses with the CONN-fMRI toolbox and SPM12 to evaluate FC in the patients with DCM and HC.This analysis computed functional connectivity between a seed and remaining voxel as the Fisher-transformed bivariate correlation coefficients between a seed BOLD signal and each voxel BOLD signal.

Statistics
We performed a two-sample t-test to compare fALFF values between CM and HC and a multiple regression analysis to investigate the relationship between fALFF values and electrophysiological measures.In multiple regression analysis, covariates of interest were right CMCT, left CMCT, right SEP amplitude, and left SEP amplitude.Moreover, age, disease duration, and JOA score was put into the model as covariates of uninterest.For these statistical tests, the thresholds were set at p < 0.001 (uncorrected, peak-level) and p < 0.05 (false discovery rate [FDR]-corrected, cluster-level).Furthermore, multiple comparison correction (Bonferroni correction) was applied to the cluster level threshold in the correlation analyses (p < 0.0125).
To investigate the relationship between FC strengths and electrophysiological measures, we performed multiple regression analysis.As in the multiple regression analysis for fALFF values, we put right CMCT, left CMCT, right SEP amplitude, or left SEP amplitude into the regression model as covariates of interest, and put age, disease duration, and JOA score into the model as covariates of uninterest.In particular, for FC of the left or right primary motor cortex, contralateral CMCT was a covariate of interest.For FC of the left or right primary somatosensory cortex or thalamus, contralateral SEP amplitude was a covariate of interest.
The thresholds were set at p < 0.001 (uncorrected, peak-level) and p < 0.05 (FDR-corrected, cluster-level).Multiple comparison correction (Bonferroni correction) was further applied to cluster-level threshold values (p < 0.00625).Moreover, for FC showing a significant correlation in the correlation analysis, we compared its strength between patients with DCM and HC using a two-sample t-test.The thresholds were set at p < 0.05 for these tests.
We created bar graphs and scatter plots with Microsoft Excel version 16.73 and used its functions to add error bars to the bar graphs and regression lines to the scatter plots and to calculate the coefficients of determination (R 2 ).Analysis of all data was performed with StatFlex Ver. 7 for Windows (Artec, Osaka, Japan; https:// www.statfl ex.net/).

Figure 1 .
Figure 1.Correlation between left central motor conduction time (CMCT) and functional connectivity (FC) strength of the right primary motor cortex.(a) A cluster showing a significant positive correlation between left CMCT and FC strength of the right primary motor cortex, b) scatter plot of left CMCT versus FC strength, and c) comparison of FC strengths between patients with degenerative cervical myelopathy (DCM) and healthy controls (HC).The dashed line in panel b is the regression line.Error bars in panel c represent standard deviations.

Figure 2 .
Figure 2. Correlation between right somatosensory evoked potentials (SEP) amplitude and functional connectivity (FC) strength of the left thalamus.(a) A cluster showing a significant positive correlation between right SEP and FC strength of the left thalamus, (b) scatter plot of right SEP versus FC strength, and (c) comparison of FC strengths between patients with degenerative cervical myelopathy (DCM) and healthy controls (HC).The dashed line in panel (b) is the regression line.Error bars in panel (c) represent standard deviations.

Figure 3 .
Figure 3. Correlation between left somatosensory evoked potentials (SEP) amplitude and functional connectivity (FC) strength of the right lateral part of the sensorimotor network.(a) A cluster showing a significant positive correlation between left SEP and FC strength of the right lateral part of the sensorimotor network, (b) scatter plot of left SEP versus FC strength, and (c) comparison of FC strengths between patients with degenerative cervical myelopathy (DCM) and healthy controls (HC).The dashed line in panel (b) is the regression line.Error bars in panel c represent standard deviations.

Figure 4 .
Figure 4. Correlation between left somatosensory evoked potentials (SEP) amplitude and functional connectivity (FC) strength of the right lateral part of the sensorimotor network.(a) A cluster showing a significant negative correlation between left SEP and FC strength of the right lateral part of the sensorimotor network, (b) scatter plot of left SEP versus FC strength, and c) comparison of FC strengths between patients with degenerative cervical myelopathy (DCM) and healthy controls (HC).The dashed line in panel (b) is the regression line.Error bars in panel (c) represent standard deviations.

Table 1 .
Differences in fALFF between patients with degenerative cervical myelopathy and healthy control.x,y, and z represent the local maximum in a cluster in the MNI coordinate.T value is the t value of a local maximum.Cluster-level threshold was p < 0.05 (false discovery rate corrected).fALFF fractional amplitude of low-frequency fluctuation, DCM degenerative cervical myelopathy, HC healthy control, R right, L left.

Table 2 .
Correlation between FC strength and electrophysiological index in patients with DCM.x, y, and z represent the local maximum in a cluster in the MNI coordinate.T value is the t value of a local maximum.Cluster-level threshold was p < 0.05 (false discovery rate corrected).Bonferroni correction was further applied to the cluster-level threshold (p < 0.00625).FC functional connectivity, DCM degenerative cervical myelopathy, CMCT central motor conduction time, SEP somatosensory evoked potential, R right, L left, BA Brodmann area.