Article Text
Abstract
Objective Peripheral nerve injury can induce immediate and long-standing remodelling of the brain cortex, which may affect outcomes of nerve repair. This study examined changes of corresponding cortical representations in patients with brachial plexus injuries.
Methods Resting-state fMRI was acquired for 13 adult patients with total brachial plexus root avulsion, three of whom underwent second scans 7 or 8 months later. The time of examination ranged from 1 to 16 months after injuries. Nine healthy adults were enrolled as control. Seed-based functional connectivity was performed for all subjects.
Results For nine patients whose first fMRI was performed from 1 to 4 months after brachial plexus injuries, images showed that their cortical maps of sensorimotor areas corresponding to the hand and arm in the hemisphere contralateral to the injured side had much weaker correlation with the supplementary motor area (SMA) than those ipsilateral to the injured side. Symmetrical maps of bilateral cortical sensorimotor areas corresponding to the hand and arm were observed in other four cases with fMRI tested from 7 to 16 months after injuries. For three of the nine patients with asymmetrical cortical representations, second scans indicated symmetric results or even stronger correlation with SMA in the cerebral cortex contralateral to the injured side.
Conclusions Total brachial plexus root avulsion causes cortical representations of the brachial plexus to undergo a change from an inactive to an active state. This implies that peripheral deafferentation after brachial plexus injuries will induce corresponding cortical representations to be occupied by adjacent non-deafferented cortical territories.
- FUNCTIONAL IMAGING
- PERIPHERAL NERVE SURGERY
- MRI
- NEUROPHYSIOL, CLINICAL
Statistics from Altmetric.com
Introduction
Total brachial plexus root avulsion is one of the commonest of brachial plexus injuries, which is usually caused by a high-speed motorcycle or car accident. In this injury, the rootlets of the C5-T1 spinal nerves connecting the central nervous system with peripheral nerves are divided, which makes the whole upper limb paralysed. Treatment of the injury is still difficult, although techniques of extraplexal nerve transfer, of free-functioning muscle transfer and even of re-plantation of avulsed roots to the spinal cord have been applied since the last five decades.1 In a rat model study of contralateral C7 transfer for treatment of total brachial plexus root avulsion, we found that the extent of transhemispheric functional reorganisation of the motor cortex correlates closely with regaining of independent motor function.2 This implies that the central nervous system plays an important role in the outcome of nerve reconstruction operations after brachial plexus injuries.
Peripheral nerve injury, which induces removal of sensory inputs and blockage of motor output activity, will render corresponding cortical representation of the injured nerve to be taken over by adjacent non-deafferented cortical areas. It was found in primates with division of the radial and median nerves that sizes of corresponding cortical areas were changed within minutes after injury.3 In monkeys with complete deafferentation of inputs from the forelimb persisting for more than 10 years, activities could be evoked in corresponding deprived cortical representations in response to stimuli from the face.4 This type of cortical reorganisation occurs when no former targets of this nerve are re-innervated, and it can be termed phase-one reorganisation. If, after repair, the transected nerve successfully regenerates and re-innervates its former targets, its original cortical area can be re-established, and this can be called phase-two reorganisation. In this stage, the original cortical representation will never be all re-established owing to misdirection of axonal regeneration in the repair site.5 In our former study of transhemispheric cortical reorganisation after contralateral C7 transfer in rats, we demonstrated that encroachment of hindlimb representations upon original forelimb representations was an important factor for prevention of transhemispheric cortical reorganisation.2 We conjecture, therefore, that preventing adjacent cortical areas from occupying the original brachial plexus cortical territory in the phase-one cortical reorganisation may facilitate the phase-two reorganisation after brachial plexus repair, the latter of which is critical to improvement of efficacy after nerve reconstruction operations. The purpose of this study was to observe dynamic changes of sensorimotor areas of the brain cortex corresponding to the hand and arm in patients with total brachial plexus root avulsion, who had no re-innervation of former targets of the injured brachial plexus. Elucidation of the time course for that reorganisation may help establish the timing of rehabilitation training to prevent cortical remodelling.
Methods
Participants
With informed consent, 13 patients and nine normal subjects who had no disorders or injuries of the central nervous system participated in this study from October 2010 to March 2012. Patients meeting the following criteria were enrolled, that is, unilateral total brachial plexus root avulsion without any spontaneous sensory or motor recovery of the upper limb, and no surgical interventions performed before their first resting-state fMRI scans. Of the 13 patients, 11 were male and two female, with age ranging from 22 to 43 years; the injured side was on the right in five and on the left in eight cases. The cause of injury was motorcycle accident in most cases (table 1). Total brachial plexus root avulsion was diagnosed by physical examinations, neurophysiological investigations and surgical exploration. According to the Peripheral Nerve Injury scoring system of Royal National Orthopedic Hospital in Stanmore,6 neuropathic pain that started within 1 week after injuries in this series was recorded as Significant in one, Moderate in five and Mild or None in seven cases. Nine normal subjects were enrolled to demonstrate the normal pattern of cortical sensorimotor areas corresponding to the hand and arm as shown by resting-state fMRI. Of them, seven were male and two female, and the age ranged from 21 to 39 years.
Resting-state fMRI data acquisition
Resting-state fMRI was acquired for all the patients and the normal subjects. Three out of the 13 patients underwent a second test at an interval of 7–8 months after their first scans. The inclusive criteria for second scans were that there was no neurophysiological or clinical evidence of re-innervation of former motor or sensory targets of the injured brachial plexus after 3 months postoperatively. The timing of the first and second tests ranged from 1 to 16 months after injuries. All subjects were scanned using resting-state blood oxygenation level-dependent (BOLD) fMRI (3.3×3.3×4.0 mm voxels; time of echo, 35 ms; time of repetition, 2000 ms) on a Siemens Magnetom Verio 3.0T MRI scanner. Scans lasted for 8 min, comprising 240 time points per subject. During scanning, subjects kept still with eyes closed but not falling asleep, and no tasks were given. T1-weighted structural imaging (time of echo, 2.93 ms; time of repetition, 1900 ms; slice number, 176; slice thickness, 1 mm; field-of-view, 250×219 mm) was used for atlas transformation. All protocols were approved by the Ethic Committee of Fudan University.
Preprocessing of imaging data
All data were preprocessed using Statistical Parametric Mapping 8 (SPM8; Wellcome Department of Imaging Neuroscience, University College London, UK) and Data Processing Assistant for Resting-State fMRI-advanced edition (DPARSFA; by Yan and Zang, State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University, China). Standard image preprocessing was carried out as follows. First 10 volumes were discarded. Images of each subject were realigned and corrected for slice timing. The T1-weighted structural image was normalised to a Montreal Neurological Institute (MNI) template. The normalisation parameters determined for the structural volume, including those of bounding box and of voxel sizes, were then applied to each functional image volume. These images were smoothed with a Gaussian kernel of 4 mm. Linear trend over each run was removed. We applied a temporal band-pass filter (0.01 Hz<f<0.08 Hz) to the time course to obtain low-frequency fluctuations, and reduced the spurious BOLD variances that were unlikely to reflect neuronal activity. These variances included signals from the cerebrospinal fluids, from the white matter and from the whole brain, as well as the six parameters obtained by rigid body head motion correction.7 ,8
Seed-based functional connectivity
The BOLD fMRI time course was extracted from spherical seed regions of interest (ROI, 3-mm radius) and then the seed-based functional connectivity map was generated. We created the seed region from the supplementary motor area (SMA), and the coordinate of the seed region in MNI space was at x=0, y=−8 and z=58 or at x=0, y=0 and z=56.9 Correlation coefficient was then calculated between the average time course of the seed region and time courses of all other voxels in the brain. We used xjView 8 (http://www.alivelearn.net/xjview) to visualise the correlation map on the platform of Matlab. Threshold was determined according to the qualitative evaluation which showed functional neuronal activity best while minimising spatially nonspecific noise that is probably of non-neuronal origin.10
We created the seed region not only from SMA but also from the Ω-like hand area11 on both sides. The BOLD fMRI time courses were extracted from each seed region. The correlation coefficient between SMA and Ω-like hand area was calculated on both sides, and a paired t test was applied to compare the bilateral data. Statistical analysis was made using SPSS (V.13.0, Chicago, Illinois, USA), and statistical significance was set at 0.05.
Results
By placing a ROI in SMA, the correlation map of resting-state activity between that seed point and every other voxel of the brain was presented after resting-state fMRI acquisition and data processing. In the nine normal subjects, cortical maps showed symmetry of bilateral sensorimotor areas including those corresponding to the hand and arm (figure 1), which are located intermediately according to the cortical homunculus found by Wilder Penfield.12 The correlation coefficient between SMA and Ω-like hand area on both sides was not statistically different (table 2).
Cortical maps in patients No. 1–9 (table 1), who acquired their first resting-state fMRI from 1 to 4 months after brachial plexus injuries, showed asymmetry of bilateral cortical sensorimotor areas corresponding to the hand and arm, where the correlation between SMA and sensorimotor areas corresponding to the hand and arm contralateral to the side of brachial plexopathy attenuated obviously as compared with that of the other side (figure 2). The correlation coefficient between SMA and Ω-like hand area contralateral to the injured side was statistically smaller than that between SMA and Ω-like hand area ipsilateral to the injured side (table 2). Four maps in patients No. 10–13, whose resting-state fMRI was acquired more than 6 months after injuries, showed symmetry of bilateral cortical sensorimotor areas corresponding to the hand and arm. The correlation coefficient between SMA and Ω-like hand area contralateral/ipsilateral to the injured side was 0.4711/0.6053, 0.5970/0.5682, 0.4030/0.4947 and 0.5788/0.4099, respectively. Three patients (No. 3–5, table 1) were followed up by resting-state fMRI from 7 to 8 months after their first scans. For patients No. 3 and 4, asymmetric correlation maps on both sides were changed to symmetric, where the correlation coefficient between SMA and Ω-like hand area contralateral/ipsilateral to the injured side was 0.4137/0.4147 and 0.6431/0.6631, respectively. For patient No. 5, cortical sensorimotor areas corresponding to the hand and arm contralateral to the injured side had stronger relationship with SMA than those ipsilateral to the injured side, and the correlation coefficient between SMA and Ω-like hand area contralateral/ipsilateral to the injured side was 0.4671/0.2762.
Illustrative case
A 28-year-old male patient (No. 3 in table 1) presented with sensory loss and motor deficit of his right upper limb immediately after a motorcycle accident. He felt pain in his right upper limb to the extent of significant, which started immediately after injuries and made him unable to work, study and enjoy hobbies. Surgical exploration confirmed that the C5-T1 spinal nerves were avulsed from intervertebral foramina, and he underwent a nerve reconstruction operation of extraplexal nerve transfer. Preoperatively, resting-state fMRI was obtained at 2 months after injuries. A spherical ROI was seeded at SMA to generate the correlation map, where cortical sensorimotor areas corresponding to the hand and arm on the left side could not be revealed, but those on the right were connected functionally with SMA (images of No. 3 in figure 2). The correlation coefficient between SMA and left cortical sensorimotor areas corresponding to the hand and arm was 0.1047, while that between SMA and right cortical sensorimotor areas corresponding to the hand and arm was 0.3819 (table 2). We also seeded a spherical ROI at the left Ω-like hand area but could not get any functional connectivity map on other regions of the whole brain (figure 3). This patient was followed up by resting-state fMRI at 7 months after operation, at which no motor and sensory improvement of the hand and arm was found neurophysiologically or clinically. Bilateral cortical sensorimotor areas corresponding to the hand and arm were then found to be symmetric, and the area correlated with the SMA on the left side increased as compared with that shown on the first scan (figure 4). The correlation coefficient between SMA and left cortical sensorimotor areas corresponding to the hand and arm was 0.4137, while that between SMA and right cortical sensorimotor areas corresponding to the hand and arm was 0.4147.
Discussion
As a non-invasive method, fMRI has been applied to the study of brain plasticity over the past several decades. The work in rodents13 and dogs14 by using contrast agents was the first step in the emergence of fMRI. This technique was extended to functional mapping for task activation in humans in 1991.15 By a longitudinal task-based fMRI study of patients with nerve reconstruction operations following brachial plexus injuries, Yoshikawa indicated that improvement of motor function could induce more activation in the sensorimotor cortex contralateral to the injured side, as compared with their preoperative status.16 Spontaneous BOLD fluctuations were identified by Biswal,17 and Fox18 applied seed-based functional connectivity to analyse these resting-state fMRI data. One of the important advantages of resting-state fMRI is that it can be acquired without the need for actively participating in a task. Strong coherence is reproducibly present between SMA and bilateral sensorimotor cortices,17 ,19 and this was also confirmed by nine normal subjects in our study. Until recently, however, little is known about brain plasticity from fMRI changes for patients who are unable to accomplish functional tasks due to peripheral nerve injury, although evidence has been presented in rats that surgical forelimb denervation causes disruption in the correlations of BOLD low-frequency fluctuations between areas of sensorimotor cortices.20 In this study, cortical changes were observed in phase-one reorganisation when no former targets of the injured brachial plexus were re-innervated so that no movements (tasks) could be performed, and this was the reason why resting-state fMRI, rather than task-based, should be applied.
In the present study, results of nine patients (No. 1–9 in table 1) whose first resting-state fMRI was performed at 1–4 months after injuries showed that their cortical sensorimotor areas corresponding to the hand and arm in the hemisphere contralateral to the injured side had much weaker correlation with SMA than those ipsilateral to the injured side (table 2). When a ROI was seeded at the Ω-like hand area in the hemisphere contralateral to the injured side, we could not get any functional connectivity map on other regions of the whole brain (figure 3). These facts meant that the corresponding cortical representations of the brachial plexus were functionless. When resting-state fMRI was repeated for three (No. 3–5) of them 9–10 months after injuries, at which no improvement of former motor or sensory targets of the injured brachial plexus was detected neurophysiologically or clinically, bilateral cortical maps of sensorimotor areas corresponding to the hand and arm were found to be symmetric, or even stronger correlation was shown between cortical sensorimotor areas corresponding to the hand and arm contralateral to the injured side and SMA. This suggested that there was a functional remodelling in cortical representations of the injured brachial plexus for the three patients, which resulted apparently from expansion of neighbouring cortical areas. Accordingly, we suggest that the symmetry of bilateral cortical sensorimotor areas corresponding to the hand and arm in the other four cases (No. 10–13) with their resting-state fMRI done from 7 to 16 months after injuries resulted from the cortical remodelling of the representations of the injured brachial plexus by extension of adjacent cortical territories.
Some researches have shown that the prevention of functional cortical reorganisation resulting from expansion of adjacent cortical territories might improve the efficacy of peripheral nerve repair and relieve pain. Lundborg5 showed that cutaneous forearm anaesthesia repeated with a local anaesthetic agent resulted in much better recovery of tactile gnosis than that with placebo in patients with median or ulnar nerve injury and repair at the wrist level. Using task-based fMRI scans for patients with upper limb amputation unilaterally, Lotze21 indicated that wearing a myoelectric prosthesis frequently could refrain from shrinking of corresponding cortical sensorimotor areas of the hand and arm, as well as further expansion of non-deafferented areas; also, the maintenance of the original cortical representations was negatively correlated with phantom limb pain. Our data indicated that after total brachial plexus root avulsion, cortical sensorimotor areas corresponding to the hand and arm could be functionless for a period of 3 or 4 months, after which functional remodelling by occupation of nearby cortical regions might begin. This implies that for patients with severe brachial plexus injuries, the rehabilitation training should commence within 4 months of the injury before cortical reorganisation takes place.
Deafferented pain is usually a serious problem in patients with total brachial plexus root avulsion.6 Studies of humans with upper limb amputation showed that painful phantom sensations sometimes arose from areas with cortical reorganisation, the extent of which correlated with the amount of pain.4 For cases No. 3 and 4 (table 1) in this series, who had Significant or Moderate pain starting immediately after injuries, silent cortical sensorimotor areas corresponding to the hand and arm on the uninjured side were shown by resting-state fMRI performed at 1 and 2 months, respectively, while for cases No. 1, 2, 5–7, 12 and 13 with Mild or None pain, the symmetry of bilateral cortical maps corresponding to the hand and arm was found. It, therefore, did not seem to indicate any correlation between pain and cortical reorganisation in this study. A further study of more patients with severe pain is required to clarify if deafferented pain correlates with cortical reorganisation in patients with total brachial plexus root avulsion.
Conclusions
From resting-state fMRI scans it was shown that total brachial plexus root avulsion results in dynamic changes in cortical sensorimotor areas corresponding to the hand and arm, which are transferred from the inactive to the active state. It implies that peripheral deafferentation resulting from brachial plexus injuries will induce corresponding cortical representations to be occupied by adjacent non-deafferented cortical territories. A further study is needed to clarify whether prevention of cortical reorganisation may improve the efficacy of nerve reconstruction operations after brachial plexus injuries.
References
Footnotes
-
Contributors Study concept and design: LC and YM; acquisition of data: T-MQ, LC, J-SW, W-JT and S-NH; analysis and interpretation of data: T-MQ, LC and YM; drafting of the manuscript: T-MQ and LC; critical revision of the manuscript for important intellectual content: Y-DG and L-FZ.
-
Competing interests None.
-
Patient consent Obtained.
-
Ethics approval The Ethics Committee of Huashan Hospital, Fudan University.
-
Provenance and peer review Not commissioned; externally peer reviewed.