Brain, Vol. 125, No. 5, 1150-1161,
May 2002
© 2002 Guarantors of Brain
Interlimb reflexes and synaptic plasticity become evident months after human spinal cord injury
1 The Miami Project to Cure Paralysis and Department of Neurological Surgery, University of Miami School of Medicine, Miami, FL 33136, USA
Correspondence to: Blair Calancie, PhD, Department of Neurosurgery, SUNY’s Upstate Medical University, 750 East Adams Street, IHP Room 1213, Syracuse, NY 13210, USA E-mail: calancib{at}upstate.edu
Received September 13, 2000. Revised December 13, 2001. Accepted January 9, 2002.
 |
Summary |
|---|
 Top Summary IntroductionMethodsResultsDiscussionReferences |
|---|
Persons with long-standing injury to the cervical spinal cord resulting in complete or partial paralysis typically develop a wide spectrum of involuntary movements in muscles receiving innervation caudal to the level of injury. We have previously shown that these movements include brief and discrete contraction of muscles in the hand and forearm in response to innocuous sensory stimulation to the feet and legs, but we have been unable to replicate these interlimb reflexes in able- bodied subjects. Properties of these muscle responses indicate that the synaptic contacts between ascending sensory fibres and motor neurones of the cervical enlargement are more efficacious than normal. If these connections are present at all times, and require the more rostrally-placed spinal cord injury to allow their emergence, one might expect their appearance relatively soon following injury, as has been shown for studies of ‘latent’ synapses. Conversely, delayed appearance of these interlimb reflexes would suggest either the development of new synaptic connections or a profound strengthening of existing circuits in the cervical spinal cord due to a combination of afferent target loss and motor neurone denervation from motor tracts originating rostral to the injury site. In this study, we used repeated examinations of persons with acute injury to the cervical spinal cord to examine the time post-injury at which interlimb reflexes are first seen. Using tibial nerve stimulation at the knee as a screening test, a total of 24 subjects were found to develop interlimb reflexes following spinal cord injury. Latencies between stimulation and EMG were as brief as 32 ms for muscles of the forearm and 44 ms for muscles in the hand. These minimal delays all but rule out a supraspinal route for these interlimb reflexes. Interlimb reflexes first became evident no sooner than
6 months following injury, and in some individuals were not seen until well over 1 year post-injury. Enhanced lower limb segmental excitability had emerged in nearly all of these subjects weeks or months prior to the first appearance of interlimb reflexes, arguing against a manifestation of traditional post-traumatic spasticity as a basis for this activity. This prolonged delay between time of injury and emergence of interlimb reflex activity lends support to the hypothesis that this activity represents an example of plasticity—and perhaps ‘regenerative sprouting’—in the human spinal cord following traumatic injury. Keywords: spinal cord injury; plasticity; regenerative sprouting; reflex; human; cervical
Abbreviations: ADM= hypothenar group of the hand; APB = thenar group of the hand; ASIA = American Spinal Injury Association; ECR = wrist extensors; FCR = wrist flexors; ILR = interlimb reflexes; Psoas = hip flexors; SCI = spinal cord injury
|
Introduction |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
There are numerous examples of axonal regeneration and synaptic reorganization of neurones (collectively referred to as ‘plasticity’) following traumatic lesions to the mammalian spinal cord (for a review, see Guth, 1974
; Steward, 1989; Goldberger et al., 1993; Schwab and Bartholdi, 1996; Mendell et al., 2001; Siddall and Loeser, 2001; Wolpaw and Tennissen, 2001). In the absence of specific interventions, such plasticity may contribute to the development of abnormal movement states (McCouch et al., 1958; Nelson and Mendell, 1979; Hiersemenzel et al., 2000) and/or sensory disturbances (Christensen and Hulsebosch, 1997; Romero et al., 2000). Alternatively, certain post-injury alterations of behaviour may reflect the collective action of synaptic connections that were present at the time of injury, but were functionally inactive (or ‘silent’). These latent connections have been reported to become ‘unmasked’ within minutes to hours of a CNS lesion (Goshgarian and Guth, 1977; Nelson et al., 1979; Wall, 1988; Goshgarian et al., 1989; but compare with Brown et al., 1984). Thus the timecourse with which certain spinal cord input/output properties emerge following CNS trauma may provide clues as to the mechanism(s) underlying those behaviours.
Many of the same mechanisms of plasticity reported from animal models of spinal cord injury (SCI) can likely be demonstrated in human subjects, but quantitative histologic data supporting the possibility of sprouting or synaptogenesis are indirect (Krassioukov et al., 1999). Behavioural studies abound though and raise the possibility for both spontaneous (Calancie et al., 1994) and task-specific (Bach and Rita, 1981; Wernig et al., 1995; Harkema et al., 1997; Barbeau et al., 1999) plasticity in the human spinal cord caudal to an injury.
Previous reports from this laboratory (Calancie, 1991; Calancie et al., 1996) have described novel ‘interlimb reflexes’ (ILR) in persons who have sustained SCI >1 year prior to study (i.e. in the chronic phase). These involuntary movements are characterized by short-latency (i.e. 40–50 ms) contractions of hand and forearm muscles following a wide variety of innocuous sensory stimuli delivered to the lower limb or limbs (including skin stroking, hair pull, tendon taps and electrical stimulation of peripheral nerves). We have suggested that such interlimb reflexes may reflect the consequences of novel synaptic connections formed between ascending first- and second-order afferent fibres and motor neurones of the cervical enlargement partially denervated due to a more rostrally-placed lesion to the spinal cord (Calancie et al., 1996). If correct, one would expect to see a significant delay between the time of injury and the time at which such ILRs become evident.
In this paper, we report findings from a group of subjects who ultimately developed interlimb reflexes after sustaining traumatic spinal cord injury. We conducted repeated measures on these subjects over a period of many months to determine the time after injury when these reflexes became evident. Our data are consistent with the hypothesis that these interlimb reflexes represent the establishment of new synaptic connections between nerve populations that do not normally interact. We suggest that ILR emergence serves as an example of CNS plasticity (or ‘regenerative sprouting’; Steward, 1989) in the adult human nervous system following traumatic injury. This conclusion is included within a much broader examination of plasticity after human SCI that was presented previously (Calancie et al., 2000).
|
Methods |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Subjects
Experiments were performed on persons with traumatic injury to the cervical spine resulting in neurologic deficit (i.e. spinal cord injury). In the majority of cases, the initial examination of a given subject took place within the first week after injury, with follow-up studies continuing over the next weeks and months post-injury. All subjects gave their informed consent to participate in this protocol, which was approved by the University of Miami’s Institutional Review Board.
Procedures
Self-adhesive surface EMG electrodes (S’Offset; Graphic Controls Corp., Buffalo NY, USA) were positioned over the biceps brachii, triceps brachii, wrist extensors (ECR), wrist flexors (FCR), thenar group of the hand (APB), hypothenar group of the hand (ADM), hip flexors (Psoas), quadriceps, hamstring, tibialis anterior (TA), soleus, and foot intrinsics of the subject’s left side. Examinations were made of voluntary individual muscle contractions, tendon reflexes and central motor conduction in response to transcranial magnetic stimulation, and interlimb reflexes through surface-applied electrical stimulation of the tibial nerve at the popliteal fossa.
Tibial stimulation was accomplished using either a Grass S88 stimulator (initial four subjects; Grass Instrument Co, Quincy, MA, USA) or a Digitimer D185 stimulator (remaining 20 subjects; Digitimer Inc, Welwyn Garden City, UK). Stimuli were delivered through pairs of self-adhesive surface electrodes (Cleartrace; ConMed Corp., Utica, NY, USA) positioned over the tibial nerve (cathode) and a site medial and distal to this site (anode). Single pulses of 
100 V were used to define the optimal stimulus site, using soleus direct muscle response (M-wave) and plantar-going ankle movement as the criteria to guide stimulation. In most cases, pressure was applied to the cathode while stimuli were delivered, pushing the electrode closer to the underlying tibial nerve in order to minimize subject discomfort (in those subjects who could feel the stimulus) while still eliciting a strong plantar-flexion. The stimulus intensity delivered via the D185 stimulator [based on readings from the stimulator’s liquid crystal display (LCD)] was not less than 125 mA in any subject tested, while trials using the Grass stimulator routinely used pulses of 150 V (the maximum capable with this device). The duration of individual stimuli within a 3-pulse train was 50 µs and 1000 µs for the D185 and S88 stimulators, respectively. (Note that the D185 pulse width cannot be adjusted from this 50 µs value, but that its maximum stimulus intensity far exceeds that of the S88, enabling the D185 output to produce strong plantar-going twitches when desired.)
Once an acceptable stimulation site was established, a series of 3-pulse stimulus trains was delivered, each pulse in the train separated from the next by 2 ms (i.e. three pulses at 2 ms each; ‘3 @ 2’). A 4-pulse train was used on occasion. Pulse trains were separated from one another by a minimum of 1 s. In most cases, the rate of pulse train stimulation was approximately 0.2 Hz, and was controlled manually (i.e. a deliberate button push was needed to trigger a stimulus). No fewer than 10 stimulus trains were delivered, and the EMG responses from the six upper limb muscles were monitored via computer (Toshiba T6400 and RC Electronics Computerscope; RC Electronics Inc, Goleta, CA, USA) and stored on digital tape [Vetter 4000a (defunct) or MicroData Instruments DT1600; MicroData Instrument Inc, South Plainfield, NJ, USA]. After collecting left-side responses to left-side stimulation, stimuli were applied to the right-side tibial nerve while still recording from left-side muscles (and the right-side soleus to confirm response properties). Additional evaluations were carried out, all electrodes were shifted to the comparable right-side muscles, and the measures repeated.
Data were analysed off-line (Cambridge Electronic Design 1401; Cambridge Electronic Design, Cambridge, UK) from the taped records. EMG from each of the upper-limb muscles being studied was offset to zero, rectified and averaged for a 200 ms post-stimulus time period. Minimum response latencies were determined relative to the onset of the first stimulus in the pulse train. Responses were considered positive if: (i) the response at threshold stimulus intensity was lost if a 2-pulse train or single stimulus was applied; and (ii) the response latency was 
110 ms (more prolonged latencies can be due to input from supraspinal elements, a demonstration of which follows). In some cases of well-defined single motor unit (SMU) discharge, peri-stimulus time histograms were produced to summarize that motor unit’s discharge properties across all stimuli. Template matching was used to define the motor unit’s discharge latency for each stimulus, the results of which were grouped into 1 ms bins.
Muscle interference patterns were graded on a 6-point scale, for which ‘0’ represents a total absence of volitional recruitment and ‘5’ represents normal recruitment (this EMG scale is fully described elsewhere by Alexeeva et al., 1997; Calancie et al., 2001). Subjects who could—on command—recruit and silence EMG in at least one lower limb muscle during attempted voluntary contractions and relaxations of that muscle were termed ‘motor-incomplete’ for the purposes of this study.
|
Results |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
A total of 24 subjects in our series, all male, were found to develop interlimb reflexes after acute spinal cord injury. Table 1 shows the cause of injury and the most rostral level of fracture for each subject, along with—at the time ILR activity was first noted—each subject’s age, the most rostrally-innervated upper limb muscle from which interlimb reflexes were seen, and the subject’s clinical status as defined by American Spinal Injury Association (ASIA) nomenclature (Ditunno et al., 1994
). Nineteen subjects were first examined within 10 days of their traumatic injury. Initial examinations on subject numbers 1, 11, 12, 22 and 23 were conducted 21, 25, 18, 30 and 42 days after their respective injury dates. These delays were due to transfer from another hospital (n = 4) or an inability to consent due to sedation (n = 1). We attempted to perform follow-up examinations at least five times during the first year after injury, but this was not always possible (due to subject discharge, transfer to another facility, moving out of town, incarceration, etc.). As few as three and as many as 10 examinations for ILR activity were performed on each subject in our population before such responses were first seen (mean ± standard deviation: 6.0 ± 1.8).
|
Original and processed data records, which reflect several elements of interlimb reflexes consistent with our experience, are illustrated in Figs 1 and 2, both taken from Subject 4. Fig. 1 shows EMG (superimposed) from 11 left-side muscles (and right-side soleus) in response to right-side tibial nerve electrical stimulation with a 3-pulse train, at a slow (0.3 Hz; Fig. 1A) and a somewhat faster (0.8 Hz; Fig. 1B) average presentation rate. Each stimulus train resulted in short-latency single motor unit recruitment in each of the (contralateral) biceps and (contralateral) ADM muscles. Note that what appear to be EMG responses in the triceps muscle were probably caused by discharge in the biceps motor unit, due to volume conduction (see Calancie et al., 2001
). The post-stimulus time histograms of the biceps and ADM motor units are shown at the bottom of Fig. 1. Contraction of the biceps tendon of insertion and movement of the intrinsic muscles adjacent to the fifth metacarpal was clearly visible with each stimulus delivery, but no other left- or right-side upper limb muscle contraction was evident in this subject in response to right-side tibial nerve stimulation.
|
|
Of the biceps brachii and ADM motor unit records illustrated in Fig. 1A, the ADM motor unit had the shorter response latency, despite the longer conduction distances involved. The response latency of the biceps motor unit showed considerably more variability from trial to trial, particularly for the slow stimulus rate. A modest increase in the average rate of stimulation from 0.3 trains/s (Fig. 1A) to 0.8 trains/s (Fig. 1B) led to a more consistent response latency of this biceps brachii motor unit, at a considerably shorter latency (as early as 55 ms at the higher stimulus rate) compared with its earliest latency of 60 ms at a slower rate of stimulation. Similar findings have been reported previously in persons with chronic SCI (Calancie, 1991
). This effect of increasing stimulus rate was not readily apparent in the initial response histogram for the ADM motor unit, but discharge at more prolonged latencies (90 ms) following the stimulus was more tightly grouped for the faster rate of stimulation in this muscle. Note that close inspection of individual stimulus-evoked responses for this ADM motor unit could not rule out the possibility that a second motor unit of nearly identical waveform shape contributed to the later periods (>80 ms) of discharge in the histograms shown in Fig. 1A and B. Figure 2 illustrates the behaviour of a single motor unit from the right-side ADM muscle of Subject 4 in response to a 3 @ 2 (Fig. 2A) or 4 @ 2 (Fig. 2B) stimulus pattern to the left (i.e. contralateral) tibial nerve, using a stimulus intensity of 200 mA. Because the rate of stimulation can affect a motor unit’s response probability and latency (as shown in Fig. 1), we restricted the trials giving rise to Fig. 2 to stimulus train intervals of 2.76–3.29 s. Using this approach, the mean stimulus train delivery rate was 0.32 Hz for the 3-pulse train (n = 12), and 0.34 Hz for the 4-pulse train (n = 15). Despite these almost identical rates of train delivery, the response latency onset for the ADM motor unit was clearly earlier for the 4-pulse train (55.7 ms) compared with the 3-pulse train (60.1 ms).
To compare their minimum reaction latency to latencies of the interlimb reflexes reported here, we asked some of our subjects with partial sensation in their lower limbs to react with a forceful voluntary extension of their wrist as soon as they felt the electrical stimulus at the back of the knee. For these trials, the audio for the EMG records was turned off and subjects were asked to close their eyes in order to eliminate auditory and visual cues of the stimulation. Figure 3 illustrates an example of this minimal reaction time in Subject 8 showing averaged, rectified EMG from multiple right-side muscles in response to a 3 @ 2 stimulus train of 180 V (216 mA) delivered to the left tibial nerve. In Figure 3A, the subject was instructed to remain as relaxed as possible throughout delivery of the 25 stimulus trains used. In Fig. 3B, the subject was asked to extend his wrist (activating primarily ECR) in response to each of the five stimulus trains. The minimum onset latencies of the interlimb reflex (53 ms) and volitional components (115 ms) of the EMG responses are indicated with arrows. For all subjects tested in this manner, none could initiate clearly defined wrist extensor EMG at latencies of <105 ms in response to tibial nerve stimulation; most showed minimum onset latencies of 115–130 ms, well beyond the latencies of the ILR activity we typically observed.
|
A chief goal of the present study was to investigate the time at which ILR activity became evident relative to each subject’s time of injury. Figure 4 summarizes our findings, showing for each subject the various examination times after injury (in years) at which ILR activity was looked for and not seen (interlimb reflexes absent; open circles) and eventually looked for and seen for the first time (interlimb reflexes present; filled circles). Results have been sorted from shortest to longest post-injury period when ILR activity was first observed. The subject numbers correspond to those used in Table 1. Persons with a motor-incomplete injury (ASIA category C or D) at the time at which ILR activity was first seen are indicated with asterisks.
|
The earliest time after which ILR activity could be evoked by tibial nerve stimulation in this population was 5 months, 3 weeks post-injury (Subjects 1 and 2). For all subjects, the average duration between date of injury and date of the first positive ILR observation was 447 ± 221 days. This number must be viewed with caution, however, since we were unable to carry out follow-up evaluations at the same time post-injury across all subjects. For the same reason, we cannot be nearly as certain as to how long post-injury ILR activity was not seen, due again to the widespread variability in the timing of follow-up examinations in this population. In most cases, subjects were discharged from our hospital by 2 months post-injury, and for many of them securing transportation back to our centre for follow-up studies on a regular basis was problematic. Despite these difficulties in the timing of follow-ups, ILR activity was looked for, and found to be absent, as late as 11 months post-injury in seven of the 24 subjects included in this study. The longest delay between injury and confirmed absence of ILR activity was 1.6 years (Subject 23).
Did the time post-injury at which ILR activity became evident correspond with the timing of the development of hyperreflexia and/or spasticity in these subjects? The answer is no; in most subjects the emergence of ILR activity occurred months after the recovery of spinal reflexes, based upon the following observations. First, all subjects who were grouped in the motor-incomplete cohort had visible, brisk responses to taps at the Achilles and/or patellar tendons from the very first examination onwards. Despite maintenance of spinal reflex excitability early after their spinal injury, ILR activity was not evident at this time in this cohort. Secondly, persons in both groups (motor-complete and motor-incomplete SCI) showed large increases in mean response amplitudes to taps at the Achilles and patellar tendons before the emergence of ILR activity. These data are summarized in Table 2, which shows the average maximum peak-to-peak amplitudes of EMG responses to Achilles and patellar tendon taps at the initial evaluation, and at the most recent evaluation prior to that in which ILR activity was first seen. Even before the time when ILR activity was first observed, most subjects in both groups showed well-defined responses to taps at both the patellar and Achilles tendons. Also, 18 of the 24 subjects studied had already been started on anti-spasmodic medication (typically baclofen) at this time (i.e. prior to emergence of interlimb reflexes), suggesting that spasticity had already been noted by their treating physicians.
|
Table 3 summarizes the distribution of muscles from which ILR activity was observed following tibial nerve stimulation in our subject population, and whether or not the site of stimulation was on the same side as the site of evoked muscle response or contralateral to the recruited muscle. Two trends are evident from Table 3. First, responses tend to be contralateral to the side stimulated [35 out of 49 instances (71%)]. Secondly, ILR activity was far more commonly seen in the intrinsic muscles of the hand, particularly the ADM muscle, than more proximal muscles in this subject population. In fact, only one case of response in the biceps brachii muscle to tibial nerve stimulation was noted in this series.
|
The evoked response latencies (onset) to tibial nerve stimulation are shown for the different upper limb muscles in Table 4. In addition to the average (and standard deviation) latency by muscle, the earliest and most prolonged latencies seen in a given subject are also shown. Two conclusions can be drawn from these data. First, the averaged response latency increases with conduction distance from the cervical spinal cord (i.e. distal hand muscle latencies are, on average, more prolonged than those for muscles in the forearm). Secondly, there is a considerable range of minimum response latencies in a given muscle (excepting biceps brachii, of course, in which only one positive response to tibial nerve stimulation was seen in the 24 subjects examined). Given this range of latencies, it is clear that the situation depicted in Fig. 1, in which the latency to response in the ADM unit shown was less than that in the biceps motor unit, is not totally unexpected.
|
Based on a 6-point EMG recruitment scale (Calancie et al., 2001
), upper limb muscles in which ILR responses were seen tended to be weaker (mean recruitment score = 1.9 ± 2.3) than those in which ILR activity was not seen (2.7 ± 3.9); this difference was statistically significant. There were no cases in which a muscle with normal EMG interference pattern (i.e. a score of ‘5’) demonstrated ILR activity. Furthermore, there were no instances in which a single motor unit under volitional control (i.e. it could be recruited and silenced at will, or its discharge rate could be modulated if spontaneously active) was also responsive to tibial nerve stimulation. |
Discussion |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
In previous studies, we showed that interlimb reflexes are commonly seen in persons with chronic cervical spinal cord injury (Calancie, 1991
; Calancie et al., 1996). In the present study, we now show that: (i) these reflexes are also evident in persons with motor-incomplete injury, contradicting one conclusion from an earlier study (Calancie 1991); and (ii) these reflexes are not evident at the acute stage, but take many months to emerge following spinal cord injury.
Origins of interlimb reflexes
We suggest that ILR activity is mediated by sprouting from ascending afferent fibres onto motor neurones of a particular upper limb muscle partially denervated following a more rostrally-placed spinal lesion. Either new contacts between afferents and these motor neurones are made, or existing contacts are dramatically strengthened through the sprouting process. While absolute proof of such sprouting is lacking, we believe that the aggregate of evidence summarized below points to this as the most probable explanation.
Distribution of interlimb reflexes
The nature and extent of ILR activity was often restricted to a few individual motor units recruited in one or several upper limb muscles. These singular activations, with highly reproducible latency, commonly resulted in visible twitches in the muscle belly or the joint being acted upon by the muscle. This form of recruitment is very different from that typically seen in response to synchronized peripheral (i.e. Ia) or central (i.e. corticospinal tract) excitatory inputs (Jones et al., 1996), whereby excitation is likely distributed widely throughout the motor neurone pool. Functional movements of the hand or forearm were never observed.
As shown in Table 1, upper limb muscles which responded to tibial nerve stimulation always derived their primary level of innervation from neurologic levels caudal to the rostral-most level of fracture. For example, the single instance of ILR activity observed in the biceps brachii muscle was from a subject with a C3 fracture (Subject 4; this subject required ventilator support for 3 weeks following his injury). Because the biceps brachii receives its primary innervation from the fifth and sixth neurological levels of the spinal cord (Kendall and McCreary, 1983), motor neurones to this muscle would be expected to be at least partially denervated by the more cranially-placed spinal injury.
Based on our findings, the ability to contract an upper limb muscle through volitional effort does not preclude the presence of an ILR response to lower limb inputs in that same muscle. We saw a number of cases in individuals with both complete and incomplete SCI where an EMG interference pattern with voluntary effort could be produced in the same muscle from which ILR activity was seen. However, muscles with positive ILR responses to lower limb innocuous stimulation were invariably weaker during attempted voluntary contractions. This suggests that a greater percentage of motor neurones to such muscles had lost innervation from axons originating cranial to the lesion, rendering their motor neurones more susceptible to regenerative sprouting from lower limb ascending first- or second-order afferent fibres. Indirectly supporting this suggestion, we did not see in this study—nor in other subjects with chronic SCI studied at other times—a single instance in which the same motor unit recruited by voluntary contraction was also activated by tibial nerve stimulation or other types of lower limb sensory inputs (Calancie, 1991; Calancie et al., 1996).
These observations argue that those upper limb motor units being recruited through tibial nerve stimulation lie towards the caudal range of the motor neurone pool for that muscle. There is abundant evidence (albeit from lower- or hind-limb studies) supporting the concept that the motor neurones innervating a given muscle are distributed in a rostral-caudal orientation spanning three or more neurologic segments (Henneman and Mendell, 1981; Ungar-Sargon and Goldberger, 1987; Phillips and Park, 1991). In the human upper limb, Kendall and McCreary (1983) have compiled findings from six groups indicating that in man, there is also widespread spinal segmental distribution to the six upper limb muscles studied herein as follows: biceps: C5–C7; triceps: C6–T1; ECR: C5–C8; FCR: C6–C8; APB: C6–T1; ADM: C7–T1.
Given the requirement for a distribution of motor neurones in a given muscle’s pool to account for both ILR and voluntary activity, this explanation also requires that the longitudinal extent of spinal cord injury be relatively circumscribed, preserving substantial grey matter caudal to the injury epicentre (at least for relatively mild injuries leading to ASIA categories C or D). Limited evidence from human histopathology studies indeed shows that traumatic injuries are often restricted in a longitudinal (i.e. cranial–caudal) extent to distances of no more than 3–4 cm within the spinal cord (Bedbrook, 1963; Bunge et al., 1993, 1997).
For sprouting to occur onto or expand across motor neurones of the cervical enlargement, somatic and/or dendritic surface area must presumably be vacated due to degeneration of synaptic contacts originating from nerve fibres axotomized at a point more rostral in the neuraxis (i.e. the site of spinal cord injury), and new synaptic contacts formed (Raisman and Field, 1973). This partial denervation is most likely to occur with motor neurones innervating muscles of the forearm and hand, whose source of motor innervation is highly dependent upon the corticospinal tract (Phillips and Porter, 1977; Kuypers, 1981; Palmer and Ashby, 1992; Maier et al., 1998) or cervical propriospinal neurones subserving fractionated movements (Pierrot-Deseilligny, 1996). Conversely, proximal upper limb muscles, such as biceps and triceps, would be less likely to undergo denervation due to the lesser dependence of their motor neurones upon corticospinal influence (either directly or via propriospinal neurones) and greater input from other pathways (Turton and Lemon, 1999), which appear to rely to a greater extent upon segmental interneurones rather than direct supraspinal projections (Lawrence and Kuypers, 1968). The present findings agree with both predictions. That is, ILR activity is most commonly seen in hand intrinsic muscles, is less common in forearm muscles and is only rarely encountered in proximal muscles of the upper limb, even in cases in which these proximal muscles receive innervation from motor neurones located within the caudal-most portions of the cervical enlargement, such as the triceps brachii (in which we have never observed ILR activity).
Time of interlimb reflex emergence post-injury
Does the time course of ILR emergence following spinal cord injury support the argument that regenerative sprouting is the most likely explanation for the ILR activity described herein? We believe the answer is ‘yes’, for the following reasons.
First, many months were needed before ILR activity was seen, whereas the time at which ‘latent’ synapses have been demonstrated following nerve injury in various models ranges from several minutes to several days (Nelson et al., 1979; Wall, 1988; Goshgarian et al., 1989; Brasil-Neto et al., 1992). Note also that 50% of the subjects had spinal reflexes that were brisk and well-defined immediately following injury (i.e. those with motor-incomplete SCI), yet these subjects had no evidence of ILR activity for many months beyond this time. The majority of the remaining, motor-complete subjects had recovered spinal reflex excitability even before the experimental session in which ILR activity was first noted. Thus, one cannot simply invoke ‘spinal shock’ as a basis for the absent interlimb reflexes until six or more months post-injury.
Secondly, the rate of cellular degeneration following human SCI is considerably slower than that associated with the rat based on histopathologic findings (Becerra et al., 1995; Bunge et al., 1997). This protracted period of cellular breakdown would thereby delay the availability of new synaptic sites on affected motor neurones.
Thirdly, nerve growth factor associated sprouting in human axons has been reported following traumatic spinal cord injury (Wang et al., 1996). In a different study, ascending afferent fibres within the dorsal columns were found to persist just below the level of a cervical injury and did not retract from the injury locus (Quencer and Bunge, 1996). Sprouting from these ascending sensory fibres may account not only for the ILR activity reported herein, but also for the development of autonomic dysreflexia in this population over a period of months following SCI (Mathias et al., 1979; Weaver et al., 1997; Krenz et al., 1999; Krassioukov et al., 1999; Teasell et al., 2000). We are currently pursuing this line of investigation.
Finally, of the more than 70 subjects with chronic, severe injury to the cervical spinal cord examined for ILR activity in our laboratory, only four have not shown such reflex behaviour when examined appropriately. We believe the absence of ILR activity in these four subjects is due to a widespread loss of spinal cord grey matter and resultant denervation of hand and forearm muscles. We have confirmed this in three of these subjects by demonstrating a complete absence of intrinsic hand muscle contraction following intense (>50 mA) electrical stimulation of median and ulnar nerves at the wrists bilaterally (results not shown). In our experience, such extensive grey matter loss after cervical SCI is rare (see Peckham et al., 1976).
Latency of interlimb reflexes
In records presented in this paper and elsewhere (Calancie, 1991; Calancie et al., 1996) the ILR onset latency to tibial nerve stimulation (at the knee) can be as short as 50 ms or less, when measured at the intrinsic muscles of the hand (note that this latency is relative to the onset of the first stimulus pulse of the brief train used). The conduction path for this activity in a subject whose height was 6 feet (as was true for Subject 4) would be approximately 2 metres in length. Of course, there must be a minimum of two synaptic delays in this path prior to the onset of EMG, giving rise to a calculated conduction velocity of the fastest fibre(s) mediating this activity of no less than 42 m/s. We use the term ‘no less’ because a high frequency 3- or 4-pulse train is often necessary to elicit interlimb reflexes, suggesting that the evoked response is elicited through the effects of temporal summation. Were latencies to be counted only from the onset of the penultimate pulse, the calculated maximum axonal conduction velocity from the example above would be at least 48 m/s.
Based on the brief response latencies shown in this study, it seems clear that the initial components of interlimb reflexes are mediated by spinal-only pathways, consistent with the hypothesis that interlimb reflexes reflect altered spinal circuitry. Figure 4 shows that while ILR activity at 50–60 ms is likely entirely of spinal origin, activity beyond a latency of 110 ms conceivably includes both spinal (relatively slowly-conducting) and supraspinal activity.
Many of the properties of interlimb reflexes suggest that they are mediated, at least in part, by second-order cells with widespread convergence from primary afferents. We showed in an earlier study that ILR response properties to repeated stimulation can show ‘wind-up’ (Mendell, 1966; Calancie et al., 1996); that is, enhanced output (intensity of discharge) for a given input. We interpret our findings that modest changes in the stimulus rate can lead to consistent reduction in ILR response latency (as shown in Fig. 1) as being consistent with recruitment of more rapidly conducting second-order afferents reflecting this state of heightened response probability underlying wind-up. Reliance upon these fibres and ascending routes also helps explain the wide range of response latencies for different muscles as reported in Table 4. While the fastest conducting afferent and efferent limbs must be responsible for the shortest latency responses seen, a combination of slowing conducting afferents (primary and secondary) and relatively slowly-conducting motor axons could easily give rise to ILR latencies of 100 ms and beyond.
We also show that the addition of extra pulses in a train can actually reduce the response latency of ILR activity. We typically used an electrical stimulus to the tibial nerve of sufficient intensity to cause a clear plantar-going twitch, such that we were likely stimulating a large number of nerve fibres (both efferent and primary afferents). The 4-pulse train in the example shown (Fig. 2) caused recruitment at 56 ms. We suggest that afferents of relatively rapid conduction velocity mediated this response, and that the fourth pulse in the stimulus train actually caused the second-order cell (or cells) to reach threshold and discharge an action potential. By eliminating one of these inputs (i.e. using a 3-pulse instead of a 4-pulse train), activity in these more rapidly conducting afferents could contribute to the second-order cell’s depolarization, but additional effects of more slowly conducting afferents were necessary to actually reach threshold, accounting for the additional latency (4.5 ms in Fig. 2).
Other studies
There have been other studies describing the influence on upper limb musculature of lower limb sensory inputs (Gassel and Ott, 1973; Kearney and Chan, 1979, 1981; Delwaide and Crenna, 1984). Several authors have shown that stimuli similar in principal to those used in the present study can cause modulation of upper limb reflex excitability at latencies that suggest spinal action in normal subjects (Kearney and Chan, 1979; Delwaide and Crenna, 1984). However, in these studies, the magnitude of effect is sub-threshold for direct observation, requiring a conditioning test paradigm to resolve; this is in sharp contrast to the ease with which ILR activity can be elicited and directly observed in persons with a history of cervical SCI. Moreover, the latencies of effects reported in these other studies are well beyond those seen in the present study once the conduction delay associated with the efferent limb of the motor circuit is taken into consideration. At the very least, the ILR activity reported in this paper may reflect pre-existing segmental connections whose efficacy is dramatically altered (i.e. enhanced) as a consequence of the more rostrally-placed spinal cord injury.
Significance
We believe these findings demonstrate segmental reorganization in the human spinal cord caudal to the level of injury, in a manner not previously reported in this population (other than in our earlier publications). The prolonged timeframe argues that this form of plasticity may continue for many months, or even years, after human injury. Whether or not the ILR activity seen in these subjects remains constant (as measured through response distribution, latency and post-stimulus time histogram analysis) through subsequent follow-up examinations in these same subjects will form the basis of a subsequent report.
These observations of presumed human spinal cord plasticity bolster arguments for repeated episodes of rehabilitation in humans with sub-acute or chronic SCI (i.e. >1 year duration) and further argue that animal models of SCI and therapeutic intervention should include prolonged (i.e. >1 year) survival durations post-injury to more accurately model the human condition.
|
Acknowledgements |
|---|
We wish to thank Natalia Alexeeva for her helpful comments related to this manuscript. This work was supported in part by the National Institutes of Health (HD31240; NS36542), and by The Miami Project to Cure Paralysis.
|
References |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Alexeeva N, Broton JG, Suys S, Calancie B. Central cord syndrome of cervical spinal cord injury: widespread changes in muscle recruitment studied by voluntary contractions and transcranial magnetic stimulation. Exp Neurol 1997; 148: 399–406.[Web of Science][Medline]
Bach Y, Rita P. Central nervous system lesions: sprouting and unmasking in rehabilitation. Arch Phys Med Rehabil 1981; 62: 413–7.[Web of Science][Medline]
Barbeau H, McCrea DA, O‘Donovan MJ, Rossignol S, Grill WM, Lemay MA. Tapping into spinal circuits to restore motor function. [Review]. Brain Res Brain Res Rev 1999; 30: 27–51.
Becerra JL, Puckett WR, Hiester ED, Quencer RM, Marcillo AE, Post MJ. MR-pathologic comparisons of Wallerian degeneration in spinal cord injury. AJNR Am J Neuroradiol 1995; 16: 125–33.[Abstract]
Bedbrook GM. Some pertinent observations on the pathology of traumatic spinal paralysis. Paraplegia 1963; 1: 215–27.
Brasil-Neto JP, Cohen LG, Pascual-Leone A, Jabir FK, Wall RT, Hallett M. Rapid reversible modulation of human motor outputs after transient deafferentation of the forearm: a study with transcranial magnetic stimulation. Neurology 1992; 42: 1302–6.
Brown AG, Fyffe RE, Noble R, Rowe MJ. Effects of hind limb nerve section on lumbosacral dorsal horn neurones in the cat. J Physiol (Lond) 1984; 354: 375–94.
Bunge RP, Puckett WR, Becerra JL, Marcillo A, Quencer RM. Observations on the pathology of human spinal cord injury. A review and classification of 22 new cases with details from a case of chronic cord compression with extensive focal demyelination. Adv Neurol 1993; 59: 75–89.[Medline]
Bunge RP, Puckett WR, Hiester ED. Observations on the pathology of several types of human spinal cord injury, with emphasis on the astrocyte response in penetrating injuries. Adv Neurol 1997; 72: 303-15.
Calancie B. Interlimb reflexes following cervical spinal cord injury in man. Exp Brain Res 1991; 85: 458–69.[Web of Science][Medline]
Calancie B, Needham-Shropshire B, Jacobs P, Willer K, Zych G, Green BA. Involuntary stepping after chronic spinal cord injury: evidence for a central rhythm generator for locomotion in man. Brain 1994; 117: 1143–59.
Calancie B, Lutton S, Broton JG. Central nervous system plasticity after spinal cord injury in man: interlimb reflexes and the influence of cutaneous stimulation. Electroencephalogr Clin Neurophysiol 1996; 101: 304–15.[Medline]
Calancie B, Molano M, Broton JG. Neural plasticity as revealed by the natural progression of movement expression—both voluntary and involuntary—in humans after spinal cord injury. [Review]. Prog Brain Res 2000; 128: 71–88.[Web of Science][Medline]
Calancie B, Molano M, Broton JG, Bean JA. Relationship between EMG and muscle force after spinal cord injury. J Spinal Cord Med 2001; 24: 19–25.[Medline]
Christensen MD, Hulsebosch CE. Chronic central pain after spinal cord injury. [Review]. J Neurotrauma 1997; 14: 517–37.[Web of Science][Medline]
Delwaide PJ, Crenna P. Cutaneous nerve stimulation and motorneuronal excitability. II: evidence for non-segmental influences. J Neurol Neurosurg Psychiatry 1984; 47: 190–6.
Ditunno JF, Young W, Donovan WH, Creasey G. The international standards booklet for neurologic and functional classification of spinal cord injury. Paraplegia 1994; 32: 70–80.[Web of Science][Medline]
Gassel MM, Ott KH. Patterns of reflex excitability change after widespread cutaneous stimulation in man. J Neurol Neurosurg Psychiatry 1973; 36: 282–7.
Goldberger ME, Murray M, Tessler A. Sprouting and regeneration in the spinal cord: their roles in recovery of function after spinal injury. In: Gorio A, editor. Neuroregeneration. New York: Raven Press; 1993. p. 241–64.
Goshgarian HG, Guth L. Demonstration of functionally ineffective synapses in the guinea pig spinal cord. Exp Neurol 1977; 57: 613–21.
Goshgarian HG, Yu XJ, Rafols JA. Neuronal and glial changes in the rat phrenic nucleus occurring within h after spinal cord injury. J Comp Neurol 1989; 284: 519–33.
Guth L. Axonal regeneration and functional plasticity in the central nervous system. Exp Neurol 1974; 45: 606–54.[Web of Science][Medline]
Harkema SJ, Hurley SL, Patel UK, Requejo PS, Dobkin BH, Edgerton VR. Human lumbosacral spinal cord interprets loading during stepping. J Neurophysiol 1997; 77: 797–811.
Henneman E, Mendell LM. Functional organization of motorneuron pool and its inputs. In: Mountcastle VB, Brookhart JM, Brooks VB, editors. Handbook of physiology, Sect. 1, Vol. 2, Pt. 1. Bethesda (MD): American Physiological Society; 1981. p. 423–507.
Hiersemenzel LP, Curt A, Dietz V. From spinal shock to spasticity. Neuronal adaptations to a spinal cord injury. Neurology 2000; 54: 1574–82.
Jones KE, Calancie B, Hall A, Bawa P. Comparison of peripheral Ia and corticomotorneuronal composite EPSPs in human motor neurones. Electroencephalogr Clin Neurophysiol 1996; 101: 431–7.[Medline]
Kearney RE, Chan CW. Reflex response of human arm muscles to cutaneous stimulation of the foot. Brain Res 1979; 170: 214–7.[Web of Science][Medline]
Kearney RE, Chan CW. Interlimb reflexes evoked in human arm muscles by ankle displacement. Electroencephalogr Clin Neurophysiol 1981; 52: 65–71.[Web of Science][Medline]
Kendall FP, McCreary EK. Muscles. Testing and function. 3rd ed. Baltimore: Williams & Wilkins; 1983.
Krassioukov AV, Bunge RP, Puckett WR, Bygrave MA. The changes in human spinal sympathetic preganglionic neurones after spinal cord injury. Spinal Cord 1999; 37: 6–13.[Web of Science][Medline]
Krenz NR, Meakin SO, Krassioukov AV, Weaver LC. Neutralizing intraspinal nerve growth factor blocks autonomic dysreflexia caused by spinal cord injury. J Neurosci 1999; 19: 7405–14.
Kuypers HGJM. Anatomy of the descending pathways. In: Mountcastle VB, Brookhart JM, Brooks VB, editors. Handbook of physiology, Sect. 1, Vol. 2, Pt. 1. Bethesda (MD): American Physiological Society; 1981. p. 597–665.
Lawrence DG, Kuypers HG. The functional organization of the motor system in the monkey. II. The effects of lesions of the descending brain-stem pathways. Brain 1968; 91: 15–36.
Maier MA, Illert M, Kirkwood PA, Nielsen J, Lemon RN. Does a C3-C4 propriospinal system transmit corticospinal excitation in the primate? An investigation in the macaque monkey. J Physiol (Lond) 1998; 511: 191–212.
Mathias CJ, Christensen NJ, Frankel HL, Spalding JM. Cardiovascular control in recently injured tetraplegics in spinal shock. Q J Med 1979; 48: 273–87.[Web of Science]
McCouch GP, Austin GM, Liu CN, Liu CY. Sprouting as a cause of spasticity. J Neurophysiol 1958; 21: 205–16.
Mendell LM. Physiological properties of unmyelinated fiber projection to the spinal cord. Exp Neurol 1966; 16: 316–32.[Web of Science][Medline]
Mendell LM. Modifiability of spinal synapses. [Review]. Physiol Rev 1984; 64: 260–324.
Mendell LM, Munson JB, Arvanian VL. Neurotrophins and synaptic plasticity in the mammalian spinal cord. [Review]. J Physiol (Lond) 2001; 533: 91–7.
Nelson SG, Mendell LM. Enhancement in Ia-motorneuron synaptic transmission caudal to chronic spinal cord transection. J Neurophysiol 1979; 42: 642–54.
Nelson SG, Collatos TC, Niechaj A, Mendell LM. Immediate increase in Ia-motorneuron synaptic transmission caudal to spinal cord transection. J Neurophysiol 1979; 42: 656–63.
Palmer E, Ashby P. Corticospinal projections to upper limb motor neurones in humans. J Physiol (Lond) 1992; 448: 397–412.
Peckham PH, Mortimer JT, Marsolais EB. Upper and lower motor neuron lesions in the upper extremity muscles of tetraplegics. Paraplegia 1976; 14: 115–21.[Medline]
Phillips LH 2nd, Park TS. Electrophysiologic mapping of the segmental anatomy of the muscles of the lower extremity. Muscle Nerve 1991; 14: 1213–8.[Web of Science][Medline]
Phillips CG, Porter R. Corticospinal neurones: their role in movement. London: Academic Press; 1977.
Pierrot-Deseilligny E. Transmission of the cortical command for human voluntary movement through cervical propriospinal premotor neurones. [Review]. Prog Neurobiol 1996; 48: 489–517.[Web of Science][Medline]
Quencer RM, Bunge RP. The injured spinal cord: imaging, histopathologic clinical correlates, and basic science approaches to enhancing neural function after spinal cord injury. Spine 1996; 21: 2064–6.[Web of Science][Medline]
Raisman G, Field PM. A quantitative investigation of the development of collateral reinnervation after partial deafferentation of the septal nuclei. Brain Res 1973; 50: 241–64.[Web of Science][Medline]
Romero MI, Rangappa N, Li L, Lightfoot E, Garry MG, Smith GM. Extensive sprouting of sensory afferents and hyperalgesia induced by conditional expression of nerve growth factor in the adult spinal cord. J Neurosci 2000; 20: 4435–45.
Schwab ME, Bartholdi D. Degeneration and regeneration of axons in the lesioned spinal cord. [Review]. Physiol Rev 1996; 76: 319–70.
Siddall PJ, Loeser JD. Pain following spinal cord injury. [Review]. Spinal Cord 2001; 39: 63–73.[Web of Science][Medline]
Steward O. Reorganization of neuronal connections following CNS trauma: principles and experimental paradigms. [Review]. J Neurotrauma 1989; 6: 99–152.[Medline]
Teasell RW, Arnold MO, Krassioukov A, Delaney GA. Cardiovascular consequences of loss of supraspinal control of the sympathetic nervous system after spinal cord injury. [Review]. Arch Phys Med Rehabil 2000; 81: 506–16.[Web of Science][Medline]
Turton A, Lemon RN. The contribution of fast corticospinal input to the voluntary activation of proximal muscles in normal subjects and in stroke patients. Exp Brain Res 1999; 129: 559–72.[Web of Science][Medline]
Ungar-Sargon J, Goldberger ME. Maintenance of specificity by sprouting and regenerating peripheral nerves. II. Variability after lesions. Brain Res 1987; 407: 124–36.[Web of Science][Medline]
Wall PD. Recruitment of ineffective synapses after injury. In: Waxman SG, editor. Functional recovery in neurological disease. Advances in neurology, Vol. 47. New York: Raven Press; 1988. p. 387–400.
Wang ZH, Walter GF, Gerhard L. The expression of nerve growth factor receptor on Schwann cells and the effect of these cells on the regeneration of axons in traumatically injured human spinal cord. Acta Neuropathol (Berl) 1996; 91: 180–4.[Medline]
Weaver LC, Cassam AK, Krassioukov AV, Llewellyn-Smith IJ. Changes in immunoreactivity for growth associated protein-43 suggest reorganization of synapses on spinal sympathetic neurones after cord transection. Neuroscience 1997; 81: 535–51.[Web of Science][Medline]
Wernig A, Müller S, Nanassy A, Cagol E. Laufband therapy based on ‘rules of spinal locomotion’ is effective in spinal cord injured persons. Eur J Neurosci 1995; 7: 823–9.[Web of Science][Medline]
Wolpaw JR, Tennissen AM. Activity-dependent spinal cord plasticity in health and disease. [Review]. Annu Rev Neurosci 2001; 24: 807–43.[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  Y. Gerasimenko, P. Musienko, I. Bogacheva, T. Moshonkina, A. Savochin, I. Lavrov, R. R. Roy, and V. R. Edgerton Propriospinal Bypass of the Serotonergic System That Can Facilitate Stepping J. Neurosci., April 29, 2009; 29(17): 5681 - 5689. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Kawashima, D. Nozaki, M. O. Abe, and K. Nakazawa Shaping Appropriate Locomotive Motor Output Through Interlimb Neural Pathway Within Spinal Cord in Humans J Neurophysiol, June 1, 2008; 99(6): 2946 - 2955. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. G. Hornby, J. H. Kahn, M. Wu, and B. D. Schmit Temporal facilitation of spastic stretch reflexes following human spinal cord injury J. Physiol., March 15, 2006; 571(3): 593 - 604. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. K. Floeter, P. Zhai, R. Saigal, Y. Kim, and J. Statland Motor Neuron Firing Dysfunction in Spastic Patients With Primary Lateral Sclerosis J Neurophysiol, August 1, 2005; 94(2): 919 - 927. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Grasso, Y. P. Ivanenko, M. Zago, M. Molinari, G. Scivoletto, V. Castellano, V. Macellari, and F. Lacquaniti Distributed plasticity of locomotor pattern generators in spinal cord injured patients Brain, May 1, 2004; 127(5): 1019 - 1034. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Frigon, D. F. Collins, and E. P. Zehr Effect of Rhythmic Arm Movement on Reflexes in the Legs: Modulation of Soleus H-Reflexes and Somatosensory Conditioning J Neurophysiol, April 1, 2004; 91(4): 1516 - 1523. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. Zijdewind and C. K. Thomas Motor Unit Firing During and After Voluntary Contractions of Human Thenar Muscles Weakened by Spinal Cord Injury J Neurophysiol, April 1, 2003; 89(4): 2065 - 2071. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||