Neuroanatomical Substrates of Functional Recovery After Experimental Spinal Cord Injury: Implications of Basic Science Research for Human Spinal Cord Injury

Lesion Development After SCI

Based on the histopathology of 50 spinal cords examined post-mortem in The Miami Project, Bunge and colleagues^7–9 broadly categorized lesions into: (1) contusion injuries in which the glial limitans and spinal cord surface remain intact or (2) maceration or laceration injuries that disrupt the glial and pial interface and may directly tear the spinal cord tissue. Contusion and some maceration injuries may be present with loss of central gray and white matter, which creates a cavity that is surrounded by a rim of intact white matter at the periphery of the spinal cord.^8,9 Of particular interest to clinicians is the fact that intact, continuous central nervous system (CNS) axons traverse the lesion center in many individuals with functionally complete SCI.^8,9

Thus, a clinically complete lesion does not necessarily indicate an anatomically complete SCI. A functionally or clinically complete lesion is typically described as the absence of sensory and motor function below the level of the lesion (American Spinal Injury Association [ASIA] Impairment Scale classification A).¹⁰ In contrast, an individual with an incomplete SCI has sensory and/or motor function below the lesion (ASIA B, C, and D).¹¹ Thus, tissue sparing after complete SCI suggests that the remaining axons are available to possibly mediate some recovery of function but do not appear to be recruited using traditional rehabilitation approaches such as compensation. The emergence of the full capacity of these spared axons may depend on the type of therapeutic interventions that are used.

Several experimental models of SCI have been developed in order to understand factors that may facilitate recovery of function after SCI in humans. The advantages of using an animal model rather than studying humans with SCI are: (1) the severity of the lesion can be controlled and reproduced across animals and experiments, (2) direct neuroanatomical evidence of lesion development over time can be attained using invasive procedures, and (3) effective parameters of therapeutic interventions can be established and refined before translating them for human use. One of the most clinically relevant experimental models of SCI is the contusion injury in rats, created by rapid impact of the dorsal surface of the spinal cord using an electromagnetic device^12,13 or a weight-drop device.^14–16 These devices create a central core lesion and a peripheral rim of spared white matter similar to that reported in humans. The area of the cord with the greatest extent of damage is referred to as the “lesion epicenter.” This review will focus on data gathered from contusion models of SCI primarily in rats.

Lesion development in SCI progresses over time and in neuroanatomical distribution. The primary phase of injury occurs during the first 18 hours and is typified by necrotic death of neurons and axons that were directly disrupted by the trauma.^17–20 The secondary phase of tissue injury can last several weeks and progresses in rostral and caudal directions, away from the lesion epicenter. Recent work suggests that the immune system plays a primary role in initiating cellular cascades that contribute to the expansion of the lesion during the secondary phase of lesion development.^21–23 In the secondary phase, immune system cells such as monocytes and macrophages are thought to emit chemical signals such as cytokines and chemokines.²² These substances act on neurons and oligodendrocytes and appear to trigger apoptosis or programmed cell death in which DNA in the nucleus of the cell is systematically broken down into fragments.^24–26 Thus, intracellular processes in neurons and oligodendrocytes lead to destruction of their nuclei, causing cell death, despite no direct trauma to the cell. Apoptosis is known to occur after contusive SCI, especially in white matter tracts great distances away from the lesion (at least 4 spinal segments above and below the lesion epicenter).^26,27 Oligodendrocytes appear to undergo programmed cell death, which results in demyelination of axons many segments away from the lesion epicenter.²⁸ Thus, early after SCI, the injury is small, focal, and localized to the point of impact, but over time the lesion spreads to distant sites and may preferentially affect myelination of axons that are otherwise intact.

One critical issue in SCI research is whether lesion size and tissue sparing are important determinants of final functional outcome after contusion injury. If the size of the lesion is an important determinant of final functional outcome in patients with SCI, then interventions that attempt to bridge the central lesion or limit the secondary progression of the lesion would appear to be quite beneficial. The action of methylprednisolone, the standard treatment after SCI, is to reduce the inflammatory response and inhibit the immune system.^29,30 Methylprednisolone, therefore, could limit lesion expansion, which might explain the marked improvement in functional outcome of those treated.^29,30 As other pharmaceutical interventions are developed that limit or prevent the secondary lesion cascade, it is likely that the incidence of functionally complete lesions will decline.

Relationship of Lesion Size and Behavioral Outcome

Are Segmental Reflexes Modified According to the Amount of Tissue Sparing?

In our studies, we determined that tissue sparing facilitates recovery of locomotion after contusion but we did not know whether the amount of sparing would differentially affect other motor behaviors such as reflexes. In another set of experiments, my collaborators and I⁵² compared the performance of the flexor withdrawal reflex in animals with substantial sparing (mild SCI) or minimal sparing (moderate SCI). The flexor withdrawal reflex was elicited by pinching the intrinsic muscles of the paw with our fingers, which elicited rapid flexion of the hip, knee, and ankle. By using our fingers rather than an implement, we were able to release the paw immediately at the onset of movement so that the motion was not impeded. We rated the pressure of the pinch on a scale from 1 to 3, with 1 being almost no pressure, 2 being the pressure necessary to elicit a response in an animal without a lesion, and 3 being a large amount of pressure. In order to assess the movement characteristics of the hind limb after stimulation, we videotaped the performance and used frame-by-frame kinematic analysis. We focused our analysis on the ankle because it appeared to have greater deficits than the other joints during locomotion.

Examination of stimulus threshold for the animals with moderate SCI showed an initial increase above normal 1 week after SCI, which decreased to below-normal levels by 4 weeks after SCI. Over time, very little pinch pressure was required to elicit a robust flexor withdrawal response. Given the manner in which we measured stimulus intensity, we have recently replicated these findings in animals with moderate SCI under blinded conditions and found a decrease in stimulus threshold compared with animals with laminectomy that served as controls.⁵³ Due to technological problems in which verbal statements of stimulus intensity recorded on videotape were erased, no stimulus data could be recovered for the animals with mild SCI. Therefore, we quantified the stimulus intensity after mild SCI (n=12) in several other studies in our laboratory and present these data as representative of animals with mild SCI (unpublished observation). One week after mild SCI, the stimulus intensity was slightly less than normal but had returned to normal levels by 4 weeks after SCI. In summary, animals with less sparing required a much lower stimulus intensity to elicit the reflex than animals with more tissue sparing and animals without lesions.

The reflex response was more robust after moderate SCI but not after mild SCI, as was evident by: (1) incorporation of pronounced trunk flexion and rotation, (2) rapid, repetitive hind-limb flexion to a single stimulus, and (3) brisk, full extension of the unstimulated hind limb. These qualitative observations suggest that hyperreflexia may have developed after moderate SCI, an interpretation also supported by precise quantitative measurements of hind-limb movement. Although flexor excursion of the ankle was unchanged for the animals with mild and moderate SCI, we found that movement speed and timing were different between groups. There was a trend that peak angular velocity of the ankle was consistently higher for animals with moderate SCI than for animals with mild SCI at 2, 3, and 4 weeks after SCI. We also found that the animals with moderate SCI had faster movement times and reached peak flexion sooner than the animals with mild SCI (Figure). Our data suggest that pronounced hyperreflexia of segmental systems in the lumbar spinal cord developed when the lesion was extensive and tissue sparing was minimal.

• Download figure
• Open in new tab
• Download powerpoint

Figure.

Mean (±SD) movement times in animals with mild spinal cord injury (SCI) and in animals with moderate SCI. Time to peak flexion was recorded before SCI and at 1-week intervals for 4 weeks after SCI. Number sign (#) indicates P<.05, significantly different than before SCI values; asterisk (*) indicates P<.05, significant difference between groups; double asterisk (**) indicates P<.01, significant difference between groups; triple asterisk (***) indicates P<.001, significant difference between groups.

In summary, tissue sparing after contusion-type SCI appears to directly mediate recovery of function in experimental models. Although the extent of locomotor recovery is related to the amount of tissue sparing, even minimal sparing induces marked reorganization of neural systems below the level of the lesion. This reorganization appears to increase segmental reflex responses and may or may not be sufficient to facilitate locomotor recovery in animals. Given the relative dependence of recovery of motor function on tissue sparing, it seems important to identify which supraspinal systems are spared after spinal cord contusion and to determine whether training in motor skill development after SCI will promote even more extensive recovery.

Identification of CNS Systems Spared After SCI

To date, the focus of research in my laboratory has been on identifying the source and extent of sparing in descending systems after spinal cord contusion in the rat. Using anatomical techniques, my collaborators and I⁵⁴ filled the lumbar spinal cord with dye (Fluorogold*) in animals without lesions as well as in animals that had recovered from mild or moderate SCI. The tracer is taken up at the axon terminals and retrogradely transported back to the cell body, where it accumulates. Thus, cell bodies with intact axons extending to the lumbar cord will be positively labeled with tracer. We examined several areas of the brain, brain stem, and cervical spinal cord known to contribute to movement. For locomotion in animals without lesions, the rubrospinal tract has been shown to control hind-limb flexion during the swing phase of locomotion^55,56 and the vestibulospinal tract has been shown to control hind-limb extension during stance.⁵⁷ Likewise, the reticulospinal system appears to be important for both flexion and extension of the hind limb during the swing and stance phases of normal locomotion, respectively.^58,59 There is also evidence that coordination of the forelimbs and hind limbs is mediated by long-descending propriospinal neurons connecting the cervical and lumbar enlargements.⁶⁰

In general, retrograde tracing demonstrated greater cell labeling after mild SCI than after moderate SCI. Despite differential sparing, both groups had labeled neurons in most of the movement-related nuclei in the CNS, including the red nucleus, medullary reticular formation, raphe nuclei, and lateral vestibular nucleus. However, there were 2 areas that had little or no labeling in either group: the sensorimotor cortex and lamina 8 of the gray matter of the cervical spinal cord. Thus, it appears that corticospinal and long-descending propriospinal tracts to the anterior horn are lost after spinal cord contusion, regardless of lesion severity. It is likely that the central position of these tracts in the rat spinal cord rendered them more susceptible to damage after contusion than tracts located at the peripheral edge of the spinal cord. Although the human corticospinal tract runs in the anterior and lateral funiculi of the spinal cord, the corticospinal tract in the rat travels in the base of the dorsal funiculus.⁶¹

In summary, the spared tissue at the lesion center following mild or moderate spinal cord contusion is composed of at least rubro-, reticulo-, vestibulo-, and raphe-spinal tracts. Given that locomotor recovery depends on tissue sparing, these descending brain-stem systems must play an important role, whereas cortico-spinal and propriospinal tracts probably play little or no role. Anatomically intact axons may not be physiologically active. However, in our model, at least some of these intact axons appear to be functional, given that transection of the spared axons results in a loss of previously recovered locomotion. Further studies are being conducted to determine whether individual descending pathways contribute to the recovery of specific features of locomotion. Alternatively, the spared systems may function as a unit, modulating descending drive to the reorganized spinal cord systems below the injury.

Justification for Locomotor Training After SCI

In the clinic, the goal of rehabilitation after SCI is re-entry into the community, functional independence, and energy conservation. Traditionally, the treatment of choice to achieve these goals has been to train the patient to perform compensatory strategies. Although these strategies effectively accomplish the essential goals of rehabilitation, they may not have taken advantage of spared descending systems or the inherent neuroplasticity of the spinal cord. Given our knowledge about SCI through basic science research, the physical therapy profession is now challenged to adapt current interventions and to create new, effective therapeutic approaches for individuals with SCI according to these scientific principles to test whether they are beneficial.

One of the first attempts to convert neuroanatomical principles from the laboratory into a therapeutic treatment approach is based on evidence of central pattern generators (CPGs) in the spinal cord. Central pattern generators produce rhythmic, oscillating activity of limb flexor and extensor muscle groups independent of any supraspinal or afferent input.^62–64 Thus, CPGs appear to be interneuron-motoneuron circuits within the spinal cord of vertebrates. It is widely held that such CPGs are located in the lumbar segments of the spinal cord in rats, cats, and dogs.^62–65 These CPGs are multifunctional and can produce a variety of locomotor behaviors. Normally, CPGs that produce overground locomotion are activated by descending supraspinal input; however, in the absence of this input, the CPGs can be triggered by an external stimulus such as a moving treadmill belt. Thus, the neural control of overground locomotion is distinct from the control of reflex locomotion on a treadmill. Animals that are unable to locomote overground after complete transection of the spinal cord perform nearly normal stepping on the treadmill.^48,49,66

Scientists have taken advantage of the principles of CPG activation to examine the capacity of the isolated lumbar spinal cord to learn. Edgerton and colleagues^67,68 have carried out a series of studies in which animals with complete, low-thoracic transection of the spinal cord were trained to step or stand with their hind limbs and their performance was compared with the spontaneous recovery of untrained animals with SCI that served as controls. The animals in the experimental group were placed in a body support sling on a treadmill and underwent extensive training to either stand in a stationary position with the treadmill turned off or to step with their hind limbs on the moving treadmill belt. Initially, the hind limbs of the animals in the step training group were manually assisted, when necessary, to complete the step. During training, the amount of body weight supported by the hind limbs and treadmill speeds were varied.

The animals in the standing group were trained by increasing afferent input to the spinal cord via cutaneous stimulation of the tail or paw. By 3 months, the animals in the standing group could stand independently and support their full body weight for 30 minutes, whereas the untrained animals were unable to stand. The animals in the stepping group were able to walk on the treadmill at relatively high speeds without assistance. Although the untrained animals had some spontaneous recovery of treadmill stepping, these animals could only tolerate treadmill speeds nearly 3 times lower than the treadmill speeds tolerated by the trained animals. Thus, each group of animals learned to perform the task on which they were trained to a higher competency than animals relying on only spontaneous recovery. More importantly, these findings demonstrate that the isolated lumbar spinal cord has the capacity to “learn” and execute motor tasks in quadrupedal animals.

In further research, Edgerton's group set out to determine whether this learning was task-specific or more generalized and could improve motor performance on other behavioral tasks.⁶⁹ Animals with complete transection of the spinal cord were trained to step or stand until performance plateaued, then they were tested on the opposite task (eg, step training followed by stand training). Animals that were trained to step could not stand in a manner similar to that of animals that were trained to stand. They were unable to support themselves for any length of time without a large amount of cutaneous stimulation. Likewise, most animals that were trained to stand could not step. Two animals took only a few aberrant steps at extremely low treadmill speeds. Thus, the general observation that animals trained to step could not stand and animals trained to stand could not step is powerful evidence of task-specific learning. Further evidence was gleaned by cross-training these animals. Animals that learned to step or stand were later trained to perform the opposite task, although they lost the ability to perform the first task. That is, the isolated spinal cord acquires complex motor skills in a context-dependant manner and requires extensive training.

Recently, treatments based on CPGs have been successfully transferred to human subjects with SCI by several researchers.^70–76 The methods of intervention are based on the principles identified in animals with SCI by Edgerton and colleagues.^67–69 Initially, the individual is partially unweighted over a treadmill, and the lower extremities are manually assisted to step. With training, people with SCI can progress to full weight support and stepping independently while on the treadmill.⁷⁰ Even individuals with long-standing, functionally complete lesions have been trained to step independently on the treadmill (see article by Behrman and Harkema⁷⁷ in this special series). Furthermore, training at increased tread-mill speeds and greater loads across the legs over time facilitated greater angular excursion of the joints, increased amplitudes of muscle activity (measured via electromyography), and facilitated appropriate muscle activation patterns relative to other muscles and to the phases of gait.⁷⁰ These findings serve as a dramatic example of the inherent neuroplasticity of the spinal cord. In addition, training can trigger this plasticity several years after the SCI, suggesting that the potential for recovery is somehow preserved in the spinal cord over time.

Summary

The focus of this review has been to highlight current advances in basic science research that may have important implications for understanding recovery of function after SCI. Research using animal models appears to be relevant to humans and can serve as a guide for rehabilitative intervention. In summary, some key factors of SCI are: (1) the initiation of secondary immunological cascades contributes to lesion development and loss of undamaged CNS tissue, (2) anatomically incomplete lesions can exist in humans where the lesion is clinically complete, (3) sparing of tissue at the lesion epicenter can dramatically enhance motor output and is likely to facilitate reorganization of neural systems below the lesion, (4) the spinal cord has the capacity to learn complex motor skills such as locomotion, (5) training paradigms must provide specific, task-dependent experience to facilitate skill acquisition after SCI, and (6) an exercise paradigm based on rhythmic stepping under low-load conditions, developed and extensively used in experimental models of SCI, appears to produce remarkable benefits in humans with SCI as well.