Hindlimb Immobilization in a Wheelchair Alters Functional Recovery Following Contusive Spinal Cord Injury in the Adult Rat

Introduction

A potential impediment to the translation of activity-based rehabilitation following spinal cord injury (SCI) from animal models to patients is that the conditions of recovery in animal models do not parallel those of patients. Most patients are immobile over the first few weeks following injury, except for muscle stretch and passive joint range-of-motion (ROM) therapies.^1,2 In contrast, animals used in experimental studies of SCI begin to move about in their cages within a few days or a week of injury. With a few exceptions,^3-5 the majority of animal studies do not mention the use of stretch or ROM therapies, yet these are widely accepted in clinical practice worldwide,^6,7 and the Consortium for Spinal Cord Medicine prescribes that physical therapy begin within the first week postinjury and continue throughout the acute phase. A recent review of clinical trials of SCI rehabilitation found these therapies to be ineffective when compared with no intervention or conventional care.^8,9 The effects of immobilization and stretching/ROM maneuvers in animal models of SCI remains largely unstudied.

Step training with weight support is a promising form of activity-based rehabilitation for SCI patients that is often used in conjunction with stretch/ROM therapies. The step-training approach is deeply rooted in the elegant studies by Grillner, Rossignol, Edgerton, and others showing that adult cats with complete spinal transections can be retrained to perform quality hindlimb stepping via repeated exposure to the activity on a treadmill. Afferent input associated with phasic limb movements, loading, and paw or foot contact^10-12 is thought to retrain spinal circuitry below the injury. However, improvements in overground ambulation for patients participating in step training are variable and far less than expected based on the preclinical feline studies.^13-15 Similar to clinical studies, step training in rodent models of incomplete SCI has shown only modest, task-specific improvements in treadmill stepping that rarely lead to improvements in overground stepping.^16-20 In contrast to the clinical situation, however, adult rats with all but the most severe of incomplete injuries show remarkable spontaneous recovery leading some authors, including ourselves, to suggest that early in-cage activity provides sufficient amounts of appropriate afferent input to bring about substantial, even maximal, retraining that is difficult to improve on using a variety of approaches including step training on the treadmill, stepping in shallow water, or swimming.^17,19,20-24

The current study was undertaken to test the hypothesis that early in-cage activity is responsible for the dramatic functional improvements seen in adult rats with incomplete SCI. We limited hindlimb activity and altered the pattern of activity-dependent afferent input by placing the animals in wheelchairs, 15 to 18 hours per day for 5 days a week starting at 4 days postinjury. In addition, we explored the influence of a daily stretch protocol on the profile of recovery in hindlimb-immobilized animals and normally housed animals as controls. We found that periodic hindlimb immobilization in a wheelchair had a negative and lasting effect on functional recovery in rats with mild SCIs and that a daily passive stretch therapy had a lasting negative impact on recovery for normally housed SCI animals, but was not a factor for those experiencing hindlimb immobilization.

Methods

Spinal Cord Injury and Experimental Design

Twenty-one female adult Sprague-Dawley rats (190-215 g) were used for this study. All procedures involving experimental animals were performed according to the guidelines of the University of Louisville Institutional Animal Care and Use Committee. Animals were assigned to 3 experimental groups: Wheelchair/Stretch (WC/SR), wheelchair/non-stretch (WC/non-SR), non-wheelchair/stretch (non-WC/SR), n=5 each or non-wheelchair/non-stretch (non-WC/non-SR) controls, n=6. Animals received contusion injuries at the T10 spinal cord level as described previously.^21,24,25

Figure 1 shows a timeline of the study and all assessments. On postinjury day 4, animals in the 2 wheelchair (WC) groups spent 15 to 18 hours per day (from approximately 4 pm to 8 am), 5 days per week for 8 weeks (dashed bar along time axis) immobilized in a 4-wheeled rat wheelchair (Figure 2). WC cage mates stayed together, paired, in large breeding cages, whereas non-WC animals remained paired in standard cages. Rats in SR groups received the enhanced daily care including the stretch protocol 5 mornings per week, for 8 weeks, beginning the morning of postinjury day 5. Both WC immobilization and SR interventions ended at 8 weeks; assessments continued until 16 weeks. Detailed methods can be found in the supplementary online material.

OPEN IN VIEWER

Figure 1.

Wheelchair immobilization and stretch protocol was for an 8-week period (dashed line) followed by 8 weeks of normal (2 per cage) housing. The schedules of injury and treatments are shown below the timeline, and the schedules of assessments are shown above the timeline.

OPEN IN VIEWER

Figure 2.

(A) An injured animal in a wheelchair. Velcro straps held the animal and hindlimbs in place. (B) A 3-dimensional kinematic analysis of the right hindlimb during wheelchair immobilization. The anatomic landmarks represented are the iliac crest (IC), hip, knee, ankle, and toe. Some movement of the knee or ankle was sometimes seen during spasms, as shown here for the knee. For (C) and (D) all data are shown as means ± standard deviation. (C) The BBB scores over time for all 4 experimental groups (WC, wheelchair; SR, stretch). All animals scored 21 preinjury. Significant differences were found for both WC groups (n = 5) compared with non-WC/non-SR controls (*, n = 6, P < .05). Animals receiving the SR intervention (non-WC/SR, n = 5) had BBB scores that were significantly different from non-WC/non-SR controls (**, n = 6, p < .05, (D) The BBB scores over time for groups based on presence (con) or absence (no-con) of ankle contractures. WC/con (n = 3) and WC/no-con (n = 4) groups have significantly lower BBB scores compared with controls for weeks 2 to 16 (*, n = 6, P < .05). Scores for the WC/con group are significantly different from the WC/no-con group at weeks 2, 6, and 10 to 16 (&, P < .05).

Results

Kinematic Assessment of Stepping

For stepping, group 3D kinematic excursion data were quantified for the hip, knee, and ankle (Figure 3A). Group averages were compared at weeks 6 and 11. At week 6, WC animals showed significantly reduced mean excursions for all 3 joints when compared with non-WC animals (Figure 3A). Hip and knee positions were relatively flexed (70° and 40°, respectively), whereas the ankles were extended (140° to 160°; Figure 3B and D). By week 11, both WC groups regained hip excursions that were not different from controls and also regained some knee excursion, which remained significantly different from non-WC animals by having a lower trough value (flexion; Figure 3C and E). Mean ankle excursions at week 11 were significantly different for the 2 WC groups compared with both non-WC groups, and also between WC groups (WC groups contracture/no contracture; Figure 3A). By week 11, WC animals without contractures had recovered sufficient ROM about the ankle to achieve weight-supported, plantar stepping (Figure 3A and C), whereas animals with ankle contractures had very low excursion ranges reflecting their very limited ankle ROM. There were no differences between non-WC groups with or without SR at any time points, indicating that deficits observed using the BBB scale were in interlimb coordination and not in joint excursions during stepping.

OPEN IN VIEWER

Figure 3.

Three-dimensional kinematic analysis of hindlimb joint excursions during overground stepping. All data are shown as group means ± standard deviation. (A) Joint excursion is presented as the range of angles through which the joint moves and is calculated as maximum peak angle (extension) minus the trough angle (flexion). Baseline excursion data represent the preinjury average for all animals. Symbols within each bar represent significant differences in average excursion; symbols above and below represent significant differences in average peak or trough angles; (*) indicates significant differences for both WC groups (WC/con, n = 3, and WC/no-con, n = 4) when compared with both non-WC groups (non-WC/SR, n = 5, and non-WC/non-SR controls, n = 6), P < .05. (&) represents significant differences between the 2 WC groups, P < .05. (B) through (D) show 2 examples of excursion plots (joint angle changes over time) and examples of stick figures. (B) and (C) show excursion plots for WC rats with and without contractures and stick figures showing extended-limb dragging (B) and flexed limb dragging (C) in WC animals with and without contractures, respectively. WC/con animals displayed cyclic hip and knee movement despite the reduced ankle ROM. (D) and (E) show excursion plots and stick figures for non-WC animals that represent near-normal stepping of the hindlimbs that was not different from baseline at either time point.

Histological Assessments

Using a 1-way ANOVA, no significant differences were found in the percentage of spared white matter (cross-sectional area), gray matter loss, or cavity volume (mm³) at the epicenter among any groups (Table 1). The amount of spared white matter at the epicenter (30%) is consistent with the functional recovery achieved by the control group as measured by the BBB scale (~18). ²⁹ BBB scores of ~10 at week 1 and terminal scores of ~18 for injured controls are higher than expected with a 200 kD contusion injury using the Infinite Horizons system. We attribute the relatively mild injuries to the custom-made system we used to stabilize the vertebrae, which was novel for us at the time. Nonetheless, all the animals were injured in a consistent manner.

OPEN IN VIEWER

Table 1.

Mean Spared White Matter Cross-Sectional Area (CSA) at the Injury Epicenter (mm²), Gray Matter Volume Loss (mm³), and Cavity Volume (mm³) ^a

Group	White Matter CSA	Gray Matter Volume	Cavity Volume
WC/SR	28.8 ± 14.13	160.2 ± 24.75	17.2 ± 7.72
WC/non-SR	21.4 ± 7.16	152.3 ± 30.03	21.6 ± 21.08
Non-WC/SR	24.4 ± 7.10	152.7 ± 52.62	36.8 ± 27.38
Non-WC/non-SR	32.4 ± 6.36	143.8 ± 25.85	53.7 ± 29.8

Abbreviations: WC, wheelchair; SR, stretch.

a

No significant differences were found among groups.

Discussion

It has been more than 3 decades since Sten Grillner and his talented trainees began referring to “the central generation of locomotion” while studying the low spinal cat model.^39,40 Since then, this model of acute transection has been used to characterize the capabilities of the lumbar spinal cord and the principles underlying locomotor retraining on the treadmill. The work of many laboratories including those of Edgerton, Rossignol, and Barbeau have identified several issues including phasic limb loading, cutaneous paw contact, and load progression as critical to retraining.^12,41-46 In addition, training is thought to be task specific in that improvements are largely limited to the trained task.^11,21,47,48

Translating these key principles to the rodent and human models, however, has not been straightforward. Inherent to these 3 models—human, cat, and rat—are a number of critical differences, as reviewed by Majczynski and Slawinska, ⁴⁹ which may hinder translation and complicate the interpretation of experimental results. For example, most rodent studies use midthoracic transections/contusions or incomplete lacerations and acute or subacute retraining. In contrast, most clinical studies involve chronic incomplete or dyscomplete cervical or high-thoracic contusions. These 3 models also have very different locomotor characteristics: bipedal versus quadrupedal, high versus low versus very low center of gravity. The characteristics of balance, trunk stability, and posture and the consequences of errors in stepping vary considerably and dramatically influence the capabilities of each species during the critical acute and subacute postinjury time period. Rodents are very active within days of injury whereas patients are largely immobile for several weeks or months, experiencing little more than transfer, stretching, and passive ROM therapies during this period. Apart from the substantial studies conducted by Dobkin and colleagues, rarely will patients enter an activity-based rehabilitation setting before 3 to 6 months postinjury.^2,14

Importantly, step training of spinal cord injured rodents consistently demonstrates task-specific changes in hindlimb function but rarely achieves frank improvements in overground stepping. Fouad and colleagues ¹⁷ used daily treadmill training in an over-hemisection rat model and were unable to uncover improvements in overground stepping using the BBB scale and kinematics as outcomes. As also alluded to earlier by de Leon et al, ⁵⁰ Fouad et al ¹⁷ concluded that spontaneous recovery resulting from self-training was substantial and that treadmill training provided no added benefit. Subsequently, Multon et al ¹⁸ used manually assisted treadmill training of rats with T7 microballoon compression injuries to demonstrate that small improvements in BBB scores could be achieved. Notably, these animals were housed individually, without enrichment. ²⁷ Invariably, studies using rats with all but the most severe incomplete injuries report substantial improvements in overground locomotion over the first few weeks postinjury.^{17-19,29,51,52} Similar reports for cats following thoracic contusion or incomplete laceration injuries are limited, ⁵³ with the majority of studies being focused on task-specific improvements of limb function on the treadmill.

Taken together, these observations suggest that the in-cage activity exhibited by rats with acute incomplete SCIs should be viewed as cumulative step training. Therefore, we hypothesized that incompletely injured rats maximally retrain themselves within the first few weeks after injury. To test this hypothesis, we designed a 4-wheeled cart or wheelchair that dramatically restricts hindlimb movement by holding the limbs in an extended-drag position and alters the normal pattern of afferent input while still permitting good in-cage mobility. Anticipating that hindlimb immobilization might induce pathologies of muscles and joints, we also developed a stretching protocol based on standard-of-care clinical physical therapy. ⁹ We found that neither intermittent hindlimb immobilization (WC) nor stretch (SR) initiated at 4 days postinjury, prior to the resolution of spinal shock, had any influence on the dramatic early increase in hindlimb function. Both SR and non-SR and WC and non-WC groups had similar BBB scores at 7 days postinjury (10.6 ± 1.43 and 11.3 ± 1.85, respectively), suggesting that this phase of recovery is likely not dependent on activity or patterns of afferent input. Thereafter, WC (SR and non-SR) animals experienced a dramatic loss of hindlimb function leaving them unable to generate weight-supporting plantar steps by 3 weeks postinjury. Poor hindlimb function, assessed using the BBB scale and kinematically, persisted for the 8-week duration of WC immobilization. When returned to normal housing at 8 weeks postinjury, function rebounded somewhat during weeks 8 to 10 but remained significantly below the non-WC/non-SR control group out to week 16, suggesting that the spinal cord circuitry was not as responsive to in-cage retraining beyond 8 weeks postinjury. Both WC/SR and WC/non-SR groups also developed ankle pathologies at a 40% to 50% rate that dramatically limited ROM. Only modest decreases in muscle weights were observed for the LG and MG only (not the TA), and the WC/SR group experienced loss similar to the WC/non-SR group. The SR protocol did appear to ameliorate a decrease in bone volume ratio seen in the WC/non-SR group; however, our group sizes were too small to achieve significance with this measure. Importantly, when the animals with ankle pathologies were excluded, a significant difference in the BBB scores and kinematic measures persisted, showing that WC immobilization had a dramatic and lasting negative influence on locomotor function.

Conceptually, there are interesting parallels between the current work, the literature discussed above, and the work of Maier et al, ⁵⁴ using a rat model of unilateral corticospinal tract injury. Maier and colleagues showed that constraining the use of one limb, thus forcing the use of the impaired limb, led to dramatic improvements in function that were correlated to increased contralateral collateralization of corticospinal fibers in the cervical gray matter. When not forced to use the forelimb on the injured side, it remains immobile and the long term functional (and anatomical) consequences of dis-use are dramatic.

In contrast to the anticipated loss of function brought about by WC hindlimb immobilization, we also observed a significant and surprising difference between the SR and non-SR groups of non-WC animals, indicating that our stretch protocol was having a detrimental effect on hindlimb function. The differences between groups developed somewhat gradually between weeks 2 and 5, suggesting that the SR protocol was bringing about a functional plateau earlier than observed in the non-SR control animals. Despite stretching/ROM being a standard-of-care practiced clinically, systematic literature reviews have suggested that these approaches have mixed benefits or the data appear inconclusive at best. ⁹ While many dubious explanations may be put forward for our observations, the most feasible is that our stretching protocol was activating barrages of afferent activity that disrupted lumbosacral circuitry and plasticity. The Grau laboratory, using an instrumental learning paradigm, has demonstrated that randomly applied (uncontrollable) noxious inputs can disrupt spinal cord plasticity associated with both the short-term establishment phase and long-term maintenance phase of instrumental learning in the fully transected rat preparation.^55,56 Importantly, they also uncovered evidence that only 6 minutes (cumulative) of uncontrollable noxious input (tail shock) can negatively alter the normal profile of locomotor recovery in a contusion model of SCI. ⁵⁷ Thus, we posit that our SR protocol, acting as an uncontrollable input, had a negative influence on the process of plasticity that leads to or allows in-cage activity-dependent functional recovery. The non-WC/SR animals in our study were high functioning; they exhibited consistent weight-supported stepping as measured by the BBB and were not different from controls when assessed kinematically while moving freely overground and swimming, indicating the stretch protocol was not debilitating. The clinical implications of these observations are significant, and future work should address the utility of stretch-based therapies for avoidance of peripheral pathologies versus the potential for negative long-term consequences on spinal learning and locomotor retraining.

WC-immobilized animals also exhibited higher levels of antagonist muscle co-contraction in response to tail pinch, as a surrogate measure of spasticity, compared with non-WC animals. In addition, we routinely observed bouts of bilateral hindlimb spasms during wheelchair immobilization, which were not quantified. The combination of ankle contractures and spasticity suggests that the immobilization was not just preventing movement-related afferent input but was possibly resulting in uncontrollable, noxious afferent input. Thus, while immobilization or inactivity may prevent optimal in-cage retraining, it remains to be determined how an absence or lack of activity-dependent afferent input compares with an abundance of activity-independent (uncontrollable), possibly noxious, input to influence functional recovery. ⁵⁷

Recently, we discovered that providing 50% to 60% body weight support with 5 cm of water allows young adult female SD rats with moderately severe (25 g-cm) T9 contusion injuries to generate high-quality plantar hindlimb stepping with near-normal kinematics as early as 1 week postinjury in the absence of an applied retraining strategy. ²³ In an elegant piece of work, Courtine and colleagues ⁵⁸ showed in fully transected rats that a combination of serotonergic agonists, epidural stimulation, and weight support allowed near-normal stepping (albeit bipedal) to be expressed at 1 week postinjury. In the current study, nightly WC immobilization on days 4, 5, and 6 postinjury, as spinal shock was resolving, had no influence on the day 7 BBB scores. Taken together, these findings suggest that a robust ability to generate a stepping pattern is present immediately postinjury and that this capacity is not dependent on an acute retraining phenomenon. Importantly, however, in our shallow-water stepping model, we found that the acute stepping capacity was not maintained in the absence of continued exposure to the activity (“retraining”). This is in good agreement with the work of Courtine et al, ⁵⁸ described above, who found that excellent patterns induced at 1 week (with the “full combination” treatment) were not present at 9 weeks in animals that did not receive treadmill training in the intervening period. Thus, it follows that hindlimb immobilization of animals with mild to moderate spinal cord injuries sufficiently disrupts the patterns of afferent input from the hindlimbs to prevent the maintenance of the acute capacity leading to a dramatic drop in locomotor function over the first few weeks postinjury. With the caveat that intense acute activity may be detrimental to traumatically injured spinal cord, ²² our results support suggestions that appropriate activity, applied acutely, may enhance or optimize overall functional outcomes following a contusive SCI.

The current findings suggest that limiting early in-cage activity can dramatically alter functional recovery of locomotion in a rodent model of mild to moderate SCI. Surprisingly, we also show that a hindlimb stretch protocol can have a negative effect on recovery in normally housed animals, preventing the development of forelimb–hindlimb coordination, while having no apparent influence on recovery in hindlimb immobilized animals or on the development of ankle contractures. These findings imply that the immobility experienced by the majority of spinal cord–injured patients, and some aspects of standard-of-care therapies, may significantly hinder functional recovery acutely and also potentially limit the responsiveness of the spinal cord circuitry to retraining efforts initiated at later time points.