| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1Brain Rehabilitation Research Center, Malcom Randall Veterans Affairs Medical Center, Gainesville, Florida; 2Rehabilitation Research and Development Center, Veterans Affairs Palo Alto Health Care System, Palo Alto, California; 3Department of Physical Therapy and 4Brooks Center for Rehabilitation Studies, University of Florida, Gainesville, Florida; and 5Department of Orthopaedic Surgery, Stanford University Medical School, Stanford, California
Submitted 15 September 2004; accepted in final form 3 December 2004
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
; Kim et al. 2003). Although significant reductions in performance have been observed in both limbs during bimanual upper extremity performance in hemiparetic persons and control subjects (Lewis and Byblow 2004), there have been conflicting suggestions regarding bimanual movement since different paradigms have been reported to improve (Cunningham et al. 2002) and constrain unilateral performance (Rice and Newell 2004). Thus the similarity of the coordination deficits characterizing uni- and bilateral performance of the paretic limb is relatively unknown.
The myriad of common activities of daily living involving the legs requires production of bilateral motor patterns. Accordingly, to subserve coordinated, task-specific activity of the legs, much of the neural circuitry that controls lower limb function is organized bilaterally. One prominent example of this bilateral neural circuitry is the crossed extension (withdrawal) reflex where target limb extension is observed in conjunction with the contralateral withdrawal reflex (Sherrington 1910). Bilateral motor output is modulated for different tasks and is further modulated within different phases of the gait cycle (e.g., Schneider et al. 2000). To meet these task-specific demands requires significant reorganization of bilateral motor output and therefore the neural circuitry controlling bilateral coordination must have substantial flexibility. The underlying premise of this research is that paretic limb motor pattern generation may be substantially influenced by contralesional neural activity related to the ipsilesional (nonparetic) leg sensorimotor state during bilateral lower limb function. The main question of interest was whether paretic leg motor patterns differ during mechanically equivalent bi- and unilateral tasks, and, if so, whether nonparetic leg participation improves or exacerbates paretic leg coordination deficits.
Our initial prediction was that nonparetic leg participation would improve paretic leg coordination deficits during bilateral tasks because all of the "appropriate" sensorimotor information associated with the nonparetic leg could be integrated by the nervous system and would contribute to a more appropriate paretic leg pattern. However, there is considerable evidence that paretic leg coordination may be disproportionately influenced by the sensorimotor state of the nonparetic leg. One such manifestation of these exaggerated interlimb influence is an associated reaction or movement (Brunnstrom 1970), in which forceful voluntary exertion of the nonparetic limb elicits involuntary activation or movement in the paretic limb. It has been proposed that associated movements observed in hemiparesis result from loss of modulation by a supraspinal inhibitory system that normally suppresses bilateral activity when unilateral activity is desired (Lazarus 1992). The sensorimotor state of the nonparetic limb can likely influence paretic leg activity in even more complex ways. For example, in the sitting position, nonparetic leg hip extension has been associated with paretic leg hip extension; this bilateral pattern contrasts with a reciprocal pattern of contralateral hip flexion during ipsilateral hip extension observed in control subjects (Gauthier et al. 1992). Brunnstrom (1970) suggests that nonparetic leg activity might be used deliberately in therapy to elicit paretic limb activity in muscles that patients have difficulty activating voluntarily. Specifically, forceful voluntary exertion of the nonparetic leg in extension or flexion elicits involuntary tensing or movement of flexion or extension, respectively, in the resting paretic leg when the subject is supine (Brunnstrom 1970). Thus it remains a question whether the influences of nonparetic leg activity on paretic leg coordination are positive or negative.
It is also unclear whether there are general principles related to interlimb influences on paretic leg coordination that generalize across different lower limb tasks or if interlimb influences are highly task specific. Lower limb function entails coordination of multiple muscles to produce force. Constraints at the foot-environment interaction determine whether motor activity of the leg moves the foot or if the foot pushes isometrically against the environment. From the neural control perspective, isometric force generation and movement represent markedly different relationships between proprioceptive feedback and muscle excitation. Thus there may be task-specific differences in interlimb influences in each case. For example, the role of movement dependent feedback will differ substantially in isometric tasks relative to its function in voluntary movement.
This study compares paretic leg coordination during several different pairs of biomechanically similar unilateral and bilateral motor tasks (e.g., with and without the participation of the nonparetic leg). We have developed a series of lower-limb tasks (isometric force generation, discrete movement, and cyclical movement) that can be performed on a modified pedaling ergometer. The ergometer allows constraint of the kinematics and kinetics of different tasks, while the loading of the legs is mechanically decoupled. Previous experiments in our lab have successfully used similar methodology in numerous studies of coordination in neurologically healthy persons (Kautz et al. 2002; Ting et al. 1998, 2000). Thus we are able to make meaningful comparisons between biomechanically equivalent uni- and bilateral tasks, while simultaneously measuring task kinematics and kinetics using a proven methodology.
In this manuscript, we report findings of the influence of the nonparetic leg on the coordination deficits observed in the paretic leg during motor tasks of progressively increasing task complexity: isometric force generation, discrete movement, and cyclical movement (i.e., pedaling). Consistent with clinical concepts of interlimb coupling of hemiparetic flexion and extension synergies (e.g., Brunnstrom 1970; Perry 1992), we hypothesized that paretic leg coordination would demonstrate specific aberrations during isometric force generation and movement that are dependent on whether the nonparetic leg generated flexion or extension force. Furthermore, we expected the aberrations associated with nonparetic leg flexion or extension to be similar in each of the lower limb tasks. In an effort to provide a rational scientific basis for therapeutic interventions, our goal is to understand how paretic leg coordination is influenced by the sensorimotor state of the nonparetic leg.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Twenty-one subjects with poststroke hemiparesis of >6-mo duration and 11 similarly aged neurologically healthy subjects were recruited from the surrounding community (Table 1). Hemiparetic subjects were selected if they had sustained a single, unilateral cerebrovascular accident (CVA) and experienced residual lower limb paresis in the absence of severe perceptual, cognitive, or sensory deficits, significant lower limb contractures or orthopedic impairment, significant cardiovascular impairments contraindicative to mild bouts of exertion (pedaling), and could tolerate sitting on a bicycle seat for 
1 h. All subjects provided informed consent as approved by the Stanford University administrative panels on human subjects and consistent with the Declaration of Helsinki. The lower extremity and balance subsections of the Fugl-Meyer Motor Assessment (Fugl-Meyer et al. 1975) were performed on the hemiparetic subjects to provide a global indicator of motor impairment. The neurologically healthy subjects showed no signs of neurological disease, lower limb orthopedic impairment, or cardiorespiratory disorders.
|
) as assessed clinically by a portion of the modified Fugl-Meyer assessment (see Synergy Performance in Table 1). Subjects scoring 
14 on the lower extremity synergy score (Fugl-Meyer et al. 1975) were able to move only within synergy patterns (n = 6; e.g., only basic limb flexion or extension synergies are performed voluntarily and are sufficiently developed to show definite joint movements), subjects scoring 15–18 were able to combine elements of the synergy patterns (n = 5; e.g., some movement combinations that deviate from basic limb synergies become available), and those scoring >18 (n = 10) were able to perform at least partial movements independent of the synergy patterns. Experimental apparatus to eliminate mechanical coupling
Our experimental design compared a hemiparetic subject's ability to perform three different exercise tasks with the paretic leg (isometric force generation, discrete limb movement, and pedaling) while the nonparetic leg was either relaxed and supported in a mid-extension position or performed the same task (i.e., bilateral isometric force generation, bilateral dynamic flexion or extension, reciprocal pedaling). Fundamental to our design is mechanical decoupling of the limbs during all tasks so that the load experienced by the paretic leg remained unchanged whether the nonparetic leg rested or executed the same task. The term "paretic," used in the following text, describes one leg in the experimental design and refers to the contralesional leg in persons with poststroke hemiparesis and to the leg randomly chosen in neurologically healthy subjects to perform the same conditions as the paretic leg of poststroke subjects.
A two-servomotor, split-axle pedaling apparatus was used to test the different exercise tasks (see Van der Loos et al. 2002 for detailed description of apparatus). Each crank was independently connected to a servomotor, thus by implementing different control modes using custom hardware and software, we were able to eliminate mechanical coupling between the two cranks. Specifically, we did not allow the pedal force generated by the paretic leg to be transmitted to the opposite crank. The cranks of the split-axle pedaling apparatus were uncoupled and each driven by independent servomotors (Kollmorgen B606A motor, D20 motor controller, 1-kHz servo loop; Kollmorgen Motion Technologies Group, Commack, NY). The data acquisition and servomotor control were implemented with LabVIEW 5.0 software. Custom real-time C software, accessible from LabVIEW, provided dual-motor servocontrol and data-acquisition for 32 analog channels and 4 encoder channels at a 1-kHz sampling rate.
In the isometric force generation trials, the servomotors fixed both crank arms in place by implementing position control. Each crank was maintained in position independently by its respective servomotor, thus force applied by the subject to one crank had no mechanical effect on the opposite crank or limb.
In the discrete movement trials, the crank arm was moved through an arc of 100° at 100°/s by the servomotors by implementing velocity control. Each servomotor independently moved the crank attached to it at the prescribed velocity in a direction that evoked either limb extension or flexion, so the force applied to the crank by a subject had no mechanical effect on the velocity of either that crank or the opposite crank.
In the pedaling trials, the servomotor emulated a mechanical load similar to that normally encountered by the paretic leg during bilateral pedaling. Servomotor control was implemented to produce loads on the pedaling leg that replicated the dynamics encountered with a conventional coupled-crank ergometer (i.e., flywheel inertia, belt friction, and freewheeling). The emulated frictional load, at 45 rpm, dissipated 60 J.
In both uni- and bilateral pedaling trials, the servomotors were programmed so that the paretic leg pedal force accelerated the paretic leg crank (as well as the nonparetic leg crank in bilateral pedaling); the nonparetic leg crank was servocontrolled 180° antiphased from the paretic leg in bilateral pedaling (i.e., 1 servomotor applied the necessary torque to always position the nonparetic leg crank antiphased to the paretic leg crank such that the nonparetic leg pedal force had no effect on the acceleration of either crank); and the load encountered by the paretic leg was similar to the load a paretic leg encounters during conventional coupled bilateral ergometer pedaling. The load encountered by a pedaling leg during conventional pedaling is not only the load from the ergometer but also the load transmitted through the shared crank axle from the contralateral pedaling leg (e.g., the contralateral leg provides an assistance crank torque to lift the leg against gravity during ipsilateral limb flexion). Therefore because the paretic leg servomotor was programmed to emulate the ergometer load experienced in two-legged pedaling it needed to also include emulation of the crank torque generated by the nonparetic leg in two-legged pedaling. This allows the paretic leg to experience the same mechanical load as during conventional coupled-crank pedaling. A generic nonparetic crank torque profile (i.e., torque as a function of crank angle) was generated from torque profiles previously recorded in two-legged pedaling in pilot work. Two profiles that varied in the amount of assistance provided by the nonparetic leg, and hence the servomotor, were developed for use by the hemiparetic subjects (one required net work of 20 J and the other required net work of 0 J). A third profile that reproduced the contribution of a normal leg was used for all neurologically healthy subjects (required net work of 60 J). In summary, the acceleration of the paretic leg crank always resulted from only the paretic leg pedal force and was unassisted by the actual nonparetic leg pedal force, instead being assisted by a software emulation of generic nonparetic leg torque production. Thus the uni- and bilateral pedaling conditions each allowed pedaling with the usual bilateral motor patterns used for coupled-crank pedaling.
Experimental setup and data collection
Subjects wore cleated ankle braces (DePuy Orthotech, Tracy, CA) on each leg that could either be locked to fix the ankle in neutral dorsiflexion or left unlocked to leave ankle motion unrestricted. The brace was fixed with the ankle locked at 10° of plantarflexion during the isometric trials. During the pedaling trials, the brace was left unlocked and analogous to a cleated cycling shoe. During the discrete movement trials the brace was fixed with the ankle locked at 10° of plantarflexion. With the brace fixed, assuming the pelvis remained stationary, the configuration of the leg (hip and knee angles) was uniquely determined by the crank angle (e.g., Redfield and Hull 1986). Pelvic motion was minimized by including a backboard (parallel to seat tube of ergometer) with shoulder and lap harnesses that stabilized the subject and minimized movement relative to the seat (Fig. 1). During trials of all three tasks, subjects were seated in a reclined posture against a backboard that minimized balance demands while on the ergometer. The cleated braces firmly attached the feet to the pedals, which were instrumented to allow measurement of both fore-aft, shear, and vertical forces (Newmiller et al. 1988). Crank and pedal angles were measured using digital optical encoders.
|
20 kHz). Note that the measurement methodology is consistent with recent studies from our laboratory and have been described previously (Ting et al. 1999, 2000). Experimental protocol
Each experimental session examined three force generation tasks: isometric, discrete movement, and cyclical movement. The primary interest was observing motor output of the paretic limb during different combinations of paretic limb force generation and nonparetic limb force generation.
ISOMETRIC FORCE GENERATION. Twelve randomly assigned trials of isometric force generation were performed. To minimize the possible effects of fatigue, all subjects were provided a minimum of 30-s rest between trials. Due to the nature of the hemiparetic subjects' physical disability, they were allowed to rest as long as they wished. In the hemiparetic subjects, rest between trials was typically between 30 s and 3 min.
The isometric force generation trials involved a randomized complete block design, which involved the combination of four ipsilateral leg (paretic leg of hemiparetic subjects) force generation directions (extension, flexion, anterior, and posterior) in combination with three contralateral leg (nonparetic leg of hemiparetic subjects) force generation states (extension, flexion, and resting). Subjects were instructed to generate one of four isometric forces with the ipsilateral leg: "push down" (extension force), "pull up" (flexion force), "push forward" (anterior force), or "pull back" (posterior force) and simultaneously to produce one of three states with the contralateral leg: "relax" (CL resting), "push down" (CL extension), or "pull up" (CL flexion). Each crank was held stationary by its servomotor at the forward horizontal position (90°). The feet were attached to the cranks by the ankle braces, which were fixed in 10° plantarflexion. With the ankle braced, no muscular effort is required by the contralateral leg to maintain a static configuration. When producing force, subjects were instructed to push at a "substantial, but not maximal level" with each leg simultaneously in the desired directions. They were informed that they should push until told to stop, which would be 5–10 s. Data collection began roughly 2–3 s after the command to initiate force generation. The investigator visually monitored the subject to confirm compliance with the test condition prior to initiating data acquisition. Three seconds of data were collected.
DISCRETE MOVEMENT.
Six discrete movement trials were performed according to a randomized complete block design. Concentric ipsilateral (paretic leg of hemiparetic subjects) leg movements were tested in two directions (extension and flexion) in combination with three contralateral (nonparetic leg of hemiparetic subjects) leg movement states (CL extension, CL flexion, and CL resting) in a block (2 directions x 3 states yields 6 trials). The presentation order of the six trials was randomized. Again, the feet were attached to the pedals using the ankle braces set to 10° plantarflexion. Subjects were instructed that the crank would not move immediately but would begin moving after 1 s. The subjects were instructed to either push down (extension movement) or pull up (flexion movement) at a "substantial but not maximal level" and to continue pushing (pulling) until the movement ceased. Cranks were held stationary by the servomotor at 20 and 160° for extension and flexion, respectively, and were moved simultaneously for 1 s at 140°/s (i.e., final positions of 160 and 20°, respectively). Note that the discrete flexion condition (i.e., from 160 to 20° with the direction of crank rotation reversed) is not identical to the flexion as performed in pedaling (i.e., from 200 to 340°). The investigator visually monitored the subject to confirm compliance with the test condition and then initiated data collection within 2–3 s. Data were collected for a 6-s period.
PEDALING.
Two randomly assigned trials of pedaling were performed. The pedaling trials involved the ipsilateral leg (paretic leg of hemiparetic subjects) pedaling against one mechanical load and two contralateral (nonparetic leg of hemiparetic subjects) leg states (CL pedaling or CL resting). When the contralateral leg was resting, the crank was held stationary by its servomotor at the forward horizontal position (90°) with the ankle brace set to 10° plantarflexion (independent of leg, the ankle brace was always unlocked during pedaling). When the contralateral leg was pedaling, the crank was offset 180° antiphased from the paretic leg crank using the position-servocontrol (note that the ipsilateral leg crank was mechanically isolated as in Experimental apparatus to eliminate mechanical coupling). An audible metronome tone was provided until a steady cadence was established at 45 rpm (10 s), then the metronome was turned off and data collection initiated. Subjects were instructed to pedal until instructed to stop, which would be about thirty seconds. Data collection began 2–3 s after the metronome was halted. Data were collected for 20 s to ensure that 10 consecutive revolutions of data in each trial would be available for processing.
Data processing
ISOMETRIC FORCE GENERATION. For each of the four muscles in each leg, average EMG magnitude was calculated bilaterally for the entire 3-s trial. These average values per condition were normalized by the greatest average EMG that occurred in any of the four unilateral paretic leg isometric force generation trials (i.e., contralateral leg resting) so that the average EMG for a particular muscle would be 1.0 for the isometric force direction that preferentially excites that muscle. Values >1.0 represent facilitation and can only occur when the nonparetic leg is also performing isometric force generation.
DISCRETE MOVEMENT.
For discrete movement trials, the average pedal forces and average EMG were only calculated during the 1 s in which the 140° movement occurred. For each muscle, these EMG data were normalized by the same value as in the preceding text for the isometric force generation trials. Thus similar values indicate similar average EMG during discrete movement and isometric force generation trials.
PEDALING.
Pedal force and EMG data from each trial of each subject were averaged as a function of crank angle over 10 consecutive crank revolutions. The average rectified EMG between 180 and 270° was calculated for each muscle and normalized by the same factor determined from the isometric force generation conditions (we define this activity in the 3rd quadrant of pedaling as RF3, VM3, BF3, and SM3). These values correspond to previously reported measures of impaired EMG timing during pedaling in persons with poststroke hemiparesis (Kautz and Brown 1998). The average tangential crank force (that component of the pedal force that is oriented perpendicular to the crank arm and acts to rotate the crank) was also calculated as a function of crank angle. The average tangential crank force between 180 and 270° was also calculated. This measure is related to the mechanical work output of the pedaling leg (work = average tangential crank force * crank length * angular distance) for that range of the pedaling cycle.
Data analysis
Four different types of statistical analyses were performed (JMP version 5.0, SAS, Cary, North Carolina). First, Student's t-test (P < 0.05) was used to test for differences in the EMG activity means between the hemiparetic and control groups for the unilateral conditions during isometric force generation and discrete movement, and for RF3, VM3, BF3, and SM3 during bilateral pedaling. Second, paired t-test (P < 0.05) were used to test for differences between isometric force generation, discrete movement and pedaling conditions within each group. Note that we only compared the CL-extension and CL-flexion condition with the CL-resting condition and not with each other. This is because we sought to elucidate differences between uni- and bilateral coordination and not differences between in-phase and reciprocal bilateral coordination. Third, Spearman's rank correlation (P < 0.05) was used to test for associations between level of motor recovery and EMG timing abnormalities during pedaling. Finally, regression analysis was used to test for correlations (P < 0.05) between changes in EMG timing abnormalities and changes in EMG amplitude.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Example raw EMG data during isometric force generation conditions are shown (Fig. 2) for control and hemiparetic subjects during CL-rest conditions. Each recorded muscle was preferentially excited (i.e., achieved greatest EMG) in the same isometric force generation condition (Fig. 3) with RF preferentially excited during anterior force generation, VM during extension force generation, and BF and SM during posterior force generation. The primary inter-group differences related to increased relative excitation by the hemiparetic subjects in muscles not preferentially excited in a particular force generation condition, as hemiparetic subjects exhibited increased excitation during anterior, extension and flexion force generation in BF; increased excitation during anterior, extension and flexion force generation in SM; and increased excitation during posterior force generation in RF (dagger P < 0.05, Fig. 3). Thus while hemiparetic subjects predominantly excited the same muscles to generate isometric force in each position, they were less likely to show selectively reduced excitation in the other muscles.
|
|
In hemiparetic subjects, ipsilateral (paretic) leg coordination of isometric extension and flexion force generation (IL extension and IL flexion), particularly excitation of the bifunctional muscles (RF, BF, and SM), differed depending on whether the contralateral (nonparetic) leg concurrently generated extension or flexion force (CL extension and CL flexion). This can be observed in the raw EMG data (Fig. 2) and the group data (Fig. 3). During IL extension, CL-flexion force increased paretic BF (89 vs. 56%, P = 0.002) and SM (84 vs. 55%, P = 0.01) excitation as compared with nonparetic extension force. During IL flexion, CL-extension force increased paretic RF (88 vs. 51%, P = 0.003) excitation. Similar relationships were found in the control subjects (Fig. 3). During IL extension, CL-flexion force increased paretic BF (34 vs. 26%, P = 0.013) excitation while during IL flexion CL-extension force increased paretic RF (67 vs. 47%, P = 0.02) excitation. However, additional relationships were also observed in the control subjects during IL flexion, with CL-flexion force increasing paretic BF (62 vs. 29%, P = 0.004) and SM (71 vs. 39%, P = 0.0009) excitation.
Ipsilateral leg coordination of anterior and posterior isometric force generation was not significantly different when the CL leg was generating extension versus flexion force. Nevertheless, there were trends in the excitation of the bifunctional muscles similar to those seen during IL-extension and IL-flexion force generation that did not reach statistical significance (Fig. 3). During anterior force generation, paretic RF excitation was increased by CL extension [113 vs. 88%, not significant (NS); P = 0.07], whereas during posterior force generation, paretic BF and SM excitation was increased with CL flexion [124 vs. 98%, NS (P = 0.105) and 119 vs. 98%, NS (P = 0.16), respectively]. Control subjects performing anterior force generation with CL-flexion force generation exhibited significant increases in BF (34 vs. 14%, P = 0.01), SM (27 vs. 12%, P = 0.02), and VM (103 vs. 64%, P = 0.006).
In summary, all interlimb influences observed in the four different paretic leg force generation conditions in both hemiparetic and control subjects were consistent with a relative facilitation of RF by CL extension and of BF and SM by CL flexion.
Discrete movement
When comparing the discrete movement CL-resting condition to the CL-resting condition isometric force generation, there were no obvious relative differences in the average EMG in any of the muscles (Fig. 4). Ipsilateral leg (paretic leg in hemiparetic subjects) coordination of active discrete extension and flexion movements was altered when the contralateral leg concurrently generated extension or flexion movements (Fig. 5). Although statistical significance was not achieved for several conditions, the trend of changes in RF, BF, and SM were identical to those in coordination of isometric force generation. During IL extension, CL flexion increased excitation of paretic BF (86 vs. 53%, P = 0.009). The changes in SM [66 vs. 55%, NS (P = 0.078)] did not achieve significance. During IL flexion, CL extension increased excitation of paretic RF (82 vs. 58%, P = 0.01). Similar trends were observed in the control subjects. CL flexion increased SM (46 vs. 26%, P = 0.043) excitation during IL extension. Changes in BF [41 vs. 22%, NS (P = 0.075)] did not achieve significance. During CL extension, the changes in RF [57 vs. 37%, NS (P = 0.064)] excitation during IL-flexion were not significant.
|
|
Consistent with previous studies of bilateral pedaling, paretic leg coordination was characterized by prolonged VM excitation, relatively increased flexion phase activity in RF, and relatively decreased flexion phase activity in BF and SM (Fig. 6). Previously, Kautz and Brown (1998) identified prolonged VM excitation (quantified as increased VM3) and phase-advanced RF, BF, and SM excitation (quantified as increased RF3, decreased BF3 and decreased SM3, respectively, as defined in Data processing) as timing abnormalities in the paretic leg and showed their importance by correlating them with reduced motor performance (total work done by paretic leg). Consistent with previous observations during bilateral pedaling, in the present study, the hemiparetic subjects showed increased VM3 (18 vs. 2%, P = 0.0003), increased RF3 (21 vs. 11%, P = 0.02), and reduced SM3 (24 vs. 32%, P = 0.003) compared with the control subjects. The change in BF3 [22 vs. 25%, NS (P = 0.26)] was not statistically significant.
|
|
|
|
While we do not believe that the same EMG coordination measures should be considered timing abnormalities when comparing bilateral to unilateral pedaling in the control subjects (Fig. 10), we note that the decreases in VM3 (2 vs. 3%, P = 0.02) and RF3 (11 vs. 16%, P = 0.001) would be considered better coordination during bilateral pedaling, whereas increases in SM3 (39 vs. 32%, P = 0.025) and BF3 [29 vs. 25%, NS (P = 0.18)] would be considered worse coordination during bilateral pedaling, if they occurred in the hemiparetic subjects. The overall amplitude of EMG activity (average EMG during entire pedaling cycle) excitation was increased for VM in bilateral pedaling; however, this was the only significant difference when comparing bilateral to unilateral pedaling [RF: 0.75 vs. 0.81, NS (P = 0.49); VM: 0.85 vs. 0.68, P = 0.03; BF: 0.53 vs. 0.55, NS (P = 0.77); and SM: 0.39 vs. 0.41, NS (P = 0.68)].
|
Effects of level of motor recovery
The observed EMG timing abnormalities during bilateral pedaling were associated with the level of motor recovery as assessed by the "synergy performance" portion of the Fugl-Meyer assessment (Fugl-Meyer et al. 1975). The synergy sub score corresponds with the ability of the lower limb to move within, combine, or move outside of "extensor/flexor synergy patterns" (Brunnstrom 1970). We have previously used synergy performance score to test for associations between the level of motor recovery and motor performance (Kautz and Brown 1998) and believe that it more directly assesses the motor abilities of a subject than does the total Fugl-Meyer for the lower limb. VM3 (P = 0.01) and RF3 (P = 0.0005) were each negatively associated with motor recovery (i.e., increased motor recovery was associated with decreased VM3 and RF3) and SM3 [NS (P = 0.057)] was positively associated with motor recovery (i.e., increased motor recovery was associated with increased SM3). There was no association with motor recovery for BF3 [NS (P = 0.36)].
The observed exacerbation of prolonged VM activity in bilateral pedaling versus unilateral pedaling (measured by subtracting VM3 for unilateral pedaling from VM3 for bilateral pedaling) was also negatively associated with motor recovery (P = 0.02). Increased amplitude of VM was also negatively associated with motor recovery (P = 0.03). There was no association between motor recovery and changes in excitation as assessed by in RF3 [NS (P = 0.18)], BF3 [NS (P = 0.29)], and SM3 [NS (P = 0.44)]. Indeed, the associations for the hamstrings were actually negative; the opposite of those expected if abnormal excitation had been exacerbated.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
). However, although these interlimb influences appeared to be operational and likely contributed to exacerbation of coordination deficits in bilateral pedaling of hemiparetic subjects, there was little evidence of similar interlimb influences during bilateral pedaling of control subjects. Thus it is possible that suppression of these interlimb influences, which are observed in isometric and discrete movement, may occur during pedaling movements in neurologically healthy subjects. The current data also strongly suggest that this suppression is absent or disrupted in persons with hemiparesis. In addition to exacerbated abnormal excitation in the bifunctional muscles, participation of the nonparetic leg (i.e., bilateral pedaling) also coincided with prolonged excitation of the VM. Although no similar influence was observed in isometric force generation or in discrete movement, this is consistent with our observations that participation of the nonparetic leg exacerbated coordination deficits during bilateral pedaling. Abnormal synergies and associated movements in hemiparesis
The common clinical observation that hemiparetic subjects often exhibit alternation of mass flexion and extension synergies (e.g., Twitchell 1951) has inspired clinical interpretations of hemiparetic gait (Brunnstrom 1970; Perry 1969; Perry et al. 1978). We originally hypothesized that the performance of lower limb tasks would be strongly influenced, perhaps even limited, by these synergies.
Our hypothesis that paretic leg coordination would show increased excitatory reciprocal influences of nonparetic flexion or extension force generation was not supported during isometric force generation and discrete movement. A strong reciprocal link had been suggested between the flexion and extension synergies of the paretic leg and flexion and extension activity of the nonparetic leg (Brunnstrom 1970). However, because we observed similar interlimb influences in neurologically healthy subjects, the influences of nonparetic leg activity on the paretic leg coordination observed during reciprocal bilateral isometric force generation in the hemiparetic persons were not unique.
Isometric and discrete movement conditions
Our main finding from isometric force generation and discrete movement trials was that excitation of the paretic limb hamstrings group during extension force was facilitated by contralateral flexion force generation. Similarly, excitation of the paretic RF during flexion force was facilitated by contralateral extension force generation. Importantly, for hemiparetic subjects, this facilitation of the paretic hamstrings group by contralateral flexion force during extension force generation was operational only during reciprocal limb use. Unlike neurologically healthy control subjects, paretic limb hamstrings activity was not facilitated by contralateral flexion force during paretic flexion force generation. Thus the paretic leg did not exhibit novel interlimb influences due to the activity of the nonparetic leg during isometric force generation and discrete movement.
Our data support our hypothesis that coordination of the paretic leg RF and hamstrings group muscles during isometric force generation and discrete movement would show specific patterns of facilitation due to contralateral flexion and extension that are consistent with EMG timing abnormalities observed in pedaling. Abnormal EMG excitation during pedaling (Kautz and Brown 1998) was found in BF and SM during leg extension (coincident with nonparetic leg flexion) and in RF during leg flexion (coincident with nonparetic leg extension). The current study confirmed that the interlimb influences during isometric force generation were consistent with the EMG timing abnormalities in pedaling, as BF and SM were facilitated during paretic extension by contralateral flexion force and RF was facilitated during paretic flexion by contralateral extension force.
Excitation of bifunctional thigh muscles
Bifunctional muscles are defined anatomically as biarticular muscles (i.e., crossing two joints) and classified as a flexor at one joint and an extensor at the other (e.g., Ingen Schenau et al. 1994; Pratt et al. 1991, 1996). Note that RF, BF, and SM are all bifunctional muscles. In acute decorticated cats that exhibit spontaneous locomotor activity, bifunctional muscles demonstrate complex activation patterns that can occur during parts of both the flexor and extensor bursts (Perret and Cabelguen 1980). During pedaling by neurologically healthy subjects, the excitation of ipsilateral bifunctional muscles is significantly modulated by contralateral motor activity (Kautz et al. 2002; Ting et al. 2000).
We have previously emphasized the importance of bifunctional muscles for transitions between limb flexion and extension (Kautz et al. 2002; Ting et al. 1999). The excitation of bifunctional thigh muscles is considered to be critical for the smooth execution of whole limb movements (Ingen Schenau et al. 1992, 1994). During normal locomotion, the activity of bifunctional thigh muscles shows both a complexity and mutability that is markedly increased compared with that shown by the unifunctional extensors and flexors (Pratt et al. 1996). Because the bifunctional muscle excitation appears to be closely linked with the intersegmental dynamics and external forces (Pratt et al. 1996), central influences are thought to be less important than peripheral influences. It has thus been suggested that a differential distribution of motion- and force-related afferent information to bi- and unifunctional muscles allows this peripheral input to play a critical role in shaping bifunctional muscle excitation (Ingen Schenau et al. 1994). Impairment of the processing of peripheral afferent information in persons with hemiparesis might contribute to the abnormal bifunctional muscle activity. From our previous pedaling biomechanics research (e.g., regions of positive muscle power in Fig. 5 of Neptune et al. 2000), we estimate that the periods of muscle lengthening are approximately from 250 to 110° for BF and SM and from 100 to 270° for RF. Thus there appears to be no simple relationship with stretch reflexes as the abnormal activity typically begins in mid-stretch, and it usually continues during shortening. The abnormal activity seems more closely tied to the limb flexion and extension cycle, which is also approximately coincident with extension and flexion of the hip. Neurophysiological studies have shown that afferent information related to hip extension is extremely important for swing phase initiation (Grillner 1981; Grillner and Rossignol 1978).
Despite the clinical focus on flexion and extension synergies, little previous research had focused on documenting the participation of individual paretic leg muscles in the synergies, especially the bifunctional muscles that are not unambiguously classified as flexors or extensors. This study shows that both extension and flexion isometric force generation and discrete movement result in excitation of paretic hamstrings and RF. Hemiparetic persons demonstrate a reduced ability to selectively recruit these bifunctional muscles when compared with neurologically healthy persons. Thus these muscles cannot be considered to belong exclusively to hemiparetic flexion or extension synergies for isometric force generation and discrete movement. Whereas only those subjects in Brunnstrom's stage 3 would be assumed to be restricted to use of the synergies in voluntary control, subjects at stages 4 and 5 are also assumed to manifest an exaggerated influence of these synergies. Note that our protocol does not determine an absolute level of excitation (cf., maximum isometric force), so it is possible that the recruitment level of the bifunctional muscles is low in all trials. Nevertheless, the measures do reflect the relative excitation of the muscles in each condition.
The interlimb influences observed in the hemiparetic and neurologically healthy subjects are consistent with the crossed extension reflex, where application of a stimulus that evokes the flexion reflex in the ipsilateral leg, also produces (facilitates) extension of the contralateral leg (Sherrington 1910). Furthermore, there is a pattern of excitation and inhibition to both legs that is reciprocally organized such that the muscles excited in contralateral extension are inhibited in ipsilateral flexion and those inhibited in contralateral flexion are excited in the ipsilateral extension. In humans, the rectus femoris is associated with limb flexion, and semimembranosus with extension, respectively, although the activity of semimembranosus is more variable (Hagbarth 1960).
Interlimb influences during rhythmic motor activity
Specific deficits in bilateral coordination of rhythmic motor activity (pedaling) observed in hemiparetic subjects appear to be impairments of the neural (interlimb) coupling and to contribute to impaired intralimb coordination on the paretic side. Normally, neural mechanisms for interlimb coupling facilitate coordination between the legs and enable execution of locomotor activities. Our data indicate that in poststroke hemiparesis, impaired coupling results in exaggerated influence of the nonparetic leg on intralimb coordination of the paretic leg. Specifically, results of the present study indicate that nonparetic leg activity reinforces and exaggerates both previously documented prolonged excitation of VM and produces an excitatory effect on antagonistic function of the paretic leg (e.g., CL extension facilitates CL flexion) in the bifunctional RF, BF, and SM. As a result, the motor coordination of the paretic leg is more impaired during bilateral than unilateral locomotor activity.
Understanding the uni- versus bilateral pedaling performance of the hemiparetic subjects requires an understanding of the differences found in neurologically healthy individuals. The results from the age-matched neurologically healthy subjects in this study (age = 59 ± 8 yr) are consistent with work previously conducted in younger neurologically healthy subjects (31 ± 7 yr in Kautz et al. 2002; Ting et al. 2000) in that the unilateral motor pattern showed increased contribution from the bifunctional muscles and a concomitant decreased contribution from the VM, resulting in less resistance to leg flexion during the pedaling upstroke (Fig. 10). Given that the common paradigm in all of these studies was to maintain biomechanical equivalence for unilateral and bilateral pedaling (i.e., workload and speed), it is axiomatic that when a steady-state pattern is achieved, increased output in some muscles needs to be compensated with decreased output in other muscles, and vice versa. Otherwise, increased power is generated by muscles such that movement speed and/or external mechanical work done by the leg would increase. Because we observe the steady-state output, it is not possible to determine if increased VM activity and decreased bifunctional muscle activity are both primary changes or if one is a secondary response while achieving steady-state in the face of the other primary response.
Our most important result is that EMG abnormalities in VM, BF, and SM previously defined in hemiparetic subjects were exacerbated in bilateral pedaling such that much more resistance to leg flexion occurred during the upstroke phase. For the individual subjects, there was also a strong relationship between exaggerated and prolonged excitation of VM (VM3) and increased VM amplitude. Subjects who produced little change in VM3 actually decreased VM amplitude. Because the aberrant increases in VM3 were also negatively correlated with stages of motor recovery (i.e., less recovered subjects showed greater increases in VM3), one possible explanation of the decreased VM amplitude in the better performing hemiparetic subjects (in contrast to increased VM amplitude shown by control subjects) might be that the increased hamstrings activity during the extension phase observed in bilateral pedaling (i.e., BF3 and SM3 decrease while overall BF and SM increase in Fig. 8) needed to be balanced by decreased motor activity elsewhere. Because workload between uni- and bilateral pedaling is kept approximately constant by the experimental paradigm, increased resistance during leg flexion has to occur in tandem with increased work during leg extension.
While our paradigm was not designed to directly observe changes in EMG activity that result in a net change in mechanical output of the paretic leg, one possible explanation consistent with our data is that motor output of the paretic leg during bilateral pedaling may be facilitated relative to its output in unilateral pedaling through interlimb pathways that couple pattern generation in the two legs (e.g., Duysens et al. 2004). In a recent review of interlimb coordination during walking, Duysens et al. (2004) suggested that current evidence in neurologically healthy persons was consistent with the premise that contralateral motor activity facilitates ipsilateral motor activity (i.e., bilaterally coupled pattern generators controlling each leg). This facilitation may have an even greater effect in persons with hemiparesis because they may have a decreased ability to activate pattern generation on the paretic side. Recent evidence in hemiparetic walking suggests that a generally reduced ability to excite muscles on the paretic side may be the primary impairment (Mulroy et al. 2003). Thus it may be the case that more impaired subjects have greater difficulty recruiting the paretic leg for a unilateral task, that this difficulty may correspond with the severity of impairment, and further, that bilateral tasks facilitate paretic leg output. However, this putative increased output in bilateral pedaling is not just a simple increase in the paretic leg motor pattern for unilateral pedaling. Rather, the paretic leg motor pattern is more disrupted as evidenced by more prolonged VM activity as the amplitude increased. Whether this increased recruitment in the paretic leg is the result of nonparetic leg efferent activity or the associated nonparetic leg afferent activity cannot be determined with our protocol. An alternative explanation to increased output of paretic leg pattern generation would be that spinal mechanisms modulate the gain of transmission of afferent input between the nonparetic and paretic leg motoneuron pools (via both afferent and descending pathways) differentially when the nonparetic leg is isometric as opposed to moving dynamically (e.g., similar to effects in normals in Peper and Carson 1999).
Similar to our results on the influence of contralateral sensorimotor state in the pedaling condition, recent studies have also looked at unilateral and bilateral stepping in clinically complete spinal cord injured (SCI) patients. Dietz et al. (2002) used a driven gait orthosis on a treadmill to induce stepping movements in patients with complete SCI and in neurologically healthy subjects. Unilateral locomotion in the patients was associated with a normal pattern of leg muscle EMG activity restricted to the moving side, whereas in the healthy subjects, a bilateral activation occurred, which they interpreted as evidence that interlimb coordination depends on supraspinal input. In another study, Ferris et al. (2004) found that rhythmic contralateral lower limb loading or movement could produce rhythmic muscle activation in the ipsilateral limb even when it is stationary and unloaded. Kawashima et al. (2005) also investigated unilateral and bilateral coordination of a locomotor-like movement in subjects with complete SCI and found that reciprocal contralateral leg motion amplified induced ipsilateral locomotor-like activity. Thus the results from both Ferris et al. (2004) and Kawashimi et al. (2004) suggest that interlimb coordination may not depend on supraspinal input. Regardless of exact role of supraspinal input, these studies in stepping are consistent with our results in pedaling and further establish the importance of contralateral sensorimotor input during locomotion.
The corticoreticular-reticulospinal-spinal interneuronal system (Matsuyama et al. 2004) has many characteristics that suggest both a potential role in bilateral coordination deficits and an avenue for contralesional neural pathways to contribute to the paretic motor pattern. Contralateral reticulospinal descending pathways that cross the midline have been implicated in dramatic improvements in ipsilateral locomotor function and corresponding electrophysiological measurements observed immediately after spinal cord hemisection in rats (Fujiki et al. 2004). Even more intriguing for the results of our study is that manipulation of contralateral spinal-level motor output (i.e., brachial root transection) facilitated ipsilateral locomotor improvements immediately after hemisection. Fujiki et al. (2004) hypothesized this phenomenon resulted from an uninjured, initially latent pathway mediating the locomotor recovery in an analogous manner to the crossed phrenic phenomenon (Goshgarian 2003) seen in the respiratory phrenic nerve. Jankowska et al. (2003) demonstrated in cats that although the direct actions of reticulospinal neurons were much more potent on ipsilateral motoneurons (note direct contralateral actions of crossed axon collaterals were found in 10% of fibers), interneuronally mediated actions were as potent contralaterally as ipsilaterally and midlumbar commissural neurons likely contribute. Also in cats, a population of midlumbar commissural interneurons (both excitatory and inhibitory) has been identified that project only to contralaterally located neurons (including motoneurons), and it was suggested that they may be important in the recruitment of groups of contralateral muscles that subserve a variety of motor synergies (Bannatyne et al. 2003). While much has been learned about reticulospinal control of locomotion, very little is known about the relative importance of these pathways for either bilateral coordination deficits or the recovery of hemiparetic locomotion after the cerebral lesion of stroke (cf., spinal lesions).
It is important to realize that what appears to be a specific bilateral coordination deficit in a person with chronic stroke might actually represent a substantial improvement in coordination from what was originally possible more immediately after the lesion. Depending on the structures damaged by the lesion, restitution of normal function may not be possible, and thus the observed bilateral coordination pattern, while aberrant, might represent marked reactive neural plasticity that subserves locomotor function. Future work needs to determine mechanisms underlying the recovery of locomotor function.
Clinical relevance
The clinical relevance of this study lies in the presentation of novel evidence of disruptions to bilateral coordination of locomotion following stroke. Paretic leg coordination deficits during locomotion do not appear to be strictly a unilateral phenomenon. Unilateral coordination appears to be much less impaired than bilateral coordination, although notable aberrations in rhythmic unilateral movements are still evident, albeit at a reduced scale. These findings may be an argument for task-specific (including bilateral) approaches to rehabilitation. Because some coordination deficits may only be manifest in bilateral tasks, activities designed to improve coordination may need to be performed bilaterally. It may also be an argument for staged or progressive advancement through approaches from unilateral-isolated to bilateral less or unconstrained movements. There are potential benefits to be gained from each approach, and it is important to recognize the physiological manifestations produced through each individual approach. That hemiparetic severity and aberrant motor activity correlated strongly suggests that some patients may be ready to progress to the task-specific approaches, whereas others may benefit from working systematically through stages (i.e., learning to activate appropriate patterns and inhibit unwanted activity, learning bilateral activation, and then incorporating these patterns into movement).
|
GRANTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
FOOTNOTES |
|---|
Address for reprint requests and other correspondence: S. A. Kautz, Brain Rehabilitation Research Center, Malcom Randall VA Medical Center (151A), 1601 SW Archer Rd., Gainesville. FL 32608-1197 (E-mail: skautz{at}phhp.ufl.edu
, significant differences between mean EMG activity during CL rest in hemiparetic and control subjects for the same isometric force generation conditions. *, significant differences for either CL Ext or CL Flex in comparison to CL Rest (P < 0.05). Note that the trend in both groups is for CL Ext (
) to facilitate RF excitation during flexion and CL Flex (
) to facilitate BF and SM excitation during extension.
