Journal of Electromyography and Kinesiology 24 (2014) 172–177
Contents lists available at ScienceDirect
Journal of Electromyography and Kinesiology journal homepage: www.elsevier.com/locate/jelekin
Loading and knee flexion after stroke: Less does not equal more Mary Murray a,b, Amy Hardee c,d, Richard L. Goldberg c, Michael D. Lewek a,e,⇑ a
Division of Physical Therapy, Department of Allied Health Sciences, University of North Carolina at Chapel Hill, USA Cleveland Clinic, Cleveland, OH, USA c Joint Department of Biomedical Engineering, University of North Carolina at Chapel Hill and North Carolina State University, USA d Teleflex Medical, Durham, NC, USA e Human Movement Science Program, University of North Carolina at Chapel Hill, USA b
a r t i c l e
i n f o
Article history: Received 19 June 2013 Received in revised form 22 August 2013 Accepted 10 October 2013
Keywords: Stroke Gait Force feedback Stiff-knee gait Rehabilitation
a b s t r a c t It is believed that force feedback can modulate lower extremity extensor activity during gait. The purpose of this research was to determine the role of limb loading on knee extensor excitability during the late stance/early swing phase of gait in persons post-stroke. Ten subjects with chronic hemiparesis poststroke participated in (1) seated isolated quadriceps reflex testing with ankle loads of 0–0.4N m/kg and (2) gait analysis on a treadmill with 0%, 20% or 40% body weight support. Muscle reflex responses were recorded from vastus lateralis (VL), rectus femoris (RF), and vastus medialis (VM) during seated testing. Knee kinematics and quadriceps activity during late stance/early swing phase of gait were compared across loading conditions. Although isolated loading of the ankle plantarflexors at 0.2 N m/kg reduced VM prolonged response (p = 0.04), loading did not alter any other measure of quadriceps excitability (all p > 0.08). During gait, the use of BWS did not influence knee kinematics (p = 0.18) or muscle activity (all p > 0.17) during late stance/early swing phase. This information suggests that load sensed at the ankle has minimal effect on the ipsilateral quadriceps of individuals post-stroke during late stance. It appears that adjusting limb loading during rehabilitation may not be an effective tool to address stiffknee gait following stroke. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction Following stroke, gait is often marked by decreased knee flexion of the paretic limb during swing phase; a pattern called ‘‘stiff-knee gait’’ (Kerrigan et al., 1991). While the exact cause of stiff-knee gait is unknown, abnormal muscle timing of the quadriceps has been implicated (Kerrigan et al., 1991; Lewek et al., 2007; Waters et al., 1979). Normally, quadriceps activation occurs during the early to mid-stance period of gait; however, individuals poststroke often demonstrate prolonged quadriceps activity (Kautz and Brown, 1998). The presence of prolonged quadriceps activity into late stance has been associated with stiff knee gait in children with cerebral palsy (Goldberg et al., 2004). Muscle activity during gait is modifiable by sensory receptors throughout the lower extremity. In particular, force (i.e., load) feedback to the central nervous system (CNS) from Golgi tendon organs (group Ib) serves a critical role in regulating extensor activ-
⇑ Corresponding author. Address: University of North Carolina at Chapel Hill, 3043 Bondurant Hall, CB#7135, Chapel Hill, NC 27599-7135, USA. Tel.: +1 919 966 9732. E-mail address: [email protected] (M.D. Lewek). 1050-6411/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jelekin.2013.10.006
ity duration during stance (Dietz et al., 2002; Hiebert and Pearson, 1999). Prolonged mechanical loading to the ankle extensors, for instance, elicits prolonged muscle activity in the homonymous muscle of decerebrate walking cats (Crone et al., 1988; Hiebert and Pearson, 1999). Humans also appear to exhibit similar positive autogenic force feedback (Dietz et al., 2002; Gordon et al., 2009; Yang et al., 2004), suggesting that the CNS is particularly sensitive to plantarflexor force feedback, and neuromotor output can be directly influenced by these signals (Ada et al., 2010). Although force feedback appears to excite homonymous muscles during gait (Conway et al., 1987; Duysens and Pearson, 1980), there is evidence for heterogenic inhibitory-force feedback from the plantarflexors to the quadriceps (Ross and Nichols, 2009; Wilmink and Nichols, 2003). This would suggest that increased limb loading (particularly into late stance) would increase the Ib sensory signals from the plantarflexors, to inhibit quadriceps activity during late stance to allow for improved knee flexion. During clinical practice, however, walking on a treadmill with body weight support (BWS) has been used extensively following stroke to facilitate walking recovery (Ada et al., 2010; Barbeau and Visintin, 2003; Visintin et al., 1998). While not shown to be superior to other forms of therapy (Duncan et al., 2011), gait training with BWS has demonstrated improvements in gait speed, endurance,
173
M. Murray et al. / Journal of Electromyography and Kinesiology 24 (2014) 172–177
and balance (Barbeau and Visintin, 2003; Visintin et al., 1998). Evidence from unimpaired adults (Colby et al., 1999; Dietz et al., 2002; Finch et al., 1991) and individuals post-stoke (Hesse et al., 1999) suggests that the use of BWS will decrease quadriceps activity during early stance. The effect on late stance, however, when plantarflexor load is greatest, remains unknown. In addition, the use of BWS appears to decrease swing phase knee flexion of unimpaired individuals (Finch et al., 1991; van Hedel et al., 2006), representing a potentially unwanted side effect of BWS for individuals post-stroke at a time when knee flexion may already be reduced. The purpose of this study was to characterize the heterogenic force-feedback response from the plantarflexors to the quadriceps in individuals with chronic (>6 months) stroke. We hypothesized that a simulated ankle load during seated isolated joint reflex testing would decrease knee extensor excitability. We then wanted to determine the presence of a relationship between ankle-load mediated reflex control of the quadriceps and the generation of abnormal quadriceps activity during walking in individuals post-stroke. Based on animal models that manipulated limb load during gait (Hiebert and Pearson, 1999), we hypothesized that walking with BWS would reduce any abnormal late stance/early swing phase quadriceps activity and consequently improve swing phase knee flexion for individuals with chronic (>6 months) stroke. Although all joints in the lower extremity are loaded and unloaded during gait, the ankle extensors are believed to contain the critical Ib afferents to modify gait (Dietz and Duysens, 2000), and thus we expected to observe a direct relationship between plantarflexorapplied, limb load mediated reflex excitability of the quadriceps and the late stance/early swing phase quadriceps activity during gait in individuals post-stroke.
2. Methods Ten subjects with chronic hemiparesis (>6 months) resulting from a hemorrhagic or ischemic stroke were recruited for testing (see Table 1). Subjects exhibited a variety of walking patterns (Mulroy et al., 2003) but were able to step independently on a treadmill. Subjects were excluded from the study if they responded affirmatively to questions about confounding medical issues (e.g., cardiac arrhythmia, uncontrolled seizures or hypertension) that would preclude the ability to ambulate safely on a treadmill, bilateral stroke involvement, vestibular or cerebellar involvement, or other neurologic and musculoskeletal disorders affecting the legs. Subjective clinical judgment was used to determine the presence of other cognitive or communication problems that inhibited understanding of study involvement. Prior to participation, all subjects signed an informed consent form approved by the University of North Carolina at Chapel Hill Institutional Review Board. All subjects performed the two parts (isolated joint reflex testing, followed by an instrumented gait analysis) of the study during a single session.
2.1. Reflex testing and analysis Isolated reflex testing was performed to determine the unique contribution of limb load (as it is sensed by the ankle plantarflexors) on knee extensor reflex excitability in a controlled setting (Wu and Schmit, 2006). Prior to testing, active surface electrodes (Motion Lab Systems, Baton Rouge, LA) were applied to the rectus femoris (RF), vastus lateralis (VL), vastus medialis (VM), and medial hamstrings (MH) to monitor reflex response. Subjects sat on an isokinetic dynamometer (Humac, CSMi, Stoughton, MA) with the back supported and the hips flexed to 80°. The paretic knee’s flexion/extension axis was aligned with the Cybex motor axis and the paretic foot was secured to the footplate of a custom designed ankle loader (see Fig. 1). The ankle loader simulated weight bearing during stance through the production of a dorsiflexion load (Wu and Schmit, 2006). An air powered linear actuator generated and transmitted the load to the footplate. A load cell (MLP-100, Transducer Techniques, Temecula, CA), in series with the actuator, monitored the dorsiflexion load applied to the subject. Dorsiflexion loads were normalized to the subject’s body mass, and were set at 0, 0.2, and 0.3–0.4 N m/kg. Some subjects had difficulty tolerating 0.4 N m/kg, so a lower load was used (i.e., 0.3 N m/kg) such that all subjects denied any discomfort during testing. These loads represent a portion of the ankle moments experienced during walking (Lewek, 2011), but are comparable to the passive dorsiflexion moment required to achieve maximum dorsiflexion (Schindler-Ivens et al., 2008). During each loading condition, ramp-hold stretches at the knee joint were performed by passively flexing the knee from 30° to 90° of flexion at 180 and 300°/s. These knee angles are slightly greater than what is observed during ‘normal’ gait, however, we chose these angles to ensure an adequate stretch of the quadriceps, and to avoid the possibility of hamstring tightness in a more extended position. The speeds were chosen to simulate joint speeds similar to those experienced during gait (Granata et al., 2000), as higher velocity stretches produce higher tonic stretch reflexes (Crone et al., 1988). Therefore, there were a total of 6 randomized conditions (three loads and two speeds). Prior to each trial, subjects pre-activated the hip flexors (approx 10%MVIC) (Lewek et al., 2007) to facilitate the production of RF responses and simulate the hip flexion force that occurs during the stance – swing transition of gait. Subjects were asked to maintain the pre-activated level of effort through the duration of the trial. During each trial, the knee’s angular position, velocity, and torque were recorded from the Cybex. EMG signals were bandpass filtered between 20 and 500 Hz and sampled at 1000 Hz to a personal computer. Analysis of EMG signals was performed with custom software (Labview 2009, National Instruments, Austin, TX). A linear envelope was created by full-wave rectifying the data, followed by low-pass filtering with a 20 Hz, phase-corrected, 8th order Butterworth filter. To avoid values greater than 1.0, the linear envelope was normalized to either the peak recorded during an MVIC collected before performing the reflex testing, or the peak muscle activity recorded during the gait analysis (Rudolph et al., 2001).
Table 1 Subject demographics. Subject
Sex
Age
Paretic side
Time since stroke (months)
Treadmill walking speed (m/s)
AFO
Assistive device
1 2 3 4 5 6 7 8 9 10
M F M F M M F M F M
45 61 72 66 51 57 62 68 56 63
L L R R L R R L L R
55 21 22 31 89 73 14 195 356 8
0.6 0.5 0.3 0.45 0.7 0.8 0.6 0.4 0.8 0.65
n n y n y n n n n n
y y y y y n y n n y
174
M. Murray et al. / Journal of Electromyography and Kinesiology 24 (2014) 172–177
Fig. 1. The custom designed ankle loader applied a dorsiflexion torque about the ankle to load the plantarflexors. The Cybex motor was used to rotate the knee from 30° to 90° of flexion at 180 and 300°/s.
The normalized EMG for each muscle was combined into an ensemble average to allow for comparisons of individual muscle responses between conditions (load level and knee flexion velocity). 2.2. Instrumented gait analysis Following isolated reflex testing, subjects walked on an instrumented dual-belt treadmill (Bertec Corp, Columbus, OH) with varying amounts of BWS. BWS was provided by a custom designed harness support system (Lewek, 2011), which used elastic elements to provide a nearly constant unweighting force. Unloading of 0%, 20%, and 40%BWS was applied in random order. Treadmill walking speed was based on the subject’s self-selected comfortable overground walking speed, from an average of three passes over a GaitRite mat (CIR Systems; Haverford, PA). Once the target treadmill speed was reached, subjects walked for approximately 1– 2 min at each level of weight support, with data collected for the final 30s of each condition. A short rest break (approx. 1 min) was provided between conditions. Limb motions were recorded with an 8-camera motion capture system (MX-40+; Vicon/Peak; Centenial, CO) recording at 120 Hz. Reflective markers were placed on the pelvis, thighs, and shanks in clusters mounted to rigid thermoplastic shells. Additional markers were placed over the 2nd metatarsal head, at the 5th metatarsal head, and on the heel counter to track the motion of the feet. Finally, markers were placed at the bilateral iliac crests, greater trochanters, femoral condyles, and malleoli to determine the ends of limb segments. Ground reaction forces and EMG muscle activity (from the RF, VL, VM and MH) were recorded simultaneously with marker trajectories at 960 Hz. Marker trajectories were low-pass filtered at 6 Hz to calculate the relative three-dimensional positions and intersegmental joint angles using a rigid body analysis and normalized to a stride cycle (Visual 3D, C-Motion Corp, Bethesda, MD). EMG data during gait trials were filtered as described in the reflex testing and an ensemble average was computed. 2.3. Data management and analysis Outcome measures from isolated stretch reflex testing included an initial peak response as well as a prolonged response (Lewek
et al., 2007). Initially, we determined baseline activity, which consisted of the average muscle activity during the 250 ms immediately preceding knee movement. Peak responses were then calculated as the largest increase in muscle activity during the 150 ms following the initiation of knee movement. Next, a prolonged response was determined as the difference between the average EMG activity after knee movement (i.e., 250 ms immediately following termination of knee movement) and the baseline activity. All statistical analyses were performed using SPSS (ver 16, Chicago, IL). Repeated-measures ANOVAs (repeated for load and speed) were used to compare the group means and standard deviations of the VL, RF, and VM muscles for both the peak and prolonged muscle responses. During gait, we determined the peak knee flexion of the paretic and non-paretic limbs during swing, as well as the integral of muscle activity from peak propulsion (during late stance) to peak knee flexion (during swing) for the VL, RF, and VM. Each of these variables was compared between BWS condition using a repeated measures ANOVA. Pearson correlations were used to relate muscle activity integrals during gait to peak and prolonged muscle activity during isolated reflex testing. A significance level of p < 0.05 was used for all statistical analyses. 3. Results 3.1. Reflex responses In response to the rapid knee flexion motion, we observed the VL, RF, and VM to increase activation, although there was little consistent effect of speed or ankle load. Although peak VM activity demonstrated significantly greater muscle activity at 300°/s compared to 180°/s (p = 0.013), no significant effects due to speed were observed for VL (p = 0.107) or RF (p = 0.228) (see Fig. 2). Peak quadriceps muscle responses were not influenced by ankle load (VL, p = 0.082; RF, p = 0.602; VM, p = 0.145) and no interactions between speed and ankle load were observed for the peak quadriceps muscle responses (VL, p = 0.414; RF, p = 0.612; VM, p = 0.131). The prolonged RF activity was greater at 300°/s compared to 180°/s (p = 0.050), however, no differences between speeds were observed for the VL (p = 0.136) or VM (p = 0.940) (see Fig. 3). The load applied to the ankle plantarflexors significantly influenced the prolonged VM response only (p = 0.038). Specifically, 0.2 N m/kg yielded a significantly reduced prolonged quadriceps muscle response compared to the 0 N m/kg (p = 0.043) and 0.4 N m/kg (p = 0.023). The prolonged quadriceps muscle response for the VL (p = 0.089) and the RF (p = 0.524) was not influenced by load. Finally, no interaction between speed and ankle load was observed for any of the prolonged quadriceps muscle responses (VL, p = 0.414; RF, p = 0.263; VM, p = 0.382).
Fig. 2. The initial peak response of the quadriceps in response to the rapid knee flexion stretch. Values are provided for the VL, RF, and VM muscles with different amounts of ankle loading (0, 0.2, and 0.4 N m/kg). A main effect for speed was observed for the VM muscle only (p = 0.013).
M. Murray et al. / Journal of Electromyography and Kinesiology 24 (2014) 172–177
175
anticipated that walking with 20% or 40%BWS would decrease quadriceps activity during late stance/early swing and consequently increase swing phase knee flexion. Our data instead suggested that the use of BWS did not influence knee kinematics or quadriceps activity during this phase of the gait cycle.
4.1. Effect of load on extensor excitability
Fig. 3. The prolonged response of the quadriceps in response to the rapid knee flexion stretch. Values are provided for the VL, RF, and VM muscles with different amounts of ankle loading (0, 0.2, and 0.4 N m/kg). A main effect for speed was observed for the RF muscle (p = 0.050) and a main effect for load was observed for VM (p = 0.038).
3.2. Gait kinematics and quadriceps muscle activity Our evaluation of gait kinematics and muscle activity revealed that subjects exhibited a range of paretic limb peak knee flexion angles during swing phase (range: 13°–49° of knee flexion; see Table 2). The paretic knee exhibited less knee flexion during swing phase than the non-paretic knee (p = 0.001), but there was no main effect of load on peak knee flexion (p = 0.175). Likewise, no interaction effect was observed between load and side (p = 0.972). We experienced difficulty in obtaining clear EMG signals during gait from some subjects, such that we analyzed VL for 7 subjects, RF for 8 subjects, and VM for all 10 subjects. The amount of BWS did not significantly alter quadriceps muscle activity during the late stance/early swing phase of gait (VL: p = 0.177; RF: p = 0.615; VM: p = 0.167). 3.3. Correlation of reflex measures and gait kinematics We did not observe substantial correlations between the muscle activity during gait and the muscle responses during isolated reflex testing (see Table 3). The only pair to show a significant correlation was the VL activity during the late stance/early swing phase of gait with the initial peak responses at 180°/s (r = 0.560, p = 0.008 (see Fig. 4)). The VM and RF integrals during gait were not significantly related to the initial or prolonged response at either 180°/s or 300°/s during isolated reflex testing. 4. Discussion The hypothesis that force feedback, as sensed by the plantarflexors, would modulate knee extensor excitability for individuals with chronic stroke was partially accepted. In particular, we expected that the application of ankle load would decrease knee extensor excitability during seated quadriceps reflex testing. However, we observed a reduction in the prolonged quadriceps response to 0.2 N m/kg load for the VM only. We further
Table 2 Kinematic and electromyographic variables during gait. 0%BWS
20%BWS
40%BWS
P-value
Kinematics Paretic PKF (°) Non-paretic PKF (°)
37.7 (9.9) 61.6 (8.3)
37.4 (9.7) 61.0 (7.9)
36.8 (10.2) 60.6 (8.0)
p = 0.516 p = 0.391
Electromyography VL Integral RF Integral VM Integral
16.4 (9.5) 17.9 (7.4) 14.9 (8.5)
23.0 (11.6) 16.7 (7.5) 15.6 (8.6)
19.4 (11.3) 17.6 (8.5) 17.4 (9.5)
p = 0.177 p = 0.615 p = 0.167
All values represent mean (SD).
In a cat model, load applied to the ankle plantarflexors creates autogenic excitation, but has no, or an inhibitory, effect on the quadriceps (heterogenic inhibition) during locomotion (Ross and Nichols, 2009; Wilmink and Nichols, 2003). We have extended this finding to adults with post-stroke hemiparesis, who demonstrated either no change or a reduction in quadriceps reflex excitability with the application of isolated ankle load. Although heterogenic force feedback seemed to have a limited effect on quadriceps excitability in our subjects post-stroke, force feedback is only one factor capable of modulating muscle activity and timing. Other factors, which were not measured, include exteroceptive input (Schmit et al., 2003), hip flexor length (Yang et al., 2004), and feed-forward inputs (Duysens et al., 2000). We also did not collect EMG data from the gastrocnemius or soleus muscles during testing, so we are unable to determine the autogenic response to the applied ankle load.
4.2. Effect of load on gait In our subjects post-stroke, we observed that quadriceps activity during late stance/early swing phase of gait was not influenced by altering limb load through BWS, at least up to 40%BWS. Others have previously examined the effect of BWS on quadriceps activity during early stance phase (Finch et al., 1991; Hesse et al., 1999). Because of our focus on stiff knee gait, we focused on the late stance/early swing phase of gait, normally considered a time that the uniarticular quadriceps muscles are inactive. Muscle activity during late stance is particularly important, however, because it can directly influence the stiff knee gait pattern (Goldberg et al., 2004). Furthermore, plantarflexor loads are largest during late stance (Lewek, 2011) making this an important period to examine the influence of heterogenic force feedback to the quadriceps. The timing of load application may also be an important factor in modulating muscle activity. Changes in VL activation have been shown with transient increases/decreases in load, but only when the load is applied during the early portion of stance (Stephens and Yang, 1999). Clinically, most BWS systems do not modulate load throughout the gait cycle. In such a case, where unloading is relatively constant, it appears that only lower leg muscle EMG of unimpaired individuals is altered, while upper leg muscles remained unchanged (Bachmann et al., 2008). The lack of change to quadriceps activity, by us and others (Bachmann et al., 2008; Colby et al., 1999; Finch et al., 1991), is consistent with our observed lack of change in knee flexion angles with varying amounts of BWS. Although our subjects post-stroke did not alter knee flexion angles with different levels of BWS (see also (Chen et al., 2005)), unimpaired subjects have exhibited decreased peak knee flexion angles with increased BWS (Finch et al., 1991; Threlkeld et al., 2003). While those levels of unloading (Finch et al., 1991; Threlkeld et al., 2003) surpassed the typical clinically applied limits of 40%BWS (Hesse et al., 1995; van Hedel et al., 2006), there is the possibility of a differential kinematic response to BWS when used with unimpaired subjects and those with neurological injury. Nevertheless, the literature supports our finding that thigh musculature is relatively unaffected by ankle modulated force feedback during gait.
176
M. Murray et al. / Journal of Electromyography and Kinesiology 24 (2014) 172–177
Table 3 Correlation coefficients between isolated reflex testing and gait. Initial response
VL (gait)
Prolonged response
180°/s
300°/s
180°/s
300°/s
r = 0.56; p = 0.008
r = 0.42; p = 0.06
r=
r = 0.10; p = 0.68
RF (gait)
r=
VM (gait)
r = 0.33; p = 0.08
0.26; p = 0.23
r=
0.24; p = 0.26
r = 0.34; p = 0.07
Fig. 4. Relationship between VL initial activity at 180°/s during isolated reflex testing and the VL activity during the late stance/early swing phase of gait (r = 0.56; p = 0.008). High load corresponds with 0%BWS and 0.4 N m/kg, moderate load corresponds with 20%BWS and 0.2 N m/kg and low load corresponds with 40%BWS and 0 N m/kg. Note that these different load conditions are scattered about the plot suggesting that the muscle activity during gait is not mediated by force feedback from the ankle.
4.3. Force feedback from ankle extensors Other studies have examined the effects of whole body (un)loading on ankle or hip kinematics and muscle activity (Gordon et al., 2009; Nakazawa et al., 2004; Stephens and Yang, 1999), yet few have studied the effects that occur at the knee (Threlkeld et al., 2003). In contrast to whole body (un)loading, our isolated reflex testing was designed to load the ankle extensors only. In this way, we attempted to determine what effect, if any, ankle loading had on quadriceps muscle activity. During gait, however, the extensor muscles throughout the limb, and not just the ankle plantarflexors, are loaded during stance. There are, therefore, numerous competing sensory signals that result in the net output to the muscle. In fact, it is possible that positive force feedback amongst the quadriceps (Wilmink and Nichols, 2003) is competing with inhibitory force feedback from the plantarflexors yielding no real change in quadriceps activity when BWS is altered. In addition to force feedback, the quadriceps are receiving input from monosynaptic and polysynaptic length-sensitive afferents (group Ia and II). During the seated isolated reflex testing, the knee velocity was constant throughout the motion, with a fixed ankle load. This differs substantially from the gait testing. As the knee flexes and extends throughout the gait cycle, such inputs are constantly altered making the relationship between isolated reflex data and functional gait performance difficult to interpret (Lewek et al., 2007). Given the inherent limitations in relating isolated experimental conditions to gait characteristics, it is not surprising that few correlations were noted between conditions. Nevertheless, a weak correlation was noted across subjects between VL activity during gait testing and seated quadriceps reflex testing. Specifically, data pairs (e.g., 0%BWS with 0.4 N m/kg; 20%BWS with 0.2 N m/kg) suggest that VL activity seen during gait may have a reflexive component, but is clearly not modulated by limb load. The observed
0.12; p = 0.62
r = 0.00; p = 0.99
r = 0.00; p = 0.99
r = 0.29; p = 0.131
r = 0.30; p = 0.12
relationships could also be a result of individual differences in spasticity (e.g., some subjects may have exhibited greater muscle activity than others in both reflex and gait testing, regardless of load level) and/or MVIC measures, which are difficult to ascertain in individuals post-stroke (Hsu et al., 2006). Some limitations of this research exist, and should be noted. We intentionally recruited individuals with a range of walking patterns (i.e., not all had stiff-knee gait pattern), which may have confounded the results. This decision was made because we expected that a heterogeneous group would allow us to better see how the reflex mediated quadriceps activity influenced gait parameters. In addition, some subjects (N = 3) had to wear their ankle foot orthosis (AFO) during the gait testing only (AFOs were removed for isolated reflex testing). We suspect that the use of an AFO has the potential to limit the force feedback sensed at the ankle. Additionally, subjects in our study only walked at each load level for a few minutes and it is possible that training with BWS for an extended period of time may affect knee flexion, but further research would be needed to test this assertion. Finally, several subjects required use of the handrails during the gait testing. Shifting a portion of the weight from the lower extremities to the upper extremity may have limited the load sensed at the ankle during gait. The information from this study suggests that the use of BWS during gait retraining will not result in immediate kinematic changes at the knee in those with stiff-knee gait following stroke. Clinically, this is important, as BWS may not be an appropriate modality if attempting to reduce stiff-knee gait during training. Nevertheless, the initially observed performance does not indicate what the final outcome will be (Winstein and Schmidt, 1990). While interventions that focus on other afferent information may prove effective, adjusting loading will likely not result in immediate improvements in knee flexion or reduced knee extensor muscle activity during late stance/early swing. Conflict of interest This work was supported in part by grants from NSF (Graduate Student Fellowship to AH) and UNC’s University Research Council (to MDL). Role of the funding source The study sponsors had no involvement in the study design, in the collection, analysis and interpretation of data; in the writing of the manuscript; and in the decision to submit the manuscript for publication References Ada L, Dean CM, Vargas J, Ennis S. Mechanically assisted walking with body weight support results in more independent walking than assisted overground walking in non-ambulatory patients early after stroke: a systematic review. J Physiother 2010;56:153–61. Bachmann V, Muller R, van Hedel HJ, Dietz V. Vertical perturbations of human gait: organisation and adaptation of leg muscle responses. Exp Brain Res 2008;186:123–30.
M. Murray et al. / Journal of Electromyography and Kinesiology 24 (2014) 172–177 Barbeau H, Visintin M. Optimal outcomes obtained with body-weight support combined with treadmill training in stroke subjects. Arch Phys Med Rehabil 2003;84:1458–65. Chen G, Patten C, Kothari DH, Zajac FE. Gait deviations associated with post-stroke hemiparesis: improvement during treadmill walking using weight support, speed, support stiffness, and handrail hold. Gait Posture 2005;22:57–62. Colby SM, Kirkendall DT, Bruzga RF. Electromyographic analysis and energy expenditure of harness supported treadmill walking: implications for knee rehabilitation. Gait Posture 1999;10:200–5. Conway BA, Hultborn H, Kiehn O. Proprioceptive input resets central locomotor rhythm in the spinal cat. Exp Brain Res 1987;68:643–56. Crone C, Hultborn H, Kiehn O, Mazieres L, Wigstrom H. Maintained changes in motoneuronal excitability by short-lasting synaptic inputs in the decerebrate cat. J Physiol 1988;405:321–43. Dietz V, Duysens J. Significance of load receptor input during locomotion: a review. Gait Posture 2000;11:102–10. Dietz V, Muller R, Colombo G. Locomotor activity in spinal man: significance of afferent input from joint and load receptors. Brain 2002;125:2626–34. Duncan PW, Sullivan KJ, Behrman AL, Azen SP, Wu SS, Nadeau SE, et al. Bodyweight-supported treadmill rehabilitation after stroke. New England J Med 2011;364:2026–36. Duysens J, Pearson KG. Inhibition of flexor burst generation by loading ankle extensor muscles in walking cats. Brain Res 1980;187:321–32. Duysens J, Clarac F, Cruse H. Load-regulating mechanisms in gait and posture: comparative aspects. Physiol Rev 2000;80:83–133. Finch L, Barbeau H, Arsenault B. Influence of body weight support on normal human gait: development of a gait retraining strategy. Physical therapy. 1991;71:842– 55 [discussion 55–6]. Goldberg SR, Anderson FC, Pandy MG, Delp SL. Muscles that influence knee flexion velocity in double support: implications for stiff-knee gait. J Biomech 2004;37:1189–96. Gordon KE, Wu M, Kahn JH, Dhaher YY, Schmit BD. Ankle load modulates hip kinetics and EMG during human locomotion. J Neurophysiol 2009;101:2062–76. Granata KP, Abel MF, Damiano DL. Joint angular velocity in spastic gait and the influence of muscle–tendon lengthening. J Bone Joint Surg Am 2000;82:174–86. Hesse S, Bertelt C, Jahnke MT, Schaffrin A, Baake P, Malezic M, et al. Treadmill training with partial body weight support compared with physiotherapy in nonambulatory hemiparetic patients. Stroke J Cereb Circulat 1995;26:976–81. Hesse S, Konrad M, Uhlenbrock D. Treadmill walking with partial body weight support versus floor walking in hemiparetic subjects. Arch Phys Med Rehabil 1999;80:421–7. Hiebert GW, Pearson KG. Contribution of sensory feedback to the generation of extensor activity during walking in the decerebrate Cat. J Neurophysiol 1999;81:758–70. Hsu WL, Krishnamoorthy V, Scholz JP. An alternative test of electromyographic normalization in patients. Muscle Nerve 2006;33:232–41. Kautz SA, Brown DA. Relationships between timing of muscle excitation and impaired motor performance during cyclical lower extremity movement in post-stroke hemiplegia. Brain 1998;121:515–26. Kerrigan DC, Gronley J, Perry J. Stiff-legged gait in spastic paresis. A study of quadriceps and hamstrings muscle activity. Am J Phys Med Rehabil 1991;70:294–300. Lewek MD. The influence of body weight support on ankle mechanics during treadmill walking. J Biomech 2011;44:128–33. Lewek MD, Hornby TG, Dhaher YY, Schmit BD. Prolonged quadriceps activity following imposed hip extension: a neurophysiological mechanism for stiffknee gait? J Neurophysiol 2007;98:3153–62. Mulroy S, Gronley J, Weiss W, Newsam C, Perry J. Use of cluster analysis for gait pattern classification of patients in the early and late recovery phases following stroke. Gait Posture 2003;18:114–25. Nakazawa K, Miyoshi T, Sekiguchi H, Nozaki D, Akai M, Yano H. Effects of loading and unloading of lower limb joints on the soleus H-reflex in standing humans. Clin Neurophysiol 2004;115:1296–304. Ross KT, Nichols TR. Heterogenic feedback between hindlimb extensors in the spontaneously locomoting premammillary cat. J Neurophysiol 2009;101:184–97. Rudolph KS, Axe MJ, Buchanan TS, Scholz JP, Snyder-Mackler L. Dynamic stability in the anterior cruciate ligament deficient knee. Knee Surg, Sports Traumatol, Arthrosc: Official J ESSKA 2001;9:62–71. Schindler-Ivens S, Desimone D, Grubich S, Kelley C, Sanghvi N, Brown DA. Lower extremity passive range of motion in community-ambulating stroke survivors. J Neurol Phys Ther 2008;32:21–31. Schmit BD, Hornby TG, Tysseling-Mattiace VM, Benz EN. Absence of local sign withdrawal in chronic human spinal cord injury. J Neurophysiol 2003;90:3232–41. Stephens MJ, Yang JF. Loading during the stance phase of walking in humans increases the extensor EMG amplitude but does not change the duration of the step cycle. Exp Brain Res 1999;124:363–70. Threlkeld AJ, Cooper LD, Monger BP, Craven AN, Haupt HG. Temporospatial and kinematic gait alterations during treadmill walking with body weight suspension. Gait Posture 2003;17:235–45. van Hedel HJ, Tomatis L, Muller R. Modulation of leg muscle activity and gait kinematics by walking speed and bodyweight unloading. Gait Posture 2006;24:35–45.
177
Visintin M, Barbeau H, Korner-Bitensky N, Mayo NE. A new approach to retrain gait in stroke patients through body weight support and treadmill stimulation. Stroke J Cereb Circulat 1998;29:1122–8. Waters RL, Garland DE, Perry J, Habig T, Slabaugh P. Stiff-legged gait in hemiplegia: surgical correction. J Bone Joint Surg Am 1979;61:927–33. Wilmink RJ, Nichols TR. Distribution of heterogenic reflexes among the quadriceps and triceps surae muscles of the cat hind limb. J Neurophysiol 2003;90:2310–24. Winstein CJ, Schmidt RA. Reduced frequency of knowledge of results enhances motor skill learning. J Exp Psychol Learn 1990;16:677–91. Wu M, Schmit BD. Spastic reflexes triggered by ankle load release in human spinal cord injury. J Neurophysiol 2006;96:2941–50. Yang JF, Lam T, Pang MY, Lamont E, Musselman K, Seinen E. Infant stepping: a window to the behaviour of the human pattern generator for walking. Can J Physiol Pharmacol 2004;82:662–74.
Mary Murray received her Doctorate in Physical Therapy from the University of North Carolina at Chapel Hill and her Bachelor’s degree in Communications from the University of Akron. She currently works with adult neurorehabilitation services at the Cleveland Clinic.
Amy Hardee received her master’s degree in biomedical engineering from the joint program at UNC-Chapel Hill and NC State University. She previously received her bachelor’s degree in biomedical engineering from NC State University. Currently, she works as a Development and Sustaining Engineer for Teleflex Medical.
Richard L. Goldberg received his doctorate in biomedical engineering from Duke University. He is currently the director of undergraduate studies in the Department of Biomedical Engineering at The University of North Carolina at Chapel Hill. He has expertise in rehabilitation engineering, device design, and assistive technology for individuals with disabilities.
Michael D. Lewek received his doctorate in biomechanics and movement science from the University of Delaware and performed a post-doctoral fellowship in neuromechanics at the Rehabilitation Institute of Chicago. He is currently an associate professor in the Division of Physical Therapy at The University of North Carolina at Chapel Hill. His research involves determination of the biomechanical, neurophysiologic, and motor learning aspects of walking recovery following stroke.
Contents lists available at ScienceDirect
Journal of Electromyography and Kinesiology journal homepage: www.elsevier.com/locate/jelekin
Loading and knee flexion after stroke: Less does not equal more Mary Murray a,b, Amy Hardee c,d, Richard L. Goldberg c, Michael D. Lewek a,e,⇑ a
Division of Physical Therapy, Department of Allied Health Sciences, University of North Carolina at Chapel Hill, USA Cleveland Clinic, Cleveland, OH, USA c Joint Department of Biomedical Engineering, University of North Carolina at Chapel Hill and North Carolina State University, USA d Teleflex Medical, Durham, NC, USA e Human Movement Science Program, University of North Carolina at Chapel Hill, USA b
a r t i c l e
i n f o
Article history: Received 19 June 2013 Received in revised form 22 August 2013 Accepted 10 October 2013
Keywords: Stroke Gait Force feedback Stiff-knee gait Rehabilitation
a b s t r a c t It is believed that force feedback can modulate lower extremity extensor activity during gait. The purpose of this research was to determine the role of limb loading on knee extensor excitability during the late stance/early swing phase of gait in persons post-stroke. Ten subjects with chronic hemiparesis poststroke participated in (1) seated isolated quadriceps reflex testing with ankle loads of 0–0.4N m/kg and (2) gait analysis on a treadmill with 0%, 20% or 40% body weight support. Muscle reflex responses were recorded from vastus lateralis (VL), rectus femoris (RF), and vastus medialis (VM) during seated testing. Knee kinematics and quadriceps activity during late stance/early swing phase of gait were compared across loading conditions. Although isolated loading of the ankle plantarflexors at 0.2 N m/kg reduced VM prolonged response (p = 0.04), loading did not alter any other measure of quadriceps excitability (all p > 0.08). During gait, the use of BWS did not influence knee kinematics (p = 0.18) or muscle activity (all p > 0.17) during late stance/early swing phase. This information suggests that load sensed at the ankle has minimal effect on the ipsilateral quadriceps of individuals post-stroke during late stance. It appears that adjusting limb loading during rehabilitation may not be an effective tool to address stiffknee gait following stroke. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction Following stroke, gait is often marked by decreased knee flexion of the paretic limb during swing phase; a pattern called ‘‘stiff-knee gait’’ (Kerrigan et al., 1991). While the exact cause of stiff-knee gait is unknown, abnormal muscle timing of the quadriceps has been implicated (Kerrigan et al., 1991; Lewek et al., 2007; Waters et al., 1979). Normally, quadriceps activation occurs during the early to mid-stance period of gait; however, individuals poststroke often demonstrate prolonged quadriceps activity (Kautz and Brown, 1998). The presence of prolonged quadriceps activity into late stance has been associated with stiff knee gait in children with cerebral palsy (Goldberg et al., 2004). Muscle activity during gait is modifiable by sensory receptors throughout the lower extremity. In particular, force (i.e., load) feedback to the central nervous system (CNS) from Golgi tendon organs (group Ib) serves a critical role in regulating extensor activ-
⇑ Corresponding author. Address: University of North Carolina at Chapel Hill, 3043 Bondurant Hall, CB#7135, Chapel Hill, NC 27599-7135, USA. Tel.: +1 919 966 9732. E-mail address: [email protected] (M.D. Lewek). 1050-6411/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jelekin.2013.10.006
ity duration during stance (Dietz et al., 2002; Hiebert and Pearson, 1999). Prolonged mechanical loading to the ankle extensors, for instance, elicits prolonged muscle activity in the homonymous muscle of decerebrate walking cats (Crone et al., 1988; Hiebert and Pearson, 1999). Humans also appear to exhibit similar positive autogenic force feedback (Dietz et al., 2002; Gordon et al., 2009; Yang et al., 2004), suggesting that the CNS is particularly sensitive to plantarflexor force feedback, and neuromotor output can be directly influenced by these signals (Ada et al., 2010). Although force feedback appears to excite homonymous muscles during gait (Conway et al., 1987; Duysens and Pearson, 1980), there is evidence for heterogenic inhibitory-force feedback from the plantarflexors to the quadriceps (Ross and Nichols, 2009; Wilmink and Nichols, 2003). This would suggest that increased limb loading (particularly into late stance) would increase the Ib sensory signals from the plantarflexors, to inhibit quadriceps activity during late stance to allow for improved knee flexion. During clinical practice, however, walking on a treadmill with body weight support (BWS) has been used extensively following stroke to facilitate walking recovery (Ada et al., 2010; Barbeau and Visintin, 2003; Visintin et al., 1998). While not shown to be superior to other forms of therapy (Duncan et al., 2011), gait training with BWS has demonstrated improvements in gait speed, endurance,
173
M. Murray et al. / Journal of Electromyography and Kinesiology 24 (2014) 172–177
and balance (Barbeau and Visintin, 2003; Visintin et al., 1998). Evidence from unimpaired adults (Colby et al., 1999; Dietz et al., 2002; Finch et al., 1991) and individuals post-stoke (Hesse et al., 1999) suggests that the use of BWS will decrease quadriceps activity during early stance. The effect on late stance, however, when plantarflexor load is greatest, remains unknown. In addition, the use of BWS appears to decrease swing phase knee flexion of unimpaired individuals (Finch et al., 1991; van Hedel et al., 2006), representing a potentially unwanted side effect of BWS for individuals post-stroke at a time when knee flexion may already be reduced. The purpose of this study was to characterize the heterogenic force-feedback response from the plantarflexors to the quadriceps in individuals with chronic (>6 months) stroke. We hypothesized that a simulated ankle load during seated isolated joint reflex testing would decrease knee extensor excitability. We then wanted to determine the presence of a relationship between ankle-load mediated reflex control of the quadriceps and the generation of abnormal quadriceps activity during walking in individuals post-stroke. Based on animal models that manipulated limb load during gait (Hiebert and Pearson, 1999), we hypothesized that walking with BWS would reduce any abnormal late stance/early swing phase quadriceps activity and consequently improve swing phase knee flexion for individuals with chronic (>6 months) stroke. Although all joints in the lower extremity are loaded and unloaded during gait, the ankle extensors are believed to contain the critical Ib afferents to modify gait (Dietz and Duysens, 2000), and thus we expected to observe a direct relationship between plantarflexorapplied, limb load mediated reflex excitability of the quadriceps and the late stance/early swing phase quadriceps activity during gait in individuals post-stroke.
2. Methods Ten subjects with chronic hemiparesis (>6 months) resulting from a hemorrhagic or ischemic stroke were recruited for testing (see Table 1). Subjects exhibited a variety of walking patterns (Mulroy et al., 2003) but were able to step independently on a treadmill. Subjects were excluded from the study if they responded affirmatively to questions about confounding medical issues (e.g., cardiac arrhythmia, uncontrolled seizures or hypertension) that would preclude the ability to ambulate safely on a treadmill, bilateral stroke involvement, vestibular or cerebellar involvement, or other neurologic and musculoskeletal disorders affecting the legs. Subjective clinical judgment was used to determine the presence of other cognitive or communication problems that inhibited understanding of study involvement. Prior to participation, all subjects signed an informed consent form approved by the University of North Carolina at Chapel Hill Institutional Review Board. All subjects performed the two parts (isolated joint reflex testing, followed by an instrumented gait analysis) of the study during a single session.
2.1. Reflex testing and analysis Isolated reflex testing was performed to determine the unique contribution of limb load (as it is sensed by the ankle plantarflexors) on knee extensor reflex excitability in a controlled setting (Wu and Schmit, 2006). Prior to testing, active surface electrodes (Motion Lab Systems, Baton Rouge, LA) were applied to the rectus femoris (RF), vastus lateralis (VL), vastus medialis (VM), and medial hamstrings (MH) to monitor reflex response. Subjects sat on an isokinetic dynamometer (Humac, CSMi, Stoughton, MA) with the back supported and the hips flexed to 80°. The paretic knee’s flexion/extension axis was aligned with the Cybex motor axis and the paretic foot was secured to the footplate of a custom designed ankle loader (see Fig. 1). The ankle loader simulated weight bearing during stance through the production of a dorsiflexion load (Wu and Schmit, 2006). An air powered linear actuator generated and transmitted the load to the footplate. A load cell (MLP-100, Transducer Techniques, Temecula, CA), in series with the actuator, monitored the dorsiflexion load applied to the subject. Dorsiflexion loads were normalized to the subject’s body mass, and were set at 0, 0.2, and 0.3–0.4 N m/kg. Some subjects had difficulty tolerating 0.4 N m/kg, so a lower load was used (i.e., 0.3 N m/kg) such that all subjects denied any discomfort during testing. These loads represent a portion of the ankle moments experienced during walking (Lewek, 2011), but are comparable to the passive dorsiflexion moment required to achieve maximum dorsiflexion (Schindler-Ivens et al., 2008). During each loading condition, ramp-hold stretches at the knee joint were performed by passively flexing the knee from 30° to 90° of flexion at 180 and 300°/s. These knee angles are slightly greater than what is observed during ‘normal’ gait, however, we chose these angles to ensure an adequate stretch of the quadriceps, and to avoid the possibility of hamstring tightness in a more extended position. The speeds were chosen to simulate joint speeds similar to those experienced during gait (Granata et al., 2000), as higher velocity stretches produce higher tonic stretch reflexes (Crone et al., 1988). Therefore, there were a total of 6 randomized conditions (three loads and two speeds). Prior to each trial, subjects pre-activated the hip flexors (approx 10%MVIC) (Lewek et al., 2007) to facilitate the production of RF responses and simulate the hip flexion force that occurs during the stance – swing transition of gait. Subjects were asked to maintain the pre-activated level of effort through the duration of the trial. During each trial, the knee’s angular position, velocity, and torque were recorded from the Cybex. EMG signals were bandpass filtered between 20 and 500 Hz and sampled at 1000 Hz to a personal computer. Analysis of EMG signals was performed with custom software (Labview 2009, National Instruments, Austin, TX). A linear envelope was created by full-wave rectifying the data, followed by low-pass filtering with a 20 Hz, phase-corrected, 8th order Butterworth filter. To avoid values greater than 1.0, the linear envelope was normalized to either the peak recorded during an MVIC collected before performing the reflex testing, or the peak muscle activity recorded during the gait analysis (Rudolph et al., 2001).
Table 1 Subject demographics. Subject
Sex
Age
Paretic side
Time since stroke (months)
Treadmill walking speed (m/s)
AFO
Assistive device
1 2 3 4 5 6 7 8 9 10
M F M F M M F M F M
45 61 72 66 51 57 62 68 56 63
L L R R L R R L L R
55 21 22 31 89 73 14 195 356 8
0.6 0.5 0.3 0.45 0.7 0.8 0.6 0.4 0.8 0.65
n n y n y n n n n n
y y y y y n y n n y
174
M. Murray et al. / Journal of Electromyography and Kinesiology 24 (2014) 172–177
Fig. 1. The custom designed ankle loader applied a dorsiflexion torque about the ankle to load the plantarflexors. The Cybex motor was used to rotate the knee from 30° to 90° of flexion at 180 and 300°/s.
The normalized EMG for each muscle was combined into an ensemble average to allow for comparisons of individual muscle responses between conditions (load level and knee flexion velocity). 2.2. Instrumented gait analysis Following isolated reflex testing, subjects walked on an instrumented dual-belt treadmill (Bertec Corp, Columbus, OH) with varying amounts of BWS. BWS was provided by a custom designed harness support system (Lewek, 2011), which used elastic elements to provide a nearly constant unweighting force. Unloading of 0%, 20%, and 40%BWS was applied in random order. Treadmill walking speed was based on the subject’s self-selected comfortable overground walking speed, from an average of three passes over a GaitRite mat (CIR Systems; Haverford, PA). Once the target treadmill speed was reached, subjects walked for approximately 1– 2 min at each level of weight support, with data collected for the final 30s of each condition. A short rest break (approx. 1 min) was provided between conditions. Limb motions were recorded with an 8-camera motion capture system (MX-40+; Vicon/Peak; Centenial, CO) recording at 120 Hz. Reflective markers were placed on the pelvis, thighs, and shanks in clusters mounted to rigid thermoplastic shells. Additional markers were placed over the 2nd metatarsal head, at the 5th metatarsal head, and on the heel counter to track the motion of the feet. Finally, markers were placed at the bilateral iliac crests, greater trochanters, femoral condyles, and malleoli to determine the ends of limb segments. Ground reaction forces and EMG muscle activity (from the RF, VL, VM and MH) were recorded simultaneously with marker trajectories at 960 Hz. Marker trajectories were low-pass filtered at 6 Hz to calculate the relative three-dimensional positions and intersegmental joint angles using a rigid body analysis and normalized to a stride cycle (Visual 3D, C-Motion Corp, Bethesda, MD). EMG data during gait trials were filtered as described in the reflex testing and an ensemble average was computed. 2.3. Data management and analysis Outcome measures from isolated stretch reflex testing included an initial peak response as well as a prolonged response (Lewek
et al., 2007). Initially, we determined baseline activity, which consisted of the average muscle activity during the 250 ms immediately preceding knee movement. Peak responses were then calculated as the largest increase in muscle activity during the 150 ms following the initiation of knee movement. Next, a prolonged response was determined as the difference between the average EMG activity after knee movement (i.e., 250 ms immediately following termination of knee movement) and the baseline activity. All statistical analyses were performed using SPSS (ver 16, Chicago, IL). Repeated-measures ANOVAs (repeated for load and speed) were used to compare the group means and standard deviations of the VL, RF, and VM muscles for both the peak and prolonged muscle responses. During gait, we determined the peak knee flexion of the paretic and non-paretic limbs during swing, as well as the integral of muscle activity from peak propulsion (during late stance) to peak knee flexion (during swing) for the VL, RF, and VM. Each of these variables was compared between BWS condition using a repeated measures ANOVA. Pearson correlations were used to relate muscle activity integrals during gait to peak and prolonged muscle activity during isolated reflex testing. A significance level of p < 0.05 was used for all statistical analyses. 3. Results 3.1. Reflex responses In response to the rapid knee flexion motion, we observed the VL, RF, and VM to increase activation, although there was little consistent effect of speed or ankle load. Although peak VM activity demonstrated significantly greater muscle activity at 300°/s compared to 180°/s (p = 0.013), no significant effects due to speed were observed for VL (p = 0.107) or RF (p = 0.228) (see Fig. 2). Peak quadriceps muscle responses were not influenced by ankle load (VL, p = 0.082; RF, p = 0.602; VM, p = 0.145) and no interactions between speed and ankle load were observed for the peak quadriceps muscle responses (VL, p = 0.414; RF, p = 0.612; VM, p = 0.131). The prolonged RF activity was greater at 300°/s compared to 180°/s (p = 0.050), however, no differences between speeds were observed for the VL (p = 0.136) or VM (p = 0.940) (see Fig. 3). The load applied to the ankle plantarflexors significantly influenced the prolonged VM response only (p = 0.038). Specifically, 0.2 N m/kg yielded a significantly reduced prolonged quadriceps muscle response compared to the 0 N m/kg (p = 0.043) and 0.4 N m/kg (p = 0.023). The prolonged quadriceps muscle response for the VL (p = 0.089) and the RF (p = 0.524) was not influenced by load. Finally, no interaction between speed and ankle load was observed for any of the prolonged quadriceps muscle responses (VL, p = 0.414; RF, p = 0.263; VM, p = 0.382).
Fig. 2. The initial peak response of the quadriceps in response to the rapid knee flexion stretch. Values are provided for the VL, RF, and VM muscles with different amounts of ankle loading (0, 0.2, and 0.4 N m/kg). A main effect for speed was observed for the VM muscle only (p = 0.013).
M. Murray et al. / Journal of Electromyography and Kinesiology 24 (2014) 172–177
175
anticipated that walking with 20% or 40%BWS would decrease quadriceps activity during late stance/early swing and consequently increase swing phase knee flexion. Our data instead suggested that the use of BWS did not influence knee kinematics or quadriceps activity during this phase of the gait cycle.
4.1. Effect of load on extensor excitability
Fig. 3. The prolonged response of the quadriceps in response to the rapid knee flexion stretch. Values are provided for the VL, RF, and VM muscles with different amounts of ankle loading (0, 0.2, and 0.4 N m/kg). A main effect for speed was observed for the RF muscle (p = 0.050) and a main effect for load was observed for VM (p = 0.038).
3.2. Gait kinematics and quadriceps muscle activity Our evaluation of gait kinematics and muscle activity revealed that subjects exhibited a range of paretic limb peak knee flexion angles during swing phase (range: 13°–49° of knee flexion; see Table 2). The paretic knee exhibited less knee flexion during swing phase than the non-paretic knee (p = 0.001), but there was no main effect of load on peak knee flexion (p = 0.175). Likewise, no interaction effect was observed between load and side (p = 0.972). We experienced difficulty in obtaining clear EMG signals during gait from some subjects, such that we analyzed VL for 7 subjects, RF for 8 subjects, and VM for all 10 subjects. The amount of BWS did not significantly alter quadriceps muscle activity during the late stance/early swing phase of gait (VL: p = 0.177; RF: p = 0.615; VM: p = 0.167). 3.3. Correlation of reflex measures and gait kinematics We did not observe substantial correlations between the muscle activity during gait and the muscle responses during isolated reflex testing (see Table 3). The only pair to show a significant correlation was the VL activity during the late stance/early swing phase of gait with the initial peak responses at 180°/s (r = 0.560, p = 0.008 (see Fig. 4)). The VM and RF integrals during gait were not significantly related to the initial or prolonged response at either 180°/s or 300°/s during isolated reflex testing. 4. Discussion The hypothesis that force feedback, as sensed by the plantarflexors, would modulate knee extensor excitability for individuals with chronic stroke was partially accepted. In particular, we expected that the application of ankle load would decrease knee extensor excitability during seated quadriceps reflex testing. However, we observed a reduction in the prolonged quadriceps response to 0.2 N m/kg load for the VM only. We further
Table 2 Kinematic and electromyographic variables during gait. 0%BWS
20%BWS
40%BWS
P-value
Kinematics Paretic PKF (°) Non-paretic PKF (°)
37.7 (9.9) 61.6 (8.3)
37.4 (9.7) 61.0 (7.9)
36.8 (10.2) 60.6 (8.0)
p = 0.516 p = 0.391
Electromyography VL Integral RF Integral VM Integral
16.4 (9.5) 17.9 (7.4) 14.9 (8.5)
23.0 (11.6) 16.7 (7.5) 15.6 (8.6)
19.4 (11.3) 17.6 (8.5) 17.4 (9.5)
p = 0.177 p = 0.615 p = 0.167
All values represent mean (SD).
In a cat model, load applied to the ankle plantarflexors creates autogenic excitation, but has no, or an inhibitory, effect on the quadriceps (heterogenic inhibition) during locomotion (Ross and Nichols, 2009; Wilmink and Nichols, 2003). We have extended this finding to adults with post-stroke hemiparesis, who demonstrated either no change or a reduction in quadriceps reflex excitability with the application of isolated ankle load. Although heterogenic force feedback seemed to have a limited effect on quadriceps excitability in our subjects post-stroke, force feedback is only one factor capable of modulating muscle activity and timing. Other factors, which were not measured, include exteroceptive input (Schmit et al., 2003), hip flexor length (Yang et al., 2004), and feed-forward inputs (Duysens et al., 2000). We also did not collect EMG data from the gastrocnemius or soleus muscles during testing, so we are unable to determine the autogenic response to the applied ankle load.
4.2. Effect of load on gait In our subjects post-stroke, we observed that quadriceps activity during late stance/early swing phase of gait was not influenced by altering limb load through BWS, at least up to 40%BWS. Others have previously examined the effect of BWS on quadriceps activity during early stance phase (Finch et al., 1991; Hesse et al., 1999). Because of our focus on stiff knee gait, we focused on the late stance/early swing phase of gait, normally considered a time that the uniarticular quadriceps muscles are inactive. Muscle activity during late stance is particularly important, however, because it can directly influence the stiff knee gait pattern (Goldberg et al., 2004). Furthermore, plantarflexor loads are largest during late stance (Lewek, 2011) making this an important period to examine the influence of heterogenic force feedback to the quadriceps. The timing of load application may also be an important factor in modulating muscle activity. Changes in VL activation have been shown with transient increases/decreases in load, but only when the load is applied during the early portion of stance (Stephens and Yang, 1999). Clinically, most BWS systems do not modulate load throughout the gait cycle. In such a case, where unloading is relatively constant, it appears that only lower leg muscle EMG of unimpaired individuals is altered, while upper leg muscles remained unchanged (Bachmann et al., 2008). The lack of change to quadriceps activity, by us and others (Bachmann et al., 2008; Colby et al., 1999; Finch et al., 1991), is consistent with our observed lack of change in knee flexion angles with varying amounts of BWS. Although our subjects post-stroke did not alter knee flexion angles with different levels of BWS (see also (Chen et al., 2005)), unimpaired subjects have exhibited decreased peak knee flexion angles with increased BWS (Finch et al., 1991; Threlkeld et al., 2003). While those levels of unloading (Finch et al., 1991; Threlkeld et al., 2003) surpassed the typical clinically applied limits of 40%BWS (Hesse et al., 1995; van Hedel et al., 2006), there is the possibility of a differential kinematic response to BWS when used with unimpaired subjects and those with neurological injury. Nevertheless, the literature supports our finding that thigh musculature is relatively unaffected by ankle modulated force feedback during gait.
176
M. Murray et al. / Journal of Electromyography and Kinesiology 24 (2014) 172–177
Table 3 Correlation coefficients between isolated reflex testing and gait. Initial response
VL (gait)
Prolonged response
180°/s
300°/s
180°/s
300°/s
r = 0.56; p = 0.008
r = 0.42; p = 0.06
r=
r = 0.10; p = 0.68
RF (gait)
r=
VM (gait)
r = 0.33; p = 0.08
0.26; p = 0.23
r=
0.24; p = 0.26
r = 0.34; p = 0.07
Fig. 4. Relationship between VL initial activity at 180°/s during isolated reflex testing and the VL activity during the late stance/early swing phase of gait (r = 0.56; p = 0.008). High load corresponds with 0%BWS and 0.4 N m/kg, moderate load corresponds with 20%BWS and 0.2 N m/kg and low load corresponds with 40%BWS and 0 N m/kg. Note that these different load conditions are scattered about the plot suggesting that the muscle activity during gait is not mediated by force feedback from the ankle.
4.3. Force feedback from ankle extensors Other studies have examined the effects of whole body (un)loading on ankle or hip kinematics and muscle activity (Gordon et al., 2009; Nakazawa et al., 2004; Stephens and Yang, 1999), yet few have studied the effects that occur at the knee (Threlkeld et al., 2003). In contrast to whole body (un)loading, our isolated reflex testing was designed to load the ankle extensors only. In this way, we attempted to determine what effect, if any, ankle loading had on quadriceps muscle activity. During gait, however, the extensor muscles throughout the limb, and not just the ankle plantarflexors, are loaded during stance. There are, therefore, numerous competing sensory signals that result in the net output to the muscle. In fact, it is possible that positive force feedback amongst the quadriceps (Wilmink and Nichols, 2003) is competing with inhibitory force feedback from the plantarflexors yielding no real change in quadriceps activity when BWS is altered. In addition to force feedback, the quadriceps are receiving input from monosynaptic and polysynaptic length-sensitive afferents (group Ia and II). During the seated isolated reflex testing, the knee velocity was constant throughout the motion, with a fixed ankle load. This differs substantially from the gait testing. As the knee flexes and extends throughout the gait cycle, such inputs are constantly altered making the relationship between isolated reflex data and functional gait performance difficult to interpret (Lewek et al., 2007). Given the inherent limitations in relating isolated experimental conditions to gait characteristics, it is not surprising that few correlations were noted between conditions. Nevertheless, a weak correlation was noted across subjects between VL activity during gait testing and seated quadriceps reflex testing. Specifically, data pairs (e.g., 0%BWS with 0.4 N m/kg; 20%BWS with 0.2 N m/kg) suggest that VL activity seen during gait may have a reflexive component, but is clearly not modulated by limb load. The observed
0.12; p = 0.62
r = 0.00; p = 0.99
r = 0.00; p = 0.99
r = 0.29; p = 0.131
r = 0.30; p = 0.12
relationships could also be a result of individual differences in spasticity (e.g., some subjects may have exhibited greater muscle activity than others in both reflex and gait testing, regardless of load level) and/or MVIC measures, which are difficult to ascertain in individuals post-stroke (Hsu et al., 2006). Some limitations of this research exist, and should be noted. We intentionally recruited individuals with a range of walking patterns (i.e., not all had stiff-knee gait pattern), which may have confounded the results. This decision was made because we expected that a heterogeneous group would allow us to better see how the reflex mediated quadriceps activity influenced gait parameters. In addition, some subjects (N = 3) had to wear their ankle foot orthosis (AFO) during the gait testing only (AFOs were removed for isolated reflex testing). We suspect that the use of an AFO has the potential to limit the force feedback sensed at the ankle. Additionally, subjects in our study only walked at each load level for a few minutes and it is possible that training with BWS for an extended period of time may affect knee flexion, but further research would be needed to test this assertion. Finally, several subjects required use of the handrails during the gait testing. Shifting a portion of the weight from the lower extremities to the upper extremity may have limited the load sensed at the ankle during gait. The information from this study suggests that the use of BWS during gait retraining will not result in immediate kinematic changes at the knee in those with stiff-knee gait following stroke. Clinically, this is important, as BWS may not be an appropriate modality if attempting to reduce stiff-knee gait during training. Nevertheless, the initially observed performance does not indicate what the final outcome will be (Winstein and Schmidt, 1990). While interventions that focus on other afferent information may prove effective, adjusting loading will likely not result in immediate improvements in knee flexion or reduced knee extensor muscle activity during late stance/early swing. Conflict of interest This work was supported in part by grants from NSF (Graduate Student Fellowship to AH) and UNC’s University Research Council (to MDL). Role of the funding source The study sponsors had no involvement in the study design, in the collection, analysis and interpretation of data; in the writing of the manuscript; and in the decision to submit the manuscript for publication References Ada L, Dean CM, Vargas J, Ennis S. Mechanically assisted walking with body weight support results in more independent walking than assisted overground walking in non-ambulatory patients early after stroke: a systematic review. J Physiother 2010;56:153–61. Bachmann V, Muller R, van Hedel HJ, Dietz V. Vertical perturbations of human gait: organisation and adaptation of leg muscle responses. Exp Brain Res 2008;186:123–30.
M. Murray et al. / Journal of Electromyography and Kinesiology 24 (2014) 172–177 Barbeau H, Visintin M. Optimal outcomes obtained with body-weight support combined with treadmill training in stroke subjects. Arch Phys Med Rehabil 2003;84:1458–65. Chen G, Patten C, Kothari DH, Zajac FE. Gait deviations associated with post-stroke hemiparesis: improvement during treadmill walking using weight support, speed, support stiffness, and handrail hold. Gait Posture 2005;22:57–62. Colby SM, Kirkendall DT, Bruzga RF. Electromyographic analysis and energy expenditure of harness supported treadmill walking: implications for knee rehabilitation. Gait Posture 1999;10:200–5. Conway BA, Hultborn H, Kiehn O. Proprioceptive input resets central locomotor rhythm in the spinal cat. Exp Brain Res 1987;68:643–56. Crone C, Hultborn H, Kiehn O, Mazieres L, Wigstrom H. Maintained changes in motoneuronal excitability by short-lasting synaptic inputs in the decerebrate cat. J Physiol 1988;405:321–43. Dietz V, Duysens J. Significance of load receptor input during locomotion: a review. Gait Posture 2000;11:102–10. Dietz V, Muller R, Colombo G. Locomotor activity in spinal man: significance of afferent input from joint and load receptors. Brain 2002;125:2626–34. Duncan PW, Sullivan KJ, Behrman AL, Azen SP, Wu SS, Nadeau SE, et al. Bodyweight-supported treadmill rehabilitation after stroke. New England J Med 2011;364:2026–36. Duysens J, Pearson KG. Inhibition of flexor burst generation by loading ankle extensor muscles in walking cats. Brain Res 1980;187:321–32. Duysens J, Clarac F, Cruse H. Load-regulating mechanisms in gait and posture: comparative aspects. Physiol Rev 2000;80:83–133. Finch L, Barbeau H, Arsenault B. Influence of body weight support on normal human gait: development of a gait retraining strategy. Physical therapy. 1991;71:842– 55 [discussion 55–6]. Goldberg SR, Anderson FC, Pandy MG, Delp SL. Muscles that influence knee flexion velocity in double support: implications for stiff-knee gait. J Biomech 2004;37:1189–96. Gordon KE, Wu M, Kahn JH, Dhaher YY, Schmit BD. Ankle load modulates hip kinetics and EMG during human locomotion. J Neurophysiol 2009;101:2062–76. Granata KP, Abel MF, Damiano DL. Joint angular velocity in spastic gait and the influence of muscle–tendon lengthening. J Bone Joint Surg Am 2000;82:174–86. Hesse S, Bertelt C, Jahnke MT, Schaffrin A, Baake P, Malezic M, et al. Treadmill training with partial body weight support compared with physiotherapy in nonambulatory hemiparetic patients. Stroke J Cereb Circulat 1995;26:976–81. Hesse S, Konrad M, Uhlenbrock D. Treadmill walking with partial body weight support versus floor walking in hemiparetic subjects. Arch Phys Med Rehabil 1999;80:421–7. Hiebert GW, Pearson KG. Contribution of sensory feedback to the generation of extensor activity during walking in the decerebrate Cat. J Neurophysiol 1999;81:758–70. Hsu WL, Krishnamoorthy V, Scholz JP. An alternative test of electromyographic normalization in patients. Muscle Nerve 2006;33:232–41. Kautz SA, Brown DA. Relationships between timing of muscle excitation and impaired motor performance during cyclical lower extremity movement in post-stroke hemiplegia. Brain 1998;121:515–26. Kerrigan DC, Gronley J, Perry J. Stiff-legged gait in spastic paresis. A study of quadriceps and hamstrings muscle activity. Am J Phys Med Rehabil 1991;70:294–300. Lewek MD. The influence of body weight support on ankle mechanics during treadmill walking. J Biomech 2011;44:128–33. Lewek MD, Hornby TG, Dhaher YY, Schmit BD. Prolonged quadriceps activity following imposed hip extension: a neurophysiological mechanism for stiffknee gait? J Neurophysiol 2007;98:3153–62. Mulroy S, Gronley J, Weiss W, Newsam C, Perry J. Use of cluster analysis for gait pattern classification of patients in the early and late recovery phases following stroke. Gait Posture 2003;18:114–25. Nakazawa K, Miyoshi T, Sekiguchi H, Nozaki D, Akai M, Yano H. Effects of loading and unloading of lower limb joints on the soleus H-reflex in standing humans. Clin Neurophysiol 2004;115:1296–304. Ross KT, Nichols TR. Heterogenic feedback between hindlimb extensors in the spontaneously locomoting premammillary cat. J Neurophysiol 2009;101:184–97. Rudolph KS, Axe MJ, Buchanan TS, Scholz JP, Snyder-Mackler L. Dynamic stability in the anterior cruciate ligament deficient knee. Knee Surg, Sports Traumatol, Arthrosc: Official J ESSKA 2001;9:62–71. Schindler-Ivens S, Desimone D, Grubich S, Kelley C, Sanghvi N, Brown DA. Lower extremity passive range of motion in community-ambulating stroke survivors. J Neurol Phys Ther 2008;32:21–31. Schmit BD, Hornby TG, Tysseling-Mattiace VM, Benz EN. Absence of local sign withdrawal in chronic human spinal cord injury. J Neurophysiol 2003;90:3232–41. Stephens MJ, Yang JF. Loading during the stance phase of walking in humans increases the extensor EMG amplitude but does not change the duration of the step cycle. Exp Brain Res 1999;124:363–70. Threlkeld AJ, Cooper LD, Monger BP, Craven AN, Haupt HG. Temporospatial and kinematic gait alterations during treadmill walking with body weight suspension. Gait Posture 2003;17:235–45. van Hedel HJ, Tomatis L, Muller R. Modulation of leg muscle activity and gait kinematics by walking speed and bodyweight unloading. Gait Posture 2006;24:35–45.
177
Visintin M, Barbeau H, Korner-Bitensky N, Mayo NE. A new approach to retrain gait in stroke patients through body weight support and treadmill stimulation. Stroke J Cereb Circulat 1998;29:1122–8. Waters RL, Garland DE, Perry J, Habig T, Slabaugh P. Stiff-legged gait in hemiplegia: surgical correction. J Bone Joint Surg Am 1979;61:927–33. Wilmink RJ, Nichols TR. Distribution of heterogenic reflexes among the quadriceps and triceps surae muscles of the cat hind limb. J Neurophysiol 2003;90:2310–24. Winstein CJ, Schmidt RA. Reduced frequency of knowledge of results enhances motor skill learning. J Exp Psychol Learn 1990;16:677–91. Wu M, Schmit BD. Spastic reflexes triggered by ankle load release in human spinal cord injury. J Neurophysiol 2006;96:2941–50. Yang JF, Lam T, Pang MY, Lamont E, Musselman K, Seinen E. Infant stepping: a window to the behaviour of the human pattern generator for walking. Can J Physiol Pharmacol 2004;82:662–74.
Mary Murray received her Doctorate in Physical Therapy from the University of North Carolina at Chapel Hill and her Bachelor’s degree in Communications from the University of Akron. She currently works with adult neurorehabilitation services at the Cleveland Clinic.
Amy Hardee received her master’s degree in biomedical engineering from the joint program at UNC-Chapel Hill and NC State University. She previously received her bachelor’s degree in biomedical engineering from NC State University. Currently, she works as a Development and Sustaining Engineer for Teleflex Medical.
Richard L. Goldberg received his doctorate in biomedical engineering from Duke University. He is currently the director of undergraduate studies in the Department of Biomedical Engineering at The University of North Carolina at Chapel Hill. He has expertise in rehabilitation engineering, device design, and assistive technology for individuals with disabilities.
Michael D. Lewek received his doctorate in biomechanics and movement science from the University of Delaware and performed a post-doctoral fellowship in neuromechanics at the Rehabilitation Institute of Chicago. He is currently an associate professor in the Division of Physical Therapy at The University of North Carolina at Chapel Hill. His research involves determination of the biomechanical, neurophysiologic, and motor learning aspects of walking recovery following stroke.
