Exp Brain Res DOI 10.1007/s00221-014-4099-2

RESEARCH ARTICLE

Intersegmental coordination of gait after hemorrhagic stroke John W. Chow · Dobrivoje S. Stokic 

Received: 23 June 2014 / Accepted: 4 September 2014 © Springer-Verlag Berlin Heidelberg 2014

Abstract  We compared gait using the planar law of intersegmental coordination between 14 hemorrhagic stroke subjects walking at a self-selected normal speed (56 ± 21 cm/s) and 15 age-matched healthy controls walking at a very slow speed (56 ± 19 cm/s). Sagittal plane elevation angles of the thigh, shank, and foot segments were submitted to principal component analysis. Additional outcome measures included the range of elevation angle and timing of peak elevation angle of the thigh, shank, and foot segments. The range of elevation angles at the shank and foot was significantly smaller in the paretic leg than nonparetic and control legs. Also, the peak elevation angle at the thigh occurred significantly later in the gait cycle in the paretic than control leg. Gait of both stroke and control subjects followed the planar law with the first two principal components explaining approximately 99 % of the variance. However, the three-dimensional trajectory of elevation angles (gait loop) in stroke subjects deviated from the typical teardrop shape bilaterally, which was more exaggerated in the paretic leg. Compared to the non-paretic and control legs, the paretic leg showed significantly increased absolute loading of the thigh elevation angle and decreased absolute loadings of the shank and foot elevation angles on the first principal component, whereas the opposite was observed for the second principal component. Despite following the planar law, the gait of chronic stroke subjects is characterized by atypical timing of the thigh motion and disrupted intersegmental coordination of both legs.

J. W. Chow (*) · D. S. Stokic  Center for Neuroscience and Neurological Recovery, Methodist Rehabilitation Center, 1350 East Woodrow Wilson Drive, Jackson, MS 39216, USA e-mail: [email protected]

Keywords  Locomotion · Biomechanics · Motor control · Planar covariation · Hemiparesis

Introduction The planar law of intersegmental coordination describes the coordination patterns among the elevation angles of the lower limb segments during bipedal locomotion (Borghese et al. 1996). When the sagittal plane elevation angles of the thigh, shank, and foot are plotted against each other, the trajectory of healthy subjects falls on a plane in threedimensional space and forms a “teardrop” loop pointed at the top and round at the bottom (Fig. 1, third row left). The principal component analysis (PCA) indicates that nearly all variance in the thigh, shank, and foot elevation angles can be explained by the first two principal components. It has been suggested that the planar law facilitates maintenance of dynamic equilibrium during forward progression and anticipatory adaptation to changing environments by means of coordinated kinematic synergies of the whole body (Lacquaniti et al. 1999; MacLellan and McFadyen 2010). The compliance with the planar law has been extensively studied in healthy subjects. The planar law is preserved in adults walking at different speeds (Bianchi et al. 1998; Ivanenko et al. 2002), conditions (backward, crouched, curve, and weight-supported walking) (Grasso et al. 1998, 2000; Courtine and Schieppati 2004; Ivanenko et al. 2002), and constraints (walking on incline, staircases, and over an obstacle) (Dominici et al. 2010; MacLellan and McFadyen 2010; Noble and Prentice 2008). Both biomechanical and physiological factors have been put forth to explain the planar law. Hicheur et al. (2006) emphasized biomechanical factors after demonstrating that planar covariation emerges

13



Exp Brain Res

Fig. 1  Average (±1 SD) joint flexion and segment elevation angles during a gait cycle, 3D planar covariation plot of mean-subtracted elevation angles, and biplot of the loadings of the first two principal components (from top to bottom) of a control (left) and the non-paretic (center) and paretic (right) limbs of a stroke subject (no ankle–foot orthosis). The horizontal axis of the angle plots is extended beyond 100 % gait cycle to show the transition at initial foot

contact (IFC). Vertical dashed lines in angle plots indicate the instant of toeoff (TO). Grids in 3D plots correspond to the best-fitting planes. Symbols in biplots are principal component scores (elevation angles projected onto the first two principal components) that are scaled to fit within the unit square. Symbols in 3D plots and biplots were plotted at 1 % gait cycle interval and the arrows indicate the progression in time. PC principal component

out of the strong correlation between the foot and shank elevation angles, with the thigh angle contributing independently to the pattern of intersegmental covariation. On the

other hand, it has been suggested that the planar covariation has physiological underpinning and simplifies the control of locomotion by reducing the effective degrees of freedom

13

Exp Brain Res Table 1  Stroke subject characteristics

1 2 3 4 5 6 7 8 9 10 11 12 13 14 Mean SD

Sex

Age (years)

Height (cm)

Mass (kg)

Injury type

Post onset (months)

Paretic side

Walking velocity (cm/s)

Ankle–foot orthosis

Assistive device

F F F M F M F F M M F F M F

42 32 51 41 52 38 47 29 59 20 36 44 46 32 40.6

155 160 178 175 157 183 173 152 191 183 160 160 180 163 169

118 109 70 93 76 90 88 71 105 82 59 76 86 70 85.2

ICH ICH ICH ICH ICH ICH SAH SAH ICH SAH ICH ICH ICH ICH

34 20 87 15 47 136 50 12 20 21 6 14 44 10 36.9

L R R R L R R R L R R R L L

35.3 52.5 49.4 78.5 39.9 64.4 70.6 67.1 52.5 55.9 107.7 20.4 39.4 49.1 55.9

n/a R n/a R L R n/a R L n/a n/a n/a L n/a

Quad cane n/a Cane n/a n/a Cane Cane Cane Quad cane n/a n/a n/a Cane n/a

10.4

13

16.7

35.9

21.4

ICH intracerebral hemorrhage, SAH subarachnoid hemorrhage, R right, L left, n/a not available/not applicable

(Lacquaniti et al. 1999, 2002). Since planar covariation emerges in toddlers with the development of mature gait pattern, it is considered to reflect a coordinated, centrally controlled behavior beyond biomechanical constraints (Cheron et al. 2001; Ivanenko et al. 2008). The planar law of intersegmental coordination has occasionally been used to assess effects of interventions on gait coordination, including medication and electrical stimulation in Parkinson’s disease (Grasso et al. 1999), intrathecal baclofen bolus injection in hereditary spastic paraparesis (Dan et al. 2000), and chemodenervation injections (Bleyenheuft et al. 2009) and ankle–foot orthosis (Bleyenheuft et al. 2013) in spastic stroke. Common to these interventions was that the shape and spatial orientation of the planar gait loops in patients became a closer approximation of the features seen in healthy controls. Information on lower limb intersegmental coordination after stroke is scarce. Aside from the above interventions studies (Bleyenheuft et al. 2009, 2013), MacLellan et al. (2013) examined six subjects with chronic stroke during level walking and obstacle crossing and concluded that segment covariance followed the planar law in both paretic and non-paretic legs. The main focus of these studies was on the percentage of variance explained by different principal components (Bleyenheuft et al. 2009) and the relationships between variance explained and mechanical work (Bleyenheuft et al. 2013) or between covariance plane and phase differences in elevation angles of adjacent segments (MacLellan et al. 2013). However, little attention was given

to the loadings of thigh, shank, and foot segments on different principal components. Thus, the first aim of this study was to determine whether the range and peak timing of elevation angles of the thigh, shank, and foot segments differ between chronic stroke and age-matched healthy controls. The second goal was to compare intersegmental coordination of the thigh, shank, and foot elevation angles using PCA. It was hypothesized that the gait of stroke subjects would follow the planar law but, in comparison to the controls, the intersegmental coordination would be disrupted, and more so in the paretic than non-paretic leg. This was tested by comparing profiles of elevation angles, gait loops, principal component loadings, and percentage of variance explained.

Methods A convenience sample of 14 subjects with chronic stroke was recruited from an outpatient clinic and the local community (Table 1). The inclusion criteria were first documented unilateral stroke, able to follow simple instructions, and ability to walk independently faster than 20 cm/s for at least 10 m with or without assistive devices. Although the sample was not intentionally limited to those with hemorrhagic stroke, they represented the majority and several subjects with ischemic stroke were not included for consistency. Stroke subjects were compared to 15 controls who did not report any orthopedic and neurological disorders

13



(seven men, age 40 ± 11 years; height 173 ± 12 cm; body mass 67 ± 14 kg). The age was not significantly different between the two groups (unpaired t test, P  = 0.851). All subjects signed the informed consent approved by the institutional review board for human research. Gait data were collected using eight digital cameras operated at 60 Hz (Motion Analysis Corp, Santa Rosa, CA, USA), five forceplates sampled at 1,200 Hz (Type 4060; Bertec Corp, Columbus, OH, USA), and the Helen Hayes marker system (Kadaba et al. 1990). Stroke subjects walked at a self-selected normal speed and controls at a selfselected very slow speed several times along a 7 m walkway. To preserve the gait patterns, the stroke subjects were allowed to wear a short, non-rigid polypropylene ankle– foot orthosis to prevent foot drop during gait and use customary assistive devices (straight or quad cane), if necessary (Table 1). An EvaRT data acquisition system (Motion Analysis Corp) was used to collect data synchronously. OrthoTrak Gait Analysis software (Motion Analysis Corporation) was used to process marker location data and to determine footfall instants based on a combination of ground reaction force and foot kinematics (Zeni et al. 2008). A gait cycle (GC) was delineated by two consecutive initial foot contacts of the same foot. Temporospatial variables were calculated as described in Chow et al. (2010). The joint range of motion was calculated as the difference between the maximum and minimum flexion angles during a GC. The hip flexion angle was the angle between the vertical and the thigh (positive when the knee was anterior to the hip). The knee flexion angle was the angle between the distal extension of the thigh and the shank. The ankle flexion angle is zero at the neutral position and positive for dorsiflexion. The elevation angle of a limb segment was computed as the inclination angle of the segment relative to the vertical in the sagittal plane, with a positive sign when the distal endpoint is anterior to the proximal endpoint of a segment (Borghese et al. 1996). The foot elevation angle is >90° when the toe marker is higher than the heel marker. For each subject, the elevation angles of thigh, shank, and foot segments were normalized to 100 % gait cycle at 1 % increment and the average angles over multiple gait cycles for each limb were used in subsequent analysis. The range of elevation angle was defined as the difference between the maximum and minimum values during a GC. The timing of maximum (peak) elevation angle was expressed as a percentage of the GC duration. PCA is a data reduction technique that creates composite variables (principal components, PCs) from the original variables with each PC representing a weighted sum of the original variables (Sainani 2014). The PCA was run using the princomp function in MATLAB (The MathWorks, Inc., Natick, MA, USA). The parameters of interest were

13

Exp Brain Res

coefficients (factor loadings of the three elevation angles) for the three PCs (PC1, PC2, and PC3 are the eigenvectors of the covariance matrix) and percent variance (%) explained by each PC (PV1, PV2, and PV3). Each PC is the weighted linear combination of the three elevations angles that explains as much of the variance of the input data as possible. The loading of each elevation angle ranges from −1 to +1, and the magnitude indicates the amount of variance in the elevation angle that is captured by the given PC (i.e., the influence of a segment elevation angle on a PC). Because the eigenvectors of the covariance matrix are unit length (i.e., the square root of the sum of squared loadings equals to one for each PC), the absolute loadings were directly compared. Covariation of elevation angles was examined by plotting thigh angle versus shank angle versus foot angle in three-dimensional (3D) space. In normal walking, the trajectory forms a loop and lies very close to the best-fitting plane (Fig. 1, third row). The loadings of PC3 define the orientation of the covariation plane (a unit vector normal to the best-fitting plane). PV1 and PV2 reflect the height and width of the gait loop, respectively, and PV3 is the planarity index of the gait loop (PV3 = 0 for a complete planar covariation) (Bleyenheuft et al. 2009). Some investigators define the planarity index as the sum of PV1 and PV2 (same as 100 % minus PV3) (Ivanenko et al. 2008). Both the loadings and principal component scores of the first two PCs can be graphically represented in a biplot to assist in the interpretation of PCA results (Fig. 1, bottom row) (De Wit et al. 2009). This way the loading of each variable is represented by a vector and the direction and length of the vector indicate how each variable contributes to PC1 and PC2. Statistical analysis Because the use of an ankle–foot orthosis may alter gait kinematics after stroke (Leung and Moseley 2003; Tyson and Kent 2013), we used a series of Mann–Whitney U tests (α = 0.01) to compare paretic leg of subjects who walked with (n  = 7) and without (n  = 7) an ankle–foot orthosis. There were no significant differences in temporospatial (P  ≥ 0.259), elevation angle (P  ≥ 0.026), and PCA (P  ≥ 0.097) parameters. The subjects with orthosis had a smaller ankle range of motion during a GC (14.3° ± 2.5°) compared to those without (20.0° ± 6.2°), but the difference was not significant (P  = 0.128). Thus, all patients were pooled for the main analysis. Unpaired t tests were used to compare gait speed, stride length, and cadence between stroke subjects and controls. A one-way analysis of variance was applied to determine the main effect of group (paretic, non-paretic, and control) for the remaining temporospatial parameters, joint range

Exp Brain Res

of motion, range and peak timing of elevation angles, and PCA outputs. When the main effect was significant, Tukey post hoc analysis was used to compare paretic and nonparetic legs to control, whereas the difference between paretic and non-paretic legs was evaluated using a paired t test. Considering the number of statistical tests performed, the P value was set at a more stringent level of 0.01.

The typical teardrop shape was observed in the paretic leg of only three stroke subjects, whereas in the remaining stroke subjects the gait loops were narrower, twisted (figure-ofeight shape) or without the tail (Fig. 1, third row right). The three stroke subjects who exhibited atypical gait loop in the non-paretic leg also had atypical gait loop in the paretic leg. Principal component analysis

Results General gait characteristics Despite no difference in gait speed (P = 0.957), the stroke subjects walked with a shorter stride length (P = 0.027) and a higher cadence (P  = 0.013) than controls. Step length, toe clearance, and stance time were significantly different (P ≤ 0.005) between the groups (Table 1). The non-paretic step length and toe clearance bilaterally were smaller in stroke than controls (P ≤ 0.004), whereas the stance time was longer in the non-paretic than either the paretic or control leg (P  ≤ 0.001), typical for hemiparetic gait. The range of motion in the knee and ankle joints, but not hip, was significantly different between the groups (P ≤ 0.001). The knee range of motion was smaller in the paretic than either the non-paretic or control leg (both P 

Intersegmental coordination of gait after hemorrhagic stroke.

We compared gait using the planar law of intersegmental coordination between 14 hemorrhagic stroke subjects walking at a self-selected normal speed (5...
4MB Sizes 2 Downloads 8 Views