- Email: [email protected]
Soleus long-latency stretch reflexes during walking in healthy and spastic humans
Clinical Neurophysiology 110 (1999) 951±959
Soleus long-latency stretch re¯exes during walking in healthy and spastic humans q Thomas Sinkjñr a,*, Jacob Buus Andersen a, Jùrgen Feldbñk Nielsen a, Hans Jacob Hansen b a
Center for Sensory-Motor Interaction, Department of Medical Informatics and Image Analysis, Aalborg University, Fredrik Bajers Vej 7D-3, DK-9220 Aalborg, Denmark b The Hospital for Treatment of Multiple Sclerosis, Ry, Denmark Accepted 15 December 1998
Abstract The present study was carried out to investigate the long-latency soleus stretch re¯exes M2 (peak latency of approximately 85 ms) and M3 (peak latency of approximately 115 ms) during walking in healthy and spastic multiple sclerosis (MS) patients. An 88 stretch was applied to the ankle extensors of the left leg in 8 healthy subjects during normal walking speed and 9 spastic MS patients and 10 age-matched healthy subjects during slow walking. When present in walking healthy subjects, M2 and M3 were modulated in a similar way and with the same amplitudes as previously described for the short latency soleus stretch re¯ex (M1). The spastic patients' soleus M1 was signi®cantly less modulated during walking. The patients' M2 long-latency response was modulated in the same way as the age-matched healthy subjects. All patients' M3 responses were absent or much suppressed during walking. The origin and functional importance of the short- and long-latency stretch re¯exes in healthy and spastic persons are discussed in relation to the above ®ndings and the behaviour of the stretch re¯exes during matched isometric contractions. M3 is argued to be part of a transcortical re¯ex in healthy subjects. q 1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Walking; Short and long-latency stretch re¯exes; Spasticity; Soleus muscle; Motor control
1. Introduction The neural mechanisms that make the activity in the leg muscles adapt to the environmental conditions during walking rely on continuous afferent input to the spinal cord. Such afferent information is believed to be particularly appropriate for correction of unanticipated perturbations. Most such compensatory mechanisms stem from proprioceptive information from the muscle, skin, and joint receptors in the legs (e.g. Dietz, 1997). In the healthy sitting human subject, an imposed dorsi¯exion of the ankle joint causes a series of distinct responses in the electromyogram (EMG) of the stretched ankle extensors (e.g. Gottlieb and Agarwall, 1979). The ®rst response is the short latency stretch re¯ex (labelled `M1' in this study) with an onset latency at 40±50 ms and a peak latency at 50± 60 ms. During an isometric contraction of the ankle extensors, the M1 re¯ex is typically followed by a long-latency q This study was ®nancially supported by The Danish National Research Foundation and The Danish Multiple Sclerosis Society. * Corresponding author. Tel.: 1 45-96-35-88-27; fax: 1 45-98-15-4008. E-mail address: [email protected] (T. Sinkjñr)
1388-2457/99/$ - see front matter q 1999 PII: S 1388-245 7(99)00034-6
Elsevier Science Ireland Ltd.
re¯ex response (M2) with a peak latency at 70±90 (Toft et al., 1991; Fellow et al., 1993). In some subjects, a second long-latency stretch re¯ex (M3) appears (Gottlieb and Agarwall, 1980; Gottlieb et al., 1983). The afferents primarily responsible for the onset of the M1 response are the stretch velocity-sensitive group Iaafferents from muscle spindles (Taylor et al., 1985; Stein et al., 1991; Matthews, 1991). The latency of the M1 response is so brief that its re¯ex arc is certainly spinal and likely monosynaptic. The origin and pathways of the M2 and M3 responses are, however, still a matter of dispute (Berardelli et al., 1982; Thilmann et al., 1991; Dietz, 1992; Fellow et al., 1993; Corna et al., 1995). Several observations support the view that the M2 response in the soleus muscle is mainly mediated by the slow conduct of the spindle group II afferent ®bres through an oligosynaptic spinal pathway (Dietz, 1992; Nardone et al., 1996). Dietz et al. (1985) showed an enhancement of the M2 during walking compared to a tonic contraction during standing when a mechanical perturbation is applied by a sudden acceleration or deceleration of a treadmill. This and the latency led them to suggest that at least the early part of the long-latency responses is a polysynaptic
. All rights reserved.
CLINPH 98656
952
T. Sinkjñr et al. / Clinical Neurophysiology 110 (1999) 951±959
spinal compensatory re¯ex response mediated mainly by muscle proprioceptive input from group II afferents (Dietz et al., 1987). Others have argued that the physiological properties of the long-latency M2 re¯exes are similar to those of the short-latency re¯ex. This makes them represent responses of the motoneurone pool to successive Ia bursts (Berardelli et al., 1982) or responses transmitted over a more complex spinal pathway (Fellow et al., 1993). The origin of the M3 response is also not clear. Gottlieb et al. (1983) showed that the appearance of a late response (seen after 90±100 ms) was reduced, but still present, in the ankle extensors after ischemia had blocked the transmission in the group Ia afferents from the lower leg. Nielsen et al. (1998) recently showed a block of the M3 soleus muscle response after ischemia and during walking. By applying transcranial stimulation, Petersen et al. (1998) demonstrated that in the tibialis anterior, the long-latency M3 response to a stretch applied to the dorsi¯exors is likely to be mediated by a transcortical re¯ex loop through the group Ia afferents. When present, the latency of the M3 response in the soleus muscle to a stretch applied to the ankle extensors occurs at similar latency as the long-latency M3 in tibialis anterior (Sinkjñr et al., 1996b). The functional importance of the long-latency re¯exes has been addressed in several studies (see Dietz, 1992, for a brief review). For subjects jumping from some height or hopping on one foot, the long-latency re¯exes were much more powerful than the spinal M1 re¯ex (Melvill Jones and Watt, 1969). This made the authors refer to the long-latency re¯exes as the `functional stretch re¯ex' which they suggested was of transcortical origin. In this study, the origin and functional importance of the long-latency re¯exes of the soleus muscle were investigated. This was done by comparing the short- and longlatency stretch re¯exes of the soleus muscle during sitting and standing, with the re¯exes during walking at matched excitation level of healthy and spastic multiple sclerosis (MS) patients' soleus muscle. Furthermore, recordings from patients are of importance in assessing the value of long-latency re¯exes in clinical diagnosis (Diener et al., 1985; Deuschl et al., 1988).
2. Materials and methods To describe the M1, M2, and M3 re¯exes, two protocols were used. In protocol 1, healthy subjects' stretch re¯exes were elicited during normal walking speed and at matched excitation levels during standing. To investigate if the longlatency stretch re¯ex M3 was of transcortical origin, we compared re¯exes in healthy subjects and spastic patients with a central lesion in protocol 2. All subjects gave their informed consent before the inclusion in this study. The study was approved by the local
ethical committee in accordance with the ethical standards in the Declaration of Helsinki. 2.1. Experimental set-up In both protocols, a portable stretch device capable of rotating the human ankle joint during walking on a treadmill, elicited the stretch re¯exes. The system consisted of a mechanical joint, mounted in level with the ankle joint that was connected to a powerful actuator system by means of two ¯exible bowden wires. Five different polypropylene plaster casts were made to give a unique interface from the mechanical joint to the ankle of the subject. For further details on the system, see Andersen and Sinkjñr (1995). In the experiment, the mechanical joint was strapped to the subject's left leg. A heel contact was placed in the subject's left shoe, and an insole was placed in the right shoe to match the different height caused by the casting in the left shoe. Bipolar EMG electrodes were placed 2 cm apart; on the soleus muscle about 10 cm above the calcanus, and on the tibialis anterior muscle about 10 cm below the patella and 2 cm from the tibia bone. A ground electrode was placed under the knee. The EMGs were ampli®ed and ®ltered from 20±1000 Hz (DISA, model 15C01). 2.2. Protocol 1 The experiment was performed on 8 healthy subjects (6 males and two females, age 22±32 years). The subjects were told to walk with a natural cadence at 3.5±4 km/h. After an adaptation period of 5±10 min, an EMG pro®le, triggered from heel contact to heel contact, was performed on the soleus and tibialis anterior muscle based on 20 steps. The step cycle was divided into 10 segments. An 88 stretch was made with a velocity of 2808/s randomly between the segments and with an interval of 4±6 steps apart until 8± 10 stretches were accomplished in each time segment. After walking, the subjects were seated in a chair with the left shoe fastened to the ground. The subjects were told to maintain a static level of the soleus EMG that corresponded to the level of walking at approximately 20% of stance phase. In order to maintain the level, the subjects were given a visual feedback of the recti®ed, lowpass ®ltered (1st order 2 Hz) soleus EMG on a monitor. The subjects were told to contract to the static level with an interval of about 4 s until 8±10 trials were averaged. The static contraction experiment was repeated during standing. 2.3. Protocol 2 The protocol included: 9 patients with multiple sclerosis (MS) with a mean age of 43 years (range 27±62) matched with 10 control subjects with a mean age of 38 years (range 25±55). The criteria for inclusion of the patient were: (1) de®nite MS according to criteria by Poser et al. (1983); (2) stable, neurological condition for at least 6 months; (3)
T. Sinkjñr et al. / Clinical Neurophysiology 110 (1999) 951±959
953
suf®cient strength, co-ordination, and range of motion around hip, knee, and ankle joints to walk on a treadmill for 100 m; (4) clinical signs of spasticity in the lower extremities with an increase in muscle tone and exaggerated tendon re¯exes; (5) no contractures or major sensory impairments. The mean duration of MS from the ®rst symptom was 11 years (range 5±25). One patient took antispastic medicine (30 mg baclofen per day). The control subjects were healthy and did not take any medication. In all subjects, the left leg was investigated. The patient recordings were made at their preferred walking speed, which ranged from 0.9±3.1 km/h (mean value: 1.6 km/h). In this protocol, the healthy control subjects walked at a matched walking speed of 1.6 km/h. For each subject the average stride length (^1 SD) was calculated from approximately 30 steps. From a switch placed under the heal of the subject, the stride length was measured as the time difference from one heel contact to the succeeding. The patients walked for 3±5 min without rest. After walking, the patients were seated in a chair with the left shoe fastened to the ground, and the experiments during sitting were repeated as described in protocol 1. 2.4. Data analysis The data were analysed on a computer as follows. The EMG activity of the soleus and tibialis anterior muscles was averaged, recti®ed, and lowpass ®ltered at 20 Hz (1st order). The re¯ex amplitudes of M1, M2, and M3 were measured as the peak amplitude minus an average background activity of 30 ms before the onset of the re¯ex response. The EMGs and ankle position during a perturbation were only averaged if the size of the stretches and velocities of the stretches were within a pre-set range. For all subjects, the stretch amplitude and velocity were kept constant throughout a step. Eight±ten stretch re¯ex responses were averaged together in each segment. 2.5. Statistics The data were tested by the Wilcoxon±Mann±Whitney test for independent samples and by the Wilcoxon signed ranks test for matched samples with a level of signi®cance of P , 0:05. Fig. 1. Example of short- and long-latency stretch re¯exes during walking. (A) Position of the ankle joint during the stance phase. The thin line shows an average of 10 individual steps with an 88 stretch elicited at time zero. The onset of the stretch corresponds to 250 ms after heel contact. The thick line is an average of 8 control steps. (B) Averaged soleus EMG re¯exes to the imposed dorsi¯exion movement shown in (A). The thin line shows the recti®ed and ®ltered soleus EMG with a short-latency re¯ex (labelled M1) followed by two long-latency re¯exes (M2, M3). The soleus EMG activity is superimposed on the EMG activity from a control step (thick line). (C) Position of the ankle joint in control steps and the stretch amplitude. (D) Velocity of the ankle joint and the imposed stretch velocity during the step. (E) Modulation of M1, M2, and M3 soleus stretch re¯exes during the entire step. The stretch is kept constant at a displacement of 88 and a displacement velocity of 2808/s. Subject DM walking at a speed of 3.8 km/h.
3. Results 3.1. Short- and long-latency stretch re¯ex modulation during normal walking speed in healthy subjects During walking, the EMG responses in the soleus muscle to the stretch of the ankle extensors consisted of 1±3 peaks. An example is shown in Fig. 1. Fig. 1A,B illustrates the time course of one healthy subject's ankle position and soleus EMG in early stance during normal walking speed. Each trace shows superimposed average responses with control
954
T. Sinkjñr et al. / Clinical Neurophysiology 110 (1999) 951±959
steps (heavy lines) and steps, in which a stretch was elicited (thin lines). At time zero, the steps of an imposed displacement de¯ected from the position of the control steps (Fig. 1A). The 08 equals the ankle joint position at heel contact. The 88 displacement had a risetime of 35 ms and kept the position nearly ®xed for 200 ms. When the stretch was released after 200 ms, the position returned to the same level as in the control step within 50 ms. Prior to the stretch, the soleus EMG matched the soleus EMG in the control steps until approximately 40 ms after stretch onset (Fig. 1B). At this time, a distinct short-latency stretch re¯ex (M1) was followed by two long-latency EMG components (M2, M3). In this subject, M1 had a peak latency of 50 ms. The peak latencies of M2 and M3 were 85 ms and 112 ms. After the M3 response, the EMG activity returned to the same level as the background EMG of the control step. No responses were seen in the tibialis anterior, except a small peak caused by the crosstalk from the prominent EMG stretch re¯ex of the soleus muscle (not shown). Fig. 1C shows ankle joint position in the control step and the amplitude, and Fig. 1D the velocity of the ankle joint and
the imposed stretch during a full step cycle for the same subject. The stretch was kept constant throughout the gait cycle at a displacement of 88 and a stretch velocity of 2808/s. Fig. 1E illustrates the short- and long-latency soleus stretch re¯exes when the soleus background EMG was subtracted for the same subject. During a step, they were all clearly modulated with high amplitudes in part of the stance phase and near zero values in the swing phase. The M1 response built up in the stance phase was completely inhibited in the transition from stance to swing and again slowly increased during the swing phase as described earlier (Sinkjñr et al., 1996b). The M2 and M3 were only present in two out of the 10 segments in this subject. The re¯exes all peaked in mid- or late-stance phases. None of the stretch re¯exes were different from zero in the transition from stance to swing (Fig. 1E). 3.2. Appearances of M1, M2, and M3 during walking A large inter-subject variation in the appearance of the re¯exes was found in protocol 1. In several subjects, the long-latency re¯exes were either not present, or only present in one or two of the 10 points during a step where the ankle extensors were stretched. Fig. 2A shows the number of persons with an observed M1, M2, and M3 for each of the 10 segments during a step. As 8 people were included, the observation 8 was the highest possible per re¯ex for each segment. For example, at 20% of the step, M1 was seen in all 8 people, M2 in 6 people, and M3 also in 6 people. 3.3. Latencies and amplitudes for M1, M2, and M3 during walking Fig. 2B shows the average peak latencies of M1, M2, and M3 for all 8 healthy subjects during a normal step cycle (protocol 1, see Section 2). The latencies measured from stretch onset to peak EMG did not depend on the time at which they were elicited during the gait cycle. The averaged latency (^1SE) of the M1 response was 56:0 ^ 0:7 ms. The average latency of the M2 response was 84:9 ^ 1:3 ms, and the average latency of the M3 response was 113:9 ^ 3:4 ms. On average, M1, M2, and M3 all modulated during a step with high amplitudes in part of the stance phase and near zero values in the swing phase. Within each segment, no differences were found in the amplitude of the re¯exes (P . 0:05). On average, they all behaved alike during a step (Fig. 2C).
Fig. 2. M1, M2, and M3 stretch re¯ex amplitudes and latencies during a normalized step. (A) Number of M1, M2, and M3 re¯exes observed in each of the 10 segments during a step. If a M1 re¯ex is observed in 6 of the 8 investigated subjects, it will be displayed as a column with the height of 6 in the segment of interest. On top of the `M1 column', the columns for M2 and M3 are shown. (B) Averaged peak latencies of the M1, M2, and M3 stretch re¯exes. (C) Averaged M1, M2, and M3 stretch re¯ex amplitudes. ^ 1 SE, 8 subjects.
3.4. M1, M2, and M3 during isometric contraction and during walking at matched EMG To quantitate how the re¯exes depend on the motor task, we measured the stretch re¯ex at matched contractions (background soleus EMG) during early stance phase of walking, in the standing subject, and during sitting, applying the same stretches in all 3 tasks in 4 subjects from protocol 1 (normal walking speed) and in 8 healthy subjects from
T. Sinkjñr et al. / Clinical Neurophysiology 110 (1999) 951±959
Fig. 3. Amplitude of the M1, M2, and M3 soleus stretch re¯exes at matched background EMG. (A) During early stance phase of normal walking speed (20% of the step), during standing, and during sitting. ^ 1 SE, 4 subjects, protocol 1. (B) Amplitude of the M1, M2, and M3 soleus stretch re¯exes at matched background EMG during early stance phase of slow walking (20% of the step) and during sitting. ^ 1 SE, 8 healthy control subjects, protocol 2.
protocol 2 (slow walking). In the healthy subjects in protocol 2, we only compared sitting and walking. Fig. 3 shows the results. At normal walking speed and at matched soleus background EMG (Fig. 3A), no differences in M1 were found for the 3 tasks, as reported earlier (Sinkjñr et al., 1996b). M2 was non-signi®cantly decreased (P . 0:05) during sitting compared with standing and walking. No M3 was found during sitting and standing. In the healthy control subjects of protocol 2, M2 was non-signi®cantly decreased during sitting compared with walking; 8 subjects. No M3 was found during sitting whereas M3 was seen in the same subjects during the stance phase of walking.
955
tude and stretch velocity were kept constant in all segments and at 88 and 2808/s for all subjects. In the control subjects, the M1 stretch re¯ex response modulated from an average maximum amplitude of 42:6 ^ 10:87 mu;V at 40% of the step declining to 4:0 ^ 2:24 mV at 60% at the end of the stance phase. As described earlier (Sinkjñr et al., 1996b), MS patients do not modulate the M1 as well as the control group. They had a maximum amplitude of 89:0 ^ 32:3 mV at 10% and a minimum of 41:8 ^ 9:23 mV at 40% of the gait cycle. The M1 stretch re¯ex response was on average shifted upwards in the patients with a signi®cantly higher level in 7 out of 9 points during the total step (P , 0:05). The average amplitude of the M1 short-latency soleus stretch is shown in Fig. 4A. No signi®cant differences were found between peak latencies of patients and control subjects M1 (P . 0:05). The averaged M2 stretch re¯ex was modulated equally in patients and the control group, with an increasing response during the stance phase reaching a maximum of 46:8 ^ 8:95 mV for control subjects and 43:5 ^ 28:6 mV for patients at 30% of the step (Fig. 4B). For both groups, the M2 response was declined to 4:1 ^ 3:2 mV for control subjects and 5:2 ^ 5:2 mV for patients at 60% of the step. Except for one point in mid-swing (at 70% of the step in Fig. 4B), the M2 for the two groups was not signi®cantly different (P . 0:05). In addition, no signi®cant differences were found between the peak latencies of patients' and control subjects' M2 stretch re¯ex (P . 0:05). The M3 stretch re¯ex was only seen in 4 observations in the patient group (in 3 patients), whereas 36 observations of M3 were made in the control group (in 5 subjects). In the control group, the M3 had a maximum amplitude of 57:3 ^ 16:0 mV at 30% with a diminishing response of 1:0 ^ 1:0 mV at 50% (Fig. 4C). During the step cycle, the background activity of the soleus and the tibialis anterior was reciprocally activated in the investigated legs of both the control and patient groups (Fig. 4D,E). In the stance phase, patients' soleus background EMG (Fig. 3D) reached a maximum of 49:8 ^ 6:4 mV at 50% of the gait cycle. This was signi®cantly higher than the maximum background EMG of the control group which reached a maximum of 31:1 ^ 5:0 mV at 40%. The average background activity of the tibialis anterior muscle in patients was not signi®cantly different at any point of the gait cycle in relation to the activity of the control group.
3.5. The M1, M2, and M3 in MS patients To get a better understanding of the origin of the longlatency re¯exes, we compared the latencies and amplitudes of the re¯exes during walking in MS patients (protocol 2 in Materials and methods). Fig. 4 illustrates the M1, M2, and M3 stretch-re¯ex modulation during an entire step for 9 MS patients (thin line) and 10 age-matched control subjects (heavy line), at similar walking speed. The stretch ampli-
4. Discussion When present, the 3 stretch re¯exes in the soleus muscle (M1, M2, and M3), were modulated in a similar fashion and with a similar amplitude that will functionally add to the torque at the ankle joint during a step, as earlier described for the short-latency stretch re¯ex (Yang et al., 1991; Sinkjñr et al., 1996a). At matched soleus EMG excitation level,
956
T. Sinkjñr et al. / Clinical Neurophysiology 110 (1999) 951±959
the amplitudes of M1 and M2 re¯exes were of comparable sizes during sitting, standing and walking. M3 is rarely present during isometric contractions during sitting and standing (as observed before by Gottlieb and Agarwall, 1979; Berardelli et al., 1982; Toft et al., 1993), but is seen consistently in 5 out of 8 subjects during normal walking. The M2 long-latency re¯ex in the patients was modulated in the same way as in the age-matched healthy subjects. The M3 stretch re¯ex was absent, or much suppressed in all patients during walking. 4.1. The short-latency M1 soleus stretch re¯ex during walking M1 was present in all subjects and most pronounced in the stance phase (Fig. 2C), as earlier described by Sinkjñr et al. (1996a). The M1 re¯ex was modulated during a step with a high amplitude in the stance phase and a zero amplitude in the transition from stance to swing. The afferents primarily responsible for the onset of the M1 stretch re¯ex are thought to be the stretch velocity sensitive group Ia-afferents from muscle spindles (Taylor et al., 1985; Matthews, 1991; Stein et al., 1991). The structures responsible for the re¯ex modulation are not known, but animal studies (Akazawa et al., 1982) showed that short-latency stretch re¯ex modulation is also present in the premammillary cat during walking. This suggests that at least part of the neural mechanisms responsible for the re¯ex modulation is located in the spinal cord. During walking and sitting at matched excitation level, the similar size of M1 responses in healthy subjects suggested that the short-latency stretch re¯ex was less sensitive to motor task speci®c presynaptic inhibition compared with the H-re¯ex. A different composition and/or temporal dispersion of the afferent volleys evoked by electrical and mechanical stimuli may explain this difference (Morita et al., 1998). 4.2. The long-latency M2 soleus stretch re¯ex during walking
Fig. 4. Averaged re¯ex modulation during an entire step for all MS patients (thin line) and all control subjects (heavy line). (A) The average amplitude of the M1 short-latency soleus re¯ex. (B) The average amplitude of the M2 long-latency soleus stretch re¯ex. (C) The average amplitude of the M3 long-latency soleus stretch re¯ex. The M1, M2, and M3 are calculated as peak EMG activity with the background EMG subtracted. (D) The average background soleus EMG in the control steps. (E) The average background anterior tibialis EMG in the control steps. Mean ^ 1 SE is shown. A q indicates that the two groups are signi®cantly different (P , 0:05)
When the ankle extensors were stretched during the midstance of walking at velocities of nearly 3008/s, a strong EMG response appeared with the latency of the M2 re¯ex (Figs. 1B and 2B). If the M2 response simply represents responses of the motoneurone pool to successive Ia bursts as proposed by, e.g. Berardelli et al., (1982), then the lack of modulation in the spastic patients' M1 (Fig. 4A) is expected also to hold for the M2 response in the same patients. This is not the case (Fig. 4B); M2 in patients modulated in a similar way as M2 in healthy subjects. The onset latency of M2 (approximately 65 ms) was too short for a transcortical loop. This supported the view that the response was mainly of spinal origin. Investigations during walking have shown that the M2 responses are preserved when the group I afferents are blocked by ischemia (Dietz, 1992; Nielsen et al., 1998). Several observations support the view that the M2 responses
T. Sinkjñr et al. / Clinical Neurophysiology 110 (1999) 951±959
in the leg are mainly mediated by muscle proprioceptive input from group II (Corna et al., 1995) (see also Dietz, 1992). These observations were recently supported by investigations in standing man (Nardone et al., 1996). The monosynaptic excitation from group II afferents to motoneurones is considered weak compared with that of group Ia (Lundberg et al., 1977), and is probably contributing little to the M2 responses seen here. The modulation of M2 could take place in the more complex polysynaptic pathways which, at least in the cat, are known to in¯uence the spinal locomotor neuronal network importantly (Edgley et al., 1988). In patients and control subjects, the similar EMG pattern at matched walking speed (Fig. 4D,E) and the patients' normal modulation of the M2 are consistent with the preservation of this part of the neural circuitry in spastic MS. 4.3. The long-latency M3 soleus stretch re¯ex during walking Yang et al. (1991), who investigated the stretch re¯ex in the early stance phase of walking, observed a double EMG soleus burst in two out of 6 subjects. The ®rst burst is consistent with the M1 short-latency stretch re¯ex described here, whereas the second burst peaked at a time (approximately 120 ms after stretch onset) which is consistent with the M3 stretch response observed in this study. In the present study, the M3 response was not seen in sitting and standing subjects (Fig. 3). This is consistent with earlier studies in which we only observed a M3 response in one out of 6 subjects during experiments in sitting subjects (Toft et al., 1991). In contrast to this, the M3 response was seen in 5 out of the 8 subjects during normal walking. A recent study showed that during isometric contractions, M3 in the dorsi¯exors is part of a transcortical re¯ex loop through motor cortex (Petersen et al., 1998), most likely evoked by muscle proprioceptive input from group I. In the present study, the peak latency of M3 (120 ms) was not signi®cantly different from the one found in the dorsi¯exors (Toft et al., 1989; Petersen et al., 1998). If M3 in the soleus muscle is part of a transcortical loop, we suggest that one of the reasons for not observing it in the patients is a large temporal dispersion between the contributing motor unit action potentials as the nerve conduction velocity decreases in MS patients (Caramia et al., 1988; Jones et al., 1991). This results in a pronounced phase cancellation between the positive phase and the negative phase of the different motor unit potentials. This is evident from Fig. 1 in Deuschl et al. (1988), in which the response to a very synchronized electrical stimuli of the medial or radial super®cial nerve is averaged more than 100 times, and still the delayed re¯ex response in MS patients' thenar muscle is not impressive. A stretch mediated input, as in the present study, will deliver a much less synchronized input to the CNS, which has to travel a much longer path in the leg compared with the
957
arm. Both issues will make it even more unlikely to detect a delayed M3 response in MS patients' leg. In another patient group with a central motor lesion after apoplexia cerebri, we have earlier observed that M3 is either absent or only seen as single points in the step cycle (Sinkjñr et al., 1997). This supports that M3 is part of a transcortical re¯ex loop. At present, however, the lack of M3 in the patients cannot be excluded to be due to an inhibition or facilitation loss of a spinal re¯ex due to an impaired supraspinal control. Further investigations (for example, by applying magnetic stimulation above the motor cortex timed to the expected arrival of the afferents volley from the receptors in the stretched ankle extensor muscles) might be able to clarify if M3 in the soleus muscles is part of a transcortical re¯ex loop. In healthy humans, the M3 response of soleus was interestingly not found during sitting and standing. If M3 is transcortical, this implied that the brain was less involved in the control of the ankle extensors to an unexpected event during such motor tasks than it was during the automatic task of walking. Contrary to this, the M3 responses in the preactivated tibialis anterior muscle are seen consistently during sitting (Toft et al., 1989) and walking (Christensen et al., 1998). Whether this implies that a transcortical re¯ex pathway plays a more important and integrated part of the control of dorsi¯exors than a transcortical re¯ex pathway does for the ankle extensors, remains to be investigated. The M2 response has been demonstrated to be likely transcortical in the upper limb, whereas the data from the leg suggest that M3, but not the M2 response, is transcortical. This difference between the upper and lower limb is due to misleading labelling, since what is called an M2 response in the upper limb corresponds in terms of the underlying mechanism to what is called an M3 response in the lower limb (Petersen et al., 1998). 4.4. Functional considerations The short-latency feedback mechanism makes the stretch re¯ex adjust the ankle joint impedance in accordance with the external constraints in the fastest possible way. If the environmental context deviates from the average natural perturbations received by the ankle during walking (e.g. if the ankle extensors are being stretched above the averaged eccentric contractions due to an incline in the surface during the stance phase), a stronger afferent input from the ankle extensor muscles is proposed to compensate for this through M1 as well as the long-latency stretch re¯exes M2 and M3. They all add to the muscle activity to stiffen up the ankle joint. Such compensation happens most likely in every step cycle with more or less distinct muscle responses depending on the environmental context. The long-latency stretch re¯exes (M2, M3) are only present when the muscle is active, and are only observed in mid-stance during walking. They were not found in all subjects in this study. Often a strong and complex perturba-
958
T. Sinkjñr et al. / Clinical Neurophysiology 110 (1999) 951±959
tion is required to elicit them (Berardelli et al., 1982; Quintern et al., 1985). When elicited, these long-latency responses are of similar size or larger than the M1 response (see also Dietz et al., 1987). With regard to standing, fast and co-ordinated postural adjustments of the lower limb to a perturbation are assumed to rely mainly on a spinal re¯ex circuit as the receiving group I muscle input (to elicit the short latency M1 re¯ex) and group II muscle input (to elicit the M2 long-latency re¯ex). As the polysynaptic group II pathways are part of a circuit with convergence from various peripheral receptors and supraspinal centres (Jankowska, 1992), ample possibilities for a functional modulation are present. The lack of M3 in the soleus muscle during sitting and standing suggests that its importance is linked to more dynamic motor tasks such as walking. The response is assumed to be part of a supraspinal, mediated compensatory re¯ex response, which in¯uences the stereotyped generated motor output when unexpected strong afferent input is detected at the supraspinal level. It is not known at present whether the M3 is more pronounced when cortical control becomes dominant with respect to the spinal cord, as might happen, for example, when a new and not yet trained walking pattern is required or in skilled locomotion. The lack of the long-latency M3 re¯ex in spastic gait might worsen the spastic patients' ability to compensate for large, unexpected perturbations impairing their overall gait performance, as it might impair their ability to do skilled walking if the M3 is important in such tasks. References Andersen JB, Sinkjñr T. An actuator system for investigating electrophysiological and biomechanical features around the human ankle joint during gait. Trans Rehab Eng 1995;3(4):299±306. Akazawa K, Aldridge JW, Steeves JD, Stein RB. Modulation of stretch re¯exes during locomotion in the mesencephalic cat. J Physiol 1982;329:553±567. Berardelli A, Hallett M, Kaugman C, Fome E, Berenberg W, Simon SR. Stretch re¯exes of triceps surae in normal man. J Neurol Neurosurg Psychiatry 1982;45:513±525. Caramia MD, Bernardi G, Zarola F, Rossini PM. Neurophysiological evaluation of the central nervous impulse propagation in patients with sensorimotor disturbances. Electroenceph clin Neurophysiol 1988;70: 16±25. Christensen LOD, Andersen JB, Sinkjñr T, Nielsen J. Evidence suggesting that a transcortical re¯ex pathway contributes to stretch re¯ex responses in the tibialis anterior muscle during walking, Abstract for Forum of European Neuroscience Association (ENA) Conference, June/July, 1998 Berlin, Germany. Corna S, Grasso M, Nardone A, Schieppati M. Selective depression of medium latency leg and foot muscle responses to stretch by an alpha 2-agonist in humans. J Physiol (Lond) 1995;484:803±809. Deuschl G, Strahl K, Schenck E, LuÈcking CH. The diagnostic signi®cance of long-latency re¯exes in multiple sclerosis. Electroenceph clin Neurophysiol 1988;70:56±61. Diener HC, Ackermann H, Dichgans J, Guschlbauer B. Medium- and longlatency responses to displacement of the ankle joint in patient with spinal and central lesions. Electroenceph clin Neurophysiol 1985;60: 407±416.
Dietz V. Human neuronal control of automatic functional movements: interaction between central programs and afferent input. Physiol Rev 1992;72(1):33±69. Dietz V. Neurophysiology of gait disorders: present and future applications. Electroenceph clin Neurophysiol 1997;103:333±355. Dietz V, Quintern J, Berger W. Afferent control of human stance and gait: evidence for blocking of group I afferents during gait. Exp Brain Res 1985;61:153±163. Dietz V, Quintern J, Sillem M. Stumbling reactions in man: signi®cance of proprioceptive and pre-programmed mechanisms. J Physiol (Lond) 1987;368:149±163. Edgley SA, Jankowska E, Shefchyk S. Evidence that mid-lumbar neurones in re¯ex pathways from group II afferents are involved in locomotion in the cat. J Physiol 1988;403:57±71. Fellow SJ, DoÈmges F, ToÈpper R, Thilmann AF, Noth J. Changes in the short- and long-latency stretch re¯ex components of the triceps surae muscle during ischaemia in man. J Physiol 1993;472:737±748. Gottlieb GL, Agarwall GC. Response to sudden torques about ankle in man: myotatic re¯ex. J Neurophysiol 1979;42:91±106. Gottlieb GL, Agarwall GC. Response to sudden torques about the ankle in man: II-post-myotatic reactions. J Neurophysiol 1980;43:86±101. Gottlieb GL, Agarwall GC, Jaeger RJ. Response to sudden torques about the ankle in man: V-Effect of peripheral ischemia. J Neurophysiol 1983;50:297±312. Jankowska E. Interneuronal relay in spinal pathways from proprioceptors. Prog Neurobiol 1992;38:335±378. Jones SM, Streletz LJ, Raab VE, Knobler RL, Lublin FD. Lower extremity motor evoked potentials in multiple sclerosis. Arch Neurol 1991;48:944±948. Lundberg A, Malmgren K, Schomburg ED. Comments on re¯ex actions evoked by electrical stimulation of group II muscle afferents. Brain Res 1977;122:551±555. Matthews PBC. The human stretch re¯ex and the motor cortex. Trends Neurosci 1991;14:87±91. Melvill Jones G, Watt DGD. Observations on the control of stepping and hopping movements in man. J Physiol (Lond) 1969;219:709±727. Morita H, Petersen N, Christensen LOD, Sinkjñr T, Nielsen J. Sensitivity of H-re¯exes and stretch re¯exes to presynaptic inhibition in humans. J Neurophysiol 1998;80:610±620. Nardone A, Grasso M, Giordano A, Schiepatti M. Different effect of height on latency of leg and foot short- and medium-latency EMG responses to pertubation of stance in humans. Neurosci Lett 1996;206:89±92. Nielsen J, Sinkjñr T, Baumgarten J, Andersen JB, Toft E, Christensen LOD, Ladouceur M, Morita H. Modulation of the soleus stretch re¯ex and H-re¯ex during human walking after block of peripheral feedback, Abstract for 28th Annual Meeting of Society for Neuroscience, 7th± 12th November 1998 Los Angeles, USA, No. 837.16: 2103. Petersen N, Christensen LOD, Sinkjñr T, Morita H, Nielsen J. Evidence suggesting a transcortical pathway from muscle afferents to tibialis anterior motoneurones in man. J Physiol 1998;512(1):267±276. Poser CM, Paty DW, Scheinberg L, et al. New diagnosis criteria for multiple sclerosis: guidelines for research protocols. Ann Neurol 1983;13(3):227±231. Quintern J, Berger W, Deitz V. Compensatory reactions to gait pertubations in man: short- and long-term effects of neuronal adaptation. Neurosci Lett 1985;62:371±376. Sinkjñr T, Andersen JB, Larsen B. Soleus stretch re¯ex modulation during gait in man. J Neurophysiol 1996;76(2):1112±1120. Sinkjñr T, Andersen JB, Nielsen JF. Impaired stretch re¯ex and joint torque modulation during spastic gait. J Neurol 1996;243:566±574. Sinkjñr T, Nielsen JF, Andersen JB. Soleus long-latency stretch re¯exes during human walking, Neural Control of Movement Meeting, April, 1997 Marriott Casa Magna, Cancun, Mexico (abstract). Stein RB, Yang J, Edamura M, Capaday C. Re¯ex modulation during normal and pathological human locomotion. In: Shimamura M, editor. Neurobiological Basis of Human Locomotion, Tokyo: Japan Scienti®c Societies Press, 1991: 335.
T. Sinkjñr et al. / Clinical Neurophysiology 110 (1999) 951±959 Taylor J, Stein RB, Murphy PR. Impulse rates and sensitivity to stretch of soleus muscle spindle afferent ®bers during locomotion in premamillary cats. J Neurophysiol 1985;53:341±360. Thilmann AF, Schwarz M, ToÈpper R, Fellows SJ, Noth J. Different mechanisms underlie the long-latency stretch re¯ex response of active human muscle at different joints. J Physiol 1991;444:631±643. Toft E, Sinkjñr T, Andreassen S. Mechanical and electromyographic responses to stretch of the human anterior tibial muscle at different contraction levels. Exp Brain Res 1989;74:213±219.
959
Toft E, Sinkjñr T, Andreassen S, Larsen K. Mechanical and electromyographic responses to stretch of the human ankle extensors. J Neurophysiol 1991;65:1402±1410. Toft E, Sinkjñr T, Andreassen S, Hansen HJ. Stretch responses to ankle rotation in multiple sclerosis patients with spasticity. Electroenceph clin Neurophysiol 1993;89:311±318. Yang JF, Stein RB, James KB. Contribution of peripheral afferents to the activation of the soleus muscle during walking in humans. Exp Brain Res 1991;87:679±687.
Soleus long-latency stretch re¯exes during walking in healthy and spastic humans q Thomas Sinkjñr a,*, Jacob Buus Andersen a, Jùrgen Feldbñk Nielsen a, Hans Jacob Hansen b a
Center for Sensory-Motor Interaction, Department of Medical Informatics and Image Analysis, Aalborg University, Fredrik Bajers Vej 7D-3, DK-9220 Aalborg, Denmark b The Hospital for Treatment of Multiple Sclerosis, Ry, Denmark Accepted 15 December 1998
Abstract The present study was carried out to investigate the long-latency soleus stretch re¯exes M2 (peak latency of approximately 85 ms) and M3 (peak latency of approximately 115 ms) during walking in healthy and spastic multiple sclerosis (MS) patients. An 88 stretch was applied to the ankle extensors of the left leg in 8 healthy subjects during normal walking speed and 9 spastic MS patients and 10 age-matched healthy subjects during slow walking. When present in walking healthy subjects, M2 and M3 were modulated in a similar way and with the same amplitudes as previously described for the short latency soleus stretch re¯ex (M1). The spastic patients' soleus M1 was signi®cantly less modulated during walking. The patients' M2 long-latency response was modulated in the same way as the age-matched healthy subjects. All patients' M3 responses were absent or much suppressed during walking. The origin and functional importance of the short- and long-latency stretch re¯exes in healthy and spastic persons are discussed in relation to the above ®ndings and the behaviour of the stretch re¯exes during matched isometric contractions. M3 is argued to be part of a transcortical re¯ex in healthy subjects. q 1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Walking; Short and long-latency stretch re¯exes; Spasticity; Soleus muscle; Motor control
1. Introduction The neural mechanisms that make the activity in the leg muscles adapt to the environmental conditions during walking rely on continuous afferent input to the spinal cord. Such afferent information is believed to be particularly appropriate for correction of unanticipated perturbations. Most such compensatory mechanisms stem from proprioceptive information from the muscle, skin, and joint receptors in the legs (e.g. Dietz, 1997). In the healthy sitting human subject, an imposed dorsi¯exion of the ankle joint causes a series of distinct responses in the electromyogram (EMG) of the stretched ankle extensors (e.g. Gottlieb and Agarwall, 1979). The ®rst response is the short latency stretch re¯ex (labelled `M1' in this study) with an onset latency at 40±50 ms and a peak latency at 50± 60 ms. During an isometric contraction of the ankle extensors, the M1 re¯ex is typically followed by a long-latency q This study was ®nancially supported by The Danish National Research Foundation and The Danish Multiple Sclerosis Society. * Corresponding author. Tel.: 1 45-96-35-88-27; fax: 1 45-98-15-4008. E-mail address: [email protected] (T. Sinkjñr)
1388-2457/99/$ - see front matter q 1999 PII: S 1388-245 7(99)00034-6
Elsevier Science Ireland Ltd.
re¯ex response (M2) with a peak latency at 70±90 (Toft et al., 1991; Fellow et al., 1993). In some subjects, a second long-latency stretch re¯ex (M3) appears (Gottlieb and Agarwall, 1980; Gottlieb et al., 1983). The afferents primarily responsible for the onset of the M1 response are the stretch velocity-sensitive group Iaafferents from muscle spindles (Taylor et al., 1985; Stein et al., 1991; Matthews, 1991). The latency of the M1 response is so brief that its re¯ex arc is certainly spinal and likely monosynaptic. The origin and pathways of the M2 and M3 responses are, however, still a matter of dispute (Berardelli et al., 1982; Thilmann et al., 1991; Dietz, 1992; Fellow et al., 1993; Corna et al., 1995). Several observations support the view that the M2 response in the soleus muscle is mainly mediated by the slow conduct of the spindle group II afferent ®bres through an oligosynaptic spinal pathway (Dietz, 1992; Nardone et al., 1996). Dietz et al. (1985) showed an enhancement of the M2 during walking compared to a tonic contraction during standing when a mechanical perturbation is applied by a sudden acceleration or deceleration of a treadmill. This and the latency led them to suggest that at least the early part of the long-latency responses is a polysynaptic
. All rights reserved.
CLINPH 98656
952
T. Sinkjñr et al. / Clinical Neurophysiology 110 (1999) 951±959
spinal compensatory re¯ex response mediated mainly by muscle proprioceptive input from group II afferents (Dietz et al., 1987). Others have argued that the physiological properties of the long-latency M2 re¯exes are similar to those of the short-latency re¯ex. This makes them represent responses of the motoneurone pool to successive Ia bursts (Berardelli et al., 1982) or responses transmitted over a more complex spinal pathway (Fellow et al., 1993). The origin of the M3 response is also not clear. Gottlieb et al. (1983) showed that the appearance of a late response (seen after 90±100 ms) was reduced, but still present, in the ankle extensors after ischemia had blocked the transmission in the group Ia afferents from the lower leg. Nielsen et al. (1998) recently showed a block of the M3 soleus muscle response after ischemia and during walking. By applying transcranial stimulation, Petersen et al. (1998) demonstrated that in the tibialis anterior, the long-latency M3 response to a stretch applied to the dorsi¯exors is likely to be mediated by a transcortical re¯ex loop through the group Ia afferents. When present, the latency of the M3 response in the soleus muscle to a stretch applied to the ankle extensors occurs at similar latency as the long-latency M3 in tibialis anterior (Sinkjñr et al., 1996b). The functional importance of the long-latency re¯exes has been addressed in several studies (see Dietz, 1992, for a brief review). For subjects jumping from some height or hopping on one foot, the long-latency re¯exes were much more powerful than the spinal M1 re¯ex (Melvill Jones and Watt, 1969). This made the authors refer to the long-latency re¯exes as the `functional stretch re¯ex' which they suggested was of transcortical origin. In this study, the origin and functional importance of the long-latency re¯exes of the soleus muscle were investigated. This was done by comparing the short- and longlatency stretch re¯exes of the soleus muscle during sitting and standing, with the re¯exes during walking at matched excitation level of healthy and spastic multiple sclerosis (MS) patients' soleus muscle. Furthermore, recordings from patients are of importance in assessing the value of long-latency re¯exes in clinical diagnosis (Diener et al., 1985; Deuschl et al., 1988).
2. Materials and methods To describe the M1, M2, and M3 re¯exes, two protocols were used. In protocol 1, healthy subjects' stretch re¯exes were elicited during normal walking speed and at matched excitation levels during standing. To investigate if the longlatency stretch re¯ex M3 was of transcortical origin, we compared re¯exes in healthy subjects and spastic patients with a central lesion in protocol 2. All subjects gave their informed consent before the inclusion in this study. The study was approved by the local
ethical committee in accordance with the ethical standards in the Declaration of Helsinki. 2.1. Experimental set-up In both protocols, a portable stretch device capable of rotating the human ankle joint during walking on a treadmill, elicited the stretch re¯exes. The system consisted of a mechanical joint, mounted in level with the ankle joint that was connected to a powerful actuator system by means of two ¯exible bowden wires. Five different polypropylene plaster casts were made to give a unique interface from the mechanical joint to the ankle of the subject. For further details on the system, see Andersen and Sinkjñr (1995). In the experiment, the mechanical joint was strapped to the subject's left leg. A heel contact was placed in the subject's left shoe, and an insole was placed in the right shoe to match the different height caused by the casting in the left shoe. Bipolar EMG electrodes were placed 2 cm apart; on the soleus muscle about 10 cm above the calcanus, and on the tibialis anterior muscle about 10 cm below the patella and 2 cm from the tibia bone. A ground electrode was placed under the knee. The EMGs were ampli®ed and ®ltered from 20±1000 Hz (DISA, model 15C01). 2.2. Protocol 1 The experiment was performed on 8 healthy subjects (6 males and two females, age 22±32 years). The subjects were told to walk with a natural cadence at 3.5±4 km/h. After an adaptation period of 5±10 min, an EMG pro®le, triggered from heel contact to heel contact, was performed on the soleus and tibialis anterior muscle based on 20 steps. The step cycle was divided into 10 segments. An 88 stretch was made with a velocity of 2808/s randomly between the segments and with an interval of 4±6 steps apart until 8± 10 stretches were accomplished in each time segment. After walking, the subjects were seated in a chair with the left shoe fastened to the ground. The subjects were told to maintain a static level of the soleus EMG that corresponded to the level of walking at approximately 20% of stance phase. In order to maintain the level, the subjects were given a visual feedback of the recti®ed, lowpass ®ltered (1st order 2 Hz) soleus EMG on a monitor. The subjects were told to contract to the static level with an interval of about 4 s until 8±10 trials were averaged. The static contraction experiment was repeated during standing. 2.3. Protocol 2 The protocol included: 9 patients with multiple sclerosis (MS) with a mean age of 43 years (range 27±62) matched with 10 control subjects with a mean age of 38 years (range 25±55). The criteria for inclusion of the patient were: (1) de®nite MS according to criteria by Poser et al. (1983); (2) stable, neurological condition for at least 6 months; (3)
T. Sinkjñr et al. / Clinical Neurophysiology 110 (1999) 951±959
953
suf®cient strength, co-ordination, and range of motion around hip, knee, and ankle joints to walk on a treadmill for 100 m; (4) clinical signs of spasticity in the lower extremities with an increase in muscle tone and exaggerated tendon re¯exes; (5) no contractures or major sensory impairments. The mean duration of MS from the ®rst symptom was 11 years (range 5±25). One patient took antispastic medicine (30 mg baclofen per day). The control subjects were healthy and did not take any medication. In all subjects, the left leg was investigated. The patient recordings were made at their preferred walking speed, which ranged from 0.9±3.1 km/h (mean value: 1.6 km/h). In this protocol, the healthy control subjects walked at a matched walking speed of 1.6 km/h. For each subject the average stride length (^1 SD) was calculated from approximately 30 steps. From a switch placed under the heal of the subject, the stride length was measured as the time difference from one heel contact to the succeeding. The patients walked for 3±5 min without rest. After walking, the patients were seated in a chair with the left shoe fastened to the ground, and the experiments during sitting were repeated as described in protocol 1. 2.4. Data analysis The data were analysed on a computer as follows. The EMG activity of the soleus and tibialis anterior muscles was averaged, recti®ed, and lowpass ®ltered at 20 Hz (1st order). The re¯ex amplitudes of M1, M2, and M3 were measured as the peak amplitude minus an average background activity of 30 ms before the onset of the re¯ex response. The EMGs and ankle position during a perturbation were only averaged if the size of the stretches and velocities of the stretches were within a pre-set range. For all subjects, the stretch amplitude and velocity were kept constant throughout a step. Eight±ten stretch re¯ex responses were averaged together in each segment. 2.5. Statistics The data were tested by the Wilcoxon±Mann±Whitney test for independent samples and by the Wilcoxon signed ranks test for matched samples with a level of signi®cance of P , 0:05. Fig. 1. Example of short- and long-latency stretch re¯exes during walking. (A) Position of the ankle joint during the stance phase. The thin line shows an average of 10 individual steps with an 88 stretch elicited at time zero. The onset of the stretch corresponds to 250 ms after heel contact. The thick line is an average of 8 control steps. (B) Averaged soleus EMG re¯exes to the imposed dorsi¯exion movement shown in (A). The thin line shows the recti®ed and ®ltered soleus EMG with a short-latency re¯ex (labelled M1) followed by two long-latency re¯exes (M2, M3). The soleus EMG activity is superimposed on the EMG activity from a control step (thick line). (C) Position of the ankle joint in control steps and the stretch amplitude. (D) Velocity of the ankle joint and the imposed stretch velocity during the step. (E) Modulation of M1, M2, and M3 soleus stretch re¯exes during the entire step. The stretch is kept constant at a displacement of 88 and a displacement velocity of 2808/s. Subject DM walking at a speed of 3.8 km/h.
3. Results 3.1. Short- and long-latency stretch re¯ex modulation during normal walking speed in healthy subjects During walking, the EMG responses in the soleus muscle to the stretch of the ankle extensors consisted of 1±3 peaks. An example is shown in Fig. 1. Fig. 1A,B illustrates the time course of one healthy subject's ankle position and soleus EMG in early stance during normal walking speed. Each trace shows superimposed average responses with control
954
T. Sinkjñr et al. / Clinical Neurophysiology 110 (1999) 951±959
steps (heavy lines) and steps, in which a stretch was elicited (thin lines). At time zero, the steps of an imposed displacement de¯ected from the position of the control steps (Fig. 1A). The 08 equals the ankle joint position at heel contact. The 88 displacement had a risetime of 35 ms and kept the position nearly ®xed for 200 ms. When the stretch was released after 200 ms, the position returned to the same level as in the control step within 50 ms. Prior to the stretch, the soleus EMG matched the soleus EMG in the control steps until approximately 40 ms after stretch onset (Fig. 1B). At this time, a distinct short-latency stretch re¯ex (M1) was followed by two long-latency EMG components (M2, M3). In this subject, M1 had a peak latency of 50 ms. The peak latencies of M2 and M3 were 85 ms and 112 ms. After the M3 response, the EMG activity returned to the same level as the background EMG of the control step. No responses were seen in the tibialis anterior, except a small peak caused by the crosstalk from the prominent EMG stretch re¯ex of the soleus muscle (not shown). Fig. 1C shows ankle joint position in the control step and the amplitude, and Fig. 1D the velocity of the ankle joint and
the imposed stretch during a full step cycle for the same subject. The stretch was kept constant throughout the gait cycle at a displacement of 88 and a stretch velocity of 2808/s. Fig. 1E illustrates the short- and long-latency soleus stretch re¯exes when the soleus background EMG was subtracted for the same subject. During a step, they were all clearly modulated with high amplitudes in part of the stance phase and near zero values in the swing phase. The M1 response built up in the stance phase was completely inhibited in the transition from stance to swing and again slowly increased during the swing phase as described earlier (Sinkjñr et al., 1996b). The M2 and M3 were only present in two out of the 10 segments in this subject. The re¯exes all peaked in mid- or late-stance phases. None of the stretch re¯exes were different from zero in the transition from stance to swing (Fig. 1E). 3.2. Appearances of M1, M2, and M3 during walking A large inter-subject variation in the appearance of the re¯exes was found in protocol 1. In several subjects, the long-latency re¯exes were either not present, or only present in one or two of the 10 points during a step where the ankle extensors were stretched. Fig. 2A shows the number of persons with an observed M1, M2, and M3 for each of the 10 segments during a step. As 8 people were included, the observation 8 was the highest possible per re¯ex for each segment. For example, at 20% of the step, M1 was seen in all 8 people, M2 in 6 people, and M3 also in 6 people. 3.3. Latencies and amplitudes for M1, M2, and M3 during walking Fig. 2B shows the average peak latencies of M1, M2, and M3 for all 8 healthy subjects during a normal step cycle (protocol 1, see Section 2). The latencies measured from stretch onset to peak EMG did not depend on the time at which they were elicited during the gait cycle. The averaged latency (^1SE) of the M1 response was 56:0 ^ 0:7 ms. The average latency of the M2 response was 84:9 ^ 1:3 ms, and the average latency of the M3 response was 113:9 ^ 3:4 ms. On average, M1, M2, and M3 all modulated during a step with high amplitudes in part of the stance phase and near zero values in the swing phase. Within each segment, no differences were found in the amplitude of the re¯exes (P . 0:05). On average, they all behaved alike during a step (Fig. 2C).
Fig. 2. M1, M2, and M3 stretch re¯ex amplitudes and latencies during a normalized step. (A) Number of M1, M2, and M3 re¯exes observed in each of the 10 segments during a step. If a M1 re¯ex is observed in 6 of the 8 investigated subjects, it will be displayed as a column with the height of 6 in the segment of interest. On top of the `M1 column', the columns for M2 and M3 are shown. (B) Averaged peak latencies of the M1, M2, and M3 stretch re¯exes. (C) Averaged M1, M2, and M3 stretch re¯ex amplitudes. ^ 1 SE, 8 subjects.
3.4. M1, M2, and M3 during isometric contraction and during walking at matched EMG To quantitate how the re¯exes depend on the motor task, we measured the stretch re¯ex at matched contractions (background soleus EMG) during early stance phase of walking, in the standing subject, and during sitting, applying the same stretches in all 3 tasks in 4 subjects from protocol 1 (normal walking speed) and in 8 healthy subjects from
T. Sinkjñr et al. / Clinical Neurophysiology 110 (1999) 951±959
Fig. 3. Amplitude of the M1, M2, and M3 soleus stretch re¯exes at matched background EMG. (A) During early stance phase of normal walking speed (20% of the step), during standing, and during sitting. ^ 1 SE, 4 subjects, protocol 1. (B) Amplitude of the M1, M2, and M3 soleus stretch re¯exes at matched background EMG during early stance phase of slow walking (20% of the step) and during sitting. ^ 1 SE, 8 healthy control subjects, protocol 2.
protocol 2 (slow walking). In the healthy subjects in protocol 2, we only compared sitting and walking. Fig. 3 shows the results. At normal walking speed and at matched soleus background EMG (Fig. 3A), no differences in M1 were found for the 3 tasks, as reported earlier (Sinkjñr et al., 1996b). M2 was non-signi®cantly decreased (P . 0:05) during sitting compared with standing and walking. No M3 was found during sitting and standing. In the healthy control subjects of protocol 2, M2 was non-signi®cantly decreased during sitting compared with walking; 8 subjects. No M3 was found during sitting whereas M3 was seen in the same subjects during the stance phase of walking.
955
tude and stretch velocity were kept constant in all segments and at 88 and 2808/s for all subjects. In the control subjects, the M1 stretch re¯ex response modulated from an average maximum amplitude of 42:6 ^ 10:87 mu;V at 40% of the step declining to 4:0 ^ 2:24 mV at 60% at the end of the stance phase. As described earlier (Sinkjñr et al., 1996b), MS patients do not modulate the M1 as well as the control group. They had a maximum amplitude of 89:0 ^ 32:3 mV at 10% and a minimum of 41:8 ^ 9:23 mV at 40% of the gait cycle. The M1 stretch re¯ex response was on average shifted upwards in the patients with a signi®cantly higher level in 7 out of 9 points during the total step (P , 0:05). The average amplitude of the M1 short-latency soleus stretch is shown in Fig. 4A. No signi®cant differences were found between peak latencies of patients and control subjects M1 (P . 0:05). The averaged M2 stretch re¯ex was modulated equally in patients and the control group, with an increasing response during the stance phase reaching a maximum of 46:8 ^ 8:95 mV for control subjects and 43:5 ^ 28:6 mV for patients at 30% of the step (Fig. 4B). For both groups, the M2 response was declined to 4:1 ^ 3:2 mV for control subjects and 5:2 ^ 5:2 mV for patients at 60% of the step. Except for one point in mid-swing (at 70% of the step in Fig. 4B), the M2 for the two groups was not signi®cantly different (P . 0:05). In addition, no signi®cant differences were found between the peak latencies of patients' and control subjects' M2 stretch re¯ex (P . 0:05). The M3 stretch re¯ex was only seen in 4 observations in the patient group (in 3 patients), whereas 36 observations of M3 were made in the control group (in 5 subjects). In the control group, the M3 had a maximum amplitude of 57:3 ^ 16:0 mV at 30% with a diminishing response of 1:0 ^ 1:0 mV at 50% (Fig. 4C). During the step cycle, the background activity of the soleus and the tibialis anterior was reciprocally activated in the investigated legs of both the control and patient groups (Fig. 4D,E). In the stance phase, patients' soleus background EMG (Fig. 3D) reached a maximum of 49:8 ^ 6:4 mV at 50% of the gait cycle. This was signi®cantly higher than the maximum background EMG of the control group which reached a maximum of 31:1 ^ 5:0 mV at 40%. The average background activity of the tibialis anterior muscle in patients was not signi®cantly different at any point of the gait cycle in relation to the activity of the control group.
3.5. The M1, M2, and M3 in MS patients To get a better understanding of the origin of the longlatency re¯exes, we compared the latencies and amplitudes of the re¯exes during walking in MS patients (protocol 2 in Materials and methods). Fig. 4 illustrates the M1, M2, and M3 stretch-re¯ex modulation during an entire step for 9 MS patients (thin line) and 10 age-matched control subjects (heavy line), at similar walking speed. The stretch ampli-
4. Discussion When present, the 3 stretch re¯exes in the soleus muscle (M1, M2, and M3), were modulated in a similar fashion and with a similar amplitude that will functionally add to the torque at the ankle joint during a step, as earlier described for the short-latency stretch re¯ex (Yang et al., 1991; Sinkjñr et al., 1996a). At matched soleus EMG excitation level,
956
T. Sinkjñr et al. / Clinical Neurophysiology 110 (1999) 951±959
the amplitudes of M1 and M2 re¯exes were of comparable sizes during sitting, standing and walking. M3 is rarely present during isometric contractions during sitting and standing (as observed before by Gottlieb and Agarwall, 1979; Berardelli et al., 1982; Toft et al., 1993), but is seen consistently in 5 out of 8 subjects during normal walking. The M2 long-latency re¯ex in the patients was modulated in the same way as in the age-matched healthy subjects. The M3 stretch re¯ex was absent, or much suppressed in all patients during walking. 4.1. The short-latency M1 soleus stretch re¯ex during walking M1 was present in all subjects and most pronounced in the stance phase (Fig. 2C), as earlier described by Sinkjñr et al. (1996a). The M1 re¯ex was modulated during a step with a high amplitude in the stance phase and a zero amplitude in the transition from stance to swing. The afferents primarily responsible for the onset of the M1 stretch re¯ex are thought to be the stretch velocity sensitive group Ia-afferents from muscle spindles (Taylor et al., 1985; Matthews, 1991; Stein et al., 1991). The structures responsible for the re¯ex modulation are not known, but animal studies (Akazawa et al., 1982) showed that short-latency stretch re¯ex modulation is also present in the premammillary cat during walking. This suggests that at least part of the neural mechanisms responsible for the re¯ex modulation is located in the spinal cord. During walking and sitting at matched excitation level, the similar size of M1 responses in healthy subjects suggested that the short-latency stretch re¯ex was less sensitive to motor task speci®c presynaptic inhibition compared with the H-re¯ex. A different composition and/or temporal dispersion of the afferent volleys evoked by electrical and mechanical stimuli may explain this difference (Morita et al., 1998). 4.2. The long-latency M2 soleus stretch re¯ex during walking
Fig. 4. Averaged re¯ex modulation during an entire step for all MS patients (thin line) and all control subjects (heavy line). (A) The average amplitude of the M1 short-latency soleus re¯ex. (B) The average amplitude of the M2 long-latency soleus stretch re¯ex. (C) The average amplitude of the M3 long-latency soleus stretch re¯ex. The M1, M2, and M3 are calculated as peak EMG activity with the background EMG subtracted. (D) The average background soleus EMG in the control steps. (E) The average background anterior tibialis EMG in the control steps. Mean ^ 1 SE is shown. A q indicates that the two groups are signi®cantly different (P , 0:05)
When the ankle extensors were stretched during the midstance of walking at velocities of nearly 3008/s, a strong EMG response appeared with the latency of the M2 re¯ex (Figs. 1B and 2B). If the M2 response simply represents responses of the motoneurone pool to successive Ia bursts as proposed by, e.g. Berardelli et al., (1982), then the lack of modulation in the spastic patients' M1 (Fig. 4A) is expected also to hold for the M2 response in the same patients. This is not the case (Fig. 4B); M2 in patients modulated in a similar way as M2 in healthy subjects. The onset latency of M2 (approximately 65 ms) was too short for a transcortical loop. This supported the view that the response was mainly of spinal origin. Investigations during walking have shown that the M2 responses are preserved when the group I afferents are blocked by ischemia (Dietz, 1992; Nielsen et al., 1998). Several observations support the view that the M2 responses
T. Sinkjñr et al. / Clinical Neurophysiology 110 (1999) 951±959
in the leg are mainly mediated by muscle proprioceptive input from group II (Corna et al., 1995) (see also Dietz, 1992). These observations were recently supported by investigations in standing man (Nardone et al., 1996). The monosynaptic excitation from group II afferents to motoneurones is considered weak compared with that of group Ia (Lundberg et al., 1977), and is probably contributing little to the M2 responses seen here. The modulation of M2 could take place in the more complex polysynaptic pathways which, at least in the cat, are known to in¯uence the spinal locomotor neuronal network importantly (Edgley et al., 1988). In patients and control subjects, the similar EMG pattern at matched walking speed (Fig. 4D,E) and the patients' normal modulation of the M2 are consistent with the preservation of this part of the neural circuitry in spastic MS. 4.3. The long-latency M3 soleus stretch re¯ex during walking Yang et al. (1991), who investigated the stretch re¯ex in the early stance phase of walking, observed a double EMG soleus burst in two out of 6 subjects. The ®rst burst is consistent with the M1 short-latency stretch re¯ex described here, whereas the second burst peaked at a time (approximately 120 ms after stretch onset) which is consistent with the M3 stretch response observed in this study. In the present study, the M3 response was not seen in sitting and standing subjects (Fig. 3). This is consistent with earlier studies in which we only observed a M3 response in one out of 6 subjects during experiments in sitting subjects (Toft et al., 1991). In contrast to this, the M3 response was seen in 5 out of the 8 subjects during normal walking. A recent study showed that during isometric contractions, M3 in the dorsi¯exors is part of a transcortical re¯ex loop through motor cortex (Petersen et al., 1998), most likely evoked by muscle proprioceptive input from group I. In the present study, the peak latency of M3 (120 ms) was not signi®cantly different from the one found in the dorsi¯exors (Toft et al., 1989; Petersen et al., 1998). If M3 in the soleus muscle is part of a transcortical loop, we suggest that one of the reasons for not observing it in the patients is a large temporal dispersion between the contributing motor unit action potentials as the nerve conduction velocity decreases in MS patients (Caramia et al., 1988; Jones et al., 1991). This results in a pronounced phase cancellation between the positive phase and the negative phase of the different motor unit potentials. This is evident from Fig. 1 in Deuschl et al. (1988), in which the response to a very synchronized electrical stimuli of the medial or radial super®cial nerve is averaged more than 100 times, and still the delayed re¯ex response in MS patients' thenar muscle is not impressive. A stretch mediated input, as in the present study, will deliver a much less synchronized input to the CNS, which has to travel a much longer path in the leg compared with the
957
arm. Both issues will make it even more unlikely to detect a delayed M3 response in MS patients' leg. In another patient group with a central motor lesion after apoplexia cerebri, we have earlier observed that M3 is either absent or only seen as single points in the step cycle (Sinkjñr et al., 1997). This supports that M3 is part of a transcortical re¯ex loop. At present, however, the lack of M3 in the patients cannot be excluded to be due to an inhibition or facilitation loss of a spinal re¯ex due to an impaired supraspinal control. Further investigations (for example, by applying magnetic stimulation above the motor cortex timed to the expected arrival of the afferents volley from the receptors in the stretched ankle extensor muscles) might be able to clarify if M3 in the soleus muscles is part of a transcortical re¯ex loop. In healthy humans, the M3 response of soleus was interestingly not found during sitting and standing. If M3 is transcortical, this implied that the brain was less involved in the control of the ankle extensors to an unexpected event during such motor tasks than it was during the automatic task of walking. Contrary to this, the M3 responses in the preactivated tibialis anterior muscle are seen consistently during sitting (Toft et al., 1989) and walking (Christensen et al., 1998). Whether this implies that a transcortical re¯ex pathway plays a more important and integrated part of the control of dorsi¯exors than a transcortical re¯ex pathway does for the ankle extensors, remains to be investigated. The M2 response has been demonstrated to be likely transcortical in the upper limb, whereas the data from the leg suggest that M3, but not the M2 response, is transcortical. This difference between the upper and lower limb is due to misleading labelling, since what is called an M2 response in the upper limb corresponds in terms of the underlying mechanism to what is called an M3 response in the lower limb (Petersen et al., 1998). 4.4. Functional considerations The short-latency feedback mechanism makes the stretch re¯ex adjust the ankle joint impedance in accordance with the external constraints in the fastest possible way. If the environmental context deviates from the average natural perturbations received by the ankle during walking (e.g. if the ankle extensors are being stretched above the averaged eccentric contractions due to an incline in the surface during the stance phase), a stronger afferent input from the ankle extensor muscles is proposed to compensate for this through M1 as well as the long-latency stretch re¯exes M2 and M3. They all add to the muscle activity to stiffen up the ankle joint. Such compensation happens most likely in every step cycle with more or less distinct muscle responses depending on the environmental context. The long-latency stretch re¯exes (M2, M3) are only present when the muscle is active, and are only observed in mid-stance during walking. They were not found in all subjects in this study. Often a strong and complex perturba-
958
T. Sinkjñr et al. / Clinical Neurophysiology 110 (1999) 951±959
tion is required to elicit them (Berardelli et al., 1982; Quintern et al., 1985). When elicited, these long-latency responses are of similar size or larger than the M1 response (see also Dietz et al., 1987). With regard to standing, fast and co-ordinated postural adjustments of the lower limb to a perturbation are assumed to rely mainly on a spinal re¯ex circuit as the receiving group I muscle input (to elicit the short latency M1 re¯ex) and group II muscle input (to elicit the M2 long-latency re¯ex). As the polysynaptic group II pathways are part of a circuit with convergence from various peripheral receptors and supraspinal centres (Jankowska, 1992), ample possibilities for a functional modulation are present. The lack of M3 in the soleus muscle during sitting and standing suggests that its importance is linked to more dynamic motor tasks such as walking. The response is assumed to be part of a supraspinal, mediated compensatory re¯ex response, which in¯uences the stereotyped generated motor output when unexpected strong afferent input is detected at the supraspinal level. It is not known at present whether the M3 is more pronounced when cortical control becomes dominant with respect to the spinal cord, as might happen, for example, when a new and not yet trained walking pattern is required or in skilled locomotion. The lack of the long-latency M3 re¯ex in spastic gait might worsen the spastic patients' ability to compensate for large, unexpected perturbations impairing their overall gait performance, as it might impair their ability to do skilled walking if the M3 is important in such tasks. References Andersen JB, Sinkjñr T. An actuator system for investigating electrophysiological and biomechanical features around the human ankle joint during gait. Trans Rehab Eng 1995;3(4):299±306. Akazawa K, Aldridge JW, Steeves JD, Stein RB. Modulation of stretch re¯exes during locomotion in the mesencephalic cat. J Physiol 1982;329:553±567. Berardelli A, Hallett M, Kaugman C, Fome E, Berenberg W, Simon SR. Stretch re¯exes of triceps surae in normal man. J Neurol Neurosurg Psychiatry 1982;45:513±525. Caramia MD, Bernardi G, Zarola F, Rossini PM. Neurophysiological evaluation of the central nervous impulse propagation in patients with sensorimotor disturbances. Electroenceph clin Neurophysiol 1988;70: 16±25. Christensen LOD, Andersen JB, Sinkjñr T, Nielsen J. Evidence suggesting that a transcortical re¯ex pathway contributes to stretch re¯ex responses in the tibialis anterior muscle during walking, Abstract for Forum of European Neuroscience Association (ENA) Conference, June/July, 1998 Berlin, Germany. Corna S, Grasso M, Nardone A, Schieppati M. Selective depression of medium latency leg and foot muscle responses to stretch by an alpha 2-agonist in humans. J Physiol (Lond) 1995;484:803±809. Deuschl G, Strahl K, Schenck E, LuÈcking CH. The diagnostic signi®cance of long-latency re¯exes in multiple sclerosis. Electroenceph clin Neurophysiol 1988;70:56±61. Diener HC, Ackermann H, Dichgans J, Guschlbauer B. Medium- and longlatency responses to displacement of the ankle joint in patient with spinal and central lesions. Electroenceph clin Neurophysiol 1985;60: 407±416.
Dietz V. Human neuronal control of automatic functional movements: interaction between central programs and afferent input. Physiol Rev 1992;72(1):33±69. Dietz V. Neurophysiology of gait disorders: present and future applications. Electroenceph clin Neurophysiol 1997;103:333±355. Dietz V, Quintern J, Berger W. Afferent control of human stance and gait: evidence for blocking of group I afferents during gait. Exp Brain Res 1985;61:153±163. Dietz V, Quintern J, Sillem M. Stumbling reactions in man: signi®cance of proprioceptive and pre-programmed mechanisms. J Physiol (Lond) 1987;368:149±163. Edgley SA, Jankowska E, Shefchyk S. Evidence that mid-lumbar neurones in re¯ex pathways from group II afferents are involved in locomotion in the cat. J Physiol 1988;403:57±71. Fellow SJ, DoÈmges F, ToÈpper R, Thilmann AF, Noth J. Changes in the short- and long-latency stretch re¯ex components of the triceps surae muscle during ischaemia in man. J Physiol 1993;472:737±748. Gottlieb GL, Agarwall GC. Response to sudden torques about ankle in man: myotatic re¯ex. J Neurophysiol 1979;42:91±106. Gottlieb GL, Agarwall GC. Response to sudden torques about the ankle in man: II-post-myotatic reactions. J Neurophysiol 1980;43:86±101. Gottlieb GL, Agarwall GC, Jaeger RJ. Response to sudden torques about the ankle in man: V-Effect of peripheral ischemia. J Neurophysiol 1983;50:297±312. Jankowska E. Interneuronal relay in spinal pathways from proprioceptors. Prog Neurobiol 1992;38:335±378. Jones SM, Streletz LJ, Raab VE, Knobler RL, Lublin FD. Lower extremity motor evoked potentials in multiple sclerosis. Arch Neurol 1991;48:944±948. Lundberg A, Malmgren K, Schomburg ED. Comments on re¯ex actions evoked by electrical stimulation of group II muscle afferents. Brain Res 1977;122:551±555. Matthews PBC. The human stretch re¯ex and the motor cortex. Trends Neurosci 1991;14:87±91. Melvill Jones G, Watt DGD. Observations on the control of stepping and hopping movements in man. J Physiol (Lond) 1969;219:709±727. Morita H, Petersen N, Christensen LOD, Sinkjñr T, Nielsen J. Sensitivity of H-re¯exes and stretch re¯exes to presynaptic inhibition in humans. J Neurophysiol 1998;80:610±620. Nardone A, Grasso M, Giordano A, Schiepatti M. Different effect of height on latency of leg and foot short- and medium-latency EMG responses to pertubation of stance in humans. Neurosci Lett 1996;206:89±92. Nielsen J, Sinkjñr T, Baumgarten J, Andersen JB, Toft E, Christensen LOD, Ladouceur M, Morita H. Modulation of the soleus stretch re¯ex and H-re¯ex during human walking after block of peripheral feedback, Abstract for 28th Annual Meeting of Society for Neuroscience, 7th± 12th November 1998 Los Angeles, USA, No. 837.16: 2103. Petersen N, Christensen LOD, Sinkjñr T, Morita H, Nielsen J. Evidence suggesting a transcortical pathway from muscle afferents to tibialis anterior motoneurones in man. J Physiol 1998;512(1):267±276. Poser CM, Paty DW, Scheinberg L, et al. New diagnosis criteria for multiple sclerosis: guidelines for research protocols. Ann Neurol 1983;13(3):227±231. Quintern J, Berger W, Deitz V. Compensatory reactions to gait pertubations in man: short- and long-term effects of neuronal adaptation. Neurosci Lett 1985;62:371±376. Sinkjñr T, Andersen JB, Larsen B. Soleus stretch re¯ex modulation during gait in man. J Neurophysiol 1996;76(2):1112±1120. Sinkjñr T, Andersen JB, Nielsen JF. Impaired stretch re¯ex and joint torque modulation during spastic gait. J Neurol 1996;243:566±574. Sinkjñr T, Nielsen JF, Andersen JB. Soleus long-latency stretch re¯exes during human walking, Neural Control of Movement Meeting, April, 1997 Marriott Casa Magna, Cancun, Mexico (abstract). Stein RB, Yang J, Edamura M, Capaday C. Re¯ex modulation during normal and pathological human locomotion. In: Shimamura M, editor. Neurobiological Basis of Human Locomotion, Tokyo: Japan Scienti®c Societies Press, 1991: 335.
T. Sinkjñr et al. / Clinical Neurophysiology 110 (1999) 951±959 Taylor J, Stein RB, Murphy PR. Impulse rates and sensitivity to stretch of soleus muscle spindle afferent ®bers during locomotion in premamillary cats. J Neurophysiol 1985;53:341±360. Thilmann AF, Schwarz M, ToÈpper R, Fellows SJ, Noth J. Different mechanisms underlie the long-latency stretch re¯ex response of active human muscle at different joints. J Physiol 1991;444:631±643. Toft E, Sinkjñr T, Andreassen S. Mechanical and electromyographic responses to stretch of the human anterior tibial muscle at different contraction levels. Exp Brain Res 1989;74:213±219.
959
Toft E, Sinkjñr T, Andreassen S, Larsen K. Mechanical and electromyographic responses to stretch of the human ankle extensors. J Neurophysiol 1991;65:1402±1410. Toft E, Sinkjñr T, Andreassen S, Hansen HJ. Stretch responses to ankle rotation in multiple sclerosis patients with spasticity. Electroenceph clin Neurophysiol 1993;89:311±318. Yang JF, Stein RB, James KB. Contribution of peripheral afferents to the activation of the soleus muscle during walking in humans. Exp Brain Res 1991;87:679±687.