Recovery of hypermetria after a cerebellar stroke occurs as a multistage process

Recovery of Hypermetria after a Cerebellar Stroke Occurs as a Multistage Process M. Manto, MD,# J. Jacquy, MD,I J. Hildebrand, MD,’ and E. Godaux, MDS

In a prospective study, we repeatedly recorded fast goal-directed wrist movements of 8 patients who had experienced an acute cerebellar hypermetria due to a stroke and who had subsequently recovered clinically. Movements and the associated agonist and antagonist electromyographic (EMG) activities were recorded before and after addition of inertial loads. Four stages characterized the recovery process. At stage 1, hypermetria was present in the basal state and was not modified by the addition of inertial loads. At stage 2, hypermetria, which was present in the basal state, was enlarged by mass addition. At stage 3, hypermetria was absent in the basal state, but was revealed by an inertial load increase. At stage 4, as in healthy subjects, there was no hypermetria without or with addition of inertial loads. At stage 1, the patients presented several defects. (1) Facing an increased inertia, they could not increase their agonist EMG activity. (2) The onset latency of their antagonist EMG activity was delayed. (3) Facing an increased inertia, they could not increase their antagonist EMG activity. Among these three defects, the first disappeared at stage 2, the second at stage 3, and the third at stage 4. Manto M, Jacquy J, Hildebrand J, Godaux E. Recovery of hypermetria after a cerebellar stroke occurs as a multistage process. Ann Neurol 1995;38:437-445

Cerebellar hypermetria, a classic sign of a lesion in the lateral cerebellum, consists of a movement that overshoots the target when a patient attempts to make a fast and accurate movement [l-41. In healthy subjects, such movements are launched by a burst of electromyographic (EMG) activity in the agonist muscle and are braked and stopped by a burst of EMG activity in the antagonist muscle [5-81. The most prominent EMG abnormality associated with hypermetria is a delayed onset of the antagonist activity [9-131. In two of our previous studies, wrist flexion movements in the basal state and after changes of the inertial load were compared in healthy subjects and in cerebellar patients. We made two observations. The first one concerned a group of patients presenting a cerebellar hypermetria in the basal state. We found that hypermetria enlarged when the inertial load of the moving hand was artificially increased [141. While a healthy subject adapts to increasing inertia by increasing both the agonist activity (the launching force) and the antagonist activity (the braking force), a patient with cerebellar disease is able to increase the agonist activity but not the antagonist activity. The second observation was performed in a group of patients with normal clinical findings in spite of a cerebellar lesion {lS]. We found that in these patients, extra loads induced hypermetria. This load-induced hypermetria was also found to be

due to an inability to adapt the antagonist activity in an appropriate way. In an attempt to fit these two observations into a general pattern, we hypothesized that they might actually correspond to two stages of the same recovery process. To test this hypothesis, a prospective study was undertaken. We followed patients who exhibited hypermetria after a cerebellar stroke. We repeatedly recorded their fast and accurate movements and the associated EMG activities in the basal state and when the inertial load was artificially increased. This study not only confirmed our hypothesis but revealed two other stages, one occurring just after the acute cerebellar injury and the other at the end of the recovery process.

From the “Department of Neurology, HBpital Erasme, Bruxelles; tDeparment of Neurology, HBpital Civil de Charleroi, Charleroi; and the Of Neurophysiolo~, of Medicine, University of Mons, Mons, Belgium.

Received Apr 13, 1995, and in revised form Jun 14. Accepted for publication Jun 14, 1995.

Materials and Methods The methodology was, for its major part, the same as that used in our previous studies 114, 151. In brief, during each recording session, the control subject or the patient was asked to make fast and accurate goal-directed wrist movements with the fingers extended. In response to a “go”signal, a horizontal wrist flexion had to be made toward an aimed target located either 15 or 50 degrees away from the start position (see Fig 1 in reference 1141).Movements were made before and after alteration of the mechanical state of the hand by fixing masses of 200 or 500 gm at the level of the metacarpophalangeal joints. In each of the three mechanical

Address correspondence to Dr Godaux,University of Mans, place du Parc, 20, 7000 Mons, Belgium.

Copyright 0 1995 by the American Neurological Association

437

states, subjects performed some practice trials before data were collected. The effect of the 200-gm load was always investigated before that of the 500-gm load.

Patients Recording sessions were carried out in 8 patients (2 women, 6 men) who had a cerebellar stroke documented by magnetic resonance imaging (MRI). All patients, who were in a hospital and were examined clinically every day, presented cerebellar signs in the acute stage, and recovered clinically thereafter. Movement recordings performed on these patients were compared to those obtained from 9 healthy subjects (3 women, 6 men). The clinical data of the patient group are summarized in Table 1. The mean age of the control group was 55 years, with a range from 36 to 79 years. The mean age of the patients of this study was 60 years, with a range from 38 to 71 years. When lesions were bilateral, only one side was investigated. When the lesion was unilateral, we only considered the limb on the side of the lesion. The patient’s informed consent was obtained following full explanation of the experimental procedures.

Motion Recording Computerized motion analysis in three dimensions was made with a Selspot 11 system (Selcom). One infrared light-emitting diode (LED) was attached to the forefinger. Two cameras, each fitted with a photodetector unit, recorded the twodimensional positioning of the LED. Sampling rate was 300 times per second and the resolution was 0.25 mm. Threedimensional positioning was computed using the two-axis camera recordings.

Electromyographic Activity: Recording, Averaging, and Calibration For each recording session, the activity of the flexor carpi radialis (agonist muscle) and of the extensor carpi radialis (antagonist muscle) was recorded by surface electrodes. EMG signals were differentially amplified, filtered (2,000 x , 30-8,000 Hz), and full-wave rectified (Pathfinder I, Nicolet Instrument, Madison, WI). During each session, 12 observations were made on both agonist and antagonist muscles of each subject in six different experimental protocols (amplitude = 15 or 50 degrees; additional mass = 0, 200, or 500 gm). These 12 observations were then averaged after aligning them to the moment when the finger crossed a light beam received by a photoelectric cell )located 4 o r 7 degrees away

from the initial position for the 15- or the 50-degree amplitude movements, respectively. EMG activities were quantified according to the method described by Gottlieb and coauthors C161. The rectified envelope of the agonist EMG activity was integrated over the interval from the onset to the first zero-crossing of the acceleration (integrated agonist activity over the acceleration phase of movement). The antagonist EMG activity was integrated from the onset to the second zero-crossing of the acceleration (integrated antagonist activity over the deceleration phase of movement). To measure the changes of EMG activities as a function of the added loads, we considered, for each subject, the integrated EMG activities recorded with no load as a reference value (100% value) and expressed the integrated EMG activities recorded with extra loads (200 or 500 gm) in percentages of these basal activities. Since we also wanted to compare the EMG activities recorded during successive sessions in the patient group, calibration became a critical factor. Indeed, the integrated EMG activities developed by a patient may vary as a function of electrode position. The activity of each muscle under examination was calibrated by asking the patient to develop an isometric contraction against a load that was intended to stretch the muscle. Calibrations for both agonist and antagonist activities were obtained by loading the wrist with 200 gm using a pulley. For both the agonist and the antagonist muscle, 15 rectified EMG activities developed during a 1second period of isometric contraction were averaged. The 1-second duration area extending from the zero-potential line to the trace of the EMG activity defined the calibration activity (measured in microvolts . second). For each muscle activity, the ratio of the corresponding integrated EMG activity (expressed in microvolts . second) and of the calibration activity (measured in microvolts . second) of this muscle was computed. These ratios were expressed in arbitrary units (a.u.). To check the reliability of this procedure, we computed the ratios described above on 3 successive days in 6 control subjects. The variation from day to day never exceeded 3% either for the agonist or for the antagonist muscle.

Statistical A nabs is In the basal state, the movement amplitudes and the onset of latencies of antagonist activity were compared in both groups (patients vs healthy subjects) by Student’s t test. The normality of the distribution of data was tested using Shapiro-Wilks

Table 1 . Clinical Data for Patients with Cerebellar Disease Patient No.

438 Annals of Neurology

Age (yr)

Sex

Diagnosis

59

M F F

Occlusion of the right superior cerebellar artery Left cerebellar ischemia Left cerebellar hemorrhage Left cerebellar hemorrhage h g h t cerebellar hemorrhage Bilateral cerebellar ischemia Left cerebellar ischemia Left cerebellar ischemia

61 66 38

M

61 68 71 57

M M M M

Vol 38 No 3 September 1995

test ( p > 0.05). When the inertia was artificially modified, the movement amplitudes and the onset latencies of the antagonist activity were compared by repeated-measureanalysis of variance (load effect). Friedman’s test was used to evaluate the statistical significance of modifications of the integrated EMG activities compared with those observed in the basal state. The EMG activities recorded in patients when they made movements with an extra mass (200 or 500 gm) were compared to those recorded in healthy subjects in the same circumstances by repeated-measure analysis of variance. Modifications of magnitude of the calibrated EMG activities as a function of the stages were assessed by repeated-measure analysis of variance (sequence effect). The probability level for significance was set at 0.05. Recordings of the movement amplitudes in the basal state were performed o n each of our healthy subjects during 3 successive days to check the variation from day to day. The repeated-measure analysis of variance showed that the movement amplitudes did not change significantly from day to day (sequence effect; significance of F: p = 0.574). Data used for intergroup comparison were those obtained during the first recording session. Among the data collected on different days during a same stage of recovery from a cerebellar stroke, only one data set was selected for statistical analysis. At stage 1, data from the day when hypermetria was the largest were selected. At stages 2 and 3, data from the day when the load-induced change of amplitude was the largest were chosen. At stage 4, data from the day when movement amplitudes were the closest to the aimed amplitude were selected.

Results Healthy Subjects Figure 1 illustrates the typical features of wrist flexion movements toward an aimed target located at 15 degrees made by a normal subject in the basal state and after addition of 500 gm to the moving hand. In the basal state, the movements were accurate without any detectable overshoot. The mean amplitude was 15.6 degrees. A first burst of EMG activity in the agonist muscle (which launches the movement) was followed by a burst of EMG activity in the antagonist muscle (which stops the movement). The onset latency of the antagonist activity (with respect to the first burst of EMG activity in the agonist muscle) was 37 msec. When an extra load of 500 gm was affixed to the hand, the index did not overshoot the aimed target. The mean movement amplitude was 15.3 degrees. The onset latency of the antagonist activity remained similar (45 msec) to that observed in the basal state (37 msec). The subject adapted to increasing inertia by increasing both the agonist activity from 100% (basal state) to 177% ( + 5 0 0 gm) and the antagonist activity from 100% (basal state) to 188% ( + 500 gm). Tables 2 and 3 list the kinematic and EMG parameters of the movements performed by the group of healthy subjects in the six experimental conditions (15 or 50 degrees; no overload or 200 gm or 500 gm).

HEALTHY SUBJECT

B

A 0 gm

2o

0

1

9

8

,

37 msec

0

-

100 msec

c+

500gm

1-

‘

I

45 msec

u 100 msec

Fig 1. Kinematic and EMG fratures of fast and accurate wrist flexion movements made by a healthy subject. The movements were perfomzed in two mechanical states of the moving hand: (A) no overload and (B) addition of an extra load of SO0 gm. Each top panel corresponds to the superimposition of the individual records of position j i r 12 flexion movements (target distance: 15 degrees). The middle and bottom panels correspond to the averages of the full-rectified EMG activities associated with these 12 movements. EMG AGO and EMG ANTA =agonist EMG activities (jexor carpi radialis) and antagonist EMG activities (extensor carpi radialis), respectively. SurJaces of full-rect$ed EM G activities are calibrated in arbitraty units (a.u.).

These values were used as a reference for subsequent comparisons. Movement amplitudes did not change significantly with increasing inertia, as attested by the repeated-measure analysis of variance (load effect; significance of F: p = 0.154). The onset latencies of antagonist activity were also unaffected by addition of extra loads, as confirmed by the repeated-measure analysis of variance (load effect; significance of F: p = 0.7 15). The integrated EMG activities of both agonist and antagonist muscles increased when an extra load (either 200 or 500 gm) was added, as demonstrated by Friedman’s test ( p < 0.001, whether the aimed amplitude was 15 or 50 degrees).

Recovery in the Patient Group O n the basis of the presence or absence of hypermetria in the basal state and of the effect of load addition

Manto et al: Recovery from Cerebellar Injury 439

Table 2. Movement Amplitudes a,vd Onset Latencies of Antagoni.rt EMG Activities in Healthy Subjects and in Patients During the First, the Second, and the Third Stags Inertial Loadb

Movement amplitudes' (aimed a.mplitude = 15 degrees)

Onset latencies of antagonist EMG activityd (aimed amplitude = 15 degrees) Movement amplitudes' (aimed amplitude = 50 degrees)

Onset latencies of antagonist EMG activityd (aimed amplitude = 50 degrees)

Groupa

No Load

Healthy subjects Patients, stage 1 Patients, stage 2 Patients, stage 3 Healthy subjects Patients, stage 1 Patients, stage 2 Patients, stage 3 Healthy subjects Patients, stage 1 Patients, stage 2 Patients, stage 3 Healthy subjects Patients, stage 1 Patients, stage 2 Patients, stage 3

15.6 23.1 23.5 15.7 41.0 98.0 98.0 38.0 51.5 56.9 57.0 51.7 44.0 97.0 99.6 36.0

t 0.9 t 2.6

2.5 -r- 0.6 ? 11.0 t 24.0 t 25.0 t 14.0 t 1.3 t 2.5 t 2.0 t 0.8 t 13.0 t 24.3 ? 24.6 ? 13.6 ?

+ 200 gm 16.1 23.2 26.6 17.9 40.0 97.0 97.0 36.0 50.8 57.4 60.1 54.5 44.0 100.0 100.0 40.8

i

t t t t t _"

rf-

0.7 2.6 2.9 0.9 11.0 23.0 21.0 9.0 1.2

t 2.7 t 2.4 t 0.8 ?

10.0

t 23.7 t 23.4 t 10.8

t 500 gm

15.6 21.0 29.1 19.8 41.0 99.5 99.0 34.0 51.2 57.1

62.8 56.5 44.0 97.8 100.0 38.3

2

t t t t t 2

t t t t t t t t 2

1.0 2.4 3.2 0.8 8.0 25.0 22.0 13.0 1.0 3.0 2.9 1.3 9.0 24.2 24.0 9.6

'The number of normal subjects included in this table is 9 and the number of patients included is 8. bValues are mean 2 standard deviation. 'Values are expressed in degrees. 'Values are expressed in milliseconds and are measured from the onset of agonist EMG activity.

Table 3 . Integrated Agonist and Antagonat EMG Activities in Healthy Subjects and in Patients During the First, the Second, and the Third StageAInertial Load"

Integrated agonist EMG activities' (aimed amplitude = 15 degrees)

Integrated antagonist EMG activities' (aimed amplitude = 15 degrees)

Integrated agonist EMG activitiid (aimed amplitude = 50 degrees)

Integrated antagonist EMG activities' (aimed amplitude = 50 degrees)

Groupa

No Load

+ 200 gm

+ 500 gm

Healthy subjects Patients, stage 1 Patients, stage 2 Patients, stage 3 Healthy subjects Patients, stage 1 Patients, stage 2 Patients, stage 3 Healthy subjects Patients, stage 1 Patients, stage 2 Patients, stage 3 Healthy subjects Patients, stage 1 Patients, stage 2 Patients, stage 3

100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100

118.4 t 9.8 105.1 t 1.8 112.3 t 5.4 119.6 t 8.3 121.3 -t 10.4 105.8 t 1.8 104.3 t 1.5 108.6 t 3.0 122.3 t 8.1 105.5 t 1.2 114.9 t 4.5 121.3 t 7.7 128.7 ? 10.5 104.5 t 1.9 106.3 t 2.0 109.5 t 3.6

126.6 t 20.9 106.4 ? 2.0 132.3 t 17.3 133.0 _" 20.0 133.9 t 22.5 106.2 t 2.3 108.0 t 5.8 111.9 t 3.6 127.9 t 15.5 106.8 t 2.2 138.9 t 14.4 132.1 i 16.9 135.7 t 13.2 105.6 -t 3.1 113.3 2 9.9 114.8 -t 6.1

~

'The number of normal sublects induded in thls table IS 9 and the number of patients included IS 8 bValues are mean k standard deviatton (except for EMG activities when no load was added) 'Values are expressed In percentagc. of the correspondlng basal EMG acrivit~es(when no load was added)

440 Annals of Neurology Vol 38 No 3 September 1995

on movement amplitudes, four stages emerged in the recovery process. Movement was considered as hypermetric when it exceeded 17.4 degrees (mean + 2 SDs of the control values) for an aimed movement of 15 degrees and 54.1 degrees (mean 2 SDs of the control values) for an aimed movement of 50 degrees. The presence or absence of a load-induced change of movement amplitude was assessed by repeatedmeasure analysis of variance performed on the 12 sets of observations made on a same patient in the three inertial conditions. For each stage, a report of observations made on a representative patient (Patient 1) will be followed by a statistical analysis performed on the patient group. During a follow-up of 50 days, 15 recording sessions were carried out in this representative patient. The features of the first stage are illustrated in Figure 2 for Patient 1. This stage was characterized by a hypermetria in the basal state which was unaffected by addition of masses. The illustrated recordings were performed on the second day. In the basal state and for an aimed target located at 15 degrees, the mean amplitude of movement was 27.4 degrees. When a mass of 500 gm was added, the mean amplitude of movement was 26.6 degrees. The patient was not able to increase appropriately either his agonist or his antagonist activity. After addition of 500 gm, the integrated agonist EMG activity amounted to only 106% of that observed in the basal state and the integrated antagonist activity amounted to only 108% of that observed in the basal condition. The onset latency of the antagonist activity was delayed in the basal condition (98 msec) and when an extra weight of 500 gm was added (97 msec). Clinically, the patient complained of weakness on the right side and a marked cerebellar syndrome was observed on the right side. The movement amplitudes of the patient group were larger than the movement amplitudes of the control group, as confirmed by Student’s test ( p < 0.001 whether the aimed amplitude was 15 or 50 degrees). In the basal state, the onset latencies of the antagonist activity were also larger in the patient group with respect to the control group (Student’s test, p < 0.001, whether the aimed amplitude was 15 or 50 degrees). Hypermetria was not modified by addition of loads, as demonstrated by repeated-measure analysis of variance (load effect; significance of F: p = 0.235). The onset latency of the antagonist activity was unaffected by loads, as demonstrated by repeated-measure analysis of variance (load effect; significance of F: p = 0.393). The integrated agonist EMG activities increased slightly in the patient group after addition of inertia, as shown by Friedman’s test ( p = 0.011 when aimed amplitude was 15 degrees and p = 0.002 when aimed amplitude was 50 degrees). However, the repeatedmeasure analysis of variance showed that the increase

CEREBELLAR SYNDROME - STAGE 1

A

B

+

0

IMVT/ = 20 a.u.

= 20

E

A

98 msec

-

100 msec

a.u.

E A 97 msec

-

100 msec

Fig 2. Kinematic and EMG features of fast and accurate uwst flexion movements made by a cerebellar patient (Patient 1 I when he was in the first stage of recozjery. Movements, EMG agonist (AGO) activity, and EMG antagonist (ANTA) actioity for flexions made uithout any extra load (AJand after addition of an extra load of 500 gm (Bi. See Figure 1 for details.

of the agonist EMG activity following loading was substantially smaller in the patient group than in the control group (group effect; significance of F: p = 0.001). The integrated antagonist EMG activities increased slightly in the patient group after addition of masses (Friedman’s test, p = 0.002 whether the aimed amplitude was 15 or 50 degrees). However, the repeatedmeasure analysis of variance showed that the increase of the antagonist EMG activity after loading was substantially smaller in the patient group than in the control group (group effect; significance of F: p < 0.001). The second stage, illustrated in Figure 3 for Patient 1, was characterized by a hypermetria in the basal state that increased with loads. The illustrated recordings were performed on the ninth day. For an aimed amplitude of 15 degrees, the mean movement amplitude was 27.1 degrees in the basal state, and 33.3 degrees with an extra mass of 500 gm. The patient was able to increase sufficiently his agonist activity but not his antagonist activity. After addition of 500 gm, the integrated agonist EMG activity was 162% of that observed in the basal state while the integrated antagonist

Manto et al: Recovery from Cerebellar Injury

441

CEREBELLAR SYNIDROME - STAGE 2

A

€3

30

0

-

1

9

0

1

9

EMG AGO

EMG ANTA

- I

’

101 msec

100 msec

- I

.

99 msec

100 mSec

Fig 3. Kinematic and EMG fiatures of fast and accurute wrist flexion movements made by a cerebel‘lar patient (Patient 1 ) when he was in the second stage of recovery. Movement, EMG agonist (AGO) activity, and EMG a,vtagonist (ANTA) activity forjexions made without any extnz load (A)and after addition of an extra load of 500 gm (B). Se,?Figure 1 for details.

activity amounted to only 121% of that observed in the basal condition. The onset latency of the antagonist activity was delayed in both the basal condition (101 msec) and after addition of an extra weight of 500 gm (99 msec). Clinically, the patient did not complain anymore of weakness but complained of clumsiness. Neurological examination sti 11 revealed a cerebellar syndrome o n the right side. At this stage, all the patients exhibited a hypermetria in the basal state. The movement amplitudes were significantly larger in the patient group than in the control group, as demonstrated by the Student’s test ( p < 0.001, whether the aimed amplitude was 15 or 50 degrees). In the basal state, the onset latencies of the antagonist activity were also larger in the patient group with respect to the control group (Student’s test, p < 0.001, whether the aimed amplitude was 15 or 50 degrees). By contrast with the first stage, hypermetria increased significantly with extra weights, as shown by the repeated-measure analysis of variance (load effect; significance of F: p < 0.001). The onset latency of

the antagonist activity was not modified by loads, as confirmed by the repeated-measure analysis of variance (load effect; significance of F: p = 0.980). The integrated agonist EMG activities markedly increased with addition of loads (Friedman’s test, p = 0.003 when aimed amplitude was 15 degrees and p < 0.001 when aimed amplitude was 50 degrees). The increase of the agonist EMG activity following loading was similar in both groups (healthy subjects vs patients), as attested by the repeated-measure analysis of variance (group effect; significance of F: p = 0.895). The integrated antagonist EMG activities increased slightly in the patient group with extra loads. This increase was significant, as shown by Friedman’s test ( p < 0.001 whether the aimed amplitude was 15 or 50 degrees). However, the increase of the antagonist EMG activity after loading was smaller in the patient group than in the control group, as attested by the repeated-measure analysis of variance (group effect; significance of F: p < 0.001). The third stage, illustrated in Figure 4 for Patient 1, was characterized by accurate movements in the basal state which became hypermetric when a load was added. The illustrated recordings were performed on the 21st day. In the basal state, the mean amplitude of movement was 16.2 degrees for an aimed target located at 15 degrees. When an extra mass of 500 gm was added, the mean amplitude of movement was 21.3 degrees. The patient was able to increase his agonist but not his antagonist activity. After addition of 500 gm, the integrated agonist EMG activity was 170% of that observed in the basal state while the integrated antagonist activity only amounted to 116% of that observed in the basal condition. The onset latency of the antagonist activity was normal both in the basal condition (42 msec) and after addition of 500 gm (35 msec). Neurological examination revealed normal findings. During the test, the patient perceived his inability to adapt himself to inertial loads but was not able to compensate for it. At this stage, the patients exhibited no hypermetria in the basal state. The movement amplitudes were similar in the patient group and in the control group, as demonstrated by Student’s test ( p = 0.728 when aimed amplitude was 15 degrees and p = 0.663 when aimed amplitude was 50 degrees). In the basal state, the onset latencies of the antagonist activity were similar in the patient group and in the control group (Student’s test, p = 0.808 when the aimed amplitude was 15 degrees and p = 0.862 when the aimed amplitude was 50 degrees). Overloading caused the appearance of hypermetria as assessed by the repeated-measure analysis of variance (load effect; significance of F: p < 0.001). The onset latency of antagonist activity was not affected by addition of masses, as attested by the repeated-measure analysis of variance (load effect; significance of F: p = 0.455). The hypermetria revealed

442 Annals of Neurology Vol 38 No 3 September 1995

-

CEREBELLAR SYNDROME - STAGE 3

CEREBELLAR SYNDROME STAGE 4

A

A

Y

42 msec

I

100 msec

-

35 msec

-

100 msec

F i g 4. Kinematic and EMG features of fast and accurate wrist flexion movements made by a cerebellar patient (Patient 1) when be was in the third stage of recovery. Movement, EMG agonist

(AGO) activity, and EMG antagonist (ANTA) activity for Jlexions made without any extra load (A)and after addition of an extra load of 500 gm (B). See Figure 1 for details.

by an artifical increase of the inertia of the moving hand was actually due to the inability of the patients to increase the intensity of their antagonist muscle. The integrated agonist EMG activity increased when an extra load (200 or 500 gm) was added, as confirmed by Friedman’s test ( p = 0.001, whether the aimed amplitude was 15 or 50 degrees). This increase of the agonist activity following addition of inertial loads was similar in the patient group and in the control group, as attested by the repeated-measure analysis of variance (group effect; significance of F: p = 0.660). At this third stage, the integrated antagonist EMG activity also increased when an extra load (200 or 500 gm) was added, as shown by Friedman’s test ( p < 0.001 when aimed amplitude was 15 degrees and p = 0.001 when aimed amplitude was 50 degrees). However, the increase of the antagonist EMG activity was substantidy smaller in the patient group than in the control group. This was demonstrated by the repeated-measure analysis of variance (group effect; significance of F: p = 0.001). The fotlrth stage, illustrated in Figure 5 for Patient 1, was characterized by accurate movements in both the basal condition and when a mass was added. The

6

A 100 msec

29 msec

I

100 msec

Fig 5 . Kinematic and EMG hatures of fast and accurate wrist pexion movements made by a cerebellar patient (Patient 1 ) when fourth stage of recoz,ery. Movement, EMG agohe was in nist(AGO) activity, and EM(; antagonist (ANTA) for Jexions made without any extra load (A) and after addition of an extra load of 500 gm (B). See Figure 1 for details.

illustrated recordings were performed on the 50th day. For an aimed target located at 15 degrees, the mean movement amplitude was 16.1 degrees in the basal state and 15.6 degrees when an extra mass of 500 gm was added to the hand. The patient could now increase both his agonist and his antagonist activity. The integrated agonist EMG activity increased from 100% (basal condition) to 171% (extra weight of SO0 gm) and the integrated antagonist EMG activity increased from 100% (basal condition) to 176% (extra weight of 500 gm). The onset latency of the antagonist activity was normal both in the basal state (30 msec), and after addition of 500 gm (29 msec). This stage was reached by only 2 patients (Patients 1 and 2). At this stage, movements were accurate in the basal state as well as after addition of an extra mass. Indeed, when the aimed amplitude was 15 degrees, the movements were smaller than 17.4 degrees in the three inertial conditions, whereas when the aimed amplitude was 50 degrees, the movements were smaller than 54.1 degrees in the three inertial conditions. The onset latency of the antagonist activity was normal in the three inertial conditions (these latencies were within the 95% confidence interval of the correspond-

Manto et al: Recovery from Cerebellar Injury 443

ing control values). Addition of masses induced an increase in the activity of the agonist as well as in the antagonist muscle (these increases were within the 95% confidence interval of the corresponding control values).

AGONIST MUSCLE

ANTAGONIST MUSCLE

B

A 95

L

7--

85 -

95 90

-; 2 -0

80-

Evolution of the EMG Parumefer.r During Recovery The calibrated EMG activity of the agonist muscle was smaller at stage 1 than in stages 2 and 3, as illustrated in Figure 6A for movements with an aimed amplitude of 50 degrees. This transient reduction appearing just after the cerebellar stroke was confirmed by the repeated-measure analysis of v,ariance (sequence effect, p < 0.02). The calibrated EMG activiry of the antagonist muscle was smaller at stage 1 than in stages 2 and 3 in all patients but 1, as illustrated in Figure 6B for movements with an aimed amplitude of 50 degrees. However, this reduction was not statistically significant (repeated-measure analysis of variance, sequence effect, p = 0.104).

Fig 6. Individual calibrated agonist (A)and antagonist (B) EMG activity associated with the fast and accurate jexions of the wrist carried out by the cerebellar patients in the basal state (without any extra load) as a function of the stage reached in the course of the recovery. Each point is the mean of the calibrated EMG associated to 12 movements when the patient was asked to make 50-degree jexion movements.

Discussion The two malor findings of this study are the following: (1) The recovery course from hypermetria after a cerebellar stroke (when the features of this latter are examined in both the basal state and after an artificial increase of the inertial load) goes through four qualitatively distinct stages. ( 2 ) During the first stage after a cerebellar stroke, the agonist activity launching the rapid and accurate movements is depressed. Actually, the four recovery stages we observed in this study can be easily interpreted in terms of agonist and antagonist activities. During the task of reaching a target as fast and as accurateily as possible, a healthy subject launches the movement by a burst of activity in the agonist muscle, and stops it by generating a burst of activity in the antagonist muscle at the appropriate time [5-8]. When the inertia of the moving limb is artificially increased, the subject adapts to the new situation by increasing both the agonist activity to put in motion the extra load, and the antagonist activity to stop the increased inertial mass on time C141. Just after the acute cerebellar injury, during the first stage, the patient presents a hypermetria due to the delayed braking activity of the antagonist muscle. When a mass is added, the patient is unable to increase either the agonist or the antagonist activity. Consequently, no additional imbalance between the launching force and the braking force occurs. This, together with the fact that the onset of the antagonist aciivity is not modified by an extra load, explains why hypermetria remains unchanged when an extra load is added. In the second stage, a hypermetria is still present in the basal state because the onset of the braking activity of the antago-

nist muscle is delayed. At this stage, when a mass is added, the patient increases the agonist activity but is unable to increase the antagonist activity to the same extent. An imbalance between the increased launching force and the unchanged braking force occurs and makes the hypermetria worse. The delay of the onset of the antagonist activity remains unchanged when an extra load is added, and therefore is not involved in the aggravation of hypermetria. In the third stage the onset latency of the antagonist activity returns to normal, so that no hypermetria occurs in the basal state. However, when an extra load is added, the patient can increase the agonist activity but not the antagonist activity. In such circumstances, the braking force is insufficient with respect to the increased launching force, causing hypermetria. The stage-by-stage recovery of hypermetria can be explained in terms of a differential recovery of the different EMG parameters controlling the execution of rapid and accurate movements. Just after an acute cerebellar injury, the patient presents several defects in the neural control of movements carried out as fast and as accurately as possible. (1) Facing an artificially increased inertial load, the patient cannot increase the agonist activity. ( 2 ) The latency of the onset of this antagonist activity is enlarged. (3) Facing an increased inertial load, the patient is not able to increase the antagonist activity. The return of an adequate intensity of the agonist muscle in the presence of an increased inertial load precedes the return of a correct onset latency of the antagonist activity (hence, the emergence of the second stage), which in turn precedes the return

75

-

70-

6560

-

Aimed amplitude 50 degrees

45

45 40 L Stage 1

444 Annals of Neurology Vol 38 No 3 September 1995

Stage 2

Stage 3

stage 1

stage 2

stage 3

of an adequate intensity of the antagonist muscle in the presence of an increased inertial load (hence, the emergence of the third stage). While it is well known that the deficits resulting from cerebellar lesions can substantially recover 117191, the underlying mechanisms are unknown. However, experimental data suggest that the use of alternative pathways plays a role. Indeed, lesioning of extracerebellar structures (notably the sensory cortex E203, the dorsal column system 1211, and the premotor cortex E22)) after recovery from a cerebellar lesion causes the reappearance of cerebellar symptoms (decompensation). Remodeling of neural circuits could also contribute to the observed recovery. Indeed, it has been demonstrated that synaptic rearrangements of remaining inputs occur in the red nucleus when the latter is deprived from its cerebellar input E233. It has also been shown that lesions of the deep cerebellar nuclei induce sprouting of axon terminals in the motor cortex 1241. When they were in the first stage, our patients presented a reduced EMG activity in their agonist muscle. In patients presenting gunshot wounds affecting cerebellum during World War I, H o h e s [l] already observed a weakness that was greater in the initial phase of the disease, and then gradually abated. H e even made of this symptom one of the fundamental cerebellar disturbances. The patients explored here also complained of this weakness during the early course of their illness. More effort was apparently necessary to perform movements, in spite of the fact that clinical examination showed no pyramidal sign and that there was no sign of brainstem lesion or compression on magnetic resonance imaging. The reduction in the intensity of the agonist muscle observed in our patients might be related to the withdrawal of the facilitory effect exerted by the cerebellum on the motor cortex r25,261. This study was supported by a grant from the Philippe and T h M s e Lefebvre’s Fund. Mario Manto was supported by a grant from the Erasme Foundation. We are grateful to Christiane Busson for secerarial assistance and to Bernard Foucart for taking care of the electronic equipment.

References 1. Holmes G. The symptoms of acute cerebellar injuries due to gunshot injuries. Brain 1917;40:461-535 2. Holmes G. The clinical symptoms of cerebellar disease and their interpretation. The Croonian lectures. 11. Lancet 1922;1:12311237 3. Holmes G. The cerebellum of man. The Hughlings Jackson memorial lecture. Brain 1939;62:1-301 4. Gilman S, Bloedel JR, Lechtenberg R. Disorders of the cerebellum. Contemporary neurology series, vol 2 1. Philadelphia: Davis, 198L 5. Meinck HM, Benecke R, Meyer W, et al. Human ballistic finger

flexion: uncoupling of the three-burst pattern. Exp Brain Res 1984;55:127-1 33 6. Hannaford B, Stark L. Roles of the elements of the triphasic control signal. Exp Neurol 1985;90:619-634 7. Wierzbicka MM, Wiegner AW, Shakani BT. Role of agonist and antagonist muscles in fast arm movements in man. Exp Brain Res 1986;63:331-340 8. Hallett M, Marsden CD. Ballistic flexion movements of the human thumb. J Physiol (Lond) 1979;294:33-50 9. Conrad B, Brooks VB. Effects of dentate cooling on rapid alternating arm movements. J Neurophysiol 1974;37:792-804 10. Flament D, Hore J. Movement and electromyographic disorders associated with cerebellar dysmetria. J Neurophysiol 1986;55: 1221-1233 11. Marsden CD, Merton PA, Morton HB, et al. Disorders of movement in cerebellar disease in man. In: Rose FC, ed. The physiological aspects of clinical neurology. Oxford, U K Blackwell, 1077:179-199 2. Hallett M, Berardelli A, Matheson J, et al. Physiological analysis of simple rapid movements in patients with cerebellar deficits. J Neurol Neurosurg Psychiatry 1991;53:124-133 3. Hallett M, Shahani BT, Young RR. EMG analysis of patients with cerebellar deficits. J Neurol Neurosurg Psychiatry 1975; 38:1163-1169 4. Manto M, Godaux E, Jacquy J. Cerebellar hypermetria is larger when the inertial load is artificially increased. Ann Neurol 1994; 35:45-52 15. Manto M, Godaux E, Jacquy J. Detection of silent cerebellar lesions by increasing the inertial load of the moving hand. Ann Neurol 1995;37:344-350 16. Gottlieb GL, Corcos DM, Agarwal GC. Organizing principles for single-joint movements. I. A speed-insensitive strategy. J Neurophysiol 1989;62:342-357 17. Goldberger ME, Growdon JH. Pattern of recovery following cerebellar deep nuclear lesions in monkey. Exp Neurol 1973;39: 307-322 18. Kase CS, Norving B, Levine SR, et al. Cerebellar infarction. Clinical and anatomic observations in 66 cases. Stroke 1993;24: 76-83 19. Amarenco P, Roullet E, Goujon C, et al. Infarction in the anterior rostral cerebellum (the territory of the lateral branch of the superior cerebellar artery). Neurology 1991;41:253-258 20. Mackel R. The role of the monkey sensory cortex in the recovery from cerebellar injury. Exp Brain Res 1987;66:638-652 2 1. Carpenter MB, Correll JW. Spinal pathways mediating cerebellar dyskinesia in rhesus monkey. J Neurophysiol 1961;24:535551 22. Aring CD, Fulton JF. Relation of the cerebrum to the cerebellum. 11. Cerebellar tremor in the monkey and its absence after removal of the principal excitable areas of the cerebral (areas 4 and 6). 111. Accentuation of cerebellar tremor following lesions of the premotor area (area 6, upper part). Arch Neurol Psychiatry 1936;35:439-466 23. Tsukahara N, Hultborn H, Murakami F, Fujito Y . Electrophysiological study of formation of new synapses and collateral sprouting in red nucleus neurons after partial denervation. J Neurophysiol 1975;38:1359-1372 24. Keller A, Arissian K, Anasuma H . Formation of new synapses in the cat motor cortex following lesions of the deep cerebellar nuclei. Exp Brain Res 1990;80:23-33 25. Sasaki K, Kawaguchi S, Oka H, et al. Electrophysiologicalstudies on the cerebellocerebral projections in monkeys. Exp Brain Res 1976;16:495-507 26. Shinoda Y, Kano M, Futami T. Synaptic organization of the cerebello-thalamo-cerebral pathway in the cat. I. Projection of individual cerebellar nudei to single pyramidal tract cells in areas 4 and 6. Neurosci Res 1985;2:133-156

Manto et al: Recovery from Cerebellar Injury

445