Action Tremor During Object Manipulation in Parkinson s Disease

Transcription

1 Movement Disorders Vol. 15, No. 2, 2000, pp Movement Disorder Society Action Tremor During Object Manipulation in Parkinson s Disease *Hans Forssberg, MD, PhD, Páll E. Ingvarsson, MD, PhD, *Nobuaki Iwasaki, MD, PhD, Roland S. Johansson, MD, and Andrew M. Gordon, PhD From the *Department of Woman and Child Health & Nobel Institute for Neurophysiology, Karolinska Institute, Stockholm, Sweden; Department of Clinical Neurosciences, Section of Neurology, University of Göteborg, Sweden; Department of Physiology, University of Umeå, Umeå, Sweden; Department of Biobehavioral Science, Teachers College; and Department of Rehabilitation Medicine, College of Physicians and Surgeons, Columbia University, New York, NY, U.S.A. Summary: In previous studies of fingertip forces during precision grip in subjects with Parkinson s disease (PD), we observed regular oscillations in isometric force. The present study characterizes the nature of these oscillations. Fingertip forces were recorded from the index finger and thumb during precision grip-lifts with a 300 g and 900 g object in 10 subjects with PD and 20 healthy control subjects. Fourier analysis confirmed that all subjects with PD exhibited force oscillations with a clearly definable frequency ( 7 11 Hz). Five of these subjects also exhibited a second lower frequency peak ( 5 Hz). Approximately half of the 20 control subjects displayed a single frequency peak ( 8 12 Hz), which was generally lower in amplitude than in the subjects with PD (representing enhanced physiological tremor), whereas the remaining control subjects had low-amplitude, broad-based spectra (representing physiological tremor). The amplitude of the force oscillations was higher for lifts with the heavier object in both the control subjects and subjects with PD. L-Dopa resulted in a decreased tremor amplitude but did not influence the frequency. The force oscillations of the two opposing digits normal to the grip surfaces were in phase, whereas the oscillations tangential to the grip surfaces were often out of phase. The results suggest that the multipeaked force rate trajectories reported previously are caused by action tremor. The similarity of force oscillations in subjects with PD and healthy control subjects suggests common tremor-generating mechanisms and supports the notion that the parkinsonian action tremor (AT) is an exaggerated form of physiological tremor. These findings provide insight into the impaired hand function observed in individuals with PD. Key Words: Action tremor Force oscillations Parkinson s disease Hand function Precision grip. Lack of manual dexterity is a common complaint in subjects with Parkinson s disease (PD), even during the early stages of the disease. However, the pathophysiological mechanisms underlying the disturbed hand functions are unclear. Recently, we have investigated the coordination of fingertip forces while subjects with PD grasped and lifted a small instrumented device using the precision grip (that is, between the tips of the thumb and the index finger 1 4 ; see also reference 5). We found that subjects with PD often were slow establishing digit contact with the object and had prolonged transitions between the various phases of the grip-lift movement. The Received March 12, 1997; revision received July 20, Accepted November 22, Address correspondence and reprint requests to Andrew M. Gordon, PhD, Department of Biobehavioral Sciences, Teachers College, Columbia University, Box 199, 525 West 120th Street, New York, NY 10027, U.S.A. force increase appeared discontinuous because the force rate profiles were not single-peaked and bell-shaped, and instead resembled the profiles seen in young children 6 and patients with hemiparesis 7 who are unable to scale the rate of force increase in advance, that is, use anticipatory control. Nevertheless, all subjects displayed appropriate parallel coordination of the grip (squeeze) and load (vertical lifting) forces and the durations of isometric force increase were unaffected by the disease. 3 Furthermore, the amplitudes of the force rates were appropriately scaled according to the object s size, weight and texture (that is, faster rates of force increase for larger, heavier, or more slippery objects), 3 suggesting that the oscillations were not the result of a lack of anticipatory control. Rather, the regularity of the oscillations suggested that these were the result of tremor. Classic resting tremor (RT) is considered to be the most common tremor type in Parkinson s disease It 244

2 ACTION TREMOR IN PD 245 mainly affects the extremities, particularly the arms and hands, 11 and typically occurs at frequencies of 4 5 Hz. 9 However, RT decreases or disappears during voluntary movement. 10,12 Several other types of tremor have been described in PD, including postural tremor, action (movement, kinetic) tremor, physiological tremor, cogwheeling, and clonus. 9,13,14 Parkinsonian action tremor (AT) is found in the distal parts of the upper limbs and tends to be more pronounced on the body side that is most affected by the disease Usually its amplitudes are smaller and its frequencies higher than those of RT, but reports on tremor frequency vary from 4.8 Hz 21 to 12 Hz. 18 AT was previously thought to be present only during movement, 12 but is currently considered to occur... during any voluntary muscle contraction, which includes postural, kinetic, isometric and task specific tremors. 22 The present study further examines the oscillations in fingertip forces in parkinsonian subjects during the grasping and lifting of an instrumented manipulandum using the precision grip. 1,3,4 Specifically, our aim was to: 1. determine the frequency and amplitude of the oscillations and compare them with oscillations observed in healthy age-matched subjects; 2. study phase relationships in the normal and tangential force oscillations recorded at the two engaged digits; 3. determine the effects of object load on the oscillations; and 4. determine the influence of dopaminergic medication on the oscillations. We will test the hypothesis that the oscillations in subjects with PD are the result of action tremor, which shares the same neural mechanisms as physiological or activated physiological tremor. 18 METHODS Subjects Ten right-handed subjects (mean age, 54.5 ± 9.8 yrs) with idiopathic Parkinson s disease (Table 1) were compared with 20 healthy subjects (mean age 42.4 ± 14.5 yrs). These subjects and 10 of the healthy subjects participated in earlier studies. 1,3,4 All subjects gave their informed consent. Immediately after each test session, the subjects with PD were clinically examined by a licensed neurologist, and scored according to Hoehn and Yahr 23 and the Motor Examination part of the Unified Parkinson s Disease Rating Scale (UPDRS). 24 For descriptive purposes, the PD subjects were divided into early, moderate, and late subgroups according to their UPDRS score when OFF medication. The UPDRS resting tremor (RT) scores, the combined postural (POT) and action (AT) tremor scores, and the sum of all UPDRS tremor ratings (maximum value 28) were also recorded separately (Table 1). The healthy subjects were divided into two reference groups (A and B) according to the total power of their oscillations (see below). Apparatus and Procedures The instrumented test object was a slightly modified version of one described previously. 25 Briefly, it had two parallel grip surfaces (35 35 mm, 20 mm apart) covered with fine (200-grit) sandpaper. By inserting appropriate exchangeable masses into a slot at the base, the weight of the object could be adjusted to either 300 g or 900 g without changing its visual appearance. The grip force (normal force) and the vertical load force (tangential force) were measured at each grip surface by straingauge transducers (DC-120 Hz). The vertical movement Patient no. Age (yrs) TABLE 1. Clinical descriptions and rating scale scores in subjects with Parkinson s disease Disease duration (yrs) Stage (H&Y)* Total score (& subgroup) UPDRS scores (in the off/on state) Right side RT - Right hand AT&POT - Right hand Tremortotal score I 11/5 (Early) 11/1 2/1 1/1 3/ II 23/1 (Early) 20/1 2/0 0/0 2/ II 26/12 (Early) 16/7 1/0 0/0 3/ II 29/7 (Moderate) 24/7 1/1 1/0 4/ III 40/13 (Moderate) 31/12 2/1 1/0 5/ III 40/17 (Moderate) 25/9 1/2 1/1 6/ II 42/10 (Moderate) 31/10 2/0 1/0 7/ III 58/3 (Late) 45/3 3/0 1/0 8/ III 70/11 (Late) 44/7 3/0 3/1 15/ IV 74/5 (Late) 52/3 4/0 3/0 22/2 Mean (2.5) 41.3/ / / / /1.0 * H&Y, assessment of severity of disease (stages I V) according to Hoehn and Yahr. 23 Motor Examination part of the Unified Parkinson s Disease Rating Scale (UPDRS) 24 : RT, resting tremor (score 0 4), according to the UPDRS protocol. AT & POT, action tremor (score 1 2) and postural tremor (score 3 4), combined into the same item on the UPDRS.

3 246 H. FORSSBERG ET AL. FIG. 1. Grip (GF) and load (LF) forces of the index finger, the vertical position, and the grip force rate (dgf/dt) and load force rate (dlf/dt) of a subject with PD during a lift of the 900-g object. The fingertip forces increase during the loading phase (T1-T2), the object is transported upward during the transition phase (T2-T3), and held in the air during the static phase (T3-T4). Note the force oscillations throughout the lift. was recorded by a camera containing a light-sensitive photoresistor (United Detector Technology, SC/28, Baltimore, MD, USA) sensing the position of an infrared light-emitting diode attached to the object (DC-110 Hz). Subjects sat on a chair in front of an adjustable table, and lifted the object with their right hand approximately 5 cm above the table surface using the precision grip, between the tips of the thumb and index finger. After 4 6 seconds, the object was replaced on the table. The task was carefully explained to each subject and several demonstration trials were performed by the experimenter. After several practice trials before each condition, subjects performed five lifts with each weight (300 g and 900 g). The subjects with PD were studied both in the practically defined OFF state after 12 hours without dopaminergic medication 26,27 and in the medicated ON state approximately 1 hour after medication. Five of the subjects were first tested OFF medication and the other five were first tested ON medication. There were no influences of medication order for any of the measured parameters (p >0.05 in all cases). Data Analysis The grip and load forces applied by each digit were sampled at 400 Hz and the position signal at 100 Hz. The signals were digitized with 12-bit resolution and stored in a laboratory computer system (SC/ZOOM, Department of Physiology, Umeå University). A graphics terminal was used to indicate the various phases of the grip-lift sequence shown in Figure 1. 4 The loading phase (T1-T2) was defined as the period of isometric load force increase until the object moved from its support. The transition phase (T2-T3) was the period when the object moved from its support, and the static phase (T3-T4) was the period the object was held in the air. The grip and load force data were collected in epochs of 3.5 sec from the static phase of the lift with the ZOOM software. Numeric computation and visualization was made in MATLAB (MathWorks, Inc, Natick, MA, USA). The power spectral density functions of the force oscillations during the static phase of the lift were estimated by Fourier transformation (the FFT algorithm) on the force rate signals (dgf/dt and dlf/dt). This time differentiation suppressed the low-frequency noise (nontremor fluctuations in force level of low frequency) and induced a whitening effect, that is, a slow filter effect that improved the balance between low- and high-frequency portions of the power spectrum. Standard preprocessing procedures were used 28 : the mean and systematic trends were subtracted from the signal, which was then tapered by the Hanning windowing method. The spectral frequency resolution was increased to 0.1 Hz by adding 0 values before and after the data epoch. If a peakfrequency below the 2-Hz range remained, the remaining low-frequency noise was filtrated at 2 or 3 Hz. The final power spectrum was obtained by averaging the power spectra for the five test trials for each weight for each subject. From these averaged power spectra, the following parameters were calculated 29 : 1. The peak frequency the frequency (within the 2 17 Hz range) at which the peak power of the spectrum occurred; in five subjects with PD two peaks were found. In this case, the peak in the higher frequency range (>6 Hz) was recorded, although subsequent analyses were performed on each peak separately; 2. the average peak power the power in a 1-Hz band centered at the peak-frequency; 3. the total power the power of all spectral components between 2 and 17 Hz; 4. the peak power ratio the ratio between the average peak power and the total power; 5. the mean tremor amplitude the square root of the total power of the spectrum (between 2 and 17 Hz); and 6. the Root Mean Square (RMS) of the force rate recordings was used to estimate the RMS amplitude from the first time derivatives of the subtracted and the added load forces measured at the index finger and thumb, respectively.

4 ACTION TREMOR IN PD 247 The latter analysis separately characterized the amplitude of a rotational tremor component and of a common vertical tremor component. FFT could not be performed on the data during the loading and transition phases of the lift because the short duration (sometimes less than 100 msec) of these phases resulted in too few tremor cycles (often two or less). We analyzed the phase relationships between the oscillations in the various force components during the static phase by calculating the cross-correlation in each trial between the two grip forces and between both load forces (from the thumb and the index finger, respectively), and between the load and grip forces in each digit. These estimates were considered meaningful only if the peak cross-correlation coefficients from all five test trials were above Statistical Analysis The following nonparametric tests were performed on subjects median values: Wilcoxon signed rank tests were used for paired within-subject comparisons, Friedman two-way analysis of variance by ranks (F r ) were used for multiple within-subject comparisons; 3 + Mann- Whitney U-tests were used for comparisons between groups, and estimates of Spearman rank order correlation coefficient (r s ) were used to determine relationships between tremor parameters. All tests were performed at the p <0.05 level. Unless stated otherwise, the reported data refer to analyses of the load force recorded from the index finger during lifts with the 900-g object and for the subjects with PD when they were tested OFF medication. RESULTS Tremor Ratings All subjects with PD exhibited a resting tremor (RT) in their most affected (right) hand when tested according to the UPDRS protocol in the OFF state (Table 1). They scored between 1 and 4 on a four-point scale (2.1 ± 1.0; mean ± standard deviation). The RT disappeared following medication in six of the 10 subjects with PD and decreased in two of the four remaining subjects (0.5 ± 0.71). The score for the combined postural and action tremor often was less than the score for the RT, and two subjects in the early subgroup had no observable postural or action tremor. This score also decreased following medication and disappeared in all but three subjects. In general, both RT and AT scores increased in parallel with the severity of disease (as scored by the UPDRS; see Table 1). FFT Profiles of Force Oscillations Despite the lack of observable postural or action tremor in some subjects OFF medication and in most subjects ON medication, all 10 subjects with PD exhibited regular, approximately sinusoidal, oscillations in force during all phases of the lifting task. This can be seen in the grip force, load force, and their derivatives shown for a representative subject with PD in Figure 1. Several of the 20 healthy subjects also exhibited such force oscillations. Figure 2 illustrates examples of grip and load force rates from the index finger (top) and corresponding frequency spectra (bottom) during a lift with FIG. 2. Examples of grip and load force rate signals from the index finger during lifts with the 900-g object, and the estimated squared mean power spectral densities in three subjects with PD OFF medication (A, B, and C) and in two healthy control subjects (D and E). Note the overlap in tremor amplitude between subjects with PD and the control subjects. (A) The subject with PD with the most pronounced tremor (PD subject no. 10) showing a high- and narrow-frequency peak in his spectrum. (B) Subject with PD (PD subject no. 9) with two simultaneously occurring frequency peaks in her force spectrum. (C) Subject with PD with the least power and a broad-based bimodal frequency spectrum (PD subject no. 8). (D) Control subject (from group B) with irregular force oscillations of low energy and a largely flat force spectrum. (D) Control subject from group A with a medium broad-based FFT spectrum and force oscillations of high amplitude and power. Note the different scales of power spectral densities in A.

5 248 H. FORSSBERG ET AL. FIG. 3. The total power (plotted on a logarithmic scale) plotted as a function of the peak power ratio of the load force oscillations from the index finger during lifts of the 900-g object. Symbols refer to individual subjects. The subjects with PD (filled symbols) are divided into subgroups according to UPDRS scores. The control subjects (open symbols) are subdivided into two groups (A and B) according to the total power of their force oscillations. the 900-g object for three subjects with PD (Fig. 2A C) and two control subjects (Fig. 2D, E). Note that to some extent oscillations can be seen in all subjects, although the frequency and amplitude varied across the subjects. In some subjects (Fig. 2A), one distinct peak could be observed. This was true of five of the subjects with PD we tested. Other subjects (n 5) displayed two frequency peaks (Fig. 2B; Table 2). One of the latter subjects with PD (who had the least power) exhibited a broad-based bimodal spectrum (Fig. 2C). There was also a wide variation in the control subjects, with some showing oscillations of low energy and a flat spectrum (Fig. 2D) whereas others had broad-based spectra with medium amplitude and power (Fig. 2E). As shown in Figure 3, the total power and peak power ratio varied greatly in the subjects with PD and control subjects, although these measures were generally higher for the subjects with PD (p <0.01 in both cases). Nevertheless, there was some overlap between the two groups (see also Table 2). The total power of the force oscillations, together with the average peak power and the peak power ratio, generally increased with the severity of disease (Table 2). An exception to this was PD subject no. 8, who was severely disabled by rigidity and slowness, had moderate tremor at rest, but only slight action tremor (1 of 4) on the UPDRS scale (see Table 1). The total power of the force oscillations correlated well with the peak power ratio for the entire group (n 30) of subjects (r s 0.80, p < ). Because it was difficult to assess a single peak frequency from broad-based frequency spectra (cf. 13), we divided the group of healthy subjects into two equal groups (reference groups A and B) according to their total power (Table 2). Unless stated otherwise, subsequent analyses were performed only on the group of 10 TABLE 2. Median values characterizing oscillations in load force data from the index finger during lifts with the 900 g object in 10 subjects with PD in the off state* PD subject Peak-frequency (Hz) Average peak power (N/s) 2 Total power (N/s) 2 Peak power ratio Amplitude (FFT) (N/s) RMS amplitude (N/s) / / (3.50/2.17) 0.14/ (1.87/1.47) / / (6.24/4.96) 0.21/ (2.50/2.23) / / (0.71/0.55) 0.11/ (0.85/0.74) / / (6.93/18.48) 0.11/ (2.63/4.30) / / (23.02/2.85) 0.38/ (4.8/1.69) 5.33 Medians (& quartiles), 9.8/ / (6.24/2.85) 0.22/ (2.50/2.09) 3.73 PD subjects (9.5/10.5) (0.76/7.25) (2.88/20.06) (0.14/0.35) (1.8/5.16) (1.97/5.0) Reference group A (8.5/11.7) (0.21/0.52) (1.28/2.77) (0.13/0.16) (1.26/1.83) (1.32/2.07) Reference group B 9.5* 0.097* * (7.1/10.9)* (0.048/0.15)* (0.32/0.90) (0.067/0.12)* (0.81/1.07) (0.88/1.22) * In five PD subjects with two major frequency peaks, data from the high-/low-frequency peaks are reported separately. Medians (and quartiles) of the individual medians are also shown for the 10 subjects with PD, reference groups A and B. Asterisks indicate uncertain data, as a result of ambiguous peak-frequency determinations in reference group B.

6 ACTION TREMOR IN PD 249 control subjects with the highest total power (hence referred to as reference group A ). Frequency of Force Oscillations As described above, the force oscillations were present as soon as the digits contacted the object and continued throughout all phases of the lift (see Fig. 1). Visual inspection of the force traces suggested that the frequency of oscillations was similar throughout all phases, although in general they appeared slightly faster in the loading and transition phase than in the subsequent static phase. However, we were only able to characterize the tremor in the static phase as a result of the short durations of these phases, as described in the Methods. Figure 4 plots the median oscillation frequencies of grip and load force rate (Fig. 4A) and the load force rate during lifts with each weight (Fig. 4B) for each subject with PD, as well as a box-whisker plot of the median, quartiles, and range of reference group A. The frequencies of force oscillations in the static phase were similar (p >0.05) in the subjects with PD (median 9.80) and the control subjects (median 9.85) in group A, although there were large individual differences in both groups of subjects (Fig. 4A). Interestingly, as described above, five subjects with PD also exhibited a second, lowerfrequency oscillation in force, generally in the range of 4 6 Hz, that is, corresponding to the frequency of resting tremor (Fig. 4, hatched symbols). The frequency of the load force oscillations was higher during lifts with 900 g than during lifts with 300 g in seven of the 10 subjects with PD (Fig. 4B), but p >0.05 in both the PD and control groups. The frequency of the lower-frequency peaks was higher for lifts of the 300-g object in all five subjects with PD with two peaks. Amplitude of Force Oscillations Not surprisingly, the subjects with PD generally exhibited a larger amplitude of force oscillations during the static phase than the subjects in reference group A (p <0.01 for the load and grip force amplitudes). This can be seen in Figure 5A, which shows the amplitude of the grip and load force oscillations (see also Table 2). Note, however, that the amplitudes of four subjects with PD, including PD subject no. 8, were within the range of the reference group. The amplitudes of the load force oscillations were lower than those in the grip force for nine of the 10 subjects with PD (p <0.01) and for eight of the 10 subjects in reference group A (p <0.05). Figure 5B shows that the amplitudes of load force oscillations were lower in the index finger than in the thumb in all subjects with PD (p <0.01), whereas no differences were found between the amplitudes of grip force from the two digits in the subjects with PD (not shown). No differences in oscillations were found between the two digits in either force in reference group A (p >0.05 in all cases). The amplitude of the force oscillations during the static phase was strongly influenced by the weight of the object. The total amplitude of the grip and load force oscillations was higher for the 900-g object than for the 300-g object for all but one of the subjects with PD (PD subject no. 10) (but p <0.1) and for the subjects in reference group A (p <0.01; Fig. 5C). For the subjects with two peaks, the amplitude of both peaks increased with weight. As mentioned above, one subject (PD subject no. 10) in the late subgroup differed strongly from the others (Fig. 5C). When this subject lifted the 300-g object, the total amplitude of the force oscillation was approximately 10 times larger than that observed for the other subjects with PD. This amplitude decreased into the range of the other subjects with PD when the object load increased from g. During lifts with the 300-g object, this subject displayed two peaks of similar am- FIG. 4. Frequency of force oscillations from the index finger. (A) Grip and load force rates (weight 900 g). (B) Load force rates while subjects lift a 900-g and a 300-g object. Symbols represent data from individual subjects with PD in the OFF state, and pairs of symbols referring to individual subjects are joined by lines. Hatched symbols and connecting lines indicate a second low-frequency tremor peak found in five of the 10 subjects with PD. Data from reference group A are represented by box-whisker plots showing medians, quartiles, and minimum and maximum values.

7 250 H. FORSSBERG ET AL. FIG. 5. Total amplitudes (the square root of the total power of the averaged FFT spectrum, see Methods) of the force oscillations recorded from the index finger during lifts of the 900-g object. (A) Load and grip force rates. (B) Load force rate in the index finger and in the thumb. (C) The effect of object load on the load force oscillations. For further details, see legend to Figure 4. All subjects with PD were in the OFF state. plitude, with the frequency of the higher peak (11.8 Hz) being twice as high as the lower one (5.9 Hz). During lifts with the 900-g object, the amplitude of the lowerfrequency peak was lower than that of the highfrequency peak (Fig. 2A; Table 2). Thus, in contrast to the other four subjects with PD with two peaks, the amplitude of both peaks decreased by loading in this subject. FIG. 6. (A) Frequency and (B) total amplitude of load force oscillations in the index finger during lifts of the 900-g object OFF and ON medication in the subjects with PD. For further details, see legend to Figure 4. FIG. 7. (A) Effect of object load and (B) medication during lifts of the 900-g object on the force oscillations of the high-frequency portion (filled symbols) and the low-frequency portion (unfilled symbols, displaced slightly to the right) of the frequency spectrum in the five subjects with PD that displayed double-frequency peaks. Load force rates from the index finger are shown. For further details, see legend to Figure 4. Note the different scales. Effects of Medication As seen in Figure 6A, there was no consistent change in the frequency of the force oscillations following L-dopa and other medications, including the subjects who exhibited both high- and low-frequency peaks. Medication dampened the total amplitude of the load force oscillations in eight of the 10 subjects with PD (but p >0.05; Fig. 6B). The effect of medication on the amplitude of the two frequency components in the five subjects with PD who exhibited double-frequency peaks is plotted in Figure 7B. A similar tendency for the amplitude of the high- and low-frequency peak to decrease following medication after treatment was seen in most subjects. Phase Coordination of Force Oscillations The pattern of force oscillations described above for the index finger was also representative for those of the thumb. Because these two digits shared a common load and were coupled to the same hand arm system, one would expect they are subjected to a common motor drive. This could be tested as our test object separately recorded the forces of the thumb and index finger, allowing analysis of the phase relationship between the oscillations of the two digits. High cross-correlation coefficients (p >0.45) allowed analysis of phase shifts be-

8 ACTION TREMOR IN PD 251 tween the two grip force rates for all 10 subjects with PD and 20 control subjects, whereas the correlation coefficients between the two load forces were more variable (significant during lifts with the 300-g object only in five subjects with PD; and during lifts with the 900-g object in one subject with PD and four of the 20 healthy subjects). Except for the thumb in two subjects with PD, no significant correlations were present between the grip and load force oscillations. The oscillation of the grip forces of the thumb and index finger were approximately in phase for both the PD and healthy subjects, that is, the opposing forces increased and decreased in parallel, like would be expected from their mechanical coupling. In contrast to the grip forces, the oscillations of the load forces of the index finger and the thumb were more variable. They were out of phase, that is, with a phase shift close to 180 in three of the five subjects with PD who had correlation coefficients >0.45 during lifts with the 300-g object, and in all four such healthy subjects. The oscillations were in phase (with less than 20 phase shift) in the remaining two subjects with PD with significant 300-g lifts and in the sole subject with PD with correlated load forces during lifts with the 900-g object. In the only two subjects with PD in whom grip and load forces were correlated, phase shifts approached 180. Figure 8 compares the rotational and the vertical components (see Methods) of the force oscillations during lifts of the 300-g and the 900-g object. During lifts with the lighter object, the rotational components of the force tremor was larger than the vertical tremor component in most subjects with PD. Both components became stronger with the heavier weight (900 g), but the vertical component increased more than the rotational one. DISCUSSION The present results support our earlier hypothesis 1,3,4 that the discontinuous force increase observed in subjects with PD is the result of action tremor. Furthermore, the similarities between the tremor observed in subjects with PD and the healthy control subjects is in agreement with the notion that action tremor and physiological tremor are of the same origin. These results provide insight into the impaired hand function observed in individuals with PD. Tremor in Parkinson s Disease All of the subjects with PD we tested had regular oscillations in fingertip forces during the precision grip-lift task. The range of frequencies of the force oscillation seen in subjects with PD with one frequency peak, as well as the high-frequency peak in subjects FIG. 8. Rotational and vertical components of the RMS amplitude of force tremor in subjects with PD in the unmedicated (OFF) state. The rotational tremor component was calculated as the first time derivative of the difference between the load forces recorded in the index finger and thumb, and the vertical component as the first time derivative of the sum of the load forces recorded in the index finger and thumb. Data points from the static phases of 300-g (unfilled symbols) and 900-g lifts (filled symbols) from each subject are joined by straight lines. with two peaks (in all cases Hz for lifts with the 900-g object), corresponds well to the frequencies described by Lance et al. for action tremor (AT). 18 In their study, more than half of the subjects with PD tested had a marked tremor during active contractions in upper and lower limb muscles at frequencies ranging between 7 Hz and 12 Hz. The tremor rhythm in all cases was correlated with a grouping of muscle potentials that reflected rhythmic co-contractions of antagonistic muscles. The force oscillations were observed throughout all phases of the lift. Compared with clinical terminology, the action tremor in the load force observed during the isometric force increase of the loading phase (T1-T2; Fig. 1) and during the transition phase (T2-T3; Fig. 1) may be considered isometric and kinetic variants of action tremor, respectively, whereas the grip force is always isometric. In contrast, load force tremor observed during the static (hold) phase (T3-T4; Fig. 1) may be considered postural tremor (see reference 22). Findley et al. 13,15,16,30 suggested that three different types of postural and action tremor exist, corresponding to the following:

9 252 H. FORSSBERG ET AL. 1. force oscillations around 6 Hz, 2. oscillations between 6.5 and 8.5 Hz, and 3. rippling within the physiological tremor range ( Hz). Our findings do not support such a subclassification. Rather, our findings are more in line with the view of Lance et al. 18 and others 19,31 that postural and action tremor may depend on the same pathophysiological mechanism producing a tremor in a wide 6 12-Hz frequency band. The slow tremor component of the double-peaked Fourier spectrum, seen in five of the subjects with PD in the present study, is more puzzling. Findley et al. have also described this double-peaked appearance of the frequency spectrum in 75% of the subjects with PD they tested. 13,15,16 The low-frequency component could be a persistent, but asymptomatic, component of resting tremor. 13 One exception to this pattern of an asymptomatic RT was PD subject no. 10. This subject exhibited a severe hand tremor that appeared to be unaffected by the lifting of the 300-g object and seemed to persist before, during, and after the lift. A 2:1 ratio was observed between the higher-amplitude 5.9-Hz peak and the 11.8-Hz peak. It is unclear as to whether the second peak is a sideband or whether the two peaks represent different tremors (see reference 32), although the large difference in frequencies (up to 5.8 Hz) suggests the latter. When the object load was increased to 900 g for this subject, the tremor amplitude decreased dramatically, whereas the tremor frequency increased to 8.0 Hz (in the AT range). It may be that a continuing RT was so strong in this subject that the motor drive while lifting the light weight was insufficient to affect it, whereas the stronger motor drive used for the lifts with the heavier object caused a switch from the RT to an AT. Similarly, Yanagisawa and Nezu 20 examined electromyography of the leg muscles in standing parkinsonian subjects. They found that the stance imposed a load heavy enough to switch from a slow tremor of Hz to a fast Hz tremor in approximately 25% of the subjects they tested. Also, when an unloaded arm 20 or finger 21,33 is lifted, the lowfrequency postural tremor reported may be generated by an RT generator. This could explain the low AT frequencies reported in these studies. Indeed, Cleeves et al. 15 point out that in certain situations, it can be difficult to differentiate an RT that persists in posture from a true postural (or other types of action) tremor. Action Tremor versus Physiological Tremor Lance et al. 18 suggest that AT in PD represents an exaggeration of normal physiological tremor. This hypothesis is supported by the present study. The peakfrequency of the oscillations observed was nearly identical for the subjects with PD and the control subjects. Furthermore, in both groups, the tremor amplitude increased with an increased load with no consistent change in frequency. The group of healthy control subjects who had the least tremor power (reference group B) displayed broad-based spectra of low energy, which is typical of physiological tremor. 13,30,34,35 The control subjects with the greater tremor (reference group A) all had an easily discernible single frequency peak, a pattern usually ascribed to an activated (or enhanced) physiological tremor. 9,13,36 Thus, the results from the present study suggest that there is a continuum in tremor intensity from physiological tremor to AT with an overlap between the subjects exhibiting enhanced physiological tremor and subjects with PD in the early stage of disease. Pathophysiological Aspects The oscillations of the grip forces in the index finger and the thumb usually correlated well and were always approximately in phase (otherwise the object would shift between the fingers). Attempts to find phase relations between the grip and load forces were more difficult. Large differences in the peak frequency of the grip and load force oscillations, the double peaks in five of the subjects with PD, and broad-based spectra in some subjects are possible reasons for the lack of correlation between the force oscillations. However, considerable phase differences (usually between 90 and 180 ) were observed in the few cases when significant crosscorrelation coefficients were present. The phase shifts indicate that the grip and load force in various digits may partly be controlled by different neural networks. For example, when two opposing digits engaged in the precision grip contact different surface textures (for example, sandpaper and silk), the ratio between the grip and load forces at each digit is adapted independently to the local frictional condition. 37,38 In the present study, the high tremor amplitude, which was seen in many subjects with PD and was often out of phase, would have resulted in pronounced object rotation if the center of gravity of our test object had not been well below the grip areas. Indeed, torque oscillations caused by this phase difference of load force oscillations (Fig. 8) may explain the pill-roll behavior considered characteristic of subjects with PD when they grasp small objects. Because of the presence of multipeaked velocity (for example, see references 39 41) or force rate 42 trajectories, it has been suggested in several previous studies that subjects with PD cannot adequately program movements in advance (for example, see references and 43), which may explain their bradykinesia. 41 However, in our

10 ACTION TREMOR IN PD 253 lifting task, the subjects with PD scale the force output during the isometric loading phase in adequate anticipation of the lift-off taking place when the load force counterbalances the weight of the object. 3 Thus, the multipeaked movement and force trajectories in subjects with PD during some tasks may be caused by the AT rather than reflect a deficit in anticipatory control of movements. Recent studies suggest that slow finger movements in humans are characterized by discontinuities (accelerations/decelerations) in the 8 10 Hz domain. 44,45 These movements can be implemented by a series of muscle force pulses, reflecting the action of a pulsatile motor control circuit (central generator). Accordingly, the large force oscillations seen in subjects with PD may represent an overexpression of such pulsatile commands. Furthermore, the lack of bradykinesia during this task suggests that the prevailing views regarding the mechanisms underlying bradykinesia 41 may need to be reevaluated. Clinical Considerations Most neurology textbooks either do not mention parkinsonian AT or if they do, consider it to be of minor importance 12,31,46 (see also references 15 and 47). However, in clinical practice, a moderate AT is regarded to be more disabling than a severe RT. AT was seen during the clinical examination in eight of the 10 subjects in this study (Table 1), and the force recordings indicated that it was present in all subjects with PD. Although tremor is one of the cardinal symptoms of PD, there is no universally accepted method to quantify it in clinical practice. RT is variable and unpredictable, and has been suggested to be omitted from the subjective rating scales. 48 Even when RT is appropriately quantified (for example, by triaxial computerized accelerometry), 49 its reliability for clinical use has been questioned. 22,50 The present study demonstrated that action tremor can be detected in the fingertip forces during grasping even when there is a lack of observable postural or action tremor when rated clinically. Given that subjects with PD do not use a larger safety margin in their grip forces to prevent slips 3,4 (see reference 27), the continuous oscillations throughout all phases of the lift could sometimes result in the grip force decreasing below the minimal force required to maintain contact, resulting in slips. This would indeed contribute to the impaired manual dexterity in PD. Thus, the present study suggests that quantifying AT may be useful for the clinical assessment of subjects with PD. Acknowledgments: This study was supported by the Svenska Parkinson Förbundets Forskningsfond, the Swedish Medical Research Council (projects 5925 and 8667), the Göteborg Medical Society, the Neurologiskt Handikappades Riksförbund, and the VIDDA Foundation. Nobuaki Iwasaki was supported by grants from Wenner-Gren Center. The authors thank Torsten Olsson, Holger Broman, and Göran Steg for helpful comments on earlier drafts of the manuscript. REFERENCES 1. Gordon AM. Object release in patients with Parkinson s disease. Neurosci Lett 1997;232: Gordon AM. Task-dependent deficits during object release in Parkinson s disease. Exp Neurol 1998;153: Gordon AM, Ingvarsson PE, Forssberg H. Anticipatory control of manipulative forces in Parkinson s disease. Exp Neurol 1997;145: Ingvarsson PE, Gordon AM, Forssberg H. Coordination of manipulative forces in Parkinson s disease. Exp Neurol 1997;145: Fellows SJ, Noth J, Schwarz M. Precision grip and Parkinson s disease. Brain 1998;121: Forssberg H, Kinoshita H, Eliasson AC, Johansson RS, Westling G, Gordon AM. Development of human precision grip. II: anticipatory control of isometric forces targeted for object s weight. Exp Brain Res 1992;90: Eliasson AC, Gordon AM, Forssberg H. Impaired anticipatory control of isometric forces during grasping by children with cerebral palsy. Dev Med Child Neurol 1992;34: Parkinson J. An Essay on the Shaking Palsy. London: Whittingham and Rowland, Marsden CD. Origins of normal and pathological tremor. In: Findley LJ, Capildeo R, eds. Movement Disorders: Tremor. New York, NY: Oxford University Press, 1984: Barbeau A. Parkinson s disease: clinical features and etiopathology. In: Vinken PJ, Bruyn GW, Klawans HL, eds. Handbook of Neurology, vol 5: Extrapyramidal disorders. Amsterdam: Elsevier, 1986: Hunker CJ, Abbs JH. Uniform frequency of parkinsonian resting tremor in the lips, jaw, tongue, and index finger. Mov Disord 1990;5: De Jong H. Action-tremor. J Nerv Ment Dis 1926;64: Findley LJ, Gresty MA, Halmagyi GM. Tremor, the cogwhell phenomenon and clonus in Parkinson s disease. J Neurol Neurosurg Psychiatry 1981;44: Stein RB, Lee RG. Tremor and clonus. In: Brooks VB, ed. Handbook of Physiology, vol 2: The Nervous System II, Motor Control. Bethesda, MD: American Physiological Society, 1981: Cleeves L, Findley LJ, Marsden CD. Odd tremors. In: Marsden CD, Fahn S, eds. Movement Disorders 3. Oxford: Butterworth- Heinemann, 1994: Findley LJ, Gresty MA. Tremor and rhythmical involuntary movements in Parkinson s disease. In: Findley LJ, Capildeo R, eds. Movement Disorders: Tremor. London: Macmillan, 1984: Hefter H, Hömberg V, Reiners K, Freund H-J. Stability of frequency during long-term recordings of hand tremor. Electromyogr Clin Neurophysiol 1987;67: Lance JW, Schwab RS, Peterson EA. Action tremor and the cogwheel phenomenon in Parkinson s disease. Brain 1963;86: Teravainen H, Calne DB. Action tremor in Parkinson s disease. J Neurol Neurosurg Psychiatry 1980;43: Yanagisawa N, Nezu A. Pathophysiology of involuntary movements in Parkinson s disease. Eur Neurol 1987;26(suppl 1): Hadar U, Rose FC. Is parkinsonian arm tremor a resting tremor? Eur Neurol 1993;33: Bain P. A combined clinical and neurophysiological approach to

11 254 H. FORSSBERG ET AL. the study of patients with tremor [Editorial]. J Neurol Neurosurg Psychiatry 1993;56: Hoehn MM, Yahr MM. Parkinsonism: onset, progression, and mortality. Neurology 1967;17: Fahn S, Elton RL. Unified Parkinson s Disease Rating Scale. In: Fahn S, Marsden CD, Goldstein M, Calne DB, eds. Recent Developments in Parkinson s Disease, vol II. Florham Park, NJ: Mac- Millan, 1987: Eliasson AC, Gordon AM, Forssberg H. Basic coordination of manipulative forces in children with cerebral palsy. Dev Med Child Neurol 1991;33: Langston JW, Widner H, Goetz CG, et al. Core Assessment Program for Intracerebral Transplantations (CAPIT). Mov Disord 1992;7: Gordon AM, Reilmann R. Getting a grasp on research: does testing treatment taint testing of parkinsonian patients? Brain 1999;122: Schwarz M, Shaw L. Signal Processing. Discrete Spectral Analysis, Detection and Estimation. New York, NY: McGraw-Hill, Homberg V, Hefter H, Reiners K, Freund HJ. Differential effects of changes in mechanical limb properties on physiological and pathological tremor. J Neurol Neurosurg Psychiatry 1987;50: Findley LJ. Differential diagnosis and pharmacology. In: Jankovic J, Tolosa E, eds. Parkinson s Disease and Movement Disorders. Baltimore, MD: Williams and Wilkins, 1993: Stiles RN, Pozos RS. A mechanical-reflex oscillator hypothesis for parkinsonian hand tremor. J Appl Physiol 1976;40: Gresty M, Buckwell D. Spectral analysis of tremor: understanding the results. J Neurol Neurosurg Psychiatry 1990;53: Koller WC, Vetere-Overfield B, Barter R. Tremors in early Parkinson s disease. Clin Neuropharmacol 1989;12: Hagbarth KE, Young RR. Participation of the stretch reflex in human physiological tremor. Brain 1979;102: Elble RJ. Physiologic and essential tremor. Neurology 1986;36: Young RR. Physiological and enhanced physiological tremor. In: Findley LJ, Capildeo R, eds. Movement Disorders: Tremor. New York, NY: Oxford University Press, 1984: Edin BB, Westling G, Johansson RS. Independent control of human finger tip forces at individual digits during precision lifting. J Physiol (Lond) 1992;450: Burstedt MKO, Birznieks I, Edin BB, Johansson RS. Control of forces applied by individual fingers engaged in restraint of an active object. J Neurophysiol 1997;78: Draper IT, Johns RJ. The disordered movement in parkinsonism and the effect of drug treatment. Bulletin of the Johns Hopkins Hospital 1964;115: Flowers KA. Visual closed-loop and open-loop characteristics of voluntary movement in patients with parkinsonism and intention tremor. Brain 1976;99: Hallett M, Khoshbin S. A physiological mechanism of bradykinesia. Brain 1980;103: Wierzbicka MM, Wiegner AW, Logigian EL, Young RR. Abnormal most-rapid isometric contractions in patients with Parkinson s disease. J Neurol Neurosurg Psychiatry 1991;54: Flash T, Inzelberg R, Schechtman E, Korczyn AD. Kinematic analysis of upper limb trajectories in Parkinson s disease. Exp Neurol 1992;118: Vallbo AB, Wessberg J. Organization of motor output in slow finger movements in man. J Physiol (Lond) 1993;469: Wessberg J, Vallbo AB. Coding of pulsatile motor output by human muscle afferents during slow finger movements. J Physiol (Lond) 1995;485: Marshall J. Tremor. In: Vinken PJ, Bruyn GW, eds. Handbook of Neurology, vol. 6: Diseases of the basal ganglia. Amsterdam: North-Holland Publishing Co, 1968: Rushworth G. The gamma system in parkinsonism. Int J Neurol 1961;2: Martinez-Martin P. Rating scales in Parkinson s disease. In: Jankovic J, Tolosa E, eds. Parkinson s Disease and Movement Disorders. Baltimore, MD: Williams and Wilkins, 1993: Jankovic J, Frost JDJ. Quantitative assessment of parkinsonian and essential tremor: clinical application of triaxial accelerometry. Neurology 1981;31: Jankovic J. Pathophysiology and clinical assessment of motor symptoms in Parkinson s disease. In: Koller WC, ed. Handbook of Parkinson s Disease. New York, NY: Marcel Dekker, 1987: