EXPERIMENTAL NEUROLOGY ARTICLE NO. 146, 159–170 (1997) EN976507 Parkinsonism Reduces Coordination of Fingers, Wrist, and Arm in Fine Motor Control Hans-Leo Teulings, José L. Contreras-Vidal, George E. Stelmach, and Charles H. Adler* Motor Control Laboratory, Arizona State University, Tempe, Arizona 85287-0404; and *Mayo Clinic, Scottsdale, Arizona 85259 This experiment investigates movement coordination in Parkinson’s disease (PD) subjects. Seventeen PD patients and 12 elderly control subjects performed several handwriting-like tasks on a digitizing writing tablet resting on top of a table in front of the subject. The writing patterns, in increasing order of coordination complexity, were repetitive back-and-forth movements in various orientations, circles and loops in clockwise and counterclockwise directions, and a complex writing pattern. The patterns were analyzed in terms of jerk normalized for duration and size per stroke. In the PD subjects, back-and-forth strokes, involving coordination of fingers and wrist, showed larger normalized jerk than strokes performed using either the wrist or the fingers alone. In the PD patients, wrist flexion (plus radial deviation) showed greater normalized jerk in comparison to wrist extension (plus ulnar deviation). The elderly control subjects showed no such effects as a function of coordination complexity. For both PD and elderly control subjects, looping patterns consisting of circles with a left-to-right forearm movement, did not show a systematic increase of normalized jerk. The same handwriting patterns were then simulated using a biologically inspired neural network model of the basal ganglia thalamocortical relations for a control and a mild PD subject. The network simulation was consistent with the observed experimental results, providing additional support that a reduced capability to coordinate wrist and finger movements may be caused by suboptimal functioning of the basal ganglia in PD. The results suggest that in PD patients fine motor control problems may be caused by a reduced capability to coordinate the fingers and wrist and by reduced control of wrist flexion. r 1997 Academic Press INTRODUCTION Parkinson’s disease (PD) is caused by the degeneration of nigrostriatal neurons resulting in a reduction of the neurotransmitter dopamine (25, 26). Apart from the well-known movement control problems, slowness, reduced movement amplitudes, and prolonged decelera- tion times, PD patients are hypothesized to suffer from difficulties in the coordination and control of various muscle systems (15). For example, PD patients show a delay of the onset of the opening of the hand relative to the initiation of the transportation of the forearm (6). Whereas normal subjects can smoothly modify an ongoing movement, PD patients initiate a corrective movement only after completing the initial movement (15). Temporal dissociation has also been observed between the left and the right arms in PD patients (22). When fingers and thumb of the same arm were analyzed, Benecke et al. (2) found that PD subjects showed a substantial impairment when performing an isotonic elbow flexion while isometrically squeezing a force transducer. Similarly, Isenberg and Conrad (20) observed that PD subjects do not initiate components of arm movements simultaneously, resulting in angular or curved movement trajectories. These data suggest that in PD patients coordination is reduced in movement patterns that require control of a large number of muscles and joints. PD leads to a disruption in the execution of practiced skills such as handwriting (25, 41). Boisseau et al. (3) observed that PD handwriting can be characterized by various types of dysfluencies: lack of control, abrupt changes of direction, tremor, slowness, hesitation, rigidity, variability of baseline, and, in some cases, micrographia. The breakdown of handwriting may come from the inadequate control of acceleration amplitude (41), which is also reflected by abnormal EMG amplitudes (42) and by reduced coordination of independent movement components (20). Handwriting-like movement patterns are ideal tasks to study motor control because they are well learned. Many potential artifacts due to muscle strength limitations, gravity, inertia, visual feedback, and insufficient practice are absent. More importantly, coordination complexity can be varied by using handwriting patterns involving different combinations of finger, wrist, and arm movement components. In Western cursive handwriting, finger movements (i.e., flexion and extension of the thumb and the opposing index and middle fingers) generate primarily up-and-down strokes (i.e., 159 0014-4886/97 $25.00 Copyright r 1997 by Academic Press All rights of reproduction in any form reserved. 160 TEULINGS ET AL. away and toward the body in the horizontal plane) while wrist flexions and extension (plus radial and ulnar deviations) generate primarily the small left-andright movements. Forearm movements generate primarily the large left-to-right progression (43). Therefore, coordination complexity can be manipulated by selecting different handwriting patterns. For example, horizontal back-and-forth movements require mainly the wrist joint, vertical back-and-forth movements require mainly the finger joints, and oblique up-and-down strokes require a combination of finger and wrist movements that are in phase (e.g., 34). Circles require independent (i.e., out of phase) control of the fingers and wrist. Loops require an additional left-to-right progression by the forearm, which further magnifies the coordination complexity, irrespective of whether the progression is continuous (e.g., 17, 35) or discontinuous (e.g., 4). Reduced coordination is defined here as improperly timed initiation and disproportionate activation of independent muscle systems resulting in multiple acceleration peaks which extend movement duration. Such movement problems contribute to increased jerk levels because jerk is the change of acceleration per time (e.g., 15). However, because jerk level depends on the size and the duration of the movements it needs to be normalized. The advantage of normalized jerk is that coordination difficulties in patterns of different shapes, sizes, and durations can be compared. We hypothesized that normalized jerk per handwriting stroke increases with the number of degrees of freedom to be coordinated (e.g., wrist, fingers, wrist plus fingers, and wrist plus fingers and arm), particularly in PD subjects relative to the elderly subjects. To help understand our experimental data, writing patterns were also simulated by a biologically plausible model of normal versus PD basal ganglia thalamocortical interactions (7–9). EXPERIMENT Subjects Seventeen PD patients (14 males and 3 females; ages 42–78 years, mean age is 65 years) and 12 elderly control subjects (7 males and 5 females; ages 53–78 years, mean age is 67) participated in this study after providing informed consent. PD and elderly subjects were paid for their participation. All PD subjects were in mild stages of the disease. All but one reported that they were unilaterally affected (8 right sided, 8 left sided, and 1 bilateral). The median duration of PD since diagnosis was 3 years. Ten PD patients reported having bradykinesia and 9 reported having rigidity. Three PD patients showed action tremors in their handwriting with frequencies between 6.3 and 6.8 Hz. Nearly all PD patients reported micrographic writing impairments; 2 did not. All subjects were right handed except 1 PD patient, who wrote with the left hand. These selfreports were not significantly correlated with stroke duration, size, or normalized jerk. Apparatus The subjects wrote on a digitizer–display (Wacom PL-100V) controlled by an Intel 80486-based personal computer. The digitizer sampled the x and y coordinates of the pen tip 200 times per second with a spatial error of 0.05 mm. The digitizer–display (28 cm wide 3 23 cm high 3 1.5 cm thick) rested horizontally on top of a table and was oriented to meet each subject’s preference. Writing Patterns and Instructions Nine writing patterns were used: back-and-forth strokes in four different orientations—left-and-right (i.e., horizontal), away-and-toward the body (i.e., vertical), and two oblique orientations (i.e., forward slanted and backward slanted, respectively); continuous circles in clockwise and counterclockwise directions; repetitive loops in clockwise and counterclockwise directions (i.e., cursive ‘‘jjjjjjjj’’ and ‘‘llllllll’’); and a complex writing pattern (i.e., cursive ‘‘ljielije’’). Subjects were asked to suppress dotting of the ‘‘i’’ and ‘‘j.’’ The left-to-right progression is estimated in the loops by comparing sequences of two loops (requiring no progression), four loops, and eight loops (requiring substantial progression). The effects due to different numbers of strokes can be estimated separately by comparing all nonprogressing patterns when performing two, four, and eight ‘‘letters.’’ The subjects were instructed not to count the number of repeated strokes while producing the patterns. As a result, some trials could have fewer or more strokes than required. The experimenter gave feedback when the number of strokes was out of range so that the trial could be redone. Once a comfortable position was adopted the subjects were requested to keep the orientation of their forearm and the digitizer constant. The angles of the forearm and of the digitizer relative to the table front were recorded. All subjects spontaneously positioned their writing arm perpendicular (615°) to the horizontal baseline. Therefore, the horizontal strokes are mainly performed by the wrist joint and the vertical strokes by the finger joints. Oblique strokes were produced by both fingers and wrist. The handwriting task was shown as standard-font letters on the display screen to motivate subjects to preprogram the movement patterns rather than copying. The subjects were asked to write at comfortable 161 COORDINATION IN PARKINSONIAN HANDWRITING size and speed. No speed instruction was given because elderly controls are likely to change their writing more than PD subjects (41). The writing area of 18 by 10 cm was centered on the digitizer and had a horizontal guide line 3 cm from the bottom. Subjects were instructed to write on this guide line. Procedure Subjects performed three blocks of 4-, 8-, and 16stroke patterns in a random sequence. Within each of these blocks the nine patterns were performed in a random sequence. Each pattern was replicated at least six times (three times by the three slowest parkinsonians) so that the duration of the experiment session was about equal for all subjects (i.e., 1 h). Each time a new pattern was to be produced for the first time the experimenter showed an example of how the pattern was to be performed. By blocking the replications and pattern-length conditions, task switching was minimized. Task switching would have selectively disadvantaged PD patients (11). The subjects were familiarized with the equipment by writing their name and a sentence (‘‘we write llll in arizona’’) two times to check for fatigue or medication effects during the experimental session. Recording of the writing movements started as soon as the pen touched the surface of the digitizer–display and ended when the pen was lifted for more than 0.5 s (2 s when writing the name or the sentence). The trace of the pen position was made visible on the digitizer– display in real time. A few seconds after the previous trial, the subject started with the next trial. If the experimenter or subject discovered that a trial was inappropriate, it was redone. Simulations Handwriting patterns similar to those used in the experiment were also generated using a neural network of parkinsonian and normal handwriting movements (7–9). In the handwriting simulations, a sequence of relative target position vectors was defined according to the individual stroke direction and amplitude for each degree of freedom as follows: finger flexion/extension producing vertical displacement (TPVy ), forearm supination/pronation producing local horizontal displacement (TPVx ), and radial flexion/ ulnar extension of the wrist joint producing the left-toright progression (TPVa ). At times of zero or peak velocity the subsequent target position vector of the motor program was fed into the VITE model, which continuously computed the difference vector (DVx, DVy, DVa ) between the target position (TPVx, TPVy, TPVa ) FIG. 1. Each basal ganglia–thalamocortical circuit controls 1 of the 3 degrees of freedom in the handwriting simulations. The basal ganglia modulate the dynamics of a central movement generator (VITE) by gating difference vector (DV) computations between target position vector (TPV) and present position vector (PPV) at a rate specified by the pallidothalamic neurons. Dopamine depletion in Parkinson’s disease has a differential effect on the direct and indirect pathways of the basal ganglia. In particular, an overactivation of striatal neurons in the indirect pathway causes inhibition of neurons of the external segment of the globus pallidus (GPe) which in turn disinhibits neurons in the subthalamic nucleus (STN). The disinhibition of STN cells causes an overactivation of neurons in the internal segment of the globus pallidus (GPi). Furthermore, a reduced activation of striatal cells in the direct pathway further increases GPi activity by disinhibition. This pathological increase in GPi activity causes inhibition of thalamic neurons that project to motor cortical areas necessary for movement production. During normal motor behavior the direct pathway is activated to initiate movement at a desired speed while the indirect pathway may be activated to terminate the movement (7). (see Appendix) and the present position vectors (PPVx, PPVy, PPVa ). The sequence of present position vectors forms the trajectory of the pen tip. The horizontal and vertical coordinates of the pen tip are derived from the present position vector (PPVx, PPVy, PPVa ) as follows (4), x 5 PPVx 3 cos (PPVa) 1 (length 1 PPVy) 3 sin (PPVa), y 5 2PPVx 3 sin (PPVa) 1 (length 1 PPVy) 3 cos (PPVa), 162 TEULINGS ET AL. FIG. 2. Examples of handwriting patterns used in the experiment, performed by an elderly control. The writing patterns were (a) horizontal back-and-forth strokes, (b) back-and-forth strokes slanted forward, (c) vertical back-and-forth strokes, (d) back-and-forth strokes slanted backward, (e) counterclockwise circles, (f ) clockwise circles, (g) a sequence of counterclockwise loops, ‘‘llllllll,’’ (h) a sequence of clockwise loops, ‘‘jjjjjjjj,’’ and (i) a complex writing pattern, ‘‘ljielije.’’ The dots mark the segmentation points between strokes. For each of the patterns we show the vertical (in the back-and-forth movements, upward or leftward along the orientation) velocity V as a function of time and the normalized jerk level per stroke. The horizontal back-and-forth movements are performed by alternating flexion plus ulnar deviation and extension plus radial deviation of the wrist joint. The normalized jerk is rather constant for successive strokes. The patterns were clipped after 2.5 s. where length 5 7.5 cm, represents the distance between wrist joint and pen tip.1 The repetitive patterns are generated by repeating the basic sequence of relative target positions. Withinsubject motor variability was simulated by adding random noise to the target positions. The noise was equal for both simulated PD and control, namely uniformly distributed noise between minus and plus 0.025 cm (for TPVx and TPVy ) and between plus and 1 The biomechanical models used in the simulation and in Wing (43) are not substantially different. In the simulation model the wrist flexion and extension has the same function as the arm movement in Wing (43). Wrist rotation has the same function as wrist extension and flexion in the aforementioned model. Note that both models include 3 degrees of freedom, which are involved in a similar hierarchy with coordination complexity. minus 3° (for TPVa ). The dynamics of the central pattern generator are modulated by a neural model of the basal ganglia–thalamus circuit which gates the output of the difference vector (see Fig. 1). Therefore, the network specifies the onsets and the speeds of each of the degrees of freedom. In the simulation of the PD subject, the dopamine depletion was set to 15%, corresponding to preclinical PD. In the simulation of the control subject the dopamine depletion was set to 0%. The reason that, in the model, a small dopamine depletion shows a measurable effect, whereas in PD patients depletion levels of 80% are required to show any effects is probably because the model lacks the compensatory mechanisms of the nervous system such as presynaptic overactivity and supersensitivity of receptors (32). The dynamical equations specifying the COORDINATION IN PARKINSONIAN HANDWRITING 163 FIGURE 2—Continued network model are explained in detail in ContrerasVidal and Stelmach (7). One hundred twenty trials of back-and-forth strokes were generated using the PD model and as many using the elderly control model. Half of them started with an upstroke and the other half started with a downstroke. This represents the amount of data generated by 20 PD subjects and 20 elderly subjects. Analysis The handwriting recordings and simulated patterns were filtered using a frequency-domain attenuation filter with a pass band between 0 and 3 Hz and a cosine transition band between 3 and 11 Hz in all subjects (39). Proper filtering was required for calculating higher order time derivatives such as jerk. To verify that the results were not caused by a particular choice of filter frequency all data were processed using a higher (3.5–12.5 Hz) and a lower (2.5–9.5 Hz) transition band. The qualitative results of this study appeared independent of the low-pass filter frequency. After filtering, the patterns were segmented into alternating up- and downstrokes where the interpolated vertical velocity crosses 0. Spurious segmentation points, such as segmentation points spaced less than 0.5 mm, were removed. Back-and-forth movement patterns were rotated to a vertical orientation: 90° counterclockwise for horizontal strokes and 45 or 245° for oblique forward or backward strokes, respectively. Subsequently, a pattern verifier was used to check the correct performance and the segmentation of each individual pattern. The first stroke was removed if it was a downward or a leftward stroke or an incomplete stroke, so that all processed patterns were aligned with a beginning 164 TEULINGS ET AL. FIG. 3. The same as Fig. 2 but from a PD patient instead of an elderly control. The wrist flexions show greater normalized jerk levels in nearly all PD subjects in comparison to the wrist extensions. The patterns were clipped after 5 s. upward or a rightward stroke. Trials with less than 2, 6, or 12 strokes in the 4-, 8-, and 16-stroke conditions, respectively, were discarded. Trials were also discarded when ‘‘e’’ or ‘‘l’’ had no counterclockwise loop or ‘‘j’’ had no clockwise loop, or when the loop areas were smaller than one of the spurious loops at the movement reversals of ‘‘i’’ and ‘‘j.’’ Finally, the complex pattern ‘‘ljielije’’ was checked for the proper sequence of stroke sizes. Replications in which the vertical sizes of ‘‘l’’ or ‘‘j’’ were smaller than the vertical sizes of a subsequent ‘‘i’’ or ‘‘e’’ were discarded. In both the elderly controls and the PD subjects 10% of the trials were discarded, leaving 5.8 and 5.5 appropriate replications per condition, respectively. The measure for fine motor control impairment adopted here is based on the integrated squared jerk (e.g., 15). Jerk is the change of acceleration, which is the third time derivative of position, and possesses a unit of length/duration3. The time-integrated squared jerk is minimal in smooth movements (19). Because jerk varies dramatically with duration and size of the stroke (e.g., 33) and because the present study attempted to measure movements under preferred stroke durations and sizes, jerk has to be normalized for different stroke durations and sizes. This was done by dividing integrated squared jerk by length2/duration5 per stroke (e.g., 21). Subsequently, the square root was taken so that normalized jerk is proportional with absolute jerk: Œ(1⁄2 e dt j 2(t) 3 duration5/length2). Normalized jerk is a unit-free measure and can be compared between strokes of different sizes and durations. For example, a harmonic back-and-forth movement yields a normalized jerk score equal to p3/22 5 7.75. This is smaller than the normalized jerk of a half-constant-velocity circular movement, which is p3/ 21.5 5 10.96, thus reflecting the higher cost of producing a curved stroke than a straight stroke. Data across replications within subjects were averaged. This generated better results than medians, COORDINATION IN PARKINSONIAN HANDWRITING 165 FIGURE 3—Continued probably because the number of replications is small (e.g., 5) so that the medians per stroke may not follow the general trend. The resulting data per subject were in general non-Gaussian (e.g., outliers tended to be far too large) and showed nonuniform variances. Therefore, the nonparametric Mann–Whitney U or Wilcoxon test between elderly and PD groups was used (e.g., 5, 13). For tests between conditions within groups the sign test for paired data with a priori probability of 0.5 was performed. RESULTS The left-hand side of Fig. 2 shows the actual handwriting-like patterns produced by a control subject: backand-forth movements in (a) horizontal, (b) forward oblique, (c) vertical, and (d) backward oblique direc- tions, counterclockwise and clockwise (e and f ) circles and (g and h) loops, and (i) the complex writing pattern. The scales are calibrated in centimeters. The right side of Fig. 2 shows the vertical velocity V (or the velocity along the upward or rightward directions in the straightstroke patterns) as a function of time and the normalized jerk per stroke (N. Jerk). Only the first 2.5 s of the movement are shown. The positive phases of the velocity refer to upward or rightward movements and the negative phases to downward or leftward movements. Each positive or negative phase is called a stroke (see Fig. 2 legend). The normalized jerk is plotted as a bar chart in which the bar widths are equal to the stroke durations so that the bar chart is aligned with the velocity patterns per stroke. The graph shows that multimodal velocity patterns, which we think signify coordination deficits, yield high values for normalized 166 TEULINGS ET AL. FIG. 4. Normalized jerk per stroke in back-and-forth movements as a function of stroke direction (a) for PD and elderly subjects and (b) for a simulated PD and a control subject. Vertical bars indicate the standard deviations of the means. To discriminate the vertical bars per subject group, the curves have been slided in horizontal direction. For each direction the predominant finger and wrist activities are given. The PD subjects show higher normalized jerk, especially in the oblique stroke directions requiring the coordination of wrist and fingers. Furthermore, PD subjects show higher normalized jerk in wrist flexions than in wrist extensions. jerk. The straight-stroke patterns (a–d) show uniformly low normalized jerk levels close to the minimum of 7.75. The circular patterns (e–i) show slightly higher normalized jerk levels of about 10.96. The complex pattern (i) shows normalized jerk levels which are generally higher. Figure 3 shows the same patterns executed by a PD subject. It can be seen that the PD subject produced the movements at a lower rate. Therefore, the time scale was extended to 5 s. The PD subject generally shows greater normalized jerk per stroke than the elderly subject. Furthermore, the normalized jerk per stroke differed largely between successive strokes. Specifically, Fig. 3a shows that normalized jerk levels in the wrist flexions were systematically higher than in the wrist extensions. The normalized jerk of the back-and-forth strokes are depicted in Fig. 4a as a function of stroke direction. In the PD subjects, strokes with oblique orientations have greater normalized jerk than the horizontal and vertical strokes. This effect cannot be explained by different stroke durations as the stroke orientation effect consists only of the backward oblique strokes being much slower than other strokes in both parkinsonians and the elderly (see the mean durations in the back-and-forth movements in Table 1). The orientation effect upon normalized jerk cannot be explained by different stroke sizes, either, as the stroke size effect consists of the vertical strokes being smaller while the horizontal strokes are larger than the oblique strokes in both PD and the elderly subjects (see the mean TABLE 1 Means and Standard Deviations of Duration, Size, and Normalized Jerk per Stroke in the Back-and-Forth Strokes in the Elderly and PD Subjects Duration (ms) Size (cm) Normalized jerk Pattern/Stroke Mean SD Mean SD Mean Elderly control subjects (a) Horizontal back-and-forth strokes Left 247 117 1.01 0.31 14.9 Right 255 106 1.03 0.31 14.6 (b) Oblique slanted forward back-and-forth strokes Upward 240 106 0.96 0.24 14.1 Downward 247 116 0.96 0.24 15.9 (c) Vertical back-and-forth strokes Upward 247 127 0.81 0.27 14.4 Downward 237 125 0.82 0.27 13.1 (d) Oblique slanted backward back-and-forth strokes Upward 294 144 0.96 0.32 17.2 Downward 268 132 0.96 0.33 15.8 PD subjects (a) Horizontal back-and-forth strokes Left 286 166 1.11 0.42 17.3 Right 306 182 1.11 0.42 19.8 (b) Oblique slanted forward back-and-forth strokes Upward 288 168 0.99 0.28 20.0 Downward 303 191 1.00 0.27 19.6 (c) Vertical back-and-forth strokes Upward 282 169 0.93 0.34 16.9 Downward 286 178 0.95 0.33 16.8 (d) Oblique slanted backward back-and-forth strokes Upward 341 221 0.98 0.39 23.1 Downward 346 236 0.99 0.38 24.6 SD 10.8 9.5 8.2 11.9 7.5 7.1 11.2 12.9 14.4 16.3 16.0 16.4 16.5 15.3 23.8 28.5 COORDINATION IN PARKINSONIAN HANDWRITING stroke sizes in the back-and-forth movements in Table 1). Therefore, the normalization of jerk for stroke size and duration differences is appropriate. Strokes in oblique directions that require the coordination of fingers and wrist had significantly greater normalized jerk than the strokes using only the wrist joint or only the finger joints, sign test(16) 5 4, P , 0.05. This especially holds for the stroke directions in which fingers flex while wrist extends and vice versa. In the elderly controls, the normalized jerk difference between oblique and other directions was much less pronounced and nonsignificant. In addition, PD patients showed significantly greater normalized jerk in the single-joint wrist flexions (plus ulnar deviation) than in the extensions (plus radial deviation), sign test(16) 5 3, P , 0.05 (see also Fig. 3a). This is supported by the observation that one of the PD subjects included in the data who wrote with his left hand showed greater normalized jerk in flexion than extension as wrist flexion in the left hand results in a stroke in the direction opposite to that of the right hand. In contrast to the PD subjects, the elderly control subjects showed no differences between wrist flexion and wrist extension (see Fig. 3a). The additional effect of the left-to-right progression component was estimated by comparing circles and loop patterns. Table 2 summarizes stroke duration, size, and normalized jerk in the circular patterns. As can be seen in Fig. 5, the counterclockwise circles and the ‘‘llllllll’’ patterns did not differ for both PD and elderly control groups, whereas the ‘‘jjjjjjjj’’ pattern showed greater jerk than the clockwise circles and this was true for both PD and elderly subjects, sign test(16) 5 1, P , 0.01 and sign test(10) 5 1, P , 0.05, TABLE 2 Means and Standard Deviations of Duration, Size, and Normalized Jerk per Stroke in the Curved Patterns in the Elderly and PD Subjects Duration (ms) Pattern Elderly control subjects (e) Circles counterclockwise (f ) Circles clockwise (g) Loops counterclockwise (h) Loops clockwise (i) Complex writing pattern PD subjects (e) Circles counterclockwise (f ) Circles clockwise (g) Loops counterclockwise (h) Loops clockwise (i) Complex writing pattern Size (cm) Mean SD Mean SD Normalized jerk Mean SD 247 262 257 332 297 131 165 122 144 177 0.62 0.67 0.68 0.86 0.50 0.36 0.36 0.24 0.36 0.30 15.1 16.9 13.7 22.9 30.9 6.7 12.6 6.4 14.6 64.0 278 321 306 417 335 128 170 167 202 158 0.72 0.74 0.74 0.88 0.48 0.24 0.24 0.22 0.34 0.27 17.8 23.1 19.6 35.6 32.6 12.3 21.1 16.5 29.4 49.4 167 FIG. 5. Normalized jerk per stroke for counterclockwise and clockwise circles and counterclockwise and clockwise loop sequences (i.e., cursive ‘‘llllllll’’ and ‘‘jjjjjjjj,’’ respectively) and the complex pattern ‘‘ljielije’’ for PD subjects and elderly controls. Vertical bars indicate the standard deviations of the means. To discriminate the vertical bars per subject group, the curves have been slided in horizontal direction. respectively. Therefore, there is no systematic additional effect of left-to-right progression of the forearm. In a separate analysis, instead of averaging across the first 12 strokes, the first up-and-down strokes in ‘‘llllllll’’ and in ‘‘ljielije’’ (i.e., ‘‘l’’) were compared. Similarly, the second down-and-up strokes in ‘‘jjjjjjjj’’ and ‘‘ljielije’’ (i.e., ‘‘j’’) were compared. The first ‘‘l’’ showed greater normalized jerk in the complex pattern than in the repetitive pattern both in the elderly subjects, sign test(11) 5 1, P , 0.05, and in the PD subjects, sign test(15) 5 3, P , 0.05, but the first ‘‘j’’ did not show any difference. Therefore, the PD subjects showed no differences with increasing task complexity. Detailed analysis showed that serial stroke position (i.e., 1 to 12) in the repetitive patterns did not affect normalized jerk nor size or duration per stroke. Furthermore, pattern length (i.e., 4, 8, or 16 strokes) had no effect upon the duration, size, or normalized jerk of the first 2 strokes in either PD or elderly group. Back-and-forth movements were also simulated using a biologically plausible model of movement production in normal and PD subjects (see Simulations; 7). The movements produced by the simulations have been analyzed analogously to the experimental data. The results are shown in Fig. 4b. As observed in the experimental data, the PD simulation showed markedly elevated normalized jerk in oblique strokes requiring coordination between wrist and fingers. However, the asymmetry between wrist flexion and extension observed in the PD subjects but not in the elderly was not replicated in the simulations as the model is concerned with differences between extension and flexion and therefore the outputs were perfectly symmetric for flexion and extension. 168 TEULINGS ET AL. DISCUSSION This research suggests that some of the fine motor control problems in PD patients are caused by a reduced capability to coordinate the fingers and wrist and by reduced control of wrist flexion. The control problems occur in handwriting strokes that require the coordination of wrist and fingers (i.e., strokes in oblique orientations) compared to strokes executed by the wrist or the fingers alone (i.e., strokes in left–right or up– down orientations). Coordination difficulties were expressed in terms of normalized jerk, which is a measure of the rate of change in acceleration. The integrated squared jerk increases with excessive accelerations and decelerations as occurring in suboptimal coordination. Integrated jerk was normalized to allow comparison between strokes of various durations and sizes. Superimposed on the stroke direction effect, PD subjects show greater normalized jerk in wrist flexions plus ulnar deviation than in wrist extensions plus radial deviation, whereas elderly control subjects show no difference. The difference between extension and flexion may be due to increased tonic activation of flexor muscles in PD patients (28). Marginal evidence supported the notion that integrating forearm movements with wrist and finger movements causes increase of normalized jerk. Forearm movements are required in the left-to-right progression of series of loops. This conclusion was based upon the observation that normalized jerk of nonprogressing circles and progressing loops showed only effects for clockwise circles and loops. Loops do require additional coordination skills compared to circles. This was shown by Mai and Marquardt (24) who observed that specific brain-damaged patients perform pairs of circles well, whereas pairs of loops were substantially impaired. The present results indicated no change of coordination during repetitive patterns. A change of normalized jerk would have been expected based on Martin et al. (27). They investigated 10-cm-large zigzags on smooth sheets with and without targets visible and found a small progressive slowing and variability increase in the PD subjects. They observed an increase of the serial position feedback in conditions where the targets are not visible, suggesting that the PD subjects are forced to rely more on position feedback, which is impaired. Stelmach and Castiello (36) studied PD and control subjects performing sequences of cursive ‘‘e’’ and ‘‘l’’ and observed that both stroke size and stroke duration progressively decreased. Stelmach et al. (38) found that PD subjects showed progressively increasing errors. In contrast, reports exist that PD subjects may also speed up repetitive movements. Finger tapping seems to cause movement hastening in PD subjects (29, 30). A mixture of opposing effects may have canceled a serial position effect. The number of strokes of a pattern (i.e., 4, 8, or 16) and the context (e.g., ‘‘l’’ in a repetitive versus nonrepetitive context) had little influence upon normalized jerk. Possibly, both PD and elderly subjects preprogrammed only 1 or 2 strokes ahead independently of the number of strokes to follow or the complexity of the strokes. The handwriting patterns generated by the simulations show effects qualitatively similar to the observed experimental data. The neural network used only information about the beginning and target positions of each stroke and its speed so that the simulations were not expected to show any differences between wrist flexion and extension. Although the amount of noise added to the target positions was the same for PD and normal movements, the simulations replicated the experimental finding that normalized jerk (i.e., motor output variability) was larger in PD than in the normal case. In the present simulations, each degree of freedom (i.e., finger, wrist, and arm) was controlled by separate basal ganglia–thalamocortical circuits (1, 7) (see Fig. 1). This is consistent with Hoover and Strick (18) who found that separate regions of the internal segment of the globus pallidus (GPi) project to distinct cortical motor areas, namely, the primary motor cortex, the supplementary motor area, and the ventral premotor area. The model presented here assumes an even greater level of segregation: parallel channels from the GPi to the motor cortex controlling individual degrees of freedom. This assumption is corroborated by singlecell studies of the primate motor cortex and putamen, showing that neurons in these areas are related to active rather than passive movements in one direction for a single joint movement (12). This suggests that the basal ganglia may also be involved in joint coordination. In the present simulations, small levels of dopamine depletion caused slightly smaller and shorter than normal pallidothalamic signals which in turn modulate the dynamics of a central movement generator (the VITE model). Due to the altered neurotransmitter dynamics in the PD model, the basal ganglia output signals show greater than normal variability causing an increased jerk in strokes requiring multiple degrees of freedom (7). In the design of the handwriting tasks and in our handwriting model, we assumed that fingers, wrist, and arm generate specific components of the writing movements. The fingers produce the vertical movement as in up–down strokes, toward and away from the body in the horizontal plane. The wrist produces the local horizontal movement as in left–right strokes. The forearm produces the left-to-right horizontal progression as in extended horizontal lines of writing (40). Some handwriting theorists have proposed that the left-to-right progression in loops occurs steadily during 169 COORDINATION IN PARKINSONIAN HANDWRITING a circular motion (e.g., 17, 35), whereas others have suggested a discontinuous progression that takes place mainly during the connecting stroke (e.g., 4). Nevertheless, it is safe to suppose that adding more components to the handwriting movement should increase the complexity. In several instances the present experiment showed no significant effects where the previous literature would have predicted effects. The absence of effects may be a result of the design of the experiment which was geared toward eliminating potential artifacts. Three of those potential artifacts are discussed here. (a) The subjects were not artificially limited in their degrees of freedom because this would be confounded with the experimental patterns having different numbers of the degrees of freedom. The disadvantage is that no certainty exists that the subjects used fingers, wrist, and arm exactly as assumed in the model. However, nearly all subjects spontaneously positioned their forearm approximately vertically relative to the baseline of the digitizer so that wrist and finger movements play roles as assumed. This assumption is supported by the large consistency among subjects in terms of larger normalized jerk in oblique directions. (b) In the present experiment no additional instructions or guidelines were presented because they would require extra attention to visual feedback resulting in additional on-line corrections. This could affect the PD and the elderly subjects differently because visual feedback is thought to play a greater role in PD subjects (16), because PD patients have greater uncertainty of limb position (10) than the elderly, and because PD patients have more difficulty initiating and executing nonrepetitive patterns (37). (c) Replications of trials were performed in a blocked fashion. This minimizes task switching which is a known problem specifically in PD subjects (11). Movement proficiency measures other than normalized jerk exist but they have several disadvantages. For example, the number of acceleration peaks per velocity peak (23) or the number of acceleration zero crossings (14, 31) can only assume integer values and do not exploit acceleration amplitude. Some other coordination measures do not yield a simple score but rather a pictorial characteristic, such as velocity versus acceleration diagrams (24). Timing and amplitude differences between the coordinated components yield nonstraight trajectories so that straightness error could also be used as a measure for coordination problems. Isenberg and Conrad (20) found that PD patients initiate arm movement components nonsimultaneously, resulting in curved and angular movement trajectories. Duration per stroke may be used as a measure of movement difficulty. Duration per stroke is correlated with normalized jerk because a slow movement is likely to have multiple acceleration peaks which result in a greater normalized jerk. However, in contrast to jerk effects, no significant stroke duration effects were observed. In summary, our data indicate that PD patients, when compared to the elderly controls, show reduced capability to coordinate the wrist and fingers in handwriting-like tasks. In addition, wrist flexion showed more irregular acceleration profiles than wrist extension. Therefore, we conclude that coordination impairments in PD patients can be detected in finger and wrist movements and even in flexions and ulnar deviations of the wrist which may contribute to the handwriting impairments observed in PD patients. APPENDIX The sequences of relative target position vectors (TPVx, TPVy, TPVa) for the simulations of the straight back-and-forth strokes were: (a) horizontal (0.47, 0.003, 0) (0, 0, 0) (20.47, 0.003, 0) (0, 0, 0), (b) oblique slanted forward (0.33, 0.33, 0) (0, 0, 0) (20.33, 0.33, 0) (0, 0, 0), (c) vertical (0.003, 0.47, 0) (0, 0, 0) (20.003, 20.47, 0) (0, 0, 0), and (d) oblique slanted backward (20.33, 0.33, 0) (0, 0, 0) (0.33,20.33, 0) (0, 0, 0). ACKNOWLEDGEMENTS This research was supported by NIH R01 NS 33173-01. The authors thank the Parkinson patients and the elderly subjects from Arizona, The Barrow’s Neurological Institute, and the Mayo Clinic Scottsdale for their willingness to participate and to cooperate. REFERENCES 1. 2. 3. 4. 5. 6. 7. Alexander, G. E., M. R. DeLong, and P. L. Strick. 1986. Parallel organization of functionally segregated circuits linking basal ganglia and cortex. Annu. Rev. Neurosci. 9: 357–381. Benecke, R., J. C. Rothwell, J. P. R. Dick, B. L. Day, and J. C. Marsden. 1986. Performance of simultaneous movements in patients with Parkinson’s disease. Brain 109: 739–757. Boisseau, M., G. Chamberland, and S. Gauthier. 1986. Handwriting analysis of several extrapyramidal disorders. [Presented at the meeting of the CSFS, Niagara Falls, Canada, September 1986] Bullock, D., S. Grossberg, and C. Mannes. 1993. A neural network model for cursive script production. Biol. Cybernet. 70: 15–28. Cassell, K., K. Shaw, and G. Stern. 1973. A computerized tracking technique for the assessment of parkinsonian motor disabilities. Brain 96: 815–826. Castiello, U., K. M. B. Bennett, and M. Scarpa. 1994. The reach to grasp movement of Parkinson’s disease subjects. Pages 215–237 in K. M. B. Bennett and U. Castiello, Eds., Insights into the Reach to Grasp Movement. North-Holland, Amsterdam. Contreras-Vidal, J. L., and G. E. Stelmach. 1995. A neural network model of basal ganglia–thalamocortical relations in 170 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. TEULINGS ET AL. normal and parkinsonian movement. Biol. Cybernet. 73: 467– 476. Contreras-Vidal, J. L., H. L. Teulings, and G. E. Stelmach. 1995. A network model of pathological cortical-subcortical interactions in abnormal handwriting. Pages 38–39 in the Proceedings of the 7th International Graphonomics Society Conference. London, Ontario. [August 1995, ISBN 0-921121-14-8] Contreras-Vidal, J. L., H. L. Teulings, and G. E. Stelmach. 1995. Micrographia in Parkinson’s disease. NeuroReport 6: 2089– 2092. Cooke, J. D., J. D. Brown, and V. B. Brooks. 1978. Increased dependence on visual information for movement control in patients with Parkinson’s disease. Can. J. Neurol. Sci. 5: 413–415. Cools, A. R., R. Jaspers, M. Schwarz, K. H. Sonntag, M. Vrijmoed-de Vries, and J. Van den Bercken. 1984. Basal ganglia and switching motor programs. Pages 513–543 in J. S. McKenzie, R. E. Kemm, and L. N. Wilcock, Eds., The Basal Ganglia: Structure and Function. Plenum, New York. Crutcher, M. D., and M. R. DeLong. 1984. Single cell studies in the primate putamen. Exp. Brain Res. 53: 233–243. Daniel, W. W. 1978. Applied Nonparametric Statistics. Houghton Mifflin, Boston. Eichhorn, T., G. Arnold, N. Mai, T. Gasser, and W. H. Oertel. 1994. Disturbance in the execution of automated handwriting movements: A sensitive marker for early Parkinsonism. Mov. Disord. 9: 121. Flash, T., R. Inzelberg, and A. D. Korsczyn. 1992. Quantitative methods for the assessment of motor performance in Parkinson’s disease. Pages 87–106 in F. Clifford Rose, Ed., Parkinson’s Disease and the Problems of Clinical Trials. Smith-Gordon, London. Flash, T., R. Inzelberg, E. Schechtman, and A. D. Korczyn. 1992. Kinematic analysis of upper limb trajectories in Parkinson’s disease. Exp. Neurol. 118: 215–226. Hollerbach, J. M. 1981. An oscillation theory of handwriting. Biol. Cybernet. 39: 139–156. Hoover, J. E., and P. L. Strick. 1993. Multiple output channels in the basal ganglia. Science 259: 819–821. Hogan, N., and T. Flash. 1987. Moving gracefully: Quantitative theories of motor coordination. Trends Neurosci. 10: 170–174. Isenberg, C., and B. Conrad. 1994. Kinematic properties in slow arm movements in Parkinson’s disease. J. Neurol. 241: 323– 330. Kitazawa, S., T. Goto, and Y. Urushihara. 1993. Quantitative evaluation of reaching movements in cats with and without cerebellar lesions using normalized integral of jerk. Pages 11–19 in N. Mano, I. Hamada, and M. R. DeLong, Eds., Role of the Cerebellum and Basal Ganglia in Voluntary Movement. Elsevier, Amsterdam. Lazarus, A., and G. E. Stelmach. 1992. Interlimb coordination in Parkinson’s disease. Mov. Disord. 7: 159–170. Maarse, F. J., R. G. J. Meulenbroek, H. L. Teulings, and A. J. W. M. Thomassen. 1987. Computational measures for ballistic handwriting. Pages 16–18 in R. Plamondon, C. Y. Suen, J. G. Deschenes, and G. Poulin, Eds., Proceedings of the 3rd International Symposium on Handwriting and Computer Applications. Ecole Polytechnique, Montreal. [ISBN 2-553-00197-5] Mai, N., and Ch. Marquardt. 1993. Analysis of handwriting movements in brain damaged patients. Pages 240–242 in 6th International Conference on Handwriting and Drawing. Telecom, Paris. Marsden, C. D. 1984. Which motor disorders in Parkinson’s disease indicate the true motor function of the basal ganglia? 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. Functions of the basal ganglia. Pages 225–241 in Ciba Foundation Symposium, Vol. 107. Pitman, London. Marsden, C. D., and J. A. Obeso. 1994. The function of the basal ganglia and the paradox of stereotaxic surgery in Parkinson’s disease. Brain 117: 877–897. Martin, K. E., J. G. Phillips, R. Iansek, and J. L. Bradshaw. 1994. Inaccuracy and instability of sequential movements in Parkinson’s disease. Exp. Brain Res. 102: 131–140. Mink, J. W., and W. T. Thach. 1991. Basal ganglia motor control. III. Pallidal ablation: Normal reaction time, muscle concentration, and slow movement. J. Neurophysiol. 65: 330–351. Nagasaki, H., and R. Nakamura. 1982. Rhythm formation and its disturbance: A study based upon periodic response of the motor output system. J. Hum. Ergol. 11: 27–142. Nakamura, R., H. Nagasaki, and H. Narabayashi. 1978. Disturbances of rhythm formation in patients with Parkinson’s disease. Part I. Characteristics of tapping response to the periodic signals. Percept. Mot. Skills 46: 63–75. Phillips, J. G., J. L. Bradshaw, E. Chiu, and J. A. Bradshaw. 1994. Characteristics of handwriting of patients with Huntington’s disease. Mov. Disord. 9: 521–530. Riederer, P., E. Sofic, C. Konradi, J. Kornhuber, H. Beckmann, M. Dieti, G. Moll, and G. Hebnestrait. 1989. The role of dopamine in the control of neurobiological functions. Pages 1–17 in P. Riederer et al., Eds., The Role of Brain Dopamine. Springer, Berlin. Schneider, K., and R. F. Zernicke. 1989. Jerk-cost modulations during the practice of rapid arm movements. Biol. Cybernet. 60: 221–230. Schomaker, L. R. B., A. J. W. M. Thomassen, and H. L. Teulings. 1989. A computational model of cursive handwriting. Pages 153–177 in R. Plamondon, C. Y. Suen, and M. Simner, Eds., Computer Recognition and Human Production of Handwriting. World Scientific, Singapore. Singer, Y., and N. Tishby. 1994. Dynamical encoding of cursive handwriting. Biol. Cybernet. 71: 227–237. Stelmach, G. E., and U. Castiello. 1992. Functional force control in Parkinson’s disease patients. Pages 401–423 in G. E. Stelmach and J. Requin, Eds., Tutorials in Motor Behavior II. North-Holland, Amsterdam. Stelmach, G. E., A. Garcia-Colera, and Z. E. Martin. 1989. Force transition control within a movement sequence in Parkinson’s disease. J. Neurol. 236: 406–410. Stelmach, G. E., C. J. Worringham, and E. A. Strand. 1987. The programming and execution of movement sequences in Parkinson’s disease. Int. J. Neurosci. 36: 55–65. Teulings, H. L., and F. J. Maarse. 1984. Digital recording and processing of handwriting movements. Hum. Mov. Sci. 3: 193– 217. Teulings, H. L., A. J. W. M. Thomassen, and F. J. Maarse. 1989. A description of handwriting in terms of main axes. Pages 193– 211 in R. Plamondon, C. Y. Suen, and M. Simner, Eds., Computer Recognition and Human Production of Handwriting. World Scientific, Singapore. Teulings, H. L., and G. E. Stelmach. 1991. Control of stroke size, peak acceleration, and stroke duration in parkinsonian handwriting. Hum. Mov. Sci. 10: 315–333. Wierzbicka, M. M., A. W. Wiegner, E. L. Logigian, and R. R. Young. 1991. Abnormal most-rapid contraction in patients with Parkinson’s disease. J. Neurol. Neurosurg. Psychiatry 54: 210– 216. Wing, A. M. 1978. Response timing in handwriting. Pages 153–172 in G. E. Stelmach, Ed., Information Processing in Motor Control and Learning. Academic Press, New York.