Objective: To study anticipatory postural adjustments (APAs) in Parkinson’s disease (PD) via a biomechanical analysis, including vertical torque (Tz).
Methods: Ten patients with PD (in the “off-drug” condition) and 10 age matched controls were included. While standing on a force platform, the subject performed a right shoulder flexion in order to grasp a handle in front of him/her, under three conditions (all at maximal velocity): movement triggered by a sound signal and loaded/non-loaded, self-paced movement. The anteroposterior coordinates of the centre of pressure (COP) and Tz were calculated.
Results: A group effect was observed for Tz and COP in patients with PD (compared with controls): the maximal velocity peak appeared later and the amplitude of the COP backward displacement and the area of the positive phase of Tz were lower, whereas the duration of the positive phase of Tz was greater. Interaction analysis showed that the area of Tz was especially affected in the triggered condition and the loaded, self-paced condition. The onset of the COP backward displacement was delayed in the triggered condition.
Conclusion: Our biomechanical analysis revealed that patients with PD do indeed perform APAs prior to unilateral arm movement, although there were some abnormalities. The reduced APA magnitude appears to correspond to a strategy for not endangering postural balance.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Voluntary movements are preceded by anticipatory postural adjustments (APAs). These responses have been studied in many types of segmental movement by recording either muscle activities with electromyography1 or trunk and leg movements using accelerometers.2 For forward oriented movements, the first mechanical event is a backward shift of the centre of pressure (COP).3
Parkinson’s disease (PD) is a neurodegenerative disorder caused by the progressive loss of dopaminergic neurons in the substantia nigra pars compacta. Akinesia, rigidity and tremor account for the main motor symptoms in PD. Furthermore, a deficit in postural reaction appears progressively, and may induce falls. This deficit reportedly involves two components: abnormal corrective, postural reactions4 and a lack of anticipatory change in the activity of postural muscles.5 Indeed, patients who are severely affected clinically tend to show the weakest compensatory postural adjustments, which may contribute to postural instability. In PD, postural preparation is also involved. Anticipatory postural adjustments tend to minimise the potential disturbance of the forthcoming movement and enable equilibrium to be maintained during movement execution. In healthy subjects, execution of an arm raising movement is preceded by a sequence of postural modifications involving recruitment of the postural (leg and trunk) muscles associated with accelerations of the various body segments.2 6 In patients with PD, contradictory results concerning APAs associated with arm movement have been published. Previous reports of anticipatory adjustments in patients with PD have suggested abnormal timing of the electromyographic (EMG) bursts of postural muscles, relative to the prime mover. In particular, a lack of anticipatory change in the activity of postural muscles during arm movement has been reported for standing subjects.5 7 In contrast, the presence of APAs in patients has been detected in other studies,8 9 although certain abnormalities were observed: Dick and colleagues8 showed that the EMG burst amplitude was lower in patients with PD (compared with controls), whereas Rogers and colleagues9 showed that the recruitment of postural muscles was less frequent and was characterised by multiple EMG bursts extending to the agonist.
APAs associated with lower limb movements have also been analysed. Postural adjustments associated with the task of rising on tip toe were investigated in a reaction time paradigm by Diener and colleagues.10 They showed that the basic pattern of preparatory and executional activities was preserved in most patients with PD. The time intervals between postural preparation and execution were also normal. In contrast, analysis of APAs associated with a lateral leg raising task revealed abnormalities.11 12 In severely affected patients with PD, the amplitude of the initial displacement of the centre of foot pressure was markedly lower, and the interval between the earliest force changes and the onset of leg elevation was longer. Furthermore, the alternating bursts and periods of inhibition observed in EMG recordings in normal subjects were replaced by continuous, tonic EMG activity in patients with PD.
APAs are usually analysed by monitoring EMG activities or COP displacements. In our previous study, we examined APAs during an arm raising movement and under different experimental conditions via a particularly useful biomechanical parameter (vertical torque (Tz)).13 14 We showed that an asymmetric movement (such as right arm raising) causes a counterclockwise twist (as viewed from above) applied to the base support, resulting in a negative deviation. This negative phase of Tz is present in all conditions (voluntary and passive movements). In our previous work, we saw that a positive phase occurred prior to the negative phase and preceded movement onset only if the movement was voluntary. The positive phase of Tz thus characterises postural preparation in terms of its latency, intensity and duration. In a reaction time condition (triggered by a sound signal), the positive phase of Tz was delayed in comparison with the self-paced condition at maximal velocity. However, Tz onset was neither modified by load nor velocity. The amplitude of Tz increased with the velocity of voluntary movement, with load and in the reaction time condition.13 In healthy elderly subjects, Tz was delayed in all voluntary movement conditions, in comparison with young subjects.15
The vertical torque Tz parameter is particularly useful because it enables characterisation of APA when voluntary arm movement is performed at low velocity.14 Indeed, as also reported by other authors,16 17 we observed that muscles with early activity prior to rapid movement either did not show any measurable activity prior to slow movement or were activated later. Thus when movement is performed slowly, one records little or no anticipatory muscle activity, whereas the rotational preparation (Tz) remains.14 Tz is the overall consequence of local postural activities and, ultimately, provides information on all anticipatory phenomena associated with voluntary movement. We believe that the contradictory reports on the existence or not of APAs in PD are probably due to the different measurement methods used. The patients may well have performed voluntary movements more slowly than healthy age matched subjects, and thus anticipatory muscle activities would not necessarily have been registered. Tz allows us to make an objective analysis of the presence or absence of APA in patients with PD, whatever the velocity of the voluntary movement. If Tz is recorded, then the characteristics of APA in PD can be studied in different movement conditions. Also, we complemented our Tz data by monitoring a more commonly studied biomechanical parameter, COP.
Ten patients (nine males, one female) with PD (according to the clinical criteria of Gibb and Lees18) were studied in the “off-drug” condition (ie, no medication was administered during the preceding 12 h). Mean (SD) age was 63.8 (8.5) years and mean time since disease onset was 10.6 (6.4) years. Mean Hoehn and Yahr stage score was 2.5 (0.5). The Unified Parkinson’s Disease Rating Scale (UPDRS) Part III score (in the off-drug condition) was 26 (7.3).19 The mean bradykinesia subscore for the (dominant) arm used in the subsequent movement paradigm (items 23, 24 and 25 of the motor UPDRS scale) was 3.6 out of 12 (1.9). The postural stability score (item 29) was 0.5 (0.5), with a posture score (item 28) of 1 (0.7) and a right hand postural tremor score (item 21) of 0.7 (1). Ten elderly age matched subjects (eight men and two women, mean age 67.3 (4.85) years) with no history of neurological or motor disorders were also studied. There was no significant difference between the two groups in terms of age. All patients and control subjects gave their informed consent to participate in this study, which was approved by the local investigational review board.
Subjects stood upright (and barefoot) on a force platform (the AMTI OR-6 from Advanced Mechanical Technology Inc, Newton Massachusetts, USA; 46.5 cm in width and 51 cm in length) with equal weight on each foot (with a constant 20 cm distance between the feet), looking ahead and with their arms hanging freely alongside the body. They were asked to flex their dominant arm in order to reach and grasp a handle located in front of them. All subjects were right-handed. This handle was placed so as to be within their grasp when the arm was held at 90° relative to the bodyline. The focal movement was executed under different conditions: (1) a self-paced task at maximal velocity, in which subjects performed the movement at any self-selected time following a cue from the experimenter to begin; (2) a reaction time task at maximal velocity, in which subjects initiated the movement immediately in response to a sound signal, as rapidly as possible; (3) a self-paced task at maximal velocity with a load of 1 kg attached to the wrist. The subjects were encouraged to make all the movements under all conditions as similar to each other as possible. After a brief practice period, each subject performed 10 trials for each condition, with a rest between trials.
Recording system and data analysis
The movement was recorded with six infrared video cameras (Vicon 370, Oxford Metrics, UK) using 16 retro-reflective spherical (25 mm diameter) markers fixed on several anatomical sites: the spinous process of the 7th cervical vertebra, the sacrum and bilaterally on the acromion, the lateral epicondyl of the elbow, the radial styloid of the wrist, the anterior superior iliac spine, the lateral tuberosity of the knee, the external malleolus of the ankle and the base of the second toe. The markers on the right acromion and wrist were used to compute the arm’s angular displacement and angular velocity (ie, the first derivative of the displacement) (fig 1).
The force plate signals (sampled at 250 Hz) provided the three components of the floor reaction forces (F) in the anteroposterior (x), lateral (y) and vertical (z) directions, as well as the three components of the moment (M). We used these data to compute the coordinates of COP in both the anteroposterior (xCOP) and lateral directions (yCOP). The vertical torque (Tz) was also calculated from the equation
Tz = Mz+yCOP×Fx−xCOP×Fy
Kinetic and kinematic data were recorded simultaneously. Kinematic data were recorded at 50 Hz and then linearly interpolated at 250 Hz, to allow synchronisation with analogue data.
The onset of the angular velocity of the acromion wrist segment was determined and defined as “time zero” (t0) for arm movement onset. All trials were realigned with respect to t0 and were analysed from 0.5 s before this time to 1.5 s afterwards. We also calculated the time and intensity of the velocity peak, as well as the movement duration (ie, the time period from t0 to return to baseline). With respect to COP, only its anteroposterior displacement was considered. The onset time of COP displacement and the magnitude and time of the maximal backward shift were determined (fig 1). The vertical torque was biphasic (a positive phase followed by a negative phase). For each trial, the latency of Tz was determined, together with the duration and total area of the positive phase (reflecting the APA). The area of the positive phase occurring before t0 was also calculated. All latencies were determined with respect to t0 (ie, a negative value means that the event preceded the movement onset).
The mean of 10 trials in each condition was calculated for each individual. We performed a Conover test (an analysis of variance for repeated measures on ranks of variables) with group (healthy subjects; patients with PD) and condition (triggered; loaded, self-paced; non-loaded, self-paced) as variables.21 For the condition effect, the non-loaded, self-paced condition was taken as the reference. The triggered condition and the loaded, self-paced conditions were thus compared with the non-loaded, self-paced condition. Main effects and interactions were further investigated using contrast analysis. A Mann–Whitney test was performed on the reaction time observed in the triggered condition. The results were considered to be significant when p<0.05. Statistical analysis was performed with SPSS software v9.0 (SPSS, Chicago, Illinois, USA).
The results of the Conover tests are presented in table 1. Median (1st and 3rd quartile) values of the characteristic features of angular velocity, Tz and COP, are presented in table 2. Typical velocity, COP and Tz curves for patients and controls subjects are displayed in fig 2, which provides a graphic illustration of the main effect of group detected by the statistical tests. For patients with PD, the time of peak velocity was higher than for controls in all conditions, whereas the amplitude of the COP backward shift and the area and duration of the positive phase of Tz were lower. Only significant results are mentioned in the text below.
There was a main effect of condition on the velocity peak. Contrast analysis showed that the magnitude of this variable was significantly smaller in the loaded, self-paced condition than in the non-loaded, self-paced condition.
Statistical analysis of the time of peak velocity revealed a main effect of condition. This time was shorter in the triggered condition than in the self-paced condition and was greater in the loaded condition than in the non-loaded, self-paced condition. There was also a main effect of group, as evidenced by an increased time of peak velocity in PD sufferers, with a (median, 1st/3rd quartile) value of (0.26 s, 0.22/0.31) in patients and (0.21 s, 0.19/0.24) in controls.
Regarding movement duration, only a main effect of condition was observed. The movement duration was shorter in the triggered condition than in the non-loaded, self-paced condition and was greater in the loaded condition. For all these parameters, there were no significant group×condition interactions.
In the triggered condition, Mann–Whitney tests did not reveal any differences between the two groups in terms of reaction time.
There was only an effect of condition for the onset of Tz and for the area of its positive phase prior to t0. The onset was delayed in the triggered condition, compared with the self-paced condition. The area of the positive phase prior to t0 was smaller in the triggered condition and greater in the loaded, self-paced condition than in the non-loaded, self-paced condition. Main effects of condition and group and a group×condition interaction for the area of positive phase were revealed. This area was greater in the triggered condition and the loaded, self-paced condition than in the non-loaded, self-paced condition. The area was lower in patients with PD (1098 N×mm/s, 774/1360) than in controls (1690 N×mm/s, 1340/2068). Interaction analysis revealed lower values in patients with PD, specifically in the triggered condition and the loaded, self-paced condition. An effect of both condition and group was observed for the duration of the positive phase, which was lower in the triggered condition than in the self-paced condition. This duration was greater in patients with PD (0.45 s, 0.41/0.53) than in controls (0.36 s, 0.32/0.4).
Centre of pressure
We observed a main effect of condition on the amplitude of the COP backward shift. This parameter was greater in the triggered condition that in the self-paced condition. There was also a main effect of group, indicating a lower amplitude in patients with PD (−6.43 mm, −9.31/−4.74) compared with controls (−12.28 mm, −15.7/−8.6).
An effect of condition and a group×condition interaction were revealed for the displacement onset. This time was delayed in the triggered condition, compared with the self-paced condition. The interaction analysis showed that the onset displacement was specifically delayed in patients with PD in the triggered condition (table 2). For the time of maximal displacement, there was only an effect of condition. Maximal displacement occurred earlier in the loaded, self-paced condition than in the non-loaded, self-paced condition.
The present study demonstrated the existence of APA (as revealed by biomechanical parameters) in patients with PD during a unilateral arm raising movement performed at maximal velocity. Indeed, earlier anteroposterior postural preparation (reflected by the COP backward shift) and a rotational postural preparation (indicating the asymmetry of the movement and reflected by the vertical torque (Tz)) were observed, although there were abnormalities in certain movement conditions. In the PD group, the onset of the COP backward shift was delayed, especially in the triggered condition. The magnitude of the two postural preparations was lower in patients with PD, and that of the rotational preparation was especially lower in the triggered condition and the loaded, self-paced condition. The duration of the positive phase of Tz was greater in patients with PD.
As seen in a previous study,13 the characteristic features of Tz and COP differed according to the movement condition. In the triggered condition, the onset of COP backward displacement was delayed and the amplitude of displacement was higher. Moreover, the postural preparation revealed by Tz was delayed, had a shorter duration and was also more intense. A number of authors have shown that APAs are longer in self-paced conditions than in reactive conditions.16 17 22–28 The increase in APA intensity observed in the triggered condition appears to compensate for the delay in anticipation. Indeed, Nougier and colleagues27 suggested that healthy subjects had a more unstable posture in a reactive condition, as shown by a broader range of COP oscillations. The solution for performing the same voluntary movement without loss of balance was thus to increase the intensity of the APA. These results thus revealed a change in instructional strategy. In our study, rotational preparation (Tz) was modified by the presence of an additional load: it was more intense but had the same latency. This modification seemed to be correlated with an increase in the postural disturbance induced by the loaded, self-paced movement. Finally, in both groups, the specific features of Tz and COP were characteristic of voluntary movement.
Comparison of the two groups revealed a number of abnormalities in patients with PD. The magnitudes of the two (anteroposterior and rotational) postural preparations were smaller in patients with PD, with a longer duration for the rotational preparation. In our previous study, we showed that the APA parameters (amplitude or area and duration) were all movement dependent.13 The decrease in the velocity peak tended to reduce the amplitude of the COP backward displacement and the amplitude (and thus the area) of the positive phase of Tz. The greater movement duration also implied a greater duration for the positive phase of Tz. In the present study, the velocity peak and duration of movement did not significantly differ between the two groups. Patients with PD performed the same movement but with a decrease in APA intensity. Our results contradict those of Horak and colleagues16 who reported that the decrease in APA amplitude (or indeed the complete absence of APA) was due to the low velocity of the voluntary movement, since the perturbing forces in slow movement do not endanger balance. These authors concluded that the reduction in postural adjustments are related to disease induced weak motor performance.16 In our study, bradykinesia was not the cause of APA impairments, especially since the mean motor UPDRS subscore in our 10 patients was low (3.6 out of 12). In contrast with the study of Horak and colleagues,16 Bazalgette and colleagues5 reported that reduced motor performance is a consequence of reduced postural adjustments, as patients in a sitting position had a shorter movement duration. Reducing potential imbalance had positive effects on the patients’ motor performance.
The paradigm’s instructions could be related to the abnormalities that we observed. Van der Fits and colleagues29 analysed APAs associated with unilateral shoulder flexion in healthy subjects. They showed that the majority of subjects followed the paradigm’s instructions and performed the arm raising with the deltoid muscle, whereas a minority activated the biceps brachii (ie, they started the movement with a slight flexion of the elbow, followed by arm elevation and extension). These different pointing strategies affected the APA. The arm trajectory consisted of a single segment movement in subjects who activated the deltoid and of a two step movement in subjects who activated the biceps brachii. The single segment movement probably induced a higher moment of inertia, which in turn resulted in earlier activation of the postural muscles in subjects who activated the deltoid.16 30 In our study, the movement strategy involving slight flexion of the elbow was sometimes observed by the experimenter and may explain the low observed APA intensity. Indeed, the patients adopted a flexed posture in inflection and had more difficulty grasping the handle with the outstretched arm, performing the movement with an elbow flexion. These data could also explain the decrease in COP anticipation, which is a parameter that did not depend on the velocity of voluntary movement. In a previous study, we showed that velocity did not influence the latency of postural preparation.13
Changes in the performance of the voluntary movement are certainly a factor but the reduction in the APA was mostly due to the disease itself, as reported by Dick and colleagues.8 The latter authors showed that patients activated three postural muscles (the erector spinae, hamstring and tibialis anterior) with the same temporal sequence as in normal subjects. Dick and colleagues8 concluded that patients deliver the complete programme for forward arm flexion with correct timing but reduced EMG activity. In other words, the form of the complete programme appeared to be preserved in patients but its content was reduced. Similarly, Rogers and colleagues9 specified that the APAs were preserved in 80% of the tests in patients with PD compared with 100% of the tests in control subjects. Equally, Latash and colleagues31 suggested that the general mechanism of feedforward postural programming is intact in patients with PD. In contrast with our study and that of Dick and colleagues,8 the results of Bazalgette and colleagues5 appear to differ for what is the same movement paradigm: postural movements were indeed recorded in patients with PD but did not precede the voluntary movement. This difference might be due to the movement, which was triggered in the study by Bazalgette and colleagues.5 In our study, anteroposterior postural preparation was generally delayed in the triggered condition but it nevertheless occurred (for most of the patients) before the movement onset.
There are several possible explanations for these contradictory results. The primary reason could be the slowness of voluntary movements (classically observed in PD), which would not require anticipatory reprogramming of EMG activities during arm movements in standing subjects.14 16 Indeed, in patients with PD, the movement duration was greater5 32 and peak velocity was lower.8 32 Another reason could be the type of voluntary movement used, especially if the latter was performed in the presence or absence of additional support; subjects were asked to pull a strap in the study by Dick and colleagues8 and to raise their arm in the study by Bazalgette and colleagues.5 This latter movement (with no handle to grasp) is more destabilising, and the presence of APA would be an additional source of disturbance: consequently, no APAs were recorded.
Lastly, the interaction analysis in our study showed that the onset of the COP backward shift was more affected in the triggered movement condition and that the area of the positive phase of Tz was more affected in the triggered condition and the self-paced, loaded condition. Different neuronal mechanisms are involved in triggered and self-paced movements. Electrophysiological data suggest that the supplementary motor area is involved in initiation of self-paced movements, whereas other cortical areas (including the premotor cortex) are involved in the generation of movements in response to visual or auditory stimuli.33 The disturbance of self-paced movement in the PD group is in agreement with this assumption, as one of the major efferences of the basal ganglia goes to the supplementary motor area via the thalamus. In our study, the COP backward shift was modified in patients exclusively in the triggered condition, whereas the area of the positive phase of Tz was modified in the triggered condition and in the loaded, self-paced condition. Finally, one major element which influenced the characteristics of Tz and the COP was the increase in inertial disturbance. The triggered condition and the loaded, self-paced condition increased postural imbalance, with a resulting reduction in intensity for the two postural preparations. These observations are similar to those seen in situations with poor stability and little support. When various movements of the arm were performed by young subjects on a freely rotating support, APAs were less intense.34 Hence APAs could be considered as an additional source of disturbance. The reduction in APA amplitude could be a protection against imbalance.
In conclusion, our results suggest that patients with Parkinson’s disease produce anticipatory postural adjustments during unilateral arm movement, as revealed by analysis of biomechanical parameters. Indeed, vertical torque is the overall consequence of local postural phenomena (such as muscle activity, joint movements and viscoelastic properties) and ultimately provides information on all the anticipatory phenomena associated with voluntary movement, some of which are undetectable with a standard method such as EMG monitoring.14 Moreover, our study revealed certain APA abnormalities in patients with PD: rotational preparation was less intense and lasted for longer. These findings will facilitate analysis of the effect of levodopa and/or subthalamic stimulation on parkinsonian APAs.
We are grateful to Dr David Fraser for helpful comments on the manuscript and revision of the English.
Competing interests: None.
Ethics approval: The study was approved by the local investigational review board.