ULMB have been identified in both unimanual and bimanual tasks in individuals with Parkinson's disease (PD) sensory-perceptual mechanisms have recently been suggested to be involved in lower limb freezing. Current study investigated the occurrence of ULMB in PD participants without ('off') and with ('on') dopamine replacement using bimanual wrist flexion-extension with external auditory cues.
ULMB have been identified in both unimanual and bimanual tasks in individuals with Parkinson's disease (PD) sensory-perceptual mechanisms have recently been suggested to be involved in lower limb freezing. Current study investigated the occurrence of ULMB in PD participants without ('off') and with ('on') dopamine replacement using bimanual wrist flexion-extension with external auditory cues.
ULMB have been identified in both unimanual and bimanual tasks in individuals with Parkinson's disease (PD) sensory-perceptual mechanisms have recently been suggested to be involved in lower limb freezing. Current study investigated the occurrence of ULMB in PD participants without ('off') and with ('on') dopamine replacement using bimanual wrist flexion-extension with external auditory cues.
The dopaminergic system in upper limb motor blocks (ULMB)
investigated during bimanual coordination in Parkinsons disease (PD) Matt J. N. Brown
Quincy J. Almeida
Fariborz Rahimi Received: 13 June 2014 / Revised: 12 September 2014 / Accepted: 19 September 2014 Springer-Verlag Berlin Heidelberg 2014 Abstract Upper limb motor blocks (ULMB) (inability to initiate or sudden discontinue in voluntary movements) have been identied in both unimanual and bimanual tasks in individuals with Parkinsons disease (PD). In particular, ULMB have been observed during rhythmic bimanual coordination when switching between phase patterns which is required (e.g. between in-phase and anti-phase). While sensory-perceptual mechanisms have recently been sug- gested to be involved in lower limb freezing, there has been no consensus on the mechanism that evokes ULMB or whether motor blocks respond to dopamine replacement like other motor symptoms of PD. The current study investigated the occurrence of ULMB in PD participants without (off) and with (on) dopamine replacement using bimanual wrist exionextension with external auditory cues. In Experiment 1, coordination was performed in either in-phase (simultaneous exion and extension) or anti-phase (asymmetrical exion and extension between the limbs) in one of three sensory conditions: no vision, normal vision or augmented vision. Cycle frequency was increased within each trial across seven cycle frequencies (0.752 Hz). In Experiment 2, coordination was initiated in either phase pattern and participants were cued to make an intentional switch between phases in the middle of trials. Trials were performed at one of two cycle frequencies (1 or 2 Hz) and one of two sensory conditions: no vision or normal vision. Healthy age-matched control participants were also investigated in both experiments for the occur- rence of motor blocks that were measured using automated detection from a computer algorithm. The results from Experiment 1 indicated that increasing cycle frequency resulted in more ULMB in individuals with PD during continuous coordinated movement, regardless of dopami- nergic status, phase pattern or sensory condition. Experi- ment 2 also conrmed an increased occurrence of ULMB with increased cycle frequency. Furthermore, a large amount of ULMB were observed when initiating anti-phase coordination at 2 Hz, as well as after both externally-cued switches and in catch trials with distracting auditory cues when no switch was required. Dopamine replacement was not found to inuence the frequency of ULMB in either experiment. Therefore, ULMB likely result from non- hypodopaminergic impairments associated with PD. Spe- cically, ULMB may be caused by an inability to shift attentional control under increased cognitive demand that could be associated with hypoactivation in motor and prefrontal areas. Keywords Parkinsons disease Basal ganglia Dopamine Freezing Motor blocks Upper limb Bimanual coordination Introduction Freezing, the inability to initiate, or a sudden interruption in voluntary movements, has been identied as one of the most debilitating symptoms of Parkinsons disease (PD) M. J. N. Brown Q. J. Almeida (&) Sun Life Financial Movement Disorders Research and Rehabilitation Centre (MDRC), Wilfrid Laurier University, 75 University Avenue West, Waterloo, ON N2L 3C5, Canada e-mail: qalmeida@wlu.ca F. Rahimi Department of Electrical Engineering, University of Waterloo, 200 University Avenue West, Waterloo, ON N2L 3G1, Canada F. Rahimi Electrical Engineering Department, University of Bonab, Velayet Highway, East Azerbaijan, Bonab 5551761167, Iran 1 3 J Neurol DOI 10.1007/s00415-014-7514-5 [1, 2]. Freezing has traditionally been identied in the lower limbs (freezing of gait, FoG) in PD during specic aspects of gait including initiation, walking through narrow spaces or during turning [36]. In addition to FoG, several studies have identied a similar phenomenon, upper limb motor blocks (ULMB), in PD during rhythmic unimanual nger tapping [7, 8] and bimanual coordination [915]. FoG is often associated with PD progression though there has been difculty in correlating freezing with a specic pathological mechanism [16, 17]. Our lab recently con- ducted two experiments in which the effects of dopamine replacement on coordination and spatiotemporal parame- ters of rhythmic bimanual movements were assessed [18, 19], although ULMB were not specically examined. The current study investigated the occurrence of ULMB in PD participants without (off) and with (on) dopamine replacement using bimanual wrist exionextension with external auditory cues. In addition, different parameters were manipulated during the bimanual movements such as the type of sensory feedback available during coordination, required cycle frequency, coordination patterns and pattern switching to evaluate the mechanisms that could evoke motor blocks in the upper limbs. The evaluation of ULMB during dopaminergic modulation as well after manipulation of these different parameters during upper limb coordina- tion could provide insight into the role of the dopaminergic system as it relates to the potential underlying mechanisms contributing to ULMB. Currently, the responsiveness of ULMB during biman- ual movements to dopamine replacement is unknown. The amount of motor blocks were not inuenced by dopamine replacement during unimanual nger tapping [8]. More recently, an investigation evaluating the number of ULMB, which they referred to as freezing episodes, during bimanual repetitive movements found that these episodes were correlated with FoG scores but not disease severity in PD, suggesting that dopaminergic system dysfunction may not be the primary contributor to this phenomenon [11]. However, this study did conclude that upper limb freez- ing and FoG may share a common mechanism that pro- vokes a breakdown in coordinated movement resulting in freezing [11]. Similarly, in a subsequent study, more upper limb freezing episodes (i.e. ULMB) were identied during small and fast rhythmic bimanual nger movements rather than when PD patients moved at a comfortable pace and amplitude [12, 14]. Thus, ULMB may be linked to spatiotemporal parameters of movement that have previ- ously been shown to be responsive to dopamine replace- ment [19]. Furthermore, if ULMB and FoG share common mechanisms then their responsiveness to dopamine replacement should be similar. A study investigating FoG revealed that levodopa signicantly decreased FoG fre- quency predominantly during turning [6]. However, in the same study, akinetic FoG (complete lack of movement) was only identied in the PD patients off levodopa medications [6]. These results suggest that FoG may be responsive to dopamine replacement. Recently, different types of FoG have been identied including off levodopa FoG that is responsive to dopamine replacement, whereas in other cases dopamine is argued to have no inuence on FoG (so-called levodopa-unresponsive FoG), or actually lead to FoG (so-called pure on FoG) [2, 6, 17, 2023]. FoG episodes that are unresponsive or evoked with dopa- mine replacement may be linked non-hypodopaminergic pathology involving mechanisms related to stress, cogni- tion or sensory-perceptual impairments [3, 17, 24, 25]. It is important to consider that several mechanisms may underlie ULMB and subsequently, only some of these mechanisms may respond to dopamine replacement. Recently, hypoactivation was identied in cortical motor and prefrontal areas during upper limb motor blocks, [13] supporting a role for theses areas in ULMB. However, it is unclear whether dopamine replacement can modulate these frontal motor and cognitive networks to reduce ULMB. The primary objective of the current study was to determine whether ULMB in PD were responsive to dopamine replacement. Secondary objectives were to: (a) evaluate the occurrence of ULMB during manipulations of frequency of movements, visual feedback, phase, and pattern switching, (b) identify clinical and demographic characteristics of individuals with PD that display ULMB, and (c) characterize the type (i.e. unilateral vs. bilateral), duration and affected limb (more vs. less affected) of ULMB. It was hypothesized that dopamine replacement would not signicantly reduce the occurrence of ULMB. Methods Participants The methods used in both experiments and participants have been described in detail elsewhere [18, 19]. In both Experiment 1 and 2, 15 (n = 15) individuals with Parkin- sons disease and 15 (n = 15) healthy age-matched adults participated. All individuals were right-hand dominant and evaluated on the Modied Mini-Mental State Examination (3-MS) for signs of dementia and to ensure all individuals had the mental capacity to perform the experiment (see Table 1 for demographic information of participants). In both experiments, PD participants performed the rst session after overnight withdrawal from dopaminergic medication. Individuals were evaluated on the motor sub- section of the Unied Parkinsons Disease Rating Scale (UPDRS-III) to document motor symptoms after with- drawal from dopamine replacement (PD off). After J Neurol 1 3 completion of the rst session, medication was self- administered. Re-evaluation of motor symptoms occurred after a 70 min waiting period to document the response to dopamine replacement (PD on) (see Table 2 for clinical characteristics of PD participants). Upper limb laterality scores were calculated and compared for both limbs from items 2025 on the UPDRS-III (which examine upper limb motor symptoms) to classify the more and less affected limbs [19]. Patient history was reviewed for symptoms of FoG. Participants were categorized as FoG or non-FoG (as per a previously established protocol) [3]. PD participants were recruited from the patient database at the Sun Life Financial Movement Disorders Research and Rehabilita- tion Centre (MDRC) at Wilfrid Laurier University. Healthy controls were recruited from family and friends of PD participants. Ethics approval for the current experiment was received from the Human Research Ethics Board (REB) at Wilfrid Laurier University. All participants pro- vided written consent prior to participation in the study in accordance with the Declaration of Helsinki. Apparatus Briey, both experiments had participants seated in a height adjustable chair with their forearms resting on a padded surface and forearms constrained to avoid unwanted movements at the elbow and shoulder joints. The forearms were pronated 90 with the palms facing inward and thumbs facing upwards. Movements were performed on two sepa- rate robotic Phantom Omni haptic devices (SensAble Technologies Inc., Woburn, MA, USA) that were syn- chronized and linked to a desktop computer (Dell Com- puter, with a g-force Intel Pentium 4 with SSE2) for data recording using MatLab R2007b (The MathWorks Inc, Nattick, MA, USA). A pen-shaped stylus was attached to a pivoting arm that allowed for three-dimensional (3-D) movements. Wrist exionextension movements were performed in rhythm with a computer-generated metro- nome using QuaRC (Quanser Inc., Markham, On, Canada). To run all the different experimental sessions, automated programs were created using Simulink in MatLab R2007b. Procedure For both experiments, participants performed wrist ex- ionextension primarily in the mediallateral direction with the hands grasping each stylus. However, the wrists were not constrained allowing 3-D movements if neces- sary. The goal of both tasks was to maintain rhythmic coordination synchronized with the metronome. Partici- pants were instructed to perform movements as large as possible with both limbs, but no specic amplitude requirements were given. In both experiments, in-phase and anti-phase coordination patterns were used [26, 27]. In- phase required the symmetrical movement of both limbs with simultaneous extension and exion of the wrists using homologous muscles. Anti-phase was performed as an asymmetrical pattern requiring exion of one wrist and extension of the opposite wrist using non-homologous muscles. In Experiment 1, three visual feedback conditions were used: no vision, normal vision, and augmented vision. In the normal vision condition participants were able to see their moving limbs. The no vision condition involved the removal of vision by blindfolding participants. In the augmented vision condition moving limbs were covered by a custom-made apparatus but information of the moving limbs was provided using a modied Lissajous gure on the computer monitor (see [19] for more details). A dynamic cycle frequency protocol was used to set the cycle frequency for each trial, where the cycle frequency was gradually increased from 0.75 to 2 Hz at set intervals. Each trial lasted 50 s beginning with a resting period of 5 s. Each phase and sensory feedback manipulation was randomly performed and resulted in a total of 18 trials within each session. Each participant performed a total of 36 trials over 2 sessions (i.e. off vs. on). In Experiment 2, participants performed in either no vision or normal vision conditions. Participants began each trial by performing continuous coordination in either in- phase or anti-phase. At the midpoint of each 23-s trial, a high-pitched auditory cue signaled individuals to perform a Table 1 Summary of demographic information for PD and healthy age-matched control (HC) participants including age, gender and Modied Mini-Mental State Examination (3-MS) scores Experiment 1 Experiment 2 PD HC PD HC N 15 15 15 15 Age (years) 68 (6.4) 65.5 (7.3) 67 (7.5) 67.8 (8.7) Gender Male = 7, female = 8 Male = 6, female = 9 Male = 8, female = 7 Male = 5, female = 10 3-MS score (out of 100) 94.1 (5.2) 96.3 (3.7) 95.5 (4.5) 95.9 (3.0) Years of education (years) 14.5 (3.8) 14.5 (3.8) 14.5 (3.3) 14.1 (3.3) J Neurol 1 3 rapid and smooth transition (intentional switch) to the opposite phase pattern without stopping. To avoid antici- pation of the voluntary switch, the experimenter randomly cued the beginning of trials with a verbal go signal 2, 4 or 6 s after the beginning of each trial (resulting in switches randomly occurring 5.5, 7.5, 9.5, or 11.5 s after the initi- ation of trials). The combination of switch, cycle frequency and sensory condition resulted in 24 trials per session. PD Table 2 Clinical characteristics of PD patients including motor subscale scores on Unied Parkinsons Disease Rating Scale (UPDRS-III), dopamine medication, disease laterality and history of freezing of gait (FoG) Participant Experiment Dopamine medication a Time off medication (in hours) UPDRS-III off (score out of 108) a Time on medication UPDRS-III on (score out of 108) a Previous history of FoG b Disease laterality c Bilaterally affected? 1 1 LD-CD 15 21 90 14.5 N L\R N 2 1 LD-CD, tri, pram 15 38 70 30.5 Y L\R Y 3 1 LD-CD 13.5 27 90 13.5 N R\L N 4 1 LD-CD 15.5 30.5 75 22 Y L\R Y 5 1 rop, ras 17 29.5 85 17.5 N R\L N 6 1 LD-CD 13.5 41.5 70 30.5 N R\L N 7 1 LD-CD 18 42.5 70 22.5 N R\L Y 8 1 LD-CD, ent, ras 16.5 18.5 70 10 N L\R N 9 1 LD-CD 12 46 90 35.5 N L\R Y 10 1 LD-CD 14.5 20 70 12.5 N L\R N 11 1 LD-CD, ent 16 21.5 75 12.5 Y L\R Y 12 1 LD-CD, pram 14.5 34 70 23 Y R\L N 13 1 LD-CD 17 26.5 70 10.5 N L\R N 14 1 LD-CD, ras, pram, 13.5 31 75 22 Y R\ Y 15 1 LD-CD 12.5 32 75 23 N \R N 16 2 LD-CD 15 32.5 70 23.5 N R\L Y 17 2 LD-CD/ent 12.5 42.5 75 32 N L\R Y 18 2 LD-CD 12 34 85 25 N L\R N 19 2 LD-CD 15 35.5 70 29 Y L\R Y 20 2 LD-CD, pram 14 31 75 25.5 Y R\L Y 21 2 LD-CD, rop, sel 12 39.5 75 24 Y R\L Y 22 2 LD-CD/ ent, ras 16 16 70 9 N L\R N 23 2 LD-CD 12.5 38 75 31 Y L\R Y 24 2 LD-CD 14.5 38.5 65 31.5 N R\L Y 25 2 LD-CD 12.5 47.5 90 37.5 Y R\L Y 26 2 Rop 12 32 75 23.5 N R\L Y 27 2 LD-CD 19 30 80 24.5 N L\R Y 28 2 LD-CD 21.5 15.5 75 10 N R\L N 29 2 Pram 17.5 28.5 70 18.5 N L\R N 30 2 LD-CD 14 26 70 19 N R\L Y LD-CD levodopa-carbidopa (L-dopa/Dopa decarboxylase inhibitor), ras rasagiline (MAO-B selective agent), pram pramipexole (dopamine receptor agonist), ent entacapone (COMT inhibitors), rop ropinirole (dopamine receptor agonist), tri trihexyphenidyl (antimuscanrinic) a UPDRS-III scores represent clinical evaluation on the motor subsection of the Unied Parkinsons Disease Rating Scale. Disease laterality was based on the sum of scores on the right side compared to the left side b Previous history of freezing of gait was obtained from patient history on database c Disease laterality was based on the sum of scores of UPDRS-III on the right side compared to the left side J Neurol 1 3 participants performed a total of 48 trials across two ses- sions and healthy controls performed 24 trials in one ses- sion. In addition, to rule out that the high-pitched auditory cue did not act as a distractor provoking ULMB, before each experimental session participants performed six catch trials (3 at 1 Hz and 3 at 2 Hz). These trials had partici- pants perform continuous coordination in anti-phase for 23 s where at the midpoint a high-pitched auditory cue would occur but without instructions to voluntarily switch between patterns. Data processing and analysis For both experiments, displacement data was collected at a rate of 1,000 Hz from each haptic device and stored for ofine analysis using MatLab. Displacement data was used to calculate coordination accuracy, coordination stability, limb frequencies and limb amplitudes [18, 19]. The movement amplitude of each limb was used for analyzing ULMB. The current study classied motor blocks using criteria that combined previous denitions. Freezing in the upper limbs have previously been classied based on at least 1 s of no change in movement amplitude [9, 10]. However, recent research in freezing of gait has indicated that total cessation in movement does not always occur with a freeze [28]. Thus, the denition for upper limb freezing was recently modied to incorporate either a reduction in amplitude (less than 50 % of regular ampli- tude) prior to a freeze or irregular cyclic movement [11, 12, 14, 28]. Thus, in both experiments ULMB were dened as a 75 % reduction of amplitude for at least 1 s. The current denition of motor blocks allowed classi- cation of ULMB that did not produce a total arrest in movement. In both experiments, detection of motor blocks was automated using a script created in MatLab R2007b. All motor blocks that were detected by the computer algorithm were visually inspected on displacement proles to ensure that ULMB were accurately detected with the automated script. Visual inspection was performed to conrm that the automated script did not incorrectly classify irregular movements as ULMB. Voluntary stops were analyzed and documented during each collection to ensure that motor blocks were not falsely identied as intentional arrests in movement (and vice versa). A voluntary stop was dened as any discontinued movement that was not preceded by a reduction in amplitude [11]. These could occur in re- establishing coordination after a transition, early termina- tion of movement at the end of a trial, or during trials due to equipment restrictions. During the testing sessions, two investigators recorded any time distinct voluntary stops occurred. In Experiment 1, ULMB were detected using the peak- to-peak amplitude of each cycle measured over a trial and compared to the reference amplitude. The reference amplitude of movement was obtained from the mean peak- to-peak amplitude within each trial when participants were moving at 1 Hz. Thus, ULMB were classied as a 75 % reduction in amplitude compared to the reference ampli- tude for at least 1 s. In Experiment 2, the reference amplitude was modied from Experiment 1 for two reasons. Unlike Experiment 1, not all trials had movements at 1 Hz and motor blocks were extremely prominent at 2 Hz. Furthermore, the investiga- tors noted that in some trials participants were blocked for nearly the whole duration of a trial, which eliminated the ability to use the reference amplitude within each trial similar to Experiment 1. As a consequence, the mean amplitude was calculated over 4 s within each trial before the cue to switch (from 6 to 10 s). The trial with the largest maximum amplitude (over these 4 s) was used as the ref- erence amplitude, separately for each limb and dopami- nergic status. Chi-square analyses compared the amount of ULMB based on dopaminergic status, sensory condition, phase and cycle frequency in both experiments. T tests were also used to evaluate whether the duration of motor blocks differed in PD patients off compared to on. In Experiment 2, ULMB were separated both before and after voluntary transitions. In addition, the amount of motor blocks was documented in catch trials after the auditory cue. An a level of 0.05 was used to dene statistical signicance. In the event of signicance between more than two variables for either cycle frequency or sensory condition, individual Chi-square tests were performed to determine which fac- tors were different. Results Occurrence and characteristics of ULMB and individuals with PD experiencing ULMB An example of a motor block is displayed in Fig. 1. In Experiment 1, at least one motor block was documented in six out of 15 PD participants (3 more affected on left side and 3 more affected on right side). Only two out of the six were classied as FoG (compared to 3 out of 9 who did not display ULMB were identied as FoG). The duration of ULMB ranged from 1.27 to 14.51 s with an average length of 2.75 (2.4) s. No signicant difference in the duration of ULMB were revealed between PD off (mean = 2.80 s) and on (mean = 2.71 s) (t (38) = 0.10 p = 0.92). Table 3 presents the breakdown of ULMB by the limb(s) they were evoked and dopaminergic status. J Neurol 1 3 In Experiment 2, ULMB were identied before the cue to switch in 5 out of 15 PD participants. Only one out of the ve was classied as a FoG and all were more affected on the left side. The duration of ULMB ranged from 1.2 to 5.2 s (mean = 2.6 1.2 s). No signicant differences in the duration of ULMB was determined between PD off (mean = 2.59) and on (mean = 2.76) (t (50) = -0.43, p = 0.67). After the cue to switch, ULMB were identied in 8 out of 15 PD participants (6 more affected on the left side and considered bilaterally affected). Only three out of eight that displayed ULMB were identied as FoG (com- pared to 2 out of 7 who did not display ULMB were FoG). The duration of ULMB ranged from 1.1 to 9.1 s (3.02 1.9 s). No signicant differences were observed in the duration of ULMB between PD off (mean = 3.09 s) compared to PD on (mean = 3.11 s) (t (95) = -0.05, p = 0.96). Tables 4 and 5 presents the breakdown of ULMB by the limb(s) they were evoked and dopaminergic status before and after the cue to switch phase patterns, respectively. It should be noted that three motor blocks were identied in healthy control participants after the cue to switch in Experiment 2. Based on visual inspection, these trials represented short voluntary stops, which were not preceded by amplitude reductions (see Fig. 2). As a con- sequence, these episodes were not classied as ULMB. In PD, 41 ULMB were identied within the transition period after the cue to switch. However, 12 out of the 41 episodes (29.3 %) were classied as delayed responses. As a con- sequence, 17 trials remained where motor blocks may have Fig. 1 (top) An example of a motor block in the more affected left limb of PD non- FOG (below) red line illustrates the point when all the criteria were met for a ULMB by the computer algorithm. Green line represents the beginning of the motor block, while the end of the motor block was represented by the end of the trial Table 3 Breakdown of ULMB detected using computer-algorithm and visually veried in PD both off and on dopaminergic medication in Experiment 1 PD off PD on More affected a Less affected a Both limbs a Total a More affected a Less affected a Both limbs a Total a Total ULMB 18 (13) 1 (1) 0 19 (14) 14 (9) 3 (2) 4 (4) 21 (15) % of total trials b 4.8 0.04 0 5.2 3.3 0.07 1.5 5.6 % of total ULMB c 40.9 2.3 0 43.2 31.8 6.8 9.1 47.7 a Out of brackets represents all motor blocks including when multiple motor blocks occurred in the same trial. In brackets, represents the total number of trials which an ULMB occurred b Percentage (%) is calculated by dividing motor blocks (excluding motor blocks when multiple blocks occurred) by total trials of each dopaminergic status (n = 270) c Percentage (%) is calculated from all motor blocks including when multiple motor blocks occurred (n = 40 motor blocks) J Neurol 1 3 Table 4 The amount of ULMB before pattern switching in PD participants in Experiment 2 PD off PD on More affected a Less affected a Both limbs a Total a More affected a Less affected a Both limbs a Total a Total ULMB 24 (20) 0 (0) 1 (1) 25 (21) 27 (22) 0 (0) 0 (0) 27 (22) % of total trials b 5.6 0 0.28 5.88 6.1 0 0 6.1 % of total ULMB c 45.3 0 1.9 47.2 50.9 0 0 50.9 a Out of brackets represents all ULMB including when multiple motor blocks occurred in the same trial. In brackets, only represents the total number of trials which an ULMB occurred b Percentage (%) is calculated by dividing motor blocks (excluding motor blocks when multiple ULMB occurred) by total trials of each dopaminergic status (n = 360) c Percentage (%) is calculated from dividing motor blocks by all ULMB (n = 52 motor blocks) including when multiple ULMB occurred in same trial and both limbs blocked at same time Table 5 The amount of ULMB after pattern switching in PD participants in Experiment 2 PD off PD on More affected a Less affected a Both limbs a Total a More affected a Less affected a Both limbs a Total a Total ULMB 41 (31) 1 (1) 5 (5) 47 (37) 49 (28) 1 (1) 0 (0) 50 (29) % of total trials b 8.6 0.06 1.7 10.6 7.8 0.06 0 8.3 % of total ULMB c 40.2 0.09 4.9 45.2 48.0 0.09 0 48.09 a Out of brackets represents all ULMB including when multiple motor blocks occurred in the same trial. In brackets, represents the total number of trials which an ULMB occurred b Percentage (%) is calculated by dividing ULMB (excluding motor blocks when multiple ULMB occurred) by total trials of each dopaminergic status (n = 360) c Percentage (%) is calculated from dividing motor blocks by all ULMB (n = 97 ULMB) including when multiple ULMB occurred in same trial and both limbs blocked at same time Fig. 2 (top) An example of a motor block detected in the right limb during a transition in a healthy control (below) red line represents the point when all the criteria were met for a ULMB by the computer algorithm. Green lines illustrate when the computer algorithm detected the motor block J Neurol 1 3 occurred within the transition period. However, none of these ULMB could clearly be identied based on visual inspection since amplitude reductions would have occurred to perform transitions. In addition, these transitional ULMB appeared similar in some circumstances to the motor blocks detected in healthy controls (see Fig. 3). Thus, ULMB after the cue to switch phase patterns in Experiment 2 only represents trials where ULMB did not occur during the transition period and could clearly be identied as ULMB (Fig. 4). In the catch trials used in Experiment 2, ULMB were identied 64 trials in PD participants after the distracting auditory cue when no switch occurred. Only 41 out of 64 were visually conrmed as ULMB. Several of these trials were veried as small voluntary stops with the auditory cue that were noted by the investigators during testing and Fig. 3 (top) An example of PD motor block detected in the less affected left limb during the transition period (below) red line represents the point when all the criteria were met for a ULMB by the computer algorithm. Green lines illustrate when the motor block was detected by the computer algorithm Fig. 4 (top) An example of a multiple ULMB after a pattern switch from anti-phase to in- phase in the more affected limb of PD (below) blue lines represent periods of decreased amplitude that did not last 1 s. Red lines illustrate the points when all the criteria were met for a ULMB by the computer algorithm. Green lines represent the beginning and end of each motor block. The second motor block terminated at the end of the trial J Neurol 1 3 veried during analysis as they were not preceded by amplitude reductions. Multiple ULMB occurred in the same trial in 8 out of 41 (19.5 %) trials. Both limbs froze in 3 out of 41 (7.3 %) of trials. As a result, ULMB were identied in 17 separate trials out of 90 (18.9 %) total catch trials of PD participants. Effects of dopamine replacement, sensory conditions, phase pattern and cycle frequency on ULMB In Experiment 1, only cycle frequency was found to sig- nicantly increase the amount of ULMB (v 2 (5) = 34.6, p \0.001). Individual comparisons using Chi-square tests revealed that more motor blocks occurred at 1.75 (n = 15) and 2 Hz (n = 14) compared to the three slowest cycle frequencies (0.75 Hz (n = 0), 1 Hz (n = 1) and 1.25 Hz (n = 3)). Additionally, more motor blocks occurred at 1.5 Hz (n = 7) compared to the two slowest cycle fre- quencies (0.75 and 1 Hz). Importantly, no effects of sen- sory condition, phase pattern or dopamine replacement (off n = 19 vs. on n = 21) were observed. In Experiment 2 before the cue to switch, analyses revealed that cycle frequency signicantly increased the amount of ULMB (v 2 (1) = 19.6, p \0.001) as more ULMB were identied in trials at 2 Hz (n = 41) compared to 1 Hz (n = 11). In addition, phase signicantly inu- enced the amount of ULMB (v 2 (1) = 14.8, p \0.001) revealing more ULMB during anti-phase (n = 40) com- pared to in-phase (n = 12). The four combinations of cycle frequency and phase were also compared. Chi-square analysis found a signicant effect (v 2 (3) = 32.4, p \0.001) demonstrating that more ULMB occurred dur- ing anti-phase coordination at 2 Hz (n = 29) compared to during anti-phase coordination at 1 Hz (n = 11) or in- phase coordination at 2 Hz (n = 12). No ULMB were documented during in-phase coordination at 1 Hz. No signicant differences on the amount of ULMB were observed with dopamine replacement (off n = 25 vs. on n = 27) or sensory condition (no vision n = 30 vs. normal vision n = 22). In Experiment 2 after the cue to switch, Chi-square analysis indicated that cycle frequency had a signicant inuence on the amount of motor blocks (v 2 (1) = 24.2, p \0.001). Overall, more ULMB were identied at 2 Hz (n = 72) compared to 1 Hz (n = 25). In addition, there was a signicant inuence of phase on the number of ULMB (v 2 (1) = 10.3, p \0.001) as more ULMB were identied after switching to and performing anti-phase (n = 64) compared to in-phase (n = 33). No effects of dopamine replacement (off n = 47vs. on n = 50) or sensory condition (no vision n = 53 vs. normal vision n = 44) were revealed on the occurrence of ULMB (p [0.05). Discussion The primary objective of the current study was to evaluate the responsiveness of ULMB to dopaminergic treatment. A secondary objective was to evaluate how potential mech- anisms contributing to ULMB might be involved in this responsiveness, by manipulating sensory feedback, required cycle frequency, phase pattern and pattern switching during coordination. The novel nding of the current study was that there was no signicant difference in the frequency or duration of ULMB between PD off and on dopamine replacement during bimanual coordination. Furthermore, increased cycle frequency, anti-phase coor- dination, cued-switching to anti-phase coordination (from in-phase) all contributed to signicant increase in ULMB. Interestingly, sensory feedback had no inuence on the occurrence of ULMB. Dopaminergic system and ULMB The current ndings demonstrating neither difference in the number nor duration of ULMB between PD off and on dopamine replacement is in agreement with previous research that identied manual motor blocks during uni- manual nger tapping were also not dopa-responsive [8]. These results seem to suggest that ULMB is primarily evoked by mechanism(s) outside the hypodopaminergic system that are not ameliorated by dopamine replacement. Although there have been conicting results in FoG [2, 6, 17, 2023], there is some evidence that dopamine replacement can decrease the occurrence and spatiotem- poral precursors of FoG [6, 20, 21]. In addition, the prev- alence of episodes of akinetic FoG (complete absence of movement) were only documented in PD off state after dopaminergic withdrawal [6]. It is likely that these epi- sodes (i.e. akinetic FoG) could be classied as FoG that is responsive to dopamine replacement whereas other types of FoG are unresponsive to levodopa or even emerge only after levodopa [17, 29]. Thus, these results would support that the mechanisms underlying ULMB are potentially shared with the mechanisms underlying FoG that are unresponsive to levodopa therapy. Neural networks contributing to ULMB Surprisingly, only ve out of the total 14 individuals with PD that displayed ULMB that were identied in both the current experiments had a history of FoG. This result is conicting with a previous report [11], which found that J Neurol 1 3 the amount of upper limb freezing episodes were corre- lated with scores on a freezing of gait questionnaire in individuals with PD, suggesting freezing shares a common mechanism. It should be noted that we did not directly examine FoG in our patients. However, some of the patients used in the current experiment have been identied as FoG in other experiments in our lab [3, 4]. Thus, our results suggest that FoG and ULMB do not necessarily share a common mechanism. It may be possible that ULMB may precede FoG in individuals with PD as pre- viously suggested [11]. If this were the case, this would support ULMB as a predictor of the future development of FoG [14]. Alternatively, ULMB and FoG could be linked to different neural mechanisms. Recently, a functional imaging study investigating motor arrests while walking in a virtual reality environment in PD patients determined that these episodes were asso- ciated with impaired processing in prefrontal (increased BOLD signals), posterior parietal (increased BOLD sig- nals), basal ganglia (decreased BOLD signal), sensorimotor (decreased BOLD signals), and brainstem mesencephalic locomotor (MCL) regions (decreased BOLD signals) [30]. In contrast, functional imaging of upper limb motor blocks in PD patients determined that these episodes were asso- ciated with altered activity in cortical motor areas (increased BOLD signals), striatum (decreased BOLD signals) and anterior prefrontal cortex (increased BOLD signals) but not brainstem MCL [13]. Based on these neuroimaging studies, it is plausible that cortical prefrontal and motor networks may be involved in both ULMB and FoG. However, specic involvement of posterior parietal cortex, STN and brainstem areas such as MCL and PPN may solely contribute to FoG. Potential mechanisms contributing to ULMB Surprisingly, the manipulation of sensory feedback had no inuence on the occurrence of ULMB in either of the current experiments. Previous research has suggested that sensory-perceptual impairments may contribute to FoG [2, 3, 17]. However, visual sensory-perceptual impairments do not appear to be a primary contributor to ULMB based on our results, nor do they appear to respond to dopaminergic treatment during bimanual coordination. In both Experi- ment 1 and Experiment 2, the largest amount of ULMB occurred when cycle frequency was at or approached 2 Hz. In Experiment 1, 36 out of the 40 ULMB occurred at cycle frequencies between 1.52 Hz. In Experiment 2, 113 out of 149 ULMB occurred at 2 Hz. Previous research has iden- tied an increased amount of upper limb freezing epi- sodes during fast (50 % faster than normal cadence) as well as fast and small (5 vs. 10 cm) bilateral alternating movements [15]. These results were in contrast to previous studies that did not nd signicant differences between the amount of upper limb freezing during normal and fast- paced bimanual nger coordination [11, 14]. However, it was previously demonstrated that upper limb freezing episodes were preceded by high frequency hastening that included an increase in frequency and decrease in ampli- tude prior to freezing episodes [14]. Based on the spa- tiotemporal parameters previously reported [19], overall amplitudes of movements in PD were small, while fre- quency of movements was similar to aged-matched con- trols. Thus, it may be possible that in particular circumstances the small amplitude-high frequency move- ments provoke ULMB. This observation would be in line with previous research in FoG that reduced stride length and increased stepping frequency precedes FoG [1, 20]. Given that in-phase coordination is well known to be the preferred mode of coordination (as compared to anti-phase) due to the tight temporal-spatial coupling between homologous muscles, one possible explanation of the increased occurrence of ULMB during anti-phase com- pared to in-phase coordination (Experiment 2) could be related to decient movement lateralization (i.e. increased mirrored activity) in PD [3134]. Several studies investi- gating unilateral upper limb movements in PD, particularly with lateralized or hemiParkisonian features, have found that there is increased movement or EMG activity in the resting limb, reective of mirror movements (MM) [31 35]. Interestingly, MM are more pronounced in the less affected limb rather than affected limb in PD, which point to increased M1 activity that results from decreased transcallosal inhibition (or increased excitation) [32, 33]. Furthermore, increased MM have been observed in the less affected limb when PD patients move at cycle frequencies between 0.52 Hz compared to 2.53 Hz [31], as in the current experiments. Thus, if impaired movement laterali- zation particularly during anti-phase coordination were responsible for evoking ULMB, then motor blocks would be expected to be more prevalent in the less rather than more affected limb during bimanual coordination. How- ever, nearly all ULMB in the current experiments were identied in the more affected limb suggesting this is not a likely mechanism for provoking motor blocks. Alternatively, the increased ULMB could be related attentional demand imposed by auditory cueing and/or anti-phase coordination. In both experiments, increased cycle frequency (Experiments 1 and 2) and anti-phase coordination (Experiment 2) were associated to increase ULMB. Previous research did not reveal increased upper limb freezing during anti-phase or high-frequency movements during bimanual nger coordination [14]. However, auditory pacing in this study [14] only occurred for the rst six movement cycles then coordination was internally generated without auditory cues unlike the J Neurol 1 3 continuous cues provided in our experiments. It has pre- viously been suggested that anti-phase coordination requires greater attentional resources than in-phase coor- dination [3638]. Furthermore, it has been suggested that external cueing during anti-phase coordination increases the complexity of the task [39]. Interestingly, in a previous study all but one upper limb freezing episodes were documented during rhythmic nger bimanual coordination after auditory cues were removed within trials [12]. Fur- thermore, PD with FoG demonstrated signicantly smaller as well as increased frequency error, frequency variability, relative phase error and relative phase variability after removal of the auditory cues suggesting that PD with FoG benet and are dependent on external cues [12]. However, based on this study [12], it is unclear how many upper limb freezing episodes may have been provoked by the removal of the auditory cue itself. In Experiment 2, the largest amount of ULMB (n = 97) were identied after the cue to transition between coordi- nation patterns. It has previously been found that voluntary pattern switching from in-phase to anti-phase (compared to anti-phase to in-phase) resulted in a greater occurrence of upper limb freezing (53.9 % compared to 15.5 %) [9]. These ndings are in line with the notion that phase switching, particularly when there was an increased dif- culty after de-stabilizing in-phase coordination, requires increased cognitive demand resulting in movement inter- ruptions [9]. However, these effects may not necessarily be unique to switching between coordination patterns as results of Experiment 2 also found that distracting auditory cues during catch trials also led to a considerable amount of ULMB (n = 41). Thus, it may be possible that the key contributor to provoking ULMB is the shifting of atten- tional resources rather than a sensory-perceptual impair- ment. It should be noted that the dynamic cycle frequency increases used in Experiment 1 would have also included attentional shifts when each new cueing frequency was implemented. Sharing or shifting of attentional resources as well as suppressing behaviours are both related to execu- tive function of the prefrontal cortex, that can be affected in PD. [40, 41] As previously reviewed, altered processing have been identied in the prefrontal cortex during both upper and lower limb motor blocks [13, 30]. Furthermore, a functional imaging study investigating the effects of high cognitive loads on motor arrests in PD patients while walking in virtual reality found that motor arrests were associated with decreased functional connectivity between both the right and left cognitive control networks (CCN) (between the dorsolateral prefrontal cortex and posterior parietal cortex) as well as basal ganglia [42]. Specically during pre-arrest periods, a decreased functional connec- tivity was revealed between the right CCN and basal ganglia [42]. It was concluded that abnormal communication between the left and right CCN as well as right CCN and basal ganglia during periods of high cog- nitive load and low dopamine underlie FoG [42]. Based on these ndings, if a similar neural network is involved in ULMB, then it would suggest that ULMB do not neces- sarily occur due to shifting between motor sets but rather due to increased attention demand that can be produced when having to shift attention between both motor and sensory demands. Conclusion The occurrence or duration of ULMB were not decreased after administration of dopamine replacement, suggesting non-hypodopaminergic PD neuropathology is likely responsible for ULMB. Furthermore, an increased occur- rence of ULMB was associated with increased cycle fre- quencies, anti-phase coordination, switching between coordination patterns, and distracting auditory cues during continuous coordination. These results suggest a signicant decit in PD in shifting attentional resources with increased motor and sensory demands may underlie ULMB. This may be associated with the decreased func- tional connectivity between right and left CCN as well as right CCN and basal ganglia that has been associated with motor arrests in the lower limbs [42]. It is likely that the inability to adequately shift attentional resources evoked a breakdown to small amplitude-high frequency movements that was previously shown to precede upper limb freezing [14]. Future neuroimaging research should be directed at examining the functional connectivity between the CCN and basal ganglia during ULMB. In addition, future research should examine whether different types of ULMB may exist and its relationship with different forms of FoG. Acknowledgments We would like to thank both Dr. Alice Nie- uwboer and Dr. Sarah Vercruysse from University of Leuven for providing their denition and examples of upper limb freezing in a personal communication before publication. This project was sup- ported by National Science and Engineering Research Council (NSERC) scholarship to the rst author, North American Society for Psychology of Sport and Physical Activity (NASPSPA) Graduate Student Award to the rst author, NSERC grant to the second author and Canadian Fund for Innovation (CFI) grant to second author. Conicts of interests On behalf of all authors, the corresponding author states that there is no conict of interest. References 1. Nieuwboer A, De Weerdt W, Desloovere K et al (2001) Abnor- malities of the spatiotemporal characteristics of gait at the onset of freezing in Parkinsons disease. Mov Disord 16:10661075. doi:10.1002/mds.1206 J Neurol 1 3 2. Nutt JG, Bloem BR, Giladi N et al (2011) Freezing of gait: moving forward on a mysterious clinical phenomenon. Lancet Neurol 10:734744. doi:10.1016/S1474-4422(11)70143-0 3. Almeida QJ, Lebold CA (2010) Freezing of gait in Parkinsons disease: a perceptual cause for a motor impairment? J Neurol Neurosurg Psychiatry 81:513518. doi:10.1136/jnnp.2008. 160580 4. Bhatt H, Pieruccini-Faria F, Almeida QJ (2013) Dynamics of turning sharpness inuences freezing of gait in Parkinsons dis- ease. Parkinsonism Relat Disord 19:181185. doi:10.1016/j.park reldis.2012.09.006 5. Giladi N, McMahon D, Przedborski S et al (1992) Motor blocks in Parkinsons disease. Neurology 42:333339 6. Schaafsma JD, Balash Y, Gurevich T et al (2003) Characteriza- tion of freezing of gait subtypes and the response of each to levodopa in Parkinsons disease. Eur J Neurol 10:391398 7. Nakamura R, Nagasaki H, Narabayashi H (1978) Disturbances of rhythm formation in patients with Parkinsons disease: part 1. Characteristics of tapping response to the periodic signals. Per- cept Mot Skills 46:6375 8. Ziv I, Avraham M, Dabby R et al (1999) Earlv-occurrence of manual motor blocks in Parkinsons disease : a quantitative assessment. Acta Neurol Scand 99:106111 9. Almeida QJ, Wishart LR, Lee TD (2003) Disruptive inuences of a cued voluntary shift on coordinated movement in Parkinsons disease. Neuropsychologia 41:442452 10. Almeida QJ, Wishart LR, Lee TD (2002) Bimanual coordination decits with Parkinsons disease : the inuence of movement speed and external cueing. Mov Disord 17:3037. doi:10.1002/ mds.10030 11. Nieuwboer A, Vercruysse S, Feys P et al (2009) Upper limb movement interruptions are correlated to freezing of gait in Parkinsons disease. Eur J Neurosci 29:14221430. doi:10.1111/j. 1460-9568.2009.06681.x 12. Vercruysse S, Spildooren J, Heremans E et al (2012) Abnor- malities and cue dependence of rhythmical upper-limb move- ments in Parkinson patients with freezing of gait. Neurorehabil Neural Repair 26:636645. doi:10.1177/1545968311431964 13. Vercruysse S, Spildooren J, Heremans E et al (2013) The neural correlates of upper limb motor blocks in Parkinsons disease and their relation to freezing of gait. Cereb Cortex. doi:10.1093/cer cor/bht170 14. Vercruysse S, Spildooren J, Heremans E et al (2012) Freezing in Parkinsons disease: a spatiotemporal motor disorder beyond gait. Mov Disord 27:254263. doi:10.1002/mds.24015 15. Williams AJ, Peterson DS, Ionno M et al (2013) Upper extremity freezing and dyscoordination in Parkinsons disease: effects of amplitude and cadence manipulations. Parkinsons Dis. doi:10. 1155/2013/595378 16. Lees AJ (2007) Unresolved issues relating to the shaking palsy on the celebration of James Parkinsons 250th birthday. Mov Disord 22:S327S334. doi:10.1002/mds.21684 17. Heremans E, Nieuwboer A, Vercruysse S (2013) Freezing of gait in Parkinsons disease: where are we now? Curr Neurol Neurosci Rep 13:350. doi:10.1007/s11910-013-0350-7 18. Brown MJN, Almeida QJ (2011) Evaluating dopaminergic sys- tem contributions to cued pattern switching during bimanual coordination. Eur J Neurosci 34:632640. doi:10.1111/j.1460- 9568.2011.07773.x 19. Almeida QJ, Brown MJN (2013) Is DOPA-responsive hypoki- nesia responsible for bimanual coordination decits in Parkin- sons disease? Front Neurol 4:117. doi:10.3389/fneur.2013. 00089 20. Iansek R, Huxham F, McGinley J (2006) The sequence effect and gait festination in Parkinson disease: contributors to freezing of gait? Mov Disord 21:14191424. doi:10.1002/mds.20998 21. Okuma Y (2006) Freezing of gait in Parkinsons disease. J Neurol 253:VII27VII32. doi:10.1007/s00415-006-7007-2 22. Bloem BR, Hausdorff JM, Visser JE, Giladi N (2004) Falls and freezing of gait in Parkinsons disease: a review of two inter- connected, episodic phenomena. Mov Disord 19:871884. doi:10. 1002/mds.20115 23. Nomoto M, Nagai M (2006) Pharmacological consideration of the symptoms resistant to dopaminergic therapy. Parkinsonism Relat Disord 12:S83S87. doi:10.1016/j.parkreldis.2006.05. 025 24. Giladi N, Hausdorff JM (2006) The role of mental function in the pathogenesis of freezing of gait in Parkinsons disease. J Neurol Sci 248:173176. doi:10.1016/j.jns.2006.05.015 25. Vercruysse S, Devos H, Munks L et al (2012) Explaining freezing of gait in Parkinsons disease: motor and cognitive determinants. Mov Disord 27:16441651. doi:10.1002/mds.25183 26. Kelso JAS, Southard DL, Goodman D (1979) On the nature of human interlimb coordination. Science 203:10291031 27. Kelso J, Holt K, Rubin P, Kugler P (1981) Patterns of human interlimb coordination emerge from the properties of non-linear, limit-cycle oscillatory processes- theory and data. J Mot Behav 13:226261 28. Giladi N, Nieuwboer A (2008) Understanding and treating freezing of gait in parkinsonism, proposed working denition, and setting the stage. Mov Disord 23(Suppl 2):S423S425. doi:10.1002/mds.21927 29. Espay AJ, Fasano A, van Nuenen BFL et al (2012) On state freezing of gait in Parkinson disease: a paradoxical levodopa- induced complication. Neurology 78:454457. doi:10.1212/ WNL.0b013e3182477ec0 30. Shine JM, Matar E, Ward PB et al (2013) Exploring the cortical and subcortical functional magnetic resonance imaging changes associated with freezing in Parkinsons disease. Brain 136:12041215. doi:10.1093/brain/awt049 31. Van den Berg C, Beek PJ, Wagenaar RC, van Wieringen PCW (2000) Coordination disorders in patients with Parkinsons dis- ease: a study of paced rhythmic forearm movements. Exp Brain Res 134:174186. doi:10.1007/s002210000441 32. Vidal JS, Derkinderen P, Vidaihet M et al (2003) Mirror move- ments of the non-affected hand in hemiparkinsonian patients: a reection of ipsilateral motor overactivity? J Neurol Neurosurg Psychiatry 74:13521353 33. Espay AJ, Li J-Y, Johnston L et al (2005) Mirror movements in parkinsonism: evaluation of a new clinical sign. J Neurol Neu- rosurg Psychiatry 76:13551358. doi:10.1136/jnnp.2005.062950 34. Cincotta M, Borgheresi A, Balestrieri F et al (2006) Mechanisms underlying mirror movements in Parkinsons disease: a trans- cranial magnetic stimulation study. Mov Disord 21:10191025. doi:10.1002/mds.20850 35. Kishore A, Espay AJ, Marras C et al (2007) Unilateral versus bilateral tasks in early asymmetric Parkinsons disease: differ- ential effects on bradykinesia. Mov Disord 22:328333. doi:10. 1002/mds.21238 36. Temprado J-J, Zanone P-G, Monno A, Laurent M (1999) Attentional load associated with performing and stabilizing pre- ferred bimanual patterns. J Exp Psychol Hum Percept Perform 25:15791594. doi:10.1037//0096-1523.25.6.1579 37. Stinear JW, Byblow WD (2001) Phase transitions and postural deviations during bimanual kinesthetic tracking. Exp Brain Res 137:467477. doi:10.1007/s002210000665 38. Ridderikhoff A, Peper CLE, Beek PJ (2008) Attentional loads associated with interlimb interactions underlying rhythmic bimanual coordination. Cognition 109:372388. doi:10.1016/j. cognition.2008.10.002 39. Johnson KA, Cunnington R, Bradshaw JL et al (1998) Bimanual co-ordination in Parkinsons disease. Brain 121:743753 J Neurol 1 3 40. Rodriguez-Oroz MC, Jahanshahi M, Krack P et al (2009) Initial clinical manifestations of Parkinsons disease: features and pathophysiological mechanisms. Lancet Neurol 8:11281139. doi:10.1016/S1474-4422(09)70293-5 41. Brown RG, Marsden CD (1988) Internal versus external cues and the control of attention in Parkinsons disease. Brain 111:323345 42. Shine JM, Matar E, Ward PB et al (2013) Freezing of gait in Parkinsons disease is associated with functional decoupling between the cognitive control network and the basal ganglia. Brain 1:111. doi:10.1093/brain/awt272 J Neurol 1 3