Aberrant cortical excitability reflects the loss of hand dexterity in musician's dystonia

Abstract

Key points

• Dystonia is a movement disorder characterized by abnormalities at multifaceted aspects of motor dexterity and neural functions.

• Evidence bridging between pathophysiology and movement abnormalities is limited.

• A novel finding was that in focal task‐specific dystonia (FTSD), an aberrantly reduced inhibition at the motor cortex was related to the temporal imprecision of the dexterous finger movements, whereas an elevated facilitation was associated with an abnormally sluggish transition of finger movements from flexion to extension.

• We newly identified two sets of behavioural–physiological covariations as hallmarks of hand FTSD, which is clinically significant because these findings provide novel evidence connecting distinct types of malfunctions within the motor cortex at rest with distinct aspects of motor dexterity degradation in FTSD patients.

Abstract

Focal task‐specific dystonia (FTSD) compromises dexterous movements. A proposed pathophysiological mechanism of FTSD involves malfunction of the motor cortex (M1). However, no evidence is yet available regarding whether and how malfunctions of M1 are responsible for the loss of motor dexterity. Here, we addressed this issue by assessing both M1 excitability and detailed movement parameters, as well as their relationships. Transcranial magnetic stimulation was applied over M1 in 20 pianists with FTSD, 20 healthy pianists and 20 non‐musicians. The patients demonstrated both reduced short‐interval intracortical inhibition (SICI) and elevated intracortical facilitation (ICF) compared with the healthy controls. This indicates that the abnormal cortical excitability reflects pathophysiology but not current skills. Hand motor dexterity was evaluated by position sensors during piano playing at two tempi. The patients showed delayed transition from finger flexion to extension at the fastest tempo and greater timing variability of the finger movements. Furthermore, multivariate analyses identified distinct sets of covariation between cortical excitability and dexterity measures. Namely, the SICI measure and ICF measure were associated with the temporal variability of the movements and the quickness of the transition from flexion to extension, respectively. Specifically, the reduced inhibition and elevated facilitation at M1 in pianists was related to the temporal imprecision and impairment of quick transitions in the sequential finger movements. The present study provides novel evidence associating M1 malfunctions with dexterity loss.

Key points

• Dystonia is a movement disorder characterized by abnormalities at multifaceted aspects of motor dexterity and neural functions.

• Evidence bridging between pathophysiology and movement abnormalities is limited.

• A novel finding was that in focal task‐specific dystonia (FTSD), an aberrantly reduced inhibition at the motor cortex was related to the temporal imprecision of the dexterous finger movements, whereas an elevated facilitation was associated with an abnormally sluggish transition of finger movements from flexion to extension.

• We newly identified two sets of behavioural–physiological covariations as hallmarks of hand FTSD, which is clinically significant because these findings provide novel evidence connecting distinct types of malfunctions within the motor cortex at rest with distinct aspects of motor dexterity degradation in FTSD patients.

Introduction

Dystonia is a movement disorder characterized by sustained or intermittent muscle contractions that cause abnormal, often repetitive, movements and/or postures (Albanese et al. 2013). Focal task‐specific dystonia (FTSD) is one form of this heterogeneous disorder that affects specific body parts and compromises motor dexterity primarily during performance of specific motor tasks. FTSD typically develops in trained professionals, such as writers, golfers, surgeons and musicians (Jankovic & Ashoori, 2008; Altenmüller et al. 2012; Mohammadi et al. 2012; Quartarone & Hallett, 2013), who perform repetitive and precise motor actions over a prolonged period (Rozanski et al. 2015). Previous behavioural studies have characterized impairments of hand motor dexterity due to FTSD. Patients with writer's cramp (WC) exhibit movement slowness (Curra et al. 2004), inaccurate force control (Nowak et al. 2005) and impaired behavioural response inhibition (Stinear & Byblow, 2004a). Musician's dystonia (MD) disrupts timing control of sequential and individuated finger movements (Jabusch et al. 2004; Furuya & Altenmüller, 2013a). In particular, pianists with FTSD showed abnormalities in motor coordination across the digits such as loss of individuated finger movements, and inaccuracy in timing control particularly regarding finger elevation during piano playing (Furuya et al. 2015). FTSD appears to affect various spatiotemporal features of movements, primarily being characterized by the loss of quickness and precise timing control of movements.

The pathophysiology of FTSD remains unclear. The primary motor cortex (M1) plays a key role in fine motor control (Schieber & Santello, 2004; Lemon, 2008), and neuroplastic adaptation in M1 underlies the acquisition, improvement and loss of skilled movements (Karni et al. 1995; Gentner et al. 2010). Given the association of FTSD with the long‐term training of skilled movement, M1 is probably involved in important aspects of dexterity loss in FTSD. However, the extent to which malfunctions exist at the level of M1 in FTSD remains controversial. Many studies have reported functional abnormalities in M1 in patients with FTSD, but contradictory observations have also been reported. Neuroimaging studies have reported abnormal over‐activation of M1 of patients with FTSD involving the hands (Pujol et al. 2000) and mouth (Haslinger et al. 2010), whereas others have failed to identify such M1 malfunctions (Murase et al. 2005; Kadota et al. 2010). Transcranial magnetic stimulation (TMS) studies often demonstrate a loss of inhibition in M1 in patients with hand FTSD (Sohn & Hallett, 2004a; Stinear & Byblow, 2004b; Beck et al. 2008). Nevertheless, interventions with repetitive TMS or unihemispheric transcranial direct current stimulation (tDCS), aimed at restoring M1 function, have failed to relieve the symptoms of FTSD (Murase et al. 2005; Buttkus et al. 2010; Benninger et al. 2011). To date, whether and in what manner malfunctioning of M1 is responsible for the loss of multifaceted aspects of motor dexterity in FTSD remain unclear. This ambiguity results at least in part from the lack of studies that bridge the gap between the neurophysiological and motor behavioural abnormalities in FTSD. Indeed, motor cortical silent period during handwriting in WC patients was not a sensitive pathological marker of motor behavioural abnormalities in FTSD (Stinear & Byblow, 2005). To address this will advance the understanding of the pathophysiological origins underlying the emergence of dystonic movements and the loss of dexterity, thereby clarifying which aspects of motor behavioural abnormalities are related to what kind of functional abnormalities of M1 in FTSD.

The present study aimed to examine both the detailed behavioural abnormalities and the physiological malfunctions of M1 to address their relationship in a cohort with FTSD, specifically MD. We used a paired‐pulse TMS technique to assess malfunctions in M1, as this is an established and unique method for assessing the excitatory and inhibitory functions of local neural circuits (Kujirai et al. 1993). Among several candidate regions with potential malfunctions such as motor and somatosensory cortices, basal ganglia, and cerebellum in patients with FTSD (Furuya & Hanakawa, 2016), the present study focused on M1 because TMS can evaluate detailed excitation and inhibition functions of this region but not others. We hypothesized that M1 excitability measured with TMS might correlate with a measure of fine control of hand movements in pianists with and without MD.

Methods

Ethical approval

In accordance with the latest revision of the Declaration of Helsinki, the experimental procedures were explained to all participants. Written informed consent was obtained from all participants prior to participation in the study, and the experimental protocol was approved by the ethics committee of the National Centre of Neurology and Psychiatry.

Participants

Twenty pianists with unilateral MD (age 40.2 ± 10.4 years; 9 males, 2 left‐handed, 19 classical and 1 popular music pianists), 20 age‐matched healthy pianists (age 37.1 ± 11.6 years; 7 males, 1 left‐handed, all classical music pianists) and 20 age‐matched healthy musically untrained individuals (i.e. non‐musicians) with no history of piano training (age 32.5 ± 10.1 years; 5 males, 1 left‐handed) participated in the study (Table 1). We studied MD as a model of FTSD for the following reasons. First, although musical performances require more dexterous movement control and carry a higher risk for developing FTSD than other tasks such as handwriting (Altenmüller et al. 2012), little is known regarding the pathophysiology of MD (Honda et al. 1998; Jankovic & Ashoori, 2008; Altenmüller et al. 2012). Second, methods have been established to evaluate motor dexterity during keyboard playing in musicians (Furuya & Altenmüller, 2013a; Furuya et al. 2015), although no previous attempts have been made to connect behavioural and physiological measures in pianists with FTSD. All pianists underwent a thorough neurological examination. The exclusion criteria were bilateral FTSD, generalized dystonia, epilepsy, a history of any other neurological diseases and a history of pharmacological intervention, including neuroleptic drugs and the injection of botulinum toxin, within at least the past 3 months. The participants had no histories of other neuropsychiatric disorders, neurosurgery, or metal or electronic implants. The patients had suffered from MD for 6.5 ± 6.0 years (range: 1–28 years). The symptom entailed in most cases exaggerated finger‐flexion or thumb‐adduction, and in some cases difficulty of finger extension due to involuntary flexor cramp pulling the finger(s) down. None of the patients reported a family history of FTSD. The average daily practice duration of the patients was 2.4 ± 1.6 h.

Table 1.

Characteristics of pianists with focal task‐specific dystonia (FTSD)

Patient	Sex	Age (years)	Handedness	Affected hand	Affected finger	Symptom	Age of starting piano playing (years)	Period affected by FTSD (years)
no1	F	47	Right	Right	R, L	hyper‐flx	5	10
no2	F	51	Right	Right	R	hard to ext	5	4
no3	F	59	Right	Right	R	hyper‐flx	4	11
no4	M	34	Right	Right	I, M	hyper‐flx	5	7
no5	M	38	Left	Right	M	hyper‐flx	12	4
no6	M	45	Right	Right	M	hyper‐flx	5	3
no7	M	29	Right	Left	I	hyper‐flx	8	6
no8	F	36	Right	Left	I	hard to ext	3	3
no9	F	49	Right	Left	I, R	hyper‐flx	3	8
no10	M	27	Right	Right	T	hyper‐add	12	5
no11	M	37	Left	Right	L	hyper‐flx	5	11
no12	F	46	Right	Right	M, R	hard to ext	7	3
no13	F	50	Right	Right	M	hyper‐flx	9	3
no14	M	24	Right	Right	T	hyper‐add	16	4
no15	M	32	Right	Left	I	hyper‐flx	4	1
no16	F	40	Right	Right	M	hyper‐flx	6	10
no17	F	23	Right	Right	M	hyper‐flx	6	3
no18	M	46	Right	Right	I	hard to ext	4	28
no19	F	56	Right	Left	M, R	hyper‐flx	3	1
no20	F	36	Right	Right	I, M, R	hyper‐flx	5	5

T, I, M, R and L indicate the thumb, index, middle, ring and little finger, respectively.

hyper‐flx: exaggerated flexion, hard to ext: difficulty of extension, hyper‐add: exaggerated adduction.

Experimental design

The entire experiment consisted of a TMS recording and a motor behavioural test, which were performed on two separate days in a randomized order across the participants. Only the non‐musicians participated in a TMS recording but not in a behavioural test, because they are musically untrained and supposedly unable to play the target task fast without practising. The experimental days of the two tests were separated from 2 to 6 weeks.

EMG recording

To record the motor evoked potentials (MEPs) and background muscular activities, surface EMG data were recorded from the abductor pollicis brevis (APB), first dorsal interosseous (FDI), abductor digiti minimi (ADM), flexor digitorum superficialis (FDS) and extensor digitorum communis (EDC) muscles using Ag/AgCl disposable electrodes with diameters of 23 mm (Biorode SDC112 GE Healthcare, Piscataway, NJ, USA). The target muscle was FDS responsible for the flexion of all fingers, and the remaining muscles were recorded primarily to detect unintentional/involuntary muscular contractions (e.g. as background EMG) in those muscles during the experiment. In addition, we analysed the results from EDC, antagonistic to FDS. Standard skin preparation was performed prior to attachment of the electrodes. For extrinsic finger muscles located in the forearm (FDS and EDC), the electrodes were mounted over the muscle belly in a bipolar montage. For intrinsic hand muscles (APB, FDI and ADM), the electrodes were placed in a belly‐tendon montage. For the FDS muscle, the electrode was carefully placed and activity from FDS was checked to minimize the crosstalk from activity of other forearm muscles, especially the flexor carpi radialis muscles. However, we acknowledge that both EMG activities and MEPs at the FDS can be contaminated by those at the other forearm flexor muscles. The EMG signals were amplified, band‐pass filtered (10–1000 Hz), sampled at 2 kHz (PowerLab, ADInstruments, Oxford, UK), and stored on a personal computer for offline analysis using LabChart software (ADInstruments). For the subsequent analyses, the measured EMG data were full‐wave rectified.

TMS recording

The participants were seated in a comfortable chair with both arms resting on a cushion while supinated. Single‐pulse and paired‐pulse TMS were delivered with a figure‐of‐eight shaped coil (wing diameter of 70 mm) connected to two high‐power Magstim 200² machines through a Bistim module (Magstim Company, Dyfed, UK). An experimenter held a coil tangentially to the scalp while keeping the handle orientated backwards at a 45‐deg angle from the sagittal plane, which yielded a posterior‐to‐anterior current direction, to activate the corticospinal system. The coil was placed over the ‘motor hot spot’ for eliciting the largest MEP in the FDS (target muscle). Additionally, we confirmed that TMS at the unconditioned stimulus intensity also elicited MEPs in the APB, FDI, ADM and EDC muscles. For ensuring proper coil placement throughout the experiment, we marked this position on the scalp with a felt‐tip pen. The majority of the participants underwent stimulation over the left M1 to elicit MEPs in the right upper limb. However, because five patients had FTSD affecting the left hand, these five patients, as well as five age‐ and gender‐matched controls in each of the bhealthy musicians and non‐musicians groups, underwent stimulation over the right M1. The resting motor threshold (RMT) for the FDS was defined as the minimal stimulus intensity required to evoke MEPs of at least 50 μV in 5 of 10 successive trials. The active motor threshold (AMT) for the FDS was defined as the minimal stimulus intensity required to evoke MEPs of at least 100 μV MEPs in 5 of 10 successive trials during a weak sustained isometric muscle contraction of the FDS of approximately of 10% of the maximum voluntary contraction. Step width for threshold determination was about 1% of stimulator output. The TMS intensity for the conditioning stimulus was set at 90% of the AMT. The entire TMS assessment procedure followed standard guidelines (Chipchase et al. 2012).

The TMS recording session consisted of single‐pulse and double‐pulse TMS conditions. In the single‐pulse TMS condition, the intensity of an unconditioned stimulus was adjusted to elicit MEPs of 0.2–0.4 mV in amplitude at rest before the experimental session was initiated. This intensity was determined on the basis of a pilot experiment in three healthy pianists (not included in the study). We failed to elicit MEP of 1 mV at FDS even with the maximum stimulation intensity of the stimulator. However, MEPs of 0.2–0.4 mV were consistently evoked by stimulation at reasonable intensities, corresponding approximately to 1.3‐ to 1.5‐fold of the RMT (for details see the Results). In the paired‐pulse TMS condition, inter‐stimulus intervals (ISIs) of 1, 2, 3, 4, 5, 6, 10 and 12 ms between the conditioning and test stimuli were used to assess short interval intracortical inhibition (SICI) and intracortical facilitation (ICF) of M1 (Kujirai et al. 1993). Here, the intensity of the test stimulus was set at the same intensity as the unconditioned stimulus in the single‐pulse TMS experiment. Additionally, an ISI of 100 ms was adopted to assess the long intracortical inhibition (LICI) of M1, only in which the conditioning stimulus intensity was the same as the test stimulus intensity. In the present study, we did not assess the other ISIs such as 150 and 200 ms to investigate the LICI, and therefore acknowledge a possibility that the appropriate ISI for assessing LICI might be overlooked.

For all conditions (i.e. single‐pulse and paired‐pulse TMS with different ISI conditions), TMS was delivered every 5–7 s while the participants completely relaxed both the hand and the forearm. Sixteen MEPs were recorded for each condition, and the experimental order was randomized across the conditions and the participants so as to intermingle the order of running the single‐pulse and double‐pulse experiments with different ISIs.

Behavioural test

Each participant played a sequence of 17 successive strikes of five adjacent piano keys (C, D, E, F, G) within a range of one octave with a specified fingering (1‐2‐3‐4‐5‐4‐3‐2‐1‐2‐3‐4‐5‐4‐3‐2‐1, where 1, 2, 3, 4 and 5 represent the thumb, index, middle, ring and little finger, respectively) unilaterally as accurately as possible. The sequence was played either in synchrony with a metronome (inter‐keystroke interval of 300 ms, i.e. the paced condition) or as fast as possible (i.e. the fastest condition). For each of the two tempi, each participant performed 10 trials. None of the pianists struck a wrong key throughout the experiment. The five patients with left‐hand MD and the five matched controls were asked to play with the left hand, whereas the others played with the right hand. The score, with the fingering, was visually presented on a computer monitor located in front of the participants. Prior to the experiment, each participant was provided with an auditory stimulus that played the target melody at the target loudness of 80 Musical Instrument Digital Interface (MIDI) velocity equivalent to mezzo‐forte (moderately loud). The participants were asked to play the target sequence with a legato touch, i.e. each key was not released until the next key was depressed. The pianists played an acoustic piano (U1, Yamaha Co., Tokyo, Japan) with MIDI sensors implemented beneath all piano keys. Using a custom‐made script, we recorded the MIDI data representing the times of the individual keystrokes and key‐releases and the key‐descent velocity (i.e. the loudness), and these parameters served as measures of finger dexterity (Furuya & Altenmüller, 2013a; Furuya et al. 2015).

Data analysis

TMS recording

As an index of the MEPs, the integral of the MEP over 30 ms, within the range of 15–45 ms following the onset of the TMS, was computed for each trial (i.e. the MEP area). We used the area rather than the amplitude because MEPs in the forearm muscles often exhibit polyphasic patterns (Byblow et al. 2007). We assessed the background EMG as root mean square over a 50‐ms window before the stimulation onset and rejected trials in which the background EMG in the FDS was greater than 20 μV (MacKinnon et al. 2005; Beck et al. 2009; Qi et al. 2011; Tremblay et al. 2016). After these exclusion procedures, the median of the MEP area was used as the summary statistics (here the median was more stable than the mean that is susceptible to large variability of the MEPs across stimulations in some individuals). To assess intracortical inhibition and facilitation, an MEP ratio was obtained through normalizing the MEP area derived from the paired‐pulse stimulation for each ISI by the unconditioned MEP obtained in the single‐pulse experiment.

Behavioural test

Using the MIDI information, we evaluated finger dexterity by computing the inter‐trial mean values and variabilities of the following five parameters: (1) the interval between two successive key‐presses (inter‐keystroke interval: IKI), (2) the interval between two successive key‐releases (inter‐key‐release interval: IRI), (3) the interval between the key‐press and key‐release of a strike (finger‐key contact duration), (4) the interval between the key‐press of a strike and the key‐release of the previous strike (overlap duration between two successive tones) and (5) the key‐descending velocity at individual strokes (loudness) (Jabusch et al. 2004; Furuya & Altenmüller, 2013a). These parameters were computed across trials for each participant and at each of the two tempi. The computed inter‐trial mean values and variabilities were then averaged across the strikes, and the results served as measures representing the different aspects of motor dexterity during piano playing.

Statistics

To assess group differences of the MEP ratios for the different ISIs, a two‐way mixed‐design ANOVA was performed. The between‐subject and within‐subject variables were ‘group’ (three levels: patients/healthy‐pianists/non‐musicians) and ‘ISI’ (eight levels: 1–12 ms), respectively. Post hoc comparisons were performed with corrections for multiple comparisons (Benjamini et al. 2006). For the 100‐ms ISI, a one‐way ANOVA was performed to compare the LICIs between the groups. We analysed the 100‐ms ISIs separately from the 1‐to‐12 ms ISIs due to the discontinuity of the ISIs and different conditioning stimulus intensity. The background EMG levels in the included trials were evaluated over 50 ms prior to the stimulus onset and compared between the ISIs and groups. The RMTs and AMTs were also evaluated and compared between the groups with one‐way ANOVAs (Table 2). To assess group differences in the MIDI‐derived behavioural variables at each tempo, two‐way mixed‐design ANOVAs were performed, in which the between‐subjects and within‐subject variables were ‘group’ and ‘tempi’ (two levels: paced and fastest), respectively. Statistical analyses were performed using R (ver. 3.2.3).

Table 2.

Neurophysiological characteristics of participants

		MD patients		Healthy pianists		Non‐musicians		ANOVA results
		Mean	SEM	Mean	SEM	Mean	SEM	F 2,57	P	η2
Resting MT (RMT) (%MSO)		55.2	(1.89)	54.4	(1.96)	50.6	(1.88)	2.31	0.11	0.07
Active MT (AMT) (%MSO)		42.8	(1.36)	40.1	(1.67)	41.3	(1.46)	0.65	0.53	0.02
Background EMG (μV)	FDI	13.1	(2.9)	8.6	(1.2)	11.7	(1.7)	1.01	0.37	0.03
	ADM	9.4	(2.2)	13.8	(3.0)	13.8	(2.2)	1.03	0.36	0.03
	APB	6.4	(1.1)	6.1	(1.2)	5.5	(0.9)	0.17	0.84	0.01
	FDS	5.9	(0.9)	7.2	(1.1)	5.5	(0.4)	1.11	0.33	0.04
	EDC	11.7	(2.0)	14.1	(2.4)	14.6	(2.7)	0.41	0.67	0.01

The number in parentheses indicates one standard deviation.

MT: motor threshold, MSO: maximum stimulator output.

Finally, to test whether the inter‐individual differences in intracortical inhibition and facilitation in M1 were associated with those in the measures of finger dexterity across pianists, we performed penalized multiple‐linear‐regression analyses with L1 norm regularization, so called LASSO (least absolute shrinkage and selection operator) regression. This algorithm selects a subset of independent variables that are associated with the dependent variables from a total pool of candidate predictors, through shrinking the regression weights toward zero (i.e. feature selection) (Tibshirani, 1996). The λ parameter that determines the overall intensity of regularization was optimized through three‐fold cross‐validation of the LASSO regression model. The dependent variables were each of the MIDI variables (e.g. the inter‐trial mean values and variabilities of the IKI, IRI, contact duration and loudness) in the ‘fastest’ tempo condition, in which all of these measures differed significantly between the groups. The independent variables were the MEP ratios at the three different ISIs, which represented the SICI (1 and 3 ms) and the ICF (10 ms) (the LICI was not included in the model due to a lack of group‐wise differences, see Results). Here we chose values of 1 and 3 ms as the representatives of SICI, based firstly on previous reports suggesting different physiological mechanisms underlying cortical inhibition phenomena between these two ISIs (Fisher et al. 2002; Roshan et al. 2003), and secondly on a group‐wise difference between the patients and healthy pianists at these two ISIs (see Results). In addition, the independent variables included both age and a period affected by FTSD of the participants. The present regression model included all participants from both the healthy and the dystonic musician groups, on a theoretical basis of a heuristic model proposing a continuum from healthy to dystonic musicians, rather than two clearly distinct groups (Altenmuller et al. 2015). One possible factor determining fine motor control is a balance between the excitatory and inhibitory local circuits in M1, which has been considered to change plastically in relation to both motor training (Perez et al. 2004) and the development of dystonia (Ridding et al. 1995a). The present regression model was thus based on an assumption that the functional balance of the inhibitory and/or excitatory circuits might relate to fine motor control in healthy and dystonic musicians. In other words, we presumed that the imbalance of the circuits, as measured through the paired pulse TMS, might characterize one of the predisposing factors of dystonia spanning over the two groups. Regression analyses were performed using MATLAB (MathWorks, Inc., Natick, MA, USA).

Results

Analysis of the patients’ profiles

Age, handedness and sex did not differ significantly between the patients, healthy pianists and non‐musicians groups (age: one‐way ANOVA, F _2,57 = 2.68, P = 0.08, η² = 0.09; handedness: χ² test, χ² = 0.43, P = 0.80; sex: X² test, χ² = 3.33, P = 0.19). The affected digits were the thumb, index, middle, ring and little fingers for 2, 7, 9, 7 and 2 of the patients, respectively (Table 1). Because the FDS is responsible for flexing the index, middle, ring and little fingers, the total number of patients whose FDS was affected was 18. Pianists with and without MD began playing piano at 6.4 ± 3.5 and 3.9 ± 1.1 years of age, respectively. A t test yielded a significant difference between these two groups (P = 0.0045), confirming a previous observation that patients with MD initiated musical training at a later age than did healthy musicians (Schmidt et al. 2013).

TMS assessment

Table 2 summarizes the group mean values of the basic TMS parameters and background EMG data of the individual muscles for each group. Statistical analyses did not reveal group‐wise differences in the RMT, AMT or background EMG activities of any of the monitored muscles (one‐way ANOVA: P > 0.05 for all comparisons, Table 2). A two‐way mixed design ANOVA using muscle and group as independent variables did not yield an interaction effect between group and muscle (F _8,228 = 1.20, P = 0.30, η² = 0.03) or a main effect of group (F _2,57 = 0.17, P = 0.85, η² = 0.002) for the background EMG activity. Hence, group‐wise differences in cortical inhibition and facilitation measures, if any, were unlikely to have resulted from differences in background EMG activities of either the target muscle or non‐target muscles between the groups. This assessment is important for fair evaluation of MEPs because the contraction of non‐target surrounding muscles can affect the cortical excitability of the target muscle (e.g. surround inhibition) (Sohn & Hallett, 2004b). There was no significant difference in group mean values of the MEP areas from the FDS at the unconditioned intensity between the three groups (0.17 ± 0.08 mV.ms for the patient group, 0.13 ± 0.11 mV.ms for the pianists group and 0.10 ± 0.03 mV.ms for non‐musicians group; a group effect by one‐way ANOVA: F _2,57 = 3.00, P = 0.06, η² = 0.09). Likewise, the unconditioned MEP areas from the antagonistic EDC did not differ between the three groups (0.30 ± 0.15 mV.ms for patients, 0.29 ± 0.28 mV.ms for pianists and 0.21 ± 0.09 mV.ms for non‐musicians group; a group effect by one‐way ANOVA: F _2,57 = 1.14, P = 0.32, η² = 0.04). The intensities of the unconditioned stimuli required to evoke 0.2–0.4 mV in the FDS ranged from 1.3‐ to 1.5‐fold of the RMT [patients: 73.2 ± 7.1% maximum stimulator output (%MSO); pianists: 81.0 ± 8.0%MSO; non‐musicians: 70.5 ± 9.4%MSO]. At these unconditioned intensities, MEPs were observed in all monitored muscles in all participants.

Figure 1 A illustrates representative MEP waveforms superimposed over five unconditioned stimuli (grey) and five paired‐pulse stimuli (black) with ISIs of 1 ms (SICI) and 10 ms (ICF). Differences in the MEP amplitudes between the unconditioned and paired‐pulse stimulations at the 1‐ms ISI (i.e. SICI) were more obvious in the healthy pianist and non‐musician than in the patient. Conversely, in the ICF condition, differences in MEP amplitudes were more evident in the patient than the healthy pianist or non‐musician. Figure 1 B displays the mean MEP areas at the different ISIs for the patients (filled circles), healthy pianists (open circles) and healthy non‐musicians (open squares). A two‐way mixed‐design ANOVA using group and ISI as the independent variables revealed a significant interaction effect (F _14,399 = 2.16, P = 0.025 following the adjustment of degrees of freedom with Greenhouse–Geisser's ε, η² = 0.03) and significant main effects of the ISI (F _7,399 = 68.73, P = 1.39 × 10⁻⁴² with the Greenhouse–Geisser adjustment, η² = 0.33). Post hoc tests with adjustments for multiple comparisons identified significant group differences between the patients and both control groups at ISIs of 1, 3 and 10 ms. At the ISI of 2 ms, a group difference was evident only between the patients and non‐musicians, whereas at the ISI of 6 ms, there was a group difference between the patients and healthy pianists. No significant group differences in SICI/ICF were found between the healthy pianists and the non‐musicians. The MEP ratio at each of these ISIs was not significantly correlated with age of starting piano playing (P = 0.26, 0.32, 0.51 and 0.89 for 1, 3, 6, and 10 ms ISI, respectively), which indicated no association of the group‐wise difference between the MEP ratio and age of starting piano playing. Overall, the patients exhibited lower SICI and more pronounced ICF values than the two healthy control groups. These results highlighted the reduced SICI and more pronounced ICF that were specific to the motor corticospinal system innervating the symptomatic muscles. In contrast, at the EDC, which is the antagonist of the FDS, a two‐way ANOVA failed to reveal an interaction effect between group and ISI (F _14,399 = 1.69, P = 0.20, η² = 0.03) or a main effect of the group (F _2,57 = 1.61, P = 0.21, η² = 0.03). This result may indicate a lack of functional abnormality in the corticospinal system innervating the finger extensor muscle in the patients. However, this interpretation requires caution because the EDC was not the target muscle of the TMS (i.e. the intensity of the conditioning stimulus was not adjusted to the EDC). This was also the case for the remaining muscles, and this fact precluded detailed analyses of the results from those non‐target muscles.

Figure 1. TMS results.

A, typical MEP waveforms from a representative patient (left) and control (right) are superimposed over five unconditioned stimuli (grey) and five paired‐pulse stimuli (black) with an ISI of 1 ms (upper panel) to demonstrate short interval intracortical inhibition (SICI). The same unconditioned MEP waveforms are superimposed with five paired‐pulse stimuli (black) with an ISI of 10 ms (lower panel) to demonstrate intracortical facilitation (ICF). B, group mean values of the MEP ratios at nine different ISIs in patients (filled circles), healthy pianists (open circles) and healthy non‐musicians (open triangles). ^* P < 0.05 (a group difference between the patients and healthy pianists), ⁺ P < 0.05 and ⁺⁺ P < 0.01 (a group difference between the patients and non‐musicians).

To assess the group differences in LICI, we evaluated the effects of paired‐pulse TMS with an ISI of 100 ms on the MEP ratio of the FDS (Fig. 1 B). A one‐way ANOVA revealed no significant group effect (F _2,57 = 1.86, P = 0.16, η² = 0.06).

Behavioural assessment

Figure 2 illustrates the mean values and variabilities of the five MIDI‐derived dexterity parameters during piano playing in the pianists with and without MD. At the paced tempo, the patients exhibited smaller mean values for contact duration, tone overlap and loudness than the controls. At the fastest tempo, the patients exhibited greater mean IKI, IRI and contact duration values and a smaller mean loudness values than the controls. At both the paced and the fastest tempi, the patients displayed greater inter‐trial variability in IRI, contact duration, tone overlap and loudness than the controls. Additionally, the IKI variability was greater in the patients than in the controls at the fastest tempo. Overall, the loss of dexterity in MD was characterized by increased temporal variability of keystrokes (IKI) and key‐releases (IRI) and decreased and more variable key‐striking velocity (loudness). Specifically, the patients exhibited slower performances and longer finger‐key contact durations, indicating delayed initiation of key‐releases (i.e. delayed finger lifting/extension). Table 3 summarizes the results of two‐way mixed‐design ANOVA using group and tempo as the independent variables. Significant interaction effects between these variables were observed for the inter‐trial mean values of all of the variables except for loudness (P = 0.005 with Bonferroni correction).

Figure 2. Behavioural results.

Horizontal grey bars representing the timing of the key‐presses and key‐releases of two successive keystrokes (i.e. the edges of the bars correspond to the timing of these events) depict each of the four variables evaluated. A, the interval between two successive key‐presses (inter‐keystroke interval: IKI); B, the interval between two successive key‐releases (inter‐key‐release interval: IRI); C, the interval between a key‐press and key‐release of a strike (finger‐key contact duration); and D, the interval between the key‐press of a strike and the key‐release of the previous strike (i.e. the overlap duration between two successive tones). The black and white bars represent the group mean values of the inter‐trial mean values (left panel) and inter‐trial variabilities (right panel) of the individual movement features during the piano keystrokes at the paced and fastest tempi (x‐axis) for the FTSD patients (black bar) and healthy controls (white bar). The unit of the variables was milliseconds for IKI, IRI, contact duration and overlap between tones, and MIDI velocity for loudness. ^* P < 0.05, ^** P < 0.01, ^*** P < 0.001.

Table 3.

Results of two‐way mixed‐design ANOVA for each of the dexterity parameters in the pianists with and without focal task‐specific dystonia (FTSD)

		Inter‐trial mean values					Inter‐trial variabilities
		IKI	IRI	CD	Overlap	Loudness	IKI	IRI	CD	Overlap	Loudness
Group	F 1,38	8.702	13.112	2.061	3.893	10.919	7.203	16.306	16.852	13.447	8.513
	P	0.005	0.001	0.159	0.056	0.002	0.011	2.52 × 10−4	2.06 × 10−4	0.001	0.006
	η2	0.105	0.160	0.032	0.079	0.213	0.106	0.239	0.254	0.211	0.125
Tempo	F 1,38	963.433	781.692	801.135	12.003	3.208	0.824	23.909	19.093	3.916	23.063
	P	1.32 × 10−28	5.96 × 10−27	3.82 × 10−27	0.001	0.081	0.370	1.87 × 10−5	9.30 × 10−5	0.055	2.46 × 10−5
	η2	0.925	0.902	0.891	0.049	0.005	0.008	0.144	0.104	0.025	0.180
Group × Tempo	F 1,38	23.494	18.071	31.899	14.424	6.863	4.083	0.120	0.105	0.679	6.224
	P	2.14 × 10−5	1.33 × 10−4	1.73 × 10−6	0.001	0.013	0.049	0.731	0.748	0.415	0.017
	η2	0.231	0.176	0.245	0.058	0.011	0.039	0.001	0.001	0.004	0.056

IKI: inter‐keystroke interval; IRI: inter‐key‐release interval; CD: finger‐key contact duration; Overlap: overlap between tones.

A bold number indicates a significant effect. P = 0.005 with Bonferroni correction.

Multiple‐regression analysis

Based on a proposal for a continuum from healthy to dystonic musicians (Altenmuller et al. 2015), it is plausible that the individual difference in the functional architecture of the excitatory and inhibitory circuits in M1 correlates with a wide spectrum of motor dexterity across pianists with and without MD. A statistical model was therefore developed using a multiple‐linear‐regression technique with L1 norm regularization (i.e. LASSO regression). Before running the analysis, one‐sample Kolmogorov–Smirnov tests were performed for each of the dependent and independent variables from 40 pianists (i.e. both the MD and the unaffected pianists), which confirmed Gaussian distributions (MEP ratio: P = 0.12, 0.10 and 0.23 for 1, 3 and 10 ms ISIs, respectively; MIDI variables: P = 0.55 and 0.92 for the IKI mean values and SDs, respectively; P = 0.46 and 0.18 for the IRI mean values and SDs, respectively). Thus, the distribution of the data was consistent with our assumption for a continuum in both M1 excitability and finger dexterity across the groups. Table 4 summarizes the results of the regression analyses, which explained inter‐individual differences in each dexterity parameter at the fastest tempo based on physiological measures representing the SICI and ICF (i.e. 1, 3 and 10 ms) and participants’ profile (age, disease duration affected by FTSD). The values shown in Table 4 indicate coefficients derived from the LASSO, and the sign indicates whether each dependent variable co‐varied positively or negatively with the independent variable. The regression analysis focused on the IKI and IRI, because these variables that are independent of each other represent quickness and accuracy of the finger flexion and extension, which differed between the patients and controls (Fig. 2). The ICF was a significant covariate of the mean value of the dexterity parameter, explaining the IRI, which positively co‐varied with the MEP ratio at the 10 ms ISI but not the 1 and 3 ms ISIs (Table 4 lower left). This positive covariation indicates that pianists with more pronounced ICF had more prolonged IRI. In contrast, the SICI (i.e. 1 and 3 ms ISIs) was associated positively with inter‐trial variability (Table 4 upper right), being a significant covariate of the inter‐trial variabilities of the IKI. This positive covariation indicates that pianists with less pronounced SICI (i.e. larger conditioned MEP measure at 1 and 3 ms ISIs) showed a larger inter‐trial variability of the keypresses. However, the ICF did not covary with the inter‐trial variabilities of any of the dexterity measures, whereas the SICI did not covary with the inter‐trial averages (i.e. coefficients were zero in Table 4). Together, the SICI and ICF provided contrasting results regarding predictive values for the variabilities and mean values, respectively, of the motor dexterity parameters. Furthermore, a disease duration affected by FTSD was also related positively to the inter‐trial variability of the IKI, which indicates that pianists suffering from FTSD for a longer duration showed greater variability in the keystrokes.

Table 4.

Results of LASSO multiple regression associating individual movement features of piano keystrokes with TMS results representing SICI and ICF and participants’ profiles

		Median across trials		Variability across trials
		Inter‐keystroke interval coef	Inter‐key‐release interval coef	Inter‐keystroke interval coef	Inter‐key‐release interval coef
Conditioned MEP measures	1 ms ISI	0	0	0.00238	0
	3 ms ISI	0	0	0.00016	0
	10 ms ISI	0	0.00410	0	0
Profiles	Disease duration affected by FTSD	0	0	0.00013	0
	age	0	0	0	0

ISI: inter‐stimulus interval of the paired‐pulse stimulation; coef: coefficient derived from the regression.

Discussion

We assessed both motor cortical excitability and hand motor dexterity in a relatively large number of pianists with unilateral hand FTSD and matched controls with and without a history of piano training. The study revealed detailed neurophysiological and behavioural abnormalities underlying hand FTSD, and these results are partly in agreement with previous observations in patients with FTSD. First, the MD patients exhibited both a reduced SICI and an elevated ICF as physiological markers of M1 malfunction compared with the healthy pianists and non‐musicians. Previous studies have separately reported that patients with WC exhibit abnormally reduced SICI (Ridding et al. 1995a) and elevated ICF (Sommer et al. 2002). To our knowledge, however, our study is the first to demonstrate abnormality in both SICI and ICF, but not LICI, at once in patients with MD through investigating a wide range of ISIs. In addition, a comparison of the patients with both skilled and unskilled individuals, for the first time, exemplified that these abnormalities of cortical excitability reflected pathophysiology of MD, but not motor expertise or musical skills. Second, the patients displayed abnormally inconsistent and prolonged keypresses during piano playing, which was more pronounced at faster tempi. This finding is not only consistent with the notion that dystonic symptoms involve slowness and timing variability of dexterous movements (Jabusch et al. 2004; Furuya & Altenmüller, 2013a, b ), but also new with respect to the disruptive effects of tempo on dexterity. Taking advantage of those rigorous measurements, our study extended the previous findings by identifying an association between dexterity loss at the behavioural level and M1 malfunctions at the physiological level. Specifically, the reduced M1 inhibition was associated with the temporal imprecision of the finger movements, while the elevated facilitation was related to the impairment of smooth and quick transitions of finger movements from flexion to extension (as explained in detail below). Namely, we newly identified two sets of behavioural–physiological covariations as hallmarks of hand FTSD, which will advance our understanding of the pathophysiology underlying FTSD because these findings provide novel evidence connecting distinct types of M1 malfunctions with distinct aspects of motor dexterity degradation in FTSD patients. Our finding linking the inter‐individual differences in multifaceted features of hand motor dexterity to a neurophysiological measure obtained at rest potentially implicates that the association between the two exists even when not performing a task. Importantly, we were able to associate behavioural abnormality with TMS measures at rest. TMS measures at rest have an advantage over TMS measures during movement because a TMS measure at rest is not influenced by the development of dystonic symptoms during the particular testing trials.

In the present behavioural analyses, pianists with MD exhibited more pronounced timing variabilities when both pressing and lifting their fingers than the healthy pianists. A similar finding was reported in a previous study that investigated finger movements during piano playing in pianists with MD (Jabusch et al. 2004). Possible mechanisms underlying increased movement variability in MD include augmented signal‐dependent noise in motor commands (Faisal et al. 2008) and/or loss of independent movement control between fingers (Furuya et al. 2015). Here, the novel behavioural findings are the prolonged IKI, IRI and finger‐key contact durations in the MD patients when playing at the fastest tempo (Fig. 2). Contact duration is prolonged primarily when the initiation of a key‐release is delayed, which can be ascribed to an abnormally sluggish transition from flexion to extension. The prolonged contact duration is unlikely to have resulted from earlier initiation of key‐pressing, due to our finding of the prolonged interval observed between successive key‐presses. Our findings therefore suggest that the initiation of finger‐lifting from the key is delayed only while playing fast, which corroborates the pronounced abnormalities of finger kinematics in MD musicians when playing fast (Furuya & Altenmüller, 2013b). The delayed initiation of lifting the affected finger also prolongs the IRI between the affected and unaffected fingers, which explains the prolonged IRI of the patients. An advantage of investigating pianists’ MD is the ability to provide detailed assessment of motor symptoms, which is difficult in other forms of FTSD such as embouchure dystonia and WC.

Intracortical inhibition within M1 plays a role in selective muscular activation (Stinear & Byblow, 2003) and in preventing unwanted activation of muscles (Coxon et al. 2007), both of which are responsible for independent movement control across fingers. Reduced SICI has been reported in patients with WC at rest (Ridding et al. 1995a) and/or during movement preparation, initiation and production (Stinear & Byblow, 2004b; Beck et al. 2008). An EEG study also reported absence of oscillatory activities associated with action inhibition in patients with hand FTSD (Hummel et al. 2002). The reduced motor cortical inhibition in musicians’ dystonia can also be related to genetic predisposition (Baumer et al. 2016). A novel finding of the present study is that the reduced SICI was related to increased variability of the timing control of sequential finger movements in pianists with MD. In a previous kinematic study of piano keystrokes, MD patients with pronounced timing variability displayed reduced independent control of movements between the affected finger and a finger striking a key (Furuya et al. 2015). Together, the reduced SICI is a candidate biomarker of an FTSD symptom, impairment of individuated finger movements, which possibly underlies the temporally variable keystrokes of pianists with MD. This idea corroborates the loss of surround inhibition in patients with FTSD (Sohn & Hallett, 2004a; Rosenkranz et al. 2005), although the relationship between SICI and surround inhibition remains unclear.

Motor cortical facilitation plays a role in the rapid performance of sequential finger movements (Karni et al. 1995). Neuroimaging studies have demonstrated elevations in M1 activity in proportion to movement speed, in order to accommodate the neuromuscular demands of producing larger muscular activity necessary for driving and taming movement (Schlaug et al. 1996) and/or the accuracy demands associated with faster movement (Furuya & Soechting, 2012). However, fast finger movements require the quick reciprocal activation and deactivation of finger flexor and extensor muscles. Indeed, the quick deactivation of the finger muscles responsible for a keystroke has been demonstrated to be necessary for nimble piano performance (Winges et al. 2013). The exaggerated ICF of M1 innervating the FDS may impair the inhibition of the finger flexors with the desired timing and thereby delay the initiation of finger extension. This effect may be particularly influential during fast piano playing, which requires the rapid reversal of movement direction from flexion to extension. This abnormality probably resulted in prolonged finger‐key contact and inter‐key‐release interval in the FTSD pianists. Indeed, pianists with hand FTSD typically exhibit limited finger extension (Furuya et al. 2015) and exaggerated finger flexion (Altenmüller et al. 2012). Intriguingly, tDCS augmenting M1 excitability prolongs and shortens the finger‐key contact durations of healthy expert pianists and non‐musicians, respectively (Furuya et al. 2014). This effect can be explained by the non‐linear, inverted U‐shape relationship between motor cortical excitability and finger movement agility (Furuya et al. 2014); in over‐practised pianists whose M1 excitability is maximally tuned for performance, simply further pushing excitability is not beneficial. We proposed that maladaptation in such an over‐practised condition may underlie the pathophysiology of FTSD (Honda et al. 1998).

Overall, our findings for disinhibited M1 in dystonia corroborate previous observations suggesting the disinhibition of M1 as its pathogenesis. Previous neuroimaging studies have reported abnormal M1 hyperactivation in patients with MD during the performance of motor tasks (Pujol et al. 2000; Haslinger et al. 2010). However, analyses of task‐related activity are inherently difficult to determine whether the aberrant neuronal activities during task performance reflect fundamental malfunctions in specific brain regions or merely epiphenomena of the symptoms. In contrast, the paired‐pulse TMS at rest provides observations independent of task performance. This evaluation can become inaccurate during performing motor tasks because active muscular contraction weakens the neuromodulatory effects of paired‐pulse stimulation (Ridding et al. 1995b). Notably, alterations in resting‐state activities have also been reported in patients with WC. Abnormal functional connectivity was found between the motor and somatosensory cortices, cerebellum, and basal ganglia (Dresel et al. 2014). A possibility remains that the M1 malfunctions observed here were also associated with an aberrant cortical–ubcortical functional networks where patients with FTSD probably have abnormality (Moore et al. 2012).

A more straightforward interpretation is to associate the present findings with abnormality in excitatory–inhibitory (EI) circuits within M1. First, strong evidence is available to ascribe our findings of M1 disinhibition to abnormality in M1 local circuits in dystonia (Hallett, 2011). Previous studies suggest that, although both SICI and ICF occur within M1, distinct circuits mediated them (Chen, 2004). For example, SICI and ICF were probably mediated by GABA_A receptors and glutamate, respectively (Liepert et al. 1997; Schwenkreis et al. 2000; Chen, 2004), and SICI but not ICF entails suppression of late I‐waves (Di Lazzaro & Rothwell, 2014). The differential (yet interacted) control of SICI and ICF is consistent with our finding that these two explained different aspects of dexterity. Little is known about the relationship between the balance of M1 local circuits and dexterity in humans. Intriguingly, a precisely balanced excitation and inhibition in a local neural circuit plays a role in noise reduction and adjustment of the signal‐to‐noise ratio of cortical activity (Logothetis, 2008). EI imbalance augments noise in cortical activity, which probably compromises fine motor control (Churchland et al. 2006; Faisal et al. 2008). The M1 EI balance, which accounts for both fine motor control and SICI/ICF, plausibly distribute in succession across healthy and affected individuals (Sommer et al. 2002). Thus, individual differences in EI balance at M1 local circuits may account for the present observation of a large individual difference in the movement variability across the healthy and affected pianists. Our findings suggest that an abnormal shift of EI balance at M1 local circuits toward disinhibition predisposes towards FTSD.

A few technical considerations and interpretational issues should be mentioned regarding abnormalities in SICI at 1 and 3 ms ISIs. First, ‘short‐latency intracortical facilitation (SICF)’ effects may contaminate the measure of SICI at 3 ms ISI. However, this is less likely because we used 90% of the AMT as a conditioning stimulus while SICF is observed when conditioning stimulus intensity is close to or above AMT (Peurala et al. 2008). Second, unlike the previous studies of WC (Stinear & Byblow, 2004b; Beck et al. 2008), we failed to find significant group effects for the SICI measure at 2 ms ISI. This discrepancy can be ascribed to methodological differences between the studies (such as the target muscle), pathophysiological differences of the patients between the studies, or both. Therefore, we consider that the present study extended previous findings in WC (Stinear & Byblow, 2004b; Beck et al. 2008) to MD, lending support for abnormal SICI in FTSD.

In conclusion, the present findings provided evidence that abnormal motor cortical excitability of the patients with MD reflects pathophysiology but not expertise or current skill level. The aberrant cortical excitability was characterized by loss of inhibition and elevation of facilitation within M1, each of which associated with the temporal variability of the movements and the quickness of the transition from flexion to extension, respectively. Identification of two sets of behavioural–physiological covariations as hallmarks of hand FTSD is clinically significant because these findings provide novel evidence connecting distinct types of malfunctions within the motor cortex at rest with distinct aspects of motor dexterity degradation in FTSD patients. Because the present study did not address a causal relationship between behavioural and neurophysiological abnormalities, future studies should investigate effects of restoration of functional abnormalities of M1 on fine motor control in patients with FTSD.

Additional information

Conflict of interest

The authors have no conflicts of interest to declare.

Author contributions

SF participated in the design of the study, carried out both neurophysiological and behavioural experiments, analysed the data, performed the statistical analyses and drafted the manuscript. KU participated in the design and coordination of the study, assisted with the neurophysiological experiments, and helped to interpret the data and draft the manuscript. TS diagnosed the patients and helped to draft the manuscript. TH participated in the design of the study, and helped to interpret the data and draft the manuscript. All authors read and approved the final manuscript. All authors agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Biographies

Shinichi Furuya is an associate researcher at Sony Computer Science Laboratories and a research associate professor at Sophia University in Japan. His research interest is neuroplastic mechanisms subserving acquisition, sophistication, loss and restoration of sensorimotor skills of musicians.

Kazumasa Uehara is a post‐doctoral researcher studying human motor control in the Neural Control of Movement Laboratory, School of Biological and Health Systems Engineering, Arizona State University, USA, from 2016. He obtained his PhD in human motor control from Hiroshima University, Japan, in 2013. His research interests are cortical control of movement by using techniques based on non‐invasive brain stimulation, neuroimaging and behavioural studies.