Brain Activity in Patients With Adductor Spasmodic Dysphonia Detected by Functional Magnetic Resonance Imaging

ARTICLE IN PRESS Brain Activity in Patients With Adductor Spasmodic Dysphonia Detected by Functional Magnetic Resonance Imaging Asanori Kiyuna, Norimoto Kise, Munehisa Hiratsuka, Shunsuke Kondo, Takayuki Uehara, Hiroyuki Maeda, Akira Ganaha, and Mikio Suzuki, Nishihara-cho, Okinawa, Japan Summary: Objectives. Spasmodic dysphonia (SD) is considered a focal dystonia. However, the detailed pathophysiology of SD remains unclear, despite the detection of abnormal activity in several brain regions. The aim of this study was to clarify the pathophysiological background of SD. Study Design. This is a case-control study. Methods. Both task-related brain activity measured by functional magnetic resonance imaging by reading the fivedigit numbers and resting-state functional connectivity (FC) measured by 150 T2-weighted echo planar images acquired without any task were investigated in 12 patients with adductor SD and in 16 healthy controls. Results. The patients with SD showed significantly higher task-related brain activation in the left middle temporal gyrus, left thalamus, bilateral primary motor area, bilateral premotor area, bilateral cerebellum, bilateral somatosensory area, right insula, and right putamen compared with the controls. Region of interest voxel FC analysis revealed many FC changes within the cerebellum-basal ganglia-thalamus-cortex loop in the patients with SD. Of the significant connectivity changes between the patients with SD and the controls, the FC between the left thalamus and the left caudate nucleus was significantly correlated with clinical parameters in SD. Conclusion. The higher task-related brain activity in the insula and cerebellum was consistent with previous neuroimaging studies, suggesting that these areas are one of the unique characteristics of phonation-induced brain activity in SD. Based on FC analysis and their significant correlations with clinical parameters, the basal ganglia network plays an important role in the pathogenesis of SD. Key Words: Spasmodic dysphonia–Resting-state functional connectivity–Disease severity–Basal ganglia network–Focal dystonia.

INTRODUCTION Spasmodic dysphonia (SD) has unique clinical characteristics, such as irregular movement of the vocal folds during speech production and a strained or strangled, hoarse, and effortful voice with breaks in pitch and phonation.1 The clinical manifestations of the disorder and various experimental results have indicated that SD is a focal dystonia.2 Although this hypothesis has been widely accepted, the detailed pathophysiology of SD remains unclear, despite the detection of abnormal activity in several brain regions. In recent years, less invasive investigation of brain activity has become possible using various functional brain imaging techniques. Numerous studies have reported the brain activity of patients with various focal dystonias,3 revealing abnormal (increased or reduced) brain activities in the primary sensory and motor cortices, accessory motor cortices, basal ganglia, thalaAccepted for publication September 16, 2016. Author contributions: A.K., N.K., and M.H.: acquisition of experimental and clinical data, supervision of the experiments, and preparation of the manuscript; S.K.: experimental studies and data acquisition; T.U., H.M., and A.G.: recruitment of patients, collection of clinical data, patient follow-ups; and M.S.: study design, supervision of experiments, and manuscript review. From the Department of Otorhinolaryngology, Head and Neck Surgery, Graduate School of Medicine, University of the Ryukyus, 207 Uehara, Nishihara-cho, Okinawa 903-0215, Japan. Address correspondence and reprint requests to Mikio Suzuki, Department of Otorhinolaryngology, Head and Neck Surgery, Graduate School of Medicine, University of the Ryukyus, 207 Uehara, Nishihara-cho, Okinawa 903-0215, Japan. E-mail: suzuki@ med.u-ryukyu.ac.jp Journal of Voice, Vol. ■■, No. ■■, pp. ■■-■■ 0892-1997 © 2016 The Voice Foundation. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jvoice.2016.09.018

mus, and cerebellum. The brain regions that are affected are consistent among the various forms of focal dystonia. An abnormality in the motor loop, namely a mismatch between sensory input and motor output, in the basal ganglia might underlie the pathophysiology of dystonia.4 Abnormal activity of the cerebellum might also be associated with dystonia.5,6 Several studies have reported the functional imaging findings of SD, indicating abnormal (increased or reduced) brain activities compared with healthy controls in the cerebellum, basal ganglia, thalamus, sensorimotor area, insula, auditory cortex, supplementary motor area (SMA), and anterior cingulate cortex (ACC).1,7–10 The findings resemble those of a task-related functional magnetic resonance imaging (fMRI) study of other types of focal dystonias.3 However, the increased and reduced brain activities were not consistent among the above reports, possibly owing to differences in the disease severity and duration in the subjects and the tasks used.11 Resting-state fMRI measures spontaneous low-frequency fluctuations in blood oxygen level–dependent (BOLD) contrast, and it can detect the functional architecture of the brain. Application of this technique has allowed for the identification of various resting-state networks and spatially distinct areas of the brain, demonstrating synchronous BOLD fluctuations at rest. 12 Neuroimaging research focusing on resting-brain activity, when subjects receive no external stimulation, has been increasing.13 A statistically significant overlap between resting-state functional connectivity (FC) and task-activation maps was obtained.14 There have been some reports of resting-state FC in patients with dystonia, but there are no such reports in patients with SD.

ARTICLE IN PRESS 2

Journal of Voice, Vol. ■■, No. ■■, 2016

TABLE 1. Case Profiles

Case 1 2 3 4 5 6 7 8 9 10 11 12

Age at fMRI Examination (y) 43 66 34 37 36 26 23 29 33 30 23 32

Sex

G Rating

Overall Severity Scale

Disease Duration (mo)

F F F F F F F F F F F F

1 3 2 2 2 3 2 1 2 1 1 3

3 6 5 5 5 7 5 3 6 4 2 7

82 74 158 120 145 82 26 12 162 121 63 24

Thus, the aim of this study was to clarify the pathophysiological background of SD by measuring task-related brain activity and resting-state FC. METHODS Subjects The task-related fMRI study and resting-state FC study enrolled 12 patients with adductor SD (ADSD) (12 women; mean age, 34.3 years old; age range, 23–66 years) and 16 healthy controls (16 women; mean age, 33.1 years old; age range, 22–51

years). None of the subjects in the control group had any previous history of neurologic, psychiatric, or voice disorders. All the participants in this study were strictly right-handed. ADSD was diagnosed by otolaryngologic examinations and speech-language assessments as follows: a choked, strained or strangled voice with intermittent breaks in phonation, no anatomic abnormality of the larynx observed on fiberoptic laryngoscopy with stroboscopy, disease duration of at least 1 year (Table 1), poor improvement despite voice therapy, and symptom alleviation using a whisper or high-pitched voice. The disease severity of patients with ADSD was evaluated by the G rating of the GRBAS classification (G, grade; R, rough; B, breathy; A, asthenic; S, strained)15 and the overall severity scale of the Unified Spasmodic Dysphonia Rating Scale,16 as shown in Table 1. The study protocol was approved by our institutional review board. All the participants provided written informed consent according to the guidelines of the Ethics Committee. This study was conducted in accordance with the principles of the Declaration of Helsinki. Task-related fMRI study Task design The experimental task consisted of alternating “phonation” and “no vocalization” conditions, as shown in Figure 1. In the phonation task, the subjects read and pronounced five-digit numbers in Japanese, for example, 1-2-3-4-5. This process is shown in Figure 1 as /xxxxx/ (five-digit number) under the speaker symbol (gray representation against a white background). All the

FIGURE 1. Schematic illustration of task design in the task-related functional magnetic resonance imaging study. The experimental task consisted of alternating vocalization (the reading of five-digit numbers; /xxxxx/) and no vocalization (rest) for 3 seconds to minimize scanner noise. Functional magnetic resonance imaging (fMRI) scans were acquired within the first 2.7 seconds (TA = acquisition time) of each interscan interval of 10 seconds. The 3-second experimental task (stimulus) occurred at six time points with a 300-ms interval within the 10-second interscan period. A prescanning delay between the end of the experimental stimulus and the start of single-volume MR scanning was varied stepwise between 1.5 and 3 seconds.

ARTICLE IN PRESS Asanori Kiyuna, et al

Brain activity in adductor spasmodic dysphonia

five-digit numbers were different from the others in a run. The cues for both phonation and no vocalization were randomly presented to subjects through goggles mounted on the head coil. Before the fMRI experiment, it was confirmed that each subject could see the visual cues and respond to them promptly. The patient’s voice was recorded during the examination to confirm phonation task performance. The subjects wore sound-attenuating earmuffs to minimize background noise from the scanner. The fMRI acquisition was externally triggered by a rectangular pulse from the first computer adjusted to the visual stimulus presentation, which was produced by the second personal computer running Presentation software (Neurobehavioral Systems, Berkeley, CA). Data acquisition and analysis Whole-brain functional (T2*-weighted) images, followed by anatomic three-dimensional images of the entire brain (repetition time [TR]/echo time [TE], 7.0 ms/3.0 ms; flip angle, 15°; matrix, 256 × 256; field of view, 240 mm; slice thickness, 1.4 mm; interslice gap, 0 mm), were acquired with a Discovery MR750 3-Tesla system (GE Healthcare Japan, Tokyo, Japan). The functional images obtained consisted of single-shot gradient echo planar imaging volumes parallel to the anterior-posterior commissure plane that was sensitive to BOLD contrast (TE, 30 ms; acquisition time, 2.7 seconds; TR, 10 seconds; flip angle, 70°; matrix size, 64 × 64; field of view, 192 mm; slice thickness, 3 mm; interslice gap, 0 mm; number of slices, 42). The relationship between the tasks and the scan timing is shown in Figure 1. A measurement protocol consisting of a short scanning period of 2.7 seconds and a long interval of 7.3 seconds was used to minimize the influence of background noise from the MR scanner and body movement due to phonation (Figure 1). We performed two functional runs per subject. Each run consisted of the acquisition of 37 whole brain volumes (=scans) at a constant TR of 10 seconds but at systematically varied intervals after a single 3-second stimulus presentation. The cue for “ON” (phonation) or “OFF” (no phonation) was presented to subjects 4.5–6.0 seconds before starting the scan, with successive cues presented with durations of 3 seconds each. Upon presentation of the “ON” cue (a gray speaker symbol against a white background), the subject vocalized for the duration of the cue presentation; for the “OFF” cue (a circle-backslash symbol), the subject rested. The first scan in each run was discarded and was not used in the analysis. The stimulus conditions were distributed pseudorandom over the scans: the phonation run comprised 18 events of the phonation condition and 18 events of the rest condition. The timing of stimulus presentation relative to the subsequent single scan acquisition was chosen to cover the expected maximum of the hemodynamic response function at 5–8 seconds after stimulus onset.1 The prescanning delay between the end of the 3-second experimental stimulus presentation and the start of single-volume MR scanning was stepwise varied between 1.5 and 3 seconds (six imaging time points with a 300-ms interval). To maintain a constant TR of 10 seconds, a postscanning delay ranging from 2.8 to 1.3 seconds was inserted between the end of MR data acquisition and the next stimulus presentation (Figure 1).

3

Image processing and statistical analyses were carried out using Statistical Parametric Mapping 8 (SPM8) software (Wellcome Department of Imaging Neuroscience, University College London, UK) and were implemented in MATLAB R2015a (MathWorks, Natick, MA). All the volumes were realigned to the second volume, corrected for motion artifacts, coregistered with the subject’s corresponding anatomic image, resliced, normalized into a standard stereotactic space (Montreal Neurological Institute [MNI] template) using nonlinear transformations, and smoothed with an 8-mm full-width-at-half-maximum Gaussian kernel. Activated voxels were detected by fitting a general linear model to the time series data. We defined as the reference waveform, without hemodynamic fluctuation related to the tasks, a design matrix consisting of BOLD contrasts modeling the alternating periods of baseline and activation in each subject using a finite impulse response model.1 A parametric map of the t statistic was generated for each voxel, and global effects were removed by proportional scaling. A random effects approach was used for statistical analysis. The corresponding contrast images were entered into a secondlevel (random effects) analysis for group comparison. Withingroup analyses were conducted by applying the one-sample t test (P < 0.001, uncorrected). Group comparisons between the SD and the control groups were calculated using the twosample t test (P < 0.001, uncorrected). Subject age was input as a covariate in the group comparison analysis. The anatomic localization of significant clusters was investigated with the SPM Anatomy Toolbox.17 The locations of significant signal increases in the group analyses were recorded in terms of the anatomic region, Brodmann area (BA), local maxima of the cluster (MNI coordinates), t value of the cluster, and cluster size. Resting-state FC During the experimental protocol of the resting-state fMRI, the subjects received no external stimulation and were instructed to rest quietly with their eyes closed and not to fall asleep during the scan. Using the same MR scanner, 150 T2-weighted echo planar images (TE, 30 ms; TR, 2 seconds; flip angle, 70°; matrix, 64 × 64; field of view, 256 mm; slice thickness, 4 mm; interslice gap, 0 mm; number of slices, 42; scan time, 5 minutes 10 seconds) covering the whole brain were acquired. Data were preprocessed in SPM8, based on MATLAB. The functional data were realigned to correct for head motion, coregistered to the structural image, and normalized to the MNI space using SPM8. The normalized images were smoothed with an isotropic Gaussian kernel of 8-mm full-width-at-halfmaximum. A region of interest (ROI)-based analysis of restingstate FC was performed in each subject using the CONN toolbox v.14.p.18 Before correlation analysis, average signals from white matter and the ventricles were removed from the data using linear regression. Cerebral blood flow volumes were filtered using a low-pass (0.008 Hz < f < 0.09 Hz) filter. Seed-to-voxel FC analysis was performed using each a priori ROI as a seed. The subsequent bilateral seed ROIs were selected from the automated anatomic labeling atlas provided by the WFU PickAtlas toolbox (http://www.nitrc.org/projects/wfu_pickatlas)19 because

ARTICLE IN PRESS 4 these areas showed abnormal activation in patients with SD in previous studies.1,7–10 The areas in question were the precentral gyrus, postcentral gyrus, SMA, inferior frontal operculum, insula, superior temporal gyrus, transverse temporal gyrus, anterior cingulate gyrus, precuneus, cerebellum, and subcortical nucleus (thalamus, caudate nucleus, putamen, and pallidum). This analysis produced Fisher r-to-z transformed correlation maps for each participant and seeds that subsequently were subjected to independent t tests. Type I error was controlled through the use of cluster-level false discovery rate correction (P < 0.05). FC values (mean z-scores) for significant clusters were extracted using the ROI extraction toolbox. The differences in FC maps between the SD and the control groups were calculated by contrasting the ROI-specific FC maps in a second-level group analysis.20 A positive FC (+FC) indicated a positive correlation of BOLD signal time courses between two areas, whereas negative FC (–FC) indicated anti-correlated time courses. Following descriptive statistics for demographic variables, characterization of the relationships of individuals’ FC values (ie, seed-to-cluster z-scores) with the G rating of the GRBAS and over-severity scale in the Unified Spasmodic Dysphonia Rating Scale was conducted using nonparametric (Spearman rank correlation coefficient)

Journal of Voice, Vol. ■■, No. ■■, 2016

correlation matrices. The relationship between individuals’ FC values and disease duration was also investigated using parametric (Pearson correlation coefficient) correlation matrices. RESULTS Task-related fMRI study Brain activity of the control group when reading the five-digit numbers In the control subjects, brain activity associated with phonation was widely observed in the bilateral cerebral cortex, subcortical nucleus, and cerebellum (Figure 2A and Table 2). These areas of activation were divided into the following areas in accordance with their function: primary and secondary motor areas (BA 4 and 6), motor speech area (BA 44), SMA (BA 6), primary and secondary somatosensory areas (BA 1, 3, and 43), somatosensory association cortex (BA 7), auditory association area (BA 41, 42, 21, and 22), ACC (BA 24, 32, and 33), insular cortex (BA 13), precuneus (BA 7), subcortical areas (thalamus, caudate nucleus, putamen, pallidum, and amygdala), and cerebellum. All of the activated areas were distributed bilaterally and symmetrically.

FIGURE 2. (A) Areas of activation in the control group when reading the five-digit numbers. (B) Areas of activation in the patients with adductor spasmodic dysphonia (SD) when reading the five-digit numbers. (C) Areas of increased activation in patients with adductor SD compared with the controls when reading the five-digit numbers. (D) Areas of reduced activation in the patients with SD compared with the controls when reading the five-digit numbers. A, anterior; Cer, cerebellum; SMA, supplementary motor area; SMC, sensorimotor cortex; R, right.

ARTICLE IN PRESS Asanori Kiyuna, et al

5

Brain activity in adductor spasmodic dysphonia

TABLE 2. Areas of Activation in Healthy Controls when Reading the Five-digit Numbers t Value

Local Maxima of Cluster MNI Coordinates (x, y, z)

Cluster Size

Location (Brodmann Area)

13.57

3289

52

8

−8

12.96

6976

16

−64

−24

12.55

942

0

8

44

12.3

3175

−48

−24

8

87 29 4 7 3 5 3

−2 −2 0 2 −20 −28 26

−52 −86 −36 −16 −22 −16 −2

72 44 76 42 78 −2 −10

5.35 4.65 4.44 4.31 4.28 4.19 3.84

Brain activity in the SD group when reading the five-digit numbers In the patients with ADSD, brain activity associated with phonation was also observed in the bilateral cerebral cortex, subcortical nucleus, and cerebellum (Figure 2B and Table 3). These areas of activation were divided into the following areas: primary and secondary motor areas (BA 4 and 6), motor speech area (BA 44), SMA (BA 6), primary and secondary somatosensory areas (BA 1, 3, and 43), somatosensory association cortex (BA 7), auditory association area (BA 41, 42, 21, and 22), ACC (BA 24, 32, and 33), insular cortex (BA 13), subcortical nucleus (thalamus, putamen, caudate nucleus, pallidum, and amygdala), and cerebellum. All of the activated areas were observed bilaterally and symmetrically and resembled those in the control group. Differences in brain activity between patients with ADSD and controls Increased brain activity in the patients with SD compared with the controls was observed in the left middle temporal gyrus (BA 21), left thalamus, bilateral primary and secondary motor areas (BA 4 and 6), right insula (BA 13), bilateral primary somatosensory area (BA 1 and 3), bilateral cerebellum (I–IV, VIII, IX), right putamen, and right SMA (BA 6) (Figure 2C and Table 4). The bilateral cerebellum (Crus I, II, VI) and left superior temporal gyrus (40) showed reduced brain activity in the patients with ADSD compared with the controls (Figure 2D and Table 5).

R superior temporal gyrus (21, 22) R postcentral gyrus (1, 3) R rolandic operculum (43) R inferior frontal operculum (44) R Heschl gyrus (41, 42) R precentral gyrus (4, 6) R insula (13) R and L cerebellum I–VIII, Crus 1, 2 R and L thalamus R and L caudate nucleus L and R supplementary motor area (6) L and R anterior cingulate gyrus (24, 32, 33) L superior temporal gyrus (21, 22) L Heschl gyrus (41, 42) L postcentral gyrus (1, 3) L inferior frontal operculum (44) L precentral gyrus (4, 6) L rolandic operculum (43) L insula (13) L and R precuneus (5, 7) L precuneus (7) R precentral gyrus (4) R anterior cingulate cortex (24) L precentral gyrus (6) L amygdala R amygdala

Resting-state FC Increased FC in patients with SD compared with controls (Table 6) An increased FC in the patients with SD relative to the controls was observed in the left thalamus to the left caudate; the right precentral gyrus to the left inferior temporal gyrus, temporal pole, and middle temporal gyrus; the left postcentral gyrus to the right frontal pole; the left inferior operculum to the right precentral gyrus and postcentral gyrus; the cerebellum (vermis I, II) to the right lateral occipital cortex and superior parietal lobule; and the right cerebellum (IX) to the right postcentral gyrus and precentral gyrus. Reduced FC in patients with SD compared with controls (Table 7) Reduced FC in the patients with SD relative to the controls was observed in the left insula to the right angular gyrus and lateral occipital cortex; the right thalamus to the left middle frontal gyrus and inferior frontal gyrus; the left precuneus to the bilateral lingual gyrus; and the right precentral gyrus to the right occipital pole. Correlations between the FC results in patients with SD and the clinical manifestations of SD The relationships of individuals’ FC values (Tables 6 and 7) with a G rating, over-severity scale, or disease duration were investigated. Of these relationships, the increased FC observed in the

ARTICLE IN PRESS 6

Journal of Voice, Vol. ■■, No. ■■, 2016

TABLE 3. Areas of Activation in Patients With Spasmodic Dysphonia when Reading the Five-digit Numbers t Value

Local Maxima of Cluster MNI Coordinates (x, y, z)

Cluster Size

Location (Brodmann Area)

15.04

3620

−64

−6

16

14.52

2810

62

−4

22

13.4 10.73

1093 956

−30 2

−62 0

−26 68

8.55

2016

14

−20

2

5.94 5.64 5.04 4.91 4.76 4.5 4.22

74 20 2 8 1 3 1

4 −2 26 24 −24 −16 16

−58 −82 −28 −12 −2 −26 −26

2 −26 72 0 −16 18 18

left thalamus to the left caudate nucleus was significantly correlated with the G rating of the GRBAS (Figure 3, ρ = 0.62). DISCUSSION The production of speech requires activation of the primary laryngeal motor cortex (LMC). In addition, it requires the integration

L rolandic operculum (43) L postcentral gyrus (1, 3) L precentral gyrus (4, 6) L superior temporal gyrus (21, 22) L inferior frontal operculum (44) L Heschl gyrus (41, 42) L insula (13) R postcentral gyrus (1, 3) R precentral gyrus (4, 6) R rolandic operculum (43) R superior temporal gyrus (21, 22) R Heschl gyrus (41, 42) R insula (13) L and R cerebellum V–VII, Crus 1 R and L supplementary motor area (6) R and L anterior cingulate gyrus (24, 32, 33) R and L thalamus R and L caudate nucleus R and L pallidum R and L cerebellum I–IV R and L cerebellum I–V L and R cerebellum VI, VII, Crus 1, 2 R precentral gyrus (6) R amygdala L amygdala L thalamus R thalamus

and coordination of multiple brain regions associated with various speech-related processes, such as auditory perception, semantic processing, memory encoding, and preparation for motor execution. These regions contain the sensorimotor, supplementary motor, and auditory association areas, as well as the insula, precuneus, thalamus, basal ganglia, and cerebellum.11,21,22

TABLE 4. Areas With Greater Activation in Patients With Spasmodic Dysphonia Compared With Healthy Controls when Reading the Five-digit Numbers t Value

Local Maxima of Cluster MNI Coordinates (x, y, z)

Cluster Size

Location (Brodmann Area)

5.49 4.56 4.33 4.22 4.22 4.13 4.02

192 40 106 72 29 108 27

−48 −26 −26 40 24 10 54

−4 −28 −16 8 −52 −54 −14

−16 14 64 0 −38 −32 50

3.91 3.64 3.6 3.55 3.5

17 4 3 3 3

−44 26 −30 −10 14

−16 18 −38 −48 8

54 4 52 −22 70

L middle temporal gyrus (21) L thalamus L precentral gyrus (6) R insula (13) R cerebellum VIII R cerebellum IX R precentral gyrus (6, 4) R postcentral gyrus (1) L precentral gyrus (6, 4) R putamen L postcentral gyrus (3) L cerebellum I–IV R supplementary motor area (6)

ARTICLE IN PRESS Asanori Kiyuna, et al

7

Brain activity in adductor spasmodic dysphonia

TABLE 5. Areas With Greater Activation in Healthy Controls Compared With Patients With Spasmodic Dysphonia when Reading the Five-digit Numbers t Value 4.45 3.63 3.71 3.75

Local Maxima of Cluster MNI Coordinates (x, y, z)

Cluster Size 56 13 9 7

−20 24 −64 −16

Although brain activities in these areas, including the LMC, were observed, the activities were found to be different between the patients with SD and the healthy controls in the present taskrelated brain examinations (Table 8). The patients with SD showed significantly higher brain activation in the bilateral insula, bilateral cerebellum, left sensorimotor cortex, right SMA, right cingulate gyrus, right primary somatosensory cortex, left middle temporal gyrus, and right putamen compared with controls in the task of reading the five-digit numbers. These results suggested that patients with SD have increased brain activity in these speech-related regions during speech. In our previous study, using the /i:/ task (continuous vowel), the left primary auditory cortex, bilateral inferior frontal gyrus, left inferior parietal lobule, bilateral insula, and bilateral cerebellum were significantly activated in patients with SD, compared with controls. In contrast, brain activities in the left somatosensory cortex, bilateral putamen, and bilateral pallidum were significantly reduced, compared with controls.10 The most prominent difference between our two studies was observed in the sensorimotor and basal ganglia, whereas the brain activities in the insula and cerebellum were consistent in both studies. Other studies of functional brain activities in patients with SD,7–10 with the exception of Haslinger et al,1 have shown increased brain activity in both the insula and the cerebellum, although the studies have used different imaging methods and tasks, and the patients have had different disease durations and ages. These results suggested that the increased brain activity in the insula and cerebellum could be a characteristic of brain activity during phonation in SD. Haslinger et al1 reported significantly lower brain activity in the primary sensorimotor and premotor cortices and SMA in an fMRI study using a continuous vowel task, as in our previous study.10 In contrast, Simonyan and Ludlow,9 using a repetitive syllable task, reported significantly higher brain activity in the primary sensorimotor cortex. The present experimental results were in accordance with Simonyan and Ludlow’s report.9 The five-digit number phonation task employed in the present study could easily induce typical dysphonia symptoms in the patients with SD, and it induced similar brain activation with the repetitive syllable task.9 The use of different tasks might be the reason for the discrepancies in sensorimotor cortex findings among the fMRI studies. Resting-state fMRI has identified several resting-state networks, consisting primarily of interconnections between cortical regions.23 Brain activity in response to phonation in SD might be influenced by many factors, such as used tasks, cognitive pro-

−90 −80 −28 −70

Location (Brodmann Area) −28 −20 18 −16

L cerebellum Crus 1, 2 R cerebellum Crus 1 L superior temporal gyrus (40) L cerebellum VI

cesses related to the associations between numbers and words, and the hearing of one’s own voice. FC detected by restingstate fMRI has been divided into several heteromodal networks: the default mode network, executive control network, dorsal attention network, salience network, sensorimotor network, auditory network, visual network, cerebellum network, and basal ganglia network.13,23,24 Reports with regard to several types of focal dystonia have shown that focal dystonia caused FC alterations in the sensorimotor network, cerebellum network, and basal ganglia network.25–29 However, the reduced or increased FC patterns in these networks were different among various types of focal dystonia and different studies. For example, in writer’s cramp, lower positive FC of the dorsolateral prefrontal cortex, thalamus, and pallidum to the symptomatic primary sensorimotor cortex was observed, compared with controls.29 In the present study, increased FC was observed between the thalamus and the basal ganglia, between the motor cortex and the auditory associated cortices, between the somatosensory cortex and the frontal lobe, between the motor speech area and the motor cortex, and between the cerebellum and the sensorimotor cortex. Reduced FC was also observed between the insular cortex and the semantic processing area, between the thalamus and the motor speech area, between emotion-related areas, and between the motor area and the visual cortex. These results in SD were considered to be FC abnormalities in the basal ganglia network, sensorimotor network, language network, and cerebellum network in agreement with task-related brain activity. Because phonation is a highly cognitive and human-specific behavior and involves the precise coordination of more than 100 laryngeal, orofacial, and respiratory muscles, the affected FC pattern (increased or decreased) of the cerebellum-basal ganglia-thalamus-cortex network in SD might be different from that in other types of focal dystonia. The thalamus is reciprocally connected with the LMC, whereas the caudate nucleus receives only projection from the LMC.30 The left caudate nucleus, which is a part of corticobasal gangliathalamic loop, is also one of the brain structures that functions as a center for language control.31 During speech, direct outputs from the LMC are relayed through the pyramidal tract to the brainstem. The indirect pathways are via the striatum to the substantia nigra and to the reticular formation in the brainstem in humans. Two extensions of these indirect pathways are through the striatum back to the motor cortex and to the subthalamic nuclei.30 In the present study, of the many significant differences in FC between the SD group and the control group, the FC between the left thalamus and the left caudate nucleus was

8

TABLE 6. Increased Functional Connectivity in Patients With Spasmodic Dysphonia Compared With Controls

Left thalamus

−20

−10

26

268

8.44

Right precentral gyrus

−50

−6

−44

236

6.59

Left postcentral gyrus Left inferior operculum

48

52

−8

304

6.23

44

−12

28

242

5.48

Cerebellum (vermis I, II)

28

−62

56

205

5

Right cerebellum (IX)

44

−20

58

209

4.93

Cluster Healthy Spasmodic Voxels P Value Control Dysphonia in % (P < 0.05 Connectivity Connectivity Region Coverage FDR) Means (SD) Means (SD)

Cluster Regions Left caudate Not assigned or less than 1% coverage Left inferior temporal gyrus, anterior division Left temporal pole Left middle temporal gyrus, anterior division Left inferior temporal gyrus, posterior division Not assigned or less than 1% coverage Right frontal pole Not assigned or less than 1% coverage Right precentral gyrus Right postcentral gyrus Not assigned or less than 1% coverage Right lateral occipital cortex Right superior parietal lobule Not assigned or less than 1% coverage Right postcentral gyrus Right precentral gyrus Not assigned or less than 1% coverage

21 247 98 17 7 7 107 241 63 103 40 99 164 39 2 118 85 6

4 — 29 1 2 1 — 3 — 2 1 — 3 3 — 4 2 —

0.04

−0.01 (0.14)

0.26 (0.14)

0.049

−0.15 (0.17)

0.12 (0.18)

0.017

−0.00 (0.15)

0.22 (0.16)

0.013

−0.10 (0.21)

0.17 (0.22)

0.025

−0.10 (0.19)

0.13 (0.2)

0.046

−0.05 (0.21)

0.19 (0.22)

Abbreviations: MNI, Montreal Neurological Institute; FDR, false discovery rate; SD, standard deviation.

TABLE 7. Reduced Functional Connectivity in Patients With Spasmodic Dysphonia Compared With Controls

Cluster t Size Value

54

−54

46

231

Right thalamus

−50

28

22

244

Left precuneus

0

−86

−10

232

Right precentral gyrus

18

−92

16

182

Left insula

Cluster Regions

−6.65 Right angular gyrus Right lateral occipital cortex Not assigned or less than 1% coverage −5.87 Left middle frontal gyrus Left inferior frontal gyrus, pars triangularis Not assigned or less than 1% coverage −5.64 Right lingual gyrus Left lingual gyrus Not assigned or less than 1% coverage −4.84 Right occipital pole Not assigned or less than 1% coverage

Abbreviations: MNI, Montreal Neurological Institute; FDR, false discovery rate; SD, standard deviation.

Spasmodic Healthy Cluster Dysphonia Control P Value Voxels (P < 0.05 Connectivity Connectivity in % Means (SD) Means (SD) FDR) Region Coverage 113 31 87 117 61 66 89 40 103 134 48

8 1 — 4 9 — 5 3 — 5 —

0.048

0.08 (0.15)

−0.16 (0.16)

0.039

0.10 (0.20)

−0.18 (0.21)

0.032

0.33 (0.22)

0.03 (0.23)

0.038

0.12 (0.25)

−0.16 (0.26)

Journal of Voice, Vol. ■■, No. ■■, 2016

SEED REGION

Local Maxima of Cluster MNI Coordinates (x, y, z)

ARTICLE IN PRESS

Seed Region

Local Maxima of Cluster MNI Coordinates Cluster t (x, y, z) Size Value

Asanori Kiyuna, et al

Present Study

Kiyuna et al (2014)

Simonyan et al (2010)

fMRI 12 89 months Five-digit number, 3 s

fMRI 6 41 months /i:/, 3 s

fMRI 11 15.1 years /i:i/, 5 s

Task cue fMRI design

Visual trigger Sparse sampling design

Increased brain activity

PrG (BA 4, 6), PoG (BA 1, 3), SMA (BA 6), MTG (BA 21), insula (BA 13), Cer (I–IV, VIII, IX), putamen, thalamus Cer (I, II, VI), STG (BA40)

Visual trigger Sparse sampling design IPL (BA 40), TTG (BA 41), MSA (BA 44, 45), IFG (BA 9), MFG (BA 46), insula (BA 13), Cer PoG (BA 3, 1, 2), basal ganglia

Acoustic trigger Sparse sampling design PMC, somatosensory cortex, STG, MTG, operculum, insula, Cer, basal ganglia, thalamus Midbrain

Decreased brain activity

PET 9 11 years Narrative speech

SMC, auditory cortex, ACC, insula, Cer

SMA, posterior supraMG, posterior MTG, PAG

Haslinger et al (2005) fMRI 12 12.5 years /i:/, 3 s Visual trigger Silent event-related design No area

SMC (BA 4, 3), ACC (BA 24, 32), SMA (BA 6), preMC (BA 6), IFG (BA 45), preFC (BA 9, 10), SP&PO (BA 7, 19), FG, paraHG, Cer

Hirano et al (2001) PET 1 >10 years Daily use sentences

Left MA, Broca area, PAA, left STG, Cer SMA

Abbreviations: ACC, anterior cingulate cortex; BA, Brodmann area; Cer, cerebellum; FG, fusiform gyrus; IFG, inferior frontal gyrus; IPL, inferior parietal lobule; MA, motor area; MFG, middle frontal gyrus; MSA, motor speech area; MTG, middle temporal gyrus; PAA, primary auditory area; PAG, periaqueductal gray matter; paraHG, parahippocampal gyrus; PMC, primary motor cortex; PoG, postcentral gyrus; preFC, prefrontal cortex; preMC, premotor cortex; PrG, precentral gyrus; SMA, somatomotor area; SMC, sensorimotor cortex; SP&PO, superior parietal and parieto-occipital cortex; STG, superior temporal gyrus; supraMG, supramarginal gyrus; TTG, transverse temporal gyrus.

ARTICLE IN PRESS

Modality No. of cases Disease duration Task

Ali et al (2006)

Brain activity in adductor spasmodic dysphonia

TABLE 8. Summary of Brain Activity in Patients With Adductor Spasmodic Dysphonia Detected by Functional Magnetic Resonance Imaging (fMRI) and Positron Emission Tomography (PET)

9

ARTICLE IN PRESS 10

FIGURE 3. Correlation between functional connectivity (FC) in spasmodic dysphonia (SD) and the clinical manifestations of SD (overall severity scale). The increased positive FC in the left thalamus to the left caudate nucleus compared with the controls was significantly correlated with the overall severity scale. L, left; ρ, Spearman rank correlation coefficient; P, P value; ROI, region of interest; SD, standard deviation. positively correlated with the clinical parameters (Figure 3). Simonyan et al32 reported that SD showed dopaminergic transmission during symptomatic speaking. Concurrent activation and connectivity between the premotor cortex and the elements of basal ganglia-thalamocortical circuitry were related to an increasing degree of voluntary control over vocalization in humans.33 These results, including those from our FC analysis, suggest that dysfunction of the basal ganglia network could be linked to the clinical features of SD. CONCLUSIONS The higher task-related brain activity in the insula and cerebellum was consistent with previous neuroimaging studies, suggesting that these areas are one of the unique characteristics of phonation-induced brain activity in SD. Based on FC analysis and their significant correlations with clinical parameters, the basal ganglia network plays an important role in the pathogenesis of SD. Acknowledgments This study was supported by the KAKENHI 26462612 grant, which was awarded by the Japan Society for the Promotion of Science to Dr. Kiyuna. We also thank the Ryukyu Society for the Promotion of Oto-Rhino-Laryngology for their writing assistance and for providing technical help. REFERENCES 1. Haslinger B, Erhard P, Dresel C, et al. “Silent event-related” fMRI reveals reduced sensorimotor activation in laryngeal dystonia. Neurology. 2005;65:1562–1569. 2. Ludlow CL, Adler CH, Berke GS, et al. Research priorities in spasmodic dysphonia. Otolaryngol Head Neck Surg. 2008;139:495–505. 3. Zoons E, Booij J, Nederveen AJ, et al. Structural, functional and molecular imaging of the brain in primary focal dystonia—a review. Neuroimage. 2011;56:1011–1020.

Journal of Voice, Vol. ■■, No. ■■, 2016 4. Kaji R, Urushihara R, Murase N, et al. Abnormal sensory gating in basal ganglia disorders. J Neurol. 2005;252(suppl 4):IV13–IV16. 5. Teo JT, van de Warrenburg BP, Schneider SA, et al. Neurophysiological evidence for cerebellar dysfunction in primary focal dystonia. J Neurol Neurosurg Psychiatry. 2009;80:80–83. 6. Sadnicka A, Hoffland BS, Bhatia KP, et al. The cerebellum in dystonia—help or hindrance? Clin Neurophysiol. 2012;123:65–70. 7. Hirano S, Kojima H, Naito Y, et al. Cortical dysfunction of the supplementary motor area in a spasmodic dysphonia patient. Am J Otolaryngol. 2001;22:219–222. 8. Ali SO, Thomassen M, Schulz GM, et al. Alterations in CNS activity induced by botulinum toxin treatment in spasmodic dysphonia: an H215O PET study. J Speech Lang Hear Res. 2006;49:1127–1146. 9. Simonyan K, Ludlow CL. Abnormal structure-function relationship in spasmodic dysphonia. Cereb Cortex. 2012;22:417–425. 10. Kiyuna A, Maeda H, Higa A, et al. Brain activity related to phonation in young patients with adductor spasmodic dysphonia. Auris Nasus Larynx. 2014;41:278–284. 11. Simonyan K, Ludlow CL. Abnormal activation of the primary somatosensory cortex in spasmodic dysphonia: an FMRI study. Cereb Cortex. 2010;20: 2749–2759. 12. Lee MH, Smyser CD, Shimony JS. Resting-state fMRI: a review of methods and clinical applications. AJNR Am J Neuroradiol. 2013;34:1866– 1872. 13. Heine L, Soddu A, Gomez F, et al. Resting state networks and consciousness: alterations of multiple resting state network connectivity in physiological, pharmacological, and pathological consciousness states. Front Psychol. 2012;3:295. 14. Biswal BB, Van Kylen J, Hyde JS. Simultaneous assessment of flow and BOLD signals in resting-state functional connectivity maps. NMR Biomed. 1997;10:165–170. 15. Hirano M. Objective evaluation of the human voice: clinical aspects. Folia Phoniatr (Basel). 1989;41:89–144. 16. Stewart CF, Allen EL, Tureen P, et al. Adductor spasmodic dysphonia: standard evaluation of symptoms and severity. J Voice. 1997;11:95–103. 17. Eickhoff SB, Stephan KE, Mohlberg H, et al. A new SPM toolbox for combining probabilistic cytoarchitectonic maps and functional imaging data. Neuroimage. 2005;25:1325–1335. 18. Whitfield-Gabrieli S, Nieto-Castanon A. Conn: a functional connectivity toolbox for correlated and anticorrelated brain networks. Brain Connect. 2012;2:125–141. 19. Tzourio-Mazoyer N, Landeau B, Papathanassiou D, et al. Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage. 2002;15: 273–289. 20. Boissoneault J, Letzen J, Lai S, et al. Abnormal resting state functional connectivity in patients with chronic fatigue syndrome: an arterial spinlabeling fMRI study. Magn Reson Imaging. 2016;34:603–608. 21. van de Ven V, Esposito F, Christoffels IK. Neural network of speech monitoring overlaps with overt speech production and comprehension networks: a sequential spatial and temporal ICA study. Neuroimage. 2009;47:1982–1991. 22. Silbert LJ, Honey CJ, Simony E, et al. Coupled neural systems underlie the production and comprehension of naturalistic narrative speech. Proc Natl Acad Sci USA. 2014;111:E4687–E4696. 23. Woodward ND, Rogers B, Heckers S. Functional resting-state networks are differentially affected in schizophrenia. Schizophr Res. 2011;130:86–93. 24. Szewczyk-Krolikowski K, Menke RA, Rolinski M, et al. Functional connectivity in the basal ganglia network differentiates PD patients from controls. Neurology. 2014;83:208–214. 25. Jin SH, Lin P, Auh S, et al. Abnormal functional connectivity in focal hand dystonia: mutual information analysis in EEG. Mov Disord. 2011;26:1274– 1281. 26. Mohammadi B, Kollewe K, Samii A, et al. Changes in resting-state brain networks in writer’s cramp. Hum Brain Mapp. 2012;33:840–848. 27. Hinkley LB, Sekihara K, Owen JP, et al. Complex-value coherence mapping reveals novel abnormal resting-state functional connectivity networks in task-specific focal hand dystonia. Front Neurol. 2013;4:149.

ARTICLE IN PRESS Asanori Kiyuna, et al

Brain activity in adductor spasmodic dysphonia

28. Delnooz CC, Pasman JW, Beckmann CF, et al. Altered striatal and pallidal connectivity in cervical dystonia. Brain Struct Funct. 2015;220:513–523. 29. Dresel C, Li Y, Wilzeck V, et al. Multiple changes of functional connectivity between sensorimotor areas in focal hand dystonia. J Neurol Neurosurg Psychiatry. 2014;85:1245–1252. 30. Simonyan K, Horwitz B. Laryngeal motor cortex and control of speech in humans. Neuroscientist. 2011;17:197–208.

11

31. Crinion J, Turner R, Grogan A, et al. Language control in the bilingual brain. Science. 2006;312:1537–1540. 32. Simonyan K, Berman BD, Herscovitch P, et al. Abnormal striatal dopaminergic neurotransmission during rest and task production in spasmodic dysphonia. J Neurosci. 2013;33:14705–14714. 33. Schulz GM, Varga M, Jeffires K, et al. Functional neuroanatomy of human vocalization: an H215O PET study. Cereb Cortex. 2005;15:1835–1847.