Altered putamen functional connectivity is associated with anxiety disorder in Parkinson's disease

In this study, we used resting state-functional magnetic resonance imaging (rs-fMRI) to explore altered putamen functional connectivity (FC) in Parkinson's disease patients with anxiety disorder. We divided 65 Parkinson's disease patients into anxiety (PD-A; n=18) and non-anxiety (PD-NA; n=45) groups based on a Hamilton Anxiety Rating Scale cutoff score of 12. The PD-A patients exhibited altered putamen FC with cortical and subcortical regions. The PD-A patients showed enhanced putamen FC with the caudatum, which correlated with increased emotional processing during anxiety. Decreased putamen FC with the orbitofrontal gyrus and cerebellum also correlated with increased anxiety in Parkinson's disease. Our findings demonstrate that anxiety disorder in Parkinson's disease is associated with abnormal putamen FC networks, especially with caudatum, orbitofrontal gyrus and cerebellum.


INTRODUCTION
Anxiety syndrome is more prevalent in Parkinson's disease (PD) patients (6%-55%) [1] than in the general population and non-PD patients [2,3]. Furthermore, anxiety in PD is associated with increased motor symptoms and negatively impacts quality of life [2]. Anxiety is clinically under-diagnosed in PD patients and is neglected while researching PD pathology.
In recent years, neuroimaging technology has been widely used to investigate functional and structural alterations of motor and non-motor symptoms in PD brains [4,5]. Positron emission computed tomography (PET), single photon emission computed tomography (SPECT) and T1 structural magnetic resonance imaging (MRI) have been used to image individual brain regions while investigating anxiety disorders in PD [4]. Most PET/SPECT reports show correlation between putamen and severity of anxiety [6][7][8][9]. However, the association between anxiety in PD and alterations in putamen FCfunctional connectivity (FC) with other brain regions is unknown.
Resting state-functional MRI (rs-fMRI) measures the resting blood oxygen level-dependent signal to determine spontaneous brain activity between functionally linked brain regions at the level of neural networks. FC indicates inter-regional temporal patterns of the blood oxygen level-dependent signal [10]. Recently, rs-fMRI has been extensively used to explore neurological and psychiatric diseases [11,12], including various motor [13] and non-motor symptoms [14][15][16] in PD.
The aim of this study was to use rs-fMRI to analyze differences in the putamen FC patterns among 3 study subject groups: , namely, PD patients with anxiety (PD-A), PD patients without anxiety (PD-NA) and healthy controls.

Study subjects
Sixty-five right-handed PD patients were recruited from the outpatient clinic of our hospital and diagnosed Clinical Research Paper with idiopathic PD according to the UK Parkinson's Disease Society Brain Bank criteria [17]. The following patients were excluded from the study: (1) patients with corticobasal degeneration, progressive supranuclear palsy, multiple system atrophy, vascular parkinsonism, and other forms of parkinsonism; (2) patients with contraindications to MRI such as claustrophobia, metallic implants or devices in the body; (3) patients with severe cognitive decline indicated by Mini Mental State Examination (MMSE) score < 24; (4) patients with clinically prominent depressive symptoms (17-item Hamilton Depression Rating Scale, HAMD score > 14) for controlling possible hybrid factors. None of the patients took anti-psychotic drugs. MRI scans and clinical examinations were performed at least 12 h after withdrawal from medications to mitigate the pharmacological effects on neural activity. Meanwhile, 24 healthy controls without psychological and neurological disturbances or neuroimaging abnormalities were recruited for this study. Written informed consent was obtained from all participants. The study was approved by the ethics committee of The First Affiliated Hospital of Nanjing Medical University.

Clinical assessment of study subjects
PD patients were classified into PD-A (n = 18) and PD-NA (n = 45) groups based on the cut off score of 12 on the Hamilton Anxiety Rating Scale (HAMA) score [18][19][20]. The HAMA scale is a fourteen-item scale (total scores ranging 0-56) that measures both psychic and somatic anxiety symptoms [18], which is useful in evaluating anxiety severity in clinical practice and research [19]. Anxiety disorders of PD patients were assessed by a trained movement disorder specialist (Kezhong Zhang), sensitized to psychiatric disorders in PD. The Unified Parkinson's Disease Rating Scale (UPDRS-III) and Hoehn & Yahr (H&Y) staging scales were used to assess PD motor severity and disease stage, respectively. Total levodopa-equivalent daily dose (LEDD), LEDD of levodopa preparations and LEDD of dopamine receptor agonists of each PD patient were calculated as previously described [21]. Disease duration of PD patients and education levels of all subjects were recorded. Cognitive function, executive function and symptoms of depression were also quantified separately by MMSE, Frontal Assessment Battery (FAB) and HAMD.

Rs-fMRI data analysis
The rs-fMRI data was analyzed by the data processing assistant for resting-state fMRI (DPARSF, http://www.restfmri.net/forum/dparsf) [22] with Statistical Parametric Mapping (SPM8, http://www.fil. ion.ucl.ac.uk). The analysis included 1) removal of the first 10 time points; 2) slice timing correction; 3) head motion correction via six-parameter rigid body spatial transformation during data acquisition; 4) nonlinear registration of the high-resolution T1 structural images to the Montreal Neurological Institute (MNI) template and segmenting them into white matter, gray matter, and cerebrospinal fluid using the DARTEL (diffeomorphic anatomical registration through exponentiated lie algebra) algorithm [23] followed by further structural analyses of the resulting segments; 5) nuisance signal removal (white matter, cerebrospinal fluid, global signal, 6-head motion parameters as covariates) via multiple regression; 6) spatial normalization to the Montreal Neurological Institute template; 7) resampling images into a spatial resolution of 3×3×3 mm 3 ; 8) spatial smoothening with a Gaussian kernel (full width at half-maximum = 4×4×4 mm 3 ); 9) temporal bandpass filtering (0.01 < f < 0.08 Hz) and linear detrended removal. We excluded subjects from further analysis if the translation or rotation of head movement was > 2 mm or 2° in any direction. Additionally, the mean head translation, mean head rotation, and framewise displacement were calculated [24]. There were no differences in head motion parameters among the three groups (p > 0.05).

Functional connectivity analysis
The left and right putamen were defined as two regions of interest (ROI) based on the Anatomical Automatic Labeling (AAL) template by selecting specific areas to study anxiety-dependent resting-state functional networks in PD patients. A voxel-wise FC analysis was performed by computing the temporal correlation www.impactjournals.com/oncotarget between the mean time series of each ROI and the time series of each voxel within the brain. Pearson's correlation coefficient maps were created for each individual subject and were converted to a z-value using Fisher's z transformation.
Analysis of covariance (ANCOVA) was performed to identify brain areas with differences between the three study subject groups with age, gender, education, HAMD scores and gray matter volume as covariates (voxel-level p < 0.05; cluster size > 113 voxels; corrected p < 0.01 as determined by AlphaSim correction). These areas were then extracted as a mask. Then, two-sample post hoc t tests were performed within this mask, with age, gender, education, HAMD scores and gray matter volume as covariates, to detect differences between study subject groups (voxel-level p < 0.05; cluster size of the left putamen > 13 voxels; cluster size of the right putamen > 16 voxels; corrected p < 0.01 as determined by AlphaSim correction). Additionally, for each ROI, the clusters that showed differences in FC between PD-A and PD-NA groups were extracted separately. Then the average FC value of each cluster was calculated to estimate the correlation of anxiety severity. The Pearson correlation between mean FC values and HAMA scores was calculated by IBM SPSS statistics v20.0.0 software (SPSS, Chicago, IL, USA) and the significance was set at p < 0.01 (two-tailed; http://afni.nimh.nih.gov/pub/dist/doc/ manual/AlphaSim.pdf).

Statistical analysis
The clinical data was analyzed with the IBM SPSS statistics v20.0.0 software and expressed as mean ± s.d. The continuous and categorical variables were analyzed by either one-way analysis of variance test, chi-square test, independent-sample t-test or nonparametric tests. p < 0.05 was considered statistically significant.

Demographic and clinical characteristics
The demographic and clinical characteristics of PD-A, PD-NA and healthy control subjects are summarized in Table 1. The three subject groups showed differences in age, gender, HAMD scores, but education levels were similar. The PD-A and PD-NA subjects had similar disease duration, H&Y staging, total LEDD, LEDD of levodopa preparations and LEDD of dopamine receptor agonists and UPDRSIII scores (p > 0.01). The MMSE and FAB scores for PD-A and PD-NA subjects was also similar. As expected, the HAMA scores were different between the three groups (p < 0.01). Next, two-sample post hoc t-tests were performed to detect pair-wise differences in FC of the putamen regions among the PD-A, PD-NA and healthy control groups. The PD-A group showed following differences with PD-NA group: (1) Figure 1).

Functional connectivity data
PD-A subjects showed following differences with the healthy control subjects: (1) reduced left putamen FC with left anterior cingulate; (2) reduced right putamen FC with left orbitofrontal gyrus and (3) increased right putamen FC with paracentral lobule ( Table 2 and Figure  1).
The PD-NA subjects showed the following differences with the healthy control subjects: (1) Figure 1). Figure 2 shows Pearson correlation analysis of HAMA scores and FC scores corresponding to differences between PD-A and PD-NA groups. The analysis revealed the following: (1) positive correlation between right putamen FC with left caudatum (r = 0.330, p = 0.008); (2) negative correlation between left putamen FC with right orbitofrontal gyrus (r = -0.367, p = 0.003) and (3) negative correlation between right putamen FC with left orbitofrontal gyrus (r = -0.332, p = 0.008) and right cerebellum (r = -0.326, p = 0.009).

DISCUSSION
In the present study, we compared the FC alterations in the putamen region of the brain in Parkinson's patients with anxiety disorder to Parkinson's patients without anxiety and healthy controls. Rs-fMRI analysis revealed significant alterations in FC in the PD-A patients compared to the PD-NA and HC groups. The PD-A patients showed decreased left putamen FC with right orbitofrontal gyrus and right putamen FC with left orbitofrontal gyrus, right cerebellum and right precuneus. On the other hand, there was increased right putamen FC with right insula, left temporal pole, left middle occipital gyrus, left caudatum and right middle cingulate gyrus compared to PD patients without anxiety. This suggested abnormal putamen FC in the cortical and subcortical regions in PD-A patients.
The HAMA scores showed positive correlation between right putamen FC and left caudatum, but negative correlation for left putamen FC with right orbitofrontal gyrus and right putamen FC with left orbitofrontal gyrus and right cerebellum. These results demonstrate that anxiety in PD patients correlates with the aberrant putamen FC in them, especially with the caudatum, orbitofrontal gyrus and cerebellum.
Neuroimaging studies with 123 I-FP-CIT SPECT, SPECT with TRODAT-1 or PET have explored anxiety severity in PD patients by assessing dopamine transporter (DAT) availability or density [6-9, 25, 26]. Most studies observed an inverse correlation between DAT availability in the putamen with the severity of anxiety in PD [6,7,9]. However, Ceravolo et al. showed that increased striatal DAT density was associated with mild anxiety symptoms in PD patients [8]. The PD rat model depleted of striatal dopamine showed increased anxiety [27,28], which improved upon L-DOPA treatment [29,30]. Furthermore, social anxiety disorder, which is common in PD patients was associated with dopaminergic-mediated striatal circuits [31] and sustained suppression of DA receptor activity [32]. These findings suggest a relationship between anxiety and dopaminergic dysfunction in PD patients. Therefore, we focused on interactions of the bilateral putamen with other brain areas in the present study.
Our results show increased putamen FC with caudate in PD-A relative to PD-NA patients. Although striatal dysfunction is typically related to motor impairment in PD [33], it is implicated in emotional processing [34], thereby suggesting a common pathophysiological mechanism. Neuroimaging by [ 18 F] fluorodeoxyglucose PET (FDG-PET) showed that anxiety symptoms in PD were inversely associated with DAT availability in the caudatum or decreased caudate metabolism [35]. Therefore, enhanced putamen FC and caudatum suggested emotional processing during anxiety in PD patients.
We also observed reduced bilateral putamen FC with orbitofrontal gyrus in PD-A compared to PD-NA patients. Previous studies in non-PD patients demonstrated that orbitofrontal gyrus as the site for emotion processing. Milad et al. showed the role of medial orbitofrontal gyrus in conditioned fear response based on various animal studies and human neuroimaging analysis [36]. Patients with agoraphobia, which is a panic disorder showed decreased gray matter volume in their left medial orbitofrontal gyrus compared to healthy controls [37]. Moreover, patients with impaired orbitofrontal gyrus showed mental disturbances such as impulsive, socially inappropriate behavior and emotional changes compared to control subjects [38]. This suggested the role of impaired orbitofrontal gyrus in anxiety disorders. Additionally, a T1-weighted MRI study in patients with early PD showed that focal orbitofrontal gyrus atrophy led to anxiety disorder [39]. Topographically, orbitofrontal gyrus projects into regions of the striatum [40] and the frontostriatal pathway has been implicated in anxiety [41]. Based on these considerations, attenuated connections between orbitofrontal gyrus and putamen suggest abnormal frontostriatal pathway, which contributes to anxiety disorder in PD.
There was also reduced FC between cerebellum and putamen in PD-A relative to PD-NA subjects, thereby  [42,43]. For example, electrical stimulation of the human cerebellum produces anxiety and fear [44]. The synaptic connections between parallel fibers and Purkinje cells of cerebellum are involved in the learned fear and their long-term potentiation contributes to fear memory [45]. In functional brain imaging studies with PET, increased blood flow was observed in the left cerebellar hemisphere during fear conditioning [46]. Therefore, both human and animal studies have shown that cerebellum contributes to fear-associated learning.
Recently, a meta-analysis of neuroimaging studies on functional topography in the human cerebellum showed involvement of the cerebellar posterior lobe in cognitive/ emotional function [42]. This is in agreement with our findings of decreased cerebellum FC, especially in the posterior lobe. Consequently, decreased communication between cerebellum and putamen plays a role in anxiety disorders observed in PD patients. This study has several limitations. Firstly, our sample size was relatively small and therefore may not have detected exact neural substrates of PD-related anxiety disorder. Secondly, anxiety and depression overlap in PD patients with similar symptoms and pathophysiology [20,47]. Hence, patients with clinically distinct depressive symptoms were excluded from our study. However, HAMD scores still showed significant differences between study subject groups and therefore, FC was analyzed. Besides, age and gender were significantly different among groups. Many studies have shown correlation between PD-related anxiety disorder with age and gender [2,48]. Therefore, these were analyzed as covariates.
Finally, although the present study defined important regions of interest that are responsible for anxiety in PD, it did not represent complete FC analysis. Therefore, further comprehensive studies are needed to identify other anxiety related networks that were not analyzed by our study.
In conclusion, our study showed that anxiety in PD was associated with altered FC network between putamen and other regions of the brain, especially caudatum, orbitofrontal gyrus and cerebellum. .