1　Introduction

       Bell’s palsy (BP), a unilateral idiopathic facial nerve palsy^[1], is a common disorder with a fair prognosis causing an acute weakness or paralysis of the facial nerve that affects facial muscles and interferes with their normal facial functions.

       Brain connectivity^[2] indicates a pattern of anatomical connectivity, functional connectivity or effective connectivity between distinct brain regions. Resting state fMRI is the functional brain imaging method used to evaluate regional interactions during non-performing an explicit task^[3].

       Default mode network (DMN) is a brain network that consists of brain regions that are active when the individual is not focused on the external environment and the brain is at wakeful rest. It is known to be related to self-referential thought^[4].

       There are few studies about BP with fMRI^{[1, 5,6,7,8]} but non of them studied the effect of BP on DMN connectivity. In this study we tends to understand the effect of BP on DMN connectivity and how these connectivity changes might affect the recovory process of BP patients.

2　Materials and Methods

2.1　Subjects

       35 right handed healthy participants (16 males, age=30.08 ± 9.94 (mean ± std) years old, age range=18—54 years old) and 52 right handed BP patients (28 males, age= 37.44 ± 14.59 years old, age range= 19—70 years old) were recruited. Healthy participants were scanned only once, while some of patients took part more than once. All healthy participants and patients signed informed consents in accordance with the Human Research Committee of the Affiliated Hospital of Anhui TCM University before taking part in the experiment. Gender, age and duration from the onset of BP were recorded into an experiment information acquisition form before the scan. For recovered patients, they got scanned when they were recognized as recovered. House-Brackmann score^[9,10] (HBS) (1: healthy facial movement - 6: no facial movement) is the most widely used facial nerve grading system which used for clinical assessment of facial function and the grading results were also recorded.

       To find the effect of BP on DMN connectivity between healthy volunteers and BP patients at different pathological stages, subjects’ scans were classified into four groups based on their HBS and duration (D) including healthy volunteers (healthy groups), patients with BP in early stages (early group) (D<14 days, HBS> I), late stages (late group) (D> 14 days, HBS> I) and recovered patients (recovery group) (D>14 days; HBS=I) (see figure 1).

点击查看大图

Fig. 1  Experimental groups. We had 4 groups of data classified based on disease duration (D) and House-Brackmann score (HBS) system. Healthy group (control), early group (with duration<14 days and HBS>1), late group (with duration>14 days and HBS>1), recovered group (with duration<14 days and HBS=1).

2.2　fMRI scanning

       fMRI experiments were performed at the Medical Imaging Center, the First Affiliated Hospital of Anhui University of Traditional Chinese Medicine with 1.5 T whole body MRI scanner (Siemens Symphony, Siemens Medical, Erlanger, Germany) and standard head coil (8-channels). A total of three sessions were performed as (1) T2-weighted (exclude subjects with apparent central nervous disease and it lasts for 1.5 minutes), (2) resting state bold scan (10 minutes and 4 seconds, EPI-BOLD sequences, TR/TE/FA of 3000 ms/30 ms/90°, FOV of 192 mm×192 mm, and resolution of 64×64, 200 volumes), and (3) T1-weighted 3D anatomical images (sagittal position and a total of 176 slices scanned, spoiled gradient echo sequence, TR/TE/FA 2100 mm/3.93 mm/13°, FOV of 250 mm×250 mm; slice thickness/gap of 1.0mm/0.5mm, resolution of 256×256, whole brain, 9 minutes).

2.3　fMRI data analysis

       Pre-processing and statistical analysis at both individual and group levels were performed using the FSL, AFNI, SUMA and FreeSurfer software packages.

       For each subject, head motion correction was performed based on six-parameter rigid-body transformation with 8th volume as reference. Brain extraction, cortical surface reconstruction, functional data registration to standard Montreal Neurological Institute (MNI) space using FMRIB’s nonlinear co-registration tool (FNIRT) and resampled to a voxel size of 2 mm × 2 mm × 2 mm were performed. Functional data were spatially smoothed using a Gaussian kernel of full width at half maximum 6 mm, and high-pass temporal filtering (f = 0.007 Hz) were performed.

2.4　Dual regression analysis

       Dual-regression independent component analysis (ICA) approach was used to investigate resting state functional connectivity changes in DMN using FSL-MELODIC tool with 25 components. DMN network have been defined and extracted based on previously defined templates provide by Beckmann et al^[11]. All individual extracted brain maps of the right BP patients were flipped across mid-sagittal plan^{[1, 7,8]} so that right hemisphere became contralateral side to paretic side and left hemisphere became ipsilateral side to paretic side.

2.5　Inter-group analysis

       Group mean maps for each group were calculated using a mixed effects statistical model. In order to find DMN changes over different stages of BP, difference maps were calculated between combination of two groups from healthy, early, late and recovered groups using unpaired t-test with mixed effect model.

       Corrrelation analysis between DMN connectivity and HBS and duration were performed using ANCOVA with mixed effect model. All statistical maps were thresholded with cluster forming threshold at Z >1.96 (voxel wise threshold P <0.05, cluster size ≥ 26 voxels).

3　Results

3.1　Clinical outcomes

       All BP patients revealed a unilateral loss of their facial movement functions. In this study, BP patients’ scans were divided to be in one of three different groups early, late and recovered groups based on patient’s disease duration and HBS at the scan time. Early group contained 39 subjects (18 right paralysis, HBS=2—4, 3.1 ± 0.63(range, mean ± SD), duration=2—11, 6.36 ± 2.38 (range, mean ± SD)) and late group had 36 subjects (18 right paralysis, HBS=2—4, 2.51 ± 0.68, duration=16—55, 36.77 ± 26.02). For recovered group (38 subjects, 22 right paralysis, HBS=1, duration=16—118, 46.91 ± 26.55), patients were able to perform facial movement efficiently (HBS=1) without synkinesis or contracture (see Table 1, Figure 1).

点击查看表格

Tab. 1  Demographic information and data groups (D: Duration. HBS: House-Brackmann score)

3.2　DMN Inter-group differences

       Significant increase in early group over healthy group was observed in right (r.) precentral gyrus (MI), r. postcentral gyrus (SI), r. middle frontal gyrus (MFG) and r. superior frontal gyrus (SFG) (see Figure 2A, Table 2).

       Significant increase was observed in late group over healthy group in r. superior parietal lobule (SPL), r. cerebellum, r. lingual gyrus, r. ventromedial frontal gyrus (VMPFC), r. middle cingulate cortex (MCC), r. posterior cingulate gyrus (PCC), left (l.) precuneus, and l. MCC (see Figure 2B, Table 2).

       And interestingly, significant increase in recovered group over healthy group in l. lingual gyrus, and l. culmen (see Figure 2C, Table 2).

点击查看大图

Fig. 2  Difference maps of Default Mode Network (Z>1.96, P<0.05). Right side is the contralateral, and the left side is ipsilateral to the paretic side of patients’ faces. A: Healthy (H) vs early (E) groups within DMN network. B: Healthy (H) vs recovered (R) within DMN network. C: Healthy (H) vs recovered (R) within DMN network. SI: primary sensory cortex. MI: primary motor cortex. MCC: middle cingulate cortex. PCC: posterior cingulate cortex. SFG: superior frontal gyrus. MFG: middel frontal gyrus. VMPFC: ventromedial prefrontal cortex. Parahipp: parahippocampal gyrus.
Fig. 3  Correlation analysis between duration and Default Mode Network of recovered groups (Z>1.96, P<0.05). SI, primary sensory cortex. SMG, Supramarginal gyrus. PCC: Posterior cingulate gyrus. DLPFC: dorsolateral prefrontal cortex. STG: superior temporal gyrus. aIns: anterior insula.

点击查看表格

Tab. 2  Brain regions with significant changes between healthy group and different Bell’s palsy groups

3.3　Correlation with duration and HBS

       No correlation between DMN brain network and HBS for patients over early and late groups were observed. Also no correlation between DMN and disease duration in early group was observed.

       Negative correlation was observed between late group and disease duration until the late scan in ipsilateral side at l. STG, l. middle occipital cortex and inferior temporal gyrus (ITG) (see Table 3).

       DMN connectivity in recovered group was Positively correlated with disease duration in ipsilateral side at l. SMG, l. angular gyrus, l. PCC, l. STG, l. precuneus, and l. SPL, while negative correlation was observed at contralateral side in r. anterior insula (aIns), and r. temporal pole (see Figure 3, Table 3).

点击查看表格

Tab. 3  Brain regions with significant correlation between different Bell’s palsy groups and disease duration

4　Discussion

       This study was aimed to investigate the effect of BP on DMN functional connectivity over different pathological stages of BP.

4.1　Overall tendency of the connectivity changes

       By observing changes of DMN functional connectivity between healthy and BP patients, quite different connectivity changes were found at different pathological stages. This connectivity changes observed in patients’ groups supported our hypothesis that BP causes DMN connectivity changes. This cortical reorganization occurred as compensatory reaction to overcome loss of facial motor control. It is inferred that brain’s connectivity status altered in BP leading to changes in patients’ groups compared to healthy individuals.

       For early group changes occurred only in contralateral side only which is responsible for paratic side, might result as a compensatory way to perform facial movements and enhance these movements to be in precise and right way.

       For late group, patients with late recovery might have many connectivity changes due to neuroplasticity as brain tries to adjust and change connectivity in the way to help patients get better. Also changes in late group were observed in regions that is related to motor planning and emotion adjustments.

       For recovered group, patients were able to perform their facial movement so they might not need to have any engagement of contralateral side connectivity changes while in ipsilateral side they had changes. These changes might occur due to neuroplasticity which leads to change of the brain states and connectivity to overcome brain function disability, nerve disease or injuries.

       Interestingly, ipsilateral regions with conne-ctivity changes in patients were significant in late and recovered groups and they might be due to neuroplasticity. According to stroke researches, the greater the involvement of the ipsilesional connectivity, the better is the recovery^[12,13]. Also it was shown as one theme for stroke recovery has been recruitment ipsilateral pathways. Strong evidence for this from evaluation of patients who have undergone hemispherectomy^[14].

4.2　DMN connectivity changes with specific areas in response to BP

4.2.1　Somatosensory and motor association cortex

       MI and SI are the primary motor and somato-sensory regions which are responsible for body parts movement and sensation. Although BP is a peripheral neuropathy which means no central nervous system lesion or injuries, connectivity of sensorimotor regions to DMN was increased in early stage of BP. Such changes in contralateral side were produced to overcome the patient’s movement impairment.

       BP is known to be deefferentation without deafferentation^[7] so the sensory function of face is not affected by BP. Also the facial expression is performed based on the synergies of facial muscles which are under control of human brain^[15] and maintained by the intactness of sensorimotor circuit to ensure the facial expression performance. Due to BP, facial muscles paralysis might cause loss of movement feedback with breaking the intactness of this synergies. SI connectivity might be increased to maintain this facial inability and find the reason of this disability.

       MI connectivity increase in contralateral side for early group was observed in facial representation part of MI. In the early stage of BP such broken synergies lead to disrupted connectivity within the affected cortical facial motor network^[5,6]. This means that facial parts of MI might have more activation as compensatory mechanism to overcome facial motor functional problems.

       MCC is known for environmental monitoring, response selection, and skeletomotor body orientation^[16]. So its increased connectivity to DMN might be resulting as mechanism for mapping new ways to perform facial movement efficiently.

4.2.2　DMN related regions

       The angular gyrus, precuneus, VMPFC and PCC are known to be associated with brain’s default mode network (DMN), which is defined as a set of regions that is spontaneously active during passive moments^[17].

       Activation in VMPFC is known to be associated with emotional precessing^[18]. Difference in VMPFC activation was shown in late group which might result to control negative emotions that patients might have due to their disability for more than two weeks. Although patients’ emotional information weren’t acquires during this study, facial palsy is known to cause negative emotions such as anxiety and depression^[19]. Also in another study patients with BP had social communication problems such as face-to-face communication and people change of attitude with them^[20]. These kinds of social problems might induce negative emotions which VMPFC activation would decrease and suppress in order to facilitate the healing process.

       Precuneus is known for its central role in highly integrated tasks such as visuo-spatial imagery, episodic memory retrieval and self-processing operations^[21]. DMN connectivity to precuneus was increased in late group of BP. This connectivity increase might be resulting from patients concern about their facial movement inability leading to reactivate the image of the intended movement to support its execution. Studies concerning motor imagery have reported that the precuneus is more activated during imagery than during the execution of movements^[22].

4.2.3　Dorsolateral prefrontal cortex (DLPFC)

       DLPFC plays important roles in executive functions such as working memory^[23], cognitive flexibility, planning, inhibition, and abstract reasoning^[24]. However, DLPFC is not fully responsible for the executive functions as complex mental activity demands additional cortical and subcortical circuits involvement. Also DLPFC is considered as the highest cortical area involved in motor planning, organization and regulation^[25]. DLPFC connectivity changes might be a complementary method as more motor mapping required in early stage to find suitable way for solving this malfunction of facial nerve and try to define the best way to perform the right motion in an efficient way.

4.3　Limitation of this study

       There are several limitations in this study such as (1) no consideration of treatment of BP patients (every patient was treated by manual acupuncture three times per week till full recovery), (2) lack of clinical outcomes (only HBS), and (3) lack of sensational and emotional information for patients.

       Acknowledgements: The authors would like to thank the Korean National Research Foundation funded by the Ministry of Science, ICT and Future Planning (NRF-2011-0028968, NRF-2009-0076345), and the Ministry of Health & Welfare and Seoul Metropolitan Government (Traditional Korean Medicine R&D Project, HI13C0700).

       Also this study was supported by National Key Basic Research and Development Program (973) (2010CB530500), the National Natural Science Foundation of China (81202768, 31171083, 31230032), Anhui Provincial Natural Science Foundation (1208085MH147), and Major Scientific Projects of Anhui Provincial Education Commission (KJ2011ZD05).