Alternations of interhemispheric functional connectivity in children with strabismus and amblyopia: a resting-state fMRI study

Previous neuroimaging studies demonstrated that patients with strabismus or amblyopia can show significant functional and anatomical changes in the brain, but alterations of interhemispheric functional connectivity (FC) have not been well studied in this population. The current study analyzed whole-brain changes of interhemispheric FC in children with strabismus and amblyopia (CSA) using voxel-mirrored homotopic connectivity (VMHC).A total of 24 CSA (16 males and 8 females) and 24 normal controls (NCs) consisting of 16 and 8 age-, sex, and education-matched males and females, respectively, underwent functional magnetic resonance imaging (fMRI) scans in the resting state. According to Gaussian random field theory, changes in the resting state FC (rsFC) between hemispheres were evaluated using the VMHC method. The relationships between mean VMHC values in multiple brain regions and behavioral performance were evaluated by Pearson correlation analysis. In contrast to NCs, the CSA group showed significantly decreased VMHC values in the bilateral cerebellum, bilateral frontal superior orbital (frontal sup orb), bilateral temporal inferior(temporal inf),and bilateral frontal superior(frontal sup). CSA have abnormal interhemispheric FC in many brain regions, which may reflect dysfunction of eye movements and visual fusion. These findings might provide insight into the underlying pathogenetic mechanisms of CSA.

Strabismus and amblyopia are common visual developmental disorders characterized by impaired vision and can occur in infancy 1,2 . Strabismus can be divided into comitant and noncomitant forms 3 . It is characterized by abnormalities in eye position and movement, can cause binocular vision impairment, and is often associated with amblyopia and stereoscopic vision loss. Strabismus is an optical manifestation of extraocular muscle discoordination 4 . The prevalence of strabismus in preschool children in eastern China is ~ 5.65%, and ~ 12.8% of the condition is associated with amblyopia 5,6 .
Amblyopia is caused by abnormal visual experiences (e.g., monocular strabismus, anisometropia, and ametropia) during vision development 7 . The prevalence of amblyopia among children in China is 2-3% 8 . It is therefore significant and necessary to explore interhemispheric functional connectivity (FC) in children with strabismus and amblyopia (CSA), which refers to children who have both strabismus and amblyopia.
Magnetic resonance imaging (MRI) techniques have evolved rapidly to provide a non-invasive neuroimaging method that can assess functional and structural changes in the brain 9 . Previous studies demonstrated that interhemispheric synchrony is closely related to visual experience 10,11 . Resting state functional MRI (rsfMRI) is a special technique first proposed by Biswal 12 that can assess consistent patterns of spontaneous fluctuation of blood oxygen level dependent signals during rest. These signals can be used to measure interhemispheric coordination 13 . rsfMRI and combined studies of functional and anatomic imaging have been applied to various ocular diseases including primary angle-closure glaucoma, congenital comitant strabismus, monocular blindness, and acute eye pain 14-17 . Quantification of interhemispheric FC between time series at a certain voxel and its mirrored counterpart in the opposite hemisphere can be accomplished with voxel-mirrored homotopic connectivity (VMHC) 18

Materials and methods
Participants. This study included 24 CSA (16 males and 8 females) who were treated in the Department of Ophthalmology of The First Affiliated Hospital of Nanchang University (Nanchang, China). All of the subjects with CSA (14 with exotropia and 10 with esotropia) met the following criteria: (1) strabismus, (2) greater than one line difference in the best-corrected visual acuity (VA ≥ 0.20 logMAR units) between the amblyopic and fellow eyes, (3) and able to perform center fixation. The exclusion criteria were: (1) children with previous ocular surgery history including intra-and extraocular surgery, (2) other disease (cardiovascular disease, psychiatric disorders, and cerebral infarction), or (3) eye disease (e.g., infection, inflammation, and ischemic disease). In addition, 24 NCs (16 males and 8 females) were matched with the CSA group in accordance with age and sex. The inclusion criteria for NCs were as follows: (1) able to undergo MRI (no pacemaker or implanted metal device), (2) no cardiovascular conditions such as heart disease and high blood pressure, (3) no mental or psychiatric disorders (depression and/or anxiety disorders), and(4) no eye disease history with VA ≤ 0 logMAR units. The study was approved by the ethics committee of the First Affiliated Hospital of Nanchang University, and all methods were applied in accordance with the Helsinki Declaration. The entire study design was provided to the parents of each child involved in the study before they signed informed consent forms.

MRI parameters.
All subjects underwent MRI in a 3-Tesla MR scanner (Trio, Siemens, Munich, Germany). They were required to lie flat on the scanning bed with the head in a neutral position, which was fixed with a foam sponge during the scanning process to prevent head movement. During examination, the subjects remained awake, closed their eyes, and relaxed while avoiding intentional thinking. All scans were performed by the same imaging physician, who observed the subjects until the process was completed successfully. Routine brain localization and T1 and T2 sequences were performed first. The specific parameters were: repeat time = 1900 ms; echo time = 2.26 ms, thickness = 1.0 mm, gap = 0.5 mm, acquisition matrix = 256 × 256, field of view = 250 mm × 250 mm, and reversal angle = 90°. No substantial brain lesions were included in the scanning process. rsfMRI data were acquired using a spoiled gradient-recalled echo sequence. The specific parameters were: repeat time = 2000 ms, echo time = 30 ms, thickness = 4.0 mm, gap = 1.2 mm, acquisition matrix = 64 × 64, flip angle = 90°, field of view = 220 mm × 220 mm. Data were collected continuously at 240 time points, and the scanning range was the whole brain.
MRI data processing. Data were obtained with functional images through MRIcro software package (www. MRIcro. com) after prefiltering. SPM8 (http:// www. fil. ion. ucl. ac. uk/ spm) and DPARSFA (http:// rfmri. org/ DPARSF) software were used to preliminarily analyze data before it was converted to the NIFTI format. Due to possible instability of the initial MRI signal, we removed the first 10 time points of the functional images to ensure the subjects had adapted to the scanning environment. The remaining images were corrected for time differences and small movements, and the signals collected at different times were modified to the same time point. A single T1-weighted magnetization-prepared rapid gradient-recalled echo structure image was transformed into average fMRI data, and then the obtained T1-weighted image was segmented with the DARTEL tool to improve the spatial accuracy of standardized fMRI data. Subjects with head movement > 1.5 mm in the x, y or, z direction or angular rotation > 1.5° were excluded. After correcting for head movement, low-frequency filtering (0.01-0.08 Hz) was modified to eliminate the influence of the physiological high-frequency noises (e.g., breathing and heartbeat). We utilized standard echo planar imaging templates to standardize the fMRI images to the Montreal Neurological Institute space and resampled all voxels to 3 mm × 3 mm × 3 mm resolution.
Statistical analysis for VMHC. To normalize the data, we transferred the VMHC maps in subjects to z-values with Fisher z-transformations in the REST software (http:// resti ng-fmri. sourc eforge. net). The child's brain was used as a mask. Two-sample t-tests were applied to evaluate z-maps in individuals to identify differences in VMHC values between the two groups using global VMHC as a covariate in a voxel-wise manner. Based on Gaussian random field theory (z > 2.3, cluster > 40 voxels, P < 0.01, family wise error [FWE] corrected), we set the voxel level (P < 0.01) as the statistical threshold to make comparisons.
Brain-behavior correlation analysis. Brain areas with different VMHC values in the CSA group were classified as different regions of interest using REST software. The relationships between mean VMHC values in different brain regions in the CSA group and clinical features were assessed with correlation analysis, with P < 0.01 considered statistically significant.
Statistical analysis. SPSS version 16.0 (SPSS Inc, Chicago, IL, USA) was used to perform statistical analyses on significant data. With the SPM8 toolkit, differences in VMHC z-maps between the CSA and NC groups were analyzed with two-sample t-tests (z > 2.3, P < 0.01, cluster > 40 voxels, FWE corrected). Pearson correlation analysis was performed to clarify the relationship between mean VMHC values of different brain regions and behavioral performance. P < 0.05 was considered statistically significant.

Results
Demographics and visual measurements. There were no remarkable differences in age (P = 0.732), weight (P = 0.814), or best-corrected VA of another amblyopic eye (P = 0.005) between groups. Details are shown in Table 1.
VMHC differences. VMHC was significantly decreased in the CSA group compared to NCs in four regions: bilateral cerebellum, bilateral frontal sup orb, bilateral temporal inf, and bilateral frontal sup (Fig. 1, Table 2). The mean VMHC values are represented in (Fig. 1c). In the CSA group, there was no association between clinical features and average VMHC values in different brain regions (all P > 0.05).

Correlation analysis.
In the CSA group, the mean hospital anxiety and depression scale (HADS) scores negatively correlated with VMHC values in the frontal sup (r = − 0.834, P < 0.001), and the HADS scores also negatively correlated with temporal inf VMHC (r = − 0.797, P < 0.001) (Fig. 2).

Discussion
VMHC is a novel measurement that can reflect changes in interhemispheric FC. The VHMC method has been applied in several ophthalmological diseases (Table 3). To our best knowledge, this is the first time it has been used to study interhemispheric FC in CSA. Compared with NCs, VMHC values were significantly decreased in the bilateral cerebellum, bilateral frontal sup orb, bilateral temporal inf, and bilateral frontal sup in the CSA group. The cerebellum is located in the lower posterior part of the brain and is involved in maintaining coordinated motor function 26 . Previous studies reported that Purkinje cells, the main output neurons of the cerebellum, can predict eye movements and that the cerebellum helps regulate precise eye movements 27,28 . The cerebellar vermis in particular plays a key role in eye movement 29 . Another group found that the posterior interposed nucleus (PIN) in the cerebellum is critical for coordinated eye movement in strabismic monkeys 30 . Similarly, studies in humans demonstrated that individuals with strabismus exhibit impaired motor behaviors 31,32 . Other groups found that comitant strabismus patients showed low mean diffusivity values in the bilateral cerebellum 33 , and voxel-wise degree centrality (DC) values were decreased in the right cerebellum of comitant exotropia strabismus patients 34 . In this study, the CSA group exhibited reduced VMHC in the cerebellum, suggesting impaired interhemispheric FC function in this brain region. Abnormal interhemispheric FC in the cerebellum might be used as a clinical maker to assess motor control in CSA.
The temporal inf is located in the anterior part of the temporal lobe and manages natural scene coding 35 . This brain region has visual selectivity and responds to three-dimensional structures defined by binocular disparity 36,37 . Temporal inf dysfunction is associated with various diseases including optic neuritis 38 , blindness 39 , and Alzheimer's disease 40 . Yuan et al. found that patients with mild cognitive impairment had significant regional homogeneity changes in the temporal inf compared to NCs and noted that this brain region plays an important role in multisensory memory and sensory integration 41 . Another group reported altered FC between the visual cortex and right temporal inf in patients with primary open angle glaucoma 42 . Many brain regions are associated with the default model network (DMN), which is activated at rest and disabled during tasks, including the middle frontal gyrus, superior frontal gyrus, inferior parietal cortex, and precuneus 43,44 . It is increasingly evident that the DMN vulnerability plays an important part in depression and anxiety 45 . In line with these findings, the present study showed that CSA had reduced VHMC in the bilateral temporal inf. Moreover, HADS scores were negatively correlated with VMHC values of the temporal inf (Fig. 2b). It may therefore be helpful to assess brain www.nature.com/scientificreports/ function changes in CSA using the VMHC method. Lower values might reflect the reduction of the role of the temporal inf in the DMN in CSA, who are thought to have dysfunctions in visual selectivity and visual memory. The frontal orbital cortex is below Brodmann area (BA) 47, which is part of the frontal cortex associated with processing language and grammar 46,47 . In addition, BA 47 is thought to control the perception of musical structure 48 . The patients with the right eye monocular blindness showed lower VMHC values in the superior parietal lobule (BA7) 21 . In the present study, CSA showed remarkably decreased VHMC in the bilateral frontal sup orb, which may be a reflection of impaired language understanding in CSA.
The frontal lobe is located in the anterior central sulcus, and injuries in this region can cause impairment of voluntary movement, language expression, memory. There is a positive correlation between the DMN and cognitive control network in the frontal gyrus 49 . Huang et al. detected alterations in the whole brain microstructure of patients with esotropia using diffusion tensor imaging and found that average diffusion coefficient was markedly reduced in the left middle frontal gyrus 33 . Another group reported decreased voxel-wise DC values in the right middle frontal gyrus were observed in patients with comitant exotropia strabismus 34 . In a former study, patients with corneal ulcer had decreased VMHC values in the medial frontal gyrus 23 . In addition, Yuan et al. found that retinal detachment patients had significantly lower VMHC values in the bilateral occipital lobe 20 . In patients with acute open globe injury, they exhibited reduced VMHC values in the lingual gyrus 19 . Furthermore, the frontal lobe of patients with comitant exotropia showed decreased white matter volumes 50 , and the middle frontal gyrus of amblyopic patients exhibited reduced gray matter density 51 . In the present study, VHMC was www.nature.com/scientificreports/ decreased in the frontal sup of the CSA group compared to NCs, indicating that this region contributes to SA and is involved in visual processing and associated eye movements (Table 4). Consequently, we hypothesize that VMHC values in these brain regions might be potential diagnostic markers for CSA. Our study should be interpreted in the context of some limitations. First of all, the sample size was relatively small. In addition, the correlation between clinical characteristics of SA and mean VMHC values require further investigation. Thus, we are looking forward to designing more experiments to further elucidate the underlying molecular mechanisms.    Depression and anxiety, involved in visual processing and associated eye movements