Introduction
Infantile esotropia (IE) is a subtype of concomitant strabismus manifesting within the first 6 months of life and is characterized by constant and large-angle esodeviation, alternate fixation, impaired stereovision and dissociated eye movements (e.g., dissociated vertical deviation and latent nystagmus), with no evidence of additional neurological disease (Brodsky, 2018). As a rare condition with 2/1,000 prevalence, IE not only interferes the normal neurodevelopment of the vision but also influences the cosmetic appearance of the children (Torp-Pedersen et al., 2017). The specific pathogenesis of IE remains poorly understood. Further understanding the neuropathological mechanism of IE is conducive for treatment option and prognosis evaluation.
Previous animal studies have revealed structural and functional alterations in visual areas of brain in various species of animals with experimentally induced strabismus (Kiorpes et al., 1998; Zhang et al., 2005; Schmidt and Löwel, 2008). In recent years, human neuroimaging studies have explored the alterations of brain structure in patients with strabismus. For instance, it was reported that patients with strabismic amblyopia exhibited tract-specific changes in mean diffusivity (MD), especially the optic radiation, vertical occipital fasciculus, frontal corpus callosum and some long-range association tracts (Duan et al., 2015). Decreased MD along the optic radiation was also observed in patients with constant exotropia (Wang et al., 2022). In addition to the structural alterations, strabismic patients also exhibited functional changes of the brain, which were demonstrated by resting-state functional MRI (rsfMRI) studies. For example, reduced functional network connectivity of the primary visual cortex with numerous brain regions related to vision processing and eye movement was found in patients with intermittent exotropia (He et al., 2021). Through an independent component analysis, altered intra- and inter- network connectivity were revealed in patients with concomitant exotropia, in terms of the visual network, sensorimotor network and auditory network (Jin et al., 2022).
However, no studies have explored the coupling metrics of brain structural and functional profiles in strabismus. As an indicator implicated in the combined analysis of structural connectivity (SC) and functional connectivity (FC) networks, SC-FC coupling is intrinsically a reflection of the structural-functional coherence, which has shown potential in quantifying the changes of structural-functional network association across age during development (Hagmann et al., 2010) and has been applied to investigate a variety of neurological disorders in recent years (Sun et al., 2017; Zhang et al., 2019; Chen et al., 2021). Evidence suggests that SC-FC coupling serves as a reliable and valuable index surpassing any single-modal network property (Zhang et al., 2011). Distinct alterations of coupling metrics were demonstrated in various conditions. For instance, alterations of SC-FC coupling strength in terms of the long connections were demonstrated in children with a family history of schizophrenia and bipolar disorder (Collin et al., 2017). In addition, abnormality in SC-FC coupling for short connections was reported in cannabis users (Kim et al., 2019). Furthermore, patients with idiopathic tinnitus exhibited altered SC-FC coupling metrics in global network connections (Chen et al., 2022). Nonetheless, it is still unclear if the patients with IE may present altered SC-FC coupling metrics for global connections, short connections or long connections. To fill in this gap and further illustrating the role of coupling metrics in the neurobiological process of IE, this study investigated the SC-FC coupling profiles for each of the three connectomes across whole brain subregions.
In the current study, our coupling analysis mainly included two steps: we first compared the SC-FC coupling in IE patients with that of healthy subjects, of which the group-age interaction was also examined to identify distinct developmental trends of coupling metrics in IE. Afterward, we assessed the correlation of the coupling measures with clinical variable, on which the age-related regulatory effect was also investigated in different brain subregions. Abnormalities of structural neuroimaging of strabismic patients were overlapped for regions implicated in visual pathway (Duan et al., 2015; Wang et al., 2022). Thus, we hypothesized that patients with IE would show unique pattern of SC-FC coupling in brain, especially in those local regions involved in visual information processing, and exhibit distinct developmental trajectory of coupling profile across age.
Materials and methods
Participants
A total of 18 patients with IE (10 males and 8 females, mean age: 3.27 ± 2.00, age range: 0.41–6.36) were enrolled in this study from Beijing Children’s Hospital. Meanwhile, 20 age- and gender- matched healthy subjects (11 males and 9 females, mean age: 3.41 ± 1.58, age range: 1.18–6.28) were also recruited as controls. The inclusion criteria for IE were as follows: (1) large-angle nasalward deviation of both eyes manifesting before 6 months of age; (2) free of other eye diseases or previous ophthalmic surgery; (3) without neuropsychiatric or cardiovascular disease; (4) full-term birth. Subjects who met one of the following conditions were excluded: (1) brain abnormality detected by conventional MRI; (2) contraindications for MR examination. The research conformed to the Declaration of Helsinki and was approved by the Medical Research Ethic Committee of Beijing Children’s Hospital. Written informed consents for each participate were obtained from their parents before any procedures of the study.
MRI data acquisition
Imaging data were collected using a 3.0 Tesla MRI system (Discovery MR 750, General Electric, Milwaukee, WI, USA) in Medical Imaging Center of Beijing Children’s Hospital. All participates underwent scanning during natural sleep. Earplugs and spongy cushions were used to cut down on the noise and minimize head motion during scanning. A camera facing to the scanner bed was applied to monitor the sleeping behaviors of the subjects. T1-weighted, sagittal three-dimension brain volume sequence was acquired with following parameters: repetition time (TR) = 8.2 ms, echo time (TE) = 3.2 ms, flip angle = 12°, field of view (FOV) = 256 mm × 256 mm, acquisition matrix = 256 × 256, slice thickness = 1 mm, voxel size = 1 mm × 1 mm × 1 mm, slice = 164. The rsfMRI data were collected with echo-planar imaging (EPI) sequence, and the scanning parameters were as follows: TR = 2,000 ms, TE = 30 ms, FOV = 224 mm × 224 mm, acquisition matrix = 64 × 64, slice thickness = 3.5 mm, voxel size = 3.5 mm × 3.5 mm × 3.5 mm, flip angle = 90°, slice = 40, data volume = 240. Axial diffusion tensor imaging (DTI) was acquired with a single-shot EPI sequence: TR = 7,200 ms, TE = 63 ms, slice thickness = 2 mm, slice gap = 0 mm, slice = 72, acquisition matrix = 128 × 128, FOV = 256 mm × 256 mm, voxel size = 2 mm × 2 mm × 2 mm, 32 diffusion gradient direction images with the b-value of 1,000 s/mm2 and a single image with the b-value of 0 s/mm2.
DTI data preprocessing
Diffusion tensor imaging brain imaging data were preprocessed by FMRIB Software Library (FSL 6.0)1 (Jenkinson et al., 2012) and FreeSurfer package2 (Fischl, 2012). The main steps were as follows: (1) skull-stripping to remove the non-brain tissue with FreeSurfer package, which is better and more stable than FSL-bet; (2) eddy current correction and motion correction with FSL-eddycorrect. Framewise displacement (FD) was used to assess the head motion, and datasets with mean FD bigger than 0.3 mm were excluded in the following analysis (Power et al., 2014); (3) tensor estimation was performed with FSL-dtifit, including fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD).
rsfMRI data preprocessing
Analysis of Functional NeuroImages (AFNI)3 was used in rsfMRI data preprocessing. The main steps of preprocessing pipeline included: (1) removing the first 5 time points; (2) slice timing correction to remove the time difference between slices; (3) head motion correction with linear registration. Scrubbing method was applied removing the volumes with FD >0.3 mm and less than 3 continuous timespoints; (4) nuisance covariates regression of white matter signal, cerebral fluid signal, global signal and Friston-24 head motion parameters (Friston et al., 1996); (5) band-pass filtering with the frequency range of 0.01–0.1 Hz; (6) spatial smoothing with a 6 mm full width at a half maximization Gaussian kernel was done after the native preprocessed images being warped into standard space.
Multimodal images normalization
Similar spatial registration steps were done for both native diffusion images and native function images using the Advanced Normalization Tools (ANTs).4 First, the b0 volume or the first function volume were skull-stripped and non-linearly warped to the individual anatomical volume with high resolution. The non-linear co-registration here was thought to better control the EPI distortion in both diffusion and function images. Second, the native anatomical volume was non-linearly warped into to the University of North Carolina (UNC) brain standard space for 2-year-olds (Shi et al., 2011). All the spatial transform files were combined into one step warping in order to avoid multiple interpolations. Although the subjects in current study were aged across 6 years, within which the brain possess a fast and complex growth, we chose a single 2-year-old template as the appropriate target space here considering two aspects: (1) newborn brain grows logarithmically with the major growth before and relatively gentle growth after about 2-year-old; (2) the 2-year-old brain is almost adultlike and only two newborns were less than 1-year-old.
Network construction
Totally 90 regions covering the brain from the Automated Anatomical Labeling (AAL) atlas (Shi et al., 2011) were labeled and used to build both the structural and functional connectivity networks. Structural connectivity was generated with DSI studio5 (Yeh et al., 2013a,b) with the deterministic fiber assignment by continuous tracking (FACT) algorithm. The fiber tracking terminated if one of these conditions were met: track length shorter than 10 or longer than 200 mm; FA threshold lower than 0.1; turning angle higher than 45°. The normalized number of fiber streamlines between paired regions by the total number of streamlines, as well as the mean DTI metrics of voxels that the tract passing, were computed, yielding five networks with different structural measures. Functional connectivity was quantified with the Fisher-Z transformed Pearson’s correlation of the preprocessed bold signals from paired regions. In order to obtain the spatial location relationship between regions, Euclidean distance between each paired regions were calculated in 2-year-old standard space.
Brain structural-functional coupling
Regional connectivity profiles were extracted from each column of a subject’s functional connectivity matrix and five structural connectivity matrixes. Multiple linear regression analysis was performed on these connectivity profiles, predicting the regional functional connectivity profile from the structural connectivity profiles. All variables were normalized to be zero mean and one standard deviation before regression modeling. Structure-function coupling score was then measured as the adjusted R-square score of the model fitting for each region (Wang et al., 2018; Baum et al., 2020). Furthermore, we grouped the whole brain connections into short connections and long connections by the Euclidean distance threshold of about 66 mm, which was the mean length between regions in 2-year-old standard space (Shi et al., 2011). Similarly, two more structure-function coupling scores for short connections and for long connections were generated for each region.
Discussion
This is the first study exploring the potential neural basis of pathogenesis of IE from the point of brain structural and functional measures interaction, through a combined DTI and rsfMRI analysis. Our main findings were as follows: (1) patients with IE exhibited aberrant SC-FC coupling in CAU compared to healthy subjects; (2) A decoupled tendency with age in structure-function relationship of CAU was initiated in newborn period and persisted throughout early childhood in IE patients; (3) For SC-FC coupling profile, it was short connectome, but not the long connectome, that was prone to be disturbed under the influence of IE; (4) The coupling measures of the sensorimotor region (i.e., PCL) showed significantly positive correlation with the strabismus degree; (5) age-related regulatory effect on the association between coupling score and strabismus degree was discriminative in brain areas related to visual or sensorimotor processing and those responsible for cognitive tasks.
It is considered that structural and functional network interact with one another. More specifically, the SC acts as the anatomical infrastructure and imposes constraints for the FC, while FC exerts effect on shaping of the SC through a plasticity mechanism (Honey et al., 2009; Guerra-Carrillo et al., 2014). The attenuated SC-FC coupling of the CAU arises from the weakened association between SC and FC. In fact, CAU plays a central role in visual processing. On the one hand, it has been reported to be involved in the static and dynamic visual information processing (Gombkötõ et al., 2013), equilibrating visual and reward information (Doi et al., 2020) and visual category learning (Nomura and Reber, 2008). Moreover, CAU is capable of receiving the visual content information and visuospatial information, and thus is an indispensable component implicated in both ventral and dorsal visual pathway (Yamamoto et al., 2012). Although a weaker structure-function coupling may contribute to the more flexible function, it may make some areas pertaining to the visual pathway (e.g., the CAU) vulnerable in the presence of heterogeneous binocular inputs, which accounts for the observed abnormality in stereopsis in IE patients (Zarkali et al., 2021). Impairment in voxel-mirrored homotopic connectivity or amplitude of low-frequency fluctuation in CAU was also proved in patients with strabismus (Yin et al., 2021; Zhang et al., 2021). Of note, it was left ACU, but not the bilateral CAU, that showed anomalous coupling metrics in IE patients. The cerebral-lateralization theory may be reasonable for this observation, but the underlying mechanism needs to be further explored in future.
On the other hand, direct pathway projections are existent between the CAU and substantia nigra pars reticulata, through an open loop pathway, the latter projects to the visual reflex center, that is, the superior colliculus, which is connected with lateral geniculate nucleus via the brachium of superior colliculus and associates reticular structure of brainstem through afferent fibers, with responsibility for the eye movement regulation and eye position control (Seger, 2013). As an important part of the visual corticostriatal loop, CAU is also closely linked with visual temporal cortex that encodes visual motion (Choi et al., 2012; Rolls et al., 2022). Overall, convergent evidences suggest that CAU not only involves in the visual corticostriatal loop (Lawrence et al., 1998), but also is implicated in subcortical visual motion circuit (Brodsky, 2019). Thus, we speculated that alterations in SC-FC coupling of CAU could interfere the visual corticostriatal loop responsible for visual processing and eye movement planning as well as the visual subcortical circuit subserving nasalward optokinesis, resulting in visuospatial impairment and nasalward deviation of eyes in IE patients, which lends neuroimaging support to the recent hypothesis that changed plasticity of subcortical visual pathway during the newborn period in IE patients induce monocular asymmetric optokinetic response to the horizontal stimulus and thus drive the nasalward rotation of both eyes (Brodsky, 2019).
An open question is that when the SC-FC coupling deficits appear in children with IE during the process of neurodevelopment. In this study, we found that the SC-FC coupling changes with age during typical development, with distinct trajectories in different brain regions, as similar to previous studies (Collin and van den Heuvel, 2013). Interestingly, the developmental trends of SC-FC coupling across age were also demonstrated in patients with IE in several brain regions, among which the CAU exhibited entirely different trajectory in coupling metrics compared to that of the healthy counterparts. More importantly, our findings also suggested that the disturbance of SC-FC coupling evolution was initiated in their early postnatal period, which enhanced the understanding for the developmental pattern of network property in patients with IE and provided evidence to explain the universal phenomenon that most IE patients present inward deviation of eyes within the first 6 months of life.
For IE patients, enhanced SC-FC coupling in PCL significantly correlated with increased strabismus degree. The heightened SC-FC coupling in PCL in patients with IE may indicate that functional connections through the PCL are more prone to rely on the intrinsic structural connections, resulting in less dynamic brain function (Dai et al., 2019). The PCL, known as an important part of the sensorimotor network, plays a central role in the sensorimotor integration, which contributes to the visuomotor coordination (Di Ciò et al., 2020). Increased functional dependence over structural connectivity in PCL as the strabismus degree increasing indicates the more serious the esotropia is, the less flexible is the brain function dynamics in PCL, which may contribute to the observed abnormalities in hand-eye coordination in IE children. Anomalous cortical thickness or voxel-wise degree centrality in PCL was also reported in patients with strabismus (Wu et al., 2020; Yin et al., 2021). Interestingly, for visual-related or sensorimotor cortical areas, the correlation between the coupling score and strabismus degree had a downtrend with the increase of age, indicating that the younger the IE patients are, the greater influence the strabismus degree exerts on SC-FC coupling of the visual or sensorimotor regions. Given the fact of early onset of IE, corrective surgery is recommended to be implemented as early as possible to relieve the disturbance of coupling arising from esotropia. On the other hand, for cortical regions pertaining to cognition (i.e., such pivotal hubs of default mode network as superior frontal gyrus, posterior cingulate gyrus, and middle temporal gyrus), the association between the coupling metrics and strabismus degree displayed a growing tendency with age, which means the older the patients are, the more serious is the cognitive impairment related to esotropia, suggesting that misalignment of eyes may bring continuous undesirable effect on the cognition development (Zhen et al., 2018).
Previous studies have demonstrated that SC-FC coupling profile is influenced by the distances between brain areas which interconnect with each other (Collin et al., 2017; Zhang et al., 2019). Likewise, our observations showed that the SC-FC coupling of patients with IE was mainly affected in short connections, but rather in long connections, across the whole brain subregions. One reasonable explanation is that, the local connectome responsible for visual processing is more vulnerable to unequal visual input from both eyes. As a congenital fusion-deficient strabismus, IE exhibits evident deficits in stereovision and eye movement, which may be attributed to the disturbed SC-FC coupling in local regions with short connections linked to visual pathway.
Several limitations should be noted when interpreting the results. First, a relatively small number of subjects was enrolled in the present study. Even so, this study has been the largest one in sample size among the advanced neuroimaging (i.e., fMRI, DTI, and so forth) studies for patients with IE up to now and, for the first time, offers insight into the interaction between structural and functional networks in IE patients. Second, the age-related regulatory effect analysis on the coupling-degree association was exploratory in nature. In this context, correction for multiple comparation was not employed in these analyses, which may inevitably bring about some false positive entries, thus some cautions are essential when illustrating these observations. Additional topological properties of the network would be explored in future works through a graph theoretical method, since some of which play an important role in the global information communication and has been reported to have associations with SC-FC coupling metrics in both healthy and disease states (Collin et al., 2014; Kim et al., 2019; Cao et al., 2020).