Alterations of regional homogeneity and functional connectivity in pituitary adenoma patients with visual impairment

Evidences have showed that the deprivation of vision can considerably alter the resting-state activity both within and beyond the visual cortices. However, the functional changes of the brain cortices related to partially vision-deprivation are still largely unknown. Using resting-state functional MR imaging, we quantitatively evaluated the regional homogeneity(ReHo) and functional connectivity(FC) changes between 25 pituitary adenoma patients with visual impairment and 25 healthy controls(HCs). Compared with HCs, PAs exhibited significant increased ReHo in the left superior occipital gyrus, bilateral middle occipital gyrus and reduced ReHo in the left inferior frontal gyrus and right middle temporal gyrus. PAs also showed decreased FC between vision-related area and higher-order cognitive brain areas. Furthermore, we identified that in the PAs group the FC between the left V1 and left V3 increased while the FC between left V2v and left V5 significantly decreased, the FC between left V4 area and the V3, V2d area increased. In our study, we identified that the ReHo and FC were altered between the vision-related cortices and other higher-order cognitive cortices along with disorganized functional connectivity within the visual system in PAs with visual impairment. These findings may provide important insights to understand the plasticity of visual network.

brain due to partially visual deprivation of the adults. There are only few studies that focused on the functional changes of the brain cortices related to partially vision-deprivation caused by the compression of the anterior visual pathway. In a study, the researchers investigated the alterations of one pituitary adenoma patient before and after the chiasm decompression using functional magnetic resonance image(fMRI) techniques 10 . They found that the visual cortex of the affected eye activation area decreased compared with the normal side, and returned to normal after the surgical decompression of the chiasm. Another study explored the changes of brain resting-state activity related to vision in pituitary adenoma patients, their findings showed significant functional changes in the vision-related cortex of pituitary adenoma patients, within and beyond the visual cortices 11 . These studies only focused on the visual cortex and its subareas, the functional changes of the cortical visual processing pathway were not further investigated.
In our study, we recruited 25 pituitary adenomas with visual impairment and 25 normal controls, aiming not only to learn about the alterations of the visual cortex subareas but also the vision processing pathways and other higher cortices beyond visual cortex. We hope to explore the alterations of local synchronization and functional connectivity in pituitary adenoma patients(PAs) with visual impairment by using resting-state functional MR imaging(rs-fMRI).

Results
Demographic characteristics and visual profile. The demographic characteristics information are listed in Table 1. Two groups of participants showed no significant difference in age (p = 0.24), gender (p = 0.77), among two groups of participants. Twenty-five pituitary adenoma patients with visual impairments and 25 ageand sex matched HCs with normal vision were recruited. The detailed clinical and ophthalmological examinations results were recorded and summarized in Table 2. All of the patients demonstrated that the corrected vision acuity below 1.0 (20/20) or visual field defect more than 25% of the visual field at least one eye.

Resting-state fMRI analysis
ReHo analysis. Compared with HCs (p < 0.05 at the cluster level, GFR corrected), PAs group showed an significant increase of ReHo in the left superior occipital gyrus and bilateral middle occipital gyrus and a reduction of ReHo in the left inferior frontal gyrus and right middle temporal gyrus (Fig. 1a, Table 3). Brain areas with significant differences are projected onto a 3D brain model (Fig. 1b).

FC analysis
Voxel-wise FC analysis. The temporal correlations between visual cortices (V1, V2) and other brain regions are calculated by correlation analysis. The FC differences between the two groups of both sides of V1, V2 with other brain regions were assessed by voxel-wise two-sample t-test. In comparison with the HCs, the PA group showed decreased FC between bilateral the Brodmann area (BA) 17 area, left inferior frontal gyrus, left middle temporal gyrus, left middle temporal gyrus and left middle occipital gyrus (Fig. 2, Table 4). Compared with HCs, the PAs exhibited decreased FC between bilateral BA18 with bilateral superior temporal gyri, left inferior frontal gyrus, left inferior temporal gyrus, left middle occipital gyrus, left medial frontal gyrus and right middle frontal gyrus (Fig. 3, Table 5).

ROI-wise FC analysis between subareas within visual network. The synchrony of time series
between the 16 nodes (8 nodes in each hemisphere) within visual network is measured by correlation analysis to determine the significant differences in FC among these nodes. Average FC of the 16 nodes is compared using two-sample t-test between Pearson correlation coefficients of two groups (Fig. 4). In the left hemisphere, compared with HCs the PAs showed increased FC between left V1 and left V3, decreased FC between left V2v and left V5. Also, in comparison with the normal controls, the FC in pituitary adenoma group significantly increased between the left V4 and V2d, V3 (Fig. 4c,d). Meanwhile, the PAs exhibited decreased FC between right VP with left V2v, and between left V5 and right V3, V3a (Fig. 4c).

Discussion
In this study, we explored the regional and circuit visual-related resting-state activity alterations in pituitary adenomas patients with visual damages. As an emerging method, the resting-state fMRI is an efficient approach to exploring brain connectivity and plasticity with reliable results, particularly in patients who cannot perform task fMRI scans 12 .
It was hypothesized that the reorganization of the brain cortex occurs when the partially impaired vision failed to facilitate brain work efficiently in respond to the compromised visual input 13 . In our study, we found remarkable changes of the ReHo values in the visual areas and other cortices, compared with the HCs, the PAs group showed increased ReHo value in the left superior occipital gyrus and bilateral middle occipital gyri and and reduced ReHo value in the left inferior frontal gyrus and right middle temporal gyrus. Furthermore, we The cortical visual system occupies the largest cortices of the brain, which is universally thought to be segregated into ventral stream and dorsal stream 14 . The ventral steam starts with visual area V1, relays to visual area V2, and finally reaches the inferior temporal cortex (IT cortex). It is also called "what stream", which is associated with the object recognition. In contrast, the dorsal stream is referred to as "where pathway" that begins with visual area V1, passes through visual area V2, and visual area MT (middle temporal/V5), finally project dorsally to the parietal lobe. The function of the dorsal stream is associated with object locations 5 . In our study, we constructed the visual systems by selecting the key nodes of the visual stream to investigate the underlying changes caused by partial visual impairment.
ReHo represents the local spontaneous coherence of neural activity and is one of the most efficient, reliable, and widely used index to inspect the brain activity 15 . And the functional connectivity (FC) is widely used to characterize functional relationships in the spatially distant brain regions and has proven to be a very reliable method 16 . The alteration of ReHo or FC reflects an abnormal changes of brain activity 17 . In our study, vision-related areas (bilateral middle occipital gyri, left precuneus) showed a significantly increase of ReHo in the patients group. We also found the ReHo of the right middle temporal gyrus decreased significantly. There were several studies investigating the changes of the resting-state activity in partially visual impaired patients with glaucoma and other ophthalmological diseases [18][19][20] , these studies found there were significant changes in FC of the visual cortex between the visual cortex and other vision-related brain regions. And a study found that visual cortex could undergo structure and/or function changes following deprived retinal input 21 . Several studies have investigated the visual cortex changes in lesioned visual pathways 22,23 , the findings of which showed the recovery of the visual field perimetry correlates with the V1 BOLD signal changes in previously visually deprived areas. A study explored the alteration of vision-related resting-state activities in patients with anterior visual pathway lesions 11 . The results showed significant vision-related cortex subareas within the visual cortex FC and ReHo alterations. The above-mentioned studies indicated that when visual input to the visual cortex is damaged, the visual system may undergo compensatory changes. However, these researches mainly focus on the alteration of the vision-related cortex without studying the visual system itself. In our study, we did not find the ReHo changes  in the primary visual cortex(V1), but the bilateral V2/V3 visual areas showed increased and the right middle temporal gyrus exhibited reduced ReHo. As we know, the intrinsic functional network is a robust system which can keep the brain function operating efficiently even under the circumstances of brain lesions. When the input of the visual cortex is compromised, the related visual areas can adjust their function to the alterations. Evidences showed that the primary visual cortex(V1), an early visual cortex, lacks of reorganization ability. However, the higher-order visual areas can make relevant changes to ensure the normal visual function 24,25 . Our results showed there were significant increased ReHo in bilateral middle occipital cortices(V2/V3), reflecting the dysfunction of the higher-order visual cortices. A previous study investigated the multisensory plasticity of visual stream functions between congenitally blind people and sighted people using task fMRI 26 . Their results showed that compared with normal sighted controls, the congenitally blind participants exhibited activated extrastriate cortex and the auditory cortex. Furthermore, their findings showed that dorsal stream functions could develop through non-visual spatial information at an early age. These findings indicated that the visual deprivation may cause compensatory intra-modal and cross-modal plasticity in the visual system. A study investigated the microstructural differences between the dorsal and ventral visual pathways in congenitally blind, late blind and normal sighted controls. They revealed that fractional anisotropy in the ventral stream of congenital and late blind individuals decreased. It seems that the blindness can selectively affect the microstructure of the ventral visual stream 27 . Another study reported that in congenitally blind individuals, the functional connectivity in ventral stream areas is more easily to be affected than that in the dorsal stream 28 . These studies mainly explored the changes of visual streams of blind participants, but the changes of the visual streams associated with partial visual impairment are still largely unknown. In line with their studies, we identified that the FC between the left V1 and left V3 increased while the FC between left V2v and left V5 significantly decreased. The results indicated that the FC in ventral stream increased while the FC dorsal stream decreased. It seems that the increased FC in the ventral stream in pituitary adenoma subjects is a result of compensatory plasticity, affecting the intrinsic connectivity and functional coupling activities in patients with partial visual loss.
The current concepts of visual processing is that the visual system does not work separately. In fact, there are extensive cross-talks between the two streams 29 . Goodale et al. proposed three possible cross-stream talk models 30 , one of which is the 'continuous cross-talk' account. In this model, information can be transferred between the two streams at multiple stages in the processing pathways. In our study, the PAs group exhibited increased FC between left V4 area and the V3, V2d area. Our results indicated increased FC between ventral stream and dorsal stream, which may reflect the enhancement of functional integration process.
The BA17 areas is also called the primary visual cortices which receives the direct visual stimuli from the retina. A resting-state study found that primary visual cortex was functionally associated with other brain areas, including the precuneus, the precentral/postcentral gyrus, the middle frontal gyrus, the fusiform gyrus, the inferior/middle temporal gyrus. They indicated that the functional connectivity may be related with processes of memory-related mental imagery and/or visual memory consolidation 31 . In our study, we identified that the FC between the BA17 area and the left middle temporal gyrus, left middle temporal gyrus, left middle occipital gyrus diminished. The decreased FC between the BA17 area and these regions in pituitary adenoma patients may thus indicate the dysfunction of these processes. Jann et al. described in visual damaged patients the FC between the BA 17 area and language network decreased 32 . The FC decreased between the BA17 area and the left inferior frontal gyrus, which is crucial for the comprehension and production of language. This finding may reflect the dysfunction of visual and language information integration. Our study also demonstrated decreased FC between the primary visual cortex (BA17) and higher visual cortices (BA19), which indicated the visual information processing pathway from primary visual area to higher visual cortices may be compromised.
Our study showed decreased FC between the BA18 area and bilateral superior temporal gyri, left inferior frontal gyrus, left inferior temporal gyrus, left middle occipital gyrus, left medial frontal gyrus and right middle frontal gyrus in the PAs compared with the HCs. The FC decreased between the BA 18 area and the left inferior temporal gyrus, which plays a key function in visual, auditory and sensory integration as it will impair the integration of visual and other sensory stimuli. Also, the PAs group showed decreased FC between the BA 18 area and some higher-order cortices. The superior temporal gyrus is responsible for processing sounds and speech 33 .
It was proved that the medial frontal gyrus may play a role in executive function 34 . Decreased FC between the BA 18 visual area and cognitive cortices may be caused by the decreased integration of visual information into these functions.
In our study, the FC between the BA17/18 area and the vermis, cerebellum decreased in PAs group. These findings may be a result of diminished input from the visual area, which may indicate the compromised integration of visual information and voluntary movement and postural balance.

Conclusions
In our study, we identified that the ReHo and FC altered between the vision-related cortices and other higher-order cognitive cortices along with disorganized functional connectivity within the visual system in PA patients with visual impairment. These findings may provide important insights to understand the plasticity of visual network.  Table 3. Group difference in ReHo between two groups.  Table 4. Brain regions with BA17 FC differences between PAs compared with HCs.    (20/20)or visual field defect of more than 25% of the visual field at least one eye. For both the pituitary adenoma patients and health controls the visual acuity was assessed with the E chart. Using the standardized automated perimetry (Octopus900 Perimetry) to obtain the visual field data. All of the participants were fully informed prior to the study and signed consent forms. The study was granted ethical approval by the Institutional Review Board of Beijing Tiantan Hospital affiliated to Capital Medical University, and the study was carried out in accordance with the relevant guidelines and regulations.

Participants.
Image acquisition. The rs-fMRI data were obtained using a 3.0 Tesla Siemens scanner with a standard head coil. The rs-fMRI data were acquired with an echo-planar image sequence (  ScIeNtIfIc REPORtS | 7: 13074 | DOI:10.1038/s41598-017-13214-5 Regional homogeneity(ReHo) analysis. As a data-driven method, ReHo was used to measure local synchronization of fluctuations of spontaneous blood oxygen level dependent(BOLD) signals with nearest neighbouring voxels 37 . We used DPABI to calculate each subject's ReHo parametric map that each voxel presents the Kendall's coefficient of concordance of its nearest neighbouring 27-voxels cluster based on unsmoothed preprocessed images. Two sample t-test of each voxel between patients and HCs on ReHo maps were performed after division by global mean value, Fisher z-transformation and spatial smoothness with a 4 * 4 * 4 mm full-width-half-maximum kernel. The gibbs random field (GRF) method was used to correct for multiple comparisons. The corrected value of p < 0.001 in voxel level and p < 0.05 in cluster level were used as the threshold.
Seed-based resting state functional connectivity(FC) analysis. As a priori-based method, seed-based FC analysis is one of the most direct ways to measure correlations between distant brain regions which includes voxel-wise and region-of-interest(ROI)-wise FC 38 . Higher-order visual processing pathway is considered to be divided into two different streams. The dorsal stream is considered as the "where pathway" and "how pathway" that associated with perception of the spatial locations of an object. The ventral stream is referred to as the "what pathway" which is responsible for object recognition and the storage of long-term memory. In this study, we defined these 16 nodes (8 nodes which includes V1, V2d, V2v, V3, V3a, V4, VP and V5 areas in each hemisphere) within 5-mm radius around the coordinates as ROIs in the visual streams to compute ROI-wise correlation matrix 39 . The coordinates of these nodes were selected and obtained from a previous study 40 . We also  Table 6. The coordinates of ROI-wise seed nodes. defined V1(Brodmann Area 17) and V2(Brodmann Area 18) areas in both hemispheres as seed ROIs to compute voxel-wise FC maps (Table 6, Fig. 5). Pearson's correlation analysis was performed between each pair of time courses in these 16 nodes and between the seed reference time course and the time series from the whole brain in a voxel. The correlation coefficients were converted to z values using Fisher's r-to-z transformation.
Data Availability. The datasets generated and analysed during the current study are available from the corresponding author on reasonable requests.