Structural Brain Abnormalities in Patients with Primary Open-Angle Glaucoma: A Study with 3T MR Imaging

Abstract

Purpose.: Weexamined changes of the central nervous system in patients with advanced primary open-angle glaucoma (POAG).

Methods.: The clinical observational study included 15 patients with bilateral advanced POAG and 15 healthy normal control subjects, matched for age and sex with the study group. Retinal nerve fiber layer (RNFL) thickness was measured by optical coherence tomography (OCT). Using a 3-dimensional magnetization-prepared rapid gradient-echo sequence (3D–MP-RAGE) of magnetic resonance imaging (MRI) and optimized voxel-based morphometry (VBM), we measured the cross-sectional area of the optic nerve and optic chiasm, and the gray matter volume of the brain.

Results.: Patients in the POAG group compared to the subjects in the control group showed a significant (P < 0.001) decrease in the bilateral gray-matter volume in the lingual gyrus, calcarine gyrus, postcentral gyrus, superior frontal gyrus, inferior frontal gyrus, and rolandic operculum, as well as in the right cuneus, right inferior occipital gyrus, left paracentral lobule, and right supramarginal gyrus. Patients in the study group showed a significant increase in the bilateral gray matter volume in the middle temporal gyrus, inferior parietal gyrus, and angular gyrus, and in the left gray matter volume in the superior parietal gyrus, precuneus, and middle occipital gyrus. In addition, the cross-sectional area of the optic nerve and optic chiasm, and RNFL thickness were significantly decreased in the POAG group.

Conclusions.: In patients with POAG, three-dimensional MRI revealed widespread abnormalities in the central nervous system beyond the visual cortex.

Introduction

Primary open angle glaucoma (POAG) has been defined formerly by intraocular morphologic changes, such as progressive retinal ganglion cell loss and defects in the retinal nerve fiber layer (RNFL), and by corresponding psychophysical abnormalities, such as visual field loss. ¹ Recent studies by various researchers, however, have suggested that the entire visual pathway may be involved in glaucoma. ^2–23 Findings from these experimental studies, human autopsy investigations, and in vivo neuro-imaging studies of glaucoma patients indicated that glaucoma is a complex disorder in which the whole visual pathway from the retina to the primary visual cortex may be involved. These studies, in particular those using neuro-imaging techniques, were limited by the age of the patients with young patients and those with advanced glaucoma usually not included into the studies, and by focusing on the visual pathway to the visual cortex mostly without examining other cerebral regions, such as the gray matter in the superior cortex. Previous studies indicated, however, that glaucoma damage in the receptive fields in the retina could eliminate the stimulation of parts of the visual cortex, resulting in the degeneration of inactive cortical neuronal tissue. ¹⁴ Therefore, it has been hypothesized that patients with advanced POAG could experience changes in the whole brain. Also, the neuro-imaging technologies, namely magnetic resonance imaging (MRI), have been refined further, resulting in higher spatial resolution. In addition, optimized voxel-based morphometry (VBM) as a whole-brain, semi-automated technique for characterizing regional cerebral differences in high-resolution MRIs with minimal operator dependence has been developed. ²⁴ A so-called optimized VBM incorporates additional spatial processing steps ahead of the statistical analysis. The optimization steps help in excluding the non-brain voxels before normalization and subsequent segmentation, and help to avoid the potential bias due to systematic differences in skull size and shape or scalp thickness between the study groups. ²⁵ In view of these new neuro-imaging technologies and in view of the possibilities of further improvement of the previous investigations, we conducted the present study to examine relatively young patients with advanced glaucomatous optic neuropathy by applying optimized VBM-assisted MRI. 

Methods

Subjects

A total of 15 patients (9 men) with an age of 40 to 50 years and bilateral advanced POAG formed the study group, and 15 control subjects matched for age and sex with the study group were included into the control group of the study. The study followed the tenets of the Declaration of Helsinki, and it was approved by the ethics committee of the West China Hospital, Sichuan University Institute, Chengdu, Sichuan Province, P.R. China. Informed consent was obtained from the subjects after explanation. All participants in this study underwent an ophthalmologic examination (visual acuity assessment, refractometry, measurement of central corneal thickness, slit-lamp–assisted biomicroscopy of the anterior and posterior segment of the eye, applanation tonometry, gonioscopy, photography of the optic nerve head and fundus, and standard automated perimetry; Octopus 101 perimeter; Interzeag, Bern, Switzerland), and a systematic body examination. All patients were followed for at least 6 months. The diagnosis of POAG was made by two glaucoma specialists (XL, XC) independently of each other. The inclusion criterion for patients in the study group was bilateral advanced POAG, as defined by a cup-to-disc diameter ratio ≥0.9, a mean perimetric defect ≥15 decibels (dB), preservation of central vision in both eyes, and open anterior chamber angles upon gonioscopy. Exclusion criteria were any other ocular, neurologic or psychiatric disorder; any history or clinical signs of auto-immune diseases, cardiovascular diseases, cerebrovascular diseases, and diabetes mellitus; and blood pressure measurements of >140/90 or antihypertensive medication. 

All study participants underwent measurement of the RNFL thickness by optical coherence tomography (OCT, Heidelberg Spectralis OCT; Heidelberg Engineering Co. Heidelberg, Germany). The fast measurement scan protocol was used (3.4 mm diameter). The scanning was completed following a 10-minute dark adaptation to ascertain that the pupils were large enough to permit imaging (usually 5 mm) without pharmacologic dilation of the pupil. An OCT scan of good quality was defined by a uniform brightness across the scan circumference. The optic disc was centered in all scans by the scanning technician. 

MRI Data Acquisition

All subjects were examined by a 3 Tesla MRI scanner (Magnetom Trio; Siemens Co., Erlangen, Germany) with an 8-channel phased-array head coil as a magnetic resonance signal receiver. The subjects were required to close their eyes and avoid any eye movements during the image acquisition. To secure the repeatability of the MRI scan, the axial plane of each MRI session was set parallel to the line from the anterior commissure to the posterior commissure on sagittal localizer images. ²⁶ The MRI protocol scan contained a transverse-axial turbo spin-echo imaging sequence with T2-weighted and a 3-dimensional magnetization-prepared rapid gradient-echo (3D–MP-RAGE) sequence. T2-weighed images were used for conventional screening with parameters: TR/TE = 4000/83 ms and slice thickness 3 mm. Sequence parameters for MP-RAGE were: TR/TE = 1900/2.27 ms, field of view (FOV) of 240 × 240 × 170, data matrix of 256 × 256 × 176, resulting in an approximate isotropic voxel size 1 × 1 × 1 mm. 

Optic Nerve Cross-Sectional Area Measurement

Using Syngo MultiModality Workplace (Series Number 3064, Version VE31A; Siemens Co.), we reconstructed axial images perpendicular to the optic nerve through the middle point along the line between the retrobulbar optic nerve and annulus tendineus of Zinn (Fig. 1A). The region of interest (ROI) of the optic nerve was drawn manually in the reconstructed plane (Figs. 1, 2). To ensure the reliability of the ROI measurement, the same procedure was performed 3 times by three radiologists (ZF, QW, XH) in a masked manner. A mean value of each optic cross-sectional area was calculated from the results of the three measurements. In the targeted MP-RAGE images, the optic nerve showed a low dark signal against the cerebrospinal fluid, while the surrounding tissue showed the bright signal. Hence, quantification of the optic nerve was facilitated by the excellent contrast of the nerve against its surroundings. The low and dark area was the ROI. 

Figure 1

 

Reconstruction of the axial images perpendicular to the optic nerve. (A) Orbital optic nerve allowing for the construction of a plane parallel to the optic nerve (red line). The middle point was identified along the marked line parallel to the optic nerve from the beginning of the retrobulbar optic nerve to annulus of Zinn. Reconstruction of the targeted plane was performed perpendicular to the optic nerve through the middle point (green line). (B) The sagittal image showing the placement of the reconstruction plane at the middle point of the optic nerve (green line). (C) The reconstructed coronal plane showing the axial targeted image of the optical nerve.

Figure 2

 

ROI of the optic nerve defined as the cross-sectional image of both eyes at a reconstructed oblique-coronal plane.

Optic Chiasm Cross-Sectional Area Measurement

The same software was applied to measure the optic chiasm cross-sectional area. A plane perpendicular to the optic chiasm through its middle point was reconstructed (Fig. 3). The ROI drawing of the optic chiasm was made manually in the reconstructed plane (Fig. 3) by the same three radiologists (ZF, QW, XH). These three radiologists performed the same procedure in a masked fashion three times. The mean value of each optic cross-sectional area was calculated from the results of the three measurements. In the targeted images, the optic chiasm yielded a high, white signal, while cerebrospinal fluid showed a gray signal. The white signaled area was the ROI. 

Figure 3

 

Reconstruction of the plane perpendicular to the optic chiasm and the ROI drawing. (A) Placement of the median sagittal plane with a full sight of the optic chiasm as the primary plane, reconstructing a plane perpendicular to the optic chiasm through the middle point. The green line showed the position of the reconstructed target plane. (B) A coronal image of the target plane. ROI drawing was along the edge of the white signaled area.

Optimized VBM

The structural images were preprocessed using the optimized VBM implemented in SPM2 (available in the public domain at www.fil.ion.ucl.ac.uk/spm), running under Matlab (MathWorks, Natick, MA). The image processing procedures already have been described in detail. ²⁷ The first step was the creation of a study-specific gray matter template image for all study participants. T1 images of each participant were normalized to the T1 template of SPM2 using an affine-only cut-off. After normalization, the images were averaged and smoothed with an 8 mm kernel, creating the customized T1 template. The normalized images then were segmented into three different tissues: gray matter, white matter, and cerebrospinal fluid. Non-brain voxels were excluded from the statistical analysis by applying a brain mask. Gray matter partitions were normalized spatially using a 12-parameter affine transformation and 7 × 8 × 7 nonlinear basis functions, which were the default normalization parameters in SPM2 to a customized gray matter template constructed from the normalized, segmented, and smoothed gray matter datasets of all study participants. The deformation parameters obtained from the normalization process were applied to the original raw images (native space) of all participants to create optimally normalized whole-brain images, which were segmented recursively and brain–tissue-extracted. The optimally processed images then were smoothed with an isotropic Gaussian kernel with full width-half maximum of 10 mm. The differences in gray matter volume of the whole brain between the two groups were assessed statistically with two-sample t-tests. All results were presented at the voxel level. In addition, a small volume correction was performed using the WFU PickAtlas software V.2 to reduce the number of voxels entering the statistical computation, ²⁷ and corrected for multiple comparisons using a family-wise error rate of P < 0.05. The anatomic location (i.e., the coordinates obtained for the peak voxels) of the maximal gray matter loss in each significant cluster was transferred into Talairach space using Matthew Brett's mni2tal routine (available in the public domain at http://imaging.mrc-cbu.cam.acuk/down load /MNI2tal). Data were analyzed on a personal computer by means of Windows XP Professional V.5.1, and Matlab7.0.1 and SPM2 (available in the public domain at http://www.fil.ion.ucl.ac.uk/spm). 

Statistical analysis was performed by using the Statistical Package for the Social Sciences (SPSS, version 20.0; IBM-SPSS, Chicago, IL). Comparison between patients and control subjects was performed using an independent-sample t-test. All P values < 0.05 were considered statistically significant based on a 2-tailed test. 

Results

The mean age of the 15 patients (9 men) in the advanced POAG study group was 43.3 ± 4.1 years (mean ± SD). The subjects in the control group were matched for age (43.9 ± 3.8 years, P = 0.68) and sex (9 men) with the POAG patients. Visual acuity, indices of visual field examinations and intraocular pressure in the glaucoma group were summarized in Table 1. The right and left eyes did not differ significantly (P > 0.05) in these parameters (Table 1). Visual acuity in the control group was at least 20/20, and intraocular pressure was within the range of 10 to 21 mm Hg. Typical visual fields of patients of the glaucoma group were presented in Figures 4 and 5. The mean thickness of the RNFL as a whole (48.5 ± 15.1 vs. 102.7 ± 22.7 μm) and measured separately in the four quadrants was significantly (all P < 0.001) lower in the study group than in the control group. 

Figure 4

 

Visual field result of the right eye of a patient with POAG.

Figure 5

 

Visual field result of the left eye of a patient with POAG.

Table 1. 

 

Ocular Characteristics (Mean ± SD) in the POAG Group

	Right Eyes	Left Eyes	P Value
Visual acuity, LogMAR	0.15 ± 0.13	0.16 ± 0.15	0.43
IOP before treatment	28.1 ± 3.9	30.1 ± 4.3	0.64
IOP after treatment	15.6 ± 3.6	15.0 ± 3.4	0.78
MD of visual field	20.6 ± 2.9	21.6 ± 3.2	0.36
sLV of visual field	5.36 ± 0.97	5.52 ± 1.02	0.62

 IOP, intraocular pressure (mm Hg); MD, mean perimetric defect; sLV, square root of the perimetric loss of variance; LogMAR, logarithm of the minimum angle of resolution; P value, statistical significance of difference between right and left eyes.

The cross-sectional areas of the optic nerve (6.67 ± 2.23 vs. 10.10 ± 3.14 mm²) and the optic chiasm (20.38 ± 4.94 vs. 35.10 ± 4.63 mm²) as measured on the MRI images were significantly (all P < 0.001) smaller in the study group than in the control group. 

The VBM analysis revealed that the glaucomatous study group compared to the control group showed a significantly decreased gray matter volume in the lingual gyrus, calcarine gyrus, postcentral gyrus, superior frontal gyrus, inferior frontal gyrus, and rolandic operculum of both sides, and in the right inferior occipital gyrus, left paracentral lobule, right supramarginal gyrus, and right cuneus (Table 2, Fig. 6). The gray matter volume was significantly larger in the study group than in the control group in both sides of the middle temporal gyrus, inferior parietal gyrus, angular gyrus, and left superior parietal gyrus, left precuneus, and left middle occipital gyrus (Table 3, Fig. 7). 

Figure 6

 

Comparison of areas of reduced gray matter among patients with an advanced POAG and in normal controls. The figure shows areas of regional changes in visual cortex area. (A) Cluster map of the whole brain. (B) Lingual gyrus and Calcarine fissure. (C) Right and left postcentral gyri. (D) Right and left superior frontal gyri. (E) right and left inferior frontal gyri. (F) Right and left Rolandic operculums. (G) Right inferior occipital gyrus. (H) Left paracentral lobule. (I) Left supramarginal gyrus.

Figure 7

 

The comparison of areas of increased gray matter among patients with advanced POAG and in normal control subjects. (A) Cluster map of the whole brain. (B) Right and left middle temporal gyri. (C) Right and left angular gyri. (D) Right inferior parietal gyrus. (E) Left inferior parietal gyrus. (F) Left superior parietal gyrus. (G) Left middle occipital gyrus. (H) Left precuneus.

Table 2. 

 

Clusters of Decreased Gray Matter Volume in Patients with POAG

Cluster Size (cm3)	Talairach x,y,z (mm)	Voxel T	Voxel P (unc)	Cluster P (unc)	Cluster P (c)	Brain Region
Control group > Glaucoma study group
16.9	8,−63,0	8.87	<0.001	<0.001	<0.001	Lingual, calcarine
	30,−90,13	7.6	<0.001			Occipital_inf_R
	14,−70,20	7.05	<0.001			Cuneus_R
7.2	60,−2,28	6.93	<0.001	<0.001	<0.001	Postcentral_R
	55,−3,10	6.29	<0.001			Rolandic_oper_R
	59,18,19	6.09	<0.001			Frontal_inf_oper_R
6.1	−33,4,15	6.16	<0.001	<0.001	<0.001	Rolandic_oper_L
	−44,5,11	5.14	<0.001			Frontal_inf_oper_L
	−63,−25,20					Supramarginal_L
	−56,−6,28	5.04	<0.001			Postcentral_L
4	−5,−20,64	5.59	<0.001	<0.001	<0.001	Paracentral_lobule_L
	15,−10,76	5.02	<0.001			Frontal_super_R
	−17,−10,73	4.96	<0.001			Frontal_super_L

Table 3. 

 

Clusters of Increased Gray Matter Volume in Patients with POAG

Cluster Size (cm3)	Talairach x,y,z (mm)	Voxel T	Voxel P (unc)	Cluster P (unc)	Cluster P (c)	Brain Region
POAG > NC
5.5	52,−38,−4	5.17	<0.001	<0.001	0.01	Temple_mid_R
	51,−58,46	4.96	<0.001			Parietal_inf_R
	37,−50,44	4.3	<0.001			Parietal_inf_R
	59,−55,25					Angular_R
1.4	−27,−50,53	4.86	<0.001	0.043	0.732	Parietal_inf_L
	−32,−56,57	4.77	<0.001			Parietal_sup_L
	−32,−67,55	4.59	<0.001			Parietal_sup_L
1.7	−9,−64,49	4.74	<0.001	0.027	0.557	Precuneus_L
	−5,−71,60	4.64	<0.001			Precuneus_L
	−8,−53,51	3.83	<0.001			Precuneus_L
1.7	−40,−77,25	4.59	<0.001	0.025	0.535	Occipital_mid_L
	−49,−69,16	3.71	0.001			Temporal_mid_L
	−36,−82,30	3.46	0.001			Occipital_mid_L
3.2	−48,−56,26	4.42	<0.001	0.004	0.111	Angular_L
	−55,−46,4	4.29	<0.001			Temporal_mid_L
	−42,−71,38	4.05	<0.001			Angular_L

Discussion

Glaucoma is an optic neuropathy characterized by apoptotic death of retinal ganglion cells. To date, the pathology of glaucoma has been examined extensively at the level of the retina, optic nerve head, intracranial optic nerves, lateral geniculate ganglion, and primary visual cortex. ^3–8 Most of these studies on glaucoma-associated changes in the human brain were postmortem investigations studies or assessed biochemical changes (e.g., cytochrome oxidase) in primates. In recent years, few neuro-imaging studies on the lateral geniculate ganglion and the visual central area were performed. Using 1.5-T vivo MRI, Gupta et al. detected an atrophy in the lateral geniculate ganglion in patients with POAG. ⁵ Applying a 3-T Diffusion-Tensor MRI, Garaci et al. noticed an increased diffusion tensor and decreased fractional anisotropy, reflecting axonal disruption of the optic nerves and optic radiations in patients with POAG. ²⁸ Kitsos et al. reported on an increase in the number of white matter hyperintensities, suggesting cerebrovascular changes in POAG. ¹² In a recent study, Zikou et al. used a conventional VBM and diffusion tension imaging, and examined the visual pathway in patients with POAG who showed a thinning of the left temporal and right nasal retinal nerve fiber layer. ²³ VBM revealed a significant reduction in the left visual cortex volume, left lateral geniculate nucleus, and chiasm. In addition, fraction analysis of diffusion tension imaging was decreased significantly in the inferior fronto-occipital fasciculus, longitudinal and inferior frontal fasciculi, putamen, caudate nucleus, anterior and posterior thalamic radiations, and anterior and posterior limbs of the internal capsule of the left hemisphere. Zikou et al. concluded that neurodegenerative changes beyond of the optic pathway could be found in patients with POAG. ²³ Besides this recent investigation by Zikou et al., who used a conventional VBM, it has remained unclear whether the higher brain cortex is affected or which other regions of the brain than those described above are affected in patients with POAG. 

VBM is a whole-brain, nonbiased, semi-automated technique for characterizing regional cerebral differences in high-resolution MRIs, suitable for the comparison of the gray or white matter. VBM is useful in characterizing subtle and gross structural changes in the brain in a variety of neurologic and psychiatric diseases, such as schizophrenia, multiple sclerosis, and Alzheimer's disease. It also has been used in studies of normal subjects, focusing on the impact of learning and practice on the brain structure. For instance, some changes were reported in the structure of the brain when humans learned to navigate and to speak a second language. ^29,30 An optimized VBM can help to quantify group differences in neuro-anatomy without any prior hypothesis. Previous studies in this area focused mainly on case reports, and none of the studies used an optimized VBM to examine damages in the higher cortex of patients with POAG. In our study, patients with advanced POAG showed a significant reduction of the gray matter in the visual cortex. Based on an optimized VBM analysis, the patients with advanced POAG demonstrated not only an extensive decrease of the gray matter volume in various regions of the brain, but also an increased gray matter volume in other specific brain areas (Tables 2 and 3, Fig. 8). According to the VBM analysis in our study, patients with POAG showed a bilateral symmetric decrease in the gray matter volume in the occipital cortex. The clusters contained the primary visual cortex (Brodmann area 17) and the secondary visual cortex (Brodmann areas 18 and 19), with the occipital pole area corresponding to the central retina. ³¹ The decreased clusters in the patients with POAG excluded the occipital pole, which might be due to the preserved central vision of the involved patients as an inclusion criterion of the study. Consistent with this result, Boucard et al. found a reduction of the cortical gray matter density in glaucoma patients in the anterior half of the medial occipital cortex. ¹³ Dai et al. showed a disrupted functional connectivity between the visual cortex and associative visual areas using resting-state functional MRI in patients with POAG. ³² Our study detected extensive structural atrophy in patients with advanced POAG. All the brain regions involved were associated with visual functions. The parietal and frontal areas, along with their association connections, represent a potential cortical network for visual reaching. ³³ The frontal lobe has an important role in the formation of the optic localization, multiple retinal correspondence, and scotoma of the fixation point. It has been known that the postcentral gyrus is an important region for processing of higher order somatosensory and visual information, and the supramarginal gyrus has a critical role in visual word recognition. ³⁴  

Figure 8

 

Possible trend of the central nerve system changes in advanced POAG.

The optimized VBM analysis showed that the patients with POAG had an increased gray matter volume in regions with different functions. It might have been due to a functional reconstruction of the brain in the sense of cerebral plasticity. Longstanding vision loss may lead to a decrease in cerebral centers primarily or secondarily related to vision, and to a transmodal compensatory increase in nonvisual functions and corresponding morphologic changes in the brain. One of the important regions involved was the middle temporal visual area, which has a critical role in visual motion processing. The parietal lobe is associated with the spatial awareness and redirecting visual attention. ³⁵ Functional imaging findings in healthy subjects suggested a central role for the precuneus in a wide spectrum of highly integrated tasks, including visuospatial imagery, episodic memory retrieval, and self-processing operations, namely first-person perspective taking and an experience of agency. ³⁶ The damage in the angular gyrus might cause dyslexia. ³⁷ Neural plasticity to compensate for the glaucoma-related loss of visual input thus may have been responsible for the changes observed in the cerebral centers beyond the visual cortex. 

With family-wise error rate rectifications, the measurements obtained in some of the clusters were statistically significant only for one side of the brain (Tables 2, 3), although the glaucomatous damage to the optic nerve and visual field were bilateral. Such a statistical phenomenon occurs commonly in VBM analysis, and future studies on a larger number of study participants may address whether a bilateral glaucomatous damage to the visual afferent system leads mostly to bilateral changes in the cerebral cortex. 

In our study, the patients with advanced POAG showed a decrease in RNFL thickness, and cross-sectional areas of the optic nerve and optic chiasm, as well as an extensive gray matter volume change. Most previous studies with routine MRI sequences usually measured the cross-sectional area of the optic chiasm by an approximation through a section not perpendicular to the optic nerve or optic chiasm. ^12,38 Other investigations applied twice or multiple times the MRI scan to obtain the targeted plane of the precise cross-section. However, the use of high resolution 3D–MP-RAGE and the reconstruction procedure helped to resolve the problem by using a single scan in the present study. The advantages of this sequence included its relatively high soft tissue contrast and decreased imaging time, leading to a lower likelihood for motion artifacts. ³⁹ It resulted in a high resolution imaging protocol for cross-sections of the optic nerve and optic chiasm with no change in scanning time. A high resolution of the MRI images with a voxel size of 1 × 1 × 1 mm made the reconstruction results identifiable. Obtaining any part of the optic nerve cross-section in both eyes and in the optic chiasm during a single scanning sequence by using the method presented was feasible. 

Potential limitations of our study should be mentioned. First, the number of patients included into the study was relatively small. Despite of this weakness, however, the differences between the study and control groups were statistically significant. Second, our study was hospital-based, so that a confounding effect by a selection bias is possible as for any hospital-based investigation. Third, some of the observed differences between the study and control groups were significant for only one side of the brain. Future studies may address whether, with a larger number of study participants, these differences become significant, or whether biologic phenomena were the reason. Fourth, the imaging techniques applied in our study had their own limitations, namely in spatial resolution. Future investigations applying the then most sophisticated neuro-imaging techniques may give further information and clues on the question whether and how far glaucomatous defects in the eye lead to more or less widespread changes in the brain. 

In conclusion, our study detected changes in the visual cortex in patients with advanced POAG by means of high-resolution 3D–MP-RAGE and VBM. Detailed assessment of the visual pathways in multiple approaches may provide insight into the specific neurologic deficits in a patient with POAG. Our study supported the hypothesis and previous studies that POAG not only is an ocular disorder, but a disease involving the whole visual pathway. 

Footnotes

 Supported by the National Natural Science Foundation (Grant Nos. 81030027, 81227002 and 81220108013), National Key Technologies R&D Program of China (Program No. 2012BAI01B03), and China Postdoctoral Science Foundation (Grant No. 2012M520329). The authors alone are responsible for the content and writing of this paper.

Footnotes

³  These authors contributed equally to the work presented here and therefore should be regarded as equivalent authors.

Footnotes

 Disclosure: W.W. Chen, None; N. Wang, None; S. Cai, None; Z. Fang, None; M. Yu, None; Q. Wu, None; L. Tang, None; B. Guo, None; Y. Feng, None; J.B. Jonas, None; X. Chen, None; X. Liu, None; Q. Gong, None