Introduction

Glaucoma is an age-related neurodegenerative disease that can lead to irreversible blindness^1,2. Recent studies have indicated potential links between neurodegenerative diseases and cerebrospinal fluid (CSF) dynamics^3,4, which can be driven by visual stimulation⁵. Nevertheless, such an impact has not been studied in glaucoma patients. Thus, our study aims to investigate the impact of visual stimulation on CSF flow in glaucoma patients using functional MRI.

Methods

Eighteen early glaucoma patients [age= 66.00±1.89 years (mean±S.E.M.); 27.78% male], twenty-four advanced glaucoma patients [age= 66.63±1.36 years; 54.17% male] and twenty-three healthy controls [age= 64.52±1.59 years; 39.13% male] underwent clinical ophthalmic testing using spectral-domain optical coherence tomography (Cirrus HD-OCT; Zeiss, Dublin, CA, USA), and the Humphrey visual field perimetry via the Swedish Interactive Thresholding Algorithm 24-2 standard (Zeiss, Dublin, CA, USA). The categorization of early versus advanced stages was determined based on visual field mean deviation (MD), using a cut-off of -6.0 dB. Then subjects underwent anatomical MRI (voxel size=0.8×0.8×0.8 mm³) and functional MRI (voxel size=2.3×2.3×2.3 mm³) inside a 3-Tesla Siemens Prisma MRI scanner. For functional MRI, we collected two runs of functional MR images using a gradient-echo echo-planar imaging (EPI) sequence (1 run= 5 min, TR/TE= 1000/32.60 ms). We placed the first imaging slice at the bottom of the fourth ventricle to detect incoming CSF flow (Figure 1a)⁶. During fMRI scan, flickering checkerboard visual stimuli were presented on a full screen (Figure 1b) for the purpose of inducing large-scale neural activity (Figure 1c).
The fMRI data was motion-corrected without spatial or temporal smoothing and registered to the individual’s structural template. Then we extracted the blood-oxygenation-level-dependent (BOLD) signals from the visual cortex mask (hOc1-hOc4v) as well as the fourth ventricle mask which we manually delineated from the first slice of fMRI imaging. We removed the linear trend and obtained the percentage change of BOLD signals for each voxel of masks across time. We also obtained from the clinical ophthalmic tests each individual’s intraocular pressure (IOP), visual field MD, and OCT’s optic nerve head cup-to-disc ratio (C/D), neuroretinal rim area, peripapillary retinal nerve fiber layer (pRNFL) thickness, and macular ganglion cell-inner plexiform layer (mGCIPL) thickness for associations with MRI measurements.

Results

We first investigated whether the stimulation-induced BOLD signal in the visual cortex would lead to anticorrelated CSF inflow in both healthy older adults and glaucoma individuals. Consistent with a recent study in young adults⁵, in our healthy individuals, the CSF inflow signal was gradually suppressed when the stimulus was presented (1~8 seconds after stimulus onset; decreasing trend across 1~8 seconds, F(1,22)=7.711, P=0.011). The CSF inflow signal reached its lowest value at 7 seconds after stimulus onset (T(22)=-3.471, P=0.002, one-sample t-test). Following this, the CSF inflow signal gradually began to increase until 14 seconds after stimulus offset (increasing trend across 1~14 seconds after stimulus offset, F(1,22)=11.068, P=0.003), reaching its peak value at 10 seconds after stimulus offset (T(22)=-3.858, P<0.001, one-sample t-test). However, these stimulus-locked CSF inflow patterns became weakened in early (decreasing trend across 1~8 seconds after stimulus onset, F(1,17)=4.952, P=0.040; increasing trend across 1~14 seconds after stimulus offset, F(1,17)=1.705, P=0.209) and advanced glaucoma patients (decreasing trend across 1~8 seconds after stimulus onset, F(1,23)=0.190, P=0.667; increasing trend across 1~14 seconds after stimulus offset, F(1,23)=0.021, P=0.886), indicating that the influence of BOLD activity on CSF inflow is reduced in glaucoma (Figure 2).
Next, we investigated the temporal relationship between visual cortex BOLD signal and CSF inflow signal by calculating the cross-correlation between two signals. The results showed that the CSF inflow signal exhibits the strongest negative correlation with the BOLD signal at a lag of 1 second (Figure 3). The absolute value of the correlation coefficient, which reflects coupling between the BOLD and CSF inflow signals, gradually decreased among early and advanced glaucoma patients (linear trend, F(1,62)=11.178, P=0.001; main effect of group, F(2,62)=6.102, P=0.004; healthy control vs early glaucoma, P>0.05; healthy control vs advanced glaucoma, P=0.004; early vs advanced glaucoma, P=0.054, Bonferroni-corrected).
Additionally, the clinical ophthalmic measures were correlated with the coupling between the BOLD and CSF inflow signals (Figure 4). Specifically, reduced coupling was associated with smaller values of MD (P=0.002), pRNFL thickness (P=0.009), mGCIPL thickness (P=0.054), rim area (P=0.006), and a higher C/D ratio (P=0.003), but not with IOP (P=0.524).

Discussion and Conclusion

Our study shows that the stimulus-locked CSF inflow signals are reduced in glaucoma. As glaucoma severity increases, the coupling between the BOLD and CSF inflow signals decreases. This finding offers a novel perspective of the glaucoma pathogenesis, suggesting that glaucoma involves alterations in the CSF dynamics apart from neurodegeneration along the visual pathway.

Acknowledgements

This work is supported in part by the National Institutes of Health R01-EY028125, R01-EY013178, and P41-EB017183 (Bethesda, Maryland), BrightFocus Foundation G2016030, G2019103, and G2021001F (Clarksburg, Maryland), and an unrestricted grant from Research to Prevent Blindness to NYU Langone Health Department of Ophthalmology (New York, New York).