Localised increase in regional cerebral perfusion in patients with visual snow syndrome: a pseudo-continuous arterial spin labelling study

Objectives We aimed to investigate changes in regional cerebral blood flow (rCBF) using arterial spin labelling (ASL) in patients with visual snow syndrome (VSS), in order to understand more about the underlying neurobiology of the condition, which remains mostly unknown. Methods We performed an MRI study in which whole-brain maps of rCBF were obtained using pseudo-continuous ASL. Twenty-four patients with VSS and an equal number of gender and age-matched healthy volunteers took part in the study. All subjects were examined with both a visual paradigm consisting of a visual-snow like stimulus, simulating key features of the snow, and a blank screen at rest, randomly presented. Results Patients with VSS had higher rCBF than controls over an extensive brain network, including the bilateral cuneus, precuneus, supplementary motor cortex, premotor cortex and posterior cingulate cortex, as well as the left primary auditory cortex, fusiform gyrus and cerebellum. These areas were largely analogous comparing patients either at rest, or when looking at a ‘snow-like’ visual stimulus. This widespread, similar pattern of perfusion differences in either condition suggests a neurophysiological signature of visual snow. Furthermore, right insula rCBF was increased in VSS subjects compared with controls during visual stimulation, reflecting a greater task-related change and suggesting a difference in interoceptive processing with constant perception of altered visual input. Conclusion The data suggest VSS patients have marked differences in brain processing of visual stimuli, validating its neurobiological basis.


Image pre-processing
Pre-processing of pCASL images was performed using Automated Software for ASL Processing (ASAP (1)) which employs the SPM software suite, version 12 (SPM 12;www.fil.ion.ucl.ac.uk/spm/). Computation of CBF maps was performed by the scanner computer using the following formula: = 600 / 1 2 1 (1 − − / 1 ) in which is the signal in the perfusion-weighted image (control-label), is the signal in the reference image, is the combined efficiency of labelling and background suppression (~65%), is the label duration (1825 ms), 1 is the 1 of arterial water, and is the postlabelling delay (2025 ms). For spatial normalization of the CBF maps to the space of the MNI within the ASAP framework, a multistep approach was used: CBF maps were co-registered to the T1-weighted structural ADNI images, after coarse alignment of the origin of both images. Unified segmentation of the T1-weighted image normalised this image to the MNI space and was used to produce a 'brain-only' binary mask which was multiplied by the co-registered rCBF map to produce an image free of extra-cerebral artefacts. The spatial transformation matrix was applied to the clean CBF images and then smoothed using an 8x8x8 mm Gaussian kernel. To measure global CBF signal, the ASAP toolbox was used to extract average CBF values from a grey matter mask for each subject. Probabilistic grey matter images in MNI space, derived from the FSL voxel-based morphometry toolbox, were thresholded to produce a mask which included all voxels with a >20% likelihood of being grey matter. A global CBF value, defined as the mean of all grey matter voxels within the mask, was computed for each individual pCASL CBF map. Mean global CBF for VSS patients and Ctrls in both conditions were compared with standard t-test. Global CBF differences were tested in an ANCOVA model, while controlling for the underlying experimental condition (i.e. baseline vs snow-like stimulus). BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s)

Evaluations of visual snow simulation
A comparison between the 'snow-like' visual simulation used as a task for the MRI sequence and the personal experience of visual snow itself, was collected from each VSS patient at the end of the scanning session. The patients were asked to rate the similarity of the simulation to their own visual static percept by rating the following parameters: density, speed, size and colour. The visual static percept could be classified as equal (=)

Global CBF analysis
Mean global CBF in grey matter values did not differ between groups, whether measured at baseline (mean ± standard error 53.5 ± 10.6 in the VSS group and 51.7 ± 12 in Ctrls; P = 0.6) when looking at the snow-like stimulus (54.9 ± 10.5 in VSS and 53.3 ± 12.8 in Ctrls; P = 0.6) or averaging both conditions (54.2 ± 10.5 in the VSS group and 52.5 ± 12.3 in Ctrls; P = 0.5).
An ANCOVA analysis confirmed no significant differences in median global CBF values between groups, when accounting for stimulus type.

Evaluation of head motion
Cumulative distance travelled was analysed as a measure of head motion, and showed no significant differences between the two groups (0.17 ± 0.12 in the VSS group and 0.16 ± 0.09 in Ctrls; P = 0.8).

F-tests
An F contrast to evaluate the main effect of group on whole-brain CBF maps showed a significant increase in regional CBF in the same nine clusters presented in the main analysis (Table 1), with the following F and k values:

Stimulus effects
When subject to the snow-like stimulus and compared to baseline (i.e. resting fixation), both groups showed an increase in rCBF in a large cluster involving the primary visual cortex, lingual gyrus and inferior temporal gyrus. In VSS patients only, the stimulation also evoked a decrease in rCBF in the mid-cingulate and posterior cingulate cortex. eFigure 1 shows these areas of increased and decreased rCBF in patients and controls, when subject to the snow-like stimulus.

eFigure 1: Stimulus effects in VSS patients and Ctrls
Areas of increases (red/yellow) and decreases (blue/green) of rCBF in VSS patients (a) and Ctrls (b) when subject to the snow-like stimulus. All areas are significant at the cluster level whole-brain analyses and corrected for cluster extent. Bars represent T-values. For deactivation cluster in a) k = 551; MNI coordinates: x = 12 y = -10 z = 46.
BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s)