Functional and structural brain network correlates of visual hallucinations in Lewy body dementia

Abstract Visual hallucinations are a common feature of Lewy body dementia. Previous studies have shown that visual hallucinations are highly specific in differentiating Lewy body dementia from Alzheimer’s disease dementia and Alzheimer–Lewy body mixed pathology cases. Computational models propose that impairment of visual and attentional networks is aetiologically key to the manifestation of visual hallucinations symptomatology. However, there is still a lack of experimental evidence on functional and structural brain network abnormalities associated with visual hallucinations in Lewy body dementia. We used EEG source localization and network based statistics to assess differential topographical patterns in Lewy body dementia between 25 participants with visual hallucinations and 17 participants without hallucinations. Diffusion tensor imaging was used to assess structural connectivity between thalamus, basal forebrain and cortical regions belonging to the functionally affected network component in the hallucinating group, as assessed with network based statistics. The number of white matter streamlines within the cortex and between subcortical and cortical regions was compared between hallucinating and not hallucinating groups and correlated with average EEG source connectivity of the affected subnetwork. Moreover, modular organization of the EEG source network was obtained, compared between groups and tested for correlation with structural connectivity. Network analysis showed that compared to non-hallucinating patients, those with hallucinations feature consistent weakened connectivity within the visual ventral network, and between this network and default mode and ventral attentional networks, but not between or within attentional networks. The occipital lobe was the most functionally disconnected region. Structural analysis yielded significantly affected white matter streamlines connecting the cortical regions to the nucleus basalis of Meynert and the thalamus in hallucinating compared to not hallucinating patients. The number of streamlines in the tract between the basal forebrain and the cortex correlated with cortical functional connectivity in non-hallucinating patients, while a correlation emerged for the white matter streamlines connecting the functionally affected cortical regions in the hallucinating group. This study proposes, for the first time, differential functional networks between hallucinating and not hallucinating Lewy body dementia patients, and provides empirical evidence for existing models of visual hallucinations. Specifically, the outcome of the present study shows that the hallucinating condition is associated with functional network segregation in Lewy body dementia and supports the involvement of the cholinergic system as proposed in the current literature.


Validation of the source localization pipeline
To validate the implemented cortical source estimation, we used the EEG data collected for a different study involving a motor task paradigm, where participants were asked to maintain isometric contraction by opposition of thumb and index. 1 We randomly chose four subjects, pre-processed the respective EEG task data, estimated the cortical sources, and generated the power spectrum topographies.
For the source localization pipeline being deemed correct, we expected a prominent power activation across time points within the β-band over sensory-motor areas. [2][3][4][5] This was in fact the case, as shown in Figure S1.
Figure S1 -Topographies of power spectra during a motor task. Topographies were obtained on four randomly chosen healthy control subjects; they all show a prominent activation over the sensorimotor areas.

Age as a nuisance covariate in the Network Based Statistics
To address the potential concern on the possible effect of age on the statistics, we also performed the NBS by including age as nuisance covariate and tested the difference in EEG network patterns between groups. As expected, we found a network component which largely resembled the one we reported in the main text (14 edges, 15 nodes, P = 0.033, Figure SII), confirming the robustness of our results. Figure SI1 -Results of the NBS with inclusion of age as a nuisance covariate. The obtained connectome resembles the results reported in the main text.

Validation of MMSE as cognition-related score
A potential limitation of our analysis lies in the choice of the MMSE as a cognition-related behavioural score. This score reportedly shows low sensitivity in characterizing the cognitive phenotype of patient groups, and may poorly represent executive function. To address any potential concern, we additionally tested whether any significant difference in the CAMCOG subdomain scores between groups exists. We first compared the CAMCOG total score, and no significant result emerged (P = 0.6442). By including the CAMCOG total score as a nuisance covariate in the NBS, we obtained a differential subnetwork which resembled the one we reported in the main text (P = 0.027, 17 edges, 18 nodes). By testing the individual subdomains, we found a significant difference only for the orientation score (P = 0.033), which was higher for the NVH group. With the inclusion of this score as a covariate in the NBS we still found a resemblant significant component, although with a weaker effect and only by choosing a more permissive primary threshold (tth = 13, P = 0.046).
Although not significantly different between groups, we also performed the NBS by including the CAMCOG visuo-perceptual score, as we believed this variable to be pertinent for the purpose of our

Potential effect of cholinergic medication on the functional connectivity analysis
To address the potential concern on the effect of cholinergic medication on the functional connectivity analysis, we additionally performed the NBS analysis also including the medication state as a nuisance covariate. Although we obtained a significant network component, this shows a reduced significance and number of connections compared to the one we reported in the main analysis (P = 0.045, 11 edges, 12 nodes, Figure SIV). This outcome might be due to the low statistical power associated with the imbalance between participants' medication states, as majority of participants were taking cholinesterase inhibitors. Future studies with larger cohorts should better address the effect of this type of medication on functional connectivity alterations associated with LBD-VH as detected with EEG.

DLB and PDD groups
In our work we grouped DLB and PDD participants together and divided them in two groups based on the existence of the visual hallucination feature. For exploratory purpose, we investigated whether any difference between DLB and PDD groups existed in terms of functional connectivity. By performing the NBS analysis, we did not find any significant differential network component. This result is in agreement with our previous publication where we investigated EEG-network properties of different types of dementia in the sensor domain. 6 Demographics of the two diagnostic groups are reported in Table SI.

EEG network differences: theta and beta frequency bands
We focused our analyses on the EEG alpha frequency band, since it is reportedly the most linked to lower level visual processing, whilst other frequency ranges such as beta are likely associated with higher level attentional processes. 7 In addition, abnormal alteration of EEG alpha rhythm features has been reported to be associated with Lewy body dementia and to show high specificity in discriminating this syndrome against other forms of dementia and the healthy condition. 8,9 For completeness, we used the Network Based Statistics (NBS) to test whether any significant differential network component emerged within the other frequency bands; specifically we tested theta (4.6-7 Hz) and beta (15-20 Hz), and no significant components were detected (P > 0.05). We also did not find any significant difference in the modularity measure.

Alternative measures of structural connectivity
In our work we chose the number of white matter streamlines as a measure of structural connectivity strength between brain regions, as we hypothesized it to be associated with the degree of pathological

EEG network coordinates
For visualization purpose, node coordinates were obtained as mass centroids of the ICBM152 headmodel 11 vertices within each of the 148 ROIs. 12 Coordinates in the MNI space are reported in Table   SII.