A resting-state functional connectivity study of visual hallucination awareness in Lewy body disease

Introduction Visual awareness encompasses the ability to distinguish internally generated sensory experience (thoughts, memories, imagination, dreams) from veridical percepts derived from the external world. Visual hallucinations (VH) are an example of internally generated experience with important clinical implications and represent a core clinical feature in Lewy body disease (LBD), with some patients aware of their hallucinations and others not. This difference in awareness provides an ideal opportunity to investigate the neural mechanisms underlying hallucination awareness (VH-awareness). To date, several theoretical approaches have been used to explain the generation of VH and have been recently synthesized within a unified framework.1 The framework incorporates three fundamental processing/conceptual levels, with either hallucinations or veridical percepts arising from their interaction: I. the visual processing of bottom-up sensory data; II. the perceptual, emotional, and memory context of the experience, referred to as reality monitoring; III. the processes by which priors and expectancies are compared with sensory data from which veridical percepts or hallucinations are derived (prediction error [PE] mechanisms). Although the framework does not explicitly address the awareness dimension of VH, it seems reasonable to hypothesize that the different processes may also be involved in different aspects of awareness. Therefore, we hypothesized that VH-awareness would be associated with functional brain alterations in terms of resting-state functional connectivity (rs-FC) within and between cerebral systems involved in VH (intra- and inter-network rs-FC). In particular, we focused on visual association areas and networks engaged in PE and RM, comparing first LBD patients with healthy controls (HC), and subsequently comparing LBD patients with preserved awareness of VH to those with impaired awareness of VH. Materials and methods Participants. We conducted a retrospective analysis of resting-state functional magnetic resonance imaging (rs-fMRI) data from DLB and PDD participants. The final sample consisted of 32 LBD patients (17 DLB, 15 PDD) with a history of complex VH and 20 HC. Phenomenology and awareness of VH were assessed using the North-East Visual Hallucinations Interview (NEVHI). The awareness question probed online VH-awareness and was as follows: “Whilst you were experiencing hallucinations, how often have you been aware that you were experiencing a hallucination?”. Possible answers were: “never” (0), “sometimes” (1), “always” (2). Patients who responded “always” were categorized as having intact awareness (VH-Aware group), while those responding “never” or “sometimes” were categorized as having impaired awareness (VH-Unaware group). Image acquisition, processing, and analyses. All subjects underwent an MRI exam using a 3T Siemens Magnetom Verio scanner. We acquired 3D high‐resolution T1‐weighted images (170 slices; in‐plane resolution = 192×256 mm; slice thickness = 1.2 mm; time repetition [TR] = 9 s; time echo [TE] = 4 ms) and resting-state T2*-weighted images using a gradient-echo imaging sequence to measure the BOLD contrast (200 scans; 50 slices; in-plane resolution = 3×3 mm; slice thickness = 3 mm; TR = 3 s; TE = 31 ms). Resting-state images were preprocessed following a standard pipeline implemented in the Functional Connectivity Toolbox (CONN 22.a) running on SPM12, performing: removal of the first 4 scans in each run, functional data realignment and unwarp, slice-timing correction, outlier identification, co-registration to individual anatomical images, structural segmentation of gray and white matter and cerebrospinal fluid, normalization to the MNI152 space, and functional smoothing using a spatial convolution with a Gaussian kernel of 8-mm FWHM. A seed-to-seed functional connectivity (FC) analysis was implemented to test intrinsic functional coupling within and between couples of the cortical networks expected to be involved in VH awareness. Three bilateral networks and their constituent regions of interest (ROIs) were defined based on previous studies. The visual association network (VIS network) included dorsal and ventral visual stream areas: the lateral occipital cortex (superior and inferior), lingual gyrus, temporo-occipital inferior temporal gyrus, temporal fusiform cortex (anterior and posterior), temporal occipital and occipital fusiform cortex. The PE network involved the insula, pars triangularis and opercularis of the inferior frontal gyrus, angular gyrus, and posterior supramarginal gyrus. The RM network consisted of the frontal medial cortex, middle frontal cortex, anterior cingulate gyrus, and hippocampus. Due to the exploratory nature of the current work and a proposed right hemispheric lateralization for symptom insight/awareness, we analyzed FC in each hemisphere separately. Average intra- and inter-network FC values were compared using the “conn_withinbetweenROItest” function implemented in CONN, which averages the BOLD signal across voxels in all ROIs within each network. This resulted in Fisher’s z-transformed bivariate Pearson correlation coefficients. Intra-network FC represents the average degree of FC among ROIs within a single network, while inter-network FC refers to the average FC between the ROIs of two distinct networks at a time. First, we compared HC and the whole LBD group using GLMs to assess differences in intra-network and pairwise inter-network FC (VIS–PE, VIS–RM, RM–PE), including age and education as covariates. Then, a Mann-Whitney U test was employed to test FC differences between VH-Aware and VH-Unaware LBD patients. The significance threshold was set at p < 0.05 corrected for multiple comparisons using the FDR method, with an additional Bonferroni correction of 3 to correct for the number of networks included (p = 0.017). RESTING-STATE FUNCTIONAL NETWORK CONNECTIVITY RESULTS During the acquisition of rs-fMRI data, no LBD patient reported experiencing VH. Therefore, the present findings pertain to the hallucination trait rather than to the hallucinating state. Functional connectivity in HC and LBD patients Intra-network FC within the VIS and RM networks did not differ between HC and LBD in either hemisphere, while reduced FC was found in LBD patients relative to HC in the right PE network (pFDR = 0.034, despite not surviving additional Bonferroni correction). Regarding inter-network connectivity, LBD patients showed lower FC between VIS and RM networks in both hemispheres compared to HC (left hemisphere: pFDR < 0.001; right hemisphere: pFDR = 0.004). Functional connectivity in VH-Aware and VH-Unaware LBD patients When comparing LBD patients with and without awareness of VH, the intra-network FC analysis did not show significant group differences for any network of interest in both hemispheres. The inter-network analysis revealed differences in the FC values between VIS and PE networks in the right hemisphere only (pFDR = 0.013), with reduced connectivity in VH-Unaware patients compared to VH-Aware patients. No group differences were found in average connectivity between VIS and RM networks and between PE and RM networks in either hemisphere. DISCUSSION The present study aimed to investigate awareness of VH, which we defined here as the capacity to discern whether a visual percept is a false product of our brain or is veridical, reflecting the external world. We hypothesized that awareness may depend on the interaction of three key components, and we investigated it by focusing on VH in the LBD spectrum. The association between alterations in VH awareness was tested studying rs-FC within the recently proposed consensus framework for VH (Collerton et al., 2023). Compared with HC, LBD patients showed lower bilateral rs-FC between VIS and RM networks, suggesting that the neural mechanisms supporting the evaluation and contextualization of visual inputs might be compromised in LBD. As the functional disruption of visual association areas is a well-recognized hallmark of DLB, the observed VIS–RM disconnection may be a consequence of this visual system vulnerability. One might speculate that these functional alterations represent a predisposing condition for the typical impairments in visuo-perceptual abilities as well as VH in LBD. The alterations may reflect a weakening of higher-order monitoring systems, which, in turn, may reduce the accuracy with which visual perceptual experiences are retrospectively attributed to internal or external sources. Additionally, we might speculate that the observed trend for LBD patients to show reduced FC within the PE network may signal an early inefficiency in dynamically updating visual representations when expectations and sensory evidence diverge. If progressive, such alterations could ultimately contribute to the transition from preserved to impaired awareness of hallucinatory experiences. Furthermore, relative to VH-Aware LBD patients, VH-Unaware patients exhibited reduced rs-FC between VIS and PE networks in the right hemisphere (Fig. 1). This suggests that VH unawareness in LBD might be supported by an alteration in implicit, online awareness processes and, specifically, may depend on the impaired functional coupling between the brain regions providing perceptual contents to the VH experience and those evaluating its congruency with environmental circumstances and internal predictions. Taken together, the results suggest that online awareness during the VH experience could be conceptualized as a higher-order cognitive process that requires the dynamic integration of domain-general (PE) and domain-specific (VIS) systems, depending on the nature of the perceptual object. In other words, the mere intact functioning of individual networks may not suffice for awareness to emerge. Although preliminary in their nature, our findings provide a foundation for further investigation into awareness of VH in LBD. Addressing these aspects of VH could have significant implications for their management in clinical practice, while, from a neuroscience perspective, future research should explore awareness with its varying levels and nuances across diverse populations in order to unveil one of the most distinctive aspects of brain function.