In this study, a modularity-based similarity approach was used to determine intermodule and intramodule comparisons between the two groups at 25% sparsity. The results revealed that FC abnormalities in the ANMDARE group were concentrated in the parieto-occipito-temporal region and the parahippocampal gyrus compared to those of HCs. In addition, abnormal FC between module regions may suggest altered brain function in maintaining cognition in patients with ANMDARE.
In the ANMDARE group, the parieto-occipito-temporal regions, as well as the parahippocampal gyrus, were divided into a module due to similar strength of FC and significantly higher FC in these regions than in the HCs. Combined with cognitive scores that were significantly lower than those of HCs and emotional deficit scores that were significantly higher than those of HCs, this suggests the presence of cognitive impairment and emotional deficits in patients with ANMDARE, which may be related to changes in brain function. In a study of the relationship between changes in modularity and cognitive function in a population of adolescents, it was found that both the increased efficiency of whole-brain functional networks and the maturation of cognitive function were associated with the development of modular topology; that is, stronger intramodule connections and weaker intermodule connections(Baum et al., 2017). Furthermore, close connections within modules are designed to ensure local brain activity, while sparse connections between modules limit the expansion of redundant information(Kaiser et al., 2007), and enhanced close connections within modules may be compensation made to maintain the efficiency of whole brain networks. Studies in individual mice have also confirmed that when the dissemination of functional information in brain regions mediating short-term memory is disrupted or blocked, redundant pathways come into play and inefficiently continue to maintain function in the damaged areas(Chen et al., 2021). This suggests that patients with ANMDARE may have intramodular dysfunction.
The occipital lobe occupies a certain percentage of FC modules. There are multiple U-shaped fiber connections between the occipital lobe and part of the temporoparietal cortex, including the occipital vertical tract, frontal suboccipital tract, and optic radiation, which are involved in object recognition, motion perception, and face recognition(Palejwala et al., 2020). Reduced occipital brain activity was found to be associated with memory and cognitive impairment by comparing changes in brain activity in patients with schizophrenia vs. HCs(Zhang et al., 2018). The occipital lobe plays an important role in visuospatial cognition, and the processing of two-dimensional visual space has been shown to rely on the regulation of connections between the occipital cortex and the posterior parietal lobe, suggesting a possible link between cognitive activity involving the occipital lobe and the parietal lobe(Beffara et al., 2022). When the gray matter volume of the occipital, temporal and frontal lobes is atrophied, it can cause cognitive impairment in Parkinson's patients(Chen et al., 2019); therefore, the abnormal FC of the parieto-occipital temporal lobe in patients with ANMDARE may cause cognitive alterations.
In addition, the most significantly closer connections between the right cuneus and the left precuneus are found in abnormally connected brain regions. The ability to regulate negative emotions is reduced in the presence of functional impairment in the precuneus(Ferri et al., 2016), and damage to this region may be associated with emotional deficits in patients with ANMDARE. It has also been reported in the literature that cortical atrophy in the cuneus and parieto-occipital regions is present in patients with hallucinations in Lewy body dementia(Delli Pizzi et al., 2014), and we speculate that this may be related to the presence of hallucinations as a clinical manifestation in some ANMDARE cases.
In addition, the parahippocampal gyrus, located inferiorly within the occipitotemporal lobe, is an important structure in which the hippocampus functions, with extensive cortical connections that play a role in higher cognitive functions, including memory encoding and retrieval and visuospatial processing(Lin et al., 2021). The results of the present study only revealed abnormal connections between the parahippocampal gyrus and the occipitoparietal lobe, and we speculate that the visuospatial cognitive impairment in ANMDARE may be due to abnormal functional changes between the parahippocampal gyrus and parietooccipital lobe.
This study also found that compared to HCs, the PC module of the right inferior frontal gyrus increased and that of the parietal and basal ganglia regions decreased. PC is a measure that describes the depth to which a single node is embedded in its local module according to the optimal community clusters, which is the ratio of the number of connections a node has within its module to the number of connections within the entire network. An increase in PC can suggest a decrease in the number of intramodule connections or an increase in the number of intermodule connections, while a decrease in PC is an increase in the PC value that can indicate a decrease in the number of connections within a module or an increase in the number of connections between modules, while a decrease in PC value is a quantitative indication of an increase in the number of connections within a module or a decrease in the number of connections between modules. The inferior frontal gyrus (IFG) has white matter connections mainly with the superior frontal gyrus, inferior parietal lobule, lateral occipital area, temporal pole, cuneate gyrus, and lingual gyrus(Briggs et al., 2019) and is involved in important cognitive functional activities, which include executive functions, encoding, long- and short-term memory, attention, and retrieval(Li et al., 2013; Prince et al., 2007). Increased connections between the IFG and other modules or reduced connections within its own modules both suggest abnormal structural separation of modules, which to some extent affects functional brain network efficiency. Combined with enhanced intramodule connections, no abnormalities in whole-brain functional network efficiency were found between the two groups, but we speculate that cognitive impairment cannot be considered lightly; conversely, we hypothesize that patients with ANMDARE would be able to maintain whole-brain network efficiency possibly with intramodule and intermodule functional modulation, but still cannot effectively maintain the original cognitive function(Yang et al., 2018), thus causing a more time-dependent cognitive recovery process(Heine et al., 2021). In contrast, increased intramodular connectivity or decreased intermodular connectivity in the parietal and basal ganglia regions suggests the presence of associated cognitive impairment(Beffara et al., 2022; Riva et al., 2018). Combined with the duration of disease of < 2.3 years (average 0.83 years) in the patients included in this study, the majority of patients still had cognitive impairment(Heine et al., 2021), which explains the changes in brain function with partial certainty, but the assessment scale is more limited in terms of targeted cognitive impairment, which is one of the possible reasons for the lack of association in the correlation analysis.
The comparison of module Q values between the two groups in this study was not statistically significant. We believe that the modularity Q value, which generally indicates the criteria for obtaining nonrandom and clear community clumps, was not significantly different between the two groups, which suggests the existence of comparable module delineation between the two groups. In contrast, the exploration between the two groups of gray matter volumes aimed to regress the linear relationship between gray matter volume changes and functional connectivity, and the results only revealed statistically significant results for the precuneus, which was not representative of gray matter volume changes in this brain region due to the voxel size of 1. This suggests that there were no significant differences in volume between the subjects involved and that there was no linear correlation with changes in FC. Unexpectedly, however, there were no changes in brain network topology in the ANMDARE group, and in combination with the analysis of previous studies that did not find global topological changes in this disease population, we speculate that brain network topology in patients with this disease still retains the more plastic and efficient small-world properties(Li et al., 2021) and could benefit more from early cognitive recovery in the recovery period after timely treatment(Heine et al., 2021). However, as no longitudinal studies have confirmed this, further studies are still needed to repeat the validation.
The shortcomings of this study still need to be addressed in future studies. First, the years of education were relatively low in the patient group, and years of education, age and gender were regressed as covariates in this study. Second, the sample size is insufficient, and the accuracy and sensitivity of the results are limited accordingly. Third, this study was conducted only on a cross-sectional basis and lacks the modular baseline status of this patient group. Finally, the cognitive recovery of patients with ANMDARE is time dependent, and this study lacks further analysis of the disease duration subgroup.