Whole-brain Functional Network Research In Subjects At High-risk Of Alzheimer’s Disease: Based On Graph-theoretical Analysis

BackgroundWith the coming of aging society, dementia in elderly is associated with a close attention because it is becoming a public health problem. Alzheimer’s disease (AD) is the most common type of dementia in elderly. Currently, the AD etiology and pathogenesis is still not fully understood, and none of the treatments are available for AD. Therefore, AD’s early diagnosis and prompt intervention remain to be the critical measure for delaying disease progression. Therefore, it is a hot spot to detect the neurobiological markers in individuals at high-risk of AD in the field of AD research.The apolipoprotein E (APOE) e4 allele is the only well-established susceptibility gene for developing AD and lows the age at onset in a dose-dependent fashion. Biological studies demonstrate that the APOE gene participates in almost all AD pathogenesis pathways. Evidence suggests that the major effect of APOE gene on the risk of developing AD is via its effect on amyloid P (AP) aggregation and clearance. These studies indicate that the APOE-ε4 allele may play an important role in the progression of AD. Moreover, patients with amnestic mild cognitive impairment (aMCI) or late-onset depression have an increased risk of progressing to AD. aMCI represents a transition state between normal ageing and AD, which has a high probability of converting to AD at a rate of 12% per year, while normal ageing is associated with a rate of 1%-2%. Similarly, patients with late-onset depression (LOD) often complicated by varying degrees of cognitive impairment, even when the affective symptoms are relieved. LOD with cognitive impairment has increased risk of conversion to AD, and it might be a preclinical stage of AD.Previous studies have demonstrated that AD is a’disconnection syndrome’. Recently, researchers further proposed the AD network dysfunction hypothesis, which indicated that aberrant network activity may interact with AD pathogenesis and casually contribute to cognitive impairment. Thus, it is critical to understand alterations of neuronal circuits underlying cognitive deficits in individuals at high-risk of AD. Currently, the resting-state functional MRI (R-fMRI) approach is an ideal tool in studying brain functional network. Previous numerous neuroimaging studies have reported the alterations of the brain’s structural and functional connectivity in AD-risk. It is worthy to note that the human brain is a complex network that continuously integrates information across distributed brain regions. In the last ten years, numerous studies have suggested that human structural and functional networks can be constructed using multimodal neuroimaging data and that their topological organization can be characterized quantitatively using various graph-theory metrics. With these network metrics, many nontrivial organizational principles, including small-worldness, modularity, and highly connected hubs, have been observed in the human brain connectome. Importantly, in patients with AD, several research groups have reported topological alterations in the whole-brain connectome, including a loss of small-worldness and a redistribution of hubs. Despite the increasing knowledge of AD, however, limited work has investigated the topological organization of whole-brain functional networks in individuals at high-risk of AD. Therefore, connectome-based studies in AD-risk would open up a new window for understanding the pathophysiological mechanisms of AD and may provide new imaging biomarkers for AD early diagnosis.This study recruited the cognitive normal elders, aMCI patients and LOD patients. For all participants, we assessed their cognitive function using multi-domain neurocognitive tests and constructed their whole-brain functional networks using the R-fMRI data. The present study yielded three main research directions. (1) We sought to determine how the APOE genotype affects the brain’s functional network architecture in cognitively healthy elders. (2) We sought to investigate the influence of aMCI and APOE-s4 on local and global functional integrity in the whole-brain connectome. (3) We aimed to investigate the topological organization of whole-brain functional networks in patients with remitted LOD, and to examine the relationship between topological aberrations and performances on neuropsychological tests. Finally, the study is expected to advance our current understanding of the neurobiological mechanism underlying the cognitive dysfunction in individuals at high-risk of AD, and shed lights on exploring network-based AD biomarker in the further. Part 1 APOE genotype modulates functional brain connectome in cognitively healthy elders Objective:We sought to determine how the APOE genotype modulates the brain’s functional network architecture in cognitively healthy elders. Methods:127 cognitively healthy elders, including 35 APOE ε2 carriers,43 APOE ε3ε3 carriers and 49 APOE s4 carriers, were recruited in this study. All participants underwent multi-domain neurocognitive assessment and R-fMRI scan. The routine imaging preprocessing was carried out using the SPM8 software. Then, to construct the brain functional network, the images of each brain was parcellated into 90 regions of interest (ROIs,45 for each hemisphere) using the automated anatomically labeling atlas. To measure interregional resting-state functional connectivity, Pearson correlation coefficients between any pair of ROIs were calculated, thus generating a 90×90 correlation matrix for each subject. Each absolute correlation matrix was thresholded into a binary matrix with a fixed sparsity level. Finally, for the constructed brain networks at each sparsity threshold, we calculated both global and regional network measures (i.e., small-world, network efficiency, and nodal efficiency). For statistical analysis, first, analyses of covariance were used to detect the among-group difference of topological metrics. Then, linear regression analyses were performed to examine the relationships between the neuropsychological measures and the topological properties in the three genotype groups separately. Results:(ⅰ) No significant difference of the neurocognitive performances was observed among the three genotype groups, (ⅱ) The three genotype groups showed small-world organization of brain functional networks expressed as γ> 1 and λ≈1. Significant group effects in the normalized clustering coefficient (γ) and small-worldness (σ) were observed among the three groups, with theAPOE ε2 carriers showing the highest γ and σ values; and the APOE ε3ε3 carriers networks had values intermediate between APOE ε2 carriers and APOE ε4 carriers, although no statistically significant changes were found. Towards to other global network measures, no significant group effects were found among the three groups. (ⅲ) As to regional network measure, significant among-group differences in nodal efficiency were observed in the right orbital part of superior frontal gyrus, right precuneus and left angular gyrus, with the APOE ε2 carriers showing the highest nodal efficiency in these areas. Moreover, compared with APOE ε3ε3 carriers, the APOE ε4 carriers showed decreased nodal efficiency in the right orbital part of superior frontal gyrus and left angular gyrus. (ⅳ) Multiple linear regression analyses further revealed that the network metrics correlated with the neuropsychological performances in the three groups. (ⅴ) Finally, we examined the network robustness of the three genotype groups in response to the random failures and targeted attack. Although the brain functional networks of the three groups exhibited similar robustness to random failures, there was significant among-group difference in robustness to targeted attacks (ε2+>ε3ε3>ε4+).Conclusion:Our findings of loss of small-world characteristics in APOE ε4 carriers reflect a less optimized topological organization in brain networks, suggesting that APOE ε4 allele might be a risk factor for AD onset. On the contrary, the organization of brain functional networks in APOE ε2 carriers indicated a more optimized pattern, suggesting a lower AD onset risk related to the APOE ε2 allele.Part 2 Mapping the effects of aMCI and APOE polymorphism on intrinsic brain network connectivityObjective:To investigate the influences of aMCI and APOE-s4 allele on intrinsic brain network connectivity.Methods:In the present study, we selected the data of 156 elderly participants, including 66 patients with aMCI and 90 healthy control (HC) subjects. Of these aMCI patients, there were 27 APOE ε4 carriers and 39 APOE ε3ε3 carriers; of these HC subjects, there were 45 APOE ε4 carriers and 45 APOE ε3ε3 carriers. All subjects underwent neuropsychological battery assessment and R-fMRI scan. In this study, we employed two voxel-wise network centrality measures, degree centrality (DC) and eigenvector centrality (EC), to quantify locally and globally functional integrity of the brain connectome. Voxel-wise two-way analyses of covariance were separately performed to examine the main effects of diagnosis (aMCI vs. HC) and APOE genotype (ε4 carriers vs. ε3ε3 carriers), and diagnosis-by-genotype interactions on DC and EC maps, with age, gender and years of education as covariates. Finally, we performed multiple linear regression analyses to examine the relationships between the neuropsychological measures and network centrality values (i.e., DC and EC) in brain areas showing significant diagnosis-by-genotype interactions.Results:(ⅰ) Significant main effect of diagnosis on each cognitive domain was observed, with the aMCI patients showing worse cognitive performances than the healthy controls. There was no significant main effect of APOE genotype on any cognitive measure. A significant interaction between diagnosis and APOE genotype was observed only on visuospatial function, with APOE-ε4 carriers showing worse performance than ε3ε3 carriers in the aMCI group but no genotype difference in the control group, (ⅱ) Significant diagnosis-by-genotype interactions on DC were observed in the left superior/middle frontal gyrus, right middle temporal gyrus and posterior cerebellar lobe, with higher DC values in the APOE-ε4 carriers than APOE-ε3ε3 carriers in the aMCI group. No regions showed significant main effects of either diagnosis or APOE genotype.(ⅲ) We further observed diagnosis-by-genotype interactions on EC, with higher values in the right middle temporal gyrus but lower values in the medial parts of default-mode network in the APOE-ε4 carriers than APOE-ε3ε3 carriers in the aMCI group. Notably, these genotype differences in DC or EC were absent in the control group. In addition, significant main effects of diagnosis on EC were observed in the regions mainly located in the default-mode network, with the aMCI patients showed lower EC values in these regions compared with healthy controls. However, no regions exhibited significant main effects of APOE genotype, (iv) Finally, multiple linear regression analyses revealed that the network connectivity DC values in the left superior/middle frontal gyrus, right middle temporal gyrus and posterior cerebellar lobe negatively correlated with the cognitive performances (i.e., episodic memory and executive function scores) in aMCI ε4-carriers.Conclusion:Global network integrity may be more preferentially affected in patients with aMCI. Further, the APOE ε4 genotype specifically modulates functional integration of brain networks in patients with aMCI, thus providing important insights into the gene-connectome interaction in this disease.Part 3 Altered topological patterns of whole-brain functional networks in rLODObjective:We aimed to investigate the topological organization of whole-brain networks in patients with rLOD combing R-fMRI and graph-theory approaches, and further to examine the relationship between topological aberrations and performances on neuropsychological tests.Methods:A total of 33 patients with rLOD and 31 healthy control subjects underwent neuropsychological battery assessment and R-fMRI scans. To construct the brain functional network, the images of each brain was parcellated into 90 regions of interest (ROIs,45 for each hemisphere) using the automated anatomically labeling atlas. To measure interregional resting-state functional connectivity, Pearson correlation coefficients between any pair of ROIs were calculated, thus generating a 90×90 correlation matrix for each subject. Each absolute correlation matrix was then thresholded into a binary matrix with a fixed sparsity level. Finally, for the constructed brain networks at each sparsity threshold, we calculated both global and regional network measures (e.g., small-world, network efficiency, and nodal efficiency). For statistical analysis, first, nonparametric permutation tests were used for group comparisons of topological metricl’s. To investigate the clinical relevance of altered brain network topologies in the rLOD group, we then conducted multiple linear regression analyses to examine the relationships between the neuropsychological measures and the topological properties.Results:(ⅰ) The rLOD group performed significantly worse than the control group in episodic memory, executive function, processing speed and visuospatial skills, (ⅱ) Both the rLOD and control groups showed small-world architecture in the brain functional networks, suggesting a balance between functional segregation and integration. Importantly, the rLOD patients exhibited abnormal global topology in their functional brain networks (i.e., increased shortest path length and decreased network efficiency) compared with the healthy controls, implying a less optimal topological organization in rLOD. (ⅲ) Moreover, the rLOD patients showed decreased nodal efficiencies, predominately in the frontal-striatal-occipital regions that are closely associated with the neuropathological changes in LOD. (iv) Intriguingly, we showed that the topological aberrations correlated with the neuropsychological performances in the rLOD patients, (ⅴ) Finally, the brain functional networks of rLOD were equally as robust to random failures as those of healthy controls, but more vulnerable against targeted attacks.Conclusion:These results demonstrate that the topological organization of functional brain networks is disrupted in rLOD and that this disruption may contribute to disturbance in cognitive function in rLOD patients.