The human connectome in Alzheimer disease - relationship to biomarkers and genetics

Abstract

The pathology of Alzheimer disease (AD) damages structural and functional brain networks, resulting in cognitive impairment. The results of recent connectomics studies have now linked changes in structural and functional network organization in AD to the patterns of amyloid-β and tau accumulation and spread, providing insights into the neurobiological mechanisms of the disease. In addition, the detection of gene-related connectome changes might aid in the early diagnosis of AD and facilitate the development of personalized therapeutic strategies that are effective at earlier stages of the disease spectrum. In this article, we review studies of the associations between connectome changes and amyloid-β and tau pathologies as well as molecular genetics in different subtypes and stages of AD. We also highlight the utility of connectome-derived computational models for replicating empirical findings and for tracking and predicting the progression of biomarker-indicated AD pathophysiology.

Fig. 1 |. Network models.

a | An example network model consists of three communities (dotted circles) linked by three connector hubs (red). The three interconnected hubs constitute a rich club. The three blue lines show the shortest path between the two blue nodes. b | An example of rich-club regions and connections forming a group-averaged human structural connectome. c | An example of a group-averaged human functional connectome consists of four communities, including the default-mode network, frontoparietal network, cingulo-opercular network and visual network. Part b is reprinted with permission from REF., The Journal of Neuroscience. Part c is adapted with permission from REF., Elsevier.

Fig. 2 |. Patterns of tau accumulation in AD.

Tau accumulation patterns in 5 young healthy adults, 17 older adults negative for amyloid-β (Aβ), 16 older adults positive for Aβ and 15 individuals with Alzheimer disease (AD). Tau accumulation was measured by ¹⁸F-AV-1451 PET imaging; Aβ status was defined based on Pittsburgh compound B PET signal. No tau accumulation was observed in the cortical areas of the young healthy adults. The older adults who were negative for Aβ had tau accumulation in some medial temporal lobe regions, such as the entorhinal cortex and parahippocampal gyrus, and in some inferior and lateral temporal regions. The older adults positive for Aβ had tau accumulation in the same regions as individuals who were negative for Aβ but to a greater degree. In addition, in the individuals who were positive for Aβ, tau accumulation was observed in the parietal regions such as the precuneus. In the individuals with AD, tau accumulation in the temporal and parietal regions was significantly higher than in the other groups of older adults; tau accumulation was also observed in broader parietal regions and in frontal and visual cortices in individuals with AD. SUVR, standardized uptake value ratios. Reprinted with permission from REF., Elsevier.

Fig. 3 |. Early Aβ accumulation in resting-state functional brain networks.

Brain regions of early amyloid-β (Aβ) accumulation (red) and the regions comprising six key resting-state functional brain networks (blue) are superimposed upon a 3D rendering of the brain surface. At the preclinical stage of Alzheimer disease, Aβ mainly accumulates in the regions of the default-mode network and, to a lesser degree, the frontoparietal network, the ventral and dorsal attention networks, and the frontotemporal network. The sensorimotor network and the visual network showed the least Aβ accumulation at the preclinical stage. Reprinted from REF., CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).

Fig. 4 |. Association between functional connectivity and covariance in tau-PeT change.

a | Associations between group-averaged functional connectivity and tau-PET covariance at the whole-brain level (top panel) and the level of seven resting-state networks (lower panels) in individuals positive for amyloid-β (Aβ) from two independent samples (53 participants from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) datasets and 41 participants from the BioFINDER datasets). The functional connectivity analysis used a resting-state connectome template derived from the functional MRI data of 500 participants from the Human Connectome Project. Functional connectivity was significantly associated with tau accumulation pattern, both at the whole-brain level and at the level of each of the resting-state networks. DAN, dorsal attention network; DMN, default-mode network; FPN, frontoparietal control network; Limbic, limbic system; SMN, sensorimotor network; VAN, ventral attention network; VIS, visual network. Reprinted from REF., CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).

Fig. 5 |. Hypothesized, empirical and predicted tau spreading patterns.

a | Hypothetical patterns of tau pathology, according to Braak staging^,. Tau initially deposits in the entorhinal cortex at Stage I, spreads to the hippocampus and parahippocampus at Stage II, spreads to the inferior and middle temporal and the cingulate (for example, the posterior cingulate cortex) cortices at Stage III, and finally extends to the frontal, visual and parietal (for example, the precuneus) regions at Stage IV. b | Patterns of tau pathology at different stages of AD from empirical tau-PET data. c | Patterns of tau pathology predicted by an epidemic spreading model (ESM) built on a structural connectome derived from diffusion tensor imaging data. d | Patterns of tau pathology predicted by an ESM built on a functional connectome derived from resting-state functional MRI data. The ESM-predicted tau spreading patterns simulated by the structural connectome (part c) are more similar to the hypothetical (part a) and empirical (part b) tau spreading patterns than those simulated by the functional connectome (part d). Note that warmer colours in the empirical and predicted tau spreading patterns represent higher levels of tau accumulation. Reprinted from REF., CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).

Fig. 6 |. Gene expression, AD pathology and the human connectomes.

In the brain, evidence indicates that regional expression levels of CLU and APP modulate amyloid-β (Aβ) accumulation, whereas the regional expression levels of MAPT and LRP1 (REF.) regulate tau accumulation and spread. In addition, regional APOE expression plays a central role in regulating both the Aβ and tau accumulation patterns. Human connectomes consist of five subnetworks: the medial temporal lobe network, the salience network, the visual network, the default-mode network and the sensorimotor network. These networks are interconnected via hubs: anterior cingulated cortex (ACC), entorhinal cortex (EN), hippocampus (HIPP), middle occipital gyrus (MOG), medial orbitofrontal cortex (MOF), posterior cingulate cortex (PCC), precuneus (PCUN) and supplementary motor area (SMA). The characteristics of the human connectome in Alzheimer disease (AD) are a result of the, as yet unclear, interactions between Aβ and tau pathologies as well as APOE expression. The gene expression mappings were created using publicly available microarray-based regional gene expression data from the Allen Human Brain Atlas as described in REF.. Warmer colours represent higher gene expression levels; the same colour in different gene expression maps might denote different gene expression values as the expression values were not normalized across the different genes. The maps of the Aβ and tau accumulation patterns were manually created according PET data from REF. and REF., respectively. Warmer colours in the brain maps of Aβ and tau represent earlier and higher levels of Aβ and tau accumulation.