Age-related neurocognitive disorders such as Alzheimer’s disease (AD) and mild cognitive impairment (MCI) involve widespread grey matter (GM) atrophy, as well as loss of white matter (WM, estimated with fractional anisotropy) in network structures (Sexton et al., 2011, Sui et al., 2015, Talwar et al., 2021, Vogt et al., 2020; L. Wang et al., 2009; Yao et al., 2012). Intrinsic brain connectivity, measure with resting-state fMRI were observed to be disrupted (reduced functional connectivity(FC)) across several brain networks (Brier et al., 2014, Talwar et al., 2021, Xiong et al., 2023). Structural and functional degenerative changes in the brain are also associated with cognitive decline (Avants et al., 2010, Dadar et al., 2022, Mallio et al., 2015, Misquitta et al., 2020, Raghavan et al., 2020, Sui et al., 2015). Given multimodal approaches better are able to discriminate AD and MCI from normal aging or other disorders (Mesrob et al., 2012, Sheelakumari et al., 2018, Vogt et al., 2020; L. Wang et al., 2009; Xiong et al., 2023), studying the interaction between GM structures and their network pathways is particularly relevant to understanding neurodegeneration.

Several multimodal neuroimaging studies in aging-related neurocognitive disorders have observed associations between GM structures and WM connections (Avants et al., 2010, Cauda et al., 2018, Raj and Powell, 2018, Sui et al., 2015, Yu et al., 2020). Diffusion tensor imaging (DTI) WM integrity (fractional anisotropy), and WM hyperintensities, have been correlated with estimates of GM atrophy in AD and MCI (Avants et al., 2010, Dadar et al., 2022, Jang et al., 2017, Misquitta et al., 2020, Sui et al., 2015, Sydykova et al., 2007; L. Wang et al., 2009). Studies have also shown that topological network properties of the WM tracts predicted global GM atrophy over time (Nir et al., 2015, Raj and Powell, 2018). Longitudinal studies similarly showed that baseline WM network structure predicted GM volume reductions over time in MCI and AD (Nir et al., 2015, Raj and Powell, 2018). In reference to a normative healthy network template, regions with the highest amount of WM connections were also those with higher annualized GM volume reduction in AD (Phillips et al., 2019). This interaction between WM networks and GM structures suggests that network disruption could play a role in GM neurodegeneration.

Explanatory degeneration models were developed to explain how toxic agents including amyloid beta (Aβ) and tau proteins may propagate through fibre pathways (Pandya et al., 2017, Raj and Powell, 2018, Zhou et al., 2012). Studies using resting-state FC have also linked progress in tau/Aβ deposit to network dysfunction in Alzheimer's disease, which may involve disordered transmission of information (Brier et al., 2014) and nodal stress, characterised by hyperactivation in a region which may intensify metabolic activity in neurons, making them vulnerable to degeneration (De Haan et al., 2012, Fornito et al., 2015, Zhou et al., 2012). In a study on several neurodegenerative and psychiatric disorders, (e.g., AD, Schizophrenia and Bipolar disorder), patterns of “co-alterations” (associated structural changes) in pairs of regional GM volumes were shown, and could be predicted by the brain network’s structural connectivity (SC) and FC (Cauda et al., 2018). Longitudinal analyses in AD also report that baseline reduced average strength in the default mode network resting-state FC predicted cortical thinning overtime (Hampton et al., 2020). This suggests both SC and FC contribute to atrophy in distant regions.

Coalterations and pathological propagation models imply that areas most affected by GM decreases in MCI and AD, in particular the medial temporal lobes, with the bilateral hippocampi, entorhinal and parahippocampal gyri (Mallio et al., 2015, Manuello et al., 2018; W.-Y. Wang et al., 2015), could lead to distant alterations through their connections. Decreased left temporal WM and decreased hippocampal volume help classify AD and MCI (Mesrob et al., 2012), which gives the hippocampus a key importance in neurodegeneration. GM volume coalterations were established in AD, highlighting that along with the parahippocampal cortices, the hippocampi, especially the left one, had a distant associated decreases (Cauda et al., 2020, Manuello et al., 2018). In AD, early reduced hippocampal grey matter predicted later medial temporal GM atrophy in AD (mutual, but stronger from the hippocampus) (Li et al., 2011); and propagation of GM atrophy over time was positively correlated with average FC from “epicentres” to the other regions, including the left angular gyrus and the right frontal gyrus, and the left hippocampus (Mutlu et al., 2017). In AD and MCI, regions with the most WM fibre streamlines connecting them to the hippocampi and the entorhinal cortices, or the topologically closest areas in the brain network, were also the regions with the most decreases in SC (Mallio et al., 2015). Specifically, the hippocampus is correlated to decline in verbal memory over time, which predicted decreased WM in the fornix (lower fractional anisotropy (FA) and mean diffusivity (MD)) (Yu et al., 2020). Therefore, the hippocampus may be of direct importance for the progress of the disease.

Considering the above literature, the present study focused on the hippocampal SC and FC to distant areas. We identified regions of GM CT alteration associated with memory impairment, and measured SC and FC from these regions to the left and right hippocampi. Then we fitted these measurements to a structural equation model to test whether SC and FC to the hippocampi predicted GM thickness in the selected regions. We hypothesized that hippocampal connectivity was associated with distant changes in the cortical thickness of regions connected to the hippocampi.