Alzheimer’s disease (AD) is a neurodegenerative disorder with a dual proteinopathy as key markers of its pathology: accumulation of amyloid-beta in the form of plaques and hyperphosphorylated-tau in the form of neurofibrillary tangles (NFTs). These changes in the brain lead to irreversible loss of neurons and an eventual decline in cognitive and functional abilities as clinical symptoms of the disease manifest (McKhann et al., 2011, Jack et al., 2018).

AD has a complex temporal evolution and the precise sequence of the spread of neurodegeneration across brain regions, especially in the initial stages, remains unclear. Widely accepted models suggest that Alzheimer’s degeneration starts in the entorhinal cortices and then spreads through the temporoparietal cortex (Heiko Braak, Braak, and Bohl 1993). This is supported by a hierarchical system that stages AD according to tau aggregation (H. Braak and Braak 1991), where the accumulation of NFT first appears in entorhinal cortices and hippocampus. However, this theory has been challenged by histological (Marsel Mesulam et al. 2004; Sassin et al. 2000; Geula and Mesulam 1996; Schliebs and Arendt 2011) and structural imaging studies (Grothe et al., 2012, Grothe et al., 2013, Xia et al., 2023), focusing on the early pathological changes to cholinergic neurons of the basal forebrain.

The main source of cholinergic projections to the cerebral cortex is the magnocellular neurons of the Nucleus basalis of Meynert (NbM), the largest cluster of cholinergic cells that constitute the basal forebrain. Postmortem studies have shown high densities of NFTs in NbM in early and presymptomatic stages of the disease (Marsel Mesulam et al. 2004). The NbM is located within a continuous band of structures including the amygdala, hippocampus, and entorhinal cortex which are all at high risk of degeneration due to AD pathology. This anatomical positioning of NbM has been speculated to be the main reason for its vulnerability to NFTs (Marsel Mesulam 2013).

Longitudinal studies focusing on basal forebrain and specifically NbM have also reported disproportionate degeneration when compared to normally aging older adults (Grothe et al., 2013, Alzheimer’s Disease Neuroimaging Initiative, 2018). Schmitz and Spreng (Schmitz et al., 2016) showed that early longitudinal grey matter loss in the region of the NbM surpassed that in the entorhinal cortex in amyloid-positive cognitively normal older adults. They also discovered that baseline NbM volumes predicted rates of structural degeneration in the entorhinal cortex. In another work the degeneration of the cholinergic projection system was further theorized to be upstream of entorhinal and neocortical degeneration (Fernández-Cabello et al. 2020).

However, the precise delineation of NbM is difficult due to limited spatial resolution and contrast in MR images. The NbM lacks strict boundaries with adjacent cell groups and its small size poses a challenge for defining this region on common 1mm3 isotropic T1w scans. To overcome this challenge, we increased the resolution of our MRI scans, before performing a deformation-based morphometry (DBM) analysis. DBM, unlike voxel-based morphometry (VBM), does not depend on an automated segmentation of the MR data into gray matter, white matter, and CSF; instead it can use image contrast directly as an explicit representation of these distributions (Ashburner and Friston, 2000, Chung et al., 2001). The improvements in nonlinear image registration algorithms allow for matching the images locally based on similarities in contrast and intensities, making DBM more sensitive than VBM for subtle differences and more resilient to erroneous tissue classification.

In this study, we precisely quantify volume loss in these structures (NbM, EC, HC) across Alzheimer’s disease trajectory. We look at cognitively normal controls (CN), including those that are amyloid-negative (CN-) and those that are amyloid-positive (CN+), as well as amyloid-positive early mild cognitive impairment (eMCI), amyloid-positive late mild cognitive impairment (lMCI) and amyloid-positive patients with dementia due to clinically probable AD. In addition to examining these groups cross sectionally, we also map all subjects into a common disease timeline to compare longitudinal differences between their continuous volume trajectories. We also verify these results by comparing these trajectories in only the CN+ and eMCI groups to ensure the later disease groups are not driving the early findings.