Genetic predisposition to Alzheimer’s disease is differentially associated with joint volumetric variations in the disease continuum

Abstract Background Imaging genetics studies commonly assess the individual association between genetic information and brain features without considering joint brain modulations. The aim of this study was to explore the structural variation of the brain at different stages of Alzheimer’s disease (AD), depending on the genetic predisposition to AD. Method A total of 351 cognitively unimpaired (CU) middle‐age participants from the ALFA (Alzheimer’s and FAmilies) study and 330 cognitively impaired participants from ADNI (Alzheimer’s Disease Neuroimaging Initiative) with magnetic resonance imaging scans, genetic information and CSF‐amyloid status were included. They were classified as CU Aβ‐, CU Aβ+, (mild cognitive impairment) MCI Aβ+ and AD Aβ+[ Table 1 ]. Freesurfer was used to obtain cortical and subcortical parcellations using the Desikan‐Killiany atlas. A polygenic risk score estimating each participant’s genetic predisposition to AD (PRS‐AD) was calculated including genetic variants at the genome‐wide suggestive level (5×10‐6). Individuals were classified into high/low genetic predisposition groups. Statistical analysis was based on compositional data analysis. First, we computed the optimal multivariate brain structural variation signature (elastic net selection). Then, a logistic regression was performed to determine the association between the optimal signature and the PRS‐AD. Models were stratified by disease stage and adjusted for age and sex. Result Individuals at different stages on the AD continuum displayed different brain volumetric modulations associated with higher genetic predisposition to AD [ Figure 1 ]. For instance, the optimal signature associated with an increased risk of AD was mainly characterized by increased hippocampal volumes in Aβ‐ individuals, but decreased volumes in MCI individuals along with the modulation of other temporal regions. Moreover, in AD patients at higher genetic predisposition of AD, the optimal signature was characterized by increased and decreased volumes in frontal regions (e.g. pars triangularis, lateral orbitofrontal vs. medial orbitofrontal, respectively) [ Figure 2 ]. Conclusion Results showed AD stage‐specific multivariate volumetric variation associated with an increased genetic risk of AD. Main results suggested that regions that modulated together, although differing according to AD stage, were implicated in the structural brain topology of specific memory networks. The analysis of the joint volumetric variation of brain subregions brings an innovative modeling perspective for neurogenetic studies of AD.