Longitudinal genetic and environmental relationships between structural‐ and diffusion‐based Alzheimer’s disease neuroimaging signatures

McKenna E. Williams,Nathan A. Gillespie,Tyler Reed Bell,Anders M. Dale,Jeremy A. Elman,Lisa T. Eyler,Christine Fennema‐Notestine,Carol E Franz,Donald J. Hagler,Michael J. Lyons,Linda K. McEvoy,Michael C. Neale,Matthew S. Panizzon,Chandra A. Reynolds,Mark E. Sanderson‐Cimino,William S. Kremen

DOI: https://doi.org/10.1002/alz.065864

2022-12-01

Abstract:Composite scores of MRI‐based brain regions associated with Alzheimer’s disease (AD) pathology, commonly termed ‘AD signatures,' are tools developed to identify brain changes specific to mild AD. In recent work from our group, we found that a novel diffusion‐based cortical mean diffusivity (MD) signature among cognitively normal adults in their 50s aided prediction of 12‐year progression to MCI beyond age and polygenic risk for AD. Results held up after accounting for general brain age. In contrast, a cortical thickness/hippocampal volume‐based signature did not improve prediction. Given the importance of genetic influences on brain structure and AD, here we examined the phenotypic and genetic relationships among diffusion and structural AD signatures and brain age, and how those relationships change across midlife and early old age. Our validated thickness/volume signature, novel MD signature, and a validated brain age measure were used in biometrical twin analyses to examine phenotypic, genetic, and environmental relationships both cross‐sectionally and longitudinally across 12 years. Participants were 736 men from three waves of the Vietnam Era Twin Study of Aging (VETSA; baseline age=56.1, SD=2.6, range=51.1‐60.2). Subsequent waves were at approximately 5.7‐year intervals. Both AD signatures and brain age are heritable (56%‐72%), but each signature captured unique phenotypic and genetic variance that is also not explained by general brain aging. Genetic correlations across time within each signature were high (0.77‐0.98). Baseline MD signatures (age=56 years) predicted thickness/volume signatures over a decade later (r= ‐0.53, 95% CI: ‐0.62, ‐0.42), but baseline thickness/volume signatures showed a significantly weaker relationship with future MD signatures (r= ‐0.13, 95% CI: ‐0.24, ‐0.01). Interestingly, by wave 2 (average age=62 years), the thickness/volume signature appears to have caught up with the MD signature in terms of predictive ability. Findings lend further support to the idea that a signature based on brain microstructure captures very early AD‐related changes that are later reflected in the macrostructural signature. This MD signature explains substantial genetic variance that is consistent throughout midlife and into early old age, supporting the utility of the MD signature as a very early AD‐related neuroimaging biomarker and its potential role in clinical trials.

clinical neurology