The early-onset Alzheimer's disease MRI signature: a replication and extension analysis in early-stage AD

Abstract

Early-onset Alzheimer's disease (EOAD) is less investigated than the more common late-onset Alzheimer's disease (LOAD) despite its more aggressive course. A cortical signature of EOAD was recently proposed and may facilitate EOAD investigation. Here, we aimed to validate this proposed MRI biomarker of EOAD neurodegeneration in an Appalachian clinical cohort. We also compared differences in EOAD signature atrophy in participants with biomarker-positive EOAD, LOAD, early-onset non-AD pathologies, and cognitively normal individuals. Cortical thinning was reliably detected in eight of nine signature areas of persons with EOAD relative to cognitively normal individuals despite very early disease stage. Additionally, individuals with EOAD showed thinner cortex in most signature regions relative to those with early-onset non-AD pathologies. EOAD and LOAD showed similar cortical atrophy within most EOAD signature regions. Whole-brain vertex-wise cortical analyses supported these findings. Furthermore, signature cortical atrophy showed expected relationships with measures of global and specific cognitive and functional status. This investigation further validates and expands upon the recently defined EOAD signature and suggests its robustness within a rural population, even at early disease stage. Larger scale and longitudinal studies employing this marker of EOAD neurodegeneration are needed to further understand clinical effects and appropriate management of persons with EOAD.

Keywords: MRI; amyloid; cortical thickness; early-onset Alzheimer’s disease; neuroimaging.

Fig. 1

Region of interest (ROI) analyses of cortical thickness in EOAD relative to CN participants. A) ROIs from the Desikan Killiany atlas comprising the EOAD signature: SFG, superior frontal gyrus; MFG, middle frontal gyrus; MTG, middle temporal gyrus; ITG, inferior temporal gyrus; IPG, inferior parietal gyrus; SPL, superior parietal lobule; PCG, posterior cingulate gyrus; PRECU, precuneus; FUSG, fusiform gyrus; and AD control region PCAL, pericalcarine gyrus. B) Partial eta-squared (η_p²) effect sizes for bihemispheric ROI differences in cortical thickness with color-scale separating non-significant η_p² < 0.03 in purple.

Fig. 2

Vertex-wise comparison of cortical thickness in EOAD versus CN participants in the left and right hemispheres separately. Results displayed at the post hoc vertex-z-threshold of P < 0.01 and Monte Carlo z-field simulation cluster threshold of P < 0.05.

Fig. 3

EOAD vs. LOAD. Results of vertex-wise comparison of cortical thickness in EOAD and LOAD participants in the left and right hemispheres separately. Results displayed at the post hoc vertex-z-threshold of P < 0.01 and Monte Carlo z-field simulation cluster threshold of P < 0.05.

Fig. 4

EOAD vs. EOnonAD. Results of vertex-wise comparison of cortical thickness in EOAD and EOnonAD participants in the left and right hemispheres separately. Results displayed at the post hoc vertex-z-threshold of P < 0.01 and Monte Carlo z-field simulation cluster threshold of P < 0.05.

Fig. 5

Correlation between thickness of the EOAD signature composite and cognitive and functional tests.