P1‐355: Patterns of Regional Gray Matter Atrophy Vary Across Cognitively‐defined Alzheimer's Disease Subgroups in Adni
Year: 2019
Type: article
Abstract: We sought to determine whether cognitively-defined subgroups of people with Alzheimer's disease (AD) had different patterns of gray matter density (GMD). We used bifactor models on cognitive data from the first study visit with AD to generate memory, executive functioning, language, and visuospatial functioning scores. We determined average cross domain scores and differences between each domain score and the cross-domain average. We used published thresholds to form subgroups: AD-Memory, AD-Executive, AD-Language, and AD-Visuospatial, AD-No domains, and AD-Multiple domains. We varied thresholds in sensitivity analyses. We analyzed MRI scans from people ≥65 years old, and similarly aged cognitively normal controls. We processed MRI scans using DARTEL implemented in SPM12. We converted each voxel's GMD findings into w-scores covaried for age, sex, intracranial volume, and field strength. We considered average w-scores from 21 regions of interest (ROI). For each ROI, we compared average w-scores for each group controlling for average w-scores outside the ROI. We also performed voxel-wide analyses in SPM12 with the final gray matter images, covaried for age, sex, education, MRI field strength, and mean global w-score. Of 543 people with AD, 177 had AD-Memory, 42 had AD-Language, 61 had AD-Visuospatial, and 229 had AD-No Domains. We omitted 12 with AD-Executive and 22 with AD-Multiple domains from analyses. Medial temporal lobe ROI omnibus p was 9.7x10-6, with AD-Memory < AD-Language by 0.15, p = 0.07; AD-Memory < AD-Visuospatial by 0.28, p = 1.4x10-4, and AD-Memory< AD-No domains by 0.23, p=4.7x10-6. The visuospatial network had an omnibus p = 0.0037, AD-Memory > AD-Visuospatial by 0.14, p=3.8x10-4. Other findings included left-sided language network (AD-Language<AD-Memory), executive control network (AD-No Domains<AD-Memory), and precunious (AD-Visuospatial<AD-Memory). In voxel-wide analyses, each group had profound differences with controls in medial temporal lobes. Differences between groups in GMD were apparent in brain regions that corresponded to each phenotypic group. Secondary thresholds impacted group size with larger numbers in AD-Executive and AD-Multiple, but regression findings were similar. Voxel-wide results with the secondary thresholds are shown in the Figure.
Source: Alzheimer s & Dementia
Language: English
Subfield: Psychiatry and Mental health
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W2980796196
