Identification and validation of a gray matter volume network in Alzheimer's disease

This study aims to identify and validate a gray matter volume network in patients with Alzheimer's disease (AD).

To identify a disease-related network, a principal component analysis-based algorithm, Scaled Subprofile Model, was applied to gray matter volume data derived from structural T1-weighted magnetic resonance imaging of the training sample that consisted of nine patients with AD (women, four; dementia, seven; mild cognitive impairment, two; age, 66.7 ± 8.8 [mean ± SD] years) with positive ¹⁸F-flutemetamol amyloid positron emission tomography and eight age-matched healthy controls obtained on-site. The network expression scores were calculated by topographic profile rating in the validation sample obtained via the Open Access Series of Imaging Studies and comprised 12 patients with AD dementia (women, four; age, 70.0 ± 3.7 years) and 12 age-matched healthy controls.

A significant network from the training sample, for which subject expression differed between the groups (permutation test, P = 0.006; sensitivity and specificity, 100%; area under the curve, 1), was identified. This network was represented by the principal components 1, 2, and 3 and showed a relative decrease in the inferior parietal lobule including angular gyrus, inferior temporal gyrus, premotor cortex, amygdala, hippocampus, and precuneus. It significantly differed between the groups with a sensitivity, specificity, and area under the curve of 83%, 91%, and 0.85, respectively, in the validation sample (P = 0.003).

An AD-related gray matter volume network that captured relevant regions was identified in amyloid positron emission tomography-positive patients and validated in an independent sample.