Biomarkers

Abstract

Background: Alzheimer's disease (AD) is associated with cerebral slowing on electroencephalography (EEG), or a shift in oscillatory power from higher to lower frequencies. This change can be captured as the spectral power ratio (SPR) of alpha and beta over delta and theta: (α+β)/(δ+θ). Prior studies have shown that compared to cognitively unimpaired older adults, the SPR was lower in persons with a clinical diagnosis of dementia due to probable AD and those with a biomarker-confirmed diagnosis of amnestic mild cognitive impairment (aMCI) due to AD, and a lower SPR was associated with worse executive functions scores. Moreover, the SPR was unaffected in individuals who had amyloid-negative aMCI, suggesting cerebral slowing may result from AD-associated pathology. We sought to test this hypothesis in a cognitively unimpaired population using plasma Aβ42/Aβ40 levels as a surrogate biomarker of brain amyloid.

Methods: Eyes-closed resting-state EEG, neuropsychological testing, and plasma Aβ42/Aβ40 levels were obtained from 34 adults (18 women, aged 65-89 years), as part of an ongoing study. EEG data were cleaned and analyzed using custom scripts built around EEGLAB and TESA toolboxes in Matlab. Aβ42/Aβ40 <0.1 was used as a cutoff to indicate a high risk of brain amyloid (Rissman et al., 2023, Alzheimer's and Dementia, DOI: 10.1002/alz.13542). An executive functions composite score was created by averaging age-normed z-scores from the digit symbol substitution, digit span backwards, semantic fluency, verbal fluency, and trail making tests.

Results: A Shapiro-Wilk test indicated SPR values were not normally distributed, so data were log10-transformed. 18 participants (53%) were classified as "high-risk of brain amyloid." An independent-samples t-test showed the SPR_log10 (mean ±SD) was lower in this group (0.23 ±0.2) than those with "low-risk of brain amyloid" (0.39 ±0.2), which was significant, t(32) = -2.6, p = 0.015. Pearson's correlation coefficients showed that executive functions were significantly correlated with SPR_log10 in the high-risk group, R(16) = 0.57, p = 0.013, but not in the low-risk group, R(14) = 0.24, p = 0.370.

Conclusions: Cerebral slowing is an early feature of amyloid-associated pathophysiology. The EEG SPR (α+β)/(δ+θ) is a sensitive and cognitively relevant marker of preclinical AD.