Single-subject gray matter networks predict future cortical atrophy in preclinical Alzheimer's disease

Highlights

• Study if gray matter network measures can predict future atrophy in preclinical AD.

• Network measures, but not other AD biomarkers, predict hippocampal atrophy rates.

• Network measures in early-Aβ accumulating regions predict the rate and location of atrophy.

• Gray matter networks detect AD-related pathological changes before overt atrophy.

Abstract

The development of preventive strategies in early-stage Alzheimer's disease (AD) requires measures that can predict future brain atrophy. Gray matter network measures are related to amyloid burden in cognitively normal older individuals and predict clinical progression in preclinical AD. Here, we show that within individuals with preclinical AD, gray matter network measures predict hippocampal atrophy rates, whereas other AD biomarkers (total gray matter volume, cerebrospinal fluid total tau, and Mini-Mental State Examination) do not. Furthermore, in brain areas where amyloid is known to start aggregating (i.e. anterior cingulate and precuneus), disrupted network measures predict faster atrophy in other distant areas, mostly involving temporal regions, which are associated with AD. When repeating analyses in age-matched, cognitively unimpaired individuals without amyloid or tau pathology, we did not find any associations between network measures and hippocampal atrophy, suggesting that the associations are specific for preclinical AD. Our findings suggest that disrupted gray matter networks may indicate a treatment opportunity in preclinical AD individuals but before the onset of irreversible atrophy and cognitive impairment.