A greater difference between magnetic resonance imaging–estimated brain age and chronologic age may be associated with progression to dementia among patients with subjective cognitive decline or mild cognitive impairment.
Investigators conducted a retrospective cohort study of 799 patients enrolled in the Amsterdam Dementia Cohort and SCIENCe project between 2002 and 2024. Among the patients, 412 had subjective cognitive decline and 387 with mild cognitive impairment.
Brain age was estimated from baseline magnetic resonance imaging (MRI) using BrainageR and cNeuro. Brain-predicted age difference (brain-PAD) represented estimated brain age minus chronologic age. Standard MRI assessment included visual ratings of medial temporal, parietal, and global cortical atrophy and white matter hyperintensities. The investigators used Cox proportional hazards models adjusted sequentially for age, sex, Mini Mental State Examination score, visual MRI ratings, and amyloid status.
During a median follow-up was 3.2 years, 29% (n = 235) of the patients progressed to dementia, including 39 with subjective cognitive decline and 196 with mild cognitive impairment. In the BrainageR analysis, each additional year of brain-PAD was associated with a 6% higher likelihood of progression after adjustment and a 4% higher likelihood after adding visual MRI ratings.
A significant interaction between brain-PAD and baseline diagnosis supported separate analyses of patients with subjective cognitive decline and mild cognitive impairment. Among those with subjective cognitive decline, each additional year of brain-PAD was associated with a 9% higher hazard of progression after adjustment. Adding brain-PAD increased the C-index from 0.77 to 0.79 in that model. When amyloid status was added in patients with subjective cognitive decline, brain-PAD improved model fit, but the C-index remained 0.84. In patients with mild cognitive impairment, BrainageR-derived brain-PAD was no longer predictive when visual MRI ratings were incorporated, whereas cNeuro-derived brain-PAD remained associated with progression when visual MRI ratings and amyloid status were included.
“Our results show that brain-PAD captures variance not explained by widely used visual rating scales,” wrote lead study author Stefan De Vries, of the Department of Radiology and Nuclear Medicine at Vrije Universiteit Amsterdam at Amsterdam UMC, and colleagues.
The investigators cautioned that scanner-specific differences affected brain-PAD estimates and that the relatively young tertiary memory-clinic cohort may limit generalizability. The cohort largely included patients who progressed to Alzheimer's disease dementia, and visual rating scales do not capture the full MRI assessment performed by a trained radiologist.
Full disclosures of the study authors can be found in the study.
Source: Neurology
