Sleep EEG-based brain-age index correlates with dementia

Reuters Health - 08/10/2020 - A brain-age index (BAI) based on sleep electroencephalography (EEG) shows promise as a biomarker for dementia, new findings show.

Using machine learning to analyze more than 5,000 polysomnograms, Dr. Alice D. Lam of Massachusetts General Hospital in Boston and her colleagues found BAI - a measure of brain age in relation to chronological age - correlated significantly with dementia and measures of global cognition.

"Many people undergo polysomnography for clinical purposes, such as to evaluate for sleep apnea or other sleep disturbances. The BAI is a method that extracts additional information from a polysomnogram, which could potentially serve as a preliminary screen for individuals who might be at risk for developing dementia," Dr. Lam told Reuters Health by email.

She and her colleagues retrospectively analyzed polysomnograms from 5,144 patients 50 and older, who they categorized into five groups: healthy (2,336 patients based on 2,799 polysomnogram studies); nondementia (3,024 patients, 3,632 studies); symptomatic (1,075 patients, 1,361 studies); mild cognitive impairment (MCI, 44 patients, 55 studies); and dementia (81 patients, 96 studies).

BAI was 0.20 in the nondementia group, 0.58 in the symptomatic patients, 1.65 in patients with MCI and 4.18 in patients with dementia (P<0.001), the researchers report in JAMA Network Open. Overall, men's mean BAI was about three years higher than women's.

Patients with dementia were more likely to have psychotic disorders, mood disorders and anxiety disorders than patients without dementia. BAI also correlated negatively with Montreal Cognitive Assessment and Mini-Mental State Examination scores.

Out of 480 total BAI features, 54 were significantly correlated with dementia and 213 with nondementia. Features indicating delta activity during stage-2 and stage-3 sleep were more likely to be negatively correlated with dementia.

BAI features linked to delta and theta activity while patients were awake or in stage-1 sleep tended to be positively correlated with dementia. The non-dementia patients had higher alpha oscillation features in all stages of sleep.

"While our study found populational differences in sleep EEG-based BAI between groups (no dementia; symptomatic; mild cognitive impairment; and dementia), further work is needed to validate this biomarker at the level of individuals before it can be used in the clinical setting," Dr. Lam said. "We are continuing to refine and validate these BAI measures, as well as to develop new sleep EEG-based measures that could be useful biomarkers in screening for neurodegenerative diseases and dementia."

She concluded: "Polysomnograms contain important physiologic information, such as the BAI, that could be extracted to identify deviations from healthy brain aging, including the presence of neurodegenerative disease."

By Anne Harding

SOURCE: https://bit.ly/34eA8rk JAMA Network Open, online September 28, 2020.



Posted on 8 October 202022 October 2020