In late-onset unexplained epilepsy, higher p-tau217 went with worse cognition and disrupted sleep

Design

Prospective cross-sectional cohort study

Population

85 adults with late-onset unexplained epilepsy (onset ≥55, non-lesional MRI), mean age 71

Primary outcome

Association of plasma p-tau217 with cognition (PACC5) and sleep microarchitecture

Effect

β −0.66 (95% CI −1.19 to −0.13) per log unit p-tau217; ~24% mediated by spindle–slow-oscillation coupling

This prospective study recruited 85 people (mean age 71) with late-onset unexplained epilepsy: new unprovoked seizures after age 55 with no cortical lesion on MRI. All had cognitive testing, plasma p-tau217 (a marker of Alzheimer pathology) and 24-hour EEG with sleep scoring.

Higher p-tau217 was associated with poorer cognition on the PACC5 composite (β −0.66 per log unit, 95% CI −1.19 to −0.13), and the effect was stronger in those with drug-refractory epilepsy (interaction β −1.49). Higher p-tau217 was also associated with weaker coupling of sleep spindles and slow oscillations, which accounted for about 24% of the link with cognition.

Late-onset epilepsy is increasingly recognised as an early sign of neurodegeneration in some patients. This small study supports checking cognition and considering Alzheimer biomarkers in these patients, and suggests seizure control may matter for cognition.

• In new epilepsy after 55 with a normal MRI, assess cognition at baseline and at follow-up.
• A raised plasma p-tau217 in such a patient suggests underlying Alzheimer pathology.
• Refractory seizures were linked with worse cognition; aim for good seizure control.
• Ask about sleep; disrupted sleep architecture may be part of the pathway.

Why it matters

It links a common presentation in older adults to a treatable-sounding mechanism and a blood test that is becoming available.

Don't overread it

This is a small cross-sectional association study; it does not show that treating seizures or sleep protects cognition.

The statistics, in plain English

The mediation estimate (24%) has a very wide 95% CI of 3% to 85%, so how much sleep explains the link is uncertain. With 85 participants, all associations are imprecise and cannot show cause.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.