Late-Onset Temporal Lobe Epilepsy Has Distinct Cognitive, CSF Profile

Key Takeaways

• In a small Italian cohort, biomarker-normal late-onset temporal lobe epilepsy was associated with cognitive deficits despite preserved structural magnetic resonance imaging (MRI) morphometry.
• Compared with mild cognitive impairment due to Alzheimer disease, late-onset temporal lobe epilepsy showed milder and less diffuse cognitive impairment across global and several domain measures.
• Cerebrospinal fluid in late-onset temporal lobe epilepsy showed lower neurofilament light chain and a lower phosphorylated-to-total tau ratio than mild cognitive impairment due to Alzheimer disease, and all LO-TLE ratios were below the previously proposed 0.17 cutoff used by the authors.
• Most late-onset temporal lobe epilepsy participants met mild cognitive impairment criteria, but the phenotype was more heterogeneous and attention remained preserved.

New-onset temporal lobe epilepsy after age 50 can create a difficult differential when seizures emerge alongside forgetfulness and other cognitive complaints. In an Italian cohort analysis, investigators at a single Italian center studied 18 patients with late-onset temporal lobe epilepsy (LO-TLE), 24 with mild cognitive impairment due to Alzheimer disease (MCI-AD), and 17 age-matched and sex-matched healthy controls. LO-TLE required epilepsy onset after age 50, International League Against Epilepsy (ILAE)-defined temporal lobe epilepsy confirmed by expert epileptologists, unilateral or bilateral temporal epileptiform activity on long-term video-electroencephalography (EEG), structural magnetic resonance imaging (MRI), and normal cerebrospinal fluid (CSF) Alzheimer biomarkers.

Participants underwent structural MRI morphometry and neuropsychological testing across episodic memory, short-term memory, language, attention, and executive function; CSF assays for amyloid-beta (Aβ), tau measures, neurofilament light chain (NfL), and the calculated phosphorylated-to-total tau (p/t-tau) ratio were available for the LO-TLE and MCI-AD groups only.

Cortical thickness and subcortical volumes in LO-TLE were comparable with healthy controls, whereas MCI-AD showed bilateral temporal and precuneus atrophy with smaller hippocampal and amygdala volumes. Despite that preserved morphometry, LO-TLE performed worse than healthy controls in episodic memory, t(53) = -7.79, pFDR < 0.001, and better than MCI-AD in global cognition, t(53) = 4.21, pFDR < 0.001. Additional deficits versus healthy controls involved short-term memory, language, executive function, and global cognition, while attention was preserved; compared with MCI-AD, LO-TLE performed better in episodic memory, short-term memory, attention, and executive function, with no significant language difference.

Most LO-TLE participants met mild cognitive impairment criteria, but the phenotype was more heterogeneous and pathologic performances clustered mainly in episodic memory measures, whereas MCI-AD showed broader deficits across domains. In CSF, LO-TLE showed lower NfL and a lower p/t-tau ratio than MCI-AD, together with lower tau markers and higher Aβ measures; for the p/t-tau ratio, F(1,40) = 23.97, pFDR < 0.001, and all LO-TLE values were below 0.17. Within LO-TLE, a higher p/t-tau ratio was associated with better global cognition, rs = 0.585, pFDR = 0.032.

This single-center cohort was intentionally restricted to LO-TLE with normal CSF Alzheimer biomarkers, so the findings do not define the full late-onset temporal lobe epilepsy spectrum. The sample was small and cross-sectional, behavioral and psychiatric features were not assessed, left-sided cases predominated, and conventional morphometry may have missed microstructural abnormalities that could help explain cognitive deficits despite normal structural MRI.

Clinician Questions

How was biomarker-normal late-onset temporal lobe epilepsy defined in this cohort?

Investigators defined LO-TLE as epilepsy onset after age 50 with ILAE-defined temporal lobe epilepsy confirmed by expert epileptologists, unilateral or bilateral temporal epileptiform activity on long-term video-EEG, structural MRI availability, and laboratory-defined normal CSF Alzheimer biomarkers: Aβ1-42 greater than 600 pg/mL, Aβ1-42/1-40 ratio greater than 0.069, total tau less than 400 pg/mL, and phosphorylated tau 181 less than 56.5 pg/mL. That definition captured a selected biomarker-normal LO-TLE subgroup rather than all late-onset epilepsy.

What cognitive pattern separated LO-TLE from MCI-AD beyond global test scores?

LO-TLE showed a more circumscribed profile, with pathologic performance concentrated mainly in episodic memory measures and attention preserved relative to healthy controls, whereas MCI-AD showed broader episodic memory, attentional, executive, and visuospatial impairment. LO-TLE mild cognitive impairment phenotypes were also more heterogeneous than the predominantly amnestic multidomain pattern seen in MCI-AD.

Why could patients with LO-TLE have cognitive deficits despite preserved structural MRI?

Cortical thickness and subcortical volumes in LO-TLE were comparable with healthy controls, yet measurable cognitive deficits were still observed in this biomarker-normal epilepsy group. The authors said conventional morphometry may miss microstructural white matter or other non-morphometric abnormalities, so normal structural MRI did not exclude cognitive involvement in this cohort.

What remains unresolved about late-onset temporal lobe epilepsy after this analysis?

The authors identified several open questions: the cohort was small, cross-sectional, and restricted to patients with normal CSF Alzheimer biomarkers; right-sided and bilateral cases were underrepresented; behavioral and psychiatric features were not assessed; and longitudinal multimodal studies are needed to determine the underlying pathology and whether this profile reflects an alternative neurodegenerative pathway or a relatively stable condition.