Authors :
Presenting Author: Fuwen Tu, MPH – Yale School of Medicine
Irene Zhou, MPhil – Yale University
Kellie Geisel, BS – Yale University
Grace Grandel, BS Candidate – Yale University
Amiyah Buan, BS – Yale University
Imran Quraishi, MD, PhD – Yale School of Medicine
Rationale:
Memory loss is the most common complaint among people with epilepsy, yet routine neuropsychological testing can miss subtle, hippocampal-dependent deficits. The Mnemonic Similarity Task (MST) indexes two complementary processes—pattern separation (Lure Discrimination Index; LDI) and corrected recognition memory (REC)—that are closely linked to hippocampal function. In particular, LDI has shown sensitivity in several clinical contexts, including aging and refractory temporal lobe epilepsy (TLE). What clinical features may contribute to hippocampal-related vulnerability in pattern separation has not yet been established in a broader population of patients with epilepsy.
Methods:
47 adults with epilepsy (female 53.2%, age median 40 (IQR: 28)) were administered MST in an outpatient setting, excluding those with cognitive/developmental impairments or brain-implanted neurostimulators. Participants first encoded a series of 200 unique object images. They then completed a memory test, where they were shown a series of 100 target images (identical to those from the encoding run), 100 lures (similar but not identical to encoded images) and 100 foils (completely novel images) in random order and were asked to judge whether each image was “old”, “similar”, or “new”, respectively. LDI reflected participants’ ability to distinguish similar lures, while REC reflected participants’ ability to distinguish old targets. Demographics (including age, sex, education) and clinical features (age of onset, duration, intractability, lateralization, localization, frequency, seizure types, status epilepticus history, antiseizure medications (ASMs), MRI lesions) were also collected. These predictors were evaluated for associations with MST outcomes (LDI & REC) using univariate and multivariate linear ridge regression models.
Results:
Based on a multivariate analysis, better mnemonic discrimination (LDI) was associated with earlier age of seizure onset (β = -0.0057, p = 0.025), no history of status epilepticus (β = -0.23, p = 0.042), and better REC (β = 0.099, p = 0.031). The age of onset effect was driven entirely by whether epilepsy began in childhood (12 years or younger); age effects were not apparent between adolescent, early adult, or later adult onset. While having medically-refractory epilepsy was not a significant main effect in this multivariate analysis, it was associated with worse performance on both LDI and REC in separate univariate models. Age, sex, level of education, duration of epilepsy, lateralization, localization by lobe, seizure frequency, seizure types, number of ASMs, and presence of MRI lesion did not affect LDI.
Conclusions:
Our initial findings suggest that pattern separation is sensitive to specific clinical characteristics in adults with epilepsy. While history of status epilepticus was a poor prognostic, childhood onset of seizures was a positive predictor even after controlling for other factors, including localization and severity. These results may reflect the capacity for cognitive reorganization following an early-life insult.
Funding: N/A