Page 87 - MEGIN Book Of Abstracts - 2023

magnetoencephalography (MEG) at rest, we studied Generating diagnostic profiles of cognitive decline
30 Alzheimer's disease patients without subclinical epi- and dementia using magnetoencephalography
leptiform activity, 20 Alzheimer's disease patients with (2022)
subclinical epileptiform activity and 35 age-matched
controls. Presence of subclinical epileptiform activ- Scheijbeler, Elliz P; Schoonhoven, Deborah N; Engels,
ity was assessed in patients with Alzheimer's disease Marjolein M A; Scheltens, Philip; Stam, Cornelis J; Gouw,
by long-term video-EEG and a 1-h resting MEG with Alida A; Hillebrand, Arjan
simultaneous EEG. Using the resting-state source-space
reconstructed MEG signal, in patients and controls we Alzheimer Center Amsterdam, Department of Neurology,
computed the global imaginary coherence in alpha Amsterdam Neuroscience, Vrije Universiteit Amsterdam,
(8-12 Hz) and delta-theta (2-8 Hz) oscillatory frequen- Amsterdam UMC, Amsterdam, The Netherlands; Department
cies. We found that Alzheimer's disease patients with of Clinical Neurophysiology and MEG Center, Department
subclinical epileptiform activity have greater reductions of Neurology, Amsterdam Neuroscience, Vrij Universiteit
in alpha imaginary coherence and greater enhance- Amsterdam, Amsterdam UMC, Amsterdam, The Netherlands.
ments in delta-theta imaginary coherence than Al- Electronic address: [email protected]
zheimer's disease patients without subclinical epilep-
tiform activity, and that these changes can distinguish ABSTRACT Accurate identification of the underlying
between Alzheimer's disease patients with subclinical cause(s) of cognitive decline and dementia is challeng-
epileptiform activity and Alzheimer's disease patients ing due to significant symptomatic overlap between
without subclinical epileptiform activity with high subtypes. This study presents a multi-class classification
accuracy. Finally, a principal component regression framework for subjects with subjective cognitive de-
analysis showed that the variance of frequency-specific cline, mild cognitive impairment, Alzheimer's disease,
neuronal synchrony predicts longitudinal changes in dementia with Lewy bodies, fronto-temporal dementia
Mini-Mental State Examination in patients and controls. and cognitive decline due to psychiatric illness, trained
Our results demonstrate that quantitative neurophysi- on source-localized resting-state magnetoencephalog-
ological measures are sensitive biomarkers of network raphy data. Diagnostic profiles, describing probability
hyperexcitability and can be used to improve diagnosis estimates for each of the 6 diagnoses, were assigned
and to select appropriate patients for the right therapy to individual subjects. A balanced accuracy rate of 41%
in the next-generation clinical trials. The current results and multi-class area under the curve value of 0.75 were
provide an integrative framework for investigating obtained for 6-class classification. Classification primar-
network hyperexcitability and network dysfunction to- ily depended on posterior relative delta, theta and beta
gether with cognitive and clinical correlates in patients power and amplitude-based functional connectivity in
with Alzheimer's disease. the beta and gamma frequency band. Dementia with
Lewy bodies (sensitivity: 100%, precision: 20%) and
Keywords: epileptiform activity in Alzheimer’s disease, Alzheimer's disease subjects (sensitivity: 51%, preci-
imaginary coherence, magnetoencephalography, net- sion: 90%) could be classified most accurately. Fronto-
work hyperexcitability, neuronal synchrony temporal dementia subjects (sensitivity: 11%, precision:
3%) were most frequently misclassified. Magnetoen-
Brain: a journal of neurology (2022), Vol. 145, No. 2 cephalography biomarkers hold promise to increase
(34919638) (8 citations) diagnostic accuracy in a noninvasive manner. Diagnos-
tic profiles could provide an intuitive tool to clinicians
and may facilitate implementation of the classifier in
the memory clinic.

 ontents  Index 66
C