Dynamic, state-dependent characteristics of cognitive fluctuations in Lewy body dementia: a magnetoencephalography...

Cognitive fluctuations are a hallmark clinical feature of Lewy body dementia (LBD), yet their underlying neural mechanisms remain poorly understood. This study aimed to identify dynamic, state-dependent neural signatures of cognitive fluctuations in LBD using magnetoencephalography and dynamic functional connectivity based on hidden Markov modelling. Resting-state magnetoencephalography data were acquired from individuals with LBD, Parkinson's disease without dementia and cognitively normal controls. Hidden Markov modelling was used to identify transient brain states followed by spectral analyses across regions and states. Additionally, associations between regional spectral power and cognitive fluctuations severity, measured by the Clinician Assessment of Fluctuation, were assessed. Patients with LBD exhibited a distinct pattern of brain dynamics, particularly in two states (States 2 and 6), characterized by increased fractional occupancy of State 2 and markedly reduced occupancy of State 6, contrasting with the more distributed state engagement observed in Parkinson's disease and normal controls. Spectral analyses revealed widespread slowing in LBD, with elevated theta/beta power ratios in frontal, parietal and visual cortices-most pronounced in States 2 and 6. Region-specific theta/beta power ratio elevations were identified in the anterior cingulate, medial prefrontal cortex, posterior cingulate, dorsal visual stream and auditory cortex. Critically, Clinician Assessment of Fluctuation scores correlated positively with spectral power in low frequency (δ and θ) and negatively with power in the high frequency (α and β), particularly in the ventral visual stream, default mode network hubs and sensorimotor regions. These findings reveal dynamic and spatially specific electrophysiological abnormalities in LBD closely linked to cognitive fluctuations severity, suggesting that magnetoencephalography-hidden-Markov-model characteristics hold promise as biomarkers for diagnosis, monitoring and therapeutic targeting in LBD.