Hypoexcitation and connectome disintegration via whole-brain semi-empirical modeling in neurodegeneration

Abstract

Abstract Background Neuroimaging biomarkers are intensively investigated in Alzheimer disease (AD) and the behavioral variant frontotemporal dementia (bvFTD). However, advanced ND biomarkers (i.e., PET and plasma) are expensive or not widely available/validated in underrepresented regions. EEG emerges as a promising alternative, due to its low-cost, non-invasiveness, portability, and wide availability in clinical research, although they lack mechanistic explications and robust modeling approaches. The aim of our work was to characterize the functional alterations in AD and bvFTD by combining EEG source-level metaconnectivity (that captures high-order interactions), anatomical priors, whole-brain modeling, and a perturbational approach. Method Demographic-matched data were collected via the Dementia Latin American Consortium (ReDLat; AD = 31, bvFTD = 18, Control = 46). From EEG source level recordings and using machine learning classifiers, we compared the pairwise functional connectivity with high-order correlations captured by metaconnectivity. Then, neural mass modeling was used to propose biophysical mechanisms specifically ascribed to AD and bvFTD. Finally, a perturbational approach was employed to characterize the brain trajectories involved in the healthy and pathological states. Results Using metaconnectivity, we found specific subnetworks of brain regions compromised in AD and bvFTD, yielding an excellent accuracy in classifying each group of patients. Brain dynamics turned more “viscous” (uncoordinated) in patients and were associated with multimodal disease progression (years of disease, cognitive impairment, atrophy). Then a whole-brain Jansen Rit model simulated the functional disturbances in AD and bvFTD. The DTI connectome disintegration (reduced structural integration) and hypoexcitability (altered excitation/inhibition balance) triggered the metaconnectivity dynamics observed in AD and bvFTD. Finally, in-silico stimulation identified key regions involved in the transition from disease (AD and bvFTD models) to a healthy brain dynamics. Conclusion Our work suggests new possible EEG markers for characterizing AD and bvFTD, linking the functional disturbances in patients to alterations in the connectome and E/I balance, and evidencing possible anatomical targets to restore the normal brain function.