Metabolic complexity in the human brain supports metastable network dynamics and declines in normative ageing

The adaptive capacity of the brain is tied to the complexity of its neural activity. While entropy of brain signals has been used to characterise this complexity in haemodynamic and electrophysiological systems, it remains unknown whether entropy contributes to the temporal dynamics of the metabolic signals of the brain (glucodynamics) and how metabolic entropy may relate to network organisation, brain ageing and cognitive decline. Here, we combined dynamic functional PET (fPET) with 18F-fluoroedeoxyglucose and simultaneously acquired functional magnetic resonance imaging (fMRI) in 84 adults to investigate multiscale entropy (MSE) in the brain. We report that metabolic complexity measured with PET MSE represents non-linear, high-frequency complexity, whereas hemodynamic complexity reflects non-linearly constrained neurovascular phase dynamics. Greater metabolic complexity was associated with greater network metastability, particularly in unimodal and higher-order associative networks. Greater regional metabolic complexity also correlated with greater cortical thickness, cerebral metabolic rates of glucose (CMRglc) and cerebral blood flow, and was lower in older than younger adults. In contrast, global and network haemodynamic MSE was not significant associated with network dynamics. Higher global metabolic complexity was associated with better cognitive performance, and voxel-level derived MSE clusters were \associated with cognitive performance in both modalities, with varied cluster sizes and spatial locations. However, these associations were largely accounted for by age variance. We conclude that temporal complexity in cerebral glucose is associated with the dynamics of the underlying metabolic signals, enables metastable network states and declines in ageing. Entropy of metabolic brain signals offers promise as a neurobiologically grounded metric of brain ageing.