Fragility and volatility of structural hubs in the human connectome

Abstract

Brain structure reflects the influence of evolutionary processes that pit the costs of its anatomical wiring against the computational advantages conferred by its complexity. We show that cost-neutral 'mutations' of the human connectome almost inevitably degrade its complexity and disconnect high-strength connections to prefrontal network hubs. Conversely, restoring the peripheral location and strong connectivity of empirically observed hubs confers a wiring cost that the brain appears to minimize. Progressive cost-neutral randomization yields daughter networks that differ substantially from one another and results in a topologically unstable phenomenon consistent with a phase transition in complex systems. The fragility of hubs to disconnection shows a significant association with the acceleration of gray matter loss in schizophrenia. Together with effects on wiring cost, we suggest that fragile prefrontal hub connections and topological volatility act as evolutionary influences on brain networks whose optimal set point may be perturbed in neuropsychiatric disorders.

Spatial patterning of tissue volume loss in schizophrenia reflects brain network architecture

“…The second null model preserves the spatial embedding (i.e., geometry) of the structural connectome [9,31,54]. Edges were first binned according to Euclidean distance.…”

“…We next seek to assess the extent to which these associations depend on network topology. To address this question, we constructed two null models: (1) a geometric null model that randomly rewires pairs of edges in the structural network, preserving the degree sequence and edge length distribution [9,31,54]; (2) a spatial autocorrelation null model that projects nodes to a sphere and randomly rotates the sphere [1] (see Methods and Materials). Fig.…”

“…The associations remain only when considering structurally connected neighbours. (E,F) Two null structural models were constructed: (1) a spatial autocorrelation null model that projects nodes to a sphere and randomly rotates the sphere [1] (i.e., spin tests; 10,000 repetitions; blue); (2) a geometric null model that randomly rewires pairs of edges in the structural network, preserving the degree sequence and edge length distribution [9,31,54] (1,000 repetitions; pink). A null functional model was also constructed by randomly reassigning resting-state functional MRI time series to each node (1,000 repetitions; grey).…”

“…The relationship between a node's and its spatially non-adjacent neighbours' deformation values | Node deformation was correlated with the mean deformation of spatially non-adjacent neighbours (i.e., non-abutting neighbours) across five parcellation resolutions, such that the spatially adjacent neighbours of a given node were excluded from the calculation of its mean neighbour deformation. The statistical significance of the correlations was assessed against the structural and functional null models: (1) a spatial autocorrelation null model that projects nodes to a sphere and randomly rotates the sphere [1] (i.e., spin tests; 10,000 repetitions; blue); (2) a geometric null model that randomly rewires pairs of edges in the structural network, preserving the degree sequence and edge length distribution [9,31,54] (1,000 repetitions; pink); (3) A null functional model was also constructed by randomly reassigning resting-state functional MRI time series to each node (1,000 repetitions; grey). Asterisks, colored according to the corresponding null model, indicate significant empirical correlations (P < 0.05; two-tailed).…”

Efficient Coding in the Economics of Human Brain Connectomics

“…In the economics of brain connectomics, natural selection optimizes network architecture for versatility, resilience, and efficiency under constraints of metabolism, materials, space, and time [2,1,3]. Networks -composed of nodes representing cortical regions and edges representing white matter tracts -strike evolutionary compromises between costs and adaptations [2,1,4,5,6,7], whereby disruptions may contribute to the development of neuropsychiatric disorders [8,9,10,11]. To understand how the brain efficiently balances resource constraints with pressures of information processing, models of information diffusion in brain networks are necessary.…”