Multi-scale Modelling of Neurosteroid-mediated Seizure Trajectories in Childhood Absence Epilepsy

Abstract

Childhood absence epilepsy (CAE) is a pediatric generalized epilepsy disorder characterized by brief episodes of impaired consciousness and distinctive 2.5--5 Hz spike-wave discharges (SWDs) on electroencephalography. With a well-established genetic aetiology, this condition tends to resolve spontaneously during adolescence in most cases. While several mechanisms have been proposed for remission, understanding remains insufficient to guide early intervention practices.

In this thesis, we first utilize a conductance-based thalamocortical network model that exhibits characteristic SWDs, to demonstrate that allopregnanolone---a progesterone metabolite known to enhance GABAa receptor-mediated inhibition---has an ameliorating effect on SWDs. To investigate the divergence between this finding and clinical observations, we developed an enhanced thalamocortical model that incorporates a layered cortical structure to explore regional cortical heterogeneity and frontocortical connectivity as potential resistance factors to ALLO-mediated recovery. Our results suggest that non-resolving CAE may be due not only to increased frontocortical connectivity but also to the composition of cell types within the network. Specifically, a higher proportion of bursting-type cells may prevent the therapeutic effects of allopregnanolone. We extended our investigation to examine whether these findings apply to CAE caused by different genetic mechanisms, particularly mutations in sodium channel genes by modelling their effects at the individual neuron level. Furthermore, we examined the degree to which these alterations lead to network-level pathological activity, as well as the influence of ALLO on these genetically distinct networks. Our results demonstrate that ALLO facilitates recovery from SWDs regardless of the underlying mutation type. However, enhanced frontocortical connectivity prevents recovery in some mutation types, particularly when mutation effects are severe.

Altogether, the multi-scale computational framework developed in this thesis demonstrates that CAE remission is determined by complex interactions between hormonal influences, genetic factors, and network connectivity patterns. The results suggest that certain genetic mutations may predispose individuals toward either remission or non-remission, which can be further modulated by connectivity profiles. In particular, enhanced frontocortical connectivity appears to be a significant factor in resistance to hormone-mediated remission. Additionally, this thesis develops techniques for analyzing transitions between distinct dynamical states in neural systems, incorporates genetic and hormonal factors into conductance-based models, and provides a computational framework to identify key parameters governing epileptic activity. These approaches not only advance our understanding of CAE specifically, but offer generalizable insights into the mathematical modelling of neurological conditions characterized by spontaneous shifts in brain dynamics.