Electrophysiologically-defined excitation-inhibition autism neurosubtypes

Excitation-inhibition (E:I) imbalance has long been considered one of the primary neurobiological theories explaining autism. However, the theory has been bolstered heavily by research on high-penetrance genetic mechanisms which are only found in a small proportion of the autism population and can manifest with potentially different directionality. How well does E:I imbalance explain idiopathic autistic males and does such E:I imbalance manifest in one particular direction? Answering this question in human patients necessitates a need for validated tools that allow for inference about E:I balance from non-invasively measured in-vivo electrophysiological (EEG, LFP) time-series data. Predictions from in-silico modelling alongside in-vivo validations from mouse electrophysiological data show that alterations of synaptic E:I balance cause changes in fractal long memory characteristics of the neural time-series that are captured with a metric known as the Hurst exponent (H). H estimated in resting state EEG data of idiopathic autistic males allows us to discover two distinct E:I neurosubtypes that generalize at high levels (>92% accuracy) in independent data. Each autism neurosubtype captures approximately half of the sample and each can be described by opposing patterns of E:I imbalance relative to a typically-developing (TD) comparison group. Autism E:I neurosubtypes also show differential relationships between H and behavioral and demographic variables consisting of age, intelligence, and autism symptomatology. This work establishes causal evidence for interpreting changes in EEG-derived H as an E:I-relevant biomarker and suggests that idiopathic autistic males can be characterized by opposing types of E:I imbalance with differential phenotypic relevance.