EEG markers of excitatory- inhibitory signalling dynamics in autistic and non-autistic people

Student thesis: Doctoral ThesisDoctor of Philosophy


Coordinated excitatory- inhibitory (E-I) signalling is fundamental to efficient brain function. Alterations in E-I circuitry are hypothesised in multiple neurodevelopmental conditions which may explain symptom overlap between conditions. Within conditions however, the nature and extent of individual E-I differences are heterogeneous. Untangling this complexity is challenging because researchers are limited in the methods available that can capture E-I information in living humans. For example, the methods available such as Positron Emission Tomography (PET) and Proton Magnetic Resonance Spectroscopy [1H] MRS are expensive and/or invasive and cannot capture E-I signalling at fast timescales due to their poor temporal resolution.

In contrast, electroencephalography (EEG) is a cheap, non-invasive method with a high temporal resolution that could offer a hugely valuable tool to study E-I. In Ahmad*, Ellis* et al. (2022), I identified several EEG metrics with potential as proxy markers of E-I signalling in humans. However, I also noted a gap between preclinical and clinical neuroscience. The underlying E-I neurobiology of candidate markers has largely been informed by animal studies without verifying that EEG recorded from the scalp in humans is sampling the same E-I processes. Therefore, the aim of this PhD was to address that gap by examining whether E-I signalling is indeed captured in candidate EEG markers in humans namely; beta power and the aperiodic 1/f exponent.

Furthermore, I identified some key obstacles to investigating the E-I mechanisms underpinning candidate markers (beta power and the aperiodic 1/f signal) in humans and provided solutions to these. First, both markers are captured in the EEG power spectral density (PSD), they co-exist and can confound one another. Therefore, I used spectral parameterisation to isolate each EEG PSD feature and examined their E-I mechanisms 3 separately. Second, E-I differences are predicted in multiple neuropsychiatric and neurodevelopmental conditions, such as autism. Yet, the E-I theory of autism has been inconsistently supported, possibly due to the limitations of current methods to capture dynamic E-I differences. Therefore, I investigated each marker, and their relationship to E-I signalling, in autistic and non-autistic people. Finally, direct and invasive techniques like optogenetic imaging are used to study E-I mechanisms in animals, but are not ethical in humans. Therefore, I investigated each marker and its relationship to E-I, non-invasively, using proton Magnetic Resonance Spectroscopy ([1H] MRS) E-I neurotransmitter concentrations and E-I pharmacological challenge with arbaclofen.

Starting with the aperiodic 1/f exponent. I observed that this is related to [1H] MRS Glx/ GABA+ concentrations (chapter 3 and 5). This relationship was the same for autistic and non-autistic people with on average no between-group difference in aperiodic exponents at baseline. However, E-I signalling is dynamic, and potential differences in E-I in autistic people may be masked by assessing average measures at baseline which do not tell us about differences in E-I ‘responsivity’ or homeostatic regulation (see Ellis*, Ahmad*, in prep). To probe E-I responsivity, I administered an E-I pharmacological challenge with arbaclofen, a GABAB receptor agonist, and evaluated whether any between-group differences in E-I responsivity differentially shift the aperiodic 1/f exponent. I discovered, for the first time in humans, that, across groups, aperiodic exponents increased in response to a high dose of arbaclofen (chapter 4 and 6). However, in response to a low dose of arbaclofen, there was a group by drug interaction where aperiodic exponents increased in the autistic individuals but decreased or did not change in non-autistic people (chapter 6).
Next, I examined how E-I signalling might be captured with beta power. Since, I found that arbaclofen alters aperiodic power it was important to control for this effect in the analyses. At baseline (chapter 7), I found that beta power was higher in autistic than non-4 autistic participants. Furthermore, there was a positive relationship between beta power-GABA+ in neurotypical people but this relationship was significantly different (negative) in autistic people. This may indicate that GABAergic activity modulates beta power differently in autistic and non-autistic people- but correlations provide limited explanatory evidence.
To directly test this hypothesis, I examined how beta power behaves when GABAergic activity is enhanced by GABAB receptor activation with arbaclofen (chapter 8). However, in contrast to prior study results, I observed no drug effects on beta power. The relationship between beta power and bulk measures of GABA as measured by spectroscopy does not appear to reflect a (GABAergic) dynamic relationship as previously assumed. Moreover, the failure of previous studies to control for co-occurring aperiodic power in their beta power measures is an important methodological flaw and spectral parameterisation is a necessary step to avoid methodological artifacts in EEG biomarker research.

In conclusion, the work described in this thesis demonstrates that the aperiodic 1/f exponent is a readily quantifiable metric; is robust to measurement error; relates to bulk [1H] MRS metrics of E-I; and is capable of revealing dynamic differences in E-I in neurodiverse cohorts (comprising autistic and non-autistic people). These discoveries provide a solid platform for the future use of this metric. The drug challenge design has also proven useful to extend our understanding of how E-I differences contribute toward neurodiversity. Finally, they may also help assess target engagement by candidate E-I based interventions at the level of the individual, paving the way for individualised pharmacological support options for those autistic people who would like that choice.
Date of Award1 Oct 2023
Original languageEnglish
Awarding Institution
  • King's College London
SupervisorDeclan Murphy (Supervisor), Jumana Ahmad (Supervisor) & Grainne McAlonan (Supervisor)

Cite this