Statistics from Altmetric.com
Understanding the mechanism of psychogenic non-epileptic seizures (PNES) is challenging. A recent review grouped currently hypothesised psychological models into four: (1) traumatic dissociation, (2) hard-wired anxious-arousal responses, (3) conversion defences and (4) conditioned behaviours, but concluded that determining which of these was correct, if any, went far beyond the available evidence.1 Patients’ subjective reports may lend striking support to some models, but are necessarily post-hoc, and susceptible to iatrogenic suggestion.2 Neuroimaging, such as functional MRI, provides more objective evidence, but is typically restricted to the inter-ictal phase because of potential movement during the scan.3 Single photon emission computed tomography (SPECT) would have distinct advantages in imaging of the peri-ictal state as the tracer injection and uptake occur at the time of the seizure while the scanning can be completed later, after the seizure has finished, overcoming the motion issues. The ictal scan could then be compared with an inter-ictal SPECT to determine the ictal contribution to cerebral blood flow. This process is standard for epilepsy surgery workup, and we combined the databases from two epilepsy surgery centres to find opportunistic cases where the ictal scans of potential surgery cases proved to be of PNES—either because the initial epilepsy diagnosis was mistaken or because the scan captured a comorbid non-epileptic seizure by accident.
A retrospective chart review (2013–2016) was conducted for patients admitted for characterisation of seizures at two major epilepsy centres in Melbourne, Australia who had an ictal SPECT scan which was subsequently confirmed to have captured a PNES (by epileptologist consensus reading of the concurrent video electroencephalograph), as well as a baseline (inter-ictal) SPECT scan. The project was approved by the Human Research Ethics Committees at Austin Health and the Royal Melbourne Hospital.
All SPECT scans used Tc 99m-ECD as the tracer. Scans at Royal Melbourne Hospital were performed on a Siemens Symbia T2 gamma camera while those at Austin Health were performed on a Siemens Symbia T16 gamma camera. For the analysis, the raw SPECT images were converted from DICOM into Analyse format using dcm2nii (www.mathworks.com). The inter-ictal images were then subtracted from the ictal images using the ISAS Bioimage suite, following their standard pipeline.4 The subtractions and the T1 MNI Template from the Bioimage suite were then imported into SPM12 (Wellcome Functional Imaging Laboratory). Images were centred to the anterior commissure, realigned and co-registered to the template. A one sample T-test was performed in a cluster-wise analysis without extended threshold, and significant clusters after correction for multiple comparisons (false-discovery rate) localised using the Talairach atlas.
Six subjects were identified (four females) aged between 18 and 54 years. Three patients were diagnosed with PNES alone, three with PNES plus comorbid temporal lobe epilepsy. The tracer was injected from 0 to 125 seconds following commencement of the ictal event. Four events were complex partial in phenomenology, one was tonic-clonic, and a seizure description was not available for one; they lasted for from 1 to 4 minutes, apart from one event which lasted for 3 hours, and one for which duration was unavailable.
Regional blood flow activation (hyperperfusion) was identified in five separate clusters at the time of the ictal event (see table 1). These were right ventromedial prefrontal cortex (vmPFC), right insula and three areas which are likely to represent artefact as they were either at the very edge of the brain image in the cerebellum (two clusters) or in white matter. There were no clusters of ictal hypoperfusion.
Others have reported SPECT ictal cerebral activation but have typically relied on visual inspection, or have performed subtraction analysis on individual scans, with very variable results.5 The largest such study found no difference in 11 patients, ictal hyperperfusion of the left insula in one patient, and of the right insula and right postero-lateral frontal cortex in a second patient.6 Clearly, there is considerable individual variation, and/or the effects may be too weak for visual inspection or single scans to detect, in which regard a group-wise, within-subject statistical analysis should be more powerful, and a subtraction design should help to account for the contribution of the considerable comorbidities of PNES. To our knowledge, ours is the first study to adopt this approach—though with a small, retrospective, heterogeneous sample, and drawn from two scanners.
Our main findings are of ictal co-activation of two interconnected right-sided cortical structures: the vmPFC and the insula. While these brain regions subserve complex functions, they are unified by a shared role in the processing of internally generated, self-referential phenomena. Activations of the vmPFC have been reported in the generation and modulation of negative affect, self-focussed thoughts, imagined feelings and emotional processing of episodic and autobiographical memory. The insula lies at the hub of integrating sensory and emotional symptoms, with its posterior division receiving somatosensory inputs which are transferred to the anterior division with input from limbic regions and the vmPFC. The core role of the anterior division is integration of interoceptive signals with top-down emotional and cognitive signals. In particular, the vmPFC-anterior insula connectivity has been hypothesised to integrate past events and subjective feelings and mediate subsequent behaviour.
Regarding the hypothesised models of PNES,1 these data fit best with Model 2 (hard-wired, anxious arousal) and particularly with a ‘panic without panic’ in which there may be a dissociation between a bodily anxious response and cognitive elements of anxiety secondary to vmPFC inhibition of amygdala (whose activation was just below significance, p=0.057). Activations of the insula and cerebellum reported here are partially consistent with Model 3 (conversion defence); however, previous studies of functional movement disorders have reported left-sided insula activations and reduced activation of the vmPFC.7 8 They provide less support for Model 4, a learnt response, since reinforcement of the seizure by the response would suggest it should have reduced any arousal, and for Model 1, as traumatic dissociation is associated with reduced activations of the vmPFC.9
Contributors JSO wrote the manuscript, LFC conducted the analysis, RAK conceived the project and wrote the manuscript; all other authors obtained the data and edited the manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.