Electrodiagnostic Applications of Somatosensory Evoked High-Frequency EEG Oscillations: Technical Considerations

Brain Research Bulletin 137 (2018) 351–355

Contents lists available at ScienceDirect

Brain Research Bulletin

journal homepage: www.elsevier.com/locate/brainresbull

Research report Electrodiagnostic applications of somatosensory evoked high-frequency EEG T oscillations: Technical considerations ⁎ A.J. Simpsona,1, M.O. Cunninghama, M.R. Bakera,b,c, a Institute of Neuroscience, The Medical School, Newcastle University, NE2 4HH, UK b Department of Neurology, Royal Victoria Inﬁrmary, Newcastle upon Tyne Hospitals NHS Foundation Trust, Newcastle upon Tyne, NE1 4LP, UK c Department of Clinical Neurophysiology, Royal Victoria Inﬁrmary, Newcastle upon Tyne Hospitals NHS Foundation Trust, Newcastle upon Tyne, NE1 4LP, UK

ARTICLE INFO ABSTRACT

Keywords: Introduction: High frequency oscillations (HFOs) embedded within the somatosensory evoked potential (SEP) are Somatosensory evoked potentials (SEPs) not routinely recorded/measured as part of standard clinical SEPs. However, HFOs could provide important High-frequency oscillations (HFOs) additional diagnostic/prognostic information in various patient groups in whom SEPs are tested routinely. One EEG area is the management of patients with hypoxic ischaemic encephalopathy (HIE) in the intensive care unit (ICU). However, the sensitivity of standard clinical SEP recording techniques for detecting HFOs is unknown. Methods: SEPs were recorded using routine clinical methods in 17 healthy subjects (median nerve stimulation; 0.5 ms pulse width; 5 Hz; maximum 4000 stimuli) in an unshielded laboratory. Bipolar EEG recordings were acquired (gain 50 k; bandpass 3Hz–2 kHz; sampling rate 5 kHz; non-inverting electrode 2 cm anterior to C3/C4; inverting electrode 2 cm posterior to C3/C4). Data analysis was performed in MATLAB. Results: SEP-HFOs were detected in 65% of controls using standard clinical recording techniques. In 3 controls without significant HFOs, experiments were repeated using a linear electrode array with higher spatial sampling frequency. SEP-HFOs were observed in all 3 subjects. Conclusions: Currently standard clinical methods of recording SEPs are not sufficiently sensitive to permit the inclusion of SEP-HFOs in routine clinical diagnostic/prognostic assessments. Whilst an increase in the number/ density of EEG electrodes should improve the sensitivity for detecting SEP-HFOs, this requires confirmation. By improving and standardising clinical SEP recording protocols to permit the acquisition/analysis of SEP-HFOs, it should be possible to gain important insights into the pathophysiology of neurological disorders and refine the management of conditions such as HIE.

1. Introduction (15–20 ms after the stimulus) and a late component (20–30 ms after the stimulus). Early HFOs are thought to represent synchronous action High frequency oscillations (HFOs) around 600 Hz are a feature of potential volleys in thalamo-cortical neurons because, unlike late HFO, the N20 (Cracco and Cracco, 1976) or N20 m (Curio et al., 1994) they are not abolished by kynurenic acid, a non-specific glutamatergic component of the somatosensory evoked potential (SEP) or somato- antagonist. Late HFOs are generated by spike bursting exclusively sensory evoked magnetic field (SEF) following median nerve stimula- within cortical networks, which involve complex network interactions tion. Experimental evidence suggests that the N20 reflects underlying between principal cells and GABAergic fast-spiking interneurons, which synchronous excitatory (glutamatergic) postsynaptic potentials in themselves form interconnected networks coupled by gap junctions. apical dendrites of area 3b pyramidal neurons, generated by the sy- SEP-HFOs have been studied in patients with a variety of disorders, naptic volley arriving from thalamic projection neurons. The origins of including Parkinson’s disease, myoclonic epilepsy (Mochizuki et al., somatosensory HFOs are much more complex (Baker et al., 2003; Ozaki 1999), dystonia (Inoue et al., 2004), benign rolandic epilepsy (Kubota and Hashimoto, 2011). Somatosensory evoked potential high frequency et al., 2004), migraine (Coppola et al., 2005; Sakuma et al., 2004) and oscillations (SEP-HFOs) divide functionally into an early component schizophrenia (Waberski et al., 2004). It has therefore been proposed

Abbreviations: EEG, electroencephalogram; HFO, high frequency oscillation; HIE, hypoxic ischaemic encephalopathy; ICU, intensive care unit; MT, motor threshold; SEP, somatosensory evoked potential; SNR, signal to noise ratio ⁎ Corresponding author at: Institute of Neuroscience, The Medical School, Newcastle University, Newcastle upon Tyne, NE2 4HH, UK. E-mail address: [email protected] (M.R. Baker). 1 Supported by an undergraduate student bursary from the British Society of Clinical Neurophysiologists (BSCN). https://doi.org/10.1016/j.brainresbull.2018.01.011 Received 25 September 2017; Received in revised form 2 January 2018; Accepted 15 January 2018 0361-9230/ © 2018 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons.org/licenses/BY/4.0/). A.J. Simpson et al. Brain Research Bulletin 137 (2018) 351–355 that SEP-HFOs might be ‘useful in various CNS diseases for testing in- hibitory function in somatosensory information processing’ (Ozaki and Hashimoto, 2011); or more broadly the integrity of interneuronal cir- cuitry within parietal cortex. One potential application for SEP-HFOs might be to improve the prognostic information provided by SEP testing on the intensive care unit in patients with traumatic and non-traumatic brain injury, including hypoxic ischaemic encephalopathy (Carter and Butt, 2005); for example, dissociation of early and late HFOs could assess the extent of thalamic and cortical dysfunction respectively. However, SEP-HFOs are not routinely measured in clinical practice and consequently no clinical standard for measuring SEP-HFOs has been established. Moreover, nothing is known regarding the sensitivity and speciﬁcity of SEP-HFOs, particularly when data is acquired in a rela- tively electrically hostile environment, as might be encountered in intensive care. Motivated by these issues we tested the sensitivity of standard clinical SEP recording methods at detecting SEP-HFOs outside the setting of a clinical neurophysiology laboratory (and therefore without the beneﬁt of electrical shielding) in an attempt to simulate an intensive care unit.

2. Methods

Experiments were approved by the Newcastle University Research Ethics Committee and informed written consent was obtained from all participants. Fig. 1. A An example of a typical clinical somatosensory evoked potential (SEP) acquired using a low pass ﬁlter. B. The same data as in A but bandpass ﬁltered (300–1000 Hz) to 2.1. Demographics and environment reveal high frequency oscillations (HFOs). Note that the same timebase has been used in A and B. Initially, 21 healthy human participants (18 right-handed; 11 male; – mean age 29.6 years; range 20 57 years) were recruited to this study. (Cambridge Electronic Design Ltd, Cambridge, UK). An electrode im- SEPs could not be tested in 4 of these subjects, either because they were pedance of less than 5 kΩ was achieved in all recordings to reduce noise unable to tolerate high frequency electrical stimulation of the median and maximise the chances of detecting HFO (Fedele et al., 2012). Whilst nerve, or because hair-type prevented EEG electrode placement. SEPs needle electrodes might have reduced recording impedance and re- were acquired in a laboratory without electrical shielding in close duced some of the high-frequency signal attenuation by the scalp, we proximity to an electrically noisy computer laboratory (to simulate to did not have ethical approval to use such electrodes in healthy controls. some extent the electrically hostile environment of a typical intensive To achieve higher spatial density in three of the controls (subjects care unit). 17, 18 and 20) we repeated SEP recordings using a linear arrangement of four electrodes (F3, C3 + 2 cm, C3 and C3–2 cm). 2.2. Stimulation

The median nerve in the dominant arm was stimulated at the wrist 2.4. Analysis via adhesive electrodes (Bio-Logic M0476; Natus Medical, Mundelein, IL, USA) with a DS7AH stimulator (Digitimer, Welwyn Garden City, EEG recordings were analysed in the MATLAB environment UK). Approximate motor threshold (MT) was first determined at a low (Mathworks Inc., Natick, USA) allowing average SEPs to be constructed repetition rate (1 Hz) by inspection and the stimulator then set at and analysed post hoc. Low frequency components of the SEP (P15, N20, 1.5MT. A total of 2000 (n = 4) or 4000 (n = 17) stimuli (0.5 ms pulse P35) were identified by applying a low pass filter of 300 Hz (See width; 5 Hz) were delivered in each subject. The stimulation frequency Fig. 1A) and for HFOs we applied an acausal 300–1000 Hz bandpass used was chosen in order to minimise the duration of experiments filter using the fir1 function in MATLAB and a filter order of 100; this without attenuating the N20 or late HFO as has been observed at sti- yielded a pass band between the −3 dB points (where power is atte- mulation frequencies > 8 Hz (Klostermann et al., 2001; Urasaki et al., nuated by 50%) of 350–950 Hz (See Fig. 1B). 2002). The latencies of the low frequency components of the SEP were identified using the findpeaks function in MATLAB. SEP-HFOs were not 2.3. Recording always clearly distinguishable from background noise. SEP-HFOs were therefore considered to be significant (and thus present) if the spectral Bipolar EEG recordings were made with adhesive electrodes density of HFOs (300Hz–1000 Hz) 15–25 ms after the stimulus was (Neuroline 720, Ambu, Denmark), with additional electrode gel more than 2 standard deviations greater than the baseline spectral (Spectra360, Parker Laboratories Inc., NJ, USA) as required, over con- density (the 15–25 ms window was chosen after measuring the duration tralateral sensorimotor cortex. The scalp was cleaned and prepared of the HFO in each subject). Mean baseline spectral density in the with alcohol and the non-inverting electrode positioned 2 cm anterior 300Hz–1000 Hz range was calculated by averaging twenty contiguous to C3/C4 (C3/C4 + 2 cm) and the inverting electrode 2 cm posterior to 1 ms bins 30–50 ms prior to the stimulus artefact. This baseline period C3/C4 (C3/C4–2 cm; international 10–20 system). A reference elec- was chosen to avoid the stimulus artefact or later components of the trode was placed on the forehead ipsilateral to the recording electrodes. response. The signal to noise ratio was defined as the mean SEP-HFO EEG was recorded using a D360 system (Digitimer, Welwyn Garden spectral density divided by the mean baseline spectral density. We did City, UK). Signals were amplified (gain 50k) and filtered (bandpass not analyse early and late components of SEP-HFOs separately as there 3Hz–2000 Hz) before being digitised (5 kHz sampling rate) using a is no recognised standard method for separating these components and Power1401 interface connected to a computer running Spike2 software our aim was simply to establish whether any SEP-HFOs could be

352