Electrophysiological testing in concussion: A guide to clinical applications

Introduction

The neurological assessment of patients with concussion – including those with mild acute traumatic brain injury (TBI) – is challenging. A key reason is that head injury typically affects multiple systems, often combining central (brain) and peripheral (nerve and sensory organ) injury. Furthermore, since brain injury often results in a blunting of perceptual systems (with or without an initial loss of consciousness), the acute history in head injured patients may be misleading, even more so when obtained retrospectively. Thus, a manifest gait ataxia could result from injury to any component of the sensorimotor feedback loop including injury to the peripheral vestibular organ, dorsal column proprioceptive pathways from cervical trauma, cerebellar injury or the disruption of cortical networks (via subcortical white matter injury or multiple cortical contusions) that control human gait and postural control. Similarly, apparently impaired cognitive performance, e.g. an apparent visuo-spatial neglect, requires an assessment of the integrity of afferent visual pathways, from the periphery to the cortex. It follows that a wide range of assessment including clinical and laboratory testing is required to delineate the full extent of injury to the brain, its sensorium and its effectors. Concern about the long-term sequelae of mild TBI and its widely cited link with chronic traumatic encephalopathy particularly with repetitive mild TBI and exposure to sub-concussive head impacts in contact sports make this a clinically important topic. ¹

A brief note on definitions and meanings is essential at this point. Concussion – from the Latin, ‘concutere’ – meaning ‘to shake together’ – can be considered an event in which kinetic energy is transferred to the head. There is much divergence in the literature as different experts use the word concussion to mean different ‘diagnoses’ – some argue that it relates only to a transient impairment in brain function whilst others argue that concussion should always involve the term ‘traumatic brain injury’. ² The issue has been generated primarily by professional sports where there is continuous monitoring of the players who have suffered a blow to the head and, particularly with the benefit of video replays, manifest neurological dysfunction, e.g. gait imbalance or apparent confusion. In this context, where there is recorded evidence of a mental obtundation, current consensus is to describe this as a mild TBI (recent Berlin guidelines). ³ However, most cases presenting to a clinician with a blow to the head occur outside of a professional sporting context, where there is limited information on the nature of the injury and rarely video footage of the event and the post-concussive behaviour. At the point of attending the clinician, the patient may have a number of symptoms (and signs following assessment) – some of which may not relate to brain injury – e.g. benign paroxysmal positional vertigo (BPPV). Hence, it is potentially unhelpful to label such cases as having had a TBI (even ‘possible TBI’) simply based upon symptoms. We argue that it is much more helpful to consider concussion an event rather than a diagnosis – hence, patients developing symptoms or signs following a blow to the head should be labelled as a post-concussion syndrome, indicated by the constellation of symptoms and signs following a concussive event, i.e. an event where there has been a transfer of kinetic energy to the head. Within such a syndrome, the clinician should then attempt to define specific diagnoses causing these symptoms and/or signs, and such diagnoses could relate to peripheral and central neurological mechanisms. In this way, all potential diagnoses are picked up and treated. This is especially important for common diagnoses such as BPPV and vestibular migraine ⁴ (which of note is currently a diagnosis of exclusion and not possible to be diagnosed by electrophysiological testing). These can cause balance problems and may impede functional recovery either directly through their effect upon balance or indirectly through the effect of vestibular dysfunction of brain cognition and psychiatric morbidity. Indeed, vestibular diagnoses increase the rates of not returning to work three-fold at six months post-TBI.^5,6

Since we do not consider concussion a diagnosis but rather an event with an associated clinical syndrome – diagnostic testing of the whole neuroaxis is critical in assessing concussion patients – and neurophysiological testing is a crucial component in this assessment. ⁷ In particular, neurophysiological testing can interrogate the functional and anatomical integrity of neural systems as a whole as well as better understand the role and contribution of peripheral disorders in concussion patients. This review focuses on the diagnostic utility of neurophysiology in mild head trauma patients. We do not include patients with moderate or severe brain injury with coma – where neurophysiology has an undoubted utility – e.g. in excluding underlying epilepsy or providing initial insight into the prognosis of brain functional recovery. In contrast, it is unclear whether neurophysiological testing can provide a similar insight into patients with concussion with or without mild brain injury.

Given the multifocal impact of head trauma, we would advocate a multimodal assessment in acute concussion patients, and the importance of a comprehensive assessment is even more important in children. This is because there may be a greater difficulty in extracting a clear-cut history due to immaturity of cognitive systems, e.g. those mediating language, acute perceptual blunting related to brain injury, use of sedative drugs or psychological reactions in acute TBI. ⁸ Electrophysiology could provide important information in understanding questions such as the therapeutic benefits of rest during acute recovery and appropriate assessment in follow-up.

This review focuses on the evidence supporting the use of neurophysiological techniques in concussion such as electroencephalogram (EEG), polysomnography (PSG), brainstem auditory evoked potentials (BAEP), electro- and videonystagmography (ENG/VNG), vestibular evoked myogenic potentials (VEMPs), visually evoked potentials (VEP), somatosensory evoked potentials (SSEP) and transcranial magnetic stimulation (TMS). The relative paucity of published data highlights an immediate requirement for research using multi-modal assessment of patients with mild head trauma as this may provide more sensitive and specific assessment toolboxes.

Methods

Relevant articles were identified through a MEDLINE search strategy using relevant terms for concussion and electrophysiology testing (see Table 1 for search terms used). A search excluding non-English language articles, animal studies and articles published between 1 January 1975 and 1 August 2017 was performed. Potentially relevant articles were then extracted and full-text accessed. Studies were excluded on the basis of being unrelated to the study topic, inclusion of non-mild TBI patients and if no original data was presented (i.e. review articles). Articles were then grouped within each electrophysiological testing parameter (see Figure 1 for flow diagram of article inclusion and exclusion process). For the articles included in this review, a quality assessment was performed, using the QUADAS-2 tool (Table S1, Supplementary material). ⁹ Records were not excluded on the basis of the quality assessment.

OPEN IN VIEWER

Figure 1.

Flow diagram of article inclusion and exclusion process.

Visual system – Visual evoked potentials

VEPs’ use in assessing head trauma has been put forward since 1970s. ¹⁰ Despite this, published literature has produced mixed results on its ability to distinguish mild TBI. Werner and Vanderzant ¹¹ found no difference in the P100 value (the large positive peak wave that occurs at about 100 ms in healthy individuals) beyond three standard deviations of normality. In this particular group, 18 head injured subjects, defined as a period of unconsciousness as less than 20 min and Glasgow Coma Scale (GCS) on admission of ≥13, had VEPs recorded within two weeks of the injury.

Similarly, Papathanasopoulos et al. ¹² observed no differences in VEPs compared with normal controls; however, they found a significant within-subject decrease in P100 latency over a period of 30 days.

Nonetheless, it has been shown for subjects with persistent post-concussion syndrome, 30% had latencies beyond a 2.5 standard deviation of the normal limit ¹³ as well as statistically significant reductions in P100 amplitudes compared to a non-head injured control group. ¹⁴ Also, significant improvements in VEP latency and amplitude of concussed individuals have been reported after optometric rehabilitation. ¹⁵

More recently, possible magnocellular deficits have been reported for individuals with a history of mild TBI compared with visually normal control subjects based on their P100 VEP responses to a high-contrast (85%) checkerboard stimulus (1.49-cycle/degree check size) at a very low luminance level (0.3% transmittance; greater P100 latency and smaller amplitude) ¹⁶ as well as to low-contrast (20%) checkerboard stimuli of varying sizes (smaller P100 amplitude). ¹⁷

Interestingly, by analysing low-level VEPs evoked by stimuli with a homogenous motion or orientation pattern in comparison to stimuli of greater complexity (i.e. textured stimuli), Lachapelle et al. ¹⁸ observed that VEP peak times to homogenous stimuli did not differ between TBI patients and controls but peak times were longer to textured stimuli in TBI patients than controls. These findings may indicate that TBI patients show spared first-order visual processing (restricted to area V1 of visual cortex) but impaired higher-order visual processing mechanisms that may originate from second-order antero-posterior cortical processes higher in the visual processing chain.

Poltavski et al. ¹⁹ investigated a subset of patients with a history of concussion and subsequent convergence abnormalities on the basis that this particular visual disorder was a frequent finding in mild TBI patients. They compared VEP latencies in this group, and subjects with a convergence abnormality and no history of concussion, finding a significant prolonged difference in the P100 measurement.

In a follow-up case study, VEP amplitude progressively decreased over seven weeks post-injury after the soccer-sustained concussion in an 8-year-old, when compared to a pre-trauma VEP. ²⁰ In this case, cognitive abnormalities resolved around three months, whereas VEP parameters took 12 months to return to normal.

Auditory system – BAEP

Investigation comparing adult patients with persistent concussion symptoms to a control cohort on several auditory processing assessments found the individuals with concussions who performed abnormally in these tests (i.e. gaps in noise, temporal processing, binaural masking level difference and time compressed speech) demonstrated delay in latencies for waves I, III and V. ²¹ Similar findings in BAEPs between controls and individuals experiencing ‘minor head injury’ have been documented previously ²² as well as from three studies conducted in the 1980s.^23–25

In contrast, a study did not find any abnormal BAEPs in 31 patients with a post-concussive syndrome who had undergone ‘blunt trauma of the head’ one year earlier compared to 12 individuals without history of head injury. ²⁶ A Swedish study comparing aged-matched boxers, soccer players and field athletes did not find significant differences in BAEPs between these groups. ²⁷

BAEPs have been utilised in patients presenting with ‘minor’ head injury at initial assessment (within 48 h) of the trauma, demonstrating prolonged waveform latencies.^28–30 However, Soustiel et al. ²⁸ found that abnormalities in BAEPs did not correlate with whether patients experienced a post-concussion syndrome at a three month follow-up. Earlier similarly conducted research produced these findings also show that although abnormal BAEPs may be found in mild TBI, this did not have any predictive value in determining patients who would go onto develop a post-concussion syndrome.^31,32

Vestibular system – ENG/VNG and VEMPs

The VEMP provides a non-invasive complementary method for assessing peripheral vestibular function – specifically otolith function ³³ – and can be used at the bedside. The VEMP test involves delivering sound pressure to the vestibular system via headphones or a bone vibrator and short latency myogenic responses are recorded, either via electromyographic (EMG) electrodes on the ipsilateral sternocleidomastoid muscle – the cervical VEMP (cVEMP) – or the inferior oblique muscle over the contralateral eye – the ocular VEMP (oVEMP). ³⁴ The generator sites for the cVEMP and oVEMP are now recognised as being the saccule and utricle, respectively, ³⁵ and as such, these tests offer a unique objective methodology to probe these structures.

One of the most common and debilitating consequences of TBI is chronic dizziness (i.e. illusory self-motion) and imbalance, with the frequency of vertingious reported to be between 47 and 78% in the literature.^36,37 Symptoms of dizziness arise from a combination of causes, combining peripheral and central causes and can be challenging to differentiate. The commonest cause of post-TBI dizziness is BPPV, affecting half of patients with dizziness in the acute setting and 30–40% who have chronic symptoms.^4,38

Other peripheral causes for post head injury vertigo include vestibular nerve injury. ³⁹ The perilymphatic fistulas are commonly quoted in the literature; however, they are very rare (<1%) in one of the authors’ (BMS) experience with head trauma patients on an acute major trauma ward. Integrity of the peripheral vestibular system can be probed using the head impulse test vHIT ⁴⁰ to assess the vestibular-ocular reflex (VOR) and a variety of techniques that record eye movements (e.g. electronystagmography or VNG) to record static and dynamic eye movements. ⁴¹ Whilst the vHIT provides a quick, objective and repeatable method of assessing high frequency of the integrity of the semicircular canal based ocular reflex arcs, it is severely limited by the fact that high acceleration head thrusts are needed to elicit the response, and therefore this test is of limited value in the acute stage. In contrast, the use of portable VNG goggles which can record eye movements with fixation removed provide an excellent tool for gathering information relating to peripheral disorders like BPPV as well as central deficits that may be manifested as aberrant eye movements.

Much of the identified literature relating to the use of ENG/VNG and VEMPs pertained to mild TBI sustained from blast injuries in military personnel.^42–44 Gattu et al. ⁴⁴ presented four Veterans, aged 29–46 years, who complained of chronic dizziness and/or postural instability following blast exposures. Three of the four individuals were found to have abnormal vestibular function on the basis of ENG/VNG and VEMP testing. In three cases, VEMPs were either unilaterally or bilaterally reduced in amplitude or absent. In one case, eye movement recordings showed prolonged latency for rightward saccades, suggestive of central pathology. Similar findings were reported by Scherer et al. ⁴³ in a study of 24 service members recovering from blast-related TBI sustained in Iraq or Afghanistan. Six of the 12 symptomatic groups demonstrated abnormal nystagmus or oculomotor findings on VNG testing. Meanwhile, a prospective cohort study on United States Marine Corps’ instructors who were exposed to repeated blast exposures during explosive breaching training found that upbeat nystagmus was common and correlated with a history of mild TBI, with the authors postulating that this nystagmus could be acute blast effect. ⁴²

Previous research, retrospectively examining the case notes of patients presenting with dizziness following a road traffic accident, has compared mild head injury versus whiplash injury using ENG. ⁴⁵ During patient assessment, it was determined whether, in addition to the whiplash injury, the patient had incurred a blow to the head apart from impact with the head rest. In patients who were considered to have a whiplash injury alone, no detectable ENG abnormalities were found, whilst in all the head injured patients, an ENG abnormality was found, although it did not go on to specify this. determined that all patients those he did finding ENG abnormalities are limited to patients who have suffered a head injury. The retrospective nature of these data is however a major limitation.

Transcranial magnetic stimulation

The neurophysiology of the primary motor cortex (M1) can be probed using non-invasive stimulation techniques such as TMS. Given the consistent topography of motor representations in M1, TMS can be used to generate cortically mediated myogenic potentials in a target muscle (e.g. the first dorsal interosseous). A variety of TMS paradigms have been demonstrated to probe different aspects of M1 function, from resting to active excitability, and intracortical excitation, which is something that has been demonstrated to be abnormal in patients with concussion.^46,47

The resting motor threshold is defined as the lowest intensity of TMS (in terms of % of maximum for a given setup) that elicits a 50 µV myogenic potential (peak-to-peak) in the target muscle at rest, in 5 out of 10 TMS trials. The motor threshold can be measured when the muscle is at rest or in a state of active contraction when the required criteria are 100 µV myogenic potential in 5/10 TMS pulses. ⁴⁸

TMS has been used safely in concussion and mild TBI as described in the literature below, with preliminary guidelines for its use in moderate-to-severe TBI also reported. ⁴⁹

Upper limb SEPs and central motor conduction time with respect to TMS were found to be unaltered in asymptomatic concussed athletes. ⁵⁰ However, when TMS was performed immediately and following a sport-related concussion (acute), the abductor digiti minimi (ADM) motor evoked potential (MEP) amplitude ratio was found to be reduced in three and five days after the injury. ⁵¹ Also, in the same study, the MEP latency with cortical stimulation was found to be prolonged on day 10 with recording from the abductor pollicis brevis muscle. In a separate study, central motor latency times and amplitude (not amplitude ratio) were normal 25 months after TBI with first dorsal interosseous recording. ⁵² Comparatively, in symptomatic patients, up to five years post mild head-injury, MEPs have been found to be delayed.^53,54

The cortical silent period (cSP) has been found to be prolonged in patients with concussion.^55–57 One study in footballers (soccer) which garnered much publicity found significant changes in this specific TMS parameter – cSP – following a heading practice session in 19 subjects. ⁵⁸ Specifically, the cSP was shown to be longer post-heading. Problematically, however, this study did not follow several experts’ recommendations when assessing the cSP (see Wolterts et al. ⁵⁹ for review) including inappropriate use of TMS intensity, the lack of a duration-stimulus intensity curve for the cSP and the lack of a defined automated algorithm for extracting the cSP to remove observer bias. Additionally, the study failed to test a control group or better a control task (in a balanced order design with blinded analysis). Finally, the significance of the primary outcome was borderline (reported at P = 0.049), which, when taken in the context of the aforementioned study limitations, leaves the findings of this study open to question. At the least, a properly controlled study is required before the basic findings of this study can be considered important for sports medicine. Nonetheless, other studies have found no significant difference in the cSP between concussion versus non-concussion participants.^60,61 In addition, using this paradigm in a paediatric population, no difference was found between mild TBI and controls. ⁶²

Somatosensory evoked potentials

Only one relevant study was retrieved relating to the use of SEPs in concussion. ⁶³ Abnormal N60 latencies within 48 h of the injury and at three months follow up were found in concussion subjects for a sample of 20 consecutive patients presenting to the emergency unit. This parameter was found to have normalised at six months.

Of note, the data from studies in more severely affected TBI cases are conflicting in terms of its ability to predict acute brain injury severity and long-term prognosis. For example, in severe TBI, short latency SEPs did not correlate with clinical disability, ⁶⁴ whereas a normal central conduction time (the time with cervical and cortical waveforms with reference to the median nerves) or normal cortical time correlated with better survival and less disability.^65,66 In addition, SEPs have been found to be better predictors of outcome than motor or pupillary responses. ⁶⁷ Comparatively, in one study, 10% of children with acute severe TBI, with bilaterally absent SEPs, had favourable outcomes. ⁶⁸ An important confound not mentioned in the above SEP studies was the routine exclusion of a peripheral neuropathy.

EEG and event-related potentials

There are many EEG studies in concussion with significant heterogeneity in methodology including their definition of concussion, time of testing post-concussion, patient groups and medical comorbidities, making comparison problematic. More recently, Munia et al.^69–71 have produced a number of works demonstrating significant differences in EEG recordings, in terms of range of frequencies and power of waveforms, in concussed individuals compared to non-concussed subjects. In addition, one study found that EEG was able to show significant differences between a group with a history of concussion versus a group without, in comparison to a standard neurocognitive test – the King Devick Test – which was not able to find any significant difference between the two groups. ⁶⁹ Moreover, a newly developed EEG-only test, using a so-called Brain Function Index (BFI), has been used as a biomarker to detect concussion in patients presenting in Emergency Departments.^72,73 The BFI is derived from EEG features that may relate to the ‘disruption in “connectivity” related to integrity of fiber tracts’, and such disconnection is common in significant head injury. The BFI is recorded as a percentile of a non-head injured population ⁷³ and has shown promise in a multi-centre Emergency Department study. ⁷²

Quantitative EEG (qEEG) has also shown significant abnormalities in mild TBI patients. qEEG technology digitises brain signals, so it can be analysed mathematically. Researchers can analyse the strength of brainwaves at different frequencies (i.e. theta, alpha, beta, gamma), which can then be processed into colour maps of brain functioning, sometimes referred to as ‘brain mapping’. McCrea et al. ⁷⁴ presented results showing that for the beta frequency band, concussed athletes exhibited increased power on the day of injury and at eight days post-injury.

Waveforms in EEGs have been found to be abnormal, with regards to hemisphere asymmetry and coherence, in individuals with persistent concussive symptoms for alpha and beta bandwidths^75,76 as well as delta and theta.^77,78 A case report of an ambulatory EEG in a patient undergoing investigation to rule out unrecognised seizures (he had no definite clinical seizures for more than a year) showed a decrease in beta and alpha waveforms following an incidental mild TBI (during the EEG recording the patient was involved in a road traffic collision). ⁷⁹ The changes did ultimately resolve after 20 min.

Earlier studies have used ERPs (event-related potentials) to demonstrate differences in EEGs in concussed athletes versus non-concussed atheles^80,81 and reported indices of ERPs as sensitive correlates of persistent post-concussion symptoms. ⁸² The ERP measure illustrates the averaged EEG signal time-locked to the provided stimulus and involves different components labeled by their amplitude polarity (i.e. P for positive and N for negative) and time range in milliseconds. ⁸⁰ Furthermore, concussed subjects who were, on average, over three years post-injury demonstrated decrements in N2 and P3b amplitudes relative to individuals without a history of concussion, ⁸³ in response to a three-stimulus oddball task, although other research has shown no difference. ⁸⁴ Nevertheless, the P3 component has been repeatedly shown to have chronic amplitude suppression among multi-concussion athletes.^85–91

However, results have been mixed with previous study of an auditory oddball and Go/No-go tasks not differentiating concussed athletes from controls,^92–94 whilst other similar research has shown differences.^95–97 Additional research suggests that ERPs are only useful with a pre-concussion baseline.^98,99 Combining EEG recording with a neurocognitive assessment – a working memory task – showed lower amplitudes for frontal N350 and parietal P300.^100–102 Furthermore, EEG recording, when standing versus sitting, demonstrated the most significant findings, of decreased power waveforms, in concussed individuals when standing relative to same individuals when sitting and non-concussed subjects.^103,104 Slobounov et al. ¹⁰⁵ also showed a persistent reduction in ERP amplitude prior to initiation of postural movement up to 30 days post-concussion (in comparison to ERP findings prior to injury).

However, other authors have highlighted a lack of clear and consistent EEG features that are unique to concussion subjects^106–108 as well as no differences in recordings being detected between head injured versus non-head injured groups. ¹⁰⁹ Further efforts have been made to explore unique features of EEG abnormalities. One such example is a measure of non-stationarity EEG signals.^107,110 This measure uses wavelet entropy to analyse transient features of non-stationary signals and is based on the shift of the dominant frequency of the EEG signal over time. It has produced values that are significantly decreased in subjects suffering from mild TBI compared to non-affected participants.^107,110

Sleep – PSG

The ‘gold standard’ electrophysiological monitoring of sleep is performed using PSG, although other measures exist to detect sleep disturbance such as actigraphy, the Multiple Sleep Latency Test and self-reports on sleep questionnaires.

Focusing on PSG, we identified a number of studies highlighting abnormalities on PSG in relation to patients with mild TBI. In the acute setting, sleep EEGs one week post TBI were abnormal in patients who had a mild TBI (defined by an admission GCS of ≥13), in comparison to a control group, showing a lower delta power. ¹¹¹ Arbour et al. ¹¹² also found significant increase in beta power in sleep EEGs, compared to controls, in 34 patients after mild TBI (sustained a mean 10 weeks earlier). Changes in these parameters were not associated with participants’ subjective evaluation of their sleep. The authors concluded that while abnormalities may be attributable to the occurrence of a brain injury, they could also be related to the presence of pain and anxiety. This was related to previous work by the same group that found lower delta power and high gamma power only in mild TBI patients with chronic pain in comparison to mild TBI patients without chronic pain. ¹¹³ Furthermore, similar abnormalities on PSG have been found in patients with post-traumatic stress disorder and mild TBI. ¹¹⁴

Abnormalities on PSG have also been found on follow-up of mild TBI patients, affecting both timing and architecture of their sleep patterns. This includes shorter REM (rapid eye movement) onset latencies, ¹¹⁵ higher light-sleep non-REM stage 2 scores and significantly lower REM sleep scores ¹¹⁶ as well as increased duration of nocturnal wakefulness ¹¹⁷ and nocturnal awakenings ¹¹⁸ all in contrast to controls. The data are not consistent however as other studies found no differences in PSG measures between mild TBI patients and healthy controls with no sleep complaints.^119,120

Conclusion

Neuro-electrophysiological testing, despite its wide availability, has been disappointing as a tool for diagnosis and monitoring recovery in concussion. At present, there is no convincing evidence to support the use of routine electrophysiological testing for concussion. There have not been any prospective controlled studies with clear inclusion and exclusion criteria and explicit outcome measures which can be correlated with neurophysiological parameters. Often, the many confounds are poorly controlled, e.g. neuroactive drugs can affect neurophysiological responses. Finally, future studies require multi-modal assessments, linking clinical, behavioural, neurophysiology and neuroimaging data in prospective fashion.

Supplemental Material