Clinical Neurophysiology 126 (2015) 1234–1245

Contents lists available at ScienceDirect

Clinical Neurophysiology journal homepage: www.elsevier.com/locate/clinph

Single motor unit responses underlying cervical vestibular evoked myogenic potentials produced by bone-conducted stimuli Sally M. Rosengren a,e,⇑, James G. Colebatch b, Dominik Straumann c, Konrad P. Weber c,d a

Neurology Department, Royal Prince Alfred Hospital, Sydney, Australia Prince of Wales Clinical School and Medical Research Institute, University of New South Wales, Sydney, Australia c Neurology Department, University Hospital Zurich, Switzerland d Ophthalmology Department, University Hospital Zurich, Switzerland e Central Clinical School, University of Sydney, Australia b

See Editorial, pages 1067–1068

a r t i c l e

i n f o

Article history: Accepted 15 July 2014 Available online 18 September 2014 Keywords: Vestibular evoked myogenic potential Otolith Bone conduction Single motor unit Sternocleidomastoid muscle VEMP Vibration

h i g h l i g h t s  Bone-conducted (BC) stimulation is a useful cVEMP stimulus in patients with conductive hearing loss

as it bypasses the middle ear.  Our single motor unit data show that the mainly inhibitory cVEMP may change polarity with different

directions of BC stimulation to become an excitatory reflex.  In some conditions the BC cVEMP is likely to receive contributions from end organs in addition to the

saccule, such as the utricle.

a b s t r a c t Objective: Cervical vestibular evoked myogenic potentials (cVEMPs) are muscle reflexes recorded from the sternocleidomastoid (SCM) neck muscles following vestibular activation with air- or bone-conducted (BC) stimulation. We investigated the effect of different forms of BC stimulation on the single motor unit response underlying the cVEMP. Methods: We tested 8 healthy human subjects with 5 different stimuli. Motor units were recorded with thin concentric needle electrodes; surface potentials were recorded simultaneously. Results: The polarity of the initial change (at approx. 15 ms) in single motor unit activity reflected the polarity of the surface cVEMPs: a short-latency decrease in activity (inhibition) was seen with the four stimuli that produced a positive surface potential (p13), while an initial increase in activity (excitation) was seen with the stimulus that produced a negative surface potential. Conclusions: BC stimulation with common clinical stimuli usually produces an inhibition in single motor unit activity in the ipsilateral SCM muscle. However the projections activated by BC stimulation are not exclusively inhibitory in nature and depend upon the shape and direction of the stimulus. Significance: The utricle is likely to contribute to some BC cVEMPs, as some stimuli clearly evoke an excitation that is not likely to be saccular in origin. Ó 2014 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.

1. Introduction ⇑ Corresponding author at: Royal Prince Alfred Hospital, Neurology Department, Level 8, Missenden Rd, Camperdown, NSW 2050, Australia. Tel.: +61 295157565. E-mail address: [email protected] (S.M. Rosengren).

Vestibular evoked myogenic potentials (VEMPs) are muscle reflexes elicited by activation of the vestibular system with short bursts of sound, vibration or galvanic stimulation. They were first

http://dx.doi.org/10.1016/j.clinph.2014.07.037 1388-2457/Ó 2014 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.

S.M. Rosengren et al. / Clinical Neurophysiology 126 (2015) 1234–1245

described in the sternocleidomastoid (SCM) neck muscles in response to stimulation with loud air-conducted (AC) clicks (Colebatch and Halmagyi, 1992) and are now often called cervical VEMPs (cVEMPs) to distinguish them from more recently reported reflexes in the extraocular (ocular VEMPs) and masseter muscles (see Rosengren et al., 2010 for review). The cVEMP is recorded from an active surface electrode placed over the middle to upper third of the SCM muscle belly and a reference over the medial clavicle. The reflex usually consists of a short-latency, biphasic positive–negative potential with peak latencies of approximately 13 and 23 ms, respectively (i.e. p13–n23). For clinical purposes, cVEMPs are most commonly evoked by AC sound stimulation and recorded in the muscle ipsilateral to the stimulated ear. cVEMPs are not dependent upon hearing and are therefore present in patients with sensorineural hearing loss. However, they are attenuated or absent in patients with conductive hearing loss, as the air-conducted stimulus requires efficient transfer through the outer and middle ear to the vestibule (Bath et al., 1999). To overcome this disadvantage of air-conducted sound, Halmagyi et al. (1995) demonstrated that cVEMPs could also be elicited by tapping the forehead with a clinical reflex hammer. Following this, Sheykholeslami et al. (2000, 2001) reported that a clinical bone-conductor normally used to test hearing could also evoke cVEMPs. The advantages of the bone-conductor are that it allows control of stimulus shape and frequency, enables threshold determination and is less ‘operator-dependent’ than a reflex hammer. Using either stimulus, the shortest-latency response has the same biphasic waveform as the AC cVEMP and is vestibular-dependent, though it is followed by a second, non-vestibular biphasic wave, now hypothesised to be produced by activation of neck stretch receptors. Importantly, the bone-conducted (BC) cVEMP is present in patients with conductive hearing loss, as the stimulus bypasses the conductive mechanism of the middle ear. As a result, BC stimulation is considered a good substitute for AC sound in patients with conductive hearing loss. BC stimulation has been used in several clinical studies, for example in patients with otitis media (e.g. Monobe and Murofushi, 2004; Seo et al., 2008; Yang and Young, 2003), but has not become a standard cVEMP stimulus, possibly due to the need for an additional amplifier to provide sufficient drive to the bone conductor. There has been a recent increase in interest in BC cVEMPs as a result of the popularity of more powerful vibrators used to elicit ocular VEMPs. These stronger vibrators have a greater effective frequency range, extending to lower frequencies than the audiological bone conductors. The most common types of stimulus delivered by these vibrators to evoke cVEMPs are: sine waves, often at 500 Hz and sometimes delivered to the forehead near Fz (e.g. Cai et al., 2011; Manzari et al., 2010, 2012); square waves, either delivered to the forehead (e.g. Taylor et al., 2011, 2012) or inion (e.g. Huang et al., 2011); and controlled taps delivered to the mastoid (stimulus drive in the form of a gamma distribution, directed either toward the mastoid [inward taps] or away from it [outward taps] Rosengren et al., 2009; Govender et al., 2011). Both air-and bone-conducted vestibular stimuli are thought to activate similar populations of irregularly firing otolith afferents (Curthoys et al., 2006). There are, however, two properties of skull vibration that render BC cVEMPs more complicated than those evoked by AC sound. First, during BC stimulation both ears are stimulated simultaneously and, second, BC stimuli can be applied to different sites on the head and therefore produce linear acceleration in different directions. When a BC stimulus is applied to the midline (e.g. the forehead), the vibration reaching the vestibule is likely to be relatively equal on both sides and deflect vestibular hairs cells in a similar direction, producing symmetric reflexes (e.g. Halmagyi et al., 1995). When applied to lateralised sites on the skull, the stimulus strength may not

1235

be equal and the direction of hair deflection will be different in each ear. For example, vibration applied at the mastoid differentially activates the two ears. When the stimulus is a sine wave of around 500–1000 Hz, cVEMPs are present bilaterally in normal subjects and have the same polarity on both sides of the neck, but the skull acceleration and reflex amplitude are usually larger on the side of the bone conductor (McNerney and Burkard, 2011; Welgampola et al., 2003). In contrast, when the stimulus has a lower dominant frequency, such as a tendon hammer tap, the skull acceleration is approximately equal on both sides but is oppositely-directed (e.g. both sides move away from the hammer, causing the ipsilateral ear to move medially and the contralateral ear to move laterally) and the cVEMPs have similar size but different polarity and/or peak latency in the ipsilateral and contralateral SCM muscles (Brantberg et al., 2002, 2003, 2008, 2009; Cai et al., 2011; Rosengren et al., 2009; Todd et al., 2008). Therefore different types of AC and BC stimulation probably activate distinct, although often overlapping, populations of vestibular otolith afferents. Given the recent interest in BC cVEMPs we wished to systematically examine the motor unit response to skull vibration in human SCM muscles. The change in muscle activity that underlies the cVEMP evoked by AC sound and galvanic stimulation was determined by Colebatch and Rothwell (2004), who recorded the responses of single motor units in SCM muscles in normal volunteers. They found that the initial surface positivity (p13) seen for the SCM ipsilateral to click or cathodal galvanic stimulation was associated with a brief decrease or gap in motor unit firing, i.e. a short inhibition of the motoneurone supplying the motor unit. This determined the basis of the cVEMP reflex and demonstrated the nature of the projection to the ipsilateral SCM evoked by AC sound and galvanic stimulation in humans. To extend these findings, we investigated the single motor unit response to BC stimulation using several commonly used stimulus types: a 500 Hz BC tone burst evoked by a traditional B-71 bone conductor on the mastoid, the same shape 500 Hz BC tone burst delivered to the forehead using a minishaker and inward and outward taps delivered to the mastoid with a minishaker. Responses to these stimuli were compared to those evoked by AC sound. All of these stimuli, except for the outward tap, have been shown to evoke cVEMPs with typical positive–negative (i.e. p13–n23) polarity (Cai et al., 2011; Colebatch et al., 1994; Rosengren et al., 2009), while the outward tap produces a cVEMP with the opposite polarity (negative–positive) (Rosengren et al., 2009). We predicted that the single motor unit responses would mirror the surface responses and change with changing stimulus type. In particular, we hypothesised that some surface cVEMPs would be associated with an increase in muscle activity, i.e. an excitation of the muscle, as the surface responses are known to sometimes have inverted polarity.

2. Methods 2.1. Subjects Eight healthy human subjects were studied over multiple sessions (4 females, 4 males; mean age 36 years, range 24–49 years). All subjects were tested with each type of stimulus, except one subject who only completed one session with sound stimulation and was not tested further. The subjects had no history of conductive hearing loss or vestibular or neurological disease. Participants were staff and students at University Hospital Zurich and all gave written informed consent according to the Declaration of Helsinki. The study was approved by the local ethics committee (ethics committee of the canton of Zurich, 2010-0177/3).

1236

S.M. Rosengren et al. / Clinical Neurophysiology 126 (2015) 1234–1245

2.2. Vestibular stimulation The vestibular stimuli included one AC sound stimulus and four different types of BC stimulation. The sound stimulus (‘‘AC sound’’) was an unshaped 500 Hz burst of 4 ms duration delivered with headphones (TDH 39, Telephonics Corp., Farmingdale, USA) and custom amplifier at 142 dB peak SPL. The BC stimuli were: (1) ‘‘BC mastoid’’ – unshaped tone bursts (500 Hz, 4 ms) delivered to the mastoid process with a B-71 bone-conductor (Radioear Corp., New Eagle, PA, USA); (2) ‘‘BC forehead’’ – tone bursts with the same shape and frequency delivered to the forehead posterior to the hairline at Fz with a hand-held minishaker (model 4810; amplifier model 2706, Brüel & Kjaer P/L, Denmark); (3 and 4) lateral taps delivered to the mastoid process with the minishaker (in the shape of a gamma distribution, similar to a tendon hammer tap; Todd et al., 2008). The minishaker was fitted with a custom plastic cylindrical rod, which was held perpendicular to the head by the experimenter with about 1–2 kg of force. The lateral tap could be either positive (i.e. the rod moved away from the minishaker and toward the subject’s head) or negative (the rod moved toward the minishaker and away from the head). As the subject’s head was not fixed, the elasticity of the neck, coupled with the tonic force, ensured that the head accelerated away from the minishaker when its displacement was positive and toward the minishaker when its displacement was negative. The taps with positive polarity therefore moved the ipsilateral mastoid inward (stimulus 3 – ‘‘Inward tap’’) and those with negative polarity moved the ipsilateral mastoid outward (toward the minishaker: stimulus 4 – ‘‘Outward tap’’). The stimuli therefore produced medial (3) or lateral (4) acceleration of the ipsilateral labyrinth. Stimulation with a tendon hammer was not attempted due to the large number of stimuli required. The mastoid stimuli were applied directly posterior to the external auditory meatus and for all sine waves the initial polarity was toward the mastoid. The intensities of the stimuli were not matched, but were set to the levels used in previous reports to mimic the typical conditions for each stimulus. For the BC mastoid (B-71) stimulus this was 136 dB peak FL (6.3 N peak). Our BC forehead (minishaker) stimulus produced an initial acceleration peak that was largest and most consistent in the interaural (y) axis (a ‘bowing out’ of the mastoids of 0.1 g), with smaller peaks in inconsistent directions in the x (0.06 g) and z (0.03 g) axes (Weber et al., 2012). For the lateral minishaker taps, the peak mastoid acceleration was approximately 0.14 g at 3.2 ms (Todd et al., 2008). All stimuli were generated with customized software using a laboratory interface (micro1401, Cambridge Electronic Design (CED)) and delivered at a rate of 7.5 Hz up to a maximum of 2000 repetitions per trial (typically 1300 per trial). 2.3. Single motor unit and surface EMG recordings Subjects were usually tested first using surface electrodes to confirm the presence of cVEMPs and determine the best side for stimulation during single unit trials. For most single motor unit recordings subjects sat upright and turned their heads away from the recorded side. Subjects were instructed to use their hand to push gently against their chin to produce a controlled contraction of the SCM. In this position a weak SCM contraction could be easily maintained for the duration of each trial (about 3–5 min). For some early recordings subjects reclined to approximately 30° above horizontal and lifted their head to activate the SCM muscle. This position produced similar results, but the muscle contraction was more difficult to control. Recordings were made in either the left or right SCM according to subject preference (or better ear) and were always ipsilateral to the ear or mastoid stimulated. An active surface electrode (Blue sensor N, Ambu A/S) was placed near the needle electrode and referred to an electrode on the medial

clavicle. An earth was placed on the sternum or lateral clavicle. Negative potentials at the active electrodes were displayed as upward deflections. We used ultra thin (0.3 mm diameter, 30G) disposable concentric needles (Neuroline, Ambu A/S) to record single motor units. The skin was first prepared with alcohol (Kodan Tinktur, Schülke & Mayr GmbH, Norderstedt, Germany). The needle was inserted near the middle or upper third of the SCM belly and held in position with tape during each recording. Audio feedback was provided to help subjects maintain constant activation of one or more discernible motor units with a flat baseline in between. In some cases visual feedback was also provided. EMG was recorded with the same micro1401 data acquisition interface and custom software as described above. Data were sampled (1902 mk4 amplifiers, CED) at 50 kHz for 100 ms (from 40 ms before to 60 ms following stimulus onset), amplified and bandpass filtered (5 Hz–10 kHz). 2.4. Data analysis Motor unit spikes were identified using a threshold level and clustered with custom software (Matlab, The MathWorks Inc.) based on an automatic algorithm using wavelets and super-paramagnetic clustering (Quiroga et al., 2004). In some trials a single unit was recorded, while in others there were multiple units, from which single units were extracted using the sorting algorithm (Fig. 1). Peri-stimulus time histograms consisting of 100 bins of 1 ms width (centered at whole numbers) were constructed for each unit. The number of spikes in each single unit histogram ranged from 407 to 2542 (mean 1162). The cluster software was set to maximize specificity rather than sensitivity in order to ensure that single unit clusters were not contaminated with extraneous spikes. The accuracy of the software was confirmed for each single unit by visual comparison of the selected spikes with the raw data. For recordings in which spikes from a single unit could not be reliably differentiated, multiple unit histograms were constructed to show the behavior of several units recorded simultaneously. The minimum number of spikes in a histogram was set to 400 to ensure that each histogram had sufficient data. This is important as the cVEMP is typically inhibitory when measured in the ipsilateral SCM and sufficient spikes in the baseline period are required in order to detect a gap in firing. Based on the spike count measured over the 40 ms pre-stimulus period the 2.5 and 97.5 quantiles were determined and used as inhibitory and excitatory threshold levels. Local maxima (peaks) and minima (troughs) were accepted as significant changes in motor unit activity if they exceeded these limits and occurred after the first 8 post-stimulus bins. We excluded histograms in which the 2.5th quantile extended below 1 spike to ensure that exceeding the threshold for decreases in activity was always possible (9 histograms). Using the above criteria, the rate of false positives detected during the pre-stimulus period was 3.8%, i.e. close to the 5% expected by chance. After the stimulus, there appeared to be a false-positive, single-bin increase or decrease in activity (peak or trough) preceding a clearer significant response in 35 histograms. To check that these responses did not exceed the number expected by chance, we calculated that they represented a false-positive rate of 3.5% over the interval from 9 ms post stimulus (from the first bin in which a response was accepted) to 16.5 ms post stimulus (the mean latency of the initial response). This interval was chosen to account for the fact that a false positive could occur at any time within the latency range of initial responses (9–25 ms) but would only precede the first cVEMP response approximately half of the time. This rate was similar to the rate for the pre-stimulus period and thus these responses were ignored. To quantify the motor unit responses in the histograms, the amplitude of each change in unit activity was expressed as a pro-

1237

S.M. Rosengren et al. / Clinical Neurophysiology 126 (2015) 1234–1245

A

E

Needle EMG

Raster plot

2000

1095

1800 1600 1090 1400

Trials

1200 1085

1000 800 600

1080 400 200 1075 -40

-20

B

40

C

0

1

Spike count

Amplitude [mV]

0

-1

D

0

1

0

20

60

40

60

40

60

30 20 10 -20

H

Unit 2b

0

20

Unit 2b 40

Spike count

Amplitude [mV]

-20

Unit 2a

0 -40

2

0.5

0

-0.5 -2

40

10

40

0.5

-0.5 -2

60

20

G

Unit 2a

40

30

0 -40

2

0 20 Time [ms]

Unit 1

40 Spike count

-1

-20

F

0

-0.5 -2

0

60

Unit 1

0.5 Amplitude [mV]

0 20 Time [ms]

-1

0 Time [ms]

1

2

30 20 10 0 -40

-20

0 20 Time [ms]

Fig. 1. Single motor unit recording from the sternocleidomastoid muscle in response to BC stimulation at the mastoid. (A) Needle EMG recording showing 20 consecutive trials out of 2000 stimuli. The stimulus was a 4 ms burst of 500 Hz vibration delivered to the mastoid with a B-71 bone conductor. Three distinct motor units can be seen: unit 1 in blue, unit 2a in red and its satellite unit 2b in orange, which always appears nearly 10 ms following the red unit. The black vertical line indicates stimulus onset. (B–D) The two distinct single motor units and the satellite unit were aligned to their peaks and identified with a sorting algorithm using wavelets and super-paramagnetic clustering (Quiroga et al., 2004). (E) Raster plot of the two independent single motor units over all 2000 stimuli. Spikes from the two units are randomly distributed over the prestimulus interval ( 40 to 0 ms). After the stimulus (0–60 ms) a gap in firing can be seen at about 13 ms. The horizontal gray band represents the 20 consecutive trials illustrated in A. (F–H) Peri-stimulus time histograms for each unit illustrate a gap or decrease in discharge at 13 ms for the blue and red units and a decrease in firing for the satellite at 22 ms (reflecting the delay in firing seen between this unit and its satellite in part A). The black bars indicate the duration of the stimulus.

portion of the mean spike count calculated over the 40 ms prestimulus period. The latency of a change in activity was taken at the first bin to exceed the 2.5 or 97.5 quantiles. The duration of a change in firing was measured by counting the consecutive bins that exceeded threshold.

For the surface recordings, amplitudes and latencies were measured at each of the response peaks. Peaks occurring after the initial vestibular-dependent response were also measured for the vibration stimuli. Latencies were adjusted by 0.5 ms to correct for a fixed delay in the recording system (1902 digital filtering).

1238

S.M. Rosengren et al. / Clinical Neurophysiology 126 (2015) 1234–1245

Table 1 Characteristics of single motor unit responses evoked by each stimulus. Stimulus

Polarity

Latency mediana

Latency minimum

Latency maximum

Amplitude medianb

Amplitude minimum

Amplitude maximum

N responsec

N totald

Initial change in activity (p/n13) AC sound Decrease BC mastoid Decrease BC forehead Decrease Tap inward Decrease Tap outward Increase

14.5 15.0 15.5 14.0 18.5

11 10 9 10 10

19 24 24 25 25

0.08 0.14 0.05 0.00 1.93

0.0 0.0 0.0 0.0 1.6

0.5 0.7 0.4 0.3 4.5

18 35 22 23 14

25 39 24 24 21

Second change in activity (n/p23) AC sound Increase BC mastoid Increase BC forehead Increase Tap inward Increase e Tap outward Decrease Increase

22.0 23.5 22.0 18.0 22.0 32.0

16 14 14 14 14 23

29 29 31 32 28 44

1.86 1.84 1.82 2.40 0.00 2.08

1.6 1.4 1.5 1.7 0.0 1.4

3.0 4.7 2.6 4.8 0.2 2.9

7 16 9 15 21 12

18 35 22 23 21 21

29 24 27 41

55 45 57 53

2.17 2.18 2.13 2.26

1.5 1.6 1.6 1.7

2.9 3.3 4.1 2.7

14 15 20 11

35 22 23 21

Third change in activity (possible stretch reflex) BC mastoid Increase 39.0 BC forehead Increase 32.0 Tap inward Increase 38.0 Tap outward Increase 49.0 a b c d

e

Latency in ms. Amplitude is expressed as a proportion of median baseline activity. Number of single unit histograms showing a significant response to the stimulus. Total number of single unit histograms. Second and third changes in activity were counted only for histograms that showed a significant initial response (e.g. for AC sound in 7 of the 18 histograms that had a significant decrease in activity, this decrease was followed by a significant increase in activity). The second response for the outward tap consisted of two changes in activity and both are described here.

The strength of the background SCM contraction was measured offline by full wave rectifying and then averaging the surface recordings from each frame. The background SCM contraction was measured from the mean rectified EMG over the 40 ms prestimulus interval. 2.5. Statistical analysis There were variable numbers of single units recorded across subjects. As we were interested in the spread of values across all observations, the descriptive statistics in Table 1 include all of the data. In contrast, to prevent subjects with greater numbers of observations unduly influencing the comparisons, inferential statistical analysis was performed on averaged data for each subject. The significance level was set at a = 0.05. We reported medians and ranges and used non-parametric statistics for independent samples (i.e. the Kruskal–Wallis H test). This test was chosen due to the small number of subjects tested (N = 8), some positive skew in the data and because there were some missing values. Post hoc tests were performed with a Mann–Whitney U test. We correlated the single and multiple unit response amplitudes and durations with the surface potentials. For these analyses the surface response amplitude was measured using the ratio of p13–n23 peak-to-peak amplitude to background contraction, i.e. as in clinical cVEMP studies; Rosengren et al., 2010. 3. Results The polarity of the initial change in single motor unit activity in the histograms reflected the cVEMPs recorded with surface electrodes. An initial decrease in activity (and positive surface potential, i.e. p13) was seen in the ipsilateral SCM in response to stimulation with AC sound, the BC mastoid and forehead stimuli and inwardly-directed mastoid taps, while an initial increase in activity (and negative surface potential) was produced by outwardly-directed mastoid taps. Typical surface and single unit responses from each stimulus in a single subject are shown in Fig. 2.

All of the subjects had surface cVEMPs in response to each type of stimulus, except two subjects who had very high thresholds to AC sound and one subject who had a poor response to BC forehead stimulation. The surface responses had the same appearance as those reported previously using the same stimuli, but were smaller than usual, around 30 lV peak-to-peak (p13–n23), because the strength of the background SCM contraction was kept relatively weak to enable tracking of only one or more single units. The background contraction did not differ between stimuli (mean 37 ± 15 lV, range 15–80 lV over all trials, H(4) = 5.5, P = 0.239). Despite their small amplitude, the cVEMPs were very well defined because the responses to more than 1000 stimuli were averaged together, resulting in a high signal-to-noise ratio. We analyzed the data from each trial in which there was a clear surface response, giving a total of 134 single units across 124 trials and 24 recording sessions. There were 43 trials in which one unique single unit could be identified, 32 trials in which 2 separate units were extracted and 9 trials in which 3 units were extracted (total = 134 units). There were an additional 40 trials in which no single unit could be extracted, and from these recordings multiple unit histograms were constructed. The results from all subjects are summarized in Tables 1 and 2.

3.1. Initial change in activity In the majority of cases (119/134 single unit histograms) there was a short-latency change in SCM activity following the vestibular stimulus. This change was a decrease in activity at around 15 ms for all stimuli except the outward tap, which showed an increase in activity. A decrease in motor unit activity was always associated with a positive surface response (i.e. p13 potential) and an increase in activity with a negative surface response. There were no significant differences in the median latency of the initial change for either the surface responses (H(4) = 3.7, P = 0.455) or the single unit responses (H(4) = 5.6, P = 0.227) for the different stimuli. There was greater variation in single unit response latency (ranging from 9 to 25 ms) compared to the surface responses (10–20 ms), reflecting the fact that the surface response represents the combined activity

1239

S.M. Rosengren et al. / Clinical Neurophysiology 126 (2015) 1234–1245

A

AC Sound 25 Surface cVEMP Needle EMG

15 0 10 5 0 −40

−10 −20

B

0

20

BC Mastoid

40

n23

60

late peak 10

30 Spike count

Amplitude [µV]

10

20

p13

0

10

Amplitude [µV]

Spike count

20

−10 0 −40

C

−20

0

20

40

60

BC Forehead 30 Amplitude [µV]

Spike count

100 20 0 10 −100 −20

0

D

20

40

60

Inward Tap

50

Spike count

40 30

0

20 10 0 −40

Amplitude [µV]

0 −40

−50 −20

E

0

20

40

60

Outward Tap n13

50

30 p23

late peak 0

20 10

Amplitude [µV]

Spike count

40

−50 0 −40

−20

0

Time [ms]

20

40

60

Fig. 2. Single motor unit histograms from a single subject showing a typical response to each type of stimulus. The stimuli were: Sine waves of 500 Hz and 4 ms duration delivered with (A) headphones (‘AC sound’), (B) a B-71 bone conductor applied to the mastoid (‘BC mastoid’) and (C) a minishaker applied to the forehead posterior to the hairline (‘BC forehead’); and lateral taps delivered to the mastoid with a minishaker, which accelerated the mastoid medially (D, ‘Inward tap’) or laterally (E, ‘Outward tap’). Each histogram shows a significant short-latency change in activity compared to the baseline pre-stimulus period. Simultaneously-recorded surface responses are shown superimposed and their amplitude is indicated in lV on the right. For the stimuli A–D the surface responses show a typical positive–negative (p13–n23) cVEMP and the histograms show an initial gap or decrease in single motor unit activity. For stimulus E, the outward-directed tap, the surface response is inverted (negative–positive polarity) and the single motor unit shows an initial increase in activity. The responses shown in parts D and E are from the same single motor unit that was stimulated with inward and outward taps over consecutive recordings. Oppositely-directed mediolateral acceleration produced responses with the same latency but opposite polarity.

1240

S.M. Rosengren et al. / Clinical Neurophysiology 126 (2015) 1234–1245

Table 2 Characteristics of surface responses evoked by each stimulus. Latency minimum

Latency maximum

Amplitude medianb

Initial change in activity (p/n13) AC sound Positivity 14.1 BC mastoid Positivity 12.9 BC forehead Positivity 15.3 Tap inward Positivity 13.4 Tap outward Negativity 13.2

10.0 11.0 13.3 12.5 12.0

16.6 20.3 18.5 17.3 14.9

8.9 6.3 17.4 13.2 8.3

2.4 1.3 2.7 3.7 4.0

20.4 18.6 55.0 29.7 28.4

Second change in activity (n/p23) AC sound Negativity 25.9 BC mastoid Negativity 24.7 BC forehead Negativity 24.8 Tap inward Negativity 21.2 c Tap outward Positivity 21.0 Negativity 32.1

22.5 20.7 20.8 19.5 17.0 25.2

28.3 30.6 28.3 25.0 24.5 37.2

9.9 9.5 29.5 20.4 30.8 41.4

3.2 3.6 7.6 1.9 7.7 3.7

31.0 20.8 96.1 48.0 92.3 81.4

Third change in activity (possible stretch reflex) BC mastoid Negativity 31.6 BC forehead Negativity 29.8 Tap inward Negativity 36.3 Tap outward Negativity 45.9

28.4 28.9 27.1 32.9

58.4 31.5 47.4 48.3

8.1 51.0 26.0 26.4

1.0 30.8 3.8 4.1

25.7 75.5 78.1 60.7

Stimulus

a b c

Polarity

Latency mediana

Amplitude minimum

Amplitude maximum

Latency in ms. Amplitude in lV. The second response for the outward tap consisted of two changes in activity and both are described here.

of many units with different latencies. The latency range of the surface cVEMPs was also greater than usual because the surface electrode was often placed away from the motor point to allow sufficient space for the needle electrode. As the stimuli each had different intensities, the number and size of responses differed between stimuli, even after the opposite polarity of the outward tap was taken into account. The strongest decrease in activity was seen with the inward tap, to which nearly all units responded and which completely abolished activity at around 14 ms in 70% of units. The weakest response occurred following the BC mastoid stimulus, where activity was abolished in only 17% of units. There was a trend toward a significant difference in surface p13 amplitude (H(3) = 7.0, P = 0.071) and a difference in the size of the single unit response across stimuli (H(3) = 11.0, P = 0.012), after excluding the outward tap stimulus. Post hoc comparisons suggested that this was due to the BC mastoid stimulus producing smaller responses than the other stimuli (significant only compared to the inward tap stimulus for the single unit data, U = 12.5, P = 0.006). The effect of changing stimulus intensity can be seen in Fig. 3, where two intensities of BC mastoid stimulus were compared in the same single unit. The maximal stimulus caused a gap in firing at 12 ms and produced a well-formed surface response. Decreasing the intensity by 10 dB eliminated the response from this single unit and decreased the overall surface response. The residual surface response indicates that the stimulus remained above threshold for other nearby motor units. The outward tap produced an initial increase in firing at short latency, opposite to the inward tap and consistent with the oppositely-directed surface responses to these stimuli. The median amplitude of this response was nearly double the baseline activity (1.93, ranging from 1.6 to 4.5 times the baseline activity). Fig. 2 (parts D and E) shows an example of the same single unit stimulated with both inward and outward taps over consecutive recordings. The initial peaks had opposite polarity but identical latency. The duration of the initial change in activity ranged from 1 to 8 ms and differed between stimuli (H(4) = 19.3, P = 0.001). The median duration was similar for the decreases in firing produced by the AC sound (3 ms), BC mastoid (2 ms) and BC forehead (3 ms) stimuli. It was slightly longer for that produced by the inward tap stimulus (4 ms; U = 11.7, P = 0.028 and U = 10.9, P = 0.024, compared to AC sound and BC mastoid, respectively).

In contrast, the duration of the increase in firing produced by the outward tap (1 ms) was significantly shorter than the initial response to the inward tap (U = 20.8, P < 0.001). For the outward tap, the decrease in activity following the initial increase was longer than the other early troughs (6 ms, H(4) = 11.3, P = 0.023). The amplitude and duration of the initial decrease in single unit motor activity were significantly correlated with each other (r = 0.40, P < 0.001), i.e. larger inhibitions had longer duration. However, neither measure was correlated with the size of the surface response (r = 0.15, P = 0.249, and r = 0.22, p = 0.087, respectively). 3.2. Second change in activity For all stimuli the earliest response was often closely followed by a change in firing with the opposite polarity. For most of the stimuli this was an increase in firing at about 22 ms, corresponding to the n23 surface potential of the cVEMP. This increase in activity exceeded the threshold in about half of all units and ranged in size from 1.4 to 4.8 times the mean baseline activity of the unit (Table 1). For the outward tap the pattern was again reversed, with the initial increase followed by a suppression of activity at a similar median latency of 22 ms. Following this suppression, a similar n23like increase in activity was seen, albeit at a correspondingly longer latency of 32 ms. When these increases in activity were compared across stimuli, there were no significant differences in size or latency, except that caused by the delayed response to the outward tap (amplitude compared across all stimuli H(4) = 2.8, P = 0.592; latency across all stimuli H(4) = 11.1, P = 0.026; latency excluding outward tap H(3) = 1.0, P = 0.805). 3.3. Third change in activity The BC stimuli all also produced an additional, late increase in activity that could be identified for about half of the units (Tables 1 and 2). This response occurred between 25 and 57 ms and had similar median amplitude to the second increase in activity reported above (1.5–4.1 times the mean baseline activity of the unit). There were no significant differences between the stimuli, except the longer median latency produced by the outward tap stimulus (amplitude compared across all stimuli H(3) = 2.2, P = 0.532; latency across all stimuli H(3) = 9.3, P = 0.025; latency

1241

S.M. Rosengren et al. / Clinical Neurophysiology 126 (2015) 1234–1245

Maximum Intensity

A

30

40

0 10

Amplitude [µV]

Spike count

20 20

−20

0 −40

−40 −20

B

0

20

40

60

−10dB Intensity 20

40 Surface cVEMP Needle EMG

10

0

Amplitude [µV]

Spike count

20

−20

0 −40

−40 −20

0

20

40

60

Time [ms] Fig. 3. The effect of stimulus intensity on single motor unit activity. A single motor unit was recorded over consecutive trials using the BC mastoid stimulus at maximal intensity (A, 136 dB peak FL) and 10 dB softer (B, 126 dB peak FL). The maximal stimulus produced a gap in firing at 12 ms and a clear surface response, while the softer stimulus abolished the response in this single unit and decreased the overall surface response. The residual surface response indicates that the stimulus remained above threshold for other nearby motor units. The strength of SCM muscle contraction was similar in both recordings (30.6 and 29.5 lV, respectively).

excluding outward tap H(2) = 4.0, P = 0.134). As can be seen in Fig. 4, the second and third changes in activity typically had a distinct peak in the single unit histograms, but sometimes overlapped in the surface recordings.

unit activity (r = 0.55, P = 0.001). However, the n23 surface peak was not correlated with the amplitude of the second change in activity (r = 0.26, P = 0.149), but was instead significantly correlated with the initial change in activity (r = 0.60, P = 0.001).

3.4. Multiple unit responses 4. Discussion As multiple unit histograms were only constructed for recordings in which no single units could be extracted, the numbers were small, the distribution was not even across stimuli and thus the results were not analyzed in detail. The 40 multiple unit histograms were similar to the corresponding single unit and surface responses, though they generally had a smoother appearance, with broader changes in activity (Fig. 5). Most multiple unit histograms (39/40) showed the same early change in activity, 19/40 showed the n23-like increase in activity and 23/40 showed the later increase in activity. The multiple unit histograms better matched the shape of the surface response, at least for the initial peak, as multiple unit histograms and surface recordings both record the combined activity of many units recorded simultaneously. Consistent with this, both the amplitude and duration of the initial decrease in motor activity seen in the multiple unit histograms were significantly correlated with the amplitude of the p13–n23 surface response (amplitude r = 0.66, P < 0.001, and duration r = 0.69, P < 0.001). Multiple unit histograms with larger or longer initial decreases in activity (i.e. stronger effects) were found in recordings with larger surface cVEMPs. In addition, the p13 surface peak amplitude was significantly correlated with the amplitude of the initial change in motor

4.1. First change in activity Our results demonstrate that bone-conducted vestibular stimulation usually produces an initial decrease in single motor unit activity in the ipsilateral SCM muscle. This decrease or gap in firing was seen for all stimuli that produced an early positive (p13) surface cVEMP, while the outward mastoid tap produced a negative surface cVEMP and an initial increase in motor unit activity. We are confident that these early single unit responses are vestibular-dependent because the surface cVEMPs produced by these stimuli have been independently shown to depend upon vestibular function (Brantberg and Tribukait, 2002; Manzari et al., 2010; Rosengren et al., 2009; Welgampola et al., 2003). In addition, in the current study, subjects who had absent surface cVEMP waveforms also showed no early change in single or multiple unit histograms (e.g. Fig. 5C). The measured responses are likely to originate in the otolith organs, as animal evidence suggests that skull vibration selectively activates irregularly-firing otolith afferents (Curthoys et al., 2006). A contribution from semicircular canal afferents, however, cannot be ruled out because there is also some evidence of canal activation with vibration (Zhu et al., 2011) and

1242

S.M. Rosengren et al. / Clinical Neurophysiology 126 (2015) 1234–1245

Separated Peaks 15

50

Spike count

Surface cVEMP Needle EMG

*

10 0 5

0 −40

Amplitude [µV]

A

−50 −20

B

0

20

40

60

Merged Peaks 30

*

80

40

0 10

Amplitude [µV]

Spike count

* 20

−40

0 −40

−80 −20

0

Time [ms]

20

40

60

Fig. 4. Differences between the second and third changes in motor unit activity in the surface recordings and histograms. Two single motor units from a single subject stimulated with the ‘BC mastoid’ stimulus are shown. The surface recording for the motor unit in part A shows a typical cVEMP, in which the p13, n23 and non-vestibular peaks can be clearly distinguished. In contrast, while the histogram for the unit in part B shows two distinct increases in activity (shown by asterisks), the corresponding peaks in the surface recording are merged. The n23 in this case is probably the ‘shoulder’ indicated by the arrow.

some of the stimuli may have caused a small amount of rotation of the head. Single unit recordings are important because they unambiguously show the behavior of the target muscle, whereas potentials recorded with surface electrodes depend upon electrode placement and can be influenced by the activity of nearby muscles. The direction of the initial change in muscle activity indicates the polarity of the earliest projection underlying the reflex: a decrease in activity corresponds to an inhibitory projection from the vestibular organs to the SCM muscle, while an increase in activity corresponds to an excitatory projection. Although the nature of our data does not allow us to distinguish between excitation and disinhibition, inhibition and disfacilitation, we will use this terminology for simplicity and consistency with other vestibular studies. Previous studies used this technique to investigate the nature of vestibulo-collic and vestibulo-masseteric projections evoked by AC sound and galvanic stimulation (Colebatch and Rothwell, 2004; Deriu et al., 2003, 2005, 2007). In the current study, we applied this technique to investigate the vestibulo-collic projections evoked by BC vestibular stimulation. Our data show that the projections to the SCM muscles evoked by BC stimulation in humans are not exclusively inhibitory. By changing the direction of head acceleration, the polarity of the motor unit response could be reversed, while the latency was unchanged. This has been proposed by previous authors based on the results of surface studies (e.g. Brantberg and Tribukait, 2002; Rosengren et al., 2009; Westin and Brantberg, 2014), but shown definitively here using direct recordings from the muscle per se.

There was very good concordance between the polarity of the single motor unit responses and the surface waveforms, suggesting that, when the traditional SCM belly-tendon cVEMP electrode montage is used, the initial surface responses are likely to be a reliable indicator of the polarity of the reflex in the SCM muscle. The latencies were also similar in the needle and surface recordings, though the range was always greater for the single motor unit responses, similar to that reported for AC sound- and galvanicevoked responses (Colebatch and Rothwell, 2004). This is partly because the surface electrodes were not always placed directly over the needle electrodes, but predominantly because the surface response represents the sum of electrical activity from the underlying muscle fibres and effectively averages out the temporal dispersion of many single motor unit responses. For this reason the multiple unit histograms more closely matched the shape of the initial change in activity in the surface response. The amplitudes of the single unit and surface responses were not correlated, due to the large variability in single unit behavior, however the size and duration of response in the multiple unit histograms predicted the size of the p13–n23 surface response. The median increase in firing (1.93) for the outward tap was greater than the maximum inhibitory effects for the other stimuli. This is because there is no upper limit to the amplitude of an excitation, while an inhibition is inherently limited to a pause in firing and the size is therefore related to the tonic level of single unit activity. This illustrates why excitatory surface potentials are commonly larger than inhibitory ones (Rosengren et al., 2008). Finally, the shorter average duration of the initial increase in firing compared to the initial decrease in firing evoked by the tap stimuli probably explains

1243

S.M. Rosengren et al. / Clinical Neurophysiology 126 (2015) 1234–1245

Strong Response

A

50 Surface cVEMP Needle EMG

30 0

20

Amplitude [µV]

Spike count

40

10

0 −40

−50 −20

B

0

20

40

60

Moderate Response 50

150 0 100

Amplitude [µV]

Spike count

200

50

0 −40

−50 −20

0

20

40

60

Absent Response

C

50

100 0 50

0 −40

Amplitude [µV]

Spike count

150

−50 −20

0

20

40

60

Time [ms] Fig. 5. Multiple unit histograms and surface responses from three subjects. Three multiple unit histograms are shown to illustrate strong (A), moderate (B) and absent (C) responses (using the ‘AC sound’ stimulus). The strong cVEMP in part A had an amplitude of 1.0 and the moderate one in part B had an amplitude of 0.4 (i.e. ratio of peak-to peak amplitude to background contraction). As multiple unit histograms contain data from many single units with different temporal characteristics, they have a broader shape than single unit histograms and better match the strength of the surface response.

the characteristically shorter latency for the initial surface negativity peak compared to the initial positive one, as well as the failure to demonstrate the surface negativity in some subjects (Rosengren et al., 2009). It is likely that relatively brief excitability changes are better detected using single unit studies than surface recordings. The laterality of the reflex is more difficult to determine due to the bilateral nature of bone-conducted stimulation in normal subjects. Previous studies describing surface potentials evoked by these stimuli in patients with unilateral dysfunction of the vestibular organs can help provide this information. For the BC mastoid, BC forehead and inward tap stimuli, prior studies show that a normal p13–n23 response is seen in the SCM ipsilateral to

the intact (stimulated) ear, suggesting the existence of an ipsilateral inhibitory projection to the SCM activated by these stimuli (Manzari et al., 2010; Rosengren et al., 2009; Welgampola et al., 2003). For the outward tap, evidence for the laterality of the early excitation is less clear. Rosengren et al. (2009) recorded a negative (excitatory) surface peak in only some normal subjects, while the subsequent positive (inhibitory) response at 20 ms was seen in all subjects. This positivity was reduced in size with outward stimulation of both the intact and affected mastoids in the patients, suggesting that it receives contributions from both ears (Rosengren et al., 2009). Such a pattern was reproduced in a group of patients with vestibular neuritis, though

1244

S.M. Rosengren et al. / Clinical Neurophysiology 126 (2015) 1234–1245

the lesions in these patients were not complete (Govender et al., 2011). A bilateral projection might explain the long duration (6 ms) of the corresponding inhibition recorded in the single units in this study. Additional evidence regarding the negative surface potential comes from Brantberg and Tribukait (2002), who applied similar stimuli to the sides of the head above the mastoids in patients with unilateral vestibular loss. They found no negative–positive response in the SCM ipsilateral to the intact ear, suggesting that any excitatory projection might originate in the contralateral ear. This pattern of results would be consistent with previous research showing that AC sound and galvanic stimulation evoke inhibitory responses in the SCM ipsilateral to the headphone/cathode and excitatory responses in the opposite SCM (Colebatch and Rothwell, 2004). It is also consistent with prior animal research using whole-nerve electrical stimulation, which has shown that all vestibular end organs have an inhibitory projection to the ipsilateral SCM and all except the saccule have an additional excitatory projection to the contralateral SCM (see Uchino and Kushiro, 2011 for review). It is therefore not likely that the excitatory responses originate from the saccule, as no excitatory projection to the SCM from this organ has been demonstrated. However, caution is warranted, as the results of animal experiments might not generalize to humans. Also, the otoliths consist of two halves containing oppositely-directed hair cells, suggesting that one ear might theoretically be capable of producing both inhibition and excitation. It is not known exactly which of these hair cells are activated by these stimuli and it is possible that different results might occur with alternate forms of stimulation.

practice of measuring the peak-to-peak (i.e. p13–n23) amplitude of the surface cVEMP. We found that both the p13 and n23 peaks were strongly correlated with the inhibition of SCM motor unit activity. Measurement of surface responses will be more robust if two peaks are used.

4.3. Third change in activity The BC stimuli also often evoked a third increase in motor unit activity, which corresponded to the late negative surface response. This peak was quite variable and detected in only about half of all single units. The late surface response was described in the earliest BC cVEMP study and is not vestibular-dependent (Halmagyi et al., 1995), but is thought to be caused by activation of neck stretch receptors. Unlike the preceding vestibular-dependent responses, the polarity did not reverse between the inward and outward taps, although the response became later (Tables 1 and 2). Our results highlight how difficult it can be to distinguish this peak from the n23 potential in surface recordings (Fig. 4). The two excitations are usually seen as clearly separate peaks in the single motor unit histograms, but in the surface responses, where many single units with different latencies contribute to the waveform, the responses are often blurred. This can be problematic in clinical settings when BC stimuli are used, as correct identification of the n23 peak, and therefore accurate measurement of cVEMP amplitude and asymmetry, can be impossible in some subjects (Halmagyi et al., 1995). We suggest that caution is required when measuring BC cVEMPs, especially when the n23 peak has a longer latency or is larger than usual (as in Fig. 4B).

4.2. Second change in activity Following the initial inhibition of muscle activity, we saw a significant increase in firing in about half of the single and multiple unit histograms, suggestive of the n23 surface response. However, the observation of this response in the needle recordings was inconsistent compared to the surface recordings, in which a clear negativity (n23) was seen in all cases. The single motor unit response at approximately 23 ms is unlikely to be caused by a separate excitatory projection from the vestibular organs, but instead probably represents the recovery of activity following the period of suppression, i.e. for some units firing was simply delayed by the inhibitory stimulus. This increase in activity probably contributes to the n23 surface response, but cannot completely account for it. Our data demonstrate that the n23-like increase in motor unit activity was not correlated with the amplitude of the surface n23 response. Instead, the early inhibition predicted the size of both the p13 and n23 surface potentials. This finding is consistent with a recent mapping study of the surface cVEMP, which found that the n23 potential behaved like a ‘standing wave’ produced by the inhibitory signal reaching the muscle–tendon junction (Colebatch, 2012). In that study the p13 increased in latency with increasing distance of the recording electrode from the motor point, representing the progressive inhibition of motor units along the muscle toward the tendons. In contrast, the n23 potential latency remained relatively constant, suggesting that it was produced by the momentary dipole generated by the arrival of the inhibitory response at the muscle–tendon junction, i.e. by a terminal standing wave (Lateva et al., 1996). The latency of this standing wave represents the time it takes for the response to reach the tendon. The response occurs at a fixed location and is seen remotely by the recording electrodes. It thus characteristically does not change latency with changing electrode position (Dumitru and King, 1991; Lateva et al., 1996). Our results therefore support the hypothesis that the n23 surface potential is caused predominantly by a standing wave with a contribution from increased motor unit activity. Importantly, this finding should not change the accepted

4.4. Clinical significance Bone-conducted stimulation is a useful cVEMP stimulus in clinical settings, as many patients have conductive hearing loss, which renders air-conducted sound stimulation ineffective and thus requires an alternate stimulus that bypasses the middle ear conductive apparatus. Although BC stimuli are likely to activate different, though probably overlapping, subsets of vestibular afferents, they provide an alternate means of testing the otoliths and vestibulo-collic reflex in patients. As BC stimuli become a more popular choice, it is important to understand their properties. Our results confirm that the BC cVEMP changes with different directions of stimulation and is not an exclusively inhibitory reflex. The utricle is likely to contribute to some vibration cVEMPs, as some stimuli clearly evoke an excitation, which is unlikely to be saccular in origin. Our comparison of surface and single unit responses shows that the surface waveform provides a good indication of the polarity of the initial reflex, while later changes in activity do not have a one-to-one relationship with the surface response.

Acknowledgements We are grateful to R. Quian Quiroga for providing open access to the code of his spike detection and sorting algorithm. Dr Sally Rosengren was supported by the National Health and Medical Research Council of Australia (GNT1058056). The study was supported by the Neuro-Otology Society of Australia, the Swiss National Science Foundation, the Betty and David Koetser Foundation for Brain Research and the Zurich Center for Integrative Human Physiology (University of Zurich). Conflict of interest statement: None of the authors have potential conflicts of interest to be disclosed.

S.M. Rosengren et al. / Clinical Neurophysiology 126 (2015) 1234–1245

References Bath AP, Harris N, McEwan J, Yardley MP. Effect of conductive hearing loss on the vestibulo-collic reflex. Clin Otolaryngol 1999;24:181–3. Brantberg K, Löfqvist L, Westin M, Tribukait A. Skull tap induced vestibular evoked myogenic potentials: an ipsilateral vibration response and a bilateral head acceleration response? Clin Neurophysiol 2008;119:2363–9. Brantberg K, Tribukait A. Vestibular evoked myogenic potentials in response to laterally directed skull taps. J Vestib Res 2002;12:35–45. Brantberg K, Tribukait A, Fransson PA. Vestibular evoked myogenic potentials in response to skull taps for patients with vestibular neuritis. J Vestib Res 2003;13:121–30. Brantberg K, Westin M, Löfqvist L, Verrecchia L, Tribukait A. Vestibular evoked myogenic potentials in response to lateral skull taps are dependent on two different mechanisms. Clin Neurophysiol 2009;120:974–9. Cai KY, Rosengren SM, Colebatch JG. Cervical and ocular vestibular evoked myogenic potentials are sensitive to stimulus phase. Audiol Neurootol 2011;16:277–88. Colebatch JG. Mapping the vestibular evoked myogenic potential (VEMP). J Ves Res 2012;22:27–32. Colebatch JG, Halmagyi GM. Vestibular evoked potentials in human neck muscles before and after unilateral vestibular deafferentation. Neurology 1992;42:1635–6. Colebatch JG, Halmagyi GM, Skuse NF. Myogenic potentials generated by a clickevoked vestibulocollic reflex. J Neurol Neurosurg Psychiatry 1994;57:190–7. Colebatch JG, Rothwell JC. Motor unit excitability changes mediating vestibulocollic reflexes in the sternocleidomastoid muscle. Clin Neurophysiol 2004;115:2567–73. Curthoys IS, Kim J, McPhedran SK, Camp AJ. Bone conducted vibration selectively activates irregular primary otolithic vestibular neurons in the guinea pig. Exp Brain Res 2006;175:256–67. Deriu F, Ortu E, Capobianco E, Giaconi E, Melis F, Aiello E, Rothwell JC, Tolu E. Origin of sound-evoked EMG responses in human masseter muscles. J Physiol 2007;580:195–209. Deriu F, Tolu E, Rothwell JC. A short latency vestibulomasseteric reflex evoked by electrical stimulation over the mastoid in health humans. J Physiol 2003;553:267–79. Deriu F, Tolu E, Rothwell JC. A sound-evoked vestibulomasseteric reflex in healthy humans. J Neurophysiol 2005;93:2739–51. Dumitru D, King JC. Far-field potentials in muscle. Muscle Nerve 1991;14:981–9. Govender S, Rosengren SM, Colebatch JG. Vestibular neuritis has selective effects on air- and bone-conducted cervical and ocular vestibular evoked myogenic potentials. Clin Neurophysiol 2011;122:1246–55. Halmagyi GM, Yavor RA, Colebatch JG. Tapping the head activates the vestibular system: a new use for the clinical reflex hammer. Neurology 1995;45:1927–9. Huang C-H, Wang S-J, Young Y-H. Localization and prevalence of hydrops formation in Ménière’s disease using a test battery. Audiol Neurootol 2011;16:41–8. Lateva ZC, McGill KC, Burgar CG. Anatomical and electrophysiological determinants of the human thenar compound muscle action potential. Muscle Nerve 1996;19:1457–68. Manzari L, Burgess AM, Curthoys IS. Effect of bone-conducted vibration of the midline forehead (Fz) in unilateral vestibular loss (uVL). Evidence for a new

1245

indicator of unilateral otolithic function. Acta Otorhinolaryngol Ital 2010;30:175–81. Manzari L, Burgess AM, McGarvie LA, Curthoys IS. Ocular and cervical vestibularevoked myogenic potentials to 500 Hz Fz bone-conducted vibration in superior semicircular canal dehiscence. Ear Hear 2012;33:508–20. McNerney KM, Burkard RF. The vestibular evoked myogenic potential (VEMP): airversus bone-conducted stimuli. Ear Hear 2011;32:e6–e15. Monobe H, Murofushi T. Vestibular neuritis in a child with otitis media with effusion; clinical application of vestibular evoked myogenic potential by boneconducted sound. Int J Pediatr Otorhinolaryngol 2004;68:1455–8. Quiroga RQ, Nadasdy Z, Ben-Shaul Y. Unsupervised spike detection and sorting with wavelets and superparamagnetic clustering. Neural Comput 2004;16:1661–87. Rosengren SM, Aw ST, Halmagyi GM, Todd NPM, Colebatch JG. Ocular vestibular evoked myogenic potentials (OVEMPs) in superior canal dehiscence. J Neurol Neurosurg Psychiatry 2008;79:559–66. Rosengren SM, Todd NPM, Colebatch JG. Vestibular evoked myogenic potentials evoked by brief interaural head acceleration: properties and possible origin. J Appl Physiol 2009;107:841–52. Rosengren SM, Welgampola MS, Colebatch JG. Vestibular evoked myogenic potentials: past, present and future. Clin Neurophysiol 2010;121:636–51. Seo T, Miyamoto A, Saka N, Shimano K, Nishida T, Hashimoto M, Sakagami M. Vestibular evoked myogenic potential induced by bone-conducted stimuli in patients with conductive hearing loss. Acta Otolaryngol 2008;128:639–43. Sheykholeslami K, Kermany MH, Kaga K. Frequency sensitivity range of the saccule to bone-conducted stimuli measured by vestibular evoked myogenic potentials. Hear Res 2001;160:58–62. Sheykholeslami K, Murofushi T, Kermany MH, Kaga K. Bone-conducted evoked myogenic potentials from the sternocleidomastoid muscle. Acta Otolaryngol 2000;120:731–4. Taylor RL, Wijewardene AA, Gibson WPR, Black DA, Halmagyi GM, Welgampola MS. The vestibular evoked-potential profile of Ménière’s disease. Clin Neurophysiol 2011;12:1256–63. Taylor RL, Zagami AS, Gibson WPR, Black DA, Watson SRD, Halmagyi GM, Welgampola MS. Vestibular evoked myogenic potentials to sound and vibration: characteristics in vestibular migraine that enable separation from Menière’s disease. Cephalagia 2012;32:213–25. Todd NPM, Rosengren SM, Colebatch JG. Ocular vestibular evoked myogenic potentials (OVEMPs) produced by impulsive transmastoid accelerations. Clin Neurophysiol 2008;119:1638–81. Uchino Y, Kushiro K. Differences between otolith- and semicircular canal-activated neural circuitry in the vestibular system. Neurosci Res 2011;71:315–27. Weber KP, Rosengren SM, Michels R, Sturm V, Straumann D, Landau K. Single motor unit activity in human extraocular muscles during the vestibulo-ocular reflex. J Physiol 2012;590:3091–101. Welgampola MS, Rosengren SM, Halmagyi GM, Colebatch JG. Vestibular activation by bone-conducted sound. J Neurol Neurosurg Psychiatry 2003;74:771–8. Westin M, Brantberg K. Mastoid and vertex low-frequency vibration-induced oVEMP in relation to medially directed acceleration of the labyrinth. Clin Neurophysiol 2014;125:615–20. Yang T-L, Young Y-H. Comparison of tone burst and tapping evocation of myogenic potentials in patients with chronic otitis media. Ear Hear 2003;24:191–4. Zhu H, Tang X, Wei W, Mustain W, Xu Y, Zhou W. Click-evoked responses in vestibular afferents in rats. J Neurophysiol 2011;106:754–63.