- Email: [email protected]
Ocular vestibular-evoked myogenic potentials to bone-conducted vibration in superior vestibular neuritis show utricular function
Otolaryngology–Head and Neck Surgery (2010) 143, 274-280
ORIGINAL RESEARCH– OTOLOGY AND NEUROTOLOGY
Ocular vestibular-evoked myogenic potentials to bone-conducted vibration in superior vestibular neuritis show utricular function Leonardo Manzari, MD, AnnaRita Tedesco, Ann M. Burgess, PhD, and Ian S. Curthoys, PhD, Rome and Cassino, Italy; and Sydney, New South Wales, Australia Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article. ABSTRACT OBJECTIVE: To determine whether the first negative component (n10) of the ocular vestibular-evoked myogenic potential (oVEMP) to bone-conducted vibration (BCV) is due primarily to activation of the utricular macula. STUDY DESIGN: The n10 was recorded in response to brief BCV at the midline of the forehead at the hairline (Fz). If the n10 is due primarily to utricular activation, then diseases that affect only the superior division of the vestibular nerve in which all utricular afferents course (i.e., superior vestibular neuritis [SVN]) should reduce or eliminate n10 beneath the contralesional eye, whereas the n10 beneath the ipsilesional eye and the sacculo-collic cervical vestibular-evoked myogenic potential (cVEMP) on the ipsilesional side should be preserved. SETTING: A prospective study at a tertiary neurotological referral center. SUBJECTS AND METHODS: The n10 component of the oVEMP was measured in 133 patients with unilateral SVN but with inferior vestibular nerve function preserved, as shown by ipsilesional cVEMPs. RESULTS: The n10 to Fz BCV of 133 SVN patients was reduced beneath the contralesional eye relative to the ipsilesional eye so that there was an n10 asymmetry that was significantly greater than the n10 asymmetry in the 50 healthy subjects. In terms of predicting the affected side (shown by canal paresis), using an n10 asymmetry ratio (asymmetry ratio for the relative size of the n10 of the oVEMPs for the two eyes [AR]) of 46.5 percent, the n10 AR has a diagnostic accuracy of 94 percent. CONCLUSION: The n10 component of the oVEMP to BCV is probably mediated by the superior vestibular nerve and so mainly by the utricular receptors. The n10 AR is almost as good as canal paresis in identifying the affected side in patients. © 2010 American Academy of Otolaryngology–Head and Neck Surgery Foundation. All rights reserved.
C
linical and experimental studies have demonstrated that vestibular-evoked myogenic potentials (VEMPs) are generated by utricular and saccular receptors in response to various acoustic and vibratory stimuli,1-10 and may be recorded by surface electrodes above activated muscles. VEMPs have been proposed as a reliable test that may supplement the current vestibular test battery. The p13 component of the cervical vestibular-evoked myogenic potential (cVEMP) is a short-latency (approximately 13 ms) positive (inhibitory) potential recorded from contracted sternocleidomastoid muscles (SCM) in response to air-conducted sound (ACS)1 or bone-conducted vibration (BCV).4 The cVEMP is an uncrossed, predominantly sacculocollic response and is used to evaluate the function of the saccular macula and the inferior vestibular nerve.1,2,4 Recently, the cVEMP has been complemented by the report of the ocular vestibular-evoked myogenic potential (oVEMP) to BCV.5-10 The first component of the oVEMP to BCV stimulation at the midline of the forehead at the hairline (Fz) is a negative (excitatory) potential at about 10-ms latency (and so-called n10), which is enhanced when the subject looks up,7-10 which brings the inferior oblique (IO) and inferior rectus (IR) closer to the recording electrode. The polarity of the electromyogram (EMG) response and its interpretation as excitatory or inhibitory depends on the electrode configuration and the connections to the differential amplifier. Colebatch and Rothwell showed that, with our configurations, a positive potential is inhibitory, and a negative potential is excitatory.11 What is the probable sequence from vibration stimulus to the n10 response? Linear acceleration measurements at the mastoid show that 500 Hz BCV at Fz causes a series of brief, rapid changes in linear acceleration (a series of jerks) at the mastoids8 and so would be expected to activate preferentially the jerk-sensitive otolith afferents, irregular otolith afferents;12 this is indeed the case. In guinea pigs, irregular otolith afferents, originating from the type 1 receptors at the striola of the maculae, are sensitive to changes
Received August 24, 2009; revised November 9, 2009; accepted March 18, 2010.
0194-5998/$36.00 © 2010 American Academy of Otolaryngology–Head and Neck Surgery Foundation. All rights reserved. doi:10.1016/j.otohns.2010.03.020
Manzari et al
Ocular vestibular-evoked myogenic potentials to bone-conducted . . .
in linear acceleration and are selectively and vigorously activated by 500-Hz bone-conducted vibration at very low intensities close to the auditory brainstem response (ABR) threshold.3 Regular otolith neurons and semicircular canal neurons are not significantly activated by 500 Hz BCV at comparable intensities.3 Such selective otolith activation will result in otolith-ocular and otolith-spinal responses, as Suzuki et al13 showed by selective electrical stimulation of otolith afferents in cats, which resulted in activation of the contralateral IO and IR. This hypothesis of utricular-induced eye movements is confirmed by the recent demonstration that stimulation by BCV of one mastoid in human subjects causes downward and torsional eye movements (upper pole of the eye rolling away from the stimulated ear,14 comparable with the direction reported by Suzuki et al13 in cats after unilateral utricular nerve stimulation). The n10 of the oVEMP to Fz BCV (arrowheads in Fig 1) reflects the muscular precursor to eye movement;5-8 the two n10s are about equal beneath both eyes in healthy subjects,7,8 but, in the patient with superior vestibular neuritis (SVN), the n10 beneath the contralesional (left) eye is very small or absent, whereas the n10 beneath the ipsilesional (right) eye
Figure 1
275
is of normal amplitude. The simplified schematic version of some of the known neural vestibulo-ocular projections, which underlie the asymmetric oVEMP response to Fz BCV after unilateral vestibular loss, is based on known anatomic projections and on physiologic results from Suzuki et al13 and Uchino et al15 showing that high-frequency electrical stimulation of the utricular nerve results in activation of the contralateral inferior oblique. Afferents from the saccular and utricular macula project to the vestibular nuclei, but the exact termination of these afferents is not presently known; Figure 1 represents this present uncertainty about the exact neural connections of these afferents within the vestibular nuclei15 as an open box. The otolithic projections to other eye muscles are not shown. The afferents from the saccular macula course predominantly in the inferior vestibular nerve and synapse on inhibitory neurons in the vestibular nucleus (black hexagons in Fig 1), which in turn project to spinal motoneurons controlling the SCM.15 Therefore, the cVEMP indicates predominantly sacculocollic function. Afferents from the utricular macula course entirely in the superior vestibular nerve and project to many sites, including indirectly to the contralateral IO and IR.13
Simplified schematic diagrams of the likely neural pathways, which mediate the oVEMP and cVEMP responses.10
276
Otolaryngology–Head and Neck Surgery, Vol 143, No 2, August 2010
The clinical interpretation of the cVEMP and oVEMP has focused on amplitude or threshold asymmetries between the myogenic potentials on the left and right sides as an indication of the likely side of the vestibular pathology.1,4 Measures of the n10 asymmetry for Fz BCV stimuli of patients with surgical unilateral vestibular loss after surgery have shown that the symmetry of the amplitude of the n10 response detects the affected side.7,9 One particular measure of n10 symmetry to Fz BCV is the n10 asymmetry ratio (n10 AR), which is defined as n10 AR ⫽ [(n10larger ⫺ n10smaller)/(n10larger ⫹ n10smaller) ⫻ 100]. For healthy subjects, the n10 AR ranged from values of 0.52 percent to 34.05 percent, with an average of 11.73 ⫾ 8.26 percent (standard deviation [SD]).8 For patients with unilateral vestibular loss, the mean n10 AR was significantly greater at 75.03 ⫾ 16.32 percent.9 In previous studies, the reliability of the n10 AR has been shown by repeated testing.7,8 A more common group of patients are individuals with SVN10 who have canal paresis (CP) on the affected side (as shown by caloric stimulation) but have residual saccular and inferior vestibular nerve function, as shown by the presence of cVEMPs in the SCM on the affected side in response to air-conducted sound stimuli.10 If the n10 of the oVEMP depends on utricular as opposed to saccular function, then patients with SVN should show a dissociation between oVEMPs and cVEMPs because all afferents from the utricular macula course in the superior vestibular nerve,16 whereas most afferents from the saccular macula course in the inferior vestibular nerve16 (Fig 2). These patients should, depending on the severity of the neuritis, have a reduced or absent n10 beneath their con-
tralesional eye (i.e., the eye opposite their affected ear), with a normal n10 beneath the ipsilesional eye (the eye on the affected side). On the other hand, the cVEMPs on their ipsilesional SCM should be normal, indicating preserved saccular function (Fig 2). Such a result was reported in a prior small study of 13 patients.10 However, the asymmetry of n10 to Fz BCV in SVN patients needs to be replicated in an independent clinic with a large group of SVN patients. Accordingly, this study set out to measure the asymmetry of n10 to Fz BCV in a large group of patients (133) with clinically identified SVN who still had cVEMPs and so likely retained saccular function. The simple questions were as follows: 1) In response to Fz BCV, do healthy subjects show symmetrical n10 responses? 2) Is it the case that n10 to 500 Hz Fz BCV is reduced or absent opposite the affected ear in a large group of patients with SVN? An additional aim was to establish the diagnostic accuracy of the n10 asymmetry ratio in determining the affected side in patients with vestibular pathology, by statistical comparison of the diagnostic accuracy of the n10 AR with respect to the CP score of the caloric test. In vestibular diagnostic testing, the first question that has to be asked is the following: which side is affected? The “gold standard” for answering that question is the caloric test, and the indicator of the affected side is the CP score. However, the asymmetry ratio of n10 of the oVEMP to Fz BCV stimulation also identifies the affected side, as demonstrated in previous studies,9,10 so an important question is as follows: what is the diagnostic accuracy of the n10 AR in comparison with the gold standard caloric test? We compared the size of the n10 AR in response to Fz BCV stimulation with the CP score from caloric stimulation in 133 patients with SVN and in a group of normal healthy individuals who volunteered to undergo caloric testing (27 of the 50 normal subjects tested). These data allowed us to identify the sensitivity and specificity of the n10 AR of the oVEMP for identifying the affected side, with respect to the CP.17
Patients and Methods
Figure 2 A schematic illustration of projections of vestibular sense organs in the superior and inferior division of the vestibular nerve.16
A total of 133 patients (aged between 21 to 86 years; mean age, 51; 57 male, 76 female) were tested. All of them were identified as having SVN according to the criteria below. These subjects were enrolled in this prospective study from October 2008 to March 2009. All procedures were in accordance with the Helsinki Declaration and were approved by the Human Ethics Committee of the University of Sydney and the MSA ENT Academy Center Institutional Review Board, and all subjects and patients gave informed consent. All patients were assessed with a series of audiovestibular tests: audiometric examination to exclude significant conductive hearing loss, tympanometry, stapedial reflexes, caloric vestibular test (modified Fitzgerald-Hallpike), and cVEMPs to ACS using binaural air-conducted sound stimulation by a 500-Hz logon at 4-Hz repetition rate. In accordance with standard diagnostic criteria,18 the criteria for selecting the patients with unilateral SVN were
Manzari et al
Ocular vestibular-evoked myogenic potentials to bone-conducted . . .
277
as follows: 1) absent or reduced unilateral horizontal canal function (absent or reduced caloric responses on one side and the presence of a head impulse sign for horizontal head rotations toward the affected ear, indicating that the superior vestibular nerve was probably not functional). The caloric response for the healthy ear was within the normal range; 2) cVEMPs were still present in response to air-conducted sound stimulation of the affected ear, indicating that those saccular otolithic afferents coursing in the inferior vestibular nerve were still functional; 3) absence of auditory signs. The results of these 133 patients were combined with results from 50 healthy subjects without any vestibular disturbances, age range of 16 to 86 years, average age of 38 years, and tested with informed consent. None of the healthy subjects reported any auditory, vestibular, neurologic, or visual problems (apart from standard refractive errors).
Method of Stimulation and Recording Fz BCV oVEMPs The methods have been described in detail previously.7,8 Briefly, BCV was delivered using a hand-held, Bruel and Kjaer (Naerum, Denmark), Mini-shaker 4810, fitted with a short bolt (2 cm long, M5) terminated in a bakelite cap 1.5 cm in diameter. The flat end of this cap was the contact point for the stimulator on the subject’s forehead at Fz (Fig 3). The 4810 was driven by computer-generated signals, usually consisting of 50 repetitions of a 500-Hz tone burst lasting a total of 7 ms (including a 1-ms rise and 1-ms fall with a zero crossing start and 5-ms duration) or a square wave of 1-ms duration. Stimulus intensity after amplification was 130 dB Force Level re 1 N as measured by an artificial mastoid (model 4930; Bruel and Kjaer). Other measures have shown that this stimulus causes a linear acceleration of the order of 0.1g at the mastoids. Unrectified EMG was sampled at 20 kHz and band-pass filtered between 3 and 500 Hz and averaged with a Medelec Amplaid MK12 (Medelec Amplaid, Milan, Italy) averager. To minimize artifacts, each lead was shielded individually, and the shielding was connected to the ground electrode attached to the chin or the sternum of the subject. The subjects or patients lay supine on a bed with their head supported on a pillow but positioned so that the head was horizontal or pitched slightly nose down, with the chin close to the chest (Fig 3). The skin beneath the eyes was cleaned very carefully with alcohol wipes (with the patient’s eyes closed), and surface EMG electrodes were applied to the skin beneath both eyes as shown in Figure 3. The self-adhesive pads around each electrode were cut to allow the active (⫹) electrode placement close to the lower eyelid, with the reference electrode (⫺) 2 cm below that, taking care that there was no electrical bridge formed between the conductive gel of the two closely juxtaposed electrodes beneath the eyes. The electrodes were positioned to be aligned with the center of the pupil as the subject looked up at a distant target exactly in the midline (i.e., it is important
Figure 3 The electrode configuration for optimum recording of oVEMPs (the subject shown is A. Tedesco, one of the authors). The location marked X shows the contact point for the 4810 stimulator on the subject’s forehead.
that the eye position in the orbit be elevated, which brings the IO and the IR close to the surface-recording electrodes on the skin beneath the eyes. During testing, the patients and the healthy subjects were instructed to maintain visual fixation on a target placed 25 degrees above their visual straight ahead (Fig 3).
cVEMP Testing and Recording Subjects laid supine on a bed. The skin over the SCM muscles was thoroughly cleaned with alcohol wipes, and surface EMG electrodes were used to record the responses from both SCMs. The subject or patient was required to lift his or her head from the pillow while the operator stimulated at Fz with the BCV stimulus.
Statistical Analysis The significance level was set at 0.05.19 SPSS version 16 (SPSS Inc., Chicago, IL) was used for all data analyses. All results are given as means ⫾ standard deviations unless otherwise noted. A receiver operating characteristic curve (ROC) was calculated using SPSS 16, plotting sensitivity versus specificity to compare n10 AR with CP. The area under the ROC and its 95% confidence interval were calculated.17
278
Figure 4
Otolaryngology–Head and Neck Surgery, Vol 143, No 2, August 2010
Recordings of cVEMPs and oVEMPs from a healthy subject (A) and a patient with right superior vestibular neuritis (B).
Results The average time series from one patient and one healthy subject (Fig 4) show the main features of the oVEMP response to 500 Hz Fz BCV. In healthy subjects, the 500-Hz brief tone burst of BCV at Fz produced a small negative potential at a latency of approximately 10 ms (n10) of approximately equal amplitude beneath both eyes. All 133 patients with SVN had, by definition, shown cVEMP responses to stimulation of their affected ear by air-conducted sound, indicating that the saccular otolithic receptors and their afferents, predominantly in the inferior vestibular nerve, were functional. However, in response to 500 Hz Fz BCV stimulation, all 133 patients had an asymmetric oVEMP response with the n10 component being markedly reduced or absent beneath the eye opposite the affected side. The n10 asymmetry ratio (AR) values for healthy subjects and patients are shown in Figure 5. The average n10 AR of BCV oVEMPs for all the SVN patients was 66.9 ⫾ 19.7 percent SD, n ⫽ 133, which was significantly greater (P ⬎ 0.001) than the n10 AR value of unselected normal subjects, 7.3 ⫾ 4.6 percent SD, n ⫽ 50. The means and two-tailed 95% confidence intervals for the mean are shown within the square, and the box plots for the medians, quartiles, and ranges are shown outside the square (Fig 5).
relation to the CP.17 A cut-off point of an n10 AR of 46.5 percent is optimal and has a sensitivity of 0.9 and a specificity of 0.8 for identifying the affected side. The area under the ROC curve was 0.94 ⫾ 0.02 (standard error) with a 95% confidence interval of 0.89-0.98. This area under the ROC is the index of test accuracy— defined as the probability of correct diagnostic classification.17 Therefore, the probability of correct diagnosis of the affected side by the n10 AR (relative to the “gold standard” of the CP score of the caloric test) is 94 percent. In other words, n10 asymmetry is almost as good as CP at identifying the affected side.
Diagnostic Accuracy Individuals were classed into two groups, normal and abnormal, according to their CP score: 40 were classed as normal in that they had a CP score less than 22 percent, and the remaining 120 were classed as abnormal with a CP ⱖ 22 percent. The n10 AR for each patient was graphed (Fig 6A), and, from these data, an ROC was calculated: the graph (Fig 6B) shows the plot of the sensitivity and specificity of values of the n10 AR for detecting the affected side in
Figure 5 The asymmetry ratios of all 133 patients with superior vestibular neuritis (closed circles) and 50 healthy subjects (open triangles).
Manzari et al
Ocular vestibular-evoked myogenic potentials to bone-conducted . . .
Figure 6
279
(A) The n10 AR values for each individual. (B) The receiver operating characteristic (ROC) curve.
Discussion The answers to the two main questions posed in the introduction are that 1), in response to Fz BCV, healthy subjects show symmetrical n10 responses; and 2), in SVN patients, the n10 of the oVEMP is reduced or absent beneath the contralesional eye while the ipsilesional cVEMP in the SCM is preserved. Therefore, by using BCV at Fz to stimulate both labyrinths simultaneously in patients with SVN, we have shown a dissociation between two indicators of otolithic function in the affected ear. The p13 of the cVEMPs on the ipsilesional SCM was still present, indicating intact saccular and inferior vestibular nerve function. However, in these same patients in response to Fz BCV, the n10 component of the oVEMP beneath the contralesional eye was greatly reduced or absent, whereas it was of normal size beneath the ipsilesional eye. This dissociation between the cVEMP and oVEMP results points very strongly to the crossed n10 being due to utricular function because saccular function in the affected ear was intact, as shown by the cVEMP. Our results confirm previous results10 and show that this Fz BCV stimulus is valuable in exploring superior vestibular nerve function. In SVN, the horizontal and anterior semicircular canals have reduced functioning; therefore, could the results here be because of those reduced semicircular canal responses as opposed to otolithic responses? There are two major grounds for interpreting the n10 responses as being because of otolithic rather than canal activation. 1) Physiology from animal studies shows that primary semicircular canal neurons are only rarely activated (9 out of 189 tested) at very high intensities by BCV, whereas irregular primary otolithic
neurons are very commonly (48 out of 58 tested) activated by the same stimulus at very low intensities— close to ABR threshold.3 There is a large difference in threshold between canal and otolithic responses to BCV. 2) In humans, conditions that should modulate semicircular canal activation do not influence n10 amplitude. The sensitivity and specificity of the n10 to Fz BCV test are shown by the ROC analysis and are high. Fz BCV merits inclusion in a vestibular test battery both because of these impressive numerical values and also the simplicity and speed of giving the test. In addition, it causes the subject little pain and distress, and can be performed even in senior subjects and young children without difficulty or distress.
Acknowledgments The authors thank NH&MRC of Australia and the Garnett Passe and Rodney Williams Memorial Foundation for their support.
Author Information From the Department of Experimental Medicine and Pathology (Dr. Manzari), “La Sapienza” University, Rome, Italy; MSA Academy ENT Center (Dr. Manzari, Ms. Tedesco), Cassino, Italy; and Vestibular Research Laboratory (Drs. Burgess and Curthoys), School of Psychology, the University of Sydney, Sydney, New South Wales, Australia. Corresponding author: Ian S. Curthoys, PhD, Vestibular Research Laboratory, School of Psychology, A 18, University of Sydney, Sydney, NSW 2006, Australia. E-mail address: [email protected].
280
Otolaryngology–Head and Neck Surgery, Vol 143, No 2, August 2010
Author Contributions Leonardo Manzari, design of the study, collection, analysis, interpretation of the data, critical revision of the manuscript, final approval of the submitted version; AnnaRita Tedesco, collection of the data, critical revision of the manuscript, final approval of the submitted version; Ann M. Burgess, analysis of the data, preparation of the graphs, critical revision of the manuscript, final approval of the submitted version; Ian S. Curthoys, design of the study, interpretation of the data, writing most of the paper, preparation of the figures, data analysis, final approval of the submitted version.
Disclosures Competing interests: Ian S. Curthoys, consultant: Otometrics. Sponsorships: Supported in part by Garnett Passe and Rodney Williams Memorial Foundation and National Health and Medical Research Council of Australia for funding some of the data analysis.
References 1. Colebatch JG, Halmagyi GM, Skuse NF. Myogenic potentials generated by a click-evoked vestibulocollic reflex. J Neurol Neurosurg Psychiatry 1994;57:190 –7. 2. Murofushi T, Curthoys IS, Topple AN, et al. Responses of guinea pig primary vestibular neurons to clicks. Exp Brain Res 1995;103:174 – 8. 3. Curthoys IS, Kim J, McPhedran SK, et al. Bone conducted vibration selectively activates irregular primary otolithic vestibular neurons in the guinea pig. Exp Brain Res 2006;175:256 – 67. 4. 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. 5. Rosengren SM, Todd NPM, Colebatch JG. Vestibular-evoked extraocular potentials produced by stimulation with bone-conducted sound. Clin Neurophysiol 2005;116:1938 – 48.
6. Todd NPM, Rosengren SM, Aw ST, et al. Ocular vestibular evoked myogenic potentials (OVEMPs) produced by air- and bone-conducted sound. Clin Neurophysiol 2007;118:381–90. 7. Iwasaki S, McGarvie LA, Halmagyi GM, et al. Head taps evoke a crossed vestibulo-ocular reflex. Neurology 2007;68:1227–9. 8. Iwasaki S, Smulders YE, Burgess AM, et al. Ocular vestibular evoked myogenic potentials to bone conducted vibration of the midline forehead at Fz in healthy subjects. Clin Neurophysiol 2008;119:2135– 47. 9. Iwasaki S, Smulders YE, Burgess AM, et al. Ocular vestibular evoked myogenic potentials in response to bone-conducted vibration of the midline forehead at Fz. A new indicator of unilateral otolithic loss. Audiol Neurotol 2008;13:396 – 404. 10. Iwasaki S, Chihara Y, Smulders Y, et al. The role of the utricular macula and the superior vestibular nerve in the generation of ocular vestibular-evoked myogenic potentials to bone conducted vibration at Fz. Clin Neurophysiol 2009;120:588 –93. 11. Colebatch JG, Rothwell JC. Motor unit excitability changes mediating vestibulocollic reflexes in the sternocleidomastoid muscle. Clin Neurophysiol 2004;115:2567–73. 12. Goldberg JM. Afferent diversity and the organization of central vestibular pathways. Exp Brain Res 2000;130:277–97. 13. Suzuki JI, Tokumasu K, Goto K. Eye movements from single utricular nerve stimulation in the cat. Acta Otolaryngol 1969;68:350 – 62. 14. Cornell ED, Burgess AM, MacDougall HG, et al. Vertical and horizontal eye movement responses to unilateral and bilateral bone conducted vibration to the mastoid. J Vestib Res 2009;19:41–7. 15. Uchino Y, Sasaki M, Sato H, et al. Otolith and canal integration on single vestibular neurons in cats. Exp Brain Res 2005;164:271– 85. 16. de Burlet HM. Zur Innervation der Macula sacculi bei Säugetieren. Anat Anzeig 1924;58:26 –32. 17. Habbema JDF, Eijkemans R, Krijnen P, et al. Analysis of data on the accuracy of diagnostic tests. In: Knottnerus JA, Buntinx F, editors. The Evidence Base of Clinical Diagnosis. 2nd ed. Chichester: WileyBlackwell; 2009. p. 118 – 45. 18. Halmagyi GM, Aw ST, Karlberg M, et al. Inferior vestibular neuritis. Ann N Y Acad Sci 2002;956:306 –13. 19. Winer BJ, Brown DR, Michels KM. Statistical principles in experimental design. New York: McGraw-Hill; 1991.
ORIGINAL RESEARCH– OTOLOGY AND NEUROTOLOGY
Ocular vestibular-evoked myogenic potentials to bone-conducted vibration in superior vestibular neuritis show utricular function Leonardo Manzari, MD, AnnaRita Tedesco, Ann M. Burgess, PhD, and Ian S. Curthoys, PhD, Rome and Cassino, Italy; and Sydney, New South Wales, Australia Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article. ABSTRACT OBJECTIVE: To determine whether the first negative component (n10) of the ocular vestibular-evoked myogenic potential (oVEMP) to bone-conducted vibration (BCV) is due primarily to activation of the utricular macula. STUDY DESIGN: The n10 was recorded in response to brief BCV at the midline of the forehead at the hairline (Fz). If the n10 is due primarily to utricular activation, then diseases that affect only the superior division of the vestibular nerve in which all utricular afferents course (i.e., superior vestibular neuritis [SVN]) should reduce or eliminate n10 beneath the contralesional eye, whereas the n10 beneath the ipsilesional eye and the sacculo-collic cervical vestibular-evoked myogenic potential (cVEMP) on the ipsilesional side should be preserved. SETTING: A prospective study at a tertiary neurotological referral center. SUBJECTS AND METHODS: The n10 component of the oVEMP was measured in 133 patients with unilateral SVN but with inferior vestibular nerve function preserved, as shown by ipsilesional cVEMPs. RESULTS: The n10 to Fz BCV of 133 SVN patients was reduced beneath the contralesional eye relative to the ipsilesional eye so that there was an n10 asymmetry that was significantly greater than the n10 asymmetry in the 50 healthy subjects. In terms of predicting the affected side (shown by canal paresis), using an n10 asymmetry ratio (asymmetry ratio for the relative size of the n10 of the oVEMPs for the two eyes [AR]) of 46.5 percent, the n10 AR has a diagnostic accuracy of 94 percent. CONCLUSION: The n10 component of the oVEMP to BCV is probably mediated by the superior vestibular nerve and so mainly by the utricular receptors. The n10 AR is almost as good as canal paresis in identifying the affected side in patients. © 2010 American Academy of Otolaryngology–Head and Neck Surgery Foundation. All rights reserved.
C
linical and experimental studies have demonstrated that vestibular-evoked myogenic potentials (VEMPs) are generated by utricular and saccular receptors in response to various acoustic and vibratory stimuli,1-10 and may be recorded by surface electrodes above activated muscles. VEMPs have been proposed as a reliable test that may supplement the current vestibular test battery. The p13 component of the cervical vestibular-evoked myogenic potential (cVEMP) is a short-latency (approximately 13 ms) positive (inhibitory) potential recorded from contracted sternocleidomastoid muscles (SCM) in response to air-conducted sound (ACS)1 or bone-conducted vibration (BCV).4 The cVEMP is an uncrossed, predominantly sacculocollic response and is used to evaluate the function of the saccular macula and the inferior vestibular nerve.1,2,4 Recently, the cVEMP has been complemented by the report of the ocular vestibular-evoked myogenic potential (oVEMP) to BCV.5-10 The first component of the oVEMP to BCV stimulation at the midline of the forehead at the hairline (Fz) is a negative (excitatory) potential at about 10-ms latency (and so-called n10), which is enhanced when the subject looks up,7-10 which brings the inferior oblique (IO) and inferior rectus (IR) closer to the recording electrode. The polarity of the electromyogram (EMG) response and its interpretation as excitatory or inhibitory depends on the electrode configuration and the connections to the differential amplifier. Colebatch and Rothwell showed that, with our configurations, a positive potential is inhibitory, and a negative potential is excitatory.11 What is the probable sequence from vibration stimulus to the n10 response? Linear acceleration measurements at the mastoid show that 500 Hz BCV at Fz causes a series of brief, rapid changes in linear acceleration (a series of jerks) at the mastoids8 and so would be expected to activate preferentially the jerk-sensitive otolith afferents, irregular otolith afferents;12 this is indeed the case. In guinea pigs, irregular otolith afferents, originating from the type 1 receptors at the striola of the maculae, are sensitive to changes
Received August 24, 2009; revised November 9, 2009; accepted March 18, 2010.
0194-5998/$36.00 © 2010 American Academy of Otolaryngology–Head and Neck Surgery Foundation. All rights reserved. doi:10.1016/j.otohns.2010.03.020
Manzari et al
Ocular vestibular-evoked myogenic potentials to bone-conducted . . .
in linear acceleration and are selectively and vigorously activated by 500-Hz bone-conducted vibration at very low intensities close to the auditory brainstem response (ABR) threshold.3 Regular otolith neurons and semicircular canal neurons are not significantly activated by 500 Hz BCV at comparable intensities.3 Such selective otolith activation will result in otolith-ocular and otolith-spinal responses, as Suzuki et al13 showed by selective electrical stimulation of otolith afferents in cats, which resulted in activation of the contralateral IO and IR. This hypothesis of utricular-induced eye movements is confirmed by the recent demonstration that stimulation by BCV of one mastoid in human subjects causes downward and torsional eye movements (upper pole of the eye rolling away from the stimulated ear,14 comparable with the direction reported by Suzuki et al13 in cats after unilateral utricular nerve stimulation). The n10 of the oVEMP to Fz BCV (arrowheads in Fig 1) reflects the muscular precursor to eye movement;5-8 the two n10s are about equal beneath both eyes in healthy subjects,7,8 but, in the patient with superior vestibular neuritis (SVN), the n10 beneath the contralesional (left) eye is very small or absent, whereas the n10 beneath the ipsilesional (right) eye
Figure 1
275
is of normal amplitude. The simplified schematic version of some of the known neural vestibulo-ocular projections, which underlie the asymmetric oVEMP response to Fz BCV after unilateral vestibular loss, is based on known anatomic projections and on physiologic results from Suzuki et al13 and Uchino et al15 showing that high-frequency electrical stimulation of the utricular nerve results in activation of the contralateral inferior oblique. Afferents from the saccular and utricular macula project to the vestibular nuclei, but the exact termination of these afferents is not presently known; Figure 1 represents this present uncertainty about the exact neural connections of these afferents within the vestibular nuclei15 as an open box. The otolithic projections to other eye muscles are not shown. The afferents from the saccular macula course predominantly in the inferior vestibular nerve and synapse on inhibitory neurons in the vestibular nucleus (black hexagons in Fig 1), which in turn project to spinal motoneurons controlling the SCM.15 Therefore, the cVEMP indicates predominantly sacculocollic function. Afferents from the utricular macula course entirely in the superior vestibular nerve and project to many sites, including indirectly to the contralateral IO and IR.13
Simplified schematic diagrams of the likely neural pathways, which mediate the oVEMP and cVEMP responses.10
276
Otolaryngology–Head and Neck Surgery, Vol 143, No 2, August 2010
The clinical interpretation of the cVEMP and oVEMP has focused on amplitude or threshold asymmetries between the myogenic potentials on the left and right sides as an indication of the likely side of the vestibular pathology.1,4 Measures of the n10 asymmetry for Fz BCV stimuli of patients with surgical unilateral vestibular loss after surgery have shown that the symmetry of the amplitude of the n10 response detects the affected side.7,9 One particular measure of n10 symmetry to Fz BCV is the n10 asymmetry ratio (n10 AR), which is defined as n10 AR ⫽ [(n10larger ⫺ n10smaller)/(n10larger ⫹ n10smaller) ⫻ 100]. For healthy subjects, the n10 AR ranged from values of 0.52 percent to 34.05 percent, with an average of 11.73 ⫾ 8.26 percent (standard deviation [SD]).8 For patients with unilateral vestibular loss, the mean n10 AR was significantly greater at 75.03 ⫾ 16.32 percent.9 In previous studies, the reliability of the n10 AR has been shown by repeated testing.7,8 A more common group of patients are individuals with SVN10 who have canal paresis (CP) on the affected side (as shown by caloric stimulation) but have residual saccular and inferior vestibular nerve function, as shown by the presence of cVEMPs in the SCM on the affected side in response to air-conducted sound stimuli.10 If the n10 of the oVEMP depends on utricular as opposed to saccular function, then patients with SVN should show a dissociation between oVEMPs and cVEMPs because all afferents from the utricular macula course in the superior vestibular nerve,16 whereas most afferents from the saccular macula course in the inferior vestibular nerve16 (Fig 2). These patients should, depending on the severity of the neuritis, have a reduced or absent n10 beneath their con-
tralesional eye (i.e., the eye opposite their affected ear), with a normal n10 beneath the ipsilesional eye (the eye on the affected side). On the other hand, the cVEMPs on their ipsilesional SCM should be normal, indicating preserved saccular function (Fig 2). Such a result was reported in a prior small study of 13 patients.10 However, the asymmetry of n10 to Fz BCV in SVN patients needs to be replicated in an independent clinic with a large group of SVN patients. Accordingly, this study set out to measure the asymmetry of n10 to Fz BCV in a large group of patients (133) with clinically identified SVN who still had cVEMPs and so likely retained saccular function. The simple questions were as follows: 1) In response to Fz BCV, do healthy subjects show symmetrical n10 responses? 2) Is it the case that n10 to 500 Hz Fz BCV is reduced or absent opposite the affected ear in a large group of patients with SVN? An additional aim was to establish the diagnostic accuracy of the n10 asymmetry ratio in determining the affected side in patients with vestibular pathology, by statistical comparison of the diagnostic accuracy of the n10 AR with respect to the CP score of the caloric test. In vestibular diagnostic testing, the first question that has to be asked is the following: which side is affected? The “gold standard” for answering that question is the caloric test, and the indicator of the affected side is the CP score. However, the asymmetry ratio of n10 of the oVEMP to Fz BCV stimulation also identifies the affected side, as demonstrated in previous studies,9,10 so an important question is as follows: what is the diagnostic accuracy of the n10 AR in comparison with the gold standard caloric test? We compared the size of the n10 AR in response to Fz BCV stimulation with the CP score from caloric stimulation in 133 patients with SVN and in a group of normal healthy individuals who volunteered to undergo caloric testing (27 of the 50 normal subjects tested). These data allowed us to identify the sensitivity and specificity of the n10 AR of the oVEMP for identifying the affected side, with respect to the CP.17
Patients and Methods
Figure 2 A schematic illustration of projections of vestibular sense organs in the superior and inferior division of the vestibular nerve.16
A total of 133 patients (aged between 21 to 86 years; mean age, 51; 57 male, 76 female) were tested. All of them were identified as having SVN according to the criteria below. These subjects were enrolled in this prospective study from October 2008 to March 2009. All procedures were in accordance with the Helsinki Declaration and were approved by the Human Ethics Committee of the University of Sydney and the MSA ENT Academy Center Institutional Review Board, and all subjects and patients gave informed consent. All patients were assessed with a series of audiovestibular tests: audiometric examination to exclude significant conductive hearing loss, tympanometry, stapedial reflexes, caloric vestibular test (modified Fitzgerald-Hallpike), and cVEMPs to ACS using binaural air-conducted sound stimulation by a 500-Hz logon at 4-Hz repetition rate. In accordance with standard diagnostic criteria,18 the criteria for selecting the patients with unilateral SVN were
Manzari et al
Ocular vestibular-evoked myogenic potentials to bone-conducted . . .
277
as follows: 1) absent or reduced unilateral horizontal canal function (absent or reduced caloric responses on one side and the presence of a head impulse sign for horizontal head rotations toward the affected ear, indicating that the superior vestibular nerve was probably not functional). The caloric response for the healthy ear was within the normal range; 2) cVEMPs were still present in response to air-conducted sound stimulation of the affected ear, indicating that those saccular otolithic afferents coursing in the inferior vestibular nerve were still functional; 3) absence of auditory signs. The results of these 133 patients were combined with results from 50 healthy subjects without any vestibular disturbances, age range of 16 to 86 years, average age of 38 years, and tested with informed consent. None of the healthy subjects reported any auditory, vestibular, neurologic, or visual problems (apart from standard refractive errors).
Method of Stimulation and Recording Fz BCV oVEMPs The methods have been described in detail previously.7,8 Briefly, BCV was delivered using a hand-held, Bruel and Kjaer (Naerum, Denmark), Mini-shaker 4810, fitted with a short bolt (2 cm long, M5) terminated in a bakelite cap 1.5 cm in diameter. The flat end of this cap was the contact point for the stimulator on the subject’s forehead at Fz (Fig 3). The 4810 was driven by computer-generated signals, usually consisting of 50 repetitions of a 500-Hz tone burst lasting a total of 7 ms (including a 1-ms rise and 1-ms fall with a zero crossing start and 5-ms duration) or a square wave of 1-ms duration. Stimulus intensity after amplification was 130 dB Force Level re 1 N as measured by an artificial mastoid (model 4930; Bruel and Kjaer). Other measures have shown that this stimulus causes a linear acceleration of the order of 0.1g at the mastoids. Unrectified EMG was sampled at 20 kHz and band-pass filtered between 3 and 500 Hz and averaged with a Medelec Amplaid MK12 (Medelec Amplaid, Milan, Italy) averager. To minimize artifacts, each lead was shielded individually, and the shielding was connected to the ground electrode attached to the chin or the sternum of the subject. The subjects or patients lay supine on a bed with their head supported on a pillow but positioned so that the head was horizontal or pitched slightly nose down, with the chin close to the chest (Fig 3). The skin beneath the eyes was cleaned very carefully with alcohol wipes (with the patient’s eyes closed), and surface EMG electrodes were applied to the skin beneath both eyes as shown in Figure 3. The self-adhesive pads around each electrode were cut to allow the active (⫹) electrode placement close to the lower eyelid, with the reference electrode (⫺) 2 cm below that, taking care that there was no electrical bridge formed between the conductive gel of the two closely juxtaposed electrodes beneath the eyes. The electrodes were positioned to be aligned with the center of the pupil as the subject looked up at a distant target exactly in the midline (i.e., it is important
Figure 3 The electrode configuration for optimum recording of oVEMPs (the subject shown is A. Tedesco, one of the authors). The location marked X shows the contact point for the 4810 stimulator on the subject’s forehead.
that the eye position in the orbit be elevated, which brings the IO and the IR close to the surface-recording electrodes on the skin beneath the eyes. During testing, the patients and the healthy subjects were instructed to maintain visual fixation on a target placed 25 degrees above their visual straight ahead (Fig 3).
cVEMP Testing and Recording Subjects laid supine on a bed. The skin over the SCM muscles was thoroughly cleaned with alcohol wipes, and surface EMG electrodes were used to record the responses from both SCMs. The subject or patient was required to lift his or her head from the pillow while the operator stimulated at Fz with the BCV stimulus.
Statistical Analysis The significance level was set at 0.05.19 SPSS version 16 (SPSS Inc., Chicago, IL) was used for all data analyses. All results are given as means ⫾ standard deviations unless otherwise noted. A receiver operating characteristic curve (ROC) was calculated using SPSS 16, plotting sensitivity versus specificity to compare n10 AR with CP. The area under the ROC and its 95% confidence interval were calculated.17
278
Figure 4
Otolaryngology–Head and Neck Surgery, Vol 143, No 2, August 2010
Recordings of cVEMPs and oVEMPs from a healthy subject (A) and a patient with right superior vestibular neuritis (B).
Results The average time series from one patient and one healthy subject (Fig 4) show the main features of the oVEMP response to 500 Hz Fz BCV. In healthy subjects, the 500-Hz brief tone burst of BCV at Fz produced a small negative potential at a latency of approximately 10 ms (n10) of approximately equal amplitude beneath both eyes. All 133 patients with SVN had, by definition, shown cVEMP responses to stimulation of their affected ear by air-conducted sound, indicating that the saccular otolithic receptors and their afferents, predominantly in the inferior vestibular nerve, were functional. However, in response to 500 Hz Fz BCV stimulation, all 133 patients had an asymmetric oVEMP response with the n10 component being markedly reduced or absent beneath the eye opposite the affected side. The n10 asymmetry ratio (AR) values for healthy subjects and patients are shown in Figure 5. The average n10 AR of BCV oVEMPs for all the SVN patients was 66.9 ⫾ 19.7 percent SD, n ⫽ 133, which was significantly greater (P ⬎ 0.001) than the n10 AR value of unselected normal subjects, 7.3 ⫾ 4.6 percent SD, n ⫽ 50. The means and two-tailed 95% confidence intervals for the mean are shown within the square, and the box plots for the medians, quartiles, and ranges are shown outside the square (Fig 5).
relation to the CP.17 A cut-off point of an n10 AR of 46.5 percent is optimal and has a sensitivity of 0.9 and a specificity of 0.8 for identifying the affected side. The area under the ROC curve was 0.94 ⫾ 0.02 (standard error) with a 95% confidence interval of 0.89-0.98. This area under the ROC is the index of test accuracy— defined as the probability of correct diagnostic classification.17 Therefore, the probability of correct diagnosis of the affected side by the n10 AR (relative to the “gold standard” of the CP score of the caloric test) is 94 percent. In other words, n10 asymmetry is almost as good as CP at identifying the affected side.
Diagnostic Accuracy Individuals were classed into two groups, normal and abnormal, according to their CP score: 40 were classed as normal in that they had a CP score less than 22 percent, and the remaining 120 were classed as abnormal with a CP ⱖ 22 percent. The n10 AR for each patient was graphed (Fig 6A), and, from these data, an ROC was calculated: the graph (Fig 6B) shows the plot of the sensitivity and specificity of values of the n10 AR for detecting the affected side in
Figure 5 The asymmetry ratios of all 133 patients with superior vestibular neuritis (closed circles) and 50 healthy subjects (open triangles).
Manzari et al
Ocular vestibular-evoked myogenic potentials to bone-conducted . . .
Figure 6
279
(A) The n10 AR values for each individual. (B) The receiver operating characteristic (ROC) curve.
Discussion The answers to the two main questions posed in the introduction are that 1), in response to Fz BCV, healthy subjects show symmetrical n10 responses; and 2), in SVN patients, the n10 of the oVEMP is reduced or absent beneath the contralesional eye while the ipsilesional cVEMP in the SCM is preserved. Therefore, by using BCV at Fz to stimulate both labyrinths simultaneously in patients with SVN, we have shown a dissociation between two indicators of otolithic function in the affected ear. The p13 of the cVEMPs on the ipsilesional SCM was still present, indicating intact saccular and inferior vestibular nerve function. However, in these same patients in response to Fz BCV, the n10 component of the oVEMP beneath the contralesional eye was greatly reduced or absent, whereas it was of normal size beneath the ipsilesional eye. This dissociation between the cVEMP and oVEMP results points very strongly to the crossed n10 being due to utricular function because saccular function in the affected ear was intact, as shown by the cVEMP. Our results confirm previous results10 and show that this Fz BCV stimulus is valuable in exploring superior vestibular nerve function. In SVN, the horizontal and anterior semicircular canals have reduced functioning; therefore, could the results here be because of those reduced semicircular canal responses as opposed to otolithic responses? There are two major grounds for interpreting the n10 responses as being because of otolithic rather than canal activation. 1) Physiology from animal studies shows that primary semicircular canal neurons are only rarely activated (9 out of 189 tested) at very high intensities by BCV, whereas irregular primary otolithic
neurons are very commonly (48 out of 58 tested) activated by the same stimulus at very low intensities— close to ABR threshold.3 There is a large difference in threshold between canal and otolithic responses to BCV. 2) In humans, conditions that should modulate semicircular canal activation do not influence n10 amplitude. The sensitivity and specificity of the n10 to Fz BCV test are shown by the ROC analysis and are high. Fz BCV merits inclusion in a vestibular test battery both because of these impressive numerical values and also the simplicity and speed of giving the test. In addition, it causes the subject little pain and distress, and can be performed even in senior subjects and young children without difficulty or distress.
Acknowledgments The authors thank NH&MRC of Australia and the Garnett Passe and Rodney Williams Memorial Foundation for their support.
Author Information From the Department of Experimental Medicine and Pathology (Dr. Manzari), “La Sapienza” University, Rome, Italy; MSA Academy ENT Center (Dr. Manzari, Ms. Tedesco), Cassino, Italy; and Vestibular Research Laboratory (Drs. Burgess and Curthoys), School of Psychology, the University of Sydney, Sydney, New South Wales, Australia. Corresponding author: Ian S. Curthoys, PhD, Vestibular Research Laboratory, School of Psychology, A 18, University of Sydney, Sydney, NSW 2006, Australia. E-mail address: [email protected].
280
Otolaryngology–Head and Neck Surgery, Vol 143, No 2, August 2010
Author Contributions Leonardo Manzari, design of the study, collection, analysis, interpretation of the data, critical revision of the manuscript, final approval of the submitted version; AnnaRita Tedesco, collection of the data, critical revision of the manuscript, final approval of the submitted version; Ann M. Burgess, analysis of the data, preparation of the graphs, critical revision of the manuscript, final approval of the submitted version; Ian S. Curthoys, design of the study, interpretation of the data, writing most of the paper, preparation of the figures, data analysis, final approval of the submitted version.
Disclosures Competing interests: Ian S. Curthoys, consultant: Otometrics. Sponsorships: Supported in part by Garnett Passe and Rodney Williams Memorial Foundation and National Health and Medical Research Council of Australia for funding some of the data analysis.
References 1. Colebatch JG, Halmagyi GM, Skuse NF. Myogenic potentials generated by a click-evoked vestibulocollic reflex. J Neurol Neurosurg Psychiatry 1994;57:190 –7. 2. Murofushi T, Curthoys IS, Topple AN, et al. Responses of guinea pig primary vestibular neurons to clicks. Exp Brain Res 1995;103:174 – 8. 3. Curthoys IS, Kim J, McPhedran SK, et al. Bone conducted vibration selectively activates irregular primary otolithic vestibular neurons in the guinea pig. Exp Brain Res 2006;175:256 – 67. 4. 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. 5. Rosengren SM, Todd NPM, Colebatch JG. Vestibular-evoked extraocular potentials produced by stimulation with bone-conducted sound. Clin Neurophysiol 2005;116:1938 – 48.
6. Todd NPM, Rosengren SM, Aw ST, et al. Ocular vestibular evoked myogenic potentials (OVEMPs) produced by air- and bone-conducted sound. Clin Neurophysiol 2007;118:381–90. 7. Iwasaki S, McGarvie LA, Halmagyi GM, et al. Head taps evoke a crossed vestibulo-ocular reflex. Neurology 2007;68:1227–9. 8. Iwasaki S, Smulders YE, Burgess AM, et al. Ocular vestibular evoked myogenic potentials to bone conducted vibration of the midline forehead at Fz in healthy subjects. Clin Neurophysiol 2008;119:2135– 47. 9. Iwasaki S, Smulders YE, Burgess AM, et al. Ocular vestibular evoked myogenic potentials in response to bone-conducted vibration of the midline forehead at Fz. A new indicator of unilateral otolithic loss. Audiol Neurotol 2008;13:396 – 404. 10. Iwasaki S, Chihara Y, Smulders Y, et al. The role of the utricular macula and the superior vestibular nerve in the generation of ocular vestibular-evoked myogenic potentials to bone conducted vibration at Fz. Clin Neurophysiol 2009;120:588 –93. 11. Colebatch JG, Rothwell JC. Motor unit excitability changes mediating vestibulocollic reflexes in the sternocleidomastoid muscle. Clin Neurophysiol 2004;115:2567–73. 12. Goldberg JM. Afferent diversity and the organization of central vestibular pathways. Exp Brain Res 2000;130:277–97. 13. Suzuki JI, Tokumasu K, Goto K. Eye movements from single utricular nerve stimulation in the cat. Acta Otolaryngol 1969;68:350 – 62. 14. Cornell ED, Burgess AM, MacDougall HG, et al. Vertical and horizontal eye movement responses to unilateral and bilateral bone conducted vibration to the mastoid. J Vestib Res 2009;19:41–7. 15. Uchino Y, Sasaki M, Sato H, et al. Otolith and canal integration on single vestibular neurons in cats. Exp Brain Res 2005;164:271– 85. 16. de Burlet HM. Zur Innervation der Macula sacculi bei Säugetieren. Anat Anzeig 1924;58:26 –32. 17. Habbema JDF, Eijkemans R, Krijnen P, et al. Analysis of data on the accuracy of diagnostic tests. In: Knottnerus JA, Buntinx F, editors. The Evidence Base of Clinical Diagnosis. 2nd ed. Chichester: WileyBlackwell; 2009. p. 118 – 45. 18. Halmagyi GM, Aw ST, Karlberg M, et al. Inferior vestibular neuritis. Ann N Y Acad Sci 2002;956:306 –13. 19. Winer BJ, Brown DR, Michels KM. Statistical principles in experimental design. New York: McGraw-Hill; 1991.