Otology & Neurotology 36:1069Y1073 Ó 2015, Otology & Neurotology, Inc.
Normal Values for Cervical Vestibular-Evoked Myogenic Potentials *Brian W. Blakley and †Veronica Wong *Department of Otolaryngology, ÞUniversity of Manitoba, Winnipeg, Manitoba, Canada
The quartile coefficients of dispersion were much less than 1.0 for all cVEMP parameters in the literature, suggesting that the variability in normal ranges across the literature is small. The distributions for threshold and symmetry ratio were similar between normal and patient groups. There is a lack of information in the literature regarding the impairment of function resulting from various degrees of abnormality of VEMP results. Conclusions: Normal values for cVEMP parameters are statistically consistent in the literature. The clinical significance of abnormal values has not been validated. For clinical purposes, cVEMP ‘‘thresholds’’ should be reported. Reporting of other parameters is optional. Keywords: DizzinessVNormal valuesVSacculeVVestibular-evoked myogenic potentials.
Objectives: To assess the variability of normal values for cervical vestibular-evoked myogenic potentials (cVEMP) testing and to provide guidance regarding which parameters should be reported for clinical practice. Study Design: Forty-eight normal subjects with no history of hearing loss or vestibular symptoms underwent cVEMP testing. Measurement parameters were tabulated and compared to other sets of cVEMP normal values in the literature. The literature was reviewed to assess the clinical significance of abnormal cVEMP results. The distributions of threshold and symmetry ratios for normal subjects were compared to the distributions of 90 patients who underwent cVEMP testing. Setting: Tertiary academic center. Results: Upper limits of 42% symmetry ratio and the range of 65 to 95 dB HL for threshold were established for our center.
Otol Neurotol 36:1069Y1073, 2015.
Vestibular testing is indirect and variable. Every year, new vestibular tests become available with normal values based on statistical considerations. Cervical vestibularevoked myogenic potentials (cVEMP) are a relatively new method of testing that is dependent on function of the saccule and was first developed by Colebatch and others (1,2). cVEMP suffers from the lack of widely accepted normal values with which to interpret vestibular function. How should VEMP results be applied for clinical, as opposed to research, purposes? What parameters should be reported? Should patients with a long latency be concerned? Do the results make a difference in patient management or diagnosis? These are the questions that we seek to address in this paper. cVEMP assesses the saccular response to a loud auditory stimulus by monitoring activity of the sternocleidomastoid
muscles. The saccule is one of one of two sensory otolith organs. Its role is to sense linear acceleration or gravity in position on a sagittal plane. Initially, the cVEMP response was thought by many to be an artifact or form of auditory brainstem response test because of the belief that a sound stimulus could not elicit a vestibular response. This notion has been disproven by a large body of evidence including the finding that profoundly deaf patients often have normal cVEMP responses (3). Although ocular VEMP testing and other forms have been proposed, this paper considers cervical or cVEMP testing only. Compared to other modes of vestibular testing, cVEMP is better tolerated by the patient, inducing less vertigo. It is easier to perform and provides more interpretable results. It may differentiate between central and peripheral defects. Recent publications have de-emphasized the role of the saccule, as opposed to differentiating between superior and inferior vestibular nerves for some reason (4), but the test still reflects saccular function. The FDA has issued a recall notice for VEMP software, indicating that it had not approved the test for clinical use in the United States, thus limiting its use. There is uncertainty about the clinical interpretation and application of the data. A wide variety of parameters are measured for the two
Address correspondence and reprint requests to Brian W. Blakley, M.D., Ph.D., FRCSC, FACS, Department of Otolaryngology, University of Manitoba, GB441-820 Sherbrook St., Winnipeg, Manitoba, Canada R3A 1R9; E-mail: [email protected] The authors report no conflicts of interest. This project was funded by the Department of Otolaryngology, University of Manitoba.
1069
Copyright © 2015 Otology & Neurotology, Inc. Unauthorized reproduction of this article is prohibited.
1070
B. W. BLAKLEY AND V. WONG
cVEMP waves, P13 and N23, which occur approximately 13 and 23 milliseconds after the stimulus. Parameters measured, often with different names, may include the following: Y P13 latency Y P13 amplitude Y N23 latency Y N23 amplitude Y P13-N23 interwave latency Y right/left ear differences in interwave latency Y right/left ear differences between P13 and N23 latencies and/or amplitudes Y right/left ear amplitude ratios for amplitude and/or asymmetry Y thresholds; absolute and relative, and others. The FDA’s concern may be valid. After all, as the number of parameters measured in a test increases, the chance that one or more will be abnormal increases, particularly when those parameters vary greatly. As more parameters are reported, normal individuals are labeled ‘‘abnormal.’’ Of course, different centers use different test parameters such as bone-conduction clicks, or pure tones of various frequencies, different test positions, data acquisition parameters, and other preferences. Activation of the sternocleidomastoid muscle is required for cVEMP testing, and there are several methods to accomplish this. There is still a paucity of data regarding the clinical meaning of cVEMP results for subjects who are ‘‘abnormal’’ in some measure. It seems valuable to consider two types of normal valuesV‘‘statistical’’ normal values and ‘‘clinical’’ normal values. The former is typically considered to be within T2 standard deviations of the mean if data are normally distributed. Clinically abnormal values refer to the measures that indicate some impact on human function. Clinically abnormal values do not exist for most vestibular tests because they are very difficult to validate. Previous reports have found cVEMP testing to be increasingly useful for evaluation of patients with Me´nie`re’s disease (5) or orthostatic hypotension (6), but the findings are nonspecific. Some believe that cVEMP testing helps in the staging of vestibular disease and guides timing of surgical intervention (5). Although the most common ‘‘abnormality’’ in many tests is reduced or poorer function, cVEMP testing has the unique finding of lower thresholds than normal in patients with superior semicircular canal dehiscence (7Y11). Other than this application, it seems that cVEMP test abnormalities are nonspecific. This study was conducted to compare statistically normal values from our center to those in the literature, review the literature regarding the clinical interpretation of cVEMP testing, and make recommendations regarding reporting parameters for clinical use. The study was approved by the University of Manitoba’s Health Research Ethics Board. METHODS Our normal sample consisted of 48 consenting subjects, 28 females and 20 males ranging in age from 23 to 64 years (mean: 36.3 TSD: 13.1 yr) who denied hearing loss or dizziness. Subjects
consisted of friends, relatives, students, and staff members who responded to an e-mail request for volunteers. All subjects had a normal otoscopic examination and were tested with the Intracoustics computerized system. We contrasted threshold and symmetry data for this normal group with 90 consecutive unselected patients who were referred for cVEMP testing over a calendar year. Subjects completed the test in the chair with the back angled at 30 degrees above the horizontal plane as for caloric testing. After skin preparation, surface EKG electrodes were placed on the forehead (ground), sternum (reference), and over the midpoint of each sternocleidomastoid muscle (active). Cervical VEMP testing requires some muscle tension for accurate recording. The sternocleidomastoid muscles were activated during stimulation by asking the subject to lift their head off the back of the chair or modify their head position until the EMG activity was within the acceptable range. Two hundred tone bursts at 500 Hz were administered to each ear through ear buds starting at 100 dB HL. The evoked responses were averaged to establish the P13 and N23 wave responses. The stimulus amplitude was decreased or increased in 5-dB steps until the greatest intensity that did not elicit a response was found. This method is analogous to the Hughes-Westlake method for audiometry. The P13 wave response was considered a positive wave that occurred between 10 and 20 milliseconds after the stimulus and the N23 wave response was considered a negative wave that occurred between 20 and 30 milliseconds after the stimulus. We recorded the following parameters: 1. Threshold (dB)Vgreatest intensity at which there was no cVEMP response 2. Symmetry ratio (SR)Vthe ratio of the difference divided by the sum of the right and left amplitudes of the P13 response to a 100-dB stimulus 3. P13 latency (ms) 4. P13 amplitude (mV) 5. N23 latency (ms 6. N23 amplitude (mV) Our results were compared to those in the literature. A PubMed search using the keywords ‘‘VEMP normative’’ or ‘‘vestibular evoked myogenic potentials’’ was employed to identify papers in the literature that reported normal values in the general population using analogous methods. The reported means and standard deviations (SD) of the parameters for each paper were entered in a database and are displayed in Figures 1, 2, 3, and 4 as normal value ranges. For our data, the overall median, 5th percentile, and 95th percentile were calculated using IBM SPSSv22.0 (Chicago). Nonparametric statistics were used because application of the Kolmogorov-Smirnov test indicated that the data were not normally distributed. We wanted to compare variability across datasets quantitatively. The coefficient of variation (CV) which is the standard deviation divided by the mean is often used as a measure to compare variability for normally distributed data. One interpretation is that a CV less than 1 can be considered acceptable variability whereas a CV greater than 1 is not. We found that the data across the papers in the literature were not normally distributed, invalidating the CV. A nonparametric equivalent is the quartile coefficient of dispersion (12) which is calculated as the difference between the third and first quartiles divided by the median. The overall quartile coefficients of dispersion of the papers in the literature were calculated for the upper and lower limits of each parameter and appear in the figure legends. ‘‘No response’’ is an important type of missing data that presents problems in quantitative analysis. This was considered
Otology & Neurotology, Vol. 36, No. 6, 2015
Copyright © 2015 Otology & Neurotology, Inc. Unauthorized reproduction of this article is prohibited.
NORMAL VALUES FOR CVEMP
FIG. 1. Comparison of some normal ranges for threshold from the literature. High and low limits of normal are shown as calculated from data in various papers as indicated by the authors. Our current data is shown as a dotted line. The quartile coefficient of dispersion was 0.3 for lower threshold and 0.02 for the upper threshold, suggesting acceptably small variability across the literature. The scale is dB SPL. Some data, including the current data, were converted from dB HL to dB SPL for comparison.
by conducting analysis using two methods. The first analysis was conducted simply excluding missing data. For the second analysis, the maximum stimulus level (100 dB) was entered as the threshold and the mean of the other parameters was assumed
FIG. 2. Normal symmetry ratios from the literature. Symmetry ratio, sometimes called ‘‘asymmetry ratio’’ or percent difference, measures the relative size of responses in the right and left ears. It is usually calculated as the ratio of the difference and the sum of the amplitudes of the P13 waves in the two ears. Data represent two standard deviations of symmetry ratio calculated from data in the papers. Some papers, not included in this graph, use different definitions. The left-most margin should be assumed to start at zero. Our current data is shown by a dotted line. The quartile coefficient of dispersion was 0.28 for the papers in the literature shown on this graph, suggesting good consistency across the literature. The normal ranges are large, suggesting that the asymmetry ratio may not be as sensitive as may be desired.
1071
FIG. 3. Normal P13 latencies in the literature. Normal ranges (T2SD) for P13 latency based on data in the papers. Our current data (dotted line) study shows our 5th and 95th percentiles, but the 2 SD criterion gave similar results. The variability of P13 latencies is large, limiting the applicability of this parameter to clinical practice. The quartile coefficient of dispersion was 0.16 for the upper limit and 0.36 for the lower limit of P13 latency, which indicates good consistency across the literature. P13 is the cVEMP wave that occurs between 10 and 20 milliseconds after stimulus. Only those normal subjects with responses are included.
as the missing data. This method of handling missing thresholds is common in audiological research. Differences in final interpretation between the two analyses with and without missing data were not significant, so we report only the results with the missing data represented as absent and identify the number of ears for which there were no responses (two ears).
FIG. 4. Normal N23 latencies in the literature. Normal range (T2 SD) for N23 latency was calculated from data in the papers. Our current study (dotted line) shows 5th and 95th percentiles, but the 2 SD criterion gave similar results. The range of N213 latencies is considerable. The quartile coefficient of dispersion was 0.18 for the upper limit and 0.21 for the lower limit of normal for N23 latency, which indicates good consistency in the literature. N213 is the cVEMP wave that occurs between 20 and 30 milliseconds after stimulus. Only those normal subjects with responses are included. Otology & Neurotology, Vol. 36, No. 6, 2015
Copyright © 2015 Otology & Neurotology, Inc. Unauthorized reproduction of this article is prohibited.
1072
B. W. BLAKLEY AND V. WONG
Finally, the Mann-Whitney U test was applied to assess the statistical significance of the differences in threshold and symmetry ratio for the 48 normal subjects (96 ears) and the 90 patients (180 ears) who were referred for cVEMP testing in the calendar year July 1, 2013 to June 30, 2014.
RESULTS There were no significant differences in cVEMP parameters between the right and left ears or between male and female subjects, so both genders and both ears were pooled. Cervical VEMP responses were present bilaterally in 46 of 48 subjects. Two subjects had no response in one ear, so data are available for 94 of 96 ears (98% response rate). Symmetry ratio calculations compare the two ears, so data for 46 subjects were available. The 5th and 95th percentiles of the parameters rounded to the nearest 5 dB level that we test at as our normal values are shown in Table 1. It is interesting to note that papers in the literature report mean T 2 SD without consideration of the distribution of the data. Nevertheless, when we analyzed our data using both parametric and nonparametric tests, the differences were small. Our normal value for ‘‘threshold’’ is 65 to 95 dB HL and for symmetry ratio the normal vale is less than 42%. Several authors have provided summaries of their normal values (13Y24). Our results are compared to values from the literature for P13 latency, N23 latency, threshold, and symmetry ratio in Figures 1 to 4 for some studies that used test parameters and conditions similar to ours. P13 and N23 amplitudes were variable and most authors do not report them, so they are not shown here. For all the other parameters, the greatest quartile coefficient of dispersion was 0.3 which is well below the cutoff of 1.0 suggested for acceptable measure of variability. More detailed statistical analysis of the differences across the literature would seem to be invalid because of different recording parameters and standards, so only the comparison graphs in Figures 1 to 4 are shown. Most authors reported thresholds in dB SPL, so for comparison purposes, those that used dB HL (19,23), including our current data, were converted to dB SPL by adding the ANSI conversion for 500 Hz of 13.5 dB (25). TABLE 1. Parameter Threshold, dBHL Symmetry ratio P13 latency, ms P13 amplitude, mV N23 latency, ms N23 amplitude, mV
Our center’s cVEMP norms 5th Percentile
95th Percentile
70 1.7 13.9 j76.7 22.9 13.1
91.3 41.4 19.2 j10.7 30.3 110.7
This table shows the 5th and 95th percentiles for each parameter. Abnormal values for clinical purposes should be the first level outside these normal values at which testing is performed to include the 95th percentile limit. In clinical practice, we deliver stimuli at 5-dB steps so the upper clinical limit for ‘‘threshold’’ is 990 dB HL and the lower limit is 70 dB HL. The abnormal range is then e65 dB or Q95 dB. The upper limit of the ‘‘symmetry ratio’’ is 42%. N = 93 ears. The upper normal value for the symmetry ratio is G42%, whereas being a ratio of the difference and the sum of the amplitude of P13, the lower limit is 0%.
Despite different protocols, centers, equipment, training, and other factors, the data are fairly consistent. We found no substantive discussion in the literature about ‘‘clinical’’ application of normal values except for the application to superior semicircular canal dehiscence (11,26Y28). The omission of data to assess the functional significance, as opposed to the statistical significance, was disconcerting. The Mann-Whitney U test results indicated that the threshold and symmetry ratio were not statistically significant between the 48 normal subjects (96 ears) and the 90 patients (180 ears). P values were p = 0.713 and p = 0.135, respectively. Missing responses were more likely in patients (56 of 180 ears = 31.19%) than normals (2 of 96 ears = 2%). DISCUSSION The response rate for normal subjects in this study at the stimulus intensity of 100dBHL was 98%, which is higher than most reports. The finding that ‘‘normal’’ subjects are reported as having no response is disconcerting. Our findings demonstrate a symmetry ratio 41.4%. We consider the upper limit of normal for asymmetry ratio to be 42% at our center. We test in 5-dB steps, so our normal limits of threshold should be between 70 and 95 dB HL. Abnormal values fall outside this range. It seems important that each center conduct its own study of normal values to apply for its own situation. We are not confident that animal testing will validate cVEMP testing. Although animal cVEMP results have been reported, we have attempted to perform cVEMP testing in animals without success. The problem is that cVEMP testing requires active, voluntary, consistent muscle activity of the sternocleidomastoid muscle which precludes the sedation that animals require for cooperation during recording. The cVEMP parameter that has the most evidencebased clinical application is ‘‘threshold.’’ Although other parameters may be useful for research purposes, ‘‘threshold’’ may give insight into the presence of superior semicircular canal dehiscence. A conservative view of reporting would suggest that the other parameters do not have evidence to justify their use in clinical application. This view is reinforced by the finding in our patients and other reports that ‘‘normal’’ subjects frequently have no response and that the distributions of parameters were not different for patients versus normals. Considering the variability of results in normal subjects, it becomes less clear which abnormal results indicate that one or both ears are abnormal or not. We realize that more detailed dissection of the patient data may identify subgroups with consistent cVEMP abnormalities and that is the subject of a future paper. Although clinicians frequently rely on tests to make diagnoses, there is little evidence that results that are statistically abnormal always relate to human dysfunction. The assumption that statistically abnormal values can be applied directly to clinical situations may arise from the
Otology & Neurotology, Vol. 36, No. 6, 2015
Copyright © 2015 Otology & Neurotology, Inc. Unauthorized reproduction of this article is prohibited.
NORMAL VALUES FOR CVEMP fact that most clinical vestibular testing laboratories are not run by clinicians who see patients. Regardless, there is a need for each center to validate their normal values for vestibular tests for themselves, with their own equipment, staff, and methods. VEMP testing may contribute information as part of a battery of tests to be considered together. Currently, there is a lack of evidence to support the reporting of cVEMP parameters other than threshold as a stand-alone test for clinical utility at this point. Other parameters could be of interest in research applications.
11. 12. 13. 14. 15.
CONCLUSIONS Our normal range of values for cVEMP threshold is between 70 and 90 dB HL. For clinical purposes, cVEMP ‘‘thresholds’’ should be reported. The clinical significance of other parameters is unknown at this time.
16. 17. 18.
REFERENCES 1. Colebatch JG, Halmagyi GM, Skuse NF. Myogenic potentials generated by a click-evoked vestibulocollic reflex. J Neurol Neurosurg Psychiatr 1994;57:190Y7. 2. Colebatch JG, Halmagyi GM. Vestibular evoked potentials in human neck muscles before and after unilateral deafferation. Neurology 1992;42:1635Y6. 3. Bath AP, Harris N, Yardley MP. The vestibulo-collic reflex. Clin Otolaryngol Allied Sci 1998;23:462Y6. 4. Curthoys IS. The interpretation of clinical tests of peripheral vestibular function. Laryngoscope 2012;122:1342Y52. doi: 1310.1002/ lary.23258. 5. Lin MC, Young YH. The use of vestibular test battery to identify the stages of delayed endolymphatic hydrops. Otolaryngol Head Neck Surg 2012;147:912Y8. 6. Lin KY, Wang SJ, Young YH. Vestibular-evoked myogenic potential tests in orthostatic dizziness. Clin Auton Res 2012;22:281Y7. 7. Welgampola MS, Myrie OA, Minor LB, Carey JP. Vestibularevoked myogenic potential thresholds normalize on plugging superior canal dehiscence. Neurology 2008;70:464Y72. 8. Zhang AS, Govender S, Colebatch JG. Tuning of the ocular vestibular evoked myogenic potential (oVEMP) to air- and boneconducted sound stimulation in superior canal dehiscence. Exp Brain Res 2012;223:51Y64. 9. Zuniga MG, Davalos-Bichara M, Schubert MC, Carey JP, Janky KL. Optimizing ocular vestibular evoked myogenic potential testing for superior semicircular canal dehiscence syndrome: electrode placement. Audiol Neurootol 2014;19:239Y47. 10. Niesten ME, Hamberg LM, Silverman JB, et al. Superior canal dehiscence length and location influences clinical presentation and
19. 20.
21. 22. 23. 24. 25. 26. 27. 28.
1073
audiometric and cervical vestibular-evoked myogenic potential testing. Audiol Neurootol 2014;19:97Y105. Rinaldi V, Portmann D. Vestibular-evoked myogenic potentials after superior semicircular canal obliteration. Rev Laryngol Otol Rhinol (Bord) 2011;132:85Y7. Bonett DG. Confidence interval for a coefficient of quartile variation. Comput Stat Data Anal 2006;50:2953Y7. Brantberg K, Fransson PA. Symmetry measures of vestibular evoked myogenic potentials using objective detection criteria. Scand Audiol 2001;30:189Y96. Cheng PW, Huang TW, Young YH. The influence of clicks versus short tone bursts on the vestibular evoked myogenic potentials. Ear Hear 2003;24:195Y7. Isaradisaikul S, Navacharoen N, Hanprasertpong C, Kangsanarak J. Cervical vestibular-evoked myogenic potentials: norms and protocols. Int J Otolaryngol 2012;2012:913515. Janky KL, Shepard N. Vestibular evoked myogenic potential (VEMP) testing: normative threshold response curves and effects of age. J Am Acad Audiol 2009;20:514Y22. Maes L, Vinck BM, De Vel E, et al. The vestibular evoked myogenic potential: a test-retest reliability study. Clin Neurophysiol 2009;120:594Y600. Mahdi P, Amali A, Pourbakht A, Karimi Yazdi A, Bassam A. Vestibular evoked myogenic potential produced by bone-conducted stimuli: a study on its basics and clinical applications in patients with conductive and sensorineural hearing loss and a group with vestibular schwannoma. Iran J Otorhinolaryngol 2013;25:141Y6. Ochi K, Ohashi T, Nishino H. Variance of vestibular-evoked myogenic potentials. Laryngoscope 2001;111:522Y7. Sanchez-Andrade IV, Soto-Varela A, Labella Caballero T, Gayoso Diz P, Santos-Perez S. Impact of subject’s position and acoustic stimulus type on vestibular evoked myogenic potentials (VEMPs) in normal subjects. Eur Arch Otorhinolaryngol 2014;271:2359Y64. Wang CT, Young YH. Earlier and later components of tone burst evoked myogenic potentials. Hear Res 2004;191:59Y66. Wang CT, Young YH. Comparison of the head elevation versus rotation methods in eliciting vestibular evoked myogenic potentials. Ear Hear 2006;27:376Y81. Welgampola MS, Colebatch JG. Characteristics of tone burstevoked myogenic potentials in the sternocleidomastoid muscles. Otol Neurotol 2001;22:796Y802. Wu CH, Murofushi T. The effect of click repetition rate on vestibular evoked myogenic potential. Acta Otolaryngol 1999;119: 29Y32. ANSI. American National Standard Specifications for Audiometers. New York: National Standards Institute, 1996. Brantberg K, Verrecchia L. Testing vestibular-evoked myogenic potentials with 90-dB clicks is effective in the diagnosis of superior canal dehiscence syndrome. Audiol Neurootol 2009;14:54Y8. Crane BT, Minor LB, Carey JP. Three-dimensional computed tomography of superior canal dehiscence syndrome. Otol Neurotol 2008;29:699Y705. Zhou G, Gopen Q, Poe DS. Clinical and diagnostic characterization of canal dehiscence syndrome: a great otologic mimicker. Otol Neurotol 2007;28:920Y6.
Otology & Neurotology, Vol. 36, No. 6, 2015
Copyright © 2015 Otology & Neurotology, Inc. Unauthorized reproduction of this article is prohibited.
Normal Values for Cervical Vestibular-Evoked Myogenic Potentials *Brian W. Blakley and †Veronica Wong *Department of Otolaryngology, ÞUniversity of Manitoba, Winnipeg, Manitoba, Canada
The quartile coefficients of dispersion were much less than 1.0 for all cVEMP parameters in the literature, suggesting that the variability in normal ranges across the literature is small. The distributions for threshold and symmetry ratio were similar between normal and patient groups. There is a lack of information in the literature regarding the impairment of function resulting from various degrees of abnormality of VEMP results. Conclusions: Normal values for cVEMP parameters are statistically consistent in the literature. The clinical significance of abnormal values has not been validated. For clinical purposes, cVEMP ‘‘thresholds’’ should be reported. Reporting of other parameters is optional. Keywords: DizzinessVNormal valuesVSacculeVVestibular-evoked myogenic potentials.
Objectives: To assess the variability of normal values for cervical vestibular-evoked myogenic potentials (cVEMP) testing and to provide guidance regarding which parameters should be reported for clinical practice. Study Design: Forty-eight normal subjects with no history of hearing loss or vestibular symptoms underwent cVEMP testing. Measurement parameters were tabulated and compared to other sets of cVEMP normal values in the literature. The literature was reviewed to assess the clinical significance of abnormal cVEMP results. The distributions of threshold and symmetry ratios for normal subjects were compared to the distributions of 90 patients who underwent cVEMP testing. Setting: Tertiary academic center. Results: Upper limits of 42% symmetry ratio and the range of 65 to 95 dB HL for threshold were established for our center.
Otol Neurotol 36:1069Y1073, 2015.
Vestibular testing is indirect and variable. Every year, new vestibular tests become available with normal values based on statistical considerations. Cervical vestibularevoked myogenic potentials (cVEMP) are a relatively new method of testing that is dependent on function of the saccule and was first developed by Colebatch and others (1,2). cVEMP suffers from the lack of widely accepted normal values with which to interpret vestibular function. How should VEMP results be applied for clinical, as opposed to research, purposes? What parameters should be reported? Should patients with a long latency be concerned? Do the results make a difference in patient management or diagnosis? These are the questions that we seek to address in this paper. cVEMP assesses the saccular response to a loud auditory stimulus by monitoring activity of the sternocleidomastoid
muscles. The saccule is one of one of two sensory otolith organs. Its role is to sense linear acceleration or gravity in position on a sagittal plane. Initially, the cVEMP response was thought by many to be an artifact or form of auditory brainstem response test because of the belief that a sound stimulus could not elicit a vestibular response. This notion has been disproven by a large body of evidence including the finding that profoundly deaf patients often have normal cVEMP responses (3). Although ocular VEMP testing and other forms have been proposed, this paper considers cervical or cVEMP testing only. Compared to other modes of vestibular testing, cVEMP is better tolerated by the patient, inducing less vertigo. It is easier to perform and provides more interpretable results. It may differentiate between central and peripheral defects. Recent publications have de-emphasized the role of the saccule, as opposed to differentiating between superior and inferior vestibular nerves for some reason (4), but the test still reflects saccular function. The FDA has issued a recall notice for VEMP software, indicating that it had not approved the test for clinical use in the United States, thus limiting its use. There is uncertainty about the clinical interpretation and application of the data. A wide variety of parameters are measured for the two
Address correspondence and reprint requests to Brian W. Blakley, M.D., Ph.D., FRCSC, FACS, Department of Otolaryngology, University of Manitoba, GB441-820 Sherbrook St., Winnipeg, Manitoba, Canada R3A 1R9; E-mail: [email protected] The authors report no conflicts of interest. This project was funded by the Department of Otolaryngology, University of Manitoba.
1069
Copyright © 2015 Otology & Neurotology, Inc. Unauthorized reproduction of this article is prohibited.
1070
B. W. BLAKLEY AND V. WONG
cVEMP waves, P13 and N23, which occur approximately 13 and 23 milliseconds after the stimulus. Parameters measured, often with different names, may include the following: Y P13 latency Y P13 amplitude Y N23 latency Y N23 amplitude Y P13-N23 interwave latency Y right/left ear differences in interwave latency Y right/left ear differences between P13 and N23 latencies and/or amplitudes Y right/left ear amplitude ratios for amplitude and/or asymmetry Y thresholds; absolute and relative, and others. The FDA’s concern may be valid. After all, as the number of parameters measured in a test increases, the chance that one or more will be abnormal increases, particularly when those parameters vary greatly. As more parameters are reported, normal individuals are labeled ‘‘abnormal.’’ Of course, different centers use different test parameters such as bone-conduction clicks, or pure tones of various frequencies, different test positions, data acquisition parameters, and other preferences. Activation of the sternocleidomastoid muscle is required for cVEMP testing, and there are several methods to accomplish this. There is still a paucity of data regarding the clinical meaning of cVEMP results for subjects who are ‘‘abnormal’’ in some measure. It seems valuable to consider two types of normal valuesV‘‘statistical’’ normal values and ‘‘clinical’’ normal values. The former is typically considered to be within T2 standard deviations of the mean if data are normally distributed. Clinically abnormal values refer to the measures that indicate some impact on human function. Clinically abnormal values do not exist for most vestibular tests because they are very difficult to validate. Previous reports have found cVEMP testing to be increasingly useful for evaluation of patients with Me´nie`re’s disease (5) or orthostatic hypotension (6), but the findings are nonspecific. Some believe that cVEMP testing helps in the staging of vestibular disease and guides timing of surgical intervention (5). Although the most common ‘‘abnormality’’ in many tests is reduced or poorer function, cVEMP testing has the unique finding of lower thresholds than normal in patients with superior semicircular canal dehiscence (7Y11). Other than this application, it seems that cVEMP test abnormalities are nonspecific. This study was conducted to compare statistically normal values from our center to those in the literature, review the literature regarding the clinical interpretation of cVEMP testing, and make recommendations regarding reporting parameters for clinical use. The study was approved by the University of Manitoba’s Health Research Ethics Board. METHODS Our normal sample consisted of 48 consenting subjects, 28 females and 20 males ranging in age from 23 to 64 years (mean: 36.3 TSD: 13.1 yr) who denied hearing loss or dizziness. Subjects
consisted of friends, relatives, students, and staff members who responded to an e-mail request for volunteers. All subjects had a normal otoscopic examination and were tested with the Intracoustics computerized system. We contrasted threshold and symmetry data for this normal group with 90 consecutive unselected patients who were referred for cVEMP testing over a calendar year. Subjects completed the test in the chair with the back angled at 30 degrees above the horizontal plane as for caloric testing. After skin preparation, surface EKG electrodes were placed on the forehead (ground), sternum (reference), and over the midpoint of each sternocleidomastoid muscle (active). Cervical VEMP testing requires some muscle tension for accurate recording. The sternocleidomastoid muscles were activated during stimulation by asking the subject to lift their head off the back of the chair or modify their head position until the EMG activity was within the acceptable range. Two hundred tone bursts at 500 Hz were administered to each ear through ear buds starting at 100 dB HL. The evoked responses were averaged to establish the P13 and N23 wave responses. The stimulus amplitude was decreased or increased in 5-dB steps until the greatest intensity that did not elicit a response was found. This method is analogous to the Hughes-Westlake method for audiometry. The P13 wave response was considered a positive wave that occurred between 10 and 20 milliseconds after the stimulus and the N23 wave response was considered a negative wave that occurred between 20 and 30 milliseconds after the stimulus. We recorded the following parameters: 1. Threshold (dB)Vgreatest intensity at which there was no cVEMP response 2. Symmetry ratio (SR)Vthe ratio of the difference divided by the sum of the right and left amplitudes of the P13 response to a 100-dB stimulus 3. P13 latency (ms) 4. P13 amplitude (mV) 5. N23 latency (ms 6. N23 amplitude (mV) Our results were compared to those in the literature. A PubMed search using the keywords ‘‘VEMP normative’’ or ‘‘vestibular evoked myogenic potentials’’ was employed to identify papers in the literature that reported normal values in the general population using analogous methods. The reported means and standard deviations (SD) of the parameters for each paper were entered in a database and are displayed in Figures 1, 2, 3, and 4 as normal value ranges. For our data, the overall median, 5th percentile, and 95th percentile were calculated using IBM SPSSv22.0 (Chicago). Nonparametric statistics were used because application of the Kolmogorov-Smirnov test indicated that the data were not normally distributed. We wanted to compare variability across datasets quantitatively. The coefficient of variation (CV) which is the standard deviation divided by the mean is often used as a measure to compare variability for normally distributed data. One interpretation is that a CV less than 1 can be considered acceptable variability whereas a CV greater than 1 is not. We found that the data across the papers in the literature were not normally distributed, invalidating the CV. A nonparametric equivalent is the quartile coefficient of dispersion (12) which is calculated as the difference between the third and first quartiles divided by the median. The overall quartile coefficients of dispersion of the papers in the literature were calculated for the upper and lower limits of each parameter and appear in the figure legends. ‘‘No response’’ is an important type of missing data that presents problems in quantitative analysis. This was considered
Otology & Neurotology, Vol. 36, No. 6, 2015
Copyright © 2015 Otology & Neurotology, Inc. Unauthorized reproduction of this article is prohibited.
NORMAL VALUES FOR CVEMP
FIG. 1. Comparison of some normal ranges for threshold from the literature. High and low limits of normal are shown as calculated from data in various papers as indicated by the authors. Our current data is shown as a dotted line. The quartile coefficient of dispersion was 0.3 for lower threshold and 0.02 for the upper threshold, suggesting acceptably small variability across the literature. The scale is dB SPL. Some data, including the current data, were converted from dB HL to dB SPL for comparison.
by conducting analysis using two methods. The first analysis was conducted simply excluding missing data. For the second analysis, the maximum stimulus level (100 dB) was entered as the threshold and the mean of the other parameters was assumed
FIG. 2. Normal symmetry ratios from the literature. Symmetry ratio, sometimes called ‘‘asymmetry ratio’’ or percent difference, measures the relative size of responses in the right and left ears. It is usually calculated as the ratio of the difference and the sum of the amplitudes of the P13 waves in the two ears. Data represent two standard deviations of symmetry ratio calculated from data in the papers. Some papers, not included in this graph, use different definitions. The left-most margin should be assumed to start at zero. Our current data is shown by a dotted line. The quartile coefficient of dispersion was 0.28 for the papers in the literature shown on this graph, suggesting good consistency across the literature. The normal ranges are large, suggesting that the asymmetry ratio may not be as sensitive as may be desired.
1071
FIG. 3. Normal P13 latencies in the literature. Normal ranges (T2SD) for P13 latency based on data in the papers. Our current data (dotted line) study shows our 5th and 95th percentiles, but the 2 SD criterion gave similar results. The variability of P13 latencies is large, limiting the applicability of this parameter to clinical practice. The quartile coefficient of dispersion was 0.16 for the upper limit and 0.36 for the lower limit of P13 latency, which indicates good consistency across the literature. P13 is the cVEMP wave that occurs between 10 and 20 milliseconds after stimulus. Only those normal subjects with responses are included.
as the missing data. This method of handling missing thresholds is common in audiological research. Differences in final interpretation between the two analyses with and without missing data were not significant, so we report only the results with the missing data represented as absent and identify the number of ears for which there were no responses (two ears).
FIG. 4. Normal N23 latencies in the literature. Normal range (T2 SD) for N23 latency was calculated from data in the papers. Our current study (dotted line) shows 5th and 95th percentiles, but the 2 SD criterion gave similar results. The range of N213 latencies is considerable. The quartile coefficient of dispersion was 0.18 for the upper limit and 0.21 for the lower limit of normal for N23 latency, which indicates good consistency in the literature. N213 is the cVEMP wave that occurs between 20 and 30 milliseconds after stimulus. Only those normal subjects with responses are included. Otology & Neurotology, Vol. 36, No. 6, 2015
Copyright © 2015 Otology & Neurotology, Inc. Unauthorized reproduction of this article is prohibited.
1072
B. W. BLAKLEY AND V. WONG
Finally, the Mann-Whitney U test was applied to assess the statistical significance of the differences in threshold and symmetry ratio for the 48 normal subjects (96 ears) and the 90 patients (180 ears) who were referred for cVEMP testing in the calendar year July 1, 2013 to June 30, 2014.
RESULTS There were no significant differences in cVEMP parameters between the right and left ears or between male and female subjects, so both genders and both ears were pooled. Cervical VEMP responses were present bilaterally in 46 of 48 subjects. Two subjects had no response in one ear, so data are available for 94 of 96 ears (98% response rate). Symmetry ratio calculations compare the two ears, so data for 46 subjects were available. The 5th and 95th percentiles of the parameters rounded to the nearest 5 dB level that we test at as our normal values are shown in Table 1. It is interesting to note that papers in the literature report mean T 2 SD without consideration of the distribution of the data. Nevertheless, when we analyzed our data using both parametric and nonparametric tests, the differences were small. Our normal value for ‘‘threshold’’ is 65 to 95 dB HL and for symmetry ratio the normal vale is less than 42%. Several authors have provided summaries of their normal values (13Y24). Our results are compared to values from the literature for P13 latency, N23 latency, threshold, and symmetry ratio in Figures 1 to 4 for some studies that used test parameters and conditions similar to ours. P13 and N23 amplitudes were variable and most authors do not report them, so they are not shown here. For all the other parameters, the greatest quartile coefficient of dispersion was 0.3 which is well below the cutoff of 1.0 suggested for acceptable measure of variability. More detailed statistical analysis of the differences across the literature would seem to be invalid because of different recording parameters and standards, so only the comparison graphs in Figures 1 to 4 are shown. Most authors reported thresholds in dB SPL, so for comparison purposes, those that used dB HL (19,23), including our current data, were converted to dB SPL by adding the ANSI conversion for 500 Hz of 13.5 dB (25). TABLE 1. Parameter Threshold, dBHL Symmetry ratio P13 latency, ms P13 amplitude, mV N23 latency, ms N23 amplitude, mV
Our center’s cVEMP norms 5th Percentile
95th Percentile
70 1.7 13.9 j76.7 22.9 13.1
91.3 41.4 19.2 j10.7 30.3 110.7
This table shows the 5th and 95th percentiles for each parameter. Abnormal values for clinical purposes should be the first level outside these normal values at which testing is performed to include the 95th percentile limit. In clinical practice, we deliver stimuli at 5-dB steps so the upper clinical limit for ‘‘threshold’’ is 990 dB HL and the lower limit is 70 dB HL. The abnormal range is then e65 dB or Q95 dB. The upper limit of the ‘‘symmetry ratio’’ is 42%. N = 93 ears. The upper normal value for the symmetry ratio is G42%, whereas being a ratio of the difference and the sum of the amplitude of P13, the lower limit is 0%.
Despite different protocols, centers, equipment, training, and other factors, the data are fairly consistent. We found no substantive discussion in the literature about ‘‘clinical’’ application of normal values except for the application to superior semicircular canal dehiscence (11,26Y28). The omission of data to assess the functional significance, as opposed to the statistical significance, was disconcerting. The Mann-Whitney U test results indicated that the threshold and symmetry ratio were not statistically significant between the 48 normal subjects (96 ears) and the 90 patients (180 ears). P values were p = 0.713 and p = 0.135, respectively. Missing responses were more likely in patients (56 of 180 ears = 31.19%) than normals (2 of 96 ears = 2%). DISCUSSION The response rate for normal subjects in this study at the stimulus intensity of 100dBHL was 98%, which is higher than most reports. The finding that ‘‘normal’’ subjects are reported as having no response is disconcerting. Our findings demonstrate a symmetry ratio 41.4%. We consider the upper limit of normal for asymmetry ratio to be 42% at our center. We test in 5-dB steps, so our normal limits of threshold should be between 70 and 95 dB HL. Abnormal values fall outside this range. It seems important that each center conduct its own study of normal values to apply for its own situation. We are not confident that animal testing will validate cVEMP testing. Although animal cVEMP results have been reported, we have attempted to perform cVEMP testing in animals without success. The problem is that cVEMP testing requires active, voluntary, consistent muscle activity of the sternocleidomastoid muscle which precludes the sedation that animals require for cooperation during recording. The cVEMP parameter that has the most evidencebased clinical application is ‘‘threshold.’’ Although other parameters may be useful for research purposes, ‘‘threshold’’ may give insight into the presence of superior semicircular canal dehiscence. A conservative view of reporting would suggest that the other parameters do not have evidence to justify their use in clinical application. This view is reinforced by the finding in our patients and other reports that ‘‘normal’’ subjects frequently have no response and that the distributions of parameters were not different for patients versus normals. Considering the variability of results in normal subjects, it becomes less clear which abnormal results indicate that one or both ears are abnormal or not. We realize that more detailed dissection of the patient data may identify subgroups with consistent cVEMP abnormalities and that is the subject of a future paper. Although clinicians frequently rely on tests to make diagnoses, there is little evidence that results that are statistically abnormal always relate to human dysfunction. The assumption that statistically abnormal values can be applied directly to clinical situations may arise from the
Otology & Neurotology, Vol. 36, No. 6, 2015
Copyright © 2015 Otology & Neurotology, Inc. Unauthorized reproduction of this article is prohibited.
NORMAL VALUES FOR CVEMP fact that most clinical vestibular testing laboratories are not run by clinicians who see patients. Regardless, there is a need for each center to validate their normal values for vestibular tests for themselves, with their own equipment, staff, and methods. VEMP testing may contribute information as part of a battery of tests to be considered together. Currently, there is a lack of evidence to support the reporting of cVEMP parameters other than threshold as a stand-alone test for clinical utility at this point. Other parameters could be of interest in research applications.
11. 12. 13. 14. 15.
CONCLUSIONS Our normal range of values for cVEMP threshold is between 70 and 90 dB HL. For clinical purposes, cVEMP ‘‘thresholds’’ should be reported. The clinical significance of other parameters is unknown at this time.
16. 17. 18.
REFERENCES 1. Colebatch JG, Halmagyi GM, Skuse NF. Myogenic potentials generated by a click-evoked vestibulocollic reflex. J Neurol Neurosurg Psychiatr 1994;57:190Y7. 2. Colebatch JG, Halmagyi GM. Vestibular evoked potentials in human neck muscles before and after unilateral deafferation. Neurology 1992;42:1635Y6. 3. Bath AP, Harris N, Yardley MP. The vestibulo-collic reflex. Clin Otolaryngol Allied Sci 1998;23:462Y6. 4. Curthoys IS. The interpretation of clinical tests of peripheral vestibular function. Laryngoscope 2012;122:1342Y52. doi: 1310.1002/ lary.23258. 5. Lin MC, Young YH. The use of vestibular test battery to identify the stages of delayed endolymphatic hydrops. Otolaryngol Head Neck Surg 2012;147:912Y8. 6. Lin KY, Wang SJ, Young YH. Vestibular-evoked myogenic potential tests in orthostatic dizziness. Clin Auton Res 2012;22:281Y7. 7. Welgampola MS, Myrie OA, Minor LB, Carey JP. Vestibularevoked myogenic potential thresholds normalize on plugging superior canal dehiscence. Neurology 2008;70:464Y72. 8. Zhang AS, Govender S, Colebatch JG. Tuning of the ocular vestibular evoked myogenic potential (oVEMP) to air- and boneconducted sound stimulation in superior canal dehiscence. Exp Brain Res 2012;223:51Y64. 9. Zuniga MG, Davalos-Bichara M, Schubert MC, Carey JP, Janky KL. Optimizing ocular vestibular evoked myogenic potential testing for superior semicircular canal dehiscence syndrome: electrode placement. Audiol Neurootol 2014;19:239Y47. 10. Niesten ME, Hamberg LM, Silverman JB, et al. Superior canal dehiscence length and location influences clinical presentation and
19. 20.
21. 22. 23. 24. 25. 26. 27. 28.
1073
audiometric and cervical vestibular-evoked myogenic potential testing. Audiol Neurootol 2014;19:97Y105. Rinaldi V, Portmann D. Vestibular-evoked myogenic potentials after superior semicircular canal obliteration. Rev Laryngol Otol Rhinol (Bord) 2011;132:85Y7. Bonett DG. Confidence interval for a coefficient of quartile variation. Comput Stat Data Anal 2006;50:2953Y7. Brantberg K, Fransson PA. Symmetry measures of vestibular evoked myogenic potentials using objective detection criteria. Scand Audiol 2001;30:189Y96. Cheng PW, Huang TW, Young YH. The influence of clicks versus short tone bursts on the vestibular evoked myogenic potentials. Ear Hear 2003;24:195Y7. Isaradisaikul S, Navacharoen N, Hanprasertpong C, Kangsanarak J. Cervical vestibular-evoked myogenic potentials: norms and protocols. Int J Otolaryngol 2012;2012:913515. Janky KL, Shepard N. Vestibular evoked myogenic potential (VEMP) testing: normative threshold response curves and effects of age. J Am Acad Audiol 2009;20:514Y22. Maes L, Vinck BM, De Vel E, et al. The vestibular evoked myogenic potential: a test-retest reliability study. Clin Neurophysiol 2009;120:594Y600. Mahdi P, Amali A, Pourbakht A, Karimi Yazdi A, Bassam A. Vestibular evoked myogenic potential produced by bone-conducted stimuli: a study on its basics and clinical applications in patients with conductive and sensorineural hearing loss and a group with vestibular schwannoma. Iran J Otorhinolaryngol 2013;25:141Y6. Ochi K, Ohashi T, Nishino H. Variance of vestibular-evoked myogenic potentials. Laryngoscope 2001;111:522Y7. Sanchez-Andrade IV, Soto-Varela A, Labella Caballero T, Gayoso Diz P, Santos-Perez S. Impact of subject’s position and acoustic stimulus type on vestibular evoked myogenic potentials (VEMPs) in normal subjects. Eur Arch Otorhinolaryngol 2014;271:2359Y64. Wang CT, Young YH. Earlier and later components of tone burst evoked myogenic potentials. Hear Res 2004;191:59Y66. Wang CT, Young YH. Comparison of the head elevation versus rotation methods in eliciting vestibular evoked myogenic potentials. Ear Hear 2006;27:376Y81. Welgampola MS, Colebatch JG. Characteristics of tone burstevoked myogenic potentials in the sternocleidomastoid muscles. Otol Neurotol 2001;22:796Y802. Wu CH, Murofushi T. The effect of click repetition rate on vestibular evoked myogenic potential. Acta Otolaryngol 1999;119: 29Y32. ANSI. American National Standard Specifications for Audiometers. New York: National Standards Institute, 1996. Brantberg K, Verrecchia L. Testing vestibular-evoked myogenic potentials with 90-dB clicks is effective in the diagnosis of superior canal dehiscence syndrome. Audiol Neurootol 2009;14:54Y8. Crane BT, Minor LB, Carey JP. Three-dimensional computed tomography of superior canal dehiscence syndrome. Otol Neurotol 2008;29:699Y705. Zhou G, Gopen Q, Poe DS. Clinical and diagnostic characterization of canal dehiscence syndrome: a great otologic mimicker. Otol Neurotol 2007;28:920Y6.
Otology & Neurotology, Vol. 36, No. 6, 2015
Copyright © 2015 Otology & Neurotology, Inc. Unauthorized reproduction of this article is prohibited.
