Correlating Vestibular Schwannoma Size With Vestibular-Evoked Myogenic Potential Results Kuan-Liang Lin,1 Chang-Mu Chen,2 Shou-Jen Wang,3 and Yi-Ho Young1 Objectives: The maximum size of the vestibular schwannoma (VS) that is compatible with preservation of the function of the vestibular nerve in performing stereotactic radiosurgery remains unclear. This study utilized ocular vestibular-evoked myogenic potential (oVEMP) and cervical VEMP (cVEMP) test results to correlate with the size of VS.
because of its lower morbidity and better functional preservation compared to microsurgery. However, the maximum size of the VS that is compatible with preserving the function of the vestibular nerve in performing SRS remains unclear. Stimulation via air-conducted sound or bone-conducted vibration enables recording of vestibular-evoked myogenic potential (VEMP) from cervical muscles (called cervical VEMP [cVEMP]) and extraocular muscles (called ocular VEMP [oVEMP]). A recent investigation showed that the cVEMP test evaluates the ipsilateral sacculo-collic reflex, via the inferior vestibular nerve (IVN) to the motor neurons of the neck muscles. In contrast, the oVEMP runs through the crossed vestibulo-ocular reflex, along the superior vestibular nerve (SVN) to the opposite extraocular muscles (Colebatch et al. 1994; Curthoys et al. 2010). Both tests thus can be adopted in VS patients to assess the afflictions of SVN and IVN (Ushio et al. 2009; Lin et al. 2013b). Previously, Day et al. (2008) reported that estimated tumor size in VS patients with abnormal caloric or cVEMP responses increased by 1.43 or 1.35 cm, respectively. However, various output expressions between caloric nystagmus and cVEMP waveforms, and the evaluation of different nerve bundles in caloric and cVEMP tests, show that the two test results cannot be compared (Su & Young 2011). Since both oVEMP and cVEMP tests display similar output waveforms and since oVEMP and caloric tests partially share the same vestibulo-ocular reflex pathway, this study utilized oVEMP and cVEMP test results to correlate with theVS size.
Design: Fifty patients with unilateral VS underwent audiometry, and caloric, oVEMP and cVEMP tests. Tumor size from magnetic resonance imaging was measured on the axial plane, and the relationships between tumor size and each test result were analyzed. Results: The pure-tone average from four frequencies did not significantly predict tumor size. Alternatively, oVEMP and cVEMP responses remained significant predictors for tumor size in the regression model, namely, tumor size (cm) = 0.62 × (oVEMP response) + 1.39 × (cVEMP response), where oVEMP and cVEMP responses were regarded as binary variables, in which 1 and 0 reflect abnormal and normal responses, respectively. This model explained 76% of the variance. Accordingly, the estimated VS size exhibiting abnormal oVEMPs and cVEMPs is >2.01 (0.62 +1.39) cm. Conclusions: When VS size is 50 µV (Chang et al. 2007). The subjects elevated their heads during testing. A total of 50 responses were averaged and recorded bilaterally. The first positive and second negative polarities of biphasic waveform were termed waves p13 and n23, respectively. At our laboratory, the norm for the latency of p13 was 14.4 ± 1.3 msec, and we defined p13 latency >17.0 msec as delayed cVEMPs. In addition, the norm for the asymmetry ratio of cVEMP was 11 ± 11%, and those with asymmetry ratio >33% were defined as abnormal.
oVEMP Test
Foam Posturography
Caloric Test
The subject was in a sitting position. Two active electrodes were placed around 1 cm below the center of the two lower eyelids. The other two reference electrodes were positioned about 1 to 2 cm below the active ones, and one ground electrode was placed on the sternum. During recording (Smart EP 3.90; Intelligent Hearing Systems, Miami, FL), the subject was instructed to look upward at a small fixed target >2 m from the eyes. The electromyographic signals were amplified and band-passfiltered between 1 and 1000 Hz. The stimulation rate was 5/s. The duration of analysis of each response was 50 msec, and 30 responses were averaged for each run. Bone-conducted vibration stimuli were delivered using a hand-held electromechanical vibrator (minishaker 4810; Bruel and Kjaer, Naerum, Denmark). The operator held the vibrator by hand and supported most of its weight such that the axis of the connected bakelite cap perpendicularly delivered a repeatable tap with little pressure on the subject’s skull at the Fz site (the midline forehead at the hairline). If oVEMP responses were not elicited, alternatively, tapping at the ipsilateral mastoid site (2 cm behind the opening of external ear canal) was subsequently performed (Tseng et al. 2012). The input signal was 500 Hz sine wave, driven by a custom amplifier. The drive voltage was adjusted and fixed to produce a peak force equivalent to 128-dB force level. The initial negative–positive biphasic waveform comprised peaks nI and pI. Consecutive runs were performed to confirm the reproducibility of peaks nI and pI, and oVEMPs
Posturography (Lucerne Measuring Plate; Happersberger Otopront, Hohenstein, Germany) consisted of a computer coupled with transducers beneath a rectangular force platform. Initially, the subject was asked to stand up straight at the appointed place on the platform and keep the body as stable as possible. The position of the center of pressure in each subject under four test conditions was measured, namely: Condition A: firm surface with eyes open; Condition B: firm surface with eyes closed; Condition C: foam pad with eyes open; and Condition D: foam pad with eyes closed (Lin et al. 2013a). The foam pad used in this study was 56.3 × 50.0 × 7.6 cm section of mediumdensity (2 lb/ft3) Sunmate foam with the compressive strength of 10 to 50 psi, which reduced the accuracy of the orientation information. Each condition lasted for 30 s or until the subject required assistance to prevent falling. Two characteristic parameters were recorded for analysis according to our previous report (Lin et al. 2013a), namely, sway area per second (the envelopment area traced by the movement of the center of pressure per second, cm2/s) and Romberg quotient for sway area (measured value from eyes closed divided by that from eyes open).
Statistical Methods The grades of vestibular function deficit between two groups were compared by Fisher exact test. Data of posturography between various grades were analyzed using one-way repeatedmeasures analyses of variance with Bonferroni-adjusted t test
Lin ET AL. / EAR & HEARING, VOL. 35, NO. 5, 571–576
for comparisons. Cochran Q test was utilized to compare the abnormalities among the vestibular tests. The relationship between the tumor size and audiovestibular function was best fit using the linear regression model. Student’s t tests were carried out to determine whether the corresponding regression coefficient is significantly different from zero. A p value 3.0 cm in four patients.
Audiometry Prior to treatment, audiometry identified class A in 12 patients (24%), class B in 11 patients, class C in 22 patients, and class D in 5 patients. The mean PTA of the lesion ears from 500, 1000, 2000, and 3000 Hz was 52 ± 26 (mean ± SD) dB. No significant correlation existed between PTA and tumor size (r = 0.184, p = 0.201). Further, using linear regression analysis, the regression coefficient of PTA did not pass the t test, indicating that PTA was not significant in predicting tumor size.
Vestibular Function Test A vestibular function test battery comprising caloric, oVEMP, and cVEMP tests was performed on all 50 VS patients before treatment. On the lesion ears, 27 patients had normal caloric responses, and 23 patients (46%) had abnormal caloric responses, including canal paresis in three ears and caloric areflexia in 20 ears. Thirteen patients had normal oVEMPs, and 37 patients (74%) showed abnormal oVEMPs consisting of absent responses in 35 and delayed responses in 2. With the cVEMP test, 8 patients exhibited normal responses, and 42 patients (84%) showed abnormal cVEMPs comprising absent responses in 36 and delayed responses in 6. There was a significantly declining trend in abnormal percentages from cVEMP test (84%), followed by oVEMP test (74%) and caloric test (46%; p < 0.01, Cochran Q test). In contrast, normal responses were observed on the opposite healthy ears despite caloric, oVEMP, or cVEMP tests (Figs. 1 and 2). Registration of vestibular deficits was based on the number of abnormal test results of the caloric, oVEMP, and cVEMP tests. Grade I indicates that one of the three test results is abnormal, while grades II and III indicate two and three tests are abnormal, respectively. Finally, grade 0 means that all three tests are normal. Accordingly, grade 0 was identified in 2 patients, grade I in 11 patients, grade II in 20 patients, and grade III in 17 patients (Table 1).
573
Based on tumor size, 50 patients were classified into two groups, namely, group A had a maximum tumor size ≤1.0 cm and comprised 12 patients, whereas group B had a maximum tumor size >1.0 cm and comprised 38 patients. Thus, grades 0, I, II, and III in group A had 1, 4, 4, and 3 patients, respectively, compared with the respective 1, 7, 16, and 14 patients in group B; therefore, the two groups did not significantly differ in the distribution of grades of vestibular deficits (p > 0.05, Fisher exact test; Table 1).
Foam Posturography Versus Grades of Vestibular Deficits To determine whether the grades of vestibular deficits affect balance, all patients underwent foam posturography before treatment. No significant differences among the grades of vestibular deficits existed in terms of sway area per second with eyes open, sway area per second with eyes closed, and Romberg quotient of sway area, regardless of firm or foam surface (p > 0.05, analyses of variance test; Table 2), indicating that the grades of vestibular deficits did not affect balance.
Tumor Size Versus Vestibular Function Testing Using simple linear regression analysis, the results of vestibular function tests were used to predict the VS size, which was significantly related to caloric response (regression coefficient = 1.84, p < 0.01, t test), oVEMP response (regression coefficient = 1.78, p < 0.01), and cVEMP response (regression coefficient = 1.85, p < 0.01; Table 3). Among the three independent variables, the cVEMP response was the most significant predictor of VS size, explaining 73% of the variance in tumor size (i.e., R2), followed by the oVEMP response (60%) and caloric response (40%; Table 3). Further, a multivariate stepwise linear regression model was employed to identify significant independent predictors of VS size. As a result, linear regression model is tumor size (cm) = 0.40 (caloric response) + 1.65 (cVEMP response). However, the regression coefficient for caloric response (0.40) failed to pass the t test (p = 0.16; Table 4). Thus, caloric response is not a significant predictor of tumor size in this regression model. Alternatively, oVEMP and cVEMP responses remained significant predictors of tumor size in the regression model after backward elimination of caloric response (p < 0.05; Table 4), namely, tumor size (cm) = 0.62 × (oVEMP response) + 1.39 × (cVEMP response), where oVEMP and cVEMP responses were regarded as binary variables, in which 1 and 0 reflect abnormal and normal responses, respectively. This final model explained 76% of the variance. Accordingly, the estimated size of VS that exhibits abnormal oVEMPs and cVEMPs is >2.01 (0.62 + 1.39) cm.
DISCUSSION In the modern era, tumor control and functional preservation have become the mainstream in management of VS. Several series have reported promising outcome of SRS in terms of tumor control and hearing preservation (Pollock et al. 2006; Sughrue et al. 2010). Most tumors treated by SRS were 0.05, Fisher exact test.
a tumor for which vestibular nerve function can be preserved when SRS is performed.
Audiometry Since most VS originate from the vestibular nerve, tumor compression or stretching of the cochlear nerve accounts for the hearing loss. However, a series of studies have documented that auditory brainstem response is not sensitive in intracanalicular tumor, with only 58% sensitivity when tumor size
because of its lower morbidity and better functional preservation compared to microsurgery. However, the maximum size of the VS that is compatible with preserving the function of the vestibular nerve in performing SRS remains unclear. Stimulation via air-conducted sound or bone-conducted vibration enables recording of vestibular-evoked myogenic potential (VEMP) from cervical muscles (called cervical VEMP [cVEMP]) and extraocular muscles (called ocular VEMP [oVEMP]). A recent investigation showed that the cVEMP test evaluates the ipsilateral sacculo-collic reflex, via the inferior vestibular nerve (IVN) to the motor neurons of the neck muscles. In contrast, the oVEMP runs through the crossed vestibulo-ocular reflex, along the superior vestibular nerve (SVN) to the opposite extraocular muscles (Colebatch et al. 1994; Curthoys et al. 2010). Both tests thus can be adopted in VS patients to assess the afflictions of SVN and IVN (Ushio et al. 2009; Lin et al. 2013b). Previously, Day et al. (2008) reported that estimated tumor size in VS patients with abnormal caloric or cVEMP responses increased by 1.43 or 1.35 cm, respectively. However, various output expressions between caloric nystagmus and cVEMP waveforms, and the evaluation of different nerve bundles in caloric and cVEMP tests, show that the two test results cannot be compared (Su & Young 2011). Since both oVEMP and cVEMP tests display similar output waveforms and since oVEMP and caloric tests partially share the same vestibulo-ocular reflex pathway, this study utilized oVEMP and cVEMP test results to correlate with theVS size.
Design: Fifty patients with unilateral VS underwent audiometry, and caloric, oVEMP and cVEMP tests. Tumor size from magnetic resonance imaging was measured on the axial plane, and the relationships between tumor size and each test result were analyzed. Results: The pure-tone average from four frequencies did not significantly predict tumor size. Alternatively, oVEMP and cVEMP responses remained significant predictors for tumor size in the regression model, namely, tumor size (cm) = 0.62 × (oVEMP response) + 1.39 × (cVEMP response), where oVEMP and cVEMP responses were regarded as binary variables, in which 1 and 0 reflect abnormal and normal responses, respectively. This model explained 76% of the variance. Accordingly, the estimated VS size exhibiting abnormal oVEMPs and cVEMPs is >2.01 (0.62 +1.39) cm. Conclusions: When VS size is 50 µV (Chang et al. 2007). The subjects elevated their heads during testing. A total of 50 responses were averaged and recorded bilaterally. The first positive and second negative polarities of biphasic waveform were termed waves p13 and n23, respectively. At our laboratory, the norm for the latency of p13 was 14.4 ± 1.3 msec, and we defined p13 latency >17.0 msec as delayed cVEMPs. In addition, the norm for the asymmetry ratio of cVEMP was 11 ± 11%, and those with asymmetry ratio >33% were defined as abnormal.
oVEMP Test
Foam Posturography
Caloric Test
The subject was in a sitting position. Two active electrodes were placed around 1 cm below the center of the two lower eyelids. The other two reference electrodes were positioned about 1 to 2 cm below the active ones, and one ground electrode was placed on the sternum. During recording (Smart EP 3.90; Intelligent Hearing Systems, Miami, FL), the subject was instructed to look upward at a small fixed target >2 m from the eyes. The electromyographic signals were amplified and band-passfiltered between 1 and 1000 Hz. The stimulation rate was 5/s. The duration of analysis of each response was 50 msec, and 30 responses were averaged for each run. Bone-conducted vibration stimuli were delivered using a hand-held electromechanical vibrator (minishaker 4810; Bruel and Kjaer, Naerum, Denmark). The operator held the vibrator by hand and supported most of its weight such that the axis of the connected bakelite cap perpendicularly delivered a repeatable tap with little pressure on the subject’s skull at the Fz site (the midline forehead at the hairline). If oVEMP responses were not elicited, alternatively, tapping at the ipsilateral mastoid site (2 cm behind the opening of external ear canal) was subsequently performed (Tseng et al. 2012). The input signal was 500 Hz sine wave, driven by a custom amplifier. The drive voltage was adjusted and fixed to produce a peak force equivalent to 128-dB force level. The initial negative–positive biphasic waveform comprised peaks nI and pI. Consecutive runs were performed to confirm the reproducibility of peaks nI and pI, and oVEMPs
Posturography (Lucerne Measuring Plate; Happersberger Otopront, Hohenstein, Germany) consisted of a computer coupled with transducers beneath a rectangular force platform. Initially, the subject was asked to stand up straight at the appointed place on the platform and keep the body as stable as possible. The position of the center of pressure in each subject under four test conditions was measured, namely: Condition A: firm surface with eyes open; Condition B: firm surface with eyes closed; Condition C: foam pad with eyes open; and Condition D: foam pad with eyes closed (Lin et al. 2013a). The foam pad used in this study was 56.3 × 50.0 × 7.6 cm section of mediumdensity (2 lb/ft3) Sunmate foam with the compressive strength of 10 to 50 psi, which reduced the accuracy of the orientation information. Each condition lasted for 30 s or until the subject required assistance to prevent falling. Two characteristic parameters were recorded for analysis according to our previous report (Lin et al. 2013a), namely, sway area per second (the envelopment area traced by the movement of the center of pressure per second, cm2/s) and Romberg quotient for sway area (measured value from eyes closed divided by that from eyes open).
Statistical Methods The grades of vestibular function deficit between two groups were compared by Fisher exact test. Data of posturography between various grades were analyzed using one-way repeatedmeasures analyses of variance with Bonferroni-adjusted t test
Lin ET AL. / EAR & HEARING, VOL. 35, NO. 5, 571–576
for comparisons. Cochran Q test was utilized to compare the abnormalities among the vestibular tests. The relationship between the tumor size and audiovestibular function was best fit using the linear regression model. Student’s t tests were carried out to determine whether the corresponding regression coefficient is significantly different from zero. A p value 3.0 cm in four patients.
Audiometry Prior to treatment, audiometry identified class A in 12 patients (24%), class B in 11 patients, class C in 22 patients, and class D in 5 patients. The mean PTA of the lesion ears from 500, 1000, 2000, and 3000 Hz was 52 ± 26 (mean ± SD) dB. No significant correlation existed between PTA and tumor size (r = 0.184, p = 0.201). Further, using linear regression analysis, the regression coefficient of PTA did not pass the t test, indicating that PTA was not significant in predicting tumor size.
Vestibular Function Test A vestibular function test battery comprising caloric, oVEMP, and cVEMP tests was performed on all 50 VS patients before treatment. On the lesion ears, 27 patients had normal caloric responses, and 23 patients (46%) had abnormal caloric responses, including canal paresis in three ears and caloric areflexia in 20 ears. Thirteen patients had normal oVEMPs, and 37 patients (74%) showed abnormal oVEMPs consisting of absent responses in 35 and delayed responses in 2. With the cVEMP test, 8 patients exhibited normal responses, and 42 patients (84%) showed abnormal cVEMPs comprising absent responses in 36 and delayed responses in 6. There was a significantly declining trend in abnormal percentages from cVEMP test (84%), followed by oVEMP test (74%) and caloric test (46%; p < 0.01, Cochran Q test). In contrast, normal responses were observed on the opposite healthy ears despite caloric, oVEMP, or cVEMP tests (Figs. 1 and 2). Registration of vestibular deficits was based on the number of abnormal test results of the caloric, oVEMP, and cVEMP tests. Grade I indicates that one of the three test results is abnormal, while grades II and III indicate two and three tests are abnormal, respectively. Finally, grade 0 means that all three tests are normal. Accordingly, grade 0 was identified in 2 patients, grade I in 11 patients, grade II in 20 patients, and grade III in 17 patients (Table 1).
573
Based on tumor size, 50 patients were classified into two groups, namely, group A had a maximum tumor size ≤1.0 cm and comprised 12 patients, whereas group B had a maximum tumor size >1.0 cm and comprised 38 patients. Thus, grades 0, I, II, and III in group A had 1, 4, 4, and 3 patients, respectively, compared with the respective 1, 7, 16, and 14 patients in group B; therefore, the two groups did not significantly differ in the distribution of grades of vestibular deficits (p > 0.05, Fisher exact test; Table 1).
Foam Posturography Versus Grades of Vestibular Deficits To determine whether the grades of vestibular deficits affect balance, all patients underwent foam posturography before treatment. No significant differences among the grades of vestibular deficits existed in terms of sway area per second with eyes open, sway area per second with eyes closed, and Romberg quotient of sway area, regardless of firm or foam surface (p > 0.05, analyses of variance test; Table 2), indicating that the grades of vestibular deficits did not affect balance.
Tumor Size Versus Vestibular Function Testing Using simple linear regression analysis, the results of vestibular function tests were used to predict the VS size, which was significantly related to caloric response (regression coefficient = 1.84, p < 0.01, t test), oVEMP response (regression coefficient = 1.78, p < 0.01), and cVEMP response (regression coefficient = 1.85, p < 0.01; Table 3). Among the three independent variables, the cVEMP response was the most significant predictor of VS size, explaining 73% of the variance in tumor size (i.e., R2), followed by the oVEMP response (60%) and caloric response (40%; Table 3). Further, a multivariate stepwise linear regression model was employed to identify significant independent predictors of VS size. As a result, linear regression model is tumor size (cm) = 0.40 (caloric response) + 1.65 (cVEMP response). However, the regression coefficient for caloric response (0.40) failed to pass the t test (p = 0.16; Table 4). Thus, caloric response is not a significant predictor of tumor size in this regression model. Alternatively, oVEMP and cVEMP responses remained significant predictors of tumor size in the regression model after backward elimination of caloric response (p < 0.05; Table 4), namely, tumor size (cm) = 0.62 × (oVEMP response) + 1.39 × (cVEMP response), where oVEMP and cVEMP responses were regarded as binary variables, in which 1 and 0 reflect abnormal and normal responses, respectively. This final model explained 76% of the variance. Accordingly, the estimated size of VS that exhibits abnormal oVEMPs and cVEMPs is >2.01 (0.62 + 1.39) cm.
DISCUSSION In the modern era, tumor control and functional preservation have become the mainstream in management of VS. Several series have reported promising outcome of SRS in terms of tumor control and hearing preservation (Pollock et al. 2006; Sughrue et al. 2010). Most tumors treated by SRS were 0.05, Fisher exact test.
a tumor for which vestibular nerve function can be preserved when SRS is performed.
Audiometry Since most VS originate from the vestibular nerve, tumor compression or stretching of the cochlear nerve accounts for the hearing loss. However, a series of studies have documented that auditory brainstem response is not sensitive in intracanalicular tumor, with only 58% sensitivity when tumor size
