Effect of Cochlear Implantation on Vestibular Evoked Myogenic Potentials and Wideband Acoustic Immittance

Abstract

Objective.

The objective of this study was to determine if absent air-conduction stimuli vestibular evoked myogenic potential (VEMP) responses found in ears after cochlear implantation can be the result of alterations in peripheral auditory mechanics rather than vestibular loss. Peripheral mechanical changes were investigated by comparing the response rates of air- and bone-conduction VEMPs as well as by measuring and evaluating wideband acoustic immittance (WAI) responses in ears with cochlear implants and normal hearing control ears. The hypothesis was that presence of a cochlear implant can lead to an air-bone gap, causing absent air-conduction stimuli VEMP responses, but present bone-conduction vibration VEMP responses (indicating normal vestibular function), with changes in WAI as compared to ears with normal hearing. Further hypotheses were that subsets of ears with cochlear implants would (a) have present VEMP responses to both stimuli, indicating normal vestibular function and either normal or near-normal WAI, or, (b) have absent VEMP responses to both stimuli, regardless of WAI, due to true vestibular loss.

Design.

27 ears with cochlear implants (age range 7–31) and 10 ears with normal hearing (age range 7–31) were included in the study. All ears completed otoscopy, audiometric testing, 226 Hz tympanometry, WAI measures (absorbance), air-conduction stimuli cervical and ocular VEMP testing through insert earphones, and bone-conduction vibration cervical and ocular VEMP testing with a mini-shaker. Comparisons of VEMP responses to air- and bone-conduction stimuli, as well as absorbance responses between ears with normal hearing and ears with cochlear implants, were completed.

Results.

All ears with normal hearing demonstrated 100% present VEMP response rates for both stimuli. Ears with cochlear implants had higher response rates to bone-conduction vibration compared to air-conduction stimuli for both cervical and ocular VEMPs; however, this was only significant for ocular VEMPs. Ears with cochlear implants demonstrated reduced low-frequency absorbance (500 – 1200 Hz) as compared to ears with normal hearing. To further analyze absorbance, ears with cochlear implants were placed into subgroups based on their cervical and ocular VEMP response patterns. These groups were 1) present air-conduction stimuli response, present bone-conduction vibration response, 2) absent air-conduction stimuli response, present bone-conduction vibration response, and 3) absent air-conduction stimuli response, absent bone-conduction vibration response. For both cervical and ocular VEMPs, the group with absent air-conduction stimuli responses and present bone-conduction vibration responses demonstrated the largest decrease in low-frequency absorbance as compared to the ears with normal hearing.

Conclusions.

Bone-conduction VEMP response rates were increased compared to air-conductive VEMP response rates in ears with cochlear implants. Ears with cochlear implants also demonstrate changes in low-frequency absorbance consistent with a stiffer system. This effect was largest for ears that had absent air-conduction but present bone-conduction VEMPs. These findings suggest that this group in particular has a mechanical change that could lead to an air-bone gap, thus, abolishing the air-conduction VEMP response due to an alteration in mechanics and not a true vestibular loss. Clinical considerations include using bone-conduction vibration VEMPs and WAI for pre-and post-operative testing in patients undergoing cochlear implantation.

INTRODUCTION

The close anatomical proximity of the cochlear and vestibular structures puts the vestibular system at risk for damage following cochlear implantation. Vestibular evoked myogenic potential (VEMP) responses are often absent in both children and adults after cochlear implantation (Jin et al. 2006; Todt et al. 2008; Licameli et al. 2009; Krause et al. 2010; De Kegal et al. 2014; Xu et al. 2015; Parkes et al. 2017; Ibrahim et al. 2017; Yong et al. 2019), suggestive of loss of otolith function. However, emerging evidence suggests that air-bone gaps may occur following cochlear implant (CI) surgery (Chole et al. 2014; Greene et al. 2015; Raveh et al. 2015; Banakis Hartl et al. 2016; Mattingly et al. 2016). Therefore, the purpose of the present study was to determine if absent air-conduction stimuli (ACS) VEMP responses found after cochlear implantation can be the result of alterations in peripheral auditory mechanics rather than vestibular loss.

Significant damage to inner ear structures has been documented in cadaveric preparations as well as in patients following CI surgery. Damage includes fibrosis in the vestibule, collapse of the saccule, decrease in ganglion cells, formation of hydrops in the inner ear, and development of scar tissue (Tien & Linthicum 2002; Choi & Oghalai 2005; Handzel et al. 2006; Roland & Wright 2006; Li et al. 2007; Fayad et al. 2009; Licameli et al. 2009). In addition, alterations in peripheral mechanics and intracochlear pressures have been found in cadavers and patients after cochlear implantation resulting in a persistent air-bone-gap (Chole et al. 2014; Greene et al. 2015; Raveh et al. 2015; Banakis Hartl et al. 2016; Mattingly et al. 2016). Due to equipment limitations, behavioral assessment of the air-bone gap in individuals with severe-to-profound hearing loss is often unable to be measured (Raveh et al. 2015). While the incidence of post-CI air-bone gaps is challenging to accurately quantify given these limitations in behavioral audiometric testing, post-operative air-bone gaps of up to 40 dB have been demonstrated in CI patient populations with residual post-surgical hearing (Chole et al. 2014; Raveh et al. 2015; Mattingly et al. 2016) and 5–10 dB air-bone gaps have been demonstrated in temporal bone preparations after insertion of an electrode array in the cochlea as well (Greene et al. 2015; Banakis Hartl et al. 2016). Several studies suggest that these air-bone gaps are largest at low frequencies (e.g., Chole et al. 2014 and Mattingly et al. 2016), though some studies have shown similarly sized air-bone gaps across frequency (Raveh et al. 2015). Regarding hearing, this air-bone gap is generally not clinically relevant as sound transduction occurs electrically with the CI (except for hybrid electroacoustic stimulation). However, if cochlear implantation results in even a small air-bone gap, this could cause reduced or absent ACS VEMP responses in the presence of normal vestibular function and result in a misdiagnosis of vestibular dysfunction. In a meta-analysis by Ibrahim et al. (2017), evidence of absent VEMP responses were present in approximately 60% (range 17 – 80%) of cochlear implant recipients’. If this rate of vestibular loss following implantation is inaccurate, this could result in inaccurate patient counseling regarding realistic effects of the surgical procedure as well as inaccurate consideration of the presence of vestibular loss when considering a second side surgery.

VEMPs are clinical tests of otolith and vestibular nerve function. The cervical VEMP (cVEMP) tests the inferior vestibular nerve and saccule and the ocular VEMP (oVEMP) tests the superior vestibular nerve and utricle (Colebatch et al. 1994; Todd et al. 2000; Govender et al. 2015). Traditionally, 125 dB SPL ACS are used to elicit VEMP responses. Higher intensity stimuli are typically not used due to equipment limitations; additionally, higher intensity stimuli result in unsafe levels secondarily delivered to the cochlea (Young 2006; Krause et al. 2010; Strömberg et al. 2015; Rodriguez et al. 2018). Normal VEMP thresholds are between 115 dB and 125 dB SPL (Janky & Shepard 2009; Rodriguez et al. 2019); thus, even a small air-bone gap of 5–10 dB could be sufficient to reduce the SPL reaching the vestibular system and abolish the ACS VEMP response. In contrast to ACS, the utility and sensitivity of utilizing bone-conduction vibration (BCV) to elicit VEMP responses has been investigated (Rosengren et al. 2019). One potential advantage of BCV is that BCV elicited VEMP responses should not be impacted by an air-bone gap (Mahdi et al. 2013; Rosengren et al. 2019).

If CI placement alters middle- or inner-ear mechanics, these alterations should be evident in the mechanical properties of the system. One method of assessing the mechanical response of the periphery (other than audiometrically through measurement of the air-bone gap) is through a non-invasive impedance-based tool called wideband acoustic immittance (WAI). WAI refers to a group of measurements (i.e., absorbance, power reflectance, admittance, and impedance, among others) used to represent the acoustic behavior of the ear. These measurements are made in the ear canal in response to wideband stimuli (e.g., a click or a chirp) and compare a sound input to the absorbed or reflected portions of that sound. As the impedance of the system changes due to pathological alterations in middle- and inner-ear mechanics, changes in WAI can, at least in some cases, be demonstrated. WAI has been shown to have clinical value in differentiating origins of conductive hearing loss, including both middle- and inner-ear causes (Feeney et al. 2003; Allen et al. 2005; Feeney et al. 2009; Shahnaz et al. 2009a; Shahnaz et al. 2009b; Nakajima et al. 2012; Voss et al. 2012; Nakajima et al. 2013; Prieve et al. 2013; Merchant et al. 2015; Merchant et al. 2016, Merchant et al., 2019). Thus, WAI may be of use in providing quantitative information about middle- and inner-ear peripheral mechanics in ears with CIs. Limited data on WAI responses in a small sample of CI ears (N=15) suggests that, at least in some cases, post-CI WAI demonstrates patterns consistent with increased stiffness below 1000 Hz (Scheperle & Hajicek 2019). However, large variability was found, and this pattern was not consistent for all CI recipients (Scheperle & Hajicek 2019), suggesting this type of change does not occur in all ears. WAI could be especially useful in CI patients, given that behavioral audiometric indications of mechanical changes to the system (i.e., air-bone gaps) generally cannot be measured in this population. Thus, WAI could help determine whether a mechanical change (and possible air-bone gap) is present, which could, in turn, lead to absent ACS VEMP responses without true vestibular loss.

The purpose of the present study was to determine if cochlear implantation alters peripheral auditory mechanics in some ears with CI, thus resulting in reduced or absent ACS VEMP responses when there is no vestibular loss present. We hypothesize that placement of a CI in the cochlear cavity leads to a mild air-bone gap in some ears. Consequently, these ears will present with reduced or absent ACS VEMP responses and changes in WAI due to a change in peripheral mechanics leading to an air-bone gap and not true vestibular loss. If this is the case, BCV VEMP responses will be present as BCV should be unaffected by the air-bone gap. A subset of CI ears will demonstrate present ACS and BCV VEMPs and normal WAI, suggesting normal or near-normal peripheral auditory mechanics and vestibular function. Last, a subset of ears will present with both absent ACS and BCV VEMP responses, regardless of WAI, due to the presence of true vestibular loss.

MATERIALS AND METHODS

Study Population

CI Sample: 27 ears with CI from 19 participants (mean age = 14.74, range 7–31, including 8 bilateral CI recipients) were included in the study population. All participants with CI received a full audiometric assessment including pure-tone air- (250–8000 Hz) and bone-conduction (250–4000 Hz) thresholds (Madsen Astera², Otometrics, Schaumberg, IL). CI participants also underwent otoscopy, 226 Hz probe tone tympanometry, and a screening questionnaire to rule out confounding middle-ear abnormalities or pathology at the time of testing. Participants were also asked whether they had any history of dizziness or imbalance. Responses were tallied as yes or no. Ears with CI had ear canal volumes (ECV) ≤ 2.0 ml and tympanometric peak pressure between −100 and 30 daPa (British Society of Audiology 2013). Strict tympanic membrane admittance criteria were not utilized in this population as the changes in auditory mechanics hypothesized here could contribute to changes observed in admittance and will be investigated directly as part of this work. Etiology of hearing loss, results of imaging, age at implantation, surgical approach, and implant type were gathered from the medical chart when available.

Normal Control Sample: 10 ears with normal hearing (NH) from 10 participants served as the control group and were age-matched to the CI participants (mean age = 16.53, range 7–31). All control participants were characterized by having unremarkable otoscopy, normal hearing sensitivity, normal middle-ear function, and no history of balance disorder, dizziness, neurologic involvement, or middle-ear surgery other than a single occurrence of tympanostomy tube placement. All participants completed a hearing screening at 25 dB HL for 1000, 2000, and 4000 Hz (ASHA 1997). To rule out the presence of middle-ear pathology, all participants underwent 226 Hz probe tone tympanometry (Titan, Interacoustics, Middlefart, DK) with the following inclusion criteria: ECV ≤ 2.0 ml, admittance ≥ 0.2 mmhos, and tympanometric peak pressure within the range of −100 to 30 daPa (British Society of Audiology 2013).

Informed consent was obtained from all participants for testing approved by the Institutional Review Board at Boys Town National Research Hospital.

Vestibular Evoked Myogenic Potential (VEMP)

VEMP methods have been reported previously (Rodriguez et al 2019). In short, both c- and oVEMPs were administered using: 1) air-conducted 500 Hz tone bursts using ER-3A insert phones (Otometrics, Schaumberg, IL) and 2) BCV 500 Hz tone bursts using the 4810 Mini-shaker (Brüel & Kjær, Nærum, DK). ACS was presented at 125 dB pSPL for ECVs > 0.8 ml or 120 dB pSPL for ECVs ≤ 0.8 ml (Rodriguez et al., 2018). BCV was presented at 147.6 dB FL. For each VEMP, participants lay supine on an exam table. For cVEMP, participants were instructed to raise their head straight up, nose to the ceiling, activating the right and left sternocleidomastoid (SCM) muscles simultaneously during the recording. For oVEMP, participants were instructed to gaze 30 degrees upward at a target on the ceiling, activating the inferior oblique muscles. For the mini-shaker, the examiner stood behind the participant, placed the mini-shaker on the midline (Fz), and ensured stability of the mini-shaker. The main outcome parameters were amplitude, corrected amplitude (raw peak-to-peak amplitude/raw EMG), and presence/absence of responses.

VEMP Stimuli and Recording Parameters

ACS and BCV 500 Hz tone bursts (Blackman gated; 1 ms rise/fall, 0 ms plateau, condensation polarity) were presented at a rate of 5.1 Hz. At least two trials of 75 repetitions were completed to determine replicability. If replication did not occur, the response was considered absent. When replication did occur, the 2 waveforms were averaged. Amplitude and corrected amplitude were measured on the averaged waveform. For cVEMPs, the ground electrode was placed on the right inner canthi, the reference electrode was placed on the sternum, and active electrodes were placed on the belly of the right and left SCM. EMG monitoring electrodes were placed just below each active electrode (amplified 5000x and band-pass filtered from 5 to 500 Hz) to ensure recordings were within 100 to 300 μV (Bogle et al. 2013). For oVEMPs, the ground electrode was placed on the sternum, the reference electrode was placed on the right inner canthi, and active electrodes were placed inferior and laterally on the inferior oblique extraocular eye muscles.

Wideband Acoustic Immittance (WAI)

WAI was measured in both ears for each participant using a Titan probe (Interacoustics, Middleafart, DK). Custom software (MATLAB, MathWorks, Inc., Natick, MA) built on the Titan Research Platform (Interacoustics) was utilized for both WAI calibration and data collection. In order to obtain the source pressure and impedance of the system, a calibration procedure (e.g., a Thévenin calibration) was completed daily prior to each test session in a four-cavity calibrator (Interacoustics) according to the Titan Research Platform specifications and published methods (Nørgaard et al. 2017). WAI was measured in response to a wideband chirp stimulus (220–8000 Hz) in an ambient condition in each ear. Responses were inspected in both real-time and post-processing for the presence of air-leaks (using the criteria reported in Groon et al. 2015). Measurements were repeated if indications of leaks were present and noted in real-time and as needed due to participant noise (e.g., if the patient yawned or spoke during the recording or if a proper seal was not achieved). A minimum of two qualitatively identical responses were collected for each ear to ensure reliability of the response, and a single tracing of these two was chosen at random for analysis. Responses were smoothed across frequency in 1/6^th octave bands.

A number of responses can be calculated from WAI measurements. These include, but are not limited to, the impedance, the pressure reflectance, the power reflectance, and the absorbance. The pressure reflectance R(f) of the ear is defined as the ratio of the reflected pressure to the forward pressure. Pressure reflectance is related to the impedance at the point of measurement by the following equation:

$$\text{R}\left(f\right)=\frac{\left(Z\left(f\right)-z_0\right)}{\left(Z\left(f\right)+z_0\right)}$$ (1)

where f is frequency, Z(f) is the frequency-dependent impedance looking into the ear canal, and z_o is the characteristic acoustic impedance of the ear canal at the measurement point. Power reflectance (PR) is the squared magnitude of the pressure reflectance |R(f)|² and is a number between 0 and 1, where PR = 0 suggests that all power is transmitted to the ear and where PR = 1 represents that all power is reflected back into the ear canal at the tympanic membrane. Absorbance is the inverse of PR and represents the sound power absorbed by the system near the tympanic membrane (Absorbance = 1 – PR). Both PR and absorbance have been investigated widely in the literature, mainly because these measures are minimally influenced by the location the probe in the ear canal, which is advantageous for clinical applications. Absorbance shares similar properties to admittance with respect to interpretation of the response, and has thus become the most common outcome parameter reported in translational studies of WAI. Therefore, the main outcome parameter utilized in this work is absorbance.

Statistics

SPSS Statistics v. 22.0 software was used to perform statistical analyses. Chi-square analyses were used to compare VEMP response rates between the ears with NH and ears with CI. A mixed groups analysis of variance (ANOVA) was completed to investigate mean differences in VEMP amplitude with group (NH, CI) as the between subjects variable and stimulus (ACS, BCV) as the within subjects variable. One-way ANOVA was completed to investigate mean differences in the percent difference between ACS cVEMP amplitudes and BCV cVEMP amplitudes and all tympanometry measures. For WAI, absorbance data were initially averaged into 1/3^rd octave bands from 220–8000 Hz, reducing the number of data points from 364 to 16 (such as in Prieve et al. 2013). A principal component analysis (PCA) was then conducted on the 1/3^rd octave band averaged absorbance data, further reducing the data points from 16 to 4 frequency bands based on the 4 PCA factors, as detailed below. PCA frequency bands were analyzed using a mixed groups ANOVA, with group (NH, CI) as the between subjects variable and PCA frequency band as the within subjects variable. Significant interactions were followed by post hoc t tests using Bonferroni-adjusted alpha levels (α = 0.05, family-wise) as needed. Effect estimates, including partial η2, Cohen’s d, and effect size r were calculated by SPSS or calculated based on the means, standard deviations, and sample sizes for each individual test.

Data Selection

37 ears were included in the final study population (27 CI ears and 10 ears with NH). However, 75 ears were assessed in the laboratory as part of the initial study population: 45 ears with CI from 28 participants (mean age = 15.29, range 7–46, including 17 bilateral CI recipients) and 30 ears with NH (mean age = 16.53, range 7–31). 38 ears were removed from the final study population for the following various reasons. During cVEMP, 2 participants were unable to generate EMG values greater than the accepted minimum value of 100 μV, which excluded 2 ears with CI. Post-processing analyses of absorbance data identified 7 ears with responses affected by acoustic air-leaks, which excluded 3 ears with NH and 4 ears with CI (Groon et al. 2015). Technical problems with the measurement setup during a single data collection session resulted in invalid data, which excluded 2 ears with CI. Previous middle-ear surgery excluded 1 ear with NH and 4 ears with CI. Presence of an active ear infection excluded 1 ear with a CI. To minimize previous tympanostomy tube placement effects on WAI (Kazikdas et al. 2006), participants were excluded if they had more than one previous tube placement, which excluded 6 ears with NH and 3 ears with CI. Hearing loss caused by meningitis was an initial exclusion criterion due to possible bone ossification leading to inner-ear stiffening (Nadol & Hsu 1991), and this criterion also excluded 2 ears with CI by medical record search after data collection. Of the remaining ears with NH, one ear was randomly selected for each participant to not bias normal statistics towards the subjects with two normal ears. In some cases, after previously listed exclusion criteria were considered, only one ear remained and thus was the ear used (ultimately excluding 10 ears with NH). In cases of bilateral implantation, each ear with CI was included. Because cochlear implantation alters the normal state of the ear, it is unlikely that implantation impacted each ear identically, thus reducing the likelihood of statistical bias. Furthermore, all bilateral recipients were sequentially (not simultaneously) implanted. The time between implants ranged between 1 and 6 years. In some instances, different internal components were implanted, different surgeons completed the surgery, and the majority of ears had an unknown etiology.

RESULTS

Demographics

Table 1A shows the clinical characteristics of the 27 ears with CI, including etiology of hearing loss, implant type, surgeon, surgical approach, and insertion depth. Table 1B shows the combined clinical characteristics of the 27 ears with CI. Two patients were reimplanted after device failure; only information after reimplantation was included in Table 1A–B. Air-bone gaps could not be quantified for any CI participant as none of the 27 ears with CI had measurable bone-conduction thresholds at any frequency on behavioral audiometric testing to the limits of the equipment. Of the 27 ears with CI, 12 (44.4%) ears were associated with a participant responding yes when asked whether they subjectively had any history of dizziness of imbalance.

Table 1A:

Clinical characteristics of 27 ears with CI

CI Ears	Etiology	Implant Type	Surgeon	Surgical Approach	Insertion Depth
1	Unknown	AB HiRes Ultra/HiFocus ms	C	Round Window	Full
2	Unknown	AB HR90K/HiFocus1J	A	Cochleostomy	Full
3	Unknown	AB HR90K/HiFocus1J	A	Cochleostomy	Full
4	Pendred	AB CII Bionic Ear/HiFocus1J	B	Cochleostomy	Full
5	Unknown	AB HR90K/HiFocus1J	A	Cochleostomy	Full
6	Unknown	AB HR90K/HiFocus1J	A	Cochleostomy	Full
7	Unknown	AB CII Bionic Ear /HiFocus1J	B	Cochleostomy	Full
8	CMV	AB HR90K/HiFocus1J	A	Cochleostomy	Full
9	Unknown	AB HR90K/HiFocus1J	A	Cochleostomy	Full
10	Unknown	AB HR90K/HiFocus1J	A	Cochleostomy	Full
11	Pendred	AB Clarion 1.2	B	Unknown	Unknown
12	Mondini Malformation	Cochlear Nucleus 24K CI24R(ST)	B	Cochleostomy	Full
13	Genetic	AB HR90K ADV/HiFocus1J	C	Cochleostomy	Full
14	Unknown	AB HR90K/HiFocus1J	B	Cochleostomy	Full
15	Unknown	AB HR90K/HiFocus1J	A	Cochleostomy	Full
16	Unknown	Cochlear Nucleus Freedom CI24RE(CA)	A	Cochleostomy	Full
17	Unknown	Cochlear Nucleus Freedom CI24RE(CA)	A	Cochleostomy	Full
18	Unknown	AB HR90K ADV/HiFocus1J	C	Cochleostomy	Full
19*	Unknown	AB HR90K ADV/HiFocus1J	C	Cochleostomy	Full
20	Connexin 26	AB CII Bionic Ear/HiFocus1J	B	Unknown	Unknown
21*	Mondini Malformation	AB HR90K ADV/HiFocus ms	C	Round Window (Initially implanted by Cochleostomy)	Full
22	Genetic	AB HR90K/HiFocus1J	A	Cochleostomy	Full
23	Genetic	AB HR90K/HiFocus1J	A	Cochleostomy	Full
24	Unknown	AB HR90K/HiFocus1J	A	Cochleostomy	Full
25	Unknown	AB HR90K/HiFocus1J	A	Cochleostomy	Full
26	Connexin 26	AB CII Bionic Ear/HiFocusIJ	B	Cochleostomy	Partial
27	Connexin 26	AB HR90K/HiFocus1J	A	Cochleostomy	Full

^*= reimplanted ear,

CMV = Cytomegalovirus, AB = Advanced Bionics, ADV = advantage

Table 1B:

Clinical characteristics totals by category

CI Ears	Etiology	Implant Type	Surgeon	Surgical Approach	Insertion Depth
N = 27	CMV = 1 Unknown = 16 Pendred = 2 Connexin = 3 Genetic = 3 Mondini Malformation = 2	AB HiRes Ultra/HiFocus ms = 1 AB HR90K/HiFocus1J = 14 AB CII Bionic Ear/HiFocus1J = 4 AB HR90k ADV/HiFocus1J = 3 Cochlear Nucleus 24K CI24(ST) = 1 Cochlear Nucleus Freedom CI24RE(CA) = 2 AB Clarion 1.2 = 1 AB HR90k ADV/HiFocus ms = 1	A = 15 B = 7 C = 5	Cochleostomy = 23 Round Window = 2 Unknown = 2	Full = 24 Partial = 1 Unknown = 2

CMV = Cytomegalovirus, AB = Advanced Bionics, ADV = advantage

VEMP

All subjects completed c- and oVEMP testing in response to 500 Hz tone bursts via ACS and BCV. Response rates for each group and stimulus type are shown in Table 2. All ears with NH had 100% response rates to both stimuli. The ears with NH had significantly higher response rates for each stimulus type compared to ears with CI (p = .036 - < .001, Table 2). While the ears with CI had higher response rates to BCV compared to ACS, this was not significant for cVEMP (X²(1) = 3.65, p = 0.056); however, this was significant for oVEMP (X²(1) = 8.33, p = 0.004).

Table 2.

Response rates for cervical and ocular VEMP to ACS and BCV stimuli.

	CERVICAL VEMP		OCULAR VEMP
GROUP	ACS	BCV	ACS	BCV
Ears with NH (n = 10)	100%	100%	100%	100%
Ears with CI (n = 27)	41%	67%	15%	52%*
Chi-Square p-value	10.41 0.001	4.41 0.036	22.51 <0.001	7.42 0.006

^*Significantly higher response rates compared to ACS (p =.004)

ACS = air conducted stimuli, BCV = bone conducted vibration, NH = normal hearing, CI = cochlear implant

A mixed groups ANOVA was completed to examine the effect of a CI on c- and oVEMP amplitudes. All ears with a no response were given an amplitude value of 0. Descriptive data can be found in Table 3. For cVEMP amplitude there was no significant interaction (F(1, 35) = 1.717, p = 0.199, partial η2 = 0.047) and no significant effect of stimulus type (F(1, 35) = 3.887, p = 0.057, partial η2 = 0.1); however, there was an effect of group (F(1, 35) = 48.095, p < 0.001, partial η2 = 0.579). Ears with NH had significantly higher cVEMP amplitudes compared to ears with CI (Table 3). The percent difference between ACS cVEMP amplitudes and BCV cVEMP amplitudes was calculated and there was no significant difference between groups (F(1,36) = 3.326, p = 0.077, Cohen’s d = 0.592; Effect Size r = 0.291, NH: 2.3%, CI: 33.2%).

Table 3.

Mean (SD) cervical and ocular VEMP amplitudes for ears with Normal Hearing (NH) and ears with a Cochlear Implant (CI) in response to 500 Hz tone burst ACS and BCV

	VEMP AMPLITUDE
	CERVICAL (μv)		CERVICAL CORRECTED		OCULAR (μv)
GROUP	ACS	BCV	ACS	BCV	ACS	BCV
Ears with NH	401.68 (144.41)	420.03 (141.9)	2.73 (1.12)	2.54 (.74)	14 (8.53)	17.39 (11.74)
Ears with CI	78.34 (110.39)	169.43 (149.67)	.57 (.83)	1.19 (1.03)	2.23 (6.64)	6.74 (10.02)

Similarly for cVEMP corrected amplitude, there was no significant interaction (F(1, 35) = 3.92, p = 0.056, partial η2 = 0.101) and no significant effect of stimulus type (F(1, 35) = 1.063, p = 0.31, partial η2 = 0.029); however, there was an effect of group (F(1, 35) = 38.965, p < 0.001, partial η2 = 0.527). Ears with NH had significantly higher corrected cVEMP amplitudes compared to ears with CI (Table 3). There was no significant difference in the percent difference between ACS cVEMP corrected amplitudes and BCV cVEMP corrected amplitudes between groups (F(1,36) = 4.026, p = 0.053, Cohen’s d = 0.651; Effect Size r = 0.317, NH: −1.76%, CI: 32.6%).

For oVEMP, there was no significant interaction (F(1, 35) = 0.059, p = 0.81, partial η2 = 0.002) and no significant effect of stimulus type (F(1, 35) = 2.897, p = 0.098, partial η2 = 0.076); however there was an effect of group (F(1, 35) = 22.148, p < 0.001, partial η2 = 0.388). Ears with NH had significantly higher oVEMP amplitudes compared to ears with CI (Table 3). There was no significant difference in the percent difference between ACS oVEMP amplitudes and BCV oVEMP amplitudes between groups (F(1,36) = 1.77, p = 0.192, Cohen’s d = 0.432; Effect Size r = 0.216, NH: 7.3%, CI: 34.6%).

These results suggest that ears with CI have reduced VEMP amplitudes compared to ears with NH; however, despite the increased response rates to BCV, there was not an effect or interaction for stimulus type. This suggests differential patterns of VEMP results in the ears with CI. Thus, ears with CI were placed in subgroups based on their c- and oVEMP response patterns: Present-Present (CI PP, present ACS response, present BCV response), Absent-Present (CI AP, absent ACS response, present BCV response), or Absent-Absent (CI AA, absent ACS response, absent BCV response), shown in Table 4. It was possible for a participant to have a Present-Absent pattern (CI PA, present ACS response and absent BCV response); however, only a single ear fell into this category in both the c- and oVEMP subgroupings (different ear in each case). Given an n of 1 for that group, that ear was excluded from analyses of subgroups. The mean percent difference between ACS and BCV is shown by group and VEMP type (Table 4); negative values indicated that ACS was greater than BCV. There were significant mean differences in the percent difference for cVEMP amplitude (F(3,35) = 65.726, p < 0.001, Cohen’s d = 0.432; Effect Size r = 0.923), cVEMP corrected amplitude (F(3,35) = 58.328, p < 0.001, Cohen’s d = 0.432; Effect Size r = 0.914) and oVEMP amplitude (F(3,35) = 45.757, p < 0.001, Cohen’s d = 0.432; Effect Size r = 0.893) between groups. For all outcomes, the CI AP group had a significantly (p < 0.001) higher percent difference (100%) compared to all other groups (NH, CI PP, and CI AA); there were otherwise no mean differences between the remaining groups (NH, CI PP, and CI AA). Of note, with respect to the screening question relating to a history of dizziness or imbalance, when ears with CI were separated into these subgroups (cVEMP subgroupings), only 20% of ears in the CI PP subgroup and 37.5% of ears in the CI AP subgroup were associated with a participant who responded yes to a history of dizziness or imbalance, while 75% of ears in the CI AA subgroup were associated with a participant who responded yes.

Table 4.

Subgroups based on cervical and ocular VEMP Response Patterns

		Ears with NH	Ears with CI Subgroup
		Present ACS Present BCV (NH)	Present ACS Present BCV (CI PP)	Absent ACS Present BCV (CI AP)	Absent ACS Absent BCV (CI AA)
cVEMP Group	n =	n = 10	n = 10	n = 8	n = 8
	ACS - BCV	2%	20%	100%	0%
oVEMP Group	n =	n = 10	n = 3	n = 11	n = 12
	ACS - BCV	3%	−7%	100%	0%

226 Hz Tympanometry

The standard clinical test of immittance is 226 Hz tympanometry. Prior to considering the WAI data, results from 226 Hz tympanometry were compared to determine whether any differences occur in admittance and/or tympanometric peak pressure between groups or subgroups. Table 5 displays the means and standard deviations (SD) for both the admittance and tympanometric peak pressure when grouped by ears with NH versus ears with CI, and when separating ears with CI into VEMP subgroups. There were no significant mean differences in admittance (p = 0.709), pressure (p = 0.213), or ECV (p = 0.623) between ears with NH and ears with CI. When separated into c- and oVEMP subgroups, there remained no significant mean differences in admittance (p = 0.916, p = .562), pressure (p = 0.212, p = 0.472), or ECV (p = 0.52, p = 0.581) between groups. Overall, this demonstrates that there are no significant differences between ears with NH and CI with respect to standard clinical measures of immittance in any form of grouping.

Table 5.

Tympanometric data displayed by group and subgroups.

Admittance (mmhos)
	Mean	SD
NH	0.77	0.29
CI	0.71	0.46
Pressure (daPa)
	Mean	SD
NH	−7.60	17.54
CI	0.81	18.05
cVEMP Group
Admittance (mmhos)
	Mean	SD
NH	0.77	0.29
CI PP	0.63	0.41
CI AP	0.72	0.55
CI AA	0.81	0.50
Pressure (daPa)
	Mean	SD
NH	−7.60	17.54
CI PP	−6.50	10.32
CI AP	9.00	26.85
CI AA	3.75	12.37
oVEMP Group
Admittance (mmhos)
	Mean	SD
NH	0.77	0.29
CI PP	0.77	0.73
CI AP	0.57	0.21
CI AA	0.85	0.57
Pressure (daPa)
	Mean	SD
NH	−7.60	17.54
CI PP	−9.67	3.79
CI AP	−1.73	11.92
CI AA	−5.17	24.01

Wideband Acoustic Immittance: Absorbance

Figure 1a displays the overall means for the ears with NH (black) and ears with CI (dashed green). The shaded regions represent +/− the standard error of the mean (SEM). Data from the ears with NH are consistent with published data (e.g., Beers et al., 2010; Rosowski et al., 2012). Absorbance in ears with CI is reduced as compared to the ears with NH between 500 and 1200 Hz and slightly increased around the peak (resonance) at 2000 Hz. To compare the ears with CI to ears with NH in more detail, Figure 1b displays the mean +/− 1 standard deviation (SD) of the ears with NH (thick black line and gray shading) with individual tracings overlying this mean for each ear with CI. This figure suggests that, while there is a trend for ears with CI to have lower absorbance from 500 Hz to 1200 Hz, there is a wide range of variability in the individual absorbance data for ears with CI, consistent with previous findings (Scheperle & Hajicek 2019).

Figure 1.

a) Means +/− the standard error of the mean (SEM) for all 10 ears with NH in gray and all 27 ears with CI in dashed green. b) NH mean +/− 1 standard deviation (SD, thick black line, gray shading) with individual absorbance tracings for each of the 27 ears with CI (thin black).

For the purpose of analyses, absorbance data were initially averaged into 1/3^rd octave bands from 250–8000 Hz, reducing the number of data points from 364 to 16. A principal component analysis (PCA) was then conducted on the 1/3^rd octave band averaged absorbance data (16 frequency bands) for all 37 ears (10 NH and 27 CI) with varimax rotation. The Kaiser-Meyer-Olkin measure verified the sampling adequacy for the analysis, KMO=0.616, which is above the acceptable limit of 0.5 (Field 2013). An initial analysis was run to obtain eigenvalues for each factor in the data. Four factors had eigenvalues over Kaiser’s criterion of 1 and in combination explained 91.08% of the variance. These four factors were retained and a varimax rotated analysis was completed on these 4 factors, resulting in the eigenvalues, variance explained, and factor loading displayed in Table 6. Only factor loadings greater than 0.5 were considered and are shown in the table. If a 1/3^rd octave frequency band had a factor loading greater than 0.5 for more than 1 factor, that band was included as part of the factor that it loaded onto more significantly. This only occurred at the frequency boundaries between factors. No band was included in more than 1 factor. The final 4 factors created four frequency bands considered for further analyses: 220–707 Hz, 708–1122 Hz, 1123–4489 Hz, and 4490–8000 Hz (Table 6).

Table 6.

Summary of data from principal component analysis on the 1/3^rd octave band averaged absorbance data (16 frequency bands) for all 37 ears (10 NH and 27 CI) with varimax rotation.

1/3rd Octave Band Center Frequency	Component
	1	2	3	4
Eigenvalue	4.766	2.713	4.183	2.911
Percent Variance Explained	29.788%	16.958%	26.14%	18.19%
250	0.954
314.980263	0.968
396.850263	0.947
500	0.879
629.960525	0.732	0.647
793.700526	0.514	0.826
1000		0.871
1259.92105		0.572	0.589
1587.40105			0.87
2000			0.926
2519.8421			0.893
3174.8021			0.853
4000			0.679	0.659
5039.6842				0.909
6349.60421				0.932
8000				0.637
Factor Frequency Range (Hz)	220–707	708–1122	1123–4489	4490–8000

The 1/3^rd octave band absorbance data was then averaged into these four frequency bands, which will hereafter be referred to as PCAf1 (220–707 Hz), PCAf2 (708–1122 Hz), PCAf3 (1123–4489 Hz), and PCAf4 (4490–8000 Hz).

To determine whether the mean absorbance for the ears with NH and CI were significantly different, a 2 × 4 mixed groups ANOVA, was performed with group (NH, CI) as the between subjects variable and PCA frequency band (PCAf1, PCAf2, PCAf3, and PCAf4) as the within subjects variable. There was a significant interaction between PCA frequency band and group (F (1,3) = 3.973; p = 0.010; partial η2 = 0.102), indicating that the pattern of absorbance across PCA frequency bands was different for the two groups. To further investigate this significant interaction, post hoc testing was performed between the two groups for the 4 PCA frequency bands (Bonferroni-corrected α = 0.05/4 = 0.0125). Significant differences were observed between the ears with NH and CI in PCAf2 (t(35) = 2.747; p = 0.009; Cohen’s d = 0.758; Effect Size r = 0.346) indicating that the two groups were significantly different at frequencies from 708 to 1122 Hz, with ears with CI demonstrating reduced absorbance as compared to ears with NH in this frequency range.

Ears with CI were then separated into subgroups based on their cVEMP and oVEMP responses to ACS and BCV stimuli (Table 4) and analyzed separately.

Figure 2 displays the absorbance means for the ears with NH (black) and for each of the CI subgroups (CI PP, large dashed red; CI AP, dotted teal; and CI AA, small dashed purple). The shaded regions represent +/− SEM. The cVEMP subgroups are displayed in panel a and the oVEMP subgroups are displayed in panel b. With the exception of the oVEMP CI PP group, the overall trend in the subgroups is consistent with what was observed in Figure 1 for the ears with CI group as a whole, demonstrating reduced absorbance from 500 Hz to 1200 Hz, and some evidence of increases or changes in absorbance near the peak (resonance) between 2000 and 4000 Hz. In the cVEMP subgroups, this trend is largest in the CI AP group (dotted teal), where absorbance is decreased by up to 30% (0.3 absorbance units) as compared to ears with NH from 200 Hz through 1500 Hz. The CI PP and CI AP group demonstrate smaller reductions in low-frequency absorbance and in a smaller frequency range (500 to 1000 Hz) than the CI AP group. In contrast, for the oVEMP subgroups, the CI PP group appears similar to the ears with NH in the low frequencies, with a trend towards reduced absorbance above 2000 Hz, whereas the CI AP and CI AA groups appear similar to each other and to the ears with CI as a whole, with reduced absorbance from 500 Hz to 1200 Hz. Note that the oVEMP CI PP group has a small sample size (n=3) as compared to the other subgroups, so most of the ears with CI end up in the CI AP or CI AA subgroups when grouped based on their oVEMP responses.

Figure 2.

Means +/− SEM for NH ears (black) and CI VEMP subgroups CI PP (large dashed red), CI AP (dotted teal) and CI AA (small dashed purple). Subgroups based on cVEMP responses are shown in panel a and subgroups based on oVEMP responses are shown in panel b.

To explore the subgrouped data in more detail, Figure 3 displays the mean of the ears with NH +/− 1 SD with individual tracings for each ear with CI for all cVEMP and oVEMP subgroups. The top 3 panels display the cVEMP data and the bottom 3 panels display the oVEMP data, with the subgroups CI PP, CI AP, and CI AA being displayed from left to right. For the cVEMP subgroups, similar to the ears with CI as a whole, the trend for reduced low-frequency absorbance is visible for all 3 subgroups, but is most apparent in the CI AP group, with wider variability in the CI PP and CI AA subgroups. For the oVEMP subgroups, the CI PP group only has an n=3, so comparisons are limited. There is a slight tendency for the CI AP group to display more individual tracings with reduced low-frequency absorbance than the CI AA group, but much like the mean results (Figure 2), the CI AP and CI AA groups when grouped based on oVEMP response rates appear qualitatively similar overall.

Figure 3.

Mean of ears with NH (n=10, thick black) +/− 1 SD displayed in all 6 panels, with individual CI data overlaid depending on VEMP subgrouping. The top 3 panels display cVEMP subgroups. The bottom 3 panels display oVEMP subgroups. The subgroups from left to right are CI PP, CI AP, and CI AA.

To determine whether the mean absorbance for the NH and CI cVEMP subgrouped ears were significantly different, a 4 (group) × 4 (PCA frequency band) mixed ANOVA, repeated on the last factor, was performed. There was a significant interaction between PCA frequency band and group (F (3, 9) = 3.223; p = 0.002; partial η2 = 0.232), indicating that the pattern of absorbance across PCA frequency bands was different across cVEMP subgroups. To further investigate this significant interaction, post hoc testing was performed between each of the groups on the 4 PCA frequency bands (Bonferroni-corrected α = 0.05/4 = 0.0125). Significant differences were observed between the NH and CI AP group in two of the four PCA frequency bands, PCAf1 (t(16) = 3.679; p = 0.002; Cohen’s d = 1.851; Effect Size r = 0.661) and PCAf2 (t(16) = 5.580; p < 0.001; Cohen’s d = 2.807; Effect Size r = 0.801) indicating that the CI AP group had significantly lower absorbance compared to the NH group at frequencies from 220 to 1122 Hz. In addition, significant differences were observed between the CI AP group and the CI AA group in PCAf2 (t(14) = 2.990; p = 0.010; Cohen’s d = 1.598; Effect Size r = 0.603), indicating the CI AP group also had significantly lower absorbance than the CI AA group from 708–1122 Hz.

To determine whether the mean absorbance for the NH and CI oVEMP subgrouped ears were significantly different, a 4 (group) × 4 (PCA frequency band) mixed ANOVA, repeated on the last factor, was performed. There was a significant interaction between PCA frequency band and group (F (3, 9) = 2.422; p = 0.016; partial η2 = 0.185), indicating that the pattern of absorbance across PCA frequency bands was different across oVEMP subgroups groups. To further investigate this significant interaction, post hoc testing was performed between each of the groups on the 4 PCA frequency bands (Bonferroni-corrected α = 0.05/4 = 0.0125). Significant differences were only observed between the NH and CI AP ears in one of the four PCA frequency bands, PCAf2 (t(19) = 3.287; p = 0.004; Cohen’s d = 1.510; Effect Size r = 0.587) indicating the CI AP group had significantly lower absorbance than the NH group from 708–1122 Hz. Interestingly, the difference between the NH group and the CI AA group, while qualitatively similar to the CI AP group, did not reach statistical significance, though it did approach it (t(20) = 2.614; p = 0.017, not significant at 0.0125 level). This may be due to the slight trend towards more individual ears in the CI AP group than the CI AA group having reduced low-frequency absorbance that was observed in Figure 3.

DISCUSSION

The purpose of the present study was to determine if absent ACS VEMP responses found after cochlear implantation can be the result of alterations in peripheral auditory mechanics rather than vestibular loss. We hypothesized that placement of a CI in the cochlear cavity would lead to an air-bone gap in some ears, thus resulting in reduced or absent ACS VEMP responses and changes to WAI. Consistent with this hypothesis, ears with a CI had higher VEMP response rates to BCV compared to ACS. While measurement of an air-bone gap was not possible in these ears due to the severity of their hearing loss, WAI demonstrated decreased low-frequency absorbance changes compared to age-matched ears with NH consistent with a stiffer system, suggesting that in some ears, a mechanical change could be present leading to a conductive component to the hearing loss.

Vestibular Function and Vestibular Evoked Myogenic Potentials

There have been various reports of vestibular function following cochlear implantation. Results of these studies have varied based on whether tests of otolith (c- and oVEMP) or canal (rotary chair, calorics, or video head impulse testing (vHIT)) function were used and whether the recipients were adults versus children. In a recent meta-analysis, Ibrahim and colleagues (2017) report no significant effect of implantation on head impulse testing; however, a significant effect of implantation was noted on both caloric and cVEMP testing. Focusing specifically on children, a systematic review and meta-analysis by Yong and colleagues (2019) report no significant effect of implantation on caloric testing; however, a significant effect of implantation was noted on cVEMP testing. While variable findings regarding canal function have been reported, the most frequently documented abnormality following implantation is reduced or absent ACS cVEMP (Jin et al. 2006; Melvin et al. 2009; Licameli et al. 2009; Krause et al. 2010; Katsiari et al. 2013; Ibrahim et al. 2017; Yong et al. 2019). The current study is consistent with these findings in that ears with NH had significantly higher c- and oVEMP amplitudes and response rates compared to ears with CI. While the ears with NH had response rates of 100% for ACS cVEMP and oVEMP, the ears with CI had an ACS cVEMP response rate of 41% and an ACS oVEMP response rate of 15%. Interestingly, response rates using BCV increased to 67% for cVEMP and 52% for oVEMP. While this was only statistically significant for oVEMP, this trend was also present for cVEMP. Given our small sample size, we are likely underpowered (r = 0.31). Nonetheless, the higher response rates using BCV suggest that using ACS may result in a false positive diagnosis of vestibular loss following cochlear implantation in some patients. This finding is also consistent with the ACS VEMP response being abolished due to a conductive component in some ears with CI. ACS VEMP responses may be particularly vulnerable to these effects, given that the literature reports larger air-bone gaps post-CI in low frequencies and the fact a 500 Hz stimulus utilized to elicit the VEMP response.

Because there was not a significant effect of stimulus type and no significant percent difference between ACS and BCV VEMP, ears with CI were placed in subgroups based on their c- and oVEMP response patterns. Of note, the response rate trends were different for cVEMP versus oVEMP; therefore, the ears in the CI PP, CI AP, and CI AA groups were different for the cVEMP and oVEMP subgroups. These differences in subgroups between cVEMP and oVEMP are likely attributed to differences in VEMP thresholds, where oVEMP thresholds are ~ 5 – 10 dB higher than cVEMP thresholds (Park et al. 2010; Rodriguez et al. 2019). As a result of this threshold difference, oVEMPs would be more susceptible to an air-bone gap following implantation, thus resulting in lower ACS response rates.

The underlying mechanism leading to vestibular loss following cochlear implantation is unknown. While several theories have been postulated, vestibular damage following cochlear implantation has been largely attributed to either the cochleostomy or insertion of the CI (Licameli et al. 2009). In turn, either of these factors could lead to vestibular damage due to acoustic trauma in the form of intralabyrinthine pressure transients from insertion of a CI, labyrinthitis, reaction to a foreign body, or endolymphatic hydrops (Kubo et al. 2001; Fina et al. 2003; Katsiari et al. 2013; Banakis Hartl et al. 2018). Alternatively, a CI may also be placed via a round window approach. Todt et al. (2008) found a lower rate of vestibular loss in patients with round window compared to cochleostomy approaches. Degree of vestibular loss has also been attributed to the size and shape of the electrode, with increased risk of vestibular loss when using larger, straight arrays (Licameli et al. 2009).

In addition to vestibular damage, emerging literature suggests the possibility of cochlear implantation causing a post-operative air-bone gap (Chole et al. 2014; Banakis Hartl et al. 2016; Mattingly et al. 2016). However, the physiologic cause of this conductive component is not well understood. Possible mechanisms include a small 5 – 10 dB loss due to the presence of the electrode array itself (leading to increased cochlear pressure), damage from the array to middle-ear or intracochlear structures, loss of perilymph during surgery, inflammation and fibrosis surrounding the electrode array, and/or a third window effect due to a cochleostomy, if applicable (Somdas et al. 2007; Donnelly et al. 2009; Seyyedi & Nadol 2014; Banakis Hartl et al. 2016). It has also been suggested that packing the cochleostomy with fascia may alter the sound transduction properties necessary for eliciting ACS VEMP, which could result in false positive indications of vestibular loss (Melvin et al. 2009). In CI patients with residual hearing, post-operative air-bone gaps of up 40 dB have been demonstrated (Chole et al. 2014; Raveh et al. 2015; Mattingly et al. 2016), with the largest air-bone gaps generally impacting the low frequencies (e.g., Chole et al. 2014; Mattingly et al. 2016). However, determining the presence of and quantifying post-cochlear implantation air-bone gaps is challenging due to the limitations of behavioral audiometric testing in this population.

226 Hz Tympanometry

To first determine whether standard clinical measures of acoustic immittance provide any indications of peripheral mechanical changes in the CI group as a whole and/or when separated by VEMP subgroups, group comparisons of tympanometric admittance, tympanometric peak pressure, and ear-canal volume were completed. No statistically significant group differences occurred for any of these comparisons, demonstrating that if mechanical changes do exist between these groups, 226 Hz tympanometry is not sensitive to these changes.

Wideband Acoustic Immittance: Absorbance

Measures of WAI were then investigated to determine whether any indications of peripheral mechanical changes were present in the ears with CI as compared to the age-matched ears with NH. Mean absorbance from the ears with NH is qualitatively consistent with previously published reports of WAI in both school-aged children and adults (e.g., Beers et al. 2010; Rosowski et al. 2012). However, our NH population includes both children and adults in a single group (mean age = 17.80, range 7–31), whereas much of the published normative data has not grouped these age ranges together. This is likely because age has been shown to impact WAI (e.g., Keefe et al. 1993; Beers et al. 2010); but, the focus of work on developmental and/or maturational changes on WAI responses has largely focused on neonates and young infants, or on comparisons of younger adults to elderly adults. The largest impact of age on WAI has been shown in the first 6 months of life, where maturational changes in the outer- and middle-ear can significantly impact acoustic responses. These responses become increasingly more adult-like between 6 and 24 months (Keefe et al. 1993; Sanford and Feeney 2008; Merchant et al. 2010). Much of this data suggests that low-frequency absorbance decreases as age increases (i.e., adult low-frequency absorbance is lower than that of infants and children), consistent with increases in stiffness due to maturation (Keefe et al. 1993; Sanford and Feeney 2008; Merchant et al. 2010). However, published comparisons between school-aged children and adults have shown a small but significant opposite effect, with adults demonstrating increased low-frequency absorbance as compared to school-aged children populations (Beers et al. 2010). The effect of age was not investigated here statistically due to limited power given our sample size. The mean age of our ears with normal hearing was 17.8 years (SD 7.84 years) and of our ears with CI was 14.74 years (SD 5.62 years). When subgrouped, mean ages per group ranged from 10.75 to 19.67 years. Details can be found in Table 7. Despite the possibility for small differences in WAI between school-aged children and adults in either direction, by age matching our ears with NH and CI, any age effects that may exist should be insignificant when it comes to the group comparisons made here. In addition, given the mean and ranges of age within groups (Table 7), it is unlikely that there are significant age effects driving the differences in WAI observed here.

Table 7.

Mean Age of NH and CI Groups and Subgroups

	All Participants			cVEMP Subgroup				oVEMP Subgroup
	NH	CI		CI PP	CI AP	CI AA		CI PP	CI AP	CI AA
Mean	17.80	14.74		15.50	18.75	10.75		19.67	14.91	13.67
SD	7.84	5.62		4.81	5.85	2.71		6.43	5.92	5.10

Results in ears with CI demonstrate decreased low-frequency absorbance as compared to age-matched ears with NH. This change in absorbance is consistent with a mechanical stiffening in the peripheral auditory system. To date, there is only one previously published report of WAI in CI recipients, the results of which are consistent with our findings (Scheperle & Hajicek 2019; a few examples of WAI in CI recipients are also shown in Wolfe et al. 2018, though they are not the focus of that work). Stiffening effects in peripheral mechanics can lead to conductive hearing loss (such as in disorders including otosclerosis and tympanosclerosis), particularly in the low frequencies, and it is possible that at least some of the proposed mechanisms described above (e.g., increased cochlear pressure due to the presence of the electrode, damage from the array to middle-ear or intracochlear structures, inflammation and fibrosis surrounding the electrode array) could result in stiffening of the system. In addition, several other reports have suggested that stiffening effects can occur post-implantation (Hernandez et al. 2018; Scheperle & Hajicek 2019). Thus, the observation of a stiffening effect on WAI is consistent with these published findings relating to a post-operative conductive loss.

Despite these overall trends, we find that there is wide variability in the post-operative absorbance response in the individual data. While the overall trend is a decrease in low-frequency absorbance as compared to normal controls, some ears with CI demonstrate normal-like absorbance responses and some even demonstrate an increase in absorbance, more consistent with an increase in admittance versus a stiffening effect. Scheperle and Hajicek (2019) also found a wide range of variability in their WAI responses in patient’s post-implantation. In addition, Donnelly and colleagues (2009) found that stapes displacement could increase, decrease, or stay the same after cochlear implantation when measured via laser Doppler vibrometry (Donnelly et al. 2009).

Overall, our data suggest that the mechanical impact of cochlear implantation on the peripheral auditory system varies from patient to patient and from ear to ear. This may be due to the fact that there are several possible causes of post-operative conductive changes, and the extent to which the system changes could vary on a case-by-case basis. Similar to our WAI findings, the presence (or absence) of an ACS VEMP response is also variable after cochlear implantation providing yet another indication that mechanical changes could be present in some cases, but not others. For this reason, we grouped our ears with CI based on their responses to ACS and BCV VEMPs. Our hypothesis was that peripheral mechanical changes that would lead to an air-bone gap would most likely be present in cases where the ACS VEMP was absent but the BCV VEMP was present (the CI AP group). The CI PP group likely has minimal to no peripheral mechanical change (at least not enough of an impact to abolish their ACS VEMP) and no true vestibular loss, and the CI AA group may or may not have alterations to their peripheral mechanics, but also has a true vestibular loss, and therefore the VEMP patterns can’t be used to determine whether a conductive component may be present.

When looking at the resulting absorbance data based on these subgroups (Figures 2 and 3), the CI AP group, regardless of whether they were grouped by cVEMP or oVEMP response patterns, was the only group with significant differences in absorbance from the NH control ears, consistent with our hypothesis. Looking at this more closely, when grouped based on cVEMP response patterns, this effect is clear, with the CI AP group showing a distinct decrease in low-frequency absorbance compared to NH controls while both the CI PP and CI AA groups demonstrate absorbance that is similar to normal (though a small but non-significant trend for a decrease in low-frequency absorbance is still observed). In contrast, when grouped based on the oVEMP response patterns, while the CI PP group continues to appear similar to NH control ears, the CI AA group appears more similar to the CI AP group, with a noticeable decrease in low-frequency absorbance. Interestingly, there were no significant differences between the CI AA oVEMP subgroup and NH controls, while there were significant differences between the CI AP oVEMP subgroup and NH controls. However, the CI AA versus NH comparison did approach significance, and there is potentially more variability in the CI AA group than the CI AP group that lead to the lack of significance in the effect. This CI AA group finding for oVEMP subgroups was not surprising for two reasons: 1) As described above, we did not expect the cVEMP and oVEMP subgroups to be identical due to differences in c- and oVEMP thresholds and 2) We did not hypothesize that the CI AA group would be similar to the NH controls (or any of the other groups for that matter), given it is possible that ears in that group could have both a conductive change as well as a true vestibular loss, but also could have no conductive change in the presence of a vestibular loss. Given the absent BCV VEMP in the CI AA group, VEMP responses cannot be used to indicate the presence of an air-bone gap in this group. Thus, the most interesting comparisons are between the NH control ears, and ears in the CI PP and CI AP groups. Unfortunately, the comparisons of these three groups are limited with the oVEMP subgroups given the small sample size of the oVEMP CI PP group.

While participants were excluded for a history of middle-ear surgery or multiple instances of tympanostomy tube placement, participants with a single instance of tympanostomy tube placement were included in this study. To rule out whether the stiffening effects observed in some ears with CI could be attributed to post-operative scarring after tube placement, we considered how many ears with tube placement fell into each group and subgroup. As shown in Table 8, the incidence of tube placement was spread out between the CI groups and is unlikely to account for CI group differences found here.

Table 8.

Number of Ears Per Group with Tympanostomy Tube Placements

All Participants			cVEMP Subgroup				oVEMP Subgroup
NH	CI		CI PP	CI AP	CI AA		CI PP	CI AP	CI AA
0	9		4	1	4		0	4	4

Limitations

First, there was not enough variability in participants or a large enough sample size to adequately compare associations between etiology, surgical technique, insertion depth, and implant type. Most of our participants had hearing loss of unknown etiology. CI patients with etiologies where established evidence suggests auditory mechanics could be impacted, including meningitis, were excluded. However, we cannot rule out that participants with other etiologies including Mondini malformation (N=2), Pendred syndrome (N=2), or other unknown etiologies could also impact the peripheral mechanical system. While there is not sufficient evidence to suggest these specific etiologies could confound our results, such that it justified participants with these etiologies be excluded, we acknowledge that the lack of adequate sample size to explore the impact of etiology is a limitation. Despite the small sample size, we looked at responses of the 4 ears with Mondini malformation and Pendred syndrome qualitatively as compared to the entire group. Figure 4 highlights the WAI responses of these 4 ears and demonstrates that none of the 4 ears had WAI responses that were outliers driving the stiffening effects observed here. Future work could also investigate pre- versus post-operative responses, where etiology effects would be present in both cases, but CI effects would only be observed post-operatively.

Figure 4.

NH mean +/− 1 standard deviation (SD, thick black line, gray shading) with individual absorbance tracings for each of the 27 ears with CI (thin black), with the two individual tracings for ears with Pendred syndrome highlighted in red and the two individual tracings for ears with Monidini malformation highlighted in light blue.

With regards to surgical technique, previous research has documented less traumatic effects with round window insertion as compared to cochleostomy (Todt et al. 2008; Richard et al. 2012). In contrast, round window insertions generally result in packing of the round window with tissue, which could alter the impedance at the round window and in turn alter the peripheral mechanics. The current study sample included ears that underwent primarily cochleostomies, therefore, the effect of surgical technique on VEMP and absorbance responses could not be adequately analyzed. Similarly, in relation to insertion depth, previous research has documented increased intracochlear trauma with deep insertions (Adunka & Kiefer 2006). The current study sample was primarily comprised of ears where a full insertion depth was achieved during surgery, as reported in the medical record. However, even implants where a full insertion depth is reported can vary in terms of true insertion depth and placement. For these reasons, the effect of insertion depth on VEMP and absorbance results could not be investigated here. Relating to implant type, there were 8 different implants amongst the 27 ears with CI, and the AB HR90K/HiFocus1J implant accounted for 14 of the ears. Although previous research has demonstrated no substantial differences in cochlear damage based on implant type (Eshraghi et al. 2009; Greene et al. 2015), a larger and more variable sample size is needed to obtain the statistical power to account for implant type in this study. Therefore, the effect of implant type on VEMP and WAI results could not be investigated statistically.

Second, to adequately account for other possible origins of changes in WAI, a detailed case history was used to exclude participants who had previous middle-ear surgery or had more than one tympanostomy tube placement. The impact of confounding pathologies and how those may also alter WAI could impact the clinical application of these findings. Future research should assess if these exclusion criteria are necessary and would, in fact, impact the utility of WAI to detect a change on the peripheral mechanics in this population.

Finally, pre-implantation VEMP responses and WAI measurements could not be completed because all participants in this study were recruited post-implantation. The average duration following CI placement in our cohort was 11.7 years (range: 1 – 23 years, SD: 4.76). Both pre-implantation and immediate post-implantation data would be of value, as within-subjects comparisons could provide information on the changes in these responses over time. This could help discern whether the vestibular dysfunction present in the CI AA group was pre-existing and related to etiology (or some other cause), or a direct result of implantation. It would also help define the actual change in WAI post-implantation. Future work should investigate differences in both VEMP responses and WAI measurements pre- and post-implantation, including at multiple time points post-implantation, given the multiple possible mechanisms for conductive changes related to post-operative healing.

CONCLUSION

BCV VEMP response rates were increased compared to ACS VEMP response rates in ears with CI. Ears with CI also demonstrate changes in low-frequency absorbance consistent with a stiffer system. This effect was largest for ears that had absent ACS but present BCV VEMPs. These findings suggest that cochlear implantation can lead to peripheral mechanical changes, at least in some ears, that could cause an air-bone gap.

While this post-operative conductive hearing loss may be of minimal significance to patients in most circumstances (given they are now receiving input electrically), and often cannot be measured audiometrically, there are several circumstances where understanding whether this is present may be clinically significant. ACS VEMP responses have been shown to be sensitive to the presence of air-bone gaps (Halymagyi et al. 1994; Bath et al. 1999). An absent ACS VEMP would suggest vestibular dysfunction; however, in a CI patient, this may lead to a misdiagnosis if the absent ACS VEMP is actually due to a conductive component and not true vestibular dysfunction. In addition to the misdiagnosis, this could also lead to inaccuracies in patient counseling if a patient is considering a second implant on the contralateral side. Patients and providers may be more conservative in their consideration of a second side, and even counsel against it, for fear of possible bilateral vestibular loss. Awareness of post-operative mechanical changes and/or a conductive component to the hearing loss may also be important for understanding and utilizing post-operative electrically evoked stapedial reflex thresholds for device programming (Scheperle & Hajicek, 2019), considerations relating to the utilization of hybrid (electro-acoustic) CIs and programming based on residual acoustic hearing, and monitoring of transient or chronic middle-ear dysfunction in CI recipients.

As a result of these findings, absent ACS VEMP responses in ears with CI should be interpreted with caution. Clinical considerations include using BCV VEMPs and WAI for pre-and post-operative testing in patients undergoing cochlear implantation.