Characterizing a Joint Reflection-Distortion OAE Profile in Humans with Endolymphatic Hydrops

Abstract

Objectives:

Endolymphatic hydrops (EH), a hallmark of Meniere’s disease, is an inner-ear disorder where the membranes bounding the scala media are distended outward due to an abnormally increased volume of endolymph. In this study we characterize the joint-otoacoustic emission (OAE) profile, a results profile including both distortion- and reflection-class emissions from the same ear, in individuals with EH and speculate on its potential utility in clinical assessment and monitoring.

Design:

Subjects were 16 adults with diagnosed EH and 18 adults with normal hearing (N) matched for age. Both the cubic distortion product (DP) OAE, a distortion-type emission, and the stimulus-frequency (SF) OAE, a reflection-type emission, were measured and analyzed as a joint-OAE profile. OAE level, level growth (input/output functions), and phase-gradient delays were measured at frequencies corresponding to the apical half of the human cochlea and compared between groups.

Results:

Normal hearers and individuals with EH shared some common OAE patterns, such as the reflection emissions being generally higher in level than distortion emissions and showing more linear growth than the more strongly compressed distortion emissions. However, significant differences were noted between the EH and N groups as well. OAE source strength (a metric based on OAE amplitude re: stimulus level) was significantly reduced, as was OAE level, at low frequencies in the EH group. These reductions were more marked for distortion than reflection emissions. Furthermore, two significant changes in the configuration of OAE input/output functions were observed in ears with EH: a steepened growth slope for reflection emissions and an elevated compression knee for distortion emissions. SFOAE phase-gradient delays at 40 dB FPL were slightly shorter in the group with EH compared to the normal group.

Conclusions:

The underlying pathology associated with EH impacts the generation of both emission types, reflection and distortion, as shown by significant group differences in OAE level, growth, and delay. However, hydrops impacts reflection and distortion emissions differently. Most notably, DPOAEs were more reduced by EH than were SFOAEs, suggesting that pathologies associated with the hydropic state do not act identically on the generation of nonlinear distortion at the hair bundle and intracochlear reflection emissions near the peak of the traveling wave. This differential effect underscores the value of applying a joint OAE approach to access both intracochlear generation processes concurrently.

INTRODUCTION

Two Classes of OAEs

Though the human cochlea is not accessible for direct measurement, the assessment of cochlear mechanics and outer hair cell function can be done noninvasively by presenting acoustic signals to the ear and recording the evoked acoustic responses, or otoacoustic emissions (OAEs). OAEs are a by-product of normal transduction. They are associated with the amplification process that allows for the acute sensitivity, frequency selectivity, and wide dynamic range that is characteristic of mammalian hearing. As such, OAEs are an invaluable tool for noninvasively probing and understanding the human cochlea.

OAEs are classified by the intracochlear processes that generate them. There are two broad classes of emissions, which arise via distinct cochlear mechanisms and may be sensitive to different cochlear properties (Shera & Guinan, 1999). Distortion emissions arise as a consequence of nonlinear distortion produced by the saturation of outer hair cell transduction currents (e.g., Hudspeth & Corey, 1977); they are thought to probe the strength and form of the cochlear nonlinearities responsible for their generation. Reflection emissions are backscattered wavelets, reflected off micromechanical irregularities along the cochlear partition (Shera & Guinan, 1999); they come from the region near the peak of the traveling wave (Lichtenhan, 2012; Berezina-Greene & Guinan, 2015; Goodman et al., 2020) and are thought to probe features characteristic of this region, such as near-threshold tuning and cochlear amplifier gain.

When measured together and analyzed relationally as a joint OAE-profile, these two classes of OAEs provide comprehensive information about hearing by accessing both cochlear generation processes (Abdala & Kalluri, 2017; Abdala et al, 2022). The dual approach to assessment may capture differences between hearing losses of distinct etiology. Unique or “signature” joint-OAE profiles could offer objective tools for diagnosis and monitoring. In this report, we characterize the joint-OAE profile in individuals with endolymphatic hydrops^*.

Endolymphatic Hydrops

Normal cochlear function requires a delicate and precise homeostasis within the membranous labyrinth of the scala media. Endolymphatic hydrops (EH), a hallmark of Meniere’s disease, is an inner-ear disorder with a prevalence of 20–40 for every 100,000 adults (Gates, 2006). In ears with EH, membranes bounding the scala media are distended outward due to an abnormally increased volume of endolymph. Hydropic ears are commonly diagnosed via symptom presentation, which is often episodic in nature. Diagnosis by an otologist considers low-frequency sensorineural hearing loss, vertigo, tinnitus, and aural fullness. Unfortunately, there are no relatively simple tests that definitively identify EH or aid in differential diagnosis. Those that exist, such as vestibular evoked myogenic potentials or magnetic resonance imaging (MRI) (Gluth, 2020), can be expensive, variable in sensitivity, and time-consuming. Additionally, successful application of the MRI for diagnosis of hydrops requires special protocols involving contrast-enhancement and sophisticated scoring methods (e.g., Han et al., 2022; Mavrommatis et al., 2022; Xiao et al., 2022)

One of the most reliable diagnostic markers of EH is fluctuating low-frequency hearing thresholds. This hearing loss can be borderline normal or mild in degree in the early stages but over time, as the hydropic state becomes chronic, more severe permanent hearing loss occurs and the higher frequencies become involved. In a hydropic ear, the fluid increase stiffens the cochlear partition, primarily in the apex, which impacts traveling-wave motion (Gluth, 2020); abnormalities in basilar membrane motion can be consequential as normal vibration is critical to transduction. Most existing studies of OAEs in human ears with Meniere’s disease have used only one class of OAE (either distortion product OAEs or transient evoked OAEs) but not both. These studies have generally reported inconsistent findings but have observed an overall reduction of OAE level and irregular phase in ears with EH (e.g., Fetterman, 2001; Magliulo et al., 2001; de Kleine et al., 2002; Magliulo et al., 2004; Rotter et al., 2008; Gerenton et al., 2015; Drexl et al., 2018).

The joint-OAE profile may be helpful in the diagnosis and monitoring of EH. While measuring reflection (stimulus-frequency, SF) and distortion (distortion-product, DP) OAEs in an initial pilot subject with diagnosed EH, we observed an interesting pattern: DPOAEs were extremely low level or non-measurable at low-frequencies, while SFOAEs were near-normal at these same frequencies (Abdala et al; 2020; Abdala, 2022). Previous work in guinea pigs supports this result. Lee et al. (2020) performed endolymphatic sac ablation surgery to induce EH in guinea pigs and recorded DPOAEs and SFOAEs at various post-surgical times. They found mid-frequency DPOAE levels were reduced by approximately 10–20 dB at all test times (day 1 through day 30). In contrast, SFOAE levels were more mildly reduced, by roughly 5–7 dB on average for the 1-day survival group only.

The mechanism by which EH impacts otoacoustic emissions is unclear. One possible mechanism is a shift in the operating point of the outer hair cell transduction function. It has been shown that abnormal endolymph volume can mechanically disturb outer hair cell transduction by altering the operating point of the transducer (Sirjani et al., 2004; Salt et al., 2009). The relationship between the hair cell bundle displacement and transduction current can be described by a response curve; and the operating point is the place on this curve where the hair cell “operates” in the absence of stimulation. Shifts of this operating point have been shown to influence the generation of even-order distortion emissions (e.g., f₂-f₁) and to a lesser degree, odd-order distortion (e.g., 2f₁-f₂) (Sirjani et al., 2004).

In this study we sought to characterize the joint-OAE profile in human ears with EH and speculate on its potential utility in clinical assessment and monitoring. We hypothesize that the two classes of OAEs, reflection and distortion, will be differentially sensitive to EH.

MATERIALS AND METHODS

Subjects

Subjects were 16 adults with diagnosed EH (9 female, 7 male; mean age 47 years) and 18 normal-hearing healthy adults (11 female, 7 male; mean age 44 years) matched within 3 years of average age. Normal hearing was defined by a modified audiometric Hughson-Westlake procedure as ≤ 15 dB HL at octave and inter-octave frequencies from 0.5–8 kHz.

Figure 1 displays the audiometric thresholds of each ear with EH as thin colored lines. The group mean thresholds for EH are shown in thick gray lines and for the Normal (N) group in thick black lines (no individual thresholds are provided for the N group). The mean differences in audiometric thresholds between the N and EH groups for target frequencies in the apical half of the cochlea (≤ 1.5 kHz) ranged from 15–22 dB. The mean audiogram from this EH group shows only slight hearing loss, but notably, there is nearly a 10 dB difference between the lowest-frequency audiometric threshold at 250 kHz and thresholds at 2–3 kHz, producing the classic rising configuration for the ears with EH. In contrast, the mean audiometric thresholds for the N group are flat across frequency.

Figure 1.

Mean audiometric hearing thresholds as a function of frequency for ears with normal-hearing (thick black line; filled circles) and endolymphatic hydrops (thick gray line; open circles). Individual thresholds are shown for the EH group only, as thin colored lines.

Audiometric thresholds were not controlled for the EH group; the only inclusion criterion was a medical diagnosis of either Meniere’s disease or EH made by an otologist. These diagnoses were made mostly from symptomology (e.g., episodic vertigo, aural fullness, rushing/roaring tinnitus, and fluctuating low-frequency hearing loss), as is most often the case in the assessment of this disease. By self-report, none of the individuals with EH were experiencing an active episode of disease during the data-collection period. Of the 16 individuals in the experimental group, 64% had been given a diagnosis of Meniere’s disease and 36% had a diagnosis of EH. Fourteen of the 16 EH subjects had been diagnosed within 1.5 years of their participation in this study; the average gap between diagnosis and data collection was 9.5 months. Hence, the majority of experimental subjects could be considered as having early-stage EH, commensurate with their slight-to mild hearing loss, mostly restricted to the low-frequencies.

All participants had normal outer- and middle-ear function as determined by otoscopy, tympanometry (GSI Tympstar Middle Ear Analyzer, Grason-Stadler Inc., Eden Prairie, MN), and air-bone gaps on the audiogram of ≤ 10 dB. Participants provided informed consent prior to data collection and were monetarily reimbursed for their participation. All procedures were conducted in accordance with guidelines of the Institutional Review Board at the University of Southern California.

Instrumentation and Calibration

A Babyface Pro USB High Speed Audio Interface (RME Audio, Germany) and ER-10X probe system (Etymōtic Research, Elk Grove Village, IL) controlled by custom software in MATLAB (Mathworks, Natick, MA) were used to generate stimulus waveforms and record the ear-canal pressures. Microphone voltages were amplified (+20 dB) and high-pass-filtered (300-Hz cutoff frequency) before A/D conversion. Testing was conducted with the subject reclined in an ergonomic chair within a sound-isolated IAC audiometric booth that met ambient noise standards (ANSI S3.1–1999). The probe cable was suspended from the ceiling and the probe tip was carefully positioned into the ear canal to achieve a relatively deep and stable fit, whereupon the cable was secured using a nylon headband. Subjects rested quietly or watched a subtitled video during testing.

Forward-pressure-level (FPL) stimulus calibration was applied to correct for the effects of ear-canal standing waves by controlling only the amplitude of the forward-traveling stimulus wave, separating it from any energy reflected off the eardrum (Scheperle et al., 2008). Prior to each participant’s arrival, Thévenin-equivalent probe parameters were obtained in the ER-10X calibrator (brass tube, inner diameter 7.9 mm) at room temperature using five settings of calibrator length (78.4, 64.8, 35.8, 29.7, and 24.6 mm) with the goal of achieving total “source-calibration errors” < 1 (Scheperle et al., 2011). The source-calibration error is a dimensionless index calculated as the sum of squared differences between the measured and predicted cavity pressures divided by the sum of the measured cavity pressures. The typical source-calibration error in our laboratory is 0.03.

With these known parameters characterizing the probe, the ear-canal acoustic impedance and corresponding characteristic impedance were derived from a wideband chirp stimulus. All stimulus levels in this experiment were measured in dB FPL. OAEs emerging from the cochlea are likewise impacted by standing wave effects in the closed ear canal and were corrected to emitted pressure level (EPL) (Charaziak & Shera, 2017). EPL-corrected emissions are equivalent to the those measured in an anechoic ear canal, not contaminated by standing wave effects in the ear canal.

OAE Stimulus Parameters and Data Collection

During testing, the probe fit was monitored by noting deviations in stimulus level re: target levels, and by documenting changes in the frequency of the half-wave resonance in the ear canal. These shifts prompted a re-fitting of the probe or a slight nudge back into the ear canal by the tester, followed by recalibration. Testers were asked to manually initiate a recalibration for observed changes of > 3 dB in the stimulus level or shifts in the half-wave resonance frequency of > 400 Hz. Re-calibration also occurred automatically between stimulus-level conditions (or, at minimum, every 6 minutes). During any fit or re-fit, an alert was provided to the tester if the presence of a probe-fit leak was suspected. This occurred when absorbance at low frequencies (A_low) was 0.2 (–7 dB) or higher (Groon et al., 2015). A_low was calculated by averaging over the frequency range spanning 100–200 Hz.

Stimulus generation and OAE acquisition were controlled using an algorithm that swept a tone upward (for DPOAEs) or downward (for SFOAEs) across a 5-octave range (0.5–16 kHz) at a rate of one octave/second. Stimulus tones that are swept in frequency allow for rapid and efficient OAE measurements with high frequency resolution (e.g., Long et al., 2008; Kalluri & Shera; 2013). Sweep rate and direction were established from past work in our labs (Kalluri & Shera, 2013; Abdala et al., 2015; Shera & Abdala, 2016). For normal hearers, the SFOAE probe was presented between 20 and 65 dB FPL and the DPOAE L₂ was presented between 25 and 75 dB FPL, both with 5 dB resolution. For subjects with impaired hearing, stimulus levels were adjusted slightly based on audiometric thresholds; that is, the lowest-level OAE stimulus presented was based on the best of the elevated audiometric thresholds (converted to dB SPL and rounded up to the nearest 5 dB). This avoided wasting time by presenting stimuli at levels too low to elicit measurable OAEs. OAE-type and stimulus-level conditions were presented in a computer-generated random order, with the exception of the first two conditions, which were always 65 dB FPL for DPOAEs and 40 dB FPL for SFOAEs.

To further expedite data collection, rather than present a single continuous sweep across the frequency range, segments of the frequency span were stacked (with 0.1 octave overlap) and presented concurrently (Abdala et al., 2018a; 2022). Three concurrent segments were presented to evoke DPOAEs for stimulus levels from 20–60 dB FPL but only two stacks could be presented for stimuli > 60 dB FPL, where presumed interaction between the segments was observed (i.e., where pilot testing determined that a single sweep level/phase deviated from the stacked sweep data by more than typical test-retest variability). The durations of each DPOAE sweep were either 1.7 seconds or 2.5 seconds (depending on the number of stacked frequency segments). Because three sweeps were required to derive a mean DPOAE measure due to phase-rotation averaging (explained in the following paragraph), the duration of one DPOAE “trial” was roughly 6–8 seconds including inter-sweep intervals. For SFOAEs, at the highest stimulus levels (60- and 65-dB FPL), two segments were presented concurrently while at moderate (50- and 55-dB FPL) and low (< 50 dB FPL) levels, a three- and five-stack approach was applied without any notable deviation in level or phase re: the single-sweep condition. For SFOAEs, the duration of each sweep varied from 2.3 to 5.3 seconds (depending on the number of stacked segments); four sweeps were required to derive two SFOAE measurements; hence, the duration of one SFOAE “trial” was roughly 5–11 seconds including inter-sweep intervals.

DPOAEs were evoked with an optimized stimulus-frequency ratio as described in Stiepan et al. (2022). This optimized strategy applies the f₂/f₁ ratio shown to produce the highest DPOAE levels on average for a given frequency and stimulus-level combination as per group data analyzed from 30 normal-hearing individuals. In general, the higher the frequency and the lower the stimulus level, the narrower the f₂/f₁. Because much of our previous work with DPOAEs has been conducted using the so-called “scissors” method, we applied this strategy here (Kummer et al., 1998; 2000) to set the primary-tone level separation. No “optimal” or standardized level-separation strategy exists at present, but it is important to note that the choice of primary-tone levels (i.e., scissors vs constant level separation across stimulus level) will impact the overall shape of the DPOAE input/output function (Zelle et al., 2015). We also applied phase-rotation averaging to cancel the primary tones before analysis (Whitehead et al., 1996). To do so, three stimulus segments with different primary-tone starting phases (ϕ) were interleaved such that the primary tones, f₁ and f_2, cancel when the responses are averaged.

SFOAEs were measured using a modified, interleaved suppression paradigm (Shera & Guinan, 1999; Abdala et al., 2018a,b). Responses to four stimulus combinations were measured: p₁ = probe tone alone, p₂ = probe and suppressor tone (+polarity), p₃ = probe tone alone, and p₄ = probe and suppressor tone (–polarity). The SFOAE time waveform was extracted from the four response waveforms using the formula: p_SFOAE = (p₁ + p₃ − p₂ − p₄)/2. The suppressor tone (frequency f_s) was presented at 50 dB FPL for probe tones between 20–35 dB FPL and at +15 dB (re: probe level) for L_p > 40–65 dB FPL. The suppressor frequency was chosen so that f_s/f_p = 0.95 (Kalluri & Shera, 2007; 2013).

SNR-guided Data Collection

A custom-designed data-collection program based on the signal-to-noise ratio (SNR) of the OAE estimated in real-time was utilized. To apply this method, the five-octave frequency range was divided into more tractable half-octave frequency bands denoted by the following center frequencies: 0.75, 1, 1.5, 2, 3, 4, 6, 8, and 12 kHz. Prior to data collection, level-dependent estimates of the number of sweeps required to achieve a target 6 dB SNR were derived from previous data collected in our laboratory, combined with published literature (Abdala et al., 2018a; Gorga et al., 1997; Ellison & Keefe, 2005). The minimum number of sweeps was always 24 and the maximum number was set to three times our estimate of the number of sweeps required to achieve criterion SNR. As an example, in a normal-hearing ear, the number of sweeps presented to measure DPOAEs could vary between 24 and 153 sweeps, and for SFOAEs, between 24 and 282 sweeps (with more sweeps presented at lower stimulus levels). The number of presented sweeps was greater for hearing-impaired ears to achieve the target SNR; the number of DPOAE sweeps presented ranged from 24 to 300 in impaired ears and the number of SFOAE sweeps ranged from 24 to 500 sweeps.

OAE data collection ceased when: (1) all center frequencies achieved at least 6 dB SNR, (2) the maximum number of sweeps was reached without achieving criterion SNR, or (3) the tester terminated averaging because it was unlikely to result in a 6 dB SNR. The decision to terminate averaging was guided by the projected increase in SNR based on the square root of the remaining number of sweeps available. However, most often, testers did not abort averaging when the choice was presented; testers were instructed to get as close to the maximum number of sweeps required for each condition, while also considering session duration. The mean duration of OAE data collection with this protocol is 43 minutes for normal-hearers (range = 26–68 minutes) and 62 minutes for EH subjects (range = 23–136 minutes).

Artifact Rejection

Following each individual sweep presentation, a least-squares fitting (LSF) estimate of the OAE was derived and the mean magnitude of the OAE was updated and displayed. Any single data point (of ~500 data points sampled across five-octaves) differing from the real-time mean OAE level by more than 4 standard deviations was labeled an artifact and automatically triggered an additional replacement sweep. During post-test analysis, the artifact-detection process was repeated, and individual data points identified as artifacts in the frequency domain were linked to the corresponding point on the time waveform. A waveform segment centered around this time point and equal in duration to 20% of the analysis-window, was eliminated. All artifactual segments were removed before the final grand mean OAE level was calculated for any given condition. Overall, (across frequency, stimulus level, and OAE type), 17% of total sweeps presented to the group with endolymphatic hydrops were “replacement sweeps”, triggered by an artifact rejection. In normal hearers, the percentage of replacement sweeps was roughly 12% of the total.

Estimating OAEs

Least Square Fit and Signal Processing

OAE level and phase were derived from the recorded ear canal signal using an LSF procedure (Long et al., 2008; Kalluri & Shera, 2013; Abdala et al., 2015). Briefly, this method estimates OAE phase/level by segmenting the OAE time waveform into moving analysis windows that shift in 0.01 octave steps. Models for the stimuli, suppressors, and OAEs are created within these windows. The amplitude and phase within each window are estimated by minimizing the sum of the squared residuals between the model and the data to achieve the best fit. LSF analysis-window durations varied continuously as a function of frequency to keep constant the number of spectral fine structure periods (for DPOAEs) or cycles of phase rotation (for SFOAEs).

For DPOAEs, the LSF was also used to separate the nonlinear distortion component of the DPOAE from the reflection component. Because the reflection component occurs at longer latencies, it can be removed by using a wide LSF analysis window (~1.75 fine-structure periods). The SFOAE was processed with a continuous wavelet transform (CWT) to eliminate short-latency energy associated with stimulus artifact and long-latency energy associated with multiple internal reflections (Moleti et al., 2012). The CWT is a time–frequency analysis tool that decomposes signals using wavelets that are frequency-scaled and time-shifted. Compared to other time–frequency analyses (e.g., the short-time Fourier transform) the CWT provides improved time resolution at high frequencies and improved frequency resolution at low frequencies. The corresponding noise floors were calculated at four points around the OAE frequency, either on the low-frequency side (0.90, 0.88, 0.86, 0.84 times the DPOAE frequency) or the high-frequency side (1.10, 1.12, 1.14, 1.16 times the SFOAE frequency) and then passed through the same signal processing as was the OAE.

Data Analysis

Band-Averaged Spectra

OAE level and corresponding noise floors were binned into half-octave-wide frequency bands denoted by the center frequency of the band: 0.75, 1, 1.5, 2, 3, 4, 6, 8, and 12 kHz. Each frequency band was comprised of a maximum of 50 data points across frequency. The first step in creating a band-averaged OAE spectrum was the elimination of OAE data embedded in excessive noise. Many of these points were caught by the artifact-rejection scheme but for more chronic noise (vs artifactual “spikes”), the following technique was used: To examine the data for excessive noise, the frequency range was broken up into 1/10^th octave sub-bands and the noise floor was calculated within each sub-band. If any of these noise-floor bands exceeded the upper boundary of a 95% confidence interval formulated from a group of 117 normal-hearing adults, the OAE measured in that frequency-band was eliminated. This elimination occurred regardless of the OAE’s SNR in this frequency band. That is, the data-cleaning was driven only by the noise to avoid poor-quality data points from contributing to the averaged spectra.

Once OAE data points contaminated by excessive noise were cleaned (as described above), mean OAE levels were then calculated in half-octave frequency bands. Our reasoning for using noise to eliminate poor-quality data instead of using SNR is as follows: SNR can be poor for two reasons, either noise is too high (which often has little to do with the generation of OAEs) or the signal is too low. We were only interested in measuring the latter because low-level exemplars are informative and eliminating them results in an over-estimation of OAE level. Although we did not eliminate any data based on SNR, we considered this factor by weighting the contribution of each single data point to the half-octave mean by its SNR. The weighting function was the square of the SNR on a linear scale (i.e., 10^SNR/10 where SNR is in dB). The SNR calculation for weighting used the median of the seven data points nearest the target frequency as the noise referent. With this analysis strategy, every subject and test condition was populated with data and, most importantly, even conditions with weak OAEs were represented in the mean.

Although OAE level is initially shown at all center frequencies for full transparency (Fig. 2), because the apical half of the cochlea is most impacted by EH, we analyzed data for only the three lowest center-frequencies in this study: 0.75, 1, and 1.5 kHz. To compare the band-averaged OAE level data from healthy normal-hearers to those with hydrops, three-way Group (EH, N) x Frequency (0.75, 1, 1.5 kHz) x OAE (SFOAE, DPOAE) ANOVAs were conducted at 40 and 65 dB FPL separately. Alpha levels were set at 0.05.

Figure 2.

Mean DPOAE (top row) and SFOAE (bottom row) band-averaged spectra for the normative group or N (blue and red) and for the group with endolymphatic hydrops or EH (cyan and gray) at two stimulus levels, 40 and 65 dB FPL. For target frequencies (≤ 1.5 kHz), DPOAEs from EH ears were more reduced than were SFOAEs. Individual subject data are shown as lighter lines and group mean noise floors are dotted lines (when noise was < −30 dB, no noise floor is visible).

Difference Values

An additional analysis sought to assess the impact of EH on the relationship between reflection and distortion emission level by subtracting SFOAE from DPOAE levels to generate an OAE-level difference value. This difference value was analyzed with a two-way ANOVA of Group (EH, N) x Frequency (0.75, 1, 1.5 kHz) for 40 and 65 dB FPL stimulus levels separately. A second difference value subtracted OAE levels between the two groups, N-EH, for these same two stimulus levels.

Input/Output Functions

DPOAE and SFOAE I/O functions were generated and evaluated at low frequencies corresponding to fine-structure peaks/plateaus in the SFOAE spectra. This method of choosing target frequencies at peak data only has been implemented by our lab for over a decade, in multiple published reports. Only peak data are considered because SFOAE spectral minima are the result of destructive phase interactions among reflected wavelets. Erratic shifts in amplitude occur at steep minima frequencies, and the test-retest reliability of the data are degraded (Abdala et al., 2018). Data at minima do not necessarily reflect cochlear health (Shera & Bergevin, 2012; Kalluri & Shera, 2013; Abdala & Kalluri, 2017). DPOAE I/O functions were generated at corresponding frequencies.

OAE I/O functions were generated with data centered around the target frequency (1/8^th octave on either side). Because OAE spectral peaks can shift slightly in frequency as stimulus level decreases, only the top 30% of data points within this band (n = 8) were used in creating the I/O function, thus allowing us to more effectively track the peak as stimulus levels decrease and avoid the inclusion of data on a flank or minimum. The noise floor was calculated from these same eight points. Because SFOAE spectral peaks occurred at idiosyncratic frequencies across subjects, the I/O function data were slotted into corresponding half-octave frequency bands (as described for band-averaged spectra). Therefore, at a center frequency of 1 kHz, for example, OAE data could come from I/O functions centered at frequencies between 0.91 to 1.23 kHz and so on.

To characterize and quantify OAE growth and compression features (which included eight data points per stimulus level), I/O functions were fit with the following function converted to dB:

$$Y=A_{\text{n}}[1+{\left(X/A_0\right)}^a/{\left(1+X/A_1\right)}^{\left(a-b\right)}],$$

where $X$ is stimulus amplitude in $\mu \mathrm{P}\mathrm{a};Y$ is OAE amplitude in $\mu \mathrm{P}\mathrm{a};A_0,A_1,A_{\mathrm{n}},a,b$ are fitting parameters; $A_0,A_1,A_{\mathrm{n}}$ are sound-pressure amplitudes in $\mu \mathrm{P}\mathrm{a}$; and $a,b$ are power-law exponents. Conceptually, $A_{\mathrm{n}}$ is the noise amplitude, $A_0$ is the stimulus amplitude where the I/O function emerges from the noise, and $A_1$ characterizes the stimulus amplitude at the onset of the high-level behavior. The following parameters were estimated from each I/O function: (1) Maximum Slope (dB/dB) is the slope of the function at its steepest point; (2) Source Strength (dB) is the OAE level re: stimulus level at maximum slope and is conceptualized as the strength of the underlying emission generation mechanism; (3) Compression Knee (dB FPL) is the stimulus level at which the slope of the function has decreased to 50% of the maximum slope; and (4) Compressive Slope (dB/dB) is the slope of the function between the compression knee and the data point at the highest stimulus level presented.

To compare the four parameters extracted from the I/O functions between groups, three-way ANOVAs for Group (EH, N) x OAE (SFOAE, DPOAE) x Frequency (0.75, 1, 1.5 kHz) were conducted separately for each parameter.

OAE Phase-Gradient Delay

OAE phase was converted to a group (or phase-gradient) delay by calculating the negative of the phase slope in milliseconds as $\tau (f)=-\text{d}{\varphi }_{\text{SFOAE}}/\text{d}f$, where f is in kHz, and ${\varphi }_{\text{SFOAE}}$ is the OAE phase in cycles. The delay data for each group was then fit with a loess trend line to elucidate the pattern or trajectory of delay across frequency. Delays were examined at two fixed stimulus levels, 40 and 65 dB FPL, for both OAE-types and subject groups, across a nearly five-octave range. In addition, the level-dependence of reflection emission phase-gradient delay was probed by comparing SFOAE delay trend lines within groups for three stimulus level conditions: 40, 50, and 60 dB FPL.

RESULTS

Band-Averaged Spectra

Figure 2 shows OAE level as a function of frequency at two stimulus levels. The dark lines with symbols depict mean OAE levels for nine half-octave frequency bands denoted by their center frequencies (0.75 to 12 kHz) while individual band-averaged spectra are shown as thinner, lighter lines. Although all frequencies are shown in this initial plot, subsequent figures and analyses focus on target frequencies from the apical half of the human cochlea: 0.75, 1, and 1.5 kHz. Data from the N group are shown in blue filled circles (DPOAE) and red x’s (SFOAE) while corresponding data from the EH group are shown in cyan filled circles (DPOAE) and gray x’s (SFOAE). Overall, one can see that OAE levels are reduced in the group with EH at both stimulus levels and for both OAEs. However, the mean difference between the N and EH groups for the distortion emission is greater than that for the reflection emission, which was more likely to overlap with the normal spectra.

Three-way ANOVAs conducted on OAE level for Group x OAE x Frequency at 40 and 65 dB FPL separately, showed a main effect of Group (F = 74.12; p < 0.0001) and OAE (F = 12.67; p < 0.0005) at 40 dB FPL. There was no main effect of Frequency (F = 2.92; p = 0.56) and no three-way interaction (F = 0.73, p = 0.31). The EH group showed significantly reduced OAEs relative to the N group and the SFOAEs were higher in level than DPOAEs always, which has been reported in previous work (Abdala et al., 2022). At 65 dB FPL, there was also a main effect of Group (F = 50.51; p < 0.0001) and OAE (F= 189.26; p < 0.0001), with trends in the same direction as the lower stimulus level. There was a main effect of Frequency at 65 dB FPL (F = 3.87; p = 0.022) but no three-way interaction (F = 0.02; p = 0.98).

These analyses confirmed that EH produces reductions in OAE level for both emission types. However, this analysis does not easily illustrate how the relationship between reflection and distortion emissions changes when hydrops is present. OAE level difference-values can illustrate changes in the relationship between SFOAEs and DPOAEs. In the EH group, DPOAE–SFOAE level difference values were calculated at 40 and 65 dB FPL and are depicted in the top plots of Figure 3. Values below zero indicate the SFOAE as the higher-level of the two emission levels. As SFOAEs are commonly higher in level than DPOAEs (Abdala et al., 2022), data below zero for both groups is not surprising as shown in Figure 3. While at 40 dB FPL, the relationship between DPOAEs and SFOAEs does not appear to change due to EH, at 65 dB FPL, the hydropic group shows DPOAE–SFOAE difference values that are farther from zero than those of the normal group, by roughly 5 dB across frequency. This suggests that the relationship between SFOAEs and DPOAEs changes in the EH group. Indeed, a two-way ANOVA (Group x Frequency) on the difference value showed a significant effect of Group at 65 dB FPL (F = 6.67; p = 0.011), which can be accounted for by a greater reduction in the DPOAE (vs SFOAE) in pathological ears.

Figure 3.

Mean DPOAE–SFOAE level differences for endolymphatic hydrops (EH) and normal (N) ears are shown in the top row for target low frequencies at two stimulus levels. Mean N–EH OAE level differences for each emission are shown in the bottom row. Error bars are +/− 1 SD for the top panels and the square root of the sum of variances for N-EH data in the bottom panels.

A second difference value was calculated by subtracting OAE level in the EH group from that in the N group (see the bottom panels of Figure 3). In this calculation, the horizontal line at zero indicates no-difference between OAE levels from the normal healthy group and the group with EH. At 65 dB FPL (but not 40 dB FPL, where emissions are equally affected), the SFOAE is closest to zero (by ~ 5 dB), indicating a trend for more “normal-like” levels (i.e., the SFOAE has retained more of its amplitude) than DPOAEs, which have been more markedly reduced by EH. This is consistent with the observation that the pathology associated with EH is interfering with the generation and/or propagation of nonlinear distortion emissions more so than that of cochlear reflection emissions.

OAE Input/Output Functions

OAE levels measured at low-mid frequencies (0.75 – 1.5 kHz) were plotted as a function of stimulus level. By quantifying how OAE amplitude changes with increasing stimulus levels, one can gain insight into the growth and compressive nature of emissions and their underlying generation mechanisms. Figure 4 shows group mean I/O functions for the three target frequencies. The means are depicted as dark lines while individual I/O functions are shown as thinner, lighter lines. The blue and cyan functions on the top row are DPOAEs from N and EH groups respectively; the red and gray functions on the bottom row are SFOAEs from N and EH groups respectively.

Figure 4.

Mean DPOAE (top row) and SFOAE (bottom row) input/output functions from each group at target low frequencies. Individual input/output are also shown as the lighter lines. Dotted lines depict mean noise.

The I/O functions from the EH group were more heterogeneous overall. This is consistent with a disease process that can be episodic and likely presents at different stages across subjects. Also evident in the I/O function data is that the EH-related reductions in OAE levels are apparent across stimulus level. Again, these reductions appear to be greater for DPOAEs than SFOAEs, which often skirt the range of low-normal. Interestingly, the shape of I/O functions from ears with EH mimic the growth pattern seen in healthy ears in many ways; That is, the DPOAE is strongly compressed with at lower stimulus levels than the SFOAE, which grows more linearly overall. Additionally, the SFOAE is generally the higher-level emission, particularly at high stimulus levels, for both normal hearers and the EH group.

Four parameters were extracted from the I/O functions to quantify the growth and compression features of the emissions (as detailed in the Methods section). Three of these parameters are plotted in Figure 5 for each group: maximum slope, source strength, and compression knee. Two-way Group x OAE ANOVAs were conducted on each parameter (summed over the three frequencies because an initial omnibus three-way analysis including Frequency as a factor failed to show any significant effect of frequency). For maximum slope, there was a main effect of Group (F = 17.2; p < 0.0001) and OAE (F = 13.6; p < 0.0003) but also, a Group x OAE interaction (F = 6.9; p < 0.01). This interaction is consistent with a steepening of the I/O function in EH ears, for the SFOAE only. This can be seen in the top right panel of Figure 5 (for SFOAE data); the EH group (gray) shows steeper slopes than does the N group (depicted in red).

Figure 5.

Mean parameters derived from DPOAE (left column) and SFOAE (right column) input/output functions: Maximum slope, Source Strength, and Compression Knee shown at target low frequencies. The red and blue lines represent normal SFOAE and DPOAE data respectively while the cyan and gray data are from the EH group. Error bars are +/− 1 SD.

Similar to OAE level, source strength (middle panels of Fig. 5) was significantly reduced in the EH group (F = 120; p < 0.0001). This is expected since source strength most often follows the trends observed for OAE level (as it is a measure of OAE magnitude re: stimulus level). A borderline significant effect of OAE was also noted (F = 3.34; p < 0.07) with a trend for the SFOAE to show slightly stronger emissions than the DPOAE for both normal and hydropic ears.

The compression knee showed main effects of both Group (F = 25.7; p < 0.0001) and OAE (F = 41.78; p < 0.0001) and a Group x OAE interaction (F = 5.6; p < 0.02). This interaction indicates that the Group effect on compression was only present for the DPOAE not the SFOAE; that is, for the DPOAE, the compression knee was elevated. This can be seen in the bottom left panel of Figure 5 as the DPOAE data from EH ears (depicted in cyan) show higher compression knee values than the N data (depicted in blue). The compressive slope is not shown because it showed only a main effect of OAE (f = 114.65; p < 0.0001), confirming that the SFOAE was always steeper in the compressive portion of the function (vs DPOAE) but no group N vs EH distinction was present (F = 2.64; p = 0.106).

In Figure 6, we generated a low-frequency joint-OAE profile by plotting all DPOAE values from EH ears against their corresponding and paired SFOAE values (purple circles) for two main I/O function parameters. Most individuals have contributed more than one data point since three frequency conditions (0.75, 1, and 1.5 kHz) have been collapsed into one low-frequency band and OAE frequencies are linked to individual SFOAE spectral peaks. The shaded ellipse encompasses ~95% of the data from our normal hearers (assuming a 2D Gaussian distribution) thus providing a normative template against which to compare ears with hearing loss or disease. The benefit of using this profile is that it illustrates the effects of EH on both reflection and distortion emissions simultaneously, elucidating relational patterns that are otherwise difficult to see.

Figure 6.

A joint-OAE profile as an x-y graph, showing DPOAE values plotted as a function of SFOAE values for source strength (top) and compression knee (bottom) in a low-frequency band (0.75 – 1.5 kHz). The shaded gray ellipses are the normative 95% confidence regions while the purple symbols are paired DPOAE-SFOAE data points from individual EH subjects. Some subjects contributed multiple data points because OAE frequency is linked to SFOAE spectral peaks, which vary among individuals.

As is evident from the top panel of Figure 6, source strength data mostly fall outside of the normal ellipse for EH ears: 42 of 62 data points are abnormally reduced as shown by data falling below the ellipse. This vertical pattern is consistent with greater spread on the y-axis, which is the DPOAE dimension. All 42 abnormal DPOAE values represent low DPOAE strength. In contrast, fewer than half of the SFOAE values are abnormal (i.e., fall outside of the ellipse) and source-strength is never below roughly –70 dB (with the exception of one outlier). This elliptical analysis allows one to visualize the more impacted DPOAE (vs SFOAE) in the EH group.

The compression knee ellipse, likewise, not only tells us which points are abnormal (a much smaller percentage than abnormalities observed in source strength) but it provides a characteristic pattern or directionality to further describe the abnormality. When compression is abnormal in ears with EH (10 data points are outside of the ellipse or borderline), it is always in the direction of an elevated compression knee and only for DPOAEs. All SFOAE compression data points remain well within the corresponding normative space. These results suggest that the presence of EH extends the linear range of the DPOAE growth function but does not impact compression features of the SFOAE significantly.

SFOAE Phase/Delay

OAE phase-vs-frequency functions were converted into group delays (see Methods) or what we refer to here as “phase-gradient delays” (Shera et al., 2000) and compared between groups. Left and middle panels of Figure 7 show the SFOAE phase-gradient delay trends at three stimulus levels for both groups. Reflection-emission delays, which have been linked to cochlear tuning (Shera et al., 2010), are known to have a strong and consistent level-dependence; that is, SFOAE delays are longer and phase gradients steeper at low stimulus levels, with delays shortening as stimulus level is increased (Schairer et al. 2006; Bergevin et al., 2012; Abdala et al., 2018a). As is evident from Figure 7, SFOAE phase-gradient delays in both groups manifest this characteristic level dependence (longer delays at lower stimulus levels), across nearly five octaves.

Figure 7.

SFOAE phase-gradient delays were fit with a loess trend line and plotted as a function of frequency for the N and EH groups. The left and middle panels illustrate that SFOAE phase-gradient delays from both groups show a characteristic level-dependence: The lower the stimulus level, the longer the SFOAE delays. The dashed lines define the confidence intervals for the loess fits. The right panel shows a group comparison for SFOAE delay trend lines at 40 dB FPL. In this comparison, EH ears had slightly shorter SFOAE delays than normal ears for frequencies below 2 kHz.

The right panel of Figure 7 shows that at the lowest stimulus level (40 dB FPL), below approximately 2 kHz, the SFOAE delay trend in the EH group is slightly (but significantly) shorter than that of the N group. This group effect for SFOAE delay was only noted at 40 dB FPL and not at the higher levels. The comparison of DPOAE phase-gradient delays were inconclusive but showed mostly overlap between the two groups. This is an area that warrants further investigation.

Results Summary

Our results show that normal hearers and individuals with EH share some common OAE patterns, such as the SFOAE being generally higher in level than DPOAEs and showing more linear growth than the strongly compressed DPOAE. However, significant differences were also noted between groups, indicating that endolymphatic hydrops can produce OAE abnormalities and impact reflection and distortion emissions in distinct ways. OAE levels and source strength were consistently reduced at low frequencies in hydropic ears. For example, OAE source strength was reduced overall by an average of 14–15 dB in ears with EH while OAE level was reduced on average by 6–9 dB at target frequencies. These EH-related OAE reductions were more marked for distortion (vs reflection emissions, which appear to retain their amplitude to a greater extent); that is, reflection emissions are relatively preserved in ears with EH when compared to the impact the disease has on distortion emissions. The presence of EH produced two significant changes in the configuration of OAE I/O functions: (1) a steepened growth slope for SFOAEs only and (2) an elevated compression knee for DPOAEs, which created an expanded region of linear growth.

DISCUSSION

It is noteworthy that the effects of endolymphatic hydrops on OAEs were sometimes restricted to either reflection or distortion emissions, and if the effects manifested in both, they were sometimes more marked on one of the two OAEs. This suggests that the application of a joint-OAE approach (i.e., the inclusion of both classes of emissions in assessment and/or study of EH) can be of value to access different intracochlear generation processes concurrently. This contrasts with a more common strategy that arbitrarily chooses either DPOAEs or transient-evoked OAEs, which are a commonly utilized reflection-class emission in clinical settings. Our results indicate that EH does not act identically on the generation of nonlinear distortion emissions at the hair bundle and the generation of intracochlear reflection emissions near the peak of the traveling wave. Hence, reflection and distortion OAEs offer non-redundant information about EH.

Why are DPOAEs More Impacted by EH than SFOAEs?

Animal studies have shown that the initial and early pathophysiology of EH affects the hair-cell bundle, particularly the lateral links among the stereocilia (Rydmarker & Horner, 1990,1991). The volumetric increase of endolymph causes an interruption in this linkage, producing a disarray of the hair bundle and likely altering its mechanical function to reduce or eliminate cochlear nonlinearities as gauged by DPOAEs. However, alterations in the hair bundle would also be expected to reduce cochlear gain, which impacts generation of both distortion and reflection emissions. This compromised gain may be responsible for the reduced SFOAE levels reported here, although these are normal or near-normal in many cases. Perhaps the impact of a hydropic cochlea (and all it entails) compromises the generation of cochlear nonlinearities more so than cochlear gain in the earlier stages of the disease process.

One route by which the loss of lateral links in the hair bundle could disrupt the generation of nonlinear distortion would be to shift the operating point of the outer hair cell transduction function, which is transducer current versus hair bundle displacement. Mechano-electrical transduction is thought to be optimized for the generation of the cubic distortion product when the outer hair cells operate symmetrically at a region of maximum slope on this transduction function. Therefore, a change of the operating point, and hence the symmetry of the cochlear amplifier, could be reflected in diminished 2f₁-f₂ DPOAE levels as observed here. Other investigators have probed this possibility in some depth (e.g., Sirjani et al., 2004; Salt et al.2009).

A third contributor to the differential effect of EH on reflection and distortion emissions may have less to do with DPOAE generation and more to do with the generation mechanisms of reflection emissions. As per models of SFOAE generation, micromechanical irregularity along the cochlear partition is a required element. It is possible that hair-bundle disarray, along with damage produced to the stria vascularis or membranes bounding the scala media (which can rupture during active phases, e.g., Salt & Plontke, 2010), will increase tissue irregularity and produce enhanced “roughness” in hydropic ears. This increase in biological roughness may result in higher cochlear reflectivity. That is, more tissue damage produced by EH will create more points of wavelet backscattering, which may increase SFOAE levels overall. This effect could compensate for the reductions in cochlear gain produced by hydrops and account for the relative preservation of SFOAE levels observed here. A comparable effect has been reported in aging ears (Abdala et al., 2018). Additionally, the Ceacam16 mutant mouse, which shows a deficit in tectorial membrane collagen resulting in poorly banded edges on this membrane, has also shown atypically large and abnormally numerous reflection emissions (Cheatham et al., 2014).

The Confound of Aging

In the present study we were careful to compare our EH subjects to age-matched controls (group mean age matched to within three years). Previous work in our laboratory (Abdala et al., 2018b) demonstrated that aging, after controlling for audiometric threshold, can produce qualitatively similar results to those reported here. Abdala et al. (2018b) showed that aging reduces both DPOAEs and SFOAEs, but the nonlinear distortion emission drops more rapidly with increasing age than does the reflection emission. In a phenomenological model of simulated SFOAEs, increasing the roughness parameter in the model with increasing age led to a qualitatively similar preservation of SFOAE level. In the present study, we may be seeing this effect in ears with EH. Eventually, if cochlear gain decreases beyond some threshold (perhaps as the disease—or aging—progresses), one expects that even increased roughness could not compensate sufficiently. Therefore, if age is not carefully matched between groups, the independent effect of aging can magnify the selective and more marked reduction in DPOAE levels attributed to EH. Indeed, our initial observation during pilot work (and the impetus for the current study) was a drastically reduced DPOAE in hydropic ears when compared to a normative group that was not matched in age. Once we completed our study and analyzed data with age-matched normal hearers, these DPOAE reductions were smaller, and the subsequent change in the DPOAE-SFOAE relationship, though it persisted, was less pronounced. Clearly, any joint-OAE analyses must take age into consideration in formulating normative datasets because reflection and distortion emissions are differently impacted by aging (Abdala et al., 2018b)

A Signature OAE Profile for Endolymphatic Hydrops?

Although we have described a joint-OAE profile for a group of individuals with documented EH, we have not addressed the most critical questions. For example: Does the EH profile described here provide a signature result that, when compared to age-matched normative groups, can be used to identify this specific pathology and differentiate it from others? In the current study, we cannot address this question because we did not include hearing losses of varied etiology. Here, our objective was to contrast ears with EH to those with fully normal peripheral function so as to determine the absence/presence and extent of hydrops-induced OAE abnormalities. However, our labs are currently working on an ongoing, large-scale project recording a joint-OAE profile from individuals with sensory hearing loss of varied etiologies, such as noise-exposure or ototoxicity, to address this intriguing question. Although we lack a definitive answer, Figure 8 can inform this question in a preliminary way.

Figure 8.

The 16 ears with EH were divided into two sub-groups: those with normal audiometric thresholds (≤ 20 dB HL) at target low frequencies (EH+NH) and those with elevated thresholds at these frequencies (EH+HL). Mean DPOAE level, source strength, and compression knee are shown for these two EH sub-groups and for the normal group (blue filled circles: N). Error bars are +/− 1 SD. In general, the degree of abnormality for each OAE metric appears linked to the degree of threshold elevation.

Figure 8 shows DPOAE levels measured at 65 dB FPL for the normative group and for two subsets of data derived from the total EH group: (1) a dataset for EH ears where audiometric thresholds were normal (though still an average of 5.5 dB worse across target frequencies than the normal-hearing control group) and (2) a dataset for EH ears where audiometric thresholds were elevated (> 20 dB HL). Data in Figure 8 show that DPOAE level reductions in the EH group are impacted by audiometric threshold, at least with the two gross threshold distinctions we have provided here. Interestingly, subjects with EH and normal low-frequency thresholds still have DPOAEs that are somewhat reduced relative to the normative group, suggesting that the disease itself produces abnormalities in the generation or detection of DPOAEs, even in the absence of threshold elevation. These apparently conflicting observations suggest complexity in how the combined effect of disease and elevated hearing thresholds impact the joint-OAE profile. The middle and bottom panels in Figure 8 also show an association between mild threshold elevation and DPOAE source strength and compression knee. The source strength metric is reduced consistent with the degree of threshold elevation and the DPOAE compression knee is elevated to a greater degree for those EH subjects with hearing loss. These same trends were observed for the SFOAE data (not shown).

From the association of elevated threshold on OAEs in these subjects with EH, one might question whether comparable elevated audiometric thresholds produced by any other type of hearing loss, noise-exposure or ototoxins for example, would show this same effect on OAEs. Do OAE levels simply follow audiometric thresholds regardless of the type of underlying damage that produced the hearing loss? This is a highly relevant question that we are exploring at present with other studies. In emerging work, however, we have observed distinct patterns in OAE level reduction from ears with noise-induced hearing loss vs presbycusis (Abdala et al., 2023). And here, we note that even in EH ears with no elevated thresholds, OAEs were abnormally reduced. However, our data show that the impact of EH on OAEs is linked to some extent on the degree of hearing loss. One could argue that individuals with more hearing loss (translating to more reduction in cochlear gain) have more advanced cases of EH. The more advanced the disease process, the more likely its impact on the generation of OAEs.

Finally, our observation of shortened SFOAE delays (Fig. 7; 40 dB FPL) may appear to conflict with the notion—recently supported by measurements of the cochlear frequency map in guinea pigs with induced hydrops (Guinan et al., 2021)—that a stiffened cochlear partition in hydropic ears will shift the peak of the traveling wave toward the apex. If SFOAE delays were determined primarily by the round-trip distance between the stapes and the peak of the traveling wave, then an apically shifted peak would produce longer delays. However, other factors can influence SFOAE delays. For example, shortened delays may reflect a basalward shift in the locus of SFOAE generation and/or generally broadened frequency tuning in the EH group (consistent with their elevated thresholds). Unfortunately, the data in this study cannot disentangle these interrelated effects. Nevertheless, the overall similarity of the SFOAE and DPOAE delays between the two groups suggests that EH did not greatly affect OAE phase features.

Further investigation is needed to describe the joint-OAE profile that characterizes various etiologies of hearing loss and to study how much overlap and/or independence exists among these patterns. It is not clear whether the joint-OAE profile we have defined here is specific for EH. Still, it will likely be useful in the clinical monitoring and tracking of this episodic disease as symptoms and hearing fluctuate during its course.