Auditory biophysics of endolymphatic hydrops

1 Introduction

Menière’s disease (MD) is a still imperfectly understood cochleo-vestibular condition, characterized by the occurrence of a triad of symptoms (sensorineural hearing loss, vertigo, tinnitus plus a feeling of aural fullness) during repeated attacks that last a few hours, and for which there is no validated treatment. One issue for understanding what causes an MD attack, sometimes every few days, is that, although hydrops is considered a hallmark of MD, its presence both postmortem and in contralateral asymptomatic ears suggests that alone it is not sufficient to elicit the symptoms [12, 14]. Intuition dictates that inflation of scala media must have been associated with pressure increase in the endolymph at some early stage, but the high mechanical compliance of the boundary between endolymph and perilymph, in particular Reissner’s membrane, means that once inflated, scala media may remain so even when what provoked the inflation has disappeared. If true, the physical trigger of an attack remains to be ascertained.

To discuss the biophysical background of MD and the possible parts played by fluid pressure or compartment inflation, objective functional tools for exploring patients are needed. In a specifically human condition, for which no animal model satisfactorily reproduces its attacks even if murine models are currently being developed [11], these attacks can be studied only in patients, which requires noninvasive explorations. New MRI techniques have revealed endolymphatic hydrops in patients [14], but as they cannot be repeated at will, they are not suitable for monitoring the size of hydrops in relation to fluctuating symptoms. Conversely, cochlear electrophysiology allows closely repeated measures near an attack. Auditory sensory cells combine several advantages over their vestibular counterparts, their subnanometric sensitivity to stereocilia displacements and their ability to be probed by routine audiological tools. Accordingly, this short article will focus on the biophysical significance of audiological findings in MD.

2 Methods of interest

A diversity of responses from the cochlea are readily available. Outer hair cells, the amplifying effectors of the organ of Corti, generate otoacoustic emissions (OAEs) and cochlear microphonic (CM). The inner hair cells (IHCs), actual transducers that synapse with auditory neurons, are normally at the origin of the summating potential (SP), a result of their membrane depolarization. Electrocochleography also detects the compound action potential (AP), which stems from synchronous action potentials in cochlear neurons at stimulus onset. In addition, immitance measures using unconventional probe frequencies are sensitive to the impedance of structures beyond the tympanic membrane including the membranous labyrinth, which may be mechanically affected by MD. All these tests have shown some ability to reveal MD related changes.

3 Diagnostic value

A first approach is phenomenological, record the signal and compare its features when the patient is near an attack, to either the same ear in the absence of clinical sign, or a healthy subject. A sensitive test should be abnormal when the patient experiences (or has just experienced) an attack, and a specific test should not raise false alarms, neither in the asymptomatic ear nor in the MD ear tested when its hearing and balance functions are back to normal. Since the 70 s, the finding on an electrocochleographic record of a SP/AP ratio >0.40 is thought abnormal [8] (Fig. 1A). The sensitivity of the SP/AP test is rather limited, of the order of 60%, while its specificity is nearer 90%, e.g., in Avan et al. [2]. To improve the sensitivity, tone burst stimuli have been attempted. Another paradigm rests on the detection of OAEs produced by OHCs, and on the analysis of their amplitude and phase at low frequencies, characteristically more affected during MD attacks. For this purpose, distortion-product OAEs (DPOAEs) make it easy to focus on low frequencies even though click-evoked OAEs are also suitable once low-pass filtered [13]. The protocol is unconventional in that, building on the finding that intracranial pressure (ICP) modulates OAEs in a distinctive manner, by shifting their phases at low frequencies, a small disturbance in ICP is applied to the subject by gentle body tilt. The working hypothesis is that if the balance of hydrostatic pressures in a system with impaired homeostasis is moderately perturbed, its responses are exaggerated. In MD patients near an attack, body-tilt induced OAE phase shift exceeded, sometimes hugely, its normal ceiling of 40 degrees (Fig. 1B). The effect has a sensitivity of 75% and a specificity of 91% [7]. Interestingly the same ears show no abnormal effect when they are asymptomatic. Interestingly too, DPOAEs can be detected using primary tones around 1 kHz and primary intensities around 70 dB SPL, despite the presence of low frequency hearing loss that could reach 60 dB at least in MD ears. Such an ear can still recover near-normal auditory thresholds after an attack. It is thus noteworthy that OAEs in the context of MD lose their traditional screening or audiometric ability to spot frequency intervals with significant hearing loss.

OPEN IN VIEWER

Fig. 1

(A) repeated ABRs in response to 90-dB clicks, with an SP/AP ratio close to 1, largely exceeding the normal upper limit. (B) monitored DPOAE phase (solid line) when an MD subject near an attack is tilted from upright to supine (vertical arrows): the normal resulting shift should not exceed 40°. Dashed line, DPOAE amplitude, which remains stable. (C) modelled change in absorbance above 1.2 kHz against ear canal pressure (simulated by an increase in eardrum stiffness), at three different cochlear masses. The distance between absorbance peaks increases with increasing mass. (D) block diagram of the lumped-element model of the ear used for building Fig. 1C.

Last, immittance measurements using probe tones chosen in a broad frequency range, up to 2 kHz, usefully explore MD, with the advantage compared to the other two techniques that they do not need a functional cochlea. Two characteristics of immitance are abnormal in MD ears, the resonance frequency and the shape of conductance ‘tympanograms’ at 2 kHz [5, 16]. The sensitivity was 56% in MD ears, while the test was negative in 95% controls, yet only in 45% of asymptomatic ears contralateral to an MD ear. Interestingly, body tilt, which increases intracranial pressure, produced similar effects in controls [4], which pinpoints pressure as a parameter of interest.

None of the aforementioned abnormalities per se is pathognomonic of MD: an excessively large SP is present when the cochlea suffers from acute ischemia, or even in the context of synaptopathy [10]. Their interest in relation to the clinical picture of MD rests on our understanding of how SP, OAEs and immittance may be affected, either by hydrops or increased pressure or any geometrical, physical or ionic factor that might trigger an attack of MD or be concomitant. In this respect, recent reports outline a consistent view.

The SP is defined by the existence of an asymmetry in the compound sound-induced cochlear responses, so that the microphonic oscillations have a shifted baseline, usually (when collected by a standard electrocochleographic montage) in the same direction as the later AP. It is widely held that normally, the microphonic potential comes from OHCs that generate symmetric membrane-potential oscillations, whereas the SP is mainly IHC-generated. In both types of hair cells, the membrane-potential vs. stereocilia deflection plot is sigmoidal, but while the operating point OP of OHCs is set around the point with maximum slope, which is the center of symmetry, IHCs operate near the left asymptote of their plot, with a large asymmetry accounting for the SP. The situation changes if OHC homeostasis is impaired so that their OP shifts. This shift has two consequences, a contribution to the SP and a decrease in cochlear amplification (as for a given sound level, the membrane potential changes less, which affects electromotility). The OP shift and its consequences are not only a theoretical concept, it is proven in a mouse model with genetically disrupted OHC-to-tectorial membrane relationship [9].

This view of a modified OHC physiology in relation to OP shift pertains to DPOAE properties in the presence of a biased organ of Corti. The characteristics of DPOAEs mathematically relate to the shape of the sigmoidal transducer function that represents membrane potential of an OHC against stereocilia displacement near its OP. The even-order DPOAEs (e.g., f2-f1) and the odd-order DPOAEs (e.g., 2f1-f2) are not sensitive to the same features [6]. The amplitude of the f2-f1 product is very small when OP is near the center of symmetry and grows sharply when OP shifts. Conversely, the amplitude of the 2f1-f2 term is little affected in the same circumstances. How the cochlear transducer function can be derived from the behavior of DPOAEs when a low-frequency modulation is used for biasing the basilar membrane has been shown by Bian et al. [3]. Elaborating on this concept, Sirjani et al. [15] have shown that volume disturbances of scala media due to injections of artificial endolymph caused changes in the OP that resulted in predictable changes in DPOAEs. Gerenton et al. [7] went one step further by showing, in patients with definite MD near an attack, how body tilt modulates differentially the 2f1-f2 and f2-f1 components. As seen earlier, the shift in phase of the 2f1-f2 DPOAE with body tilt and concomitant elevation of intracranial pressure was excessive in MD ears only near an attack. In these ears, body tilt provoked oscillations in the amplitude of the f2-f1 component, when present, not in the amplitude at 2f1-f2. Different behaviors of f2-f1 and 2f1-f2 DPOAE amplitudes also occur during an acute endolymphatic hydrops in guinea pig [17]. In MD ears near an attack, results are consonant with the presence of an already shifted OP in the upright position, moving to either a more displaced or a more normal position after body tilt, thus vindicating the idea of an abnormal operation of OHCs and of an excessive modulation by small ICP manipulations. The associated presence in most of these ears of an excessive SP/AP also supports the concept of SP as a marker of an asymmetric OHC transducer function due to a displaced OP. These changes in the function of OHCs cannot be explained by the mere presence of an increased volume of scala media, with a distended Reissner’s membrane yet, because of its high compliance, hardly any pressure difference between endolymph and perilymph. A displaced OP is more easily understood with a bent basilar membrane requiring a significant endolymphatic pressure, and an attendant change in geometry of the organ of Corti.

Whereas the aforementioned effects happen around 1 kHz, principally near attacks according to several reports, and with features that implicate stiffness and pressure, the immittance changes described in MD ears occur at 2 kHz, above the resonance frequency of the ear. This suggests an inertial rather than stiffness-related effect, which might pinpoint increased mass of scala media due to hydrops. Preliminary modeling efforts are being made (Avan, unpublished, Fig. 1C) using a lumped-element model of the ear (Fig. 1D, [1]) to predict tympanometric shapes against probe frequency. An increase in mass in the cochlear branch of the model is needed to increase the separation between peaks in the reflectance of the ear as a function of ear-canal pressure, a characteristic feature of MD [5].

5 Conclusion

Until recently, the notion of hydrops was merely a post-mortem observation, but the improvements in MRI techniques now reveal hydrops [14]. Discrepancies between symptoms and imaging data call for an explanation. Should one pay attention to hydrops if it is asymptomatic, and what does it mean, in particular when the contralateral ear suffers from definite MD? This short review aimed to discuss whether and how biophysics can make sense of functional and imaging data, and of the patterns they form when combined. It supports the idea that instead of expecting that all tests of MD corroborate one another, one should comprehensively hoard their outcomes in the hope that their agreements and contradictions will soon find a logical mechanistic explanation.