The middle ear muscle reflex: Current and future role in assessing noise-induced cochlear damage^a)

A. MEMR pathway

The neural pathway of the MEMR has been studied across multiple species (Borg, 1973; Lyon, 1978; Lyon and Malmgren, 1982). The validity and reproducibility of the MEMR as it relates to various lesions along the auditory pathway was initially assessed in rabbits (Borg, 1973, 1977). Subsequent studies found that the neuronal organization of the reflex pathway in rats (Rouiller et al., 1989) and cats (Lyon, 1978; Lyon and Malmgren, 1982) were comparable to the findings in rabbits. These studies, as well as others, led to the current understanding of the MEMR pathway. The pathway begins when sound energy progresses through the external ear canal and transfers to the middle ear via vibration of the tympanic membrane. The middle ear ossicles then transmit the vibrational energy to the fluid-filled cochlea, where sensory cells that communicate with the auditory nerve are stimulated to send signals to the central auditory nervous system. The MEMR pathway extends from the auditory periphery to the central auditory nervous system by ascending to the ventral cochlear nucleus and then progressing through two sets of neural fibers (Moller, 1974a; Mukerji et al., 2010). The first neural fiber track activates the ipsilateral facial nerve motor neuron. The second set of neural fibers travel from the ventral cochlear nucleus through the brainstem and then descend to the contralateral facial nerve motor neuron. The facial nerve neuron traverses back through the internal auditory meatus and out to the middle ear, where a branch of the facial nerve is sent to the stapedius muscle (Mukerji et al., 2010). This ipsilateral and contralateral pathway activation results in bilateral contraction of the stapedius muscle that “stiffens” the tympanic membrane and preferentially attenuates low frequency sounds (Borg and Zakrisson, 1974; Moller, 1974a; Mukerji et al., 2010). Any disruption along this pathway can eliminate or elevate the reflex response. The threshold and amplitude of the MEMR, as elicited by an external sound, is dependent on both the intensity and frequency content of the incoming acoustic signal as well as the acoustic parameters of the MEMR response probe (Olsen et al., 1975; Stach, 1987; Margolis, 1993).

B. Clinical application of MEMR

The MEMR response is used extensively in clinical audiology as an objective cross-check test to identify various pathologies along the auditory pathway. The MEMR response encompasses both a threshold and amplitude measure. The MEMR response can be effective in identifying the possible presence of middle ear, cochlear, or retrocochlear pathologies (Borg, 1977). Studies have shown that there is a reduced or absent MEMR response in individuals with conductive hearing loss from medical conditions such as otosclerosis or third window disorders (Keefe et al., 2017). The MEMR, in conjunction with tympanometry and behavioral audiometry, is often used to screen and identify conductive hearing loss (Hong et al., 2016; Keefe et al., 2017). The MEMR has also been used in the assessment of cochlear hearing loss. In individuals with mild sensorineural hearing loss (20–40 dB HL thresholds), the MEMR can be elicited by stimulus levels of ∼90 dB HL. However, as hearing thresholds increase beyond ∼40 dB HL, MEMR thresholds increase (Margolis, 1993). For example, in individuals with moderate hearing loss (50–60 dB HL thresholds), the MEMR can be elicited at 35–45 dB sensation level (SL), or at approximately 95 dB HL (Katz et al., 2002). With severe-profound sensorineural hearing loss (>80 dB HL thresholds), the MEMR response is either absent or present at the highest presentation levels (95–105 dB HL). Consequently, the presence or absence of the MEMR response largely depends on the degree and configuration of hearing loss.

In addition to its use in standard clinical hearing assessments, the MEMR response is also used as part of test batteries for the diagnosis of retrocochlear pathology. Studies have found that the MEMR amplitudes can be reduced or absent in cases of acoustic neuromas (>15 mm in size), although the sensitivity and specificity can be quite variable and is affected by the location and size of the tumor (Mangham and Miller, 1979; Mangham et al., 1980; Bergenius et al., 1983; Hunter et al., 1999). Abnormalities in the MEMR response are also found in patients with facial nerve palsy. Inflammation and swelling of the CNVII results in temporary MEMR amplitude weakness or paralysis (Kopala and Kukwa, 2016). The complex anatomy and physiology of the MEMR increases its susceptibility to multiple pathologies, including middle ear problems, cochlear damage, and neural pathway disruption. For these reasons, the MEMR response is often tested as part of wider battery of tests, and MEMR response interpretation can be challenging in the absence of established evidence-based protocols. For research purposes related to hearing loss, the MEMR response should be used in conjunction with other tests of cochlear or auditory function.

C. Noninvasive MEMR assessment parameters

The ipsilateral MEMR response is assessed using a multifunction probe that creates a hermetic seal in the ear canal, presents an acoustic elicitor stimulus, and records the resulting change in tympanic membrane compliance to a separate acoustic probe stimulus. Of these functions, both the elicitor stimuli and probe parameters have the potential to impact the results of the test. For example, elicitor stimulus type and frequency content can affect the MEMR response. Pure tone stimuli at 0.5, 1, and 2 kHz are commonly used in clinic and reliably elicit the MEMR response. In contrast, MEMR responses to 4 kHz pure tones can be elevated or absent in some individuals with normal hearing. It has been speculated that this elevation or absence of the MEMR response at 4 kHz may be due to more rapid onset latencies (Jerger et al., 1986) and subsequently more rapid adaptation rates observed in higher frequency stimuli (Wilson et al., 1978). In a study that compared multiple different elicitor stimuli on the MEMR response, the lowest proportion of MEMR responses was elicited using brief (100 ms) 4 kHz pure tones in a population of normal hearing, adult participants; conversely, brief 1 kHz pure tones elicited the highest proportion of MEMR responses (Deiters et al., 2019). In addition to pure tone stimuli, broadband noise (BBN) and filtered noise have also been used as elicitors. Noise elicitors have been shown to produce overall lower MEMR thresholds than pure tone stimuli in adults (Silman, 1979; Thompson et al., 1980; Margolis, 1993) as noise stimuli simultaneously activates a broader frequency region of the cochlea relative to single frequency elicitors (Peterson and Liden, 1972; Wilson and McBride, 1978). Taken together, the type and frequency spectra of MEMR elicitor stimuli have clear implications for the resulting response.

Studies have also shown that elicitor duration can have a significant effect on the MEMR response (Woodford et al., 1975; Barry and Resnick, 1976; Korabic and Cudahy, 1984; Cacace et al., 1991). These studies suggest that MEMR threshold and amplitude measures adhere to the temporal integration effects observed in psychoacoustic studies (Barry and Resnick, 1976). For example, a study that varied elicitor duration (10, 20, 50, 100, 200, 500 ms) using multiple pure tone stimuli (0.5, 1, 2, 3, 4 kHz) found that as elicitor duration decreased, MEMR threshold increased in normal hearing adults (Woodford et al., 1975). In a similar study, MEMR threshold increased as elicitor duration decreased, and short duration stimuli (20 ms) produced overall higher contralateral MEMR thresholds (Cacace et al., 1991). In adults with sensorineural hearing loss, increasing stimulus duration produced MEMR threshold temporal integration functions that were relatively steeper than normal hearing adult participants, suggesting that elicitor duration parameters may be useful for assessing differences among individuals (Korabic and Cudahy, 1984).

In Sec. II B, the effects of elicitor parameters on the MEMR response were discussed, including elicitor type, frequency content, and duration. However, the frequency of the probe tone can also influence the MEMR response. In adults, the MEMR is typically measured using a low frequency (e.g., 220 or 226 Hz) probe tone. A low frequency, low intensity probe tone was originally selected because it did not risk activating the reflex, while allowing MEMR response assessment by the probe microphone to elicitor stimuli (Terkildsen and Nielsen, 1960; Silman, 1984). In contrast, higher frequency probes are used in pediatric assessment. Higher frequency probe tones such as 660 or 1000 Hz have been shown to be more effective at measuring the MEMR response in this population (Sprague et al., 1985; Yang and Liu, 2020). This is in part due to smaller ear canal volumes and overall higher compliance in infant ears (Holte et al., 1991; Hunter and Margolis, 1992; Swanepoel et al., 2007). Although both low and high single-frequency probe tones are used in research and in the clinic, studies have shown that the probe frequency capable of measuring maximal changes in compliance can differ across individuals (Bharadwaj et al., 2019). For that reason, probe measurements that are capable of assessing changes in compliance across a range of frequencies, such as wideband probe stimuli, are emerging diagnostic and clinical research tools.

Wideband probe measures can be used in conjunction with different elicitor stimuli to measure changes in admittance across multiple frequencies simultaneously. These wideband probe measurements have been suggested to be more sensitive to compliance differences among individuals and may be more sensitive to changes in the MEMR response overall (Feeney et al., 2003). In humans, the use of wideband probes yields thresholds that can be 20–24 dB lower than those measured using single frequency probes (Feeney and Keefe, 2001; Schairer et al., 2007). Although single frequency probes produce higher thresholds, it is important to note that even with the standard 226 Hz probe, MEMR thresholds in humans are considered reliable and repeatable when using a minimum amplitude criteria of 0.02 ml or greater reduction in compliance and a 10 dB down, 2 dB up stair wise threshold procedure (Guest et al., 2019b). MEMR assessment with wideband click probes has also been shown to have high test-retest reliably when elicitor stimuli presentation levels are between 70 and 90 dB SPL. However, when the presentation level of the elicitor stimuli is reduced to 60–65 dB SPL, reliability becomes moderate and at 45–50 dB SPL, reliability becomes low (Kamerer et al., 2019). As such, MEMR elicitor presentation levels for both single frequency and wideband probes are an important consideration for both clinical and research purposes. In order to briefly summarize the MEMR parameters of the references cited in this paper, a table has been provided (Table I).

Given the ensuing impact on the MEMR response and the reliability of the MEMR, it is critical to consider both elicitor and measurement probe parameters. When these factors are considered, emerging research suggests that the clinical utility of the MEMR response could expand and may provide an objective and reliable proxy measure of cochlear health.

A. Effects of noise-induced hearing loss on the MEMR

Noise-induced hearing loss (NIHL) is a well-documented consequence that occurs as a result of exposure to hazardous noise levels. Numerous animal studies have shown that within the cochlea, high levels of noise produce metabolic and mechanical stress that can permanently damage the sensory cells of the inner ear (Kurabi et al., 2017; Ding et al., 2019; Frye et al., 2019). For example, noise-induced cochlear damage is correlated with permanent audiometric threshold shifts (PTS), tinnitus, and hyperacusis (Eggermont and Roberts, 2015; Turner and Larsen, 2016; Ding et al., 2019). In general, when hearing thresholds are elevated, whether due to aging, noise exposure, or other cochlear pathologies, MEMR thresholds are also elevated (Katz et al., 2015). In many cases, NIHL is differentiated from other causes of hearing loss by a “notch” in the audiogram, where hearing thresholds are elevated at the frequencies that appear to be most susceptible to noise exposure (3–6 kHz) (Coles et al., 2000; McBride and Williams, 2001; Kirchner et al., 2012). However, notched audiograms can occur in individuals without corresponding noise exposure history (Nondahl et al., 2009), and NIHL may be present in audiograms that are otherwise not distinguishable from traditional sensorineural hearing loss (Lie et al., 2017). Thus, it has been suggested that the MEMR response may have the potential to differentiate between sensorineural hearing loss and NIHL (Lindgren et al., 1983).

Early research that investigated the MEMR response in relation to NIHL suggested that the MEMR response could be used to differentiate NIHL from other sensorineural pathologies in instances where the audiometric “noise notch” is unclear. In two studies the MEMR response was elicited with 4000 Hz pure tone stimuli in individuals with 60 dB HL or higher pure tone thresholds at 4000 Hz. One of the studies measured MEMR thresholds in individuals with NIHL and found that the MEMR was reliably elicited in 90% of the participants despite high audiometric thresholds at that frequency (60–80 dB HL) (Lindgren et al., 1983). In contrast, the second study assessed the MEMR response in individuals with other cochlear pathologies and similar 60 dB HL pure tone thresholds at 4000 Hz, but found that the MEMR response could not be elicited with 4000 Hz pure tone stimuli (Jerger et al., 1972). The differences observed between studies suggest that pure-tone thresholds alone may fail to account for differences in the underlying cochlear pathology across patients. While studies that assess the MEMR response as a tool to differentially diagnose cochlear hearing loss are limited, some studies support the possibility of using the MEMR response in conjunction with other clinical measures to distinguish between NIHL and other causes of cochlear hearing loss.

B. Effects of noise induced IHC synaptic damage on the MEMR

The discovery that noise exposure can substantially damage cochlear IHC synapses before there is evidence of permanent threshold elevation has significantly altered how we think about NIHL, NIHL risk, and NIHL assessment (Kujawa and Liberman, 2009; Furman et al., 2013). Studies in multiple species, including mice, guinea pigs, chinchillas, and rhesus monkeys (Song et al., 2016; Hickox et al., 2017; Valero et al., 2017), have shown that noise-induced loss of IHC synaptic terminals is correlated with permanent reductions in auditory brainstem response (ABR) wave-1 amplitudes at suprathreshold levels (Kujawa and Liberman, 2009). The pattern of synaptic damage preferentially affects low spontaneous rate (SR) auditory nerve fibers that are activated at high intensity sound levels, but has little effect on high SR fibers activated at lower intensity sound levels (Furman et al., 2013; Marmel et al., 2015), though this effect may be species-dependent (Suthakar and Liberman, 2021). Consequently, impaired afferent communication between low SR fibers and IHC does not produce threshold elevation but is believed to compromise suprathreshold hearing (Chen et al., 2021b; Cildir et al., 2022; Shehabi et al., 2022).

Cochlear synaptopathy is believed to produce a number of suprathreshold deficits including poorer temporal resolution, poorer word recognition in competing background noise, tinnitus, and hyperacusis (Hickox and Liberman, 2014; Chen et al., 2021a). The proposed deficits associated with synaptopathy have been labeled as “hidden hearing loss” (Schaette and McAlpine, 2011; Bajin et al., 2022; Valderrama et al., 2022), or threshold-independent (Kamerer et al., 2019; Trevino and Lobarinas, 2022), due to the lack of changes in auditory thresholds. Thus far, the most reliable physiological correlate of synaptopathy in animals has been the reduction of ABR wave-1 amplitude (Kujawa and Liberman, 2006, 2009; Furman et al., 2013; Kujawa and Liberman, 2015). In humans, however, ABR wave-I amplitudes alone can have significant variability across individuals. For example, head size and distance to early ABR wave generators can affect wave-I amplitude (Bramhall et al., 2019; Bramhall, 2021). In addition, unlike animal studies, there is typically no baseline assessment prior to noise exposure, precluding measurement of potential noise-induced changes in MEMR threshold or amplitude. To mitigate this limitation, other electrophysiological measures have been suggested as objective markers of noise-induced synaptopathy. These include ABR wave I/V ratios (Grose et al., 2017), SP/AP ratios (Liberman et al., 2016), and envelope following response measures (Mepani et al., 2021). The reliability and specificity of these measures are currently under investigations across a number of laboratories with promising results in both human and animal studies (Bramhall et al., 2021; Bharadwaj et al., 2022). Regardless of which evoked potential metric is selected, there is no agreed upon definition for clinically significant change (if a baseline is available) or clinically significant deficit (if no baseline is available).

One of the key advantages of the MEMR response is that it is a rapid, objective assessment that is elicited at higher external sound levels. The higher-level activation may reflect the afferent cochlear activity that constitutes the initial portion of the reflex circuit that is postulated to be highly dependent on low SR fibers (Kobler et al., 1992; Valero et al., 2018). If the MEMR response is dependent on low SR fibers [as suggested by Kobler et al. (1992) and Valero et al. (2018)] and if synaptopathy degrades low SR fibers [either preferentially or in combination of degradation of medium or high SR fibers as shown in Furman et al. (2013), Marmel et al. (2015), and Suthakar and Liberman (2021)], then loss of low SR fibers could reduce the amplitude of the MEMR. This hypothesis may be supported by studies in mice with evidence of noise-induced synaptopathy, where it has been reported that synaptopathic noise exposures can significantly reduce the amplitude of the MEMR (Valero et al., 2016; Valero et al., 2018). Importantly, the MEMR amplitude was only reduced in frequency regions with moderate IHC synapse loss, but not in regions with minimal synaptic loss. Additionally, the reduced MEMR amplitudes were observed when using wideband probe measurements, but not with single frequency probes that are often used clinically. It has been suggested that individuals that have similar MEMR thresholds and amplitudes to single frequency probes (e.g., 226 Hz) could have significant differences at other frequencies (Feeney et al., 2017; Bharadwaj et al., 2019).

The lack of sensitivity of the 226 Hz probe to noise exposure history, noise-induced synaptopathy or IHC damage have been discussed across multiple studies. In humans, MEMR response assessment using the 226 Hz probe tone did not correlate with lifetime noise exposure (Causon et al., 2020) or tinnitus and speech-in-noise perception (Guest et al., 2019a). Interestingly, 226 Hz MEMR threshold and amplitude growth were related to middle ear compliance, where higher compliance yielded lower thresholds and steeper growth functions (Causon et al., 2020). Thus, middle ear compliance could play a greater role in MEMR reflexes and thresholds among individuals with normal or near normal thresholds regardless of their noise exposure history. An additional study in humans also failed to find a correlation between hearing-in-noise ability and the 226 Hz probe MEMR response, and instead suggested that poorer hearing-in-noise ability was likely the result of reduced top-down processes, rather than peripheral bottom up deficits (Saiz-Alia et al., 2019). More recently, an animal study destroyed 50%–80% of IHC pharmacologically, a level of deafferentation much greater than that produced by noise exposure, found no effect on the 226 Hz probe MEMR response (Trevino et al., 2022). Collectively, the human and animal studies to date suggest limitations of the commonly used 226 Hz probe in MEMR measurement as an assay of IHC synapse and cochlear damage.

In contrast to the aforementioned results from 226 Hz probe MEMR response studies, more recent studies using wideband MEMR probes suggest that these may be more sensitive to cochlear damage. For example, in one human study, reduced MEMR amplitudes elicited with wideband probes were shown to correlate with both increased estimates of lifetime noise-exposure and poorer word recognition in competing noise as a function of increasing presentation level (74–104 dBA) (Shehorn et al., 2020). In another study, elevated MEMR thresholds assessed with wideband probes were related to poorer word recognition in noise in some, but not all tasks, and were also correlated with poorer performance for time compressed speech (Mepani et al., 2020). Reduced MEMR amplitude strength using wideband probes has also been found in individuals with tinnitus and otherwise normal or near normal hearing. Evidence of weakened MEMR response, when assessed using click probe stimuli with a flat spectrum (0.25–12 kHz) and broadband noise elicitor stimuli, were found in individuals with tinnitus suspected of having synaptopathy (Wojtczak et al., 2017). Taken together, these studies suggest a potential research and clinical role of the wideband MEMR response with respect to cochlear synaptopathy and suprathreshold deficits; however, additional research is warranted.

C. Otoprotectants

NIHL is widespread across active military and veteran populations (Yankaskas, 2013; Theodoroff et al., 2015; Moore et al., 2019) as well as workers in industries with hazardous noise levels (Nelson et al., 2005; Engdahl and Tambs, 2010). In these environments, hearing protection devices may not be used due to safety concerns, but also due to lack of compliance to safety regulations (Casali et al., 2009; Green et al., 2021). For this reason, there has been substantial effort towards investigating potential pharmaceutical interventions that could minimize risk and improve compliance (Le Prell, 2022). An ideal otoprotective drug would minimize the damaging effects of noise and preserve hearing, either by prevention (prior to exposure) or by rescue (after exposure).

Currently, the efficacy of otoprotectants is often evaluated by typical markers of NIHL, such as changes or shifts in distortion product otoacoustic emissions (DPOAE), ABR thresholds, or behavioral audiometric thresholds. Given the growing body of data suggesting that noise exposure may also produce suprathreshold deficits (Hickox and Liberman, 2014; Lobarinas et al., 2017; Le Prell, 2022), otoprotective drug studies lack reliable and sensitive objective measures to detect these changes. MEMR responses have been posited as a potential measure of synaptopathic damage, NIHL, and the resulting suprathreshold symptoms that is both noninvasive and objective; however, additional validation and agreed on definitions for clinically significant change are needed if MEMR is to be used in clinical trials as a measure of otoprotectant efficacy. It is important to note that the MEMR responses can be temporarily reduced or eliminated by muscle relaxants, sedatives, and general anesthetics (Borg and Moller, 1975; Uhles et al., 2000; Arslan et al., 2017).

Moreover, although the presence of the MEMR is common, it fails to meet the criteria of being pervasive, as defined by 95% confidence of being observed in at least 95% of the people. An analysis of data from the National Health and Nutrition Examination Survey (NHANES) in 2017 found that in over 15 000 participants, the MEMR failed to be pervasive with prevalence of 86.9% in adults 18–30 with normal hearing, 85.3% in all adults 18–30, and 74/6% across all participants (McGregor et al., 2018). These findings suggest that the MEMR is unlikely to be effective as a singular metric for any clinical study. For studies evaluating experimental medicines, the FDA stipulates that the outcome of the treatment yields clinically significant results (Le Prell, 2021, 2022). The lack of prevalence and criteria for clinical significance of the MEMR at this time present a significant limitation for the use of the MEMR as a singular metric for intervention studies for hearing loss. However, despite these limitations, the MEMR may prove useful as an objective measure and correlate of clinically significant suprathreshold auditory deficits. In individuals with normal hearing sensitivity but reported suprathreshold auditory problems, widely accepted objective correlates of these deficits have yet to be established. The emerging studies appear to suggest that in a subset of these patients, the MEMR may provide measurable and reliable changes that can be used to evaluate intervention among specific clinical populations.