Skip to main content
NCBI home page
Journal of Neurophysiology logoLink to Journal of Neurophysiology
. 2017 Jun 7;118(2):1164–1172. doi: 10.1152/jn.00142.2017

Efferent inhibition strength is a physiological correlate of hyperacusis in children with autism spectrum disorder

Uzma S Wilson 1, Kate M Sadler 1, Kenneth E Hancock 2,3, John J Guinan Jr 2,3, Jeffery T Lichtenhan 1,✉
PMCID: PMC5547266  PMID: 28592687

Children with autism spectrum disorder (ASD) are a heterogeneous group, some with hyperacusis and some without. Our research shows that hyperacusis can be estimated in children with ASD by using medial olivocochlear (MOC) reflex measurements. By establishing that an objective measure correlates with attributes of hyperacusis, our results enable future work to enable subtyping of children with ASD to provide improved individualized treatments to at-risk children and those without adequate language to describe their hyperacusis symptoms.

Keywords: cochlea, otoacoustic emissions, medial olivocochlear reflex, sound level tolerance

Abstract

Autism spectrum disorder (ASD) is a developmental disability that is poorly understood. ASD can influence communication, social interaction, and behavior. Children with ASD often have sensory hypersensitivities, including auditory hypersensitivity (hyperacusis). In adults with hyperacusis who are otherwise neurotypical, the medial olivocochlear (MOC) efferent reflex is stronger than usual. In children with ASD, the MOC reflex has been measured, but without also assessing hyperacusis. We assessed the MOC reflex in children with ASD by measuring the strength of MOC-induced inhibition of transient-evoked otoacoustic emissions (TEOAEs), a noninvasive physiological measure that reflects cochlear amplification. MOC activity was evoked by contralateral noise. Hyperacusis was assessed subjectively on the basis of the children’s symptoms. We found a significant correlation between hyperacusis scores and MOC strength in children with ASD. When children were divided into ASD-with-severe-hyperacusis (ASDs), ASD-with-not-severe-hyperacusis (ASDns), and neurotypical (NT) groups, the last two groups had similar hyperacusis and MOC reflexes, whereas the ASDs group, on average, had hyperacusis and MOC reflexes that were approximately twice as strong. The MOC inhibition of TEOAEs averaged larger at all frequencies in the ASDs compared with ASDns and NT groups. The results suggest that the MOC reflex can be used to estimate hyperacusis in children with ASD and might be used to validate future questionnaires to assess hyperacusis. Our results also provide evidence that strong MOC reflexes in children with ASD are associated with hyperacusis and that hyperacusis is a comorbid condition and is not a necessary, integral part of the abnormal neural processing associated with ASD.

NEW & NOTEWORTHY Children with autism spectrum disorder (ASD) are a heterogeneous group, some with hyperacusis and some without. Our research shows that hyperacusis can be estimated in children with ASD by using medial olivocochlear (MOC) reflex measurements. By establishing that an objective measure correlates with attributes of hyperacusis, our results enable future work to enable subtyping of children with ASD to provide improved individualized treatments to at-risk children and those without adequate language to describe their hyperacusis symptoms.

autism spectrum disorder (ASD) is a neurological condition characterized by a range of impairments in the sensory, behavioral, language, and social-communication domains. Within the sensory domain, abnormally low sound tolerance or auditory hypersensitivity (termed “hyperacusis”) affects at least 40% of children with autism (Rimland and Edelson 1995). Hyperacusis can be behaviorally manifested in responses to moderately loud sounds in children with ASD by actions such as covering the ears, crying, or running away, and these aversive behaviors can disrupt academic and social functioning of individuals with ASD (World Health Organization 2001). Although hyperacusis itself can lead to serious maladaptations in children, other aspects of ASD such as poor emotional and social regulation can exacerbate these behaviors (American Psychiatric Association 1994). Therefore, it is important to understand what influence hyperacusis may have on a child’s daily life. Ascertaining this role requires measuring hyperacusis in children with ASD. A test for hyperacusis that is tolerated by children with ASD would aid in the diagnosis and treatment planning for children with ASD as well as adding to our understanding of ASD.

A typical test to quantify hyperacusis requires slowly increasing the level of a sound until the individual reports that it is uncomfortable. This loudness discomfort level test might be done successfully by neurotypical (NT) children, but it is unsuitable for many children with ASD because their reporting ability may be compromised (Cascio et al. 2016). Furthermore, when the sound becomes too loud, children with ASD may respond by exhibiting extreme behaviors, such as aggression. In adults, hyperacusis can be assessed using subjective questionnaires; however, no such tool has been fully validated for children with ASD (Egelhoff and Lane 2013; Robertson and Simmons 2013). NT adults with hyperacusis show increased responses to acoustic stimuli in the inferior colliculus, thalamus, and auditory cortical regions, increases that have been measured by functional magnetic resonance imaging (fMRI) (Gu et al. 2010). However, fMRI is not suitable for children with ASD, because many would not tolerate the loud scanner noise and having to stay motionless in the confined space of an fMRI machine. In contrast to the above methods, an approach that does seem suitable for children with ASD is the measurement of the strength of medial olivocochlear (MOC) reflex inhibition on otoacoustic emissions (OAEs). MOC reflex strength has been shown to be greater in NT adults with hyperacusis compared with NT controls without hyperacusis (Knudson et al. 2014).

The MOC efferents are a neural pathway by which the brain adjusts the gain of mechanical amplification within the cochlea (termed “cochlear amplification”). OAEs are sounds that result from cochlear amplification and that can be measured noninvasively with a sensitive, low-noise microphone placed in the external ear canal. MOC activity reduces cochlear amplification, and thereby reduces OAE amplitudes (Guinan 2006). Several prior reports deal with MOC effects on OAEs in children with ASD, but their conclusions conflict with each other and their methods were not always adequate (Collet et al. 1993; Danesh and Kaf 2012; Kaf and Danesh 2013; see discussion). Furthermore, in these studies, hyperacusis was not measured. What is needed is a study of MOC effects on OAEs in children with and without ASD that uses adequate methods and takes into account the degree of hyperacusis in each ASD child.

MATERIALS AND METHODS

Participants

Twenty ASD males (average age = 13.1 yr, range 10–16 yr) were recruited from the Special School District of Saint Louis on the basis of initially meeting ASD Educational Eligibility (DSM-V; American Psychiatric Association 2013; http://idea.ed.gov/ or http://www.naset.org/fileadmin/user_upload/Forms_Checklist_Etc/IEP_Committee/Eligibility_Criteria_Chklt_Diagnosis_Autism.pdf). ASD is more prevalent in males than females by a factor of 4:1 (Rubenstein and Merzenich 2003). Only male children were recruited to avoid possible sex differences. The measurements were done in the summer of 2014; all children with ASD had a medical diagnosis of ASD made before 2013. Complete sets of MOC reflex data could not be collected from two children with severe ASD: one ASD child was echolalic and could not sit without verbal reciting, and the other had extreme behavioral and communication difficulties that may have originated, in part, from hyperacusis to the test stimuli. MOC reflex data were obtained from the remaining 18 children with ASD and from 14 NT male controls. NT children were recruited from the greater St. Louis metropolitan area (average age = 13.2 yr, range 10–16 yr). All children had evidence of normal auditory status as determined by pure tone audiometric screening at 20-dB hearing level (HL) at 1–8 kHz with the use of a screening audiometer (Earscan, model ES-AMN; Micro Audiometrics, Murphy, NC) or the presence of otoacoustic emissions in response to moderate-level click stimuli (used when a child with ASD could not participate in behavioral audiometric screening). Most children were able to complete the study within a single 1.5-h testing session, although some children with ASD required multiple sessions to complete the study. Parents and children voluntarily consented to all procedures, which were approved by the Institutional Review Board of Washington University in St. Louis (201404145).

Hyperacusis Assessment

The hyperacusis assessment was subjective. We assessed hyperacusis in a semi-structured interview by asking questions from published hyperacusis questionnaires (Egelhoff and Lane 2013; Khalfa et al. 2002) that were modified for each child to accommodate the language skills of the children with ASD. The questions were administered by the second author (K.M.S., an ASD expert with 15 yr of experience in special education) to parents and simultaneously to children when they had the language ability to answer. The questions elicited the recall of various attributes of hyperacusis and defensiveness such as sounds the children were fearful of (e.g., the noise of a flushing toilet, vacuum cleaner, hairdryer) and associated behavioral responses (running away, crying, covering the ears), and the answers were recorded. These responses reported behavioral responses to sounds but not the sound levels that evoked these behaviors. The two authors who interacted with the children (U.S.W., K.M.S.) arrived at a consensus hyperacusis score for each child using the interview answers and their personal observations of the children. A similar process was used to score the NT children (the scorers knew who were children with ASD and who were not). Hyperacusis scoring was done before the authors knew the results of the MOC reflex strength tests. Hyperacusis scores could not be obtained in one child with ASD because he withdrew from the study and in one NT child who could not be recontacted following an initial visit.

Division into hyperacusis groups.

To enable us to see patterns that are not visible in the data from a single child, we averaged data across groups of children. Knudson et al. (2014) found that in adults with hyperacusis, increased MOC inhibition occurs mainly for adults with severe hyperacusis. This guided us to divide the children with ASD into two groups, those with severe hyperacusis and those with not-severe hyperacusis. The questionnaire and hyperacusis evaluation process used by Knudson et al. (2014) were very different from our assessment procedure that was used on a population with special needs, and thus Knudson et al.’s hyperacusis criterion could not be directly translated to our hyperacusis scores. On the basis of our assessment of which scores would lead to severe impairment in everyday life, we arbitrarily chose severe hyperacusis to be scores of 5 and higher (the maximum score among NT children was 2). This criterion was chosen before the MOC reflex data were analyzed. It put 5 children with ASD into the severe hyperacusis group (ASDs) and the 12 remaining scored and MOC-tested children with ASD into a not-severe hyperacusis group (ASDns). The 13 scored and MOC-tested NT children formed a separate group (NT).

MOC Reflex Measurements Using TEOAEs With and Without Contralateral Noise

We measured the effect of the MOC reflex on transient evoked otoacoustic emissions (TEOAEs). TEOAEs are faint sounds recorded in the ear canal following click stimulation, and their amplitude reflects the amount of cochlear amplification. Stimulus calibration, stimulus generation, data acquisition, and data storage were performed with a National Instruments PXI-1031 chassis and the Eaton-Peabody-Laboratory Cochlear Function Test Suite. Custom-written software in MATLAB (The MathWorks, Natick, MA) was used for offline analyses. Ipsilateral click stimuli for evoking TEOAEs and contralateral acoustic stimulation (CAS) broadband noise (0.1–10 kHz, 60 dB SPL) for activating the MOC reflex were routed through a TDT SA-1 amplifier (Tucker-Davis Technologies) and delivered via separate ER-10C sound sources in each ear. TEOAEs were recorded with the ER-10C microphone (Etymotic Research, Elk Grove Village, IL) coupled to the right ear canal using a foam tip. Sound calibrations using in-ear chirps were performed at the beginning of each session and between MOC reflex measurements to account for drift in stimulus sound pressure levels (Lichtenhan et al. 2016). Following calibrations, 100-µs, 40/s clicks were delivered to the right ear. TEOAE responses were recorded with and without CAS for click levels ranging from 50 to 95 dB SPL. At each click level, data were obtained in 4 blocks of 820 responses in the following sequence: no-CAS, with-CAS, no-CAS, and with-CAS (blocks termed A, B, C, and D). All data were streamed to disk for posttest signal processing.

Data Analysis

Noise rejection.

Not all children were able to sit quietly for an entire TEOAE measurement session. Because of this, we conducted extensive posttest data processing to reject noisy measurements and ensure adequate signal-to-noise ratios (SNRs). Raw data stored on disk were processed offline in MATLAB using custom noise rejection criteria described in detail below. From the data that passed the noise criteria we calculated various CAS-induced MOC effects. Many sets of criteria were tried (varying one criterion at a time) with the goal of optimizing the SNR of the accepted data while not making the criteria so strict that many children would have all their data rejected.

Individual responses (traces) in each block of data were filtered (1–10 kHz, 4th-order zero-phase-change digital filter) and averaged, and this average was subtracted from each filtered trace. This yielded 820 “trace residuals” that were quantified by their root mean square (rms) values. The 820 sequentially displayed residual rms values ranged from having steady, low-amplitude variations to having huge bursts (>>10 times the average) to occasionally showing breathing-rate cyclical variations. Traces that had residual rms values that were too high or too low were rejected in a two-step procedure. First, traces were removed that had residual rms amplitude more than 2.5 times the rms amplitude of traces in the 20th percentile from the bottom of the distribution of residual rms amplitudes from that block. Second, from only the traces remaining after the first cut, traces were removed whose residual rms deviated from the new average residual by more than ±1.8 times the standard deviation. The filtered traces corresponding to the residuals that passed these criteria were averaged to get the recorded click and TEOAE waveforms of the block, and were averaged again with alternate traces reversed in polarity (even number used) to get the noise waveform of the block.

MEM reflex and acoustic drift.

Probe movement or contraction of the stapedius muscle can result in inaccurate estimation of the MOC reflex. Variations in the recorded sound pressure levels of the click itself were used to measure any drift or abrupt changes in the acoustic assembly fit and to monitor for possible middle-ear-muscle (MEM) contractions (for the rationale, see Guinan 2006). Click amplitudes were measured from the largest click peak (the same peak throughout a block) using only traces that passed the criteria described above. The resulting sequentially arranged click amplitudes were three-point-smoothed four times. From the resulting click amplitudes, we calculated the average value, the maximum upward deviation, the maximum downward deviation, and the deviation range. A group of four blocks was rejected if one block had an upward or downward deviation >10% of the mean or a deviation range that was >10% of the mean. To ensure that CAS-induced MEM contractions had little or no influence on our results, we calculated the CAS-induced change in click amplitude and removed data that might have come from trials with MEM contractions (see Guinan 2006). The percentage change was calculated from block-average click amplitudes using the formula ΔclickAmp = 100 × (A − B + C − D)/[(A + B + C + D)/4]. Similarly, the percentage drift was calculated as drift = 100 × (A + B − C − D)/[(A + B + C + D)/4]. A block group was rejected if ΔclickAmp was >2.33% (0.2 dB) or the drift was >10%. In previous work with cooperative NT adults (Lichtenhan et al. 2016), we used a stricter MEM criterion of 1.2%; however, that criterion was impractical in the present study because children with ASD often produced seemingly random (i.e., not related to the CAS) changes in click amplitude, presumably due to movement, and the previous MEM criterion would have removed too much data. By applying a particular criterion level, we cannot rule out the possibility that weaker MEM contractions may have occurred. Our 0.2-dB change criterion ensures that any MEM contractions that may have been included would have had a minimal influence on the results.

Signal processing.

The data from each block group that passed all of the above criteria were consolidated by adding each corresponding waveform pair (A + C, B + D) weighted by the number of traces remaining in their average waveforms. This yielded four waveforms for each group (no-CAS and with-CAS TEOAE waveforms and no-CAS and with-CAS noise waveforms). From these waveforms, TEOAE and noise amplitudes were calculated from the rms values in a time window 6 to 16 ms after the click electrical onset. When calculating TEOAEs from the time waveforms, we used the probe tube correction magnitudes linearly averaged from 1 to 8 kHz. When calculating TEOAE values in half-octave frequency ranges, we selected the TEOAE time region by windowing the click-response waveform using a window with raised cosine-shaped edges of 1-ms duration (to avoid the frequency splatter produced by a window with sharp edges). The window-selected TEOAE waveform was fast-Fourier transformed (FFT) to the frequency domain, where each frequency was multiplied by the magnitude of the corresponding frequency response of the probe tube correction. The result was divided into half-octave bands by linearly averaging the rms magnitudes of each FFT frequency within the band. After the above exclusions, 191 of the original 308 sets of measurements from scored and MOC-tested children remained, and all data from 2 NT children and from 1 ASDns child were eliminated. The percentage of measurements rejected was high because many children could not sit quietly throughout the measurements. However, rejecting the noisy data enabled us to achieve SNRs that averaged 52.6 and 52.2 dB for the no-CAS and with-CAS data, respectively.

Statistical Analyses

The Pearson product-moment correlation coefficient (MATLAB corrcoef) “r” was used for correlation analysis between hyperacusis score and CAS-induced decibel change in TEOAE amplitudes. Analysis of variance (MATLAB anovan) was performed to estimate the significance of TEOAE differences across frequency, group, and frequency × group interaction. P < 0.05 was considered significant.

RESULTS

Hyperacusis Severity Varied Among Children with ASD

We assessed the degree of hyperacusis in 17 children with ASD and in 13 NT control children. The children with ASD had a wide range of hyperacusis scores (0 to 8), whereas the NT control children had much lower scores (0 to 2) (Fig. 1). Examples of the kinds of differences we found are that most of the children with ASD (10/17) reported being startled in response to loud sounds, whereas few NT controls did (2/13). Children with ASD reported covering ears (7/14) and/or crying (4/13) in response to loud sounds, with the sounds of vacuum cleaners and toilet flushing being most bothersome, but NT children seldom reported these behaviors (1/13 reported covering ears and 0/13 reported crying).

Fig. 1.

Fig. 1.

Open in a new tab

Histograms of hyperacusis scores for children with autism spectrum disorder (ASD; left; n = 17) and neurotypical (NT) children (right; n = 13). Cross (×) signifies children with severe hyperacusis.

TEOAE Results Overview

TEOAE measurements that passed our noise rejection criteria were from 11 NT, 11 ASDns, and 5 ASDs children. These measurements, as a function of sound level, are shown in Fig. 2, A, C, and E. The data are separated into the three groups, and for each group, regression lines have been fit separately to the no-CAS and the with-CAS data. For each data set, the with-CAS lines are at lower TEOAE amplitudes than the no-CAS lines, which is consistent with there being CAS-induced MOC inhibition in each group. The CAS-induced TEOAE reductions were similar for ASDns and NT children, and were larger for ASDs children. Statistical tests were not applied to these data because such tests would be biased by overweight of children with measurements at more levels.

Fig. 2.

Fig. 2.

Open in a new tab

A, C and E: transient-evoked otoacoustic emission (TEOAE) amplitude as a function of click level without contralateral acoustic stimulation (CAS; circles) and with CAS (triangles). B, D, and F: the CAS-induced decrease in TEOAE (ΔTEOAE) as a function of click level. Each point in B, D, and F is the difference between a pair of points in corresponding panels A, C, and E. A and B: neurotypical children (NT). C and D: children with ASD and not-severe hyperacusis (ASDns). E and F: children with ASD and severe hyperacusis (ASDs). Shown are all of the data that passed the noise criteria from 11 NT, 11 ASDns, and 5 ASDs children. Diagonal lines are regressions through the points (in A, C, and E: solid lines for no-CAS and dotted lines for with-CAS). The changes in TEOAE for A–F, in order, are 1.51, 1.51, 1.63, 1.64, 2.45, and 2.34 dB. For A, C, and E, the TEOAE change is the distance between the lines at 70 dB SPL. For B, D, and F, it is the average of all points in each panel.

The differences between the no-CAS and with-CAS TEOAE amplitudes (ΔTEOAE) are shown in Fig. 2, B, D, and F. Regression analysis of the data (lines in Fig. 2, B, D, and F) indicate that the MOC-induced change in the TEOAE (ΔTEOAE) decreased as sound level increased. The slope of the decrease with sound level was greater for the ASDs group, but this group also had the largest average ΔTEOAE (Fig. 2F). Values of the slope divided by the ΔTEOAE (evaluated at 70 dB SPL) were little different for the three groups (−1.3, −1.9, and −2.0 for NT, ASDns and ASDs groups, respectively). This is consistent with there being different overall MOC effects across groups but that all groups have similar dependencies of the MOC effect on sound level.

MOC Reflex Strength is Positively Correlated with the Hyperacusis Score

Because ΔTEOAE can change with sound level, the ideal way to evaluate the data would be from ΔTEOAE measurements obtained at the same sound level. This would remove any bias due to differences in the number of measurements made at low vs. high sound levels. Although our data were scattered across levels, an evaluation at one level could be done by fitting lines to each child’s data plotted as level functions (e.g., like the group growth-function lines in Fig. 2, A, C, and E) or as ΔTEOAEs (like the lines in Fig. 2, B, D, and F). However, after noisy data were rejected, some children had good ΔTEOAE measurements only at one sound level, and a growth-function line cannot be derived from one sound level. As a first evaluation method that includes data from as many children as possible, for each child we averaged all of the ΔTEOAE measurements. This gave us a single ΔTEOAE metric for each child. This ΔTEOAE is plotted against the hyperacusis score for each child in Fig. 3, A and B. NT children had no significant correlation between their ΔTEOAEs and their hyperacusis scores [Pearson’s r(10) = −0.18, P = 0.58; Fig. 3A], largely because most NT children had a hyperacusis score of 0. In contrast, children with ASD had a statistically significant correlation between their ΔTEOAEs and their hyperacusis scores [Pearson’s r(15) = 0.58, P = 0.018; Fig. 3B]. Although this assessment includes the greatest number of subjects, it may be biased by subjects being measured at different sound levels and the MOC effect changing with sound level.

Fig. 3.

Fig. 3.

Open in a new tab

CAS-induced reductions in TEOAE amplitudes as functions of the child’s hyperacusis score. A and B: ΔTEOAE values from averaging each child’s ΔTEOAE measurements from all sound levels in NT children (A; n = 11) and children with ASD (B; n = 16). C and D: TEOAE vertical amplitude reductions (C) and horizontal sound-level shifts (D) from fitting lines to the growth function of each child with ASD, as in Fig. 2, A, C, and E (n = 15). In B–D the correlations between ΔTEOAE and hyperacusis score are statistically significant: B, Pearson’s r = 0.58, P = 0.018, n = 16; C, r = 0.76, P = 0.0068, n = 11; and D, r = 0.78, P = 0.0050, n = 11.

An alternate method for measuring each child’s CAS-induced TEOAE change that avoids this bias is to fit lines to their no-CAS and with-CAS TEOAEs (like the growth-function lines in Fig. 2, A, C, and E) and to compare all children at the same sound level. The distance between a child’s growth-function lines yields two metrics: the TEOAE amplitude reduction (vertical change) and the TEOAE shift along the sound-level axis (horizontal change). TEOAE amplitude reductions are the most common metric in the literature. However, sound-level shifts are the preferred metric, because the shift (the amount that the sound level must be increased to compensate for the TEOAE reduction from MOC inhibition) equals the reduction in cochlear amplification. Because fitting a line to the data requires at least two good points for each child, we were only able to apply this method to 11 children with ASD. These data should be more accurate than the prior ΔTEOAE metric because they all have at least two good points, and they were all evaluated at the same sound level (70 dB SPL). Both metrics showed statistically significant correlations between the TEOAE changes and hyperacusis scores [Fig. 3C amplitude reductions: Pearson’s r(10) = 0.76, P = 0.0068; Fig. 3D level shifts: Pearson’s r(10) = 0.78, P = 0.0050]. Because it appeared that these correlations might be strongly influenced by the data from one subject who had a hypersensitivity score of 8, we calculated the correlations without that point. The resulting correlations were, for amplitude reductions: Pearson’s r(9) = 0.49, P = 0.152, and for level shifts: Pearson’s r(9) = 0.58, P = 0.08. The correlations are no longer statistically significant (although the level-shift correlation was almost significant), and the correlation values are reduced to near half. We emphasize that there is no reason to doubt the validity of the data from this subject, but the lack of statistical significance at the 0.05 level without this point does show the thinness of our data.

When the growth-function metrics were divided into the three groups, the NT and ASDns groups had similar changes, whereas the ASDs group had much greater changes (Fig. 4). The dividing line between the two ASD groups was somewhat arbitrary (i.e., it was not justified by an independent set of observations), so doing statistical tests with the two ASD groups considered as separate entities would have questionable statistical validity. The main point of Fig. 4 is that results for the children with ASD but without severe hyperacusis are very similar to those for NT children.

Fig. 4.

Fig. 4.

Open in a new tab

CAS-induced TEOAE level shift vs. reduction in TEOAE amplitude (ΔTEOAE), calculated from each child’s TEOAE growth functions, for the 3 groups of children (NT: circle, n = 11; ASDns: square, n = 10; ASDs: triangle, n = 5). Error bars are ±SE. Diagonal line has slope = 2.

Another thing that Fig. 4 shows is that the level shift is approximately twice as great as the amplitude reduction. Thus the common metric of TEOAE amplitude reduction considerably underestimates the CAS-induced MOC efferent reduction in cochlear amplification, which averaged close to 8 dB for the ASDs children.

MOC Inhibition Across Frequency

To determine the frequency distributions of the TEOAE reductions, we used the data averaged for all of the children in each of the three groups and analyzed the data in half-octave frequency bands from 0.5 to 8 kHz. In all frequency bands, the average CAS-induced reduction in TEOAE amplitude for the ASDs group was higher than that for the other two groups (Fig. 5). An ANOVA on the CAS-induced changes in TEOAE amplitude with the three groups and eight frequency bands as factors showed a highly statistically significant difference across frequency [F(7) = 10.16, P < 10−6] and across group [F(2) = 9.76, P = 0.0001]. ANOVA showed no significant interaction between group and frequency [F(14) = 0.57, P = 0.88)]. Although the differences between the ASDs group and the other two groups were larger at low frequencies than at mid and high frequencies, the error bars are also larger at low frequencies. The low P values were obtained in this case because the differences across children were generally consistent across frequency. Note that calculating statistical significance across the two artificially created ASD groups is not valid, as pointed out earlier. The justification for saying that the MOC effect is larger for children with higher hyperacusis scores is the statistically significant correlations shown in Fig. 3.

Fig. 5.

Fig. 5.

Open in a new tab

CAS-induced change in TEOAE amplitude (ΔTEOAE) as a function of frequency for the 3 groups of children (NT: circle, n = 11; ASDns: square, n = 11; ASDs: triangle, n = 5). Points are the mean TEOAE amplitude reductions, averaged across children, in half-octave frequency bands shown at the logarithmic center frequency of the band. Error bars are ±SE.

Differences in MOC Effects Are Not Produced by Differences in Cochlear Sensitivity

We explored whether the MOC effects (i.e., the CAS-induced decreases in TEOAE amplitudes) might be attributable to different behavioral thresholds and/or different baseline (no-CAS) TEOAE amplitudes across groups. We do not have behavioral thresholds for all children whose data were used. There were, however, no marked differences between NT children and children with ASD in the behavioral thresholds to clicks (averages ± SD: NT = 40.4 ± 5.7 dB SPL, n = 14; and ASD = 35.4 ± 4.0 dB SPL, n = 12). We do have TEOAE levels for all children, and these provide objective measures of cochlear sensitivity, which we use in the following analysis. To assess possible differences in baseline TEOAE levels across the three groups, we must consider that different click levels were used for different children. To account for this, each child’s baseline TEOAE level was determined by the regression line fit to the no-CAS TEOAE amplitude vs. click level data (as was done for the data in Fig. 2, A, C, and E) and was evaluated at 70 dB SPL. This yielded average TEOAE baseline levels of 9.42, 6.14, and 8.85 dB SPL for the NT, ASDns, and ASDs groups, respectively.

To interpret these baseline TEOAE differences, we need to know how much the differences in baseline levels influenced the CAS-induced decreases in TEOAE levels. An estimate of this influence was calculated from a scatter plot that included all the data from all groups of the CAS-induced decreases in TEOAE levels vs. the corresponding baseline TEOAE levels. This scatter plot had an overall slope of 0.0444 dB/dB, i.e., on average, a 1-dB increase in the baseline TEOAE level was accompanied by a 0.0444-dB increase in the CAS-induced decrease in TEOAE level. To determine if differences in baseline TEOAE levels may account for the difference in CAS-induced TEOAE reductions between the ASDns and ASDs groups (the main group difference that might be explained by differences in baseline TEOAE levels), we did the following calculation: the 2.71-dB difference from the ASDns to ASDs group in baseline TEOAE level multiplied by the slope of 0.0444 dB/dB yielded the estimate that the baseline level difference would produce a 0.120-dB increase in CAS-induced TEOAE reduction from the ASDns to the ASDs group. Thus the difference in baseline TEOAE levels would account for only 15% of the 0.78 dB greater CAS-induced decrease in TEOAE level found in the ASDs group compared with the ASDns group.

DISCUSSION

We found that for children with ASD, there was a significant correlation between the subjectively assessed hyperacusis score and the CAS-induced MOC inhibition of TEOAEs. This was true for all three of our approaches to quantifying TEOAE changes. When children with ASD were divided into groups with severe hyperacusis and not-severe hyperacusis, the TEOAE changes were highest for the ASDs group and were similar in the ASDns and NT groups. Finally, the ASDns and NT children had similar ΔTEOAE vs. frequency functions, whereas the ASDs children had frequency functions that were more elevated. Although the children often did not sit still and produced many noisy measurements, our custom-designed artifact rejection scheme enabled us to achieve high TEOAE SNRs so that only accurate ΔTEOAE measurements were used. These ΔTEOAEs can be attributed to MOC reduction of cochlear amplification, because the click sound pressure level was unchanged by CAS, which indicates that there were negligible middle-ear-muscle contractions (Guinan 2006).

Subjective Hyperacusis Findings

The greatest weakness of our method is that the hyperacusis scores were subjective. However, the scoring was done before the scorers knew the MOC results, so the scoring was independent of the MOC results. The correlation between hyperacusis score and MOC effect indicates that the subjective scoring process accessed, or was associated with, some physiological aspect of the ASD children’s auditory neural processing of sound. Because the hyperacusis scoring was done by evaluators who worked with the children with ASD, their scoring cannot be considered as agnostic to their knowledge of the child. Indeed, this knowledge might have been an important factor in the score. Notes about the children’s answers to the questions were taken in an informal interview and/or from direct observations and were used when the scores were decided. However, the question wordings varied, and the answers were not recorded in the formal way necessary to validate a questionnaire. Because of this, we are not proposing that our current interviewing method with questions adjusted for each child can be successfully used by others. However, the correlation between our hyperacusis score and the MOC effect suggests that the MOC effect could be used in future studies to get a validated questionnaire. The resulting questionnaire would be “validated” in providing questions that correlate with there being an MOC effect. On the basis of our findings, the questionnaire results would provide some knowledge about the hyperacusis of the subject, but additional work is required to determine exactly how well any questionnaire shows that there is hyperacusis. Considering the wide range of linguistic and other abilities of children with ASD, it is not clear what fraction of ASD children would be able to respond appropriately to a traditional questionnaire (one with questions having a fixed wording and a limited range of acceptable answers).

Our hyperacusis scores reflect both a child’s increased sensitivity to loud sounds and his or her increased reaction toward annoying sounds. An increased score due to heightened sound sensitivity is exactly what is desired for a hyperacusis score. In contrast, the acting-out propensity may only be weakly related to an underlying hyperacusis. A questionnaire might distinguish how much the two factors (increased sensitivity vs. increased reactions) are correlated with increased MOC effect. Questions should be designed to assess each factor separately (e.g., by asking about the tendency to act-out following displeasure from some nonauditory annoyance). The correlation with MOC effects of questions targeted to hyperacusis vs. nonauditory triggers might then tell which factor was more important in eliciting MOC activity.

Prevalence of Hyperacusis in Children with ASD

We found no hyperacusis in NT children. In the 13 scored children with ASD, 5 had severe hyperacusis and 8 had not-severe or no hyperacusis, and we were unable to test 2 children because of their unusual sound-induced behavior. The prevalence of hyperacusis is reported to range from 18% to 69% among children with ASD (cf. Danesh et al. 2015; Rimland and Edelson 1995; Rosenhall et al. 1999). Comparison of these numbers must be done cautiously, because they may not represent measurements from the same ASD subject pool. The challenge in testing children with ASD is not simply to test a large number of children but to test a high percentage of the whole ASD spectrum. Although we attempted to cover the entire ASD range, we did not succeed in measuring from all severely impacted children with ASD.

Published Work on OAEs and MOC Effects from Children with ASD

In our sample, NT children and children with ASD had similar baseline TEOAE levels and behavioral thresholds. Gravel et al. (2006) and Tharpe et al. (2006) reported no significant differences in behavioral thresholds, auditory brain stem responses, or OAE levels between ASD and NT children. More recently, Bennetto et al. (2017) measured OAEs in children with ASD but reported only SNRs, and not OAE levels. Cochlear function cannot be assessed from SNR levels alone.

MOC reflex effects have been measured previously in children with ASD, but the results were inconclusive. In a brief letter, Collet et al. (1993) reported lower MOC effects in subjects with ASD compared with normal subjects, but not enough methodological detail was given to evaluate the reliability of the ASD diagnosis or the MOC effects. Furthermore, Collet et al. (1993) did not consider differences in degree of hyperacusis. Danesh and Kaf (2012) and Kaf and Danesh (2013) measured MOC effects in children with ASD, but they reported only the change in SNR. It is impossible to determine from these reports whether the SNR change was from a MOC effect or a change in the noise (CAS may elicit more motion and noise in children with ASD). Last, we found MOC-induced inhibition of TEOAEs was greatest at frequencies below 3 kHz (Fig. 5). This is consistent with previous results from NT adults (Francis and Guinan 2010; Giraud et al. 1996; Knudson et al. 2014).

Relation Between Hyperacusis and the MOC Reflex

It has been hypothesized that children with ASD have generalized hypersensitivity and impaired inhibition of sensory inputs (Lucker and Doman 2015; Rubenstein and Merzenich 2003). Our results are consistent in that children with ASD reported greater signs of hyperacusis than NT children. MOC activity reduces cochlear amplification, responses in afferent auditory-nerve fibers, and responses in subsequent auditory pathways. Consequently, MOC activity should reduce hyperacusis. A reduction of peripheral output over hours to months can induce an increase in central neural activity. The MOC activation measured in the present study changes over a 100-ms timescale (Backus and Guinan 2006). Although the MOC reflex may help by transiently reducing hyperacusis, it does not appear to help enough to remove the hyperacusis.

One proposed connection between hyperacusis and increased MOC effect is through an increased gain in the central auditory pathways (the “central gain” hypothesis; see reviews, Knipper et al. 2013; Pienkowski et al. 2014; Tyler et al. 2014). This hypothesis proposes that certain brain stem neural mechanisms (e.g., decreased inhibition) lead to increased neural responses to sound, and these increased neural responses produce both a hypersensitive behavioral response to sound and an increased MOC reflex. A change in brain stem auditory gain might come about from abnormal top-down influences on brain stem auditory circuits. Such a top-down influence may or may not be related to an underlying cause of ASD. The findings that ASDns and NT children had similar MOC effects whereas ASDs children had much larger MOC effects suggest that hyperacusis with an increased MOC effect is not a symptom of ASD but is a comorbid neurophysiological anomaly of the auditory system (i.e., a child can have ASD without having hyperacusis). The central gain hypothesis is strengthened by the evidence that hyperacusic adults without ASD also demonstrate a stronger MOC reflex compared with those without hyperacusis (Knudson et al. 2014). However, an increased MOC reflex may be a common compensatory response to the problem of hyperacusis so that its presence does not mean that the hyperacusis has the same origin in different people.

Potential Uses of the MOC Reflex in Diagnosing Hyperacusis in ASD

We found a significant correlation between behaviorally assessed hyperacusis and physiological measurements of the MOC reflex in children with ASD. The correlation suggests that MOC reflex strength could potentially be used as an objective assessor of hyperacusis and could facilitate the diagnosis of hyperacusis in children with ASD. Assessing hyperacusis has been done by obtaining loudness discomfort level (LDL) scores, but Zaugg et al. (2016) reported that LDL scores are poorly correlated with subjective hyperacusis scores and concluded that LDLs do not accurately represent a patient’s ability to tolerate loud sounds. Because we found a significant correlation between subjective hyperacusis scores and MOC effects, a MOC-effect test may be better than the LDL test in indicating a patient’s ability to tolerate loud sounds.

Currently, there are no validated questionnaires to quantify hyperacusis in children with ASD. A validated questionnaire would extend the ability to determine hyperacusis in children for whom a MOC test would be difficult. A MOC test could be used to validate preliminary questionnaires. To do this, the MOC test must produce accurate measurements for each child (see Guinan 2006; Marshall et al. 2014). Because the MOC test requires taking the difference between two OAE measurements (OAEs with and without CAS), each measurement must have a much higher SNR (e.g., SNR >25 dB; see Goodman et al. 2013) than a standard single OAE measurement. The MOC test should be powered to detect a 1-dB change, or less, for each child. Measurements done with low SNRs will produce scattered MOC effects that will wash out possible correlations. Our high OAE SNRs may be an important reason why we found significant correlations of MOC effect and hyperacusis score. Achieving a high SNR can be challenging in an ASD child, and more than the usual artifact rejection methods may be necessary.

ASD is a developmental disorder, yet the majority of ASD research is completed on people after they have already developed symptoms. A potential study area is to determine if the MOC reflex can predict which at-risk children (e.g., those born into families with ASD) will have hyperacusis. Accurate prediction could guide early intervention.

GRANTS

This research was funded by the McDonnell Center for Systems Neuroscience (to J. T. Lichtenhan), the American Otological Society (to J. T. Lichtenhan), and National Institute on Deafness and Other Communication Disorders Grant R01DC005977 (to J. J. Guinan Jr).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

U.S.W., J.J.G., and J.T.L. conceived and designed research; U.S.W. and K.M.S. performed experiments; U.S.W., K.M.S., K.E.H., J.J.G., and J.T.L. analyzed data; U.S.W., K.M.S., J.J.G., and J.T.L. interpreted results of experiments; U.S.W. and J.J.G. prepared figures; U.S.W. and J.T.L. drafted manuscript; U.S.W., J.J.G., and J.T.L. edited and revised manuscript; U.S.W., K.M.S., K.E.H., J.J.G., and J.T.L. approved final version of manuscript.

ACKNOWLEDGMENTS

We thank Dr. Martin Pienkowski and two anonymous reviewers for their productive criticisms.

Present affiliation for U. S. Wilson: Northwestern University.

Present affiliation for K. M. Sadler: University of Missouri-Columbia.

REFERENCES

  1. American Psychiatric Association Diagnostic and Statistical Manual of Mental Disorders (4th ed.). Washington, DC: American Psychiatric Association, 1994. [Google Scholar]
  2. American Psychiatric Association Diagnostic and Statistical Manual of Mental Disorders (5th ed.). Washington, DC: American Psychiatric Association, 2013. [Google Scholar]
  3. Backus BC, Guinan JJ Jr. Time-course of the human medial olivocochlear reflex. J Acoust Soc Am 119: 2889–2904, 2006. doi: 10.1121/1.2169918. [DOI] [PubMed] [Google Scholar]
  4. Bennetto L, Keith JM, Allen PD, Luebke AE. Children with autism spectrum disorder have reduced otoacoustic emissions at the 1 kHz mid-frequency region. Autism Res 10: 337–345, 2017. doi: 10.1002/aur.1663. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Cascio CJ, Lorenzi J, Baranek GT. Self-reported pleasantness ratings and examiner-coded defensiveness in response to touch in children with ASD: effects of stimulus material and bodily location. J Autism Dev Disord 46: 1528–1537, 2016. doi: 10.1007/s10803-013-1961-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Collet L, Roge B, Descouens D, Moron P, Duverdy F, Urgell H. Objective auditory dysfunction in infantile autism. Lancet 342: 923–924, 1993. doi: 10.1016/0140-6736(93)91969-S. [DOI] [PubMed] [Google Scholar]
  7. Danesh AA, Kaf WA. DPOAEs and contralateral acoustic stimulation and their link to sound hypersensitivity in children with autism. Int J Audiol 51: 345–352, 2012. doi: 10.3109/14992027.2011.626202. [DOI] [PubMed] [Google Scholar]
  8. Danesh AA, Lang D, Kaf W, Andreassen WD, Scott J, Eshraghi AA. Tinnitus and hyperacusis in autism spectrum disorders with emphasis on high functioning individuals diagnosed with Asperger’s Syndrome. Int J Pediatr Otorhinolaryngol 79: 1683–1688, 2015. doi: 10.1016/j.ijporl.2015.07.024. [DOI] [PubMed] [Google Scholar]
  9. Egelhoff K, Lane AE. Brief report: preliminary reliability, construct validity and standardization of the Auditory Behavior Questionnaire (ABQ) for children with autism spectrum disorders. J Autism Dev Disord 43: 978–984, 2013. doi: 10.1007/s10803-012-1626-5. [DOI] [PubMed] [Google Scholar]
  10. Francis NA, Guinan JJ Jr. Acoustic stimulation of human medial olivocochlear efferents reduces stimulus-frequency and click-evoked otoacoustic emission delays: Implications for cochlear filter bandwidths. Hear Res 267: 36–45, 2010. doi: 10.1016/j.heares.2010.04.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Giraud AL, Perrin E, Chéry-Croze S, Chays A, Collet L. Contralateral acoustic stimulation induces a phase advance in evoked otoacoustic emissions in humans. Hear Res 94: 54–62, 1996. doi: 10.1016/0378-5955(96)00002-0. [DOI] [PubMed] [Google Scholar]
  12. Goodman SS, Mertes IB, Lewis JD, Weissbeck DK. Medial olivocochlear-induced transient-evoked otoacoustic emission amplitude shifts in individual subjects. J Assoc Res Otolaryngol 14: 829–842, 2013. doi: 10.1007/s10162-013-0409-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Gravel JS, Dunn M, Lee WW, Ellis MA. Peripheral audition of children on the autistic spectrum. Ear Hear 27: 299–312, 2006. doi: 10.1097/01.aud.0000215979.65645.22. [DOI] [PubMed] [Google Scholar]
  14. Gu JW, Halpin CF, Nam E-C, Levine RA, Melcher JR. Tinnitus, diminished sound-level tolerance, and elevated auditory activity in humans with clinically normal hearing sensitivity. J Neurophysiol 104: 3361–3370, 2010. doi: 10.1152/jn.00226.2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Guinan JJ., Jr Olivocochlear efferents: anatomy, physiology, function, and the measurement of efferent effects in humans. Ear Hear 27: 589–607, 2006. doi: 10.1097/01.aud.0000240507.83072.e7. [DOI] [PubMed] [Google Scholar]
  16. Kaf WA, Danesh AA. Distortion-product otoacoustic emissions and contralateral suppression findings in children with Asperger’s Syndrome. Int J Pediatr Otorhinolaryngol 77: 947–954, 2013. doi: 10.1016/j.ijporl.2013.03.014. [DOI] [PubMed] [Google Scholar]
  17. Khalfa S, Dubal S, Veuillet E, Perez-Diaz F, Jouvent R, Collet L. Psychometric normalization of a hyperacusis questionnaire. ORL J Otorhinolaryngol Relat Spec 64: 436–442, 2002. doi: 10.1159/000067570. [DOI] [PubMed] [Google Scholar]
  18. Knipper M, Van Dijk P, Nunes I, Rüttiger L, Zimmermann U. Advances in the neurobiology of hearing disorders: recent developments regarding the basis of tinnitus and hyperacusis. Prog Neurobiol 111: 17–33, 2013. doi: 10.1016/j.pneurobio.2013.08.002. [DOI] [PubMed] [Google Scholar]
  19. Knudson IM, Shera CA, Melcher JR. Increased contralateral suppression of otoacoustic emissions indicates a hyperresponsive medial olivocochlear system in humans with tinnitus and hyperacusis. J Neurophysiol 112: 3197–3208, 2014. doi: 10.1152/jn.00576.2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Lichtenhan JT, Wilson US, Hancock KE, Guinan JJ Jr. Medial olivocochlear efferent reflex inhibition of human cochlear nerve responses. Hear Res 333: 216–224, 2016. doi: 10.1016/j.heares.2015.09.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Lucker JR, Doman A. Neural mechanisms involved in hypersensitive hearing: helping children with ASD who are overly sensitive to sounds. Autism Res Treat 2015: 369035, 2015. doi: 10.1155/2015/369035. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Marshall L, Lapsley Miller JA, Guinan JJ, Shera CA, Reed CM, Perez ZD, Delhorne LA, Boege P. Otoacoustic-emission-based medial-olivocochlear reflex assays for humans. J Acoust Soc Am 136: 2697–2713, 2014. doi: 10.1121/1.4896745. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Pienkowski M, Tyler RS, Roncancio ER, Jun HJ, Brozoski T, Dauman N, Coelho CB, Andersson G, Keiner AJ, Cacace AT, Martin N, Moore BCJ. A review of hyperacusis and future directions: part II. Measurement, mechanisms, and treatment. Am J Audiol 23: 420–436, 2014. doi: 10.1044/2014_AJA-13-0037. [DOI] [PubMed] [Google Scholar]
  24. Rimland B, Edelson SM. Brief report: a pilot study of auditory integration training in autism. J Autism Dev Disord 25: 61–70, 1995. doi: 10.1007/BF02178168. [DOI] [PubMed] [Google Scholar]
  25. Robertson AE, Simmons DR. The relationship between sensory sensitivity and autistic traits in the general population. J Autism Dev Disord 43: 775–784, 2013. doi: 10.1007/s10803-012-1608-7. [DOI] [PubMed] [Google Scholar]
  26. Rosenhall U, Nordin V, Sandström M, Ahlsén G, Gillberg C. Autism and hearing loss. J Autism Dev Disord 29: 349–357, 1999. doi: 10.1023/A:1023022709710. [DOI] [PubMed] [Google Scholar]
  27. Rubenstein JL, Merzenich MM. Model of autism: increased ratio of excitation/inhibition in key neural systems. Genes Brain Behav 2: 255–267, 2003. doi: 10.1034/j.1601-183X.2003.00037.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Tharpe AM, Bess FH, Sladen DP, Schissel H, Couch S, Schery T. Auditory characteristics of children with autism. Ear Hear 27: 430–441, 2006. doi: 10.1097/01.aud.0000224981.60575.d8. [DOI] [PubMed] [Google Scholar]
  29. Tyler RS, Pienkowski M, Roncancio ER, Jun HJ, Brozoski T, Dauman N, Coelho CB, Andersson G, Keiner AJ, Cacace AT, Martin N, Moore BCJ. A review of hyperacusis and future directions: part I. Definitions and manifestations. Am J Audiol 23: 402–419, 2014. doi: 10.1044/2014_AJA-14-0010. [DOI] [PubMed] [Google Scholar]
  30. World Health Organization The World Health Report 2001: Mental Health: New Understanding, New Hope. Geneva: World Health Organization, 2001. [Google Scholar]
  31. Zaugg TL, Thielman EJ, Griest S, Henry JA. Subjective reports of trouble tolerating sound in daily life versus Loudness discomfort levels. Am J Audiol 25: 359–363, 2016. doi: 10.1044/2016_AJA-15-0034. [DOI] [PubMed] [Google Scholar]

Articles from Journal of Neurophysiology are provided here courtesy of American Physiological Society

RESOURCES