Keywords
hearing loss - CAEP - auditory gating - current density reconstruction - sLORETA -
inhibition
Sensory gating is a measure used to study degraded central inhibitory function underlying
various disorders, including autism spectrum disorder, schizophrenia, and tinnitus.[1]
[2]
[3]
[4] The role of inhibition in gating function is to filter out nonnovel input, leaving
adequate resources for the brain to process relevant information.[5] This process appears to be regulated by endogenous mechanisms through the action
of nicotinic receptors (nAChR) on inhibitory interneurons in the hippocampus, as well
as distributed nAChR activity in temporoparietal, temporofrontal, and prefrontal cortical
networks.[6]
[7]
[8] In certain disorders, these underlying inhibitory systems may be deficient, allowing
for abnormal sensory perceptions to occur.[5]
Through auditory gating, inhibition is evaluated using cortical auditory evoked potentials
(CAEPs) recorded via electroencephalography (EEG) in response to repeated pairs of
identical acoustic stimuli (S1 and S2).[5] Normal gating is observed through amplitude suppression of the CAEP S2 response
in relation to CAEP S1 amplitude, as the second stimulus in the pair is deemed nonnovel.[5] Inhibition can then be quantified through the peak amplitude ratio (P50 S2/P50 S1)
and difference indices (P50 S1–P50 S2). P50 ratio values less than 0.500 and difference
values greater than 0 µV are reflective of normal suppression,[9] and act as a biomarker of deficient central inhibition.[5]
While endogenous mechanisms play a significant role in gating processes, exogenous
factors should be considered as well. For instance, peripheral auditory deafferentation,
or sensorineural hearing loss (HL), has been found to adversely impact central inhibitory
mechanisms through a reduction of inhibitory inputs and subsequent imbalance between
excitatory and inhibitory systems.[10]
[11]
[12] Indeed, CAEP studies in humans have described increased amplitude of peak components
to correlate with measures of HL, specifically via the CAEP P2 component, which may
arise from decreased neural inhibition.[13]
[14]
[15]
[16]
[17] Thus, it may be hypothesized that HL would alter aspects of the gating response,
which is composed of CAEP amplitude indices.
However, to our knowledge, the specific effects of HL on sensory gating remain unknown,
as studies routinely exclude participants with auditory deficits.[18] Because this measure is used to study inhibitory deficits in clinical populations,[1]
[4]
[18] including recent research targeting tinnitus,[2]
[3] it is necessary to understand the possible effects of HL on the gating response.
Therefore, we examined gating performance in adults with normal hearing (NH) and mild–moderate
high-frequency (HF) sensorineural HL. CAEPs were recorded via high-density EEG in
response to tonal pairs. Peak amplitude and latency were compared within groups, while
the peak amplitude ratio and difference indices were compared between groups. It was
hypothesized that decreased gating function would be observed in HL, and would correlate
with auditory thresholds. The second goal of the study was to assess group differences
in cortical gating networks via current density reconstructions (CDRs). We hypothesized
that the HL group would present with incomplete activation of those temporal, frontal,
and prefrontal gating networks that have been identified in individuals with NH.[6]
[7]
Methods
Participants
This study was approved by the University of Texas Institutional Review Board. Twenty-two
adults provided informed written consent and were grouped according to pure-tone thresholds
above 25 dB HL. A requirement for study inclusion was that no participant report tinnitus
or score above 0 on the Tinnitus Handicap Inventory.[3]
[19] There was no significant age difference between NH (n = 11, mean age/standard deviation = 47.546 ± 7.967 years) or HL groups (n = 11, mean age/standard deviation = 56.273 ± 13.871 years) [F(1, 20) = 3.274, p = 0.085]. Nine females and one male were included in the NH group, while four females
and seven males made up the HL group. Seventeen participants denied smoking, while
two NH and three HL participants did not answer this question.[6]
[7] One NH participant reported an unspecified psychological diagnosis, one reported
migraine, and one HL participant reported migraine.
Audiometry and Speech Perception
Audiometric thresholds were measured bilaterally at 0.250 to 8 kHz via 3M E-A-R TONE
GOLD 3A insert earphones. The criterion for HL was 25 dB HL[15] ([Fig. 1]). Bone conduction thresholds were obtained to verify type of HL. Pure-tone averages
(PTAs) were calculated at 0.500, 1, and 2 kHz, and HF PTAs at 4 and 8 kHz. No participant
reported history of using an amplification device.
Fig. 1 Audiometric thresholds. Pure-tone thresholds for NH (black, n = 11) and HL groups (red, n = 11). The criterion for hearing loss was 25 dB HL (horizontal black line). Error bars represent one standard deviation. HL, hearing loss; NH, normal hearing.
Signal-to-noise ratio (SNR) loss was assessed using the QuickSIN Speech-in-Noise Test
(Etymotic Research) to determine a relationship between gating indices and auditory
performance in a degraded condition. Recorded sentences were delivered at a level
of 70 dB HL through a speaker placed at 0° azimuth, while background noise varied
from 25 to 0 dB SNR. Individual SNR loss was calculated from the average of two lists.
Auditory Gating Paradigm
Participants were fit with a 128-channel electrode net (Electrical Geodesics, Inc.)
and underwent testing in a sound booth. The EEG sampling rate was 1 kHz, with a band-pass
filter set at 0.100 to 200 Hz. Ocular electrodes were utilized for offline rejection
of myogenic activity.
Stimuli consisted of 0.25 kHz tonal pairs, a frequency chosen to ensure audibility
regardless of hearing status at higher frequencies, with a duration of 50 milliseconds
and 10 milliseconds-linear rise/fall times. The interstimulus interval was 500 milliseconds
and the intertrial interval was 7 seconds. Seven hundred tonal pairs were presented
at a level of 50 dB HL through two speakers placed at ± 45° azimuth while participants
watched a muted movie with subtitles. Please see Campbell et al[2]
[3] for additional details regarding the paradigm.
EEG Analyses
A 1-Hz high-pass filtered was applied and event epochs were created with −100 milliseconds
prestimulus and 350 milliseconds poststimulus periods. Data were exported to EEGLAB[20] and baseline-corrected according to the prestimulus period. Channels with amplitude
greater than a standard deviation value of ± 3 µV were deleted, followed by artifact
rejection using the same criterion. Deleted channels were interpolated using a spherical
interpolation algorithm. A total of 529.909 sweeps were included in the NH CAEP S1
average, and 525.546 sweeps in the CAEP S2 average. There were 460.546 sweeps accepted
in the HL CAEP S1 average, and 465.909 sweeps in the CAEP S2 average. There was no
significant difference between groups for the number of sweeps in the CAEP S1 average
[F(1, 20) = 3.586, p = 0.073] or the CAEP S2 average [F(1, 20) = 3.008, p = 0.098].
Thirteen electrodes were averaged to create a frontal region of interest (3, 4, 5,
9 [Fp2], 10, 11 [Fz], 12, 15, 16, 18, 19, 22 [Fp1], 23).[2]
[3] Baseline to peak amplitude and latency of the CAEP response components were compared
statistically within each group and marked as follows: P50 45–85 milliseconds, N1
90–140 milliseconds, and P2 140–220 milliseconds. Peak amplitude ratio and difference
gating indices were compared between groups.
Statistical Analyses
Multiple comparisons were corrected for using the Benjamini–Hochberg procedure, with
a false discovery rate of 0.1.[21] Within-group and between-group differences were tested using analysis of variance
(ANOVA). A one-tailed Pearson correlation was calculated to assess the relationship
between PTA (worse ear), HF PTA (worse ear), SNR loss, and gating indices across participants.
Current Density Reconstruction
Underlying components accounting for the greatest percent variance in CAEP peaks in
response to S1 and S2 stimuli were identified using independent component analysis.[22] Retained components for P50, N1, and P2 S1 and S2 peaks were averaged by participant
group in CURRY Scan 7 Neuroimaging Suite (Compumedics Neuroscan).[15] Grand difference averages were created for each peak by subtraction of S2 from S1
waveforms, and CDR was performed on peak difference averages using standardized low-resolution
brain electromagnetic tomography (sLORETA).[23] Group head models were generated using the boundary element method,[24] and sLORETA results plotted using an F-statistic distributed color scale on an average
magnetic resonance image (MRI; [Fig. 4]).
Results
Audiometry and Speech Perception
Eleven participants had thresholds at or above 25 dB HL at two or more frequencies
in either ear, and were categorized as the HL group. One NH participant presented
with a mild conductive component at 0.250 Hz in the right ear, but remained in the
NH group as speech perception in noise performance was within normal limits. Two HL
participants responded with a threshold above 25 dB HL at only one frequency, but
were retained in the HL group due to speech perception in noise performance outside
of normal limits (mild SNR loss). One individual in the HL group showed significant
interaural asymmetry, or a difference of greater than 15 dB between ears at two or
more frequencies. On average, HL participants showed mild–moderate HF HL ([Fig. 1]), with significantly worse PTA (worse ear) (F(1, 20) = 4.611, p < 0.050), and HF PTA (worse ear) thresholds than NH participants (F(1, 20) = 45.444, p < 0.001). No significant difference in SNR loss was found [F(1, 20) = 0.823, p = 0.375].
Auditory Gating in Hearing Loss
Within-group comparisons of P50, N1, and P2 amplitudes and latencies showed no significant
differences between CAEP S1 and S2 waveforms for the NH or HL groups ([Fig. 2A, B]). This finding is typically indicative of abnormal inhibitory processes,[5] but may be due to the mean age (47.546 years) of the NH group. Age has been reported
to negatively impact gating function,[25] and we have previously observed typical gating function in a young adult cohort
(20–22 years) using this paradigm.[2]
[3] However, there was a visual trend for amplitude suppression in NH participants for
the P2 peak, as well as for a higher amplitude P50 S2 response and similar amplitude
response between P2 S1 and S2 peaks in the HL group.
Fig. 2 Auditory gating in normal hearing (NH) and hearing loss (HL). (A) NH (n = 11) CAEP responses to a 0.250-kHz tonal pair. The vertical axis represents amplitude
in µV and the horizontal axis milliseconds. (B) HL (n = 11) CAEP responses to a 0.250-kHz tonal pair. (C) NH and HL CAEP comparison. Mean bar graphs illustrate differences in group P2 amplitude
gating indices. Error bars illustrate one standard deviation, and one asterisk indicates
significance at p < 0.050. CAEP, Cortical auditory evoked potential.
P50, N1, and P2 amplitude ratio and difference indices were calculated to quantify
inhibitory function. Only the P2 amplitude difference was significantly decreased
in the HL group [F(1, 20) = 5.383, p < 0.050], indicative of inhibitory deficits. The P2 amplitude ratio approached significance
for abnormal gating [F(1, 20) = 3.811, p = 0.065]. Both results support the hypothesis that central inhibition may be reduced
in adults with HL.
Auditory Gating Correlations
A significant negative correlation was observed between PTA and P2 amplitude gating
difference (r = −0.412, p < 0.050). A similar relationship was found between HF PTA and P2 amplitude gating
difference (r = −0.417, p < 0.050; [Fig. 3A]), as well as a significant positive correlation with P2 amplitude gating ratio (r = 0.379, p < 0.050; [Fig. 3B]). These results demonstrate that as auditory thresholds increased, there was an
associated decrease in inhibitory processes reflected by the P2 component, consistent
with the hypothesis that HL may coincide with deficits in gating function.
Fig. 3 Threshold correlations with auditory gating. (A) A negative correlation between P2 amplitude gating differences and HF PTA (n = 22). The Pearson correlation coefficient and level of significance are located
in the right corner. P2 amplitude difference values are shown on the horizontal axis
in µV and HF PTA on the vertical axis in dB HL. (B) A positive correlation between P2 amplitude gating ratios and HF PTA. HF, high frequency;
HL, hearing loss; PTA, pure-tone average.
Current Density Reconstruction in Auditory Gating
[Fig. 4] shows CDR results for P50, N1, and P2 difference averages in both groups, with a
table listing cortical regions in approximate order from highest activation. Sources
of the NH P50 gating component included right superior temporal gyrus (STG) and middle
temporal gyrus (MTG). Lateralization of this response was likely due to the use of
tonal stimuli.[26] In addition, right frontal and prefrontal regions such as inferior frontal gyrus
(IFG), middle frontal gyrus (MFG), superior frontal gyrus (SFG), and Brodmann area
(BA) 6 showed activation. These results are comparable with typical gating function
in NH listeners.[6]
[7]
[27] In the HL group, IFG, MFG, and SFG responses were observed, although activation
shifted to include mainly temporal regions. Furthermore, an absence of activation
in prefrontal cortex was apparent, with left supramarginal and angular gyri responsive
instead. A similar lack of prefrontal activation underlying gating deficits has been
reported in clinical populations.[28]
Fig. 4 Auditory gating current density reconstructions. (A) NH and HL CDR for P50, N1, and P2 difference (S1–S2) averages on sagittal MRI slices.
The F-distribution scale is located in the right corner, with yellow being the highest
level of activation, and MNI (Montreal Neurological Institute) coordinates are listed
below each MRI slice. (B) Activated cortical areas, listed in approximate order of highest level of activation.
CDR, Current density reconstruction; HL, hearing loss; MRI, magnetic resonance imaging;
NH, normal hearing.
Cortical activation for the NH N1 gating component included temporal and frontal regions,
as well as precentral gyrus ([Fig. 4]), congruent with normal N1 gating function.[25] The HL group showed a shift of activation to temporal areas, as well as lateralization
of the temporal response to the left hemisphere. This lateralization during N1 activity
is consistent with atypical inhibitory function in elderly participants.[25]
NH P2 gating activity involved the right temporal regions of STG, MTG, and ITG ([Fig. 4]), and left frontal areas (MFG, SFG). In contrast, the HL group presented with only
right temporal activation. A similar reduction in left hemispheric activation underlying
the P2 S2 component has been reported in individuals at risk for inhibitory-related
deficits.[29] This result is also consistent with the finding of reduced P2 amplitude gating indices
in HL ([Fig. 2C]).
Discussion
We examined auditory gating function in adults with mild–moderate HL versus NH controls
to determine whether HL may adversely affect sensory inhibitory mechanisms. Three
findings are described: (1) significantly reduced P2 amplitude gating difference indices,
consistent with atypical inhibitory function in HL, (2) an association between decreased
P2 gating and increased auditory thresholds, and (3) a possible compensatory temporoparietal
gating network in HL.
It should be noted that typical gating function was not observed in the NH group,
possibly due to the age of the cohort. In previous studies using the same paradigm,[2]
[3] young, NH adults demonstrated expected gating function consistent with the literature.
However, despite an apparent lack of gating in the current NH group, P50 gating generators
typically present in individuals with normal gating performance were successfully
localized.[6]
[7]
[27] This finding suggests that central inhibitory function may have been normal in the
NH adults, albeit unobserved via the CAEP measure. Future research should examine
the effect of extended age ranges using the described gating protocol, particularly
as age has been reported to affect gating performance.[25]
While decreased inhibition was identified in HL, it is important to mention that it
is unlikely this result is due to reduced audibility.[30] First, a 0.250-kHz tonal pair was presented at a suprathreshold level (50 dB HL)
within a frequency region where most subjects demonstrated normal auditory thresholds.
Furthermore, a reduction of CAEP amplitude in conjunction with reduced audibility
has been reported in the literature,[31] while we observed either similar or larger CAEP S1 and S2 amplitudes in the HL group
versus the NH group ([Fig. 2C]). Rather, this finding is consistent with the hypothesis that there is a corresponding
decrement in central inhibition associated with HL, possibly due to a reduction in
lateral inhibition resulting from auditory deprivation in adjacent frequency bands.[11]
In addition, we found only the P2 gating component to specifically reflect inhibitory
deficits related to HL. Although the P50 gating component is considered a biomarker
of inhibition,[5] both N1 and P2 gating components have been reported to be sensitive to inhibitory
deficits.[32] Furthermore, several CAEP studies have identified the P2 component to be representative
of cortical plasticity in HL. P2 amplitude appears to increase in conjunction with
elevated thresholds, leading to the hypothesis that larger amplitude may arise from
decreased inhibition associated with HL and subsequent “effortful listening.”[13]
[14]
[15]
[17] Thus, the P2 gating component is likely representative of plasticity in cortical
inhibitory function related to acquired HL.
In support of this hypothesis, CDR results revealed a qualitative difference between
inhibitory networks in NH and HL adults ([Fig. 4]). NH gating function was localized in temporal, frontal, and prefrontal cortical
regions, including activation of BA 6, consistent with previous gating studies.[6]
[7]
[27] In contrast, HL gating function was decreased in frontal and prefrontal regions,
with a shift to temporal and parietal areas. Similar gating activity in parietal regions
was observed in typical subjects,[6]
[27] suggesting that those with HL may use an incomplete or secondary gating network
in comparison with NH adults. Along these lines, there was a reduction in frontal
activation underlying the HL P2 gating response. This specific finding could be a
result of resource reallocation in HL. For instance, Campbell and Sharma[15] showed that HL adults reallocate passive processing of auditory input from temporal
to frontal cortex, indicative of recruitment of cognitive resources for the processing
of degraded lower order auditory information. Thus, in HL, frontal cortical resources
required for typical gating function may not be available, leading to the development
of a compensatory temporoparietal gating network.
Study Limitations
While HL appeared to be related to decreased inhibition in the present study, it will
be important in future research to tightly control for and examine the effects of
specific types of HL, including configuration, laterality, and severity, on the CAEP
amplitude gating response. For instance, the participants included in the HL group
demonstrated a range of hearing thresholds across varying frequencies, which may have
precluded findings of atypical inhibition specific to patterns of HL. Along these
lines, it will also be necessary to better understand possible age effects and interactions
with HL on CAEP amplitude gating indices. The age range included in this study was
broad, and may have acted as a confounding variable in observing a typical gating
response for the NH group. However, post-hoc analyses showed no significant correlation
between CAEP amplitude gating indices and age across groups, suggesting that group
differences in gating function were indeed due to HL rather than age effects. In any
case, additional research is required to better understand and appropriately implement
auditory gating as a clinical measure for central disorders such as tinnitus,[2]
[3] especially in the presence of HL.
Conclusion
This study examined the effects of mild–moderate sensorineural HL on auditory gating,
an inhibitory measure used in clinical populations, to better understand the impact
of reduced auditory input on sensory inhibitory mechanisms. Using CAEPs recorded via
high-density EEG in response to tonal pairs, we found decreased inhibition in HL reflected
by P2 amplitude gating indices, which correlated with auditory thresholds. In NH adults,
the gating response was localized in temporal, frontal, and prefrontal cortices, while
adults with HL showed mainly temporal and parietal responses. Overall, our results
suggest decreased inhibition underlying auditory gating in HL. These outcomes should
be considered when implementing auditory gating as a clinical measure, especially
in disorders such as tinnitus.
