1. Introduction
Subjective tinnitus describes the perception of a tonal and more complex sound
without the presence of an external source. Therefore, it is often referred to as
a
phantom noise, in analogy to phantom pain. A description as a circumscribed
phenomenon of the auditory pathway or auditory system therefore seems obvious.
Tinnitus can occur as a result of any injury to the auditory pathway, such as
hearing loss or presbycusis. Likewise, injury to the auditory nerve, such as from
a
vestibular schwannoma, can cause tinnitus. These impairments lead to changes in
cortical activity and ultimately to perception of tinnitus. However, the correlation
of lesion or hearing loss and tinnitus is complex. Hearing loss does not necessarily
lead to tinnitus, and not every patient with chronic tinnitus has an abnormal
hearing threshold. Therefore, it has been assumed that any combination of an
alteration in auditory or somatosenory input together with altered central nervous
activity or structures may produce tinnitus [1]. Based on neurophysiological studies, it has been suggested that, as a
consequence of altered activity, tinnitus is associated with reorganization of
tonotopic maps [2]. Tonotopic organization is
a hallmark of the auditory system in mammals, originating in the cochlea and
continuing to the neocortex in humans [3]
[4].
Already this brief overview shows that tinnitus is a very heterogeneous disorder.
In
addition to the above-mentioned impairment of the auditory pathway (see also the
current S3 guideline [5]), chronic tinnitus
(persisting for more than 3 months) is accompanied by significant cognitive and
affective disorders in most cases. In the cognitive domain, abnormalities are most
notable in attention [6]
[7]
[8],
executive functions, and memory functions. These impairments of central processes
can also be the cause of disorders of language comprehension. A recent study
revealed that the perception of linguistic signals is impaired, but not the
perception of speakers or voices [9]. Thus,
the impairment of perception cannot be explained by a too weak or unclear
presentation of the auditory signal per se.
Besides these abnormalities of cognition, affective disorders in particular are
observed in tinnitus (see Mazurek et al.; further presentation for the DGHNO meeting
in 2023). A recent study compared four groups with regard to psychopathological
impairments: decompensated tinnitus patients, compensated tinnitus patients,
patients with major depression without tinnitus, and unimpaired control subjects
[10]. Evaluated questionnaires on anxiety,
depressive and psychosomatic symptoms were collected from all groups. The four
groups could be classified using a canonical discriminant analysis based on two
factors. Factor 1 was termed “general psychopathology” because most
questionnaires responded strongly to this factor. With regard to this factor,
patients with decompensated tinnitus and patients with major depression were equally
and more impaired than patients with compensated tinnitus, while the latter were
also significantly more impaired than healthy controls. Both tinnitus groups
(compensated and decompensated) scored higher than the other two groups on factor
2,
“somatization”. Consistent with previous trials, this study could
demonstrate the strong psychopathological burden in compensated, but especially in
decompensated tinnitus. By the quantitative approach, the increased burden could
also be revealed in compensated tinnitus patients, even if they were not clinically
conspicuous in the actual sense.
In summary, it can already be seen from these examples that tinnitus is an extremely
heterogeneous disorder that affects different physical, emotional, and cognitive
domains. To briefly anticipate what is to come, current neurophysiological findings
point in the same direction. These findings were obtained by means of various
methods, their strengths and weaknesses will be described in the following
section.
2. Neurological, neurophysiological, and neurocognitive examination
methods
Neurophysiological imaging methods can basically be described on two dimensions,
temporal and spatial. Current examination procedures show their strengths and
weaknesses on these dimensions. Magnetic resonance imaging (MRI), as a method
for detecting structural and functional properties of the brain, has very high
spatial accuracy and allows localization of structures and corresponding
activation in the cubic millimeter range at typical magnetic field strength to
1.5–3 Tesla. MRI is one of the most commonly used noninvasive methods,
but it is expensive and requires specially trained staff. Using voxel-based
morphometry (VBM), the entire brain can be depicted in volumetric pixels
(voxels), allowing morphometric differences to be mapped individually or across
groups and quantified for statistical analysis. The necessary structural images
in the MRI scanner usually take only a few minutes. Diffusion tensor imaging
(DTI) methods can also be performed based on such structural images. This allows
diffusion movements of water molecules in the brain to be imaged and quantified.
Frequently, DTI is used to determine the course, strength, and effectiveness of
large nerve fiber bundles. Commonly used measures include fractional anisotropy
(FA, the directionality of white fiber matter) and axial diffusivity (AD, the
strength of diffusion in fiber direction) and radial diffusivity (RD, the
strength of diffusivity perpendicular to the principal direction). These
measures can be used to detect the integrity or affectedness of axons and their
myelination.
In addition to structural measurements, MRI also allows the determination of
functional neuronal activity. This is based on the “blood oxygenation
level dependent” (BOLD) effect, with which the regional distribution of
highly oxygenated blood in the brain, and thus the activity, can be measured.
While the spatial accuracy is also very high, the temporal resolution is in the
range of several seconds. Measurements in MRI scanners involve considerable
noise exposure, which can be quite stressful for tinnitus patients.
Electroencephalography and magnet encephalography (EEG and MEG) are methods that
record the synchronized activity of larger cell clusters directly and completely
non-invasively. The spatial accuracy for determining the origin of the signal
can be measured less precisely than with MRI but increases with the number of
sensors used. The localization of activity is based on mathematical modeling
techniques that incorporate, for example, head shape, electrical conductivity
properties in the head, and the underlying number or neuronal sources. In
contrast to the rather limited spatial resolution capability, the temporal
resolution is in the range of milliseconds. In particular, EEG is a very widely
used technique that is inexpensive and requires few staff.
The aforementioned methods for measuring brain activity allow the recording of
spontaneous activity, i. e., without external stimuli and without the
subjects being involved in a task to be performed (so-called “resting
state” activity), as well as the recording of evoked activity. Evoked
activity measures the brain’s responses to a sensory stimulus that may
be associated with the performance of a task (e. g. key press).
3. Structural changes in tinnitus
Structural changes have been reported in tinnitus not only in auditory and
sensory, but also in limbic areas [11]
[12]. However, different
areas with increased or decreased gray matter volume (GMV) were found in
different studies. GMV in Heschl’s gyrus and superior temporal gyrus
showed changes in both directions in tinnitus patients compared to healthy
controls [13]
[14]
[15]
[16], while inferior
colliculus volume (Landgrebe et al., 2009) was decreased and medial geniculus
volume (Muhlau et al., 2006) was increased in these patients. As for the
non-auditory limbic brain structures, decreased GMV was reported in the
ventromedial (vmPCF) and dorsomedial (dmPFC) prefrontal cortex, nucleus
accumbens, anterior (ACC), posterior cingulate cortex, hippocampus, and
supramarginal gyrus [15]
[17]
[18]
[19]. However, the changes
in some of these areas were modulated by hearing loss [13]
[14]
[20]. In some studies,
regional volumes have been associated with depressive and affective comorbidity,
such as insula, cerebellum, and ACC [19]
[21].
A recent study investigated the issue of psychiatric comorbidity in relation to
structural changes in the brain caused by tinnitus (see Besteher et al., 2019,
parallel to Ivansic et al., 2019). While hypothesis-based (region of
interest-based) analysis of brain areas involved in tinnitus (specifically
parahippocampal cortex, ACC, and superior temporal/transverse temporal
cortex) showed main effects of tinnitus in parahippocampal cortex and
trend-level findings in ACC and superior/transverse temporal cortices,
several other findings at the whole-brain level (e. g. in the precuneus)
did not survive a more conservative correction for multiple statistical
comparisons. A reduction in parahippocampal matter was also found in a
comparative analysis of tinnitus patients without psychiatric comorbidity
compared with health controls, supporting the findings described above. This
means that even if patients do not meet the diagnostic criteria for a
psychiatric disorder, they are still more distressed than healthy controls. The
most important finding of this study concerning the brain structural
associations of this condition in tinnitus patients appears to be the particular
importance of limbic structures (anterior and posterior cingulate gyri and
parahippocampal gyri).
A recent study [23] investigated structural
changes after auditory training and compared patients who improved with those
who did not benefit from training. Compared to patients with improvement,
patients without benefit had a significant decrease in gray matter volume in the
right middle frontal gyrus (MFG) as well as the right precentral gyrus (PreCG).
Thus, improvement in auditory perception is particularly associated with changes
in frontal rather than typically auditory structures. A very recent
meta-analysis compared groups of patients with and without tinnitus in whom
hearing loss was or was not measurable [24]. The authors reported a minor reduction in gray matter in the
left inferior temporal gyrus in normal-hearing tinnitus patients compared with
groups of hearing subjects without tinnitus. In contrast, tinnitus was
associated with increased gray matter in the bilateral lingual gyrus and
bilateral precuneus in the groups with hearing loss. This study suggests that
changes in gray matter in individuals with and without tinnitus are driven by
hearing loss.
4. Changes of the white fiber matter in tinnitus
A first DTI study found reduced FA in the right prefrontal area, the left
inferior and superior longitudinal fasciculus, and the anterior thalamic
radiation in tinnitus patients compared to healthy control subjects [25]. In contrast, other authors showed
increased FA in similar regions such as the inferior longitudinal fasciculus and
anterior thalamic radiation [26], as well
as in auditory and limbic areas [27].
Other DTI studies dealt with the correlation between perceived loudness of
tinnitus and measures of white matter integrity. A positive correlation between
perceived loudness and FA and a negative correlation between loudness and RD and
AD were reported in the anterior thalamic radiation and ventromedial prefrontal
cortex [19]
[28]. An increase in RD at the level of the
lateral lemniscus and inferior colliculus indicated the presence of
demyelination processes [29]. This study
emphasized that hearing loss but not tinnitus per se was associated with white
matter changes (Lin et al., 2008). An increase in AD in the left superior,
middle, and inferior temporal white matter similarly indicated axonal
degeneration in tinnitus patients compared with control subjects [30]. Overall, recent DTI trials emphasize
that the differences between tinnitus patients and controls are less due to the
tinnitus itself, but can be mainly explained by age and an altered hearing
threshold of the patients [31]. These
factors, as well as experienced tinnitus distress and existing comorbidities,
may at least partially explain the above contradictory results and should be
captured in future studies to elucidate the anatomical heterogeneity of tinnitus
patients (Schmidt et al., 2018). A recently published trial with a very tight
control of these aspects failed to detect changes in white matter that correlate
with perceived loudness of tinnitus. However, the severity of hearing loss and
tinnitus distress did correlate with changes in acoustic radiation and arcuate
fasciculus [32].
5. Functional changes in tinnitus
Magnetic resonance imaging (MRI) studies of tinnitus patients have also been used
to depict functional changes (fMRI) that allow determination of dysfunctional
network activity. Compared to all other methods, fMRI is the most commonly used
imaging technique. Also due to the short duration and simplicity of the
examination, “resting-state fMRI” (rs-fMRI) represents a very
commonly used method in which the subject is not involved in a task. These
rs-fMRI scans allowed the identification of several so-called resting-state
networks across studies populations, analysis methods, and recording protocols:
sensorimotor network, auditory network, limbic network, visual and extrastriate
visual network, insular-temporal/anterior cingulate salience network
(ACC), left and right lateralized frontoparietal network (attention), dorsal and
ventral attention networks, default mode network (DMN), and a network for
frontal executive functions [33]. Based on
six available studies, an earlier review [34] came to the conclusion that the limbic DMN and auditory-limbic
functional connectivity in particular are increased in resting state of
tinnitus. A very recent paper [35]
provides a review of 29 studies, 26 of which report abnormalities in tinnitus
patients compared to controls in resting state networks. They include the
auditory network (19 studies), DMN (17 studies), visual network (14 studies),
dorsal (7 studies) and ventral (1 study) attention network, the executive
function network (9 studies) and the limbic system (8 studies). The authors
emphasized that the findings depended on the regions of interest (ROIs),
i. e. whether or not a priori specific regions were included in the
analyses. The authors therefore suggested that future studies should prioritize
replicability of outcomes. In the peer-reviewed studies, strong heterogeneity
was again evident and confounding variables were not adequately controlled.
Nonetheless, the overall results suggest that multiple overlapping networks are
involved in tinnitus. However, it remains unclear which changes are primarily
due to tinnitus and which may be secondary to tinnitus.
Based on these studies, another model, so-called “triple network
model” suggests that the default mode network and the executive function
network are anti-correlated in acute tinnitus. In chronic tinnitus, this
anti-correlation disappears, and a dysfunctional triple network emerges
consisting of the DSM, executive, and salience networks underlying
tinnitus-associated distress and cognitive executive impairments [36].
A recent study investigated tonotopic changes in tinnitus and, in this context,
did not evaluate resting-state activity but brain activity in response to
sinusoidal tones with frequencies between 0.25 and 8 kHz [37]. Subjects with and without tinnitus,
but all with bilateral hearing loss, and a control group were evaluated.
Activity in bilateral regions of the auditory cortex was higher in the groups
with hearing loss compared to the control group. This was most evident in the
group without tinnitus. Similarly, the tonotopic maps of the group with hearing
loss without tinnitus did not differ from the controls. These results indicate
that higher activation and reorganization of tonotopic maps are a feature of
hearing loss and not of tinnitus.
6. Evidence by electro and magnetic encephalography for presentation of
dysfunctional network activity in tinnitus
Complementary to resting-state in tinnitus patients using fMRI, EEG and MEG
studies achieved comparable conclusions. An MEG study with source localization
investigated extended cortical connections in tinnitus patients compared to
controls [38]. The study looked for
so-called “hubs” within ramified networks connecting different
regions. The prefrontal cortex, orbitofrontal cortex, and parieto-occipital
regions were shown to be central structures in the tinnitus network, and the
information flow toward the temporal cortex correlated with the severity of
tinnitus distress. Changes in functional connectivity were demonstrated in an
EEG study [39] that investigated the role
of stress in the perception of tinnitus. This emphasized the role of the
parahippocampus as the node of a network that additionally included the
posterior and anterior cingulate cortex, insula, and auditory cortex regions. In
another EEG trial, a reorganization of the entire tinnitus network was observed,
reflected by a decrease in the strength and efficiency of information transfer
between fronto-limbic and medial temporal regions [40]. These regions also represented the
main nodes of the tinnitus network. Parts of this network, and specifically the
connections from the left hippocampus/parahippocampus to the subgenual
anterior cingulate cortex, were strongly correlated with tinnitus distress.
An MEG study yielded very similar results but could show, using more complex
connectivity measures that the role of the left parahippocampal area is
modulated by the dorsomedial prefrontal cortex, a region typically attributed to
the dorsal attention network and involved in the regulation of emotional
processing [41]. In addition, this method
of analysis across the entire cortex provided new insight into the role of the
left inferior parietal cortex, which modulated the activity of the right
superior temporal gyrus (see [Fig.
1]).
Fig. 1 Significantly higher connectivity in tinnitus patients
compared to control subjects. The arrows represent connections between
areas with greater connectivity in tinnitus patients, with the thickness
of the arrows representing the strength of connectivity. These results
show that there is a cluster of connections in the dorsal prefrontal
cortex, left medial cortex, parahippocampal regions, left inferior
temporal gyrus, lateral occipital gyrus, and right intraparietal lobe
(for details [41]).
In addition to these studies based on resting-state measurements, a number of
trials have been published that focused on brain activity as an evoked response
to auditory stimulation. Repeatedly, a tinnitus-related increase in neuronal
excitability was reported, reflected for example in the amplitude of the
auditory N1 component. In this context, this early auditory evoked component is
often used to objectively assess stimulus-associated EEG/MEG signals or
as a biomarker to indicate typical and atypical cortical development (for a
review, see Tomé et al., 2015; Foxe et al., 2011). Tinnitus patients had
higher N1 amplitude in response to a frequency-specific tone outside the area of
hearing loss, typically 500 Hz or 1 kHz, compared to healthy
controls (Pantev, 1989; Hoke et al., 1998; Weisz et al., 2005), or even in
response to tinnitus frequency tones (Kadner et al., 2002; Pineda et al., 2008).
In contrast, however, other authors demonstrate significantly smaller N1
amplitudes in tinnitus patients compared to normal-hearing controls [45]
[46] or showed no response to a 1-kHz tone [47]
[48]. The inconsistency of these results may be due to the relatively
small sample sizes (<30 subjects) and different methodological
strategies, such as different and/or varying number of a priori defined
ROIs, or by constraints on a small number of sources in brain activity modeling.
When source modeling methods were used that allowed for a large number of
possible sources, an interaction of temporal, frontal, and parietal regions was
reported within the N1 time window [49]
[50].
7. Conclusion and outlook
Tinnitus is a common symptom with a prevalence in Europe of about 15%, with
1–2% of the population suffering from severe tinnitus [51]. The aging of societies and the associated
hearing loss are expected to lead to a further increase in prevalence. The resulting
healthcare expenses are already substantial [52]
[53] and will continue to
increase. In particular, tinnitus of high severity is characterized by high
comorbidity, which can manifest in diverse physical, emotional, and cognitive
symptoms. Already for this reason, it is obvious that tinnitus cannot be
conceptualized as a narrowly defined purely auditory phenomenon. The results using
the neurophysiological methods described here allow very similar conclusions to be
drawn. Regardless of the methods used, the analysis procedures, the heterogeneity
of
the samples, the involvement of subjects in tasks, it was shown that tinnitus is
characterized by complex dysfunctional network activity. In addition to auditory
temporal regions, parietal, frontal, and especially limbic systems are particularly
frequently affected. Whereas earlier analysis methods could only detect simultaneous
activation, newer methods allow establishing correlations between specific
activation patterns and regions, so that the direction of connectivity can be
determined. It has been emphasized several times in this review article that in
order to understand chronic tinnitus, a number of confounding variables must be
controlled, including especially hearing loss and comorbid symptomatology. It would
be desirable for future studies to focus on replication, an appropriate number of
subjects, and comprehensive interdisciplinary diagnostics [54].
This review article was intended to provide a synopsis of the most common
non-invasive procedures. Those techniques have in common the ability to capture the
structural and functional properties of the brain over as large an area as possible.
Near-infrared spectroscopy (NIRS) is another non-invasive technique that measures
electromagnetic radiation in the infrared range. This method is used to determine
blood volume and blood flow as well as oxygen content of various tissues such as the
brain. Similar to EEG and MEG, the temporal resolution is very high, and the spatial
resolution is rather moderate (again depending on the number of sensors applied).
Compared to the above methods, NIRS is new and has been little used, especially in
research on chronic tinnitus. Currently, the number of sensors used is relatively
small, so that a priori decisions must be made as to which regions’ activity
will be recorded and evaluated. Initial studies indicate that cortical changes in
tinnitus are not limited to auditory regions [55], but also affect emotion-relevant regions [56].
To determine the causal role of specific areas and networks on the development of
chronic tinnitus, longitudinal studies and especially studies on the transition from
an acute to a chronic state in patients are desirable. Currently, these studies are
lacking, which is also a major gap in the literature from a clinical perspective.
A
combination of neurophysiological procedures would also be desirable, e. g.
to exploit the advantages of different procedures. This would allow conclusions
about spatial and temporal aspects of brain activation. Another gap in the
literature is longitudinal studies in the course of therapeutic interventions. Since
in particular cognitive behavioral therapy is recommended according to the S3
guideline for the treatment of chronic tinnitus, it seems obvious to investigate the
neurophysiological changes in the course of this therapeutic procedure.
