Introduction Tinnitus is characterised by the perception of sounds (e.g., a tone, hissing, or roaring noise, and sometimes combinations of such perceptions) in the absence of any objective physical sound source. For the affected individual, this condition often causes a considerable amount of distress. Although it is now widely accepted that the generation of tinnitus has a central basis [1–3], the neurophysiological mechanism is still not well understood. Typically, tinnitus is accompanied by an audiometrically measurable hearing loss, and the considerable overlap between hearing loss and tinnitus spectra [4] suggests that these phenomena may be related. Thus, a widespread assumption is that neuronal response properties in the auditory system change following damage to it. These changes may persist even after recovery from the peripheral lesion. However, there is no general agreement as to which changes are responsible for the perception of this auditory phantom phenomenon. Studies in humans and animals indicate that neurons in the deprived regions of the auditory cortex change both their receptive field [5–8] and their spontaneous activity [1,9,10]. Furthermore, recent positron emission tomography studies in particular have pointed to the brain areas involved in attentional and emotional regulation [11–14]. An example of such a neuroimaging study was conducted by Lockwood et al. [14], who investigated individuals with tinnitus who were able to enhance or reduce the perceived loudness of their tinnitus via oral–facial movements. Besides changes in auditory cortical activity contralateral to the affected ear, the authors reported changes in hippocampal activity related to loudness changes. The authors interpreted this as evidence for limbic system involvement. In addition to this study implicating the limbic system (and temporal) structures, studies by Mirz et al. [11,12] indicate that frontal areas have a role in tinnitus. Surprisingly, even though spontaneous activity has been a frequent research target in animal models of tinnitus, studies in humans have been rare. Data from a group study for abnormal activity of the cochlear nerve were reported by Martin [15]. An abnormal peak in spectral power could be observed close to 200 Hz in a majority of participants undergoing cerebellopontine angle surgery, a majority of whom had tinnitus (in the present study, however, we mean cortical activity when speaking of spontaneous activity). When extracted from spontaneous magnetoencephalography (MEG)–produced images in perisylvian regions, the values of the largest Lyapunov exponent—a measure for the predictability of the time series—were generally larger for the participants with tinnitus than for control individuals, indicating different nonlinear temporal dynamics in spontaneous activity for tinnitus and control participants [16]. A further study aiming to use information offered by spontaneous ongoing brain rhythms was published by Shulman and Goldstein, who reported temporal and frontotemporal changes (increases and decreases) of relative power and coherence irregularities in severely disabled individuals with tinnitus [17]. Otherwise, attempts to study tinnitus in humans have focussed on the use of designs that measure neurophysiologic responses following sounds [8,18–23] or experimental manipulations that enhance or reduce the perceived loudness [13,14,24,25]. In the present study, we examined the power spectrum of neuromagnetic oscillatory activity during rest. The generation of tinnitus, in most cases, can be linked to damage to the auditory system, usually to receptors of the inner ear [1–3], probably even in cases where an impairment cannot be assessed audiometrically [26]. In this sense, one can speak of the presence of a deafferentation in the system since certain areas of the brain are now deprived from their normal input. Various sources of evidence indicate that deprivation of primary input leads to the functioning of the system in a slow-wave modus, i.e., analysis of neuroelectric signals in the frequency domain reflect an enhanced power in the delta frequency range ( 5, p 0.4 for all). Amount and slope of hearing loss were also not significantly correlated with neuromagnetic data (0.12 < r < 0.25). Discussion To our knowledge, this is the first group study on tinnitus in humans (there have been reports on single cases [35,36]) to show marked relative enhancements in delta and concomitant reductions in alpha spontaneous cortical activity. Notable group differences are a bilateral relative enhancement in delta and an accompanying reduction in alpha power over temporal areas (extending into posterior regions in the right hemisphere) in individuals with tinnitus. These findings are in agreement with a study conducted by Shulman and Goldstein [17], who report abnormalities in spontaneous activity in all 21 investigated patients. The definition of abnormality, however, was rather broad, meaning enhancements, reductions, or coherence irregularities in different bands. Another difference from our study is that these authors investigated severely disabled patients, whereas only a minority (four out of 17; see Table 1) of our participants could be classified in a similar manner. Our results are in agreement with ideas put forward by Jeanmonod and Llinas and colleagues [35,36], who stated that positive symptoms are a consequence of a hyperpolarisation of thalamic neurons following deafferentation, leading to spike bursts around 4 Hz. A look at the power spectra in Figure 1 shows that a reduced alpha cannot be caused by normalisation, because in this case the peak is almost absent—a pattern we encounter frequently in tinnitus participants. Concerning delta, it is very unlikely that the enhancement relative to alpha is an effect of the reduced alpha, as enhancements should then not be restricted to slow oscillations but should also be seen in other frequency bands (e.g., beta; note, e.g., that the artefactual 16-Hz activation is identical for both groups). Apart from this, the general pattern (i.e., alpha reduction and delta enhancement) is also seen in the non-normalised data (available upon request). It is unclear whether this relative enhancement of delta activity compared to alpha is the “abnormal activity” that is perceived as tinnitus [44]. The underlying neural substrate of tinnitus perception could also be an edge effect, such as gamma activity that arises in regions between abnormal (delta) and normal awake activity [45]. The (methodological) problem in the latter case lies, however, in the exact definition of gamma activity. However, it would be interesting to investigate the effects of treatments presumably inducing tinnitus on slow waves in animals. The fact that regions with an increase in slow-wave activity are also the regions of decreased alpha activity resembles results found during slow-wave sleep [32] and supports the idea that the changes in spontaneous brain activity might be mediated by sensory deprivation, i.e., partial hearing loss in this case. Additional evidence is provided by the observation that delta enhancement and alpha reduction were strongly correlated with tinnitus-related distress variables with a focus on the right temporal and also left frontal cortex. The association of the abnormality index—which sets alpha and delta power in relationship to one another—and tinnitus-related distress demonstrates that the effects reported are not an epiphenomenon of normalisation. Furthermore, a deviant abnormality index can be seen at an almost individual level. We are thus confident that the main effect reported here is not due to methodological flaws concerning data analysis. However, we can not exclude with certainty the possibility that the effects are not specific to tinnitus, a limitation of our study. One interesting finding is that compared to the normal hearing control group, the tinnitus group also had a high-frequency hearing loss. So, theoretically, the ideal control group would have exactly the same type of hearing loss without a tinnitus sensation. However, to find such a group of an appropriate sample size would be very difficult as both phenomena are strongly associated. For example, in an attempt to investigate the influence of high-frequency hearing loss (similar to that of our study) on cortical reorganisation, Dietrich et al. [8] noticed that all hearing-impaired participants also had tinnitus. Yet it cannot be disputed that occasionally individuals exist who have a similar kind of deficiency in clinical audiograms as the participants in our study, but without having a tinnitus sensation. Simply having a reduced threshold does not seem to be sufficient to trigger reorganisational processes leading to tinnitus. Why these individuals (although rare) do not develop this symptom is a very important question for understanding tinnitus. A possible way to tackle this question is to analyse underlying hearing damage in more detail (e.g., amount of inner and outer hair-cell damage) by comparing individuals with hearing loss with and without tinnitus. Another strategy would be to investigate possible predisposing factors, e.g., psychological or genetic factors. As parameters of hearing loss (depth and slope) appear to be uncorrelated with distress scores and neurophysiological data, we assume its distorting effects to be minimal. This hypothesis is further supported by the fact that only a small proportion of the distress score was attributed to hearing loss. Overall, it should be emphasised that the issue of hearing loss is less serious in our study than in studies in which individuals with tinnitus (either human or animals) are acoustically stimulated (especially in the hearing-loss range). The association of our neurophysiologic data with distress suggests that the right temporal and left frontal cortex might be involved in a tinnitus-related cortical network, in which the temporal region is associated more with perceptual issues (i.e., aspects concerning the character of the sound, e.g., tonal or noise-like, loudness), and the left frontal region more with affective distress and motivational attention of tinnitus (i.e., the tinnitus becoming a signal of high importance, so that it draws attention of the individual). Without reference to lateralisation, Jastreboff describes the prefrontal cortex as a “candidate for the integration of sensory and emotional aspects of tinnitus” [44]. Our data lend support to this notion. Concerning the stronger effect for the right temporal area than for the left, one has to consider the higher frequency of left-sided tinnitus in our study. Thus, it cannot be excluded that this asymmetry would vanish if more individuals with right-sided tinnitus were included in the analysis. However, the fact that tinnitus is generally more common for the left ear [46] certainly opens up possibilities to speculate about potential asymmetries, either on a peripheral or central level, that could account for this finding. The underlying physiological reasons for this asymmetry on an epidemiological level have not been a matter of research so far. Left frontal activation has been linked with positive, and right frontal activation with negative, affect [47–49]. In the context of this framework, enhanced alpha (indicating hypoactivation) in the left frontal cortex should be indicative of depression. Our alpha–distress association, however, is negative, thus implying that the results obtained cannot be explained by an effect of higher depressive mood in individuals with tinnitus. This simple explanation would also not fit with recent data from our group [52], which demonstrated a negative association between left frontal delta (measured via dipole density) and depression. In future studies, we hope to elucidate the function of the (left) frontal area, as it may point us to the role of top-down influences that presumably play a role in the perception and perhaps even generation of tinnitus. An important aspect of these results is that they potentially have clinical implications, which are currently being tested by our group. Reducing the abnormal spontaneous activity pattern reported in this study via neurofeedback might cause concomitant reductions in tinnitus distress and/or intensity (often reported to be unrelated; [51]). Besides experimentally demonstrating to what extent our results reflect a “genuine” signature of tinnitus sensation in ongoing brain activity, these investigations could ultimately be of benefit for persons affected by this condition. Supporting Information Data Acquisition and Signal Analysis Five minutes of MEG under a resting condition was recorded (sampling rate: 678.17 Hz; 0.1–200 Hz band-pass) using a 148-channel neuromagnetometer (Magnes 2500 WH, Biomagnetic Technologies, San Diego, California, United States). The participant was requested to keep eyes open and to maintain gaze on a fixation mark at the ceiling of the recording chamber. Eye movements were monitored with four electrodes attached to the left and right outer canthi and above and below the right eye. In the first step of data analysis, sampling points were reduced by a factor of ten. After noise reduction, eye-movement correction was undertaken using the algorithm proposed by Berg and Scherg [52]. As an explorative step, spectral power was calculated for each sensor via mean fast Fourier transformations (window length, 7.55 s; 50% overlap between cosine squared windows). This step was taken to focus on specific frequency bands of interest. As the emphasis of the study was on alterations of spontaneous activity patterns within the tinnitus group, data were scaled by dividing each value by the overall mean power (gained by averaging over all sensors). As can be seen in Figure 1, participants with tinnitus showed a markedly reduced alpha peak accompanied by an enhancement in the slow frequency (delta) range. The second step consisted of the investigation of the underlying source activity via application of the minimum norm estimate [53,54] to the eye-movement-corrected continuous data. This linear estimation technique yields a solution for the current density of a configuration of dipoles (here: 197 evenly distributed) located on a spherical shell by multiplying the pseudo-inverse of the lead-field matrix (which describes the sensitivity of each sensor to the sources) with the obtained data. The lead-field matrix was computed for each participant, based on information about the centre of a fitted sphere to the digitised head shape, and the positions of the MEG sensors relative to the head. In order to be able to average the obtained minimum norm estimate solutions over different participants, we employed a shell with a fixed radius of 6 cm. The radius of 6 cm was chosen as a tradeoff between depth sensitivity and spatial resolution [53]. Each of the 197 dipole locations consisted of two perpendicular dipoles oriented tangentially to the shell surface. The source-space transformed continuous data were then entered into the same mean fast Fourier transformation algorithm as described above. For both orientations of each dipole, power was calculated in the following frequency bands: delta (1.5–4 Hz), theta (4–8 Hz), and alpha (8–12 Hz). In order to gain a single value for each dipole, the square root of the sum of squares of the power for the two orientations was calculated and scaled in the same manner as described above (using mean power of each dipole instead of each sensor). Oscillatory activity was then analysed (a) averaged over all 197 dipoles and (b) in specific regions by grouping clusters of dipoles on the sphere and averaging their values. These regions (always bilateral) were prefrontal, frontal, temporal, frontocentral, parietal (anterior and posterior), and occipital cortex. Patient Summary Background Tinnitus—hearing sounds such as hissing or roaring that are not really present—is a common condition and can be very distressing. Little is know about why tinnitus starts. Though people who have it often have some hearing loss, the hearing loss does not cause the tinnitus. There are few good treatments for tinnitus. What Did the Authors Do? They measured spontaneous activity in the brains of 17 people with tinnitus and 16 without. The found that one type of brain activity—alpha—was decreased and another—delta—was increased in people with tinnitus, particularly in one part of the brain, the temporal region, where the brain processes sounds. The changes were particularly pronounced in people who were most distressed by the tinnitus. What Do These Findings Mean? It seems that tinnitus is associated with one type of abnormal spontaneous activity in the brain. One way of treating tinnitus therefore might be to try to alter the activity by providing feedback about the ongoing brain activity and training the individual to interfere with this activity, i.e., by attempting to enhance alpha activity and reduce delta activity. Where Can I Get More Information? MedlinePlus has a Web page on tinnitus: http://www.nlm.nih.gov/medlineplus/tinnitus.html The Welcome Gateways has a selection of Web sites on tinnitus: http://omni.ac.uk/browse/mesh/D014012.html The American Tinnitus Association also offers information on tinnitus: http://www.ata.org
