The neural network of phantom sound changes over time

14 Pages·2011·1.11 MB·English

European Journal of Neuroscience
EuropeanJournal of Neuroscience, pp. 1–14,2011 doi:10.1111/j.1460-9568.2011.07793.x
The neural network of phantom sound changes over time: a
comparison between recent-onset and chronic tinnitus
patients
Sven Vanneste,1 Paul van de Heyning2 and Dirk De Ridder1
1TRITinnitusClinic,Brai2n,TRI&DepartmentofNeurosurgery,UniversityHospitalAntwerp,Wilrijkstraat10,2650Edegem,Belgium
2Brai2n,TRI &ENT,University Hospital Antwerp,Edegem,Belgium
Keywords:network, sLORETA,time,tinnitus
Abstract
Tinnitus is characterized by an ongoing conscious perception of a sound in the absence of any external sound source. Chronic
tinnitusisnotoriouslycharacterizedbyitsresistancetotreatment.Inthepresentstudytheobjectivewastoverifywhethertheneural
generators and⁄or the neural tinnitus network, evaluated through EEG recordings, change over time as previously suggested by
MEG. We therefore analyzed the source-localized EEG recordings of a very homogenous group of left-sided narrow-band noise
tinnitus patients. Results indicate that the generators involved in tinnitus of recent onset seem to change over time with increased
activityin severalbrainareas [auditorycortex,supplementarymotorareaand dorsalanterior cingulatecortex (dACC) plusinsula],
associated with a decrease in connectivity between the different auditory and nonauditory brain structures. An exception to this
generalconnectivitydecreaseisanincreaseingamma-bandconnectivitybetweentheleftprimaryandsecondaryauditorycortexand
the left insula, and also between the auditory cortices and the right dorsal lateral prefrontal cortex. These networks are both
connectedtotheleftparahippocampalarea.Thusacuteandchronictinnitusarerelatedtodifferentialactivityandconnectivityina
network comprising theauditory cortices, insula,dACC and premotorcortex.
Introduction
Tinnitus is characterized by an ongoing conscious perception of a At least two different hypotheses can be proposed to explain why
sound in the absence of any physical sound source. Hence it is tinnitusismoredifficulttosuppressifthetinnitusdurationincreases.
considered an auditory phantom phenomenon similar to central Firstly, it is possible that the neural generators change over time, i.e.
neuropathicpain(Tonndorf,1987;Moller,2000).Tinnitusisperceived new areas become involved in tinnitus generation and the auditory
continuouslyby(cid:2)5–15%oftheadultpopulation.Theneurobiological cortex becomes less involved over time. Another possibility can be
basisoftinnitusischaracterizedbyanongoingabnormalspontaneous proposed in which the neural connectivity within the same network
activity and reorganization of the auditory central nervous system changes over time. In a recent magnetoencephalographic (MEG)
(Weisz etal., 2005, 2007a,b; Moazami-Goudarzi etal., 2010), but connectivity study with selected ROIs it was shown that in patients
nonauditory brain structures such as the dorsal prefrontal cortex, with recent tinnitus onset (i.e. <4years) the gamma network
anterior cingulate cortex, and the (para)hippocampus (Mirz etal., connectionsareconcentratedinthelefttemporalarea(auditorycortex)
2000; De Ridderetal., 2006; Muhlau etal., 2006; Landgrebe etal., whileinchronictinnituspatients(i.e.>4years)thenetworkiswidely
2009)arealsoinvolvedintinnitus,andhowtheseareasco-activatehas distributedovertheentirecortex,withlessinvolvementoftheauditory
recentlybeeninvestigated(Schleeetal.,2009). temporalareas(Schleeetal.,2009).However,thetwopropositionsare
Several studies have shown that focal stimulation of the temporal notmutuallyexclusive.Itispossiblethatboththeneuralgeneratorand
cortex by repetitive transcranial magnetic stimulation (rTMS) (De neuralconnectivitychangewithincreasingtinnitusduration.
Ridderetal.,2005b;Kleinjungetal.,2007;Plewniaetal.,2007a)can In the present research the objectives were to verify whether new
suppresstinnitusperception.However,theamountofmaximaltinnitus neural generators arise and⁄or whether the neural tinnitus network
suppression by rTMS decreases with time (De Ridder etal., 2005b; changes over time. We used the source-localized electroencephalo-
Plewniaetal.,2007a;Kleinjunget al.,2007;Khedretal.,2008). graphic (EEG) recordings of a homogenous group of unilateral left-
sided narrow-band noise tinnitus patients and analyzed the spectral
componentsrelatedtothetinnitusdurationaswellastheconnectivity
Correspondence:SvenVanneste,asabove. withinagroupofselectedregionsofinterest(ROIs).TheROIswere
E-mail:[email protected] chosen based on areas involved in tinnitus according to the positron
Received20February2011,revised22May2011,accepted13June2011 emission tomography (PET), single-photon emission computed
ª2011TheAuthors.EuropeanJournalofNeuroscienceª2011FederationofEuropeanNeuroscienceSocietiesandBlackwellPublishingLtd
2 S. Vanneste et al.
tomography (SPECT), magnetoencephalography (MEG), EEG and Table1. Patients’characteristics
voxel-basedmorphometry (VBM) tinnitus literature.
Duration
Recentonset Chronic P-value
Materials and methods
Patients Sex
Male 5 4 –
Eighteenpatients(N=18;ninemalesandninefemales)withstrictly
Female 5 4
unilateral left-sided narrow-band noise tinnitus with a mean age of
60.82yearswereselectedfromthemultidisciplinaryTinnitusResearch Age
M 59.78 62.12 0.74
Initiative(TRI)ClinicoftheUniversityHospitalofAntwerp,Belgium.
SD 10.62 12.43
NoneofthepatientsincludedinthisstudyhadbeentreatedwithrTMS
priortothestudy.Individualswithpulsatiletinnitus,Me´nie`re’sdisease, Tinnitusduration
M 1.72 8.00 0.001
otosclerosis,chronicheadacheandneurologicaldisorderssuchasbrain
SD 0.62 1.85
tumors, and individuals being treated for mental disorders, were
excludedfromthestudyinordertoobtainaveryhomogeneoussample. VASintensity
M 6.43 7.43 0.21
Assuch,10patientswithtinnitusofrecentonset(<4 years)andeight
SD 1.40 1.27
patientswithchronictinnitus(>4but<10years)wereselected.The4-
year cutoff was based on the available tinnitus literature. Results of TQ
M 53.00 53.80 0.79
microvascular decompression (Moller et al., 1993; Brookes, 1996;
SD 10.73 9.86
Jannetta,1997;Ryuetal.,1998;DeRidderetal.,2004,2005a,2007,
2010)andtranscranialmagneticstimulation(DeRidderetal.,2005b; Tinnitusfrequency(Hz)
M 4937.50 4720.83 0.85
Plewnia etal., 2007a; Kleinjung et al., 2007; Khedr etal., 2008)
SD 2771.14 2777.08
suggestthattinnitusresultsworsenovertime,withaturningpointof
efficacy between 3 and 5years. This fits with a magnetoencephalo- Hearingloss*(dBHL)
M 26.83 35.00 0.28
graphicstudythatdemonstratesthatafter4yearsthetinnitus-generating
SD 20.55 15.37
networkchangesitsconnectivity(Schleeetal.,2009).
All patients were investigated for the extent of hearing loss using *MeanHLatthetinnitusfrequency
audiograms. Tinnitus matching was performed looking for tinnitus
pitch(frequency)andtinnitusintensity.Participantswererequestedto collected with eyes closed (sampling rate 1024Hz, bandpass-filtered
refrain from alcohol consumption 24h prior to recording, and from 0.15–200Hz).Datawereresampledto128Hz,bandpass-filtered(fast
caffeinated beverage consumption on the day of recording. Each Fouriertransformfilter)to2–44Hzandsubsequentlytransposedinto
patient’s subjective tinnitus loudness perception was obtained on a Eureka! Software (Congedo, 2002), plotted and carefully inspected
visual analogue scale (VAS) from 0 to 10 and a validated Dutch withmanualrejectionofartifacts.Allepisodicartifactsincludingeye
translationofthetinnitusquestionnaire(TQ;Meeuset al.,2007)was blinks, eye movements, teeth clenching, body movement and ECG
used to asses tinnitus related distress. A Mann–Whitney U-test was artifacts wereremovedfrom thestream of theEEG.AverageFourier
conductedtoverifydifferencesbetweenbothgroupsforage,tinnitus cross-spectral matrices were computed for bands delta (2–3.5Hz),
duration,VAS,TQ,andHearingloss.Nosignificantdifferenceswere theta (4–7.5Hz), alpha1 (8–10Hz), alpha2 (10–12Hz), beta1 (13–
foundbetweenrecent-onsetandchronictinnituspatientsforage,VAS 18Hz), beta2 (18.5–21Hz), beta3 (21.5–30 Hz) and gamma (30.5–
or TQ. Gender was equally balanced. See Table1 for overview. No 44Hz).
significantdifferenceswerefoundforhearingloss,asmeasuredbythe
lossin decibels (dBSPL) atthe tinnitus frequency.
Also, an age- and gender-matched control group (N=12; Source localization
mean±SD, 61. 32±10.91years; six males and six females) was Standardizedlow-resolutionbrainelectromagnetictomography(sLO-
collected. None of these subjects was known to suffer from tinnitus. RETA) was used to estimate the intracerebral electrical sources that
Exclusion criteria were known psychiatric or neurological illness, generated the scalp-recorded activity in each of the eight frequency
psychiatric history, abuse of drugs or alcohol, history of head injury bands(Pascual-Marqui,2002).sLORETAcomputeselectricneuronal
(withlossofconsciousness),seizures,headache,orphysicaldisability activity as current density (A⁄m2) without assuming a predefined
inaparticipant. Toour knowledgethe tinnitus patientsdidnotdiffer
number of active sources. The sLORETA solution space consists of
fromthe control groupin anyaspect excepttheir tinnitus.
6239 voxels (voxel size 5·5·5mm) and is restricted to cortical
gray matter and hippocampi, as defined by digitized Montreal
Neurological Institute (MNI)152 template (Fuchs et al., 2002). Scalp
EEGdatacollection electrode coordinates on the MNI brain are derived from the
EEGrecordings(Mitsar-201;NovaTechhttp://www.novatecheeg.com/) international 5% system(Jurcak etal., 2007).
wereobtainedinaquietanddimlylightedroomwitheachparticipant The tomography sLORETA has received considerable validation
sitting upright on a small but comfortable chair. This EEG system is from studies combining LORETAwith other more established local-
usedbyseveral research groupsover the world (Sherlin etal., 2007; izationmethods,suchasfunctionalmagneticresonanceimaging(fMRI;
Muelleretal.,2010;Phlypo&Congedo,2010).Theactualrecording Vitaccoet al.,2002;Mulertetal.,2004),structuralMRI(Worrelletal.,
lasted(cid:2)5min.TheEEGwassampled with19electrodes (Fp1,Fp2, 2000) andPET(Dierks etal., 2000; Pizzagalli etal.,2004; Zumsteg
F7,F3,Fz,F4,F8,T7,C3,Cz,C4,T8,P7,P3,Pz,P4,P8,O1O2)in etal.,2005),andhasbeenusedinpreviousstudiestodetectforexample
thestandard10–20Internationalplacement,referencedtolinkedears, activity in the auditory cortex (Zaehle etal., 2007; Vanneste etal.,
and impedances were checked as remaining below 5kX. Data were 2011). Further sLORETAvalidation has been based on accepting as
ª2011TheAuthors.EuropeanJournalofNeuroscienceª2011FederationofEuropeanNeuroscienceSocietiesandBlackwellPublishingLtd
EuropeanJournalofNeuroscience,1–14
Theneural network of phantomsoundchanges overtime 3
groundtruththelocalizationfindingsobtainedfrominvasive,implanted Table2. Regionsofinterest
depthelectrodes,ofwhichthereareseveralstudiesinepilepsy(Zumsteg
et al., 2006a,c) and cognitive event-related potentials (Volpe etal., Authors
2007).Itisworthemphasizingthatdeepstructuressuchastheanterior
Amygdala DeRidderetal.(2006)and
cingulate cortex (Pizzagalli et al., 2001) and mesial temporal lobes
Landgrebeetal.(2009)
(Zumsteget al.,2006b)canbecorrectlylocalizedwiththesemethods.
InthecurrentimplementationofsLORETA,computationsweremadein Anteriorcingulatecortex
Dorsal Plewniaetal.(2007b),Vannesteetal.(2010)
arealisticheadmodel(Fuchsetal.,2002),usingtheMNI152template
andSchleeetal.(2009)
(Mazziotta et al., 2001), with the three-dimensional solution space Subgenual Muhlauetal.(2006)and
restricted to cortical gray matter, as determined by the probabilistic Vannesteetal.(2010)
Talairachatlas(Lancasteretal.,2000).Thestandardelectrodepositions
Auditorycortex Muhlnickeletal.(1998),Schneideretal.(2009),
on the MNI152 scalp were taken from (Jurcak et al., 2007) and Smitsetal.(2007)andWeiszetal.(2007a,b)
(Oostenveld & Praamstra, 2001). The intracerebral volume is parti- Dorsallateralprefrontal Mirzetal.(2000)
tioned into 6239 voxels at 5-mm spatial resolution. Thus, sLORETA cortex
images represent the standardized electrical activity at each voxel in Insula Smitsetal.(2007)andVannesteetal.(2010)
neuroanatomic MNI space as the exact magnitude of the estimated Parahippocampus Landgrebeetal.(2009)
currentdensity.AnatomicallabelsasBrodmannareasarealsoreported
usingMNIspace,withcorrectiontoTalairachspace(Brettet al.,2002).
the left and right secondary auditory cortex (BA21 and BA22), and
werepredefinedbasedontheMNIcoordinates,belongingtoaspecific
Brodmannareaonastandardbrain.ROIanalysiswerecomputedfor
Functional connectivity
the different frequency bandsseparately.
Brainconnectivitycanrefertoapatternofanatomicallinks(anatomical Amultivariateanova(i.e.Wilks’Lambda)wasusedforthedifferent
connectivity),ofstatisticaldependencies(functionalconnectivity)orof frequency bands with the respective ROIs [i.e. left and right primary
causalinteractions(effectiveconnectivity)betweendistinctunitswithin auditorycortex(BA40andBA41)andleftandrightsecondaryauditory
anervoussystem.Thepresentresearchfocusesonfunctionalconnec- cortex(BA21andBA22)]asdependentvariablesanddifferentgroups
tivitywhichcapturesdeviationsfromstatisticalindependencebetween (recent-onset,chronicandcontrolsubjects)asindependentvariables.A
distributed and often spatially remote neuronal units. Statistical Bonferronicorrectionwasappliedformultiplecomparisons.
dependencemaybeestimatedbymeasuringcorrelationorcovariance,
spectral coherence or phase-locking. Functional connectivity is often
calculatedbetweenallelementsofasystem,regardlessofwhetherthese Statistical analyses
elements are connected by direct structural links. Unlike structural Themethodologyusedisnonparametric.Itisbasedonestimating,via
connectivity,functionalconnectivityishighlytime-dependent.Statis- randomization, the empirical probability distribution for the max-
tical patterns between neuronal elements fluctuate on multiple time statistic, under the null hypothesis comparisons (Nichols & Holmes,
scales,someasshortastensorhundredsofmilliseconds.Itshouldbe 2002). This methodology corrects for multiple testing (i.e. for the
notedthatfunctionalconnectivitydoesnotmakeanyexplicitreference collectionoftestsperformedforallvoxels,andforallfrequencybands).
tospecificdirectionaleffectsortoanunderlyingstructuralmodel. Duetothenonparametricnatureofthemethod,itsvaliditydoesnotrely
Coherence and phase synchronization between time-series corre- onanyassumptionofGaussianity(Nichols&Holmes,2002).
sponding to different spatial locations are usually interpreted as sLORETAstatisticalcontrastmapswerecalculatedthroughmultiple
indicatorsoftheconnectivity.However,anymeasureofdependenceis voxel-by-voxelcomparisonsinalogarithmoftheF-ratio.Thesignif-
highly contaminated with an instantaneous, nonphysiological contri- icancethresholdwasbasedonapermutationtestwith5000permuta-
butionduetovolumeconduction(Pascual-Marqui,2007b).However, tions.Acomparisonwasmadebetweenthetinnitusgroup(recent-onset
Pascual-Marqui (2007a) introduced a new technique (i.e. Hermitian andchronictinnitusgroup)andcontrolsubjectsaswellasbetweenthe
covariance matrices) that removes this confounding factor. As such, chronictinnitusgroupandthegroupwithtinnitusofrecentonset.
this measure of dependence can be applied to any number of brain Connectivitycontrastmapswerecalculatedthroughmultiplevoxel-
areas jointly, i.e. distributed cortical networks, whose activity can be by-voxel comparisons using t-statistics. The significance threshold
estimatedwithsLORETA.Measuresoflineardependence(coherence) was based on a permutation test with 5000 permutations. Again, a
between the multivariate time-series are defined. The measures are comparison was made between the tinnitus group (recent-onset and
expressed as the sum of lagged dependence and instantaneous chronictinnitusgroup)andcontrolsubjectsaswellasbetweenchronic
dependence. The measures are non-negative and take the value zero tinnitus andtinnitus with recent onset.
only when there is independence, and are defined in the frequency To further confirm our connectivity analysis we conducted a split-
domain:delta(2–3.5Hz),theta(4–7.5Hz),alpha1(8–10Hz),alpha2 halfreliabilityanalysis.Werandomlydividedindividualsforeachof
(10–12Hz), beta1 (13–18Hz), beta2 (18.5–21Hz), beta3 (21.5– the recent-onset and chronic groups into two groups and conducted
30Hz) and gamma (30.5–45Hz). Based on this principle lagged similar connectivity analyses between the recent-onset and chronic
linear connectivity was calculated. ROIs were defined based on groupfor both groups.
previous brain research ontinnitus (see Table2 for overview).
Results
ROI analysis Neural generators–tinnitus vs.control
Thelog-transformedelectricalcurrentdensitywasaveragedacrossall Acomparisonbetweentinnituspatientsandcontrolsubjectsrevealed
voxels belonging to the ROI. ROIs were the left and right primary significant differences for theta, beta2, beta3 and gamma frequency
auditorycortices[Brodmannarea(BA)40andBA41respectively]and bands. Increased activity could be found for tinnitus patients in,
ª2011TheAuthors.EuropeanJournalofNeuroscienceª2011FederationofEuropeanNeuroscienceSocietiesandBlackwellPublishingLtd
EuropeanJournalofNeuroscience,1–14
4 S. Vanneste et al.
respectively,thesupplementarymotorarea(BA6)fortheta,thedorsal Table3. Region of interest comparison between patient with recent onset
anterior cingulate (BA24) for beta2 and beta3 and the right tinnitus, chronic tinnitus and control subjects for the primary and secondary
parahippocampal area (BA37) for gamma (see Fig.1; P<0.05). No auditorycortexingammabandfrequency
significantdifferencescouldberetrievedforthedelta,alpha1,alpha2
Recentonset Chronic Control
orbeta1 frequencies.
primaryAC
left 1.77a 2.22a 1.33b
Neuralgenerators–chronictinnitusvs.tinnitusofrecentonset right 1.72a 2.49a 1.42b
The same frequencies as when comparing tinnitus patients with the secondaryAC
controlgroupshowedsignificantdifferencesbetweenchronictinnitus left 3.04a 3.24b 1.84c
andtinnituswithrecentonset.Increasedspontaneouselectricalactivity right 2.54a 3.55b 1.88a
couldbefoundforchronictinnituspatientsincomparisontotinnitus
a,b,cSignificantdifferencesbetweenrecentonsettinnitus,chronictinnitusand
patientswithrecentonset,inthesupplementarymotorarea(BA6)for
controlsubjects(Bonferronicorrectionformultiplecomparisons).
theta, the dorsal anterior cingulate (BA24 and BA32) for beta2 and
beta3,theinsula(BA13)forbeta3andtheauditorycortex(BA21and
frequencies (see Figs5 and 6; P<0.05). Decreased phase-lagged
BA22) for gamma (see Fig.2; P<0.05). No significant differences
connectivitycouldbefoundingeneralforthesethreefrequencybands
couldberetrievedindelta,alpha1,alpha2orbeta1frequencies.
forchronictinnituspatientsincomparisontotinnitusofrecentonset.
Furthermore,asignificantincreaseinphase-laggedconnectivitycould
ROIanalysis befoundforthegammabandbetweentheleftparahippocampalarea
(BA37),leftprimaryandnon-primaryauditorycortices(BA21,BA22,
Asignificanteffectwasfoundforthelog-transformedcurrentdensity
BA41 and BA42), the left insula (BA13) and the right dorsal lateral
for the different groups on the ROI for the gamma frequency band
prefrontal cortex (DLPFC; BA9 and BA46) for chronic tinnitus
(F =2.43, P<0.05; see Table3). Including all frequency bands
8,48 patients.Twoclosedconnectivityloopswerediscerned.Thefirstwas
nexttothegammafrequencybandinoneanalysisdidnotresultina
betweentheinsulaandBAs21,22,41and42oftheauditorycortex.
significanteffect.However,thissignificanteffectmightremainifthe
A second closed connectivity loop was found between DLPFC, and
samplesize increased.
BA21, BA22 and BA41, BA42 of the auditory cortex (See Fig.3,
Univariateanovafurtheryieldedsignificanteffectsforleftprimary
below panel).
auditorycortex(F =3.54,P<0.05),rightprimaryauditorycortex
2,27 The split-half reliability estimate showed similar results for both
(F =5.32,P<0.05),leftsecondaryauditorycortex(F =4.84,
2,27 2,27 theta (see Supporting Information Fig.S1) and gamma bands (see
P<0.05) and right secondary auditory cortex (F = 9.30,
2,27 Supporting Information Fig.S3) between the chronic group and the
P<0.001). A Bonferroni multiple comparison analysis (P<0.05)
acute tinnitus group. That is, decreased phase-lagged connectivity
revealed that the control subjects had significant lower log-averaged
could be found in general for these frequency bands for chronic
current density than either recent-onset tinnitus or chronic tinnitus
tinnituspatientsincomparisonwithtinnituspatientswithrecentonset.
patientsfortheleftandrightprimaryauditorycortex.Chronictinnitus
Foralpha2similarresultswerealsoobtainedalthoughwithadecrease
was significantly different from tinnitus with recent onset for the left
in phase-lagged connectivity for chronic tinnitus patients in compar-
andrightsecondaryauditorycortex.Nosignificanteffectwasobtained
isontotinnituspatientswithrecentonset(seeSupportingInformation
forthe other frequency bands(P>0.25).
Fig.S2).Somedifferenceswiththefrontalareaswerenotedbetween
the two split-half groups but these disappeared when the two groups
were combined. In addition, significant increases in functional
Neuralconnectivity–tinnitus vs. control
connectivity could be found for the gamma band between: (i) the
A comparison between tinnitus patients and control subjects demon- left-sided parahippocampal area and primary and non-primary audi-
stratedsignificantdifferencesfortheta,alpha2andgammafrequency torycortices,and(ii)theinsulatothetinnitussideandtheright-sided
bands(seeFigs3and4;P<0.05).Decreasedfunctionalconnectivity DLPFC for chronic tinnitus patients in both groups, with two closed
couldbefoundingeneralforthesethreefrequencybandsfortinnitus connectivityloops.Theselatterfindingsfurtherconfirmthereliability
patients in comparison to control subjects. Increased synchronized of our results.
activity could be found for tinnitus patients in comparison to control
subjects within the theta (4–7.5Hz) band between the left and right
parahippocampalarea(BA37)andinalpha2(10–12Hz)connectivity
between the left parahippocampal area (BA37) and the left auditory Discussion
cortex(BA21,BA22),andtherightparahippocampalarea(BA37)and Themainobjectiveofthisstudywastocharacterizethedifferencesin
the right auditory cortex (BA21, BA22; see Fig.3; P<0.05). In neural generators and the neural network between tinnitus of recent
addition,gamma(30–45Hz)connectivitywasincreasedbetweenthe onset and chronic tinnitus in a very homogenous group. When
left parahippocampal area (BA37) and the left and right auditory comparing the tinnitus group with a control group, we found
cortex (BA21, BA22) for tinnitus patients in comparison to control differences within the anterior cingulate, medial premotor cortex and
subjects (see Fig.4; P< 0.05). No significant differences could be parahippocampal area. Differences were demonstrated in auditory
retrieved indelta, alpha1,beta1, beta2 or beta3 frequencies. areas (primary auditory cortex) as well as in nonauditory areas
(anteriorcingulate,insulaandmedialpremotorcortex)forthechronic
tinnitusgroupincomparisontothetinnitusofrecentonsetgroup.ROI
Neuralconnectivity–chronictinnitusvs.tinnitusofrecentonset analysis further revealed that the control subjects had significantly
Connectivityanalysisyieldedasignificantdifferencebetweenchronic lowerlog-averagedcurrentdensityincomparisontobothrecent-onset
tinnitus and tinnitus with recent onset for theta, alpha2 and gamma tinnitus and chronic tinnitus patients for the left and right primary
ª2011TheAuthors.EuropeanJournalofNeuroscienceª2011FederationofEuropeanNeuroscienceSocietiesandBlackwellPublishingLtd
EuropeanJournalofNeuroscience,1–14
Theneural network of phantomsoundchanges overtime 5
A
B
C
D
Fig.1. sLORETAcontrastanalysisbetweentinnituspatientsandcontrolgroup(P<0.05).Increasedneuralsynchronizationwithin(A)theta(4–7.5Hz),(B)beta2
(18.5–21Hz),(C)beta3(21.5–30Hz),and(D)gamma(30.5–45Hz)in,respectively,(A)supplementarymotorarea(BA6),(B)rightdorsalanteriorcingulate(BA24
andBA32),(C)rightdorsalanteriorcingulate(BA24),and(D)parahippocampalarea(BA19andBA37)fortinnituspatients.L,left;R,right;A,anterior;P,posterior.
ª2011TheAuthors.EuropeanJournalofNeuroscienceª2011FederationofEuropeanNeuroscienceSocietiesandBlackwellPublishingLtd
EuropeanJournalofNeuroscience,1–14
6 S. Vanneste et al.
A
B
C
D
Fig.2. sLORETAcontrastanalysisbetweenchronictinnituspatientsandtinnituspatientswithrecentonset(P<0.05).Increasedneuralsynchronizationwithin(A)
theta(4–7.5Hz),(B)beta2(18.5–21Hz),(C)beta3(21.5–30Hz),and(D)gamma(30.5–45Hz)in,respectively,(A)rightlateralpremotorarea(BA6),(B)right
dorsalanteriorcingulate(BA24andBA32),(C)rightdorsalanteriorcingulate(BA24andBA32)andrightinsula(BA13),and(D)rightauditorycortex(BA21and
BA22)forchronictinnituspatients.L,left;R,right;A,anterior;P,posterior.
ª2011TheAuthors.EuropeanJournalofNeuroscienceª2011FederationofEuropeanNeuroscienceSocietiesandBlackwellPublishingLtd
EuropeanJournalofNeuroscience,1–14
Theneural network of phantomsoundchanges overtime 7
auditory cortex. Furthermore, chronic tinnitus patients had signifi- cortical dysrhythmia as the pathophysiological model for the devel-
cantly higher log-averaged current density in comparison to tinnitus opmentofgamma-bandactivityrelatedtothetinnituspercept(Llina´s
with recent onsetfor the left andrightsecondary auditory cortex. etal., 1999). According to this model tinnitus is caused by an
Differenceswerealsofoundintheneuralnetworkwithadecreased abnormal, spontaneous and constant gamma-band activity (>30Hz)
functionalconnectivitybetweenthedifferentROIsfortheta,alphaand generatedasaconsequenceofdeafferentation-inducedhyperpolariza-
gamma frequency band for tinnitus patients in comparison to the tionofthe medialgeniculate body.Innormal circumstancesauditory
control subjects as well as in the chronic tinnitus patients in stimuli increase thalamocortical rhythms to gamma-band activity
comparison to the tinnitus patients of recent onset. Increased phase- (Joliot etal., 1994). In the deafferented state, however, oscillatory
laggedconnectivitycould beseenfortinnitus patientsincomparison alpha activity decreases to theta-band activity (4–7Hz; Steriade,
to control subjects in the theta frequency band between the left and 2006).Asaresultlateralinhibitionisreduced,inducingasurrounding
right parahippocampal area and in alpha2 between the left parahip- gamma-bandactivityknownasthe‘edgeeffect’(Llina´setal.,2005).
pocampal area and the left auditory cortex, and the right parahippo- Thisabnormallypersistentcoupledtheta–gammabanddysrhythmiais
campal area and the right auditory cortex. Increased phase-lagged relayed to the cortex, selectively in the deafferented thalamocortical
connectivity was found in the gamma band between the left columns.Ithasbeenproposedthatsynchronizedgamma-bandactivity
parahippocampal area and the left and right auditory cortex for in the auditory cortex binds auditory events into one coherent
tinnituspatientsincomparisontocontrolsubjects.Increasedgamma- conscious auditory percept (Ribary etal., 1991; Llinas et al., 1998;
band connectivity was also found between the left parahippocampal Crone etal., 2001). The decrease in alpha activity and associated
area and primary auditory cortex, the left auditory cortex and right gamma-band activity has been shown to exist in tinnitus (Lorenz
auditorycortex,andtheleftauditorycortexandleftinsula,aswellas etal., 2009). However, why there was more activity in the right
between the left auditory cortices and the right DLPFC for chronic auditory cortex for chronic left-sided tinnitus patients it not yet
tinnitus in comparison to tinnitus of recent onset. established,butahypothesiscanbeproposed.Ourresultsrevealthat
there was increased gamma connectivity between the left and right
auditorycortexviatheleftparahippocampalareafortinnituspatients
in comparison to control subjects. ROI analysis further demonstrated
Auditory cortex,parahippocampal areaand supplementary
that both the left and right auditory cortex have a higher gamma
motor area
currentdensityinchronictinnitusincomparisontothecontrolgroup
Increased gamma activity was found for tinnitus patients in compar- and that the left auditory cortex for chronic tinnitus patients is
ison to control subjects in the parahippocampal area. This might be involved in closed loop with the insula as well as with the DLPFC.
related to the constant updating of the tinnitus percept, preventing The reason why only the right auditory cortex is hyperactive in the
habituation(DeRidderetal.,2006).Cellsinthehumanhippocampus gamma band and not the left auditory cortex in the global brain
andparahippocampalareasrespondtonovelstimuliwithanincrease analysismightberelatedtothefactthattinnitusintensityiscorrelated
in firing. However, even on the second presentation of a stimulus, to increased contralateral gamma-band activity (van der Loo etal.,
neurons in these regions show very different firing patterns. In the 2009),increasingtherightgamma-bandactivityevenmoreintheright
parahippocampal region there is dramatic decrease in the number of auditory cortex.
cells responding to the stimuli (Viskontas etal., 2006), suggesting a For tinnitus patients increased theta activity was also found in the
rapid habituation. This rate of response decrement during trains of supplementarymotorareaincomparisontoacontrolgroup,aswellas
several stimulus repetitions is linear for acoustic responses (Wilson intherightsupplementarymotorareaforpatientswithchronictinnitus
et al., 1984). In contrast to the rapid auditory habituation in the in comparison to patients with recent-onset tinnitus. The (visual)
parahippocampal area, in the hippocampus there is recruitment of a global workspace model has been suggested as a model for the
largesubsetofneuronsshowinginhibitoryresponses(Viskontasetal., awareness component of consciousness (Dehaene et al., 2006). It
2006). Thus a novel stimulus is normally associated with parahippo- proposesthatconsciousperceptionofsensoryeventsrequiressensory
campal habituation and active hippocampal inhibitory activity. cortex activation embedded in a larger cortical network, the global
Repetitive auditory stimuli both in animals (Bickford etal., 1993) workspace, extending beyond the primary sensory regions including
andhumans(Boutrosetal.,2008)leadtoattenuationofevent-related prefrontal, parietal and cingulate cortices (Dehaene etal., 2006).
potentials, but with differences in hippocampal and parahippocampal Therefore, to be aware of an auditory stimulus, activation of the
areas,asearlyhippocampalevent-relatedpotentialsarenotattenuated, primaryauditorycortexisaprerequisitebutisnotsufficient(Laureys
inaccordancewith theabovementionedsingle-cellrecordings.Based etal., 2000; Boly etal., 2005). Studies performed on patients in
onthesedataitcanbehypothesizedthatintinnitusthismechanismis vegetative state who do not have conscious auditory percepts reveal
disrupted with persistent parahippocampal activity, preventing habit- thatauditorystimulistillactivatetheprimaryauditorycortexbutthat
uation.Ithasbeenhypothesizedthattheparahippocampalareaplaysa there is no functional connectivity to frontal areas in these patients
central role in memory recollection, sending information from the (Bolyetal.,2004). Primary auditorycortex activation might beonly
hippocampus to the association areas, and a dysfunction in this relatedtointensitycoding(Janckeetal.,1998)andnottheperceptper
mechanismispositedasanexplanationforcomplexauditoryphantom se. Melloni andcolleagues foundthat theta oscillations in the frontal
perceptssuchasauditoryhallucinations(Diederenetal.,2010).Inline regionsareessentialforconsciousperceptionduringthemaintenance
with this hypothetical mechanism, tinnitus can be seen as a result of intervalofvisualstimuli(Mellonietal.,2007).Takingthesefindings
constantsendingofstoredauditoryinformationfromthehippocampus together,weproposethatsynchronizedgammaactivityintheauditory
totheauditoryassociationareasbypersistentparahippocampalactivity. cortex is responsible for the tinnitus intensity (van der Loo etal.,
For chronic tinnitus patients increased gamma activity was also 2009), while synchronized theta activity in the supplementary motor
found in the right auditory cortex in comparison to patients with areamightbeaccountablefortheconsciousperceptionofthephantom
recent-onset tinnitus, and an ROI analysis demonstrated increased sound, analogous to the conscious perception for somatosensory
gammacurrentdensityintinnituspatientsbothinleft-andright-sided stimuli (de Lafuente & Romo, 2002, 2005, 2006). The functional
primaryandsecondaryauditorycortices.MEGdatasupportthalamo- heterogeneityofthesupplementalmotorarea(SMA)hasbeenshown
ª2011TheAuthors.EuropeanJournalofNeuroscienceª2011FederationofEuropeanNeuroscienceSocietiesandBlackwellPublishingLtd
EuropeanJournalofNeuroscience,1–14
8 S. Vanneste et al.
A
B
Fig.3. Connectivitycontrastanalysisbetweentinnituspatientsandcontrolgroup(P<0.05).Decreasedneuralphase-laggedsynchronizationwithin(A)theta(4–
7.5Hz)and(B)alpha2(10–12Hz)connectivityintinnituspatients.Anincreasedneuralphase-laggedtheta(4–7.5Hz)synchronizationcanbeseenbetweentheleft
andrightparahippocampalareaandanincreasedneuralphase-laggedalpha2(10–12Hz)synchronizationisevidentbetweentheleftparahippocampalareaandthe
leftauditorycortexandtherightparahippocampalareaandtherightauditorycortex.L,left;R,right;A,anterior;P,posterior;S,superior.
before and demonstrates that the SMA has multiple functions, auditory imagery and auditory hallucinations (Linden etal., 2011),
including motor, sensory, word generation and working memory two conditions of auditory perception without and external sound
functions (Chung et al., 2005). The SMA is also activated in both source, analogous to tinnitus. This might be related to illusionary
ª2011TheAuthors.EuropeanJournalofNeuroscienceª2011FederationofEuropeanNeuroscienceSocietiesandBlackwellPublishingLtd
EuropeanJournalofNeuroscience,1–14
Theneural network of phantomsoundchanges overtime 9
Fig.4. Connectivitycontrastanalysisbetweenbetweentinnituspatientsandcontrolgroup(P<0.05).Decreasedphase-laggedgamma(30–45Hz)synchronization
for tinnitus patients is found, except for increased phase-lagged gamma (30–45Hz) synchronization between the left parahippocampal area, the left and right
auditorycortex.L,left;R,right;A,anterior;P,posterior;S,superior.
continuity, filling in missed auditory input, as unconscious sensory Connectivity
repair for missed speech occurs in Broca’s area, bilateral anterior
Both decreased and increased connectivity were found in tinnitus
insulaandpre-SMA(Shahinetal.,2009).Thisisinaccordancewitha
patientsincomparisontocontrolsubjectsbetweenthedifferentROIs
PET study of tinnitus (Mirz et al., 2000) which found a right
for theta, alpha and gamma activity. For the theta frequency band,
prefrontal–temporal network subserving tinnitus perception. In
increased connectivity was demonstrated between the left and right
chronic tinnitus theta activity is increased in the lower part of the
parahippocampal areas. For the alpha2 frequency band, increased
premotor cortex. This might be related to tonal memory which in
connectivitywasfoundbetweentheleftparahippocampalareaandthe
fMRIisrelatedtoactivationofthelowerpremotorcortex(encroach-
leftauditorycortex,andbetweentherightparahippocampalareaandthe
ingonBroca’sarea),theauditorycortex,theinferiorparietallobeand
rightauditorycortex.Increasedconnectivitywasalsorevealedwithin
the anterior insula.Asin chronic tinnitus the auditory cortex and the
thegammafrequencybandbetweentheleftparahippocampalareaand
anterior insula are also more activated in comparison to tinnitus of
theleftandrightauditorycortexfortinnituspatientsincomparisonto
recent onset this could suggest that chronic tinnitus is mediated by
control subjects. Previous research has already demonstrated that a
similar areastothose implicated intonal memory.
directconnectionbetweentheauditorycortexandtheparahippocampal
area is necessary for auditory sensory gating (Grunwald etal., 2003;
Boutros etal., 2005, 2008; Korzyukov etal., 2007). With respect to
Dorsal anteriorcingulate cortex andinsula
tinnitus,patientsshowedapatternofdecreasedfunctionalconnectivity
Our result also revealed that the dorsal anterior cingulate cortex indicating that there are fewer connections formed within the brain,
(dACC)andinsulabecomemoreinvolvedovertime.Anexplanation similartothefindingsobtainedwithMEG(Schleeetal.,2009).
cannot be given yet for their involvement but a hypothesis can be Chronictinnitusincomparisontotinnitusofrecentonsetgoesalong
proposed. withadecreaseinthetaandalphaconnectivity,aswellasadecreasein
Whether an auditory stimulus is perceived consciously or not gamma connectivity. However, there is also an increase in gamma
depends on fluctuations in the dACC and insula (Sadaghiani etal., connectivitybetweenleftparahippocampalcortex,leftauditorycortex,
2009), analogous to what has been described for pain (Boly etal., left insula and right DLPFC. A recent MEG study has indicated
2007). The dACC plus anterior insula are thought to be part of the decreased connectivity in the alpha frequency band together with an
salience network (Seeley etal., 2007), encoding behaviourally increase in gamma-band connectivity for longer lasting tinnitus
importantandfunctionallysignificantinputs.Thusitisplausiblethat (>4 years; Schlee etal., 2009). The connections between DLPFC,
dACCandanteriorinsulasignifyalackofhabituationtotheconstant primary and secondary auditory cortex are interesting, as are those
phantom sound, analogous to what has been described for pain betweentheinsulaandtheprimaryandsecondaryauditorycortex.In
(Mobascheretal.,2010).Theseresultshavetobeseeninthelightof addition, reduced gamma phase-lagged connectivity was also found
connectivity. between the left primary and secondary auditory cortex and the left
ª2011TheAuthors.EuropeanJournalofNeuroscienceª2011FederationofEuropeanNeuroscienceSocietiesandBlackwellPublishingLtd
EuropeanJournalofNeuroscience,1–14
10 S.Vannesteet al.
A
B
Fig.5. Connectivitycontrastanalysisbetweenchronictinnituspatientsand tinnituspatientswithrecentonset(P<0.05).Decreased (A)theta(4–7.5Hz)and
(B)alpha2(10–12Hz)connectivityforchronictinnituspatients.L,left;R,right;A,anterior;P,posterior;S,superior.
parahippocampalarea.Thisisalsoinaccordancewiththefindingsof Asmentionedbefore,theinvolvementoftheparahippocampalarea
Schlee etal. (2009) who argue that the left temporal lobe shows a in the network might be responsible for constantly updating the
major hub in patients with short tinnitus duration, but this decreases tinnitus, preventing habituation (De Ridder etal., 2006) due to its
with longer tinnitus duration. sensory gating function, which has been shown in both animals
ª2011TheAuthors.EuropeanJournalofNeuroscienceª2011FederationofEuropeanNeuroscienceSocietiesandBlackwellPublishingLtd
EuropeanJournalofNeuroscience,1–14