Original
article
Abnormal
brain
oscillations
persist
after
recovery
from
bipolar
depression
P.
Canali
a,*
,
S.
Casarotto
b,
M.
Rosanova
b,e,
G.
Sferrazza-Papa
a,
A.G.
Casali
f,
O.
Gosseries
c,d,
M.
Massimini
b,
E.
Smeraldi
a,
C.
Colombo
a,
F.
Benedetti
aa
Departmentofclinicalneurosciences,scientificinstituteOspedaleSanRaffaele,universityVita-SaluteSanRaffaele,SanRaffaeleTurro,20,viaStamira d’Ancona,20127Milano,Italy
b
Departmentofbiomedicalandclinicalsciences‘‘L.Sacco’’,universita` degliStudidiMilano,Milano,Italy
cComasciencegroup,GIGAresearch&neurologydepartment,universityhospitalofLie`ge,Lie`ge,Belgium d
Centerforsleepandconsciousness,Postlelaboratory,departmentofpsychologyandpsychiatry,universityofWisconsin,Madison,WI,USA
e
FondazioneEuropeadiRicercaBiomedica,FERBOnlus,Milan,Italy
f
Instituteofscienceandtechnology,FederaluniversityofSa˜oPaulo,330,RuaTalim,Sa˜oJose´ dosCampos,Brazil
1. Introduction
Currently, bipolardisorder(BD)is thesixth leadingcauseof disability [1,2] and affects nearly 1–2% of the population worldwide[3].DuringillnessepisodesofBDpatientsexperience pervasivechangesinmoodandcognition,anddeficitsinexecutive functions, attention, psychomotor speed, verbal and visual memory often persist in euthymia [4], suggesting persistent changes in brain structure and function [5]. Identifying trait
markersofpersistentlyabnormalbrainfunctionisthenapriority toidentifynewtargetsfortreatmentofthesedysfunctions[6].
High frequency brain oscillations are rhythmic electrical phenomena,whicharegeneratedspontaneouslyandinresponse tostimuli,andwhichparallelsthenaturalmechanismforcarrying neuralinformationamongbrainareas[7]andintegratingcortical modules [8]. They are modified in many neuropsychiatric conditions, and in cognitive impairment [9]. Accordingly, they arealsomarkedlyreducedinBD.Cross-sectionalstudiessuggest thatalterationsintheGABA/glutamatergicsystems,andinneural circuits that regulate cognitive processing, may be reflected throughinalteredbrainoscillationsinBD[10]:evenineuthymic conditions,patientsshowedreducedgammaoscillations[11,12],
ARTICLE INFO Articlehistory: Received19July2016
Receivedinrevisedform15October2016 Accepted16October2016
Availableonline3February2017 Keywords: TMS/EEG Gammaoscillations Biomarkers Bipolardisorder GABAergiccircuits ABSTRACT
Whendirectlyperturbedinhealthysubjects,premotorcorticalareasgenerateelectricaloscillationsin
thebetarange(20–40Hz).Inschizophrenia,majordepressivedisorderandbipolardisorder(BD),these
oscillationsaremarkedlyreduced,intermsofamplitudeandfrequency.However,itstillremainsunclear
whethertheseabnormalitiescanbemodulatedovertime,oriftheycanbestillobservedaftertreatment.
Here,weemployedtranscranialmagneticstimulation(TMS)combinedwithEEGtoassessthefrontal
oscillatoryactivityineighteenBDpatientsbefore/afterantidepressanttreatments(sleepdeprivation
andlighttherapy),relativetoninehealthycontrols.Inordertodetectdominantfrequencies,event
relatedspectralperturbations(ERSP)werecomputedforeachTMS/EEGsessioninallparticipants,using
waveletdecomposition.Thenaturalfrequencyatwhichthecorticalcircuitoscillateswascalculatedas
thefrequencyvaluewiththelargestpoweracross300mspost-stimulustimeinterval.Severityof
depressionmarkedlydecreasedaftertreatmentwith12patientsachievingresponseandninepatients
achievingremission.TMS/EEGresultedinasignificantactivationofthebeta/gammabandresponse(21–
50Hz)inhealthycontrols.Inpatients,themainfrequenciesofpremotorEEGresponsestoTMSdidnot
significantlychangebefore/aftertreatmentandwerealwayssignificantlylowerthanthoseofcontrols
(11–27Hz)andcomparableinpatientsachievingremissionandinthosenotrespondingtotreatment.
TheseresultssuggestthatthereductionofnaturalfrequenciesisatraitmarkerofBD,independentfrom
theclinicalstatusofthepatients.Thepresentfindingsshedlightontheneurobiologicalunderpinningof
severe psychiatric disorders and demonstrate that TMS/EEG represents a unique tool to develop
biomarkersinpsychiatry.
ß2016ElsevierMassonSAS.Allrightsreserved.
* Correspondingauthor.Tel.:+390226433156;fax:+390226433265. E-mailaddress:canali.paola@gmail.com(P.Canali).
ContentslistsavailableatScienceDirect
European
Psychiatry
j our na l ho me p a ge : ht t p: / / w ww . e ur opsy -j ou rna l . c om
http://dx.doi.org/10.1016/j.eurpsy.2016.10.005
reduced long distance gamma coherence between frontal and temporoparietalregions[11],anddecreasedbetasynchronization inthefrontalregion[13].
The combination of transcranial magnetic stimulation with electroencephalogram(TMS/EEG)represents anon-invasive, per-turbationalapproachtoprecisely identify theintegrity of thala-mocorticalcircuitsbydirectlychallengingthe brain’scapacityto produceandsustainoscillatoryactivity[14–16].WithTMS/EEG,we previouslyreportedthateachcorticalregiontendstooscillateata specificnaturalfrequency [17],andthatthe mainfrequenciesof frontalEEGresponsestoTMSweresignificantlyreducedinpatients withBD,majordepressivedisorder,andschizophreniarelativeto healthysubjects(11–27Hzvs21–50Hz,respectively)[18].
Itremainsuncleariftheseabnormalitieschangeovertime,and nolongitudinalstudyhasyetassessedhigh-frequencyoscillations before and afterresponse toantidepressants. Sleepdeprivation and light therapy (SD+LT) provide a model of antidepressant treatment which allows to study the biological correlates of psychopathologyatclosetimepointsandwithouttheconfounding effects of drugs [19]. Using this model antidepressant, we previouslyshowedthatresponseassociatesboth,withTMS/EEG evokedmeasuresofcorticalexcitability[20],withcorticalvolumes andfunction,andconcentrationsofneurotrophins[21].Here,we aimed at investigating the oscillatoryproperties of the frontal cortex by TMS/EEG before and after treatment with combined chronotherapeutictechniques(SD+LT).
2. Materialsandmethods
2.1. Participants,treatmentanddatacollection
Westudied18consecutivelyadmittedinpatients(14females; meanSDage:42.69.6;ageatonsetofillness:27.97.4;yearsat school:13.54.3;previousdepressiveepisodes:6.15.3;previous manicepisodes:3.12.2)sufferingfromamajordepressiveepisode, withoutpsychoticfeatures, affected byBD(DSM-IV criteria, SCID interview).Inclusioncriteriawere abaseline Hamiltondepression ratingscale(HDRS)scoreof18orhigher;absenceofotherdiagnosis onaxisIandofmentalretardationonaxisII;absenceofpregnancy, history of epilepsy, or major medical and neurologic disorders; absenceofahistoryofdrugoralcoholdependencewithinthelast 6months;notreatmentwithlong-actingneurolepticdrugsinthelast 3monthsbeforeadmission.Ninehealthyparticipants(6females,age 38.910.5)servedascontrols.Afteracompletedescriptionofthe study, awritten informedconsent wasobtained. Allthe research activitieswereapprovedbythelocalethicalcommittee.
AllpatientsweretreatedforoneweekwithSD+LT[22].They weretotallysleepdeprivedondays1,3and5,from7:00amto 7:00pmofthesubsequentday;andwereallowedtorecoversleep ondays2,4,and6.Allpatientswereadministereda10,000-lux whitelightfor30minutes,givenat3:00amduringtheSDnight and in the morning after recovery sleep, half an hour after awakening. Five patients were on ongoing lithium treatment (meanSDdailydose:750251mg),andcontinuedit;thirteen startedittogetherwiththechronotherapeuticprocedure(600mg/ day) to enhance its effect and prevent relapse [22]. No other psychotropicdrugwasadministeredduringthestudy.
Severityofdepressionwasratedatbaseline(day0)andafter treatment(day7)onthe21-itemHDRS.
2.2. TMS/EEGprocedure
TMS/EEGwasperformedbeforeandaftertreatment(day0and 5,at08:30am).Stimulationparameters(location,intensity,angle, coil orientation) were maintained constant and reproducible througha neuronavigationsystem (Nexstim,Helsinki,Finland).
Spontaneous EEG was continuously recorded for about 3min beforeeachTMS/EEGrecordingsession.
Prior to the TMS/EEG recording sessions, anatomical whole headimagesofeachpatientwereobtainedwitha3.0-Tscanner (Gyroscan Intera, Philips, Netherlands; T1-weighted MPRAGE sequence;TR2500ms,TE4.6ms,yielding220transversalslices with a thickness of 0.8mm). The acquired volume was then segmentedtoobtaina3Dmodelofthesurfaceofthescalpandof thecortex,tobeuploadedinthebrainnavigationsoftware.
TheexperimentalsetupincludedTMSwithaFocalBipulse8-Coil (EximiaTMSstimulator;NexstimLtd.,Helsinki,Finland)equipped withanavigatedbrainstimulationsystem(NBS;NexstimLtd.)and a3D-infraredtrackingpositionsensorunit(Polaris,NorthemDigital Inc.,Waterloo,Canada).EEGwasrecordedwitha60-channel TMS-compatibleEEGamplifier(NexstimLtd,Helsinki,Finland)equipped withsample-and-holdcircuitsthatpreventtherecordingfromthe powerful TMS-related artifacts [23]. EEG cap was repositioned beforeeachsession,controllingforreproducibilityoflocationusing theNBSsystem.Impedanceswerekeptbelow5k
V
.EEGsignals were band-pass filtered between 0.1–500Hz, and sampled at 1.450Hzwith16-bitresolution.Electro-oculogramwasrecorded withtwoadditionalelectrodesontheforeheadtomeasureocular movementsandblinks.ThisequipmentprovidesinrealtimetheTMScoilpositionand subject’shead,withinthereferencespaceofindividualmagnetic resonance imaging (MRI) by the co-registration between the fiducialspoints(nasion,lefttragusandrighttragus)selectedonthe individualMRIwiththecorrespondingdigitizedscalplandmarks. The exact location of thestimulation site wasadjusted on the individualMRIinordertopreventaccidentalmuscletwitchesthat often affectEEGrecordings, and toestimatethe electrical field induced by TMS pulses, which depends on the stimulation intensity (V/m). The TMS intensity was adjusted according to themaximumelectricfieldintensity(expressedinV/m)estimated on thecorticalsurface,rather thanrelying onindividual motor thresholdoronthepercentageofmaximumstimulatoroutput.
To ensuresignificantEEG responses[24]TMS intensitywas always>90V/masestimatedbytheNBSsystem,foreachpatient. TMSwasdeliveredontheconvexityofthemiddlecaudalportionof the superior frontal gyrus close to the midline (Brodmann’s areas6),withthecurrentperpendiculartoitsmainaxis.These brainareasshowedthehighestchangesofmetabolicrateandEEG correlatesbetweenwakeandsleep[25]andhavebeenassociated withtheantidepressanteffectsofSD[19].
ToobtainsignificantTMSevokedpotentials(TEP)withagood signal-to-noise ratio,about 200–300stimuliwere delivered for eachsessionatfrequencyrandomlydistributedbetween1.5–1.8s (equivalent toabout0.5–0.6Hz).Thisstimulationratedoes not inducesignificantreorganization/plasticityprocessesthat might possibly interferewithlongitudinalmeasurements[26].During TMSstimulationpatientswerelayingonanergonomicchair,with eyesopenlookingatafixationpointonascreen,andworeinserted earplugscontinuouslyplayingamaskingnoisethatabolishedthe auditorypotentialselicitedbyTMS-associatedclick[27]. 2.3. Dataanalysis
Data analysis was carried out using Matlab (2007b, The Mathworks Inc.,Natick,MA). TMSevoked potentialscontaining activity from sources other than neural, such as spontaneous muscles activity or ocular movements, were automatically identified and rejected using a semi-automatic algorithm (EOG>70
m
V or absolute power of EEG channel F8 above 25Hz, >0.9m
V2) [24]. Thereafter, single trials and channels contaminated by residual artifacts werevisually inspected and excluded from further analysis. Selected trials were band-passfiltered between 2–80Hz, down-sampled to 725Hz, and re-referencedtothecommon averagereference.EachTMS-evoked responsewasobtainedbyaveraging150–250artifactfreetrials.
In order to quantify the responses in the time-frequency domain[28],fromeachTMS/EEGsession,wemeasuredthe event-related spectral perturbation (ERSP) changes in the power spectrum using wavelet decomposition (3.5 oscillation cycles) acrosssingle-trialsatthechannelclosesttothestimulationsite (selectedchannelforsinglesubject:Fzn=1;FC1n=8;FCzn=8; FC2n=4;C1n=1;Czn=4;C2n=1).TheERSPwasnormalizedby subtractingthemeanbaselinepowerspectrum.SignificantERSP wereevaluatedbyapplyingabootstrapstatisticalmethodbased on a surrogate distribution randomly derived from the pre-stimulusonset ( 700 50ms).Statisticalsignificancelevel was set at P<0.01 and only significant values were considered in theanalysis. Averaged ERSPvalues acrossalltrials ofa session werecalculatedbetween8and50Hz(1Hzbinresolution)overa 20–300 millisecond time window, corresponding to the main EEG activity evoked by TMS. The natural frequency was computedasthefrequencybinwiththelargestcumulatedERSP overtime[17].
Data were analyzed with Student’s t-test and Pearson’s correlation.Moreover,weperformedarepeatedmeasuresANOVA onTMSevokednatural frequenciesbefore/aftertreatment,with timeandresponsetotreatmentasindependentfactors.Analyses wereperformedinthecontextofthegenerallinearmodel(GLM)
[29,30].Thesignificance oftheeffectofthesingleindependent factoron eachdependent variablewasestimated(leastsquares method) by parametric estimates of predictor variables and followingstandardcomputationalprocedures[31].
3. Results
Severity of depression markedly decreased after treatment (HDRSday0:23.25.8;day7:8.57.5;t=8.85,P<0.00001),with 12patients(66.6%)achievingresponse(HDRS50%reduction) and 9patients(50%)achievingremission(HDRS<8).
Data obtained with the TMS/EEG procedure are showed in
Fig.1.ToppanelshowstheaverageEEGresponsestoTMS(grey traces)forthechannelclosesttothestimulationsite(blacktrace) over the premotor area. To detect the natural frequency, we measuredtheevent-relatedspectralperturbation(ERSP)foreach
Fig.1.Toppanel:averageEEGresponsestoTMS(greytracesrepresentthe60recordingchannels)forthechannelclosesttothestimulationsite(blacktrace)overthe premotorarea;color-coded:event-relatedspectralperturbation(ERSP)plotsreflectthesignificantTMSrelatedchangesinamplitudeandtheirduration.Dottedlinesshow thefrequencywiththehighestactivity(naturalfrequency).DataareshownfromarepresentativehealthysubjectandforaBDpatient.Bottompanel:individualnatural frequencyvaluesforhealthycontrolsubjectsandpatientswithbipolardisorderbefore/aftertheantidepressanttreatments(SD+LT).
single subject. Specifically, we cumulated the ERSP between 8–50Hzand20–300msevokedbyTMS.Themainfrequencyat whichasystemoscillateswasselectedbythefrequencyshowing thelargestactivityacrosstime.DatashowninFig.1arefroma representativesubject.
TMSresultedinasignificantactivationofthebeta/gammaband response in healthy controls (27.03.0Hz, range 23–33). In patients,themainfrequenciesoffrontalEEGresponsestoTMSdid not significantly change before/after treatment (day 0: 19.445.41Hz, range 10–27; day 7: 19.304.99Hz, range 13– 27;t=0.391,P=0.700),andatbothtimepointsweresignificantly lower than those of controls (day 0: t=3.87, P=0.0006; day 7: t=4.22,P=0.0002).Valueswerecloselysimilarinpatients eventu-allyachievingremission,ornot(remitters:day0,20.375.50and day7,20.214.93;nonremitters:day0,18.515.46andday7, 18.405.18).Valuesbefore/aftertreatmentwerehighlycorrelated (r=0.9616,P<0.0001).
Arepeated measuresANOVAgave no effects of timeand of responseto treatment on EEG measuresbefore/aftertreatment (time: F1,16=0.14, P=0.709; response: F1,16=0.56, P=0.467; interaction:F1,16=0.006,P=0.941).Moreover,naturalfrequencies didcorrelateneitherwithHDRSscoresbefore/aftertreatment,nor with their delta change, nor with clinical and demographic characteristicsofthepatients.
4. Discussion
Inthisstudy,weassessednaturalfrequenciesofcorticalcircuits beforeandafterantidepressanttreatmentinBD.Notably,neither changes over time nor any correlation with the severity of depression were observed. Evoked brain oscillations remained lowerthanthoseofhealthycontrols,andcomparableinpatients achievingremissionandinthosenotrespondingtotreatment.This suggeststhatthereductionofnaturalfrequenciesisatraitmarker ofBD,independentfromthecurrentclinicalstatusofthepatients. Thepresentobservation thatsuccessfulantidepressant treat-ment is unable to normalize thalamocortical circuits in BD is consistentwithpreviouslyreportedreducedgammacoherencein euthymic patients withBD [11].Recently, we showedthat the reductionofnaturalfrequenciesisdetectableinschizophrenia,BD, and major depressive disorder [18]. The observation of its persistence in BD after treatment now suggests that this trait markercouldbelinkedtopersistenttraitcharacteristicsofbrain structureandfunctioninBD.Basedonexistingliterature,themost likely candidates are GABAergicand glutamatergic neurotrans-mission.
FastactivityofparvalbuminGABAinterneuronsisneededfor high-frequency oscillations [32,33], with gamma suppression resultingfromtheirinhibition[34].GABAinhibitoryinterneurons produceandsustaincomplexlarge-scalenetworkoscillationsin fastfrequencybands(40–100Hz)[35],alsogeneratinginhibitory potentialsinexcitatorypyramidalneurons[36].ThisGABAergic activityresultsingammaoscillations[37–40],andisreducedin BD.Post-mortemstudiesshowedareduceddensityofGABAergic neuronsinthecortexofpatientswithBD[35],withareductionin thenumericaldensityofparvalbumin-andsomatostatin-positive interneurons[41]andintheprefrontalcortexreducedmarkersof the parvalbumin subpopulation [42]. In vivo, MR spectroscopy confirmedlowGABAlevelsintheprefrontalcortex[43].
Inthegenerationandmaintenanceofhighfrequency oscilla-tions,GABAergicmechanismsarelikelytointeractwith glutama-tergic mechanisms, involving NMDA and AMPA receptors
[44].Glutamatelevelsinbraintissuewereincreasedbothinvivo
[45],andpost-mortem[46]inpatientswithBD.Geneticablationof NMDA receptors in parvalbumin interneurons resulted in increasedgamma power [47]. The decreaseof excitatory input
tofastspikingparvalbumininterneurons,inducedbytheNMDA antagonist ketamine [48,49], lead toincreased gamma activity
[50–52]. AMPA receptorantagonists inhibit gammaoscillations
[32].KetaminehasrapidandmarkedantidepressanteffectsinBD
[53], and it is not the only therapeutic agent to act on brain oscillations. Evoked beta responses from lithium-treated BD patients werehigherthanthose inbothdrug-freeeuthymicBD patientsandhealthycontrols[54],andinterestingly,post-mortem GABAlevelswereincreasedafterlithiumtreatment[46].
Altogether, thesedata supportthehypothesisthat abnormal GABAergicandglutamatergicneurotransmissioncouldbecritical to explain the abnormal gamma oscillations observed in our patients[34,55],andthattheseabnormalitiesarepersistentand detected both duringthelifespan,and post-mortem.Alternative interpretationscannot beruledout;forinstance,EEGpower in highfrequencybandshasbeenpositivelycorrelatedwitharterial spin labelling MRI measures of resting cerebral perfusion in healthysubjects[56].
Fastoscillatorydynamicsareneededforlarge-scaleintegration and synchronous communication between brain regions, are thought to parallel the emergence of coherent behavior and cognition [57], and are implicated in many brain functions including the processing of sensory stimuli [36], language comprehension [58], cognitive skills [59], cognitive processing inrecognitionmemory[60].Itistemptingtohypothesizethattheir persistentabnormalityinBDcouldmarkthepersistentdeficitsin higher cognitive functions and brain network connectivity associatedwiththedisorder[4,5,61].Furtherresearchisneeded toclarifythisissue.
Strengths of the present study include a focused research question,state-of-the-artTMS/EEGmethods,andstraightforward effects.However,ourexperimentalsettingdidnotallowtodirectly assesstheroleofdeeperstructures,suchashippocampus,which contributetogammaoscillations[62].Weobtainedanexcellent power tostudygroupdifferences,but couldnotconsiderother biological markers, gene variants, and their interaction with clinicalvariables.Patientswerenondrug-naı¨ve.Recruitmentwas inasingleethnicgroup,thusraisingthepossibilityofpopulation stratificationslimitingthegeneralizabilityofthefindings. Authorsandcontributors
F.B.,P.C.,M.M.andM.R.designedthestudy.E.S.,F.B.andM.M. obtained the funding. C.C., F.B. and P.C. were involved in participants’ recruitment and selection. P.C. and G.S.P.collected theclinicaldatawiththesupervisionofF.B.P.C.,S.C.,G.S.P.andO.G. carriedouttheTMS-EEGexperimentalprocedures.P.C.,M.R.,M.M., S.C.,G.S.P.,A.C.andOGdesignedthedataanalysesandcarrieditout withcontributionsfromF.B.P.C.andF.B.wrotethefirstdraftofthe manuscript,withotherauthorscontributingtodatainterpretation andfinalmanuscriptpreparation.F.B.andP.C.hadfullaccesstoall ofthedatainthestudyandtakeresponsibilityfortheintegrityof thedataandtheaccuracyofthedataanalysis.Allauthorstakefinal responsibilityforthedecisiontosubmitforpublication.
Funding
Theauthorsdeclarenofundingsourceforthepresentwork. Disclosureofinterest
Theauthorsdeclarethattheyhavenocompetinginterest. Acknowledgments
References
[1]BauerMS,KirkGF,GavinC,WillifordWO.Determinantsoffunctionaloutcome andhealthcarecostsinbipolardisorder:ahigh-intensityfollow-upstudy.J AffectDisord2001;65:231–41.
[2]MurrayCJ,LopezAD.Evidence-basedhealthpolicy–lessonsfromtheGlobal burdenofdiseasestudy.Science1996;274:740–3.
[3]KetterTA.Nosology,diagnosticchallenges,andunmetneedsinmanaging bipolardisorder.JClinPsychiatry2010;71:e27.
[4]PolettiS,SferrazzaPapaG,LocatelliC,ColomboC,BenedettiF. Neuropsycho-logicaldeficitsinbipolardepressionpersistaftersuccessfulantidepressant treatment.JAffectDisord2014;156:144–9.
[5]PolettiS,BollettiniI,MazzaE,LocatelliC,RadaelliD,VaiB,etal.Cognitive performancesassociatewithmeasuresofwhitematterintegrityinbipolar disorder.JAffectDisord2015;174:342–52.
[6]BiomarkersDefinitions Working G. Biomarkers and surrogateendpoints: preferred definitions and conceptual framework. Clin Pharmacol Ther 2001;69:89–95.
[7]BuzsakiG,DraguhnA.Neuronaloscillationsincorticalnetworks.Science 2004;304:1926–9.
[8]NikolicD,FriesP,SingerW.Gammaoscillations:precisetemporal coordina-tionwithoutametronome.TrendsCognSci2013;17:54–5.
[9]Bas¸arE.Brainoscillationsinneuropsychiatricdisease.DialoguesClinNeurosci 2013;15:291–300.
[10]Ozerdem A,KocaaslanS, TuncaZ,Basar E.Event related oscillationsin euthymicpatientswithbipolardisorder.NeurosciLett2008;444:5–10.
[11]OzerdemA,GuntekinB,AtagunI,TurpB,BasarE.Reducedlongdistance gamma(28-48Hz)coherenceineuthymicpatientswithbipolardisorder.J AffectDisord2011;132:325–32.
[12]ShawA,BrealyJ,RichardsonH,MuthukumaraswamySD,EddenRA,John EvansC,etal.Markedreductionsinvisualevokedresponsesbutnot gamma-aminobutyric acid concentrationsor gamma-band measuresin remitted depression.BiolPsychiatry2013;73:691–8.
[13]ChenSS,TuPC,SuTP,HsiehJC,LinYC,ChenLF.Impairedfrontal synchroni-zation ofspontaneousmagnetoencephalographicactivityinpatientswith bipolardisorder.NeurosciLett2008;445:174–8.
[14]BuzsakiG,WatsonBO.Brainrhythmsandneuralsyntax:implicationsfor efficientcodingofcognitivecontentandneuropsychiatricdisease.Dialogues ClinNeurosci2012;14:345–67.
[15]CanaliP.AroleforTMS/EEGinneuropsychiatricdisorders.NeurolPsychiatry BrainRes2014;20:37–40.
[16]Rosanova M,Casarotto S, PigoriniA, Canali P, CasaliAG, MassiminiM. Combiningtranscranialmagneticstimulationwithelectroencephalography tostudyhumancorticalexcitabilityandeffectiveconnectivity. In:FellinT, HalassaM,editors.Neuronalnetworkanalysisconceptsandexperimental approaches.NewYork,Heidelberg:Springer;2012.p.435–57.
[17]RosanovaM,CasaliA,BellinaV,RestaF,MariottiM,MassiminiM.Natural frequenciesofhumancorticothalamiccircuits.JNeurosci2009;29:7679–85.
[18]CanaliP,SarassoS,RosanovaM,CasarottoS,Sferrazza-PapaG,GosseriesO, etal.Sharedreductionofoscillatorynaturalfrequenciesinbipolardisorder, majordepressivedisorderandschizophrenia.JAffectDisord2015;184:111–5.
[19]BenedettiF,BernasconiA,BlasiV,CadioliM,ColomboC,FaliniA,etal.Neural andgeneticcorrelatesofantidepressantresponsetosleepdeprivation–A functionalmagneticresonanceimagingstudyofmoralvalencedecision,in bipolardepression.ArchGenPsychiatry2007;64:179–87.
[20]CanaliP,SferrazzaPapaG,CasaliAG,FecchioM,PigoriniA,SchienaG,etal. Changesofcorticalexcitabilityasbiomarkersofantidepressantresponsein bipolardepression.BipolarDisord2014;16:809–19.
[21]BenedettiF,PolettiS,HoogenboezemTA,LocatelliC,AmbreeO,deWitH,etal. Stemcellfactor(SCF)isaputativebiomarkerofantidepressantresponse.J NeuroimmunePharmacol2016.
[22]BenedettiF,RiccaboniR,LocatelliC,PolettiS,DallaspeziaS,ColomboC.Rapid treatmentresponseofsuicidalsymptomstolithium,sleepdeprivation,and lighttherapy(chronotherapeutics)indrug-resistantbipolardepression.JClin Psychiatry2014;75:133–40.
[23]VirtanenJ,RuohonenJ,NaatanenR,IlmoniemiRJ.Instrumentationforthe measurementofelectricbrainresponsestotranscranialmagneticstimulation. MedBiolEngComput1999;37:322–6.
[24]CasaliAG,CasarottoS,RosanovaM,MariottiM,MassiminiM.Generalindices tocharacterizetheelectricalresponseofthecerebralcortextoTMS. Neuroi-mage2010;49:1459–68.
[25]HorneJA.Humansleep,sleeplossandbehaviour.Implicationsforthe pre-frontalcortexandpsychiatricdisorder.BrJPsychiatry1993;162:413–9.
[26]CasarottoS,RomeroLauroLJ,BellinaV,CasaliAG,RosanovaM,PigoriniA,etal. EEGresponsestoTMSaresensitivetochangesintheperturbationparameters andrepeatableovertime.PlosOne2010;5:e10281.
[27]MassiminiM,FerrarelliF,HuberR,EsserSK,SinghH,TononiG.Breakdownof corticaleffectiveconnectivityduringsleep.Science2005;309:2228–32.
[28]DelormeA,MakeigS.EEGLAB:anopensourcetoolboxforanalysisof single-trialEEGdynamicsincludingindependentcomponentanalysis.JNeurosci Methods2004;134:9–21.
[29]McCullochCE,SearleSR,NeuhausJM.Generalized,linear,andmixedmodels, 2nded.,NewYork:JohnWiley&Sons;2008.
[30]TimmN,KimK.Univariateandmultivariategenerallinearmodels:theoryand applicationswithSAS, 2nded.,Berlin:HeidelbergSpringer;2006.
[31]Hill T,LewickiP. Statistics:methods andapplications. Acomprehensive referencefor science,industry, anddata mining.GeneralLinear Models, StatSoft,Tulsa(OK).2006;Chapter18:245–76.
[32]BartosM,VidaI, JonasP.Synaptic mechanisms ofsynchronizedgamma oscillationsin inhibitory interneuronnetworks. Nat Rev Neurosci 2007; 8:45–56.
[33]UhlhaasPJ,HaenschelC,NikolicD,SingerW.Theroleofoscillationsand synchronyincorticalnetworksandtheirputativerelevanceforthe patho-physiologyofschizophrenia.SchizophrBull2008;34:927–43.
[34]SohalVS,ZhangF,YizharO,DeisserothK.Parvalbuminneuronsandgamma rhythmsenhancecorticalcircuitperformance.Nature2009;459:698–702.
[35]BenesFM,BerrettaS.Gabaergicinterneurons.implicationsfor understand-ingschizophreniaandbipolardisorder.Neuropsychopharmacology2001; 25:1–27.
[36]WhittingtonMA,CunninghamMO,LeBeauFE,RaccaC,TraubRD.Multiple originsofthecorticalgammarhythm.DevNeurobiol2011;71:92–106.
[37]GrayCM,McCormick DA.Chatteringcells:superficialpyramidalneurons contributingtothegenerationofsynchronousoscillationsinthevisualcortex. Science1996;274:109–13.
[38]TraubRD,BibbigA,LeBeauFE,CunninghamMO,WhittingtonMA.Persistent gammaoscillationsinsuperficiallayersofratauditoryneocortex:experiment andmodel.JPhysiol2005;562:3–8.
[39]Gonzalez-BurgosG,LewisDA.GABAneuronsandthemechanismsofnetwork oscillations:implicationsforunderstandingcorticaldysfunctionin schizo-phrenia.SchizophrBull2008;34:944–61.
[40]BrennerCA,KieffaberPD,ClementzBA,JohannesenJK,ShekharA,O’Donnell BF,etal.Event-relatedpotentialabnormalitiesinschizophrenia:afailureto ‘‘gatein’’salientinformation?SchizophrRes2009;113:332–8.
[41]WangAY,LohmannKM,YangCK,ZimmermanEI,PantazopoulosH,HerringN, etal.Bipolardisordertype1andschizophreniaareaccompaniedbydecreased densityofparvalbumin-andsomatostatin-positiveinterneuronsinthe para-hippocampalregion.ActaNeuropathol2011;122:615–26.
[42]SibilleE,MorrisHM,KotaRS,LewisDA.GABA-relatedtranscripts inthe dorsolateralprefrontalcortexinmooddisorders.IntJNeuropsychopharmacol 2011;14:721–34.
[43]BhagwagarZ,WylezinskaM,JezzardP,EvansJ,AshworthF,SuleA,etal. Reductioninoccipitalcortexgamma-aminobutyricacidconcentrationsin medication-freerecoveredunipolardepressedandbipolarsubjects.Biol Psy-chiatry2007;61:806–12.
[44]UhlhaasPJ,SingerW.High-frequencyoscillationsandtheneurobiologyof schizophrenia.DialoguesClinNeurosci2013;15:301–13.
[45]ChittyKM,LagopoulosJ,LeeRS,HickieIB,HermensDF.Asystematicreview andmeta-analysisofprotonmagneticresonancespectroscopyandmismatch negativityinbipolardisorder.EurNeuropsychopharmacol2013;23:1348–63.
[46]LanMJ,McLoughlinGA,GriffinJL,TsangTM,HuangJT,YuanP,etal. Meta-bonomicanalysisidentifiesmolecularchangesassociatedwiththe patho-physiology and drug treatment of bipolar disorder. Mol Psychiatry 2009;14:269–79.
[47]KorotkovaT,FuchsEC,PonomarenkoA,vonEngelhardtJ,MonyerH.NMDA receptorablationonparvalbumin-positiveinterneuronsimpairshippocampal synchrony, spatial representations, and working memory. Neuron 2010; 68:557–69.
[48]HongLE,SummerfeltA,BuchananRW,O’DonnellP,ThakerGK,WeilerMA,etal. Gammaanddeltaneuraloscillationsandassociationwithclinicalsymptoms
under subanesthetic ketamine. Neuropsychopharmacology 2010;35:
632–40.
[49]HakamiT,JonesNC,TolmachevaEA,GaudiasJ,ChaumontJ,SalzbergM,etal. NMDAreceptorhypofunctionleadstogeneralizedandpersistentaberrant gammaoscillationsindependentofhyperlocomotionandthestateof con-sciousness.PlosOne2009;4:e6755.
[50]HomayounH,MoghaddamB.NMDAreceptorhypofunctionproducesopposite effectsonprefrontalcortexinterneuronsandpyramidalneurons.JNeurosci 2007;27:11496–500.
[51]SpencerKM.Thefunctionalconsequencesofcorticalcircuitabnormalitieson gammaoscillationsinschizophrenia:insightsfromcomputationalmodeling. FrontHumNeurosci2009;3:33.
[52]YizharO,FennoLE,DavidsonTJ,MogriM,DeisserothK.Optogeneticsinneural systems.Neuron2011;71:9–34.
[53]ZarateJrCA,BrutscheNE,IbrahimL,Franco-ChavesJ,DiazgranadosN, Crav-chikA,etal.Replicationofketamine’santidepressantefficacyinbipolar depression: a randomized controlledadd-on trial. BiolPsychiatry 2012; 71:939–46.
[54]OzerdemaA,GuntekindB,AtaguneMI,BasarE.Brainoscillationsinbipolar disorderinsearchofnewbiomarkers.SupplClinNeurophysiol2013;62: 207–21.
[55]WhittingtonMA,FaulknerHJ,DohenyHC,TraubRD.Neuronalfastoscillations asatargetsiteforpsychoactivedrugs.PharmacolTher2000;86:171–90.
[56]O’GormanRL,PoilSS,BrandeisD,KlaverP,BollmannS,GhisleniC,etal. Coupling between resting cerebral perfusion and EEG. Brain Topogr 2013;26:442–57.
[57]VarelaF,LachauxJP,RodriguezE,MartinerieJ.Thebrainweb:phase synchro-nizationandlarge-scaleintegration.NatRevNeurosci2001;2:229–39.
[58]LewisAG,WangL,BastiaansenM.Fastoscillatorydynamicsduringlanguage comprehension:unificationversusmaintenanceandprediction?BrainLang 2015;148:51–63.
[59]BenasichAA,GouZ,ChoudhuryN,HarrisKD.Earlycognitiveandlanguage skillsarelinkedtorestingfrontalgammapoweracrossthefirst3years.Behav BrainRes2008;195:215–22.
[60]KucewiczMT,CimbalnikJ,MatsumotoJY,BrinkmannBH,BowerMR,VasoliV, etal.Highfrequencyoscillationsareassociatedwithcognitiveprocessingin humanrecognitionmemory.Brain2014;137:2231–44.
[61]BenedettiF,BollettiniI.Recentfindingsontheroleofwhitematterpathology inbipolardisorder.HarvRevPsychiatry2014;22:338–41.
[62] Basar E, Schurmann M, Basar-Eroglu C, Demiralp T. Selectively
distributedgammabandsystemofthebrain.IntJPsychophysiol2001; 39:129–35.