Default mode network connectivity is linked to cognitive functioning and CSF Aβ1-42 levels in Alzheimer's disease

Default

mode

network

connectivity

is

linked

to

cognitive

functioning

and

CSF

A

b

1

–42

levels

in

Alzheimer

’s

disease

Ozlem

Celebi

a

,

Andac

Uzdogan

b

,

Kader

Karli

Oguz

c

,

Arzu

Ceylan

Has

d

,

Anil

Dolgun

e

,

Gul

Yalcin

Cakmakli

a

,

Filiz

Akbiyik

b

,

Bulent

Elibol

f

,

Esen

Saka

f,

*

a_{HacettepeUniversity,InstituteofHealthSciences,DepartmentofNeurologicalSciencesandPsychiatry,Ankara,Turkey} b

HacettepeUniversityFacultyofMedicine,DepartmentofMedicalBiochemistry,Ankara,Turkey

c

HacettepeUniversityFacultyofMedicine,DepartmentofRadiology,Ankara,Turkey

d

BilkentUniversity,NationalMagneticResonanceResearchCenter,Ankara,Turkey

e

HacettepeUniversityFacultyofMedicine,DepartmentofBiostatistics,Ankara,Turkey

f

HacettepeUniversityFacultyofMedicine,DepartmentofNeurology,Ankara,Turkey

ARTICLE INFO Articlehistory: Received11May2015

Receivedinrevisedform25September2015 Accepted28September2015

Availableonline17October2015 Keywords:

Alzheimer’sdisease Biomarker Cognition

Restingfunctionalmagneticresonance imaging

Posteriorcingulatecortex Retrosplenialcortex

ABSTRACT

Background:Changesinthedefaultmodenetwork(DMN)activityareearlyfeaturesofAlzheimer’s disease(AD)andmaybelinkedtoAD-specificAbpathology.

Methods:Cognitiveprofiles;DMNconnectivityalterations;andcerebrospinalfluid(CSF)amyloidbeta (Ab)1–42,totaltau,phosphorylatedtau181,anda-synucleinlevelswerestudiedin21patientswithAD

and10controls.

Results:DMNactivityisalteredinAD.Posteriorcingulatecortex(PCC)functionalconnectivitywithother partsofDMNwasrelatedtocognitivefunctionscores.ThereductionofconnectivityofthedorsalPCC withtheretrosplenialcortexontherightsidewascloselyrelatedtodecreasedCSFAb1–42levelsin

patientswithAD.

Conclusions:ThedorsalPCCandretrosplenialcortexmayhavespecialimportanceinthepathogenesis andcognitivefindingsofAD.

ã2015ElsevierIrelandLtd.Allrightsreserved.

1.Introduction

Alzheimer’sdisease(AD)isthemostcommon neurodegenera-tivedisease,which ischaracterized bydistinctclinicalfeatures, pathology,andpathogenesis.AnearlydiagnosisofADcanbemade using well-established biomarkers and imaging markers. These markersareexpectedtobetherationaluseoftherapeutics for halting the progression of the AD-specific pathophysiological cascadeandthuspreventingAlzheimer’sdementia.Concomitant withthis idea, increasing efforts have beenmade to find new markersandtoimprovetheexistingonesusedforADandother neurodegenerativedisorders.

A proposed model of the AD pathophysiological cascade suggeststhattheinitialdetectablepathophysiologicalchangein ADistheaccumulationofamyloidbeta(A

b

)inthebrain(Hardy, 1992;Selkoe,2002).Accordingly,thedetectionofA

b

depositionon positronemissiontomography(PET)anddecreasedcerebrospinal

fluid (CSF) A

b

levels are expected to be the earliest signs for diagnosing AD in vivo (Blennow, Zetterberg, & Fagan, 2012;

Sperlingetal.,2011).Measuresofsynapticdysfunction,suchasa regional decrease in glucose metabolism, demonstrated by fluorodeoxyglucose (FDG)PETand alteredconnectivitypatterns in functionalmagnetic resonanceimaging (fMRI) closely follow these A

b

changes. Likewise,specificalterations inresting-state networks, in particular changes in the default mode network (DMN), are evident in prodromal AD. On the other hand, the elevated CSF total-tau (t-tau) and phosphorylated tau 181 (p-tau181)levelsarebelievedtobeindicatorsofneuronalinjury.

Therefore, theyoccurrelativelylater inthedevelopmentof AD pathology(Blennowetal.,2012).

Resting-statenetworksareactiveduringrestandcanbeeither taskpositiveortasknegative(Rektorova,2014).DMNisa task-negative network measured by resting-state fMRI. It is closely related to higher cognitive functionsand is largelybelieved to reflectthetransitofearlydisease-specificmolecularalterationsto evident neurodegenerationindegenerativediseases suchasAD (Barkhof,Haller,&Rombouts,2014).Ithasalreadybeenshownthat resting-statenetworks, suchas DMNand thecentral executive network,arealteredinAD(Gouretal.,2014;Hahnetal.,2013).

* Correspondingauthor.HacettepeUniversityFacultyofMedicine,Departmentof Neurology,Ankara06100,Turkey.

E-mailaddress:esensaka@hacettepe.edu.tr(E.Saka).

http://dx.doi.org/10.1016/j.archger.2015.09.010

0167-4943/ã2015ElsevierIrelandLtd.Allrightsreserved.

ContentslistsavailableatScienceDirect

Archives

of

Gerontology

and

Geriatrics

DMN is the most extensively studied resting-state network, revealingsimilaralteredpatternsoffunctionalconnectivity(FC) in patients with both sporadic and autosomal dominant AD (ADAD).SeveralstudieshavealsoanalyzedthevalidityofDMN activitychangesindiagnosingADandreportedaspecificityand sensitivityof70–80%(Balthazar,deCampos,Franco,Damasceno,& Cendes,2014;Kochetal.,2012;Lietal.,2012).Thedisruptionof DMNhasbeendemonstratedasaveryearlyfeatureinAD;eventhe asymptomaticcarriersofADADmutationsandindividualsatrisk ofAD,suchaspatientswithmildcognitiveimpairment, elderly carriers of ApoE

e

4, and cognitively normal patients with PET demonstratingA

b

deposition,experiencedecreasesintheDMN connectivity(Chhatwaletal.,2013;Elmanetal.,2014;Jacketal., 2013;Kochetal.,2014;Sperlingetal.,2009).

TheNationalInstituteonAgingandtheAlzheimer’sAssociation (NIA-AA)recentlyestablishedareviseddiagnosticcriteriaforAD dementia(McKhannetal.,2011)andrecommendedtheuseofCSF biomarkers,suchasdecreasedCSFA

b

1–42levelsandelevatedCSF

t-tauandp-tau181levels,inadditiontoPETormagneticresonance

imaging(MRI),fordetectingAD,whenthereisaneedtoconfirm that the dementia syndrome is caused by AD. There are a considerablenumberofstudiescomparingorcombiningdifferent biomarkers for diagnosing AD, although there are only a few studiesthatanalyzebothDMNconnectivityandCSFbiomarkers andtheirrelationshipin AD (Wang etal., 2013;Liet al.,2013; Shelineetal.,2010).Inthepresentstudy,wehypothesizedthatthe biomarkersofADindifferentmodalities(i.e.,proteinmarkersand imagingmarkers)and themagnitudeofclinicaldysfunctionare relatediftheproposedmodeloftheADpathophysiologicalcascade wasvalid.Furthermore,findingasignificantcorrelationbetween thewell-establishedCSFbiomarkersandchangesinDMNactivity, particularlyinthespecificregionsofthebrain,couldincreasethe diagnosticutilityofDMNactivitystudiesinAD.Therefore,inthe presentstudy, we analyzed theCSF A

b

1–42, t-tau,and p-tau181

levelsaswell-knownbiomarkersofADpathologyaswellasthe CSF

a

-synuclein levels as another marker of synaptic and/or neuronalinjuryinagroupofnewlydiagnosedpatientswithADin comparisonwiththecontrols.DMNactivityandcognitiveprofiles ofthesamesubjectswerealsocomparativelyexaminedinrelation tothesebiomarkers.

2.Methods

2.1.Subjectsandclinicalassessment

Twenty-one patientswithprobable ADand 10age-matched healthy controls participated in the present study. Written informed consent was obtained from all participants and/or a familymember.ThestudywasapprovedbytheAnkaraUniversity FacultyofMedicineEthicalBoardforClinicalStudies.Theclinical diagnosisofADwasbasedontheNationalInstituteofNeurological and Communicative Disorders and Stroke–Alzheimer’s Disease andRelatedDisordersAssociationcriteria(McKhannetal.,1984, 2011).Patients with AD were prospectivelyrecruited fromthe NeurologyoutpatientclinicsattheHacettepeUniversityHospitals betweenSeptember2012andJune2014.Controlswererecruited fromthecommunity.Healthycontrolshadnocognitivecomplaints or a history of neurological disease. The mini-mental state examination (MMSE), geriatric depression scale (GDS), and a neuropsychologicaltestbattery,includingenhanced cuedrecall, semanticfluency, phonemic fluency, trailmaking test A and B, recitingmonths backwards,and clock drawing tests(evaluated accordingtoa4-pointscoringsystem),wereadministeredtoall subjects.Theadministrationand scoringof thesetestswereas previouslydescribed (Cangoz, Karakoc,&Selekler,2006;Celebi, Temucin,Elibol,&Saka,2014;Saka,Mihci, Topcuoglu,&Balkan,

2006). Clinical assessment, MRI data acquisition, and CSF collection(onlyforpatientswithAD)wereperformedwithina 1-weekperiodineachsubject.

2.2.CSFcollectionandanalysisofbiomarkers

CSFwas collectedaccordingtonewconsensus-based recom-mendations for preanalytical issues with AD and Parkinson’s disease(PD)CSF biomarkeranalysis(delCampoetal.,2012).In brief, atraumatic lumbar puncture was performed using a 22-gaugeneedleintheL4–5orL3–4intervertebralspacetoremove 12mLCSF.No seriousadverseeventwasreported.CSF samples were centrifuged at 2000g for 10min within 30min of collection. Samples were aliquoted into 0.5-mL polypropylene storage tubes and stored in a 80C freezer until analysis. All archived samples were analyzed at the Clinical Biochemistry LaboratoryatHacettepeUniversityHospitals.

CSFA

b

1–42,t-tau,andp-tau181levelswereanalyzedusing

plate-basedenzyme-linked immunosorbent assay(ELISA)(INNOTEST; Innogenetics, Ghent, Belgium). CSF

a

-synuclein was measured using ELISA (catalog no. SIG 38974; Covance, Dedham, MA) according to the manufacturer’s instructions. CSF hemoglobin levelswerealsodeterminedtoevaluatebloodcontaminationand tocontrol the possibleconfounding effect of hemolysison the CSF

a

-synuclein level. CSF hemoglobinwas analyzedusing the ELISAmethod withreagentsobtained fromBethyl Laboratories according to the manufacturer’s instructions. CSF protein concentrationsweredeterminedusingacolorimetricmethodby BeckmanCoulteranalyzers.

2.3.Imagingmethods

2.3.1.Dataacquisitionandpreprocessing

Allimagingdataofthebrainwereobtainedona3TMRscanner (Magnetom,TrioTIMsystem;Siemens,Germany)equippedwith an 8-channelphased-arrayheadcoil. A T2*-weighted gradient-echospiralpulsesequence(repetitiontime[TR]2000ms;timeto echo[TE],30ms;flip angle,80_{)was usedfortheresting-state}

functionalscans.It comprised150dynamic seriesthatrequired 5min.Thefieldofview(FOV)was230230mm2_{,andthematrix}

sizewas6464,givinganin-planespatialresolutionof3.6mm. The subjectskepttheireyes closed and remained still without concentrating on anything specific. For anatomical data, all subjectsalsounderwentstructuralT1-weighted three-dimension-al (3D) high-resolution images with 0.9-mm isotropic voxels [magnetization-preparedrapidgradient-echo-(MPRAGE)](TR/TE: 1900/3.4ms; FA:90;FOV: 256mm; matrix:224256;distance factor:50%).

Data were preprocessed using SPM8 software (Statistical Parametric Mapping SPM Software, 2015). Motion correction using least-squares minimization without higher-order correc-tions forspinhistoryand normalization(Fristonetal.,1995)of bothfunctionalandstructural3DT1-weightedMPRAGEdatatothe MontrealNeurologicalInstitute(MNI)templatewereperformed. Imageswerethenresampledevery2-mmisotropicvoxelsizeusing sincinterpolationandsmoothedwitha4-mmfullwidthathalf maximum(FWHM)Gaussiankerneltodecreasethespatialnoise. Resampling and smoothing were done in three dimensions yielding a 2-mm3 _{resolution and effective spatial smoothness}

(FWHM) of 7.27.18.4mm. Normalized data of each subject wereusedasinputforMELODICindependentcomponentanalysis (Beckmann& Smith, 2004) for identifyingand removingnoise components.Temporalfilteringwasappliedusingbothhigh-pass (Gaussian-weighted least-squares straight line fitting, with sigma=100.0s)andlow-passfilters(Gaussianlow-passtemporal filtering:half-widthathalf-maximum,2.8s).

We performed individual seed-based connectivity for the analysisofFCinindividualsandanalysisoftheseed-voxelgroup for building a composite map for the entire group of similar subjectsinFig.2.Allofthepresenteddata,exceptthedatainFig.2, aretheresultsofindividualseed-basedconnectivityanalysis. 2.3.2.FCanalysis(individualseed-basedconnectivity)

For individual seed-based connectivity, FMRIB’s Software Library(www.fmrib.ox.ac.uk/fsl)(FSL,2015)wasusedtoexamine functionalcouplingrespectivelybetweentheposteriorcingulate cortex(PCC),leftdorsalPCC,rightdorsalPCC,leftventralPCC,right ventralPCC(SupplementaryTable1),andsevenotherregionsof interest(ROI) (Supplementary Table 2)locatedin DMNregions during the resting state. Each subject’s time series was trans-formed into MNI space using 12 of freedom linear affine transformation implementedin FLIRT(voxelsize=111mm) toobtainthetimeseriesforeachseed,foreachsubject.ROIwas definedassphereswitha 6-mmradius.Toforma spheremask aroundavoxelofinterest,thefslmathscommandswereused.The mean time series was extractedfrom a normalized and trans-formed fMRI for all of the sphere-masked voxels using the fslmeantscommand.TheFEATtoolbox(FEAT,2015)wasusedwith theseedtimeseriesfilethatwasjustcreatedtoperformmultiple regressionanalyses(Dietal.,2008).Toextracttheuniquevariance thatthetimeseriesforeachseedmaskreflects,timeseriesforeach seedwereorthogonalizedwithrespecttoeach otherusing the Gram–Schmidt process. Individual subject-level maps of all positivelyandnegativelypredictedvoxelsforeachregressorwere createdby thisanalysis, and Z-scorevalues werecalculatedfor eachsubject(Marguliesetal.,2007).

2.3.3.FCanalysis(seed-voxelgroupanalysis)

UsingtheCONNtoolboxinMATLABR2008a(www.nitrc.org/ projects/conn),principalcomponentsassociatedwithsegmented whitematterandCSFforeachsubjectwereidentified,andwhite matter, CSF, and realignment parameters were entered as confounders in a first-level analysis. The data were band-pass filteredto0.008–0.09Hz.Foragroupanalysis,seed-voxelanalysis was then used, and connectivity patterns were separately specifiedforfive6-mmsphericalclusters;fiveseedROIlocated inPCC,theleftdorsalPCC,therightdorsalPCC,theleftventral PCC, and the right ventral PCC. Using the specified five seed ROI, temporal correlations were computed between these seeds and all other voxels in the brain. Seed-to-voxel results were reported as significant at a voxel-wise threshold level of p<0.001 uncorrected and a cluster-level threshold of p<0.05 corrected for thefalse discovery rate (FDR). The t-test and Fisher’s Z-transformed correlations were used to compute differences in FC between the patients with AD and healthy controls.

2.4.Statisticalanalysis

Mean_standarderrorofthemeanvaluesareusedtodescribe thequantitativevariables.Frequenciesandpercentagesaregiven fornominaldata.Normalityassumptionwascheckedusing the Shapiro–Wilktest.IndependentsamplesttestandMann–Whitney Utestwereusedtocomparethegroupsintermsofquantitative variables. The sex distribution among the study groups was analyzedusingFisher’sexacttest.Anexploratoryfactoranalysis (using principal component analysis for factor extraction) was applied tocombine several quantitative variables into a single factor to reduce high-dimensional data. Accordingly, PCC (the connectivityofPCCwiththeothernodesoftheDMN),ventralPCC (theconnectivityoftherightandleftventralPCCwiththeother nodesofDMN),dorsalPCC(theconnectivityoftherightandleft

dorsalPCCwiththeothernodesoftheDMN),memory(enhanced cued recall and semanticfluency), and executivefunction (trial makingB,recitingmonthsbackwards,andverbalfluency)factors werebuilt.Forallextractedfactors,explainedvariancesweremore than 50% with a single factor. The nonparametric Spearman correlationanalysis(withoutapplyinganycovariateadjustment) was then used to analyze the relationship between both the quantitativevariablesandextractedfactors.TheFDRcontrolwas applied for all analyses to correct for multiple comparisons (Benjamini&Hochberg,1995).IBM-SPSSversion21.0(IBMCorp., Armonk,NY)andRpackage“p.adjust”(RCoreTeam,2013)version 3.0.2(TheRFoundationforStatisticalComputing)wereused,and thestatisticalsignificancewassetatp<0.05.

3.Results

3.1.Demographicsandcognitiveprofile

The demographics and cognitive features of subjects are summarized in Table 1. Patients with AD and controls were matchedforage,education,andsex.InpatientswithAD,disease severity was mild to moderate, with a mean MMSE score of 18.11.4,whereasthemeanMMSEscoreofthecontrolgroupwas 27.50.7.TheperformanceofpatientswithADwaslowerthan that of the controls in all of the neuropsychological tests administered(all pvalues<0.05).TheGDS scorewashigher in patientswithADthancontrols(p=0.025).

3.2.CSFbiomarkers

CSFA

b

1–42,t-tau,p-tau181,and

a

-synucleinlevelsandratioof

a

-synucleintototalproteininpatientswithADarepresentedin

Table 2. These values were compared with another group of control subjects (n=24) who underwent diagnostic lumbar puncture for headache or peripheral nervous system disorders. None of them had neurodegenerative disease. As expected, patientswithAD had alower CSFA

b

1–42and higher CSF t-tau

andp-tau181levelsthancontrols(p<0.05).PatientswithADhad

higher CSF

a

-synucleinlevels than controls(2035.2266.5 vs. 1425.9182.5pg/mL,respectively; p<0.05) aswellas a higher ratio of

a

-synuclein to total protein (51.86.5 vs. 35.54.7; p<0.05).Thisratiocorrelatedwiththelevelofp-tau181inpatients

withAD.

Table1

ThedemographicsandthecognitivefeaturesofthesubjectswithADandcontrols.

AD (21) meanSEM Control (10) meanSEM pValue* Age(yrs) 66.41.9 66.62.3 0.852 Education(yrs) 9.31.3 10.31.3 0.917 Gender(F/M) 8/13 8/2 0.054

Diseaseduration(yrs) 2.80.3 – N/A

MMSE (range) 18.11.4(11-23) 27.50.7(24-30) 0.002 ECR 16.22.9 46.10.7 0.002 Semanticfluency 8.40.7 15.71.2 0.002 Verbalfluency 5.71.0 9.71.5 0.028 TrailmakingA 165.625.3 66.89.6 0.024 TrailmakingB 293.416.3 144.026.0 0.002 Recitingmonthsbackwards 167.828.6 80.937.1 0.024 Clockdrawingtest 2.20.4 3.80.1 0.024

GDS 4.41.1 1.50.5 0.025

AD,Alzheimer’sdisease;MMSE,mini-mentalstateexamination;ECR,enhanced cuedrecalltest;GDS,geriatricdepressionscale;SEM,standarderrorofmean.

*

AdjustedpvaluesaccordingtotheFalseDiscoveryRate(FDR);N/A,statistical analysisisnotavailable.

3.3.CorrelationofcognitivetestperformancesandCSFbiomarkers The cognitive test results were not related to any of the AD-related CSF biomarkers (A

b

1–42, t-tau, and p-tau181) or

a

-synucleinlevels.

3.4.ComparisonofDMNFCacrossgroups

We used individualseed-based connectivityanalysis forthe comparisonofDMNFCacrossgroups.ADMNconnectivitymap derivedfromPCC(PCCasawholeanddorsalPCCandventralPCC separatelyonbothsides)showeddecreasedFCwiththemedial prefrontalcortex,medialtemporallobes,inferiorparietallobules, and retrosplenial cortices in patients with AD relative to age-matchedcontrols(Fig.1;p<0.05).Wealsoperformedseed-voxel groupanalysisforthevisualcomparisonofADandhealthycontrol groups(Fig.2).

3.5.CorrelationofDMNintegritychangeswithcognitivefunctions Dataderivedfromindividualseed-basedconnectivityanalyses wereusedforcorrelationstudies.PatientswithADandhealthy controlswereanalyzedas a whole.To avoid multiple compar-isons,theconnectivityofPCCwiththeothernodesofDMNasa single unit (the PCC factor, which was extracted using factor analysis)wasanalyzed.Wealsoclassifiedneuropsychologicaltest scores into four cognitive domains: memory (enhanced cued recall and semantic _fluency), visuospatial (clock drawing), attention(trialmakingA),andexecutivefunctions(trialmaking B, reciting months backwards, and verbal fluency) and asked whetherthePCCfactorwasrelatedtothecognitiveperformance of study subjects. This analysis showed that the PCC factor was relatedto the MMSE score (r=0.546, p=0.001), attention (r= 0.392, p=0.029), executive function factor (r= 0.429, p=0.016),GDSscore(r= 0.522,p=0.003),andmoreprominently withmemoryfactor(r=0.647,p=0.0001)(Table3).

TheresultsoffurthercorrelationanalysesofPCCconnectivity withtheothernodesofDMNseparatelyandindividualcognitive testscoresaresummarizedinTable4.Thisanalysisdemonstrated the presence of significant correlations between individual cognitivetestscoresandconnectivityoftheirfunctionallyrelated DMNnodeswithPCC.

3.6.CorrelationofDMNintegritychangeswithCSFbiomarkers Dataderivedfromindividualseed-basedconnectivityanalyses wereusedforcorrelationstudies.BecauseCSFbiomarkerswere notstudiedinhealthycontrols,thisanalysiswasconductedonlyin patientswithAD.WefirstanalyzedwhetherPCCconnectivityasa whole(PCCfactor)wasrelatedtoanyoftheCSFmarkersofADand foundthatCSFA

b

1–42levelstendedtocorrelatewiththelevelof

connectivity of PCC, although this did not reach statistical significance(r=0.421,p=0.058).

Next, we soughtthe level of connectivityof thedorsal and ventral PCC with each DMN node individually. This analysis showed that among otherDMN nodes, theconnectivity of the dorsal PCC with theretrosplenial cortexon the rightside was significantlyrelatedtoCSFA

b

1–42levels(r=0.499,p=0.021).This

findingshows that theamountof decreaseof A

b

1–42 inCSF is

related to a decrease in dorsal PCC and retrosplenial cortex connectivity.

4.Discussion

In thepresent study,weshowed that,in arestingstate,the connectivity of PCC with the medial prefrontal cortex, medial temporallobes,inferiorparietallobules,andretrosplenialcortices isdecreasedinpatientswithnewlydiagnosedmildtomoderate AD.Asexpected,thesepatientshadalowerCSFA

b

1–42andhigher

CSFt-tauandp-tau181levelsthancontrols.Weadditionallyfound

thatCSF

a

-synucleinlevelswereincreasedinthesepatients.The comparativeanalysisof allthesebiologicalandimaging param-eterstogetherwithcognitiveprofilingdepictedsomeexpectedas wellasnovelfindings.WhenpatientswithADandhealthycontrols wereanalyzedasawhole,thescoresofcognitivefunctions,suchas attention, executive functions, and more prominently memory domains,wererelatedtothemagnitudeofPCCconnectivitywith other parts of DMN. Moreover, as a novel finding, in DMN, decreased connectivityofthe dorsalPCC withtheretrosplenial cortexwas foundtobeclosely relatedtodecreased CSFA

b

1–42

levelsinpatientswithAD.

DMN includes medialtemporal, frontal,and parietal cortical areas.Itis active whenattention is notfocused butdeactivates during task-relatedactivities(Buckner,Andrews-Hanna,&Schacter,2008). InstudiesofAD,changesinDMNactivityparticularlyinvolvedFC betweenPCCandhippocampus/medialtemporallobes(Greicius, Srivastava,Reiss,&Menon,2004;Wangetal.,2013).Whenstudying PCCconnectivity,wealsoanalyzeddorsalandventralPCCseparately becauseFCofthedorsalandventralPCCmaydifferintheresting stateandduringcognitivetasks.Forexample,similartotheventral PCC,thedorsalPCChasstrongconnectivitywithinDMN;however, unliketheventralPCC,italsohasconnectivitytoawiderangeof otherintrinsicconnectivitynetworks,includingfrontalandparietal regionsinvolved incognitivecontrol(Leech&Sharp, 2014).We foundherethatFCofPCC(ventral,dorsal,ortotal)withtheother partsofDMNisalteredinpatientswithAD.Wefurthershowedthat thestrength ofPCCFCis lower in subjectswithlower cognitivescores andhigherinsubjectswithhighercognitivefunction.Thisfindingis inaccordancewithpreviousstudiesthatshowedthealterationof PCCFCtobeproportionaltotheseverityofthedisease(Chhatwal etal.,2013).

The widely accepted and largely proved pathophysiological cascadeofADsuggeststhatalteredA

b

productionanditsdeposition isaveryearlystepandleadstootherAD-relatedneuronalandcircuit changes (Hardy, 1992; Sperling et al., 2011). Preferentially, A

b

-depositedareasincludethenodesofDMNsuchasPCC,retrosplenial cortex,precuneus,andparietal andtemporallobes(Wangetal., 2013; Buckner et al., 2005). Several studies have reported an associationbetweenCSFbiomarkersandDMNconnectivitychanges (Wangetal.,2013;Lietal.,2013;Shelineetal.,2010).Amongthem,

Wangetal.(2013)foundassociationsofdecreasedCSFA

b

1–42levels

andincreased CSF p-tau levelswithdecreases of FCof PCC andmedial temporal lobes in a large sample of cognitively normal older individuals. Liet al.(2013) showed thattheratioof A

b

42to

p-tau181correlatedwithFCwithinDMNintheleftprecuneusinAD. Basedonthisbackground,weexpectedchangesinDMNinpatients withADtoberelatedtoA

b

depositionandCSFA

b

levelsinour subjects. Indeed, wefoundthat CSFA

b

levelswererelated toFC ofthe dorsalPCCwiththeretrosplenialcortexontherightside.Inother

Table2

CSFbiomarkers,resultsofthesubjectswithADandcontrols. AD(n=21) MeanSEM Control(n=24) MeanSEM pValue Ab1–42(pg/ml) 772.364.5 1089.244.8 <0.0001 t-tau(pg/ml) 681.260.5 278.631.4 <0.0001 p-tau181,(pg/ml) 90.66.5 41.63.9 <0.0001 a-Synuclein(pg/ml) 2035.2266.5 1425.9182.5 0.047

a-Synucleintototalprotein,ratio 51.86.5 35.54.7 0.047 AD,Alzheimer’sdisease;SEM,standarderrorofmean.

Fig.1.Box-and-whiskerplotsofthefunctionalconnectivityvaluesbetween(A)theposteriorcingulatecortex(PCC)andtheotherdefaultmodenetwork(DMN)regions(B) rightventralPCCandtheotherDMNregions(C)LeftventralPCCandotherDMNregions(D)rightdorsalPCCandotherDMNregions(E)leftdorsalPCCandotherDMNregions. FigureshowsreducedconnectivityofthePCCandventral/dorsalPCCwiththemedialprefrontalcortex(MPFC),medialtemporallobes(MTL),inferiorparietallobules(IPL), andretrosplenialcortices(RSC)inAlzheimer’sdisease(AD)patients.*p<0.05,accordingtotheFalseDiscoveryRate(FDR)adjustment.IntheBox-and-whiskerplot,thebox representsthevaluesfromthelowertoupperquartile(theinterquartilerange)andthemiddlelinerepresentsthemedianofthefunctionalconnectivityvalues.

Fig.2.FunctionalconnectivitybetweenthePCCandallotherbrainregionsinthecontrolgroup(A),inADgroup(B).Figureshowsreducedfunctionalconnectivitywith medialprefrontalcortex,medialtemporallobes,inferiorparietallobulesandretrosplenialcorticesinADpatientsrelativetoage-matchedcontrols(twosidedFDR*p<0.05, seed-levelcorrected*p<0.05).

words,insubjectswithlowerCSFA

b

levels,FCofthedorsalPCCwith theretrosplenialcortex waslower.Thisfinding isnovelbutnot unexpected.PreviousFDGPETstudiesshoweddecreasedglucose metabolisminthevery early stagesofADinPCCandtheretrosplenial cortex(Buckneretal.,2005;Frings,Spehl,Weber,Hull,&Meyer, 2013).Itwasalsoshowntobeassociatedwithincreasedamyloid depositionand/oratrophyintheseregionsandthebrainregions anatomicallyandfunctionallyconnectedtothem(Buckneretal., 2005;Desgrangesetal.,2002;Fringsetal.,2013).Theretrosplenial cortex functions in spatial navigation and in acquiring new information, particularly autobiographical memory (Vann, Aggleton, & Maguire, 2009). It has connections with the hippocampal formation,parahippocampalregion,andanteriorandlateraldorsal thalamicnucleiaswellaswithPCC.Owingtothosefindingsand others,theretrosplenialcortexinadditiontoPCCisbelievedtoplaya majorroleincognitivedeficitsassociatedwithAD(Pengasetal., 2012).TheinvolvementofdorsalbutnottheventralPCCmaybe becauseofthemoredeliberateuseofthedorsalPCC,asitisbelieved to link networks that are functionally distinct but need to be coordinatedforefficientcognitiveprocessing(Leech&Sharp,2014). Consistentwiththisidea,Sperlingetal.foundthatincreasedA

b

depositionisassociatedwithhigherDMNactivityduringamemory task(Sperlingetal.,2009).

Additionally,weanalyzedCSF

a

-synucleinlevelsasaprotein marker of synaptic degeneration not specific to AD. In fact,

a

-synuclein is being studied as a biomarker of PD and other

a

-synucleinopathies. It is decreased in PD, multiple system atrophy, and Lewy body dementia (Mollenhauer et al., 2008; Mondello et al., 2014). However,

a

-synuclein is reported to be elevated in Creutzfeldt–Jacob disease (Kasai et al., 2014; Mollenhaueretal.,2008),traumaticbraininjury(Mondello,Buki,

Italiano,&Jeromin,2013),andAD(Slaetsetal.,2014).Ourfinding of increased CSF

a

-synuclein levels in patients with AD is in accordance withthosestudies.Furthermore,we foundthatthe

a

-synuclein levelspositivelycorrelated withincreasedp-tau181

in patients with AD. These findings further support the view that the decreased CSF

a

-synuclein levels can be a marker of synucleinopathies;however,anincreaseinCSF

a

-synucleinlevels may indicate neurodegeneration irrespective of the underlying neuropathology.

Thisstudyhasseverallimitations.Thecontrolgroupwassmall andmostlycomposedofwomen.Forethicalreasons,wecouldnot obtainCSFsamplesfromhealthycontrols;hence,allthecorrelative data betweenchanges in DMNand CSF biomarkerlevelswere derivedfrompatientswithAD.Furtherstudiesinalargersampleof bothcontrolandpatientgroupscoveringprodromaltoadvanced stages of AD are required. These studies mayallow a reliable assessmentofthediagnosticpowerofchangesinDMNactivityin diagnosingADcomparedwithCSFbiomarkersorothervalidated biomarkers.

5.Conclusions

ThedatashowthatDMNactivitychangesinADandisclosely related to cognitive functions and A

b

pathology. The findings furtherindicatethatthedorsalPCCandretrosplenialcorticesmay havespecialimportanceinthepathogenesisandcognitivefindings ofAD.

Disclosurestatement

Theauthorshavenoactualorpotentialconflictsofinterest.

Table3

Correlationofthedefaultmodenetwork(DMN)functionalconnectivity(FC)valueswiththecognitivetestscoresinstudysubjects.

MMSE Memoryfactor Visuo-spatialfunction Attention Executivefunctionfactor GDS

r p r p r p r p r p r p PCCfactor 0.55 0.001 0.65 <0.0001 0.32 NS 0.39 0.029 0.43 0.016 0.52 0.003 VentralPCCfactor 0.64 <0.0001 0.69 <0.0001 0.42 0.02 0.38 0.035 0.50 0.004 0.45 0.011 DorsalPCCfactor 0.58 0.001 0.66 <0.0001 0.37 0.041 0.35 NS 0.48 0.006 044 0.014 PCC-R/LMTL 0.52 0.002 0.52 0.003 0.42 0.018 0.24 NS 0.30 NS 0.48 0.007 R/LDorsalPCC-R/LMTL 0.51 0.003 0.53 0.002 0.36 0.045 0.26 NS 0.34 NS 0.39 0.029 R/LVentralPCC-R/LMTL 0.51 0.003 0.55 0.001 0.36 NS 0.22 NS 0.33 NS 0.40 0.026

MMSE,mini-mentalstateexamination;GDS,geriatricdepressionscale;PCC,posteriorcingulatecortex;R,right;L,left;MTL,medialtemporallobe;NS,non-significant;PCC factor,theconnectivityofPCCwiththeothernodesofDMNasasinglefactorwhichisextractedbyusingfactoranalysis;VentralPCCfactor,theconnectivityofRandLventral PCCwiththeothernodesofDMNasasinglefactorwhichisextractedbyusingfactoranalysis;DorsalPCCfactor,theconnectivityofRandLdorsalPCCwiththeothernodesof DMNasasinglefactorwhichisextractedbyusingfactoranalysis;Memoryfactor,asinglefactorofenhancedcuedrecallandsemanticfluency;Executivefunctionfactor,a singlefactoroftrialmakingB,recitingmonthsbackwards,verbalfluency.

Table4

CorrelationanalysisofthePCCfunctionalconnectivitywiththeotherDMNnodesandtheneuropsychologicaltestscoresinthestudygroups.

PCC-MPFC PCC-RMTL PCC-LMTL PCC-RIPL PCC-LIPL PCC-RRSC PCC-LRSC r p r p r p r p r p r p r p MMSE 0.38 0.033 0.44 0.013 0.65 <0.0001 0.44 0.013 0.24 NS 0.44 0.012 0.33 NS ECR 0.53 0.002 0.63 <0.0001 0.69 <0.0001 0.60 <0.0001 0.40 0.025 0.44 0.015 0.32 NS Semanticfluency 0.29 NS 0.24 NS 0.46 0.010 0.53 0.002 0.35 NS 0.48 0.006 0.50 0.004 Verbalfluency 0.11 NS 0.002 NS 0.25 NS 0.38 0.038 0.07 NS 0.33 NS 0.11 NS TrailmakingA 0.13 NS 0.08 NS 0.47 0.007 0.36 0.049 0.07 NS 0.49 0.006 0.23 NS TrailmakingB 0.24 NS 0.43 0.015 0.62 <0.0001 0.58 0.001 0.17 NS 0.45 0.011 0.22 NS

Recitingmonthsbackwards 0.11 NS 0.12 NS 0.40 0.026 0.31 NS 0.03 NS 0.39 0.032 0.32 NS

Clockdrawing 0.20 NS 0.33 NS 0.52 0.003 0.24 NS 0.15 NS 0.28 NS 0.10 NS

GDS 0.47 0.008 0.40 0.027 0.54 0.002 0.34 NS 0.27 NS 0.16 NS 0.28 NS

MMSE,mini-mentalstateexamination;ECR,enhancedcuedrecall;GDS,geriatricdepressionscale;PCC,posteriorcingulatecortex;R,right;L,left;MTL,medialtemporallobe; IPL,inferiorparietallobul;MPFC,medialprefrontalcortex;RSC,retrosplenialcortex.

Acknowledgements

ThestudywasfundedbyTUBITAK(SBAG112S360),andispart oftheBIOMARKAPDprojectintheJPNDprogramme.

AppendixA.Supplementarydata

Supplementarydataassociatedwiththisarticlecanbefound, intheonlineversion,at http://dx.doi.org/10.1016/j.archger.2015. 09.010.

References

Balthazar,M.L.,deCampos,B.M.,Franco,A.R.,Damasceno,B.P.,&Cendes,F.(2014). Wholecorticalanddefaultmodenetworkmeanfunctionalconnectivityas potentialbiomarkersformildAlzheimer’sdisease.PsychiatryRes.,221,37–42.

Barkhof,F.,Haller,S.,&Rombouts,S.A.(2014).Resting-statefunctionalmrimaging: anewwindowtothebrain.Radiology,272,29–49.

Beckmann,C.F.,&Smith,S.M.(2004).Probabilisticindependentcomponent analysisforfunctionalmagneticresonanceimaging.IEEETrans.Med.Imaging, 23,137–152.

Benjamini,Y.,&Hochberg,Y.(1995).Controllingthefalsediscoveryrate:apractical andpowerfulapproachtomultipletesting.J.R.Stat.Soc.,57(SeriesB),289–300.

Blennow,K.,Zetterberg,H.,&Fagan,A.M.(2012).Fluidbiomarkersinalzheimer disease.ColdSpringHarborPerspect.Med.,2,a006221.

Buckner,R.L.,Snyder,A.Z.,Shannon,B.J.,LaRossa,G.,Sachs,R.,Fotenos,A.F.,etal. (2005).Molecular,structural,andfunctionalcharacterizationofAlzheimer’s disease:evidenceforarelationshipbetweendefaultactivity,amyloid,and memory.J.Neurosci.,25,7709–7717.

Buckner,R.L.,Andrews-Hanna,J.R.,&Schacter,D.L.(2008).Thebrain’sdefault network:anatomy,function,andrelevancetodisease.Ann.N.Y.Acad.Sci.,1124, 1–38.

Cangoz,B.,Karakoc,E.,&Selekler,K.(2006).Thenormdeterminationandvalidity– reliabilitystudiesofClockDrawingTestonTurkishAdultsandElderlys(ages 50andover).Turk.J.Geriatr.,9,136–142(inTurkish).

Celebi,O.,Temucin,C.M.,Elibol,B.,&Saka,E.(2014).Cognitiveprofilinginrelation toshortlatencyafferentinhibitionoffrontalcortexinmultiplesystematrophy. ParkinsonismRelat.Disord.,20,632–636.

Chhatwal,J.P.,Schultz,A.P.,Johnson,K.,Benzinger,T.L.,Jack,C.Jr,Ances,B.M.,etal. (2013).Impaireddefaultnetworkfunctionalconnectivityinautosomal dominantAlzheimerdisease.Neurology,81,736–744.

RCoreTeam(2013).Alanguageandenvironmentforstatisticalcomputing. Vienna, Austria:RFoundationforStatisticalComputingISBN3-900051-07-0.

Desgranges,B.,Baron,J.C.,Lalevee,C.,Giffard,B.,Viader,F.,deLaSayette,V.,etal. (2002).TheneuralsubstratesofepisodicmemoryimpairmentinAlzheimer’s diseaseasrevealedbyfdg-pet:relationshiptodegreeofdeterioration.Brain, 125,1116–1124.

Di,M.,artino,A.,Scheres,A.,Margulies,D.S.,Kelly,A.M.,Uddin,L.Q.,etal.(2008). Functionalconnectivityofhumanstriatum:arestingstatefmristudy.Cereb. Cortex,18,2735–2747.

Elman,J.A.,Madison,C.M.,Baker,S.L.,Vogel,J.W.,Marks,S.M.,Crowley,S.,etal. (2014).Effectsofbeta-amyloidonrestingstatefunctionalconnectivitywithin andbetweennetworksreflectknownpatternsofregionalvulnerability.Cereb. Cortex.

FEAT(2015).<http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FEAT>Accessed01.27.15. FSL(2015).<http://fsl.fmrib.ox.ac.uk/fsl/>Accessed01.27.15.

Frings,L.,Spehl,T.S.,Weber,W.A.,Hull,M.,&Meyer,P.T.(2013).Amyloid-betaload predictsmedialtemporallobedysfunctioninalzheimerdementiaJournalof nuclearmedicine:officialpublication.Soc.Nucl.Med.,54,1909–1914.

Friston,K.J.,Holmes,A.P.,Poline,J.B.,Grasby,P.J.,Williams,S.C.,Frackowiak,R.S., etal.(1995).Analysisoffmritime-seriesrevisited.Neuroimage,2,45–53.

Gour,N.,Felician,O.,Didic,M.,Koric,L.,Gueriot,C.,Chanoine,V.,etal.(2014). Functionalconnectivitychangesdifferinearlyandlate-onsetAlzheimer’s disease.Hum.BrainMapp.,35,2978–2994.

Greicius,M.D.,Srivastava,G.,Reiss,A.L.,&Menon,V.(2004).Default-modenetwork activitydistinguishesAlzheimer’sdiseasefromhealthyaging:evidencefrom functionalMRI.Proc.Natl.Acad.Sci.U.S.A.,101,4637–4642.

Hahn,K.,Myers,N.,Prigarin,S.,Rodenacker,K.,Kurz,A.,Forstl,H.,etal.(2013). Selectivelyandprogressivelydisruptedstructuralconnectivityoffunctional brainnetworksinAlzheimer’sdisease—revealedbyanovelframeworkto analyzeedgedistributionsofnetworksdetectingdisruptionswithstrong statisticalevidence.Neuroimage,81,96–109.

Hardy,J.(1992).An‘anatomicalcascadehypothesis’forAlzheimer’sdisease.Trends Neurosci.,15,200–201.

Jack,C.R.Jr,Knopman,D.S.,Jagust,W.J.,Petersen,R.C.,Weiner,M.W.,Aisen,P.S.,et al.(2013).TrackingpathophysiologicalprocessesinAlzheimer’sdisease:an updatedhypotheticalmodelofdynamicbiomarkers.LancetNeurol.,12,207– 216.

Kasai,T.,Tokuda,T.,Ishii,R.,Ishigami,N.,Tsuboi,Y.,Nakagawa,M.,etal.(2014). Increasedalpha-synucleinlevelsinthecerebrospinalfluidofpatientswith creutzfeldt-jakobdisease.J.Neurol.,261,1203–1209.

Koch,W.,Teipel,S.,Mueller,S.,Benninghoff,J.,Wagner,M.,Bokde,A.L.,etal.(2012). DiagnosticpowerofdefaultmodenetworkrestingstatefMRIinthedetectionof Alzheimer’sdisease.Neurobiol.Aging,33,466–478.

Koch,K.,Myers,N.E.,Gottler,J.,Pasquini,L.,Grimmer,T.,Forster,S.,etal.(2014). Disruptedintrinsicnetworkslinkamyloid-betapathologyandimpaired cognitioninprodromalAlzheimer’sdisease.Cereb.Cortex.

Leech,R.,&Sharp,D.J.(2014).Theroleoftheposteriorcingulatecortexincognition anddisease.Brain,137,12–32.

Li,R.,Wu,X.,Fleisher,A.S.,Reiman,E.M.,Chen,K.,&Yao,L.(2012).Attention-related networksinAlzheimer’sdisease:arestingfunctionalMRIstudy.Hum.Brain Mapp.,33,1076–1088.

Li,X.,Li,T.Q.,Andreasen,N.,Wiberg,M.K.,Westman,E.,&Wahlund,L.O.(2013). Ratioofabeta42/p-tau181pincsfisassociatedwithaberrantdefaultmode networkinad.Sci.Rep.,3,1339.

Margulies,D.S.,Kelly,A.M.,Uddin,L.Q.,Biswal,B.B.,Castellanos,F.X.,&Milham,M. P.(2007).Mappingthefunctionalconnectivityofanteriorcingulatecortex. Neuroimage,37,579–588.

McKhann,G.,Drachman,D.,Folstein,M.,Katzman,R.,Price,D.,&Stadlan,E.M. (1984).ClinicaldiagnosisofAlzheimer’sdisease:reportofthenincds-adrda workgroupundertheauspicesofdepartmentofhealthandhumanservices taskforceonAlzheimer’sdisease.Neurology,34,939–944.

McKhann,G.M.,Knopman,D.S.,Chertkow,H.,Hyman,B.T.,Jack,C.R.Jr.,Kawas,C. H.,etal.(2011).ThediagnosisofdementiaduetoAlzheimer’sdisease: recommendationsfromtheNationalInstituteonAging-Alzheimer’s AssociationworkgroupsondiagnosticguidelinesforAlzheimer’sdisease. AlzheimersDement.,7,263–269.

Mollenhauer,B.,Cullen,V.,Kahn,I.,Krastins,B.,Outeiro,T.F.,Pepivani,I.,etal. (2008).Directquantificationofcsfalpha-synucleinbyelisaandfirst cross-sectionalstudyinpatientswithneurodegeneration.Exp.Neurol.,213,315–325.

Mondello,S.,Buki,A.,Italiano,D.,&Jeromin,A.(2013).Alpha-synucleinincsfof patientswithseveretraumaticbraininjury.Neurology,80,1662–1668.

Mondello,S.,Constantinescu,R.,Zetterberg,H.,Andreasson,U.,Holmberg,B.,& Jeromin,A.(2014).Csfalpha-synucleinanduch-l1levelsinParkinson’sdisease andatypicalparkinsoniandisorders.ParkinsonismRelat.Disord.,20,382–387.

Pengas,G.,Williams,G.B.,Acosta-Cabronero,J.,Ash,T.W.,Hong,Y.T., Izquierdo-Garcia,D.,etal.(2012).Therelationshipoftopographicalmemoryperformance toregionalneurodegenerationinalzheimer’sdisease.Front.AgingNeurosci.,4 (17).

Rektorova,I.(2014).Resting-statenetworksinAlzheimer’sdiseaseandParkinson’s disease.Neuro-DegenerativeDis.,13,186–188.

Saka,E.,Mihci,E.,Topcuoglu,M.A.,&Balkan,S.(2006).Enhancedcuedrecallhasa highutilityasascreeningtestinthediagnosisofAlzheimer’sdiseaseandmild cognitiveimpairmentinturkishpeople.Arch.Clin.Neuropsychol.,21,745–751.

Selkoe,D.J.(2002).Decipheringthegenesisandfateofamyloidbeta-proteinyields noveltherapiesforalzheimerdisease.J.Clin.Investig.,110,1375–1381.

Sheline,Y.I.,Morris,J.C.,Snyder,A.Z.,Price,J.L.,Yan,Z.,D’Angelo,G.,etal.(2010). Apoe4alleledisruptsrestingstatefmriconnectivityintheabsenceofamyloid plaquesordecreasedcsfabeta42.J.Neurosci.,30,17035–17040.

Sperling,R.A.,Laviolette,P.S.,O’Keefe,K.,O’Brien,J.,Rentz,D.M.,Pihlajamaki,M.,et al.(2009).Amyloiddepositionisassociatedwithimpaireddefaultnetwork functioninolderpersonswithoutdementia.Neuron,63,178–188.

Sperling,R.A.,Jack,C.R.Jr.,Black,S.E.,Frosch,M.P.,Greenberg,S.M.,Hyman,B.T.,et al.(2011).Amyloid-relatedimagingabnormalitiesinamyloid-modifying therapeutictrials:recommendationsfromtheAlzheimer’sassociationresearch roundtableworkgroup.AlzheimersDemen.,7,367–385.

StatisticalParametricMappingSPMSoftware(2015).<http://www.fil.ion.ucl.ac.uk/ spm/software/spm8/>Accessed01.27.15.

Vann,S.D.,Aggleton,J.P.,&Maguire,E.A.(2009).Whatdoestheretrosplenialcortex do?Nat.Rev.Neurosci.,10,792–802.

Wang,L.,Brier,M.R.,Snyder,A.Z.,Thomas,J.B.,Fagan,A.M.,Xiong,C.,etal.(2013). Cerebrospinalfluidabeta42,phosphorylatedtau181,andresting-state functionalconnectivity.JAMANeurol.,70,1242–1248.

delCampo,M.,Mollenhauer,B.,Bertolotto,A.,Engelborghs,S.,Hampel,H., Simonsen,A.H.,etal.(2012).Recommendationstostandardizepreanalytical confoundingfactorsinAlzheimer’sandParkinson’sdiseasecerebrospinalfluid biomarkers:anupdate.BiomarkersMed.,6,419–430.

Slaets,S.,Vanmechelen,E.,LeBastard,N.,Decraemer,H.,Vandijck,M.,Martin,J.J.,et al.(2014).Increasedcsfalpha-synucleinlevelsinAlzheimer’sdisease: correlationwithtaulevels.AlzheimersDemen.,10(Suppl.5),S290–8.