• Non ci sono risultati.

Simultaneous PET/MR head-neck cancer imaging: Preliminary clinical experience and multiparametric evaluation

N/A
N/A
Protected

Academic year: 2021

Condividi "Simultaneous PET/MR head-neck cancer imaging: Preliminary clinical experience and multiparametric evaluation"

Copied!
8
0
0

Testo completo

(1)

ContentslistsavailableatScienceDirect

European

Journal

of

Radiology

jo u r n al ho me p a g e : w w w . e l s e v i e r . c o m / l o c a t e / e j r a d

Simultaneous

PET/MR

head–neck

cancer

imaging:

Preliminary

clinical

experience

and

multiparametric

evaluation

M.

Covello

a,∗

,

C.

Cavaliere

a

,

M.

Aiello

a

,

M.S.

Cianelli

a

,

M.

Mesolella

b

,

B.

Iorio

b

,

A.

Rossi

a

,

E.

Nicolai

a

aIRCCSSDN,ViaE.Gianturco,111-11380143,Naples,Italy

bDepartmentofOtorhinolaryngoiatry,FedericoIIUniversity,Naples,Italy

a

r

t

i

c

l

e

i

n

f

o

Articlehistory: Received17March2014

Receivedinrevisedform26March2015 Accepted11April2015 Keywords: Hybridimaging PET/MR SUV DCE DWI

Headandneckcancer

a

b

s

t

r

a

c

t

Purpose:ToevaluatetheroleofsimultaneoushybridPET/MRimagingandtocorrelatemetabolicPETdata withmorpho-functionalparametersderivedbyMRIinpatientswithhead–neckcancer.

Methods:Forty-fourpatients,withhistologicallyconfirmedheadand neckmalignancy(22primary tumorsand22follow-up)werestudied.Patientsinitiallyreceivedaclinicalexamandendoscopywith directbiopsy.NextpatientsunderwentwholebodyPET/CTfollowedbyPET/MRofthehead/neckregion. PETandMRIstudieswereseparatelyevaluatedbytwoblindedgroups(bothincludedoneradiologist andonenuclearphysician)inordertodefinethepresenceorabsenceoflesions/recurrences.Regionsof interest(ROIs)analysiswasconductedontheprimarylesionatthelevelofmaximumsizeonmetabolic (SUVandMTV),diffusion(ADC)andperfusion(Ktrans,V

e,kepandiAUC)parameters.

Results:PET/MRexaminationsweresuccessfullyperformedonall44patients.Agreementbetweenthe twoblindedgroupswasfoundinanatomicallocationoflesionsbyPET/MR(Primarytumors:Cohen’s kappa0.93;Follow-up:Cohen’skappa0.89).TherewasasignificantcorrelationbetweenCT-SUV meas-uresandMR(e.g.,CT-SUVVOIvs.MR-SUVVOI:=0.97,p<0.001fortheentiresample).Therewas alsosignificantpositivecorrelationsbetweentheROIarea,SUVmeasures,andthemetabolicparameters (SUVandMTV)obtainedduringbothPET/CTandPET/MR.Asignificantnegativecorrelationwasobserved betweenADCandKtransvaluesintheprimarytumors.Inaddition,asignificantnegativecorrelationexisted

betweenMRSUVandADCinrecurrenttumors.

Conclusion:OurstudydemonstratesthefeasibilityofPET/MRimagingforprimarytumorsandrecurrent tumorsevaluationsofhead/neckmalignantlesions.WhenassessingHNC,PET/MRallowssimultaneous collectionofmultiparametricmetabolicandfunctionaldata.Thistechniquethereforeallowsforamore completecharacterizationofmalignantlesions.

©2015ElsevierIrelandLtd.Allrightsreserved.

1. Introduction

Headandneckcancers(HNC)accountforapproximately4–5%

ofallmalignantdisease.Ofthese,headandnecksquamouscell

car-cinoma(HNSCC)comprisesthevastmajority(90%)[1].Diagnosis

ofheadandneckcanceristypicallyachievedbyacombinationof

physicalexaminationandeithernasopharyngoscopyand/or

laryn-goscopywithdirectbiopsies.However,thefirstmanifestationof

the diseasecommonly involvesthe cervical lymph nodes, also

thepresenceofdistantmetastasesanddifferentprimariesmust

beconsidered.Followingdiagnosis,accuratestagingiscriticalfor

∗ Correspondingauthor.fax:+39081668841. E-mailaddress:[email protected](M.Covello).

selectionoftheappropriatetreatmentstrategy.Forearly-stage

dis-ease,surgeryorradiotherapyarethemostcommontreatments,

withasuccessrateofabout80%[2].Postoperativeradiotherapyis

typicallyrecommendedwhentheriskforlocoregionalrecurrence

exceeds20%[3].Forpatientswhosediseaseisnotcontrolledwith

definitiveradiotherapy,salvagesurgeryisrecommended[3].For

regionallyadvancedHNSCC,concurrentchemo-radiotherapyhas

beenwidely used[4].Currently, positronemission tomography

(PET)andmagneticresonance(MR)arewell-establishedimaging

techniquesforthestaging[5]andfollow-upofHNC.MRtechniques,

suchasdiffusionweightedimaging(DWI)anddynamiccontrast

enhancement (DCE), along with fluorodeoxyglucose (FDG)-PET

are thebasilar imaging techniques usedto assess angiogenesis

andmetabolisminHNSCC.Developmentofnoninvasiveimaging

biomarkersmightassistinplanningoptimaltreatmentstrategies,

http://dx.doi.org/10.1016/j.ejrad.2015.04.010

(2)

whichmayleadtoa morefavorabletherapeuticoutcome [6,7].

Therecentcommercialintroductionof hybridPET/MRscanners

offernewpossibilitiesforoncologicimaging,combiningthewide

rangeofphysiologicalinformation(glucosemetabolism,cell

prolif-eration,hypoxia,cellreceptorexpression)achievablebydifferent

PETtracerswiththehighspatialandcontrastresolutionofMRI.

Moreover, advanced MRI techniques (DWI, DCE and MR

spec-troscopy)yieldfunctionalinformationthatmaycomplementthe

findingsof PET imaging. Todate, severalPET/MRI studies have

shownpromisingresultsforclinicaloncologicapplications[8,9].

ThebestapplicationsforPET/MRIimagingexistforHNC.Duetothe

complexanatomyofthisregionthesuperbsoft-tissuecontrast

res-olutionofMRIisextremelyuseful.Forthisreason,PET/MRIseems

tobehighlyaccurateintheevaluationofprimaryextension

(T-staging).However,ithasnosignificantbenefitwithlymphnodes

involvement(N-staging)andlimitedimprovementformetastasis

detection(M-staging)whencomparedtoPET/CTexamination[10].

Therefore,theaims ofthestudy weretoevaluatefeasibility

ofsimultaneoushybridPET/MRimaginginpatientswithHNC.In

addition,metabolicdataobtainedfromPETandmorphologicand

functionaldata(DWI,DCE)obtainedbyMRIonprimarytumors

and recurrentlesionswere correlated.In this context, thenew

technologyofsimultaneousPET/MRscanningcanbesuitablefor

acomprehensivemultiparametricassessment.Wepresentour

ini-tialclinicalexperiencewithsimultaneousPET/MRimaginginthe

evaluationofpatientsaffectedbyHNC.

2. Materialsandmethods

2.1. PatientpopulationandImagingprotocol

InaccordancewiththeinstitutionalreviewboardandEthical

Committee,44consecutivepatients(meanage57±12;23males

and9females)withhistologicallyconfirmedheadandneck

malig-nancy(22primarytumorsand22recurrenttumors)werestudied

betweenMarch,2012toJune,2013.Exclusioncriteriaconsisted

of>150mg/dLbloodglucose,pregnancy,andstandard

contraindi-cationforMRI.Forprimarytumors,notherapyhasbeenstarted

atthetimeofMRimaging.Forrecurrentlesions,imagingsession

hasbeenperformed6–8weeksaftertheendoftreatment

(includ-ing12patientstreatedwithsurgery,4withradiotherapy,andthe

remainingwithacombination ofsurgeryand radiotherapy).All

patientsunderwent aonedaysingle-injectionimaging protocol

includingPET/CTandsubsequentPET/MR.

Briefly,patientswerefastedforatleast6hbeforescanning,

fol-lowedbybloodglucoseassessmentpriortoFDGinjection.Adoseof

406±40MBqof[18F]-FDGwasinjectedbasedonthepatient’sbody

weight.Followinganuptakeperiod(81±15min),patients

under-wentaPET/CTscan.AftercompletionofthePET/CTscan,patients

thenreceivedaheadandneckPET/MRexamination.

2.2. PET/CTacquisition

PET/CTacquisitionwasperformedonaGeminiTF(Philips

Med-icalSystems,Best,TheNetherlands)tomographdesignedwitha

multi-ringLYSOblockdetectorssystemandanominalaxial

reso-lutionnearthecenteroffieldofviewof4.8mm(FWHM)withan

absolutesensitivityof6.6kcps/MBq,aspreviouslyreported[11].

Accordingtoourroutineprotocolforoncologicalstudies,PET

data was acquired in sinogram mode for 15min with

ama-trix size of 144×144. PET data were reconstructed using the

LOR-TF-RAMLAalgorithm,therefore datawaspostfilteredwith

a three-dimensional isotropic gaussian of 4mm at FWHM.

CT-basedattenuationmaps werecreatedwitha low doseCTscan

(120kV,80mA)usingcommonlyemployedbi-linearscaling.The

acquisitiontimewas3minperbedposition(BP),with5\6BPs(each

21cm)coveringthetrunkofthepatientsstartingfromthepelvis

andmoveduptowardthehead.Followingwholebodyacquisition

anadditionalBPwasacquiredfocusingonthehead/neckregion.

Theresultwasatotalacquisitiontimeofapproximately18mins

perpatientforPET/CT.

Thepatientswerepositionedsupinewiththeirarmsbrought

togetherabovetheheadfortotalbodyacquisitionandwiththeir

armsrestingatthesidesforhead/neckacquisition.

2.3. PET/MRAcquisition

PET/MRwasperformedontheBiographmMR(Siemens

Health-care,Erlangen,Germany).Thissystemconsistsofa3TMRIscanner

featuring high-performance gradient systems (45mT/m) and a

slewrateof200T/m/s.ThePET/MRsystemisequippedwithTotal

ImagingMatrixcoiltechnology(SiemensHealthcare),coveringthe

entirebodywithmultipleintegratedradiofrequencysurfacecoils.

Thecoils,patienttable,andcableshavebeenredesignedforPET/MR

inordertominimizetheirattenuationandtoallowunimpairedPET

acquisitionwiththecoilsinplace.AfullyfunctionalPETsystem

equippedwiththeavalanchephotodiodetechnologyisembedded

intothemagneticresonancegantry.

ThePETscannerhasaspatialresolutionof4.1mm(FWHM)at

1cmandof5.0mm(FWHM)at10cmfromthetransverseFOVand

asensitivityof11.72kcps/MBqatthecenteroftheFOV.

Onaverage,thePET/MRscanstartedabout20minafterthestart

ofthePET/CTacquisition.Bedpositionwasestablishedinorderto

getfullcoverageofthehead/neckregion.Aftercorrectpositioning

wasestablished,thecombinedPET/MRacquisitionwasperformed.

First,acoronal2-pointDixon3-dimensionalvolumetric

inter-polatedbreath-holdT1-weightedMRIsequencewasacquiredand

used for the generation of attenuationmaps and for anatomic

allocation of the PET results. The software of theMRI scanner

automaticallygenerates4differentimages:T1-weightedin-phase,

T1-weighted out-of-phase, water-only, and fat-only.

Simulta-neouslywiththestartoftheDixonMRIsequence,PETacquisition

began toensure correct temporal and regional correspondence

betweenMRIandPETdata.ThePETacquisitiontimewas4minand

consideringsimultaneousMRacquisitionthetotalPET/MR

exami-nationlastedabout30min.

2.4. PETdatareconstruction

PETdataobtainedonthePET/CTand PET/MRscannerswere

processed withcomparable reconstructionand correction

algo-rithms. For both modalities, emission data were corrected for

randoms, dead time, scatter, and attenuation. A 3-dimensional

attenuation-weightedordered-subsetsexpectationmaximization

iterativereconstructionalgorithm(AWOSEM3D)wasappliedwith

3iterationsand21 subsets,gaussiansmoothing of4mminfull

widthathalfmaximum,andazoomof1.Attenuationmapswere

obtainedfromtheCT databybilineartransformation,as

imple-mentedinthepost-processingsoftwareofthePET/CTscannerand

wereusedforattenuationcorrectionofthePET/CTdata.

2.5. MRI-basedattenuationcorrection

PETattenuationcorrection(AC)isacriticalissueforintegrated

PET/MRIscanners.Contrarily toX-rayCT, which inherently

pro-videsapatient-specificmeasurementoftheelectrondensity,MRI

signalisonlyrelatedtoprotondensity.Therefore,MRIis

intrinsi-callyunabletomeasureattenuationfactorsduetoelectrondensity.

Proposedsolutionsaresubstantiallybasedonthecombinationof

custom-tailoredMRIacquisitionswithimageprocessing,pattern

(3)

Inthiswork,clinicallyapprovedACisperformedbymeansofthe

attenuationmapsgeneratedonthebasisofthe2-pointDixonMRI

sequencesobtainedforeveryBP.Thisapproachhasrecentlybeen

demonstratedtoprovideresultscomparabletothoseof

conven-tionalattenuationcorrectionbylowdoseCT[12].Theprocedure

hasbeenimplementedinthepost-processingsoftwareofthe

scan-nerandoperatesautomatically.TheDixonfat-andwater-weighted

imageswereused tocreate anattenuationmap with4distinct

tissue-classes:background,lungs,fat,andsofttissue.Attenuation

ofthePETsignalcausedbyinstrumentation,suchasthepatient

bedandthefixedMRIcoils,isautomaticallyintegratedintothe

attenuationmaps[13].

It is worth noting that suchAC approach ignores the

pres-enceofbonystructures,thereforelimitingquantitativeaccuracyin

anatomicaldistrictscontaining(orcloseto)bones[14].Toaddress

thisissue,morerefinedACtechniquesarecurrentlyunder

investi-gationandclinicalvalidation[15].

2.6. MRacquisition

TheMRIprotocolwasperformedwithadedicated16channel

headandneckcoilincluding:coronalSTIR,axialFSET2-weighted,

axial FSET1-weighted,axial diffusion-weightedimaging (DWI),

andasingle-shotechoplanar2DSPAIR.Inthissequencethreeb

valueswereapplied:0,500and800s/mm2.

Perfusionstudiesweredynamicallyacquiredimmediatelyafter

intravenousadministrationofparamagneticcontrast(Magnevist®,

Bayer-Schering,Berlin,Germany)with50measurementsofvolume

InterpolatedGREprecededbytwopre-contrastmeasurementsof

thesamesequenceatavariableflipangle(2◦ and15◦).The

con-trastagentadministration(0.2ml/kgbodyweight)wasdonewith

aflow-rateof3.5ml/s,followedbysalineflush,usingapower

injec-tor.

Inaddition,afterdynamicacquisitions,avolumeInterpolated

GREisotropicathighresolutionandaT1-weightedturbospinecho

withfatsuppressionwereacquired.

Table1 shows technicaldetails of theabove-mentionedMR sequences.

InordertoobtaingoodMRIimagequality,allpatientswere

care-fullyinstructedtorefrainfromvigorousswallowingorbreathing

duringimageacquisitionespeciallyduringtheacquisitionofDWI

sequences.Smallbreaksweremadebetweenindividualsequences

toallowpatientstoclearthethroatofmucoussecretions,cough,

orswallow.

2.7. Dataprocessingandmultiparametricanalysis

PETdataobtainedsimultaneouslywithCTandMRexamination

wereprocessed withcomparable reconstruction and correction

algorithms.Forbothmodalities,emissiondatawerecorrectedfor

randoms,deadtime,scatter,andattenuation.

Tumortissuewasidentifiedasanyvoxelinthe3Ddatasetwith

countsgreaterthana fixedthresholdfractionofthepeak

activ-ityinthetumor.Thethresholdlevelfortumorcharacterization

wasselectedasa StandardizedUptake Value(SUV)equal toor

higher than 3.5 [16]. The SUVs were calculated automatically

by the software (Syngo.via, Siemens Medical Systems) using

the body weight method: SUV=[decay corrected tissue

activ-ity (kBq/ml)]/[injected18F-FDG doseper body weight(kBq/g)].

ThemaximumSUV(SUVmax)andmeanSUV(SUVmean)ofthe

tumortissuewerederivedautomaticallybythesoftwareusinga

voxel-of-interest(VOI)methodoftissuedelineationforboththe

acquisitions.Moreover,basedonSUVvalues,avolumetric

charac-terizationoflesionburdenwasmadeassuggestedinaprevious

study[17]consideringametabolictumorvolume(MTV)witha

thresholdof40%ofthemaximumsignalintensity(MTV40).

Twogroupsofoneseniorradiologistexperiencedinheadand

neckimaging(8and10yearsexperience)andonenuclearmedicine

specialist(9and7yearsexperience)wereblindedtothePET/CT

findings and consensuallyreviewed tumor findings on PET/MR

imagesinconsensus.Primarytumorsizeandinfiltrationof

neigh-boringstructureswereassessed(Fig.1A).Acomparisonbetween

PET/CTandPET/MRfindingshasnotbeenperformedconsidering

thatCTexaminationwaswithoutcontrastagentinjectionandthis

couldbiastheanalysisbyunderestimatingCTdiagnosticaccuracy

(Fig.1B).

Subsequently,theDCE-MRimagesweretransferredfor

post-processing to a workstation running commercially available

softwarefortissueperfusionestimation(Tissue4D,Siemens

Med-icalSystems).Aftermotioncorrectionandregistrationofthe

pre-and post-contrast acquisitions, T1 mapping was automatically

performed and a freehand region-of-interest (ROI)was plotted

aroundthetumorincludingtheneighboringvessels(carotid

arter-ies,jugularvein). Thepharmacokineticmodeling wasbased on

atwo-compartmentmodelthatallowsforthecalculationofthe

following:transferconstantbetweenvascular,extravascular,and

extracellularspace(EES)(Ktrans);thevolumeofEES(

v

e);the

con-stantbetweenEESandbloodplasma(kep);theinitialareaunder

theconcentrationcurve(iAUC)[18].Ktransisaparameterrelated

tovesselpermeabilityandtissuebloodflow.TheleakagespaceVe

isamarkerofcelldensity,thekepisatransferconstantfromthe

extracellular/extravascularspacetoplasma,and iAUC isrelated

tothebloodvolumeinthetissueofinterest[18].Arterial input

function(AIF)wasrelated togadolinium doseinjectedand was

modeledbyabi-exponentialfunctionusinganintermediatemode

providedbythesoftware.For thesubsequenttumorROI

analy-sis,PET/MRdatasets(PETacquisition,axialT1post-contrastand

axialT2sequences,axialADCmap,andsingleperfusionmapsfor

Ktrans,

v

e,kep andiAUC)wereevaluatedintoaunified

measure-mentsframeworkcustomizedintotheSyngo.viasoftwareplatform

(SiemensMedicalSolutions)allowingavisualcomparisonofthe

multiparametric data.Post-contrast T1-weightedas wellas

T2-weightedimageswereutilizedtodrivethetumoroutlineinmajor

diameterlesionslice,whichavoidsnecroticareasandlarge

feed-ingvessels.Therefore,ROIareavalue,referredasthetumorsize

inthemajordiameterlesionslicewasextractedforeachpatient.

Formultiparametriccomparisons,theROIoutlinewasdrawnwith

thesamepositionandextentoneachmaptoautomaticallyextract

maximum and mean values for each parameter (SUVmax and

SUVmeanforthePET; ADCmaxandADCmeanforthediffusion;

Ktrans,k

ep,

v

e,andiAUC,respectivelymaxandmeanforperfusion

maps)(Fig.1A).

Table1

SummaryofMRacquisitionprotocolandsequences.

Sequencename Orientation Repetitiontime(ms) Echotime(ms) Flipangle(◦) Voxelsize(mm3) Acquisitiontime(min:s)

STIR Coronal 5000 84 125 1.1×0.8×3.5 2:02

FSET2-w Axial 5000 117 90 0.7×0.7×3.0 3:10

FSET1-w Axial 590 9.9 150 0.9×0.6×3.0 4:32

DWI Axial 13,800 70 2.8×2.8×4.0 2:11

VolumeInterpolatedGRE Axial 5.37 1.78 15 1.7×1.2×3.6 0:10

VolumeInterpolatedGREISO Axial 4.93 2.29 9 1×1×1 4:41

(4)

Fig.1.InA,ROIplacementformultiparametricanalysisina70y.o.malewithHNSCClesioninprimarystaging.ROI(inred)wasmanuallyoutlinedonpost-contrast T1-weighted(T1+C)andT2-weighted(T2)imagesinmajordiameterlesionslice,andfollowingdrawnwithsamepositionandextentoneachmap.InB,acoronalviewofthe samepatientfrombothPET/MRandPET/CTacquisitions.

Moreover, standardized ROIs (1cm2) were placed in the

paraspinalmuscleofeach patientin order toobtainSUV,ADC, andDCE-MRIestimatesinthenormaltissueandprovidesignificant differencescomparedwiththetumortissue.

2.8. Statisticalanalysis

Multiparametricdata wereexportedand analyzedusingthe Sigmaplotv10.0programwithSigma-Statintegrationv3.2(SPSS,

(5)

Erkrath,Germany).Ap-valueoflessthan0.05wasconsidered sta-tisticallysignificant.Thevaluesarepresentedasmean±standard deviation(SD).

TheKolmogorov–Smirnovtestwasfirstusedtoexamineifthe datawasnormallydistributed.Cohen’skappawasusedto eval-uateagreementforprimarytumordetectioninPET/MRbetween thetwogroupsofobservers,withoutaccountingforlymphnodes involvementandputativedistantmetastases.Anydifferencesin theparametersestimationsbetweenthenormalandtumortissue andbetweentheprimarytumorsandrecurrentlesionswere deter-minedwitha t-studenttest. Spearman’scorrelationcoefficients werecalculatedtocompareMRIandCTbasedSUVvaluesandto comparemetabolicdiffusionandperfusionparameters.A correla-tioncoefficientof≤0.35wasconsideredlow,0.36–0.67moderate, 0.68–0.90high,and≥0.90tobeexcellentcorrelations[19].

3. Results

3.1. Feasibilityandinterobserveragreement

AllPET/CTandPET/MRstudieswereacquiredsuccessfullyand

weresuitableforfurtherevaluation.

Satisfactoryagreementbetweenthetwoobservergroupswas

foundinanatomicallocationoflesionsbyPET/MR(T-staging)with

nosignificantdifferencesbetweenthetwoclinicallydefinedgroups

(Primarytumors:97%,Cohen’skappa0.93;Recurrenttumors:93%,

Cohen’skappa0.89).

3.2. Multiparametriccomparisonintheoverallsampleand

subgroupsdefinition

ThesummarystatisticsfortheFDG-metabolism,diffusion,and

perfusion measurements in the healthy muscle tissue and the

tumorsaredemonstratedinTable2.

Representativecasesofmultiparametrichead–neckimagingare

showninFigs.2and3.

Whengroupingallpatientstogether(n=44)therewasa

signifi-cantpositivecorrelationsbetweentheROIarea,theSUVmeasures,

and the SUV estimations. Conversely, there was a significant

negativecorrelationbetweenMRSUVROI,andADCMean.There

wasalsoasignificantnegativecorrelationbetweenADCMeanand

KtransMean(Table3).

Inparticular,thereweretwosubgroupsidentifiedforpatients

inprimaryorrecurrenttumors.Primarytumors(n=22)included

11laryngeal,3oral,1sinonasal,2oropharingeal,3parotid,and2

rhinopharingealcancers.

Recurrenttumors(n=22)included14casesoflaryngeal,3oral,

2sinonasal,1parotid,and2rhinopharingealcancers.Inparticular,

11patientsreceivedonlysurgicaltreatment,4radiotherapy,and7

bothchemo-andradiotherapy.

3.3. Multiparametriccomparisoninprimarytumors

Asforasprimarytumorsthecorrelationanalysisdemonstrated

asignificantcorrelationbetweentheROIarea,theSUVmeasures,

andtheSUVestimations.

Conversely,therewasasignificantnegativecorrelationbetween

ADCMeanandKtransMean(Fig.2;Table3).

3.4. Multiparametriccomparisoninrecurrenttumors

Asforasrecurrenttumors,theSpearman’scoefficientanalysis

amongthedifferentparametersinthetumorsitesdemonstrateda

significantcorrelationbetweentheROIareaandtheMRSUVROI.

However,thecorrelationwasnotsignificantwithCTSUVVOIorMR

SUVVOI(p=0.35and.23,respectively),whichwasdemonstratedin

primarytumors.Asignificantcorrelationwasalsofoundbetween

theSUVmeasures,MRSUVVOI,andMRSUVROI.Finally,a

signifi-cantnegativecorrelationwasfoundbetweenMRSUVROIandADC

mean(Fig.3;Table3).

No significant differences were detected between the two

groupsformetabolic,diffusionandperfusionparameters.SUV

esti-mationandallperfusionparameters(Ktrans,k

ep,

v

e,andiAUC)in

tumorsweresignificantlyhighercomparedtonormalmuscletissue

inboththegroups(Primarytumors:p<0.001;Recurrenttumors

p<0.001)(Figs.1and2,Table2).Conversely,theADCmeanvalue

intumorswassignificantlylowercomparedtonormalparaspinal

Table2

Summaryofmetabolic,diffusionandperfusionparametersinboththegroups.

Parameter Overallsample Primarytumors Recurrenttumors Muscle

Mean SD Mean SD Mean SD Mean SD

AGE 60.94 2.23 62.52 12.35 59.36 12.93

AREA–Tumorsize (cm2) 2.73 0.37 2.99 4.90 2.47 1.73 1 0 CTSUVMean 5.77 1.76 4.53 4.53 7.01 5.75 0.715** 0.065 CTMTV (cm3) 14.42 1.62 15.56 19.48 13.27 14.24 MRSUVMean 7.78 2.39 6.09 5.05 9.47 8.91 0.721** 0.077 MRMTV (cm3) 11.92 1.58 13.03 16.59 10.81 11.78 SUVMean2D 20.36 1.35 19.40 24.10 21.32 17.66 0.724** 0.079 ADCMean (␮m2/s) 1193.26 0.14 1193.14 481.68 1193.4 187.09 1483.58* 108.2

KtransMean(min−1)

(×1000) 268.09 13.60 277.71 167.29 258.48 150.82 51.18* 10.36 kepMEAN (min−1)(×100) 83.99 5.98 88.22 62.30 79.76 31.73 51.59* 14.36 iAUCMean (mMmin)(×50) 901.54 17.73 914.08 472.24 889.01 539.96 166.21* 30.27 veMean (×1000) 326.35 25.84 308.08 124.99 344.62 133.11 103.52* 22.03

Abbreviations:CT:computedtomography;SUV:standardizeduptakevalue;MR:magneticresonance;MTV:metabolictumorvolume;ADC:apparentdiffusioncoefficient; Ktrans:transferconstantbetweenvascularandEES;kep:constantbetweenEESandbloodplasma;iAUC:initialareaundertheconcentrationcurvein60s;ve:thevolumeof

EESperunitofvolumeoftissue.

*p0.05; **p0.01.

(6)

Fig.2.Multiparametricevaluationofmorphological(T1,andT1+C),metabolic(PET)andfunctional(DWI,ADC,iAUC,ve,kep,andKtrans)ina57y.omalewithrightvocalcord

HNSCCandanteriorcommissureinfiltrationinprimarystaging.Theincreaseofthicknessandenhancementofthetumorsiteinthemorphologicalacquisitions,theincrease ofFDG-uptake,andthereduceddiffusivityinADCwithanincreaseofperfusionparametersweredetected.

Fig.3. Multiparametricevaluationofmorphological(T1,andT1+C),metabolic(PET)andfunctional(DWI,ADC,iAUC,ve,kep,andKtrans)ina69y.o.malerecurrencyafter

surgeryforaHNSCCofthelarynx.Theincreaseofthicknessandenhancementofthetumorsiteinthemorphologicalacquisitions,theincreaseofFDG-uptake,andthe reduceddiffusivityinADCwithanincreaseofperfusionparametersweredetected.

(7)

Table3

Significantcorrelationsinmultiparametricanalysis.

Parameters Spearman’s

coefficient

Significance(p value) Overallsample

ROIareavs.CTSUVVOI 0.66 <0.001

ROIareavs.MRSUVVOI 0.61 <0.001

ROIareavs.MRSUVROI 0.61 <0.001

CTSUVVOIvs.MRSUVVOI 0.97 <0.001

CTSUVVOIvs.MRSUVROI 0.88 <0.001

MRSUVVOIvs.MRSUVROI 0.91 <0.001

CTMTVvs.MRMTV 0.94 <0.001

MRSUVROIvs.ADCMean −0.36 0.04

ADCMeanvs.KtransMean −0.5 <0.001

Primarytumors

ROIareavs.CTSUVVOI 0.51 0.04

ROIareavs.MRSUVVOI 0.45 0.04

ROIareavs.MRSUVROI 0.53 0.01

CTSUVVOIvs.MRSUVVOI 0.95 <0.001

CTSUVVOIvs.MRSUVROI 0.81 <0.001

MRSUVVOIvs.MRSUVROI 0.83 <0.001

CTMTVvs.MRMTV 0.93 <0.001

ADCMeanvs.KtransMean −0.42 0.04

Recurrenttumors

ROIareavs.MRSUVROI 0.61 0.04

MRSUVVOIvs.MRSUVROI 0.99 0.01

MRSUVROIvs.ADCmean −0.72 0.01

Abbreviations:ROIarea:tumorsizein2Dmajorlesionsize;Abbreviations:CT: computedtomography;SUV:standardizeduptakevalue;MTV:metabolictumor volume;ADC:apparentdiffusioncoefficient;Ktrans:transferconstantbetween vas-cularandEES.

muscle(Primarytumors:p<0.001;Recurrenttumors:p<0.001) (Table2).

4. Discussion

ThisstudydemonstratedthefeasibilityofPET/MRexamination

fortheevaluationoftheT-stadiuminpatientswithheadandneck

cancerduringstagingorfollow-up.ThisstudyalsofoundPET/MR

examinationtohaveahighinter-observeragreementforlesion

detectionandsignificantrelationshipsamongseveralmetabolic,

diffusion,andperfusionparametersintheprimarytumor.

InthepastthreedecadesPEThasdemonstrateditsvaluein

pro-vidingnon-invasiveinformationoftissueatthemolecularlevel.

The valueof PET lies in itshigh sensitivity tracking of in vivo

biomarkers, however it lacks precise anatomical information,

whichhasbeenovercomewiththeintroductionofthePET/CT

scan-ners.18F-labeledfluorodeoxyglucosepositronemission

tomogra-phy([18F]FDGPET)iscommonlyusedinHNSCCfortumorstaging,

monitoringoftreatmentresponses,detectionofrecurrences,and

radiotherapyplanning[5].AlthoughPET/CTiscurrentlythegold

standardforoncologicalimaging,severallimitationsexist,which

include:substantialamountsofionizingradiationfromrepeated

PET/CTscansduringdiagnosticandfollow-upexaminations;the

anatomicalcontributionofCTislimitedinsofttissuesandareas

withcomplexanatomy(i.e.theheadandneck&pelvisregion).

Theseissuescanbeovercomewithrecentlylaunchedfully

inte-gratedhybridPET/MRscanners,whichcanperformsimultaneous

acquisitionofPETandMRI.BycombininganatomicalMRI,DW-MRI,

DCE-MRIandPETimagingtoobtainconcertedinformationabout

cellularity,perfusionandbiologicalactivityofthetumor(i.e.byuse

ofFDG,FLTorCholine),wehypothesizethatthesensitivityandthe

specificityinoncologystagingcanbeimproved.However,currently

nospecificclinicalindicationforPET/MRhasbeenestablished.

PET/CTand PET/MRstudieswereacquiredin all44 patients

without side effects and were suitable for further evaluation.

Althoughseveralstudieshavedemonstratedthefeasibilityofthis

new powerful tool in oncological imaging, the head and neck

regionincludescomplexanatomywithmanydifferenttissuesin

asmallregionandisoftenaffectedbyinvoluntarymotionartifacts

duringimagingacquisition.Theabilitytoperformsimultaneous

acquisition decreases the acquisition time bypassing putative

co-registrationproblems,especiallyforhigh-temporalresolution

phenomenalikeperfusion.

Indeed,softwarefusionofPETandMRIdataischallenginginthe

neckandthesimultaneousacquisitionoftheMRIandPETdata

pro-videsimprovedcoregistration.ThefeasibilityofPET/MRinthehead

andneckregionhasalsobeenreportedbyothergroups,

demon-stratingagoodMRimagequalitywithoutimpairmentofthePET

signal[20,21].

ThehighcontrastresolutionofMRIcomparedtoCTcreateda

morefinedifferentiationofheadandnecktissues,whichprovideda

correctstagingandfollow-upofHNSCC,mainlyforTstadium

char-acterization[10].Arecentarticlethatexaminedthirtypatientswith

HNCcanceratprimarytumorsfoundahigherdiagnosticaccuracy

forTevaluationsfromretrospectivelyfusedPET/MRcomparedto

PET/CTandacomparableaccuracyforN-staging[22].Ourcurrent

studyalsosupportsthispreviousreportbydemonstratingexcellent

agreementbetweenthetwoobservergroupsfortheanatomic

allo-cationoflesionsbyPET/MR(T-staging).Thishasalsobeenfound

inotherstudies,examiningdifferenttypesofcancer(lymphnodes

anddistantmetastases)revealinginter-readeragreementbetween

observersofabout0.5[23]and0.7[24].

Asfor the correlationanalysis, there wasa significant

posi-tiverelationshipbetweentheROIarea(tumorsizeinthemajor

axiallesion diameter)and theSUV.Thiscorrelationshowsthat

biggerlesionshavehighermetabolicactivity,althoughthe

pres-enceofnecroticintralesionalregionscanbiasSUVestimation.In

this study, we limit this bias performinga manual tumor

out-linein majordiameterlesion slice, avoiding necrotic areasand

largefeedingvessels.Althoughtherearecontroversialresultswith

respecttothecorrelationbetweenthelesionsizeandSUV

measure-ment[25,26]thedifferencesfoundbetweenprimaryandrecurrent

tumors couldbealsodue totheeffect of differenttherapies on

lesionhomogeneity.Fruehwald-Pallamaretal.[26] reportedno

significantdifferences in ADCvalues andSUVmax betweenthe

variousTstagesin31patients.Thepossibleinclusionof

inflamma-tion,fibrosisand/ornecrosiswithintheROIlesionareameasured

alongthemajoraxisisalikelyconsequenceofsurgery,

radiothe-rapy,chemotherapyordifferentcombinedtherapeuticapproaches

duringthefollow-up.Forthesereasons,inordertoreduceeffects

related to treatments, 6–8 weeks from the end of treatment

passedbeforetheimagingstudyofrecurrentlesions.Alsoprevious

reportsandthecurrentstudyhavefoundcomparableestimatesof

metabolicparameters(SUVandMTV)betweenPET/CTandPET/MR

mainly in primary tumors and theoverall sample. In a recent

report[23],averyhighcorrelationbetweentheSUVvaluein

18F-FDG-PET/MRand18F-FDG-PET/CTimaginghasbeendetectedfor

primarytumors,lymphnode,anddistantmetastasesin14patients.

Otherauthors[12,24]foundnostatisticallysignificantdifferencein

diagnosticcapabilitybetweenPET/CTandPET/MRin17patients.

AlthoughMRI-basedACisstillunderdebate[14],2-pointDixon

approachforMRIattenuationhasrecentlyprovidedresults

com-parable tothose of conventional attenuationcorrectionby low

doseCT[12].HighcorrelationbetweenSUVestimatedbyPET/CT

andPET/MRresultingfromthisstudysupportstheseconclusions,

accordingtoapreviouswork[14].Inthecurrentstudywhen

exam-iningmetabolic(SUV),diffusion(ADC)andperfusionparameters

(Ktrans,k

ep,iAUC,

v

e),wedemonstratedasignificantnegative

corre-lationbetweenMRSUVROIandADCMeaninrecurrenttumorsand

theoverallsample.Althoughthecelldensity(estimatedbyADC)

andglucosemetabolism(estimatedbySUV)ofatumoraredifferent

physiologicalparameters,onerecentreporthasshownapositive

(8)

[26,28]haveobservednosignificantcorrelationbetweenthesetwo

parameterswithMRIandPET/CT.Thenegativecorrelationresulting

fromthisstudyshowsthathighercellularityofthelesions,

reflect-ingonalowADC,parallelswithanincreasedglucoseuptakeand

cellularmetabolism.Controversialresults[26–28],instead,canbe

duetotumorinhomogeneityandcontextualnecroticareas,where

thefreediffusionofwaterprotonsresultsinnoADCrestriction[26].

Inaddition,therewasasignificantnegativecorrelationbetween

ADCMean and Ktrans Mean in primary tumors and the overall

sample. Although few studies have investigated the

relation-ship among these parameters, the resultsappear controversial

[26–28].AninversecorrelationbetweenADCandKtranshasalso

beendetectedfor othertumors(i.e.brain)[27],whichsuggests

anintricaterelationshipbetweenvascularpermeability andthe

tumormicroenvironment.Inthis context, twotumoral features

mayexplainthenegativecorrelationbetweentheseparameters.

Fromoneside,tumorproliferationincreasescelldensityandADC

restrictionwhile,fromtheotherhand,neoangiogenesisincreases

immature tumoral vessels and, thereby, vascular permeability

expressedbyKtrans.

Finally,perfusionparameterssuchasKtranshavebeen

demon-stratedtobepotentiallypredictiveofpatientresponsetotherapy,

howevercontroversialresultscanbedebated[29].Anydifferences

maybeattributedtotheappliedpharmacokineticmodelling(e.g.

(extended)Toftsmodelormoreelaboratedmodelslikethe

shutter-speedone)[18].Inanycase,Ktransseemstobesuitabletomonitor

therapy(relative)changesintumorsthoughthecorrectnessofthe

absolutevaluesmaybenotalwaysclaimed[30].

5. Conclusions

OurstudydemonstratesthefeasibilityofPET/MRimaging in

theT-stadiumevaluationofHNClesionsduringstagingor

follow-up.Thisintroducesamorecompletecharacterizationofmetabolic,

diffusion,andperfusioncharacteristicsofprimarylesions.

The strength and the originality of this study exists in the

involvementofmultiparametricMRI,whichincludebothdiffusion

andperfusionevaluationsthatarecollectedsimultaneouslyduring

PET/MRacquisition.

Although further improvements should be performed to

increasethevalueofhybridacquisitionofPET/MR(improved

atten-uationcorrectionalgorithmfor bone segmentation,and surface

dedicatedcoilsforheadandneckregion),webelievethis

analy-sismayplayacrucialroleinprognosisandtreatmentofpatients

whenconsideringthepredictivevalueoftheseparameters[29].

Conflictofinterest

Allauthorshavenoconflictsofinterestandnodisclosuresof

financialinteresttoreport.

References

[1]JansenJFA,KoutchgerJA,Shukla-DaveA.Non-invasiveimagingofangiogenesis inheadandnecksquamouscellcarcinoma.Angiogenesis2010;13(2):149–60. [2]ShangJ,GuJ,HanQ,XuY,YuX,WangK.Chemoradiotherapyissuperior toradiotherapyaloneaftersurgeryinadvancedsquamouscellcarcinomaof theheadandneck:asystematicreviewandmeta-analysis.IntJClinExpMed 2014;7:2478–87.

[3]SchoderH,FuryM,LeeN,KrausD.PETmonitoringoftherapyresponseinhead andnecksquamouscellcarcinoma.JNuclMed2009;50:74S–88S.

[4]CooperJS,PajakTF,ForastiereAA,JacobsJ,CampbellBH,SaxmanSB,etal. Post-operativeconcurrentradiotherapyandchemotherapyforhigh-risk squamous-cellcarcinomaoftheheadandneck.NEnglJMed2004;350:1937–44. [5]BernierJ, DomengeC, OzsahinM,Matuszewska K,LefèbvreJ-L, Greiner,

RH,etal.Postoperativeirradiationwithorwithoutconcomitant chemother-apyforlocallyadvancedhead andneckcancer. NEnglJMed 2004;350: 1945–52.

[6]VandecaveyeV,DirixP,DeKeyzerF,deBeeckKO,VanderPoortenV,RoebbenI, etal.Predictivevalueofdiffusionweightedmagneticresonanceimagingduring chemoradiotherapyforheadandnecksquamouscellcarcinoma.EurRadiol 2010;20(7):1703–14.

[7]ChawlaS,KimS,LoevnerLA,HwangWT,WeinsteinG,ChalianA,etal.Prediction ofdisease-freesurvivalinpatientswithsquamouscellcarcinomasofthehead andneckusingdynamiccontrastenhancedMRImaging.AJNRAmJNeuroradiol 2011;32(4):778–84.

[8]PartoviS,KohanA,RubbertC,Vercher-ConejeroJL,GaetaC,YuhR,etal.Clinical oncologicapplicationsofPET/MRI:anewhorizon.AmJNuclMedMolImaging 2014;4(2):202–12.

[9]PartoviS,RobbinMR,SteinbachOC,KohanA,RubbertC,Vercher-ConejeroJL, etal.InitialexperienceofMR/PETinaclinicalcancercenter.JMagnResImaging 2014;39:768–80.

[10]BuchbenderC,HeusnerTA,LauensteinTC,BochischA,AntochG.Oncologic PET/MRI.Part1:tumorsofthebrain,headandneck,chest,abdomenandpelvis. JNuclMed2012;53:928–38.

[11]SurtiS,KuhnA,WernerME,PerkinsAE,KolthammerJ,KarpJS.Performance ofPhilipsGeminiTFPET/CTscannerwithspecialconsiderationforits time-of-flightimagingcapabilities.JNuclMed2007;48(3):471–80.

[12]SeemannMD.Whole-bodyPET/MRI:thefutureinoncologicalimaging.Technol CancerResTreat2005;4:577–82.

[13]DelsoG,FurstS,JackobyB,LadebeckR,GanterC,NekollaSG,etal.Performance measurementsoftheSiemensmMRintegratedwhole-bodyPET/MRscanner.J NuclMed2011;52(12):1914–22.

[14]VaroquauxA,RagerO,PoncetA,DelattreBM,RatibO,BeckerCD,etal.Detection andquantificationoffocaluptakeinheadandnecktumours:18F-FDGPET/MR versusPET/CT.EurJNuclMedMolImaging2014;41(3):462–75.

[15]ArabiH,RagerO,AlemA,VaroquauxA,BeckerM,ZaidiH.Clinicalassessmentof MR-guided3-classand4-classattenuationcorrectioninPET/MR.MolImaging Biol2014:1–13.

[16]BaekCH,ChungMK,SonYI,ChoiJY,KimHJ,YimYJ,etal.Tumorvolume assess-mentby18F-FDGPET/CTinpatientswithoralcavitycancerwithdentalartifacts onCTorMRimages.JNuclMed2008;49:1422–8.

[17]ErdiYE,MawlawiO,LarsonSM,ImbriacoM,YeungH,FinnR,etal. Segmenta-tionoflunglesionvolumebyadaptivepositronemissiontomographyimage thresholding.Cancer1997;80:2505–9.

[18]ToftsPS,KermodeAG.Measurementoftheblood-brainbarrierpermeability andleakagespaceusingdynamicMRimaging1.Fundamentalconcepts.Magn ResonMed1991;17:357–67.

[19]PaceL,NicolaiE,Luongo,A,AielloM,CatalanoOA,SoricelliA,etal.Comparison ofwhole-bodyPET/CTandPET/MRIinbreastcancerpatients:lesionsdetection andquantificationof18F-deoxyglucoseuptakeinlesionsandinnormalorgan tissues.EurJRadiol2014;83:289–96.

[20]PlatzekI,Beuthien-BaumannB,SchneiderM,GudziolV,LangnerJ,SchrammG, etal.PET/MRIinheadandneckcancer:initialexperience.EurJNuclMedMol Imaging2013;40:6–11.

[21]BossA,SteggerL,BisdasS,KolbA,SchwenzerN,PfisterM,etal.Feasibility ofsimultaneousPET/MRimagingintheheadandupperneckarea.EurRadiol 2011;21:1439–46.

[22]KanadaT,KitajimaK,SuenagaY,KonishiJ,SasakiR,MorimotoK,etal.Valueof retrospectiveimagefusionof18F-FDGPETandMRIforpreoperativestagingof headandneckcancer:comparisonwithPET/CTandcontrast-enhancedneck MRI.EurJRadiol2013;82:2005–10.

[23]PatroviS,KohanA,Vercher-ConejeroJL,RubbertC,MargeviciusS,Schluchter MD,etal.Qualitativeandquantitativeperformanceof18F-FDG-PET/MRIversus 18F-FDG-PET/CTinpatientswithheadandneckcancer.AJNRAmJNeuroradiol 2014;35(10):1970–5.

[24]KubiessaK,PurzS,GawlitzaM,KuhnA,FuchsJ,SteinhoffKG,etal.Initial clinicalresultsofsimultaneous18F-FDGPET/MRIincomparisonto18F-FDG PET/CTinpatienswithheadandneckcancer.EurJNuclMedMolImaging 2014;41:639–48.

[25]KhalafM,Abdel-NabiH,BakerJ,ShaoY,LamonicaD,GonaJ.Relationbetween nodulesizeand18F-FDG-PETSUVformalignantandbenignpulmonary nod-ules.JHematolOncol2008;1:13,http://dx.doi.org/10.1186/1756-8722-1-13. [26]Fruehwald-Pallamar J, Czerny C, Mayerhoefer ME, Halpern BS,

Heder-CzembirekC,BrunnerM,etal.Functionalimaginginheadandnecksquamous cellcarcinoma:correlationofPET/CTanddiffusion-weightedimagingat3 Tesla.EurJNuclMedMolImaging2011;38(6):1009–19.

[27]ChuJP,MakHK,YauKK,ZhangL,TsangJ,ChanQ,etal.Pilotstudyon evalua-tionofanycorrelationbetweenMRperfusion(Ktrans)anddiffusion(apparent diffusioncoefficient)parametersinbraintumorsat3Tesla.CancerImaging 2012;12:1–6,http://dx.doi.org/10.1102/1470-7330.2012.0001.

[28]VaroquauxA,RagerO,LovbladK-O,MastersonK,DulquerovP,RatibO,etal. Functionalimagingofheadandnecksquamouscellcarcinomawith diffusion-weightedMRIandFDGPET/CT:quantitativeanalysisofADCandSUV.EurJ NuclMedMolImaging2013;40(6):842–52.

[29]BernsteinJM,HomerJJ,WestCM.Dynamiccontrast-enhancedmagnetic res-onance imagingbiomarkersin headand neckcancer:potential to guide treatment?Asystematicreview.Oraloncology2014Oct;50(10):963–70. [30]BisdasS,SmrdelU,BajrovicFF,Surlan-PopovicK.Assessmentof

progression-free-survivalinglioblastomasbyintratreatmentdynamiccontrast-enhanced MRI. Clin Neuroradiol 2014;(August), http://dx.doi.org/10.1007/s00062-014-0328-0.

Riferimenti

Documenti correlati

measures being used in the study. The analysis of the data revealed several measures at both individual and organizational levels. All the identified measures were captured under

SPECT evaluation of stroke is a promising technique in that it images physiology and thus may eventually serve as a surrogate for the triage of stroke victims concerning

In most patients with squamous cell carcinoma of the head and neck (HNSCC) a staging PET examination will be done after having obtained cytological/ his- tological proof of

Axial T2-weighted image demonstrates high het- erogeneous signal from the mass lesion in the right parotid gland

In diagnostic nuclear medicine, radioactive substances, termed radiophar- maceuticals, are administered to patients and an image is obtained by placing the patient under a

Anatomic imaging evaluation with head and neck computed tomography (CT) and/or magnetic resonance imaging (MRI) with intravenous contrast is often utilized either prior to

2[F]-Fluoro-2- deoxy-D-glucose positron emission tomography is a sensitive tool for the detection of occult primary cancer (carcinoma of unknown primary syndrome) with head and

1. The role of Ga-67 scintigraphy in evaluating the results of therapy of lymphoma patients. Kostakoglu L, Leonard JP, Kuji I, et al. Comparison of fluorine-18 fluorodeoxyglucose