Availableonlineatwww.sciencedirect.com
ScienceDirect
EuropeanJournalofRadiologyOpen2(2015)66–70
Comparing
hepatic
2D
and
3D
magnetic
resonance
elastography
methods
in
a
clinical
setting
–
Initial
experiences
Mikael
F.
Forsgren
a,c,b,∗,
Bengt
Norén
c,d,
Johan
Kihlberg
c,d,
Olof
Dahlqvist
Leinhard
b,c,
Stergios
Kechagias
e,
Peter
Lundberg
c,baWolframMathCoreAB,Linköping,Sweden
bRadiationPhysics,DivisionofRadiologicalSciences,DepartmentofMedicalandHealthSciences(IMH),LinköpingUniversity,SE-58183Linköping,Sweden cCenterforMedicalImageScienceandVisualization(CMIV),LinköpingUniversity,SE-58185Linköping,Sweden
dRadiology,DivisionofRadiologicalSciences,DepartmentofMedicalandHealthSciences(IMH),LinköpingUniversity,SE-58183Linköping,Sweden eDivisionofCardiovascularMedicine,DepartmentofMedicalandHealthSciences(IMH),LinköpingUniversity,SE-58183Linköping,Sweden
Received26February2015;receivedinrevisedform13April2015;accepted15April2015 Availableonline28April2015
Abstract
Purpose: Continuousmonitoringofliverfibrosisprogressioninpatientsisnotfeasiblewiththecurrentdiagnosticgoldenstandard(needlebiopsy).
Recently,magneticresonanceelastography(MRE)hasemergedasapromisingmethodforsuchcontinuousmonitoring.Sincetherearedifferent MREmethodsthatcouldbeusedinaclinicalsettingthereisaneedtoinvestigatewhethermeasurementsproducedbytheseMREmethods arecomparable.Hence,thepurposeofthispilotstudywastoevaluatewhetherthemeasurementsoftheviscoelasticpropertiesproducedby2D (stiffness)and3D(elasticityand‘Gabs,Elastic’)MREarecomparable.
Materialsandmethods: SevenpatientswithdiffuseorsuspectdiffuseliverdiseasewereexaminedinthesamedaywiththetwoMREmethods.2D
MREwasperformedusinganacousticpassivetransducer,witha1.5TGE450WMRsystem.3DMREwasperformedusinganelectromagnetic activetransducer,witha1.5TPhilipsAchievaMRsystem.Finally,meanviscoelasticvalueswereextractedfromthesameanatomicalregionfor bothmethodsbyanexperiencedradiologist.
Results: Stiffnesscorrelatedwellwiththeelasticity,R2=0.96(P<0.001;slope=1.08,intercept=0.61kPa),aswellaswith‘G
abs,Elastic’R2=0.96
(P<0.001;slope=0.95,intercept=0.28kPa).
Conclusion: ThispilotstudyshowsthatdifferentMREmethodscanproducecomparablemeasurementsoftheviscoelasticpropertiesofthe
liver.Theexistenceofsuchcomparablemeasurementsisimportant,bothfromaclinicalaswellasaresearchperspective,sinceitallowsfor equipment-independentmonitoringofdiseaseprogression.
©2015TheAuthors.PublishedbyElsevierLtd.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/ licenses/by-nc-nd/4.0/).
Keywords: Liver;Rheology;Elastography;Fibrosis;MRE;MRI
1. Introduction
Thelong-term prognosisof chronicliverdiseases,caused, for example, by alcohol, viral hepatitis, non-alcoholic fatty liver disease (NAFLD), andautoimmune or metabolic disor-ders,dependsprincipallyontheextentandprogressionofliver
∗Corresponding author at: Linköpings Universitet, Avdför radiologiska
vetenskaper,Institutionenförmedicinochhälsa,SE-58183Linköping,Sweden. Tel.:+460101032790.
E-mailaddress:mikael.f.forsgren@liu.se(M.F.Forsgren).
fibrosis.Histopathologicalexaminationofaliverbiopsy, con-ventionally consideredto bethe goldstandard for evaluating hepaticfibrosis,hasseveraldrawbacks.Theseincludetherisk ofcomplications,inter-andintra-observervariability,inaccurate stagingduetospatialsamplingerrorsandthefactthat hetero-geneousdistributionandrateof fibrosisprogression, whichis commonly not constant over time, may not be reflected ina singlebiopsy[1–7].
Animalmodelsaswellasdataonhumanliverdiseasehave demonstratedthatfibrosis,andevencirrhosis,maybereversible. Theseobservationshavestimulatedeffortstofindnon-invasive http://dx.doi.org/10.1016/j.ejro.2015.04.001
2352-0477/© 2015TheAuthors.Published byElsevierLtd.This is anopen accessarticleunder theCCBY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/).
alternativesallowingaclosemonitoringofpatientsand facili-tatingclinicaldecision-making[8,9].
Serum biomarkers,ultrasound elastography anda number ofmagneticresonance(MR)applicationshavebeenproposed toreplaceliverbiopsy,eitherassinglemethods,orinvarious combinations.AmongtheMRtechniques,magneticresonance elastography(MRE)appearstobereliableinidentifying signif-icantfibrosis(stage≥F2)andAUROCvalues>0.90havebeen reported[10–12].
MREisaphase-contrast-basedMRIimagingtechnique.In principleitconsistsofthreekeysteps:(i)mechanicalmotion(or shearwaves)isappliedtothetissue,eitherfromanexternaloran internalsourcesuchasheartmotion.(ii)Thetissueresponseto thestressofthismotionisimagedusingphase-contrastMRIwith motionencodinggradients.(iii)Theimagedataareprocessed toobtaininformationaboutviscoelasticpropertiesoftheliver [13,14].
InhepaticMRE,varioustypesofexternaldrivershavebeen usedtoinducethemechanicalwaves;acoustic,piezoelectricor pneumatic[14].Inthisstudy,twosuchdesignsareused,acoustic andelectromagnetic.Fortheimagingtechniquesboth2Dand3D methodshavebeenproposedandthepresentlyavailable com-mercialsystemusesthe2DMREdatasamplingtechnique.The useofeither2Dor3Daffectstheabilitytousemoreorless com-plexpost-processingalgorithmstoderivethemechanical proper-tiesofthetissue.The3Dsamplingtechnique,whichpresentlyis notcommerciallyavailable,allowsforseparationofthecomplex shearmodulusintotwobasiccomponents,elasticityand viscos-ity,whereasthecommerciallyavailable2DMREpresentsthe shearstiffness.Furthermorebyusinga2Dacquisitiontechnique thereisalsoanimplicitassumptionthattheshearwavesinduced bythedriveronlypropagateintheselectedimagingslice,which isnotthecaseinthe3Dtechnique,wherethealgorithmsolves thewavepropagationinallthreedimensions[13].
SincetherearedifferentMREmethodsthatcouldbeusedin aclinicalsettingthereisaneedtoinvestigateifthe measure-mentsoftheviscoelasticpropertiesoftheliverproducedbythese differentMREmethodsare comparable.Thus,the purposeof thispilotstudywastocompare2Dand3DMRE,usinga com-merciallyavailable 2DMRE systemanda3DMRE research system,withrespecttoliverstiffnessandelasticityinpatients withdiffuseorsuspecteddiffuseliverdisease.
2. Materialsandmethods
2.1. Patients
Inthisstudy,sevenpatientswereexaminedinthecourseof oneday(2012).Thepatientswereseparatelyrecruitedfroman on-goingstudy[15].Thesepatientswereselectedduetotheir elevatedserumalanineaminotransferase(ALT)and/oralkaline phosphatase(ALP)levels.Physicalexaminationandlaboratory testsrevealed no signsof liver cirrhosis. The biopsyand the histopathologicalgradingofthesepatientsweregainedfromthe recordsofthepreviouslymentionedon-goingstudy[15].Table1 presentsbasicdescriptiveparametersaswellasthetimebetween theinitialbiopsyandtheMREexaminations(”Biopsyage”in Table1).ThetimebetweeneachMREacquisition(2Dand3D MRE)wasdependentonhowlongittookforthepatienttomove betweenthetwoMRsystemswithinthehospital(typicallyless than10min).
Allparticipantsgavetheirinformedconsentbeforethestart of the study. The study was approved by the regional ethics committee(ReferenceNo.M72-07,T92-08).
2.2. Dataacquisitionandimageanalysis 2.2.1. 2DMRE
The2DMRE(MR-Touch,GEMedicalSystems,Milwaukee, US)wasperformedbytransmittingmechanicalwavesat60Hz intotherightsideoftheliverbyapassivetransducer(acoustic) placed on the anterior chest wall to the right of the xiphoid processof thepatient, whowaslyinginasupineposition.A 1.5TGE450WMRsystem(GEHealthcare,Milwaukee,US) wasused,alongwithaphasedarraybodycoil(HD8Torso,using all8coilelements).Thequantitativeshearstiffnessmapswere generatedbyprocessingtheacquiredimageswithapreviously describedlocalfrequencyestimationinversionalgorithm[16].
2.2.2. 3DMRE
Theprincipleofthe3DMREmethodusedhasbeendescribed previously[17,18].Inshort,mechanicalwavesof56Hz were transmitted into theright sideof the liverby anactive trans-ducer (electromagnetic) thatwas placedon theanterior chest wall to the right of the xiphoid process of the patient, who
Table1
Demographicvariables,fibrosisstaginganddiagnosis.
Patient Gender Agea(year) BMI(kg/m2) Biopsyageb(year) Biopsylocalizationc Fibrosisstaged Diagnosise
1 F 80 22.9 2.4 Right 2 PSC 2 M 33 22.8 3.4 Left 3 PSC 3 M 69 27.8 4.6 Left 4 NAFLD 4 F 41 20.7 0.9 Left 3 AIH 5 F 75 19.6 3.3 Right 2 AIH,PBC 6 F 44 25.6 1.3 Right 2 AIH 7 F 51 27.9 1.9 Left 1 AIH
a Agereferstothepatient’sagewhentheMREexaminationswereperformed.
b Biopsyagereferstothetimein-betweentheinitialbiopsyandwhentheMREexaminationswereperformed. c Biopsylocalizationreferstoeithertheleftortherightliverlobe.
d FibrosisstageaccordingtoBatts–Ludwigfibrosisscoring;1=portalfibrosis,2=periportalfibrosis,3=septalfibrosis,4=cirrhosis.
Table2
SummaryoftheMRprotocols.
2DMRE,GRE 3DMRE,GRE
MRsystem GE450W,1.5T PhilipsAchieva,1.5T
Fieldofview 440mm×440mm 320mm×256mm Matrix 256×64 80×38 Slicethickness 10mm 4mm #Slices 4 9 Flipangle 30◦ 15◦ TR 50ms 112ms TE 21.7ms 9.21ms
Acceleration ASSET=2 SENSE=2
#Breath-hold 4 4
was lying ina supineposition. A 1.5T PhilipsAchieva MR system(PhilipsHealthCare,Best,TheNetherlands)wasused, alongwithaphasedarraybodycoil(SenseTorsoXL,usingall 16 coil elements). The shearwaveswere obtained by apply-ingthecurloperatorandusingtheVoigtrheologicalmodelto obtainshearelasticitymaps. Detailsoftheelasticity/viscosity mapcalculationscanbefoundelsewhere[17,18].
RelevantprotocolparametersaresummarizedinTable2.
2.2.3. Imageanalysisoftheviscoelasticmaps
Foreachpatient,aregionofinterest(ROI)wasplaced man-uallybyaradiologist(withmorethan20years’experiencein abdominalradiology;BN)inanappropriatesingle10mmslice acquiredusing2DMRE.TheshapeandsizeoftheROIwere limitedbytheuncertaintymeshcalculatedbythe2DMRE sys-tem.ThereafteracorrespondingROIforthe3DMREwasplaced manuallyoverthreeslicessuchthatitcoveredthesame anatom-icalregionaswiththe2DMREmeasurement(each3DMRE slicehadathicknessof4mm).Thisyieldedatotalcranio-caudal coverage of theROIs equalto10mm(for the 2DMRE) and 12mm(forthe3DMRE).Subsequently,themeanandstandard deviations(unitkPa)ofthestiffness(2DMRE),elasticity(3D
MRE) and‘Gabs,Elastic’ (3D MRE) bothinunits of kPa were
calculatedforeachROIandpatient.‘Gabs,Elastic’istheabsolute
valueoftheshearmodulus,whichinprincipleisequivalentto theviscoelasticpropertyshearstiffness.
2.3. Statisticsandcomputersoftware
Statisticalanalysesincludedalinearregressionwitha95% confidenceintervalonthefittedregressionparameters;R2was used asameasureof model fit anddata correlation.The sta-tistical analyseswere performed using Mathematica (9.0.1.0, WolframResearchInc.,Champaign,IL,U.S.),theROIdrawing andelastogramquantification forthe datafrom the 2DMRE wasperformedonaPACS-system(PACSIDS7,15.1.10.8, Sec-traAB, Linköping,Sweden), andfor the 3D MRE the ROIs were placed and analysed using a custom software package implementedinROOT(5.30/01,CERN,Geneva,Switzerland) generouslyprovidedbyR.Sinkus(KingsCollege,London,UK).
3. Results
ThemeasuredvaluesforallpatientsarepresentedinTable3. An example of a ROI placement in the viscoelasticmaps is showninFig.1.Boththeelasticityandthe‘Gabs,Elastic’
corre-latedwellwiththestiffnessmeasurementcarriedoutonthe2D MRE system(Fig. 2aandb), as was shownbythe elasticity and stiffness correlation R2=0.96 (P<0.001), slope=1.08
(P<0.001, 95% CI={0.83;1.33}), intercept=0.61kPa (P=0.08, 95% CI={−0.12;1.34}), ‘Gabs,Elastic’ and stiffness
correlation R2=0.96 (P<0.001), slope=0.95 (P<0.001, 95% CI={0.72;1.18}), intercept=0.28kPa (P=0.43, 95% CI={−0.56;1.12}).
4. Discussion
The main result was that there was a very good correla-tion betweentheelastograms obtained fromthetwo different MRE methods, one using a passive acoustic transducer and 2D acquisitionandoneusinganactiveelectromagnetic trans-ducer and3Dacquisition;inbothcomparisons(i.e.,elasticity and ‘Gabs,Elastic’ vs. stiffness) the linear factor inthe
regres-sionwasclosetoone.Iftheseobservationsholdtrueinlarger studies, theexistence ofrobust andreliableabsolute quantifi-cation tools from different manufacturers producing directly comparable data is encouraging,both from apatient as well asaresearchperspective.However,careshouldbetakenwhen comparingtheelasticityandstiffnessmeasures,sincetheyhave slightly different physical interpretations, and a standardized way of representing/processing MRE would be very benefi-cial. In recent publications the authors have argued that the use of multi-frequency MRE (MMRE) could allow for such a robust standardization of MRE [19,20] – this would be a promising enhancement for the diagnosis and monitoring of diseaseprogressioninaclinicalsetting.Anaddedvalueofthe MMREapproachisthatitpotentiallyalsoallowsforassessment and differentiationof the inflammatoryprocesses inthe liver parenchymafromfibrosis[19–21].Thispossibilitytoalsoassess theinflammatorycomponentindiffuseliverdiseasewouldbean importantstepforwardtowardsthecreationofacomprehensive non-invasiveMR-toolsetformeasuringandmonitoringvarious histopathologicalparameters.
Table3
Measuredviscoelasticproperties. Patient Fibrosis
stagea
3DMREb 2DMREb
Elasticity(kPa) Gabs,Elastic(kPa) Stiffness(kPa)
1 2 1.96±0.37 2.34±0.41 3.03±0.55 2 3 0.77±0.67 1.76±0.36 1.63±0.25 3 4 5.72±2.58 7.03±2.45 7.06±1.62 4 3 2.18±0.29 2.50±0.44 2.56±0.55 5 2 1.96±0.02 2.49±0.54 3.18±0.62 6 2 1.99±0.26 2.54±0.41 2.88±0.54 7 1 3.13±0.64 3.86±0.86 3.44±0.80
aFibrosisstagebasedonhistopathologicalexamination(seeTable1). bViscoelasticdatapresentedasmean±onestandarddeviation.
Fig.1.ViscoelasticmapsandROIpositionsforpatient7.(A)AconventionalTHRIVEimage,acquiredpriortothe3DMRE.The3DMREelasticitymapisshown infalsecolourin(B),withtheROImarkedbyaredoutline(alsoshownoverlaidontheTHRIVEimagein(A)).(C)The2DMREstiffnessmapinfalsecolour, andin(D)theconfidencemeshisshownoverlaidonthe2DMREstiffnessmap(regionsexcludedfromthemeshcorrespondtoregionswithhighconfidenceinthe calculations),aswellastheROImarkedbyawhiteoutline.Ascanbeseenin(A)and(D)theROIsareplacedinthesameanatomicalregionoftheliverforboth
2Dand3DMRE.
Thereareafew potentiallimitationswheninterpretingthe resultsofthepresentstudy:(i)Arelativelysmallsamplesize was used. (ii) The biopsies were in some cases performed up to 4 years prior to the MRE. Importantly, the biopsies wereincluded inthisworktoshowthe spreadof thefibrosis stages in the patient cohort, and not to gauge the diagnos-tic power of the MRE methods. (iii) The biopsies were not spatially correlated with the MRE ROIs. (iv) The patients were not required to fast prior to the examination, and the postprandialeffecthasbeenshowntoaffecttheabsolute stiff-ness/elasticity values [22]. However, this postprandial effect
probablyhadverylittleeffectonthecomparisonsbetweenthe two systems sincethe patient examinations wereobtained as closeintimeaswaspossible.(v)Thedriverfrequencieswere not identical (although very close) and this could also have influenced the observed absolute values [23]. Despite these limitations, we believe that this is an important addition to the body of knowledge on using MRE in a clinical setting with different MRE methods. A central aspect of this small patient group is that the spread in the fibrosis stages is pre-sumablyrepresentativeforpatientswithsuspecteddiffuseliver disease in need of diagnostic workup. Also,the aetiologyin
Fig.2.Correlationanalysisoftheviscoelasticproperties.Inbothpanelsthedataarepresentedasthemeanvalueoftheviscoelasticproperties,theerrorbars correspondtoonestandarddeviation,andthedashedlinescorrespondtothe95%confidenceinterval(CI)ofthelinearregression.(A)Thecorrelationofelasticity andstiffness,and(B)thecorrelationof‘Gabs,Elastic’andstiffness.InbothcomparisonsR2wasequalto0.96,andonlytheslopeofthelinearregressionwassignificant
thispatientgroupisheterogeneous,whichisatypicalclinical situation.
Insummary,thispilotstudyshowsthatdifferentMRE meth-odscanproducecomparablemeasurementsoftheviscoelastic propertiesoftheliver.Theexistenceofsuchcomparable mea-surementsisimportant,bothfromaclinicalaswellasaresearch perspective,sinceitallowsforequipment-independent monitor-ingofdiseaseprogression.
Conflictofinterest
Nonedeclared.
Acknowledgements
We gratefully acknowledge funding from the Swedish ResearchCouncilforMedicineandHealth(VR/MH,Dnr 2007-2884),aswellasfundingfromtheUniversityHospitalResearch FundsandLinköpingUniversity.Wearealsogratefulfor gener-oushelpanddiscussionwithDr.RalphSinkus(KingsCollege, London, UK) and Dr. Philippe Garteiser (INSERM, Paris, France).
References
[1]GoldinRD,GoldinJG,BurtAD,DhillonPA,HubscherS,WyattJ,etal. Intra-observerandinter-observervariationinthehistopathological assess-mentofchronicviralhepatitis.JHepatol1996;25(5):649–54.
[2]KleinerDE,BruntEM,VanNattaM,BehlingC,ContosMJ,Cummings OW,etal.Designandvalidationofahistologicalscoringsystemfor non-alcoholicfattyliverdisease.Hepatology2005;41(6):1313–21.
[3]BedossaP,DargereD,ParadisV.Samplingvariabilityofliverfibrosisin chronichepatitisC.Hepatology2003;38(6):1449–57.
[4]RousseletMC,MichalakS,DupreF,CroueA,BedossaP,Saint-AndreJP, etal.Sourcesofvariabilityinhistologicalscoringofchronicviralhepatitis. Hepatology2005;41(2):257–64.
[5]PiccininoF,SagnelliE,PasqualeG,GiustiG.Complicationsfollowing percutaneousliverbiopsy.Amulticentreretrospectivestudyon68,276 biopsies.JHepatol1986;2(2):165–73.
[6]RegevA,BerhoM,JeffersLJ,MilikowskiC,MolinaEG,PyrsopoulosNT, etal.Samplingerrorandintraobservervariationinliverbiopsyinpatients withchronicHCVinfection.AmJGastroenterol2002;97(10):2614–8.
[7]MartinezSM,CrespoG,NavasaM,FornsX.Noninvasiveassessmentof liverfibrosis.Hepatology2011;53(1):325–35.
[8]RockeyDC.Currentand futureanti-fibrotictherapiesforchronicliver disease.ClinLiverDis2008;12(4):939–62,xi.
[9]Cohen-NaftalyM,FriedmanSL.Currentstatusofnovelantifibrotic ther-apies in patients with chronic liver disease. Ther Adv Gastroenterol 2011;4(6):391–417.
[10]WangQB,ZhuH,LiuHL,ZhangB.Performanceofmagneticresonance elastographyanddiffusion-weightedimagingforthestagingofhepatic fibrosis:Ameta-analysis.Hepatology2012;56(1):239–47.
[11]YinM,TalwalkarJA,GlaserKJ,ManducaA,GrimmRC,RossmanPJ,etal. Assessmentofhepaticfibrosiswithmagneticresonanceelastography.Clin GastroenterolHepatol2007;5(10),1207.e2–1213.e2.
[12]HuwartL,SempouxC,VicautE,SalamehN,AnnetL,DanseE,etal. Mag-neticresonanceelastographyforthenoninvasivestagingofliverfibrosis. Gastroenterology2008;135(1):32–40.
[13]GlaserKJ,ManducaA,EhmanRL.ReviewofMRelastography applica-tionsandrecentdevelopments.JMagnResonImaging2012;36(4):757–74.
[14]GodfreyEM,MannelliL,GriffinN,LomasDJ.Magneticresonance elas-tographyinthediagnosisofhepaticfibrosis.SeminUltrasoundCTMR 2013;34(1):81–8.
[15]NorenB,Forsgren MF,DahlqvistLeinhardO,DahlstromN,Kihlberg J,RomuT,etal.Separationofadvancedfrommildhepaticfibrosisby quantificationofthehepatobiliaryuptakeofGd-EOB-DTPA.EurRadiol 2013;23(1):174–81.
[16]ManducaA,OliphantTE,DresnerMA,MahowaldJL,KruseSA,Amromin E,etal.Magneticresonanceelastography:non-invasivemappingoftissue elasticity.MedImageAnal2001;5(4):237–54.
[17]HuwartL,PeetersF,SinkusR,AnnetL,SalamehN,terBeekLC,etal.Liver fibrosis:non-invasiveassessmentwithMRelastography.NMRBiomed 2006;19(2):173–9.
[18]SinkusR,TanterM,XydeasT,CathelineS,BercoffJ,FinkM.Viscoelastic shearpropertiesofinvivobreastlesionsmeasuredbyMRelastography. MagnResonImaging2005;23(2):159–65.
[19]GarteiserP,D’AssigniesG,LeitaoH,SahebjavaherRS,LambertSA,Mouri F,etal.MultifrequencyvsmonofrequencyMRelastographyforthe char-acterizationofliverfibrosisandinflammation.In:EleventhInternational TissueElasticityConference.2012.p.67.
[20]VanBeersBE,DaireJL,GarteiserP.Newimagingtechniquesforliver diseases.JHepatol2015;62(3):690–700.
[21]GarteiserP,D’AssigniesG,LeitãoHS,MouriF,VilgrainV,etal.The influenceofinflammationandfibrosisonmultifrequencyand monofre-quencyMRelastographyparameters:astudyin47patientswithchronic viralhepatitis.In:ProcIntlSocMagResMed.2014.p.373.
[22]YinM,TalwalkarJA,GlaserKJ,VenkateshSK,ChenJ,ManducaA,etal. DynamicpostprandialhepaticstiffnessaugmentationassessedwithMR elastographyinpatientswithchronicliverdisease.AJRAmJRoentgenol 2011;197(1):64–70.
[23]AsbachP,KlattD,HamhaberU,BraunJ,SomasundaramR,HammB,etal. AssessmentofliverviscoelasticityusingmultifrequencyMRelastography. MagnResonMed2008;60(2):373–9.