QUANTITATIVE IN VIVO MAGNETIC RESONANCE IMAGING OF SPIO-LABELED CELLS IN A CERVICAL

CANCER MOUSE MODEL Zoe O’Brien-Moran1,2, Marie-Laurence Tremblay2, Christa Davis2, James Rioux1, Kimberly D. Brewer1,2

1Dalhousie University; 2 Biomedical Translational Imaging Centre (BIOTIC) at IWK Health Centre & QEII Health Sciences Centre; Halifax, NS, Canada

INTRODUCTION

Background

IMAGING METHODS

GENERAL METHODS

1.BrewerKDetal.ClearanceofdepotvaccineSPIO-labeledantigenandsubstratevisualizedusingMRI.Vaccine2014,32:6956-6.2.RiouxJAetal.QuantificationofsuperparamagneticironoxidewithlargedynamicrangeusingTurboSPI.JournalofMagneticResonance2012,216:152-60.3.MajumdarS.etal.Theinfluenceofpulsesequenceontherelaxationeffectsofsuperparamagneticironoxidecontrastagents.MagneticResonanceinMedicine,1989,102:289-301.4.RiouxJAetal.3DsinglepointimagingwithcompressedsensingprovideshightemporalresolutionR2*mappingforinvivopreclinicalapplication.MAGMA2016;epubaheadofprint.

IMAGING RESULTS

v  C57BL/6 miceareimplantedwith5x105C3cellsin100ulv CD8+ cells are isolated from diseasematchedGFPmice &cultured invitro for7daysv Culturedcellsare loadedwithultra-smallSPIObypassiveincubationfor24hoursv Cellsaredeliveredviatailveininjection24hoursaregivenforinitialcellmigrationv MiceareimagedaccordingtotheImagingMethodsprotocol

v  Understandingtheroleofimmunecellsis invivo isimportantforanumberofdiseaseprocesses,particularlywithcancerandimmunotherapies.

v  Magneticresonanceimaging(MRI)canbeusedtomonitorimmunecells invivobylabelingwithasuperparamagneticironoxide(SPIO)contrastagent.

v  Thesemethods allow us to detect immune cell recruitment andmigrationpatternsbyacquiringMRIsandanalyzingchangesinimagecontrastwithinaregionofinterest.

v  Traditionally,wehaveusedabSSFPMRIpulsesequencetoperformstudiesevaluating SPIO labeled components1. Unfortunately, this sequence lacksspecificity in detecting the contrast effects of SPIO. Labeled cells can bedetectedbyan increase innegative (dark) contrast,butnecrosis results insimilar contrast. Thiswas confirmed using simultaneous positron emissiontomography(PET)/MRI(Fig.1).

v  Additionally, the previous method does not support true quantitativeevaluationofcellularrecruitment.

v  Weproposeuseofasecondsequence,TurboSPI2,thathasgreaterspecificityandsupportscellularquantification.

IMAGE ANALYSIS

RESULTS & DISCUSSION

CONCLUSION

FIGURE1:Upperrow:MRIobtainedusingbSSFPinsagittal(a),coronal(b),andaxial (c) planes. Negative contrast in the center of the flank tumor suggestsSPIOuptakeornecrosis.Bottomrow: 18F-FDGPET/MRIoverlay insagittal (d),axial(e),andcoronal(f)planesusingPETdataobtainedsimultaneouslywiththeMRdata.Anecroticcoreisconfirmedbylow18F-FDGactivityalignedwiththeregionofnegativeMRcontrast.

v  SPIO nanoparticles have high magnetic susceptibility. They locally alter themagneticfieldintheMRIscanner.TheR2*(s-1)relaxationratesofprotonsnearSPIOareincreased.

v  All MRIs are obtained on a 3T pre-clinical Agilent (Menlo Park, CA, USA)scanner:

v  Balanced steady state free precession (bSSFP) is performed first for high-resolutionanatomicalimages(TR/TE=8/4ms,α=30°,4phasecycles,4averages,256x170x170,FOV=38.4mmx25.5mmx25.5mm,t=64min)

v  highsignaltonoiseratio(SNR)v  highresolutionv  sensitivetoSPIOeffects3

v  TurboSPI isperformedsecond(TR/TEeffective=250/10ms,ETL=8,ESP=10ms,96x96x48,FOV=30mmx30mmx30mm,t=28min)

v  quantificationcapabilitiesv  higherSPIOspecificity

v  TurboSPIisamulti-echosinglepointimaging(SPI)sequencethatisacceleratedthrough compressed sensing techniques. R2* maps are calculated fromTurboSPIdata4.

v  PET data (Fig. 1) was obtained for several mice, using the NuPET (Cubresa,Winnipeg,MB,CA)insertforsimultaneousPET/MR.

QUANTIFICATION

FIGURE2:Asummaryschematicofthegeneralmethodshowingthestepsfromtumorimplantationtoimaging

VERIFICATION v  Immunohistochemistry

was performed on thee x c i s e d t umo r t oconfirmtheexistenceofinjected (SPIO labeled)cells. Cellswere labeledwith DAPI (nuclearstain),andironlabelhasRhodamineBattached.

v  TheR2*mappingthatisperformedusingTurboSPIoffersmoreinformationthanthebSSFPimages.ThespecificityoftheTurboSPIanalysisresultsinmore“truepositive” regions and therefore more accurate evaluation of cellular recruitment. Furthermore, the R2* mapping and calibration provides superior cellquantificationtoprevioustechniques.WearecurrentlyevaluatingthequantitativeresultsofCD8+Tcellrecruitmentinatumormodel

v  We suggest that the imaging protocol for SPIO labeled cell tracking is most powerful when the two sequences, bSSFP and TurboSPI are used together,incorporatingthestrengthsofeachtechnique:thespecificityandquantificationfromTurboSPI,andtheresolutionandSNRofbSSFP.

v  Parallelworkiscurrentlybeingperformedusingthesecelltrackingtechniquestoinvestigatethecellularresponsetodifferentimmunotherapytreatments(seeposterbyTremblayetal.).CD8+Tcellsanddendriticcellsarebeingstudied,particularlytheirresponsetoimmunotherapy,specificallyapeptide-basedvaccine,acheckpointinhibitorandacombinationofboththerapies.

FIGURE3:Upperrow:a)isthebSSFPMRIreconstructedintheaxialplane,b)isan R2* map (pixel intensity in s-1) overlaid on the bSSFP slice that shows adistributionofR2*inthesameplane.Region1indicatesapixelwithasignificantamount of iron in the tumor, Region 2 indicates signal from pure fat(oscillations are fat specific) and 3 suggests a combination of iron and fat.Bottomrow:theMRsignaltimecourseforregion1(c),region2(d),andregion3(e)

v  After data acquisition, the under-sampled TurboSPI images were reconstructedusingasupplementalguideimageacquiredimmediatelybeforetheTurboSPIscan

v  R2* relaxation rate maps were calculated from the reconstructed multi-echoTurboSPIdata.ThesemapsarealignedwiththeaxialslicesofthebSSFPimageandthesetareanalyzedintandem.

v  Figure 3 shows an example image set using this technique. The R2* map in b)showsR2*valuebypixelintensity.Itisnotanironspecificmetric,butanalyzingR2*valuecombinedwiththesignaltimecoursegivesusinformationonironcontentinregionsofinterest.

v  Thetimecourseofthesignalinc)showsavoxelwithsuspectedSPIO-labeledcells,d)showsaperiodictimecoursefromafatvoxel,e)showsatimecourseindicatingthepresenceofbothfatandironfromavoxelinthefatpadanteriortothetumor.

v  Interestingly, labeled CD8+ T cells seem to be primarily located at the tumorperiphery,notinthecentralregions

FIGURE4:R2*calibrationfordifferentconcentrationsofSPIOloadedcells.

v  R2* calibration was performed forcellular quantification purposes &Samples were prepared with knowncell concentrations and the bulk R2*wascalculatedusinglinewidth

v  Cellular concentrations are calculatedfrom the voxel intensities in the R2*mapandcanbeconvertedtonumbersofcells/voxelusingthevoxelsize

v  For example, the voxel indicated byr e g i o n 1 i n F i g u r e 3 h a sapproximately540cells/voxel

FundingProvidedby:IWK Student Scholarship (ZOBM), NSHRF Establishment Grant (KDB),IWKCategoryBgrant(KDB),NSERCDiscoveryGrant(KDB),ACOAAIF(salarysupportKDB)

FIGURE 4: a) DAPI stained cells in a tumor atweek 4 b) Rhodamine B positive (i.e. SPIO-labeled)cellsinthesectionedtumor

0 1000 2000 3000 4000 5000Concentration (cell/mm 3)

240

260

280

300

320

340

R2*

(s-1

)

R2* CalibrationR2* = 0.018929c + 243.9122