BORTEZOMIB(BOR)PHARMACOGENOMICS(PG)IDENTIFYMECHANISMS
OFDRUGRESISTANCEANDPREDICTSURVIVALINMULTIPLEMYELOMA
(MM)TREATEDWITHTOTALTHERAPY3(TT3)
JohnDShaughnessyJr,PingpingQu,JeffHaessler,JohnCrowley
andBartBarlogie
MyelomaInsPtuteforResearchandTherapy,UniversityofArkansasforMedical
Sciences,LiUleRock,AR;CancerResearchandBiostaPsPcs,SeaUle,WA
Abstract
Background:Theprognosisofpa.entswithMMisbestcapturedbygeneexpressionprofiling(GEP)analysisofCD138purifiedplasma
cells(PC),dis.nguishingahighriskgroupof15%withdismalsurvivalusinga70genebaselineriskmodel(BLR).Transla.onalresearch
inTT3wasdesignedtoinves.gatewhethershorttermBORinducedGEPaltera.onscouldadvanceourunderstandingofBOR'snovel
mechanismofac.on.PaPentsandMethods:PGstudieswereperformedaspartoftwoTT3trials(TT3a,n=303;TT3b,n=177),
obtainingPCpriortoand48hraVeraBORtestdose(1.0mg/m2),whichwasaccomplishedin142pa.entsinTT3a(trainingset)and
127inTT3b(testset).Among1051genessignificantlyalteredpostBORinTT3a,80wereiden.fiedasbeingsignificantlyassociated
withEFS.Acon.nuousriskscorewascalculatedandanop.malcutpointforEFSsepara.ondetermined.Theindependentprognos.c
powerofthebinaryriskscorewastestedinTT3b.Mul.variateanalyses(MV)wereemployedtodeterminepostBORrisk(PBR)in
rela.onshiptostandardprognos.cvariablesandBLR.Results:ThediscriminatorypowerinTT3a(3yrOS:95%v45%,p<0.0001;3yr
EFS:90%v35%,p<0.0001)wasconfirmedinTT3b(18moOS:100%v65%,p=0.0004;18moEFS:95%v45%,p<0.0001).Evalua.ng
PBRinthecontextofBLR,12/26inTT3aand7/21inTT3bdeemedashavinglowBLRhadhighPBR;conversely,8/126inTT3aand
14/106inTT3bdeemedashavinghighBLRhadlowPBR.Inthecontextofour8molecularsubgroupmodel,highPBRwasover
representedintheProlifera.on(PR)subgroup(7/15inTT3a,8/18inTT3b)andabsentintheLowBonedisease(LB)group(0/28).On
MV,PBRwasanindependentadversevariableforbothOSandEFSinTT3a(OS:HR=3.17,p=0.006,R2=55%;EFS:HR=4.40,p<0.001,
R2=48%)andinTT3b(OS:HR=13.00,p=0.002,R2=48%;EFS:HR=15.57,p<0.001,R2=55%).Proteasomegenesrankedfirstamong
thosedifferen.allyupregulatedbyBOR.Conclusions:PGiden.fiedapowerful80genePBRmodelwithunprecedentedprognosis
discrimina.ngpower,dispellingBLRfromMVanalysisbyalteringBLRdesigna.onmainlyfromlowtohighrisk.HighPBR(18%)could
betracedtoupregula.onofproteasomegenes,thetargetofBOR.
BackgroundandSignificance
· A70genesignatureinnewlydiagnoseddiseasecandefinehighandlowriskinmulPpletreatmentse]ngs
· However,asubsetoflowriskcasesthatfollowanaggressiveclinicalcourseconvertfromlowtohighrisk
overPme
· ThesedatasuggeststhatMMexhibitsintraclonalheterogeneitywithrespecttodrugsensiPvityandthatclonal
evoluPon,oremergenceofanundetectableaggressiveclonealreadypresentatdiagnosis,canemergeinthe
absenceofeffecPvetherapies
· Thus,moreaccurateriskassessmentisneededtodesignearlytreatmentstrategiesthatmightpreventthis
transformaPon.
· WehavepreviouslyreportedonthefeasibilityofperformingpharmacogenomicstudiesinMMandthe
idenPficaPonofpredicPvegeneexpressionsignaturesintumorcellsfollowingshorttermexposureto
thalidomide,lenalidomideanddexamethasone.
· BortezomibisafirstinclassproteasomeinhibitorthatisFDAapprovedforthetreatmentofMM
· BortezomibhasbeenincorporatedintoTotalTherapy3,afrontlinetreatmentprotocolfornewlydiagnosedMM.
Hypothesis
Pharmacogenomicsontumorcellsand/orthetumormicroenvironmentpriortoandfollowing
shortterminvivotestdosingofsingleagentbortezomibmayprovideinsightsinto
mechanismsofacPonandaidinthedevelopmentofmorerobustpredictorsofsensiPvity/
resistanceaffordedbybaselineclinicalandmolecularvariables
PaPentsandMethods
PaPents
Trainingset:
142consecuPvenewlydiagnosedcasestreatedonUAMS200333
ValidaPonset:
127consecuPvenewlydiagnosedcasestreatedonUAMS200666
Treatment
INDUCTION:VTDPACEX2
HDT:MEL200mg/m2X2withautologousstemcellsupport
CONSOLODATION:VTDPACEX2
MAINTENANCE:VTD/VRD
Methodology
·CD138immunomagnePcbeadselecPonofMMtumorcellswasappliedtobone
marrowaspiratesatdiagnosisand48hrfollowingsingleagentinvivoBOR(1.0mg/
m2).CellpuritywasassessedbytwocolorCD38/CD45flowcytometryandallsamples
containedgreaterthat85%tumorcells
·TotalRNAwasisolated,convertedtolabeledcRNA,andhybridizedtoU133Plus2.0
Microarrays,datawasMAS5.0normalizedandanalyzed.
·AliquotsofcellsusedinmicroarrayweretrypsindigestedandpepPdeswereanalyzed
intriplicatebynanoLC/MS/MSusinganLTQOrbitrapmassspectrometer.
·ThepepPde/proteinabundancedatawasprocessedusingtheElucidatorsolware
packagefromRoseUaBiosolware.
DerivaPonofthePostBORGenomicPredictor
· Inthetrainingset1051(P<.005)of~25,000uniquegeneswerefoundtobe
differenPallyexpressedfollowingsingleagentBOR
· CoxproporPonalhazardsmodelwasusedtocorrelate
1. changeinexpression,adjusPngforbaselineexpression,withEFS
2. postdrugexpressionwithEFS
· Bycomparingtheqvalues,thepostdrugexpressionvaluesweredeemedmore
significantand80geneswereretainedthatsaPsfiedasignificance(P<=.05).
· GeometricmeanofgeneswithhazardraPos(HR)of<1.0(changefavorable)andgenes
withHR>=1(changeunfavorable)wasusedtocreateascoreforeachpaPent.
· Thehigherthescorethepooreroutcome.
· Usingthescore,wedefinedopPmalcutoffvaluestoclassifypaPentsintohighand
lowriskgroupstomaximizethesurvivaldifferencebetweenthegroups.
· Developedin142,itwasvalidatedin127
· ToverifytheindependentprognosPcpowerofthescore,italongwithotherprognosPc
factorswereexaminedbymulPvariateanalysis(MVA)ofEFSandOS.
Results
PostBORGEPSignatureofHighRiskCharacterizedby
IncreasedExpressionofProteasomeGenes
HiRisk
LoRisk
Proteasome
Genes
PreTreatment
Proteasome
Genes
Post48hr
invivoBOR
ElevatedExpressionofProteasomeSubunitsFollowingShortTermBORSignificant
ComponentofPostBortezomibModelofHighRiskMM
Ingenuity Pathway Analysis
Red filled shapes represent
genes upregulated in high
risk
PostBOR80GeneModelisaSignificant
DiscriminatorofRisk
EFS
OS
1.0
1.0
0.8
0.8
0.6
P = 3.0e-9
P = 8.8e-8
0.6
0.4
0.4
0.2 Low Risk (18/116)
Low Risk (14/116)
oporPonofCases
0.2
High Risk (17/26)
oporPonofCases
High Risk (14/26)
Pr 0.0
Pr 0.0
0
1
2
3
4
5
0
1
2
3
4
5
YearsFromStartofTherapy
YearsFromStartofTherapy
PostBOR80GeneModelCapturesRiskOutliers
From70GeneBaselineModelin200333
EFS
OS
1.0
LowLow(15/108)
1.0
LowLow(12/108)
0.8
0.8
HighLow(3/8)
HighLow(2/8)
0.6
0.6
LowHigh(6/12)
LowHigh(5/12)
0.4
0.4
HighHigh(11/14)
HighHigh(9/14)
oporPonofCases 0.2
oporPonofCases 0.2
Pr
Pr
0.0
0.0
0
1
2
3
4
5
0
1
2
3
4
5
YearsFromStartofTherapy
YearsFromStartofTherapy
Lowlowvslowhigh:p<.0001
Lowlowvslowhigh:p=.002
Lowlowvshighlow:p=.069
Lowlowvshighlow:p=.24
Lowlowvshighhigh:p<.0001
Lowlowvshighhigh:p<.0001
Lowhighvshighlow:p=.46
Lowhighvshighlow:p=.43
Lowhighvshighhigh:p=.18
Lowhighvshighhigh:p=.31
Highlowvshighhigh:p=.067
Highlowvshighhigh:p=.081
Note:SamplegroupsinKManalysisaredefinedbybeinghighorlowriskbybaseline70genemodel(lel)
andhighorlowriskbypostBOR80genemodel(right)
DistribuPonofPostBOR80GeneScoresisSimilarin
TrainingandValidaPonSets
TT3:Training
TT3b:ValidaPon
HighRiskCutOff
Density
Score
PostBOR80GeneModelValidatedin200666
LowRisk
HighRisk
EFS
OS
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
oporPonofCases 0.2
Pr
oporPonofCases 0.2
Pr
0.00
1
2
3
4
5
0.00
1
2
3
4
5
YearsFromStartofTherapy
YearsFromStartofTherapy
SimilarSurvivalEsPmatesfor
PostBORDefinedHighandLowRiskin
TrainingandValidaPonSets
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
oporPonofCases 0.2
oporPonofCases 0.2
Pr
Pr
0.0
0.0
0
1
2
3
4
5
0
1
2
3
4
5
YearsFromStartofTherapy
YearsFromStartofTherapy
TrendSuggeststhatasin200333,PostBOR
ModelRefinesPredicPonAffordedby70Gene
Modelin200666
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
oporPonofCases 0.2
oporPonofCases 0.2
Pr
Pr
0.0
0.0
0
1
2
3
4
5
0
1
2
3
4
5
YearsFromStartofTherapy
YearsFromStartofTherapy
Lowlowvslowhigh:p=.065
Lowlowvslowhigh:p=.016
Lowlowvshighlow:p=.98
Lowlowvshighlow:p=.655
Lowlowvshighhigh:p<.0001
Lowlowvshighhigh:p<.0001
Lowhighvshighlow:p=.317
Lowhighvshighlow:p=.346
Lowhighvshighhigh:p=.49
Lowhighvshighhigh:p=.723
Highlowvshighhigh:p=.04
Highlowvshighhigh:p=.092
MulPvariateAnalysisin200666RevealsLow
ExpressionofTP53 (P53HIGH)andPostBORModel
(PV80G2)AssociatedwithPoorOutcome
ProteasomeProteinsAlsoIncreaseFollowingInvivoBOR
CD1 CD1
HY
MF
CD2 HY
CD1
CD1
HY
MF
CD2 HY
PtA PtB
PtC PtD
PtE PtF
PtA
PtB
PtC
PtD
PtE
PtF
BL PB BL PB BL PB BL PB BLPBBL PB
BL PB BL PB BL PB BL PB BL PB BL PB
PSMA5
PSMA7
PSMA6
PSMA4
PSMA3
PSMB2
PSMB4
PSMA1
eins
3.44
2.62
1.99
2.02
1.65
1.54 PBScore
ot
0.92
0.65
1.3
0.96 0.03
.051 BLScore
1400Pr
2.550
0
2.550
Zscore
ProteinExpressionofPSMComponentsin80Gene
ModelGenerallyFollowsmRNAChanges
CD1 CD1 PR/HY MF CD2 HY/MY
PtA
PtB
PtC PtD PtE
PtF
BL PB BL PB BL PB BLPB BL PB BL PB
PSMA5
PSMA7
PSMA6
Protein
PSMA4
PSMA3
PSMB2
PSMB4
PSMA1
PSMA7
PSMA4
mRNA
PSMA2
PSMB4
PSMB4
3.44 2.62 1.99 2.02 1.65 1.54
80GeneScore
0.92 0.65 1.3 0.96 0.03 .051
70GeneScore
Conclusions
AsingleinvivotestdoseofBOR(1.3mg/m2)causesGEPchangesintumor
cellsthatarerelatedtosurvival
Highriskisassociatedwithincreasedexpressionofanumberofproteasome
genesand proteinsfollowingBOR
80genepostBORscorecanpredictEFSandOSintwoindependentdatasets
80genescorecanidenPfyhighriskinlowriskcasesasdefined70genemodel
80genescoreisasignificantindependentriskfactorthattogetherwithother
variablesincreasesR2togreaterthan50%
FutureQuesPons
IstheincreaseinproteasomegenesfollowingBORdueto:
ashilincellpopulaPons(preferenPalkillofcellswithlowriskGEP)
Druginducedgeneandproteinexpression?
80genescorecanbereadoutoftargetengagement.Canalteringthedoseand/or
scheduleofBORovercome(highrisk/proteasomehyperacPvaPon)resistance
IstheaggresomealsohyperacPvatedinhighrisk?BOR+HDACiinthisdisease?
Dogeneswithinthemicroenvironmentchangeandarerelatedtooutcome?InPBLs?
IshyperacPvaPongenePcallyprogrammedortumorcellspecific?
DoesinvivotesPngprovideadvantages(tumorhostinteracPons)orcouldthistestbe
performedexvivo
CanpostBORscorepredictefficacyofnonBORcontainingregimensorotherprotocols?
Isthissignaturelinkedtogeneralresistance?