450 • J Clin Pharmacol 2010;50:450-458

Cytochrome P450 2D6 (CYP2D6) is involved inthe metabolism of approximately 20% of all

drugsinclinicaluse,includingβ-blockers,anti-arrhythmics, antidepressants, and antipsychotics.1,2

Itisalsoresponsibleforthemetabolicactivationoftamoxifen and opioids like codeine, hydrocodone,andoxycodone.3,4

The metabolic activity of CYP2D6 varies up to1000-fold among individuals and ethnic popula-tion.5-7ThegeneticcontributiontothisvariabilityinCYP2D6 activity has been well documented.8CYP2D6 is a highly polymorphic gene with morethan 70 allelic variants identified (http://www.cypalleles.ki.se/), which encode for proteins withfull(CYP2D6*1,*2,*33,*35),reduced(CYP2D6 *9, *10, *17, *29, *37, *41, *45, *46),ornoenzymaticactivity(CYP2D6*3-*8,*11-*16,*18-*21,*38,*40,*42,*44).9MultiplecopiesofCYP2D6alleles(ie,*1,*2, *35) have been reported in subjects with theultrarapid metabolism (UM) phenotype.10,11 ThislargeinterindividualvariabilityinCYP2D6capacity

Accurate assessment of CYP2D6 phenotypes from geno-type is inadequate in patients taking CYP2D6 substrate together with CYP2D6 inhibitors. A novel CYP2D6 scoring system is proposed that incorporates the impact of con-comitant medications with the genotype in calculating the CYP2D6 activity score. Training (n = 159) and validation (n = 81) data sets were obtained from a prospective cohort tamoxifen pharmacogenetics registry. Two inhibitor fac-tors were defined: 1 genotype independent and 1 genotype based. Three CYP2D6 gene scoring systems, and their combination with the inhibitor factors, were compared. These 3 scores were based on Zineh, Zanger, and Gaedigk’s approaches. Endoxifen/NDM-Tam plasma ratio was used as the phenotype. The overall performance of the 3 gene scoring systems without consideration of CYP2D6-inhibiting medications in predicting CYP2D6 phenotype was poor in both the training set (R2 = 0.24, 0.22, and 0.18)

and the validation set (R2 = 0.30, 0.24, and 0.15). Once the CYP2D6 genotype-independent inhibitor factor was inte-grated into the score calculation, the R2 values in the train-ing and validation data sets were nearly twice as high as the genotype-only scoring model: (0.44, 0.43, 0.38) and (0.53, 0.50, 0.41), respectively. The integration of the inhibitory effect of concomitant medications with the CYP2D6 genotype into the composite CYP2D6 activity score doubled the ability to predict the CYP2D6 pheno-type. However, endoxifen phenotypes still varied substan-tially, even with incorporation of CYD2D6 genotype and inhibiting factors, suggesting that other, as yet unidenti-fied factors must be involved in tamoxifen activation.

Keywords: Activity score; CYP2D6; inhibition; tamoxifenJournalofClinicalPharmacology,2010;50:450-458

©2010theAmericanCollegeofClinicalPharmacology

From the Division of Clinical Pharmacology (Dr Borges, Dr Desta, Dr Jin, Mr Robarge, Mr Philip, Ms Nguyen, Dr Skaar, Dr Flockhart, Dr Li) and Division of Biostatistics (Mr Faouzi, Dr Li), Department of Medicine, Indiana University School of Medicine, Indianapolis, Indiana; Division of Hematology Oncology, Department of Internal Medicine, University of Michigan Medical Center, Ann Arbor, Michigan (Mr Hayes, Mr Rae); and Department of Oncology, Johns Hopkins University, Baltimore, Maryland (Mr Stearns). Submitted for publica-tion December 13, 2008; revised version accepted November 11, 2009. Address for correspondence: Lang Li, PhD, Division of Biostatistics/Clinical Pharmacology, Indiana University, School of Medicine, 410 W. 10th St., HITS 3000, Indianapolis, IN 46202; e-mail: lali@iupui.edu.DOI:10.1177/0091270009359182

Composite Functional Genetic and Comedication CYP2D6 Activity Score in

Predicting Tamoxifen Drug Exposure Among Breast Cancer Patients

Silvana Borges, MD, Zeruesenay Desta, PhD, Yan Jin, MD, Azzouz Faouzi, MS, Jason D. Robarge, MS, Santosh Philip, MS, Anne Nguyen, BSc, Vered Stearns,

Daniel Hayes, James M. Rae, Todd C. Skaar, PhD, David A. Flockhart, MD, PhD, and Lang Li, PhD

Drug InteractIons

SCORING SYSTEM FOR CYP2D6

DRUg INTeRACTIONS 451

has important clinical consequences.1,4 The poormetabolism (PM) status has repeatedly been associ-ated with an increased risk of adverse effects inpatients taking drugs whose metabolism is largelydependentonCYP2D6.12-17PMsalsoshowdecreasedefficacy when treated with drugs that require meta-bolic activation by CYP2D6.16,18 On the other hand,UMs have been reported to have a poor therapeuticresponse to certain CYP2D6 substrates13,19-21 orincreased opiates toxicity, including fatalities.22-26Furthermore,thecostoftreatingpatientswithextremesin CYP2D6 activity (PMs and UMs) has been esti-matedtobe$4000to$6000peryeargreaterthanthecostof treatingpatients in theextensivemetabolism(EM)andintermediatemetabolism(IM)groups.27

CYP2D6 activity can be affected by nongeneticfactorsaswell.Forexample,manycommonlypre-scribedmedications areknown to inhibit the cata-lytic activity of CYP2D6.28 Pregnancy and diseaseconditions also affect CYP2D6 activity. As withCYP2D6 genetic variations, drug interactions mayhave important clinical consequences, as shownrecentlyfortamoxifenbyourgroup.18,29,30

Agrowingbodyofevidenceisnowavailableshow-ingthatintersubjectvariabilityofCYP2D6causedbygenetic and nongenetic factors is indeed a determi-nant of response to some of the drugs primarilymetabolizedbyCYP2D6.Despitethis,itisstilldiffi-culttointegratesuchinformationtothedrugdevelop-ment process and to guide decisions in clinicalpractice.ItfollowsthatpreciseknowledgeofCYP2D6metabolicstatusmighthelpresearchersdesignappro-priatetrialsoridentifyasubgroupofpatientswhoareat high risk for failure of therapy or adverse effectsduringinitiationoftherapywithCYP2D6substrates.Traditionally, different groups of phenotypes havebeenused todescribe rangeofCYP2D6activity (eg,PMs,EMs),andthesephenotypescouldbemeasuredthroughadministrationofaprobedrugorpredictedfromgenotypes.However,giventhecomplexityoftheCYP2D6 gene structure, large numbers of uniqueCYP2D6 diplotypes with uncertain phenotype out-comehavebeenidentified.30-32AccurateassessmentofCYP2D6phenotypesfromgenotypeis inadequateinpatients taking a CYP2D6 substrate together withinhibitorsoftheenzyme.Becauseoftheselimitations,researchershaverecognizedtheneedtoimprovetra-ditionalwaysofassessingCYP2D6metabolicstatus.Severalsuchapproacheshavebeenproposedrecently,among them the genotype–phenotype relationshipdescribed by Zanger et al8 and the CYP2D6 activityscoresystemproposedbyZinehetal33andGaedigket al.34 There are distinct advantages of using an

activityscoreforgroupingCYP2D6genotypes.Intheresearcharena,suchascorepermitscombinationsofmanydifferentdiplotypesintoa fewactivityscores,thereforedecreasingthenumberofgroupsforcompari-soningenotype–phenotypeassociationstudies.Inclin-icalpractice,clinicallaboratoriescaneasilycalculateaCYP2D6activityscore,thussimplifyingtheinterpreta-tion of genotype results. However, these approacheshavenotincorporatedtheimpactofconcomitantmedi-cationsincalculatingtheactivityscore.TheimpactofinhibitiononCYP2D6activityvarieswiththeCYP2D6genotype and the potency of the CYP2D6 inhibitor(weak,moderate,andstronginhibition).29,30

Ourpriorwork29,30,35hasdemonstratedthatthecon-versionoftamoxifentoitsactivemetabolite,endoxifen,isdependentonCYP2D6activity.Thepurposeofthisstudyistousetherichgenotype,concomitantmedica-tion,andphenotypedatafromaprospectivetamoxifenpharmacogenetics cohort study to (1) develop aCYP2D6 scoring system that accounts for CYP2D6genotypeandCYP2D6-mediateddruginteractionsand(2) compare the performance of our scoring methodwithpreviouslypublishedmethods.

METHODS

Dataonmedicationhistory,genotypefor29CYP2D6alleles,andplasmaconcentrationsoftamoxifenanditsmetaboliteswereobtainedfrom159womenwhopar-ticipatedinourtamoxifenpharmacogeneticstrialpub-lished elsewhere.30 This sample was used as thetrainingdatasetforthiswork.Asecondgroupof81patientswhowereenrolledatalaterstageinthesamestudywereusedasavalidationdataset.Inthereferredstudyontamoxifenpharmacogenetics,theendoxifen/N-desmethyltamoxifen(endoxifen/NDM)plasmaratioreflectedbettertheCYP2D6activitythantheendoxifenplasmaconcentrationalone. Inthecurrentstudy,weused the endoxifen/NDM plasma ratio as an indexmeasureofCYP2D6metaboliccapacity.

CYP2D6 Gene Score

Basedontheavailablephenotypicdata,CYP2D6alle-leswereassignedavalue that reflects theexpectedactivityoftheCYP2D6enzymeforwhichtheycode.InTableI,fullyfunctionalCYP2D6alleles(eg,*1,*2,*33,*35)wereassignedascoreof1.CYP2D6allelesassociatedwithreducedenzymeactivity(eg,*9,*10,*17,*29,*37,*41,*45,*46)werescoredas0.5.ThenullCYP2D6alleles(eg,*3-*8,*11-*16,*18-*20,*38,*40,*42,*44,*56)andtheirduplicationsreceivedavalue of 0. Duplications of fully active alleles (eg,

452 • J Clin Pharmacol 2010;50:450-458

BORGES ET AL

*1×N,*2×N,*35×N)wereassignedavalue+1,ren-deringascoreof2foreachoftheseduplicatedvari-ants.Duplicatedreducedactivityalleles (eg,*41×Nand*45×N)wereassignedavalueof+0.5,yieldingascoreof1.ThisproposedgenescoreissimilartothatdescribedbyGaedigketal.34

System 1 (Table I) is defined by Zineh and col-leagues’33 approach: PM = 0, IM (*10, *17, *37) =0.5,IM(a*9,*29,*41,*45,*46)=0.75,EM=1,UM(*41xN,*45xN)=1.5,andUM(*1xN,*2xN,*35xN)=2. System 2 (Table I) is defined by Zanger andcolleagues’8approach:PM=0,IM=1,EM=2,andUM=3.Thescoreofabi-allelicCYP2D6genotypeisdefinedas

Genescore=allele1score+allelle2score. (1)

CYP2D6 Inhibitor Factor

Once we calculated the expected activity of theCYP2D6enzymeaccording togenotype,we incorpo-ratedthepossibleeffectsofknownCYP2D6inhibitorson theperformanceof theCYP2D6protein.CYP2D6inhibitorswereclassifiedaseitherweakorstrong inaccordancewithdefinitionoftheUSFoodandDrugAdministration.36Weused2differentmethodstocal-culatethemagnitudeofthereductionofCYP2D6activ-ity caused by the concomitant use of a CYP2D6inhibitor.First,weconsideredtheinhibitoryeffecttobeindependentof thegenotype,nowrepresentedbythe CYP2D6 gene score. We assumed that weakCYP2D6inhibitorswoulddecreasebyhalftheCYP2D6metabolic capacity in all individuals, and strongCYP2D6 inhibitors would completely abolish theCYP2D6activity, resulting inphenocopy (convertingall individuals into poor metabolizers). Therefore,weakandstrongCYP2D6inhibitorswereassignedaninhibitorfactorof0.5and0,respectively,withnocon-sideration of the CYP2D6 gene score (Table II, noCYP2D6 genotype input). In the second method, wetookintoaccountthatindividualswithhighCYP2D6enzymecapacityexperiencea larger inhibitoryeffectthan individuals with decreased CYP2D6 enzymeactivitywhenexposedtoCYP2D6inhibitors.Followingthispremise,wemodifiedtheinhibitorfactoraccord-ing to the individualCYP2D6gene scores, assigninglow inhibitor factor values to individuals with highCYP2D6genescoresandviceversa.

For example, participants taking a CYP2D6 weakinhibitor receivedaCYP2D6 inhibitor factorof0.5 ifthey had a CYP2D6 gene score of 2 and a receivedCYP2D6inhibitorfactorof0.75iftheirCYP2D6genescorewas1.Weagaintookintoaccountthatmostindi-vidualstakingstrongCYP2D6inhibitorsareconvertedto the poor metabolizer status, therefore receiving a

Table I CYP2D6StarAlleleScoreDefinition

CYP2D6 Alleles Function System 1 (Zineh et al33) System 2 (Zanger et al8) Proposed Gene Score System

*3-*8,*11-*16,*18-*20,*36,*38,*40,*42,*44,*56,*4×N

PM 0 0 0

*10,*17,*37 IM 0.5 1 0.5*9,*29,*41,*45,*46 IM 0.75 1 0.5*1,*2,*33,*35 EM 1 2 1*41×N,*45×N UM 1.5 3 1*1×N,*2×N,*35×N UM 2 3 2

EM,extensivemetabolism;IM,intermediatemetabolism;PM,poormetabolism;UM,ultrarapidmetabolism.ProposedgenescoresystemissimilartoGaedigk’s34system:theadditionisassigning*41×Nas1.

Table II CYP2D6InhibitorFactorDefinition

CYP2D6 Inhibitor Factor Calculation Method

CYP2D6 Inhibitor

Factor

NoCYP2D6genotypeinput

NoCYP2D6inhibitor 1CYP2D6weakinhibitora 0.5CYP2D6stronginhibitorb 0

WithCYP2D6genotypeinput

NoCYP2D6inhibitor 1

WeakCYP2D6inhibitora

CYP2D6genescore≥1.75 0.5 CYP2D6genescore0.5-1.5 0.75 CYP2D6genescore0 1StrongCYP2D6inhibitorb

CYP2D6genescore>2 0.25 CYP2D6genescore0.5-2 0 CYP2D6genescore0 1

a.Weakinhibitorsincludesertraline,venlafaxine,citalopram,fluoxet-ine,buproprion,razadone,andcelecoxib.b.Stronginhibitorsincludeparoxetine,amiodarone,andmetoclopro-mide.

SCORING SYSTEM FOR CYP2D6

DRUg INTeRACTIONS 453

CYP2D6 inhibitor factor of 0. The exceptions to thisrulewereparticipantswithCYP2D6genescoresgreaterthan2(ie,UMs),whotendtomaintainsomeresidualCYP2D6activitydespitetheexposuretostrongCYP2D6inhibitors.30,37 These individuals were assigned aCYP2D6inhibitorfactorof0.25(TableII,withCYP2D6genotypeinput).

CYP2D6 Activity Score

TheCYP2D6activityscoreisexpressedastheprod-uct of multiplying the CYP2D6 gene score by theinhibitorfactor,whichisdefinedinEquation2:

Activityscore=genescoreinhibitionfactor.(2)

wherethegenescoreisdefinedinTableIandinhibi-tionfactorisdefinedinTableII.TheCYP2D6activ-ity score represents an estimate of the actualperformanceoftheCYP2D6enzymeinvivoasdeter-mined by the CYP2D6 genotype and the effect ofconcomitantuseofCYP2D6inhibitors.

Statistical Analysis

Descriptiveanalysisofdemographicvariablesisper-formed and presented as sample mean, standarddeviation, median, and range or in categories. Thepredictive performances of various CYP2D6 scoreson the log(endoxifen/NDM) are assessed with R2,andtheirsignificancesareevaluatedwithPvaluesthrough linear regressions. All of the analyses areconductedinSAS9.1(SASInstitute,Cary,NC).

RESULTS

Demographics

Thetrainingandvalidationdatasetswerecomposedof159and81participants,respectively.Demographiccharacteristics suchasage,bodymass index (BMI),frequency of CYP2D6 genotypes, and ethnicity inbothdatasetsarepresentedinTableIII.Bothtrainingandvalidationdatasetshadsimilaragedistributions,withmeanagesbetween53and54years,aswellassimilarBMIs:28.5kg/m2inthetrainingsetand27.59kg/m2inthevalidationset.Inbothgroups,themajor-ityofpatientswerewhite:90.6%and96.3%,respec-tively.Most important, bothdata setshadbalancedPM/PMandEM/EMfrequencies:4.4%versus2.5%in PM/PM and 40.3% versus 44.5% in EM/EM.Therefore, demographic characteristics among thetrainingandvalidationdatasetsweresimilar.

CYP2D6 Gene Score as a Predictor of CYP2D6 Phenotype

All 3 scoring systems (ie, our proposed CYP2D6gene score and the system 1 and 2 gene scores)showedaweakcorrelationwiththeendoxifen/NDMplasmaratiointhetrainingdataset(R2=0.24,0.22,and 0.18, respectively; P < .00001), although ourproposed scoring system was slightly better thansystem1and2genescores.Inthevalidationset,ourproposed CYP2D6 gene score consistently showedthe strongest association with the CYP2D6 indexmeasure (R2 = 0.30 vs 0.24 and 0.15, respectively;

Table III DemographicsofParticipantsinaTrainingDataSet,aValidationDataSet,andBothSetsCombined

Data Sets

Training, n = 159 Validation, n = 81 Combined, n = 240

Age,y Mean 53.23 53.40 53.32 Median(range) 54(42,78) 55(43,81) 54(42,81)Meanbodymassindex,kg/m2(range) 28.5(23.6,34.5) 27.6(20.3,36.9) 28.2(20.3,36.9)Ethnicity,n(%) White 146(91.83) 78(96.30) 224(93.33) Black 9(5.66) 2(2.47) 11(4.58) Hispanic 1(0.63) 0(0.00) 1(0.42) Unknown 3(1.89) 1(2.47) 4(1.67)CYP2D6genotype,n(%) PM/PM 7(4.40) 2(2.47) 9(3.75) EM/EM 64(40.25) 36(44.45) 100(41.67)

EM,extensivemetabolism;PM,poormetabolism.

454 • J Clin Pharmacol 2010;50:450-458

BORGES ET AL

–1 0 1 2 3 4

0.02

0.05

0.10

0.20

0.50

Proposed CYP2D6 Gene Score

En

dox

ifen

/ND

M p

lasm

a ra

tio

r² = 0.24

–1 0 1 2 3 4

0.02

0.05

0.10

0.20

0.50

System 1 CYP2D6 Gene Score

r² = 0.22

–1 0 1 2 3 4

0.02

0.05

0.10

0.20

0.50

System 2 CYP2D6 Gene Score

r² = 0.18

–1 0 1 2 3 4

0.01

0.02

0.05

0.10

0.20

Proposed CYP2D6 Gene Score

En

dox

ifen

/ND

M p

lasm

a ra

tio

r² = 0.3

–1 0 1 2 3 4

0.01

0.02

0.05

0.10

0.20

System 1 CYP2D6 Gene Score

r² = 0.24

–1 0 1 2 3 4

0.01

0.02

0.05

0.10

0.20

System 2 CYP2D6 Gene Score

r² = 0.15

A

B

Figure 1. Gene score alone. (A) training, (B) validation. NDM, N-desmethyltamoxifen.

P<.00001).Inthevalidationset,themagnitudeofthe correlation remained low when the CYP2D6genescorealonewasused(Figure1).

CYP2D6 Activity Score as a Predictor of CYP2D6 Phenotype

We evaluated the ability of our proposed CYP2D6activityscore toestimate invivoCYP2D6activityasmeasuredbytheendoxifen/NDMplasmaratio.Becauseweused2differentmethodstocalculatetheCYP2D6inhibitor factor (withandwithoutCYP2D6genotypeinput),thereare2groupsofCYP2D6activityscoresforall3scoringsystems.InFigure2wepresentthecor-relation of endoxifen/NDM plasma ratio with the 3CYP2D6 activity score systems when the CYP2D6inhibitor factor iscalculatedwithoutconsidering theCYP2D6genotype.Asexpected,wheninformationontheconcomitantuseofCYP2D6inhibitorsisincorpo-ratedincalculationoftheCYP2D6activityscore,theassociationofall3scores(proposedsystem,system1,andsystem2)withtheendoxifen/NDMplasmaratio

almostdoubledinthetrainingdataset(R2=0.44,0.43,and 0.38 vs 0.24, 0.22, and 0.18, respectively). Inthevalidationdataset,asimilarpatternwasobserved(R2 = 0.53, 0.50, and 0.41 vs 0.30, 0.24, and 0.15).WhentheCYP2D6genotypewasconsideredindeter-mining the CYP2D6 inhibitor factor, performance oftheCYP2D6activityscoredidnotimprovecomparedwith the genotype-independent inhibition factor(Figure3),showingR2valuesalmostidenticaltothosedepictedinFigure2fortheproposedsystem,system1,andsystem2inthetraining(R2=0.43,0.42,and0.38,respectively)andvalidation(R2=0.52,0.48,and0.38,respectively)datasets.Table4showsasummaryoftheperformanceoftheproposedCYP2D6geneandactiv-ityscorescomparedwithsystems1and2inthetrain-ingandvalidationdatasetsandinbothsetscombined.

DISCUSSION

CYP2D6geneticvariationandtheeffectofCYP2D6inhibitorsonthepharmacokineticsofdrugsthatusethismetabolicroutehavebeenintensivelystudied.

SCORING SYSTEM FOR CYP2D6

DRUg INTeRACTIONS 455

–1 0 1 2 3 4

0.02

0.05

0.10

0.20

0.50

Proposed CYP2D6 Gene Score

En

dox

ifen

/ND

M p

lasm

a ra

tio

r² = 0.44

–1 0 1 2 3 4

0.02

0.05

0.10

0.20

0.50

System 1 CYP2D6 Gene Score

r² = 0.43

–1 0 1 2 3 4

0.02

0.05

0.10

0.20

0.50

System 2 CYP2D6 Gene Score

r² = 0.38

–1 0 1 2 3 4

0.01

0.02

0.05

0.10

0.20

Proposed CYP2D6 Gene Score

En

dox

ifen

/ND

M p

lasm

a ra

tio

r² = 0.53

–1 0 1 2 3 4 –1 0 1 2 3 4

0.01

0.02

0.05

0.10

0.20

System 1 CYP2D6 Gene Score

r² = 0.5 0.01

0.02

0.05

0.10

0.20

System 2 CYP2D6 Gene Score

r² = 0.41

A

B

Figure 2. Activity score no genetic input. NDM, N-desmethyltamoxifen.

However,somecharacteristicsoftheCYP2D6geneticpolymorphismand theCYP2D6gene–environmentinteractions have become important limitations totheclinicaluseofCYP2D6geneticinformation.TheincreasingnumberofCYP2D6allelesdescribedandtheirmanyalleliccombinationsmakeitdifficultformost clinicians to interpret CYP2D6 genotyperesults.Inaddition,withtheprobableexceptionofCYP2D6 poor metabolizers, the functional conse-quenceoftheCYP2D6geneticvariationismodulatedby other factors, such as physiological states likepregnancy, some herbal products, and concomitantuseofcertaindrugs.Therefore,inintegratingCYP2D6geneticinformationintothedecision-makingprocessin clinical practice, clinicians should interpret theCYP2D6genotypeintheclinicalcontextoftheindi-vidualpatient,takingintoaccountallpossibledeter-minants of CYP2D6 metabolic activity. Althoughsomeofthesedeterminantsareunknownorpoorlyunderstood,theCYP2D6genotypeandtheconcomi-tantuseofCYP2D6inhibitorsarewelldocumented

andarereadilyavailableforpatientsandtheirdoc-tors. Physicians often identify the importance ofCYP2D6 genetic variation and drug interactionswhen prescribing a CYP2D6 substrate. However,thereisnocurrentinstrumenttoquantifythisinter-action. Some progress has been made in groupingCYP2D6genotypesbyexpectedphenotypicexpres-sion,8,30,33,34 but this research has not consideredconcomitantuseofmedications.

Inthepresentstudy,weanalyzedforthefirsttimetheutilityofaCYP2D6scoringsystemthatincorpo-rates CYP2D6 genetic variation and CYP2D6-mediated drug–drug interactions. In our previousstudies, we have shown that tamoxifen is sequen-tiallymetabolizedbythecytochromeP450systemtoendoxifen. Specifically, we have shown thatendoxifenisformedfromNDM,themajorcirculat-ingmetaboliteoftamoxifen,byCYP2D6.Therefore,theratioofendoxifentoNDMwasusedasamarkerofCYP2D6activity.WehavedevelopedaCYP2D6genescorebasedontheanticipatedeffectofgenetic

456 • J Clin Pharmacol 2010;50:450-458

BORGES ET AL

–1 0 1 2 3 4

0.02

0.05

0.10

0.20

0.50

Proposed CYP2D6 Gene Score

En

dox

ifen

/ND

M p

lasm

a ra

tio

r² = 0.43

–1 0 1 2 3 4

0.02

0.05

0.10

0.20

0.50

System 1 CYP2D6 Gene Score

r² = 0.42

–1 0 1 2 3 4

0.02

0.05

0.10

0.20

0.50

System 2 CYP2D6 Gene Score

r² = 0.38

–1 0 1 2 3 4

0.01

0.02

0.05

0.10

0.20

Proposed CYP2D6 Gene Score

En

dox

ifen

/ND

M p

lasm

a ra

tio

r² = 0.52

–1 0 1 2 3 4

0.01

0.02

0.05

0.10

0.20

System 1 CYP2D6 Gene Score

r² = 0.48

–1 0 1 2 3 4

0.01

0.02

0.05

0.10

0.20

System 2 CYP2D6 Gene Score

r² = 0.38

A

B

Figure 3. Activity score genetic input. NDM, N-desmethyltamoxifen.

variationsontheCYP2D6enzymaticcapabilityandevaluated its correlation with the endoxifen/NDMplasmaratioasanindexmeasureofCYP2D6activ-ity.Asexpected,theCYP2D6genescorealonewaspoorlyassociatedwiththeCYP2D6phenotype.Thatwas also the case for the other CYP2D6 scoringapproaches(systems1and2)thatusedgeneticinfor-mationalone.However,whentheexpectedCYP2D6inhibitory effect was integrated with the CYP2D6genotypeintothecompositeCYP2D6activityscore,ourabilitytoestimatetheCYP2D6phenotypedou-bled.Interestingly,asimilarincreaseinthecorrela-tion of the score with the CYP2D6 index measurewasseenwiththeinclusionoftheCYP2D6inhibitorfactorintothegenotype-basedscoresdevelopedbyotherauthors.Theseresultshighlighttheheteroge-neityoftheCYP2D6phenotypeandtheimportanceofconsideringotherfactorsinadditiontothegeno-typeintheestimationoftheCYP2D6performanceinvivo.Intheprocessofintegratinginformationaboutthe CYP2D6 inhibitors into the CYP2D6 activity

score,wehypothesizedthattakingintoaccountthatCYP2D6genotypeinfluencesthecapacityofCYP2D6inhibitors to decrease metabolic activity of theCYP2D6enzymewouldallowamoreaccurateesti-mationofthisactivity.However,theCYP2D6geno-type input did not seem to alter the ability of theCYP2D6activityscoretopredicttheCYP2D6pheno-type.ApossibleexplanationisthatthegreatimpactoftheadditionoftheCYP2D6inhibitorfactortothegenescoredilutesthesmalleffectthatthegenotypeinputhasincalculatingtheCYP2D6inhibitorfactor.Therefore, the definition of the CYP2D6 inhibitorfactor without considering the CYP2D6 genotypeappears to be a simple and reliable method. Rightnow,theselectionofinhibitionfactorvaluesissub-jective.Abetterstrategycouldbeapharmacokinet-ics drug interaction model–based prediction thatintegrates genetics and inhibition mechanisms inscoreconstruction.Inthisarticle,wegroupedweakinhibitorsandmoderateinhibitorstogetherbecauseoftheirsubtledifference.Thedefinitionofweakor

SCORING SYSTEM FOR CYP2D6

DRUg INTeRACTIONS 457

Table IV CorrelationoftheProposedCYP2D6GeneandActivityScoresWithEndoxifen/N-DesmethyltamoxifenPlasmaRatioComparedWithSystem133andSystem28inaDerivation

DataSet,aValidationDataSet,andBothSetsCombined

Data Set

Derivation, n = 159

Validation, n = 81

Combined, n = 240

CYP2D6genescoresalone(noinputfromCYP2D6inhibitorfactor) ProposedCYP2D6genescoresystem 0.24 0.30 0.25 System1 0.22 0.24 0.22 System2 0.18 0.15 0.16CYP2D6activityscores(inputfromCYP2D6inhibitorfactornotgenotype-derived) ProposedCYP2D6genescoresystem 0.44 0.53 0.45 System1 0.43 0.50 0.43 System2 0.38 0.41 0.36CYP2D6activityscores(inputfromgenotype-derivedCYP2D6inhibitorfactor) ProposedCYP2D6genescoresystem 0.43 0.52 0.44 System1 0.42 0.48 0.42 System2 0.38 0.38 0.34

ResultsarepresentedinR2values.AllthePvaluesare<.00001forthesecorrelations.

moderate inhibitorsometimesdependsonthesub-stratesfromvariouspublishedinvitro/invivostud-iesandneedsfurtherexploration.

OurproposedCYP2D6activityscoreisanimprovedtooltoestimateCYP2D6performanceinvivo.Ourfind-ingshavebeenreplicatedinasecondsampleofindi-viduals.TheR2valuesinthevalidationdatasetwerecomparable to the training data set when testing theCYP2D6 gene score alone and both CYP2D6 activityscores.Inthevalidationsample,thecorrelationoftheendoxifen/NDMplasmaratiowiththeCYP2D6activityscorewasalsotwiceashighaswiththeCYP2D6genescore alone across all 3 scoring systems tested. Thisconsistency in the results in 2 data sets reflects thestrengthofthecorrelationbetweentheCYP2D6activityscoreandtheCYP2D6phenotype.However,otherfac-torsthatinfluenceCYP2D6activityarenotbeingcon-sidered, making the correlation still imperfect. Forexample, it has been shown that CYP2D6 genotypeshaveadifferentphenotypicexpressioninCaucasianscomparedwithAfricanAmericans.34Anethnicityfac-torcouldnotbetestedinoursamplesgiventhelackofethnicdiversityinourstudypopulation.Inthefuture,ethnicity,pregnancy,herbalproducts,andothervaria-bles known to influence CYP2D6 activity could beincluded in calculating the CYP2D6 activity score.Anotherlimitation,whichappliestoanyscoringsys-temusingCYP2D6geneticinformation,isthefactthatthe functionalconsequenceofmanyof thedescribedCYP2D6allelesisunknown,hencetheuncertaintyinestimating the CYP2D6 phenotype from genotype.

Despite these limitations, the CYP2D6 activityscorepresentedinthisworkshowsanimportantcor-relation with the CYP2D6 phenotype and representsa significant improvement over the use of CYP2D6genotypealone.

This study was supported in part by a PharmacogeneticsResearch Network grant, 2U-01 GM61373 (DAF, DFH, VS),K24RR020815 (DAF), GM74217 (LL); a clinical pharmacologytraininggrant5T32-GM-08425(DAF)fromtheNationalInstituteof General Medical Sciences, Bethesda, Maryland; grantM01-RR000042fromtheNationalCenterforResearchResources(NCRR),acomponentoftheNationalInstitutesofHealth(NIH);Damon Runyon-Lilly Clinical Investigator award CI-3 from theDamonRunyonCancerResearchFoundation (VS); andFashionFootwearFoundation/QVCPresentsShoesonSale(DFH).

REFERENCES

1. Ingelman-Sundberg M, Johansson I, Persson I, et al. GeneticpolymorphismofcytochromeP450:functionalconsequencesandpossible relationship to disease and alcohol toxicity. EXS.1994;71:197-207.2.Ingelman-SundbergM.PharmacogeneticsofcytochromeP450anditsapplicationsindrugtherapy:thepast,presentandfuture.Trends Pharmacol Sci.2004;25:193-200.3.DestaZ,WardBA,SoukhovaNV,FlockhartDA.Comprehensiveevaluation of tamoxifen sequential biotransformation by thehuman cytochrome P450 system in vitro: prominent roles forCYP3AandCYP2D6.J Pharmacol Exp Ther.2004;310:1062-1075.4.TyndaleRF,DrollKP,SellersEM.GeneticallydeficientCYP2D6metabolismprovidesprotectionagainstoralopiatedependence.Pharmacogenetics.1997;7:375-379.

458 • J Clin Pharmacol 2010;50:450-458

BORGES ET AL

5. Dahl ML, Johansson I, Palmertz MP, Ingelman-Sundberg M,SjoqvistF.AnalysisoftheCYP2D6geneinrelationtodebrisoquinand desipramine hydroxylation in a Swedish population. Clin Pharmacol Ther.1992;51:12-17.6.HouZY,PickleLW,MeyerPS,WoosleyRL.Salivaryanalysisfor determination of dextromethorphan metabolic phenotype.Clin Pharmacol Ther.1991;49:410-419.7. Sachse C, Brockmoller J, Bauer S, Roots I. Cytochrome P4502D6 variants in a Caucasian population: allele frequencies andphenotypicconsequences.Am J Hum Genet.1997;60:284-295.8. Zanger UM, Raimundo S, Eichelbaum M. Cytochrome P4502D6:overviewandupdateonpharmacology,genetics,biochemis-try.Naunyn Schmiedebergs Arch Pharmacol.2004;369:23-37.9.Ingelman-Sundberg M, Daly AK, Nebert DW. HumanCytochrome P450 (CYP) Allele Nomenclature Committee Website.www.imm.ki.se/CYPalleles.AccessedDecember24,2009.10. Dahl ML, Johansson I, Bertilsson L, Ingelman-Sundberg M,SjoqvistF.UltrarapidhydroxylationofdebrisoquineinaSwedishpopulation:analysisofthemoleculargeneticbasis.J Pharmacol Exp Ther.1995;274:516-520.11. Lundqvist E, Johansson I, Ingelman-Sundberg M. GeneticmechanismsforduplicationandmultiduplicationofthehumanCYP2D6geneandmethods fordetectionofduplicatedCYP2D6genes.Gene.1999;226:327-338.12. Michelson D, Read HA, Ruff DD, Witcher J, Zhang S,McCracken J. CYP2D6 and clinical response to atomoxetine inchildrenandadolescentswithADHD.J Am Acad Child Adolesc Psychiatry.2007;46:242-251.13.RauT,WohllebenG,WuttkeH,etal.CYP2D6genotype:impactonadverseeffectsandnonresponseduring treatmentwithantide-pressants—apilotstudy.Clin Pharmacol Ther.2004;75:386-393.14.SchillevoortI,deBoerA,vanderWeideJ,etal.Antipsychotic-inducedextrapyramidalsyndromesandcytochromeP4502D6geno-type:acase-controlstudy.Pharmacogenetics.2002;12:235-240.15.ShamsME,ArnethB,HiemkeC,etal.CYP2D6polymorphismandclinicaleffectoftheantidepressantvenlafaxine.J Clin Pharm Ther.2006;31:493-502.16.SomogyiAA,BarrattDT,CollerJK.Pharmacogeneticsofopio-ids.Clin Pharmacol Ther.2007;81:429-444.17.WuttkeH,RauT,HeideR,etal. Increasedfrequencyofcyto-chromeP4502D6poormetabolizersamongpatientswithmetopro-lol-associatedadverseeffects.Clin Pharmacol Ther.2002;72:429-437.18. Goetz MP, Knox SK, Suman VJ, et al. The impact of cyto-chrome P450 2D6 metabolism in women receiving adjuvanttamoxifen.Breast Cancer Res Treat.2007;101:113-121.19.KawanishiC,FurunoT,Kishida I,MatsumuraT,KosakaK.Apatientwithtreatment-resistantschizophreniaandcytochromeP4502D6geneduplication.Clin Genet.2002;61:152-154.20.KawanishiC,LundgrenS,AgrenH,BertilssonL.Increasedinci-denceofCYP2D6geneduplicationinpatientswithpersistentmooddisorders: ultrarapid metabolism of antidepressants as a cause ofnonresponse:apilotstudy.Eur J Clin Pharmacol.2004;59:803-807.21.KirchheinerJ,NickchenK,BauerM,etal.Pharmacogeneticsofantidepressantsandantipsychotics:thecontributionofallelicvariations to the phenotype of drug response. Mol Psychiatry.2004;9:442-473.

22.DalenP,FrengellC,DahlML,SjoqvistF.Quickonsetofsevereabdominal pain after codeine in an ultrarapid metabolizer ofdebrisoquine.Ther Drug Monit.1997;19:543-544.23.deLeonJ,DinsmoreL,WedlundP.AdversedrugreactionstooxycodoneandhydrocodoneinCYP2D6ultrarapidmetabolizers.J Clin Psychopharmacol.2003;23:420-421.24.GascheY,DaaliY,FathiM,etal.Codeineintoxicationassoci-ated with ultrarapid CYP2D6 metabolism. N Engl J Med.2004;351:2827-2831.25.KirchheinerJ,SchmidtH,TzvetkovM,etal.Pharmacokineticsofcodeineanditsmetabolitemorphineinultra-rapidmetabolizersduetoCYP2D6duplication.Pharmacogenomics J.2007;7:257-265.26. Koren G, Cairns J, Chitayat D, Gaedigk A, Leeder SJ.Pharmacogeneticsofmorphinepoisoninginabreastfedneonateofacodeine-prescribedmother.Lancet.2006;368:704.27.ChouWH,YanFX,deLeonJ,etal.Extensionofapilotstudy:impact from the cytochrome P450 2D6 polymorphism on out-come and costs associated with severe mental illness. J Clin Psychopharmacol.2000;20:246-251.28. Flockhart DA. Cytochrome P450 drug interactions. http://medicine.iupui.edu/clinpharm/ddis/.AccessDecember24,2009.29.JinY,DestaZ,StearnsV,etal.CYP2D6genotype,antidepres-santuse,andtamoxifenmetabolismduringadjuvantbreastcan-certreatment.J Natl Cancer Inst.2005;97:30-39.30.BorgesS,DestaZ,LiL,etal.QuantitativeeffectofCYP2D6genotype and inhibitors on tamoxifen metabolism: implicationforoptimizationofbreastcancertreatment.Clin Pharmacol Ther.2006;80:61-74.31.CaiWM,NikoloffDM,PanRM,etal.CYP2D6geneticvariationinhealthyadultsandpsychiatricAfrican-Americansubjects:impli-cationsforclinicalpracticeandgenetictesting.Pharmacogenomics J.2006;6:343-350.32. Li L, Pan RM, Porter TD, et al. New cytochrome P4502D6*56 allele identified by genotype/phenotype analysis ofcryopreservedhumanhepatocytes.Drug Metab Dispos.2006;34,1411-1416.33.ZinehI,BeitelsheesAL,GaedigkA,etal.PharmacokineticsandCYP2D6 genotypes do not predict metoprolol adverse events orefficacyinhypertension.Clin Pharmacol Ther.2004;76:536-544.34.GaedigkA,SimonSD,PearceRE,BradfordLD,KennedyMJ,LeederJS.TheCYP2D6activityscore:translatinggenotypeinfor-mationintoaqualitativemeasureofphenotype.Clin Pharmacol Ther.2008;83:234-242.35. Stearns V, Johnson MD, Rae JM, et al. Active tamoxifenmetabolite plasma concentrations after coadministration oftamoxifenandtheselectiveserotoninreuptakeinhibitorparoxet-ine.J Natl Cancer Inst.2003;95:1758-1764.36.FoodandDrugAdministration.Guidanceforindustrydruginter-actionstudies—studydesign,dataanalysis,andimplicationsfordos-ingandlabeling(DraftGuidance).September2006.http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm072101.pdf.AccessedDecember24,2009.37. Laine K, Tybring G, Hartter S, et al. Inhibition of cytochromeP4502D6activitywithparoxetinenormalizestheultrarapidmetabo-lizerphenotypeasmeasuredbynortriptylinepharmacokineticsandthedebrisoquintest.Clin Pharmacol Ther.2001;70:327-335.