ebook img

Race, Ethnicity and Ancestry in Unrelated Transplant Matching for the National Marrow Donor PDF

15 Pages·2015·0.39 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Race, Ethnicity and Ancestry in Unrelated Transplant Matching for the National Marrow Donor

RESEARCHARTICLE Race, Ethnicity and Ancestry in Unrelated Transplant Matching for the National Marrow Donor Program: A Comparison of Multiple Forms of Self-Identification with Genetics JillA.Hollenbach1*,AliyaSaperstein2,MarkAlbrecht3,CynthiaVierra-Green4, PeterParham5,6,PaulJ.Norman5,6,MartinMaiers3 1 DepartmentofNeurology,UniversityofCalifornia,SanFranciscoSchoolofMedicine,SanFrancisco, California,UnitedStatesofAmerica,2 DepartmentofSociology,StanfordUniversity,Stanford,California, UnitedStatesofAmerica,3 BioinformaticsResearch,NationalMarrowDonorProgram,Minneapolis, Minnesota,UnitedStatesofAmerica,4 CenterforInternationalBloodandMarrowTransplantResearch, Minneapolis,Minnesota,UnitedStatesofAmerica,5 DepartmentofStructuralBiology,StanfordUniversity SchoolofMedicine,Stanford,California,UnitedStatesofAmerica,6 DepartmentofMicrobiology& Immunology,StanfordUniversitySchoolofMedicine,Stanford,California,UnitedStatesofAmerica OPENACCESS * [email protected] Citation:HollenbachJA,SapersteinA,AlbrechtM, Vierra-GreenC,ParhamP,NormanPJ,etal.(2015) Race,EthnicityandAncestryinUnrelatedTransplant Abstract MatchingfortheNationalMarrowDonorProgram:A ComparisonofMultipleFormsofSelf-Identification withGenetics.PLoSONE10(8):e0135960. Weconductedanationwidestudycomparingself-identificationtogeneticancestryclassifi- doi:10.1371/journal.pone.0135960 cationsinalargecohort(n=1752)fromtheNationalMarrowDonorProgram.Wesoughtto Editor:TesfayeBMersha,CincinnatiChildren's determinehowvariousmeasuresofself-identificationintersectwithgeneticancestry,with HospitalMedicalcenter,UNITEDSTATES theaimofimprovingmatchingalgorithmsforunrelatedbonemarrowtransplant.Multiple Received:March1,2015 dimensionsofself-identification,includingrace/ethnicityandgeographicancestrywere comparedtoclassificationsbasedonancestryinformativemarkers(AIMs),andthehuman Accepted:July28,2015 leukocyteantigen(HLA)genes,whicharerequiredfortransplantmatching.Nearly20%of Published:August19,2015 responseswereinconsistentbetweenreportingrace/ethnicityversusgeographicancestry. Copyright:©2015Hollenbachetal.Thisisanopen DespitestrongconcordancebetweenAIMsandHLA,nomeasureofself-identification accessarticledistributedunderthetermsofthe showscompletecorrespondencewithgeneticancestry.Incertaincasesgeographicances- CreativeCommonsAttributionLicense,whichpermits unrestricteduse,distribution,andreproductioninany tryreportingmatchesgeneticancestrynotreflectedinrace/ethnicityidentification,butin medium,providedtheoriginalauthorandsourceare othercasesgeographicancestriesshowlittlecorrespondencetogeneticmeasures,with credited. importantdifferencesbygender.However,whenrespondentsassignancestrytograndpar- DataAvailabilityStatement:Ourinformedconsent ents,weobservesub-groupsofindividualswithwell-definedgeneticancestries,including stipulatedthatsurveydatawillonlybereportedin importantdifferencesinHLAfrequencies,withimplicationsfortransplantmatching.While aggregateform.SNPallelefrequenciesandHLA weadvocatefortailoredquestioningtoimproveaccuracyofancestryascertainment,collec- haplotypefrequenciesforthefulldatasetwillbemade availableuponrequest([email protected]). tionofdonorgrandparents’informationwillimprovethechancesoffindingmatchesfor manypatients,particularlyformixed-ancestryindividuals. Funding:ThisworkwasfundedbytheOfficeof NavalResearchN00014-12-1-0142(MA,CVG,MM). Thefundershadnoroleinstudydesign,data collectionandanalysis,decisiontopublish,or preparationofthemanuscript. PLOSONE|DOI:10.1371/journal.pone.0135960 August19,2015 1/15 Race,EthnicityandAncestryinUnrelatedTransplantMatching CompetingInterests:Theauthorshavedeclared Introduction thatnocompetinginterestsexist. TheNationalMarrowDonorProgram(NMDP)registry[1]istherepositoryfordataforthe highlypolymorphichumanleukocyteantigen(HLA)genesfromovertwelvemillionpotential donorsforhematopoieticstemcelltransplant(HSCT).However,duetochangesovertimein technologyandthecostsassociatedwithhighresolutionHLAgenotyping,theregistrydatafor potentialdonorsisoftenoflowerresolutionthanrequiredformatchingwithpatients.Over tenthousandHLAallelesarecurrentlyrecognized,withmorethan2000allelesforsomegenes [2],andworldwidedistributionpatternscorrespondtospecificpopulationswithshareddemo- graphichistory[3].Accordingly,theNMDPfacilitatesinitialselectionofpotentiallyHLA- matcheddonorsbyclassifyingthemusingtheirself-identifiedrace/ethnicityfromastandard- izedquestionnaire,incorporatingthisinformationinabioinformaticsprocessthatprojects donors’mostlikelyhigh-resolutionHLAhaplotypesandpatientmatchprobabilities[4]. However,racialandethnicself-identificationamongAmericansisacomplexprocessthat drawsoninformationaboutknownancestry,appearance,howpeoplewereraised,wherethey grewup,whethertheyexperienceddiscrimination,andbywhom;andanindividual’s responsescanalsochangeovertime[5–9].SomeAmericansdonotreportallfacetsoftheir ancestrywhentheyareaskedtoself-identify,eveniftheyknowthem(e.g.,althoughPresident BarackObamaoftenacknowledgeshiswhiterelatives[10],heidentifiedonlyasblackonthe 2010Census[11]).Othersmayadjusttheirresponsesbasedonwhattheyassumeisthepurpose ofthedatacollection[12].Reportingofgeographicoriginsorancestrycanbeequallysubjective, asmoreAmericansreportcommonorigins(e.g.Irish)thanishistoricallyfeasible[13].Thus, theextenttowhichself-identifiedrace/ethnicityorgeographicancestrywillcorrespondto geneticancestryisalsolikelytovaryand,apriori,itisimportanttonotprivilegeanyonemea- sureofrace/ethnicityorancestryoveranyotherwithoutformal,rigoroustesting.Although manystudieshaveexaminedtherelationshipbetweenself-identifiedrace/ethnicityandgenetic ancestry[14–20],previousworkhasinvolvedonlyasingledimensionofself-identification withoutconsiderationofalternativemeasures. Here,wepresentthefirstsimultaneousassessmentofmultipleapproachestomeasuring self-reportedancestrywithdirectcomparisonofsurveyresponseswithgeneticmarkers,and examineimplicationsforunrelateddonorHLAmatchinginHSCT.Weconductedanation- widestudycomparingseveralformsofself-identificationtogeneticclassifications,inalarge cohortfromtheNMDP.Wecapturedmultiplefacetsofself-identification,includingrace/eth- nicityandgeographicancestry.FormanyAmericansthesedimensionswillyieldconsistent responses,butforotherseachmeasureproducesadifferentresponse[21].Ouraimwastochar- acterizetheextenttowhichthesedifferentself-identificationmethodscorrespondnotonlyto eachotherbutalsotogeneticmeasures,withtheobjectiveofimprovingcurrentpracticesfor selectingunrelateddonorsforbonemarrowtransplantsintheUnitedStates. MaterialsandMethods Ourcohortincludes1752individualswhorespondedpositivelytoaletterofrequestforpartici- pationinthestudybymailingbackbothacheekswabandacompletedpaper-and-pencilsur- veyquestionnaire.Recipientsoftheinvitationletterwereselectedrandomlywithrespectto originalself-identificationfromtheNMDPdonorpool.Inordertoassessconcordanceofself- identificationwithhigh-resolutionHLAgenotype,onlydonorsforwhomsequence-leveldata wereavailableforfiveHLAloci(HLA-A,-B,-C,DRB1,DQB1)wererecruited.Becauseweare interestedinimprovingclassificationfordonorswithrareHLAtypes,weoversampled(2:1) individualswhoseHLAgenotypeisunderrepresentedintheregistry(fewerthan11copies among12million).TheNationalMarrowDonorPrograminstitutionalreviewboardapproved PLOSONE|DOI:10.1371/journal.pone.0135960 August19,2015 2/15 Race,EthnicityandAncestryinUnrelatedTransplantMatching Table1. Surveyrespondentdemographics. Total Female Male Meanage Allrespondents 1752 1324 428 31 Originalregistryclassification AfricanAmerican 27 22 5 30 White 1414 1090 324 31 AsianorPacificIslander 65 33 32 30 Hispanic 100 70 30 31 NativeAmerican 4 4 0 30 Multi-race 142 105 37 30 RespondentUSborn No 100 61 39 33 Yes 1647 1259 388 30 ParentsUSBorn Neither 159 95 64 31 One 145 114 31 31 Both 1444 1113 331 31 GrandparentsUSborn None 177 111 66 32 One 28 23 5 29 Two 172 134 38 32 Three 133 102 31 29 Four 1225 944 281 31 doi:10.1371/journal.pone.0135960.t001 thestudy(Study#0292);allsubjectsgavewritteninformedconsent.Responseratesbyoriginal registryrace/ethnicityclassificationandgenderaregiveninSupplementalmaterials(S1Table). Thebaselinedemographicsofsurveyrespondents,includingtheoriginalrace/ethnicityclassifi- cation(basedontheexistingregistryquestionnaire;S1Fig),nativityofrespondent,parents andgrandparents,ageandgenderaregiveninTable1.Althoughpresentedontheoriginal donorformasaseparateethnicityselection,individualsselecting“Hispanic”aretreatedas equivalenttothemajorracialcategoriesbytheregistry,andareconsideredassuchhere. Inonepartofthequestionnaire(S2Fig),weaskedrespondentshowtheyself-identifywith respecttoraceandethnicity:“Markoneormoreboxestoshowtheracialorethnicgroup(s)you usetodescribeyourself.”Thelistofpossibleresponsesincludedthefivemajorracialcategories (AmericanIndianorAlaskaNative;Asian;BlackorAfricanAmerican;NativeHawaiianor otherPacificIslander;White)recommendedbytheOfficeofManagementandBudget(OMB) inits1997directive[22].Wealsoincluded“HispanicorLatino”andgavetheoptiontomark “other.”Asimilarinstrumentiscurrentlybeingtestedforthe2020Census[12].Inthesame section,weaskedrespondentstotellustheirperceivedracialclassificationbyothers(“Howdo otherpeopleinthiscountrytypicallyclassifyyou?”).Thisquestionwasbasedonaninstrument developedbytheCenterforDiseaseControl’sMeasuresofRacismWorkingGroupforthe annualBehavioralRiskFactorSurveillanceSystemsurvey(http://www.cdc.gov/brfss/). Wealsoaskedwhetherrespondents,theirparentsortheirgrandparentswereborninthe UnitedStates,followedbyaquestionabouttheirgeographicancestry:“Fromwhatcountriesor partsoftheworlddidyourancestorscome?”Theseinstrumentsweremodeledonitemsusedby theGeneralSocialSurveysince1972[23].Respondentswereaskedtochooseasmanycatego- riesasnecessarytodescribetheirancestryfromalistofcountries(e.g.,Cuba,Vietnam), regionsoftheworld(e.g.,MiddleEast,Caribbean),andothercommonancestriesamong PLOSONE|DOI:10.1371/journal.pone.0135960 August19,2015 3/15 Race,EthnicityandAncestryinUnrelatedTransplantMatching Americans(“AfricanAmerican”or“NorthAmericanIndian”).Thelistof24countriescom- binedthemostcommonancestriesreportedinthemostrecentAmericanCommunitySurvey, andthecountriesfromwhichthemostimmigrantscametotheUnitedStatesinthelast decade.Weofferedregionsoftheworldtocapturecountriesnotspecificallyincluded,aswell astoprovideoptionsforpeoplewhomightonlyknowgeneralinformationaboutfamilyori- gins.Finally,weaskedrespondentstoprovidethespecificcountry,regionorancestryfromthe samelistthatappliedtoeachoftheirfourgrandparents(“Ifpossible,intheboxesbelow,enter thebestorigincodeforeachofyourgrandparents”)usingnumericalcodesadjacenttoeachof thegeographicancestryoptions.Inordertocomparethisself-identifiedgeographicancestryto self-identifiedrace/ethnicityandmeasuresofgeneticancestry,weaggregatedspecificcountry orregionresponsesintobroadercategories(S2Table),inkeepingwithfederalguidelineson racial/ethnicclassificationintheUnitedStates[22]Forexample,wehavecombinedallAsian geographicancestries(South,Southeast,East)underoneterm,“Asian,”asisdonebytheU.S. Census[24]. Awell-characterizedpanel[25]of93ancestryinformativemarkers(AIMs)[26,27](S3 Table)wasgenotypedforallrespondentsusingtheSequenomiPLEXassay.Failedsinglenucle- otidepolymorphism(SNP)callsunderwentasecondgenotypingusingtheTaqmanassay (ABI).Fivepercentofthesamplesservedasqualitycontrolsandweretypedblindinduplicate. AllsamplesandSNPsmetstandardqualitycontrolchecks,includingfittoexpectationsunder Hardy-WeinbergEquilibriumandlessthan10%missingnessorfailurerates.WiththeHuman GenomeDiversityPanel(HGDP)publicdataset[28]asareference,SNPgenotypeswereused inancestryestimationswithStructuresoftware[29].ParametersfortheStructurerunwereas follows:k=4,popflag=1,burnin=10,000andreps=10,000.The‘popflag’optionenableduse oftheHGDPpopulationsasthetrainingset,whilemultiplerunsindicatedthatk=4clusters allowedthebestresolutionofbroadcontinentalgroups(African,European[clusteredwith MiddleEastern],AsianandAmerindian),facilitatedbyuseoftheHGDPpopulations.Aplot oftheStructurerunshowingfourdistinctclustersisgiveninS3FigWedefinedsubpopulations byrace/ethnicityorgeographicancestryresponses(e.g.,allindividualsthatselected“Hispanic orLatino”aloneorincombinationwithotherrace/ethnicitychoices;orallindividualsreport- inganyEuropeangeographicancestry).Themeanproportionofeachgeneticancestry(based ontheAIMs)wascomputedforthesesubpopulations,andtheWilcoxonMann-Whitneytest usedtotestdifferencesindistributionsofancestryproportionsbetweengroups.Allstatistical analysesandplottingwereperformedintheRlanguageandenvironmentforstatisticalcom- puting[30]. Intransplant,theprimarygenesofinterestareHLA.Toexaminecorrespondencebetween self-identification,AIMsandHLA,aBayesianclassifier[31]customscriptedinRwasusedto assignthemostprobablecontinentaloriginforsubjects’HLAhaplotypes.Priorprobabilities formarginaldistributionofclassificationswerecomputedfromproportionsoftheoriginalreg- istryclassificationsforthecohort.Foreachsubpopulationdefinedbyquestionnaireresponses, frequenciesofAfrican,European.AsianandAmerindianHLAhaplotypeswerecalculated. Correlationsbetweenself-identificationmeasuresandgeneticancestryproportions(calculated fromAIMs)andHLAhaplotypeorigins,aswellascorrelationbetweengeneticancestrypro- portionsandHLAorigins,werecomputedusingthe‘corr.test’functionintheRpackage ‘psych.’Foreachcorrelationcoefficient,95%confidenceintervalsandp-valueswerecomputed. Allp-valueswerecorrectedformultipletests[32]. Throughoutthemanuscript,wereferto“race/ethnicity”and“geographicancestry”to describemeasuresofself-identification.Theterm“geneticancestry”referstocontinental ancestryproportionsascertainedthroughAIMsandStructureanalysis;and“HLAorigin” referstothecontinentaloriginforHLAhaplotypesdeterminedviatheBayesianclassification. PLOSONE|DOI:10.1371/journal.pone.0135960 August19,2015 4/15 Race,EthnicityandAncestryinUnrelatedTransplantMatching Afulldescriptionofthesetermsandtheirpossiblevaluesisgiveninsupplementalmaterials (S4Table). Results Self-identificationoftenvariesbyreportingformat Whereasallsubjectsself-identifiedbyrace/ethnicity,3%didnotprovideresponsesregarding geographicancestry,and19%gavewhatcouldbeconsideredinconsistentresponsesforthese twoquestions(S5Table).ApproximatelyhalfoftheinconsistentresponsesreportedAmerican Indianancestrywithoutidentifyingtheirrace/ethnicityasAmericanIndian.Mostoften,lack ofconcordancebetweengeographicancestryandrace/ethnicityreportingoccurswhenindivid- ualsacknowledgeparticulargeographicancestriesbutdonotexplicitlyidentifywithacorre- spondingracial/ethnicgroup.Lesscommonisrace/ethnicityself-identificationwithout reportingacorrespondinggeographicancestry.Consequently,manymoreindividualsreported multiplegeographicancestries(19%)thanmultiplerace/ethnicities(7%).Thevastmajorityof individualswhoreportedasinglerace/ethnicityalsoreportedthatotherpeopletypicallyper- ceivethemthesameway.However,consistentwithevidencefromothernationallyrepresenta- tivesurveys[9],themajorityofindividualswhoreportedmultiplerace/ethnicitiesreported thatotherpeopletypicallyperceivethemasWhite. Correspondenceofself-identificationmeasureswithgeneticancestry andHLAorigin Nomeasureofself-identificationshowscompletecorrespondencewithanyspecificpatternof geneticancestry,underscoringtheunclearboundariesandimprecisenatureofthesecategori- zations,whichareoftenpresumedtorepresentseparateandidentifiablegroups.Wefindevi- denceofEuropeangeneticancestryamongnearlyallsub-populations,asexpectedinaU.S. cohort(Table2).However,theproportionofEuropeangeneticancestryrangesbroadly,with lowestaveragevaluesobservedamongindividualswhoidentifytheirrace/ethnicityAsian.Afri- cangeneticancestryproportionsrangefrom20–95%forindividualswhoidentifytheirrace/ ethnicityas“BlackorAfricanAmerican,”withtheremainingcomponentlargelyEuropean. Forindividualswhoself-identifytheirrace/ethnicityas“HispanicorLatino”weobservemean EuropeanandAmerindiangeneticancestriesof71%and21%,respectively,butthedefining featureoftheseclassificationsisthattheproportionofeachgeneticancestryvariesbroadly betweenindividualsevenwhentheyself-identifythesameway. Acrossallclassificationsandquestionformats,wefindveryhighcorrelationbetweenthe meanproportionofAfrican,European,AsianandAmerindiangeneticancestryineachsub- population,andthefrequencyoftherespectiveHLAhaplotypesoriginsinthoseindividuals (r=0.97,0.79,0.97,0.95,respectively;Table2).Thisdemonstratesconcordanceofpopulation levelgeneticancestryassessedviaAIMsandHLAorigin.Thestrongestcorrelationbetween anytwomeasuresisbetweenself-reportedAsianrace/ethnicityandself-reportedAsiangeo- graphicancestry(r=0.94;CI=0.93–0.94;p<10−15),andeachofthesemeasuresissimilarly correlatedwithAsianHLAoriginandgeneticancestry(Table3).Sub-SaharanAfricangeo- graphicancestryandracial/ethnicself-identificationas“BlackorAfricanAmerican”arealso highlycorrelated(r=0.83;CI=0.81–0.84;p<10−15),butthelattertrackssomewhatmore closelywithAfricanHLAoriginandgeneticancestry(Table3).Wefindlower,yetsignificant correspondencebetweenEuropeangeographicancestryandself-identificationasWhite (r=0.68;CI=0.66–0.71;p<10−15),withrace/ethnicitybeingslightlymorepredictiveofEuro- peanHLAoriginandgeneticancestry(Table3).IndividualsreportingLatinAmerican PLOSONE|DOI:10.1371/journal.pone.0135960 August19,2015 5/15 Race,EthnicityandAncestryinUnrelatedTransplantMatching Table2. MeangeneticancestryproportionsdeterminedviaancestryinformativemarkersandHLAhaplotypeoriginfrequenciesforsubpopula- tionsdefinedbyreportedrace/ethnicityorgeographicancestry. N Meangeneticancestryproportion HLAhaplotypeorigin African European Asian Amerindian African European Asian Amerindian Multi- American origin* Race/EthnicitySelf-ID:Single 1624 choicerespondents BlackorAfricanAmerican 27 0.70 0.27 0.03 0.01 63% 11% 4% - 22% White 1443 0.00 0.99 0.00 0.00 - 82% - - 17% Asian 64 0.01 0.22 0.76 0.01 - - 97% - 3% HispanicorLatino 71 0.05 0.63 0.05 0.28 - 13% 1% 52% 34% AmericanIndianorAlaska 2 0.00 0.99 0.01 0.00 - 50% - - 50% Native NativeHawaiianorother 2 0.01 0.00 0.97 0.02 - - 100% - - PacificIslander Geographicancestry:Single 1373 originrespondents AfricaorAfricanAmerican 15 0.71 0.25 0.03 0.02 67% 13% 7% - 13% Europe 1236 0.00 0.99 0.00 0.00 - 82% - - 17% Asia/Pacific 63 0.01 0.21 0.77 0.01 - - 97% - 3% NorthAmericanIndian 6 0.00 0.91 0.01 0.08 - 20% - - 80% LatinAmerica 36 0.04 0.51 0.07 0.38 - - - 72% 28% Caribbean 17 0.19 0.71 0.03 0.07 6% - 6% 12% 77% Race/EthnicitySelf-ID: 124 0.04 0.84 0.06 0.05 5% 48% 6% 6% 35% Multiplechoicerespondents BlackorAfricanAmerican 13 0.36 0.60 0.03 0.02 54% 8% - - 39% White 120 0.03 0.85 0.06 0.05 3% 48% 7% 6% 36% Asian 18 0.00 0.66 0.31 0.02 - 22% 33% 6% 39% HispanicorLatino 46 0.03 0.84 0.03 0.11 4% 37% - 13% 46% AmericanIndianorAlaska 53 0.04 0.91 0.02 0.03 6% 66% 2% 4% 23% Native NativeHawaiianorother 5 0.01 0.71 0.23 0.05 - 20% 60% - 20% PacificIslander Geographicancestry:Multiple 379 0.04 0.89 0.03 0.04 4% 63% 3% 6% 25% originrespondents AfricaorAfricanAmerican 27 0.40 0.57 0.02 0.02 41% 22% - - 37% Europe 318 0.02 0.91 0.03 0.04 2% 64% 4% 5% 25% Asia/Pacific 25 0.01 0.70 0.28 0.02 - 28% 36% 4% 32% NorthAmericanIndian 49 0.04 0.91 0.02 0.03 2% 72% 1% 3% 23% LatinAmerica 68 0.05 0.78 0.02 0.15 6% 40% - 22% 32% Caribbean 21 0.19 0.75 0.04 0.02 14% 48% 5% - 33% Classificationbyothers BlackorAfricanAmerican 31 0.71 0.25 0.03 0.01 68% 10% - - 23% White 1545 0.00 0.98 0.01 0.01 - 80% - - 18% Asian 70 0.01 0.24 0.74 0.01 - 7% 90% 1% 1% HispanicorLatino 73 0.05 0.63 0.05 0.28 1% 10% 1% 52% 36% AmericanIndianorAlaska 6 0.01 0.74 0.11 0.14 - 67% - - 33% Native NativeHawaiianorother 3 0.01 0.67 0.31 0.01 - 33% 33% - 33% PacificIslander (Continued) PLOSONE|DOI:10.1371/journal.pone.0135960 August19,2015 6/15 Race,EthnicityandAncestryinUnrelatedTransplantMatching Table2. (Continued) N Meangeneticancestryproportion HLAhaplotypeorigin African European Asian Amerindian African European Asian Amerindian Multi- American origin* Other 23 0.10 0.71 0.13 0.05 9% 26% 26% 4% 35% *AllindividualshavetwoHLAhaplotypes.Multi-originhaplotypeclassificationindicatesthatoneoftheindividual’shaplotypesiscloselyassociatedwith onecontinentaloriginwhiletheotherhaplotypeisassociatedwithadifferentcontinentalorigin. doi:10.1371/journal.pone.0135960.t002 geographicancestry(e.g.,Mexico,Guatemala,CentralorSouthAmerica)tendtoself-identify theirrace/ethnicityas“HispanicorLatino”(r=0.76;CI=0.73–0.77;p<10−15)andhaveAmer- indianHLAoriginandgeneticancestry(Table3),whilethosereportingCaribbeangeographic ancestry(e.g.,PuertoRico)donotidentifytheirrace/ethnicityasHispanicasfrequently (r=0.37;CI=0.33.0.41;p<10−15),havemuchlowerproportionsofAmerindiangeneticances- try,andarelesslikelytohaveAmerindianHLAorigin(Table3).NorthAmericanIndiangeo- graphicancestryiscorrelatedwithreportingmultiplerace/ethnicitiesorancestries(r=0.80; CI=0.78–0.82;p<10−15),butnotanyspecificgeneticancestry. WhiledistinguishingbetweenreportedLatinAmericanandCaribbeangeographicancestry affordsadvantageoverHispanicracial/ethnicidentificationinrepresentinggeneticancestry, othergeographicancestriesaremoreweaklyrelatedtocomparablegeneticclassifications.For example,3%ofrespondentsreportedEuropeangeographicancestryinadditiontootherori- gins,butdidnotidentifyasWhitewithrespecttorace/ethnicity(n=52).Ofthose,nearlytwo- Table3. Correlationbetweenself-identificationmeasuresandHLAoriginsandgeneticancestry. Self-identification HLAorigins Geneticancestry(AIMs) measure African European Asian Amerindian African European Asian Amerindian Race/ethnicity BlackorAfrican 0.71(0.68–0.74) 0.90(0.89–0.91) American White 0.74(0.72–0.77) 0.77(0.74–0.79) Asian 0.84(0.82–0.85) 0.84(0.82–0.86) HispanicorLatino 0.58(0.54–0.62) 0.66(0.63–0.69) NativeHawaiianor 0.14(0.09–0.19) 0.15(0.10–0.21) otherPacific Islander Geographicancestry AfricanorAfrican 0.69(0.66–0.72) 0.82(0.80–0.83) American European 0.72(0.69–0.75) 0.75(0.73–0.77) Asian 0.83(0.81–0.85) 0.83(0.81–0.85) LatinAmerican 0.54(0.50–0.58) 0.69(0.66–0.72) Caribbean 0.13(0.07–0.18) 0.20(0.14–0.25) Grandparentsancestry AfricanorAfrican 0.72(0.69–0.74) 0.86(0.85–0.88) American European 0.77(0.75–0.79) 0.81(0.79–0.83) Asian 0.88(0.87–0.89) 0.89(0.88–0.90) LatinAmerican 0.62(0.59–0.66) 0.80(0.78–0.82) Caribbean 0.10(0.05–0.16) 0.17(0.120.23) 0.18(0.12–0.23) Thecorrelationcoefficient(r)isgivenwith95%confidenceintervals(parentheses).Onlystatisticallysignificant(p<0.05aftercorrection)valuesareshown. doi:10.1371/journal.pone.0135960.t003 PLOSONE|DOI:10.1371/journal.pone.0135960 August19,2015 7/15 Race,EthnicityandAncestryinUnrelatedTransplantMatching thirdsself-identifiedtheirrace/ethnicityasHispanic(n=28)orAfricanAmerican(n=4); however,meanEuropeangeneticancestryforself-identifiedHispanicsandAfricanAmericans whoreportedEuropeangeographicancestry(0.73and0.34,respectively)isnotstatisticallysig- nificantlyhigherthanthatforindividualswiththesameself-identifiedrace/ethnicitywhodo notreportEuropeangeographicancestry(0.63and0.27). GenderdifferencesinreportingconsistencyhaveimplicationsforHLA matching Insomecaseswhengeographicancestryyieldslessconsistentclassificationsrelativetogenetic ancestrythanthosebasedonrace/ethnicity,wefindthedifferencesalsovarybygender.More than10percentofourcohortreportedNorthAmericanIndiangeographicancestrybutdid notself-identifyasAmericanIndianwithrespecttorace/ethnicity(n=189).ThemeanAmer- indiangeneticancestryproportion(0.018)fortheseindividualsissimilartothosenotreporting NorthAmericanIndianancestry(0.019).Strikingly,significantlymorewomen(n=164;12%) thanmen(n=25;6%)reportNorthAmericanIndianancestryinadditiontoanothergeo- graphicancestrywithoutalsoidentifyingtheirrace/ethnicityasAmericanIndian(p<0.001). Further,andofmostimportanceintransplantmatching,selectionofmorethanonerace/eth- nicityorgeographicancestrybymenismuchmorepredictiveofmulti-originHLAthanmulti- originreportingamongwomen,eventhoughmenarelesslikelytoidentifywithmultiplerace/ ethnicitiesorgeographicancestries.Forexample,28%ofmenwhoself-identifywithtwoor morerace/ethnicitygroupshavemulti-originHLAhaplotypes;incontrast,just9%ofwomen selectingmorethanonerace/ethnicitygrouphavemulti-originHLAhaplotypes(p<0.05). Grandparents’geographicancestrymostcloselycorrelatedwithHLA haplotypeorigins Incontrasttoreportinggenericfamilyorigins,whenrespondentsselectoriginsforspecific grandparentstheresultingclassificationusingthenumberofgrandparentswithagivengeo- graphicancestryidentifiessub-groupsofrespondentswithmoreuniformgeneticbackgrounds. Sub-classificationbythenumberofgrandparentswithSub-SaharanAfricanorAfricanAmeri- canancestry,forexample,showsaclearpatternofincreasingAfricangeneticancestrypropor- tionswithincreasingnumberofreportedAfrican-ancestrygrandparents(Fig1A).The substructureismirroredinHLAoriginfrequencies,identifyinggroupsofindividualswithdif- ferentcombinationsofHLAhaplotypes.Thisresulthasimportantimplicationsforthetrans- plantmatchalgorithm’sassignmentoflikelyhighresolutionHLAhaplotypes.Forexample, welloverhalfoftheindividualswhoreportedfourAfrican-ancestrygrandparentshavetwo AfricanHLAhaplotypes;incontrast,threeofthefourindividualswhoreportedtwoAfrican- ancestrygrandparentshaveonlyoneAfricanHLAhaplotype(Fig1B).Theresultsaresimilar forindividualswhoreportedgrandparentsfromLatinAmericanorigins,withrespecttoAmer- indiangeneticancestryproportionsandHLA(Fig1Cand1D).ThesefindingsholdforEuro- peanandAsiangeneticancestryproportionsandHLAoriginfrequenciesaswell(notshown), demonstratingthevalueincollectinggrandparents’ancestryforallindividualsasasubstantial improvementtocurrentracial/ethnicself-identificationpracticesforprospectivedonors. Discussion Inmedicalpracticeandbiomedicalresearch,subjects’self-identifiedrace/ethnicityisfre- quentlycollected[33–35],oftenservingasaproxyforgeneticancestry.Atthesametime,there isongoingdiscourseamongphysicians,researchersandthepublicabouthowbesttocollect PLOSONE|DOI:10.1371/journal.pone.0135960 August19,2015 8/15 Race,EthnicityandAncestryinUnrelatedTransplantMatching Fig1.AfricanandAmerindiangeneticancestryproportions(a,c)andHLAoriginfrequencies(b,d)are correlatedwiththenumberofgrandparentswithreportedAfricanandLatinAmericanancestry, respectively.GeneticancestryproportionswereestimatedfromAIMsdatausingStructureandtheHGDP referenceset.Forthisanalysis,k=4,reflectingbroadcontinentalancestry.Foreachsubpopulationdefined bythenumberofrespondents’grandparentswithSub-SaharanAfrican/AfricanAmericanorLatinAmerican ancestry,thepercentageofindividualswithzero,oneortwoAfricanorAmerindianHLAhaplotypes, respectively,wascalculated.Individualsreportingone,two,threeorfourAfrican-ancestrygrandparents: n=4;4;2;21,respectively.Individualsreportingone,two,threeorfourLatinAmerican-ancestry grandparents:n=16;21;7;38,respectively. doi:10.1371/journal.pone.0135960.g001 thiscriticalinformation[36,37],therelationshipbetweenracialclassificationandgeneticances- try[38–46],itspossibleconfoundingroleinbiomedicalresearch[16,47]anditsapplicationin healthcaresettings.UsingAIMs,manystudieshaveshowngoodcorrelationbetweengenetics andself-identificationforsomeracialandethniccategoriesintheUnitedStates[16,18],buta limitedcorrespondenceforindividualswhoidentifyasHispanicormulti-racial[14,19,48]. Geneticclusteringofworldwidehumanpopulations[27]suggeststhatidentificationaccording togeographicancestrymaybemorebiologicallyrelevantthanthestandardrace/ethnicitycate- goriesusedinfederaldatacollectionandreporting.Yet,manyremainwaryofassumptions aboutinherentdifferenceandinternalhomogeneitythathaveoftencharacterizedtheuseof raceandancestryidentificationinbiomedicalfields[49,50].Toooften,however,thesediscus- sionstakeplaceatanabstractlevel–aboutwhatshouldandshouldnotbedoneingeneral,or underidealcircumstances.Equallyimportantiscarefulexaminationofthepracticalityandrel- evanceofdifferentmeasuresofancestryidentificationforspecificmedicalapplicationswhere thesedataarecurrentlyrelieduponforpatientcare:suchasmatchingdonorsandrecipients fortransplant. Withtheaimofimprovingmatchingalgorithmsforbonemarrowtransplant,thisstudy soughttodeterminehowvariousmeasuresofself-identificationintersectwithgeneticancestry. Ourfindingthatnearlyoneinfiverespondentsprovidedanswersinconsistentbetween reportedrace/ethnicityandgeographicancestryhighlightsthesubjectivenatureofself-identifi- cation.Incomparingmeasuresofself-identification,wedonotmeantoimplythatwethink responsesshouldalwaysbeconsistentacrossmeasuresorthatonetypeofresponseismore PLOSONE|DOI:10.1371/journal.pone.0135960 August19,2015 9/15 Race,EthnicityandAncestryinUnrelatedTransplantMatching “correct”thananother.Rather,wewanttounderscorehowdifferentmeasuresofself-identifi- cationrevealdifferentinformation.Moreover,itisapparentthatthereisnoclear‘winner’in thecomplexprocessofself-identificationintheU.S.withrespecttogeneticancestry;bestprac- ticesforcollectingthisinformationforsomepopulationsmaybeineffectiveforothers.Unrav- elingthemyriadhistorical,politicalandsocialfactorscontributingtothesedifferencesamong measuresofself-identification,andbetweenself-reportedandgeneticancestrymeasures,is beyondthescopeofthiswork.Instead,ourresultspointtotheneedformoreresearchtoiden- tifythespecificnatureofthisvariationandrefinehowwecollectandinterpretmeasuresof race/ethnicityandgeographicancestryinmedicalcontexts. Asexpected,incertaincases,geographicancestryreportingmorecloselycorrespondsto geneticancestrycomparedtoself-identificationusingstandardrace/ethnicitycategories.While oftenclassedtogetheras“Hispanic”byregistriesandinmuchbiomedicalresearch,individuals reportingLatinAmericanoriginsdisplaymarkedlydifferentproportionsofAfricanandAmer- indiangeneticancestrythanthosewhoreportCaribbeanorigins[51,52];ourresultsareconcor- dantwiththesefindingsand,mostimportantfortheregistry,mirroredinthefrequencyof HLAsubclasses.Incontrast,reportedEuropeangeographicancestryprovideslimitedaddi- tionalinsightintogeneticancestryproportionsformost.ThereportingofNorthAmerican Indiangeographicancestryisalsogenerallynotconsistentwithgeneticclassifications.These individualsmaybereportingverydistantancestrythathassurvivedinfamilylore,orthismay reflectpeopleselectingsymbolicresponsesnotbasedonspecificknowledgeoffamilyorigins [6,13].ThatmostrespondentsreportingAfricanAmericanracialidentificationdidnotreport sub-SaharanAfricangeographicancestryalsosuggeststhatgeographicancestryisnotaform ofself-identificationthatresonateswitheveryone. Startingin2000,theU.S.Censusquestionnaireallowedrespondentstoselectmorethanone self-identifiedrace.In2010,over9millionAmericansidentifiedwithtwoormoreraces,a32% increasefromthepreviousdecade,andasubstantialfractionofourcohortreportsmultiple race/ethnicitiesorgeographicancestries.Likewise,individualswhoself-identifyasHispanic nowcomprisethelargestracial/ethnicgroupwithintheUnitedStatesafternon-Hispanic Whites[53,54].Inourstudy,grandparents’ancestryisthemostinformativepieceofdatawith respecttomorepreciselycharacterizinggeneticancestryforindividualsfrompopulationslong recognizedashighlyheterogeneous[48,52],andcollectionofthisinformationshouldimprove theefficiencyandreliabilityoffindingHLA-identicaldonorsformanypatients.Althoughnot currentlycollectedbytheregistry,inthehealthcaresetting,orformostbiomedicalresearch, inoursample>70%respondentsprovidedageographicancestryforallfourgrandparents. Becausereportingforeachgrandparentwaslimitedtoonlyoneoriginorancestry,these responsesmaybeparticularlysalient,derivingfromdirectknowledgeofacloserelative.We suggestthatregistriesconsidercollectingthesevaluabledataroutinely.Incontrast,whenasked to“checkallthatapply”todescribeone’sfamilyoriginsingeneral,thethresholdforclaiming particulargeographicancestriesmaybemuchlower,andtheresponsesarenotconstrainedto fourwithequalweight.However,giventhatnotallrespondentsareabletoprovidedatafor theirgrandparents,thismeasureshouldbeusedincombinationwithothers.Providingthe meansforindividualstoquantifyorotherwisedescribetherelativesalienceoftheirancestries mayalsobeofvalueinfuturestudies. Amajorlimitationhereisover-representationofindividualswhoself-identifyasWhite comparedtothecurrentU.S.populationandanover-representationofindividualswithEuro- pean-originHLAtypescomparedtotheoverallmakeupoftheregistry;additionalworkinthis areawillprofitfromstudyoflargersamplesofsubpopulationswithnon-Europeanoriginsand ancestries.Wefocusedonlow-frequencyHLAhaplotypesinthisstudyunderaworkingprem- isethatindividualswithcomplexancestries,whichmightnotbefullycapturedbythecurrent PLOSONE|DOI:10.1371/journal.pone.0135960 August19,2015 10/15

Description:
try reporting matches genetic ancestry not reflected in race/ethnicity identification, but in R: A language and environment for statistical computing.
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.