The effect of HLA on immunological response to hepatitis B vaccine in healthy people: A meta-analysis

(1)

ContentslistsavailableatSciVerseScienceDirect

Vaccine

j o ur na l h o me pa g e : w w w . e l s e v i e r . c o m / l o c a te / v a c c i n e

The

effect

of

HLA

on

immunological

response

to

hepatitis

B

vaccine

in

healthy

people:

A

meta-analysis

Zheng-Kang

Li

1

_,

_Jing-Jing

_Nie

1

_,

_Jie

_Li

∗

_,

_Hui

_Zhuang

∗

DepartmentofMicrobiologyandInfectiousDiseaseCenter,SchoolofBasicMedicalSciences,PekingUniversityHealthScienceCenter,Beijing100191,China

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received1March2013

Receivedinrevisedform23April2013

Accepted28June2013

Available online 23 July 2013 Keywords: HLA-DRB1 HLA-DQB1 Healthypeople Meta-analysis HepatitisBvaccine

a

b

s

t

r

a

c

t

Backgroundandaim:Evidenceisaccumulatingthatseveralmarkersinthehumanleukocyteantigen (HLA)regionhavebeenassociatedwithdecreasedorincreasedantibodyresponsetohepatitisBvaccine indifferentindividuals.Thismeta-analysisistoassesstheassociationsofHLAclassIIDRB1andDQB1 alleleswithimmunologicresponsetohepatitisBvaccineinhealthypeople.

Methods:Asystematicreviewofcohortstudiesinhealthypeoplewasperformed.Wesearcheddatabases forrelevantstudiesthatwerepublishedinEnglishorChineseuptoFebruary17,2012.Oddsratios (ORs)withcorresponding95%confidenceintervals(CIs)ofHLAallelesresponsetohepatitisBvaccine werepooledbyusingofafixed-effectsorrandom-effectsmodeldependingonabsenceorpresenceof significantheterogeneity.Allstatisticaltestsweretwo-sided.

Results:Fifteenstudieswereincludedinthismeta-analysisafterscanning774potentiallyrelevant arti-cles.Atotalof2308subjects(including1215responders,873nonrespondersand220controlpopulations) wereincluded.ForDRB1alleles,pooledORsshowedthatthreeHLAvariants,DRB1*01,DRB1*1301and DRB1*15wereassociatedwithasignificantincreaseantibodyresponsetohepatitisBvaccine,their pooledORswere2.73,5.94and2.29respectively.WhileDRB1*03(DRB1*0301),DRB1*04,DRB1*07 andDRB1*1302wereopposite,theirpooledORswere0.55(0.42),0.57,0.24and0.25respectively.And forDQB1alleles,pooledORsshowedthatDQB1*05(DQB1*0501),DQB1*06,DQB1*0602wereassociated withasignificantincreaseantibodyresponsetohepatitisBvaccine.TheirmergerORswere1.85,2.35, 2.34and3.32respectively.WhileDQB1*02(pooledOR=0.27)wasadverse.Sensitivityandspecificity analysisofHLAallelesshowedthatDRB1*1301andDQB1*0602hadhighspecificity(94.2%and90.1%) butlowsensitivity(25.1%and26.3%),respectively.

Conclusion:ItwassuggestedthatspeciﬁcHLAclassIIalleles(DRB1andDQB1)wereassociatedwith antibodyresponsetoHepB.

1. Introduction

Sincetheearly1980s,hepatitisBvaccine(HepB)hasbeen

avail-able.Its effectiveness and safetyhave been provedsufﬁciently.

Despite ofthetype/dose of vaccine,orroute ofadministration,

about 5%–15% of allindividuals fail toseroconvert toanti-HBs

>10mIU/mLwiththestandardvaccinationschedule[1].Several

Abbreviations: HBV,hepatitisBvirus;HLA,humanleukocyteantigen;HepB,

hepatitisBvaccine;ORs,Oddsratios;CIs,conﬁdenceintervals;MHC,major

histo-compatibilitycomplex;IHWS,InternationalHistocompatibilityWorkshop.

∗ Correspondingauthors.Tel.:+861082805133;fax:+861082805133.

E-mailaddresses:jielipku@gmail.com(J.Li),zhuangbmu@126.com(H.Zhuang).

1 _These_authors_contributed_equally_to_this_article.

factorsarepossibletoaffectadverselytheantibodyresponseto

HepBincludingmalegender,advancingage,overweight,

smok-ing,immunosuppression,andimmunodeﬁciency[2].Inaddition,

evidencesare accumulatingthatseveralmarkers in thehuman

leukocyteantigen(HLA)regionhavebeenassociatedwithantibody

responsetoHepBindifferentindividuals[3–5].

TheHLAarecoded bythemajorhistocompatibilitycomplex

(MHC)groupofgeneslocatedonchromosomesixinthehuman

genome.HLAmoleculesaredividedintothreeclasses,HLAclassI,

classIIandclassIII.Amongthem,HLAclassIImoleculespresent

antigensfromoutsideofthecelltoT-lymphocytes.These

partic-ularantigensstimulatethemultiplicationofT-helpercells,which

inturnstimulateantibody-producingB-cellstoproduce

antibod-iestothatspeciﬁcantigen.Someinvestigatorshavesuggestedthat

severalHLAclassIIallelesareinvolvedintheimmuneresponseto

hepatitisBantigen.Forexample,individualsharboredHLAclassII

alleleDRB1*0701,DRB1*0301werelesslikelytohavea

serolog-icalresponsetoHepB[6],whileindividualsharboredDRB1*1301

http://dx.doi.org/10.1016/j.vaccine.2013.06.108

Open access under CC BY-NC-ND license.

(2)

Table1

Qualitycriteriafortheincludedstudies.

Qualityparameter Score

2 1 0

No.ofcases >100 50—100 <50

QualitycontrolofPCR DNAsequencingorRepeat Internalcontrol ND

HLAcontrolgroup Yes No Notscored

Explanationofparticipantsselected Detail Simple Unknown

Matchingofresonpendersandnonresponders

Confoundergroup1 HBsAgandHBVDNA HBsAgorHBVDNA ND

Confoundergroup2 Ageandsex Ageorsex ND

Note:Detail:participantsshouldbeexcludedwithpreviousoractivehepatitisBvirus(HBV)infection,hepatitisCvirusinfection,HIVinfection,chronicliverorkidney disease,autoimmunedisease,cancer,pregnancyorlactationandallergytovaccinecomponents,smoking(20/day),alcoholintake(>40g/day),intravenoussubstanceabuse orregularimmunosuppressantmedication;simple:participantsincludedwerenegativeforHBsAg,HBeAg,anti-HBc,andHBVDNA,ruledoutthepossibilitiesofprevious oractivehepatitisBvirus(HBV)andotherhepatitisvirusinjection,gotridofotherchronicdiseases;unknown:onlyhealthypeoplewithoutanyotherinformations;DNA consequence:thesequencingresultswerecomparedwiththecorrespondingnucleotidesequencesoftheNCBIGenBank(BLAST);repeat:checkingtheresultwithmore thanonemethod;internalcontrol:agenedisplayedhighlyuniformexpressioninlivingorganismsduringvariousphasesofdevelopmentandunderdifferentenvironment conditions;ND:Undetectedorunknown.

weremorelikelytohaveaserologicalresponsetoHepB[6].

How-ever,otherresearchersdidnotﬁndsuchanassociation[5,7].The

inconsistenceofthesedatamightbecauseofthesamplesizeand

statisticalmethodsappliedinthesestudies.

In this study, the meta-analysis was adopted to provide

enhancedstatisticalpowerfortheinterpretationofpublisheddata,

andtogenerateasummaryestimateoftheeffectofdifferentHLA

onHepBresponse.

2. Methods

2.1. Searchstrategyandselectioncriteria

We searched MEDLINE (1966 to February 17, 2011),

EMBASE (1950 to February 17, 2011), Cochrane library

(http://www.cochrane.org), China National Knowledge

Infor-mation database (1985 to February 17, 2011) and Chinese

Wanfang Database using the Medical Subject Heading (MeSH)

terms“HepatitisBVaccines,”“HLAAntigens,”“Histocompatibility

Antigens,”andtheindividualcorrespondingfreeterms.In

addi-tion,we performedmanualsearches byscanning thereference

listsoftheincludedarticlestolocateadditionalpapersrelatedto

thetopic.We alsocontactedtheauthorsofsomestudieswhen

necessary.

Thefollowinginclusivecriteriaweresetandreviewedbytwo

independentinvestigators(ZKLiandJJNie):(1)recombinant

vac-cine;(2)study’smethodbasedonacohortstudydesign;(3)the

studiesshowinginterestsintheassociationbetweenHLAalleles

andresponsetoHepB(thevaccineswereclassiﬁedintoresponders

(anti-HBs≥10IU/L)and nonresponders(anti-HBs<10IU/L)after

thewholevaccineschedule;(4)followingthestandardvaccination

withcurrentlylicensedhepatitisBrecombinantvaccines

(immu-nizedat0,1,6monthsoraddedonedoseafterthreedoses);and

(5)oddsratios(ORs)incohortstudiesreportedwiththe95%

conﬁ-denceintervals(CI)(or,if95%CIswerenotavailable,thereported

dataweresufﬁcienttocalculatethem).Weexcludedthestudies:(1)

notpublishedasafulltext;(2)thenumbersofincludedsubjects

werelessthan10;(3)studieswithoutgroupsofnonresponders

orwithinconformitygroup(thevalueofcut-offofanti-HBsdid

notmeetourscriterion);(4)thesubjectswereunhealthy;and(5)

reviews,notoriginalpapers.Ifmorethanoneversionofthesame

studywasretrieved,onlythemostrecentpublishedstudyorthe

studywithlargernumberofparticipantswasused.Iftheearlier

paperfocusedonagivenHLAallelesandthesubsequentpaper(s)

examinedadditionalHLAallelesbythesameauthorinthesame

studypopulation,thetwopaperswereregardedasonestudy[8,9].

2.2. Dataextraction

Datawereindependently extractedandcrosscheckedbytwo

researchers(ZKLiandJJNie),usingastandardizedform.The

dis-crepancieswereresolvedbyconsensus.Thefollowingdatawere

collectedfromeach study:ﬁrstauthor;publication year;

coun-tryorarea;studysubjects;numberofresponders,nonresponders

andcontrols; HLAgenotypingandanti-HBsdetection methods;

HLAclassIIalleles(HLA-DRB1andHLA-DQB1)andpotential

con-founders(age,sex,HBsAgandHBVDNA).

2.3. Assessmentofstudyquality

Twoinvestigators(ZKLiand JJNie)independentlyratedthe

qualityofeachretrievedstudybyusinga12-pointscoringsystem

thatwasdesignedforthismeta-analysis.Thescoringsystemwas

basedonfactorsthatmightaffectthequalityoftheseobservational

studies(Table1)[10],includingcasenumber,qualitycontrolofPCR,

HLAcontrolgroup,explanationofparticipantsselectedand

match-ingofrespondersandnonresponders.Studieswithanoverallscore

of9orhigherwereclassiﬁedashigh-qualitystudies.Thosewith

anoverallscoreof5–8wereclassiﬁedasmedium-qualitystudies.

Andstudieswithanoverallscoreof4orlowerwereclassiﬁed

low-qualitystudies.Akappaidentitytestwascarriedouttocheckthe

consistency.Disagreementwasresolvedbyconsensus.

2.4. Statisticalanalysis

ThestatisticalworkwasperformedbyusingRevManManager

5.0software(CochraneCollaboration).Heterogeneitywas

evalu-atedbythemeansofX2 _test,_P_values_and_I2_statistics._I2 _values

of25%,50%and 75%wereassigned aslow,moderate,and high

estimates,respectively[11].Studieswereconsideredtohaveno

statistical heterogeneitywhen P≥0.1, and a ﬁxedeffect model

(Mantel–Haenszelmethod)wasusedtomergedata.Studieswere

consideredtohaveheterogeneousifP<0.1,andarandomeffects

mode (DerSimonian and Laird method) was used [12]. When

signiﬁcantheterogeneitywasdetected intheallele (numberof

studies>2),sensitivityanalysiswasconductedbyremovingone

studyata time toevaluatewhethertheresultsweremarkedly

affectedbyasinglestudy.Pooledoddsratioswiththe

correspond-ing95%conﬁdenceintervals(95%CIs)werecalculated.Publication

biaseswereassessesbyEgger’stest(STATAversion12.0)and

fun-nelplots.ThecriteriaforEgger’stextswererequirementofatleast

threeormorestudies.APvalueof<0.05wasconsideredstatistical

(3)

765

citations identified

by database search and

9

citation

s identified

by citation list

774 records screened

104

full-text articles assessed for e

ligibility

670 articles excluded after

title and abstract review

(including duplicate studies)

77 articles excluded:

1 duplication of data

2 not relevant to hepatitis B vaccine or HLA

3 reportsnot retrieved

5 not recombinant vaccine

8 non-healthy population

12 In vitro experiments or animal

experiments

13 letters or editorials or systematic

reviews or conference papers

27 studies included in qualitative analysis

12 articles excluded:

2 immune strategy not meet the

inclusion criteria

4 detection of anti-HBs not meet the

inclusion criteria

6 not suitable for meta-analysis

15 studies included in the meta-analysis

Fig.1.Flowchartofarticleselection.

3. Results

3.1. Thecharacteristicsandtherankoftheincludedstudies

Weidentiﬁed774potentiallyrelevantarticlesthrough

search-ingthepublishedliterature.Atotalof15articleswereincludedin

thismeta-analysis(Fig.1).Themainfeaturesofthestudiesincluded

inthismeta-analysisareshowninTable2.Ofthe15included

arti-cles,10werefromChina[7–9,13–19],4fromEuropean[6,20–22]

and1fromIran[5].Sevenstudiesexcludedthepossibilityofthe

lowlevelofHBVDNAinfectionleadingtononresponsethrough

detectingHBVDNA.Factorsofageandsexwerematchedortested

insigniﬁcantlyin11studies,othersmatchedonlyonefactorornot

atall.Atotalof2308(including1215responders,873

nonrespon-dersand220controlpopulation)subjectswereenrolledin this

study.ThemostcommonlyreportedHLAallelesassociatedwith

responsetoHepBwereHLA-DRB1andHLA-DQB1alleles.According

totheassessingcriteriaofstudyquality,6studieswereclassiﬁed

ashigh-qualitystudies,6weresortedasmedium-qualitystudies,

and2wereregardedaslow-qualitystudies.Thevalueofkappatest

foragreementbetweenthetworatersonoverallquality

evalua-tionwas0.784.OnlystudiespublishedinEnglishandChinesewere

includedinthismeta-analysis. _Table

2 Characteristics of studies included in the meta-analysis. First author Country or area Study subjects Matching factors (year) [reference] Sex (male/female) Mean age, y HBsAg HBV DNA R/NR/control HLA genotyping HLA alleles Detection of anti-HBs Quality score Hohler T (2002) [6] Germany Adults 102/24 36 (R) 38(NR) + – 53/73/89 PCR-SSP DRB1 – ≥ 9 Hohler T (1998) [20] Germany Caucasian adults – – – – 53/73 PCR-SSP DRB1 RIA 1–4 McDermott AB (1997) [22] UK Caucasian – 19–67 – – 117/86/131 PCR-SSP; PCR-SSO DRB1; DQB1 EIA 5–8 Lango-Warensjo A (1998) [21] Sweden Nonsmoking individuals 3/119 40.6 ± 10.3 (R); 41.8 ± 10.1 (NR) – – 69/53 PCR-SSO; PCR-RFLP DRB1; DQB1 EIA 5–8 Zhang JL (2011) [7] China College students 56/106 18–22 + + 78/84 PCR-SSP DRB1 ELISA ≥ 9 Wei YH (2009) [8,9] China College students 108/127 21–24 + + 136/99 PCR-SSP DRB1 ELISA ≥ 9 Yang X (2009) [15] China College students 81/80 – + + 96/65 PCR-SSP DRB1 ELISA 5–8 Liao XY (2006) [13] China Children + 9.23 + + 91/8 – DRB1; DQB1 RIA 5–8 Amirzargar AA (2008) [5] Iran Adults 48/10 27 (R); 31 (NR) + – 46/12 PCR-SSP DRB1; DQB1 ELISA 5–8 Hu DB (2009) [14] China Adults + + + – 20/20 PCR-SSP DRB1; DQB1 RIA 1–4 Gong XG (2009) [16] China College students 98/75 18–21 + – 148/25 PCR-SSP DRB1 ELISA ≥ 9 Yu YQ (2009) [17] China College students 64/78 21–24 + + 62/80 PCR-SSP DRB1 ELISA ≥ 9 Li LL(2010) [18] China College students 82/96 21–24 + + 101/77 PCR-SSP DRB1 ELISA ≥ 9 Qian Y (2002) [19] China College students + 17–20 + + 145/118 PCR-SSP DRB1 ELISA 5–8 Note: + matched or texted insignificantly; –: data not available; R: responders; NR: nonresponders; PCR-SSP: polymerase chain reaction-sequence specific primers; INNO-LIPA: reverse hybridization-based line probe assay; PCR-RFLP: polymerase chain reaction-restriction fragment length polymorphism analysis; PCR-SSO: polymerase chain reaction-sequence specific oligonucleotide; RIA: radioimmunoassay; EIA: enzyme immunoassay; ELISA: enzyme-linked immunosorbent assay.

(4)

Table3

SummaryofassociationsbetweenHLAallelesandresponsetohepatitisBvaccine,pooledoddsratios(OR)withcorresponding95%conﬁdenceintervals(95%CI)ofthe

publishedresultswerecalculatedwheremorethanonestudyinvestigatedtheallele.

HLAtypes No.of Heterogeneity Model Associationwithresponse Egger’s2-tailedbias

Study X2_value _P_value _I2 _P_value _OR _95%CI _P_value _95%CI _T_value

DRB1*01 3 0.00 1.00 0 Fixed 0.0005 2.73 1.55–4.81 0.411 −121.79to98.74 −1.33 DRB1*03 4 1.12 0.77 0 Fixed 0.009 0.55 0.35–0.86 0.189 −32.62to12.22 −1.96 DRB1*0301 2 0.00 1.00 0 Fixed 0.01 0.42 0.21–0.84 ND DRB1*04 5 0.29 0.99 0 Fixed 0.009 0.57 0.37–0.87 0.448 −21.00to11.99 −0.87 DRB1*07 8 16.03 0.02 56 Random <0.00001 0.24 0.14–0.43 0.155 −202.28to40.72 −1.63 DRB1*11 6 7.39 0.19 32 Fixed 0.12 1.42 0.91–2.22 0.785 −81.93to66.35 −0.29 DRB1*13 7 14.23 0.03 58 Random 0.20 1.58 0.79–3.19 0.124 −427.51to69.96 −1.85 DRB1*1301 3 1.51 0.47 0 Fixed <0.0001 5.94 2.65–13.32 0.777 −366.98to346.51 −0.36 DRB1*1302 2 1.06 0.30 5 Fixed 0.007 0.25 0.09–0.68 ND DRB1*14 6 17.45 0.004 71 Random 0.94 1.04 0.31–3.56 0.341 −345.19to151.71 −1.08 DRB1*15 7 7.54 0.27 20 Fixed <0.00001 2.29 1.61–3.27 0.949 −29.54to31.13 0.07 DQB1*02 4 5.30 0.15 43 Fixed <0.00001 0.28 0.18–0.42 0.327 −40.31to21.75 −1.29 DQB1*03 4 0.91 0.82 0 Fixed 0.24 1.27 0.85–1.91 0.956 −2.96to3.04 0.06 DQB1*0301 2 0.44 0.51 0 Fixed 0.93 1.02 0.59–1.77 ND DQB1*0302 3 5.09 0.08 61 Random 0.52 0.74 0.29–1.88 0.475 −2402.69to2025.52 −1.08 DQB1*0303 3 0.02 0.99 0 Fixed 0.19 1.91 0.72–5.05 0.475 −2402.69to2025.52 −1.08 DQB1*04 4 1.98 0.58 0 Fixed 0.94 0.97 0.39–2.40 0.472 −1100.87to727.09 −0.88 DQB1*05 4 0.65 0.88 0 Fixed 0.008 1.85 1.17–2.93 0.029 2.67to18.92 5.72 DQB1*0501 2 0.22 0.64 0 Fixed 0.03 2.35 1.09–5.05 ND DQB1*0503 2 2.59 0.11 61 Fixed 0.98 1.02 0.23–4.51 ND DQB1*06 4 7.76 0.05 61 Random 0.03 2.34 1.09–5.01 0.529 −38.23to26.83 −0.75 DQB1*0602 3 0.20 0.91 0 Fixed 0.0001 3.32 1.80–6.15 0.917 −225.53to220.91 −0.13 DQB1*0603 2 3.79 0.05 74 Random 0.44 1.92 0.36–10.24 ND DQB1*0604 3 6.15 0.05 67 Random 0.45 0.52 0.10–2.80 0.600 −1067.92to1197.53 0.73

Note:ND:notperformedwhenthereislessthanthreestudies;Boldlineindicatessigniﬁcantassociation.

3.2. AssociationofHLA-DRB1alleleswithantibodyresponseto

HepB

Eighttwo-digitnumberingofHLA-DRB1alleleswereincludedin

themeta-analysisasshowninTable3,andthefour-digit

number-ingofDRB1*03,DRB1*13allelesarealsolistedinTable3.Signiﬁcant

heterogeneitywereobservedinthreealleles(DRB1*07,DRB1*13,

DRB1*14),andtheseallelesweremoderateheterogeneity.Pooled

ORsshowedthatthree HLAvariants(DRB1*01, DRB1*1301 and

DRB1*15)wereassociatedwithasigniﬁcantincreaseofantibody

responsetoHepB,whileDRB1*03(DRB1*0301),DRB1*04,DRB1*07

andDRB1*1302wereassociatedwithasigniﬁcantdecreaseof

anti-bodyresponsetoHepB.Statisticallysigniﬁcantpublicationbiaswas

notfoundinotherDRB1alleles.Thefunnelplotswere

symmetri-cal.ThepublicationbiasofDRB1*0301andDRB1*1302couldnot

beperformedduetotheinsufﬁcientnumbersofarticles(lessthan

threestudies).

Theresultsofsensitivityanalysisforalleleswithsigniﬁcant

het-erogeneityareshowninTable4.Twostudies(Gongetal.[16]and

Zhangetal.[7])wereresponsiblefortheheterogeneityofDRB1*07,

thestudyofHohleretal.[6]wasthesourceoftheheterogeneityof

DRB1*13,andthestudyofZhangetal.[7]ledtotheheterogeneity

forDRB1*14.Thestatisticalresultsdidnotchangeafterremoving

correspondingstudies.

3.3. AssociationofHLA-DQB1alleleswithantibodyresponseto

HepB

Fivetwo-digitnumberingofHLA-DQB1alleleswereincluded

in the meta-analysis as shown in Table 3, and the four-digit

numbering ofDQB1*03, DQB1*05 and DQB1*06 allelesare also

listed in Table 3. Signiﬁcant heterogeneity of DQB1*0302 and

DQB1*06(DQB1*0603andDQB1*0604)wasobserved.Thesealleles

hadmoderate heterogeneity.PooledORsshowedthat DQB1*05

(DQB1*0501)andDQB1*06(DQB1*0602)wereassociatedwitha

signiﬁcantincreaseofantibodyresponsetoHepB,whileDQB1*02

wereassociatedwithasigniﬁcantdecreaseofantibodyresponse

toHepB.ByusingEgger’stest,statisticallysigniﬁcantpublication

biaswasfoundinDQB1*05.ThepublicationbiasesofDQB1*0301,

DQB1*0501,DQB1*0503andDQB1*0603couldnotbeperformed

becauseoftheinsufﬁcientnumbersofarticles(lessthanthree

stud-ies).

TheresultsofsensitivityanalysiswereshowninTable4.Two

studies(Lango-Warensjoetal.[21]andMcdermottetal.[22])were

responsiblefortheheterogeneityofDQB1*0302andDQB1*0604.

Andthe studyofAmirzargaret al.[5] ledtotheheterogeneity

ofDQB1*06.ThestatisticalresultsofDQB1*0302andDQB1*0604

werenot applicable for heterogeneity.The statistical resultsof

DQB1*06didnotchangeafterremovingcorrespondingstudiesin

sensitivityanalysis.

3.4. Thesubgroupanalysisbasedondifferentethnic

Consideringthepossibilitythatethnicdifferencemightaffect

the response to HepB [23,24]. HLA alleles associated with

responsetoHepBwereanalyzedindifferentethnicitysubgroups

as follows. Considering the number of references and

sam-ples,DRB1*03,DRB1*04,DRB1*07,DRB1*11,DRB1*13,DRB1*14,

DRB1*15,DQB1*02, DQB1*03,DQB1*04, DQB1*05and DQB1*06

alleles were included in the subgroup analysis. The results

showedthatDRB1*03,DRB1*04,DRB1*11,DQB1*02,DQB1*05and

DQB1*06werestatisticalsigniﬁcantinCaucasiangroup,butnot

inAsian ethnicity.However,theresultofDRB1*15wasadverse

withtheabovealleles.Inaddition,inbothraces,thesame

statis-ticallysigniﬁcantresultswereobservedintheallelesofDRB1*07,

DRB1*13,DRB1*14,DQB1*03andDQB1*04.TheORvaluesdidnot

change(OR>1or<1)inthealleleswithstatisticalsigniﬁcancein

subgroupanalysis.Soethnicitywasanimportantfactortoinﬂuence

theresponsetoHepBinthegeneticanalysis,whichisconsistent

withsomeresearches[25–27].

3.5. SensitivityandspeciﬁcityofHLAallelesinhealthypopulation

forthepredictionofresponsetoHepB

The sensitivity and speciﬁcity analysis of HLA alleles

(5)

Table4

Resultsofsensitivityanalysis.

Alleles Originofheterogeneity OR(95%CI) PvalueforOR Pvalueforheterogeneitytext I2

DRB1*07 Gongetal.[16],Zhangetal.[7] 0.25(0.17–0.37) <0.00001 0.59 0

DRB1*13 Hohleretal.[6] 1.14(0.74–1.76) 0.55 0.80 0

DRB1*14 Zhangetal.[7] 1.85(1.00–3.45) 0.05 0.31 16

DQB1*0302 Lango-Warensjoetal.[21] NA

Mcdermottetal.[22]

DQB1*06 Amirzargaretal.[5] 1.84(1.18–2.89) 0.008 0.39 0

DQB1*0604 Lango-Warensjoetal.[21]Mcdermottetal.[22] NA

NA:notapplicableforheterogeneity.

predictionofresponsetoHepB.Weevaluatedthefrequenciesof

DRB1*01,DRB1*03,DRB1*0301,DRB1*04,DRB1*07,DRB1*1301,

DRB1*1302,DRB1*15,DQB1*02,DQB1*05,DQB1*0501,DQB1*06

and DQB1*0602 (which were proved to be statistically

signif-icant in Table 3) alone using data extracted from all included

studies as potential biomarkers for the prediction of response

to HepB. Among them, DRB1*1301 and DQB1*0602 had high

speciﬁcity(94.2%[95%CI=90.3–98.1],90.1%[95%CI=85.3–94.9]),

butlowsensitivity(25.1%[95%CI=18.5–31.7],26.3%[20.6–32.0],

respectively)(Table5).

4. Discussion

HLAassociationswithresponsetoHepBhavebeenintensively

investigatedacrosstheglobalpopulations[28].Hohleretal.[29]

conductedastudyon202twinpairsindicatingthatgenetic

fac-torsmightaccountfornearly60%ofimmuneresponsivenessto

HepB,ofwhichHLAgeneswerethemajorcontributors.Wehave

attemptedtoreviewthepublishedliteratures withthepurpose

ofidentifyingcommonHLAalleleaffectingtheantibodyresponse

of HepB. Our systematic meta-analysis showed that DRB1 *03

(DRB1*0301),DRB1*04,DRB1*07,DRB1*1302andDQB1*02were

associatedwithnonresponsetoHepB,whileDRB1*01,DRB1*1301,

DRB1*15,DQB1*05(DQB1*0501)andDQB1*06(DQB1*0602)were

associated with a signiﬁcant increase of antibody response to

HepB. For the sake of evaluating thepredicted value of above

describedalleles,theanalysisofsensitivityandspeciﬁcityofthese

alleleswerecarriedout.TheresultsshownthatDRB1*1301and

DQB1*0602hadhighspeciﬁcitytopredictingtheantibodyresponse

toHepB,butallofthesealleleshadlowsensitivitytopredictingthe

antibodyresponsetoHepB.BecausesingleHLAallelewasnot

suit-ableasapredictivevalue,HLAhaplotypesmightbeamoresuitable

tool.However,alldiscussionsregardingtheHLAhaplotypeswere

basedondesignationsaccordingtothemostprobable

combina-Table5

SensitivityandspeciﬁcityofHLAallelesinhealthypopulationforthepredictionofresponsetohepatitisBvaccine.

HLAalleles Responders Nonresponders Sensitivity,%(95%CI) Speciﬁcity,%(95%CI)

DRB1*01 Positive 56 21 25.9(20.1–31.7) 87.7(82.8–92.6) Negative 160 150 DRB1*03 59 63 19.2(14.8–23.6) 64.8(57.8–71.8) 248 116 DRB1*0301 20 34 20.2(12.2–28.2) 60.0(49.6–70.4) 79 51 DRB1*04 73 74 22.3(17.8–26.8) 62.8(56.1.0–69.5) 254 125 DRB1*07 76 141 10.2(8.0–12.4) 72.7(68.9–76.5) 658 376 DRB1*1301 42 8 25.1(18.5–31.7) 94.2(90.3–98.1) 125 130 DRB1*1302 7 14 6.1(1.7–10.5) 78.5(68.5–88.5) 107 51 DRB1*15 141 53 24.2(20.7to27.7) 88.0(85.0–91.0) 442 389 DQB1*02 104 105 31.2(26.2–36.2) 34.0(26.6–41.4) 219 54 DQB1*05 131 35 40.6(35.2–46) 78.0(71.6–84.4) 192 124 DQB1*0501 39 11 33.9(25.2–42.6) 83.1(74–92.2) 76 54 DQB1*06 130 47 40.2(34.9–45.5) 75.6(69.5–81.7) 193 146 DQB1*0602 61 15 26.3(20.6–32.0) 90.1(85.3–94.9) 171 136

(6)

tions[30].Itisdifﬁculttoconductthemeta-analysisofhaplotypes

duetodifferentcombinationamongstudies.Establishmentofthe

exacthaplotypesbyfamilystudiesisneededinthefuture.

Onlythetwo-digitnumbering and thefour-digit numbering

ofalleleswereincludedinourstudybecausetheﬁfthandsixth

digits(silentsubstitutions)werenotexpectedtoaltertheantigenic

expression[31].Andthroughthismeta-analysis,wealsofoundthat

DRB1*1301and1302hadoppositeeffectontheresponsetoHepB,

whichwasagreedwiththepreviousreport[21].Therewasonlyone

aminoaciddifferenceatallele86inthepeptide-bindinggroove.At

thisposition,DRB1*1301hasaVal,whileDRB1*1302hasaGly,and

thesinglepositionchangemaybehasanimportantinﬂuenceon

HBsAgproduction[21].Therefore,comparedwithtwo-digit

num-beringofalleles,four-digitalnumberingofHLAalleleswillbemore

preciousforanalysistherelationshipbetweenHLAandresponse

toHepB.

Accordingtoourqualityassessmentcriteria,boththestudies

of Hohler et al.[20] and Huet al. [14] were regarded as

low-qualitystudies.Thesetwostudieswereexcludedwhenanalyzing

thealleles of DRB1*01, DRB1*04,DRB1*07, DRB1*11, DRB1*14,

DRB1*15andDQB1*04.Thestatisticalresultsdidnotchangeafter

excludingthesestudies,theadjustedpooledoddsratiowasmuch

the same. The causes couldbe that the assessment criteria of

across-sectional studywerenotadmittedbythegreat massof

researchers,ourassessmentcriteriawereestablishedbasedonthe

professionalknowledge,somerelatedreportsandthedemandof

ourpaper.Sotheso-calledlowqualitystudiesdidnotreallyreﬂect

thequalityofstudy.However,thestudyassessmentwas

neces-sary,whichcanmakeclearthestudyforthepeoplefromdifferent

languagecountries.Inaddition,highqualitystudieswere

represen-tativeofgooddesign,clearbackground,andstrictqualitycontrol.

TheshapeoffunnelplotandEgger’stestindicatedthatthe

pub-licationbiasmight existin someHLAalleles. Andtheresult of

sensitivityanalysisforalleleswithsigniﬁcantheterogeneity

con-vincedtheexistenceofpublicationbias.Thesourceofbiasmight

beduetoeditorsmorereadytoforgiveauthorsforﬂawsinstudy

designwhen statisticallysigniﬁcantassociationswerereported.

Poormethodologicalqualityofsmallerstudies,orcontrolgroup

couldnot be trulyrepresentative of the HLAfrequency in the

locality.Socontrolgroupwasnecessarytobesetfortesting

repre-sentativeofHLAfrequenciesthroughcomparingwithfrequencies

publishedbyInternationalHistocompatibilityWorshop(IHWS).

Ourstudyincluded2308subjects(1215responders,873

non-responders and 220 control population). Sample collection of

nonrespondersispractically difﬁcult,sothestatistical poweris

lowduetosmallsamplesizeinasinglestudy.However,the

meta-analysiscanprovideenhancedstatisticalpowerforintegratingthe

publisheddata.Moreover,ourmeta-analysistriedtomake

assess-mentofstudyquality,whichwasseldomfoundinsimilaranalysis.

Thesubjectsincludedinourpaperwerehealthypopulations,and

theconfounderfactorsweremanagedstrictly.

Thelimitationofthispaperwasthesmallnumbersofincluded

studies,sotheethnicsubgroupanalysiswasdifﬁculttoevaluate

insomealleles.Anotherlimitationwasonlysinglealleleanalyzed

inthisstudy.HLAalleledistributionmaybeinﬂuencedbylinkage

disequilibriumofgenesandinteractionsbetweendifferentalleles,

sothemulticenterstudiesandhaplotypesanalysisareneededin

thefuturestudy.

Productionofanti-HBsantibodyfollowingvaccinationor

nat-ural infection is helper T cell dependent. CD4+ _T _lymphocytes

recognizedifferentepitopeswithintheHBsAgmoleculethatcanbe

presentedbycertainHLAclassIIantigens[32].Despiteconﬂicting

ﬁndings from previous studies, this meta-analysis strengthens

thecasethatHLAclassIIallelesareassociatedwithanimmune

responsetoHBvaccine,whichishelpfultoexplorethemechanism

offailingtoresponsetoHepB.Moreover,ourstudyalsoprovided

somesuggestionforfuturestudieswhoconcentrateonefﬁcacyof

hepatitisBvaccinethatCertainHLAclassIItypesshouldbe

consid-eredinthestudyofantibodynonresponse.

Acknowledgement

ThisstudywassupportedbyMajorScienceandTechnology

Spe-cialProjectofChinatwelfthFive-yearPlan(2012ZX10002001).

Conﬂictsofinterest:Theauthorsdeclarethattheyhaveno

con-ﬂictofinterest.

References

[1]Kukka C. HBV. Preventing Motherto-Child Hepatitis B Infection; 2004.

www.hcvadvocate.org

[2]ZuckermanJN.Protectiveefﬁcacy,immunotherapeuticpotential,andsafetyof hepatitisBvaccines.JMedVirol2006;78(2):169–77.

[3]DasK,GuptaRK,KumarV,SinghS,KarP.AssociationofHLAphenotypewith primarynon-responsetorecombinanthepatitisBvaccine:astudyfromnorth India.TropGastroenterol2004;25(3):113–5.

[4]LiY,NiR,SongW,ShaoW,ShresthaS,AhmadS,etal.Clearandindependent associationsofseveralHLA-DRB1alleleswithdifferentialantibodyresponses tohepatitisBvaccinationinyouth.HumGenet2009;126(5):685–96. [5]AmirzargarAA,MohseniN,ShokrgozarMA,ArjangZ,AhmadiN,YouseﬁBM,

etal.HLA-DRB1,DQA1andDQB1allelesandhaplotypesfrequenciesin Ira-nianhealthyadultrespondersandnon-responderstorecombinanthepatitisB vaccine.IranJImmunol2008;5(2):92–9.

[6]HohlerT,Stradmann-BellinghausenB,StarkeR,SangerR,VictorA,RittnerC, etal.C4AdeﬁciencyandnonresponsetohepatitisBvaccination.JHepatol 2002;37(3):387–92.

[7]ZhangJL,LiN,SongYG,BiSL.AssociationofHLA-DRB1alleleswithimmune responsetohepatitisBvaccineinChineseHanpeopleofJilinregion.ChinJ Immunol2011;27(2):140–3.

[8]WeiYH,WuJZ,WuJL,LiGJ,ChenMW,YuYH,etal.Correlationbetweenimmune responsetohepatitisBvaccineandHLA-DRB1*14,15genes.JGuangxiMedUniv 2009;26(2):213–5.

[9]WeiYH,WuJZ,WuJL,LiGJ,ChenMW,YuYH,etal.Aninvestigationforthe effectofHLA-DRBl*07,13allelicgenesontheimmuneresponsetohepatitisB vaccination.ChinJImmunol2009;25(11):996–1002.

[10]LiuS,ZhangH,GuC,YinJ,HeY,XieJ,etal.Associationsbetweenhepatitis Bvirusmutationsandtheriskofhepatocellularcarcinoma:ameta-analysis.J NatlCancerInst2009;101(15):1066–82.

[11]HigginsJP,ThompsonSG,DeeksJJ,AltmanDG.Measuringinconsistencyin meta-analyses.BMJ2003;327(7414):557–60.

[12]LauJ,IoannidisJP,SchmidCH.Quantitativesynthesisinsystematicreviews. AnnInternMed1997;127(9):820–6.

[13]Liao XY, Liang ZL,Li YP, Li RC, Liu JY, Wu XY, et al. The inﬂuence of cellularimmunityandgeneticfactorontheeffectofmothertoinfant trans-mission interrupted with hepatitis Bvaccine. ChinJ Microbiol lmmunol 2006;26(12):1070–5.

[14]HuDB,LiuSK,ZhaoLL,XuAP,HongYZ,CaoPY.Analysistherelationofadults HBVvicinalfailureandTcellsubset,HLA-DRranggenephenotype.ChinJExp ClinVirol2009;23(1):17–9.

[15]YangX,WuJZ,WeiYH,LiGJ,WuJL,ChenMW,etal.Associationofnon-or hypo-responsetohepatitisBvaccinationwithHLA-DRB1*11alleleinthepopulation ofGuangxiprovince.InternMedChina2009;4(5):676–8.

[16]GongXG,WuJZ,WeiYH,LiGJ,ChenMW,WuJL,etal.Comparativestudy ofrelationshipbetweenHLA-DRB1*07geneandimmuneresponseto hepati-tisBvaccineinZhuangandHannationalityfromGuangxi.InternMedChina 2009;4(2):181–4.

[17]YuYQ,WuJZ,WeiYH,LiGJ,WuJL,ChenMW,etal.Associationoftheimmune responsetohepatitisBvaccinationwithHLA-DRB1*15alleleaswellasthe expressionofThl/Th2cytokines.InternMedChina2009;4(5):673–5. [18]LiLL,WuJL,WeiYH,WuJL,GongXG,YuYQ,etal.Inﬂuenceontheimmune

responsetohepatitisBvaccineanddependencywithhla-drbl*15allelefor tgf-␤lcytokine.InternMedChina2010;5(2):113–5.

[19]QianY,ZhangL,LiangXM,HouJL,LuoKX.Associationofimmuneresponseto hepatitisBvaccinewithHLA-DRB1*02,07,09genesinthepopulationofHan nationalityinGuangdongProvince.JFirstMilMedUniv2002;22(1):67–9. [20]HohlerT,MeyerCU,NotghiA,Stradmann-BellinghausenB,SchneiderPM,

StarkeR,etal.TheinﬂuenceofmajorhistocompatibilitycomplexclassIIgenes andT-cellVbetarepertoireonresponsetoimmunizationwithHBsAg.Hum Immunol1998;59(4):212–8.

[21]Lango-WarensjoA,CardellK,LindblomB.Haplotypescomprisingsubtypes oftheDQB1*06alleledirecttheantibodyresponseafterimmunisationwith hepatitisBsurfaceantigen.TissueAntigens1998;52(4):374–80.

[22]McDermottAB,ZuckermanJN,SabinCA,MarshSG,MadrigalJA.Contributionof humanleukocyteantigenstotheantibodyresponsetohepatitisBvaccination. TissueAntigens1997;50(1):8–14.

[23]MilichDR,Leroux-RoelsGG.ImmunogeneticsoftheresponsetoHBsAg vacci-nation.AutoimmunRev2003;2(5):248–57.

(7)

[24]WangC,TangJ,SongW,LobashevskyE,WilsonCM,KaslowRA.HLAand cytokinegenepolymorphismsareindependentlyassociatedwithresponses tohepatitisBvaccination.Hepatology2004;39(4):978–88.

[25]HsuLC,LinSR,HsuHM,ChaoWH,HsiehJT,WangMC,etal.Ethnicdifferencesin immuneresponsestohepatitisBvaccine.AmJEpidemiol1996;143(7):718–24. [26]ChenJB,ShiL,YaoYF,ShiL,YuL,LinKQ,etal.Analysisofthe14bp inser-tion/deletionpolymorphisminhumanleukocyteantigen-GgeneinChinese DaiandHanethnicgroupsinYunnanProvince.YiChuan2010;32(6):577–82. [27]EdinurHA,Zafarina Z,SpinolaH,NurhaslindawatyAR,PanneerchelvamS,

NorazmiMN.HLApolymorphisminsixMalaysubethnicgroupsinMalaysia. HumImmunol2009;70(7):518–26.

[28]SinghR,KaulR,KaulA,KhanK.AcomparativereviewofHLAassociationswith hepatitisBandCviralinfectionsacrossglobalpopulations.WorldJ Gastroen-terol2007;13(12):1770–87.

[29]HohlerT, ReussE, EversN,DietrichE, RittnerC, FreitagCM,etal. Dif-ferentialgenetic determinationof immuneresponsivenessto hepatitisB surfaceantigenandtohepatitisAvirus:avaccinationstudyintwins.Lancet 2002;360(9338):991–5.

[30]LangoA,LindblomB.HLADQA-DQBhaplotypesinaSwedishpopulation.EurJ Immunogenet1993;20(6):453–60.

[31]HoldsworthR,HurleyCK,MarshSG,LauM,NoreenHJ,KempenichJH,etal. TheHLAdictionary2008:asummaryofHLA-A,-B,-C,-DRB1/3/4/5,and-DQB1 allelesandtheirassociationwithserologicallydeﬁnedHLA-A,-B,-C,-DR,and -DQantigens.TissueAntigens2009;73(2):95–170.

[32]DesombereI,GijbelsY, VerwulgenA,Leroux-RoelsG.Characterizationof theTcellrecognitionofhepatitisBsurfaceantigen(HBsAg)bygoodand poorresponders tohepatitis Bvaccines. ClinExp Immunol2000;122(3): 390–9.