Liver International. 2019;00:1–11. wileyonlinelibrary.com/journal/liv  | 1© 2019 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd

Received:11December2018  |  Revised:12June2019  |  Accepted:28July2019DOI: 10.1111/liv.14207

O R I G I N A L A R T I C L E

Predicting HBsAg clearance in genotype A chronic hepatitis B using HBsAg epitope profiling: A biomarker for functional cure

Renae Walsh1 | Rachel Hammond1 | Lilly Yuen1 | Joshua Deerain1 | Tanya O’Donnell1 | Thomas Leary2 | Gavin Cloherty2 | Anuj Gaggar3 | Kathryn Kitrinos3 | Mani Subramanian3 | Darren Wong4  | Stephen Locarnini1

Abbreviations:aa,aminoacid;ALT,alanineaminotransferase;CHB,chronichepatitisB;CP,clearanceprofile;EBV,Epstein‐Barrvirus;EOS,endofstudy;HBV,hepatitisBvirus;HCC,hepatocellularcarcinoma;IndCP,inducedclearanceprofile;mAbs,monoclonalantibodies;NA,nucleos(t)ideanalogue;NCP,non‐clearanceprofile;NR,non‐responder;RFU,relativefluorescentunits;SL,HBsAgloss;TDF,tenofovir.

1DivisionofMolecularResearchandDevelopment,VictorianInfectiousDiseasesReferenceLaboratory,DohertyInstitute,Melbourne,Victoria,Australia2AbbottLaboratories,Chicago,Illinois,USA3GileadSciences,FosterCity,California,USA4DepartmentofGastroenterology,StVincent’sHospital,Fitzroy,Victoria,Australia

CorrespondenceStephenLocarnini,DivisionofMolecularResearchandDevelopment,VictorianInfectiousDiseasesReferenceLaboratory,DohertyInstitute,792ElizabethSt,Melbourne,Victoria,Australia.Email:stephen.locarnini@mh.org.au

Funding informationGileadSciencesfundedandconductedtheoriginalGS‐US‐174‐0103clinicaltrial,includingthecollectionandstorageofserumsamples.Forthestudydescribedhere,allclinicaldataweremadeavailabletotheinvestigators.NeitherGileadSciencesnorAbbottLaboratorieswereinvolvedintheanalysisorinterpretationofdata,thedecisiontosubmitforpublication,anddidnotfundthestudy.

HandlingEditor:AlessioAghemo

AbstractBackground and Aim: FunctionalcureisthemajorgoalofchronichepatitisB(CHB)therapythoughfewbiomarkerspredictthisoutcome.HBsAgepitopeoccupancycanbe influencedby therapeutic and immunepressure. The aimof this studywas tomaptheHBsAgepitopeprofilesduringlong‐termnucleos(t)ideanaloguetherapyinpatientswithgenotypeACHB,inthecontextofHBsAgloss(SL)/seroconversion.Methods: Weevaluated25genotypeACHBpatients intheGS‐US‐174‐0103trialofHBeAg‐positiveCHBpatientstreatedwithtenofoviroradefovirfor4years,14whoachievedSLwhilst11hadnochange.WeepitopemappedthemajordomainsofHBsAgtoidentifythosepatientswithHBsAgclearanceprofile(CP)(lossofbindingatbothloops1and2epitopesofthe ‘a’determinant)vsnon‐clearanceprofile (nochangeinepitoperecognition,orlossofepitopebindingatonelooponly),correlat‐ingthistoon‐treatmentHBsAgresponses.Complexedanti‐HBswasalsomeasured.Results: AnalysisoftheHBsAgepitopeprofilesofthe25patientsatbaselineidenti‐fiednopredictivecorrelationwithSL. Incontrast,analysisatweek48andendofstudy(week192)orpriortoSLidentifiedsignificantpredictiveassociationsbetweendevelopmentofHBsAgCPsandoutcomeoffunctionalcure.ThedetectionofaCPalsocorrelatedwiththedevelopmentofanalanineaminotransferaseflareanddetec‐tionofanti‐HBscomplexedwithHBsAg.Conclusion: The detection ofHBsAgCPs by epitopemapping represents a novelviral biomarker, reflecting an emerging anti‐HBs selection pressure prior to func‐tionalcure.

K E Y W O R D S

ALTflare,antiviraltherapy,functionalcure,genotypeAinfectedpatients,HBsAgepitopemapping,HBsAgloss

2  |     WALSH et AL.

1  | INTRODUC TION

Approximately2billionpeoplehavebeeninfectedwiththehep‐atitis B virus (HBV), of whom 260‐350 million have developedchronicdisease, resulting inmore than780000deathsannuallyattributedtotheclinicaloutcomesofcirrhosisandhepatocellularcarcinoma(HCC).1CurrentantiviraltherapiesforchronichepatitisB (CHB)arenot curative, and theavailablepreventativevaccinehasnoimpactonexistinginfections.Nucleos(t)ideanalogue(NA)therapies effectively achieve viral DNA suppression (HBV DNAundetectable),butnottheclearanceof theHBVsurfaceantigen(HBsAg), which is associated with ongoing risk for developingHCC.2,3FunctionalcureisthecurrentgoalforCHBtherapy,4butisrarelyachieved,occurringinonly1%‐2%ofuntreatedpatientsannually and onlymarginal improvementwhen treatedwithNAtherapy.4 The endpoint of HBsAg loss (SL) ± seroconversion toanti‐HBs,withundetectableserumHBVDNA,isconsideredtobeafunctionalcureforCHB,4asitallowsnotonlyforthecessationoftherapybutisalsoassociatedwithsignificantreductionintheratesofcirrhosisandthedevelopmentofHCC,andanoverallin‐creasedsurvivalrate.2‐4

Themechanism(s)associatedwithSLand/orseroconversionareunknown,butthebasisforthisclearanceispresumablytheselec‐tivepressureofaneffectivehostantiviral response.AneffectiveB‐cellresponsehasbeenshowntobeessentialtomaintainafunc‐tionalhepatitisBcure,withB‐celldepletingtherapies(egrituximab,anti‐CD20)beingassociatedwithHBVreactivationleadingtoliverfailure and death.5 Clinical and diagnostic observations suggestthatB‐cellclonesencodinganti‐HBswithlowaffinitytothehomol‐ogousHBsAgexist inmostpatientswithCHB.6Concurrentlyex‐pressed,orcocirculatinganti‐HBsinHBsAg‐positivepatientswithCHBfurthersuggeststhattheanti‐HBsresponseisquantitativelyand/orqualitativelyinsufficientorinadequatetoovercomechronicinfection,butdoes indicatethatHBsAgspecificantibodyproduc‐tionisnotcompletelyblocked.7DuringspontaneousresolutionofacutehepatitisBinfection,acoordinatedinteractionofHBV‐spe‐cific T cells and B cells eliminate infected hepatocytes, suppressviralreplicationvianon‐cytolyticpathways,andneutralisevirionsthrough the production of virus‐specific antibodies (anti‐HBs).Whetherthesesameorsimilareventsoccurduringfunctionalcureisnotknown.

TheHBVtranslatesthreemembrane‐associatedsurfaceoren‐velopeproteins(L,MandS)thatallshareacommonC‐terminalorSregion,whichencodethemajoranti‐HBsbindingdomain,alsocalledthe ‘a’ determinant, spanning residues 99 to 160 ofHBsAg.8 TheHBsAg ‘a’ determinant is a conformationally dynamic protein do‐main,incorporatingahighdensityofconservedcysteineandprolineresidues resulting in the formationofmultiple loopdomains.8TheHBsAg isvulnerabletobothstructuralalterationsandalsotoepi‐topeavailability fromapplied immunepressure,whichcanmodifytheHBsAgepitopeavailabilityprofile.9HBsAgepitopeavailabilityoroccupancyareinfluencedbyimmunepressuresuchasa‘clearing’anti‐HBs response targeting the HBsAg antigenic ‘a’ determinant

region or major loop domains therein (loop1: amino acids [aa]107‐138;loop2:aa139‐149).Atpresent,nobiomarkerscanreliablypredictfunctionalcure.WehavepreviouslycharacterisedauniquecohortofNAtreatedgenotypeApatientswithCHBwhohaveun‐dergone functional hepatitis B cure10 and also demonstrated thatalanineaminotransferase(ALT)flarescanpredictSL.11Theavailabil‐ityof this cohort provided theopportunity to studyHBV‐specifichumoral immune responses directed against epitopes within theHBsAgantigenic ‘a’determinant,andtodetermine if lossof theseepitopeswaspredictiveofafunctionalcure.

2  | METHODS

2.1 | Patients: cohort selection and characteristics

Patientswere selected from theGS‐US‐174‐0103 (G103) study,10 a randomiseddouble‐blind,phase3 clinical trial ofNA treatment‐naïve, HBeAg positive, immune clearance phase patients, treatedwith tenofovir (TDF) or adefovir dipivoxil for 48weeks, followedby TDF monotherapy for an additional 144 weeks (Figure S1).CompletedatatoW192wereavailable for146patients, including42(29%)genotypeApatients.Intotal,141(96%)attainedantiviralsuppressionand22(15%)achievedSL.InthegenotypeAsubgroup,33%(14/42)attainedHBsAgclearance.Westudiedasubsetof25genotypeA(HBeAg+)CHBpatientsfromtheG103study,14whoachievedHBsAgclearance±seroconversion,whilst11realisednochange inHBsAg (<0.5 logdecline)byW192(endofstudy [EOS]).Demographicalandclinicalcomparisonofthesetwogroupsofpa‐tientsrevealedtheywereessentiallyidenticalexceptforSLornot(resultsnotshown10).

2.2 | Diagnostic serology

Studysampleswereserologicallyanalysed forquantitativeHBsAg(IU/mL;LLOD0.05IU/mL),HBeAg(PEIU/mL),ALT(U/L),and‘free’anti‐HBsantibody(IU/L)usingElecsyskitdiagnosticassaysontheRocheCobasplatform.

Key pointsNew treatments are being developed for patients withchronichepatitisB(CHB)withthegoalofachievingfunc‐tionalcure.Theavailabilityofadiagnosticbiomarkerthatwould enable the reliable prediction of functional curewouldhelpexpandthesearchforbetterCHBtreatments.Thisstudyhasidentifiedspecificchangesintheshapeandconformation of the hepatitis B virus envelope proteinwhich can be used as amarker for predicting functionalcureingenotypeAinfectedpatients,therebyguidingfu‐ture studies that could lead to improved therapeutic ap‐proachesforthesepatients.

     |  3WALSH et AL.

2.3 | HBsAg epitope mapping

A 19plexHBsAg epitopemapping assay on a Bioplex®200 plat‐form (Figure1A), targetinganti‐HBsepitopesacross fiveHBsAgdomains has been described previously12 (Figure 1B and C).Briefly, epitope display (antigen conformation) and occupancy(‘native’ anti‐HBs recovery) influence theHBsAg profile.HBsAgprofile isreportedasfoldchange(95%CI:±0.5‐fold) incompari‐sontothegenotypebackboneandnormalisedtothepretreatmentprofile(Figure2and3).12

2.4 | HBsAg CP analysis

TheHBsAgepitopeprofilewasthenmappedusingthis19plexim‐munoassay targeting the key epitopes across theHBsAg coveringthe‘a’determinant(Figure1).HBsAgprofileswereclassifiedeitheras aHBsAg clearance profile (CP: reduced epitope recognition atbothloops1and2epitopesofthe ‘a’determinant8) (Figure2B)oranHBsAgnon‐clearanceprofile(NCP:noprofilechange,orreducedepitoperecognitionatonelooponly)(Figure2A).ResultswerethenrelatedtotheserumHBsAglevel.

F I G U R E 1  DiagrammaticrepresentationoftheHBVsurfaceantigen(HBsAg)epitopeprofileassaytargetingtheHBsAgantigenicregion,andanassayoutcomeofanHBsAgCPvsNCP.A,19plexHBsAgepitopeprofileassaydesignusingtheBioplex200multiplexplatform.Individualfluorescentlyidentifiedmagneticbeadsetswhichhavebeenpreconjugatedwithoneof19differentanti‐HBsmAbsareplexedtogetherandincubatedwithpatientHBsAg‐positiveserasamplespriortodetectionviaasecondaryreportedantibody.InteractionsarereportedinRFU.RawRFUdataareanalysedthroughanin‐housebioinformaticspipelineandexpressedaspositiveornegativefoldchangeinepitoperecognitionwithrespecttoHBVgenotypebackboneandpatientbaselinepretreatmentsample.B,IllustrationoftheHBsAgSantigenic‘a’determinantandC‐terminaldomains,spanningresidues99‐226.Potentialloop1andloop2domainsareindicated,asarethelocationsofconformationallyinfluentialcysteine(orange)andproline(Blue)residues.Theepitopespecificityofthe19plexHBsAgepitopemappingmAbstoparticularHBsAgdomains(N‐terminal,Loop1,Loop2,C‐terminal,Loop1/2combinational,conformational)isindicated.C,SchematicexamplesoftheHBsAgepitopemappingassayoutcomeclassificationofHBsAgprofilesasanHBsAgclearanceprofile(CP:reducedepitoperecognitionatbothloops1and2epitopes)oranHBsAgnon‐clearanceprofile(NCP:noprofilechange,orreducedepitoperecognitionatonelooponly)

4  |     WALSH et AL.

F I G U R E 2  ExampleofHBsAgepitopemappingpatientdata,illustratingapretreatment/baselineHBsAgNCPandconversiontoanHBsAgCPpriortoHBsAgloss(SL).A,Analysisofstudypatient'spretreatmentorbaseline(BL)HBsAgepitopeprofileswasperformedandreportedasa±foldchangewithrespecttoHBVgenotypebackbone.ThispatientdisplaysapretreatmentHBsAgNCP,withNOreductionofepitoperecognition(>0.5‐fold)detectedateitherloop1orloop2HBsAgepitopes.B,DuringthestudythepatientdevelopedanHBsAgCPwithreducedepitoperecognition(>0.5‐fold)atbothloop1andloop2HBsAgepitopes,priortoachievinganoutcomeofHBsAgloss.CP,clearanceprofile;HBsAg,HBVsurfaceantigen;NCP,non‐clearanceprofile

     |  5WALSH et AL.

F I G U R E 3  RepresentativeHBsAgepitopeprofiles(CPandNCP)acrossstudytimepointsforHBsAgSLvsandHBsAgNRpatient.A,AnHBsAgSLpatientwithapretreatmentHBsAgNCPdevelopedanHBsAgCP(reducedrecognitionatBOTHloop1andloop2HBsAgepitopes)duringthestudyfromweek10,coincidingwithHBsAgdeclinefromweek8.AnHBsAgCPwasmaintainedfromweek10toweek120(finalHBsAgpositivetimepoint)priortoSLatweek144.B,AnHBsAgNRpatientwithapretreatmentHBsAgNCP(nochangeinHBsAgepitoperecognition,orreducedbindingatonlyoneloopregion)thatwasmaintainedthroughoutthestudywithn=6timepointsassessedfromweek12toweek192/EOSandnoHBsAgdeclinedetected.CP,clearanceprofile;HBsAg,HBVsurfaceantigen;NCP,non‐clearanceprofile;NR,non‐responder;SL,HBsAgloss

6  |     WALSH et AL.

2.5 | Complexed anti‐HBs (with HBsAg): concurrent anti‐HBs and HBsAg

An in‐houseenzyme immunoassayfordetectionofanti‐HBscom‐plexedwithHBsAg(concurrentantibodyandantigen),whichisnotabletobereliablydetectedusingcurrentdiagnosticassays,wasalsodeveloped(seeFigureS2).InthisassayHBsAgiscapturedby2plexmagneticbeadsetconjugatedwithbroadlyspecificanti‐HBsmono‐clonalantibodies(mAbs),andconcomitantpatient‐derivedanti‐HBsidentifiedindirectlyviahumanIgGFcdomaindetection.Thisassayuses a fluorescent readout based on detection of the complexedanti‐HBsFcdomain.13

2.6 | Induced anti‐HBs CP analysis

Theeffectofanti‐HBsantibody(includingpatientanti‐HBspost‐se‐roclearance) on theHBsAg epitope profilewas next examined bypre‐incubationof theanti‐HBsantibodypositive sample,with thewild‐typereferenceHBsAgreagentpriortotestinginthe19plexim‐munoassayforthesubsequentHBsAgepitopeprofileasdescribedabove(FigureS3).Theresultinganti‐HBs ‘induced’HBsAgprofileswere classified either as an Induced Clearance Profile (IndCP: re‐duced epitope recognition at both loops 1 and 2 epitopes) or anInducednon‐clearanceprofile(noprofilechange,orreducedepitoperecognitionatonelooponly)andtheresultswererelatedtotheclin‐icaloutcomeofthesefunctionalcurepatients.

2.7 | Peptide chip scanning

Pan‐genotypicHBsAg(L,MandSdomains;389residuesintotal)cy‐clisedandlinearpeptidechipsetsweredesignedandorderedfromPEPperCHIP.Linearpeptidechipswereprintedwithatotalof1953peptides induplicate(3906peptides intotal),coveringHBVgeno‐typesA‐HofHBsAgas13merpeptides,overlappingby12residues.Cyclicpeptidechipswereprintedwithatotalof1614cyclisedpep‐tidesinduplicate(3228peptidesintotal),coveringHBVgenotypesA‐DofHBsAgasboth15merand10merpeptides,withashiftoftworesidues.RedundantpeptidesacrossHBVgenotypeswereex‐cluded,andthetargetpeptidearraysonbothchipsetswereframedby internal controlFlag (DYKDDDDKGG)andHA (YPYDVPDYAG)peptides.Peptidechipswereblockedwith1%BSAinPBS‐T(PBS,pH7.4with0.05%Tween20),priortoovernightincubationwithpatientserumdilutedat1in1000,andstainingwithsecondarygoatanti‐human IgGFcCy3antibody (ThermoFisherScientific).ChipswerescannedusingaGenepix4000Bscannerwithfluorescenceintensi‐tiesquantifiedbyPepSlideAnalyzersoftwareandreportingasrela‐tivefluorescenceunits(RFU)signals.Positivebindingfluorescencewas determined by normalisation of the sample red foregroundmean RFU to twice the red background mean RFU, with values>0RFUindicatingpositivebindingreactivityandvalues>1000RFUindicativeofstrongbindingbetweensampleantibodyandthechipimmobilisedpeptidespot.

2.8 | Statistical methods

Statisticalanalysesofcategoricaloutcomevariableswereconductedusing theFisherexact test, asappropriate forapplication tosmallsamplesizes.AP‐valueof.05wasconsideredstatisticallysignificant.

3  | RESULTS

3.1 | HBsAg CP/epitope profile

From the G103 cohort, 181 longitudinal samples (pretreatment,and timepoints to SL orW192) from the 25 genotype A patientswereanalysedbymappingeachsampletimepoint'sHBsAgepitopeprofile. Inthissubset,14patientsachievedfunctionalcureand11showednochangeinqHBsAgbyW192.UsingthemultiplexHBsAgepitopeprofileimmunoassaypanelof19anti‐HBsmAbs,divergentantigenicprofilesbetweenthe14SLseroconvertersand11non‐re‐sponders(NR)wereidentified.Twopatternsweredefined:eitheranHBsAgCPwhichincludedalossofanti‐HBsepitopeavailabilityatbothloop1andloop2ofthe‘a’determinant,orHBsAgNCPwhichincludednochangeinanti‐HBsepitoperecognition,oralossofrec‐ognitionatonlyasingleHBsAgepitopeloopinthe‘a’determinant.TheHBsAgepitopesweredeterminedacrossmultiplesampletime‐pointstoSLortoW192/EOS,anddevelopmentofHBsAgCPwasestablished(Figure3).

AnHBsAg CPwas identified byW48 of treatment in 10/14SLpatients(Table1).Therewasstrongandstatisticallysignificantassociation (P‐value .015; PPV 83%) between the developmentofthisHBsAgCPandtheoutcomeofHBsAgclearance/serocon‐version(Table1).ThestatisticallysignificantassociationbetweenHBsAgCPandSLwasenhanced(P‐value.001)bytheEOSorthefinalavailablesampletimepointbeforeSLandreachedaPPVof92% and an NPV of 77% (Table 1). The baseline HBsAg profilewasnotpredictiveofaclinicaloutcomeofSL(P‐value.697).TheHBsAgprofilesidentifiedwithinthematchedNRpatientssharedsimilaritywith thoseprofilesobtainedwith referenceclonedes‐capeprofilesidentifiedpreviouslyinstudiesoftheHBsAgvariants

TA B L E 1  DevelopmentofanHBsAgCPinstudypatientsbyweek48andEOS,orpriortoSL

Development of HBsAg CP At baseline

By W48 or prior to SL

By EOS or prior to SL

SL,n=14 5 10(2/10maintainedfromBL)

11

NR,n=11 5 2(maintainedfromBL)

1

P‐value .697 .015 .001

PPV n/a 83% 92%

NPV n/a 69% 77%

Abbreviations:CP,clearanceprofile;EOS,endofstudy;HBsAg,HBVsurfaceantigen;NR,non‐responder;SL,HBsAgloss.

     |  7WALSH et AL.

sG145R,avaccineescapevariant,andsD144EanHBIgtreatmentescapevariant (datanot shown). Thedevelopmentof anHBsAgCPwith an outcome of SL could also be correlated to the levelofHBsAgdecline (Table2),with theassociationbeing strength‐ened with each log IU/mL decline in HBsAg response (P‐value.003‐.0005),aswellastoaserumALTelevation(P‐value.017)pre‐cedingHBsAgdeclineorSL(Table2).11Noneoftheserafromthespecificityserapanel reacted inanyof theCP‐associatedassays(datanotshown).

3.2 | Complexed anti‐HBs development: association with HBsAg CP and HBsAg response

To further understand the development of anti‐HBs responsesleadinguptoHBsAgclearanceandseroconversion,thecomplexedanti‐HBs assaywhichmeasures anti‐HBsbound toHBsAg (FigureS2)wasutilised.Thisassayidentifiestheserologicalprofileofcon‐currentorco‐existinganti‐HBsantibodywithHBsAgduringCHB.Thedevelopmentofcomplexedanti‐HBsmayrepresentrecoveryoftheimmuneresponseand/orreductionof‘free’HBsAgtoincreasetheratioofcomplexedanti‐HBstotheHBsAg.AnalysisoftheG103genotypeASL (n=14)vsNR (n=11)patientsubsetsdeterminedthat complexed anti‐HBs was not necessarily associated with SL.Complexedanti‐HBswithHBsAgwasdetectedpriortoSLorEOSin11/14(79%)patientswhoachievedSL,butalsoin7/11(64%)HBsAgNRsubjects.Thedevelopmentofcomplexedanti‐HBsdid,however,coincidewithanHBsAgCP(Table3),andwassignificantlyassoci‐ated(P‐value.004)withanoutcomeofHBsAgclearance/serocon‐version,withaPPVof91%andanNPVof71%(Table3).

Inaddition,thisassaycouldidentifythosestudypatientsexperi‐encingincompletefunctionalcure,whereHBsAgpersistedbeyondthe lowerdetection limitofHBsAgdiagnosticassays (<0.05 IU/L),or in a complex with antibody. Complexed anti‐HBs with HBsAgwas also detected in study patient serum post‐seroclearance ofHBsAg, indicating incompleteHBsAgclearance.Thissubsetofpa‐tientscouldbeinastateoffluxortransitiontofunctionalcureandcouldbemoreprone to reactivationof theirCHBstatus.AnalysisoftheG103cohortrevealedthatHBsAgincomplexwithanti‐HBspersistedin57%(8/14)patientspost‐SL,reducingto36%(5/14)byEOS.Thus,30%to50%ofSLpatientsdisplayedaserologicalprofileindicativeof‘diagnostic‐assay’clearance(usingstandarddiagnosticserologyassays)butnottrueorcompleteclearanceofHBsAgfromtheperipheralcompartment.

3.3 | CP‐associated anti‐HBs development: Induced HBsAg CP analysis

Athirdassaywasnextutilisedtomeasuretheabilityoftheserumsam‐plescollectedfromthesefunctionallycured,seroconvertedpatients,toinduceaCPinthewild‐typereferenceHBsAgreagentusedintheBioplexPlatform.Inthe14genotypeASLgroupfromtheG103cohort,12/14(86%)seroconvertedtoanti‐HBsasdeterminedbytheconven‐tionaldiagnosticassay. InducedCPanalysisofmultiplesampletime‐pointsbetweenSLandW192fromthese12anti‐HBsantibodypositivesubjects detected a CP‐associated ‘clearing’ anti‐HBs antibody re‐sponseorIndCPfor10/12(83%).Theanti‐HBsresponsepresentinthemajorityofG103studypatientswhoachievedSLwasassociatedwithaHBsAgCPbeforeSL.Analysisofthe2/14anti‐HBsnegativepatientsforanInducedCPwasalsoperformedonsampletimepointsfollowingHBsAgclearanceandnotsurprisingly,thesesamplesdidnotreturnapositiveresultforanIndCPprofileof‘clearing’anti‐HBsantibody.

3.4 | Anti‐HBs antibody peptide chip screening

Thepeptidechipscreenwasdesignedtocomprehensivelyscancontin‐uousepitopes,bothlinearandcyclised,spanningthecompleteHBsAgL,MandSsequenceacrossHBVgenotypesA‐H(linearpeptidechip)orA‐D(cyclicpeptidechip).All14genotypeASLsubjectsfromtheG103cohortwerescreenedonthepeptidechipsusingaselectedanti‐HBspositive sample timepoint post‐SL for the 12/14 subjectswho sero‐converted,orasamplebetweenHBsAgclearanceandEOSfor2/14non‐seroconvertedpatients.Onlyonesubjectsample(1/14)returnedapositiveepitope‘hit’bypeptidechipscreeninginthecyclicpeptidechip,whichcomprised15mercyclicpeptidesfromgenotypeA‐D.Thisreac‐tivitywasdirectedtothecontinuouscyclicepitopewithintheHBsAgLdomain,residues20‐32(NPLGFFPDHQLDP),associatedwiththees‐sentialdomainforNTCPreceptorbinding.14ThepeptidescreeningalsofailedtoidentifyanyconsistentcontinuousepitopeswithintheHBsAg,including theHBsAg S antigenic domain (HBsAg S residues 99‐169),whichwerethetargetoftheanti‐HBsresponsedevelopedpost‐HBsAgclearanceinthiscohortofCHBfunctionalcurepatients.However,thepeptidechipscannotscreenfordiscontinuousepitopes(non‐linear,butspatiallyadjacentresidues),whicharemoreapplicabletotheHBsAgSdomaingiven itshighlyconformationalstructurebecauseofmultiplecysteine andproline residues.Without an accurate3D structure forHBsAg(virus‐likeparticlesandvirions),designandscanningofdiscon‐tinuousorconformationalHBsAgepitopescannotbeachieved.

HBsAg CP coinciding with HBsAg decline or ALT flare

0.5 log10 HBsAg decline

1 log10 HBsAg decline

2 log10 HBsAg decline

Coinciding with ALT flare

SL,n=14 6 8 10 12

NR,n=11 2 0 0 4

P‐value .234 .003 .0005 .017

Abbreviations:ALT,alanineaminotransferase;CP,clearanceprofile;HBsAg,HBVsurfaceantigen;NR,non‐responder;SL,HBsAgloss.

TA B L E 2  DevelopmentofanHBsAgCPthatcoincidedwithHBsAgdeclineorALTflare

8  |     WALSH et AL.

3.5 | Relationship between changes in serum ALT, CP, complexed anti‐HBs, HBsAg levels and functional cure

The relationship between the development of a serum ALT flareand HBsAg seroclearance in this G103 cohort has been recently

published.11Table2summarisesthedevelopmentofanHBsAgCPwithHBsAgdeclineandALTflare.AmodelfortheclinicaltimelinesofthevirologicalanddiagnosticmarkersofHBVDNA,HBsAgandALT together with the development of the HBsAg CP biomarker,alongwiththedetectionofcomplexedanti‐HBswithHBsAgandtheassociationofseroconvertedanti‐HBsresponsewithanInducedCP,isoutlinedinFigure4forthefunctionalcureSLpatientsandFigure5forNRCHBpatients.

4  | DISCUSSION

Inthisstudy,threenovelassayshavesuccessfullybeenappliedtothestudyofacohortofpatientswithCHB10 inordertomonitorHBsAgepitopeprofilechangesforthedevelopmentofanHBsAgCP,thepresenceofcomplexedanti‐HBsandfinally,theemergenceofanantibodyresponsenotpreviouslyrecognised,butcapableofinducing a CP leading to functional cure. Collectively, these as‐says identified quite divergent antigenic profiles and serological

TA B L E 3  Detectionofcomplexedanti‐HBsantibody(withHBsAg)thatcoincidedwithdevelopmentofanHBsAgCP

Association of complexed anti‐HBs with HBsAg CP

Week 192 (EOS) or prior to SL

SL,n=14 10

NR,n=11 1

P‐value .004

PPV 91%

NPV 71%

Abbreviations:CP,clearanceprofile;EOS,endofstudy;HBsAg,HBVsurfaceantigen;NR,non‐responder;SL,HBsAgloss.

F I G U R E 4  HBsAgSLpatient:amodelforHBsAgclearancebasedonclinicalserologyanddevelopedbiomarkerprofiles.Adeclineinserologicalviralmarkers(HBVDNA,HBsAgandHBeAg)coincidedwithanALTflare(week44‐62)indicativeofimmuneactivityintheliver.ThedevelopmentofanHBsAgCPpredicativeofSLfromweek12precededtheviralserologyresponses.Thedetectionofcomplexedanti‐HBs(withHBsAg)indicativeofadevelopingantibodyresponsecoincidedwithimmunere/activation(ALTflare)andalsoHBsAgCPdevelopment.ThepatientachievedaclinicalfunctionalcureoutcomeofHBsAgclearanceandseroconversionatweek96‐108.ALT,alanineaminotransferase;CP,clearanceprofile;HBsAg,HBVsurfaceantigen;SL,HBsAgloss

     |  9WALSH et AL.

responsesbetweenpatientswithSLcomparedtoNRpatients.Thecohort consisted of an important and valuable subset of 14 pa‐tientswhoachievedSL/seroconversion (functional cure), and11closelymatchedpatientsinwhomnochangeinHBsAg(<0.5log10 IU/mLdecline)wasobserved10(FigureS1).EpitopemappingoftheHBsAgprofileresolvedthis25patientcohortintothosewhoei‐therdemonstratedaHBsAgCPoraHBsAgNCP(Figures1and2).By linking these findings to theHBsAg log decline/response onNAtherapy,itwaspossibletodemonstratethatanalteredepitopeavailability in theHBsAgantigenicdomain is likelybecauseof adevelopinganti‐HBsselectionpressure,capableofinducingaCP.StudiesofthesegenotypeACHBpatientsidentifiedastrongandstatistically significant association between the development ofthisHBsAgCPandtheoutcomeofHBsAgclearanceandserocon‐version (P‐value .015andPPVof83%)basedonHBsAgepitopemapping (Table 1). The development of an HBsAg CP based onlossofspecificepitopetotheloop1andloop2regions(becauseofantibodyoccupancyorocclusion)werealsostronglyassociated

(P‐values.003‐.0005)withasignificantdecline(>1.0logIU/mL)orloss inHBsAg (P‐value .001,PPV92%) (Tables1and2). Indeed,in 9/14 patients who achieved SL, an HBsAg CP preceded anyHBsAg response,while for 3/14SLpatients anHBsAgCP coin‐cidedwith HBsAg response (>1 log decline), whilst 2/14 SL pa‐tients anHBsAgCPwas not detected. Furthermore, complexedanti‐HBswithHBsAgwhencorrelatedwithanHBsAgCP(Table3),wasstronglyassociatedwithHBsAgclearance/seroconversion(P‐value .004;PPV91%;NPV71%).Thecorrelationof thedevelop‐mentofanHBsAgCPpredictivebiomarkerwithALTflare(P‐value.017)(Table2)furthersupportstheconceptthattheHBsAgCPisprobablydrivenbythedevelopmentofatargeted‘clearing’anti‐HBs response to clearance‐sensitive HBsAg epitopes presentedon the surface of hepatocytes.We propose that theHBsAgCPreflectsepitopeoccupancyoravailability,particularlywithin themajor loop domains (loop1: aa 107‐138; loop2: aa 139‐149), be‐causeofarecoveringhumoralimmuneresponse.Finally,inthose12/14 (86%)HBsAg clearance patientswho also seroconverted,

F I G U R E 5  HBsAgNRpatient:amodelforCHBpersistencebasedonclinicalserologyanddevelopedbiomarkerprofiles.AdeclineinserologicalviralmarkersofHBVDNAandHBeAgfollowedNAtreatmentandabriefALTflarefromBLtoweek8.TheimmuneeventwasinsufficientorinappropriateanddidnotresultinanHBsAgdecline.AnHBsAgNCPprofilewasmaintainedthroughoutthestudy,predictiveofCHBpersistence.Thedetectionofcomplexedanti‐HBs(withHBsAg)wastransient,indicativeofaninsufficientorweakdevelopingantibodyresponse.ToEOSthepatientmaintainedaconstantHBsAgloadandpersistentCHBinfection.ALT,alanineaminotransferase;BL,baseline;CHB,chronichepatitisB;EOS,endofstudy;HBsAg,HBVsurfaceantigen;HBV,hepatitisBvirus;NA,nucleos(t)ideanalogue;NCP,non‐clearanceprofile;NR,non‐responder

10  |     WALSH et AL.

the resulting anti‐HBswas shown to be capable of inducing anHBsAgCP, and thus can be considered a ‘clearing’ antibody re‐sponseorclearance‐associatedanti‐HBsantibody.

Surprisingly, therewasadistinctabsenceofHBsAgSdomainepitope reactivity of the 12/14 SLwith seroconversion subjectswhen analysed against a comprehensive pan‐genotypic peptidechip screening panel of linear and cyclised continuous epitopesacrossHBsAg L,M and S protein. Not a single positive reactiveregion within the HBsAg S domain was found in any of the 12functionalcuresubjects.ThesefindingsstronglyindicatethattheHBsAg epitopes which need to be targeted to achieve develop‐mentofaclearinganti‐HBsresponse(andsatisfyanHBsAgCP)arediscontinuous, spanning fragmented linear sequences displayedinastructurallyandspatiallydependantmanneronoracrosstheHBsAgsubunit/s(Figure1B).TheHBsAgisrecognisedasahighlydynamic conformational protein subunit forming multisubunitparticles encompassing anunusually largenumberofhighly con‐served conformationally influential cysteine and proline residuesthroughout the ‘a’ determinant antigenic domain.8 These factorsare thought to influence formation of numerous loop structuresandcontribute toastructural formatwhich isnot readilycondu‐civetoresolution.12

ThisstudyrevealedthatthedetectionofaHBsAgCPprecededorcoincidedwithHBsAgdeclineandseroconversion.Amodellink‐ingpatientsserologyresponsesandbiomarkerprofiles leadinguptoHBsAgclearanceandantibodyseroconversion toachieveCHBfunctionalcure (Figure4)vsHBsAgNRCHBgenotypeA infectedpatients(Figure5)canbeproposed.InpatientsundergoingSL/se‐roconversion a decline in serologicalmarkers (HBVDNA,HBeAgandHBsAg)toundetectablelevelscancoincidewithanALTflare11 (Figure4); theHBsAgCPprecedesorcoincideswithanALT flare(93%) andwith complexed anti‐HBs (71%) (Figure 4 and Table 3)implyingsomeformofimmunereactivityandrecovery.InpatientswhoareHBsAgNR,thereisnoHBsAgdeclineandaHBsAgCPisnotdetected(seeFigure5).SerumALTflaresandcomplexedanti‐HBsare inconsistently detected (Figure 5 and Table 3). These profilescouldhavepotentialapplicationasbiomarkersforpredictingfunc‐tionalcureoutcomes.Futurestudiestoelucidatethecharacteristicsofthisanti‐HBsresponsedevelopedbyCHBpatientsinfectedwithotherHBVgenotypescouldofferabroadbasedclinicallyrelevantdiagnostic forguiding treatmentstrategies inachieving functionalcure.

This studywas limited to the analysis ofHBV genotypeA in‐fectedCHBpatients,arestrictionappliedsimplybyaccesstoavail‐ableserafromasufficientlysizedcohortofNAtreatedCHBpatientswhoachievedHBsAgseroclearance±seroconversiontoantibody.However,futurestudieswillbeaimedatexpandingthispredictivebiomarkeranalysisandvalidation,andtestthemodelofCP‐associ‐atedantibodymediatedHBsAgclearanceculminatinginfunctionalcuretoadditionalnon‐HBVgenotypeAcohortsofCHBpatients.Inaddition, futurestudieswill involvestudiestofurthercharacterisetheHBsAgCPepitopestargetedbyaclearingantibodyresponseandhopefullyprovideanunderstandingofthemechanismofantibody

mediatedCHBclearancethatdrivesfunctionalcure.Inconclusion,thedevelopmentofaHBsAgCPprecededorcoincidedwithdeclinein HBsAg and was predictive of HBsAg clearance and functionalcure.Theidentificationofcomplexedanti‐HBsandinducedCPanti‐HBswas indicativeofanemergingandclearing immune responsedevelopedbypatientswhoachievedfunctionalcure,providingnewinsightsintotheeventsleadingtoSL.

ACKNOWLEDG EMENTS

We thank Prof Howard Thomas (UCL), Abbott Diagnostics, bi‐oMerieuxandXTLPharmaceuticalsfortheirkinddonationoftheanti‐HBs antibodies utilised in these reported studies. We alsothankGileadSciencesfortheGS‐US‐174‐0103(G103)cohortpa‐tientserasamples.

CONFLIC T OF INTERE S T

Thomas Leary and Gavin Cloherty are employees of AbbottLaboratoriesandownstockandsharesinAbbottLaboratories.AnujGaggar,KathrynKitrinosandManiSubramanianareemployeesofGileadSciences,andownstockandsharesinGileadSciences.Theremainingauthorshavenopersonalor financialdisclosuresof rel‐evancetothisstudy.

ORCID

Darren Wong https://orcid.org/0000‐0003‐1490‐0547

Stephen Locarnini https://orcid.org/0000‐0001‐7253‐8932

R E FE R E N C E S

1. LozanoR,NaghaviM,ForemanK,etal.Globalandregionalmortal‐ityfrom235causesofdeathfor20agegroupsin1990and2010:asystematicanalysisfortheGlobalBurdenofDiseaseStudy2010.Lancet.2012;380(9859):2095‐2128.

2. FattovichG,GiustinaG,Sanchez‐TapiasJ,etal.Delayedclearanceof serum HBsAg in compensated cirrhosis B: relation to inter‐feronalphatherapyanddiseaseprognosis.EuropeanConcertedAction on Viral Hepatitis (EUROHEP). Am J Gastroenterol. 1998;93(6):896‐900.

3. YuenM,WongDK,FungJ,etal.HBsAgSeroclearanceinchronichepatitisBinAsianpatients:replicativelevelandriskofhepatocel‐lular carcinoma. Gastroenterology.2008;135(4):1192‐1199.

4. LokAS,ZoulimF,DusheikoG,GhanyMG.HepatitisBcure:fromdis‐coverytoregulatoryapproval.Hepatology.2017;66(4):1296‐1313.

5. ShouvalD,ShiboletO.ImmunosuppressionandHBVreactivation.Semin Liver Dis.2013;33(2):167‐177.

6. Gerlich WH. The enigma of concurrent hepatitis B surfaceantigen (HBsAg) and antibodies to HBsAg. Clin Infect Dis. 2007;44(9):1170‐1172.

7. ZhangJ‐M,XuY,WangX‐Y,etal.CoexistenceofhepatitisBsurfaceantigen (HBsAg) andheterologous subtype‐specific antibodies toHBsAgamongpatientswithchronichepatitisBvirusinfection.Clin Infect Dis.2007;44(9):1161‐1169.

     |  11WALSH et AL.

8. AlavianSM,CarmanWF,JazayeriSM.HBsAgvariants:diagnostic‐escapeanddiagnosticdilemma.J Clin Virol.2013;57(3):201‐208.

9. Ijaz S, SzypulskaR, AndrewsN, Tedder RS. Investigating the im‐pactofhepatitisBvirussurfacegenepolymorphismonantigenicityusingexvivophenotyping.J Gen Virol.2012;93(Pt11):2473‐2479.

10. Marcellin P, ButiM, Krastev Z, et al. Kinetics of hepatitis B sur‐face antigen loss in patients with HBeAg‐positive chronic hep‐atitis B treated with tenofovir disoproxil fumarate. J Hepatol. 2014;61(6):1228‐1237.

11. WongD,LittlejohnM,EdwardsR,etal.ALTflaresduringnucleo‐tideanaloguetherapyareassociatedwithHBsAglossingenotypeAHBeAg‐positivechronichepatitisB.Liver Int.2018;38:1760‐1769.

12. HyakumuraM,WalshR,Thaysen‐AndersenM,etal.Modificationof asparagine‐linked glycan density for the design of hepatitis Bvirus virus‐like particles with enhanced immunogenicity. J Virol. 2015;89(22):11312‐11322.

13. WalshR,HammondR,YuenL,etal.PredictingHBsAgclearanceresponsesduringARC‐520RNAinterference(RNAi)therapybasedonHBsAgepitopeprofileanalysis.J Hepatol.2016;64:S602.

14. GlebeD,KonigA.MolecularvirologyofhepatitisBvirusandtar‐getsforantiviralintervention.Intervirology.2014;57(3–4):134‐140.

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle.

How to cite this article:WalshR,HammondR,YuenL,etal.PredictingHBsAgclearanceingenotypeAchronichepatitisBusingHBsAgepitopeprofiling:Abiomarkerforfunctionalcure. Liver Int. 2019;00:1–11. https://doi.org/10.1111/liv.14207