http://www.diva-portal.org
This is the published version of a paper published in Vaccine.
Citation for the original published paper (version of record):
Gessner, B D., Wilder-Smith, A. (2016)
Estimating the public health importance of the CYD-tetravalent dengue vaccine: vaccine preventable disease incidence and numbers needed to vaccinate.
Vaccine, 34(20): 2397-2401
https://doi.org/10.1016/j.vaccine.2016.03.017
Access to the published version may require subscription.
N.B. When citing this work, cite the original published paper.
Permanent link to this version:
http://urn.kb.se/resolve?urn=urn:nbn:se:umu:diva-127477
ContentslistsavailableatScienceDirect
Vaccine
jo u rn al h om ep a g e :w w w . e l s e v i e r . c o m / l o c a t e / v a c c i n e
Estimating the public health importance of the CYD-tetravalent
dengue vaccine: Vaccine preventable disease incidence and numbers needed to vaccinate
Bradford D. Gessner
a, Annelies Wilder-Smith
b,c,∗aAgencedeMédecinePreventive,Ferney-Voltaire,France
bLeeKongChianSchoolofMedicine,NanyangTechnologicalUniversity,Singapore
cInstituteofPublicHealth,UniversityofHeidelberg,Germany
a r t i c l e i n f o
Articlehistory:
Received28November2015 Receivedinrevisedform1March2016 Accepted9March2016
Availableonline5April2016
Keywords:
Dengue Denguevaccine CYD-TDV Vaccineefficacy Vaccineeffectiveness
Vaccinepreventablediseaseincidence Numberneededtovaccinate
a b s t r a c t
Background:ToevaluatethepotentialpublichealthimpactoftheliveattenuatedtetravalentSanofiPas- teurdenguevaccine(CYD-TDV)weanalyzeddatafromthereportedclinicaltrialstocalculatevaccine preventablediseaseincidence(VPDI)andnumberneededtovaccinate(NNV)basedonthelicensure indicationforpersonsage9yearsandabove.
Methods:VPDIisdefinedasincidenceinanunvaccinatedpopulationXvaccineefficacy(VE),andthus incorporatesbothVEandtheunderlyingburdenofdisease.NNVwascalculatedas100,000dividedby VPDIdividedby2-yearlengthofstudy.Wecomparedthesevaluestodataforthreenewervaccinesthat arecurrentlyintegratedintosomenationalimmunizationprogramsinAsiaandLatinAmerica,namely pneumococcalconjugate,Haemophilusinfluenzaetypeb,androtavirusvaccines.
Results:IntheAsian-Pacifictrial,inthefirst25monthsafterthefirstdoseofthedenguevaccine,CYD-TDV preventedannually2639casesofvirologicallyconfirmeddengueforevery100,000personsvaccinated, foranNNVof18.IntheLatinAmericantrial,giventheoveralllowerannualdengueincidencecompared toAsia,VPDIwas1707,andNNV28.FortheAsian-PacificandLatinAmericanstudies,theVPDIsfor hospitalizedvirologicallyconfirmeddiseaseatthetrials’endwere638and239per100,000population peryear,respectively,withNNVsof75and201.VPDIforconfirmeddenguehospitalizationwashigher thanthatforHibvaccineagainstHibmeningitisorallcauseseverepneumoniawhilelowerthanthatfor rotavirusvaccineagainstsevererotavirusgastroenteritis.
Conclusions:OuranalysisfoundthattheCYD-TDVdenguevaccinehadfavorableVPDIandNNV,also whencomparedtoexistingvaccinesusedinLatinAmericaandAsia.VPDIandNNVvariedbyserotype distribution,extentofpriordengueexposure(baselineseroprevalence)andcountry.Thesefindingswill helppolicy-makersdecidewhereandhowtointroducethisvaccinepost-licensure.
©2016TheAuthors.PublishedbyElsevierLtd.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction
Dengueisanarboviraldiseasethatposesasignificantpublic healthburdeninmostcountriesinthetropicsandsubtropics.With some100millioncasesestimatedtooccurannually,manyofwhich leadtohospitalizations,dengueoutbreakscanoverwhelmalready
Abbreviations:Hib,Haemophilusinfluenzaetypeb;NNV,numberneededtovac- cinate;RCT,randomizedclinicaltrial;VE,vaccineefficacy;VPDI,vaccinepreventable diseaseincidence.
∗ Correspondingauthorat:LeeKongChianSchoolofMedicine,NanyangTechno- logicalUniversity,Singapore.Tel.:+6583328532.
E-mailaddress:anneliesws@gmail.com(A.Wilder-Smith).
fragilehealthcaresystems[1].Theoftenunpredictablenatureof dengueoutbreaksfurtheraggravatesthepublichealthimpact.The increasingincidenceandgeographicexpansionofdenguetrans- missioninthepasttwodecades,accompaniedbytheincreasing socioeconomicburdencompoundedbycostlyyetstillineffective vectorcontrolstrategies,underpintheurgentneedforadengue vaccine[2].
The live attenuated recombinant tetravalent Sanofi Pasteur denguevaccine CYD-TDVwasassessedduringa 25-montheffi- cacysurveillancephase(Phase3trial).Conductedintenendemic countriesinAsiaandLatinAmerica,twomulti-centerefficacytri- alsinvolvedmorethan31,000subjectswithanagerangefrom2 to16years[3,4].Theoverallefficacyforallagegroupsinbothtri- alswas54%forvirologicallyconfirmeddengueofanyseverityor
http://dx.doi.org/10.1016/j.vaccine.2016.03.017
0264-410X/©2016TheAuthors.PublishedbyElsevierLtd.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.
0/).
2398 B.D.Gessner,A.Wilder-Smith/Vaccine34(2016)2397–2401
serotype[5].Theefficacyagainstvirologicallyconfirmeddengue ofanyseverityandserotype,hospitalizations,andseveredisease wasconsistentlyhigherinthoseaged9–16yearsthanyoungersub- jects[5].Becauseofthelowerefficacyandthetransientreversed risk:benefitobservedinthethirdyearinyoungerchildren[5]the agegroupanticipatedtobenefitmostfromthisvaccine,andthe agegroupforwhichthemanufacturerisseekinglicensure,isindi- vidualsfromtheageof9yearsandabove[6].Inthisagegroup,the overallefficacyinthefirst25months(13monthsafterthethird dose)againstvirologicallyconfirmeddenguewas65%,againsthos- pitalization81%andagainstseveredisease93%[5].Follow-upis stillongoingtoassesslong-termefficacyandsafety.Meanwhile, thevaccinehasbeenlicensedinatleast4countriesfortheage rangeof9–45years.
Efficacy results obtained from randomized controlled trials (RCTs)are importantfor licensure.In an RCT,efficacyprovides a measureof proportionatereduction at individuallevel. How- ever, evidence-informed introduction of vaccines into national programs,onceadenguevaccineisindeedlicensed,cannotonly bedrivenbyefficacybecauseefficacyindicateswhetheravaccine worksagainstaspecificoutcomeratherthanprovidinginformation onthevaccine’spublichealthimpact.Inadditiontotheprevention ofinfectionatindividuallevel,theultimategoalofvaccinationis todecreasethepublichealthburdenofdiseaseatthepopulation level.Hence,anadditionalmeasuretoefficacythatmoredirectly establishesavaccine’spublichealthimportanceisthevaccinepre- ventablediseaseincidence(VPDI)[7].VPDIistheincidenceofgiven diseasesyndromepreventablebyvaccineinagivencontext[7].
Furthermore,thenumberneededtovaccine(NNV)isoftenused asametricofthevalueofvaccinationprograms,andcanalsobe usedforcosteffectivenessstudies.NNVisameasuretoquantify thenumberofpeople,orthenumberofvaccinedoses,neededto preventoneeventduetodiseaseandallowthecalculation[8].
ToevaluatethepotentialpublichealthimpactoftheCYD-TDV denguevaccine beyondtheefficacydataalready published, we
analyzeddatafromthepreviouslyreportedclinicaltrials[3–5]to calculateVPDIandNNVbasedonthetargetindicationofages9 yearsand above.Toprovidecontexttothesefindings,for Latin Americawecomparethesevaluestodataforthreevaccineswith clinicaltrialdatafromLatinAmericaandthatarecurrentlyinte- gratedintomostnationalimmunizationprogramsintheregion, namelypneumococcalconjugate[9],Haemophilusinfluenzaetype b[10],androtavirusvaccines[11].
2. Methods
MeasurementofVPDIisdefinedas:incidenceinanunvacci- natedpopulationX vaccineefficacy(VE),and thusincorporates bothVEandtheunderlyingburdenofdisease[12].Thisismath- ematicallyequivalenttotheincidenceinthecontrolgroupminus theincidenceintheinterventiongroup.InprincipleVPDIisbest calculatedfromcommunityrandomizedtrialsasthisallowsincor- porationofthevaccine’sabilitytopreventdiseasethroughboth directandindirectmechanisms.Additionally,itisbestcalculated for clinically ratherthan etiologically definedendpoints as this adjustsfortheinevitablefailuretoconfirmallpreventedoutcomes.
However,currentlypublisheddata[3–5]arelimitedtotheindivid- uallyrandomizedtrialsthatreportVEagainstetiologicallydefined outcomes,whichwillprovidealowerboundoftrueVPDI.
Whileincidencedensitiesforallvirologicallyconfirmeddengue werepresentedintheseparateregionalreports[3,4],thesereports didnotpresentdatathatwouldallowincidencecalculationsfor severeorhospitalizeddengueandtheAsianmanuscriptdidnot presentdataseparatelyforthe9–16yearoldagegroup.Conse- quently,we usedthepooled analysis[5] for thecurrentstudy, based on intention-to-treat analysis, as this presented data on allandhospitalizedvirologicallyconfirmeddenguestratifiedby agegroup.Forchildrenage9–16years,dataforalldenguewere obtainedfromFigure2andforhospitalizeddenguefromFigure 3 [5]. Becausesummary years of follow-up were notreported,
Table1
Calculationofdenguevaccinepreventablediseaseincidence(VPDI)andnumberneededtovaccinate(NNV)(withconfidenceintervals)basedontwo-yearfollow-updata forvirologicallyconfirmeddenguecases,calculatedfromdataabstractedfromFigures2and3ofRef.[5].
Outcome Region Intervention
cases
Control cases
Intervention incidencea
Control incidencea
Vaccine efficacy
VPDI*(95%CI)a NNV(95%CI)
Hospitalizeddengue
Allhospitalized Bothregions 27 70 75 391 81% 316(226,422) 152(114,213)
Asian-Pacific 10 27 151 815 82% 638(365,1008) 75(48,132)
LatinAmerica 17 43 61 310 80% 239(154,346) 201(139,313)
Severe hospitalized
Bothregions 3 22 8 123 93% 114(72,178) 419(270,667)
Asian-Pacific 2 11 29 319 91% 290(130,528) 166(91,370)
LatinAmerica 1 11 3 76 96% 73(34,134) 661(357,1429)
Alldenguecases
Allserotypes Bothregions 367 521 1022 2909 66% 1887(1632,2160) 25(22,29)
Asian-Pacific 90 136 1357 4106 68% 2639(1968,3360) 18(14,24)
LatinAmerica 277 385 995 2774 65% 1707(1440,2016) 28(24,33)
Serotype1 Asian-Pacific 36 52 543 1570 66% 986(576,1488) 49(32,83)
LatinAmerica 99 109 356 785 55% 412(264,576) 116(83,182)
Serotype2 Asian-Pacific 33 26 498 785 37% 276(0,672) 174(71,undefined)
LatinAmerica 84 84 302 605 50% 291(158,437) 165(110,303)
Serotype3 Asian-Pacific 11 18 166 543 70% 363(134,672) 132(71,357)
LatinAmerica 55 106 198 764 74% 543(403,720) 88(67,119)
Serotype4 Asian-Pacific 10 41 151 1238 88% 1044(720,1488) 46(32,67)
LatinAmerica 32 83 115 598 81% 464(346,624) 104(77,139)
Baseline seropos.b
Asian-Pacific 7 17 690 3251 79% 2561(1162,4416) 19(11,41)
LatinAmerica 8 23 358 2156 84% 1798(1008,2832) 27(17,48)
Baseline seroneg.b
Asian-Pacific 7 8 2605 6508 62% 3904(−149,9360) 12(5,undefined)
LatinAmerica 9 9 1674 2899 43% 1225(−768,3768) 39(13,undefined)
aAnnualincidenceper100,000persons.
b Baselineseropositivitywasdeterminedforonlyasubsetofenrolledsubjects;consequently,thedenominatorfortheseoutcomesisdifferentthanforallotheroutcomes.
DenominatorsfortheAsian-Pacificregionwere3316and1656forvaccinatedandcontrolsubjects,respectively,forallenrolledsubjectsandforLatinAmerica13,914and 6940.Forbaselineseropositive,denominatorsfortheAsian-Pacificregionwere487and251forvaccinatedandcontrolsubjectsandforLatineAmerica1073and512.For baselineseronegative,denominatorsfortheAsian-Pacificregionwere129and59forvaccinatedandcontrolsubjectsandforLatinAmerica258and149.
Table2
Calculationofdenguevaccinepreventablediseaseincidence(VPDI)andnumber neededtovaccinate(NNV)percountry,inLatinAmericaabasedon25months follow-updataforvirologicallyconfirmeddenguecases.
Country Controlincidence density
Vaccineefficacy(VE) VPDIb NNV
Brazil 3.7 77.5% 2.9 17
Colombia 2.7 67.5% 1.8 27
Honduras 4.0 71.1% 2.8 18
Mexico 2.5 31.3% 0.8 64
PuertoRico 1.6 57.6% 0.9 54
aTheonlypublisheddatastratifiedbycountryfortheAsian-Pacificregionare inAppendixTable1[3]andtheseincludeallagegroupsdowntoage2years.
Consequently,wedonotpresentcountry-stratifieddataforthisregion.
bVPDIforLatinAmericawascalculatedfromTableS1[4]ascontrolgroupinci- dencedensitymultipledbyvaccineefficacy.Rawdatawerenotpresented,so confidencelimitswerenotcalculated.
weestimatedincidenceascases dividedbynumberofsubjects enrolleddividedbylengthofthestudy,inthiscase25monthsor 2.1years,times100,000toprovidecasespreventedper100,000 personsperyear.UnlikeVPDI,NNVisnotaratebutinsteadthe overallnumberofcasespreventedforagivennumberofpersons vaccinated,andthusincorporatesthelengthofthetrial.NNVwas calculatedas(100,000dividedbyVPDIdividedby2.1-yearlength ofstudy).
Tocalculatethe95%confidenceintervalsforVPDIinTable1,we usedVassarstats (website:http://vassarstats.net/prop2ind.html, lastaccessedFebruary28,2016)asinthiscase VPDIwascalcu- latedasthedifferencebetweenannualincidences.ForTable2,we used OpenEpi (website: http://www.openepi.com/PersonTime2/
PersonTime2.htm,lastaccessedFebruary28,2016)tocalculatethe VPDI95%confidenceintervalssinceinthiscaseincidencedensities wereavailable.
Decision-makers do not judge a vaccine’s importance in a vacuumbutratheragainstotheroptionsforpublichealthinter- ventions.Toprovidecontext,forLatinAmericawecomparedthe calculationofVPDIandNNVfordenguevaccinetothatforpneu- mococcalconjugate[9],Hib[10],androtavirus[11]vaccines.For Asia,weusedcomparisondataforHib[13]androtavirus[14]vac- cines.Thesevaccineswereselectedfortworeasons:(1)clinical trialsexistedspecificallyfromLatinAmericaorAsiathatallowed calculationofVPDIandinallcasesbuttheAsianHibvaccinestudy forNNV;(2)someormostcountriesintherespectiveregionhave
includedthesevaccinesintheirnationalimmunizationprograms and thus have already concluded that they represent efficient useofresources.Aspecificsubtletywasthatsomestudiesused person-yearsofobservationandothersusedpersonsvaccinatedto calculateincidencesinstudygroups,andthisisnotedinthedatawe present.CalculationofNNVinallcaseswasdoneonlyifthenumber ofpersonsvaccinatedbyinterventionandcontrolpopulationswas available.
3. Results
Fortheprimaryendpointofallvirologicallyconfirmeddengue casesinsubjectsaged9yearsandabove,basedonVEdatareported atthe2.1yearfollow-up,theAsianPacificstudyreportedcontrol andinterventiongroupannualincidences of3942and1303per 100,000vaccinatedsubjects,respectively,whichtranslatesintoan annualVPDI of2639per100,000(Table1).Phraseddifferently, theCYD-TDVdenguevaccineprevented2639casesofvirologically confirmeddengueyearlyforevery100,000personsvaccinated,for anNNVoverthe2.1yearstudyperiodof18.Similarresultsfor controlandinterventiongroupincidencesandVPDIintheLatin Americanstudywere2663,956,and1707per100,000population;
theNNVwas28.FortheAsian-PacificandLatinAmericanregions, theannualVPDIsforhospitalizedvirologicallyconfirmeddisease attheendoftwoyearswere638and239per100,000population, respectively,withNNVsof75and201,withlowerVPDIsandhigher NNVsforseverehospitalizeddengue.
Serotype4hadthemostfavorableVPDIandNNV,followedby serotype1(Table1).Amongthesubgroupwithbaselineseropreva- lencedetermined,VEwaslowerinbothAsiaandLatinAmericafor thosewhowereseronegative.Despitethis,inAsiaVPDIwashigher andNNVlowerforpersonsseronegativeatbaselinewhiletheoppo- sitewastrueinLatinAmerica;thisresultistemperedhoweverby thewideconfidenceintervalsintheseronegativegroup.Variations inVPDIandNNValsooccurredbyindividualcountryinLatinAmer- ica(Table2).MexicoforexamplehadalowerVPDIandhigherNNV thanBrazil,duetoalowerbaselinecontrolgroupincidencecom- binedwithalowervaccineefficacy;thelatterinturnmayrelatetoa lowerbaselinedengueseropositivitystatusinMexicoanddifferent circulatingserotypes.
Compared to studies of pneumococcal, Hib, and rotavirus vaccines in LatinAmerica, dengue vaccineefficacy was similar
Table3
Comparisonofvaccinepreventablediseaseincidence(VPDI)andnumberneededtovaccinate(NNV)fordenguevaccinecomparedtoothervaccinesevaluatedandintroduced intonationalimmunizationprogramsinLatinAmerica.VPDIreportedascasesper100,000vaccinatedpersonsperyearexceptwhereotherwisenoted.
Etiology Outcome Vaccineefficacy(95%CI) VPDI NNV
Dengue[5]a Allvirologicallyconfirmedclinicalcases 65%(59,70) 1707 28
Allvirologicallyconfirmedhospitalizedcases 80%(65,89) 239 201 Allvirologicallyconfirmedseverehospitalized
cases
96%(69,100) 73 661
Rotavirus[11]b Confirmedrotavirushospitalization 85%(70,94) 870 200
Allcausegastroenteritishospitalization 42%(29,53) 1790 97
Allcauseseveregastroenteritishospitalization 40%(28,50) 2080 84 Pneumococcus
[9]c
Vaccineserotypeinvasivepneumococcaldisease 100%(77,100) 152 1779 Consolidatedcommunityacquiredpneumonia
(CAP)
22%(8,34) 600 448
CAPwithradiographicconfirmationof consolidationorpleuraleffusion
10%(2,18) 800 306
ClinicallysuspectedCAP 9%(4,14) 1800 135
Haemophilusinfluenzaetypeb(Hib)[10]d Allpneumoniahospitalizationswithconsolidation, effusion,bronchialbreathsounds,orelevated erythrocytesedimentationrate
26%(7,44) 250 Notavailable
aDatacalculatedforpersons9to16yearsofageand2yearfollow-upperiod.
bDatafrom10LatinAmericancountriesplusFinland,forinfantsfollowedfrominfantimmunizationtoage1year.
c DatafromthreeLatinAmericancountries,forchildrenfollowedfrominfantimmunizationtoaverageofalmost3years.VPDIreportedascasesper100,000person-years ofobservation(PYO).
d DatafromChile,forchildrenfollowedinfantimmunizationtoage2years.VPDIreportedasPYO.
2400 B.D.Gessner,A.Wilder-Smith/Vaccine34(2016)2397–2401
Table4
Comparisonofvaccinepreventablediseaseincidence(VPDI)andnumberneededtovaccinate(NNV)fordenguevaccinecomparedtoothervaccinesevaluatedandintroduced intonationalimmunizationprogramsintheAsian-Pacificregion.VPDIreportedascasesper100,000vaccinatedpersonsperyearexceptwhereotherwisenoted.
Etiology Outcome Vaccineefficacy(95%CI) VPDI NNV
Dengue[5]a Allvirologicallyconfirmedclinicalcases 68%(58,76) 2639 18
Allvirologicallyconfirmedhospitalizedcases 82%(61,92) 638 75
Allvirologicallyconfirmedseverehospitalizedcases 91%(58,99) 290 166
Rotavirus[14]b Severerotavirusgastroenteritis(Vesikariscore11+) 48%(22,66) 3000 29
Allcauseseveregastroenteritis 27%(2,46) 3000 38
Haemophilus influenzaetype b(Hib)[13]c
Hibmeningitishospitalization 86% 16 Notavailable
Allcausemeningitishospitalization 22% 158 Notavailable
Allcauseseverepneumonia 5% 264 Notavailable
Allcauseclinicalpneumonia 4% 1561 Notavailable
aDatacalculatedforpersons9–16yearsofageand2yearfollow-upperiod.
b DatafromBangladeshandVietnam,forchildrenfollowedfrominfantimmunizationtoage2years,VPDIreportedascasesper100,000person-yearsofobservation(PYO).
c DatafromIndonesiaforchildrenfollowedinfantimmunizationtoage2years.VPDIreportedasPYO.
againstetiologicallyconfirmeddisease(Table3).Unlikethedengue vaccine trial, trials of the other three vaccines also presented dataonVEagainstclinicalsyndromesincludingpneumoniaout- comesforpneumococcusandHibandallcausegastroenteritisfor rotavirus.Asexpected,VEwasloweragainsttheseclinicalsyn- dromesreflectingthatotheretiologieswereinvolved.However, VPDIwasrelativelyhighindicatingthatevenwithrelativelylow efficacy,vaccinescanhavehighimpactwhenbackgrounddisease ratesarehigh.VPDIforconfirmeddenguehospitalizationwassim- ilartothatfor pneumococcalconjugate vaccineagainstvaccine serotypeinvasivepneumococcaldiseaseandHibvaccineagainst allcausepneumoniabutwaslowerthanthatforrotavirusvaccine againstallcauseseveregastroenteritisandpneumococcalconju- gatevaccineagainstallcausepneumonia.
Compared to studies of Hib and rotavirus vaccines in Asia, denguevaccineefficacyagainwassimilaragainstetiologicallycon- firmeddisease(Table4).VPDIforconfirmeddenguehospitalization washigherthanthatforHibvaccineagainstHibmeningitisorall causeseverepneumoniawhilelowerthanthatforrotavirusvaccine againstsevererotavirusgastroenteritis.
4. Discussion
Policymakersconsistentlystatethatnationaldiseaseburdenis themostimportantfactorinsettingprioritiesforvaccinestobe introducedintopublicsectorimmunizationprograms[15].Despite this,resultsfromvaccineclinicaltrialsfocusontheregulatorycon- cernsofvaccineefficacyandsafetyamongindividualvaccinated subjectsandrarelypresentdatainawaythatallowsassessmentof theexpectedburdenreduction,andthuspublichealthimportance, thatvaccinescanachieve.Intheabsenceofthisinformation,policy- makersandpublichealthadvisorygroupsinaffectedcountriesmay havedifficultymakingrationalrecommendationsanddecisionson whetherandhowtointroducenewvaccines.
Herewepresentananalysisofdatafromdenguevaccineclin- ical trials to illustrate the utility of VPDI, a measure recently describedindetail[7,12].Thisoutcomeprovidesameasureofa vaccine’spublichealth impactbydefining howmanyoutcomes canbepreventedoveracertaintimeperiodbydeliveringadefined quantityofvaccine. Therelated measureof NNVnot onlypro- videsameasureofimmediaterelevancetopolicy-makers,butalso incorporatestheconceptofcasespotentiallypreventedovermul- tipleyears followingprimaryimmunization.Bothmeasureswill varywithunderlyingdiseaseepidemiology,suchasbaselinebur- den,seasonality,agedistributionandwhererelevantserotypeor serogroup distribution.Less wellappreciated is that VE also is notaninvariantqualityandcanvarybymeasuredoutcome(e.g., lowerdengueVEagainstnon-severethanhospitalizeddisease)and geography(e.g.,lowerrotavirusVEinMalawithanSouthAfrica [16]).
OuranalysisfoundthattheCYD-TDVdenguevaccinehadfavor- ableVPDIandNNVwhencomparedtoexistingvaccinesusedin LatinAmericaandAsia.Forexample,inAsia,theCYD-TDVdengue vaccinehadaVPDIforseverehospitalizeddiseaseapproximately equaltotheVPDIforsevereHibpneumonia.InLatinAmerica,while severediseaseVPDIwasrelativelylow,theVPDIforallhospitalized denguewasapproximatelyequaltothesumofinvasiveHibdisease andseverepneumonia.
Moreover,denguevaccinehadahighVPDIagainstlesssevere disease,whichmayhavesubstantialimplicationsforhealthservice utilization.DespitethelowerVEagainstserotype2,therelatively higherincidenceofthisserotypeledtoaVPDIwithintherangeof otherserotypeswiththenotableexceptionsofserotypes1and4 intheAsian-Pacificregion.Lastly,wefoundvariationbybaseline seropositivity(amongthesubgroupthathadseroprevalencedeter- mined)butnotco-linearwithVE.Forexample,inAsiatheVPDIwas higherandtheNNVloweramongthosewhowereseronegativeat baselinedespiteasubstantiallylowerVEamongthisgroup.
This occurred despite available data being limited to etio- logicallyconfirmeddiseasefordengue whilecomparisonswere madetoVDPIsbasedonsyndromicdisease.VPDIfordenguewas calculated foretiologically confirmeddiseasewhile comparison conditionswerecalculatedinpartagainstsyndromicdisease.VPDI calculatedforetiologicallyconfirmeddiseaselikelywillbelower than that for syndromic disease because diagnostic tests have imperfectsensitivity,notallpersonswithsuspecteddengueare tested,andsystemerrors(e.g.,delaysbetweenspecimencollection andprocessing)canleadtofalsenegatives.InFinland,forexample, rotavirusvaccinepreventedovertwiceasmanycasesofallcause asrotavirus-confirmedinpatientacutegastroenteritis[17].Addi- tionally,theindividuallyrandomizednatureofthedenguetrials preventedinclusionofindirecteffects.Consequently,thevalues presentedhereshouldbeconsideredalowerboundonthedengue vaccine’sabilitytoreduceburden.
VPDIis animportant measurebut otherissues affecta vac- cine’spublichealthimportance[18].Forexample,denguevaccine inprinciplecanpreventdiseaseacrossallagegroupswhilethe benefitsofrotavirusandlargelyHibvaccinearelimitedtoearly childhood.Furthermore,adenguevaccinecanpreventthepoten- tiallylargeeffectsofdengueoutbreaksonhealth,healthsystems, andtheeconomy[19].Ontheotherhand,denguerarelycausessig- nificantlong-termsequelaeandhasarelativelylowmortalityrate [2],similartorotavirusbutdistinctfromthedevastatingsequelae andhighmortalityoftencausedbyinvasivepneumococcalandHib disease.
Ourstudywasnotsetouttoaddresssafetyissuesandislimited tothefirst2yearsofthetrial.DuringthethirdyearoftheCYD- TDVtrial,ahighernumberofhospitalizationswereobservedinthe vaccinatedgroupcomparedtotheunvaccinatedgroupforthose individualsof2to5yearsofage.Thereversedrisk:benefitratiois
ofconcernandfurtherlong-termfollow-upofthephase3study participantsisongoing. Nosuchreversed risk:benefitratiowas observedinsubjectsaged9–16years–theagegroupforwhich licensurehasbeensought[20].Becauseoftheongoing concern aboutpotentialantibody-dependentenhancementatthetimeof waningefficacy,theWorldHealthOrganizationrecommendsthat surveillanceforpossibleimmuneenhanceddiseaseshouldbecon- tinuedfor5yearsaftervaccination[21].Waningefficacyaftertwo yearsmayoccur[22].However,suchwaningwillnotaffectthe VPDIandNNVestimatespresentedhereastheseestimatesderived directlyfromtheclinicaltrialresultsduringthetwo-yearreporting period.WaningimmunitymaydiminishVPDIandNNVbeyondthe two-yearperiod.StudiesarecurrentlybeingintroducedbySanofi Pasteurtoevaluatetheneedandtimingofvaccineboosters.
OurdataonVPDIandNNVcanaidincosteffectivenessstud- iesoncethepriceofthevaccineisknown.Otherissuesthatwill needto beaddressedinclude issues suchas immune duration, variabledistributionofdiseaseburdenwithincountries,andpro- grammaticissues suchasvaccineschedulerequirements. Some oftheseremainingquestionsforCYD-TDVcanbeaddressedrea- sonablywithexistingdatafromthePhase3trialsorfuturedata.
For example, if the existing trials collected vaccine impact on syndromicdisease,this shouldbeanalyzed and reported.Post- licensuresurveillance alsowillhelp tofurtherquantifyvaccine effectiveness at the population level, confirm VPDI estimates reported here, assess long-term safety, aid in determining the besttimingforboosterdoses,andmeasuretheindirecteffectof thevaccine[18].Moregenerally,wesuggestthatVPDIandNNV should bepresented alongsideVE in primary trialreports and includedintrialdesignsandanalyticplans.Forexample,vaccines predictedtohavesubstantialindirecteffectscouldhaveacommu- nity(orcluster)ratherthanindividuallyrandomizeddesign.Trials shouldassessbothetiologicallyconfirmedandclinicallydefined outcomes.
Insummary,ouranalysisdocumentsareasonably highVPDI andlow NNV, varyingfromcountrytocountry, and dependent onserotypedistributionandbaselineseropositivity.Thesefindings willhelppublichealthadvisorygroupsandpolicy-makersdecide whereandhowtointroducethefirstdenguevaccine.
Conflictofintereststatement:BDGworksforAgencedeMédecine Preventive(AMP),whichcurrentlyreceivesgrantspecificsupport fromGSK,Merck,Pfizer,andSanofiPasteurandduringthepast3 yearshasreceivedsupportfromCrucell,HillemanLaboratories,and Novartis;noneofthissupportisorwasrelatedtodengueordengue vaccine.BDGservesonadenguevaccineadvisorygroupforSanofi Pasteurbutneitherhenorhisorganizationreceiveremuneration forthisparticipation.AWSreportsnoconflictofinterestsince2011;
from2008to2010shewasthePrincipalInvestigatorfortheadult cohortinthePhase2trialoftheCYD-TDVvaccineinSingapore.
Funding:Nofinancialsupportwasreceivedforthisstudy.
References
[1]BhattS,GethingPW,BradyOJ,MessinaJP,FarlowAW,MoyesCL,etal.The globaldistributionandburdenofdengue.Nature2013;496:504–7.
[2]Wilder-SmithA,MacaryP.Dengue:challengesforpolicymakersandvaccine developers.CurrInfectDisRep2014;16:404.
[3]CapedingMR,TranNH,HadinegoroSR,IsmailHI,ChotpitayasunondhT,Chua MN,etal.Clinicalefficacyandsafetyofanoveltetravalentdenguevaccinein healthychildreninAsia:aphase3,randomised,observer-masked,placebo- controlledtrial.Lancet2014;384:1358–65.
[4]VillarL,DayanGH,Arredondo-GarciaJL,RiveraDM,CunhaR,DesedaC,etal.
EfficacyofatetravalentdenguevaccineinchildreninLatinAmerica.NEnglJ Med2015;372:113–23.
[5]HadinegoroSR,Arredondo-GarciaJL,CapedingMR,DesedaC, Chotpitaya- sunondhT,DietzeR,etal.Efficacyandlong-termsafetyofadenguevaccinein regionsofendemicdisease.NEnglJMed2015;373(13):1195–206.
[6]Wilder-SmithA,MassadE.Agespecificdifferencesinefficacyandsafetyforthe CYD-tetravalentdenguevaccine.ExpertRevVaccines2016:1–5.
[7]GessnerBD,FeikinDR.Vaccinepreventablediseaseincidenceasacomplement tovaccineefficacyforsettingvaccinepolicy.Vaccine2014;32:3133–8.
[8]KellyH,AttiaJ,AndrewsR,HellerRF.Thenumberneededtovaccinate(NNV) andpopulationextensionsoftheNNV:comparisonofinfluenzaandpneu- mococcalvaccineprogrammesforpeopleaged65yearsandover.Vaccine 2004;22:2192–8.
[9]TregnaghiMW,Saez-LlorensX,LopezP,AbateH,SmithE,PóslemanA,etal.
EfficacyofpneumococcalnontypableHaemophilusinfluenzaeproteinDcon- jugatevaccine(PHiD-CV)inyoungLatinAmericanchildren:adouble-blind randomizedcontrolledtrial.PLoSMed2014;11:e1001657.
[10]LevineOS,LagosR,MunozA,VillaroelJ,AlvarezAM,LevineMM,etal.Defin- ingtheburdenofpneumoniainchildrenpreventablebyvaccinationagainst Haemophilusinfluenzaetypeb.PediatrInfectDisJ1999;18:1060–4.
[11]Ruiz-PalaciosGM,Perez-SchaelI,VelazquezFR,AbateH,BreuerT,Clemens SC,etal.Safetyandefficacyofanattenuatedvaccineagainstsevererotavirus gastroenteritis.NEnglJMed2006;354:11–22.
[12]FeikinDR,ScottJA,GessnerBD.Useofvaccinesasprobestodefinedisease burden.Lancet2014;383:1762–70.
[13]GessnerBD,SutantoA,LinehanM,DjelantikIG,FletcherT,GerudugIK,etal.
Incidencesofvaccine-preventableHaemophilusinfluenzaetypebpneumonia andmeningitisinIndonesianchildren:hamlet-randomisedvaccine-probetrial.
Lancet2005;365:43–52.
[14]ZamanK,DangDA,VictorJC,ShinS,YunusM,DallasMJ,etal.Efficacyofpen- tavalentrotavirusvaccineagainstsevererotavirusgastroenteritisininfantsin developingcountriesinAsia:arandomised,double-blind,placebo-controlled trial.Lancet2010;376:615–23.
[15]MuniraSL,FritzenSA.Whatinfluencesgovernmentadoptionofvaccinesin developingcountries?Apolicyprocessanalysis.SocSciMed2007;65:1751–64.
[16]MadhiSA,CunliffeNA,SteeleD,WitteD,KirstenM,LouwC,etal.Effectof humanrotavirusvaccineonseverediarrheainAfricaninfants.NEnglJMed 2010;362:289–98.
[17]LeinoT,OllgrenJ,SaloH,TiihonenP,KilpiT.Firstyearexperienceofrotavirus immunisationprogrammeinFinland.Vaccine2012;31:176–82.
[18]Wilder-SmithA,GublerDJ.PUBLICHEALTH.Denguevaccinesatacrossroad.
Science2015;350:626–7.
[19]BarnighausenT,BloomDE,CafieroET,O’BrienJC.Valuingthebroaderbenefits ofdenguevaccination,withapreliminaryapplicationtoBrazil.SeminImmunol 2013;25:104–13.
[20]Wilder-SmithA,MassadE.Agespecificdifferencesinefficacyandsafetyforthe CYD-tetravalentdenguevaccine.ExpertRevVaccines2016;15(4):437–41.
[21]EdelmanR,HombachJ.Guidelinesfortheclinicalevaluationofdenguevac- cinesinendemicareas:summaryofaWorldHealthOrganizationTechnical Consultation.Vaccine2008;26:4113–9.
[22]VelumaniS,TohYX,BalasingamS,ArchuletaS,LeoYS,GanVC,etal.Lowanti- bodytitersfiveyearsaftervaccinationwiththeCYD-TDVdenguevaccinein bothpre-immuneandnaivevaccines.HumVaccinImmunother2016.
