M A J O RA R T I C L EFall in Human Papillomavirus PrevalenceFollowing a National Vaccination ProgramSepehr N. Tabrizi,1,2,3,4Julia M. L. Brotherton,5,8John M. Kaldor,9S. Rachel Skinner,8Eleanor Cummins,4Bette Liu,9Deborah Bateson,10Kathleen McNamee,6,7Maria Garefalakis,11and Suzanne M. Garland1,2,3,41Regional World Health Organization Human Papillomavirus Laboratory Network, Department of Microbiology and Infectious Diseases, The RoyalWomens Hospital, Victoria, Australia; 2Department of Obstetrics and Gynecology, University of Melbourne, Australia; 3Department of Microbiology,and 4Murdoch Childrens Research Institute, Royal Childrens Hospital, Parkville, Australia; 5Registries, Victorian Cytology Service, East Melbourne,Australia; 6Family Planning Victoria, Box Hill, and 7Department of Obstetrics and Gynecology, Monash University, Clayton, Victoria, Australia;8Sydney University Discipline of Pediatrics and Child Health, Childrens Hospital Westmead, Australia; 9The Kirby Institute, University of New SouthWales, Darlinghurst; 10Family Planning New South Wales, Asheld; and 11Family Planning Western Australia, Northbridge, Western Australia(See the Editorial Commentary by Hariri and Markowitz, on pages 16335.)Background. InApril 2007, Australiabecamethe rst countrytointroduceanational government-fundedhumanpapillomavirus (HPV) vaccinationprogram. We evaluatedthe programs impact ongenotype-specicHPV infection prevalence through a repeat survey of women attending clinical services.Methods. HPVgenoprevalenceinwomenaged1824yearsattendingfamilyplanningclinicsintheprevac-cine period (20052007) was compared with prevalence among women of the same age group in the postvaccineperiod (20102011). The same recruitment and testing strategies were utilized for both sets of samples, and com-parisons were adjusted for potentially confounding variables.Results. The prevalence of vaccine HPV genotypes (6, 11, 16, and 18) was signicantly lower in the postvac-cinesamplethanintheprevaccinesample(6.7%vs28.7%; P < .001), withlowerprevalenceobservedinbothvaccinatedandunvaccinatedwomencomparedwiththeprevaccinepopulation(5.0%[adjustedoddsratio, 0.11;95%condenceinterval, 0.060.21]and15.8%[adjustedoddsratio, 0.42; 95%condenceinterval, 0.190.93],respectively). AslightlylowerprevalenceofnonvaccineoncogenicHPVgenotypeswasalsofoundinvaccinatedwomen (30.8% vs 37.6%; adjusted odds ratio, 0.68; 95% condence interval, 0.460.99).Conclusions. Four years after thecommencement of theAustralianHPVvaccinationprogram, asubstantialdecrease in vaccine-targeted genotypes is evident and should, in time, translate into reductions in HPV-related lesions.Australia was therst country to implement a nationalgovernment-funded vaccination programto preventcervical cancer. FromApril 2007throughDecember2008, a school-based delivery strategy was used tooffer free human papillomavirus (HPV) vaccination togirls aged1217years, andfromJuly2007throughDecember2009, general practitionersandothercom-munityproviders offeredfree vaccinationtowomenaged 26years. Since2009, routineHPVvaccinationhascontinuedfor girls in therst year of high school(age, 1213years)aspartoftheNational Immuniza-tionSchedule. During20072009, anestimated83%of girls aged1217receivedat least 1doseof HPVvaccineand70%completedthe3-doseHPVvaccina-tioncourse[1]. Amongwomenaged1826years, re-cordedcoverageinthisperiodwas55%and32%for1-doseand3-dosecoverage, respectively[2]. Thiswasless than true coverage because notication in this agegroup to the national vaccine register was not compul-sory, although it did attract a small noticationpayment per dose [2]. The Australian HPVvaccinationprogramhasexclusivelyusedthequadri-valentvaccine, which protects against HPVgenotypesReceived17April 2012; accepted15June2012; electronicallypublished19October 2012.A portion of this data was presented at Preventing Cervical Cancer: IntegratingScreeningandVaccinationConference, on911November 2011inMelbourne,Australia.Correspondence: Sepehr N. Tabrizi, PhD, Department of Microbiology and Infec-tious Diseases, The Royal Womens Hospital, Locked Bag 300, Parkville, Victoria3052, Australia (sepehr.tabrizi@thewomens.org.au).The Journal of Infectious Diseases 2012;206:164551 The Author 2012. Published by Oxford University Press on behalf of the InfectiousDiseases Society of America. All rights reserved. For Permissions, please e-mail:journals.permissions@oup.com.DOI: 10.1093/infdis/jis590Postvaccine HPV Prevalence Reduction JID2012:206(1 December)1645 by guest on August 10, 2015http://jid.oxfordjournals.org/Downloaded from 16 and 18 [3], the main causes of high-grade cervical intraepi-thelial neoplasia(CINgrades2and3)inAustralianwomen,and genotypes 6 and 11, which cause most genital warts [4, 5].Denitiveproof of theprogramseffectivenessinreducingcervical cancer will not be evident for several decades.However, there have already been early signs of success, mani-fested as substantially lower numbers of both newgenitalwarts presentations at sexual health clinics [6, 7] and histologi-callyconrmeddiagnoses of high-grade cervical lesions [8]amongyoungwomeninthevaccinecohort, comparedwiththeir predecessors at the same age.Another outcomemeasurethatisincreasinglybeingrecog-nized as relevant for HPV vaccine evaluation is the prevalenceof infectionwithvaccine-targetedgenotypes. Withplacebo-controlledtrialsnolongerethicallyfeasible, genotypepreva-lence has been endorsedas an appropriate surrogate outcomeforcomparativetrialsof newHPVvaccineconstructs[9], aswellasprovidinganimportantbasisforpopulationmonitor-ing of vaccine program effectiveness [1012].We conductedAustralias rst studyof HPVgenopreva-lence, known as WHINURS (Women, Human papillomavirusprevalence, Indigenous, Non-Indigenous, Urban, Rural Study),just prior to the implementation of the national HPV vaccina-tionprogram. Thisstudyinvolvedrecruitment of womenat34 sentinel clinical sites from around the country and provid-edestimatesof theprevalenceof specicgenital HPVgeno-types[13].With 4 yearshaving elapsedsince implementationof the vaccination program, we initiated a follow-up survey toassesschangesintheprevalenceofHPVgenotypesinyoungAustralian women. Here we report on therst analysis ofnd-ings in women aged 1824 years at participating clinics in 3 ofAustralias major cities.METHODSStudy Design and ParticipantsWe used a repeat cross-sectional design to compare HPVprevalence in2 samples of womenrecruitedfromsentinelclinical sites. The rst or prevaccine implementationsample was made up of women aged 1824 years (at the timeofrecruitment)whowererecruitedfromparticipatingfamilyplanning clinics (FPCs) for Papanicolaou screening during therecruitmentperiodof20052007andhadparticipatedintheWHINURS study[13]in Sydney, Melbourne, andPerth, 3ofAustralias 4largest cities andhome tonearlyone-half theAustralianpopulation. All FPCsinAustraliaaregovernmentfundedandprovideawiderangeof reproductiveandsexualhealth services, including clinical screening, diagnosis andtreatment, professional education, and health promotion. EachofAustraliasmajor citieshaseither1or2FPCsites thatarerecognized as attracting a wide range of clients from across thesociodemographic spectrum.The second or postvaccine implementation sample com-prisedwomenaged1824years whohadattendedFPCs inthesamecitiesduringtherecruitmentperiod20102011forPapanicolaouscreening. These womenwouldhavebeenaged1420yearsatthetimetheywerevaccinated, if theypartici-patedintheschool-orcommunity-basedcomponentsoftheHPV vaccination programbetween 2007 and 2009. Anyimpact measured in this age group would most closely approx-imate the benet that might be expected for future cohorts ofadolescents. Based on vaccine registry data, the recorded HPVvaccine coverage among women aged 1824 years in 2011 was74%for1doseand55%for3doses[1, 2]. UnderAustralianguidelines, cervical screening begins at age 18 years or 2 yearsafter rst intercourse, whichever is later. Therefore, onthebasisof theirage, thewomeninthesecondsamplewereinboththevaccinatedcohortandeligibleforcervical screeningat thetimeof recruitment. Therecruitment forthepostvac-cinesampleisongoing, andthecurrentstudyisbasedonaninterim data analysis.Of 2 FPCs in Melbourne, only 1 participated in the prevac-cine phase, whereas both joined the postvaccine phase. Partic-ipantswererecruitedfrombothPerthFPCsinbothphases.The Sydney FPC that had participated in the prevaccine phasemovedpremises soonafterwardtoanearbysitewithinthecity. Both this site and a second major Sydney FPC participat-ed in the postvaccine phase. All the clinics participating in theprevaccine and postvaccine phases were located withinthesame metropolitan areas.The procedures to recruit participants were identical for theprevaccine and postvaccine samples. Clinic staff identied age-eligible womenat the time of a consultationthat includedroutine Papanicolaou testing. Invitation to participate depend-ed on judgment by the clinician that there was sufcient timetoraisethestudyinthecontextoftheclinical visit. Written,informed consent was obtained from those who agreed to par-ticipate. At the time of the Papanicolaoutest, eachwomanhad a sample of exfoliated cervical cells collected in Preservcyt(ThinPrep, Cytyc) for HPVDNAtesting. Informationonage, current useof hormonal contraception, smokingstatus,and postcode of residence was collected. Those enrolled in thepostvaccinegroupwerealsoaskedtoprovideinformationre-lating to HPVvaccination status, sexual and reproductivehistory, and knowledge of risk factors for cervical cancer. Ap-proval toconduct thestudywasobtainedfromresearchandethics committees at each of the sites that enrolledparticipants.Laboratory TestingIdentical laboratory methods for HPVgenotype detectionwere used for boththe prevaccine and postvaccine groups[13]. Briey, 1 mLof cervical cellsinPreservcytwaspelletedandresuspendedin200 Lof phosphate-bufferedsalinefor1646JID2012:206(1 December) Tabrizi et al by guest on August 10, 2015http://jid.oxfordjournals.org/Downloaded from DNAextractionusingtheautomatedMagNAPureLCisola-tionandpuricationsystem(RocheMolecularSystems)withtheDNA-Iisolationkit [14]. Followingnucleicacidisolation,all samples were initially assessed for the presence of 13 high-risk HPV (HR-HPV) genotypes using the Amplicor HPV testkit (Roche Molecular Systems). Anyspecimenshowntobenegative for HR-HPVtypes was testedfor the presence ofmucosal HPV DNA using L1 consensus primer set PGMY09-PGMY11 [15]. Polymerase chain reaction (PCR) productswere detected by enzyme-linked immunosorbent assay(ELISA)usingagenericbiotin-labeledprobefordetectionofthepresenceof mucosal HPVsequencesinthesample[16].SamplespositiveforHPVbyeithertheAmplicortestorthePGMY09-PGMY11PCR-ELISA weresubsequentlygenotypedusing the Linear Array HPV genotyping test (Roche) modiedby using an automated blot processor, BeeBlot(Bee Robotics),for the hybridization and washing steps, as previously validat-edbyourlaboratory[1719]. HPVgenotypingprolesweremanually interpretedandveriedusing the HPVreferenceguide provided with each test kit. Any sample positive for theHPV52/33/35/58 probe line onthe Linear Array test wasfurther testedto conrmthe presence or absence of HPVtype 52 [20].Statistical AnalysisThe primary analyses compared women in the prevaccine andpostvaccine samples, withthe latter being further stratiedinto2groupsbasedonwhetherornotparticipantsreportedhavingreceivedatleast1doseofHPVvaccine. Comparisonsof sociodemographic characteristics and HPV prevalence werethenmadeacrossthe3groups ( prevaccine, postvaccineun-vaccinated, andpostvaccinevaccinated). It washypothesizedthatifthegroupshadcomparablesociodemographicproles,wewouldseealowerprevalenceof vaccinetypesinwomenwho had received the vaccine, compared with both the prevac-cine women and the postvaccine women who had not receivedthe vaccine.We analyzed HPV prevalence as (1) any HPV genotype; (2)anyHR-HPVgenotype(16, 18, 31, 33, 35, 39, 45, 51, 52, 56,58, 59, or 68); (3) any HR-HPVgenotype excluding thevaccine genotypes 16 and 18; (4) vaccine-preventable HPV ge-notypes (6, 11, 16,and 18); and(5) other prespecied group-ingsofnonvaccinegenotypes. Vaccineeffectiveness(VE)wascalculated by comparing the prevalence of HPVinfectionbetweenthevaccinatedandunvaccinatedgroupinthepost-vaccine sample using the standard formula VE = 1 RR,where RR is the ratio of the prevalences in these 2 postvaccinesamples.We also compared the groups of women according to socio-demographic variables including age, socioeconomic status(upperorlower50thpercentile, basedontheSocioeconomicIndexes for Areas, a score that ranks Australianresidentialpostcodesaccordingto informationfromthe 2006Censusonindicators including median income, education level, andhousehold numbers) [21], residential area (major city or other,based on the Accessibility/Remoteness Index of Australia[ARIA], classies residential postcodes into major cities) [22],currentuseofhormonalcontraceptives, andcurrentsmokingstatus.In addition, for postvaccine study participants, we alsocompared vaccinated and unvaccinated women with regard totheir reported age at rst sexual intercourse (in years), educa-tion(higher or lower, basedoncompletionof at least highschool or technical college), and whether they were Australianborn. To test for differences in proportions or means betweenthegroupsofwomen, 2testsorttestsandanalysisofvari-ancewere used, respectively. We usedlogistic regressiontoadjustcomparisonsof prevalencebysociodemographicchar-acteristics that were found to vary between the groups ofwomenandreportadjustedoddsratios(ORs). Statistical sig-nicancewastestedat thelevel of P < .05. All analyseswereconducted using Stata software (version 10.1).RESULTSDuring the prevaccine phase of recruitment, metropolitanFPCsinSydney, Melbourne, andPerthrecruited202womenaged 1824 years. During the postvaccine phase, FPCs recruit-ed 404 women aged 1824 years. In the postvaccine sample, 9womenreportedthattheydidnotknowwhether ornottheyhad received any doses of the HPV vaccine and were excludedfrom analyses by vaccine status.Women fromthe prevaccine sample were slightly olderthanthosefromthepostvaccinesample(meanage, 21.6vs21.2 years; P = .005) and less likely to be using hormonal con-traceptives at the time of recruitment (60.9% vs 74.2%;P = .001), but other characteristics were similar betweengroups (Table 1). Of women recruited to the postvaccinesample, 85.6% (n = 338) reported that they had been vaccinat-ed with at least 1 dose, and of those who recalled the numberofdoses, 279(86.4%) of323womenstatedthattheyreceivedall 3, giving a self-reported3-dose coverage rate of 70.6%.Withinthepostvaccinesample, womenwhostatedthat theyhadbeenvaccinatedwereslightlyyoungerthanunvaccinatedwomen(meanage, 21.1vs21.8years; P = .009), hadahighereducationstatus (95.0%vs 87.7%completedhighschool ortechnical college; P = .03), andweremore likelyto be born inAustralia (88.5%vs 59.7%; P < .001). These 2 groups weresimilarfor othercharacteristicsincludingtheir ageat rstre-ported sexual intercourse (mean age, 16.5 vs 16.7 years;P = .3), socioeconomic status, useofhormonalcontraceptives,and smoking status.Acrossthe3categoriesofwomen( prevaccine, postvaccineunvaccinated, and postvaccine vaccinated), the onlyPostvaccine HPV Prevalence Reduction JID2012:206(1 December)1647 by guest on August 10, 2015http://jid.oxfordjournals.org/Downloaded from characteristicswhichdifferedsignicantlywereageandhor-monal contraceptiveuse(Table1); wethereforeadjustedforthese when comparing HPV prevalence across these 3 groups.There were highly signicant differences in crude HPVprevalence betweenthe prevaccine andpostvaccine samples(Figure 1). The prevalence of any HPV genotype was 59.9% vs48.0%intheprevaccineandpostvaccinegroups, respectively(P = .006). For the vaccine-targeted HPV genotypes, the diffe-rence was even greater at 28.7% vs 6.7%, respectively(P < .001). When the postvaccine sample was stratied accord-ing to self-reported vaccination status, there were highly signif-icant differences across the 3groups inthe proportions ofwomenpositive for anytype of HPV, HR-HPVtypes, andHPV vaccine genotypes (Table 2). Vaccinated womenconsistentlyhadthelowestprevalencesof HPV, andthedif-ferences were particularly striking for the HPVgenotypescoveredbythevaccine, withonly5.0%ofvaccinatedwomeninthe postvaccine samplehaving1of these genotypes. Wealsofound signicant differences betweenthe 3 groupsin theprevalenceofHPVtypes6, 16, and18(Table2)andnosig-nicant differences for selected other HR-HPVgenotypes.Vaccineeffectivenessagainstvaccine-typeHPVinfectionwas73% (95% condence interval [CI], 48%86%; P < .001).After adjusting for age and use of hormonal contraceptives,compared with women in the prevaccine sample, those in thepostvaccine sample who reported being vaccinated were signif-icantlylesslikelytobepositiveforanyHPVgenotype(OR,0.49; 95% CI, 0.340.71), HR-HPVgenotypes (OR, 0.48; 95%Table 1. Characteristics of Women According to Study and Vaccination StatusCharacteristicPrevaccineGroup(n =202)PostvaccineGroup(n = 404) PaPostvaccineUnvaccinatedGroup (n = 57)PostvaccineVaccinatedGroup (n =338) PbAge, mean years (SD) 21.6 (1.8) 21.2 (1.8) .005c21.8 (1.9) 21.1 (1.8) .001cResiding in major city 192 (95.1) 362 (90.9) .08c49 (87.5) 304 (91.3) .1Higher socioeconomicgroup159 (78.7) 312 (78.0) .8 40 (71.4) 266 (79.4) .4Hormonal contraceptive use 123 (60.9) 299 (74.2) .001c36 (63.2) 257 (76.3)