This randomized, multicenter, double-blind, parallel, three-group controlled, superiority trial tested the efficacy of single-dose bivalent (HPV 16/18) and single-dose nonavalent (HPV 16/18/31/33/45/52/58/6/11) HPV vaccination, as described in the published protocol paper17 and in the report of the primary results10. The study was conducted at three Kenya Medical Research Institute (KEMRI) clinical sites in Kisumu, Thika and Nairobi.

In GAVI-funded countries, including Kenya, multi-dose HPV vaccination is offered to 9–14-year-old girls through the national immunization program. Catch-up vaccination for adolescent girls and young women 15 years of age and older is not provided. Cervical cancer screening is offered to older women instead. We conducted a clinical trial to test the efficacy of single-dose HPV vaccination among young women aged 15-20 years within the context of cytological screening for dysplastic lesions. This was determined to be ethical, as vaccination for this age group in Kenya and many low-HDI countries is not currently supported through national programs or global immunization bodies18.

Participants were eligible for the study if they were born female, aged 15–20 years old inclusive, were sexually active with one to five sexual partners reported in their lifetime, and planned to reside in the study area for 37 months. The exclusion criteria were people living with HIV for whom few data on single-dose HPV VE are available, history of previous HPV vaccination, allergies to vaccine components of latex, pregnancy, hysterectomy, history of autoimmune, degenerative or genetic diseases, and investigator discretion regarding participant safety. Sex assigned at birth was assessed through participant self-report at screening. Participants were recruited through community outreach. All participants, and their parents/guardians in the case of minors, provided written informed consent, which included counseling about randomization, risks and benefits of participation, study procedures and their rights as research participants.

Meningococcal vaccination was chosen as the control because meningococcal antibodies offer potential clinical benefits and do not impact HPV outcomes. Participants were randomized to (1) nonavalent HPV vaccination (Gardasil-9), (2) bivalent HPV vaccination (Cervarix) or (3) meningococcal (control) vaccination. Following randomization, a single dose of each vaccine was administered.

An unblinded statistical analyst generated the randomization sequence using SAS v.9.4. Randomization was stratified by site, using a fixed block size of 15 and a 1:1:1 allocation. Blinded study assignment was implemented via www.randomize.net. Study staff, participants, investigators, clinic staff, laboratory technicians, the end points adjudication committee members and other study team members did not have access to the randomization codes, except for the unblinded statistical analysts and unblinded pharmacists at each site. At the conclusion of the enrollment visit, an unblinded pharmacist entered the participant identification number (PTID) on randomize.net, obtained the next sequential intervention assignment, recorded the PTID and randomization identifier on an eCRF, drew up the vaccine in a masked syringe and administered the vaccination via the intramuscular route. An independent observer, not on the study team, observed the masked vaccination to assess the success of masking.

Potential participants completed eligibility screening with a provider, including a detailed medical history, collection of external genital (labial/vulvar/perineal), lateral vaginal, and cervical swabs for HPV DNA testing, and serum for HPV antibody testing. Participants received cytological screening for cervical cancer screening at enrollment. Sexual and reproductive health services (contraception, STI diagnosis and treatment, HIV testing and HIV pre-exposure prophylaxis) were offered at enrollment and every visit. Participants also received counseling including services for mental health. All questionnaires used electronic case report forms (eCRFs) (DFexplore Software, DF/Net Research).

Participants had study visits at months 3, 6 and then every 6 months for up to 36 months. Providers administered clinical questionnaires and collected a cervical swab at each 6-month visit. Participants self-collected vaginal swabs using validated instructions at month 3; self-collected swabs, which have similar accuracy compared to provider-collected cervical swabs25, were available at subsequent follow-up visits by participant choice or to comply with COVID-19 research restrictions.

Following dissemination and WHO review of the month 18 primary results, participants were offered vaccination at their next study visit, which was at either month 30 or 36, so as not to delay vaccine receipt. Participants provided a final analysis cervical swab before vaccination. Participants in the meningococcal group received the nonavalent HPV vaccine and those in the HPV vaccine groups received the meningococcal vaccine.

HPV DNA genotyping was conducted using the Anyplex II HPV28 assay (Seegene), a multiplexed type-specific real-time PCR-based assay that detects 28 HPV types28,29 at the University of Washington (UW) East Africa STI Laboratory, Mombasa, Kenya with standard proficiency testing30. For HPV-positive samples, a low (+), intermediate (++), or high (+++) positivity was indicated; + or greater was considered positive. The assay runs included negative and positive controls and the housekeeping human gene, β-globin, as an internal control. Runs were performed with CFX96 Real-time PCR System (Bio-Rad).

Serum specimens were shipped to the UW and tested at the Galloway Laboratory, Fred Hutchinson Cancer Research Center. HPV IgG antibodies were detected using a multiplex Luminex assay31,32. The mean pre-established fluorescent intensity seropositivity cutoffs for HPV 16/18/31/33/45/52/58 were used10.

Sexually transmitted infections (N. gonorrhoeae, C. trachomatis or T. vaginalis) were assessed by nucleic acid amplification testing (APTIMA; Hologic/GenProbe) at the UW East Africa STI Laboratory; HSV-2 was evaluated by the Focus ELISA and BV was evaluated using the Nugent score at the National Quality Control Laboratory, Nairobi, Kenya.

The primary trial end point was incident-persistent cervical vaccine type-specific HPV infection among participants who were vaccine-type HPV naive at enrollment. Persistent HPV infection, a surrogate marker for cervical dysplasia/precancer, was defined as high-risk vaccine-type-specific HPV (HPV 16/18 for the bivalent vaccine and HPV 16/18/31/33/45/52/58 for the nonavalent vaccine) detected at two consecutive visits after the month 3 visit, which were obtained no less than 4 months apart (with the same HPV type at both time points). Cervical swabs were tested for the primary end point; vaginal swabs were substituted if necessary. This analysis included follow-up through month 36 to evaluate the durability of VE.

Secondary analyses assessed VE in the sensitivity cohorts and subgroup analyses. The prespecified subgroups were the presence of co-infections (chlamydia, gonorrhea, HSV-2, trichomoniasis, syphilis and BV), self-reported condom use, number of self-reported lifetime sexual partners (1 versus 2+) and contraception method use.

Safety was assessed through adverse event reporting following Division of Allergy and Infectious Diseases Guidelines33. Participants were monitored for adverse events 30 min after vaccination, asked about adverse events at each study visit and reported adverse events outside of study visits. The study clinical monitor followed SAEs, including permission to access medical records. SAEs were recorded on a CRF. The study Principal Investigator and clinical monitor determined the relatedness of SAEs to vaccination. We followed the KEMRI Scientific Ethics Review Unit’s (SERU’s) guidance for reporting SAEs.

The study was powered for the 18-month analysis, which included follow-up through 18 months and has been previously published10. The sample size calculations assumed a combined persistent HPV 16/18/31/33/45/52/58 annual incidence of 5%, single-dose VE of 75%, and loss-to-follow-up of 10% with a fixed follow-up time of 12 months. Sample size calculations assumed that 52% of participants would meet the requirements for inclusion in the 18-month analysis based on the observed prevalence of HPV infection in similar settings34. Assuming a proportional hazards model (seqDesign in R) with 80% power to detect 75% efficacy, a sample size of 2,250 participants was planned.

We used Cox proportional hazards models stratified by study site to estimate the hazard ratios (HRs) of the interventions versus control for the primary and sensitivity analyses. Models for the sensitivity analyses used crude incidence rate ratios instead of the Cox model when no events were observed in a group. Follow-up was calculated as days since the month 3 visit for the primary analysis and days since month 6 for the extended-sensitivity analysis until the first persistent infection. Participants who did not reach this outcome were censored at the last study visit with HPV testing where they did not meet the criteria for persistent infection. VE was expressed as 1 − HR (or relative risk). The log-rank test stratified by study site was used to calculate the P value for each comparison (one degree of freedom). Cumulative incidence curves of time to infection were calculated by intervention group using Kaplan–Meier methods. Efficacy analyses were performed on the month 36 mITT cohorts. In post hoc analysis, we evaluated the absolute difference in cumulative incidence of HPV from the Kaplan–Meier survival estimates at month 36. We calculated the rates of chlamydia and gonorrhea during follow-up by assigned group.

Participants included in the primary analysis tested HPV DNA negative (external genital/lateral vaginal, and cervical swabs) at enrollment and at month 3, by self-collected vaginal swab and HPV antibody negative at enrollment in the mITT cohort. The ITT included all randomized participants. For inclusion in the HPV 16/18 mITT cohort, participants were HPV 16/18 naive. Similarly, for the HPV 16/18/31/33/45/52/58 mITT cohort, participants were HPV 16/18/31/33/45/52/58 naive. Participants without swabs after month 3 did not contribute follow-up time in the primary analysis. Participants in the bivalent vaccine group were excluded from the HPV 16/18/31/33/45/52/58 analysis as the study was not powered to detect cross-protection. Participants who seroconverted to HIV during follow-up are included in analyses.

Sensitivity analyses were planned for the following subsets: participants who tested HPV DNA negative at enrollment and month 3, regardless of antibody status at enrollment (sensitivity cohort) and participants who tested HPV DNA negative at enrollment, month 3, and month 6 and antibody negative at enrollment (extended-sensitivity cohort); the sensitivity cohort was a less conservative definition of an HPV-naive cohort and the extended-sensitivity cohort more closely matched the analysis cohort for HPV vaccine licensure trials. The extended-sensitivity cohort excluded participants who might have had prevalent HPV infection at enrollment that was not detected. Safety was assessed among all participants; the three groups were compared using Fisher’s exact test. Exploratory analysis evaluated cross-protection of the bivalent vaccine against HPV 31/33/45. We performed all analyses using SAS software, v.9.4 (SAS Institute) and R (v.4.2.2).

An independent Data Safety and Monitoring Board was constituted to review study progress, participant safety and the primary outcome, and met annually. The trial is registered at ClinicalTrials.gov (NCT03675256).

Data for this study, including from Kenya, were collected via eCRFs in Kenya. Seven colleagues (M.A.O., E.A.B., B.N., I.W., C.B., S.K. and N.R.M.), including the senior author (N.R.M.) are from Kenya, a low-and-middle-income country and one other (R.V.B.) is South African and is now based in a high-income country. We fully endorse and are committed to the Nature Portfolio journals’ guidance on low-and-middle-income country authorship and inclusion.

This research is locally relevant to Kenya and other countries that have not achieved the 90% HPV vaccine coverage goal.

The KEMRI SERU (nos. 3745 and 3741) and the Massachusetts General Hospital Institutional Review Board (no. 2022P001178) approved the study. Study participation may have carried stigmatization associated with vaccination. The data collection and analysis techniques employed raised no risks pertaining to incrimination, discrimination, animal welfare, the environment, health, safety, security or other personal risks. All HPV and STI testing was conducted at local laboratories. Serum for specialized HPV antibody testing was shipped to Seattle for testing. No cultural artifacts or associated traditional knowledge has been transferred out of any country. In preparing the manuscript, the authors reviewed relevant studies from Kenya.

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.