Quadrivalent Vaccine against Human Papillomavirus to Prevent Anogenital Diseases

Study Design

The Females United to Unilaterally Reduce Endo/Ectocervical Disease (FUTURE) I study is an ongoing, double-blind, placebo-controlled, randomized trial sponsored by Merck; the study was designed and managed and the results were analyzed by the sponsor in conjunction with external academic investigators and an external data and safety monitoring board. From January 2002 through March 2003, 6463 women underwent screening for eligibility; of these, we enrolled a total of 5455 (84%) subjects between the ages of 16 and 24 years at 62 study sites in 16 countries. The institutional review board at each site approved the protocol. Written informed consent was obtained from each subject. The study population was drawn primarily from communities near universities. Healthy women who were not pregnant and had no history of genital warts or abnormal results on cervical cytologic testing and had a lifetime number of no more than four sex partners were eligible. The women were required to use effective contraception during the vaccination period (day 1 through month 7) of the study.

The sponsor collated the data and monitored the conduct of the study. The cutoff date for this manuscript was June 15, 2006. The sponsor and the academic authors proposed the statistical analyses, which were performed by the sponsor. All authors had full access to these analyses and approved the final manuscript. The manuscript was drafted by employees of the sponsor in collaboration with academic authors. All authors vouch for the completeness and accuracy of the data presented.

Vaccine and Randomization

The quadrivalent HPV-6/11/16/18 L1 virus-like–particle vaccine with amorphous aluminum hydroxyphosphate sulfate (Gardasil, Merck) as an adjuvant and the aluminum-containing placebo were visually indistinguishable and have been described elsewhere.²³ A description of the randomization procedure used in the trial can be found in the Supplementary Appendix, available with the full text of this article at www.nejm.org. Vaccine or placebo was administered after a negative result on a pregnancy test of the urine or blood.

The women were observed for 30 minutes after vaccination. Each woman recorded oral temperatures on a vaccination report card 4 hours after receiving the injection and once daily for the next 4 days. Adverse events were recorded with the use of the vaccination report cards for 15 days after vaccination, as were all serious adverse events potentially related to the procedure or vaccine, deaths, and pregnancy outcomes throughout the trial.

Follow-up

At day 1 (the day of the first injection of vaccine or placebo), month 7 (1 month after administration of the third dose), and months 12, 24, 36, and 48, the women underwent a gynecologic examination. At day 1, month 3 (1 month after administration of the second dose), and months 7, 12, 18, 24, 30, 36, and 48, they underwent a comprehensive anogenital examination at which an endocervical and ectocervical swab specimen (considered to be one specimen), a combined labial–vulvar–perineal swab specimen, and a perianal swab specimen (pooled as one specimen) were collected. For a description of the anti-HPV testing and HPV DNA testing, see the Supplementary Appendix. At day 1 and months 7, 12, 18, 24, 30, 36, and 48, cervical samples for Papanicolaou cytologic testing (ThinPrep, Cytyc) were collected.

Lesions

Each study site was provided with a detailed protocol describing the anogenital examination, which included visual inspection of the perianal, vulvar, and vaginal areas observed with the unaided eye, a magnifying glass, or a colposcope. Overt lesions were photographed. Biopsy specimens of lesions considered to be clinically associated with HPV or of unknown cause were obtained. Repeated biopsy of recurrent external anogenital or vaginal lesions was not performed. The treatment of lesions was based on local standards of care. Cervical cytologic specimens were classified according to the Bethesda system (2001) and read at a central laboratory (Diagnostic Cytology Laboratories).²⁴ For the criteria for referral for colposcopy, see the Supplementary Appendix. All biopsy specimens were processed independently to avoid contamination of HPV DNA and were read in a blinded fashion first for clinical management by pathologists at the central laboratory, then for end-point adjudication by a panel of four gynecologic pathologists, as described in the report by the FUTURE II study group elsewhere in this issue of the Journal.²⁵

Case Definition and Primary Hypotheses

There were two efficacy hypotheses. The first stated that, as compared with placebo, the quadrivalent vaccine reduces the combined incidence of anogenital warts, vulvar or vaginal intraepithelial neoplasia grades 1 to 3, or cancer associated with vaccine-type HPV; the second stated that, as compared with placebo, the quadrivalent vaccine reduces the combined incidence of cervical intraepithelial neoplasia grades 1 to 3, adenocarcinoma in situ, or cancer associated with vaccine-type HPV. A case was defined as a consensus diagnosis by the pathology panel of one of these end-point events, with vaccine-type HPV DNA detected in an adjacent histologic section of the same biopsy specimen (Fig. 1 in the Supplementary Appendix). The subject and the investigator were informed of the need for further treatment if the panel rendered a more clinically significant diagnosis. Clinical impressions were recorded but were not used to determine an end point in the efficacy analysis.

Statistical Analysis

A fixed-event design was used for the statistical analysis. At least 38 cases of external anogenital or vaginal lesions associated with vaccine-type HPV and at least 38 cases of cervical lesions associated with vaccine-type HPV were required for the study to have 91% power to declare the vaccine efficacious against at least one of the primary composite end points with a one-sided alpha level of significance of 0.0125 (incorporating multiplicity adjustment), assuming a true vaccine efficacy of at least 80%. To reach the requisite numbers of each of the end points by month 36 in the primary analysis population, and an assumed event rate of 1% per year for each of the coprimary end points, approximately 5400 participants were required. An analysis was conducted when 38 external anogenital and vaginal end-point events and 37 cervical end-point events had occurred (at approximately 1.5 years of follow-up after administration of the third dose, and including all data from visits occurring on or before July 15, 2005). This analysis was included in the application for vaccine licensure, which was approved by the Food and Drug Administration on June 8, 2006. The efficacy and safety analyses presented here include an additional year of follow-up from this ongoing study and data from visits that occurred on or before June 15, 2006.

Figure 1. Figure 1. Populations Included in the Efficacy Analyses.

For each of the primary composite end points, the primary efficacy analysis (Figure 1) was conducted in the HPV-type–specific per-protocol susceptible populations, defined as subjects who were negative on polymerase-chain-reaction (PCR) analysis and serologic testing for the relevant vaccine-type HPV at enrollment, remained negative for the same vaccine-type HPV through 1 month after administration of the third dose of vaccine or placebo, received three doses of vaccine or placebo within 1 year, and did not have protocol violations. Follow-up for case ascertainment started 1 month after administration of the third dose. An exact conditional procedure was used to evaluate vaccine efficacy, assuming that the numbers of cases in the vaccine group and the placebo group were independent Poisson random variables.²⁶ A point estimate of vaccine efficacy and the 95% confidence interval (CI) were calculated on the basis of the observed split in cases between the vaccine and placebo recipients and the accrued person-time. For subjects who had more than one end-point event, only the first event in a category was counted as a case (defined as a consensus diagnosis), but a subject with more than one end-point event could be counted in more than one category of end-point events.

The analyses were performed with respect to the primary composite end points and were further characterized according to type of lesion and vaccine-type HPV. A woman who had a single lesion containing more than one vaccine-type HPV or multiple lesions of different histologic grades was counted once toward the composite end points and once toward the type-specific and lesion-specific end points. For example, for a woman found by consensus diagnosis to have a cervical intraepithelial neoplasia grade 2 lesion associated with HPV-16 at month 12, this end-point event would be counted as a case at month 12. Subsequently, the same subject may have been diagnosed with an HPV-16–associated cervical intraepithelial neoplasia grade 3 lesion and an HPV-18–associated cervical intraepithelial neoplasia grade 2 lesion at month 19. Overall, on the basis of these consensus diagnoses, this subject would be counted once toward the primary composite end point, once toward HPV-16–associated end points, once toward HPV-18–associated end points, once toward cervical intraepithelial neoplasia grade 2, and once toward cervical intraepithelial neoplasia grade 3.

A description of the two additional populations that were included in the prespecified analyses supportive of the primary analysis is shown in Figure 1. Vaccine efficacy was estimated in an unrestricted susceptible population that included all women who were negative on PCR analysis and serologic testing for the relevant vaccine-type HPV at enrollment. We also estimated vaccine efficacy in an intention-to-treat population that included all subjects who had undergone randomization, regardless of their baseline HPV status or evidence of HPV-associated anogenital disease. Follow-up for end-point ascertainment in these two additional populations started after day 1. The analyses in the intention-to-treat population evaluated vaccine efficacy against diseases associated with vaccine-type HPV and the overall effect of vaccination on the number of cases of prevalent and incident disease (caused by vaccine-type HPV or other HPV types). For the intention-to-treat analysis, the cumulative incidence distribution²⁷ was computed for subjects in the vaccine and placebo groups for whom follow-up data were available and was presented graphically for the composite end points, with 95% CIs calculated at 6-month intervals.

Geometric mean antibody titers for vaccine-type HPV and seroconversion rates were measured in subjects included in the type-specific, per-protocol immunogenicity analysis. These subjects were members of the per-protocol population from whom serum samples had been collected during predefined periods.

Adverse events were summarized as frequencies and percentages according to study group and the type of adverse event reported at all visits for the administration of a dose of vaccine or placebo. Risk differences and the associated 95% CIs were calculated by comparing the vaccine and placebo groups for all vaccination visits. A 95% CI (unadjusted for multiplicity testing) that does not include 0 indicates a nominally significant difference at an alpha level of 0.05 (two-sided).

Table 1. Table 1. Baseline Characteristics of the Subjects.

Table 2. Table 2. Subjects Included in and Excluded from the Different Analyses.

From January 2002 through March 2003, a total of 6463 women between the ages of 16 and 24 years were screened for eligibility at 62 study sites in the Asia–Pacific region, Europe, and North, Central, and South America. Of these women, 5455 met the inclusion criteria, and 2723 women were randomly assigned to receive quadrivalent vaccine and 2732 were assigned to receive placebo. Only a small percentage (8 of 5388 subjects [0.15%]) had serologic evidence or HPV DNA evidence of infection with all the HPV types covered by the quadrivalent vaccine (Table 1). The two study groups were well balanced with regard to baseline characteristics and were also similar in the numbers of subjects excluded from the populations analyzed. Subjects in whom vaccine-type HPV was detected at baseline were excluded from the analyses for prophylactic efficacy (Table 2).

Table 3. Table 3. Vaccine Efficacy against External Anogenital, Vaginal, and Cervical Lesions Associated with HPV-6, HPV-11, HPV-16, or HPV-18 or Regardless of HPV Type.

Subjects in this ongoing trial were followed for an average of 3 years after administration of the first dose of vaccine or placebo. At least 83% of those who underwent randomization were included in one or more of the type-specific, per-protocol susceptible populations for external anogenital or vaginal lesions (2261 subjects in the vaccine group and 2279 in the placebo group). The HPV vaccine was 100% effective (95% CI, 94 to 100; 0 cases in the vaccine group vs. 60 cases in the placebo group) in preventing vaginal, vulvar, perineal, and perianal intraepithelial lesions or warts in association with vaccine-type HPV. In the type-specific, per-protocol susceptible populations, in the analysis for cervical end points, among 2241 subjects in the vaccine group and 2258 in the placebo group, the vaccine was 100% effective (95% CI, 94 to 100; 0 vs. 65 cases, respectively) in preventing cervical intraepithelial neoplasia of grades 1 to 3 or adenocarcinoma in situ in association with vaccine-type HPV (Table 3). The estimates of vaccine efficacy made on the basis of the diagnoses at the central laboratory were similar to the estimates made by the pathology panel (data not shown).

More than 95% of the subjects who underwent randomization were included in one or more of the type-specific, unrestricted susceptible populations. The vaccine efficacy was 95% when all grades of external anogenital or vaginal lesions were combined (4 cases in the vaccine group vs. 81 cases in the placebo group), 98% when all grades of cervical lesions were combined (2 vs. 89 cases, respectively), with an efficacy of 91% for high-grade vulvar or vaginal lesions (1 vs. 11 cases, respectively), and 100% for adenocarcinoma in situ (0 vs. 6 cases, respectively) (Table 3). Overall, more than 95% of the subjects received the complete regimen of three doses of vaccine or placebo.

Figure 2. Figure 2. Times to End-Point Events in the Intention-to-Treat Population.

Of the women who underwent randomization, 2672 of 2723 subjects in the vaccine group received at least 1 dose and had at least one follow-up visit for the analysis for external anogenital end-point events and 2669 of 2732 subjects in the placebo group received at least 1 dose and had at least one follow-up visit for the analysis for external anogenital end-point events; 2609 subjects in the vaccine group and 2615 subjects in the placebo group received at least one dose of vaccine or placebo and had at least one follow-up visit for the analysis of cervical end-point events. I bars represent confidence intervals. CIN denotes cervical intraepithelial neoplasia.

Because the public health benefit of a safe and effective HPV vaccine will be measured by its effect in all vaccinated women, we estimated vaccine efficacy in an intention-to-treat population, regardless of the baseline HPV status of the subjects included in the analysis (Table 3). The analysis included women with prevalent infections with vaccine-type HPV or diseases associated with these HPV types. The efficacy against vaccine-type HPV was 73% (95% CI, 58 to 83) when all grades of external anogenital or vaginal lesions were combined (28 cases in the vaccine group vs. 102 cases in the placebo group) and 55% (95% CI, 40 to 66) when all grades of cervical lesions were combined (71 vs. 155 cases, respectively) (Table 3). In the intention-to-treat population, in the placebo group, the incidence of external anogenital or cervical disease associated with vaccine-type HPV continued to increase over time, whereas in the vaccine group the incidence began to plateau (Figure 2). There was no clear evidence that vaccination altered the course of disease or infection present before administration of the first dose (Table 2 in the Supplementary Appendix). No cancers associated with any vaccine-type HPV were identified.

A second intention-to-treat analysis in the population of all women who underwent randomization was performed to evaluate the effectiveness of the vaccine against all anogenital disease (i.e., caused by either a vaccine-type HPV or one not covered by the vaccine). For the primary composite disease end points, in the vaccine group there was a reduction of 34% (95% CI, 15 to 49; 104 cases in the vaccine group vs. 157 cases in the placebo group) in the incidence of external anogenital or vaginal lesions and a reduction of 20% (95% CI, 8 to 31; 344 vs. 421 cases, respectively) in cervical lesions, regardless of the causal HPV type (Table 3 and Figure 2).

For each HPV type covered by the quadrivalent vaccine, at least 99.5% of the subjects in the respective per-protocol immunogenicity cohort had seroconversion at 1 month after the third dose. (For details on the persistence of immune titers, see the Supplementary Appendix.) Five of six recipients of the quadrivalent vaccine in the unrestricted susceptible population who had a genital lesion associated with vaccine-type HPV had antibody titers for an anti-HPV response that were similar to the corresponding anti-HPV antibody response in the per-protocol immunogenicity population. The sixth subject had incorrectly received three doses of placebo.

Table 4. Table 4. Adverse Events.

Vaccine recipients (87%) were more likely than placebo recipients (77%) to have adverse events at the injection site, the most common of these being pain at the site (risk difference, 10 percentage points; 95% CI, 7.8 to 12.1). Erythema, pruritus, and swelling at the injection site were also more common among vaccine recipients than among placebo recipients (Table 4). With respect to systemic adverse events, a nominally higher proportion of vaccine recipients (13.3%), as compared with placebo recipients (10.3%), reported fever between 100°F (37.8°C) and 102°F (38.9°C) (risk difference, 3.0; 95% CI, 1.3 to 4.8) (Table 4). Similar proportions of vaccine and placebo recipients reported a serious adverse event. All systemic and serious adverse events, categorized according to organ system and treatment group, are shown in Tables 3 and 4 in the Supplementary Appendix.

Among the subjects who were seropositive for one or more of the four HPV types at day 1, the profile of adverse events was similar to that of the entire study cohort. For example, of 529 vaccine recipients, 452 (85%) reported one or more injection-site adverse events, as compared with 388 of 507 placebo recipients (77%), with injection-site pain reported as the most frequent adverse event in 84% of the vaccine recipients and 74% of the placebo recipients (risk difference, 10.1; 95% CI, 5.2 to 15.1) (Table 4). One subject in this subgroup had a serious vaccine-related adverse event (bronchospasm 1 day after receipt of the third dose). No multiplicity adjustments were made for these comparisons. For a summary of the pregnancy outcomes in the combined phase 3 quadrivalent vaccine studies, see the report by the FUTURE II study group (and Tables 2, 3, and 4 in that article's Supplementary Appendix).²⁵

These results of the FUTURE I study show that a prophylactic quadrivalent HPV vaccine is highly effective in preventing clinical disease, including anogenital warts and intraepithelial neoplasia of the cervix, vagina, and vulva, associated with HPV-6, HPV-11, HPV-16, and HPV-18. There appears to be no interference among the four HPV types covered by the vaccine, since 100% HPV-type–specific efficacy was observed in the per-protocol analysis. There were relatively few adverse events among the vaccine recipients.

Several steps were taken to ensure that this study would provide a high level of confidence in the safety and efficacy of the quadrivalent vaccine. A diverse population of young women participating in developed and developing countries were enrolled. Data on serious vaccine-related or procedure-related adverse events and pregnancy outcomes were collected for the entire follow-up period. Since HPV-associated vaginal and vulvar lesions may have an ambiguous clinical presentation, an intensive visit schedule with aggressive regimens for genital inspection, biopsy of suspect lesions and cytologic screening, and colposcopy with biopsy were used to ensure a high sensitivity for HPV-associated lesions. High diagnostic accuracy for end-point determinations was provided by a panel of expert pathologists who were unaware of the treatment assignments. Estimates of vaccine efficacy were high whether the diagnoses were made by the pathology panel or by the central laboratory, indicating robust efficacy measurements.

When we evaluated the quadrivalent vaccine's effectiveness against disease associated with the HPV types covered by the vaccine in the intention-to-treat population (as compared with the unrestricted susceptible population), all additional cases detected in the vaccine group occurred in subjects who were infected with vaccine-type HPV before vaccination. Among all randomized subjects with an end-point event within the first year of follow-up, both the vaccine group (57 [97%] cases of disease in 59 subjects) and the placebo group (59 [81%] cases of disease in 73 subjects) had evidence of infection or disease that was prevalent at enrollment. During the second year of follow-up, the incidence of disease associated with vaccine-type HPV in the placebo group continued to increase, whereas in the vaccine group the incidence appeared to reach a plateau, as cases of disease due to prevalent infection were no longer detected and vaccine appeared to reduce the incidence of new infections and associated disease.

A decrease (unadjusted for multiplicity) in overall rates of anogenital disease, regardless of causal HPV type, was also observed. The development of vulvar cancer, though rare, in one subject in the vaccine group highlights the importance of continued screening. Since vaccinated women remain at risk for cervical and genital disease resulting from infections with vaccine-type HPV that might be present at the time of vaccination and from newly acquired infections with HPV types that are not targeted by the quadrivalent HPV vaccine, such women should continue to undergo regular cervical screening for cancer and genital examination, as clinically indicated.

A limitation of our study is the lack of long-term follow-up. The duration of the efficacy of the quadrivalent HPV vaccine and whether boosters are needed are not known. Similarly, to date, no minimum protective anti-HPV antibody titers have been identified. A phase 2 trial of the quadrivalent vaccine showed that at 5 years the vaccine is highly efficacious against infection and disease associated with vaccine-type HPV and that vaccine-induced anti-HPV antibody levels are maintained at or above the levels observed in natural infection.^28,29 At 5 years in this phase 2 study, an antigen challenge resulted in strong anamnestic responses, with sharp rises in antibody titer, indicating the presence of strong, long-lived immune memory.²⁹ The FUTURE I study had limited power to definitively address individual components of the composite study end points; however, the consistency of the results for all components and the results of the FUTURE II study²⁵ are encouraging.

As in other trials of prophylactic vaccines,³⁰ our primary analysis focused on women who at baseline were not infected with vaccine-type HPV. Under conditions that may reflect deviations in vaccine dosing intervals, such as in the unrestricted susceptible population, the efficacy against these HPV types remained high — 95% for external anogenital or vaginal lesions and 98% for cervical lesions. These data also suggest that there is some flexibility in the timing of the vaccination regimen. Adolescent men and women mount higher antibody responses to the quadrivalent vaccine³¹ than do young adult women, but whether the quadrivalent HPV vaccine will prevent genital infection and lesions in men is unknown.

Our data demonstrate the efficacy of a prophylactic quadrivalent HPV vaccine against lesions caused by all the targeted types of HPV. There was also a reduction in the overall incidence of anogenital lesions in the vaccine group. Widespread vaccination of young women and adolescent girls should reduce the incidence of cervical and external anogenital disease associated with HPV-6, HPV-11, HPV-16, and HPV-18. Further research is needed to evaluate the effect of large-scale vaccination programs on the overall burden of HPV disease.

The following investigators participated in the Females United to Unilaterally Reduce Endo/Ectocervical Disease (FUTURE) I study (listed in alphabetical order): Data and Safety Monitoring Board: M. Boulos, J.T. Cox, F. Langmark, J. Modlin (chair), A. Muñoz, V. Odlind, E. Wilkinson; Pathology Panel: A. Ferenczy, R. Kurman, B. Ronett, M. Stoler; Principal Investigators (with asterisks indicating coauthors): Australia — S.M. Garland*; Austria — S. Leodolter*; Brazil — G.I. Andreoni, R.L. Costa, R.P. De Andrade, R. Ferriani, D. Lewi, C. Petta; Canada — N. Ayotte, C. Bouchard, M. Boucher, L. Gilbert, M. Steben,* M. Tolszczuk; Colombia — J. Luna, G. Perez*; Czech Republic — V. Dvorak, B. Micanik, J. Stepan; Germany — E. Barthell, Y. Garnier, T. Grubert, F. Jaenicke, W. Lichtenegger; Hong Kong — G.W.K. Tang*; Italy — G. Carosi, S. Greggi, L. Mariani, M. Moscarini, A. Perino; Mexico — M. Hernandez-Avila,* E. Lazcano; New Zealand — S. Bagshaw, H. Roberts; Peru — L. Jefferson; Puerto Rico — R. Barnes, J. Romaguera; Russia — I. Manukhin, N. Mikhailova, N. Tsvetkova; Thailand — P. Pitisuttithum; United Kingdom — K. Neal, C. Lacey, A. Wade; United States — M. Akin, K. Beutner, D.G. Ferris,* S. Gall, M. Gold, D.M. Harper,* W. Koltun, L.A. Koutsky,* K. Kreutner, W. Ledger, C. Livengood III, S.G. McNeeley, Jr., W. Nebel, P. Rogge, S. Sharma, S. Trupin, Y. Wade, E. Wegner, C.M. Wheeler,* D.M. Whitaker, and M. Yardley; Statistical and Programming Support: K. McCarroll, L. Zhang, H. Zhou.