HPVs infect the stratified squamous epithelia of skin and mucous membranes, where they cause benign lesions, some of which have the potential to progress to invasive cancer. (2–4). HPVs are small, nonenveloped viruses whose approximately 8-kb circular genome encodes 2 structural proteins, L1 and L2, that form the viral capsid, plus several nonstructural proteins that are important for the virus life cycle but are not incorporated into virions. To establish infection, microtrauma or erosion of the overlying epithelial layers is thought to enable HPVs to infect cells of the basal epithelial layer, where the stem cells and other long-lived cells are found (Figure ​(Figure1).1). HPV infections tend to last months or years because the viral genome successfully parasitizes these cells and because the virus evades the immune system by limiting most viral gene expression and viral replication to suprabasal cell layers. Most infections are self-limited, presumably because the host eventually mounts a successful immune response.

The benign lesions induced by HPVs include nongenital and anogenital skin warts, oral and laryngeal papillomas, and anogenital mucosal condylomata (Figure ​(Figure2).2). Anogenital infections are almost always transmitted sexually. Long-term infection by a subset of HPVs can lead to malignant anogenital tumors, including cancers of the anus, penis, vulva, vagina, and cervix (5–7). A proportion of oral cancer is also attributable to HPV (8, 9). While HPV infection has been associated on limited occasions with esophageal cancer and skin cancer, a frequent causal link, although plausible, remains more tenuous (6, 10).

Among the cancers attributable to HPV infection, cervical cancer has received the most attention (11), as it accounts for about 10% of all cancers in women worldwide. Cervical cancer is the second most common cause of death from cancer, after breast cancer, among women worldwide. The interval between the acquisition of HPV infection and malignant progression usually takes at least 10 years (Figure ​(Figure3)3) and is frequently longer (7, 12). Cervical cancer is therefore very uncommon in women under 25; the incidence rises progressively for women over 25 and is highest for women over 40. About 80% of cervical cancers occur in less-developed countries, primarily because they lack sufficient resources for high-quality cervical cancer screening programs that detect cervical abnormalities via Pap smear testing or testing for the presence of cervical HPV DNA (13). Virtually all cases of cervical cancer are attributable to sexually acquired HPV infection (14), while infection by HPV accounts for a variable proportion of the other tumors in which the virus has been implicated etiologically. Of all cancer cases linked etiologically to HPV, cervical cancer accounts for about two-thirds of them. This cancer can result from infection by any 1 of about 15 oncogenic HPV types, but HPV16 and HPV18 predominate, accounting for about 50% and 20% of cervical cancer, respectively (15). These 2 types account for an even higher proportion of the other genital and mucosal cancers attributable to HPV infection (6).

The recognition of HPV as the cause of cervical cancer and other diseases therefore implied that an effective HPV vaccine should be able to interfere with the benign and malignant conditions attributable to HPV infection. However, approved prophylactic vaccines have been directed against infectious agents that cause systemic disease, and efforts to develop vaccines against sexually transmitted agents such as HPV whose disease results from local infection had not proven successful. It is believed that neutralizing antibodies form the cornerstone of most prophylactic vaccines, and viruses that cause disease only after passing through the circulation are accessible to the neutralizing antibodies present in the blood (22). Another limitation was that the presence of oncogenes in HPVs suggested that a subunit vaccine approach would theoretically be preferable to an inactivated vaccine or an attenuated live-virus vaccine, and it was unclear whether a subunit HPV vaccine would have the potential to be effective against the local infections caused by genital HPVs.

Despite these uncertainties, 2 pharmaceutical companies, Merck and GlaxoSmithKline, have recently reported a remarkable degree of protection by candidate prophylactic HPV vaccines (see below). The vaccines that both companies are developing are subunit virus-like particle (VLP) vaccines composed of a single viral protein, L1, which is the major structural (capsid) protein of the virus and contains the immunodominant neutralization epitopes of the virus. The vaccines are based primarily on preclinical research showing that (a) when expressed in cells, L1 has the intrinsic ability to self-assemble into VLPs (Figure ​(Figure4)4) (23–26) that can induce high levels of neutralizing antibodies (23, 27, 28); (b) in animal models of animal papillomavirus infection, parenteral vaccination with L1 VLPs protects from high-dose challenge with homologous virus (29–31), while animals are not protected by systemic immunization with denatured L1 or L1 VLPs from a heterologous papillomavirus because L1 neutralization epitopes are conformationally dependent and predominantly type specific (29); and (c) protection can be passively transferred by immune IgG (29, 30). The VLPs from the Merck vaccine are produced in yeast (32–34), while the VLPs from the GlaxoSmithKline vaccine are produced in insect cells via recombinant baculovirus (35). Merck uses alum as an adjuvant in its vaccine, while GlaxoSmithKline uses AS04, a proprietary adjuvant composed of alum plus monophosphoryl lipid A (a detoxified form of lipopolysaccharide). Both vaccines use purified particles, which are given as 3 intramuscular injections over a 6-month period.

Of course, it would be desirable for an HPV vaccine to have the ability to prevent all cases of cervical cancer. However, although the 15 oncogenic types implicated in cervical cancer are more closely related to each other phylogenetically than they are to the HPVs that cause nongenital skin lesions (warts) (36), the immunodominant epitopes in L1 VLPs induce neutralizing antibodies that are predominantly type specific (37). It has therefore been necessary, at least for the first generation HPV vaccines, to focus on the HPV types found most frequently in cervical cancer. This consideration has led both companies to focus on HPV16 and HPV18, which, as noted above, account for about 70% of cases of cervical cancer. The Merck vaccine also targets HPV6 and HPV11, which together account for about 90% of external genital warts (38); the latter 2 types also infect the cervix, but are not implicated in cervical cancer. Thus, the GlaxoSmithKline vaccine that is currently in phase III trials is a bivalent vaccine composed of VLPs from HPV16 and HPV18, while the vaccine that Merck has used for its phase III trials is a quadrivalent vaccine that contains VLPs from HPV6, HPV11, HPV16, and HPV18.

When considering appropriate end points for determining vaccine efficacy, the most relevant end points recommended by an FDA vaccine advisory panel (39) were a reduction in the incidence of vaccine type–specific persistent infections and of associated moderate- and high-grade cervical dysplasias and carcinomas in situ, which together are referred to as cervical intraepithelial neoplasia 2+ (CIN2+; CIN is graded as CIN1, CIN2, and CIN3 for low-, moderate-, and high-grade dysplasia, respectively). HPV DNA testing results have been shown to be substantially more reproducible than the pathological diagnosis of dysplasia (40), but moderate- and high-grade dysplasias represent clinical end points that trigger therapeutic intervention. It is important to note that it would be unethical to use cervical cancer as a primary end point for vaccine efficacy trials, as cervical cancer screening can prevent the vast majority of cancer through the identification of precancers, which are then treated. Also, the interval between infection and the development of invasive cancer usually takes more than 10 years (7, 12).

Following the observations that systemic vaccination with a monovalent HPV16 L1 VLP vaccine was safe and highly immunogenic (41), even without adjuvant, a Merck-sponsored proof-of-principle efficacy trial of an HPV16 L1 VLP vaccine reported that fully vaccinated women who were HPV negative throughout the vaccination period were completely protected against the development of persistent incident infection with HPV16 when followed for an average of 17 months (Table ​(Table11 and ref. 32). This vaccine cohort has now been followed for an average of 3.5 years, and the high level of protection was maintained throughout this period (Table ​(Table11 and ref. 34). After the initial peak, the levels of serum antibodies appeared to decline approximately 10-fold over the first 2 years following vaccination but then remained stable for the remainder of the follow-up period. If the level of serum antibodies represents a surrogate for protection against infection, the stability of the antibody titers would suggest that high-level protection may continue substantially beyond 3.5 years. Preliminary efficacy results from Merck’s quadrivalent vaccine were also published and showed excellent protection against the viruses targeted by the vaccine (Table ​(Table11 and ref. 33). In unpublished studies, Merck has reported at scientific meetings on the clinical efficacy of the multinational phase III trial of their quadrivalent vaccine, with an average follow-up of 1.5 years (42, 43). When fully vaccinated women who remained negative for infection throughout the vaccination period were analyzed against the comparable placebo cohort, the vaccine was 100% effective in preventing CIN2+ associated with HPV16 or HPV18 and also in preventing external genital warts associated with HPV6 or HPV11. Even when the efficacy was compared starting 1 month after the first immunization, protection for these end points was greater than 90%. Thus systemic immunization with a subunit HPV vaccine can achieve a high degree of protection in women against benign and premalignant diseases induced by this sexually transmitted local infection of the genital mucosa or skin.

The Merck-sponsored trials have not reported on the comparative incidence of infection by other HPV types not included in the vaccines. However, the proof-of-principle monovalent HPV16 trial has reported on the total number of patients with various grades of cervical dysplasia, and whether the dysplasias were associated with HPV16, for the fully vaccinated women who were HPV negative throughout the vaccination period (32). In contrast to the complete protection seen against HPV16-associated dysplasias, there was no difference between the placebo and vaccinated groups in the number of non–HPV16-associated dysplasias in the initial report on this cohort. In the 3.5-year follow-up report (34), there was also no evidence that non–HPV16-associated dysplasias in the vaccinated group were less frequent in number than those in the nonvaccinated group. These results strongly suggest that the protection conferred by Merck HPV16 monovalent vaccine was predominantly HPV type specific.

Several issues about the vaccine remain to be addressed. It will be important to confirm that the strong safety profile remains intact as more individuals receive the vaccine. As noted above, it remains to be determined how long the high level of type-specific protection is maintained, as this issue will have implications for whether and when booster injections might be advisable and will also contribute to the cost-effectiveness of the vaccine. It remains to be seen whether the cross-protection against HPV types not in the vaccine, as noted in the GlaxoSmithKline proof-of-principle trial, will be confirmed in the large-scale trials of this vaccine and whether the Merck vaccine possesses similar properties. A substantial degree of cross-protection could increase the potential impact of the vaccine by further reducing the incidence of serious genital HPV infections and by reducing the number of abnormal Pap tests and the cost of their follow-up. On the other hand, protection against heterologous HPV types is likely to wane more quickly than that against the HPV types specifically targeted by the vaccine, which could also have implications for boosting.

Another unanswered question is whether vaccination might alter the natural history of prevalent HPV infection by reducing the incidence of persistent infection or cytological abnormalities. The large vaccine trials may have a sufficient number of prevalent infections attributable to the HPV types in the vaccine to address this question. If such a “therapeutic effect” were seen, it could provide an added rationale to vaccinate sexually active women who might have prevalent infection with one of the types in the vaccine. The most likely explanation for such an effect would be that the vaccine had reduced the efficiency of transmission of an early infection from one genital site to other genital sites, presumably via specific antibodies in the genital tract (46). It is also possible that the vaccine could have direct therapeutic effects against established lesions. However, this possibility seems less likely, as persistent infection is usually attributed to the presence of the viral genome in long-lived basal epithelial cells, which do not express L1. The experimental evidence also does not support this possibility: when the effect of the VLP vaccine was tested on established lesions in an animal model (bovine papillomavirus type 4 [BPV-4], which induced benign oral mucosal lesions), it did not induce their regression, although the vaccine was very effective when given prophylactically (31).

Although the immune response of men to the vaccine is similar to that of women (41), it is not yet known whether the vaccine will be protective in men. Many vaccines have comparable efficacy in males and females. However, a subunit vaccine for type 2 herpes simplex virus (HSV gD vaccine), another sexually transmitted viral infection, was found to be effective in women but not in men, which raises the possibility that an analogous difference might be seen with the HPV vaccine (47). As HSV infection is more likely to be mucosal in women and cutaneous in men, it was speculated that the difference in protection from the HSV vaccine might be attributable to higher antibody titers in mucosa than in skin. Even if this explanation is relevant to HSV, the high level of protection of the Merck vaccine against cutaneous genital warts in women would be expected to apply to men as well, since these warts appear on cornified skin in both sexes. It may also be relevant that the protection induced in women by the HSV vaccine was less robust than that induced by the HPV vaccine (47). Efficacy trials of the HPV vaccine in men will directly address this issue.