Abstract
The human papillomavirus (HPV) is an important causal agent of premalignant cervical epithelial changes and cervical cancers. These cancers account for ∼5% of all cancers globally and kill more than a quarter million women annually. HPV infections also associate with certain anogenital and oropharyngeal cancers. Events leading to the development of HPV vaccines to prevent associated cancers are described, with a further discussion of goals that must be met to achieve full virus eradication.
Key Background Information
• Cervical cancer accounts for about 5% of the global cancer burden, killing >250,000 women annually (9)
• 99% of cervical cancers are causally associated with infection with one of ∼10 high-risk α-papillomaviruses (PVs), from among >200 known human PVs (15)
• 70% of cervical cancers are associated with the two most common high-risk α-PV genotypes: HPV16, 18 (49)
• >50% of women and men acquire a genital infection with at least one high-risk α-PV (5)
• >95% of infections with high-risk α-PV resolve spontaneously over 1–5 years (69)
• Persisting high-risk α-PV infection is associated with premalignant cervical epithelial changes (CIN2,3), which can occur within 2 years of infection, and can persist lifelong without treatment or progress to cancer (in 30%) with a median time of 10–15 years (69)
• High-risk α-PV is associated with other anogenital and some oropharyngeal cancers (66)
A Half-Full Glass: PV Vaccines Are a Product of the Molecular Biology Revolution
Cervical cancer has been recognized to model as a sexually transmitted infection since the 1840s (58). However, demonstration that the causal agent of cervical cancer is a PV required development of the tools and technologies of molecular genetics. Unlike many viruses, PVs cannot be propagated effectively in the laboratory, as they require a maturing epithelium to enable the different stages of their reproduction (17). Production of PV genes that promote viral DNA replication occurs in dividing basal epithelial cells (52), while PV structural protein production and DNA packaging are restricted to nondividing mature epithelial cells by selective promoter use (50), and by matching of viral codon usage to the availability of corresponding tRNAs (72).
PVs were recognized as oncogenic in animals and as a cause of warts in the genital tract in humans. However, without a method for PV genotyping or production in the laboratory, viral serology was impractical. Establishment of a link between cervical cancer and PV infection thus became possible only with the development of molecular techniques for distinguishing different PVs (32), allowing the detection of a novel PV DNA in cervical cancer samples (25). DNA sequencing technology established that a group of HPVs associated with cervical cancer (known as high-risk α-PVs) were distinct from those responsible for skin and genital warts (10). Proof of the oncogenic potential of these high-risk PVs followed demonstration that two PV nonstructural proteins (E6 and E7) from high-risk PVs were together sufficient to immortalize and transform somatic cells (4), and PVs thus became the first viruses demonstrated unequivocally to be linked to cancer initiation in humans.
Recognition that PVs were oncogenic resulted in considerable interest in developing vaccines against these viruses. However, there were two key problems. First, the detection of PV infection was difficult, and thus the epidemiology and natural history of PV infections were uncertain. It was unclear whether high-risk α-PV infections were common or rare, when and how they were transmitted, and how long infections lasted. Second, the standard approaches to production of viral vaccines, using killed or live attenuated viruses, were not possible for PV, as PVs could not be propagated in vitro, and testing for protection following immunization was also not possible, as PVs could not be grown from tissue samples. New techniques for gene cloning and expression, however, gave solutions to both these challenges to PV vaccine development in the 1980s.
Studies of PV epidemiology (41) and testing for protection against infection following immunization (37) were each enabled by the development of DNA hybridization techniques that could detect high-risk PVs in cells or tissues from the anogenital tract. Epidemiological studies based on these tests established that PV infections were extremely common after the onset of sexual intercourse (53) and most infections were transient: furthermore, only persisting infection gave a risk of cervical precancer (CIN2,3) and its progression to cancer, and the cervical cancer and precancer cells also tested positive for PV DNA (10). At the same time, genetic engineering also produced PV proteins enabling serology studies, using a range of technologies (54,67). However, while the PV nonstructural proteins were not highly immunogenic following infection (3), PV structural proteins became the basis of serology once a method for producing conformationally correct proteins was developed (18), and serology using viral capsid proteins was then shown to be a useful, if relatively, insensitive indicator of past or current PV infection (22).
PV vaccine development required production of PV proteins. Initial attempts using prokaryotic systems allowed production of PV proteins (48), but the capsid proteins were not correctly conformed, and did not present viral epitopes capable of inducing virus-neutralizing antibodies. Development in the late 1980s of eukaryotic gene expression systems allowed production of correctly conformed viral capsid proteins (13). Expression of the major L1 capsid protein of the HPV16 virus, with initiation of translation from the appropriate start codon in the L1 gene open reading frame, resulted in correctly conformed L1 protein and assembly of the protein into virus-like particles (VLPs). VLPs comprised multiple copies of the L1 protein and resembled immunologically the empty shell of the PV. VLP production was achieved initially with vaccinia vectors in epithelial cells (75) and subsequently with baculovirus vectors in insect cells (39), and with transgenic yeast (35). PV VLPs induced genotype specific antibody (55), and could be shown for animal PVs to protect against viral challenge (11). Human clinical trials required scaling of the manufacturing process and production of good manufacturing practice materials free of DNA (45), a technical challenge as VLPs are able to package DNA (73). Development of protocols for testing for protection against human infection enabled clinical trials of VLP based PV vaccines. These were initially small scale for safety and immunogenicity (26), but became of necessity in large scale, when efficacy proof of protection against precancer was required, because of the low risk of developing persisting infection and of its progression to precancer.
Following pivotal clinical trials showing near 100% efficacy at prevention of not only PV infection but also of PV-associated cervical precancer (57), two PV vaccines were licensed for use worldwide from 2007. Cervarix™ includes HPV16 and HPV18 VLPs, while Gardasil™ also incorporates HPV6 and HPV11, responsible for genital warts. More recently, Gardasil-9 has been developed and incorporates a further five high-risk α-PV (HPV31, 33, 45, 52, and 58) (51). Over 50 million young women, generally between the ages of 10 and 15, have been immunized (14), and, increasingly, vaccine is also being delivered to young men (12). Data from Australia, one of the first countries to introduce universal immunization of female children, have shown a dramatic reduction in new HPV infections and a corresponding reduction in associated cervical premalignancy, particularly in women immunized before the age of 14 (24). Data from the developing world have shown that where vaccine is provided and there is strong government support, near universal immunization can also be achieved (23). Thus, in principle, as infections with high-risk PVs are protected against with near 100% efficacy by immunization, and as the virus is genetically stable and there is no nonhuman reservoir of infection, eradication of high-risk PV infection should be achievable. The remainder of this article is therefore given over to outlining the challenges to achieving this goal.
A Half-Empty Glass: The Known and Unknown Unknowns
Basic science
What are the genetic/stochastic determinants of persisting infection with high-risk HPVs?
Infection with high-risk PVs is near universal. Behavioral and environmental contributions to risk of persisting infection and cancer, given HPV infection, including other infections, a smoking habit, sexual behavior, and use of oral contraceptives, appear to be of minimal impact (40). Immune suppression increases risk of progression of PV infection to cancer, and there is increasing evidence that immune response to the virus is under genetic control and progression of HPV infection to malignancy is associated with particular variants of the immune response genes, MHC 1 and MHC 2 (43). These observations are consistent with a familial tendency to develop cervical cancer (34). While immunization strategies are unlikely to be based on prescreening for risk, identification of higher risk individuals would enable more selective ongoing monitoring for the development of persisting infection and cancer.
Can we make a better/cheaper prophylactic HPV vaccine using alternate technologies?
VLP-based vaccine technologies have become an integral part of viral vaccine development. One citation on VLP technology for vaccine development in PubMed in 1990 (33) has increased to over 150 per year in 2016. However, VLPs are neither cheap nor simple to mass produce, and vaccine cost of goods is still too high for universal introduction of PV immunization programs in the developing world (63). Simpler PV vaccines based on recombinant bacterial proteins (56) or on the protein products of other PV genes (36) are being explored, and some have recently reached clinical trials (
Can the blocks to effective immunotherapy for existing HPV infection be overcome?
The pivotal clinical trials of the current HPV vaccines showed clearly that even women with asymptomatic existing HPV infection received no protection against subsequent development of cervical precancer from available PV prophylactic vaccines (1). Many trials of immunotherapy for existing HPV infection have been undertaken with only limited success and no licensed products to date (70), despite induction of the host-protective cell mediated immunity believed to be necessary. There is evidence that HPV-associated gene products can induce local innate and adaptive immune suppression mechanisms, which may explain the limited success (62). Better understanding of these mechanisms might lead to an immunotherapeutic that could be given on a population basis to reduce risk of future cancer development for those already infected with a high-risk HPV.
Public health
What most hinders HPV vaccine introduction in the developing world?
Many pilot programs for HPV vaccination have been successfully conducted in the developing world (28) and have shown that with provision of vaccine, universal immunization can be achieved with most efficacy through school-based programs (23). However, funded vaccine programs need support from stable government policy, new infrastructure to deliver vaccine to teenagers, education programs for students, parents, and government, and reliable local information on safety and field efficacy. Further research to better understand the practical limitations to vaccine delivery would likely facilitate universal immunization.
Is a “one dose, all ages, one time” epidemic vaccination policy appropriate/effective?
Data from the pivotal PV vaccine trials show that two and three dose regimes of PV vaccine delivery are effective at protecting against HPV infection and subsequent development of cervical premalignancy, and recent analysis of the data from these pivotal studies suggests that a single immunization may protect for up to 5 years against acquisition of a new infection (42). If this can be confirmed, an approach to vaccination more usually used in epidemic viral infection (20), employing universal single-dose immunization of at-risk individuals of all ages across communities or countries, might become feasible as a means to eliminate HPV infection from a community. This would overcome many of the logistic blocks to HPV eradication, as well as substantially reducing the cost.
What percentage of the community needs to be immunized to eradicate HPV infection permanently from a community?
The efficacy of the Australian HPV immunization program at reducing cervical abnormalities (30), when only a fraction of girls were immunized, suggests that the infectivity ratio for HPV propagation is low. Better defining this ratio would enable practical decisions about whether there was a need to immunize boys as well as girls to achieve general protection against HPV-associated cancers in countries where overall efficacy of delivery is lower than can be achieved in the developed world.
Clinical
How, when, and where are nongenital oncogenic HPV infections acquired?
A significant part of the burden of HPV-associated cancer is nongenital, particularly oropharyngeal cancer (19). There is evidence of vertical (intrapartum) infection with low-risk anogenital PV infections (60), which can result in recurrent respiratory papillomatosis, and the possibility therefore exists that oropharyngeal cancer may arise not only from sexual transmission but also from intrapartum delivery of high-risk α-HPVs. This would, if proven, have significant implication for public health strategies to prevent this cancer as school-based programs might be insufficient.
How long does vaccination give protection?
Most global vaccine programs are delivered to preschool infants, and there would be considerable advantage for HPV vaccine programs if incorporation of HPV vaccines into existing childhood immunization schedules could be justified. However, as exposure to PV infection is not likely in the first 10 years of life, and could occur over many years of adult life, it would be necessary to establish how long protection might last, not only with the current immunization schedules but also with single- or two-dose schedule, delivered to preschool infants.
How widespread is cross-protection across HPV genotypes?
High-risk α-PVs include at least 10 viral genotypes, which are largely serologically distinct (31). However, from the studies to date, some degree of cross-protection across genotypes is observed for the two- and four-valent HPV vaccines (21), each of which incorporates the two high-risk α-PVs most commonly associated with cervical cancer. The newer nine-valent vaccine incorporates seven high-risk α-PVs and has demonstrated broader protection against CIN2, 3 in studies to date (46). However, the duration and breadth of cross-protection remain to be established, and will become an issue of increasing significance as the more common HPV types become scarcer with immunization.
Vaccine development and testing
Testing in vitro—what is the right assay?
Serological assays were initially developed using recombinant capsid protein from the virus as antigen, in standard ELISA assays (59). However, it was realized that there was a fair amount of denatured protein in recombinant capsid protein preparations, and antibody to denatured protein was unlikely to be virus neutralizing. Several approaches were adopted to overcome this: a switch to Elisa capture assays based on neutralizing monoclonal antibodies of defined epitope specificity allowed removal of background signals from denatured protein. Displacement by immune serum of binding of monoclonal antibody to VLPs allowed epitope-specific antibody quantitation (61). A pseudovirion neutralizing assay was developed, which measured prevention of infection of a susceptible cell line by a viral capsid packaging a reporter gene construct (68). Evaluation of these three assays in parallel by multiple laboratories worldwide allowed the conclusion that, by and large, immune sera from infected subjects and immune sera from immunized subjects demonstrated similar potency for binding or neutralization in each of these assays. This comparison was also used to designate a WHO reference serum for HPV16 (27). As with all virus antibody assays, no particular titer or antibody specificity in blood or at the mucosal site of infection has been demonstrated to be indicative of host protection against infection following immunization. While T cell assays to viral capsid proteins have been developed, there has been no systematic exploration of the relevance of T cell responses to HPV antigens to host protective immunity following immunization.
Testing new vaccines in vivo—are animal PV infection models informative?
Proof of concept of the likely efficacy of vaccines against PV was obtained by assessing protection against bovine PV (BPV) challenge in cattle using BPV vaccines (37), against canine oral PV (COPV) challenge in dogs using vaccines based on formalin fixed COPV (6), and against Cottontail Rabbit PV (CRPV) using CRPV VLPs (44). Subsequently, a mouse challenge model has been established for a murine PV MmuPV (38). Each of these has given proof of concept that vaccines against PV might work, but no animal or in vitro model using authentic human PV has been established, a consequence of the twin problems of HPV production in vitro and HPV species specificity.
Choice of virus types—are genotypes and serotypes identical
PVs are classified by gene sequence, using various sets of criteria to determine how different a virus DNA sequence needed to be to count as a distinct genotype (7), and this classification has recently been updated. (65) The α-PV group is largely tropic for genital skin, includes sublineages within which the majority of viruses associated with anogenital and oropharyngeal malignancy lie, and includes separate sublineages that are associated with genital warts (64). β PV produces skin warts and may be initiators of squamous skin cancer, particularly in immunosuppressed patients (2,29). The extent to which PV genotypes are also serotypes was clarified by definition of the crystal structure of the virus and the deduction that the sites to which neutralizing monoclonal antibodies bound were on the outer face of the L1 protein, where the maximum sequence variability lay (8). Some cross-protection has been observed following development and testing of vaccines containing multiple genotypes, and is most obvious between the most closely sequence-related genotypes. The degree of cross-reactivity has been confirmed from the serotype-specific protection offered by the two-, four-, and nine-valent vaccines in the pivotal clinical trials of the vaccines. Minor sequence variations that define subtypes of HPV16 are well described, but do not appear to define significant serological differences (16).
Choice of expression system
The initial development of VLPs was undertaken using vaccinia (75) and then baculovirus (39) expression vectors, and it was noted that there were posttranslational modifications, particularly glycosylation (74), although these did not seem to be particularly important either for virus assembly or for induction of neutralizing antibody. Subsequent production of the vaccine in yeast-based systems (71) and baculovirus/insect cell expression systems appears to produce VLPs of similar immunogenicity and potential for generating virus-neutralizing antibody in animals and humans.
Regulatory requirements
Discussions between the vaccine development companies and the clinical and research experts in the field of cervical cancer, and the regulatory authorities were held under the auspices of the WHO over the period of preclinical and early clinical vaccine development, and led to a consensus opinion that the endpoint of the pivotal vaccine studies of vaccines targeting HPV16 and HPV18 should be more stringent than protection against HPV infection, and should include protection against and reduction in HPV type-specific premalignancy (CIN 3) (47). More recent studies of multivalent vaccines could not be sufficiently powered to reach this endpoint for each available vaccine phenotype, as the proportion of CIN2,3 caused by these additional types is small (10). Composite endpoints of overall protection against disease and protection against infection with individual genotypes have since been used to validate the nine-valent vaccine for clinical use (51).
Conclusion
HPV vaccination programs have supported a dramatic reduction in new HPV infections and a corresponding reduction in associated cervical premalignancy. Now that infections with high-risk PVs can be protected against with near 100% efficacy by immunization, eradication of these infections should be achievable. Improvements in numerous aspects of vaccine development (e.g., basic research, vaccine design, vaccine dosing, vaccine distribution, response durability, immunotherapy, and regulatory requirements) may be considered to aid prospects for full eradication of high-risk PV.
Footnotes
Author Disclosure Statement
The University of Queensland receives royalty payments from the sale of HPV vaccines, from which the author receives a share.
