Human papillomavirus infection and the primary and secondary prevention of cervical cancer§

Authors

  • Douglas R. Lowy MD,

    Corresponding author
    1. Laboratory of Cellular Oncology, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, Maryland
    • Laboratory of Cellular Oncology, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Building 37, Room 4106, MSC 4263, Bethesda, MD 20892
    Search for more papers by this author
    • Fax: (301) 480-5322

    • Drs. Schiller and Lowy are named inventors on US government-owned human papillomavirus vaccine patents that are licensed to Merck and GlaxoSmithKline and are entitled to limited royalties as specified by federal law.

  • Diane Solomon MD,

    1. Division of Cancer Prevention, National Cancer Institute, National Institutes of Health, Bethesda, Maryland
    Search for more papers by this author
  • Allan Hildesheim PhD,

    1. Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Bethesda, Maryland
    Search for more papers by this author
  • John T. Schiller PhD,

    1. Laboratory of Cellular Oncology, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, Maryland
    Search for more papers by this author
    • Drs. Schiller and Lowy are named inventors on US government-owned human papillomavirus vaccine patents that are licensed to Merck and GlaxoSmithKline and are entitled to limited royalties as specified by federal law.

  • Mark Schiffman MD, MPH

    1. Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Bethesda, Maryland
    Search for more papers by this author

  • GlaxoSmithKline provided vaccine for the trial under a Clinical Trials Agreement with the National Cancer Institute.

  • All of the coauthors are investigators in a phase 3 clinical trial to evaluate the safety and efficacy of a human papillomavirus type 16/18 bivalnent vaccine.

  • §

    This article is a US Government work and, as such, is in the public domain in the United States of America.

Abstract

A wealth of evidence has led to the conclusion that virtually all cases of cervical cancer are attributable to persistent infection by a subset of human papillomavirus (HPV) types, especially HPV type 16 (HPV-16) and HPV-18. These HPV types also cause a proportion of other cancers, including vulvar, vaginal, anal, penile, and oropharyngeal cancers. Although cervical cancer screening, primarily with the Papanicolaou (Pap) smear, has reduced the incidence of this cancer in industrialized countries, cervical cancer remains the second most common cause of death from cancer in women worldwide, because the developing world has lacked the resources for widespread, high-quality screening. In addition to advances in Pap smear technology, the identification of HPV as the etiologic agent has produced 2 recent advances that may have a major impact on approaches to reduce the incidence of this disease. The first is the development of a preventive vaccine, the current versions of which appear to prevent close to 100% of persistent genital infection and disease caused by HPV-16 and HPV-18; future second-generation vaccines may be able to protect against oncogenic infections by a broader array of HPV types. The second is the incorporation of HPV testing into screening programs. In women aged >30 years, HPV testing can identify high-grade cervical intraepithelial neoplasia earlier than Pap smears with acceptable rates of specificity. These results, together with the high sensitivity of HPV testing, suggest that such testing could permit increased intervals for screening. An inexpensive HPV test in development, if successful, may be incorporated as part of an economically viable ‘screen-and-treat’ approach in the developing world. The manner in which vaccination and screening programs are integrated will need to be considered carefully so that they are efficient in reducing theoverall incidence of cervical cancer. Cancer 2008;113(7 suppl):1980–93. Published 2008 by the American Cancer Society.

The past 6 decades have witnessed a dramatic change in the approach to cervical cancer and the understanding of its pathogenesis. The emphasis has switched to the prevention of invasive cancer before it occurs. The introduction of the Papanicolaou (Pap) smear in the 1940s and its validation permitted the identification of easily treated cervical intraepithelial neoplasia (CIN) and early cervical cancer, initiating a transition toward giving progressively greater emphasis to the prevention of this cancer. Beginning in the 1950s, widespread Pap smear implementation led to reductions in the incidence of cervical cancer in communities with the resources for high-quality screening programs. However, the incidence of this cancer has remained high in the developing world, which lacks the resources for widespread screening programs as practiced in nations with greater resources. A second achievement was the recognition, from observations initiated in the 1970s that reached their fruition in the 1980s and early 1990s, that human papillomaviruses (HPVs) were linked etiologically to cervical cancer. This major advance in our understanding of the etiology and pathogenesis of cervical cancer also led to 2 important clinical advances: a preventive HPV vaccine for the primary prevention of cervical cancer and HPV assays to improve secondary prevention (screening programs). The availability of these etiology-based interventions for the primary and secondary prevention of cervical cancer has provided an opportunity for even greater and more efficient reductions in the incidence of this cancer in settings with established secondary screening programs and may offer the possibility of bringing cost-effective cervical cancer prevention strategies to the developing world. There is an urgent need for such interventions, because cervical cancer remains the second most common cause of death from cancer among women worldwide, accounting for >250,000 deaths each year.1

This review is divided into 3 parts. First, we summarize the current understanding of cervical cancer pathogenesis; then, we consider the use of vaccination and other approaches aimed at the primary prevention of cervical cancer. Finally, we discuss strategies for secondary prevention in vaccinated and unvaccinated populations.

HPV INFECTION AND CERVICAL CANCER PATHOGENESIS

HPV infection and disease

HPVs are a group of more than 150 related DNA viruses that infect cutaneous and mucosal epithelia, in which acute infection causes benign cutaneous lesions, such as nongenital and genital warts, flat cervical condylomas, or low-grade CIN.2 HPV genomes are quite stable genetically compared with the relatively high mutation rate of many RNA viruses.3 A subset of approximately 15 HPVs that infect the genital tract have the potential to cause malignant tumors, most commonly in the cervix.4, 5 The cancer-associated HPV types are designated high-risk (or oncogenic) types, whereas the HPV types that are not associated with cervical cancer are designated low-risk types. Two closely related low-risk types (HPV-6 and HPV-11) cause most cases of genital warts (condyloma acuminatum), but many other low-risk types may cause virtually no pathology. The high-risk types are related phylogenetically,5 and HPV-associated carcinomas arise as the result of long-term, persistent infection with high-risk types. HPV DNA is found almost universally in primary cervical tumors (regardless of histology) and their metastases. The HPV types associated with cervical cancer are similar throughout the world, although minor regional differences in frequency have been noted.6 HPV-16 is the type identified most frequently in all regions, accounting for approximately 50% of all cervical cancers. In most regions, HPV-18 is the next most common type and typically is found in from 15% to 20% of squamous cell cancers and in a greater proportion of adenocarcinomas. HPVs also are implicated in the development of a variable proportion of vulvar, vaginal, anal, penile, and oropharyngeal cancers, with HPV-16 accounting for the vast majority of the HPV-associated tumors at these sites.1 Worldwide, cervical cancer represents approximately 80% of the cancers attributable to genital-mucosal HPV infection. However, in countries with effective cervical cancer screening programs, the noncervical cancers, which tend not to be subject to widespread screening, represent a higher proportion of these cancers.

Genital HPV infection is believed to be the most common sexually transmitted infection.7 Young women are at particularly high risk of acquiring HPV soon after initiating sexual activity. Women in their early 20s have point prevalence rates on cross-sectional screening that range from 20% to 40%, with a roughly equal distribution between high-risk and low-risk HPVs.8 The cumulative incidence depends on the frequency of sampling. One longitudinal study with semiannual visits demonstrated that sexually active women ages 15 to 19 years had a 3-year cumulative incidence >40%.9 The cumulative lifetime risk of infection cannot be estimated accurately but is most likely ≥75% for 1 or more genital HPV infections. The great majority of these infections are self-limited or controlled by the immune system, and the prevalence of HPV infection among women aged >30 years is substantially lower than among women soon after the average age of first sexual intercourse,10 although age-specific prevalence at older ages varies for unclear reasons. Clearance of infection is believed to be immune-mediated and largely type-specific, as evidenced by the association between degree of immunosuppression and rates of infection among immunocompromised individuals, altered immune profiles of women with persistent infections, and the independent clearance of specific HPV types among women with ≥2 HPV infections.11, 12 Neutralizing antibodies that develop in response to infection are type-restricted, with limited evidence of cross-reaction observed for some closely related HPV types.13, 14

The importance of persistent infection and viral oncogenes

Women who, instead of clearing their infection, become persistently infected with a high-risk HPV type are at increased risk of developing cervical cancer.15 The risk of progression to high-grade CIN and invasive cancer is greater for women who are infected with HPV-16 and HPV-18 than for women who are infected with other high-risk types (Fig. 1).16 Certain variants of HPV-16 and HPV-18 may be associated with a different risk of progression to high-grade CIN or cancer.17 Among HPV-infected women, some etiologic cofactors may be associated with an increased risk of persistent infection and/or progression to high-grade CIN or invasive cancer. The most likely cofactors include cigarette smoking, multiparity, long-term hormonal contraceptive use, and human immunodeficiency virus infection or other causes of long-term immunosuppression.11, 18–20

Figure 1.

Cumulative incidence of cervical cancer/precancer in women over 30 during a 10-year period. Women with normal cytology were tested once, at enrollment, for HPV16, HPV18, and Hybrid Capture II (HC2, a cocktail of multiple high-risk HPV types, including HPV16 and HPV18). Each woman was classified as being positive for HPV16 (HPV16+), HPV18 (HPV18+), HC2 positive but negative for HPV16 or HPV18 (HC2+), or negative for HPV (HPV-), and followed prospectively for 10 years.

Characteristic features of most cervical cancers that contain HPV-16 and nearly all cervical cancers with HPV-18 DNA are 1) the viral DNA is integrated in the host genome; and 2) usually, only the viral E6 and E7 genes are expressed, and the other viral genes are deleted or mutated.21, 22 (Tumors with unintegrated viral DNA are discussed below.) These findings suggest that E6 and E7 are important viral oncogenes, and a variety of experimental findings have validated this possibility. For example, HPV-16 E6 and E7 together can induce cervical cancer in transgenic mice, and their tumorigenic activity is substantially greater when both genes are expressed together compared with the expression of 1 gene alone.23 The E6 and E7 genes of high-risk HPVs also possess in vitro biologic properties and associated biochemical activities that are either lacking or less prominent in the E6 and E7 genes of low-risk HPVs. When they are expressed together, high-risk E6 and E7 cooperate to immortalize primary human keratinocytes. E6 and E7 expression also appears to be necessary for maintenance of the transformed phenotype, because suppression of their expression in cervical cancer cell lines leads to growth arrest or apoptosis.21, 22 High-risk E6 and E7 proteins each encode multiple biochemical activities; key features include the ability of E6 to inactivate the p53 tumor suppressor protein and the ability of E7 to inactivate the retinoblastoma tumor suppressor protein pRb.24 A characteristic cellular response to pRb inactivation is expression of the p16 tumor suppressor gene, which, thus, often is detected in benign and malignant lesions.25 The biologic importance of p53 and pRb inactivation is underscored by the observations that E6 and E7 mutants lacking these activities are deficient in their ability to contribute to keratinocyte immortalization, as are low-risk E6 and E7. Thus, E6 and E7 are important determinants of the oncogenicity of high-risk HPVs. Currently, it is not clear whether differences in E6/E7 properties among high-risk HPV types exist or whether they explain the predominance of HPV-16/HPV-18 in cancers.

Cancer only develops after many years of persistent infection with a high-risk type.26 Low-grade cytopathic changes (atypical squamous cells of undetermined significance [ASC-US] or low-grade squamous intraepithelial lesion [LSIL] cytology, CIN-1 histology) often develop shortly after infection.27 Of these minor abnormalities, LSIL cytology is the most reproducible sign of HPV28; both ASC-US cytology and CIN-1 histology are reproduced poorly in interpathologist studies.29 Whether or not minor lesions occur, acute HPV infection has a high likelihood of regressing without intervention.12 High-grade CIN (most clearly represented as CIN-3, because CIN-2 may represent a mixture of acute infection and proto-CIN-3) typically is detected from 5 to 15 years after infection, depending on the intensity and sensitivity of screening.15 Progression of CIN-3 to invasive cancer generally takes many years or even decades. The interval from first infection to high-grade CIN is usually less than the time from high-grade CIN to cancer.26 Overall, the long interval between first infection and cancer implies that, although the virus may initiate the chain of events that lead to cancer, a series of cellular alterations in the target cell and/or stroma that collaborate with the virus occurs during the process of carcinogenesis. Consistent with this hypothesis, many genetic and epigenetic changes have been observed in tumors.22, 30–32 These changes in the tumors, which are a consequence of long-term infection and/or exposure to exogenous cofactors, include the activation of oncogenes and antiapoptotic genes and the inactivation of tumor suppressors, proapoptotic genes, and genes implicated in antigen processing and presentation.

The coexpression of HPV-16 E6 and E7 induces genomic instability, which is believed to contribute to aneuploidy and viral DNA integration.31 For HPV-16–associated lesions, which have been examined in the greatest detail, aneuploidy is detected very uncommonly in low-grade CIN but may be present in approximately 33% of high-grade CIN and in the vast majority of invasive cancers. The frequency of viral DNA integration follows a similar pattern; however, in high-grade CIN, integration in 1 study was only approximately 50% as frequent as aneuploidy, whereas the difference in frequency was not observed for cancers.33 It appears likely that aneuploidy usually precedes viral DNA integration, because nearly all lesions with integrated viral DNA are aneuploid, whereas a proportion of aneuploid lesions lack detectable integrated viral DNA.

Although the integration of viral DNA is detected in the majority of tumors associated with HPV-16, HPV-18, and HPV-45 (which is very closely related phylogenetically to HPV-18), integration is not an obligatory step for the development of invasive cancer.34 This point recently was emphasized by using a sensitive molecular assay to demonstrate that the majority of tumors associated with HPV-31 and HPV-33 did not contain detectable, integrated viral DNA.35 The tumors associated with the latter HPV types also appeared to develop more slowly. Integrated viral DNA reportedly is associated with increased expression of E6 and E7, which may further enhance the degree of genomic instability and foster a faster rate of cellular alterations.36

It is unclear which viral properties may account for differences in oncogenicity observed between subtypes and variants of a given high-risk HPV type or between the different high-risk HPV types, including the degree to which the differences may be attributable to biologic activities of the virus or to host response. Obtaining insight into these issues would be highly worthwhile given the medical importance of HPV as a carcinogen and the opportunity to correlate experimental analyses with ongoing prospective natural history studies of genital infection by various HPVs. Because sequence divergence between HPV types is distributed throughout their genomes, it appears likely that the particular oncogenic properties of a given virus could be attributable to quantitative differences in several genes.

Primary Prevention

Introduction

Until very recently, cervical cancer prevention has involved mainly secondary prevention, specifically, screening based on the Pap test. However, the recognition that virtually all cases of cervical cancer are attributable to HPV infection implies that primary prevention of a high proportion of the infections that lead to cervical cancer could represent a powerful complementary approach to reducing the incidence of this cancer and other cancers attributable to HPV infection.

The current high incidence of genital HPV infection suggests that traditional approaches to reduce infection rates have had limited efficacy. One reason is that condoms, as used by most individuals, afford limited protection against HPV, in contrast to their relatively high protection against pregnancy and efficacy against several other sexually transmitted infections.37–39 There is evidence that circumcision for men may reduce the incidence of infection among sexual partners.40, 41 Nevertheless, the incidence of HPV infection among circumcised men and their partners remains high. Reducing the number of sexual partners also could reduce incidence, although the high frequency of HPV exposure among sexually experienced individuals and the apparent high transmissibility of infection suggests that other approaches to primary prevention are needed.

HPV vaccination: Theoretical considerations

In this regard, approaches that are directed more specifically toward preventing HPV infection could have a more substantial impact on infection rates and subsequent cancer. Theoretically, the long interval between infection and cancer development implies that a therapeutic vaccine could have high utility. However, it has proven easier to develop effective preventive vaccines, as indicated by the finding that most approved vaccines against other infections are preventive rather than therapeutic. Furthermore, preventive vaccination historically has been identified as a cost-effective approach that can reduce dramatically the incidence of many infections.42

Preventive vaccines may be composed of virions (virus particles) that have been inactivated chemically to render them noninfectious, as with the Salk poliovirus vaccine; attenuated live viruses, such as the Sabin poliovirus vaccine; or a subunit vaccine that is not infectious because it lacks some components required for infection, such as the hepatitis B virus vaccine.43, 44 Such subunit vaccines can be delivered directly or by using live vectors, with direct delivery perhaps posing fewer theoretical regulatory issues. Currently, the subunit approach is the most appropriate for an HPV vaccine, because the presence of HPV oncogenes, the difficulty of making preparative amounts of authentic HPV, and the lack of an animal model for testing the pathogenic activity of an HPV (because of its species specificity) make other approaches less suitable.

Preventive HPV vaccine

It is believed that neutralizing antibodies, the class of antibodies that inhibit infection by binding to viral proteins in infectious virions, are the main protective activity induced by preventive vaccines.45, 46 Such antibodies typically are induced by virion proteins, of which papillomaviruses have 2, L1 and L2, which are designated the major and minor structural proteins, respectively, because of their relative abundance in the virion. Both L1 and L2 can induce neutralizing antibodies, with levels of antibodies induced by L1 being substantially higher; the current commercial subunit HPV vaccines are composed of L1.47 These noninfectious, protein-based vaccines are based on the preclinical observations that multiple copies of the L1 protein will self-assemble into virus-like particles (VLPs) that induce high levels of neutralizing antibodies and are highly protective in animal papillomavirus models.48 Passive transfer of immune immunoglobulin G can confer protection to naive animals, implying that neutralizing antibodies are most likely the main protective activity induced by the vaccine. The particulate nature of the immunogen appears to be important, because the neutralizing antibodies are directed against L1 epitopes that have conformational dependence; denaturation of VLPs leads to an L1 immunogen that does not induce neutralizing antibodies and was ineffective as a vaccine in a preclinical model.48 The repetitive structure of the VLPs most likely contributes to their high immunogenicity.49 Because the neutralizing antibodies induced by VLPs are predominantly type-specific,13, 50 commercial versions of the vaccine contain VLPs from more than 1 HPV type.

There are 2 commercial versions of the vaccine. One, which is produced by GlaxoSmithKline (GSK; Cervarix), is a bivalent vaccine composed of L1 VLPs from HPV-16 and HPV-18.51 These 2 HPV types are responsible for approximately 70% of cervical cancer. The other vaccine, which is produced by Merck (Gardasil), is a quadrivalent vaccine. It contains HPV-16 and HPV-18 VLPs and also has VLPs from HPV-6 and HPV-11, which, together, cause approximately 90% of genital warts.52, 53 Both vaccines are administered in 3 intramuscular doses given over a 6-month period. The Merck vaccine is produced in yeast and uses a simple aluminum salt as an adjuvant. The GSK vaccine is produced in insect cells and uses a proprietary adjuvant, AS04, which contains an aluminum salt and monophosphoryl lipid A. Both vaccines induce seroconversion in >99.5% of vaccinees.

Vaccine efficacy

Published clinical efficacy trials of the 2 vaccines have been performed in women ages 15 to 26 years (Table 1). In controlled trials, both vaccines demonstrated a high level of protection against incident persistent infection and disease caused by the HPV types in the respective vaccine (vaccine types), and both had a good safety record. In fully vaccinated women, according to a protocol analysis in which women who were positive for a given HPV type at enrollment or during the vaccination period were excluded from analysis for that type, protection was close to 100% against incident endpoints involving vaccine types. For the Merck vaccine, in their interim analysis of phase 3 trials, such protection was demonstrated for HPV-16– and HPV-18–associated cases of moderate- and high-grade CIN and moderate- and high-grade vulvar and vaginal dysplasia as well as for external genital warts associated with any of the 4 vaccine types.53, 54 The majority of the genital warts, as expected, were associated with HPV-6 and HPV-11. A similar degree of protection against incident persistent infection or high-grade CIN attributable to HPV-16 or HPV-18 infection was demonstrated for the GSK vaccine in fully vaccinated women.55 The interim analysis of the GSK phase 3 trials was conducted as a modified intention to treat (MITT) protocol in which women who were positive for a given HPV type at enrollment were excluded from analysis for that type, but they were not excluded if they became positive during the vaccination period.51 The MITT analysis produced somewhat lower protection rates, which may reflect decreased protection during the vaccination period rather than an actual reduction in vaccine potency. Phase 2 trials have demonstrated that prophylactic protection from the vaccines lasts at least 5 years.55, 56 To our knowledge to date, the women in the phase 3 trials, which, of necessity, started after the phase 2 trials, have been followed for a shorter period.

Table 1. Prophylactic Efficacy of Virus-like Particle Vaccines Against Vaccine Targeted Human Papillomavirus Types
Outcome (ATP or MITT)Test VaccineReferenceControlsVaccineesEfficacy [95% CI]
  • ATP indicates according to protocol (fully vaccinated women who were negative for a given vaccine type at enrollment and throughout the vaccination period); MITT, modified intention to treat (women who were negative for a given vaccine type at enrollment); CI, confidence interval; FUTURE, Females United to Unilaterally Reduce Endo/Ectocervical Disease; CIN, cervical intraepithelial neoplasia; HPV, human papillomavirus; GSK, GlaxoSmithKline.

  • *

    GSK vaccine: bivalent HPV types 16 and 18.

  • Merck vaccine: quadrivalent HPV types 6, 11, 16, and 18.

  • With regard to MITT, the number of months indicates the time between positive HPV tests required for the infection to be defined as persistent.

CIN-2+
 MITTGSK*Harper 2006555/4700/481100 (−7-100)
 MITTGSKPaavonen 20075121/78382/778890 (53-99)
 ATPMerckGarland 20075244/22580/2241 
 ATPMerckFUTURE II Study Group5842/52601/530598 (86-100)
CIN-1+
 MITTGSKHarper 2006558/4700/481100 (42-100)
 MITTGSKPaavonen 20075128/78383/778889 (59-99)
 ATPMerckGarland 20075265/22580/2241100 (94-100)
Persistent HPV DNA
 ATP, 12 moGSKHarper 2006559/3850/414100 (61-100)
 MITT, 12 moGSKHarper 20065516/4701/48194 (78-99)
 MITT, 12 moGSKPaavonen 20075146/343711/338676 (48-90)
 ATP, 4 moMerckVilla 20065645/2332/23596 (83-100)
 MITT, 4 moMerckVilla 20065648/2544/25694 (83-98)
External genital warts
 ATPMerckGarland 20075248/22790/2261100 (92-100)

Despite their efficacy in preventing incident infection and disease, the vaccines do not appear to influence the rate of clearance of prevalent HPV-16 or HPV-18 infections and/or CIN.57, 58 This observation is consistent with preclinical animal papillomavirus studies in which the VLP vaccine did not induce regression of established lesions.59 Therefore, it is not surprising that, when the Merck vaccine was been analyzed by an intention to treat protocol, which includes a mixture of prevalent and incident infections, because women who were positive for a given HPV type at enrollment were not excluded from analysis for that type, its level of protection was much lower than when only incident infections were counted.52, 58

In addition to protecting against incident infection and disease caused by the vaccine types, limited cross-protection against other HPVs has been observed, as published in the GSK phase 2 and phase 3 trials51, 55 and as reported at meetings for the Merck vaccine. In the published studies, cross-protection has been limited to those HPV types that are related most closely to HPV-16 and HPV-18. These results indicate that protection is type-restricted rather than strictly type-specific. The cross-protection apparently increases the overall level of protection against high-grade CIN associated with any HPV type, but approximately 25% of such lesions may not be prevented by the vaccine. In vitro assays indicate that neutralization titers against a given vaccine type is at least 10-fold higher than against even a closely related HPV type, raising the possibility that the duration of protection against nonvaccine HPV types might wane sooner than that against vaccine types. Therefore, it is noteworthy that the published phase 2 trial data indicate that the cross-protection lasts several years.55 This observation raises the possibility that protection against the vaccine types could last substantially longer if it is assumed that the in vitro neutralization results are clinically relevant.

HPV vaccine: Regulatory aspects and recommendations

The Merck vaccine was approved in the US and the European Union in 2006 and in many other countries. The GSK vaccine was approved in the European Union in 2007 and in several other countries. GSK also has applied for licensure in the US, and a decision is anticipated in 2008. In the US and in many other countries, approval (of the Merck vaccine) has been limited to young women ages 9 to 26 years. Published efficacy studies have been limited to women ages 16 to 26 years, as noted earlier. Immunologic bridging studies to adolescent girls ages 9 to 15 years, which showed an even more robust immune response than that among the young women in the efficacy trials, were accepted by the US Food and Drug Administration (FDA) as implying that they also would be protected. Two relevant considerations were the widespread recognition that it would be extremely difficult to perform efficacy trials in young adolescents and data from the US and many other countries indicating that young adolescents should be the main target group for the vaccine. That is because the vaccine should be most cost-effective (ie, prevent the most cases of high-grade CIN and cancer for a given public expenditure) if it is given before women become sexually active; and US behavioral surveys indicate that approximately 25% of young women aged 15 years have been sexually active, with this proportion increasing to approximately 70% of young women aged 18 years.

The European Union approved the Merck vaccine for young men ages 9 to 15 years in addition to young women ages 9 to 26 years based on their acceptance of immunologic bridging studies of young men ages 9 to 15 year, which produced results that were virtually identical to the results produced among young women ages 9 to 15 years. The FDA presumably will consider approval for vaccination of men only after efficacy has been demonstrated in them, perhaps because it was demonstrated that an experimental herpes simplex virus subunit vaccine conferred partial protection in women, but none in men.60 This sex-specific response raised the possibility that prophylactic vaccination against mucosal genital infections may have reduced efficacy among men. Efficacy trials of the HPV vaccine in men have been initiated.

In the US, the Advisory Committee on Immunization Practices (ACIP) of the Centers for Disease Control and Prevention makes national recommendations for approved vaccines, although implementation is at the state and local level. The ACIP recommended routine vaccination of girls aged 11 and 12 years as the main target group for the vaccine with catch-up vaccination for girls and women ages 13 to 26 years, and they recommended vaccination of girls ages 9 and 10 years at the discretion of the medical personnel involved. However, the recommendation for catch-up vaccination for women aged ≥19 years has not been embraced universally.61

The ACIP also recommended that the federal government purchase vaccine through its Vaccine for Children (VFC) program, which now can provide the HPV vaccine for girls aged ≤18 years who come from poor families. This aspect of vaccine implementation may be particularly important for public health, because women from this socioeconomic background tend to have less access to screening when they are older and, thus, are at greater risk of developing cervical cancer.62 Traditional vaccines that are part of the VFC program have achieved broad coverage for eligible children, especially when the vaccines are required.63

Because the vaccine may not prevent at least 25% of infections that may lead to cervical cancer, at least for now, it is recommended that vaccinated women should continue to follow the same cervical cancer screening guidelines as nonvaccinated women (for information on integrating screening and vaccination, see below). The vaccine is in the process of being made widely available in the industrialized world. However, its current high cost ($120 per dose) most likely means that it will not be implemented widely in the developing world in the near future, even with tiered pricing.

Implementation may be divided into vaccination of the target group and catch-up vaccination for girls and young women aged ≤26 years. The number of cases of cervical cancer prevented will be greater if the vaccine is administered before women become active sexually given the high risk for infection soon after sexual activity is initiated and the lack of vaccine efficacy against established infection. An instructive theoretical model has been developed for girls in Finland,64 although a somewhat lower percentage of Finnish girls aged ≤18 years are sexually active (approximately 10% of those aged 15 years in Finland vs 25% in the US and 65% of those aged 18 years in Finland vs 69% in the US).65 The model predicts the percentage of cervical cancers attributable to HPV-16 that would be prevented if 70% of Finnish girls/young women were vaccinated when they were aged 12 years, 15 years, 18 years, or 21 years, with vaccination postulated to be followed by lifelong protection (Fig. 2). The number of cases protected is similar for vaccination at ages 12 years and 15 years, because so few Finnish girls are sexually active at these time points. However, if the vaccine were not given until they were aged 18 years, it would prevent only approximately 50% as many cases; and, if it were given when they were aged 21 years, it would prevent only approximately 20% as many cases. Merck recently presented results at conferences suggesting that the vaccine may be as efficacious in preventing incident infection and disease in women ages 27 to 45 years who happen to be naive to HPV-16, HPV-18, HPV-6, or HPV-11 as it is in similarly unexposed, younger women. However, as the number of unexposed women decreases greatly with age, the number of cancers prevented is likely to be inversely proportional to the age at which women are vaccinated. Independent of choices by individual women, this population-wide phenomenon is likely to influence public health-oriented implementation recommendations for women aged >26 years.

Figure 2.

Model, from Finnish women, of the proportion of annual incident HPV16-associated cervical cancer cases prevented with different ages at vaccination, if coverage is 70% of females only and is initiated in 2008. Reprinted by permission from Macmillan Publisher Ltd: British Journal of Cancer, 2007;96:514–518, Copyright 2007.

It will be important to determine whether the vaccine will maintain its excellent safety record and to monitor the actual duration of protection afforded by the vaccines. If it is observed that protection wanes, the 1 or more booster vaccination(s) most likely will be needed during a woman's lifetime to maintain a high level of protection. The administration of boosters, of course, would increase the overall cost and logistical complexity of vaccination.

Another important issue is the public health benefits of vaccinating men if it is determined that the vaccine is protective in them. Vaccination of women is the top priority for public health, because, worldwide, 90% of HPV-associated cancers occur in women, and modeling has suggested that, if there is high vaccine coverage of women, then the benefit from vaccinating men would be rather limited.64 However, if a country has sufficient resources and only a minority of the target population of women actually is being vaccinated, then the vaccination of men most likely would contribute to reductions in infection of both men and women, although it would do so less efficiently than if those vaccinations were given to women.

Potential second-generation vaccines

Several second-generation, prophylactic vaccines are under consideration and/or active development.66 However, with the exception of upper respiratory delivery of purified VLPs, these approaches have not been tested clinically.67 Second-generation vaccines seek to address 1 or more of the inherent limitation(s) of the current vaccines, such as high cost of production and implementation and type-restricted efficacy. In a simple extension of the current approach, both companies are considering increasing the number of VLP types in their vaccines. Increasing VLP valency likely would increase the percentage of cervical cancers that would be prevented by the vaccines. However, after HPV-16 and HPV-18, no single type accounts for more than a few percentages of cancers. From a broad public health perspective, therefore, this approach would be cost-effective primarily if it did not constitute a further increase in the cost of vaccination or if could lead to decreased costs of cervical cancer screening (see below). Although such a multivalent vaccine most likely would be of value in the industrialized world, the increased manufacturing complexity could delay further the time when the vaccine may be affordable in the developing world.

Live bacterial vectors that express L1 may be inexpensive to produce and deliver if they were given orally. Live L1-recombinants of the widely used Salmonella typhi vaccine strain Ty21a induce high titers of neutralizing antibodies in mice,68 and a clinical trial of this vaccine is under consideration. Decreased production and delivery costs also might be achieved with VLPs produced in plants69 or with L1 capsomeres produced in bacteria.70, 71

Most approaches for second-generation vaccines are based on L1, but neutralizing antibodies also can be generated against L2, the minor capsid protein. Unlike the case for L1, relatively short L2 polypeptides can induce neutralizing antibodies.72 It has been observed that these antibodies are broadly cross-neutralizing against divergent HPV types and can induce protection against heterologous virus challenge in animal papillomavirus models,73, 74 raising the possibility of a simple monovalent vaccine that would prevent a broad-spectrum of HPV infections. However, the titers of L2-neutralizing antibodies achieved to date are considerably lower than the neutralizing titers induced by L1 VLPs, and current efforts are focused on increasing the immunogenicity of L2 vaccines.

Secondary Prevention

Introduction

Despite the high efficacy of the vaccine against the high-risk HPV types that cause the greatest morbidity and mortality, cervical cancer screening will continue to play a critical role in the control of cervical cancer.75 It will take at least 2 decades for a prophylactic HPV vaccine program targeting a cohort of adolescent girls to reduce substantially their incidence of invasive cervical cancer in middle age and beyond (Fig. 2).76 In addition, even if worldwide vaccine coverage were achievable today, secondary prevention would be required, because the vaccine does not appear to alter the natural history of prevalent infections, and some cancer-causing infections will not be prevented by the vaccine.57, 58

Nonetheless, it is important to start thinking now about HPV vaccination and cervical screening programs together, in part because successful vaccination programs will affect screening performance much sooner than they reduce cancer rates. The vaccines decrease the high-grade cancer precursors that screening is designed to detect and treat. And, as vaccine coverage increases, more polyvalent vaccines presumably are introduced, and screening programs will have to change to remain cost-effective (see below). In the interim, the use of improved cervical screening methods that recognize the central etiologic role of persistent HPV infection could save hundreds of thousands of lives, especially in low-resource regions.77

Current cytology- and colposcopy-based programs

Current conventional cervical cancer prevention programs include repeated rounds of screening of women in the general population, triage of equivocal screening results by additional testing or heightened surveillance, histologic diagnosis of abnormal screening results by colposcopically directed biopsy, postcolposcopic follow-up if no immediate treatment is performed, and assessment of cure if treatment is performed.78 Currently, these programs rely mainly on the microscopic (cytologic and histologic) and visual (colposcopic) correlates of HPV infection to predict the cervical cancer risk of women with different screening or diagnostic results.

New technologies for high-resource settings

New technologies provide several options for the improvement of cervical cancer secondary prevention. Improvements in cytology slide preparation and interpretation have increased screening efficiency and may increase sensitivity.79 However, to our knowledge to date, no computer-assisted technology has been proven to dramatically increase accuracy.80, 81

In a few countries, testing for HPV directly by molecular assays already is being used with cytology to provide better risk prediction and the possibility of fewer cycles of screening.82–84 HPV DNA testing is more reproducible than cytologic screening and colposcopy for the detection of existent and incipient cervical precancerous conditions and cancer.85–87 A negative test for carcinogenic HPV types, as a corollary of the high sensitivity of HPV testing, provides a degree and duration of reassurance not achievable by any other diagnostic method. However, especially among young women, a single positive HPV test has low specificity and poor positive predictive value.16 Conversely, HPV persistence of approximately ≥2 years predicts the substantial risk of a diagnosis of high-grade CIN within the subsequent 5 to 10 years.15, 26 Few women or physicians will want to wait to know whether an HPV infection clears. It would be very useful to have biomarkers that can predict carcinogenic HPV persistence and the risk of progression to high-grade CIN, obviating the need for follow-up. Validation of such a biomarker with high sensitivity and high specificity even may permit the use of molecular testing (with or without HPV assays) at younger ages at which HPV testing alone is nonspecific. For example, it has been proposed that p16INK4 assays88 or E6/E7 expresson89 may serve this role, but sufficient validation trials have not been conducted to date.

With the data already at hand, the use of HPV testing to complement or even replace cervical cytology for screening is likely to be accelerated by the results from a series of randomized controlled trials reported in the last year, all of which demonstrated that testing for high-risk HPV types is more sensitive than cytology for the early detection of high-grade CIN.90–93 It is important to note that HPV testing detected the same number of cases as cytology in those trials, but earlier, suggesting that the molecular tests are detecting true cancer precursors. In addition, these data also suggested that the choice of HPV testing at longer intervals, versus repeated cytology at shorter intervals, is largely a matter of programmatic efficiency and cost, at least in high-resource regions where patients infrequently are lost to follow-up.

Similarly, in comparative trials, it has been demonstrated that HPV testing is more efficient than repeat cytology in the triage of equivocal cytologic interpretations.94 Because persistent infections with high-risk HPV genotypes cause cervical cancer and all true precursor lesions, HPV testing identifies possible cervical cancer precursors and avoids unnecessary follow-up or treatment of the great number of ‘look-alike’ lesions.95 Finally, HPV testing is being proposed, based on recent studies, for use in following women postcolposcopy and post-treatment.87, 96

Applicability of HPV-based methods in low-resource settings

New technologies tend to be expensive and difficult to implement; however, because >80% of all cervical cancer and its related mortality occur in low-resource settings,1 it is desirable to adapt new technologies for these underserved populations. Cost, infrastructure, and acceptability must be addressed to achieve widespread use. In low-resource regions, screening is most effective if women are reached at the ages of peak risk of treatable precancerous conditions attributable to persistent infection (10–15 years after the population median age of sexual debut) and before the average age at which most frankly invasive cancers occur.77, 97 It has been demonstrated that simple visual inspection with acetic acid reduces cervical cancer incidence in low-resource settings.98 Although this technique saves lives, it does not appear to be sensitive or specific enough to be considered the optimal, long-term approach.

It is appealing intuitively, if feasible, to use secondary prevention methods for cervical cancer based directly on HPV detection. Candidate HPV-based tests are being developed currently into rapid, robust, easy-to-use, and inexpensive formats.99 In low-resource regions, screening might target women a total of once or twice in the age range of peak risk of treatable, high-grade CIN and earliest cancer.97 One-visit ‘screen-and-treat’ strategies would minimize loss to follow-up that frequently reduces the effectiveness of screening programs. Two limitations of screen-and-treat strategies still remain. It will be difficult to fully rule out lesions that require advanced care.100 Also, full success of this innovative strategy awaits development of improved safe, inexpensive, and effective outpatient treatment for HPV-positive women.101, 102

The effect of HPV vaccination on all forms of cervical screening

In any setting, the amount of public funding to be used for HPV vaccination will be balanced against the need to continue screening. This balancing act will be made more difficult because screening gradually will become less cost-effective after the widespread use of HPV vaccines, as described below.103 Although the elimination of CIN-3 and cancers attributable to HPV-16 and HPV-18 will be welcome, vaccination will leave behind more equivocal and less predictive abnormalities, because the most evident and risky abnormal results determined by cytology, HPV tests, and colposcopy are caused by HPV-16. For vaccinated women, therefore, the positive predictive value of a positive result for CIN-3 and cancer will decrease for cytology, HPV testing, and colposcopy.103

For cytology, vaccination will reduce the number of high-grade cytologic findings disproportionately to the reduction in equivocal and low-grade abnormalities.104 The effect of vaccination on testing protocols that rely on HPV tests that pool all carcinogenic types will parallel the issues raised for cytology, because the power of HPV testing derives in large part from the detection of HPV-16 and HPV-18.16 With HPV testing, as with cytology, the number of HPV infections identified that predict cancer or high-grade CIN will diminish relative to the number of HPV-positive women who are not destined to get cervical cancer or even high-grade CIN. Successful vaccination also will affect colposcopy (and, presumably, its lower cost replacements, such as visual assessment with acetic acid) as strongly as, if not more than, it will affect cytology.105 The visual appearance of cervical HPV infections, as evaluated by colposcopy or derivative techniques, is even more highly variable and difficult to classify than cytopathic effects. In fact, the lack of reproducibility and accuracy of colposcopy represents a considerable but underappreciated clinical challenge.106, 107 HPV-16 is associated with the highest probability of clearly recognizable lesions, including those lesions that lead to a histologic diagnosis of high-grade CIN.108 Therefore, the removal of HPV-16 by vaccination will leave an even greater challenge for colposcopists approaching the already difficult task of targeting lesions for biopsy diagnosis.

The considerations discussed earlier suggest that vaccination duplicates important parts of screening, doing a large part of the same job of raising safety and removing the most dangerous cervical abnormalities. Vaccination, as described above, will reduce some of the underlying value and efficiency of screening, but it will not eliminate entirely the need for screening. However, if the long-term intensity of screening were to remain unchanged indefinitely for vaccinated cohorts, then it would mean that the cost of vaccination simply had been added to the current cost of screening, except for the relatively modest savings resulting from the decrease in the number of positive screening tests.

One potentially cost-effective, long-term strategy to consider would be to raise the age of initiation of screening for vaccinated women. Emerging data suggest that the risk of early cancers in young women may be linked preferentially to HPV-16 and HPV-18.35 If that is true, and if data demonstrate further reductions in early cancer risk for vaccinated women, it may be justifiable medically not to begin screening until the middle 20s or even later in some regions of the world. In addition to initiating screening later, we may want to stretch out screening intervals if it is proven that vaccine durability is truly long-term, if it is demonstrated that boosters are cost-effective, and especially if HPV testing is added routinely to cervical cytology.

Conclusions

The justifiable excitement over primary prevention by vaccination should be coordinated with screening efforts. Further understanding of HPV as a uniquely powerful human carcinogen also remains an important research goal. In the postvaccination era, screening must continue but will need to be changed to preserve the cost-effectiveness of the total program. The challenge will be to screen women in high-resource regions appropriately while applying the latest advances rationally and equitably to low-resource regions.

Ancillary